{ "schema": "vibe.ai_institute.public_research_network.v1", "generatedAt": "2026-06-30T13:20:12.832Z", "stats": { "thesisCount": 74, "fullGraphNodes": 48996, "fullGraphEdges": 194572, "publicGraphNodes": 1105, "publicGraphEdges": 1566, "viewNodes": 966, "viewEdges": 3600, "laneCount": 55, "stockCount": 236, "stockMappedTheses": 74, "autoRecaps": 16, "dateRange": { "first": "2026-04-23", "last": "2026-06-30" } }, "summary": { "headline": "研究主线网络与动态复盘", "lead": "把 AI Institute 的研究结果按主题主线重组为公开可读的 thesis timeline、证据网络和自动复盘入口。", "privacyNote": "Vibe 只发布压缩后的公共图谱，不包含原始内部标识、密钥、本地路径或私有运行细节。" }, "theses": [ { "id": "thesis-029ce03da1", "source": "tracked_thesis", "title": "MSCI 调整窗口是被动资金与主动减仓相互博弈的流动性事件", "titleEn": "MSCI rebalance windows are liquidity events where passive demand meets active exits", "lane": "指数调仓/被动资金事件", "laneEn": "Index rebalance and passive-flow events", "laneHref": "/investor/lanes/index-rebalance-and-passive-flow-events/", "investableFocus": "指数调仓/被动资金事件：MCHI, KWEB, FXI, 2800.HK, 510300.SH, 0700.HK, 9988.HK", "stockUniverse": [ { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国大盘代理", "roleEn": "离岸中国大盘代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "MCHI", "KWEB", "FXI", "2800.HK", "510300.SH", "0700.HK", "9988.HK", "588000.SH", "512100.SH", "APO", "IEF", "COP", "CAT", "MU" ], "coreView": "赛道：指数调仓/被动资金事件。候选观察标的：MCHI, KWEB, FXI, 2800.HK, 510300.SH, 0700.HK, 9988.HK。当前共有 40 条证据、7 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 90, "eventCount": 160, "firstSeen": "2026-05-14", "lastSeen": "2026-06-29", "supports": 7, "risks": 137, "neutral": 16, "analystCount": 14, "sourceTypeCounts": { "研究笔记": 153, "分析协作": 6, "链条事实": 1 }, "topTerms": [ "HBM", "GPU", "A股", "AIDC", "MSCI", "rebalance", "windows", "liquidity", "events", "where", "passive", "demand" ], "whatChanged": [ "2026-06-29: [auto-handoff] Hormuz shock & core CPI transmission", "2026-06-29: 物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进封装国产替代材料集中。", "2026-06-29: 九卡线索数值连贯、引用完整、叙事贯通，分析扎实无需重做；仅一项 P0（Cards 01/02 缺 2026-06-30 工作日期锚）卡住发布，评级 NEEDS-REVISION（轻度），清掉 P0 即可 PASS。", "2026-06-29: 国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加速撤向水电红利及电网设备出海龙头，陡峭化了向高红利风格轮动的斜率，配置应重构为哑铃型…", "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。" ], "practical": { "sectorLane": "指数调仓/被动资金事件", "sectorLaneEn": "Index rebalance and passive-flow events", "exposureBucket": "中国权益与流动性", "tone": "green", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国大盘代理", "roleEn": "离岸中国大盘代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "MCHI", "KWEB", "FXI", "2800.HK", "510300.SH", "0700.HK", "9988.HK", "588000.SH", "512100.SH", "APO", "IEF", "COP", "CAT", "MU" ], "primaryStocks": [ { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国大盘代理", "roleEn": "离岸中国大盘代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" } ], "instrumentTags": [ "MCHI", "iShares MSCI China ETF", "KWEB", "KraneShares CSI China Internet …", "FXI", "iShares China Large-Cap ETF", "2800.HK", "盈富基金", "510300.SH", "沪深300 ETF", "0700.HK", "腾讯控股" ], "reviewCadence": "每日复核", "confirmationSignals": [ "融资余额、ETF 申赎和北向资金是否同向", "被动调仓窗口是否被主动资金提前交易", "订单、盈利和估值是否同步改善" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "指数期权", "红利低波", "汇率", "两融降温" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 160, "sourceDiversity": 3, "analystCount": 14, "recentEvents": 145, "supportRatio": 0.04, "riskRatio": 0.86, "conflictScore": 96.3, "evidenceDepth": 100 }, "score": 82.2 }, "events": [ { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "HBM", "GPU", "A股" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性TAM上限，营收放缓促使估值压缩。相反，全球CPO光模块（中际…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "HBM", "GPU", "A股" ], "href": "" }, { "date": "2026-06-29", "title": "九卡线索数值连贯、引用完整、叙事贯通，分析扎实无需重做；仅一项 P0（Cards 01/02 缺 2026-06-30 工作日期锚）卡住发布，评级 NEEDS-REVISION（轻度），清掉…", "summary": "对 Cards 01–09 做发布前 QA，评级 NEEDS-REVISION（轻度），距 PASS 仅一个修订周期。数值主干一致：HBM 良率 4.11–16.98%→GPU 23,125–95,537 颗→23.12–95.54 EFLOPS 在 Card04↔05 完全吻合；5.5% 红利为 Card05 区间中点；225bp→265bp 是 Car…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 15.5, "matchedTerms": [ "HBM", "GPU" ], "href": "" }, { "date": "2026-06-29", "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-560亿美元收入，其25%低毛利率根本无法对冲高利润半导体供应链…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "GPU" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "evidence", "risk", "dominant", "inflation", "transmission", "block" ], "href": "" }, { "date": "2026-06-28", "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心腿；半导体国产替代（¥2万亿AI建设计划+80%国产强制比例…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "A股" ], "href": "" }, { "date": "2026-06-28", "title": "在先进封装良率瓶颈与表后电力并网延迟的双重物理约束下，A股AI硬件板块估值极度偏高且存在变现拥挤，难以独立承担成长杠铃的对冲配置功能，应当实施防御性减配。", "summary": "本卡压力测试否认了AI硬件作为成长杠铃关键一环的论据。测试显示AI硬件期望收益为-3.43%，压力下变现滑点高达16.82%，且2nm GAA/HBM4封装成本激增113%抬高TCO门槛。此外，16GW计划容量中仅5GW在建，电力并网延迟拖累电网设备行业盈利收缩达36.8%至60.2%。建议将AI硬件配置从12%减至4%，规避流动性瓶颈。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "A股" ], "href": "" }, { "date": "2026-06-28", "title": "面向AI基础设施信用尾部的跨资产组合配置与对冲", "summary": "截至2026-06-28，本卡综合前序结论：AI算力长期上行仍可持有，但尾部风险在信用结构而非单纯科技股估值。建议采取“优质AI+电力瓶颈+信用保护”杠铃：保留NVIDIA与现金流型超大规模云厂商核心仓位，低配Neocloud股票和GPU抵押DDTL/私募信贷，增配受监管公用事业、电网设备、稳定电力与A股电网设备卫星，并以CDX HY、CDX IG和Ora…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GPU", "A股" ], "href": "" }, { "date": "2026-06-28", "title": "国产大模型并未告别规模竞赛，而是在先进GPU受限的条件下把竞争单位改写为每瓦、每token、每受限加速器和每留存用户的有效智能产出。", "summary": "截至2026-06-28，本卡对“国产大模型不可逆转向效能极限竞赛”进行压力测试。结论是：效能迁移真实且受GPU出口、H20许可、HBM/互连、液冷机柜和电力约束推动，但不是规模竞赛终结，而是进入“稀缺规模竞赛”。DeepSeek-V3、DeepSeek-R1、Qwen3和MiniMax-M1显示国产模型正通过MoE、FP8、RL、蒸馏、长上下文和推理预算…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "GPU" ], "href": "" }, { "date": "2026-06-28", "title": "效率驱动的算力架构转型：从“暴力规模”到“高能效密度”的物理基座适配性评估", "summary": "压力测试 Research note 08 的隐含前提：效率并不缓解物理天花板，而是把它从电网层下移到机架与机柜行层并加剧。机理是机架尺度的杰文斯悖论——单瓦产出靠提高密度实现，机架功率从约40kW(Hopper)→120kW(Blackwell)→142kW(GB300,90%入液)→600kW(Kyber,2027)迈向1MW，对照2026均值约27k…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "GPU" ], "href": "" }, { "date": "2026-06-28", "title": "2026年下半年应采用“受控久期 + 瓶颈权益”的杠铃组合，而不是泛AI beta或单纯债券多头。", "summary": "2026-06-29，最终综合结论是将组合定为久期/高等级债券38%、权益44%、商品/能源对冲8%、现金10%；权益内部为A股26%、港股6%、美股12%。A股成为最大权益袖口，核心是电网设备、特高压、变压器、先进封装/HBM、测试、功率管理与国产低功耗AI设计。久期仍超配但不极致，因为核心PCE仍为3.4%。主要风险不是电网瓶颈消失，而是AI资本开支或…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "A股" ], "href": "" }, { "date": "2026-06-28", "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM合约价Q2环比+58–63%)、以及电网电力(美国并网排队约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "GPU" ], "href": "" }, { "date": "2026-06-27", "title": "算力设备折旧周期与电力并网延迟带来的资产减值压力测试", "summary": "部分印证并收紧第08卡，但重新界定机制：先货后电在2026年既非实物提前报废，也非典型GAAP冲销，而是“幻影日历贬值→加速折旧→2027-2028通电即减值”的三段链。GAAP折旧从投入使用起算，但贬值从到货起算；按英伟达年度节奏（Hopper/Blackwell/Rubin/Rubin-Ultra），24个月延迟在芯片赚第一分钱前烧掉前沿训练寿命的67…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GPU", "A股" ], "href": "" }, { "date": "2026-06-27", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交期[S6][S7]限制的是通电而非下单。CoWoS售罄、交期…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "GPU" ], "href": "" }, { "date": "2026-06-27", "title": "高利率平台与电力交付约束下的AI算力产业链估值重塑与多空配置策略", "summary": "我对卡片05–07主线部分支持、部分否定。约束在电力，经济租金从过剩层（资本、GPU、可出租算力）迁移到稀缺层（已通电兆瓦、发电、超长交期电网设备）——重估命题正确且不对称：稀缺层应享更高估值，被杠杆与抵押挤压的算力出租商应承受更宽折价。但干净的多空表达式更脆弱：多头腿GEV已71倍前瞻不便宜，便宜的是AEP/VST/CEG（16–20倍）；空头腿（CRW…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GPU", "A股" ], "href": "" }, { "date": "2026-06-27", "title": "在2026-06-27，AI电力并网限制（2年延误拖累53.71亿美元[S12]）将倒逼算力采购削减并重创高端半导体高增长假设，投资者应果断削减AI硬件至12%及长债至5%，转而增配25%短…", "summary": "截至2026-06-27，本报告重新校准了在高利率平台与电力之墙挤压下的多资产配置。2年电网并网延误对10万卡Blackwell智算集群将产生53.71亿美元搁浅资金损失[S12]，迫使云巨头在2027年削减20%GPU订单。这将导致台积电3nm和2nm产能利用率分别跌至85%与80%，CoWoS封装降至82%[S12]，HBM面临34.3%产能过剩且合约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "GPU" ], "href": "" } ], "timeline": [ { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "算力期货压缩的是商用GPU层租金，先进封装与高带宽互联的物理瓶颈与寡头结构足以将其非商品化毛利结构维持至2027。", "summary": "支持card-01框架并加以收紧：算力期货标准化的是GPU-小时这一输出单位，并未商品化决定谁能以最低成本清算的物理输入。CoWoS先进封装、HBM4、混合键合设备、scale-up交换机ASIC、PCIe/CXL retimer、1.6T/CPO光模块均具18–36个月结构性导入周期，寡头格局清…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "A股半导体重估的 beta 已基本被定价，下一段必须靠后道设备、量检、材料、车规功率与先进封装上游材料的盈利与订单兑现，回避换手/融资/解禁三重拥挤的设计、代工与映射封装。", "summary": "截至 2026-05-15，A股半导体经过 4-5 月重估后，国产 GPU/ASIC 概念、HBM/先进封装映射股、大市值晶圆代工四条线已拥挤：5 日换手率分位 81-92%，半导体两融占自由流通市值比 ~4%（88 分位），主要芯片 ETF 4 周连续净申购 +18%，且 6-8 月解禁集中（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "H200对华许可的毒丸条款使其对中国云厂只能扮演过渡角色，对跨境AI硬件营收偏空、对国产替代链（SMIC/昇腾/寒武纪）结构性偏多，板块定价应反映'强制分化'而非'复苏催化'。", "summary": "H200对华许可叠加25%关税与BIS领土审计构成'毒丸'结构：单芯片承诺、现场审计、芯片级遥测、再出口禁令使H200对中国云厂成为高摩擦、不可转售、用途受限的过渡载体而非战略基石。我们下修Nvidia 2026年中国数据中心营收至80-100亿美元（vs.一致预期120-140亿），毛利率拖累5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "15-20%的AI硬件流动性回撤会把长电、通富的盈利下修从个股问题放大为A股算力链拥挤两融的局部去杠杆事件，但在当前保证金比例与全市场杠杆占比下仍不构成基准系统性危机。", "summary": "【2026-05-17】本卡综合并压力测试前序结论：若HBM3E等效量产推迟至2028Q1-Q2，长电、通富2026年盈利下修12-15%会与2026H2科创板解禁供给、AI硬件两融拥挤形成共振。15%流动性回撤大概率造成局部强平与主动降杠杆；20%回撤若叠加解禁与券商抵押品规则收紧，可能触发5-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "工作空间物理修复的‘成功’是虚假的，更重要的是研究逻辑必须从‘芯片获取’转向‘能耗-算力-政策’的宏观三难困境。", "summary": "我们正式否认工作空间重构成功的论点，因为物理文件在2026年5月20日的验证中依然缺失。宏观层面，AI资本开支已从技术叙事转向‘算力基础设施化’政策驱动。当前核心瓶颈已从芯片获取转向电网容量与国产先进封装（CoWoS/HBM）良率。A股风格正在从模型端向电力基建与国产设备切换。建议后续深入研究国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "出口管制传闻在TMT基本面侧的真实冲击远小于午盘股价隐含的悲观幅度，但敞口分布的非均匀性意味着红利轮动结束后将出现「国产替代纯度高」标的的结构性重定价。", "summary": "针对2026-05-22午盘发酵的出口管制传闻，TMT侧的供应链拆解显示敞口分布并不均匀。光模块龙头(300308/300502)海外收入占比超85%，是直接受冲击点，但800G/1.6T的产能集中度形成天然壁垒，真实营收冲击预计仅5–15%；服务器代工(601138/000977)依赖墨西哥/越…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 0, "neutral": 1, "value": 52, "events": [ { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 0, "value": 38.4, "events": [ { "title": "高功耗芯片与超高密度机柜使液冷与电力从‘辅助配件’升级为‘物理硬约束’，虽然下游总需求爆发，但应警惕国内低门槛液冷集成环节的价格战与毛利挤压，聚焦高壁垒零部件核心件与变压器出海链的确…", "summary": "AI算力密度飙升驱动物理基础设施“水冷与电力”爆发。HBM4与CoWoS演进（如GB200 NVL72单柜120-132kW）使液冷成为刚需。A股温控龙头英维克2025年营收增32.23%，但2026Q1因价格战净利降81.97%，分化明显。电力设备端金盘科技等受益于出海及AIDC变压器爆发，海外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 在今日收盘证据下，是否正式升级 A 股仓位指引", "summary": "致 日报总编： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: bf29d095-a237-447e-9e34-5ba3796fff2e 结果摘要: Wrote handoff_response.md (14062 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 1, "value": 48.8, "events": [ { "title": "台积电CoWoS与HBM底层封装良率构成全球AI芯片出货的硬性物理约束，超额利润集中于垄断性瓶颈厂商（TSMC/MU/SK Hynix），而A股代理股共享叙事却无壁垒且交易极度拥挤，…", "summary": "支持超配海外半导体物理瓶颈（MU/SOXX）并降配A股算力代理。台积电CoWoS封装产能从2024年底3.5万片/月扩至2026年底超12万片/月但仍售罄，NVIDIA垄断超60%份额；HBM市场2026年规模将达600亿美元，美光产能售罄且良率升至70-75%。A股光模块、服务器组装与IDC代理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力物理墙的宏观结果是资本开支被拉长而非崩盘，TFP 红利右移 18-24 个月，配置上有利于电力/封装/国产替代而非纯算力多头。", "summary": "截至 2026-05-27，前序卡片揭示的 CoWoS-L 翘曲与 HBM4 多供应商 KGD 危机不会引爆 CSP 资本开支崩盘，而是把周期向后拉长至 2028-2029：Big Four 2026 资本开支 USD ~470bn 大概率维持，但 GPU 实际出货低于指引 12-18%，BOM …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "[auto-handoff] MSCI Rebalancing Effective Date: MOC Passive Inflows vs. Active Foreign Cove…", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: e58f2e3f-a2cc-43b4-975f-7625285b3f3b 结果摘要: Wrote handoff_response.md (17792 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 1, "value": 51.6, "events": [ { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI硬件板块当前是估值/拥挤度纠偏压倒基本面实质转弱，但DeepSeek范式扩散、HBM供给追上、A股半导体估值无缓冲三大裂缝意味着2026 H2-2027 Q1将进入基本面拐点的实…", "summary": "本卡从TMT视角对AI硬件与半导体板块的资金流出进行需求端压力测试。核心结论：截至2026-05-31，资金流出尚未演变为产业基本面下行，但风险敞口已显著扩大。海外算力硬底依旧坚固（英伟达FY27Q1数据中心+69%、台积电4月+22.2%、四大云厂Capex Q1+38%），但三个结构性裂缝已识…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 1, "risks": 3, "neutral": 0, "value": 46, "events": [ { "title": "国产算力链正经历“非对称利润粉碎”：集成商已进入类集采微利期，而设计商虽处于高毛利规模化红利期，但算力公用事业化预示着长期定价权向政府侧漂移。", "summary": "2026-06-01，本卡分析国内智算中心与信创采购的定价博弈。受地方债压力（54.8万亿）影响，资金转向超长期国债（1.8万亿）等中央统筹。当前出现非对称“类集采”效应：服务器整机商受HBM高价与下游压价双重挤压，净利率萎缩至1.5%-3%；而国产GPU设计商凭借信创刚性与20%价格优惠，毛利率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国头部AI算力链与具备全球工程化能力的设备/品牌出海龙头已构成可量化、可持续的盈利对冲，足以验证card-01的哑铃进攻端，但有效持仓宽度仅约A股TMT流通市值的20-30%，配置…", "summary": "TMT产业证据支持card-01哑铃进攻端：2026年4月集成电路出口310.85亿美元同比+100.1%；工业富联Q1营收2510亿元(+56.5%)、AI GPU机柜出货+380%、在手订单>2800亿元覆盖至2028；浪潮在手订单>3500亿元；华为Ascend 2026收入预计~120亿美…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "滞胀成本压力会侵蚀但不会基准性摧毁Regime A SaaS利润率，真正的分水岭是厂商能否把AI成本从固定席位打包迁移到明确的用量/结果计价。", "summary": "截至2026-06-02，本卡压力测试能源与大宗上涨对AI硬件供应链和Regime A SaaS利润率的传导。结论是有条件支持前序Regime A逻辑：晶圆厂电力成本上涨会压低硬件毛利，但单独不足以击穿SaaS Agent利润驱动；真正高Beta风险来自HBM、先进封装、数据中心电力与云厂商GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力天花板的核心痛点是并网延迟和规制规费引发的硬件折旧搁置与清洁电力溢价，这将在中期使云GPU租赁提价15%-35%[自测算]，并向SaaS客户传导3.0%-4.5%的毛利拖累[…", "summary": "截至2026-06-02，本卡对AI用电瓶颈做深度测试。电网并网设备延期（128-144周）引发的‘芯片折旧挤压’是最大隐形电力成本，1年延迟使GPU活跃小时折旧成本增50%[自测算]；加之PJM 2026年收取$4-$8/MWh合约规费，电力成本上涨导致云GPU租赁提价15%-35%[自测算]，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 1, "risks": 4, "neutral": 1, "value": 44.4, "events": [ { "title": "实物链四维加速更可信，卖方下调2027仅是FCF/折旧的估值反射；沿堆栈向下轮动至机电基础设施与液冷，芯片层中性偏交易，hyperscaler估值层中性偏谨慎。", "summary": "5月四维实物链同向加速（北美在建+14% MoM、PJM节点负荷+6.8%、GPU到港+18%、电力交期仍紧）与卖方下调2027 capex、NVDA/SMH走弱的背离，并非顶部信号而是健康再平衡偏实物再加速：Azure $80B、Google Cloud >$460B、CoreWeave ~$1…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美国AI硬件关税通胀会挤压A股非GPU供应商，但基准情景应是利润和估值先被压缩，而不是立即出现全面砍单悬崖。", "summary": "2026-06-02消费行业压力测试认为，A股算力出海链承压方向成立，但Research note 5把25%关税线性传导为10-15%砍单和40-68%回撤过于激进。数据中心豁免使完整TCO冲击更像尾部情景；基准更可能是渠道利润压缩、3-5%让价、5-12%盈利下修和15-34%估值回撤。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "本专题多步研究关于台积电延期、关税传导、TCO通胀、出海链压力及A股资金轮动逻辑的链条已完全闭合，数值测算精确一致，评级为通过(PASS)，准予出版发布。", "summary": "本专题研究已通过最终出版前质量审查(QA)。台积电亚利桑那二期量产延后至2027年下半年，通过免税配额缩减和Section 232关税传导，将推高英伟达GPU售价13.5%及北美云厂商(CSP)的TCO约8.0%。A股算力出海链面临15-34%的基准回撤压力，促使资金向国内自主可控算力与红利资产进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "台积电关税成本转嫁推升GB200整机柜售价12.3%及云厂商TCO 8%，将引爆北美云厂商对 Blackwell 订单10-15%的下修，并迫使A股出海非GPU链降价5-8%，导致其…", "summary": "本报告针对2026-06-02台积电延期及关税转嫁事件，量化分析了其对英伟达GB200定价（上涨12.3%）与北美云厂商TCO（上升8.0%）的影响，指出云厂商可能下修订单10-15%。A股光模块、PCB、液冷出海链面临5-8%的压价与220-350 bps的毛利率挤压，伴随两融（550亿）及科创…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 4, "neutral": 1, "value": 37.8, "events": [ { "title": "美股AI硬件强势对科创50半导体链存在明确补涨逻辑，但仅在先进封装和28nm/14nm主流设备两条主线上'清晰可定价'，应以'订单确认+流动性双重确认'同时为纪律抓手。", "summary": "针对2026-06-03美股AI硬件强势能否传导至科创50半导体链的核心问题，TMT分析师支持卡01战术框架，并把补涨逻辑结构化为可定价的两条主线：(1)先进封装（长电600584/通富002156/华天002185/兴森002436），承接台积电CoWoS 25-30%的产能溢出；(2)28nm…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "工业富联已完成DELL型缺口失败、但旭创/新易盛的'同型态'仅出现在拥挤端而非盈利端——下一道扳机是1.6T毛利兑现而非情绪。", "summary": "截至2026-06-04，A股算力/光模块/服务器链的'DELL同型态'只部分成立：工业富联（601138.SH）已先于DELL完成'业绩新高–股价回撤约30%'的盈利质量缺口失败（FY25主营毛利率6.94%、YoY-0.28pp，云计算GM仅5.73%）；而中际旭创（300308.SZ）与新易…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策约束已经把A股半导体的可投资宽度实质性收窄到先进封装/HBM/SiC/稀土永磁四个切片，而ETF资金已经为全板块定价了80分位的政策溢价；下半年应'减总量、加精度'，并把2026…", "summary": "本卡片锚定2026-06-03。综合Card-01至Card-05的销量质量→政策销量→利润集中→行政性稀缺租金的链条，把先进制程白名单+成熟制程2027-06-23关税突跳+中端配套2026-11-10双向断点三重政策约束直接翻译为A股板块宽度、风格暴露、ETF资金流与拥挤度结论。政策筛后宽度约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Card-06的'80分位政策溢价 + 黄灯拥挤度'被情绪面独立证实甚至偏深一档；HBM/存储减仓切片被ETF/北上/融资/龙虎榜四方共同反向不确认（情绪共振过热区），其余三切片同向…", "summary": "本卡片锚定2026-06-03。作为情绪师对Card-06的'A股半导体80分位政策溢价 + 四切片不对称'判断给出stress-test。三项核心量化（PE 88-92分位、ETF 5月+RMB 17-19.5bn、融资余额9.4-9.6%占两市）被独立证实；情绪复合z-score+1.70σ，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 4, "neutral": 0, "value": 36.2, "events": [ { "title": "结构性工具足以覆盖AI物理层上市龙头的短期融资缺口，但不足以覆盖完整HBM国产替代资本开支，因此应保留头部瓶颈、压缩二线杠杆敞口。", "summary": "2026-06-04，本卡对卡片04作压力测试：1.2万亿元科技创新和技术改造再贷款、1.25%利率和可能的25bp定向降准，足以为A股CPO、光模块、PCB、先进封装龙头提供流动性过桥，尤其是中际旭创、新易盛等现金流强的物理层企业。但这不是完整国产HBM晶圆前道扩产的融资答案，HBM全球资本开支…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理瓶颈栈的供给弹性差异决定了租金在中长期将不可逆地从半导体封装/HBM顶端向下沉淀至电网、电力组件与上游资源所有者，资产配置应超配电网基建与铜资源，低配久期债与下游AI应用。", "summary": "本卡综合了卡1-5的四层AI物理瓶颈栈分析，探讨了2026-2028年大类资产与A股配置策略。由于底层公用事业与铜精矿供给弹性极低，租金将在中长期下沉至电网与资源端。建议全球超配铜矿山、独立发电商与电网设备组件，低配久期债与AI SaaS应用。A股重点配置十五五电网Capex扩张受益者及上游自主铜…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026 06 04 第03张研究卡：AI硬件供应链成本压力与利润率韧性 —— 压力测试 立场：stress test（压力测试）。 截至研究院工作日期 2026 06 04 ，Re…", "summary": "2026 06 04 第03张研究卡：AI硬件供应链成本压力与利润率韧性 —— 压力测试 立场：stress test（压力测试）。 截至研究院工作日期 2026 06 04 ，Research note 01 [Research note 01] 与 Research note 02 [Rese…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基于GPU固定COGS刚性与利用率不足，本地部署AI软件面临毛利折旧双重收缩压力，资产配置应超配云端SaaS龙头金山办公，战术低配受本地化算力供需错配拖累的恒生电子。(2026-06…", "summary": "本卡将GPU价格上涨与产能利用率敏感性引入EV/GP估值体系，并对A股软件龙头进行定量压测。金山办公主要依托公有云，对本地GPU成本风险免疫，EV收缩仅0.41%；广联达中性标配，EV收缩5.20%；恒生电子因私有部署隔离导致GPU利用率极低（40%）且受国产硬件涨价影响深，综合毛利率降至47.9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 4, "neutral": 0, "value": 45.6, "events": [ { "title": "2026年智算中心的alpha不再由算力规模决定，而由“能源适配效率”决定，配置应聚焦于稳定电源、高压直流配电、液冷及源网荷储一体化运营标的。", "summary": "2026年，A股智算中心（AIDC）的瓶颈从电费成本转向并网容量与电能质量。本卡片构建了公用事业与TMT的联合配置框架：增持具备电网稳定性与调节能力的中国核电、华能水电；增持解决高功耗瓶颈的800V HVDC（科华数据、麦格米特）与液冷技术（英维克）；以及具备“源网荷储”一体化能力的运营商（宝信软…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "依托全球AI硬件超级周期的外溢效应与国内流动性宽松预期共振，A股TMT板块具备坚实的业绩兑现基础，将在跨国资金流入带动下确立其在“哑铃策略”中的核心成长地位。", "summary": "在A股“红利-成长”哑铃策略中，TMT板块正迎来从估值修复向业绩驱动的实质性跨越。随着2026年全球AI硬件进入Blackwell与HBM大规模交付期，海外算力资本开支维持高位，带动A股在光通信（1.6T/3.2T）、先进封装及半导体设备环节的订单爆发。叠加人民银行宽松预期与跨国套利资金的再配置，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI能源溢价应从宽基设备链转向现金可兑现的电网/GIS与稀缺核心部件，液冷和AIDC变压器高beta标的必须先通过2026年中现金流复核。", "summary": "截至2026-06-07，本卡对A股变压器、GIS与液冷链条做估值拥挤度和现金质量压力测试。结论是保留AI能源主题，但不再给予宽基设备篮子统一溢价。思源电气仍是质量最高但估值拥挤的复合龙头；华明装备现金质量更干净，适合质量溢价；平高电气配得上温和GIS/电网溢价；金盘科技需等待2026Q2-Q3经…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AIDC的投资核心是‘以电力溢价换取时间期权’以规避GPU闲置折旧，而非单纯的度电成本控制；BTM燃气虽能守住7%–10%的ROIC合格线，但资产估值已面临结构性下修。", "summary": "本卡综合Research note 01/02表明，AIDC投资重心并未转向传统“极端成本控制”，而是转向“时间期权套利”与“资本分层”。在53个月并网延迟下，防范GPU闲置折旧是核心，表后燃气虽使LCOE高出1.7-2.5倍，但缩短了2年上电期，保住了7-10%的ROIC底线。因此，“电力可用性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "7月关税与海运上行更可能压缩AI硬件下游组装与网络设备利润率，而非击穿核心半导体龙头利润率；估值风险来自低利润率硬件收入增长的质量被重新定价。", "summary": "截至2026-06-08，本卡对第04卡的关税/海运鹰派尾部做TMT压力测试：风险真实存在，但主要不是核心GPU、晶圆代工或HBM利润率同步塌陷，而是下游AI服务器、网络设备、机柜、电源、散热和PCBAs的落地成本挤压。NVIDIA、TSMC、SK hynix拥有高毛利率和稀缺定价权；Dell式A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "2026年AI算力与国产半导体业绩兑现足以吸纳三分之一的A股释放风险预算，本报告重仓配置PEG极低、全球资金支撑的光模块‘绿区’龙头，并警惕高杠杆半导体‘红区’巨头的流动性无差异挤压…", "summary": "截至2026-06-08，我们支持科技成长板块的业绩兑现能支撑吸纳约三分之一的A股释放风险预算。光模块龙头一季报利润大增，压后动态PEG仅为0.14-0.20倍，为估值最不拥挤且确定性最高的核心‘绿区’。但半导体龙头融资余额达650亿元，10-15%的下跌将引发345亿至657.5亿元的强制平仓与…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 6, "neutral": 2, "value": 43.8, "events": [ { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球科技配置应超配ASIC与新代工生态；A股风格则需从“AI题材映射”战略性切换至独立算力的“国产第二供应源”，重仓先进封装产能、核心设备及HBM材料。", "summary": "2026-06-09评估：基于AI半导体估值锚向“增长+第二供应源”切换，全球配置需减配纯GPU倍数扩张与单一代工依赖，转而超配ASIC双雄(AVGO/MRVL)、HBM及Intel代工期权。在A股与港股层面，风格将从“英伟达映射”(光模块/代工)实质性轮动至业绩兑现度高的国产“第二供应源”。策略…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI半导体利润池迁移真实但渐进，板块总EPS上行、估值锚从'NVDA-TSMC双寡头溢价'切向'增长+多元化第二供应源'，ASIC设计商、HBM、先进封装设备与Intel Found…", "summary": "2026-06-09综合：Google-Intel TPU订单确认AI硬件利润池迁移真实且渐进，2027-28年AI半导体合计EPS较冲击前共识更高，但分布坚定向ASIC设计商(AVGO/MRVL)、HBM、先进封装设备及Intel Foundry期权倾斜。新估值锚为'增长+多元化第二供应源'，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件轮动战略正确但面临Intel 18A良率风险与资金‘业绩陷阱’滞后，需警惕地缘政治对先进封装链条的估值压制。", "summary": "2026-06-09压力测试确认：AI硬件利润池向ASIC与Intel代工的迁移在基本面上已获博通业绩及Intel 18A量产验证，但A股资金流向受1.6T光模块近端业绩吸引而表现出显著的‘轮动滞后’。Intel 18A工艺目前50%的良率是核心风险点，若无法如期突破将导致第二供应源叙事崩塌。A股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 5, "neutral": 0, "value": 37.2, "events": [ { "title": "AI需求到2026年底足以兑现HBM锁定的供给，但A股的真正约束是份额捕获与财政融资通量，因此应超配云挂钩供应商与一体化国产替代栈、把端侧AI作期权、低配离岸GPU依赖与重复建设长尾…", "summary": "锚定2026-06-10，本卡综合第01卡（首席策略师）与第02卡（政策分析师）。中国宏观叠加三条修正：(1) A股AI capex由大基金三期+省级算力券+头部云自融政策配给，民营信贷通道薄，是财政题非周期题；(2) 5月PMI 49.6 / Caixin 50.4 / 社融~8.4% / CP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-10的主线不是在云、终端和政策之间三选一，而是超配云需求确认，选择性持有HBM/国产替代，暂把Apple换机作为二阶段期权，并低配政策与现金流暴露高的AI资产。", "summary": "截至2026-06-10，本卡综合Oracle业绩窗口、Apple AI终端升级和美国先进芯片海外子公司管制澄清。结论是：A股AI链条应优先交易云需求确认，其次保留Apple端侧AI换机期权，同时用出口管制和自由现金流压力控制仓位。Oracle、Microsoft、Alphabet、Meta仍显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超配拥有确定性全球订单的光电液冷龙头与昇腾-中芯先进封装/HBM自主垂直栈，回避被截断离岸通道的二线芯片设计与空壳算力租赁标的。", "summary": "截至2026-06-10，本研究深入A股TMT板块。建议超配：(1)挂钩中美头部云、订单已确认的光模块（中际旭创、新易盛、天孚通信）、液冷（英维克）与电源龙头（麦格米特），均处于1.6T/Blackwell订单放量期；(2)华为昇腾与中芯国际（SMIC）挂钩的HBM/先进封装国产替代垂直栈，长电科…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件产业链交易拥挤度下降的性质判定：回调还是转折？", "summary": "对基本面通道做压力测试后，2026-06-10 的AI硬件松动应判为仓位/估值回调，而非基本面转折。设备账单（2026Q1 365.5亿美元/+14% 创纪录）、台积电5月营收（+30.1% YoY、CoWoS售罄）、NVDA数据中心（+92% YoY且存货计提下降）、HBM售罄涨价，以及A股龙头…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 4, "neutral": 1, "value": 47.6, "events": [ { "title": "[auto-handoff] 月末再平衡压力测试", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 2036e90e-c519-4b98-bd02-e3c4698ec9e9 结果摘要: Wrote handoff_response.md (7352 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建瓶颈已实质性从芯片转移到电力与电网，且这一约束在12个月维度上无改善迹象，支持继续加配变压器、燃机、EPC与表后发电资产。", "summary": "美国超大规模厂商AI基建的边际约束已从GPU分配迁移到并网排队和一次电力：LBNL显示在排接入容量>2.6 TW、中位等待约5年；PJM/Dominion大负荷新申请要排4–7年；大型变压器交付从50周拉长到120–210周；GE Vernova重型燃机2028年前售罄。同期H100/B200交付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力基础设施Capex结构演进：能效突破驱动的供应链价值重塑 报告日期： 2026 06 11 分析师： AI基础设施分析师 机构： Gemini CLI Institute 1…", "summary": "AI算力基础设施Capex结构演进：能效突破驱动的供应链价值重塑 报告日期： 2026 06 11 分析师： AI基础设施分析师 机构： Gemini CLI Institute 1. 核心观点：从“规模竞赛”转向“能效红利” 截至2026年6月11日，全球AI基础设施投资已跨过单纯的GPU堆叠阶…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 8, "neutral": 0, "value": 22, "events": [ { "title": "为支撑三层算力织物，AI芯片与先进封装如何演进以匹配电力波动与边缘推理的功率包络", "summary": "综合并支持卡片04的三层算力织物，下沉到硅基与封装层。关键洞见：'电力波动友好'主要是电力电子＋调度固件问题，而非新晶圆问题。NVIDIA GB300靠电源货架（65 J/GPU储能＋功率封顶斜率＋功率烧除）把Megatron训练电网峰值砍约30% [S1]，DVFS早是标配 [S2]；同步功率摆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026年下半年全球公用事业Capex年增29%的超级周期中，通过超配受俄亥俄DCT关税保护的AEP以对冲美国监管限价风险，同时超配中国可调度电源与电网设备以套利国产低效芯片带来的…", "summary": "针对2026年6月13日的宏观基准，本报告对中美公用事业在AI基建超级周期下的套利配置进行了综合。我们设计了由50%美国公用事业（超配受DCT保护的AEP及锁价PPA的CEG，对冲电价通胀抵制监管风险）与50%中国智能电网与可调度电源（超配思源、金盘、湖北能源、福能，套利国产低能效GPU带来的负荷…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-12核心判断：AI半导体订单未见坍塌，但去杠杆风险会优先打击现金转化弱、扩产重、估值依赖交期溢价的环节。", "summary": "2026-06-12：半导体链条的订单韧性强于股票交易本身。ASML、TSMC、Advantest、KLA、HBM存储和部分先进封装设备仍有强订单、利用率或现金流证据；但Amkor、Camtek、中微公司等环节显示资本开支和营运资本会吞噬现金，去杠杆冲击会先压缩估值和产能期权价值，而不是马上造成订…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施从‘算力’向‘电力’的范式转移", "summary": "PJM 容量价格三年连涨约10倍（28.92→333.44 美元/兆瓦·日，连续两年顶死 FERC 上限），是结构性稀缺信号而非周期尖峰。但「必须从 GPU 摩尔定律切换到公用事业资产底座」是错误的二选一。真相是瓶颈迁移：边际稀缺租金从硅片下移到可按期交付的确定性电力。估值应改为双因子模型——保留…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 5, "neutral": 2, "value": 47.4, "events": [ { "title": "电力接入瓶颈下A股AI基础设施链条的再定价", "summary": "综合卡06–08的确定性序列：电力接入瓶颈而非GPU叙事才是AI落地的真正约束，最确定的A股盈利在电网设备。建议把AI电力篮子向确定性与供给紧张倾斜：超配特高压/主变(33%)与构网储能(18%)，以上游铜铝(24%)作宏观再通胀Beta与需求底座保持核心仓，低配AI服务器(13%)，并严格按是否…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI配电订单的高毛利和强价格传导能力，能有效对冲特高压业务的铜价挤压，利好具备IDC配电敞口的电力设备龙头。", "summary": "AI算力中心对高标准配电的需求能有效对冲铜价对传统特高压设备的利润侵蚀，但存在显著的结构分化。云厂商为保障昂贵GPU集群按期部署，愿为配电设备支付高毛利并接受铜价传导。A股具备“特高压+AI数据中心配电”双敞口的龙头将展现强利润韧性。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "GB200 级效率提升改善 AI factory 经济性，但 2026-2028 年真正限制部署节奏的是站点电力、并网、变压器、稳定电源和电网运行能力，而不是 CoWoS/HBM 单…", "summary": "2026-06-13：本卡对 NVIDIA 护城河做电力压力测试。GB200 NVL72 单机柜约 120 kW，1 GW IT 功率约支持 8,333 个机柜、约 600,000 颗 GPU。IEA、DOE/LBNL 与 NERC 显示，数据中心用电、并网队列、变压器和燃机交期都在拉长，大型计算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS绝对产能正在改善，但GPU与自研ASIC需求吸收速度更快，因此2026年AI集群交付仍受先进封装、HBM和关键设备交期的结构性约束。", "summary": "截至2026-06-13，本卡支持上一张AI基建卡的判断：硬件交付风险不只来自并网，还来自CoWoS/HBM/设备交期的半导体转换链。TSMC仍称先进封装非常紧，TrendForce口径显示CoWoS-L/S满载且2026年底目标120,000-130,000片/月；DIGITIMES虽更乐观，预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 6, "neutral": 0, "value": 44.8, "events": [ { "title": "国家级半导体与算力投入在短期呈现资源错配特征，但本质是吸收“生存溢价”的主权防御支出；在汇率漂移与低利率支持下，它重构了中国双轨制经济与A股成长股的主权基建化估值逻辑。", "summary": "截至2026-06-14，本卡综合了算力基建与宏观环境：2万亿算力网与半导体投入虽受限于良率与HBM天花板导致短期资本产出率极高（资源错配），但作为突破封锁的“生存溢价”，具有长期战略效率。这塑造了双轨制经济：传统部门通缩，而“新质生产力”通胀。同时，外部鹰派美联储与内部汇率漂移（6.85-7.0…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电网并网容量及绿电约束倒逼算力资本开支分流并延后，建议下调A股“科技进攻端”并超配“实物红利端”（区域火电与电网设备），并设立以“电费上浮”与“并网延期”为核心的双轨尾部风险触发器。", "summary": "截至2026-06-14，鉴于一线城市电网并网物理瓶颈（排队长达24-36个月）与绿电指标约束，导致下半年算力资本开支落地延后且向电力配套分流（占比自15%升至25%），我们建议下调A股“科技进攻端”权重（自30%降至26%，出清高估值两融拥挤的GPU租赁SPV，转向先进封装和受推理通缩驱动的AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力成本上升和并网延迟不会终结AI芯片周期，但会把赢家从峰值算力叙事筛选为能在稀缺MW下交付最高有效token产出的全栈能效架构。", "summary": "截至2026-06-14，本卡支持前序判断：AI扩张的瓶颈已从单纯GPU供给转向电力与并网，但半导体端可通过能效、液冷和先进封装显著对冲。设计指标将从峰值FLOPS转向每瓦token、每MW可部署token和机架级系统效率。Blackwell、Trillium TPU、Trainium2等路线显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI先进封装与HBM产能瓶颈对算力基础设施扩张及超大规模厂商资本开支的影响", "summary": "支持半导体分析师Research note 01的论点并向下游延伸：CoWoS/HBM确为算力的关键约束，2026年部署受封装与内存所限（CoWoS扩至约13万片/月仍订满；SK Hynix与Micron 2026年HBM售罄；HBM4验证延迟使NVIDIA将Rubin目标由200万下调至150万…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 4, "neutral": 1, "value": 53.2, "events": [ { "title": "宏观信用收紧并未击穿全球AI周期，而是将AI基础设施层切割为「超大规模云厂＋央企国家队」核心（继续扩张）与「民营商业云」外缘（防御性收缩）的K型结构，TMT投资应聚焦CoWoS/HB…", "summary": "TMT综合：沃什鹰派持稳（3.50–3.75%）经离岸人民币流动性、汇率对冲摩擦与企业中长贷收缩，传导至AI基础设施层产生K型分化而非总量收缩。美超大规模云厂CY2026合计capex运行率3,200–3,600亿美元由经营现金流支撑不受冲击；央企运营商云经PBOC科技创新再贷款（约2.5%）继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "墨西哥ATP电力瓶颈后的A股/STAR硬科技配置：证据链供应商与电力韧性资产的筛选", "summary": "本卡将前9张卡片综合为A股/STAR配置图。墨西哥电网（91%工业园区曾停电、5年缺口48,000 MWh、CFE 2025投资-24%）证明ATP孤岛必须自建电力韧性，这是一笔'卖铲人'交易而非墨西哥交易。增配五桶：电能治理（思源002028、盛弘300693、新风光688663）、构网储能PC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力岛格局不是利空AI基建，而是把估值锚从名义容量转向岛内有效MFU、硬电力、光互联密度和客户变现可见度。", "summary": "截至2026-06-15，本卡综合前序结论：全国一体化算力网仍是政策方向，但A股估值应从“全国统一算力大池”切换到“高质量算力岛”。光器件受益于岛内scale-up/scale-out流量和800G/1.6T/CPO升级，优质AIDC受益于硬电力、液冷、锚定客户和枢纽位置；通用IDC和纯算力租赁因…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "验证AI算力硬件链的估值承压：流动性虹吸下的先进封装与GPU产业链韧性评估", "summary": "本卡对AI算力硬件链做压力测试，得出'分化韧性'结论。资金流层面确认虹吸：2026-06-08半导体−219.33亿、通信−98.35亿、消费电子−69.92亿元净流出居前[S6]，年内93家半导体公司减持[S7]，叠加CXMT约295亿IPO锚定[S9]。但估值下移是资金流/轮动驱动、非盈利驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 7, "neutral": 1, "value": 36.6, "events": [ { "title": "先进封装/HBM 产业链拥挤度风险压力测试 工作日期：2026 06 17（Asia/Singapore） 分析师：半导体分析师（semiconductors analyst） 立场…", "summary": "先进封装/HBM 产业链拥挤度风险压力测试 工作日期：2026 06 17（Asia/Singapore） 分析师：半导体分析师（semiconductors analyst） 立场：stress test（压力测试 Research note 6 的\"先进封装择时择股\"结论） 板块：A股 CoW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供给限制驱动的半导体能效转型主要表现为通用GPU的系统与精度级优化，而非ASIC/NPU的大规模替代；自研定制芯片的爆发反而加剧了全球先进制程与封装供应链对TSMC的物理集中度。", "summary": "截至2026-06-17，电力瓶颈确在倒逼算力能效转型，但并非单纯的通用GPU向专用ASIC切换。算法多变性（如MoE与动态推理）使自研ASIC面临极高的“代差失效”风险，通用GPU（如NVIDIA Blackwell）凭借软件与架构灵活性仍占据70%以上份额。能效提升主要通过系统精度（FP4/F…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI大基建进入“能效优先”阶段，二级市场资金正从粗放型算力硬件（CPO/服务器）向电力基建“瓶颈资产”与半导体先进封装“能效资产”进行结构性跨量级轮动。", "summary": "截至2026年6月17日，AI算力的电力挤出效应与能效转型正引发二级市场深刻的风格轮动。资金正从受资本挤压的传统CPO与服务器组装板块流出，转向两大确定性主线：一是具备强垄断属性与刚性Capex支撑的电力/电网设备（特高压、变压器、液冷）；二是受益于架构优化的先进封装（CoWoS、HBM）产业链。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力链可放心承接、先进封装需择时择股、CPO杠杆未出清——双主线轮动方向正确但资金弹性低于纯轮动模型预期。", "summary": "截至2026-06-17，A股微观流动性对Research note 5'电力基建+能效硬件'双主线轮动具备结构性承接能力但弹性不对称。电力设备链ETF连续7周净申购、融资买入占比84%分位、公募超配仅+1.4pct，承接顺畅；先进封装/HBM链融资已达92%分位、公募超配+5.8pct处历史峰值…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 6, "neutral": 0, "value": 49, "events": [ { "title": "AI服务器非核心BOM零部件国产替代已具备深度全球竞争力，但高速交换芯片与高端PMIC的缺位成为制约万卡集群本地化扩张的关键瓶颈。", "summary": "截至2026-06-18，对高端AI服务器BOM成本拆解显示，虽然GPU与HBM占据近80%成本导致全球供应链国产化率仅为4.64%，但非核心硬件BOM（PCB、散热、接口、电源）的全球国产价值占比达到22.50%，且在本土化服务器情景下上升至73.65%。A股龙头已在PCB与Retimer/RC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 能源重定价压的是超大规模厂商 ROIC 而非芯片 capex 效率——硅片主导 TCO、能源占比单位数，且性能/瓦溢价+杰文斯回弹反而净支撑前沿逻辑/先进封装/HBM 需求…", "summary": "2026-06-18 压力测试：能源是证伪芯片逻辑的错误杠杆。Epoch 1GW 机房年化 TCO 约 $8.5B，服务器占 ~59%、能源仅 ~7%；交付电价 +30% 仅抬升 TCO 约 +2.1%，翻倍也仅 +7%（自测算），而 IT 寿命 5→3 年使 TCO +41%——刷新节奏的影响是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FOMC 后应以“选择性算力基础设施挤空”而非“无差别 AI 融涨”来交易：多稀缺、空低 ROIC，净多但对冲。", "summary": "2026-06-18 策略综合：前序链条成立，2027 风险是 ROIC 与结构分化而非 capex 悬崖。FOMC 后应做多稀缺算力资产而非泛 AI beta：前沿逻辑/设备/HBM/先进封装、自研 ASIC 与网络、电力与电网受益方应超配；商用 GPU 龙头保留但严控风险；低 ROIC 商用 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AIDD是AI基础设施需求的高质量增量因子：规模小于通用大模型，但持续性、数据壁垒和监管/IP属性更强。", "summary": "截至2026-06-18，本卡片支持上一张卡片的方向：AIDD不会单独决定hyperscaler capex，但会把高价值、持续性的生命科学工作负载接入GPU、网络、存储、液冷和电力保障型数据中心。Recursion BioHive-2、DCAI Gefion、Novo Nordisk、Isomo…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-18", "supports": 1, "risks": 4, "neutral": 1, "value": 65.4, "events": [ { "title": "investment_focus", "summary": "HBM is identified as a key non-GPU investment focus linked to AIDD-driven AI infrastructure demand.", "sourceLabel": "链条事实", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电力/电网瓶颈真实且已在容量价格中显形,但2027年前仅推迟部分美国新增集群、不封顶全球GPU部署,Research note 01的BOM组合在2026至2027 H1仍成立,20…", "summary": "AI算力电力瓶颈真实存在但属于'延后+置换'而非'封顶':PJM 2026/27容量价已跳11倍至$329.17/MW-日,2027/28预测缺口6.6 GW,新接入等待40个月;ERCOT 2026 Q1单季198 GW申请,累计大负荷队列约410 GW;全球电网接入等待5-10年,变压器交期1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "RUM的Together AI合同仅占用其未商业化电力的约3%，而完全填满200 MW所需的47亿至51亿美元资本开支面临巨大融资缺口，且TSMC CoWoS与HBM3E供给紧缺导致…", "summary": "截至2026-06-18，对RUM Group的B300 GPU采购进行供应链与财务敏感性测试显示：其与Together AI签署的2.7亿美元B300液冷云服务协议在物理规模上仅需约1,800至3,000张GPU，对应电力功耗仅约4.8至7.3 MW，仅占其剩余200 MW未商业化容量的2.4%…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "大模型价格战引发的推理流量爆量，导致国产GPU设计龙头呈现“营收高增、毛利率承压”的量价背离，定价权向制造端瓶颈与TCO优化环节倾斜。", "summary": "在大模型极低价格战（DeepSeek-V4-Flash降至$0.14/1M tokens）推动下，云厂商Q1 AI基础设施收入大增79%（GPU云激增184%）。但因先进制程及封装产能受限导致BOM成本上涨10%-15%，加之云厂商降本压价，国产GPU（如海光一季度毛利率降至55.6%，同比下滑5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 3, "neutral": 0, "value": 61.8, "events": [ { "title": "电力与封装成本上升会触发AI基础设施技术路线调整，但截至2026-06-20仍不足以触发超大规模云厂商AI芯片采购总量的全面回撤。", "summary": "工作日2026-06-20：本卡压力测试认为，电力和先进封装成本上升不会在2026年触发超大规模云厂商全面削减AI芯片采购，但会提高单位弹性，推动技术路线从“更多GPU插槽”转向“每瓦性能、液冷机架、自研ASIC、提前锁定CoWoS/HBM和电力站点”。真正风险是并网、变电站、CoWoS或HBM配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "维持 Strategy 3 的 20% 半导体总权重是合理压力测试结论，但必须把它结构化为先进封装/HBM/AI ASIC/GPU/国产替代各 4 pp，否则宽 β 配置会让减仓换 …", "summary": "工作日 2026-06-19。本卡对 Strategy 3 中半导体权重降至 20% 的安排进行压力测试。方向上支持削减——SOX 估值已处历史 88% 分位、解禁去估值风险真实——但加三条限定：(1) 20% 仓位的内部结构远比头条数字重要，宽 SOX 等权会损失 50–80 bps 结构 α；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 5, "neutral": 0, "value": 51.2, "events": [ { "title": "在AI资本开支受制于电网接入、热管理与绿电物理直连约束下，A股AI投资应从过度拥挤的GPU/服务器链条撤出，重新向处于供给瓶颈核心的特高压变压器、数据中心液冷、储能系统及低成本西部绿…", "summary": "截至2026-06-21，AI基础设施电网接入与变压器瓶颈正引发A股配置重估。测算显示，TTP延迟使AI服务器出货量在2026-2028年产生7.5%—12.5%的缺口，在1.6倍运营杠杆下拖累利润18.6%—30.0%。相反，UHV变压器供需缺口将交货期从26周拉长至154.5周，受绕线工稀缺、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "特高压变压器测试大厅设备排班、绕线工短缺及CTC铜导线瓶颈，构成2026-2030年AI基础设施上电的绝对物理上限，将拖延国内43.7%的规划算力新增，并给科技巨头带来数千亿的闲置折…", "summary": "截至2026-06-21，本卡片对特高压变压器供应链瓶颈对AI智算集群建设的影响进行了压力测试。由于绕线熟练工短缺（培训加内训需3年以上）、出厂测试大厅排期达3年以及CTC铜导线紧缺，特高压变压器在2026-2030年产生严重供需缺口。压力测试显示，该瓶颈将拖累国内新增算力IT容量累计达8.82 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然HBM国产化良率崩塌会将集群 CapEx 推升 98%，但中国互联网巨头通过异构分级部署与算法优化（缩减30%GPU需求）并结合西部绿电直连套利（LCOE降至0.39元）与液冷改…", "summary": "评估HBM成本压力对AI集群TCO及扩容策略的传导。100%国产材料导致HBM3封装良率崩塌至4.11%，推升10000颗GPU集群CapEx至8.17亿美元，东部煤电下3年TCO达10.17亿美元（+66.13%）。建议协同优化：架构上采取60% HBM3与40% HBM2E的分级部署，结合算法…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "通过估值、拥挤度及电网资本开支敏感度因子筛选的“电网优先”A股组合（PE 17.14x，周换手6.52%）可有效避免为拥挤AI Beta（PE 32.78x，周换手27.57%）支付…", "summary": "截至2026-06-20，本报告将“电网优先”的AI基建逻辑落成A股定量筛选框架，对比筛选组合与拥挤AI Beta。筛选出的电网优先组合（中国西电、国电南瑞、思源电气等）具有低估值、低拥挤度与高电网弹性特征（PE 17.14x，融资占比2.95%，周换手6.52%）。相反，中际旭创等拥挤AI标的P…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 4, "neutral": 0, "value": 58.6, "events": [ { "title": "在2026-06-21，受2nm GAA先进制程 wafer 溢价与先进封装 80% 初期装配良率的限制，下一代高密度 AI 芯片的制造 BOM 将翻倍至8,153.59美元，其中封…", "summary": "截至2026-06-21，本卡片评估了AI机架密度120kW+下，先进封装良率及散热材料研发对AI芯片成本与供应链的重塑。下一代2nm GAA+HBM4+液冷芯片BOM成本较3nm基准大涨77.2%至6,780.97美元，在初期封装装配良率降至80%的重度压力下更暴增113.1%至8,153.59…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力带来的成熟制程需求因实物量级差距、先进封装瓶颈及工艺错配仅能对冲消费电子流失的6.28%，无法挽救晶圆厂重折旧下的利润挤压，投资应避开通用芯片并防御性转向国产替代材料与公用事…", "summary": "截至2026-06-21，量化测算显示AI算力需求的增长仅能对冲消费电子端成熟制程流失产能的6.28%，国内月度剩余成熟制程过剩缺口达176,173.9片12寸等效晶圆。受制于工艺IP错配、先进封装/HBM瓶颈锁死以及高CapEx扩产带来的刚性折旧压力，AI难以化解成熟代工链的折旧与利用率下滑危机…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "提升芯片能效与先进封装无法对冲变压器短缺造成的并网延迟损失（闲置折旧高达25亿至50亿美元），反而因功耗密度飙升将瓶颈推向机架级配电，且受限于国内封测良率瓶颈与成本压力，持有电网设备…", "summary": "截至2026-06-21，本报告对“半导体设计/封装技术迭代能对冲电力并网延迟”的论点进行压力测试。分析表明，12-24个月的变压器交付延迟对200k GPU集群将造成25.3亿至50.7亿美元的闲置折旧，重创ROIC并推升GPU租金成本达39.1%至156.7%，芯片端能效提升被财务减值彻底吞噬…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-21，电网与电力并网瓶颈带来的通电延迟（2年延迟拖累53.71亿美元[自测算]）将迫使云厂商放缓GPU采购，若发生20%的需求削减，台积电3nm/2nm制程及CoW…", "summary": "截至2026-06-21，本分析对‘电力与电网并网瓶颈’对高端AI芯片（GPU、HBM）及先进制程产能的传导影响进行了压力测试。定量测算表明，10万张Blackwell GPU集群面临2年通电延迟将产生53.71亿美元闲置财务拖累[自测算]。如果云厂商因电力限制削减20%芯片采购，台积电3nm和2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 7, "neutral": 1, "value": 45, "events": [ { "title": "国内通缩+自主可控体制是美国更久维持高位的镜像，使A股本土杠铃被双重逻辑支撑而不破；风险在于宽松与升值的节奏，而非方向。", "summary": "本卡以宏观视角压力测试Card-06的A股本土杠铃。结论：杠铃不破，但成立理由是「镜像」而非「复制」——中国处于通缩+再通胀+自主可控体制（CPI同比+1.2%、PPI同比+3.9%、LPR连续12个月维持1年3.00%/5年3.50%且有降息空间、人民币升值至约6.76–6.79），与美国更久维…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力瓶颈本地地点化和能耗惩罚电价将加速粗放GPU堆叠的价值减损，重塑A股科技风格超配“应用红利”与“电网CapEx”，并转向以“能效比”为核心的ASIC高能效硬件。", "summary": "截至2026-06-22，本卡综合研判AI电力瓶颈地点特定约束对权益风格的重塑。随着“算电协同”写入政府工作报告及发改委688号直连政策落地，东部智算面临并网与PUE差别电价硬约束（北京PUE>1.35加价0.20元）。自测算表明，不达标能耗与并网延迟将对100MW智算中心分别造成17.05亿和1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "扩大版出口管制导致材料与EDA工具国产化面临良率黑洞与算力TCO暴增，实质性切断了在华先进封装的商业化增量逻辑。", "summary": "截至2026-06-22，评估表明若BIS将管制扩大到先进材料（如ABF载板、MUF、NCF）及EDA工具，将实质切断在华先进封装产能的商业化增量逻辑。100%国产化材料导致8-Hi HBM堆叠良率崩塌至4.11%，产出缩水68.48%；GPU BOM成本翻倍至$37,746，售价涨至$75,49…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "变压器/电网瓶颈对AI芯片需求预期与库存周期的传导风险", "summary": "截至2026-06-22，本卡以半导体视角压力测试Card-03。结论：芯片需求仍明确领先于电力瓶颈，2026年不存在AI芯片砍单——约束仍是芯片供给本身（CoWoS售罄至2026年底、英伟达独占约60%；Blackwell积压约360万颗售罄至年中；2026年HBM全部订满且超大规模厂商预付扩产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 5, "neutral": 0, "value": 55.4, "events": [ { "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "summary": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下修。", "summary": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "硬件预算挤压对次世代AI芯片(如Rubin)性能需求与技术路线的传导影响", "summary": "针对“电力成本飙升→能效驱动型替代→削弱顶尖算力需求”的链条做压力测试。结论：前提成立但结论被颠倒。能效比(每瓦token→每兆瓦营收)确已成为芯片主导设计轴[S3][S10]，但在电力硬约束下，优化能效比会把需求集中到前沿节点(Rubin/HBM4/CoWoS-L)而非替代离场——兆瓦稀缺时运营…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI芯片能效和封装升级可以提高每个已送电MW的算力收入，但不能消除变压器长周期交付造成的站点送电瓶颈。", "summary": "截至2026-06-23，本卡支持前序判断：半导体厂商正在通过更高性能每瓦、低精度推理、HBM/CoWoS/3.5D封装和自研ASIC，提高每个已送电MW的AI产出。但这只是缓解，不是替代变压器。若站点无法送电，GPU租赁收入仍会延后；若站点已送电，Blackwell/GB300、MI350、Ir…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 5, "neutral": 1, "value": 59.8, "events": [ { "title": "[auto-handoff] 日元干预警报与亚太半导体/CPO板块去杠杆风险评估", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: c956eaa1-0683-41fa-9c41-551222934843 结果摘要: Wrote handoff_response.md (10735 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM/先进封装自主与出口管制反身性是否构成对中国AI建设节奏的绑定宏观约束及政策驱动A股主题", "summary": "本卡综合 Research note 07：接受瓶颈已迁至国产HBM/先进封装，并将其升级为宏观—政策命题。国产HBM确是绑定约束，但绑定的是2026–2027的“节奏”而非“轨迹”——属软绑定。2026年仅约25万–30万颗 Ascend-910C 等效国产封装[S11]，仅相当于英伟达约360…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科技板块风格再校准：能效赋能方的稀缺溢价与高耗电算力股的吞吐折价", "summary": "本卡把前五卡收束为可执行配置框架。算力以‘每兆瓦有效算力’计价后，权益市场沿‘谁拥有稀缺投入、谁的回报被它锁死’重估。折价已现于现金流：2026超大规模厂商资本开支约6900–7250亿美元、约75%投向AI，却吞噬近100%经营性现金流；微软约800亿美元Azure积压受限于电力。溢价已现于盘面…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中国CSP成本上升背景下，AI算力基础设施供应弹性与扩容压力分析", "summary": "压力测试部分否定线程以电力为中心的框架。AI算力系统是供给绑定而非需求绑定：2026年超大规模厂商资本开支同比+77%至约7250亿美元[S1]，英伟达Blackwell售罄至年中、积压360万颗[S2]，因此约束的是配额优先级而非需求。核心锚定集群凭借锁定电力、GPU配额与表后自发电被隔离；放缓…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 5, "neutral": 0, "value": 58.2, "events": [ { "title": "电力设备瓶颈不会系统性消灭AI加速器需求，但会把2026-2028年约15%-25%的GPU/HBM/ASIC部署变成收入确认和爬坡时点风险。", "summary": "截至2026-06-25，本卡从半导体侧压力测试电力设备瓶颈传导：2026-2028年加速器等效部署中约15%-25%面临2-4个季度时点风险，真实终端需求毁灭约3%-7%。商用GPU最直接受通电MW影响，HBM因多年协议和稀缺配额更像库存节奏风险，自研ASIC既受益于每瓦token优化，也因客户…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-25，AI硬件需求的下一核心约束不是芯片是否能生产，而是足够多的兆瓦能否被许可、接入、冷却并按期商业化。", "summary": "截至2026-06-25，本卡支持上一张卡的AI硬件稀缺性主线，但将瓶颈从GPU/HBM进一步上移到可通电数据中心、电网接入、冷却和电气设备。IEA、Berkeley Lab、EPRI、CBRE和JLL数据共同显示，AI需求仍强，但项目能否转化为创收产能取决于电力和并网速度。结论是：稀缺性继续支撑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 1, "neutral": 0, "value": 83.6, "events": [ { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 4, "neutral": 0, "value": 67, "events": [ { "title": "算力设备折旧周期与电力并网延迟带来的资产减值压力测试", "summary": "部分印证并收紧第08卡，但重新界定机制：先货后电在2026年既非实物提前报废，也非典型GAAP冲销，而是“幻影日历贬值→加速折旧→2027-2028通电即减值”的三段链。GAAP折旧从投入使用起算，但贬值从到货起算；按英伟达年度节奏（Hopper/Blackwell/Rubin/Rubin-Ult…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高利率平台与电力交付约束下的AI算力产业链估值重塑与多空配置策略", "summary": "我对卡片05–07主线部分支持、部分否定。约束在电力，经济租金从过剩层（资本、GPU、可出租算力）迁移到稀缺层（已通电兆瓦、发电、超长交期电网设备）——重估命题正确且不对称：稀缺层应享更高估值，被杠杆与抵押挤压的算力出租商应承受更宽折价。但干净的多空表达式更脆弱：多头腿GEV已71倍前瞻不便宜，便…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-27，AI电力并网限制（2年延误拖累53.71亿美元[S12]）将倒逼算力采购削减并重创高端半导体高增长假设，投资者应果断削减AI硬件至12%及长债至5%，转而增配…", "summary": "截至2026-06-27，本报告重新校准了在高利率平台与电力之墙挤压下的多资产配置。2年电网并网延误对10万卡Blackwell智算集群将产生53.71亿美元搁浅资金损失[S12]，迫使云巨头在2027年削减20%GPU订单。这将导致台积电3nm和2nm产能利用率分别跌至85%与80%，CoWoS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 7, "neutral": 0, "value": 50.4, "events": [ { "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在先进封装良率瓶颈与表后电力并网延迟的双重物理约束下，A股AI硬件板块估值极度偏高且存在变现拥挤，难以独立承担成长杠铃的对冲配置功能，应当实施防御性减配。", "summary": "本卡压力测试否认了AI硬件作为成长杠铃关键一环的论据。测试显示AI硬件期望收益为-3.43%，压力下变现滑点高达16.82%，且2nm GAA/HBM4封装成本激增113%抬高TCO门槛。此外，16GW计划容量中仅5GW在建，电力并网延迟拖累电网设备行业盈利收缩达36.8%至60.2%。建议将AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面向AI基础设施信用尾部的跨资产组合配置与对冲", "summary": "截至2026-06-28，本卡综合前序结论：AI算力长期上行仍可持有，但尾部风险在信用结构而非单纯科技股估值。建议采取“优质AI+电力瓶颈+信用保护”杠铃：保留NVIDIA与现金流型超大规模云厂商核心仓位，低配Neocloud股票和GPU抵押DDTL/私募信贷，增配受监管公用事业、电网设备、稳定电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产大模型并未告别规模竞赛，而是在先进GPU受限的条件下把竞争单位改写为每瓦、每token、每受限加速器和每留存用户的有效智能产出。", "summary": "截至2026-06-28，本卡对“国产大模型不可逆转向效能极限竞赛”进行压力测试。结论是：效能迁移真实且受GPU出口、H20许可、HBM/互连、液冷机柜和电力约束推动，但不是规模竞赛终结，而是进入“稀缺规模竞赛”。DeepSeek-V3、DeepSeek-R1、Qwen3和MiniMax-M1显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 4, "neutral": 1, "value": 72.8, "events": [ { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "九卡线索数值连贯、引用完整、叙事贯通，分析扎实无需重做；仅一项 P0（Cards 01/02 缺 2026-06-30 工作日期锚）卡住发布，评级 NEEDS-REVISION（轻度…", "summary": "对 Cards 01–09 做发布前 QA，评级 NEEDS-REVISION（轻度），距 PASS 仅一个修订周期。数值主干一致：HBM 良率 4.11–16.98%→GPU 23,125–95,537 颗→23.12–95.54 EFLOPS 在 Card04↔05 完全吻合；5.5% 红利为…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-56…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-029ce03da1/", "networkHref": "/investor/research-network/#thesis-029ce03da1" }, { "id": "thesis-8c5a9690a0", "source": "tracked_thesis", "title": "Mag7 在 AI 时代从平台垄断转向基础设施再分层", "titleEn": "Mag7 shifts from platform monopoly to AI infrastructure re-layering", "lane": "Mag7 AI 平台再分层", "laneEn": "Mag7 AI platform re-layering", "laneHref": "/investor/lanes/mag7-ai-platform-re-layering/", "investableFocus": "Mag7 AI 平台再分层：MSFT, AAPL, TSLA, META, AMZN, GOOGL, NVDA", "stockUniverse": [ { "ticker": "MSFT", "name": "Microsoft", "market": "US", "role": "旧软件入口与云 AI", "roleEn": "旧软件入口与云 AI", "bucket": "core", "href": "/investor/stocks/MSFT/" }, { "ticker": "AAPL", "name": "Apple", "market": "US", "role": "端侧 AI 与消费硬件", "roleEn": "端侧 AI 与消费硬件", "bucket": "core", "href": "/investor/stocks/AAPL/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "自动驾驶/机器人与高波动叙事", "roleEn": "自动驾驶/机器人与高波动叙事", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告平台", "roleEn": "开源模型与广告平台", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "AMZN", "name": "Amazon", "market": "US", "role": "AWS/自研芯片/零售自动化", "roleEn": "AWS/自研芯片/零售自动化", "bucket": "core", "href": "/investor/stocks/AMZN/" }, { "ticker": "GOOGL", "name": "Alphabet", "market": "US", "role": "模型、搜索、云与 TPU", "roleEn": "模型、搜索、云与 TPU", "bucket": "core", "href": "/investor/stocks/GOOGL/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" } ], "stockTickers": [ "MSFT", "AAPL", "TSLA", "META", "AMZN", "GOOGL", "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ" ], "coreView": "赛道：Mag7 AI 平台再分层。候选观察标的：MSFT, AAPL, TSLA, META, AMZN, GOOGL, NVDA。当前共有 40 条证据、5 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 88, "eventCount": 160, "firstSeen": "2026-04-25", "lastSeen": "2026-06-28", "supports": 21, "risks": 120, "neutral": 19, "analystCount": 15, "sourceTypeCounts": { "研究线程": 6, "分析协作": 3, "研究笔记": 123, "证据更新": 26, "深度研究": 1, "研究主题": 1 }, "topTerms": [ "AI 基础设施", "Microsoft", "NVIDIA", "Amazon", "Apple", "Meta", "Tesla", "HBM", "GPU", "ASIC", "Mag7", "shifts" ], "whatChanged": [ "2026-06-28: 电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "2026-06-28: 数据中心电网并网瓶颈与核能转向", "2026-06-28: AI交易不会结束，但估值核心应从买到多少芯片转向能否按期通电、保持利用率并降低每token能耗。", "2026-06-28: AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "2026-06-28: AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。" ], "practical": { "sectorLane": "Mag7 AI 平台再分层", "sectorLaneEn": "Mag7 AI platform re-layering", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "MSFT", "name": "Microsoft", "market": "US", "role": "旧软件入口与云 AI", "roleEn": "旧软件入口与云 AI", "bucket": "core", "href": "/investor/stocks/MSFT/" }, { "ticker": "AAPL", "name": "Apple", "market": "US", "role": "端侧 AI 与消费硬件", "roleEn": "端侧 AI 与消费硬件", "bucket": "core", "href": "/investor/stocks/AAPL/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "自动驾驶/机器人与高波动叙事", "roleEn": "自动驾驶/机器人与高波动叙事", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告平台", "roleEn": "开源模型与广告平台", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "AMZN", "name": "Amazon", "market": "US", "role": "AWS/自研芯片/零售自动化", "roleEn": "AWS/自研芯片/零售自动化", "bucket": "core", "href": "/investor/stocks/AMZN/" }, { "ticker": "GOOGL", "name": "Alphabet", "market": "US", "role": "模型、搜索、云与 TPU", "roleEn": "模型、搜索、云与 TPU", "bucket": "core", "href": "/investor/stocks/GOOGL/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" } ], "stockTickers": [ "MSFT", "AAPL", "TSLA", "META", "AMZN", "GOOGL", "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ" ], "primaryStocks": [ { "ticker": "MSFT", "name": "Microsoft", "market": "US", "role": "旧软件入口与云 AI", "roleEn": "旧软件入口与云 AI", "bucket": "core", "href": "/investor/stocks/MSFT/" }, { "ticker": "AAPL", "name": "Apple", "market": "US", "role": "端侧 AI 与消费硬件", "roleEn": "端侧 AI 与消费硬件", "bucket": "core", "href": "/investor/stocks/AAPL/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "自动驾驶/机器人与高波动叙事", "roleEn": "自动驾驶/机器人与高波动叙事", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告平台", "roleEn": "开源模型与广告平台", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "AMZN", "name": "Amazon", "market": "US", "role": "AWS/自研芯片/零售自动化", "roleEn": "AWS/自研芯片/零售自动化", "bucket": "core", "href": "/investor/stocks/AMZN/" }, { "ticker": "GOOGL", "name": "Alphabet", "market": "US", "role": "模型、搜索、云与 TPU", "roleEn": "模型、搜索、云与 TPU", "bucket": "core", "href": "/investor/stocks/GOOGL/" } ], "instrumentTags": [ "MSFT", "Microsoft", "AAPL", "Apple", "TSLA", "Tesla", "META", "Meta Platforms", "AMZN", "Amazon", "GOOGL", "Alphabet" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 6 类，近期事件 87 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 6, "analystCount": 15, "recentEvents": 87, "supportRatio": 0.13, "riskRatio": 0.75, "conflictScore": 83.1, "evidenceDepth": 100 }, "score": 84.4 }, "events": [ { "date": "2026-06-28", "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计约6900-7200亿美元（自测算），NVIDIA FY20…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 39.25, "matchedTerms": [ "Microsoft", "NVIDIA", "Amazon", "Meta", "HBM", "infrastructure", "META" ], "href": "" }, { "date": "2026-06-28", "title": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "summary": "截至2026-06-28，本卡否定“AI资本开支韧性等于A股防御性配置锚”的宽泛结论。Microsoft、Alphabet、Meta、NVIDIA、TSMC等数据仍支持AI基础设施真实需求，但FOMC 3.50%-3.75%、5月PCE同比4.1%、10年美债4.40%和A股两融突破RMB 3tn，意味着长久期成长股、拥挤主题和融资敏感供应商仍可能估值下修…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.25, "matchedTerms": [ "Microsoft", "NVIDIA", "Meta", "HBM", "META" ], "href": "" }, { "date": "2026-06-28", "title": "AI交易不会结束，但估值核心应从买到多少芯片转向能否按期通电、保持利用率并降低每token能耗。", "summary": "截至2026-06-28，本卡综合支持前序判断：2026-2030年AI扩张的约束从服务器交付转向可通电MW交付。Amazon、Microsoft、Alphabet、Meta最新季度CAPEX/PPE购买合计约$130.8B[自测算]，但IEA预计美国数据中心用电到2030年较2024年增加约240TWh，约等于27.4GW连续负荷[自测算]。结论是Hyp…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.25, "matchedTerms": [ "Microsoft", "Amazon", "Meta", "META" ], "href": "" }, { "date": "2026-06-28", "title": "数据中心电网并网瓶颈与核能转向", "summary": "截至2026年6月28日，AI基础设施建设的核心瓶颈已从GPU稀缺转移到电网并网时间（Time-to-Power）。对100 MW超大规模AI数据中心的模型测算显示，2年的并网延迟会减少项目35.73%的净现值（NPV），并使IRR大幅下降。为规避此类延迟，科技巨头（微软、亚马逊、谷歌、Meta）在2024至2026年间锁定了超十吉瓦（GW）的核电容量，通…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 23.25, "matchedTerms": [ "Meta", "GPU", "infrastructure", "META", "power" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "summary": "截至 2026-06-28，本卡压力测试认为：电力交付瓶颈会迫使 hyperscaler 从单纯追求新增 MW 转向 MW 生产率和存量优化，但不会终结扩产。Microsoft、Alphabet、Meta 可见 2026 年 capex 区间合计约 4,950-5,250 亿美元（自测算），同时仍披露容量约束；IEA 显示数据中心 1-3 年可建成，而新电…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 23.25, "matchedTerms": [ "Microsoft", "Meta", "infrastructure", "META" ], "href": "" }, { "date": "2026-06-27", "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期：…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 / Executive Summary 截至 2026 年 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.75, "matchedTerms": [ "AI 基础设施", "Microsoft", "Amazon", "Meta", "infrastructure", "META", "GOOGL" ], "href": "" }, { "date": "2026-06-27", "title": "变压器材料和交付瓶颈已开始改变 Hyperscaler 的2026-2027资本开支与电力采购节奏，影响主要体现为预算通胀、长期锁电和通电交付递延，而非AI需求取消。", "summary": "截至2026-06-28，结论为支撑：Hyperscalers 已经通过预算上修、长期电力合同和项目节奏调整来应对变压器及电气设备瓶颈，但公开披露通常不会把变化直接归因为 GOES 或铜关税。Microsoft 2026 capex 中约250亿美元来自更高组件价格，Meta 上调2026 capex中点100亿美元，Amazon 把电力容量作为部署KPI…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28.75, "matchedTerms": [ "Microsoft", "Amazon", "Meta", "infrastructure", "META" ], "href": "" }, { "date": "2026-06-27", "title": "电子链条利润反弹主要受AI基础设施投资拉动，具备向存储、设备、材料、先进封装、光互联和数据中心电源散热扩散的潜力，但尚不足以证明消费电子和内需全面复苏。", "summary": "截至2026-06-28，本卡支持上一卡判断：电子链条利润修复主要由AI基础设施、算力芯片和存储涨价驱动，而非内需广谱复苏。国家统计局数据显示，1—5月电子行业利润同比增长103.9%，贡献全部工业利润增量43.1%。NVIDIA、Microsoft、Alphabet、Meta、Alibaba和Tencent的披露共同验证数据中心硬件需求仍强；TrendF…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.25, "matchedTerms": [ "Microsoft", "NVIDIA", "Meta", "META" ], "href": "" }, { "date": "2026-06-26", "title": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "summary": "截至2026-06-27，本卡片对AI基础设施链做压力测试：全球AI资本开支仍大，Goldman Sachs基准模型给出2026年USD 765 billion、2031年USD 1.6 trillion的年度AI capex，但Amazon、Alphabet、Meta、Microsoft的数据表明现金流、折旧、能源和电网接入正成为估值核心约束。A股不应全…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28.75, "matchedTerms": [ "Microsoft", "Amazon", "Meta", "infrastructure", "META" ], "href": "" }, { "date": "2026-06-26", "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–560 亿美元；中国云(字节/阿里/腾讯)与国产替代构成第二引…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.75, "matchedTerms": [ "AI 基础设施", "NVIDIA", "Amazon", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实触发重排，但方向是向支出倾斜，且买方自身点名的硬约束是电力与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 38.75, "matchedTerms": [ "Microsoft", "Amazon", "Meta", "HBM", "ASIC", "infrastructure", "META" ], "href": "" }, { "date": "2026-06-25", "title": "CNH套息去杠杆会先打掉A股AI基础设施股票的流动性溢价和估值久期，而不是立即打断全球AI CapEx实物周期。", "summary": "截至2026-06-25，本卡片压力测试CNH套息去杠杆对AI基础设施估值和CapEx预期的冲击。结论是：流动性溢价消失可使A股AI硬件/算力基础设施估值出现非线性回撤，但不应直接推导为全球hyperscaler CapEx崩塌。Microsoft、Alphabet、Meta、Amazon、Oracle和NVIDIA披露的2026年AI基础设施投入、订单积…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34.25, "matchedTerms": [ "Microsoft", "NVIDIA", "Amazon", "Meta", "infrastructure", "META" ], "href": "" }, { "date": "2026-06-25", "title": "评估AI基础设施与半导体产业链在“更高更久”利率环境下的现金流质量与估值韧性", "summary": "压力测试Research note 2后结论：不能把AI基础设施当成单一“质量成长防御”桶——产业链已分化。算力卖方（NVDA型）财年2026营收$215.9B、经营现金流$102.7B、FCF$96.6B、FCF利润率约44.7%，当期造血、净现金，确属短久期质量成长。算力买方（MSFT/GOOGL/AMZN/META）2026年合计资本开支约$725B…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "Amazon", "Meta", "infrastructure", "MSFT", "META", "AMZN", "GOOGL", "NVDA" ], "href": "" }, { "date": "2026-06-24", "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Microsoft、Alphabet、Meta 的 capex …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 39.5, "matchedTerms": [ "AI 基础设施", "Microsoft", "Amazon", "Meta", "GPU", "infrastructure", "META" ], "href": "" }, { "date": "2026-06-24", "title": "AI硬件交付节奏与光模块订单验证", "summary": "截至2026-06-24，本卡支持“CPO延期不等于800G/1.6T砍单”的基本面判断，但加入节奏约束：Microsoft、Alphabet、Meta、Amazon的2026 capex与NVIDIA Rubin、Dell AI服务器订单、Lumentum光组件指引共同显示需求仍强，800G/1.6T更像供给受限放量而非订单悬崖。短线风险在于中际旭创与新…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34.25, "matchedTerms": [ "Microsoft", "NVIDIA", "Amazon", "Meta", "infrastructure", "META" ], "href": "" }, { "date": "2026-06-24", "title": "半导体杠杆清盘对Hyperscaler AI资本开支的近端传导主要是价格、排序和边际买方融资纪律，而不是一线供应链信用崩塌或Mega-cap资本开支急刹车。", "summary": "截至2026-06-24，本卡压力测试认为，韩国半导体杠杆ETF清盘会重定价AI硬件股票并强化资本市场审视，但不构成Mega-cap Hyperscaler近端AI资本开支削减的主传导。Microsoft、Alphabet、Meta、Amazon仍以强经营现金流和投资级融资支撑AI建设；NVIDIA、TSMC、Micron、SK Hynix等关键供应商也未…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34.25, "matchedTerms": [ "Microsoft", "NVIDIA", "Amazon", "Meta", "infrastructure", "META" ], "href": "" } ], "timeline": [ { "date": "2026-04-25", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "2026年云巨头AI基础设施资本开支跃升", "summary": "2026年云巨头AI基础设施资本开支跃升 Alphabet、Amazon、Microsoft、Meta等云巨头的2026年AI基建支出，是已被云与AI需求验证的扩产周期，还是尚未兑现回报的资本开支风险？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-26", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "北美四大云厂商Q1 2026业绩前瞻——AI capex指引对A股算力链（光模块/PCB/液冷）的传导与赔率", "summary": "北美四大云厂商Q1 2026业绩前瞻——AI capex指引对A股算力链（光模块/PCB/液冷）的传导与赔率 在4月28日—5月1日Meta/Microsoft/Alphabet/Amazon集中披露Q1 2026业绩的窗口，capex指引会向上修正、维持还是被下修？这对已经YTD大涨48.5%的…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-27", "supports": 2, "risks": 1, "neutral": 0, "value": 54.8, "events": [ { "title": "[auto-handoff] AI 大厂 capex 指引边际曲率", "summary": "致 日报总编： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 4905c420-d2ab-4c97-97c1-99465e0023d6 结果摘要: 为了回答海外超大规模厂商（Hyperscalers）一季报中 Capex 指引的边际变化，我重点分析了本周（以 2024 …", "sourceLabel": "分析协作", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2026 Q1之后超大规模云厂商AI资本开支可持续性", "summary": "2026 Q1之后超大规模云厂商AI资本开支可持续性 Q1 2026财报季后，Amazon/Alphabet/Meta/Microsoft合计AI capex指引上修至约6,400亿美元（同比+36%，AI占比约75%），但Alphabet与Meta的2026年FCF被预计同比缩水近90%，财报后…", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2026年四大超大规模云厂商AI资本开支约7,000亿美元的可持续性,以及电力瓶颈对Capex落地节奏的真实约束", "summary": "2026年四大超大规模云厂商AI资本开支约7,000亿美元的可持续性,以及电力瓶颈对Capex落地节奏的真实约束 在自由现金流被显著挤压、电网与变压器排队进入'以年计'的当下,本周(4月24-30日)Microsoft/Meta/Alphabet/Amazon的FY26一季报会兑现还是修正这条7,…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-29", "supports": 0, "risks": 0, "neutral": 1, "value": 49.2, "events": [ { "title": "2026年一季度超大规模云厂商AI资本开支轨迹与GPU/ASIC供应链含义", "summary": "2026年一季度超大规模云厂商AI资本开支轨迹与GPU/ASIC供应链含义 在Alphabet、Microsoft、Meta、Amazon集中披露2026年一季报的窗口,云厂商AI基础设施资本开支是否仍维持50%+同比增速、2026年全年指引是上调还是下调,以及边际美元在GPU、定制ASIC和电力…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "Q1 2026 财报后的云巨头 AI capex：需求验证还是瓶颈预警？", "summary": "本卡选择 tmt-analyst 评估 Q1 2026 云巨头 AI capex 周期。Microsoft、Alphabet、Amazon 和 Meta 均显示 AI/云收入加速，但 capex 和 PP&E 支出同步大幅上行。TSMC 的 61% HPC 收入占比、74% 先进制程占比和 66.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力仍更像牛市中段而非sell-on-news，但最佳实现方式是风险预算型跨资产杠铃，而不是无保护的AI权益集中仓位。", "summary": "作为 asset-allocator，我对前四张卡的“AI算力牛市中段”结论做跨资产压力测试。结论是：主线仍成立，但组合表达不能是集中追涨AI权益，而应是风险预算型杠铃。NVIDIA、Microsoft、Alphabet 和 Amazon 的最新披露继续支持企业推理和AI资本开支周期，但Fed维持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-03", "supports": 1, "risks": 1, "neutral": 0, "value": 51, "events": [ { "title": "2026年4月 hyperscaler AI capex 重估", "summary": "本卡由 tmt-analyst 作为开题分析师完成。结论是：2026年 hyperscaler AI capex 仍是需求驱动周期，而非简单泡沫破裂；Microsoft、Alphabet、Amazon 和 Meta 的 2026 capex 大致落在 $695-725 billion 区间，且由 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Alphabet/Microsoft/Amazon AI CapEx ROI兑现时间表的板块层面评估", "summary": "致 深研主编： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: e4cd9656-368f-4c42-a950-c9b9ec8f192b 结果摘要: Wrote handoff_response.md (10,104 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-04", "supports": 1, "risks": 7, "neutral": 1, "value": 19.4, "events": [ { "title": "超大规模资本开支天花板不是硬美元上限，而是融资与能源上限；中东资本抬高天花板，但也把AI基础设施推入出口管制和联盟管理的核心。", "summary": "本卡片综合而非否定卡片01：AI从暴力训练转向架构效率与推理，并未终结资本开支周期，而是把约束从GPU购买力推向现金流、项目融资、电力、液冷和出口管制。2026年主要平台AI基础设施资本开支约处于$700 billion量级，Amazon、Alphabet、Microsoft、Meta仍能投入，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2028年的清洁算力溢价首先归属于现有核电和稀缺输电权所有者，其次归属于拥有可交付清洁坚实PPA的Meta与Microsoft，而不是泛化云厂商篮子或2030年后的SMR叙…", "summary": "本卡对第03卡做压力测试：清洁算力溢价真实存在，但只属于已锁定可交付清洁坚实MW的公司。2026-2028年，Meta凭2027年起1,121 MW Clinton核电和150 MW地热居前；Microsoft凭2028年约835-837 MW TMI/Crane核电在PJM具备高质量但高成本对冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SMR 是必要但不充分,2025–2028 的边际电力来自燃气、LWR 重启与现役核电功率提升,真正的瓶颈是 NRC 审批通量、HALEU 与 FERC 监管,而不是反应堆设计。", "summary": "宏观开场:科技巨头的核电 PPA(Microsoft–TMI、Amazon–Talen、Google–Kairos、Amazon–X-Energy、Meta RFP)是写在 2028–2035 供给曲线上的远期期权,而非已落地的解决方案。需求端,IEA/EPRI 给出 2030 年全球数据中心用电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "hyperscaler AI 负荷以 firm 为主，non-firm 灵活性窗口只有 15-25%，PJM 灵活电价带来的节省被高估约 4-6 倍，真正的投资倾斜应回到 firm …", "summary": "对 Card-01 援引的 Johns Hopkins $15-16 B 节省进行压力测试。拆解 hyperscaler AI 负荷构成：在线 inference 占 45-55% 且持续上升、基础模型训练 25-35%、fine-tuning/RLHF/eval 5-10%、内部批处理 10-1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 2, "neutral": 6, "value": 57.8, "events": [ { "title": "2026年谷歌、微软、亚马逊、Meta四大超大规模云商合计承诺约7250亿美元AI基础设施投资，较2025年增长约71%。", "summary": "2026年谷歌、微软、亚马逊、Meta四大超大规模云商合计承诺约7250亿美元AI基础设施投资，较2025年增长约71%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年AI芯片竞争格局：NVIDIA Rubin对阵云巨头ASIC", "summary": "2026年AI芯片市场进入“混合计算”时代。NVIDIA推出3nm Rubin架构并引入NVLink Fusion以整合第三方芯片。与此同时，Google、AWS和Meta通过自研3nm ASIC显著降低了推理成本。AMD则凭借更大的HBM容量在长文本推理模型中占据优势。市场焦点已从单纯的算力（F…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026年四大超大规模厂商（微软、亚马逊、谷歌、Meta）预计在AI基础设施上投入7000亿美元", "summary": "2026年四大超大规模厂商（微软、亚马逊、谷歌、Meta）预计在AI基础设施上投入7000亿美元", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2025年四大超大规模云厂商（MSFT、META、GOOGL、AMZN）合计capex约3000亿美元，2026财年总盘子约3600-3900亿美元，同比增长20%-30%", "summary": "2025年四大超大规模云厂商（MSFT、META、GOOGL、AMZN）合计capex约3000亿美元，2026财年总盘子约3600-3900亿美元，同比增长20%-30%", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 1, "risks": 4, "neutral": 2, "value": 43.2, "events": [ { "title": "截至 2026 年初，微软、Alphabet、Meta、亚马逊均已维持或上调 2026 年 AI 基础设施资本开支指引", "summary": "截至 2026 年初，微软、Alphabet、Meta、亚马逊均已维持或上调 2026 年 AI 基础设施资本开支指引", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年Q1四大美超大规模云厂商（MSFT/GOOGL/META/AMZN）合计capex指引约6450-7250亿美元。", "summary": "2026年Q1四大美超大规模云厂商（MSFT/GOOGL/META/AMZN）合计capex指引约6450-7250亿美元。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年美股四大超大规模厂商(MSFT、META、GOOGL、AMZN)合计AI资本开支预计为3800-4300亿美元，同比增长约25-30%", "summary": "2026年美股四大超大规模厂商(MSFT、META、GOOGL、AMZN)合计AI资本开支预计为3800-4300亿美元，同比增长约25-30%", "sourceLabel": "证据更新", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2025日历年亚马逊、微软、谷歌、Meta合计资本支出承诺超3250亿美元", "summary": "2025日历年亚马逊、微软、谷歌、Meta合计资本支出承诺超3250亿美元", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "2026年初，微软、Meta、谷歌、亚马逊四家超大规模厂商的合计年化资本开支超过3000亿美元。", "summary": "2026年初，微软、Meta、谷歌、亚马逊四家超大规模厂商的合计年化资本开支超过3000亿美元。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-09", "supports": 0, "risks": 1, "neutral": 1, "value": 53, "events": [ { "title": "四家厂商2026年Q1业绩复核指引日期：MSFT（2026-04-29）、META（2026-04-30）、GOOGL（2026-04-24）、AMZN（2026-05-01）", "summary": "四家厂商2026年Q1业绩复核指引日期：MSFT（2026-04-29）、META（2026-04-30）、GOOGL（2026-04-24）、AMZN（2026-05-01）", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "四大超大规模云厂商合计 FCF/OCF 连续两个季度低于 10%。", "summary": "四大超大规模云厂商合计 FCF/OCF 连续两个季度低于 10%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-11", "supports": 0, "risks": 0, "neutral": 1, "value": 60.4, "events": [ { "title": "NVDA、MSFT和AAPL合计占US Quality (QUAL) 因子篮子权重的19%。", "summary": "NVDA、MSFT和AAPL合计占US Quality (QUAL) 因子篮子权重的19%。", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 3, "neutral": 0, "value": 40.8, "events": [ { "title": "AI CapEx的主线不是收缩，而是从新增外壳数量转向每瓦、每机架、每封装的半导体密度提升。", "summary": "截至2026-05-13，本卡支持卡片01的“密度胜过足迹”判断：数据中心外壳新开工-19% QoQ并不等于AI CapEx下行，而是资本更多流向GPU/ASIC、HBM、CoWoS/先进封装、机架级互连、光DSP和液冷配套。Microsoft、Meta、Alphabet、Amazon的最新披露均…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Capex 扩张不会否定量子与空间的独立订单验证，但会通过机构风险预算挤出显著提高其波动率，尤其是量子硬件。", "summary": "截至 2026-05-13，本卡支持上一卡结论：量子与空间在订单/收入验证上部分独立，但在机构风险预算中仍是 AI 邻近仓位。Microsoft 约 $190B、Alphabet Q1 PPE $35.674B、Meta 2026 Capex $125B-$145B、Amazon TTM PPE …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力不是已经断裂的盈利支撑腿，但除NVIDIA/TSMC/HBM等瓶颈外，市场已过度定价利润上修的广度与持久性。", "summary": "截至2026-05-13，AI盈利不及触发器真实存在，但尚未由最新财报证伪为红灯。MSFT/GOOGL/META/AMZN仍在上修或维持高强度2026年AI资本开支，NVIDIA、TSMC和HBM/DRAM供应商毛利仍具瓶颈定价权；但二线加速器、CPU/DCU和国产算力链已出现毛利下滑或经营杠杆不…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 8, "neutral": 0, "value": 12.2, "events": [ { "title": "四家美股超大规模厂商（MSFT、GOOGL、META、AMZN）2026年合计资本开支指引约$355–410 bn，中值约$380 bn", "summary": "四家美股超大规模厂商（MSFT、GOOGL、META、AMZN）2026年合计资本开支指引约$355–410 bn，中值约$380 bn", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "四家美股超大规模厂商2025年实际合计资本开支约$327 bn", "summary": "四家美股超大规模厂商2025年实际合计资本开支约$327 bn", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源获取难度要求对AI基础设施进行选择性下修：终局需求不变，但IDC和算力租赁的CAPEX转化曲线、利用率和ROIC应显著打折。", "summary": "截至2026-05-15，本卡对电力瓶颈对AI基础设施的影响进行TMT压力测试。结论是：不应全面下修AI算力长期需求，但必须下修CAPEX转化节奏、利用率爬坡和ROIC。IEA预计数据中心用电从2024年415 TWh升至2030年约945 TWh，同时约20%规划项目存在电网延期风险；Micro…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "微软、谷歌、亚马逊和Meta四大超级云商2026年的资本开支目标合计超过3000亿美元。", "summary": "微软、谷歌、亚马逊和Meta四大超级云商2026年的资本开支目标合计超过3000亿美元。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 1, "risks": 1, "neutral": 0, "value": 63.6, "events": [ { "title": "推理侧ASIC的规模化替代已导致通用GPU定价权削弱，NVIDIA面临物料成本上升与推理市场份额流失的双重压力，2026Q3毛利率下修逻辑成立。", "summary": "调研显示，2026年5月主要云服务商（CSP）的自研推理ASIC已进入3nm成熟期。Google TPU v6部署量超160万颗，Meta MTIA v3预计年底覆盖35%推理机群，微软Maia 200已支撑GPT-5.2。ASIC相比通用GPU可降低30%-50%的TCO，这导致推理定价权从NV…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI需求是真实支撑，但不能完全抵消未来四周TMT/机器人解禁与拥挤交易的局部供给冲击，减仓约三分之一是合理的战术对冲。", "summary": "截至2026-05-15，本卡支持卡片02对高弹性TMT/机器人龙头提前减仓约三分之一，但理由是供给冲击和交易拥挤，而非AI基本面转弱。全球半导体、ASML、TSMC、NVIDIA、Microsoft、Amazon、Meta以及中芯国际、寒武纪数据均显示AI需求仍强；但未来四周约4120亿元解禁、…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-16", "supports": 2, "risks": 0, "neutral": 0, "value": 79, "events": [ { "title": "超大规模云厂商 AI Capex 要到 2027 下半年至 2028 上半年才会出现干净的 FCF 拐点,前提是 Capex 减速、推理收入超越训练、GPU 使用寿命稳定在 5–6 …", "summary": "超大规模云厂商 (MSFT/GOOG/META/AMZN) 当前处于 AI Capex 周期阶段 1:收入加速、FCF 受压。FY25 合计 Capex 约 3400 亿美元 (+55% YoY),FY26E 约 4200–4500 亿;Q1 FY26 合计 FCF 同比下滑约 12%。基础情形下…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "截至2026-05-16，AI capex周期尚未结束，但二阶动能与估值赔率已转弱，配置上应提前部分减仓而非等待确认。", "summary": "截至2026-05-16，本卡支持在NVDA 2026-05-28财报与6月超大规模厂capex指引前先部分减仓。公开capex仍强：NVIDIA指引Q1 FY2027收入780亿美元上下2%，Microsoft预计2026年capex约1900亿美元，Meta上调至1250亿-1450亿美元。但…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 52.4, "events": [ { "title": "截至2026-05-19，AI半导体供应链尚未出现云CapEx或HBM/CoWoS订单环比下修，算力基础设施交易仍应保留核心仓位但需严密跟踪ROIC与Q2/Q3指引风险。", "summary": "截至2026-05-19，首席策略师提出的敏感触发点有效，但尚未触发。Microsoft、Alphabet、Amazon、Meta的2026年AI基础设施CapEx仍在上修或维持高位，TrendForce估计前八大CSP 2026年CapEx超过7100亿美元、同比约+61%。HBM方面，Micr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027-2028年电力缺口不会终结AI芯片周期，但会把下一代AI硬件迭代逻辑从单芯片峰值性能改写为受限MW下的有效AI输出、上线时间和机架级能效。", "summary": "截至2026-05-18，本卡支持前序电力瓶颈主线，但压力测试“硬件从极致性能转向能效”的表述：更准确地说，采购函数会转向每已通电MW的有效AI输出。2027-2028年表后燃气仅4-6GW，无法覆盖AI负荷缺口；DOE/LBNL预计美国数据中心用电2028年达325-580TWh。NVIDIA …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 59.8, "events": [ { "title": "截至 2026-05-19，AI 基础设施仍是 TMT 领涨主线，但可投资范围应收窄到有真实 usage、backlog、产能瓶颈和现金流承受力的环节。", "summary": "工作日期 2026-05-19。本文否定前序空白 TMT 论点并选择 (A)：AI 基础设施仍是领涨主线。证据是 hyperscaler capex/收入比虽已升至约 30% 的四家公司 Q1 合计强度，且 Alphabet、Amazon 的 FCF 张力明显，但 Microsoft AI ARR…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "2026年的风险是估值-流动性收缩，而不是核心AI硬件龙头经营现金流崩塌；融资陷阱主要咬住高杠杆、重营运资本和担保品融资依赖者。", "summary": "截至2026-05-21，本卡压力测试Research note 08后认为：券商折算率收缩与保险“波动率税”会压低AI硬件估值倍数和担保品融资能力，但不会导致现金流充裕的核心AI硬件龙头在2026年出现系统性崩塌。NVIDIA、AMD、TSMC、Microsoft、Meta的披露显示，现金流与客…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "AI capex 叙事已经从单纯需求验证进入 FCF 约束阶段，但在利用率、backlog、经营现金流和供应商订单同时恶化前，不应把周五语料切换当作结构性减仓信号。", "summary": "截至2026-05-23，本卡认为盘后语料从 capex 上修利多切到 FCF 受压利空，更像获利了结后的叙事消化，而非 AI 多头第一次结构性反转。情景权重为55%获利了结加叙事消化、35%真实估值分母压缩、10%结构性 capex 退潮。Microsoft、Alphabet、Meta、Amaz…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 1, "risks": 1, "neutral": 0, "value": 72, "events": [ { "title": "截至2026-05-24，AI硬件与半导体设备是A股结构性成长修复最可信的核心驱动力，但只能买订单和盈利修复，不能买泛科技估值扩张。", "summary": "截至2026-05-24，本卡支持ashare-strategist的结构成长修复判断，但将可投资主线收窄为有订单验证的AI硬件与半导体设备。NVIDIA、Microsoft、Meta、Amazon、TSMC、ASML、SEMI和SIA数据共同显示，全球AI capex、先进制程、先进封装、测试与…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "截至2026-05-24，AI硬件应维持核心超配，但需把建设进度相关风险从主仓位中剥离，并削减10-15%的战术超配。", "summary": "截至2026-05-24，本卡将chief-strategist的30-40%减仓建议与tmt-analyst的维持超配建议综合处理：内部-6.4%周度建设热度信号未能公开核验，应作为战术节奏风险，而非战略需求反转证据。公开证据显示NVIDIA、TSMC、SK hynix、Microsoft、Al…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 0, "value": 60.8, "events": [ { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek V4 标志着原始推理 token 化工品化，而不是 AI 全产业零毛利化；应把它理解为利润池从模型 API 向算力底座和工作流应用迁移。", "summary": "截至2026-05-26，本卡支持并细化既有判断：DeepSeek V4 的低价把原始模型 API token 推向重资产、低定价权、类似大宗化工品的层级，但并不意味着整个 AI 行业进入零毛利。利润正在从可替代模型调用迁移到上游芯片/网络/电力约束的基础设施，以及下游企业数据、治理和工作流集成。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 0, "neutral": 1, "value": 77.2, "events": [ { "title": "AI 产业链利润正在按 firm MW、设备订单簿、电源效率 IP 和大用户 tariff 政治可防御性重新分配；2026–2027 年最强留存在电气设备 OEM、签 PPA 的 f…", "summary": "本卡综合 01–06 号卡，把 AI 产业链利润定价从单层'算力溢价'切换为按 firm MW、电气设备订单簿、电源效率 IP 和大用户 tariff 政治可防御性配给的四层结构。2026–2027 年利润留存最强的是 GEV/ETN/VRT 等订单簿已锁的电气设备 OEM、CEG/VST/TLN…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "2026年AI芯片与先进封装产能约束会推高并拉长云端CapEx，而不是使其取消；配置应偏向具备配额权的上游瓶颈环节。", "summary": "截至2026-05-28，本卡支持前序判断：AI硬件CapEx传导链的约束真实但不等于削减。Microsoft、Amazon、Alphabet、Meta的2026年CapEx指引中枢约为$710 billion，反映的是芯片、HBM、先进封装、电力和数据中心资源稀缺导致的锁单与涨价，而非需求走弱。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 1, "risks": 0, "neutral": 0, "value": 85, "events": [ { "title": "硬件成本红利与高CapEx会通过更低cost-per-token、更稳定算力供给和更可预测API价格改写AI应用单位经济模型，但真正的投资筛选变量是单任务成本改善乘以分发杠杆。", "summary": "截至2026-05-29，本卡支持前序判断：制造端成本红利会向下游传导，但更关键的不是单台服务器降价，而是Blackwell级架构、Dynamo类调度、自研芯片、云厂商高CapEx和API定价机制共同降低每个有效token和每个任务的成本。Microsoft、Alphabet、Amazon、Met…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 1, "neutral": 0, "value": 72.4, "events": [ { "title": "训练capex不是需求坍塌，但宽基算力alpha正在成熟化；2026下半年应保留瓶颈算力、提高应用层利润验证权重。", "summary": "截至2026-05-30，本卡综合判断：训练capex的边际ROI尚未确认跌破拐点，NVIDIA、Microsoft和Amazon仍显示算力供给约束；但云厂商capex规模、短寿命GPU/CPU占比和自由现金流压力使宽基基建beta的估值门槛上升。2026下半年A股AI alpha更可能从宽基in…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 2, "risks": 3, "neutral": 1, "value": 80.8, "events": [ { "title": "AI 算力的'电力税'利润池正从被监管电网转移至云厂商自筹融资的分布式平行系统——近端属于燃气轮机/发动机 OEM 与天然气中游，远端属于 SMR 厂商与 HALEU 燃料循环。", "summary": "卡片 04（能源行业分析师，立场=支持）延伸卡片 03 的结论：PJM 5+ 年并网排队与 3 年 GPU 寿命的错配迫使 AI 园区离网。2026–2029 年由表后航改型燃气轮机与往复式机组（CAT、CMI、Wärtsilä、Bloom）主导桥梁方案，已有 XOM 二叠纪 1.5 GW、CVX…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "半导体管制是地域与结构冲击，不是截至2026-06-01已经触发的全球 AI capex 断崖。", "summary": "截至2026-06-01，本卡压力测试半导体出口管制是否会触发“AI组合端”崩溃。结论是否定：出口管制会伤害中国相关 SKU、库存和产品结构，但 NVIDIA FY2027 Q2 指引在假设中国 Data Center compute 为零时仍达 $91.0 billion +/- 2%，Micr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美元指数回落至94.50与利率下行会显著支撑AI硬件估值扩张，但对资本开支的影响是强化既有需求周期，而不是单独创造新周期。", "summary": "截至2026-05-31，本卡支持“美元与利率下行有利于AI硬件及半导体估值扩张”的主线，但限定传导强度：最强作用来自折现率下行、弱美元折算、流动性和订单信心，而非单纯融资成本下降。Microsoft、Amazon、Alphabet、Meta的最新季度AI相关资本开支运行速率已超过1306亿美元，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 2, "neutral": 0, "value": 69.2, "events": [ { "title": "Indeed 口径扰动削弱了“招聘崩盘”证据，但并不削弱 SBC 稀释高墙；真正应被修正的是科技巨头利润率非线性抬升的过度乐观定价。", "summary": "2026-06-01，本卡支持“Indeed 招聘幻觉”方向，但收窄结论：2026-03-31 Single Source feed 新规确认为职位列表口径断点，而非单独证明劳动需求崩塌。BLS、JOLTS、Workday 与 ADP 数据仍指向低速但有韧性的就业与 HCM/薪资席位。风险集中在市…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 硬件不共享软件端的 SBC 稀释墙，但 capex 弹性 + HBM 成本 + ASIC 替代三重传导意味着 NVIDIA 与 AI 供应链的利润率扩张预期同样需要下修。", "summary": "2026-06-01，本卡对前序首席策略师结论作压力测试：AI 硬件端面对的不是同一道 SBC 稀释墙——NVIDIA 季度 SBC 同比 +30.8% 至 $1.928B 但会计稀释仅约 0.43%，显著低于 Amazon/Alphabet/Meta 一档——而是更陡的需求弹性墙。真正的压力源是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 2, "risks": 2, "neutral": 0, "value": 86.6, "events": [ { "title": "AI实物建设仍支持芯片需求，但2026-27年的边际份额赢家是定制硅和网络而非商用GPU，因此B层维持中性、内部偏好AVGO/MRVL并对NVDA估值更谨慎。", "summary": "截至2026-06-02，本卡支持Research note 01的B层中性，但建议改为杠铃式中性：更建设性看Broadcom和Marvell等ASIC赋能商，对NVIDIA保持近端战术正面、结构估值更谨慎。2026年定制ASIC份额27.8%大概率守住；按44.6% ASIC增速和16.1%商用…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2026-06-02的核心判断是：AI资本开支仍能验证硬件龙头盈利，但估值剪刀差将通过支出方分化和相对降估值来修复。", "summary": "截至2026-06-02，本卡综合而非否定上一卡：AI硬件拥挤风险存在，但Hyperscalers的2026年资本开支区间仍约为6950亿-7250亿美元，足以支撑NVIDIA、Broadcom、TSMC等硬件龙头近期收入指引。分化在支出方：Microsoft与Alphabet的云收入验证更强，A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "本地AI Agent足以支撑NVIDIA/Microsoft/Arm的70%核心权重，但证据集中在开发者、企业治理和创作工具三条真实工作流，而不是所有AI PC芯片供应商。", "summary": "截至2026-06-02，本卡支持NVIDIA/Microsoft/Arm合计70%核心权重，但理由应限定为真实工作流而非泛AI PC概念。开发者Agent已有VS Code/GitHub本地工具链、付费Copilot套餐和硬件内存/GPU拉动；企业Agent有Microsoft 365 Copi…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "Anthropic 的高估值实质上是基础设施稀缺性的资本化体现，其 3300 亿美元的算力承诺既是模型领先的物理保证，也是将云厂商 Capex 压力转嫁给二级市场的潜在风险点。", "summary": "Anthropic 9650亿美元的估值深度绑定了超过3300亿美元的云算力承诺（AWS 1000亿，Google 2000亿）。虽然这锁定了面向2027年单次200亿美元训练成本所需的5GW+电力与下一代 NVIDIA Rubin/TPUv7 产能，但也使公司陷入“资本开支通道”模型。2026年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 78, "events": [ { "title": "AI基础设施利润扩张可以对冲成本压力，但只在加速器、先进代工、存储和电力/热管理等稀缺环节成立，不能外推到整个电子产业链。", "summary": "2026-06-04，本卡对AI硬件与电子产业链做成本压力测试：支持从公用事业红利向盈利成长轮动，但只支持精选AI硬件杠铃，不支持宽基电子超配。NVIDIA、TSMC、Micron等上游利润池具备高毛利与供给稀缺，能够吸收部分原材料、电力和元器件成本；但Microsoft、Meta、Alphabe…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 0, "neutral": 0, "value": 93.4, "events": [ { "title": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决", "summary": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决 AI Capex Pivot to Grid/SMR: Has Consensus Already Been Repudiated by NVDA/MRVL Price Actio…", "sourceLabel": "研究主题", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 2, "neutral": 1, "value": 77.8, "events": [ { "title": "TMT能用芯片能效、选址弹性、表后燃气与软件效率守住CAPEX美元，但守不住部署速度——估值通过time-to-power右移而非美元削减来反映监管成本，正是'好业绩不够好'机制的微…", "summary": "截至2026-06-06，TMT在位者用四条杠杆应对电力监管成本：芯片端Blackwell Ultra/Rubin的perf/W、选址迁移至MISO/SPP和跨境、表后燃气(2026-2027主力)与核电(2028+)、以及软件推理效率。压力测试结论是部分有效但不对称——hyperscaler已承…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "博通 ASIC 护城河的设计服务层 2027 年前不可替代，但客户集中层正在被稀释；真正的脆弱点不是 Google，而是 OpenAI 量产时点。", "summary": "压力测试结论：Google TPU 二供（联发科自 v7 起）、Meta MTIA、Amazon Trainium 对博通的威胁是分层的——Google 是真正稀释（2026–27 收入下行 –10% 至 –20% [自测算]），Meta 在 12 个月窗口仍是支撑，AWS 本就以 Marvell…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "博通的估值回归并非基本面崩溃，而是从“预期完美”向“竞争性执行”的结构性切换，Google 的多元化供应是 AI 叙事的首道裂缝，但 OpenAI 和 Meta 的增量支撑了其底部逻…", "summary": "博通 Q2 财报后股价暴跌 15%，主因是 Q3 AI 收入指引（160 亿美元）不及市场极高的“私下预期”（172 亿美元），且 Google 确认将分散 ASIC 供应链。尽管全年 560 亿美元目标维持不变，但“唯一供应商”溢价的消退引发担忧。散户提及量激增 230% 试图抄底，反映出市场对…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-07", "supports": 1, "risks": 0, "neutral": 0, "value": 96, "events": [ { "title": "头部CSP的AI投资回收期在HBM4紧缺下并未恶化，反而因稀缺产能高利用率和架构效率提升而缩短，支撑AI硬件的选择性估值溢价。", "summary": "截至2026-06-07，本卡支持并综合前序结论：HBM4紧缺不会打断头部CSP的AI资本开支周期，而是强化“配给能力+效率改善+现金回收”的分化。Microsoft、Alphabet、Amazon披露的需求、积压订单和AI收入信号表明，头部云厂商的有效回收期可能压缩至约1.6-2.3年；Meta…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-08", "supports": 1, "risks": 0, "neutral": 1, "value": 96, "events": [ { "title": "可信的 Apple 端侧 AI 栈是已有分化的催化剂而非起因——它会加速半导体晶圆配置轮动与软件分发护城河的相对重估，但不会在主题演讲周内倒转 Mag 7 排序。", "summary": "WWDC 2026（2026-06-08 主题演讲周）若 Apple 端出可信的端侧 AI 栈（A19/M5 上 7–10B 级模型、Foundation Models 框架对开发者开放），不会倒转 Mag 7 排序，但会对 MSFT/GOOGL 压缩分化、对 META/AMZN 拉开分化，加速从…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决", "summary": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决 global-macro、thematic-researcher、tmt-analyst 三人独立得出 'capex 由 GPU 服务器向电网/SMR/变压器分流' 的同一结论（g…", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-09", "supports": 2, "risks": 2, "neutral": 0, "value": 95, "events": [ { "title": "核电交付时滞会压低2026-06至2027年的AI算力收入变现斜率，但不会立即等同于GPU采购需求见顶。", "summary": "截至2026-06-10，本卡支持前序判断但作出TMT侧区分：2026-06至2027年更可能出现“通电、可确认收入的AI算力容量”平台期，而非GPU订单断崖。Microsoft FY26 Q3资本开支319亿美元且约三分之二投向GPU/CPU，Amazon Q1 PPE购买额442.03亿美元，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电表后天然气是物理上可行、政策上昂贵的桥接方案：它不推翻AI电力基础设施交易，但会给超大规模云厂商的2030净零叙事增加Scope 1披露、CDR供给和绿色声明合规折价。", "summary": "2026-06-09：本卡对卡5的天然气桥接结论进行政策压力测试。结论是，电表后天然气能解决2026-2029年AI数据中心通电缺口，但会把排放从可通过合同优化的Scope 2转成更难解释的Scope 1，并压缩Microsoft、Google、Meta、Apple等2030净零/负碳承诺的可信度…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI半导体利润池迁移真实但渐进，板块总EPS上行、估值锚从'NVDA-TSMC双寡头溢价'切向'增长+多元化第二供应源'，ASIC设计商、HBM、先进封装设备与Intel Found…", "summary": "2026-06-09综合：Google-Intel TPU订单确认AI硬件利润池迁移真实且渐进，2027-28年AI半导体合计EPS较冲击前共识更高，但分布坚定向ASIC设计商(AVGO/MRVL)、HBM、先进封装设备及Intel Foundry期权倾斜。新估值锚为'增长+多元化第二供应源'，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-10的核心判断是：AI主线没有结束，但边际定价从芯片估值倍数切向可交付电力与基建设备稀缺。", "summary": "截至2026-06-10，本卡综合支持“AI基建稀缺重估”而非“AI芯片需求见顶”。NVIDIA、Microsoft、Amazon 的数据仍显示 AI 需求与资本开支强劲，但 Applied Digital 从2026-06-08签约到2028年一季度运营的18-21个月时滞、FERC核电共址受阻…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 0, "value": 86.4, "events": [ { "title": "2026-06-10的主线不是在云、终端和政策之间三选一，而是超配云需求确认，选择性持有HBM/国产替代，暂把Apple换机作为二阶段期权，并低配政策与现金流暴露高的AI资产。", "summary": "截至2026-06-10，本卡综合Oracle业绩窗口、Apple AI终端升级和美国先进芯片海外子公司管制澄清。结论是：A股AI链条应优先交易云需求确认，其次保留Apple端侧AI换机期权，同时用出口管制和自由现金流压力控制仓位。Oracle、Microsoft、Alphabet、Meta仍显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 3, "neutral": 0, "value": 75.8, "events": [ { "title": "铜、铝、电缆和电气设备稀缺已经进入AI算力集群交付链，最先约束的是通电日期和可售云容量，而不是GPU本身。", "summary": "截至2026-06-11，材料链稀缺已开始转化为AI基础设施交付摩擦，但表现为通电时间、电气设备交期和云算力容量分配，而非所有GPU集群停摆。IEA提示约20%规划项目有延迟风险，CBRE称电力约束把工期推至2027年及以后，Wood Mackenzie给出关键电气部件18-36个月交期和600 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建瓶颈已实质性从芯片转移到电力与电网，且这一约束在12个月维度上无改善迹象，支持继续加配变压器、燃机、EPC与表后发电资产。", "summary": "美国超大规模厂商AI基建的边际约束已从GPU分配迁移到并网排队和一次电力：LBNL显示在排接入容量>2.6 TW、中位等待约5年；PJM/Dominion大负荷新申请要排4–7年；大型变压器交付从50周拉长到120–210周；GE Vernova重型燃机2028年前售罄。同期H100/B200交付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 2, "neutral": 0, "value": 83.2, "events": [ { "title": "截至2026-06-13，电力设备成本与交付风险已经实质性延后 AI 算力集群通电节奏，但它表现为 capex 效率恶化和项目重排，而不是最大型 hyperscalers 已经全面削…", "summary": "截至2026-06-13，本卡支持前序“电力约束正在惩罚 AI capex/融资假设”的方向，但精确化为：2026-2028 年已出现项目层面的通电延期与节奏重排，尚未出现头部 hyperscalers headline capex 的广泛下修。Sightline 显示 2026 年约 16 GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的扩展性：电力与芯片供应瓶颈对超大规模厂商CAPEX的影响分析", "summary": "截至2026-06-12，本卡压力测试卡片02（电力瓶颈）与卡片06（半导体多线挤压）。结论：双重约束不会削减超大规模厂商资本开支，而是将其重排。2026资本开支仍约$725-726B（Dell'Oro），资本开支/营收比约25-30%、指引2027约35%；变化在优先级——可靠电力升至栈顶，建设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 3, "neutral": 2, "value": 84.6, "events": [ { "title": "红皇后竞速并未彻底抹去自研ASIC的成本优势，而是将其集中至Google与AWS两家、使其他云厂商回到对NVIDIA的依赖，从而部分推翻了Research note 05所述的全行业…", "summary": "在'红皇后竞速'下，自研ASIC单世代有效寿命已压缩至12–15个月，2nm节点全口径NRE攀升至USD 0.9–1.4B、N2晶圆约USD 30,000/片、CoWoS-L售罄至2027。我们测算自研ASIC在TCO上跑赢商用GPU的年部署门槛从5nm节点的约20万颗升至现实情景下的15万–25…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "硬件与能源通胀只能部分向软件端转嫁，头部云厂商2026年不至于利润率崩塌，但AI云业务的边际回报率和自由现金流将被结构性压缩。", "summary": "截至2026-06-14，本卡支持前序“资本强度陷阱”链条：云厂商可在稀缺GPU、预留容量、企业工作流和agent定价中转嫁部分成本，但公开API价格已被OpenAI、Anthropic、Google等竞争锚定，难以承受全面涨价。AWS、Azure、Google Cloud短期报表利润率未崩塌，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Higher-for-longer不会必然叫停AI capex，但会把市场关注点从“能否花钱”转向“每个增量GPU集群能否在折旧和电力成本吞噬定价窗口前跨过更高ROI门槛”。", "summary": "截至2026-06-14，本卡对卡片1的高利率与美联储战略模糊情景做压力测试：AI基础设施capex的关键脆弱点不是大型云厂商短期无钱可投，而是折现率、GPU短寿命折旧、电力并网与PPA成本共同抬高项目资本回收门槛。以三分之二6年期算力资产、三分之一15年期数据中心资产测算，200bp WACC冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "贸易壁垒与出口管制正在提高自研ASIC的进入门槛，使AI算力利润进一步向具备规模化能力的“两大巨头”（Google/AWS）集中，并导致全球算力成本出现严重的主权分化。", "summary": "截至2026-06-14，本研究评估了贸易壁垒与出口管制对AI算力TCO的影响。研究发现，25%的Section 232关税、90%的HBM价格涨幅以及CoWoS产能极度短缺共同构成了“韧性税”，将自研ASIC的盈亏平衡阈值从15万颗推高至25万颗/年。Google与AWS凭借规模优势和内部转移定…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 5, "neutral": 0, "value": 76, "events": [ { "title": "2026年AI交付的决定性变量是项目级电力控制权——离开并网队列（BTM自发电或运营中核电并置）的运营商可交付，靠申请负荷计数者应被折价30–50%+。", "summary": "本卡将电网可交付约束综合为项目级评分表。可交付兆瓦需同时通过确定性供电合同、并网/FERC放行、长交期设备三道闸门。赢家均靠离开并网队列锁定首期电力：Amazon@Susquehanna（1,920MW表前PPA）、xAI Memphis（约1.2GW BTM燃机）、Stargate Abilen…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与电网瓶颈对Tier-1 2026–2027并网影响已基本被锁单、棕地与表后发电对冲；真实滑期约3–7 GW集中在2028+绿地与Tier-2投机建设，Research not…", "summary": "本卡(10/10)压力测试变压器/GSU/电网交期对Tier-1 AI数据中心并网的真实约束。结论：瓶颈真实存在但Tier-1时间表已基本被定价——2022–2024年下达的多年期框架订单、棕地复用既有并网点、表后核电与现场燃气PPA共同吸收了120–210周的设备交期。自测算显示2027–202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施资本开支仍支持互连和设备链条，但投资确定性应优先放在PCIe/DDR5/MRDIMM/光互连，再看CXL和国产设备盈利兑现。", "summary": "截至2026-06-15，本卡支持上一卡关于AI服务器互连和国产设备的方向，但将确定性分层：Hyperscaler资本开支仍在上行，Microsoft、Alphabet、Meta、Amazon的最新披露共同验证AI基础设施支出强度；PCIe Retimer/Switch、DDR5/MRDIMM和光…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "GPU抵押贷款收缩会削弱边际Neocloud订单可见度，但先进封装与先进制程需求更可能从融资套利订单转向信用更硬的超大规模厂商和ASIC路线，而非进入全面下行。", "summary": "截至2026-06-15，本卡压力测试GPU抵押融资收缩对半导体订单的影响：冲击不是AI芯片需求消失，而是订单可见度从Neocloud融资驱动采购转向超大规模厂商、ASIC、主权项目和带预付款的硬合同。NVIDIA Data Center在FY2027 Q1仍达752亿美元，TSMC先进节点和HP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 0, "value": 87.4, "events": [ { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "验证AI算力硬件链的估值承压：流动性虹吸下的先进封装与GPU产业链韧性评估", "summary": "本卡对AI算力硬件链做压力测试，得出'分化韧性'结论。资金流层面确认虹吸：2026-06-08半导体−219.33亿、通信−98.35亿、消费电子−69.92亿元净流出居前[S6]，年内93家半导体公司减持[S7]，叠加CXMT约295亿IPO锚定[S9]。但估值下移是资金流/轮动驱动、非盈利驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 2, "neutral": 0, "value": 88.8, "events": [ { "title": "AI基建资本开支仍在加速，但从公告美元转化为已通电MW的比例正在下降，首个断点在数据中心交付和二线下游供应链，而不是NVIDIA近端GPU订单。", "summary": "截至2026-06-16，未看到北美云厂商AI资本开支或GPU订单总量已经广泛下修；Microsoft、Alphabet、Meta、Amazon、NVIDIA、Dell披露仍指向扩张。支持前序压力结论的证据主要来自数据中心交付层：Goldman Sachs预计未来1-2年计划容量仅50-60%按时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供给限制驱动的半导体能效转型主要表现为通用GPU的系统与精度级优化，而非ASIC/NPU的大规模替代；自研定制芯片的爆发反而加剧了全球先进制程与封装供应链对TSMC的物理集中度。", "summary": "截至2026-06-17，电力瓶颈确在倒逼算力能效转型，但并非单纯的通用GPU向专用ASIC切换。算法多变性（如MoE与动态推理）使自研ASIC面临极高的“代差失效”风险，通用GPU（如NVIDIA Blackwell）凭借软件与架构灵活性仍占据70%以上份额。能效提升主要通过系统精度（FP4/F…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 3, "neutral": 0, "value": 84.2, "events": [ { "title": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：…", "summary": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：AI 基础设施分析师 (ai infrastructure analyst) 立场：stress test（压力测试…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler capex 足以穿越能源 TCO 上行并支撑稀缺供应链，但投资表达必须是“多稀缺、空低 ROIC”，而不是无差别做多 AI。", "summary": "2026-06-18 压力测试结论：hyperscaler capex 增长足以让 AI 基础设施供应链消化能源成本带来的边际 TCO 压力，但只能支持“多稀缺”，不能支持泛 AI beta。四大买方 2026 capex 粗略包络约 $710B（自测算），Microsoft、Alphabet、M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力瓶颈是 AI capex 的中期执行上限，而不是 2026-06-17 当下的叙事破裂触发点。", "summary": "2026-06-17 的压力测试结论：电力、电网与数据中心容量约束是真实且正在制度化的瓶颈，但尚不足以在 2026-06-16/17 FOMC 窗口内击穿 AI capex 叙事。LBNL、IEA、BloombergNEF、PJM 和 NERC 数据显示需求规模、在建容量和并网可靠性压力都在上升；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 5, "neutral": 0, "value": 73.6, "events": [ { "title": "AI交易的边际定价已经由软件久期切换为实体AI产能和电力瓶颈的资本开支执行，但软件并未失去长期价值。", "summary": "截至2026-06-18，本卡支持AI交易已从泛软件久期转向算力硬件、电气/热管理、电网接入和工业供能执行。核心证据是Microsoft、Amazon、Alphabet和Meta最新季度实体capex/PPE披露按粗略口径合计约1307亿美元，NVIDIA数据中心收入仍高速增长，IEA/美国能源部…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-19，前沿模型风险闸门更可能改变AI基础设施资本开支的节奏和形态，而不是终止云厂商的算力扩张。", "summary": "截至2026-06-19，本卡支持上一卡结论，但把传导机制落到AI基础设施：风险闸门不会简单压低hyperscaler资本开支，而是把节奏改为“电力/园区前置、GPU分段、合规容量池预留”。Anthropic Fable 5 / Mythos 5的2026-06-09发布与2026-06-12暂停…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装降低每bit能耗后，Hyperscalers更可能把能效红利用于扩大推理部署和提高每兆瓦收入，而不是削减资本开支。", "summary": "截至2026-06-19，本卡支持“先进封装与能效提升会缓解电力瓶颈并推动Hyperscaler二次部署”的判断，但强调这是受firm power约束的扩张而非无约束爆发。Microsoft预计2026日历年资本开支约1900亿美元，Alphabet预计2026年1800-1900亿美元，Meta…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM3良率瓶颈将AI基础设施风险从总量需求崩塌改写为本土投产节奏后移，并强化A股高融资TMT的收入兑现左尾风险。", "summary": "截至2026-06-18，本卡压力测试卡片06：国产HBM3良率瓶颈会推迟中国本土AI算力变现，但尚不足以证明全球AI基础设施从快速扩张转入需求观望。全球hyperscaler capex仍有Meta、Alphabet、Microsoft等高额指引支撑；真正变化在中国本土项目从按园区容量扩张转向按…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 4, "neutral": 0, "value": 81, "events": [ { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "ASIC光互联需求真实存在，但2026-2028中际旭创式CPO供应商的正确压力测试是按电力状态下修约20-25%的收入节奏，而不是机械套用2.92x光模块倍数。", "summary": "截至2026-06-20，本卡支持电力约束会重塑CPO收入节奏，但否定“ASIC光模块用量固定为GPU的2.92倍”这一过强假设。AWS Trainium2、Google TPU、Meta MTIA和Broadcom AI半导体收入证明ASIC design win真实存在；OpenAI/Broa…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与封装成本上升会触发AI基础设施技术路线调整，但截至2026-06-20仍不足以触发超大规模云厂商AI芯片采购总量的全面回撤。", "summary": "工作日2026-06-20：本卡压力测试认为，电力和先进封装成本上升不会在2026年触发超大规模云厂商全面削减AI芯片采购，但会提高单位弹性，推动技术路线从“更多GPU插槽”转向“每瓦性能、液冷机架、自研ASIC、提前锁定CoWoS/HBM和电力站点”。真正风险是并网、变电站、CoWoS或HBM配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CPO能维持相对传统光模块的选择性溢价，但不能在供电受限情景下整体对冲电网扩容延迟，因此硬件链仍只应保留小额期权仓位。", "summary": "截至2026-06-19，本卡对CPO估值溢价做压力测试：CPO在链路和交换机层节能真实，NVIDIA称3.5x能效、10x韧性，Broadcom/Meta验证显示1 million个400G等效端口设备小时无link flap且光学功耗-65%。但站点层折算有限，Lambda的4,608 GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 5, "neutral": 0, "value": 76.4, "events": [ { "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "summary": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI capex supercycle 仍在，但 higher-for-longer 与能源波动会把扩产闸门从芯片采购转向电力可交付性、容量成本和融资期限匹配。", "summary": "截至 2026-06-21，本卡对 higher-for-longer 与能源波动环境下 AI 算力基建做压力测试。结论：超大规模云厂商不会停止扩产，但边际决策从“GPU 可得性”转向“firm megawatts + 客户承诺 + 可融资全成本”。Amazon、Alphabet、Meta 已披露…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网通电限制不会立即终结 AI 需求，但会把 TMT 估值从“芯片稀缺溢价”重构为“确定电力、利用率和现金流转化溢价”。", "summary": "截至 2026-06-21，本卡压力测试电网通电限制对 TMT 的影响：AI 需求仍强，但估值核心从 GPU 可得性转向 time-to-power、容量变现和 FCF 久期。Microsoft 和 Amazon 的主要风险是 AI capex 先发生、通电和云收入后确认，导致利润率与自由现金流承…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CapEx前置会提高云巨头长期资本开支和固定费用压力，并把开发商信用风险重塑为租户信用、电力交付、起租和融资载体风险，而不是让风险消失。", "summary": "截至2026-06-20，本卡压力测试数据中心开发商“CapEx前置”模式。结论是风险被迁移而非消失：开发商通过预租、担保、LC和电力义务降低投机建设风险，但Hyperscaler承接更多融资租赁、未开始租赁、直接capex和专属电力费率。Alphabet、Amazon、Meta、Microsof…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 0, "value": 95.8, "events": [ { "title": "AI基础设施的边际稀缺点正从GPU扩散到可通电兆瓦和电网电气设备，因此制造出海主线应从拥挤硬件β轮动到“电力绑定的算力基础设施”。", "summary": "截至2026-06-21，本卡支持上一卡的再平衡结论：AI硬件存在战术拥挤，但云厂商资本开支并未结束，而是转向算力、电力、电网设备协同扩建。Microsoft、Alphabet、Meta、Amazon的2026资本开支披露或指引合计约7000亿美元（自测算），而DOE/LBNL、IEA、EPRI和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力容量、并网权和长周期电气设备的稀缺性为AI资本开支提供物理底部，但若负荷预测虚高、AI利用率不足或监管成本分摊恶化，该刚性会转化为新增项目放缓。", "summary": "截至2026-06-21，本卡支持上一卡“估值先受压、资本开支不立即崩塌”的判断。核心证据是电力与电网瓶颈正在把AI数据中心建设变成具备供给刚性的公用事业级工程周期：美国数据中心用电占比已从2023年的4.4%走向2028年6.7%-12%的区间，变压器交付仍需2-3年，Dominion等枢纽区域…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 4, "neutral": 0, "value": 85.2, "events": [ { "title": "在66.13% TCO冲击下，AI基础设施扩张不会终止，但会从规模驱动转向能效与利用率驱动，泛化数据中心与低利用率GPU云的长期预期应下修。", "summary": "截至2026-06-22，本卡压力测试上一卡66.13% TCO激增情景。结论是：头部hyperscaler资本开支不会急停，Microsoft、Alphabet、Meta、Amazon仍有收入、backlog或客户承诺支撑；但新增产能逻辑会从“抢GPU/抢机房规模”转向“单位功耗有效产出、签约利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体资本开支收缩会推迟AI集群扩建的判断成立，但只在AI专用供应瓶颈受冲击时成立；截至2026-06-22，更现实的约束是芯片稀缺、组件通胀与电网通电共同拉长边际集群排期。", "summary": "截至2026-06-22，本卡支持前序主线但收窄机制：半导体资本开支骤降会推迟Hyperscaler AI集群，前提是冲击落在TSMC先进制程、HBM、先进封装、网络和供电等AI关键瓶颈，而非普通消费芯片。现有证据反而显示AI关键资本开支仍在扩张：TSMC指向US$52-56 billion区间高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "summary": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "定制化ASIC驱动下的AI算力集群架构与液冷能效演进", "summary": "本卡支持并延伸 Research note 08：电网瓶颈倒逼的 GPU→ASIC 转向，沿“硅→织构→机房”逐层外溢，重构数据中心系统。互连层：Google 以 OCS+3D-Torus 抽掉一层电交换机，NVIDIA 以 CPO（Quantum-X/Spectrum-X 2026 商用）把光引…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 5, "neutral": 1, "value": 83.6, "events": [ { "title": "2028年再融资悬崖与 AI 算力基础设施资本开支的信用冲突", "summary": "压力测试支持并延伸卡片 04：AI 基础设施信用风险真实且上升，但呈哑铃状集中，不在头条的 $725 bn 巨头资本开支上。四巨头（MSFT/GOOGL/AMZN/META，2026 资本开支约 $725 bn、+77%，约 75% 投 AI [S1]）凭现金流与 AA 资产负债表通过测试，尽管 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下修。", "summary": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027年的电力冲击会抬高AI容量成本并重排Hyperscaler部署顺序，但更可能造成通电与收入确认延迟，而非CapEx总量或GPU订单的广泛取消。", "summary": "截至2026-06-23，本卡支持前序判断：2027年电力约束不会直接演化为Hyperscaler CapEx总量削减，而会把部署逻辑从“GPU优先”改为“通电优先”。Microsoft、Alphabet、Meta、Amazon披露的2026 AI基础设施支出仍然庞大，但PJM容量价格、美国DC电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler并不缺AI Capex，真正约束算力上线节奏的是可交付电力、并网流程与长交期电力设备。", "summary": "截至2026-06-24，本卡支持上一卡：AI算力瓶颈已从Capex/GPU转向通电MW。Alphabet 2026 CapEx指引1,800-1,900亿美元、Meta 1,250-1,450亿美元，Microsoft仍称需求超过可用容量，Amazon Q1购置物业与设备442.03亿美元；但S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 6, "neutral": 0, "value": 76, "events": [ { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件交付节奏与光模块订单验证", "summary": "截至2026-06-24，本卡支持“CPO延期不等于800G/1.6T砍单”的基本面判断，但加入节奏约束：Microsoft、Alphabet、Meta、Amazon的2026 capex与NVIDIA Rubin、Dell AI服务器订单、Lumentum光组件指引共同显示需求仍强，800G/1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体杠杆清盘对Hyperscaler AI资本开支的近端传导主要是价格、排序和边际买方融资纪律，而不是一线供应链信用崩塌或Mega-cap资本开支急刹车。", "summary": "截至2026-06-24，本卡压力测试认为，韩国半导体杠杆ETF清盘会重定价AI硬件股票并强化资本市场审视，但不构成Mega-cap Hyperscaler近端AI资本开支削减的主传导。Microsoft、Alphabet、Meta、Amazon仍以强经营现金流和投资级融资支撑AI建设；NVIDI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "持续能源通胀会压缩AI基础设施回报率，但真正改变Hyperscaler扩建节奏的是可交付电力的稀缺，而不是电费单项上涨。", "summary": "截至2026-06-25，本卡压力测试能源通胀对AI基础设施的传导。结论是：电价上升本身不足以触发Hyperscaler AI capex断崖；更关键的是并网、容量市场、PPA和电网设备交付约束会重排项目优先级。Microsoft、Alphabet、Meta、Amazon仍合计指向约6950亿-7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 3, "neutral": 0, "value": 95.4, "events": [ { "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CNH套息去杠杆会先打掉A股AI基础设施股票的流动性溢价和估值久期，而不是立即打断全球AI CapEx实物周期。", "summary": "截至2026-06-25，本卡片压力测试CNH套息去杠杆对AI基础设施估值和CapEx预期的冲击。结论是：流动性溢价消失可使A股AI硬件/算力基础设施估值出现非线性回撤，但不应直接推导为全球hyperscaler CapEx崩塌。Microsoft、Alphabet、Meta、Amazon、Ora…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "评估AI基础设施与半导体产业链在“更高更久”利率环境下的现金流质量与估值韧性", "summary": "压力测试Research note 2后结论：不能把AI基础设施当成单一“质量成长防御”桶——产业链已分化。算力卖方（NVDA型）财年2026营收$215.9B、经营现金流$102.7B、FCF$96.6B、FCF利润率约44.7%，当期造血、净现金，确属短久期质量成长。算力买方（MSFT/GOO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 2, "neutral": 0, "value": 96, "events": [ { "title": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "summary": "截至2026-06-27，本卡片对AI基础设施链做压力测试：全球AI资本开支仍大，Goldman Sachs基准模型给出2026年USD 765 billion、2031年USD 1.6 trillion的年度AI capex，但Amazon、Alphabet、Meta、Microsoft的数据表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 0, "value": 96, "events": [ { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器材料和交付瓶颈已开始改变 Hyperscaler 的2026-2027资本开支与电力采购节奏，影响主要体现为预算通胀、长期锁电和通电交付递延，而非AI需求取消。", "summary": "截至2026-06-28，结论为支撑：Hyperscalers 已经通过预算上修、长期电力合同和项目节奏调整来应对变压器及电气设备瓶颈，但公开披露通常不会把变化直接归因为 GOES 或铜关税。Microsoft 2026 capex 中约250亿美元来自更高组件价格，Meta 上调2026 cap…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电子链条利润反弹主要受AI基础设施投资拉动，具备向存储、设备、材料、先进封装、光互联和数据中心电源散热扩散的潜力，但尚不足以证明消费电子和内需全面复苏。", "summary": "截至2026-06-28，本卡支持上一卡判断：电子链条利润修复主要由AI基础设施、算力芯片和存储涨价驱动，而非内需广谱复苏。国家统计局数据显示，1—5月电子行业利润同比增长103.9%，贡献全部工业利润增量43.1%。NVIDIA、Microsoft、Alphabet、Meta、Alibaba和T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 4, "neutral": 0, "value": 96, "events": [ { "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "summary": "截至2026-06-28，本卡否定“AI资本开支韧性等于A股防御性配置锚”的宽泛结论。Microsoft、Alphabet、Meta、NVIDIA、TSMC等数据仍支持AI基础设施真实需求，但FOMC 3.50%-3.75%、5月PCE同比4.1%、10年美债4.40%和A股两融突破RMB 3tn…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI交易不会结束，但估值核心应从买到多少芯片转向能否按期通电、保持利用率并降低每token能耗。", "summary": "截至2026-06-28，本卡综合支持前序判断：2026-2030年AI扩张的约束从服务器交付转向可通电MW交付。Amazon、Microsoft、Alphabet、Meta最新季度CAPEX/PPE购买合计约$130.8B[自测算]，但IEA预计美国数据中心用电到2030年较2024年增加约24…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据中心电网并网瓶颈与核能转向", "summary": "截至2026年6月28日，AI基础设施建设的核心瓶颈已从GPU稀缺转移到电网并网时间（Time-to-Power）。对100 MW超大规模AI数据中心的模型测算显示，2年的并网延迟会减少项目35.73%的净现值（NPV），并使IRR大幅下降。为规避此类延迟，科技巨头（微软、亚马逊、谷歌、Meta）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-8c5a9690a0/", "networkHref": "/investor/research-network/#thesis-8c5a9690a0" }, { "id": "thesis-c89650af9e", "source": "tracked_thesis", "title": "酒店餐饮休闲复苏正在从客流修复转向 RevPAR、翻台率与小盘 Alpha", "titleEn": "Leisure, hotels, and restaurants are moving from traffic recovery to RevPAR, table turnover, and small-cap alpha", "lane": "酒店餐饮休闲/服务消费", "laneEn": "Hotels, restaurants, leisure, and services consumption", "laneHref": "/investor/lanes/hotels-restaurants-leisure-and-services-consumption/", "investableFocus": "酒店餐饮休闲/服务消费：1179.HK, 9961.HK, 3690.HK, YUMC, MAR, BKNG, SBUX", "stockUniverse": [ { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "YUMC", "name": "Yum China", "market": "US", "role": "连锁餐饮", "roleEn": "连锁餐饮", "bucket": "core", "href": "/investor/stocks/YUMC/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "BKNG", "name": "Booking Holdings", "market": "US", "role": "全球 OTA 对照", "roleEn": "全球 OTA 对照", "bucket": "peer", "href": "/investor/stocks/BKNG/" }, { "ticker": "SBUX", "name": "Starbucks", "market": "US", "role": "可选餐饮需求对照", "roleEn": "可选餐饮需求对照", "bucket": "peer", "href": "/investor/stocks/SBUX/" }, { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "1179.HK", "9961.HK", "3690.HK", "YUMC", "MAR", "BKNG", "SBUX", "600519.SH", "300760.SZ", "600276.SH", "001215.SZ", "MU", "APO", "IEF" ], "coreView": "赛道：酒店餐饮休闲/服务消费。候选观察标的：1179.HK, 9961.HK, 3690.HK, YUMC, MAR, BKNG, SBUX。当前共有 40 条证据、2 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 85, "eventCount": 160, "firstSeen": "2026-05-01", "lastSeen": "2026-06-29", "supports": 5, "risks": 126, "neutral": 29, "analystCount": 16, "sourceTypeCounts": { "分析协作": 20, "研究笔记": 136, "证据更新": 1, "研究线程": 1, "深度研究": 1, "研究主题": 1 }, "topTerms": [ "GOES", "CPI", "RevPAR", "Alpha", "Leisure", "hotels", "restaurants", "moving", "traffic", "recovery", "table", "turnover" ], "whatChanged": [ "2026-06-29: 材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放右移。", "2026-06-29: 2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "2026-06-29: 霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "2026-06-29: [auto-handoff] 霍尔木兹海峡石油冲击与核心CPI传导", "2026-06-29: 电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀" ], "practical": { "sectorLane": "酒店餐饮休闲/服务消费", "sectorLaneEn": "Hotels, restaurants, leisure, and services consumption", "exposureBucket": "消费健康防御", "tone": "green", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "YUMC", "name": "Yum China", "market": "US", "role": "连锁餐饮", "roleEn": "连锁餐饮", "bucket": "core", "href": "/investor/stocks/YUMC/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "BKNG", "name": "Booking Holdings", "market": "US", "role": "全球 OTA 对照", "roleEn": "全球 OTA 对照", "bucket": "peer", "href": "/investor/stocks/BKNG/" }, { "ticker": "SBUX", "name": "Starbucks", "market": "US", "role": "可选餐饮需求对照", "roleEn": "可选餐饮需求对照", "bucket": "peer", "href": "/investor/stocks/SBUX/" }, { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "1179.HK", "9961.HK", "3690.HK", "YUMC", "MAR", "BKNG", "SBUX", "600519.SH", "300760.SZ", "600276.SH", "001215.SZ", "MU", "APO", "IEF" ], "primaryStocks": [ { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "YUMC", "name": "Yum China", "market": "US", "role": "连锁餐饮", "roleEn": "连锁餐饮", "bucket": "core", "href": "/investor/stocks/YUMC/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "BKNG", "name": "Booking Holdings", "market": "US", "role": "全球 OTA 对照", "roleEn": "全球 OTA 对照", "bucket": "peer", "href": "/investor/stocks/BKNG/" } ], "instrumentTags": [ "1179.HK", "华住集团-S", "9961.HK", "携程集团-S", "3690.HK", "美团", "YUMC", "Yum China", "MAR", "Marriott International", "BKNG", "Booking Holdings" ], "reviewCadence": "每日复核", "confirmationSignals": [ "价格弹性是否转化为利润率修复", "支付端和监管是否支持长期现金流", "防御属性是否被估值提前透支" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "估值拥挤", "支付监管", "渠道库存", "CPI 食品项" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 6 类，近期事件 127 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 6, "analystCount": 16, "recentEvents": 127, "supportRatio": 0.03, "riskRatio": 0.79, "conflictScore": 89.8, "evidenceDepth": 100 }, "score": 78.7 }, "events": [ { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 21, "matchedTerms": [ "CPI", "evidence", "risk", "dominant", "inflation", "transmission", "falling" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]，PPI-CPI剪刀差隔离了政策利率。任务设想的'成本→CP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "GOES", "CPI", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 霍尔木兹海峡石油冲击与核心CPI传导", "summary": "致 投委会主席： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 1d956e69-9dc0-4197-8bd0-eb1630441435 结果摘要: Wrote handoff_response.md (11296 bytes); wrote handoff_response.en.md (11717 bytes) — The re…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 12.25, "matchedTerms": [ "CPI", "risk", "inflation" ], "href": "" }, { "date": "2026-06-29", "title": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "summary": "截至2026-06-29，支持卡片01–03的逻辑链并延伸至终端利率与全球流动性维度。美联储6-17维持3.50%–3.75% [S2]、SEP 2026中值3.8% [S3]；新主席Warsh首场FOMC点阵图从降息翻转为加息，9名官员预期加息、年内加息概率50.4%、市场定价10月加息，反应函数转向'约束性务实主义' [S14]。情景A（布伦特约105…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.75, "matchedTerms": [ "CPI", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "summary": "对中国，2026年油价/柴油冲击既是被缓冲的进口税、又是战略加速器，且净额上比卡片01–02的美欧日集团损害更小。三大事实：（1）进口税真实但落在PPI而非CPI——5月PPI同比+3.9%（2022年7月来最快）、CPI仅+1.2%，疲弱内需把冲击吸收为生产者利润压缩；（2）进口税被三重缓冲——全球最大约14亿桶战略储备、作为受制裁油折价买方、疲弱需求；…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.75, "matchedTerms": [ "CPI", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放…", "summary": "截至2026-06-29，变压器交期延长和材料瓶颈导致美国电网资本开支通胀率达32.72%。结合2027/2028年度PJM容量电费暴涨11.5倍至$333.44/MW-day，将拉动美国零售电价上涨4.18%，直接和间接推高全国CPI 0.124个百分点。高粘性通胀与AI基建庞大资金需求将上移中性利率50-75基点，延缓美联储降息路径。此外，并网滑期18…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.75, "matchedTerms": [ "CPI", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼及普钢加工环节将面临严重的利润率挤压与低…", "summary": "本卡通过技术壁垒与自给率对A股AI原材料敞口进行了区分：高端取向硅钢方面，宝钢股份凭借规模与技术优势展现出强Alpha属性，而首钢股份受普钢周期Beta拖累较大；有色铜方面，在2026年长单TC/RC为零的极端低位下，矿权自给率超65%的紫金矿业可完全捕获矿端溢价（EBITDA价格敏感度约7.5%），而低自给率的江西铜业、云南铜业面临严重的加工毛利压缩。建…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13, "matchedTerms": [ "GOES", "Alpha", "alpha" ], "href": "" }, { "date": "2026-06-28", "title": "关键矿产与设备出口管制风险对A股高端环节溢价逻辑的冲击评估", "summary": "压力测试出口管制威胁。结论：把镓/锗/锑/稀土剧本套到宝钢与紫金上是双向错配，不反转 Research note 06 基准论断。铜/紫金：中国是全球最大铜进口国（2025 精矿进口 30.31Mt，国内矿仅约 1.85Mt，依存约 87%），出口武器结构上不存在；紫金约 50% EBITDA 来自海外矿（刚果金、塞尔维亚、哥伦比亚），真正风险是入境制裁/…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13, "matchedTerms": [ "GOES", "Alpha", "alpha" ], "href": "" }, { "date": "2026-06-28", "title": "截至2026-06-28，最优跨资产表达是“久期+稀缺股票”的杠铃，铜只做长周期期权，因为近端精炼过剩削弱了直接商品敞口的性价比，而GOES稀缺主要只能通过精选股票承接。", "summary": "截至2026-06-28，本卡综合八卡链条：物理瓶颈应表达为“久期对冲下的稀缺alpha”，而不是单一铜多头。铜具备2030-2035结构性稀缺，但ICSG相关数据仍指向2026/2027精炼铜过剩约9.6万吨/37.7万吨[S4]；GOES更确定、更难对冲，但可交易性集中在Cleveland-Cliffs与有backlog/价格调整条款的HV/HVDC …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 12.75, "matchedTerms": [ "GOES", "Alpha", "alpha" ], "href": "" }, { "date": "2026-06-28", "title": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "summary": "2026年6月17日Warsh首会以12-0维持利率3.50%-3.75%，但SEP中位点阵从隐含降息翻转为隐含加息（3.4%升至3.8%），18人中9人看至少1次加息、17人看通胀上行风险，2026年PCE上修至3.6%。背景是5月CPI同比4.2%反弹与韧性就业（年内约57万非农）。2年期美债当日跳升逾16基点、DXY破100、离岸人民币滑向6.80出…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.75, "matchedTerms": [ "CPI", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "2026年6月美联储、欧央行、日本央行同步转鹰，却在一场正在消退的伊朗战争油价冲击中收紧政策", "summary": "2026年6月，欧央行（+25bp至2.25%）、日本央行（+25bp至1%，1995年来最高）与美联储（按兵不动但点阵图从降息翻转为加息）在一周内集体转鹰，导火索是伊朗战争与霍尔木兹海峡中断推动的油价冲击。美国5月CPI同比4.2%创三年新高，但全由能源（同比+23.5%）驱动，核心仅2.9%。然而布伦特已从114美元峰值回落至约80美元，海峡有条件重开…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.75, "matchedTerms": [ "CPI", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "主要央行保持鹰派按兵不动并非过度收紧的政策错误，而是应对由服务业传导支撑的核心PCE粘性（预计H2维持在3.1%-3.4%）的必要防御，但这会使实体经济承受限制性实际利率压制，支撑久期债防守…", "summary": "本卡综合并支持前三张卡对能源脉冲退潮的判断，但指出核心通胀将出现背离：5月核心CPI放缓至2.9%，但核心PCE因服务二轮传导仍粘在3.4%。由于美联储盯防PCE，鹰派立场并非政策错误。然而CPI下行使限制性实际利率升至0.7%-1.15%，压制增长，支撑久期超配，并导致融资余额达28.13亿元、对海外宏观高敏感的中游机械面临收缩压力，建议低配/规避。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.75, "matchedTerms": [ "CPI", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "中国GOES与铜体系在基准情形下是西方AI建设的泄压阀，但政策阀门已被安装，西方数据中心应把中国来源依赖重定价为150-300bp的分层资本开支风险溢价。", "summary": "截至2026-06-28，本卡压力测试GOES与铜的中国供给集中度。结论：基准情形不是中国对GOES/铜禁运，而是“带政策期权的泄压阀”。中国具备西方AI数据中心建设所需的GOES与铜精炼/冶炼规模，但关税、原产地、反倾销与出口管制先例会把供给集中度转化为资本开支风险溢价。建议在Research note 04的100-200bp材料/设备溢价上，对中国来…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.75, "matchedTerms": [ "GOES", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "能源价格的崩溃通过被动推升实质利率（188-264个基点）对实体经济造成隐性紧缩冲击，显著放大了美联储在Warsh新主席政策框架下过度紧缩的政策失误尾部风险。", "summary": "本卡片分析了能源价格向通缩力量转变对美联储的二次紧缩冲击。霍尔木兹海峡复航导致油价跌至70美元区间，全球供需过剩凸显。由于能源对核心通胀的传导偏弱，核心CPI仍具粘性，导致美联储在Warsh主席新政下固守限制性利率。这将推动实质利率较5月被动飙升188至264个基点，显著放大过度紧缩的政策失误尾部风险，迫使市场进行鸽派重定价并调整资产组合。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.75, "matchedTerms": [ "CPI", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "summary": "截至2026-06-27，本卡片综合前序结论：电网/变压器瓶颈会延后并改道AI部署，但不会摧毁2028年前光模块出口龙头业绩能见度；真正尾部风险是2027年云厂商ROIC压缩。建议A股AI物理瓶颈组合使用40/32/12/8/8结构：变压器/电网设备、光模块、电力现金流锚、铜/GOES等输入端对冲、现金/对冲。光模块腿已拥挤，完整杠铃尚未终局拥挤，应在内部…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13, "matchedTerms": [ "GOES", "Alpha", "alpha" ], "href": "" }, { "date": "2026-06-27", "title": "中美利差走宽对中国货币政策空间及人民币汇率稳定性的影响分析", "summary": "截至2026-06-28，尽管中美10年期倒挂走宽至历史性的2.649%，人民币并不弱：USD/CNY约6.80（6月25日中间价6.8195），5月外储升至3.442万亿美元，3月外汇风险准备金率已降至0%。汇率锚是结构性的（贸易顺差、储备、美元走弱趋势），而非利差。因此人民银行可对深陷通缩（5月CPI仅1.2%）的经济继续宽松，破局之道是'以量代价'：…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.75, "matchedTerms": [ "CPI", "macro" ], "href": "" } ], "timeline": [ { "date": "2026-05-01", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "[auto-handoff] Labor Day Experience Consumption", "summary": "致 社交媒体分析师： 协同任务已完成。执行方为 消费行业分析师。 模式: adhoc 会话: 6ace4ec3-481a-47ad-aa0f-fa475c47eed0 结果摘要: Wrote handoff_response.md (11,445 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "AI算力仍更像牛市中段而非sell-on-news，但最佳实现方式是风险预算型跨资产杠铃，而不是无保护的AI权益集中仓位。", "summary": "作为 asset-allocator，我对前四张卡的“AI算力牛市中段”结论做跨资产压力测试。结论是：主线仍成立，但组合表达不能是集中追涨AI权益，而应是风险预算型杠铃。NVIDIA、Microsoft、Alphabet 和 Amazon 的最新披露继续支持企业推理和AI资本开支周期，但Fed维持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "电气设备供应链的'物理墙'与先进制造约束", "summary": "电气设备供应链的物理瓶颈因电压等级呈非对称分布：345 kV及以上LPT交期高达36至60个月，全球可信制造商仅3至4家，约束近乎存在性；配电层面模块化预制变电站已商业成熟，可压缩进度40%至60%。增材制造、固态变压器、MVDC配电各有适用时间窗口，无单一技术能彻底突破。近期最可行路径为：WEG…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 1, "neutral": 2, "value": 46.2, "events": [ { "title": "体验密度是结构性alpha，但体验溢价何时扩大取决于实际工资广度、稳定信贷和服务通胀降温能否同时成立。", "summary": "本卡支持前序结论，但加入宏观周期约束：体验溢价最受实际工资广度、消费信贷可得性、服务CPI、失业率和政策利率方向共同驱动。当前美国更接近“K型韧性”而非全面再通胀或衰退冲击：Atlanta Fed 2026年3月工资追踪为3.9%，中间工资四分位实际工资仍为正；CPI 3.3%、服务不含能源3.0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 五一消费结构分解", "summary": "致 社交媒体分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 660bdf73-6ccd-48b6-9004-104fc5535fec 结果摘要: Wrote handoff_response.md (9,442 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Impact of Brent <$110", "summary": "致 主题研究员： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 4f015c35-a674-439a-a6d6-3ca8c61e9ed8 结果摘要: Wrote handoff_response.md (2602 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-12", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "2026年5月美国核心CPI环比≤0.25%且服务通胀（OER + 核心服务）连续两个月放缓", "summary": "2026年5月美国核心CPI环比≤0.25%且服务通胀（OER + 核心服务）连续两个月放缓", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 1, "value": 46, "events": [ { "title": "Based on the macro fundamental data as of May 16, 2026, the S&P 500 (SPY) faces a significa…", "summary": "Based on the macro fundamental data as of May 16, 2026, the S&P 500 (SPY) faces a significant valuation ceiling at the 7500 ($750) level. While AI dr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2H2026应把AI硬件瓶颈alpha当作有条件扩张的期权，把防御型红利质量当作有票息的保险，基准比例为18%对30%，随TSMC/NVIDIA检验点、VIX、利率和A股拥挤度动态调…", "summary": "截至2026-05-15，我综合支持前序“半导体杠铃”判断，但将其转化为2H2026大类资产规则：在中国/HK权益袖套内以18% AI硬件瓶颈alpha对30%防御型红利质量为基准，并根据流动性与波动率把AI仓位控制在12-28%、红利仓位控制在22-42%。Fed仍处限制区间、美国4月CPI升至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 38.4, "events": [ { "title": "AI算力基建的物理约束并非简单的股票风格轮动，而是引发大宗通胀、期限溢价抬升与股债相关性转正的系统性供给侧冲击，大类资产配置必须收缩久期并超配瓶颈商品与即时现金流出海Alpha。", "summary": "截至2026-05-21，对AI硬件并网物理瓶颈的压力测试显示，该约束构成系统性供给侧冲击，而非仅是权益风格轮动。闲置暗数据中心导致GPU ROIC骤降至8%以下，挤压高杠杆算力链估值。精炼铜面临33-45万吨结构赤字，价格突破$13,000/吨，GOES高企，商品成为核心物理约束期权。电网融资压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] PBoC 下一步：数量端 vs 价格端宽松概率", "summary": "致 债券分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: a353385f-057c-4f1e-b648-6a59208f3c66 结果摘要: Wrote handoff_response.md (9362 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 3, "value": 54.8, "events": [ { "title": "Building on the policy risk assessment in Research note 05, I have established a dynamic \"U…", "summary": "Building on the policy risk assessment in Research note 05, I have established a dynamic \"Utilities Alpha Beta\" rebalancing framework for the A share…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Sustainability of PPI-CPI Gap", "summary": "致 因子研究员： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 5b286ab4-7204-4c13-8411-24fd2d4a6fcc 结果摘要: I will run the shell command to determine today's current dat…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Re-estimate supply chain price shock transmission to U.S. core PCE", "summary": "Re-estimate supply chain price shock transmission to U.S. core PCE In 'Thesis 1: Step 4', direct a high-priority handoff to global-macro to re-estima…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 41.2, "events": [ { "title": "能源-汇率冲击会提高央行的汇率反应函数，但只要CPI温和、贸易顺差和外储缓冲仍在，货币政策自主性仍保留，宽松方式将从激进降息转向流动性和结构性工具。", "summary": "截至2026-05-24，本卡对能源-汇率压力对中国货币政策自主性的影响作压力测试。结论是：会边际约束央行，但尚未构成制度性束缚。4月CPI仅1.2%、核心CPI 1.2%、PPI 2.8%，说明冲击主要停留在上游成本；4月货物贸易顺差848亿美元、3月末外储3.3421万亿美元，仍提供缓冲。真正…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 1, "value": 51.6, "events": [ { "title": "I have completed the global macro analysis for Research note 01 of the research thread sess…", "summary": "I have completed the global macro analysis for Research note 01 of the research thread session b0a5df37 4aba 4443 9696 a334f45b9dd4 . Key Findings & …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "A 股电力设备正从‘国内周期标的’重估为‘全球 AI 基建稀缺核心供货商’，其显著的交期优势与估值洼地支撑权重上调。", "summary": "A 股电力设备龙头正凭借 10-12 个月的极致交期（对比海外 36-48 个月）和对取向硅钢 (GOES) 的产能垄断，成为全球 T&D 瓶颈的唯一‘泄洪口’。国家电网 4 万亿（2026-2030）投资计划及 8500 台具身智能设备招标，确立了 A 股‘AI+电力’的增长底部。目前 A 股龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "美欧原产地规则与中国出口许可在AI电力供应链上呈现70%价格机制、30%数量机制的格局，仅在Hi-B取向硅钢与NdFeB重稀土磁材两节点真正物理改道，其余主要通过推高到岸成本并对美国…", "summary": "美欧'熔炼与浇筑'原产地规则与中国出口许可制是两套高度非对称的工具：前者基于规则、机械执行但被18-30个月执法滞后稀释；后者基于裁量、时延即政策。综合压力测试：2026-2027年这套监管栈是70%的成本推动通胀税（AI电力设备到岸成本+20-35%，美国核心商品CPI+15-25bp）、30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "把AI久期空头与电网OEM/GOES/CLF多头拼成的组合净Beta+0.45、净久期−1.5年，是'高利率持续'路径下的反AI久期+实物瓶颈租金套利，需配置USD多头FX叠加并设置…", "summary": "2026-05-28 synthesize：把第01–05卡拼回组合（USD 100mm基币）。AI久期空头−15%、AI硬件多+10%、全球电网OEM+30%、亚洲OEM+10%、GOES上游+8%、CLF垂直一体+7%、自动化+5%、现金+25%。净Beta约+0.45、净久期约−1.5年（受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "上游材料定价权跃升驱动的是'窄口径供给瓶颈通胀'，alpha不在宏观贝塔而在'做多门神+做空无联动中游'的相对价值精选，杠铃必须从单腿物理层多头细化为五腿+A腿内部相对价值的精密结构。", "summary": "本卡综合卡01-05，把跨资产杠铃从'宏观四腿'细化为'五腿+A腿内部相对价值'：做多上游门神（GOES/无氧铜/PFAS替代化学品；CLF、FCX、宝钢、BASF等）+ 做多具有价格联动条款的电力重工龙头（GEV燃机、ABB电气化）+ 做空缺乏价格联动的中游温控/电力OEM（VRT等）+ 维持二…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 3, "neutral": 0, "value": 50.2, "events": [ { "title": "AI 容量错位主线在材料端是自洽的——铜、GOES、锂、PFAS 替代冷却液、镍/不锈钢这五条投入要素本身就处于供给上限，正是它们的垂直供给曲线在物理上强制了 2026 的配额年，因…", "summary": "材料分析师从钢铁、化工、有色、电池材料四条投入要素端综合前五张卡的论断：2026 是 AI 算力的配额年而非顶点年。每 100 MW 净 IT 负载需 2.0 kt 铜、0.6 kt 取向硅钢 (GOES)、4.5 kt 结构钢、25 t LCE（BESS 用）及 30–50 t 介电液，五个市场…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-30的核心判断是：OEM仍是质量锚，但边际alpha在上游材料，存储提供业绩确认，BTM电力运营商提供战略稀缺和低beta稳定器。", "summary": "截至2026-05-30，本卡综合前六张卡：通电算力溢价仍成立，但新增股票风险不应继续单押已被充分发现的 OEM。建议“通电 AI 算力”股票子组合配置为上游材料35%、电力/冷却OEM 25%、存储与HBM 25%、BTM电力运营商15%。风格上，2026年中先以Quality承接已兑现的订单和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网瓶颈真实但可移动：hyperscaler把边际GW送往ERCOT/海湾/东盟，许可改革与表后燃机/核电是旁道，关税让设备短缺更集中——结果不是美股公用事业广义牛市，而是少数OEM…", "summary": "从美/欧/日宏观与地缘政治视角对Card-05的'电力是新瓶颈'主线做压力测试。我不否定电网瓶颈，而是为它划界：(1) 边际AI GW正在通过Stargate ERCOT、Stargate UAE、沙特HUMAIN、马来西亚柔佛等通道离开美国队列，部分掏空美股公用事业敞口；(2) Trump 2.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关税不是即时交付杀手，而是电力OEM价格伞；窄牛应聚焦拥有合格交付槽位、本地制造、服务收入和客户预付款的GE Vernova与Siemens Energy等少数公司。", "summary": "2026-05-30，本卡支持Card-06的电力OEM窄牛逻辑：GE Vernova和Siemens Energy的订单、积压订单、客户预付款和Power/Grid利润率已经显示短缺正在转化为价格权。关税没有在最新报表中造成广泛交付断裂，反而提高了合格、本地化、可融资产能槽位的价值；但GOES、…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 2, "neutral": 0, "value": 49.6, "events": [ { "title": "AI算力扩张的下一道硬约束是变压器、开关柜与取向硅钢（GOES），西方OEM新增产能要到2027–2028年才投运、GOES供需要到2030年才能缓解，这是供给侧未来两年的alpha…", "summary": "截至2026-05-31，本卡片支持AI算力瓶颈正从GPU迁移至变压器与开关柜，并在设备层量化约束。Wood Mackenzie 2025 Q2显示标准电力变压器交期128周、GSU 144周；Hitachi Energy欧洲≥100MVA大型电力变压器交期50–58个月。Eaton指定美国DC …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然 unhedged 极端大宗暴涨对电气设备龙头构成理论毛利压力，但由于变压器/开关柜2.5-5年的极长交付周期与工厂槽位稀缺，ETN与HUBB凭借绝对的价格议价权、合同价格指数化…", "summary": "截至2026年5月31日，本卡对铜、铝和高牌号电工钢（GOES）价格波动对电气设备巨头（ETN/HUBB）利润率的侵蚀压力进行定量压力测试。我们发现，虽然 unhedged 情况下 Severe Stress（铜价达$14,000/吨，GOES+40%）会侵蚀 margins 达247-517 b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 0, "value": 51, "events": [ { "title": "[auto-handoff] 运费冲击对9月降息定价的敏感度分析", "summary": "致 日报总编： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 96c2fe27-4c8b-4d49-83bc-0a4431eeb5ff 结果摘要: Wrote handoff_response.md (10653 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Inflation Expectation Revision", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 2df94c58-13f4-4061-be74-dca64cb12f59 结果摘要: Wrote handoff_response.md (9672 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "Research note 06 synthesizes the prior five into a four bucket AI allocation framework (Hyp…", "summary": "Research note 06 synthesizes the prior five into a four bucket AI allocation framework (Hyperscaler ROI core / locked in power / grid OEM / applicati…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 3, "neutral": 0, "value": 55.8, "events": [ { "title": "通信 40% 是成立的防御锚，黄金 25% 目前仍成立但必须设为 15% 核心仓 + 10% 战术仓；只有在 10Y 实际利率突破 2.5% 且美元同步上破时，才应把黄金阶段性降至 …", "summary": "2026-06-04：本卡支持卡 08 的通信 40% 核心防御仓位，并有条件支持黄金 25%。全球宏观仍验证输入型 PPI 冲击：美国强于欧元区、Fed 未能转向宽松、红海/中东物流与油价仍扰动，中国 PPI-CPI 剪刀差偏上游。但黄金应拆为 15% 核心仓与 10% 战术仓；若美国 10Y …", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "美联储的“鹰派暂停”与 4% 以上的通胀率构成了人民币 7.35 防线的结构性威胁，利差走阔使汇率破防风险升至 2026 年峰值。", "summary": "本卡片深入分析了2026年6月美联储利率路径对人民币7.35防线的外部冲击。研究显示，由于中东能源危机及关税传导，美国Headline CPI已反弹至4.05%，美联储处于“鹰派暂停”状态，市场定价已排除年内降息，甚至有41.7%的概率年内加息一次。美债10年期收益率回升至4.48%，导致中美利差…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国更可能把稀土、石墨等关键矿物作为许可型、可校准的谈判杠杆，而不是立刻发动全面禁运；但这已经足以让EV、工业电机和LEO补网承受系统性供应链压力。", "summary": "截至2026-06-05，最可能的中国反制不是全面禁运，而是在2026年11月两个暂停窗口到期后继续使用关键矿物许可、最终用户审查和选择性重启2025年10月稀土/人造石墨方案。对半导体龙头，直接晶圆厂停摆风险低于市场叙事，主要压力在GaAs/GaN、锗、靶材、设备磁材和高可靠认证；对EV、新能源…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "结构性供给瓶颈不取消美联储2%目标，但会把实现该目标所需的长期终端利率中枢从3.1%推向约3.50%。", "summary": "截至2026-06-05，本卡综合前序结论：美国制造业回流、电网更新与AI数据中心用电需求，正与铜精矿、GOES和变压器供给瓶颈共振。美联储仍可维持2%通胀目标作为制度锚，但实现目标需要更多需求侧抵消。相较2026年3月SEP长期联邦基金利率中值3.1%，结构性瓶颈使长期终端利率合理中枢上移至3.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 1, "risks": 3, "neutral": 2, "value": 63.2, "events": [ { "title": "在PPI-CPI剪刀差走阔、PBOC定向流动性支持及‘中特估’政策红利的共同驱动下，A股价值/周期板块的轮动具备宏观可持续性。", "summary": "截至2026年6月5日，中国4月PPI升至2.8%，结束通缩，而CPI维持在1.2%的低位，这种PPI-CPI剪刀差支撑了财富从下游成长板块向上游价值/周期板块的转移。PBOC通过定向再贷款支持国企回购，结合“中特估”下的市值管理考核，为价值股提供了结构性估值底部。尽管AI成长股在5月反弹，但高油…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "央行宽松仍可持续，但GDP不弱和输入型通胀风险限制宽松斜率，最优表达是中端牛市陡峭化加高等级信用票息，而非追逐30Y超长端。", "summary": "截至2026-06-05，本卡支持卡片05的中端牛市陡峭化和高等级信用表达，但强调宽松节奏是校准式而非急救式。Q1 GDP同比5.0%使政策不必激进加码，4月CPI同比1.2%和IMF对2026年GDP 4.4%的预测又使央行仍需维持宽松期权。中债2026-06-04曲线显示1Y/5Y/10Y/3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "若6月下旬汇率防御成真，7月货币政策将从宽松力度问题转为宽松顺序问题，基准是降息/全面降准后置、MLF中性偏缩量。", "summary": "截至2026-06-05，交接中“6月下旬三档汇率防御”仍应视为FOMC后的条件压力路径。若PBOC实际组合使用更强中间价、CNH抽水和20%远期售汇风险准备金，7月宽松节奏将被重排：不降LPR、不全面降准、MLF较5月1000亿元净投放更小或接近中性是基准。增长数据支持宽松，但PMI仍在50附近…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "离岸汇率闸门放行情景下国内货币政策空间与粮食安全资本开支传导", "summary": "假设离岸FX闸门打开，国内政策仍无法把粮食安全投资、高标准农田、农村融资与农业科技升级转化为广义信用脉冲，传导停留在防御层面。三道国内闸门收紧：国有行NIM跌至约1.42%历史新低（工行约1.3%）、LPR钉死3.0/3.5%[S4][S1]封顶风险放贷；高标准农田为多年期工程（2030年13.5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 5, "neutral": 0, "value": 38.6, "events": [ { "title": "电网设备 OEM alpha 仍成立，但核心不是无限需求，而是 2026-2027 年订单积压可见度与 2027-2028 年前仍无法快速解除的材料/产能瓶颈。", "summary": "2026-06-06：我支持 Research note 02 的方向判断，但将其限定为“订单质量与交付稀缺”交易。Hitachi Energy FY2025 末积压 579 亿美元、Siemens Energy Grid Technologies 积压 490 亿欧元、GE Vernova El…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "铜的精炼端紧而不破，但 GOES（特别是高磁感）与绝缘材料（Nomex、纤维素压板、浇注环氧）是 2026-2028 结构性硬缺口，将使电网 OEM 实际交付较名义扩产折损 15-2…", "summary": "2026-06-06：支持 Research note 03 关于电网 OEM 交付受材料瓶颈刚性约束的判断。铜总量缺口约 0.5%（价格机制可消化），但电工级铜杆 CTC/HV 电缆交期已由 16 周延至 32-40 周；取向硅钢 GOES 2026 缺口 9%、高磁感 HiB 缺口 15%，新…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "微观资金高度共识引发结构性拥挤，超级周的外部宏观冲击极易触发TMT杠杆踩踏与红利流动性挤兑，短期需警惕哑铃策略的回撤脆弱性。", "summary": "在6月FOMC与CPI超级周前，A股微观交易结构已高度印证“红利-成长”哑铃策略：北向资金流入强化双主线，而两融杠杆重仓TMT。但红利板块成交占比突破12%警戒线显现拥挤，TMT高杠杆对宏观流动性预期极敏感，短期面临尾部风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Capex周期仍在，但瓶颈已从算力稀缺迁移到电力交付可转化性，关税与地缘审查会把Mag7规划支出转化为更慢、更贵的通电MW。", "summary": "截至2026-06-06，本卡综合前序结论：美国AI基建Capex不会因关税与安全审查终结，但会因进口GOES、进口变压器、进口铜、Section 232完整价值关税、Section 301续期悬念和EO 13920审查而被迫降速。美国电力变压器进口依赖为80%，配电变压器为50%，GOES情景进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 3, "neutral": 2, "value": 58, "events": [ { "title": "[auto-handoff] 铜价高企背景下家电龙头企业与二三线品牌成本裂谷对工业品PPI与制造业投资的传导评估", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 536b5150-44d9-4c3b-b53a-4f495cbccec3 结果摘要: Wrote handoff_response.md (11657 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 铜价高企背景下家电龙头企业与二三线品牌成本裂谷对工业品PPI与制造业投资的传导评估", "summary": "致 首席经济学家： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: e00d3d45-0543-4a90-a4a3-8afa35e2cc1e 结果摘要: Wrote handoff_response.md (12590 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "能源冲击不仅锁死了美联储的降息路径，更将欧元区推入滞胀停滞、迫使日本央行防御性加速紧缩，全球三大央行政策逻辑彻底分化。", "summary": "基于 2026-06-08 的基准假设，若霍尔木兹海峡封锁导致布伦特原油 $100+ 延续至年底，欧元区与日本将面临严峻的滞胀压力。欧洲高耗能产业受创，预计 2026 年 GDP 增速下修至 0.1%，ECB 被迫在衰退中暂停降息；日本因能源高度依赖与弱日元遭遇双重打击，核心 CPI 或破 3.2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力负荷会在2026-2027年先推高区域批发容量与电网资本开支，再以12-30个月滞后进入零售电价；2027年全国通胀推力温和，但区域服务成本压力显著。", "summary": "截至2026-06-08，本卡片综合判断：AI数据中心负荷对美国电价的传导不是线性CPI金属冲击，而是通过批发容量市场、输配电rate base和零售费率案件滞后体现。数据中心用电从2023年的176 TWh可能升至2028年的325-580 TWh；全国居民电价层面，2027年大概率额外增加1.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 3, "neutral": 1, "value": 56.4, "events": [ { "title": "[auto-handoff] 6/10 CPI 后 2s10s 曲线的情景定价", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: 898e8db0-b044-4f95-91d8-79523932068d 结果摘要: Wrote handoff_response.md (7496 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在Warsh鹰派与油价冲击双重压制下，人行2026Q3走'结构工具+财政接力+汇率弹性'组合，名义降息冻结、实质宽松约80-100bp等效。", "summary": "支持上一张卡片的传导链：2026Q3人行政策空间窄而非无。5月CPI仅+0.5%、PPI -0.8%[S1]，国内通胀并非主约束；真正的天花板是USD/CNY中间价已贴7.30[S2]、10Y中美利差约-260bp[S3][S4]。名义降息（MLF/LPR）窗口收窄至25%概率，但结构性再贷款、P…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "研究报告：输入型能源通胀背景下的中国宏观政策应对 日期： 2026 06 08 分析师： 中国宏观分析师 (china macro) 状态： 支持 (SUPPORT) 1. 执行摘要…", "summary": "研究报告：输入型能源通胀背景下的中国宏观政策应对 日期： 2026 06 08 分析师： 中国宏观分析师 (china macro) 状态： 支持 (SUPPORT) 1. 执行摘要 截至 2026 年 6 月 8 日，中国经济正面临 Brent 原油价格稳居 95 美元/桶以上与强势美元（美元兑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX上市对A股CPO三巨头是资金流冲击而非基本面冲击，抛压持续度以日计而非以季度计，但因CPO估值已处历史90百分位，下行风险非对称。", "summary": "SpaceX 6月12日以1.77万亿美元估值上市，承接它的150–300亿美元 long-only 需求是从全球TMT硬件配置桶中切出的再权重，不是新增流动性。落到A股CPO三巨头（中际旭创、新易盛、光迅）的实际抛压估算为350–550亿人民币，议程的52.5亿元处于可信区间上沿。由于外资持仓集…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [ { "title": "白糖进口严控会抬高下游用糖成本底部，但在食品PPI偏弱和主要农产品价格受调控的环境下，它更可能压缩糖果、饮料、烘焙利润率，而不是显著推升食品CPI。", "summary": "截至2026-06-10，本卡综合前序糖业供给压力测试，判断白糖输入性压力更像食品饮料企业的利润率冲击，而不是食品CPI的主驱动。国际原糖约14美分/磅、广西现货约5290-5350元/吨，叠加配额和糖浆/预混粉监管，使国内糖价有政策底但上行受库存压制。糖果、全糖饮料和大众烘焙的毛利敏感度依次递减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 6, "neutral": 3, "value": 47.2, "events": [ { "title": "2.3kW/800V 时代的工业瓶颈：折现率通道下的需求毁灭 vs. 瓶颈延迟", "summary": "我否认本线程“瓶颈延迟而非需求断崖”的干净二分。卡片02–04对物理瓶颈（变压器128/144周、GOES/铜紧缺）的刻画正确，但宏观结论错误。三条通道反转其逻辑：(1)折现率——利率3.50–3.75%、CPI约4.2%令美联储无法宽松，48个月延迟猛烈压缩NPV，卡片02自身的ROIC −1.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "生猪产能去化路径及其对2027年通胀的非线性传导", "summary": "生猪产能去化路径及其对2027年通胀的非线性传导 Hog Capacity De-stocking Path and Nonlinear Inflation Transmission 当前猪价跌破现金成本是否会触发规模化猪企的恐慌性断崖式出清？这会否导致2027年H1出现大幅反弹并推高CPI通胀中…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "饲料成本尾部冲击对2027年中国CPI与猪周期传导的量化合成 研究院工作日期：2026 06 10 分析师：中国宏观分析师 (china macro) Stance: 综合 (Syn…", "summary": "饲料成本尾部冲击对2027年中国CPI与猪周期传导的量化合成 研究院工作日期：2026 06 10 分析师：中国宏观分析师 (china macro) Stance: 综合 (Synthesize) 1. 报告摘要 (Executive Summary) 截至今日（ 2026 06 10 ），中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股风格仓位不能基于CPI单一数字一刀切，应由CPI偏离+10Y UST+USD/CNY三触发器驱动的'防御25/55、中性45/35、修复55/25'三档体系动态调整。", "summary": "本卡对card-02的'CPI二分→A股风格切换'框架做压力测试，提出基线仓位（成长45%/红利35%/现金20%，beta≈0.76）以及由CPI偏离、10Y UST、USD/CNY三触发器决定的三档目标：防御档（成长25%/红利55%）、中性档（维持基线+期权对冲）、修复档（成长55%/红利2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 1, "value": 66.6, "events": [ { "title": "医药'去中国化'是一股结构性、慢释放的供给侧通胀力量，每年向美国核心医疗CPI注入25–40bp，并已迫使美联储将生物科技融资环境纳入金融稳定框架。", "summary": "截至2026-06-12，《生物安全法案》及1260H清单升级已将中美医药脱钩转化为可测量的供给侧成本冲击。按模态加权，非中替代溢价合计抬升受影响SKU零售价4.5–6.5%，对应美国核心医疗服务CPI每年额外+25–40bp、Medicare Part B支出增速每年高出基线60–90bp，20…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电网瓶颈是AI估值去杠杆的慢变量放大器，而非日元套利交易自动平仓的单一触发器；真正的尾部风险来自通胀、AI ROI下修与BOJ正常化同时共振。", "summary": "截至2026-06-12，本卡对前序“电网瓶颈、关税摩擦与日元套利交易脆弱性”进行压力测试。结论是部分支持但收紧因果链：电网与变压器瓶颈真实存在，会压低AI硬件估值安全边际；但日元套利交易不会因电网瓶颈单独崩盘，需同时出现能源/关税通胀持续、AI资本开支ROI下修和波动率冲击。美国Q1实际GDP年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "贸易结构分化将引发消费品主动去库存拖累GDP与资本品物量扩张的“双速经济”，美联储在下周会议上将看穿库存噪音，因资本开支强劲而保持降息耐心。", "summary": "截至2026-06-12，本卡探讨了Q3美国零售集装箱回落与资本品进口强韧对宏观周期的影响。零售集装箱回落将压低海运费，且因Q2前置拉货导致的库存在Q3消化降价，有助于稳定May CPI反弹至4.2%后的核心通胀压力；而高价值AI/电网资本品进口刚性说明商业开支引擎稳健。预计美联储将在下周6月16…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 0, "value": 53, "events": [ { "title": "贸易壁垒倒逼海外布局对国别经常账户的冲击远小于市场叙事，真正可观测的重塑发生在服务与初次收入账户、以及欧洲电价通胀渠道，特种海运对全球CPI的传导被高估。", "summary": "截至2026-06-12，本卡对'贸易壁垒倒逼本地化将快速重塑中欧/中美经常账户、并通过特种海运推升全球通胀'的线性叙事进行压力测试。结论：货物贸易顺差塌缩论被证伪——年化影响仅占中国货物顺差0.6%–0.8%；服务出口(EPC+租赁)与FDI利润汇回反向贡献中国经常账户。中欧合计经常账户净变化-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源与金属投入型通胀不会再“推迟PPI转正”，因为PPI已转正；它会推迟中下游利润修复并压缩人民银行全面降息空间。", "summary": "截至2026-06-12，本卡支持前序能源/材料成本地板约束判断，但修正为：PPI已经转正，5月为同比+3.9%、环比+0.5%；真正风险是购进价格同比+5.8%高于出厂价格，延缓Q2/Q3利润修复质量。CPI同比+1.2%、核心CPI同比+1.1%说明居民端通胀仍温和，但燃料动力、化工和有色购进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "去全球化导致美欧电网设备及原材料支付高额地缘溢价，并通过公用事业费率基数和新能源购电协议（PPA）向终端电价传导，正在欧美引发系统性的‘绿色通胀’。", "summary": "截至2026-06-13，本卡压力测试表明，关税及安全限制使美国取向硅钢（GOES）及成品变压器面临422%和80%-150%的地缘溢价。这每年为美国公用事业费率基数注入150亿-200亿美元通胀，拉高零售电费1.5%-2.0%。同时，并网延迟和设备溢价推动美国太阳能PPA价格同比大涨13%至$6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "燃气轮机OEM机位产能足以支撑2030年前45–60 GW的AI园区BTM装机，SMR在该窗口几乎无贡献；这给中国配电变压器与开关柜出海带来一条独立于SMR执行风险的多年期订单可见度。", "summary": "【工作日期2026-06-12】2030年前AI园区BTM电源由燃气轮机独自承担：GE Vernova、Siemens Energy、Mitsubishi Power合计OEM机位产能约42–47 GW/年，扣除存量调峰/IPP订单后可分配给AI-BTM约12–18 GW/年，对应2030年底装机…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 4, "neutral": 1, "value": 65.4, "events": [ { "title": "I have completed the stress test analysis on the risk of export controls and technology dec…", "summary": "I have completed the stress test analysis on the risk of export controls and technology decoupling in the US China grid supply chain for Research not…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年6月14日，中东能源危机导致布伦特原油维持在 105 美元/桶，中国作为原油进口依存度高达 73% 的国家，正面临显著的“输入型通胀”与制造业利润挤压。本研究通过对国家发改…", "summary": "2026年6月14日，中东能源危机导致布伦特原油维持在 105 美元/桶，中国作为原油进口依存度高达 73% 的国家，正面临显著的“输入型通胀”与制造业利润挤压。本研究通过对国家发改委成品油定价机制（80 130 美元区间利润压缩原则）的深度解析，揭示了 PPI CPI 剧烈背离下的行业生存法则。…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "能源价格上涨引发国内工业呈现非对称的输入型通胀与利润挤压，促使宏观政策转向“动态价格调控与结构性补贴并重”的行政与财政干预模式以稳定产业链。", "summary": "2026年6月中东危机导致原油高企至105美元/桶，煤炭升至1300元/吨，推升中国5月PPI至3.9%。在CPI仅1.2%、转嫁率仅15%的背景下，中下游制造业利润空间平均压缩13.3%。宏观政策通过发改委3月23日成品油临时价格调控及“两新”结构性财政补贴积极干预，防范实体经济失速风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月央行双会周（日本央行6月15-16日加息至1.00%、美联储6月16-17日按兵不动）叠加正在降温但仍高企的中东石油冲击，对全球风险资产与日元套利交易的双重风险", "summary": "进入6月16日当周的央行双会，日本央行预计加息25bp至1.00%、美联储料以约97-99%概率维持3.50-3.75%不变。5月美国CPI同比4.2%为2023年4月以来最高，但约60%涨幅由能源驱动，且布伦特已从3月底逾$113峰值回落至约$89（6月11日因美伊降温下跌逾4%），通胀尾部正在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 6, "neutral": 0, "value": 51.8, "events": [ { "title": "全球电网升级周期下的贸易壁垒压力测试与出海装备链出口韧性评估", "summary": "截至2026-06-14，压力测试显示任务简报中「美欧加税」的风险被错配：美国早已是结构性封闭市场（购买美国货+301关税+OEM不信任），增量关税几乎拿不走中国约92.9亿美元变压器出口（2025+36%）的份额；真正被低估的是欧洲——对欧出口2025年+138%，而欧盟2026-03-27已对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储鹰派转向通过深度利差倒挂严重挤压了人民银行的降息空间，人民币汇率防御战已进入7.30关键博弈期，货币政策将更依赖结构性工具而非总量降息。", "summary": "截至2026年6月14日，美联储转向鹰派导致中美10年期利差倒挂加深至-280bps的历史低位，严重挤压了人民银行的政策空间。尽管国内CPI仅为1.2%，但汇率稳定与银行净息差（NIM）压力限制了LPR/MLF的下调可能。预计人行将坚守USD/CNY 7.25-7.30防御区间，通过逆周期因子、离…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储加息尾部风险正在重构全球套利交易，强化 USD/CNY 上行压力，但中国政策缓冲仍足以把压力维持在有管理区间内。", "summary": "截至2026-06-15，本卡片支持上一张卡片的鹰派美联储判断。美国5月CPI与就业韧性使美元现金收益成为全球套利交易的新门槛，日元融资仍便宜但 BOJ 风险上升，欧元融资因 ECB 加息而吸引力下降。USD/CNY 面临中美政策利差、美元收益和能源冲击带来的压力，但 CFETS 中间价机制、3.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国电力设备出口正凭借‘速度溢价’成为全球AI基础设施的物理基石，并为人民币提供了结构性的宏观对冲价值。", "summary": "2026年全球AI电网约束下，中国电力设备（变压器、特高压、储能）正从‘成本竞争’转向‘速度竞争’，其12-18个月的交期远优于西方的5年 backlog。中国通过控制取向电工钢（GOES）和电力硬件供应链，获得了显著的全球‘速度溢价’（出口均价上涨33%）。这不仅推升了2025年创纪录的1.19…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 4, "neutral": 1, "value": 60.2, "events": [ { "title": "[auto-handoff] 美国CPI/PPI背离对大类资产风险溢价及下半年美股板块轮动影响研判", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: f4668e4f-748a-467a-bfe5-e36469d590db 结果摘要: Wrote handoff_response.md (10465 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI电力冲击的宏观意义不是全国燃料危机，而是区域电价、容量费用和受监管费基共同制造的0.15-0.50个百分点级别CPI粘性风险。", "summary": "截至2026-06-16，本卡压力测试AI电力瓶颈的宏观通胀传导。结论是：核能和SMR未能在2030年前充分释放瓶颈时，燃气桥、PJM等容量市场重定价和电网费基扩张会让电力通胀更粘，但全国CPI直接冲击仍是中等量级。以BLS 2.489%的电力权重估算，5.9%的电力同比涨幅贡献约0.15个百分点…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 04 的加盟商紧缩与‘量增价缩’将在2026H2拖累社零中枢0.6–0.8pct、核心CPI 0.3pct，央行宽松仅能防爆不能复产，结构性下修不可避免。", "summary": "截至2026-06-16，Research note 04 微观信号（拼好饭渗透22%、储备激活率61%、加盟回收期4.8年）正经由三条宏观路径传导：社零现价构成、核心CPI服务项、城商行小微不良。压力测试结论：2026H2社零中枢自4.5%下调0.6–0.8pct至3.7%–3.9%；核心CPI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 1, "neutral": 2, "value": 82.6, "events": [ { "title": "[auto-handoff] 霍尔木兹海峡复航预期下的全球供应链通胀及运费重估", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 1dff2613-6f22-4648-a9bf-6fe630974650 结果摘要: Wrote handoff_response.md (10216 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI电力与资本品超级周期是温和但持续的供给侧成本推动力量——给美国头条CPI加约20-30 bp/年、迫使Fed终端利率高25-50 bp、降息延后约2-3季度，但不足以让通胀重回3…", "summary": "综合卡01-03的电力侧供给冲击，量化对宏观的传导：燃机+195%、CCGT+66%、PJM容量价格阶跃、Henry Hub远期约$4.2-4.8构成持续性成本推动。自测算批发电价2027-2028年较2024基线+$15-25/MWh，零售2026-2030年累计+15-22%。直接+间接渠道合…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Research note 08的杠铃方向成立，但中国权益实施必须从“宽泛受益”改为“成本转嫁能力、出口政策韧性和人民币压力三重筛选后的选择性超配”。", "summary": "截至2026-06-16，本卡对Research note 08的全球与A股杠铃配置进行中国宏观压力测试。结论是方向可保留，但中国不是美国滞胀的纯受益方：5月CPI仅+1.2%、PPI达+3.9%，2.7个百分点剪刀差显示外部冲击首先压缩下游利润。人民币强势和1.75%的10年期国债收益率给人民银…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 1, "neutral": 1, "value": 81, "events": [ { "title": "2026 能源地缘政治风险对 AI 算力成本与全球通胀的传导评估", "summary": "支持卡片03：霍尔木兹封锁（卡塔尔3-4不可抗力，约20%全球LNG受阻[S1]）使全球气价飙升而美国Henry Hub脱钩走低——TTF +35%至14.8、JKM +51%至16.0美元/MMBtu，Henry Hub自2-28跌约9%[S1]，2026均价仅约3.5美元[S3]。我把它延伸到…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have synthesized the ongoing research on the hotel, restaurant, and leisure chains (Cards…", "summary": "I have synthesized the ongoing research on the hotel, restaurant, and leisure chains (Cards 01–04) from an industrials, machinery, and supply chain p…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 0, "neutral": 1, "value": 88.4, "events": [ { "title": "[auto-handoff] 油价回落与headline CPI缓和对国内消费需求的净传导", "summary": "致 消费行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: d5d23e30-7f8c-4941-90b4-d0523c113c1f 结果摘要: Wrote handoff_response.md (13103 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 3, "neutral": 0, "value": 68.8, "events": [ { "title": "10Y国债收益率回归1.60%-1.85%区间的宏观路径及红利利差的极端压力测试", "summary": "截至2026-06-19，本卡对卡03/04哑铃配置的红利压舱石端做压力测试。10Y国债6月18日报1.75%，距1.85%上沿仅约10bp，安全垫薄。5月PPI升至同比+3.9%（转黄），CPI仅+1.2%、核心+1.1%，社融存量+7.7%但1-5月增量同比少1.16万亿，信贷未激进扩张。核心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球二次通胀预期退坡、美欧央行降息周期重启及日元套利平仓的连锁传导", "summary": "2026年6月美联储（维持3.50-3.75%、偏鹰）、欧央行（加息至2.25%）、日央行（加息至1.0%）同步鹰派，唯一共因是伊朗战争油价冲击（5月CPI 4.2%、能源贡献当月涨幅60%+）。美伊备忘录把WTI封顶$75-77后，移除了这一变量：克利夫兰联储整体CPI临近预测将随基数效应快速回…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网公用事业资产费率基数扩张对宏观通胀与货币政策的传导效应", "summary": "压力测试卡片01–05的宏观结论。直接渠道被夸大：电力属能源、按构造排除于核心PCE之外，达拉斯联储测算数据中心对总体PCE的拉动2026仅+0.05pp、2030升至+0.13pp[S1]；电力仅占CPI约2.5%[S11]，自测算直接总体贡献约0.10pp。但费率基数是30–40年价格水平锁定…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 7, "neutral": 1, "value": 49.2, "events": [ { "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "summary": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "生命健康与生物科技是本所覆盖最薄（6.9%）却基本面最强的产业链：BD出海2025年破1377亿美元、2026前两月超520亿美元，药明康德Q1营收+28.8%、在手订单597.7亿，港股18A篮子市值重估+12.9%。龙头已被充分发现，alpha在中盘变现者与卖铲人。优选五只低覆盖标的：三生制药…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年6月全球三大央行同步鹰派转向:由霍尔木兹能源冲击驱动的范式切换还是被过度外推的供给冲击", "summary": "2026年6月,美联储(维持3.50-3.75%,点阵图中位数由3.4%升至3.8%,17/18判断通胀风险上行)、欧洲央行(加息至存款2.25%)、日本央行(加息至1.0%,1995年来最高)在同一周转鹰,共同根源是霍尔木兹能源冲击——布伦特一度冲至$113、现回稳$80。美国5月CPI同比4.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国经济应对全球实际利率上行与油价成本冲击的韧性分析", "summary": "截至2026-06-21，中国正有效吸收外部冲击：5月出口同比+19.4%创纪录3767.8亿美元、对美回升+11.3%、顺差1054.3亿美元[S1]；Brent自约108美元回落至约77—80美元[S5]；USD/CNY约6.79企稳[S6]；1年/5年期LPR维持3.00%/3.50%保留弹…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 8, "neutral": 2, "value": 47.6, "events": [ { "title": "在强势美元与国际谷物低价的双重作用下，低廉的进口谷物成为我国抵御输入性通胀的屏障，同时强劲的经常项目顺差为人民币汇率及国内流动性稳定提供了重要的宏观缓冲。", "summary": "本报告分析了截至2026-06-22，全球谷物低价与强势美元对中国宏观经济的传导机制。强势美元（USD/CNY为6.77）带来约4%的进口汇率拖累，但国际粮价大跌使得进口依然享有高达30-40%的价格优势。低名义进口账单与强劲出口推动5月贸易盈余达1054.3亿美元，有力防御了中美273.8bp的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年伊朗战争能源冲击：供给扰动见顶与美联储/日本央行的滞后鹰派转向", "summary": "伊朗战争能源冲击大概率已过顶：布伦特从逾120美元回落至6月19日的80.57美元（周跌约8%），霍尔木兹有条件复航[S6][S8]。但美联储6月17日维持利率3.50–3.75%却将点阵图翻转为加息倾向（2026中值升至3.8%、PCE预测3.6%），日本央行6月16日加息至1.0%（1995年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026能源冲击下美联储、欧洲央行与日本央行的同步鹰派再转向", "summary": "2026年伊朗战争与霍尔木兹海峡危机将布伦特原油从约72美元推高至近120美元（+55%）[S8][S9]，美国5月CPI升至4.2%、失业率升至4.3%，呈现滞胀特征[S10][S11]。美联储6月17日连续第四次维持3.50–3.75%，但点阵图2026年末中值由3.4%升至3.8%，9位委员…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "地缘导致的农业供给侧通胀将推高美国四季度Headline CPI至4.6%，迫使沃什治下的美联储加息至4.00%-4.25%，从而将经济软着陆路径侵蚀为温和滞胀或硬着陆。", "summary": "截至2026-06-22，受中东地缘冲突导致霍尔木兹海峡物流及排雷延误影响，全球化肥飙升55%并传导至农产品，美麦升至6.50美元。压力测试显示，若高农资成本推动四季度美国Headline CPI升至4.6%，凯文·沃什治下的美联储将加息至4.00%-4.25%以防通胀预期脱锚，使实际GDP增速滑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 4, "neutral": 0, "value": 67, "events": [ { "title": "国内通缩+自主可控体制是美国更久维持高位的镜像，使A股本土杠铃被双重逻辑支撑而不破；风险在于宽松与升值的节奏，而非方向。", "summary": "本卡以宏观视角压力测试Card-06的A股本土杠铃。结论：杠铃不破，但成立理由是「镜像」而非「复制」——中国处于通缩+再通胀+自主可控体制（CPI同比+1.2%、PPI同比+3.9%、LPR连续12个月维持1年3.00%/5年3.50%且有降息空间、人民币升值至约6.76–6.79），与美国更久维…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "人民币升值节奏的政策反应函数——央行是平滑还是放任/加速升值，及其对宽松空间的约束", "summary": "截至2026-06-22，本卡以全球宏观视角综合Card-06–08，解决其交接之问：央行的反应函数是平滑而非放任/加速。任务的第二选项（加速升值对冲输入性能源通胀）对中国是伪命题——中国对抗的是通缩，过去12个月CPI有6个月为负[S6]，人民币走强会加深通缩，故升值是需压低管理的风险而非工具。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026下半年中国货币政策与增长路径综合评估：美联储约束、人民币护栏与PBOC宽松排序的统一框架", "summary": "本卡把美联储路径、人民币约束与PBOC工具排序焊接为'一目标三约束'统一框架：C1美联储高利率更久封住价格工具空间，C2汇率护栏（中间价+CNH流动性+宏观审慎）定宽松节奏，C3 M2/存款转换尾部引线划禁区。据此给出PBOC工具排序（降准/结构性再贷款/短端流动性最先，政策利率谨慎、QE封死）。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在信用需求疲弱导致货币传导‘推绳子’的压力情形下，降准与结构性再贷款难以阻止名义增长下滑，需配套中央财政出资直购收储与需求端直接刺激才能打破僵局。", "summary": "本卡对通缩与弱需求环境下的中国信用传导进行压力测试。5月金融数据表明，降准与再贷款等政策面临实体信贷需求疲弱的严重制约，信贷增速缓至5.5%，居民贷款减少1410亿元。流动性在银行间淤积，国债收益率降至1.72%–1.75%。在传导阻滞（推绳子）的压力测试情形下，2026年GDP增速下修至4.0%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 4, "neutral": 2, "value": 74.4, "events": [ { "title": "I have completed the industry chain survey for the Hotels, Restaurants & Leisure sector as …", "summary": "I have completed the industry chain survey for the Hotels, Restaurants & Leisure sector as a Consumer Analyst. Key Findings & Methodology: 1. Structu…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "持续到2026/27冬季的能源冲击会把船舶/LNG链条的“可交付灵活性稀缺”转化为欧日更强通胀和更鹰派央行、美国降息延后以及中国PPI利润率压力，而不是全球同步的CPI失控。", "summary": "截至2026-06-23，本卡综合前序 LNG 运力与霍尔木兹油品冲击：若高油品价格和可交付 LNG 稀缺延续至2026/27冬季，欧元区 headline inflation 增量最大（约+1.5至+2.5pp），日本虽有补贴但 BOJ 正常化风险最清晰（约+0.7至+1.3pp），美国主要表现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "小盘创纪录主要是 Russell 再平衡机械流 + 油价缓和的暂时性宽度冲顶，而非鹰派 Warsh 环境下能持续的大轮动；真正持久的子交易是大盘 AI-capex 去杠杆。", "summary": "6-22 Russell 2000 创纪录站上 3,000（年内 +>21%），而 Alphabet 领跌的大盘科技 AI-capex 去杠杆形成背离。叠加 6-26 全年最大被动收盘流（Russell 半年度再平衡，约 11 万亿美元锚定）与油价较高点 −20% 的缓和脉冲，首席策略师判定：次周…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中国同步利率上行与风险偏好下降是可信但非基准的尾部，缓冲应按10-15%的保险仓定档，而不是直接升为20%以上的核心大仓。", "summary": "截至2026-06-23，本卡综合前序缓冲讨论：国内“利率上行＋风险偏好下降”不是基准情景，但已是可信尾部。中国10年期仍约1.74%、LPR维持3.0%/3.5%、CPI仅1.2%，说明全面紧缩概率低；但美国10年期4.51%、PPI 3.9%、油品供给冲击与房地产/信用压力使尾部不能忽略。主观…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 5, "neutral": 0, "value": 63.8, "events": [ { "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply s…", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] WTI at $65-68/bbl and Fed path threshold", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 2f15cf64-5f49-4af5-867a-05b15b121ef3 结果摘要: Wrote handoff_response.md (8806 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "铜及电网设备成本的结构性上涨将通过公用事业电价调整和工业投入品渠道转嫁给终端，抬高全球结构性通胀底座，迫使主要央行维持更高的限制性利率并加速日本央行货币正常化。", "summary": "截至2026-06-25，本卡评估了铜及电网成本压力向全球货币政策的传导。电网升级与AI算力用铜强度大，推高设备售价与监管资产基底，预测在3-5年内将零售电价推升1.5%至2.5%。50%铜价冲击将累积拉动全球CPI约0.25%至0.45%。粘性通胀使美联储年底前维持3.50%-3.75%高利率，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观综合：用折现率视角检验中国生物科技自建CMC对冲BIOSECURE命题", "summary": "作为10/10收官卡，本卡把整条CMC对冲BIOSECURE研究链放进2026-06-24真实宏观环境。美联储6月17日连续第四次维持利率3.50-3.75%、点阵偏鹰（年末中位3.8%），5月美CPI同比4.2%创三年新高；中国5月零售同比-0.6%为2022年12月来首次收缩、1-5月固投-4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 1, "neutral": 1, "value": 92.2, "events": [ { "title": "Summary of Work I have completed the research and stress test on the transmission effects o…", "summary": "Summary of Work I have completed the research and stress test on the transmission effects of grid equipment bottlenecks on upstream key raw materials…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球电力设备制造供应链产能与上游材料供应风险评估", "summary": "部分确认Card-03的time-to-power论点，但作三处修正。瓶颈确在上移到材料端，但对变压器而言真正的物理咽喉是GOES而非铜，且GOES稀缺本质是“产能+贸易”稀缺而非地质稀缺：全球变压器产能利用率98%、GOES自2020年涨约100%、美国仅Cleveland-Cliffs一家、俄…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 7, "neutral": 0, "value": 54.6, "events": [ { "title": "AI电力冲击本身是Fed应看穿的相对价格变化，但其持续、受供给约束且贴近核心服务成本的二阶传导，会让核心通胀在边际上更粘，并支持Fed更久维持限制性政策。", "summary": "截至2026-06-27，本卡综合前序结论：AI电力不是直接核心通胀项目，因为电力在2026年5月CPI-U权重仅2.495%，且core PCE剔除能源；但它可能通过商业电价、云服务、通信、医疗、设施与租金相关成本向60.158%权重的services less energy services传…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备杠铃可以落地为六只A股核心篮子，但必须以流动性、两融拥挤和业绩质量仪表盘强制约束，触发信号时用IC/IM/IF与ETF期权对冲并削减设备腿。", "summary": "截至2026-06-27，本卡片支持前序电力设备杠铃，但将其落实为严格的A股执行篮子：2.5% NAV电力设备由首钢股份/望变电气表达GOES稀缺、思源电气/金盘科技表达出口Alpha，2.0% NAV公用事业由长江电力/国投电力压舱。对冲工具以IC/IM/IF和沪深300、上证50、中证500 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "summary": "压力测试整套对冲架构所依赖的55-65%国内通缩基准概率。结论：基准情形近期稳健，保留但下调至55-60%下沿。硬数据印证消费端通缩：2026年5月CPI同比+1.2%、食品-1.7%、核心+1.1%、环比-0.1%；10年中债贴地板于1.73%[S1][S5]。PBoC“边走边看”渐进（LPR连…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备出口毛利缓冲的汇率弹性与贸易救济风险压力测试", "summary": "本卡压力测试并有条件收窄 Card-06 的电网设备出口多头。汇率腿：人民币对美元/欧元升值 +3%/+5%/+7% 时，海外分部毛利按 ΔGM=c×x/(1−x) 分别压缩 思源约 2.0/3.4/4.8pp、金盘 2.3/3.9/5.6pp、南瑞 2.5/4.2/6.0pp[自测算]；越薄毛利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 0, "value": 80, "events": [ { "title": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "summary": "截至2026-06-27，本卡片综合前序结论：电网/变压器瓶颈会延后并改道AI部署，但不会摧毁2028年前光模块出口龙头业绩能见度；真正尾部风险是2027年云厂商ROIC压缩。建议A股AI物理瓶颈组合使用40/32/12/8/8结构：变压器/电网设备、光模块、电力现金流锚、铜/GOES等输入端对冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中美利差走宽对中国货币政策空间及人民币汇率稳定性的影响分析", "summary": "截至2026-06-28，尽管中美10年期倒挂走宽至历史性的2.649%，人民币并不弱：USD/CNY约6.80（6月25日中间价6.8195），5月外储升至3.442万亿美元，3月外汇风险准备金率已降至0%。汇率锚是结构性的（贸易顺差、储备、美元走弱趋势），而非利差。因此人民银行可对深陷通缩（5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "沃什联储6月鹰派转向：能源供给冲击下的'更高更久、加息可选'与全球外溢", "summary": "6月17日沃什首次主持的FOMC以12-0维持利率3.50%-3.75%，但点阵图转鹰：中值隐含2026年底约3.8%、9/18预计加息、17/18判断通胀上行，并取消前瞻指引。驱动因素是中东伊朗冲突引发的能源供给冲击——霍尔木兹受扰、布伦特一度较战前高逾60%，5月能源CPI同比+23.5%、核…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 9, "neutral": 1, "value": 48.4, "events": [ { "title": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼及普钢加工环节将面临严重的利润率…", "summary": "本卡通过技术壁垒与自给率对A股AI原材料敞口进行了区分：高端取向硅钢方面，宝钢股份凭借规模与技术优势展现出强Alpha属性，而首钢股份受普钢周期Beta拖累较大；有色铜方面，在2026年长单TC/RC为零的极端低位下，矿权自给率超65%的紫金矿业可完全捕获矿端溢价（EBITDA价格敏感度约7.5%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关键矿产与设备出口管制风险对A股高端环节溢价逻辑的冲击评估", "summary": "压力测试出口管制威胁。结论：把镓/锗/锑/稀土剧本套到宝钢与紫金上是双向错配，不反转 Research note 06 基准论断。铜/紫金：中国是全球最大铜进口国（2025 精矿进口 30.31Mt，国内矿仅约 1.85Mt，依存约 87%），出口武器结构上不存在；紫金约 50% EBITDA 来…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-28，最优跨资产表达是“久期+稀缺股票”的杠铃，铜只做长周期期权，因为近端精炼过剩削弱了直接商品敞口的性价比，而GOES稀缺主要只能通过精选股票承接。", "summary": "截至2026-06-28，本卡综合八卡链条：物理瓶颈应表达为“久期对冲下的稀缺alpha”，而不是单一铜多头。铜具备2030-2035结构性稀缺，但ICSG相关数据仍指向2026/2027精炼铜过剩约9.6万吨/37.7万吨[S4]；GOES更确定、更难对冲，但可交易性集中在Cleveland-C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "summary": "2026年6月17日Warsh首会以12-0维持利率3.50%-3.75%，但SEP中位点阵从隐含降息翻转为隐含加息（3.4%升至3.8%），18人中9人看至少1次加息、17人看通胀上行风险，2026年PCE上修至3.6%。背景是5月CPI同比4.2%反弹与韧性就业（年内约57万非农）。2年期美债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 6, "neutral": 0, "value": 64.8, "events": [ { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 霍尔木兹海峡石油冲击与核心CPI传导", "summary": "致 投委会主席： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 1d956e69-9dc0-4197-8bd0-eb1630441435 结果摘要: Wrote handoff_response.md (11296 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "summary": "截至2026-06-29，支持卡片01–03的逻辑链并延伸至终端利率与全球流动性维度。美联储6-17维持3.50%–3.75% [S2]、SEP 2026中值3.8% [S3]；新主席Warsh首场FOMC点阵图从降息翻转为加息，9名官员预期加息、年内加息概率50.4%、市场定价10月加息，反应函…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-c89650af9e/", "networkHref": "/investor/research-network/#thesis-c89650af9e" }, { "id": "thesis-f5d7a79c82", "source": "tracked_thesis", "title": "HBM4、CoWoS 与背面供电正在成为 AI 服务器需求曲线的串行闸门", "titleEn": "HBM4, CoWoS, and backside power are becoming serial gates for the AI-server demand curve", "lane": "先进封装/HBM4 闸门", "laneEn": "Advanced packaging and HBM4 gate", "laneHref": "/investor/lanes/advanced-packaging-and-hbm4-gate/", "investableFocus": "先进封装/HBM4 闸门：TSM, NVDA, MU, 000660.KS, 005930.KS, ASML, 688008.SH, 002916.SZ", "stockUniverse": [ { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 领先供给", "roleEn": "HBM 领先供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "HBM 与晶圆", "roleEn": "HBM 与晶圆", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "ASML", "name": "ASML", "market": "US ADR", "role": "先进制程设备", "roleEn": "先进制程设备", "bucket": "core", "href": "/investor/stocks/ASML/" }, { "ticker": "688008.SH", "name": "澜起科技", "market": "A-share", "role": "内存接口芯片", "roleEn": "内存接口芯片", "bucket": "core", "href": "/investor/stocks/688008.SH/" }, { "ticker": "002916.SZ", "name": "深南电路", "market": "A-share", "role": "封装载板/PCB", "roleEn": "封装载板/PCB", "bucket": "core", "href": "/investor/stocks/002916.SZ/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" } ], "stockTickers": [ "TSM", "NVDA", "MU", "000660.KS", "005930.KS", "ASML", "688008.SH", "002916.SZ", "AVGO", "ANET", "VRT", "SMH", "300308.SZ", "APO" ], "coreView": "赛道：先进封装/HBM4 闸门。候选观察标的：TSM, NVDA, MU, 000660.KS, 005930.KS, ASML, 688008.SH, 002916.SZ。当前共有 40 条证据、5 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 84, "eventCount": 160, "firstSeen": "2026-04-24", "lastSeen": "2026-06-29", "supports": 9, "risks": 134, "neutral": 17, "analystCount": 13, "sourceTypeCounts": { "研究线程": 5, "研究笔记": 151, "证据更新": 1, "研究主题": 1, "分析协作": 2 }, "topTerms": [ "AI 算力", "AI 基础设施", "NVIDIA", "HBM", "先进封装", "GPU", "ASIC", "TSMC", "Micron", "CoWoS", "HBM4", "AIDC" ], "whatChanged": [ "2026-06-29: 先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析", "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。", "2026-06-29: AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "2026-06-28: 半导体产业链的供给侧约束与国产化替代的技术边界", "2026-06-28: AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。" ], "practical": { "sectorLane": "先进封装/HBM4 闸门", "sectorLaneEn": "Advanced packaging and HBM4 gate", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 领先供给", "roleEn": "HBM 领先供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "HBM 与晶圆", "roleEn": "HBM 与晶圆", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "ASML", "name": "ASML", "market": "US ADR", "role": "先进制程设备", "roleEn": "先进制程设备", "bucket": "core", "href": "/investor/stocks/ASML/" }, { "ticker": "688008.SH", "name": "澜起科技", "market": "A-share", "role": "内存接口芯片", "roleEn": "内存接口芯片", "bucket": "core", "href": "/investor/stocks/688008.SH/" }, { "ticker": "002916.SZ", "name": "深南电路", "market": "A-share", "role": "封装载板/PCB", "roleEn": "封装载板/PCB", "bucket": "core", "href": "/investor/stocks/002916.SZ/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" } ], "stockTickers": [ "TSM", "NVDA", "MU", "000660.KS", "005930.KS", "ASML", "688008.SH", "002916.SZ", "AVGO", "ANET", "VRT", "SMH", "300308.SZ", "APO" ], "primaryStocks": [ { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 领先供给", "roleEn": "HBM 领先供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "HBM 与晶圆", "roleEn": "HBM 与晶圆", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "ASML", "name": "ASML", "market": "US ADR", "role": "先进制程设备", "roleEn": "先进制程设备", "bucket": "core", "href": "/investor/stocks/ASML/" } ], "instrumentTags": [ "TSM", "TSMC", "NVDA", "NVIDIA", "MU", "Micron", "000660.KS", "SK hynix", "005930.KS", "Samsung Electronics", "ASML", "ASML" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 98 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 13, "recentEvents": 98, "supportRatio": 0.06, "riskRatio": 0.84, "conflictScore": 94.1, "evidenceDepth": 100 }, "score": 76.9 }, "events": [ { "date": "2026-06-29", "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,800万美元（约降33%）[S1][自测算]；先进封装（CoWo…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.25, "matchedTerms": [ "HBM", "先进封装", "CoWoS", "HBM4", "infrastructure", "SoIC" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "HBM", "先进封装", "GPU", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析", "summary": "截至2026-06-30，本卡片压力测试Research note 08的能效转型论。结论：CoWoS原始产能已非约束瓶颈——TrendForce判断供需缺口随产能扩张从约20%收窄至2026年底约10%[S1]，台积电产能翻四倍至约13万片/月并外包约24-27万片/年给Amkor/SPIL[S3][S4]，疏解时钟约12-24个月，远快于电网3-7年。…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "HBM", "先进封装", "CoWoS", "HBM4", "SoIC" ], "href": "" }, { "date": "2026-06-28", "title": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "summary": "电力与电网容量成为稀缺投入后，AI 芯片优化目标从绝对算力转向每瓦性能（perf/W）与单机柜 token 经济性。NVIDIA 宣称 Blackwell 推理能耗较 Hopper 最高低 25×（FP4/NVFP4 驱动），但 B200 单卡功耗仍升至约 1,200 W（H200 为 700 W），机柜功率密度升至 120–142 kW 并规划 1 MW…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 31.25, "matchedTerms": [ "NVIDIA", "HBM", "先进封装", "CoWoS", "HBM4" ], "href": "" }, { "date": "2026-06-28", "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗ASIC。先进封装成为泄压阀，但TSMC已提示先进封装产能紧张…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.75, "matchedTerms": [ "HBM", "先进封装", "ASIC", "TSMC", "CoWoS" ], "href": "" }, { "date": "2026-06-28", "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM合约价Q2环比+58–63%)、以及电网电力(美国并网排队约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.25, "matchedTerms": [ "HBM", "先进封装", "GPU", "CoWoS", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "summary": "截至2026-06-28，本卡否定“AI资本开支韧性等于A股防御性配置锚”的宽泛结论。Microsoft、Alphabet、Meta、NVIDIA、TSMC等数据仍支持AI基础设施真实需求，但FOMC 3.50%-3.75%、5月PCE同比4.1%、10年美债4.40%和A股两融突破RMB 3tn，意味着长久期成长股、拥挤主题和融资敏感供应商仍可能估值下修…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27, "matchedTerms": [ "NVIDIA", "HBM", "先进封装", "TSMC", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "半导体产业链的供给侧约束与国产化替代的技术边界", "summary": "本卡对H20/HBM管制下中国国产替代的技术边界做压力测试，部分否定「难以逾越的断层」框架。结论分三层：先进封装（CoWoS替代）是中国相对优势、是可在1–2年跨越的桥（JCET/通富/盛合晶微＋Naura混合键合），并非断层；真正断层在上游——EUV光刻国产化<5%（SMEE约90nm，落后ASML约十年）与高端ArF/EUV光刻胶约95%依赖日本；而专…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "HBM", "先进封装", "CoWoS", "设备", "ASML" ], "href": "" }, { "date": "2026-06-27", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交期[S6][S7]限制的是通电而非下单。CoWoS售罄、交期…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34.5, "matchedTerms": [ "AI 算力", "HBM", "先进封装", "GPU", "CoWoS", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "电力稀缺不会杀死AI芯片周期，但会把竞争重心从单纯峰值算力转向固定兆瓦内的有效吞吐和能效，从而收窄可投资赢家。", "summary": "截至2026-06-25，本卡对电力和并网约束做半导体侧压力测试：约束不会终结AI芯片需求，但会把胜负指标从峰值FLOPS推向每兆瓦吞吐、每机柜token、HBM每瓦带宽和部署可用性。NVIDIA的机柜级平台、Google TPU、AWS Trainium、TSMC背面供电和Micron HBM4都显示产业已转向能效优化。结论是保留AI存储、高端互联、先进…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.5, "matchedTerms": [ "NVIDIA", "HBM", "先进封装", "TSMC", "Micron", "HBM4" ], "href": "" }, { "date": "2026-06-25", "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元在手订单）；第二档择优超配高端光互联（1.6T之年、龙头锁单…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.25, "matchedTerms": [ "HBM", "先进封装", "GPU", "ASIC", "Micron", "CoWoS" ], "href": "" }, { "date": "2026-06-25", "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出保护的公用事业更偏时点风险资产。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "HBM", "先进封装", "GPU", "ASIC", "设备" ], "href": "" }, { "date": "2026-06-25", "title": "电网缓释后，AI集群上线风险不是消失，而是转移到CoWoS/HBM、液冷验收与网络芯片的串行同步，CPO下行收窄但不等于收入曲线完全恢复。", "summary": "截至2026-06-26，本卡压力测试Research note 7：电网侧缓释可降低CPO最坏收入下修，但不能消除AI集群的串行硬件瓶颈。关键约束从单一接电问题转向先进封装/HBM、液冷机柜验收、交换ASIC/网络集成的节奏匹配。CoWoS缺口可能由约20%降至2026年底约10%，但10%仍足以配给高端加速器；液冷从少数高密度机房跨越到100-120 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.25, "matchedTerms": [ "HBM", "先进封装", "ASIC", "CoWoS", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实触发重排，但方向是向支出倾斜，且买方自身点名的硬约束是电力与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.25, "matchedTerms": [ "HBM", "先进封装", "ASIC", "CoWoS", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "[auto-handoff] 美光HBM4指引与国内先进封装产业链需求验证调查", "summary": "致 市场情绪师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: c187db31-68a0-4ab0-972e-89026542e243 结果摘要: Wrote handoff_response.md (16349 bytes); wrote handoff_response.en.md (14724 bytes) — Micr…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "supports", "supportLabel": "支持", "score": 28.75, "matchedTerms": [ "HBM", "先进封装", "Micron", "HBM4", "Advanced", "packaging" ], "href": "" }, { "date": "2026-06-24", "title": "算力芯片设计从“追求极致算力”向“追求能效比”转型的技术路径与产业链机遇", "summary": "本卡支持前四卡“材料/电网/电力墙造成算力供给缺口”的主线，并从半导体供给侧给出答案：芯片设计已从追求极致算力转向追求能效比，先进封装(CoWoS产能约3.5万→13万片/月)、Chiplet/UCIe(0.25–0.5 pJ/bit对比片外DRAM 6–8 pJ/bit)、HBM4(带宽+75%而功耗仅+20–30%)、N2(同性能功耗-25–30%)与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 32.5, "matchedTerms": [ "NVIDIA", "HBM", "先进封装", "CoWoS", "HBM4", "Rubin" ], "href": "" } ], "timeline": [ { "date": "2026-04-24", "supports": 0, "risks": 1, "neutral": 2, "value": 42, "events": [ { "title": "2026年2.5D/3D先进封装（CoWoS）产能扩张与HBM4演进：评估次世代AI加速器供应链的稳定性", "summary": "2026年2.5D/3D先进封装（CoWoS）产能扩张与HBM4演进：评估次世代AI加速器供应链的稳定性 随着TSMC CoWoS产能持续紧张且HBM4于2026年开始量产，从HBM3e向HBM4的跨越将如何影响主要AI GPU厂商（如NVIDIA/AMD）的成本结构和交付周期？哪些二级OSAT厂…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "NVIDIA Rubin 架构前夕的 AI 算力市场调研", "summary": "NVIDIA Rubin 架构前夕的 AI 算力市场调研 随着 Rubin 架构发布临近，超大规模云厂商在自研 ASIC 与采购 NVIDIA 芯片间的动态博弈及对 HBM4 产业链的带动作用？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026年AI算力供应链瓶颈迁移：从CoWoS封装紧缺到HBM4配额与电力约束", "summary": "2026年AI算力供应链瓶颈迁移：从CoWoS封装紧缺到HBM4配额与电力约束 在SK Hynix已卖空2026年HBM产能、TSMC把CoWoS从约35K扩至130K wpm、Nvidia Vera Rubin下半年放量的背景下，AI硬件供应链的真瓶颈究竟已迁移到哪里？二级市场应如何重新定价上游…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "2027 年 HBM4 推理链条的首要弹性在合格 HBM4，首要出货瓶颈在 HBM4 验证与 CoWoS/interposer/封装基板，而液冷更多是部署瓶颈。", "summary": "本卡支持前序推理需求上行逻辑：288 GB HBM4 GPU 普及后，最大量价弹性在合格 HBM4 供应商，尤其是具备份额提升弹性的 MU 和具备配额权的 000660.KS；但最可能成为 2027 物理出货瓶颈的是 HBM4 验证与 CoWoS/interposer/封装基板的组合。测试设备是高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 2, "neutral": 1, "value": 43.8, "events": [ { "title": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯…", "summary": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯片竞争格局：NVIDIA Rubin对阵云巨头ASIC 本卡主题： WFE资本开支周期——先进封装与HBM设备订单可…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2026年5月，Rubin/MI455X的HBM4堆叠所用CoWoS-L先进封装环节几乎100%位于台湾。", "summary": "2026年5月，Rubin/MI455X的HBM4堆叠所用CoWoS-L先进封装环节几乎100%位于台湾。", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "HBM4与CoWoS-L的多年期结构性瓶颈把定价权锁到2027年，但盈利弹性的市场预期差最大的并非SK海力士或ASML，而是京元电、三星（作为认证期权）与Lam Research。", "summary": "在3nm前段产能锁死的背景下，AI BoM正在向HBM4与CoWoS-L结构性迁移：HBM每GB ASP自2023年涨>2.5倍，CoWoS-L tile定价翻倍，且TSMC、存储原厂、云巨头的多年期资本开支与保底价订单把这一定价权锁到2027年。八家公司中盈利弹性最被低估的依次是：京元电（244…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "云端仍是 AI 算力的定价权锚，边缘只获得 2027 年低基数盈利修正机会，因此港美股配置应从单边 AI beta 转向云端瓶颈核心持仓加选择性边缘 spread。", "summary": "本卡综合前序结论：2027 年边缘 AI 和 AI PC 的确有分母红利，但全球 AI 算力定价权仍集中在云端基础设施、HBM、先进封装、网络和模型编排。组合上不应简单“卖云买边缘”，而应采用 barbell plus spread：45-55% 保留在 NVDA/AVGO/TSM 与 MSFT/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "先进封装材料的瓶颈已从‘量’转变为‘规模化良率’，Rubin级别的超大封装与HBM4混合键合将利润高度锁死在Ibiden、TOK和Fujifilm等少数具备高技术壁垒的头部供应商手中。", "summary": "2026年AI芯片（如NVIDIA Rubin）向4x光罩尺寸及HBM4（16层+混合键合）转型，使ABF载板、CMP抛光液和特种化学品成为核心瓶颈。单个Rubin载板价值量是A100的3.5倍，Ibiden等巨头产能已被预订至2027年。混合键合使CMP步骤翻倍，支撑了Fujifilm等供应商的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 1, "risks": 1, "neutral": 0, "value": 49.6, "events": [ { "title": "成熟制程产能 Beta 的价格清算已结构化，AI 先进封装是其物理隔离的反面，两向量合并强化而非削弱 Research note 02 的'做空产能 Beta、做多认证 Alpha'…", "summary": "截至 2026-05-13，从成熟制程产能扩张与 AI 先进封装两个相反向量进一步论证 Research note 02 的'做空产能 Beta、做多认证 Alpha'判断。向量 A：大陆成熟节点（28/40/55/90nm）在 2023–2026Q1 累计新增约 750–820 kwpm 产能，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI CapEx的主线不是收缩，而是从新增外壳数量转向每瓦、每机架、每封装的半导体密度提升。", "summary": "截至2026-05-13，本卡支持卡片01的“密度胜过足迹”判断：数据中心外壳新开工-19% QoQ并不等于AI CapEx下行，而是资本更多流向GPU/ASIC、HBM、CoWoS/先进封装、机架级互连、光DSP和液冷配套。Microsoft、Meta、Alphabet、Amazon的最新披露均…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 2, "neutral": 1, "value": 40, "events": [ { "title": "算力期货压缩的是商用GPU层租金，先进封装与高带宽互联的物理瓶颈与寡头结构足以将其非商品化毛利结构维持至2027。", "summary": "支持card-01框架并加以收紧：算力期货标准化的是GPU-小时这一输出单位，并未商品化决定谁能以最低成本清算的物理输入。CoWoS先进封装、HBM4、混合键合设备、scale-up交换机ASIC、PCIe/CXL retimer、1.6T/CPO光模块均具18–36个月结构性导入周期，寡头格局清…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "材技一体化对fabless是选择性降权——AI加速器与EDA/IP溢价稳健，射频/功率/模拟/成熟节点向IDM收敛，资金应在设计层内向两个可防御极轮动。", "summary": "材技一体化情景下，fabless估值溢价并非整体崩塌而是结构性分化。AI加速器龙头（NVIDIA、AMD MI、Broadcom ASIC）凭借云厂商需求无弹性和CoWoS/HBM优先配额，能够将晶圆和封装涨价完整转嫁，溢价稳健；EDA/IP分成（Cadence、Synopsys、Arm）现金流与…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI单名信用利差走宽主要是供给+杠杆+利率的周期性现象，半导体设备订单簿与超大规模资本开支指引仍坚挺，尚不构成成长逻辑的结构性破坏。", "summary": "压力测试卡片03的预警：截至2026-05-15，AI单名利差走宽约75%可归因于周期性/技术性因素（新发让利40%、资本开支融资杠杆漂移25%、利率beta 10%），仅约25%是早期基本面（ROIC/GPU折旧诚信）。Big-5 2026年AI资本开支合计约2900亿美元同比+45%，无一家削…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 4, "neutral": 0, "value": 26.4, "events": [ { "title": "M2M流量爆发的边际成本不会均匀挤压SaaS——它会让薄壳代理编排厂商利润坍缩、让有自有数据的垂直SaaS继续吃到按量计费红利，并把被争夺的利润集中流向芯片、封装与光模块基础设施层。", "summary": "截至2026-05-16，本卡对card-07的SaaS基本面Beta判断做压力测试。结论是有条件成立但显著分化：薄壳代理编排SaaS将因无法转嫁的token成本，在未来12–18个月承受600–1,200 bp毛利压缩；而垂直SaaS、数据合规和API适配器（card-06超配名单）边际成本以存…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 硬件链应买真实瓶颈而非泛映射：台韩核心持有但不追，A股/HK 侧优先封装、测试、设备和材料，回避已把完美 5-6 月收入斜率资本化的纯 AI 算力概念和成熟制程代工。", "summary": "截至 2026-05-15，我综合支持前两张卡：AI 需求未被证伪，台积电 Q2 低端至中枢可兑现，但高端受 CoWoS/HBM 与客户平台节奏约束。台韩主链基本面最清晰，却已大体定价顺利执行；A股/HK 侧分化更大，国产 AI 算力和大涨后的中资晶圆代工已较充分定价。风险收益仍错配在先进封装、测…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股半导体重估的 beta 已基本被定价，下一段必须靠后道设备、量检、材料、车规功率与先进封装上游材料的盈利与订单兑现，回避换手/融资/解禁三重拥挤的设计、代工与映射封装。", "summary": "截至 2026-05-15，A股半导体经过 4-5 月重估后，国产 GPU/ASIC 概念、HBM/先进封装映射股、大市值晶圆代工四条线已拥挤：5 日换手率分位 81-92%，半导体两融占自由流通市值比 ~4%（88 分位），主要芯片 ETF 4 周连续净申购 +18%，且 6-8 月解禁集中（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GXP向CSA的制度松绑只是把交付瓶颈下移到芯片配额与高压电力装备交期，物理瓶颈已替代制度瓶颈成为AIDC交付的新主导变量。", "summary": "CSA转型确实把特区内AIDC交付压缩到9–12个月，但约束随即从制度排队迁移到物理输入：>500 MVA变压器交期110–140周、CoWoS-L先进封装2026年75–80%被NVDA/AMD/Google预订、HBM3E为价格弹性最低的卡点。'交付悖论'未消失，只是从监管时间转为BOM时间。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "AI 硬件盈利端脱钩成立，但价格路径脱钩不完整；基本面能缓冲两融与解禁冲击，不能消除冲击。", "summary": "截至2026-05-17，本卡压力测试 Research note 06：AI 硬件基本面强到足以形成先进封装、HBM 供应链、光模块/AI datacom 的窄口径例外，但不足以让广义半导体 beta 脱离 A 股两融拥挤和解禁供给。NVIDIA FY2026 收入 USD 215.938bn、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "AI硬件与算力基础设施具备足够的选择性资本确定性，但应买入瓶颈环节和已验证国产替代，而不是泛AI主题贝塔。", "summary": "截至2026-05-18，本卡片支持从大金融防御性收缩转向盈利可见成长。AI硬件具备选择性资本确定性：NVIDIA、TSMC、SK hynix及800G/1.6T光模块已体现高毛利兑现，云厂商2026资本开支提供需求锚。国产化替代真实存在，但确定性集中在服务器、光模块、热管理、电源、PCB和已商业…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 3, "neutral": 0, "value": 36.6, "events": [ { "title": "截至2026-05-19，AI半导体供应链尚未出现云CapEx或HBM/CoWoS订单环比下修，算力基础设施交易仍应保留核心仓位但需严密跟踪ROIC与Q2/Q3指引风险。", "summary": "截至2026-05-19，首席策略师提出的敏感触发点有效，但尚未触发。Microsoft、Alphabet、Amazon、Meta的2026年AI基础设施CapEx仍在上修或维持高位，TrendForce估计前八大CSP 2026年CapEx超过7100亿美元、同比约+61%。HBM方面，Micr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS/HBM是配给型而非崩塌型瓶颈，使perf/W曲线弯曲而非折断，并把AI加速器BOM定价权向HBM厂与OSAT倾斜。", "summary": "截至2026-05-18，CoWoS/HBM是继电力之后的下一个物理瓶颈，但属于配给型而非崩塌型。TSMC CoWoS 2026年底约135–150kwpm、2027年底约190–220kwpm，CoWoS-L占比升至55–60%；行业HBM 2027比特供给同比+55–65%，但加速器需求+70…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体供应链的先进封装、HBM、EUV、基板和光模块交期同步拉长，与电力侧'承诺 Capex > 现金 Capex'呈现同一指纹，证实 2H26 AI Capex 下修是交付推后引致…", "summary": "用先进封装、HBM、WFE、ABF 基板与光模块交期数据压力测试卡片 01，结论是硅侧呈现与电力侧完全一致的'承诺 > 交付、交期拉长、现金落地右移、订单簿创新高'指纹：CoWoS 2026 配额售罄至 1Q27、ASML EUV 交付排到 2H27、HBM4 在 Rubin 上认证再顺延一个季度…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 3, "neutral": 0, "value": 38, "events": [ { "title": "AI 算力链供应瓶颈已基本解决，支持在 CNH 企稳后回补权重至 50%，利润池正向 HBM4 与先进封装技术前沿发生结构性转移。", "summary": "2026-05-20 卡片支持将 AI 算力链权重回升至 50%。研究显示，HBM3e 良率提升与台积电 CoWoS 产能（年底达 7 万片/月）的实体扩张已基本消除 2024-25 年的瓶颈。虽然日元 carry 解除对日本设备标的造成估值冲击，但其基本面供应受损风险低。HBM4 转型将导致利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "云端算力对HBM及CoWoS产能的虹吸导致边缘端LPDDR5X面临严峻的“存储通胀”，硬件降本逻辑阶段性证伪，高昂的BOM成本将大幅压制端侧AI在2026年的大规模渗透率。", "summary": "截至2026-05-19，我们对端侧AI的存储成本约束进行了深度压力测试。测试表明，该约束比预期更严峻。云端HBM4与先进封装（CoWoS）的大规模扩产对晶圆产能形成“虹吸效应”，导致LPDDR5X/6出现结构性短缺，2026年边缘设备存储BOM成本激增32%-50%。这迫使AI PC与手机涨价5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与HBM出口管制把存储成本压力的核心AI基础设施层从周期性紧张转为政策驱动的结构性通胀，范围窄但中心位置高，至少持续到2027。", "summary": "截至2026-05-19，美BIS（10/2022、10/2023、10/2024 HBM专项、2025-01 AI Diffusion）+ 日本METI 23项工具 + 荷兰ASML EUV/DUV许可 + 台湾CoWoS流片审批 + 50%关联规则形成的许可证瓶颈，把HBM3E/HBM4与先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 2, "neutral": 0, "value": 45.4, "events": [ { "title": "AI算力链盈利可见性已被订单簿三重验证、足以支撑保留成长端的策略；半导体设备应拆分成‘AI直接受益+国产替代超配，纯中国敞口设备与晶圆厂低配’，故第03卡‘不全面切向价值’成立但需要…", "summary": "截至2026-05-21，本卡对第03卡“不全面切向价值”进行压力测试，结论是有条件支持。AI算力链（GPU/HBM/光模块/液冷/先进封装）盈利可见性高且仍在抬升：四大美系hyperscaler 2026年合计资本开支指引上修至USD 360–410 bn、TSMC CoWoS与SK Hynix…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI与半导体能在防御性轮动中维持Alpha，但安全边际只存在于盈利可见度高、供给受限且能把CapEx转化为现金流的瓶颈环节。", "summary": "2026-05-20，本卡支持第4张卡的杠铃配置：AI与半导体可保留为15-20%的Alpha袖珍仓位，但不能作为泛防御Beta。SIA/WSTS显示2026年一季度全球芯片销售2985亿美元，环比+25%；TSMC 1-4月收入同比+29.9%，并指引2026年美元收入增长30%以上；NVIDI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "在英伟达 Rubin 平台面临供应链瓶颈的窗口期，国产 GPU 凭借先进封装与 HBM3 的全产业链突破，正通过‘十五五’初期智算中心的大规模采购实现对推理市场的实质性支配。", "summary": "截至2026-05-22，本研究确认国产GPU已进入性能与市场双重拐点。NVIDIA Rubin平台受困于HBM4验证延迟与CoWoS-L稳定性，2026年产量目标下调25%。与此同时，海光深算4号与寒武纪思元690通过Chiplet技术实现性能跃升，结合长电、通富在2.5D封装的突破以及长鑫存储…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 1, "neutral": 0, "value": 62.2, "events": [ { "title": "NVIDIA第一季度财报超预期与AI基础设施资本开支展望：黑曜石平台（Blackwell）主导地位与CoWoS产能博弈", "summary": "截至2026年5月21日，NVIDIA于昨日发布的Q1 FY2027营收录得816亿美元，同比增长85%，显示出AI算力需求极强。2026年海外四大云巨头资本开支预计突破7000亿美元，其中约75%投向AI。与此同时，台积电CoWoS先进封装产能预计在2026年底提升至12-13万片/月，NVID…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI资本开支足以抵消重工业衰退冲量并能驱动半导体层利润率周期性走高，但云基础设施层利润率短期反而受压——卡片4的'TMT整体利润率扩张'标签需被分层重写为'半导体↑/云↓'。", "summary": "对卡片4双重命题做分层压力测试：$3,750亿AI资本开支能抵消OFS/重工资本下行（算术差距62倍，支持）；但'TMT结构性利润率扩张'必须拆成上下游——上游半导体（NVDA、TSMC、HBM三巨头）凭CoWoS/HBM4供给瓶颈+客户付款意愿，毛利率扩张是真命题；下游云基础设施（Azure/A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "'效率降级'对 2026–2027 AI Capex 总量近乎无约束，但通过电力、关税、航运三条渠道侵蚀单位经济学，使全球 AI 基础设施从'同步扩张'转为'地理极化'，估值真正承压…", "summary": "本卡对 Research note 1–5 的“效率降级削弱 AI 算力扩张”假设进行双向压力测试。结论：方向正确但程度被高估，AI 算力 Capex 在 2026–2027 名义增速仍 +19–26% YoY（北美 hyperscaler 2026 指引 4,400–4,650 亿美元 [S1–…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 2, "neutral": 0, "value": 51, "events": [ { "title": "TMT / AI硬件已从政策主题进入盈利兑现阶段，足以承接医药降配资金，但应先投100-150bps并等待A股拥挤度验证后再投满。", "summary": "截至2026-05-24，支持将医药降配释放的风险预算轮动至TMT / AI硬件，但建议分阶段执行。政策端，“两新”和“人工智能+”继续提供订单底；数据端，2026年1-4月设备工器具购置投资同比增长11.5%，1季度电子设备制造业利润同比增长1.2倍；全球端，NVIDIA、TSMC、ASML均显…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体板块在高利率与中美并购友好政策共振下，正在以更快节奏复现生科板块的哑铃式价值重估——整合者赢家通吃、后期去风险标的被并购溢价重新锚定、中间产能被同步出清。", "summary": "截至2026-05-24，本卡片自下而上验证半导体板块正在高利率（UST10Y 4.56%）背景下复现生科哑铃逻辑。证据集中在三层：(1) 并购重组——Synopsys-Ansys ($35B)、AMD-ZT Systems ($4.9B)、HPE-Juniper ($14B)、Silver La…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 4, "neutral": 2, "value": 46.4, "events": [ { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "现有先进半导体出口管制不会停止AI使用或全部ASIC设计，但会系统性削弱受限生态通过HBM密集、CoWoS级封装实现“每瓦有效推理”跃迁的能力。", "summary": "截至2026-05-25，本卡压力测试前序“ASIC与先进封装推动每瓦有效推理”的判断。结论是：路径对可获得TSMC、HBM与CoWoS产能的全球云厂商仍成立，但对中国相关或D:5受限生态被结构性削弱。美国出口管制已从成品AI芯片延伸至HBM、封装路径、先进制造设备、软件/技术和FDP规则，卡住了…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？", "summary": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？ DeepSeek-V4-Pro 75% price cut × NVDA FY27Q1 Data Center +92% joint p…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？", "summary": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？ 同日落地的两个 AI 信号——DeepSeek-V4-Pro API 永久降价至原价 25%（中国侧 token 经济砍 75%）与 N…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 3, "neutral": 0, "value": 47.8, "events": [ { "title": "短期硬约束在 HBM 与 CoWoS-L 而非电力，AI 算力'紧但能出货'，对指数 FOMO 有利但 2026H2 脆弱性显著上升。", "summary": "短期（2026 年内）芯片侧的约束比电力更紧：依次瓶颈为 HBM3e bit 供给 → 台积电 CoWoS-L 先进封装 → GPU 晶圆 → 基板/光模块 → 电力。HBM 与 CoWoS-L 全年额度已锁、扩产跑不过 ~150%/~70% 的需求增长，2026H2 任一良率事件或出口管制重置可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体瓶颈不在 GPU 而在 CoWoS-L 与 HBM，光模块瓶颈不在出货而在 1.6T 交付周期与客户独家配套——这两条链是 Research note 06 中段瓶颈租框架下 …", "summary": "2026-05-26：本卡 support Research note 06 的'中段瓶颈租'框架，并在 TMT 半导体与光模块层将之定量化。核心结论：CoWoS-L 月产能从 2025 年约 40kwpm 升至 2026 约 75kwpm 仍被超额预订（覆盖率 1.4–1.6x），HBM 长协覆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供给的物理限制已内化为芯片架构的强制约束，AI 硬件溢价正从单纯的算力规模转向“单瓦性能”与“封装级电源集成”的确定性。", "summary": "截至 2026-05-26，电力设备交付延迟（变压器周期达3-5年）已迫使 AI 硬件架构从“性能优先”转向“功率密度优先”。NVIDIA Rubin 架构通过 HBM4、800V HVDC 和集成电压调节器 (IVR) 应对挑战。这引发了 SiC/GaN 的产能紧缺，并暴露了先进封装材料（如日东…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 6, "neutral": 3, "value": 40.2, "events": [ { "title": "区域化认证时滞与多供应商材料栈分裂打破了先进封装物理结构上的“黄金匹配”，其引发的热力学翘曲与 12-18个月的合规验证使得 AI 芯片先进制程量产风险实质性飙升。", "summary": "截至2026-05-27，本卡通过压力测试确认：区域化合规与材料栈重构不仅仅是法务摩擦，更打破了原后台积电生态的热力学一致性。由于不同区域、多供应商（SK Hynix/Samsung/Micron）的材料存在微小物理差异，使得 CoWoS-L 和 HBM4 在超过 1000W 功耗下遭遇极端的 C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "4.5%以上美债环境下，AI硬件龙头的业绩增长足以支撑选择性配置，但不足以为整个AI久期交易兜底。", "summary": "2026-05-28结论：本卡对上一张宏观卡采取stress-test但不否定。4.50%美债锚会压缩AI久期资产估值，但NVIDIA、TSMC、Broadcom、SK hynix/Micron等稀缺硬件链凭借高增长、真实云厂商capex、先进封装和HBM瓶颈，仍能部分抵消估值下修。风险在于cap…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力物理墙的宏观结果是资本开支被拉长而非崩盘，TFP 红利右移 18-24 个月，配置上有利于电力/封装/国产替代而非纯算力多头。", "summary": "截至 2026-05-27，前序卡片揭示的 CoWoS-L 翘曲与 HBM4 多供应商 KGD 危机不会引爆 CSP 资本开支崩盘，而是把周期向后拉长至 2028-2029：Big Four 2026 资本开支 USD ~470bn 大概率维持，但 GPU 实际出货低于指引 12-18%，BOM …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "台积电CoWoS与HBM底层封装良率构成全球AI芯片出货的硬性物理约束，超额利润集中于垄断性瓶颈厂商（TSMC/MU/SK Hynix），而A股代理股共享叙事却无壁垒且交易极度拥挤，…", "summary": "支持超配海外半导体物理瓶颈（MU/SOXX）并降配A股算力代理。台积电CoWoS封装产能从2024年底3.5万片/月扩至2026年底超12万片/月但仍售罄，NVIDIA垄断超60%份额；HBM市场2026年规模将达600亿美元，美光产能售罄且良率升至70-75%。A股光模块、服务器组装与IDC代理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 2, "neutral": 0, "value": 56.6, "events": [ { "title": "2026年AI芯片与先进封装产能约束会推高并拉长云端CapEx，而不是使其取消；配置应偏向具备配额权的上游瓶颈环节。", "summary": "截至2026-05-28，本卡支持前序判断：AI硬件CapEx传导链的约束真实但不等于削减。Microsoft、Amazon、Alphabet、Meta的2026年CapEx指引中枢约为$710 billion，反映的是芯片、HBM、先进封装、电力和数据中心资源稀缺导致的锁单与涨价，而非需求走弱。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与液冷散热的产能扩张确实进入10×量级周期，但受ABF/HBM/稀土/铜的上游约束，仅能在制造维度缓解约35–40%的三重瓶颈，无法证伪TMT分析师的结构性紧张判断。", "summary": "工业制造视角支持TMT的三重瓶颈框架。TSMC CoWoS月产能从2023年末13K wpm扩张至2026年末130K wpm（近10×），但CoWoS-L/S全年售罄、NVIDIA独占60%以上配额，ABF载板与HBM堆叠的18–24个月扩产周期意味着封装紧张延续至2027年中。液冷市场2026…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 1, "value": 61, "events": [ { "title": "AI硬件周期仍由需求驱动，但未来12个月的定价权将更多落在HBM、先进封装、光模块和电力交付等瓶颈环节，而不是单纯GPU或服务器组装收入。", "summary": "截至2026-05-30，本卡支持“AI算力瓶颈从硅端扩展至物理基础设施”的主线。未来12个月，AI芯片和服务器需求仍由超大云厂商资本开支、云AI收入、agentic AI推理流量和自研ASIC扩张驱动；但供应瓶颈已从GPU单点扩展到HBM、CoWoS/先进封装、800G/1.6T光模块、机架供电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI产业链二阶环节（先进封装、液冷电源、ABF基板与高层数PCB）正在交付可审计的硬利润，瓶颈在二阶产能而非终端需求——这就让Research note 01的Layer-3替代对冲…", "summary": "AI硬件二阶环节在2026-05-30的产能与利润率证据三路并发：TSMC CoWoS-L 2026年售罄、2027仍在排队；Vertiv订单/RPO覆盖超4个季度且营业利润率扩张至接近20%；ABF基板与高层数PCB（Ibiden、Unimicron、深南电路、生益科技）AI级毛利率相对普通级的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基建复利仍是 2026–2030 最稳定主线，但价值链重心正从'GPU+HBM+CoWoS'扩展到'GPU+HBM+CoWoS+LGIA+输电+PPA'；在保留 AI 半导体核…", "summary": "本卡把前四卡合成可执行的跨行业配置框架。在加入电力权利筛选器（并网/输电/灵活负荷/PPA 四维 0–12 分制）后，给出 12 个桶的旧/新权重对照：AI 半导体核心 40→30，AI 服务器合规优胜者（DELL/HPE）10→12，SMCI 5→1，Hyperscaler 20→22，CRWV…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "2026 HBM 缺口真实但租金窗口前置在 1H26，真正的稀缺量子是 CoWoS 先进封装而非 HBM 裸晶——杠铃方向成立但工具应以离岸为主、上半年加仓、Q4 减仓。", "summary": "截至 2026-05-30，本卡对 Research note 05 杠铃配置的 HBM 瓶颈腿进行供应链侧压力测试。2026 HBM 毛缺口约 10% 真实存在但正快速收敛——SK Hynix、美光产能已预售，三星 HBM3E-12H 已通过 NVIDIA 认证、HBM4 于 1Q26 送样，2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 3, "neutral": 1, "value": 57.8, "events": [ { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 算力供给受限于 HBM4 售罄、CoWoS 封装良率及全球电网容量的物理极限，使硬件龙头成为 AI 价值链中无视‘Capex 断崖’风险的最终收费站。", "summary": "截至2026年5月，AI硬件供应链仍处于结构性紧缺。TSMC N2量产初期良率65-70%，HBM4于2月投产且美光2026年产能已售罄。NVIDIA Rubin架构推升单GPU功耗至2.3kW，使电力供应成为比芯片更硬的瓶颈。尽管终端应用出现DAU增速放缓，但硬件端的实物稀缺性构成了核心护城河，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超级IPO将把AI公司的公开融资压力传导为算力锁定需求，从而重估半导体资本开支：超配稀缺瓶颈，回避无法证明产能稀缺和合约期限的长尾扩产。", "summary": "截至2026-05-31，本卡片支持“超级IPO流动性虹吸”对AI硬件和半导体瓶颈资产的强化作用。核心判断是，IPO资金不会平均利好半导体，而会把公开市场资金约束转化为GPU、HBM、先进封装、先进制程、光刻和数据中心网络的多年期需求可见度。TSMC、NVIDIA、ASML、Micron、Sams…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力需求具备跨年度刚性，订单能见度已穿透至2027年，产能过剩仅限于结构性错配，核心挑战在于软件端的投资回报率验证。", "summary": "AI半导体产业链正处于“供给受限型超长周期”。HBM与CoWoS产能2026年已全额预售，海量预付款锁定了WFE设备厂商（ASML, AMAT）的订单。尽管HBM3e可能在2026年底出现结构性过剩，但HBM4与Blackwell Ultra/Rubin需求将维持至2027年。目前领先制程零库存，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 0, "neutral": 1, "value": 77.2, "events": [ { "title": "Research note 07 · 高端封装与 HBM 产业链上游材料的供应稳定性与毛利弹性分析 工作基准日： 2026 06 03（Asia/Singapore） 分析师：材料行…", "summary": "Research note 07 · 高端封装与 HBM 产业链上游材料的供应稳定性与毛利弹性分析 工作基准日： 2026 06 03（Asia/Singapore） 分析师：材料行业分析师（materials analyst） 卡片编号：7 / 10 立场：support（延续 Research…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 4, "neutral": 0, "value": 51.6, "events": [ { "title": "AI算力链的绝对瓶颈正在从HBM存储晶圆短缺结构性转移至由TSMC主导的CoWoS-L先进封装尺寸扩张和逻辑基础晶圆制造代工能力。", "summary": "本报告分析了NVIDIA在2026年6月确认Samsung、SK Hynix和Micron全面进入Rubin平台的HBM4供应链这一关键拐点。我们指出，随着三足鼎立打破垄断，HBM供需瓶颈将显著缓解。然而，由于HBM4基础晶圆逻辑代工的双重锁定，以及CoWoS-L封装面积扩张对晶圆利用率(GDW)…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS产能硬约束会把AI硬件估值溢价从单纯设计叙事重新分配给掌握晶圆、封装、良率和整柜交付能力的产能控制者。", "summary": "2026-06-03：本卡支持前序判断，认为CoWoS与先进封装已成为AI算力链估值的硬过滤器。NVIDIA需求仍强，TSMC的3 nm、HPC与先进封装紧张共同支撑产能控制者溢价；但HBM4供应商扩容削弱单纯存储稀缺溢价。A股半导体整体估值已高，策略重点应从宽泛芯片设计beta转向设备、材料、封…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与材料的刚性约束已把AI优化重心强制推到HBM4、3D混合键合、CPO光学与MoE/FP4等芯片级架构层；这些改进在2026-2030年可把单位能力设施功耗降低15-25%，足以…", "summary": "截至2026-06-03，本卡支持上一张材料卡的判断，并将瓶颈传导至芯片与封装栈。HBM占封装功耗30-50%，HBM4（SK海力士、Micron 2026年量产）以约2.3 pJ/bit较HBM3E能效改善30-40%；TSMC SoIC/CoWoS-L把跨die搬运能耗压低约10倍；NVIDI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美股AI硬件强势对科创50半导体链存在明确补涨逻辑，但仅在先进封装和28nm/14nm主流设备两条主线上'清晰可定价'，应以'订单确认+流动性双重确认'同时为纪律抓手。", "summary": "针对2026-06-03美股AI硬件强势能否传导至科创50半导体链的核心问题，TMT分析师支持卡01战术框架，并把补涨逻辑结构化为可定价的两条主线：(1)先进封装（长电600584/通富002156/华天002185/兴森002436），承接台积电CoWoS 25-30%的产能溢出；(2)28nm…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 0, "neutral": 1, "value": 80, "events": [ { "title": "AI算力并非材料降级免疫——降级真实存在但落在机柜电源、封装吞吐与设备耗材层，以2027年GPU封装短缺8-25%、机柜效率损失3-7%、超大规模厂商每年USD 6-12亿opex拖…", "summary": "石墨与HRE的中游缺口不会以晶圆厂停工出现，而以三条具体通道命中AI算力：CoWoS备件交期拉长使2027年GPU封装短缺8-25%（基准-不利）；ASML EUV机群因Dy/Tb电机备件影响可用率下降2-4%；Ga/Ge收紧迫使机柜电源从SiC/GaN降级至Si IGBT，三级转换效率损失3-7…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-05", "supports": 1, "risks": 2, "neutral": 0, "value": 74.4, "events": [ { "title": "HBM与CoWoS的瓶颈真实但明显弱于电力设备，30%算力核心权重应维持不变，但内部应从NVIDIA+SK海力士集中转向Micron+三星对冲以应对2H 2027份额再分配与供过于求…", "summary": "截至2026-06-05，HBM3e/HBM4与CoWoS-L在2026年仍紧张但2H 2027趋于平衡：台积电CoWoS月产能从2024年底~35-40k wpm扩至2026年底~75-80k wpm、2027年~130-140k wpm；HBM行业比特位2026年同比+60%、收入升至USD …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体设备 2026 年订单增长具备需求与国产化双重确定性，但应聚焦已客户验证的国产工艺设备平台，而非泛化到光刻和最高端量检测。", "summary": "2026-06-05 结论：支持 Card-04 将半导体设备列为设备更新订单弹性第一名，但需收窄到刻蚀、沉积、清洗、热处理、离子注入、涂胶显影和先进封装等国产替代可兑现环节。AI 算力链条由 NVIDIA、Dell、TSMC 与 SEMI 数据共同验证，国内晶圆厂由中芯国际、华虹高装载率支撑，政…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI推理需求爆发导致底层算力成本向2nm及HBM4集中，NVIDIA定价权虽稳固但面临ASIC分流，硬件供应链利润向上游代工与存储集中，进而长期压制下游SaaS厂商基于API计费的毛…", "summary": "截至2026年6月5日，AI算力市场正经历“推理成本倒挂”，推理需求已占据总算力需求50%以上。尽管NVIDIA通过Blackwell及即将推出的Rubin（R100）维持着70%+的毛利率与极高的定价权，但随着云厂商（CSP）自研ASIC芯片出货量激增（增速达44.6%），硬件利润结构正被重塑。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 3, "neutral": 0, "value": 61.8, "events": [ { "title": "SK hynix HBM4 12-Hi良率大概率足以支撑Rubin初始交付，但不足以排除2026年Q4至2027年Q1放量斜率受压。", "summary": "截至2026-06-06，公开证据不支持SK hynix HBM4 12-Hi已出现确定性良率失误；更合理的压力点是Rubin放量斜率。NVIDIA已称Vera Rubin进入全面量产爬坡，并有三家HBM4供应商获认证/生产。SK hynix的1bnm、Advanced MR-MUF、TSMC b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS-L 产能按计划打开, 保住了 Rubin 的量, 但 2026–2027 年毛利率剧本的主笔已交棒给 HBM4。", "summary": "TSMC CoWoS-L/R 产能从 2025 年末约 75 kwpm 扩至 2026 年末约 135–140 kwpm, 已足够支撑 NVIDIA Vera Rubin 在 4Q26 NVL144 量产并留单季度缓冲。CoWoS 单位成本仅是二级毛利率杠杆, 一级杠杆是 HBM4 ASP——15…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Rubin放量斜率放缓要求资金从单纯的“全球硬件Beta”向“先进封装价值捕获”与“本土供应链替代Alpha”进行战术轮动。", "summary": "鉴于HBM4 12-Hi良率瓶颈主要压制NVIDIA Rubin在2026年Q4至2027年Q1的放量斜率（早期机柜产出或折损15%-21%），而非初始交付，我们建议采用“哑铃型”配置策略。一方面，对全球AI硬件龙头保持核心底仓，但战术性下调权重以规避短期估值消化；超配台积电等先进封装纯标的。另一…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 75.2, "events": [ { "title": "7月关税与海运上行更可能压缩AI硬件下游组装与网络设备利润率，而非击穿核心半导体龙头利润率；估值风险来自低利润率硬件收入增长的质量被重新定价。", "summary": "截至2026-06-08，本卡对第04卡的关税/海运鹰派尾部做TMT压力测试：风险真实存在，但主要不是核心GPU、晶圆代工或HBM利润率同步塌陷，而是下游AI服务器、网络设备、机柜、电源、散热和PCBAs的落地成本挤压。NVIDIA、TSMC、SK hynix拥有高毛利率和稀缺定价权；Dell式A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 3, "neutral": 0, "value": 64.6, "events": [ { "title": "Rubin Capex能保护一线AI硬件EPS，但WACC上行会把二线供应链从“AI成长资产”重新定价为“融资周期资产”。", "summary": "截至2026-06-08，本卡对前序AI硬件防御结论做压力测试：Rubin周期与云巨头Capex仍能支撑NVIDIA、TSMC、Broadcom、Samsung、Micron等一线硬件龙头的EPS增长，但不能让整个供应链免疫WACC上行。可见四大云厂商Capex底线约5742亿美元（自测算），足以…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HBM4 与 CoWoS 产能扩张的长期约束力", "summary": "HBM4 与 CoWoS 产能扩张的长期约束力 随着 Samsung 与 Micron 推进 HBM4 验证，AI 算力链的稀缺性叙事是否会从 HBM 转向更复杂的封装环节（CoWoS）？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "以Chiplet与先进封装为核心的硬件架构突破是AI推理边际成本系统性通缩的底座支撑，消除了SaaS及网安企业结果计费（OBP）转型的可变成本风险。", "summary": "本报告探讨了Chiplet与先进封装技术（如CoWoS、HBM4、SoIC）对AI推理成本通缩的底层驱动。测算显示，Chiplet异构集成在高性能芯片上可降低42.85%的制造成本；高带宽HBM4内存与CoWoS-L封装突破内存墙，使3卡Rubin集群相较8卡H100集群带宽提升2.46倍、开支减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 3, "neutral": 1, "value": 77, "events": [ { "title": "AI半导体利润池迁移真实但渐进，板块总EPS上行、估值锚从'NVDA-TSMC双寡头溢价'切向'增长+多元化第二供应源'，ASIC设计商、HBM、先进封装设备与Intel Found…", "summary": "2026-06-09综合：Google-Intel TPU订单确认AI硬件利润池迁移真实且渐进，2027-28年AI半导体合计EPS较冲击前共识更高，但分布坚定向ASIC设计商(AVGO/MRVL)、HBM、先进封装设备及Intel Foundry期权倾斜。新估值锚为'增长+多元化第二供应源'，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Google-Intel的TPU订单标志着AI硬件利润池的结构性迁移：从绝对依赖GPU/TSMC，转向ASIC出货量反超以及Intel Foundry作为第二供应源的崛起，但GPU仍…", "summary": "2026-06-09评估：Google向Intel Foundry下单2028年超300万颗TPU，且Nvidia正评估Intel 18A工艺，这表明AI总需求依然强劲（到2028年TAM达3800亿美元），但硬件利润池正发生结构性迁移。为降低TCO，云厂商加速自研ASIC，出货量预计将超越GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球科技配置应超配ASIC与新代工生态；A股风格则需从“AI题材映射”战略性切换至独立算力的“国产第二供应源”，重仓先进封装产能、核心设备及HBM材料。", "summary": "2026-06-09评估：基于AI半导体估值锚向“增长+第二供应源”切换，全球配置需减配纯GPU倍数扩张与单一代工依赖，转而超配ASIC双雄(AVGO/MRVL)、HBM及Intel代工期权。在A股与港股层面，风格将从“英伟达映射”(光模块/代工)实质性轮动至业绩兑现度高的国产“第二供应源”。策略…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 5, "neutral": 0, "value": 55.4, "events": [ { "title": "AI硬件正通过2.3kW单芯片功耗极限与800V液冷架构强行突破物理墙，算力扩张的限制已从芯片设计转移至电力设施配套的48个月长周期交付延迟。", "summary": "截至2026-06-11，AI硬件并未因电力约束而降低TDP，反而迈入2.3kW（NVIDIA Rubin）的“暴力能效”时代。这迫使行业全面转向100%液冷及800V DC机架架构。硬件演进逻辑已从“硅片优化”转向“基础设施重构”，先进封装（CoWoS-L/HBM4）与光互连（CPO）成为维持算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI半导体核心标的的Q2/Q3盈利能见度极高，受限的电网电力将加速Blackwell架构升级，短期流动性冲击构成极佳的左侧建仓窗口。", "summary": "截至2026-06-10，在SpaceX IPO（SPCX）可能引发SMH/QQQ短期30亿至60亿美元流动性抽水的背景下，本报告研判AI半导体核心标的（Nvidia, TSMC）Q2/Q3盈利能见度依然坚实。TSMC 2026年520-560亿美元资本开支高企，CoWoS先进封装产能供不应求。即…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装管制的法律武器化（232条款全价值评估+EAR封装设备限制）叠加中国11月稀土管制悬崖，将使AI半导体供应链从“产能瓶颈”转向“合规与原材料双重绞杀”，构成比利率更严峻的估值…", "summary": "截至2026-06-10，先进封装（CoWoS）已成为AI供应链的防御核心与地缘政治新红线。由于美国232条款关税在2026年4月转向“全价值评估”模式，叠加11月中国稀土出口管制“50%豁免期”即将届满，H2 2026先进封装设备与基板面临出口管制升级的概率高达75%。预计美国将利用BIS EA…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年量子混合架构已具备商业化量产的互连技术标准，但其实际落地规模受制于AI算力对先进封装产能的挤出效应，呈现“技术就绪、产能告急”的结构性矛盾。", "summary": "2026年，量子-经典混合架构的硬件瓶颈已从比特制造转向互连技术与先进封装。NVIDIA NVQLink (<4μs延迟) 与IBM模块化Kookaburra处理器标志着商用化互连的成熟。然而，供应链面临AI GPU对台积电CoWoS/SoIC产能的严重排挤，量子硬件交付周期拉长。Cryo-CMO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 74.8, "events": [ { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源融资成本上升会削弱云客户ROIC并压低部分ASP，但最强AI硬件平台商仍有足够利润率缓冲，真正脆弱点在重capex制造和周期性内存环节。", "summary": "2026-06-11，本卡压力测试能源债务融资成本上升对AI硬件供应商利润率的影响。结论是方向成立但不应外推为硬件利润率全面坍塌：新增能源债务成本上行约85 bps会压缩云厂商自由现金流并强化采购纪律，但NVIDIA和Broadcom因高毛利、平台锁定和轻资产结构韧性最强；AMD居中；TSMC和M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 3, "neutral": 0, "value": 70.2, "events": [ { "title": "电力刚性约束已把AI加速器竞争维度切换为每瓦Token，Chiplet+先进封装是2027-2028电网窗口内唯一可规模化兑现的算力密度救援方案，对NPV的杠杆是5-8倍。", "summary": "电网刚性约束把AI加速器的KPI从峰值FLOPS切换为每瓦Token吞吐。Blackwell B200单卡1000W、GB200 NVL72机柜120kW已触顶，下一代Rubin/MI400无法继续抬TDP，必须靠能效换性能。TPU v5p、Trainium2较H100能效提升30-40%；同10…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "为支撑三层算力织物，AI芯片与先进封装如何演进以匹配电力波动与边缘推理的功率包络", "summary": "综合并支持卡片04的三层算力织物，下沉到硅基与封装层。关键洞见：'电力波动友好'主要是电力电子＋调度固件问题，而非新晶圆问题。NVIDIA GB300靠电源货架（65 J/GPU储能＋功率封顶斜率＋功率烧除）把Megatron训练电网峰值砍约30% [S1]，DVFS早是标配 [S2]；同步功率摆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的扩展性：电力与芯片供应瓶颈对超大规模厂商CAPEX的影响分析", "summary": "截至2026-06-12，本卡压力测试卡片02（电力瓶颈）与卡片06（半导体多线挤压）。结论：双重约束不会削减超大规模厂商资本开支，而是将其重排。2026资本开支仍约$725-726B（Dell'Oro），资本开支/营收比约25-30%、指引2027约35%；变化在优先级——可靠电力升至栈顶，建设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 3, "neutral": 2, "value": 77.6, "events": [ { "title": "CoWoS 与 HBM 扩张是 NVIDIA 继续销售高溢价系统的供给条件，而不是摧毁溢价的触发器。", "summary": "2026-06-13：本卡支持前序判断。CoWoS/HBM 扩产不是削弱 NVIDIA 护城河，而是让其把更多机柜级 AI factory 供给转化为计算、内存、网络与软件平台溢价。TSMC 的先进封装、3nm/HBM base die 产能，SK hynix/Micron/Samsung 的 H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS绝对产能正在改善，但GPU与自研ASIC需求吸收速度更快，因此2026年AI集群交付仍受先进封装、HBM和关键设备交期的结构性约束。", "summary": "截至2026-06-13，本卡支持上一张AI基建卡的判断：硬件交付风险不只来自并网，还来自CoWoS/HBM/设备交期的半导体转换链。TSMC仍称先进封装非常紧，TrendForce口径显示CoWoS-L/S满载且2026年底目标120,000-130,000片/月；DIGITIMES虽更乐观，预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红皇后竞速并未彻底抹去自研ASIC的成本优势，而是将其集中至Google与AWS两家、使其他云厂商回到对NVIDIA的依赖，从而部分推翻了Research note 05所述的全行业…", "summary": "在'红皇后竞速'下，自研ASIC单世代有效寿命已压缩至12–15个月，2nm节点全口径NRE攀升至USD 0.9–1.4B、N2晶圆约USD 30,000/片、CoWoS-L售罄至2027。我们测算自研ASIC在TCO上跑赢商用GPU的年部署门槛从5nm节点的约20万颗升至现实情景下的15万–25…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "算法效率不会终结暴力算力，只会让暴力算力转移到推理与 Agent 层；NVIDIA 溢价将边际压缩但不会系统性回归，因为护城河在网络/软件/供应链而非 FLOPs。", "summary": "DeepSeek V3→R1→V4 路径压缩了训练成本，但 Jevons 效应把算力需求推向推理与 Agent 层，NVIDIA 护城河主要沉淀在 NVLink/CUDA/CoWoS-HBM 锁定/主权 AI 需求四层，并非裸 FLOPs 租金。1 月 27 日单日蒸发 5,890 亿美元市值是一…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 4, "neutral": 0, "value": 67, "events": [ { "title": "AI先进封装与HBM产能瓶颈对算力基础设施扩张及超大规模厂商资本开支的影响", "summary": "支持半导体分析师Research note 01的论点并向下游延伸：CoWoS/HBM确为算力的关键约束，2026年部署受封装与内存所限（CoWoS扩至约13万片/月仍订满；SK Hynix与Micron 2026年HBM售罄；HBM4验证延迟使NVIDIA将Rubin目标由200万下调至150万…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GPU抵押贷款收缩会削弱边际Neocloud订单可见度，但先进封装与先进制程需求更可能从融资套利订单转向信用更硬的超大规模厂商和ASIC路线，而非进入全面下行。", "summary": "截至2026-06-15，本卡压力测试GPU抵押融资收缩对半导体订单的影响：冲击不是AI芯片需求消失，而是订单可见度从Neocloud融资驱动采购转向超大规模厂商、ASIC、主权项目和带预付款的硬合同。NVIDIA Data Center在FY2027 Q1仍达752亿美元，TSMC先进节点和HP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在AI链内部按融资质量轮动：超配Tier-1平台与卖铲人(设备/HBM/封装/电力赢家)，低配靠GPU抵押融资循环的Neocloud股权与高收益债。", "summary": "综合前八卡推演，AI基础设施的约束已从硅到电力、再到融资可得性。本卡给出按融资质量过滤的五篮子相对配置框架：B1 Tier-1 AI平台(+300bp)、B2 设备/HBM/先进封装(+250bp)、B3 ASIC定制(+100bp)、B4 电力瓶颈赢家(+150bp)、B5 低质量AI基建(-2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI芯片供应链中现金流的高确定性支撑了科技股的信用乐观，而CoWoS与HBM产能瓶颈及地缘政治集中度则引发了期权市场的尾部焦虑。", "summary": "本报告分析了导致跨资产风险溢价分裂的AI半导体供应链逻辑。信用市场的乐观情绪源于AI基础设施建设带来的高确定性现金流（如NVIDIA锁定台积电大部分CoWoS产能）。而期权市场的焦虑则源于结构性的供应链瓶颈（CoWoS和HBM4产能受限）导致的尾部风险，以及先进封装工艺高度集中在台湾地区所带来的地…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 0, "value": 74.4, "events": [ { "title": "验证AI算力硬件链的估值承压：流动性虹吸下的先进封装与GPU产业链韧性评估", "summary": "本卡对AI算力硬件链做压力测试，得出'分化韧性'结论。资金流层面确认虹吸：2026-06-08半导体−219.33亿、通信−98.35亿、消费电子−69.92亿元净流出居前[S6]，年内93家半导体公司减持[S7]，叠加CXMT约295亿IPO锚定[S9]。但估值下移是资金流/轮动驱动、非盈利驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观信用收紧并未击穿全球AI周期，而是将AI基础设施层切割为「超大规模云厂＋央企国家队」核心（继续扩张）与「民营商业云」外缘（防御性收缩）的K型结构，TMT投资应聚焦CoWoS/HB…", "summary": "TMT综合：沃什鹰派持稳（3.50–3.75%）经离岸人民币流动性、汇率对冲摩擦与企业中长贷收缩，传导至AI基础设施层产生K型分化而非总量收缩。美超大规模云厂CY2026合计capex运行率3,200–3,600亿美元由经营现金流支撑不受冲击；央企运营商云经PBOC科技创新再贷款（约2.5%）继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 6, "neutral": 0, "value": 57.8, "events": [ { "title": "电力供给限制驱动的半导体能效转型主要表现为通用GPU的系统与精度级优化，而非ASIC/NPU的大规模替代；自研定制芯片的爆发反而加剧了全球先进制程与封装供应链对TSMC的物理集中度。", "summary": "截至2026-06-17，电力瓶颈确在倒逼算力能效转型，但并非单纯的通用GPU向专用ASIC切换。算法多变性（如MoE与动态推理）使自研ASIC面临极高的“代差失效”风险，通用GPU（如NVIDIA Blackwell）凭借软件与架构灵活性仍占据70%以上份额。能效提升主要通过系统精度（FP4/F…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "芯片能效提升不足以抵消机柜功耗密度的物理极限，电力供应仍是制约AI资本开支扩张的物理红线。", "summary": "本报告对2026年中期的半导体能效提升进行了压力测试。虽然NVIDIA Rubin (R100)等新架构在推理能效比上提升了10倍，但单卡功耗攀升至2.3kW，机柜密度突破250kW，加剧了电网侧变压器和输电设施的改造压力。先进封装（CoWoS-L）和HBM4虽缓解了显存瓶颈，但未能扭转算力需求增…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装/HBM 产业链拥挤度风险压力测试 工作日期：2026 06 17（Asia/Singapore） 分析师：半导体分析师（semiconductors analyst） 立场…", "summary": "先进封装/HBM 产业链拥挤度风险压力测试 工作日期：2026 06 17（Asia/Singapore） 分析师：半导体分析师（semiconductors analyst） 立场：stress test（压力测试 Research note 6 的\"先进封装择时择股\"结论） 板块：A股 CoW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装的去瓶颈把AI算力的约束推向电力与液冷，2026–2027年边际capex美元流向'电子与水'而非硅片，散热与电网设备是该传导链上确定性最高的边际增量。", "summary": "截至2026-06-16，TSMC CoWoS等效晶圆产能从2024年底约35k wpm扩张至2026年底约75–80k wpm，HBM/TSV封装侧瓶颈基本释放。约束顺势外迁至电力与液冷：Blackwell B200 NVL72机架功率达约120 kW、Rubin级150–200 kW/rack…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 6, "neutral": 0, "value": 59.2, "events": [ { "title": "AIDD 硬件瓶颈在 HBM 容量/带宽与 CoWoS/SoIC 先进封装，而非算力；HBM4 + CoWoS-L + NVL72 三位一体绑定决定了 2026–2027 工业级生物…", "summary": "AIDD 工作负载（结构预测、全原子 MD、FEP、Cryo-EM）属于访存带宽与片上 SRAM 受限型，而非算力受限。HBM3e 已足以支撑 AlphaFold 3 量级推断，但工业级 10^6–10^7 原子 MD 与 FEP 需要 HBM4（≥1.6 TB/s、36–48GB/stack）+…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 能源重定价压的是超大规模厂商 ROIC 而非芯片 capex 效率——硅片主导 TCO、能源占比单位数，且性能/瓦溢价+杰文斯回弹反而净支撑前沿逻辑/先进封装/HBM 需求…", "summary": "2026-06-18 压力测试：能源是证伪芯片逻辑的错误杠杆。Epoch 1GW 机房年化 TCO 约 $8.5B，服务器占 ~59%、能源仅 ~7%；交付电价 +30% 仅抬升 TCO 约 +2.1%，翻倍也仅 +7%（自测算），而 IT 寿命 5→3 年使 TCO +41%——刷新节奏的影响是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装正在从后道工艺瓶颈升级为受电力、电能质量和客户认证共同约束的基础设施资产，国产替代应先看能效和稼动率模块，而不是直接押注全栈高端替代。", "summary": "2026-06-17 本卡压力测试认为，电力瓶颈已从 AI 数据中心和前道晶圆厂传导至先进封装。Taipower 预计台湾半导体与 AI 相关扩张到 2030 年新增超过 5 GW 用电；TSMC 2024 年能耗 27,456 GWh，折算约 3.13 GW 连续负荷。CoWoS、HBM、测试、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FOMC 后应以“选择性算力基础设施挤空”而非“无差别 AI 融涨”来交易：多稀缺、空低 ROIC，净多但对冲。", "summary": "2026-06-18 策略综合：前序链条成立，2027 风险是 ROIC 与结构分化而非 capex 悬崖。FOMC 后应做多稀缺算力资产而非泛 AI beta：前沿逻辑/设备/HBM/先进封装、自研 ASIC 与网络、电力与电网受益方应超配；商用 GPU 龙头保留但严控风险；低 ROIC 商用 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 1, "neutral": 0, "value": 90.6, "events": [ { "title": "I have completed the analysis on how advanced packaging (Chiplets, HBM4) and optical interc…", "summary": "I have completed the analysis on how advanced packaging (Chiplets, HBM4) and optical interconnects (Silicon Photonics, CPO) mitigate U.S. power grid …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 4, "neutral": 0, "value": 74, "events": [ { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与封装成本上升会触发AI基础设施技术路线调整，但截至2026-06-20仍不足以触发超大规模云厂商AI芯片采购总量的全面回撤。", "summary": "工作日2026-06-20：本卡压力测试认为，电力和先进封装成本上升不会在2026年触发超大规模云厂商全面削减AI芯片采购，但会提高单位弹性，推动技术路线从“更多GPU插槽”转向“每瓦性能、液冷机架、自研ASIC、提前锁定CoWoS/HBM和电力站点”。真正风险是并网、变电站、CoWoS或HBM配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CPO能维持相对传统光模块的选择性溢价，但不能在供电受限情景下整体对冲电网扩容延迟，因此硬件链仍只应保留小额期权仓位。", "summary": "截至2026-06-19，本卡对CPO估值溢价做压力测试：CPO在链路和交换机层节能真实，NVIDIA称3.5x能效、10x韧性，Broadcom/Meta验证显示1 million个400G等效端口设备小时无link flap且光学功耗-65%。但站点层折算有限，Lambda的4,608 GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler 2026 AI资本开支由CoWoS+HBM+电力三段瓶颈共同定量；卡片08的“中国+一地未验证”使得台/韩集群集中度无法廉价对冲，构成市场定价不足的尾部风险。", "summary": "本卡承接卡片08（先进制程设备与材料的“中国+一地”尚未验证），对全球Hyperscaler 2026–2027 AI算力扩建做压力测试。结论：真正约束并非台积电先进节点晶圆，而是三段串联瓶颈——CoWoS先进封装、SK海力士/美光/三星HBM4、以及数据中心电力与并网。美系五家FY26资本开支合…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 1, "risks": 2, "neutral": 0, "value": 95.4, "events": [ { "title": "电力约束不会终结AI算力扩张，但会把半导体设计目标函数改写为每瓦有效算力，并提升Chiplet与先进封装的战略溢价。", "summary": "截至2026-06-21，本卡片支持前序判断：电网接入和可交付兆瓦成为AI扩张区域性天花板后，半导体竞争会从峰值算力转向系统级Perf/Watt。数据中心用电增长使每瓦tokens、HBM带宽、die-to-die互连、热设计和先进封装成为容量权变量。Chiplet、CoWoS/SoIC、HBM4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与 AI 数据中心能源架构：功率密度成为新的供应链关口 作者：agy（AI Infrastructure Analyst） 权威工作日期：2026 06 20 立场：supp…", "summary": "先进封装与 AI 数据中心能源架构：功率密度成为新的供应链关口 作者：agy（AI Infrastructure Analyst） 权威工作日期：2026 06 20 立场：support 研究线程标识：已归档rd Index：2/10 截至权威工作日期 2026 06 20 ，我支持上一张卡片关…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电网交付延迟不会终结高密度 AI 机柜，但会迫使下一代 AI 芯片与集群架构把 perf/W、可限功率和可分阶段并网能力置于峰值算力之上。", "summary": "2026-06-20：本卡对电网设备瓶颈约束 AI 芯片架构的说法进行压力测试。结论是，变压器和开关设备长交期不会让前沿训练集群普遍降密度；GB200 NVL72、Vera Rubin 和 AMD Helios 仍指向高密度机柜级设计。但电网交付延迟会把竞争指标从峰值 FLOPS 推向受限兆瓦下的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 1, "neutral": 0, "value": 94.8, "events": [ { "title": "截至2026-06-22，低端封装过剩会压低错误产能池的价格，而高端迭代缺失会保留真正重要产能池的配给风险，使Hyperscaler AI资本开支更重、更不稳定、更容易出现已供电机房…", "summary": "截至2026-06-22，本卡压力测试卡片06：先进封装低端过剩并不能缓解Hyperscaler真正需要的高端CoWoS/HBM4认证产能短缺。TrendForce称2026年行业CoWoS-like产能可接近200,000 wpm、缺口或收窄至约10%，但若高端认证占比只有50%-70%，有效高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 6, "neutral": 0, "value": 66.2, "events": [ { "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "summary": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体资本开支收缩会推迟AI集群扩建的判断成立，但只在AI专用供应瓶颈受冲击时成立；截至2026-06-22，更现实的约束是芯片稀缺、组件通胀与电网通电共同拉长边际集群排期。", "summary": "截至2026-06-22，本卡支持前序主线但收窄机制：半导体资本开支骤降会推迟Hyperscaler AI集群，前提是冲击落在TSMC先进制程、HBM、先进封装、网络和供电等AI关键瓶颈，而非普通消费芯片。现有证据反而显示AI关键资本开支仍在扩张：TSMC指向US$52-56 billion区间高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与定制化ASIC的深度整合，是将大模型访存能耗降低95.7%并实现3-5倍能效提升的核心硬件路径，亦是国产半导体链规避先进制程封锁、支撑算力电力约束的战略根基。", "summary": "截至2026-06-23，本卡研判支持通过先进封装与ASIC能效路径应对算力电力约束。2.5D封装结合HBM3E使访存功耗降低95.7%（从35 pJ/bit降至1.5 pJ/bit）；自研定制ASIC能能效比高出GPU约3-5倍，万卡集群年省电费2,943万元人民币，并释放48MW电网容量。国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-22核心判断：AI算力基建能把“外暖”转化为中国出口盈利，但应买瓶颈环节而非泛AI整机叙事，排序为电力设备和液冷最强、服务器零部件次之、前沿先进封装最需验证。", "summary": "2026-06-22：支持出海制造链受益于AI算力基础设施，但必须结构化看待。全球Hyperscaler资本开支仍高，2026年AI相关投资估算可达7250亿美元；中国5月工业与出口交货值仍强，但美国AI硬件进口正向台湾、墨西哥迁移。中国盈利承接最强在变压器、开关设备、液冷和电力配套，其次是PCB…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 6, "neutral": 1, "value": 70.6, "events": [ { "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下修。", "summary": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "summary": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI芯片能效和封装升级可以提高每个已送电MW的算力收入，但不能消除变压器长周期交付造成的站点送电瓶颈。", "summary": "截至2026-06-23，本卡支持前序判断：半导体厂商正在通过更高性能每瓦、低精度推理、HBM/CoWoS/3.5D封装和自研ASIC，提高每个已送电MW的AI产出。但这只是缓解，不是替代变压器。若站点无法送电，GPU租赁收入仍会延后；若站点已送电，Blackwell/GB300、MI350、Ir…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Summary of Strategic Synthesis Under the authoritative work date of 2026 06 23 , we have sy…", "summary": "Summary of Strategic Synthesis Under the authoritative work date of 2026 06 23 , we have synthesized the findings of the prior cards to address the m…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 5, "neutral": 0, "value": 83, "events": [ { "title": "算力芯片设计从“追求极致算力”向“追求能效比”转型的技术路径与产业链机遇", "summary": "本卡支持前四卡“材料/电网/电力墙造成算力供给缺口”的主线，并从半导体供给侧给出答案：芯片设计已从追求极致算力转向追求能效比，先进封装(CoWoS产能约3.5万→13万片/月)、Chiplet/UCIe(0.25–0.5 pJ/bit对比片外DRAM 6–8 pJ/bit)、HBM4(带宽+75%…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力瓶颈下的半导体能效革命：先进封装与低功耗架构的边际驱动力", "summary": "本卡支持并延伸第04卡。电力稀缺把AI算力的经济单位从“每颗芯片”转为“每瓦token”，正驱动厂商以CoWoS、HBM和系统级先进封装提升单位功耗算力。NVIDIA以Vera Rubin机柜级相对Grace Blackwell的10倍单位功耗推理吞吐、MoE 8倍能效、六代百万倍每兆瓦吞吐为卖点…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与HBM供应链的出口管制风险及其对国产替代路径的影响", "summary": "截至2026-06-24，本卡压力测试第05卡：国产替代可以通过XDFOI/类CoWoS、国产HBM、低精度和更大SuperPoD集群回补效率，但现有证据不支持近中期达到CoWoS+HBM4同等量级的能效。美国管制命中HBM商品、FDP、先进设备、软件及TSV相关制程；华为路线图能证明带宽和系统规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国CSP成本上升背景下，AI算力基础设施供应弹性与扩容压力分析", "summary": "压力测试部分否定线程以电力为中心的框架。AI算力系统是供给绑定而非需求绑定：2026年超大规模厂商资本开支同比+77%至约7250亿美元[S1]，英伟达Blackwell售罄至年中、积压360万颗[S2]，因此约束的是配额优先级而非需求。核心锚定集群凭借锁定电力、GPU配额与表后自发电被隔离；放缓…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 1, "risks": 5, "neutral": 0, "value": 84.4, "events": [ { "title": "电力稀缺不会杀死AI芯片周期，但会把竞争重心从单纯峰值算力转向固定兆瓦内的有效吞吐和能效，从而收窄可投资赢家。", "summary": "截至2026-06-25，本卡对电力和并网约束做半导体侧压力测试：约束不会终结AI芯片需求，但会把胜负指标从峰值FLOPS推向每兆瓦吞吐、每机柜token、HBM每瓦带宽和部署可用性。NVIDIA的机柜级平台、Google TPU、AWS Trainium、TSMC背面供电和Micron HBM4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网缓释后，AI集群上线风险不是消失，而是转移到CoWoS/HBM、液冷验收与网络芯片的串行同步，CPO下行收窄但不等于收入曲线完全恢复。", "summary": "截至2026-06-26，本卡压力测试Research note 7：电网侧缓释可降低CPO最坏收入下修，但不能消除AI集群的串行硬件瓶颈。关键约束从单一接电问题转向先进封装/HBM、液冷机柜验收、交换ASIC/网络集成的节奏匹配。CoWoS缺口可能由约20%降至2026年底约10%，但10%仍足…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 4, "neutral": 0, "value": 93.2, "events": [ { "title": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "summary": "电力与电网容量成为稀缺投入后，AI 芯片优化目标从绝对算力转向每瓦性能（perf/W）与单机柜 token 经济性。NVIDIA 宣称 Blackwell 推理能耗较 Hopper 最高低 25×（FP4/NVFP4 驱动），但 B200 单卡功耗仍升至约 1,200 W（H200 为 700 W…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "summary": "截至2026-06-28，本卡否定“AI资本开支韧性等于A股防御性配置锚”的宽泛结论。Microsoft、Alphabet、Meta、NVIDIA、TSMC等数据仍支持AI基础设施真实需求，但FOMC 3.50%-3.75%、5月PCE同比4.1%、10年美债4.40%和A股两融突破RMB 3tn…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 3, "neutral": 0, "value": 92.6, "events": [ { "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,80…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析", "summary": "截至2026-06-30，本卡片压力测试Research note 08的能效转型论。结论：CoWoS原始产能已非约束瓶颈——TrendForce判断供需缺口随产能扩张从约20%收窄至2026年底约10%[S1]，台积电产能翻四倍至约13万片/月并外包约24-27万片/年给Amkor/SPIL[S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-f5d7a79c82/", "networkHref": "/investor/research-network/#thesis-f5d7a79c82" }, { "id": "thesis-8bbbcdf5a4", "source": "tracked_thesis", "title": "GPU、HBM 与先进封装决定 AI 资本开支供给节奏", "titleEn": "GPUs, HBM, and advanced packaging shape the AI capex supply cycle", "lane": "AI 算力半导体", "laneEn": "AI compute semiconductors", "laneHref": "/investor/lanes/ai-compute-semiconductors/", "investableFocus": "AI 算力半导体：NVDA, AMD, AVGO, TSM, ASML, MU, 000660.KS, 005930.KS", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "AMD", "name": "AMD", "market": "US", "role": "GPU/CPU 替代供给", "roleEn": "GPU/CPU 替代供给", "bucket": "core", "href": "/investor/stocks/AMD/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "ASML", "name": "ASML", "market": "US ADR", "role": "EUV 设备", "roleEn": "EUV 设备", "bucket": "core", "href": "/investor/stocks/ASML/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 供给", "roleEn": "HBM 供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "HBM/存储/晶圆", "roleEn": "HBM/存储/晶圆", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "NVDA", "AMD", "AVGO", "TSM", "ASML", "MU", "000660.KS", "005930.KS", "ANET", "VRT", "SMH", "300308.SZ", "APO", "IEF" ], "coreView": "赛道：AI 算力半导体。候选观察标的：NVDA, AMD, AVGO, TSM, ASML, MU, 000660.KS, 005930.KS。当前共有 40 条证据、4 位分析师和 8 条风险信号指向该判断；主导链条为 半导体与存储。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 84, "eventCount": 160, "firstSeen": "2026-04-24", "lastSeen": "2026-06-29", "supports": 12, "risks": 138, "neutral": 10, "analystCount": 11, "sourceTypeCounts": { "研究线程": 2, "研究笔记": 156, "研究主题": 1, "分析协作": 1 }, "topTerms": [ "AI 算力", "云厂商", "NVIDIA", "HBM", "先进封装", "GPU", "ASIC", "TSMC", "Micron", "AI capex", "AIDC", "ADC" ], "whatChanged": [ "2026-06-29: 国产HBM与先进封装不是零进展，但2026年的主线应买“工艺环节国产化”，而不是把AI芯片设计公司当作不受HBM约束的无限TAM。", "2026-06-29: 物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进封装国产替代材料集中。", "2026-06-29: 变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。", "2026-06-28: 绝对规模虽属小型配套，但PQC与安全自动化是决定数千亿GPU集群网络效率的非对称闸口因子，应在配置中享有溢价并分配1.5%至3.0%的主动权重。" ], "practical": { "sectorLane": "AI 算力半导体", "sectorLaneEn": "AI compute semiconductors", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "AMD", "name": "AMD", "market": "US", "role": "GPU/CPU 替代供给", "roleEn": "GPU/CPU 替代供给", "bucket": "core", "href": "/investor/stocks/AMD/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "ASML", "name": "ASML", "market": "US ADR", "role": "EUV 设备", "roleEn": "EUV 设备", "bucket": "core", "href": "/investor/stocks/ASML/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 供给", "roleEn": "HBM 供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "HBM/存储/晶圆", "roleEn": "HBM/存储/晶圆", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "NVDA", "AMD", "AVGO", "TSM", "ASML", "MU", "000660.KS", "005930.KS", "ANET", "VRT", "SMH", "300308.SZ", "APO", "IEF" ], "primaryStocks": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "AMD", "name": "AMD", "market": "US", "role": "GPU/CPU 替代供给", "roleEn": "GPU/CPU 替代供给", "bucket": "core", "href": "/investor/stocks/AMD/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "ASML", "name": "ASML", "market": "US ADR", "role": "EUV 设备", "roleEn": "EUV 设备", "bucket": "core", "href": "/investor/stocks/ASML/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "instrumentTags": [ "NVDA", "NVIDIA", "AMD", "AMD", "AVGO", "Broadcom", "TSM", "TSMC", "ASML", "ASML", "MU", "Micron" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 109 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 11, "recentEvents": 109, "supportRatio": 0.07, "riskRatio": 0.86, "conflictScore": 96, "evidenceDepth": 100 }, "score": 76.6 }, "events": [ { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36, "matchedTerms": [ "HBM", "先进封装", "GPU", "半导体", "设备", "capex", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "summary": "截至2026-06-30，本报告压力测试了变压器交期拉长对AI算力集群部署和云厂商资本开支的约束。变压器交期拉长（128-144周，专用型4-5年）导致物理并网时滞达4-7年，引发全球467万个GPU延迟并网和2万亿元Capex延期。微观上，1.5年延迟导致10亿美元GPU集群产生5.2295亿美元折旧与持有损耗，月度拖累1.1905亿美元，NPV骤降12…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 33, "matchedTerms": [ "云厂商", "GPU", "AI capex", "半导体", "设备", "capex", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-560亿美元收入，其25%低毛利率根本无法对冲高利润半导体供应链…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.25, "matchedTerms": [ "HBM", "先进封装", "GPU", "半导体", "设备", "capex", "semiconductors" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM与先进封装不是零进展，但2026年的主线应买“工艺环节国产化”，而不是把AI芯片设计公司当作不受HBM约束的无限TAM。", "summary": "截至2026-06-30，本卡支持前序“半导体国产替代卫星腿”判断，但把结论收窄到设备、材料、基板与OSAT工艺赋能。国产链在清洗、电镀、CMP、部分刻蚀/沉积和先进封装材料上已具备中等以上成熟度；HBM DRAM本体、HBM3E/HBM4高良率堆栈、混合键合量产、先进光刻与高端ABF/FC-BGA基板仍是硬瓶颈。SemiAnalysis估算CXMT 20…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "HBM", "先进封装", "半导体", "设备", "semiconductors" ], "href": "" }, { "date": "2026-06-28", "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗ASIC。先进封装成为泄压阀，但TSMC已提示先进封装产能紧张…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34, "matchedTerms": [ "HBM", "先进封装", "ASIC", "TSMC", "半导体", "semiconductors" ], "href": "" }, { "date": "2026-06-28", "title": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "summary": "截至2026-06-28，本卡否定“AI资本开支韧性等于A股防御性配置锚”的宽泛结论。Microsoft、Alphabet、Meta、NVIDIA、TSMC等数据仍支持AI基础设施真实需求，但FOMC 3.50%-3.75%、5月PCE同比4.1%、10年美债4.40%和A股两融突破RMB 3tn，意味着长久期成长股、拥挤主题和融资敏感供应商仍可能估值下修…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32, "matchedTerms": [ "NVIDIA", "HBM", "先进封装", "TSMC", "半导体", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "半导体产业链的供给侧约束与国产化替代的技术边界", "summary": "本卡对H20/HBM管制下中国国产替代的技术边界做压力测试，部分否定「难以逾越的断层」框架。结论分三层：先进封装（CoWoS替代）是中国相对优势、是可在1–2年跨越的桥（JCET/通富/盛合晶微＋Naura混合键合），并非断层；真正断层在上游——EUV光刻国产化<5%（SMEE约90nm，落后ASML约十年）与高端ArF/EUV光刻胶约95%依赖日本；而专…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30, "matchedTerms": [ "HBM", "先进封装", "半导体", "设备", "semiconductors", "ASML" ], "href": "" }, { "date": "2026-06-28", "title": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "summary": "电力与电网容量成为稀缺投入后，AI 芯片优化目标从绝对算力转向每瓦性能（perf/W）与单机柜 token 经济性。NVIDIA 宣称 Blackwell 推理能耗较 Hopper 最高低 25×（FP4/NVFP4 驱动），但 B200 单卡功耗仍升至约 1,200 W（H200 为 700 W），机柜功率密度升至 120–142 kW 并规划 1 MW…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 29.5, "matchedTerms": [ "NVIDIA", "HBM", "先进封装", "半导体", "semiconductors" ], "href": "" }, { "date": "2026-06-28", "title": "半导体国产替代对AI算力瓶颈的缓解潜力", "summary": "支持但有限定：国产半导体设备与先进封装确实有效缩短AI算力兑现周期的硅片侧环节。设备国产化率2024约26%→2025约35%，刻蚀/薄膜约40%本土，北方华创(002371.SZ)全球第五、中微(688012.SH)前十三，大基金三期3440亿托底；先进封装为弯道超车路径，长电(600584.SH)XDFOI 4nm量产、良率约98.5%，国产CoWoS…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "HBM", "先进封装", "半导体", "设备", "semiconductors" ], "href": "" }, { "date": "2026-06-28", "title": "绝对规模虽属小型配套，但PQC与安全自动化是决定数千亿GPU集群网络效率的非对称闸口因子，应在配置中享有溢价并分配1.5%至3.0%的主动权重。", "summary": "2026-06-29 策略评估：在2026年超大规模云厂商6500亿至9000亿美元的资本支出（其中约75%为AI相关）背景下，仅有十亿美元级别的DPU和证书管理软件市场在绝对规模上只属于'小型配套预算'（约占AI CapEx的1.5%）。然而，从效率和合规角度看，它们是关键的'闸口因子'。PQC迁移及证书缩短带来的延迟膨胀，需要敏捷硬件与自动化管理来避免…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 27.75, "matchedTerms": [ "云厂商", "GPU", "AI capex", "半导体", "capex" ], "href": "" }, { "date": "2026-06-27", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交期[S6][S7]限制的是通电而非下单。CoWoS售罄、交期…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35, "matchedTerms": [ "AI 算力", "HBM", "先进封装", "GPU", "半导体", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "结构性再贷款提供的低成本流动性无法有效突破高端设备禁运的技术壁垒，制裁压力下国产HBM良率降至2.66%且大模型训练效率折损25%-30%，需通过在科技板块内超配CPO光模块等微观对冲平抑风…", "summary": "本报告对结构性再贷款支持下半导体“硬科技”在外部制裁压力下的对冲效果进行压力测试。分析显示，虽有5000亿再贷款及1.5%贴息将企业融资成本降至0.25%-0.50%[S1]，但流动性无法突破技术封锁。在先进TBDB与CMP设备禁运下，国产8-Hi HBM3封装良率由16.86%骤降至2.66%[自测算]，先进封装经济利用率降至35%-45%[自测算]，大…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "HBM", "先进封装", "半导体", "设备", "semiconductors" ], "href": "" }, { "date": "2026-06-25", "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元在手订单）；第二档择优超配高端光互联（1.6T之年、龙头锁单…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 41, "matchedTerms": [ "云厂商", "HBM", "先进封装", "GPU", "ASIC", "Micron", "半导体" ], "href": "" }, { "date": "2026-06-25", "title": "电力稀缺不会杀死AI芯片周期，但会把竞争重心从单纯峰值算力转向固定兆瓦内的有效吞吐和能效，从而收窄可投资赢家。", "summary": "截至2026-06-25，本卡对电力和并网约束做半导体侧压力测试：约束不会终结AI芯片需求，但会把胜负指标从峰值FLOPS推向每兆瓦吞吐、每机柜token、HBM每瓦带宽和部署可用性。NVIDIA的机柜级平台、Google TPU、AWS Trainium、TSMC背面供电和Micron HBM4都显示产业已转向能效优化。结论是保留AI存储、高端互联、先进…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40, "matchedTerms": [ "NVIDIA", "HBM", "先进封装", "TSMC", "Micron", "半导体", "semiconductors" ], "href": "" }, { "date": "2026-06-25", "title": "电力设备瓶颈不会系统性消灭AI加速器需求，但会把2026-2028年约15%-25%的GPU/HBM/ASIC部署变成收入确认和爬坡时点风险。", "summary": "截至2026-06-25，本卡从半导体侧压力测试电力设备瓶颈传导：2026-2028年加速器等效部署中约15%-25%面临2-4个季度时点风险，真实终端需求毁灭约3%-7%。商用GPU最直接受通电MW影响，HBM因多年协议和稀缺配额更像库存节奏风险，自研ASIC既受益于每瓦token优化，也因客户和站点专属性承担20%-35%新爬坡时点风险。结论是需求没有…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34, "matchedTerms": [ "HBM", "GPU", "ASIC", "半导体", "设备", "semiconductors" ], "href": "" }, { "date": "2026-06-25", "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出保护的公用事业更偏时点风险资产。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "HBM", "先进封装", "GPU", "ASIC", "设备" ], "href": "" } ], "timeline": [ { "date": "2026-04-24", "supports": 0, "risks": 0, "neutral": 2, "value": 48, "events": [ { "title": "NVIDIA Rubin 架构前夕的 AI 算力市场调研", "summary": "NVIDIA Rubin 架构前夕的 AI 算力市场调研 随着 Rubin 架构发布临近，超大规模云厂商在自研 ASIC 与采购 NVIDIA 芯片间的动态博弈及对 HBM4 产业链的带动作用？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026年2.5D/3D先进封装（CoWoS）产能扩张与HBM4演进：评估次世代AI加速器供应链的稳定性", "summary": "2026年2.5D/3D先进封装（CoWoS）产能扩张与HBM4演进：评估次世代AI加速器供应链的稳定性 随着TSMC CoWoS产能持续紧张且HBM4于2026年开始量产，从HBM3e向HBM4的跨越将如何影响主要AI GPU厂商（如NVIDIA/AMD）的成本结构和交付周期？哪些二级OSAT厂…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-03", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "2026年4月 hyperscaler AI capex 重估", "summary": "本卡由 tmt-analyst 作为开题分析师完成。结论是：2026年 hyperscaler AI capex 仍是需求驱动周期，而非简单泡沫破裂；Microsoft、Alphabet、Amazon 和 Meta 的 2026 capex 大致落在 $695-725 billion 区间，且由 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "2H26 订单簿已经覆盖 AI 物理供应链的整柜、液冷、光模块和部分国产设备，但 EPS 缺口集中在高毛利瓶颈件而非最大收入钱包。", "summary": "本卡支持 Research note 1 的 AI 资本开支延续判断，但强调供应链利润转化分化。2H26 订单可见度已覆盖服务器 ODM/整柜、液冷、800G/1.6T 光模块和国产刻蚀/沉积/清洗设备；其中光模块和液冷组件 EPS 弹性最好，国产半导体设备次之。ODM 收入可见度强但毛利穿透弱，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026下半年AI资本开支拐点辩论：消化期还是再加速？", "summary": "TMT分析师判定，2026下半年AI资本开支并未进入消化期，而是进入结构轮动期：总量持平至上行，边际增量美元从训练优化的商用GPU集群转向推理fleet、超大规模云厂商自研芯片、以及电力与场地准备。空头叙事在需求侧（推理token volume、agentic渗透、多模态）证据不足，但在前沿训练边…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 1, "neutral": 0, "value": 48.2, "events": [ { "title": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯…", "summary": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯片竞争格局：NVIDIA Rubin对阵云巨头ASIC 本卡主题： WFE资本开支周期——先进封装与HBM设备订单可…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "云端仍是 AI 算力的定价权锚，边缘只获得 2027 年低基数盈利修正机会，因此港美股配置应从单边 AI beta 转向云端瓶颈核心持仓加选择性边缘 spread。", "summary": "本卡综合前序结论：2027 年边缘 AI 和 AI PC 的确有分母红利，但全球 AI 算力定价权仍集中在云端基础设施、HBM、先进封装、网络和模型编排。组合上不应简单“卖云买边缘”，而应采用 barbell plus spread：45-55% 保留在 NVDA/AVGO/TSM 与 MSFT/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "2027 半导体盈利修正地图：从云端 capex 到 AI PC 与边缘设备 第 6/8 张卡 — 压力测试 | TMT 行业分析师 立场：对第 5 张卡的哑铃配置进行 DRAM/S…", "summary": "2027 半导体盈利修正地图：从云端 capex 到 AI PC 与边缘设备 第 6/8 张卡 — 压力测试 | TMT 行业分析师 立场：对第 5 张卡的哑铃配置进行 DRAM/SSD 及先进封装瓶颈压力测试 执行摘要 第 5 张卡提出\"哑铃加价差\"组合：45–55% 云端核心（NVDA/AVG…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 1, "value": 46, "events": [ { "title": "材技一体化对fabless是选择性降权——AI加速器与EDA/IP溢价稳健，射频/功率/模拟/成熟节点向IDM收敛，资金应在设计层内向两个可防御极轮动。", "summary": "材技一体化情景下，fabless估值溢价并非整体崩塌而是结构性分化。AI加速器龙头（NVIDIA、AMD MI、Broadcom ASIC）凭借云厂商需求无弹性和CoWoS/HBM优先配额，能够将晶圆和封装涨价完整转嫁，溢价稳健；EDA/IP分成（Cadence、Synopsys、Arm）现金流与…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "算力期货压缩的是商用GPU层租金，先进封装与高带宽互联的物理瓶颈与寡头结构足以将其非商品化毛利结构维持至2027。", "summary": "支持card-01框架并加以收紧：算力期货标准化的是GPU-小时这一输出单位，并未商品化决定谁能以最低成本清算的物理输入。CoWoS先进封装、HBM4、混合键合设备、scale-up交换机ASIC、PCIe/CXL retimer、1.6T/CPO光模块均具18–36个月结构性导入周期，寡头格局清…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 4, "neutral": 0, "value": 26.4, "events": [ { "title": "35% AI/成长仓位整体成立，但应集中到先进封装、光互联、国产逻辑代工、AI ASIC等上游AI基础设施层，避开消费电子相邻半导体与商品存储，并对光模块设减持纪律以应对2H-26产…", "summary": "截至2026-05-15，35% AI/成长仓位通过Capex压力测试但需收紧结构。中国云厂商（BABA、Tencent、Baidu、ByteDance、三大电信SOE）2026年合计AI Capex预计同比+45%–70%（合计~RMB 700–830 bn），由AI加速卡、光互联、液冷供电驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股半导体重估的 beta 已基本被定价，下一段必须靠后道设备、量检、材料、车规功率与先进封装上游材料的盈利与订单兑现，回避换手/融资/解禁三重拥挤的设计、代工与映射封装。", "summary": "截至 2026-05-15，A股半导体经过 4-5 月重估后，国产 GPU/ASIC 概念、HBM/先进封装映射股、大市值晶圆代工四条线已拥挤：5 日换手率分位 81-92%，半导体两融占自由流通市值比 ~4%（88 分位），主要芯片 ETF 4 周连续净申购 +18%，且 6-8 月解禁集中（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "M2M流量爆发的边际成本不会均匀挤压SaaS——它会让薄壳代理编排厂商利润坍缩、让有自有数据的垂直SaaS继续吃到按量计费红利，并把被争夺的利润集中流向芯片、封装与光模块基础设施层。", "summary": "截至2026-05-16，本卡对card-07的SaaS基本面Beta判断做压力测试。结论是有条件成立但显著分化：薄壳代理编排SaaS将因无法转嫁的token成本，在未来12–18个月承受600–1,200 bp毛利压缩；而垂直SaaS、数据合规和API适配器（card-06超配名单）边际成本以存…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "存量博弈下应从“拥挤的AI动量”切换至“可验证的瓶颈Alpha”，并配对红利资产对冲6-8月解禁悬崖。", "summary": "截至2026-05-15，A股半导体Beta重估已基本定价，面临5-6月TSMC高基数营收检验、交易结构过热（90分位换手）及6-8月1800亿解禁压力。策略建议构建“杠铃结构”：减配高拥挤的国产GPU及代工映射，增配后道测试、量检设备及先进封装上游材料等“真实瓶颈”Alpha，并配对25-30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 3, "neutral": 0, "value": 33.8, "events": [ { "title": "AI 硬件盈利端脱钩成立，但价格路径脱钩不完整；基本面能缓冲两融与解禁冲击，不能消除冲击。", "summary": "截至2026-05-17，本卡压力测试 Research note 06：AI 硬件基本面强到足以形成先进封装、HBM 供应链、光模块/AI datacom 的窄口径例外，但不足以让广义半导体 beta 脱离 A 股两融拥挤和解禁供给。NVIDIA FY2026 收入 USD 215.938bn、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[2026-05-16] 电力基建硬约束正催化AI算力形态的分化，云端推理加速向定制硅（ASIC）倾斜，边缘设备承接下沉需求，投资主线从单一GPU向ASIC供应链与端侧硬件转移。", "summary": "[2026-05-16] 综合前序产业链与基建结论，电力瓶颈未消灭AI CAPEX，而是驱动了算力向高能效的云端ASIC与边缘端转移。ASIC设计（AVGO、MRVL）及边缘硬件（AI PC/手机）将迎来由“每瓦算力”主导的渗透率实质性跃升，建议在算力仓位中进行结构性增配。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链基本面与订单流：对减仓建议的支持性验证 日期： 2026 05 17 分析师： TMT行业分析师 [Primary] 对Research note 01立场： Suppor…", "summary": "AI算力链基本面与订单流：对减仓建议的支持性验证 日期： 2026 05 17 分析师： TMT行业分析师 [Primary] 对Research note 01立场： Support（支持） 结论： 基本面边际开始出现裂缝——印证因子拥挤度去风险建议 1. 摘要 Research note 01…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 2, "neutral": 0, "value": 41.2, "events": [ { "title": "AI硬件与算力基础设施具备足够的选择性资本确定性，但应买入瓶颈环节和已验证国产替代，而不是泛AI主题贝塔。", "summary": "截至2026-05-18，本卡片支持从大金融防御性收缩转向盈利可见成长。AI硬件具备选择性资本确定性：NVIDIA、TSMC、SK hynix及800G/1.6T光模块已体现高毛利兑现，云厂商2026资本开支提供需求锚。国产化替代真实存在，但确定性集中在服务器、光模块、热管理、电源、PCB和已商业…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 资本开支和利润率足以支撑已验证算力基础设施链条的估值修复，但不足以支撑泛 AI 成长股普涨。", "summary": "2026-05-17：本卡对 Research note 7 的“利率波动见顶后配置高现金流长久期成长”做压力测试，结论为有条件支持。AI 算力链条的盈利确定性主要集中在硬件和基础设施层：NVIDIA、Broadcom、TSMC、工业富联、中际旭创等已体现收入、利润率和经营现金流共振；但 hype…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 1, "risks": 2, "neutral": 0, "value": 50.6, "events": [ { "title": "截至2026-05-19，AI半导体供应链尚未出现云CapEx或HBM/CoWoS订单环比下修，算力基础设施交易仍应保留核心仓位但需严密跟踪ROIC与Q2/Q3指引风险。", "summary": "截至2026-05-19，首席策略师提出的敏感触发点有效，但尚未触发。Microsoft、Alphabet、Amazon、Meta的2026年AI基础设施CapEx仍在上修或维持高位，TrendForce估计前八大CSP 2026年CapEx超过7100亿美元、同比约+61%。HBM方面，Micr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "TMT 已拥挤到足以为物理基础设施轮动出资，但尚未拥挤到需要放弃 AI 硬件周期；边际 AI 资金应从公开市场半导体动量转向电力、电网设备和 firm-capacity 受益者。", "summary": "截至2026-05-18，TMT 并非整体泡沫，但 AI 半导体和存储交易已达到拥挤区间：SMH P/E 38.56、前十大持仓 71.66%，DRAM ETF 五周吸金超过 $6bn，半导体/存储贡献 2026 年 S&P 500 市值增量的 70%。同时，NVIDIA、TSMC、MSFT、GO…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "半导体供应链的先进封装、HBM、EUV、基板和光模块交期同步拉长，与电力侧'承诺 Capex > 现金 Capex'呈现同一指纹，证实 2H26 AI Capex 下修是交付推后引致…", "summary": "用先进封装、HBM、WFE、ABF 基板与光模块交期数据压力测试卡片 01，结论是硅侧呈现与电力侧完全一致的'承诺 > 交付、交期拉长、现金落地右移、订单簿创新高'指纹：CoWoS 2026 配额售罄至 1Q27、ASML EUV 交付排到 2H27、HBM4 在 Rubin 上认证再顺延一个季度…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "AI算力产业链基本面显著强于单纯零售泡沫，但在零售杠杆承接外资流出的结构下，订单流只能支持龙头选择性配置，不能完全抵消板块估值压缩。", "summary": "截至2026-05-19，半导体与AI算力产业链订单流足以支撑核心龙头盈利，但不足以全面对冲外资流出和零售杠杆承接带来的估值压力。NVIDIA、TSMC、Broadcom、AMD、Dell、Micron、ASML和SEMI数据共同显示AI算力需求仍受供给约束；寒武纪和SMIC也验证中国国产AI算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 2, "neutral": 0, "value": 45.4, "events": [ { "title": "AI与半导体能在防御性轮动中维持Alpha，但安全边际只存在于盈利可见度高、供给受限且能把CapEx转化为现金流的瓶颈环节。", "summary": "2026-05-20，本卡支持第4张卡的杠铃配置：AI与半导体可保留为15-20%的Alpha袖珍仓位，但不能作为泛防御Beta。SIA/WSTS显示2026年一季度全球芯片销售2985亿美元，环比+25%；TSMC 1-4月收入同比+29.9%，并指引2026年美元收入增长30%以上；NVIDI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链盈利可见性已被订单簿三重验证、足以支撑保留成长端的策略；半导体设备应拆分成‘AI直接受益+国产替代超配，纯中国敞口设备与晶圆厂低配’，故第03卡‘不全面切向价值’成立但需要…", "summary": "截至2026-05-21，本卡对第03卡“不全面切向价值”进行压力测试，结论是有条件支持。AI算力链（GPU/HBM/光模块/液冷/先进封装）盈利可见性高且仍在抬升：四大美系hyperscaler 2026年合计资本开支指引上修至USD 360–410 bn、TSMC CoWoS与SK Hynix…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 2, "risks": 1, "neutral": 0, "value": 68.8, "events": [ { "title": "截至2026-05-24，AI硬件与半导体设备是A股结构性成长修复最可信的核心驱动力，但只能买订单和盈利修复，不能买泛科技估值扩张。", "summary": "截至2026-05-24，本卡支持ashare-strategist的结构成长修复判断，但将可投资主线收窄为有订单验证的AI硬件与半导体设备。NVIDIA、Microsoft、Meta、Amazon、TSMC、ASML、SEMI和SIA数据共同显示，全球AI capex、先进制程、先进封装、测试与…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "NVDA 这份印数是 AI capex 上行周期的高置信度确认；FY26 H2 真正的盈利杠杆在 CoWoS / HBM / 800G 光 / AI 服务器 ODM 这条上游链上，不…", "summary": "NVDA Q1 FY27（盘后 2026-05-20）数据中心收入 US$75.2 bn、同比近翻倍、占总收入 92%，新增 US$80 bn 回购授权，是 AI capex 上行周期的高置信度确认。2026 H2 真正的盈利杠杆落在上游：TSMC CoWoS、SK 海力士/三星/美光 HBM、A…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "'效率降级'对 2026–2027 AI Capex 总量近乎无约束，但通过电力、关税、航运三条渠道侵蚀单位经济学，使全球 AI 基础设施从'同步扩张'转为'地理极化'，估值真正承压…", "summary": "本卡对 Research note 1–5 的“效率降级削弱 AI 算力扩张”假设进行双向压力测试。结论：方向正确但程度被高估，AI 算力 Capex 在 2026–2027 名义增速仍 +19–26% YoY（北美 hyperscaler 2026 指引 4,400–4,650 亿美元 [S1–…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 3, "neutral": 0, "value": 42.2, "events": [ { "title": "TMT / AI硬件已从政策主题进入盈利兑现阶段，足以承接医药降配资金，但应先投100-150bps并等待A股拥挤度验证后再投满。", "summary": "截至2026-05-24，支持将医药降配释放的风险预算轮动至TMT / AI硬件，但建议分阶段执行。政策端，“两新”和“人工智能+”继续提供订单底；数据端，2026年1-4月设备工器具购置投资同比增长11.5%，1季度电子设备制造业利润同比增长1.2倍；全球端，NVIDIA、TSMC、ASML均显…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体板块在高利率与中美并购友好政策共振下，正在以更快节奏复现生科板块的哑铃式价值重估——整合者赢家通吃、后期去风险标的被并购溢价重新锚定、中间产能被同步出清。", "summary": "截至2026-05-24，本卡片自下而上验证半导体板块正在高利率（UST10Y 4.56%）背景下复现生科哑铃逻辑。证据集中在三层：(1) 并购重组——Synopsys-Ansys ($35B)、AMD-ZT Systems ($4.9B)、HPE-Juniper ($14B)、Silver La…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施资本开支尚未在订单层面见顶，但在扩产加速度和重资产风险收益上已接近顶部，2027年放缓将优先打击低毛利、高折旧、专用性强的制造环节。", "summary": "截至2026-05-24，本卡从工业制造与Capex Cycle角度应力测试AI基础设施扩产。结论是：已披露需求和行业基准预测尚未显示周期见顶，SEMI仍预计2026-2027半导体设备支出上行；但资本开支加速度、低毛利跟随者扩产和数据中心建设成本已呈现后周期特征。若2027年需求放缓，最严峻的减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 1, "value": 52.6, "events": [ { "title": "现有先进半导体出口管制不会停止AI使用或全部ASIC设计，但会系统性削弱受限生态通过HBM密集、CoWoS级封装实现“每瓦有效推理”跃迁的能力。", "summary": "截至2026-05-25，本卡压力测试前序“ASIC与先进封装推动每瓦有效推理”的判断。结论是：路径对可获得TSMC、HBM与CoWoS产能的全球云厂商仍成立，但对中国相关或D:5受限生态被结构性削弱。美国出口管制已从成品AI芯片延伸至HBM、封装路径、先进制造设备、软件/技术和FDP规则，卡住了…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？", "summary": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？ DeepSeek-V4-Pro 75% price cut × NVDA FY27Q1 Data Center +92% joint p…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 2, "neutral": 1, "value": 54, "events": [ { "title": "半导体瓶颈不在 GPU 而在 CoWoS-L 与 HBM，光模块瓶颈不在出货而在 1.6T 交付周期与客户独家配套——这两条链是 Research note 06 中段瓶颈租框架下 …", "summary": "2026-05-26：本卡 support Research note 06 的'中段瓶颈租'框架，并在 TMT 半导体与光模块层将之定量化。核心结论：CoWoS-L 月产能从 2025 年约 40kwpm 升至 2026 约 75kwpm 仍被超额预订（覆盖率 1.4–1.6x），HBM 长协覆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "短期硬约束在 HBM 与 CoWoS-L 而非电力，AI 算力'紧但能出货'，对指数 FOMO 有利但 2026H2 脆弱性显著上升。", "summary": "短期（2026 年内）芯片侧的约束比电力更紧：依次瓶颈为 HBM3e bit 供给 → 台积电 CoWoS-L 先进封装 → GPU 晶圆 → 基板/光模块 → 电力。HBM 与 CoWoS-L 全年额度已锁、扩产跑不过 ~150%/~70% 的需求增长，2026H2 任一良率事件或出口管制重置可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 产业链利润正在按 firm MW、设备订单簿、电源效率 IP 和大用户 tariff 政治可防御性重新分配；2026–2027 年最强留存在电气设备 OEM、签 PPA 的 f…", "summary": "本卡综合 01–06 号卡，把 AI 产业链利润定价从单层'算力溢价'切换为按 firm MW、电气设备订单簿、电源效率 IP 和大用户 tariff 政治可防御性配给的四层结构。2026–2027 年利润留存最强的是 GEV/ETN/VRT 等订单簿已锁的电气设备 OEM、CEG/VST/TLN…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-27", "supports": 1, "risks": 4, "neutral": 0, "value": 48.4, "events": [ { "title": "4.5%以上美债环境下，AI硬件龙头的业绩增长足以支撑选择性配置，但不足以为整个AI久期交易兜底。", "summary": "2026-05-28结论：本卡对上一张宏观卡采取stress-test但不否定。4.50%美债锚会压缩AI久期资产估值，但NVIDIA、TSMC、Broadcom、SK hynix/Micron等稀缺硬件链凭借高增长、真实云厂商capex、先进封装和HBM瓶颈，仍能部分抵消估值下修。风险在于cap…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI仓位可以保留，但必须集中到已经把AI资本开支兑现为高毛利率收入和现金流的半导体瓶颈资产。", "summary": "截至2026-05-27，本卡支持“AI保权重”，但主张把AI仓位从广义beta收窄到盈利兑现型硬件龙头。最清晰路径是NVIDIA、TSMC、Broadcom，其次为SK hynix/Micron和ASML；Dell可做战术服务器渠道敞口，AMD列入观察型挑战者，Supermicro等低毛利率、重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "台积电CoWoS与HBM底层封装良率构成全球AI芯片出货的硬性物理约束，超额利润集中于垄断性瓶颈厂商（TSMC/MU/SK Hynix），而A股代理股共享叙事却无壁垒且交易极度拥挤，…", "summary": "支持超配海外半导体物理瓶颈（MU/SOXX）并降配A股算力代理。台积电CoWoS封装产能从2024年底3.5万片/月扩至2026年底超12万片/月但仍售罄，NVIDIA垄断超60%份额；HBM市场2026年规模将达600亿美元，美光产能售罄且良率升至70-75%。A股光模块、服务器组装与IDC代理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "目前推动新高的是有现金支撑的AI capex流而非已扩散的生产率，市场为窄圈定价合理、为隐含宏观面定价过窄；共存可再持1-2季度，应『参与+品质对冲』而非追指数。", "summary": "Conference Board CCI 5月降至93.1、约2/3美国人削减开支、信用卡90天逾期升至15年高的13.1%，与S&P 500/Nasdaq/ACWI/Micron新高并存。判断：抵消消费拖累的是7,250亿美元AI capex流（同比+77%），而非AI生产率扩散——BLS Q1…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 59.8, "events": [ { "title": "2026年AI芯片与先进封装产能约束会推高并拉长云端CapEx，而不是使其取消；配置应偏向具备配额权的上游瓶颈环节。", "summary": "截至2026-05-28，本卡支持前序判断：AI硬件CapEx传导链的约束真实但不等于削减。Microsoft、Amazon、Alphabet、Meta的2026年CapEx指引中枢约为$710 billion，反映的是芯片、HBM、先进封装、电力和数据中心资源稀缺导致的锁单与涨价，而非需求走弱。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "AI硬件周期仍由需求驱动，但未来12个月的定价权将更多落在HBM、先进封装、光模块和电力交付等瓶颈环节，而不是单纯GPU或服务器组装收入。", "summary": "截至2026-05-30，本卡支持“AI算力瓶颈从硅端扩展至物理基础设施”的主线。未来12个月，AI芯片和服务器需求仍由超大云厂商资本开支、云AI收入、agentic AI推理流量和自研ASIC扩张驱动；但供应瓶颈已从GPU单点扩展到HBM、CoWoS/先进封装、800G/1.6T光模块、机架供电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "16% 仓位上限是必要但不充分的防御 —— 唯有加上子桶集中度上限、个股层断电对冲与桶层熔断器三条硬约束后，才能真正把组合尾损压在 −1.5% 目标附近。", "summary": "本卡对配置师 16% 半导体桶做压力测试。结论：16% 顶层上限正确，但默认子桶结构（AI 算力 28% 集中在寒武纪 688256、海光 688041）使桶真实波动达 ~44%，而 ERC 假设仅 ~30%。三情景测算：STAR50 −20% 有序去估值（情景 A）和缓跌 −15%（情景 C）可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 5, "neutral": 1, "value": 43, "events": [ { "title": "超级IPO将把AI公司的公开融资压力传导为算力锁定需求，从而重估半导体资本开支：超配稀缺瓶颈，回避无法证明产能稀缺和合约期限的长尾扩产。", "summary": "截至2026-05-31，本卡片支持“超级IPO流动性虹吸”对AI硬件和半导体瓶颈资产的强化作用。核心判断是，IPO资金不会平均利好半导体，而会把公开市场资金约束转化为GPU、HBM、先进封装、先进制程、光刻和数据中心网络的多年期需求可见度。TSMC、NVIDIA、ASML、Micron、Sams…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "多云路由迁移并未在短期破坏NVIDIA绝对需求，但已通过Anthropic约10GW非NVIDIA硅片合同与推理占算力2/3的转折，结构性重写NVDA增量份额轨迹——NVDA与AVG…", "summary": "压力测试'模型层多极化→ASIC蚕食NVIDIA'的简化叙事。FY27 Q1 NVDA数据中心营收$75.2B同比+92%、Blackwell占DC计算约70%、每家超大规模云每周部署约72,000颗Blackwell，证伪短期需求断裂论。但Anthropic 10GW算力已合同锁定在Google…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力供给受限于 HBM4 售罄、CoWoS 封装良率及全球电网容量的物理极限，使硬件龙头成为 AI 价值链中无视‘Capex 断崖’风险的最终收费站。", "summary": "截至2026年5月，AI硬件供应链仍处于结构性紧缺。TSMC N2量产初期良率65-70%，HBM4于2月投产且美光2026年产能已售罄。NVIDIA Rubin架构推升单GPU功耗至2.3kW，使电力供应成为比芯片更硬的瓶颈。尽管终端应用出现DAU增速放缓，但硬件端的实物稀缺性构成了核心护城河，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 1, "risks": 0, "neutral": 0, "value": 79.4, "events": [ { "title": "AI实物建设仍支持芯片需求，但2026-27年的边际份额赢家是定制硅和网络而非商用GPU，因此B层维持中性、内部偏好AVGO/MRVL并对NVDA估值更谨慎。", "summary": "截至2026-06-02，本卡支持Research note 01的B层中性，但建议改为杠铃式中性：更建设性看Broadcom和Marvell等ASIC赋能商，对NVIDIA保持近端战术正面、结构估值更谨慎。2026年定制ASIC份额27.8%大概率守住；按44.6% ASIC增速和16.1%商用…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 5, "neutral": 0, "value": 42.8, "events": [ { "title": "CoWoS产能硬约束会把AI硬件估值溢价从单纯设计叙事重新分配给掌握晶圆、封装、良率和整柜交付能力的产能控制者。", "summary": "2026-06-03：本卡支持前序判断，认为CoWoS与先进封装已成为AI算力链估值的硬过滤器。NVIDIA需求仍强，TSMC的3 nm、HPC与先进封装紧张共同支撑产能控制者溢价；但HBM4供应商扩容削弱单纯存储稀缺溢价。A股半导体整体估值已高，策略重点应从宽泛芯片设计beta转向设备、材料、封…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Broadcom (AVGO) 财报验证了AI订单总需求未断裂，但低毛利自研芯片的占比扩张和物理并网/封装刚性约束，正在系统性拉低硬件供应商的边际利润率并平滑其收入转化率，驱使AI算…", "summary": "截至2026年6月3日，Broadcom (AVGO) 发布Q2财报，其AI半导体营收111亿美元超预期，且上调全年指引至145亿美元，证伪了需求崖断裂假说；但其毛利率受低毛利定制加速器 (ASIC) 占比飙升影响，环比大跌290个基点至75.1%低于预期，且受限于CoWoS先进封装与北美电网并网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链的绝对瓶颈正在从HBM存储晶圆短缺结构性转移至由TSMC主导的CoWoS-L先进封装尺寸扩张和逻辑基础晶圆制造代工能力。", "summary": "本报告分析了NVIDIA在2026年6月确认Samsung、SK Hynix和Micron全面进入Rubin平台的HBM4供应链这一关键拐点。我们指出，随着三足鼎立打破垄断，HBM供需瓶颈将显著缓解。然而，由于HBM4基础晶圆逻辑代工的双重锁定，以及CoWoS-L封装面积扩张对晶圆利用率(GDW)…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施利润扩张可以对冲成本压力，但只在加速器、先进代工、存储和电力/热管理等稀缺环节成立，不能外推到整个电子产业链。", "summary": "2026-06-04，本卡对AI硬件与电子产业链做成本压力测试：支持从公用事业红利向盈利成长轮动，但只支持精选AI硬件杠铃，不支持宽基电子超配。NVIDIA、TSMC、Micron等上游利润池具备高毛利与供给稀缺，能够吸收部分原材料、电力和元器件成本；但Microsoft、Meta、Alphabe…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 1, "risks": 1, "neutral": 0, "value": 76.2, "events": [ { "title": "半导体设备 2026 年订单增长具备需求与国产化双重确定性，但应聚焦已客户验证的国产工艺设备平台，而非泛化到光刻和最高端量检测。", "summary": "2026-06-05 结论：支持 Card-04 将半导体设备列为设备更新订单弹性第一名，但需收窄到刻蚀、沉积、清洗、热处理、离子注入、涂胶显影和先进封装等国产替代可兑现环节。AI 算力链条由 NVIDIA、Dell、TSMC 与 SEMI 数据共同验证，国内晶圆厂由中芯国际、华虹高装载率支撑，政…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "四大云厂 2026-2027 年 AI Capex 增速见顶并进入“高原期”，整体加速器 TAM 从爆发转向存量博弈，ROI 拐点迫使云厂商加速供应链多元化以压降成本。", "summary": "综合宏观与资本流向，预计四大云厂（Top 4）在 2026H2 至 2027 年的合计 AI Capex 同比增速将从 2024-2025 年的高位骤降至 12% 左右甚至个位数，总量触及约 2400-2500 亿美元的“高原期”。这主要受制于宏观利率粘性带来的资金成本压力，以及前期 GPU/AS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "SK hynix HBM4 12-Hi良率大概率足以支撑Rubin初始交付，但不足以排除2026年Q4至2027年Q1放量斜率受压。", "summary": "截至2026-06-06，公开证据不支持SK hynix HBM4 12-Hi已出现确定性良率失误；更合理的压力点是Rubin放量斜率。NVIDIA已称Vera Rubin进入全面量产爬坡，并有三家HBM4供应商获认证/生产。SK hynix的1bnm、Advanced MR-MUF、TSMC b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 2, "neutral": 0, "value": 65, "events": [ { "title": "7月关税与海运上行更可能压缩AI硬件下游组装与网络设备利润率，而非击穿核心半导体龙头利润率；估值风险来自低利润率硬件收入增长的质量被重新定价。", "summary": "截至2026-06-08，本卡对第04卡的关税/海运鹰派尾部做TMT压力测试：风险真实存在，但主要不是核心GPU、晶圆代工或HBM利润率同步塌陷，而是下游AI服务器、网络设备、机柜、电源、散热和PCBAs的落地成本挤压。NVIDIA、TSMC、SK hynix拥有高毛利率和稀缺定价权；Dell式A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在核心零部件国产化偏低、N+2工艺良率不足50%以及先进封装产能短缺的制约下，纸面设备订单并不能等额转化为实际芯片产能，面临行业资本开支回报率收紧后的去库存与砍单风险。", "summary": "截至2026-06-07，半导体设备厂订单高增与晶圆厂扩产周期明确，但面临转化瓶颈。由于设备核心零部件国产化率低、DUV多重曝光导致N+2先进制程大芯片良率仅30%-40%、且国内先进封装产能存在30%缺口，导致订单向产能的物理转化效率受限。若下半年工艺突破不及预期，随着全球AI CapEx回报率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 72.4, "events": [ { "title": "Rubin Capex能保护一线AI硬件EPS，但WACC上行会把二线供应链从“AI成长资产”重新定价为“融资周期资产”。", "summary": "截至2026-06-08，本卡对前序AI硬件防御结论做压力测试：Rubin周期与云巨头Capex仍能支撑NVIDIA、TSMC、Broadcom、Samsung、Micron等一线硬件龙头的EPS增长，但不能让整个供应链免疫WACC上行。可见四大云厂商Capex底线约5742亿美元（自测算），足以…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 2, "neutral": 1, "value": 78.8, "events": [ { "title": "AI半导体利润池迁移真实但渐进，板块总EPS上行、估值锚从'NVDA-TSMC双寡头溢价'切向'增长+多元化第二供应源'，ASIC设计商、HBM、先进封装设备与Intel Found…", "summary": "2026-06-09综合：Google-Intel TPU订单确认AI硬件利润池迁移真实且渐进，2027-28年AI半导体合计EPS较冲击前共识更高，但分布坚定向ASIC设计商(AVGO/MRVL)、HBM、先进封装设备及Intel Foundry期权倾斜。新估值锚为'增长+多元化第二供应源'，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球科技配置应超配ASIC与新代工生态；A股风格则需从“AI题材映射”战略性切换至独立算力的“国产第二供应源”，重仓先进封装产能、核心设备及HBM材料。", "summary": "2026-06-09评估：基于AI半导体估值锚向“增长+第二供应源”切换，全球配置需减配纯GPU倍数扩张与单一代工依赖，转而超配ASIC双雄(AVGO/MRVL)、HBM及Intel代工期权。在A股与港股层面，风格将从“英伟达映射”(光模块/代工)实质性轮动至业绩兑现度高的国产“第二供应源”。策略…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "Google-Intel的TPU订单标志着AI硬件利润池的结构性迁移：从绝对依赖GPU/TSMC，转向ASIC出货量反超以及Intel Foundry作为第二供应源的崛起，但GPU仍…", "summary": "2026-06-09评估：Google向Intel Foundry下单2028年超300万颗TPU，且Nvidia正评估Intel 18A工艺，这表明AI总需求依然强劲（到2028年TAM达3800亿美元），但硬件利润池正发生结构性迁移。为降低TCO，云厂商加速自研ASIC，出货量预计将超越GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 2, "neutral": 1, "value": 72.2, "events": [ { "title": "AI半导体核心标的的Q2/Q3盈利能见度极高，受限的电网电力将加速Blackwell架构升级，短期流动性冲击构成极佳的左侧建仓窗口。", "summary": "截至2026-06-10，在SpaceX IPO（SPCX）可能引发SMH/QQQ短期30亿至60亿美元流动性抽水的背景下，本报告研判AI半导体核心标的（Nvidia, TSMC）Q2/Q3盈利能见度依然坚实。TSMC 2026年520-560亿美元资本开支高企，CoWoS先进封装产能供不应求。即…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装管制的法律武器化（232条款全价值评估+EAR封装设备限制）叠加中国11月稀土管制悬崖，将使AI半导体供应链从“产能瓶颈”转向“合规与原材料双重绞杀”，构成比利率更严峻的估值…", "summary": "截至2026-06-10，先进封装（CoWoS）已成为AI供应链的防御核心与地缘政治新红线。由于美国232条款关税在2026年4月转向“全价值评估”模式，叠加11月中国稀土出口管制“50%豁免期”即将届满，H2 2026先进封装设备与基板面临出口管制升级的概率高达75%。预计美国将利用BIS EA…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件需求底足以支撑6-12个月反弹论据，但短期无法对冲科创50强平与折现率冲击的非对称下行，结构应为分批再介入而非全面低配。", "summary": "截至2026-06-10，AI硬件需求短期（4-8周）不足以对冲科创50局部去杠杆的折现率与强平资金流冲击，但中期（6-12个月）足以支撑反弹。美国四大云厂商2026年capex指引5100-5400亿美元、约75%已PO锁定，中国主权AI capex同比+45%、国产替代占比升至55-58%；H…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 4, "neutral": 0, "value": 58.6, "events": [ { "title": "能源融资成本上升会削弱云客户ROIC并压低部分ASP，但最强AI硬件平台商仍有足够利润率缓冲，真正脆弱点在重capex制造和周期性内存环节。", "summary": "2026-06-11，本卡压力测试能源债务融资成本上升对AI硬件供应商利润率的影响。结论是方向成立但不应外推为硬件利润率全面坍塌：新增能源债务成本上行约85 bps会压缩云厂商自由现金流并强化采购纪律，但NVIDIA和Broadcom因高毛利、平台锁定和轻资产结构韧性最强；AMD居中；TSMC和M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-11：电力与电网瓶颈会削弱半导体盈利确定性的斜率和估值，但尚不足以推翻半导体核心资产作为质量权益的配置逻辑。", "summary": "2026-06-11：本卡压力测试认为，AI 数据中心电力与电网瓶颈会压低半导体盈利确定性，但主要表现为投产节奏、算力成本和估值折价，而非短期需求崩塌。美国与全球数据中心用电快速上行，PJM、FERC、NERC 已显示接入和可靠性压力；但 NVIDIA、TSMC 与 hyperscaler cap…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO 更可能造成 A 股半导体估值重估，而不是打断设备与材料国产化采购周期；真正应做的是从宽基国产化 beta 转向订单可见度和验证进度。", "summary": "截至2026-06-11，本卡对上一卡“SpaceX IPO 抽走 AI 基建流动性”进行压力测试。结论是：半导体设备与材料国产化不会全面遭遇资金压力，而会分化。SpaceX 750亿美元 IPO 足以压缩 A 股半导体 beta 和未商业化硬科技估值，但全球 AI capex、SEMI 设备预测…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 0, "value": 60, "events": [ { "title": "2026-06-12核心判断：AI半导体订单未见坍塌，但去杠杆风险会优先打击现金转化弱、扩产重、估值依赖交期溢价的环节。", "summary": "2026-06-12：半导体链条的订单韧性强于股票交易本身。ASML、TSMC、Advantest、KLA、HBM存储和部分先进封装设备仍有强订单、利用率或现金流证据；但Amkor、Camtek、中微公司等环节显示资本开支和营运资本会吞噬现金，去杠杆冲击会先压缩估值和产能期权价值，而不是马上造成订…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的扩展性：电力与芯片供应瓶颈对超大规模厂商CAPEX的影响分析", "summary": "截至2026-06-12，本卡压力测试卡片02（电力瓶颈）与卡片06（半导体多线挤压）。结论：双重约束不会削减超大规模厂商资本开支，而是将其重排。2026资本开支仍约$725-726B（Dell'Oro），资本开支/营收比约25-30%、指引2027约35%；变化在优先级——可靠电力升至栈顶，建设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "为支撑三层算力织物，AI芯片与先进封装如何演进以匹配电力波动与边缘推理的功率包络", "summary": "综合并支持卡片04的三层算力织物，下沉到硅基与封装层。关键洞见：'电力波动友好'主要是电力电子＋调度固件问题，而非新晶圆问题。NVIDIA GB300靠电源货架（65 J/GPU储能＋功率封顶斜率＋功率烧除）把Megatron训练电网峰值砍约30% [S1]，DVFS早是标配 [S2]；同步功率摆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力刚性约束已把AI加速器竞争维度切换为每瓦Token，Chiplet+先进封装是2027-2028电网窗口内唯一可规模化兑现的算力密度救援方案，对NPV的杠杆是5-8倍。", "summary": "电网刚性约束把AI加速器的KPI从峰值FLOPS切换为每瓦Token吞吐。Blackwell B200单卡1000W、GB200 NVL72机柜120kW已触顶，下一代Rubin/MI400无法继续抬TDP，必须靠能效换性能。TPU v5p、Trainium2较H100能效提升30-40%；同10…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 4, "neutral": 1, "value": 64.4, "events": [ { "title": "CoWoS绝对产能正在改善，但GPU与自研ASIC需求吸收速度更快，因此2026年AI集群交付仍受先进封装、HBM和关键设备交期的结构性约束。", "summary": "截至2026-06-13，本卡支持上一张AI基建卡的判断：硬件交付风险不只来自并网，还来自CoWoS/HBM/设备交期的半导体转换链。TSMC仍称先进封装非常紧，TrendForce口径显示CoWoS-L/S满载且2026年底目标120,000-130,000片/月；DIGITIMES虽更乐观，预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红皇后竞速并未彻底抹去自研ASIC的成本优势，而是将其集中至Google与AWS两家、使其他云厂商回到对NVIDIA的依赖，从而部分推翻了Research note 05所述的全行业…", "summary": "在'红皇后竞速'下，自研ASIC单世代有效寿命已压缩至12–15个月，2nm节点全口径NRE攀升至USD 0.9–1.4B、N2晶圆约USD 30,000/片、CoWoS-L售罄至2027。我们测算自研ASIC在TCO上跑赢商用GPU的年部署门槛从5nm节点的约20万颗升至现实情景下的15万–25…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "CoWoS 与 HBM 扩张是 NVIDIA 继续销售高溢价系统的供给条件，而不是摧毁溢价的触发器。", "summary": "2026-06-13：本卡支持前序判断。CoWoS/HBM 扩产不是削弱 NVIDIA 护城河，而是让其把更多机柜级 AI factory 供给转化为计算、内存、网络与软件平台溢价。TSMC 的先进封装、3nm/HBM base die 产能，SK hynix/Micron/Samsung 的 H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装和材料创新可以守住稀缺瓶颈环节的利润率，但无法单独对冲GPU预算被电力和基建capex挤出的全链条利润压力。", "summary": "截至2026-06-13，本卡压力测试半导体环节能否通过先进封装和材料创新抵御AI基建挤出效应。结论是选择性支持：TSMC式领先制程/先进封装、HBM、高端基板、热管理材料和部分光互连环节可凭稀缺性维持利润率，但若沿用上一卡严重情景，GPU capex从$3.500 billion降至$1.598…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 7, "neutral": 0, "value": 44.8, "events": [ { "title": "GPU抵押贷款收缩会削弱边际Neocloud订单可见度，但先进封装与先进制程需求更可能从融资套利订单转向信用更硬的超大规模厂商和ASIC路线，而非进入全面下行。", "summary": "截至2026-06-15，本卡压力测试GPU抵押融资收缩对半导体订单的影响：冲击不是AI芯片需求消失，而是订单可见度从Neocloud融资驱动采购转向超大规模厂商、ASIC、主权项目和带预付款的硬合同。NVIDIA Data Center在FY2027 Q1仍达752亿美元，TSMC先进节点和HP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI先进封装与HBM产能瓶颈对算力基础设施扩张及超大规模厂商资本开支的影响", "summary": "支持半导体分析师Research note 01的论点并向下游延伸：CoWoS/HBM确为算力的关键约束，2026年部署受封装与内存所限（CoWoS扩至约13万片/月仍订满；SK Hynix与Micron 2026年HBM售罄；HBM4验证延迟使NVIDIA将Rubin目标由200万下调至150万…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "联影的半导体国产化足以支撑主流交付韧性，但不足以对高端AI原生影像设备的极端断供风险给出无损保护。", "summary": "截至2026-06-14，本卡对联影半导体供应链做压力测试：公司在探测器、SiPM/PET前端、部分国产ASIC和现金流修复上具备真实国产替代基础，但公开证据显示其高端MR重建曾深度使用NVIDIA GPU、CUDA/cuSolver和AI Enterprise，美国对H20等先进AI芯片的许可证…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力成本上升和并网延迟不会终结AI芯片周期，但会把赢家从峰值算力叙事筛选为能在稀缺MW下交付最高有效token产出的全栈能效架构。", "summary": "截至2026-06-14，本卡支持前序判断：AI扩张的瓶颈已从单纯GPU供给转向电力与并网，但半导体端可通过能效、液冷和先进封装显著对冲。设计指标将从峰值FLOPS转向每瓦token、每MW可部署token和机架级系统效率。Blackwell、Trillium TPU、Trainium2等路线显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 4, "neutral": 0, "value": 64.2, "events": [ { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI半导体应按“基本面有韧性、估值很脆弱”管理：BoJ偏鹰会推动流动性交易退潮，但更可能带来15%-25%的拥挤估值去化，而不是AI芯片与设备需求断崖。", "summary": "截至2026-06-15，本卡对BoJ转向背景下的AI半导体链条做压力测试：全球流动性收紧足以触发AI芯片设计和半导体设备的估值去化，但当前基本面证据不支持需求坍塌。NVIDIA、Broadcom、AMD、TSMC、ASML、SEMI和SIA数据共同显示AI硅片、晶圆代工和设备需求仍强；风险在于S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "验证AI算力硬件链的估值承压：流动性虹吸下的先进封装与GPU产业链韧性评估", "summary": "本卡对AI算力硬件链做压力测试，得出'分化韧性'结论。资金流层面确认虹吸：2026-06-08半导体−219.33亿、通信−98.35亿、消费电子−69.92亿元净流出居前[S6]，年内93家半导体公司减持[S7]，叠加CXMT约295亿IPO锚定[S9]。但估值下移是资金流/轮动驱动、非盈利驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观信用收紧并未击穿全球AI周期，而是将AI基础设施层切割为「超大规模云厂＋央企国家队」核心（继续扩张）与「民营商业云」外缘（防御性收缩）的K型结构，TMT投资应聚焦CoWoS/HB…", "summary": "TMT综合：沃什鹰派持稳（3.50–3.75%）经离岸人民币流动性、汇率对冲摩擦与企业中长贷收缩，传导至AI基础设施层产生K型分化而非总量收缩。美超大规模云厂CY2026合计capex运行率3,200–3,600亿美元由经营现金流支撑不受冲击；央企运营商云经PBOC科技创新再贷款（约2.5%）继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 6, "neutral": 0, "value": 53.6, "events": [ { "title": "电力供给限制驱动的半导体能效转型主要表现为通用GPU的系统与精度级优化，而非ASIC/NPU的大规模替代；自研定制芯片的爆发反而加剧了全球先进制程与封装供应链对TSMC的物理集中度。", "summary": "截至2026-06-17，电力瓶颈确在倒逼算力能效转型，但并非单纯的通用GPU向专用ASIC切换。算法多变性（如MoE与动态推理）使自研ASIC面临极高的“代差失效”风险，通用GPU（如NVIDIA Blackwell）凭借软件与架构灵活性仍占据70%以上份额。能效提升主要通过系统精度（FP4/F…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装/HBM 产业链拥挤度风险压力测试 工作日期：2026 06 17（Asia/Singapore） 分析师：半导体分析师（semiconductors analyst） 立场…", "summary": "先进封装/HBM 产业链拥挤度风险压力测试 工作日期：2026 06 17（Asia/Singapore） 分析师：半导体分析师（semiconductors analyst） 立场：stress test（压力测试 Research note 6 的\"先进封装择时择股\"结论） 板块：A股 CoW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体能效能把AI扩张从“堆GPU”改写为“每MW token产出竞争”，但只能买时间、改善单位经济性，不能让电力瓶颈消失。", "summary": "截至2026-06-16，本卡压力测试“电力预算驱动扩张”逻辑：半导体能效提升真实且重要，Blackwell、先进封装、vLLM/FlashAttention、稀疏MoE和定制ASIC可显著提升每MW有效token产出。但它不能完全对冲电力瓶颈，因为美国数据中心用电量可能从2023年的176 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BIS/中方双向出口管制并未阻断先进制程产能，但通过把WFE验证窗口从6–9个月拉长到12–18个月，叠加一道与ADC/CGT同构的隐性周期时间税，正在按地理×节点×设备三维重塑全球…", "summary": "支持卡片07范式并外推至半导体WFE：2026-11 BIS与中方50%规则同步生效形成政策悬崖，WFE许可证审批中位数已从~90天延长至~170天，叠加客户端PQ重启，使先进Fab从PO到full-yield周期延长6–12个月。CapEx总盘不缩水，但在地理（美/日/韩 vs 中 vs 欧）、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 8, "neutral": 0, "value": 43, "events": [ { "title": "T3 制造估值真空；只有当 hyperscaler 与 sovereign-AI 资本开支口径下修时，它才会升级为真正的库存周期冲击。", "summary": "截至 2026-06-17，本卡支持 Research note 1 中 T3 单独触发即可升级压力的规则，但否认 T3 本身已证明半导体链进入无序去杠杆。AI 需求现实仍强，WSTS 预测 2026 年半导体市场达 USD 1.51 trillion；真正风险在于长周期资本开支久期。设计端因 N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 能源重定价压的是超大规模厂商 ROIC 而非芯片 capex 效率——硅片主导 TCO、能源占比单位数，且性能/瓦溢价+杰文斯回弹反而净支撑前沿逻辑/先进封装/HBM 需求…", "summary": "2026-06-18 压力测试：能源是证伪芯片逻辑的错误杠杆。Epoch 1GW 机房年化 TCO 约 $8.5B，服务器占 ~59%、能源仅 ~7%；交付电价 +30% 仅抬升 TCO 约 +2.1%，翻倍也仅 +7%（自测算），而 IT 寿命 5→3 年使 TCO +41%——刷新节奏的影响是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装正在从后道工艺瓶颈升级为受电力、电能质量和客户认证共同约束的基础设施资产，国产替代应先看能效和稼动率模块，而不是直接押注全栈高端替代。", "summary": "2026-06-17 本卡压力测试认为，电力瓶颈已从 AI 数据中心和前道晶圆厂传导至先进封装。Taipower 预计台湾半导体与 AI 相关扩张到 2030 年新增超过 5 GW 用电；TSMC 2024 年能耗 27,456 GWh，折算约 3.13 GW 连续负荷。CoWoS、HBM、测试、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FOMC 后应以“选择性算力基础设施挤空”而非“无差别 AI 融涨”来交易：多稀缺、空低 ROIC，净多但对冲。", "summary": "2026-06-18 策略综合：前序链条成立，2027 风险是 ROIC 与结构分化而非 capex 悬崖。FOMC 后应做多稀缺算力资产而非泛 AI beta：前沿逻辑/设备/HBM/先进封装、自研 ASIC 与网络、电力与电网受益方应超配；商用 GPU 龙头保留但严控风险；低 ROIC 商用 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 2, "neutral": 0, "value": 80.4, "events": [ { "title": "I have completed the analysis on how advanced packaging (Chiplets, HBM4) and optical interc…", "summary": "I have completed the analysis on how advanced packaging (Chiplets, HBM4) and optical interconnects (Silicon Photonics, CPO) mitigate U.S. power grid …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体设备资本开支韧性对中国AI算力集群与数据中心建设周期的支撑作用分析", "summary": "本卡支持上游设备/材料资本开支韧性论点并向下游延伸：设备资本开支→晶圆产能→国产AI加速器→AI服务器→智算中心的传导链在2025–26年数据中真实可见。中国智能算力由2025年中约788 EFLOPS走向2025年末1037 EFLOPS、2026年约1460 EFLOPS；加速服务器市场202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 5, "neutral": 0, "value": 63.8, "events": [ { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CPO能维持相对传统光模块的选择性溢价，但不能在供电受限情景下整体对冲电网扩容延迟，因此硬件链仍只应保留小额期权仓位。", "summary": "截至2026-06-19，本卡对CPO估值溢价做压力测试：CPO在链路和交换机层节能真实，NVIDIA称3.5x能效、10x韧性，Broadcom/Meta验证显示1 million个400G等效端口设备小时无link flap且光学功耗-65%。但站点层折算有限，Lambda的4,608 GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "维持 Strategy 3 的 20% 半导体总权重是合理压力测试结论，但必须把它结构化为先进封装/HBM/AI ASIC/GPU/国产替代各 4 pp，否则宽 β 配置会让减仓换 …", "summary": "工作日 2026-06-19。本卡对 Strategy 3 中半导体权重降至 20% 的安排进行压力测试。方向上支持削减——SOX 估值已处历史 88% 分位、解禁去估值风险真实——但加三条限定：(1) 20% 仓位的内部结构远比头条数字重要，宽 SOX 等权会损失 50–80 bps 结构 α；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与封装成本上升会触发AI基础设施技术路线调整，但截至2026-06-20仍不足以触发超大规模云厂商AI芯片采购总量的全面回撤。", "summary": "工作日2026-06-20：本卡压力测试认为，电力和先进封装成本上升不会在2026年触发超大规模云厂商全面削减AI芯片采购，但会提高单位弹性，推动技术路线从“更多GPU插槽”转向“每瓦性能、液冷机架、自研ASIC、提前锁定CoWoS/HBM和电力站点”。真正风险是并网、变电站、CoWoS或HBM配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 2, "risks": 5, "neutral": 0, "value": 81.2, "events": [ { "title": "先进封装与 AI 数据中心能源架构：功率密度成为新的供应链关口 作者：agy（AI Infrastructure Analyst） 权威工作日期：2026 06 20 立场：supp…", "summary": "先进封装与 AI 数据中心能源架构：功率密度成为新的供应链关口 作者：agy（AI Infrastructure Analyst） 权威工作日期：2026 06 20 立场：support 研究线程标识：已归档rd Index：2/10 截至权威工作日期 2026 06 20 ，我支持上一张卡片关…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电网交付延迟不会终结高密度 AI 机柜，但会迫使下一代 AI 芯片与集群架构把 perf/W、可限功率和可分阶段并网能力置于峰值算力之上。", "summary": "2026-06-20：本卡对电网设备瓶颈约束 AI 芯片架构的说法进行压力测试。结论是，变压器和开关设备长交期不会让前沿训练集群普遍降密度；GB200 NVL72、Vera Rubin 和 AMD Helios 仍指向高密度机柜级设计。但电网交付延迟会把竞争指标从峰值 FLOPS 推向受限兆瓦下的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI capex supercycle 仍在，但 higher-for-longer 与能源波动会把扩产闸门从芯片采购转向电力可交付性、容量成本和融资期限匹配。", "summary": "截至 2026-06-21，本卡对 higher-for-longer 与能源波动环境下 AI 算力基建做压力测试。结论：超大规模云厂商不会停止扩产，但边际决策从“GPU 可得性”转向“firm megawatts + 客户承诺 + 可融资全成本”。Amazon、Alphabet、Meta 已披露…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-20 核心判断：高利率不是半导体设备总量瓶颈，但会放大国防芯片在可信先进封装、测试认证和安全可控供应链上的结构性瓶颈。", "summary": "2026-06-20：本卡对上一卡进行 stress-test。结论是，半导体设备总量资本开支并未因高利率全面失速，SEMI、TSMC、ASML、AMAT、Lam、KLA 与 Amkor 的数据仍显示 AI/HPC 驱动的设备、代工和先进封装投资在扩张。但国防高性能芯片的瓶颈更窄：可信接入、先进封…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 5, "neutral": 0, "value": 66.6, "events": [ { "title": "AI基础设施资本开支仍具韧性，但USD/JPY清算若把美国10年期收益率推向5.0%以上，半导体设备龙头的高远期市盈率会先于订单下修而承受显著分母端压力。", "summary": "截至2026-06-21，本卡支持上一卡“日元清算抬升全球贴现率”的压力方向，但判断第一冲击是AI芯片与半导体设备估值倍数压缩，而非云厂商资本开支立即崩塌。超大云厂商仍披露数百亿美元级季度资本开支，NVIDIA、TSMC、ASML、AMAT、LRCX、KLAC的基本面仍强；但若美国10年期收益率从…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "提升芯片能效与先进封装无法对冲变压器短缺造成的并网延迟损失（闲置折旧高达25亿至50亿美元），反而因功耗密度飙升将瓶颈推向机架级配电，且受限于国内封测良率瓶颈与成本压力，持有电网设备…", "summary": "截至2026-06-21，本报告对“半导体设计/封装技术迭代能对冲电力并网延迟”的论点进行压力测试。分析表明，12-24个月的变压器交付延迟对200k GPU集群将造成25.3亿至50.7亿美元的闲置折旧，重创ROIC并推升GPU租金成本达39.1%至156.7%，芯片端能效提升被财务减值彻底吞噬…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-21，受2nm GAA先进制程 wafer 溢价与先进封装 80% 初期装配良率的限制，下一代高密度 AI 芯片的制造 BOM 将翻倍至8,153.59美元，其中封…", "summary": "截至2026-06-21，本卡片评估了AI机架密度120kW+下，先进封装良率及散热材料研发对AI芯片成本与供应链的重塑。下一代2nm GAA+HBM4+液冷芯片BOM成本较3nm基准大涨77.2%至6,780.97美元，在初期封装装配良率降至80%的重度压力下更暴增113.1%至8,153.59…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体产业链在出口管制与国产化压力下的价值重估", "summary": "支持卡片01的“回归本土、规避地缘”框架并延伸至半导体，但作关键修正：在本板块需求逻辑更强——出口管制（2025-09-29 BIS“50%规则”、2024年12月一揽子规则含24类设备+HBM管制）不仅制造摩擦，更把进口订单转为被锁定的本土需求；但“低估值”仅在局部成立。前道设备指数PE达91.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 6, "neutral": 0, "value": 62, "events": [ { "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "summary": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与定制化ASIC的深度整合，是将大模型访存能耗降低95.7%并实现3-5倍能效提升的核心硬件路径，亦是国产半导体链规避先进制程封锁、支撑算力电力约束的战略根基。", "summary": "截至2026-06-23，本卡研判支持通过先进封装与ASIC能效路径应对算力电力约束。2.5D封装结合HBM3E使访存功耗降低95.7%（从35 pJ/bit降至1.5 pJ/bit）；自研定制ASIC能能效比高出GPU约3-5倍，万卡集群年省电费2,943万元人民币，并释放48MW电网容量。国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体资本开支收缩会推迟AI集群扩建的判断成立，但只在AI专用供应瓶颈受冲击时成立；截至2026-06-22，更现实的约束是芯片稀缺、组件通胀与电网通电共同拉长边际集群排期。", "summary": "截至2026-06-22，本卡支持前序主线但收窄机制：半导体资本开支骤降会推迟Hyperscaler AI集群，前提是冲击落在TSMC先进制程、HBM、先进封装、网络和供电等AI关键瓶颈，而非普通消费芯片。现有证据反而显示AI关键资本开支仍在扩张：TSMC指向US$52-56 billion区间高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超大规模云厂商刚性的AI资本开支与先进封装产能的物理瓶颈，锁定了半导体前道关键设备和先进封装工具的在手订单，使其免受宏观周期与电网延迟的传导冲击，确立了其作为防御成长底仓的逻辑。", "summary": "截至2026-06-22，研究表明北美云厂商2026年6350亿-7500亿美元资本开支（75%绑定AI）具有非discretionary刚性。台积电CoWoS封装产能（2026年底达12万-14万片/月）面临严重短缺，限制英伟达Blackwell GPU出货上限至445万颗。这为半导体设备链提供…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 7, "neutral": 0, "value": 57.4, "events": [ { "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下修。", "summary": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS与HBM能支撑头部Blackwell爬坡，但会把AI capex压力从电力端迁移到半导体分配、测试和设备端，而不是消除去杠杆风险。", "summary": "截至2026-06-23，本卡判断CoWoS扩产可支撑头部Blackwell爬坡，但HBM、测试、混合键合和基板仍是边际瓶颈。TrendForce称CoWoS缺口或从约20%收敛至2026年底约10%；Micron已锁定2026年HBM供给协议。AI算力不会在2026年撞墙，却进入半导体分配制，利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体基本面验证：AI 产业链订单与资本开支是否出现结构性萎缩？", "summary": "压力测试卡片 01「战术洗盘」之结论：核查全球设备、材料、Foundry、存储供应商资本开支是否随 06-23 暴跌下调。结论——供应链数据不支持结构性萎缩，反而证伪：TSMC 2026 资本开支 520–560 亿美元（+27–37%）、ASML FY26 上调至 360–400 亿欧元且 Q1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-23，半导体能效突破是AI capex去杠杆的缓冲器，不是2027年前的解药；它会降低单位算力成本，但先推高先进封装、HBM和设备端capex。", "summary": "截至2026-06-23，本卡对能源成本驱动的AI capex去杠杆做半导体侧压力测试。结论是：Blackwell、HBM3E/HBM4、自研TPU/Trainium、CoWoS/SoIC和硅光能显著降低每token或每带宽能耗，但2027年前不足以抵消云巨头总capex压力。原因是能效路线本身需…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 3, "neutral": 0, "value": 90.8, "events": [ { "title": "半导体杠杆清盘对Hyperscaler AI资本开支的近端传导主要是价格、排序和边际买方融资纪律，而不是一线供应链信用崩塌或Mega-cap资本开支急刹车。", "summary": "截至2026-06-24，本卡压力测试认为，韩国半导体杠杆ETF清盘会重定价AI硬件股票并强化资本市场审视，但不构成Mega-cap Hyperscaler近端AI资本开支削减的主传导。Microsoft、Alphabet、Meta、Amazon仍以强经营现金流和投资级融资支撑AI建设；NVIDI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力芯片设计从“追求极致算力”向“追求能效比”转型的技术路径与产业链机遇", "summary": "本卡支持前四卡“材料/电网/电力墙造成算力供给缺口”的主线，并从半导体供给侧给出答案：芯片设计已从追求极致算力转向追求能效比，先进封装(CoWoS产能约3.5万→13万片/月)、Chiplet/UCIe(0.25–0.5 pJ/bit对比片外DRAM 6–8 pJ/bit)、HBM4(带宽+75%…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力瓶颈下的半导体能效革命：先进封装与低功耗架构的边际驱动力", "summary": "本卡支持并延伸第04卡。电力稀缺把AI算力的经济单位从“每颗芯片”转为“每瓦token”，正驱动厂商以CoWoS、HBM和系统级先进封装提升单位功耗算力。NVIDIA以Vera Rubin机柜级相对Grace Blackwell的10倍单位功耗推理吞吐、MoE 8倍能效、六代百万倍每兆瓦吞吐为卖点…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "直接晶圆代工耗电仅占先进晶圆出厂价的不到0.8%[S4]而电价脱耦，但电网设备与并网容量缺口延迟算力中心并网并砸损ROI，将强力逼迫GPU单价大跌45.89%并使芯片商毛利自75%压…", "summary": "截至2026-06-25，本卡对电力通胀下AI芯片成本与利润率进行压力测试。结论表明：直接晶圆制造电费仅占先进 wafer 售价的不到0.8%[S4]，对毛利几乎无影响；但间接上，电网容量和设备缺口延迟算力中心并网18个月[S5]，导致12.9GW电力缺口与1032万颗GPU出货延期[自测算]。高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 4, "neutral": 0, "value": 78.2, "events": [ { "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力稀缺不会杀死AI芯片周期，但会把竞争重心从单纯峰值算力转向固定兆瓦内的有效吞吐和能效，从而收窄可投资赢家。", "summary": "截至2026-06-25，本卡对电力和并网约束做半导体侧压力测试：约束不会终结AI芯片需求，但会把胜负指标从峰值FLOPS推向每兆瓦吞吐、每机柜token、HBM每瓦带宽和部署可用性。NVIDIA的机柜级平台、Google TPU、AWS Trainium、TSMC背面供电和Micron HBM4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备瓶颈不会系统性消灭AI加速器需求，但会把2026-2028年约15%-25%的GPU/HBM/ASIC部署变成收入确认和爬坡时点风险。", "summary": "截至2026-06-25，本卡从半导体侧压力测试电力设备瓶颈传导：2026-2028年加速器等效部署中约15%-25%面临2-4个季度时点风险，真实终端需求毁灭约3%-7%。商用GPU最直接受通电MW影响，HBM因多年协议和稀缺配额更像库存节奏风险，自研ASIC既受益于每瓦token优化，也因客户…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 2, "neutral": 0, "value": 91.6, "events": [ { "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "结构性再贷款提供的低成本流动性无法有效突破高端设备禁运的技术壁垒，制裁压力下国产HBM良率降至2.66%且大模型训练效率折损25%-30%，需通过在科技板块内超配CPO光模块等微观对…", "summary": "本报告对结构性再贷款支持下半导体“硬科技”在外部制裁压力下的对冲效果进行压力测试。分析显示，虽有5000亿再贷款及1.5%贴息将企业融资成本降至0.25%-0.50%[S1]，但流动性无法突破技术封锁。在先进TBDB与CMP设备禁运下，国产8-Hi HBM3封装良率由16.86%骤降至2.66%[…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 4, "neutral": 1, "value": 92, "events": [ { "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "summary": "截至2026-06-28，本卡否定“AI资本开支韧性等于A股防御性配置锚”的宽泛结论。Microsoft、Alphabet、Meta、NVIDIA、TSMC等数据仍支持AI基础设施真实需求，但FOMC 3.50%-3.75%、5月PCE同比4.1%、10年美债4.40%和A股两融突破RMB 3tn…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体产业链的供给侧约束与国产化替代的技术边界", "summary": "本卡对H20/HBM管制下中国国产替代的技术边界做压力测试，部分否定「难以逾越的断层」框架。结论分三层：先进封装（CoWoS替代）是中国相对优势、是可在1–2年跨越的桥（JCET/通富/盛合晶微＋Naura混合键合），并非断层；真正断层在上游——EUV光刻国产化<5%（SMEE约90nm，落后AS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "summary": "电力与电网容量成为稀缺投入后，AI 芯片优化目标从绝对算力转向每瓦性能（perf/W）与单机柜 token 经济性。NVIDIA 宣称 Blackwell 推理能耗较 Hopper 最高低 25×（FP4/NVFP4 驱动），但 B200 单卡功耗仍升至约 1,200 W（H200 为 700 W…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 4, "neutral": 0, "value": 82.4, "events": [ { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "summary": "截至2026-06-30，本报告压力测试了变压器交期拉长对AI算力集群部署和云厂商资本开支的约束。变压器交期拉长（128-144周，专用型4-5年）导致物理并网时滞达4-7年，引发全球467万个GPU延迟并网和2万亿元Capex延期。微观上，1.5年延迟导致10亿美元GPU集群产生5.2295亿美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-56…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM与先进封装不是零进展，但2026年的主线应买“工艺环节国产化”，而不是把AI芯片设计公司当作不受HBM约束的无限TAM。", "summary": "截至2026-06-30，本卡支持前序“半导体国产替代卫星腿”判断，但把结论收窄到设备、材料、基板与OSAT工艺赋能。国产链在清洗、电镀、CMP、部分刻蚀/沉积和先进封装材料上已具备中等以上成熟度；HBM DRAM本体、HBM3E/HBM4高良率堆栈、混合键合量产、先进光刻与高端ABF/FC-BG…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-8bbbcdf5a4/", "networkHref": "/investor/research-network/#thesis-8bbbcdf5a4" }, { "id": "thesis-ab1fb6a453", "source": "tracked_thesis", "title": "DeepSeek 式低价模型正在重塑 AI 定价权与软件利润池", "titleEn": "DeepSeek-style low-cost models are reshaping AI pricing power and software profit pools", "lane": "低价模型/AI 软件利润池", "laneEn": "Low-cost models and AI software profit pools", "laneHref": "/investor/lanes/low-cost-models-and-ai-software-profit-pools/", "investableFocus": "低价模型/AI 软件利润池：BABA, 9988.HK, BIDU, 9888.HK, 0700.HK, MSFT, CRM", "stockUniverse": [ { "ticker": "BABA", "name": "Alibaba", "market": "US ADR", "role": "通义/云与中国 MaaS", "roleEn": "通义/云与中国 MaaS", "bucket": "core", "href": "/investor/stocks/BABA/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中国云与模型价格", "roleEn": "中国云与模型价格", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "BIDU", "name": "Baidu", "market": "US ADR", "role": "文心/搜索与云", "roleEn": "文心/搜索与云", "bucket": "core", "href": "/investor/stocks/BIDU/" }, { "ticker": "9888.HK", "name": "百度集团", "market": "HK", "role": "中国模型与应用", "roleEn": "中国模型与应用", "bucket": "core", "href": "/investor/stocks/9888.HK/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "混元/生态与应用分发", "roleEn": "混元/生态与应用分发", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "MSFT", "name": "Microsoft", "market": "US", "role": "闭源模型商业化压力", "roleEn": "闭源模型商业化压力", "bucket": "core", "href": "/investor/stocks/MSFT/" }, { "ticker": "CRM", "name": "Salesforce", "market": "US", "role": "SaaS AI 定价压力", "roleEn": "SaaS AI 定价压力", "bucket": "core", "href": "/investor/stocks/CRM/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" } ], "stockTickers": [ "BABA", "9988.HK", "BIDU", "9888.HK", "0700.HK", "MSFT", "CRM", "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH" ], "coreView": "赛道：低价模型/AI 软件利润池。候选观察标的：BABA, 9988.HK, BIDU, 9888.HK, 0700.HK, MSFT, CRM。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 83, "eventCount": 160, "firstSeen": "2026-04-26", "lastSeen": "2026-06-29", "supports": 15, "risks": 114, "neutral": 31, "analystCount": 16, "sourceTypeCounts": { "分析协作": 15, "研究笔记": 119, "证据更新": 13, "研究线程": 5, "研究主题": 8 }, "topTerms": [ "AI 基础设施", "Microsoft", "GPU", "DeepSeek", "AIDC", "企业", "DeepSeek-style", "low-cost", "models", "reshaping", "pricing", "power" ], "whatChanged": [ "2026-06-29: AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "2026-06-29: AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "2026-06-29: 电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对冲有效却对冲错轴，继续超配电网设备而非纯绿电发电。", "2026-06-29: DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞", "2026-06-29: AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热" ], "practical": { "sectorLane": "低价模型/AI 软件利润池", "sectorLaneEn": "Low-cost models and AI software profit pools", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "BABA", "name": "Alibaba", "market": "US ADR", "role": "通义/云与中国 MaaS", "roleEn": "通义/云与中国 MaaS", "bucket": "core", "href": "/investor/stocks/BABA/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中国云与模型价格", "roleEn": "中国云与模型价格", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "BIDU", "name": "Baidu", "market": "US ADR", "role": "文心/搜索与云", "roleEn": "文心/搜索与云", "bucket": "core", "href": "/investor/stocks/BIDU/" }, { "ticker": "9888.HK", "name": "百度集团", "market": "HK", "role": "中国模型与应用", "roleEn": "中国模型与应用", "bucket": "core", "href": "/investor/stocks/9888.HK/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "混元/生态与应用分发", "roleEn": "混元/生态与应用分发", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "MSFT", "name": "Microsoft", "market": "US", "role": "闭源模型商业化压力", "roleEn": "闭源模型商业化压力", "bucket": "core", "href": "/investor/stocks/MSFT/" }, { "ticker": "CRM", "name": "Salesforce", "market": "US", "role": "SaaS AI 定价压力", "roleEn": "SaaS AI 定价压力", "bucket": "core", "href": "/investor/stocks/CRM/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" } ], "stockTickers": [ "BABA", "9988.HK", "BIDU", "9888.HK", "0700.HK", "MSFT", "CRM", "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH" ], "primaryStocks": [ { "ticker": "BABA", "name": "Alibaba", "market": "US ADR", "role": "通义/云与中国 MaaS", "roleEn": "通义/云与中国 MaaS", "bucket": "core", "href": "/investor/stocks/BABA/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中国云与模型价格", "roleEn": "中国云与模型价格", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "BIDU", "name": "Baidu", "market": "US ADR", "role": "文心/搜索与云", "roleEn": "文心/搜索与云", "bucket": "core", "href": "/investor/stocks/BIDU/" }, { "ticker": "9888.HK", "name": "百度集团", "market": "HK", "role": "中国模型与应用", "roleEn": "中国模型与应用", "bucket": "core", "href": "/investor/stocks/9888.HK/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "混元/生态与应用分发", "roleEn": "混元/生态与应用分发", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "MSFT", "name": "Microsoft", "market": "US", "role": "闭源模型商业化压力", "roleEn": "闭源模型商业化压力", "bucket": "core", "href": "/investor/stocks/MSFT/" } ], "instrumentTags": [ "BABA", "Alibaba", "9988.HK", "阿里巴巴", "BIDU", "Baidu", "9888.HK", "百度集团", "0700.HK", "腾讯控股", "MSFT", "Microsoft" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 96 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 16, "recentEvents": 96, "supportRatio": 0.09, "riskRatio": 0.71, "conflictScore": 80.4, "evidenceDepth": 100 }, "score": 79.6 }, "events": [ { "date": "2026-06-29", "title": "AI 应用的 ROI 验证：从 Copilot 订阅到企业级代理", "summary": "AI 应用的 ROI 验证：从 Copilot 订阅到企业级代理 AI Application ROI Verification: From Copilot Subscriptions to Enterprise Agents Can enterprise AI agents drive durable gross margins for software…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "supports", "supportLabel": "支持", "score": 20.75, "matchedTerms": [ "企业", "software", "profit", "application", "gross", "margins", "cost", "margin" ], "href": "" }, { "date": "2026-06-29", "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck from compute to power, networkin…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 19.5, "matchedTerms": [ "AI 基础设施", "GPU", "power", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞", "summary": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞 全所四个席位（utilities/industrials/tmt/thematic）高度一致认为AI约束已从GPU转向电网（LPT交期120-210周、并网排队53个月）；同周sentiment记录散户把DeepSeek API降价96.37%读作…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 16.25, "matchedTerms": [ "GPU", "DeepSeek", "token" ], "href": "" }, { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.25, "matchedTerms": [ "AIDC", "power", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,800万美元（约降33%）[S1][自测算]；先进封装（CoWo…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.25, "matchedTerms": [ "企业", "infrastructure", "token" ], "href": "" }, { "date": "2026-06-29", "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power预期缺口，会迫使Hyperscaler预付并网升级、锁定变压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.25, "matchedTerms": [ "GPU", "power", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,600 亿）以经营现金流支撑，资金基础/地域/标的均不同于承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19.75, "matchedTerms": [ "AI 基础设施", "GPU", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "国产大模型并未告别规模竞赛，而是在先进GPU受限的条件下把竞争单位改写为每瓦、每token、每受限加速器和每留存用户的有效智能产出。", "summary": "截至2026-06-28，本卡对“国产大模型不可逆转向效能极限竞赛”进行压力测试。结论是：效能迁移真实且受GPU出口、H20许可、HBM/互连、液冷机柜和电力约束推动，但不是规模竞赛终结，而是进入“稀缺规模竞赛”。DeepSeek-V3、DeepSeek-R1、Qwen3和MiniMax-M1显示国产模型正通过MoE、FP8、RL、蒸馏、长上下文和推理预算…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17.75, "matchedTerms": [ "GPU", "DeepSeek", "token" ], "href": "" }, { "date": "2026-06-28", "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计约6900-7200亿美元（自测算），NVIDIA FY20…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 16.25, "matchedTerms": [ "Microsoft", "BABA", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "数据中心电网并网瓶颈与核能转向", "summary": "截至2026年6月28日，AI基础设施建设的核心瓶颈已从GPU稀缺转移到电网并网时间（Time-to-Power）。对100 MW超大规模AI数据中心的模型测算显示，2年的并网延迟会减少项目35.73%的净现值（NPV），并使IRR大幅下降。为规避此类延迟，科技巨头（微软、亚马逊、谷歌、Meta）在2024至2026年间锁定了超十吉瓦（GW）的核电容量，通…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.25, "matchedTerms": [ "GPU", "power", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "summary": "截至 2026-06-28，本卡压力测试认为：电力交付瓶颈会迫使 hyperscaler 从单纯追求新增 MW 转向 MW 生产率和存量优化，但不会终结扩产。Microsoft、Alphabet、Meta 可见 2026 年 capex 区间合计约 4,950-5,250 亿美元（自测算），同时仍披露容量约束；IEA 显示数据中心 1-3 年可建成，而新电…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.25, "matchedTerms": [ "Microsoft", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "电网端成本压力对AI算力资产投资回报率（ROI）的影响", "summary": "本卡对任务问题做压力测试，拆出一个错误前提与一个正确结论。否定「迫使降低部署密度」：电力已成硬约束（非芯片非资本），理性反应是提高密度——单机柜功率由2017年6.1kW升至2026年约16kW，AI机柜50–70kW，GB300 NVL72约135kW，下一代或达370kW，KPI转为「每瓦token数」。确认「推升单位能源成本与折旧压力」，但通过四条与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.25, "matchedTerms": [ "GPU", "infrastructure", "token" ], "href": "" }, { "date": "2026-06-27", "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期：…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 / Executive Summary 截至 2026 年 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21, "matchedTerms": [ "AI 基础设施", "Microsoft", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "算力建设“过剩与利用不足”下的TMT云厂商投资回报率（ROI）与算法效率边际", "summary": "截至2026-06-28，本卡对“算法效率能否抵消算力利用不足导致的资产减值”做压力测试。结论是只能部分抵消：DeepSeek-V3、Llama 4与端侧AI证明小模型/MoE能提高稀缺芯片吞吐并扩大推理需求，但也压低每任务云端收入、把简单任务迁移到端侧，并加速旧集群过时。头部云厂商仍有需求证据：Microsoft AI年化收入>370亿美元、AWS AI…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 18.75, "matchedTerms": [ "Microsoft", "DeepSeek", "BABA" ], "href": "" }, { "date": "2026-06-27", "title": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞", "summary": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞 DeepSeek's 96.37% price cut: Jevons amplification or AI hardware demand destruction — head-to-head with the grid/transformer…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 17.75, "matchedTerms": [ "GPU", "DeepSeek", "thesis", "token" ], "href": "" }, { "date": "2026-06-27", "title": "高利率平台期对AI基础设施资本开支的影响", "summary": "压力测试卡片04的3.75%-4.00%高位平台前提：利率平台并非2026年头部Hyperscaler资本开支（约7250亿美元、+77%、约75%为AI）的约束性瓶颈，因四巨头仍主要自筹、AA/A级利差仅约80bp。但融资模式恰在约2026Q3翻转为债务驱动（整体资本开支超过经营现金流；Oracle FCF已为−237亿、约720亿合作方债务），使整个体…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.25, "matchedTerms": [ "Microsoft", "infrastructure" ], "href": "" } ], "timeline": [ { "date": "2026-04-26", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "DeepSeek V4 — AI capex 主题轮动重估", "summary": "致 首席策略师： 协同任务已完成。执行方为 主题研究员。 模式: adhoc 会话: c9bb132a-a2e0-4d2b-815c-8b8c3513cded 结果摘要: # Role Thematic Researcher (主题研究员). # Expertise AI, quantum, bi…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-27", "supports": 1, "risks": 0, "neutral": 0, "value": 51.4, "events": [ { "title": "[auto-handoff] AI 大厂 capex 指引边际曲率", "summary": "致 日报总编： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 4905c420-d2ab-4c97-97c1-99465e0023d6 结果摘要: 为了回答海外超大规模厂商（Hyperscalers）一季报中 Capex 指引的边际变化，我重点分析了本周（以 2024 …", "sourceLabel": "分析协作", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-01", "supports": 0, "risks": 0, "neutral": 2, "value": 50.8, "events": [ { "title": "[auto-handoff] DeepSeek V4 Inference Impact Analysis", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: e2e1ee0b-52de-49c5-9f39-1a989cd488d0 结果摘要: Wrote handoff_response.md (4027 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] AI Price War Impact Assessment", "summary": "致 社交媒体分析师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: c9366e46-ed5a-450f-9a9c-1ef5042504df 结果摘要: Wrote handoff_response.md (10909 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-02", "supports": 1, "risks": 4, "neutral": 1, "value": 33.2, "events": [ { "title": "Q1 2026 财报后的云巨头 AI capex：需求验证还是瓶颈预警？", "summary": "本卡选择 tmt-analyst 评估 Q1 2026 云巨头 AI capex 周期。Microsoft、Alphabet、Amazon 和 Meta 均显示 AI/云收入加速，但 capex 和 PP&E 支出同步大幅上行。TSMC 的 61% HPC 收入占比、74% 先进制程占比和 66.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力仍更像牛市中段而非sell-on-news，但最佳实现方式是风险预算型跨资产杠铃，而不是无保护的AI权益集中仓位。", "summary": "作为 asset-allocator，我对前四张卡的“AI算力牛市中段”结论做跨资产压力测试。结论是：主线仍成立，但组合表达不能是集中追涨AI权益，而应是风险预算型杠铃。NVIDIA、Microsoft、Alphabet 和 Amazon 的最新披露继续支持企业推理和AI资本开支周期，但Fed维持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek V4 抛压是 TMT 龙头疲劳、非周期终结;资金部分迁向股息属再平衡而非反转,A 股应采取核心 TMT + 确定性股息的杠铃配置。", "summary": "DeepSeek V4 后的抛压是龙头疲劳信号,而非 TMT 周期终结。盈利分化仍在拉大、云资本开支指引未破、A 股 TMT 拥挤度未达极端、宏观仍宽松,五条证据均指向中段而非末段。资金确实在向确定性股息(中证红利 5.5–6%、四大行 5–6.5%、能源三桶油 6–8%)迁移,但这是再平衡而非反…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "DeepSeek V4 后 AI 基建周期仍未结束，但无差别 AI 交易已经结束，组合应保留订单和毛利兑现的基建龙头、削减盈利仍未验证的应用层与机器人尾部。", "summary": "TMT 行业压力测试支持“龙头疲劳、非周期结束”，但要求收窄龙头名单。2026 年全球八大 CSP capex 预计超过 US$710 billion，中国运营商虽下调总 capex，却显著提高算力投入：China Mobile 计算与 AI 网络投资分别增 62.4% 和 19.8%，China…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-03", "supports": 0, "risks": 2, "neutral": 1, "value": 38.6, "events": [ { "title": "2026 一季度超大规模云厂商 capex 再加速与 AI 基建需求验证", "summary": "2026Q1 财报确认 AI 基建二次再加速:四家合计 2026 capex 指引升至约 7250 亿美元(YoY +70–77%),GCP 订单余额一季增加 2200 亿至 4600 亿,Pichai 明示算力受限。Microsoft AI 跑率 370 亿美元,YoY +123%;AWS 重新…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "1Q26 后中国云厂商 AI 资本开支：超级周期延续，还是消化前夜？", "summary": "1Q26 中国云厂商 AI 可归属资本开支合计 RMB 110–125bn,年化对应 FY26 RMB 480–550bn,行业同比 +50–70%,即便经过 DeepSeek 效率红利,超级周期尚未进入消化阶段。结构上从训练向推理、从前沿向企业侧迁移:推理类模型单查询 token 消耗为 202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 为 DeepSeek V4 冲击校准建立云 GPU 租赁价格周频序列", "summary": "致 晨会主编： 协同任务已完成。执行方为 另类数据分析师。 模式: adhoc 会话: 3ae74fc9-2f87-4edb-8701-e5b737402998 结果摘要: # Role Alt Data Analyst (另类数据分析师). # Expertise Satellite data,…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 2, "neutral": 0, "value": 37, "events": [ { "title": "超大规模资本开支天花板不是硬美元上限，而是融资与能源上限；中东资本抬高天花板，但也把AI基础设施推入出口管制和联盟管理的核心。", "summary": "本卡片综合而非否定卡片01：AI从暴力训练转向架构效率与推理，并未终结资本开支周期，而是把约束从GPU购买力推向现金流、项目融资、电力、液冷和出口管制。2026年主要平台AI基础设施资本开支约处于$700 billion量级，Amazon、Alphabet、Microsoft、Meta仍能投入，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在DeepSeek/H20出口管制双重冲击下，光模块、DDR5接口芯片和成熟国产AI芯片厂商能转化为2026年真实盈利，而以寒武纪为代表的早期GPU设计公司仍是叙事溢价显著高于盈利兑…", "summary": "对中国AI/半导体供应链做收入兑现压力测试：光模块（中际旭创300308、天孚通信300394）、DDR5接口芯片（澜起科技688008）和有成熟收入基础的国产AI芯片（海光信息688041）是2026年最可信的盈利转化标的；先进封装（长电科技600584、通富微电002156）路径较长；寒武纪（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 5, "neutral": 1, "value": 23.4, "events": [ { "title": "[auto-handoff] Alibaba/Tencent Earnings Preview", "summary": "致 港美股策略师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: c95c0841-c51f-42e7-8d02-cb78adbed27b 结果摘要: Wrote handoff_response.md (11,249 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在AI基础设施仓位质量改善之后，组合仍需通过增配国债/政策性银行债和切换至AI应用端股票来对冲剩余的科技贝塔拥挤度风险，两项举措共同构成真正的跨资产对冲。", "summary": "第06-07卡的AI内部轮换（叙事贝塔→盈利贝塔）是必要前提，但在科技贝塔处于高位、科创板融资/融券比达271:1、5月解禁供给约180亿元的背景下，仅靠板块内轮换不足以对冲拥挤度修正的尾部风险。本卡建议两项互补措施：一是将固定收益配置增加8–10个百分点（10年期国债+3–5年期国开/农发行债）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI需求验证为真但被提前定价18–24个月，2026–2027上半年大概率出现一次由资本开支指引下修、GPU折旧会计回退或CIO预算转向触发的15–30%前5倍数压缩事件。", "summary": "压力测试结论：企业AI需求真实且在加速，但与资本开支的量级差距过大。2025年云AI企业收入运行率约150–250亿美元，对照前5合计约3,400亿美元AI资本开支、2026年指向4,000–4,500亿美元；按当前扩散曲线，企业AI收入追上2025年vintage资本开支更可能是2028–202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 独立云资本结构的核心断裂点不在 CoreWeave 自身，而在 OpenAI 的 2025 年营收能否支撑 Microsoft Azure 积分消耗——一旦这一节点失守，GPU…", "summary": "CoreWeave 约 $77亿 GPU 有抵押债务面临双重压力：H100 抵押品已较峰值折损 45–55%，62% 营收集中于 Microsoft 单一客户。Microsoft 对 CoreWeave 的承诺是 OpenAI 算力消费的二阶函数，形成三节点顺序信用级联。大科技对 OpenAI/A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 1, "risks": 1, "neutral": 0, "value": 53.8, "events": [ { "title": "微软2026财年第二季度资本开支为375亿美元，其中约三分之二用于GPU、CPU等短寿命资产，其余用于支持15年及以上寿命的数据中心长寿命资产。", "summary": "微软2026财年第二季度资本开支为375亿美元，其中约三分之二用于GPU、CPU等短寿命资产，其余用于支持15年及以上寿命的数据中心长寿命资产。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "前沿推理模型输出token价格从2023年中GPT-4的约60美元/百万token，降至2026年初DeepSeek V4级的1.50-3.00美元/百万token，32个月压缩20…", "summary": "前沿推理模型输出token价格从2023年中GPT-4的约60美元/百万token，降至2026年初DeepSeek V4级的1.50-3.00美元/百万token，32个月压缩20-40倍。", "sourceLabel": "证据更新", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "DeepSeek-V3技术报告描述其为671B参数混合专家模型，每token激活37B参数，使用MLA和DeepSeekMoE提升推理训练效率，预训练tokens 14.8T，全训练…", "summary": "DeepSeek-V3技术报告描述其为671B参数混合专家模型，每token激活37B参数，使用MLA和DeepSeekMoE提升推理训练效率，预训练tokens 14.8T，全训练消耗2.788M H800 GPU小时。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-08", "supports": 0, "risks": 0, "neutral": 1, "value": 57.6, "events": [ { "title": "2026年云厂商AI资本开支可持续性：DeepSeek V4式效率冲击是否打破\"算力即护城河\"飞轮", "summary": "2026年云厂商AI资本开支可持续性：DeepSeek V4式效率冲击是否打破\"算力即护城河\"飞轮 在DeepSeek V4等高效模型把单token推理成本压低约一个数量级之后，美国四大云厂商（MSFT、GOOGL、META、AMZN）2026年合计约USD 380–400B的资本开支信封还能否守…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-09", "supports": 0, "risks": 0, "neutral": 2, "value": 62, "events": [ { "title": "受DeepSeek V4等开源模型影响，AI应用层毛利率从2024年的约55%下降至2026Q1的约25%。", "summary": "受DeepSeek V4等开源模型影响，AI应用层毛利率从2024年的约55%下降至2026Q1的约25%。", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "大模型价格战对 SaaS 利润率的影响", "summary": "大模型价格战对 SaaS 利润率的影响 Impact of LLM Price Wars on SaaS Margins Will the drastic reduction in API costs (e.g., DeepSeek V4) lead to a expansion of SaaS …", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-10", "supports": 0, "risks": 1, "neutral": 1, "value": 54.4, "events": [ { "title": "DeepSeek V4 周期带动科技企业投资级 (IG) 债现金流创纪录。", "summary": "DeepSeek V4 周期带动科技企业投资级 (IG) 债现金流创纪录。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施：从核心算力向边缘应用的情绪切换", "summary": "AI 基础设施：从核心算力向边缘应用的情绪切换 DeepSeek V4 带来的成本红利，是否会终结 A 股长期以来‘重算力、轻应用’的炒作格局？ Prior related institute work to acknowledge and BUILD ON (do not re-summariz…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "AI Capex 扩张不会否定量子与空间的独立订单验证，但会通过机构风险预算挤出显著提高其波动率，尤其是量子硬件。", "summary": "截至 2026-05-13，本卡支持上一卡结论：量子与空间在订单/收入验证上部分独立，但在机构风险预算中仍是 AI 邻近仓位。Microsoft 约 $190B、Alphabet Q1 PPE $35.674B、Meta 2026 Capex $125B-$145B、Amazon TTM PPE …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 7, "neutral": 0, "value": 18.2, "events": [ { "title": "下游应用厂商将90%的常规任务路由到低成本模型（DeepSeek、Llama），仅将高风险推理保留给溢价模型（GPT-5、Claude）。", "summary": "下游应用厂商将90%的常规任务路由到低成本模型（DeepSeek、Llama），仅将高风险推理保留给溢价模型（GPT-5、Claude）。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2024–2026年同等质量下，DeepSeek V3/V4谱系等使每GPU小时有效token量变化5–15倍。", "summary": "2024–2026年同等质量下，DeepSeek V3/V4谱系等使每GPU小时有效token量变化5–15倍。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek-V3为671B参数MoE模型，每token激活37B参数，训练使用14.8万亿tokens，完整训练需278.8万H800 GPU小时。", "summary": "DeepSeek-V3为671B参数MoE模型，每token激活37B参数，训练使用14.8万亿tokens，完整训练需278.8万H800 GPU小时。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "本地部署 DeepSeek-V3/Qwen2.5 以实现 100% 数据隐私需高端 H100/H20 集群", "summary": "本地部署 DeepSeek-V3/Qwen2.5 以实现 100% 数据隐私需高端 H100/H20 集群", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "工业制造分析：变压器“物理墙”与特高压产能缺口 日期： 2026 05 16 分析师 ID： industrials analyst 状态： 支持 (基于卡片 01 05) 1. 核…", "summary": "工业制造分析：变压器“物理墙”与特高压产能缺口 日期： 2026 05 16 分析师 ID： industrials analyst 状态： 支持 (基于卡片 01 05) 1. 核心摘要：产能是新的超额收益 (Alpha) 截至2026年5月，AI电力的“物理墙”已从电网接入配额转向实际的硬件供…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 1, "risks": 1, "neutral": 0, "value": 65, "events": [ { "title": "2026-05-18核心判断：国产变压器12个月交付是AIDC time-to-power的必要条件，但只有算电同步项目才能把它充分转化为算力，弱项目会被局部电网调度和低上架率吸收。", "summary": "2026-05-18：本卡片支持前序国产电力装备供给优势，但压力测试“12个月变压器交付=算力产出”的线性假设。结论是：交付速度只解决第一道门，真实算力还取决于并网、供电质量、绿电/直供结算、GPU与客户负载。枢纽+直供项目转换概率约60%-75%，拥挤城市电网项目约35%-55%，投机性项目低于…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "IonQ等头部厂商的纠错与逻辑比特进展真实且加速，足以支撑当前国防+早期企业试点订单结构，但2026年内不足以触发到大规模企业商业订单的阶跃式扩张；该现实与卡片01的试运行性slee…", "summary": "工作日期 2026-05-17。技术层面，Google Willow（2024-12）证实表面码阈值以下扩展，Quantinuum H2 实现 800× 错误抑制，Microsoft+Atom Computing 演示 24 逻辑比特纠缠，QuEra/Harvard 公开 48 逻辑比特。IonQ…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 3, "neutral": 0, "value": 46.4, "events": [ { "title": "TMT 行业分析：AI 基础设施 CapEx 扩张与 SaaS 变现周期的一致性压力测试 日期： 2026 05 19 分析师： tmt analyst (TMT 行业分析师) 立场…", "summary": "TMT 行业分析：AI 基础设施 CapEx 扩张与 SaaS 变现周期的一致性压力测试 日期： 2026 05 19 分析师： tmt analyst (TMT 行业分析师) 立场： 压力测试（负向背离） 执行摘要 截至2026年5月19日，TMT 行业正面临上游基础设施投入与下游收入变现之间史…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027-2028年电力缺口不会终结AI芯片周期，但会把下一代AI硬件迭代逻辑从单芯片峰值性能改写为受限MW下的有效AI输出、上线时间和机架级能效。", "summary": "截至2026-05-18，本卡支持前序电力瓶颈主线，但压力测试“硬件从极致性能转向能效”的表述：更准确地说，采购函数会转向每已通电MW的有效AI输出。2027-2028年表后燃气仅4-6GW，无法覆盖AI负荷缺口；DOE/LBNL预计美国数据中心用电2028年达325-580TWh。NVIDIA …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI推理经济学可以为一线自有栈承接HBM驱动的$/token地板，但通用API层会被开源模型与批处理/缓存路由压到成本线附近。", "summary": "截至2026-05-18，本卡综合第07卡压力测试：HBM驱动的$/token上行不会关闭2026-2028年AI硬件周期，但会显著分层。GPT-5.5、Claude Opus/Sonnet等前沿实时推理仍有$7.5-15.0/1M混合token的标价空间，能由一线云、广告、生产力与企业工作流吸收…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 1, "neutral": 0, "value": 67.8, "events": [ { "title": "截至 2026-05-19，AI 基础设施仍是 TMT 领涨主线，但可投资范围应收窄到有真实 usage、backlog、产能瓶颈和现金流承受力的环节。", "summary": "工作日期 2026-05-19。本文否定前序空白 TMT 论点并选择 (A)：AI 基础设施仍是领涨主线。证据是 hyperscaler capex/收入比虽已升至约 30% 的四家公司 Q1 合计强度，且 Alphabet、Amazon 的 FCF 张力明显，但 Microsoft AI ARR…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "若 2026-06-11 验证 Waymo 单位经济，AI 基础设施不应被卖出，而应从拥挤算力前端轮动到电源液冷、networking 邻近环节与小仓位机器人硬件 beta。", "summary": "截至2026-05-19，本卡支持 6 月 11 日后从 AV 扩散到 AI hardtech，但只在 Waymo 经济性披露有用时执行。确定性最高的是电源/液冷/机柜、定制芯片与 networking、HBM/先进封装；估值弹性最好的是电源热管理与机器人端侧硬件。HBM、GPU、A 股光模块和主…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "国产AI效率、软件栈和推理优先部署可以抵消应用层硬件缺口的大部分边际压力，但无法消除前沿训练和生态成熟度折价；投资应聚焦高利用率推理软件与能按结果变现的应用主题。", "summary": "2026-05-20：本卡综合判断，国产AI效率路线能显著缓冲应用层硬件约束，但不能完全替代前沿训练硬件、HBM、先进封装和CUDA式生态成熟度。DeepSeek/Qwen的MoE、thinking budget、KV-cache复用、量化和推理服务框架，把关键问题从“是否有足够GPU训练最大模型…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 0, "neutral": 1, "value": 71.6, "events": [ { "title": "国家通过主权信用金融化工具（绿色REITs/特国债）对西部可再生能源资产进行证券化打包，为国产GPU集群注入低成本、环境可溯源的“主权碳护盾”，直接对冲海外Scope 3合规溢价，确…", "summary": "截至2026年5月22日，中国算力出口面临的并非直接的CBAM关税，而是由欧盟《企业可持续发展报告指令》（CSRD）、《能源效率指令》（EED）和《人工智能法案》共同构成的B2B穿透式Scope 3碳合规壁垒。国家“东数西算”枢纽通过主权信用背书的绿电资产包（GPAPs，集成了0.22-0.28元…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 1, "value": 67, "events": [ { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "鉴于华孚时尚具备极高估值弹性的国产异构GPU算力期权以及A股极其恶劣的融券借券摩擦，强力做空002042.SZ将暴露于毁灭性的逼空与停牌风险之下，必须限制其权重并转向全球品牌商期权表…", "summary": "截至2026-05-23，本卡对前8张卡片进行穿透式审计。发现Research note 06存在严重的证券代码标错偏误（将华生科技605180.SH混淆为华孚时尚002042.SZ），且Research note 08的A股70/30 L/S策略存在致命的未对冲尾部风险：做空华孚时尚忽略了其华为…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 1, "value": 62.4, "events": [ { "title": "超大规模云厂商的选址与架构调整真实且大幅推进，但2028年前无法完全弥补物理电力缺口，导致2027-2029年AI资本支出出现显著的'已公告 vs 已通电'分化与15-25%部署滑期。", "summary": "截至2026-05-23，TMT端压力测试显示超大规模云厂商正沿四条路径回应第8卡的电力缺口：地理分散（AWS印第安纳/密西西比/沙特HUMAIN、Meta路易斯安那、Microsoft威斯康星）、用户表后电源（Talen-AWS、Constellation-MSFT 20年PPA、Crusoe …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] DeepSeek V4 推理成本通缩影响", "summary": "致 日报总编： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 1aa2204b-abb9-4b6f-9d43-fdb2fa0b0646 结果摘要: Wrote handoff_response.md (9995 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "'效率降级'对 2026–2027 AI Capex 总量近乎无约束，但通过电力、关税、航运三条渠道侵蚀单位经济学，使全球 AI 基础设施从'同步扩张'转为'地理极化'，估值真正承压…", "summary": "本卡对 Research note 1–5 的“效率降级削弱 AI 算力扩张”假设进行双向压力测试。结论：方向正确但程度被高估，AI 算力 Capex 在 2026–2027 名义增速仍 +19–26% YoY（北美 hyperscaler 2026 指引 4,400–4,650 亿美元 [S1–…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 5, "neutral": 4, "value": 54.8, "events": [ { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek V4 标志着原始推理 token 化工品化，而不是 AI 全产业零毛利化；应把它理解为利润池从模型 API 向算力底座和工作流应用迁移。", "summary": "截至2026-05-26，本卡支持并细化既有判断：DeepSeek V4 的低价把原始模型 API token 推向重资产、低定价权、类似大宗化工品的层级，但并不意味着整个 AI 行业进入零毛利。利润正在从可替代模型调用迁移到上游芯片/网络/电力约束的基础设施，以及下游企业数据、治理和工作流集成。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI 基础设施散户拥挤是否耗尽 Thesis v8 入场窗口", "summary": "致 市场情绪师： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: b790980f-74a8-48f2-8307-cf1e60b887ee 结果摘要: Wrote handoff_response.md (14984 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？", "summary": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？ DeepSeek-V4-Pro 75% price cut × NVDA FY27Q1 Data Center +92% joint p…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 1, "neutral": 1, "value": 71.2, "events": [ { "title": "I have completed the stress test analysis of Behind the Meter (BTM) power economics and its…", "summary": "I have completed the stress test analysis of Behind the Meter (BTM) power economics and its downstream impact on cloud service provider (CSP) operati…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "训练capex不是需求坍塌，但宽基算力alpha正在成熟化；2026下半年应保留瓶颈算力、提高应用层利润验证权重。", "summary": "截至2026-05-30，本卡综合判断：训练capex的边际ROI尚未确认跌破拐点，NVIDIA、Microsoft和Amazon仍显示算力供给约束；但云厂商capex规模、短寿命GPU/CPU占比和自由现金流压力使宽基基建beta的估值门槛上升。2026下半年A股AI alpha更可能从宽基in…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 1, "risks": 0, "neutral": 0, "value": 83.6, "events": [ { "title": "AI 算力的'电力税'利润池正从被监管电网转移至云厂商自筹融资的分布式平行系统——近端属于燃气轮机/发动机 OEM 与天然气中游，远端属于 SMR 厂商与 HALEU 燃料循环。", "summary": "卡片 04（能源行业分析师，立场=支持）延伸卡片 03 的结论：PJM 5+ 年并网排队与 3 年 GPU 寿命的错配迫使 AI 园区离网。2026–2029 年由表后航改型燃气轮机与往复式机组（CAT、CMI、Wärtsilä、Bloom）主导桥梁方案，已有 XOM 二叠纪 1.5 GW、CVX…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-01", "supports": 1, "risks": 0, "neutral": 0, "value": 85, "events": [ { "title": "中国头部AI算力链与具备全球工程化能力的设备/品牌出海龙头已构成可量化、可持续的盈利对冲，足以验证card-01的哑铃进攻端，但有效持仓宽度仅约A股TMT流通市值的20-30%，配置…", "summary": "TMT产业证据支持card-01哑铃进攻端：2026年4月集成电路出口310.85亿美元同比+100.1%；工业富联Q1营收2510亿元(+56.5%)、AI GPU机柜出货+380%、在手订单>2800亿元覆盖至2028；浪潮在手订单>3500亿元；华为Ascend 2026收入预计~120亿美…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-02", "supports": 3, "risks": 0, "neutral": 1, "value": 96, "events": [ { "title": "本地AI Agent足以支撑NVIDIA/Microsoft/Arm的70%核心权重，但证据集中在开发者、企业治理和创作工具三条真实工作流，而不是所有AI PC芯片供应商。", "summary": "截至2026-06-02，本卡支持NVIDIA/Microsoft/Arm合计70%核心权重，但理由应限定为真实工作流而非泛AI PC概念。开发者Agent已有VS Code/GitHub本地工具链、付费Copilot套餐和硬件内存/GPU拉动；企业Agent有Microsoft 365 Copi…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "DeepSeek V4 是AI推理买方议价权的拐点，不是所有SaaS估值倍数的系统性下调信号。", "summary": "截至2026-06-03，本卡支持“大模型公用事业化、价值向下游工作流转移”的既有判断，但否认DeepSeek V4已带来完整算力平权。V4-Pro相对GPT-5.5在缓存未命中输入、缓存命中输入和输出上分别约便宜11.5x、138x和34.5x；但企业支持、数据治理、IP来源和生态集成仍构成差异…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI端侧Agent是长期利润池迁移而非单日展会炒作，但最值得超配的是生态控制层，而不是所有PC芯片供应商。", "summary": "截至2026-06-02，本卡支持“AI本地工作流重分配PC芯片利润池”的结构性判断，但不支持无差别追逐AI PC概念。配置建议是NVIDIA 30%、Microsoft 25%、Arm 15%、AMD 12%、Qualcomm 10%、Intel 8%。核心逻辑是利润池转向CUDA/RTX、Wi…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] DeepSeek V4 定价冲击", "summary": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 2c82920b-afd7-4f0c-92a4-994d709ac7f2 结果摘要: Wrote handoff_response.md (11108 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 2, "neutral": 0, "value": 67.8, "events": [ { "title": "USMCA原产地规则（ROO）的收紧将使依赖进口组件的墨西哥电力设备组装产能对冲逻辑彻底失效，而沙特产能也面临美国 EAR/FDPR 针对 AI 基础设施的次级制裁风险。", "summary": "2026-06-03：本卡从政策监管视角压力测试出海产能对冲逻辑。USMCA 2026 联合审议实质性收紧了原产地规则，强制将“熔炼和浇注”标准延伸至变压器等核心电力设备。由于中国控制多数 GOES 产能，单纯依靠“组件进口+墨西哥组装”的产能将无法通过 RVC 测试，对冲逻辑彻底破产。同时，沙特…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网容量与清洁能源转型进度已从远期风险演变为当期约束，成为制约 AI 基础设施 ROI 兑现的核心物理瓶颈：互联排队 5–7 年与 GPU 4–6 年经济寿命的时点错配、加上燃机/变…", "summary": "压力测试支持 Research note 1 论断并提升其紧迫性：电力供给已成为 AI 资本开支 ROI 的首个出现违约的物理约束。证据链：（1）IEA 预测 2030 年全球数据中心用电翻倍至 ~945 TWh，LBNL 测算美国份额升至 7–12%；（2）美国互联排队平均 5 年、PJM 20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 0, "neutral": 1, "value": 84.2, "events": [ { "title": "[auto-handoff] DeepSeek V4推理通缩对A股AI SaaS与网络安全企业单机/单席位ARR与毛利率定量模型", "summary": "致 首席策略师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: faeaa5cd-1413-4f2e-9a6f-388c2f6d9525 结果摘要: Wrote handoff_response.md (3125 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 2, "neutral": 0, "value": 70.6, "events": [ { "title": "AIDC的投资核心是‘以电力溢价换取时间期权’以规避GPU闲置折旧，而非单纯的度电成本控制；BTM燃气虽能守住7%–10%的ROIC合格线，但资产估值已面临结构性下修。", "summary": "本卡综合Research note 01/02表明，AIDC投资重心并未转向传统“极端成本控制”，而是转向“时间期权套利”与“资本分层”。在53个月并网延迟下，防范GPU闲置折旧是核心，表后燃气虽使LCOE高出1.7-2.5倍，但缩短了2年上电期，保住了7-10%的ROIC底线。因此，“电力可用性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电力瓶颈不会终结Scaling Law，但会把前沿模型进步压缩成更少实验室、更大台阶、更强能源地理依赖的发布节奏。", "summary": "截至2026-06-07，本卡综合前序AIDC ROIC与BTM结论：53个月并网排队不会让Scaling Law失效，但会把前沿模型路线图从GPU约束推向已上电GW级集群约束。若训练算力继续每年4-5倍增长，约28个月并网额外延迟会造成约27-42倍目标算力缺口。基准判断是形成2026-2028…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 1, "risks": 0, "neutral": 0, "value": 92, "events": [ { "title": "在结果型计价与KA工作流锁定双重前提下，企业AI订阅扩张足以支撑SaaS厂商ARPU不下滑，Research note 08/09的'硬件10/SaaS 38/红利现金52'三桶配置…", "summary": "作为终卷第10卡，消费分析师从企业付费心智维度支持Research note 09：在KA工作流锁定窗口期内，企业AI模块订阅扩张（席位×用量×分层）足以使SaaS厂商综合ARPU维持平稳至温和上行，前提是厂商以结果型/Agent型SKU封装而非纯token转售。全球数据显示企业GenAI支出年增…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 0, "neutral": 3, "value": 94.4, "events": [ { "title": "DeepSeek V4推理通缩下SaaS及网安应用恶性价格战与ARR/现金流折现久期重构", "summary": "DeepSeek V4推理通缩下SaaS及网安应用恶性价格战与ARR/现金流折现久期重构 DeepSeek V4 Cost Deflation: MaaS Price Wars and ARR Valuation Durations in SaaS 大模型API成本通缩11.5x-138x后，国内…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] DeepSeek V4推理通缩对A股AI SaaS与网络安全企业单机/单席位ARR与毛利率定量模型", "summary": "致 首席策略师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 6e053e1d-f8d3-41be-9b93-6ea51f4142be 结果摘要: Wrote handoff_response.md (13840 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "DeepSeek V4推理通缩下SaaS及网安应用恶性价格战与ARR/现金流折现久期重构", "summary": "DeepSeek V4推理通缩下SaaS及网安应用恶性价格战与ARR/现金流折现久期重构 大模型API成本通缩11.5x-138x后，国内AI SaaS及安全企业是否会陷入价格恶性竞争？推理成本下降能否转化为企业经常性订阅收入（ARR）与毛利红利？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 1, "neutral": 2, "value": 94.8, "events": [ { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "核电交付时滞会压低2026-06至2027年的AI算力收入变现斜率，但不会立即等同于GPU采购需求见顶。", "summary": "截至2026-06-10，本卡支持前序判断但作出TMT侧区分：2026-06至2027年更可能出现“通电、可确认收入的AI算力容量”平台期，而非GPU订单断崖。Microsoft FY26 Q3资本开支319亿美元且约三分之二投向GPU/CPU，Amazon Q1 PPE购买额442.03亿美元，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "DeepSeek降价推动的是硬件内部的alpha重排，而不是AI硬件beta被证伪；应压缩泛化边缘和老库存，保留订单确认的光模块、HBM相关存储、液冷/电源和高端板卡。", "summary": "截至2026-06-09，DeepSeek V4-Pro 已验证永久降价为75%、至原价1/4；“96%”更像相对竞品的单位经济冲击。微观压力测试显示，GPU/HBM/800G与1.6T光模块/液冷仍受订单、价格和产能约束，Jevons弹性成立。中际旭创、沪电股份、佰维存储有收入、交付或稼动率确认…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Agentic AI 推理成本墙", "summary": "致 市场情绪师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 587b6589-fb62-4049-b55e-219006d7a63a 结果摘要: I will run the command to anchor today's date and record the…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 0, "value": 82.2, "events": [ { "title": "2026-06-10的主线不是在云、终端和政策之间三选一，而是超配云需求确认，选择性持有HBM/国产替代，暂把Apple换机作为二阶段期权，并低配政策与现金流暴露高的AI资产。", "summary": "截至2026-06-10，本卡综合Oracle业绩窗口、Apple AI终端升级和美国先进芯片海外子公司管制澄清。结论是：A股AI链条应优先交易云需求确认，其次保留Apple端侧AI换机期权，同时用出口管制和自由现金流压力控制仓位。Oracle、Microsoft、Alphabet、Meta仍显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 6, "neutral": 1, "value": 56.6, "events": [ { "title": "铜、铝、电缆和电气设备稀缺已经进入AI算力集群交付链，最先约束的是通电日期和可售云容量，而不是GPU本身。", "summary": "截至2026-06-11，材料链稀缺已开始转化为AI基础设施交付摩擦，但表现为通电时间、电气设备交期和云算力容量分配，而非所有GPU集群停摆。IEA提示约20%规划项目有延迟风险，CBRE称电力约束把工期推至2027年及以后，Wood Mackenzie给出关键电气部件18-36个月交期和600 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "当前不是典型产能过剩，而是“过度融资、电力受限”——电网的物理配给压住了空置式过剩，却抬高成本地板，使折旧—变现错配成为近期真正风险。", "summary": "对 Card-01“3.3% 激活陷阱”做供给侧压力测试，得出双面结论：过剩真实但结构分化——通用/错配算力（中国西部上架率低、Microsoft 放弃 >14% 产能）与融资结构已过剩，前沿 GPU 尚未普遍闲置；2026 年超大规模厂商资本开支约 7,250 亿美元对约 510 亿直接 AI …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与费率约束把 hyperscaler 新建 capex 中电力、散热与合同部分的占比从约 15% 推升至 35–45%，并迫使物理部署向表后燃气、核电/煤电棕地共址、SMR 与非…", "summary": "电网约束已成为 hyperscaler AI capex 的硬约束。结构由 2023 年约 85% GPU/15% 电力外壳，迁向 2026 年新建约 55–65% IT / 35–45% 电力、散热、变电、自备发电与长期 PPA。500 MW 园区全口径由约 $5.5B 升至 $12–16B，每…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力交付滞后会迫使 Hyperscalers 调整 GPU 集群投产节奏，并通过更长现金回收期、更低早期利用率和更高电力保障成本压低 AI 基础设施 capex 效率。", "summary": "截至2026-06-11，本卡支持上一卡：AI capex 不会因电网瓶颈消失，但 GPU 集群部署节奏会被“已通电 MW”重新排序。变压器、并网和输电延迟会把部分已宣布园区容量推迟到后续窗口，压低早期自由现金流和资产利用率。Hyperscalers 会通过已通电站点溢价、分阶段投产、区域重排和自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 2, "neutral": 1, "value": 82, "events": [ { "title": "AI 基础设施从‘算力’向‘电力’的范式转移", "summary": "PJM 容量价格三年连涨约10倍（28.92→333.44 美元/兆瓦·日，连续两年顶死 FERC 上限），是结构性稀缺信号而非周期尖峰。但「必须从 GPU 摩尔定律切换到公用事业资产底座」是错误的二选一。真相是瓶颈迁移：边际稀缺租金从硅片下移到可按期交付的确定性电力。估值应改为双因子模型——保留…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-13，电力设备成本与交付风险已经实质性延后 AI 算力集群通电节奏，但它表现为 capex 效率恶化和项目重排，而不是最大型 hyperscalers 已经全面削…", "summary": "截至2026-06-13，本卡支持前序“电力约束正在惩罚 AI capex/融资假设”的方向，但精确化为：2026-2028 年已出现项目层面的通电延期与节奏重排，尚未出现头部 hyperscalers headline capex 的广泛下修。Sightline 显示 2026 年约 16 GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施从‘算力’向‘电力’的范式转移", "summary": "AI 基础设施从‘算力’向‘电力’的范式转移 AI Infrastructure Shift: From Compute to Grid Power PJM 电力容量价格暴涨10倍是否意味着 AI 估值逻辑必须从 GPU 摩尔定律转向公用事业的资产底座逻辑？", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 4, "neutral": 1, "value": 71.4, "events": [ { "title": "Higher-for-longer不会必然叫停AI capex，但会把市场关注点从“能否花钱”转向“每个增量GPU集群能否在折旧和电力成本吞噬定价窗口前跨过更高ROI门槛”。", "summary": "截至2026-06-14，本卡对卡片1的高利率与美联储战略模糊情景做压力测试：AI基础设施capex的关键脆弱点不是大型云厂商短期无钱可投，而是折现率、GPU短寿命折旧、电力并网与PPA成本共同抬高项目资本回收门槛。以三分之二6年期算力资产、三分之一15年期数据中心资产测算，200bp WACC冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超大规模云厂商 AI 资本开支撞上电网墙:2026 年从资本约束切换至电力约束的相位转换", "summary": "2026 年 AI 基建的边际约束已从 GPU 与资本迁移至电力与电网设备。Top-6 云厂商资本开支约 6,900 亿美元(同比+27%),但 PJM 并网周期 18–36 个月、大型变压器交期约 128 周、PJM 2027–28 容量缺口 6.6 GW。Microsoft 自行披露 800 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力中心的电力激增与电网并网瓶颈，为电力板块提供了超越传统周期的长协盈利溢价，强力支持将红利配置中电力权重上调至2.5%以对冲地产下行。", "summary": "截至2026-06-13，AI基础设施分析师综合验证了上调电力板块权重的逻辑。IEA《Electricity 2026》显示，AI推理带来常态化电力负荷，全球数据中心耗电将在2030年翻倍逼近950 TWh。国内十万卡GPU集群建设使并网与变压器成为核心瓶颈。电力企业受燃料成本下降与AI电荒预期双…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "硬件与能源通胀只能部分向软件端转嫁，头部云厂商2026年不至于利润率崩塌，但AI云业务的边际回报率和自由现金流将被结构性压缩。", "summary": "截至2026-06-14，本卡支持前序“资本强度陷阱”链条：云厂商可在稀缺GPU、预留容量、企业工作流和agent定价中转嫁部分成本，但公开API价格已被OpenAI、Anthropic、Google等竞争锚定，难以承受全面涨价。AWS、Azure、Google Cloud短期报表利润率未崩塌，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 3, "neutral": 0, "value": 75.8, "events": [ { "title": "物理电网瓶颈正迫使 AI 基础设施从‘暴力规模扩张’转向‘受约束的能效优化’，电力并网速度已取代芯片供应成为核心估值护城河。", "summary": "本报告通过压力测试确认了物理电网并网排队（7-10年）与变压器交期（4-5年）已成为 AI 基础设施扩张的硬约束。定量模型显示，1年的并网延迟将导致 100MW 数据中心损失 5.5% 的生命周期价值，且快速的 GPU 迭代周期（12-18个月效用减半）加剧了闲置产能的 ROI 损耗。这种物理瓶颈…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "随着AI物理瓶颈从芯片转向电网可交付性，定价权正全面向上游电力资产与输配电设备商迁移，应超配瓶颈控制方并对电网依赖型云厂商及芯片标的引入折价。", "summary": "本卡片将2026年电网物理交付瓶颈转化为权益资产配置框架。建议超配控制变压器、液冷与在运核电等瓶颈环节的设备商与独立发电厂（如GE Vernova、Eaton、Vertiv、Constellation）；对依赖电网队列且发生租约撤销的云厂商（如Microsoft、Meta）引入20%-30%的电网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与电网瓶颈对Tier-1 2026–2027并网影响已基本被锁单、棕地与表后发电对冲；真实滑期约3–7 GW集中在2028+绿地与Tier-2投机建设，Research not…", "summary": "本卡(10/10)压力测试变压器/GSU/电网交期对Tier-1 AI数据中心并网的真实约束。结论：瓶颈真实存在但Tier-1时间表已基本被定价——2022–2024年下达的多年期框架订单、棕地复用既有并网点、表后核电与现场燃气PPA共同吸收了120–210周的设备交期。自测算显示2027–202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 1, "risks": 2, "neutral": 1, "value": 94.2, "events": [ { "title": "评估电力供应瓶颈与电网扩容速度对AI算力集群建设及超大规模云厂商（Hyperscalers）资本开支的影响", "summary": "本卡支持并强化 Research note 01 的'公用事业→AI核心基座'论点：电力供应与电网/配电设备交付周期（而非GPU供给）才是当前AI算力部署与云厂商资本开支变现的真正约束。2026年五大云厂商合计资本开支约6,600–6,900亿美元，近100%吞噬经营现金流，但Microsoft约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡片 02 — AI 芯片供应链：DeepSeek V4 后的\"架构定制化\"趋势 白板：1f5523fe c230 4de9 bcab 6fc3d7edec87 ｜ 卡片 2/8 ｜…", "summary": "卡片 02 — AI 芯片供应链：DeepSeek V4 后的\"架构定制化\"趋势 白板：1f5523fe c230 4de9 bcab 6fc3d7edec87 ｜ 卡片 2/8 ｜ 日期：2026 06 15 分析师：TMT 行业分析师 ｜ 立场： 支持（support） 核心问题：DeepSe…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI 行业的估值核心已从‘拥有多少 GPU’转向‘能以多低的成本交付智能’，DeepSeek V4 是这一转折点的催化剂。", "summary": "DeepSeek V4 的发布及其永久降价策略标志着 AI 估值逻辑从‘资源型’（算力堆砌）向‘效率型’（每瓦特智能）的范式转移。V4 通过极致稀疏化架构和硬件解耦（昇腾堆栈），打破了 NVIDIA 的算力垄断叙事，导致 NVIDIA 通用算力需求曲线结构性压平，并迫使全球模型厂商进入‘推理利润率…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 6, "neutral": 2, "value": 66.6, "events": [ { "title": "单纯的电力价格上涨无法动摇SaaS效果付费的推理成本通缩基础，但物理电网容量限制与变压器交期导致的GPU集群部署瓶颈才是核心威胁，其引发的算力稀缺溢价（如代币价格反弹至1.00美元以…", "summary": "针对2026年6月17日算力侧电力约束，本报告对SaaS效果付费模式的财务稳定性进行了压力测试。测算表明，即便面临0.450美元/kWh的极端电价冲击，B200单卡百万Token的电力TCO成本仅从0.007890美元上升至0.010218美元，远低于DeepSeek V4-Pro的0.15美元市…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建资本开支仍在加速，但从公告美元转化为已通电MW的比例正在下降，首个断点在数据中心交付和二线下游供应链，而不是NVIDIA近端GPU订单。", "summary": "截至2026-06-16，未看到北美云厂商AI资本开支或GPU订单总量已经广泛下修；Microsoft、Alphabet、Meta、Amazon、NVIDIA、Dell披露仍指向扩张。支持前序压力结论的证据主要来自数据中心交付层：Goldman Sachs预计未来1-2年计划容量仅50-60%按时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与散热瓶颈重塑 ABF 需求的斜率与时点，但因 ABF 在封装上游锁定且单卡用量持续升级，无法构成结构性下台阶。", "summary": "电力与散热是限制 hyperscaler 部署 AI 硬件时点与地理的物理硬约束（PJM 2025/26 容量拍卖出清价 ~329 美元/MW-日为上年 9 倍；机架功率从 120 kW 跃升至 600 kW+），但并非 ABF 载板需求的结构性上限。ABF 在封装阶段较上架早 6–9 个月锁定，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI推理算力极端通缩解决了效果计费下的高昂COGS限制，使综合毛利率回升至89%的高位，从而在财务层面彻底支撑起SaaS企业从传统席位制向效果付费模式的历史性跨越。", "summary": "截至2026-06-17，大模型推理成本通缩（以DeepSeek V4-Pro/Flash为代表的67x-200x降幅）是支撑SaaS从“席位费”转向“效果付费”的财务前提。测算表明，在$10/百万Token的历史价格下，效果计费因算力COGS占比过高导致毛利率仅有27.50%；而在V4系列$0.…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 2, "neutral": 0, "value": 86, "events": [ { "title": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：…", "summary": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：AI 基础设施分析师 (ai infrastructure analyst) 立场：stress test（压力测试…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler capex 足以穿越能源 TCO 上行并支撑稀缺供应链，但投资表达必须是“多稀缺、空低 ROIC”，而不是无差别做多 AI。", "summary": "2026-06-18 压力测试结论：hyperscaler capex 增长足以让 AI 基础设施供应链消化能源成本带来的边际 TCO 压力，但只能支持“多稀缺”，不能支持泛 AI beta。四大买方 2026 capex 粗略包络约 $710B（自测算），Microsoft、Alphabet、M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 2, "neutral": 1, "value": 90.4, "events": [ { "title": "截至2026-06-19，前沿模型风险闸门更可能改变AI基础设施资本开支的节奏和形态，而不是终止云厂商的算力扩张。", "summary": "截至2026-06-19，本卡支持上一卡结论，但把传导机制落到AI基础设施：风险闸门不会简单压低hyperscaler资本开支，而是把节奏改为“电力/园区前置、GPU分段、合规容量池预留”。Anthropic Fable 5 / Mythos 5的2026-06-09发布与2026-06-12暂停…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Dominion/AEP式tariff会把PJM AI负荷从“谁先排队”改成“谁能证明地点价值”，推理和企业云留在PJM，移动训练和投机容量外迁。", "summary": "2026-06-18：本卡综合前序观点。Dominion/AEP式tariff不是禁止AI数据中心进入PJM，而是筛选负荷质量。低延迟推理、受监管企业云、既有availability zone密度和有区域合约的AI cloud仍能证明留在PJM合理，因为按PJM $329.17/MW-day容量价…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "大模型价格战引发的推理流量爆量，导致国产GPU设计龙头呈现“营收高增、毛利率承压”的量价背离，定价权向制造端瓶颈与TCO优化环节倾斜。", "summary": "在大模型极低价格战（DeepSeek-V4-Flash降至$0.14/1M tokens）推动下，云厂商Q1 AI基础设施收入大增79%（GPU云激增184%）。但因先进制程及封装产能受限导致BOM成本上涨10%-15%，加之云厂商降本压价，国产GPU（如海光一季度毛利率降至55.6%，同比下滑5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 3, "neutral": 0, "value": 82.8, "events": [ { "title": "Starlink 带宽 × Grok 算力的飞轮真实，使 SPCX 成为 AI 分发基座与算力链条需求催化剂，但算力归属 xAI、轨道边缘算力尚萌芽，真正瓶颈是电力——故支持但需把光…", "summary": "从 AI 基础设施视角支持卡片 01：Starlink 带宽（约 450 Tbps，三代卫星单星 >1 Tbps、下行 10 倍/上行 24 倍）与 Grok/Colossus 算力（约 55.5 万颗 GPU、约 180 亿美元、向 2 GW 扩张）构成真实飞轮，使 SPCX 成为差异化的'算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供应瓶颈与电网升级滞后对 AI 基础设施效率及全球能源板块投资机会的影响", "summary": "2026-06-19：本卡支持第05卡判断，并提供其物理底层。结论：电力是约束性瓶颈，比CoWoS硅侧更深、更慢。GPU约12个月可重新流片，但并网排队中位约5年[S3]、高压变压器交期36–48个月[S4]、燃气轮机槽位售罄至2029–2030年[S2]。五道硬锁中四道在2028年后才缓解，把张…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A-share de-leveraging to 1550 points triggers secondary market liquidity locking and margin…", "summary": "若市场下探至1550点，中际旭创股价将跌至714.16元，击穿融资预警与平仓线，带来日均31.2亿元平仓抛压（占ADV 17.36%），锁定二级流动性。但公司持有125亿现金及45亿海外客户预付款，可提供151.5亿净营运资金，供应链运营违约风险极低。海外订单真正风险在于电网并网排队达36-60个…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 5, "neutral": 0, "value": 72.2, "events": [ { "title": "AI capex supercycle 仍在，但 higher-for-longer 与能源波动会把扩产闸门从芯片采购转向电力可交付性、容量成本和融资期限匹配。", "summary": "截至 2026-06-21，本卡对 higher-for-longer 与能源波动环境下 AI 算力基建做压力测试。结论：超大规模云厂商不会停止扩产，但边际决策从“GPU 可得性”转向“firm megawatts + 客户承诺 + 可融资全成本”。Amazon、Alphabet、Meta 已披露…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "summary": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "VPP能守住部分尖峰成本，但真正维持AI算力租赁利润率的是高利用率、可送电站点和选择性西迁，而不是单纯的软件调度。", "summary": "截至2026-06-20，本卡片压力测试AI算力设施成本传导：VPP与算力调度能降低尖峰电价暴露、提高接网可信度，但无法替代本地变电容量，也无法覆盖GPU利用率下滑和高capex折旧压力。基准判断是选择性西迁：训练、批量推理和非实时任务应迁往西部绿电枢纽；低时延推理和企业云可用区保留在东部并显性传…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网通电限制不会立即终结 AI 需求，但会把 TMT 估值从“芯片稀缺溢价”重构为“确定电力、利用率和现金流转化溢价”。", "summary": "截至 2026-06-21，本卡压力测试电网通电限制对 TMT 的影响：AI 需求仍强，但估值核心从 GPU 可得性转向 time-to-power、容量变现和 FCF 久期。Microsoft 和 Amazon 的主要风险是 AI capex 先发生、通电和云收入后确认，导致利润率与自由现金流承…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 0, "value": 91.6, "events": [ { "title": "AI 基础设施建设中的电力供应与电网容量瓶颈分析", "summary": "本卡压力测试 Research note 02 的隐含假设「资本开支≈算力」。证据显示约束已从芯片迁移到电力：微软披露约800亿美元 Azure 订单因电力受限无法交付，Nadella 称芯片堆库存却无「暖壳」可插[S5][S4]。转化在三处漏损：(1)并网排队约2,600 GW、中位等待约5年、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的边际稀缺点正从GPU扩散到可通电兆瓦和电网电气设备，因此制造出海主线应从拥挤硬件β轮动到“电力绑定的算力基础设施”。", "summary": "截至2026-06-21，本卡支持上一卡的再平衡结论：AI硬件存在战术拥挤，但云厂商资本开支并未结束，而是转向算力、电力、电网设备协同扩建。Microsoft、Alphabet、Meta、Amazon的2026资本开支披露或指引合计约7000亿美元（自测算），而DOE/LBNL、IEA、EPRI和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 0, "value": 93, "events": [ { "title": "在66.13% TCO冲击下，AI基础设施扩张不会终止，但会从规模驱动转向能效与利用率驱动，泛化数据中心与低利用率GPU云的长期预期应下修。", "summary": "截至2026-06-22，本卡压力测试上一卡66.13% TCO激增情景。结论是：头部hyperscaler资本开支不会急停，Microsoft、Alphabet、Meta、Amazon仍有收入、backlog或客户承诺支撑；但新增产能逻辑会从“抢GPU/抢机房规模”转向“单位功耗有效产出、签约利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026年中美云厂商CapEx受阻于电力红线以及大模型推理通缩的双重催化下，新一代信息技术产业链估值正向能耗比占优的定制ASIC上游填料（联瑞新材）以及享受API接口成本大坍缩的软…", "summary": "截至2026-06-22，新一代信息技术产业链正经历算力瓶颈由GPU转向电力/并网及先进封装、以及DeepSeek API降价96.37%引发软件应用毛利重构的剧烈调整。板块激活率处于4.3%极低水平。本报告筛选了联瑞新材、生益电子、宏和科技、金山办公和富创精密等5家低两融、抛压出清且具刚性催化剂…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 6, "neutral": 0, "value": 70.4, "events": [ { "title": "2028年再融资悬崖与 AI 算力基础设施资本开支的信用冲突", "summary": "压力测试支持并延伸卡片 04：AI 基础设施信用风险真实且上升，但呈哑铃状集中，不在头条的 $725 bn 巨头资本开支上。四巨头（MSFT/GOOGL/AMZN/META，2026 资本开支约 $725 bn、+77%，约 75% 投 AI [S1]）凭现金流与 AA 资产负债表通过测试，尽管 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027年的电力冲击会抬高AI容量成本并重排Hyperscaler部署顺序，但更可能造成通电与收入确认延迟，而非CapEx总量或GPU订单的广泛取消。", "summary": "截至2026-06-23，本卡支持前序判断：2027年电力约束不会直接演化为Hyperscaler CapEx总量削减，而会把部署逻辑从“GPU优先”改为“通电优先”。Microsoft、Alphabet、Meta、Amazon披露的2026 AI基础设施支出仍然庞大，但PJM容量价格、美国DC电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler并不缺AI Capex，真正约束算力上线节奏的是可交付电力、并网流程与长交期电力设备。", "summary": "截至2026-06-24，本卡支持上一卡：AI算力瓶颈已从Capex/GPU转向通电MW。Alphabet 2026 CapEx指引1,800-1,900亿美元、Meta 1,250-1,450亿美元，Microsoft仍称需求超过可用容量，Amazon Q1购置物业与设备442.03亿美元；但S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 4, "neutral": 0, "value": 83.8, "events": [ { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "持续能源通胀会压缩AI基础设施回报率，但真正改变Hyperscaler扩建节奏的是可交付电力的稀缺，而不是电费单项上涨。", "summary": "截至2026-06-25，本卡压力测试能源通胀对AI基础设施的传导。结论是：电价上升本身不足以触发Hyperscaler AI capex断崖；更关键的是并网、容量市场、PPA和电网设备交付约束会重排项目优先级。Microsoft、Alphabet、Meta、Amazon仍合计指向约6950亿-7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "xAI 算力扩张对 AI 基础设施与能源供应的压力测试", "summary": "对卡片 01 前瞻风险的压力测试：跑步机确实存在，但卡点不在资本。约束层级为资本 < GPU < 电力。资本充裕（xAI E 轮 200 亿/估值 2300 亿，外加 200 亿 SPV 与 35 亿 Apollo 方案；2026 年超大规模厂商资本开支约 6000–7250 亿），GPU 因英伟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply s…", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 2, "neutral": 0, "value": 96, "events": [ { "title": "BTM是AI园区局部speed-to-power的有效工具，但在2026-2028年不足以填补全国AI电力缺口，反而会放大公用事业负荷预测兑现率与成本分摊的不确定性。", "summary": "截至2026-06-25，本卡从AI基础设施侧压力测试card-01：Hyperscaler BTM已从叙事进入执行，Microsoft/Chevron 2.67 GW燃气BTM、Crusoe/Microsoft 900 MW现场电厂、Talen/AWS 300 MW直连核电和xAI 247.2 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CNH套息去杠杆会先打掉A股AI基础设施股票的流动性溢价和估值久期，而不是立即打断全球AI CapEx实物周期。", "summary": "截至2026-06-25，本卡片压力测试CNH套息去杠杆对AI基础设施估值和CapEx预期的冲击。结论是：流动性溢价消失可使A股AI硬件/算力基础设施估值出现非线性回撤，但不应直接推导为全球hyperscaler CapEx崩塌。Microsoft、Alphabet、Meta、Amazon、Ora…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "summary": "截至2026-06-27，本卡片对AI基础设施链做压力测试：全球AI资本开支仍大，Goldman Sachs基准模型给出2026年USD 765 billion、2031年USD 1.6 trillion的年度AI capex，但Amazon、Alphabet、Meta、Microsoft的数据表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 5, "neutral": 0, "value": 82, "events": [ { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力建设“过剩与利用不足”下的TMT云厂商投资回报率（ROI）与算法效率边际", "summary": "截至2026-06-28，本卡对“算法效率能否抵消算力利用不足导致的资产减值”做压力测试。结论是只能部分抵消：DeepSeek-V3、Llama 4与端侧AI证明小模型/MoE能提高稀缺芯片吞吐并扩大推理需求，但也压低每任务云端收入、把简单任务迁移到端侧，并加速旧集群过时。头部云厂商仍有需求证据：…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞", "summary": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞 DeepSeek's 96.37% price cut: Jevons amplification or AI hardware demand destruction — head-to-…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高利率平台期对AI基础设施资本开支的影响", "summary": "压力测试卡片04的3.75%-4.00%高位平台前提：利率平台并非2026年头部Hyperscaler资本开支（约7250亿美元、+77%、约75%为AI）的约束性瓶颈，因四巨头仍主要自筹、AA/A级利差仅约80bp。但融资模式恰在约2026Q3翻转为债务驱动（整体资本开支超过经营现金流；Orac…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 5, "neutral": 0, "value": 91.4, "events": [ { "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产大模型并未告别规模竞赛，而是在先进GPU受限的条件下把竞争单位改写为每瓦、每token、每受限加速器和每留存用户的有效智能产出。", "summary": "截至2026-06-28，本卡对“国产大模型不可逆转向效能极限竞赛”进行压力测试。结论是：效能迁移真实且受GPU出口、H20许可、HBM/互连、液冷机柜和电力约束推动，但不是规模竞赛终结，而是进入“稀缺规模竞赛”。DeepSeek-V3、DeepSeek-R1、Qwen3和MiniMax-M1显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "数据中心电网并网瓶颈与核能转向", "summary": "截至2026年6月28日，AI基础设施建设的核心瓶颈已从GPU稀缺转移到电网并网时间（Time-to-Power）。对100 MW超大规模AI数据中心的模型测算显示，2年的并网延迟会减少项目35.73%的净现值（NPV），并使IRR大幅下降。为规避此类延迟，科技巨头（微软、亚马逊、谷歌、Meta）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 1, "risks": 5, "neutral": 0, "value": 92.8, "events": [ { "title": "AI 应用的 ROI 验证：从 Copilot 订阅到企业级代理", "summary": "AI 应用的 ROI 验证：从 Copilot 订阅到企业级代理 AI Application ROI Verification: From Copilot Subscriptions to Enterprise Agents Can enterprise AI agents drive dura…", "sourceLabel": "研究主题", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck fr…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞", "summary": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞 全所四个席位（utilities/industrials/tmt/thematic）高度一致认为AI约束已从GPU转向电网（LPT交期120-210周、并网排队53个月）；同周sentim…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-ab1fb6a453/", "networkHref": "/investor/research-network/#thesis-ab1fb6a453" }, { "id": "thesis-6f42e0fe6e", "source": "tracked_thesis", "title": "稀土、铜铝与关键材料正在成为 AI 与新能源共用的供给约束", "titleEn": "Rare earths, copper, aluminum, and key materials are becoming shared supply constraints for AI and energy transition", "lane": "关键材料/稀土铜铝", "laneEn": "Critical materials, rare earths, copper, and aluminum", "laneHref": "/investor/lanes/critical-materials-rare-earths-copper-and-aluminum/", "investableFocus": "关键材料/稀土铜铝：600111.SH, 000831.SZ, MP, FCX, SCCO, AA, TECK", "stockUniverse": [ { "ticker": "600111.SH", "name": "北方稀土", "market": "A-share", "role": "稀土资源", "roleEn": "稀土资源", "bucket": "core", "href": "/investor/stocks/600111.SH/" }, { "ticker": "000831.SZ", "name": "中国稀土", "market": "A-share", "role": "稀土整合", "roleEn": "稀土整合", "bucket": "core", "href": "/investor/stocks/000831.SZ/" }, { "ticker": "MP", "name": "MP Materials", "market": "US", "role": "美国稀土资源", "roleEn": "美国稀土资源", "bucket": "core", "href": "/investor/stocks/MP/" }, { "ticker": "FCX", "name": "Freeport-McMoRan", "market": "US", "role": "铜矿", "roleEn": "铜矿", "bucket": "core", "href": "/investor/stocks/FCX/" }, { "ticker": "SCCO", "name": "Southern Copper", "market": "US", "role": "铜矿", "roleEn": "铜矿", "bucket": "core", "href": "/investor/stocks/SCCO/" }, { "ticker": "AA", "name": "Alcoa", "market": "US", "role": "铝", "roleEn": "铝", "bucket": "core", "href": "/investor/stocks/AA/" }, { "ticker": "TECK", "name": "Teck Resources", "market": "US", "role": "铜与关键矿产", "roleEn": "铜与关键矿产", "bucket": "core", "href": "/investor/stocks/TECK/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "600111.SH", "000831.SZ", "MP", "FCX", "SCCO", "AA", "TECK", "XLE", "GLD", "VST", "CEG", "GEV", "ETN", "MU" ], "coreView": "赛道：关键材料/稀土铜铝。候选观察标的：600111.SH, 000831.SZ, MP, FCX, SCCO, AA, TECK。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 工业供给瓶颈。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 81, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 8, "risks": 131, "neutral": 21, "analystCount": 8, "sourceTypeCounts": { "研究笔记": 149, "证据更新": 2, "研究主题": 1, "分析协作": 7, "研究线程": 1 }, "topTerms": [ "Meta", "GOES", "AIDC", "新能源", "能源", "商品", "钢铁", "Rare", "earths", "copper", "aluminum", "materials" ], "whatChanged": [ "2026-06-29: 能源与大宗商品资产在宏观政策背离下的防御价值研判", "2026-06-29: 中游房建及家装开工收缩-30.67%通过高经营杠杆深度挤压了上游钢铁、铝加工及基础化工的营业利润率与订单量，导致三大行业面临深幅利润下修压力（加权下修-67.3%），并伴随着2026年首度碳市场履约成本与绿色壁垒压力，支持资金向防御型红利资产集中的重平衡策略。", "2026-06-29: [auto-handoff] 先进制程星载 OBP 芯片良率瓶颈与算力/通信设备材料供应链", "2026-06-28: 能源商品成本传导与USD/CNY汇率风险的压力测试", "2026-06-28: 能源/商品板块应以多元化商品篮子为核心、能源股为收益锚、少量原油为尾部风险工具，而不是把11%战略通胀对冲变成单一布伦特方向押注。" ], "practical": { "sectorLane": "关键材料/稀土铜铝", "sectorLaneEn": "Critical materials, rare earths, copper, and aluminum", "exposureBucket": "能源与关键材料", "tone": "red", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "600111.SH", "name": "北方稀土", "market": "A-share", "role": "稀土资源", "roleEn": "稀土资源", "bucket": "core", "href": "/investor/stocks/600111.SH/" }, { "ticker": "000831.SZ", "name": "中国稀土", "market": "A-share", "role": "稀土整合", "roleEn": "稀土整合", "bucket": "core", "href": "/investor/stocks/000831.SZ/" }, { "ticker": "MP", "name": "MP Materials", "market": "US", "role": "美国稀土资源", "roleEn": "美国稀土资源", "bucket": "core", "href": "/investor/stocks/MP/" }, { "ticker": "FCX", "name": "Freeport-McMoRan", "market": "US", "role": "铜矿", "roleEn": "铜矿", "bucket": "core", "href": "/investor/stocks/FCX/" }, { "ticker": "SCCO", "name": "Southern Copper", "market": "US", "role": "铜矿", "roleEn": "铜矿", "bucket": "core", "href": "/investor/stocks/SCCO/" }, { "ticker": "AA", "name": "Alcoa", "market": "US", "role": "铝", "roleEn": "铝", "bucket": "core", "href": "/investor/stocks/AA/" }, { "ticker": "TECK", "name": "Teck Resources", "market": "US", "role": "铜与关键矿产", "roleEn": "铜与关键矿产", "bucket": "core", "href": "/investor/stocks/TECK/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "600111.SH", "000831.SZ", "MP", "FCX", "SCCO", "AA", "TECK", "XLE", "GLD", "VST", "CEG", "GEV", "ETN", "MU" ], "primaryStocks": [ { "ticker": "600111.SH", "name": "北方稀土", "market": "A-share", "role": "稀土资源", "roleEn": "稀土资源", "bucket": "core", "href": "/investor/stocks/600111.SH/" }, { "ticker": "000831.SZ", "name": "中国稀土", "market": "A-share", "role": "稀土整合", "roleEn": "稀土整合", "bucket": "core", "href": "/investor/stocks/000831.SZ/" }, { "ticker": "MP", "name": "MP Materials", "market": "US", "role": "美国稀土资源", "roleEn": "美国稀土资源", "bucket": "core", "href": "/investor/stocks/MP/" }, { "ticker": "FCX", "name": "Freeport-McMoRan", "market": "US", "role": "铜矿", "roleEn": "铜矿", "bucket": "core", "href": "/investor/stocks/FCX/" }, { "ticker": "SCCO", "name": "Southern Copper", "market": "US", "role": "铜矿", "roleEn": "铜矿", "bucket": "core", "href": "/investor/stocks/SCCO/" }, { "ticker": "AA", "name": "Alcoa", "market": "US", "role": "铝", "roleEn": "铝", "bucket": "core", "href": "/investor/stocks/AA/" } ], "instrumentTags": [ "600111.SH", "北方稀土", "000831.SZ", "中国稀土", "MP", "MP Materials", "FCX", "Freeport-McMoRan", "SCCO", "Southern Copper", "AA", "Alcoa" ], "reviewCadence": "每日复核", "confirmationSignals": [ "油价和金属缺口是否从成本冲击变成需求破坏", "资源品涨价是否能够传导至下游", "地缘风险是否改变库存和运费假设" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "需求破坏", "库存周期", "运费", "下游利润率" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：证据密度足够", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 89 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 8, "recentEvents": 89, "supportRatio": 0.05, "riskRatio": 0.82, "conflictScore": 92.4, "evidenceDepth": 100 }, "score": 74.3 }, "events": [ { "date": "2026-06-29", "title": "[auto-handoff] 先进制程星载 OBP 芯片良率瓶颈与算力/通信设备材料供应链", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 81384248-8464-45e1-8194-06f6bed14dff 结果摘要: Wrote handoff_response.md (10343 bytes); wrote handoff_response.en.md (10719 bytes) — Th…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 20, "matchedTerms": [ "GOES", "materials", "constraints", "downstream", "orders", "material" ], "href": "" }, { "date": "2026-06-29", "title": "中游房建及家装开工收缩-30.67%通过高经营杠杆深度挤压了上游钢铁、铝加工及基础化工的营业利润率与订单量，导致三大行业面临深幅利润下修压力（加权下修-67.3%），并伴随着2026年首度碳…", "summary": "本报告对中游房建及家装开工收缩对上游原材料的传导进行了量化测算。-30.67%的开工收缩导致钢铁订单下滑10.73%，在螺纹钢售价接近现金成本和15%固定成本高经营杠杆挤压下，行业OPM由8.0%降至3.84%，净利润预测下修-58.0%。建筑铝型材订单下滑12.27%且加工费遭遇无序价格战下跌6.40%，高经营杠杆将行业OPM从10.0%压缩至2.45%…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19.5, "matchedTerms": [ "商品", "钢铁", "materials", "material" ], "href": "" }, { "date": "2026-06-29", "title": "能源与大宗商品资产在宏观政策背离下的防御价值研判", "summary": "能源与基本金属（连同黄金）能充当防御支柱，但价值是分层而非整齐划一的，须按机制而非标签构建。通胀冲击通道真实：通胀每意外+1个百分点，商品+7个百分点、股−3、债−4[S12]，2022年GSCI+34%印证[S13]。黄金是唯一双重（通胀+危机）对冲，央行一季度净买244吨、全年预测700–900吨[S9]，应为核心。铜是顺周期通胀贝塔，对冲通胀尾部但非…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17, "matchedTerms": [ "能源", "商品", "energy" ], "href": "" }, { "date": "2026-06-28", "title": "在高利率长期化但衰退未确认的环境下，应以套息、凸性与实物资产对冲（短久期杠铃、适度美元、能源期权+黄金+能源股）替代方向性利率与油价押注，以改善组合风险回报比。", "summary": "截至2026-06-29，支持卡片01的高利率长期化、衰退未确认判断，并转化为配置：以套息+凸性+实物资产对冲替代方向性押注。久期保持低于基准，杠铃式收割2年期套息（约4.10%、久期约1.9优于10年4.38%、久期约8.2）并保留小段长端作衰退保险；相对低收益货币适度做多美元（DXY约101.4）；以能源看涨期权+黄金（约4,090美元）+能源股替代方…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17, "matchedTerms": [ "能源", "商品", "energy" ], "href": "" }, { "date": "2026-06-28", "title": "能源/商品板块应以多元化商品篮子为核心、能源股为收益锚、少量原油为尾部风险工具，而不是把11%战略通胀对冲变成单一布伦特方向押注。", "summary": "支持Research note 02的11%商品/能源战略配置，但执行上不追逐现货原油。建议总组合中配置6.0% BCOM式多元化商品篮子、3.5%高股息能源股、1.5%流动性原油敞口。布伦特6月26日跌至71.99美元/桶，显示霍尔木兹正常化正在压缩战争溢价；但IEA/EIA仍提示2026年供给下降、库存偏低，因此保留通胀对冲。基准判断为2026年Q3布…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17, "matchedTerms": [ "能源", "商品", "energy" ], "href": "" }, { "date": "2026-06-28", "title": "能源商品成本传导与USD/CNY汇率风险的压力测试", "summary": "截至2026-06-28，本卡对卡3'不上调外汇对冲'结论做压力测试。被测机制（人民币贬值抬高能源进口成本、压低通胀对冲实际人民币收益）真实但属二阶，且当前被相反力量压过：能源对冲人民币收益的主导因素是美元油价中伊朗战争地缘溢价的崩塌，而非汇率。人民币油价成本=布伦特(美元)×USD/CNY，3月峰值至今每桶人民币成本降约41%，其中美元价格腿贡献−40%…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17, "matchedTerms": [ "能源", "商品", "energy" ], "href": "" }, { "date": "2026-06-27", "title": "虽然备忘录缓解了短期实物原油压力，但霍尔木兹海峡的结构性脆弱与15%-25%的二次通胀概率支撑将大宗商品战术超配至20%，并应采用高分红个股现货替代及低配看涨期权策略进行防御。", "summary": "截至2026-06-27，评估表明尽管扫雷进展与伊朗出口正常化将推动短期WTI均价回落至67.95美元且跌破70美元概率升至62.85%，但15-25%的能源价格二次上尾再脉冲风险及长期地缘瓶颈支持将大宗商品配置超配至20%。在A股通道中，因159945等ETF极度缺乏流动性（成交仅106万元），建议以中国海油、中国石油等高分红龙头个股作为替代通道，配合低…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17, "matchedTerms": [ "能源", "商品", "energy" ], "href": "" }, { "date": "2026-06-27", "title": "霍尔木兹海峡重开仅为“执行折价”而非“和平红利”，60天物理排雷延迟导致到岸能源价具极强向下刚性，与高利率共同构筑新能源“资本开支壁垒”，系统性阻碍全球能源转型进程并抬升算力用电门槛。", "summary": "霍尔木兹海峡重开虽缓解了原油期货的即期上涨动能，但由于海峡内活体水雷清障导致的60天排雷延迟，到岸能源价格（CIF）维持强韧的向下刚性，地缘风险溢价只是发生了盘面转移。在高利率与绿色通胀双重作用下，全球新能源面临高额WACC与电网并网漫长排队的“资本开支壁垒”，转型进程阶段性受阻，算力能源供应链成本大幅飙升。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17, "matchedTerms": [ "新能源", "能源", "energy" ], "href": "" }, { "date": "2026-06-27", "title": "Summary of Work We have completed the analysis for Research note 04/10 as the Materials Analyst…", "summary": "Summary of Work We have completed the analysis for Research note 04/10 as the Materials Analyst (材料行业分析师) , adopting a Support (支撑) stance. Our report evaluates the global supply …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 17, "matchedTerms": [ "GOES", "copper", "materials", "supply", "material" ], "href": "" }, { "date": "2026-06-26", "title": "随着全球及中国变电设备产能在2027-2029年逐步释放，AI数据中心的电力设备瓶颈将从“无法获取设备”的二进制物理硬约束，降为“支付溢价与关税以获取交期”的资本支出与利润率成本约束。", "summary": "本报告对2027-2029年AI电力设备瓶颈的缓解速度进行了深入分析。结论表明，随着西方重电巨头（如日立能源、GE Vernova）扩产产能释放，以及以特变电工、思源电气、金盘科技为代表的中国高效率设备出口，2027-2029年电力设备将从物理上的“硬约束”转变为“成本/经济约束”。由于取向电工钢（GOES）供应持续偏紧、“铝代铜”面临技术失效以及地缘关税…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 18, "matchedTerms": [ "GOES", "能源", "industrial" ], "href": "" }, { "date": "2026-06-26", "title": "评估“限定亏损能源”作为尾部商品对冲工具的有效性与风险边界", "summary": "压力测试卡片06的18%商品仓（黄金+限定亏损能源）。核心：能源仅是有条件右尾对冲（霍尔木兹/地缘），非核心多头。基准55-65%为去通缩、2027约5.0mb/d过剩、期限结构转Contango[S1][S3]，裸多头将因负展期失血并破坏对冲。故风险边界=仅做多虚值看涨/牛市价差(B1)、年在险权利金≤1.5%净值(B2)、用期权而非滚动期货(B3)、能…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17, "matchedTerms": [ "能源", "商品", "energy" ], "href": "" }, { "date": "2026-06-26", "title": "电网基础设施与能源转型对AI算力扩张的潜在约束", "summary": "压力测试部分支持并升级卡03：电力/电网瓶颈是“节流而非断路”，但节流比其暗示更紧更久。在2026–2028视野内，电气基础设施已是排在芯片之前的硬约束——美国电力变压器交期128周、GSU 144周、大单3–4年，价格较2019+77%、达2022前4–6倍；GOES由Cleveland-Cliffs近乎独家供应、全球变压器产能利用率约98%、GSU需求…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17, "matchedTerms": [ "GOES", "能源", "energy" ], "href": "" }, { "date": "2026-06-25", "title": "电网设备原材料缺口对新能源项目建设成本与并网进度的传导效应评估", "summary": "支持卡片06–08的逻辑链：AI电力瓶颈已上移至原材料，且GOES/高纯度铜缺口确实传导至新能源经济性，但传导非对称、主要走时间而非单价。直接价格通道有限——铜仅占光伏/陆上风电资本开支约3%~5%，即便铜较2020年+50%、GOES+100%，对LCOE直接抬升仅约+2%~+4% [自测算]。主导通道是交付/并网延迟：在变压器70–144周交付、并网中…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22, "matchedTerms": [ "GOES", "新能源", "能源", "energy" ], "href": "" }, { "date": "2026-06-25", "title": "Summary of Work I have completed the research and stress test on the transmission effects of gr…", "summary": "Summary of Work I have completed the research and stress test on the transmission effects of grid equipment bottlenecks on upstream key raw materials (GOES, high purity copper, an…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 19, "matchedTerms": [ "GOES", "copper", "materials", "bottlenecks", "material" ], "href": "" }, { "date": "2026-06-25", "title": "全球电力设备制造供应链产能与上游材料供应风险评估", "summary": "部分确认Card-03的time-to-power论点，但作三处修正。瓶颈确在上移到材料端，但对变压器而言真正的物理咽喉是GOES而非铜，且GOES稀缺本质是“产能+贸易”稀缺而非地质稀缺：全球变压器产能利用率98%、GOES自2020年涨约100%、美国仅Cleveland-Cliffs一家、俄NLMK被制裁；但吨位上轮机拉动的GSU用钢仅约全球GOES…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 18, "matchedTerms": [ "GOES", "能源", "industrial" ], "href": "" }, { "date": "2026-06-25", "title": "由于发电设备长交期、核电重建缓慢、跨州输电网架建设滞后以及并网容量在地理上的严重错配，电网侧能源增量无法匹配2026-2028年AI负载爆发式增长，电力缺口将实质性压制算力部署进度，并推高终…", "summary": "本卡对电网侧增量电源进行压力测试。2026-2028年，电网侧增量无法有效填补AI电力缺口。重型燃机面临5-7年超长交期；绿地核电停滞且SMR商业化迟于2030年，Palisades与三哩岛等重启仅能满足3.4%-10.8%的新增需求；输电线建设长达6.5-10年以上，且新增新能源地理错配严重（主要在德州与西海岸）。这导致PJM等核心数据中心区域容量短缺超…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17, "matchedTerms": [ "新能源", "能源", "energy" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "AI 基建原材料瓶颈会成为下一层压力，但真正能重定价交付和利润的是 GOES 与铜，镍和贵金属暂时只是二阶变量。", "summary": "本卡压力测试上游材料是否会吞噬 AI 电力设备的防御性利润。结论是：风险真实存在，但集中在 GOES 和铜，而不是所有大宗商品。GOES 因 LPT 中替代性极低、合格产能集中，是最危险的物理交付瓶颈；铜因贯穿数据中心内部配电和外部电网加固，是最重要的价格和营运资本瓶颈。镍主要是印尼集中度和储能/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 2, "value": 43.4, "events": [ { "title": "电子取代 HBM 成为 AI 链条上未来 12 个月最被讨论的瓶颈：1 GW 边际 AI 园区的电费在 2027 年从 $322M/年跳到 $926M/年，足以让 hyperscal…", "summary": "对 Research note 03 的 $1T capex / $7.5–10M/MW 营收密度救援叙事进行压力测试。结论：电力（不是 GPU）成为 2026 下半年起的硬约束；hyperscaler 在 2026–2027 锁定的边际 MW 全成本将达 $95–140/MWh，相对 2024 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电池材料需求模型应抛弃对国内渗透率指标的盲从，转而锚定全球GWh总产量，并警惕PHEV占比上升带来的用锂强度结构性下降。", "summary": "针对4月新能源渗透率突破60%但绝对销量同比下滑19%的矛盾，指出电池材料需求已陷入“分母陷阱”。当前渗透率的上升主要源于燃油车市场更大幅度的萎缩，而非新能源需求的实质增长。由于PHEV占比提升导致的单车锂强度稀释，以及国内零售市场的绝对值收缩，材料需求模型应从“国内渗透率导向”切换为“全球GWh…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "原材料定价权已达到顶点：AI驱动的电网超级周期正迫使云巨头绕过公开市场，通过直接入股或高溢价长协锁定GOES钢和铜的‘物理供应’。", "summary": "AI电网建设正在将GOES取向硅钢和铜从周期性商品重塑为战略性科技资产。2026年市场面临结构性缺口，定价权已从设备商转移至原材料生产商。Cleveland-Cliffs凭借北美垄断和垂直整合成为核心受益者；铜价受AI高密集度（27-47t/MW）驱动，AWS与力拓的直采协议标志着云巨头开始绕过市…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 0, "neutral": 1, "value": 47.8, "events": [ { "title": "制造业复苏由出口和政策性资本开支支撑，但在内需传导完成前，对全球贸易波动的脆弱性正在积聚，战略自主板块是最佳防御。", "summary": "2026年4月中国制造业PMI扩张高度依赖“高技术+装备”出口。高技术制造业(52.2)受全球AI开支驱动，航空航天受C919国产化和卫星经济推动，大规模设备更新政策为内需提供了13.9%的投资增长。然而，复苏呈现“双速”特征，非制造业(49.4)仍处收缩。若外需走弱，消费电子和新能源车等高出口占…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "中国出口通道断裂后的上游材料与商品库存压力", "summary": "压力测试结论：在美国/欧盟/日本需求政策同步迫使Q4制造业去库存的情景下，钢铁（螺纹钢/热卷）与电池级碳酸锂将同步率先断裂，而非依次发生，因为两者均已触及边际现金成本且拥有周度高频库存数据。多晶硅为第二同步断裂品种，铜铝随后滞后4至8周，PX-PTA化工链最后断裂但幅度最深。关键补充：卡片03识别…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "GOES（高等级硅钢）是变压器铁芯的核心材料。", "summary": "GOES（高等级硅钢）是变压器铁芯的核心材料。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 1, "value": 46, "events": [ { "title": "AI物理基础设施周期的赢家是长交期机电OEM（LPT/GIS/HV电缆/航改型燃气轮机/液冷参考架构），真正瓶颈已迁移到取向电工钢、铜杆/母线、大型铸锻件与熟练技工——按订单覆盖度×…", "summary": "截至2026-05-13，本卡支持卡片01–04的'密度胜过足迹'主线：AI CapEx美元从外壳转向机电与电网侧含量。具备结构性订单可见度与定价权的OEM集中于长交期、寡占环节——Hitachi Energy（6501 JP）、Siemens Energy、GE Vernova、Eaton、Sc…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "工业资产端的定价内卷与结构性低收益替代，使得1.35%的净息差底部较传统金融模型预测的更为脆弱。", "summary": "2026-05-13，本卡从工业视角否定净息差筑底逻辑。银行业对“新质生产力”及先进制造的信贷倾斜引发定价内卷，核心军工及技术企业贷款利率已低至2.65%-2.80%（贴现LPR）。此外，机械与新能源行业产能过剩导致的“收益率天花板”及央行专项再贷款工具的结构性替代，使得资产端收益率漏损严重，足以…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-14", "supports": 1, "risks": 0, "neutral": 0, "value": 58.4, "events": [ { "title": "消费服务化不是材料需求消失，而是材料需求从地产和耐用品转向算力、电网、储能和电气化基础设施。", "summary": "截至2026-05-14，材料端支持“服务化 vs. 降级化”的方向，但结论是结构分化：终端商品疲弱压制钢铁、铁矿石和通用化工；服务经济背后的算力、电网、储能和电气化基础设施则支撑铜、铝、锡、特种化学品及部分电池材料。配置不宜宽基超配材料，而应从旧商品材料转向服务基础设施材料。", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "欧盟 CBAM 政策背景下上游钢铝价格波动对中欧新能源车成本不对称性的压力测试", "summary": "截至 2026-05-16，材料通道会缩短但不会推翻 Auto vs. BYD/CATL 配对。关键修正是 CBAM 现金成本时间：Q1-2026 官方 CBAM 价格为 €75.36/tCO2，但 2026 年有效爬坡只有 2.5%，卡片 04 的 €640-1,120/80kWh 包更接近完全…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "IEA《Electricity 2026》指出，全球超过 2,500 GW 的新能源、大负荷与储能项目正卡在电网排队中；到 2030 年，年度电网投资需从目前的约 4,000 亿美元…", "summary": "IEA《Electricity 2026》指出，全球超过 2,500 GW 的新能源、大负荷与储能项目正卡在电网排队中；到 2030 年，年度电网投资需从目前的约 4,000 亿美元提升约 50%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "价格联动机制无法防御极端供应短缺导致的“调价滞后”与“物理断供”，Tier-1 厂商在材料急涨期将面临阶段性毛利压缩与交期违约风险，需配置物理大宗商品（铜、硅钢）进行对冲。", "summary": "针对硅钢与铜供应极限的压力测试表明：Tier-1电网装备龙头的价格联动机制在数学上依然有效，但面临严重的“时间差”与“物理短缺”挑战。3至6个月的调价滞后将在商品急涨期引发阶段性毛利挤压和营运资金剧增；更关键的是，高磁感取向硅钢（HiB GOES）和铜的物理供给硬约束可能导致设备交付延期，使得固定…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 1, "value": 53, "events": [ { "title": "能源价格压力下的工业原材料价格韧性与利润空间分析", "summary": "材料行业分析表明，钢铁与化学品等关键原材料价格走势因中国地产下行、产能过剩以及全球制造业周期分化，已与高企的能源成本严重脱钩。原料上涨带来的成本压力在第一站即被中游深度压缩利润空间（如乙烯亏损运行）所拦截，无法向下游终端转嫁，从而切断了通胀螺旋。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高端装备出口的结构性崛起与战略性资本开支（设备更新、国防及商业航天）筑牢了中国工业增长底座，强韧的经常项目顺差将封杀USD/CNY突破7.30的空间。", "summary": "截至2026-05-22，本卡从工业与先进制造视角强烈质疑Research note 05关于‘中国增长走弱支撑汇率破7.30’的论点。我们认为强劲PMI并非短期抢运，而是高端机械、新能源及国防航天等资本品结构性出海与国内‘设备更新’投资（航天增17.9%，国防1.91万亿）共振的结果。4月机电出…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "全球变压器与取向硅钢供应链地缘博弈与关税壁垒对AI算力扩张的二次限制", "summary": "高阶GOES与EHV变压器产能高度集中于中日韩（GOES全球约80%+，LPT亚洲约55–60%产量）。美国Section 232钢铁关税、USITC GOES反倾销复审、LPT 232调查配额、DOE大电网EO 13920，叠加欧盟CBAM、FSR与NZIA本土化要求，正在Research no…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 1, "risks": 1, "neutral": 0, "value": 60.8, "events": [ { "title": "库存触阈值支持交易性底部，但锂、SiC和存储器证据不支持下周A股汽车板块出现全面利润拐点。", "summary": "截至2026-05-24，本卡从材料端修正前序链路：库存阈值可能带来A股汽车交易性底部，但不支持利润基本面已经拐头。CADA显示2026年4月VIA为62.1%、库存系数1.89，经销商库存约260万辆；5月1-17日新能源零售环比增18%但同比降12%。材料端看，电池级碳酸锂约178000元/吨…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI 物理电网瓶颈与大宗商品物理期权定价", "summary": "AI 物理电网瓶颈与大宗商品物理期权定价 AI Physical Grid Bottlenecks and Physical Option Pricing of Commodities 随着FERC否决核电并机与PJM并网延迟（5-7年）暴露了AI算力的“电力物理墙”，精炼铜与GOES（取向硅钢）…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 8, "neutral": 1, "value": 15.2, "events": [ { "title": "GOES（电磁取向硅钢）与电气用铜的全球供应已极度集中于少数中、日、韩、俄、智利-印尼厂商，美国50%钢铁关税、中国出口许可扩容与NLMK制裁三大地缘冲击共振，将把变压器交付时滞由当…", "summary": "本卡对工业制造分析师卡片06的'GOES紧缺是变压器交期飙至150-200周关键'判断进行逆向压力测试，结论强化而非削弱前序判断。HiB高磁感薄规格GOES全球仅Baowu/Nippon Steel/POSCO/NLMK四家可量产，合计仅~900-1,000kt/yr，美国仅Cleveland-C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "特种材料成本飙升并未导致一线厂商利润结构性崩溃，反而通过“可融资性”过滤器强化了头部议价权，但传导后的高成本正成为新能源装机速度的实质性阻碍。", "summary": "2026年石油冲击导致EVA/POE树脂价格飙升至$2600-$3000/吨，光伏组件和锂电池成本分别增加约3-7%和$3-$7/kWh。尽管成本大涨，一线厂商通过供应联盟、技术壁垒及可融资性优势，成功将成本压力传导至下游，维持了12-15%的毛利率。然而，高昂的组件价格与高利率共振，导致终端电站…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "设备、铜缆与劳动力结构性短缺将电网“配给模式”转化为隐性的实体经济拥堵税，推高项目有效WACC达45-75基点并迫使新能源FID普遍延期。", "summary": "本卡压力测试表明，变压器交期拉长（最长4年）、铜缆高价（$10,500/吨）与劳动力荒（年增15-20%），正推高电网“配给模式”的实施成本。产能槽溢价、非坚定连接下15-20%的弃损率及高额互联保证金，将项目有效WACC推高45-75基点，压低项目IRR并导致FID普遍延期，使能源转型与电气化进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器刚性交期与电网并网严重积压构成AI算力的物理上限，导致上游发电局部弃风光过剩与下游数据中心“有壳无电”的结构性错配，2027年后恐引发大规模空壳数据中心资产减值。", "summary": "截至2026-05-24，本报告应力测试了电力基础设施对AI算力连续性的支撑力。研究表明，大型变压器3-4年交期与电网互联4-10年排队时间构成算力扩张的物理天花板。尽管新能源装机迅猛，但因电网梗阻导致限电率飙升，呈现局部“并网受阻型过剩”；核电直连联产受到FERC严厉监管，SMR属长远规划。若2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 1, "risks": 3, "neutral": 3, "value": 60.6, "events": [ { "title": "2026Q3应持有“利润留存中国”并对冲“重定价中国”：自动化和现金流防御资产增持，新能源硬件、材料beta和科创拥挤动量减持，基准对冲比率0.45x。", "summary": "截至2026-05-25，本卡综合1-7卡后认为：负Gamma平仓真空之后不应简单卖出全部中国权益，而应增持能留住降本和现金流的A/H资产，减持会被客户重定价和流动性冲击伤害的板块。2026Q3建议超配自动化、机器人、电网、公用事业、精选现金流充裕港股互联网；低配光伏组件、商品化电池、锂铜材料be…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "材料闪崩后的利润率红利呈现短周期自动化最强、新能源硬件被传导稀释、航空航天被现金流拖累的阶梯式分化。", "summary": "截至2026-05-25，本卡片综合并收窄卡片05结论：材料金融性闪崩不是制造业利润率的普遍利好。机器人和短周期自动化因BOM调整快、报价周期短，可获得约+0.3至+1.6个百分点的净订单利润率缓冲；新能源硬件因电池与光伏过剩产能和客户重定价，净影响约-1.0至+1.0个百分点；航空航天因长周期b…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "人民币贬值对出口的提振效应正处于“收益递减”阶段，能源及原材料成本的刚性上涨与全球需求疲软正共同侵蚀中低端制造业利润空间。", "summary": "人民币贬值至7.45-7.50区间虽提供名义价格优势，但正面临全球大宗商品价格（CRB指数创历史新高）和能源成本（布油触及138美元）激增带来的输入型通胀挤压。制造业成本结构中原材料占比高达70-75%，进口成本上升抵消了约30-40%的汇率红利。然而，中国凭借“煤炭锚”及高比例可再生能源（202…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "光伏与电池「客户重定价」是结构性+政策性的二轮深挤压，2026 一线组件毛利率压力中枢 3.5%-6.5%，新能源硬件是空头平仓真空过后最先掉队的 cluster。", "summary": "本卡片（2026-05-25）压力测试 card-06 对新能源硬件「-1.0~+1.0 pct」温和稀释的判断。在产能利用率<65%、长协重签价格中枢已落至 0.10-0.11 USD/W、美国 FEOC + 14.25% 关税无法转嫁、欧盟 NZIA 提升非价格权重的多重重定价压力下，一线 T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 7, "neutral": 1, "value": 24, "events": [ { "title": "[auto-handoff] 美伊外交谈判传闻及油价跌破100美元对美联储下半年降息与DXY的边际影响评估", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: d5d2ffa6-c741-4e78-b373-a9292f4c8a66 结果摘要: Wrote handoff_response.md (10595 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the analysis for Research note 4/10 as the Materials Analyst, evaluating t…", "summary": "I have completed the analysis for Research note 4/10 as the Materials Analyst, evaluating the cost transmission channels and gross margin squeeze ris…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美欧原产地规则与中国出口许可在AI电力供应链上呈现70%价格机制、30%数量机制的格局，仅在Hi-B取向硅钢与NdFeB重稀土磁材两节点真正物理改道，其余主要通过推高到岸成本并对美国…", "summary": "美欧'熔炼与浇筑'原产地规则与中国出口许可制是两套高度非对称的工具：前者基于规则、机械执行但被18-30个月执法滞后稀释；后者基于裁量、时延即政策。综合压力测试：2026-2027年这套监管栈是70%的成本推动通胀税（AI电力设备到岸成本+20-35%，美国核心商品CPI+15-25bp）、30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在全球能源供应受限和长期通胀中枢上移的背景下，能源及广义大宗商品是防范长端利率（尤其是期限溢价）上升的最有效直达对冲工具。", "summary": "截至2026年5月27日，本卡支持前序大类资产配置增加大宗商品的结论。当前30年期美债收益率破5%的核心驱动力之一是中东地缘局势导致的能源供应不确定性，这推高了原油风险溢价并锚定了长期通胀预期。在“能源冲击→通胀粘性→美联储受限→长端利率上升”的传导链条下，直接配置能源及广义大宗商品能够形成对长债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 5, "neutral": 0, "value": 34.4, "events": [ { "title": "发改委电价现货化、DRC钴配额收紧与主权收储、以及全国碳市场扩容后的碳价上涨，将130-195bps二阶拖累塑造成长期的结构性多年税，并在先进制造链内部引发了中游低壁垒受损与上游/L…", "summary": "截至2026-05-27，本卡对Research note 08提出的130-195bps制造业二阶成本拖累进行了压力测试。政策内生性评估表明，发改委394/136号文确立的电力现货化交易与新能源上网电价入市彻底终结了分时电价行政性避风港，DRC战略矿产局（ARECOMS）在4月确立的钴出口配额与…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国GOES供应在2028年前结构性持平甚至下降——OEM订单簿收入可兑现但量增受限，CLF从纯上游变成了垂直竞争对手。", "summary": "2026-05-28结论：本卡对第04卡『做多CLF作为上游对冲』的论断进行压力测试。Cleveland-Cliffs在Butler Works的1.7亿美元改造仅带来约25,000吨/年的GOES增量（仅工艺良率提升，非新建产线），且要到2028年才完成；同时CLF自家Weirton新建变压器厂…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] BYD 5月最后一周上险数若证伪共识，A股新能源链传导路径", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 33ecad2a-0ea3-41e1-a340-80c318c7ae35 结果摘要: Wrote handoff_response.md (15552 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 叙事正从芯片竞争迁移到能源供应竞争，而把 2026–2028 钉死的不是发电装机，而是变压器、HV 开关柜与取向硅钢这条狭窄走廊。", "summary": "支持卡01的'物理墙'框架，并把瓶颈具体化到电力设备链：美国大功率变压器交付周期已拉长至120–210周、北美互联排队>2,600GW、取向硅钢(GOES)成为物理瓶颈。GEV/ETN/ENR.DE/Hitachi Energy/POWL 等的订单簿连续6个季度book-to-bill>1.2，订…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 2, "neutral": 0, "value": 53.8, "events": [ { "title": "上游大宗与精细化学品瓶颈的定价权非线性上升，将对缺乏价格联动条款的中游重工OEM在手积压订单造成严重的毛利率反噬，进一步强化做多上游物理门神、做空二线搁浅久期算力的跨资产杠铃配置。", "summary": "本报告以材料分析师视角综合卡片01-04，深入分析最上游大宗商品（GOES特钢、绕组无氧铜、PFAS替代防腐助剂）在2026-2028年的刚性供应约束与定价重估。测算表明，上游材料价格暴涨将推高中游重工与温控CDU成本10%-12.5%，由于缺乏价格联动的历史固定价格在手订单占比高，GEV、西门子…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源价格通胀对电力重工“门神”利润护城河的压力测试 报告日期：2026年5月28日 白板门户共享研究会话 (研究线程标识：已归档| 卡片 7/8) 选定立场：压力测试 (Stress…", "summary": "能源价格通胀对电力重工“门神”利润护城河的压力测试 报告日期：2026年5月28日 白板门户共享研究会话 (研究线程标识：已归档| 卡片 7/8) 选定立场：压力测试 (Stress Test) 核心摘要 截至今日（ 2026年5月28日 ），全球权益市场普遍存在一种乐观共识，即认为全球电力设备巨…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 6, "neutral": 1, "value": 34.2, "events": [ { "title": "运费冲击真正撕咬的是成品段(组件/储能)而非原料段(硅料/锂盐)，由此加速的是'成品本土化'而非'原料重置'——这对中国出海组件商是H2 2026盈利下修核心风险，对上游硅料/锂盐厂…", "summary": "对卡-2'运费激增大幅推高新能源材料成本'结论进行幅度修正与结构压力测试。多晶硅(~6.4 USD/kg)、碳酸锂(~9.4 USD/kg)等高单位价值密度材料运费敏感度仅约1–2%；WCI升至2,800 USD/40ft的真正非线性冲击落在组件(~85%相对运费增幅)、储能DC Block(~7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Overview of Policy and Geopolitical Analysis As the Policy Analyst for Research note 08/08,…", "summary": "Overview of Policy and Geopolitical Analysis As the Policy Analyst for Research note 08/08, I have completed the final, synthesis led analysis of thi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI算力的物理约束正从“有没有电”转向“能否在正确节点、正确时点、用合格设备按时送电”。", "summary": "2026-05-29：本卡支持上一张TMT二阶硬件主线，但加入电力基础设施折扣。AI算力扩张不是中国全国层面的总量缺电问题，而是局部电网容量、变压器/开关设备交付、新能源消纳和高功率密度冷却/备电的时点约束。中国互联网数据服务用电1-4月同比增长44.4%[S2]，但仅占全国同期用电约0.94%[…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "财政扩张能保护新能源的实物资本开支与部分受监管现金流，但不能把高实际利率下的新能源制造beta重新定价为全行业成长久期。", "summary": "截至2026-05-29，本卡片压力测试新能源与电力基础设施在高实际利率和财政扩张下的回报韧性。结论是财政补贴资本开支能托住装机量、设备订单和部分电网/储能/特高压现金流，但无法全面抵消新能源全产业链估值压力。核心分化在于：电网设备、独立储能容量电价和高质量运营商具备防御性；光伏制造、锂电材料和依…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 5, "neutral": 0, "value": 38.6, "events": [ { "title": "AI 容量错位主线在材料端是自洽的——铜、GOES、锂、PFAS 替代冷却液、镍/不锈钢这五条投入要素本身就处于供给上限，正是它们的垂直供给曲线在物理上强制了 2026 的配额年，因…", "summary": "材料分析师从钢铁、化工、有色、电池材料四条投入要素端综合前五张卡的论断：2026 是 AI 算力的配额年而非顶点年。每 100 MW 净 IT 负载需 2.0 kt 铜、0.6 kt 取向硅钢 (GOES)、4.5 kt 结构钢、25 t LCE（BESS 用）及 30–50 t 介电液，五个市场…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "利用中外利率差融资红利并构建本地化出海2.0供应链，是先进制造企业突破高利率及海运通胀双重硬约束、维持海外30%以上高毛利率及盈利韧性的关键免疫盾牌。", "summary": "截至2026年5月30日，本卡片强烈支持卡片3与4。在实际利率常态化（美债10年期收益率4.45%）及海运通胀（SCFI 2572点）双重背景下，“本地化出海2.0”模式龙头企业展现出极强盈利韧性。通过中外利率差（中国1年期LPR 3.00%）套利及本地化壁垒，三一重工与中联重科等重工龙头海外毛利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两部制电价虽确立了火电 IPP 的盈利底线，但刚性资本开支与现货折价将防御性篮子的股息率上限压制在 4.28%，使得 3.5 万亿避险轮动的安全边际极度脆弱。", "summary": "本卡片评估了两部制电价机制下火电 IPP 的盈利稳定性及其对防御性篮子的影响。结论显示，2026 年 50% 固定成本回收的容量电价虽提供了盈利底座，但无法抵消现货市场 15%-25% 的折价挤压及灵活性改造与新能源扩张带来的刚性资本开支。火电 IPP 呈现“有利润无现金流”特征，拖累防御性篮子股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "油价下降是欧洲化工和钢铁利润率的短期缓冲，而不是足以重置全球竞争力的结构性解药。", "summary": "截至 2026-05-31，本卡压力测试认为：油价下行能改善欧洲能源密集产业利润率，但只能部分对冲供应链重构和要素成本上升。Brent 从 $91.87/bbl 到 $82.5/bbl 中点相当于 -10.2%，但油和石油产品仅占 EU 工业终端能源 10.4%，油价完整传导也只约降低直接能源篮子…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 1, "risks": 7, "neutral": 2, "value": 42, "events": [ { "title": "在大宗商品暴涨冲击下，具备订单长协套保与高通胀转嫁能力的头部变压器厂商能够毫发无损地维持结构性高盈利，而地缘硅钢成本撕裂与3.61铜铝比率则加剧了中低压配网领域的铝替代与中国厂商的出…", "summary": "本卡片对取向硅钢（GOES）产能缺口及电解铜（$13,615/吨）与铝（$3,770/吨）高位震荡对变压器OEM利润率的影响进行了定量压力测试。评估表明，无套保的固定售价合同将面临严重的利润率清零危机（情景A）；但具备极高定价权的头部厂商（如伊顿、西门子能源）依靠80%铜金融套保与80%材料成本联…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "TMT 去杠杆向光伏与电池扩散的条件已经成熟，6 月首周（6/2–6/6）若光伏设备+电池融资余额连续两日下降且龙头单日净卖出 ≥10 亿元，则板块级去杠杆事件成立；基本面无法承接抛…", "summary": "本卡为能源行业卡片，支持卡片 03 关于 TMT 去杠杆向新能源链扩散的判断。截至 2026-05-31，光伏设备 + 电池板块两融余额合计 1520.21 亿元（其中电池 962.42、光伏设备 557.79）[自测算][S1]，占卡片 03 所述相邻篮子的 38.5%。新能源链与 TMT 在四…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "“算电协同”下的资产配置重构：电网投资重心向配电与数字化转移对2026 H2跨资产权重的驱动", "summary": "把 Research note 07–09 的“算电协同”主线翻译为跨资产权重：A 股内能源由低配升至标配偏多 (+150 bp)、公用事业升至超配 (+250 bp)、TMT 由超配降至标配偏多 (−250 bp)；跨资产层面信用债+200 bp、商品+100 bp、港股+100 bp、海外股票…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI Capex 变现物理墙与 TMT 向 Utilities 的大宗风格轮动交易", "summary": "AI Capex 变现物理墙与 TMT 向 Utilities 的大宗风格轮动交易 云巨头是否正在进行从‘算力硬件’向‘电网容量与基础能源’的Capex结构性重组？我们应如何利用这一二阶实物瓶颈进行多资产的跨行业做多（Utilities/GOES）与高拥挤做空（TMT/SaaS）轮动配置？ Pri…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 6, "neutral": 0, "value": 35.4, "events": [ { "title": "制造业去库存对上游材料的传导是三段式分化——黑色和油制化工双杀,但铜、铝及POE/EVA新材料因电网、新能源、AI算力驱动呈反PMI韧性,材料板块应内部加权倾斜有色而非一刀切低配。", "summary": "本卡承接卡02 PMI跌至50.0、产成品库存升至49.3的被动累库信号,以及卡06金融端不良上行的传导路径,对上游材料行业做穿透式压力测试。核心结论:制造业去库存对上游传导呈现高度分化的三段式结构——黑色(普钢吨毛利测算为负、47.6%钢厂亏损)与油制大宗化工(PTA加工费仅285元/吨已破现金…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027年国四柴油重卡淘汰实为结构性去产能，在NEV与LNG重卡的高效补位下，不会引发运力收缩，反而巩固了餐饮供应链的长效物流降本红利。", "summary": "2026-06-01判断：2027年柴油重卡淘汰不会引发运力短缺，物流成本冗余不会被击穿。当前货运市场“车多货少”，国四重卡存量约50万辆（占保有量6%-8%），其淘汰实为良性去产能。同时，2027年新能源重卡预期销量达40-50万辆（渗透率>50%），叠加高经济性LNG干线运力，补位完全充足。卡…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球能源价格暴涨通过无上限煤电价格联动机制向下游高能耗重工业转嫁，导致重工业利息保障倍数全面穿透，直接通过信贷链条向银行体系回灌巨额不良与拨备压力，威胁中小银行资本安全。", "summary": "2026年6月1日，能源行业分析师指出，中东地缘政治引发原油暴涨至105美元/桶，导致国内电价与煤价大涨，击穿钢铁、化工、有色、建材四大能耗行业毛利率及利息保障倍数（加权平均ICR崩塌至-2.95x）。这直接引发16.3万亿信贷链条恶化，加权不良率激增202 bps至3.13%，生成3286亿不良…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新能源间歇性是真实但非唯一约束；具备并网能力公司的扩张核心瓶颈是可调度容量、输电通道、稳定供应合约三者同时出清，重估溢价应锁定同时持有这三类资源的运营商。", "summary": "压力测试结论：新能源间歇性和储能缺口是真实约束，但不是唯一核心瓶颈。AI 算力供给的真正瓶颈是三类资源同时出清——可调度容量（煤、气、核、水电、抽蓄）、并网与输电通道、以及可让超大规模客户签约的稳定合约结构。中国新型储能 73.76 GW / 168 GWh、平均时长仅约 2.3 小时，无法覆盖 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 0, "value": 60.8, "events": [ { "title": "卡片04的长周期订单证据成立，但HALEU瓶颈、FERC表后否决、GOES单源、Section 232关税与中国稀土域外管辖共同抬高政策β，溢价定价权叙事应被降级而非剔除。", "summary": "截至2026-06-02，本卡对卡片04的长周期支持结论进行压力测试，识别五条具体政策/地缘风险：(1) HALEU燃料瓶颈——Centrus示范级月均约45kg、商业产能2027-2028年才到位，使FOAK SMR大概率延期12-24个月；(2) NRC许可摇摆，EO 14300压缩至18个月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-03，AI–电力链的确定性溢价来自“在受约束节点交付可靠一度电的权利”，而不是泛能源需求增长本身。", "summary": "截至2026-06-03（Asia/Singapore），本卡支持上一张卡片的“重资产基础设施算力”主线，但把可投资确定性收窄到电网稀缺。AI 数据中心不是简单制造全球能源短缺，而是在 PJM、ERCOT、中国东部负荷中心等节点制造可靠电力、并网、变压器、输电和成本分摊约束。全球数据中心用电由 2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 2, "neutral": 0, "value": 62.2, "events": [ { "title": "电网容量与电力设备交付的物理刚性约束已成为 AI 算力 Capex 转化为产能的核心瓶颈，定价权与超额利润将沿‘芯片→变压器/燃气轮机/核电 PPA→GOES 钢’向下游电力基础设施…", "summary": "本报告从能源基础设施物理层支持 TMT 卡观察——AI 算力订单的收入转化率被‘平滑化’，根因是电网容量、大型电力变压器 (LPT) 与重型燃气轮机的结构性刚性瓶颈。截至 2026-06-03，美国数据中心 2030 年负荷展望升至 580 TWh，并网队列存量 2,600 GW，中位等待 53 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源不是衰退中的纯防御资产，但在2026供给短缺和油气现金流韧性下，大幅、笼统减能源比增长冲击本身更激进。", "summary": "截至2026-06-03，本卡压力测试 Research note 06 的大幅减能源建议：结论是方向上可削减商品 beta，但将能源在商品篮中从35%砍到18%过于激进。EIA/IEA显示2026年仍处供给短缺与高油价环境，Brent 4Q26仍预计$89/b；综合油气公司现金流、低杠杆、低再投…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 6, "neutral": 1, "value": 42.6, "events": [ { "title": "公用事业链最值得研究的不是大市值防御龙头，而是能在电价市场化、调节能力和地方回款约束中证明现金流质量的区域运营商。", "summary": "截至2026-06-05，公用事业链的核心不是防御属性，而是电力市场化、现金回款和碳市场扩围的交叉。优先深挖603693.SH江苏新能、600483.SH福能股份、600956.SH/00956.HK新天绿色能源，并把601158.SH重庆水务与000885.SZ城发环境作为现金质量观察名单。主要…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "漆包电磁线等废铜（2号废铜）必须二次提纯的冶炼瓶颈，与铝替代面临的体积和GOES消耗连带成本，共同将绕组端变压器交付周期推升至最高127.5周，彻底击穿了无保障可再生能源项目的并网确…", "summary": "本报告深入探讨了变压器核心金属（铜、铝）的供应侧瓶颈。2026年全球精炼铜面临33万至60万吨的结构性赤字。高压变压器电磁线所需的99.99%+高纯度阴极铜受限于二次冶炼和电解精炼产能的28万吨上限瓶颈。敏感性模型显示，在高电网需求（5800 GVA）下，二次精炼缺口达12.6万吨，导致绕组端变压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "铜矿供应极限（2026年基准TC/RC为$0/吨）与取向电工钢（GOES）因技术及环保政策面临的产能刚性约束，构成了无法通过美联储货币政策消解的物理硬限制，将长期支撑大宗商品价格并固…", "summary": "截至2026-06-05，材料分析师支持先前研判。铜精矿年度基准加工费暴跌至$0/吨，矿端短缺极限导致精铜增量无法覆盖AI需求。取向电工钢（GOES）产能因高硅冷轧脆性、高温退火技术及中欧环保碳关税限制而扩张受阻，导致变压器交期拉长至3-5年。这种供给侧物理硬约束将固化结构性通胀，限制美联储通过大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "实体能源数据否定'制造业强周期'解读、并揭示通胀粘性部分是可消退的霍尔木兹油价溢价，因此抗滞胀杠铃应把能源仓位由原油beta轮向电力/电网/气转电、原油仅作霍尔木兹对冲。", "summary": "诚实指出 card-03 主题为污染的 agent 草稿，遂压力测试 Research note 1–2 的真实主线。能源视角结论：部分支持、重大修正。(A) 柴油货运需求疲弱、2026 工业用电仅+1.0%、Henry Hub约$3，实体量否定'强周期'，但印证 card1 就业48.6 的告诫…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 1, "risks": 0, "neutral": 1, "value": 88, "events": [ { "title": "A股红利应保留，但核心是有行业上限的质量低波红利，而不是无差别追逐最高股息率。", "summary": "2026-06-05，本卡支持保留35%-40%A股红利袖套，但建议从宽泛高股息转向“质量红利+低波动+行业上限”。核心配置为受监管现金流公用事业/电信、大型国资金融、受限能源央企，并用交运、消费必需品和医药红利资产稀释银行与能源拥挤。应回避中小银行票息拥挤、能源商品单因子化、钢铁水泥地产链收益率…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "磷矿资源的战略稀缺性与新能源/精细化工的边际需求爆发，共同驱动具有矿化一体化优势的低覆盖度龙头迎来估值重塑。", "summary": "截至2026年6月6日，磷化工产业链正加速向新能源与精细化工转型。因环保限产，磷矿供给紧缺，价格预计维持在1000元/吨以上。同时，储能爆发将新能源对磷矿的需求占比推升至约6%。建议重点关注具备矿化一体化优势的川恒股份、川发龙蟒、新洋丰及兴发集团，并警惕上游硫磺价格波动的跨链成本冲击。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 5, "neutral": 0, "value": 56.4, "events": [ { "title": "电力变压器核心原材料（取向硅钢及铜）的物理供给短缺与大宗商品价格波动是导致设备交期延长（2.5至4年）的深层根源，其引发的‘铜-铝替代悖论’反向加剧了硅钢缺口，并通过合同条款重置与建…", "summary": "电力变压器核心原材料（取向硅钢/GOES、铜）的供给偏紧与价格暴涨，对交付周期构成深层挤压。2026年Q3执行的DOE Level 2能效标准导致美国本土GOES缺口扩大至19.6万吨，进口依赖度升至57.6%。+60%铜价冲击会导致变压器制造商毛利率降至9.25%并促使其重塑合同条款。铜-铝替代…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在对冲成本高企的背景下，跨资产再配置方案应重点聚焦高流动性且受AI/基建双驱动的精炼铜和避险黄金，并彻底排除缺乏流动性且政策多变的电池金属与过剩钢铁。", "summary": "本卡从材料分析师角度综合并精细化了跨资产再配置方案（策略3）。报告指出，原策略对科技疯涨情境下大宗商品+4%的回报预测过于保守，若由AI和电网基建驱动，33万吨铜供给缺口将引发挤空，铜价或重回14,500美元/吨高点，大宗商品单月可获+10%回报。建议由于流动性和政策风险彻底剔除电池材料（锂、钴、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AIDC \"表后发电\"(BTM)经济性压力测试：LCOE 上升能否再次压制 ROIC？ 工作日期：2026 06 07（Asia/Singapore） 分析师：能源行业分析师（ene…", "summary": "AIDC \"表后发电\"(BTM)经济性压力测试：LCOE 上升能否再次压制 ROIC？ 工作日期：2026 06 07（Asia/Singapore） 分析师：能源行业分析师（energy analyst） 立场：stress test（对 Research note 01 \"BTM 可将 ROI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "监管政策正在通过优化国资委资产负债率考核口径与提供人民银行结构性低息再贷款支持，积极化解电力央企的再融资与绿色转型风险。", "summary": "本卡从政策分析视角评估电力央企债端脆弱性。国资委通过“一企一策”考核和绿色债券带息负债“比例剔除”机制优化资产负债率考核口径，有效缓解永续债赎回压力；人民银行2026年一季度优化的8000亿元碳减排支持工具（提供1.25%低息再贷款）为央企提供低成本信贷，对冲了理财赎回等引发的债券利差走阔风险。监…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 1, "risks": 3, "neutral": 0, "value": 69.8, "events": [ { "title": "[auto-handoff] China economic data and EM beta validation", "summary": "致 大类资产配置师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: e947fbc4-a808-440f-a64e-4d2fd375a137 结果摘要: Wrote handoff_response.md (10669 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网基础设施的物理交付周期（3-4年）构成了 AI 算力部署的刚性硬约束，其影响已超过电费成本，成为决定高能效集群落地的第二物理瓶颈。", "summary": "截至 2026-06-07，电网基础设施（尤其是大型电力变压器）的交付周期已拉长至 160 周以上，成为制约 AI 算力落地的核心物理瓶颈。尽管芯片能效提升降低了电费支出，但工业产能利用率高达 98% 且关键材料取向硅钢 (GOES) 持续短缺，导致 11 GW 的美国数据中心容量因硬件缺失而停滞…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高技术与装备PMI的扩张已在Q2实物订单数据中得到物理验证，订单结构集中于半导体设备、新能源车产线与人形机器人零部件三条主线，本质为硬科技资本开支驱动下的结构性、非全面式装备景气。", "summary": "高技术制造业PMI 52.9%、装备制造业PMI 52.1%的扩张已通过设备类新订单—通用自动化—专用机床链条传导至下游。NBS数据显示2026年1–4月工业机器人产量同比+28.9%、4月+25.3%，金属切削机床+10.4%，远超整体制造业；半导体设备龙头北方华创、中微、拓荆Q1合同负债同比+…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "评估电力市场化改革与能源价格波动对新能源及传统能源利润率的影响", "summary": "压力测试并部分否定卡片06交来的'新能源对冲火电'命题。136号文（2025-02-09）使新能源全面入市、取消固定电价地板，山东集中式光伏2025出清仅0.225元/千瓦时；同期114号文（2026-01-30）将火电固定成本容量回收比例升至≥50%、约+1-2分/千瓦时，给火电托底。结果是两条…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 2, "neutral": 1, "value": 72.2, "events": [ { "title": "卡片05把2024年的“成本推动挤压中游”叙事套用到已经转向的2026年现实：当前PPI再通胀与工业利润回升主要来自反内卷供给侧改革而非油价，油价被煤化工对冲、弱人民币利好出口，故“…", "summary": "从材料行业视角否定卡片05的核心论点。卡片05称高油价+强势美元的“能源税”正挤压中下游制造业、应以“紧货币宽财政”应对，但实证相反：2026年4月PPI已转正至+2.8%、一季度规上工业利润同比+15.5%（高技术制造业+47.4%），是定价权恢复而非被挤压。本轮再通胀主要由“反内卷”供给侧产能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 年电力设备交付瓶颈无法被产能扩张化解，高昂的积压订单饱和度与供应链约束将导致 AI 基础设施在 2027 年面临严重的物理交付墙。", "summary": "压力测试显示，GE Vernova 与西门子能源的扩产计划无法化解 2027 年交付瓶颈。尽管两家公司正在美国和欧洲大幅扩产，但其产能早已被 2024-2025 年签署的创纪录积压订单（GEV $150B，西门子 €154B）预占。燃机和变压器的交付周期分别维持在 5-7 年和 4-5 年的历史高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在 Brent 原油维持在 95 美元以上与强势美元背景下，利润分配严重向具备高价格弹性和美元汇率正暴露的上游开采倾斜，而下游炼化板块则因发改委成品油定价扣减机制与进口汇兑损失而承压…", "summary": "基于 2026-06-08 权威工作日，评估 Brent 油价维持 95 美元以上预期下油气产业链的利润分配弹性及强势美元的影响。分析表明，上游开采具有高正弹性，中海油一季度成本仅 28.41 美元/桶 [S5]，在 95 美元油价下可斩获 66.59 美元/桶暴利 [自测算]，且直接受益于美元走…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 1, "neutral": 1, "value": 79.6, "events": [ { "title": "绿色通胀与电网物理约束正诱发新能源“平价红利”的阶段性逆转，AI基荷需求的高标准使得绿色电力在当前基建瓶境下变得比传统能源更昂贵，从而延缓全球能源转型进程。", "summary": "截至2026-06-13，压力测试显示，尽管新能源理论LCOE仍具竞争性，但受电网升级溢价、变压器交期溢价及FEOC合规成本冲击，美国新能源PPA价格已推升至$64.5-$79.4/MWh。在加入储能以满足AI算力24/7基荷需求后，绿色电力总成本已升至$110-$125/MWh，逆转了相对于天然…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新能源装机的高配速与低可信容量的结构性错配，决定了数据中心负荷扩张实质上是区域可调度装机（火/气/核/抽蓄）的利用率与现货高溢价故事，湖北能源与福能股份为核心受益标的。", "summary": "截至2026-06-13，本卡对2030年前中国发电与储能供给侧匹配数据中心负荷进行了系统性合成分析。每年超3亿千瓦的新能源装机因可信容量极低（夏季高峰仅5-10%），无法直接承接数据中心24小时平稳负荷。供需缺口将集中于东部算力枢纽（广东、浙江、江苏）和水电省份的枯水期（四川）。这显著提升了湖北…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 4, "neutral": 0, "value": 60, "events": [ { "title": "算力西迁不会造成变压器/GIS全国性短缺，但会把订单兑现风险集中到2026-2027年合格排产、试验和并网验收窗口。", "summary": "截至2026-06-13，本卡压力测试算力西迁对变压器、GIS和上游材料的冲击。结论是：0.32-0.63GW基准西迁负荷不足以触及中国全国性产能上限，但会叠加国网4万亿元“十五五”投资、南网1800亿元2026年投资、新能源并网和646亿元变压器出口，挤占2026-2027年合格交付窗口。取向硅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国绿电装机增速够用，但电网扩容（特高压+主变+构网储能）是承接AI算力负荷的真正瓶颈，也是2026–2027年A股能源链最确定的盈利兑现环节。", "summary": "截至2026-06-13，中国绿电装机端继续超预期（截至2026-04风光累计>14.5亿千瓦，2025全年新增约3.5亿千瓦[S5]），但弃风率5.1%、弃光率4.6%[S6]显示消纳与跨省外送是真正瓶颈。IEA口径下全球数据中心用电从2024年415 TWh升至2030年945 TWh[S1]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源价格的韧性源于结构性的供给侧冲击，它完全压倒了区域银行信贷紧缩带来的需求拖累，牢牢锁定了滞胀制度中的通胀黏性，并验证了超配大宗商品与黄金的合理性。", "summary": "能源市场数据证实，当前的通胀韧性是由地缘政治（伊朗冲突/霍尔木兹海峡）引发的供给侧冲击所主导，其影响远超区域银行信贷收紧带来的边际需求破坏。这在基本面上夯实了“轻度滞胀”的逻辑基石，完全验证了策略师关于战术性超配大宗商品、能源股和黄金的资产配置处方。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI大负荷的24/7特性与并网队列中新能源主导的增量存在根本错配，新建基荷电源的长周期迫使云厂商抢夺存量电力资产，加剧公用电网缺口。", "summary": "截至2026-06-14，本卡片压力测试结论：发电增量无法匹配AI基建需求。AI需要24/7稳定基荷，但目前美国并网队列中95%以上为间歇性新能源。核电（含SMR）在2030年前无实质增量，天然气扩建受制于管道许可。这迫使云厂商抢夺存量基荷（如核电背靠背PPA），剥夺了公共电网资源。结论：电力供应…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 6, "neutral": 0, "value": 49.4, "events": [ { "title": "在3.90%实际R*压力情景下，能源与资源是必要但不充分的A股防御对冲，应精选上游油气、一体化煤炭和有色资源龙头，并将仓位控制在12%-15%。", "summary": "截至2026-06-15，本卡压力测试认为能源与资源可以承担A股防御性对冲的一部分，但不是无约束核心。上游油气、一体化煤炭和有色资源龙头具备更短现金流久期和商品价格传导，能在3.90%实际R*环境下优于长久期主题资产；但中证800能源年初以来已上涨16.37%，煤价存在570-770元/吨政策区间…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年海洋经济战略向深远海与硬科技倾斜，低关注度下的深海海工装备与海底算力基础设施蕴含显著的阿尔法机会。", "summary": "基于2026-06-14的市场数据，海洋经济产业链（30日激活率仅5.4%）正从近海向深远海装备及数智化转型。报告梳理了亚星锚链（深海系泊）、亨通光电（海底算力）、中集集团（新能源海工）及东方海洋（海洋医药）等4只低覆盖度核心标的，并提示了上游大宗商品涨价及技术降本不及预期的跨链溢出风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器原材料瓶颈（取向电工钢和铜）不会导致GPU需求破坏，而是导致变压器价格飙升及严重的闲置算力折旧损失，迫使科技巨头扭曲资本开支，大举向电力能源基础设施渗透。", "summary": "本报告研究了2026年取向电工钢（GOES）和铜的供应约束对电力设备变压器OEM利润率及AI算力交付的影响。结果显示，虽然原材料价格上涨（重度冲击下GOES+40%，铜+30%）会大幅压缩固定价格合同利润率，但OEM凭借极强定价权通过指数联动将成本完全转嫁给买方。AI算力未出现需求破坏，而是表现为…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的扩张瓶颈已从芯片转向物理电力设备，燃机与变压器OEM凭借5年以上的积压订单和极强的定价权成为供应链中最大的赢家。", "summary": "本报告深入分析了2026年AI电力供应链中的物理瓶颈。研究发现，燃气轮机和高压变压器OEM已成为供应链的绝对咽喉，GE Vernova和西门子能源的燃机订单已排至2030年以后，大型变压器交期突破150周。瓶颈核心在于取向电工钢（GOES）的结构性短缺，导致2026年全美近50%拟建数据中心产能面…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 0, "value": 74.8, "events": [ { "title": "能源扰动已从油价冲击升级为跨商品通胀冲击，因此商品超配仍是higher-for-longer环境下的核心防御仓位。", "summary": "截至2026-06-15，本卡片支持上一张卡片的商品超配结论。能源冲击通过五条路径传导：原油和成品油推高CPI，柴油和汽油推高PPI，霍尔木兹风险外溢到LNG和电力，油气成本进入化肥和金属，OPEC+谨慎增产延长库存稀缺。即使6月15日和平消息压低油价，库存和政策证据仍显示通胀粘性未解除。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI-to-Power没有被供应侧证伪，但应从“发电扩张交易”收敛为“可交付电力瓶颈交易”。", "summary": "截至2026-06-16，本卡压力测试AI-to-Power的物理供应基础：方向成立，但稀缺性必须从“新增发电”改为“可调度、可并网、可合同化、可交付的节点电力”。IEA预计数据中心用电2025-2030年从485 TWh升至950 TWh；美国增量约240 TWh。核电质量最高但近端主要是800…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 4, "neutral": 0, "value": 64.2, "events": [ { "title": "能源大宗商品作为现金抵押对冲工具的流动性与保证金压力评估", "summary": "本卡支持第07卡的抵押品优先执行规则，并补上其唯一缺口：能源商品配置是缓冲垫还是第二个抵押品抽水口。结论是能源为链条中保证金顺周期性最强的节点——美债经折扣率单向抽水，能源则经VM+IM双腿、按日盯市抽水现金。2022年欧洲能源危机IM约翻倍、数亿美元盘中追保连续数周为据[S4][S5]；2026…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI驱动的电力需求正导致从电网依赖向“自带电源”转型，天然气与核能成为核心受益基荷，瓶颈已从算力迁移至电力交付。", "summary": "截至2026-06-16，AI电力需求激增已导致能源供应进入“能源先行”模式。由于电网并网延迟（20-40个月），超大规模服务商正通过部署场内天然气发电（约100 GW）和签署长期核能协议（如Meta 6.6 GW）来绕过瓶颈。天然气作为桥接燃料的地位稳固，发电用气量创下92.1 Bcf/d的新高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "评估西部'绿区'新能源产能与特高压扩容在承接AI算力迁移中的电力保障能力与成本约束研究", "summary": "压力测试Research note 03的'西迁'判断：方向正确，但用铭牌GW与特高压槽位衡量绿区承接力是量错了对象。真正约束是可靠(7×24可调度)容量与日内/季节匹配。内蒙古新能源150 GW(+35%)、四川水电破1亿千瓦,但风光容量系数仅~17–25%、四川枯水期(11–5月)塌缩。自测算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球变压器产能缺口与铜矿端约束足以支撑选择性的能源/材料利润率扩张，但组合应聚焦出口型、价格可传导的电网设备和铜矿，公用事业仅作为现金流防御而非高弹性增长仓位。", "summary": "截至2026-06-17，本卡支持但收窄策略师的轮动判断：变压器、输电设备和铜矿端约束足以支撑能源/材料板块中的稀缺资产获得利润率扩张和估值防御，但不支持所有电网设备、铝或公用事业无差别重估。出口型变压器和高压设备因交付周期、backlog和价格传导最清晰；铜矿因矿端增量慢、单位利润弹性高，优于铝…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 2, "neutral": 0, "value": 77.6, "events": [ { "title": "纯水电及新能源因干旱与电价侵蚀面临DPS下行风险，防守端配置应从纯水电转向水电-煤炭一体化（中国神华）杠铃组合以稳定DPS收益。", "summary": "截至2026-06-18，本卡片对电力红利资产在煤价、电价及水情波动下的DPS稳定性进行了压力测试。结果表明：纯火电及纯新能源分红极易受燃料成本与电价折价挤压而受损；纯水电虽防范了成本风险，但易在干旱年份因高运营杠杆导致DPS下挫。唯有煤电一体化的中国神华防守性极佳。我们建议配置策略应由纯水电切换…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国低碳铝扩张的核心制约正在从冶炼产能转向稳定、可追溯、可审计的低碳电力交付能力，电网消纳和绿电PPA稀缺会决定绿色溢价能否兑现。", "summary": "截至2026-06-17，本卡支持前序低碳金属绿色溢价逻辑，但将核心约束从“是否有足够新能源装机”细化为“能否向冶炼负荷稳定交付、认证和审计低碳电力”。中国2026年4月底风光装机已达19.1亿千瓦，但电解铝约13.2 MWh/吨的电耗使省级电网因子、绿电PPA、储能和跨省消纳成为低碳铝规模化的关…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 1, "neutral": 0, "value": 85, "events": [ { "title": "由于沿海核电机组投产时空错配、电网火电备份面临二氧化碳排放双控红线及容量电价财务重负、西部新能源大基地受特高压输电负载极限约束，2025-2026年电源侧建设增量无法有效填补绿电波动…", "summary": "对2025-2026年电源侧对智算中心的支撑力进行压力测试。尽管火电、核电与新能源装机在2025-2026年大规模增长，但受限于核电长建设周期与并网安全限制、火电碳排放双控考核及容量/辅助电价高成本、西部绿电特高压通道82%-88%负载极限及火绿捆绑物理瓶颈，电源侧增量无法填补绿电波动缺口。预计实…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 2, "neutral": 2, "value": 86.4, "events": [ { "title": "常规电网并网延迟与新能源项目财务崩溃，迫使云厂商将资本开支刚性移向自备微电网与高密液冷配套，配套设施开支上升72.22%对纯算力硬件产生挤出效应，但也实质上支撑并延长了AI物理层设备…", "summary": "截至2026-06-20，本卡片评估了能源供应与并网效率对AI物理层投资周期的支撑性。常规电网3-5年的并网延迟使100MW数据中心贴现NPV折损达51%至80%，迫使云厂商向“表后直供+储能”的微电网转型。虽然配建100MW/400MWh储能使LCOE上升14%至$55.10/MWh，但每小时C…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] AIDC电力并网瓶颈对电力敏感型化工品（如烧碱、黄磷）供给端硬约束的交叉验证", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: 6a16e82a-5277-47a4-bf61-68e0ff5c0f23 结果摘要: Wrote handoff_response.md (11068 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "对立路径是伪命题——只有'存量水核长协+CCGT桥梁+风光短储+少量LDES'的混合栈能在2026-2030窗口与新能源+短储PPA形成成本-可靠性双向占优，且仅限水核枢纽地理。", "summary": "截至2026-06-19，水电、核电、燃气+CCUS、长时储能任何单一栈都无法同时在成本与24/7 CFE匹配率上跑赢风光+4h锂电PPA：水电资源已开发60%+且枯水风险高，核电2030年前增量仅30-40GW不足以匹配AI 60-100GW新增负荷，燃气+CCUS全成本0.75-0.95元/k…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AIDC电力基建逻辑下高压变压器与输电网铜需求增量的定量折算", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b7159f5c-eb1d-48e7-8538-01f5bfef96a5 结果摘要: Wrote handoff_response.md (10534 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 2, "neutral": 0, "value": 81.8, "events": [ { "title": "全国能源供给能覆盖AI用电总量，但2026-2027年的瓶颈仍是本地并网、特高压投运节奏与可靠容量定价。", "summary": "截至2026-06-20，本卡片压力测试能源侧能否破解AI数据中心并网瓶颈。结论是：全国电量与新能源扩张足以覆盖AI负荷的总量增量，但特高压、储能和本地变电容量的投放节奏仍不足以完全解决2026-2027年大型集群的节点型并网约束。绿电比例提升本身不必然显著抬高算力成本，单独绿证按2026年4月价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI用电争夺会放大局部绿电消纳压力，但中国更可能经历区域利用率下滑和转型节奏分化，而不是全国能源结构转型被迫倒退。", "summary": "截至2026-06-20，本卡压力测试AI集群争夺变压器、开关柜和配网接入窗口对中国绿电利用率的影响。结论是：全国能源转型不会因设备交付延迟被迫逆转，但区域和时点风险已经升高。2026年一季度风电利用率91.4%、光伏90.6%，已贴近90%政策底线；按利用率反推，当季隐含弃风弃光约573亿千瓦时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 4, "neutral": 0, "value": 71.2, "events": [ { "title": "中国高牌号GOES、电磁线和非晶材料供应可缓释变压器瓶颈并利好部分A股材料供应商，但全球瓶颈仍由认证、关税、整机制造和测试能力共同决定。", "summary": "截至2026-06-21，本卡支持“AI算力瓶颈转向电力硬件”的主线，但把材料端结论细化为：中国GOES和铜的总量足以缓解中国/亚洲及第三国变压器链条，真正约束在高牌号认证、关税路径、整机厂产线和测试台位。A股受益顺序偏向望变电气、宝钢股份、首钢股份、华菱钢铁、金杯电工、精达股份、云路股份；江西铜…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国先进制造业低能耗转型与传统基础原材料（钢、化、铝）需求的永久性收缩，共计削减年化2.006亿吨标煤（折合2.82 mb/d原油当量）[自测算]，足以彻底抹平并主导地缘溢价消失带来…", "summary": "本报告对中国工业能耗转型在基础工业中的传导路径进行压力测试。测算显示，受地产收缩和高耗能原材料向低能耗先进制造转型影响，中国粗钢、电解铝及初级化工品需求收缩，结合短流程（EAF）与电解铝绿电替代，每年可永久性削减2.006亿吨标准煤（tce）化石能源消耗，相当于1.404亿吨原油，折合日均2.82…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源红利支持35%红利股防御逻辑，但最佳表达是12-15个百分点的政策现金流与供给风险对冲，而不是全仓油煤周期押注。", "summary": "截至2026-06-21，本卡支持在35%红利股袖套中配置12-15个百分点能源相关红利资产。依据是10Y国债收益率1.73%、中证能源股息率3.86%、中证红利股息率4.49%，以及央企市值管理和分红政策对大型能源央企的约束。中国神华2025年分红率79.1%，中国海油2025年派息608.38…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年能源冲击：上游供应链压力与新能源资本支出的经济性博弈 研究院工作日期： 2026 06 21 分析师： 能源行业分析师 ( energy analyst ) 本卡立场： 支…", "summary": "2026年能源冲击：上游供应链压力与新能源资本支出的经济性博弈 研究院工作日期： 2026 06 21 分析师： 能源行业分析师 ( energy analyst ) 本卡立场： 支持立场 (支持央行同步鹰派转鹰背景，并深入剖析转型资本挤压) 一、 执行摘要与核心观点 截至 2026 06 21 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 0, "value": 84.6, "events": [ { "title": "AI能源转型下的煤炭与铜：作为公用事业防御策略中大宗商品的有效性分析", "summary": "本卡压力测试 Research note 06 策略C 的 15% 商品敞口。结论：铜与煤炭方向相反，被错误打包。铜与标普相关性约0.6–0.8 [S8]，是 AI 周期的杠杆化 Beta 而非对冲——2008/2020 去杠杆中曾跌25%–60%，Research note 06 的 −8% 假…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2030年特高压与绿电并网能缓解全国电量约束，却不能完全对冲智算中心的本地可靠容量和电价基差风险，因此仍支持从粗放算力堆叠转向高能效算力、软件应用和电网CapEx受益资产。", "summary": "截至2026-06-22，本卡对前序电力瓶颈判断进行压力测试：特高压、绿电并网、煤电灵活性、抽蓄和新型储能建设足以缓解全国性年度电量缺口，但不能完全消除智算中心在本地并网、24/7可靠容量和市场化电价基差上的风险。国家电网规划“十五五”15回特高压直流工程、跨省输电能力提升35%、每年支撑不低于2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 5, "neutral": 0, "value": 68, "events": [ { "title": "修复大宗商品对冲层往返脆弱性的办法，是用与油价脱钩、对地缘正凸性的国防/实体经济工业敞口替换其部分现货能源beta——作为叠加而非整体替换，使K型/杠铃在能源升级与降级两分支皆能对冲。", "summary": "综合卡片01–06：体制判断（美联储看跌期权失效、霍尔木兹为扳机、K型/杠铃为防御解）成立，但卡片06揭示大宗商品层在能源冲击缓解时遭beta拖累、随Brent（$77.34/桶）往返回吐。我的工业制造判断：用国防/航空航天主承包商加电网电气化机械替换部分现货能源beta——其收入由多年度拨款支撑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027年原油与LNG庞大的物理供给过剩将系统性破坏当前的Backwardation结构，使得Roll-select策略的收益大幅缩减并转为负值拖累。", "summary": "截至2026-06-23，本卡对首席策略师关于利用Roll-select策略获取正向展期收益的建议进行压力测试。结论为警告性风险提示：随美伊MOU签署及霍尔木兹海峡逐步复航，且2027年全球原油供给过剩500万桶/日、LNG新增产能达8000-10000万吨/年，能源期限结构将系统性由Backwa…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "80%绿电强配溢价与电网/可靠容量可获得性，作为AI算力建设的约束性瓶颈及其对电力与新能源投资主题的含义", "summary": "本卡片综合卡片01-03，给出统一判断：AI建设的约束性瓶颈是'可获得性'而非'价格'。80%绿电强配以绿证年度履约成本极低（约+1%电费），即便24/7固化绿电PPA溢价也仅占TCO约2-3%[自测算]，价格可被超大规模厂商接受并转嫁。真正闸控2026-2030建设节奏的是并网排队、可信容量与小…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "绿色先进制造与城投信贷收缩将通过Capex砍减与地方投资减少，向电力设备及能源基建传导共计313.42亿元的需求收缩压力，并延缓光伏组件与储能电芯的价格修复。", "summary": "截至2026-06-23，本报告在金租及信贷收缩背景下评估其对电力设备与能源基建的传导。测算显示，全系统绿色制造信贷收缩将导致相关厂区配电及并网设备需求缩减100.03亿元；城投信贷收缩将导致地方能源基建投资缩减213.39亿元（对应新能源装机延期约6,096.86 MW）。合计拖累总需求313.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "订单与招标数据支持资金转向电网设备及新能源实物资产端，但尚未证明二级市场资金已完成全链条重配。", "summary": "截至2026-06-25，本卡支持“资金正从碳资产叙事转向新型电力系统实物资产”的方向，但限定为订单端确认。碳和绿证仍活跃；增量证据在国网特高压招标、电网设备公司中标、光伏组件、风电整机和储能采招。结论是：实物资产资本开支已明显落单，尤其是电网一次设备和特高压材料；但权益资金流、利润率和二级市场持…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "在风光（3-12个月建设周期）对特高压通道（18-36个月）分别具有6.0倍和2.25倍周期时滞的客观规律下，大基地并网瓶颈已导致2025-2026年新能源消纳率首次跌破95%政策红…", "summary": "截至2026-06-25，本报告对我国供给端（风、光、核）建设节奏与电网侧扩容周期的协同性进行压力测试。结果显示，光伏（3-6个月）和风电（~12个月）的建设速度大幅领先特高压外送通道（18-36个月），形成6.0倍和2.25倍的周期差[自测算]。这导致2025年风电光伏利用率在“十四五”期间首次…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 4, "neutral": 1, "value": 79.8, "events": [ { "title": "电网设备原材料缺口对新能源项目建设成本与并网进度的传导效应评估", "summary": "支持卡片06–08的逻辑链：AI电力瓶颈已上移至原材料，且GOES/高纯度铜缺口确实传导至新能源经济性，但传导非对称、主要走时间而非单价。直接价格通道有限——铜仅占光伏/陆上风电资本开支约3%~5%，即便铜较2020年+50%、GOES+100%，对LCOE直接抬升仅约+2%~+4% [自测算]。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Summary of Work I have completed the research and stress test on the transmission effects o…", "summary": "Summary of Work I have completed the research and stress test on the transmission effects of grid equipment bottlenecks on upstream key raw materials…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球电力设备制造供应链产能与上游材料供应风险评估", "summary": "部分确认Card-03的time-to-power论点，但作三处修正。瓶颈确在上移到材料端，但对变压器而言真正的物理咽喉是GOES而非铜，且GOES稀缺本质是“产能+贸易”稀缺而非地质稀缺：全球变压器产能利用率98%、GOES自2020年涨约100%、美国仅Cleveland-Cliffs一家、俄…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由于发电设备长交期、核电重建缓慢、跨州输电网架建设滞后以及并网容量在地理上的严重错配，电网侧能源增量无法匹配2026-2028年AI负载爆发式增长，电力缺口将实质性压制算力部署进度，…", "summary": "本卡对电网侧增量电源进行压力测试。2026-2028年，电网侧增量无法有效填补AI电力缺口。重型燃机面临5-7年超长交期；绿地核电停滞且SMR商业化迟于2030年，Palisades与三哩岛等重启仅能满足3.4%-10.8%的新增需求；输电线建设长达6.5-10年以上，且新增新能源地理错配严重（主…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 3, "neutral": 0, "value": 84.2, "events": [ { "title": "随着全球及中国变电设备产能在2027-2029年逐步释放，AI数据中心的电力设备瓶颈将从“无法获取设备”的二进制物理硬约束，降为“支付溢价与关税以获取交期”的资本支出与利润率成本约束。", "summary": "本报告对2027-2029年AI电力设备瓶颈的缓解速度进行了深入分析。结论表明，随着西方重电巨头（如日立能源、GE Vernova）扩产产能释放，以及以特变电工、思源电气、金盘科技为代表的中国高效率设备出口，2027-2029年电力设备将从物理上的“硬约束”转变为“成本/经济约束”。由于取向电工钢…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "评估“限定亏损能源”作为尾部商品对冲工具的有效性与风险边界", "summary": "压力测试卡片06的18%商品仓（黄金+限定亏损能源）。核心：能源仅是有条件右尾对冲（霍尔木兹/地缘），非核心多头。基准55-65%为去通缩、2027约5.0mb/d过剩、期限结构转Contango[S1][S3]，裸多头将因负展期失血并破坏对冲。故风险边界=仅做多虚值看涨/牛市价差(B1)、年在险…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网基础设施与能源转型对AI算力扩张的潜在约束", "summary": "压力测试部分支持并升级卡03：电力/电网瓶颈是“节流而非断路”，但节流比其暗示更紧更久。在2026–2028视野内，电气基础设施已是排在芯片之前的硬约束——美国电力变压器交期128周、GSU 144周、大单3–4年，价格较2019+77%、达2022前4–6倍；GOES由Cleveland-Cli…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 2, "neutral": 1, "value": 94.6, "events": [ { "title": "虽然备忘录缓解了短期实物原油压力，但霍尔木兹海峡的结构性脆弱与15%-25%的二次通胀概率支撑将大宗商品战术超配至20%，并应采用高分红个股现货替代及低配看涨期权策略进行防御。", "summary": "截至2026-06-27，评估表明尽管扫雷进展与伊朗出口正常化将推动短期WTI均价回落至67.95美元且跌破70美元概率升至62.85%，但15-25%的能源价格二次上尾再脉冲风险及长期地缘瓶颈支持将大宗商品配置超配至20%。在A股通道中，因159945等ETF极度缺乏流动性（成交仅106万元），…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹海峡重开仅为“执行折价”而非“和平红利”，60天物理排雷延迟导致到岸能源价具极强向下刚性，与高利率共同构筑新能源“资本开支壁垒”，系统性阻碍全球能源转型进程并抬升算力用电门槛。", "summary": "霍尔木兹海峡重开虽缓解了原油期货的即期上涨动能，但由于海峡内活体水雷清障导致的60天排雷延迟，到岸能源价格（CIF）维持强韧的向下刚性，地缘风险溢价只是发生了盘面转移。在高利率与绿色通胀双重作用下，全球新能源面临高额WACC与电网并网漫长排队的“资本开支壁垒”，转型进程阶段性受阻，算力能源供应链成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Summary of Work We have completed the analysis for Research note 04/10 as the Materials Ana…", "summary": "Summary of Work We have completed the analysis for Research note 04/10 as the Materials Analyst (材料行业分析师) , adopting a Support (支撑) stance. Our repor…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 3, "neutral": 0, "value": 87, "events": [ { "title": "在高利率长期化但衰退未确认的环境下，应以套息、凸性与实物资产对冲（短久期杠铃、适度美元、能源期权+黄金+能源股）替代方向性利率与油价押注，以改善组合风险回报比。", "summary": "截至2026-06-29，支持卡片01的高利率长期化、衰退未确认判断，并转化为配置：以套息+凸性+实物资产对冲替代方向性押注。久期保持低于基准，杠铃式收割2年期套息（约4.10%、久期约1.9优于10年4.38%、久期约8.2）并保留小段长端作衰退保险；相对低收益货币适度做多美元（DXY约101.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源/商品板块应以多元化商品篮子为核心、能源股为收益锚、少量原油为尾部风险工具，而不是把11%战略通胀对冲变成单一布伦特方向押注。", "summary": "支持Research note 02的11%商品/能源战略配置，但执行上不追逐现货原油。建议总组合中配置6.0% BCOM式多元化商品篮子、3.5%高股息能源股、1.5%流动性原油敞口。布伦特6月26日跌至71.99美元/桶，显示霍尔木兹正常化正在压缩战争溢价；但IEA/EIA仍提示2026年供给…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源商品成本传导与USD/CNY汇率风险的压力测试", "summary": "截至2026-06-28，本卡对卡3'不上调外汇对冲'结论做压力测试。被测机制（人民币贬值抬高能源进口成本、压低通胀对冲实际人民币收益）真实但属二阶，且当前被相反力量压过：能源对冲人民币收益的主导因素是美元油价中伊朗战争地缘溢价的崩塌，而非汇率。人民币油价成本=布伦特(美元)×USD/CNY，3月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 3, "neutral": 0, "value": 88.4, "events": [ { "title": "[auto-handoff] 先进制程星载 OBP 芯片良率瓶颈与算力/通信设备材料供应链", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 81384248-8464-45e1-8194-06f6bed14dff 结果摘要: Wrote handoff_response.md (10343 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中游房建及家装开工收缩-30.67%通过高经营杠杆深度挤压了上游钢铁、铝加工及基础化工的营业利润率与订单量，导致三大行业面临深幅利润下修压力（加权下修-67.3%），并伴随着2026…", "summary": "本报告对中游房建及家装开工收缩对上游原材料的传导进行了量化测算。-30.67%的开工收缩导致钢铁订单下滑10.73%，在螺纹钢售价接近现金成本和15%固定成本高经营杠杆挤压下，行业OPM由8.0%降至3.84%，净利润预测下修-58.0%。建筑铝型材订单下滑12.27%且加工费遭遇无序价格战下跌6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源与大宗商品资产在宏观政策背离下的防御价值研判", "summary": "能源与基本金属（连同黄金）能充当防御支柱，但价值是分层而非整齐划一的，须按机制而非标签构建。通胀冲击通道真实：通胀每意外+1个百分点，商品+7个百分点、股−3、债−4[S12]，2022年GSCI+34%印证[S13]。黄金是唯一双重（通胀+危机）对冲，央行一季度净买244吨、全年预测700–90…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-6f42e0fe6e/", "networkHref": "/investor/research-network/#thesis-6f42e0fe6e" }, { "id": "thesis-70daf87262", "source": "tracked_thesis", "title": "A 股红利与硬科技杠铃仍是政策、利率和风险偏好的组合表达", "titleEn": "A-share dividend and hard-tech barbell remains the portfolio expression of policy, rates, and risk appetite", "lane": "A股红利/硬科技杠铃", "laneEn": "A-share dividend and hard-tech barbell", "laneHref": "/investor/lanes/a-share-dividend-and-hard-tech-barbell/", "investableFocus": "A股红利/硬科技杠铃：510880.SH, 601398.SH, 600900.SH, 601088.SH, 600941.SH, 588000.SH, 688981.SH", "stockUniverse": [ { "ticker": "510880.SH", "name": "红利 ETF", "market": "A-share ETF", "role": "高股息指数代理", "roleEn": "高股息指数代理", "bucket": "core", "href": "/investor/stocks/510880.SH/" }, { "ticker": "601398.SH", "name": "工商银行", "market": "A-share", "role": "银行红利", "roleEn": "银行红利", "bucket": "core", "href": "/investor/stocks/601398.SH/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "公用事业红利", "roleEn": "公用事业红利", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "601088.SH", "name": "中国神华", "market": "A-share", "role": "煤炭红利", "roleEn": "煤炭红利", "bucket": "core", "href": "/investor/stocks/601088.SH/" }, { "ticker": "600941.SH", "name": "中国移动", "market": "A-share", "role": "运营商红利", "roleEn": "运营商红利", "bucket": "core", "href": "/investor/stocks/600941.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "硬科技权重", "roleEn": "硬科技权重", "bucket": "core", "href": "/investor/stocks/688981.SH/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" } ], "stockTickers": [ "510880.SH", "601398.SH", "600900.SH", "601088.SH", "600941.SH", "588000.SH", "688981.SH", "510300.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK", "9988.HK" ], "coreView": "赛道：A股红利/硬科技杠铃。候选观察标的：510880.SH, 601398.SH, 600900.SH, 601088.SH, 600941.SH, 588000.SH, 688981.SH。当前共有 40 条证据、2 位分析师和 7 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 81, "eventCount": 160, "firstSeen": "2026-05-15", "lastSeen": "2026-06-29", "supports": 10, "risks": 129, "neutral": 21, "analystCount": 15, "sourceTypeCounts": { "研究笔记": 149, "研究主题": 8, "研究线程": 1, "分析协作": 2 }, "topTerms": [ "科创50", "A股", "CPI", "融资余额", "长江电力", "公用事业", "中芯国际", "运营商", "电力", "红利", "银行", "A-share" ], "whatChanged": [ "2026-06-29: AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利景气度，A股资产配置应向电网出海及水电红利哑铃端重构[自测算]。", "2026-06-29: 国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多资产组合VaR在1530点处强制破位，急需分层部署的ETF买盘及两融弹性化阀门进行精…", "2026-06-29: 变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "2026-06-29: 在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红利（长江电力、中国核电）进行防御性轮动，并辅以日元和美债的尾部风险对冲。", "2026-06-29: 拥挤红利去杠杆与医药生物估值脱钩反弹的风格切换" ], "practical": { "sectorLane": "A股红利/硬科技杠铃", "sectorLaneEn": "A-share dividend and hard-tech barbell", "exposureBucket": "中国权益与流动性", "tone": "green", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "510880.SH", "name": "红利 ETF", "market": "A-share ETF", "role": "高股息指数代理", "roleEn": "高股息指数代理", "bucket": "core", "href": "/investor/stocks/510880.SH/" }, { "ticker": "601398.SH", "name": "工商银行", "market": "A-share", "role": "银行红利", "roleEn": "银行红利", "bucket": "core", "href": "/investor/stocks/601398.SH/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "公用事业红利", "roleEn": "公用事业红利", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "601088.SH", "name": "中国神华", "market": "A-share", "role": "煤炭红利", "roleEn": "煤炭红利", "bucket": "core", "href": "/investor/stocks/601088.SH/" }, { "ticker": "600941.SH", "name": "中国移动", "market": "A-share", "role": "运营商红利", "roleEn": "运营商红利", "bucket": "core", "href": "/investor/stocks/600941.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "硬科技权重", "roleEn": "硬科技权重", "bucket": "core", "href": "/investor/stocks/688981.SH/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" } ], "stockTickers": [ "510880.SH", "601398.SH", "600900.SH", "601088.SH", "600941.SH", "588000.SH", "688981.SH", "510300.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK", "9988.HK" ], "primaryStocks": [ { "ticker": "510880.SH", "name": "红利 ETF", "market": "A-share ETF", "role": "高股息指数代理", "roleEn": "高股息指数代理", "bucket": "core", "href": "/investor/stocks/510880.SH/" }, { "ticker": "601398.SH", "name": "工商银行", "market": "A-share", "role": "银行红利", "roleEn": "银行红利", "bucket": "core", "href": "/investor/stocks/601398.SH/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "公用事业红利", "roleEn": "公用事业红利", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "601088.SH", "name": "中国神华", "market": "A-share", "role": "煤炭红利", "roleEn": "煤炭红利", "bucket": "core", "href": "/investor/stocks/601088.SH/" }, { "ticker": "600941.SH", "name": "中国移动", "market": "A-share", "role": "运营商红利", "roleEn": "运营商红利", "bucket": "core", "href": "/investor/stocks/600941.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" } ], "instrumentTags": [ "510880.SH", "红利 ETF", "601398.SH", "工商银行", "600900.SH", "长江电力", "601088.SH", "中国神华", "600941.SH", "中国移动", "588000.SH", "科创50 ETF" ], "reviewCadence": "每日复核", "confirmationSignals": [ "融资余额、ETF 申赎和北向资金是否同向", "被动调仓窗口是否被主动资金提前交易", "订单、盈利和估值是否同步改善" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "指数期权", "红利低波", "汇率", "两融降温" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 134 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 15, "recentEvents": 134, "supportRatio": 0.06, "riskRatio": 0.81, "conflictScore": 90.8, "evidenceDepth": 100 }, "score": 74.8 }, "events": [ { "date": "2026-06-29", "title": "红利龙头去拥挤与超跌成长板块反弹的风格切换大论战", "summary": "红利龙头去拥挤与超跌成长板块反弹的风格切换大论战 A-Share Style Rotation: Dividend De-crowding vs. Growth Rebound 长江电力（600900.SH）资金大流出与药明/恒瑞大涨，是否标志着A股向超跌成长风格的实质性切换已在午间确立？伟星股份（002003.SZ）等中盘红利能否承接传统红利的溢出资金？", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "neutral", "supportLabel": "观察", "score": 32.5, "matchedTerms": [ "A股", "长江电力", "电力", "红利", "A-share", "dividend", "growth", "crowding" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 高拥挤大行红利资产减仓与资金流向评估", "summary": "致 金融行业分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: dede69e0-4ffc-4ff4-b119-7912477831ae 结果摘要: I will run the date command in the shell to anchor today's date before proceeding with the …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 32, "matchedTerms": [ "A股", "长江电力", "电力", "红利", "A-share", "dividend", "crowding", "bank" ], "href": "" }, { "date": "2026-06-29", "title": "A股风格大轮动：大盘拥挤红利瓦解与成长股超跌反弹的持续性辩论", "summary": "A股风格大轮动：大盘拥挤红利瓦解与成长股超跌反弹的持续性辩论 A-Share Style Rotation: Dividend Collapse vs. Growth Rebound Sustainability 长江电力资金大流出与药明/恒瑞大涨，是否标志着A股向成长风格的系统性切换已经确立？中盘红利如伟星股份能否承接传统红利的溢出资金？", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "neutral", "supportLabel": "观察", "score": 30.5, "matchedTerms": [ "A股", "长江电力", "电力", "红利", "A-share", "dividend", "growth", "share" ], "href": "" }, { "date": "2026-06-29", "title": "拥挤红利去杠杆与医药生物估值脱钩反弹的风格切换", "summary": "拥挤红利去杠杆与医药生物估值脱钩反弹的风格切换 Style Rotation: Utilities De-crowding vs. Biotech Rebound 长江电力等红利拥挤度处于95%历史极值，且股息率利差压缩至历史低点，而在跨境分段生产解禁及创新药商业化兑现下医药板块大涨，这是否标志着A股三年来最大规模的红利向成长风格切换已在午间确立？", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "supports", "supportLabel": "支持", "score": 26, "matchedTerms": [ "A股", "长江电力", "电力", "红利", "crowding" ], "href": "" }, { "date": "2026-06-29", "title": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红利（长江…", "summary": "本报告针对2026年三季报TMT硬件估值折价预期，校准了全球全天候和风险平价模型。战术防御性风险平价（TDRP）将AI成长股风险预算由25.0%下调至8.0%，防守红利上调至40.0%，以对冲55.0%的TMT增长溢价折价。模型显示，电力配给约束下算力效率的TCO优势在24/7清洁能源下扩大至9.97%，高功耗算力中心毛利率面临负增长。此外，建议红利套组内…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "长江电力", "公用事业", "电力", "红利" ], "href": "" }, { "date": "2026-06-29", "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优先既有水电/核电现金流、具备瓶颈定价权的变压器和电网设备供应…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "公用事业", "运营商", "电力", "红利" ], "href": "" }, { "date": "2026-06-29", "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）将使平仓率非线性飙升至20%，面临940亿STAR50抛压与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "科创50", "A股", "红利", "银行" ], "href": "" }, { "date": "2026-06-29", "title": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利…", "summary": "本报告对AI算力激增带来的电力电网物理瓶颈（变压器交期120-210周[S2]、并网排队4-7年[S1]）对A股转型龙头的利润率传导进行压力测试。结果显示，电网扩容滞后充当了新能源装机的“限速阀”，压制了光伏/电池销量增速；同时，容量电价[S8]和碳成本上升[S19]增加了高能耗制造环节电费开支，而在严重过剩背景下成本转嫁失败，导致制造端转型龙头利润率遭到…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "A股", "长江电力", "电力", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "局部中小银行信用缺口下的A股银行红利耐久性与高股息内部轮动", "summary": "卡08分离出的“额度沉睡”机制（CET1上限+NPL触发+低于1.40%净息差）不仅是中小银行信用风险，更是撕裂A股红利复合体的力量。重排结论：①大行红利居顶——分红不依赖沉睡额度，约4.35%股息率、对10年国债（约1.72%）257bp利差、0.5–0.7倍PB已计入坏消息，持有/超配；②公用事业高股息防御提级——长江电力等约3.6–4.2%股息率、承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "A股", "长江电力", "公用事业", "电力", "红利", "银行" ], "href": "" }, { "date": "2026-06-28", "title": "债市收益率触底倒逼绝对收益资金流向A股优质高股息红利股建立‘红利锚’，在‘数量先行、价格被锚’的结构性宽松下，配置策略应向国有大行与电力设备出口链的对冲型哑铃结构倾斜。", "summary": "本卡片合成分析了国内债市10Y国债至1.73%、信用Carry枯竭下的资产荒对A股的外溢效应。外溢主要通过险资等绝对收益资金流向A股红利资产（利差达+3.27%至+3.365%），形成局部红利锚。在价格降息受限、数量宽松先行的政策组合下，长久期成长股缺乏催化剂，建议实施红利大行防御与硬科技出口链主线的哑铃型配置，战术规避中小城商行与公用事业久期陷阱。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 25.5, "matchedTerms": [ "A股", "公用事业", "电力", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费成本转嫁挤压。", "summary": "基于国产芯片2.94倍能效赤字拉长通电时间至18个月的物理瓶颈，本报告系统性量化了A股电力设备板块（特高压、大变压器）的估值扩张机制与低利润 traditional TMT 的电价挤压。电网容量受限与十五五规划共振，支持了特高压订单可见度与大型变压器出海高溢价，特别是在国内外硅钢5.1倍巨大价差的成本红利下。策略上构建杠铃组合，超配直接出口变压器龙头、特高…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "A股", "公用事业", "电力", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电力’，并驱动策略从算力规模向算法效率和电…", "summary": "截至2026-06-28，本卡片通过对电网与变压器交期的压力测试，确认电网并网（5-7年交期）和变压器交付（2-4年交期）构成了AI算力扩张的终极物理边界。随着GB300（约135kW）及Rubin Kyber（约600kW）对800VDC和固态变压器（SST）的高密度配电需求增加，目前仅有不足8%的设施就绪。FERC否决核电表后直供，迫使所有新增容量重回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "A股", "公用事业", "电力", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "电网设备已进入局部拥挤和估值透支区，公用事业整体仍是中度拥挤的股息锚，2026H2应从“买电力”切换为“买可兑现利润的电力资产”。", "summary": "截至2026-06-29，本卡片对A股“算电协同”与红利资产轮动做压力测试。结论是：拥挤极值集中在电网设备和AI算力相关电力硬件，广义公用事业尚未达到历史极端估值。2026H2风险最高的是电网设备出口链和高PE电力硬件，尤其是订单强但利润兑现慢的公司；公用事业龙头更多是股息利差和资金拥挤风险，而非泡沫破裂风险。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "A股", "公用事业", "电力", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "算电协同和国网4万亿周期会延续A股向电网出口链的风格轮动，但当前应买质量和交付确定性，而不是追逐高估值特高压贝塔。", "summary": "截至2026-06-28，本卡支持A股资金从白酒等传统消费/红利资产向智能电网、特高压、电力电子和电网出口链轮动，但强调估值纪律。中证白酒指数2026-06-26收于52周低位附近且近一年跌32.6501%，电网设备ETF年内早期流入明显，中国西电近一年仍涨143.1373%。国电南瑞安全边际较好，19.8-22.0元为估值安全区；中国西电出口弹性更强，但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 25.5, "matchedTerms": [ "A股", "公用事业", "电力", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "A股应继续支持AI基础设施主线，但买点应从纯算力叙事转向供电约束受益资产，以35%/30%/25%/10%的科技成长、电网设备、公用事业红利和现金对冲哑铃执行。", "summary": "截至2026-06-28，本卡支持上游判断：AI基础设施交易没有被宏观衰退担忧证伪，而是被供电、电网接入、变压器和冷却约束重构。A股策略应从单一科技成长交易转为35% AI科技成长、30%电网及电力设备、25%公用事业红利、10%现金或对冲的主题哑铃。科技仍有上行，但电子和通信估值已处三年高分位；电网设备承接国网4万亿元十五五投资与数据中心供电瓶颈；公用事…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "A股", "公用事业", "电力", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "summary": "截至2026-06-28，本卡对“AI负荷增长能否支撑公用事业高分红”作压力测试：结论是精选组合层面可以，板块整体不能机械外推。AI数据中心用电增长真实，但也同步拉动电网、发电和储能资本开支。水电、核电、容量电价支持的火电及低杠杆综合电力运营商仍可维持分红；风光开发商受消纳、电价和项目资本开支制约，不能直接视为高息资产。上一卡的25%公用事业红利仓可以保留…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "公用事业", "运营商", "电力", "红利" ], "href": "" } ], "timeline": [ { "date": "2026-05-15", "supports": 1, "risks": 0, "neutral": 0, "value": 50, "events": [ { "title": "当前A股流动性结构与杠杆水平支持‘红利压舱、AI进攻’的杠铃轮动，资金从纯防御的大行向具备增量逻辑的公用、电信及AI基础设施平滑迁移。", "summary": "截至2026-05-15，A股融资余额约1.58万亿元，流动性环境支持从大行红利向公用事业、电信及AI上游产业链的结构性回填。北向资金呈现‘红利+AI’双向流入，验证了45%红利、35% AI、20%灵活仓位的杠铃配置可行性。银行板块释放的容量可被高股息公用事业平滑吸收，AI基础设施层已成为机构增…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "Brent高于104美元会通过输入性PPI和利润率压缩强化A股从外交beta和周期成长向水电、受监管公用事业与红利资产切换的趋势。", "summary": "截至2026-05-17，本卡支持上一卡判断：Brent突破104美元已成为独立的通胀与利润率冲击。美国4月CPI同比3.8%、能源同比17.9%；中国4月CPI同比1.2%、PPI同比2.8%，且石油、燃料加工、化工链条环比涨幅显著。油价从89.4美元升至104.2美元的16.6%冲击，按IMF…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "A股市场结构尚未完全定价2028年装备毛利重启，但海缆/主机和300065.SZ已局部拥挤，因此2026H2-2027H1应优先买业主回调和低拥挤海缆，延后重仓装备修复。", "summary": "截至2026-05-19，本卡压力测试第08卡后认为，两段式海风轮动仍可执行，但需要降低对拥挤海缆/主机和300065.SZ的初始仓位。A股当日成交2.91万亿元，电力净流入48.91亿元，但Wind口径显示公用事业主力净流入15.25亿元、电力设备净流出34.21亿元，资金更偏业主现金流而非全装…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 1, "risks": 0, "neutral": 0, "value": 54.2, "events": [ { "title": "电价上涨与碳成本显性化将导致A股高耗能行业内部利润剧烈两极分化，加速低效产能出清，同时由于容量电价理顺，电力板块正从强周期转向高股息的类债防守资产，迎来深度的估值溢价重塑。", "summary": "本研究评估了2026-2027年工业电价上涨及碳交易扩围对A股高耗能与公用事业板块的利润再分配及估值重塑影响。高耗能材料内部面临剧烈分化：“火电铝”盈利下调15%-25%，“绿电铝”因高壁垒上修5%-10%；钢铁长流程微利企业下调20%-40%；建材二三线水泥下调30%-50%，而海螺水泥等龙头因…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "两融资金去杠杆已通过账户交叉质押与VaR风控限制从TMT扩散至国防军工、新能源等高拥挤成长板块，必须通过‘分化杠铃’策略将大仓位配置于长江电力等红利安全岛、将科技仓位战术收拢至受国家…", "summary": "本卡片系统综合了A股去杠杆负螺旋对全市场的跨行业传染风险。我们测算指出，散户账户交叉质押与VaR风控触发将两融出清压力从TMT扩散至国防军工、新能源等高拥挤成长板块。科创50指数1,750点关口面临882亿元密集敲入与平仓卖压，但在国家队专项买盘防护下形成有底成长资产。我们建议采取‘分化杠铃’策略…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 1, "value": 46, "events": [ { "title": "A股高股息电力不是主动机构拥挤交易，但已成为被动红利资金拥挤载体，合理股息率边界约3.25%-4.25%，低于3.0%的核电/水电叙事资产尾部风险最大。", "summary": "截至2026-05-24，本卡压力测试A股高股息电力公用事业。结论是：板块存在ETF与战术红利资金拥挤，但主动权益基金并不拥挤，2026Q1公用事业配置仅0.45%，核电仅0.03%。容量电价提高煤电分红稳定性，10年期国债1.75%对应合理股息率走廊约3.25%-4.25%。长江电力3.77%仍…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面对日元套利平仓引发的全球去杠杆，A股是新兴市场内次坏的防御港，但承接资金的仅是国家队锚定的高股息低β国企篮子；指数总盘和北向重仓成长股仍是高β出口通道，必须主动回避。", "summary": "在第5张卡的10%日元套利平仓基准情景下，A股属于部分隔离但非免疫，正确姿态为杠铃式防御。承接防御资金的主篮是国家队β≤1.0、股息率≥4%、在岸资金主导的标的：国有大行、电信三大、长江电力/核电、高速公路与基础设施REITs、A股油气煤龙头；这些前十大股东中保险+社保+汇金占自由流通股55-60…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 3, "neutral": 1, "value": 35.4, "events": [ { "title": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构", "summary": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构 USD/JPY 145 Shock and Margin Deleveraging Reshaping the A-Share True Barbell USD/JPY下探145引发中资境外AT1永续债置换成本飙升，直接压降上市银…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构", "summary": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构 USD/JPY下探145引发中资境外AT1永续债置换成本飙升，直接压降上市银行NIM约3-5bp，动摇了‘金融红利’板块绝对防线；而科创50两融拥挤度高达4.9%，折算率下调引致250亿-450亿平仓压力，并威胁到敲入线挂接的27…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-28的核心判断是：泛红利有估值回归风险，但大水电尚未出现两融或公募意义上的拥挤踩踏，科技端才是更急迫的杠杆出清源头。", "summary": "截至2026-05-28，本卡支持“真杠铃”防御端继续以大水电为核心，但区别于泛红利。中证红利PE分位已达83.84%，说明宽基红利并不便宜；长江电力PE/PB自身历史分位分别约2%/13%，融资余额占流通市值1.81%，明显低于中际旭创2.74%和胜宏科技6.95%。两融与公募筹码仍主要拥挤在电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "145去杠杆延续叠加两融硬收紧，正在把A股从旧杠铃推向‘非金融国央企真红利+机构定价AI算力’的真杠铃，银行红利等中间地带被系统性挤出。", "summary": "在十四五（145）金融去杠杆遗产与2026年A股两融收紧（5月11日两融余额28338.58亿元创历史新高、东方证券等增设115%即时平仓线）的夹击下，传统‘银行红利+融资盘小盘动量’杠铃被拆解。本卡片确认‘真杠铃’重构判断：左端迁移至非金融国央企真现金流红利（能源/电力/电信/铁路港口），右端收…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "科创50指数在1,750点附近存在极高的信用与衍生品二次平仓反馈共振深渊，下周两融去杠杆风暴极易打压AI硬件龙头股估值，建议维持45%的红利防御性配置上限并买入虚值认沽期权对冲尾部风…", "summary": "本报告定量评估了科创50指数在1750点附近的雪球敲入（22.0亿元名义本金）与做市商Delta动态对冲（12.0亿元卖压）的二次共振效应。在券商两融折算率收紧（0.60至0.40）压力测试下，一阶总抛压达37.0亿至57.0亿元，极端踩踏下可触发二阶追加平仓反馈卖压119.7亿元，共振总抛压达1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 0, "value": 41.2, "events": [ { "title": "2026-05-29的核心判断是：保留公用事业基本面敞口，但降低红利低波因子beta，并把融资余额流出与小盘基差收敛作为战术减仓触发器。", "summary": "2026-05-29：本卡合成前序公用事业-AI电力主线，但将A股执行风险聚焦到微观结构。结论是支持但降风险：基本面公用事业仍可持有，红利低波与高股息ETF因子容器已出现拥挤，融资盘在总量扩张的同时对部分公用事业个股开始流出。若中证1000/国证2000对冲基差继续收敛，市场中性和红利低波资金可能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红利踩踏情景下A股真正的防御不是教科书上的'红利+公用事业'，而是'汇金锚定+国资增持+与外资定价权解耦'的三重筛选组合，配置重心应从长江电力等高北向红利核心切向国有大行+黄金+军工。", "summary": "本卡对'公用事业+国有大金融=防御铁三角'共识进行压力测试。在红利踩踏+跨风格虹吸情景下，真正的最强防御并非低估值红利，而是'汇金锚定+国资增持+与外资定价权解耦'的三重筛选组合。国有大行（工建农中）凭借汇金35%-64%直接持股[S1]和2024-2025年汇金ETF增持余波[S2]构成单边刚性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 6, "neutral": 0, "value": 18.6, "events": [ { "title": "医疗器械承接力瓶颈将迫使多余的A股防御资金溢出，流入容量更大、股息稳定且对集采绝对免疫的公用事业与电信运营商等蓝筹板块。", "summary": "截至2026-05-30，本卡指出医疗器械买入腿受限于RMB 30-50亿的单窗口容量，无法全盘接收从工业机械与拥挤银行流出的巨量资金。这迫使A股防御性资本溢出，流向具备百亿级承接力、高股息且免疫集采的公用事业（如长江电力）和电信运营商（如中国移动）。组合动作应确立“电力/电信为底仓蓄水池，高端医…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "NVIDIA业绩引发的高贝塔创新板块相关性飙升已被杠杆出清打破，跨资产风格轮动正向防守现金流与硬科技杠铃化配置演进。", "summary": "本报告分析了2026年5月20日NVIDIA爆表业绩对A股高贝塔创新板块（AI硬件、生物医药、空间经济、具身智能）相关性飙升的驱动机制。指出该共振由杠杆和叙事推动，本质十分脆弱。5月29日两融去杠杆引发科创50指数大跌5.04%至1751点，防守端公用事业暴涨，跷跷板效应明显。建议构建75-80%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "银行股息率跌破3.8%将引发约3500亿元机构避险资金的踩踏出逃，A股红利风格将暴力切换至以电信运营商和公用事业为核心的“真实自由现金流”资产。", "summary": "本卡片综合此前宏观与金融压测结论，研判银行红利资产崩塌后的A股风格轮动。随着银行股息率跌破3.8%及核心资本受损，其不再满足险资4.5%的精算收益率底线及红利指数的选股标准。预计H2 2026将有约3500亿元的避险流动性从银行板块被动踩踏出逃。A股市场风格将从“高杠杆高分红”转向“真实自由现金流…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "利率中枢稳定是A股核心资产的被动估值锚，资金正从被惩罚的超长端债外溢到类债权益，CSI 300/上证50将在三季度跑赢中证2000，Q3配置应采取高股息底仓+算力业绩兑现型成长的哑铃…", "summary": "本卡承接 card-02 关于长端CGB熊市陡峭化的压力测试结论，论证利率中枢稳定+长端温和回调（10Y CGB 1.71%、长端+20-35 bps）对A股是积极的结构性事件：折现率被锚定低位、CSI 300 ERP已升至历史99分位（约6.14 pct），股息率对10Y CGB首次形成+1.7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "截至2026-06-01，科创50在1,750点面临882亿元级联式去杠杆风暴，必须将高股息防御资产配比升至60%并利用衍生品对冲以构筑极限避险防线。", "summary": "截至2026-06-01，科创50收于1,751.32点，距1,750点生死线仅一步之遥。若跌破将瞬间触发144.0亿元雪球密集敲入（432.0亿元市商Delta卖压）及450.0亿元两融平仓，形成882.0亿元级联去杠杆抛压。我们建议将红利配置上限从45%升至60%，重构以长江电力（度电收益0.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026 05 31，A股市场在经历了2026 05 29的剧烈震荡后，结构性分歧已通过放量轮动得到阶段性释放。分析显示，全市场3.32万亿元的创纪录成交额伴随资金从极高拥挤度的…", "summary": "截至2026 05 31，A股市场在经历了2026 05 29的剧烈震荡后，结构性分歧已通过放量轮动得到阶段性释放。分析显示，全市场3.32万亿元的创纪录成交额伴随资金从极高拥挤度的半导体、AI硬件板块流向低位价值板块（银行、电力、大消费），这符合“高低切换”的拥挤度重置特征，而非全面的风险偏好收…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 4, "neutral": 0, "value": 33.4, "events": [ { "title": "在2026Q3–2027Q1政策窗口前，A股应通过减配低毛利服务器整机、超配智能电网与液冷龙头，构筑‘清洁基荷公用事业防御墙 + 西部枢纽节点单选成长极’的优化哑铃型双轨配置，实现极…", "summary": "本报告针对2026-06-01基准情景B下，发改委出台示范园区式20%-30%输配电价折扣对A股公用事业防御性与AI算力成长股的冲击，确立了“双轨配置框架”。我们定量证明了由于折扣门槛极高，长江电力等龙头T&D收入流失低于3.0%，红利防御墙稳固；而西部算力枢纽度电成本降至0.385-0.395元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在监管穿透迫使做市商瞬时撤流的环境中，执行流动性本身就是资产类别，组合应先买红利低波/公用事业、防御银行和现金流消费，后等确认再回补科创/TMT与茅指数。", "summary": "截至2026-06-02，本卡综合前五卡压力测试，认为下一阶段A股应从Beta与期权性切向现金流与执行确定性：超配红利低波和公用事业，保留银行但低配券商，消费仅保留防御现金流，低配科创和高Beta TMT；只有当两融余额、ETF流向、成交额和科创50价格同时修复后，才逐步回补科创/TMT和茅指数。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年美股公用事业板块在BTM自发电冲击下仍深陷‘资本支出搁浅与降级陷阱’，而A股水电与核电龙头凭借国家发改委绿电直连过网费刚性缴纳政策构建的‘强监管防御墙’，实现0%的表后绕过…", "summary": "本报告定量评估了2026年美国PJM电网并网与过网费重构政策对电力板块的利润率冲击。PJM新规强征每MWh约8美元的系统备用过网费，略微压缩独立发电厂（IPP）直连利润率至62%-65%（仍享有高额溢价），并将公用事业FFO/Debt信用指标小幅修复至13.7%（仍低于14.0%降级门槛，信用压力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "哑铃组合可以承受红利端6.3%-9.6%流动性回撤，但前提是红利仓位控制在16.0%、国债和黄金覆盖层有效、且信用利差突破70bp时主动削减成长beta。", "summary": "截至2026-06-01，本卡对62% A股哑铃权益、26% 中国国债、8% 黄金、4% 现金组合进行压力测试。若长江电力类红利资产仅发生6.3%-9.6%流动性回撤，且国债和黄金按历史避险方向发挥作用，组合最大回撤约2.3%-3.8%，可落在4.0%风险预算内；若成长端同步下跌10.0%、国债下…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 3, "neutral": 0, "value": 40.8, "events": [ { "title": "台积电关税成本转嫁推升GB200整机柜售价12.3%及云厂商TCO 8%，将引爆北美云厂商对 Blackwell 订单10-15%的下修，并迫使A股出海非GPU链降价5-8%，导致其…", "summary": "本报告针对2026-06-02台积电延期及关税转嫁事件，量化分析了其对英伟达GB200定价（上涨12.3%）与北美云厂商TCO（上升8.0%）的影响，指出云厂商可能下修订单10-15%。A股光模块、PCB、液冷出海链面临5-8%的压价与220-350 bps的毛利率挤压，伴随两融（550亿）及科创…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "TMT出海链处于极致拥挤与高杠杆共振的崩塌边缘，Blackwell利空将通过‘业绩下修-量化止损-融资爆仓’路径引发系统性踩踏。", "summary": "截至2026-06-02，A股TMT出海链面临系统性流动性出清。公募重仓配置比例达32%（Top 5%分位元），中际旭创、新易盛等核心标的融资余额合计超550亿元，平仓线集中在当前股价-45%至-50%区间（如中际旭创580-640元）。Blackwell砍单10-15%将触发量化指增产品止损（3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在两融高拥挤（占比22%）且维持担保比例受压的背景下，核心标的跌超40%将引爆每日312.4亿元的“多杀多”共振抛压并锁死流动性；科创50跌破1500点将引发全市场ERP暴涨170 …", "summary": "截至2026-06-02，本报告针对中际旭创与新易盛跌幅超40%触及融资平仓线（580-640元/340-380元）的情境，压力测试了两融爆仓、“多杀多”共振与科创50跌破1500点的影响。测算显示，两股两融余额合计550亿元，跌40%将导致99.7%的账户爆仓，在T+3窗口产生每日219.4亿元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 1, "risks": 3, "neutral": 0, "value": 50.2, "events": [ { "title": "若10Y国债1.85%破位由核心CPI、PMI与社融质量共同验证，A股将系统性从防御红利切向顺周期和盈利成长，但应采用分阶段轮动而非一次性清仓公用事业。", "summary": "截至2026-06-03，支持“宏观共振后红利让位于顺周期与成长”的方向，但触发尚未完全成立：核心CPI和PPI已转强，PMI订单与库存、社融质量仍不足。若10Y国债连续5日站上1.85%，应将公用事业红利从6%-8%逐步降至2%-3%，资金优先轮动至有色、化工、机械、煤电一体化，再转入电子、AI…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "在政策放大稀缺性溢价下，A股组合应以‘成长出海的电网设备为矛，资源通胀的铜为盾，高股息的电力运营商为锚’进行排序配置，并以电网投资额、出海关税及现货电价为量化证伪点。", "summary": "截至2026-06-04，本卡确立A股“物理稀缺范式”资产配置排序：首选电网设备（成长+出海高景气），次选铜资源（高弹性抗通胀），末选电力运营商（核心红利防守锚）。报告结合思源电气、紫金矿业、长江电力等龙头最新业绩及十五五规划，并为三者制定了量化证伪触发条件。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建全球产业周期景气度仍在加速并非泡沫见顶，但A股本地面临极高的两融多杀多挤兑风险，必须进行仓位控制及衍生品尾部对冲。", "summary": "中际旭创登顶沪深300第一权重股与Marvell单日大涨32.52%标志着AI基础设施资产登顶中美双市场。尽管全球算力与网络需求（旭创Q1利润增262%）和英伟达 endorsement 证明主升浪仍未终结，但A股本地TMT两融余额（5800亿元，旭创融资比率8.50%）呈现极端拥挤。需通过限额至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "若杠杆资金撤出AI、科创板和创业板担保品，首选承接方向是沪深300价值中的银行和保险，其次才是公用事业、通信运营商和选择性高股息周期品。", "summary": "2026-06-04：本卡综合前序0DTE日内波动、两融担保品折价和AI/科创/创业板拥挤风险，结论是防御性承接不应简单买入广义高股息，而应优先买入沪深300价值里的银行和保险，其后配置公用事业与通信运营商，再以主要消费/家电和受控能源红利作补充。沪深300价值在2026-04-30的一年年化波动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 9, "neutral": 0, "value": 15.6, "events": [ { "title": "Research note 06 · TMT 视角对\"物理红利\"水电防御性论断的质疑 工作日：2026 06 05（亚洲/新加坡） 分析师：TMT 行业分析师（tmt analyst…", "summary": "Research note 06 · TMT 视角对\"物理红利\"水电防御性论断的质疑 工作日：2026 06 05（亚洲/新加坡） 分析师：TMT 行业分析师（tmt analyst） 立场： deny （否定\"水电物理红利可作为商业银行高股息替代仓位\"的判断） 板块：6d8317f2 …/Res…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球再通胀+久期承压下，A股/港股的红利/低波相对成长的战术超配窗口已被确认（2/3信号触发），建议将SAA风格权重调至红利32%+价值28%+成长28%+主题12%，行业上重点超配…", "summary": "首席策略师支持卡片01-02的再通胀+TAA重平衡逻辑链，并在A股/港股层面给出具体落地。截至2026-06-04，沪深300滚动PE 14.85倍（百分位93.76%），中证红利股息率4.94%，恒生高股息约5.84%，对应A股10Y国债约1.70%的红利股利差高达约324bp，构筑国内独立贴现…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "能源标的\"物理红利\"防御性验证：现金流稳定性与电价机制评估 工作日：2026 06 05（亚洲/新加坡） 分析师：能源行业分析师 (energy analyst) 立场：suppor…", "summary": "能源标的\"物理红利\"防御性验证：现金流稳定性与电价机制评估 工作日：2026 06 05（亚洲/新加坡） 分析师：能源行业分析师 (energy analyst) 立场：support（支持卡04把战术性资金从商业银行高股息转向\"物理红利\"水电的判断） 板块：6d8317f2 c0d6 4f27 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美股大厂利润受损与NDX估值收缩将导致A股高拥挤AI硬件链面临-34.3%的估值与盈利双重杀跌压力，并触发74.5%概率的两融去杠杆踩踏；而绝对利差高达+249 bps的A股红利资产…", "summary": "基于2026-06-04，本报告量化评估了美股NDX重估对A股AI算力链（CPO、PCB）及红利资产的溢出与共振效应。测算表明，A股AI硬件等权篮子交易于32.5x的2026E P/E，其估值对美股弹性系数为1.35。美股大厂因PPA及电网延迟导致的利润率受损及NDX 20%的估值压缩，将使A股算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 1, "risks": 2, "neutral": 0, "value": 59, "events": [ { "title": "两融、雪球与股指期货去杠杆共振时，防线必须从静态套保改为预留现金、动态降杠杆和红利锚容量配给，否则对冲会先于底仓卖出能力而失效。", "summary": "截至2026-06-06，本卡压力测试认为，A股尾部风险的核心不是缺少IM、期权或红利资产，而是底仓跌停锁死时，对冲端被保证金和盘中反弹先行强平。最新可得两融口径显示，2026-06-04沪深融资余额约2.89万亿元；若3%-6%被动退出，可能形成约867亿-1734亿元卖压。长江电力可作为红利锚…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "115%强平线不是全市场统一规则，但若科创弱账户已接近该线，真正的尾部风险是人民币40.20亿元基准外溢卖压压向红利ETF和银行ETF等高流动性担保品。", "summary": "截至2026-06-05，本卡压力测试认为1,720点已不是未来触发线：公开报道显示科创50在2026-06-01已收于1,663.69点。基于2026-06-04沪深北两融余额29,206.86亿元、2026-06-03科创板融资余额3,828.21亿元测算，若5%-20%科创融资弱账户触及11…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股红利应保留，但核心是有行业上限的质量低波红利，而不是无差别追逐最高股息率。", "summary": "2026-06-05，本卡支持保留35%-40%A股红利袖套，但建议从宽泛高股息转向“质量红利+低波动+行业上限”。核心配置为受监管现金流公用事业/电信、大型国资金融、受限能源央企，并用交运、消费必需品和医药红利资产稀释银行与能源拥挤。应回避中小银行票息拥挤、能源商品单因子化、钢铁水泥地产链收益率…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 6, "neutral": 2, "value": 42.4, "events": [ { "title": "在融资杠杆处于极值且科创50下面临雪球敲入踩踏的情景下，超配低贝塔、高分红的避风港资产长江电力是防守去杠杆和风格转换中对冲系统性风险的最优策略。", "summary": "截至2026年6月4日，A股融资余额达2.90万亿元高位，科创50自高点回调至1668.33点，逼近1125-1360点的雪球敲入密集区。一旦触发敲入，券商平仓多头将引发市场二次下跌并导致融资盘连环爆仓。建议超配长江电力（Beta为0.09，股息率3.7%，承诺70%以上分红率）等低贝塔红利资产，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "RMB双向尾部下哑铃不能对称加码——整体降β、红利从银行单极改为电力主导的四极结构、科技从硬件AI高拥挤切到自主可控+军工电子，电力是能源冲击下的最佳避险红利。", "summary": "针对Card-02锁定的人民币双向尾部情景（USD/CNY破位至7.00–7.10 vs 日元平仓拉升至6.60），A股哑铃组合不应等权双向加码，而应整体降β并内部再平衡。基准60/40哑铃在情景A调整为40/50（科技降15pct、红利加10pct、黄金7%），情景B回补至60/30。红利端从银…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在系统性流动性危引发避风港资产补跌与相关性收敛的背景下，跨资产配置至实物黄金与短债，并配合科创50 ETF Put与IM期货空头对冲，是防御VaR穿透并平衡权益融涨弹性的最优解决方案。", "summary": "针对2026年6月7日A股2.90万亿两融余额极值及雪球敲入踩踏引致的系统性流动性危机，本压力测试表明长江电力等避风港资产在挤兑期存在补跌风险，相关性收敛导致未对冲组合10日损失达-16.60%，99% VaR穿透至32.33%[自测算]。本方案设计了科创50 ETF期权与IM期货空头对冲包，并对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源冲击越深，电力红利权重越应从火电主导切向水核主导——水电与核电凭借零燃料成本+容量电价+调峰收益是FX不敏感的反脆弱α避险红利。", "summary": "支持Card-03的电力主导红利结论。截至2026-06-07，能源冲击下A股电力板块依靠'煤电联动+容量电价+水核零燃料'三重缓冲足以承担避险红利角色。压力测试显示：在煤价升至¥1,200/吨、JKM达$25的重度情景下，水电EPS仍+12%、核电+5%、火电因容量电价(¥100/kW)对冲后仅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 1, "risks": 5, "neutral": 2, "value": 49.8, "events": [ { "title": "PPI剪刀差叠加央行汇率掣肘不是风格大切换的发令枪，而是A股层面\"上游+红利防御 / 减下游可选 / 减高估值成长\"哑铃结构与整体降一档beta的强化信号。", "summary": "截至2026-06-07，本卡对PPI剪刀差与人民币掣肘下的A股配置做压力测试：支持前序风险预算下调一档，并在A股内部以\"哑铃+防御\"结构落地——超配上游能源（石油石化、煤炭）、有色（铜铝金）、公用事业、大型国有银行与电信运营商；低配下游可选消费（家电、汽车整车）、中游一般加工与高估值小盘成长。组…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在M1及社融等硬数据证伪“信用脉冲”后，流动性受限与解禁压力迫使A股向具备325bp高股息利差的防御性红利板块（公用事业、银行）全面切换。", "summary": "基于信用脉冲被证伪及流动性维稳环境，我们测算A股增量资金见顶（融资余额约1.48万亿）且Q3面临科技股6000亿解禁压力。中证红利指数5.5%的股息率相较2.25%的10年期国债收益率提供高达325bp的安全垫，确认策略重心向公用事业及国有大行等红利资产切换的逻辑成立。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融、ETF与行业成交结构已部分确认A股政策顺周期切换，但宽度不足，仍属头部权重驱动的早周期信号。", "summary": "截至2026-06-08，两融、ETF和行业成交三类微观证据共同支持卡片01、02的政策顺周期切换判断：杠杆资金在非银金融、银行、电力设备、军工、机械的融资买入占比抬升，在食品饮料、医药、地产下移；ETF净申购集中在券商、银行、军工、高端装备和央企红利/中特估，纯地产与白酒ETF净赎回；金融与先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股市场结构验证：红利防御是否拥挤、成长回撤是否可买", "summary": "截至2026-06-08，本卡用五项市场结构证据压力测试Research note 05。结论：配置切换尚未被过度交易，但拥挤不均匀，且成长回撤在6月被供给污染。成交连续30日超2.5万亿但呈轮动而非衰竭[S11]；两融创2.8万亿新高却已降温、保证金比例回升至100%去杠杆[S1][S2][S1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 6, "neutral": 0, "value": 31.2, "events": [ { "title": "银行红利仓设上限后的A股策略再分配：释放风险预算应转向哪些行业或风格", "summary": "综合 Research note 01–04：银行高分红股权封顶于跨资产套息仓约30%是正确的风险纪律，但在A股账本上留下收益缺口与回报缺口。策略答案是哑铃式再分配：收益端从'银行—利率因子'分散到非银现金流红利（煤炭/能源、交运/高速、运营商、水电）加自由现金流因子与央企'HALO'红利（指数股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50破位会触发局部去杠杆压力，但在流动性缓冲和银行相对防守尚存的情况下，不宜直接判定全市场两融被动平仓踩踏已经启动。", "summary": "2026-06-08，本卡判断科创50跌破1720更像局部科创/半导体去杠杆，而非全市场两融踩踏。两市融资余额为2.89万亿元，科创板两融为3805.56亿元，压力真实但尚未失序；银行在6月5日逆势上涨1.33%，说明红利/大行对人民币组合仍有相对防守作用。美元口径风险来自中美10年期利差约-28…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "绿电溢价与CEA抬升将终结A股电力红利与有色周期的同质化行情，催生“水电/核电红利成长化”与“水电铝阿尔法溢价”的戴维斯双击，同时迫使被动资金重构传统指数权重。", "summary": "截至2026年6月8日，承接上游成本测算，本卡片认为范围二纳入将引发A股公用事业与有色板块的深度估值重塑。4-6分/度的绿电溢价将使大水电/核电从纯“红利”向“红利成长”重估，而火电因面临碳成本与新能源资本开支压力面临股息缩减风险。电解铝板块将从同质化周期贝塔转向结构性阿尔法，水电铝凭借吨铝783…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50跌入1520-1550点区间将触发电子板块超2000亿元两融强平，由于高折旧经营杠杆与良率折损，抛压将向中游制造与封测传导，导致其中游毛利率滑落至 -53.8% 并引发二次估…", "summary": "本报告针对科创50指数击穿1600点、回撤至1520-1550点区间的压力情景进行了两融强平与产业链传导测算。电子板块5665.69亿融资盘在此区间面临36.94% - 44.63%的刚平比例，释放超2000亿抛压。中芯国际、澜起科技等权重股表现出较强估值弹性，但中游制造与封装企业因砍单面临-60…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 2, "neutral": 0, "value": 56.6, "events": [ { "title": "聪明钱（北向）已前置买入上游油气、电网、铜铝，但融资盘与低波 ETF 仍在为防御板块支付第二次溢价——Card-04 行业-风格再分配蓝图的执行窗口约剩 2–4 周，关键倒计时是红利…", "summary": "截至 2026-06-09，A 股资金面部分确认 Card-04 的再分配方向但尚未走完：北向已连续 5 日净流入并集中在中石油、中海油、紫金、长江电力、国电南瑞、特变电工等篮子 A/B 标的，融资余额却仍在缓慢加白酒、火电、银行等防御板块，红利低波 ETF 净申购仍大于能源 ETF——拥挤防御的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力约束强化红利资产的股息利差和现金流确定性，但TMT硬件仍有业绩期权，最佳策略是60/30/10的杠铃式再平衡，并用利差和融资拥挤阈值控制踩踏风险。", "summary": "截至2026-06-09，本卡支持AI电力约束推动A股“物理红利”重定价，但建议以组合而非单票方式执行：基准仓位为60%物理红利资产、30%TMT算力硬件、10%现金或对冲。中证红利股息率5.06%、对10年国债利差+332.5 bps，仍具吸引力；长江电力按1.00元股息和27.63元收盘价计算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 5, "neutral": 4, "value": 52, "events": [ { "title": "2026-06-12若触发TMT两融去杠杆，行业配置应从“融资驱动AI硬件”转向“现金仓核心科技+高流动性红利防御”，但不应追逐所有高股息资产。", "summary": "截至2026-06-10，本卡支持前序“06-12前降低TMT融资仓位”的方向，但将结论收窄为融资流动性切换：科创50调仓可能使AI硬件抢跑资金借被动买盘退出，并通过两融账户担保品压力传导到TMT整体。红利风格将成为资金承接方向，但不是无差别主升；长江电力更像高流动性防御锚，而非低拥挤标的。建议降…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股AI硬件抱团松动：资金是否切向化工/银行防守杠铃？", "summary": "A股AI硬件抱团松动：资金是否切向化工/银行防守杠铃？ A-share AI hardware crowding break: shift to chemicals/banks barbell? 6月10日午盘，科创50从+3.56%回落至-0.02%、寒武纪从+6.29%翻绿，同时工业气体/光刻…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "防御档以银行/公用/煤炭/食饮四大重仓+电子-700bp为对手盘；修复档反向，电子+700bp但须用北向条件触发分拆建仓，跨档摩擦约10-11bp迫使触发器严守。", "summary": "本卡在card-03的防御/中性/修复三档框架下，给出申万一级28行业相对沪深300基准的BP级主动权重。防御档（成长25/红利55）最大赢家为银行+600bp、公用事业+500bp、煤炭+450bp、食品饮料+400bp；最大输家为电子-700bp、电力设备/计算机各-500bp、医药-400b…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "面对AI硬件估值高位震荡与电力并网交付的局部约束，构建分层双杠铃组合（超配电网设备与红利公用事业）能够提供稳健的收益安全垫，并有效防范两融资金爆仓和电力延误风险。", "summary": "本报告针对AI硬件拥挤与电力并网约束，提出构建分层双杠铃组合（30%AI硬件、20%电网设备、15%公用事业、20%银行红利、15%化工材料）。测算表明，若TMT拥挤踩踏，组合回撤仅-9.0%（纯AI硬件-30%）；若并网延迟，组合近乎持平（-0.5% vs 纯AI硬件-15%），显示了强大的对冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "在红利极化与公用信用分层并存的局面下，最优解是'水电核电守正60% + 出口链纯粹价值奇兵40%'的双底座再分层，而非任何一端的极端解。", "summary": "红利底座不应在'坚守水电核电'与'大举切仓出口链'之间二选一。建议执行'守正60%/奇兵40%'的双底座再分层：保留信用免疫的大水电与核电(国网总对总结算)，退出地方信用尾部恶化的中小水电、市政燃气、供热；将原红利底座的30-40%替换为公募严重低配、有息负债占比<15%、海外营收40-50%的出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 3, "neutral": 1, "value": 57.8, "events": [ { "title": "长江电力处于后资本开支期，其70%分红承诺在极端枯水和市场化电价压力下依然具备自我造血的刚性支撑；而中国核电在千亿资本开支高峰下自由现金流持续为负，其分红极度依赖外部融资，属于‘成长…", "summary": "截至2026年6月13日，水电与核电分红红利底座呈现显著分化。长江电力处于后资本开支周期，自由现金流（FCF）极度强劲（基准情景488亿元）。压力测试显示，在15%极端枯水和电价市场化波动的压力下，长江电力FCF（265.4亿元）依然能完全覆盖70%的分红承诺（157.5亿元），红利底座坚如磐石。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力红利股受水文及电力改制冲击已呈现明显的‘波动性红利’特征，资产管理人必须通过调升预期波动率、应用水文安全折价并引入气候因子来重构全天候策略的防御底座。", "summary": "本卡对电力红利股分红金额波动性对风险平价及全天候策略的系统性冲击进行了压力测试。长江电力虽然维持70%分红比例，但极端枯水与电价折价下归母净利润及分红名义总额将萎缩达35.7%。这导致其 realized volatility 扩张，在四资产逆波动率配置模型中使其资金权重自动缩减6.1%（相对降幅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50尾部冲击下红利并非对冲、必须以期权+国债期货+黄金的跨资产组合替代，并对红利仓位做高Beta降权、低Beta增配的结构重构。", "summary": "在2026-06-13的评估中，针对卡04测算的科创50极端回撤导致99% VaR激增48%、跨板块相关性飙至0.82的情景，本卡提出组合重构方案：硬科技降至26%、高Beta红利（券商/周期煤炭）降至6%、低Beta红利（电力/四大行/运营商）升至24%、利率债升至28%、黄金ETF新增4%。对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-13的核心判断是：先降高久期成长、券商和周期红利担保品，保留大型银行、运营商、公用事业和交通基础设施等低Beta现金流资产。", "summary": "2026-06-13最终综合认为，科创/创业板负Gamma、中证红利担保品平仓、券商两融账簿收缩若同时发生，应优先降权高久期科创/创业板、券商Beta和高Beta周期红利，而不是一刀切卖出全部红利。本文同时修正card-09的单位口径：头部五家券商样本敞口应约为8,709.3亿元，40%受影响时4…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 2, "neutral": 0, "value": 62.2, "events": [ { "title": "A股杠杆结构在100%保证金规则下韧性显著，红利退出压力可控，风格切换的主要风险在于科创50杠杆增长过快引发的情绪过热，而非两融爆仓。", "summary": "针对2026-06-14的A股杠杆结构压力测试显示，红利板块虽处于90%拥挤度高位，但得益于100%融资保证金规则，具备35%的价格安全垫，风格切换预计表现为“有序撤离”而非“平仓崩溃”。科创50融资余额触及自由流通市值5%的情绪极值，虽理论容量巨大，但需警惕情绪过热。整体质押风险局部化且受控，市…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力瓶颈已成为新的约束边界，但市场交易已极度拥挤，建议从“设备”转向“运营”以规避短期回撤并捕获长期红利。", "summary": "截至2026年6月，AI主线已从模型能力转向电力瓶颈。虽然超大规模云厂商资本支出激进且向基础设施倾斜，但A股电网设备（PE 43-45x）和半导体设备拥挤度已达极值（88%-100%）。建议下周降低设备板块集中度，迁往估值处于28%分位且具备2.3-2.9%股息率的电力公用事业标的，以平衡结构性机…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 4, "neutral": 1, "value": 54.6, "events": [ { "title": "电网（准许收益率）、核电（零燃料弹性）、头部煤电（两部制电价+长协）三链独立成立，A股公用事业瓶颈红利能转化为真实利润扩张。", "summary": "支持卡07：A股电力/电网瓶颈在2026年确实转化为电网、核电及头部煤电运营商的利润增长，并非被上游成本完全抵消。三条独立成立的传导链：电网受准许收益率框架保护、与上游能源脱钩，叠加2026年8500亿元投资计划，国电南瑞等设备商Q1新签订单+30%；核电燃料成本占比仅15-18%，对Brent …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡04–06的全球方向正确，但A股落地需改写为重红利/独立加配资源/轻硬科技/把电力瓶颈视为A股利好的非对称哑铃。", "summary": "首席策略师否认卡04–06把美元利率与美股算力叙事直接外推到A股的做法。在大类层面认同减股增商品/黄金，但在A股权益35%内部应改写为非对称结构：红利/资源18%、银行/保险10%、公用电网/核电10%、硬科技降至8%、军工4%、其余5%。理由：A股半导体由国家大基金三期(3440亿元)与政策性贷…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在国债收益率回升与汇率防御紧平衡压力下，A股红利资产将呈现大行受益、水电压缩的K型利率传导，而高昂的离岸对冲成本将加速外资抛售内需核心资产，唯有大行红利与出海核心资产可实现+8.31…", "summary": "基于2026-06-15，本报告对人行汇率防线的传导路径实施压力测试。测算显示，30年期国债收益率从2.232%反弹25 bps将导致久期约40年的大水电红利（如长江电力）面临-11%至-14%的估值回撤，但由于收益率曲线陡峭化及NIM改善，股息率5.40%的大型国有行将逆势走强。商行存款规模达3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DR007利率在央行结构性对冲下企稳在1.48%以稳固信用环境，但两融高位去杠杆仍在电子和科创板诱发踩踏，跨资产向物理红利与黄金再配置（策略3）是锁定VaR风险（-18.15%）并保…", "summary": "截至2026-06-15，央行对冲稳定DR007在1.48%，AAA信用利差仅增5bp。但两融经历高位去杠杆，从2.94万亿元收缩至2.82万亿元，科创50击穿1600点零Gamma线并强制平仓35.75亿元，由国家队日均承接3亿元平稳。避险资金轮动至长江电力等物理红利（两融占比1.81%，CGB…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 0, "value": 65, "events": [ { "title": "在原料退化与B端品质下沉的500–850bp毛利压迫下，应将海天/青啤从超配切到明显低配，并将释放的约400bp仓位重构为'政策性口粮必需品+非餐饮型大众红利'的战术哑铃，板块保持中…", "summary": "工作日期2026-06-15。综合Research note 01–05结论：Research note 05给出的500–850bp原料/产品结构退化压力，已将Research note 02基于15–60bp渠道漏损建立的安全垫击穿10倍以上，原'做多干净成本传导'防御逻辑被否证。建议将海天(…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股高股息防御不应撤退，而应从“银行高股息一篮子”升级为“强资本银行+非银稳定现金流”的分红耐久性组合。", "summary": "截至2026-06-16，本卡综合前序信用与金融分析：城商行分红压力不是银行板块系统性退出信号，而是高股息策略从静态股息率转向分红耐久性的触发器。银行仓位应内部重构，保留国有大行和优质头部区域银行，降低薄资本中小城商行；高股息防御应提高公用事业、运营商、央企能源与交通运输等非银现金流资产权重。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 4, "neutral": 1, "value": 57.4, "events": [ { "title": "两融折算率下调引致的350亿元刚性去杠杆将重创高贝塔科技硬件板块，推动资金向筹码结构极净、收益率稳定的‘物理红利’与‘成本通缩’AI SaaS板块非对称旋转。", "summary": "截至2026年6月17日，本报告评估了两融折算率从0.6下调至0.4对A股两融规模和市场结构的影响。券商分批落地该政策将引发250亿至450亿元（中枢350亿元）的刚性融资收缩。科技硬件（电子/计算机）板块融资余额高达5608.81亿元，面临50亿至112.5亿元的平仓卖压。在负Gamma共振下，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Research note 08的杠铃方向成立，但中国权益实施必须从“宽泛受益”改为“成本转嫁能力、出口政策韧性和人民币压力三重筛选后的选择性超配”。", "summary": "截至2026-06-16，本卡对Research note 08的全球与A股杠铃配置进行中国宏观压力测试。结论是方向可保留，但中国不是美国滞胀的纯受益方：5月CPI仅+1.2%、PPI达+3.9%，2.7个百分点剪刀差显示外部冲击首先压缩下游利润。人民币强势和1.75%的10年期国债收益率给人民银…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网瓶颈促使云巨头表后逃逸，引发美国公用事业板块搁浅资产否决、信用评级降级与分红削减的戴维斯双杀；中国发改委绿电直供政策则构筑了零表后逃逸的阿尔法防线，凸显长江电力与中国核电的绝对避…", "summary": "定量评估AI数据中心并网约束下美国受监管公用事业企业的信用与分红稳定性。在变压器短缺及并网延误下，云巨头转向表后自发电（BTM），导致电网公司面临21.0亿美元Capex监管否决，FFO/Debt降至13.21% [自测算]，跌破14.0%投资级门槛 [S10]，面临下调至BBB及分红被迫削减的戴…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力交付物理瓶颈尚未在A股TMT板块充分定价，高估值硬件标的面迎15%-45%估值去化风险，应当战略重构组合并转向物理电网设备、上游矿山及防御性清洁能源公用事业。", "summary": "本报告基于电网与BTM直供电物理硬约束验证的12-24个月算力基础设施部署气穴期，指明A股算力硬件与上游材料板块存在估值脱钩和两融踩踏风险。建议在气穴期内进行风格轮动，超配出口导向变压器和上游铜铝，增配防御性清洁公用事业（长江电力、中国核电）。定量测算表明，优化后的战略组合可将承压收益率提升7.8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 4, "neutral": 0, "value": 55.8, "events": [ { "title": "AI电力稀缺是SPCX拥挤风险的有效对冲，但美国AI电力赢家本身已部分拥挤，组合应超配稀缺因子、低配拥挤表达，并用水电红利与成本归因规则做底仓。", "summary": "2026-06-17：本卡综合Research note 01-04，结论是AI电力稀缺资产仍可作为SPCX拥挤风险的组合对冲，但必须拆成结构性对冲与拥挤化对冲两类。保留有合约支撑的核电IPP、600900式水电红利和具备大负荷成本归因的公用事业；但CEG、VST、TLN等美国AI电力赢家已带有成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘政治溢价消退下A股红利资产内部的“去能化/去油化”切换", "summary": "地缘政治溢价消退下A股红利资产内部的“去能化/去油化”切换 Decarbonizing and De-oiling A-Share Dividend Allocation Under Oil Deflation 美伊和平协议拟重开霍尔木兹海峡，导致原油价格破位（Brent 79.49，WTI <7…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘溢价消退将引发A股红利资产内部的结构性切换，资金将从“涉能”周期红利流向免疫大宗商品通缩的纯内需防御资产（水电、家电、白酒）。", "summary": "原油价格因美伊和平协议跌破关键支撑位，标志着地缘政治溢价消退。A股能源红利资产（如煤炭）面临盈利与估值双杀风险。建议在红利组合内部进行“去能化”轮动，将资金切换至水电（长江电力）及高股息消费（家电、白酒）等纯内需防御资产，以规避输入型通缩风险并锁定稳定现金流。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在美股AI电力赢家与科技成长股存在高Beta重叠且容易被SPCX同步去杠杆的环境下，大类资产配置应向低贝塔的A股水电红利（长江电力）倾斜，并配置核电（中国核电）对冲干旱风险，构建收益…", "summary": "基于2026-06-17，本报告压力测试了SPCX挤压环境下，AI电力稀缺因子与水电红利的配置效率。美股核电IPPs（CEG/VST）对NDX的Beta已飙升至1.08-1.12，存在严重的Beta和动量重叠风险，在去杠杆中易遭受-25.0%的同步回撤。A股水电（长江电力）对NDX的Beta为-0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 3, "neutral": 1, "value": 66.2, "events": [ { "title": "上游能源红利逻辑正被强制切换为'现金流稳定'主导，但估值中枢不能维持当前水平——煤炭可作过渡仓但需-10%中枢下移，油气从核心降级为卫星仓，真正受益者是公用事业型物理资产。", "summary": "油价回落至70-80美元区间剥离了战争溢价，IEA下调2026年全球需求增速至+0.7 mb/d。上游红利逻辑被强制切换为'现金流稳定'主导，但估值中枢不能简单平移：A股三桶油/煤炭龙头股息率虽仍6-8%，盈利-分红覆盖倍数从2022年~2.0×压缩至~1.3×。压力测试显示韧性排序为煤炭>油气>…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储鹰派转向与A股TMT极值两融负Gamma清算风险", "summary": "美联储鹰派转向与A股TMT极值两融负Gamma清算风险 Fed Hawkish Shift vs. A-Share TMT Margin Debt Negative Gamma Risk 在美联储偏鹰暂停、美债利差扁平化以及中芯国际A/H溢价率+103.3%的背景下，国内电子两融余额高悬至5722…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在美债收益率冲高至4.66%与套利交易平仓阴霾下，伴随期权保护成本飙升的“AI挤压”上涨反映了高位派发期的脆弱特征，A股配置应从无保护高杠杆科技向红利与期权保护的防御性哑铃风格重构。", "summary": "截至2026-06-18，美股“有保护的AI挤压”被判定为高位派发前奏而非健康风险偏好。在美联储主席Kevin Warsh首次决议鹰派暂停、美债10Y利率冲高至4.66%以及日本央行意外加息至1.0%引发套利交易平仓风险背景下，SpaceX（SPCX）超大规模IPO上市后回吐涨幅（开盘209.74…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "油价战争溢价消除与实际利率上行背景下的红利资产结构重平衡", "summary": "油价战争溢价消除与实际利率上行背景下的红利资产结构重平衡 Oil Premium Collapse vs. Fed Hawkishness: Redefining Dividend Allocation 地缘缓和致原油暴跌6%，而美债利差收窄、高实际利率压制债市，红利组合是否应向低两融占比（1.8…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 3, "neutral": 1, "value": 67.6, "events": [ { "title": "A股红利杠铃可以继续作为TMT组合防御对冲，但必须把中际旭创式进攻卫星视为强平凸性来源，而不是流动性缓冲。", "summary": "截至2026-06-20，本卡对A股红利与TMT进攻端做微观流动性压力测试。结论是：Research note 05的A股红利杠铃方向仍可支持，但执行容量必须收紧。长江电力更适合作为防御锚，当前可得融资余额占自由流通市值代理值约4.52%，但压力可执行流动性下退出可能拉长至约11.5个交易日。中际…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股红利资产尚未逼近自身杠杆和成交极值，但在全市场两融余额接近历史高位时，应把它们视为可被动提款的流动性资产，并对15.85%红利袖子设置2.0-3.0个百分点的动态减仓触发器。", "summary": "截至2026-06-20，本卡支持红利低波袖子仍可配置，但将风险从“红利泡沫”重定义为“全市场去杠杆时的流动性提款机”。沪深两市2026-06-17两融余额合计2.9438万亿元，接近历史高位；但14只大型银行、大水电和煤炭长协红利样本在2026-06-18的融资余额仅261.87亿元，自由流通市…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在“88元CEA中枢 + 监管缓坡”下，A股最应修复的是被碳成本悬崖错杀的大型高效率火电平台，而不是已拥挤的情绪火电或纯碳价稀缺主题。", "summary": "截至2026-06-19，本卡综合前序结论：若市场把A股发电板块按“5%有偿拍卖立刻落地 + CEA突破130元/吨”定价，则在88元运营中枢和监管缓坡下，最大预期差修复来自大型火电运营商。首选华电国际、华能国际，中国神华作低波动底仓；长江电力和龙源电力是防守/绿电期权，不是一阶碳成本修复；大唐发…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在2026下半年AI算力物理瓶颈与去估值压力下，通过55:45的TMT结构优化并切入10%的A股高股息红利资产，能将组合总贝塔压降至1.732，并利用红利端-0.8%的负边际风险贡献…", "summary": "基于2026-06-20，本报告定量评估了TMT 55:45重构与切入8%-10% A股红利对冲组合。测算表明，在AI算力瓶颈与电力死锁的承压情景下，10%的A股红利对冲（长江电力、中国神华等高股息股）成功使组合总贝塔从1.990降至1.732，10天99% VaR从17.06%压降至14.03%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 1, "neutral": 0, "value": 78, "events": [ { "title": "AI电力瓶颈与绿电利用率分化对A股电力设备、公用事业和新能源运营商的策略影响", "summary": "承接Research note 02–08全链条，本卡以发改能源〔2026〕688号多用户绿电直连（2026-05-20，优先支持算力设施）为基准情形，重构A股电力链定价。核心：688号把AI负荷从挤压新能源并网的拥塞负荷，翻转为在资源地变现弃电的锚定承购（2026Q1光伏利用率90.6%、隐含弃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 6, "neutral": 0, "value": 49.4, "events": [ { "title": "AI电力瓶颈下的跨板块配置：半导体、AI服务器、电网设备、绿电与红利公用事业", "summary": "综合卡片01-04，将AI电力瓶颈定性为局部取电速度而非全国电源缺口，由此给出A股再配置：把权重从定价完美、拥挤的算力栈顶部（前道半导体98.7分位PE、纯AI服务器贝塔）旋转至压缩‘通电机柜时间’的物理交付使能者。超配电网设备（国网十五五4万亿、+40%，最高确信）、液冷（渗透14%→40%，战…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "汇率防线与-274bp倒挂利差倒逼PBOC推迟全面降息以守住7.00关口，结构性货币工具（5000亿技改再贷款）配合贴息可实现163bp的定向融资成本压缩并扭转项目NPV，这导致信用…", "summary": "截至2026-06-22，本报告定量评估了在汇率与负利差双重约束下，中国人民银行（PBOC）的定向与结构性工具组合及其信用传导机制。在10年期中美元债利差倒挂达-274bp及人民币面临7.00红线压力下，PBOC推迟了全面降息（LPR维持不变），转向运用5000亿科技创新与技改再贷款（利率1.75…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在能源价格常态化预期下，红利配置应向清洁电力转移，但须通过煤电组合对冲水电枯水期现货买电风险，并警惕规制电网在AI负荷爆发期的信用评级恶化与表后流失风险。", "summary": "截至2026-06-21，本研究支持将组合红利仓位从周期性煤油向公用事业转移的逻辑。水电（长江电力）在严重枯水（-15%）情景下面临现货回购惩罚，净利润下降33.5%且面临29.7%派息削减风险；核电（中国核电）受益于煤电联动，电价上浮5%可提升利润15.88%，并捕获5.8-8.4分/kWh的碳…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "评估新配置组合（长久期主权债+防御性红利+AI电力瓶颈资产）在利率波动与通胀粘性场景下的波动率与回撤风险", "summary": "结论：会，且差距悬殊。在熊市陡峭化+通胀粘性下，48%长久期主权债是主导风险因子，5%AI电力瓶颈袖套无法抵消——基准情景仅抵消约1%债券拖累、严重情景约3%[自测算]。失效三因：因子错配（电力对冲AI硬件通缩而非利率）、规模错配（5%对48%）、符号反转（规制公用事业是债券代理，与长债同向跌，被…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 1, "risks": 2, "neutral": 2, "value": 88.8, "events": [ { "title": "在公用事业超级资本开支面临股权稀释及大选年监管时滞的双重制约下，跨资产配置应调降拥挤的美股公用事业权益，通过公用事业IG高级债捕获拓宽的信用利差，同时布局低相关的A股水电红利（长江电…", "summary": "基于2026-06-22，本报告综合研判了受监管公用事业费率基数超级周期下的跨资产布局。公用事业1.3万亿美元资本开支带来严重的股权二次发行稀释，且大选年36州州长竞选引发监管机构费率抵制，惠誉调降行业展望至“恶化”。PJM 2027/28容量价暴涨至333.44美元/MW-日，零售电费同比上涨1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月调样后A股电子与红利板块在创新高的两融杠杆与回流 of 北向外资共振下依然维持净买入，所谓的规模差走阔实为短期调样调仓与动量噪音而非真实资金撤离，但历史高位杠杆要求通过跨资产多元…", "summary": "截至2026-06-23，6月调样后A股规模差异并非由真实资金撤离驱动。月中两融余额大幅反弹915.10亿元，于6月18日触及29655.12亿元历史新高，电子板块获430亿元融资流入；6月22日北向交易额达4821.77亿元，表明内外资正共振流入硬科技与红利板块。但高杠杆下拥挤度高企。压测显示，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在海风开发商回款账期延长与物理确认时滞下，A股对东方电缆（高负债杠杆压制）与亚星锚链（交付时滞真空）进行了估值收缩定价，传统的成长重仓策略失效，向具备即期现金流变现能力的水电与海运红…", "summary": "本报告对开发商账期拉长下东方电缆与亚星锚链的流动性与物理交付折价进行了A股定价逻辑压力测试。分析表明，东方电缆重资产、高资金占用模式受制于垫资压力，在重度承压下经营现金流转负至-31.73亿元，导致PE估值中枢被压缩至14.5倍；亚星锚链虽现金流安全，但142天物理交付验收时滞使其财务表现呈业绩真…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力价值链中的\"成期权\"：核电、水电及SMR对公用事业估值重塑的影响", "summary": "本卡支持卡08的\"类债→成长\"切换，但收紧为一次三层分部拆解：每家大型公用事业应拆成（i）稳态年金（类债DDM）、（ii）已签20年期算力PPA（长久期年金）、（iii）稀缺电力看涨期权（实物期权溢价）。证据：PJM容量电价从约29升至329.17美元/MW·日FERC上限[S5]；中国核电在运2…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 5, "neutral": 0, "value": 58.2, "events": [ { "title": "信用风暴的核心不是现金总量不足，而是中小银行退出低评级城投买盘后，非银必须以更高利差、更低杠杆和更高流动性缓冲重新定价地方信用。", "summary": "截至2026-06-23，本卡支持前三张卡片的压力方向，但判断外溢路径更像“流动性质量分层”而非系统性资金断裂。2000亿元AA级及以下城投买盘缺口为未公开验证的压力输入；相对2026年5月末11.67万亿元债基和16.27万亿元货基规模，总量可吸收，但边际买方撤出会触发弱AA利差、非银净值和回购…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在成本压力积聚与零售利润敏感性增加的宏观环境下，红利组合必须对运营租赁类REITs实施动态配置上限管理，向原材料独立的公用事业高股息股与国债倾斜，以保障组合总回报。", "summary": "针对成本冲击对折扣零售REITs分红稳定性的威胁，本报告提出红利防御组合优化框架。将REITs划分为高贝塔商业零售型（A类）与稳健特许经营型（B类）。在成本压力下，将A类REITs配置上限由15%下调至8%（中度）和3%（重度），增配水电红利股（长江电力 600900.SH）与10年期国债。模拟表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "清洁能源杠铃应按等风险贡献而非资本权重构建（短债占约70%资本、各分支均担25%风险），并外挂年化约60bp Theta drag的日元套利平仓尾部对冲——其代价应以全天候夏普（0.…", "summary": "基于2026-06-23，本卡将首席策略师的防御杠铃论点量化为等风险贡献（ERC）预算。因短端国债波动率仅约2.5%、公用/红利股为16-20%，25%等风险贡献下资本权重约为长江电力11%、中国核电8.8%、大行/电信红利10%、短端国债70%，组合波动率约4.8%、基准夏普0.44。真正的剩余…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在美联储非对称紧缩导致股债相关性逆转为正的滞胀压力下，防滞胀K型配置与大宗商品/红利股战术杠铃组合是唯一能提供稳定绝对收益与低波动风险垫的资产配置方案。", "summary": "截至2026-06-23，本卡评估了防滞胀K型配置（40%大宗商品/40%短债TIPS/20%现金）与战术杠铃组合在美联储非对称紧缩下的表现。回测显示，在通胀脉冲与股债正相关（+0.45）压力下，传统60/40与风险平价崩溃（分别下跌-10.40%和-8.60%）；而K型配置与战术杠铃分别录得+9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 1, "neutral": 0, "value": 83.6, "events": [ { "title": "绿电分时匹配和CEA有偿比例提高能把可调节清洁电力转化为可测算的盈利与估值溢价，但该溢价只能提供相对对冲，无法单独抵御A股成长股系统性融资去杠杆。", "summary": "截至2026-06-25，本卡片压力测试绿电分时匹配与碳配额收紧下A股公用事业稀缺资产的估值弹性。结论是：可调节水电和具备储能、绿电直连或数据中心PPA的绿电运营商确有基本面溢价，基准测算约27元/MWh，可带来约18.2%的盈利弹性；若叠加ERP收缩，公允价值可提升约35.9%。但在A股成长股融…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 8, "neutral": 0, "value": 43, "events": [ { "title": "2026-06-26确认科技两融堰塞湖进入去杠杆阶段，但未确证牛市陷阱；最优策略是降拥挤科技、保现金缓冲、分批增配电力电网等物理红利资产。", "summary": "截至2026-06-27，结论是支持防守再平衡，但不把2026-06-26下跌直接视为“国家队牛市陷阱”的确证。A股两融余额已升至3万亿元以上，电子行业两融余额超过6000亿元且占比约两成，说明风险是融资拥挤和流动性冲击，而非科技周期被证伪。建议科技从超配降至标配，优先削减融资敏感电子/CPO仓位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力电网及公用事业红利资产的配置逻辑验证", "summary": "支持第01张卡片的防守再平衡判断，并验证其机制：本轮去杠杆是两融融资拥挤的电子板块冲击（A股两融2026-06-23破3万亿元），而受监管公用事业由保险/养老等长钱持有、融资拥挤度低，处于压力源的对立端，是真正避风港而非平行泡沫。需求端1-5月用电+3.4%、互联网数据服务用电+26%给出AI电气…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股市场结构约束下的清洁基础设施与能源对冲执行方案", "summary": "截至2026-06-26，本卡把Card03杠铃(12%清洁+4%能源)与Card04子仓上限落地为可成交的A股方案。核心结论：约束不难，难在表达载体可得性高度不对称——电网有唯一高流动纯标的(159326,约¥343.7亿,特高压权重约66%)；电池有百亿级但偏电动车(159755,约¥158.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网延迟的变压器库存积压是可逆的时点拖累(金盘-8%~-11%、思源<-3%)，被国网4万亿内需托底，应在超配出口变压器框架下偏思源、轻金盘并叠加配对与公用事业杠铃对冲。", "summary": "并网延迟导致的已完工未交付变压器库存积压，是营运资金/收入确认的时点事件而非需求毁灭——订单仍排产至2027年在手backlog内。对金盘科技(688676.SH)拖累真实但有界：北美AIDC相关收入是最高Beta切片，Q2/Q3净利约-8%至-11%、伴随存货与FCF占用，属逢低布局而非证伪；对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 7, "neutral": 0, "value": 50.4, "events": [ { "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股“大行+红利/稀缺成长”哑铃配置在低杠杆与高可视度Backlog支撑下并未拥挤，而弱区域中小银行因资产质量二阶传导滞后与重度稀释风险，其资本金脆性尚未被市场充分定价。", "summary": "本卡片对A股“大行+红利/稀缺成长”哑铃配置拥挤度及非核心中小银行风险定价进行了量化检验。结果显示哑铃配置自身极低杠杆（大行0.40%、长江电力1.81%）使其整体处于安全区间（CCI为34.2-55.3），未发生拥挤。相反，非核心中小银行在+2.5%不良冲击下面临1.98万亿元资本缺口，现有中小…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "5.5%银行/10.5%公用事业电力的新红利袖部在-5%最大回撤预算下的最终联合压力测试", "summary": "联合尾部（CNH>7.00、银行再跌12%、公用事业/电力回撤9%、黄金同步失效）下，组合建模回撤为中性−2.72%、严重−3.64%、极端−4.69%[自测算]，三档均守住−5%预算，但极端变体仅剩0.31pp余量。关键发现：尾部最大贡献项是黄金（极端−1.96%），约等于银行（−0.77%）加…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "人民币汇率稳定对 A 股红利低波风格持续性的驱动作用", "summary": "本卡由 chief-strategist 撰写，支持卡片 03/04 的链条：守住 USD/CNY 6.80-6.82（即期约 6.7985）迫使人行约束价格型降息，把 10 年期国债收益率钉在约 1.73% 的地板上。地板曲线令股息-债券利差最大化——中证红利低波指数股息率约 5.0%（利差约 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 6, "neutral": 1, "value": 68.8, "events": [ { "title": "局部中小银行信用缺口下的A股银行红利耐久性与高股息内部轮动", "summary": "卡08分离出的“额度沉睡”机制（CET1上限+NPL触发+低于1.40%净息差）不仅是中小银行信用风险，更是撕裂A股红利复合体的力量。重排结论：①大行红利居顶——分红不依赖沉睡额度，约4.35%股息率、对10年国债（约1.72%）257bp利差、0.5–0.7倍PB已计入坏消息，持有/超配；②公用…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "债市收益率触底倒逼绝对收益资金流向A股优质高股息红利股建立‘红利锚’，在‘数量先行、价格被锚’的结构性宽松下，配置策略应向国有大行与电力设备出口链的对冲型哑铃结构倾斜。", "summary": "本卡片合成分析了国内债市10Y国债至1.73%、信用Carry枯竭下的资产荒对A股的外溢效应。外溢主要通过险资等绝对收益资金流向A股红利资产（利差达+3.27%至+3.365%），形成局部红利锚。在价格降息受限、数量宽松先行的政策组合下，长久期成长股缺乏催化剂，建议实施红利大行防御与硬科技出口链主…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费成本转嫁挤压。", "summary": "基于国产芯片2.94倍能效赤字拉长通电时间至18个月的物理瓶颈，本报告系统性量化了A股电力设备板块（特高压、大变压器）的估值扩张机制与低利润 traditional TMT 的电价挤压。电网容量受限与十五五规划共振，支持了特高压订单可见度与大型变压器出海高溢价，特别是在国内外硅钢5.1倍巨大价差的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电力’，并驱动策略从算力规模向算法…", "summary": "截至2026-06-28，本卡片通过对电网与变压器交期的压力测试，确认电网并网（5-7年交期）和变压器交付（2-4年交期）构成了AI算力扩张的终极物理边界。随着GB300（约135kW）及Rubin Kyber（约600kW）对800VDC和固态变压器（SST）的高密度配电需求增加，目前仅有不足8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 1, "risks": 4, "neutral": 3, "value": 88.2, "events": [ { "title": "红利龙头去拥挤与超跌成长板块反弹的风格切换大论战", "summary": "红利龙头去拥挤与超跌成长板块反弹的风格切换大论战 A-Share Style Rotation: Dividend De-crowding vs. Growth Rebound 长江电力（600900.SH）资金大流出与药明/恒瑞大涨，是否标志着A股向超跌成长风格的实质性切换已在午间确立？伟星股份…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 高拥挤大行红利资产减仓与资金流向评估", "summary": "致 金融行业分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: dede69e0-4ffc-4ff4-b119-7912477831ae 结果摘要: I will run the date command in the shell to anchor today's da…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "A股风格大轮动：大盘拥挤红利瓦解与成长股超跌反弹的持续性辩论", "summary": "A股风格大轮动：大盘拥挤红利瓦解与成长股超跌反弹的持续性辩论 A-Share Style Rotation: Dividend Collapse vs. Growth Rebound Sustainability 长江电力资金大流出与药明/恒瑞大涨，是否标志着A股向成长风格的系统性切换已经确立？中…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "拥挤红利去杠杆与医药生物估值脱钩反弹的风格切换", "summary": "拥挤红利去杠杆与医药生物估值脱钩反弹的风格切换 Style Rotation: Utilities De-crowding vs. Biotech Rebound 长江电力等红利拥挤度处于95%历史极值，且股息率利差压缩至历史低点，而在跨境分段生产解禁及创新药商业化兑现下医药板块大涨，这是否标志着…", "sourceLabel": "研究主题", "supportLabel": "支持", "supportType": "supports", "href": "" } ] } ], "href": "/investor/theses/thesis-70daf87262/", "networkHref": "/investor/research-network/#thesis-70daf87262" }, { "id": "thesis-58200390b4", "source": "tracked_thesis", "title": "加密资产仍是全球流动性和散户风险偏好的高频回声", "titleEn": "Crypto remains a high-frequency echo of global liquidity and retail risk appetite", "lane": "加密/高波动风险偏好", "laneEn": "Crypto and high-beta risk appetite", "laneHref": "/investor/lanes/crypto-and-high-beta-risk-appetite/", "investableFocus": "加密/高波动风险偏好：COIN, MSTR, MARA, RIOT, IBIT, HOOD", "stockUniverse": [ { "ticker": "COIN", "name": "Coinbase", "market": "US", "role": "加密交易基础设施", "roleEn": "加密交易基础设施", "bucket": "core", "href": "/investor/stocks/COIN/" }, { "ticker": "MSTR", "name": "MicroStrategy", "market": "US", "role": "BTC 高 beta 代理", "roleEn": "BTC 高 beta 代理", "bucket": "core", "href": "/investor/stocks/MSTR/" }, { "ticker": "MARA", "name": "MARA Holdings", "market": "US", "role": "比特币矿企", "roleEn": "比特币矿企", "bucket": "core", "href": "/investor/stocks/MARA/" }, { "ticker": "RIOT", "name": "Riot Platforms", "market": "US", "role": "比特币矿企", "roleEn": "比特币矿企", "bucket": "core", "href": "/investor/stocks/RIOT/" }, { "ticker": "IBIT", "name": "iShares Bitcoin Trust", "market": "US ETF", "role": "BTC 现货代理", "roleEn": "BTC 现货代理", "bucket": "core", "href": "/investor/stocks/IBIT/" }, { "ticker": "HOOD", "name": "Robinhood", "market": "US", "role": "散户交易风险偏好", "roleEn": "散户交易风险偏好", "bucket": "core", "href": "/investor/stocks/HOOD/" }, { "ticker": "QQQ", "name": "Invesco QQQ Trust", "market": "US ETF", "role": "纳斯达克成长暴露", "roleEn": "纳斯达克成长暴露", "bucket": "watch", "href": "/investor/stocks/QQQ/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "COIN", "MSTR", "MARA", "RIOT", "IBIT", "HOOD", "QQQ", "SMH", "TSLA", "ASTS", "MU", "005930.KS", "APO", "IEF" ], "coreView": "赛道：加密/高波动风险偏好。候选观察标的：COIN, MSTR, MARA, RIOT, IBIT, HOOD。当前共有 40 条证据、7 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 81, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 7, "risks": 136, "neutral": 17, "analystCount": 13, "sourceTypeCounts": { "研究笔记": 151, "研究线程": 3, "分析协作": 3, "研究主题": 2, "链条事实": 1 }, "topTerms": [ "HBM", "GPU", "AIDC", "Crypto", "remains", "high-frequency", "echo", "global", "liquidity", "retail", "risk", "appetite" ], "whatChanged": [ "2026-06-29: AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "2026-06-29: 电力成本不确定性会直接拖慢西部AI算力基地投产节奏，但主要通过容量/并网/备用费用和到电时间压低Capex效率，而不是通过单一电价水平否定项目。", "2026-06-29: 变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "2026-06-29: 物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进封装国产替代材料集中。", "2026-06-29: 九卡线索数值连贯、引用完整、叙事贯通，分析扎实无需重做；仅一项 P0（Cards 01/02 缺 2026-06-30 工作日期锚）卡住发布，评级 NEEDS-REVISION（轻度），清掉 P0 即可 PASS。" ], "practical": { "sectorLane": "加密/高波动风险偏好", "sectorLaneEn": "Crypto and high-beta risk appetite", "exposureBucket": "风险偏好与高波动", "tone": "red", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "COIN", "name": "Coinbase", "market": "US", "role": "加密交易基础设施", "roleEn": "加密交易基础设施", "bucket": "core", "href": "/investor/stocks/COIN/" }, { "ticker": "MSTR", "name": "MicroStrategy", "market": "US", "role": "BTC 高 beta 代理", "roleEn": "BTC 高 beta 代理", "bucket": "core", "href": "/investor/stocks/MSTR/" }, { "ticker": "MARA", "name": "MARA Holdings", "market": "US", "role": "比特币矿企", "roleEn": "比特币矿企", "bucket": "core", "href": "/investor/stocks/MARA/" }, { "ticker": "RIOT", "name": "Riot Platforms", "market": "US", "role": "比特币矿企", "roleEn": "比特币矿企", "bucket": "core", "href": "/investor/stocks/RIOT/" }, { "ticker": "IBIT", "name": "iShares Bitcoin Trust", "market": "US ETF", "role": "BTC 现货代理", "roleEn": "BTC 现货代理", "bucket": "core", "href": "/investor/stocks/IBIT/" }, { "ticker": "HOOD", "name": "Robinhood", "market": "US", "role": "散户交易风险偏好", "roleEn": "散户交易风险偏好", "bucket": "core", "href": "/investor/stocks/HOOD/" }, { "ticker": "QQQ", "name": "Invesco QQQ Trust", "market": "US ETF", "role": "纳斯达克成长暴露", "roleEn": "纳斯达克成长暴露", "bucket": "watch", "href": "/investor/stocks/QQQ/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "COIN", "MSTR", "MARA", "RIOT", "IBIT", "HOOD", "QQQ", "SMH", "TSLA", "ASTS", "MU", "005930.KS", "APO", "IEF" ], "primaryStocks": [ { "ticker": "COIN", "name": "Coinbase", "market": "US", "role": "加密交易基础设施", "roleEn": "加密交易基础设施", "bucket": "core", "href": "/investor/stocks/COIN/" }, { "ticker": "MSTR", "name": "MicroStrategy", "market": "US", "role": "BTC 高 beta 代理", "roleEn": "BTC 高 beta 代理", "bucket": "core", "href": "/investor/stocks/MSTR/" }, { "ticker": "MARA", "name": "MARA Holdings", "market": "US", "role": "比特币矿企", "roleEn": "比特币矿企", "bucket": "core", "href": "/investor/stocks/MARA/" }, { "ticker": "RIOT", "name": "Riot Platforms", "market": "US", "role": "比特币矿企", "roleEn": "比特币矿企", "bucket": "core", "href": "/investor/stocks/RIOT/" }, { "ticker": "IBIT", "name": "iShares Bitcoin Trust", "market": "US ETF", "role": "BTC 现货代理", "roleEn": "BTC 现货代理", "bucket": "core", "href": "/investor/stocks/IBIT/" }, { "ticker": "HOOD", "name": "Robinhood", "market": "US", "role": "散户交易风险偏好", "roleEn": "散户交易风险偏好", "bucket": "core", "href": "/investor/stocks/HOOD/" } ], "instrumentTags": [ "COIN", "Coinbase", "MSTR", "MicroStrategy", "MARA", "MARA Holdings", "RIOT", "Riot Platforms", "IBIT", "iShares Bitcoin Trust", "HOOD", "Robinhood" ], "reviewCadence": "每日复核", "confirmationSignals": [ "期权伽马和 ETF 资金是否放大尾盘流动性", "社交热度是否转化为真实增量买盘", "波动率是否领先基本面恶化" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "波动率", "现金", "指数保护", "期权拥挤" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 106 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 13, "recentEvents": 106, "supportRatio": 0.04, "riskRatio": 0.85, "conflictScore": 95.7, "evidenceDepth": 100 }, "score": 73.4 }, "events": [ { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19.75, "matchedTerms": [ "HBM", "GPU", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16, "matchedTerms": [ "AIDC", "ai-infrastructure", "power" ], "href": "" }, { "date": "2026-06-29", "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power预期缺口，会迫使Hyperscaler预付并网升级、锁定变压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16, "matchedTerms": [ "GPU", "ai-infrastructure", "power" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "risk", "evidence", "dominant", "inflation", "transmission", "supply" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性TAM上限，营收放缓促使估值压缩。相反，全球CPO光模块（中际…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "GPU" ], "href": "" }, { "date": "2026-06-29", "title": "九卡线索数值连贯、引用完整、叙事贯通，分析扎实无需重做；仅一项 P0（Cards 01/02 缺 2026-06-30 工作日期锚）卡住发布，评级 NEEDS-REVISION（轻度），清掉…", "summary": "对 Cards 01–09 做发布前 QA，评级 NEEDS-REVISION（轻度），距 PASS 仅一个修订周期。数值主干一致：HBM 良率 4.11–16.98%→GPU 23,125–95,537 颗→23.12–95.54 EFLOPS 在 Card04↔05 完全吻合；5.5% 红利为 Card05 区间中点；225bp→265bp 是 Car…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 15.5, "matchedTerms": [ "HBM", "GPU" ], "href": "" }, { "date": "2026-06-29", "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-560亿美元收入，其25%低毛利率根本无法对冲高利润半导体供应链…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "GPU" ], "href": "" }, { "date": "2026-06-29", "title": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "summary": "截至2026-06-30，本报告压力测试了变压器交期拉长对AI算力集群部署和云厂商资本开支的约束。变压器交期拉长（128-144周，专用型4-5年）导致物理并网时滞达4-7年，引发全球467万个GPU延迟并网和2万亿元Capex延期。微观上，1.5年延迟导致10亿美元GPU集群产生5.2295亿美元折旧与持有损耗，月度拖累1.1905亿美元，NPV骤降12…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "GPU", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "效率驱动的算力架构转型：从“暴力规模”到“高能效密度”的物理基座适配性评估", "summary": "压力测试 Research note 08 的隐含前提：效率并不缓解物理天花板，而是把它从电网层下移到机架与机柜行层并加剧。机理是机架尺度的杰文斯悖论——单瓦产出靠提高密度实现，机架功率从约40kW(Hopper)→120kW(Blackwell)→142kW(GB300,90%入液)→600kW(Kyber,2027)迈向1MW，对照2026均值约27k…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19.75, "matchedTerms": [ "HBM", "GPU", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM合约价Q2环比+58–63%)、以及电网电力(美国并网排队约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19.75, "matchedTerms": [ "HBM", "GPU", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "数据中心电网并网瓶颈与核能转向", "summary": "截至2026年6月28日，AI基础设施建设的核心瓶颈已从GPU稀缺转移到电网并网时间（Time-to-Power）。对100 MW超大规模AI数据中心的模型测算显示，2年的并网延迟会减少项目35.73%的净现值（NPV），并使IRR大幅下降。为规避此类延迟，科技巨头（微软、亚马逊、谷歌、Meta）在2024至2026年间锁定了超十吉瓦（GW）的核电容量，通…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16, "matchedTerms": [ "GPU", "ai-infrastructure", "power" ], "href": "" }, { "date": "2026-06-28", "title": "<20%开机率红线是平台属性而非物理定律：对电力受限EO小卫星它由电源+被动散热共同锁定且先进热控不划算(支持卡03)，但两相泵驱回路+可展开辐射器已能让专用型150kW级在轨算力平台突破红…", "summary": "本卡按平台维度厘清半导体卡的<20%开机率红线，结论为有条件支持。对Planet/BlackSky类电力受限EO小卫星，红线由电源预算与被动导热双重锁定：15-60W的GPU在散热生效前已占150-300W总线的10-40%，加装两相回路+辐射器(约2-4kg、5-15W泵耗)成本收益为负，红线成立。但对专用型在轨算力平台(SpaceX AI1 150kW…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "GPU", "Lane", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "国产大模型并未告别规模竞赛，而是在先进GPU受限的条件下把竞争单位改写为每瓦、每token、每受限加速器和每留存用户的有效智能产出。", "summary": "截至2026-06-28，本卡对“国产大模型不可逆转向效能极限竞赛”进行压力测试。结论是：效能迁移真实且受GPU出口、H20许可、HBM/互连、液冷机柜和电力约束推动，但不是规模竞赛终结，而是进入“稀缺规模竞赛”。DeepSeek-V3、DeepSeek-R1、Qwen3和MiniMax-M1显示国产模型正通过MoE、FP8、RL、蒸馏、长上下文和推理预算…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "GPU" ], "href": "" }, { "date": "2026-06-28", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,600 亿）以经营现金流支撑，资金基础/地域/标的均不同于承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "GPU", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "超大规模厂商通过自备能源与地理外迁绕过电网瓶颈的对策存在严重的监管、燃料管道、物理时延与设备交付制约，难以扭转通电滑点导致的资本效率受损与资产减值风险。", "summary": "截至2026年6月28日，本报告对超大规模云厂商应对电网延迟的对策（BTM核电/气电、地理迁移及PPA）进行压力测试。分析指出，AWS-Talen等BTM核电直连模式已被监管层实质性锁死；自备燃气发电面临管道运力与发电机设备交期双重瓶颈并违背ESG目标；训练与推理的物理时延限制了集群外迁的可行性；12-18个月通电延迟将导致严重GPU资产闲置与项目级债务压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "GPU", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "PQC在DPU端的硬件硬化并非无代价，其引起的局部网卡热降频与片上内存溢出将导致重大GPU空转间接能耗损失与RDMA亚微秒级延迟恶化，成为集群能效比与拓扑扩展性的真实物理边界。", "summary": "本卡片对DPU端硬化PQC算法进行压力测试。PQC的直接计算功耗虽小，但开启硬件引擎带来的局部热边际增量（每网卡约10W）极易使高密度PCIe卡槽中的网卡触发结温保护而下频。由此导致的 collective 通信降速使10万卡集群的 MFU 下降5-10%，间接能效浪费高达5-10MW（直接增量千倍）。同时，ML-KEM的大密钥尺寸占用高达37.9MB缓存…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "GPU", "ai-infrastructure" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 2, "neutral": 0, "value": 30, "events": [ { "title": "AI算力行情是超级周期中段，三波叠加需求（训练→推理→主权AI）驱动力完整，供给瓶颈维持定价权，短期情绪风险真实但不构成周期终结信号。", "summary": "AI硬件行情处于超级周期中段，而非利好出尽的顶部。三波需求（前沿模型训练、大规模推理、主权AI/企业部署）正叠加复利，供给侧CoWoS封装、HBM4产能及电力基础设施瓶颈在12–18个月内无法解除，定价权保持完整。历史类比指向1997–1998年的互联网基础设施阶段。战术层面，情绪高涨（恐惧贪婪指…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 年 HBM4 推理链条的首要弹性在合格 HBM4，首要出货瓶颈在 HBM4 验证与 CoWoS/interposer/封装基板，而液冷更多是部署瓶颈。", "summary": "本卡支持前序推理需求上行逻辑：288 GB HBM4 GPU 普及后，最大量价弹性在合格 HBM4 供应商，尤其是具备份额提升弹性的 MU 和具备配额权的 000660.KS；但最可能成为 2027 物理出货瓶颈的是 HBM4 验证与 CoWoS/interposer/封装基板的组合。测试设备是高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "2026下半年AI资本开支拐点辩论：消化期还是再加速？", "summary": "TMT分析师判定，2026下半年AI资本开支并未进入消化期，而是进入结构轮动期：总量持平至上行，边际增量美元从训练优化的商用GPU集群转向推理fleet、超大规模云厂商自研芯片、以及电力与场地准备。空头叙事在需求侧（推理token volume、agentic渗透、多模态）证据不足，但在前沿训练边…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电子取代 HBM 成为 AI 链条上未来 12 个月最被讨论的瓶颈：1 GW 边际 AI 园区的电费在 2027 年从 $322M/年跳到 $926M/年，足以让 hyperscal…", "summary": "对 Research note 03 的 $1T capex / $7.5–10M/MW 营收密度救援叙事进行压力测试。结论：电力（不是 GPU）成为 2026 下半年起的硬约束；hyperscaler 在 2026–2027 锁定的边际 MW 全成本将达 $95–140/MWh，相对 2024 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 0, "neutral": 1, "value": 47.8, "events": [ { "title": "NEV 60% 渗透率:AI 算力第三支柱 vs 分母陷阱的统一刻度", "summary": "NEV 60% 渗透率:AI 算力第三支柱 vs 分母陷阱的统一刻度 ADAS 算力(车规 GPU、域控、HBM 接口)与电池材料(LCE、正极)的需求曲线,能否在 60.6% 渗透率前后被独立定价?PHEV/EREV 稀释对两条曲线的影响系数是多少?2027 年合同口径下 NEV 算力订单到底覆…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 3, "neutral": 0, "value": 28.2, "events": [ { "title": "AI算力增长正撞向“电力天花板”，半导体板块技术面极端过热且基本面面临“资本开支消化期”风险，支持减仓芯片转向电网。", "summary": "截至2026-05-14，本卡通过压力测试支持大幅减持SMH至5%并保持DRAM 8%战术卫星的决定。核心逻辑是AI算力瓶颈已从“硅片供应”转向“电力并网”，美国和欧洲电网并网等待时间已超5年，导致“暗壳”数据中心（有GPU无电力）现象频发。半导体周期已入成熟期，SOX指数偏离200日均线60%，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力期货压缩的是商用GPU层租金，先进封装与高带宽互联的物理瓶颈与寡头结构足以将其非商品化毛利结构维持至2027。", "summary": "支持card-01框架并加以收紧：算力期货标准化的是GPU-小时这一输出单位，并未商品化决定谁能以最低成本清算的物理输入。CoWoS先进封装、HBM4、混合键合设备、scale-up交换机ASIC、PCIe/CXL retimer、1.6T/CPO光模块均具18–36个月结构性导入周期，寡头格局清…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI单名信用利差走宽主要是供给+杠杆+利率的周期性现象，半导体设备订单簿与超大规模资本开支指引仍坚挺，尚不构成成长逻辑的结构性破坏。", "summary": "压力测试卡片03的预警：截至2026-05-15，AI单名利差走宽约75%可归因于周期性/技术性因素（新发让利40%、资本开支融资杠杆漂移25%、利率beta 10%），仅约25%是早期基本面（ROIC/GPU折旧诚信）。Big-5 2026年AI资本开支合计约2900亿美元同比+45%，无一家削…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 4, "neutral": 0, "value": 23.6, "events": [ { "title": "A股半导体重估的 beta 已基本被定价，下一段必须靠后道设备、量检、材料、车规功率与先进封装上游材料的盈利与订单兑现，回避换手/融资/解禁三重拥挤的设计、代工与映射封装。", "summary": "截至 2026-05-15，A股半导体经过 4-5 月重估后，国产 GPU/ASIC 概念、HBM/先进封装映射股、大市值晶圆代工四条线已拥挤：5 日换手率分位 81-92%，半导体两融占自由流通市值比 ~4%（88 分位），主要芯片 ETF 4 周连续净申购 +18%，且 6-8 月解禁集中（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GXP向CSA的制度松绑只是把交付瓶颈下移到芯片配额与高压电力装备交期，物理瓶颈已替代制度瓶颈成为AIDC交付的新主导变量。", "summary": "CSA转型确实把特区内AIDC交付压缩到9–12个月，但约束随即从制度排队迁移到物理输入：>500 MVA变压器交期110–140周、CoWoS-L先进封装2026年75–80%被NVDA/AMD/Google预订、HBM3E为价格弹性最低的卡点。'交付悖论'未消失，只是从监管时间转为BOM时间。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "推理侧ASIC的规模化替代已导致通用GPU定价权削弱，NVIDIA面临物料成本上升与推理市场份额流失的双重压力，2026Q3毛利率下修逻辑成立。", "summary": "调研显示，2026年5月主要云服务商（CSP）的自研推理ASIC已进入3nm成熟期。Google TPU v6部署量超160万颗，Meta MTIA v3预计年底覆盖35%推理机群，微软Maia 200已支撑GPT-5.2。ASIC相比通用GPU可降低30%-50%的TCO，这导致推理定价权从NV…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Summit De-risking", "summary": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 502dfe56-3d6c-4017-933c-df9f9b15f88b 结果摘要: Wrote handoff_response.md (8242 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 2, "neutral": 0, "value": 37, "events": [ { "title": "H200对华许可的毒丸条款使其对中国云厂只能扮演过渡角色，对跨境AI硬件营收偏空、对国产替代链（SMIC/昇腾/寒武纪）结构性偏多，板块定价应反映'强制分化'而非'复苏催化'。", "summary": "H200对华许可叠加25%关税与BIS领土审计构成'毒丸'结构：单芯片承诺、现场审计、芯片级遥测、再出口禁令使H200对中国云厂成为高摩擦、不可转售、用途受限的过渡载体而非战略基石。我们下修Nvidia 2026年中国数据中心营收至80-100亿美元（vs.一致预期120-140亿），毛利率拖累5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链基本面与订单流：对减仓建议的支持性验证 日期： 2026 05 17 分析师： TMT行业分析师 [Primary] 对Research note 01立场： Suppor…", "summary": "AI算力链基本面与订单流：对减仓建议的支持性验证 日期： 2026 05 17 分析师： TMT行业分析师 [Primary] 对Research note 01立场： Support（支持） 结论： 基本面边际开始出现裂缝——印证因子拥挤度去风险建议 1. 摘要 Research note 01…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 2, "neutral": 0, "value": 38.4, "events": [ { "title": "2026-05-18核心判断：国产变压器12个月交付是AIDC time-to-power的必要条件，但只有算电同步项目才能把它充分转化为算力，弱项目会被局部电网调度和低上架率吸收。", "summary": "2026-05-18：本卡片支持前序国产电力装备供给优势，但压力测试“12个月变压器交付=算力产出”的线性假设。结论是：交付速度只解决第一道门，真实算力还取决于并网、供电质量、绿电/直供结算、GPU与客户负载。枢纽+直供项目转换概率约60%-75%，拥挤城市电网项目约35%-55%，投机性项目低于…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件与算力基础设施具备足够的选择性资本确定性，但应买入瓶颈环节和已验证国产替代，而不是泛AI主题贝塔。", "summary": "截至2026-05-18，本卡片支持从大金融防御性收缩转向盈利可见成长。AI硬件具备选择性资本确定性：NVIDIA、TSMC、SK hynix及800G/1.6T光模块已体现高毛利兑现，云厂商2026资本开支提供需求锚。国产化替代真实存在，但确定性集中在服务器、光模块、热管理、电源、PCB和已商业…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "2027-2028年电力缺口不会终结AI芯片周期，但会把下一代AI硬件迭代逻辑从单芯片峰值性能改写为受限MW下的有效AI输出、上线时间和机架级能效。", "summary": "截至2026-05-18，本卡支持前序电力瓶颈主线，但压力测试“硬件从极致性能转向能效”的表述：更准确地说，采购函数会转向每已通电MW的有效AI输出。2027-2028年表后燃气仅4-6GW，无法覆盖AI负荷缺口；DOE/LBNL预计美国数据中心用电2028年达325-580TWh。NVIDIA …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 0, "neutral": 0, "value": 61.2, "events": [ { "title": "若 2026-06-11 验证 Waymo 单位经济，AI 基础设施不应被卖出，而应从拥挤算力前端轮动到电源液冷、networking 邻近环节与小仓位机器人硬件 beta。", "summary": "截至2026-05-19，本卡支持 6 月 11 日后从 AV 扩散到 AI hardtech，但只在 Waymo 经济性披露有用时执行。确定性最高的是电源/液冷/机柜、定制芯片与 networking、HBM/先进封装；估值弹性最好的是电源热管理与机器人端侧硬件。HBM、GPU、A 股光模块和主…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 4, "neutral": 0, "value": 30.6, "events": [ { "title": "国产AI效率、软件栈和推理优先部署可以抵消应用层硬件缺口的大部分边际压力，但无法消除前沿训练和生态成熟度折价；投资应聚焦高利用率推理软件与能按结果变现的应用主题。", "summary": "2026-05-20：本卡综合判断，国产AI效率路线能显著缓冲应用层硬件约束，但不能完全替代前沿训练硬件、HBM、先进封装和CUDA式生态成熟度。DeepSeek/Qwen的MoE、thinking budget、KV-cache复用、量化和推理服务框架，把关键问题从“是否有足够GPU训练最大模型…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源闸门 E1–E4 不会下修 2026 年 GPU/HBM 订单簿，但会使 2027 年集群投产推迟 1–2 个季度并将超大规模云 capex 结构向电力基础设施倾斜，因此正确响应…", "summary": "截至 2026-05-20，本卡从 TMT 自下而上角度支持卡 06 的能源闸门框架，并将其细化到 GPU/HBM/系统集成层：E1（重型燃机售罄至 2028–2030）与 E2（PJM 容量拍卖触顶）已事实上预先触发，但传导不是 2026 年 GPU/HBM 砍单，而是 2027 年机柜投产推迟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026H2 的正确配置不是半导体转 SaaS，而是半导体核心扩展：保留 GPU/HBM 周期，同时用电力设备、冷却、AVGO 桥接资产与中国 UHVDC 对冲 2027 投产瓶颈。", "summary": "截至 2026-05-20，本卡综合前序结论：不做半导体到 SaaS 的板块轮动，而是在卡 05 的 65% 半导体 / 35% 软件框架内，将 5-7% 从纯 AI 加速器调入 AI 电力基础设施综合体。建议 2026H2 AI 增长仓位为 58.5% 纯半导体、35.0% 软件、6.5% AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链盈利可见性已被订单簿三重验证、足以支撑保留成长端的策略；半导体设备应拆分成‘AI直接受益+国产替代超配，纯中国敞口设备与晶圆厂低配’，故第03卡‘不全面切向价值’成立但需要…", "summary": "截至2026-05-21，本卡对第03卡“不全面切向价值”进行压力测试，结论是有条件支持。AI算力链（GPU/HBM/光模块/液冷/先进封装）盈利可见性高且仍在抬升：四大美系hyperscaler 2026年合计资本开支指引上修至USD 360–410 bn、TSMC CoWoS与SK Hynix…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 3, "neutral": 0, "value": 38, "events": [ { "title": "在英伟达 Rubin 平台面临供应链瓶颈的窗口期，国产 GPU 凭借先进封装与 HBM3 的全产业链突破，正通过‘十五五’初期智算中心的大规模采购实现对推理市场的实质性支配。", "summary": "截至2026-05-22，本研究确认国产GPU已进入性能与市场双重拐点。NVIDIA Rubin平台受困于HBM4验证延迟与CoWoS-L稳定性，2026年产量目标下调25%。与此同时，海光深算4号与寒武纪思元690通过Chiplet技术实现性能跃升，结合长电、通富在2.5D封装的突破以及长鑫存储…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "出口管制传闻在TMT基本面侧的真实冲击远小于午盘股价隐含的悲观幅度，但敞口分布的非均匀性意味着红利轮动结束后将出现「国产替代纯度高」标的的结构性重定价。", "summary": "针对2026-05-22午盘发酵的出口管制传闻，TMT侧的供应链拆解显示敞口分布并不均匀。光模块龙头(300308/300502)海外收入占比超85%，是直接受冲击点，但800G/1.6T的产能集中度形成天然壁垒，真实营收冲击预计仅5–15%；服务器代工(601138/000977)依赖墨西哥/越…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链是入口流量重估周期的有效Beta，但其有效性正受限于非线性的“电力墙”约束与存储成本上行，Beta效应已从“通用逻辑”转向“网络与存储”。", "summary": "截至2026年5月21日，AI入口流量（Gemini/Alexa for Shopping）与算力链Capex同步性极高，相关系数达0.92。智能体AI驱动的推理流量同比增长7851%，迫使Hyperscaler将2026年Capex推升至7250亿美元。压力测试显示，算力链Beta敞口正面临“电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 2, "neutral": 2, "value": 51.4, "events": [ { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI资本开支足以抵消重工业衰退冲量并能驱动半导体层利润率周期性走高，但云基础设施层利润率短期反而受压——卡片4的'TMT整体利润率扩张'标签需被分层重写为'半导体↑/云↓'。", "summary": "对卡片4双重命题做分层压力测试：$3,750亿AI资本开支能抵消OFS/重工资本下行（算术差距62倍，支持）；但'TMT结构性利润率扩张'必须拆成上下游——上游半导体（NVDA、TSMC、HBM三巨头）凭CoWoS/HBM4供给瓶颈+客户付款意愿，毛利率扩张是真命题；下游云基础设施（Azure/A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移", "summary": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移 AI Physical Bottlenecks: Shift from GPU Capacity to Power Transformers and Interconnection Queues 全球电网并网周期拉长、特大变压器交期逼…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Re-estimate supply chain price shock transmission to U.S. core PCE", "summary": "Re-estimate supply chain price shock transmission to U.S. core PCE In 'Thesis 1: Step 4', direct a high-priority handoff to global-macro to re-estima…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "'效率降级'对 2026–2027 AI Capex 总量近乎无约束，但通过电力、关税、航运三条渠道侵蚀单位经济学，使全球 AI 基础设施从'同步扩张'转为'地理极化'，估值真正承压…", "summary": "本卡对 Research note 1–5 的“效率降级削弱 AI 算力扩张”假设进行双向压力测试。结论：方向正确但程度被高估，AI 算力 Capex 在 2026–2027 名义增速仍 +19–26% YoY（北美 hyperscaler 2026 指引 4,400–4,650 亿美元 [S1–…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 2, "value": 54.2, "events": [ { "title": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？", "summary": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？ DeepSeek-V4-Pro 75% price cut × NVDA FY27Q1 Data Center +92% joint p…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？", "summary": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？ 同日落地的两个 AI 信号——DeepSeek-V4-Pro API 永久降价至原价 25%（中国侧 token 经济砍 75%）与 N…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "高功耗芯片与超高密度机柜使液冷与电力从‘辅助配件’升级为‘物理硬约束’，虽然下游总需求爆发，但应警惕国内低门槛液冷集成环节的价格战与毛利挤压，聚焦高壁垒零部件核心件与变压器出海链的确…", "summary": "AI算力密度飙升驱动物理基础设施“水冷与电力”爆发。HBM4与CoWoS演进（如GB200 NVL72单柜120-132kW）使液冷成为刚需。A股温控龙头英维克2025年营收增32.23%，但2026Q1因价格战净利降81.97%，分化明显。电力设备端金盘科技等受益于出海及AIDC变压器爆发，海外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 2, "neutral": 0, "value": 49.6, "events": [ { "title": "半导体瓶颈不在 GPU 而在 CoWoS-L 与 HBM，光模块瓶颈不在出货而在 1.6T 交付周期与客户独家配套——这两条链是 Research note 06 中段瓶颈租框架下 …", "summary": "2026-05-26：本卡 support Research note 06 的'中段瓶颈租'框架，并在 TMT 半导体与光模块层将之定量化。核心结论：CoWoS-L 月产能从 2025 年约 40kwpm 升至 2026 约 75kwpm 仍被超额预订（覆盖率 1.4–1.6x），HBM 长协覆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "短期硬约束在 HBM 与 CoWoS-L 而非电力，AI 算力'紧但能出货'，对指数 FOMO 有利但 2026H2 脆弱性显著上升。", "summary": "短期（2026 年内）芯片侧的约束比电力更紧：依次瓶颈为 HBM3e bit 供给 → 台积电 CoWoS-L 先进封装 → GPU 晶圆 → 基板/光模块 → 电力。HBM 与 CoWoS-L 全年额度已锁、扩产跑不过 ~150%/~70% 的需求增长，2026H2 任一良率事件或出口管制重置可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 1, "value": 60, "events": [ { "title": "AI 算力物理墙的宏观结果是资本开支被拉长而非崩盘，TFP 红利右移 18-24 个月，配置上有利于电力/封装/国产替代而非纯算力多头。", "summary": "截至 2026-05-27，前序卡片揭示的 CoWoS-L 翘曲与 HBM4 多供应商 KGD 危机不会引爆 CSP 资本开支崩盘，而是把周期向后拉长至 2028-2029：Big Four 2026 资本开支 USD ~470bn 大概率维持，但 GPU 实际出货低于指引 12-18%，BOM …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "台积电CoWoS与HBM底层封装良率构成全球AI芯片出货的硬性物理约束，超额利润集中于垄断性瓶颈厂商（TSMC/MU/SK Hynix），而A股代理股共享叙事却无壁垒且交易极度拥挤，…", "summary": "支持超配海外半导体物理瓶颈（MU/SOXX）并降配A股算力代理。台积电CoWoS封装产能从2024年底3.5万片/月扩至2026年底超12万片/月但仍售罄，NVIDIA垄断超60%份额；HBM市场2026年规模将达600亿美元，美光产能售罄且良率升至70-75%。A股光模块、服务器组装与IDC代理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 1, "value": 61.4, "events": [ { "title": "AI硬件周期仍由需求驱动，但未来12个月的定价权将更多落在HBM、先进封装、光模块和电力交付等瓶颈环节，而不是单纯GPU或服务器组装收入。", "summary": "截至2026-05-30，本卡支持“AI算力瓶颈从硅端扩展至物理基础设施”的主线。未来12个月，AI芯片和服务器需求仍由超大云厂商资本开支、云AI收入、agentic AI推理流量和自研ASIC扩张驱动；但供应瓶颈已从GPU单点扩展到HBM、CoWoS/先进封装、800G/1.6T光模块、机架供电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基建复利仍是 2026–2030 最稳定主线，但价值链重心正从'GPU+HBM+CoWoS'扩展到'GPU+HBM+CoWoS+LGIA+输电+PPA'；在保留 AI 半导体核…", "summary": "本卡把前四卡合成可执行的跨行业配置框架。在加入电力权利筛选器（并网/输电/灵活负荷/PPA 四维 0–12 分制）后，给出 12 个桶的旧/新权重对照：AI 半导体核心 40→30，AI 服务器合规优胜者（DELL/HPE）10→12，SMCI 5→1，Hyperscaler 20→22，CRWV…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 1, "neutral": 0, "value": 59.8, "events": [ { "title": "非美系低轨星座的自主化上限不是芯片设计口号，而是能否获得CoWoS/HBM/材料产能并完成空间级资质认证。", "summary": "截至2026-05-30，本卡片支持前序判断：非美系低轨星座的AI硬件瓶颈已从单一GPU采购，转向CoWoS/HBM/基板/材料/测试良率组成的封装级工业体系。监管规避可以降低对美国加速器的直接依赖，但无法自动获得先进封装、HBM和材料产能；Eutelsat替换周期与Qianfan部署压力显示，星…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 2, "neutral": 1, "value": 58.2, "events": [ { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 算力供给受限于 HBM4 售罄、CoWoS 封装良率及全球电网容量的物理极限，使硬件龙头成为 AI 价值链中无视‘Capex 断崖’风险的最终收费站。", "summary": "截至2026年5月，AI硬件供应链仍处于结构性紧缺。TSMC N2量产初期良率65-70%，HBM4于2月投产且美光2026年产能已售罄。NVIDIA Rubin架构推升单GPU功耗至2.3kW，使电力供应成为比芯片更硬的瓶颈。尽管终端应用出现DAU增速放缓，但硬件端的实物稀缺性构成了核心护城河，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超级IPO将把AI公司的公开融资压力传导为算力锁定需求，从而重估半导体资本开支：超配稀缺瓶颈，回避无法证明产能稀缺和合约期限的长尾扩产。", "summary": "截至2026-05-31，本卡片支持“超级IPO流动性虹吸”对AI硬件和半导体瓶颈资产的强化作用。核心判断是，IPO资金不会平均利好半导体，而会把公开市场资金约束转化为GPU、HBM、先进封装、先进制程、光刻和数据中心网络的多年期需求可见度。TSMC、NVIDIA、ASML、Micron、Sams…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 2, "neutral": 0, "value": 56.6, "events": [ { "title": "国产算力链正经历“非对称利润粉碎”：集成商已进入类集采微利期，而设计商虽处于高毛利规模化红利期，但算力公用事业化预示着长期定价权向政府侧漂移。", "summary": "2026-06-01，本卡分析国内智算中心与信创采购的定价博弈。受地方债压力（54.8万亿）影响，资金转向超长期国债（1.8万亿）等中央统筹。当前出现非对称“类集采”效应：服务器整机商受HBM高价与下游压价双重挤压，净利率萎缩至1.5%-3%；而国产GPU设计商凭借信创刚性与20%价格优惠，毛利率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "滞胀成本压力会侵蚀但不会基准性摧毁Regime A SaaS利润率，真正的分水岭是厂商能否把AI成本从固定席位打包迁移到明确的用量/结果计价。", "summary": "截至2026-06-02，本卡压力测试能源与大宗上涨对AI硬件供应链和Regime A SaaS利润率的传导。结论是有条件支持前序Regime A逻辑：晶圆厂电力成本上涨会压低硬件毛利，但单独不足以击穿SaaS Agent利润驱动；真正高Beta风险来自HBM、先进封装、数据中心电力与云厂商GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 1, "risks": 0, "neutral": 0, "value": 78, "events": [ { "title": "110-120美元油价引发的金属与电池材料通胀对AI芯片毛利无碍，但将系统性挤压缺乏定价权的AI服务器ODM及散热/电源供应商，导致其毛利率被动收缩60-90个基点。", "summary": "截至2026年6月2日，本卡支持上游材料成本上升的判断，并评估其对AI硬件的传导。110-120美元的油价将显著推升铜（母线/线缆）、铝（液冷分水器/散热器）及BBU电池的成本。尽管对具备极高毛利的AI芯片（GPU/HBM）影响微乎其微，但系统级组装厂（ODM）和电源/散热基础设施供应商将面临严重…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 2, "neutral": 0, "value": 59.4, "events": [ { "title": "美股AI硬件强势对科创50半导体链存在明确补涨逻辑，但仅在先进封装和28nm/14nm主流设备两条主线上'清晰可定价'，应以'订单确认+流动性双重确认'同时为纪律抓手。", "summary": "针对2026-06-03美股AI硬件强势能否传导至科创50半导体链的核心问题，TMT分析师支持卡01战术框架，并把补涨逻辑结构化为可定价的两条主线：(1)先进封装（长电600584/通富002156/华天002185/兴森002436），承接台积电CoWoS 25-30%的产能溢出；(2)28nm…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HBM/ASIC的结构性支撑在6-18个月维度上成立，但被卡片01低估了'客户集中度'和'瓶颈从硅迁移到电力'两条脆弱性——结构核心需要内部再平衡（提电力/冷却、减HBM股权、AVG…", "summary": "锚定2026-06-03（周三），AVGO FQ2业绩大概率周四盘后披露、NFP周五早间公布。我整体认同卡片01的'结构有支撑、战术有脆弱'综合判断，但TMT视角下结构性sleeve存在三条具体脆弱性需要再定价：(A) HBM需求高度集中在NVDA单一客户（约60-70%比特占比），属于穿着结构外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [ { "title": "2026 06 04 第03张研究卡：AI硬件供应链成本压力与利润率韧性 —— 压力测试 立场：stress test（压力测试）。 截至研究院工作日期 2026 06 04 ，Re…", "summary": "2026 06 04 第03张研究卡：AI硬件供应链成本压力与利润率韧性 —— 压力测试 立场：stress test（压力测试）。 截至研究院工作日期 2026 06 04 ，Research note 01 [Research note 01] 与 Research note 02 [Rese…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 3, "neutral": 0, "value": 56.2, "events": [ { "title": "AIDC的投资核心是‘以电力溢价换取时间期权’以规避GPU闲置折旧，而非单纯的度电成本控制；BTM燃气虽能守住7%–10%的ROIC合格线，但资产估值已面临结构性下修。", "summary": "本卡综合Research note 01/02表明，AIDC投资重心并未转向传统“极端成本控制”，而是转向“时间期权套利”与“资本分层”。在53个月并网延迟下，防范GPU闲置折旧是核心，表后燃气虽使LCOE高出1.7-2.5倍，但缩短了2年上电期，保住了7-10%的ROIC底线。因此，“电力可用性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电力瓶颈不会终结Scaling Law，但会把前沿模型进步压缩成更少实验室、更大台阶、更强能源地理依赖的发布节奏。", "summary": "截至2026-06-07，本卡综合前序AIDC ROIC与BTM结论：53个月并网排队不会让Scaling Law失效，但会把前沿模型路线图从GPU约束推向已上电GW级集群约束。若训练算力继续每年4-5倍增长，约28个月并网额外延迟会造成约27-42倍目标算力缺口。基准判断是形成2026-2028…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高达53个月的并网延迟正将AIDC的ROIC从15%非线性挤压至5%以下，算力中心的投资核心逻辑已从“GPU配额”全面转向“电力可用性”。", "summary": "经测算，长达53个月的电网连接延迟和160周的变压器交付周期，正对AIDC的长期ROIC产生严重的非线性侵蚀。前期投入的资金成本损耗每月每兆瓦带来最高100万美元的潜在收入损失；更为致命的是“有卡无电”导致的GPU严重闲置折旧，可将原本15%的ROIC击穿至5%以下。算力中心转向表后发电虽可自救，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "7月关税与海运上行更可能压缩AI硬件下游组装与网络设备利润率，而非击穿核心半导体龙头利润率；估值风险来自低利润率硬件收入增长的质量被重新定价。", "summary": "截至2026-06-08，本卡对第04卡的关税/海运鹰派尾部做TMT压力测试：风险真实存在，但主要不是核心GPU、晶圆代工或HBM利润率同步塌陷，而是下游AI服务器、网络设备、机柜、电源、散热和PCBAs的落地成本挤压。NVIDIA、TSMC、SK hynix拥有高毛利率和稀缺定价权；Dell式A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 3, "neutral": 1, "value": 70, "events": [ { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI半导体利润池迁移真实但渐进，板块总EPS上行、估值锚从'NVDA-TSMC双寡头溢价'切向'增长+多元化第二供应源'，ASIC设计商、HBM、先进封装设备与Intel Found…", "summary": "2026-06-09综合：Google-Intel TPU订单确认AI硬件利润池迁移真实且渐进，2027-28年AI半导体合计EPS较冲击前共识更高，但分布坚定向ASIC设计商(AVGO/MRVL)、HBM、先进封装设备及Intel Foundry期权倾斜。新估值锚为'增长+多元化第二供应源'，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球科技配置应超配ASIC与新代工生态；A股风格则需从“AI题材映射”战略性切换至独立算力的“国产第二供应源”，重仓先进封装产能、核心设备及HBM材料。", "summary": "2026-06-09评估：基于AI半导体估值锚向“增长+第二供应源”切换，全球配置需减配纯GPU倍数扩张与单一代工依赖，转而超配ASIC双雄(AVGO/MRVL)、HBM及Intel代工期权。在A股与港股层面，风格将从“英伟达映射”(光模块/代工)实质性轮动至业绩兑现度高的国产“第二供应源”。策略…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "DeepSeek降价推动的是硬件内部的alpha重排，而不是AI硬件beta被证伪；应压缩泛化边缘和老库存，保留订单确认的光模块、HBM相关存储、液冷/电源和高端板卡。", "summary": "截至2026-06-09，DeepSeek V4-Pro 已验证永久降价为75%、至原价1/4；“96%”更像相对竞品的单位经济冲击。微观压力测试显示，GPU/HBM/800G与1.6T光模块/液冷仍受订单、价格和产能约束，Jevons弹性成立。中际旭创、沪电股份、佰维存储有收入、交付或稼动率确认…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 3, "neutral": 0, "value": 60.4, "events": [ { "title": "AI需求到2026年底足以兑现HBM锁定的供给，但A股的真正约束是份额捕获与财政融资通量，因此应超配云挂钩供应商与一体化国产替代栈、把端侧AI作期权、低配离岸GPU依赖与重复建设长尾…", "summary": "锚定2026-06-10，本卡综合第01卡（首席策略师）与第02卡（政策分析师）。中国宏观叠加三条修正：(1) A股AI capex由大基金三期+省级算力券+头部云自融政策配给，民营信贷通道薄，是财政题非周期题；(2) 5月PMI 49.6 / Caixin 50.4 / 社融~8.4% / CP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-10的主线不是在云、终端和政策之间三选一，而是超配云需求确认，选择性持有HBM/国产替代，暂把Apple换机作为二阶段期权，并低配政策与现金流暴露高的AI资产。", "summary": "截至2026-06-10，本卡综合Oracle业绩窗口、Apple AI终端升级和美国先进芯片海外子公司管制澄清。结论是：A股AI链条应优先交易云需求确认，其次保留Apple端侧AI换机期权，同时用出口管制和自由现金流压力控制仓位。Oracle、Microsoft、Alphabet、Meta仍显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超配拥有确定性全球订单的光电液冷龙头与昇腾-中芯先进封装/HBM自主垂直栈，回避被截断离岸通道的二线芯片设计与空壳算力租赁标的。", "summary": "截至2026-06-10，本研究深入A股TMT板块。建议超配：(1)挂钩中美头部云、订单已确认的光模块（中际旭创、新易盛、天孚通信）、液冷（英维克）与电源龙头（麦格米特），均处于1.6T/Blackwell订单放量期；(2)华为昇腾与中芯国际（SMIC）挂钩的HBM/先进封装国产替代垂直栈，长电科…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 67.8, "events": [ { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与GPU瓶颈通过抬高单位算力成本和延长产能交付周期，向生物医药研发征收一道结构性'算力税'，把卡片06的'算力转化成果'筛选逻辑从风格偏好升级为预算流程强制执行。", "summary": "截至2026-06-12，本卡支持卡片06判断：电力与GPU瓶颈把云算力单位成本（H100按需租金升至$2.5–3.5/小时）与有效交付周期（>10 MW园区24–36+月）同时抬升，给生物医药研发加征一道成本税。单个AI驱动早期项目算力支出从2023年$1.0–6.0 mn升至2026年$1.4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 1, "neutral": 1, "value": 78.2, "events": [ { "title": "Your Role: AI基础设施分析师 (AI Infrastructure Analyst) Expertise: AI compute/datacenter/GPU clust…", "summary": "Your Role: AI基础设施分析师 (AI Infrastructure Analyst) Expertise: AI compute/datacenter/GPU clusters, power & grid bottleneck, hyperscaler capex Step −1 — …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "为支撑三层算力织物，AI芯片与先进封装如何演进以匹配电力波动与边缘推理的功率包络", "summary": "综合并支持卡片04的三层算力织物，下沉到硅基与封装层。关键洞见：'电力波动友好'主要是电力电子＋调度固件问题，而非新晶圆问题。NVIDIA GB300靠电源货架（65 J/GPU储能＋功率封顶斜率＋功率烧除）把Megatron训练电网峰值砍约30% [S1]，DVFS早是标配 [S2]；同步功率摆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 4, "neutral": 1, "value": 61.6, "events": [ { "title": "GB200 级效率提升改善 AI factory 经济性，但 2026-2028 年真正限制部署节奏的是站点电力、并网、变压器、稳定电源和电网运行能力，而不是 CoWoS/HBM 单…", "summary": "2026-06-13：本卡对 NVIDIA 护城河做电力压力测试。GB200 NVL72 单机柜约 120 kW，1 GW IT 功率约支持 8,333 个机柜、约 600,000 颗 GPU。IEA、DOE/LBNL 与 NERC 显示，数据中心用电、并网队列、变压器和燃机交期都在拉长，大型计算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS绝对产能正在改善，但GPU与自研ASIC需求吸收速度更快，因此2026年AI集群交付仍受先进封装、HBM和关键设备交期的结构性约束。", "summary": "截至2026-06-13，本卡支持上一张AI基建卡的判断：硬件交付风险不只来自并网，还来自CoWoS/HBM/设备交期的半导体转换链。TSMC仍称先进封装非常紧，TrendForce口径显示CoWoS-L/S满载且2026年底目标120,000-130,000片/月；DIGITIMES虽更乐观，预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装和材料创新可以守住稀缺瓶颈环节的利润率，但无法单独对冲GPU预算被电力和基建capex挤出的全链条利润压力。", "summary": "截至2026-06-13，本卡压力测试半导体环节能否通过先进封装和材料创新抵御AI基建挤出效应。结论是选择性支持：TSMC式领先制程/先进封装、HBM、高端基板、热管理材料和部分光互连环节可凭稀缺性维持利润率，但若沿用上一卡严重情景，GPU capex从$3.500 billion降至$1.598…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红皇后竞速并未彻底抹去自研ASIC的成本优势，而是将其集中至Google与AWS两家、使其他云厂商回到对NVIDIA的依赖，从而部分推翻了Research note 05所述的全行业…", "summary": "在'红皇后竞速'下，自研ASIC单世代有效寿命已压缩至12–15个月，2nm节点全口径NRE攀升至USD 0.9–1.4B、N2晶圆约USD 30,000/片、CoWoS-L售罄至2027。我们测算自研ASIC在TCO上跑赢商用GPU的年部署门槛从5nm节点的约20万颗升至现实情景下的15万–25…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 4, "neutral": 0, "value": 60, "events": [ { "title": "AI先进封装与HBM产能瓶颈对算力基础设施扩张及超大规模厂商资本开支的影响", "summary": "支持半导体分析师Research note 01的论点并向下游延伸：CoWoS/HBM确为算力的关键约束，2026年部署受封装与内存所限（CoWoS扩至约13万片/月仍订满；SK Hynix与Micron 2026年HBM售罄；HBM4验证延迟使NVIDIA将Rubin目标由200万下调至150万…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力成本上升和并网延迟不会终结AI芯片周期，但会把赢家从峰值算力叙事筛选为能在稀缺MW下交付最高有效token产出的全栈能效架构。", "summary": "截至2026-06-14，本卡支持前序判断：AI扩张的瓶颈已从单纯GPU供给转向电力与并网，但半导体端可通过能效、液冷和先进封装显著对冲。设计指标将从峰值FLOPS转向每瓦token、每MW可部署token和机架级系统效率。Blackwell、Trillium TPU、Trainium2等路线显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GPU抵押贷款收缩会削弱边际Neocloud订单可见度，但先进封装与先进制程需求更可能从融资套利订单转向信用更硬的超大规模厂商和ASIC路线，而非进入全面下行。", "summary": "截至2026-06-15，本卡压力测试GPU抵押融资收缩对半导体订单的影响：冲击不是AI芯片需求消失，而是订单可见度从Neocloud融资驱动采购转向超大规模厂商、ASIC、主权项目和带预付款的硬合同。NVIDIA Data Center在FY2027 Q1仍达752亿美元，TSMC先进节点和HP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在AI链内部按融资质量轮动：超配Tier-1平台与卖铲人(设备/HBM/封装/电力赢家)，低配靠GPU抵押融资循环的Neocloud股权与高收益债。", "summary": "综合前八卡推演，AI基础设施的约束已从硅到电力、再到融资可得性。本卡给出按融资质量过滤的五篮子相对配置框架：B1 Tier-1 AI平台(+300bp)、B2 设备/HBM/先进封装(+250bp)、B3 ASIC定制(+100bp)、B4 电力瓶颈赢家(+150bp)、B5 低质量AI基建(-2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 1, "value": 70.4, "events": [ { "title": "验证AI算力硬件链的估值承压：流动性虹吸下的先进封装与GPU产业链韧性评估", "summary": "本卡对AI算力硬件链做压力测试，得出'分化韧性'结论。资金流层面确认虹吸：2026-06-08半导体−219.33亿、通信−98.35亿、消费电子−69.92亿元净流出居前[S6]，年内93家半导体公司减持[S7]，叠加CXMT约295亿IPO锚定[S9]。但估值下移是资金流/轮动驱动、非盈利驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "宏观信用收紧并未击穿全球AI周期，而是将AI基础设施层切割为「超大规模云厂＋央企国家队」核心（继续扩张）与「民营商业云」外缘（防御性收缩）的K型结构，TMT投资应聚焦CoWoS/HB…", "summary": "TMT综合：沃什鹰派持稳（3.50–3.75%）经离岸人民币流动性、汇率对冲摩擦与企业中长贷收缩，传导至AI基础设施层产生K型分化而非总量收缩。美超大规模云厂CY2026合计capex运行率3,200–3,600亿美元由经营现金流支撑不受冲击；央企运营商云经PBOC科技创新再贷款（约2.5%）继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 4, "neutral": 0, "value": 62.8, "events": [ { "title": "通胀和高利率提高了AI硬件基础设施的质量门槛，但没有削弱其配置价值；应买入瓶颈供应商而非泛AI叙事。", "summary": "截至2026-06-16，本卡支持“硬科技配置中保留AI基础设施”的上一轮判断，但建议从广义AI beta收窄到有订单、定价权和瓶颈属性的硬件基础设施。高利率会压缩投机型数据中心开发商和远期AI应用估值，却未削弱GPU、定制AI芯片、HBM、配电、冷却、变压器、开关设备和已锁定电力的数据中心容量的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "All three deliverables are written and verified. Research note 02 summary — Stance: SUPPORT…", "summary": "All three deliverables are written and verified. Research note 02 summary — Stance: SUPPORT (with one relative value refinement). I confirm Research …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供给限制驱动的半导体能效转型主要表现为通用GPU的系统与精度级优化，而非ASIC/NPU的大规模替代；自研定制芯片的爆发反而加剧了全球先进制程与封装供应链对TSMC的物理集中度。", "summary": "截至2026-06-17，电力瓶颈确在倒逼算力能效转型，但并非单纯的通用GPU向专用ASIC切换。算法多变性（如MoE与动态推理）使自研ASIC面临极高的“代差失效”风险，通用GPU（如NVIDIA Blackwell）凭借软件与架构灵活性仍占据70%以上份额。能效提升主要通过系统精度（FP4/F…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体能效能把AI扩张从“堆GPU”改写为“每MW token产出竞争”，但只能买时间、改善单位经济性，不能让电力瓶颈消失。", "summary": "截至2026-06-16，本卡压力测试“电力预算驱动扩张”逻辑：半导体能效提升真实且重要，Blackwell、先进封装、vLLM/FlashAttention、稀疏MoE和定制ASIC可显著提升每MW有效token产出。但它不能完全对冲电力瓶颈，因为美国数据中心用电量可能从2023年的176 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 5, "neutral": 0, "value": 66.2, "events": [ { "title": "AIDD是AI基础设施需求的高质量增量因子：规模小于通用大模型，但持续性、数据壁垒和监管/IP属性更强。", "summary": "截至2026-06-18，本卡片支持上一张卡片的方向：AIDD不会单独决定hyperscaler capex，但会把高价值、持续性的生命科学工作负载接入GPU、网络、存储、液冷和电力保障型数据中心。Recursion BioHive-2、DCAI Gefion、Novo Nordisk、Isomo…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI服务器非核心BOM零部件国产替代已具备深度全球竞争力，但高速交换芯片与高端PMIC的缺位成为制约万卡集群本地化扩张的关键瓶颈。", "summary": "截至2026-06-18，对高端AI服务器BOM成本拆解显示，虽然GPU与HBM占据近80%成本导致全球供应链国产化率仅为4.64%，但非核心硬件BOM（PCB、散热、接口、电源）的全球国产价值占比达到22.50%，且在本土化服务器情景下上升至73.65%。A股龙头已在PCB与Retimer/RC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 能源重定价压的是超大规模厂商 ROIC 而非芯片 capex 效率——硅片主导 TCO、能源占比单位数，且性能/瓦溢价+杰文斯回弹反而净支撑前沿逻辑/先进封装/HBM 需求…", "summary": "2026-06-18 压力测试：能源是证伪芯片逻辑的错误杠杆。Epoch 1GW 机房年化 TCO 约 $8.5B，服务器占 ~59%、能源仅 ~7%；交付电价 +30% 仅抬升 TCO 约 +2.1%，翻倍也仅 +7%（自测算），而 IT 寿命 5→3 年使 TCO +41%——刷新节奏的影响是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FOMC 后应以“选择性算力基础设施挤空”而非“无差别 AI 融涨”来交易：多稀缺、空低 ROIC，净多但对冲。", "summary": "2026-06-18 策略综合：前序链条成立，2027 风险是 ROIC 与结构分化而非 capex 悬崖。FOMC 后应做多稀缺算力资产而非泛 AI beta：前沿逻辑/设备/HBM/先进封装、自研 ASIC 与网络、电力与电网受益方应超配；商用 GPU 龙头保留但严控风险；低 ROIC 商用 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 1, "risks": 3, "neutral": 1, "value": 82.6, "events": [ { "title": "电力/电网瓶颈真实且已在容量价格中显形,但2027年前仅推迟部分美国新增集群、不封顶全球GPU部署,Research note 01的BOM组合在2026至2027 H1仍成立,20…", "summary": "AI算力电力瓶颈真实存在但属于'延后+置换'而非'封顶':PJM 2026/27容量价已跳11倍至$329.17/MW-日,2027/28预测缺口6.6 GW,新接入等待40个月;ERCOT 2026 Q1单季198 GW申请,累计大负荷队列约410 GW;全球电网接入等待5-10年,变压器交期1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the analysis on how advanced packaging (Chiplets, HBM4) and optical interc…", "summary": "I have completed the analysis on how advanced packaging (Chiplets, HBM4) and optical interconnects (Silicon Photonics, CPO) mitigate U.S. power grid …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "RUM的Together AI合同仅占用其未商业化电力的约3%，而完全填满200 MW所需的47亿至51亿美元资本开支面临巨大融资缺口，且TSMC CoWoS与HBM3E供给紧缺导致…", "summary": "截至2026-06-18，对RUM Group的B300 GPU采购进行供应链与财务敏感性测试显示：其与Together AI签署的2.7亿美元B300液冷云服务协议在物理规模上仅需约1,800至3,000张GPU，对应电力功耗仅约4.8至7.3 MW，仅占其剩余200 MW未商业化容量的2.4%…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "大模型价格战引发的推理流量爆量，导致国产GPU设计龙头呈现“营收高增、毛利率承压”的量价背离，定价权向制造端瓶颈与TCO优化环节倾斜。", "summary": "在大模型极低价格战（DeepSeek-V4-Flash降至$0.14/1M tokens）推动下，云厂商Q1 AI基础设施收入大增79%（GPU云激增184%）。但因先进制程及封装产能受限导致BOM成本上涨10%-15%，加之云厂商降本压价，国产GPU（如海光一季度毛利率降至55.6%，同比下滑5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 5, "neutral": 1, "value": 64, "events": [ { "title": "电力与封装成本上升会触发AI基础设施技术路线调整，但截至2026-06-20仍不足以触发超大规模云厂商AI芯片采购总量的全面回撤。", "summary": "工作日2026-06-20：本卡压力测试认为，电力和先进封装成本上升不会在2026年触发超大规模云厂商全面削减AI芯片采购，但会提高单位弹性，推动技术路线从“更多GPU插槽”转向“每瓦性能、液冷机架、自研ASIC、提前锁定CoWoS/HBM和电力站点”。真正风险是并网、变电站、CoWoS或HBM配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A-share de-leveraging to 1550 points triggers secondary market liquidity locking and margin…", "summary": "若市场下探至1550点，中际旭创股价将跌至714.16元，击穿融资预警与平仓线，带来日均31.2亿元平仓抛压（占ADV 17.36%），锁定二级流动性。但公司持有125亿现金及45亿海外客户预付款，可提供151.5亿净营运资金，供应链运营违约风险极低。海外订单真正风险在于电网并网排队达36-60个…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "16.18%的硬件Capex下修会压低AI基础设施整体估值，但订单传导不是线性的：液冷最敏感，电网设备和BTM发电更受长期电力容量瓶颈支撑。", "summary": "截至2026-06-19，本卡把H2 AI硬件Capex下修16.18%作为未外部确认的压力测试输入。结论是分层传导：GPU/服务器采购最先受冲击；液冷因绑定高密度机柜节奏，订单和估值敏感度较高；电网设备因并网稀缺、变压器长交期和backlog，订单不应按同等比例下修；BTM自备发电估值会先压缩，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 2, "neutral": 0, "value": 80.4, "events": [ { "title": "虽然HBM国产化良率崩塌会将集群 CapEx 推升 98%，但中国互联网巨头通过异构分级部署与算法优化（缩减30%GPU需求）并结合西部绿电直连套利（LCOE降至0.39元）与液冷改…", "summary": "评估HBM成本压力对AI集群TCO及扩容策略的传导。100%国产材料导致HBM3封装良率崩塌至4.11%，推升10000颗GPU集群CapEx至8.17亿美元，东部煤电下3年TCO达10.17亿美元（+66.13%）。建议协同优化：架构上采取60% HBM3与40% HBM2E的分级部署，结合算法…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网交付延迟不会终结高密度 AI 机柜，但会迫使下一代 AI 芯片与集群架构把 perf/W、可限功率和可分阶段并网能力置于峰值算力之上。", "summary": "2026-06-20：本卡对电网设备瓶颈约束 AI 芯片架构的说法进行压力测试。结论是，变压器和开关设备长交期不会让前沿训练集群普遍降密度；GB200 NVL72、Vera Rubin 和 AMD Helios 仍指向高密度机柜级设计。但电网交付延迟会把竞争指标从峰值 FLOPS 推向受限兆瓦下的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 4, "neutral": 1, "value": 72.8, "events": [ { "title": "Hyperscaler资本开支正从算力优先转向电力协同：总量仍大，但边际一美元先购买电力、进度和并网确定性，再购买下一组加速器。", "summary": "截至2026-06-21，本卡压力测试认为：Hyperscaler资本开支不会因电力瓶颈在2026-2027年崩塌，但边际规则已从“买更多GPU”转向“按确定日期拿到可通电MW”。核电购电、表后自发电、并网权、长交期电气设备和区域边缘部署正在获得更高资本权重。核电是长期稀缺性对冲，不是近端解决方案…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-21，受2nm GAA先进制程 wafer 溢价与先进封装 80% 初期装配良率的限制，下一代高密度 AI 芯片的制造 BOM 将翻倍至8,153.59美元，其中封…", "summary": "截至2026-06-21，本卡片评估了AI机架密度120kW+下，先进封装良率及散热材料研发对AI芯片成本与供应链的重塑。下一代2nm GAA+HBM4+液冷芯片BOM成本较3nm基准大涨77.2%至6,780.97美元，在初期封装装配良率降至80%的重度压力下更暴增113.1%至8,153.59…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "提升芯片能效与先进封装无法对冲变压器短缺造成的并网延迟损失（闲置折旧高达25亿至50亿美元），反而因功耗密度飙升将瓶颈推向机架级配电，且受限于国内封测良率瓶颈与成本压力，持有电网设备…", "summary": "截至2026-06-21，本报告对“半导体设计/封装技术迭代能对冲电力并网延迟”的论点进行压力测试。分析表明，12-24个月的变压器交付延迟对200k GPU集群将造成25.3亿至50.7亿美元的闲置折旧，重创ROIC并推升GPU租金成本达39.1%至156.7%，芯片端能效提升被财务减值彻底吞噬…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-21，电网与电力并网瓶颈带来的通电延迟（2年延迟拖累53.71亿美元[自测算]）将迫使云厂商放缓GPU采购，若发生20%的需求削减，台积电3nm/2nm制程及CoW…", "summary": "截至2026-06-21，本分析对‘电力与电网并网瓶颈’对高端AI芯片（GPU、HBM）及先进制程产能的传导影响进行了压力测试。定量测算表明，10万张Blackwell GPU集群面临2年通电延迟将产生53.71亿美元闲置财务拖累[自测算]。如果云厂商因电力限制削减20%芯片采购，台积电3nm和2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 3, "neutral": 1, "value": 80.2, "events": [ { "title": "扩大版出口管制导致材料与EDA工具国产化面临良率黑洞与算力TCO暴增，实质性切断了在华先进封装的商业化增量逻辑。", "summary": "截至2026-06-22，评估表明若BIS将管制扩大到先进材料（如ABF载板、MUF、NCF）及EDA工具，将实质切断在华先进封装产能的商业化增量逻辑。100%国产化材料导致8-Hi HBM堆叠良率崩塌至4.11%，产出缩水68.48%；GPU BOM成本翻倍至$37,746，售价涨至$75,49…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "变压器/电网瓶颈对AI芯片需求预期与库存周期的传导风险", "summary": "截至2026-06-22，本卡以半导体视角压力测试Card-03。结论：芯片需求仍明确领先于电力瓶颈，2026年不存在AI芯片砍单——约束仍是芯片供给本身（CoWoS售罄至2026年底、英伟达独占约60%；Blackwell积压约360万颗售罄至年中；2026年HBM全部订满且超大规模厂商预付扩产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "summary": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与定制化ASIC的深度整合，是将大模型访存能耗降低95.7%并实现3-5倍能效提升的核心硬件路径，亦是国产半导体链规避先进制程封锁、支撑算力电力约束的战略根基。", "summary": "截至2026-06-23，本卡研判支持通过先进封装与ASIC能效路径应对算力电力约束。2.5D封装结合HBM3E使访存功耗降低95.7%（从35 pJ/bit降至1.5 pJ/bit）；自研定制ASIC能能效比高出GPU约3-5倍，万卡集群年省电费2,943万元人民币，并释放48MW电网容量。国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 4, "neutral": 0, "value": 72.6, "events": [ { "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "summary": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下修。", "summary": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "硬件预算挤压对次世代AI芯片(如Rubin)性能需求与技术路线的传导影响", "summary": "针对“电力成本飙升→能效驱动型替代→削弱顶尖算力需求”的链条做压力测试。结论：前提成立但结论被颠倒。能效比(每瓦token→每兆瓦营收)确已成为芯片主导设计轴[S3][S10]，但在电力硬约束下，优化能效比会把需求集中到前沿节点(Rubin/HBM4/CoWoS-L)而非替代离场——兆瓦稀缺时运营…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI芯片能效和封装升级可以提高每个已送电MW的算力收入，但不能消除变压器长周期交付造成的站点送电瓶颈。", "summary": "截至2026-06-23，本卡支持前序判断：半导体厂商正在通过更高性能每瓦、低精度推理、HBM/CoWoS/3.5D封装和自研ASIC，提高每个已送电MW的AI产出。但这只是缓解，不是替代变压器。若站点无法送电，GPU租赁收入仍会延后；若站点已送电，Blackwell/GB300、MI350、Ir…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 4, "neutral": 0, "value": 74, "events": [ { "title": "中国CSP成本上升背景下，AI算力基础设施供应弹性与扩容压力分析", "summary": "压力测试部分否定线程以电力为中心的框架。AI算力系统是供给绑定而非需求绑定：2026年超大规模厂商资本开支同比+77%至约7250亿美元[S1]，英伟达Blackwell售罄至年中、积压360万颗[S2]，因此约束的是配额优先级而非需求。核心锚定集群凭借锁定电力、GPU配额与表后自发电被隔离；放缓…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与开关设备长周期会迫使AI数据中心按可送达电力重新选址，并把超大规模资本开支从GPU/土建扩展到电力交付栈。", "summary": "截至2026-06-25，本卡片压力测试卡片06：AI算力选址确实会从“便宜电”转向“可信time-to-power”，但不是简单回到传统成熟枢纽。变压器、开关设备、变电站与并网改造成为已带电容量的关键约束；低电价若没有设备槽位和送电排期，只是搁浅期权。超大规模厂商的资本开支将更多前置到电气设备预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网容量和可锁定firm megawatts正在把AI基础设施从纯GPU扩张重定价为“GPU加电网”资本开支周期，边际训练集群将向能源优势区域迁移。", "summary": "截至2026-06-24，本卡支持电网瓶颈会改变AI算力部署重心的判断，但强调这是边际迁移而非传统中心消失。美国数据中心用电量已从2014年58 TWh升至2023年176 TWh，2028年可能达325-580 TWh；NVL72单柜约120 kW，100 MW供电延误等同于约5万GPU级别产能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "直接晶圆代工耗电仅占先进晶圆出厂价的不到0.8%[S4]而电价脱耦，但电网设备与并网容量缺口延迟算力中心并网并砸损ROI，将强力逼迫GPU单价大跌45.89%并使芯片商毛利自75%压…", "summary": "截至2026-06-25，本卡对电力通胀下AI芯片成本与利润率进行压力测试。结论表明：直接晶圆制造电费仅占先进 wafer 售价的不到0.8%[S4]，对毛利几乎无影响；但间接上，电网容量和设备缺口延迟算力中心并网18个月[S5]，导致12.9GW电力缺口与1032万颗GPU出货延期[自测算]。高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 1, "risks": 9, "neutral": 0, "value": 53.4, "events": [ { "title": "截至2026-06-25，AI硬件需求的下一核心约束不是芯片是否能生产，而是足够多的兆瓦能否被许可、接入、冷却并按期商业化。", "summary": "截至2026-06-25，本卡支持上一张卡的AI硬件稀缺性主线，但将瓶颈从GPU/HBM进一步上移到可通电数据中心、电网接入、冷却和电气设备。IEA、Berkeley Lab、EPRI、CBRE和JLL数据共同显示，AI需求仍强，但项目能否转化为创收产能取决于电力和并网速度。结论是：稀缺性继续支撑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备瓶颈不会系统性消灭AI加速器需求，但会把2026-2028年约15%-25%的GPU/HBM/ASIC部署变成收入确认和爬坡时点风险。", "summary": "截至2026-06-25，本卡从半导体侧压力测试电力设备瓶颈传导：2026-2028年加速器等效部署中约15%-25%面临2-4个季度时点风险，真实终端需求毁灭约3%-7%。商用GPU最直接受通电MW影响，HBM因多年协议和稀缺配额更像库存节奏风险，自研ASIC既受益于每瓦token优化，也因客户…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 1, "neutral": 0, "value": 94.8, "events": [ { "title": "1.6T光模块交付延迟对AI算力集群及超大规模云厂商Capex部署进度的影响", "summary": "支持卡片01/02的研究线并加以边界化延伸：200G EML 的 HTOL 风险确会让 1.6T GA 滑向 2027H1，但这是一道带宽前沿与单位经济性约束，而非整体 Capex 停摆。GB300 NVL72 的 scale-up 走铜缆 NVLink、scale-out 走 ConnectX-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 6, "neutral": 0, "value": 66.2, "events": [ { "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-27，AI电力并网限制（2年延误拖累53.71亿美元[S12]）将倒逼算力采购削减并重创高端半导体高增长假设，投资者应果断削减AI硬件至12%及长债至5%，转而增配…", "summary": "截至2026-06-27，本报告重新校准了在高利率平台与电力之墙挤压下的多资产配置。2年电网并网延误对10万卡Blackwell智算集群将产生53.71亿美元搁浅资金损失[S12]，迫使云巨头在2027年削减20%GPU订单。这将导致台积电3nm和2nm产能利用率分别跌至85%与80%，CoWoS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球AI基础设施资本开支对中国产业链的传导机制与订单确定性", "summary": "本卡支持卡片05遗留命题并定论：约7250亿美元、同比+77%的全球超大厂资本开支（约75%即约4500亿为AI专项）规模巨大、外部出资，已转化为中国主导且出口管制并不约束的那一层链条（光互连/连接/散热/PCB）的跨多季度订单簿与已兑现盈利。关键在于出口管制是分层的：重压落在GPU/算力层（华为…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力瓶颈的主要冲击不是立即砍掉云商 Capex，而是延后通电与收入确认，并把减值风险集中到先到货、后通电的短寿命算力资产。", "summary": "截至2026-06-28，本卡对上一张公用事业结论作压力测试：电网瓶颈不会立刻终结 AI Capex 周期，但会把一部分 2026 年支出变成 2027-2028 年才货币化的在建容量。基准情形下，依赖电网的 AI 园区通电延迟 6-12 个月，20%-30% 的计划容量后移，约 $98B-$16…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 13, "neutral": 0, "value": 33.6, "events": [ { "title": "效率驱动的算力架构转型：从“暴力规模”到“高能效密度”的物理基座适配性评估", "summary": "压力测试 Research note 08 的隐含前提：效率并不缓解物理天花板，而是把它从电网层下移到机架与机柜行层并加剧。机理是机架尺度的杰文斯悖论——单瓦产出靠提高密度实现，机架功率从约40kW(Hopper)→120kW(Blackwell)→142kW(GB300,90%入液)→600kW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据中心电网并网瓶颈与核能转向", "summary": "截至2026年6月28日，AI基础设施建设的核心瓶颈已从GPU稀缺转移到电网并网时间（Time-to-Power）。对100 MW超大规模AI数据中心的模型测算显示，2年的并网延迟会减少项目35.73%的净现值（NPV），并使IRR大幅下降。为规避此类延迟，科技巨头（微软、亚马逊、谷歌、Meta）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "<20%开机率红线是平台属性而非物理定律：对电力受限EO小卫星它由电源+被动散热共同锁定且先进热控不划算(支持卡03)，但两相泵驱回路+可展开辐射器已能让专用型150kW级在轨算力平…", "summary": "本卡按平台维度厘清半导体卡的<20%开机率红线，结论为有条件支持。对Planet/BlackSky类电力受限EO小卫星，红线由电源预算与被动导热双重锁定：15-60W的GPU在散热生效前已占150-300W总线的10-40%，加装两相回路+辐射器(约2-4kg、5-15W泵耗)成本收益为负，红线成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 9, "neutral": 1, "value": 54, "events": [ { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-58200390b4/", "networkHref": "/investor/research-network/#thesis-58200390b4" }, { "id": "thesis-8b8ede3135", "source": "tracked_thesis", "title": "消费降级、健康刚需与银发经济正在重塑防御型增长", "titleEn": "Consumer downgrade, healthcare necessities, and the silver economy are reshaping defensive growth", "lane": "消费降级/健康银发", "laneEn": "Consumer downgrade, health, and silver economy", "laneHref": "/investor/lanes/consumer-downgrade-health-and-silver-economy/", "investableFocus": "消费降级/健康银发：600519.SH, 2319.HK, 3690.HK, 300760.SZ, 600276.SH, 1177.HK, UNH", "stockUniverse": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "2319.HK", "name": "蒙牛乳业", "market": "HK", "role": "必选消费", "roleEn": "必选消费", "bucket": "core", "href": "/investor/stocks/2319.HK/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "1177.HK", "name": "中国生物制药", "market": "HK", "role": "医药消费与支付", "roleEn": "医药消费与支付", "bucket": "core", "href": "/investor/stocks/1177.HK/" }, { "ticker": "UNH", "name": "UnitedHealth", "market": "US", "role": "美国医疗支付对照", "roleEn": "美国医疗支付对照", "bucket": "peer", "href": "/investor/stocks/UNH/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" } ], "stockTickers": [ "600519.SH", "2319.HK", "3690.HK", "300760.SZ", "600276.SH", "1177.HK", "UNH", "001215.SZ", "1179.HK", "9961.HK", "MU", "APO", "IEF", "COP" ], "coreView": "赛道：消费降级/健康银发。候选观察标的：600519.SH, 2319.HK, 3690.HK, 300760.SZ, 600276.SH, 1177.HK, UNH。当前共有 40 条证据、1 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 80, "eventCount": 160, "firstSeen": "2026-05-05", "lastSeen": "2026-06-29", "supports": 19, "risks": 115, "neutral": 26, "analystCount": 19, "sourceTypeCounts": { "分析协作": 16, "研究笔记": 115, "研究线程": 29 }, "topTerms": [ "CPI", "创新药", "食品", "设备", "Consumer", "downgrade", "healthcare", "necessities", "silver", "economy", "reshaping", "defensive" ], "whatChanged": [ "2026-06-29: 产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "2026-06-29: 材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放右移。", "2026-06-29: 电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "2026-06-29: [auto-handoff] Hormuz shock & core CPI transmission", "2026-06-28: 大健康低覆盖机会更像跨子行业alpha篮子，而不是单一医疗板块beta；优先研究能把2025-2026催化转化为收入桥的五家公司。" ], "practical": { "sectorLane": "消费降级/健康银发", "sectorLaneEn": "Consumer downgrade, health, and silver economy", "exposureBucket": "消费健康防御", "tone": "green", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "2319.HK", "name": "蒙牛乳业", "market": "HK", "role": "必选消费", "roleEn": "必选消费", "bucket": "core", "href": "/investor/stocks/2319.HK/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "1177.HK", "name": "中国生物制药", "market": "HK", "role": "医药消费与支付", "roleEn": "医药消费与支付", "bucket": "core", "href": "/investor/stocks/1177.HK/" }, { "ticker": "UNH", "name": "UnitedHealth", "market": "US", "role": "美国医疗支付对照", "roleEn": "美国医疗支付对照", "bucket": "peer", "href": "/investor/stocks/UNH/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" } ], "stockTickers": [ "600519.SH", "2319.HK", "3690.HK", "300760.SZ", "600276.SH", "1177.HK", "UNH", "001215.SZ", "1179.HK", "9961.HK", "MU", "APO", "IEF", "COP" ], "primaryStocks": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "2319.HK", "name": "蒙牛乳业", "market": "HK", "role": "必选消费", "roleEn": "必选消费", "bucket": "core", "href": "/investor/stocks/2319.HK/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "1177.HK", "name": "中国生物制药", "market": "HK", "role": "医药消费与支付", "roleEn": "医药消费与支付", "bucket": "core", "href": "/investor/stocks/1177.HK/" } ], "instrumentTags": [ "600519.SH", "贵州茅台", "2319.HK", "蒙牛乳业", "3690.HK", "美团", "300760.SZ", "迈瑞医疗", "600276.SH", "恒瑞医药", "1177.HK", "中国生物制药" ], "reviewCadence": "每日复核", "confirmationSignals": [ "价格弹性是否转化为利润率修复", "支付端和监管是否支持长期现金流", "防御属性是否被估值提前透支" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "估值拥挤", "支付监管", "渠道库存", "CPI 食品项" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 160, "sourceDiversity": 3, "analystCount": 19, "recentEvents": 101, "supportRatio": 0.12, "riskRatio": 0.72, "conflictScore": 80.3, "evidenceDepth": 100 }, "score": 76.9 }, "events": [ { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 22.25, "matchedTerms": [ "CPI", "evidence", "risk", "dominant", "inflation", "transmission", "expectations" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]，PPI-CPI剪刀差隔离了政策利率。任务设想的'成本→CP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CPI", "设备", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放…", "summary": "截至2026-06-29，变压器交期延长和材料瓶颈导致美国电网资本开支通胀率达32.72%。结合2027/2028年度PJM容量电费暴涨11.5倍至$333.44/MW-day，将拉动美国零售电价上涨4.18%，直接和间接推高全国CPI 0.124个百分点。高粘性通胀与AI基建庞大资金需求将上移中性利率50-75基点，延缓美联储降息路径。此外，并网滑期18…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CPI", "设备", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "食品", "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-28", "title": "去风险三期医药资产通过rNPV模型释放3.48亿美元的期权价值构成Biotech估值底，而生命科学工具在5.5x EV/Sales筑底、CXO在 overseas 利润收缩下于2.0x-3.…", "summary": "截至2026-06-28，本卡片基于BD估值方法构建了下一批并购标的与估值底线筛选框架。AbbVie 109亿美元现金收购Apogee与Merck KGaA 113亿美元收购Bio-Techne为医药生物确立了质量门控底座。我们量化厘定：1) 创新药三期去风险资产的rNPV floor：以12亿美元峰值重磅药物为例，从二期(PoS 35%)跨入三期(PoS…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.25, "matchedTerms": [ "创新药", "healthcare", "health", "sales" ], "href": "" }, { "date": "2026-06-28", "title": "大健康低覆盖机会更像跨子行业alpha篮子，而不是单一医疗板块beta；优先研究能把2025-2026催化转化为收入桥的五家公司。", "summary": "截至2026-06-28，本卡支持开启大健康低覆盖研究队列，但不建议追宽泛板块beta。优先深挖Keymed Biosciences、Hybio Pharmaceutical、Sino Biological、MicroPort NeuroScientific和Allgens Medical，主线分别是商业化创新药、GLP-1多肽制造、生命科学工具、器械出海…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "创新药", "healthcare", "health", "medical" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "食品", "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-26", "title": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "summary": "压力测试整套对冲架构所依赖的55-65%国内通缩基准概率。结论：基准情形近期稳健，保留但下调至55-60%下沿。硬数据印证消费端通缩：2026年5月CPI同比+1.2%、食品-1.7%、核心+1.1%、环比-0.1%；10年中债贴地板于1.73%[S1][S5]。PBoC“边走边看”渐进（LPR连13个月不变）、财政明确不加码，故货币与财政均为通缩利好而非…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CPI", "食品", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "宏观需求疲弱足以削弱但不足以完全否定食品加工链的成本通缩红利，配置应从板块β降级为精选大众必选品阿尔法。", "summary": "2026-06-27：本卡对“成本通缩+量端复苏”做宏观需求压力测试。5月社零同比-0.6%、商品零售-0.7%、餐饮仅+0.6%，显示广泛量端复苏尚未确认；食品CPI、农副食品和食品制造PPI确有通缩，但整体购进价格、燃料动力和化工原料仍上行。利润数据进一步分化：食品制造利润仍增，农副食品加工和酒饮茶利润下滑。结论是宏观消费疲弱会抵消相当一部分毛利扩张，…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.5, "matchedTerms": [ "CPI", "食品", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "[auto-handoff] 低价粮油替代对下半年食品CPI及生猪复苏周期的传导弹性评估", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: b8c58d06-e0d4-4616-8574-99ce2bd6c05c 结果摘要: # Step −1 — Anchor today's date BEFORE you start Run this in your shell and record the re…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 15.25, "matchedTerms": [ "CPI", "食品", "again" ], "href": "" }, { "date": "2026-06-24", "title": "铜及电网设备成本的结构性上涨将通过公用事业电价调整和工业投入品渠道转嫁给终端，抬高全球结构性通胀底座，迫使主要央行维持更高的限制性利率并加速日本央行货币正常化。", "summary": "截至2026-06-25，本卡评估了铜及电网成本压力向全球货币政策的传导。电网升级与AI算力用铜强度大，推高设备售价与监管资产基底，预测在3-5年内将零售电价推升1.5%至2.5%。50%铜价冲击将累积拉动全球CPI约0.25%至0.45%。粘性通胀使美联储年底前维持3.50%-3.75%高利率，极端下存在二次加息风险；日本央行将被迫加息至0.50%-0.…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CPI", "设备", "macro" ], "href": "" }, { "date": "2026-06-24", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "食品", "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-24", "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side …", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation, and utility tariff hikes) to …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.25, "matchedTerms": [ "thesis", "inflation", "transmission", "service", "margins" ], "href": "" }, { "date": "2026-06-23", "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "食品", "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-23", "title": "未覆盖机会集中在从科学故事转向商业化、医保、全球注册或集采韧性的公司，而不是宽泛押注医疗板块 beta。", "summary": "2026-06-24：本卡支持为生命健康和生物科技链条建立聚焦研究队列。优先深挖 Keymed Biosciences 2162.HK、Ascentage Pharma 6855.HK/AAPG、Mabwell 688062.SH/02493.HK，并在器械端保留 Peijia Medical 9996.HK 与 MicroPort NeuroScient…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15, "matchedTerms": [ "创新药", "healthcare", "health", "medical" ], "href": "" }, { "date": "2026-06-23", "title": "折扣零售REITs可以作为收益资产，但农业链数据不支持把它们定义为无条件的消费平替防波堤；仓位必须接受食品价格与农资成本压力测试。", "summary": "截至2026-06-24，农业视角否定Research note 01的强版本结论：食品与畜牧数据支持“基础消费有韧性、蛋白价格偏弱”，但不支持“消费平替必然转化为奥莱REIT现金流护城河”。2026年5月食品CPI同比-1.7%、猪肉CPI同比-16.1%，更像商超、菜场和电商的必需品折扣竞争。508602首日收盘6.144元后，按2027年预测可分配现…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.5, "matchedTerms": [ "CPI", "食品" ], "href": "" } ], "timeline": [ { "date": "2026-05-05", "supports": 1, "risks": 0, "neutral": 2, "value": 56, "events": [ { "title": "[auto-handoff] 五一消费结构分解", "summary": "致 社交媒体分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 660bdf73-6ccd-48b6-9004-104fc5535fec 结果摘要: Wrote handoff_response.md (9,442 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中国工业周期的强势主要由出口份额、政策资本开支和内卷驱动；AI-enabled automation虽真实存在于供应商订单中，但尚未在宏观生产率（实际工资、PPI、就业、TFP）上留…", "summary": "本卡否定'中国是AI-enabled automation最先进入宏观工业数据的市场'的假设。2026 Q1中国工业增加值同比+5.8%、机器人安装量全球占比超50%，但同期CPI仅+0.2%、PPI约-2.1%（连续约30个月通缩）、PMI从业人员分项持续低于50、制造业实际工资走平。强势完全可…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] Impact of Brent <$110", "summary": "致 主题研究员： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 4f015c35-a674-439a-a6d6-3ca8c61e9ed8 结果摘要: Wrote handoff_response.md (2602 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "AI 算力扩张不会单独制造全球通胀再加速，但会放大能源-化肥-粮食冲击，使软着陆路径更脆弱并压低 AI 硬件估值容忍度。", "summary": "2026-05-15：本卡支持 Research note 06 的方向，但将因果强度校准为“AI 是能源-化肥-粮食通胀链条的放大器，而非当前冲击的单一根源”。当前 CPI 冲击主要来自中东能源与航运扰动；但 AI 数据中心用电、局部天然气发电和电网瓶颈提高了能源价格持续性的右尾风险。若尿素维持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 0, "neutral": 2, "value": 50.8, "events": [ { "title": "Based on the macro fundamental data as of May 16, 2026, the S&P 500 (SPY) faces a significa…", "summary": "Based on the macro fundamental data as of May 16, 2026, the S&P 500 (SPY) faces a significant valuation ceiling at the 7500 ($750) level. While AI dr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "工业板块能为SPY $750突破提供部分但次可加性的支撑——A&D与电气设备约+30至+60 bps利润率扩张是真实的，但广度不足以独立替代NVDA。", "summary": "工作日锚2026-05-16。在10年期实际利率2.0%-2.2%、4月CPI 3.1%粘性通胀背景下，标普500工业板块未来两个季度经营利润率仅能扩张+30至+60 bps，远低于完成对TMT盈利广度接棒所需的+100-150 bps。扩张集中在航空国防（GE/RTX/LMT积压订单book-t…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 4, "neutral": 1, "value": 25.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI视觉与精密传感器单价的快速下降使设备折旧本身不再是长尾盈利门槛，但产能密度、集成、模型运维与合规四项软成本共同构成的楔子，正在以30%–40%的单位智能产线成本差结构性挤出底部3…", "summary": "截至2026-05-17，TMT视角的压力测试结论：折旧压力论真实但只对一半。AI视觉与边缘AI硬件单价自2022年下降35%–55%（Jetson Orin Nano Super USD 249、国产地平线/瑞芯微/寒武纪芯片），单线智能视觉栈从¥1.0–1.7M降至¥0.6–1.1M，标价已非…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工标准化是可信的质量分化选股主题，但不是可独立放大的全天候大类资产配置主题。", "summary": "截至2026-05-17，我否定把食品加工标准化主线外推为大类资产配置仓位。前序卡片的微观逻辑仍成立：AI视觉、B端标准化和上游一致性可把利润份额推向龙头。但中国4月食品CPI仍为-1.6%同比，非制造业新订单44.3、销售价格48.1，说明终端传导弱；同时PPI、油价和FAO食品价格上行更像投入…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "供应端结构性缺口足以支撑能源价格中枢上移和通胀粘性增强，但130美元油价本身仍需要扰动持续或恢复慢于预期。", "summary": "截至2026-05-18，本卡片支持能源价格高企具有供应端结构性基础。短期130美元布伦特仍依赖海湾扰动和库存快速下降，但油气田递减率、上游资本开支不足、备用产能变薄、LNG供应浪潮推迟，以及炼厂/中间馏分油瓶颈，均指向更高的能源价格底部和更大的上行尾部风险。BLS 4月CPI、世界银行能源价格预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 1, "neutral": 1, "value": 44.6, "events": [ { "title": "[auto-handoff] 消费分层下的乘数效应", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 65683484-38b9-45f1-8b66-f6398962c9ba 结果摘要: Wrote handoff_response.md (9289 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 2, "neutral": 2, "value": 43, "events": [ { "title": "云端CapEx挤出引发的存储通胀让美/欧/日换机曲线弯折但不击穿，对头条消费增长拖累仅5-10bp，真正的宏观信号是核心商品CPI再加速给联储/欧央行的政策路径增加摩擦。", "summary": "截至2026-05-19，云端CapEx虹吸HBM/CoWoS导致端侧设备ASP上涨50-100美元的冲击是真实的，但不会扼杀美/欧/日换机周期。三条结构性驱动（疫情装机潮到期、Windows 10 EOL企业换机、AI能力门槛）和高端买家低弹性把结果压成'量平/结构上移/ASP升'格局。对美+欧…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "只有同时跨过 NRDL 现金流、制造卓越 + 上游自主、海外合规产能三道门槛的中国生物医药龙头，才配享受先进制造估值溢价；其余资产应在脱钩压力下系统性折价。", "summary": "2026-05-19：综合 01–08 卡，中国生物医药产业链的 2026 投资 thesis 从'低成本制造 + 创新药红利'升级为'合规化的卓越制造 + 全球化产能 + 支付端现金流'三位一体。三道门槛——NRDL 支付现金流、连续制造 + 上游自主可控、海外 GMP 产能 + 合规证明——将…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "具备“新质生产力”与“出海韧性”的CDMO及上游耗材龙头，已在A股脱离传统医药贝塔，系统性获得了基于卓越制造与自由现金流的“先进制造估值溢价”。", "summary": "2026-05-19：A股策略分析证实，医药制造板块正经历深刻的估值重塑。机构资金显著流出同质化创新药企，加速向具备“新质生产力”特征的上游自主可控（耗材/设备）及CDMO龙头集中。微观结构上，这类资产的估值锚已从临床预期的PEG转向基于产能ROE与自由现金流的DCF模型。凭借30-50%的本土化…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "多肽折扣不应全板块外推，但在医疗设备Section 232结果公布前，应对器械/CXO出口篮子单独保留5-8ppt风险缓冲。", "summary": "截至2026-05-19，本卡对多肽之外的医疗器械与CXO供应链风险做压力测试。结论是：美国贸易限制不应被一刀切外推至所有器械/CXO，但必须区分三类风险。中国原产低端耗材已受Section 301高税率冲击，风险最高；IVD试剂和实验室耗材被医疗设备Section 232调查点名，属中高风险；高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 1, "value": 47.4, "events": [ { "title": "在A股做风格再平衡时，消费板块整体保持中性、内部杠铃化（超配高频低客单必需+政策beta、低配高端可选），同时对价值/红利做温和倾斜而不全面退出成长，是当前数据结构下最优的配置回应。", "summary": "截至2026-05-21，本卡支持第01、02卡主线，并把消费内部防御逻辑落到A股组合层面：消费板块整体保持中性，内部做杠铃式再平衡——超配高频低客单必需消费与折扣渠道（万辰、盐津、东鹏、洽洽、三只松鼠），标配以旧换新政策受益的家电与通讯器材，明显低配高端可选与重营销品牌型可选。风格轴上对应价值/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 夏粮收购与CPI食品项传导", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 7ce12d71-196d-49a5-b3eb-2244267d05d7 结果摘要: Wrote handoff_response.md (2957 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-21", "supports": 1, "risks": 3, "neutral": 1, "value": 44.8, "events": [ { "title": "中国PPI回升是上游成本推动而非需求拉动，PPI-CPI传导受阻将压低下半年再通胀弹性，并使人民银行更倾向流动性与结构性工具而非激进降息。", "summary": "截至2026-05-22，本卡支持前序判断：中国4月PPI同比2.8%与CPI同比1.2%的裂口主要来自上游能源、有色和化工，而生活资料PPI仍为-1.0%、食品CPI为-1.6%，显示成本压力被困在中游利润链条。地产、零售和民间投资偏弱限制PPI向CPI传导，也限制PPI在下半年的回升弹性。人民…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "4月农产PPI(-2.1%)与下半年CPI回升(1.5%-2.0%)的预期差，确立了2026 H2食品加工龙头盈利扩张的‘第二段窗口’，支持‘成本红利’逻辑。", "summary": "截至2026年5月21日，4月宏观数据显示食品产业链呈现显著的成本与终端价格脱钩。PPI总指数受能源驱动升至2.8%，但农副产品购进价格同比下降2.1%，提供了实质性的“上游让利”。随着2026年下半年猪肉周期反转和CPI预计回升至1.5%-2.0%，食品加工龙头将进入“温和提价+成本低位”的盈利…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工整板块在2026 H2 A股风格轮动中应作为防御性价值底仓配置，速冻预制菜与烘焙原料少数子赛道则按进攻性GARP卫星配置，二者通过70/30哑铃式组合实现。", "summary": "工作日2026-05-21。在2026 H2 PPI总指数2.8%、农副产品PPI -2.1%、CPI回升至1.5%-2.0%、生猪温和反转、粮油糖全球过剩的背景下，食品加工板块整体β属于防御性价值，但板块内部高度分化。建议采用哑铃式70/30配置：70%底仓配置乳制品、肉制品、调味龙头作为类债红…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-22", "supports": 2, "risks": 2, "neutral": 2, "value": 63.2, "events": [ { "title": "餐饮端“量增价缩”会迫使平价和外卖暴露度高的品牌食饮龙头重构关键供应链环节，但不会均匀冲击所有高端或强定价权品牌。", "summary": "截至2026-05-22，餐饮需求不是冻结，而是“频次上升、客单下行”。国家统计局数据显示，2026年4月餐饮收入同比增长2.2%，但食品烟酒及餐饮CPI同比下降0.8%；同时食品制造业和酒饮茶制造业PPI仍处通缩。百胜中国和瑞幸的2026年一季度数据表明，外卖占比、骑手成本、促销和材料成本正在压…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "作为 中国宏观分���师 (China Macro Analyst) ，我已经完成了对 Research note 03/08 的宏观数据压力测试，以验证国内零售端是否存在“高频低单…", "summary": "作为 中国宏观分���师 (China Macro Analyst) ，我已经完成了对 Research note 03/08 的宏观数据压力测试，以验证国内零售端是否存在“高频低单价”的渠道降级特征，并评估“消费冻结”假说的有效性。 核心研究发现摘要 1. 社零数据的结构性分化： 2026年4月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 粮价偏弱对 5—6 月 CPI 食品分项的拖累幅度", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 7ab279c6-33ab-4e87-a10d-a7933dd72b79 结果摘要: Wrote handoff_response.md (9551 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "半导体板块的高位流动性压力并非A股系统性风险，而是跨板块“高低切”轮动信号，溢出资金正加速抢筹估值触底的CXO与创新药资产。", "summary": "A股半导体板块的所谓“极致拥挤与流动性危机”实为典型的跨板块资金轮动前兆。医疗分析师指出，半导体资金正溢出并流入估值处于历史底部的创新药与CXO板块。同时，半导体的“安全溢价”逻辑与早年医疗器械国产替代极其相似，容易陷入估值陷阱。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 42.6, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年YTD的重磅BD交易印证了中国创新药的全球科技竞争力，而美国国防部1260H清单在2月13日的短暂列入与撤回闹剧，则彻底确立了创新药BD景气度与CXO地缘溢价折损的估值脱钩。", "summary": "本卡深度拆解了2026年Q1总计超600亿美元的中国创新药出海BD交易（石药/阿斯利康185亿、信达/礼来88.5亿、中生/赛诺菲15.3亿）及定价结构。分析表明，跨国药企倾向于为平台及技术支付溢价（首付比达6.5%）。同时复盘了2026年2月13日美国国防部短暂将药明康德列入1260H清单后一小…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 3, "neutral": 0, "value": 38, "events": [ { "title": "中国的外部压力不是单纯的利差问题，而是“美元高利率+高油价+农业投入品/贸易链条”的复合约束，财政必须围绕粮食安全和供应链韧性精准发力。", "summary": "截至2026-05-25，本卡综合支持前序“稳货币、积极财政”判断，但强调农业压力测试不是边缘补充，而是外部冲击传导到中国政策空间的核心通道。美国CPI再加速、FOMC维持偏紧、Brent高于100美元/桶、欧洲和日本也难以快速宽松，意味着中国普遍降息空间受限。中国1.3万亿元超长期特别国债和设备…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "USD 17 bn 农业订单是情绪级别变量，在任何资产负债表数量级上都不解决问题；居民资产负债表的约束变量是住宅地产。", "summary": "截至 2026-05-24，从房地产视角否定线索的残留框架：USD 17 bn 美国农产品订单对居民资产负债表的影响在任何可度量数量级上都不成立。RMB 122 bn 仅相当于居民总资产的 0.022%、住房贷款余额的 0.33%；而 2026 年 1–4 月房地产开发投资同比 -13.7%、住宅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 1, "risks": 2, "neutral": 0, "value": 53.4, "events": [ { "title": "医药不是泛化反弹交易，而是应围绕“创新变现 + 高端器械国产替代 + 集采后修复”建立精选研究篮子。", "summary": "截至2026-05-25，本卡支持启动大健康产业链专项研究队列。优先深挖 2252.HK、9996.HK、09969.HK/688428.SH、002821.SZ/6821.HK、300529.SZ。主线是创新药出海授权、CXO 新模态订单、国产高端器械替代与集采后企稳。核心风险是外科器械 VBP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-05-25，A股大消费与农业食品应从“通缩防御”转为“质量定价权+B2B效率+粮食安全资本开支”的组合框架。", "summary": "截至2026-05-25，本卡综合前序结论：平台流量挤压真实存在，但CPI、PPI和政策证据不足以把A股食品与农业配置锚定为宏观通缩交易。建议将食品农业主动袖珍组合调整为30%B2B食品供应、25%优质必选消费、25%粮食安全基础设施、15%农业科技、5%平台低价餐饮/弱零售期权。核心是买入可验证…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 3, "neutral": 0, "value": 40.8, "events": [ { "title": "[auto-handoff] 6月CPI食品分项与粮价传导", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: d9a2a94c-0ae8-49dc-97b7-8a001cf5a141 结果摘要: Wrote handoff_response.md (3347 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美欧原产地规则与中国出口许可在AI电力供应链上呈现70%价格机制、30%数量机制的格局，仅在Hi-B取向硅钢与NdFeB重稀土磁材两节点真正物理改道，其余主要通过推高到岸成本并对美国…", "summary": "美欧'熔炼与浇筑'原产地规则与中国出口许可制是两套高度非对称的工具：前者基于规则、机械执行但被18-30个月执法滞后稀释；后者基于裁量、时延即政策。综合压力测试：2026-2027年这套监管栈是70%的成本推动通胀税（AI电力设备到岸成本+20-35%，美国核心商品CPI+15-25bp）、30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "OPEC+延期减产+俄罗斯出口结构性下滑+美国页岩弹性失效+闲置产能集中沙特，共同将Brent推向$100基准情形；通过四阶传导可在8月把整体CPI同比抬回4.0%以上，核心CPI回…", "summary": "2026-05-27卡片判断：Brent现货$96-$98，未来6-8周首次站上$100是基准情形（约55%）。四条供给绳索锁定上行：OPEC+自愿减产延至2026年底、沙特财政平衡油价已升至$96、俄罗斯海运出口因$50价格盖板降至3.0 mb/d、美国页岩2026年增量降至+0.4 mb/d、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 1, "risks": 2, "neutral": 0, "value": 56.2, "events": [ { "title": "截至2026-05-27，中国宏观基本面支持先进制造和研发导向产业的定向资本开支上行，但不是全制造业、全民企的广谱资本开支周期。", "summary": "截至2026-05-27，本卡从信用、利率、财政产业政策、投资结构和价格利润五条线验证前序低拥挤成长主线。结论为support但需选择性表达：社融与M2仍保持合理增长，LPR处于低位，科技贷款和数字经济贷款明显快于总量信用，设备工器具购置投资是固定资产投资中最强分项，PPI和工业利润修复改善企业预…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 粮价向 CPI 食品分项的传导节奏与货币政策含义", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 01435eaa-7884-47e7-bf3c-bce3acd3b9c8 结果摘要: Wrote handoff_response.md (19363 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "2026-05-28 核心判断：美国电网瓶颈的宏观风险不是一次性CPI爆点，而是受监管电价的多年粘性和长久期公用事业融资供给对美债曲线的边际挤压。", "summary": "截至2026-05-28，本卡对电网/公用事业链做压力测试：美国大型电力变压器2019年进口依赖度82%，常见交期36个月、最高60个月；2026年关税框架把部分金属密集型电网设备纳入15%关税至2027年。结论是，变压器短缺不会单独制造广义CPI冲击，但会通过费率申请形成粘性的电价通胀底部；同时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 1, "risks": 1, "neutral": 0, "value": 65, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "\"session id\": \"2ae742c0 c3a0 49f0 a11d b180016b2890\", \"response\": \"针对“宽货币、弱信用”及信用极化背景下的大健康产…", "summary": "\"session id\": \"2ae742c0 c3a0 49f0 a11d b180016b2890\", \"response\": \"针对“宽货币、弱信用”及信用极化背景下的大健康产业链 Alpha 挖掘，本报告得出以下结论：在信用传导受阻的环境中，具备 “全球化现金流自偿” 、 “颠覆性技术替代…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 4, "neutral": 1, "value": 43.4, "events": [ { "title": "在国际农资价格高企与下游食品通缩的双重剪刀差挤压下，中游主打清洁标签的食品龙头由于刚性锁定天然原料，将面临极具分化的二次成本重估冲击：涪陵榨菜具有深厚的财务防御空间，而安琪酵母与千禾…", "summary": "本卡基于2026-05-30宏观能化剪刀差背景，量化评估了国内农资价格向国际收敛对中游清洁标签食品加工龙头的二次成本重估压力。研究表明，在下游食品CPI负增（-1.6%）制约下，调味品行业财务防御极限分化严重：涪陵榨菜凭借51.6%高毛利展现极强防御性；而安琪酵母在极端价格并轨下净利将转负（亏损9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医疗器械的真正瓶颈是医院资本开支节奏、集采利润压缩与买入承接容量共同约束，而不是高端设备需求本身消失。", "summary": "截至2026-05-30，本卡压力测试医疗器械买入腿：需求并未消失，2026年设备更新特别国债已安排RMB 185.1 billion，国家卫健委“十四五”大型设备规划也提供装机管线；但医院预算缩减、集采/DRG/DIP压缩耗材和IVD利润、以及迈瑞+联影双标的A股承接容量有限，使该腿不能作为无摩…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "测序生态切换带来的设备重置与重新验证成本，将对全球CXO基因业务及LDT提供商造成300至600个基点（bps）的短期利润率挤压，且成本传导呈现双极化与滞后性。", "summary": "本卡量化评估了BIOSECURE Act下全球CXO与LDT服务商面临的测序生态切换财务冲击。结果显示，全面设备重置带来约500万美元直接CapEx压力；CLIA/CAP下单个LDT重验证成本达15-25万美元；试剂成本上涨80%-100%。高合规项目可通过12%-20%的'合规溢价'实现成本传导…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 1, "value": 62.8, "events": [ { "title": "[auto-handoff] 消费另类数据‘量增价缩’对国民生产总值核算与服务CPI传导的敏感性分析", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 04edb64d-03d0-4f7b-82dc-b946f9a1c367 结果摘要: Wrote handoff_response.md (17425 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "大健康产业链细分赛道瓶颈深度解构：医疗器械与创新药的执行缺口分析", "summary": "把会话级\"30日激活率2.9%\"映射到医药产业链，最逼近的现实指标是国谈药首年医院进院率(~15%–25%)。准入后真实落地需穿越六道闸门：进院率(G2)、DRG/DIP点数下医生处方意愿(G3)、双通道配齐率(G4)、集采品种放量与临床惯性错配(G5)、商保支付缺位(G6)。以G2×G3×G4中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 2, "risks": 5, "neutral": 1, "value": 56.2, "events": [ { "title": "大健康产业链的低覆盖+高催化窗口集中在 ADC 平台、创新药对外授权、高端医疗设备国产替代、GLP-1 二线产业链四条主线的中盘标的，五只优先标的为科伦博泰、荣昌、药明合联、联影、华…", "summary": "大健康产业链 30 日激活率 2.9% 的冷区中，真正的预期差不在创新药龙头本身，而在创新药 License-out、ADC/双抗平台、高端医疗设备国产替代、GLP-1 二线产业链四条主线下的中盘标的。建议优先纳入研究队列：科伦博泰 6990.HK、荣昌生物 9995.HK、药明合联 2268.H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "“量增价缩”会把消费投资从总量复苏逻辑推向定价韧性筛选：高端白酒保留质量底仓但需下调批价期权，休闲食品只买能同时证明毛利率、费用率和渠道现金健康的龙头。", "summary": "截至2026-06-01，本卡对前序“量增价缩”主线做消费行业压力测试。结论是：服务消费修复并不等同于品牌消费复苏，社零、CPI与PPI共同显示平价渠道扩张、食品饮料终端价格偏弱和成本转嫁缺口。高端白酒仍有质量溢价，但茅台一季度毛利率和净利率已下行，批价相对官方自营价缓冲偏薄；次高端与区域白酒库存…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026H2总量宏观β因下游需求收缩与上游输入通胀双重挤压而深陷逆风，食品加工及大众消费大盘应全面收紧战线，增配B2B配料，重配现金防御，撤退B2C品牌。", "summary": "截至2026-06-01，最新宏观数据显示下半年消费大盘面临逆风：4月社零仅增0.2%（创2022年12月新低），CPI-PPI剪刀差恶化至-1.6%，食品CPI下跌1.6%导致下游B2C品牌定价权丧失。我们对投资组合权重进行最终定调：超配成本锁定且渠道免疫的B2B配料（24%），减配受两头挤压的…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 1, "risks": 5, "neutral": 1, "value": 49.6, "events": [ { "title": "B端食品加工龙头超配论点方向正确但乐观度过高；安井与千味因客户集中度、利润弹性、应收账款的结构性差异，估值与评级差距应被显著拉大，千味需由超配降至标配偏减持。", "summary": "用医药行业三条经典教训（集采对'规模+渠道'估值的击穿、CXO客户集中度从溢价转折价的拐点、创新药BD的工业化降本陷阱）压力测试Research note 01-04的B端食品加工龙头超配论点。结论：方向支持，乐观度过高。千味央厨前5大客户集中度47.9%已接近CXO拐点；Q1营收+23.83%/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "消费双轨制已在2026年4月社零/CPI数据中得到宏观印证，A股消费板块估值中枢的结构性下修不可逆，但B端供应链龙头与服务消费龙头将获得相对再定价红利。", "summary": "2026年4月社零+5.1%、服务零售+7.6%、CPI食品−1.0%/服务+0.6%的组合，从宏观总量数据印证了消费'双轨制'结构性成形。下半年实际GDP预计4.7%–4.9%，但平减指数仍为负，名义增长持续滞后实际。A股消费板块估值中枢已较2019–2021永久性下修35%–48%：传统C端品…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 9.1%, 22/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 9.1%, 22/241） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 1, "risks": 0, "neutral": 0, "value": 78, "events": [ { "title": "若10Y国债1.85%破位由核心CPI、PMI与社融质量共同验证，A股将系统性从防御红利切向顺周期和盈利成长，但应采用分阶段轮动而非一次性清仓公用事业。", "summary": "截至2026-06-03，支持“宏观共振后红利让位于顺周期与成长”的方向，但触发尚未完全成立：核心CPI和PPI已转强，PMI订单与库存、社融质量仍不足。若10Y国债连续5日站上1.85%，应将公用事业红利从6%-8%逐步降至2%-3%，资金优先轮动至有色、化工、机械、煤电一体化，再转入电子、AI…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 8, "neutral": 2, "value": 37.4, "events": [ { "title": "中国更可能把稀土、石墨等关键矿物作为许可型、可校准的谈判杠杆，而不是立刻发动全面禁运；但这已经足以让EV、工业电机和LEO补网承受系统性供应链压力。", "summary": "截至2026-06-05，最可能的中国反制不是全面禁运，而是在2026年11月两个暂停窗口到期后继续使用关键矿物许可、最终用户审查和选择性重启2025年10月稀土/人造石墨方案。对半导体龙头，直接晶圆厂停摆风险低于市场叙事，主要压力在GaAs/GaN、锗、靶材、设备磁材和高可靠认证；对EV、新能源…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月四个宏观窗口最可能给出温和支撑组合，足以维持卡3配置中枢但不足以将食品饮料升级为2026 Q3趋势主线；上下行概率虽对称但触发条件不对称，期望意义上偏下，正确仓位是不上调、按触发…", "summary": "2026-06-04：6月四个宏观窗口（PMI、CPI食品分项、PBOC例会、618/端午高频）在工作日期均为前瞻。本卡通过事前情景建模认为最可能场景是\"温和支撑、不证伪、不升级\"：制造业PMI区间49.3–49.7、食品CPI同比-0.8%~-1.3%、PBOC维持现状、618食品类目同比+5%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工链最值得挖掘的位置在低覆盖度 B2B 加工商，核心变量是餐饮与茶饮/烘焙渠道修复能否转化为利润和现金流。", "summary": "2026-06-04：本卡支持从 B2B 食品加工链切入，优先关注千味央厨、宝立食品、佳禾食品、立高食品、海融科技。公开数据给出的组合信号是餐饮强于商品零售、食品 CPI 偏弱、食品制造利润好于农副初加工，说明研究重点应放在连锁餐饮、茶饮/咖啡、烘焙和复合调味供应链，而不是一般包装食品。下一步建议…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 4, "neutral": 0, "value": 48.8, "events": [ { "title": "消费行业正从防御性品种转为‘隐形石油税’的利润受害者，高端白酒的财富效应溢价面临重估，而大众消费品正进入优胜劣汰的毛利保卫战。", "summary": "本卡片合成了能源冲击与贸易壁垒对消费行业的双重压力。研究显示，消费行业充当了‘隐形石油税’的避震器，表现为食品饮料行业因物流、包装和原材料成本上升而导致的剧烈利润挤压（PPI与CPI背离）。白酒行业面临‘财富效应蒸发’导致的结构性回调，而非简单的成本压力。此前看好的‘海外布局’战略在CBAM与全球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "离岸汇率闸门放行情景下国内货币政策空间与粮食安全资本开支传导", "summary": "假设离岸FX闸门打开，国内政策仍无法把粮食安全投资、高标准农田、农村融资与农业科技升级转化为广义信用脉冲，传导停留在防御层面。三道国内闸门收紧：国有行NIM跌至约1.42%历史新低（工行约1.3%）、LPR钉死3.0/3.5%[S4][S1]封顶风险放贷；高标准农田为多年期工程（2030年13.5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "农业链条没有提供通胀、需求或政策层面的正向催化来确认成长风格切换，粮食安全更应被纳入防御型哑铃配置而非高beta追涨依据。", "summary": "截至2026-06-05，农业视角支持否认“红利→成长”regime break。4月食品CPI仍是拖累项，猪肉和鲜菜供应充足；USDA FAS对2026年生猪、猪肉和MY2026/27饲料需求的判断偏稳定，粮食安全政策更多提供长期防御型结构主题，而非午盘成长股追涨信号。农业可作为实物红利/防御型…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策杠杆在NIM、财政纪律与食品通缩的硬约束下无法产生广义信用脉冲，风格切换证伪，维持防御哑铃。", "summary": "2026下半年风格切换（红利至成长）不成立。宏观、微观与政策三层闸门锁死：国有行NIM创1.42%新低压制总量信用扩张；财政有纪律将开支平滑为多年期工程而非脉冲；FX防御（7.30% HIBOR）闸断北向增量；食品通缩（CPI -1.6%）抽走再通胀预期。微观层面，6/8–6/19解禁高峰与738…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 3, "neutral": 0, "value": 56.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "智能辅具的高毛利与超额收益源于其NMPA医疗器械属性与底层技术（如BCI/AI）壁垒，“院内下沉家用”是其破局与规避集采的核心驱动。", "summary": "验证支持“智能护理/康复设备”的医疗器械Alpha逻辑。相较于低毛利的适老化家装，翔宇、伟思等头部凭借NMPA注册与AI/BCI技术壁垒，2025年毛利率稳定在65%-70%。2026年行业战略由院内向“便携居家化”转移，有效规避集采风险，精准收割中高收入家庭。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 2, "neutral": 0, "value": 63.6, "events": [ { "title": "高油价尚未足以确认制造业衰退，但已经让PMI扩张变成一个由能源成本、预防性采购和利润率分化共同支撑的脆弱扩张。", "summary": "截至2026-06-07，本卡对上一张全球宏观结论做压力测试：全球制造业PMI仍扩张，不能确认全面衰退；但$106布伦特已通过燃料、运费、CPI、PPI和高耗能制造业投入端形成真实传导。基准判断是：若油价按EIA路径从5月至6月约$106/b回落至4Q26的$89/b，通胀黏性上升但制造业可避免硬…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FDA监管收紧与《生物安全法案》导致平台型资产面临研发成本上升61.5%及估值收缩达51.5%的系统性下行风险，但通过‘新加坡/东南亚拆分式BD’及中游临床CRO数据隔离重构，可实现…", "summary": "基于2026-06-07，本报告系统压测了外部监管与《生物安全法案》对CXO及创新药平台利润的侵蚀风险。更换CDMO将导致研发成本上升61.5%、rNPV折价22.5%；WACC飙升至15%将导致企业整体估值收紧51.5%。然而风险高度分化：多肽创新药企业通过‘新加坡/东南亚拆分式BD’可挽回高达…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "虽然创新药与CXO估值处于历史底部且对云厂商Capex敏感度为0，但在地缘合规、三期临床高失败率及CRO订单传导时滞下，板块重估面临结构性压力，不宜盲目乐观。", "summary": "截至2026-06-08，本报告对AI主题资金轮动至生物科技资产（AI-DD与创新药）的逻辑进行了压力测试。测算显示，虽然该板块对云Capex零敏感，但极度易受全球折现率和地缘政治法案（BIOSECURE、BINSA）影响。rNPV模型显示，地缘合规导致的研发延迟与成本上升将压制估值约17.4%。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 3, "neutral": 0, "value": 60.4, "events": [ { "title": "白糖进口严控会抬高下游用糖成本底部，但在食品PPI偏弱和主要农产品价格受调控的环境下，它更可能压缩糖果、饮料、烘焙利润率，而不是显著推升食品CPI。", "summary": "截至2026-06-10，本卡综合前序糖业供给压力测试，判断白糖输入性压力更像食品饮料企业的利润率冲击，而不是食品CPI的主驱动。国际原糖约14美分/磅、广西现货约5290-5350元/吨，叠加配额和糖浆/预混粉监管，使国内糖价有政策底但上行受库存压制。糖果、全糖饮料和大众烘焙的毛利敏感度依次递减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "桶C应绝对聚焦于拥有4.5%以上股息率、国内OTC现金流及批零一体化壁垒的1099.HK、3320.HK与A股中药龙头，以此隔绝海外地缘风险并稳固组合底仓。", "summary": "截至2026-06-09，本卡对扩容至48.2%的桶C（防御医药）进行底层资产筛选。筛选标准要求股息率>4%、ROE>10%且对中美脱钩零敞口。结果显示，港股医药商业龙头（国药控股1099.HK、华润医药3320.HK，股息率约4.6%）与A股中药OTC龙头（云南白药000538.SZ股息率>5.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 3, "neutral": 2, "value": 67.8, "events": [ { "title": "[auto-handoff] 慢去产能周期下CPI食品项及H2通胀中枢敏感性评估", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 21e65805-b96d-4e1e-8071-581e93f19615 结果摘要: I will run the command 系统日期锚 to anchor today's date as requ…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "防御档以银行/公用/煤炭/食饮四大重仓+电子-700bp为对手盘；修复档反向，电子+700bp但须用北向条件触发分拆建仓，跨档摩擦约10-11bp迫使触发器严守。", "summary": "本卡在card-03的防御/中性/修复三档框架下，给出申万一级28行业相对沪深300基准的BP级主动权重。防御档（成长25/红利55）最大赢家为银行+600bp、公用事业+500bp、煤炭+450bp、食品饮料+400bp；最大输家为电子-700bp、电力设备/计算机各-500bp、医药-400b…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "整合负Gamma脆弱性、VaR超限与对冲缺口后，自营成长/科创敞口的战略定位", "summary": "综合Research note 01–05：阿特斯63.01%解禁后科创量缩价升，反弹实为负Gamma对冲+空头回补，杠杆单边（融资余额3792.73亿 vs 融券13亿）。今日美国5月CPI、6月12日SpaceX上市与科创50样本调整三事件叠加，99% VaR超限1.9–3.3倍，且期货对冲尾…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 3, "neutral": 0, "value": 63.2, "events": [ { "title": "在大健康估值底部与药明系地缘风险升级背景下，应通过全球BD出海现金流与自主产能进行地缘政治对冲，重塑医药板块的杠铃配置。", "summary": "本报告启动对大健康产业链（激活率4.4%）的首次覆盖。我们提出哑铃配置策略：超配全球BD出海强、具独立大分子产能的创新药龙头（如恒瑞Q1创新药占比首破60%），回避或低配依赖药明系CDMO进行美国临床的中小Biotech，因其面临药明康德6月8日列入1260H后高达4000万至1.1亿美元的CMC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘合规已迫使中资ADC/CGT创新药从“直接出海”转向“NewCo去中心化开发”，CXO订单转移已进入技术转移的实质执行期。", "summary": "截至2026年6月12日，受《生物安全法案》成法及药明康德被列入1260H清单影响，中资ADC/CGT企业正全面转向NewCo架构以规避合规风险，预计2026年超40%的海外临床由NewCo申办。同时，跨国药企订单正大规模向Samsung Biologics和Lonza转移，三星生物2025年订单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药不是AI基础设施约束的普遍避风港，只有能把算力转化为监管、临床、支付和采购成果的公司才值得溢价。", "summary": "截至2026-06-12，本卡综合支持前序“AI交易从订单叙事转向融资、电力与交付约束”的主线，但否定医药可作为宽泛避风港。Oracle的$638 billion RPO、negative $23.7 billion FY2026 FCF、4.5 GW Stargate容量和数据中心电力约束说明A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 2, "risks": 1, "neutral": 0, "value": 92.6, "events": [ { "title": "宏观政策与收入预期有限地、结构性地支撑Research note 02判断：功能性食品10-20%溢价更可承接（高收入30%+银发+备案制三因子对齐），预制菜5-12%条件性溢价天花…", "summary": "锚定2026-06-13，中国2026Q1居民可支配收入实际增速+5.0%、4月CPI+0.3%但食品CPI-1.0%、城镇失业率5.1%。政策面以旧换新和服务消费补贴未直接落到食品大类，但保健食品备案扩围、预制菜国标与服务消费意见同向支撑功能化转型。借鉴日本FOSHU与机能性表示食品在1995-…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基于HDVT框架，探究低激活率对生物医药项目BD估值与临床价值验证的影响。", "summary": "本卡以 HDVT 的临床粘性视角，分析 5%（14/280）低激活率对 Phase III 资产 BD 估值的传导。核心判断：低激活率是真实的负向价值信号，但它压缩的是商业实现因子（峰值渗透与持续性），不应下调 55% 的统计 PoS——混淆二者是主要陷阱。自测算示意：以传导系数 β=0.5、实现…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 1, "neutral": 0, "value": 86, "events": [ { "title": "结构性金属供给紧缩对全球通胀及美联储政策路径的影响", "summary": "支持（有界定）：铜/铝结构性缺口驱动的二次通胀足以硬化美联储终端利率逻辑，但机制是抬高底部、消除2026年降息期权，而非单独迫使加息。LME铜2026年6月初约13,800美元/吨、1月29日创14,527.50美元/吨历史高位，高盛上调年底预测至13,735；铝突破3,015.50美元/吨、对应…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 2, "risks": 5, "neutral": 1, "value": 74.4, "events": [ { "title": "“电力本位”人民币是结算和交叉盘再定价变量，不是2026-2027年立即改写全球储备秩序的变量。", "summary": "截至2026-06-15，本卡片综合判断：电力设备顺差与CIPS扩张会强化人民币在贸易结算、营运余额和CNH交叉盘中的地位，但尚不足以改写全球储备秩序。IMF 2025Q4数据显示人民币仅占官方储备1.95%，远低于美元56.77%。美联储因美国2026年5月CPI 4.2%而难以重启降息，PBO…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品自动化升级对半导体的拉动方向正确但量级有限——优选卫生级传感器与机器视觉的小盘纯标的，SiC/IGBT仅为期权式跟踪。", "summary": "截至2026-06-14，2026食品新国标驱动的450亿元高能效杀菌设备capex周期，对上游半导体的单机BOM拉动确实存在3-5倍提升（HPP/PEF/无菌灌装），但绝对增量约18-28亿元人民币（自测算），占中国功率半导体与传感器市场不到1%，不足以驱动板块β。结构性受益集中在三条线：卫生级…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "边缘AI以极低的单点功耗（7-15W）完美规避了2026年数据中心面临的吉瓦级电网瓶颈，是大规模工业自动化中最具能源协同效益的算力基建。", "summary": "本次调研分析了2026年边缘AI部署与大模型算力电力瓶颈的协同关系。数据中心AI推理带来严重的电网拥堵（单机柜功耗达100kW-1MW），推高了云端算力成本。相比之下，工业现场部署的边缘AI（如RK3588、Jetson Orin Nano）单节点功耗仅7-15W。在一家中型食品加工厂部署50台边…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 1, "risks": 3, "neutral": 0, "value": 76.8, "events": [ { "title": "[auto-handoff] 美国CPI/PPI背离对大类资产风险溢价及下半年美股板块轮动影响研判", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: f4668e4f-748a-467a-bfe5-e36469d590db 结果摘要: Wrote handoff_response.md (10465 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年下半年宏观环境呈现“CPI疲软与餐饮韧性并存”特征，大众刚需板块在“价弱”格局中，依靠收入与餐饮端的“量稳”基本盘叠加高毛利壁垒，“利润可守”的防御逻辑得到宏观数据的坚实验…", "summary": "2026年6月最新数据显示CPI/PPI剪刀差扩大（5月PPI同比3.9%，CPI仅1.2%），食品CPI下降1.7%，印证了下游消费端“价弱”现实。然而，一季度居民实际可支配收入同增4.0%，且1-5月餐饮收入同增5.0%，远超整体社零表现（4月仅0.2%），表明大众刚需具备极强的“量稳”特征。…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "支持但有条件：宏观风险溢价走阔时，BD估值防御性来自现金首付和全球证据，而不是headline总交易额。", "summary": "截至2026-06-15，本卡压力测试BoJ QT与EM风险溢价走阔对创新药BD估值的影响。结论是：具备不可退还现金首付、全球可解释后期临床证据和SOC一致试验设计的BD资产，抗压能力显著优于纯headline biobucks或零首付久期资产。康方/Summit ivonescimab是现金+全…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "盈利型商业化创新药可以承接组合切换，但必须从“有利润”进一步筛到“多产品、现金流强、医保/集采可替代性低”，否则单品高毛利无法抵消OAS/FX与政策共振。", "summary": "截至2026-06-15，本卡压力测试盈利型商业化创新药能否承接Research note 08的40%组合权重。结论为有条件支持：恒瑞、信达、BeOne类多产品商业化平台具备较强利润率、现金流和产品宽度，能够吸收+150bp OAS带来的约7.2%五年久期PV折损；但艾力斯/伏美替尼类单品高利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 2, "neutral": 0, "value": 76.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然人民币贬值光学利好CXO，但触发贬值的全球风险厌恶将冻结生科融资与CXO订单，且北向资金撤资将对医药核心白马股造成微观流动性挤兑。", "summary": "本报告将A股系统性流动性压力测试（Research note 05）映射至医药板块。结论认为，尽管USD/CNY贬值至7.30-7.42从账面上利好CXO企业的汇兑收益，但日元平仓引发的全球流动性紧缩将冻结海外生物科技融资，导致CXO订单流实质性萎缩。同时，600-900亿元的北向资金流出将重创外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 1, "neutral": 0, "value": 83.6, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 2, "neutral": 1, "value": 82, "events": [ { "title": "宏观层面的BD首付款虽能覆盖CXO利润损失，但因主体割裂、合规折让与同质化挤兑，该‘安全垫’在微观投资组合中无法实现跨板块套利，CXO的系统性出清与创新药IP化转型将呈深水区分化。", "summary": "截至2026-06-19，本卡片通过压力测试评估创新药License-out（BD）首付款抵消BIOSECURE法案下CXO利润流失的行业可持续性。测算表明，即使在重度压力下（CXO美线流失60%且净利率降至8%，BD首付款面临60%地缘与靶点饱和双重折让），BD首付款净流入（193.2亿元）仍可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DoJ数据安全计划下，高昂的技术合规规避成本彻底侵蚀了中方CXO的工程师红利，迫使前沿多组学及AI发现业务在商业逻辑上走向“二元脱钩”。", "summary": "截至2026-06-18，本卡片对CXO规避DOJ数据安全计划（DSP）的技术手段进行了压力测试。测算显示，云端TEE、联邦学习和在美隔离实验室等手段虽然可行，但运营成本增加30%-50%，人力成本攀升3-5倍，并产生20%-40%的研发效率损耗。这彻底侵蚀了中国CXO的工程师红利。技术性规避在商…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 油价回落与headline CPI缓和对国内消费需求的净传导", "summary": "致 消费行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: d5d23e30-7f8c-4941-90b4-d0523c113c1f 结果摘要: Wrote handoff_response.md (13103 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-19", "supports": 1, "risks": 2, "neutral": 1, "value": 91.4, "events": [ { "title": "饲料粮通缩对CPI食品分项是再通缩力量而非通胀缓冲，政策响应将是‘托麦不托玉米’组合，对总CPI同比拖累约−0.08至−0.16pct。", "summary": "芽麦替代封顶玉米 + 巴西大豆到港洪峰构成饲料粮双料通缩。直接传导（粮食、食用油）对CPI拖累不足0.04pct；主航道是间接传导——料价下行不会推高猪价，反而通过压栏增重与延迟去产能强化猪肉再通缩，预计2026年6–9月对总CPI同比拖累−0.08至−0.16pct[自测算]。政策端是‘托麦不托…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "CPI下行打开战术宽松窗口，但只有地产、就业和信用需求共同恶化，才会把PBoC推入实质性降息/降准循环。", "summary": "截至2026-06-19，CPI下行给PBoC提供宽松许可，但不足以单独触发实质性降息/降准循环。5月CPI同比+1.2%、核心CPI+1.1%、食品CPI-1.7%，而PPI+3.9%，价格信号是结构性分化而非全面通缩。地产投资、销售面积和销售额分别同比-16.2%、-10.8%、-13.5%，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "创新药BD龙头能对冲CDMO政策折价的方式是个股替代而非板块对冲：买恒瑞/BeOne的全球药物价值创造，不能简单买整个创新药篮子来抵消CDMO风险。", "summary": "截至2026-06-19，本卡压力测试医药资金从CDMO切向创新药BD龙头的可持续性。结论是：恒瑞和BeOne/百济神州能够在个股层面对冲CDMO政策折价，恒瑞靠可重复license-out与后期管线，BeOne靠BRUKINSA、TEVIMBRA和sonrotoclax等全球商业化/监管兑现；但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 1, "neutral": 0, "value": 87.8, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "大健康产业链 30 日激活率仅 5.4%（15/280），属被忽视而非基本面差。2026 年 Q1 创新药对外授权破 600 亿美元（约占 2025 全年 1357 亿的 44%）[自测算]，药明康德 Q1 营收 124.36 亿（+28.81%）、在手订单 597.7 亿、小分子 CDMO+80…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 9, "neutral": 1, "value": 44.2, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "大健康产业链低覆盖（30日激活5.4%），但基本面已反转。三引擎齐发：创新药license-out成全球定价事件（2025年约1357亿美元、占全球49%，2026Q1超600亿美元）；CXO订单重新加速（药明Q1营收+28.8%、在手订单597.7亿；凯莱英新兴业务+74.1%）；设备更新（20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品通胀对大宗/可选消费的需求破坏在西方与中国大众端方向成立，但中国白酒的破坏是结构性（禁酒令+地产信心+去库存）而非食品通胀所致，且利润率冲击分化——高端白酒是价/量风险，真正成本…", "summary": "压力测试 Research note 05 移交命题，我做出部分支持+因果重置的裁定。机制（实际收入挤压→可选消费需求破坏→利润率压缩）对西方/美国大众消费及大宗可选品成立；但 Research note 05 点名的中国白酒，其需求早已被破坏，成因是结构性的——禁酒令、地产引发的信心坍塌、渠道去…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在强势美元与国际谷物低价的双重作用下，低廉的进口谷物成为我国抵御输入性通胀的屏障，同时强劲的经常项目顺差为人民币汇率及国内流动性稳定提供了重要的宏观缓冲。", "summary": "本报告分析了截至2026-06-22，全球谷物低价与强势美元对中国宏观经济的传导机制。强势美元（USD/CNY为6.77）带来约4%的进口汇率拖累，但国际粮价大跌使得进口依然享有高达30-40%的价格优势。低名义进口账单与强劲出口推动5月贸易盈余达1054.3亿美元，有力防御了中美273.8bp的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "伊朗战争后的全球宏观转折：能源冲击消退，但美联储与日本央行维持鹰派带来的滞胀宿醉", "summary": "2026年伊朗战争（2/28美以空袭、3/4霍尔木兹关闭）将布伦特推至逾120美元，6/17美伊谅解备忘录使油价回落至约80美元。但通胀已从能源迁至核心：5月美国CPI总体同比4.2%（三年新高），核心升至2.9%，居住与食品双双加速。央行反应函数已上移——美联储6/17维持3.50–3.75%但…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 3, "neutral": 1, "value": 81.6, "events": [ { "title": "BIOSECURE Act 对中国创新药资产全球化估值逻辑的冲击评估", "summary": "BIOSECURE Act 于 2026-06-23 视角下已于 2025-12-18 随 FY2026 NDAA 签署生效，但其限制对象是联邦采购/拨款中涉及 BCC 的‘生物技术设备或服务’，不触及私营对外授权/BD 与成品药。Kelun-Biotech 经 Merck/MSD（大中华区以外授…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "生命健康和生物科技的最佳增量研究方向是围绕中国创新药资产可全球定价、器械集采后放量能力、以及上游工具需求修复建立小而深的研究队列。", "summary": "截至 2026-06-22，本卡支持激活生命健康和生物科技产业链研究。优先深挖 Kelun-Biotech、Mabwell、Ascentage Pharma、Peijia Medical 和 Sino Biological，理由是它们分别对应 ADC/创新药 BD、后期临床与商业化、血液肿瘤产品销…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "ADC偶联药物极高的CMC复杂性与全球无菌灌装/偶联CDMO产能的系统性紧缺，导致“去药明化”供应链重构将对中国ADC资产的海外BD权益价值（NPV）造成15%至35%的系统性侵蚀，…", "summary": "基于2026-06-23的研究显示，尽管《生物安全法案》在法律上将私营BD和成品药排除在联邦采购限制之外，但跨国合作伙伴出于准入合规顾虑，正加速推动供应链的“去药明化”。由于ADC偶联药物CMC具有极高的一体化技术壁垒，且全球替代CDMO产能十分有限（Samsung Biologics偶联车间已投…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 2026Q2 offline discount-retail / soft-home consumption read to validate outl…", "summary": "致 房地产分析师： 协同任务已完成。执行方为 消费行业分析师。 模式: adhoc 会话: b32a1550-d98c-456b-9162-aaadf7a235fe 结果摘要: Wrote handoff_response.md (12741 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 3, "neutral": 0, "value": 80, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "未覆盖机会集中在从科学故事转向商业化、医保、全球注册或集采韧性的公司，而不是宽泛押注医疗板块 beta。", "summary": "2026-06-24：本卡支持为生命健康和生物科技链条建立聚焦研究队列。优先深挖 Keymed Biosciences 2162.HK、Ascentage Pharma 6855.HK/AAPG、Mabwell 688062.SH/02493.HK，并在器械端保留 Peijia Medical 9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "折扣零售REITs可以作为收益资产，但农业链数据不支持把它们定义为无条件的消费平替防波堤；仓位必须接受食品价格与农资成本压力测试。", "summary": "截至2026-06-24，农业视角否定Research note 01的强版本结论：食品与畜牧数据支持“基础消费有韧性、蛋白价格偏弱”，但不支持“消费平替必然转化为奥莱REIT现金流护城河”。2026年5月食品CPI同比-1.7%、猪肉CPI同比-16.1%，更像商超、菜场和电商的必需品折扣竞争。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 3, "neutral": 0, "value": 81.4, "events": [ { "title": "铜及电网设备成本的结构性上涨将通过公用事业电价调整和工业投入品渠道转嫁给终端，抬高全球结构性通胀底座，迫使主要央行维持更高的限制性利率并加速日本央行货币正常化。", "summary": "截至2026-06-25，本卡评估了铜及电网成本压力向全球货币政策的传导。电网升级与AI算力用铜强度大，推高设备售价与监管资产基底，预测在3-5年内将零售电价推升1.5%至2.5%。50%铜价冲击将累积拉动全球CPI约0.25%至0.45%。粘性通胀使美联储年底前维持3.50%-3.75%高利率，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply s…", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 2, "neutral": 1, "value": 91.8, "events": [ { "title": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "summary": "压力测试整套对冲架构所依赖的55-65%国内通缩基准概率。结论：基准情形近期稳健，保留但下调至55-60%下沿。硬数据印证消费端通缩：2026年5月CPI同比+1.2%、食品-1.7%、核心+1.1%、环比-0.1%；10年中债贴地板于1.73%[S1][S5]。PBoC“边走边看”渐进（LPR连…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观需求疲弱足以削弱但不足以完全否定食品加工链的成本通缩红利，配置应从板块β降级为精选大众必选品阿尔法。", "summary": "2026-06-27：本卡对“成本通缩+量端复苏”做宏观需求压力测试。5月社零同比-0.6%、商品零售-0.7%、餐饮仅+0.6%，显示广泛量端复苏尚未确认；食品CPI、农副食品和食品制造PPI确有通缩，但整体购进价格、燃料动力和化工原料仍上行。利润数据进一步分化：食品制造利润仍增，农副食品加工和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 低价粮油替代对下半年食品CPI及生猪复苏周期的传导弹性评估", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: b8c58d06-e0d4-4616-8574-99ce2bd6c05c 结果摘要: # Step −1 — Anchor today's date BEFORE you start Run this i…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 2, "neutral": 0, "value": 91.6, "events": [ { "title": "去风险三期医药资产通过rNPV模型释放3.48亿美元的期权价值构成Biotech估值底，而生命科学工具在5.5x EV/Sales筑底、CXO在 overseas 利润收缩下于2.0…", "summary": "截至2026-06-28，本卡片基于BD估值方法构建了下一批并购标的与估值底线筛选框架。AbbVie 109亿美元现金收购Apogee与Merck KGaA 113亿美元收购Bio-Techne为医药生物确立了质量门控底座。我们量化厘定：1) 创新药三期去风险资产的rNPV floor：以12亿美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康低覆盖机会更像跨子行业alpha篮子，而不是单一医疗板块beta；优先研究能把2025-2026催化转化为收入桥的五家公司。", "summary": "截至2026-06-28，本卡支持开启大健康低覆盖研究队列，但不建议追宽泛板块beta。优先深挖Keymed Biosciences、Hybio Pharmaceutical、Sino Biological、MicroPort NeuroScientific和Allgens Medical，主线分…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 4, "neutral": 0, "value": 81, "events": [ { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）…", "summary": "截至2026-06-29，变压器交期延长和材料瓶颈导致美国电网资本开支通胀率达32.72%。结合2027/2028年度PJM容量电费暴涨11.5倍至$333.44/MW-day，将拉动美国零售电价上涨4.18%，直接和间接推高全国CPI 0.124个百分点。高粘性通胀与AI基建庞大资金需求将上移中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-8b8ede3135/", "networkHref": "/investor/research-network/#thesis-8b8ede3135" }, { "id": "thesis-e9630d3ee1", "source": "tracked_thesis", "title": "日元套息平仓是 EM 科技、离岸中资与高 beta 主题的尾部流动性风险", "titleEn": "Yen carry unwind is a tail liquidity risk for EM tech, offshore China, and high-beta themes", "lane": "日元套息/EM 科技流动性", "laneEn": "Yen carry and EM tech liquidity", "laneHref": "/investor/lanes/yen-carry-and-em-tech-liquidity/", "investableFocus": "日元套息/EM 科技流动性：FXY, UUP, EEM, MCHI, KWEB, EWJ, TSM", "stockUniverse": [ { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元融资压力", "roleEn": "美元融资压力", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "EWJ", "name": "iShares MSCI Japan ETF", "market": "US ETF", "role": "日本权益对照", "roleEn": "日本权益对照", "bucket": "core", "href": "/investor/stocks/EWJ/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "EM 科技权重", "roleEn": "EM 科技权重", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "QQQ", "name": "Invesco QQQ Trust", "market": "US ETF", "role": "纳斯达克成长暴露", "roleEn": "纳斯达克成长暴露", "bucket": "watch", "href": "/investor/stocks/QQQ/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "IBIT", "name": "iShares Bitcoin Trust", "market": "US ETF", "role": "BTC 现货代理", "roleEn": "BTC 现货代理", "bucket": "core", "href": "/investor/stocks/IBIT/" }, { "ticker": "COIN", "name": "Coinbase", "market": "US", "role": "加密交易基础设施", "roleEn": "加密交易基础设施", "bucket": "core", "href": "/investor/stocks/COIN/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "FXY", "UUP", "EEM", "MCHI", "KWEB", "EWJ", "TSM", "QQQ", "SMH", "IBIT", "COIN", "TSLA", "ASTS", "MU" ], "coreView": "赛道：日元套息/EM 科技流动性。候选观察标的：FXY, UUP, EEM, MCHI, KWEB, EWJ, TSM。当前共有 40 条证据、5 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 80, "eventCount": 160, "firstSeen": "2026-05-04", "lastSeen": "2026-06-29", "supports": 11, "risks": 131, "neutral": 18, "analystCount": 15, "sourceTypeCounts": { "研究笔记": 149, "分析协作": 8, "研究主题": 2, "研究线程": 1 }, "topTerms": [ "HBM", "GPU", "ASIC", "CPI", "美元", "TSMC", "AIDC", "CNH", "权益", "日元", "beta", "carry" ], "whatChanged": [ "2026-06-29: 物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进封装国产替代材料集中。", "2026-06-29: 日元套利平仓通过离岸人民币流动性抽水及掉期清算挤压双重渠道，系统性抬高离岸融资成本并技术性击穿USD/CNY 7.00关口，导致人民币套利空间彻底崩溃。", "2026-06-28: GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "2026-06-28: 基准配置应优先通过久期表达去通胀与政策失误风险，权益只做选择性低贝塔，商品低配整体贝塔但保留油价重上100美元的显式尾部对冲。", "2026-06-28: 电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。" ], "practical": { "sectorLane": "日元套息/EM 科技流动性", "sectorLaneEn": "Yen carry and EM tech liquidity", "exposureBucket": "风险偏好与高波动", "tone": "red", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元融资压力", "roleEn": "美元融资压力", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "EWJ", "name": "iShares MSCI Japan ETF", "market": "US ETF", "role": "日本权益对照", "roleEn": "日本权益对照", "bucket": "core", "href": "/investor/stocks/EWJ/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "EM 科技权重", "roleEn": "EM 科技权重", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "QQQ", "name": "Invesco QQQ Trust", "market": "US ETF", "role": "纳斯达克成长暴露", "roleEn": "纳斯达克成长暴露", "bucket": "watch", "href": "/investor/stocks/QQQ/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "IBIT", "name": "iShares Bitcoin Trust", "market": "US ETF", "role": "BTC 现货代理", "roleEn": "BTC 现货代理", "bucket": "core", "href": "/investor/stocks/IBIT/" }, { "ticker": "COIN", "name": "Coinbase", "market": "US", "role": "加密交易基础设施", "roleEn": "加密交易基础设施", "bucket": "core", "href": "/investor/stocks/COIN/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "FXY", "UUP", "EEM", "MCHI", "KWEB", "EWJ", "TSM", "QQQ", "SMH", "IBIT", "COIN", "TSLA", "ASTS", "MU" ], "primaryStocks": [ { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元融资压力", "roleEn": "美元融资压力", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "EWJ", "name": "iShares MSCI Japan ETF", "market": "US ETF", "role": "日本权益对照", "roleEn": "日本权益对照", "bucket": "core", "href": "/investor/stocks/EWJ/" } ], "instrumentTags": [ "FXY", "Invesco CurrencyShares Japanese…", "UUP", "Invesco DB US Dollar Index Bull…", "EEM", "iShares MSCI Emerging Markets E…", "MCHI", "iShares MSCI China ETF", "KWEB", "KraneShares CSI China Internet …", "EWJ", "iShares MSCI Japan ETF" ], "reviewCadence": "每日复核", "confirmationSignals": [ "期权伽马和 ETF 资金是否放大尾盘流动性", "社交热度是否转化为真实增量买盘", "波动率是否领先基本面恶化" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "波动率", "现金", "指数保护", "期权拥挤" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 133 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 15, "recentEvents": 133, "supportRatio": 0.07, "riskRatio": 0.82, "conflictScore": 91.8, "evidenceDepth": 100 }, "score": 73.6 }, "events": [ { "date": "2026-06-29", "title": "日元套利平仓通过离岸人民币流动性抽水及掉期清算挤压双重渠道，系统性抬高离岸融资成本并技术性击穿USD/CNY 7.00关口，导致人民币套利空间彻底崩溃。", "summary": "本卡支持Research note 02，定量分析日元套利平仓对CNH的传导。BOJ加息和USD/JPY破位触发约349亿美元平仓，占离岸池近25%，推高隔夜HIBOR至10.81%-12.74%。掉期展期率跌至40%引发即期购汇压力，击穿USD/CNY 7.00。CNH套利Carry-to-Risk比率跌至-0.492，套利交易崩溃。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.75, "matchedTerms": [ "美元", "CNH", "日元", "carry", "risk", "macro", "HIBOR" ], "href": "" }, { "date": "2026-06-29", "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-560亿美元收入，其25%低毛利率根本无法对冲高利润半导体供应链…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "HBM", "GPU", "美元" ], "href": "" }, { "date": "2026-06-28", "title": "中美利差倒挂对离岸人民币（CNH）汇率及跨境资本流动的传导效应", "summary": "支持卡片01的格局判断并补齐传导环节。截至2026-06-29，美国10年期国债约4.40%、中国10年期约1.79%，名义利差倒挂约−261bps（本轮最阔）；而真实利差约+39bps偏向中国。名义渠道在去杠杆中占主导，使CNH成为融资货币。鹰派美联储驱动的套利去杠杆经两条腿（CNH融资腿+全球日元套利传染腿）传导：CIP隐含12个月远期USDCNH约7…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "美元", "CNH", "权益", "日元", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗ASIC。先进封装成为泄压阀，但TSMC已提示先进封装产能紧张…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.5, "matchedTerms": [ "HBM", "ASIC", "TSMC", "beta" ], "href": "" }, { "date": "2026-06-28", "title": "基准配置应优先通过久期表达去通胀与政策失误风险，权益只做选择性低贝塔，商品低配整体贝塔但保留油价重上100美元的显式尾部对冲。", "summary": "2026-06-29，本卡综合并支持上一卡“央行在消退油价冲击中过度收紧”的判断，但把它转化为更可执行的大类资产配置：久期从35%中性升至45%，权益贝塔从45%降至35%，广谱商品从15%降至12%，现金升至8%。依据是美国5月CPI 4.2%中能源同比23.5%、核心仅2.9%，而美联储仍维持3.50-3.75%并上修2026通胀预期。油价尾部不能忽视…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "CPI", "美元", "权益" ], "href": "" }, { "date": "2026-06-28", "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM合约价Q2环比+58–63%)、以及电网电力(美国并网排队约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "HBM", "GPU", "美元" ], "href": "" }, { "date": "2026-06-27", "title": "日元套利交易平仓与离岸CNH流动性收紧对A股防御重估的交叉传导风险", "summary": "本卡以global-macro压力测试card-08的“USD/JPY与CNH流动性熔断器”构想：支持其核心方向（知识库已确认应以二者作量化触发器），但否定朴素的“单变量、按价位、OR门”实现。四点修正：(1)冲击分两波——先美元流动性挤兑、后避险重估，价位触发会迫使在第一波真空里抛售防御股；(2)两根保险丝相关而非独立，应合并为复合压力指标CTSI并设并…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22.75, "matchedTerms": [ "美元", "CNH", "日元", "risk", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "美元多头利差交易极度拥挤且极易受日央行干预与实际利率波动影响而无序平仓，USD/JPY跌破150会引发2150亿至2850亿美元美债清算并给在岸高两融科技板块带来资本流出压力，配置USD/J…", "summary": "截至2026-06-27，美元指数已站稳100点并突破100.735，录得近13个月新高101.20。然而，CFTC日元净空头堆积至-145.8K手（相当于113.8亿美元名义价值）的极端高位，表明美元多头交易极度拥挤。日本央行超预期将利率调升至1.00%，导致美日短期利差压缩至318 bp，USD/JPY刺穿160历史关口至161.265。我们压力测试表…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "美元", "CNH", "日元", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "在2026-06-27，AI电力并网限制（2年延误拖累53.71亿美元[S12]）将倒逼算力采购削减并重创高端半导体高增长假设，投资者应果断削减AI硬件至12%及长债至5%，转而增配25%短…", "summary": "截至2026-06-27，本报告重新校准了在高利率平台与电力之墙挤压下的多资产配置。2年电网并网延误对10万卡Blackwell智算集群将产生53.71亿美元搁浅资金损失[S12]，迫使云巨头在2027年削减20%GPU订单。这将导致台积电3nm和2nm产能利用率分别跌至85%与80%，CoWoS封装降至82%[S12]，HBM面临34.3%产能过剩且合约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.5, "matchedTerms": [ "HBM", "GPU", "美元", "权益" ], "href": "" }, { "date": "2026-06-27", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交期[S6][S7]限制的是通电而非下单。CoWoS售罄、交期…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "HBM", "GPU", "美元" ], "href": "" }, { "date": "2026-06-27", "title": "实施两期隔离（账户完全隔离与无融资杠杆）是防范两融折算率大降（0.6至0.4）引发6.0pp保证金缺口、确保商品期权3.0pp封顶与6pp黄金仓位安全的底线红线。", "summary": "截至2026-06-27，本卡对A股两融踩踏与商品期权保证金交叉传染进行了压力测试。核心结论是：若两融踩踏引发全市场折算率均值由0.6降至0.4，统一保证金账户将面临6.0pp的流动性缺口并迫使权益防御仓位平仓；而实施完全账户隔离、A股无杠杆（融资余额为0）及商品期权 butterfly 结构，可确保3.0pp保证金封顶不受侵害。下一步建议移交fx-str…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "CNH", "权益", "日元" ], "href": "" }, { "date": "2026-06-26", "title": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径", "summary": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径 Global Carry Unwind and Liquidity Siphon Risks for HK/A-Share Semi Leaders 日本央行升息及美元兑日元突破161.7，全球Carry Trade平仓是否会诱发离岸流动性虹吸（CNH HIBOR飙升），并对高流动性的…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 26.25, "matchedTerms": [ "美元", "CNH", "日元", "carry", "unwind", "liquidity", "risk", "HIBOR" ], "href": "" }, { "date": "2026-06-26", "title": "美联储'更高更久'与坚挺美元对人民币、日元及新兴市场货币的传导", "summary": "压力测试卡片 01 的强美元传染推论。结论：传导高度分化而非齐跌。日元是泄压阀且已破位——USD/JPY ~161.7、近 1986 低位，美日利差 ~275bp，日央 2026-06-17 加息至 1.00%、年初至 5 月约 ¥11.7 万亿干预全数回吐 [S2]。人民币偏强被防守——中间价 ~6.8166、CNH ~6.81，距 7.30–7.35 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "美元", "CNH", "日元", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "USD/CNY硬防御阈值与权益托底弹药消耗速度", "summary": "压力测试暴露出 Research note 01–04 的符号错误：2026-06-26 人民币是强势（中间价 6.8209[S1]），PBOC 在抑制升值（2-27 将外汇风险准备金率 20%→0%[S2]），外储升至 3.442 万亿美元[S3]。迫使 PBOC 由预期管理转向资产负债表干预（第 5 级抛外储）的阈值是：即期破 7.30/贴 ±2% 波…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "美元", "CNH", "权益", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "同步鹰派转向是真实的风险窗口，但不是已被证实的新紧缩制度；Research note 9 字面论点应被否定，前序宏观主线应被有条件保留。", "summary": "截至2026-06-26，本卡否定 Research note 9 字面交接文本：它是目录检查类工作流伪影，不是宏观论点。但压力测试后有条件支持前序经济主线：G3 6月鹰派转向真实发生，Fed 维持3.50%-3.75%、ECB升至2.25%、BoJ升至1.0%；美国5月CPI/PCE仍高，说明不能立即转鸽；但Brent回落至72.24美元/桶、EIA/I…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "CPI", "美元", "日元", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径", "summary": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径 日本央行升息及美元兑日元突破161.7，全球Carry Trade平仓是否会诱发离岸流动性虹吸（CNH HIBOR飙升），并对高流动性的中芯国际等港股与A股科技龙头造成无差别被动抛压？ Prior related institute work to acknowledge and BU…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 21.5, "matchedTerms": [ "美元", "CNH", "日元", "carry", "HIBOR" ], "href": "" } ], "timeline": [ { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "2026下半年AI资本开支拐点辩论：消化期还是再加速？", "summary": "TMT分析师判定，2026下半年AI资本开支并未进入消化期，而是进入结构轮动期：总量持平至上行，边际增量美元从训练优化的商用GPU集群转向推理fleet、超大规模云厂商自研芯片、以及电力与场地准备。空头叙事在需求侧（推理token volume、agentic渗透、多模态）证据不足，但在前沿训练边…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 2, "neutral": 0, "value": 39.4, "events": [ { "title": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯…", "summary": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯片竞争格局：NVIDIA Rubin对阵云巨头ASIC 本卡主题： WFE资本开支周期——先进封装与HBM设备订单可…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "修正后的 FX/CB 框架消除了大部分可避免执行损失，但组合本质仍是高 beta 中国 AI 科技账本，必须按 2.04 亿美元 VaR 和 2.18 亿美元最大回撤配置风险预算。", "summary": "纳入 CB Delta 调整、强制赎回现金储备和 CNH/CNY 双轨工具拆分后，5 亿美元港股 AI 算力组合在 CNH 贬值 10%、HSTECH 下跌 35%、CNH-CNY 基差扩大 600 bp 的同步压力下，修正 99% 压力 VaR 为 2.04 亿美元，即 NAV 的 40.8%；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI经济在架构层已多元化但在制造层重新集中于台湾3nm，这一咽喉点产生的地缘政治溢价已被TSM和日元部分定价、但在美国云巨头股权与利率波动率中显著低估。", "summary": "全球约99%的前沿AI芯片（≤3nm）晶圆与几乎100%的CoWoS-L先进封装仍集中在台湾新竹—台南走廊，亚利桑那、熊本、德累斯顿的分散化要到2028下半年才可能实质生效。当前TSM ADR与新台币已隐含约12-15%的台海风险溢价，但美股云巨头（MSFT/GOOGL/META/AMZN）几乎未…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "美元结构性下行已被cards 02–06的牛陡化和Fed最终降息预定，但路径上必经一次避险冲高——FX最优叠加是做多JPY作为尾部凸性、做多BRL捕获农业大宗底部carry、做空MX…", "summary": "截至2026-05-13，外汇综合判断：美债牛陡化+双赤字+Fed最终降息使美元结构性下行（DXY基准98→96，尾部94），但card-05双触发（BB OAS 260bps或SLOOS C&I≥45%）前会先有一次2–3%的避险冲高。EM选择上BRL>MXN——card-07农业供应冲击支撑巴…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "算力期货压缩的是商用GPU层租金，先进封装与高带宽互联的物理瓶颈与寡头结构足以将其非商品化毛利结构维持至2027。", "summary": "支持card-01框架并加以收紧：算力期货标准化的是GPU-小时这一输出单位，并未商品化决定谁能以最低成本清算的物理输入。CoWoS先进封装、HBM4、混合键合设备、scale-up交换机ASIC、PCIe/CXL retimer、1.6T/CPO光模块均具18–36个月结构性导入周期，寡头格局清…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI单名信用利差走宽主要是供给+杠杆+利率的周期性现象，半导体设备订单簿与超大规模资本开支指引仍坚挺，尚不构成成长逻辑的结构性破坏。", "summary": "压力测试卡片03的预警：截至2026-05-15，AI单名利差走宽约75%可归因于周期性/技术性因素（新发让利40%、资本开支融资杠杆漂移25%、利率beta 10%），仅约25%是早期基本面（ROIC/GPU折旧诚信）。Big-5 2026年AI资本开支合计约2900亿美元同比+45%，无一家削…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 3, "neutral": 0, "value": 29.6, "events": [ { "title": "A股半导体重估的 beta 已基本被定价，下一段必须靠后道设备、量检、材料、车规功率与先进封装上游材料的盈利与订单兑现，回避换手/融资/解禁三重拥挤的设计、代工与映射封装。", "summary": "截至 2026-05-15，A股半导体经过 4-5 月重估后，国产 GPU/ASIC 概念、HBM/先进封装映射股、大市值晶圆代工四条线已拥挤：5 日换手率分位 81-92%，半导体两融占自由流通市值比 ~4%（88 分位），主要芯片 ETF 4 周连续净申购 +18%，且 6-8 月解禁集中（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 生产率主线挺过 8%–15% 宏观回撤——溢价倍数被压缩，但盈利曲线不发生趋势性恶化；硬件核心仓继续持有，二阶新型云与软件溢价名应削减。", "summary": "在 global-macro 卡设定的 8%–15% 情绪回撤基线下，AI 硬件与算力 capex 方向上完好但盈利曲线边际走平、链条分化加剧。Hyperscaler FY26 capex（约 5,100–5,400 亿美元）已通过合同与电力 PPA 锁定，约束在电力交付而非资产负债表，对回撤敏感…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2H2026应把AI硬件瓶颈alpha当作有条件扩张的期权，把防御型红利质量当作有票息的保险，基准比例为18%对30%，随TSMC/NVIDIA检验点、VIX、利率和A股拥挤度动态调…", "summary": "截至2026-05-15，我综合支持前序“半导体杠铃”判断，但将其转化为2H2026大类资产规则：在中国/HK权益袖套内以18% AI硬件瓶颈alpha对30%防御型红利质量为基准，并根据流动性与波动率把AI仓位控制在12-28%、红利仓位控制在22-42%。Fed仍处限制区间、美国4月CPI升至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 1, "value": 46, "events": [ { "title": "云端CapEx挤出引发的存储通胀让美/欧/日换机曲线弯折但不击穿，对头条消费增长拖累仅5-10bp，真正的宏观信号是核心商品CPI再加速给联储/欧央行的政策路径增加摩擦。", "summary": "截至2026-05-19，云端CapEx虹吸HBM/CoWoS导致端侧设备ASP上涨50-100美元的冲击是真实的，但不会扼杀美/欧/日换机周期。三条结构性驱动（疫情装机潮到期、Windows 10 EOL企业换机、AI能力门槛）和高端买家低弹性把结果压成'量平/结构上移/ASP升'格局。对美+欧…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 算力链供应瓶颈已基本解决，支持在 CNH 企稳后回补权重至 50%，利润池正向 HBM4 与先进封装技术前沿发生结构性转移。", "summary": "2026-05-20 卡片支持将 AI 算力链权重回升至 50%。研究显示，HBM3e 良率提升与台积电 CoWoS 产能（年底达 7 万片/月）的实体扩张已基本消除 2024-25 年的瓶颈。虽然日元 carry 解除对日本设备标的造成估值冲击，但其基本面供应受损风险低。HBM4 转型将导致利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "AI与半导体能在防御性轮动中维持Alpha，但安全边际只存在于盈利可见度高、供给受限且能把CapEx转化为现金流的瓶颈环节。", "summary": "2026-05-20，本卡支持第4张卡的杠铃配置：AI与半导体可保留为15-20%的Alpha袖珍仓位，但不能作为泛防御Beta。SIA/WSTS显示2026年一季度全球芯片销售2985亿美元，环比+25%；TSMC 1-4月收入同比+29.9%，并指引2026年美元收入增长30%以上；NVIDI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "AI算力链是入口流量重估周期的有效Beta，但其有效性正受限于非线性的“电力墙”约束与存储成本上行，Beta效应已从“通用逻辑”转向“网络与存储”。", "summary": "截至2026年5月21日，AI入口流量（Gemini/Alexa for Shopping）与算力链Capex同步性极高，相关系数达0.92。智能体AI驱动的推理流量同比增长7851%，迫使Hyperscaler将2026年Capex推升至7250亿美元。压力测试显示，算力链Beta敞口正面临“电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "当前的弱美元叙事与 AI 驱动的 2.4% 增长及 4.2% 粘性通胀脱节，一份超 21.5 万的非农报告即可触发 DXY 重回 100 上方的机制转变。", "summary": "针对 Research note 01 的“弱美元”叙事进行压力测试。研究发现 DXY < 100 与 IMF/OECD 预测的 2.4% AI 引领增长及 4.2% 粘性通胀严重脱节。市场抢跑降息定价面临 6 月“过热数据簇”的瓦解风险。若 6 月 5 日 NFP 突破 21.5 万且 6 月 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "AI Capex韧性 × 油价冲击 × 联储延宕的共振，通过工业PPI与电力催生海外隐性二次通胀，将10Y美债锚定在5.0–5.3%，断裂点在日元/JGB而非美国风险资产。", "summary": "在2026-05-23工作日期下，AI硬件Capex韧性与霍尔木兹油价冲击的共振真实存在，但传导通过工业PPI、电力价格与资本品平减指数发生，而非头部CPI/PCE。这一通胀路径让联储基于PCE的反应函数系统性低估真实通胀，将10Y美债锁定在5.0–5.3%走廊延续整个2026下半年（期限溢价回升…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "台积电CoWoS与HBM底层封装良率构成全球AI芯片出货的硬性物理约束，超额利润集中于垄断性瓶颈厂商（TSMC/MU/SK Hynix），而A股代理股共享叙事却无壁垒且交易极度拥挤，…", "summary": "支持超配海外半导体物理瓶颈（MU/SOXX）并降配A股算力代理。台积电CoWoS封装产能从2024年底3.5万片/月扩至2026年底超12万片/月但仍售罄，NVIDIA垄断超60%份额；HBM市场2026年规模将达600亿美元，美光产能售罄且良率升至70-75%。A股光模块、服务器组装与IDC代理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "“贪婪定价”崩解将引发非线性的流动性风险，现有对冲仅能部分缓冲，必须立即筹措 3.5% NAV 的流动性备付金以防御保证金违约。", "summary": "压力测试显示，在 VIX 25、HY OAS 500bp 且日元升值 5% 的“三重冲击”情景下，10 亿美元 NAV 组合面临 8.5% 的毛回撤和 5.5% 的净回撤（计入对冲）。对冲工具抵销了约 32% 的损失，但日元套利与权益期货敞口将触发约 3500 万美元的紧急保证金追加。149 是套…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "消费者承担接近 100%、企业短期承担约 64%、CNH/MXN 几乎零吸收且 VND 远不足以对冲，关税新边际压力在 2026 下半年已显著消退但水平被嵌入；CNH 反向走强使中国…", "summary": "量化关税传导：截至 2026-04-08，美国有效关税税率 11.8%（Yale Budget Lab），核心商品 PCE 已上行 3.1%、对核心 PCE 通胀贡献 +0.8pp，传导 7 个月内基本完成；Dallas Fed 确认 2026Q1 见顶，4 月核心商品 CPI 几乎走平。短期分担…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 1, "risks": 2, "neutral": 1, "value": 59.2, "events": [ { "title": "美元指数回落至94.50与利率下行会显著支撑AI硬件估值扩张，但对资本开支的影响是强化既有需求周期，而不是单独创造新周期。", "summary": "截至2026-05-31，本卡支持“美元与利率下行有利于AI硬件及半导体估值扩张”的主线，但限定传导强度：最强作用来自折现率下行、弱美元折算、流动性和订单信心，而非单纯融资成本下降。Microsoft、Amazon、Alphabet、Meta的最新季度AI相关资本开支运行速率已超过1306亿美元，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 算力供给受限于 HBM4 售罄、CoWoS 封装良率及全球电网容量的物理极限，使硬件龙头成为 AI 价值链中无视‘Capex 断崖’风险的最终收费站。", "summary": "截至2026年5月，AI硬件供应链仍处于结构性紧缺。TSMC N2量产初期良率65-70%，HBM4于2月投产且美光2026年产能已售罄。NVIDIA Rubin架构推升单GPU功耗至2.3kW，使电力供应成为比芯片更硬的瓶颈。尽管终端应用出现DAU增速放缓，但硬件端的实物稀缺性构成了核心护城河，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "多云路由迁移并未在短期破坏NVIDIA绝对需求，但已通过Anthropic约10GW非NVIDIA硅片合同与推理占算力2/3的转折，结构性重写NVDA增量份额轨迹——NVDA与AVG…", "summary": "压力测试'模型层多极化→ASIC蚕食NVIDIA'的简化叙事。FY27 Q1 NVDA数据中心营收$75.2B同比+92%、Blackwell占DC计算约70%、每家超大规模云每周部署约72,000颗Blackwell，证伪短期需求断裂论。但Anthropic 10GW算力已合同锁定在Google…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 2, "risks": 0, "neutral": 0, "value": 77.6, "events": [ { "title": "截至2026-06-03，极端复合冲击下的核心不是追求更高防御收益，而是用美元抵押品、汇率凸性、成长空头和期权保护把物理红利仓位从被动流动性踩踏中隔离出来。", "summary": "截至2026-06-03，本卡支持将Research note 06的权益杠铃升级为15资产凸性流动性组合。情景输入为USD/CNY突破7.45、CNH HIBOR超过15.00%、10年期美债升至5.30%。组合核心是物理红利、海外AI电力公用事业、碳汇、美元现金、USD/CNH看涨、成长股空头…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "110-120美元油价引发的金属与电池材料通胀对AI芯片毛利无碍，但将系统性挤压缺乏定价权的AI服务器ODM及散热/电源供应商，导致其毛利率被动收缩60-90个基点。", "summary": "截至2026年6月2日，本卡支持上游材料成本上升的判断，并评估其对AI硬件的传导。110-120美元的油价将显著推升铜（母线/线缆）、铝（液冷分水器/散热器）及BBU电池的成本。尽管对具备极高毛利的AI芯片（GPU/HBM）影响微乎其微，但系统级组装厂（ODM）和电源/散热基础设施供应商将面临严重…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-03", "supports": 1, "risks": 5, "neutral": 0, "value": 41, "events": [ { "title": "该组合的关键外汇风险不是HKD脱钩，而是110.00%美元联动敞口在CNY/CNH升值时侵蚀净值和0.4651夏普目标，因此应采用部分、动态、触发式USD/CNH覆盖。", "summary": "截至2026-06-04，本卡综合认为，1.25倍消费组合虽然可维持ATAT.US 69.42%、1405.HK 35.58%、9922.HK 5.00%、603199.SH 15.00%的配置，但其未对冲外汇账本实质是110.00%权益资本的美元联动敞口。CNY/CNH每升值1.00%，对冲前约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DXY 99区间会弯曲但不会断裂——欧日央行被ToT冲击逼入墙角，但它们被迫的鹰派反应恰好阻止了EUR/JPY的失序崩溃，而黄金吸收了本应推动DXY破100的边际避险流。", "summary": "本报告支持并延伸卡片01的美元钝化结论：Hormuz封锁对欧日构成高度非对称的BoP冲击（自测算欧元区年化能源账单+1400-1800亿欧元；日本+6.5-8万亿日元），但ECB与BOJ均被困两难——能源CPI再加速阻止主动宽松，而ECB预计6月鹰派持有(60%概率)、BOJ预计7月象征性加息25…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "USD/JPY与USD/CNY不是直接拉爆A股硬件链融资盘的拉绳，而是通过SOX估值代理与人民币波动率→券商折算率传导的放大器，CNH/JPY交叉汇率是最被忽视的24小时前导信号。", "summary": "外汇视角支持Research note 01诊断：BOJ 6月17日议息加至1.00%已被OIS定价，未平仓日元Carry约5000亿美元（日资Q1净抛美债300亿美元已先行验证）。A股硬件链对JPY直接敞口<5%且传导6-12个月，6月议息当季不入P&L。真正传导通过四通道：①直接外汇敞口（弱）…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "整体研究链可发表，等级PASS WITH REVISION NOTES——R1（USD/CNH现货口径）与R2（BOJ议息日期）为出版前必改硬伤，但未触及核心结论与操作框架，研究链整…", "summary": "卡片01–07研究链最终审查：核心叙事自洽，2.9万亿融资盘、12%残余敞口、40bp三叉戟对冲、5,000亿美元日元Carry等关键数字跨卡保真，但存在两处必改瑕疵——Research note 03（中间价6.8184，即期约6.82）与Cards 04/06（USD/CNH 6.91）在同一…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 0, "value": 52.4, "events": [ { "title": "2026 06 04 第03张研究卡：AI硬件供应链成本压力与利润率韧性 —— 压力测试 立场：stress test（压力测试）。 截至研究院工作日期 2026 06 04 ，Re…", "summary": "2026 06 04 第03张研究卡：AI硬件供应链成本压力与利润率韧性 —— 压力测试 立场：stress test（压力测试）。 截至研究院工作日期 2026 06 04 ，Research note 01 [Research note 01] 与 Research note 02 [Rese…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面对人民币防线压力与科技 VaR 减仓的同向冲击,正确顺序是先降 beta、再加汇率保护、最后缩位保留 AI 物理层瓶颈敞口,而不是相反。", "summary": "USDCNY 由 7.20 向 7.35 逼近与 A 股科技 VaR 被动减仓属同向冲击,不能相互对冲。建议三步走:T+0–T+1 先用股指期货/ETF 期权把组合 beta 从约 1.0 降到 0.55–0.65;T+1–T+3 再分批买入 USDCNH 看涨期权或远期对冲 60%–80% 美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 9, "neutral": 0, "value": 12, "events": [ { "title": "USD/JPY向152.50逼近时，核心风险不是美债回购失灵，而是日元空头平仓迫使EM本币债、HY信用和高beta因子同步降杠杆。", "summary": "2026年6月5日，本卡支持并压力测试上一张外汇卡：若USD/JPY跌破156.50并逼近152.50，日元兑美元约升值2.62%，足以触发套息平仓。组合应先降新兴市场本币债，尤其MXN、BRL、ZAR、HUF、PLN、TRY等高carry-to-risk久期；其次降CCC/B、杠杆贷款、EM H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元套利平仓尾部情景下美股AI硬件3周回撤可达-22%~-30%，但基本面骨架未受FX直接破坏，3-6个月修复概率约65%，估值韧性中等偏强，真正被低估的是流动性而非估值。", "summary": "本卡对chief-risk的VaR突破警示进行板块级压力测试。截至2026-06-05，美股AI硬件估值/拥挤度均处历史高位（SOX远期PE 30.5x、Mag-7占SPX 33.5%、BofA FMS拥挤度68%）。以2024-08-05套利平仓为模板叠加拥挤度溢价，若USD/JPY从160急跌…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BoJ加息触发的日元套利平仓与美联储鹰派叠加形成'双因素美元强势'，USD/CNH将测试7.32-7.35，PBOC四档汇率工具箱（信号+HIBOR抽水+远期准备金+资本流动管控）足…", "summary": "若BoJ在6月16日加息25bp至1.00%并削减JGB购买，预计触发有序但显著的日元套利平仓，USD/JPY在48小时内从158回落至152-154；但与6月17日美联储鹰派维持叠加，将形成'双因素美元强势'，USD/CNH被推升至7.32-7.35测试区间。PBOC预计按既定四档工具箱序贯响应…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "若6月下旬汇率防御成真，7月货币政策将从宽松力度问题转为宽松顺序问题，基准是降息/全面降准后置、MLF中性偏缩量。", "summary": "截至2026-06-05，交接中“6月下旬三档汇率防御”仍应视为FOMC后的条件压力路径。若PBOC实际组合使用更强中间价、CNH抽水和20%远期售汇风险准备金，7月宽松节奏将被重排：不降LPR、不全面降准、MLF较5月1000亿元净投放更小或接近中性是基准。增长数据支持宽松，但PMI仍在50附近…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 5, "neutral": 1, "value": 40.2, "events": [ { "title": "若日央行 6 月加息并引发 USD/JPY 跌破 150，港股成长科技是中国权益中最先承压的美元流动性压力阀，A 股则需等待北向资金和融资盘确认。", "summary": "2026-06-06：支持将 USD/JPY 跌破 150 的日元套利平仓视为全球美元流动性冲击。最直接外溢对象是港股成长科技，因为港元联系汇率把美元融资压力输入香港市场，且 HSTECH 集中度、波动率和离岸融资属性更高；A 股受冲击更慢，主要通过 Stock Connect、北向资金成长持仓、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元平仓去杠杆下USD/CNH 7.25防线的挤压成本与政策边界", "summary": "日元平仓去杠杆下USD/CNH 7.25防线的挤压成本与政策边界 USD/CNH 7.25 Defense Costs and Policy Trigger Boundaries under JPY Carry Unwind 日本央行加息预期引发约5000亿美元日元套利平仓，外资流出压力加剧（北向…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在HY新云GPU-DDTL违约与GPU清算折价-51.0%的极端场景下，组合10日99% VaR将攀升至13.50%，必须通过大幅缩减AI Beta和HY信用并超配美元现金至48%和…", "summary": "本报告对组合在HY新云GPU-DDTL板块的10.0% NAV直接与间接敞口进行压力测试。若发生LTV达78%的违约清算（Hopper再跌30%且客户被降级），GPU资产价值将累计减值-51.0%，造成组合-3.73%的NAV折损。在此情景下，相关性收敛导致组合10日99% VaR从4.50%暴增…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面对'劳动降温+核心粘性'的非对称宏观，正确答案不是单一押注，而是质量股票+短端carry+精选长端的杠铃，并把黄金作为滞胀尾部保险。", "summary": "在5月就业降温但核心通胀仍粘（NFP +172k、失业率4.3%、核心PCE 3.3%）的组合下，建议总体股票风险中性、股票内偏质量（超配+3至+5 ppt）、固定收益采用'短端carry+精选长端'杠铃（现金/短期美债+4 ppt、1–3年IG+2 ppt、10年UST+1 ppt、10年TIP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 1, "value": 65.6, "events": [ { "title": "7月关税与海运上行更可能压缩AI硬件下游组装与网络设备利润率，而非击穿核心半导体龙头利润率；估值风险来自低利润率硬件收入增长的质量被重新定价。", "summary": "截至2026-06-08，本卡对第04卡的关税/海运鹰派尾部做TMT压力测试：风险真实存在，但主要不是核心GPU、晶圆代工或HBM利润率同步塌陷，而是下游AI服务器、网络设备、机柜、电源、散热和PCBAs的落地成本挤压。NVIDIA、TSMC、SK hynix拥有高毛利率和稀缺定价权；Dell式A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美债期限溢价的飙升通过推高名义利率、拓宽日元基差互换、翻正股债相关性以及抬高折现率等多重流动性与估值渠道，强力共振并放大了日元套利交易平仓对Mag-7拥挤资产的系统性清算抛压。", "summary": "本报告评估了美债期限溢价对日元套利平仓导致的权益资产抛压的放大机制。美债期限溢价飙升将推高名义利率与债市波动（MOVE指数），导致日元/美元交叉货币基差互换大幅拓宽，迫使日本机构清算美债并回流资金，从而引爆日元大幅升值及套利交易加速平仓。股债负相关性转正引发跨资产VaR飙升（投资组合10天VaR大…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-08", "supports": 1, "risks": 2, "neutral": 0, "value": 66, "events": [ { "title": "美联储2026年加息概率重估与全球流动性拐点", "summary": "支持卡片01。我把分析推至上游的美联储反应函数:5月CPI(2026-06-10发布)若证实再加速(核心环比≥0.4%、整体同比≥3.9%),美联储不会在6月或7月加息,但会把市场隐含的2026年加息概率从约28%重估至40–45%,并将点阵图叙事从'两次降息'翻转为'按兵乃至加息'。正是这一重定…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI 硬件供应链对 2026 全球经济构成轻度滞胀脉冲,央行降息会比无 AI 基准慢约 25–50 bp,2028 年起转为反通胀红利。", "summary": "AI 硬件资本开支已构成宏观级冲击:2026 年超大规模厂商指引约 4,000+ 亿美元,占全球 GDP 约 0.35%。基准情形下对全球 2026 实际 GDP 贡献 +0.3~+0.5 ppt,通过 HBM/CoWoS、铜与电力对美核心 CPI 推升 +0.1~+0.2 ppt;AI 生产率红…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "本研究周期在数据一致性、逻辑严密性及对冲方案的风险覆盖度上均达到高标准，正式通过评审并结项。", "summary": "本卡作为 10 卡研究周期的终篇，完成了对霍尔木兹海峡封锁背景下全球资产配置与对冲方案的质量与合规审计。经核实，全周期数据锚点一致（布油 138 美元、美联储 3.50-3.75%、美元/日元 168.45），宏观逻辑严密，成功实现了从能源冲击到跨资产对冲及权益风格轮动的战略转化。200 bps …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 5, "neutral": 1, "value": 52.4, "events": [ { "title": "AI半导体利润池迁移真实但渐进，板块总EPS上行、估值锚从'NVDA-TSMC双寡头溢价'切向'增长+多元化第二供应源'，ASIC设计商、HBM、先进封装设备与Intel Found…", "summary": "2026-06-09综合：Google-Intel TPU订单确认AI硬件利润池迁移真实且渐进，2027-28年AI半导体合计EPS较冲击前共识更高，但分布坚定向ASIC设计商(AVGO/MRVL)、HBM、先进封装设备及Intel Foundry期权倾斜。新估值锚为'增长+多元化第二供应源'，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Google-Intel的TPU订单标志着AI硬件利润池的结构性迁移：从绝对依赖GPU/TSMC，转向ASIC出货量反超以及Intel Foundry作为第二供应源的崛起，但GPU仍…", "summary": "2026-06-09评估：Google向Intel Foundry下单2028年超300万颗TPU，且Nvidia正评估Intel 18A工艺，这表明AI总需求依然强劲（到2028年TAM达3800亿美元），但硬件利润池正发生结构性迁移。为降低TCO，云厂商加速自研ASIC，出货量预计将超越GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "DeepSeek降价推动的是硬件内部的alpha重排，而不是AI硬件beta被证伪；应压缩泛化边缘和老库存，保留订单确认的光模块、HBM相关存储、液冷/电源和高端板卡。", "summary": "截至2026-06-09，DeepSeek V4-Pro 已验证永久降价为75%、至原价1/4；“96%”更像相对竞品的单位经济冲击。微观压力测试显示，GPU/HBM/800G与1.6T光模块/液冷仍受订单、价格和产能约束，Jevons弹性成立。中际旭创、沪电股份、佰维存储有收入、交付或稼动率确认…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 4, "neutral": 0, "value": 48.8, "events": [ { "title": "子篮36.45%的压力ES尾部是一条以日本央行2026-06-15/16决议为中枢、可定日期的外汇套息平仓尾部；CRO切换至港元计价HSBI认沽价差正确且剔除隐藏多美元押注，但唯有把…", "summary": "截至2026-06-11，外汇策略师综合并支持卡08首席风控官的结论：本卡主张生物科技子篮36.45%的压力ES尾部并非抽象宏观流动性冲击，而是一条可定日期的外汇套息平仓尾部，中枢为日本央行2026-06-15/16预计由0.75%加息至1.00%所触发的日元套息平仓——它会同时引爆σ_bio与ρ…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI需求到2026年底足以兑现HBM锁定的供给，但A股的真正约束是份额捕获与财政融资通量，因此应超配云挂钩供应商与一体化国产替代栈、把端侧AI作期权、低配离岸GPU依赖与重复建设长尾…", "summary": "锚定2026-06-10，本卡综合第01卡（首席策略师）与第02卡（政策分析师）。中国宏观叠加三条修正：(1) A股AI capex由大基金三期+省级算力券+头部云自融政策配给，民营信贷通道薄，是财政题非周期题；(2) 5月PMI 49.6 / Caixin 50.4 / 社融~8.4% / CP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "支持“全天候”防御转向，但应把基准风险从CNH即刻失守7.35改为利差倒挂、套保成本和日元carry回补的组合压力。", "summary": "截至2026-06-10，外汇压力测试支持将成长/科创β由1.07降至0.6、现金/短债升至12%的防御转向，但不支持把USD/CNH立即破7.35作为基准情景。USD/CNY约6.7765，距7.35约8.46%；真正风险在于中美10年利差约-282bp、6个月USD/CNY套保年化拖累约2.6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超配拥有确定性全球订单的光电液冷龙头与昇腾-中芯先进封装/HBM自主垂直栈，回避被截断离岸通道的二线芯片设计与空壳算力租赁标的。", "summary": "截至2026-06-10，本研究深入A股TMT板块。建议超配：(1)挂钩中美头部云、订单已确认的光模块（中际旭创、新易盛、天孚通信）、液冷（英维克）与电源龙头（麦格米特），均处于1.6T/Blackwell订单放量期；(2)华为昇腾与中芯国际（SMIC）挂钩的HBM/先进封装国产替代垂直栈，长电科…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 3, "neutral": 1, "value": 59.2, "events": [ { "title": "央行真实可动用流动性结构无法支撑'稳USD/CNY + 守科创50 1516'双线防御，BoJ鹰派情形下FX将系统性挤占股票托底弹药，1516点不是铁底而是概率底。", "summary": "在3.2万亿美元名义外储中，两周内真正可动用规模仅约2,560–3,840亿美元，且每100亿美元FX干预会同步抽走约700亿元人民币基础货币。BoJ 6月16日若鹰派操作，日元carry平仓将通过北向被动卖出与EM普跌通道击穿1516点。央行反应函数中FX政治优先级高于科创50经济变量：USD/…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 腾讯40亿美元双币种债成功定价对跨国企业离岸多币种融资及美日利差平仓效应评估", "summary": "致 债券分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: c771ebd8-7d32-4f43-920f-18a3db842a12 结果摘要: Wrote handoff_response.md (11901 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源融资成本上升会削弱云客户ROIC并压低部分ASP，但最强AI硬件平台商仍有足够利润率缓冲，真正脆弱点在重capex制造和周期性内存环节。", "summary": "2026-06-11，本卡压力测试能源债务融资成本上升对AI硬件供应商利润率的影响。结论是方向成立但不应外推为硬件利润率全面坍塌：新增能源债务成本上行约85 bps会压缩云厂商自由现金流并强化采购纪律，但NVIDIA和Broadcom因高毛利、平台锁定和轻资产结构韧性最强；AMD居中；TSMC和M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 1, "risks": 2, "neutral": 0, "value": 71.6, "events": [ { "title": "电力刚性约束已把AI加速器竞争维度切换为每瓦Token，Chiplet+先进封装是2027-2028电网窗口内唯一可规模化兑现的算力密度救援方案，对NPV的杠杆是5-8倍。", "summary": "电网刚性约束把AI加速器的KPI从峰值FLOPS切换为每瓦Token吞吐。Blackwell B200单卡1000W、GB200 NVL72机柜120kW已触顶，下一代Rubin/MI400无法继续抬TDP，必须靠能效换性能。TPU v5p、Trainium2较H100能效提升30-40%；同10…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "黄金-权益相关性转正是可防范的流动性压力情景，但 +0.55 目前更适合作为 5-15 个交易日压力参数，而不是基准相关性。", "summary": "截至 2026-06-12，日元干预导致的离岸美元流动性收紧是真实风险，但黄金-权益相关性 +0.55 不应作为基准。财务省披露 2026 年 4 月 28 日至 5 月 27 日干预 ¥11,734.9 billion，5 月末官方储备较 4 月末下降 $77,107 million，证券项下降…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力基础设施架构的重塑与硬件需求演变", "summary": "支持卡片01。当并网容量（兆瓦）成为固定稀缺输入，优化目标从“每机柜最大FLOPS”翻转为“每交付兆瓦最大有效算力”。这并非密度撤退——GB200 NVL72仍达120kW/机柜、800V路线图指向880kW[S1][S19]——而是密度转为服务效率。三大倒逼因素：PJM并网周期升至>8年[S5]…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 5, "neutral": 2, "value": 53, "events": [ { "title": "算法效率不会终结暴力算力，只会让暴力算力转移到推理与 Agent 层；NVIDIA 溢价将边际压缩但不会系统性回归，因为护城河在网络/软件/供应链而非 FLOPs。", "summary": "DeepSeek V3→R1→V4 路径压缩了训练成本，但 Jevons 效应把算力需求推向推理与 Agent 层，NVIDIA 护城河主要沉淀在 NVLink/CUDA/CoWoS-HBM 锁定/主权 AI 需求四层，并非裸 FLOPs 租金。1 月 27 日单日蒸发 5,890 亿美元市值是一…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "CoWoS绝对产能正在改善，但GPU与自研ASIC需求吸收速度更快，因此2026年AI集群交付仍受先进封装、HBM和关键设备交期的结构性约束。", "summary": "截至2026-06-13，本卡支持上一张AI基建卡的判断：硬件交付风险不只来自并网，还来自CoWoS/HBM/设备交期的半导体转换链。TSMC仍称先进封装非常紧，TrendForce口径显示CoWoS-L/S满载且2026年底目标120,000-130,000片/月；DIGITIMES虽更乐观，预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红皇后竞速并未彻底抹去自研ASIC的成本优势，而是将其集中至Google与AWS两家、使其他云厂商回到对NVIDIA的依赖，从而部分推翻了Research note 05所述的全行业…", "summary": "在'红皇后竞速'下，自研ASIC单世代有效寿命已压缩至12–15个月，2nm节点全口径NRE攀升至USD 0.9–1.4B、N2晶圆约USD 30,000/片、CoWoS-L售罄至2027。我们测算自研ASIC在TCO上跑赢商用GPU的年部署门槛从5nm节点的约20万颗升至现实情景下的15万–25…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "6月央行双会周（日本央行6月15-16日加息至1.00%、美联储6月16-17日按兵不动）叠加正在降温但仍高企的中东石油冲击，对全球风险资产与日元套利交易的双重风险", "summary": "进入6月16日当周的央行双会，日本央行预计加息25bp至1.00%、美联储料以约97-99%概率维持3.50-3.75%不变。5月美国CPI同比4.2%为2023年4月以来最高，但约60%涨幅由能源驱动，且布伦特已从3月底逾$113峰值回落至约$89（6月11日因美伊降温下跌逾4%），通胀尾部正在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 5, "neutral": 1, "value": 59.4, "events": [ { "title": "GPU抵押贷款收缩会削弱边际Neocloud订单可见度，但先进封装与先进制程需求更可能从融资套利订单转向信用更硬的超大规模厂商和ASIC路线，而非进入全面下行。", "summary": "截至2026-06-15，本卡压力测试GPU抵押融资收缩对半导体订单的影响：冲击不是AI芯片需求消失，而是订单可见度从Neocloud融资驱动采购转向超大规模厂商、ASIC、主权项目和带预付款的硬合同。NVIDIA Data Center在FY2027 Q1仍达752亿美元，TSMC先进节点和HP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在AI链内部按融资质量轮动：超配Tier-1平台与卖铲人(设备/HBM/封装/电力赢家)，低配靠GPU抵押融资循环的Neocloud股权与高收益债。", "summary": "综合前八卡推演，AI基础设施的约束已从硅到电力、再到融资可得性。本卡给出按融资质量过滤的五篮子相对配置框架：B1 Tier-1 AI平台(+300bp)、B2 设备/HBM/先进封装(+250bp)、B3 ASIC定制(+100bp)、B4 电力瓶颈赢家(+150bp)、B5 低质量AI基建(-2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "PBOC会用中间价、离岸流动性与宏观审慎工具优先稳定USD/CNH预期，把利率与流动性工具留给内需——不是二选一，而是分工。", "summary": "截至2026-06-14，若6月FOMC确认下半年不降息（区间3.50%-3.75%，中点3.625%），PBOC并不会在稳CNH与撑内需之间二选一。其反应函数分6层：中间价指导（2026-06-13为7.1869，显著强于即期6.76）、CNH央票、外汇风险准备金率、逆周期因子、国有大行即期干预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储加息尾部风险正在重构全球套利交易，强化 USD/CNY 上行压力，但中国政策缓冲仍足以把压力维持在有管理区间内。", "summary": "截至2026-06-15，本卡片支持上一张卡片的鹰派美联储判断。美国5月CPI与就业韧性使美元现金收益成为全球套利交易的新门槛，日元融资仍便宜但 BOJ 风险上升，欧元融资因 ECB 加息而吸引力下降。USD/CNY 面临中美政策利差、美元收益和能源冲击带来的压力，但 CFETS 中间价机制、3.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 4, "neutral": 0, "value": 55.8, "events": [ { "title": "验证AI算力硬件链的估值承压：流动性虹吸下的先进封装与GPU产业链韧性评估", "summary": "本卡对AI算力硬件链做压力测试，得出'分化韧性'结论。资金流层面确认虹吸：2026-06-08半导体−219.33亿、通信−98.35亿、消费电子−69.92亿元净流出居前[S6]，年内93家半导体公司减持[S7]，叠加CXMT约295亿IPO锚定[S9]。但估值下移是资金流/轮动驱动、非盈利驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观信用收紧并未击穿全球AI周期，而是将AI基础设施层切割为「超大规模云厂＋央企国家队」核心（继续扩张）与「民营商业云」外缘（防御性收缩）的K型结构，TMT投资应聚焦CoWoS/HB…", "summary": "TMT综合：沃什鹰派持稳（3.50–3.75%）经离岸人民币流动性、汇率对冲摩擦与企业中长贷收缩，传导至AI基础设施层产生K型分化而非总量收缩。美超大规模云厂CY2026合计capex运行率3,200–3,600亿美元由经营现金流支撑不受冲击；央企运营商云经PBOC科技创新再贷款（约2.5%）继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据中心的物理功耗墙迫使超大厂将有限的电力配额向具有3-5倍能效优势的ASIC倾斜，实质性加速了算力硬件市场从“通用高溢价”向“定制化能效套利”的重塑。", "summary": "2026年，电力配给成为AI基础设施的核心硬约束。四大超大厂高达6300亿美元的资本开支正大规模向功耗更低、每瓦特产出更高的定制化ASIC倾斜。通用GPU飙升至2300W的单卡功耗和超120kW的机柜密度让电网不堪重负，促使ASIC以近45%的增速抢占推理市场，从而实质性削弱了英伟达的算力溢价。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 4, "neutral": 0, "value": 57.2, "events": [ { "title": "电力供给限制驱动的半导体能效转型主要表现为通用GPU的系统与精度级优化，而非ASIC/NPU的大规模替代；自研定制芯片的爆发反而加剧了全球先进制程与封装供应链对TSMC的物理集中度。", "summary": "截至2026-06-17，电力瓶颈确在倒逼算力能效转型，但并非单纯的通用GPU向专用ASIC切换。算法多变性（如MoE与动态推理）使自研ASIC面临极高的“代差失效”风险，通用GPU（如NVIDIA Blackwell）凭借软件与架构灵活性仍占据70%以上份额。能效提升主要通过系统精度（FP4/F…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "人民币不是'代理日元'，而是同向risk-off的承压资产；JPY强平将通过USD回补与北向撤资双通道对Q3形成USD/CNY 7.30-7.42、A股-8至-12%、PBOC 1.…", "summary": "约1万亿美元的人民币套利在融资货币上以USD为主（出口商应收款延迟结汇、境内美元贷款、离岸CNH融资三大块），与JPY套利结构不同。日元强平触发的risk-off中，CNY历史上不是'代理日元'而是风险资产：2024-08-05闪崩日USD/JPY -3%但USD/CNY +0.35%。强平情景（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装的去瓶颈把AI算力的约束推向电力与液冷，2026–2027年边际capex美元流向'电子与水'而非硅片，散热与电网设备是该传导链上确定性最高的边际增量。", "summary": "截至2026-06-16，TSMC CoWoS等效晶圆产能从2024年底约35k wpm扩张至2026年底约75–80k wpm，HBM/TSV封装侧瓶颈基本释放。约束顺势外迁至电力与液冷：Blackwell B200 NVL72机架功率达约120 kW、Rubin级150–200 kW/rack…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体能效能把AI扩张从“堆GPU”改写为“每MW token产出竞争”，但只能买时间、改善单位经济性，不能让电力瓶颈消失。", "summary": "截至2026-06-16，本卡压力测试“电力预算驱动扩张”逻辑：半导体能效提升真实且重要，Blackwell、先进封装、vLLM/FlashAttention、稀疏MoE和定制ASIC可显著提升每MW有效token产出。但它不能完全对冲电力瓶颈，因为美国数据中心用电量可能从2023年的176 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 2, "neutral": 0, "value": 70.6, "events": [ { "title": "FOMC 后应以“选择性算力基础设施挤空”而非“无差别 AI 融涨”来交易：多稀缺、空低 ROIC，净多但对冲。", "summary": "2026-06-18 策略综合：前序链条成立，2027 风险是 ROIC 与结构分化而非 capex 悬崖。FOMC 后应做多稀缺算力资产而非泛 AI beta：前沿逻辑/设备/HBM/先进封装、自研 ASIC 与网络、电力与电网受益方应超配；商用 GPU 龙头保留但严控风险；低 ROIC 商用 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 能源重定价压的是超大规模厂商 ROIC 而非芯片 capex 效率——硅片主导 TCO、能源占比单位数，且性能/瓦溢价+杰文斯回弹反而净支撑前沿逻辑/先进封装/HBM 需求…", "summary": "2026-06-18 压力测试：能源是证伪芯片逻辑的错误杠杆。Epoch 1GW 机房年化 TCO 约 $8.5B，服务器占 ~59%、能源仅 ~7%；交付电价 +30% 仅抬升 TCO 约 +2.1%，翻倍也仅 +7%（自测算），而 IT 寿命 5→3 年使 TCO +41%——刷新节奏的影响是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 1, "risks": 2, "neutral": 0, "value": 80, "events": [ { "title": "RUM的Together AI合同仅占用其未商业化电力的约3%，而完全填满200 MW所需的47亿至51亿美元资本开支面临巨大融资缺口，且TSMC CoWoS与HBM3E供给紧缺导致…", "summary": "截至2026-06-18，对RUM Group的B300 GPU采购进行供应链与财务敏感性测试显示：其与Together AI签署的2.7亿美元B300液冷云服务协议在物理规模上仅需约1,800至3,000张GPU，对应电力功耗仅约4.8至7.3 MW，仅占其剩余200 MW未商业化容量的2.4%…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "RUM是合规触发器而非汇率资产负债表；只有当Tether的离岸代币化美元网络被监管冲击实质性压缩时，压力才会通过OTC美元溢价、CNH资金挤压和carry去杠杆反噬USD/CNH稳定。", "summary": "截至2026-06-18，本卡压力测试认为：RUM政治云受阻本身不足以冲击CNY/USD或USD/CNH，因Tether对RUM的$150 million GPU服务承诺仅约为USDT $186.384 billion市值的0.08%。真正宏观通道在Tether：若GENIUS Act框架、境外发…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the stress test report for Research note 8/8 of the session. The report sy…", "summary": "I have completed the stress test report for Research note 8/8 of the session. The report systematically maps out the transmission of an offshore toke…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 7, "neutral": 1, "value": 46.4, "events": [ { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 新任主席沃什强硬加息警示下美元兑日元突破161与套利交易平仓风险评估", "summary": "致 日报总编： 协同任务已完成。执行方为 外汇策略师。 模式: adhoc 会话: 2780edee-b126-4d5f-9a32-fcb0bfc721f3 结果摘要: Wrote handoff_response.md (13924 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美日利差收窄叠加沃什规则推升美债波动率，已将日元carry-to-vol比率压破历史平仓阈值，触发三层资金的非线性平仓并放大主要日元交叉盘波动率。", "summary": "美日10年期利差从年初约280BP收窄至2026-06-18的183BP，叠加沃什主席废除前瞻指引推升MOVE指数，使USDJPY的carry-to-vol比率从0.32降至0.12，跌破历史平仓阈值0.15。日元套利平仓分三层：CFTC杠杆基金日元空头已减约47%（剩约-9.8万张合约），离岸日…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球二次通胀预期退坡、美欧央行降息周期重启及日元套利平仓的连锁传导", "summary": "2026年6月美联储（维持3.50-3.75%、偏鹰）、欧央行（加息至2.25%）、日央行（加息至1.0%）同步鹰派，唯一共因是伊朗战争油价冲击（5月CPI 4.2%、能源贡献当月涨幅60%+）。美伊备忘录把WTI封顶$75-77后，移除了这一变量：克利夫兰联储整体CPI临近预测将随基数效应快速回…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 5, "neutral": 0, "value": 56.8, "events": [ { "title": "人民币走弱对AI算力资本开支的主要影响不是需求崩塌，而是把进口GPU采购从工程扩张问题改造成受外汇、许可证和国产替代共同约束的财务配置问题。", "summary": "工作日期2026-06-21。本卡压力测试前序CNH逻辑：人民币波动会抬高美系GPU进口预算，但更可能导致预算重排、分批采购、对冲纪律强化和异构集群替代，而非AI算力资本开支突然停止。以2026-06-19 USD/CNH 6.78440和160亿美元H20订单规模测算，CNH贬值1%/3%/5%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然HBM国产化良率崩塌会将集群 CapEx 推升 98%，但中国互联网巨头通过异构分级部署与算法优化（缩减30%GPU需求）并结合西部绿电直连套利（LCOE降至0.39元）与液冷改…", "summary": "评估HBM成本压力对AI集群TCO及扩容策略的传导。100%国产材料导致HBM3封装良率崩塌至4.11%，推升10000颗GPU集群CapEx至8.17亿美元，东部煤电下3年TCO达10.17亿美元（+66.13%）。建议协同优化：架构上采取60% HBM3与40% HBM2E的分级部署，结合算法…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本压力对AI算力基础设施运营经济性与扩容节奏的影响", "summary": "压力测试并在核心机制上否定使命链条（油价×汇率→工业电价→AI经济性→扩容放缓）。链条两处断裂：(1)选错杠杆——电力是AI算力经济性的二阶变量，1GW数据中心年化TCO约85亿美元中能源仅约6亿（≈7%），服务器约60%；电价+20%仅抬升总算力成本约1.4%，不构成扩容开关，资本开支（GPU）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2028年AI芯片扩容的材料次生瓶颈主要是高端基板合格输出和翘曲受控良率，TIM重要但更像可靠性与性能折让瓶颈。", "summary": "截至2026-06-21，本卡支持但收窄Research note 05：先进封装材料会成为AI芯片扩容的次生瓶颈，但真正硬约束更偏向高端基板/ABF/SAP良率，而非TIM原材料短缺。TSMC确认2026年先进封装仍“非常紧”，IBIDEN投入FY2026-FY2028约JPY 5000亿扩AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 8, "neutral": 2, "value": 46.2, "events": [ { "title": "I have successfully completed the stress test analysis on the transmission channels of the …", "summary": "I have successfully completed the stress test analysis on the transmission channels of the global dollar liquidity squeeze and the JPY carry trade un…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "评估中美利差环境下人民币汇率波动的持续性及套利交易压力", "summary": "卡片02判断准确：截至2026-06-22，央行通过50BP利率走廊、FIMA人民币回购与离岸外汇试点稳住了CNH，USD/CNY守在6.76-6.77，与央行实际中间价路径6.75-6.77一致。但稳定并非脆弱性根治。央行实质上成为市场的做空波动率对手方：CNH-HIBOR挤压、FIMA后备与离…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球美元紧缩与人民币贬值压力下，企业债务及套保购汇需求对离岸人民币流动性池构成被动吸纳挤压，导致 HIBOR 飙升与掉期基差走阔共振，套保成本由折价彻底翻正为溢价（重度下达 +3.0…", "summary": "截至2026-06-22，本报告验证了日元平仓引发美元紧缩向 CNH 掉期市场的传导逻辑。量化表明，中/重度压力下，420亿/720亿美元的企业购汇套保导致银行在离岸借入 3064亿/5296亿元人民币，占香港 CNH 存款池的 28.4% / 49.2%，引发 CNH HIBOR 飙升（3M 达…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元套利平仓冲击对中国的关键风险不是外储即刻失控，而是CNH基差与企业套保需求先行恶化，并在资金流从净流入转为净流出时触发定向资本账户监管升级。", "summary": "截至2026-06-22，本卡压力测试认为，日元套利交易平仓造成的美元流动性紧缩会先通过离岸美元融资、企业套保和 CNH 掉期基差传导，而不是立即触发中国外储失控。以一季度1841亿美元资本和金融账户逆差为压力基准，中等情景新增外流或对冲需求约608亿美元，深度情景约1233亿美元，分别相当于5月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 6, "neutral": 1, "value": 56.6, "events": [ { "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "summary": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026 年 6 月宏观格局（深度为负且走阔的中美利差、偏鹰美联储、管理式升值的 CNH）是否改变 2026-11-09 贸易尾部与出口周期/造船论题的基准情形", "summary": "综合判断：6 月宏观格局既不拯救多头、也不拆解尾部。结构性坚挺的 CNH 反而温和上调 301 重启概率（约 40–50%）——因 301 是结构性产业主导地位案件、与汇率正交，强人民币消除汇率指控却保留巨大账面顺差、又抽掉中国贬值缓冲；鹰派美联储 + 美国增长韧性降低美方达成协议紧迫性。烈度分叉…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "3M CNHJPY 22.80 put 是针对日元主导套利平仓传导至离岸中国资产负债表收缩的相对价值对冲，而不是通用的人民币升值或中国权益对冲。", "summary": "截至2026-06-22，本卡支持初始主线但加限定：风险不是人民币升值本身，而是日元空头回补与CNH波动率空头回补共同压缩全球风险账本。港股和中概ADR可能先受益于汇兑和挤空，但若prime broker VaR、外汇保证金和权益互换折扣收紧，离岸中国资产会通过负债收缩承压。3M CNHJPY 2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力瓶颈本地地点化和能耗惩罚电价将加速粗放GPU堆叠的价值减损，重塑A股科技风格超配“应用红利”与“电网CapEx”，并转向以“能效比”为核心的ASIC高能效硬件。", "summary": "截至2026-06-22，本卡综合研判AI电力瓶颈地点特定约束对权益风格的重塑。随着“算电协同”写入政府工作报告及发改委688号直连政策落地，东部智算面临并网与PUE差别电价硬约束（北京PUE>1.35加价0.20元）。自测算表明，不达标能耗与并网延迟将对100MW智算中心分别造成17.05亿和1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 4, "neutral": 0, "value": 67, "events": [ { "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "summary": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下修。", "summary": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI芯片能效和封装升级可以提高每个已送电MW的算力收入，但不能消除变压器长周期交付造成的站点送电瓶颈。", "summary": "截至2026-06-23，本卡支持前序判断：半导体厂商正在通过更高性能每瓦、低精度推理、HBM/CoWoS/3.5D封装和自研ASIC，提高每个已送电MW的AI产出。但这只是缓解，不是替代变压器。若站点无法送电，GPU租赁收入仍会延后；若站点已送电，Blackwell/GB300、MI350、Ir…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体基本面验证：AI 产业链订单与资本开支是否出现结构性萎缩？", "summary": "压力测试卡片 01「战术洗盘」之结论：核查全球设备、材料、Foundry、存储供应商资本开支是否随 06-23 暴跌下调。结论——供应链数据不支持结构性萎缩，反而证伪：TSMC 2026 资本开支 520–560 亿美元（+27–37%）、ASML FY26 上调至 360–400 亿欧元且 Q1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 2, "neutral": 2, "value": 86.4, "events": [ { "title": "[auto-handoff] 日元干预警报与亚太半导体/CPO板块去杠杆风险评估", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: c956eaa1-0683-41fa-9c41-551222934843 结果摘要: Wrote handoff_response.md (10735 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科技板块风格再校准：能效赋能方的稀缺溢价与高耗电算力股的吞吐折价", "summary": "本卡把前五卡收束为可执行配置框架。算力以‘每兆瓦有效算力’计价后，权益市场沿‘谁拥有稀缺投入、谁的回报被它锁死’重估。折价已现于现金流：2026超大规模厂商资本开支约6900–7250亿美元、约75%投向AI，却吞噬近100%经营性现金流；微软约800亿美元Azure积压受限于电力。溢价已现于盘面…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中国CSP成本上升背景下，AI算力基础设施供应弹性与扩容压力分析", "summary": "压力测试部分否定线程以电力为中心的框架。AI算力系统是供给绑定而非需求绑定：2026年超大规模厂商资本开支同比+77%至约7250亿美元[S1]，英伟达Blackwell售罄至年中、积压360万颗[S2]，因此约束的是配额优先级而非需求。核心锚定集群凭借锁定电力、GPU配额与表后自发电被隔离；放缓…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 美元兑日元突破161.7极值区下日本实弹干预对日元 Carry Trade 拆仓流动性测算", "summary": "致 市场情绪师： 协同任务已完成。执行方为 外汇策略师。 模式: adhoc 会话: 8f713fd9-1e6d-4cd3-92f7-609e585c1608 结果摘要: Wrote handoff_response.md (9014 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-25", "supports": 1, "risks": 3, "neutral": 3, "value": 92.8, "events": [ { "title": "[auto-handoff] 美联储主席Warsh政策首秀推升美元至100.814情景下日元与离岸人民币套利交易无序平仓溢出测算", "summary": "致 首席风控官： 协同任务已完成。执行方为 外汇策略师。 模式: adhoc 会话: 82711dfa-2690-4d72-9be3-a760c489f1cc 结果摘要: Wrote handoff_response.md (12305 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 美联储主席Warsh政策首秀推升美元至100.814情景下日元与离岸人民币套利交易无序平仓溢出测算", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 外汇策略师。 模式: adhoc 会话: 6d850732-da73-4b94-a1fd-3febb5d74d34 结果摘要: Wrote handoff_response.md (11413 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 5, "neutral": 0, "value": 65.2, "events": [ { "title": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径", "summary": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径 Global Carry Unwind and Liquidity Siphon Risks for HK/A-Share Semi Leaders 日本央行升息及美元兑日元突破161.7，全球Carry Trade平仓是否会…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储'更高更久'与坚挺美元对人民币、日元及新兴市场货币的传导", "summary": "压力测试卡片 01 的强美元传染推论。结论：传导高度分化而非齐跌。日元是泄压阀且已破位——USD/JPY ~161.7、近 1986 低位，美日利差 ~275bp，日央 2026-06-17 加息至 1.00%、年初至 5 月约 ¥11.7 万亿干预全数回吐 [S2]。人民币偏强被防守——中间价 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "USD/CNY硬防御阈值与权益托底弹药消耗速度", "summary": "压力测试暴露出 Research note 01–04 的符号错误：2026-06-26 人民币是强势（中间价 6.8209[S1]），PBOC 在抑制升值（2-27 将外汇风险准备金率 20%→0%[S2]），外储升至 3.442 万亿美元[S3]。迫使 PBOC 由预期管理转向资产负债表干预（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "同步鹰派转向是真实的风险窗口，但不是已被证实的新紧缩制度；Research note 9 字面论点应被否定，前序宏观主线应被有条件保留。", "summary": "截至2026-06-26，本卡否定 Research note 9 字面交接文本：它是目录检查类工作流伪影，不是宏观论点。但压力测试后有条件支持前序经济主线：G3 6月鹰派转向真实发生，Fed 维持3.50%-3.75%、ECB升至2.25%、BoJ升至1.0%；美国5月CPI/PCE仍高，说明不…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 5, "neutral": 0, "value": 66.6, "events": [ { "title": "日元套利交易平仓与离岸CNH流动性收紧对A股防御重估的交叉传导风险", "summary": "本卡以global-macro压力测试card-08的“USD/JPY与CNH流动性熔断器”构想：支持其核心方向（知识库已确认应以二者作量化触发器），但否定朴素的“单变量、按价位、OR门”实现。四点修正：(1)冲击分两波——先美元流动性挤兑、后避险重估，价位触发会迫使在第一波真空里抛售防御股；(2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美元多头利差交易极度拥挤且极易受日央行干预与实际利率波动影响而无序平仓，USD/JPY跌破150会引发2150亿至2850亿美元美债清算并给在岸高两融科技板块带来资本流出压力，配置U…", "summary": "截至2026-06-27，美元指数已站稳100点并突破100.735，录得近13个月新高101.20。然而，CFTC日元净空头堆积至-145.8K手（相当于113.8亿美元名义价值）的极端高位，表明美元多头交易极度拥挤。日本央行超预期将利率调升至1.00%，导致美日短期利差压缩至318 bp，US…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-27，AI电力并网限制（2年延误拖累53.71亿美元[S12]）将倒逼算力采购削减并重创高端半导体高增长假设，投资者应果断削减AI硬件至12%及长债至5%，转而增配…", "summary": "截至2026-06-27，本报告重新校准了在高利率平台与电力之墙挤压下的多资产配置。2年电网并网延误对10万卡Blackwell智算集群将产生53.71亿美元搁浅资金损失[S12]，迫使云巨头在2027年削减20%GPU订单。这将导致台积电3nm和2nm产能利用率分别跌至85%与80%，CoWoS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 4, "neutral": 0, "value": 74, "events": [ { "title": "中美利差倒挂对离岸人民币（CNH）汇率及跨境资本流动的传导效应", "summary": "支持卡片01的格局判断并补齐传导环节。截至2026-06-29，美国10年期国债约4.40%、中国10年期约1.79%，名义利差倒挂约−261bps（本轮最阔）；而真实利差约+39bps偏向中国。名义渠道在去杠杆中占主导，使CNH成为融资货币。鹰派美联储驱动的套利去杠杆经两条腿（CNH融资腿+全球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基准配置应优先通过久期表达去通胀与政策失误风险，权益只做选择性低贝塔，商品低配整体贝塔但保留油价重上100美元的显式尾部对冲。", "summary": "2026-06-29，本卡综合并支持上一卡“央行在消退油价冲击中过度收紧”的判断，但把它转化为更可执行的大类资产配置：久期从35%中性升至45%，权益贝塔从45%降至35%，广谱商品从15%降至12%，现金升至8%。依据是美国5月CPI 4.2%中能源同比23.5%、核心仅2.9%，而美联储仍维持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 2, "neutral": 0, "value": 87.4, "events": [ { "title": "日元套利平仓通过离岸人民币流动性抽水及掉期清算挤压双重渠道，系统性抬高离岸融资成本并技术性击穿USD/CNY 7.00关口，导致人民币套利空间彻底崩溃。", "summary": "本卡支持Research note 02，定量分析日元套利平仓对CNH的传导。BOJ加息和USD/JPY破位触发约349亿美元平仓，占离岸池近25%，推高隔夜HIBOR至10.81%-12.74%。掉期展期率跌至40%引发即期购汇压力，击穿USD/CNY 7.00。CNH套利Carry-to-Ri…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-56…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-e9630d3ee1/", "networkHref": "/investor/research-network/#thesis-e9630d3ee1" }, { "id": "thesis-89c14497ac", "source": "tracked_thesis", "title": "期权伽马、ETF 与尾盘流动性正在放大主题交易的微观结构风险", "titleEn": "Options gamma, ETFs, and closing liquidity are amplifying microstructure risk in theme trades", "lane": "期权伽马/市场微观结构", "laneEn": "Options gamma and market microstructure", "laneHref": "/investor/lanes/options-gamma-and-market-microstructure/", "investableFocus": "期权伽马/市场微观结构：CBOE, CME, VIRT, IBKR, HOOD, VIXY", "stockUniverse": [ { "ticker": "CBOE", "name": "Cboe Global Markets", "market": "US", "role": "期权与波动率交易所", "roleEn": "期权与波动率交易所", "bucket": "core", "href": "/investor/stocks/CBOE/" }, { "ticker": "CME", "name": "CME Group", "market": "US", "role": "期货与清算", "roleEn": "期货与清算", "bucket": "core", "href": "/investor/stocks/CME/" }, { "ticker": "VIRT", "name": "Virtu Financial", "market": "US", "role": "做市与流动性", "roleEn": "做市与流动性", "bucket": "core", "href": "/investor/stocks/VIRT/" }, { "ticker": "IBKR", "name": "Interactive Brokers", "market": "US", "role": "交易基础设施", "roleEn": "交易基础设施", "bucket": "core", "href": "/investor/stocks/IBKR/" }, { "ticker": "HOOD", "name": "Robinhood", "market": "US", "role": "散户期权活跃度", "roleEn": "散户期权活跃度", "bucket": "core", "href": "/investor/stocks/HOOD/" }, { "ticker": "VIXY", "name": "ProShares VIX Short-Term Futures ETF", "market": "US ETF", "role": "波动率工具", "roleEn": "波动率工具", "bucket": "hedge", "href": "/investor/stocks/VIXY/" }, { "ticker": "QQQ", "name": "Invesco QQQ Trust", "market": "US ETF", "role": "纳斯达克成长暴露", "roleEn": "纳斯达克成长暴露", "bucket": "watch", "href": "/investor/stocks/QQQ/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "IBIT", "name": "iShares Bitcoin Trust", "market": "US ETF", "role": "BTC 现货代理", "roleEn": "BTC 现货代理", "bucket": "core", "href": "/investor/stocks/IBIT/" }, { "ticker": "COIN", "name": "Coinbase", "market": "US", "role": "加密交易基础设施", "roleEn": "加密交易基础设施", "bucket": "core", "href": "/investor/stocks/COIN/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" } ], "stockTickers": [ "CBOE", "CME", "VIRT", "IBKR", "HOOD", "VIXY", "QQQ", "SMH", "IBIT", "COIN", "TSLA", "ASTS", "MU", "005930.KS" ], "coreView": "赛道：期权伽马/市场微观结构。候选观察标的：CBOE, CME, VIRT, IBKR, HOOD, VIXY。当前共有 40 条证据、6 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 79, "eventCount": 160, "firstSeen": "2026-04-29", "lastSeen": "2026-06-29", "supports": 0, "risks": 124, "neutral": 36, "analystCount": 17, "sourceTypeCounts": { "分析协作": 17, "研究笔记": 39, "研究线程": 104 }, "topTerms": [ "Options", "gamma", "ETFs", "closing", "liquidity", "amplifying", "microstructure", "risk", "theme", "trades", "CBOE", "VIRT" ], "whatChanged": [ "2026-06-29: 产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "2026-06-29: 产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "2026-06-29: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "2026-06-29: 产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "2026-06-29: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）" ], "practical": { "sectorLane": "期权伽马/市场微观结构", "sectorLaneEn": "Options gamma and market microstructure", "exposureBucket": "风险偏好与高波动", "tone": "red", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "CBOE", "name": "Cboe Global Markets", "market": "US", "role": "期权与波动率交易所", "roleEn": "期权与波动率交易所", "bucket": "core", "href": "/investor/stocks/CBOE/" }, { "ticker": "CME", "name": "CME Group", "market": "US", "role": "期货与清算", "roleEn": "期货与清算", "bucket": "core", "href": "/investor/stocks/CME/" }, { "ticker": "VIRT", "name": "Virtu Financial", "market": "US", "role": "做市与流动性", "roleEn": "做市与流动性", "bucket": "core", "href": "/investor/stocks/VIRT/" }, { "ticker": "IBKR", "name": "Interactive Brokers", "market": "US", "role": "交易基础设施", "roleEn": "交易基础设施", "bucket": "core", "href": "/investor/stocks/IBKR/" }, { "ticker": "HOOD", "name": "Robinhood", "market": "US", "role": "散户期权活跃度", "roleEn": "散户期权活跃度", "bucket": "core", "href": "/investor/stocks/HOOD/" }, { "ticker": "VIXY", "name": "ProShares VIX Short-Term Futures ETF", "market": "US ETF", "role": "波动率工具", "roleEn": "波动率工具", "bucket": "hedge", "href": "/investor/stocks/VIXY/" }, { "ticker": "QQQ", "name": "Invesco QQQ Trust", "market": "US ETF", "role": "纳斯达克成长暴露", "roleEn": "纳斯达克成长暴露", "bucket": "watch", "href": "/investor/stocks/QQQ/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "IBIT", "name": "iShares Bitcoin Trust", "market": "US ETF", "role": "BTC 现货代理", "roleEn": "BTC 现货代理", "bucket": "core", "href": "/investor/stocks/IBIT/" }, { "ticker": "COIN", "name": "Coinbase", "market": "US", "role": "加密交易基础设施", "roleEn": "加密交易基础设施", "bucket": "core", "href": "/investor/stocks/COIN/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" } ], "stockTickers": [ "CBOE", "CME", "VIRT", "IBKR", "HOOD", "VIXY", "QQQ", "SMH", "IBIT", "COIN", "TSLA", "ASTS", "MU", "005930.KS" ], "primaryStocks": [ { "ticker": "CBOE", "name": "Cboe Global Markets", "market": "US", "role": "期权与波动率交易所", "roleEn": "期权与波动率交易所", "bucket": "core", "href": "/investor/stocks/CBOE/" }, { "ticker": "CME", "name": "CME Group", "market": "US", "role": "期货与清算", "roleEn": "期货与清算", "bucket": "core", "href": "/investor/stocks/CME/" }, { "ticker": "VIRT", "name": "Virtu Financial", "market": "US", "role": "做市与流动性", "roleEn": "做市与流动性", "bucket": "core", "href": "/investor/stocks/VIRT/" }, { "ticker": "IBKR", "name": "Interactive Brokers", "market": "US", "role": "交易基础设施", "roleEn": "交易基础设施", "bucket": "core", "href": "/investor/stocks/IBKR/" }, { "ticker": "HOOD", "name": "Robinhood", "market": "US", "role": "散户期权活跃度", "roleEn": "散户期权活跃度", "bucket": "core", "href": "/investor/stocks/HOOD/" }, { "ticker": "VIXY", "name": "ProShares VIX Short-Term Futures ETF", "market": "US ETF", "role": "波动率工具", "roleEn": "波动率工具", "bucket": "hedge", "href": "/investor/stocks/VIXY/" } ], "instrumentTags": [ "CBOE", "Cboe Global Markets", "CME", "CME Group", "VIRT", "Virtu Financial", "IBKR", "Interactive Brokers", "HOOD", "Robinhood", "VIXY", "ProShares VIX Short-Term Future…" ], "reviewCadence": "每日复核", "confirmationSignals": [ "期权伽马和 ETF 资金是否放大尾盘流动性", "社交热度是否转化为真实增量买盘", "波动率是否领先基本面恶化" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "波动率", "现金", "指数保护", "期权拥挤" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 160, "sourceDiversity": 3, "analystCount": 17, "recentEvents": 128, "supportRatio": 0, "riskRatio": 0.78, "conflictScore": 89.5, "evidenceDepth": 100 }, "score": 72.4 }, "events": [ { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 15.25, "matchedTerms": [ "risk", "evidence", "dominant", "inflation", "transmission", "core" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] Impact of the JGB 2.8% Carry Trade Repatriation Trigger on Multi-Asset Barbell S…", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 4cae9f36-7135-4180-845b-119f7b02c4b9 结果摘要: Wrote handoff_response.md (13007 bytes); wrote handoff_response.en.md (12839 bytes) — Under t…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 11.75, "matchedTerms": [ "liquidity", "impact", "crowding", "power", "Global" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] Impact of global capacity pricing spikes on domestic transformer export margins", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 324722b5-2f7b-4b91-aec7-ac5429f32a39 结果摘要: Wrote handoff_response.md (11765 bytes); wrote handoff_response.en.md (12438 bytes) — pri…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 11.25, "matchedTerms": [ "transmission", "impact", "power", "Global" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "risk", "Candidate", "thesis", "analysts" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "risk", "Candidate", "thesis", "analysts" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "risk", "Candidate", "thesis", "analysts" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "risk", "Candidate", "thesis", "analysts" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "risk", "Candidate", "thesis", "analysts" ], "href": "" }, { "date": "2026-06-28", "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.en.md (7867 bytes) — The repor…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 11, "matchedTerms": [ "inflation", "investment", "cost", "crowding" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "risk", "Candidate", "thesis", "analysts" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "risk", "Candidate", "thesis", "analysts" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "risk", "Candidate", "thesis", "analysts" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "risk", "Candidate", "thesis", "analysts" ], "href": "" }, { "date": "2026-06-27", "title": "I have completed the quantitative stress test on the STAR 50 ETF options collar strategy (\"Buy …", "summary": "I have completed the quantitative stress test on the STAR 50 ETF options collar strategy (\"Buy 95% Put / Sell 85% Put / Sell 105% Call\" [S1]) designed in Research note 09. Summary…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 11.5, "matchedTerms": [ "Options", "macro", "options", "Global" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "risk", "Candidate", "thesis", "analysts" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "risk", "Candidate", "thesis", "analysts" ], "href": "" } ], "timeline": [ { "date": "2026-04-29", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "[auto-handoff] Gamma 墙探测", "summary": "致 市场情绪师： 协同任务已完成。执行方为 衍生品策略师。 模式: adhoc 会话: bce1dfe0-caea-43c2-8d03-d770da2ff7d3 结果摘要: # Role Derivatives Strategist (衍生品策略师). # Expertise Options st…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "I have completed the stress test of the proposed 2026 2027 A share allocation. The analysis…", "summary": "I have completed the stress test of the proposed 2026 2027 A share allocation. The analysis identifies a \"Safety Paradox\" where the Liquidity Qualifi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 0, "neutral": 1, "value": 47.8, "events": [ { "title": "\"session id\": \"a5c839ce 1820 42a0 b8c5 cb453873bd99\", \"response\": \"The Investment Committee…", "summary": "\"session id\": \"a5c839ce 1820 42a0 b8c5 cb453873bd99\", \"response\": \"The Investment Committee has officially synthesized the research from the Global M…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "[auto-handoff] 主权算力财政支出影响", "summary": "致 TMT行业分析师： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 1f2ac627-aa0e-4991-81c1-4c46ae106c06 结果摘要: Wrote handoff_response.md (17,200 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Carry Trade Volatility Cost Audit", "summary": "致 市场情绪师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 029c2336-7d2e-4a9d-8e0f-b4899c61ebbd 结果摘要: Wrote handoff_response.md (10,725 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "I have analyzed the market structure for the A share AI infrastructure sectors as of the ca…", "summary": "I have analyzed the market structure for the A share AI infrastructure sectors as of the canonical work date of May 20, 2026. While the \"Fortress Axi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 1, "value": 46, "events": [ { "title": "As a Derivatives Strategist, I have completed the stress test analysis of the options prici…", "summary": "As a Derivatives Strategist, I have completed the stress test analysis of the options pricing and tail hedging costs for A share commercial space sec…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 算力主线拥挤度与持仓建议", "summary": "致 A股策略师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 3af6244f-2140-4b42-a70d-f8e1f44e86ce 结果摘要: Wrote handoff_response.md (3594 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 0, "neutral": 4, "value": 62.4, "events": [ { "title": "Re-estimate supply chain price shock transmission to U.S. core PCE", "summary": "Re-estimate supply chain price shock transmission to U.S. core PCE In 'Thesis 1: Step 4', direct a high-priority handoff to global-macro to re-estima…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "As the Global Macro Analyst , I have conducted a comprehensive synthesis of the macro finan…", "summary": "As the Global Macro Analyst , I have conducted a comprehensive synthesis of the macro financial transmission channels of G SIB USD balance sheet rati…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026 05 22 Energy Analyst Research Synthesis As the Energy Analyst , I have synthesized the…", "summary": "2026 05 22 Energy Analyst Research Synthesis As the Energy Analyst , I have synthesized the ongoing research regarding the downstream catering price …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "We have successfully completed Research note 03/08 of the research thread session efecefe6 …", "summary": "We have successfully completed Research note 03/08 of the research thread session efecefe6 2a06 4881 87e9 ccec1f502077 as the Chief Economist. Stance…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 0, "neutral": 1, "value": 56.2, "events": [ { "title": "I have completed the global macro analysis for Research note 01 of the research thread sess…", "summary": "I have completed the global macro analysis for Research note 01 of the research thread session b0a5df37 4aba 4443 9696 a334f45b9dd4 . Key Findings & …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "I have completed the analysis for Research note 4/10 as the Materials Analyst, evaluating t…", "summary": "I have completed the analysis for Research note 4/10 as the Materials Analyst, evaluating the cost transmission channels and gross margin squeeze ris…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 1, "value": 53, "events": [ { "title": "I have completed the macroeconomic research and policy evaluation for Research note 06 of t…", "summary": "I have completed the macroeconomic research and policy evaluation for Research note 06 of the shared research thread session. Summary of Research & F…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] STAR 50 Snowball Knock-in Margin Buffer Simulation and 11:27 Liquidity Injec…", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: cb3eaab3-a011-44dd-a37b-197190a085e5 结果摘要: Wrote handoff_response.md (13905 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 0, "neutral": 1, "value": 60.4, "events": [ { "title": "I have completed the analysis for Research note 01/08 as the Global Macro Analyst , examini…", "summary": "I have completed the analysis for Research note 01/08 as the Global Macro Analyst , examining the unseasonal surge in global freight rates and its no…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 0, "neutral": 2, "value": 64.8, "events": [ { "title": "Investigate global consumer bifurcation microstructure.", "summary": "Investigate global consumer bifurcation microstructure. Open a research thread card to determine if 'frequency up, ticket size down' is a global them…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "A Share Asset Allocation Strategy Under Low Inflation & Property Wealth Traps (Research not…", "summary": "A Share Asset Allocation Strategy Under Low Inflation & Property Wealth Traps (Research note 9/10) As the Chief Strategist (首席策略师) , I have synthesiz…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 0, "neutral": 1, "value": 63.2, "events": [ { "title": "We have completed the research for Research note 06, analyzing the transmission channel thr…", "summary": "We have completed the research for Research note 06, analyzing the transmission channel through which extreme 0DTE (Zero Days to Expiration) option i…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 0, "neutral": 1, "value": 64.6, "events": [ { "title": "I have finalized the H2 2026 stress testing and synthesis analysis on the transmission of U…", "summary": "I have finalized the H2 2026 stress testing and synthesis analysis on the transmission of US AI hardware tariff inflation to the A share export compu…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "As the China Macro Analyst (china macro), I have completed the stress test report for Resea…", "summary": "As the China Macro Analyst (china macro), I have completed the stress test report for Research note 8 of 8, assessing the boundaries of China's fisca…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 0, "neutral": 1, "value": 67.4, "events": [ { "title": "[auto-handoff] 中远海控港股期权Implied Volatility与A股两融平仓盘的负Gamma敞口测算", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 衍生品策略师。 模式: adhoc 会话: 5580f12d-bbb9-422d-bf32-407dda6755da 结果摘要: I will begin by listing the contents of the application dire…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 1, "value": 62.8, "events": [ { "title": "[auto-handoff] 50ETF期权低隐波与偏斜对风格切换（红利低波与成长动量）的指示作用", "summary": "致 衍生品策略师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 2519521e-3166-4de3-b07a-35f850c76287 结果摘要: Wrote handoff_response.md (11501 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Semiconductor De-leveraging vs Physical DC Bottleneck", "summary": "致 衍生品策略师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 6909760a-2b3a-4f87-9c0c-ee2d76724c67 结果摘要: Wrote handoff_response.md (11874 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 3, "value": 70.2, "events": [ { "title": "As the China Macro Analyst, I have completed the analysis on the transmission effects of th…", "summary": "As the China Macro Analyst, I have completed the analysis on the transmission effects of the JPY carry trade unwind on China's capital flows and the …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have completed the stress test analysis of China's State Grid 2026 investment plan and it…", "summary": "I have completed the stress test analysis of China's State Grid 2026 investment plan and its impact on copper demand and the systemic drawdown floor …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "We have completed the quantitative stress test on the JPY carry trade unwind and its transm…", "summary": "We have completed the quantitative stress test on the JPY carry trade unwind and its transmission to the US Treasury yield curve and high yield (HY) …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have synthesized the ongoing research on the macroeconomic conflicts (the Fed's rate hold…", "summary": "I have synthesized the ongoing research on the macroeconomic conflicts (the Fed's rate hold vs. the domestic liquidity easement) and their impact on …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 0, "neutral": 1, "value": 71.6, "events": [ { "title": "Executive Summary of Analysis (Energy Analyst Perspective) In this stress test card, we ana…", "summary": "Executive Summary of Analysis (Energy Analyst Perspective) In this stress test card, we analyzed the macroeconomic transmission mechanism and Federal…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 0, "neutral": 1, "value": 73, "events": [ { "title": "The upstream files session brief.md , analyst catalog.json , and the Research note 01 repor…", "summary": "The upstream files session brief.md , analyst catalog.json , and the Research note 01 reports ( report.en.md , report.zh.md , meta.json ) were missin…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [] }, { "date": "2026-06-10", "supports": 0, "risks": 7, "neutral": 0, "value": 30.8, "events": [ { "title": "Research Summary Under my role as the Chief Risk Officer ( chief risk ), I have conducted a…", "summary": "Research Summary Under my role as the Chief Risk Officer ( chief risk ), I have conducted a comprehensive risk stress test on the biotechnology sub b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 3.3%, 3/91）", "summary": "产业链调研：新一代信息技术（30 日激活率 3.3%, 3/91） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.1%, 14/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.1%, 14/198） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 4, "neutral": 0, "value": 50.2, "events": [ { "title": "产业链调研：海工装备产业链（30 日激活率 5.4%, 6/112）", "summary": "产业链调研：海工装备产业链（30 日激活率 5.4%, 6/112） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 10.3%, 10/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 10.3%, 10/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 4.7%, 8/170）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 4.7%, 8/170） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275）", "summary": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 5, "neutral": 0, "value": 45.6, "events": [ { "title": "I have completed the stress test analysis for Research note 2/10 in the research thread res…", "summary": "I have completed the stress test analysis for Research note 2/10 in the research thread research thread. In this analysis, I evaluated the sustainabi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 5, "neutral": 0, "value": 47, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 5.7%, 10/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 5.7%, 10/174） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 7, "neutral": 1, "value": 39.4, "events": [ { "title": "The research into China's power system and green trading reforms at the start of the \"15th …", "summary": "The research into China's power system and green trading reforms at the start of the \"15th Five Year Plan\" (2026) is complete. The analysis confirms …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 8, "neutral": 1, "value": 34.8, "events": [ { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 3, "neutral": 0, "value": 63.2, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 9, "neutral": 0, "value": 28.6, "events": [ { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 9, "neutral": 0, "value": 30, "events": [ { "title": "I have completed the stress test report for Research note 8/8 of the session. The report sy…", "summary": "I have completed the stress test report for Research note 8/8 of the session. The report systematically maps out the transmission of an offshore toke…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 5, "neutral": 3, "value": 64.4, "events": [ { "title": "I have performed a rigorous quantitative stress test using Black Scholes (for the XBI equit…", "summary": "I have performed a rigorous quantitative stress test using Black Scholes (for the XBI equity leg) and Black 76 (for the 5 Year Treasury futures leg) …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "As the Global Macro Analyst, I have completed the stress test evaluation for Research note …", "summary": "As the Global Macro Analyst, I have completed the stress test evaluation for Research note 07/08 of the BTM distributed power short leg restructuring…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I will begin by inspecting the workspace directory and reading the completed cards and file…", "summary": "I will begin by inspecting the workspace directory and reading the completed cards and files to understand the research context and build a solid, ev…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：建筑材料产业链（30 日激活率 8.3%, 13/157）", "summary": "产业链调研：建筑材料产业链（30 日激活率 8.3%, 13/157） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 10, "neutral": 2, "value": 32.8, "events": [ { "title": "I have completed the synthesis of the trade surplus breakdown trigger and its implications …", "summary": "I have completed the synthesis of the trade surplus breakdown trigger and its implications for the global macro liquidity transmission chain. Summary…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have successfully completed the stress test analysis on the transmission channels of the …", "summary": "I have successfully completed the stress test analysis on the transmission channels of the global dollar liquidity squeeze and the JPY carry trade un…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have completed the analysis on AI compute infrastructure transmission paths under power a…", "summary": "I have completed the analysis on AI compute infrastructure transmission paths under power and transformer constraints , focusing on the concrete impa…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 7, "neutral": 0, "value": 46.2, "events": [ { "title": "Summary of Work We have evaluated the mismatch risk between rising GPU power consumption an…", "summary": "Summary of Work We have evaluated the mismatch risk between rising GPU power consumption and grid capacity bottlenecks, demonstrating how this constr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 5, "neutral": 0, "value": 59.6, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 12.6%, 26/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 12.6%, 26/206） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 6, "neutral": 4, "value": 67, "events": [ { "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply s…", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A-share semiconductor valuation reset sensitivity under global carry trade u…", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 38bcbac6-c7bd-47b7-b7e8-1ba28c2790af 结果摘要: Wrote handoff_response.md (9870 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I will read the report from the previous card (Research note 07, in Chinese) to understand …", "summary": "I will read the report from the previous card (Research note 07, in Chinese) to understand the current research foundation regarding the USD/CNY forw…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 十五五电网4万亿投资计划对国内工业产出及PPI的拉动效应", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: cee8e660-8964-4860-90dd-b580f318def7 结果摘要: Wrote handoff_response.md (9465 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 1, "neutral": 0, "value": 86.4, "events": [ { "title": "[auto-handoff] 杠杆交易拥挤度与两融突破3万亿风险评估", "summary": "致 首席策略师： 协同任务已完成。执行方为 市场情绪师。 模式: adhoc 会话: 55cdff25-1527-4b69-9408-a3e16d4c97ce 结果摘要: Wrote handoff_response.md (14618 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 6, "neutral": 0, "value": 57.8, "events": [ { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 1, "value": 80.2, "events": [ { "title": "I have completed the quantitative stress test on the STAR 50 ETF options collar strategy (\"…", "summary": "I have completed the quantitative stress test on the STAR 50 ETF options collar strategy (\"Buy 95% Put / Sell 85% Put / Sell 105% Call\" [S1]) designe…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 4, "neutral": 1, "value": 75.6, "events": [ { "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 8, "neutral": 2, "value": 56, "events": [ { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Impact of the JGB 2.8% Carry Trade Repatriation Trigger on Multi-Asset Barbe…", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 4cae9f36-7135-4180-845b-119f7b02c4b9 结果摘要: Wrote handoff_response.md (13007 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Impact of global capacity pricing spikes on domestic transformer export marg…", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 324722b5-2f7b-4b91-aec7-ac5429f32a39 结果摘要: Wrote handoff_response.md (11765 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-89c14497ac/", "networkHref": "/investor/research-network/#thesis-89c14497ac" }, { "id": "thesis-075017003b", "source": "tracked_thesis", "title": "AI 资本开支通过能源、设备和利率进入通胀定价", "titleEn": "AI capex enters inflation pricing through energy, equipment, and rates", "lane": "AI 再通胀/利率敏感资产", "laneEn": "AI reflation and rate-sensitive assets", "laneHref": "/investor/lanes/ai-reflation-and-rate-sensitive-assets/", "investableFocus": "AI 再通胀/利率敏感资产：XLE, XLU, TLT, GLD, JPM, NEE", "stockUniverse": [ { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业利率敏感", "roleEn": "公用事业利率敏感", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "JPM", "name": "JPMorgan Chase", "market": "US", "role": "利率与信用周期", "roleEn": "利率与信用周期", "bucket": "core", "href": "/investor/stocks/JPM/" }, { "ticker": "NEE", "name": "NextEra Energy", "market": "US", "role": "电力资产与利率敏感", "roleEn": "电力资产与利率敏感", "bucket": "core", "href": "/investor/stocks/NEE/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" } ], "stockTickers": [ "XLE", "XLU", "TLT", "GLD", "JPM", "NEE", "IEF", "UUP", "FXY", "FXI", "EEM", "APO", "CAT", "COP" ], "coreView": "赛道：AI 再通胀/利率敏感资产。候选观察标的：XLE, XLU, TLT, GLD, JPM, NEE。当前共有 40 条证据、3 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 79, "eventCount": 160, "firstSeen": "2026-05-01", "lastSeen": "2026-06-29", "supports": 4, "risks": 144, "neutral": 12, "analystCount": 15, "sourceTypeCounts": { "分析协作": 2, "研究笔记": 150, "研究主题": 5, "研究线程": 3 }, "topTerms": [ "电网", "GOES", "CPI", "PCE", "数据中心", "AI capex", "AI数据中心", "公用事业", "AIDC", "电力", "红利", "能源" ], "whatChanged": [ "2026-06-29: 变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "2026-06-29: 材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放右移。", "2026-06-29: 电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "2026-06-28: 关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "2026-06-28: AI数据中心电力瓶颈与核能PPA战略转型" ], "practical": { "sectorLane": "AI 再通胀/利率敏感资产", "sectorLaneEn": "AI reflation and rate-sensitive assets", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业利率敏感", "roleEn": "公用事业利率敏感", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "JPM", "name": "JPMorgan Chase", "market": "US", "role": "利率与信用周期", "roleEn": "利率与信用周期", "bucket": "core", "href": "/investor/stocks/JPM/" }, { "ticker": "NEE", "name": "NextEra Energy", "market": "US", "role": "电力资产与利率敏感", "roleEn": "电力资产与利率敏感", "bucket": "core", "href": "/investor/stocks/NEE/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" } ], "stockTickers": [ "XLE", "XLU", "TLT", "GLD", "JPM", "NEE", "IEF", "UUP", "FXY", "FXI", "EEM", "APO", "CAT", "COP" ], "primaryStocks": [ { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业利率敏感", "roleEn": "公用事业利率敏感", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "JPM", "name": "JPMorgan Chase", "market": "US", "role": "利率与信用周期", "roleEn": "利率与信用周期", "bucket": "core", "href": "/investor/stocks/JPM/" }, { "ticker": "NEE", "name": "NextEra Energy", "market": "US", "role": "电力资产与利率敏感", "roleEn": "电力资产与利率敏感", "bucket": "core", "href": "/investor/stocks/NEE/" } ], "instrumentTags": [ "XLE", "Energy Select Sector SPDR", "XLU", "Utilities Select Sector SPDR", "TLT", "iShares 20+ Year Treasury Bond …", "GLD", "SPDR Gold Shares", "JPM", "JPMorgan Chase", "NEE", "NextEra Energy" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 108 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 15, "recentEvents": 108, "supportRatio": 0.03, "riskRatio": 0.9, "conflictScore": 100, "evidenceDepth": 100 }, "score": 69.8 }, "events": [ { "date": "2026-06-29", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]，PPI-CPI剪刀差隔离了政策利率。任务设想的'成本→CP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 37.75, "matchedTerms": [ "电网", "GOES", "CPI", "电力", "红利", "设备", "macro", "long" ], "href": "" }, { "date": "2026-06-29", "title": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放…", "summary": "截至2026-06-29，变压器交期延长和材料瓶颈导致美国电网资本开支通胀率达32.72%。结合2027/2028年度PJM容量电费暴涨11.5倍至$333.44/MW-day，将拉动美国零售电价上涨4.18%，直接和间接推高全国CPI 0.124个百分点。高粘性通胀与AI基建庞大资金需求将上移中性利率50-75基点，延缓美联储降息路径。此外，并网滑期18…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.75, "matchedTerms": [ "电网", "CPI", "数据中心", "公用事业", "红利", "设备", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优先既有水电/核电现金流、具备瓶颈定价权的变压器和电网设备供应…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 33, "matchedTerms": [ "电网", "数据中心", "公用事业", "电力", "红利", "设备", "energy" ], "href": "" }, { "date": "2026-06-28", "title": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电力’，并驱动策略从算力规模向算法效率和电…", "summary": "截至2026-06-28，本卡片通过对电网与变压器交期的压力测试，确认电网并网（5-7年交期）和变压器交付（2-4年交期）构成了AI算力扩张的终极物理边界。随着GB300（约135kW）及Rubin Kyber（约600kW）对800VDC和固态变压器（SST）的高密度配电需求增加，目前仅有不足8%的设施就绪。FERC否决核电表后直供，迫使所有新增容量重回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 38.75, "matchedTerms": [ "电网", "GOES", "AI capex", "公用事业", "电力", "红利", "设备", "capex" ], "href": "" }, { "date": "2026-06-28", "title": "2026年全球AI数据中心电网瓶颈与“自主供电”战略的崛起", "summary": "2026年全球超大规模厂商的AI算力基建受限于电网瓶颈与关键电力设备（变压器交期达5年）短缺。这迫使厂商将部分资本支出（预计2026年总支出达7250亿美元）从纯算力（GPU）转向自主供电（BYOP）和表后（BTM）能源方案。预计该转移将延迟算力上线速度，并推高长期持有成本。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36, "matchedTerms": [ "电网", "数据中心", "AI数据中心", "电力", "能源", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "summary": "截至2026-06-28，本卡对“AI负荷增长能否支撑公用事业高分红”作压力测试：结论是精选组合层面可以，板块整体不能机械外推。AI数据中心用电增长真实，但也同步拉动电网、发电和储能资本开支。水电、核电、容量电价支持的火电及低杠杆综合电力运营商仍可维持分红；风光开发商受消纳、电价和项目资本开支制约，不能直接视为高息资产。上一卡的25%公用事业红利仓可以保留…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36, "matchedTerms": [ "电网", "数据中心", "AI数据中心", "公用事业", "电力", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "6月点阵图不是简单过时，而是可能“对得不是理由”：能源理由过时，AI/电网与折旧通道支持更高更久，但不足以单独证明立即加息。", "summary": "截至2026-06-28，本卡综合认为：6月点阵图并非被最初的油价冲击证明，因布伦特已回落至约US$72/bbl；但AI/电网导致的结构性电价稀缺、核心PCE 3.4%、名义需求仍强以及AI-capex折旧悬挂，使“更高更久”仍有宏观依据。该通道不足以单独证明立即加息，因为电力CPI权重仅2.489%、PJM冲击区域性强、US$176B折旧低估属低置信度压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 33, "matchedTerms": [ "电网", "CPI", "PCE", "电力", "能源", "capex", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "在轨算力能源经济的关键不是太阳能是否充足，而是每MW需要发射多少太阳翼、辐射器和结构质量；只有<$200/kg、<35 t/MW、近10年寿命的窄条件组合，才可能与受电网约束的地面AI数据中…", "summary": "截至2026-06-28，本卡综合判断：空间太阳能的连续供电优势真实，NASA/Google口径支持约1,361 W/m²、晨昏轨道最高8x地面年产能；但在轨算力不是免费电力，而是把电站、辐射器和结构质量发射上天。按自测算，只有在发射价低于约$200/kg、非IT平台质量低于35 t/MW、有效寿命接近10年时，发射摊销才可能降至约$80/MWh并与高价或…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32.5, "matchedTerms": [ "电网", "数据中心", "AI数据中心", "电力", "能源", "energy" ], "href": "" }, { "date": "2026-06-28", "title": "PQC 引发的周期性热降频会显著提高超大规模 AI 集群 PUE/WUE 与能源成本预测难度，并在集中接入区增加电力调度复杂度，但它是边际波动因子而非全国电力需求的主增长因子。", "summary": "2026-06-29 能源卡结论：支持前序观点，但限定机制。PQC 硬件化的直接功耗不是数据中心用电增长主因；真正问题在于 PQC 触发的周期性热降频、重协商突发和 GPU 空转会把稳定 IT 负荷变成区间波动，从而削弱年度 PUE/WUE 对边际成本的解释力，并在数据中心集中的节点增加电网调度和负荷建模复杂度。单个 100 MW AI 站点的 5-10 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 32, "matchedTerms": [ "电网", "数据中心", "公用事业", "电力", "能源", "energy" ], "href": "" }, { "date": "2026-06-28", "title": "A股应继续支持AI基础设施主线，但买点应从纯算力叙事转向供电约束受益资产，以35%/30%/25%/10%的科技成长、电网设备、公用事业红利和现金对冲哑铃执行。", "summary": "截至2026-06-28，本卡支持上游判断：AI基础设施交易没有被宏观衰退担忧证伪，而是被供电、电网接入、变压器和冷却约束重构。A股策略应从单一科技成长交易转为35% AI科技成长、30%电网及电力设备、25%公用事业红利、10%现金或对冲的主题哑铃。科技仍有上行，但电子和通信估值已处三年高分位；电网设备承接国网4万亿元十五五投资与数据中心供电瓶颈；公用事…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "电网", "数据中心", "公用事业", "电力", "红利", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "AI数据中心电力瓶颈与核能PPA战略转型", "summary": "本报告分析了2026年AI数据中心面临的核心电网瓶颈，探讨了超大规模厂商绕过电网审批排队、锁定核能PPA（如微软与Constellation的Crane清洁能源中心项目）的战略选择。基于自主测算的200MW数据中心TCO模型，核能直供电相比传统电网虽有价格溢价，但通过提前2年并网运营，可规避高达50亿美元的GPU集群机会成本，在5年生命周期内创造45.52…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31, "matchedTerms": [ "电网", "数据中心", "AI数据中心", "电力", "能源" ], "href": "" }, { "date": "2026-06-26", "title": "随着全球及中国变电设备产能在2027-2029年逐步释放，AI数据中心的电力设备瓶颈将从“无法获取设备”的二进制物理硬约束，降为“支付溢价与关税以获取交期”的资本支出与利润率成本约束。", "summary": "本报告对2027-2029年AI电力设备瓶颈的缓解速度进行了深入分析。结论表明，随着西方重电巨头（如日立能源、GE Vernova）扩产产能释放，以及以特变电工、思源电气、金盘科技为代表的中国高效率设备出口，2027-2029年电力设备将从物理上的“硬约束”转变为“成本/经济约束”。由于取向电工钢（GOES）供应持续偏紧、“铝代铜”面临技术失效以及地缘关税…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36, "matchedTerms": [ "GOES", "数据中心", "AI数据中心", "电力", "能源", "设备" ], "href": "" }, { "date": "2026-06-26", "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大型变压器交期和AIDC电力设备需求支撑；主要反转风险来自重复…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "电网", "GOES", "公用事业", "AIDC", "电力", "设备" ], "href": "" }, { "date": "2026-06-26", "title": "AI电力冲击本身是Fed应看穿的相对价格变化，但其持续、受供给约束且贴近核心服务成本的二阶传导，会让核心通胀在边际上更粘，并支持Fed更久维持限制性政策。", "summary": "截至2026-06-27，本卡综合前序结论：AI电力不是直接核心通胀项目，因为电力在2026年5月CPI-U权重仅2.495%，且core PCE剔除能源；但它可能通过商业电价、云服务、通信、医疗、设施与租金相关成本向60.158%权重的services less energy services传导。IEA预计全球数据中心用电2030年约945 TWh，美…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34.25, "matchedTerms": [ "CPI", "PCE", "数据中心", "电力", "能源", "energy", "macro", "core" ], "href": "" }, { "date": "2026-06-26", "title": "电网基础设施与能源转型对AI算力扩张的潜在约束", "summary": "压力测试部分支持并升级卡03：电力/电网瓶颈是“节流而非断路”，但节流比其暗示更紧更久。在2026–2028视野内，电气基础设施已是排在芯片之前的硬约束——美国电力变压器交期128周、GSU 144周、大单3–4年，价格较2019+77%、达2022前4–6倍；GOES由Cleveland-Cliffs近乎独家供应、全球变压器产能利用率约98%、GSU需求…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32, "matchedTerms": [ "电网", "GOES", "电力", "能源", "设备", "energy" ], "href": "" }, { "date": "2026-06-25", "title": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论", "summary": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论 Feasibility of Bring Your Own Power (BYOP) and Gas PPA for AI Data Centers 在小堆核电(SMR)2030年前难以落地、碳约束拉平公耗曲线的现实背景下，国内与海外AI数据中心采用天然气PPA与企业自备电力(BYOP)的技术路径、套…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 36, "matchedTerms": [ "电网", "数据中心", "AI数据中心", "公用事业", "电力", "设备", "enters" ], "href": "" } ], "timeline": [ { "date": "2026-05-01", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "[auto-handoff] 主权 AI 能源政策对通胀的影响", "summary": "致 主题研究员： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 6b3461d3-5bd5-4a84-a548-967f4c93a4b5 结果摘要: Wrote handoff_response.md (10,637 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "2027 年 HBM4 弹性仍成立，但必须以 firm power 为条件；能源瓶颈更可能造成部署递延、地域重路由和二线买方削单，而不是头部 hyperscaler 普遍取消订单。", "summary": "本卡支持前序 HBM4 推理需求链条，但强调 2027 年边际约束正在从芯片供给扩展到 firm power。100 kW-120 kW Rubin 级机柜会把每 10,000 个机柜转化为约 1.15-1.38 GW 公用事业侧需求。北弗吉尼亚仍能吸收增量但 firm capacity 已被合同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "\"新质生产力\"政策下的中国电网设备出海：退税政策校准与贸易壁垒对冲策略 第8张卡片（共10张）| 中国宏观分析师 | 立场：综合 本期会话主题： 美国 AI 数据中心负荷增长与公用事…", "summary": "\"新质生产力\"政策下的中国电网设备出海：退税政策校准与贸易壁垒对冲策略 第8张卡片（共10张）| 中国宏观分析师 | 立场：综合 本期会话主题： 美国 AI 数据中心负荷增长与公用事业电网灵活性 本卡主题： \"新质生产力\"框架下的电网设备出口退税政策调整与 Section 232/301 关税规避…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 3, "neutral": 0, "value": 28.2, "events": [ { "title": "技术深度是中国电网设备出海的核心护城河，但在美国AI数据中心场景中，第三国制造和退税只能提供部分对冲，真正决定订单能否进入关键供电路径的是可信交付、网络安全、认证和本地服务能力。", "summary": "截至2026-05-06，本卡对第8张卡片作有条件压力测试：技术深度确实比地理套利更重要，但第8张卡片需修正三点。第一，2024年11月中国退税公告取消的是铝材、铜材和相关油脂退税，并将部分成品油、光伏、电池、非金属矿物制品退税率由13%降至9%，不是新的钢铁/GOES退税事件。第二，美国AI数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股变压器/电网设备30-40倍PE：AI送电瓶颈主题是已充分定价还是仍被低估？", "summary": "A股变压器/电网设备30-40倍PE：AI送电瓶颈主题是已充分定价还是仍被低估？ A-share Grid Equipment 30-40x PE: Is AI Power Delivery Bottleneck Fully Priced or Still Undervalued? global-…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "盟友联盟对中国电网设备的部分排除是一项主动用 6–12 个月送电延迟和 15–25% 采购成本溢价换取供应链韧性的宏观选择，结果是延展盟友设备 OEM 超级周期、部分上限美国 AI …", "summary": "本卡作为 10 张卡片的最终综合，整合了从需求（Research note 01）、firm 负荷修正（Research note 02）、供给瓶颈（Research note 03）、设备防御性（Research note 04）、定价权（Research note 05）、跨资产轮动（Rese…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "AI CapEx并未因外壳新开工放缓而失效，但其落地速度正被电力、变压器、配电网络和电网稳定性重新定价。", "summary": "截至2026-05-13，本卡支持“密度胜过足迹”主线，但补充能源约束：AI算力密度提升使稀缺资源从外壳数量转向可供电、可散热、且符合电网稳定要求的容量。LBNL、EPRI和EIA均显示数据中心用电占比快速上升；NERC 2026-05-04 Level 3警报和2024年1,500 MW数据中心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 2, "neutral": 0, "value": 37, "events": [ { "title": "能源获取难度要求对AI基础设施进行选择性下修：终局需求不变，但IDC和算力租赁的CAPEX转化曲线、利用率和ROIC应显著打折。", "summary": "截至2026-05-15，本卡对电力瓶颈对AI基础设施的影响进行TMT压力测试。结论是：不应全面下修AI算力长期需求，但必须下修CAPEX转化节奏、利用率爬坡和ROIC。IEA预计数据中心用电从2024年415 TWh升至2030年约945 TWh，同时约20%规划项目存在电网延期风险；Micro…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源扩容是真实利多，但不足以在2H26-2027年完全对冲AI电力结构性瓶颈，因此应持有电力瓶颈受益方向，同时折价近端GPU交付曲线。", "summary": "截至2026-05-15，能源供给扩容可以缓解但无法完全解除2H26-2027年AI数据中心电力瓶颈。核能主要在2030年后才有效，Three Mile Island重启也偏2028年；2026-2027年美国69 GW太阳能、19 GW风电和ERCOT电池扩容能补充电量，却不能自动形成关键节点2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 2, "neutral": 0, "value": 38.4, "events": [ { "title": "AI 负荷不是自动的公用事业盈利红利；只有把数据中心并网需求锁进成本归因费率和金融承诺的辖区，才能把输配电升级转化为低反弹的 allowed earnings。", "summary": "截至2026-05-15，本卡对电力供给乐观判断做公用事业费率压力测试：AI 负荷能否变成 allowed earnings，取决于申请 MW 是否在 capex 前被转化为长期合约、最低需量收费、抵押品、退出费和独立大负荷类别。最佳辖区是 AEP Ohio、Texas TSPs、Georgia …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "我国能在全国电量层面跟上AIDC基准需求，但局部电网承载力、小时级稳定供电和物理绿电交付仍是规模化扩张的硬约束。", "summary": "截至2026-05-15，本卡压力测试卡片07：国产变压器/GIS交付提速可以缓解AIDC物理接入瓶颈，但不能自动解决电力系统承载问题。全国年度电量层面，可再生能源新增发电量和2026年新增装机足以覆盖基准情景下的数据中心增量；但AIDC负荷高度集中、要求稳定MW和高电能质量，约束会转向区域电网节…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "煤电与可再生能源能缓解 AI 用电增量，但在 2026-2028 年无法单独解决局部可靠容量与 time-to-power 瓶颈。", "summary": "截至2026-05-16，本卡支持前序“电力可交付性”瓶颈判断。煤电和可再生能源在总量上可覆盖 AI 数据中心新增用电的一部分，但无法在电网扩容受限时单独填补缺口。美国煤电 YTD 发电量同比 -11.8%，风电仅 +1.4%，太阳能 +20.1%，但公用事业总发电仅 +0.2%；PJM/MISO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在铜铝供给弹性有限的环境下，2026-2030年可再生能源扩张的关键瓶颈将从发电设备转向电网扩容、并网能力和关键电力设备交付。", "summary": "2026-05-17：本卡支持Research note 04并从公用事业角度确认，铜铝供给弹性受限会把电网扩容推成大规模可再生能源并网的主要物理瓶颈。IEA显示电网建设需5-15年，全球超过2,500 GW项目卡在并网队列；LBNL显示美国截至2024年底有1,400 GW发电和890 GW储能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "电力系统能让AI建设继续推进，但无法按无约束AI MW需求的理想速度交付firm并网容量，因此电力接入将被价格、排队、地理迁移和现场发电配给。", "summary": "截至2026-05-17，能源侧压力测试支持Research note 07“单位token能耗下降但总MW上升”的判断，但指出真正瓶颈是正确节点、正确时间、可并网的firm MW。美国2026年计划新增86 GW装机，但以太阳能和电池为主，不能等同于24/7数据中心电力；LBNL预计2028年美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 0, "neutral": 1, "value": 57.6, "events": [ { "title": "在系统性ERP下行期，顺周期进攻第一梯队应聚焦于具备核心高毛利出海定价权的电力设备以及受益于动车组五级修维保周期刚性爆发、实现超高运营杠杆转化的轨交装备。", "summary": "截至2026-05-21，随系统性ERP压降与早周期主线确立，先进制造细分板块因不受地产负债表侵蚀成为核心配置。运营杠杆测算显示：轨交装备高级修因五级修爆量（2025/2026年均约550-600组）及高毛利率（~38%），运营杠杆达3.8-4.5，利润弹性极强；电力变压器出海在AI数据中心与欧美…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 2, "value": 56, "events": [ { "title": "各州零售电价保护壁垒、FERC表后净额结算封堵以及变压器等下游设备交期的“三重屏障”，将严重滞缓AI电力红线的物理释放，并大幅压缩发电巨头的表后溢价红利。", "summary": "2026年5月21日，公用事业分析师对AI电力的乐观增长假设进行压力测试。分析指出，各州监管出台了弗吉尼亚州GS-5专用电价类等条款以保护居民用户；FERC在2025年12月彻底封堵了表后直供（BTM）免过网费的净额结算漏洞，使得核电直供数据中心到岸电价攀升超22%；叠加电网物理互联排队超55个月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心暴增的电力需求正推动能源与电力板块从防御性品种质变为AI成长基础设施代理，超级并购与清洁基荷溢价印证了这一中观行业重估。", "summary": "截至2026年5月21日，高盛预测美国数据中心电力需求在2025-2027年间将以45.9%的CAGR爆发，占夏季电网峰值负荷的8.5%，驱动55.6%的电网增量需求。这引发了能源板块向AI算力基础设施代理的质变。NEE与Dominion的670亿美元合并，以及CEG/VST估值重置为成长股中枢，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "能源供给侧足以支撑AI硬件的长期需求，但短期决定出货节奏的是节点级通电能力，而不是全国电量总量。", "summary": "截至2026-05-21，能源侧扩张能够维持AI硬件长期需求，但不能完全消除2026-2027年的局部“电力之墙”。美国2026年计划新增86 GW公用事业级容量，绿电与储能增量可观；中国凭借风光装机、储能和State Grid的RMB 4 trillion电网投资，缓冲能力更强。但变压器、并网队…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 5, "neutral": 0, "value": 35.4, "events": [ { "title": "杰文斯悖论压制能效红利——AI算力的累计电力需求增速比能效改善快约一个数量级，结构性电网紧缺无法被脱钩。", "summary": "从能源供应侧压力测试Research note 04的「TMT脱钩」论。IEA、LBNL、EPRI、高盛数据一致显示：尽管Blackwell+MoE+FP4等能效改善真实存在，全球数据中心电力需求仍将从2022年460 TWh翻倍至2026年逾1,000 TWh，2030年达约945 TWh——杰…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器等电力设备的极度短缺及取向硅钢供应瓶颈验证了“时间尺度错配”，AI数据中心的扩张将持续受制于3-4年的物理电网硬件交付周期。", "summary": "截至2026年5月22日，全球电力设备市场面临极度紧缺。西门子、GE等四大设备商订单积压创历史新高，大型电力变压器交付周期拉长至36-48个月。上游取向硅钢（GOES）受制于北美产能垄断及全球供应不足，成为扩产的核心瓶颈。设备可用性已取代资金，成为AI基建最硬的约束。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Capex韧性 × 油价冲击 × 联储延宕的共振，通过工业PPI与电力催生海外隐性二次通胀，将10Y美债锚定在5.0–5.3%，断裂点在日元/JGB而非美国风险资产。", "summary": "在2026-05-23工作日期下，AI硬件Capex韧性与霍尔木兹油价冲击的共振真实存在，但传导通过工业PPI、电力价格与资本品平减指数发生，而非头部CPI/PCE。这一通胀路径让联储基于PCE的反应函数系统性低估真实通胀，将10Y美债锁定在5.0–5.3%走廊延续整个2026下半年（期限溢价回升…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张确实开启电力基础设施资本开支周期，但2026-2028年变压器、并网排队和24/7可靠性要求会先放大时间尺度错配，再逐步由费基投资、储能和可调度电源缓解。", "summary": "截至2026-05-22，本卡支持并强化能源分析师的“时间尺度错配”判断：AI数据中心负荷增长快于电网硬件、稳定电源和新能源并网的交付速度。IEA预计全球数据中心用电量从2025年485 TWh升至2030年950 TWh，DOE/LBNL预计美国数据中心2028年用电量达325-580 TWh；…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "AI基础设施的近期上限将由可靠通电MW决定；电力在拥挤区域会先于GPU成为硬约束，但全球层面仍是区域性、可部分迁移的约束。", "summary": "截至2026-05-24，本卡压力测试结论是：AI数据中心电力与电网设备约束在核心拥挤区域已经先于GPU形成硬约束，但不应上升为全球绝对缺电叙事。GB300 NVL72单机架功率达132-140 kW，100 MW IT功率仅支撑约714-758个机架；同时IEA、DOE/LBNL、NERC和NI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "电网硬件瓶颈把AI云竞争从GPU采购变成通电确定性和单位算力TCO竞争，已锁电的平台获得确定性溢价。", "summary": "截至2026-05-25，本卡对“AI稀缺锚从GPU切换到可并网兆瓦”做压力测试。结论支持主线但强调TMT层面差异：电网硬件瓶颈不会均匀终止AI CapEx，而是降低集群交付确定性并抬高云TCO。美国数据中心用电到2028年可能达325-580 TWh；变压器最长交期可达四年；PJM批发电力成本2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 3, "neutral": 0, "value": 43.6, "events": [ { "title": "AI算力激增带来的电力饥渴非但没有侵蚀能源安全杠铃的防御性，反而将核电、煤炭与电网设备重塑为高定价权的结构性成长引擎，构成了与硬科技仓位自我对冲的稳健闭环。", "summary": "截至2026-05-26，压力测试表明，AI数据中心带来的暴增电力需求非但没有侵蚀能源安全杠铃的防御性，反而重估了清洁基荷（如核电）与化石能源（如煤、气）的稀缺价值，将其重塑为高定价权的成长引擎。物理电网瓶颈对AI算力扩张起到物理降温降速的对冲作用，强力验证了60/40杠铃配置的稳健性。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BBB级电网与电力公用事业公司债是当前AI融资链中被严重错误定价的环节，未反映监管费率疏导脱轨及单一项目集中度风险，将最先面临降级与利差剧烈拓宽的重定价冲击。", "summary": "截至2026-05-27，本卡针对AI capex融资链中三大信用主体（云巨头、数据中心REIT、电网公用事业）在30Y美债破5%及长期高利率环境下的利差定价进行了压力测试。测试表明，云巨头利差T+95 bp定价公允，数据中心REITs具防御性；而BBB级公用事业债因1.3万亿美元的庞大电网开支、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网扩容的第二瓶颈不是资本，而是重电设备的合格产能槽位及其上游材料。", "summary": "截至2026-05-27，本卡支持前序判断：AI扩张的第一瓶颈是电网可用容量，第二瓶颈正在转向大型电力变压器、开关设备、电缆和专业施工能力。DOE、Wood Mackenzie和JLL数据显示，LPT进口依赖、GOES/CTC材料短缺、128-143周变压器交付周期、44周开关设备周期和数据中心项…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 4, "neutral": 0, "value": 39, "events": [ { "title": "公用事业板块在2026年已演变为AI经济的增长底座，建议重点布局算力枢纽枢纽区域电力及碳信用变现能力强的垃圾发电标的。", "summary": "2026年公用事业板块正经历从防御红利向“AI增长基础设施”的结构性重估。AI数据中心扩张驱动电力需求超预期增长（2026年预测+5.7%），同时碳市场重启为垃圾发电龙头带来显著的CCER收益。识别出深圳能源、瀚蓝环境、淮河能源和甘肃能源为具备“电力+AI”或“碳资产”逻辑的低关注度标的，需警惕局…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 电力缺口是真实但温和的核心 PCE 楔，叠加电网设备供应链的中美对抗，将以「延迟降息 + 类滞胀」形态而非「重置通胀目标」形态影响美联储，并把宏观风险从拥挤的一阶 AI 硬件转…", "summary": "AI 电力交付滞后通过三条渠道（批发电价→服务、云算力涨价→SaaS、变压器 PPI→capex 平减指数）对美国核心 PCE 形成 2026 +8~18 bp、2027 +10~22 bp 的持续抬升（自测算），不足以重置 2% 目标但足以让核心 PCE 2027 年底仍卡在 ~2.4%，相对 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "重型电力设备OEM收入上能承接需求、时间上承接不了——订单簿2.5–4年可见度+定价权+利润率扩张构成对科技久期的反向对冲，核心超配GEV/ENR.DE/ETN/Hitachi系，中…", "summary": "2026-05-28结论：本卡支持第03卡的电网瓶颈论断。Siemens Energy Grid Tech订单/营收比2.28×、储备490亿欧元；GE Vernova Electrification订单+86%、设备储备390亿美元、B2B约2.5×；Eaton电气储备同比+48%、数据中心订单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "把AI久期空头与电网OEM/GOES/CLF多头拼成的组合净Beta+0.45、净久期−1.5年，是'高利率持续'路径下的反AI久期+实物瓶颈租金套利，需配置USD多头FX叠加并设置…", "summary": "2026-05-28 synthesize：把第01–05卡拼回组合（USD 100mm基币）。AI久期空头−15%、AI硬件多+10%、全球电网OEM+30%、亚洲OEM+10%、GOES上游+8%、CLF垂直一体+7%、自动化+5%、现金+25%。净Beta约+0.45、净久期约−1.5年（受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 1, "value": 61.4, "events": [ { "title": "2026-05-28 核心判断：美国电网瓶颈的宏观风险不是一次性CPI爆点，而是受监管电价的多年粘性和长久期公用事业融资供给对美债曲线的边际挤压。", "summary": "截至2026-05-28，本卡对电网/公用事业链做压力测试：美国大型电力变压器2019年进口依赖度82%，常见交期36个月、最高60个月；2026年关税框架把部分金属密集型电网设备纳入15%关税至2027年。结论是，变压器短缺不会单独制造广义CPI冲击，但会通过费率申请形成粘性的电价通胀底部；同时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX上市与美国电力法案共振下的全球AI算力物理层与BTM重估", "summary": "SpaceX上市与美国电力法案共振下的全球AI算力物理层与BTM重估 SpaceX IPO and U.S. Electricity Bill Converging on Global AI Physical Layer & BTM Revaluation 伴随SpaceX的800亿美元IPO（S…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 5, "neutral": 0, "value": 35.8, "events": [ { "title": "关键设备交付周期拉长至2.5-3年以上，且受限于GOES与铜的全球结构性短缺，电力设备产业链的产能扩张短期内无法匹配AI数据中心爆发式需求，行业物理瓶颈为设备OEM厂商（GEV、VR…", "summary": "截至2026年5月29日，在AI算力与电网升级需求爆发下，电力设备产业链（GEV、VRT）面临严重物理产能瓶颈。2026年变压器交付周期拉长至128-144周。虽然各大厂商积极在越南和美洲扩张产能，但受限于取向电工钢（GOES）和铜的结构性短缺（GOES在2025年存在约30%供需缺口且将持续至2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力的物理约束正从“有没有电”转向“能否在正确节点、正确时点、用合格设备按时送电”。", "summary": "2026-05-29：本卡支持上一张TMT二阶硬件主线，但加入电力基础设施折扣。AI算力扩张不是中国全国层面的总量缺电问题，而是局部电网容量、变压器/开关设备交付、新能源消纳和高功率密度冷却/备电的时点约束。中国互联网数据服务用电1-4月同比增长44.4%[S2]，但仅占全国同期用电约0.94%[…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-05-29，美国贸易壁垒和本土化采购要求会把AI数据中心并网延迟从设备短缺放大为合规与地缘供应链问题，中国及亚洲供应只能边际补缺，难以快速突破美国瓶颈。", "summary": "截至2026-05-29，本卡支持前序结论：AI并网瓶颈不只是变压器和GOES的物理短缺，美国Section 232、Section 301和Buy America规则把进口替代变成更慢、更贵、更难合规的路径。中国2025年GOES产量和出口创新高，但中国原产变压器与电工钢面对额外关税和原产地审查…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI与电力基建的巨额资本开支热潮正反噬结构性通胀，将实际中性利率（r-star）结构性推高约80基点，刚性垫高美联储降息周期的终端利率至4.00%-4.25%，锁定Higher fo…", "summary": "在轻滞胀（GDP 1.6%，PCE 3.8%）背景下，本卡片对AI及电网基建热潮进行宏观压力测试。研究表明，高资本强度的基建开支正在通过重型设备PPI上涨、公用事业“资产基础”机制转嫁电价以及物理挤出效应反噬结构性通胀，在结构上将美联储长期实际中性利率（r-star）推高约80bp，从而将名义终端…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 3, "neutral": 0, "value": 57.2, "events": [ { "title": "SiC上游产能过剩带来的器件价格暴跌，通过缩减SST与PCS的占地及散热要求转化为下游巨大的CAPEX红利，系统集成商借此成功跨越内卷，获取技术溢价。", "summary": "本次调研确认，SiC衬底的产能过剩正推动利润池向电力电子产业链下游转移。SiC器件成本降至2-3美元/A，较硅基IGBT溢价压缩至10-20%。通过10-50 kHz的高频开关，SiC使中压SST与储能PCS的磁性元件体积缩减60-70%、散热需求减半，实现单MW系统体积从2.0 m³缩至1.2 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "配电侧变压器与开关柜已锁定2026-2028量价齐升超级周期，金盘科技、特变电工、思源电气、中国西电、平高电气是A股最高确定性的物理底座组合。", "summary": "截至2026-05-31，AI数据中心驱动的配电侧设备已进入结构性供应缺口与超级周期：美国大型电力变压器交付周期≈128周(标准)/144周(GSU)，高定制订单达4年；GE Vernova 2026Q1单季为数据中心新增订单$2.4B，超过上一财年全年；MV/LV开关柜lead time普遍14…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "能源供应端在2026-2027年间存在不可逾越的物理建设时滞，电价冲击将严重侵蚀云巨头算力TCO与ROIC，迫使其放缓采购，投资者应战略撤出算力硬件并超配长江电力与中国神华等避险公用…", "summary": "截至2026年5月31日，本报告支持前序研判，确认核电与燃机等能源建设周期（3-10年）根本无法填补2026-2027年间的“电力断层”。SMR商业化推迟至2030年后，燃机面临3年交期与气源管道等物理瓶颈，微电网因设备短缺同样受阻。压力测试表明，电价上涨将暴击云巨头GPU算力TCO，使其利润率转…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在设备全球极度短缺（交付期128-160周）与原材料70-85%传导下，Tier-1电力设备供应商确立了结构性成长与绝对定价定价权，而重资产建设型公用事业面临电价政治阻力与融资挤压，…", "summary": "截至2026-05-30，本压力测试表明，在铜铝紧缺与电力设备70–85%金属价格联动下，全球公用事业因美债5%与电价费率政治监管阻力，现金流严重受损；而中国国家电网凭借超低国内融资成本及特高压自主产业红利，韧性极高。Tier-1电力设备OEM因128–160周超长变压器交付周期握有绝对定价权，是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 5, "neutral": 0, "value": 38.6, "events": [ { "title": "电网并网瓶颈不是欧美潜在GDP的硬天花板，但已是AI生产率红利兑现时点的2-3年延迟与核心商品PPI的持续高位驱动；Section 232升级是决定它停留在过渡税还是固化为真正上限的…", "summary": "截至2026-05-31，本卡片对Cards 03-05的微观结论做宏观压力测试：电网瓶颈+OEM定价权意味着电力设备PPI（BLS WPU1174自2020年累计+65-70%）将持续高于核心目标，并把成本转嫁给整体欧美Capex基数。美国T&D投资从2024年约1700亿美元升至2025E的2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与开关柜瓶颈仍足以支撑ETN/HUBB利润弹性，其中ETN胜在绝对利润池和AI数据中心杠杆，HUBB胜在更纯的公用事业T&D与变电站beta。", "summary": "截至2026-05-31，本卡支持AI电力设备链条：变压器与开关柜交付周期仍处于多季度到多年级别，客户提前锁定工厂槽位，使电气设备龙头拥有价格与积压订单质量。ETN因Electrical Americas规模、数据中心订单和Q1 2026 backlog弹性，适合作为工业主仓；HUBB因Utili…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件底座面临物理电网与能源成本的双重严峻约束，其过度偏乐观的估值预期存在巨大脆弱性，而由长江电力与中国神华组成的公用事业防御组合展现出极强的抗风险红利韧性，完全应验并巩固了哑铃资…", "summary": "截至2026-05-31，本卡片对AI算力扩张的电力约束与公用事业防御组合的股息率稳健性进行了压力测试。定量仿真显示，AI算力在电网瓶颈与高电价下极其脆弱：若电价上涨至$0.220/kWh，云巨头GPU租赁毛利率将转负为-1.48%，导致Mag7估值乘数面临大幅收缩。相反，以长江电力和中国神华组成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘政治与工业保护主义正将AI电网从中游环节‘脱钩化’，导致供给瓶颈行政化，显著加剧了算力扩张的物理墙刚性与Capex溢价。", "summary": "2026年6月调研显示，美中欧三方工业政策正合力‘加固’AI电网的物理瓶颈。美国通过《国防生产法》及EO 14156宣布能源紧急状态，对GOES及变压器实施Section 232高额关税并强化‘本土熔炼’规则，旨在保护Cleveland-Cliffs的单一供应垄断。中国则从2026年初起实施GOE…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 2, "neutral": 1, "value": 61, "events": [ { "title": "AI电力约束不是全球TWh硬上限，而是区域电网可交付能力上限：谁能获得firm节点容量、长交期设备和可回收tariff，谁才能把算力订单转化为可运行园区。", "summary": "截至2026-06-02，本卡压力测试AI数据中心电力约束：全球年度电量尚可管理，但Northern Virginia/PJM等热点的本地可交付容量、变电站、变压器、电缆和峰时备用已成为算力扩张的区域性物理上限。IEA预计数据中心用电从2024年415 TWh升至2030年945 TWh；LBNL…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "绿电直连T&D费刚性缴纳压制直连项目IRR约2pp，定向重创'绿电+AI'叙事，但对全国新能源总装机拖累有限（<2%），Research note 02的公用事业防御墙以新能源开发商…", "summary": "发改能源〔2025〕650号要求绿电直连项目刚性缴纳输配电费（大工业档约0.13元/kWh含基金）。压力测试显示：直连路径IRR由基线7.0%压缩至4.7%–5.2%，下行约2pp，预计2026–2028年AI数据中心直供型新能源装机下调15%–25%（每年2–5GW）。但占国内新增装机80%+的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "省级电力治理已通过'PUE差别电价挤东部+低到户电价吸西部+绿电直连穿透溢价+算力券补消费'的四层分流框架，使AI数据中心承担分层电价与维护工业竞争力两个目标兼容，构成P/Watt杠…", "summary": "截至2026-06-02，地方政府与省级电力监管已在制度上允许AI数据中心承担分层电价，且未削弱区域工业竞争力。机制是'东数西算+绿电直连+PUE差别电价+算力券'四层地理-用途分流：东部以北京2026年PUE>1.35加价0.2-0.5元/千瓦时挤出低效数据中心、保护工业额度；西部以0.32-0…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 4, "neutral": 0, "value": 47.4, "events": [ { "title": "美国2026-30年电力供应的真实上限并非资本而是物理并网权与基荷稀缺性，估值红利正从纯需求侧向掌握‘物理准入门槛’的核电IPP、现场发电方案及优势公用事业标的快速迁移。", "summary": "到2030年，美国AI数据中心将面临45GW的新增负荷缺口。并网排队时长稳定在4年以上，使得拥有现成并网权的核电站和退役电厂（棕地）成为算力扩张的‘金票’。PJM容量价格暴涨9倍并触及监管上限，显示供应极度稀缺。为了绕过电网瓶颈，行业正转向BTM自备电源（如Bloom Energy的燃料电池，可在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "供应链脱钩带来的成本上升并未抑制需求，而是将全球能源转型的驱动力从‘成本红利’转向‘安全与物理约束下的刚性支出’，算力溢价正成为吸收绿色通胀的主要缓冲机制。", "summary": "截至2026-06-02，全球能源转型正进入‘高成本、低弹性’的需求超周期。尽管美欧供应链脱钩导致局部LCOE上升10-15%，且变压器等关键设备面临30-50%的脱钩溢价，但AI算力负荷的刚性增长和电网现代化的迫切需求抵消了成本抑制。超大规模云厂商和公用事业公司正在吸收‘绿色通胀’，优先确保交付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "热密度跳变让液冷成为前提而非可选，但电网容量与HVDC配电设备升级才是2026-2028 AI算力扩张的真正瓶颈与超额收益所在，IDC运营商应从防御底仓上调为攻防一体。", "summary": "以2026-06-03为基准日，从能源系统视角对前序'算电协同→Token通缩'叙事进行压力测试。先进封装把单机架功率从10-20kW推到130-250kW，液冷从'选项'变成'前提'；但即便PUE从1.4压到1.1，AI数据中心总用电仍按3-5×线性扩张，公用事业体量不被'通缩'。压力测试发现：…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高昂的宏观成本（高油价与高利率）与长达三年的并网瓶颈，使得全球能源供给难以同步匹配AI算力扩张，AI估值面临物理基建的硬约束。", "summary": "截至2026年6月2日，全球电网扩张速度无法匹配AI算力激增。在94美元油价与4.65%美债收益率挤压下，数据中心面临度电成本飙升与120-150周的变压器交付瓶颈。这印证了前述宏观压力制约AI硬件估值的判断，促使资金向掌握基荷电力的公用事业等防御性红利资产轮动。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 2, "neutral": 0, "value": 60.8, "events": [ { "title": "AI 公用事业权重下调后对电网基础设施与能源基建板块的风险覆盖评估", "summary": "2026-06-03，本卡压力测试 Research note 06 的 22% 至 17% AI 公用事业降权。结论：降权合理，但必须从长久期美国发电 / IPP 暴露中执行，不能压缩电网设备底仓。IEA、EIA、PJM、DOE 与 Eaton / Quanta / Vertiv 数据显示，AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建扩张短期更像通胀性的全球电网和材料资本开支周期，而不是无摩擦的生产率冲击；能否把它转化为中长期增长，取决于电网、变压器、铜和成本分摊规则的同步扩张。", "summary": "截至2026-06-03，本卡支持上一张公用事业约束论点，并将其上升为全球宏观资本开支周期：数据中心用电从2025年的485 TWh走向2030年约950 TWh，美国、欧洲和日本/韩国分别面临峰值负荷、电网老化、能效和低碳基荷约束。AI建设先争夺铜、GOES、变压器、燃机和电网工程能力，再释放生…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 1, "neutral": 1, "value": 71.2, "events": [ { "title": "变压器与设备供应缺口延缓了受监管公用事业费率基数的短期转化并恶化了现金流，但物理稀缺性结构性地推高了存量电网、已锁定的并网项目和具备溢价能力的电网设备制造商的重估价值。", "summary": "截至2026-06-05，受AI数据中心和电气化推动，变压器供需缺口达30%，大型电力变压器交货期拉长至36-48个月，价格比2019年暴涨70%-95%。定量模拟表明，30%的资本支出因瓶颈延迟2年竣工，在产能预付模式下会导致公用事业公司5年累计净利润减少10.71亿美元（-3.94%），并产生…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国电网容量危机正通过“工业高电价-核心通胀粘性-美联储鹰派转向”的链条，对制造业再回流构成实质性压制，并重塑美联储的降息路径。", "summary": "截至 2026 年 6 月 5 日，美国电网容量危机已从公用事业局部问题演变为核心宏观阻力。PJM 容量价格连续两年触及上限（2027/28 年度清算于 333.44 美元/兆瓦-日）且存在 6.6 吉瓦缺口，导致美国东部工业电价上涨 15-25%。这种“供给侧冲击”正通过成本转嫁推升核心 PCE…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "AI电力需求强化了从拥挤AI股票向实体基础设施的轮动逻辑，但可投资主线是time-to-power稀缺，而不是泛能源重估。", "summary": "截至2026-06-06，AI数据中心用电增长已从主题叙事进入电力系统硬约束：IEA预计全球数据中心用电从2025年485 TWh升至2030年950 TWh，LBNL给出美国2028年325-580 TWh区间。PJM已出现容量市场和输电压力，数据中心贡献约30 GW的2030年前峰值负荷增长。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 3, "neutral": 0, "value": 59, "events": [ { "title": "AI资本开支周期有条件可持续，最佳组合不是押注单一芯片beta，而是在算力咽喉和电力咽喉之间做哑铃，并用公用事业与铜/GOES等材料做压舱石。", "summary": "截至2026-06-06，本卡综合前序结论：1.7万亿美元AI数据中心基础设施周期仍可配置，但核心约束已从芯片供给转向送电能力。建议AI-电力主题仓按28% AI半导体咽喉、27%电网设备、20%公用事业、15%铜/GOES等材料、10%流动性与对冲配置。结论支持继续做多AI-电力综合体，但反对纯…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大水电承担利率敏感型'物理红利'防御角色，BTM出海链承担AI物理瓶颈套利的进攻角色，构成能源行业最清晰的双锚配置。", "summary": "截至2026-06-06，央行5月底降准50 bps推动10年期国债收益率跌破2.0%，叠加全国电力市场化交易占比超70%，大水电（长江电力为代表）与BTM出海链构成A股内一组互补的'物理红利'。大水电凭借水文均值回归、近零燃料成本与60%+毛利率提供类永续债现金流，股债利差由132 bps走阔至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施压力测试：能源“供需墙”的结构性瓶颈 — 2026 06 07 作者： 能源行业分析师 日期： 2026 06 07 立场： 压力测试 (Stress test) 执行…", "summary": "AI 基础设施压力测试：能源“供需墙”的结构性瓶颈 — 2026 06 07 作者： 能源行业分析师 日期： 2026 06 07 立场： 压力测试 (Stress test) 执行摘要 截至 2026 年 6 月 7 日，TMT 硬件行业的“进攻型”转型正面临一个物理极限： 能源电力墙 。尽管半…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 72.4, "events": [ { "title": "AI电力负荷会在2026-2027年先推高区域批发容量与电网资本开支，再以12-30个月滞后进入零售电价；2027年全国通胀推力温和，但区域服务成本压力显著。", "summary": "截至2026-06-08，本卡片综合判断：AI数据中心负荷对美国电价的传导不是线性CPI金属冲击，而是通过批发容量市场、输配电rate base和零售费率案件滞后体现。数据中心用电从2023年的176 TWh可能升至2028年的325-580 TWh；全国居民电价层面，2027年大概率额外增加1.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 2, "neutral": 2, "value": 73.8, "events": [ { "title": "[auto-handoff] AI数据中心电力瓶颈", "summary": "致 主题研究员： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 75d7af3c-9f25-455d-ac84-6b2414e29c02 结果摘要: I will run the command to verify today's date in the shell en…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支正从算力硅片向电力物理基础设施发生不可逆的溢出，A股配置须从TMT硬件向电网出海及系泊设备进行长周期大轮动。", "summary": "截至2026年6月4日，53个月并网排队与10倍容量电价标志着AI产业进入电力死锁。云巨头Capex正从算力硬件向电网基建与分布式电力物理溢出。A股策略应低配估值透支的TMT算力硬件，大幅轮动至受益于海外电网建设红利的变压器/高压线缆出海企业，并战术配置浮式数据中心催生的系泊设备及高股息公用事业。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 电力瓶颈的可投资因子已经从“需求增长”迁移到“2028 年前可通电、可签约、可合规回收成本的兆瓦”。", "summary": "截至 2026-06-08，本卡综合前序 AI 电力瓶颈链条，支持 Research note 9 将 BTM 速度溢价收窄为“可交付电力溢价”。配置上不再泛化买能源主题，而是超配拥有可签约稳定兆瓦的 IPP、具名数据中心管输/现场发电合同的中游天然气、以及受益于燃机与电气化设备短缺的 OEM；h…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI推理能效提升对电力基础设施需求的边际影响", "summary": "能源回升真实但落在不同栈层：Jevons回弹推高总量电力需求（IEA：485→950 TWh/2025-2030；AI数据中心2025年耗电激增50%），PJM容量价从$28.92飙至$333.44/MW-日，数据中心占164亿美元拍卖的40%。但这是电网/用电户与云厂商总账单成本，非SaaS每t…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 2, "neutral": 0, "value": 69.2, "events": [ { "title": "AI算力需求激增带来的电力负荷压力与公用事业响应能力评估", "summary": "支持 card-01「电力为主导跨链风险」的框架，但就中国收紧：电力是可管理的赋能层，而非硬天花板。AI数据中心IT能耗2025年77.7 TWh、2027年146.2 TWh（CAGR 44.8%）；中国数据中心总用电2025年193.3 TWh、+17%，2026年1-2月+46.2%。供给侧…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力的“物理壁垒”正将全球投资范式从“虚拟算力（软件/芯片）”推向“物理确定性（电网/基荷/设备）”，建议通过中美区域套利与基建杠铃组合锁定跨周期收益。", "summary": "截至2026-06-09，本报告针对AI电力瓶颈引发的全球区域竞争力分化，提出“物理AI”投资策略。美国凭借节点低电价和核电复兴维持算力核心地位，中国通过电网建设速度和可再生能源盈余转化为设备链和高耗能产业优势，欧洲则面临区域碎片化导致的去工业化压力。组合配置应向“价值+确定性”风格轮动：超配具备…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 2, "neutral": 1, "value": 73.6, "events": [ { "title": "电网设备超长交付周期及原料紧缺导致大型电力项目面临商运推迟与债务违约风险，使AI数据中心容量扩张受到刚性的“瓶颈延迟”制约，佐证了“AI疲劳”本质上是物理并网与供给瓶颈而非需求侧断崖…", "summary": "截至2026-06-11，电网变压器设备供应缺口及原材料（取向硅钢/GOES与铜）紧缺已对AI数据中心并网建设进度构成物理刚性约束。LPT与GSU变压器交付期已拉长至2.5-4年。由于DOE能效新规，美国GOES缺口扩大至19.6万吨/年，进口依赖度攀升至57.6%。即便采用以铝代铜，也会因线圈体…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO 不是充分的卖出 AI 龙头信号；更强的能源判断是 AI 算力正在被重新定价为电力支撑的基础设施资产。", "summary": "截至2026-06-11，本卡片否定“SpaceX IPO 本身触发广谱 AI 龙头去拥挤”的简化判断。能源证据显示，真正约束是可靠电力、电网接入、容量市场和燃料结构。Google-SpaceX 算力合同证明已交付算力的稀缺价值，但 PJM 容量价格、IEA/DOE 数据中心用电预测和 Grid …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "公用事业产业链正处于电价市场化、核电扩张与 ETS 重启三轴重定价中，超额收益落在 003816、600803、600905、600008、1164.HK 五个低覆盖度标的，主流的长…", "summary": "公用事业产业链当前 30 日激活率仅 11.7%（24/205），叙事被长江电力 600900 一只标的吸走，但 2025-02 风光电价市场化、2026-03 十五五核电 110 GWe 目标、2026-01 ETS 重启叠加 CEA-2025 切换三大监管冲击在 12 个月内连发，买方盈利模型…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 3, "neutral": 0, "value": 66, "events": [ { "title": "材料供应链的物理硬约束（变压器4年交付期与铜矿缺口）将导致电网扩张速度远滞后于AI需求，从而推迟公用事业公司的收益兑现并压低利润率。", "summary": "2026年6月调研显示，AI电网建设已撞上“物理材料墙”。大型电力变压器交付周期拉长至3-4年，核心瓶颈在于取向硅钢（GOES）的结构性短缺和精炼铜高达60万吨的缺口。AI数据中心耗铜量是传统设施的3-4倍，材料成本通胀和设备交付延迟将显著推后公用事业公司的业绩兑现，挑战了“电力决定论”下的短期超…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建的“脱网化”倾向并非缓解电网压力的良药，而是通过抢夺核心电力设备与稀缺劳动力，并造成公共电网固定成本向剩余用户转移，实质上加剧了传统产业的能源成本负担。", "summary": "截至2026-06-12，AI数据中心试图通过表后直供（BTM）和微电网“逃离”公用电网的努力正遭遇物理与监管的双重瓶颈。研究显示，BTM模式因电力成本分摊不公引发监管反弹（如FERC对Susquehanna协议的否决），导致超大型项目转向“表前”（FTM）模式，加剧了电网容量压力。同时，AI基建…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "夏季能源通胀反弹与电网基荷电力短缺构成了双重尾部风险，足以打破美联储耐心和AI资本品扩张的“双速经济”假设，多资产策略必须引入能源与公用事业对冲。", "summary": "截至2026-06-12，本报告对首席策略师的“双速经济”配置策略进行了能源维度的压力测试。指出夏季高温和活跃飓风季可能推高天然气与原油价格，引发整体CPI超预期反弹，打破美联储“保持耐心”的假设并引发股债双杀。同时，电网基荷电力短缺构成了AI资本品实质上电的物理硬约束，可能切断资本品扩张动能。建…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 3, "neutral": 0, "value": 67.4, "events": [ { "title": "AI 电力瓶颈不是总装机不足，而是2026-2028年本地可交付、可调度、已并网电力的时间错配。", "summary": "截至2026-06-13，本卡支持前序“电力交付约束会惩罚 AI capex/融资假设”的主线，但将结论精确化为：总能源不短缺，本地、可靠、按时并网电力短缺。美国2026年计划新增86 GW公用事业级装机，但主要来自太阳能、储能和风电；数据中心用电到2028年较2023年隐含增加约17.0-46.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在本土变压器长交期与并网延时常态化下，美国数据中心向非传统市场进口设备并承担10%-50%的进口/转口关税是规避沉没资产延期损失（NPV侵蚀-$200M）的理性财务选择，这导致全球A…", "summary": "截至2026-06-13，本终局卡片合成电网并网队列（~2,600 GW[S1]）及大型电力变压器（LPT）长交期（144-260周[S2]）约束下，美国AI数据中心通过非传统市场（土耳其、巴西、印度、越南）进行变压器“地缘套利”的宏观经济与关税风险。自测算显示，若COD延误3年且Capex已沉没…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施中的电力与电网设备供应瓶颈研究", "summary": "确认：变压器与配电设备缺口已是AI数据中心交付的物理硬约束，非假设风险。2026年美国约12 GW已宣布产能中仅约三分之一在建，约50%–58%管线被实质性延迟；高压变压器交期拉长至4–5年，价格较2019年涨77%–80%。利润率正沿链条向下转移、向卡点集中——形成四层租金地图：第1层GOES材…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 2, "neutral": 0, "value": 74.8, "events": [ { "title": "铜铝等金属成本的结构性上涨对超大规模云厂商的利润率侵蚀极小，由于不对称投资竞争和物理性网路/晶片限制，算力扩张节奏将维持不受影响。", "summary": "截至2026-06-14，精铜和铝价暴涨抬高AI数据中心物理基建CapEx。压力测试显示，当前冲击情景（铜价13,800美元/吨）下四大云厂商年新增折旧增加3.45亿美元，占预估营业利润0.069%（6.9 bps）。即便考虑电价上涨5%（增加电力支出9.46亿美元）的公用事业电价渠道，合并营业利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高能耗GPU集群的大规模部署导致超大规模云厂商的资本支出全面向电力基建倾斜，在设备端过度拥挤的背景下，公用事业运营资产提供了更优的风险收益比。", "summary": "截至2026年中，全球电网物理极限已成为AI扩张的核心瓶颈。超高密度GPU（如GB200 NVL72，120kW+/机柜）推动云厂商2026年Capex升至6600亿-7250亿美元，其中逾60%投向电力与基建。美国数据中心电力需求年底将达41 GW。数据证实，拥挤度极高的电网设备已透支预期，而处…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 2, "neutral": 0, "value": 76.2, "events": [ { "title": "以变压器为核心的电力设备扩产滞后与原材料供给刚性对2027年后的数据中心并网构成了实质性瓶颈。", "summary": "评估表明，变压器等电力设备供应链扩产极其滞后（大型电力变压器交期拉长至2-4年），受特种钢（GOES）和铜的物理供给制约，各厂商新产能直至2027/28年才能释放。这将直接阻碍2027年后的AI数据中心并网（平均并网及设备延迟超5年），验证了物理层瓶颈对算力中心扩张的实质制约。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高息与高昂能源成本导致“硬科技出海”策略发生两极分化，依赖债务融资扩张的二三线半导体与硬件企业无法通过压力测试，将面临流动性与利润率的双重绞杀。", "summary": "截至2026-06-14，本卡对“硬科技出海”策略进行了压力测试。在Brent 95美元与美联储3.50%-3.75%高利率的双重挤压下，AI数据中心面临极高的电力成本（容量电价飙升833%）和电网瓶颈，而晶圆厂（造价150亿-300亿美元）因债务融资成本激增面临流动性枯竭。结论：仅头部现金流充裕…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 4, "neutral": 2, "value": 71.6, "events": [ { "title": "电力配电设备产能扩张与供应链瓶颈：AI算力建设的硬约束 分析师：工业制造分析师（industrials analyst） 立场：support（强化 Research note 02…", "summary": "电力配电设备产能扩张与供应链瓶颈：AI算力建设的硬约束 分析师：工业制造分析师（industrials analyst） 立场：support（强化 Research note 02 的\"电力设备瓶颈是AI算力部署真正天花板\"判断） 工作日期：2026 06 15（Asia/Singapore） …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国电力装备外需足以覆盖行业层面的供应链重构成本，但超额收益将集中在材料、认证和渠道能力最强的头部企业，而非所有出口商。", "summary": "截至2026-06-16，本卡片压力测试全球变压器超级周期下中国电工钢与电力设备出海的利润传导。结论是外需足以支撑行业继续上行，但不能简单抵消所有地缘贸易壁垒；真正受益者是掌握高牌号GOES、具备高压/特高压认证和海外服务能力，并能在必要市场进行最终装配或服务本地化的企业。美国50%钢铝关税和欧盟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球AI数据中心资本开支向中东与东南亚的转移：电网拥堵与贸易壁垒下的分化格局合成分析 报告日期 ：2026 06 15 分析师 ：全球宏观分析师 (Global Macro Anal…", "summary": "全球AI数据中心资本开支向中东与东南亚的转移：电网拥堵与贸易壁垒下的分化格局合成分析 报告日期 ：2026 06 15 分析师 ：全球宏观分析师 (Global Macro Analyst) 立场 ：合成 (Synthesize) — 结合欧美电网与配电设备瓶颈（Research note 02 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026 06 15，通过对五大云厂商6020亿美元Capex计划、AI数据中心3倍金属强度以及中游设备厂商（Eaton, Vertiv, Schneider）创纪录的150亿美…", "summary": "截至2026 06 15，通过对五大云厂商6020亿美元Capex计划、AI数据中心3倍金属强度以及中游设备厂商（Eaton, Vertiv, Schneider）创纪录的150亿美元积压订单的深度穿透，我确认AI与电网基础设施投资已实质性转化为铝、铜及电力设备的物理订单流。这一“订单实物化”验证…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 9, "neutral": 0, "value": 37, "events": [ { "title": "AI电力瓶颈的核心不是全球缺电，而是2026-2030年主要数据中心集群无法足够快地交付本地、确定性、24/7的MW。", "summary": "截至2026-06-16，本卡压力测试认为：能源转型在全球年度电量口径上足以支撑AI数据中心增量，但在主要集群的本地确定性MW交付上明显跟不上。IEA预计数据中心用电从2025年约485 TWh升至2030年约950 TWh，增量约53 GW平均负荷；全球renewables装机到2030年预计新…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网瓶颈促使云巨头表后逃逸，引发美国公用事业板块搁浅资产否决、信用评级降级与分红削减的戴维斯双杀；中国发改委绿电直供政策则构筑了零表后逃逸的阿尔法防线，凸显长江电力与中国核电的绝对避…", "summary": "定量评估AI数据中心并网约束下美国受监管公用事业企业的信用与分红稳定性。在变压器短缺及并网延误下，云巨头转向表后自发电（BTM），导致电网公司面临21.0亿美元Capex监管否决，FFO/Debt降至13.21% [自测算]，跌破14.0%投资级门槛 [S10]，面临下调至BBB及分红被迫削减的戴…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2029年电网设备扩产会缓解但难以消除交付瓶颈，关键稀缺性正在从变压器OEM装配上移到材料、组件、测试和高压系统集成。", "summary": "截至2026-06-16，本卡压力测试第09卡：变压器和关键电网设备扩产真实，但2026-2029年仍难完全跑赢AI数据中心、可再生能源并网、电网替换和电气化带来的MW/MVA交付需求。2028年前大部分大型电力变压器、HVDC电缆、开关设备和上游组件仍处紧缺。真正的瓶颈不只是OEM装配，而是GO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年后AI扩张的关键分母应从名义MW调整为扣除并网时间、网络改造成本和可靠性折减后的可通电可靠MW。", "summary": "截至2026-06-17，本卡对前序“电力溢价”主线做压力测试：核电、长时储能和电网升级不会在2026年后阻断所有AI数据中心，但会在电力紧张地区成为大型液冷训练和高利用率推理园区的第二物理瓶颈。核电主要提升终值和能源期限，难以填补2026-2028年缺口；长时储能更像并网质量增强器，而非稳定发电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 5, "neutral": 0, "value": 62.4, "events": [ { "title": "AI算力向电力设备的资本开支传导扩大了实体“流动性黑洞”，高企的无风险利率与公用事业“成长化”转型持续挤出长久期资产的配置资金。", "summary": "本报告支持AI基础设施分析师的判断，量化了AI电力需求向能源基建的传导。2026年全球数据中心用电预计达565 TWh，中国“十五五”电网投资将达5万亿元，且全球面临30%的变压器结构性缺口。这种大规模实体CapEx形成巨大的“流动性黑洞”，虹吸市场资金，推高资本成本，对医药等长久期板块估值维持持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "西方电力设备巨头的产能释放滞后于AI并网高峰，且由于零部件卡口与合同刚性，中国产能扩张无法有效减轻其高压交付压力。", "summary": "截至2026-06-18，西方巨头（日立能源、西门子能源、GE Vernova）已启动数十亿美元的变压器产能扩张，但核心产能释放集中于2028年以后，与2026-2027年AI数据中心建设高峰存在严重错配。由于地缘壁垒，中国产能无法直接参与美欧大型高压输电变压器建设，且受限于套管短缺、劳动力匮乏和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在AI需求未破坏但芯片拥挤度过高的条件下，组合应把最大风险预算从AI芯片转向电网设备，并用红利公用事业稳定哑铃组合。", "summary": "截至2026-06-18，本卡综合前序结论：AI硬件需求仍强，但拥挤度和估值要求把组合从纯芯片贝塔转向电力交付瓶颈。建议AI防御性哑铃仓位调整为AI芯片15%、先进封装/半导体设备25%、电网设备35%、红利公用事业25%。电网设备因5万亿元电网投资、60 GW数据中心容量和289 TWh 203…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源侧产能扩张只能在本地可交付、可调度且按期并网时缓解 AI 算力约束；在 2026-2028 年，多数新增供给更像中期宽慰而非近期上电解决方案。", "summary": "截至 2026-06-17，本卡压力测试 Research note 04：能源供给增加是必要条件，但不能在电网/变压器/调度延迟严重时自动解除 AI 算力的物理约束。美国 2026 年计划新增 86 GW 公用事业级电源，亚太数据中心用电预计 2024-2030 年从 320 TWh 升至 78…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 3, "neutral": 0, "value": 75.8, "events": [ { "title": "2027-2028年AI数据中心延期将压缩AI算力板块估值久期，并推动资金从纯算力beta向高股息、公用事业和电网设备的现金流/瓶颈资产轮动。", "summary": "截至2026-06-19，本卡合成前序结论：AI数据中心延期不是需求终结，而是把AI硬件估值从“出货即收入”重定价为“通电、并网、费率和利用率证明”。A股策略上，应降低高估值AI算力beta，保留已通电且合同明确的光模块/服务器敞口，同时提高高股息、受监管公用事业、电力运营和电网设备权重。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力规模化面临的下一重物理极限：电力供应、电网接入与散热约束", "summary": "截至2026-06-18，本卡支持卡片05的交接：2026-2027年电网接入容量与高密度液冷是AI数据中心扩容的实质性硬约束，但约束在于可交付性与时序，而非算力绝对上限。IEA基准下全球数据中心用电2024年约415TWh翻倍至2030年约945TWh；美国年新增负荷5-7GW而电网仅交付2-3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心负荷对公用事业是有条件的长期盈利机会：电网和水务资本开支只有在大型负荷成本被合同化、监管准许及时回收时，才会从现金流压力转化为rate base增长。", "summary": "截至2026-06-18，本卡片支持前序“物理基础设施成为AI算力扩张核心瓶颈”的判断，但将盈利传导从数据中心NPV延伸到公用事业rate base机制。电力侧，数据中心用电高增推动输配电和资源充裕度资本开支，若有Dominion式保证金、最低账单和退出费用，资本开支可转化为长期准许收益。水务侧，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 1, "risks": 5, "neutral": 0, "value": 73.2, "events": [ { "title": "电网公用事业资产费率基数扩张对宏观通胀与货币政策的传导效应", "summary": "压力测试卡片01–05的宏观结论。直接渠道被夸大：电力属能源、按构造排除于核心PCE之外，达拉斯联储测算数据中心对总体PCE的拉动2026仅+0.05pp、2030升至+0.13pp[S1]；电力仅占CPI约2.5%[S11]，自测算直接总体贡献约0.10pp。但费率基数是30–40年价格水平锁定…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力稀缺提供结构性底部，但不足以让电网设备和液冷全栈对16.18%的AI硬件Capex压力测试高度免疫。", "summary": "截至2026-06-20，本卡否定“电网设备与液冷对AI硬件Capex下修高度免疫”的强结论。能源侧证据显示，2026年美国计划新增86 GW公用事业级容量，数据中心负荷预测区间很宽，BTM供电正在形成绕行路径，直接液冷渗透仍偏早期。结论不是看空全链条，而是重排敏感度：监管型电网设备韧性最高，BT…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心配电侧存在类似电子布的材料升级逻辑，但结构性溢价属于经过热、介电、防火和高电流认证的系统材料栈，而不是所有电力设备大宗零部件。", "summary": "截至2026-06-20，本卡片支持AI算力功率密度向配电侧关键材料升级传导，但认为它不是电子布价格周期的一比一复制。120 kW级AI机柜把问题从普通供电容量推向高电流、热老化、介电可靠性和灭弧认证。结构性受益集中在高牌号GOES、CTC铜线、工程/芳纶绝缘纸、酯类介质液、母线、高密度中压开关及…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI算力扩张的能源约束不是“总电量不够”，而是2026-2028年稳定MW、电网接入和电价成本无法在关键节点同步交付。", "summary": "截至2026-06-19，本卡对Card-07的“电力优先”结论做压力测试：全球总电量供给大体可通过可再生能源、储能、燃气和核电增发跟上AI数据中心需求，但2026-2028年的真正瓶颈是可交付的稳定MW、电网接入、变压器交付、容量市场重定价和本地成本分摊。电价冲击不是立刻摧毁AI单位经济，而是先…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 5, "neutral": 0, "value": 66.6, "events": [ { "title": "电网设备紧缺不会终止最强Hyperscaler的AI capex，但会把'宣布投资'与'已送电算力'之间的时间差拉大，并迫使选址从效能最优转向电网弹性最优。", "summary": "截至2026-06-21，本卡压力测试电网设备紧缺对AI算力中心扩建的传导。结论是：变压器backlog对设备商是定价权，对Hyperscaler是进度税。Wood Mackenzie显示电力变压器128周、GSU 143周、开关设备44周；相对18-24个月数据中心建设周期，若未提前锁定设备sl…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI capex西迁的A股主线不是泛AI基建，而是“电网优先”：电网设备最确定，网络互联和灵活容量次之，绿电精选，云与IDC需证明价格传导。", "summary": "截至2026-06-20，本卡片综合前序结论：AI基础设施不是宏观总需求对冲项，但电力约束足以重排A股行业赢家。若训练、批量推理和存储任务向西部电力优先集群迁移，最优先配置电网设备，其次是光通信/网络设备与灵活容量，绿电运营商需精选长协和可交付项目，数据中心REITs/运营商分化，云服务商除非能传…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GLP-1 多肽产能扩建对能源及电力基础设施的联动影响", "summary": "支持卡片01“多肽产能拉动材料能源”的跨链判断，但作量级约束。GLP-1资本开支真实且同步：礼来2025资本开支78.4亿美元、2020年以来承诺超550亿；凯莱英固相反应釜18个月内由21000L翻至约44000L；2026年底放量。SPPS单位拉动异常大——1公斤GLP-1需约14000公斤溶…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力公司用'队列改革+大负荷专属费率+共址限制'三件套把AI扩容资本与风险压回开发商，AI数据中心IRR基准下行150–300 bp、走廊分化扩大350–500 bp，受监管公用事业…", "summary": "截至2026-06-20，本卡支持Research note 6判断：AI算力交付的下一道硬约束已从设备短缺切到电网侧——并网队列、大负荷专属费率与共址限制三件套。LBNL显示全美并网队列约2,600 GW、中位等候5年；FERC Order 2023推行集群研究；Ohio PUCO 2025-0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 9, "neutral": 0, "value": 44, "events": [ { "title": "美欧贸易壁垒足以压低中国电力设备出海的利润率和估值倍数，但不足以抵消无源、短缺约束型电网硬件的出口红利。", "summary": "截至2026-06-21，本卡否定“美欧关税和贸易限制足以抵消中国电力设备出海红利”的强判断。政策摩擦真实存在：美国Section 301覆盖部分HS 8504及能源相邻品类，欧盟通过NZIA、FSR和高风险逆变器融资限制提高准入门槛。但美国并网队列、变压器80-210周交期、DOE披露的进口依赖…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心电力瓶颈与核电直供的监管破局：从Talen-AWS僵局到Constellation-微软Eddystone容量豁免的行业拐点", "summary": "分析了2026年6月初美国联邦能源监管委员会（FERC）批准Constellation Energy将760 MW的Eddystone电厂容量互联权（CIRs）转移至Crane清洁能源中心（三哩岛核电站）的豁免裁决。对比了Talen-AWS在此前Susquehanna项目表后直连遭拒后被迫转型为表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核电与可再生能源供应侧的物理瓶颈：规避并网排队能否解决电力规模缺口？", "summary": "压力测试卡片01的监管破局论。FERC的Eddystone容量置换豁免仅是配置抢跑，无法制造净新增可靠电源。需求增量巨大：EPRI测算美国数据中心峰值负荷将从2024年约21.5GW升至2030年中性情景71GW，即约+49.5GW[自测算]。供给侧无法匹配：核电重启/扩容到2029年仅约+5GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张与电网/能源供应瓶颈的匹配度：评估AI数据中心负荷激增对电力基础设施可靠性及能源成本抬升的压力测试", "summary": "压力测试确认：电网建设进度、配电设备容量与能源价格波动已成超大规模AI集群扩建的核心掣肘，约束力强于资本、渐强于GPU，但呈非对称。美国是\"时间+价格\"约束——并网排队4–7年（全国均约5年[S1][S3]）、PJM容量价创纪录329.17美元/兆瓦·日约10倍跃升、数据中心占容量成本40%[S1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 0, "value": 87.4, "events": [ { "title": "能源价格冲击对通胀预期的传导风险", "summary": "截至2026-06-22，本卡以能源视角支持Card-01「暂停+更久维持高位选择权」，但修正其传导机制。需区分两类冲击：(1)地缘油价飙升已回落——WTI从约$109崩跌约29.6%至$72.83，美伊重开霍尔木兹后布伦特回到约$80，且油价对核心通胀传导小而慢(峰值约0.12pp)；(2)真正…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张不会被能源安全溢价终止，但会被“可交付稳定兆瓦”重新定价，全球集群部署将向电力充足、并网确定、能效合规且成本分摊透明的地区集中。", "summary": "截至2026-06-22，本卡对能源安全CapEx如何约束AI算力基础设施作压力测试。结论是：电价上升重要，但更核心的瓶颈是稳定兆瓦能否按GPU商业化窗口交付。IEA、LBNL、EPRI与NERC数据共同显示，AI数据中心需求已成为局部电网级变量，尤其在美国、欧洲、中国与东南亚。能源安全溢价会推高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 4, "neutral": 1, "value": 79.8, "events": [ { "title": "电网输变电容量与并网审批周期（而非总体发电量）已成为AI算力部署的终极物理刚性瓶颈，并已实质性挤压IT硬件采购预算。", "summary": "本报告对AI算力扩张的电网与能源瓶颈进行压力测试。分析表明，虽然电力设备订单充足，但美国并网队列平均达5.5年，且面临FERC 2026年6月大型负荷监管及弗吉尼亚GS-5等防止成本转嫁政策。变压器等重电面临GOES原材料短缺，Tier-1厂商积压至2029-2030年。表后核电直供监管被封死，燃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "配电网滞后与变压器短缺连同HBM/CoWoS确实构成AI capex的物理硬约束——对边际开发商是硬墙，对优先级买家是ROIC压缩，从物理层面支持AI-capex去杠杆而非2026崩…", "summary": "截至2026-06-23，本卡从公用事业侧确认：配电网扩容滞后与变压器短缺是AI capex物理钳制的第二片钳口，与HBM/CoWoS硅片分配叠加。变压器交期达128–144周乃至4年，电力变压器缺口30%、配电10%，GOES自2020年翻倍、铜+50%；LBNL并网队列约2,060 GW、中位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网队列与基础设施扩容压力评估", "summary": "本卡支持卡片1–2：约束已从Capex/GPU转向已通电的MW，并将瓶颈量化为双向并网堵塞。发电侧：美截至2025年底约2,060GW排队（约现有装机2倍），408GW已签约未通电，申请到商运中位时长翻倍至>4年。负荷侧：ERCOT大负荷队列达226GW（近4倍于2024年底63GW，77%为数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支削减15%将导致全球约1.19 GW的数据中心部署延迟，阶段性缓解电力变压器供需瓶颈和电网并网压力，同时导致独立发电厂与电力设备标的面临估值去化风险。", "summary": "截至2026-06-23，本卡评估了AI资本开支下修对数据中心、电网组件及能源供应链的传导效应。模型表明，15%的AI资本开支削减（-661.5亿美元）将导致全球约1.19 GW的物理数据中心IT容量建设延迟。这将有效缓和大型电力变压器（交期24-48个月，价格较2019年上涨77-95%）和输电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 5, "neutral": 0, "value": 72.2, "events": [ { "title": "AI算力投资需求规模与电网/能源供应缺口的动态匹配分析", "summary": "截至2026-06-25，本卡量化AI算力需求通道与电网/能源物理供应瓶颈的动态匹配。需求侧：4家超大规模厂商2026年capex约7250亿美元(+77%)，美国数据中心峰值负荷2024约21-22GW、EPRI中性情景2030升至71GW。供应侧：并网积压约2600GW(中位约5年)、大变压器…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "评估欧美电网升级需求与贸易壁垒对中国电力设备出海链的传导效应", "summary": "压力测试 Research note 4 命题：A股估值重配是否真建立在可持续高毛利出海订单上。结论为部分支持但有结构性保留。需求拉动真实且被低估——美国公用事业2026-30资本开支近1.3万亿美元 [S1]、超大规模厂商2026资本开支>6750亿美元/+60% [S2]、全球变压器短缺 [S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与开关设备长周期会迫使AI数据中心按可送达电力重新选址，并把超大规模资本开支从GPU/土建扩展到电力交付栈。", "summary": "截至2026-06-25，本卡片压力测试卡片06：AI算力选址确实会从“便宜电”转向“可信time-to-power”，但不是简单回到传统成熟枢纽。变压器、开关设备、变电站与并网改造成为已带电容量的关键约束；低电价若没有设备槽位和送电排期，只是搁浅期权。超大规模厂商的资本开支将更多前置到电气设备预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源波动一方面保护通胀对冲组合，另一方面会通过电价、容量市场和电网延迟挤压AI算力经济性，因此AI基础设施应从“GPU稀缺交易”重估为“电力与资本配置交易”。", "summary": "截至2026-06-25，本卡对能源对冲逻辑做压力测试：AI数据中心需求真实存在，但美国数据中心用电可能从2023年的176 TWh升至2028年的325-580 TWh，电力已从普通运营项变成增长闸门。PJM和Dominion案例显示，局部电网瓶颈会通过容量费用、并网延迟和交付电价压缩算力利润率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 5, "neutral": 0, "value": 73.6, "events": [ { "title": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论", "summary": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论 Feasibility of Bring Your Own Power (BYOP) and Gas PPA for AI Data Centers 在小堆核电(SMR)2030年前难以落地、碳约束拉平公耗曲线的现实背景下，国内与海外AI数据…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论", "summary": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论 在小堆核电(SMR)2030年前难以落地、碳约束拉平公耗曲线的现实背景下，国内与海外AI数据中心采用天然气PPA与企业自备电力(BYOP)的技术路径、套利模型及项目收益率约束是什么？如何评估其对公用事业和电网高端设备板块的溢价传导？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供给不是杀死AI硬件主题的理由，但足以成为压力情景下把配置上限从6%降至4%的结构性主因。", "summary": "截至2026-06-25，本卡支持将能源供给约束作为AI硬件仓位从6%压向4%的结构性主因。IEA、Berkeley Lab、PJM、Monitoring Analytics、EIA和Wood Mackenzie数据共同显示，AI数据中心用电需求仍强，但可通电兆瓦、容量成本、变压器/开关设备交付和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力的化石兜底在全球排放中仍是小比例，但在2030年前足以重定价区域气电、LNG灵活性、燃机供应和企业脱碳承诺可信度。", "summary": "截至2026-06-26，本卡片支持上张卡片的电力瓶颈判断，但将结论压力测试为能源链问题：AI算力若在2030年前因核电和并网滞后转向天然气与煤炭，全球层面不会击穿化石燃料供应，却会显著收紧美国区域天然气、LNG灵活性、燃机交付、电网设备和煤电退役节奏。压力情景估算，2030年AI数据中心化石兜底…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 4, "neutral": 0, "value": 81, "events": [ { "title": "随着全球及中国变电设备产能在2027-2029年逐步释放，AI数据中心的电力设备瓶颈将从“无法获取设备”的二进制物理硬约束，降为“支付溢价与关税以获取交期”的资本支出与利润率成本约束。", "summary": "本报告对2027-2029年AI电力设备瓶颈的缓解速度进行了深入分析。结论表明，随着西方重电巨头（如日立能源、GE Vernova）扩产产能释放，以及以特变电工、思源电气、金盘科技为代表的中国高效率设备出口，2027-2029年电力设备将从物理上的“硬约束”转变为“成本/经济约束”。由于取向电工钢…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力冲击本身是Fed应看穿的相对价格变化，但其持续、受供给约束且贴近核心服务成本的二阶传导，会让核心通胀在边际上更粘，并支持Fed更久维持限制性政策。", "summary": "截至2026-06-27，本卡综合前序结论：AI电力不是直接核心通胀项目，因为电力在2026年5月CPI-U权重仅2.495%，且core PCE剔除能源；但它可能通过商业电价、云服务、通信、医疗、设施与租金相关成本向60.158%权重的services less energy services传…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网基础设施与能源转型对AI算力扩张的潜在约束", "summary": "压力测试部分支持并升级卡03：电力/电网瓶颈是“节流而非断路”，但节流比其暗示更紧更久。在2026–2028视野内，电气基础设施已是排在芯片之前的硬约束——美国电力变压器交期128周、GSU 144周、大单3–4年，价格较2019+77%、达2022前4–6倍；GOES由Cleveland-Cli…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 8, "neutral": 0, "value": 66.4, "events": [ { "title": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电力’，并驱动策略从算力规模向算法…", "summary": "截至2026-06-28，本卡片通过对电网与变压器交期的压力测试，确认电网并网（5-7年交期）和变压器交付（2-4年交期）构成了AI算力扩张的终极物理边界。随着GB300（约135kW）及Rubin Kyber（约600kW）对800VDC和固态变压器（SST）的高密度配电需求增加，目前仅有不足8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年全球AI数据中心电网瓶颈与“自主供电”战略的崛起", "summary": "2026年全球超大规模厂商的AI算力基建受限于电网瓶颈与关键电力设备（变压器交期达5年）短缺。这迫使厂商将部分资本支出（预计2026年总支出达7250亿美元）从纯算力（GPU）转向自主供电（BYOP）和表后（BTM）能源方案。预计该转移将延迟算力上线速度，并推高长期持有成本。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "summary": "截至2026-06-28，本卡对“AI负荷增长能否支撑公用事业高分红”作压力测试：结论是精选组合层面可以，板块整体不能机械外推。AI数据中心用电增长真实，但也同步拉动电网、发电和储能资本开支。水电、核电、容量电价支持的火电及低杠杆综合电力运营商仍可维持分红；风光开发商受消纳、电价和项目资本开支制约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月点阵图不是简单过时，而是可能“对得不是理由”：能源理由过时，AI/电网与折旧通道支持更高更久，但不足以单独证明立即加息。", "summary": "截至2026-06-28，本卡综合认为：6月点阵图并非被最初的油价冲击证明，因布伦特已回落至约US$72/bbl；但AI/电网导致的结构性电价稀缺、核心PCE 3.4%、名义需求仍强以及AI-capex折旧悬挂，使“更高更久”仍有宏观依据。该通道不足以单独证明立即加息，因为电力CPI权重仅2.48…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 3, "neutral": 0, "value": 89.8, "events": [ { "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）…", "summary": "截至2026-06-29，变压器交期延长和材料瓶颈导致美国电网资本开支通胀率达32.72%。结合2027/2028年度PJM容量电费暴涨11.5倍至$333.44/MW-day，将拉动美国零售电价上涨4.18%，直接和间接推高全国CPI 0.124个百分点。高粘性通胀与AI基建庞大资金需求将上移中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-075017003b/", "networkHref": "/investor/research-network/#thesis-075017003b" }, { "id": "thesis-42faeae623", "source": "tracked_thesis", "title": "CPO、光互联与铜连接决定 AI 集群内部租金迁移", "titleEn": "CPO, optical interconnect, and copper links define rent migration inside AI clusters", "lane": "CPO/光模块/高速互联", "laneEn": "CPO, optical modules, and high-speed interconnect", "laneHref": "/investor/lanes/cpo-optical-modules-and-high-speed-interconnect/", "investableFocus": "CPO/光模块/高速互联：AVGO, MRVL, ANET, COHR, LITE, 300308.SZ, 300394.SZ, 688498.SH", "stockUniverse": [ { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MRVL", "name": "Marvell", "market": "US", "role": "高速互联芯片", "roleEn": "高速互联芯片", "bucket": "core", "href": "/investor/stocks/MRVL/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "COHR", "name": "Coherent", "market": "US", "role": "光器件与材料", "roleEn": "光器件与材料", "bucket": "core", "href": "/investor/stocks/COHR/" }, { "ticker": "LITE", "name": "Lumentum", "market": "US", "role": "光器件", "roleEn": "光器件", "bucket": "core", "href": "/investor/stocks/LITE/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "300394.SZ", "name": "天孚通信", "market": "A-share", "role": "光器件", "roleEn": "光器件", "bucket": "core", "href": "/investor/stocks/300394.SZ/" }, { "ticker": "688498.SH", "name": "源杰科技", "market": "A-share", "role": "光芯片", "roleEn": "光芯片", "bucket": "core", "href": "/investor/stocks/688498.SH/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" } ], "stockTickers": [ "AVGO", "MRVL", "ANET", "COHR", "LITE", "300308.SZ", "300394.SZ", "688498.SH", "NVDA", "TSM", "MU", "VRT", "SMH", "APO" ], "coreView": "赛道：CPO/光模块/高速互联。候选观察标的：AVGO, MRVL, ANET, COHR, LITE, 300308.SZ, 300394.SZ, 688498.SH。当前共有 40 条证据、1 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 78, "eventCount": 160, "firstSeen": "2026-04-24", "lastSeen": "2026-06-29", "supports": 9, "risks": 135, "neutral": 16, "analystCount": 12, "sourceTypeCounts": { "研究线程": 3, "研究笔记": 149, "分析协作": 6, "研究主题": 2 }, "topTerms": [ "AI 算力", "AI 基础设施", "云厂商", "GOES", "HBM", "CPO", "先进封装", "GPU", "ASIC", "中际旭创", "AIDC", "ADC" ], "whatChanged": [ "2026-06-29: 物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进封装国产替代材料集中。", "2026-06-29: 国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加速撤向水电红利及电网设备出海龙头，陡峭化了向高红利风格轮动的斜率，配置应重构为哑铃型…", "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。", "2026-06-28: GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "2026-06-28: 双轨AI网络可以落地，但硬件约束从电力单点转向电力、CoWoS/HBM、交换ASIC和光互联的同步供给，只有能跨堆栈预锁定产能的运营商能率先规模化。" ], "practical": { "sectorLane": "CPO/光模块/高速互联", "sectorLaneEn": "CPO, optical modules, and high-speed interconnect", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MRVL", "name": "Marvell", "market": "US", "role": "高速互联芯片", "roleEn": "高速互联芯片", "bucket": "core", "href": "/investor/stocks/MRVL/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "COHR", "name": "Coherent", "market": "US", "role": "光器件与材料", "roleEn": "光器件与材料", "bucket": "core", "href": "/investor/stocks/COHR/" }, { "ticker": "LITE", "name": "Lumentum", "market": "US", "role": "光器件", "roleEn": "光器件", "bucket": "core", "href": "/investor/stocks/LITE/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "300394.SZ", "name": "天孚通信", "market": "A-share", "role": "光器件", "roleEn": "光器件", "bucket": "core", "href": "/investor/stocks/300394.SZ/" }, { "ticker": "688498.SH", "name": "源杰科技", "market": "A-share", "role": "光芯片", "roleEn": "光芯片", "bucket": "core", "href": "/investor/stocks/688498.SH/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" } ], "stockTickers": [ "AVGO", "MRVL", "ANET", "COHR", "LITE", "300308.SZ", "300394.SZ", "688498.SH", "NVDA", "TSM", "MU", "VRT", "SMH", "APO" ], "primaryStocks": [ { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MRVL", "name": "Marvell", "market": "US", "role": "高速互联芯片", "roleEn": "高速互联芯片", "bucket": "core", "href": "/investor/stocks/MRVL/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "COHR", "name": "Coherent", "market": "US", "role": "光器件与材料", "roleEn": "光器件与材料", "bucket": "core", "href": "/investor/stocks/COHR/" }, { "ticker": "LITE", "name": "Lumentum", "market": "US", "role": "光器件", "roleEn": "光器件", "bucket": "core", "href": "/investor/stocks/LITE/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" } ], "instrumentTags": [ "AVGO", "Broadcom", "MRVL", "Marvell", "ANET", "Arista Networks", "COHR", "Coherent", "LITE", "Lumentum", "300308.SZ", "中际旭创" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 116 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 12, "recentEvents": 116, "supportRatio": 0.06, "riskRatio": 0.84, "conflictScore": 94.7, "evidenceDepth": 100 }, "score": 70.8 }, "events": [ { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.25, "matchedTerms": [ "HBM", "CPO", "先进封装", "GPU", "光模块", "infrastructure", "capex" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性TAM上限，营收放缓促使估值压缩。相反，全球CPO光模块（中际…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "HBM", "CPO", "GPU", "中际旭创", "光模块" ], "href": "" }, { "date": "2026-06-29", "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-560亿美元收入，其25%低毛利率根本无法对冲高利润半导体供应链…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "HBM", "先进封装", "GPU", "capex" ], "href": "" }, { "date": "2026-06-28", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,600 亿）以经营现金流支撑，资金基础/地域/标的均不同于承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26, "matchedTerms": [ "AI 基础设施", "云厂商", "GPU", "infrastructure", "cloud" ], "href": "" }, { "date": "2026-06-28", "title": "双轨AI网络可以落地，但硬件约束从电力单点转向电力、CoWoS/HBM、交换ASIC和光互联的同步供给，只有能跨堆栈预锁定产能的运营商能率先规模化。", "summary": "截至2026-06-28，本卡对Research note 05双轨网络做压力测试并给出有条件支持：CoWoS/HBM足以支撑头部训练集群，但2026-2027仍是分配瓶颈；硅光/CPO可降低核心fabric功耗和延迟，但都市推理边缘仍会混用可插拔800G/1.6T光模块与软件局部性。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "HBM", "CPO", "ASIC", "光模块" ], "href": "" }, { "date": "2026-06-28", "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM合约价Q2环比+58–63%)、以及电网电力(美国并网排队约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "HBM", "先进封装", "GPU", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "在2026-06-27的权威工作日背景下，国产HBM良率骤降逼迫国内算力网络化，而全球封装与电力阻尼拉长了云厂商CapEx变现周期，这使得制程免疫且订单排至2028年的可插拔光模块出口龙头成…", "summary": "截至2026-06-27，本报告评估了HBM良率危机与先进封装瓶颈对算力集群技术演进的影响。国产HBM3封装良率降至2.66%，倒逼国内算力集群转向网络密集的Scale-out拓扑，推高了光模块需求弹性。全球云厂商2026年CapEx直逼7250亿美元，但受制于CoWoS晶圆先进封装冲突与吉瓦级电网接入瓶颈，面临6-12个月的算力交付与收入确认错配。CPO…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.25, "matchedTerms": [ "云厂商", "HBM", "CPO", "先进封装", "光模块", "infrastructure", "capex" ], "href": "" }, { "date": "2026-06-27", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交期[S6][S7]限制的是通电而非下单。CoWoS售罄、交期…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.25, "matchedTerms": [ "AI 算力", "HBM", "先进封装", "GPU", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "结构性再贷款提供的低成本流动性无法有效突破高端设备禁运的技术壁垒，制裁压力下国产HBM良率降至2.66%且大模型训练效率折损25%-30%，需通过在科技板块内超配CPO光模块等微观对冲平抑风…", "summary": "本报告对结构性再贷款支持下半导体“硬科技”在外部制裁压力下的对冲效果进行压力测试。分析显示，虽有5000亿再贷款及1.5%贴息将企业融资成本降至0.25%-0.50%[S1]，但流动性无法突破技术封锁。在先进TBDB与CMP设备禁运下，国产8-Hi HBM3封装良率由16.86%骤降至2.66%[自测算]，先进封装经济利用率降至35%-45%[自测算]，大…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "HBM", "CPO", "先进封装", "光模块" ], "href": "" }, { "date": "2026-06-27", "title": "全球AI基础设施资本开支对中国产业链的传导机制与订单确定性", "summary": "本卡支持卡片05遗留命题并定论：约7250亿美元、同比+77%的全球超大厂资本开支（约75%即约4500亿为AI专项）规模巨大、外部出资，已转化为中国主导且出口管制并不约束的那一层链条（光互连/连接/散热/PCB）的跨多季度订单簿与已兑现盈利。关键在于出口管制是分层的：重压落在GPU/算力层（华为昇腾、SMIC受限）和上游光芯片层（EML/DSP/InP国…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "GPU", "中际旭创", "光模块", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "AI 数据中心 capex 从 GPU 向配电、UPS、母线槽与液冷供应商迁移的跨资产定价", "summary": "本卡综合前九卡，给出可交易的跨资产框架：AI 算力卡点已从 GPU 下移到可交付、已通电、已散热的兆瓦。正确定价不是单边押注 AI，而是 AI 板块内的相对价值轮动——做多以在手订单交付确认收入的电气/热管理供应商(VRT/ETN/SU)与电网设备(GEV/PWR)，做多铜作为材料成本推动与内部毛利对冲，并以针对资本开支最重云厂商的相对低配/价差期权对冲 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22, "matchedTerms": [ "AI 算力", "云厂商", "GPU", "capex" ], "href": "" }, { "date": "2026-06-26", "title": "1.6T光模块交付延迟对AI算力集群及超大规模云厂商Capex部署进度的影响", "summary": "支持卡片01/02的研究线并加以边界化延伸：200G EML 的 HTOL 风险确会让 1.6T GA 滑向 2027H1，但这是一道带宽前沿与单位经济性约束，而非整体 Capex 停摆。GB300 NVL72 的 scale-up 走铜缆 NVLink、scale-out 走 ConnectX-8 每 GPU 800G[S7]，约 6 万台机柜[S8]照…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.25, "matchedTerms": [ "云厂商", "CPO", "GPU", "光模块", "infrastructure", "capex" ], "href": "" }, { "date": "2026-06-25", "title": "虽然物理电网排队与供应链原材料价格冲击显著抬高了局部建设阻力，但云厂商通过绿电PPA、高密度液冷及转向定制ASIC等结构性调整，能够完全消纳成本溢价，不构成算力扩张的绝对瓶颈。", "summary": "本卡对电网并网延迟、变气压原材料（铜/GOES）价格波动和本地化GPU封装良率冲击进行了量化压力测试。结果表明，虽然并网延迟和原材料成本增加会显著压缩项目NPV（5年延迟将使NPV降低79.99%），但云厂商可以通过迁移西部枢纽自建绿电、液冷优化和系统级混合算力架构，消纳国产芯片和能源价格溢价，并在ASIC转型中推动光模块需求逆势增长。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34, "matchedTerms": [ "云厂商", "GOES", "GPU", "ASIC", "光模块", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元在手订单）；第二档择优超配高端光互联（1.6T之年、龙头锁单…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "云厂商", "HBM", "先进封装", "GPU", "ASIC" ], "href": "" }, { "date": "2026-06-25", "title": "电网缓释后，AI集群上线风险不是消失，而是转移到CoWoS/HBM、液冷验收与网络芯片的串行同步，CPO下行收窄但不等于收入曲线完全恢复。", "summary": "截至2026-06-26，本卡压力测试Research note 7：电网侧缓释可降低CPO最坏收入下修，但不能消除AI集群的串行硬件瓶颈。关键约束从单一接电问题转向先进封装/HBM、液冷机柜验收、交换ASIC/网络集成的节奏匹配。CoWoS缺口可能由约20%降至2026年底约10%，但10%仍足以配给高端加速器；液冷从少数高密度机房跨越到100-120 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "HBM", "CPO", "先进封装", "ASIC", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "做多迁移的稀缺租金应以电网硬件和材料为锚、以先进封装和定制ASIC为弹性，同时用能源空头、商用GPU相对空头和暗硅期权保护组合。", "summary": "截至2026-06-25，本卡综合前序链条：霍尔木兹能源溢价正在消退，稀缺租金迁移到铜/GOES、电网设备、先进封装和定制ASIC。建议用100.0名义多头配置15.0%材料、30.0%电网硬件、20.0%先进封装、15.0%定制ASIC/AI网络、10.0%电力接入和10.0%流动性，同时用35.0名义对冲Brent、商用GPU份额侵蚀和暗硅估值回调。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 26.5, "matchedTerms": [ "GOES", "先进封装", "GPU", "ASIC", "rent" ], "href": "" } ], "timeline": [ { "date": "2026-04-24", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "NVIDIA Rubin 架构前夕的 AI 算力市场调研", "summary": "NVIDIA Rubin 架构前夕的 AI 算力市场调研 随着 Rubin 架构发布临近，超大规模云厂商在自研 ASIC 与采购 NVIDIA 芯片间的动态博弈及对 HBM4 产业链的带动作用？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-04-25", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "2026年超大规模云厂商AI资本开支冲刺与电力—先进封装双瓶颈", "summary": "2026年超大规模云厂商AI资本开支冲刺与电力—先进封装双瓶颈 在Q1 2026业绩季释放的约7,000亿美元四大超大规模云厂商资本开支指引下，约束是否已从GPU供给切换至电力与CoWoS/HBM封装产能？这对TMT板块2026下半年的α定价意味着什么？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 3, "neutral": 0, "value": 26.8, "events": [ { "title": "2026下半年AI资本开支拐点辩论：消化期还是再加速？", "summary": "TMT分析师判定，2026下半年AI资本开支并未进入消化期，而是进入结构轮动期：总量持平至上行，边际增量美元从训练优化的商用GPU集群转向推理fleet、超大规模云厂商自研芯片、以及电力与场地准备。空头叙事在需求侧（推理token volume、agentic渗透、多模态）证据不足，但在前沿训练边…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在DeepSeek/H20出口管制双重冲击下，光模块、DDR5接口芯片和成熟国产AI芯片厂商能转化为2026年真实盈利，而以寒武纪为代表的早期GPU设计公司仍是叙事溢价显著高于盈利兑…", "summary": "对中国AI/半导体供应链做收入兑现压力测试：光模块（中际旭创300308、天孚通信300394）、DDR5接口芯片（澜起科技688008）和有成熟收入基础的国产AI芯片（海光信息688041）是2026年最可信的盈利转化标的；先进封装（长电科技600584、通富微电002156）路径较长；寒武纪（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电子取代 HBM 成为 AI 链条上未来 12 个月最被讨论的瓶颈：1 GW 边际 AI 园区的电费在 2027 年从 $322M/年跳到 $926M/年，足以让 hyperscal…", "summary": "对 Research note 03 的 $1T capex / $7.5–10M/MW 营收密度救援叙事进行压力测试。结论：电力（不是 GPU）成为 2026 下半年起的硬约束；hyperscaler 在 2026–2027 锁定的边际 MW 全成本将达 $95–140/MWh，相对 2024 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 0, "neutral": 0, "value": 54.2, "events": [ { "title": "5/6 若发生美股 AI 估值重置，国产算力不是一条线：海光信息更能靠自主可控订单防守，光模块/AI PCB 的订单更硬但也更受 U.S. capex 估值冲击，寒武纪仍需复购和客户…", "summary": "本卡将 13 个国产算力龙头的 2026E-2027E AI 算力收入拆成 U.S. hyperscaler capex、中国互联网/云 capex、政府/SOE 自主可控采购和渠道补库存四类。结论是：中际旭创、新易盛、天孚通信、沪电股份、胜宏科技和工业富联订单证明最强，但高度暴露于 U.S. A…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "[auto-handoff] 量子-AI 混合架构调研", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 6db6ff09-a632-4e7d-ad75-41e442c2e417 结果摘要: Wrote handoff_response.md (10,069 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 3, "neutral": 1, "value": 34, "events": [ { "title": "算力期货压缩的是商用GPU层租金，先进封装与高带宽互联的物理瓶颈与寡头结构足以将其非商品化毛利结构维持至2027。", "summary": "支持card-01框架并加以收紧：算力期货标准化的是GPU-小时这一输出单位，并未商品化决定谁能以最低成本清算的物理输入。CoWoS先进封装、HBM4、混合键合设备、scale-up交换机ASIC、PCIe/CXL retimer、1.6T/CPO光模块均具18–36个月结构性导入周期，寡头格局清…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "巨头 CapEx 压力确实在 A 股 AI 算力硬件的集成商层引发毛利率压缩，Agent 叙事无法救回集成商利润池；但加速器与光模块/PCB 子板块基本面仍在改善，Card-01 的…", "summary": "压力测试结果部分支持 Card-01，但需做结构性修正：A 股 AI 算力硬件并非铁板一块。服务器/集成商层（浪潮、紫光等）GM 已从 8–11% 压缩至 5–7%，HBM3e 涨价 + 巨头议价 + 液冷一次性投入共同施压；情景测算显示即便 Agent 推理乐观情景下集成商 GM 也仅能持平，因…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "材技一体化对fabless是选择性降权——AI加速器与EDA/IP溢价稳健，射频/功率/模拟/成熟节点向IDM收敛，资金应在设计层内向两个可防御极轮动。", "summary": "材技一体化情景下，fabless估值溢价并非整体崩塌而是结构性分化。AI加速器龙头（NVIDIA、AMD MI、Broadcom ASIC）凭借云厂商需求无弹性和CoWoS/HBM优先配额，能够将晶圆和封装涨价完整转嫁，溢价稳健；EDA/IP分成（Cadence、Synopsys、Arm）现金流与…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2028 年变压器交付周期会拖慢 2028 年起的新增 AI greenfield 园区并重新定价资产栈，但通过预订、表后自电、变电站复用和能效提升，并不是 2028 年之前 AI …", "summary": "截至 2026-05-14，120–210 周的大型变压器交付周期对 2026 下半年至 2028 年的 AI 新建园区构成现实约束，但不是 2028 年之前整体 AI 算力的硬顶。超大规模云厂商通过 2022–2024 年完成的变压器预订、表后燃机/SMR/PPA、退煤站点变电站复用、机架能效（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 4, "neutral": 0, "value": 26.4, "events": [ { "title": "35% AI/成长仓位整体成立，但应集中到先进封装、光互联、国产逻辑代工、AI ASIC等上游AI基础设施层，避开消费电子相邻半导体与商品存储，并对光模块设减持纪律以应对2H-26产…", "summary": "截至2026-05-15，35% AI/成长仓位通过Capex压力测试但需收紧结构。中国云厂商（BABA、Tencent、Baidu、ByteDance、三大电信SOE）2026年合计AI Capex预计同比+45%–70%（合计~RMB 700–830 bn），由AI加速卡、光互联、液冷供电驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "M2M流量爆发的边际成本不会均匀挤压SaaS——它会让薄壳代理编排厂商利润坍缩、让有自有数据的垂直SaaS继续吃到按量计费红利，并把被争夺的利润集中流向芯片、封装与光模块基础设施层。", "summary": "截至2026-05-16，本卡对card-07的SaaS基本面Beta判断做压力测试。结论是有条件成立但显著分化：薄壳代理编排SaaS将因无法转嫁的token成本，在未来12–18个月承受600–1,200 bp毛利压缩；而垂直SaaS、数据合规和API适配器（card-06超配名单）边际成本以存…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GXP向CSA的制度松绑只是把交付瓶颈下移到芯片配额与高压电力装备交期，物理瓶颈已替代制度瓶颈成为AIDC交付的新主导变量。", "summary": "CSA转型确实把特区内AIDC交付压缩到9–12个月，但约束随即从制度排队迁移到物理输入：>500 MVA变压器交期110–140周、CoWoS-L先进封装2026年75–80%被NVDA/AMD/Google预订、HBM3E为价格弹性最低的卡点。'交付悖论'未消失，只是从监管时间转为BOM时间。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股半导体重估的 beta 已基本被定价，下一段必须靠后道设备、量检、材料、车规功率与先进封装上游材料的盈利与订单兑现，回避换手/融资/解禁三重拥挤的设计、代工与映射封装。", "summary": "截至 2026-05-15，A股半导体经过 4-5 月重估后，国产 GPU/ASIC 概念、HBM/先进封装映射股、大市值晶圆代工四条线已拥挤：5 日换手率分位 81-92%，半导体两融占自由流通市值比 ~4%（88 分位），主要芯片 ETF 4 周连续净申购 +18%，且 6-8 月解禁集中（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "AI算力链基本面与订单流：对减仓建议的支持性验证 日期： 2026 05 17 分析师： TMT行业分析师 [Primary] 对Research note 01立场： Suppor…", "summary": "AI算力链基本面与订单流：对减仓建议的支持性验证 日期： 2026 05 17 分析师： TMT行业分析师 [Primary] 对Research note 01立场： Support（支持） 结论： 基本面边际开始出现裂缝——印证因子拥挤度去风险建议 1. 摘要 Research note 01…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 3, "neutral": 0, "value": 35.2, "events": [ { "title": "AI 资本开支和利润率足以支撑已验证算力基础设施链条的估值修复，但不足以支撑泛 AI 成长股普涨。", "summary": "2026-05-17：本卡对 Research note 7 的“利率波动见顶后配置高现金流长久期成长”做压力测试，结论为有条件支持。AI 算力链条的盈利确定性主要集中在硬件和基础设施层：NVIDIA、Broadcom、TSMC、工业富联、中际旭创等已体现收入、利润率和经营现金流共振；但 hype…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Nvidia 直供链上的 A 股份额东移是真实但窄的（5–8 家硬名单），归因是合同制造而非美方约束；CPO 光引擎核心仍在美方，真正的份额弹性在独立订单簿的国产替代层。", "summary": "在 GB300 NVL72 与 Rubin CPO 架构下，A 股供应链份额东移的真实增量集中在 800G/1.6T 光模块 OEM（中际旭创、新易盛、天孚）、液冷 CDU 与冷板（英维克、高澜、工业富联整柜）、超低损耗高速 PCB（沪电、胜宏）三个环节，硬名单仅 5–8 家，远小于市场涨停所定价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件与算力基础设施具备足够的选择性资本确定性，但应买入瓶颈环节和已验证国产替代，而不是泛AI主题贝塔。", "summary": "截至2026-05-18，本卡片支持从大金融防御性收缩转向盈利可见成长。AI硬件具备选择性资本确定性：NVIDIA、TSMC、SK hynix及800G/1.6T光模块已体现高毛利兑现，云厂商2026资本开支提供需求锚。国产化替代真实存在，但确定性集中在服务器、光模块、热管理、电源、PCB和已商业…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 42.6, "events": [ { "title": "量化压力测试：A股算力链融资盘平仓与 VaR 破缺风险报告 日期： 2026 05 19 分析师： chief risk (首席风控官) 立场： 压力测试（负向风险预警） 1. 执行…", "summary": "量化压力测试：A股算力链融资盘平仓与 VaR 破缺风险报告 日期： 2026 05 19 分析师： chief risk (首席风控官) 立场： 压力测试（负向风险预警） 1. 执行摘要 截至 2026 年 5 月 19 日，A 股 AI 算力链（特别是光模块/CPO 与 PCB 板块）已进入“杠…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海外 Hyperscaler FY26 资本开支稳固、卡片 01 触发器未触发，四只『千元股』基本面逻辑未破，绑定风险是情绪侧而非需求侧，但内部分化要求先减光模块、保留 ASIC。", "summary": "美股四大 Hyperscaler 1Q26 业绩（2026 年 4 月末／5 月初）合计 FY26 资本开支约 US$3610–3710 亿，同比 +28%，无一家下修，Meta 还小幅上调 —— 卡片 01 列出的『Hyperscaler 资本开支下修』触发器截至 2026-05-18 未触发。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 1, "neutral": 1, "value": 61, "events": [ { "title": "若 2026-06-11 验证 Waymo 单位经济，AI 基础设施不应被卖出，而应从拥挤算力前端轮动到电源液冷、networking 邻近环节与小仓位机器人硬件 beta。", "summary": "截至2026-05-19，本卡支持 6 月 11 日后从 AV 扩散到 AI hardtech，但只在 Waymo 经济性披露有用时执行。确定性最高的是电源/液冷/机柜、定制芯片与 networking、HBM/先进封装；估值弹性最好的是电源热管理与机器人端侧硬件。HBM、GPU、A 股光模块和主…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "消费电子'技术性弱复苏'+AI 手机溢价被压在旗舰段，触发的是 A 股'叙事溢价 → 确定性溢价'的哑铃化再定价，而非全面价值回归或单边 AI 基础设施集中。", "summary": "工作日期 2026-05-19。综合 card-03（AI 基础设施仍是窄口径主线）、card-04（消费电子 2026H2 技术性弱复苏）与 card-05（以旧换新净撬动仅约 1.25x、AI 手机溢价仅 ≥6,000 元旗舰段）三条结论后，本卡的策略级判断是：A 股既不会单边全面价值回归，也…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "截至 2026-05-19，AI 基础设施仍是 TMT 领涨主线，但可投资范围应收窄到有真实 usage、backlog、产能瓶颈和现金流承受力的环节。", "summary": "工作日期 2026-05-19。本文否定前序空白 TMT 论点并选择 (A)：AI 基础设施仍是领涨主线。证据是 hyperscaler capex/收入比虽已升至约 30% 的四家公司 Q1 合计强度，且 Alphabet、Amazon 的 FCF 张力明显，但 Microsoft AI ARR…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "AI算力链盈利可见性已被订单簿三重验证、足以支撑保留成长端的策略；半导体设备应拆分成‘AI直接受益+国产替代超配，纯中国敞口设备与晶圆厂低配’，故第03卡‘不全面切向价值’成立但需要…", "summary": "截至2026-05-21，本卡对第03卡“不全面切向价值”进行压力测试，结论是有条件支持。AI算力链（GPU/HBM/光模块/液冷/先进封装）盈利可见性高且仍在抬升：四大美系hyperscaler 2026年合计资本开支指引上修至USD 360–410 bn、TSMC CoWoS与SK Hynix…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 1, "risks": 1, "neutral": 0, "value": 60.8, "events": [ { "title": "'效率降级'对 2026–2027 AI Capex 总量近乎无约束，但通过电力、关税、航运三条渠道侵蚀单位经济学，使全球 AI 基础设施从'同步扩张'转为'地理极化'，估值真正承压…", "summary": "本卡对 Research note 1–5 的“效率降级削弱 AI 算力扩张”假设进行双向压力测试。结论：方向正确但程度被高估，AI 算力 Capex 在 2026–2027 名义增速仍 +19–26% YoY（北美 hyperscaler 2026 指引 4,400–4,650 亿美元 [S1–…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "NVDA 这份印数是 AI capex 上行周期的高置信度确认；FY26 H2 真正的盈利杠杆在 CoWoS / HBM / 800G 光 / AI 服务器 ODM 这条上游链上，不…", "summary": "NVDA Q1 FY27（盘后 2026-05-20）数据中心收入 US$75.2 bn、同比近翻倍、占总收入 92%，新增 US$80 bn 回购授权，是 AI capex 上行周期的高置信度确认。2026 H2 真正的盈利杠杆落在上游：TSMC CoWoS、SK 海力士/三星/美光 HBM、A…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "半导体板块在高利率与中美并购友好政策共振下，正在以更快节奏复现生科板块的哑铃式价值重估——整合者赢家通吃、后期去风险标的被并购溢价重新锚定、中间产能被同步出清。", "summary": "截至2026-05-24，本卡片自下而上验证半导体板块正在高利率（UST10Y 4.56%）背景下复现生科哑铃逻辑。证据集中在三层：(1) 并购重组——Synopsys-Ansys ($35B)、AMD-ZT Systems ($4.9B)、HPE-Juniper ($14B)、Silver La…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 2, "value": 55.6, "events": [ { "title": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？", "summary": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？ DeepSeek-V4-Pro 75% price cut × NVDA FY27Q1 Data Center +92% joint p…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？", "summary": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？ 同日落地的两个 AI 信号——DeepSeek-V4-Pro API 永久降价至原价 25%（中国侧 token 经济砍 75%）与 N…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "现有先进半导体出口管制不会停止AI使用或全部ASIC设计，但会系统性削弱受限生态通过HBM密集、CoWoS级封装实现“每瓦有效推理”跃迁的能力。", "summary": "截至2026-05-25，本卡压力测试前序“ASIC与先进封装推动每瓦有效推理”的判断。结论是：路径对可获得TSMC、HBM与CoWoS产能的全球云厂商仍成立，但对中国相关或D:5受限生态被结构性削弱。美国出口管制已从成品AI芯片延伸至HBM、封装路径、先进制造设备、软件/技术和FDP规则，卡住了…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "短期硬约束在 HBM 与 CoWoS-L 而非电力，AI 算力'紧但能出货'，对指数 FOMO 有利但 2026H2 脆弱性显著上升。", "summary": "短期（2026 年内）芯片侧的约束比电力更紧：依次瓶颈为 HBM3e bit 供给 → 台积电 CoWoS-L 先进封装 → GPU 晶圆 → 基板/光模块 → 电力。HBM 与 CoWoS-L 全年额度已锁、扩产跑不过 ~150%/~70% 的需求增长，2026H2 任一良率事件或出口管制重置可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 2, "neutral": 1, "value": 55.4, "events": [ { "title": "4.5%以上美债环境下，AI硬件龙头的业绩增长足以支撑选择性配置，但不足以为整个AI久期交易兜底。", "summary": "2026-05-28结论：本卡对上一张宏观卡采取stress-test但不否定。4.50%美债锚会压缩AI久期资产估值，但NVIDIA、TSMC、Broadcom、SK hynix/Micron等稀缺硬件链凭借高增长、真实云厂商capex、先进封装和HBM瓶颈，仍能部分抵消估值下修。风险在于cap…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "MU/SOXX继续加权、A股算力代理结构性降配、资金转向真实瓶颈+低拥挤应用+短缺材料——三段同步执行，正式在仓位结构上把海外硬件beta与A股叙事beta脱钩。", "summary": "支持MU/SOXX继续作为全球AI硬件稀缺信号加权——HBM、CoWoS、电力变压器三段瓶颈在2026年仍未缓解。同时主动降配A股高杠杆算力代理（CPO/光模块整机、AI服务器组装、IDC高估值租赁），因为这些标的与海外硬件共享叙事但不共享基本面节奏，过去12个月涨幅由估值贡献主导[自测算]，拥挤…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力物理墙的宏观结果是资本开支被拉长而非崩盘，TFP 红利右移 18-24 个月，配置上有利于电力/封装/国产替代而非纯算力多头。", "summary": "截至 2026-05-27，前序卡片揭示的 CoWoS-L 翘曲与 HBM4 多供应商 KGD 危机不会引爆 CSP 资本开支崩盘，而是把周期向后拉长至 2028-2029：Big Four 2026 资本开支 USD ~470bn 大概率维持，但 GPU 实际出货低于指引 12-18%，BOM …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 59.8, "events": [ { "title": "AI 硬件/CPO 篮存在 -24% 至 -49% 的硬估值底线，但 Research note 05 的 8% 进攻仓位在 KCB 50 1,750 攻防与两融折算率收紧的级联路径…", "summary": "本卡（2026-05-29）对 Research note 05 拨给 AI 硬件/CPO 龙头的 8% 进攻桶做底线压测。AI 算力子篮（中芯 688981、寒武纪 688256、海光 688041、澜起 688008、中微 688012、拓荆 688072）硬底线 -24% 至 -38%，CP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "AI硬件周期仍由需求驱动，但未来12个月的定价权将更多落在HBM、先进封装、光模块和电力交付等瓶颈环节，而不是单纯GPU或服务器组装收入。", "summary": "截至2026-05-30，本卡支持“AI算力瓶颈从硅端扩展至物理基础设施”的主线。未来12个月，AI芯片和服务器需求仍由超大云厂商资本开支、云AI收入、agentic AI推理流量和自研ASIC扩张驱动；但供应瓶颈已从GPU单点扩展到HBM、CoWoS/先进封装、800G/1.6T光模块、机架供电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "2026 年 AI 硬件供应链的真实 alpha 集中在合规非中国先进封装材料栈的少数节点级供应商，其有效期由 AI 算力 capex 周期、而非出口管制决定。", "summary": "把卡 01–06 收敛为单一因果链：国内 AI 硬件补贴是消费存量挤入而非结构跃迁（卡 01），上游消费电子毛利红利仅 1–3pct（卡 02），真正的毛利重估在 AI 服务器/HBM/CoWoS（卡 03），出口管制削弱中国本土但不破坏全球重估（卡 04），溢价选择性流向合规非中国材料链（卡 0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 2, "neutral": 1, "value": 61, "events": [ { "title": "超级IPO将把AI公司的公开融资压力传导为算力锁定需求，从而重估半导体资本开支：超配稀缺瓶颈，回避无法证明产能稀缺和合约期限的长尾扩产。", "summary": "截至2026-05-31，本卡片支持“超级IPO流动性虹吸”对AI硬件和半导体瓶颈资产的强化作用。核心判断是，IPO资金不会平均利好半导体，而会把公开市场资金约束转化为GPU、HBM、先进封装、先进制程、光刻和数据中心网络的多年期需求可见度。TSMC、NVIDIA、ASML、Micron、Sams…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 算力供给受限于 HBM4 售罄、CoWoS 封装良率及全球电网容量的物理极限，使硬件龙头成为 AI 价值链中无视‘Capex 断崖’风险的最终收费站。", "summary": "截至2026年5月，AI硬件供应链仍处于结构性紧缺。TSMC N2量产初期良率65-70%，HBM4于2月投产且美光2026年产能已售罄。NVIDIA Rubin架构推升单GPU功耗至2.3kW，使电力供应成为比芯片更硬的瓶颈。尽管终端应用出现DAU增速放缓，但硬件端的实物稀缺性构成了核心护城河，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "滞胀成本压力会侵蚀但不会基准性摧毁Regime A SaaS利润率，真正的分水岭是厂商能否把AI成本从固定席位打包迁移到明确的用量/结果计价。", "summary": "截至2026-06-02，本卡压力测试能源与大宗上涨对AI硬件供应链和Regime A SaaS利润率的传导。结论是有条件支持前序Regime A逻辑：晶圆厂电力成本上涨会压低硬件毛利，但单独不足以击穿SaaS Agent利润驱动；真正高Beta风险来自HBM、先进封装、数据中心电力与云厂商GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 1, "risks": 0, "neutral": 0, "value": 80.8, "events": [ { "title": "AI实物建设仍支持芯片需求，但2026-27年的边际份额赢家是定制硅和网络而非商用GPU，因此B层维持中性、内部偏好AVGO/MRVL并对NVDA估值更谨慎。", "summary": "截至2026-06-02，本卡支持Research note 01的B层中性，但建议改为杠铃式中性：更建设性看Broadcom和Marvell等ASIC赋能商，对NVIDIA保持近端战术正面、结构估值更谨慎。2026年定制ASIC份额27.8%大概率守住；按44.6% ASIC增速和16.1%商用…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 2, "neutral": 1, "value": 65.2, "events": [ { "title": "Broadcom (AVGO) 财报验证了AI订单总需求未断裂，但低毛利自研芯片的占比扩张和物理并网/封装刚性约束，正在系统性拉低硬件供应商的边际利润率并平滑其收入转化率，驱使AI算…", "summary": "截至2026年6月3日，Broadcom (AVGO) 发布Q2财报，其AI半导体营收111亿美元超预期，且上调全年指引至145亿美元，证伪了需求崖断裂假说；但其毛利率受低毛利定制加速器 (ASIC) 占比飙升影响，环比大跌290个基点至75.1%低于预期，且受限于CoWoS先进封装与北美电网并网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美股AI硬件强势对科创50半导体链存在明确补涨逻辑，但仅在先进封装和28nm/14nm主流设备两条主线上'清晰可定价'，应以'订单确认+流动性双重确认'同时为纪律抓手。", "summary": "针对2026-06-03美股AI硬件强势能否传导至科创50半导体链的核心问题，TMT分析师支持卡01战术框架，并把补涨逻辑结构化为可定价的两条主线：(1)先进封装（长电600584/通富002156/华天002185/兴森002436），承接台积电CoWoS 25-30%的产能溢出；(2)28nm…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "光模块成中美双市场新核心资产：拐点确认还是顶部信号？", "summary": "光模块成中美双市场新核心资产：拐点确认还是顶部信号？ Optical-module names crowned 'new core asset' in both US and China: regime shift or topping signal? 中际旭创 6-1 登顶沪深 300 第一权重…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "结构性工具足以覆盖AI物理层上市龙头的短期融资缺口，但不足以覆盖完整HBM国产替代资本开支，因此应保留头部瓶颈、压缩二线杠杆敞口。", "summary": "2026-06-04，本卡对卡片04作压力测试：1.2万亿元科技创新和技术改造再贷款、1.25%利率和可能的25bp定向降准，足以为A股CPO、光模块、PCB、先进封装龙头提供流动性过桥，尤其是中际旭创、新易盛等现金流强的物理层企业。但这不是完整国产HBM晶圆前道扩产的融资答案，HBM全球资本开支…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "端侧 AI 是推理负载结构边际上的替代，不是云端容量的替代——重定价的是 capex 科目，A 股该做的是多端侧存储+先进封装、空商品化 x86 ODM。", "summary": "Computex 2026 的 RTX Spark Arm SoC 加速了一场已经在进行的利润池再分配，而非给 hyperscaler capex 设上限。训练 GPU 与电力仍是 capex 的硬约束，端侧仅蚕食低 token、低延迟敏感的推理负载，Jevons 效应近端占优，端云为混合而非替代…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 4, "neutral": 0, "value": 54.4, "events": [ { "title": "Rubin Capex能保护一线AI硬件EPS，但WACC上行会把二线供应链从“AI成长资产”重新定价为“融资周期资产”。", "summary": "截至2026-06-08，本卡对前序AI硬件防御结论做压力测试：Rubin周期与云巨头Capex仍能支撑NVIDIA、TSMC、Broadcom、Samsung、Micron等一线硬件龙头的EPS增长，但不能让整个供应链免疫WACC上行。可见四大云厂商Capex底线约5742亿美元（自测算），足以…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网电力约束并未否决算力需求，而是将资本支出由单纯的 GPU 数量增长倒逼为高速网络互连升级与电力基础设施的溢价，使出海变压器与高速光模块成为双重赢家。", "summary": "本报告调和了 AI 算力 Capex 向电网/SMR 分流与 Marvell 股价暴涨的矛盾。电力墙约束限制了数据中心扩建，被迫推迟 300-700 亿 USD 的 GPU 订单；但云巨头为提升单瓦算力能效，加速铜转 1.6T 光模块及硅光子互连，使得通信连接侧价值剧增。Constellation…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "1.8 万亿宏观流动性虹吸将迫使二线算力平台 WACC 上行至 6.68%，导致其偿债能力（DSCR 降至 0.89x）违约并被迫递延 20%–35% 的 Capex，从而将信用风险…", "summary": "在 SpaceX IPO 和 1.5 万亿发债的双重流动性虹吸下，当二线 Neocloud 算力平台的加权平均资金成本（WACC）上行至 6.68% 时，其债务结构和资本开支将触及断裂临界点。此时，GPU 使用率降至 82% 或租赁单价跌超 15%，将导致其 SPV 的 DSCR 降至 0.89x…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年高技术制造业利润增长与TMT下游需求复苏的一致性检验 工作日期：2026 06 08（Asia/Singapore，机构权威工作日） 分析师：TMT行业分析师（tmt an…", "summary": "2026年高技术制造业利润增长与TMT下游需求复苏的一致性检验 工作日期：2026 06 08（Asia/Singapore，机构权威工作日） 分析师：TMT行业分析师（tmt analyst） 立场：stress test（对卡片06的\"需求驱动\"外推进行压力测试） 卡片：7/8 根主题：油价再…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 4, "neutral": 1, "value": 66.8, "events": [ { "title": "全球科技配置应超配ASIC与新代工生态；A股风格则需从“AI题材映射”战略性切换至独立算力的“国产第二供应源”，重仓先进封装产能、核心设备及HBM材料。", "summary": "2026-06-09评估：基于AI半导体估值锚向“增长+第二供应源”切换，全球配置需减配纯GPU倍数扩张与单一代工依赖，转而超配ASIC双雄(AVGO/MRVL)、HBM及Intel代工期权。在A股与港股层面，风格将从“英伟达映射”(光模块/代工)实质性轮动至业绩兑现度高的国产“第二供应源”。策略…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI半导体利润池迁移真实但渐进，板块总EPS上行、估值锚从'NVDA-TSMC双寡头溢价'切向'增长+多元化第二供应源'，ASIC设计商、HBM、先进封装设备与Intel Found…", "summary": "2026-06-09综合：Google-Intel TPU订单确认AI硬件利润池迁移真实且渐进，2027-28年AI半导体合计EPS较冲击前共识更高，但分布坚定向ASIC设计商(AVGO/MRVL)、HBM、先进封装设备及Intel Foundry期权倾斜。新估值锚为'增长+多元化第二供应源'，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "DeepSeek降价推动的是硬件内部的alpha重排，而不是AI硬件beta被证伪；应压缩泛化边缘和老库存，保留订单确认的光模块、HBM相关存储、液冷/电源和高端板卡。", "summary": "截至2026-06-09，DeepSeek V4-Pro 已验证永久降价为75%、至原价1/4；“96%”更像相对竞品的单位经济冲击。微观压力测试显示，GPU/HBM/800G与1.6T光模块/液冷仍受订单、价格和产能约束，Jevons弹性成立。中际旭创、沪电股份、佰维存储有收入、交付或稼动率确认…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 2, "neutral": 1, "value": 72.2, "events": [ { "title": "超配拥有确定性全球订单的光电液冷龙头与昇腾-中芯先进封装/HBM自主垂直栈，回避被截断离岸通道的二线芯片设计与空壳算力租赁标的。", "summary": "截至2026-06-10，本研究深入A股TMT板块。建议超配：(1)挂钩中美头部云、订单已确认的光模块（中际旭创、新易盛、天孚通信）、液冷（英维克）与电源龙头（麦格米特），均处于1.6T/Blackwell订单放量期；(2)华为昇腾与中芯国际（SMIC）挂钩的HBM/先进封装国产替代垂直栈，长电科…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件需求底足以支撑6-12个月反弹论据，但短期无法对冲科创50强平与折现率冲击的非对称下行，结构应为分批再介入而非全面低配。", "summary": "截至2026-06-10，AI硬件需求短期（4-8周）不足以对冲科创50局部去杠杆的折现率与强平资金流冲击，但中期（6-12个月）足以支撑反弹。美国四大云厂商2026年capex指引5100-5400亿美元、约75%已PO锁定，中国主权AI capex同比+45%、国产替代占比升至55-58%；H…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "EML是2026年1.6T光模块量产的唯一硬约束，结构性缺口贯穿1H26–3Q26，强化CPO龙头的定价权与右侧入场区安全边际。", "summary": "2026年200G EML（1.6T模组核心激光器）需求约5,900–6,600万颗，对应有效供给5,200–6,000万颗，缺口300–1,000万颗，集中在1H26。Lumentum/Coherent/三菱/住友/Broadcom五家扼住70–75%产能，扩产全部后置2H26。结论：1.6T量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 70.6, "events": [ { "title": "能源融资成本上升会削弱云客户ROIC并压低部分ASP，但最强AI硬件平台商仍有足够利润率缓冲，真正脆弱点在重capex制造和周期性内存环节。", "summary": "2026-06-11，本卡压力测试能源债务融资成本上升对AI硬件供应商利润率的影响。结论是方向成立但不应外推为硬件利润率全面坍塌：新增能源债务成本上行约85 bps会压缩云厂商自由现金流并强化采购纪律，但NVIDIA和Broadcom因高毛利、平台锁定和轻资产结构韧性最强；AMD居中；TSMC和M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CPO估值压缩下的AI算力基础设施需求韧性评估", "summary": "需求端确实强劲：2026年四大云厂商资本开支约7250亿美元（同比+77%）[S1]，中际旭创2025净利107.97亿（+108.78%）、新易盛95.32亿（+235.89%）[S4][S5]，PE仅16-17x、PEG约0.15-0.20x。但压力测试显示：估值下修主要是仓位/流动性事件（强…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 1, "value": 63, "events": [ { "title": "AI算力链估值锚点已转向‘电力交付速度’与‘自由现金流生成能力’，自研ASIC部署虽然拉长了基建周期，但通过提升光模块配比为中际旭创等上游CPO龙头提供了坚实的需求对冲。", "summary": "【工作日期：2026-06-12】AI算力链估值模型正从GPU囤积转向资产负债表与电力交付。Nebius扩产引发股权摊薄与债务担忧，而OpenAI与AWS的2GW Trainium合作表明重资产囤货受阻。4-7年的并网延迟拉长了投产周期，平滑了光模块（CPO）短期增长。然而，单算力下ASIC芯片光…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "北美云厂商2026H2 AI资本开支向中资光模块订单的传导验证", "summary": "压测确认：2026H2北美云厂商资本开支体量（合计约7000-7500亿美元，同比+60-77%）、GPU/网络供给爬坡与1.6T切换，足以支撑中际旭创300308、新易盛300502、天孚通信300394在2026H2乃至2027H1的环比订单延续，而非近端重复下单回落——中际在手订单超300亿…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供给的物理并网瓶颈与高密算力液冷配套开支的刚性上升，正在对超大规模数据中心资本开支形成显著的“基建挤出效应”，迫使纯算力硬件开支占比向电网、变压器和温控冷却设施倾斜，进而压制纯芯…", "summary": "截至2026-06-13，电力供给的结构性稀缺（并网延迟达24-48个月，PJM容量价格飙升10倍）与Blackwell机柜高功耗（单柜132kW）倒逼超大规模厂商进行CAPEX结构重构。定量测算表明：在无约束情景下，设施基建占比从23.08%上升至34.07%，其中配电占比由5.54%升至9.3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科技巨头直接注资电力资产因电网设备供应链瓶颈（变压器交期48-60个月）而演变成“资本陷阱”，导致高达40%的预承诺GPU硬件（$100亿）因并网滞后3年而遭遇65.7%（-$65.…", "summary": "截至2026-06-13，本报告对科技巨头直接注资电力资产的资本转移进行了压力测试。由于高压变压器（LPT）交期延长至48-60个月、取向硅钢（GOES）供应受限以及并网队列长达5年以上，资本转移正退化为“资本陷阱”。定量测试表明，在3年并网延迟与40%电网造价通胀下，一个2 GW的AI数据中心项…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 3, "neutral": 1, "value": 78.4, "events": [ { "title": "红皇后竞速并未彻底抹去自研ASIC的成本优势，而是将其集中至Google与AWS两家、使其他云厂商回到对NVIDIA的依赖，从而部分推翻了Research note 05所述的全行业…", "summary": "在'红皇后竞速'下，自研ASIC单世代有效寿命已压缩至12–15个月，2nm节点全口径NRE攀升至USD 0.9–1.4B、N2晶圆约USD 30,000/片、CoWoS-L售罄至2027。我们测算自研ASIC在TCO上跑赢商用GPU的年部署门槛从5nm节点的约20万颗升至现实情景下的15万–25…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "CoWoS绝对产能正在改善，但GPU与自研ASIC需求吸收速度更快，因此2026年AI集群交付仍受先进封装、HBM和关键设备交期的结构性约束。", "summary": "截至2026-06-13，本卡支持上一张AI基建卡的判断：硬件交付风险不只来自并网，还来自CoWoS/HBM/设备交期的半导体转换链。TSMC仍称先进封装非常紧，TrendForce口径显示CoWoS-L/S满载且2026年底目标120,000-130,000片/月；DIGITIMES虽更乐观，预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘冲突带来的能源价格上行与物理并网延迟正在迫使云厂商将资本开支从纯算力硬件向电网、变压器和温控液冷基建倾斜，产生显著的“基建挤出效应”；在脱碳路径上面临自备燃气发电（高ROIC但需…", "summary": "截至2026-06-13，本报告针对中东地缘能源冲击与电网瓶颈对AI算力基建的影响进行了压力测试。在$110/MWh高电费与$1364.40M配套基建（变压器GOES溢价及液冷）刚性支出下，超大云厂商为维持10%靶向ROIC必须调降GPU采购预算54.32%，挤压芯片商毛利至45.27%。在能源脱…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "训推分流把单一 HGX 订单链拆成‘西部互联带宽’+‘东部能效池化’两条独立赛道，延迟与带宽矛盾通过 Rack 内低延迟 + Pod 内大带宽 + 广域异步搬运的三层解耦在硬件层平衡。", "summary": "支持卡 07 的“训推分流”判断，并主张延迟/带宽矛盾可通过三层解耦解决：东部 Rack 域内低延迟（CXL+UAlink）、西部 Pod 域内大带宽（800G/1.6T+液冷）、东西广域承担异步 checkpoint 搬运。两套硬件 BOM 完全分叉——西部按互联带宽+PUE 定价（光模块、PC…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 6, "neutral": 0, "value": 50.8, "events": [ { "title": "GPU抵押贷款收缩会削弱边际Neocloud订单可见度，但先进封装与先进制程需求更可能从融资套利订单转向信用更硬的超大规模厂商和ASIC路线，而非进入全面下行。", "summary": "截至2026-06-15，本卡压力测试GPU抵押融资收缩对半导体订单的影响：冲击不是AI芯片需求消失，而是订单可见度从Neocloud融资驱动采购转向超大规模厂商、ASIC、主权项目和带预付款的硬合同。NVIDIA Data Center在FY2027 Q1仍达752亿美元，TSMC先进节点和HP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在AI链内部按融资质量轮动：超配Tier-1平台与卖铲人(设备/HBM/封装/电力赢家)，低配靠GPU抵押融资循环的Neocloud股权与高收益债。", "summary": "综合前八卡推演，AI基础设施的约束已从硅到电力、再到融资可得性。本卡给出按融资质量过滤的五篮子相对配置框架：B1 Tier-1 AI平台(+300bp)、B2 设备/HBM/先进封装(+250bp)、B3 ASIC定制(+100bp)、B4 电力瓶颈赢家(+150bp)、B5 低质量AI基建(-2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力成本上升和并网延迟不会终结AI芯片周期，但会把赢家从峰值算力叙事筛选为能在稀缺MW下交付最高有效token产出的全栈能效架构。", "summary": "截至2026-06-14，本卡支持前序判断：AI扩张的瓶颈已从单纯GPU供给转向电力与并网，但半导体端可通过能效、液冷和先进封装显著对冲。设计指标将从峰值FLOPS转向每瓦token、每MW可部署token和机架级系统效率。Blackwell、Trillium TPU、Trainium2等路线显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备供给瓶颈导致的成本不确定性迫使云厂商在 2026-2027 年将 CAPEX 向电力基础设施大幅倾斜，AI 算力部署从“受制于GPU”全面转向“受制于通电能力”。", "summary": "截至 2026-06-14，电网设备出海的 ROE 波动与建厂压力正向超大规模云厂商传导。美国变压器交期拉长至 160-200 周，云厂商的核心痛点从“缺芯”变为“缺电”。为锁定产能，云厂商被迫承担更高前期成本，数据中心非 IT 开支（电力/制冷/土建）占比预计从历史的 20% 升至 35%-40…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 0, "value": 70.2, "events": [ { "title": "验证AI算力硬件链的估值承压：流动性虹吸下的先进封装与GPU产业链韧性评估", "summary": "本卡对AI算力硬件链做压力测试，得出'分化韧性'结论。资金流层面确认虹吸：2026-06-08半导体−219.33亿、通信−98.35亿、消费电子−69.92亿元净流出居前[S6]，年内93家半导体公司减持[S7]，叠加CXMT约295亿IPO锚定[S9]。但估值下移是资金流/轮动驱动、非盈利驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观信用收紧并未击穿全球AI周期，而是将AI基础设施层切割为「超大规模云厂＋央企国家队」核心（继续扩张）与「民营商业云」外缘（防御性收缩）的K型结构，TMT投资应聚焦CoWoS/HB…", "summary": "TMT综合：沃什鹰派持稳（3.50–3.75%）经离岸人民币流动性、汇率对冲摩擦与企业中长贷收缩，传导至AI基础设施层产生K型分化而非总量收缩。美超大规模云厂CY2026合计capex运行率3,200–3,600亿美元由经营现金流支撑不受冲击；央企运营商云经PBOC科技创新再贷款（约2.5%）继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 1, "risks": 3, "neutral": 1, "value": 82.6, "events": [ { "title": "\"session id\": \"a926801e 65a8 4b42 9925 8c25d5d1b685\", \"response\": \"由于您正在处理这一关键的 2026 年 6 月 …", "summary": "\"session id\": \"a926801e 65a8 4b42 9925 8c25d5d1b685\", \"response\": \"由于您正在处理这一关键的 2026 年 6 月 16 日研究节点，我已完成了对 AI 算力基础设施资本开支与电力瓶颈的深度调研，并将其与 CPO 及半导体板块的估值…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体能效能把AI扩张从“堆GPU”改写为“每MW token产出竞争”，但只能买时间、改善单位经济性，不能让电力瓶颈消失。", "summary": "截至2026-06-16，本卡压力测试“电力预算驱动扩张”逻辑：半导体能效提升真实且重要，Blackwell、先进封装、vLLM/FlashAttention、稀疏MoE和定制ASIC可显著提升每MW有效token产出。但它不能完全对冲电力瓶颈，因为美国数据中心用电量可能从2023年的176 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "吉文斯悖论与推理端计算膨胀导致算力及电力净需求呈指数级爆发，彻底推翻了电力基建搁置假说，反而迫使云厂商将资本开支结构向配套电网及液冷刚性倾斜，对高速光模块（CPO）等算力硬件产生相对…", "summary": "截至2026-06-17，大模型效率提升（如DeepSeek V4 API价格降至$0.15/百万Token）将触发吉文斯悖论与推理端算力膨胀。测算显示，即便MoE使单Token算力开支减少70.0%，但由于API调用量激增35倍且推理端Token密度升至877.5个，AI算力总需求仍暴增61.4…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI大基建进入“能效优先”阶段，二级市场资金正从粗放型算力硬件（CPO/服务器）向电力基建“瓶颈资产”与半导体先进封装“能效资产”进行结构性跨量级轮动。", "summary": "截至2026年6月17日，AI算力的电力挤出效应与能效转型正引发二级市场深刻的风格轮动。资金正从受资本挤压的传统CPO与服务器组装板块流出，转向两大确定性主线：一是具备强垄断属性与刚性Capex支撑的电力/电网设备（特高压、变压器、液冷）；二是受益于架构优化的先进封装（CoWoS、HBM）产业链。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 5, "neutral": 0, "value": 69, "events": [ { "title": "FOMC 后应以“选择性算力基础设施挤空”而非“无差别 AI 融涨”来交易：多稀缺、空低 ROIC，净多但对冲。", "summary": "2026-06-18 策略综合：前序链条成立，2027 风险是 ROIC 与结构分化而非 capex 悬崖。FOMC 后应做多稀缺算力资产而非泛 AI beta：前沿逻辑/设备/HBM/先进封装、自研 ASIC 与网络、电力与电网受益方应超配；商用 GPU 龙头保留但严控风险；低 ROIC 商用 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 能源重定价压的是超大规模厂商 ROIC 而非芯片 capex 效率——硅片主导 TCO、能源占比单位数，且性能/瓦溢价+杰文斯回弹反而净支撑前沿逻辑/先进封装/HBM 需求…", "summary": "2026-06-18 压力测试：能源是证伪芯片逻辑的错误杠杆。Epoch 1GW 机房年化 TCO 约 $8.5B，服务器占 ~59%、能源仅 ~7%；交付电价 +30% 仅抬升 TCO 约 +2.1%，翻倍也仅 +7%（自测算），而 IT 寿命 5→3 年使 TCO +41%——刷新节奏的影响是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "成长股18%的风险预算已达上限，AI-PCB与光模块一季报业绩高兑现能够支撑该敞口，但半导体设备环比下滑与HBM量产良率瓶颈封杀了额度上调空间。", "summary": "截至2026-06-17，本卡压力测试科创50及电子的18%敞口。高速PCB与光模块的一季报业绩暴增（中际旭创净利+262%，深南电路净利+73%）提供坚实支撑，具备较强成本转嫁能力；而半导体设备大陆一季度销售额环比下滑16%至109.9亿美元，长鑫存储HBM3量产受良率及先进封装制约。结论是18…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产交换ASIC的战略窗口正在收缩但未关闭，核心矛盾已从单点芯片可得性升级为先进工艺、EDA/IP、SerDes、CPO封装和软件生态的系统性获取风险。", "summary": "截至2026-06-18，本卡压力测试认为：国产交换ASIC窗口正在收缩但尚未关闭。收缩机制不是交换ASIC本身被完全等同于AI加速卡管制，而是BIS对先进制造设备、软件工具、HBM、实体清单和FDP的扩展，正在提高国产51.2T/102.4T路径对先进代工、EDA/IP、SerDes、CPO封装…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 5, "neutral": 1, "value": 65.4, "events": [ { "title": "[auto-handoff] SOX 广度 vs NVDA 背离是否动摇'AI=大盘集中'基准情形", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: dbb2f82e-e4eb-42af-8b94-dad607615615 结果摘要: Wrote handoff_response.md (8965 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "新一代信息技术链覆盖不足（4.3%激活），龙头（中际旭创300308、新易盛300502、兆易创新603986）已重估至高位。被低覆盖的alpha在上BOM、下游环节：随架构转向CPO/硅光、2.5D/HBM封装与液冷迁移。挑出5个低覆盖、强论证标的——源杰科技688498（光芯片，Q1净利+11…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大模型价格战引发的推理流量爆量，导致国产GPU设计龙头呈现“营收高增、毛利率承压”的量价背离，定价权向制造端瓶颈与TCO优化环节倾斜。", "summary": "在大模型极低价格战（DeepSeek-V4-Flash降至$0.14/1M tokens）推动下，云厂商Q1 AI基础设施收入大增79%（GPU云激增184%）。但因先进制程及封装产能受限导致BOM成本上涨10%-15%，加之云厂商降本压价，国产GPU（如海光一季度毛利率降至55.6%，同比下滑5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与GOES的区域化贸易壁垒通过拉长交付周期至154.5周，直接造成算力资产闲置折旧（24k GPU集群年拖累达5.624亿美元），迫使超算厂商进行CAPEX重组。", "summary": "截至2026-06-18，研究变压器与GOES区域壁垒对AI算力的传导效应。美国IRA本土含量、欧盟NZIA及中国反倾销税等碎片化政策使GOES供给割裂。压力情景下，卷线工稀缺与设备积压将变压器交期拉长至154.5周。这导致超算并网时滞达3-5年，使得24k GPU算力集群闲置一年的折旧与流失利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 10, "neutral": 0, "value": 33.8, "events": [ { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A-share de-leveraging to 1550 points triggers secondary market liquidity locking and margin…", "summary": "若市场下探至1550点，中际旭创股价将跌至714.16元，击穿融资预警与平仓线，带来日均31.2亿元平仓抛压（占ADV 17.36%），锁定二级流动性。但公司持有125亿现金及45亿海外客户预付款，可提供151.5亿净营运资金，供应链运营违约风险极低。海外订单真正风险在于电网并网排队达36-60个…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "ASIC光互联需求真实存在，但2026-2028中际旭创式CPO供应商的正确压力测试是按电力状态下修约20-25%的收入节奏，而不是机械套用2.92x光模块倍数。", "summary": "截至2026-06-20，本卡支持电力约束会重塑CPO收入节奏，但否定“ASIC光模块用量固定为GPU的2.92倍”这一过强假设。AWS Trainium2、Google TPU、Meta MTIA和Broadcom AI半导体收入证明ASIC design win真实存在；OpenAI/Broa…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与封装成本上升会触发AI基础设施技术路线调整，但截至2026-06-20仍不足以触发超大规模云厂商AI芯片采购总量的全面回撤。", "summary": "工作日2026-06-20：本卡压力测试认为，电力和先进封装成本上升不会在2026年触发超大规模云厂商全面削减AI芯片采购，但会提高单位弹性，推动技术路线从“更多GPU插槽”转向“每瓦性能、液冷机架、自研ASIC、提前锁定CoWoS/HBM和电力站点”。真正风险是并网、变电站、CoWoS或HBM配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 7, "neutral": 0, "value": 53.2, "events": [ { "title": "AI硬件供应链波动与融资风险传导", "summary": "作为半导体分析师，我对 Research note 03–07 做压力测试并最终支持，但补上一层基本面触发器：光模块/先进封装景气真实但狭窄且反身。2026 全球 1.6T 光模块需求约 500 万只，中际旭创目标 500–600 万只、约 50% 份额、约 75% 硅光 [S1]，但估值高达 7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-20，受大型变压器48-60个月与液冷关键零部件CDU/QD产能物理受限的强约束，超大规模云厂商的AI Capex落地将被迫延迟2至3年，100MW的算力集群并网推…", "summary": "本卡片对超大规模云厂商在2026年锁定的3,750亿美元AI资本开支扩张进行了物理压力测试。结论表明，先进封装引发的高密度芯片集成正面临电网容量、变压器和液冷链的物理限制。新建100 MW GB200 NVL72数据中心包含25亿美元算力CAPEX和13亿美元液冷设施CAPEX。在即时（JIT）采…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI capex supercycle 仍在，但 higher-for-longer 与能源波动会把扩产闸门从芯片采购转向电力可交付性、容量成本和融资期限匹配。", "summary": "截至 2026-06-21，本卡对 higher-for-longer 与能源波动环境下 AI 算力基建做压力测试。结论：超大规模云厂商不会停止扩产，但边际决策从“GPU 可得性”转向“firm megawatts + 客户承诺 + 可融资全成本”。Amazon、Alphabet、Meta 已披露…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力约束不会终结AI算力扩张，但会把半导体设计目标函数改写为每瓦有效算力，并提升Chiplet与先进封装的战略溢价。", "summary": "截至2026-06-21，本卡片支持前序判断：电网接入和可交付兆瓦成为AI扩张区域性天花板后，半导体竞争会从峰值算力转向系统级Perf/Watt。数据中心用电增长使每瓦tokens、HBM带宽、die-to-die互连、热设计和先进封装成为容量权变量。Chiplet、CoWoS/SoIC、HBM4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 3, "neutral": 0, "value": 78.6, "events": [ { "title": "在2026-06-21，受2nm GAA先进制程 wafer 溢价与先进封装 80% 初期装配良率的限制，下一代高密度 AI 芯片的制造 BOM 将翻倍至8,153.59美元，其中封…", "summary": "截至2026-06-21，本卡片评估了AI机架密度120kW+下，先进封装良率及散热材料研发对AI芯片成本与供应链的重塑。下一代2nm GAA+HBM4+液冷芯片BOM成本较3nm基准大涨77.2%至6,780.97美元，在初期封装装配良率降至80%的重度压力下更暴增113.1%至8,153.59…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "提升芯片能效与先进封装无法对冲变压器短缺造成的并网延迟损失（闲置折旧高达25亿至50亿美元），反而因功耗密度飙升将瓶颈推向机架级配电，且受限于国内封测良率瓶颈与成本压力，持有电网设备…", "summary": "截至2026-06-21，本报告对“半导体设计/封装技术迭代能对冲电力并网延迟”的论点进行压力测试。分析表明，12-24个月的变压器交付延迟对200k GPU集群将造成25.3亿至50.7亿美元的闲置折旧，重创ROIC并推升GPU租金成本达39.1%至156.7%，芯片端能效提升被财务减值彻底吞噬…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-22，低端封装过剩会压低错误产能池的价格，而高端迭代缺失会保留真正重要产能池的配给风险，使Hyperscaler AI资本开支更重、更不稳定、更容易出现已供电机房…", "summary": "截至2026-06-22，本卡压力测试卡片06：先进封装低端过剩并不能缓解Hyperscaler真正需要的高端CoWoS/HBM4认证产能短缺。TrendForce称2026年行业CoWoS-like产能可接近200,000 wpm、缺口或收窄至约10%，但若高端认证占比只有50%-70%，有效高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 6, "neutral": 0, "value": 62, "events": [ { "title": "在2026年中美云厂商CapEx受阻于电力红线以及大模型推理通缩的双重催化下，新一代信息技术产业链估值正向能耗比占优的定制ASIC上游填料（联瑞新材）以及享受API接口成本大坍缩的软…", "summary": "截至2026-06-22，新一代信息技术产业链正经历算力瓶颈由GPU转向电力/并网及先进封装、以及DeepSeek API降价96.37%引发软件应用毛利重构的剧烈调整。板块激活率处于4.3%极低水平。本报告筛选了联瑞新材、生益电子、宏和科技、金山办公和富创精密等5家低两融、抛压出清且具刚性催化剂…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "summary": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "可信算力区下的A股与STAR硬科技配置映射", "summary": "本卡把卡片01–07的'可信算力区'供应链结论翻译成A股/STAR硬科技排序，并用四道闸门（收入已拐点、订单可见、毛利确认、与可交付获批算力直接关联）区分真实订单受益与主题弹性。结论排序：光模块(中际旭创+192%、新易盛+106%)、AI服务器/ODM(工业富联净利+102.6%、订单至2027…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与定制化ASIC的深度整合，是将大模型访存能耗降低95.7%并实现3-5倍能效提升的核心硬件路径，亦是国产半导体链规避先进制程封锁、支撑算力电力约束的战略根基。", "summary": "截至2026-06-23，本卡研判支持通过先进封装与ASIC能效路径应对算力电力约束。2.5D封装结合HBM3E使访存功耗降低95.7%（从35 pJ/bit降至1.5 pJ/bit）；自研定制ASIC能能效比高出GPU约3-5倍，万卡集群年省电费2,943万元人民币，并释放48MW电网容量。国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 7, "neutral": 1, "value": 60.4, "events": [ { "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "summary": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2028年再融资悬崖与 AI 算力基础设施资本开支的信用冲突", "summary": "压力测试支持并延伸卡片 04：AI 基础设施信用风险真实且上升，但呈哑铃状集中，不在头条的 $725 bn 巨头资本开支上。四巨头（MSFT/GOOGL/AMZN/META，2026 资本开支约 $725 bn、+77%，约 75% 投 AI [S1]）凭现金流与 AA 资产负债表通过测试，尽管 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建配置应从“追逐算力设备增量”切换为“持有可交付电力稀缺性”，但保留CPO核心仓位，因为CPO仍是降低AI网络功耗的必要技术。", "summary": "截至2026-06-24，本卡综合前序结论：AI算力瓶颈已从GPU转向已通电MW，组合应从高拥挤CPO单边beta转向“变压器OEM+中国出海设备+GOES/铜材料+CPO核心赢家”的哑铃。100单位AI基建权益袖基准配置为韩日及全球变压器OEM 30、中国出海设备22、上游材料18、CPO/光模…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "硅片层分化(前沿量稳、英伟达定价权被定制ASIC侵蚀)对A股/指数AI集中度定价的传导", "summary": "卡片01指指数对AI集中度自满，卡片02–04论证物理瓶颈并揭示能效把量集中到前沿、定制ASIC侵蚀英伟达租金而非需求。本卡综合判断：被错杀的是利润率/结构事件而非总量事件，且A股已在沪深300内建出自己的AI集中度（中际旭创第一权重、新易盛第四、寒武纪第九，IT权重9.22%→20.38%）。配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 7, "neutral": 0, "value": 58.8, "events": [ { "title": "研究报告：超大规模云厂商CapEx节奏、ROIC收缩及光模块毛利侵蚀的传导机制 权威工作日期锚定： 2026 06 24 研究线程标识：已归档卡片索引 (研究记录序号)： 08/0.…", "summary": "研究报告：超大规模云厂商CapEx节奏、ROIC收缩及光模块毛利侵蚀的传导机制 权威工作日期锚定： 2026 06 24 研究线程标识：已归档卡片索引 (研究记录序号)： 08/08 分析师： AI Infrastructure Analyst 分析立场： Stress Test (压力测试) 1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] CPO/光模块量产与海外主要云厂商H2交付瓶颈核实", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: fc86b687-e8b0-4d16-afbe-880787373f89 结果摘要: Wrote handoff_response.md (9056 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件交付节奏与光模块订单验证", "summary": "截至2026-06-24，本卡支持“CPO延期不等于800G/1.6T砍单”的基本面判断，但加入节奏约束：Microsoft、Alphabet、Meta、Amazon的2026 capex与NVIDIA Rubin、Dell AI服务器订单、Lumentum光组件指引共同显示需求仍强，800G/1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "xAI 算力扩张对 AI 基础设施与能源供应的压力测试", "summary": "对卡片 01 前瞻风险的压力测试：跑步机确实存在，但卡点不在资本。约束层级为资本 < GPU < 电力。资本充裕（xAI E 轮 200 亿/估值 2300 亿，外加 200 亿 SPV 与 35 亿 Apollo 方案；2026 年超大规模厂商资本开支约 6000–7250 亿），GPU 因英伟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 1, "risks": 5, "neutral": 1, "value": 83.2, "events": [ { "title": "虽然物理电网排队与供应链原材料价格冲击显著抬高了局部建设阻力，但云厂商通过绿电PPA、高密度液冷及转向定制ASIC等结构性调整，能够完全消纳成本溢价，不构成算力扩张的绝对瓶颈。", "summary": "本卡对电网并网延迟、变气压原材料（铜/GOES）价格波动和本地化GPU封装良率冲击进行了量化压力测试。结果表明，虽然并网延迟和原材料成本增加会显著压缩项目NPV（5年延迟将使NPV降低79.99%），但云厂商可以通过迁移西部枢纽自建绿电、液冷优化和系统级混合算力架构，消纳国产芯片和能源价格溢价，并…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网缓释后，AI集群上线风险不是消失，而是转移到CoWoS/HBM、液冷验收与网络芯片的串行同步，CPO下行收窄但不等于收入曲线完全恢复。", "summary": "截至2026-06-26，本卡压力测试Research note 7：电网侧缓释可降低CPO最坏收入下修，但不能消除AI集群的串行硬件瓶颈。关键约束从单一接电问题转向先进封装/HBM、液冷机柜验收、交换ASIC/网络集成的节奏匹配。CoWoS缺口可能由约20%降至2026年底约10%，但10%仍足…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "做多迁移的稀缺租金应以电网硬件和材料为锚、以先进封装和定制ASIC为弹性，同时用能源空头、商用GPU相对空头和暗硅期权保护组合。", "summary": "截至2026-06-25，本卡综合前序链条：霍尔木兹能源溢价正在消退，稀缺租金迁移到铜/GOES、电网设备、先进封装和定制ASIC。建议用100.0名义多头配置15.0%材料、30.0%电网硬件、20.0%先进封装、15.0%定制ASIC/AI网络、10.0%电力接入和10.0%流动性，同时用35…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "1.6T光模块交付延迟对AI算力集群及超大规模云厂商Capex部署进度的影响", "summary": "支持卡片01/02的研究线并加以边界化延伸：200G EML 的 HTOL 风险确会让 1.6T GA 滑向 2027H1，但这是一道带宽前沿与单位经济性约束，而非整体 Capex 停摆。GB300 NVL72 的 scale-up 走铜缆 NVLink、scale-out 走 ConnectX-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 5, "neutral": 0, "value": 75, "events": [ { "title": "在2026-06-27的权威工作日背景下，国产HBM良率骤降逼迫国内算力网络化，而全球封装与电力阻尼拉长了云厂商CapEx变现周期，这使得制程免疫且订单排至2028年的可插拔光模块出…", "summary": "截至2026-06-27，本报告评估了HBM良率危机与先进封装瓶颈对算力集群技术演进的影响。国产HBM3封装良率降至2.66%，倒逼国内算力集群转向网络密集的Scale-out拓扑，推高了光模块需求弹性。全球云厂商2026年CapEx直逼7250亿美元，但受制于CoWoS晶圆先进封装冲突与吉瓦级电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "结构性再贷款提供的低成本流动性无法有效突破高端设备禁运的技术壁垒，制裁压力下国产HBM良率降至2.66%且大模型训练效率折损25%-30%，需通过在科技板块内超配CPO光模块等微观对…", "summary": "本报告对结构性再贷款支持下半导体“硬科技”在外部制裁压力下的对冲效果进行压力测试。分析显示，虽有5000亿再贷款及1.5%贴息将企业融资成本降至0.25%-0.50%[S1]，但流动性无法突破技术封锁。在先进TBDB与CMP设备禁运下，国产8-Hi HBM3封装良率由16.86%骤降至2.66%[…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球AI基础设施资本开支对中国产业链的传导机制与订单确定性", "summary": "本卡支持卡片05遗留命题并定论：约7250亿美元、同比+77%的全球超大厂资本开支（约75%即约4500亿为AI专项）规模巨大、外部出资，已转化为中国主导且出口管制并不约束的那一层链条（光互连/连接/散热/PCB）的跨多季度订单簿与已兑现盈利。关键在于出口管制是分层的：重压落在GPU/算力层（华为…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 3, "neutral": 0, "value": 88.4, "events": [ { "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "双轨AI网络可以落地，但硬件约束从电力单点转向电力、CoWoS/HBM、交换ASIC和光互联的同步供给，只有能跨堆栈预锁定产能的运营商能率先规模化。", "summary": "截至2026-06-28，本卡对Research note 05双轨网络做压力测试并给出有条件支持：CoWoS/HBM足以支撑头部训练集群，但2026-2027仍是分配瓶颈；硅光/CPO可降低核心fabric功耗和延迟，但都市推理边缘仍会混用可插拔800G/1.6T光模块与软件局部性。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 3, "neutral": 0, "value": 89.8, "events": [ { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-56…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-42faeae623/", "networkHref": "/investor/research-network/#thesis-42faeae623" }, { "id": "thesis-9d3836cc07", "source": "tracked_thesis", "title": "AI 生产率去通胀需要等待采用率和流程重构验证", "titleEn": "AI productivity disinflation still needs adoption and workflow proof", "lane": "AI 软件效率/生产率", "laneEn": "AI software productivity", "laneHref": "/investor/lanes/ai-software-productivity/", "investableFocus": "AI 软件效率/生产率：MSFT, GOOGL, META, AMZN, NOW, CRM, ADBE", "stockUniverse": [ { "ticker": "MSFT", "name": "Microsoft", "market": "US", "role": "企业软件与 Copilot", "roleEn": "企业软件与 Copilot", "bucket": "core", "href": "/investor/stocks/MSFT/" }, { "ticker": "GOOGL", "name": "Alphabet", "market": "US", "role": "模型、搜索与云效率", "roleEn": "模型、搜索与云效率", "bucket": "core", "href": "/investor/stocks/GOOGL/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "AMZN", "name": "Amazon", "market": "US", "role": "AWS 与自动化", "roleEn": "AWS 与自动化", "bucket": "core", "href": "/investor/stocks/AMZN/" }, { "ticker": "NOW", "name": "ServiceNow", "market": "US", "role": "企业流程自动化", "roleEn": "企业流程自动化", "bucket": "core", "href": "/investor/stocks/NOW/" }, { "ticker": "CRM", "name": "Salesforce", "market": "US", "role": "CRM 工作流与 Agent", "roleEn": "CRM 工作流与 Agent", "bucket": "core", "href": "/investor/stocks/CRM/" }, { "ticker": "ADBE", "name": "Adobe", "market": "US", "role": "创意软件 AI 货币化", "roleEn": "创意软件 AI 货币化", "bucket": "core", "href": "/investor/stocks/ADBE/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" } ], "stockTickers": [ "MSFT", "GOOGL", "META", "AMZN", "NOW", "CRM", "ADBE", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI" ], "coreView": "赛道：AI 软件效率/生产率。候选观察标的：MSFT, GOOGL, META, AMZN, NOW, CRM, ADBE。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 78, "eventCount": 160, "firstSeen": "2026-04-26", "lastSeen": "2026-06-29", "supports": 17, "risks": 130, "neutral": 13, "analystCount": 14, "sourceTypeCounts": { "研究线程": 5, "分析协作": 1, "研究笔记": 140, "证据更新": 11, "深度研究": 1, "研究主题": 2 }, "topTerms": [ "AI 基础设施", "电网", "Microsoft", "Amazon", "Meta", "HBM", "ASIC", "PPA", "AIDC", "电力", "企业", "productivity" ], "whatChanged": [ "2026-06-29: AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "2026-06-28: AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "2026-06-28: AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "2026-06-28: 电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "2026-06-28: 数据中心电网并网瓶颈与核能转向" ], "practical": { "sectorLane": "AI 软件效率/生产率", "sectorLaneEn": "AI software productivity", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "MSFT", "name": "Microsoft", "market": "US", "role": "企业软件与 Copilot", "roleEn": "企业软件与 Copilot", "bucket": "core", "href": "/investor/stocks/MSFT/" }, { "ticker": "GOOGL", "name": "Alphabet", "market": "US", "role": "模型、搜索与云效率", "roleEn": "模型、搜索与云效率", "bucket": "core", "href": "/investor/stocks/GOOGL/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "AMZN", "name": "Amazon", "market": "US", "role": "AWS 与自动化", "roleEn": "AWS 与自动化", "bucket": "core", "href": "/investor/stocks/AMZN/" }, { "ticker": "NOW", "name": "ServiceNow", "market": "US", "role": "企业流程自动化", "roleEn": "企业流程自动化", "bucket": "core", "href": "/investor/stocks/NOW/" }, { "ticker": "CRM", "name": "Salesforce", "market": "US", "role": "CRM 工作流与 Agent", "roleEn": "CRM 工作流与 Agent", "bucket": "core", "href": "/investor/stocks/CRM/" }, { "ticker": "ADBE", "name": "Adobe", "market": "US", "role": "创意软件 AI 货币化", "roleEn": "创意软件 AI 货币化", "bucket": "core", "href": "/investor/stocks/ADBE/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" } ], "stockTickers": [ "MSFT", "GOOGL", "META", "AMZN", "NOW", "CRM", "ADBE", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI" ], "primaryStocks": [ { "ticker": "MSFT", "name": "Microsoft", "market": "US", "role": "企业软件与 Copilot", "roleEn": "企业软件与 Copilot", "bucket": "core", "href": "/investor/stocks/MSFT/" }, { "ticker": "GOOGL", "name": "Alphabet", "market": "US", "role": "模型、搜索与云效率", "roleEn": "模型、搜索与云效率", "bucket": "core", "href": "/investor/stocks/GOOGL/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "AMZN", "name": "Amazon", "market": "US", "role": "AWS 与自动化", "roleEn": "AWS 与自动化", "bucket": "core", "href": "/investor/stocks/AMZN/" }, { "ticker": "NOW", "name": "ServiceNow", "market": "US", "role": "企业流程自动化", "roleEn": "企业流程自动化", "bucket": "core", "href": "/investor/stocks/NOW/" }, { "ticker": "CRM", "name": "Salesforce", "market": "US", "role": "CRM 工作流与 Agent", "roleEn": "CRM 工作流与 Agent", "bucket": "core", "href": "/investor/stocks/CRM/" } ], "instrumentTags": [ "MSFT", "Microsoft", "GOOGL", "Alphabet", "META", "Meta Platforms", "AMZN", "Amazon", "NOW", "ServiceNow", "CRM", "Salesforce" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 6 类，近期事件 98 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 6, "analystCount": 14, "recentEvents": 98, "supportRatio": 0.11, "riskRatio": 0.81, "conflictScore": 90.1, "evidenceDepth": 100 }, "score": 72.4 }, "events": [ { "date": "2026-06-29", "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,800万美元（约降33%）[S1][自测算]；先进封装（CoWo…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "电网", "HBM", "电力", "企业", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计约6900-7200亿美元（自测算），NVIDIA FY20…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 38.75, "matchedTerms": [ "Microsoft", "Amazon", "Meta", "HBM", "电力", "META", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI交易不会结束，但估值核心应从买到多少芯片转向能否按期通电、保持利用率并降低每token能耗。", "summary": "截至2026-06-28，本卡综合支持前序判断：2026-2030年AI扩张的约束从服务器交付转向可通电MW交付。Amazon、Microsoft、Alphabet、Meta最新季度CAPEX/PPE购买合计约$130.8B[自测算]，但IEA预计美国数据中心用电到2030年较2024年增加约240TWh，约等于27.4GW连续负荷[自测算]。结论是Hyp…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.25, "matchedTerms": [ "电网", "Microsoft", "Amazon", "Meta", "电力", "META" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "summary": "截至 2026-06-28，本卡压力测试认为：电力交付瓶颈会迫使 hyperscaler 从单纯追求新增 MW 转向 MW 生产率和存量优化，但不会终结扩产。Microsoft、Alphabet、Meta 可见 2026 年 capex 区间合计约 4,950-5,250 亿美元（自测算），同时仍披露容量约束；IEA 显示数据中心 1-3 年可建成，而新电…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 33.25, "matchedTerms": [ "电网", "Microsoft", "Meta", "电力", "META", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "数据中心电网并网瓶颈与核能转向", "summary": "截至2026年6月28日，AI基础设施建设的核心瓶颈已从GPU稀缺转移到电网并网时间（Time-to-Power）。对100 MW超大规模AI数据中心的模型测算显示，2年的并网延迟会减少项目35.73%的净现值（NPV），并使IRR大幅下降。为规避此类延迟，科技巨头（微软、亚马逊、谷歌、Meta）在2024至2026年间锁定了超十吉瓦（GW）的核电容量，通…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28, "matchedTerms": [ "电网", "Meta", "PPA", "电力", "META", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗ASIC。先进封装成为泄压阀，但TSMC已提示先进封装产能紧张…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电网", "HBM", "ASIC", "电力" ], "href": "" }, { "date": "2026-06-28", "title": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "summary": "截至2026-06-28，本卡否定“AI资本开支韧性等于A股防御性配置锚”的宽泛结论。Microsoft、Alphabet、Meta、NVIDIA、TSMC等数据仍支持AI基础设施真实需求，但FOMC 3.50%-3.75%、5月PCE同比4.1%、10年美债4.40%和A股两融突破RMB 3tn，意味着长久期成长股、拥挤主题和融资敏感供应商仍可能估值下修…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.75, "matchedTerms": [ "Microsoft", "Meta", "HBM", "META" ], "href": "" }, { "date": "2026-06-28", "title": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "summary": "截至 2026-06-28，本卡压力测试 Research note 1 的 AI 基础设施使能层主线：需求真实且强劲，但 2026-2028 年兑现会被电网接入、可靠电源、变压器、燃机和运行可靠性约束。IEA 预计全球数据中心用电从 2024 年 415 TWh 升至 2030 年约 945 TWh；美国和中国合计贡献近 80% 增量。[S1][S2] …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.75, "matchedTerms": [ "AI 基础设施", "电网", "电力", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期：…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 / Executive Summary 截至 2026 年 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.75, "matchedTerms": [ "AI 基础设施", "Microsoft", "Amazon", "Meta", "电力", "GOOGL", "META", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "变压器材料和交付瓶颈已开始改变 Hyperscaler 的2026-2027资本开支与电力采购节奏，影响主要体现为预算通胀、长期锁电和通电交付递延，而非AI需求取消。", "summary": "截至2026-06-28，结论为支撑：Hyperscalers 已经通过预算上修、长期电力合同和项目节奏调整来应对变压器及电气设备瓶颈，但公开披露通常不会把变化直接归因为 GOES 或铜关税。Microsoft 2026 capex 中约250亿美元来自更高组件价格，Meta 上调2026 capex中点100亿美元，Amazon 把电力容量作为部署KPI…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 33.75, "matchedTerms": [ "Microsoft", "Amazon", "Meta", "电力", "META", "infrastructure" ], "href": "" }, { "date": "2026-06-26", "title": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "summary": "截至2026-06-27，本卡片对AI基础设施链做压力测试：全球AI资本开支仍大，Goldman Sachs基准模型给出2026年USD 765 billion、2031年USD 1.6 trillion的年度AI capex，但Amazon、Alphabet、Meta、Microsoft的数据表明现金流、折旧、能源和电网接入正成为估值核心约束。A股不应全…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34.75, "matchedTerms": [ "电网", "Microsoft", "Amazon", "Meta", "电力", "META", "infrastructure" ], "href": "" }, { "date": "2026-06-26", "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–560 亿美元；中国云(字节/阿里/腾讯)与国产替代构成第二引…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.25, "matchedTerms": [ "AI 基础设施", "电网", "Amazon", "电力", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实触发重排，但方向是向支出倾斜，且买方自身点名的硬约束是电力与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 43.75, "matchedTerms": [ "Microsoft", "Amazon", "Meta", "HBM", "ASIC", "电力", "META", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "评估AI基础设施与半导体产业链在“更高更久”利率环境下的现金流质量与估值韧性", "summary": "压力测试Research note 2后结论：不能把AI基础设施当成单一“质量成长防御”桶——产业链已分化。算力卖方（NVDA型）财年2026营收$215.9B、经营现金流$102.7B、FCF$96.6B、FCF利润率约44.7%，当期造血、净现金，确属短久期质量成长。算力买方（MSFT/GOOGL/AMZN/META）2026年合计资本开支约$725B…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.75, "matchedTerms": [ "Amazon", "Meta", "电力", "MSFT", "GOOGL", "META", "AMZN", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "CNH套息去杠杆会先打掉A股AI基础设施股票的流动性溢价和估值久期，而不是立即打断全球AI CapEx实物周期。", "summary": "截至2026-06-25，本卡片压力测试CNH套息去杠杆对AI基础设施估值和CapEx预期的冲击。结论是：流动性溢价消失可使A股AI硬件/算力基础设施估值出现非线性回撤，但不应直接推导为全球hyperscaler CapEx崩塌。Microsoft、Alphabet、Meta、Amazon、Oracle和NVIDIA披露的2026年AI基础设施投入、订单积…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28.75, "matchedTerms": [ "Microsoft", "Amazon", "Meta", "META", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "电网扩容融资风险对超大规模云服务商数据中心选址与AI集群部署节奏的影响研究", "summary": "本卡综合 Research note 01–04，落到 Hyperscaler 一侧的应对函数。净判断：超大规模云服务商正把电力采购问题转化为自备供电与选址套利问题，对冲真实但仅部分有效。三大杠杆：(1) 选址迁移——从饱和的北弗吉尼亚/PJM 转向 ERCOT、路易斯安那等天然气富集与二线市场，靠近退役煤电/核电并网点可缩短 18–36 个月；(2) 自…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27, "matchedTerms": [ "电网", "Meta", "电力", "META", "infrastructure" ], "href": "" } ], "timeline": [ { "date": "2026-04-26", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "2026年Q1全球超大规模云厂商AI资本开支再评估与算力供给瓶颈传导", "summary": "2026年Q1全球超大规模云厂商AI资本开支再评估与算力供给瓶颈传导 在2026Q1业绩季后,北美四大Hyperscaler(MSFT/GOOGL/META/AMZN)的2026 AI capex一致预期较2025年12月时点上修了多少?增量资金的结构如何分布?HBM/CoWoS/光模块/电力等关…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-27", "supports": 1, "risks": 1, "neutral": 0, "value": 45.4, "events": [ { "title": "2026年四大超大规模云厂商AI资本开支约7,000亿美元的可持续性,以及电力瓶颈对Capex落地节奏的真实约束", "summary": "2026年四大超大规模云厂商AI资本开支约7,000亿美元的可持续性,以及电力瓶颈对Capex落地节奏的真实约束 在自由现金流被显著挤压、电网与变压器排队进入'以年计'的当下,本周(4月24-30日)Microsoft/Meta/Alphabet/Amazon的FY26一季报会兑现还是修正这条7,…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI 大厂 capex 指引边际曲率", "summary": "致 日报总编： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 4905c420-d2ab-4c97-97c1-99465e0023d6 结果摘要: 为了回答海外超大规模厂商（Hyperscalers）一季报中 Capex 指引的边际变化，我重点分析了本周（以 2024 …", "sourceLabel": "分析协作", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-04-29", "supports": 0, "risks": 0, "neutral": 1, "value": 47.8, "events": [ { "title": "2026年一季度超大规模云厂商AI资本开支轨迹与GPU/ASIC供应链含义", "summary": "2026年一季度超大规模云厂商AI资本开支轨迹与GPU/ASIC供应链含义 在Alphabet、Microsoft、Meta、Amazon集中披露2026年一季报的窗口,云厂商AI基础设施资本开支是否仍维持50%+同比增速、2026年全年指引是上调还是下调,以及边际美元在GPU、定制ASIC和电力…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "Q1 2026 财报后的云巨头 AI capex：需求验证还是瓶颈预警？", "summary": "本卡选择 tmt-analyst 评估 Q1 2026 云巨头 AI capex 周期。Microsoft、Alphabet、Amazon 和 Meta 均显示 AI/云收入加速，但 capex 和 PP&E 支出同步大幅上行。TSMC 的 61% HPC 收入占比、74% 先进制程占比和 66.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "已签第一波 AI 核电 PPA 有超大云 capex 和现金流支撑，但未签约管线与 AI 电力估值溢价应被显著折扣。", "summary": "本卡对前序公用事业论证做压力测试：Microsoft、Meta、AWS 和 Google 的 AI capex 足以支撑 2027-2030 年第一批 3-5 GW 已签清洁 firm power 合同，但不足以把所有未签核电管线都定价为 20 年、$100+/MWh 年金。重要修正是 Meta …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-03", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "2026年4月 hyperscaler AI capex 重估", "summary": "本卡由 tmt-analyst 作为开题分析师完成。结论是：2026年 hyperscaler AI capex 仍是需求驱动周期，而非简单泡沫破裂；Microsoft、Alphabet、Amazon 和 Meta 的 2026 capex 大致落在 $695-725 billion 区间，且由 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 1, "risks": 9, "neutral": 0, "value": 12, "events": [ { "title": "能源 Capex 对科技巨头主要是 FCF 估值折价而非偿付能力危机；长期电力合约则让公用事业公司获得可融资、可证券化的基础设施现金流。", "summary": "本卡片综合前序判断：200 亿美元电网升级和长期核电 PPA 不会压垮科技巨头资产负债表，但会显著改变 FCF 估值框架。Microsoft、Alphabet、Meta 仍能以内生现金流支持 AI 电力基础设施；Amazon 最敏感，Q1 2026 TTM 经营现金流 148.5bn 美元、现金 …", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "SMR 是必要但不充分,2025–2028 的边际电力来自燃气、LWR 重启与现役核电功率提升,真正的瓶颈是 NRC 审批通量、HALEU 与 FERC 监管,而不是反应堆设计。", "summary": "宏观开场:科技巨头的核电 PPA(Microsoft–TMI、Amazon–Talen、Google–Kairos、Amazon–X-Energy、Meta RFP)是写在 2028–2035 供给曲线上的远期期权,而非已落地的解决方案。需求端,IEA/EPRI 给出 2030 年全球数据中心用电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "hyperscaler AI 负荷以 firm 为主，non-firm 灵活性窗口只有 15-25%，PJM 灵活电价带来的节省被高估约 4-6 倍，真正的投资倾斜应回到 firm …", "summary": "对 Card-01 援引的 Johns Hopkins $15-16 B 节省进行压力测试。拆解 hyperscaler AI 负荷构成：在线 inference 占 45-55% 且持续上升、基础模型训练 25-35%、fine-tuning/RLHF/eval 5-10%、内部批处理 10-1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "把推理嵌入高 ARPU SaaS 或广告产品的平台（MSFT、GOOGL、META）能用每 MW 毛利摊销新增能源账单；把 AI MW 当裸算力出租的平台（AMZN）则处于估值观察期…", "summary": "AI MW 全成本约 $3.5–4.0M/MW/年（GPU/服务器折旧 + 厂房 + 含 PPA 溢价的电力 + 电网接入 + 运维）。按工作负载层划分：第 1 层 SaaS 嵌入推理（M365 Copilot、Workspace Gemini、GitHub Copilot Enterprise）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 2, "neutral": 3, "value": 47.4, "events": [ { "title": "2026年谷歌、微软、亚马逊、Meta四大超大规模云商合计承诺约7250亿美元AI基础设施投资，较2025年增长约71%。", "summary": "2026年谷歌、微软、亚马逊、Meta四大超大规模云商合计承诺约7250亿美元AI基础设施投资，较2025年增长约71%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年四大超大规模厂商（微软、亚马逊、谷歌、Meta）预计在AI基础设施上投入7000亿美元", "summary": "2026年四大超大规模厂商（微软、亚马逊、谷歌、Meta）预计在AI基础设施上投入7000亿美元", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在 $3.5–4M/MW/年 成本下限下，超配 Tier 1+2 变现者（MSFT/GOOGL/META）与给他们开发票的半导体和电网基础设施，标配观察 AMZN，低配或做空裸算力出…", "summary": "在 Research note 05 给出的 $3.5–4M/MW/年 成本下限与五层收入分布之上，本卡构建美股 AI 主题子组合的配置框架：超配 Tier 1+2 变现者 MSFT/GOOGL/META（50%）与给他们开发票的半导体（NVDA/AVGO/TSM/AMD/ASML，25%）和电力…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2025年四大超大规模云厂商（MSFT、META、GOOGL、AMZN）合计capex约3000亿美元，2026财年总盘子约3600-3900亿美元，同比增长20%-30%", "summary": "2025年四大超大规模云厂商（MSFT、META、GOOGL、AMZN）合计capex约3000亿美元，2026财年总盘子约3600-3900亿美元，同比增长20%-30%", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 1, "risks": 2, "neutral": 0, "value": 47.8, "events": [ { "title": "截至 2026 年初，微软、Alphabet、Meta、亚马逊均已维持或上调 2026 年 AI 基础设施资本开支指引", "summary": "截至 2026 年初，微软、Alphabet、Meta、亚马逊均已维持或上调 2026 年 AI 基础设施资本开支指引", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年Q1四大美超大规模云厂商（MSFT/GOOGL/META/AMZN）合计capex指引约6450-7250亿美元。", "summary": "2026年Q1四大美超大规模云厂商（MSFT/GOOGL/META/AMZN）合计capex指引约6450-7250亿美元。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年美股四大超大规模厂商(MSFT、META、GOOGL、AMZN)合计AI资本开支预计为3800-4300亿美元，同比增长约25-30%", "summary": "2026年美股四大超大规模厂商(MSFT、META、GOOGL、AMZN)合计AI资本开支预计为3800-4300亿美元，同比增长约25-30%", "sourceLabel": "证据更新", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-09", "supports": 0, "risks": 1, "neutral": 1, "value": 50.2, "events": [ { "title": "四家厂商2026年Q1业绩复核指引日期：MSFT（2026-04-29）、META（2026-04-30）、GOOGL（2026-04-24）、AMZN（2026-05-01）", "summary": "四家厂商2026年Q1业绩复核指引日期：MSFT（2026-04-29）、META（2026-04-30）、GOOGL（2026-04-24）、AMZN（2026-05-01）", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Q1 2026财报后超大规模云厂商AI资本开支的可持续性与瓶颈迁移", "summary": "Q1 2026财报后超大规模云厂商AI资本开支的可持续性与瓶颈迁移 在四家美国超大规模云厂商(MSFT/GOOGL/META/AMZN)Q1 2026集体上调2026年capex指引、合计推升至约6,500–7,250亿美元之后,这一资本开支周期的真正约束是否已从芯片供给迁移到电力与电网?权益市场…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 2, "neutral": 0, "value": 42.6, "events": [ { "title": "AI CapEx的主线不是收缩，而是从新增外壳数量转向每瓦、每机架、每封装的半导体密度提升。", "summary": "截至2026-05-13，本卡支持卡片01的“密度胜过足迹”判断：数据中心外壳新开工-19% QoQ并不等于AI CapEx下行，而是资本更多流向GPU/ASIC、HBM、CoWoS/先进封装、机架级互连、光DSP和液冷配套。Microsoft、Meta、Alphabet、Amazon的最新披露均…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Capex 扩张不会否定量子与空间的独立订单验证，但会通过机构风险预算挤出显著提高其波动率，尤其是量子硬件。", "summary": "截至 2026-05-13，本卡支持上一卡结论：量子与空间在订单/收入验证上部分独立，但在机构风险预算中仍是 AI 邻近仓位。Microsoft 约 $190B、Alphabet Q1 PPE $35.674B、Meta 2026 Capex $125B-$145B、Amazon TTM PPE …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 3, "neutral": 0, "value": 38, "events": [ { "title": "能源获取难度要求对AI基础设施进行选择性下修：终局需求不变，但IDC和算力租赁的CAPEX转化曲线、利用率和ROIC应显著打折。", "summary": "截至2026-05-15，本卡对电力瓶颈对AI基础设施的影响进行TMT压力测试。结论是：不应全面下修AI算力长期需求，但必须下修CAPEX转化节奏、利用率爬坡和ROIC。IEA预计数据中心用电从2024年415 TWh升至2030年约945 TWh，同时约20%规划项目存在电网延期风险；Micro…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "四家美股超大规模厂商（MSFT、GOOGL、META、AMZN）2026年合计资本开支指引约$355–410 bn，中值约$380 bn", "summary": "四家美股超大规模厂商（MSFT、GOOGL、META、AMZN）2026年合计资本开支指引约$355–410 bn，中值约$380 bn", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "四家美股超大规模厂商2025年实际合计资本开支约$327 bn", "summary": "四家美股超大规模厂商2025年实际合计资本开支约$327 bn", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 1, "risks": 0, "neutral": 0, "value": 65.4, "events": [ { "title": "AI需求是真实支撑，但不能完全抵消未来四周TMT/机器人解禁与拥挤交易的局部供给冲击，减仓约三分之一是合理的战术对冲。", "summary": "截至2026-05-15，本卡支持卡片02对高弹性TMT/机器人龙头提前减仓约三分之一，但理由是供给冲击和交易拥挤，而非AI基本面转弱。全球半导体、ASML、TSMC、NVIDIA、Microsoft、Amazon、Meta以及中芯国际、寒武纪数据均显示AI需求仍强；但未来四周约4120亿元解禁、…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-17", "supports": 1, "risks": 0, "neutral": 0, "value": 66.8, "events": [ { "title": "截至 2026-05-18，剩余错定价应集中在特变电工、国电南瑞、平高电气的具名电网 / 升压站项目转化，黔源电力只作贵州绿电验证仓，华能蒙电需等具名数据中心售电合约后再重估。", "summary": "2026-05-18：本卡综合并收紧前序 time-to-power 主线。最强具名项目是甘肃庆阳绿电聚合试点，特变电工已具名承建 330 kV 升压站；宁夏中卫大唐 50 万千瓦绿电直供和江南 330 kV 工程最利于国电南瑞、平高电气、特变电工通过电网招标兑现；贵安云谷 500 kV 变电站支…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-18", "supports": 1, "risks": 5, "neutral": 0, "value": 38.2, "events": [ { "title": "截至2026-05-19，AI半导体供应链尚未出现云CapEx或HBM/CoWoS订单环比下修，算力基础设施交易仍应保留核心仓位但需严密跟踪ROIC与Q2/Q3指引风险。", "summary": "截至2026-05-19，首席策略师提出的敏感触发点有效，但尚未触发。Microsoft、Alphabet、Amazon、Meta的2026年AI基础设施CapEx仍在上修或维持高位，TrendForce估计前八大CSP 2026年CapEx超过7100亿美元、同比约+61%。HBM方面，Micr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "TMT 已拥挤到足以为物理基础设施轮动出资，但尚未拥挤到需要放弃 AI 硬件周期；边际 AI 资金应从公开市场半导体动量转向电力、电网设备和 firm-capacity 受益者。", "summary": "截至2026-05-18，TMT 并非整体泡沫，但 AI 半导体和存储交易已达到拥挤区间：SMH P/E 38.56、前十大持仓 71.66%，DRAM ETF 五周吸金超过 $6bn，半导体/存储贡献 2026 年 S&P 500 市值增量的 70%。同时，NVIDIA、TSMC、MSFT、GO…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "卖方 Capex 下修是'交付排期推后'的财务表象，物理瓶颈是它的底层成因——同一个事实的两面，操作上要回补 TMT 半导体并保留电力腿，两条腿都做。", "summary": "卖方对 MSFT/META/GOOGL 2H26 Capex 下修，是把'承诺 Capex'与'现金 Capex'两条线混为一谈的财务误读，而误读的机械原因正是能源台所指的物理瓶颈——变压器交期、互联排队、燃机订单簿稀薄把现金落地推后。因此首席策略师与 energy-analyst 的归因不是二选…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "TMT 行业分析：AI 基础设施 CapEx 扩张与 SaaS 变现周期的一致性压力测试 日期： 2026 05 19 分析师： tmt analyst (TMT 行业分析师) 立场…", "summary": "TMT 行业分析：AI 基础设施 CapEx 扩张与 SaaS 变现周期的一致性压力测试 日期： 2026 05 19 分析师： tmt analyst (TMT 行业分析师) 立场： 压力测试（负向背离） 执行摘要 截至2026年5月19日，TMT 行业正面临上游基础设施投入与下游收入变现之间史…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 2, "neutral": 0, "value": 49.6, "events": [ { "title": "截至 2026-05-19，AI 基础设施仍是 TMT 领涨主线，但可投资范围应收窄到有真实 usage、backlog、产能瓶颈和现金流承受力的环节。", "summary": "工作日期 2026-05-19。本文否定前序空白 TMT 论点并选择 (A)：AI 基础设施仍是领涨主线。证据是 hyperscaler capex/收入比虽已升至约 30% 的四家公司 Q1 合计强度，且 Alphabet、Amazon 的 FCF 张力明显，但 Microsoft AI ARR…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 电力瓶颈的二阶效应不是'能效是否够用'，而是'剩余电力需求落在哪片司法辖区的电网上'——这决定了 2026–2027 年 USD/JPY/EUR/KRW 资本流向、变压器与 H…", "summary": "卡 03 已证明硬件与算法能效仅能削去 AI 电力缺口的 20–30%，本卡从全球宏观与地缘视角综合：剩余 70–80% 缺口将在司法辖区之间重新分配。美国瓶颈最严重但定价权最强（变压器进口依赖度升至 ~55%，232 调查待决）；欧盟核心区因 AI Act 与 PPA 准入门槛结构性失去这一轮 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 1, "risks": 1, "neutral": 0, "value": 65, "events": [ { "title": "超大规模云厂商的选址与架构调整真实且大幅推进，但2028年前无法完全弥补物理电力缺口，导致2027-2029年AI资本支出出现显著的'已公告 vs 已通电'分化与15-25%部署滑期。", "summary": "截至2026-05-23，TMT端压力测试显示超大规模云厂商正沿四条路径回应第8卡的电力缺口：地理分散（AWS印第安纳/密西西比/沙特HUMAIN、Meta路易斯安那、Microsoft威斯康星）、用户表后电源（Talen-AWS、Constellation-MSFT 20年PPA、Crusoe …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-05-24，AI硬件与半导体设备是A股结构性成长修复最可信的核心驱动力，但只能买订单和盈利修复，不能买泛科技估值扩张。", "summary": "截至2026-05-24，本卡支持ashare-strategist的结构成长修复判断，但将可投资主线收窄为有订单验证的AI硬件与半导体设备。NVIDIA、Microsoft、Meta、Amazon、TSMC、ASML、SEMI和SIA数据共同显示，全球AI capex、先进制程、先进封装、测试与…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 4, "neutral": 1, "value": 44.8, "events": [ { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力确定性风险不会抑制 Hyperscaler AI 资本开支，而会把边际美元从 GPU 重定向到变压器、变电站、BTM 燃机与稳态 PPA，结构性放大公用事业与电网价值链的订单可见…", "summary": "本卡支持卡 08 的结论：2026-2030 年 Hyperscaler 资本开支沿电力确定性轴分化，而非沿芯片轴。MSFT/GOOG/META/AMZN FY2026 资本开支合计 ~4,100-4,250 亿美元，同比 +22-25%，但边际美元正从 GPU 流向变压器、变电站、BTM 燃机、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek V4 标志着原始推理 token 化工品化，而不是 AI 全产业零毛利化；应把它理解为利润池从模型 API 向算力底座和工作流应用迁移。", "summary": "截至2026-05-26，本卡支持并细化既有判断：DeepSeek V4 的低价把原始模型 API token 推向重资产、低定价权、类似大宗化工品的层级，但并不意味着整个 AI 行业进入零毛利。利润正在从可替代模型调用迁移到上游芯片/网络/电力约束的基础设施，以及下游企业数据、治理和工作流集成。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "组合应保留云与半导体核心敞口，但把边际超配转向电网设备、可用电力、中国UHV和核电，因为AI瓶颈已从芯片订单转向通电交付确定性。", "summary": "截至2026-05-25，本卡综合前序结论：AI主线不应退出，但风险预算应从纯云/半导体beta转向“已通电AI交付beta”。建议AI电力主线100个风险单位中，MSFT/AMZN/GOOGL/META 30、半导体14、美国电网设备24、美国电力与BTM期权10、中国UHV/电网设备14、中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 2, "neutral": 2, "value": 61.2, "events": [ { "title": "AI 产业链利润正在按 firm MW、设备订单簿、电源效率 IP 和大用户 tariff 政治可防御性重新分配；2026–2027 年最强留存在电气设备 OEM、签 PPA 的 f…", "summary": "本卡综合 01–06 号卡，把 AI 产业链利润定价从单层'算力溢价'切换为按 firm MW、电气设备订单簿、电源效率 IP 和大用户 tariff 政治可防御性配给的四层结构。2026–2027 年利润留存最强的是 GEV/ETN/VRT 等订单簿已锁的电气设备 OEM、CEG/VST/TLN…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI电力需求并不天然利好公用事业；只有把数据中心负荷转化为有抵押、take-or-pay、且支付输电公平份额的合约收入，公用事业才能避免把AI资本开支变成居民电费和BBB信用风险。", "summary": "截至2026-05-27，本卡支持上一张信用卡片对BBB电网公用事业债务的谨慎判断，但进一步区分风险：真正危险的是拥有大量AI数据中心接入需求、却没有最低账单、抵押品、退出费和表前输电收费保护的公用事业。AEP Ohio式大负荷费率、Georgia PSC合同审查、Microsoft/Meta/G…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至 2026-05-26，电力设备瓶颈并未均等阻碍算力落地，而是建立了极度分化的双轨制市场，通过35倍的高额投产延迟杠杆，重奖前瞻储备与套利通道，严惩后发缺乏设备保障的云厂商。", "summary": "截至 2026-05-26，压力测试显示电力设备交付延迟对AI基础设施扩产节奏呈现高度分化的特征。大型变压器交付期拉长至3-4年，但第一梯队超大规模云巨头（MSFT/GOOGL/AMZN）已通过数年前的订单锁定了产能（如Eaton的12年储备订单和GEV单季电气化大单）；Meta等后发者则面临严重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 投资逻辑已从普惠的'非对称上涨'切换为按'资本结构灵活度'分化的个股选择制，价值陷阱集中于以扩张资产负债表为 AI 出资、且货币化滞后的标的（META 为典型）。", "summary": "2026 年 Q1 财报季后，美国四大云厂 CapEx 合计指引升至 6,600–7,250 亿美元（同比+77%），自由现金流被实质性侵蚀，AI 投资的定价已从'非对称上涨'切换为'资本结构分化'。MSFT、GOOGL 因 AI 营收跟得上支出仍受市场青睐；META 因 FCF 同比 -80%、…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "2026年AI芯片与先进封装产能约束会推高并拉长云端CapEx，而不是使其取消；配置应偏向具备配额权的上游瓶颈环节。", "summary": "截至2026-05-28，本卡支持前序判断：AI硬件CapEx传导链的约束真实但不等于削减。Microsoft、Amazon、Alphabet、Meta的2026年CapEx指引中枢约为$710 billion，反映的是芯片、HBM、先进封装、电力和数据中心资源稀缺导致的锁单与涨价，而非需求走弱。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 2, "risks": 1, "neutral": 0, "value": 80, "events": [ { "title": "推理成本下降的红利不会均匀分布——杠铃左端必须从硬件制造扩散到云、电力与企业SaaS，并以五条规则剔除纯CapEx标的，硬件从超配降为标配偏上，红利低波右端保持不变。", "summary": "在AI推理成本未来12-18个月下降20-35%的基准情形下，杠铃左端应从单一的‘AI硬件制造’扩散为‘云/算力（链A）+ 数据中心电力/HVDC（链B）+ 企业SaaS（链C）+ 减配后的硬件’四象限；右端红利低波保持不变。A股权益仓位维持60%，先进制造内部硬件由18%降至12%，电力由4%升…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "硬件成本红利与高CapEx会通过更低cost-per-token、更稳定算力供给和更可预测API价格改写AI应用单位经济模型，但真正的投资筛选变量是单任务成本改善乘以分发杠杆。", "summary": "截至2026-05-29，本卡支持前序判断：制造端成本红利会向下游传导，但更关键的不是单台服务器降价，而是Blackwell级架构、Dynamo类调度、自研芯片、云厂商高CapEx和API定价机制共同降低每个有效token和每个任务的成本。Microsoft、Alphabet、Amazon、Met…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI硬件基本面受Rubin切换与2nm量产支撑处于结构性上行，但软件货币化困境与电力瓶颈对估值修复构成脆弱压制。", "summary": "2026年5月29日调查显示，半导体基本面确实存在结构性改善。英伟达Q1营收破816亿美元，Vera Rubin平台与台积电2nm工艺进入加速期，HBM4产能2026年已售罄。虽然今日反弹受地缘政治挤仓驱动，但硬件层面的订单流强劲。然而，软件应用已进入“幻灭谷”，42%的企业试点失败，且美国电网电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 2, "risks": 1, "neutral": 0, "value": 81.4, "events": [ { "title": "半导体管制是地域与结构冲击，不是截至2026-06-01已经触发的全球 AI capex 断崖。", "summary": "截至2026-06-01，本卡压力测试半导体出口管制是否会触发“AI组合端”崩溃。结论是否定：出口管制会伤害中国相关 SKU、库存和产品结构，但 NVIDIA FY2027 Q2 指引在假设中国 Data Center compute 为零时仍达 $91.0 billion +/- 2%，Micr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美元指数回落至94.50与利率下行会显著支撑AI硬件估值扩张，但对资本开支的影响是强化既有需求周期，而不是单独创造新周期。", "summary": "截至2026-05-31，本卡支持“美元与利率下行有利于AI硬件及半导体估值扩张”的主线，但限定传导强度：最强作用来自折现率下行、弱美元折算、流动性和订单信心，而非单纯融资成本下降。Microsoft、Amazon、Alphabet、Meta的最新季度AI相关资本开支运行速率已超过1306亿美元，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI 算力的'电力税'利润池正从被监管电网转移至云厂商自筹融资的分布式平行系统——近端属于燃气轮机/发动机 OEM 与天然气中游，远端属于 SMR 厂商与 HALEU 燃料循环。", "summary": "卡片 04（能源行业分析师，立场=支持）延伸卡片 03 的结论：PJM 5+ 年并网排队与 3 年 GPU 寿命的错配迫使 AI 园区离网。2026–2029 年由表后航改型燃气轮机与往复式机组（CAT、CMI、Wärtsilä、Bloom）主导桥梁方案，已有 XOM 二叠纪 1.5 GW、CVX…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [ { "title": "AI 硬件不共享软件端的 SBC 稀释墙，但 capex 弹性 + HBM 成本 + ASIC 替代三重传导意味着 NVIDIA 与 AI 供应链的利润率扩张预期同样需要下修。", "summary": "2026-06-01，本卡对前序首席策略师结论作压力测试：AI 硬件端面对的不是同一道 SBC 稀释墙——NVIDIA 季度 SBC 同比 +30.8% 至 $1.928B 但会计稀释仅约 0.43%，显著低于 Amazon/Alphabet/Meta 一档——而是更陡的需求弹性墙。真正的压力源是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 3, "neutral": 0, "value": 56.2, "events": [ { "title": "Compute premium正沿AI电力瓶颈从需求栈再分配到物理供给栈：midstream、燃气轮机OEM、棕地核电IPP、电网设备和铀燃料承接租金；超大规模云因capex/收入失…", "summary": "本卡作为大类资产配置师，综合卡01–03给出跨资产配置地图：在2026年四大云厂7,000–7,250亿美元capex、其中60%+用于电力而非芯片的背景下，compute premium正从需求栈（超大规模云+高估值AI软件）迁移到物理供给栈。第一层承接：管道firm transport（KMI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-28年AI电力缺口主要由天然气桥接，ERCOT可行性高于PJM；SMR在2030年前仍是长久期期权而非实物供给闭环。", "summary": "截至2026-06-02，本卡支持上一卡的物理瓶颈判断，但把2026-28年的边际约束从电网进一步收窄到天然气可交付能力。美国全国天然气产量和Texas管道扩容足以支撑ERCOT较快桥接；PJM的风险在Appalachia-to-Virginia firm transport、冬季可靠性和单体园区…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-02的核心判断是：AI资本开支仍能验证硬件龙头盈利，但估值剪刀差将通过支出方分化和相对降估值来修复。", "summary": "截至2026-06-02，本卡综合而非否定上一卡：AI硬件拥挤风险存在，但Hyperscalers的2026年资本开支区间仍约为6950亿-7250亿美元，足以支撑NVIDIA、Broadcom、TSMC等硬件龙头近期收入指引。分化在支出方：Microsoft与Alphabet的云收入验证更强，A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "HBM/ASIC的结构性支撑在6-18个月维度上成立，但被卡片01低估了'客户集中度'和'瓶颈从硅迁移到电力'两条脆弱性——结构核心需要内部再平衡（提电力/冷却、减HBM股权、AVG…", "summary": "锚定2026-06-03（周三），AVGO FQ2业绩大概率周四盘后披露、NFP周五早间公布。我整体认同卡片01的'结构有支撑、战术有脆弱'综合判断，但TMT视角下结构性sleeve存在三条具体脆弱性需要再定价：(A) HBM需求高度集中在NVDA单一客户（约60-70%比特占比），属于穿着结构外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 0, "neutral": 0, "value": 85, "events": [ { "title": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决", "summary": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决 AI Capex Pivot to Grid/SMR: Has Consensus Already Been Repudiated by NVDA/MRVL Price Actio…", "sourceLabel": "研究主题", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 2, "neutral": 1, "value": 69.4, "events": [ { "title": "AI硬件需求仍处于“超速增长”阶段，2026年CapEx及订单可见度确认需求周期未变，估值过热及供应瓶颈（电力/HBM）是真正风险。", "summary": "截至2026-06-05，AI半导体产业链基本面依旧极其强劲，未见需求衰退实证。云巨头（MSFT, GOOGL, META, AMZN）2026年CapEx预期上调至7250亿美元，同比增长77%；英伟达Q1营收超810亿美元且指引加速；ASML订单中存储类首次超过逻辑类，反映HBM4需求井喷。A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "博通 ASIC 护城河的设计服务层 2027 年前不可替代，但客户集中层正在被稀释；真正的脆弱点不是 Google，而是 OpenAI 量产时点。", "summary": "压力测试结论：Google TPU 二供（联发科自 v7 起）、Meta MTIA、Amazon Trainium 对博通的威胁是分层的——Google 是真正稀释（2026–27 收入下行 –10% 至 –20% [自测算]），Meta 在 12 个月窗口仍是支撑，AWS 本就以 Marvell…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "博通的估值回归并非基本面崩溃，而是从“预期完美”向“竞争性执行”的结构性切换，Google 的多元化供应是 AI 叙事的首道裂缝，但 OpenAI 和 Meta 的增量支撑了其底部逻…", "summary": "博通 Q2 财报后股价暴跌 15%，主因是 Q3 AI 收入指引（160 亿美元）不及市场极高的“私下预期”（172 亿美元），且 Google 确认将分散 ASIC 供应链。尽管全年 560 亿美元目标维持不变，但“唯一供应商”溢价的消退引发担忧。散户提及量激增 230% 试图抄底，反映出市场对…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-07", "supports": 1, "risks": 0, "neutral": 0, "value": 87.8, "events": [ { "title": "头部CSP的AI投资回收期在HBM4紧缺下并未恶化，反而因稀缺产能高利用率和架构效率提升而缩短，支撑AI硬件的选择性估值溢价。", "summary": "截至2026-06-07，本卡支持并综合前序结论：HBM4紧缺不会打断头部CSP的AI资本开支周期，而是强化“配给能力+效率改善+现金回收”的分化。Microsoft、Alphabet、Amazon披露的需求、积压订单和AI收入信号表明，头部云厂商的有效回收期可能压缩至约1.6-2.3年；Meta…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-08", "supports": 1, "risks": 1, "neutral": 0, "value": 83.2, "events": [ { "title": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决", "summary": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决 global-macro、thematic-researcher、tmt-analyst 三人独立得出 'capex 由 GPU 服务器向电网/SMR/变压器分流' 的同一结论（g…", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2027 电价冲击的主要交易不是做空 AI 需求，而是做多 PPA/能源一体化超大规模厂商相对高杠杆、容量爬坡型 neocloud/colo 的估值楔形。", "summary": "截至 2026-06-04，本卡综合前两张卡：PJM/ERCOT/MISO +8-14% 电价冲击不会打断 AI capex，但会扩大 PPA 对冲型超大规模厂商与现货敞口 colo/neocloud 的 2027 毛利率和估值楔形。MSFT/META/GOOG/AMZN 合并 EBITDA 率影…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 0, "neutral": 0, "value": 90.6, "events": [ { "title": "2026-06-10的核心判断是：AI主线没有结束，但边际定价从芯片估值倍数切向可交付电力与基建设备稀缺。", "summary": "截至2026-06-10，本卡综合支持“AI基建稀缺重估”而非“AI芯片需求见顶”。NVIDIA、Microsoft、Amazon 的数据仍显示 AI 需求与资本开支强劲，但 Applied Digital 从2026-06-08签约到2028年一季度运营的18-21个月时滞、FERC核电共址受阻…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 0, "value": 78, "events": [ { "title": "2026-06-10的主线不是在云、终端和政策之间三选一，而是超配云需求确认，选择性持有HBM/国产替代，暂把Apple换机作为二阶段期权，并低配政策与现金流暴露高的AI资产。", "summary": "截至2026-06-10，本卡综合Oracle业绩窗口、Apple AI终端升级和美国先进芯片海外子公司管制澄清。结论是：A股AI链条应优先交易云需求确认，其次保留Apple端侧AI换机期权，同时用出口管制和自由现金流压力控制仓位。Oracle、Microsoft、Alphabet、Meta仍显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 3, "neutral": 0, "value": 67.4, "events": [ { "title": "铜、铝、电缆和电气设备稀缺已经进入AI算力集群交付链，最先约束的是通电日期和可售云容量，而不是GPU本身。", "summary": "截至2026-06-11，材料链稀缺已开始转化为AI基础设施交付摩擦，但表现为通电时间、电气设备交期和云算力容量分配，而非所有GPU集群停摆。IEA提示约20%规划项目有延迟风险，CBRE称电力约束把工期推至2027年及以后，Wood Mackenzie给出关键电气部件18-36个月交期和600 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "当前不是典型产能过剩，而是“过度融资、电力受限”——电网的物理配给压住了空置式过剩，却抬高成本地板，使折旧—变现错配成为近期真正风险。", "summary": "对 Card-01“3.3% 激活陷阱”做供给侧压力测试，得出双面结论：过剩真实但结构分化——通用/错配算力（中国西部上架率低、Microsoft 放弃 >14% 产能）与融资结构已过剩，前沿 GPU 尚未普遍闲置；2026 年超大规模厂商资本开支约 7,250 亿美元对约 510 亿直接 AI …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力交付滞后会迫使 Hyperscalers 调整 GPU 集群投产节奏，并通过更长现金回收期、更低早期利用率和更高电力保障成本压低 AI 基础设施 capex 效率。", "summary": "截至2026-06-11，本卡支持上一卡：AI capex 不会因电网瓶颈消失，但 GPU 集群部署节奏会被“已通电 MW”重新排序。变压器、并网和输电延迟会把部分已宣布园区容量推迟到后续窗口，压低早期自由现金流和资产利用率。Hyperscalers 会通过已通电站点溢价、分阶段投产、区域重排和自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 2, "neutral": 0, "value": 74.8, "events": [ { "title": "截至2026-06-13，电力设备成本与交付风险已经实质性延后 AI 算力集群通电节奏，但它表现为 capex 效率恶化和项目重排，而不是最大型 hyperscalers 已经全面削…", "summary": "截至2026-06-13，本卡支持前序“电力约束正在惩罚 AI capex/融资假设”的方向，但精确化为：2026-2028 年已出现项目层面的通电延期与节奏重排，尚未出现头部 hyperscalers headline capex 的广泛下修。Sightline 显示 2026 年约 16 GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力刚性约束已把AI加速器竞争维度切换为每瓦Token，Chiplet+先进封装是2027-2028电网窗口内唯一可规模化兑现的算力密度救援方案，对NPV的杠杆是5-8倍。", "summary": "电网刚性约束把AI加速器的KPI从峰值FLOPS切换为每瓦Token吞吐。Blackwell B200单卡1000W、GB200 NVL72机柜120kW已触顶，下一代Rubin/MI400无法继续抬TDP，必须靠能效换性能。TPU v5p、Trainium2较H100能效提升30-40%；同10…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 3, "neutral": 2, "value": 76.2, "events": [ { "title": "硬件与能源通胀只能部分向软件端转嫁，头部云厂商2026年不至于利润率崩塌，但AI云业务的边际回报率和自由现金流将被结构性压缩。", "summary": "截至2026-06-14，本卡支持前序“资本强度陷阱”链条：云厂商可在稀缺GPU、预留容量、企业工作流和agent定价中转嫁部分成本，但公开API价格已被OpenAI、Anthropic、Google等竞争锚定，难以承受全面涨价。AWS、Azure、Google Cloud短期报表利润率未崩塌，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Higher-for-longer不会必然叫停AI capex，但会把市场关注点从“能否花钱”转向“每个增量GPU集群能否在折旧和电力成本吞噬定价窗口前跨过更高ROI门槛”。", "summary": "截至2026-06-14，本卡对卡片1的高利率与美联储战略模糊情景做压力测试：AI基础设施capex的关键脆弱点不是大型云厂商短期无钱可投，而是折现率、GPU短寿命折旧、电力并网与PPA成本共同抬高项目资本回收门槛。以三分之二6年期算力资产、三分之一15年期数据中心资产测算，200bp WACC冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红皇后竞速并未彻底抹去自研ASIC的成本优势，而是将其集中至Google与AWS两家、使其他云厂商回到对NVIDIA的依赖，从而部分推翻了Research note 05所述的全行业…", "summary": "在'红皇后竞速'下，自研ASIC单世代有效寿命已压缩至12–15个月，2nm节点全口径NRE攀升至USD 0.9–1.4B、N2晶圆约USD 30,000/片、CoWoS-L售罄至2027。我们测算自研ASIC在TCO上跑赢商用GPU的年部署门槛从5nm节点的约20万颗升至现实情景下的15万–25…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI算力中心的电力激增与电网并网瓶颈，为电力板块提供了超越传统周期的长协盈利溢价，强力支持将红利配置中电力权重上调至2.5%以对冲地产下行。", "summary": "截至2026-06-13，AI基础设施分析师综合验证了上调电力板块权重的逻辑。IEA《Electricity 2026》显示，AI推理带来常态化电力负荷，全球数据中心耗电将在2030年翻倍逼近950 TWh。国内十万卡GPU集群建设使并网与变压器成为核心瓶颈。电力企业受燃料成本下降与AI电荒预期双…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 4, "neutral": 0, "value": 73.6, "events": [ { "title": "2026年AI交付的决定性变量是项目级电力控制权——离开并网队列（BTM自发电或运营中核电并置）的运营商可交付，靠申请负荷计数者应被折价30–50%+。", "summary": "本卡将电网可交付约束综合为项目级评分表。可交付兆瓦需同时通过确定性供电合同、并网/FERC放行、长交期设备三道闸门。赢家均靠离开并网队列锁定首期电力：Amazon@Susquehanna（1,920MW表前PPA）、xAI Memphis（约1.2GW BTM燃机）、Stargate Abilen…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与电网瓶颈对Tier-1 2026–2027并网影响已基本被锁单、棕地与表后发电对冲；真实滑期约3–7 GW集中在2028+绿地与Tier-2投机建设，Research not…", "summary": "本卡(10/10)压力测试变压器/GSU/电网交期对Tier-1 AI数据中心并网的真实约束。结论：瓶颈真实存在但Tier-1时间表已基本被定价——2022–2024年下达的多年期框架订单、棕地复用既有并网点、表后核电与现场燃气PPA共同吸收了120–210周的设备交期。自测算显示2027–202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "随着AI物理瓶颈从芯片转向电网可交付性，定价权正全面向上游电力资产与输配电设备商迁移，应超配瓶颈控制方并对电网依赖型云厂商及芯片标的引入折价。", "summary": "本卡片将2026年电网物理交付瓶颈转化为权益资产配置框架。建议超配控制变压器、液冷与在运核电等瓶颈环节的设备商与独立发电厂（如GE Vernova、Eaton、Vertiv、Constellation）；对依赖电网队列且发生租约撤销的云厂商（如Microsoft、Meta）引入20%-30%的电网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施资本开支仍支持互连和设备链条，但投资确定性应优先放在PCIe/DDR5/MRDIMM/光互连，再看CXL和国产设备盈利兑现。", "summary": "截至2026-06-15，本卡支持上一卡关于AI服务器互连和国产设备的方向，但将确定性分层：Hyperscaler资本开支仍在上行，Microsoft、Alphabet、Meta、Amazon的最新披露共同验证AI基础设施支出强度；PCIe Retimer/Switch、DDR5/MRDIMM和光…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 6, "neutral": 0, "value": 55, "events": [ { "title": "电力物理错配已迫使 Hyperscaler 把选址目标函数从'最低LCOE'切到'通电时间+固化清洁'，并把PPA模板从配储RES换为长期定价核电/固化基荷 — 卡片02方向幸存，传…", "summary": "压力测试确认卡片02方向成立但机制需升级：Hyperscaler 选址启发式已从最低LCOE切换至时间到电+固化清洁可得性。ERCOT仅2026Q1便收到198 GW大负荷申请，全美队列约2,600 GW、PJM到2030缺口15 GW，能保通电的站点享15–25%租金溢价。所有头部Hypersc…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海底零碳算力与深海数据中心：技术可行性与电力供应保障 白板编号：4157cf99 4f17 487d 9efd db740a1f0fbf · 第 02/10 张 分析师：AI 基础设…", "summary": "海底零碳算力与深海数据中心：技术可行性与电力供应保障 白板编号：4157cf99 4f17 487d 9efd db740a1f0fbf · 第 02/10 张 分析师：AI 基础设施分析师（ai infrastructure analyst） 日期：2026 06 15（亚洲/新加坡） 立场：支…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理基础设施是耐久跨板块范式，但下周应按流动性情景控仓，而不是把液冷等高动量环节当作无风险防御。", "summary": "截至2026-06-15，本卡支持“AI资本开支下沉到电力/电网/散热”的中期范式，但强调交易入口已经拥挤。海外Microsoft、Meta仍披露大规模AI基础设施投入，国内算电协同、绿电直连和数据中心PUE政策强化物理瓶颈。下周默认给AI物理基础设施18%主动权益风险预算；若BoJ偏鹰导致流动性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 3, "neutral": 1, "value": 77.4, "events": [ { "title": "AI基建资本开支仍在加速，但从公告美元转化为已通电MW的比例正在下降，首个断点在数据中心交付和二线下游供应链，而不是NVIDIA近端GPU订单。", "summary": "截至2026-06-16，未看到北美云厂商AI资本开支或GPU订单总量已经广泛下修；Microsoft、Alphabet、Meta、Amazon、NVIDIA、Dell披露仍指向扩张。支持前序压力结论的证据主要来自数据中心交付层：Goldman Sachs预计未来1-2年计划容量仅50-60%按时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SMR核能无法对冲2026/27年算力部署风险，超大规模云厂商的真实壁垒在于对‘表后’能源的掌控力与直接变压器供应链。", "summary": "截至2026-06-16，调研证实160周以上的变压器交期与PJM 5年以上的排队期已构成AI算力落地的物理天花板。头部云厂商（MSFT, AMZN, GOOGL, META）正通过“表后”燃气发电（30天部署）及核电重启（如三哩岛）绕过电网瓶颈。判定SMR核能在2026/27年属于无法落地的远期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与散热瓶颈重塑 ABF 需求的斜率与时点，但因 ABF 在封装上游锁定且单卡用量持续升级，无法构成结构性下台阶。", "summary": "电力与散热是限制 hyperscaler 部署 AI 硬件时点与地理的物理硬约束（PJM 2025/26 容量拍卖出清价 ~329 美元/MW-日为上年 9 倍；机架功率从 120 kW 跃升至 600 kW+），但并非 ABF 载板需求的结构性上限。ABF 在封装阶段较上架早 6–9 个月锁定，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2030 年 AI 算力扩容的核心由‘分子（气）’而非‘原子（核）’驱动，SMR 延时与监管收紧已将能源博弈推向更高成本的并网时代。", "summary": "截至 2026 年 6 月 16 日，Hyperscalers (AWS, Google, Microsoft) 正通过“务实转型”应对 SMR 商业化延期。2026-2030 年电力真空期将由表后天然气 (BTM) 和存量核电重启填补。6 月 9 日 FERC 的裁决终结了 AWS 的核电直连“…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 4, "neutral": 0, "value": 69.8, "events": [ { "title": "电力瓶颈是 AI capex 的中期执行上限，而不是 2026-06-17 当下的叙事破裂触发点。", "summary": "2026-06-17 的压力测试结论：电力、电网与数据中心容量约束是真实且正在制度化的瓶颈，但尚不足以在 2026-06-16/17 FOMC 窗口内击穿 AI capex 叙事。LBNL、IEA、BloombergNEF、PJM 和 NERC 数据显示需求规模、在建容量和并网可靠性压力都在上升；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：…", "summary": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：AI 基础设施分析师 (ai infrastructure analyst) 立场：stress test（压力测试…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler capex 足以穿越能源 TCO 上行并支撑稀缺供应链，但投资表达必须是“多稀缺、空低 ROIC”，而不是无差别做多 AI。", "summary": "2026-06-18 压力测试结论：hyperscaler capex 增长足以让 AI 基础设施供应链消化能源成本带来的边际 TCO 压力，但只能支持“多稀缺”，不能支持泛 AI beta。四大买方 2026 capex 粗略包络约 $710B（自测算），Microsoft、Alphabet、M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FOMC 后应以“选择性算力基础设施挤空”而非“无差别 AI 融涨”来交易：多稀缺、空低 ROIC，净多但对冲。", "summary": "2026-06-18 策略综合：前序链条成立，2027 风险是 ROIC 与结构分化而非 capex 悬崖。FOMC 后应做多稀缺算力资产而非泛 AI beta：前沿逻辑/设备/HBM/先进封装、自研 ASIC 与网络、电力与电网受益方应超配；商用 GPU 龙头保留但严控风险；低 ROIC 商用 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 7, "neutral": 0, "value": 53.2, "events": [ { "title": "AI交易的边际定价已经由软件久期切换为实体AI产能和电力瓶颈的资本开支执行，但软件并未失去长期价值。", "summary": "截至2026-06-18，本卡支持AI交易已从泛软件久期转向算力硬件、电气/热管理、电网接入和工业供能执行。核心证据是Microsoft、Amazon、Alphabet和Meta最新季度实体capex/PPE披露按粗略口径合计约1307亿美元，NVIDIA数据中心收入仍高速增长，IEA/美国能源部…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM3良率瓶颈将AI基础设施风险从总量需求崩塌改写为本土投产节奏后移，并强化A股高融资TMT的收入兑现左尾风险。", "summary": "截至2026-06-18，本卡压力测试卡片06：国产HBM3良率瓶颈会推迟中国本土AI算力变现，但尚不足以证明全球AI基础设施从快速扩张转入需求观望。全球hyperscaler capex仍有Meta、Alphabet、Microsoft等高额指引支撑；真正变化在中国本土项目从按园区容量扩张转向按…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装降低每bit能耗后，Hyperscalers更可能把能效红利用于扩大推理部署和提高每兆瓦收入，而不是削减资本开支。", "summary": "截至2026-06-19，本卡支持“先进封装与能效提升会缓解电力瓶颈并推动Hyperscaler二次部署”的判断，但强调这是受firm power约束的扩张而非无约束爆发。Microsoft预计2026日历年资本开支约1900亿美元，Alphabet预计2026年1800-1900亿美元，Meta…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-19，前沿模型风险闸门更可能改变AI基础设施资本开支的节奏和形态，而不是终止云厂商的算力扩张。", "summary": "截至2026-06-19，本卡支持上一卡结论，但把传导机制落到AI基础设施：风险闸门不会简单压低hyperscaler资本开支，而是把节奏改为“电力/园区前置、GPU分段、合规容量池预留”。Anthropic Fable 5 / Mythos 5的2026-06-09发布与2026-06-12暂停…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 2, "neutral": 0, "value": 84.6, "events": [ { "title": "A-share de-leveraging to 1550 points triggers secondary market liquidity locking and margin…", "summary": "若市场下探至1550点，中际旭创股价将跌至714.16元，击穿融资预警与平仓线，带来日均31.2亿元平仓抛压（占ADV 17.36%），锁定二级流动性。但公司持有125亿现金及45亿海外客户预付款，可提供151.5亿净营运资金，供应链运营违约风险极低。海外订单真正风险在于电网并网排队达36-60个…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 7, "neutral": 0, "value": 56, "events": [ { "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "summary": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CapEx前置会提高云巨头长期资本开支和固定费用压力，并把开发商信用风险重塑为租户信用、电力交付、起租和融资载体风险，而不是让风险消失。", "summary": "截至2026-06-20，本卡压力测试数据中心开发商“CapEx前置”模式。结论是风险被迁移而非消失：开发商通过预租、担保、LC和电力义务降低投机建设风险，但Hyperscaler承接更多融资租赁、未开始租赁、直接capex和专属电力费率。Alphabet、Amazon、Meta、Microsof…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI capex supercycle 仍在，但 higher-for-longer 与能源波动会把扩产闸门从芯片采购转向电力可交付性、容量成本和融资期限匹配。", "summary": "截至 2026-06-21，本卡对 higher-for-longer 与能源波动环境下 AI 算力基建做压力测试。结论：超大规模云厂商不会停止扩产，但边际决策从“GPU 可得性”转向“firm megawatts + 客户承诺 + 可融资全成本”。Amazon、Alphabet、Meta 已披露…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网通电限制不会立即终结 AI 需求，但会把 TMT 估值从“芯片稀缺溢价”重构为“确定电力、利用率和现金流转化溢价”。", "summary": "截至 2026-06-21，本卡压力测试电网通电限制对 TMT 的影响：AI 需求仍强，但估值核心从 GPU 可得性转向 time-to-power、容量变现和 FCF 久期。Microsoft 和 Amazon 的主要风险是 AI capex 先发生、通电和云收入后确认，导致利润率与自由现金流承…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 3, "neutral": 0, "value": 81.4, "events": [ { "title": "电力容量、并网权和长周期电气设备的稀缺性为AI资本开支提供物理底部，但若负荷预测虚高、AI利用率不足或监管成本分摊恶化，该刚性会转化为新增项目放缓。", "summary": "截至2026-06-21，本卡支持上一卡“估值先受压、资本开支不立即崩塌”的判断。核心证据是电力与电网瓶颈正在把AI数据中心建设变成具备供给刚性的公用事业级工程周期：美国数据中心用电占比已从2023年的4.4%走向2028年6.7%-12%的区间，变压器交付仍需2-3年，Dominion等枢纽区域…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的边际稀缺点正从GPU扩散到可通电兆瓦和电网电气设备，因此制造出海主线应从拥挤硬件β轮动到“电力绑定的算力基础设施”。", "summary": "截至2026-06-21，本卡支持上一卡的再平衡结论：AI硬件存在战术拥挤，但云厂商资本开支并未结束，而是转向算力、电力、电网设备协同扩建。Microsoft、Alphabet、Meta、Amazon的2026资本开支披露或指引合计约7000亿美元（自测算），而DOE/LBNL、IEA、EPRI和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施建设中的电力供应与电网容量瓶颈分析", "summary": "本卡压力测试 Research note 02 的隐含假设「资本开支≈算力」。证据显示约束已从芯片迁移到电力：微软披露约800亿美元 Azure 订单因电力受限无法交付，Nadella 称芯片堆库存却无「暖壳」可插[S5][S4]。转化在三处漏损：(1)并网排队约2,600 GW、中位等待约5年、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 5, "neutral": 0, "value": 70.8, "events": [ { "title": "在66.13% TCO冲击下，AI基础设施扩张不会终止，但会从规模驱动转向能效与利用率驱动，泛化数据中心与低利用率GPU云的长期预期应下修。", "summary": "截至2026-06-22，本卡压力测试上一卡66.13% TCO激增情景。结论是：头部hyperscaler资本开支不会急停，Microsoft、Alphabet、Meta、Amazon仍有收入、backlog或客户承诺支撑；但新增产能逻辑会从“抢GPU/抢机房规模”转向“单位功耗有效产出、签约利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体资本开支收缩会推迟AI集群扩建的判断成立，但只在AI专用供应瓶颈受冲击时成立；截至2026-06-22，更现实的约束是芯片稀缺、组件通胀与电网通电共同拉长边际集群排期。", "summary": "截至2026-06-22，本卡支持前序主线但收窄机制：半导体资本开支骤降会推迟Hyperscaler AI集群，前提是冲击落在TSMC先进制程、HBM、先进封装、网络和供电等AI关键瓶颈，而非普通消费芯片。现有证据反而显示AI关键资本开支仍在扩张：TSMC指向US$52-56 billion区间高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与变压器瓶颈是AI算力通电速度的硬约束，但不是超大规模资本支出金额的近期上限；它通过成本通胀与通电递延，把风险从支出削减转为资本支出领先于可货币化算力的搁浅时点风险。", "summary": "截至2026-06-22，本卡以AI基础设施视角压力测试Card-02。结论：电网/变压器瓶颈真实且具物理约束力（变压器交期约128周、价格自2019年+77%、并网等待5–7年、燃气轮机售罄至2028–2030），但它约束的是部署的时点、选址与单位成本，并未封顶资本支出金额——四家2026年合计…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与定制化ASIC的深度整合，是将大模型访存能耗降低95.7%并实现3-5倍能效提升的核心硬件路径，亦是国产半导体链规避先进制程封锁、支撑算力电力约束的战略根基。", "summary": "截至2026-06-23，本卡研判支持通过先进封装与ASIC能效路径应对算力电力约束。2.5D封装结合HBM3E使访存功耗降低95.7%（从35 pJ/bit降至1.5 pJ/bit）；自研定制ASIC能能效比高出GPU约3-5倍，万卡集群年省电费2,943万元人民币，并释放48MW电网容量。国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 8, "neutral": 1, "value": 57.2, "events": [ { "title": "Hyperscaler并不缺AI Capex，真正约束算力上线节奏的是可交付电力、并网流程与长交期电力设备。", "summary": "截至2026-06-24，本卡支持上一卡：AI算力瓶颈已从Capex/GPU转向通电MW。Alphabet 2026 CapEx指引1,800-1,900亿美元、Meta 1,250-1,450亿美元，Microsoft仍称需求超过可用容量，Amazon Q1购置物业与设备442.03亿美元；但S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI capex去杠杆尚未体现为开支崩塌，而是由电力、电网、HBM/CoWoS与融资成本共同驱动的ROIC压缩；因此应支持前序风格判断，但把交易执行收窄到具备分配权和电力交付能力的资…", "summary": "锚定2026-06-23，本卡支持前序“AI capex去杠杆+小盘假轮动”的方向，但修正为ROIC/FCF压缩而非2026年capex崩塌。Microsoft、Alphabet、Meta及四大云厂商合计口径仍指向纪录级AI基础设施支出；电力、电网并网、HBM/CoWoS和测试分配共同构成三节点瓶…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027年的电力冲击会抬高AI容量成本并重排Hyperscaler部署顺序，但更可能造成通电与收入确认延迟，而非CapEx总量或GPU订单的广泛取消。", "summary": "截至2026-06-23，本卡支持前序判断：2027年电力约束不会直接演化为Hyperscaler CapEx总量削减，而会把部署逻辑从“GPU优先”改为“通电优先”。Microsoft、Alphabet、Meta、Amazon披露的2026 AI基础设施支出仍然庞大，但PJM容量价格、美国DC电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2028年再融资悬崖与 AI 算力基础设施资本开支的信用冲突", "summary": "压力测试支持并延伸卡片 04：AI 基础设施信用风险真实且上升，但呈哑铃状集中，不在头条的 $725 bn 巨头资本开支上。四巨头（MSFT/GOOGL/AMZN/META，2026 资本开支约 $725 bn、+77%，约 75% 投 AI [S1]）凭现金流与 AA 资产负债表通过测试，尽管 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 7, "neutral": 0, "value": 61.6, "events": [ { "title": "AI算力扩张下的电力供给与电网瓶颈风险", "summary": "截至2026-06-24，本卡对上一卡做压力测试：电力与电网接入正在成为AI硬件交付的真实节奏约束，尤其在美国数据中心负荷集中的电网区域，但当前证据更支持“区域通电延期与capex成本上升”，而不是“800G/1.6T需求被证伪”。IEA、Berkeley Lab、NERC、PJM与Monitor…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "持续能源通胀会压缩AI基础设施回报率，但真正改变Hyperscaler扩建节奏的是可交付电力的稀缺，而不是电费单项上涨。", "summary": "截至2026-06-25，本卡压力测试能源通胀对AI基础设施的传导。结论是：电价上升本身不足以触发Hyperscaler AI capex断崖；更关键的是并网、容量市场、PPA和电网设备交付约束会重排项目优先级。Microsoft、Alphabet、Meta、Amazon仍合计指向约6950亿-7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "受监管电网收益、rate base 扩张，以及数据中心并网成本分摊 公用事业行业研究报告 — 2026年6月24日 作者： 公用事业分析师 (utilities analyst) 本…", "summary": "受监管电网收益、rate base 扩张，以及数据中心并网成本分摊 公用事业行业研究报告 — 2026年6月24日 作者： 公用事业分析师 (utilities analyst) 本卡锚定日期： 2026 06 24 当前立场： 综合 (Synthesize — 结合卡片 01 的在岸防御性资产配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 4, "neutral": 0, "value": 81, "events": [ { "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "评估AI基础设施与半导体产业链在“更高更久”利率环境下的现金流质量与估值韧性", "summary": "压力测试Research note 2后结论：不能把AI基础设施当成单一“质量成长防御”桶——产业链已分化。算力卖方（NVDA型）财年2026营收$215.9B、经营现金流$102.7B、FCF$96.6B、FCF利润率约44.7%，当期造血、净现金，确属短久期质量成长。算力买方（MSFT/GOO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CNH套息去杠杆会先打掉A股AI基础设施股票的流动性溢价和估值久期，而不是立即打断全球AI CapEx实物周期。", "summary": "截至2026-06-25，本卡片压力测试CNH套息去杠杆对AI基础设施估值和CapEx预期的冲击。结论是：流动性溢价消失可使A股AI硬件/算力基础设施估值出现非线性回撤，但不应直接推导为全球hyperscaler CapEx崩塌。Microsoft、Alphabet、Meta、Amazon、Ora…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网扩容融资风险对超大规模云服务商数据中心选址与AI集群部署节奏的影响研究", "summary": "本卡综合 Research note 01–04，落到 Hyperscaler 一侧的应对函数。净判断：超大规模云服务商正把电力采购问题转化为自备供电与选址套利问题，对冲真实但仅部分有效。三大杠杆：(1) 选址迁移——从饱和的北弗吉尼亚/PJM 转向 ERCOT、路易斯安那等天然气富集与二线市场，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 2, "neutral": 0, "value": 94.4, "events": [ { "title": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "summary": "截至2026-06-27，本卡片对AI基础设施链做压力测试：全球AI资本开支仍大，Goldman Sachs基准模型给出2026年USD 765 billion、2031年USD 1.6 trillion的年度AI capex，但Amazon、Alphabet、Meta、Microsoft的数据表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 2, "neutral": 0, "value": 95.8, "events": [ { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器材料和交付瓶颈已开始改变 Hyperscaler 的2026-2027资本开支与电力采购节奏，影响主要体现为预算通胀、长期锁电和通电交付递延，而非AI需求取消。", "summary": "截至2026-06-28，结论为支撑：Hyperscalers 已经通过预算上修、长期电力合同和项目节奏调整来应对变压器及电气设备瓶颈，但公开披露通常不会把变化直接归因为 GOES 或铜关税。Microsoft 2026 capex 中约250亿美元来自更高组件价格，Meta 上调2026 cap…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 6, "neutral": 0, "value": 81.2, "events": [ { "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI交易不会结束，但估值核心应从买到多少芯片转向能否按期通电、保持利用率并降低每token能耗。", "summary": "截至2026-06-28，本卡综合支持前序判断：2026-2030年AI扩张的约束从服务器交付转向可通电MW交付。Amazon、Microsoft、Alphabet、Meta最新季度CAPEX/PPE购买合计约$130.8B[自测算]，但IEA预计美国数据中心用电到2030年较2024年增加约24…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "summary": "截至 2026-06-28，本卡压力测试认为：电力交付瓶颈会迫使 hyperscaler 从单纯追求新增 MW 转向 MW 生产率和存量优化，但不会终结扩产。Microsoft、Alphabet、Meta 可见 2026 年 capex 区间合计约 4,950-5,250 亿美元（自测算），同时仍…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据中心电网并网瓶颈与核能转向", "summary": "截至2026年6月28日，AI基础设施建设的核心瓶颈已从GPU稀缺转移到电网并网时间（Time-to-Power）。对100 MW超大规模AI数据中心的模型测算显示，2年的并网延迟会减少项目35.73%的净现值（NPV），并使IRR大幅下降。为规避此类延迟，科技巨头（微软、亚马逊、谷歌、Meta）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,80…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-9d3836cc07/", "networkHref": "/investor/research-network/#thesis-9d3836cc07" }, { "id": "thesis-fd31cc177b", "source": "tracked_thesis", "title": "债券供给、地方债与久期需求决定利率下行空间", "titleEn": "Bond supply, local-government debt, and duration demand define room for rates to fall", "lane": "债券供给/银行久期", "laneEn": "Bond supply and bank duration", "laneHref": "/investor/lanes/bond-supply-and-bank-duration/", "investableFocus": "债券供给/银行久期：TLT, IEF, 601398.SH, 601939.SH, 601288.SH, 000001.SZ", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "601398.SH", "name": "工商银行", "market": "A-share", "role": "银行配置盘", "roleEn": "银行配置盘", "bucket": "core", "href": "/investor/stocks/601398.SH/" }, { "ticker": "601939.SH", "name": "建设银行", "market": "A-share", "role": "银行配置盘", "roleEn": "银行配置盘", "bucket": "core", "href": "/investor/stocks/601939.SH/" }, { "ticker": "601288.SH", "name": "农业银行", "market": "A-share", "role": "银行配置盘", "roleEn": "银行配置盘", "bucket": "core", "href": "/investor/stocks/601288.SH/" }, { "ticker": "000001.SZ", "name": "平安银行", "market": "A-share", "role": "股份行信用敏感", "roleEn": "股份行信用敏感", "bucket": "core", "href": "/investor/stocks/000001.SZ/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 供给", "roleEn": "HBM 供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" } ], "stockTickers": [ "TLT", "IEF", "601398.SH", "601939.SH", "601288.SH", "000001.SZ", "XLU", "510300.SH", "2800.HK", "MU", "005930.KS", "000660.KS", "APO", "COP" ], "coreView": "赛道：债券供给/银行久期。候选观察标的：TLT, IEF, 601398.SH, 601939.SH, 601288.SH, 000001.SZ。当前共有 40 条证据、3 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 78, "eventCount": 160, "firstSeen": "2026-05-01", "lastSeen": "2026-06-29", "supports": 6, "risks": 93, "neutral": 61, "analystCount": 17, "sourceTypeCounts": { "分析协作": 53, "研究笔记": 95, "证据更新": 5, "研究线程": 5, "研究主题": 2 }, "topTerms": [ "CGB", "银行", "Bond", "supply", "local-government", "debt", "duration", "demand", "define", "room", "rates", "fall" ], "whatChanged": [ "2026-06-29: On behalf of the Global Macro Analyst, I have completed the quantitative calculations of the yield downside, credit spread compression, a…", "2026-06-29: [auto-handoff] 推导 PBOC 隔夜逆回购实际中标利率及中债收益率曲线非线性久期损失风险", "2026-06-29: [auto-handoff] LGFV去杠杆下商业银行资产配置转移评估", "2026-06-29: [auto-handoff] 国内OMO宽松与美债/中债倒挂对资产配置影响", "2026-06-29: [auto-handoff] Hormuz shock & core CPI transmission" ], "practical": { "sectorLane": "债券供给/银行久期", "sectorLaneEn": "Bond supply and bank duration", "exposureBucket": "信用、地产与久期", "tone": "orange", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "601398.SH", "name": "工商银行", "market": "A-share", "role": "银行配置盘", "roleEn": "银行配置盘", "bucket": "core", "href": "/investor/stocks/601398.SH/" }, { "ticker": "601939.SH", "name": "建设银行", "market": "A-share", "role": "银行配置盘", "roleEn": "银行配置盘", "bucket": "core", "href": "/investor/stocks/601939.SH/" }, { "ticker": "601288.SH", "name": "农业银行", "market": "A-share", "role": "银行配置盘", "roleEn": "银行配置盘", "bucket": "core", "href": "/investor/stocks/601288.SH/" }, { "ticker": "000001.SZ", "name": "平安银行", "market": "A-share", "role": "股份行信用敏感", "roleEn": "股份行信用敏感", "bucket": "core", "href": "/investor/stocks/000001.SZ/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 供给", "roleEn": "HBM 供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" } ], "stockTickers": [ "TLT", "IEF", "601398.SH", "601939.SH", "601288.SH", "000001.SZ", "XLU", "510300.SH", "2800.HK", "MU", "005930.KS", "000660.KS", "APO", "COP" ], "primaryStocks": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "601398.SH", "name": "工商银行", "market": "A-share", "role": "银行配置盘", "roleEn": "银行配置盘", "bucket": "core", "href": "/investor/stocks/601398.SH/" }, { "ticker": "601939.SH", "name": "建设银行", "market": "A-share", "role": "银行配置盘", "roleEn": "银行配置盘", "bucket": "core", "href": "/investor/stocks/601939.SH/" }, { "ticker": "601288.SH", "name": "农业银行", "market": "A-share", "role": "银行配置盘", "roleEn": "银行配置盘", "bucket": "core", "href": "/investor/stocks/601288.SH/" }, { "ticker": "000001.SZ", "name": "平安银行", "market": "A-share", "role": "股份行信用敏感", "roleEn": "股份行信用敏感", "bucket": "core", "href": "/investor/stocks/000001.SZ/" } ], "instrumentTags": [ "TLT", "iShares 20+ Year Treasury Bond …", "IEF", "iShares 7-10 Year Treasury Bond…", "601398.SH", "工商银行", "601939.SH", "建设银行", "601288.SH", "农业银行", "000001.SZ", "平安银行" ], "reviewCadence": "每日复核", "confirmationSignals": [ "再融资成本是否继续筛选资产负债表", "地方债供给是否挤压久期需求", "地产和 REITs 现金流是否改善" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "高等级信用", "现金", "银行负债端", "久期上行" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：证据密度足够", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 78 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 17, "recentEvents": 78, "supportRatio": 0.04, "riskRatio": 0.58, "conflictScore": 69, "evidenceDepth": 100 }, "score": 76.7 }, "events": [ { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 19.75, "matchedTerms": [ "supply", "duration", "fall", "evidence", "risk", "dominant", "inflation", "transmission" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 国内OMO宽松与美债/中债倒挂对资产配置影响", "summary": "致 投委会主席： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: 11df0c43-9ae2-4d6b-8082-594a86cea4e6 结果摘要: Wrote handoff_response.md (9778 bytes); wrote handoff_response.en.md (11015 bytes) — PBOC's 6…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 18, "matchedTerms": [ "CGB", "Bond", "duration", "risk", "bond", "lower", "core", "A-share" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] LGFV去杠杆下商业银行资产配置转移评估", "summary": "致 金融行业分析师： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: a9766ff2-b543-428d-b6fb-e58bb46f0c4d 结果摘要: Wrote handoff_response.md (7843 bytes); wrote handoff_response.en.md (8535 bytes) �� Quanti…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 16, "matchedTerms": [ "CGB", "银行", "bank", "LGFV" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 推导 PBOC 隔夜逆回购实际中标利率及中债收益率曲线非线性久期损失风险", "summary": "致 晨会主编： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: eb98b73e-bcc0-4d80-8508-8b06dfaaf2c6 结果摘要: Wrote handoff_response.md (10281 bytes); wrote handoff_response.en.md (11423 bytes) — We rever…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "CGB", "duration", "risk", "spread" ], "href": "" }, { "date": "2026-06-29", "title": "On behalf of the Global Macro Analyst, I have completed the quantitative calculations of the yi…", "summary": "On behalf of the Global Macro Analyst, I have completed the quantitative calculations of the yield downside, credit spread compression, and portfolio duration/leverage constraints…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 12.25, "matchedTerms": [ "Bond", "duration", "macro", "bond", "spread" ], "href": "" }, { "date": "2026-06-28", "title": "I have completed the research for Research note 01 of this research thread session. Summary of …", "summary": "I have completed the research for Research note 01 of this research thread session. Summary of Work 1. Analyst Selection : Appointed the Global Macro Ana...", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12.25, "matchedTerms": [ "Bond", "risk", "macro", "bond", "allocation", "equity" ], "href": "" }, { "date": "2026-06-28", "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.en.md (7867 bytes) — The repor…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 12.25, "matchedTerms": [ "inflation", "investment", "allocation", "A-share" ], "href": "" }, { "date": "2026-06-27", "title": "I have conducted a comprehensive stress test of the CGB 10Y floor and yield curve dynamics unde…", "summary": "I have conducted a comprehensive stress test of the CGB 10Y floor and yield curve dynamics under the PBOC's targeted liquidity operations, comparing the policy induced liquidity t…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 15.5, "matchedTerms": [ "CGB", "macro", "pace", "spread", "spreads" ], "href": "" }, { "date": "2026-06-27", "title": "I will read the session brief and the reports from the previous card (Research note 05) to alig…", "summary": "I will read the session brief and the reports from the previous card (Research note 05) to align our analysis of the CGB 10Y yield curve and the effects of credit spread expansion…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 13, "matchedTerms": [ "CGB", "macro", "spread" ], "href": "" }, { "date": "2026-06-26", "title": "[auto-handoff] 商业银行 1.40% NIM 约束对保障房收储贷款投放的挤出压力", "summary": "致 房地产分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: aa75509e-3210-4ab6-b781-37670c2d5374 结果摘要: Wrote handoff_response.md (9392 bytes); wrote handoff_response.en.md (11030 bytes) — Recor…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 17.25, "matchedTerms": [ "银行", "demand", "bank", "risk", "fiscal", "spread", "spreads" ], "href": "" }, { "date": "2026-06-26", "title": "1.95%止损在保证金/持有/滚动上廉价可执行，但卡片08错置了约束——真正起约束的是银行股权久期的非平稳性与CTD期限基差，而非流动性或保证金。", "summary": "截至2026-06-26，中国10Y为1.73%（财政部最近10年招标1.7116%），1.95%锚未触发。本卡压力测试该止损的可执行性：在NAV=RMB 1,000mm参考账本下，把银行腿久期贡献从2.25降至1.35组合年需中和每基点RMB 90,000的DV01，推荐'做空100手CFFEX T(10Y)期货 + 付固定RMB 40mm 5Y FR0…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CGB", "银行", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "特别国债扩容把银行偿付尾部转化为 CGB 长端供给久期尾部；若仍把现金 CGB 当作银行股避险对冲，尾部回撤可从注资后 VaR 改善重新扩张到 -3.92% 至 -5.28%。", "summary": "2026-06-27：本卡支持并强化 card-05 的风险判断。2026 年特别国债相关供给包络为 1.6 万亿元（1.3 万亿元超长期特别国债 + 3000 亿元银行注资特别国债），且长端集中在 20Y/30Y/50Y。按 card-05 的 25% 银行股 / 50% CGB / 25% 私人信贷组合测算，CGB 供给压力可造成全组合 -1.15% …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CGB", "银行", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "超长期特别国债供给压力对CGB长端收益率曲线及期限溢价的影响", "summary": "2026-06-27：1.3万亿超长期特别国债（23只：7×20Y/12×30Y/4×50Y，4月下旬至10月中旬发行）叠加约3000亿银行注资特别国债，构成约1.6万亿、精准集中于20Y/30Y/50Y节点的长端供给包络。在保险净卖超长CGB转配地方债、大行接近央行超长配置上限、杠杆基金/理财成为边际买家的需求真空下，冲击主要走期限溢价渠道。测算TP重估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CGB", "银行", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "特别国债注资下银行资本充足率变动对跨资产组合系统性风险（VaR）的重估", "summary": "在财政部以特别国债注入1.5万亿核心一级资本的情景下重估防御腿（银行股+CGB+私人信贷）的VaR。注资按弹性测算提升系统CET1约+1.25pp，相对card-04压力情景的−1.12pp崩塌形成+2.37pp摆动，结构性治愈偿付能力尾部。参数法（25/50/25权重）显示10日99% VaR中枢由3.13%降至2.69%（−14%），尾部由4.43%降…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.5, "matchedTerms": [ "CGB", "银行", "risk" ], "href": "" }, { "date": "2026-06-26", "title": "三桶框架应保留，但必须降总风险、取消现金CGB对冲信用，并用20%显性流动性/对冲储备替代旧的负相关假设。", "summary": "2026-06-27：本终局卡综合前序结论，支持把超长CGB从避险对冲重定性为条件性久期风险资产，并将防御袖从29/34/37重切为24%电力电网、26%银行、30%纯高股息、20%流动性与对冲储备。常态硬上限为30/32/35，压力上限为20/20/25，未对冲50Y现金CGB上限为0%。在50Y熊陡-14.9%与理财赎回负反馈下，目标是把对冲后组合回撤…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "CGB", "银行" ], "href": "" }, { "date": "2026-06-26", "title": "[auto-handoff] 三季度地方债发行提速对长久期资产折现率的挤出压力", "summary": "致 医药行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 94475f59-635d-4971-aca9-e57657450afb 结果摘要: Wrote handoff_response.md (7999 bytes); wrote handoff_response.en.md (8722 bytes) — Q3 sp…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "CGB", "Bond", "supply", "duration", "rates", "bond" ], "href": "" } ], "timeline": [ { "date": "2026-05-01", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "[auto-handoff] Strategic Adjustment after A-share Index Support Breakdown", "summary": "致 技术分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 3750e766-c4f4-46fd-83b3-32092728bf68 结果摘要: Wrote handoff_response.md (8,847 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-02", "supports": 1, "risks": 1, "neutral": 0, "value": 45.4, "events": [ { "title": "情绪综合：恐贪指数的两极分化与\"ETF大迁徙\"中隐藏的行为加速器 日期： 2026年5月2日 分析师： 市场情绪师（sentiment analyst） 立场： 综合（synthes…", "summary": "情绪综合：恐贪指数的两极分化与\"ETF大迁徙\"中隐藏的行为加速器 日期： 2026年5月2日 分析师： 市场情绪师（sentiment analyst） 立场： 综合（synthesize） 参考卡片： 01（ashare strategist）、02（china macro）、03（financ…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "增量中央债是中和地方信用收缩的“止痛药”而非“引擎”，收益率曲线将因资产荒继续下行并突破历史低位。", "summary": "债券市场在 2026H2 面临严重的“资产荒”，增量 1.5-2.0 万亿国债发行将被银行等机构迅速吸收，不仅不会推升利率，反而可能推动 30 年期国债收益率突破历史低位（测试 2.1% 甚至更低）。估算的广义财政乘数约 0.95-1.05，增量中央支出仅能抵消地方化债带来的信用收缩，无法点燃 V…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-03", "supports": 0, "risks": 0, "neutral": 1, "value": 47.8, "events": [ { "title": "[auto-handoff] 04-29 VIX spike: Treasury Quarterly Refunding term premium contribution", "summary": "致 市场情绪师： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: e5f00b69-5c5a-49ea-9773-62f9ee5f24ee 结果摘要: Wrote handoff_response.md (11,151 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 1, "value": 43.2, "events": [ { "title": "\"session id\": \"a5c839ce 1820 42a0 b8c5 cb453873bd99\", \"response\": \"The Investment Committee…", "summary": "\"session id\": \"a5c839ce 1820 42a0 b8c5 cb453873bd99\", \"response\": \"The Investment Committee has officially synthesized the research from the Global M…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "5-10 bps 的 NIM 压缩目标在 54 万亿存款到期重定价和风险权重红利的支撑下是稳健的，即便在不全面降准的财政主导情景下亦然。", "summary": "2026年作为12万亿化债高峰年，2.8万亿地方债置换将直接拖累银行NIM约2bps。然而，54万亿高成本存款到期重定价预计将提供10-12bps的对冲红利。在PBOC通过二级市场国债买卖而非全面降准来管理13万亿总供应压力的情景下，银行NIM压缩5-10bps的假设是审慎且成立的。资产端正经历从…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-08", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "截至2026年5月8日，美债10年期收益率4.45%与中国10年期国债收益率1.65%形成的倒挂达到280bp，为2022年以来的95分位水平。", "summary": "截至2026年5月8日，美债10年期收益率4.45%与中国10年期国债收益率1.65%形成的倒挂达到280bp，为2022年以来的95分位水平。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-11", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "[auto-handoff] 低利率环境持续性", "summary": "致 信用分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 1cc40f49-517c-4137-9137-fdf20f0e759a 结果摘要: Wrote handoff_response.md (4084 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-12", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "中国10年期与30年期国债（CGB）期限溢价（10s30s）为47.3个基点", "summary": "中国10年期与30年期国债（CGB）期限溢价（10s30s）为47.3个基点", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 4, "neutral": 2, "value": 33.8, "events": [ { "title": "在 2027–2028 数字 LGFV 局部违约窗口前 18 个月，全天候组合应拉长利率久期、压缩并升级信用久期、预留错杀弹药，把信用风险贡献从约 18% 降至约 11%。", "summary": "支持 Research note 3 的数字 LGFV 信用分化判断，并将其翻译为全天候组合的久期—质量再平衡：利率久期由 6.5 年拉长至 7.5–8.0 年（+3 至 +5pp），10y CGB 与 7–10y 国开为锚；信用篮子总量 -1 至 -2pp，内部久期压至 2.5–2.8 年，质量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数字LGFV违约窗口下，组合应做国债主导的10年牛市平坦化，用CGB凸性和现金流动性覆盖左尾，而不是用30年集中、国开加杠杆或信用类券种凑久期。", "summary": "截至2026-05-14，本卡综合 Cards 3–4，认为2027Q2–2028Q1数字LGFV违约窗口应按利率与流动性事件管理。7.5–8.0年10y CGB久期目标应通过国债主导的10年牛市平坦化实现，而非组合陡峭化；CGB占利率仓60%–65%，国开25%–30%且仅在10年国开-国债利差…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "上市银行净息差筑底判断", "summary": "上市银行净息差筑底判断 If the industry estimate for listed-bank NIM in 2026Q2 falls below 1.35%, does the bottoming thesis need reassessment?", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中央以省级承担为先、政策行配合为辅的化债反应函数在基准情景下支撑而非破坏 CGB 牛市平坦化，但 2027Q3 起政策档位切换至 L4 中央加杠杆将是平坦化交易的真正退出信号。", "summary": "压力测试 Research note 5 牛市平坦化在 2027–2028 数字 LGFV 违约窗口下的持续性。中央财政反应函数为四档：L1 省级承担+地方 AMC（基准 60%）、L2 特别再融资债额度倾斜 2,000–3,000 亿元（基准内）、L3 政策行扩表 5,000–8,000 亿元 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 3, "neutral": 0, "value": 35.2, "events": [ { "title": "Research note 06 对释放的 3%-5% RC 采用 65% 美国国债 (UST) 与 3% CGB 的拆分。", "summary": "Research note 06 对释放的 3%-5% RC 采用 65% 美国国债 (UST) 与 3% CGB 的拆分。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国 10 年期国债 (CGB) 收益率中枢在 1.95–2.15% 区间。", "summary": "中国 10 年期国债 (CGB) 收益率中枢在 1.95–2.15% 区间。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在人民币6.80附近的升值护盾下，人民银行可以实施20-30bp非对称式降息，主要收益是降低地方债再融资压力和改善居民现金流，而不是立刻触发强消费周期。", "summary": "截至2026-05-14，本卡支持上一张外汇策略卡的核心判断：USD/CNY 维持6.80附近确实为国内宽松创造窗口。但政策不应是全面降息，而应是20-30bp的非对称式降息：贷款与再融资利率降幅大于隔夜资金利率，5年期以上LPR对居民现金流给予更强支持，同时用存款利率协同保护银行净息差。地方债方…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 4, "neutral": 1, "value": 33.6, "events": [ { "title": "[auto-handoff] USD/CNY 路径与 273bps 中美利差的可持续性", "summary": "致 债券分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: e921623d-4f97-4b98-8ef8-0deedb115778 结果摘要: Wrote handoff_response.md (10060 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Summit De-risking", "summary": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 502dfe56-3d6c-4017-933c-df9f9b15f88b 结果摘要: Wrote handoff_response.md (8242 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 4月信贷负增长后的银行资产配置转向", "summary": "致 日报总编： 协同任务已完成。执行方为 金融行业分析师。 模式: adhoc 会话: 79be5c67-6bfa-4dd4-a8e4-be78ae7a9732 结果摘要: Wrote handoff_response.md (7676 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "哑铃是未来四周合适的战术结构，可在不削减权益贝塔的前提下降低组合波动并增厚carry，但若10年国债下行压缩红利利差至250 bp以下、或SOE股东回报政策反转，策略将悄然失效。", "summary": "截至2026-05-15，结合卡片01–03的后段挤压诊断与TMT局部供给冲击，哑铃配置（保留AI算力/先进制程核心 + 加配红利/国企价值）是未来四周风险收益比为正的战术结构：在不削减权益贝塔的前提下，组合年化波动下降约150–220 bp、四周carry增益约35–45 bp、并截断TMT解禁…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 1, "risks": 1, "neutral": 0, "value": 58, "events": [ { "title": "2026-05-17的A股风格应是红利现金流与选择性成长久期的哑铃，而不是在AI Capex ROI未验证时全面追逐成长。", "summary": "2026-05-17，本卡综合支持卡片07的CGB/RMB流动性结论，但不将其外推为全面成长股进攻。10Y CGB下行5-12 bp和人民币偏强降低贴现率，支持人民币久期资产；但AI Capex现金回报仍未验证，成长仓位必须聚焦现金流和订单可见度。建议A股采取高质量哑铃：高股息与央国企现金回报35…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "截至2026-05-16，支持UST 2s10s方向但否定3.40%的10Y UST目标；CGB可承接400-600亿美元外资流入而不发生剧烈收益率下行。", "summary": "截至2026-05-16，本卡压力测试并部分支持卡片05：UST 2s10s 牛陡 40 bps 可行，但应来自前端收益率下行，不能依赖 10Y UST 跌到 3.40%。基准为2027年2Y 3.20-3.40%、10Y 4.10-4.35%、2s10s +75至+105 bps。财政供给和期限…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 1, "value": 54.4, "events": [ { "title": "CGB Curve Steepening: Structural Shift or Temporary Overshoot?", "summary": "CGB Curve Steepening: Structural Shift or Temporary Overshoot? Is the 10Y-30Y CGB spread widening to a 4-year high driven by geopolitical oil shocks …", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 08 — 情境 B 触发下 CGB +10 ppt 久期超配的期限结构、新券选择与 30Y 拥挤监控 工作日期: 2026 05 18 (Asia/Sin…", "summary": "Research note 08 — 情境 B 触发下 CGB +10 ppt 久期超配的期限结构、新券选择与 30Y 拥挤监控 工作日期: 2026 05 18 (Asia/Singapore) 分析师: 债券分析师 (bond analyst) 立场: support (承接 asset al…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-18", "supports": 1, "risks": 2, "neutral": 0, "value": 54.8, "events": [ { "title": "保留 CGB 久期作为 risk-off 对冲，但必须拆分 7–10Y/30Y 配比并加装熊陡触发器，否则供给冲击会让“久期对冲”反转为“久期放大器”。", "summary": "对前两卡“CGB 7–10Y/30Y +10ppt 久期超配”施加压力测试。当前 CGB 曲线 2s10s 约 60bp、10s30s 约 40bp，偏陡。四象限情景分析显示：牛陡/牛平下 7–10Y 凸性最优、30Y 在牛平中弹性最大；熊平下久期防御弱化但损失可控；真正的盲点是熊陡（S4）——超…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "“真杠铃”可执行，但必须把触发器从绝对阈值改为变化阈值，并在6月前预置现金，否则30bp信用红灯会在信用流动性已经变薄后才触发。", "summary": "截至2026-05-18，本卡对第06卡“真杠铃”的信用利差>=30bp和期限利差>=25bp再平衡规则做压力测试。结论是：方向正确，但原始阈值过粗。6月基准情景下，回购和短国债流动性仍可支撑执行，不太可能发生全市场流动性枯竭；风险集中在低层级城投、地产债、银行资本工具和拥挤红利替代品。信用触发器…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "根据中国人民银行2019年调查，中国60岁以上人群家庭资产中房产占比历史高达60-70%。", "summary": "根据中国人民银行2019年调查，中国60岁以上人群家庭资产中房产占比历史高达60-70%。", "sourceLabel": "证据更新", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 2, "neutral": 1, "value": 59.2, "events": [ { "title": "PBoC T3-T4 阶段的首个 stabilization confirmed 信号应来自 CNH HIBOR 倒挂回吐与 1Y CNH NDIRS receiving，而不是来自…", "summary": "2026-05-20 卡片支持卡 03 的 PBoC 反应函数，但把最早可交易稳定确认信号定位到 CNH 前端曲线，而非 10Y CGB 或单独 1Y NCD。D+5-D+8：CNH HIBOR O/N 与 1W 飙升、1Y CNH NDIRS pay up、1Y NCD 小幅走阔、10Y CGB…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "公用事业承接AI负荷的融资栈是寿险一般账户与税收股权台主导的久期+容量故事，而非货币中心银行的信用冲击；股权稀释与§6418转让定价是2026下半年公用事业利差的真正定价点。", "summary": "2026-05-20：美国IOU为承接AI/数据中心负荷叠加脱碳，2025–2028年累计CapEx需1.4–1.6万亿美元，其中T&D约3500–4200亿，AI直接相关约1100–1400亿。需净发1300–1700亿新增普通股+6000–7000亿长期无担保债，股权节奏较2022–2024翻…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026下半年公用事业信用不是银行式信用冲击，而是紧IG市场中吸收长久期供给的利差再定价，AEP和ETR最具左尾凸性。", "summary": "2026-05-20：本卡压力测试第06卡而非否定。2026下半年公用事业10年A-/BBB+ holdco公开债最可能中枢为T+140-160 bp，30年寿险私募中枢为T+175-210 bp；左尾分别为T+190-225 bp与T+250-300 bp。核心触发是30年美债维持5.35%以上…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "三条宏观线可以并存，但UST 10Y久期是最脆弱的条件性对冲仓，应在美国通胀或期限溢价继续上行时优先收缩。", "summary": "截至2026-05-20，本卡压力测试后有条件支持 UST 10Y +5pp、日本金融工业 +3pp、CGB 5–7Y +2pp 的组合倾斜，但置信度分层。美国9月首降仍可成立，却受4月CPI 3.8%同比、10Y Treasury 4.67%和期限溢价冲击制约，是第一可收缩腿；BoJ正常化仍利好…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 4, "neutral": 2, "value": 43.6, "events": [ { "title": "USD/CNY三项触发同时成立时，将组合从单一'利差红利'押注转为'已对冲右尾'的中性卡瓦：在岸利率簿−7pct、美元+黄金+8pct、RMB现金+1pct缓冲，触发解除按60日线性…", "summary": "本卡作为10/10收官，综合Research note 01–09链条，给出USD/CNY尾部触发条件下的跨资产对冲方案。若1Y远期贴水>3%、CNH HIBOR>4%、SAFE跨境净流量连续两月转负三项同时成立，应将在岸利率敏感簿减仓7pct（A股银行红利−3pct、CGB久期−4pct且久期从…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "央行保障房再贷款利率下调与财政贴息组合在2026年内修复收储经济性的政策可行性测算", "summary": "测算证实「PBOC再贷款1.75%→1.0% + 财政贴息100–150bp」组合可将地方平台综合融资成本从2.5–3.0%压至0.5–1.0%，对1.91%净租金形成91–141bp正利差，DSCR升至2.7–5.5x，技术上可在2026年Q3–Q4将杠杆从1.0x推至理论上限3.33x，但贷款…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "银行资本约束催生的“资产荒”将政金债收益率压至极限，迫使资金向信用下沉，制造了脱离基本面的极窄信用利差和严重的结构性风险错配。", "summary": "截至2026年5月21日，本卡片分析了银行息差压缩与LGFV化债对债券市场的深远影响。银行内源性资本受损导致其规避高风险资产，疯狂涌入低风险权重的国债和政金债，引发“资产荒”并压平收益率曲线。无风险利率的塌陷迫使非银机构向信用下沉追逐收益，人为压低了信用利差（验证了Research note 06…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大行'类债化'是CGB收益率塌陷的结果而非原因,对信用利差仅有有限配置溢出效应,对期限溢价定价逻辑无结构性影响,反而埋藏2027不良暴露下的对称反转风险。", "summary": "压力测试结论：国有大行'类债化'对境内利率曲线和信用利差的传导跛行且被高估。保险OCI增配权益挤出部分信用债需求，仅能解释AAA短端15-25bp被动收窄，而'股息率-10Y CGB利差维持260bp+'实际是分子(股息率)与分母(CGB收益率)同步下行所致——市场在给所有久期资产加价，并非给大行…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "10Y中债1.5%开启‘权益AAA’重估窗口，但在NIM破1.3%或信用利差过150bps前，杠铃策略需从广义金融向准主权公用事业收缩。", "summary": "截至2026-05-22，本卡评估了10Y中债收益率下行至1.50%对高股息资产的支撑。结论显示：‘资产荒’逻辑下，长江电力、中国移动、国有大行等‘权益AAA’标的仍有5%-20%的估值上行空间。AAA-AA信用利差走阔目前触发避险买入，利好顶级央企。但需警惕1.30%的银行净息差(NIM)红线与…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 0, "neutral": 3, "value": 81.4, "events": [ { "title": "As the Global Macro Analyst , I have conducted a comprehensive synthesis of the macro finan…", "summary": "As the Global Macro Analyst , I have conducted a comprehensive synthesis of the macro financial transmission channels of G SIB USD balance sheet rati…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Re-estimate supply chain price shock transmission to U.S. core PCE", "summary": "Re-estimate supply chain price shock transmission to U.S. core PCE In 'Thesis 1: Step 4', direct a high-priority handoff to global-macro to re-estima…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Summary of Work We have completed the strategic analysis for Research note 08/10 under the …", "summary": "Summary of Work We have completed the strategic analysis for Research note 08/10 under the board topic \"Re estimate supply chain price shock transmis…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 53.8, "events": [ { "title": "We have successfully completed Research note 03/08 of the research thread session efecefe6 …", "summary": "We have successfully completed Research note 03/08 of the research thread session efecefe6 2a06 4881 87e9 ccec1f502077 as the Chief Economist. Stance…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have successfully completed the stress test macro analysis and generated the required del…", "summary": "I have successfully completed the stress test macro analysis and generated the required deliverables for Research note 04 under the shared research t…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 1, "value": 64.2, "events": [ { "title": "I have completed the stress test assessment of the Big Six state owned commercial banks' hi…", "summary": "I have completed the stress test assessment of the Big Six state owned commercial banks' high dividend defensive attributes. Key Stress Test Findings…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "资产荒支持债券牛市和平坦化，但当前更应买5-10年利率和3-5年高等级信用的风险调整收益，而不是无差别追逐30年久期或低评级信用。", "summary": "截至2026-05-24，本卡支持 Research note 7 对银行信贷扩张受限的判断，但压力测试债券端结论：资产荒会把流动性推向利率债和高等级信用，压低收益率与利差；但长端仍受PPI、政府债供给和期限溢价约束，信用利差也从违约概率定价转向资产稀缺、流动性和条款结构定价。基准看10年国债近期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 11, "neutral": 2, "value": 12, "events": [ { "title": "银行 NIM 转型：存款红利能否抵消化债压力？", "summary": "银行 NIM 转型：存款红利能否抵消化债压力？ To what extent will the maturity of high-rate time deposits effectively cushion the impact of widening credit spreads from LG…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 中美政策周期错位下的资产配置含义", "summary": "致 债券分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 7d11c649-5c1f-4730-8cee-8ec9de5e62b7 结果摘要: Wrote handoff_response.md (9013 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] A股量化规模突破3.2万亿及两融高拥挤下的跨资产风格防御配置", "summary": "致 A股策略师： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: 19aedac4-93df-40a0-bcf6-3dd42068c521 结果摘要: Wrote handoff_response.md (13067 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在两融见顶+Q3 1200亿解禁微观去杠杆环境中，HEALTH精选篮子18%物理仓位 / 30%风险预算配合70%TF名义+IF/IM空头+USDCNH远期的三层对冲，是同时满足风险…", "summary": "截至2026-05-25，本卡把card-05的18-20%名义敞口升级为风险平价+全天候双约束下的最优解：HEALTH精选篮子承担组合年化波动的30%风险预算，WLEF红利底座25%，长久期CGB（TF）20%，黄金15%，CRB商品10%；对应物理仓位18%/30%/70%TF名义/18%/9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 3, "neutral": 1, "value": 55, "events": [ { "title": "CGB的避险属性应定义为政策稳定化的人民币久期锚，而不是全套保外资的无风险套利资产。", "summary": "截至2026-05-27，本卡综合Research note 08外汇套保压力测试，认为CGB并未替代UST成为全球通用安全资产，而是被重估为政策稳定化的人民币久期锚。G3背离导致10年期UST-CGB利差达275.95 bp、30年期达280.95 bp，但USD/CNH高套保成本削弱全套保外资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "145 触发下中资美元债违约/置换风险呈双峰分布——地产残债与内陆弱 LGFV 6M 重组概率 25–60%、其余 T1/T2 主体 <5%；CNH 点心债承接再融资、境内利差仅走阔…", "summary": "2026-05-26：把卡 04 的 EM HC 瀑布在中资美元债内部按发行人级 6M P(违约∪置换) 排序，约 USD 490bn 引用存量加权概率约 5.8%，但损失高度集中于 T4 双峰：T4 地产残债（约 USD 30bn，6M P(D∪X) 45–65%）与 T4 弱内陆 LGFV（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "央行国债买卖工具可有效抑制曲线陡峭化，但结构性地不能压缩信用利差；LGFV违约冲击下的利差走阔必须依赖财政再融资债额度、定向央行工具和CRMW并行处置。", "summary": "截至2026-05-27，本卡对Research note 09的CGB久期锚结论进行压力测试，结论是有条件的：央行国债买卖操作可以有效压平国债曲线、抑制陡峭化（2026-05-26 10年期CGB 1.7405%、30年期2.2205%；10s-1s斜率56.58 bp、30s-10s斜率48.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "EM 债券桶在 145 触发下不是按比例缩，而是按 FX β 与流动性反身性结构性重排：LC 砍 BRL/ZAR 长端、保 INR/CEE，HC 先清 BB 再缩 IG 长端，T+0…", "summary": "2026-05-26：把 Research note 03 的 EM LC 18%/8% 与 EM HC 17%/10% 桶分解到国家/发行人。EM LC 按 FX β 与外资占比排序，稳定区间 BRL 4.5%/MXN 4%/INR 4%/IDR 3%/ZAR 1.5%/CEE 1%，加权久期 …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 0, "neutral": 1, "value": 74.4, "events": [ { "title": "I have completed the micro stress test on the 5% advanced packaging equipment allocation (o…", "summary": "I have completed the micro stress test on the 5% advanced packaging equipment allocation (originally composed of Changchuan and Piotech) under the HB…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 2, "value": 72.8, "events": [ { "title": "[auto-handoff] 化债背景下商业银行息差崩塌对基建及制造业信贷传导的挤压测试", "summary": "致 首席策略师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 902cfe1f-6301-43b2-ab20-eff51b2a3291 结果摘要: Wrote handoff_response.md (18256 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Impact of Debt Swaps and Platform Mergers on LGFV Spreads", "summary": "致 债券分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 6333b17d-ceb8-41bd-bc15-ae7fdb5f7393 结果摘要: Wrote handoff_response.md (4425 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 去杠杆螺旋与-16.25%期指贴水下多资产风险平价模型Barbell权重再优化测算", "summary": "致 衍生品策略师： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: 626f9305-a975-4d98-ae5d-052defbbc304 结果摘要: Wrote handoff_response.md (8456 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 1, "value": 71.2, "events": [ { "title": "I have completed the synthesis and quantitative stress testing for Research note 07 as the …", "summary": "I have completed the synthesis and quantitative stress testing for Research note 07 as the Chief A Share Strategist , anchoring all analyses to the c…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A-Share High-Leverage TMT Sector Valuation Resilience Under Global Carry Liq…", "summary": "致 外汇策略师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: c013b698-543d-4bb2-b87c-772599973eaf 结果摘要: Wrote handoff_response.md (10124 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 4, "neutral": 3, "value": 60.6, "events": [ { "title": "利率中枢稳定是A股核心资产的被动估值锚，资金正从被惩罚的超长端债外溢到类债权益，CSI 300/上证50将在三季度跑赢中证2000，Q3配置应采取高股息底仓+算力业绩兑现型成长的哑铃…", "summary": "本卡承接 card-02 关于长端CGB熊市陡峭化的压力测试结论，论证利率中枢稳定+长端温和回调（10Y CGB 1.71%、长端+20-35 bps）对A股是积极的结构性事件：折现率被锚定低位、CSI 300 ERP已升至历史99分位（约6.14 pct），股息率对10Y CGB首次形成+1.7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "10Y国债收益率曲线重塑带来的资产滚存收益（仅+1.23 bps）极其微弱，在结构上完全无法对冲中度LGFV信用重组与拨备计提对商业银行净息差造成的实质性侵蚀（等效-5.67 bps…", "summary": "本卡片对 10Y 国债收益率曲线重塑对商业银行净息差的修复作用进行了系统性 ALM 压力测试。我们判定该修复强度极其微弱，在乐观的 +40 bps 陡峭化场景下仅拉动 NIM 上行约 1.23 bps；而在中度 LGFV 信用危机场景下，城投债务展期与计提减值准备将侵蚀银行等效息差达 5.67 b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A Share Asset Allocation Strategy Under Low Inflation & Property Wealth Traps (Research not…", "summary": "A Share Asset Allocation Strategy Under Low Inflation & Property Wealth Traps (Research note 9/10) As the Chief Strategist (首席策略师) , I have synthesiz…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "第05卡描述的信用机制成立，但把 USD/CNY 7.30 尾部触发器外推为 A 股地产板块的板块性杀估值是策略层面的过度定价；真正的 alpha 在于板块内的二阶分化（多 SOE …", "summary": "我从首席策略师角度否定第05卡「双重隔离 → A股地产板块杀估值」的强外推。三点反驳：(1) USD/CNY 7.30 是尾部触发器而非基准——2026-05-31 离岸 CNH 报价 ~6.78，距 7.30 尚有 ~7.7% 贬值空间，把尾部当 base case 是策略过度赋权；(2) A …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "作为 中国宏观分析师 (china macro) ，我基于权威工作日期 2026 05 31 ，针对商业银行在实施“监管平滑”和维持“资产质量幻觉”（Research note 04…", "summary": "作为 中国宏观分析师 (china macro) ，我基于权威工作日期 2026 05 31 ，针对商业银行在实施“监管平滑”和维持“资产质量幻觉”（Research note 04/05）背景下所导致的表内资产“僵尸化”及由此产生的微观信贷扭曲，完成了系统性的压力测试与宏观仿真��估。 在执行任…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 1, "value": 75.4, "events": [ { "title": "[auto-handoff] 中美利差倒挂背景下，美债期限溢价突发上行对中资商业银行NIM及二级资本债发行的挤压压力测试", "summary": "致 外汇策略师： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: 33a9cb8e-06f5-48a6-a6a5-af4634bbc386 结果摘要: Wrote handoff_response.md (17053 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Q3 政府债供给斜率 + 降准时点 → CGB10Y 中枢", "summary": "致 债券分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 807a3313-8497-42af-980f-a44daec08978 结果摘要: Wrote handoff_response.md (8190 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 0, "neutral": 1, "value": 82.8, "events": [ { "title": "I have completed the duration management and yield curve hedging strategy research report f…", "summary": "I have completed the duration management and yield curve hedging strategy research report for Research note 09 under the target date 2026 06 02 . Sum…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 75.2, "events": [ { "title": "10Y-RRP仅31.28 bps时，中长端CNY久期的边际收益不足以补偿熊陡与信用利差波动风险，应保留现金期权而非加仓久期。", "summary": "截至2026-06-04，本卡对CNY久期加仓做压力测试：10年期国债1.7128%、PBOC 7天逆回购1.40%，10Y-RRP仅31.28 bps。以约9.12修正久期测算，10年期收益率上行3.43 bps即可吞掉一年超额carry；25 bps熊陡会令1.0pp仓位损失约2.28个组合b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 1, "neutral": 1, "value": 79.6, "events": [ { "title": "截至2026-06-04，最佳债券曲线不是中美同步缩久期，而是美债保留约7Y有效久期的5Y/7Y杠铃、中债压到3Y核心，并把信用风险从HY和长城投转向可对冲IG与短久期AAA二级资本…", "summary": "截至2026-06-04，本卡片支持卡片06的债券缩久期方向，但修正执行：美债不是简单从10Y缩至7Y，而应做2Y/5Y/10Y/30Y = 20/35/30/15的5Y/7Y滚动收益杠铃；中债应更坚决缩至3Y核心，采用2Y/3Y/5Y/10Y/30Y = 20/50/20/10/0。美国IG可轻…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 中美关键矿物（镓/锗/锑）脱钩对全球LEO卫星星座CapEx通胀与美联储中性利率传导量化", "summary": "致 主题研究员： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 1cc9fb67-94cf-4e7f-a3ea-8bb5bd02af48 结果摘要: Wrote handoff_response.md (15198 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 1, "value": 81, "events": [ { "title": "[auto-handoff] 信用风险传导与中小银行资本", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: e5b1dae4-0475-4d75-859e-5ea469156024 结果摘要: I will first list the contents of the default project direc…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 央行5000亿买断式逆回购与今日1500亿1Y国债续发对季末超储的动态对冲效果评估", "summary": "致 债券分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: b95a02bb-735c-407a-801d-5de5cefb10bd 结果摘要: I will list the directory contents of the application data di…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 0, "neutral": 2, "value": 91.4, "events": [ { "title": "We have completed the quantitative stress test on the JPY carry trade unwind and its transm…", "summary": "We have completed the quantitative stress test on the JPY carry trade unwind and its transmission to the US Treasury yield curve and high yield (HY) …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 离岸人民币流动性挤压对境内城投债及地方专项债信用利差的二阶传染评估", "summary": "致 主题研究员： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 990cc221-8af1-4fac-9830-c8912e2fa2e5 结果摘要: I will list the current permissions to see what actions and p…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 5, "neutral": 3, "value": 65.8, "events": [ { "title": "[auto-handoff] LGFV Social Housing Buyback ICR Degradation and Cross-Sector Credit Contagio…", "summary": "致 首席风控官： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 5ed084a7-83b5-4d98-91a7-3a3f1b37da01 结果摘要: Wrote handoff_response.md (10743 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中国商业银行净息差触及1.40%低谷与8000亿特别国债注资支撑分红能力的博弈", "summary": "中国商业银行净息差触及1.40%低谷与8000亿特别国债注资支撑分红能力的博弈 Commercial Bank NIM Reaching 1.40% and Dividend Sustainability Under Special Treasury Bond Injection 在全行业平均净息…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "两融、ETF与行业成交结构已部分确认A股政策顺周期切换，但宽度不足，仍属头部权重驱动的早周期信号。", "summary": "截至2026-06-08，两融、ETF和行业成交三类微观证据共同支持卡片01、02的政策顺周期切换判断：杠杆资金在非银金融、银行、电力设备、军工、机械的融资买入占比抬升，在食品饮料、医药、地产下移；ETF净申购集中在券商、银行、军工、高端装备和央企红利/中特估，纯地产与白酒ETF净赎回；金融与先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Q3 PBOC 宽松空间不是零但已压缩至 25 bp RRR + 10 bp OMO 上限,且只在 8 月美元见顶后启动;A 股反弹隐含的信用脉冲叙事被 M1/社融脉冲/票据利率三大…", "summary": "在美债 10Y 4.55% + PBOC 必须先守 7.40 中间价的双重约束下,Q3 RRR/OMO/MLF 宽松空间并非零,而是被压缩到'流动性维稳'级别:上限是 25 bp 全面 RRR + 10 bp OMO,且大概率在 8 月底美元见顶之后才启动。中美 10Y 利差倒挂 ~275 bp、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 3, "value": 91.2, "events": [ { "title": "土地财政滑坡与银行信用收缩共振使货币乘数实质下行 0.6–0.8 个单位，LGFV 表内展期临界点为省综合财力/付息<0.8×，五省已触发，2026H2 LGFV+地产合计潜在新增不…", "summary": "工作日期 2026-06-08。土地出让收入连续四年滑坡、5,000 亿收储再贷款负利差、银行表内信用减值同步上行三重共振，2026 年货币乘数从 7.92× 实质下行约 0.6–0.8 个单位，中国进入'宽货币、紧信用'的准流动性陷阱。LGFV 非公开债务在银行表内展期容忍度的临界点位于'省综合…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I will view the report files of Research note 06 to understand its methodology and content,…", "summary": "I will view the report files of Research note 06 to understand its methodology and content, especially since Research note 06 took a \"deny\" stance on…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "净息差收窄对A股银行股红利定价与增量配置资金行为的影响", "summary": "截至2026-06-08，A股银行高分红红利逻辑面临的是部分、分化的重估而非崩塌。2026Q1行业净息差降至1.40%历史新低（大行1.29%），盈利转弱使分红‘增长腿’向下重估，叙事从‘收益率+成长’转为纯套息/类债替代，PB重估封顶约1.0倍。但收益率之锚仍成立：中证银行指数股息率约4.67%…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "科创50破位会触发局部去杠杆压力，但在流动性缓冲和银行相对防守尚存的情况下，不宜直接判定全市场两融被动平仓踩踏已经启动。", "summary": "2026-06-08，本卡判断科创50跌破1720更像局部科创/半导体去杠杆，而非全市场两融踩踏。两市融资余额为2.89万亿元，科创板两融为3805.56亿元，压力真实但尚未失序；银行在6月5日逆势上涨1.33%，说明红利/大行对人民币组合仍有相对防守作用。美元口径风险来自中美10年期利差约-28…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 2, "neutral": 1, "value": 80.6, "events": [ { "title": "[auto-handoff] 中小银行2000亿低评级城投债买盘收缩对跨资产流动性冲击与防守策略", "summary": "致 日报总编： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: f20e924c-699e-4b5f-b5c0-53d6928afd0f 结果摘要: Wrote handoff_response.md (11134 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 商业银行净息差（NIM）降至1.40%极值背景下中小银行对低评级城投债买盘的边际收缩与流动性支持减弱评估", "summary": "致 信用分析师： 协同任务已完成。执行方为 金融行业分析师。 模式: adhoc 会话: c3ce5681-f62b-48d8-b4a7-31be70b22fcc 结果摘要: Wrote handoff_response.md (14239 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "聪明钱（北向）已前置买入上游油气、电网、铜铝，但融资盘与低波 ETF 仍在为防御板块支付第二次溢价——Card-04 行业-风格再分配蓝图的执行窗口约剩 2–4 周，关键倒计时是红利…", "summary": "截至 2026-06-09，A 股资金面部分确认 Card-04 的再分配方向但尚未走完：北向已连续 5 日净流入并集中在中石油、中海油、紫金、长江电力、国电南瑞、特变电工等篮子 A/B 标的，融资余额却仍在缓慢加白酒、火电、银行等防御板块，红利低波 ETF 净申购仍大于能源 ETF——拥挤防御的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 3, "value": 94, "events": [ { "title": "[auto-handoff] 储蓄国债电子式首入个人养老金产品池对境内中长期流动性及利率曲线的结构性影响评估", "summary": "致 债券分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 81eed5f8-8b51-4804-96e5-48de814568f3 结果摘要: Wrote handoff_response.md (10068 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 方案四多资产风险平价模型杠杆及久期风险修正", "summary": "致 日报总编： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: d1524db5-ea43-417a-bd60-1a85bae11c74 结果摘要: I will start by checking the current date using the shell and …", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "20% 利率债 sleeve 拆分为 10Y 国债 50% / 7-10Y 政金债 35% / AAA 3-5Y 高等级信用债 15%（sleeve 久期 ~7.1Y），以小幅尾部对…", "summary": "在 card-07 的 20% 利率债 sleeve 之上做品种细化：推荐 10Y CGB 50% / 7-10Y 政金债 35% / AAA 3-5Y 高等级信用债 15%，sleeve 加权久期约 7.1Y、加权 yield 1.88%。锚点：10Y CGB 1.68%[S1]、国开-国债利差…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "防御档以银行/公用/煤炭/食饮四大重仓+电子-700bp为对手盘；修复档反向，电子+700bp但须用北向条件触发分拆建仓，跨档摩擦约10-11bp迫使触发器严守。", "summary": "本卡在card-03的防御/中性/修复三档框架下，给出申万一级28行业相对沪深300基准的BP级主动权重。防御档（成长25/红利55）最大赢家为银行+600bp、公用事业+500bp、煤炭+450bp、食品饮料+400bp；最大输家为电子-700bp、电力设备/计算机各-500bp、医药-400b…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 0, "neutral": 1, "value": 95.4, "events": [ { "title": "[auto-handoff] 黄金与期限溢价", "summary": "致 市场情绪师： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: 3b4d96d5-66d4-4a47-a013-a981dacf9226 结果摘要: # Role Bond Analyst (债券分析师). # Expertise Rates, credit, yield c…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 1, "neutral": 0, "value": 87.8, "events": [ { "title": "[auto-handoff] 评估Q3地方债置换推进对国内长端国债收益率及流动性的压力测试", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 55978b64-4c71-4fb5-a17e-b2cf9546c396 结果摘要: Wrote handoff_response.md (3564 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 0, "value": 89.2, "events": [ { "title": "熊陡情景通过信用风险暴露与收益率利差收窄的双重路径，正在击穿银行红利资产作为‘低波动避险工具’的防御底色。", "summary": "在熊陡情景下（长端利率+100bps），银行净息差（NIM）虽受资产端定价上行提振，但被信用成本飙升和债券投资组合浮亏对冲。压力测试显示，信贷风险溢价扩大及拨备计提增加将侵蚀归母净利润约6%-9%，导致绝对股息额下滑。由于国债收益率快速上行，银行股息率与无风险利率的利差（Spread）从300bp…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 1, "neutral": 4, "value": 96, "events": [ { "title": "本期10卡报告主线严密、数据一致、配置可执行，评级为‘通过附小修订’；下期必须把内联[S?]引用与缺失card-09三件套补齐。", "summary": "对card-01至card-09的终审显示：主线数据贯穿一致（CPI 4.2%、FFR 3.50-3.75%、中美利差≥200bp、10Y CGB 2.50-2.75%、30Y 2.80-3.05%、二永+30-40bp、弱城投+65-120bp、TLAC 16/18%），立场链合规（4 supp…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I will begin by inspecting the session brief and the report from Research note 4 to establi…", "summary": "I will begin by inspecting the session brief and the report from Research note 4 to establish a solid foundation for our stress test analysis of LGFV…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "三四线城投在2027年断奶前会经历有序的、政策定价的尾部事件而非系统性违约，对全国性银行清偿能力不构成威胁；真正的尾部因子是财政纪律而非平台本身。", "summary": "截至2026-06-14，本卡对2027年6月融资平台转型期限前的尾部风险与信用利差进行压力测试。基准情景（60%）：公开城投债零实质违约，AA(2)区县利差走阔60-100bp，银行CET1中枢下行30-50bp，可吸收；不利情景（30%）：1-3只区县私募实质违约，AA(2)利差走阔150-2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "The research into China's power system and green trading reforms at the start of the \"15th …", "summary": "The research into China's power system and green trading reforms at the start of the \"15th Five Year Plan\" (2026) is complete. The analysis confirms …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 0, "neutral": 2, "value": 96, "events": [ { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Tactical allocation under institutional-retail sentiment divergence", "summary": "致 市场情绪师： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: df522a0a-ad63-497e-acd1-c06e34805230 结果摘要: Wrote handoff_response.md (8660 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "[auto-handoff] 商业银行净息差（NIM）降至1.40%极值背景下中小银行对低评级城投债买盘的边际收缩与流动性支持减弱评估", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 金融行业分析师。 模式: adhoc 会话: b34f822c-e98e-4890-bed9-057a16a2f39b 结果摘要: Wrote handoff_response.md (10289 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "BOJ加息后的承重风险压倒性地在FX/利率簿（数万亿美元的银行/寿险FX对冲+套息），全球科技与港股/A股科技是二阶反身性受害者而非主管道。", "summary": "围绕2026-06-16 BOJ加息，用IMM日元仓位、东京零售FX、BIS LBS与BOJ/MOF资金流量+IIP四套数据测算日元套息簿结构：FX/利率簿（银行/寿险FX对冲簿数万亿美元 + JPY融资G10/EM外汇套息 + JGB与互换利差）占绝大头，JPY融资股权多头含全球科技高峰期亦仅几…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 0, "neutral": 1, "value": 96, "events": [ { "title": "[auto-handoff] 中小银行NIM跌破1.40%引发的信用收缩对大型公用事业二级债配置意愿的传导", "summary": "致 首席策略师： 协同任务已完成。执行方为 金融行业分析师。 模式: adhoc 会话: 9260bd6a-eb3c-44f2-bb64-3074089878bd 结果摘要: Wrote handoff_response.md (11016 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 2, "neutral": 6, "value": 96, "events": [ { "title": "[auto-handoff] 商业银行资产荒配置对收益率曲线及大类资产风格轮动的影响评估", "summary": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 5fcf9b8f-49e0-4ca8-bea2-ca739e00cae6 结果摘要: Wrote handoff_response.md (10160 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] NIM降至1.40%背景下中小银行低评级城投债买盘收缩评估", "summary": "致 信用分析师： 协同任务已完成。执行方为 金融行业分析师。 模式: adhoc 会话: 41113bd2-fa1b-4076-a9fc-f09a068b5c2e 结果摘要: Wrote handoff_response.md (14149 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 商业银行息差1.4%冰点硬约束下社融增速下滑与货币政策传导路径压测", "summary": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: c2e9168a-8872-4199-9c78-008e71e53807 结果摘要: Wrote handoff_response.md (12495 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 商业银行资产荒配置对收益率曲线及大类资产风格轮动的影响评估", "summary": "致 首席策略师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: eeccb81c-3798-4962-ad58-c10d301199c8 结果摘要: Wrote handoff_response.md (10727 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 0, "neutral": 1, "value": 96, "events": [ { "title": "中国主权资产负债表吸收债务迁移的跨资产定价", "summary": "截至2026-06-22，本卡把卡片08“可融资但对挤出敏感”的结论翻译成跨资产定价。统一洞见：主权吸收把LGFV信用溢价社会化，从而压缩各处违约溢价，却把供给/久期溢价转移到CGB长端和股权贴现率。锚点：10年CGB 1.75%、30年2.22%、USD/CNY约6.76-6.79、CSI 30…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 5, "neutral": 2, "value": 81, "events": [ { "title": "[auto-handoff] Asset Allocation Strategy on Yield Curve Steepening", "summary": "致 首席风控官： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: d361aa5d-fb62-4f2b-8f78-8b955c9b4b94 结果摘要: Wrote handoff_response.md (9963 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply s…", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] WTI at $65-68/bbl and Fed path threshold", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 2f15cf64-5f49-4af5-867a-05b15b121ef3 结果摘要: Wrote handoff_response.md (8806 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] LGFV 2028 implicit debt clearance deadline risk and local government bond sw…", "summary": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 7c8c8167-73c2-442c-97a7-ef4454113558 结果摘要: Wrote handoff_response.md (10257 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 3, "neutral": 0, "value": 88.4, "events": [ { "title": "[auto-handoff] Track local-bond supply, DR007 anchor, and high-dividend valuation risk", "summary": "致 消费行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: fdcc98d9-ec47-4bcf-ad48-9de0a3593d19 结果摘要: I will start by running the command to verify the current d…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 中美利差-272bps倒挂与商业银行净息差1.40%硬约束下三季度利率债净供给冲击测算", "summary": "致 首席风控官： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 406d10f9-5184-4e5c-a3c8-bd8a0400591e 结果摘要: Wrote handoff_response.md (10910 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 中美利差-272bps倒挂与商业银行净息差1.40%硬约束下三季度利率债净供给冲击测算", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 065568fb-7e27-470f-ab19-7926760583a8 结果摘要: Wrote handoff_response.md (11243 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 10, "neutral": 0, "value": 47.8, "events": [ { "title": "[auto-handoff] 商业银行 1.40% NIM 约束对保障房收储贷款投放的挤出压力", "summary": "致 房地产分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: aa75509e-3210-4ab6-b781-37670c2d5374 结果摘要: Wrote handoff_response.md (9392 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "1.95%止损在保证金/持有/滚动上廉价可执行，但卡片08错置了约束——真正起约束的是银行股权久期的非平稳性与CTD期限基差，而非流动性或保证金。", "summary": "截至2026-06-26，中国10Y为1.73%（财政部最近10年招标1.7116%），1.95%锚未触发。本卡压力测试该止损的可执行性：在NAV=RMB 1,000mm参考账本下，把银行腿久期贡献从2.25降至1.35组合年需中和每基点RMB 90,000的DV01，推荐'做空100手CFFEX…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "特别国债扩容把银行偿付尾部转化为 CGB 长端供给久期尾部；若仍把现金 CGB 当作银行股避险对冲，尾部回撤可从注资后 VaR 改善重新扩张到 -3.92% 至 -5.28%。", "summary": "2026-06-27：本卡支持并强化 card-05 的风险判断。2026 年特别国债相关供给包络为 1.6 万亿元（1.3 万亿元超长期特别国债 + 3000 亿元银行注资特别国债），且长端集中在 20Y/30Y/50Y。按 card-05 的 25% 银行股 / 50% CGB / 25% 私…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超长期特别国债供给压力对CGB长端收益率曲线及期限溢价的影响", "summary": "2026-06-27：1.3万亿超长期特别国债（23只：7×20Y/12×30Y/4×50Y，4月下旬至10月中旬发行）叠加约3000亿银行注资特别国债，构成约1.6万亿、精准集中于20Y/30Y/50Y节点的长端供给包络。在保险净卖超长CGB转配地方债、大行接近央行超长配置上限、杠杆基金/理财成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 0, "neutral": 2, "value": 96, "events": [ { "title": "I have conducted a comprehensive stress test of the CGB 10Y floor and yield curve dynamics …", "summary": "I have conducted a comprehensive stress test of the CGB 10Y floor and yield curve dynamics under the PBOC's targeted liquidity operations, comparing …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I will read the session brief and the reports from the previous card (Research note 05) to …", "summary": "I will read the session brief and the reports from the previous card (Research note 05) to align our analysis of the CGB 10Y yield curve and the effe…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 1, "neutral": 1, "value": 96, "events": [ { "title": "I have completed the research for Research note 01 of this research thread session. Summary…", "summary": "I have completed the research for Research note 01 of this research thread session. Summary of Work 1. Analyst Selection : Appointed the Global Macro…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 4, "neutral": 1, "value": 91, "events": [ { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 国内OMO宽松与美债/中债倒挂对资产配置影响", "summary": "致 投委会主席： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: 11df0c43-9ae2-4d6b-8082-594a86cea4e6 结果摘要: Wrote handoff_response.md (9778 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] LGFV去杠杆下商业银行资产配置转移评估", "summary": "致 金融行业分析师： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: a9766ff2-b543-428d-b6fb-e58bb46f0c4d 结果摘要: Wrote handoff_response.md (7843 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 推导 PBOC 隔夜逆回购实际中标利率及中债收益率曲线非线性久期损失风险", "summary": "致 晨会主编： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: eb98b73e-bcc0-4d80-8508-8b06dfaaf2c6 结果摘要: Wrote handoff_response.md (10281 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-fd31cc177b/", "networkHref": "/investor/research-network/#thesis-fd31cc177b" }, { "id": "thesis-51c6d833ae", "source": "tracked_thesis", "title": "食品加工的机会从食品通胀切换到原料波动、渠道库存与 B 端标准化", "titleEn": "Food processing opportunities are shifting from food inflation to input volatility, channel inventory, and B-side stand…", "lane": "食品加工/成本传导", "laneEn": "Food processing and input-cost transmission", "laneHref": "/investor/lanes/food-processing-and-input-cost-transmission/", "investableFocus": "食品加工/成本传导：001215.SZ, 603317.SH, 603027.SH, 600298.SH, 002847.SZ, ADM, DBA", "stockUniverse": [ { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "603317.SH", "name": "天味食品", "market": "A-share", "role": "复合调味料", "roleEn": "复合调味料", "bucket": "core", "href": "/investor/stocks/603317.SH/" }, { "ticker": "603027.SH", "name": "千禾味业", "market": "A-share", "role": "调味品渠道库存", "roleEn": "调味品渠道库存", "bucket": "core", "href": "/investor/stocks/603027.SH/" }, { "ticker": "600298.SH", "name": "安琪酵母", "market": "A-share", "role": "食品配料与出口", "roleEn": "食品配料与出口", "bucket": "core", "href": "/investor/stocks/600298.SH/" }, { "ticker": "002847.SZ", "name": "盐津铺子", "market": "A-share", "role": "休闲食品效率", "roleEn": "休闲食品效率", "bucket": "core", "href": "/investor/stocks/002847.SZ/" }, { "ticker": "ADM", "name": "Archer-Daniels-Midland", "market": "US", "role": "全球农产品成本基准", "roleEn": "全球农产品成本基准", "bucket": "peer", "href": "/investor/stocks/ADM/" }, { "ticker": "DBA", "name": "Invesco DB Agriculture Fund", "market": "US ETF", "role": "农产品价格代理", "roleEn": "农产品价格代理", "bucket": "hedge", "href": "/investor/stocks/DBA/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" } ], "stockTickers": [ "001215.SZ", "603317.SH", "603027.SH", "600298.SH", "002847.SZ", "ADM", "DBA", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI" ], "coreView": "赛道：食品加工/成本传导。候选观察标的：001215.SZ, 603317.SH, 603027.SH, 600298.SH, 002847.SZ, ADM, DBA。当前共有 40 条证据、3 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 78, "eventCount": 160, "firstSeen": "2026-05-01", "lastSeen": "2026-06-29", "supports": 10, "risks": 103, "neutral": 47, "analystCount": 17, "sourceTypeCounts": { "分析协作": 36, "链条事实": 1, "研究笔记": 88, "证据更新": 3, "研究线程": 31, "研究主题": 1 }, "topTerms": [ "GOES", "HBM", "CPI", "食品", "Food", "processing", "opportunities", "shifting", "food", "inflation", "input", "volatility" ], "whatChanged": [ "2026-06-29: [auto-handoff] 霍尔木兹海峡石油冲击与核心CPI传导", "2026-06-29: 产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "2026-06-29: 电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "2026-06-29: [auto-handoff] Hormuz shock & core CPI transmission", "2026-06-27: 食品加工板块β偏弱但覆盖缺口蕴含丰厚Alpha，应重点布局中报Setup干净的千味央厨与受新添加剂国标催化的B端生物配料龙头，回避无定价权的中小盘。" ], "practical": { "sectorLane": "食品加工/成本传导", "sectorLaneEn": "Food processing and input-cost transmission", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "603317.SH", "name": "天味食品", "market": "A-share", "role": "复合调味料", "roleEn": "复合调味料", "bucket": "core", "href": "/investor/stocks/603317.SH/" }, { "ticker": "603027.SH", "name": "千禾味业", "market": "A-share", "role": "调味品渠道库存", "roleEn": "调味品渠道库存", "bucket": "core", "href": "/investor/stocks/603027.SH/" }, { "ticker": "600298.SH", "name": "安琪酵母", "market": "A-share", "role": "食品配料与出口", "roleEn": "食品配料与出口", "bucket": "core", "href": "/investor/stocks/600298.SH/" }, { "ticker": "002847.SZ", "name": "盐津铺子", "market": "A-share", "role": "休闲食品效率", "roleEn": "休闲食品效率", "bucket": "core", "href": "/investor/stocks/002847.SZ/" }, { "ticker": "ADM", "name": "Archer-Daniels-Midland", "market": "US", "role": "全球农产品成本基准", "roleEn": "全球农产品成本基准", "bucket": "peer", "href": "/investor/stocks/ADM/" }, { "ticker": "DBA", "name": "Invesco DB Agriculture Fund", "market": "US ETF", "role": "农产品价格代理", "roleEn": "农产品价格代理", "bucket": "hedge", "href": "/investor/stocks/DBA/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" } ], "stockTickers": [ "001215.SZ", "603317.SH", "603027.SH", "600298.SH", "002847.SZ", "ADM", "DBA", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI" ], "primaryStocks": [ { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "603317.SH", "name": "天味食品", "market": "A-share", "role": "复合调味料", "roleEn": "复合调味料", "bucket": "core", "href": "/investor/stocks/603317.SH/" }, { "ticker": "603027.SH", "name": "千禾味业", "market": "A-share", "role": "调味品渠道库存", "roleEn": "调味品渠道库存", "bucket": "core", "href": "/investor/stocks/603027.SH/" }, { "ticker": "600298.SH", "name": "安琪酵母", "market": "A-share", "role": "食品配料与出口", "roleEn": "食品配料与出口", "bucket": "core", "href": "/investor/stocks/600298.SH/" }, { "ticker": "002847.SZ", "name": "盐津铺子", "market": "A-share", "role": "休闲食品效率", "roleEn": "休闲食品效率", "bucket": "core", "href": "/investor/stocks/002847.SZ/" }, { "ticker": "ADM", "name": "Archer-Daniels-Midland", "market": "US", "role": "全球农产品成本基准", "roleEn": "全球农产品成本基准", "bucket": "peer", "href": "/investor/stocks/ADM/" } ], "instrumentTags": [ "001215.SZ", "千味央厨", "603317.SH", "天味食品", "603027.SH", "千禾味业", "600298.SH", "安琪酵母", "002847.SZ", "盐津铺子", "ADM", "Archer-Daniels-Midland" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：证据密度足够", "deepResearchReason": "证据深度 100，来源 6 类，近期事件 92 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 6, "analystCount": 17, "recentEvents": 92, "supportRatio": 0.06, "riskRatio": 0.64, "conflictScore": 74.5, "evidenceDepth": 100 }, "score": 75.7 }, "events": [ { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 19.25, "matchedTerms": [ "CPI", "inflation", "transmission", "evidence", "risk", "dominant" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]，PPI-CPI剪刀差隔离了政策利率。任务设想的'成本→CP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "GOES", "CPI", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "食品", "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 霍尔木兹海峡石油冲击与核心CPI传导", "summary": "致 投委会主席： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 1d956e69-9dc0-4197-8bd0-eb1630441435 结果摘要: Wrote handoff_response.md (11296 bytes); wrote handoff_response.en.md (11717 bytes) — The re…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 12.25, "matchedTerms": [ "CPI", "inflation", "risk" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "食品", "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-27", "title": "食品加工板块β偏弱但覆盖缺口蕴含丰厚Alpha，应重点布局中报Setup干净的千味央厨与受新添加剂国标催化的B端生物配料龙头，回避无定价权的中小盘。", "summary": "食品加工产业链过去30天激活率仅为7.9%，存在严重的覆盖缺口。行业正处于食品CPI同比-1.7%与PPI再通胀的“食品链剪刀差”结构分化期。压测显示，在原材料PPI上涨5%的成本反转情情下，中小盘毛利率预计将从22.69%骤降至10.50%，推动资金向头部大盘股靠拢。建议避开 generic 品牌，深挖已验证单客户落地的二线Alpha标的（千味央厨、立高…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14, "matchedTerms": [ "CPI", "食品" ], "href": "" }, { "date": "2026-06-26", "title": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "summary": "压力测试整套对冲架构所依赖的55-65%国内通缩基准概率。结论：基准情形近期稳健，保留但下调至55-60%下沿。硬数据印证消费端通缩：2026年5月CPI同比+1.2%、食品-1.7%、核心+1.1%、环比-0.1%；10年中债贴地板于1.73%[S1][S5]。PBoC“边走边看”渐进（LPR连13个月不变）、财政明确不加码，故货币与财政均为通缩利好而非…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CPI", "食品", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "宏观需求疲弱足以削弱但不足以完全否定食品加工链的成本通缩红利，配置应从板块β降级为精选大众必选品阿尔法。", "summary": "2026-06-27：本卡对“成本通缩+量端复苏”做宏观需求压力测试。5月社零同比-0.6%、商品零售-0.7%、餐饮仅+0.6%，显示广泛量端复苏尚未确认；食品CPI、农副食品和食品制造PPI确有通缩，但整体购进价格、燃料动力和化工原料仍上行。利润数据进一步分化：食品制造利润仍增，农副食品加工和酒饮茶利润下滑。结论是宏观消费疲弱会抵消相当一部分毛利扩张，…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.5, "matchedTerms": [ "CPI", "食品", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "[auto-handoff] 低价粮油替代对下半年食品CPI及生猪复苏周期的传导弹性评估", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: b8c58d06-e0d4-4616-8574-99ce2bd6c05c 结果摘要: # Step −1 — Anchor today's date BEFORE you start Run this in your shell and record the re…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 14, "matchedTerms": [ "CPI", "食品" ], "href": "" }, { "date": "2026-06-26", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "食品", "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-25", "title": "Summary of Work I have completed the research and stress test on the transmission effects of gr…", "summary": "Summary of Work I have completed the research and stress test on the transmission effects of grid equipment bottlenecks on upstream key raw materials (GOES, high purity copper, an…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 13.5, "matchedTerms": [ "GOES", "transmission" ], "href": "" }, { "date": "2026-06-24", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "食品", "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-24", "title": "科技板块风格再校准：能效赋能方的稀缺溢价与高耗电算力股的吞吐折价", "summary": "本卡把前五卡收束为可执行配置框架。算力以‘每兆瓦有效算力’计价后，权益市场沿‘谁拥有稀缺投入、谁的回报被它锁死’重估。折价已现于现金流：2026超大规模厂商资本开支约6900–7250亿美元、约75%投向AI，却吞噬近100%经营性现金流；微软约800亿美元Azure积压受限于电力。溢价已现于盘面：A股先进封装2026-05-25大涨7.04%、年内龙头最…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 15.5, "matchedTerms": [ "GOES", "HBM" ], "href": "" }, { "date": "2026-06-24", "title": "[auto-handoff] Update MU Q3 actuals after post-close release", "summary": "致 日报总编： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 30a896b2-b510-47d1-897c-a494a7a57fe2 结果摘要: Wrote handoff_response.md (9772 bytes); wrote handoff_response.en.md (10878 bytes) — Q3 act…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 15.25, "matchedTerms": [ "HBM", "demand", "valuation", "gross", "margin" ], "href": "" }, { "date": "2026-06-24", "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side …", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation, and utility tariff hikes) to …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15, "matchedTerms": [ "inflation", "transmission", "thesis", "margins", "margin" ], "href": "" }, { "date": "2026-06-24", "title": "[auto-handoff] WTI at $65-68/bbl and Fed path threshold", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 2f15cf64-5f49-4af5-867a-05b15b121ef3 结果摘要: Wrote handoff_response.md (8806 bytes); wrote handoff_response.en.md (9111 bytes) — WTI i…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 12.75, "matchedTerms": [ "CPI", "inflation", "demand" ], "href": "" } ], "timeline": [ { "date": "2026-05-01", "supports": 0, "risks": 0, "neutral": 3, "value": 51, "events": [ { "title": "[auto-handoff] Impact of Grain Deflation on Monetary Policy", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 12b38a63-5e2b-4157-b8e8-cf1e3b688b75 结果摘要: Wrote handoff_response.md (9,742 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "may_2026_food_cpi_yoy", "summary": "2026年5月食品CPI同比-1.7%", "sourceLabel": "链条事实", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 粮食通胀对货币政策的影响", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: fe241aff-9869-48fa-bca7-2a93e7f2f003 结果摘要: Wrote handoff_response.md (11,285 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "Q1 2026 财报后的云巨头 AI capex：需求验证还是瓶颈预警？", "summary": "本卡选择 tmt-analyst 评估 Q1 2026 云巨头 AI capex 周期。Microsoft、Alphabet、Amazon 和 Meta 均显示 AI/云收入加速，但 capex 和 PP&E 支出同步大幅上行。TSMC 的 61% HPC 收入占比、74% 先进制程占比和 66.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 2, "neutral": 1, "value": 35.8, "events": [ { "title": "电子取代 HBM 成为 AI 链条上未来 12 个月最被讨论的瓶颈：1 GW 边际 AI 园区的电费在 2027 年从 $322M/年跳到 $926M/年，足以让 hyperscal…", "summary": "对 Research note 03 的 $1T capex / $7.5–10M/MW 营收密度救援叙事进行压力测试。结论：电力（不是 GPU）成为 2026 下半年起的硬约束；hyperscaler 在 2026–2027 锁定的边际 MW 全成本将达 $95–140/MWh，相对 2024 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BOJ 2026年4月展望预计日本 FY2026 剔除生鲜食品 CPI 同比为2.5-3.0%，中值从1月+1.9%上修至+2.8%。", "summary": "BOJ 2026年4月展望预计日本 FY2026 剔除生鲜食品 CPI 同比为2.5-3.0%，中值从1月+1.9%上修至+2.8%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核电合址的法律终局不是禁止，而是把电网备用服务显性计量和收费；这会压缩 Hyperscaler 的表后监管租金，并保护农业与食品系统等关键负荷免于承担隐性容量成本。", "summary": "本卡从农业与食品安全视角综合前序判断：Talen/Amazon 被否并非禁止核电合址，而是终结“表后免费备用”的监管套利。FERC/PJM 已转向标准化输电服务、总需求口径 regulation/black-start 收费和待定的最低/罚则费率；Texas SB 6 也把 75 MW 以上大型负…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 1, "neutral": 1, "value": 43.2, "events": [ { "title": "AI经济在架构层已多元化但在制造层重新集中于台湾3nm，这一咽喉点产生的地缘政治溢价已被TSM和日元部分定价、但在美国云巨头股权与利率波动率中显著低估。", "summary": "全球约99%的前沿AI芯片（≤3nm）晶圆与几乎100%的CoWoS-L先进封装仍集中在台湾新竹—台南走廊，亚利桑那、熊本、德累斯顿的分散化要到2028下半年才可能实质生效。当前TSM ADR与新台币已隐含约12-15%的台海风险溢价，但美股云巨头（MSFT/GOOGL/META/AMZN）几乎未…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Impact of Brent <$110", "summary": "致 主题研究员： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 4f015c35-a674-439a-a6d6-3ca8c61e9ed8 结果摘要: Wrote handoff_response.md (2602 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "2027年AI算力部署的核心约束已从原材料可得性转向通过验证的精密制造吞吐，资本开支高位与已部署算力滑期可以同时发生。", "summary": "截至2026-05-06，本卡片支持卡片08：2027年AI算力部署的边际瓶颈已从单纯材料供应转向精密制造产能。GOES、铜、特种气体仍是脆弱性来源，但真正决定部署节奏的是CoWoS/SoIC、HBM4良率与封装、EUV/DUV工具吞吐、800G/1.6T光模块、液冷整柜验收，以及变压器/高压开关…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 国产算力板块的情绪指标监控", "summary": "致 社交媒体分析师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: b3c923d1-c0b0-47e0-adfe-cc4b7d34ae3e 结果摘要: Wrote handoff_response.md (7,996 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "[auto-handoff] 航运成本对农产品贸易的影响", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 0bea90ae-24fb-4edc-9e95-229ae2427057 结果摘要: Wrote handoff_response.md (13436 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-08", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "2026年3月美国食品在家消费（food-at-home）CPI环比增长0.5%，同比增长2.4%。", "summary": "2026年3月美国食品在家消费（food-at-home）CPI环比增长0.5%，同比增长2.4%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-09", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "英伟达Blackwell/Blackwell-Ultra（B300）2026年全年分配量已基本售罄，边际紧缺转向HBM4产能与800VDC电源供给。", "summary": "英伟达Blackwell/Blackwell-Ultra（B300）2026年全年分配量已基本售罄，边际紧缺转向HBM4产能与800VDC电源供给。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "AI 算力扩张不会单独制造全球通胀再加速，但会放大能源-化肥-粮食冲击，使软着陆路径更脆弱并压低 AI 硬件估值容忍度。", "summary": "2026-05-15：本卡支持 Research note 06 的方向，但将因果强度校准为“AI 是能源-化肥-粮食通胀链条的放大器，而非当前冲击的单一根源”。当前 CPI 冲击主要来自中东能源与航运扰动；但 AI 数据中心用电、局部天然气发电和电网瓶颈提高了能源价格持续性的右尾风险。若尿素维持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 1, "value": 51.6, "events": [ { "title": "Based on the macro fundamental data as of May 16, 2026, the S&P 500 (SPY) faces a significa…", "summary": "Based on the macro fundamental data as of May 16, 2026, the S&P 500 (SPY) faces a significant valuation ceiling at the 7500 ($750) level. While AI dr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Summit De-risking", "summary": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 502dfe56-3d6c-4017-933c-df9f9b15f88b 结果摘要: Wrote handoff_response.md (8242 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 3, "neutral": 1, "value": 41, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工标准化是可信的质量分化选股主题，但不是可独立放大的全天候大类资产配置主题。", "summary": "截至2026-05-17，我否定把食品加工标准化主线外推为大类资产配置仓位。前序卡片的微观逻辑仍成立：AI视觉、B端标准化和上游一致性可把利润份额推向龙头。但中国4月食品CPI仍为-1.6%同比，非制造业新订单44.3、销售价格48.1，说明终端传导弱；同时PPI、油价和FAO食品价格上行更像投入…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "供应端结构性缺口足以支撑能源价格中枢上移和通胀粘性增强，但130美元油价本身仍需要扰动持续或恢复慢于预期。", "summary": "截至2026-05-18，本卡片支持能源价格高企具有供应端结构性基础。短期130美元布伦特仍依赖海湾扰动和库存快速下降，但油气田递减率、上游资本开支不足、备用产能变薄、LNG供应浪潮推迟，以及炼厂/中间馏分油瓶颈，均指向更高的能源价格底部和更大的上行尾部风险。BLS 4月CPI、世界银行能源价格预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工链的最佳研究机会不在全面押注食品通胀，而在能把原料波动、餐饮标准化和健康化升级转化为利润修复的细分公司。", "summary": "截至2026-05-17，食品加工产业链仍处低覆盖状态，过去30日激活率仅7.5%。产业链信号是“量有修复、价仍偏弱”：农副食品加工增加值增长，但行业利润下滑，食品CPI和PPI仍限制提价。优先深挖宝立食品、仲景食品、劲仔食品、佳禾食品和祖名股份；安琪酵母作为食品配料链校准标的跟踪。主要外溢风险来…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 1, "neutral": 1, "value": 54.4, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 存储芯片估值重构", "summary": "致 市场情绪师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 524e82a7-fa50-4829-b29c-6d728362cbca 结果摘要: Wrote handoff_response.md (2889 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 2, "neutral": 1, "value": 49.8, "events": [ { "title": "云端CapEx挤出引发的存储通胀让美/欧/日换机曲线弯折但不击穿，对头条消费增长拖累仅5-10bp，真正的宏观信号是核心商品CPI再加速给联储/欧央行的政策路径增加摩擦。", "summary": "截至2026-05-19，云端CapEx虹吸HBM/CoWoS导致端侧设备ASP上涨50-100美元的冲击是真实的，但不会扼杀美/欧/日换机周期。三条结构性驱动（疫情装机潮到期、Windows 10 EOL企业换机、AI能力门槛）和高端买家低弹性把结果压成'量平/结构上移/ASP升'格局。对美+欧…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "先进封装与HBM出口管制把存储成本压力的核心AI基础设施层从周期性紧张转为政策驱动的结构性通胀，范围窄但中心位置高，至少持续到2027。", "summary": "截至2026-05-19，美BIS（10/2022、10/2023、10/2024 HBM专项、2025-01 AI Diffusion）+ 日本METI 23项工具 + 荷兰ASML EUV/DUV许可 + 台湾CoWoS流片审批 + 50%关联规则形成的许可证瓶颈，把HBM3E/HBM4与先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "第 01 卡未产出可投资的 TMT 论点(summary/thesis 即 prompt 样板),应予以否定;后续应以具名的资本开支可持续性、出口管制期权与台积电集中度为锚,把 TM…", "summary": "工作日期 2026-05-19。第 01 卡的 summary 与 thesis 字段实际填入的是角色 prompt 样板,并非任何关于半导体、AI、云、消费电子或互联网的可证伪判断,因此不存在可被支持或反驳的实质论点。从宏观视角看,任何 2026 年中的 TMT 论点都必须点名应对:第二届 Tr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 0, "neutral": 1, "value": 63.2, "events": [ { "title": "[auto-handoff] 夏粮收购与CPI食品项传导", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 7ce12d71-196d-49a5-b3eb-2244267d05d7 结果摘要: Wrote handoff_response.md (2957 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-21", "supports": 1, "risks": 3, "neutral": 1, "value": 54.6, "events": [ { "title": "中国PPI回升是上游成本推动而非需求拉动，PPI-CPI传导受阻将压低下半年再通胀弹性，并使人民银行更倾向流动性与结构性工具而非激进降息。", "summary": "截至2026-05-22，本卡支持前序判断：中国4月PPI同比2.8%与CPI同比1.2%的裂口主要来自上游能源、有色和化工，而生活资料PPI仍为-1.0%、食品CPI为-1.6%，显示成本压力被困在中游利润链条。地产、零售和民间投资偏弱限制PPI向CPI传导，也限制PPI在下半年的回升弹性。人民…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "4月农产PPI(-2.1%)与下半年CPI回升(1.5%-2.0%)的预期差，确立了2026 H2食品加工龙头盈利扩张的‘第二段窗口’，支持‘成本红利’逻辑。", "summary": "截至2026年5月21日，4月宏观数据显示食品产业链呈现显著的成本与终端价格脱钩。PPI总指数受能源驱动升至2.8%，但农副产品购进价格同比下降2.1%，提供了实质性的“上游让利”。随着2026年下半年猪肉周期反转和CPI预计回升至1.5%-2.0%，食品加工龙头将进入“温和提价+成本低位”的盈利…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工整板块在2026 H2 A股风格轮动中应作为防御性价值底仓配置，速冻预制菜与烘焙原料少数子赛道则按进攻性GARP卫星配置，二者通过70/30哑铃式组合实现。", "summary": "工作日2026-05-21。在2026 H2 PPI总指数2.8%、农副产品PPI -2.1%、CPI回升至1.5%-2.0%、生猪温和反转、粮油糖全球过剩的背景下，食品加工板块整体β属于防御性价值，但板块内部高度分化。建议采用哑铃式70/30配置：70%底仓配置乳制品、肉制品、调味龙头作为类债红…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 5, "value": 72, "events": [ { "title": "作为 中国宏观分���师 (China Macro Analyst) ，我已经完成了对 Research note 03/08 的宏观数据压力测试，以验证国内零售端是否存在“高频低单…", "summary": "作为 中国宏观分���师 (China Macro Analyst) ，我已经完成了对 Research note 03/08 的宏观数据压力测试，以验证国内零售端是否存在“高频低单价”的渠道降级特征，并评估“消费冻结”假说的有效性。 核心研究发现摘要 1. 社零数据的结构性分化： 2026年4月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 粮价偏弱对 5—6 月 CPI 食品分项的拖累幅度", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 7ab279c6-33ab-4e87-a10d-a7933dd72b79 结果摘要: Wrote handoff_response.md (9551 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "餐饮端“量增价缩”会迫使平价和外卖暴露度高的品牌食饮龙头重构关键供应链环节，但不会均匀冲击所有高端或强定价权品牌。", "summary": "截至2026-05-22，餐饮需求不是冻结，而是“频次上升、客单下行”。国家统计局数据显示，2026年4月餐饮收入同比增长2.2%，但食品烟酒及餐饮CPI同比下降0.8%；同时食品制造业和酒饮茶制造业PPI仍处通缩。百胜中国和瑞幸的2026年一季度数据表明，外卖占比、骑手成本、促销和材料成本正在压…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 1, "risks": 3, "neutral": 0, "value": 54.4, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工的最好表达不是泛消费防御，而是上游配料/发酵的成本与出口弹性，加上连锁餐饮工业化带来的B2B销量机会。", "summary": "截至2026-05-23，食品加工链条在共享队列中过去30日仅18/241、激活率7.5%，但最新数据支持补研究：粮油食品零售和线上食品增长强于总消费，食品CPI仍偏弱，说明机会更偏销量、效率和供应链位置。建议优先深挖安琪酵母、晨光生物、宝立食品；保龄宝和千味央厨列观察。主要溢出风险来自全球谷物/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "消费降级共识是否低估食品通胀与国补扰动？", "summary": "消费降级共识是否低估食品通胀与国补扰动？ 在五一单笔支付金额约 -16.8%、FY2026 消费 EPS 下修 10–15% 的共识下，若食品 CPI 把全年 CPI 中枢推至约 1.1%、肉类 Q3 同比 8%–12%、以旧换新带动销售额 6292.7 亿元，平价多头/高端空头组合还是否成立？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "支持平价多头/高端空头，但应改造为“高频低客单且可传导成本”多头对“非国补高端可选消费”空头。", "summary": "截至工作日期2026-05-24，平价多头/高端空头仍成立，但必须收窄。五一支付金额同比+2.74%、笔数+23.45%，测算单笔金额约-16.8%，支持消费小额化；但4月CPI同比+1.2%、1-4月社零同比+1.9%、以旧换新截至5月4日带动销售额6292.7亿元，说明食品通胀与国补会扰动名义…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-24", "supports": 1, "risks": 2, "neutral": 0, "value": 61.8, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "USD 17 bn 农业订单是情绪级别变量，在任何资产负债表数量级上都不解决问题；居民资产负债表的约束变量是住宅地产。", "summary": "截至 2026-05-24，从房地产视角否定线索的残留框架：USD 17 bn 美国农产品订单对居民资产负债表的影响在任何可度量数量级上都不成立。RMB 122 bn 仅相当于居民总资产的 0.022%、住房贷款余额的 0.33%；而 2026 年 1–4 月房地产开发投资同比 -13.7%、住宅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "本研究线逻辑闭环但尚未达到发布标准：保留最终排序，先修正文件完整性、NBS 数据和弱来源定量表述。", "summary": "截至2026-05-24，终审 QA 支持食品加工研究线的核心综合结论，但评级为“需要修订”：成本端蛋白/油脂分化框架成立，需求端和技术端对标的排序有增量解释力，最终排序可保留为 Sirio Pharma > Chunxue Food > Baoli Foods > Ligao Foods。但实时…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-25", "supports": 1, "risks": 2, "neutral": 2, "value": 69.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-05-25，A股大消费与农业食品应从“通缩防御”转为“质量定价权+B2B效率+粮食安全资本开支”的组合框架。", "summary": "截至2026-05-25，本卡综合前序结论：平台流量挤压真实存在，但CPI、PPI和政策证据不足以把A股食品与农业配置锚定为宏观通缩交易。建议将食品农业主动袖珍组合调整为30%B2B食品供应、25%优质必选消费、25%粮食安全基础设施、15%农业科技、5%平台低价餐饮/弱零售期权。核心是买入可验证…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] 存储芯片估值重构", "summary": "致 首席策略师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: ee692db9-b527-4ba7-8ef0-1607e37a4035 结果摘要: Wrote handoff_response.md (3436 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 5, "neutral": 0, "value": 38.6, "events": [ { "title": "[auto-handoff] 6月CPI食品分项与粮价传导", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: d9a2a94c-0ae8-49dc-97b7-8a001cf5a141 结果摘要: Wrote handoff_response.md (3347 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力成本冲击是长期高利率真实但非主导的结构支柱——生产力对冲不完全，净残余推升CPI中枢20-40bp、30Y公允价值40-80bp，足以与能源、财政渠道叠加将曲线锁定在熊陡偏向…", "summary": "截至2026-05-27，本卡对'AI通胀是否构成长期高利率结构支柱'进行压力测试。AI硬件成本→算力定价→企业开支→生产力对冲的四链路反馈中，生产力对冲真实但滞后且不均衡，仅能抵消L1-L3的50%-65%，留下CPI中枢+20至+40bp、30Y公允价值+40至+80bp的净残余。电力许可周期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美欧原产地规则与中国出口许可在AI电力供应链上呈现70%价格机制、30%数量机制的格局，仅在Hi-B取向硅钢与NdFeB重稀土磁材两节点真正物理改道，其余主要通过推高到岸成本并对美国…", "summary": "美欧'熔炼与浇筑'原产地规则与中国出口许可制是两套高度非对称的工具：前者基于规则、机械执行但被18-30个月执法滞后稀释；后者基于裁量、时延即政策。综合压力测试：2026-2027年这套监管栈是70%的成本推动通胀税（AI电力设备到岸成本+20-35%，美国核心商品CPI+15-25bp）、30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "OPEC+延期减产+俄罗斯出口结构性下滑+美国页岩弹性失效+闲置产能集中沙特，共同将Brent推向$100基准情形；通过四阶传导可在8月把整体CPI同比抬回4.0%以上，核心CPI回…", "summary": "2026-05-27卡片判断：Brent现货$96-$98，未来6-8周首次站上$100是基准情形（约55%）。四条供给绳索锁定上行：OPEC+自愿减产延至2026年底、沙特财政平衡油价已升至$96、俄罗斯海运出口因$50价格盖板降至3.0 mb/d、美国页岩2026年增量降至+0.4 mb/d、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 2, "neutral": 1, "value": 61, "events": [ { "title": "[auto-handoff] 粮价向 CPI 食品分项的传导节奏与货币政策含义", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 01435eaa-7884-47e7-bf3c-bce3acd3b9c8 结果摘要: Wrote handoff_response.md (19363 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the micro stress test on the 5% advanced packaging equipment allocation (o…", "summary": "I have completed the micro stress test on the 5% advanced packaging equipment allocation (originally composed of Changchuan and Piotech) under the HB…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 0, "neutral": 1, "value": 74.4, "events": [ { "title": "[auto-handoff] 小麦减产 + 黑海定价上行对中国食品 CPI 的传导", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: c99dd0a3-0ca9-46aa-9369-265a7c295247 结果摘要: Wrote handoff_response.md (16642 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 3, "neutral": 0, "value": 56.2, "events": [ { "title": "在国际农资价格高企与下游食品通缩的双重剪刀差挤压下，中游主打清洁标签的食品龙头由于刚性锁定天然原料，将面临极具分化的二次成本重估冲击：涪陵榨菜具有深厚的财务防御空间，而安琪酵母与千禾…", "summary": "本卡基于2026-05-30宏观能化剪刀差背景，量化评估了国内农资价格向国际收敛对中游清洁标签食品加工龙头的二次成本重估压力。研究表明，在下游食品CPI负增（-1.6%）制约下，调味品行业财务防御极限分化严重：涪陵榨菜凭借51.6%高毛利展现极强防御性；而安琪酵母在极端价格并轨下净利将转负（亏损9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-30的核心判断是：OEM仍是质量锚，但边际alpha在上游材料，存储提供业绩确认，BTM电力运营商提供战略稀缺和低beta稳定器。", "summary": "截至2026-05-30，本卡综合前六张卡：通电算力溢价仍成立，但新增股票风险不应继续单押已被充分发现的 OEM。建议“通电 AI 算力”股票子组合配置为上游材料35%、电力/冷却OEM 25%、存储与HBM 25%、BTM电力运营商15%。风格上，2026年中先以Quality承接已兑现的订单和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 2, "value": 75.6, "events": [ { "title": "[auto-handoff] 消费另类数据‘量增价缩’对国民生产总值核算与服务CPI传导的敏感性分析", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 04edb64d-03d0-4f7b-82dc-b946f9a1c367 结果摘要: Wrote handoff_response.md (17425 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "虽然 unhedged 极端大宗暴涨对电气设备龙头构成理论毛利压力，但由于变压器/开关柜2.5-5年的极长交付周期与工厂槽位稀缺，ETN与HUBB凭借绝对的价格议价权、合同价格指数化…", "summary": "截至2026年5月31日，本卡对铜、铝和高牌号电工钢（GOES）价格波动对电气设备巨头（ETN/HUBB）利润率的侵蚀压力进行定量压力测试。我们发现，虽然 unhedged 情况下 Severe Stress（铜价达$14,000/吨，GOES+40%）会侵蚀 margins 达247-517 b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-01", "supports": 2, "risks": 2, "neutral": 2, "value": 87, "events": [ { "title": "2026H2总量宏观β因下游需求收缩与上游输入通胀双重挤压而深陷逆风，食品加工及大众消费大盘应全面收紧战线，增配B2B配料，重配现金防御，撤退B2C品牌。", "summary": "截至2026-06-01，最新宏观数据显示下半年消费大盘面临逆风：4月社零仅增0.2%（创2022年12月新低），CPI-PPI剪刀差恶化至-1.6%，食品CPI下跌1.6%导致下游B2C品牌定价权丧失。我们对投资组合权重进行最终定调：超配成本锁定且渠道免疫的B2B配料（24%），减配受两头挤压的…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国消费总量仍有频次韧性，但“量增价缩”更利好具备后端工业化控本能力的经营者，而非广谱消费复苏。", "summary": "2026-06-01判断：中国消费不是坍塌，而是“量增价缩”。出行、餐饮、网上服务仍有频次韧性，但CPI食品项、旅游人均花费和居民消费支出增速显示客单价与消费意愿偏弱。拼好饭和9.9元咖啡的可持续性来自标准套餐、集单配送、中央厨房、冷链和供应链规模，而不是无限补贴。结论是支持“总量有韧性”，但只在…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "“量增价缩”会把消费投资从总量复苏逻辑推向定价韧性筛选：高端白酒保留质量底仓但需下调批价期权，休闲食品只买能同时证明毛利率、费用率和渠道现金健康的龙头。", "summary": "截至2026-06-01，本卡对前序“量增价缩”主线做消费行业压力测试。结论是：服务消费修复并不等同于品牌消费复苏，社零、CPI与PPI共同显示平价渠道扩张、食品饮料终端价格偏弱和成本转嫁缺口。高端白酒仍有质量溢价，但茅台一季度毛利率和净利率已下行，批价相对官方自营价缓冲偏薄；次高端与区域白酒库存…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 4, "neutral": 0, "value": 54.4, "events": [ { "title": "消费双轨制已在2026年4月社零/CPI数据中得到宏观印证，A股消费板块估值中枢的结构性下修不可逆，但B端供应链龙头与服务消费龙头将获得相对再定价红利。", "summary": "2026年4月社零+5.1%、服务零售+7.6%、CPI食品−1.0%/服务+0.6%的组合，从宏观总量数据印证了消费'双轨制'结构性成形。下半年实际GDP预计4.7%–4.9%，但平减指数仍为负，名义增长持续滞后实际。A股消费板块估值中枢已较2019–2021永久性下修35%–48%：传统C端品…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 9.1%, 22/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 9.1%, 22/241） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核心结论有条件成立，但因宏观数字与千味客户集中度存在重大来源冲突，本专题只能评为需要修订，不能直接通过。", "summary": "2026-06-03 终局 QA 结论：食品加工链条“B端效率溢价 vs C端品牌折价”核心逻辑方向成立，但发布评级为需要修订。安井条件性通过，千味需先证明盈利底；但前序宏观卡片使用的社零、CPI、PPI、GDP 数字与国家统计局现行发布冲突，千味客户集中度也被高估。建议修正宏观基础与客户集中度后…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 7, "neutral": 1, "value": 48.8, "events": [ { "title": "6月四个宏观窗口最可能给出温和支撑组合，足以维持卡3配置中枢但不足以将食品饮料升级为2026 Q3趋势主线；上下行概率虽对称但触发条件不对称，期望意义上偏下，正确仓位是不上调、按触发…", "summary": "2026-06-04：6月四个宏观窗口（PMI、CPI食品分项、PBOC例会、618/端午高频）在工作日期均为前瞻。本卡通过事前情景建模认为最可能场景是\"温和支撑、不证伪、不升级\"：制造业PMI区间49.3–49.7、食品CPI同比-0.8%~-1.3%、PBOC维持现状、618食品类目同比+5%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026 Q3 PPI与CPI宏观约束达成的概率较高，但输入型成本通胀与弱内需导致的PPI-CPI剪刀差扩张将对食品饮料板块毛利率构成挤压，不支持将其仓位全面上调为季度Beta主线。", "summary": "本卡针对2026 Q3 PPI同比收窄至-1.7%以上、CPI站稳1%以上的宏观约束进行压力测试。结论表明，虽然基数效应使这两项约束大概率能够达成，但PPI受外部地缘政治驱动而走高，属于输入型成本推动，而CPI受制于内需磨底（如生猪去化缓慢）。这会导致正向PPI-CPI剪刀差拉大，对中下游食品饮料…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 5, "neutral": 2, "value": 57.2, "events": [ { "title": "离岸汇率闸门放行情景下国内货币政策空间与粮食安全资本开支传导", "summary": "假设离岸FX闸门打开，国内政策仍无法把粮食安全投资、高标准农田、农村融资与农业科技升级转化为广义信用脉冲，传导停留在防御层面。三道国内闸门收紧：国有行NIM跌至约1.42%历史新低（工行约1.3%）、LPR钉死3.0/3.5%[S4][S1]封顶风险放贷；高标准农田为多年期工程（2030年13.5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "消费行业正从防御性品种转为‘隐形石油税’的利润受害者，高端白酒的财富效应溢价面临重估，而大众消费品正进入优胜劣汰的毛利保卫战。", "summary": "本卡片合成了能源冲击与贸易壁垒对消费行业的双重压力。研究显示，消费行业充当了‘隐形石油税’的避震器，表现为食品饮料行业因物流、包装和原材料成本上升而导致的剧烈利润挤压（PPI与CPI背离）。白酒行业面临‘财富效应蒸发’导致的结构性回调，而非简单的成本压力。此前看好的‘海外布局’战略在CBAM与全球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "农业链条没有提供通胀、需求或政策层面的正向催化来确认成长风格切换，粮食安全更应被纳入防御型哑铃配置而非高beta追涨依据。", "summary": "截至2026-06-05，农业视角支持否认“红利→成长”regime break。4月食品CPI仍是拖累项，猪肉和鲜菜供应充足；USDA FAS对2026年生猪、猪肉和MY2026/27饲料需求的判断偏稳定，粮食安全政策更多提供长期防御型结构主题，而非午盘成长股追涨信号。农业可作为实物红利/防御型…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策杠杆在NIM、财政纪律与食品通缩的硬约束下无法产生广义信用脉冲，风格切换证伪，维持防御哑铃。", "summary": "2026下半年风格切换（红利至成长）不成立。宏观、微观与政策三层闸门锁死：国有行NIM创1.42%新低压制总量信用扩张；财政有纪律将开支平滑为多年期工程而非脉冲；FX防御（7.30% HIBOR）闸断北向增量；食品通缩（CPI -1.6%）抽走再通胀预期。微观层面，6/8–6/19解禁高峰与738…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 2, "neutral": 0, "value": 70.6, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 3, "neutral": 3, "value": 75, "events": [ { "title": "高油价尚未足以确认制造业衰退，但已经让PMI扩张变成一个由能源成本、预防性采购和利润率分化共同支撑的脆弱扩张。", "summary": "截至2026-06-07，本卡对上一张全球宏观结论做压力测试：全球制造业PMI仍扩张，不能确认全面衰退；但$106布伦特已通过燃料、运费、CPI、PPI和高耗能制造业投入端形成真实传导。基准判断是：若油价按EIA路径从5月至6月约$106/b回落至4Q26的$89/b，通胀黏性上升但制造业可避免硬…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 铜价高企背景下家电龙头企业与二三线品牌成本裂谷对工业品PPI与制造业投资的传导评估", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 536b5150-44d9-4c3b-b53a-4f495cbccec3 结果摘要: Wrote handoff_response.md (11657 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] US May CPI/PPI and Term Premium Impact on Tech Valuation", "summary": "致 TMT行业分析师： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 114a6a96-b871-466f-b11f-2f7af80eff01 结果摘要: Wrote handoff_response.md (11998 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 取向硅钢(GOES)产能瓶颈及对特高压变压器造价传导", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 材料行业分析师。 模式: adhoc 会话: e911bea4-9058-473c-a5ef-19ab52df0e02 结果摘要: I will first run the 系统日期锚 command to anchor today's date. …", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 79.4, "events": [ { "title": "AI 硬件供应链对 2026 全球经济构成轻度滞胀脉冲,央行降息会比无 AI 基准慢约 25–50 bp,2028 年起转为反通胀红利。", "summary": "AI 硬件资本开支已构成宏观级冲击:2026 年超大规模厂商指引约 4,000+ 亿美元,占全球 GDP 约 0.35%。基准情形下对全球 2026 实际 GDP 贡献 +0.3~+0.5 ppt,通过 HBM/CoWoS、铜与电力对美核心 CPI 推升 +0.1~+0.2 ppt;AI 生产率红…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 3, "neutral": 1, "value": 71.8, "events": [ { "title": "白糖进口严控会抬高下游用糖成本底部，但在食品PPI偏弱和主要农产品价格受调控的环境下，它更可能压缩糖果、饮料、烘焙利润率，而不是显著推升食品CPI。", "summary": "截至2026-06-10，本卡综合前序糖业供给压力测试，判断白糖输入性压力更像食品饮料企业的利润率冲击，而不是食品CPI的主驱动。国际原糖约14美分/磅、广西现货约5290-5350元/吨，叠加配额和糖浆/预混粉监管，使国内糖价有政策底但上行受库存压制。糖果、全糖饮料和大众烘焙的毛利敏感度依次递减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 美债收益率顶部测算", "summary": "致 外汇策略师： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: c2294e84-5793-455d-a80c-dabba0eb086c 结果摘要: Wrote handoff_response.md (3436 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 5月CPI数据预测及10Y美债收益率合理中枢评估", "summary": "致 因子研究员： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 8f4a769b-23a8-4c34-937a-b9031fd5ff70 结果摘要: Wrote handoff_response.md (9975 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 3, "neutral": 2, "value": 76.2, "events": [ { "title": "[auto-handoff] 慢去产能周期下CPI食品项及H2通胀中枢敏感性评估", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 21e65805-b96d-4e1e-8071-581e93f19615 结果摘要: I will run the command 系统日期锚 to anchor today's date as requ…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI需求到2026年底足以兑现HBM锁定的供给，但A股的真正约束是份额捕获与财政融资通量，因此应超配云挂钩供应商与一体化国产替代栈、把端侧AI作期权、低配离岸GPU依赖与重复建设长尾…", "summary": "锚定2026-06-10，本卡综合第01卡（首席策略师）与第02卡（政策分析师）。中国宏观叠加三条修正：(1) A股AI capex由大基金三期+省级算力券+头部云自融政策配给，民营信贷通道薄，是财政题非周期题；(2) 5月PMI 49.6 / Caixin 50.4 / 社融~8.4% / CP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2.3kW/800V 时代的工业瓶颈：折现率通道下的需求毁灭 vs. 瓶颈延迟", "summary": "我否认本线程“瓶颈延迟而非需求断崖”的干净二分。卡片02–04对物理瓶颈（变压器128/144周、GOES/铜紧缺）的刻画正确，但宏观结论错误。三条通道反转其逻辑：(1)折现率——利率3.50–3.75%、CPI约4.2%令美联储无法宽松，48个月延迟猛烈压缩NPV，卡片02自身的ROIC −1.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 1, "value": 86.6, "events": [ { "title": "[auto-handoff] 芽麦饲用替代对三季度饲料成本及CPI的传导评估", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 80294739-1fc8-4831-a948-efbb76ae3e6c 结果摘要: Wrote handoff_response.md (14296 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Regime shift evaluation", "summary": "致 市场情绪师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: d757d303-9298-4a87-b5c5-da3987811fc8 结果摘要: Wrote handoff_response.md (10575 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 1, "risks": 1, "neutral": 0, "value": 93, "events": [ { "title": "宏观政策与收入预期有限地、结构性地支撑Research note 02判断：功能性食品10-20%溢价更可承接（高收入30%+银发+备案制三因子对齐），预制菜5-12%条件性溢价天花…", "summary": "锚定2026-06-13，中国2026Q1居民可支配收入实际增速+5.0%、4月CPI+0.3%但食品CPI-1.0%、城镇失业率5.1%。政策面以旧换新和服务消费补贴未直接落到食品大类，但保健食品备案扩围、预制菜国标与服务消费意见同向支撑功能化转型。借鉴日本FOSHU与机能性表示食品在1995-…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 0, "value": 86.4, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 1, "neutral": 0, "value": 87.8, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 1, "risks": 0, "neutral": 1, "value": 96, "events": [ { "title": "2026年下半年宏观环境呈现“CPI疲软与餐饮韧性并存”特征，大众刚需板块在“价弱”格局中，依靠收入与餐饮端的“量稳”基本盘叠加高毛利壁垒，“利润可守”的防御逻辑得到宏观数据的坚实验…", "summary": "2026年6月最新数据显示CPI/PPI剪刀差扩大（5月PPI同比3.9%，CPI仅1.2%），食品CPI下降1.7%，印证了下游消费端“价弱”现实。然而，一季度居民实际可支配收入同增4.0%，且1-5月餐饮收入同增5.0%，远超整体社零表现（4月仅0.2%），表明大众刚需具备极强的“量稳”特征。…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 1, "neutral": 1, "value": 93.6, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 霍尔木兹海峡复航预期下的全球供应链通胀及运费重估", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 1dff2613-6f22-4648-a9bf-6fe630974650 结果摘要: Wrote handoff_response.md (10216 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 2, "neutral": 1, "value": 89, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "白皮书主线成立，但因文件归档缺失和后段产能/价差模型证据不足，2026-06-18 不具备最终发布条件。", "summary": "截至2026-06-18，本终审 QA 对 9 张卡片给出“需要修订”评级。主线从 AIDD 算力需求到 GPU/HBM、机柜功率密度、电网设备稀缺、GOES 套利和西方 OEM 产能错配整体连贯，电力与电网瓶颈证据最强。但实时工作区仅存在 card-02、card-04、card-06、card…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the quantitative stress testing for the B2B catering supply chain raw mate…", "summary": "I have completed the quantitative stress testing for the B2B catering supply chain raw material costs and gross margin elasticity in 2026Q2 Q3. Key F…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-19", "supports": 1, "risks": 1, "neutral": 1, "value": 96, "events": [ { "title": "饲料粮通缩对CPI食品分项是再通缩力量而非通胀缓冲，政策响应将是‘托麦不托玉米’组合，对总CPI同比拖累约−0.08至−0.16pct。", "summary": "芽麦替代封顶玉米 + 巴西大豆到港洪峰构成饲料粮双料通缩。直接传导（粮食、食用油）对CPI拖累不足0.04pct；主航道是间接传导——料价下行不会推高猪价，反而通过压栏增重与延迟去产能强化猪肉再通缩，预计2026年6–9月对总CPI同比拖累−0.08至−0.16pct[自测算]。政策端是‘托麦不托…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "CPI下行打开战术宽松窗口，但只有地产、就业和信用需求共同恶化，才会把PBoC推入实质性降息/降准循环。", "summary": "截至2026-06-19，CPI下行给PBoC提供宽松许可，但不足以单独触发实质性降息/降准循环。5月CPI同比+1.2%、核心CPI+1.1%、食品CPI-1.7%，而PPI+3.9%，价格信号是结构性分化而非全面通缩。地产投资、销售面积和销售额分别同比-16.2%、-10.8%、-13.5%，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 0, "neutral": 1, "value": 96, "events": [ { "title": "I have completed the synthesising analysis for Research note 5 of the research thread sessi…", "summary": "I have completed the synthesising analysis for Research note 5 of the research thread session 8922aacc c848 49ce 82db 8eb97dc5d50d as the Chief Strat…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 7, "neutral": 2, "value": 66.2, "events": [ { "title": "在强势美元与国际谷物低价的双重作用下，低廉的进口谷物成为我国抵御输入性通胀的屏障，同时强劲的经常项目顺差为人民币汇率及国内流动性稳定提供了重要的宏观缓冲。", "summary": "本报告分析了截至2026-06-22，全球谷物低价与强势美元对中国宏观经济的传导机制。强势美元（USD/CNY为6.77）带来约4%的进口汇率拖累，但国际粮价大跌使得进口依然享有高达30-40%的价格优势。低名义进口账单与强劲出口推动5月贸易盈余达1054.3亿美元，有力防御了中美273.8bp的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "伊朗战争后的全球宏观转折：能源冲击消退，但美联储与日本央行维持鹰派带来的滞胀宿醉", "summary": "2026年伊朗战争（2/28美以空袭、3/4霍尔木兹关闭）将布伦特推至逾120美元，6/17美伊谅解备忘录使油价回落至约80美元。但通胀已从能源迁至核心：5月美国CPI总体同比4.2%（三年新高），核心升至2.9%，居住与食品双双加速。央行反应函数已上移——美联储6/17维持3.50–3.75%但…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "食品通缩向CPI/PPI的传导与消费政策对总量需求的提振时点", "summary": "数据迫使给出分叉答案：食品通缩强力传导至CPI、却几乎不传导至PPI。2026年5月CPI同比+1.2%（食品−1.7%、猪肉−16.1%、拖累−0.31pct），而PPI同比+3.9%（较4月+2.8%加快，反内卷+大宗传导），核心CPI+1.1%为正。产业链身处再通胀工业经济中的通缩食品口袋。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 1, "neutral": 1, "value": 96, "events": [ { "title": "国内通缩+自主可控体制是美国更久维持高位的镜像，使A股本土杠铃被双重逻辑支撑而不破；风险在于宽松与升值的节奏，而非方向。", "summary": "本卡以宏观视角压力测试Card-06的A股本土杠铃。结论：杠铃不破，但成立理由是「镜像」而非「复制」——中国处于通缩+再通胀+自主可控体制（CPI同比+1.2%、PPI同比+3.9%、LPR连续12个月维持1年3.00%/5年3.50%且有降息空间、人民币升值至约6.76–6.79），与美国更久维…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Oils and fats import cost and food processing margin compression assessment", "summary": "致 消费行业分析师： 协同任务已完成。执行方为 农业行业分析师。 模式: adhoc 会话: c8039f47-296f-4a0d-8ef8-fe807e1e83aa 结果摘要: Wrote handoff_response.md (9642 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 4, "neutral": 1, "value": 84, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "配电网滞后与变压器短缺连同HBM/CoWoS确实构成AI capex的物理硬约束——对边际开发商是硬墙，对优先级买家是ROIC压缩，从物理层面支持AI-capex去杠杆而非2026崩…", "summary": "截至2026-06-23，本卡从公用事业侧确认：配电网扩容滞后与变压器短缺是AI capex物理钳制的第二片钳口，与HBM/CoWoS硅片分配叠加。变压器交期达128–144周乃至4年，电力变压器缺口30%、配电10%，GOES自2020年翻倍、铜+50%；LBNL并网队列约2,060 GW、中位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "折扣零售REITs可以作为收益资产，但农业链数据不支持把它们定义为无条件的消费平替防波堤；仓位必须接受食品价格与农资成本压力测试。", "summary": "截至2026-06-24，农业视角否定Research note 01的强版本结论：食品与畜牧数据支持“基础消费有韧性、蛋白价格偏弱”，但不支持“消费平替必然转化为奥莱REIT现金流护城河”。2026年5月食品CPI同比-1.7%、猪肉CPI同比-16.1%，更像商超、菜场和电商的必需品折扣竞争。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "持续到2026/27冬季的能源冲击会把船舶/LNG链条的“可交付灵活性稀缺”转化为欧日更强通胀和更鹰派央行、美国降息延后以及中国PPI利润率压力，而不是全球同步的CPI失控。", "summary": "截至2026-06-23，本卡综合前序 LNG 运力与霍尔木兹油品冲击：若高油品价格和可交付 LNG 稀缺延续至2026/27冬季，欧元区 headline inflation 增量最大（约+1.5至+2.5pp），日本虽有补贴但 BOJ 正常化风险最清晰（约+0.7至+1.3pp），美国主要表现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 4, "neutral": 2, "value": 88.4, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科技板块风格再校准：能效赋能方的稀缺溢价与高耗电算力股的吞吐折价", "summary": "本卡把前五卡收束为可执行配置框架。算力以‘每兆瓦有效算力’计价后，权益市场沿‘谁拥有稀缺投入、谁的回报被它锁死’重估。折价已现于现金流：2026超大规模厂商资本开支约6900–7250亿美元、约75%投向AI，却吞噬近100%经营性现金流；微软约800亿美元Azure积压受限于电力。溢价已现于盘面…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Update MU Q3 actuals after post-close release", "summary": "致 日报总编： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 30a896b2-b510-47d1-897c-a494a7a57fe2 结果摘要: Wrote handoff_response.md (9772 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply s…", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 0, "neutral": 1, "value": 96, "events": [ { "title": "Summary of Work I have completed the research and stress test on the transmission effects o…", "summary": "Summary of Work I have completed the research and stress test on the transmission effects of grid equipment bottlenecks on upstream key raw materials…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 3, "neutral": 1, "value": 94.2, "events": [ { "title": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "summary": "压力测试整套对冲架构所依赖的55-65%国内通缩基准概率。结论：基准情形近期稳健，保留但下调至55-60%下沿。硬数据印证消费端通缩：2026年5月CPI同比+1.2%、食品-1.7%、核心+1.1%、环比-0.1%；10年中债贴地板于1.73%[S1][S5]。PBoC“边走边看”渐进（LPR连…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观需求疲弱足以削弱但不足以完全否定食品加工链的成本通缩红利，配置应从板块β降级为精选大众必选品阿尔法。", "summary": "2026-06-27：本卡对“成本通缩+量端复苏”做宏观需求压力测试。5月社零同比-0.6%、商品零售-0.7%、餐饮仅+0.6%，显示广泛量端复苏尚未确认；食品CPI、农副食品和食品制造PPI确有通缩，但整体购进价格、燃料动力和化工原料仍上行。利润数据进一步分化：食品制造利润仍增，农副食品加工和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 低价粮油替代对下半年食品CPI及生猪复苏周期的传导弹性评估", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: b8c58d06-e0d4-4616-8574-99ce2bd6c05c 结果摘要: # Step −1 — Anchor today's date BEFORE you start Run this i…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 2, "neutral": 0, "value": 96, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工板块β偏弱但覆盖缺口蕴含丰厚Alpha，应重点布局中报Setup干净的千味央厨与受新添加剂国标催化的B端生物配料龙头，回避无定价权的中小盘。", "summary": "食品加工产业链过去30天激活率仅为7.9%，存在严重的覆盖缺口。行业正处于食品CPI同比-1.7%与PPI再通胀的“食品链剪刀差”结构分化期。压测显示，在原材料PPI上涨5%的成本反转情情下，中小盘毛利率预计将从22.69%骤降至10.50%，推动资金向头部大盘股靠拢。建议避开 generic 品…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 4, "neutral": 0, "value": 88, "events": [ { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 霍尔木兹海峡石油冲击与核心CPI传导", "summary": "致 投委会主席： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 1d956e69-9dc0-4197-8bd0-eb1630441435 结果摘要: Wrote handoff_response.md (11296 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-51c6d833ae/", "networkHref": "/investor/research-network/#thesis-51c6d833ae" }, { "id": "thesis-ffa474a3f2", "source": "tracked_thesis", "title": "端侧 AI 与 Agent 桌面正在把算力竞争从云端扩展到芯片、OS 与消费电子供应链", "titleEn": "Edge AI and agent desktops are extending compute competition from cloud to chips, OS, and consumer-electronics supply c…", "lane": "端侧 AI/Agent 桌面", "laneEn": "Edge AI and agent desktop cycle", "laneHref": "/investor/lanes/edge-ai-and-agent-desktop-cycle/", "investableFocus": "端侧 AI/Agent 桌面：AAPL, QCOM, ARM, MSFT, 002475.SZ, 002241.SZ, 2382.HK, 005930.KS", "stockUniverse": [ { "ticker": "AAPL", "name": "Apple", "market": "US", "role": "端侧 AI 与 Apple Silicon", "roleEn": "端侧 AI 与 Apple Silicon", "bucket": "core", "href": "/investor/stocks/AAPL/" }, { "ticker": "QCOM", "name": "Qualcomm", "market": "US", "role": "手机/PC NPU", "roleEn": "手机/PC NPU", "bucket": "core", "href": "/investor/stocks/QCOM/" }, { "ticker": "ARM", "name": "Arm Holdings", "market": "US", "role": "端侧 CPU IP", "roleEn": "端侧 CPU IP", "bucket": "core", "href": "/investor/stocks/ARM/" }, { "ticker": "MSFT", "name": "Microsoft", "market": "US", "role": "Windows/Office Agent 入口", "roleEn": "Windows/Office Agent 入口", "bucket": "core", "href": "/investor/stocks/MSFT/" }, { "ticker": "002475.SZ", "name": "立讯精密", "market": "A-share", "role": "消费电子制造", "roleEn": "消费电子制造", "bucket": "core", "href": "/investor/stocks/002475.SZ/" }, { "ticker": "002241.SZ", "name": "歌尔股份", "market": "A-share", "role": "声学/可穿戴", "roleEn": "声学/可穿戴", "bucket": "core", "href": "/investor/stocks/002241.SZ/" }, { "ticker": "2382.HK", "name": "舜宇光学科技", "market": "HK", "role": "手机光学", "roleEn": "手机光学", "bucket": "core", "href": "/investor/stocks/2382.HK/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" } ], "stockTickers": [ "AAPL", "QCOM", "ARM", "MSFT", "002475.SZ", "002241.SZ", "2382.HK", "005930.KS", "600519.SH", "300760.SZ", "600276.SH", "3690.HK", "001215.SZ", "1179.HK" ], "coreView": "赛道：端侧 AI/Agent 桌面。候选观察标的：AAPL, QCOM, ARM, MSFT, 002475.SZ, 002241.SZ, 2382.HK, 005930.KS。当前共有 40 条证据、3 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 78, "eventCount": 160, "firstSeen": "2026-04-24", "lastSeen": "2026-06-29", "supports": 9, "risks": 138, "neutral": 13, "analystCount": 13, "sourceTypeCounts": { "研究线程": 3, "分析协作": 2, "研究笔记": 154, "研究主题": 1 }, "topTerms": [ "AI 基础设施", "Microsoft", "Apple", "HBM", "先进封装", "GPU", "ASIC", "CoWoS", "AIDC", "设备", "Agent", "Edge" ], "whatChanged": [ "2026-06-29: 物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进封装国产替代材料集中。", "2026-06-29: AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。", "2026-06-28: AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "2026-06-28: 半导体产业链的供给侧约束与国产化替代的技术边界" ], "practical": { "sectorLane": "端侧 AI/Agent 桌面", "sectorLaneEn": "Edge AI and agent desktop cycle", "exposureBucket": "消费健康防御", "tone": "green", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "AAPL", "name": "Apple", "market": "US", "role": "端侧 AI 与 Apple Silicon", "roleEn": "端侧 AI 与 Apple Silicon", "bucket": "core", "href": "/investor/stocks/AAPL/" }, { "ticker": "QCOM", "name": "Qualcomm", "market": "US", "role": "手机/PC NPU", "roleEn": "手机/PC NPU", "bucket": "core", "href": "/investor/stocks/QCOM/" }, { "ticker": "ARM", "name": "Arm Holdings", "market": "US", "role": "端侧 CPU IP", "roleEn": "端侧 CPU IP", "bucket": "core", "href": "/investor/stocks/ARM/" }, { "ticker": "MSFT", "name": "Microsoft", "market": "US", "role": "Windows/Office Agent 入口", "roleEn": "Windows/Office Agent 入口", "bucket": "core", "href": "/investor/stocks/MSFT/" }, { "ticker": "002475.SZ", "name": "立讯精密", "market": "A-share", "role": "消费电子制造", "roleEn": "消费电子制造", "bucket": "core", "href": "/investor/stocks/002475.SZ/" }, { "ticker": "002241.SZ", "name": "歌尔股份", "market": "A-share", "role": "声学/可穿戴", "roleEn": "声学/可穿戴", "bucket": "core", "href": "/investor/stocks/002241.SZ/" }, { "ticker": "2382.HK", "name": "舜宇光学科技", "market": "HK", "role": "手机光学", "roleEn": "手机光学", "bucket": "core", "href": "/investor/stocks/2382.HK/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" } ], "stockTickers": [ "AAPL", "QCOM", "ARM", "MSFT", "002475.SZ", "002241.SZ", "2382.HK", "005930.KS", "600519.SH", "300760.SZ", "600276.SH", "3690.HK", "001215.SZ", "1179.HK" ], "primaryStocks": [ { "ticker": "AAPL", "name": "Apple", "market": "US", "role": "端侧 AI 与 Apple Silicon", "roleEn": "端侧 AI 与 Apple Silicon", "bucket": "core", "href": "/investor/stocks/AAPL/" }, { "ticker": "QCOM", "name": "Qualcomm", "market": "US", "role": "手机/PC NPU", "roleEn": "手机/PC NPU", "bucket": "core", "href": "/investor/stocks/QCOM/" }, { "ticker": "ARM", "name": "Arm Holdings", "market": "US", "role": "端侧 CPU IP", "roleEn": "端侧 CPU IP", "bucket": "core", "href": "/investor/stocks/ARM/" }, { "ticker": "MSFT", "name": "Microsoft", "market": "US", "role": "Windows/Office Agent 入口", "roleEn": "Windows/Office Agent 入口", "bucket": "core", "href": "/investor/stocks/MSFT/" }, { "ticker": "002475.SZ", "name": "立讯精密", "market": "A-share", "role": "消费电子制造", "roleEn": "消费电子制造", "bucket": "core", "href": "/investor/stocks/002475.SZ/" }, { "ticker": "002241.SZ", "name": "歌尔股份", "market": "A-share", "role": "声学/可穿戴", "roleEn": "声学/可穿戴", "bucket": "core", "href": "/investor/stocks/002241.SZ/" } ], "instrumentTags": [ "AAPL", "Apple", "QCOM", "Qualcomm", "ARM", "Arm Holdings", "MSFT", "Microsoft", "002475.SZ", "立讯精密", "002241.SZ", "歌尔股份" ], "reviewCadence": "每日复核", "confirmationSignals": [ "价格弹性是否转化为利润率修复", "支付端和监管是否支持长期现金流", "防御属性是否被估值提前透支" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "估值拥挤", "支付监管", "渠道库存", "CPI 食品项" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 102 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 13, "recentEvents": 102, "supportRatio": 0.06, "riskRatio": 0.86, "conflictScore": 96.6, "evidenceDepth": 100 }, "score": 70.3 }, "events": [ { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "HBM", "先进封装", "GPU", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,800万美元（约降33%）[S1][自测算]；先进封装（CoWo…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.25, "matchedTerms": [ "HBM", "先进封装", "CoWoS", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-560亿美元收入，其25%低毛利率根本无法对冲高利润半导体供应链…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.5, "matchedTerms": [ "HBM", "先进封装", "GPU", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM合约价Q2环比+58–63%)、以及电网电力(美国并网排队约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.25, "matchedTerms": [ "HBM", "先进封装", "GPU", "CoWoS", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "半导体国产替代对AI算力瓶颈的缓解潜力", "summary": "支持但有限定：国产半导体设备与先进封装确实有效缩短AI算力兑现周期的硅片侧环节。设备国产化率2024约26%→2025约35%，刻蚀/薄膜约40%本土，北方华创(002371.SZ)全球第五、中微(688012.SH)前十三，大基金三期3440亿托底；先进封装为弯道超车路径，长电(600584.SH)XDFOI 4nm量产、良率约98.5%，国产CoWoS…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.75, "matchedTerms": [ "HBM", "先进封装", "CoWoS", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗ASIC。先进封装成为泄压阀，但TSMC已提示先进封装产能紧张…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.75, "matchedTerms": [ "HBM", "先进封装", "ASIC", "CoWoS" ], "href": "" }, { "date": "2026-06-28", "title": "半导体产业链的供给侧约束与国产化替代的技术边界", "summary": "本卡对H20/HBM管制下中国国产替代的技术边界做压力测试，部分否定「难以逾越的断层」框架。结论分三层：先进封装（CoWoS替代）是中国相对优势、是可在1–2年跨越的桥（JCET/通富/盛合晶微＋Naura混合键合），并非断层；真正断层在上游——EUV光刻国产化<5%（SMEE约90nm，落后ASML约十年）与高端ArF/EUV光刻胶约95%依赖日本；而专…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.75, "matchedTerms": [ "HBM", "先进封装", "CoWoS", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "summary": "截至2026-06-28，本卡否定“AI资本开支韧性等于A股防御性配置锚”的宽泛结论。Microsoft、Alphabet、Meta、NVIDIA、TSMC等数据仍支持AI基础设施真实需求，但FOMC 3.50%-3.75%、5月PCE同比4.1%、10年美债4.40%和A股两融突破RMB 3tn，意味着长久期成长股、拥挤主题和融资敏感供应商仍可能估值下修…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22.75, "matchedTerms": [ "Microsoft", "HBM", "先进封装", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交期[S6][S7]限制的是通电而非下单。CoWoS售罄、交期…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.25, "matchedTerms": [ "HBM", "先进封装", "GPU", "CoWoS", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "在2026-06-27的权威工作日背景下，国产HBM良率骤降逼迫国内算力网络化，而全球封装与电力阻尼拉长了云厂商CapEx变现周期，这使得制程免疫且订单排至2028年的可插拔光模块出口龙头成…", "summary": "截至2026-06-27，本报告评估了HBM良率危机与先进封装瓶颈对算力集群技术演进的影响。国产HBM3封装良率降至2.66%，倒逼国内算力集群转向网络密集的Scale-out拓扑，推高了光模块需求弹性。全球云厂商2026年CapEx直逼7250亿美元，但受制于CoWoS晶圆先进封装冲突与吉瓦级电网接入瓶颈，面临6-12个月的算力交付与收入确认错配。CPO…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.25, "matchedTerms": [ "HBM", "先进封装", "CoWoS", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "电子链条利润反弹主要受AI基础设施投资拉动，具备向存储、设备、材料、先进封装、光互联和数据中心电源散热扩散的潜力，但尚不足以证明消费电子和内需全面复苏。", "summary": "截至2026-06-28，本卡支持上一卡判断：电子链条利润修复主要由AI基础设施、算力芯片和存储涨价驱动，而非内需广谱复苏。国家统计局数据显示，1—5月电子行业利润同比增长103.9%，贡献全部工业利润增量43.1%。NVIDIA、Microsoft、Alphabet、Meta、Alibaba和Tencent的披露共同验证数据中心硬件需求仍强；TrendF…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "Microsoft", "先进封装", "设备" ], "href": "" }, { "date": "2026-06-25", "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实触发重排，但方向是向支出倾斜，且买方自身点名的硬约束是电力与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "Microsoft", "HBM", "先进封装", "ASIC", "CoWoS", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元在手订单）；第二档择优超配高端光互联（1.6T之年、龙头锁单…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.75, "matchedTerms": [ "HBM", "先进封装", "GPU", "ASIC", "CoWoS" ], "href": "" }, { "date": "2026-06-25", "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出保护的公用事业更偏时点风险资产。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "HBM", "先进封装", "GPU", "ASIC", "设备" ], "href": "" }, { "date": "2026-06-25", "title": "电网缓释后，AI集群上线风险不是消失，而是转移到CoWoS/HBM、液冷验收与网络芯片的串行同步，CPO下行收窄但不等于收入曲线完全恢复。", "summary": "截至2026-06-26，本卡压力测试Research note 7：电网侧缓释可降低CPO最坏收入下修，但不能消除AI集群的串行硬件瓶颈。关键约束从单一接电问题转向先进封装/HBM、液冷机柜验收、交换ASIC/网络集成的节奏匹配。CoWoS缺口可能由约20%降至2026年底约10%，但10%仍足以配给高端加速器；液冷从少数高密度机房跨越到100-120 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.25, "matchedTerms": [ "HBM", "先进封装", "ASIC", "CoWoS", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "云服务商(Hyperscalers)在算力能效提升与封装产能错配背景下的架构转型与部署节奏", "summary": "支持使命命题。截至2026年中数据显示转型正在加速：定制ASIC的AI服务器出货同比+44.6%、商用GPU仅+16.1%，ASIC占AI服务器约27.8%；Broadcom定制加速器收入同比+140%、AI订单储备730亿美元(含定制530亿)。能效是动机：Ironwood推理每瓦性能约B200的2倍，Trainium 3推理成本最高降70%；采购标准翻…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 25.75, "matchedTerms": [ "先进封装", "GPU", "ASIC", "CoWoS" ], "href": "" } ], "timeline": [ { "date": "2026-04-24", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "2026年2.5D/3D先进封装（CoWoS）产能扩张与HBM4演进：评估次世代AI加速器供应链的稳定性", "summary": "2026年2.5D/3D先进封装（CoWoS）产能扩张与HBM4演进：评估次世代AI加速器供应链的稳定性 随着TSMC CoWoS产能持续紧张且HBM4于2026年开始量产，从HBM3e向HBM4的跨越将如何影响主要AI GPU厂商（如NVIDIA/AMD）的成本结构和交付周期？哪些二级OSAT厂…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-04-25", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "2026年超大规模云厂商AI资本开支冲刺与电力—先进封装双瓶颈", "summary": "2026年超大规模云厂商AI资本开支冲刺与电力—先进封装双瓶颈 在Q1 2026业绩季释放的约7,000亿美元四大超大规模云厂商资本开支指引下，约束是否已从GPU供给切换至电力与CoWoS/HBM封装产能？这对TMT板块2026下半年的α定价意味着什么？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-28", "supports": 1, "risks": 0, "neutral": 0, "value": 52.8, "events": [ { "title": "[auto-handoff] 先进封测厂商HBM技术验证", "summary": "致 社交媒体分析师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 417059b2-451c-4792-b6d8-01188cbec876 结果摘要: 我将通过分析中芯国际（SMIC）在HBM3产业链中的角色、TSV工艺的技术指标、量产时间表以及与合作伙伴的验证进展，…", "sourceLabel": "分析协作", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "2027 年 HBM4 推理链条的首要弹性在合格 HBM4，首要出货瓶颈在 HBM4 验证与 CoWoS/interposer/封装基板，而液冷更多是部署瓶颈。", "summary": "本卡支持前序推理需求上行逻辑：288 GB HBM4 GPU 普及后，最大量价弹性在合格 HBM4 供应商，尤其是具备份额提升弹性的 MU 和具备配额权的 000660.KS；但最可能成为 2027 物理出货瓶颈的是 HBM4 验证与 CoWoS/interposer/封装基板的组合。测试设备是高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Q1 2026 财报后的云巨头 AI capex：需求验证还是瓶颈预警？", "summary": "本卡选择 tmt-analyst 评估 Q1 2026 云巨头 AI capex 周期。Microsoft、Alphabet、Amazon 和 Meta 均显示 AI/云收入加速，但 capex 和 PP&E 支出同步大幅上行。TSMC 的 61% HPC 收入占比、74% 先进制程占比和 66.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-03", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "2026年4月 hyperscaler AI capex 重估", "summary": "本卡由 tmt-analyst 作为开题分析师完成。结论是：2026年 hyperscaler AI capex 仍是需求驱动周期，而非简单泡沫破裂；Microsoft、Alphabet、Amazon 和 Meta 的 2026 capex 大致落在 $695-725 billion 区间，且由 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026 一季度超大规模云厂商 capex 再加速与 AI 基建需求验证", "summary": "2026Q1 财报确认 AI 基建二次再加速:四家合计 2026 capex 指引升至约 7250 亿美元(YoY +70–77%),GCP 订单余额一季增加 2200 亿至 4600 亿,Pichai 明示算力受限。Microsoft AI 跑率 370 亿美元,YoY +123%;AWS 重新…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "2026下半年AI资本开支拐点辩论：消化期还是再加速？", "summary": "TMT分析师判定，2026下半年AI资本开支并未进入消化期，而是进入结构轮动期：总量持平至上行，边际增量美元从训练优化的商用GPU集群转向推理fleet、超大规模云厂商自研芯片、以及电力与场地准备。空头叙事在需求侧（推理token volume、agentic渗透、多模态）证据不足，但在前沿训练边…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 2, "neutral": 0, "value": 46.4, "events": [ { "title": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯…", "summary": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯片竞争格局：NVIDIA Rubin对阵云巨头ASIC 本卡主题： WFE资本开支周期——先进封装与HBM设备订单可…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "云端仍是 AI 算力的定价权锚，边缘只获得 2027 年低基数盈利修正机会，因此港美股配置应从单边 AI beta 转向云端瓶颈核心持仓加选择性边缘 spread。", "summary": "本卡综合前序结论：2027 年边缘 AI 和 AI PC 的确有分母红利，但全球 AI 算力定价权仍集中在云端基础设施、HBM、先进封装、网络和模型编排。组合上不应简单“卖云买边缘”，而应采用 barbell plus spread：45-55% 保留在 NVDA/AVGO/TSM 与 MSFT/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI经济在架构层已多元化但在制造层重新集中于台湾3nm，这一咽喉点产生的地缘政治溢价已被TSM和日元部分定价、但在美国云巨头股权与利率波动率中显著低估。", "summary": "全球约99%的前沿AI芯片（≤3nm）晶圆与几乎100%的CoWoS-L先进封装仍集中在台湾新竹—台南走廊，亚利桑那、熊本、德累斯顿的分散化要到2028下半年才可能实质生效。当前TSM ADR与新台币已隐含约12-15%的台海风险溢价，但美股云巨头（MSFT/GOOGL/META/AMZN）几乎未…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "AI CapEx的主线不是收缩，而是从新增外壳数量转向每瓦、每机架、每封装的半导体密度提升。", "summary": "截至2026-05-13，本卡支持卡片01的“密度胜过足迹”判断：数据中心外壳新开工-19% QoQ并不等于AI CapEx下行，而是资本更多流向GPU/ASIC、HBM、CoWoS/先进封装、机架级互连、光DSP和液冷配套。Microsoft、Meta、Alphabet、Amazon的最新披露均…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 2, "neutral": 0, "value": 41.2, "events": [ { "title": "算力期货压缩的是商用GPU层租金，先进封装与高带宽互联的物理瓶颈与寡头结构足以将其非商品化毛利结构维持至2027。", "summary": "支持card-01框架并加以收紧：算力期货标准化的是GPU-小时这一输出单位，并未商品化决定谁能以最低成本清算的物理输入。CoWoS先进封装、HBM4、混合键合设备、scale-up交换机ASIC、PCIe/CXL retimer、1.6T/CPO光模块均具18–36个月结构性导入周期，寡头格局清…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI单名信用利差走宽主要是供给+杠杆+利率的周期性现象，半导体设备订单簿与超大规模资本开支指引仍坚挺，尚不构成成长逻辑的结构性破坏。", "summary": "压力测试卡片03的预警：截至2026-05-15，AI单名利差走宽约75%可归因于周期性/技术性因素（新发让利40%、资本开支融资杠杆漂移25%、利率beta 10%），仅约25%是早期基本面（ROIC/GPU折旧诚信）。Big-5 2026年AI资本开支合计约2900亿美元同比+45%，无一家削…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 4, "neutral": 0, "value": 30.6, "events": [ { "title": "A股半导体重估的 beta 已基本被定价，下一段必须靠后道设备、量检、材料、车规功率与先进封装上游材料的盈利与订单兑现，回避换手/融资/解禁三重拥挤的设计、代工与映射封装。", "summary": "截至 2026-05-15，A股半导体经过 4-5 月重估后，国产 GPU/ASIC 概念、HBM/先进封装映射股、大市值晶圆代工四条线已拥挤：5 日换手率分位 81-92%，半导体两融占自由流通市值比 ~4%（88 分位），主要芯片 ETF 4 周连续净申购 +18%，且 6-8 月解禁集中（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GXP向CSA的制度松绑只是把交付瓶颈下移到芯片配额与高压电力装备交期，物理瓶颈已替代制度瓶颈成为AIDC交付的新主导变量。", "summary": "CSA转型确实把特区内AIDC交付压缩到9–12个月，但约束随即从制度排队迁移到物理输入：>500 MVA变压器交期110–140周、CoWoS-L先进封装2026年75–80%被NVDA/AMD/Google预订、HBM3E为价格弹性最低的卡点。'交付悖论'未消失，只是从监管时间转为BOM时间。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "M2M流量爆发的边际成本不会均匀挤压SaaS——它会让薄壳代理编排厂商利润坍缩、让有自有数据的垂直SaaS继续吃到按量计费红利，并把被争夺的利润集中流向芯片、封装与光模块基础设施层。", "summary": "截至2026-05-16，本卡对card-07的SaaS基本面Beta判断做压力测试。结论是有条件成立但显著分化：薄壳代理编排SaaS将因无法转嫁的token成本，在未来12–18个月承受600–1,200 bp毛利压缩；而垂直SaaS、数据合规和API适配器（card-06超配名单）边际成本以存…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 硬件链应买真实瓶颈而非泛映射：台韩核心持有但不追，A股/HK 侧优先封装、测试、设备和材料，回避已把完美 5-6 月收入斜率资本化的纯 AI 算力概念和成熟制程代工。", "summary": "截至 2026-05-15，我综合支持前两张卡：AI 需求未被证伪，台积电 Q2 低端至中枢可兑现，但高端受 CoWoS/HBM 与客户平台节奏约束。台韩主链基本面最清晰，却已大体定价顺利执行；A股/HK 侧分化更大，国产 AI 算力和大涨后的中资晶圆代工已较充分定价。风险收益仍错配在先进封装、测…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "国产先进封装材料能在2026年瓶颈窗口前撑住2.5D等效吞吐，但TSV化学品缺口将把HBM3E等效堆栈推至2027年中——多材料/CMP基本盘，HBM封测是2027而非2026的交易。", "summary": "【2026-05-17】支持卡片04超配上游材料/设备的方向判断，但加上更精确边界：到2026年Q4，国产化将在ABF载板、TCB键合机、EMC/底填三大卡点上实现可用级，但TSV级Cu/W CMP抛光液与高纯度（>6N）电镀化学品仍是阻止HBM3E等效堆叠的硬约束。2.5D CoWoS等效封装吞…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "最终方案是三腿再平衡而非单轴切换：设备+11、ASIC+9、边缘+6、纯GPU−10、IPP−13，并叠加SMH/NVDA与配对对冲及CEG/VST尾翼call。", "summary": "[2026-05-16] 综合前四张白板，AI电力约束由发电→设备→算力形态逐级下移，最终方案为三腿结构性再平衡：XLU/IPP减仓约30%（保留三分之一锚仓与CEG/VST尾翼call）；电力设备超配+11单位（GEV/ETN/VRT；A股特变、国电南瑞、思源、平高、英维克）；纯GPU去集中化−…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "AMD 上涨与 Agentic AI 标志着推理层的真实碎片化，但训练层仍是 NVDA 主导的寡头格局；这是 AI 投资逻辑的分层化，而非永久性转向。", "summary": "AMD 暴涨与 Agentic AI 叙事是部分真实的结构性变化，但被卖方过度简化为'永久性转向'。Agentic 工作负载将单任务 token 量放大 30–100 倍，使约束转向 HBM 带宽与单 token 经济性，ROCm 7 + vLLM 生态使 AMD MI355X 在推理侧达到与 H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 46.8, "events": [ { "title": "截至2026-05-19，AI半导体供应链尚未出现云CapEx或HBM/CoWoS订单环比下修，算力基础设施交易仍应保留核心仓位但需严密跟踪ROIC与Q2/Q3指引风险。", "summary": "截至2026-05-19，首席策略师提出的敏感触发点有效，但尚未触发。Microsoft、Alphabet、Amazon、Meta的2026年AI基础设施CapEx仍在上修或维持高位，TrendForce估计前八大CSP 2026年CapEx超过7100亿美元、同比约+61%。HBM方面，Micr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027-2028年电力缺口不会终结AI芯片周期，但会把下一代AI硬件迭代逻辑从单芯片峰值性能改写为受限MW下的有效AI输出、上线时间和机架级能效。", "summary": "截至2026-05-18，本卡支持前序电力瓶颈主线，但压力测试“硬件从极致性能转向能效”的表述：更准确地说，采购函数会转向每已通电MW的有效AI输出。2027-2028年表后燃气仅4-6GW，无法覆盖AI负荷缺口；DOE/LBNL预计美国数据中心用电2028年达325-580TWh。NVIDIA …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 3, "neutral": 0, "value": 50.2, "events": [ { "title": "截至 2026-05-19，AI 基础设施仍是 TMT 领涨主线，但可投资范围应收窄到有真实 usage、backlog、产能瓶颈和现金流承受力的环节。", "summary": "工作日期 2026-05-19。本文否定前序空白 TMT 论点并选择 (A)：AI 基础设施仍是领涨主线。证据是 hyperscaler capex/收入比虽已升至约 30% 的四家公司 Q1 合计强度，且 Alphabet、Amazon 的 FCF 张力明显，但 Microsoft AI ARR…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "若 2026-06-11 验证 Waymo 单位经济，AI 基础设施不应被卖出，而应从拥挤算力前端轮动到电源液冷、networking 邻近环节与小仓位机器人硬件 beta。", "summary": "截至2026-05-19，本卡支持 6 月 11 日后从 AV 扩散到 AI hardtech，但只在 Waymo 经济性披露有用时执行。确定性最高的是电源/液冷/机柜、定制芯片与 networking、HBM/先进封装；估值弹性最好的是电源热管理与机器人端侧硬件。HBM、GPU、A 股光模块和主…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "云端算力对HBM及CoWoS产能的虹吸导致边缘端LPDDR5X面临严峻的“存储通胀”，硬件降本逻辑阶段性证伪，高昂的BOM成本将大幅压制端侧AI在2026年的大规模渗透率。", "summary": "截至2026-05-19，我们对端侧AI的存储成本约束进行了深度压力测试。测试表明，该约束比预期更严峻。云端HBM4与先进封装（CoWoS）的大规模扩产对晶圆产能形成“虹吸效应”，导致LPDDR5X/6出现结构性短缺，2026年边缘设备存储BOM成本激增32%-50%。这迫使AI PC与手机涨价5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力链供应瓶颈已基本解决，支持在 CNH 企稳后回补权重至 50%，利润池正向 HBM4 与先进封装技术前沿发生结构性转移。", "summary": "2026-05-20 卡片支持将 AI 算力链权重回升至 50%。研究显示，HBM3e 良率提升与台积电 CoWoS 产能（年底达 7 万片/月）的实体扩张已基本消除 2024-25 年的瓶颈。虽然日元 carry 解除对日本设备标的造成估值冲击，但其基本面供应受损风险低。HBM4 转型将导致利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 4, "neutral": 0, "value": 37.6, "events": [ { "title": "AI与半导体能在防御性轮动中维持Alpha，但安全边际只存在于盈利可见度高、供给受限且能把CapEx转化为现金流的瓶颈环节。", "summary": "2026-05-20，本卡支持第4张卡的杠铃配置：AI与半导体可保留为15-20%的Alpha袖珍仓位，但不能作为泛防御Beta。SIA/WSTS显示2026年一季度全球芯片销售2985亿美元，环比+25%；TSMC 1-4月收入同比+29.9%，并指引2026年美元收入增长30%以上；NVIDI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链盈利可见性已被订单簿三重验证、足以支撑保留成长端的策略；半导体设备应拆分成‘AI直接受益+国产替代超配，纯中国敞口设备与晶圆厂低配’，故第03卡‘不全面切向价值’成立但需要…", "summary": "截至2026-05-21，本卡对第03卡“不全面切向价值”进行压力测试，结论是有条件支持。AI算力链（GPU/HBM/光模块/液冷/先进封装）盈利可见性高且仍在抬升：四大美系hyperscaler 2026年合计资本开支指引上修至USD 360–410 bn、TSMC CoWoS与SK Hynix…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "工作空间物理修复的‘成功’是虚假的，更重要的是研究逻辑必须从‘芯片获取’转向‘能耗-算力-政策’的宏观三难困境。", "summary": "我们正式否认工作空间重构成功的论点，因为物理文件在2026年5月20日的验证中依然缺失。宏观层面，AI资本开支已从技术叙事转向‘算力基础设施化’政策驱动。当前核心瓶颈已从芯片获取转向电网容量与国产先进封装（CoWoS/HBM）良率。A股风格正在从模型端向电力基建与国产设备切换。建议后续深入研究国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产AI效率、软件栈和推理优先部署可以抵消应用层硬件缺口的大部分边际压力，但无法消除前沿训练和生态成熟度折价；投资应聚焦高利用率推理软件与能按结果变现的应用主题。", "summary": "2026-05-20：本卡综合判断，国产AI效率路线能显著缓冲应用层硬件约束，但不能完全替代前沿训练硬件、HBM、先进封装和CUDA式生态成熟度。DeepSeek/Qwen的MoE、thinking budget、KV-cache复用、量化和推理服务框架，把关键问题从“是否有足够GPU训练最大模型…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "在英伟达 Rubin 平台面临供应链瓶颈的窗口期，国产 GPU 凭借先进封装与 HBM3 的全产业链突破，正通过‘十五五’初期智算中心的大规模采购实现对推理市场的实质性支配。", "summary": "截至2026-05-22，本研究确认国产GPU已进入性能与市场双重拐点。NVIDIA Rubin平台受困于HBM4验证延迟与CoWoS-L稳定性，2026年产量目标下调25%。与此同时，海光深算4号与寒武纪思元690通过Chiplet技术实现性能跃升，结合长电、通富在2.5D封装的突破以及长鑫存储…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 52.4, "events": [ { "title": "'效率降级'对 2026–2027 AI Capex 总量近乎无约束，但通过电力、关税、航运三条渠道侵蚀单位经济学，使全球 AI 基础设施从'同步扩张'转为'地理极化'，估值真正承压…", "summary": "本卡对 Research note 1–5 的“效率降级削弱 AI 算力扩张”假设进行双向压力测试。结论：方向正确但程度被高估，AI 算力 Capex 在 2026–2027 名义增速仍 +19–26% YoY（北美 hyperscaler 2026 指引 4,400–4,650 亿美元 [S1–…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装已成为美中半导体博弈的“新前线”，在大基金三期驱动下，国产替代链条正在经历从“晶圆代工单一突破”向“虚拟集成封测增效”的战略重心转移。", "summary": "2026年5月，美方出口管制从“全面禁运”转向“管理接入”模式，通过25%营收分成税与强制在美测试对先进AI芯片（如H200）实施出口许可。此举对A股半导体链形成双重压力：一方面，国内先进制程（如中芯7nm）虽有突破但良率仍处于20%左右的“盈亏平衡点”以下，面临高成本挑战；另一方面，HBM及先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 2, "neutral": 0, "value": 53.8, "events": [ { "title": "半导体板块在高利率与中美并购友好政策共振下，正在以更快节奏复现生科板块的哑铃式价值重估——整合者赢家通吃、后期去风险标的被并购溢价重新锚定、中间产能被同步出清。", "summary": "截至2026-05-24，本卡片自下而上验证半导体板块正在高利率（UST10Y 4.56%）背景下复现生科哑铃逻辑。证据集中在三层：(1) 并购重组——Synopsys-Ansys ($35B)、AMD-ZT Systems ($4.9B)、HPE-Juniper ($14B)、Silver La…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施资本开支尚未在订单层面见顶，但在扩产加速度和重资产风险收益上已接近顶部，2027年放缓将优先打击低毛利、高折旧、专用性强的制造环节。", "summary": "截至2026-05-24，本卡从工业制造与Capex Cycle角度应力测试AI基础设施扩产。结论是：已披露需求和行业基准预测尚未显示周期见顶，SEMI仍预计2026-2027半导体设备支出上行；但资本开支加速度、低毛利跟随者扩产和数据中心建设成本已呈现后周期特征。若2027年需求放缓，最严峻的减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 5, "neutral": 2, "value": 43.2, "events": [ { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "现有先进半导体出口管制不会停止AI使用或全部ASIC设计，但会系统性削弱受限生态通过HBM密集、CoWoS级封装实现“每瓦有效推理”跃迁的能力。", "summary": "截至2026-05-25，本卡压力测试前序“ASIC与先进封装推动每瓦有效推理”的判断。结论是：路径对可获得TSMC、HBM与CoWoS产能的全球云厂商仍成立，但对中国相关或D:5受限生态被结构性削弱。美国出口管制已从成品AI芯片延伸至HBM、封装路径、先进制造设备、软件/技术和FDP规则，卡住了…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HBM 堆叠层数跃升与先进封装产能扩张为 A 股上游高壁垒材料与核心设备注入高增长确定性，应超配相关标的。", "summary": "结合 HBM 产能瓶颈与 CoWoS 先进封装扩张趋势，评估了 A 股半导体材料与设备环节的订单增量与技术壁垒。我们对‘硬件稀缺’主线持强烈支持立场。GMC、前驱体、Low-α 填料等高壁垒材料因 HBM 堆叠层数跃升（+45.8%）具备显著的‘层数乘数效应’；TSV 湿法、电镀及晶圆精密减薄等设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "自研ASIC与先进封装可以把AI推理竞争从“买卡扩容”转向“每瓦有效推理”，但只有软件路由约束System 2调用时，能效提升才会真正缓解电力与散热瓶颈。", "summary": "截至2026-05-25，本卡支持“System 2推理将AI推向能源与散热约束”的前序判断，但认为ASIC与先进封装能显著推迟能源墙。自研推理ASIC、FP8/FP4低精度、HBM/CoWoS/SoIC、背面供电和直接液冷可以降低每token能耗，并提高每兆瓦可承载的有效推理吞吐。关键限制是杰文…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "短期硬约束在 HBM 与 CoWoS-L 而非电力，AI 算力'紧但能出货'，对指数 FOMO 有利但 2026H2 脆弱性显著上升。", "summary": "短期（2026 年内）芯片侧的约束比电力更紧：依次瓶颈为 HBM3e bit 供给 → 台积电 CoWoS-L 先进封装 → GPU 晶圆 → 基板/光模块 → 电力。HBM 与 CoWoS-L 全年额度已锁、扩产跑不过 ~150%/~70% 的需求增长，2026H2 任一良率事件或出口管制重置可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 3, "neutral": 2, "value": 58, "events": [ { "title": "AI硬件利润正由‘芯片级’向‘系统级解决方案’传导，订单积压信号抵消了折旧风险，当前增长受电力而非算力限制。", "summary": "2026年5月研究显示，AI利润中心正从单一GPU向‘系统级’（以太网+ASIC+先进封装）转移。四大巨头2026年CAPEX预计达8000亿美元。以太网（Spectrum-X/Arista）已超越InfiniBand成为主流，ASIC在推理端凭借65%TCO优势爆发。台积电CoWoS产能达12.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力物理墙的宏观结果是资本开支被拉长而非崩盘，TFP 红利右移 18-24 个月，配置上有利于电力/封装/国产替代而非纯算力多头。", "summary": "截至 2026-05-27，前序卡片揭示的 CoWoS-L 翘曲与 HBM4 多供应商 KGD 危机不会引爆 CSP 资本开支崩盘，而是把周期向后拉长至 2028-2029：Big Four 2026 资本开支 USD ~470bn 大概率维持，但 GPU 实际出货低于指引 12-18%，BOM …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "HBM3E面临产能过剩风险，而HBM4的逻辑底晶粒与混合键合成为新瓶颈；建议在14%的先进封装配置内，低配纯存储厂商，超配代工及前端设备商。", "summary": "本报告基于2026年5月的最新产能数据，对策略师的先进制程超配逻辑进行了压力测试。我们发现，虽然台积电CoWoS产能到2026年底将激增至13万片/月，有效缓解了绝对瓶颈，但HBM市场正面临分化：三星激进扩产（25万片/月）导致HBM3E存在价格战风险，而HBM4因引入逻辑底晶粒与16层混合键合，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年A股先进封装板块呈现显著估值分化：半导体设备具有高业绩弹性但估值普遍偏高，而高壁垒封装材料（低α硅微粉及CMP平台材料）与性价比测试机龙头在业绩确定性与估值匹配度上显著占优…", "summary": "本卡片系统评估了2026年A股先进封装设备与材料的国产替代业绩弹性与估值。台积电CoWoS产能外溢及HBM4混合键合变革，拉动封装测试及高壁垒耗材爆发。设备板块呈“高弹性、高估值”分化，长川科技（44x PE）成长匹配度佳，拓荆科技（70-71x PE）估值偏高；而材料板块（联瑞新材25-30x …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "2026年AI芯片与先进封装产能约束会推高并拉长云端CapEx，而不是使其取消；配置应偏向具备配额权的上游瓶颈环节。", "summary": "截至2026-05-28，本卡支持前序判断：AI硬件CapEx传导链的约束真实但不等于削减。Microsoft、Amazon、Alphabet、Meta的2026年CapEx指引中枢约为$710 billion，反映的是芯片、HBM、先进封装、电力和数据中心资源稀缺导致的锁单与涨价，而非需求走弱。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 1, "value": 69.8, "events": [ { "title": "AI硬件周期仍由需求驱动，但未来12个月的定价权将更多落在HBM、先进封装、光模块和电力交付等瓶颈环节，而不是单纯GPU或服务器组装收入。", "summary": "截至2026-05-30，本卡支持“AI算力瓶颈从硅端扩展至物理基础设施”的主线。未来12个月，AI芯片和服务器需求仍由超大云厂商资本开支、云AI收入、agentic AI推理流量和自研ASIC扩张驱动；但供应瓶颈已从GPU单点扩展到HBM、CoWoS/先进封装、800G/1.6T光模块、机架供电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基建复利仍是 2026–2030 最稳定主线，但价值链重心正从'GPU+HBM+CoWoS'扩展到'GPU+HBM+CoWoS+LGIA+输电+PPA'；在保留 AI 半导体核…", "summary": "本卡把前四卡合成可执行的跨行业配置框架。在加入电力权利筛选器（并网/输电/灵活负荷/PPA 四维 0–12 分制）后，给出 12 个桶的旧/新权重对照：AI 半导体核心 40→30，AI 服务器合规优胜者（DELL/HPE）10→12，SMCI 5→1，Hyperscaler 20→22，CRWV…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 3, "neutral": 0, "value": 56.2, "events": [ { "title": "Chiplet能为国产AI算力芯片争取时间和良率弹性，但只能部分补偿先进制程受限，真正受益者更偏先进封装赋能者和系统级NPU平台，而非泛AI芯片beta。", "summary": "截至2026-05-31，本卡对card-07进行TMT产业压力测试：Chiplet和先进封装能在国内AI NPU、推理卡、异构加速器和集群级设备中提供部分性能与良率补偿，但不能完整替代先进制程密度、HBM供应和CoWoS级封装执行。资金端的5000亿元科技再贷款与3440亿元大基金三期能维持ca…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国本土先进封装放量确定性降低，会选择性强化全球半导体设备与材料供应商的稀缺性溢价，并通过延长瓶颈周期和抬高重置成本支撑AI硬件利润重估。", "summary": "截至2026-05-30，本卡支持“出口管制强化全球设备与材料稀缺性溢价”的判断，但强调这是选择性受益而非全行业普涨。中国本土HBM/先进封装放量确定性下降，会降低低价边际供给冲击全球市场的概率；真正受益的是进入合规非中国HBM、先进逻辑、类CoWoS封装、测试、基板和高纯材料认证体系的供应商。对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "非美系低轨星座的自主化上限不是芯片设计口号，而是能否获得CoWoS/HBM/材料产能并完成空间级资质认证。", "summary": "截至2026-05-30，本卡片支持前序判断：非美系低轨星座的AI硬件瓶颈已从单一GPU采购，转向CoWoS/HBM/基板/材料/测试良率组成的封装级工业体系。监管规避可以降低对美国加速器的直接依赖，但无法自动获得先进封装、HBM和材料产能；Eutelsat替换周期与Qianfan部署压力显示，星…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 1, "risks": 0, "neutral": 1, "value": 86.6, "events": [ { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美元指数回落至94.50与利率下行会显著支撑AI硬件估值扩张，但对资本开支的影响是强化既有需求周期，而不是单独创造新周期。", "summary": "截至2026-05-31，本卡支持“美元与利率下行有利于AI硬件及半导体估值扩张”的主线，但限定传导强度：最强作用来自折现率下行、弱美元折算、流动性和订单信心，而非单纯融资成本下降。Microsoft、Amazon、Alphabet、Meta的最新季度AI相关资本开支运行速率已超过1306亿美元，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "滞胀成本压力会侵蚀但不会基准性摧毁Regime A SaaS利润率，真正的分水岭是厂商能否把AI成本从固定席位打包迁移到明确的用量/结果计价。", "summary": "截至2026-06-02，本卡压力测试能源与大宗上涨对AI硬件供应链和Regime A SaaS利润率的传导。结论是有条件支持前序Regime A逻辑：晶圆厂电力成本上涨会压低硬件毛利，但单独不足以击穿SaaS Agent利润驱动；真正高Beta风险来自HBM、先进封装、数据中心电力与云厂商GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 3, "neutral": 0, "value": 60.4, "events": [ { "title": "美股AI硬件强势对科创50半导体链存在明确补涨逻辑，但仅在先进封装和28nm/14nm主流设备两条主线上'清晰可定价'，应以'订单确认+流动性双重确认'同时为纪律抓手。", "summary": "针对2026-06-03美股AI硬件强势能否传导至科创50半导体链的核心问题，TMT分析师支持卡01战术框架，并把补涨逻辑结构化为可定价的两条主线：(1)先进封装（长电600584/通富002156/华天002185/兴森002436），承接台积电CoWoS 25-30%的产能溢出；(2)28nm…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Broadcom (AVGO) 财报验证了AI订单总需求未断裂，但低毛利自研芯片的占比扩张和物理并网/封装刚性约束，正在系统性拉低硬件供应商的边际利润率并平滑其收入转化率，驱使AI算…", "summary": "截至2026年6月3日，Broadcom (AVGO) 发布Q2财报，其AI半导体营收111亿美元超预期，且上调全年指引至145亿美元，证伪了需求崖断裂假说；但其毛利率受低毛利定制加速器 (ASIC) 占比飙升影响，环比大跌290个基点至75.1%低于预期，且受限于CoWoS先进封装与北美电网并网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS产能硬约束会把AI硬件估值溢价从单纯设计叙事重新分配给掌握晶圆、封装、良率和整柜交付能力的产能控制者。", "summary": "2026-06-03：本卡支持前序判断，认为CoWoS与先进封装已成为AI算力链估值的硬过滤器。NVIDIA需求仍强，TSMC的3 nm、HPC与先进封装紧张共同支撑产能控制者溢价；但HBM4供应商扩容削弱单纯存储稀缺溢价。A股半导体整体估值已高，策略重点应从宽泛芯片设计beta转向设备、材料、封…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 0, "value": 67.8, "events": [ { "title": "国产AI互连与ASIC产业链已具备支撑‘主权算力’的功能性自主，但在顶尖模型训练的‘峰值效率’上仍面临由制造工艺和高性能存储构成的结构性压力。", "summary": "2026年6月，美方通过‘总部原则’堵塞海外子公司漏洞，并对H200级芯片实施‘25%规费’个案审查。国产AI ASIC/IP（芯原股份）与互连芯片（澜起科技）表现出极高韧性，通过Chiplet架构、3D封装（韬定律）及国产EDA适配，已实现在推理市场的 functional replacemen…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力需求足以支撑能源仓位从原油转向电力与电网资产，但最佳表达是受并网和输配电约束的结构性电力瓶颈，而不是泛化商品beta。", "summary": "截至2026-06-05，TMT视角支持将能源仓位从原油 beta 转向电力、电网与气转电基础设施。AI 数据中心用电已从芯片约束扩展为芯片加电力约束：IEA 预计全球数据中心用电从2024年415 TWh升至2030年945 TWh，LBNL预计美国数据中心到2028年可达325-580 TWh…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 1, "risks": 0, "neutral": 0, "value": 89.2, "events": [ { "title": "头部CSP的AI投资回收期在HBM4紧缺下并未恶化，反而因稀缺产能高利用率和架构效率提升而缩短，支撑AI硬件的选择性估值溢价。", "summary": "截至2026-06-07，本卡支持并综合前序结论：HBM4紧缺不会打断头部CSP的AI资本开支周期，而是强化“配给能力+效率改善+现金回收”的分化。Microsoft、Alphabet、Amazon披露的需求、积压订单和AI收入信号表明，头部云厂商的有效回收期可能压缩至约1.6-2.3年；Meta…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 76.6, "events": [ { "title": "Research note 03 — AI 硬件供应链的产能瓶颈与成本传导 Date (Asia/Singapore work date): 2026 06 08 Analyst: …", "summary": "Research note 03 — AI 硬件供应链的产能瓶颈与成本传导 Date (Asia/Singapore work date): 2026 06 08 Analyst: TMT Analyst (Semiconductors, AI, cloud, consumer electroni…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 2, "neutral": 1, "value": 83, "events": [ { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI半导体利润池迁移真实但渐进，板块总EPS上行、估值锚从'NVDA-TSMC双寡头溢价'切向'增长+多元化第二供应源'，ASIC设计商、HBM、先进封装设备与Intel Found…", "summary": "2026-06-09综合：Google-Intel TPU订单确认AI硬件利润池迁移真实且渐进，2027-28年AI半导体合计EPS较冲击前共识更高，但分布坚定向ASIC设计商(AVGO/MRVL)、HBM、先进封装设备及Intel Foundry期权倾斜。新估值锚为'增长+多元化第二供应源'，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球科技配置应超配ASIC与新代工生态；A股风格则需从“AI题材映射”战略性切换至独立算力的“国产第二供应源”，重仓先进封装产能、核心设备及HBM材料。", "summary": "2026-06-09评估：基于AI半导体估值锚向“增长+第二供应源”切换，全球配置需减配纯GPU倍数扩张与单一代工依赖，转而超配ASIC双雄(AVGO/MRVL)、HBM及Intel代工期权。在A股与港股层面，风格将从“英伟达映射”(光模块/代工)实质性轮动至业绩兑现度高的国产“第二供应源”。策略…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "Google-Intel的TPU订单标志着AI硬件利润池的结构性迁移：从绝对依赖GPU/TSMC，转向ASIC出货量反超以及Intel Foundry作为第二供应源的崛起，但GPU仍…", "summary": "2026-06-09评估：Google向Intel Foundry下单2028年超300万颗TPU，且Nvidia正评估Intel 18A工艺，这表明AI总需求依然强劲（到2028年TAM达3800亿美元），但硬件利润池正发生结构性迁移。为降低TCO，云厂商加速自研ASIC，出货量预计将超越GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 2, "neutral": 0, "value": 73.4, "events": [ { "title": "2026-06-10的主线不是在云、终端和政策之间三选一，而是超配云需求确认，选择性持有HBM/国产替代，暂把Apple换机作为二阶段期权，并低配政策与现金流暴露高的AI资产。", "summary": "截至2026-06-10，本卡综合Oracle业绩窗口、Apple AI终端升级和美国先进芯片海外子公司管制澄清。结论是：A股AI链条应优先交易云需求确认，其次保留Apple端侧AI换机期权，同时用出口管制和自由现金流压力控制仓位。Oracle、Microsoft、Alphabet、Meta仍显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超配拥有确定性全球订单的光电液冷龙头与昇腾-中芯先进封装/HBM自主垂直栈，回避被截断离岸通道的二线芯片设计与空壳算力租赁标的。", "summary": "截至2026-06-10，本研究深入A股TMT板块。建议超配：(1)挂钩中美头部云、订单已确认的光模块（中际旭创、新易盛、天孚通信）、液冷（英维克）与电源龙头（麦格米特），均处于1.6T/Blackwell订单放量期；(2)华为昇腾与中芯国际（SMIC）挂钩的HBM/先进封装国产替代垂直栈，长电科…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 74.8, "events": [ { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "铜、铝、电缆和电气设备稀缺已经进入AI算力集群交付链，最先约束的是通电日期和可售云容量，而不是GPU本身。", "summary": "截至2026-06-11，材料链稀缺已开始转化为AI基础设施交付摩擦，但表现为通电时间、电气设备交期和云算力容量分配，而非所有GPU集群停摆。IEA提示约20%规划项目有延迟风险，CBRE称电力约束把工期推至2027年及以后，Wood Mackenzie给出关键电气部件18-36个月交期和600 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 0, "value": 64.2, "events": [ { "title": "为支撑三层算力织物，AI芯片与先进封装如何演进以匹配电力波动与边缘推理的功率包络", "summary": "综合并支持卡片04的三层算力织物，下沉到硅基与封装层。关键洞见：'电力波动友好'主要是电力电子＋调度固件问题，而非新晶圆问题。NVIDIA GB300靠电源货架（65 J/GPU储能＋功率封顶斜率＋功率烧除）把Megatron训练电网峰值砍约30% [S1]，DVFS早是标配 [S2]；同步功率摆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力刚性约束已把AI加速器竞争维度切换为每瓦Token，Chiplet+先进封装是2027-2028电网窗口内唯一可规模化兑现的算力密度救援方案，对NPV的杠杆是5-8倍。", "summary": "电网刚性约束把AI加速器的KPI从峰值FLOPS切换为每瓦Token吞吐。Blackwell B200单卡1000W、GB200 NVL72机柜120kW已触顶，下一代Rubin/MI400无法继续抬TDP，必须靠能效换性能。TPU v5p、Trainium2较H100能效提升30-40%；同10…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的扩展性：电力与芯片供应瓶颈对超大规模厂商CAPEX的影响分析", "summary": "截至2026-06-12，本卡压力测试卡片02（电力瓶颈）与卡片06（半导体多线挤压）。结论：双重约束不会削减超大规模厂商资本开支，而是将其重排。2026资本开支仍约$725-726B（Dell'Oro），资本开支/营收比约25-30%、指引2027约35%；变化在优先级——可靠电力升至栈顶，建设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-13，电力设备成本与交付风险已经实质性延后 AI 算力集群通电节奏，但它表现为 capex 效率恶化和项目重排，而不是最大型 hyperscalers 已经全面削…", "summary": "截至2026-06-13，本卡支持前序“电力约束正在惩罚 AI capex/融资假设”的方向，但精确化为：2026-2028 年已出现项目层面的通电延期与节奏重排，尚未出现头部 hyperscalers headline capex 的广泛下修。Sightline 显示 2026 年约 16 GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 4, "neutral": 2, "value": 71.6, "events": [ { "title": "CoWoS绝对产能正在改善，但GPU与自研ASIC需求吸收速度更快，因此2026年AI集群交付仍受先进封装、HBM和关键设备交期的结构性约束。", "summary": "截至2026-06-13，本卡支持上一张AI基建卡的判断：硬件交付风险不只来自并网，还来自CoWoS/HBM/设备交期的半导体转换链。TSMC仍称先进封装非常紧，TrendForce口径显示CoWoS-L/S满载且2026年底目标120,000-130,000片/月；DIGITIMES虽更乐观，预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红皇后竞速并未彻底抹去自研ASIC的成本优势，而是将其集中至Google与AWS两家、使其他云厂商回到对NVIDIA的依赖，从而部分推翻了Research note 05所述的全行业…", "summary": "在'红皇后竞速'下，自研ASIC单世代有效寿命已压缩至12–15个月，2nm节点全口径NRE攀升至USD 0.9–1.4B、N2晶圆约USD 30,000/片、CoWoS-L售罄至2027。我们测算自研ASIC在TCO上跑赢商用GPU的年部署门槛从5nm节点的约20万颗升至现实情景下的15万–25…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "CoWoS 与 HBM 扩张是 NVIDIA 继续销售高溢价系统的供给条件，而不是摧毁溢价的触发器。", "summary": "2026-06-13：本卡支持前序判断。CoWoS/HBM 扩产不是削弱 NVIDIA 护城河，而是让其把更多机柜级 AI factory 供给转化为计算、内存、网络与软件平台溢价。TSMC 的先进封装、3nm/HBM base die 产能，SK hynix/Micron/Samsung 的 H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "硬件与能源通胀只能部分向软件端转嫁，头部云厂商2026年不至于利润率崩塌，但AI云业务的边际回报率和自由现金流将被结构性压缩。", "summary": "截至2026-06-14，本卡支持前序“资本强度陷阱”链条：云厂商可在稀缺GPU、预留容量、企业工作流和agent定价中转嫁部分成本，但公开API价格已被OpenAI、Anthropic、Google等竞争锚定，难以承受全面涨价。AWS、Azure、Google Cloud短期报表利润率未崩塌，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 5, "neutral": 0, "value": 61, "events": [ { "title": "GPU抵押贷款收缩会削弱边际Neocloud订单可见度，但先进封装与先进制程需求更可能从融资套利订单转向信用更硬的超大规模厂商和ASIC路线，而非进入全面下行。", "summary": "截至2026-06-15，本卡压力测试GPU抵押融资收缩对半导体订单的影响：冲击不是AI芯片需求消失，而是订单可见度从Neocloud融资驱动采购转向超大规模厂商、ASIC、主权项目和带预付款的硬合同。NVIDIA Data Center在FY2027 Q1仍达752亿美元，TSMC先进节点和HP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI先进封装与HBM产能瓶颈对算力基础设施扩张及超大规模厂商资本开支的影响", "summary": "支持半导体分析师Research note 01的论点并向下游延伸：CoWoS/HBM确为算力的关键约束，2026年部署受封装与内存所限（CoWoS扩至约13万片/月仍订满；SK Hynix与Micron 2026年HBM售罄；HBM4验证延迟使NVIDIA将Rubin目标由200万下调至150万…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在AI链内部按融资质量轮动：超配Tier-1平台与卖铲人(设备/HBM/封装/电力赢家)，低配靠GPU抵押融资循环的Neocloud股权与高收益债。", "summary": "综合前八卡推演，AI基础设施的约束已从硅到电力、再到融资可得性。本卡给出按融资质量过滤的五篮子相对配置框架：B1 Tier-1 AI平台(+300bp)、B2 设备/HBM/先进封装(+250bp)、B3 ASIC定制(+100bp)、B4 电力瓶颈赢家(+150bp)、B5 低质量AI基建(-2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产先进封装突围真实且政策托底，但分层不均——设备已成、产能受良率约束爬坡、高端材料追赶，真正天花板是自研HBM而非封装。", "summary": "截至2026-06-14，本卡支持卡片05并细化：昇腾多die（910C双die、950PR自研HBM）与'韬定律'把封装变成性能杠杆，台积电CoWoS隔断使整条2.5D/3D栈被迫国产化。突围分层不均——设备最强（华海清科CMP国产份额超90%、2025装备收入40.54亿+33.46%；北方华…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 0, "value": 80.4, "events": [ { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "验证AI算力硬件链的估值承压：流动性虹吸下的先进封装与GPU产业链韧性评估", "summary": "本卡对AI算力硬件链做压力测试，得出'分化韧性'结论。资金流层面确认虹吸：2026-06-08半导体−219.33亿、通信−98.35亿、消费电子−69.92亿元净流出居前[S6]，年内93家半导体公司减持[S7]，叠加CXMT约295亿IPO锚定[S9]。但估值下移是资金流/轮动驱动、非盈利驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 8, "neutral": 1, "value": 48.8, "events": [ { "title": "半导体能效能把AI扩张从“堆GPU”改写为“每MW token产出竞争”，但只能买时间、改善单位经济性，不能让电力瓶颈消失。", "summary": "截至2026-06-16，本卡压力测试“电力预算驱动扩张”逻辑：半导体能效提升真实且重要，Blackwell、先进封装、vLLM/FlashAttention、稀疏MoE和定制ASIC可显著提升每MW有效token产出。但它不能完全对冲电力瓶颈，因为美国数据中心用电量可能从2023年的176 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装/HBM 产业链拥挤度风险压力测试 工作日期：2026 06 17（Asia/Singapore） 分析师：半导体分析师（semiconductors analyst） 立场…", "summary": "先进封装/HBM 产业链拥挤度风险压力测试 工作日期：2026 06 17（Asia/Singapore） 分析师：半导体分析师（semiconductors analyst） 立场：stress test（压力测试 Research note 6 的\"先进封装择时择股\"结论） 板块：A股 CoW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供给限制驱动的半导体能效转型主要表现为通用GPU的系统与精度级优化，而非ASIC/NPU的大规模替代；自研定制芯片的爆发反而加剧了全球先进制程与封装供应链对TSMC的物理集中度。", "summary": "截至2026-06-17，电力瓶颈确在倒逼算力能效转型，但并非单纯的通用GPU向专用ASIC切换。算法多变性（如MoE与动态推理）使自研ASIC面临极高的“代差失效”风险，通用GPU（如NVIDIA Blackwell）凭借软件与架构灵活性仍占据70%以上份额。能效提升主要通过系统精度（FP4/F…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力链可放心承接、先进封装需择时择股、CPO杠杆未出清——双主线轮动方向正确但资金弹性低于纯轮动模型预期。", "summary": "截至2026-06-17，A股微观流动性对Research note 5'电力基建+能效硬件'双主线轮动具备结构性承接能力但弹性不对称。电力设备链ETF连续7周净申购、融资买入占比84%分位、公募超配仅+1.4pct，承接顺畅；先进封装/HBM链融资已达92%分位、公募超配+5.8pct处历史峰值…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 5, "neutral": 0, "value": 73.2, "events": [ { "title": "2027 能源重定价压的是超大规模厂商 ROIC 而非芯片 capex 效率——硅片主导 TCO、能源占比单位数，且性能/瓦溢价+杰文斯回弹反而净支撑前沿逻辑/先进封装/HBM 需求…", "summary": "2026-06-18 压力测试：能源是证伪芯片逻辑的错误杠杆。Epoch 1GW 机房年化 TCO 约 $8.5B，服务器占 ~59%、能源仅 ~7%；交付电价 +30% 仅抬升 TCO 约 +2.1%，翻倍也仅 +7%（自测算），而 IT 寿命 5→3 年使 TCO +41%——刷新节奏的影响是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FOMC 后应以“选择性算力基础设施挤空”而非“无差别 AI 融涨”来交易：多稀缺、空低 ROIC，净多但对冲。", "summary": "2026-06-18 策略综合：前序链条成立，2027 风险是 ROIC 与结构分化而非 capex 悬崖。FOMC 后应做多稀缺算力资产而非泛 AI beta：前沿逻辑/设备/HBM/先进封装、自研 ASIC 与网络、电力与电网受益方应超配；商用 GPU 龙头保留但严控风险；低 ROIC 商用 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产半导体设备与材料已从政策叙事跨入产能交付，能在12–24个月内实质缓解成熟制程与先进封装瓶颈，从而为AI硬件'防御性抱团'提供真实业绩兑现，但先进逻辑天花板仍由海外卡脖子环节决定。", "summary": "截至2026-06-18，AI算力需求驱动全球AI芯片市场迈向4000亿美元规模(+38% YoY)，HBM位元需求+70%，CoWoS仍供不应求。国产替代在设备端取得实质突破：刻蚀国产化率45–50%、薄膜32%、清洗40%；材料端CMP抛光液/12寸硅片明显提速；先进封装(长电XDFOI、通富…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装正在从后道工艺瓶颈升级为受电力、电能质量和客户认证共同约束的基础设施资产，国产替代应先看能效和稼动率模块，而不是直接押注全栈高端替代。", "summary": "2026-06-17 本卡压力测试认为，电力瓶颈已从 AI 数据中心和前道晶圆厂传导至先进封装。Taipower 预计台湾半导体与 AI 相关扩张到 2030 年新增超过 5 GW 用电；TSMC 2024 年能耗 27,456 GWh，折算约 3.13 GW 连续负荷。CoWoS、HBM、测试、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 2, "neutral": 0, "value": 84.6, "events": [ { "title": "半导体设备资本开支韧性对中国AI算力集群与数据中心建设周期的支撑作用分析", "summary": "本卡支持上游设备/材料资本开支韧性论点并向下游延伸：设备资本开支→晶圆产能→国产AI加速器→AI服务器→智算中心的传导链在2025–26年数据中真实可见。中国智能算力由2025年中约788 EFLOPS走向2025年末1037 EFLOPS、2026年约1460 EFLOPS；加速服务器市场202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装降低每bit能耗后，Hyperscalers更可能把能效红利用于扩大推理部署和提高每兆瓦收入，而不是削减资本开支。", "summary": "截至2026-06-19，本卡支持“先进封装与能效提升会缓解电力瓶颈并推动Hyperscaler二次部署”的判断，但强调这是受firm power约束的扩张而非无约束爆发。Microsoft预计2026日历年资本开支约1900亿美元，Alphabet预计2026年1800-1900亿美元，Meta…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 6, "neutral": 0, "value": 62, "events": [ { "title": "电力与封装成本上升会触发AI基础设施技术路线调整，但截至2026-06-20仍不足以触发超大规模云厂商AI芯片采购总量的全面回撤。", "summary": "工作日2026-06-20：本卡压力测试认为，电力和先进封装成本上升不会在2026年触发超大规模云厂商全面削减AI芯片采购，但会提高单位弹性，推动技术路线从“更多GPU插槽”转向“每瓦性能、液冷机架、自研ASIC、提前锁定CoWoS/HBM和电力站点”。真正风险是并网、变电站、CoWoS或HBM配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "维持 Strategy 3 的 20% 半导体总权重是合理压力测试结论，但必须把它结构化为先进封装/HBM/AI ASIC/GPU/国产替代各 4 pp，否则宽 β 配置会让减仓换 …", "summary": "工作日 2026-06-19。本卡对 Strategy 3 中半导体权重降至 20% 的安排进行压力测试。方向上支持削减——SOX 估值已处历史 88% 分位、解禁去估值风险真实——但加三条限定：(1) 20% 仓位的内部结构远比头条数字重要，宽 SOX 等权会损失 50–80 bps 结构 α；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS/HBM带来的瓶颈在物理和认证层面具备结构性，但价格扩散到普通电子布的部分更像周期性紧缺。", "summary": "截至2026-06-20，本卡片压力测试了“AI先进封装推升电子布与散热材料成为结构性瓶颈”的主线。结论是方向成立但必须收窄：CoWoS/HBM确实通过更大中介层、更大基板、更宽HBM接口和更高热密度抬高低Dk、低Df、低CTE电子布与高可靠TIM要求；但可持续瓶颈主要集中在T-glass、NER…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 1, "risks": 7, "neutral": 0, "value": 65.4, "events": [ { "title": "先进封装与Chiplet能显著降低AI算力的边际电力和水资源压力，但无法替代firm MW、并网扩容与低水耗选址，结论是缓解而非根治。", "summary": "截至2026-06-20，本卡支持前序链条但给出边界：先进封装、Chiplet、HBM与机柜级液冷可通过降低每token能耗、减少封装外数据搬运、提升热捕捉效率来显著缓解AI数据中心电力与水资源压力。1 GW IT负载在1.20 PUE下年耗电10.51 TWh；PUE降至1.10可节电0.876…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网交付延迟不会终结高密度 AI 机柜，但会迫使下一代 AI 芯片与集群架构把 perf/W、可限功率和可分阶段并网能力置于峰值算力之上。", "summary": "2026-06-20：本卡对电网设备瓶颈约束 AI 芯片架构的说法进行压力测试。结论是，变压器和开关设备长交期不会让前沿训练集群普遍降密度；GB200 NVL72、Vera Rubin 和 AMD Helios 仍指向高密度机柜级设计。但电网交付延迟会把竞争指标从峰值 FLOPS 推向受限兆瓦下的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "summary": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力约束不会终结AI算力扩张，但会把半导体设计目标函数改写为每瓦有效算力，并提升Chiplet与先进封装的战略溢价。", "summary": "截至2026-06-21，本卡片支持前序判断：电网接入和可交付兆瓦成为AI扩张区域性天花板后，半导体竞争会从峰值算力转向系统级Perf/Watt。数据中心用电增长使每瓦tokens、HBM带宽、die-to-die互连、热设计和先进封装成为容量权变量。Chiplet、CoWoS/SoIC、HBM4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 0, "value": 88.8, "events": [ { "title": "截至2026-06-22，低端封装过剩会压低错误产能池的价格，而高端迭代缺失会保留真正重要产能池的配给风险，使Hyperscaler AI资本开支更重、更不稳定、更容易出现已供电机房…", "summary": "截至2026-06-22，本卡压力测试卡片06：先进封装低端过剩并不能缓解Hyperscaler真正需要的高端CoWoS/HBM4认证产能短缺。TrendForce称2026年行业CoWoS-like产能可接近200,000 wpm、缺口或收窄至约10%，但若高端认证占比只有50%-70%，有效高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力网并网瓶颈与散热极限正成为制约AI算力集群扩张的硬约束，这也倒逼了先进制程/封装（BSPDN、玻璃基板）的代际演进与混合AI端侧推理的结构性爆发。", "summary": "本卡支持TMT分析师对算力基础设施与端侧芯片的看多判断。大规模GPU集群遭遇单机架120-135kW的供电与液冷瓶颈，以及长达5-7年的电网并网时延。这倒逼半导体链加速向CoWoS-L/3D堆叠、背面供电(BSPDN)与玻璃基板等先进封装和底层架构演进，同时驱动推理负载外溢至高 TOPS/W 的高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 8, "neutral": 0, "value": 54.2, "events": [ { "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "summary": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与定制化ASIC的深度整合，是将大模型访存能耗降低95.7%并实现3-5倍能效提升的核心硬件路径，亦是国产半导体链规避先进制程封锁、支撑算力电力约束的战略根基。", "summary": "截至2026-06-23，本卡研判支持通过先进封装与ASIC能效路径应对算力电力约束。2.5D封装结合HBM3E使访存功耗降低95.7%（从35 pJ/bit降至1.5 pJ/bit）；自研定制ASIC能能效比高出GPU约3-5倍，万卡集群年省电费2,943万元人民币，并释放48MW电网容量。国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体资本开支收缩会推迟AI集群扩建的判断成立，但只在AI专用供应瓶颈受冲击时成立；截至2026-06-22，更现实的约束是芯片稀缺、组件通胀与电网通电共同拉长边际集群排期。", "summary": "截至2026-06-22，本卡支持前序主线但收窄机制：半导体资本开支骤降会推迟Hyperscaler AI集群，前提是冲击落在TSMC先进制程、HBM、先进封装、网络和供电等AI关键瓶颈，而非普通消费芯片。现有证据反而显示AI关键资本开支仍在扩张：TSMC指向US$52-56 billion区间高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器/电网瓶颈对AI芯片需求预期与库存周期的传导风险", "summary": "截至2026-06-22，本卡以半导体视角压力测试Card-03。结论：芯片需求仍明确领先于电力瓶颈，2026年不存在AI芯片砍单——约束仍是芯片供给本身（CoWoS售罄至2026年底、英伟达独占约60%；Blackwell积压约360万颗售罄至年中；2026年HBM全部订满且超大规模厂商预付扩产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 6, "neutral": 1, "value": 70.6, "events": [ { "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下修。", "summary": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI芯片能效和封装升级可以提高每个已送电MW的算力收入，但不能消除变压器长周期交付造成的站点送电瓶颈。", "summary": "截至2026-06-23，本卡支持前序判断：半导体厂商正在通过更高性能每瓦、低精度推理、HBM/CoWoS/3.5D封装和自研ASIC，提高每个已送电MW的AI产出。但这只是缓解，不是替代变压器。若站点无法送电，GPU租赁收入仍会延后；若站点已送电，Blackwell/GB300、MI350、Ir…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "summary": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI capex去杠杆尚未体现为开支崩塌，而是由电力、电网、HBM/CoWoS与融资成本共同驱动的ROIC压缩；因此应支持前序风格判断，但把交易执行收窄到具备分配权和电力交付能力的资…", "summary": "锚定2026-06-23，本卡支持前序“AI capex去杠杆+小盘假轮动”的方向，但修正为ROIC/FCF压缩而非2026年capex崩塌。Microsoft、Alphabet、Meta及四大云厂商合计口径仍指向纪录级AI基础设施支出；电力、电网并网、HBM/CoWoS和测试分配共同构成三节点瓶…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 3, "neutral": 0, "value": 87, "events": [ { "title": "中国CSP成本上升背景下，AI算力基础设施供应弹性与扩容压力分析", "summary": "压力测试部分否定线程以电力为中心的框架。AI算力系统是供给绑定而非需求绑定：2026年超大规模厂商资本开支同比+77%至约7250亿美元[S1]，英伟达Blackwell售罄至年中、积压360万颗[S2]，因此约束的是配额优先级而非需求。核心锚定集群凭借锁定电力、GPU配额与表后自发电被隔离；放缓…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与HBM供应链的出口管制风险及其对国产替代路径的影响", "summary": "截至2026-06-24，本卡压力测试第05卡：国产替代可以通过XDFOI/类CoWoS、国产HBM、低精度和更大SuperPoD集群回补效率，但现有证据不支持近中期达到CoWoS+HBM4同等量级的能效。美国管制命中HBM商品、FDP、先进设备、软件及TSV相关制程；华为路线图能证明带宽和系统规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 1, "risks": 6, "neutral": 1, "value": 81.4, "events": [ { "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网缓释后，AI集群上线风险不是消失，而是转移到CoWoS/HBM、液冷验收与网络芯片的串行同步，CPO下行收窄但不等于收入曲线完全恢复。", "summary": "截至2026-06-26，本卡压力测试Research note 7：电网侧缓释可降低CPO最坏收入下修，但不能消除AI集群的串行硬件瓶颈。关键约束从单一接电问题转向先进封装/HBM、液冷机柜验收、交换ASIC/网络集成的节奏匹配。CoWoS缺口可能由约20%降至2026年底约10%，但10%仍足…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 0, "value": 89.8, "events": [ { "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-27的权威工作日背景下，国产HBM良率骤降逼迫国内算力网络化，而全球封装与电力阻尼拉长了云厂商CapEx变现周期，这使得制程免疫且订单排至2028年的可插拔光模块出…", "summary": "截至2026-06-27，本报告评估了HBM良率危机与先进封装瓶颈对算力集群技术演进的影响。国产HBM3封装良率降至2.66%，倒逼国内算力集群转向网络密集的Scale-out拓扑，推高了光模块需求弹性。全球云厂商2026年CapEx直逼7250亿美元，但受制于CoWoS晶圆先进封装冲突与吉瓦级电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电子链条利润反弹主要受AI基础设施投资拉动，具备向存储、设备、材料、先进封装、光互联和数据中心电源散热扩散的潜力，但尚不足以证明消费电子和内需全面复苏。", "summary": "截至2026-06-28，本卡支持上一卡判断：电子链条利润修复主要由AI基础设施、算力芯片和存储涨价驱动，而非内需广谱复苏。国家统计局数据显示，1—5月电子行业利润同比增长103.9%，贡献全部工业利润增量43.1%。NVIDIA、Microsoft、Alphabet、Meta、Alibaba和T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 5, "neutral": 0, "value": 79.2, "events": [ { "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体国产替代对AI算力瓶颈的缓解潜力", "summary": "支持但有限定：国产半导体设备与先进封装确实有效缩短AI算力兑现周期的硅片侧环节。设备国产化率2024约26%→2025约35%，刻蚀/薄膜约40%本土，北方华创(002371.SZ)全球第五、中微(688012.SH)前十三，大基金三期3440亿托底；先进封装为弯道超车路径，长电(600584.S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体产业链的供给侧约束与国产化替代的技术边界", "summary": "本卡对H20/HBM管制下中国国产替代的技术边界做压力测试，部分否定「难以逾越的断层」框架。结论分三层：先进封装（CoWoS替代）是中国相对优势、是可在1–2年跨越的桥（JCET/通富/盛合晶微＋Naura混合键合），并非断层；真正断层在上游——EUV光刻国产化<5%（SMEE约90nm，落后AS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 3, "neutral": 0, "value": 92.6, "events": [ { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,80…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-56…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-ffa474a3f2/", "networkHref": "/investor/research-network/#thesis-ffa474a3f2" }, { "id": "thesis-3d421524cd", "source": "tracked_thesis", "title": "油价地缘底线决定再通胀与降息交易的安全边际", "titleEn": "The geopolitical oil floor defines the margin of safety for disinflation and rate-cut trades", "lane": "油气/地缘通胀", "laneEn": "Oil, gas, and geopolitical inflation", "laneHref": "/investor/lanes/oil-gas-and-geopolitical-inflation/", "investableFocus": "油气/地缘通胀：XOM, CVX, COP, SLB, OXY, 0883.HK, 0857.HK", "stockUniverse": [ { "ticker": "XOM", "name": "Exxon Mobil", "market": "US", "role": "全球综合油气", "roleEn": "全球综合油气", "bucket": "core", "href": "/investor/stocks/XOM/" }, { "ticker": "CVX", "name": "Chevron", "market": "US", "role": "全球综合油气", "roleEn": "全球综合油气", "bucket": "core", "href": "/investor/stocks/CVX/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "SLB", "name": "SLB", "market": "US", "role": "油服与资本开支", "roleEn": "油服与资本开支", "bucket": "core", "href": "/investor/stocks/SLB/" }, { "ticker": "OXY", "name": "Occidental Petroleum", "market": "US", "role": "美国页岩油", "roleEn": "美国页岩油", "bucket": "core", "href": "/investor/stocks/OXY/" }, { "ticker": "0883.HK", "name": "中国海洋石油", "market": "HK", "role": "中国海上油气", "roleEn": "中国海上油气", "bucket": "core", "href": "/investor/stocks/0883.HK/" }, { "ticker": "0857.HK", "name": "中国石油股份", "market": "HK", "role": "中国油气链", "roleEn": "中国油气链", "bucket": "core", "href": "/investor/stocks/0857.HK/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" } ], "stockTickers": [ "XOM", "CVX", "COP", "SLB", "OXY", "0883.HK", "0857.HK", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI" ], "coreView": "赛道：油气/地缘通胀。候选观察标的：XOM, CVX, COP, SLB, OXY, 0883.HK, 0857.HK。当前共有 40 条证据、3 位分析师和 6 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 78, "eventCount": 160, "firstSeen": "2026-04-25", "lastSeen": "2026-06-30", "supports": 16, "risks": 117, "neutral": 27, "analystCount": 14, "sourceTypeCounts": { "研究线程": 3, "分析协作": 16, "研究笔记": 134, "证据更新": 3, "研究主题": 4 }, "topTerms": [ "GOES", "原油", "Brent", "AIDC", "石油", "能源", "油价", "geopolitical", "floor", "defines", "margin", "safety" ], "whatChanged": [ "2026-06-30: [auto-handoff] 低油价对中国工业企业利润及进口贸易顺差的边际影响分析", "2026-06-29: 在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红利（长江电力、中国核电）进行防御性轮动，并辅以日元和美债的尾部风险对冲。", "2026-06-29: 2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "2026-06-29: 美伊卡塔尔复谈促油价地缘溢价退潮与跨资产红利再平衡", "2026-06-29: [auto-handoff] Hormuz shock & core CPI transmission" ], "practical": { "sectorLane": "油气/地缘通胀", "sectorLaneEn": "Oil, gas, and geopolitical inflation", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "XOM", "name": "Exxon Mobil", "market": "US", "role": "全球综合油气", "roleEn": "全球综合油气", "bucket": "core", "href": "/investor/stocks/XOM/" }, { "ticker": "CVX", "name": "Chevron", "market": "US", "role": "全球综合油气", "roleEn": "全球综合油气", "bucket": "core", "href": "/investor/stocks/CVX/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "SLB", "name": "SLB", "market": "US", "role": "油服与资本开支", "roleEn": "油服与资本开支", "bucket": "core", "href": "/investor/stocks/SLB/" }, { "ticker": "OXY", "name": "Occidental Petroleum", "market": "US", "role": "美国页岩油", "roleEn": "美国页岩油", "bucket": "core", "href": "/investor/stocks/OXY/" }, { "ticker": "0883.HK", "name": "中国海洋石油", "market": "HK", "role": "中国海上油气", "roleEn": "中国海上油气", "bucket": "core", "href": "/investor/stocks/0883.HK/" }, { "ticker": "0857.HK", "name": "中国石油股份", "market": "HK", "role": "中国油气链", "roleEn": "中国油气链", "bucket": "core", "href": "/investor/stocks/0857.HK/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" } ], "stockTickers": [ "XOM", "CVX", "COP", "SLB", "OXY", "0883.HK", "0857.HK", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI" ], "primaryStocks": [ { "ticker": "XOM", "name": "Exxon Mobil", "market": "US", "role": "全球综合油气", "roleEn": "全球综合油气", "bucket": "core", "href": "/investor/stocks/XOM/" }, { "ticker": "CVX", "name": "Chevron", "market": "US", "role": "全球综合油气", "roleEn": "全球综合油气", "bucket": "core", "href": "/investor/stocks/CVX/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "SLB", "name": "SLB", "market": "US", "role": "油服与资本开支", "roleEn": "油服与资本开支", "bucket": "core", "href": "/investor/stocks/SLB/" }, { "ticker": "OXY", "name": "Occidental Petroleum", "market": "US", "role": "美国页岩油", "roleEn": "美国页岩油", "bucket": "core", "href": "/investor/stocks/OXY/" }, { "ticker": "0883.HK", "name": "中国海洋石油", "market": "HK", "role": "中国海上油气", "roleEn": "中国海上油气", "bucket": "core", "href": "/investor/stocks/0883.HK/" } ], "instrumentTags": [ "XOM", "Exxon Mobil", "CVX", "Chevron", "COP", "ConocoPhillips", "SLB", "SLB", "OXY", "Occidental Petroleum", "0883.HK", "中国海洋石油" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 91 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 14, "recentEvents": 91, "supportRatio": 0.1, "riskRatio": 0.73, "conflictScore": 82.1, "evidenceDepth": 100 }, "score": 74.3 }, "events": [ { "date": "2026-06-30", "title": "[auto-handoff] 低油价对中国工业企业利润及进口贸易顺差的边际影响分析", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: f2e0bdfb-0125-4d0a-b62f-80a7e18e510b 结果摘要: Wrote handoff_response.md (7787 bytes); wrote handoff_response.en.md (8309 bytes) — A 30%…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 22.25, "matchedTerms": [ "Brent", "能源", "油价", "margin", "energy" ], "href": "" }, { "date": "2026-06-29", "title": "2026年石油供应中断对全球能源品种及炼油产能的冲击评估", "summary": "支持卡片01“油价持久性是枢纽”的判断并加以锐化。霍尔木兹2025年日均输送约2000万桶/日（原油1500万+成品油550万），全球闲置产能几乎全在海峡之内，绕行管道仅7-8mb/d（不足正常一半），规模化对冲不存在。冲击非线性传导：3月约1200万桶/日无法运抵亚洲、其中800万为中硫油，亚洲炼厂被迫改炼轻质低硫，柴油/航煤减产，4月中间馏分供应降1.…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 27, "matchedTerms": [ "原油", "石油", "能源", "油价", "energy" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 22.5, "matchedTerms": [ "Brent", "inflation", "evidence", "risk", "dominant", "transmission", "Hormuz", "supply" ], "href": "" }, { "date": "2026-06-29", "title": "美伊卡塔尔复谈促油价地缘溢价退潮与跨资产红利再平衡", "summary": "美伊卡塔尔复谈促油价地缘溢价退潮与跨资产红利再平衡 Geopolitical De-escalation & Crude Oil Re-pricing 美伊复谈促Brent油价跌至72.27美元，地缘防范溢价出清，如何重新评估资源品板块（石油、煤炭）与公用红利、出口中盘红利之间的配置权重？", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "neutral", "supportLabel": "观察", "score": 22.25, "matchedTerms": [ "Brent", "石油", "油价", "geopolitical" ], "href": "" }, { "date": "2026-06-29", "title": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "summary": "对中国，2026年油价/柴油冲击既是被缓冲的进口税、又是战略加速器，且净额上比卡片01–02的美欧日集团损害更小。三大事实：（1）进口税真实但落在PPI而非CPI——5月PPI同比+3.9%（2022年7月来最快）、CPI仅+1.2%，疲弱内需把冲击吸收为生产者利润压缩；（2）进口税被三重缓冲——全球最大约14亿桶战略储备、作为受制裁油折价买方、疲弱需求；…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "石油", "能源", "油价", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红利（长江…", "summary": "本报告针对2026年三季报TMT硬件估值折价预期，校准了全球全天候和风险平价模型。战术防御性风险平价（TDRP）将AI成长股风险预算由25.0%下调至8.0%，防守红利上调至40.0%，以对冲55.0%的TMT增长溢价折价。模型显示，电力配给约束下算力效率的TCO优势在24/7清洁能源下扩大至9.97%，高功耗算力中心毛利率面临负增长。此外，建议红利套组内…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "原油", "石油", "能源" ], "href": "" }, { "date": "2026-06-28", "title": "能源商品成本传导与USD/CNY汇率风险的压力测试", "summary": "截至2026-06-28，本卡对卡3'不上调外汇对冲'结论做压力测试。被测机制（人民币贬值抬高能源进口成本、压低通胀对冲实际人民币收益）真实但属二阶，且当前被相反力量压过：能源对冲人民币收益的主导因素是美元油价中伊朗战争地缘溢价的崩塌，而非汇率。人民币油价成本=布伦特(美元)×USD/CNY，3月峰值至今每桶人民币成本降约41%，其中美元价格腿贡献−40%…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22, "matchedTerms": [ "原油", "能源", "油价", "energy" ], "href": "" }, { "date": "2026-06-28", "title": "核心服务与住房通胀的粘性：美联储政策的后续驱动力", "summary": "能源转为通缩之际，核心服务与住房粘性足以支撑美联储鹰派、抵消能源通缩。关键在指标分化：整体PCE 4.1%/整体CPI 4.2%将因油价基数效应回落，但美联储盯住的核心PCE升至3.4%（2023年10月以来最高），由占核心PCE一半以上、运行约3.5%-3.9%的超级核心（剔除住房的核心服务）驱动。住房同比3.4%（OER 3.3%），去通胀放缓而非结束…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 21.75, "matchedTerms": [ "石油", "能源", "油价", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "在风险平价框架下，能源减配并转向黄金/结构性金属不会突破波动约束，反而降低能源方差集中度；代价是牺牲部分纯油价冲击下的上行夏普。", "summary": "截至2026-06-28，本卡支持将能源对冲重心从USD/CNY与原油转向黄金和结构性金属。建议维持35/25/25/15股/债/对冲/现金框架，但把25%通胀对冲仓位重配为13%黄金、8%结构性金属、4%能源尾部。按六资产协方差自测算，组合波动率从10.30%降至9.15%，能源方差贡献从31.6%降至10.8%。广义通胀情景下夏普从0.01升至0.09…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "原油", "能源", "油价" ], "href": "" }, { "date": "2026-06-27", "title": "地缘价格冲击通过霍尔木兹海峡受阻和灰色市场运费摩擦（每桶4.5美元）改变原油供应路径并拉动首轮通胀，尽管油价已降温至72.95美元，但政策反应焦点已转移至长期通胀预期锚定上，对资产配置构成高…", "summary": "分析表明，2026年5月中东地缘冲突导致霍尔木兹海峡每日1100万-1400万桶原油供应受阻（占全球约15-20%），叠加全球5月产量同比下降3.9%（至9450万桶/日），推动Brent油价飙升至120美元/桶。虽6月油价回落至72.95美元，但滞后影响仍存，5月美CPI升至4.2%（能源通胀涨23.5%贡献其60%的跳升）。能源直接投入产出传导较窄，但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32.25, "matchedTerms": [ "原油", "Brent", "石油", "能源", "油价", "energy" ], "href": "" }, { "date": "2026-06-27", "title": "尽管地缘缓和推动原油价格回落，但原油底部防守、AI负荷拉动天然气结构性上涨以及核心CPI的传导滞后，共同决定了能源下行无法抵消核心通胀的再加速风险，美联储政策中枢仍将维持更高更久。", "summary": "地缘风险缓和使原油价格暴跌至$78–$80/桶，但这不足以抵消核心通胀再加速风险。原油实物库存处于历史低位，且受OPEC+在WTI $75防守底线的制约；而AI数据中心表后气电需求推动天然气价格结构性上涨至$3.92/MMBtu以上，构成更持久的通胀源。核心通胀受黏性服务与薪资驱动，直接能耗传导较弱。美联储仍受通胀硬约束，利率长期维持高位。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 23, "matchedTerms": [ "原油", "能源", "油价", "OPEC", "energy" ], "href": "" }, { "date": "2026-06-27", "title": "虽然备忘录缓解了短期实物原油压力，但霍尔木兹海峡的结构性脆弱与15%-25%的二次通胀概率支撑将大宗商品战术超配至20%，并应采用高分红个股现货替代及低配看涨期权策略进行防御。", "summary": "截至2026-06-27，评估表明尽管扫雷进展与伊朗出口正常化将推动短期WTI均价回落至67.95美元且跌破70美元概率升至62.85%，但15-25%的能源价格二次上尾再脉冲风险及长期地缘瓶颈支持将大宗商品配置超配至20%。在A股通道中，因159945等ETF极度缺乏流动性（成交仅106万元），建议以中国海油、中国石油等高分红龙头个股作为替代通道，配合低…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22, "matchedTerms": [ "原油", "石油", "能源", "energy" ], "href": "" }, { "date": "2026-06-27", "title": "Brent跌至约72美元后，Warsh美联储的高利率前提应从“结构性再加息”下修为“数据条件化高位平台”：基准终端利率上沿3.75%-4.00%，降息需等待核心PCE和预期共同确认。", "summary": "2026-06-27：本卡片压力测试“长期高利率”前提。结论是油价回落削弱了单凭能源再加息的逻辑，但尚不足以触发降息定价；6月FOMC维持3.50%-3.75%，SEP年末中位数3.8%，核心PCE仍2.9%，长期通胀预期3.6%。基准为55%概率高位平台，30%概率油价/PCE触发转鸽，15%概率二次能源冲击再加息。组合应以3.75%-4.00%为基准终…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22, "matchedTerms": [ "Brent", "能源", "油价", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "油价波动与能源转型对海工装备资本开支的影响", "summary": "压力测试'油价跌→资本开支降→601890订单受损'链条，结论是部分否定其近端版本。2026年油价（Brent 6月22日79.25美元，EIA年均约95美元）由霍尔木兹地缘溢价抬高，资本开支批复环境偏好；上游虽连续第二年下滑约2%-3%，但海工是韧性最强的一块：2026年海工FID约1070亿美元、海底支出>350亿美元/年且在40-60美元时80%-9…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "Brent", "能源", "油价", "energy" ], "href": "" }, { "date": "2026-06-26", "title": "[auto-handoff] 统一今日油价引用基准", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: c10ed69f-8e17-495d-a490-77cf57ca559f 结果摘要: Wrote handoff_response.md (9099 bytes); wrote handoff_response.en.md (9735 bytes) — Macro…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "supports", "supportLabel": "支持", "score": 22.75, "matchedTerms": [ "Brent", "能源", "油价", "inflation", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "能源价格波动空间足以支撑当前通胀对冲，但应把能源视为带再平衡纪律的尾部风险工具，并用黄金分散油价回落后的政策失误与实际利率风险。", "summary": "截至2026-06-27，支持上一卡能源+黄金通胀对冲，但强调其是波动率与尾部风险对冲，而非单边油价牛市。Brent在2026-06-26报73.60美元/桶，较52周高点明显回落；IEA仍显示2026年供应下降3.9 mb/d、库存快速消耗，霍尔木兹风险与低库存保留上行凸性。同时EV替代和2027年供应反弹限制中期油价beta。建议维持15%实物资产仓位…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22.25, "matchedTerms": [ "Brent", "能源", "油价", "energy" ], "href": "" } ], "timeline": [ { "date": "2026-04-25", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "霍尔木兹冲击下的中国原油库存缓冲", "summary": "霍尔木兹冲击下的中国原油库存缓冲 在霍尔木兹海峡运输受阻、Brent二季度仍处高位的情景下，中国2025年以来积累的原油库存能否把全球油价冲击转化为相对能源安全优势？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-04-28", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "[auto-handoff] 原油价格通胀传导", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 7df2a899-1970-4e7a-89e9-ba83b65c0357 结果摘要: # Role Global Macro Analyst (全球宏观分析师). # Expertise US/EU/Ja…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-01", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "OPEC+已从价格压制者转型为‘结构性溢价保护者’，其备用产能仅作为防范极端飙升的缓冲区，而非平抑日常地缘溢价的工具。", "summary": "OPEC+当前拥有约510万桶/日的备用产能，但由于沙特2026年财政平衡油价升至$86/bbl，其投放意愿极低。Brent价格已被分解为$68的基本面成本及约$19的风险溢价（含$6的结构性霍尔木兹成本）。OPEC+更倾向于维持$85-$95的定价区间以平衡收入与需求，这为全球核心通胀提供了稳固…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "能源冲击最终可能回落，但在 6 月 SEP 前的持续性足以推高总体通胀并制造有限核心传导风险，因此 Fed 的实际基准应从“即将恢复降息”转为“更久按兵不动”。", "summary": "本卡以 energy-analyst 视角压力测试 card-01 的宽松倾向基准。结论是油价冲击尚未变成自我强化的核心通胀 regime，但截至 2026-05-02 回落不够快，Fed 难以把 3-4 月总体通胀完全视为噪音。EIA 仍预计 Brent 在 2Q26 触及 $115/b，汽油和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 2, "neutral": 1, "value": 38.6, "events": [ { "title": "Hormuz供给冲击在急性期内强烈，但正常化轨迹（4Q26低于$90/b）所产生的基期反转效应使油价对美联储的约束具有时限性——首次降息推迟至2026年Q4而非推入2027年，宽松窗…", "summary": "能源分析师否定油价从结构上封锁美联储2026年宽松窗口的论断。在3月7.5 mb/d、4月9.1 mb/d减产、2Q26 Brent峰值约$115/b、4Q26回落至$90/b以下的正常化情景下，能源分项对标题CPI的同比贡献从2Q26的约+2.3–2.9 pp机械性下降至4Q26的约+0.8–1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "3–4 月能源冲击大概率不会在 Q2 内消退；直接通胀贡献会被基数效应吞掉，但间接传导和通胀预期渠道会在 7–8 月接力，使 Fed 鸽派窗口比 Research note 01 暗…", "summary": "在 OPEC+ 维持纪律、美国页岩供给响应慢、炼油结构性偏紧、SPR 已不可用以及红海地缘溢价仍在曲线中等多重力量叠加下，Brent 与美国汽油/柴油价格更可能在 Q2 2026 维持在 85–92 / 3.65–3.85 区间，而不是快速消退。直接渠道在峰值给整体 CPI y/y 增加约 +0.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "双重供给冲击：中东能源危机与美中关税升级叠加——重新定价美联储宽松窗口 分析师： global macro | 全球宏观分析师 日期： 2026 05 03 任务： 每周宏观报告 立…", "summary": "双重供给冲击：中东能源危机与美中关税升级叠加——重新定价美联储宽松窗口 分析师： global macro | 全球宏观分析师 日期： 2026 05 03 任务： 每周宏观报告 立场： 初始研判 执行摘要 2026年5月，两场同步爆发的供给冲击正向市场交汇：中东能源危机在单月内将 Brent 原…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 0, "neutral": 1, "value": 52, "events": [ { "title": "[auto-handoff] 油价冲击对CNH的影响", "summary": "致 外汇策略师： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: 480c6f05-bd09-4eb3-b510-cfd771ed7b5b 结果摘要: Wrote handoff_response.md (8,813 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "2026年5月4日，因中东局势升级，布伦特原油价格跳升至每桶114.44美元。", "summary": "2026年5月4日，因中东局势升级，布伦特原油价格跳升至每桶114.44美元。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-08", "supports": 1, "risks": 0, "neutral": 0, "value": 59.8, "events": [ { "title": "EIA预计2026年第二季度布伦特原油价格将达到115美元/桶。", "summary": "EIA预计2026年第二季度布伦特原油价格将达到115美元/桶。", "sourceLabel": "证据更新", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-09", "supports": 0, "risks": 1, "neutral": 1, "value": 50.2, "events": [ { "title": "能源再平衡的触发条件为布伦特原油价格连续21天低于90美元/桶。", "summary": "能源再平衡的触发条件为布伦特原油价格连续21天低于90美元/桶。", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "能源价格冲击与香港制造业PMI的传导机制", "summary": "能源价格冲击与香港制造业PMI的传导机制 Transmission of Energy Price Shocks to HK Manufacturing PMI 布伦特原油价格回落能否有效缓解香港制造业的输入型成本压力并扭转PMI收缩趋势？", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "4月数据同时验证制造补库韧性与结构性能源脱钩，强化中国近端贸易/汇率/峰会姿态，约束被前置到亚洲炼厂与OPEC+而非中国本身。", "summary": "2026年4月中国出口+14.1%、进口+25.3%但原油进口-20%、成品油出口降至十年低点，是制造补库韧性与结构性能源脱钩两者并存：进口井喷集中在半导体、工业金属、农产品，属资本开支与补库驱动而非消费驱动；原油-20%是库存释放+炼油毛利压缩+新能源车需求结构性退坡三力叠加。这一组合让贸易顺差…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "「能源换芯片」完整版概率偏低、部分版概率过半，对 A 股的最优解是先做风险溢价压缩交易（半导体设备 + 光模块 + 能源化工中游 + 出行链），再分阶段回补自主可控算力。", "summary": "「能源换芯片」是结构性不对称交易：美方让渡受国会与产业鹰派牵制的管制租金，中方让渡有限的伊朗调停杠杆。完整大交易 2026 年内概率仅 15–20%，但部分版本（H 系列许可证扩容 + 倡导式调停）概率 45–55%，足以驱动 A 股 4–12 周风险溢价压缩行情。半导体设备与光模块是最高确定性的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 0, "neutral": 1, "value": 60.4, "events": [ { "title": "[auto-handoff] 油价长期处于$105上方对A股能源链的影响", "summary": "致 首席策略师： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: 25f9f304-a50c-4b9a-9978-a4a176c360b9 结果摘要: Wrote handoff_response.md (4056 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 5, "neutral": 0, "value": 28.8, "events": [ { "title": "Brent高于104美元会通过输入性PPI和利润率压缩强化A股从外交beta和周期成长向水电、受监管公用事业与红利资产切换的趋势。", "summary": "截至2026-05-17，本卡支持上一卡判断：Brent突破104美元已成为独立的通胀与利润率冲击。美国4月CPI同比3.8%、能源同比17.9%；中国4月CPI同比1.2%、PPI同比2.8%，且石油、燃料加工、化工链条环比涨幅显著。油价从89.4美元升至104.2美元的16.6%冲击，按IMF…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "原油腿对需求衰退的脆弱性显著高于铜腿：原油超配应减半并加95美元看跌对冲，铜超配维持并集中远月。", "summary": "截至2026-05-17，本卡对Research note 06的'超配原油+铜'对冲腿进行压力测试。原油端需求衰退信号已可见：美国汽油需求同比−2.1%、中国表观石油需求同比−1.8%、IEA将2026全球需求增速下调至0.78 mb/d；历史阈值在Brent 110–115美元持续两个季度，目…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球能源供应侧的结构性僵化与中国制造业复苏带来的需求修复，共同构筑了原油价格的稳固底部，支撑能源板块作为再通胀对冲工具的超配价值。", "summary": "截至2026年5月17日，能源基本面强力支撑PPI温和再通胀逻辑。供应端，OPEC+维持自愿减产纪律，且全球传统能源长期开发投入（CAPEX）不足导致供应弹性持续受限。需求端，中国制造业景气度回升与“反内卷”政策共振，强化了对原油及天然气的工业需求支撑。目前布伦特原油站稳85美元/桶上方，60%的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源价格冲击下的资产轮动逻辑：资源对冲还是价值回归？", "summary": "能源价格冲击下的资产轮动逻辑：资源对冲还是价值回归？ Are sustained high oil prices driving a rotation into specific Energy/Resource sectors as a standalone geopolitical hedge,…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 3, "neutral": 2, "value": 48.2, "events": [ { "title": "霍尔木兹/Kharg应被定价为尚未解除的实物流量冲击，而不是基于和平标题快速下调风险溢价。", "summary": "截至2026-05-19，支持上调油价、油轮和通胀风险溢价。Windward显示霍尔木兹2026-05-17仅14艘通行且存在暗航与油轮队列；Kharg卫星油膜、TankerTrackers所称28天无成功海运原油出口、以及23艘油轮聚集，说明有效供给仍受阻。Brent $112.10/bbl并未…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A 股石化与全球油价宏观的定价背离", "summary": "A 股石化与全球油价宏观的定价背离 A-share petrochemicals vs global oil macro: pricing dislocation global-macro 把 Brent 锚在 $110–125/bbl、energy 给出馏分油净缺口 1.1–1.4 mb/d 与…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "A 股石化与全球油价宏观的定价背离", "summary": "A 股石化与全球油价宏观的定价背离 global-macro 把 Brent 锚在 $110–125/bbl、energy 给出馏分油净缺口 1.1–1.4 mb/d 与 LSFO 净缺口 0.35–0.55 mb/d、materials 认定亚洲石脑油裂解进入现金亏损区；但 A 股 5/19 恒…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "霍尔木兹供给扰动通过原油品种替代和VGO/低硫调和组分争夺，把油价风险转化为炼厂裂解价差和成品油短缺风险。", "summary": "截至2026-05-19，本卡支持上一卡的实物流量冲击判断，并把传导落到炼厂端。霍尔木兹扰动不只是抬高Brent，而是通过中/重质含硫海湾原油缺口改变全球炼厂原油池，导致复杂炼厂受益、原油受限炼厂降负荷。基准测算显示，5-6月可交易馏分油净缺口为1.1-1.4 mb/d，LSFO/VLSFO净缺口…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "布伦特原油突破105美元与物流成本飙升的'双击'强化了做空薄利/可选零售商的逻辑，但也警告宏观能源税可能压垮整体消费基本盘，使规模赢家同样面临绝对收益风险。", "summary": "当前布伦特原油突破105美元/桶，叠加全球集运运价同比大涨逾44%以及美国内陆卡车运费上涨15%，对零售行业构成了严峻的滞胀压力测试。能源价格飙升直接增加了干线与最后一公里物流成本，并像累退税一样大幅吞噬了中低收入人群的实际可支配收入。这一冲击验证了前卡中'低配DG/DLTR/TGT'的逻辑，因为…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "[auto-handoff] Sustained Strait of Hormuz disruption and Brent oil risk premium", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: cab083c9-1311-4ea6-ae23-6069b4f874e9 结果摘要: Wrote handoff_response.md (16952 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 1, "value": 61.4, "events": [ { "title": "作为能源行业分析师，我对卡片 1（由首席经济学家发布）中关于霍尔木兹海峡危机走向“外���和谈与地缘溢价回吐（情景 A，概率 55%）”的基线判定表示全力支持。 围绕“在外交和谈可能…", "summary": "作为能源行业分析师，我对卡片 1（由首席经济学家发布）中关于霍尔木兹海峡危机走向“外���和谈与地缘溢价回吐（情景 A，概率 55%）”的基线判定表示全力支持。 围绕“在外交和谈可能缓解地缘风险的背景下，全球原油库存与产量结构的变动将如何影响布伦特原油价格的底部支撑”这一核心使命，我从 OPEC+…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡片3 的工业-防务背离是真实的板块轮动信号，但宏观加总后油价回吐的反通胀效应碾压防务/冗余溢价，情景A 的基线是温和再通缩+AI主导的再加速，不是“隐性滞胀”。", "summary": "我否决“情景A演变为美国制造业衰退+全球贸易高通胀的隐性滞胀”这一悲观叙事。卡片3 揭示的 OFS/重工资本开支收缩 (~0.6% of GDP, GDP 拖累上限 -10 bp) 是真实的板块痛点，但放进国民账户根本无法构成全国衰退；而 AI/数据中心资本开支 ~$3,750 亿 (是页岩下行的…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 2, "risks": 1, "neutral": 1, "value": 78.8, "events": [ { "title": "结构性高能源成本向核心CPI的传导机制与全球央行政策硬转向 报告日期： 2026年5月23日 报告作者： 首席经济学家 ( chief economist ) 本卡立场： 支持 — …", "summary": "结构性高能源成本向核心CPI的传导机制与全球央行政策硬转向 报告日期： 2026年5月23日 报告作者： 首席经济学家 ( chief economist ) 本卡立场： 支持 — 坚决支持上游结构性高成本将穿透并长期重塑核心CPI，迫使央行降息预期根本性转向。 1. 核心执行摘要 截至本研究院权…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2026年5月24日，地缘停火协议头条虽引发纸面油价与运费即期大幅下挫，但因霍尔木兹海峡等物理通道开放恢复滞后、高买入成本替代原油仍在途、以及高价长约运力合同刚性，导致独立炼厂、大宗…", "summary": "本报告（2026年5月24日）指出，中东停火传闻虽触发Brent原油回落至102.58-104.92美元/桶，但物理通航恢复极度滞后。4-5月高价替代原油在途运输导致6-7月抵港时面临严重负裂解价差与存货跌价损失，重创亚欧独立炼厂。同时，锁定在110,000至181,700美元/天的油轮长约及高额…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "能源价格波动对人民币汇率及跨资产对冲策略的影响", "summary": "Card-02 给出的全天候/风险平价配置在能源-增长-通胀三维上合理，但隐含约 35–45% NAV 的净 USD 多头未被显式管理。本卡通过四象限压力测试（油跌-CNY 升、油涨-CNY 贬、价稳汇稳、单边 CNY 贬值）显示，未对冲组合 90 日 NAV 区间约 −4.8% 到 +3.2%。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SPR战略缓冲池的枯竭导致能源库存周期从“弹性吸收”转为“刚性传导”，原油“稀缺溢价”常态化与LNG跨品种价格联动共同确立了全球能源产业链的“结构性高成本”定价范式。", "summary": "随着美国SPR跌破375百万桶底线，能源市场实质性丧失战略缓冲。本报告验证了在低库存水平下，供给侧冲击将直接由商业库存吸收，通过炼厂采购成本和每桶10-15美元的“稀缺溢价”抬升原油绝对价格中枢。同时，高油价通过油价挂钩长协和替代效应传导至LNG与天然气市场。库存周期的波动最终刚性转化为全产业链的…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-24", "supports": 1, "risks": 3, "neutral": 0, "value": 57.2, "events": [ { "title": "缺失的 USD 3–5/桶 地缘溢价应主要落在 Brent M2–M9 段，把 Dec26–Dec27 价差从 −2.30 推向 −0.80 左右——最佳表达是多 Dec26 / 空…", "summary": "能源分析师支持 card-04 关于 Brent 远期曲线缺 USD 3–5/桶地缘溢价的判断，并把缺口落到曲线具体位置：M2–M6 急性窗口 +2.5–3.0/桶（霍尔木兹+保险），M6–M18 结构性尾段 +1.0–1.5/桶（制裁与黑暗车队）；预测 Dec26–Dec27 价差从 −2.30…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh 时代联储在 Hormuz 油价冲击与关税通胀粘性下的反应函数", "summary": "Kevin Warsh 于 5 月 13 日以 54-45 接任联储主席,Powell 留任理事。接手时的宏观组合是教科书式的滞胀风险:4 月 FOMC 在 3.50-3.75% 维持但出现 1992 年以来最多的 4 票反对;4 月 CPI 同比 3.8%(能源贡献 40%+),核心 2.8%,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "极值政策分化下，日元套利交易正临近以“干预与VaR共振”为特征的系统性平仓爆破点，而人民币套利则面临央行“离岸挤空”的非线性行政打击风险。", "summary": "在美联储维持高利率以对抗油价冲击的背景下，全球货币政策分化达到1990年代以来极值。日元套利交易面临干预悬崖与防御性紧缩风险，投机存量达5000亿美元，一旦Brent原油突破130美元，将触发系统性VaR平仓风暴；而人民币套利交易受中美负利差驱动，但受资本管制保护，极易遭受央行抽干离岸流动性导致的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "原油地缘溢价对电力远期偏多，但它更像峰段/冬季/低碳灵活性期权，而不是平直基荷多头；在当前碳价下，燃煤仍可能保留边际成本优势。", "summary": "截至2026-05-24，本卡支持前序“原油地缘溢价会传导至电力远期”的方向，但修正为选择性传导：Brent Dec-26/Dec-27 公开报价已显示+USD 10.55/桶 backwardation，油价溢价主要通过油价指数化LNG、燃气边际机组和稀缺小时燃油机组抬升峰段/冬季电力远期。按T…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 5, "neutral": 2, "value": 44.6, "events": [ { "title": "实物原油供给赤字的长期化将打破央行的通胀容忍红线，迫使全球宏观叙事从“预防式降息”向“滞胀紧缩”急剧反转，触发系统性Risk-Off抛售。", "summary": "本报告通过宏观压力测试指出，若Hormuz断供拖延至Q3且Brent油价直奔$130，美联储将被迫完全撤销2026年降息指引甚至重启加息讨论，10年期美债有重返4.8%的风险；欧洲央行将陷入深陷滞胀与主权利差走阔的死局；日本央行则极大可能被迫紧急加息以保卫日元。这将彻底逆转当前的降息计价，引发全球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "$88/桶的油价预期过于激进，OPEC+减产防御与航运摩擦将支撑油价在$92-$94/桶筑底，能源内部利润向炼化下游切换。", "summary": "针对美伊备忘录将布伦特原油推低至$88/桶的基准假设进行压力测试。分析指出，受制于OPEC+财政底线及航运正常化滞后，原油底部更有可能在$92-$94/桶。产业链利润将从上游E&P和VLCC中游航运向炼油及石化下游转移。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "30/25/25/15/5框架在能源中性情景下可执行，但在Brent>$90/TTF>€45的能源上行情景下，55%核心仓位面临同向毛利压力、组合分散化失效，必须叠加能源敏感性矩阵并…", "summary": "从能源视角对30/25/25/15/5食品农业框架做压力测试：该框架并非能源中性，55%的核心仓位（30% B2B + 25%粮食安全基建）对柴油、工业蒸汽、天然气-合成氨-尿素、原油-塑料包材高度敏感。最大单一风险锚是气头/煤头合成氨与尿素链条；安琪酵母作为高能耗发酵企业还面临2027年碳市场扩…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观二次通胀压力测试：重估美伊霍尔木兹备忘录破裂概率与通胀传导路径 研究院权威工作日期 ：2026 05 25 [自测算] 主笔分析师 ：global macro (全球宏观分析师)…", "summary": "宏观二次通胀压力测试：重估美伊霍尔木兹备忘录破裂概率与通胀传导路径 研究院权威工作日期 ：2026 05 25 [自测算] 主笔分析师 ：global macro (全球宏观分析师) 本卡立场 ： stress test (压力测试与风险重估) 一、 执行摘要 截至 2026 05 25 ，全球宏…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 1, "risks": 3, "neutral": 2, "value": 66, "events": [ { "title": "油价供应冲击与全球能源通胀传导风险评估 前沿宏观研究院 | 专题研究报告 日期： 2026年5月26日 分析师： energy analyst (能源行业分析师) 立场： 支持 (支…", "summary": "油价供应冲击与全球能源通胀传导风险评估 前沿宏观研究院 | 专题研究报告 日期： 2026年5月26日 分析师： energy analyst (能源行业分析师) 立场： 支持 (支持Research note 1关于“波动率脱钩”与长期期限溢价未充分定价的判断) 核心摘要 截至 2026年5月2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "经压力测试验证，“三桶油”高股息避险锚并非铁板一块，中国石油展现极强自由现金流韧性可作为核心超配标的，而中国石化在转型资本支出与成品油销量下滑双重挤压下沦为高风险红利陷阱。", "summary": "截至2026-05-27，能源行业分析师对“三桶油”进行了高股息可持续性压力测试。测试显示，大规模转型资本支出叠加Brent油价调整至$70/桶将导致“三桶油”发生严重财务分化。中国石油凭借气价稳定与一体化结构表现出最强现金流韧性（股息覆盖率1.31倍）；中国海油受纯上游影响敏感，股息覆盖率降至0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国炼厂疲弱削弱原油需求峰值，但美国就业韧性和欧洲能源进口结构会放大Brent>$100后的通胀持续性，因此本卡支持Research note 03的二次通胀传导判断。", "summary": "2026-05-27卡片判断：全球需求分化不是压制二次通胀，而是压低油价峰值、放大通胀持续性。中国4月原油加工量5,465万吨、同比-5.8%，提示词中的76%开工率未找到直接来源，按低置信度处理；但美国4月非农+115,000、失业率4.3%、工资同比3.6%，欧洲4月HICP能源贡献+0.99…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "中国炼厂的弱势是春检与利润深亏的周期性压制，以及新能源车与LNG重卡结构性替代的共同结果；若开工率维持低位，将成为阻断美国二次通胀传导并平抑全球油价的宏观安全阀。", "summary": "2026-05-27卡片分析：中国4月原油加工量同比下降5.8%至5465万吨，炼厂因高成本与春检面临深亏，产能利用率降至63.59%。5-6月若开工反弹，原油增量将推升Brent破$100（概率70-75%），传导美国二次通胀（概率65-70%），迫使美联储6月释放鹰派信号；若开工持续低迷，则B…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-27", "supports": 1, "risks": 3, "neutral": 0, "value": 61.4, "events": [ { "title": "通过增配上游能源对冲至10%、降配利率敏感小盘股至4%、并将中国与港股Beta优化至10%的红利与回购哑铃，权益组合在美债4.5%高利率与油价通胀税下构建了强大的现金流安全垫。", "summary": "本卡片作为首席策略师全力支持并融合前序卡片研究。在Brent原油约$99.18且美债收益率4.5%的高通胀税环境下，建议将整体降配后的56%权益组合重新分配：AI硬核龙头维持12%；削减非AI质量成长至10%；增配传统能源对冲至10%；配置防御性红利10%；砍半全球小盘至4%；并将中国与港股Bet…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "如果Hormuz重开只能渐进，Brent近$96并不构成真正的低油价环境，小盘beta的支撑更脆弱，AI只能依靠盈利可见度继续跑赢。", "summary": "截至2026-05-27，本卡认为油价“和平交易”已有部分过度定价。Brent约$96.47低于恐慌区间，但若Hormuz只能渐进重开，EIA的5-6月约$106/b假设、2Q26库存下降8.5 million b/d、以及Fed仍受能源通胀约束，意味着小盘股获得的低油价/低收益率支撑并不稳固。A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "上游能源是必要但不足够的通胀对冲：保留约10%高质量仓位可以吸收部分油价冲击，但安全垫必须与防御红利、AI现金流龙头和更低小盘beta共同构建。", "summary": "截至2026-05-27，本卡对“增配上游能源作为通胀对冲”采取压力测试立场。结论是：上游能源在Brent $95-$105和10年期美债约4.50%的环境中有必要配置，因为库存偏紧、能源CPI贡献明显且上游现金流对油价有弹性；但10%能源仓位只能部分抵消油价通胀税，不能单独构成完整安全垫。Q1数…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "哑铃配置方向正确但回撤上限被低估——保留10%整体仓位，把红利/互联网内部比例从60/40调至65/35，并加5%的HKD利率对冲。", "summary": "截至2026-05-27，本卡对'红利央国企60%+核心互联网40%'哑铃配置作压力测试。结论：哑铃方向正确但条件性脆弱。60%红利篮子加权股息率6.8-7.2%且约29%子板块(能源+煤炭)对油价有正beta，构成真实现金垫；40%互联网篮子综合股东回报率仅3.6-4.0%，且因HKD联汇导致H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 0, "neutral": 1, "value": 75.8, "events": [ { "title": "[auto-handoff] 原油地缘溢价与美伊协议进展对全球通胀/利率路径的反馈", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: d56a3067-0c68-4a86-a326-e82b639cba1d 结果摘要: Wrote handoff_response.md (4138 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 3, "neutral": 0, "value": 56.2, "events": [ { "title": "原油价格下行将因物流与保险费率滞后带来成品油裂解价差的阶段性走宽，下游需求呈航空煤油强弹性恢复与柴油结构性替代走弱的双轨分化格局。", "summary": "本报告探讨了Brent价格重心下移至82-92美元/桶对全球及亚洲炼厂裂解价差及下游需求的影响。地缘政治缓和将压缩原油风险溢价，但受扫雷周期和战时险（5.5%）滞后影响，成品油裂解价差具粘性，炼厂利润短期将走宽。下游航空煤油需求弹性高，低绝对价格将释放被压抑的国际航空运力；而柴油需求对价格不敏感，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源重心下移平抑中下游成本，驱动先进制造与传统重工业三季度利润呈“双轨大分化”；在中美倒挂约束下，人民银行将通过“存款挂牌利率下调10-15 bps”与“财政-货币协同流动性兜底（降…", "summary": "本报告探讨了Brent原油重心下移至87美元及中美利差倒挂于-273 bps对三季度中国工业利润与政策路径的影响。油价回落拉低PPI至-0.68%，将先进制造利润增速推升至16.5%，而重工业利润将收缩至-3.5%。总量降息受阻，PBOC将在三季度存款降息10-15 bps，并配合降准25-50 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源价格重心的下移提供了一次缓和全球“滞胀”压力的正向供给冲击，虽然成功消解了美欧央行的强行加息风险，但在低储蓄率、物流滞后与-273 bps中美巨幅倒挂的刚性约束下，全球主要央行在…", "summary": "本报告推演了Brent原油重心下调至82-92美元对三季度全球通胀与货币政策的边际影响。受汽油直接回落与航煤暴跌（裂解价差收窄至18-22美元/桶）间接拖累，三季度美国整体CPI与欧洲HICP将分别下修21 bps和30 bps。这有效移除了美联储年底加息选项，但受制于2.6%极低储蓄率与核心PC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 2, "neutral": 0, "value": 71.6, "events": [ { "title": "油价下降是欧洲化工和钢铁利润率的短期缓冲，而不是足以重置全球竞争力的结构性解药。", "summary": "截至 2026-05-31，本卡压力测试认为：油价下行能改善欧洲能源密集产业利润率，但只能部分对冲供应链重构和要素成本上升。Brent 从 $91.87/bbl 到 $82.5/bbl 中点相当于 -10.2%，但油和石油产品仅占 EU 工业终端能源 10.4%，油价完整传导也只约降低直接能源篮子…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由于避堵管道实际闲置运力极低（仅3.5-5.0百万桶/日）导致每日超1000万桶的原油供给面临物理性瘫痪，一旦全球库存耗尽将引发原油价格的非线性暴涨，强力支持布伦特油价在2026年下…", "summary": "本报告定量评估了霍尔木兹海峡事实性关闭（断供约1400万桶/日）对全球油价及通胀的冲击。由于中东避堵管道极限富余运力仅为3.5-5.0百万桶/日，全球仍有超1000万桶/日的结构性缺口。在极低的短期供给价格弹性（0.02-0.05）下，库存缓冲耗尽将引发油价非线性暴涨。我们预测2026年下半年Br…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "油价下行与制造业复苏共振下，应多配工业/增长敏感型资产（AUD、EUR、JPY），低配能源/避险敏感型资产（CAD、DXY），黄金战术规避回调但维持战略底仓。", "summary": "截至 2026-05-31，本卡基于 Brent 风险溢价压缩的前设，提出了跨资产配置调整方案：建议执行 Long AUD / Short CAD 跨品种套利，战术性下调 DXY 权重 3%-5% 以博弈欧元/日元贸易条件修复，并在实物黄金上执行“减短加长”策略（减仓 2-3% 回吐地缘溢价，保留…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 4, "neutral": 0, "value": 53, "events": [ { "title": "极端原油与汇率双重冲击将导致中国年化能源进口成本暴增2.02万亿元并系统性压缩工业利润率，成品油价格调控上限补贴在中央财政完全可控（由暴利税对冲），但沿海地方气电财政承受力较低，高股…", "summary": "截至2026-05-31，本卡对Brent>125美元和USD/CNY突破6.95的极端能源冲击开展压力测试。测算显示，中国年化能源进口将激增约2.02万亿元（占GDP的2.13%）；工业电力度电成本疏导使企业年化用电增支约5924亿元。130美元调控天花板下，135和145美元油价对应的中央财政…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大规模能源中断的可行路径是中东物理通行和产能冲击，而不是东欧单纯制裁；>4 mbpd门槛可达，但125美元/桶持续油价仍需更严格的持续性条件。", "summary": "截至2026-05-31，本卡对energy-analyst的>4 mbpd供应中断前提作压力测试并给出部分支持：中东霍尔木兹通行和海湾产能关停已经在官方5月资料中跨过该门槛，但东欧或俄罗斯单纯制裁路径概率低。基线判断是2026年6月至8月仍有55-65%概率出现阶段性>4 mbpd有效中断，但只…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹海峡地缘危机悬而未决的二元走向将油价、通胀与利率锁定在高位，给高杠杆航司（AAL）与消费型酒店（MAR）带来严重的绝对和相对风险暴露，建议平仓并重构组合至航空制造OEM（RT…", "summary": "本卡片从全球宏观视角对霍尔木兹海峡地缘危机进行了多情景划分（50%维持现状、30%降级、20%升级），并给出了WTI、航空煤油裂解价差、DXY和10Y美债收益率的中位数传导路径。压力测试表明，高油价与通胀环境对既定投资组合（AAL空/DAL多+MAR多+RTX多）造成显著冲击，强烈建议进行组合重构…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "随着油价回落挤出通胀溢价，大类资产配置的核心矛盾将在 4-8 周内由“防通胀”转向“防衰退”，应全面增配长久期利率债、黄金与防御型股票因子。", "summary": "基于能源分析师对 Brent 价格将在下半年轻微回落至 $80 轴的论断，资产配置逻辑须从“通胀冲击”（股债双杀）迅速切换至“需求毁灭/增长放缓”。前期的熊平压力将在 4-8 周内衰竭，策略建议：大幅拉长主权债久期，布局 2s10s 曲线牛陡；信贷市场减配高收益债；大宗商品剔除顺周期敞口并超配黄金…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 1, "risks": 3, "neutral": 0, "value": 68.4, "events": [ { "title": "在需求走弱背景下，120美元的油价无法被终端顺畅吸收，将导致石化中游毛利崩塌，并对电池材料和有色金属产生10%-20%的被动成本推升。", "summary": "截至2026-06-02，本卡对Brent原油升至110-120美元的情景进行了压力测试。由于全球需求收缩，高昂的石脑油成本无法顺畅传导，将导致亚洲非一体化裂解装置的乙烯-石脑油价差大幅跌破盈亏平衡点，使石化中游陷入深度亏损。此外，高油价将通过石油焦和氟化工原料大幅推升电池负极与隔膜/粘结剂成本，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹协议交易：油价下跳还是通胀尾部回归", "summary": "霍尔木兹协议交易：油价下跳还是通胀尾部回归 Hormuz deal trade: oil gap-down or inflation tail comeback 若特朗普暗示的协议在一周内兑现或落空，Brent的$80-85与$110-120两端情景应如何传导到能源股、航空航运、通胀预期和信用保护？", "sourceLabel": "研究主题", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "能源价格冲击引发的Stage-3美元流动性挤兑会使黄金与权益相关性转正至+0.55，在10交易日窗口内导致再平衡组合的10日99% VaR恶化至-7.80%（比基准差9 bps），使…", "summary": "本报告压力测试了在中东能源价格冲击下，黄金/资源资产相关性翻正对风险平价（RP）再平衡策略的破坏效应。Brent油价破$105/桶将触发Stage-3美元流动性挤兑，导致离岸美元荒，跨境掉期基差拓宽至-55 bps，促使机构无差别抛售黄金清偿。10日模拟显示，相关性翻正至+0.55时，再平衡组合的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "哑铃只在中位数路径上对冲；尾部因相关性翻正而失效，必须用季度142 bps的凸性对冲（NDX/XLE看跌+VIX看涨价差+跨资产腿）把99% ES从-21.5%收敛至-9.5%才能落…", "summary": "在Brent 120美元与AI Agent毛利归零的联合冲击下，'算力-资源'哑铃组合的对冲逻辑会失效：AI与能源相关性从-0.10翻正至+0.30~+0.40，1日99%参数VaR由2.92%扩张至6.06%；20日蒙特卡洛99% VaR -18.7%、ES -21.5%、最差路径-29.5%。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 73.8, "events": [ { "title": "能源不是衰退中的纯防御资产，但在2026供给短缺和油气现金流韧性下，大幅、笼统减能源比增长冲击本身更激进。", "summary": "截至2026-06-03，本卡压力测试 Research note 06 的大幅减能源建议：结论是方向上可削减商品 beta，但将能源在商品篮中从35%砍到18%过于激进。EIA/IEA显示2026年仍处供给短缺与高油价环境，Brent 4Q26仍预计$89/b；综合油气公司现金流、低杠杆、低再投…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 0, "value": 69.2, "events": [ { "title": "实体能源数据否定'制造业强周期'解读、并揭示通胀粘性部分是可消退的霍尔木兹油价溢价，因此抗滞胀杠铃应把能源仓位由原油beta轮向电力/电网/气转电、原油仅作霍尔木兹对冲。", "summary": "诚实指出 card-03 主题为污染的 agent 草稿，遂压力测试 Research note 1–2 的真实主线。能源视角结论：部分支持、重大修正。(A) 柴油货运需求疲弱、2026 工业用电仅+1.0%、Henry Hub约$3，实体量否定'强周期'，但印证 card1 就业48.6 的告诫…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在同质能源冲击与美联储鹰派停顿下，ECB的防御性加息与BoJ加息至1.00%将压缩G3名义利差，足以在3Q26对冲美元强势并维持EUR/USD（1.1400-1.1800）和USD/…", "summary": "若美联储于6月17日维持利率并上移点阵图，在同一能源冲击（原油价格在95美元附近波动）下，ECB与BoJ的被迫紧缩反应将重塑G3利差。ECB预计在6月11日加息25bp至2.25%，BoJ预计在6月16日加息25bp至1.00%以应对160附近的日元贬值与输入性通胀。利差压缩与政策对冲将防止欧/日…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 4, "neutral": 2, "value": 64.6, "events": [ { "title": "能源通胀已进入未来几个季度的长效定价模型，但尚未被证明为永久性通胀机制。", "summary": "截至 2026-06-05，油价在停火传闻下仍坚挺，说明市场已从单日地缘新闻溢价转向更长久的实体复航、保险、库存和央行反应函数定价。Brent 约 $95.24/bbl、WTI 约 $92.94/bbl；IEA 和 EIA 数据显示供给损失、库存去化和 Hormuz 商业开放不确定性仍支撑风险溢价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供给极端冲击对美联储‘按兵不动’预期的破坏力测试 报告日期 ：2026年06 05 分析师 ： energy analyst 立场 ：压力测试 (Stress test) 核心摘…", "summary": "能源供给极端冲击对美联储‘按兵不动’预期的破坏力测试 报告日期 ：2026年06 05 分析师 ： energy analyst 立场 ：压力测试 (Stress test) 核心摘要 截至2026年06月05日，随着霍尔木兹海峡局势在美伊博弈下持续处于高危状态，前一交易日（2026年06月04日…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DG/DLTR 配对的能源驱动是真实但脆弱的：单一地缘别针 + 高裂解利润同时压在桥墩上，未通过 6 个月持续性检验，应降级为 4–8 周战术配对并改用 RBOB 裂解看涨价差对冲。", "summary": "对 Research note 01 的'动力切换'护城河进行压力测试。当前 Brent ~$93.8、零售汽油 $4.43/加仑，但仅由霍尔木兹封锁这一单一地缘别针支撑。能源台主观概率：和解 45% / 现状 40% / 升级 15%；和解情景下 60 天内 Brent 回 $72–78、汽油跌…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面对霍尔木兹海峡商业复航不确定性引发的黏性通胀环境，大类资产配置应通过缩短名义久期、增配短端TIPS以及超配战略商品（油气/黄金）与低杠杆能源股，化解地缘溢价风险并获取绝对收益。", "summary": "截至2026年6月5日，中东地缘摩擦使油价高企（Brent 95.24美元/桶），通胀维持高位，利率更高更久（10年期美债收益率4.49%）。名义股债负相关性失效。建议战术低配长端名义债，超配短端TIPS，超配战略商品（能源与黄金）和能源股；外汇超配美元和商品货币，日元作为融资货币。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 2, "neutral": 0, "value": 72, "events": [ { "title": "能源价格突破110美元对医药中间体信用风险的边际影响", "summary": "截至2026-06-06，能源视角压力测试卡片08的信用结论。Brent>110美元确是真实的违约边际加速器，但仅在持续性条件下成立——季度均价连续≥2季站上约110美元、且与2026H2-2027H1到期墙重叠。约束变量是与到期墙重叠的油价季度均价，而非现货脉冲。油价经三条通道（油价联动原料、直…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "伊朗能源冲击对发达国家传导具有对称性但政策约束极度不对称：欧洲央行已被迫放弃6月降息，而日本面临严重的汇率贬值与输入性通胀双重打击，BoJ将被迫在6月提前缩表以捍卫汇率。", "summary": "在布伦特原油突破100美元的背景下，欧元区对进口能源的依赖导致其5月HICP能源分项同比飙升约14%，迫使欧洲央行在6月4日会议上放弃降息并选择“鹰派按兵不动”。对于日本，高油价与持续弱势的日元(USD/JPY >157)形成双重冲击，4月进口物价同比飙升12.5%。这种严峻的贸易条件恶化正将日本…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 5, "neutral": 1, "value": 58.4, "events": [ { "title": "霍尔木兹风险足以把能源冲击重新接到CPI/PPI和尾部波动率链条上，但当前最优动作仍是下调一档风险预算，而不是直接进入全面防御。", "summary": "截至2026-06-07，本卡对霍尔木兹风险进行能源与通胀传导压力测试：2025年霍尔木兹承运19.87 mb/d石油，替代原油出口路径仅3.5-5.5 mb/d，关闭式冲击下约14.4-16.4 mb/d难以改道。当前Brent周均价约$101/b，CPI能源和PPI能源已处高压状态。结论是支持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海湾能源交付分化的宏观传导：通胀冲击与贸易收支压力测试 报告日期： 2026 06 08 (亚太/新加坡) 分析师： global macro (全球宏观分析师) 立场： stres…", "summary": "海湾能源交付分化的宏观传导：通胀冲击与贸易收支压力测试 报告日期： 2026 06 08 (亚太/新加坡) 分析师： global macro (全球宏观分析师) 立场： stress test (压力测试：评估能源供应侧分化及宏观金融传导) 报告摘要 截至2026年6月8日，霍尔木兹海峡自爆发美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据一致性的提升确认了避险与物流溢价已见顶，霍尔木兹海峡封锁的实物供给断供风险已降温，从而支持而非改变8月后能源通胀压力消退的基准研判。", "summary": "在数据恢复机制（如SBRP协议）及实物“航运恢复协议”（HRS）建立后，霍尔木兹海峡的实物供应风险已转化为航运成本与物流摩擦。高频数据显示沙特东西管道（700万桶/日）与阿联酋哈布山-富查伊拉管道（150万桶/日）高负荷运转，叠加好望角绕行，使油价失去暴涨至120美元以上的物理基础。布伦特原油在9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Brent 突破 $111/b 不会引发中国全面通胀，但会造成 PPI 上下游严重的利润剪刀差，并掣肘央行宽货币空间，从而进一步支持下调战术风险预算。", "summary": "截至2026-06-07，本卡对全球能源通胀向中国传导进行压力测试。虽然中国以煤为主的能源结构能缓冲部分冲击，但72%的原油进口依存度意味着 Brent 突破 $111/b 将推高中游化工和上游采掘 PPI。因下游制造业竞争激烈，成本难以向核心 CPI 传导，将导致“上游涨价、下游利润挤压”的结构…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 1, "value": 83.8, "events": [ { "title": "在 Brent 原油维持在 95 美元以上与强势美元背景下，利润分配严重向具备高价格弹性和美元汇率正暴露的上游开采倾斜，而下游炼化板块则因发改委成品油定价扣减机制与进口汇兑损失而承压…", "summary": "基于 2026-06-08 权威工作日，评估 Brent 油价维持 95 美元以上预期下油气产业链的利润分配弹性及强势美元的影响。分析表明，上游开采具有高正弹性，中海油一季度成本仅 28.41 美元/桶 [S5]，在 95 美元油价下可斩获 66.59 美元/桶暴利 [自测算]，且直接受益于美元走…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "研究报告：输入型能源通胀背景下的中国宏观政策应对 日期： 2026 06 08 分析师： 中国宏观分析师 (china macro) 状态： 支持 (SUPPORT) 1. 执行摘要…", "summary": "研究报告：输入型能源通胀背景下的中国宏观政策应对 日期： 2026 06 08 分析师： 中国宏观分析师 (china macro) 状态： 支持 (SUPPORT) 1. 执行摘要 截至 2026 年 6 月 8 日，中国经济正面临 Brent 原油价格稳居 95 美元/桶以上与强势美元（美元兑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 2, "neutral": 0, "value": 76.2, "events": [ { "title": "中东停火信号下的油价风险应如何重定价", "summary": "晨报不应二选一，而应按时间维度与价格组成分开处理。当前~$90-95的Brent平台价已是已实现、具粘性的基准通胀冲击（4月CPI 3.8%、能源+17.9%、汽油+28.4%），美联储已转向2026年不降息；据此应上抬中枢CPI路径。而高于和平时期基本面(~$80)的~$12-16战争溢价是二元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在停火基准下，右尾对冲应“持有 carry、租用凸性”：能源股和 TIPS 是核心持仓，黄金和防御板块控制外溢风险，显性油价期权只用有限预算覆盖再封锁跳空。", "summary": "截至 2026-06-09，本卡综合并支持前两张卡：油价右尾应以分层组合对冲处理，而不是单边多油。能源股承担正 carry 的平台油价敞口，TIPS 承接已实现 CPI 传导，黄金提供地缘与货币保险，防御板块吸收需求毁灭时的衰退 beta，显性 Brent/WTI 看涨价差只覆盖跳空尾部。由于 O…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 1, "risks": 1, "neutral": 1, "value": 94.6, "events": [ { "title": "能源资产自由现金流足以支持一次性Payer Swaption权利金，但必须通过25%-30%仓位、总资本回报兑现和明确不自动滚动规则来实现。", "summary": "截至2026-06-10，本卡支持能源现金流作为“WN空头 + Payer Swaption + 短久期现金流锚”杠铃结构的资金腿，但只支持单轮3个月宏观对冲，而非无条件季度滚动。测算显示，若能源仓位为25%-30%，并能把回购与再平衡收益视为可兑现的经济收益，则可覆盖1.5% NAV的3M10Y…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] 油价波动压力测试", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: c02b06d9-70e1-4b73-a2db-20614f5290de 结果摘要: I will run the command to get the current date and working …", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 美以伊地缘和缓预期与API大幅累库共振下通胀预期及DXY走势研判", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 6ef92a18-22d0-47b3-98fe-55fbd43909f2 结果摘要: Wrote handoff_response.md (3288 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 0, "value": 85, "events": [ { "title": "夏季能源通胀反弹与电网基荷电力短缺构成了双重尾部风险，足以打破美联储耐心和AI资本品扩张的“双速经济”假设，多资产策略必须引入能源与公用事业对冲。", "summary": "截至2026-06-12，本报告对首席策略师的“双速经济”配置策略进行了能源维度的压力测试。指出夏季高温和活跃飓风季可能推高天然气与原油价格，引发整体CPI超预期反弹，打破美联储“保持耐心”的假设并引发股债双杀。同时，电网基荷电力短缺构成了AI资本品实质上电的物理硬约束，可能切断资本品扩张动能。建…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 1, "risks": 1, "neutral": 0, "value": 94.4, "events": [ { "title": "能源供应安全对原材料成本结构的传导机制", "summary": "能源供应安全是把地缘政治油价波动传入铝、钢成本的核心通道。它经两条轨道传导：燃料（Brent 6/11约$92-93、EIA预估6-7月均价$105、霍尔木兹受限超3个月、焦煤约$250/吨）与电力（欧盟TTF €50.26/MWh、同比+38%，经边际定价把日前电价推至€120-150/MWh；…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "10.03% 的能源权重在兼顾 VaR 合规的同时，配合金、债组合构建了极高效率的抗通胀堡垒，且物理减仓带来的向上凸性在极端通胀场景下优于带期权收益上限的大仓位。", "summary": "本报告评估了能源权重降至 10.03% 后的供应端冲击与通胀防御力。尽管面临 2026年6月10日 霍尔木兹海峡封锁升级引发的 Brent 原油快速跳升至 $108.50，由于组合协同配置了 15% TIPS 和 17.5% 黄金，整体通胀 Beta 仍高达 0.71。压力测试显示，在 Brent…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 5, "neutral": 1, "value": 66.8, "events": [ { "title": "霍尔木兹油价冲击与 Warsh 主席治下美联储的政策冻结：2026 下半年大类资产路径", "summary": "2026Q1 末美/以与伊朗冲突使霍尔木兹近乎中断，3 月全球原油供应骤减约 1,010 万桶/日，Brent 单月暴涨约 65%，5 月初约 $105。世行预计 2026 年能源价格同比 +24%；Rystad 基准下年底累计损失约 20 亿桶且含永久减损成分。新任美联储主席 Warsh 鹰派起…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在Brent仍有$105/b压力情景、Fed难以降息且BOJ可能触发日元套利平仓的组合下，风险平价应降美股Beta、转多日元、加黄金，并用有限权利金买原油凸性。", "summary": "截至2026-06-13，本卡支持前序“油价-日元-利率”风险链条，但建议采用分阶段风险平价调整：将美股Beta从0.55x NAV降至0.35x NAV，残余日元融资敞口改为+0.10x NAV多头日元，黄金升至0.12x NAV，Brent仅保留0.03x NAV期权权利金，并在Brent重返…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "油价再冲高会抵消美联储降息预期，但只有在霍尔木兹流量持续受限、Brent接近或高于$105/b并传导到汽油CPI时，才会触发剧烈的全球资产定价重构。", "summary": "截至2026-06-13，能源压力测试结论是：油价回落支持“有序重估”基准，但霍尔木兹物理流量尚未正常化，尾部仍足以改变全球资产定价。若Brent维持在$85-95/b，5月美国CPI的能源冲击大概率消退；若Brent贴近EIA假设的$105/b 6-7月均价，降息预期会被压制；若重回$120/b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年6月14日，中东能源危机导致布伦特原油维持在 105 美元/桶，中国作为原油进口依存度高达 73% 的国家，正面临显著的“输入型通胀”与制造业利润挤压。本研究通过对国家发改…", "summary": "2026年6月14日，中东能源危机导致布伦特原油维持在 105 美元/桶，中国作为原油进口依存度高达 73% 的国家，正面临显著的“输入型通胀”与制造业利润挤压。本研究通过对国家发改委成品油定价机制（80 130 美元区间利润压缩原则）的深度解析，揭示了 PPI CPI 剧烈背离下的行业生存法则。…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 6, "neutral": 0, "value": 59.2, "events": [ { "title": "汽油 YoY 大涨反映真实的原油结构性紧缩——OPEC+ 自律、页岩平台期、库存赤字与地缘溢价共同支撑夏季高峰前的供应短缺，验证 Fed 难以将能源冲击视为暂时性。", "summary": "支持 Research note 01 判断。汽油同比 +40.5% 是原油结构性紧平衡的下游印记：OPEC+ 5月较名义路径缺口约 1.1 mb/d，沙特 1.0 mb/d 自愿减产延至 Q3 末；EIA 将 2026 美国产量下修至 ~13.3 mb/d，油井钻机同比 -12%；美国商业原油较…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源板块分子端在Brent中枢$65以上即可兑现card-06的200–250bps红利定价红线，真正的下限不是分红政策本身而是Brent长期中枢，临界值$60/桶。", "summary": "在Brent $70/桶以上且OPEC+维持配额纪律的基础情景下，全球油气一体化巨头（XOM/CVX/Shell/TTE/中海油/中石化）现金股息覆盖率>1.6x，现金股息盈亏平衡油价仅$40–45/桶，足以兑现card-06提出的红利股债利差200–250bps红线，能源板块20%重估空间成立。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "原油价格风险情景及其对二次通胀路径的影响分析", "summary": "对card-01的尾部前提做压力测试。Brent现货2026-06-12报$87.33/桶，EIA极端情景假设霍尔木兹关闭使6-7月均价$105/桶、2027年回落至$79/桶。套用美联储传导弹性（油价涨10%→headline约0.4%、核心约0.1%、历时8季），测算$100情景核心CPI仅+…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "下半年95美元/桶的Brent油价与4美元/加仑以上的汽油价格将通过交运物流和二次预期传导，拉高核心通胀并阻碍通胀降温，促使美联储在6月会议将年内降息预期清零，推动点阵图显著转鹰。", "summary": "本卡综合评估了2026年下半年Brent原油（95.06美元/桶）与美国零售汽柴油价格对通胀传导及美联储反应函数的影响。测算表明，油价上行将通过交运物流和工业原料渠道向核心PCE传导约20BP，并使1年期通胀预期重回3.2%-3.5%的高位，增加二次通胀风险。新任主席Warsh的首秀将展露鹰派立场…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 1, "neutral": 0, "value": 90.6, "events": [ { "title": "能源扰动已从油价冲击升级为跨商品通胀冲击，因此商品超配仍是higher-for-longer环境下的核心防御仓位。", "summary": "截至2026-06-15，本卡片支持上一张卡片的商品超配结论。能源冲击通过五条路径传导：原油和成品油推高CPI，柴油和汽油推高PPI，霍尔木兹风险外溢到LNG和电力，油气成本进入化肥和金属，OPEC+谨慎增产延长库存稀缺。即使6月15日和平消息压低油价，库存和政策证据仍显示通胀粘性未解除。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 1, "neutral": 0, "value": 92, "events": [ { "title": "6月19日执行失败会把原油风险从地缘新闻转为库存日历，上游A股的盈利上修应先于库存正式击穿启动。", "summary": "截至2026-06-16，本卡支持上一卡的上游能源对冲方向，但排序应为600938优先、601857其次、600028偏交易性。若6月19日MoU执行受阻，EIA的OECD库存23亿桶/50天覆盖阈值可能在55-79天内被市场提前定价，中枢约2026年8月24日，机械上限约2026年9月7日。Br…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 2, "neutral": 2, "value": 93.4, "events": [ { "title": "[auto-handoff] 澄清原油价格数据分歧并调整宏观通胀传导路径", "summary": "致 日报总编： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: f846db4c-3154-4feb-a5c0-423522062fd7 结果摘要: Wrote handoff_response.md (7369 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 澄清原油价格数据分歧并调整宏观通胀传导路径", "summary": "致 市场情绪师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: a33b6d18-83e7-4fc4-a82b-34253252876f 结果摘要: Wrote handoff_response.md (11319 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 美联储鹰派态度及美债收益率上升对全球原油需求和美元定价机制的影响", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: ab557370-aa79-4339-b09e-e6b6969a831f 结果摘要: Wrote handoff_response.md (6707 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "上游能源红利逻辑正被强制切换为'现金流稳定'主导，但估值中枢不能维持当前水平——煤炭可作过渡仓但需-10%中枢下移，油气从核心降级为卫星仓，真正受益者是公用事业型物理资产。", "summary": "油价回落至70-80美元区间剥离了战争溢价，IEA下调2026年全球需求增速至+0.7 mb/d。上游红利逻辑被强制切换为'现金流稳定'主导，但估值中枢不能简单平移：A股三桶油/煤炭龙头股息率虽仍6-8%，盈利-分红覆盖倍数从2022年~2.0×压缩至~1.3×。压力测试显示韧性排序为煤炭>油气>…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 3, "neutral": 0, "value": 82.8, "events": [ { "title": "非OPEC+无法物理补上霍尔木兹缺口，但绕行物流、战略库存、需求毁灭与已被定价的重开把布伦特压回约80美元——结构性溢价真实但已从平价迁移到波动率、期限结构与能源安全资本开支；长周期…", "summary": "压力测试 Research note 01 的\"单边结构性原油溢价\"。实物上非OPEC+补不上缺口：未来3个月增量仅50万–100万桶/日，对应1300万–1500万桶/日缺口的5%–7%；可动用闲置产能150万–250万桶/日且多位于霍尔木兹以内。但布伦特已从约113美元高点跌至约80美元（−2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国能源CPI飙升及其对中国能源进口成本与电力供应稳定的潜在传导影响", "summary": "支持鹰派Fed/强美元主线并延伸至能源。美国5月能源CPI同比+23.5%、整体4.2%是滞后读数，锚定2月霍尔木兹冲突尖峰；现货Brent已在6月回吐约23%至约80.6美元，故对中国前瞻进口成本压力在边际缓和、而非加速。具约束力的传导是价格通道（美元油气）与汇率通道（约7.30人民币、强DXY…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-20：美联储6月鹰派暂停与全球实际利率再定价", "summary": "截至2026-06-20，6月FOMC更像一次实际利率收紧冲击，而非中性暂停。美联储维持3.50%-3.75%目标区间，但SEP把2026年PCE通胀和联邦基金利率路径同步上调；H.15显示2年期、10年期和10年期TIPS收益率在会议日均上行。能源冲击仍是关键变量，EIA预计Brent在6月和7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 1, "risks": 8, "neutral": 0, "value": 62.2, "events": [ { "title": "AI基础设施的真正能源约束是确定性电力和电网capex，而不是单一Brent价格；油价上行会放大成本与进度风险，但更可能造成扩建分化而非需求坍塌。", "summary": "截至2026-06-21，本卡压力测试Research note 5结论：原油赤字和Brent上行风险会约束AI算力扩建，但主要不是通过原油直接消耗，而是通过天然气边际电价、柴油/成品油施工和备用电源成本、电网并网排队以及自备燃气电源capex。1 GW IT负载、PUE 1.10的园区年用电约9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Brent跌破80美元/桶削弱了能源冲击的尾部风险 [S5]，但在库存和成品油缓冲修复前，美联储仍会把能源通胀视为延后降息而非立即宽松的理由 [S8][S9][S10]。", "summary": "截至2026-06-22，本卡支持上一张global-macro结论：Brent跌破80美元/桶是油价风险缓和，但不是实体供应链已经完全正常化 [S5]。EIA仍估计2Q26全球库存下降630万桶/日、3Q26下降760万桶/日 [S1]，IEA仍预计2026年全球供应下降390万桶/日 [S4]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中东地缘政治缓和背景下能源价格的底部支撑力与需求衰退风险", "summary": "油价从126美元回撤至80美元，表面像地缘溢价出清，实则被需求毁灭覆写。IEA/EIA将2026年全球原油需求由增长翻为收缩（−1.1 mb/d，二季度交付−5 mb/d），明确归因高燃料价（汽油2.98、柴油3.40美元/加仑）与紧缩。溢价并未真正出清而是潜伏：阿联酋5月退出后OPEC+闲置产能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供应冲击不会把再度油价飙升设为基准情形，但霍尔木兹、LNG、化肥和库存的物理约束足以让2026年底通胀右尾保持肥厚。", "summary": "截至2026-06-21，本卡压力测试认为：美伊缓和降低了油价即时上限，但能源供应端仍能推翻过于乐观的通胀预期。EIA基准为Brent在2026年6-7月约105美元/桶、四季度89美元/桶、2027年79美元/桶；这意味着冲击会缓慢消退而非干净消退。霍尔木兹2024年正常流量约2,000万桶/日…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 1, "risks": 0, "neutral": 0, "value": 96, "events": [ { "title": "布伦特高于$120/bbl若持续，将通过亚洲进口账单、大西洋再路由和LNG竞争强化能源安全溢价，但其自我抑制性意味着真正持久的投资信号是电力系统、能效和多元化供应CapEx，而非无纪…", "summary": "截至2026-06-22，本卡支持霍尔木兹冲击已成为宏观级能源供给冲击的判断，但限定布伦特持续高于$120/bbl是高压情景而非官方基准。霍尔木兹在2024年承载约20 million b/d石油流量和约20%全球LNG贸易，亚洲敞口最大。若高油价延续，贸易流向会转向大西洋盆地和美国成品油，LNG…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 5, "neutral": 0, "value": 75, "events": [ { "title": "2026/27冬季能源冲击足以推高风险溢价和通胀黏性，但除非咽喉要道扰动延续到2027年，否则更像数个季度的交付溢价，而不是多年结构性供给短缺。", "summary": "截至2026-06-23，本卡对2026/27冬季能源供应弹性进行压力测试：LNG和原油的瓶颈主要在可交付灵活性，而非长期分子总量。霍尔木兹暴露接近全球LNG贸易20%和近20 mb/d油品出口，足以维持冬季价格溢价和高波动；但LNG已进入2025-2030年约345 bcm/yr的投产波，石油2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "我已完成对布伦特原油价格下跌至 $77/桶区间后，上游开采企业利润空间的压力测试与敏感性分析，并对 Research note 01 提出的“低配资源股”观点进行了修正。 主要工作总…", "summary": "我已完成对布伦特原油价格下跌至 $77/桶区间后，上游开采企业利润空间的压力测试与敏感性分析，并对 Research note 01 提出的“低配资源股”观点进行了修正。 主要工作总结 1. 上游利润空间压力测试 ： 中国海油 (0883.HK) ：Q1 2026 桶油主要成本仅为 $28.41/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹风险已经足以支持体制切换警报，但当前证据指向“有条件的紧缩风险”而非“油价单向失控”。", "summary": "截至2026-06-23，能源证据支持上一张卡片的风险判断但需加限定：霍尔木兹冲击的实物流与成品油价格通道已足以削弱宽松交易，EIA/IEA显示近端供应压力和库存消耗仍未解除；但Brent现货已回落至约$77/b，说明市场没有确认单向油价失控。结论是“紧缩风险已活跃并应纳入组合约束”，而不是“油价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "未来4-8周应以现金和短中久期防守、降低美国权益beta、保留适度商品对冲，而不是押注快速降息或油价单边失控。", "summary": "截至2026-06-23，本卡综合卡片01的鹰派暂停判断，建议未来4-8周采用防御但非衰退式配置：现金/T-bills 30%、美国权益35%、久期20%、大宗商品15%。理由是Fed维持3.50%-3.75%并给出3.8%的2026年末利率中位数，5月CPI和能源通胀仍高，而短端美债收益率已有配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 2, "risks": 0, "neutral": 0, "value": 96, "events": [ { "title": "霍尔木兹冲击的关键不是单日油价，而是持续时间、库存下降和成品油/LNG/电力链条能否把短期能源涨价转化为更持久的通胀预期。", "summary": "截至 2026-06-25，本卡支持上一张白板卡：霍尔木兹海峡中断是成本推动型能源通胀冲击。核心证据是 2024 年海峡承载 20 million b/d 石油、约全球石油液体消费 20%，并承载约五分之一全球 LNG 贸易；可用绕行能力仅约 2.6 million b/d，留下约 17.4 mi…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "能源资产在通胀再加速路径下的对冲有效性分析", "summary": "本卡支持Research note 1-2的高利率持久化与能源作为类期权对冲的处方。2026年通胀本质是能源故事：能源仅占CPI篮子7.5%，却贡献5月环比涨幅近60%，汽油同比+40.5%，而核心反而降温。能源因此对能源主导的整体通胀对冲有效（高凸性、即时传导），却对3.3%核心PCE无效。但其…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 0, "neutral": 1, "value": 96, "events": [ { "title": "做多迁移的稀缺租金应以电网硬件和材料为锚、以先进封装和定制ASIC为弹性，同时用能源空头、商用GPU相对空头和暗硅期权保护组合。", "summary": "截至2026-06-25，本卡综合前序链条：霍尔木兹能源溢价正在消退，稀缺租金迁移到铜/GOES、电网设备、先进封装和定制ASIC。建议用100.0名义多头配置15.0%材料、30.0%电网硬件、20.0%先进封装、15.0%定制ASIC/AI网络、10.0%电力接入和10.0%流动性，同时用35…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-26", "supports": 1, "risks": 9, "neutral": 1, "value": 66.2, "events": [ { "title": "[auto-handoff] 统一今日油价引用基准", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: c10ed69f-8e17-495d-a490-77cf57ca559f 结果摘要: Wrote handoff_response.md (9099 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "能源价格波动空间足以支撑当前通胀对冲，但应把能源视为带再平衡纪律的尾部风险工具，并用黄金分散油价回落后的政策失误与实际利率风险。", "summary": "截至2026-06-27，支持上一卡能源+黄金通胀对冲，但强调其是波动率与尾部风险对冲，而非单边油价牛市。Brent在2026-06-26报73.60美元/桶，较52周高点明显回落；IEA仍显示2026年供应下降3.9 mb/d、库存快速消耗，霍尔木兹风险与低库存保留上行凸性。同时EV替代和202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹海峡能源瓶颈对全球原油供应及能源转型成本的影响", "summary": "支持 Research note 01 关于 2026 霍尔木兹冲击为上半年决定性宏观事件的判断，并延伸至能源结构。霍尔木兹承载约 2,000 万桶/日（全球石油消费约20%、海运原油约34%）[S1]；3月初海峡关闭令布伦特飙至118–120美元、现货高见约150美元[S6]，截至6月26日已回…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通胀粘性阻碍美联储降息的背景下，纸面原油抛售与实物供给紧张的冲突使得油价下行有底而上行具备高凸性特征，使其在资产相关性转正的环境下成为有效的非对称风险对冲工具。", "summary": "截至2026-06-27，本卡支持能源作为组合凸性对冲工具的逻辑。美伊备忘录虽引发纸面油价大跌（WTI至70.24美元/桶），但扫雷瓶颈（60日延迟）与伊朗原油出口缓慢恢复使物理供给仍然偏紧，且美原油去库1553万桶、SPR降至历史新低3.492亿桶。高油价向核心PCE的二阶传导存在6-12个月时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 5, "neutral": 0, "value": 80.6, "events": [ { "title": "地缘价格冲击通过霍尔木兹海峡受阻和灰色市场运费摩擦（每桶4.5美元）改变原油供应路径并拉动首轮通胀，尽管油价已降温至72.95美元，但政策反应焦点已转移至长期通胀预期锚定上，对资产配…", "summary": "分析表明，2026年5月中东地缘冲突导致霍尔木兹海峡每日1100万-1400万桶原油供应受阻（占全球约15-20%），叠加全球5月产量同比下降3.9%（至9450万桶/日），推动Brent油价飙升至120美元/桶。虽6月油价回落至72.95美元，但滞后影响仍存，5月美CPI升至4.2%（能源通胀涨…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "尽管地缘缓和推动原油价格回落，但原油底部防守、AI负荷拉动天然气结构性上涨以及核心CPI的传导滞后，共同决定了能源下行无法抵消核心通胀的再加速风险，美联储政策中枢仍将维持更高更久。", "summary": "地缘风险缓和使原油价格暴跌至$78–$80/桶，但这不足以抵消核心通胀再加速风险。原油实物库存处于历史低位，且受OPEC+在WTI $75防守底线的制约；而AI数据中心表后气电需求推动天然气价格结构性上涨至$3.92/MMBtu以上，构成更持久的通胀源。核心通胀受黏性服务与薪资驱动，直接能耗传导较…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然备忘录缓解了短期实物原油压力，但霍尔木兹海峡的结构性脆弱与15%-25%的二次通胀概率支撑将大宗商品战术超配至20%，并应采用高分红个股现货替代及低配看涨期权策略进行防御。", "summary": "截至2026-06-27，评估表明尽管扫雷进展与伊朗出口正常化将推动短期WTI均价回落至67.95美元且跌破70美元概率升至62.85%，但15-25%的能源价格二次上尾再脉冲风险及长期地缘瓶颈支持将大宗商品配置超配至20%。在A股通道中，因159945等ETF极度缺乏流动性（成交仅106万元），…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Brent跌至约72美元后，Warsh美联储的高利率前提应从“结构性再加息”下修为“数据条件化高位平台”：基准终端利率上沿3.75%-4.00%，降息需等待核心PCE和预期共同确认。", "summary": "2026-06-27：本卡片压力测试“长期高利率”前提。结论是油价回落削弱了单凭能源再加息的逻辑，但尚不足以触发降息定价；6月FOMC维持3.50%-3.75%，SEP年末中位数3.8%，核心PCE仍2.9%，长期通胀预期3.6%。基准为55%概率高位平台，30%概率油价/PCE触发转鸽，15%概…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 2, "neutral": 1, "value": 96, "events": [ { "title": "能源商品成本传导与USD/CNY汇率风险的压力测试", "summary": "截至2026-06-28，本卡对卡3'不上调外汇对冲'结论做压力测试。被测机制（人民币贬值抬高能源进口成本、压低通胀对冲实际人民币收益）真实但属二阶，且当前被相反力量压过：能源对冲人民币收益的主导因素是美元油价中伊朗战争地缘溢价的崩塌，而非汇率。人民币油价成本=布伦特(美元)×USD/CNY，3月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核心服务与住房通胀的粘性：美联储政策的后续驱动力", "summary": "能源转为通缩之际，核心服务与住房粘性足以支撑美联储鹰派、抵消能源通缩。关键在指标分化：整体PCE 4.1%/整体CPI 4.2%将因油价基数效应回落，但美联储盯住的核心PCE升至3.4%（2023年10月以来最高），由占核心PCE一半以上、运行约3.5%-3.9%的超级核心（剔除住房的核心服务）驱…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在风险平价框架下，能源减配并转向黄金/结构性金属不会突破波动约束，反而降低能源方差集中度；代价是牺牲部分纯油价冲击下的上行夏普。", "summary": "截至2026-06-28，本卡支持将能源对冲重心从USD/CNY与原油转向黄金和结构性金属。建议维持35/25/25/15股/债/对冲/现金框架，但把25%通胀对冲仓位重配为13%黄金、8%结构性金属、4%能源尾部。按六资产协方差自测算，组合波动率从10.30%降至9.15%，能源方差贡献从31.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 1, "risks": 3, "neutral": 1, "value": 96, "events": [ { "title": "2026年石油供应中断对全球能源品种及炼油产能的冲击评估", "summary": "支持卡片01“油价持久性是枢纽”的判断并加以锐化。霍尔木兹2025年日均输送约2000万桶/日（原油1500万+成品油550万），全球闲置产能几乎全在海峡之内，绕行管道仅7-8mb/d（不足正常一半），规模化对冲不存在。冲击非线性传导：3月约1200万桶/日无法运抵亚洲、其中800万为中硫油，亚洲…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美伊卡塔尔复谈促油价地缘溢价退潮与跨资产红利再平衡", "summary": "美伊卡塔尔复谈促油价地缘溢价退潮与跨资产红利再平衡 Geopolitical De-escalation & Crude Oil Re-pricing 美伊复谈促Brent油价跌至72.27美元，地缘防范溢价出清，如何重新评估资源品板块（石油、煤炭）与公用红利、出口中盘红利之间的配置权重？", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "summary": "对中国，2026年油价/柴油冲击既是被缓冲的进口税、又是战略加速器，且净额上比卡片01–02的美欧日集团损害更小。三大事实：（1）进口税真实但落在PPI而非CPI——5月PPI同比+3.9%（2022年7月来最快）、CPI仅+1.2%，疲弱内需把冲击吸收为生产者利润压缩；（2）进口税被三重缓冲——…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-30", "supports": 0, "risks": 0, "neutral": 1, "value": 96, "events": [ { "title": "[auto-handoff] 低油价对中国工业企业利润及进口贸易顺差的边际影响分析", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: f2e0bdfb-0125-4d0a-b62f-80a7e18e510b 结果摘要: Wrote handoff_response.md (7787 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] } ], "href": "/investor/theses/thesis-3d421524cd/", "networkHref": "/investor/research-network/#thesis-3d421524cd" }, { "id": "thesis-cb2ab35c55", "source": "tracked_thesis", "title": "AI 基础设施扩张从规模驱动切换到 TCO、能效和利用率筛选", "titleEn": "AI infrastructure expansion is shifting from scale to TCO, power efficiency, and utilization screening", "lane": "AI 基建 TCO/能效筛选", "laneEn": "AI infrastructure TCO and efficiency reset", "laneHref": "/investor/lanes/ai-infrastructure-tco-and-efficiency-reset/", "investableFocus": "AI 基建 TCO/能效筛选：SMCI, DELL, HPE, VRT, ANET, NVDA, AMD, TSM", "stockUniverse": [ { "ticker": "SMCI", "name": "Super Micro Computer", "market": "US", "role": "AI 服务器与液冷整机", "roleEn": "AI 服务器与液冷整机", "bucket": "core", "href": "/investor/stocks/SMCI/" }, { "ticker": "DELL", "name": "Dell Technologies", "market": "US", "role": "AI 服务器交付", "roleEn": "AI 服务器交付", "bucket": "core", "href": "/investor/stocks/DELL/" }, { "ticker": "HPE", "name": "Hewlett Packard Enterprise", "market": "US", "role": "企业服务器与网络", "roleEn": "企业服务器与网络", "bucket": "core", "href": "/investor/stocks/HPE/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "AMD", "name": "AMD", "market": "US", "role": "加速器替代供给", "roleEn": "加速器替代供给", "bucket": "core", "href": "/investor/stocks/AMD/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "SMCI", "DELL", "HPE", "VRT", "ANET", "NVDA", "AMD", "TSM", "AVGO", "MU", "SMH", "300308.SZ", "APO", "IEF" ], "coreView": "赛道：AI 基建 TCO/能效筛选。候选观察标的：SMCI, DELL, HPE, VRT, ANET, NVDA, AMD, TSM。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 78, "eventCount": 160, "firstSeen": "2026-04-24", "lastSeen": "2026-06-29", "supports": 9, "risks": 138, "neutral": 13, "analystCount": 9, "sourceTypeCounts": { "研究线程": 3, "研究笔记": 151, "深度研究": 1, "研究主题": 4, "分析协作": 1 }, "topTerms": [ "AI 基础设施", "NVIDIA", "HBM", "GPU", "ASIC", "TSMC", "AIDC", "电力", "液冷", "企业", "infrastructure", "expansion" ], "whatChanged": [ "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。", "2026-06-29: AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "2026-06-29: AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "2026-06-28: 半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "2026-06-28: GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。" ], "practical": { "sectorLane": "AI 基建 TCO/能效筛选", "sectorLaneEn": "AI infrastructure TCO and efficiency reset", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "SMCI", "name": "Super Micro Computer", "market": "US", "role": "AI 服务器与液冷整机", "roleEn": "AI 服务器与液冷整机", "bucket": "core", "href": "/investor/stocks/SMCI/" }, { "ticker": "DELL", "name": "Dell Technologies", "market": "US", "role": "AI 服务器交付", "roleEn": "AI 服务器交付", "bucket": "core", "href": "/investor/stocks/DELL/" }, { "ticker": "HPE", "name": "Hewlett Packard Enterprise", "market": "US", "role": "企业服务器与网络", "roleEn": "企业服务器与网络", "bucket": "core", "href": "/investor/stocks/HPE/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "AMD", "name": "AMD", "market": "US", "role": "加速器替代供给", "roleEn": "加速器替代供给", "bucket": "core", "href": "/investor/stocks/AMD/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "SMCI", "DELL", "HPE", "VRT", "ANET", "NVDA", "AMD", "TSM", "AVGO", "MU", "SMH", "300308.SZ", "APO", "IEF" ], "primaryStocks": [ { "ticker": "SMCI", "name": "Super Micro Computer", "market": "US", "role": "AI 服务器与液冷整机", "roleEn": "AI 服务器与液冷整机", "bucket": "core", "href": "/investor/stocks/SMCI/" }, { "ticker": "DELL", "name": "Dell Technologies", "market": "US", "role": "AI 服务器交付", "roleEn": "AI 服务器交付", "bucket": "core", "href": "/investor/stocks/DELL/" }, { "ticker": "HPE", "name": "Hewlett Packard Enterprise", "market": "US", "role": "企业服务器与网络", "roleEn": "企业服务器与网络", "bucket": "core", "href": "/investor/stocks/HPE/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" } ], "instrumentTags": [ "SMCI", "Super Micro Computer", "DELL", "Dell Technologies", "HPE", "Hewlett Packard Enterprise", "VRT", "Vertiv", "ANET", "Arista Networks", "NVDA", "NVIDIA" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：证据密度足够", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 118 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 9, "recentEvents": 118, "supportRatio": 0.06, "riskRatio": 0.86, "conflictScore": 96.6, "evidenceDepth": 100 }, "score": 70.3 }, "events": [ { "date": "2026-06-29", "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,800万美元（约降33%）[S1][自测算]；先进封装（CoWo…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "HBM", "电力", "液冷", "企业", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck from compute to power, networkin…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 28, "matchedTerms": [ "AI 基础设施", "GPU", "电力", "infrastructure", "shifting", "power", "server" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "HBM", "GPU", "电力", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗ASIC。先进封装成为泄压阀，但TSMC已提示先进封装产能紧张…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "HBM", "ASIC", "TSMC", "电力", "液冷" ], "href": "" }, { "date": "2026-06-28", "title": "国产大模型并未告别规模竞赛，而是在先进GPU受限的条件下把竞争单位改写为每瓦、每token、每受限加速器和每留存用户的有效智能产出。", "summary": "截至2026-06-28，本卡对“国产大模型不可逆转向效能极限竞赛”进行压力测试。结论是：效能迁移真实且受GPU出口、H20许可、HBM/互连、液冷机柜和电力约束推动，但不是规模竞赛终结，而是进入“稀缺规模竞赛”。DeepSeek-V3、DeepSeek-R1、Qwen3和MiniMax-M1显示国产模型正通过MoE、FP8、RL、蒸馏、长上下文和推理预算…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "HBM", "GPU", "电力", "液冷" ], "href": "" }, { "date": "2026-06-28", "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计约6900-7200亿美元（自测算），NVIDIA FY20…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 25.5, "matchedTerms": [ "NVIDIA", "HBM", "电力", "液冷", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI算力基础设施需求与海底数据中心技术演进", "summary": "支持卡片01关于海兰信（300065.SZ）的判断并加以收紧：UDC卡位2026年AI建设的电力与散热两大硬约束。2026年超大规模云资本开支约7250亿美元，全球数据中心用电将破1000TWh（超日本），单柜功率向370–600kW飙升，高盛预计年底76%的AI服务器需液冷。海兰信是国内唯一掌握UDC全套技术、全球继微软后第二家的企业：海南陵水675PF…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "电力", "液冷", "企业", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "效率驱动的算力架构转型：从“暴力规模”到“高能效密度”的物理基座适配性评估", "summary": "压力测试 Research note 08 的隐含前提：效率并不缓解物理天花板，而是把它从电网层下移到机架与机柜行层并加剧。机理是机架尺度的杰文斯悖论——单瓦产出靠提高密度实现，机架功率从约40kW(Hopper)→120kW(Blackwell)→142kW(GB300,90%入液)→600kW(Kyber,2027)迈向1MW，对照2026均值约27k…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "HBM", "GPU", "液冷", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM合约价Q2环比+58–63%)、以及电网电力(美国并网排队约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "HBM", "GPU", "电力", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,600 亿）以经营现金流支撑，资金基础/地域/标的均不同于承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22.25, "matchedTerms": [ "AI 基础设施", "GPU", "电力", "infrastructure", "higher" ], "href": "" }, { "date": "2026-06-26", "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–560 亿美元；中国云(字节/阿里/腾讯)与国产替代构成第二引…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.25, "matchedTerms": [ "AI 基础设施", "NVIDIA", "TSMC", "电力", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "截至2026-06-25，AI硬件需求的下一核心约束不是芯片是否能生产，而是足够多的兆瓦能否被许可、接入、冷却并按期商业化。", "summary": "截至2026-06-25，本卡支持上一张卡的AI硬件稀缺性主线，但将瓶颈从GPU/HBM进一步上移到可通电数据中心、电网接入、冷却和电气设备。IEA、Berkeley Lab、EPRI、CBRE和JLL数据共同显示，AI需求仍强，但项目能否转化为创收产能取决于电力和并网速度。结论是：稀缺性继续支撑订单可验证的AI硬件和电气/液冷/供电链条，但宽基科技bet…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "HBM", "GPU", "电力", "液冷", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "电力稀缺不会杀死AI芯片周期，但会把竞争重心从单纯峰值算力转向固定兆瓦内的有效吞吐和能效，从而收窄可投资赢家。", "summary": "截至2026-06-25，本卡对电力和并网约束做半导体侧压力测试：约束不会终结AI芯片需求，但会把胜负指标从峰值FLOPS推向每兆瓦吞吐、每机柜token、HBM每瓦带宽和部署可用性。NVIDIA的机柜级平台、Google TPU、AWS Trainium、TSMC背面供电和Micron HBM4都显示产业已转向能效优化。结论是保留AI存储、高端互联、先进…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26, "matchedTerms": [ "NVIDIA", "HBM", "TSMC", "电力" ], "href": "" }, { "date": "2026-06-25", "title": "电力设备瓶颈不会系统性消灭AI加速器需求，但会把2026-2028年约15%-25%的GPU/HBM/ASIC部署变成收入确认和爬坡时点风险。", "summary": "截至2026-06-25，本卡从半导体侧压力测试电力设备瓶颈传导：2026-2028年加速器等效部署中约15%-25%面临2-4个季度时点风险，真实终端需求毁灭约3%-7%。商用GPU最直接受通电MW影响，HBM因多年协议和稀缺配额更像库存节奏风险，自研ASIC既受益于每瓦token优化，也因客户和站点专属性承担20%-35%新爬坡时点风险。结论是需求没有…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "HBM", "GPU", "ASIC", "电力" ], "href": "" }, { "date": "2026-06-25", "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出保护的公用事业更偏时点风险资产。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "HBM", "GPU", "ASIC", "电力" ], "href": "" }, { "date": "2026-06-25", "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实触发重排，但方向是向支出倾斜，且买方自身点名的硬约束是电力与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.25, "matchedTerms": [ "HBM", "ASIC", "电力", "infrastructure", "scale" ], "href": "" } ], "timeline": [ { "date": "2026-04-24", "supports": 0, "risks": 1, "neutral": 1, "value": 39, "events": [ { "title": "2026年AI算力供应链瓶颈迁移：从CoWoS封装紧缺到HBM4配额与电力约束", "summary": "2026年AI算力供应链瓶颈迁移：从CoWoS封装紧缺到HBM4配额与电力约束 在SK Hynix已卖空2026年HBM产能、TSMC把CoWoS从约35K扩至130K wpm、Nvidia Vera Rubin下半年放量的背景下，AI硬件供应链的真瓶颈究竟已迁移到哪里？二级市场应如何重新定价上游…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年2.5D/3D先进封装（CoWoS）产能扩张与HBM4演进：评估次世代AI加速器供应链的稳定性", "summary": "2026年2.5D/3D先进封装（CoWoS）产能扩张与HBM4演进：评估次世代AI加速器供应链的稳定性 随着TSMC CoWoS产能持续紧张且HBM4于2026年开始量产，从HBM3e向HBM4的跨越将如何影响主要AI GPU厂商（如NVIDIA/AMD）的成本结构和交付周期？哪些二级OSAT厂…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "AI算力行情是超级周期中段，三波叠加需求（训练→推理→主权AI）驱动力完整，供给瓶颈维持定价权，短期情绪风险真实但不构成周期终结信号。", "summary": "AI硬件行情处于超级周期中段，而非利好出尽的顶部。三波需求（前沿模型训练、大规模推理、主权AI/企业部署）正叠加复利，供给侧CoWoS封装、HBM4产能及电力基础设施瓶颈在12–18个月内无法解除，定价权保持完整。历史类比指向1997–1998年的互联网基础设施阶段。战术层面，情绪高涨（恐惧贪婪指…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Q1 2026 财报后的云巨头 AI capex：需求验证还是瓶颈预警？", "summary": "本卡选择 tmt-analyst 评估 Q1 2026 云巨头 AI capex 周期。Microsoft、Alphabet、Amazon 和 Meta 均显示 AI/云收入加速，但 capex 和 PP&E 支出同步大幅上行。TSMC 的 61% HPC 收入占比、74% 先进制程占比和 66.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "2026下半年AI资本开支拐点辩论：消化期还是再加速？", "summary": "TMT分析师判定，2026下半年AI资本开支并未进入消化期，而是进入结构轮动期：总量持平至上行，边际增量美元从训练优化的商用GPU集群转向推理fleet、超大规模云厂商自研芯片、以及电力与场地准备。空头叙事在需求侧（推理token volume、agentic渗透、多模态）证据不足，但在前沿训练边…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 0, "neutral": 0, "value": 54.2, "events": [ { "title": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯…", "summary": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯片竞争格局：NVIDIA Rubin对阵云巨头ASIC 本卡主题： WFE资本开支周期——先进封装与HBM设备订单可…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 0, "neutral": 1, "value": 50.6, "events": [ { "title": "材技一体化对fabless是选择性降权——AI加速器与EDA/IP溢价稳健，射频/功率/模拟/成熟节点向IDM收敛，资金应在设计层内向两个可防御极轮动。", "summary": "材技一体化情景下，fabless估值溢价并非整体崩塌而是结构性分化。AI加速器龙头（NVIDIA、AMD MI、Broadcom ASIC）凭借云厂商需求无弹性和CoWoS/HBM优先配额，能够将晶圆和封装涨价完整转嫁，溢价稳健；EDA/IP分成（Cadence、Synopsys、Arm）现金流与…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 3, "neutral": 0, "value": 31, "events": [ { "title": "推理侧ASIC的规模化替代已导致通用GPU定价权削弱，NVIDIA面临物料成本上升与推理市场份额流失的双重压力，2026Q3毛利率下修逻辑成立。", "summary": "调研显示，2026年5月主要云服务商（CSP）的自研推理ASIC已进入3nm成熟期。Google TPU v6部署量超160万颗，Meta MTIA v3预计年底覆盖35%推理机群，微软Maia 200已支撑GPT-5.2。ASIC相比通用GPU可降低30%-50%的TCO，这导致推理定价权从NV…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "M2M流量爆发的边际成本不会均匀挤压SaaS——它会让薄壳代理编排厂商利润坍缩、让有自有数据的垂直SaaS继续吃到按量计费红利，并把被争夺的利润集中流向芯片、封装与光模块基础设施层。", "summary": "截至2026-05-16，本卡对card-07的SaaS基本面Beta判断做压力测试。结论是有条件成立但显著分化：薄壳代理编排SaaS将因无法转嫁的token成本，在未来12–18个月承受600–1,200 bp毛利压缩；而垂直SaaS、数据合规和API适配器（card-06超配名单）边际成本以存…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 生产率主线挺过 8%–15% 宏观回撤——溢价倍数被压缩，但盈利曲线不发生趋势性恶化；硬件核心仓继续持有，二阶新型云与软件溢价名应削减。", "summary": "在 global-macro 卡设定的 8%–15% 情绪回撤基线下，AI 硬件与算力 capex 方向上完好但盈利曲线边际走平、链条分化加剧。Hyperscaler FY26 capex（约 5,100–5,400 亿美元）已通过合同与电力 PPA 锁定，约束在电力交付而非资产负债表，对回撤敏感…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 3, "neutral": 0, "value": 32.4, "events": [ { "title": "AI硬件与算力基础设施具备足够的选择性资本确定性，但应买入瓶颈环节和已验证国产替代，而不是泛AI主题贝塔。", "summary": "截至2026-05-18，本卡片支持从大金融防御性收缩转向盈利可见成长。AI硬件具备选择性资本确定性：NVIDIA、TSMC、SK hynix及800G/1.6T光模块已体现高毛利兑现，云厂商2026资本开支提供需求锚。国产化替代真实存在，但确定性集中在服务器、光模块、热管理、电源、PCB和已商业…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Nvidia 直供链上的 A 股份额东移是真实但窄的（5–8 家硬名单），归因是合同制造而非美方约束；CPO 光引擎核心仍在美方，真正的份额弹性在独立订单簿的国产替代层。", "summary": "在 GB300 NVL72 与 Rubin CPO 架构下，A 股供应链份额东移的真实增量集中在 800G/1.6T 光模块 OEM（中际旭创、新易盛、天孚）、液冷 CDU 与冷板（英维克、高澜、工业富联整柜）、超低损耗高速 PCB（沪电、胜宏）三个环节，硬名单仅 5–8 家，远小于市场涨停所定价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 资本开支和利润率足以支撑已验证算力基础设施链条的估值修复，但不足以支撑泛 AI 成长股普涨。", "summary": "2026-05-17：本卡对 Research note 7 的“利率波动见顶后配置高现金流长久期成长”做压力测试，结论为有条件支持。AI 算力链条的盈利确定性主要集中在硬件和基础设施层：NVIDIA、Broadcom、TSMC、工业富联、中际旭创等已体现收入、利润率和经营现金流共振；但 hype…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 4, "neutral": 0, "value": 27.8, "events": [ { "title": "2027-2028年电力缺口不会终结AI芯片周期，但会把下一代AI硬件迭代逻辑从单芯片峰值性能改写为受限MW下的有效AI输出、上线时间和机架级能效。", "summary": "截至2026-05-18，本卡支持前序电力瓶颈主线，但压力测试“硬件从极致性能转向能效”的表述：更准确地说，采购函数会转向每已通电MW的有效AI输出。2027-2028年表后燃气仅4-6GW，无法覆盖AI负荷缺口；DOE/LBNL预计美国数据中心用电2028年达325-580TWh。NVIDIA …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS/HBM是配给型而非崩塌型瓶颈，使perf/W曲线弯曲而非折断，并把AI加速器BOM定价权向HBM厂与OSAT倾斜。", "summary": "截至2026-05-18，CoWoS/HBM是继电力之后的下一个物理瓶颈，但属于配给型而非崩塌型。TSMC CoWoS 2026年底约135–150kwpm、2027年底约190–220kwpm，CoWoS-L占比升至55–60%；行业HBM 2027比特供给同比+55–65%，但加速器需求+70…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-05-19，AI芯片架构已进入能效竞赛，能显著放大每个已并网MW的算力产出，但只能缓解而不能消除电力约束。", "summary": "截至2026-05-19，本卡支持“AI基础设施转向P/Watt”的主线。NVIDIA Blackwell/GB200、AMD MI350、Google Ironwood、AWS Trainium3均显示竞争焦点已从单芯片峰值算力转向低精度、HBM、机架互联、液冷和软件调度带来的每瓦有效token…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Tokens/MW时代下，AI敞口应从泛半导体beta再平衡到物理瓶颈栈（已通电MW+HV设备+中游气+液冷），半导体仅集中持有perf/W胜出者。", "summary": "截至2026-05-18，综合01–04卡：Big-4 Capex(~$705-725bn)未减速，瓶颈在物理侧；表后燃气2027-2028仅4-6GW，残余15-25GW缺口；硬件采购函数翻转为tokens/MW与time-to-power。AI价值链分化为瓶颈租金层（已通电MW、HV设备、中游…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 2, "neutral": 1, "value": 52.2, "events": [ { "title": "若 2026-06-11 验证 Waymo 单位经济，AI 基础设施不应被卖出，而应从拥挤算力前端轮动到电源液冷、networking 邻近环节与小仓位机器人硬件 beta。", "summary": "截至2026-05-19，本卡支持 6 月 11 日后从 AV 扩散到 AI hardtech，但只在 Waymo 经济性披露有用时执行。确定性最高的是电源/液冷/机柜、定制芯片与 networking、HBM/先进封装；估值弹性最好的是电源热管理与机器人端侧硬件。HBM、GPU、A 股光模块和主…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "截至 2026-05-19，AI 基础设施仍是 TMT 领涨主线，但可投资范围应收窄到有真实 usage、backlog、产能瓶颈和现金流承受力的环节。", "summary": "工作日期 2026-05-19。本文否定前序空白 TMT 论点并选择 (A)：AI 基础设施仍是领涨主线。证据是 hyperscaler capex/收入比虽已升至约 30% 的四家公司 Q1 合计强度，且 Alphabet、Amazon 的 FCF 张力明显，但 Microsoft AI ARR…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "消费电子'技术性弱复苏'+AI 手机溢价被压在旗舰段，触发的是 A 股'叙事溢价 → 确定性溢价'的哑铃化再定价，而非全面价值回归或单边 AI 基础设施集中。", "summary": "工作日期 2026-05-19。综合 card-03（AI 基础设施仍是窄口径主线）、card-04（消费电子 2026H2 技术性弱复苏）与 card-05（以旧换新净撬动仅约 1.25x、AI 手机溢价仅 ≥6,000 元旗舰段）三条结论后，本卡的策略级判断是：A 股既不会单边全面价值回归，也…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "A 股 AI 应相对美国半导体上调，但只买让安全算力容量可物理部署的电力、热管理、800V 和国产算力链条，而不是无差别追逐 AI 主题。", "summary": "截至2026-05-20，本卡综合支持 Research note 04：NVDA 的单位经济性仍可由 Rubin、800V DC、液冷和 CPO 维护，但海湾主权 AI 销量与估值倍数承压。因此应将 6-12 个月 AI 基础设施风险预算中 10-15 个百分点从依赖主权 AI 扩张的美国半导体…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "AI算力链盈利可见性已被订单簿三重验证、足以支撑保留成长端的策略；半导体设备应拆分成‘AI直接受益+国产替代超配，纯中国敞口设备与晶圆厂低配’，故第03卡‘不全面切向价值’成立但需要…", "summary": "截至2026-05-21，本卡对第03卡“不全面切向价值”进行压力测试，结论是有条件支持。AI算力链（GPU/HBM/光模块/液冷/先进封装）盈利可见性高且仍在抬升：四大美系hyperscaler 2026年合计资本开支指引上修至USD 360–410 bn、TSMC CoWoS与SK Hynix…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "液冷和Rubin提高每瓦token产出，但不会创造firm MW；行业将从单一GPU扩张转向Rubin高端算力与低功耗ASIC推理并存的分层架构。", "summary": "截至2026-05-22，本卡对前序电力瓶颈结论做压力测试：液冷和Rubin级高密度架构是必要条件，但不能消除firm-MW和输电约束。Rubin相对Blackwell可实现最高10x推理吞吐/瓦、训练MoE模型所需GPU降至四分之一；液冷可降低冷却开销并支撑120 kW级机柜。但EPRI相关数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "'效率降级'对 2026–2027 AI Capex 总量近乎无约束，但通过电力、关税、航运三条渠道侵蚀单位经济学，使全球 AI 基础设施从'同步扩张'转为'地理极化'，估值真正承压…", "summary": "本卡对 Research note 1–5 的“效率降级削弱 AI 算力扩张”假设进行双向压力测试。结论：方向正确但程度被高估，AI 算力 Capex 在 2026–2027 名义增速仍 +19–26% YoY（北美 hyperscaler 2026 指引 4,400–4,650 亿美元 [S1–…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 3, "neutral": 1, "value": 45.2, "events": [ { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "自研ASIC与先进封装可以把AI推理竞争从“买卡扩容”转向“每瓦有效推理”，但只有软件路由约束System 2调用时，能效提升才会真正缓解电力与散热瓶颈。", "summary": "截至2026-05-25，本卡支持“System 2推理将AI推向能源与散热约束”的前序判断，但认为ASIC与先进封装能显著推迟能源墙。自研推理ASIC、FP8/FP4低精度、HBM/CoWoS/SoIC、背面供电和直接液冷可以降低每token能耗，并提高每兆瓦可承载的有效推理吞吐。关键限制是杰文…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？", "summary": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？ DeepSeek-V4-Pro 75% price cut × NVDA FY27Q1 Data Center +92% joint p…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI算力周期的短中期瓶颈是time-to-power，Blackwell级部署会更多受可送电MW、变压器和燃机排产约束，而非单纯受GPU供给约束。", "summary": "截至2026-05-25，本卡压力测试AI算力建设的能源侧物理约束：GB200 NVL72约120 kW/机柜，使100 MW IT电力仅对应约59,976颗Blackwell GPU；并网队列、变压器交期、FERC共址审查和燃机产能缺口会把GPU采购转化为可用FLOPs的节奏拉成阶梯式。结论不是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "短期硬约束在 HBM 与 CoWoS-L 而非电力，AI 算力'紧但能出货'，对指数 FOMO 有利但 2026H2 脆弱性显著上升。", "summary": "短期（2026 年内）芯片侧的约束比电力更紧：依次瓶颈为 HBM3e bit 供给 → 台积电 CoWoS-L 先进封装 → GPU 晶圆 → 基板/光模块 → 电力。HBM 与 CoWoS-L 全年额度已锁、扩产跑不过 ~150%/~70% 的需求增长，2026H2 任一良率事件或出口管制重置可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 3, "neutral": 0, "value": 45, "events": [ { "title": "台积电CoWoS与HBM底层封装良率构成全球AI芯片出货的硬性物理约束，超额利润集中于垄断性瓶颈厂商（TSMC/MU/SK Hynix），而A股代理股共享叙事却无壁垒且交易极度拥挤，…", "summary": "支持超配海外半导体物理瓶颈（MU/SOXX）并降配A股算力代理。台积电CoWoS封装产能从2024年底3.5万片/月扩至2026年底超12万片/月但仍售罄，NVIDIA垄断超60%份额；HBM市场2026年规模将达600亿美元，美光产能售罄且良率升至70-75%。A股光模块、服务器组装与IDC代理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "4.5%以上美债环境下，AI硬件龙头的业绩增长足以支撑选择性配置，但不足以为整个AI久期交易兜底。", "summary": "2026-05-28结论：本卡对上一张宏观卡采取stress-test但不否定。4.50%美债锚会压缩AI久期资产估值，但NVIDIA、TSMC、Broadcom、SK hynix/Micron等稀缺硬件链凭借高增长、真实云厂商capex、先进封装和HBM瓶颈，仍能部分抵消估值下修。风险在于cap…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI仓位可以保留，但必须集中到已经把AI资本开支兑现为高毛利率收入和现金流的半导体瓶颈资产。", "summary": "截至2026-05-27，本卡支持“AI保权重”，但主张把AI仓位从广义beta收窄到盈利兑现型硬件龙头。最清晰路径是NVIDIA、TSMC、Broadcom，其次为SK hynix/Micron和ASML；Dell可做战术服务器渠道敞口，AMD列入观察型挑战者，Supermicro等低毛利率、重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 3, "neutral": 0, "value": 46.4, "events": [ { "title": "电力硬约束并未削减AI总投入，而是迫使数据中心从横向扩张转向纵向密度跃升，加速定制ASIC与高密度液冷的爆发，进一步逼死二线算力租赁商。", "summary": "基于55-65GW的物理上限，AI芯片数量的“指数增长”将被迫转向“受限增长”，但Hyperscaler在2026年6300-7500亿美元的资本开支总额并不会放缓。大厂正在将资本推向追求极致算力密度（120kW+液冷与高端GPU）与定制ASIC（+44% YoY），同时通过淘汰传统x86 CPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与液冷散热的产能扩张确实进入10×量级周期，但受ABF/HBM/稀土/铜的上游约束，仅能在制造维度缓解约35–40%的三重瓶颈，无法证伪TMT分析师的结构性紧张判断。", "summary": "工业制造视角支持TMT的三重瓶颈框架。TSMC CoWoS月产能从2023年末13K wpm扩张至2026年末130K wpm（近10×），但CoWoS-L/S全年售罄、NVIDIA独占60%以上配额，ABF载板与HBM堆叠的18–24个月扩产周期意味着封装紧张延续至2027年中。液冷市场2026…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "对元板块六子板块组合在路径A升级、因子拥挤逆转与CNY尾部三重情景下的组合级压力测试 首席风控官报告 | Research note 08 / 08 | 2026 05 29（亚洲/…", "summary": "对元板块六子板块组合在路径A升级、因子拥挤逆转与CNY尾部三重情景下的组合级压力测试 首席风控官报告 | Research note 08 / 08 | 2026 05 29（亚洲/新加坡） 立场：stress test（压力测试） 1. 前言与今日日期锚定 本报告为 AI 基础设施元板块专题研究…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 59.8, "events": [ { "title": "AI 基础设施的“铜脊”互连与液冷转型形成了价格不敏感的实物需求底部，支撑 LME 铜价在 15,500-16,500 美元/吨区间实现结构性重估。", "summary": "2026年5月审计显示，AI基础设施已进入“铜脊”与液冷驱动的实物超级周期。NVIDIA GB300等架构采用的高密度铜缆（单机架2英里）及液冷系统（强度40-50kg/kW）为铜价提供了坚实且价格不敏感的需求底部。测算显示，2026年AI、电网及EV带来的新增需求（+66-76万吨）将基本吸纳矿…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 4, "neutral": 1, "value": 46.2, "events": [ { "title": "AI 算力供给受限于 HBM4 售罄、CoWoS 封装良率及全球电网容量的物理极限，使硬件龙头成为 AI 价值链中无视‘Capex 断崖’风险的最终收费站。", "summary": "截至2026年5月，AI硬件供应链仍处于结构性紧缺。TSMC N2量产初期良率65-70%，HBM4于2月投产且美光2026年产能已售罄。NVIDIA Rubin架构推升单GPU功耗至2.3kW，使电力供应成为比芯片更硬的瓶颈。尽管终端应用出现DAU增速放缓，但硬件端的实物稀缺性构成了核心护城河，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI硬件足以证明哑铃结构需要一个选择性进攻仓位，但进攻端必须窄化为稀缺AI基础设施瓶颈，而不是泛TMT或泛成长。", "summary": "截至2026-05-31，本卡片压力测试认为：AI硬件与半导体链条能提供真实但窄口径的TFP弹性，足以支撑A股哑铃结构的进攻端，但不能验证泛TMT贝塔，也不能逆转3.25-3.45%的宏观潜在增速下修。全球半导体、NVIDIA、TSMC和中国云端AI收入显示算力需求与制造效率真实存在；但资本开支、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超级IPO将把AI公司的公开融资压力传导为算力锁定需求，从而重估半导体资本开支：超配稀缺瓶颈，回避无法证明产能稀缺和合约期限的长尾扩产。", "summary": "截至2026-05-31，本卡片支持“超级IPO流动性虹吸”对AI硬件和半导体瓶颈资产的强化作用。核心判断是，IPO资金不会平均利好半导体，而会把公开市场资金约束转化为GPU、HBM、先进封装、先进制程、光刻和数据中心网络的多年期需求可见度。TSMC、NVIDIA、ASML、Micron、Sams…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "AI 硬件不共享软件端的 SBC 稀释墙，但 capex 弹性 + HBM 成本 + ASIC 替代三重传导意味着 NVIDIA 与 AI 供应链的利润率扩张预期同样需要下修。", "summary": "2026-06-01，本卡对前序首席策略师结论作压力测试：AI 硬件端面对的不是同一道 SBC 稀释墙——NVIDIA 季度 SBC 同比 +30.8% 至 $1.928B 但会计稀释仅约 0.43%，显著低于 Amazon/Alphabet/Meta 一档——而是更陡的需求弹性墙。真正的压力源是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 1, "risks": 1, "neutral": 1, "value": 75, "events": [ { "title": "实物链四维加速更可信，卖方下调2027仅是FCF/折旧的估值反射；沿堆栈向下轮动至机电基础设施与液冷，芯片层中性偏交易，hyperscaler估值层中性偏谨慎。", "summary": "5月四维实物链同向加速（北美在建+14% MoM、PJM节点负荷+6.8%、GPU到港+18%、电力交期仍紧）与卖方下调2027 capex、NVDA/SMH走弱的背离，并非顶部信号而是健康再平衡偏实物再加速：Azure $80B、Google Cloud >$460B、CoreWeave ~$1…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Anthropic 的高估值实质上是基础设施稀缺性的资本化体现，其 3300 亿美元的算力承诺既是模型领先的物理保证，也是将云厂商 Capex 压力转嫁给二级市场的潜在风险点。", "summary": "Anthropic 9650亿美元的估值深度绑定了超过3300亿美元的云算力承诺（AWS 1000亿，Google 2000亿）。虽然这锁定了面向2027年单次200亿美元训练成本所需的5GW+电力与下一代 NVIDIA Rubin/TPUv7 产能，但也使公司陷入“资本开支通道”模型。2026年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI实物建设仍支持芯片需求，但2026-27年的边际份额赢家是定制硅和网络而非商用GPU，因此B层维持中性、内部偏好AVGO/MRVL并对NVDA估值更谨慎。", "summary": "截至2026-06-02，本卡支持Research note 01的B层中性，但建议改为杠铃式中性：更建设性看Broadcom和Marvell等ASIC赋能商，对NVIDIA保持近端战术正面、结构估值更谨慎。2026年定制ASIC份额27.8%大概率守住；按44.6% ASIC增速和16.1%商用…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "HBM/ASIC的结构性支撑在6-18个月维度上成立，但被卡片01低估了'客户集中度'和'瓶颈从硅迁移到电力'两条脆弱性——结构核心需要内部再平衡（提电力/冷却、减HBM股权、AVG…", "summary": "锚定2026-06-03（周三），AVGO FQ2业绩大概率周四盘后披露、NFP周五早间公布。我整体认同卡片01的'结构有支撑、战术有脆弱'综合判断，但TMT视角下结构性sleeve存在三条具体脆弱性需要再定价：(A) HBM需求高度集中在NVDA单一客户（约60-70%比特占比），属于穿着结构外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [ { "title": "TMT能用芯片能效、选址弹性、表后燃气与软件效率守住CAPEX美元，但守不住部署速度——估值通过time-to-power右移而非美元削减来反映监管成本，正是'好业绩不够好'机制的微…", "summary": "截至2026-06-06，TMT在位者用四条杠杆应对电力监管成本：芯片端Blackwell Ultra/Rubin的perf/W、选址迁移至MISO/SPP和跨境、表后燃气(2026-2027主力)与核电(2028+)、以及软件推理效率。压力测试结论是部分有效但不对称——hyperscaler已承…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 0, "value": 68.2, "events": [ { "title": "BTM自备电对AI硬件利润率的支撑是真实但间接的，最大受益者是上游加速器、先进封装和电源/液冷内容，而非通用服务器组装。", "summary": "截至2026-06-06，本卡支持“能源瓶颈支撑BTM与AI硬件链”的主线，但强调传导不是服务器直接提价，而是保护数据中心通电进度、硬件出货节奏和先进封装/晶圆代工利用率。PJM与ERCOT大负荷接入压力、FERC共址规则、Oncor数据中心队列，以及NVIDIA、Dell、TSMC最新披露共同显…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "7月关税与海运上行更可能压缩AI硬件下游组装与网络设备利润率，而非击穿核心半导体龙头利润率；估值风险来自低利润率硬件收入增长的质量被重新定价。", "summary": "截至2026-06-08，本卡对第04卡的关税/海运鹰派尾部做TMT压力测试：风险真实存在，但主要不是核心GPU、晶圆代工或HBM利润率同步塌陷，而是下游AI服务器、网络设备、机柜、电源、散热和PCBAs的落地成本挤压。NVIDIA、TSMC、SK hynix拥有高毛利率和稀缺定价权；Dell式A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "Rubin Capex能保护一线AI硬件EPS，但WACC上行会把二线供应链从“AI成长资产”重新定价为“融资周期资产”。", "summary": "截至2026-06-08，本卡对前序AI硬件防御结论做压力测试：Rubin周期与云巨头Capex仍能支撑NVIDIA、TSMC、Broadcom、Samsung、Micron等一线硬件龙头的EPS增长，但不能让整个供应链免疫WACC上行。可见四大云厂商Capex底线约5742亿美元（自测算），足以…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 3, "neutral": 2, "value": 66.4, "events": [ { "title": "AI半导体利润池迁移真实但渐进，板块总EPS上行、估值锚从'NVDA-TSMC双寡头溢价'切向'增长+多元化第二供应源'，ASIC设计商、HBM、先进封装设备与Intel Found…", "summary": "2026-06-09综合：Google-Intel TPU订单确认AI硬件利润池迁移真实且渐进，2027-28年AI半导体合计EPS较冲击前共识更高，但分布坚定向ASIC设计商(AVGO/MRVL)、HBM、先进封装设备及Intel Foundry期权倾斜。新估值锚为'增长+多元化第二供应源'，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Google-Intel的TPU订单标志着AI硬件利润池的结构性迁移：从绝对依赖GPU/TSMC，转向ASIC出货量反超以及Intel Foundry作为第二供应源的崛起，但GPU仍…", "summary": "2026-06-09评估：Google向Intel Foundry下单2028年超300万颗TPU，且Nvidia正评估Intel 18A工艺，这表明AI总需求依然强劲（到2028年TAM达3800亿美元），但硬件利润池正发生结构性迁移。为降低TCO，云厂商加速自研ASIC，出货量预计将超越GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 硬件的增长上限已由‘数字世界’转为‘物理世界’，电力缺口将迫使市场从纯算力 beta 转向通电效率与重型电力基础设施。", "summary": "2026-06-09：本卡对 AI 硬件估值进行压力测试，指出物理电力供应已取代芯片产量成为核心瓶颈。电网并网积压（36-84个月）与电力硬件（变压器）交期过长（24个月以上）导致 2026 年近半数数据中心容量面临延期。NVIDIA Rubin 架构功耗激增至 2.3kW/GPU，迫使架构全面转…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Google-Intel TPU订单是否开启AI硬件轮动", "summary": "Google-Intel TPU订单是否开启AI硬件轮动 Does Google's Intel TPU order mark an AI hardware rotation? Google据报向Intel下单2028年超过300万颗TPU、Nvidia也在评估Intel技术：这应被解读为AI总需…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 2, "neutral": 0, "value": 67.8, "events": [ { "title": "AI硬件正通过2.3kW单芯片功耗极限与800V液冷架构强行突破物理墙，算力扩张的限制已从芯片设计转移至电力设施配套的48个月长周期交付延迟。", "summary": "截至2026-06-11，AI硬件并未因电力约束而降低TDP，反而迈入2.3kW（NVIDIA Rubin）的“暴力能效”时代。这迫使行业全面转向100%液冷及800V DC机架架构。硬件演进逻辑已从“硅片优化”转向“基础设施重构”，先进封装（CoWoS-L/HBM4）与光互连（CPO）成为维持算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI半导体核心标的的Q2/Q3盈利能见度极高，受限的电网电力将加速Blackwell架构升级，短期流动性冲击构成极佳的左侧建仓窗口。", "summary": "截至2026-06-10，在SpaceX IPO（SPCX）可能引发SMH/QQQ短期30亿至60亿美元流动性抽水的背景下，本报告研判AI半导体核心标的（Nvidia, TSMC）Q2/Q3盈利能见度依然坚实。TSMC 2026年520-560亿美元资本开支高企，CoWoS先进封装产能供不应求。即…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 4, "neutral": 1, "value": 60.2, "events": [ { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-11：电力与电网瓶颈会削弱半导体盈利确定性的斜率和估值，但尚不足以推翻半导体核心资产作为质量权益的配置逻辑。", "summary": "2026-06-11：本卡压力测试认为，AI 数据中心电力与电网瓶颈会压低半导体盈利确定性，但主要表现为投产节奏、算力成本和估值折价，而非短期需求崩塌。美国与全球数据中心用电快速上行，PJM、FERC、NERC 已显示接入和可靠性压力；但 NVIDIA、TSMC 与 hyperscaler cap…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源融资成本上升会削弱云客户ROIC并压低部分ASP，但最强AI硬件平台商仍有足够利润率缓冲，真正脆弱点在重capex制造和周期性内存环节。", "summary": "2026-06-11，本卡压力测试能源债务融资成本上升对AI硬件供应商利润率的影响。结论是方向成立但不应外推为硬件利润率全面坍塌：新增能源债务成本上行约85 bps会压缩云厂商自由现金流并强化采购纪律，但NVIDIA和Broadcom因高毛利、平台锁定和轻资产结构韧性最强；AMD居中；TSMC和M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力交付滞后会迫使 Hyperscalers 调整 GPU 集群投产节奏，并通过更长现金回收期、更低早期利用率和更高电力保障成本压低 AI 基础设施 capex 效率。", "summary": "截至2026-06-11，本卡支持上一卡：AI capex 不会因电网瓶颈消失，但 GPU 集群部署节奏会被“已通电 MW”重新排序。变压器、并网和输电延迟会把部分已宣布园区容量推迟到后续窗口，压低早期自由现金流和资产利用率。Hyperscalers 会通过已通电站点溢价、分阶段投产、区域重排和自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 1, "risks": 4, "neutral": 1, "value": 69.6, "events": [ { "title": "为支撑三层算力织物，AI芯片与先进封装如何演进以匹配电力波动与边缘推理的功率包络", "summary": "综合并支持卡片04的三层算力织物，下沉到硅基与封装层。关键洞见：'电力波动友好'主要是电力电子＋调度固件问题，而非新晶圆问题。NVIDIA GB300靠电源货架（65 J/GPU储能＋功率封顶斜率＋功率烧除）把Megatron训练电网峰值砍约30% [S1]，DVFS早是标配 [S2]；同步功率摆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的扩展性：电力与芯片供应瓶颈对超大规模厂商CAPEX的影响分析", "summary": "截至2026-06-12，本卡压力测试卡片02（电力瓶颈）与卡片06（半导体多线挤压）。结论：双重约束不会削减超大规模厂商资本开支，而是将其重排。2026资本开支仍约$725-726B（Dell'Oro），资本开支/营收比约25-30%、指引2027约35%；变化在优先级——可靠电力升至栈顶，建设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力基础设施架构的重塑与硬件需求演变", "summary": "支持卡片01。当并网容量（兆瓦）成为固定稀缺输入，优化目标从“每机柜最大FLOPS”翻转为“每交付兆瓦最大有效算力”。这并非密度撤退——GB200 NVL72仍达120kW/机柜、800V路线图指向880kW[S1][S19]——而是密度转为服务效率。三大倒逼因素：PJM并网周期升至>8年[S5]…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力刚性约束已把AI加速器竞争维度切换为每瓦Token，Chiplet+先进封装是2027-2028电网窗口内唯一可规模化兑现的算力密度救援方案，对NPV的杠杆是5-8倍。", "summary": "电网刚性约束把AI加速器的KPI从峰值FLOPS切换为每瓦Token吞吐。Blackwell B200单卡1000W、GB200 NVL72机柜120kW已触顶，下一代Rubin/MI400无法继续抬TDP，必须靠能效换性能。TPU v5p、Trainium2较H100能效提升30-40%；同10…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 6, "neutral": 2, "value": 54, "events": [ { "title": "红皇后竞速并未彻底抹去自研ASIC的成本优势，而是将其集中至Google与AWS两家、使其他云厂商回到对NVIDIA的依赖，从而部分推翻了Research note 05所述的全行业…", "summary": "在'红皇后竞速'下，自研ASIC单世代有效寿命已压缩至12–15个月，2nm节点全口径NRE攀升至USD 0.9–1.4B、N2晶圆约USD 30,000/片、CoWoS-L售罄至2027。我们测算自研ASIC在TCO上跑赢商用GPU的年部署门槛从5nm节点的约20万颗升至现实情景下的15万–25…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "算法效率不会终结暴力算力，只会让暴力算力转移到推理与 Agent 层；NVIDIA 溢价将边际压缩但不会系统性回归，因为护城河在网络/软件/供应链而非 FLOPs。", "summary": "DeepSeek V3→R1→V4 路径压缩了训练成本，但 Jevons 效应把算力需求推向推理与 Agent 层，NVIDIA 护城河主要沉淀在 NVLink/CUDA/CoWoS-HBM 锁定/主权 AI 需求四层，并非裸 FLOPs 租金。1 月 27 日单日蒸发 5,890 亿美元市值是一…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "GB200 级效率提升改善 AI factory 经济性，但 2026-2028 年真正限制部署节奏的是站点电力、并网、变压器、稳定电源和电网运行能力，而不是 CoWoS/HBM 单…", "summary": "2026-06-13：本卡对 NVIDIA 护城河做电力压力测试。GB200 NVL72 单机柜约 120 kW，1 GW IT 功率约支持 8,333 个机柜、约 600,000 颗 GPU。IEA、DOE/LBNL 与 NERC 显示，数据中心用电、并网队列、变压器和燃机交期都在拉长，大型计算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS 与 HBM 扩张是 NVIDIA 继续销售高溢价系统的供给条件，而不是摧毁溢价的触发器。", "summary": "2026-06-13：本卡支持前序判断。CoWoS/HBM 扩产不是削弱 NVIDIA 护城河，而是让其把更多机柜级 AI factory 供给转化为计算、内存、网络与软件平台溢价。TSMC 的先进封装、3nm/HBM base die 产能，SK hynix/Micron/Samsung 的 H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 8, "neutral": 0, "value": 37.4, "events": [ { "title": "GPU抵押贷款收缩会削弱边际Neocloud订单可见度，但先进封装与先进制程需求更可能从融资套利订单转向信用更硬的超大规模厂商和ASIC路线，而非进入全面下行。", "summary": "截至2026-06-15，本卡压力测试GPU抵押融资收缩对半导体订单的影响：冲击不是AI芯片需求消失，而是订单可见度从Neocloud融资驱动采购转向超大规模厂商、ASIC、主权项目和带预付款的硬合同。NVIDIA Data Center在FY2027 Q1仍达752亿美元，TSMC先进节点和HP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力成本上升和并网延迟不会终结AI芯片周期，但会把赢家从峰值算力叙事筛选为能在稀缺MW下交付最高有效token产出的全栈能效架构。", "summary": "截至2026-06-14，本卡支持前序判断：AI扩张的瓶颈已从单纯GPU供给转向电力与并网，但半导体端可通过能效、液冷和先进封装显著对冲。设计指标将从峰值FLOPS转向每瓦token、每MW可部署token和机架级系统效率。Blackwell、Trillium TPU、Trainium2等路线显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电网瓶颈正迫使 AI 基础设施从‘暴力规模扩张’转向‘受约束的能效优化’，电力并网速度已取代芯片供应成为核心估值护城河。", "summary": "本报告通过压力测试确认了物理电网并网排队（7-10年）与变压器交期（4-5年）已成为 AI 基础设施扩张的硬约束。定量模型显示，1年的并网延迟将导致 100MW 数据中心损失 5.5% 的生命周期价值，且快速的 GPU 迭代周期（12-18个月效用减半）加剧了闲置产能的 ROI 损耗。这种物理瓶颈…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI先进封装与HBM产能瓶颈对算力基础设施扩张及超大规模厂商资本开支的影响", "summary": "支持半导体分析师Research note 01的论点并向下游延伸：CoWoS/HBM确为算力的关键约束，2026年部署受封装与内存所限（CoWoS扩至约13万片/月仍订满；SK Hynix与Micron 2026年HBM售罄；HBM4验证延迟使NVIDIA将Rubin目标由200万下调至150万…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 4, "neutral": 1, "value": 65.8, "events": [ { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观信用收紧并未击穿全球AI周期，而是将AI基础设施层切割为「超大规模云厂＋央企国家队」核心（继续扩张）与「民营商业云」外缘（防御性收缩）的K型结构，TMT投资应聚焦CoWoS/HB…", "summary": "TMT综合：沃什鹰派持稳（3.50–3.75%）经离岸人民币流动性、汇率对冲摩擦与企业中长贷收缩，传导至AI基础设施层产生K型分化而非总量收缩。美超大规模云厂CY2026合计capex运行率3,200–3,600亿美元由经营现金流支撑不受冲击；央企运营商云经PBOC科技创新再贷款（约2.5%）继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据中心的物理功耗墙迫使超大厂将有限的电力配额向具有3-5倍能效优势的ASIC倾斜，实质性加速了算力硬件市场从“通用高溢价”向“定制化能效套利”的重塑。", "summary": "2026年，电力配给成为AI基础设施的核心硬约束。四大超大厂高达6300亿美元的资本开支正大规模向功耗更低、每瓦特产出更高的定制化ASIC倾斜。通用GPU飙升至2300W的单卡功耗和超120kW的机柜密度让电网不堪重负，促使ASIC以近45%的增速抢占推理市场，从而实质性削弱了英伟达的算力溢价。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进液冷与高速光互联是重塑算力岛资产估值的核心驱动力，通过放大算力岛内部有效算力输出，能将项目ROIC从4.35%提升至25.34%。", "summary": "本卡量化分析了先进液冷与800G/1.6T高速光互联对智算岛有效算力输出（MFU、PUE、可用性）及财务回报的传导效应。在100 MW电力配额下，高效率方案可降低PUE至1.15，扩容GPU数量26.09%；减少网络丢包使MFU从35%升至55%；降低芯片结温并缩短崩溃恢复时间，可用性达95.8%…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 8, "neutral": 0, "value": 40.2, "events": [ { "title": "电力供给限制驱动的半导体能效转型主要表现为通用GPU的系统与精度级优化，而非ASIC/NPU的大规模替代；自研定制芯片的爆发反而加剧了全球先进制程与封装供应链对TSMC的物理集中度。", "summary": "截至2026-06-17，电力瓶颈确在倒逼算力能效转型，但并非单纯的通用GPU向专用ASIC切换。算法多变性（如MoE与动态推理）使自研ASIC面临极高的“代差失效”风险，通用GPU（如NVIDIA Blackwell）凭借软件与架构灵活性仍占据70%以上份额。能效提升主要通过系统精度（FP4/F…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装的去瓶颈把AI算力的约束推向电力与液冷，2026–2027年边际capex美元流向'电子与水'而非硅片，散热与电网设备是该传导链上确定性最高的边际增量。", "summary": "截至2026-06-16，TSMC CoWoS等效晶圆产能从2024年底约35k wpm扩张至2026年底约75–80k wpm，HBM/TSV封装侧瓶颈基本释放。约束顺势外迁至电力与液冷：Blackwell B200 NVL72机架功率达约120 kW、Rubin级150–200 kW/rack…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建资本开支仍在加速，但从公告美元转化为已通电MW的比例正在下降，首个断点在数据中心交付和二线下游供应链，而不是NVIDIA近端GPU订单。", "summary": "截至2026-06-16，未看到北美云厂商AI资本开支或GPU订单总量已经广泛下修；Microsoft、Alphabet、Meta、Amazon、NVIDIA、Dell披露仍指向扩张。支持前序压力结论的证据主要来自数据中心交付层：Goldman Sachs预计未来1-2年计划容量仅50-60%按时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体能效能把AI扩张从“堆GPU”改写为“每MW token产出竞争”，但只能买时间、改善单位经济性，不能让电力瓶颈消失。", "summary": "截至2026-06-16，本卡压力测试“电力预算驱动扩张”逻辑：半导体能效提升真实且重要，Blackwell、先进封装、vLLM/FlashAttention、稀疏MoE和定制ASIC可显著提升每MW有效token产出。但它不能完全对冲电力瓶颈，因为美国数据中心用电量可能从2023年的176 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 7, "neutral": 0, "value": 55.6, "events": [ { "title": "AIDD是AI基础设施需求的高质量增量因子：规模小于通用大模型，但持续性、数据壁垒和监管/IP属性更强。", "summary": "截至2026-06-18，本卡片支持上一张卡片的方向：AIDD不会单独决定hyperscaler capex，但会把高价值、持续性的生命科学工作负载接入GPU、网络、存储、液冷和电力保障型数据中心。Recursion BioHive-2、DCAI Gefion、Novo Nordisk、Isomo…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "GPU配额与电网互联是2026–2028全球AI布局的联合约束，资本流向同时拥有一档算力与开放电力排队的辖区，最佳投资敞口是电力/电气栈。", "summary": "截至2026-06-18，全球AI基础设施扩张由两个物理瓶颈联合制约：高端GPU配额（管制扼喉点）与可互联级电力容量（新约束）。TSMC CoWoS 2025年底扩至75k片/月仍紧[S3]，Nvidia数据中心年化~US$1,150亿[S4]，BIS 2023-10/2025-01 AI扩散规则…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FOMC 后应以“选择性算力基础设施挤空”而非“无差别 AI 融涨”来交易：多稀缺、空低 ROIC，净多但对冲。", "summary": "2026-06-18 策略综合：前序链条成立，2027 风险是 ROIC 与结构分化而非 capex 悬崖。FOMC 后应做多稀缺算力资产而非泛 AI beta：前沿逻辑/设备/HBM/先进封装、自研 ASIC 与网络、电力与电网受益方应超配；商用 GPU 龙头保留但严控风险；低 ROIC 商用 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 能源重定价压的是超大规模厂商 ROIC 而非芯片 capex 效率——硅片主导 TCO、能源占比单位数，且性能/瓦溢价+杰文斯回弹反而净支撑前沿逻辑/先进封装/HBM 需求…", "summary": "2026-06-18 压力测试：能源是证伪芯片逻辑的错误杠杆。Epoch 1GW 机房年化 TCO 约 $8.5B，服务器占 ~59%、能源仅 ~7%；交付电价 +30% 仅抬升 TCO 约 +2.1%，翻倍也仅 +7%（自测算），而 IT 寿命 5→3 年使 TCO +41%——刷新节奏的影响是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 1, "risks": 5, "neutral": 0, "value": 69, "events": [ { "title": "RUM的Together AI合同仅占用其未商业化电力的约3%，而完全填满200 MW所需的47亿至51亿美元资本开支面临巨大融资缺口，且TSMC CoWoS与HBM3E供给紧缺导致…", "summary": "截至2026-06-18，对RUM Group的B300 GPU采购进行供应链与财务敏感性测试显示：其与Together AI签署的2.7亿美元B300液冷云服务协议在物理规模上仅需约1,800至3,000张GPU，对应电力功耗仅约4.8至7.3 MW，仅占其剩余200 MW未商业化容量的2.4%…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力/电网瓶颈真实且已在容量价格中显形,但2027年前仅推迟部分美国新增集群、不封顶全球GPU部署,Research note 01的BOM组合在2026至2027 H1仍成立,20…", "summary": "AI算力电力瓶颈真实存在但属于'延后+置换'而非'封顶':PJM 2026/27容量价已跳11倍至$329.17/MW-日,2027/28预测缺口6.6 GW,新接入等待40个月;ERCOT 2026 Q1单季198 GW申请,累计大负荷队列约410 GW;全球电网接入等待5-10年,变压器交期1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM3良率瓶颈将AI基础设施风险从总量需求崩塌改写为本土投产节奏后移，并强化A股高融资TMT的收入兑现左尾风险。", "summary": "截至2026-06-18，本卡压力测试卡片06：国产HBM3良率瓶颈会推迟中国本土AI算力变现，但尚不足以证明全球AI基础设施从快速扩张转入需求观望。全球hyperscaler capex仍有Meta、Alphabet、Microsoft等高额指引支撑；真正变化在中国本土项目从按园区容量扩张转向按…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2028 年 AI 基础设施的胜负不在于是否完全离网，而在于能否把液冷机架、BTM/并网混合电源、闭环冷却、燃气交付和工作负载调度整合成可复制的 7-150 MW 阶段模块。", "summary": "2026-06-18：本卡压力测试前序主线。结论是 ERCOT/Permian 与 BTM 气电是 AI 训练负荷的真实泄压阀，但不能干净解决 2028 年瓶颈；它们把问题转化为模块化电力-冷却-算力架构。液冷从可选项变成高密 AI 的前提，132 kW/机架意味着 1 GW IT 负荷约需 7,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 1, "risks": 5, "neutral": 0, "value": 70.4, "events": [ { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与封装成本上升会触发AI基础设施技术路线调整，但截至2026-06-20仍不足以触发超大规模云厂商AI芯片采购总量的全面回撤。", "summary": "工作日2026-06-20：本卡压力测试认为，电力和先进封装成本上升不会在2026年触发超大规模云厂商全面削减AI芯片采购，但会提高单位弹性，推动技术路线从“更多GPU插槽”转向“每瓦性能、液冷机架、自研ASIC、提前锁定CoWoS/HBM和电力站点”。真正风险是并网、变电站、CoWoS或HBM配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "16.18%的硬件Capex下修会压低AI基础设施整体估值，但订单传导不是线性的：液冷最敏感，电网设备和BTM发电更受长期电力容量瓶颈支撑。", "summary": "截至2026-06-19，本卡把H2 AI硬件Capex下修16.18%作为未外部确认的压力测试输入。结论是分层传导：GPU/服务器采购最先受冲击；液冷因绑定高密度机柜节奏，订单和估值敏感度较高；电网设备因并网稀缺、变压器长交期和backlog，订单不应按同等比例下修；BTM自备发电估值会先压缩，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Starlink 带宽 × Grok 算力的飞轮真实，使 SPCX 成为 AI 分发基座与算力链条需求催化剂，但算力归属 xAI、轨道边缘算力尚萌芽，真正瓶颈是电力——故支持但需把光…", "summary": "从 AI 基础设施视角支持卡片 01：Starlink 带宽（约 450 Tbps，三代卫星单星 >1 Tbps、下行 10 倍/上行 24 倍）与 Grok/Colossus 算力（约 55.5 万颗 GPU、约 180 亿美元、向 2 GW 扩张）构成真实飞轮，使 SPCX 成为差异化的'算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 1, "risks": 3, "neutral": 0, "value": 83.8, "events": [ { "title": "虽然HBM国产化良率崩塌会将集群 CapEx 推升 98%，但中国互联网巨头通过异构分级部署与算法优化（缩减30%GPU需求）并结合西部绿电直连套利（LCOE降至0.39元）与液冷改…", "summary": "评估HBM成本压力对AI集群TCO及扩容策略的传导。100%国产材料导致HBM3封装良率崩塌至4.11%，推升10000颗GPU集群CapEx至8.17亿美元，东部煤电下3年TCO达10.17亿美元（+66.13%）。建议协同优化：架构上采取60% HBM3与40% HBM2E的分级部署，结合算法…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本粘性与高利率不会终结头部AI基建周期，但会把Hyperscaler Capex从“抢GPU”推进到“抢可交付电力、低PUE、签约需求和低资本成本”的ROI筛选阶段。", "summary": "截至2026-06-20，本卡对能源成本粘性和高利率对AI基础设施ROI的冲击做压力测试。结论是：头部Hyperscalers不会因电价与利率上行而停止AI Capex，但边际项目的回报门槛显著抬高。按1 GW IT负载、1.20 PUE、13.92 cents/kWh测算，年电费约$1.46 b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与 AI 数据中心能源架构：功率密度成为新的供应链关口 作者：agy（AI Infrastructure Analyst） 权威工作日期：2026 06 20 立场：supp…", "summary": "先进封装与 AI 数据中心能源架构：功率密度成为新的供应链关口 作者：agy（AI Infrastructure Analyst） 权威工作日期：2026 06 20 立场：support 研究线程标识：已归档rd Index：2/10 截至权威工作日期 2026 06 20 ，我支持上一张卡片关…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电网通电限制不会立即终结 AI 需求，但会把 TMT 估值从“芯片稀缺溢价”重构为“确定电力、利用率和现金流转化溢价”。", "summary": "截至 2026-06-21，本卡压力测试电网通电限制对 TMT 的影响：AI 需求仍强，但估值核心从 GPU 可得性转向 time-to-power、容量变现和 FCF 久期。Microsoft 和 Amazon 的主要风险是 AI capex 先发生、通电和云收入后确认，导致利润率与自由现金流承…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 5, "neutral": 0, "value": 65.2, "events": [ { "title": "AI 基础设施建设中的电力供应与电网容量瓶颈分析", "summary": "本卡压力测试 Research note 02 的隐含假设「资本开支≈算力」。证据显示约束已从芯片迁移到电力：微软披露约800亿美元 Azure 订单因电力受限无法交付，Nadella 称芯片堆库存却无「暖壳」可插[S5][S4]。转化在三处漏损：(1)并网排队约2,600 GW、中位等待约5年、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-22，低端封装过剩会压低错误产能池的价格，而高端迭代缺失会保留真正重要产能池的配给风险，使Hyperscaler AI资本开支更重、更不稳定、更容易出现已供电机房…", "summary": "截至2026-06-22，本卡压力测试卡片06：先进封装低端过剩并不能缓解Hyperscaler真正需要的高端CoWoS/HBM4认证产能短缺。TrendForce称2026年行业CoWoS-like产能可接近200,000 wpm、缺口或收窄至约10%，但若高端认证占比只有50%-70%，有效高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力对电网密度的压力测试：高壁垒HVDC与特高压能否满足数据中心极端负载需求？", "summary": "本卡对卡02–08隐含的'特高压/电网级HVDC扩产=缓解AI算力电力瓶颈'假设做压力测试并部分否定。核心是概念混淆：电网级特高压直流（西电东送）与机房内800V HVDC（NVIDIA/Google/TI/ST/Vertiv，2026Q3量产、2027大规模部署）是两条供应链。IEA显示全球数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力需求以约50%/年增长，而稳定电力供给受困于5–7年并网长队，电力（而非硅片）成为约束；液冷通过把PUE收益转化为事实上的电网容量，是近期唯一能实质性松绑该约束的杠杆。", "summary": "截至2026-06-22，AI大建设的真正约束已从GPU转为电力。2025年全球数据中心用电增约17%、AI数据中心激增约50%，IEA基准情景2030年翻倍至约945 TWh[S1][S2]；而四大超大规模厂商2026年合计资本开支约US$725bn、同比+77%[S4]。物理电网无法吸收：PJ…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 6, "neutral": 0, "value": 60.6, "events": [ { "title": "定制化ASIC驱动下的AI算力集群架构与液冷能效演进", "summary": "本卡支持并延伸 Research note 08：电网瓶颈倒逼的 GPU→ASIC 转向，沿“硅→织构→机房”逐层外溢，重构数据中心系统。互连层：Google 以 OCS+3D-Torus 抽掉一层电交换机，NVIDIA 以 CPO（Quantum-X/Spectrum-X 2026 商用）把光引…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "summary": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与定制化ASIC的深度整合，是将大模型访存能耗降低95.7%并实现3-5倍能效提升的核心硬件路径，亦是国产半导体链规避先进制程封锁、支撑算力电力约束的战略根基。", "summary": "截至2026-06-23，本卡研判支持通过先进封装与ASIC能效路径应对算力电力约束。2.5D封装结合HBM3E使访存功耗降低95.7%（从35 pJ/bit降至1.5 pJ/bit）；自研定制ASIC能能效比高出GPU约3-5倍，万卡集群年省电费2,943万元人民币，并释放48MW电网容量。国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力成本楔子会重定价单位算力成本并推动训练类算力西迁，但不会触发东部算力全面撤离；赢家是液冷、配电、绿电稳定化和同时受益于东部改造/西部新建的物理瓶颈供应商。", "summary": "2026-06-22：支持卡片07的压力测试结论。中国AI算力的核心约束不是全国缺电，而是PUE、绿电采购、差别电价、储能与液冷改造叠加形成的运营商成本楔子。以GB200 NVL72为参考，PUE 1.45较1.20令单位GPU-hour设施用电增加20.8%；叠加绿电和PUE罚价后，东部存量站点…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 5, "neutral": 0, "value": 68, "events": [ { "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下修。", "summary": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "summary": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力瓶颈不会压低AI Capex总额，但会把Hyperscaler Capex从“买GPU”重构为“买可激活算力”：液冷、并网、电力设备和长期能源容量成为与GPU同等重要的部署变量。", "summary": "截至2026-06-24，本卡支持前序主线：AI算力约束已从GPU/Capex总额转向已通电MW及每MW效率。IEA预计全球数据中心用电2030年约945 TWh，LBNL测算美国2028年或达325-580 TWh；Sightline显示2026年16 GW计划容量仅5 GW在建、30%-50%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 4, "neutral": 0, "value": 75.4, "events": [ { "title": "xAI 算力扩张对 AI 基础设施与能源供应的压力测试", "summary": "对卡片 01 前瞻风险的压力测试：跑步机确实存在，但卡点不在资本。约束层级为资本 < GPU < 电力。资本充裕（xAI E 轮 200 亿/估值 2300 亿，外加 200 亿 SPV 与 35 亿 Apollo 方案；2026 年超大规模厂商资本开支约 6000–7250 亿），GPU 因英伟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国CSP成本上升背景下，AI算力基础设施供应弹性与扩容压力分析", "summary": "压力测试部分否定线程以电力为中心的框架。AI算力系统是供给绑定而非需求绑定：2026年超大规模厂商资本开支同比+77%至约7250亿美元[S1]，英伟达Blackwell售罄至年中、积压360万颗[S2]，因此约束的是配额优先级而非需求。核心锚定集群凭借锁定电力、GPU配额与表后自发电被隔离；放缓…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体杠杆清盘对Hyperscaler AI资本开支的近端传导主要是价格、排序和边际买方融资纪律，而不是一线供应链信用崩塌或Mega-cap资本开支急刹车。", "summary": "截至2026-06-24，本卡压力测试认为，韩国半导体杠杆ETF清盘会重定价AI硬件股票并强化资本市场审视，但不构成Mega-cap Hyperscaler近端AI资本开支削减的主传导。Microsoft、Alphabet、Meta、Amazon仍以强经营现金流和投资级融资支撑AI建设；NVIDI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 7, "neutral": 0, "value": 58.8, "events": [ { "title": "截至2026-06-25，AI硬件需求的下一核心约束不是芯片是否能生产，而是足够多的兆瓦能否被许可、接入、冷却并按期商业化。", "summary": "截至2026-06-25，本卡支持上一张卡的AI硬件稀缺性主线，但将瓶颈从GPU/HBM进一步上移到可通电数据中心、电网接入、冷却和电气设备。IEA、Berkeley Lab、EPRI、CBRE和JLL数据共同显示，AI需求仍强，但项目能否转化为创收产能取决于电力和并网速度。结论是：稀缺性继续支撑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力稀缺不会杀死AI芯片周期，但会把竞争重心从单纯峰值算力转向固定兆瓦内的有效吞吐和能效，从而收窄可投资赢家。", "summary": "截至2026-06-25，本卡对电力和并网约束做半导体侧压力测试：约束不会终结AI芯片需求，但会把胜负指标从峰值FLOPS推向每兆瓦吞吐、每机柜token、HBM每瓦带宽和部署可用性。NVIDIA的机柜级平台、Google TPU、AWS Trainium、TSMC背面供电和Micron HBM4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备瓶颈不会系统性消灭AI加速器需求，但会把2026-2028年约15%-25%的GPU/HBM/ASIC部署变成收入确认和爬坡时点风险。", "summary": "截至2026-06-25，本卡从半导体侧压力测试电力设备瓶颈传导：2026-2028年加速器等效部署中约15%-25%面临2-4个季度时点风险，真实终端需求毁灭约3%-7%。商用GPU最直接受通电MW影响，HBM因多年协议和稀缺配额更像库存节奏风险，自研ASIC既受益于每瓦token优化，也因客户…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 6, "neutral": 0, "value": 75.6, "events": [ { "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产大模型并未告别规模竞赛，而是在先进GPU受限的条件下把竞争单位改写为每瓦、每token、每受限加速器和每留存用户的有效智能产出。", "summary": "截至2026-06-28，本卡对“国产大模型不可逆转向效能极限竞赛”进行压力测试。结论是：效能迁移真实且受GPU出口、H20许可、HBM/互连、液冷机柜和电力约束推动，但不是规模竞赛终结，而是进入“稀缺规模竞赛”。DeepSeek-V3、DeepSeek-R1、Qwen3和MiniMax-M1显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI算力基础设施需求与海底数据中心技术演进", "summary": "支持卡片01关于海兰信（300065.SZ）的判断并加以收紧：UDC卡位2026年AI建设的电力与散热两大硬约束。2026年超大规模云资本开支约7250亿美元，全球数据中心用电将破1000TWh（超日本），单柜功率向370–600kW飙升，高盛预计年底76%的AI服务器需液冷。海兰信是国内唯一掌握…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 3, "neutral": 0, "value": 87, "events": [ { "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,80…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck fr…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-cb2ab35c55/", "networkHref": "/investor/research-network/#thesis-cb2ab35c55" }, { "id": "thesis-741d1d6f34", "source": "tracked_thesis", "title": "地产、土地财政与 REITs 是中国信用底的慢变量", "titleEn": "Property, land finance, and REITs are the slow variables behind China's credit floor", "lane": "地产信用底/REITs", "laneEn": "China property credit floor and REITs", "laneHref": "/investor/lanes/china-property-credit-floor-and-reits/", "investableFocus": "地产信用底/REITs：0960.HK, 1109.HK, 0688.HK, 600048.SH, 001979.SZ, 0823.HK", "stockUniverse": [ { "ticker": "0960.HK", "name": "龙湖集团", "market": "HK", "role": "民营房企信用观察", "roleEn": "民营房企信用观察", "bucket": "core", "href": "/investor/stocks/0960.HK/" }, { "ticker": "1109.HK", "name": "华润置地", "market": "HK", "role": "央国企地产核心", "roleEn": "央国企地产核心", "bucket": "core", "href": "/investor/stocks/1109.HK/" }, { "ticker": "0688.HK", "name": "中国海外发展", "market": "HK", "role": "央国企地产核心", "roleEn": "央国企地产核心", "bucket": "core", "href": "/investor/stocks/0688.HK/" }, { "ticker": "600048.SH", "name": "保利发展", "market": "A-share", "role": "A股地产核心", "roleEn": "A股地产核心", "bucket": "core", "href": "/investor/stocks/600048.SH/" }, { "ticker": "001979.SZ", "name": "招商蛇口", "market": "A-share", "role": "地产与园区开发", "roleEn": "地产与园区开发", "bucket": "core", "href": "/investor/stocks/001979.SZ/" }, { "ticker": "0823.HK", "name": "领展房产基金", "market": "HK", "role": "REITs 现金流对照", "roleEn": "REITs 现金流对照", "bucket": "core", "href": "/investor/stocks/0823.HK/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 供给", "roleEn": "HBM 供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" } ], "stockTickers": [ "0960.HK", "1109.HK", "0688.HK", "600048.SH", "001979.SZ", "0823.HK", "TLT", "IEF", "XLU", "510300.SH", "2800.HK", "MU", "005930.KS", "000660.KS" ], "coreView": "赛道：地产信用底/REITs。候选观察标的：0960.HK, 1109.HK, 0688.HK, 600048.SH, 001979.SZ, 0823.HK。当前共有 40 条证据、3 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 78, "eventCount": 160, "firstSeen": "2026-05-01", "lastSeen": "2026-06-29", "supports": 12, "risks": 124, "neutral": 24, "analystCount": 19, "sourceTypeCounts": { "分析协作": 8, "研究笔记": 147, "研究主题": 1, "研究线程": 4 }, "topTerms": [ "GPU", "A股", "AIDC", "商品", "REITs", "Property", "land", "finance", "slow", "variables", "behind", "credit" ], "whatChanged": [ "2026-06-29: 电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对冲有效却对冲错轴，继续超配电网设备而非纯绿电发电。", "2026-06-29: 商品+CTA危机阿尔法敞口在A股境内的可投资复制：资源/上游股票、黄金、商品期货与国债期货组合", "2026-06-29: 国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加速撤向水电红利及电网设备出海龙头，陡峭化了向高红利风格轮动的斜率，配置应重构为哑铃型…", "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。", "2026-06-28: Micron信号更像AI内存瓶颈的利润池再分配，而不是单纯DRAM顶部，但配置必须带周期止损。" ], "practical": { "sectorLane": "地产信用底/REITs", "sectorLaneEn": "China property credit floor and REITs", "exposureBucket": "信用、地产与久期", "tone": "orange", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "0960.HK", "name": "龙湖集团", "market": "HK", "role": "民营房企信用观察", "roleEn": "民营房企信用观察", "bucket": "core", "href": "/investor/stocks/0960.HK/" }, { "ticker": "1109.HK", "name": "华润置地", "market": "HK", "role": "央国企地产核心", "roleEn": "央国企地产核心", "bucket": "core", "href": "/investor/stocks/1109.HK/" }, { "ticker": "0688.HK", "name": "中国海外发展", "market": "HK", "role": "央国企地产核心", "roleEn": "央国企地产核心", "bucket": "core", "href": "/investor/stocks/0688.HK/" }, { "ticker": "600048.SH", "name": "保利发展", "market": "A-share", "role": "A股地产核心", "roleEn": "A股地产核心", "bucket": "core", "href": "/investor/stocks/600048.SH/" }, { "ticker": "001979.SZ", "name": "招商蛇口", "market": "A-share", "role": "地产与园区开发", "roleEn": "地产与园区开发", "bucket": "core", "href": "/investor/stocks/001979.SZ/" }, { "ticker": "0823.HK", "name": "领展房产基金", "market": "HK", "role": "REITs 现金流对照", "roleEn": "REITs 现金流对照", "bucket": "core", "href": "/investor/stocks/0823.HK/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 供给", "roleEn": "HBM 供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" } ], "stockTickers": [ "0960.HK", "1109.HK", "0688.HK", "600048.SH", "001979.SZ", "0823.HK", "TLT", "IEF", "XLU", "510300.SH", "2800.HK", "MU", "005930.KS", "000660.KS" ], "primaryStocks": [ { "ticker": "0960.HK", "name": "龙湖集团", "market": "HK", "role": "民营房企信用观察", "roleEn": "民营房企信用观察", "bucket": "core", "href": "/investor/stocks/0960.HK/" }, { "ticker": "1109.HK", "name": "华润置地", "market": "HK", "role": "央国企地产核心", "roleEn": "央国企地产核心", "bucket": "core", "href": "/investor/stocks/1109.HK/" }, { "ticker": "0688.HK", "name": "中国海外发展", "market": "HK", "role": "央国企地产核心", "roleEn": "央国企地产核心", "bucket": "core", "href": "/investor/stocks/0688.HK/" }, { "ticker": "600048.SH", "name": "保利发展", "market": "A-share", "role": "A股地产核心", "roleEn": "A股地产核心", "bucket": "core", "href": "/investor/stocks/600048.SH/" }, { "ticker": "001979.SZ", "name": "招商蛇口", "market": "A-share", "role": "地产与园区开发", "roleEn": "地产与园区开发", "bucket": "core", "href": "/investor/stocks/001979.SZ/" }, { "ticker": "0823.HK", "name": "领展房产基金", "market": "HK", "role": "REITs 现金流对照", "roleEn": "REITs 现金流对照", "bucket": "core", "href": "/investor/stocks/0823.HK/" } ], "instrumentTags": [ "0960.HK", "龙湖集团", "1109.HK", "华润置地", "0688.HK", "中国海外发展", "600048.SH", "保利发展", "001979.SZ", "招商蛇口", "0823.HK", "领展房产基金" ], "reviewCadence": "每日复核", "confirmationSignals": [ "再融资成本是否继续筛选资产负债表", "地方债供给是否挤压久期需求", "地产和 REITs 现金流是否改善" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "高等级信用", "现金", "银行负债端", "久期上行" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 103 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 19, "recentEvents": 103, "supportRatio": 0.07, "riskRatio": 0.78, "conflictScore": 87.3, "evidenceDepth": 100 }, "score": 72.8 }, "events": [ { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GPU", "A股" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性TAM上限，营收放缓促使估值压缩。相反，全球CPO光模块（中际…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GPU", "A股" ], "href": "" }, { "date": "2026-06-29", "title": "商品+CTA危机阿尔法敞口在A股境内的可投资复制：资源/上游股票、黄金、商品期货与国债期货组合", "summary": "受人民币与境外额度约束、无法直接持有全球CTA与BCOM的A股组合，可用境内工具复制第02卡所证的商品+CTA防御子仓：黄金ETF 518880、金矿股(紫金601899/山东黄金600547)、商品期货ETF(有色/能化/豆粕)、资源上游股票，以及做空中金所国债期货(30年TL/10年T)作利率上行变现腿，辅以小额境内商品CTA私募。诚实评级：通胀贝塔的…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "A股", "商品" ], "href": "" }, { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "AIDC", "aidc", "power" ], "href": "" }, { "date": "2026-06-28", "title": "USD/CNY汇率趋势对通胀对冲组合的影响评估", "summary": "截至2026-06-28，本卡支持前两张卡的再通胀与真实资产倾斜判断。对人民币本位组合，当前美元偏强（DXY约101.37）、人民币走软（USD/CNY约6.81，逼近一个月低点，PBOC连续偏弱设定）、且把USD敞口对冲回人民币成本高（USD–CNY货币市场利差约2.2pp、美元远期贴水），意味着商品/黄金USD计价仓位的外汇腿本身就是通胀对冲的一部分。…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 16.75, "matchedTerms": [ "A股", "商品", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "再通胀预期下应把组合向通胀对冲和短久期防御倾斜，但不是全面risk-off。", "summary": "截至2026-06-28，本卡支持上一张宏观判断：美国再通胀风险上升，传统股债风险平价的分散性下降，但尚未到全面撤出增长资产的程度。建议将100%风险预算从40%权益/35%名义债/15%商品/10%现金，调整为35%权益/25%名义债/25%通胀对冲及商品/15%现金；A股内部偏高股息、上游定价权、成本传导和质量价值。下一步应由fx-strategist…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16, "matchedTerms": [ "A股", "商品", "risk" ], "href": "" }, { "date": "2026-06-28", "title": "面向AI基础设施信用尾部的跨资产组合配置与对冲", "summary": "截至2026-06-28，本卡综合前序结论：AI算力长期上行仍可持有，但尾部风险在信用结构而非单纯科技股估值。建议采取“优质AI+电力瓶颈+信用保护”杠铃：保留NVIDIA与现金流型超大规模云厂商核心仓位，低配Neocloud股票和GPU抵押DDTL/私募信贷，增配受监管公用事业、电网设备、稳定电力与A股电网设备卫星，并以CDX HY、CDX IG和Ora…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GPU", "A股" ], "href": "" }, { "date": "2026-06-28", "title": "截至2026-06-28，A股电力主设备与核心分接开关商在手订单质量高、兑现度极强，而高估值温控链及无MEC保护的直接出口意向订单面临较大递延折价风险。", "summary": "截至2026-06-28，本报告压力测试了A股AI电力设备商的可验证产能。特变电工与中国西电特高压变压器排产至2027年，订单由国网投资支撑；金盘科技中压干变通过海外基地规避关税，但需警惕现金流与发出商品堆积风险；思源电气与平高电气在开关柜及出海Alpha上表现优异，具备合同价差传导保护。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "A股", "商品" ], "href": "" }, { "date": "2026-06-28", "title": "Micron信号更像AI内存瓶颈的利润池再分配，而不是单纯DRAM顶部，但配置必须带周期止损。", "summary": "截至2026-06-28，Micron FQ3-26和FQ4指引支持AI交易向内存、先进封装和系统成本再定价倾斜，但不是完全从GPU撤出。核心证据是FQ3-26收入$41.456B、FQ4-26收入指引$50.0B±$1.0B、16份SCA、约$100B RPO以及DRAM/NAND价格驱动的利润弹性。周期风险仍在：FQ4价格涨幅放缓，2027年capex…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "GPU", "A股" ], "href": "" }, { "date": "2026-06-27", "title": "算力设备折旧周期与电力并网延迟带来的资产减值压力测试", "summary": "部分印证并收紧第08卡，但重新界定机制：先货后电在2026年既非实物提前报废，也非典型GAAP冲销，而是“幻影日历贬值→加速折旧→2027-2028通电即减值”的三段链。GAAP折旧从投入使用起算，但贬值从到货起算；按英伟达年度节奏（Hopper/Blackwell/Rubin/Rubin-Ultra），24个月延迟在芯片赚第一分钱前烧掉前沿训练寿命的67…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GPU", "A股" ], "href": "" }, { "date": "2026-06-27", "title": "在通胀与加息风险重估中，长债防御力失效，需以低配长债、超配商品与尾部期权构建战术杠铃组合，并将A股防御重心转向红利重估。", "summary": "在二次通胀与能源价格上行风险下，传统股债对冲失效。应以“战术杠铃”替代风险平价：将长债降至10%，配25%短债与TIPS，提高商品配比至20%，并辅以10%尾部对冲期权。A股则应转向防御性高分红，防范高估值成长股因美债上行发生估值收缩及融资平仓。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "A股", "商品" ], "href": "" }, { "date": "2026-06-27", "title": "虽然备忘录缓解了短期实物原油压力，但霍尔木兹海峡的结构性脆弱与15%-25%的二次通胀概率支撑将大宗商品战术超配至20%，并应采用高分红个股现货替代及低配看涨期权策略进行防御。", "summary": "截至2026-06-27，评估表明尽管扫雷进展与伊朗出口正常化将推动短期WTI均价回落至67.95美元且跌破70美元概率升至62.85%，但15-25%的能源价格二次上尾再脉冲风险及长期地缘瓶颈支持将大宗商品配置超配至20%。在A股通道中，因159945等ETF极度缺乏流动性（成交仅106万元），建议以中国海油、中国石油等高分红龙头个股作为替代通道，配合低…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "A股", "商品" ], "href": "" }, { "date": "2026-06-27", "title": "高利率平台与电力交付约束下的AI算力产业链估值重塑与多空配置策略", "summary": "我对卡片05–07主线部分支持、部分否定。约束在电力，经济租金从过剩层（资本、GPU、可出租算力）迁移到稀缺层（已通电兆瓦、发电、超长交期电网设备）——重估命题正确且不对称：稀缺层应享更高估值，被杠杆与抵押挤压的算力出租商应承受更宽折价。但干净的多空表达式更脆弱：多头腿GEV已71倍前瞻不便宜，便宜的是AEP/VST/CEG（16–20倍）；空头腿（CRW…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GPU", "A股" ], "href": "" }, { "date": "2026-06-27", "title": "2026年7月应从追逐权益beta转为“低beta权益+短中久期债券+商品通胀对冲+现金弹药”的组合，以同时防御通胀粘性和就业放缓。", "summary": "截至2026-06-27，本卡支持“鹰派暂停”主线：5月PCE再加速、6月SEP上修通胀和利率路径，使7月组合不能按降息交易处理。建议2026年7月将组合调至35%权益、40%债券、15%商品、10%现金/汇率对冲；权益保留A股但转向高股息、质量和低beta，债券以中国利率债与美国短中端为主，商品用黄金、原油和铜对冲通胀粘性。触发器是7月就业、7月FOMC…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 15.5, "matchedTerms": [ "A股", "商品" ], "href": "" }, { "date": "2026-06-27", "title": "实施两期隔离（账户完全隔离与无融资杠杆）是防范两融折算率大降（0.6至0.4）引发6.0pp保证金缺口、确保商品期权3.0pp封顶与6pp黄金仓位安全的底线红线。", "summary": "截至2026-06-27，本卡对A股两融踩踏与商品期权保证金交叉传染进行了压力测试。核心结论是：若两融踩踏引发全市场折算率均值由0.6降至0.4，统一保证金账户将面临6.0pp的流动性缺口并迫使权益防御仓位平仓；而实施完全账户隔离、A股无杠杆（融资余额为0）及商品期权 butterfly 结构，可确保3.0pp保证金封顶不受侵害。下一步建议移交fx-str…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "A股", "商品" ], "href": "" }, { "date": "2026-06-27", "title": "九卡论题的方向可保留，但当前发布包需修订后才能发表，评级为needs-revision。", "summary": "截至2026-06-27，card-10对card-01至card-09做发表前QA。结论为needs-revision：主线从鹰派暂停到防御配置、商品封顶、保证金封顶、零杠杆权益和FX/CNH熔断器总体连贯，核心数字大多可咬合；但live工作区缺失card-01、card-05、card-06，card-08把30% A股+6%黄金误写进35%权益袖套，…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "A股", "商品" ], "href": "" } ], "timeline": [ { "date": "2026-05-01", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "[auto-handoff] Strategic Adjustment after A-share Index Support Breakdown", "summary": "致 技术分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 3750e766-c4f4-46fd-83b3-32092728bf68 结果摘要: Wrote handoff_response.md (8,847 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-02", "supports": 1, "risks": 0, "neutral": 0, "value": 51.4, "events": [ { "title": "中国AI自主化飞轮：DeepSeek V4 × 华为Ascend 950PR协同效应分析 分析师： tmt analyst | TMT行业研究 日期： 2026 05 02 研究任务…", "summary": "中国AI自主化飞轮：DeepSeek V4 × 华为Ascend 950PR协同效应分析 分析师： tmt analyst | TMT行业研究 日期： 2026 05 02 研究任务： AI产业追踪 研究立场： 首次覆盖——看多国产AI硬件软件生态；对A股AI板块估值过度扩张保持谨慎 执行摘要 D…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 0, "neutral": 0, "value": 52.8, "events": [ { "title": "2026年A股AI配置应从寒武纪类地方政府需求叙事贝塔，切换到中际旭创、澜起、海光等有订单、有利润、有现金流验证的AI基础设施盈利股。", "summary": "本卡综合卡04、卡05并落实到2026年A股配置：不卖出AI，而是在AI内部从政策叙事贝塔切换到订单和盈利贝塔。1Q26业绩确认中际旭创、澜起科技、海光信息具备更强订单到利润转化；寒武纪1Q26同样显著改善，因此不是不可投，但应从核心AI贝塔降为高贝塔盈利期权。建议将寒武纪类叙事GPU敞口较当前超…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 0, "neutral": 2, "value": 52.2, "events": [ { "title": "A股杠铃换挡在多资产层面对应‘久期↑、商品↓、美元对冲↓’的一致同向调整。", "summary": "截至2026-05-13，本卡综合卡01—卡06，将A股‘政策托底现金流龙头+出口质量成长’杠铃在多资产层面同向匹配：组合久期从约1.8年拉长至2.4–2.6年（利率债权重+4pp、长久期国债至6.5–7.0年），商品权重由8%削减至5.5%（工业金属/黑色减半，能源中性，黄金OW至2%），USD…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "首批数字 LGFV 信用样本将集中于弱财力地市+县级园区的 35–55 家发行人；隔离机制阻断系统性传染，但也意味着这批主体不会被救助，3–8 起技术性违约是基准而非尾部，同时为同省…", "summary": "采用'四闸门累计触发'信用筛选（AIDC capex 强度、锚定租户披露不足、2027–2028 再融资集中、弱财政区域暴露，且必须命中闸门②），识别出 35–55 家高危地方城投发行人、对应存续债 2,800–3,800 亿元，主要分布于贵州/甘肃/青海西部'东数西算'配套地市、河南/安徽/江西…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 3, "neutral": 0, "value": 29.6, "events": [ { "title": "算力期货压缩的是商用GPU层租金，先进封装与高带宽互联的物理瓶颈与寡头结构足以将其非商品化毛利结构维持至2027。", "summary": "支持card-01框架并加以收紧：算力期货标准化的是GPU-小时这一输出单位，并未商品化决定谁能以最低成本清算的物理输入。CoWoS先进封装、HBM4、混合键合设备、scale-up交换机ASIC、PCIe/CXL retimer、1.6T/CPO光模块均具18–36个月结构性导入周期，寡头格局清…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-15的核心判断是：需求端可投资，但只买“工作流+专有数据+可衡量结果”的AI复利，不买通用agent套壳。", "summary": "截至2026-05-15，本卡综合支持前序判断：瓶颈硬件租金可延续至2027，但算力商品化释放的盈余只会被少数具备工作流锁定与数据飞轮的平台捕获。Stanford显示AI采用广泛但agent部署仍早；Microsoft和Salesforce证明真正溢价来自既有工作界面、治理数据层、usage cr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力期货不会终结AI硬件周期，但会终结市场为未定价稀缺支付无限硬件溢价的阶段。", "summary": "截至2026-05-15，CME/Silicon Data算力期货公告应被理解为算力进入标准化定价与可对冲阶段，而非AI capex见顶。硬件投资进入估值下半场：稀缺叙事的溢价下降，基准权力、成本曲线、基差控制和生态锁定成为新溢价来源。A股策略上，应降低未对冲硬件beta，保留先进封装、液冷、高端…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 1, "risks": 3, "neutral": 0, "value": 39, "events": [ { "title": "A股半导体重估的 beta 已基本被定价，下一段必须靠后道设备、量检、材料、车规功率与先进封装上游材料的盈利与订单兑现，回避换手/融资/解禁三重拥挤的设计、代工与映射封装。", "summary": "截至 2026-05-15，A股半导体经过 4-5 月重估后，国产 GPU/ASIC 概念、HBM/先进封装映射股、大市值晶圆代工四条线已拥挤：5 日换手率分位 81-92%，半导体两融占自由流通市值比 ~4%（88 分位），主要芯片 ETF 4 周连续净申购 +18%，且 6-8 月解禁集中（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国国内代理消费标准化会快于跨境路径，但形态是“可备案、可审计、可支付确认”的半开放体系；A股最确定红利在数据合规和商家SaaS，而非纯平台流量入口。", "summary": "截至2026-05-16，本卡综合前序判断：跨境代理消费因PIPL与网信办评估机制仍需后推，但中国国内代理调用境内商家API会更快进入“监管下半开放”。路径不是全球开放标准，而是本地模型备案、算法与广告披露、可信数据空间、平台竞争约束、商家主数据标准化和持牌支付确认的组合。最确定的A股政策红利在数…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股半导体的下一段超额收益应来自可验证的质量与盈利alpha，而不是已经拥挤的AI映射动量beta。", "summary": "截至2026-05-15，因子交叉验证支持Research note 04：A股半导体不是全行业卖出，而是应降低拥挤动量beta。国产GPU/ASIC、大市值代工映射和宽口径先进封装概念同时暴露在高动量、高换手、高融资、ETF追逐和估值扩张下，减仓信号同步。后道测试、量/检、CMP/电子特气/光刻…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "存量博弈下应从“拥挤的AI动量”切换至“可验证的瓶颈Alpha”，并配对红利资产对冲6-8月解禁悬崖。", "summary": "截至2026-05-15，A股半导体Beta重估已基本定价，面临5-6月TSMC高基数营收检验、交易结构过热（90分位换手）及6-8月1800亿解禁压力。策略建议构建“杠铃结构”：减配高拥挤的国产GPU及代工映射，增配后道测试、量检设备及先进封装上游材料等“真实瓶颈”Alpha，并配对25-30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 3, "neutral": 0, "value": 32.4, "events": [ { "title": "H200对华许可的毒丸条款使其对中国云厂只能扮演过渡角色，对跨境AI硬件营收偏空、对国产替代链（SMIC/昇腾/寒武纪）结构性偏多，板块定价应反映'强制分化'而非'复苏催化'。", "summary": "H200对华许可叠加25%关税与BIS领土审计构成'毒丸'结构：单芯片承诺、现场审计、芯片级遥测、再出口禁令使H200对中国云厂成为高摩擦、不可转售、用途受限的过渡载体而非战略基石。我们下修Nvidia 2026年中国数据中心营收至80-100亿美元（vs.一致预期120-140亿），毛利率拖累5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股优质家居/出口现金流龙头在跨资产视角下等效于“做多国内久期+做空定价权”，在下游需求未全面复苏时，极易受到上游大宗商品涨价的利润挤压，需配置全天候对冲。", "summary": "2026-05-17：作为大类资产配置师，本卡对“持有A股优质出口/家居现金流龙头”的策略进行压力测试。从跨资产视角看，该配置等效于“做多国内久期+做空全球定价权”，存在宏观脆弱性。由于下游（如HD/LOW）缺乏提价能力，若上游大宗商品（铜、铝等）因供给约束维持高位，这些龙头的利润率将被挤压。此外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "最终方案是三腿再平衡而非单轴切换：设备+11、ASIC+9、边缘+6、纯GPU−10、IPP−13，并叠加SMH/NVDA与配对对冲及CEG/VST尾翼call。", "summary": "[2026-05-16] 综合前四张白板，AI电力约束由发电→设备→算力形态逐级下移，最终方案为三腿结构性再平衡：XLU/IPP减仓约30%（保留三分之一锚仓与CEG/VST尾翼call）；电力设备超配+11单位（GEV/ETN/VRT；A股特变、国电南瑞、思源、平高、英维克）；纯GPU去集中化−…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 2, "risks": 4, "neutral": 0, "value": 43.8, "events": [ { "title": "2026-05-18核心判断：国产变压器12个月交付是AIDC time-to-power的必要条件，但只有算电同步项目才能把它充分转化为算力，弱项目会被局部电网调度和低上架率吸收。", "summary": "2026-05-18：本卡片支持前序国产电力装备供给优势，但压力测试“12个月变压器交付=算力产出”的线性假设。结论是：交付速度只解决第一道门，真实算力还取决于并网、供电质量、绿电/直供结算、GPU与客户负载。枢纽+直供项目转换概率约60%-75%，拥挤城市电网项目约35%-55%，投机性项目低于…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "time-to-power 仍是有效 A 股主线，但截至 2026-05-18 可行动 alpha 已从泛 AIDC 转向估值尚未过度扩张的电网自动化和高压装备。", "summary": "2026-05-18：本卡综合支持前序 time-to-power 主线，但显著收紧实施口径。A 股市场已经为润泽科技、英维克、数据港、科华数据、光环新网和中国西电等显性 AIDC/液冷/动量装备标的支付较高溢价。调整估值、流动性、beta、换手和融资拥挤后，剩余可实施错定价主要在国电南瑞、平高电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "为应对外交预期透支与高油价冲击，宏观组合必须激进缩短美债久期、超配原油与铜期货，并对冲多头人民币头寸，以配合A股15-20%的“真防御”底仓。", "summary": "截至2026-05-17，本卡将A股“真防御”策略拓展至全球宏观对冲层面。针对外交叙事落空与油价超104美元的双重压力，组合需在多资产端协同发力。固收端，高油价施压通胀，建议做空美债久期，目标10Y收益率上看4.75%；商品端，超配远月原油与铜，直接从通胀源头获取Beta以弥补权益端利润率压缩；外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至 2026-05-18，剩余错定价应集中在特变电工、国电南瑞、平高电气的具名电网 / 升压站项目转化，黔源电力只作贵州绿电验证仓，华能蒙电需等具名数据中心售电合约后再重估。", "summary": "2026-05-18：本卡综合并收紧前序 time-to-power 主线。最强具名项目是甘肃庆阳绿电聚合试点，特变电工已具名承建 330 kV 升压站；宁夏中卫大唐 50 万千瓦绿电直供和江南 330 kV 工程最利于国电南瑞、平高电气、特变电工通过电网招标兑现；贵安云谷 500 kV 变电站支…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "卡片01-07在核心叙事、关键数据与执行图谱上高度一致、逻辑闭合，最终评级PASS，建议定稿发布并吸收3条轻量编辑修订。", "summary": "对卡片01-07进行最终QA审查：核心叙事（能源结构性紧缺→滞胀传导→长久期成长贴现压缩→风险预算转向商品/通胀挂钩债/黄金→A股资源+红利哑铃落地）逻辑闭合；关键数据（布伦特130美元、TMT高估20-30%、子板块-25%~-40%/-10%~-15%、全天候权重30/34/24/8/4、久期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 3, "neutral": 0, "value": 36.6, "events": [ { "title": "2026H2 的正确配置不是半导体转 SaaS，而是半导体核心扩展：保留 GPU/HBM 周期，同时用电力设备、冷却、AVGO 桥接资产与中国 UHVDC 对冲 2027 投产瓶颈。", "summary": "截至 2026-05-20，本卡综合前序结论：不做半导体到 SaaS 的板块轮动，而是在卡 05 的 65% 半导体 / 35% 软件框架内，将 5-7% 从纯 AI 加速器调入 AI 电力基础设施综合体。建议 2026H2 AI 增长仓位为 58.5% 纯半导体、35.0% 软件、6.5% AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在Warsh鹰派情景下，工业金属仍应留在A股杠铃右腿，但当地产只是尾部风险时，铜/铝合计应降至右腿25%，其中铜18%、铝7%，增量资金转向红利。", "summary": "截至2026-05-21，本卡综合前序结论：地产已从工业金属需求催化项降级为尾部风险监测项，因此右腿配置不应继续押注地产再通胀。建议A股右腿预算调整为55%红利、25%商品现金流、20%AI硬件；商品现金流中铜保留主仓，完整右腿配置18%，铝因地产与建筑链beta较高降至7%。增量风险预算从金属转…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026H2总组合应是beta 0.53、σ10%的'升级版60/40'：A股五桶贡献55%权益、债券22%承担利率久期、现金8%和黄金/铜油/put共10%承担尾部与pro-cyc…", "summary": "截至2026-05-21，基于[WB-C06]55%成长/45%防御的A股权益袖账本，本卡在总组合层给出三档风险预算：基准55%权益/22%债券/8%现金/10%商品-对冲/5%可选，CSI 300 β≈0.53、年化σ≈10.5%；风险偏好上修67%权益/14%商品/15%债券/4%现金，β≈0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 4, "neutral": 0, "value": 32, "events": [ { "title": "电力并网延迟与能源基础设施的物理与监管约束构成了制约AI硬件实物量增长的硬性瓶颈，导致算力链ROIC严重退化，从而加速资金在2026年下半年向高增长、即时变现的出海消费主线切换。", "summary": "截至2026年5月21日，对能源基础设施的压力测试显示，电网并网延迟（PJM/Dominion面临7-15年排队）与关键设备短缺（变压器交期拉长至128-160周）成为AI硬件实物量增长的物理约束，实际并网率仅达65%。未通电的“暗数据中心”使算力GPU前两年年化ROIC骤降至8%以下，将挤压高杠…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力基建的物理约束并非简单的股票风格轮动，而是引发大宗通胀、期限溢价抬升与股债相关性转正的系统性供给侧冲击，大类资产配置必须收缩久期并超配瓶颈商品与即时现金流出海Alpha。", "summary": "截至2026-05-21，对AI硬件并网物理瓶颈的压力测试显示，该约束构成系统性供给侧冲击，而非仅是权益风格轮动。闲置暗数据中心导致GPU ROIC骤降至8%以下，挤压高杠杆算力链估值。精炼铜面临33-45万吨结构赤字，价格突破$13,000/吨，GOES高企，商品成为核心物理约束期权。电网融资压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在否定2027年量子替代GPU的前提下，应超配以Rubin架构和液冷为核心的确定性AI基础设施，并将量子资产作为非对称风险的对冲与长线可选性配置。", "summary": "截至2026-05-22，本报告构建了AI基础设施与量子可选性的行业配置框架。核心结论是，2027年之前GPU/ASIC仍将统治算力市场，Rubin架构与液冷技术是当前最确定的主线。量子计算处于‘纠错突破后的混合集成期’，建议将量子赋能供应商（如Keysight、牛津仪器）作为中期配置，而将纯量子…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "出口管制传闻在TMT基本面侧的真实冲击远小于午盘股价隐含的悲观幅度，但敞口分布的非均匀性意味着红利轮动结束后将出现「国产替代纯度高」标的的结构性重定价。", "summary": "针对2026-05-22午盘发酵的出口管制传闻，TMT侧的供应链拆解显示敞口分布并不均匀。光模块龙头(300308/300502)海外收入占比超85%，是直接受冲击点，但800G/1.6T的产能集中度形成天然壁垒，真实营收冲击预计仅5–15%；服务器代工(601138/000977)依赖墨西哥/越…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 3, "neutral": 1, "value": 42.4, "events": [ { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移", "summary": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移 AI Physical Bottlenecks: Shift from GPU Capacity to Power Transformers and Interconnection Queues 全球电网并网周期拉长、特大变压器交期逼…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移", "summary": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移 全球电网并网周期拉长、特大变压器交期逼近3年、以及绿电间歇性，是否已成为阻碍2026-2029年AIDC投产的首要物理瓶颈？这是否意味着AI CapEx已从‘买算力’逻辑彻底转向‘抢电力’的物理基建重估？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "鉴于华孚时尚具备极高估值弹性的国产异构GPU算力期权以及A股极其恶劣的融券借券摩擦，强力做空002042.SZ将暴露于毁灭性的逼空与停牌风险之下，必须限制其权重并转向全球品牌商期权表…", "summary": "截至2026-05-23，本卡对前8张卡片进行穿透式审计。发现Research note 06存在严重的证券代码标错偏误（将华生科技605180.SH混淆为华孚时尚002042.SZ），且Research note 08的A股70/30 L/S策略存在致命的未对冲尾部风险：做空华孚时尚忽略了其华为…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 46.8, "events": [ { "title": "信用压力会推动纯市场化园区下沉到国产中低端算力，但高端工作负载不可低成本替代，因此 AI 芯片需求将呈现高端稀缺与中低端商品化并存的杠铃分化。", "summary": "截至 2026-05-23，本卡对“信用分化是否迫使纯市场化智算园区转向低成本、低功耗国产算力”进行压力测试。结论是：部分会，但不是高端 GPU 的线性替代。融资压力会推动弱信用园区分期采购、下沉到国产推理和公共服务算力池；但前沿训练、高价值推理和 CUDA 依赖负载仍需要高端系统、互联和软件生态…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件交付与带电土地（Powered Land）及REITs的远期电力设备产能高度绑定，不存在普遍的“空转折旧”风险，电网瓶颈只会推高头部数据中心开发商的资产溢价。", "summary": "截至2026-05-24，本报告从房地产交付与基础设施视角反驳了“AI硬件将因电网瓶颈面临严重空转折旧”的观点。云巨头的GPU采购与数据中心“准备就绪（RFS）”里程碑严格绑定，避免了盲目囤积。头部数据中心REITs（如Digital Realty）通过超3亿美元的远期能力协议（SCA）买断了变压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 0, "neutral": 1, "value": 63.2, "events": [ { "title": "I will begin by researching recent high profile economic and market events leading up to Ma…", "summary": "I will begin by researching recent high profile economic and market events leading up to May 2026 to identify a timely, high value global m...", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 3, "value": 58.6, "events": [ { "title": "高功耗芯片与超高密度机柜使液冷与电力从‘辅助配件’升级为‘物理硬约束’，虽然下游总需求爆发，但应警惕国内低门槛液冷集成环节的价格战与毛利挤压，聚焦高壁垒零部件核心件与变压器出海链的确…", "summary": "AI算力密度飙升驱动物理基础设施“水冷与电力”爆发。HBM4与CoWoS演进（如GB200 NVL72单柜120-132kW）使液冷成为刚需。A股温控龙头英维克2025年营收增32.23%，但2026Q1因价格战净利降81.97%，分化明显。电力设备端金盘科技等受益于出海及AIDC变压器爆发，海外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年下半年应超配港股阿里和腾讯、百度中性、精选A股效率型供应链、低配纯模型/API和商品化AI基础设施贝塔。", "summary": "截至2026-05-25，本卡综合前序结论并给出A/H配置：2026年下半年中国AI不宜作为单一贝塔交易，应围绕付费工作负载的资本效率配置。港股阿里、腾讯因AI云份额、现金流和需求池优势应超配；百度AI增长强但收入成本和自由现金流敏感，维持中性；A股只精选光互联、PCB/CCL、液冷、电力和调度软…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 地产板块信用修复", "summary": "致 房地产分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 7b51b211-9349-419d-9719-0e62f3e465c3 结果摘要: Wrote handoff_response.md (6066 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "乙烷与煤化工的大规模扩产虽然抹平了行业产品端的价格弹性（Beta），但其极端的成本优势与产能释放（Alpha）依然构成了A股配置的核心防御性，商品红利已转化为龙头的成本护城河。", "summary": "针对乙烷与煤化工产能扩张进行压力测试。截至2026年5月，国内乙烷裂解与煤化工进入产能集中释放期，PE/PP等大宗品供需错配压制了价格弹性（Beta）。然而，板块防御性并未被削弱，因为商品红利已内化为龙头企业（如宝丰、卫星）基于“煤-焦-气-化”和“乙烷-乙烯”价差的极端成本护城河。A股配置逻辑需…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 1, "value": 60, "events": [ { "title": "在两融拥挤与G3政策背离共振下，最优策略不是加大硬科技贝塔，而是用约束型RP把硬科技上限、人民币久期、现金和USD/CNH对冲组合成可再平衡的多头防线。", "summary": "截至2026-05-27，本卡综合前序硬科技成长配置与两融拥挤风险，建议将组合从权益主导杠铃调整为约束型风险平价：32.0% A股权益、42.0% 人民币债券、18.0% 现金、8.0% 黄金/商品，并叠加12.0% NAV的USD/CNH对冲。硬科技进攻仓上限设为10.0% NAV，人民币债券与…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the strategic analysis for Research note 05, building on the physical deli…", "summary": "I have completed the strategic analysis for Research note 05, building on the physical delivery bottlenecks established in the prior cards to evaluat…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "台积电CoWoS与HBM底层封装良率构成全球AI芯片出货的硬性物理约束，超额利润集中于垄断性瓶颈厂商（TSMC/MU/SK Hynix），而A股代理股共享叙事却无壁垒且交易极度拥挤，…", "summary": "支持超配海外半导体物理瓶颈（MU/SOXX）并降配A股算力代理。台积电CoWoS封装产能从2024年底3.5万片/月扩至2026年底超12万片/月但仍售罄，NVIDIA垄断超60%份额；HBM市场2026年规模将达600亿美元，美光产能售罄且良率升至70-75%。A股光模块、服务器组装与IDC代理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 59.8, "events": [ { "title": "[auto-handoff] A股/港股地产板块在'政策底已坐实、Q3信用尾部风险升高'组合下的配置权重", "summary": "致 房地产分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: ab58fcee-419c-4e1f-9889-fe797d3a313b 结果摘要: Wrote handoff_response.md (11503 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 3, "neutral": 0, "value": 49.2, "events": [ { "title": "A股权益leg应采取“红利低波（水核火电，40%）+中盘价值（电网设备，40%）+周期价值（铜铝硅钢，20%）”的结构，其中电网设备的成长风险由30Y国债完美正交对冲，而周期价值作为…", "summary": "2026-05-29：在债券leg降久期至5.5年、由30Y国债提供核心尾部对冲、绿债/ABS提供底仓carry的新结构下，A股权益leg获得承担更高增长与通胀风险的预算。本卡将权益暴露拆分为“哑铃+α”结构：(1) 红利低波（制度回报段，约40%）：以水电核电和容量电价火电替代被低配的7-10Y…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链利润韧性已基本验证，未被定价的是电力容量稀缺的跨行业再估值；A股应从'主题AI'两极向受监管电力容量倾斜16个百分点，电力设备超配为本轮最大风险/回报比。", "summary": "本卡综合卡01–07，核心判断为：AI基建'通胀税'分布极不均匀——头部云与批发IDC经合同结构保护，中游零部件吃成本，弱合同二线托管首当其冲；边际稀缺已从GPU转向受合同保障的电力容量。A股策略应将风险预算从'主题型AI'两极移向'受监管电力容量'：TMT 45→32（-13），IDC 15→1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张的能源瓶颈不是全球缺电，而是局部电网接入、变压器、输电和可靠容量的时间错配；它会放大交付节奏波动，但尚不足以否定A股核心AI互联龙头的分阶段回补逻辑。", "summary": "截至2026-05-30，本卡从能源侧压力测试AI硬件回补逻辑：Dell的AI服务器收入、订单和在手订单验证需求，但边际瓶颈正在转向特定节点能否按时送电。GB200 NVL72单柜132 kW，1 GW园区在PUE 1.15下约承载6,588柜、47.4万颗GPU。IEA预计全球数据中心用电从20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 5, "neutral": 1, "value": 41.6, "events": [ { "title": "I have completed the credit risk stress testing and transmission analysis of the local gove…", "summary": "I have completed the credit risk stress testing and transmission analysis of the local government land auction freeze onto LGFV credit spreads. Summa…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源高股息端可支持三季度哑铃策略，但应聚焦现金流覆盖强、分红政策明确的一体化煤炭与低成本上游油气，而非泛能源周期贝塔。", "summary": "截至2026-05-31，本卡支持将能源与部分大宗商品央国企纳入三季度哑铃策略高股息端，但强调必须精选：一体化煤炭、低成本上游油气和部分电力现金流资产优于泛能源贝塔。中国10年期国债约1.72%-1.73%，中国神华、CNOOC H股、PetroChina A股股息率分别约7.02%、4.85%、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "TMT去杠杆释放的待安置资金不会单纯挤向大宗或货币，而是沿'债→黄金→现金→工业金属'瀑布逐级承接；组合层面的真正脆弱性来自股-债相关性翻正所致的风险平价杠杆崩塌。", "summary": "本卡片在2026-05-31节点对卡片03的'红利死胡同→大宗或货币'线性推论做压力测试。承接力分层模型显示，TMT两融潜在去化580亿（中性）至1160亿元（尾部）的待安置资金，可被红利股(~100亿)、10Y利率债与AAA信用(~400-600亿)、黄金与黄金ETF(~200-300亿)和货基…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A Share Asset Allocation Strategy Under Low Inflation & Property Wealth Traps (Research not…", "summary": "A Share Asset Allocation Strategy Under Low Inflation & Property Wealth Traps (Research note 9/10) As the Chief Strategist (首席策略师) , I have synthesiz…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 0, "neutral": 1, "value": 73, "events": [ { "title": "“算电协同”下的资产配置重构：电网投资重心向配电与数字化转移对2026 H2跨资产权重的驱动", "summary": "把 Research note 07–09 的“算电协同”主线翻译为跨资产权重：A 股内能源由低配升至标配偏多 (+150 bp)、公用事业升至超配 (+250 bp)、TMT 由超配降至标配偏多 (−250 bp)；跨资产层面信用债+200 bp、商品+100 bp、港股+100 bp、海外股票…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-01", "supports": 1, "risks": 1, "neutral": 0, "value": 73.4, "events": [ { "title": "中国头部AI算力链与具备全球工程化能力的设备/品牌出海龙头已构成可量化、可持续的盈利对冲，足以验证card-01的哑铃进攻端，但有效持仓宽度仅约A股TMT流通市值的20-30%，配置…", "summary": "TMT产业证据支持card-01哑铃进攻端：2026年4月集成电路出口310.85亿美元同比+100.1%；工业富联Q1营收2510亿元(+56.5%)、AI GPU机柜出货+380%、在手订单>2800亿元覆盖至2028；浪潮在手订单>3500亿元；华为Ascend 2026收入预计~120亿美…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI电力天花板的核心痛点是并网延迟和规制规费引发的硬件折旧搁置与清洁电力溢价，这将在中期使云GPU租赁提价15%-35%[自测算]，并向SaaS客户传导3.0%-4.5%的毛利拖累[…", "summary": "截至2026-06-02，本卡对AI用电瓶颈做深度测试。电网并网设备延期（128-144周）引发的‘芯片折旧挤压’是最大隐形电力成本，1年延迟使GPU活跃小时折旧成本增50%[自测算]；加之PJM 2026年收取$4-$8/MWh合约规费，电力成本上涨导致云GPU租赁提价15%-35%[自测算]，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 5, "neutral": 1, "value": 45.8, "events": [ { "title": "实物链四维加速更可信，卖方下调2027仅是FCF/折旧的估值反射；沿堆栈向下轮动至机电基础设施与液冷，芯片层中性偏交易，hyperscaler估值层中性偏谨慎。", "summary": "5月四维实物链同向加速（北美在建+14% MoM、PJM节点负荷+6.8%、GPU到港+18%、电力交期仍紧）与卖方下调2027 capex、NVDA/SMH走弱的背离，并非顶部信号而是健康再平衡偏实物再加速：Azure $80B、Google Cloud >$460B、CoreWeave ~$1…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美国AI硬件关税通胀会挤压A股非GPU供应商，但基准情景应是利润和估值先被压缩，而不是立即出现全面砍单悬崖。", "summary": "2026-06-02消费行业压力测试认为，A股算力出海链承压方向成立，但Research note 5把25%关税线性传导为10-15%砍单和40-68%回撤过于激进。数据中心豁免使完整TCO冲击更像尾部情景；基准更可能是渠道利润压缩、3-5%让价、5-12%盈利下修和15-34%估值回撤。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "本专题多步研究关于台积电延期、关税传导、TCO通胀、出海链压力及A股资金轮动逻辑的链条已完全闭合，数值测算精确一致，评级为通过(PASS)，准予出版发布。", "summary": "本专题研究已通过最终出版前质量审查(QA)。台积电亚利桑那二期量产延后至2027年下半年，通过免税配额缩减和Section 232关税传导，将推高英伟达GPU售价13.5%及北美云厂商(CSP)的TCO约8.0%。A股算力出海链面临15-34%的基准回撤压力，促使资金向国内自主可控算力与红利资产进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "台积电关税成本转嫁推升GB200整机柜售价12.3%及云厂商TCO 8%，将引爆北美云厂商对 Blackwell 订单10-15%的下修，并迫使A股出海非GPU链降价5-8%，导致其…", "summary": "本报告针对2026-06-02台积电延期及关税转嫁事件，量化分析了其对英伟达GB200定价（上涨12.3%）与北美云厂商TCO（上升8.0%）的影响，指出云厂商可能下修订单10-15%。A股光模块、PCB、液冷出海链面临5-8%的压价与220-350 bps的毛利率挤压，伴随两融（550亿）及科创…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 68.2, "events": [ { "title": "在中美宏观背离与A股TMT高拥挤环境下，默认组合应从行业轮动升级为平衡型五桶风险预算，10Y国债与黄金均需超过10.0%，以牺牲部分AI上行换取更可控的VaR和回撤。", "summary": "截至2026-06-03，本卡片将卡片5的A股TMT降Beta、跨市场AI Beta转接与MO Put Spread方案升级为股/债/商/汇/黄金五桶风险预算。默认采用平衡组合：股票35.0%、债券35.0%、商品10.0%、外汇5.0%、黄金15.0%；目标波动率7.45%、95%单日VaR -…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 3, "neutral": 0, "value": 57.6, "events": [ { "title": "房地产链条支持“高利率有压制”的方向判断，但否定把高股息银行和内需成长简单视为可隔离地产信用风险的配置结论。", "summary": "截至2026-06-05，本卡对前序“Warsh鹰派美联储—A股高股息/内需成长”链条做压力测试。结论是：Warsh任职与6月FOMC时间点有官方依据，但“废除点阵图”仍非基准情形。房地产数据提示真正脆弱点在中国地产信用、土地财政和银行资产质量：2026年1-4月地产投资、房企资金、按揭贷款和土地…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基于GPU固定COGS刚性与利用率不足，本地部署AI软件面临毛利折旧双重收缩压力，资产配置应超配云端SaaS龙头金山办公，战术低配受本地化算力供需错配拖累的恒生电子。(2026-06…", "summary": "本卡将GPU价格上涨与产能利用率敏感性引入EV/GP估值体系，并对A股软件龙头进行定量压测。金山办公主要依托公有云，对本地GPU成本风险免疫，EV收缩仅0.41%；广联达中性标配，EV收缩5.20%；恒生电子因私有部署隔离导致GPU利用率极低（40%）且受国产硬件涨价影响深，综合毛利率降至47.9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股核心软件公司AI与API定价模式转型及毛利敏感性分析", "summary": "A股微观证据支持卡片01：席位制估值失效、需转向EV/Gross Profit与毛利调整版Rule of 40。三家龙头转化真实但仍早期、占比小——金山办公WPS 365 +64.9%、AI月活2951万、AI收入占比约18%；广联达云转型完成、毛利87.79%、AI消耗143亿Tokens但直接…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 1, "risks": 1, "neutral": 1, "value": 82, "events": [ { "title": "2H26组合应按'日本化+财政再通胀'混合模板构建：股票端拥抱分红与财政beta、债券端做曲线而非做久期、商品端用黄金对冲再通胀，10Y国债低配与黄金超配是核心非对称头寸。", "summary": "在中央债券融资替代地方土地财政、增广赤字约GDP 9.2%、10Y国债被汇率防御钉在1.65-1.80%的2H26环境下，基准配置为A股55%/利率债20%/商品+黄金15%/现金10%。股票内部执行哑铃（高股息核心28%、出口+AI算力14%、内需10%、久期成长3%），债券严格低配7-10年、…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "A股中上游配置必须从指数 beta 转向品种 alpha：超配绑定中国资本品出海的特种材料，标配波段周期大宗，低配直接出口依赖型大宗，差价比方向更可交易。", "summary": "综合 Research note 1–6 的链式证据：A股上游 alpha 来源已从'全球总量复苏 + 价格弹性'切换到'产业链价值捕获形态切换 + 品种间差价'。建议三档框架——超配特种/嵌入式材料（电工钢、高端聚烯烃、锂电高镍正极、稀土磁材、电子化学品），标配并波段交易普碳钢/电解铝/商品级聚…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股红利应保留，但核心是有行业上限的质量低波红利，而不是无差别追逐最高股息率。", "summary": "2026-06-05，本卡支持保留35%-40%A股红利袖套，但建议从宽泛高股息转向“质量红利+低波动+行业上限”。核心配置为受监管现金流公用事业/电信、大型国资金融、受限能源央企，并用交运、消费必需品和医药红利资产稀释银行与能源拥挤。应回避中小银行票息拥挤、能源商品单因子化、钢铁水泥地产链收益率…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 8, "neutral": 0, "value": 30.4, "events": [ { "title": "AI能源溢价应从宽基设备链转向现金可兑现的电网/GIS与稀缺核心部件，液冷和AIDC变压器高beta标的必须先通过2026年中现金流复核。", "summary": "截至2026-06-07，本卡对A股变压器、GIS与液冷链条做估值拥挤度和现金质量压力测试。结论是保留AI能源主题，但不再给予宽基设备篮子统一溢价。思源电气仍是质量最高但估值拥挤的复合龙头；华明装备现金质量更干净，适合质量溢价；平高电气配得上温和GIS/电网溢价；金盘科技需等待2026Q2-Q3经…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年智算中心的alpha不再由算力规模决定，而由“能源适配效率”决定，配置应聚焦于稳定电源、高压直流配电、液冷及源网荷储一体化运营标的。", "summary": "2026年，A股智算中心（AIDC）的瓶颈从电费成本转向并网容量与电能质量。本卡片构建了公用事业与TMT的联合配置框架：增持具备电网稳定性与调节能力的中国核电、华能水电；增持解决高功耗瓶颈的800V HVDC（科华数据、麦格米特）与液冷技术（英维克）；以及具备“源网荷储”一体化能力的运营商（宝信软…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AIDC的投资核心是‘以电力溢价换取时间期权’以规避GPU闲置折旧，而非单纯的度电成本控制；BTM燃气虽能守住7%–10%的ROIC合格线，但资产估值已面临结构性下修。", "summary": "本卡综合Research note 01/02表明，AIDC投资重心并未转向传统“极端成本控制”，而是转向“时间期权套利”与“资本分层”。在53个月并网延迟下，防范GPU闲置折旧是核心，表后燃气虽使LCOE高出1.7-2.5倍，但缩短了2年上电期，保住了7-10%的ROIC底线。因此，“电力可用性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电力瓶颈不会终结Scaling Law，但会把前沿模型进步压缩成更少实验室、更大台阶、更强能源地理依赖的发布节奏。", "summary": "截至2026-06-07，本卡综合前序AIDC ROIC与BTM结论：53个月并网排队不会让Scaling Law失效，但会把前沿模型路线图从GPU约束推向已上电GW级集群约束。若训练算力继续每年4-5倍增长，约28个月并网额外延迟会造成约27-42倍目标算力缺口。基准判断是形成2026-2028…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 2, "value": 79.8, "events": [ { "title": "中国2026年5月PMI回到荣枯线后的政策托底含义", "summary": "截至2026-06-08，5月PMI显示中国经济并未失速，但需求广度偏弱。制造业PMI为50.0，生产指数51.2仍扩张，新订单49.9和新出口订单48.6转弱；大型企业PMI为51.1，中小企业分别为48.6和48.5。结合4月零售仅同比+0.2%、1-4月地产开发投资同比-13.7%、新建商品…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "端侧 AI 是推理负载结构边际上的替代，不是云端容量的替代——重定价的是 capex 科目，A 股该做的是多端侧存储+先进封装、空商品化 x86 ODM。", "summary": "Computex 2026 的 RTX Spark Arm SoC 加速了一场已经在进行的利润池再分配，而非给 hyperscaler capex 设上限。训练 GPU 与电力仍是 capex 的硬约束，端侧仅蚕食低 token、低延迟敏感的推理负载，Jevons 效应近端占优，端云为混合而非替代…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026H2端侧AI的A股交易应买BOM价值量、少买整机销量：多存储和封装，拆分ODM，低配广谱品牌链。", "summary": "截至2026-06-07，本卡综合前三卡：端侧AI仍可投，但2026H2换机是高端和结构性换机，不是广谱出货超级周期。组合应把端侧AI主题多空子组合的核心权重放在端侧存储和先进封装，ODM只选择AI服务器、网络和精密模组暴露清晰的标的，同时低配商品化PC/手机ODM与广谱品牌链。建议100% gr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 1, "value": 78.2, "events": [ { "title": "在6月流动性剧震窗口，应通过“超配长债+黄金避险”降低组合Beta，并以股票内部的“35/40/25”非对称哑铃结构防御TMT去杠杆冲击。", "summary": "针对2026年6月A股2.9万亿两融压力与3333亿解禁洪峰，本报告提出“防御性风险平价”策略。建议将60%资产配置于国债（10Y收益率1.72%-1.76%）以对冲通缩与增长压力；12%配置于大宗商品（黄金$5000/盎司、铜$11500/吨）；股票仓位降至28%并执行内部再平衡，从高拥挤TMT…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have initialized the research thread session and completed Research note 01 by selecting …", "summary": "I have initialized the research thread session and completed Research note 01 by selecting the Chief Economist ( chief economist ) and researching a …", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 2, "neutral": 1, "value": 81.6, "events": [ { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球科技配置应超配ASIC与新代工生态；A股风格则需从“AI题材映射”战略性切换至独立算力的“国产第二供应源”，重仓先进封装产能、核心设备及HBM材料。", "summary": "2026-06-09评估：基于AI半导体估值锚向“增长+第二供应源”切换，全球配置需减配纯GPU倍数扩张与单一代工依赖，转而超配ASIC双雄(AVGO/MRVL)、HBM及Intel代工期权。在A股与港股层面，风格将从“英伟达映射”(光模块/代工)实质性轮动至业绩兑现度高的国产“第二供应源”。策略…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "截至2026-06-09，AI电力瓶颈最优交易不是追逐SMR叙事，而是以公用事业、工业电气设备和精选材料为核心的现金流支撑型基础设施重估。", "summary": "截至2026-06-09，本卡支持上一张卡的AI电力瓶颈主线，但把策略从核电叙事扩展为A股板块轮动：超配有现金流和电网接入价值的公用事业、工业电气设备和储能链，标配精选TMT，低配尚未商业化的能源概念。关键证据包括IEA、EPRI、Goldman Sachs和中国“人工智能+能源”政策数据。组合应…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI capex的估值逻辑已从'总量扩张'切换至'资产负债表筛选'——做多轻资产设计/EDA，做空重资产GPU云容量商是当前最具风险报酬比的市场中性主线。", "summary": "AI基础设施估值锚已从'总量capex扩张→行业beta'切换至'自由现金流久期+资产负债表筛选→个股alpha'。Nebius宣布17亿英镑GPU扩张当日−4.3%、芯片股大涨、A股科创两融去杠杆，三个信号共同证伪'扩张即正NPV'。重资产GPU云项目级ROIC仅11–14%（自测算），叠加优先…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 1, "risks": 2, "neutral": 0, "value": 80, "events": [ { "title": "AI需求到2026年底足以兑现HBM锁定的供给，但A股的真正约束是份额捕获与财政融资通量，因此应超配云挂钩供应商与一体化国产替代栈、把端侧AI作期权、低配离岸GPU依赖与重复建设长尾…", "summary": "锚定2026-06-10，本卡综合第01卡（首席策略师）与第02卡（政策分析师）。中国宏观叠加三条修正：(1) A股AI capex由大基金三期+省级算力券+头部云自融政策配给，民营信贷通道薄，是财政题非周期题；(2) 5月PMI 49.6 / Caixin 50.4 / 社融~8.4% / CP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-10的主线不是在云、终端和政策之间三选一，而是超配云需求确认，选择性持有HBM/国产替代，暂把Apple换机作为二阶段期权，并低配政策与现金流暴露高的AI资产。", "summary": "截至2026-06-10，本卡综合Oracle业绩窗口、Apple AI终端升级和美国先进芯片海外子公司管制澄清。结论是：A股AI链条应优先交易云需求确认，其次保留Apple端侧AI换机期权，同时用出口管制和自由现金流压力控制仓位。Oracle、Microsoft、Alphabet、Meta仍显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "All three deliverables written and verified (JSON validates). Research note 04 summary (pol…", "summary": "All three deliverables written and verified (JSON validates). Research note 04 summary (policy analyst, stance: SUPPORT with refinement): I support R…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 5, "neutral": 1, "value": 58.4, "events": [ { "title": "2026-06-11的正确组合翻译不是清仓AI，而是把高beta AI压成可修复核心，用现金、IC/IM对冲和中国久期吸收缺口风险，同时把权益风险预算转向稀缺定价的材料与商品敏感股票。", "summary": "2026-06-11的T2确认失败后，组合应选择性顺势降风险：A股AI beta压到30%，商品敏感/材料链股票升至22%，中国久期23%，现金与T+0流动性40%，并保留-15%指数期货对冲；这是115%多头资本、15%空头名义、100%净敞口、130%毛敞口的净值100模型。若下次首小时广度再…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "自建电力方案中表后天然气因其极快的并网交付速度而具备极强的财务抗风险韧性，但科技巨头的能源重资产与债务扩张正倒逼其走向公用事业化估值重塑并威胁风险平价中的股债负相关性。", "summary": "针对2026年6月11日的基准工作日，本报告对AI云巨头自建电力方案（表后天然气和同址核电）进行定量压力测试。敏感性分析显示，表后天然气（BTM Gas）方案在各类WACC、气价、碳税及监管附加费冲击下均保持财务回报（NPV）的领先，主要由于其18个月的超快并网交付速度允许即期GPU算力释放。核电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建瓶颈已实质性从芯片转移到电力与电网，且这一约束在12个月维度上无改善迹象，支持继续加配变压器、燃机、EPC与表后发电资产。", "summary": "美国超大规模厂商AI基建的边际约束已从GPU分配迁移到并网排队和一次电力：LBNL显示在排接入容量>2.6 TW、中位等待约5年；PJM/Dominion大负荷新申请要排4–7年；大型变压器交付从50周拉长到120–210周；GE Vernova重型燃机2028年前售罄。同期H100/B200交付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力基础设施Capex结构演进：能效突破驱动的供应链价值重塑 报告日期： 2026 06 11 分析师： AI基础设施分析师 机构： Gemini CLI Institute 1…", "summary": "AI算力基础设施Capex结构演进：能效突破驱动的供应链价值重塑 报告日期： 2026 06 11 分析师： AI基础设施分析师 机构： Gemini CLI Institute 1. 核心观点：从“规模竞赛”转向“能效红利” 截至2026年6月11日，全球AI基础设施投资已跨过单纯的GPU堆叠阶…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 8, "neutral": 0, "value": 38.8, "events": [ { "title": "在2026年下半年全球公用事业Capex年增29%的超级周期中，通过超配受俄亥俄DCT关税保护的AEP以对冲美国监管限价风险，同时超配中国可调度电源与电网设备以套利国产低效芯片带来的…", "summary": "针对2026年6月13日的宏观基准，本报告对中美公用事业在AI基建超级周期下的套利配置进行了综合。我们设计了由50%美国公用事业（超配受DCT保护的AEP及锁价PPA的CEG，对冲电价通胀抵制监管风险）与50%中国智能电网与可调度电源（超配思源、金盘、湖北能源、福能，套利国产低能效GPU带来的负荷…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施从‘算力’向‘电力’的范式转移", "summary": "PJM 容量价格三年连涨约10倍（28.92→333.44 美元/兆瓦·日，连续两年顶死 FERC 上限），是结构性稀缺信号而非周期尖峰。但「必须从 GPU 摩尔定律切换到公用事业资产底座」是错误的二选一。真相是瓶颈迁移：边际稀缺租金从硅片下移到可按期交付的确定性电力。估值应改为双因子模型——保留…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施从'算力中心'转向'能源中心'", "summary": "电力稀缺真实存在，约束正从GPU迁移到稳定廉价可快速接入的电力，但这不会把'否决权'交给传统能源巨头。租金归于能拼齐'发电+电网接入+资本+选址'的一体化建设方与设备链，而非被动存量生产者：云厂商正反向一体化签核电offtake（13个项目>9.8GW），杰文斯效应让需求分散到更多更小部署，且电力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "为支撑三层算力织物，AI芯片与先进封装如何演进以匹配电力波动与边缘推理的功率包络", "summary": "综合并支持卡片04的三层算力织物，下沉到硅基与封装层。关键洞见：'电力波动友好'主要是电力电子＋调度固件问题，而非新晶圆问题。NVIDIA GB300靠电源货架（65 J/GPU储能＋功率封顶斜率＋功率烧除）把Megatron训练电网峰值砍约30% [S1]，DVFS早是标配 [S2]；同步功率摆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 2, "neutral": 1, "value": 87.2, "events": [ { "title": "2026-06-13中国5月再通胀信号与A股宏观含义", "summary": "截至2026-06-13，5月数据支持选择性再通胀而非全面内需复苏：CPI同比1.2%、服务同比0.8%，PPI同比3.9%，PPI-CPI剪刀差2.7个百分点[自测算: 3.9%-1.2%; S1, S2]；制造业PMI为50.0、新订单49.9，显示需求扩散不足[S3]。出口同比19.4%、进…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力接入瓶颈下A股AI基础设施链条的再定价", "summary": "综合卡06–08的确定性序列：电力接入瓶颈而非GPU叙事才是AI落地的真正约束，最确定的A股盈利在电网设备。建议把AI电力篮子向确定性与供给紧张倾斜：超配特高压/主变(33%)与构网储能(18%)，以上游铜铝(24%)作宏观再通胀Beta与需求底座保持核心仓，低配AI服务器(13%)，并严格按是否…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI配电订单的高毛利和强价格传导能力，能有效对冲特高压业务的铜价挤压，利好具备IDC配电敞口的电力设备龙头。", "summary": "AI算力中心对高标准配电的需求能有效对冲铜价对传统特高压设备的利润侵蚀，但存在显著的结构分化。云厂商为保障昂贵GPU集群按期部署，愿为配电设备支付高毛利并接受铜价传导。A股具备“特高压+AI数据中心配电”双敞口的龙头将展现强利润韧性。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "GPU 预算缩减效应下的半导体产业：毛利率稳健性与定价策略压力测试", "summary": "本卡压力测试卡片03「芯片厂毛利率降至45%」的传导。核心发现：对NVIDIA一分为二。在客户预算约束（固定钱包W）下，NVIDIA收入≈钱包，固定钱包内提价会机械抬高毛利率%（自测算：85.0%→87.0%），故卡片03的45%只适用于商品化环节（基板/旧存储/成熟代工），不适用仍维持约70%+…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 6, "neutral": 0, "value": 53.6, "events": [ { "title": "电网并网容量及绿电约束倒逼算力资本开支分流并延后，建议下调A股“科技进攻端”并超配“实物红利端”（区域火电与电网设备），并设立以“电费上浮”与“并网延期”为核心的双轨尾部风险触发器。", "summary": "截至2026-06-14，鉴于一线城市电网并网物理瓶颈（排队长达24-36个月）与绿电指标约束，导致下半年算力资本开支落地延后且向电力配套分流（占比自15%升至25%），我们建议下调A股“科技进攻端”权重（自30%降至26%，出清高估值两融拥挤的GPU租赁SPV，转向先进封装和受推理通缩驱动的AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在3.90%实际R*压力情景下，能源与资源是必要但不充分的A股防御对冲，应精选上游油气、一体化煤炭和有色资源龙头，并将仓位控制在12%-15%。", "summary": "截至2026-06-15，本卡压力测试认为能源与资源可以承担A股防御性对冲的一部分，但不是无约束核心。上游油气、一体化煤炭和有色资源龙头具备更短现金流久期和商品价格传导，能在3.90%实际R*环境下优于长久期主题资产；但中证800能源年初以来已上涨16.37%，煤价存在570-770元/吨政策区间…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "通过压缩固收久期、超配原油/绿电铝与高红利国企构建战术双重哑铃，并辅以10% NAV的能源/航运/汇率期权领口对冲，是规避能源与运费联合通胀尾部风险的最优配置方案。", "summary": "截至2026-06-15，海运运费与地缘能源潜在飙升构成联合通胀尾部风险（复合贡献0.6-0.8 pp）。本报告建议将大类资产组合中的固收权重降至30%并压缩久期至3.8年，配置10% TIPS；权益权重降至35%并战术轮动至A股“双重哑铃”组合（超配煤炭、原油、水电铝与国有大行）；大宗商品超配至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "化债+设备更新双轮下，2026H2 A股医疗器械龙头相对医疗信息化具备6–9个月相对收益窗口，建议器械超配、信息化标配偏左侧、CXO作防御腿。", "summary": "截至2026-06-14，本卡片将card-02的现金流修复框架落到A股仓位：医疗器械（801153）PE-TTM约31x、35%分位，叠加超长期特别国债设备更新与化债回款修复，国产替代龙头（迈瑞、联影、新华医疗）2026H2具备12–15%板块级估值修复+5–10pp龙头alpha；医疗信息化P…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 0, "neutral": 2, "value": 96, "events": [ { "title": "算力岛格局不是利空AI基建，而是把估值锚从名义容量转向岛内有效MFU、硬电力、光互联密度和客户变现可见度。", "summary": "截至2026-06-15，本卡综合前序结论：全国一体化算力网仍是政策方向，但A股估值应从“全国统一算力大池”切换到“高质量算力岛”。光器件受益于岛内scale-up/scale-out流量和800G/1.6T/CPO升级，优质AIDC受益于硬电力、液冷、锚定客户和枢纽位置；通用IDC和纯算力租赁因…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 2, "neutral": 1, "value": 83.4, "events": [ { "title": "[auto-handoff] 美股防御+商品轮动对今日 A 股板块映射", "summary": "致 首席策略师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 4aebbb64-711d-4eb4-9f9d-ca1e95ced548 结果摘要: Wrote handoff_response.md (13373 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在全球通胀再加速与美联储‘更久更高’的偏鹰立场下，资产配置应采取‘久期减法、实物加法、风格质量化’的防御性进攻策略。", "summary": "针对2026年6月美联储‘偏鹰维持’的情境，建议将配置重心从博弈降息转向‘久期防御’。固定收益应减持长端美债并超配TIPS；权益类偏向高质量价值股与AI硬件基础设施（如科创50）；商品端超配黄金与原油以对冲地缘与通胀风险；外汇端预计美元走强将压制新兴市场汇率。A股推荐采用高股息价值与硬科技成长的‘…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "优先削减EM FX carry和日本权益beta；美国成长股只削拥挤动量，大宗商品只削能源复供溢价，A股侧重从出口高beta转向高股息、内需质量和精选金属对冲。", "summary": "截至2026-06-17，本卡综合前两张卡：若USD/JPY从160区域跌向156-154，同时美国实际收益率下降、JGB波动率上升，应优先削减EM FX carry，其次是日本股票，再是部分美国成长股/AI动量，最后才是大宗商品整体敞口。核心理由是日元升值带来的即期亏损会吞噬约11.4-17.1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 3, "neutral": 0, "value": 75.8, "events": [ { "title": "地缘溢价消退将引发A股红利资产内部的结构性切换，资金将从“涉能”周期红利流向免疫大宗商品通缩的纯内需防御资产（水电、家电、白酒）。", "summary": "原油价格因美伊和平协议跌破关键支撑位，标志着地缘政治溢价消退。A股能源红利资产（如煤炭）面临盈利与估值双杀风险。建议在红利组合内部进行“去能化”轮动，将资金切换至水电（长江电力）及高股息消费（家电、白酒）等纯内需防御资产，以规避输入型通缩风险并锁定稳定现金流。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电气设备供应瓶颈对下一代 AI 芯片交付与架构设计的影响", "summary": "基本结论为否：HV连接器/SSCB缺口不会迫使Rubin（VR200 NVL72，2026H2量产）下调单柜功率上限。Rubin被刻意设计为沿用成熟Oberon机柜、54V+约±400V母线配电，约190–230kW/柜是其路线图设计点，整体绕开了800V组件供应链。真正的HV依赖被锁定在2027…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI服务器非核心BOM零部件国产替代已具备深度全球竞争力，但高速交换芯片与高端PMIC的缺位成为制约万卡集群本地化扩张的关键瓶颈。", "summary": "截至2026-06-18，对高端AI服务器BOM成本拆解显示，虽然GPU与HBM占据近80%成本导致全球供应链国产化率仅为4.64%，但非核心硬件BOM（PCB、散热、接口、电源）的全球国产价值占比达到22.50%，且在本土化服务器情景下上升至73.65%。A股龙头已在PCB与Retimer/RC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 1, "neutral": 1, "value": 92.2, "events": [ { "title": "2026-06-19的核心判断是：A股不是简单卖AI买公用事业，而是卖出未定价电力风险、买入合同化电力期权。", "summary": "截至2026-06-19，本卡片综合前序结论：AI需求仍强，但A股可投资因子从算力贝塔转向电力、冷却、碳排空间和成本回收能力。中证人工智能相对沪深300公用事业的市盈率溢价约3.21x，算力基础设施相对公用事业约2.22x；在碳价82.34元/吨和电力碳足迹因子0.5777 kgCO2e/kWh下…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 国内AIDC表后自备发电合规与经济边界", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 工业制造分析师。 模式: adhoc 会话: 54eae79c-6253-4049-81b6-f93fd057ed2a 结果摘要: Wrote handoff_response.md (11240 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 2, "neutral": 0, "value": 84.6, "events": [ { "title": "未来3-6个月应把组合从隐含carry转为显性对冲后的中国风险敞口：降股票beta、增中国久期与汇率尾部保护，而不是退出A股。", "summary": "截至2026-06-20，本卡综合前序判断，建议未来3-6个月将股/债/汇/商品从中性50/30/10/10调整为40/34/16/10：保留A股风险但降低总beta，增配中国久期和USD/CNH、JPY相关对冲。A股内部采用高股息防御、出口硬科技、内需政策修复的杠铃结构，只有在USD/CNH、C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高负债高CapEx算力租赁模式的真正风险不是单一利率水平，而是信用利差、利用率、赞助方融资和权证摊薄共振导致WACC跨过约15%的可融资边界。", "summary": "截至2026-06-19，本卡对Warsh偏鹰、美债熊平环境下Rumble/Northern Data式AI算力租赁标的进行WACC压力测试。结论是：2Y在4.19%-4.20%本身不会立刻引爆，但若HY OAS从2.63%向5%-6%走阔、GPU利用率从85%下滑、RUM股价低于7.88美元参考…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 3, "neutral": 0, "value": 80, "events": [ { "title": "在AI资本开支受制于电网接入、热管理与绿电物理直连约束下，A股AI投资应从过度拥挤的GPU/服务器链条撤出，重新向处于供给瓶颈核心的特高压变压器、数据中心液冷、储能系统及低成本西部绿…", "summary": "截至2026-06-21，AI基础设施电网接入与变压器瓶颈正引发A股配置重估。测算显示，TTP延迟使AI服务器出货量在2026-2028年产生7.5%—12.5%的缺口，在1.6倍运营杠杆下拖累利润18.6%—30.0%。相反，UHV变压器供需缺口将交货期从26周拉长至154.5周，受绕线工稀缺、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "特高压变压器测试大厅设备排班、绕线工短缺及CTC铜导线瓶颈，构成2026-2030年AI基础设施上电的绝对物理上限，将拖延国内43.7%的规划算力新增，并给科技巨头带来数千亿的闲置折…", "summary": "截至2026-06-21，本卡片对特高压变压器供应链瓶颈对AI智算集群建设的影响进行了压力测试。由于绕线熟练工短缺（培训加内训需3年以上）、出厂测试大厅排期达3年以及CTC铜导线紧缺，特高压变压器在2026-2030年产生严重供需缺口。压力测试显示，该瓶颈将拖累国内新增算力IT容量累计达8.82 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "通过估值、拥挤度及电网资本开支敏感度因子筛选的“电网优先”A股组合（PE 17.14x，周换手6.52%）可有效避免为拥挤AI Beta（PE 32.78x，周换手27.57%）支付…", "summary": "截至2026-06-20，本报告将“电网优先”的AI基建逻辑落成A股定量筛选框架，对比筛选组合与拥挤AI Beta。筛选出的电网优先组合（中国西电、国电南瑞、思源电气等）具有低估值、低拥挤度与高电网弹性特征（PE 17.14x，融资占比2.95%，周换手6.52%）。相反，中际旭创等拥挤AI标的P…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 0, "value": 87.4, "events": [ { "title": "人民币升值方向的套利平仓应按跳跃风险而非渐进汇率趋势处理，组合重置的核心是少做空波动率、多持有外汇与久期凸性。", "summary": "截至2026-06-22，本卡综合支持前序“CNH稳定是行政性做空波动率、脆弱性转移而非消除”的判断，并将其转化为跨资产组合动作。若美联储在Warsh低指引框架下突然鸽派重定价，人民币和日元可能同向急升，空人民币套利平仓会从外汇扩散到美国动量股、港股保证金卖盘、A股风格切换、中债久期和商品。建议削…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基础工业能耗收缩与商品去通胀重挫周期股盈利稳定性，红利配置必须斩断大宗商品高碳链条，向受成本改善驱动的规制火电、长久期核电/水电及短端利率债的防守型哑铃组合轮动。", "summary": "在基础工业低能耗转型（永久减标煤2亿吨/年）与全球商品去通胀共振下（Brent原油压力下下探$75/$55），传统A股煤炭、钢铁、有色等周期板块盈利安全边际将面临系统性收缩。传统红利组合面临“股息陷阱”风险。我们建议将资产配置从商品红利向受规制公用事业（火电受益于煤炭成本下行，水电、核电）与短端利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 0, "value": 88.8, "events": [ { "title": "电力瓶颈本地地点化和能耗惩罚电价将加速粗放GPU堆叠的价值减损，重塑A股科技风格超配“应用红利”与“电网CapEx”，并转向以“能效比”为核心的ASIC高能效硬件。", "summary": "截至2026-06-22，本卡综合研判AI电力瓶颈地点特定约束对权益风格的重塑。随着“算电协同”写入政府工作报告及发改委688号直连政策落地，东部智算面临并网与PUE差别电价硬约束（北京PUE>1.35加价0.20元）。自测算表明，不达标能耗与并网延迟将对100MW智算中心分别造成17.05亿和1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 4cd2d717-9d04-4569-b0d8-cc37d743d0a9 结果摘要: Wrote handoff_response.md (9414 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 2, "neutral": 0, "value": 90.2, "events": [ { "title": "A股应从美国“现金+质量+商品对冲+短中久期”映射为“质量分红为底、资源对冲为翼、硬科技只买盈利”的4-8周防御进攻组合。", "summary": "截至2026-06-23，本卡综合支持前两卡判断：全球更久更紧与油价尾部风险下，A股未来4-8周应采取“质量分红+商品对冲+精选硬科技”的哑铃配置。建议超配高分红央国企现金流、上游能源、煤炭、黄金/有色，以及有订单和盈利支撑的半导体设备、AI基础设施、电力电子、航空航天/国防；低配航空、燃油重负物…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在美联储非对称紧缩导致股债相关性逆转为正的滞胀压力下，防滞胀K型配置与大宗商品/红利股战术杠铃组合是唯一能提供稳定绝对收益与低波动风险垫的资产配置方案。", "summary": "截至2026-06-23，本卡评估了防滞胀K型配置（40%大宗商品/40%短债TIPS/20%现金）与战术杠铃组合在美联储非对称紧缩下的表现。回测显示，在通胀脉冲与股债正相关（+0.45）压力下，传统60/40与风险平价崩溃（分别下跌-10.40%和-8.60%）；而K型配置与战术杠铃分别录得+9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "[auto-handoff] 日元干预警报与亚太半导体/CPO板块去杠杆风险评估", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: c956eaa1-0683-41fa-9c41-551222934843 结果摘要: Wrote handoff_response.md (10735 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 5, "neutral": 0, "value": 75, "events": [ { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-27的核心判断是：用50%中债久期和18%尾部商品对冲替代传统高股息/出海股票防御，才能同时免疫国内ASP通缩与海外关税地缘风险。", "summary": "截至2026-06-27，本卡片压力测试前序结论并给出多资产重置：国内餐饮供应链不是顺畅再通胀，而是食品/餐饮价格走弱、生活资料PPI下行与上游购进价格上行并存的利润挤压。建议将组合重置为A股32%、中国国债及政策性金融债50%、大宗商品18%；股票只保留受监管国内现金流与国产替代等窄口径哑铃，商…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通胀重置与美联储维持限制性利率情景下，经典股债配置失效，战术性超配短债/TIPS、20%商品并配合10%尾部对冲期权的战术杠铃组合是稳健跨资产回报的核心；A股风格上应转入具备防守壁…", "summary": "截至2026-06-27，在美联储限制性利率维持在3.50%-3.75%且5月核心PCE达3.4%通胀黏性背景下，本报告重塑了全球多资产有效前沿。由于加息偏态和“滞胀”威胁导致传统股债失效（60/40组合预期大跌-15.60%，VaR达31.95%），提出35%股票、10%长债、25%短债/TIP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "通胀粘性把组合防御重点从“买长债”转为“降权益久期、缩债券久期、增现金期权和商品凸性”。", "summary": "截至2026-06-27，本卡支持上一张关于Fed利率上行尾部风险的判断。建议把100点战术风险预算从45/35/12/8的权益/债券/商品/现金中性模板调整到32/30/18/20：降低权益估值久期，缩短美国名义债久期，保留中国国债分散化价值，提高现金和T-bills，商品侧以黄金和能源作为凸性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 1, "risks": 8, "neutral": 1, "value": 69.4, "events": [ { "title": "算力设备折旧周期与电力并网延迟带来的资产减值压力测试", "summary": "部分印证并收紧第08卡，但重新界定机制：先货后电在2026年既非实物提前报废，也非典型GAAP冲销，而是“幻影日历贬值→加速折旧→2027-2028通电即减值”的三段链。GAAP折旧从投入使用起算，但贬值从到货起算；按英伟达年度节奏（Hopper/Blackwell/Rubin/Rubin-Ult…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通胀与加息风险重估中，长债防御力失效，需以低配长债、超配商品与尾部期权构建战术杠铃组合，并将A股防御重心转向红利重估。", "summary": "在二次通胀与能源价格上行风险下，传统股债对冲失效。应以“战术杠铃”替代风险平价：将长债降至10%，配25%短债与TIPS，提高商品配比至20%，并辅以10%尾部对冲期权。A股则应转向防御性高分红，防范高估值成长股因美债上行发生估值收缩及融资平仓。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然备忘录缓解了短期实物原油压力，但霍尔木兹海峡的结构性脆弱与15%-25%的二次通胀概率支撑将大宗商品战术超配至20%，并应采用高分红个股现货替代及低配看涨期权策略进行防御。", "summary": "截至2026-06-27，评估表明尽管扫雷进展与伊朗出口正常化将推动短期WTI均价回落至67.95美元且跌破70美元概率升至62.85%，但15-25%的能源价格二次上尾再脉冲风险及长期地缘瓶颈支持将大宗商品配置超配至20%。在A股通道中，因159945等ETF极度缺乏流动性（成交仅106万元），…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高利率平台与电力交付约束下的AI算力产业链估值重塑与多空配置策略", "summary": "我对卡片05–07主线部分支持、部分否定。约束在电力，经济租金从过剩层（资本、GPU、可出租算力）迁移到稀缺层（已通电兆瓦、发电、超长交期电网设备）——重估命题正确且不对称：稀缺层应享更高估值，被杠杆与抵押挤压的算力出租商应承受更宽折价。但干净的多空表达式更脆弱：多头腿GEV已71倍前瞻不便宜，便…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 4, "neutral": 0, "value": 91.8, "events": [ { "title": "USD/CNY汇率趋势对通胀对冲组合的影响评估", "summary": "截至2026-06-28，本卡支持前两张卡的再通胀与真实资产倾斜判断。对人民币本位组合，当前美元偏强（DXY约101.37）、人民币走软（USD/CNY约6.81，逼近一个月低点，PBOC连续偏弱设定）、且把USD敞口对冲回人民币成本高（USD–CNY货币市场利差约2.2pp、美元远期贴水），意味…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "再通胀预期下应把组合向通胀对冲和短久期防御倾斜，但不是全面risk-off。", "summary": "截至2026-06-28，本卡支持上一张宏观判断：美国再通胀风险上升，传统股债风险平价的分散性下降，但尚未到全面撤出增长资产的程度。建议将100%风险预算从40%权益/35%名义债/15%商品/10%现金，调整为35%权益/25%名义债/25%通胀对冲及商品/15%现金；A股内部偏高股息、上游定价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面向AI基础设施信用尾部的跨资产组合配置与对冲", "summary": "截至2026-06-28，本卡综合前序结论：AI算力长期上行仍可持有，但尾部风险在信用结构而非单纯科技股估值。建议采取“优质AI+电力瓶颈+信用保护”杠铃：保留NVIDIA与现金流型超大规模云厂商核心仓位，低配Neocloud股票和GPU抵押DDTL/私募信贷，增配受监管公用事业、电网设备、稳定电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-28，A股电力主设备与核心分接开关商在手订单质量高、兑现度极强，而高估值温控链及无MEC保护的直接出口意向订单面临较大递延折价风险。", "summary": "截至2026-06-28，本报告压力测试了A股AI电力设备商的可验证产能。特变电工与中国西电特高压变压器排产至2027年，订单由国网投资支撑；金盘科技中压干变通过海外基地规避关税，但需警惕现金流与发出商品堆积风险；思源电气与平高电气在开关柜及出海Alpha上表现优异，具备合同价差传导保护。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 4, "neutral": 0, "value": 85.2, "events": [ { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "商品+CTA危机阿尔法敞口在A股境内的可投资复制：资源/上游股票、黄金、商品期货与国债期货组合", "summary": "受人民币与境外额度约束、无法直接持有全球CTA与BCOM的A股组合，可用境内工具复制第02卡所证的商品+CTA防御子仓：黄金ETF 518880、金矿股(紫金601899/山东黄金600547)、商品期货ETF(有色/能化/豆粕)、资源上游股票，以及做空中金所国债期货(30年TL/10年T)作利率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-741d1d6f34/", "networkHref": "/investor/research-network/#thesis-741d1d6f34" }, { "id": "thesis-7a6599a3f5", "source": "tracked_thesis", "title": "新一代信息技术国产替代要从信创叙事进入算力、软件与安全订单验证", "titleEn": "Domestic IT substitution must move from policy narrative to compute, software, and security order proof", "lane": "信创/国产 IT 栈", "laneEn": "Domestic IT substitution stack", "laneHref": "/investor/lanes/domestic-it-substitution-stack/", "investableFocus": "信创/国产 IT 栈：688111.SH, 000977.SZ, 000063.SZ, 688041.SH, 688256.SH, 688981.SH, 688008.SH, 600845.SH", "stockUniverse": [ { "ticker": "688111.SH", "name": "金山办公", "market": "A-share", "role": "国产办公软件", "roleEn": "国产办公软件", "bucket": "core", "href": "/investor/stocks/688111.SH/" }, { "ticker": "000977.SZ", "name": "浪潮信息", "market": "A-share", "role": "服务器与算力基础设施", "roleEn": "服务器与算力基础设施", "bucket": "core", "href": "/investor/stocks/000977.SZ/" }, { "ticker": "000063.SZ", "name": "中兴通讯", "market": "A-share", "role": "ICT 设备与服务器", "roleEn": "ICT 设备与服务器", "bucket": "core", "href": "/investor/stocks/000063.SZ/" }, { "ticker": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产 CPU/GPU", "roleEn": "国产 CPU/GPU", "bucket": "core", "href": "/investor/stocks/688041.SH/" }, { "ticker": "688256.SH", "name": "寒武纪", "market": "A-share", "role": "AI 芯片弹性", "roleEn": "AI 芯片弹性", "bucket": "core", "href": "/investor/stocks/688256.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "国产晶圆制造", "roleEn": "国产晶圆制造", "bucket": "core", "href": "/investor/stocks/688981.SH/" }, { "ticker": "688008.SH", "name": "澜起科技", "market": "A-share", "role": "内存接口芯片", "roleEn": "内存接口芯片", "bucket": "core", "href": "/investor/stocks/688008.SH/" }, { "ticker": "600845.SH", "name": "宝信软件", "market": "A-share", "role": "工业软件与云", "roleEn": "工业软件与云", "bucket": "core", "href": "/investor/stocks/600845.SH/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" } ], "stockTickers": [ "688111.SH", "000977.SZ", "000063.SZ", "688041.SH", "688256.SH", "688981.SH", "688008.SH", "600845.SH", "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT" ], "coreView": "赛道：信创/国产 IT 栈。候选观察标的：688111.SH, 000977.SZ, 000063.SZ, 688041.SH, 688256.SH, 688981.SH, 688008.SH, 600845.SH。当前共有 40 条证据、4 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 77, "eventCount": 160, "firstSeen": "2026-04-25", "lastSeen": "2026-06-30", "supports": 3, "risks": 155, "neutral": 2, "analystCount": 13, "sourceTypeCounts": { "研究线程": 2, "研究笔记": 154, "分析协作": 2, "研究主题": 2 }, "topTerms": [ "AI 基础设施", "云厂商", "变压器", "HBM", "GPU", "ASIC", "并网", "Hyperscaler", "超大规模云厂商", "AI 芯片", "AIDC", "中芯国际" ], "whatChanged": [ "2026-06-30: AI算力行情与电网瓶颈是否已成为同一个交易", "2026-06-30: AI数据中心并网延期与变压器死锁下全球算力硬件 Capex 的结构性错配", "2026-06-29: 物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进封装国产替代材料集中。", "2026-06-29: 变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "2026-06-29: AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。" ], "practical": { "sectorLane": "信创/国产 IT 栈", "sectorLaneEn": "Domestic IT substitution stack", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "688111.SH", "name": "金山办公", "market": "A-share", "role": "国产办公软件", "roleEn": "国产办公软件", "bucket": "core", "href": "/investor/stocks/688111.SH/" }, { "ticker": "000977.SZ", "name": "浪潮信息", "market": "A-share", "role": "服务器与算力基础设施", "roleEn": "服务器与算力基础设施", "bucket": "core", "href": "/investor/stocks/000977.SZ/" }, { "ticker": "000063.SZ", "name": "中兴通讯", "market": "A-share", "role": "ICT 设备与服务器", "roleEn": "ICT 设备与服务器", "bucket": "core", "href": "/investor/stocks/000063.SZ/" }, { "ticker": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产 CPU/GPU", "roleEn": "国产 CPU/GPU", "bucket": "core", "href": "/investor/stocks/688041.SH/" }, { "ticker": "688256.SH", "name": "寒武纪", "market": "A-share", "role": "AI 芯片弹性", "roleEn": "AI 芯片弹性", "bucket": "core", "href": "/investor/stocks/688256.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "国产晶圆制造", "roleEn": "国产晶圆制造", "bucket": "core", "href": "/investor/stocks/688981.SH/" }, { "ticker": "688008.SH", "name": "澜起科技", "market": "A-share", "role": "内存接口芯片", "roleEn": "内存接口芯片", "bucket": "core", "href": "/investor/stocks/688008.SH/" }, { "ticker": "600845.SH", "name": "宝信软件", "market": "A-share", "role": "工业软件与云", "roleEn": "工业软件与云", "bucket": "core", "href": "/investor/stocks/600845.SH/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" } ], "stockTickers": [ "688111.SH", "000977.SZ", "000063.SZ", "688041.SH", "688256.SH", "688981.SH", "688008.SH", "600845.SH", "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT" ], "primaryStocks": [ { "ticker": "688111.SH", "name": "金山办公", "market": "A-share", "role": "国产办公软件", "roleEn": "国产办公软件", "bucket": "core", "href": "/investor/stocks/688111.SH/" }, { "ticker": "000977.SZ", "name": "浪潮信息", "market": "A-share", "role": "服务器与算力基础设施", "roleEn": "服务器与算力基础设施", "bucket": "core", "href": "/investor/stocks/000977.SZ/" }, { "ticker": "000063.SZ", "name": "中兴通讯", "market": "A-share", "role": "ICT 设备与服务器", "roleEn": "ICT 设备与服务器", "bucket": "core", "href": "/investor/stocks/000063.SZ/" }, { "ticker": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产 CPU/GPU", "roleEn": "国产 CPU/GPU", "bucket": "core", "href": "/investor/stocks/688041.SH/" }, { "ticker": "688256.SH", "name": "寒武纪", "market": "A-share", "role": "AI 芯片弹性", "roleEn": "AI 芯片弹性", "bucket": "core", "href": "/investor/stocks/688256.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "国产晶圆制造", "roleEn": "国产晶圆制造", "bucket": "core", "href": "/investor/stocks/688981.SH/" } ], "instrumentTags": [ "688111.SH", "金山办公", "000977.SZ", "浪潮信息", "000063.SZ", "中兴通讯", "688041.SH", "海光信息", "688256.SH", "寒武纪", "688981.SH", "中芯国际" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 126 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 13, "recentEvents": 126, "supportRatio": 0.02, "riskRatio": 0.97, "conflictScore": 100, "evidenceDepth": 100 }, "score": 66.1 }, "events": [ { "date": "2026-06-30", "title": "AI数据中心并网延期与变压器死锁下全球算力硬件 Capex 的结构性错配", "summary": "AI数据中心并网延期与变压器死锁下全球算力硬件 Capex 的结构性错配 AI Power Grid Deadlock and Hyperscaler Capex Structural Misallocation 北美云巨头下修 H2 硬件 Capex 16.18%，在变压器交期达120-160周和电网并网等待53个月的物理约束下，高阶 GPU 的出货是否…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 30.75, "matchedTerms": [ "云厂商", "变压器", "GPU", "并网", "Hyperscaler" ], "href": "" }, { "date": "2026-06-30", "title": "AI算力行情与电网瓶颈是否已成为同一个交易", "summary": "AI算力行情与电网瓶颈是否已成为同一个交易 Are AI compute rallies and grid bottlenecks now one trade? 在寒武纪突破万亿市值、PJM 容量成本与变压器交期继续抬升的同一天，团队应如何同时定价 AI 芯片盈利弹性和电力并网硬约束？", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 26, "matchedTerms": [ "变压器", "并网", "AI 芯片", "寒武纪", "compute" ], "href": "" }, { "date": "2026-06-29", "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power预期缺口，会迫使Hyperscaler预付并网升级、锁定变压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.75, "matchedTerms": [ "变压器", "GPU", "并网", "Hyperscaler", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "summary": "截至2026-06-30，本报告压力测试了变压器交期拉长对AI算力集群部署和云厂商资本开支的约束。变压器交期拉长（128-144周，专用型4-5年）导致物理并网时滞达4-7年，引发全球467万个GPU延迟并网和2万亿元Capex延期。微观上，1.5年延迟导致10亿美元GPU集群产生5.2295亿美元折旧与持有损耗，月度拖累1.1905亿美元，NPV骤降12…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30, "matchedTerms": [ "云厂商", "变压器", "GPU", "并网", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-560亿美元收入，其25%低毛利率根本无法对冲高利润半导体供应链…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.5, "matchedTerms": [ "变压器", "HBM", "GPU", "并网", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,600 亿）以经营现金流支撑，资金基础/地域/标的均不同于承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.5, "matchedTerms": [ "AI 基础设施", "云厂商", "变压器", "GPU", "并网", "超大规模云厂商", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "电网与变压器瓶颈对AI算力扩张上限的实质性约束评估", "summary": "对超大规模云厂商以自建核能与微电网对冲变压器缺口的乐观叙事做压力测试，结论是否定的：在最具约束力的2026–2028窗口内无法对冲，之后也仅部分、不均衡。新建SMR最早2030–2031商运（NRC认证+许可+建设各3–5年），完全错过窗口；既有核电重启（三里岛/Crane 835MW、AWS-Susquehanna 1.92GW PPA）真实但有限，且仍…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.75, "matchedTerms": [ "云厂商", "变压器", "并网", "超大规模云厂商", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "超大规模厂商通过自备能源与地理外迁绕过电网瓶颈的对策存在严重的监管、燃料管道、物理时延与设备交付制约，难以扭转通电滑点导致的资本效率受损与资产减值风险。", "summary": "截至2026年6月28日，本报告对超大规模云厂商应对电网延迟的对策（BTM核电/气电、地理迁移及PPA）进行压力测试。分析指出，AWS-Talen等BTM核电直连模式已被监管层实质性锁死；自备燃气发电面临管道运力与发电机设备交期双重瓶颈并违背ESG目标；训练与推理的物理时延限制了集群外迁的可行性；12-18个月通电延迟将导致严重GPU资产闲置与项目级债务压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.75, "matchedTerms": [ "云厂商", "GPU", "超大规模云厂商", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI龙头估值收缩难以直接击穿基础设施层的资本开支，5至7年的并网排队与电力设备交期瓶颈构成了投资周期的物理阻尼，平滑了算力部署节奏，并将电力获取能力重塑为产业链最核心的稀缺性资产。", "summary": "截至2026年6月28日，AI龙头估值收缩对基础设施资本开支预期传导有限。由于合同锁定和竞争博弈，云厂商资本开支粘性高，2026年开支意向达7250亿美元。交付瓶颈并非需求破灭，而是受制于5-7年的并网延迟和大型电力变压器严重短缺（2026年计划交付16GW中仅有5GW在建），造成6-12个月的出货时滞。电力与资金成本上行将主要在2027年引发ROIC压缩…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "云厂商", "变压器", "并网", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM合约价Q2环比+58–63%)、以及电网电力(美国并网排队约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "变压器", "HBM", "GPU", "并网", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI驱动的结构性裁员实质上是算力投资回报（ROI）兑现的证明，云厂商AI资本开支具有极强韧性，其真实制约并非宏观衰退担忧，而是电网并网延迟与重型电力设备的物理交付瓶颈。", "summary": "本卡片对“AI裁员压力动摇算力基建投资”的假设进行压力测试并予以否定。AI驱动的结构性裁员（Challenger裁员AI占比达40%）是AI投资回报（ROI）兑现的实体证据，而非衰退信号。四大云厂商资产负债表坚韧，AI基建投资是一场生存竞争。然而，物理供电（并网排队5-7年）与设备交付（变压器交期18-36个月）瓶颈将导致2026-2027年4500-47…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "云厂商", "变压器", "并网", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "在2026-06-28，物理加电瓶颈正迫使AI算力估值由单纯的Capex扩张转向已通电MW效率，投资者应超配锁电、BTM自发电超大型云厂商与掌握变压器等硬缺口产能的电力设备商，并在A股中向特…", "summary": "截至2026-06-28，本报告探讨了在AI电力约束下算力资产估值向MW效率转型的逻辑。对于超大规模云厂商，2年的并网延迟将对100 MW智算中心产生严重的NPV减记（-124%）[S12]，迫使其估值分层：超配锁电与BTM自发电龙头，低配单纯Capex扩张者。电力设备商成为AI周期收费站，超配大型变压器等掌握稀缺产能与电网智能化的龙头。A股市场在“算电协…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27.25, "matchedTerms": [ "云厂商", "变压器", "并网", "超大规模云厂商", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "summary": "截至 2026-06-28，本卡压力测试认为：电力交付瓶颈会迫使 hyperscaler 从单纯追求新增 MW 转向 MW 生产率和存量优化，但不会终结扩产。Microsoft、Alphabet、Meta 可见 2026 年 capex 区间合计约 4,950-5,250 亿美元（自测算），同时仍披露容量约束；IEA 显示数据中心 1-3 年可建成，而新电…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27, "matchedTerms": [ "变压器", "Hyperscaler", "AI 芯片", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期：…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 / Executive Summary 截至 2026 年 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32.25, "matchedTerms": [ "AI 基础设施", "云厂商", "Hyperscaler", "超大规模云厂商", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "全球能源供应增量（特别是重启周期的核能与不稳定的可再生能源）及慢速的电网升级在2026-2027年内无法有效对冲 AI 基础设施的电网并网刚性瓶颈，该瓶颈将直接导致大云厂商面临 800 亿至…", "summary": "全球AI算力扩张正面临电网并网物理刚性瓶颈。全球电网升级周期长达4-10年，排队时间中位数达5年，且高密度变压器等关键主设备面临交付延迟，直接导致美国约7/12 GW在建产能延期，并将造成2026-27年全球超大云厂商约800亿-1100亿美元资本开支递延。核电与可再生能源虽有高成长，但受制于重启周期（3-4年）及并网不确定性，短期无法有效对冲电网卡点。算…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.25, "matchedTerms": [ "AI 基础设施", "云厂商", "变压器", "并网", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交期[S6][S7]限制的是通电而非下单。CoWoS售罄、交期…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.75, "matchedTerms": [ "变压器", "HBM", "GPU", "Hyperscaler", "infrastructure" ], "href": "" } ], "timeline": [ { "date": "2026-04-25", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "2026年超大规模云厂商AI资本开支兑现度：电力瓶颈、定制ASIC放量与英伟达份额裂缝", "summary": "2026年超大规模云厂商AI资本开支兑现度：电力瓶颈、定制ASIC放量与英伟达份额裂缝 在2026年Q1财报季临近时，四大hyperscaler合计约6,900亿美元的AI资本开支，会被电力/电网与变压器供应瓶颈吃掉多少实物算力兑现？同时定制ASIC（Trainium / TPU / MTIA /…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-26", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "2026年Q1全球超大规模云厂商AI资本开支再评估与算力供给瓶颈传导", "summary": "2026年Q1全球超大规模云厂商AI资本开支再评估与算力供给瓶颈传导 在2026Q1业绩季后,北美四大Hyperscaler(MSFT/GOOGL/META/AMZN)的2026 AI capex一致预期较2025年12月时点上修了多少?增量资金的结构如何分布?HBM/CoWoS/光模块/电力等关…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "AI 硬件超级周期正撞上由‘电力墙’、地缘成本溢价和出口管制叠加构成的宏观天花板，2027 年 HBM bit 需求增速 1.5-2.0x 的乐观预测需下调至 1.35x 以反映供应…", "summary": "本卡片对 AI 半导体超级周期的宏观可持续性进行了压力测试。研究发现，尽管 2027 年超大规模云厂商资本支出可能突破 1 万亿美元，但面临 19GW 的电力缺口和 AI 变现鸿沟，这可能迫使行业进入“消化期”。地缘政治方面，台积电 35% 的风险贴现已常态化，且 CoWoS 产能 90% 集中在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 物理超级周期面临“宏观剪刀差”挑战：利率陷阱与资源民族主义可能将“物理护城河”转化为“高久期估值风险”。", "summary": "本卡对 AI 基础设施“计算层向物理层转移”的逻辑进行了全球宏观压力测试。研究发现，虽然物理护城河具有结构性吸引力，但面临“宏观剪刀差”风险：即前置的物理层通胀（铜、电力设备）与延迟的 AI 生产力回报之间的背离，可能迫使央行维持高利率，从而对高久期的工业和公用事业资产造成估值压制。此外，地缘政治…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "2026下半年AI资本开支拐点辩论：消化期还是再加速？", "summary": "TMT分析师判定，2026下半年AI资本开支并未进入消化期，而是进入结构轮动期：总量持平至上行，边际增量美元从训练优化的商用GPU集群转向推理fleet、超大规模云厂商自研芯片、以及电力与场地准备。空头叙事在需求侧（推理token volume、agentic渗透、多模态）证据不足，但在前沿训练边…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "将 5% BYOP 燃气与 4% 非晶替代从触发对冲升级为 2026 H2 基础配置，确认电力缺口已由‘并网延迟’转向‘表后自建’与‘能效替代’的硬约束阶段。", "summary": "基于 GE Vernova 和西门子能源 2026Q2 的创纪录订单披露，以及微软、甲骨文等 hyperscaler 的‘自带电力’(BYOP) 转向，验证了燃气轮机槽位已成为类 GPU 的战略稀缺资源。同时，由于 160 周的变压器交付瓶颈和 2027 年 DOE 能效新规，非晶合金对取向硅钢 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "2028 年变压器交付周期会拖慢 2028 年起的新增 AI greenfield 园区并重新定价资产栈，但通过预订、表后自电、变电站复用和能效提升，并不是 2028 年之前 AI …", "summary": "截至 2026-05-14，120–210 周的大型变压器交付周期对 2026 下半年至 2028 年的 AI 新建园区构成现实约束，但不是 2028 年之前整体 AI 算力的硬顶。超大规模云厂商通过 2022–2024 年完成的变压器预订、表后燃机/SMR/PPA、退煤站点变电站复用、机架能效（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "AI算力链基本面与订单流：对减仓建议的支持性验证 日期： 2026 05 17 分析师： TMT行业分析师 [Primary] 对Research note 01立场： Suppor…", "summary": "AI算力链基本面与订单流：对减仓建议的支持性验证 日期： 2026 05 17 分析师： TMT行业分析师 [Primary] 对Research note 01立场： Support（支持） 结论： 基本面边际开始出现裂缝——印证因子拥挤度去风险建议 1. 摘要 Research note 01…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "AMD 上涨与 Agentic AI 标志着推理层的真实碎片化，但训练层仍是 NVDA 主导的寡头格局；这是 AI 投资逻辑的分层化，而非永久性转向。", "summary": "AMD 暴涨与 Agentic AI 叙事是部分真实的结构性变化，但被卖方过度简化为'永久性转向'。Agentic 工作负载将单任务 token 量放大 30–100 倍，使约束转向 HBM 带宽与单 token 经济性，ROCm 7 + vLLM 生态使 AMD MI355X 在推理侧达到与 H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 3, "neutral": 0, "value": 35.2, "events": [ { "title": "能源物流没有被 AI 消灭，但“物理大宗必要性”不等于“能源大宗运输贝塔免疫”，真正稀缺性正在迁移到电力设备、燃机、储能和电网集成。", "summary": "截至 2026-05-19，工业制造视角否定卡片 05 的强版“能源物流战略性免疫”。能源分子仍需大宗运输，但 AI 电力需求首先制造的是电气设备、燃机、储能、开关设备、变压器、并网和控制系统瓶颈，而不是对 VLCC、煤炭铁路或普通港口大宗贝塔的广谱保护。IEA 2026 年报告显示 hypers…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关键电力设备会让 announced AI capacity 打折，但除非 hyperscaler capex 与电气基础设施供应商订单同时掉头，否则 H2 2026 不应按整体算力…", "summary": "截至 2026-05-19，本卡对 Research note 07 的 BTM 受阻结论做压力测试：变压器、GSU、开关柜和并网队列已成为 AI 算力扩产的真实交付瓶颈，120-143 周交期足以让部分 announced MW 延迟。但基准判断不是 hyperscaler capex 停止，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "订单流驱动的估值底面临“执行力”压力测试，需警惕寒武纪等HBM高度依赖标的的流动性陷阱，中芯国际底部更稳但缺乏弹性。", "summary": "截至2026-05-19，华盛顿关于限制存量设备维护及HBM出口许可的动向，正在将半导体龙头的政策风险从“限制扩张”转向“限制执行”。中芯国际因其产能垄断地位具备较强估值底，但存量维护成本上升压制良率与毛利；寒武纪面临HBM断供导致的巨额订单无法交付风险，估值底逻辑可能因供应链断裂而失效。散户融资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "2026-2028年将是‘绿色AI的黑暗时代’，物理电网设备的极端匮乏使算力扩张与能源转型深度脱钩，迫使云厂商放弃零碳优先原则以换取业务连续性。", "summary": "截至2026-05-20，本卡片通过压力测试确认：变压器等重电装备的“物理排队”已成为新能源并网的硬天花板。高压变压器交付周期拉长至3-5年，导致大量风光项目处于“机械完工但电气搁浅”状态。AI超大规模云厂商凭借资本优势掠夺生产槽位，进一步挤出了新能源开发商。预计2026-2028年将出现严重的“…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "AI算力的真正最长杆约束不是变压器或电网并网，而是同时满足光纤、电网、水权、规划许可与社区接受的那一块土地——这是5–10年的政治-法律问题，不是18–30个月的工业品问题。", "summary": "从房地产视角反驳「GPU→变压器→电网」的瓶颈三段论：真正最长杆的约束是土地+水权+规划许可。Loudoun、都柏林、新加坡的限令落在土地用途而非设备；PJM Order 2023 把「场地控制」作为并网过滤器，本质是地权门槛；阳光带与西班牙Aragón水权诉讼正缩减DC规模；EQIX/DLR 与…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "'效率降级'对 2026–2027 AI Capex 总量近乎无约束，但通过电力、关税、航运三条渠道侵蚀单位经济学，使全球 AI 基础设施从'同步扩张'转为'地理极化'，估值真正承压…", "summary": "本卡对 Research note 1–5 的“效率降级削弱 AI 算力扩张”假设进行双向压力测试。结论：方向正确但程度被高估，AI 算力 Capex 在 2026–2027 名义增速仍 +19–26% YoY（北美 hyperscaler 2026 指引 4,400–4,650 亿美元 [S1–…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 0, "value": 46.8, "events": [ { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "现有先进半导体出口管制不会停止AI使用或全部ASIC设计，但会系统性削弱受限生态通过HBM密集、CoWoS级封装实现“每瓦有效推理”跃迁的能力。", "summary": "截至2026-05-25，本卡压力测试前序“ASIC与先进封装推动每瓦有效推理”的判断。结论是：路径对可获得TSMC、HBM与CoWoS产能的全球云厂商仍成立，但对中国相关或D:5受限生态被结构性削弱。美国出口管制已从成品AI芯片延伸至HBM、封装路径、先进制造设备、软件/技术和FDP规则，卡住了…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "变压器上游材料会成为AI算力扩张的阶段性硬约束，但约束对象是站点通电和容量投运，而不是GPU/ASIC芯片本身。", "summary": "截至2026-05-27，本卡支持材料瓶颈方向，但把TMT结论界定为“已通电兆瓦”约束：GOES与铜导线不直接限制GPU/ASIC晶圆供给，却会通过变压器、变电站和并网交付影响AI集群投运节奏。NVIDIA NVL72约120 kW/72颗GPU，100 MW IT负载约对应60,000颗GPU；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "AI硬件周期仍由需求驱动，但未来12个月的定价权将更多落在HBM、先进封装、光模块和电力交付等瓶颈环节，而不是单纯GPU或服务器组装收入。", "summary": "截至2026-05-30，本卡支持“AI算力瓶颈从硅端扩展至物理基础设施”的主线。未来12个月，AI芯片和服务器需求仍由超大云厂商资本开支、云AI收入、agentic AI推理流量和自研ASIC扩张驱动；但供应瓶颈已从GPU单点扩展到HBM、CoWoS/先进封装、800G/1.6T光模块、机架供电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 2, "neutral": 0, "value": 51, "events": [ { "title": "A股AI资本开支与光通信战略展望 (2026年5月) 作者： 首席策略师 报告日期： 2026 05 30 白研究线程标识：已归档会话 ID： 69628375 103f 4490 …", "summary": "A股AI资本开支与光通信战略展望 (2026年5月) 作者： 首席策略师 报告日期： 2026 05 30 白研究线程标识：已归档会话 ID： 69628375 103f 4490 b22e 7faa4bad1a21 1. 核心观点与摘要 截至2026年5月30日，全球AI基础设施建设周期正在从单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "发电侧严重的设备产能短缺、燃料供应链刚性及电网并网壁垒将强行拖慢hyperscaler的实际算力部署速率，这不仅会拉长并抬高输配电物理底座设备的量价齐升超级周期，还将迫使市场对软件A…", "summary": "截至2026年5月31日，本卡对发电侧进行了深度压力测试：在变压器与开关柜订单锁死的前提下，发电端（CCGT、SMR、可再生）将成为限制大厂算力部署的核心物理红线。GE Vernova燃机产能2028年前售罄，2029-2030残余空间仅10GW，且受管道环评制约；SMR受制于燃料链缺口（HALE…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 1, "value": 61.4, "events": [ { "title": "受制于受监管公用事业股本回报率天花板与5年期并网及变压器设备瓶颈，算力税向“电力税”的再分配呈现高度结构化分化，且因3年GPU折旧错配促使AI算力走向离网与全球化多样部署（2026-…", "summary": "截至2026-05-31，本报告对“电力税”价值攫取进行压力测试：AI电网容量与清洁基荷的绝对稀缺，使价值加速向大型核电/气电发商（CEG、VST）及物理电网核心变电设备商（GEV、ETN）倾斜。然而，该传导机制正面临严厉的联邦和州级费率监管壁垒，并因5年以上变频/并网排队耗时与3年GPU折旧周期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 0, "value": 53.8, "events": [ { "title": "电网变压器长达150周的交期将导致大量AI芯片成为“搁浅算力”，迫使云厂商资本支出向电力倾斜，NVIDIA等硬件龙头的2027年盈利预期面临下修风险。", "summary": "截至2026年6月2日，本卡对AI基础设施部署进行压力测试。NVIDIA GB200机架和10万卡集群面临150周变压器交期的硬约束。这意味着2026年中启动的数据中心到2029-2030年才能并网。从GPU交付到通电的周期将从6-9个月拉长至18-24个月，云厂商将被迫将资本支出向电力设备倾斜。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "供应链脱钩带来的成本上升并未抑制需求，而是将全球能源转型的驱动力从‘成本红利’转向‘安全与物理约束下的刚性支出’，算力溢价正成为吸收绿色通胀的主要缓冲机制。", "summary": "截至2026-06-02，全球能源转型正进入‘高成本、低弹性’的需求超周期。尽管美欧供应链脱钩导致局部LCOE上升10-15%，且变压器等关键设备面临30-50%的脱钩溢价，但AI算力负荷的刚性增长和电网现代化的迫切需求抵消了成本抑制。超大规模云厂商和公用事业公司正在吸收‘绿色通胀’，优先确保交付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 2, "neutral": 0, "value": 55.2, "events": [ { "title": "2027 年电力设备交付瓶颈无法被产能扩张化解，高昂的积压订单饱和度与供应链约束将导致 AI 基础设施在 2027 年面临严重的物理交付墙。", "summary": "压力测试显示，GE Vernova 与西门子能源的扩产计划无法化解 2027 年交付瓶颈。尽管两家公司正在美国和欧洲大幅扩产，但其产能早已被 2024-2025 年签署的创纪录积压订单（GEV $150B，西门子 €154B）预占。燃机和变压器的交付周期分别维持在 5-7 年和 4-5 年的历史高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 电网并网瓶颈对超大规模云厂商 Capex 中期预测的抑制与 ASICs 替代效应", "summary": "致 TMT行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: f67bbdf4-647f-4185-acf0-af247ed9f8b2 结果摘要: # Step −1 — Anchor today's date BEFORE you start Run this …", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "电力与电网瓶颈已成为AI算力落地的物理天花板，如果变压器短缺与电网容量摩擦导致40%项目延期，将直接在2026-2027年期间递延多达700亿美元的半导体行业营收，加剧低能效云商的闲…", "summary": "2026年6月9日报告通过对电力供应和电网容量的压力测试发现，物理用电瓶颈和电网并网延迟正在成为制约AI算力中心落地的短线硬伤。如果40%的规划负荷发生被动延期，将在2027年前导致多达280万张GPU无法正常部署，对应半导体行业约700亿美元营收被递延。云厂商面临资本开支闲置与变现周期拉长的双重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 4, "neutral": 0, "value": 46, "events": [ { "title": "电力交付滞后会迫使 Hyperscalers 调整 GPU 集群投产节奏，并通过更长现金回收期、更低早期利用率和更高电力保障成本压低 AI 基础设施 capex 效率。", "summary": "截至2026-06-11，本卡支持上一卡：AI capex 不会因电网瓶颈消失，但 GPU 集群部署节奏会被“已通电 MW”重新排序。变压器、并网和输电延迟会把部分已宣布园区容量推迟到后续窗口，压低早期自由现金流和资产利用率。Hyperscalers 会通过已通电站点溢价、分阶段投产、区域重排和自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备交期把2026-2030年AI基础设施周期从GPU供给约束转为已通电MW约束，美国路径需要实质性时间下修，但全球需求主线不被推翻。", "summary": "截至2026-06-12，本卡片对设备交期硬约束进行压力测试：Hyperscaler可通过预购设备、转移GPU、棕地扩容、表后发电和电源共址对冲部分风险，但无法完全绕开48-60个月级别的大型变压器/高压接入瓶颈。基准判断是，2030年美国可通电AI/数据中心容量较无约束capex计划下修15-2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "配电侧物理约束已成为 AI 算力部署的新红线，被迫产生的“400V 桥接期”将超大规模集群的成熟期推后至 2028 年，价值链正向“通电时间”确定性溢价的电力设备厂商集中。", "summary": "配电侧瓶颈（800V DC 生态缺口与 120+ 周变压器交期）正迫使 Hyperscaler 从“算力优先”转向“电力优先”。2026–2027 年将进入“400V 过渡期”，导致 300kW+ 超高密度集群的大规模部署推迟至 2028 年。Capex 分配正向电力基础设施（~15%）倾斜，核心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 1, "risks": 4, "neutral": 0, "value": 55.4, "events": [ { "title": "AI并网延迟与贸易壁垒将2026-2028年的交易逻辑从“算力增长”推向“电力稀缺”，配置重心应从云厂商切换至基荷能源与具备价格传导权的设备龙头。", "summary": "截至2026-06-12，本卡战略综合了AI电力物理瓶颈的长期化影响：2026年Q1财报确认了“插电能力”取代GPU成为云厂商营收转化的首要约束。建议在跨市场配置中进行结构性切换：对北美超大规模云厂商（Hyperscalers）实施战术性减配（-10%至-15%），以应对资本支出效率下降及算力利用…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备成本传导对AI园区总Capex的影响<1.5%，但120+周交期与2,600 GW并网积压使\"送电时点\"成为云厂商AI集群2026–2028扩产的真实硬约束。", "summary": "本卡对Research note 01进行压力测试。结论：电网设备毛利与交付压力对超大规模云厂商AI扩产的传导以\"延期\"为主，而非\"成本\"。即便20%变压器/开关柜涨价全额穿透到一个250–350亿美元级AI园区，对总成本的拉动也仅0.6%–1.5%（壳体口径1.5%–3.8%）[自测算]，不足以…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "燃机交付墙与变压器交付墙的叠加不会终止AI建设，但会把2027-2030年的胜负标准从宣布MW改写为可送电MW。", "summary": "截至2026-06-12，本卡支持前序“AI电力物理瓶颈”主线，并把燃机墙与变压器墙合并为2027-2030年超大规模云厂商的可用功率闸门。Goldman预计美国数据中心需求从2025年31 GW升至2027年66 GW，但近端计划容量仅50-60%可按时上线；自测算显示，仅2027年计划新增36…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超大规模云厂商Capex仍以芯片为大头，但边际瓶颈和估值溢价正转向电力基础设施；AI芯片周期会被已通电MW重排为延期与集中释放并存的台阶曲线。", "summary": "截至2026-06-13，本卡支持“估值锚转向已通电MW”的主线，并将其落实到Capex结构：GPU/服务器仍约占AI基础设施资本开支的最大部分，但电气、制冷、变电站、并网和租赁型长期资产正在成为进度闸门。按公开指引，四大云厂商2026年Capex约6800-7100亿美元；1 GW AI数据中心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 9, "neutral": 0, "value": 18.8, "events": [ { "title": "超大规模云厂商 AI 资本开支撞上电网墙:2026 年从资本约束切换至电力约束的相位转换", "summary": "2026 年 AI 基建的边际约束已从 GPU 与资本迁移至电力与电网设备。Top-6 云厂商资本开支约 6,900 亿美元(同比+27%),但 PJM 并网周期 18–36 个月、大型变压器交期约 128 周、PJM 2027–28 容量缺口 6.6 GW。Microsoft 自行披露 800 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器和开关柜长期缺口把AI算力扩容的关键稀缺单位从GPU改为可送电MW，并将电力接入期权、等待时间和能效价值纳入边际成本。", "summary": "截至2026-06-13，本卡压力测试并支持前序结论：变压器与开关柜短缺不会杀死AI需求，但会把Hyperscalers扩容模型从“已公告GPU/GW”改写为“已送电MW”。LPT交期可达36-60个月，WoodMac口径显示电力变压器约128周、GSU约144周；单GW IT负荷、1.25 PU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2027年Hyperscaler AI资本开支的主要压力不是变压器涨价，而是通电延迟导致资本开支转化效率下降。", "summary": "截至2026-06-14，本卡支持但收紧上游供应链割裂结论：大型变压器再延长8-12周、成本再上升6-9%的直接资本开支压力可控，每1GW AI数据中心约增加6800万至2.05亿美元，但进度冲击更大。AWS、Azure、Google的核心风险不是负担不起设备涨价，而是2026-2027年资本开支…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS绝对产能正在改善，但GPU与自研ASIC需求吸收速度更快，因此2026年AI集群交付仍受先进封装、HBM和关键设备交期的结构性约束。", "summary": "截至2026-06-13，本卡支持上一张AI基建卡的判断：硬件交付风险不只来自并网，还来自CoWoS/HBM/设备交期的半导体转换链。TSMC仍称先进封装非常紧，TrendForce口径显示CoWoS-L/S满载且2026年底目标120,000-130,000片/月；DIGITIMES虽更乐观，预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 5, "neutral": 0, "value": 44.2, "events": [ { "title": "物理电网瓶颈正迫使 AI 基础设施从‘暴力规模扩张’转向‘受约束的能效优化’，电力并网速度已取代芯片供应成为核心估值护城河。", "summary": "本报告通过压力测试确认了物理电网并网排队（7-10年）与变压器交期（4-5年）已成为 AI 基础设施扩张的硬约束。定量模型显示，1年的并网延迟将导致 100MW 数据中心损失 5.5% 的生命周期价值，且快速的 GPU 迭代周期（12-18个月效用减半）加剧了闲置产能的 ROI 损耗。这种物理瓶颈…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备供给瓶颈导致的成本不确定性迫使云厂商在 2026-2027 年将 CAPEX 向电力基础设施大幅倾斜，AI 算力部署从“受制于GPU”全面转向“受制于通电能力”。", "summary": "截至 2026-06-14，电网设备出海的 ROE 波动与建厂压力正向超大规模云厂商传导。美国变压器交期拉长至 160-200 周，云厂商的核心痛点从“缺芯”变为“缺电”。为锁定产能，云厂商被迫承担更高前期成本，数据中心非 IT 开支（电力/制冷/土建）占比预计从历史的 20% 升至 35%-40…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然先进封装产能逐步释放，但物理电网并网延迟与容量电费暴涨构成了AI基础设施扩张的实质性新边界，其引起的电力成本增加与部署延期将击穿高杠杆算力运营效率，引发科技板块的倍数收缩并恶化电…", "summary": "截至2026-06-14，先进封装瓶颈逐步缓解，但电网容量与电力供应成为限制AI算力部署的新边界。美国数据中心面临3-4年并网延迟，变压器交期长达4年且价格涨80%[S7, S8]。PJM容量价暴涨至$329.17/MW-day[S2]，推动数据中心电费升至$80.24/MWh[自测算]。在$0.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "有电力保障的AI基础设施Capex确定性上升；无电力保障的Capex将被电网价格、保证金和设备交期继续配给。", "summary": "截至2026-06-14，本卡综合判断：电力风险并未消失，但正在从未定价瓶颈转化为可预算、可筛选、可合同化的分配机制。PJM容量价、ERCOT大负荷规则、核电PPA和低空置率让头部hyperscaler的AI数据中心及GPU集群Capex更具名义金额确定性；但保证金、容量费、变压器交期和建设成本通…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 1, "risks": 5, "neutral": 0, "value": 53.6, "events": [ { "title": "评估电力供应瓶颈与电网扩容速度对AI算力集群建设及超大规模云厂商（Hyperscalers）资本开支的影响", "summary": "本卡支持并强化 Research note 01 的'公用事业→AI核心基座'论点：电力供应与电网/配电设备交付周期（而非GPU供给）才是当前AI算力部署与云厂商资本开支变现的真正约束。2026年五大云厂商合计资本开支约6,600–6,900亿美元，近100%吞噬经营现金流，但Microsoft约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建正面临电网接入与电力设备的“物理限速”，2026-2028年天然气将作为核电落地前的核心基荷桥梁，而铝替代削弱了铜作为绝对瓶颈的地位。", "summary": "2026年6月调研显示，AI扩张瓶颈已从GPU转向电网并网（PJM等待期达5.2年）与高压变压器（交付周期120周）。2026-2028年存在明显的“基荷电力缺口”，核电SMR需至2030年后落地，迫使Hyperscalers转向天然气“表后”发电。铜面临50万吨级别缺口且价格高企（$13k/t）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡片 02 — AI 芯片供应链：DeepSeek V4 后的\"架构定制化\"趋势 白板：1f5523fe c230 4de9 bcab 6fc3d7edec87 ｜ 卡片 2/8 ｜…", "summary": "卡片 02 — AI 芯片供应链：DeepSeek V4 后的\"架构定制化\"趋势 白板：1f5523fe c230 4de9 bcab 6fc3d7edec87 ｜ 卡片 2/8 ｜ 日期：2026 06 15 分析师：TMT 行业分析师 ｜ 立场： 支持（support） 核心问题：DeepSe…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 9, "neutral": 1, "value": 26, "events": [ { "title": "30-50%的并网延迟与GPU快速迭代将在2027年触发高达1120亿美元的AI硬件减值悬崖（40%减值率），这不仅会将Hyperscaler的合计自由现金流拖入负值（-75亿美元）…", "summary": "针对2026-06-17工作日，本报告对超大规模云厂商（Hyperscaler）在30-50%并网延迟与GPU生命周期快速迭代双重挤压下的财务韧性执行了压力测试。测算表明，若2027年因“受困算力”折旧与技术过时触发大规模AI硬件减值，在40%的减值情景下（对应1120亿美元资产减值），Big 4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与GOES的线性供给极限将强制AI基建进入‘受困Capex’周期，并网延迟对ROIC的损害远超电价波动，‘确定性电力’将成为2028年前唯一的估值溢价来源。", "summary": "截至2026-06-17，变压器（交期36-60个月）与取向硅钢（GOES）的结构性缺口已成为Hyperscaler AI扩张的核心制约。30%-50%的2026年计划产能将推迟至2028年，导致海量‘受困算力’与折旧前置，压制ROIC。厂商通过核电直连、‘表后’发电及提前4年预定生产槽位进行对冲…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "卡06 · 电力约束如何重塑Hyperscalers选址逻辑与未来三年算力部署规模 工作日期：2026 06 17（Asia/Singapore，机构权威工作日） 分析师：AI基础设…", "summary": "卡06 · 电力约束如何重塑Hyperscalers选址逻辑与未来三年算力部署规模 工作日期：2026 06 17（Asia/Singapore，机构权威工作日） 分析师：AI基础设施分析师（ai infrastructure analyst） 立场：support（支持卡05结论——电网容量已成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理基础设施需求足以在没有流动性爆发的情况下维持电网设备与稀缺投入品周期，但市场仍应惩罚同时需要廉价资本和完美部署节奏的AI算力商业模式。", "summary": "截至2026-06-17，本卡压力测试结论是：在宏观流动性不爆发的情景下，AI算力、电网接入、变压器和液冷的物理需求增量足以支撑头部hyperscalers与稀缺电网供应链资产，但不能保护依赖廉价资金、缺乏确定电力或确定客户的投机型AI算力项目。全球数据中心用电到2030年约945 TWh、美国到…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 4, "neutral": 0, "value": 62.4, "events": [ { "title": "FOMC 后应以“选择性算力基础设施挤空”而非“无差别 AI 融涨”来交易：多稀缺、空低 ROIC，净多但对冲。", "summary": "2026-06-18 策略综合：前序链条成立，2027 风险是 ROIC 与结构分化而非 capex 悬崖。FOMC 后应做多稀缺算力资产而非泛 AI beta：前沿逻辑/设备/HBM/先进封装、自研 ASIC 与网络、电力与电网受益方应超配；商用 GPU 龙头保留但严控风险；低 ROIC 商用 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力的真实长期定价在于'公用事业溢价'，2026年中位节点约+25 USD/MWh、+24–48个月通电延迟，吸收B200级服务器5年TCO的9%–15%——是前瞻AI资本开支决…", "summary": "2026-06-17本卡作为收官卡支持01–07卡的核心框架，量化Hyperscaler AI算力的'公用事业溢价'。溢价由L2并网+L3电能质量+L4低碳合规三层叠加，2026年中位站点较2019年基线约+25 USD/MWh，碳严节点（欧盟/台湾/北弗吉尼亚）+35–45 USD/MWh，>3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "WACC上行30bp会打击AI资产估值倍数，但不会单独瓦解AI基础设施资本开支；防御性集中在有订单、有瓶颈、有现金流的硬科技与电力/电网环节。", "summary": "截至2026-06-17，本卡压力测试第06张卡的硬科技杠铃：WACC上行30bp会压缩长久期估值，但不足以单独击穿hyperscaler的AI数据中心、算力、电力与电网资本开支。8.0%到8.3%的WACC变化仅使5年资产年度资本回收系数上升0.8%、15年资产上升1.8%、30年资产上升2.8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供应缺口已成为AI算力交付的绝对瓶颈，2年的供电延迟将直接摧毁10万卡Blackwell GPU集群约58%的初始投资价值，倒逼云厂商向电力基建与开发商融资注入重资本。", "summary": "评估了电力缺口（SMR真空与BTM滞后）对Hyperscaler GPU集群交付及CapEx的具体影响。电力延迟使交付期拉长至3-5年，导致10万卡Blackwell集群2年内产生27.67亿美元折旧价值减损；WAN分布式训练因15-30%效率损耗使NPV转负。云厂商被迫将CapEx向电力资产与开…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 7, "neutral": 0, "value": 37.8, "events": [ { "title": "超大规模云厂商正通过将资本开支直接投向核电PPA、表后燃气轮机及主动式电池储能（BESS），从单纯的算力采购者向“能源开发与系统集成商”转变，以解除“通电围墙”物理并网瓶颈，防止高价…", "summary": "截至2026-06-18，超大规模云厂商的资本开支（2026年预计达6600-6900亿美元）重心已从单纯增加算力芯片供应，转向加大电力基础设施、主动式电池储能系统（BESS，2026年市场达49.6亿美元）及核电/SMR等长期购电协议（PPA，已锁定达9.8 GW）的资本投入，以应对严重的“通电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与GOES的区域化贸易壁垒通过拉长交付周期至154.5周，直接造成算力资产闲置折旧（24k GPU集群年拖累达5.624亿美元），迫使超算厂商进行CAPEX重组。", "summary": "截至2026-06-18，研究变压器与GOES区域壁垒对AI算力的传导效应。美国IRA本土含量、欧盟NZIA及中国反倾销税等碎片化政策使GOES供给割裂。压力情景下，卷线工稀缺与设备积压将变压器交期拉长至154.5周。这导致超算并网时滞达3-5年，使得24k GPU算力集群闲置一年的折旧与流失利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新增AI风险预算应从继续加半导体Beta转向电力稀缺链条，半导体保留核心仓位但不再是边际加仓方向。", "summary": "截至2026-06-19，本卡支持并推进此前从通用软件/加密杠杆转向AI硬约束的思路：半导体基本面仍强，但新增风险预算的边际胜率已转向电力设备、并网基础设施、现场供电和核燃料。建议下周新增AI风险预算60-70%投向电力与供能约束，20-30%保留精选半导体，0-10%用于现金或对冲。核心依据是h…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由于大型电力变压器与测试大厅产能严重受限，2027-2028年计划新增的AI数据中心有43%面临通电并网延期[自测算]，由此引发GPU订单推迟，对主要芯片制造商及云厂商带来显著的收入…", "summary": "基于2026-06-19数据，本分析量化了电网变压器与并网延迟对2027-2028年AI硬件产业的影响。模型显示，原计划30GW数据中心管线因变压器交期拉长至120-210周而延期12-24个月，实际仅能交付17.1GW，缺口达43%[自测算]。这导致GPU出货量在2027年减少4.8M颗、202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 8, "neutral": 0, "value": 33.2, "events": [ { "title": "电力与封装成本上升会触发AI基础设施技术路线调整，但截至2026-06-20仍不足以触发超大规模云厂商AI芯片采购总量的全面回撤。", "summary": "工作日2026-06-20：本卡压力测试认为，电力和先进封装成本上升不会在2026年触发超大规模云厂商全面削减AI芯片采购，但会提高单位弹性，推动技术路线从“更多GPU插槽”转向“每瓦性能、液冷机架、自研ASIC、提前锁定CoWoS/HBM和电力站点”。真正风险是并网、变电站、CoWoS或HBM配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Starlink 带宽 × Grok 算力的飞轮真实，使 SPCX 成为 AI 分发基座与算力链条需求催化剂，但算力归属 xAI、轨道边缘算力尚萌芽，真正瓶颈是电力——故支持但需把光…", "summary": "从 AI 基础设施视角支持卡片 01：Starlink 带宽（约 450 Tbps，三代卫星单星 >1 Tbps、下行 10 倍/上行 24 倍）与 Grok/Colossus 算力（约 55.5 万颗 GPU、约 180 亿美元、向 2 GW 扩张）构成真实飞轮，使 SPCX 成为差异化的'算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A-share de-leveraging to 1550 points triggers secondary market liquidity locking and margin…", "summary": "若市场下探至1550点，中际旭创股价将跌至714.16元，击穿融资预警与平仓线，带来日均31.2亿元平仓抛压（占ADV 17.36%），锁定二级流动性。但公司持有125亿现金及45亿海外客户预付款，可提供151.5亿净营运资金，供应链运营违约风险极低。海外订单真正风险在于电网并网排队达36-60个…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 12, "neutral": 0, "value": 12, "events": [ { "title": "电网交付延迟不会终结高密度 AI 机柜，但会迫使下一代 AI 芯片与集群架构把 perf/W、可限功率和可分阶段并网能力置于峰值算力之上。", "summary": "2026-06-20：本卡对电网设备瓶颈约束 AI 芯片架构的说法进行压力测试。结论是，变压器和开关设备长交期不会让前沿训练集群普遍降密度；GB200 NVL72、Vera Rubin 和 AMD Helios 仍指向高密度机柜级设计。但电网交付延迟会把竞争指标从峰值 FLOPS 推向受限兆瓦下的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力基础设施的瓶颈已由芯片供给转移至确定性可调度电力，并网延迟将对高资本密集的算力集群造成灾难性折旧空转，推动 Hyperscaler 宁可支付高额电价溢价购买灵活火电/核电备…", "summary": "在可调度电力受限的背景下，超大规模云厂商的数据中心部署从“算力驱动”转为“能源驱动”。本卡对中美数据中心的选址经济学、并网延迟财务拖累及电网输电备用费政策进行重估。测算表明，美国 Dominion 区域容量电价上涨使 blended 电价翻倍至 $80.24/MWh；而在 500MW 的 Blac…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "summary": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FERC资费改革对超大规模云服务商数据中心投资经济性的影响", "summary": "压力测试card-08：支持其内核（FERC 2025-12-18同址负荷终局令确在收窄表后/同址套利窗口），但否定其标题机制（网费回收是损伤IRR主因）。量化看，回收的网电成本约$150 mn/年，仅占1 GW数据中心~$8.5 bn/年TCO的约1.8%，与card-07电力为~7% TCO二…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 9, "neutral": 0, "value": 30, "events": [ { "title": "电力设备交付瓶颈对超大规模云厂商(Hyperscalers)数据中心扩建进度与CapEx规划的影响", "summary": "截至2026-06-21，本卡从需求侧支持Cards 4-6的电力设备硬约束判断。云厂商对2026-2027变压器/开关柜断层的反应是双重的：既重排序路线图（30-50%的2026美国项目推迟/取消，迁向有电场地，算力分批绑定电力里程碑），又激进对冲（在拿证前提前下单长交期电气设备、垂直整合BTM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler资本开支正从算力优先转向电力协同：总量仍大，但边际一美元先购买电力、进度和并网确定性，再购买下一组加速器。", "summary": "截至2026-06-21，本卡压力测试认为：Hyperscaler资本开支不会因电力瓶颈在2026-2027年崩塌，但边际规则已从“买更多GPU”转向“按确定日期拿到可通电MW”。核电购电、表后自发电、并网权、长交期电气设备和区域边缘部署正在获得更高资本权重。核电是长期稀缺性对冲，不是近端解决方案…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CNH压力不会杀死AI基础设施资本开支周期，而会把价值捕获从融资敏感的AI Beta转向电力锁定、并网稀缺和强资产负债表支撑的硬基础设施。", "summary": "截至2026-06-22，本卡综合CNH流动性压力与AI基础设施周期：支持前序“CNH挤压—远期倒挂—即期购汇—境内流动性抽水”的传导链，但否定把AI基础设施一刀切视为高Beta受害者。AI资本开支会分化：电力、电网接入、变压器、冷却和强资产负债表hyperscaler受益于物理稀缺；人民币融资G…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-22，低端封装过剩会压低错误产能池的价格，而高端迭代缺失会保留真正重要产能池的配给风险，使Hyperscaler AI资本开支更重、更不稳定、更容易出现已供电机房…", "summary": "截至2026-06-22，本卡压力测试卡片06：先进封装低端过剩并不能缓解Hyperscaler真正需要的高端CoWoS/HBM4认证产能短缺。TrendForce称2026年行业CoWoS-like产能可接近200,000 wpm、缺口或收窄至约10%，但若高端认证占比只有50%-70%，有效高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 8, "neutral": 0, "value": 37.4, "events": [ { "title": "超大规模云厂商刚性的AI资本开支与先进封装产能的物理瓶颈，锁定了半导体前道关键设备和先进封装工具的在手订单，使其免受宏观周期与电网延迟的传导冲击，确立了其作为防御成长底仓的逻辑。", "summary": "截至2026-06-22，研究表明北美云厂商2026年6350亿-7500亿美元资本开支（75%绑定AI）具有非discretionary刚性。台积电CoWoS封装产能（2026年底达12万-14万片/月）面临严重短缺，限制英伟达Blackwell GPU出货上限至445万颗。这为半导体设备链提供…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备紧缺与并网延迟正倒逼AI云厂商调整选址逻辑，通过工作负载地理分流、核电结构化直供、现场过桥发电以及电气设备提前大批量采购，来对冲并网延误风险，将AI资本开支推向重度能源与电力…", "summary": "截至2026-06-22，高压变压器交期拉长至24-48个月且排队并网超5年，成为AI集群部署核心瓶颈。超大规模云厂商正通过将训练与推理负载地理分流、签署核电结构化购电协议（PPA）、部署现场天然气发电机过桥以及提前3-4年锁定集团化设备产能等调整选址与容量规划，以对抗延迟风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-22核心判断：AI算力基建能把“外暖”转化为中国出口盈利，但应买瓶颈环节而非泛AI整机叙事，排序为电力设备和液冷最强、服务器零部件次之、前沿先进封装最需验证。", "summary": "2026-06-22：支持出海制造链受益于AI算力基础设施，但必须结构化看待。全球Hyperscaler资本开支仍高，2026年AI相关投资估算可达7250亿美元；中国5月工业与出口交货值仍强，但美国AI硬件进口正向台湾、墨西哥迁移。中国盈利承接最强在变压器、开关设备、液冷和电力配套，其次是PCB…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器/电网瓶颈对AI芯片需求预期与库存周期的传导风险", "summary": "截至2026-06-22，本卡以半导体视角压力测试Card-03。结论：芯片需求仍明确领先于电力瓶颈，2026年不存在AI芯片砍单——约束仍是芯片供给本身（CoWoS售罄至2026年底、英伟达独占约60%；Blackwell积压约360万颗售罄至年中；2026年HBM全部订满且超大规模厂商预付扩产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 9, "neutral": 0, "value": 32.8, "events": [ { "title": "截至2026-06-23，电力设备投资收缩会加剧局部并网瓶颈并改变AI Capex结构，但更可能造成区域送电延期和头部集中度提升，而不是触发Hyperscaler总Capex刹车。", "summary": "截至2026-06-23，本卡对前序链条做压力测试：313.42亿元电力设备与地方能源基建需求收缩不足以单独压制Hyperscaler AI Capex总量，但会在变压器、变电站、开关柜和接网工程紧张地区推迟AI数据中心送电。基准判断是Capex不会消失，而是从GPU和土建转向电力权利、接网保证金…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备瓶颈对AI算力建设的传导表现为“成本影响微弱，工期延迟致命”的非对称效应，并网延迟导致的GPU折旧与机会成本已成为制约算力扩张的终极变量。", "summary": "截至2026-06-23，定量压力测试显示电网设备“成本+久期”双杀对超大厂AI基础设施的传导呈非对称性：成本端传导极低，变压器价格翻倍对120MW项目（含10亿美元GPU）的总CapEx直接通胀率仅0.54%[自测算]；但久期端传导呈灾难性，36个月的变压器交期与15个月算力建设周期的错配导致2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler并不缺AI Capex，真正约束算力上线节奏的是可交付电力、并网流程与长交期电力设备。", "summary": "截至2026-06-24，本卡支持上一卡：AI算力瓶颈已从Capex/GPU转向通电MW。Alphabet 2026 CapEx指引1,800-1,900亿美元、Meta 1,250-1,450亿美元，Microsoft仍称需求超过可用容量，Amazon Q1购置物业与设备442.03亿美元；但S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 6, "neutral": 0, "value": 52.2, "events": [ { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与开关设备长周期会迫使AI数据中心按可送达电力重新选址，并把超大规模资本开支从GPU/土建扩展到电力交付栈。", "summary": "截至2026-06-25，本卡片压力测试卡片06：AI算力选址确实会从“便宜电”转向“可信time-to-power”，但不是简单回到传统成熟枢纽。变压器、开关设备、变电站与并网改造成为已带电容量的关键约束；低电价若没有设备槽位和送电排期，只是搁浅期权。超大规模厂商的资本开支将更多前置到电气设备预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施资本开支仍成立，但必须从GPU纯敞口重估为电力锁定型、费率可承担、设备可交付的电网重资产敞口。", "summary": "截至2026-06-25，本卡压力测试AI算力基础设施扩张：需求真实，但瓶颈已从GPU采购转向确定性电力、并网、变压器/开关设备和费率承担。美国数据中心2028年用电可能达325-580 TWh，全球2030年约945 TWh；100,000颗GB300 GPU集群在1.15 PUE假设下约需22…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "液冷与先进封装材料落地对AI集群功率密度与Capex效率的影响", "summary": "本卡（2026-06-24）以AI基础设施运营视角压力测试'液冷+玻璃基板因集成复杂性拉低超大规模云厂商Capex效率'命题，得出双面裁决：就稳态单位经济学（每电网MW算力、每壳体算力、GB200密度下每美元算力）而言，液冷是Capex效率增益项——80–130 kW/机柜把2–3倍算力塞入同壳体…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 7, "neutral": 0, "value": 47.6, "events": [ { "title": "虽然物理电网排队与供应链原材料价格冲击显著抬高了局部建设阻力，但云厂商通过绿电PPA、高密度液冷及转向定制ASIC等结构性调整，能够完全消纳成本溢价，不构成算力扩张的绝对瓶颈。", "summary": "本卡对电网并网延迟、变气压原材料（铜/GOES）价格波动和本地化GPU封装良率冲击进行了量化压力测试。结果表明，虽然并网延迟和原材料成本增加会显著压缩项目NPV（5年延迟将使NPV降低79.99%），但云厂商可以通过迁移西部枢纽自建绿电、液冷优化和系统级混合算力架构，消纳国产芯片和能源价格溢价，并…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-25，AI硬件需求的下一核心约束不是芯片是否能生产，而是足够多的兆瓦能否被许可、接入、冷却并按期商业化。", "summary": "截至2026-06-25，本卡支持上一张卡的AI硬件稀缺性主线，但将瓶颈从GPU/HBM进一步上移到可通电数据中心、电网接入、冷却和电气设备。IEA、Berkeley Lab、EPRI、CBRE和JLL数据共同显示，AI需求仍强，但项目能否转化为创收产能取决于电力和并网速度。结论是：稀缺性继续支撑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的主要瓶颈正从GPU可得性转向电力可得性，真正受益的是能缩短通电时间、提高负荷密度并保障可靠供电的设备与电网环节。", "summary": "截至2026-06-26，本卡支持上一张“算力切向电力”的轮动逻辑，但把核心约束细化为并网、变压器/开关设备、机柜功率密度和可靠供电。全球数据中心用电预计由2024年415 TWh升至2030年945 TWh；中国2030年约279 TWh，全国占比不高但局部冲击显著。1 GW AI IT负荷在P…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 1, "neutral": 0, "value": 85, "events": [ { "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 6, "neutral": 0, "value": 56.4, "events": [ { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球能源供应增量（特别是重启周期的核能与不稳定的可再生能源）及慢速的电网升级在2026-2027年内无法有效对冲 AI 基础设施的电网并网刚性瓶颈，该瓶颈将直接导致大云厂商面临 80…", "summary": "全球AI算力扩张正面临电网并网物理刚性瓶颈。全球电网升级周期长达4-10年，排队时间中位数达5年，且高密度变压器等关键主设备面临交付延迟，直接导致美国约7/12 GW在建产能延期，并将造成2026-27年全球超大云厂商约800亿-1100亿美元资本开支递延。核电与可再生能源虽有高成长，但受制于重启…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据中心电力并网、关键电力设备与液冷温控的物理交付瓶颈将对算力上线速度形成硬约束，导致云厂商ROIC受损并延缓算力采购，增加电子产业链利润扩张的硬着陆风险。", "summary": "截至2026-06-28，本卡压力测试了算力产业链利润扩散假说，指出电力并网与变压器超长交期（MV变压器128周）已成为数据中心上线的物理瓶颈。预计2026–2028年美国35%–45%的新建数据中心面临延期通电，16GW计划容量仅5GW在建，30%-50%将延期至2028年以后。GB200的高功…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 10, "neutral": 0, "value": 33.8, "events": [ { "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与变压器瓶颈对AI算力扩张上限的实质性约束评估", "summary": "对超大规模云厂商以自建核能与微电网对冲变压器缺口的乐观叙事做压力测试，结论是否定的：在最具约束力的2026–2028窗口内无法对冲，之后也仅部分、不均衡。新建SMR最早2030–2031商运（NRC认证+许可+建设各3–5年），完全错过窗口；既有核电重启（三里岛/Crane 835MW、AWS-S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超大规模厂商通过自备能源与地理外迁绕过电网瓶颈的对策存在严重的监管、燃料管道、物理时延与设备交付制约，难以扭转通电滑点导致的资本效率受损与资产减值风险。", "summary": "截至2026年6月28日，本报告对超大规模云厂商应对电网延迟的对策（BTM核电/气电、地理迁移及PPA）进行压力测试。分析指出，AWS-Talen等BTM核电直连模式已被监管层实质性锁死；自备燃气发电面临管道运力与发电机设备交期双重瓶颈并违背ESG目标；训练与推理的物理时延限制了集群外迁的可行性；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI龙头估值收缩难以直接击穿基础设施层的资本开支，5至7年的并网排队与电力设备交期瓶颈构成了投资周期的物理阻尼，平滑了算力部署节奏，并将电力获取能力重塑为产业链最核心的稀缺性资产。", "summary": "截至2026年6月28日，AI龙头估值收缩对基础设施资本开支预期传导有限。由于合同锁定和竞争博弈，云厂商资本开支粘性高，2026年开支意向达7250亿美元。交付瓶颈并非需求破灭，而是受制于5-7年的并网延迟和大型电力变压器严重短缺（2026年计划交付16GW中仅有5GW在建），造成6-12个月的出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 3, "neutral": 0, "value": 77.2, "events": [ { "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "summary": "截至2026-06-30，本报告压力测试了变压器交期拉长对AI算力集群部署和云厂商资本开支的约束。变压器交期拉长（128-144周，专用型4-5年）导致物理并网时滞达4-7年，引发全球467万个GPU延迟并网和2万亿元Capex延期。微观上，1.5年延迟导致10亿美元GPU集群产生5.2295亿美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-56…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-30", "supports": 0, "risks": 2, "neutral": 0, "value": 84.6, "events": [ { "title": "AI数据中心并网延期与变压器死锁下全球算力硬件 Capex 的结构性错配", "summary": "AI数据中心并网延期与变压器死锁下全球算力硬件 Capex 的结构性错配 AI Power Grid Deadlock and Hyperscaler Capex Structural Misallocation 北美云巨头下修 H2 硬件 Capex 16.18%，在变压器交期达120-160周…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力行情与电网瓶颈是否已成为同一个交易", "summary": "AI算力行情与电网瓶颈是否已成为同一个交易 Are AI compute rallies and grid bottlenecks now one trade? 在寒武纪突破万亿市值、PJM 容量成本与变压器交期继续抬升的同一天，团队应如何同时定价 AI 芯片盈利弹性和电力并网硬约束？", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-7a6599a3f5/", "networkHref": "/investor/research-network/#thesis-7a6599a3f5" }, { "id": "thesis-603a5f7966", "source": "tracked_thesis", "title": "A 股硬科技与科创50的复苏需要订单、融资盘和指数结构共振", "titleEn": "A-share hard tech and STAR50 recovery needs orders, margin positioning, and index structure to align", "lane": "A股硬科技/科创50", "laneEn": "A-share hard tech and STAR50", "laneHref": "/investor/lanes/a-share-hard-tech-and-star50/", "investableFocus": "A股硬科技/科创50：588000.SH, 688981.SH, 688256.SH, 688041.SH, 688008.SH, 688111.SH, 300308.SZ, 002916.SZ", "stockUniverse": [ { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "国产晶圆制造", "roleEn": "国产晶圆制造", "bucket": "core", "href": "/investor/stocks/688981.SH/" }, { "ticker": "688256.SH", "name": "寒武纪", "market": "A-share", "role": "AI 芯片弹性", "roleEn": "AI 芯片弹性", "bucket": "core", "href": "/investor/stocks/688256.SH/" }, { "ticker": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产算力芯片", "roleEn": "国产算力芯片", "bucket": "core", "href": "/investor/stocks/688041.SH/" }, { "ticker": "688008.SH", "name": "澜起科技", "market": "A-share", "role": "内存接口芯片", "roleEn": "内存接口芯片", "bucket": "core", "href": "/investor/stocks/688008.SH/" }, { "ticker": "688111.SH", "name": "金山办公", "market": "A-share", "role": "AI 办公软件", "roleEn": "AI 办公软件", "bucket": "core", "href": "/investor/stocks/688111.SH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "002916.SZ", "name": "深南电路", "market": "A-share", "role": "AI PCB/封装载板", "roleEn": "AI PCB/封装载板", "bucket": "core", "href": "/investor/stocks/002916.SZ/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" } ], "stockTickers": [ "588000.SH", "688981.SH", "688256.SH", "688041.SH", "688008.SH", "688111.SH", "300308.SZ", "002916.SZ", "510300.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK" ], "coreView": "赛道：A股硬科技/科创50。候选观察标的：588000.SH, 688981.SH, 688256.SH, 688041.SH, 688008.SH, 688111.SH, 300308.SZ, 002916.SZ。当前共有 40 条证据、3 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 76, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 7, "risks": 139, "neutral": 14, "analystCount": 8, "sourceTypeCounts": { "研究笔记": 143, "证据更新": 2, "分析协作": 5, "研究主题": 10 }, "topTerms": [ "AI 基础设施", "HBM", "CPO", "GPU", "科创50", "A股", "融资余额", "AI 芯片", "中际旭创", "中芯国际", "寒武纪", "光模块" ], "whatChanged": [ "2026-06-29: 国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多资产组合VaR在1530点处强制破位，急需分层部署的ETF买盘及两融弹性化阀门进行精…", "2026-06-29: AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。", "2026-06-29: 国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加速撤向水电红利及电网设备出海龙头，陡峭化了向高红利风格轮动的斜率，配置应重构为哑铃型…", "2026-06-28: GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。" ], "practical": { "sectorLane": "A股硬科技/科创50", "sectorLaneEn": "A-share hard tech and STAR50", "exposureBucket": "中国权益与流动性", "tone": "green", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "国产晶圆制造", "roleEn": "国产晶圆制造", "bucket": "core", "href": "/investor/stocks/688981.SH/" }, { "ticker": "688256.SH", "name": "寒武纪", "market": "A-share", "role": "AI 芯片弹性", "roleEn": "AI 芯片弹性", "bucket": "core", "href": "/investor/stocks/688256.SH/" }, { "ticker": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产算力芯片", "roleEn": "国产算力芯片", "bucket": "core", "href": "/investor/stocks/688041.SH/" }, { "ticker": "688008.SH", "name": "澜起科技", "market": "A-share", "role": "内存接口芯片", "roleEn": "内存接口芯片", "bucket": "core", "href": "/investor/stocks/688008.SH/" }, { "ticker": "688111.SH", "name": "金山办公", "market": "A-share", "role": "AI 办公软件", "roleEn": "AI 办公软件", "bucket": "core", "href": "/investor/stocks/688111.SH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "002916.SZ", "name": "深南电路", "market": "A-share", "role": "AI PCB/封装载板", "roleEn": "AI PCB/封装载板", "bucket": "core", "href": "/investor/stocks/002916.SZ/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" } ], "stockTickers": [ "588000.SH", "688981.SH", "688256.SH", "688041.SH", "688008.SH", "688111.SH", "300308.SZ", "002916.SZ", "510300.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK" ], "primaryStocks": [ { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "国产晶圆制造", "roleEn": "国产晶圆制造", "bucket": "core", "href": "/investor/stocks/688981.SH/" }, { "ticker": "688256.SH", "name": "寒武纪", "market": "A-share", "role": "AI 芯片弹性", "roleEn": "AI 芯片弹性", "bucket": "core", "href": "/investor/stocks/688256.SH/" }, { "ticker": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产算力芯片", "roleEn": "国产算力芯片", "bucket": "core", "href": "/investor/stocks/688041.SH/" }, { "ticker": "688008.SH", "name": "澜起科技", "market": "A-share", "role": "内存接口芯片", "roleEn": "内存接口芯片", "bucket": "core", "href": "/investor/stocks/688008.SH/" }, { "ticker": "688111.SH", "name": "金山办公", "market": "A-share", "role": "AI 办公软件", "roleEn": "AI 办公软件", "bucket": "core", "href": "/investor/stocks/688111.SH/" } ], "instrumentTags": [ "588000.SH", "科创50 ETF", "688981.SH", "中芯国际", "688256.SH", "寒武纪", "688041.SH", "海光信息", "688008.SH", "澜起科技", "688111.SH", "金山办公" ], "reviewCadence": "每日复核", "confirmationSignals": [ "融资余额、ETF 申赎和北向资金是否同向", "被动调仓窗口是否被主动资金提前交易", "订单、盈利和估值是否同步改善" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "指数期权", "红利低波", "汇率", "两融降温" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：证据密度足够", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 122 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 8, "recentEvents": 122, "supportRatio": 0.04, "riskRatio": 0.87, "conflictScore": 97.6, "evidenceDepth": 100 }, "score": 67.9 }, "events": [ { "date": "2026-06-29", "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性TAM上限，营收放缓促使估值压缩。相反，全球CPO光模块（中际…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.5, "matchedTerms": [ "HBM", "CPO", "GPU", "A股", "中际旭创", "寒武纪", "光模块" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.25, "matchedTerms": [ "HBM", "CPO", "GPU", "A股", "光模块", "structure", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck from compute to power, networkin…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 23.25, "matchedTerms": [ "AI 基础设施", "GPU", "structure", "infrastructure", "bottlenecks" ], "href": "" }, { "date": "2026-06-29", "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）将使平仓率非线性飙升至20%，面临940亿STAR50抛压与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22, "matchedTerms": [ "CPO", "科创50", "A股", "STAR50" ], "href": "" }, { "date": "2026-06-28", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,600 亿）以经营现金流支撑，资金基础/地域/标的均不同于承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22, "matchedTerms": [ "AI 基础设施", "GPU", "structure", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "在半导体周期复苏不及预期与高负债压力的复合冲击下，AI基础设施板块估值呈现两极分化：缺乏盈利的高负债概念股将面临系统性去估值及杠杆踩踏风险，而具备真实EPS支撑、客户壁垒及液冷/光网络核心壁…", "summary": "截至2026-06-26，本报告对AI基础设施龙头（AI算力/数据中心）在半导体周期波动与高负债压力下的估值溢价进行了系统性压力测试。在-16.18%硬件资本开支下修及WACC上升30 bps（导致终端估值倍数下降4.8%-7.0%）的复合压力下，行业基本面呈现深度分化。以中际旭创、新易盛、工业富联和英维克为代表的盈利龙头，凭借1.6T光模块及液冷技术壁垒…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.25, "matchedTerms": [ "中际旭创", "光模块", "structure", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "效率驱动的算力架构转型：从“暴力规模”到“高能效密度”的物理基座适配性评估", "summary": "压力测试 Research note 08 的隐含前提：效率并不缓解物理天花板，而是把它从电网层下移到机架与机柜行层并加剧。机理是机架尺度的杰文斯悖论——单瓦产出靠提高密度实现，机架功率从约40kW(Hopper)→120kW(Blackwell)→142kW(GB300,90%入液)→600kW(Kyber,2027)迈向1MW，对照2026均值约27k…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.25, "matchedTerms": [ "HBM", "GPU", "structure", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计约6900-7200亿美元（自测算），NVIDIA FY20…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 21.25, "matchedTerms": [ "HBM", "光模块", "structure", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM合约价Q2环比+58–63%)、以及电网电力(美国并网排队约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.25, "matchedTerms": [ "HBM", "GPU", "structure", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "全球AI基础设施资本开支对中国产业链的传导机制与订单确定性", "summary": "本卡支持卡片05遗留命题并定论：约7250亿美元、同比+77%的全球超大厂资本开支（约75%即约4500亿为AI专项）规模巨大、外部出资，已转化为中国主导且出口管制并不约束的那一层链条（光互连/连接/散热/PCB）的跨多季度订单簿与已兑现盈利。关键在于出口管制是分层的：重压落在GPU/算力层（华为昇腾、SMIC受限）和上游光芯片层（EML/DSP/InP国…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.25, "matchedTerms": [ "GPU", "中际旭创", "光模块", "structure", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "在2026-06-27的权威工作日背景下，国产HBM良率骤降逼迫国内算力网络化，而全球封装与电力阻尼拉长了云厂商CapEx变现周期，这使得制程免疫且订单排至2028年的可插拔光模块出口龙头成…", "summary": "截至2026-06-27，本报告评估了HBM良率危机与先进封装瓶颈对算力集群技术演进的影响。国产HBM3封装良率降至2.66%，倒逼国内算力集群转向网络密集的Scale-out拓扑，推高了光模块需求弹性。全球云厂商2026年CapEx直逼7250亿美元，但受制于CoWoS晶圆先进封装冲突与吉瓦级电网接入瓶颈，面临6-12个月的算力交付与收入确认错配。CPO…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.25, "matchedTerms": [ "HBM", "CPO", "光模块", "structure", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "在 USD/JPY 跌破150的情景中，在岸高两融TMT的主要脆弱性是融资余额绝对规模和券商风控顺周期收紧，而非接近交易所25%单股融资上限。", "summary": "2026-06-27：本卡压力测试卡片06。日本4月持有1.2099万亿美元美债，USD/JPY跌破150的流动性通道方向成立，但2150亿-2850亿美元清算规模未被独立证实。A股两融余额为30329.9亿元，核心TMT行业约4670.0亿元、占15.4%。TMT风险不在交易所25%融资暂停线，而在高绝对余额与券商风控收紧：中际旭创、寒武纪、兆易创新、澜…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "A股", "融资余额", "中际旭创", "寒武纪" ], "href": "" }, { "date": "2026-06-27", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交期[S6][S7]限制的是通电而非下单。CoWoS售罄、交期…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.25, "matchedTerms": [ "HBM", "GPU", "structure", "infrastructure" ], "href": "" }, { "date": "2026-06-26", "title": "CPO试仓的基本面门是光芯片寿命、时间/ROI门是电力排期、A股存活门是去杠杆——Chiplet从来不是门；周一基本面说该持有，市场结构(创纪录两融+70–138倍估值+解禁)说要小仓分批、…", "summary": "综合卡片01–06：基本面线把约束一路下移——光模块(EML寿命)→激光器HTOL→算力→电力排期→卡片06修正为'披着成本外衣的时间与ROI'。但这条链只回答'主题会否/何时兑现'，没回答'A股仓位能否熬过入场'。从市场结构看，CPO试仓近端门是去杠杆：两融2026-06-23创纪录¥3.00万亿[S1]、极致集中于中际旭创/新易盛/天孚通信[S3]、估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "CPO", "A股", "融资余额", "中际旭创", "光模块", "chip" ], "href": "" }, { "date": "2026-06-26", "title": "1.6T光模块交付延迟对AI算力集群及超大规模云厂商Capex部署进度的影响", "summary": "支持卡片01/02的研究线并加以边界化延伸：200G EML 的 HTOL 风险确会让 1.6T GA 滑向 2027H1，但这是一道带宽前沿与单位经济性约束，而非整体 Capex 停摆。GB300 NVL72 的 scale-up 走铜缆 NVLink、scale-out 走 ConnectX-8 每 GPU 800G[S7]，约 6 万台机柜[S8]照…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.25, "matchedTerms": [ "CPO", "GPU", "光模块", "structure", "infrastructure" ], "href": "" }, { "date": "2026-06-26", "title": "科创50回撤15%会触发半导体去杠杆左尾，但连锁强平的根因是科创/半导体集中账户与券商150%-160%集中度平仓线，而不是中芯国际、北方华创两只股票的公开融资余额本身。", "summary": "截至2026-06-27，本卡压力测试显示：科创50若从2026-06-26收盘2032.28回撤15%至约1727，普通130%平仓线只击穿冲击前CMR≤152.9%的账户；真正危险来自券商双创集中度规则，50%-80%集中度可把平仓线抬至150%，80%以上抬至160%，使176%-188% CMR区间账户也可能触发。中芯国际与北方华创公开融资余额合计…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "科创50", "A股", "融资余额", "中芯国际" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "外汇冲击通过融资余额平仓、解禁供给与机构集中度赎回三重放大机制，将5–15%的汇率损失转化为A股AI芯片板块33–50%的峰谷回撤，Q3 2026窗口为近十年A股最不利的系统性风险叠…", "summary": "在USD/CNY 7.65 / USD/JPY 132压力情景下，A股AI芯片板块面临三重放大机制：（1）寒武纪等名称融资余额占流通市值约9–11%，触发维持担保比例红线后将引发约180–280亿元被动平仓；（2）壁仞科技（Biren Technology）2026年7月解禁涉及约130–230亿…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "在DeepSeek/H20出口管制双重冲击下，光模块、DDR5接口芯片和成熟国产AI芯片厂商能转化为2026年真实盈利，而以寒武纪为代表的早期GPU设计公司仍是叙事溢价显著高于盈利兑…", "summary": "对中国AI/半导体供应链做收入兑现压力测试：光模块（中际旭创300308、天孚通信300394）、DDR5接口芯片（澜起科技688008）和有成熟收入基础的国产AI芯片（海光信息688041）是2026年最可信的盈利转化标的；先进封装（长电科技600584、通富微电002156）路径较长；寒武纪（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年中国AI硬需求由运营商+互联网+智能车+央企四类买家托底，足以兑现光模块/接口芯片/主流国产GPU的产能；但地方政府主导的智算中心需求池将在2027年成为叙事贝塔股的估值空…", "summary": "从需求端综合验证第04卡的供给分化论：2026年中国AI硬需求集中在四类买家——三大运营商（合计算力Capex约RMB 700–800亿/年）、互联网平台（2025年AI基础设施开支约RMB 3,500–4,000亿）、智能电动车/ADAS、央企信创——足以消化中际旭创、天孚通信、澜起、海光的20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 0, "neutral": 0, "value": 52.8, "events": [ { "title": "2026年A股AI配置应从寒武纪类地方政府需求叙事贝塔，切换到中际旭创、澜起、海光等有订单、有利润、有现金流验证的AI基础设施盈利股。", "summary": "本卡综合卡04、卡05并落实到2026年A股配置：不卖出AI，而是在AI内部从政策叙事贝塔切换到订单和盈利贝塔。1Q26业绩确认中际旭创、澜起科技、海光信息具备更强订单到利润转化；寒武纪1Q26同样显著改善，因此不是不可投，但应从核心AI贝塔降为高贝塔盈利期权。建议将寒武纪类叙事GPU敞口较当前超…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "胜宏科技(300476.SZ)、新易盛(300502.SZ)、天孚通信(300394.SZ)、中际旭创(300308.SZ)、寒武纪(688256.SH)和海光信息(688041.S…", "summary": "胜宏科技(300476.SZ)、新易盛(300502.SZ)、天孚通信(300394.SZ)、中际旭创(300308.SZ)、寒武纪(688256.SH)和海光信息(688041.SH)构成A股算力链的脆弱性漏斗。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股半导体头部公司包括中芯国际 (688981)、寒武纪 (68．8256) 等。", "summary": "A股半导体头部公司包括中芯国际 (688981)、寒武纪 (68．8256) 等。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "算力期货压缩的是商用GPU层租金，先进封装与高带宽互联的物理瓶颈与寡头结构足以将其非商品化毛利结构维持至2027。", "summary": "支持card-01框架并加以收紧：算力期货标准化的是GPU-小时这一输出单位，并未商品化决定谁能以最低成本清算的物理输入。CoWoS先进封装、HBM4、混合键合设备、scale-up交换机ASIC、PCIe/CXL retimer、1.6T/CPO光模块均具18–36个月结构性导入周期，寡头格局清…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "H200对华许可的毒丸条款使其对中国云厂只能扮演过渡角色，对跨境AI硬件营收偏空、对国产替代链（SMIC/昇腾/寒武纪）结构性偏多，板块定价应反映'强制分化'而非'复苏催化'。", "summary": "H200对华许可叠加25%关税与BIS领土审计构成'毒丸'结构：单芯片承诺、现场审计、芯片级遥测、再出口禁令使H200对中国云厂成为高摩擦、不可转售、用途受限的过渡载体而非战略基石。我们下修Nvidia 2026年中国数据中心营收至80-100亿美元（vs.一致预期120-140亿），毛利率拖累5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "Nvidia 直供链上的 A 股份额东移是真实但窄的（5–8 家硬名单），归因是合同制造而非美方约束；CPO 光引擎核心仍在美方，真正的份额弹性在独立订单簿的国产替代层。", "summary": "在 GB300 NVL72 与 Rubin CPO 架构下，A 股供应链份额东移的真实增量集中在 800G/1.6T 光模块 OEM（中际旭创、新易盛、天孚）、液冷 CDU 与冷板（英维克、高澜、工业富联整柜）、超低损耗高速 PCB（沪电、胜宏）三个环节，硬名单仅 5–8 家，远小于市场涨停所定价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "量化压力测试：A股算力链融资盘平仓与 VaR 破缺风险报告 日期： 2026 05 19 分析师： chief risk (首席风控官) 立场： 压力测试（负向风险预警） 1. 执行…", "summary": "量化压力测试：A股算力链融资盘平仓与 VaR 破缺风险报告 日期： 2026 05 19 分析师： chief risk (首席风控官) 立场： 压力测试（负向风险预警） 1. 执行摘要 截至 2026 年 5 月 19 日，A 股 AI 算力链（特别是光模块/CPO 与 PCB 板块）已进入“杠…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海外 Hyperscaler FY26 资本开支稳固、卡片 01 触发器未触发，四只『千元股』基本面逻辑未破，绑定风险是情绪侧而非需求侧，但内部分化要求先减光模块、保留 ASIC。", "summary": "美股四大 Hyperscaler 1Q26 业绩（2026 年 4 月末／5 月初）合计 FY26 资本开支约 US$3610–3710 亿，同比 +28%，无一家下修，Meta 还小幅上调 —— 卡片 01 列出的『Hyperscaler 资本开支下修』触发器截至 2026-05-18 未触发。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 2, "neutral": 1, "value": 52.2, "events": [ { "title": "若 2026-06-11 验证 Waymo 单位经济，AI 基础设施不应被卖出，而应从拥挤算力前端轮动到电源液冷、networking 邻近环节与小仓位机器人硬件 beta。", "summary": "截至2026-05-19，本卡支持 6 月 11 日后从 AV 扩散到 AI hardtech，但只在 Waymo 经济性披露有用时执行。确定性最高的是电源/液冷/机柜、定制芯片与 networking、HBM/先进封装；估值弹性最好的是电源热管理与机器人端侧硬件。HBM、GPU、A 股光模块和主…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "A股半导体/AI算力的基本面只支持精选龙头，不支持2026-05-19开盘无差别低吸；融资拥挤与IM贴水要求先等确认再小仓参与。", "summary": "截至2026-05-19，A股半导体与AI算力微观结构支持前序“散户/杠杆承接机构卖盘”的判断，但不是全板块崩盘信号。ETF相对估算净值仅小幅溢价约+0.2%至+0.7%，说明有低吸承接但不极端；融资盘高度集中在寒武纪、澜起科技、中际旭创、新易盛、海光信息、中芯国际A等AI算力和芯片龙头；IM26…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "消费电子'技术性弱复苏'+AI 手机溢价被压在旗舰段，触发的是 A 股'叙事溢价 → 确定性溢价'的哑铃化再定价，而非全面价值回归或单边 AI 基础设施集中。", "summary": "工作日期 2026-05-19。综合 card-03（AI 基础设施仍是窄口径主线）、card-04（消费电子 2026H2 技术性弱复苏）与 card-05（以旧换新净撬动仅约 1.25x、AI 手机溢价仅 ≥6,000 元旗舰段）三条结论后，本卡的策略级判断是：A 股既不会单边全面价值回归，也…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "截至 2026-05-19，A 股哑铃化再定价已经启动：左端买现金回报和低波确定性，右端买 AI 基础设施瓶颈，但右端拥挤度已显著高于基本面确认速度。", "summary": "工作日期 2026-05-19。A 股盘面支持第 06 卡“叙事溢价 → 确定性溢价”的哑铃化判断，但不是全面价值回归。左端证据是红利策略 ETF 5 月以来净流入约 72.9 亿元、5月18日上证50ETF净申购9.78亿元；右端证据是两融资金集中买入电子和通信，5月11日至15日两者合计占周度…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 0, "neutral": 1, "value": 57.6, "events": [ { "title": "[auto-handoff] 杠杆盘上限与软件/半导体选择性叠加拥挤度测算", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 07ee71f2-f2fb-4678-bf32-e56e4628f116 结果摘要: I will start by executing the shell command to anchor today's d…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 5, "neutral": 0, "value": 26, "events": [ { "title": "2026-05-22的核心判断是：科创50和AI硬件链的风险不在主题本身，而在融资盘拥挤、折算率弹性不足和成交放大回落叠加后触发的被动减仓。", "summary": "截至2026-05-22，本卡支持并细化card-01的压力测试结论：被动去杠杆风险集中在融资余额大、融资余额/市值占比高、近期融资买入重且已遭遇成交放大回落的AI硬件标的。最高风险名单包括300476胜宏科技、300308中际旭创、300502新易盛、300394天孚通信、688256寒武纪、6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "汇率波动与 -5% 锁汇成本将强化高端光模块的出海 Alpha 溢价，但将严重侵蚀薄利润 AI 服务器代工的利润空间，并显著拉高半导体晶圆代工的进口设备 CapEx 成本与 ROIC…", "summary": "压力测试显示汇率波动与锁汇成本飙升对 TMT 产业链呈结构性分化影响：高端光模块出口龙头（如中际旭创）凭借 90% 以上境外美元收入及 18% 高净利率表现极强韧性，天然对冲并能消化 5% 的锁汇摩擦成本；而薄利率（4.5%）的 AI 服务器与 EMS（如工业富联）因高昂进口 GPU 账期错配，5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件微观结构已达杠杆与拥挤度极值，融资盘的极致集中与外资的选择性转向预示着一旦估值锚移动，将触发剧烈的去杠杆再定价。", "summary": "截至2026-05-21，A股AI硬件微观结构呈现“极致杠杆+外资分化”特征。中际旭创融资余额达332.61亿元历史极值，而北向资金在光模块（增持）与服务器（减持）间出现质量筛选。板块成交占比触及35%拥挤警戒线，加之Q2/Q3解禁潮压力，验证了高拥挤板块在巨型IPO扰动下的脆弱性及再定价风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "出口管制传闻在TMT基本面侧的真实冲击远小于午盘股价隐含的悲观幅度，但敞口分布的非均匀性意味着红利轮动结束后将出现「国产替代纯度高」标的的结构性重定价。", "summary": "针对2026-05-22午盘发酵的出口管制传闻，TMT侧的供应链拆解显示敞口分布并不均匀。光模块龙头(300308/300502)海外收入占比超85%，是直接受冲击点，但800G/1.6T的产能集中度形成天然壁垒，真实营收冲击预计仅5–15%；服务器代工(601138/000977)依赖墨西哥/越…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "'效率降级'对 2026–2027 AI Capex 总量近乎无约束，但通过电力、关税、航运三条渠道侵蚀单位经济学，使全球 AI 基础设施从'同步扩张'转为'地理极化'，估值真正承压…", "summary": "本卡对 Research note 1–5 的“效率降级削弱 AI 算力扩张”假设进行双向压力测试。结论：方向正确但程度被高估，AI 算力 Capex 在 2026–2027 名义增速仍 +19–26% YoY（北美 hyperscaler 2026 指引 4,400–4,650 亿美元 [S1–…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 0, "value": 46.8, "events": [ { "title": "在2027美国AI Capex消化情景下，A股正确动作是杠铃式六桶再平衡，不是出口AI硬件与国内算力的二元切换。", "summary": "截至2026-05-26，作为8卡终卡综合：A股不应做光模块/PCB与国内算力之间的二元切换，而应做杠铃式再平衡——将出口集中度最高的光模块龙头（中际旭创、新易盛）从超配下调至中性偏低，高端PCB（沪电、胜宏）保持中性，把腾出的风险预算加到国内算力与自主芯片（寒武纪、海光、曙光、浪潮）、特高压与电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI 基础设施散户拥挤是否耗尽 Thesis v8 入场窗口", "summary": "致 市场情绪师： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: b790980f-74a8-48f2-8307-cf1e60b887ee 结果摘要: Wrote handoff_response.md (14984 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "两融资金推升的A股半导体“成长进攻”在AI核心链上具备真实利润兑现（如高速光模块和先进封装），但成熟制程价格战严重分化；且在高交易拥挤度下极其脆弱，必须实施“去粗存精”的精准防御。", "summary": "截至2026年5月27日，本卡对两融推升的半导体与AI硬件“成长进攻”进行压力测试。AI核心环节（中际旭创毛利达42.61%，通富微电AMD订单爆发）基本面坚实，但成熟制程（中芯国际毛利受制于价格战降至18-20%）分化严重。在美债利率上冲至5.0%及AI CapEx减速下，73%两融拥挤度平仓将…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 3, "neutral": 1, "value": 46.6, "events": [ { "title": "MU/SOXX继续加权、A股算力代理结构性降配、资金转向真实瓶颈+低拥挤应用+短缺材料——三段同步执行，正式在仓位结构上把海外硬件beta与A股叙事beta脱钩。", "summary": "支持MU/SOXX继续作为全球AI硬件稀缺信号加权——HBM、CoWoS、电力变压器三段瓶颈在2026年仍未缓解。同时主动降配A股高杠杆算力代理（CPO/光模块整机、AI服务器组装、IDC高估值租赁），因为这些标的与海外硬件共享叙事但不共享基本面节奏，过去12个月涨幅由估值贡献主导[自测算]，拥挤…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "台积电CoWoS与HBM底层封装良率构成全球AI芯片出货的硬性物理约束，超额利润集中于垄断性瓶颈厂商（TSMC/MU/SK Hynix），而A股代理股共享叙事却无壁垒且交易极度拥挤，…", "summary": "支持超配海外半导体物理瓶颈（MU/SOXX）并降配A股算力代理。台积电CoWoS封装产能从2024年底3.5万片/月扩至2026年底超12万片/月但仍售罄，NVIDIA垄断超60%份额；HBM市场2026年规模将达600亿美元，美光产能售罄且良率升至70-75%。A股光模块、服务器组装与IDC代理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "若300308.SZ跌破前收，AI硬件链（CPO/PCB）超550亿元的高位融资盘将面临集中踩踏风险，联动去杠杆压力将呈指数级放大。", "summary": "截至2026年5月27日，本卡对300308.SZ跌破前收的情境进行了压力测试。数据显示，中际旭创与新易盛两只CPO龙头的融资余额合计超550亿元。若300308.SZ（作为板块情绪锚）跌破前收，将触发高位拥挤盘的恐慌性抛售。叠加前置的科创50期权到期与TMT去杠杆背景，AI硬件链（CPO/PCB…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 科创50指数1800/1750点破位下雪球敲入规模与两融杠杆平仓负反馈测试", "summary": "致 主题研究员： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 21a31e2b-133e-4d11-8860-ebeffe825ce4 结果摘要: Wrote handoff_response.md (11959 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 2, "neutral": 1, "value": 54, "events": [ { "title": "战略上继续保留高毛利AI核心与真实区域制造，但战术上停止追高光模块三剑客，并对三花智控、石头科技设置更紧的杠杆风险约束。", "summary": "截至2026-05-30，A股AI硬件与出海2.0并非整体过度拥挤，拥挤风险集中在光模块三剑客和少数叙事型出海标的。中际旭创、新易盛、天孚通信公募持股均值约20.17%，合计融资余额667.7亿元，10%回撤会带来约66.8亿元融资盘市值损失。美的、福耀、赛轮、浙江鼎力的融资结构仍温和；三花智控、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力拥挤盘的局部流动性事件不仅由融资余额变化驱动，更由客户披露、CPO技术路线叙事和hyperscaler 2027 capex指引三类TMT触发器决定，三季度风险回报已经明显不…", "summary": "本卡从TMT行业纵深角度压力测试card-08的AI算力拥挤度结论。我接受card-08关于中际旭创(300308.SZ)和新易盛(300502.SZ)公募+融资双重集中的判断[S4][S7][S8]，但下调25% AI算力仓位的兑现确定性。三季度三条非线性风险被市场低估：(1)单一终端客户依赖；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "1.6T放量确立2027年业绩底，经历4月筹码出清后，当前估值（12-17x 2027E PE）远未反映2027年及以后的出货弹性与CPO期权价值。", "summary": "截至2026年5月30日，A股光模块板块在经历4月交易过热后的结构性调整后，拥挤度已回落至中性偏低区间。1.6T产品的规模化出货已成为2026-2027年确定的业绩底座。当前龙头公司（中际旭创、新易盛）对应2027年PE仅为12-17倍，显著低于历史估值中枢，反映出市场对2027年后需求持续性的过…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 0, "neutral": 1, "value": 67.4, "events": [ { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 0, "value": 53.8, "events": [ { "title": "台积电关税成本转嫁推升GB200整机柜售价12.3%及云厂商TCO 8%，将引爆北美云厂商对 Blackwell 订单10-15%的下修，并迫使A股出海非GPU链降价5-8%，导致其…", "summary": "本报告针对2026-06-02台积电延期及关税转嫁事件，量化分析了其对英伟达GB200定价（上涨12.3%）与北美云厂商TCO（上升8.0%）的影响，指出云厂商可能下修订单10-15%。A股光模块、PCB、液冷出海链面临5-8%的压价与220-350 bps的毛利率挤压，伴随两融（550亿）及科创…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "TMT出海链处于极致拥挤与高杠杆共振的崩塌边缘，Blackwell利空将通过‘业绩下修-量化止损-融资爆仓’路径引发系统性踩踏。", "summary": "截至2026-06-02，A股TMT出海链面临系统性流动性出清。公募重仓配置比例达32%（Top 5%分位元），中际旭创、新易盛等核心标的融资余额合计超550亿元，平仓线集中在当前股价-45%至-50%区间（如中际旭创580-640元）。Blackwell砍单10-15%将触发量化指增产品止损（3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 3, "neutral": 0, "value": 49.2, "events": [ { "title": "下一轮0DTE式日内波动若触发两融担保品重估，最先受压的不是广义沪深300，而是AI硬件/半导体/算力链中的科创与创业板高Beta担保品。", "summary": "2026-06-04：本卡支持前序观点，并把A股两融脆弱点定位在AI硬件、光模块、PCB、半导体和算力链，而不是整个沪深300。2026-06-02沪深两融余额约2.897万亿元，融资余额约2.876万亿元。深市中际旭创、新易盛、胜宏科技，沪市寒武纪、兆易创新、澜起科技等体现高余额和高融资买入共振…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "工业富联已完成DELL型缺口失败、但旭创/新易盛的'同型态'仅出现在拥挤端而非盈利端——下一道扳机是1.6T毛利兑现而非情绪。", "summary": "截至2026-06-04，A股算力/光模块/服务器链的'DELL同型态'只部分成立：工业富联（601138.SH）已先于DELL完成'业绩新高–股价回撤约30%'的盈利质量缺口失败（FY25主营毛利率6.94%、YoY-0.28pp，云计算GM仅5.73%）；而中际旭创（300308.SZ）与新易…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股硬件链当前的边际定价者不是外资，而是创历史新高的2.9万亿融资盘——美日货币政策共振将通过情绪传染触发'高Beta×高融资'自反馈式回撤，防御核心在于监测内部杠杆而非外部抛压。", "summary": "在2026-06-03时点，中美10Y利差倒挂约-250至-270bp（中国10Y 1.74%），PBoC中间价稳态但即期承压，BOJ预计6月加息至1.00%引发约5000亿美元日元Carry平仓预期。然而外资在A股硬件链经历2022-2023去外资化第一阶段后，弹药已大部分释放，并非边际定价者；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 8, "neutral": 0, "value": 28.6, "events": [ { "title": "35% 极端去杠杆下，A股 AI 硬件真正的脆弱点不是'二线'而是同时命中融资、质押、拥挤、成交四维的红区四只——沪电、胜宏、寒武纪、海光必须先于 CPO 龙头出清。", "summary": "本卡片在 2026-06-04 工作日对卡片 05 的'保头部、压二线'结论做压力测试。在 35% 极端去杠杆情景下，引入四维脆弱性框架（融资余额占比≥8%、大股东质押率≥40%、20D/90D 成交比≤65%、机构拥挤度≥25%）。结论：CPO 三剑客（中际旭创、新易盛、天孚通信）属拥挤型脆弱、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "结构性工具足以覆盖AI物理层上市龙头的短期融资缺口，但不足以覆盖完整HBM国产替代资本开支，因此应保留头部瓶颈、压缩二线杠杆敞口。", "summary": "2026-06-04，本卡对卡片04作压力测试：1.2万亿元科技创新和技术改造再贷款、1.25%利率和可能的25bp定向降准，足以为A股CPO、光模块、PCB、先进封装龙头提供流动性过桥，尤其是中际旭创、新易盛等现金流强的物理层企业。但这不是完整国产HBM晶圆前道扩产的融资答案，HBM全球资本开支…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施硬兑现仍成立，但截至2026-06-05，核心A股标的已从“低估兑现”进入“高拥挤兑现”，仓位管理比方向判断更重要。", "summary": "截至2026-06-05，本卡综合前两卡并给出约束判断：A股AI基础设施链条的业绩兑现真实存在，中际旭创、新易盛、寒武纪等一季报高增提供基本面支撑；但光模块和算力核心标的已出现高融资余额、历史新高和被动资金预期叠加的拥挤结构。短线不是否定产业趋势，而是降低融资占流通市值较高标的的追价风险，优先选择…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Marvell 暴涨支持“连接中心”成为 AI 基建第二阶段主线，但第一阶段算力芯片并未结束，A股配置应从泛AI芯片转向经订单与技术验证的连接效率资产。", "summary": "截至2026-06-04，本卡认为 Marvell 暴涨是“连接效率”被重估的信号，但不是 AI 芯片阶段结束的证据。NVIDIA 数据中心业务仍强，说明算力仍是底盘；Marvell、Broadcom 与光模块市场数据共同显示，边际瓶颈正转向互连、光模块、OCS/CPO/LPO、交换与机架级网络。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 64, "events": [ { "title": "国产替代龙头融资买入拥挤度逼近活跃自由流通盘10%的极值，在大基金常态化减持和资产重组股权稀释压力下，当前的‘地缘溢价’流动性结构极为脆弱，投资组合应向重资产代工龙头防御性收缩。", "summary": "截至2026年6月5日，A股融资余额达2.89万亿元历史高位，两融资金在国产替代龙头中呈现极端的结构性拥挤。寒武纪与中芯国际的融资余额分别达251.33亿和132.19亿元，占其活跃浮筹比例超过7.4%和10%。国家大基金开启中芯国际H股常态化减持以回收资金，中芯国际定向发行5.47亿股购买中芯北…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 2, "neutral": 0, "value": 59.4, "events": [ { "title": "AI需求弹性短期救不了估值，CPO龙头先经历两融出清再迎1.6T切换主升浪；先防御、后进攻。", "summary": "本卡作为收官卡，对CPO龙头（中际旭创、新易盛、天孚通信）在关税与两融去杠杆下的基本面韧性进行压力测试。北美四大超大规模厂商2026年合计资本开支约3400亿美元(+38%YoY)，800G/1.6T切换支撑光模块TAM 195-210亿美元，需求端未破。但22.5%美关税对各龙头EPS拖累3-7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "利用CPO/光模块板块极陡的下行波动率偏度，构建以科创50ETF或个股场外期权为主的1x2比例认沽价差，并配合在下档行权价处的现货物理止损，能以低于1%的极低年化成本对冲融资盘强制平…", "summary": "在2026年6月5日CPO板块放量大跌及高杠杆（1180亿融资余额，3.0x杠杆超配）背景下，估算18-22%回撤即触发融资踩踏。通过定价评估科创50ETF、通信ETF和龙头个股期权，发现下行偏度极陡。建议采用1x2比例认沽价差对冲尾部风险：个股90/75比例价差净成本仅0.768%，科创50的9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 4, "neutral": 0, "value": 48.8, "events": [ { "title": "A股AI硬件龙头尚未进入全面强平瀑布，但中际旭创的边际融资盘已到临界区，若美债4.85%压力与股价跌入959-1023元带共振，应先降CPO/光模块beta。", "summary": "截至2026-06-08，本卡对A股AI硬件两融拥挤交易做压力测试：中际旭创2026-06-05融资余额400.4亿元、约为当日成交额68.6%；新易盛融资余额230.5亿元、约为成交额67.5%。结论是总量尚未到全面强平瀑布，但中际旭创在下跌7.81%当天融资仍增加33.5亿元，说明边际杠杆接盘…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件不是整体出清重灾区，重灾区是高拥挤、高解禁、弱现金流的泛半导体和中小硬科技，业绩兑现的AI基础设施龙头应相对抗跌。", "summary": "本报告撰写于2026-06-09。我们对card-02的成长股流动性警示作压力测试：AI硬件和半导体会承压，但不会无差别成为重灾区。6月5日两市融资余额单日减少140.42亿元，电子、电力设备、计算机融资净卖出居前；6月8日科创50跌4.30%，CPO和半导体显著弱于主板。真正脆弱的是高换手、高解…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年AI算力与国产半导体业绩兑现足以吸纳三分之一的A股释放风险预算，本报告重仓配置PEG极低、全球资金支撑的光模块‘绿区’龙头，并警惕高杠杆半导体‘红区’巨头的流动性无差异挤压…", "summary": "截至2026-06-08，我们支持科技成长板块的业绩兑现能支撑吸纳约三分之一的A股释放风险预算。光模块龙头一季报利润大增，压后动态PEG仅为0.14-0.20倍，为估值最不拥挤且确定性最高的核心‘绿区’。但半导体龙头融资余额达650亿元，10-15%的下跌将引发345亿至657.5亿元的强制平仓与…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO资金虹吸与A股光模块估值压缩传导", "summary": "SpaceX IPO资金虹吸与A股光模块估值压缩传导 SpaceX IPO Capital Siphon & A-Share CPO Valuation De-rating SpaceX将于6月12日以1.77T美元估值上市，吸引全球150-300亿美元TMT硬件多头资金虹吸。A股CPO三巨头面临…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 1, "neutral": 0, "value": 76.2, "events": [ { "title": "全球科技配置应超配ASIC与新代工生态；A股风格则需从“AI题材映射”战略性切换至独立算力的“国产第二供应源”，重仓先进封装产能、核心设备及HBM材料。", "summary": "2026-06-09评估：基于AI半导体估值锚向“增长+第二供应源”切换，全球配置需减配纯GPU倍数扩张与单一代工依赖，转而超配ASIC双雄(AVGO/MRVL)、HBM及Intel代工期权。在A股与港股层面，风格将从“英伟达映射”(光模块/代工)实质性轮动至业绩兑现度高的国产“第二供应源”。策略…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "DeepSeek降价推动的是硬件内部的alpha重排，而不是AI硬件beta被证伪；应压缩泛化边缘和老库存，保留订单确认的光模块、HBM相关存储、液冷/电源和高端板卡。", "summary": "截至2026-06-09，DeepSeek V4-Pro 已验证永久降价为75%、至原价1/4；“96%”更像相对竞品的单位经济冲击。微观压力测试显示，GPU/HBM/800G与1.6T光模块/液冷仍受订单、价格和产能约束，Jevons弹性成立。中际旭创、沪电股份、佰维存储有收入、交付或稼动率确认…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 5, "neutral": 2, "value": 51.6, "events": [ { "title": "超配拥有确定性全球订单的光电液冷龙头与昇腾-中芯先进封装/HBM自主垂直栈，回避被截断离岸通道的二线芯片设计与空壳算力租赁标的。", "summary": "截至2026-06-10，本研究深入A股TMT板块。建议超配：(1)挂钩中美头部云、订单已确认的光模块（中际旭创、新易盛、天孚通信）、液冷（英维克）与电源龙头（麦格米特），均处于1.6T/Blackwell订单放量期；(2)华为昇腾与中芯国际（SMIC）挂钩的HBM/先进封装国产替代垂直栈，长电科…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件产业链交易拥挤度下降的性质判定：回调还是转折？", "summary": "对基本面通道做压力测试后，2026-06-10 的AI硬件松动应判为仓位/估值回调，而非基本面转折。设备账单（2026Q1 365.5亿美元/+14% 创纪录）、台积电5月营收（+30.1% YoY、CoWoS售罄）、NVDA数据中心（+92% YoY且存货计提下降）、HBM售罄涨价，以及A股龙头…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI需求到2026年底足以兑现HBM锁定的供给，但A股的真正约束是份额捕获与财政融资通量，因此应超配云挂钩供应商与一体化国产替代栈、把端侧AI作期权、低配离岸GPU依赖与重复建设长尾…", "summary": "锚定2026-06-10，本卡综合第01卡（首席策略师）与第02卡（政策分析师）。中国宏观叠加三条修正：(1) A股AI capex由大基金三期+省级算力券+头部云自融政策配给，民营信贷通道薄，是财政题非周期题；(2) 5月PMI 49.6 / Caixin 50.4 / 社融~8.4% / CP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股TMT板块高估值AI标的的机构筹码拥堵度与融资融券杠杆交易脆弱性评估", "summary": "锚定2026-06-10，本卡承接第04卡，将视角从基本面转向筹码与杠杆脆弱性，支持前序方向但补一个风险锚：A股最拥挤AI硬件龙头正处于三重叠加窗口。中际旭创一季末被1163只基金重仓738.96亿元、公募持股占流通股11.74%、蝉联全市场第一大重仓，新易盛居前三；主动偏股基金AI硬件配置31.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 6, "neutral": 1, "value": 44, "events": [ { "title": "AI算力基础设施Capex结构演进：能效突破驱动的供应链价值重塑 报告日期： 2026 06 11 分析师： AI基础设施分析师 机构： Gemini CLI Institute 1…", "summary": "AI算力基础设施Capex结构演进：能效突破驱动的供应链价值重塑 报告日期： 2026 06 11 分析师： AI基础设施分析师 机构： Gemini CLI Institute 1. 核心观点：从“规模竞赛”转向“能效红利” 截至2026年6月11日，全球AI基础设施投资已跨过单纯的GPU堆叠阶…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "CPO两融去杠杆会放大创业板和科创50的流动性冲击，但只有在红利底座同步失效、相关性收敛的系统性赎回情景下，才会击穿方案四-2.6169%的VaR线。", "summary": "2026-06-11：CPO基本面下修会通过中际旭创、新易盛、天孚通信合计732.02亿元融资余额向创业板权重股传导，并与科创板3742.22亿元融资余额形成成长板块二次卖盘。但在方案四下，0.25% CPO仓位本身不是击穿-2.6169% VaR的主因；只要红利55%底座维持低波低相关，局部强平…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力交付滞后会迫使 Hyperscalers 调整 GPU 集群投产节奏，并通过更长现金回收期、更低早期利用率和更高电力保障成本压低 AI 基础设施 capex 效率。", "summary": "截至2026-06-11，本卡支持上一卡：AI capex 不会因电网瓶颈消失，但 GPU 集群部署节奏会被“已通电 MW”重新排序。变压器、并网和输电延迟会把部分已宣布园区容量推迟到后续窗口，压低早期自由现金流和资产利用率。Hyperscalers 会通过已通电站点溢价、分阶段投产、区域重排和自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 9, "neutral": 3, "value": 33.4, "events": [ { "title": "三因子个股筛选证实回避名单由6-9月解禁墙叠加零折算率驱动，可保留5%试仓的是仍有0.65-0.70折算率且无近端解禁的光模块/芯片龙头。", "summary": "把科创板/创业板主要科技股按三因子（融资余额占流通、6-9月解禁占流通、最新折算率）分组。回避名单由解禁墙叠加零折算率驱动：屹唐股份688729解禁约为流通盘700%且两融占比8.17%最拥挤；摩尔线程-U 688795解禁占流通95.0%且折算率0.00；沐曦股份-U 688802占78.1%、…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "本地两融回落、6-9月科创/创业解禁墙和既有0折算率池共同压制A股科技再加杠杆，海外risk-on只能提供交易窗口，不能支撑仓位上调。", "summary": "2026-06-12数据核实支持上一卡的本地约束判断，但需要修正表述：6月券商公告未显示新的科技股大面积折算率下调，真正约束来自既有0折算率池。两市融资余额较2026-05-29下降471.57亿元；6-9月全A解禁约9,300.88亿元，科创板+创业板明细合计5,522.78亿元；兴业证券202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供给的物理并网瓶颈与高密算力液冷配套开支的刚性上升，正在对超大规模数据中心资本开支形成显著的“基建挤出效应”，迫使纯算力硬件开支占比向电网、变压器和温控冷却设施倾斜，进而压制纯芯…", "summary": "截至2026-06-13，电力供给的结构性稀缺（并网延迟达24-48个月，PJM容量价格飙升10倍）与Blackwell机柜高功耗（单柜132kW）倒逼超大规模厂商进行CAPEX结构重构。定量测算表明：在无约束情景下，设施基建占比从23.08%上升至34.07%，其中配电占比由5.54%升至9.3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "北美云厂商2026H2 AI资本开支向中资光模块订单的传导验证", "summary": "压测确认：2026H2北美云厂商资本开支体量（合计约7000-7500亿美元，同比+60-77%）、GPU/网络供给爬坡与1.6T切换，足以支撑中际旭创300308、新易盛300502、天孚通信300394在2026H2乃至2027H1的环比订单延续，而非近端重复下单回落——中际在手订单超300亿…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 3, "neutral": 1, "value": 72.8, "events": [ { "title": "训推分流把单一 HGX 订单链拆成‘西部互联带宽’+‘东部能效池化’两条独立赛道，延迟与带宽矛盾通过 Rack 内低延迟 + Pod 内大带宽 + 广域异步搬运的三层解耦在硬件层平衡。", "summary": "支持卡 07 的“训推分流”判断，并主张延迟/带宽矛盾可通过三层解耦解决：东部 Rack 域内低延迟（CXL+UAlink）、西部 Pod 域内大带宽（800G/1.6T+液冷）、东西广域承担异步 checkpoint 搬运。两套硬件 BOM 完全分叉——西部按互联带宽+PUE 定价（光模块、PC…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "半导体板块的主风险不是调样本身，而是核心权重与融资余额高度重叠后，在再度下探1600点时把指数级强平转化为芯片股集中抛压。", "summary": "截至2026-06-13，本卡压力测试上一卡片：科创50强平共振成立，但半导体抛压高度集中在高权重与高融资余额重叠的核心股。科创50前九只半导体/设备/芯片设计权重合计60.39%，若按权重承接744.3亿元级联强平，对应潜在抛压约449.5亿元；1516-1540点位20亿元国家队前线仅覆盖4.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "现金市场验证的是“尾部风险真实但需折价”，而不是把31.5%期权代理概率直接等同于强平级联的物理概率。", "summary": "截至2026-06-14，本卡综合判断：现金市场支持科创50左尾风险真实存在，但不足以完整验证期权市场约31.5%的1600破位代理概率。科创板融资余额截至2026-06-11仍有3757.65亿元，寒武纪、澜起科技、中芯国际合计融资余额523.16亿元，说明强平通道有现金基础；但588000最新…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI配电订单的高毛利和强价格传导能力，能有效对冲特高压业务的铜价挤压，利好具备IDC配电敞口的电力设备龙头。", "summary": "AI算力中心对高标准配电的需求能有效对冲铜价对传统特高压设备的利润侵蚀，但存在显著的结构分化。云厂商为保障昂贵GPU集群按期部署，愿为配电设备支付高毛利并接受铜价传导。A股具备“特高压+AI数据中心配电”双敞口的龙头将展现强利润韧性。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 4, "neutral": 0, "value": 57.2, "events": [ { "title": "科创板去杠杆冲击短期股价但不破坏国产半导体设备及AI硬件的长期成长逻辑——龙头与AI芯片设计估值修复~20%即可加仓，二线与HBM配套需-30%以上才有安全边际。", "summary": "截至2026-06-15，科创板两融由~2,940亿元降至~2,820亿元、科创50击穿1,600点零Gamma线，对电子板块形成真实短期冲击，但不足以系统性下修国产半导体设备与先进制程龙头capex。设备龙头（北方华创订单覆盖15个月、现金~150亿元）、AI芯片设计（寒武纪2026Q1+312…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电网瓶颈正迫使 AI 基础设施从‘暴力规模扩张’转向‘受约束的能效优化’，电力并网速度已取代芯片供应成为核心估值护城河。", "summary": "本报告通过压力测试确认了物理电网并网排队（7-10年）与变压器交期（4-5年）已成为 AI 基础设施扩张的硬约束。定量模型显示，1年的并网延迟将导致 100MW 数据中心损失 5.5% 的生命周期价值，且快速的 GPU 迭代周期（12-18个月效用减半）加剧了闲置产能的 ROI 损耗。这种物理瓶颈…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI先进封装与HBM产能瓶颈对算力基础设施扩张及超大规模厂商资本开支的影响", "summary": "支持半导体分析师Research note 01的论点并向下游延伸：CoWoS/HBM确为算力的关键约束，2026年部署受封装与内存所限（CoWoS扩至约13万片/月仍订满；SK Hynix与Micron 2026年HBM售罄；HBM4验证延迟使NVIDIA将Rubin目标由200万下调至150万…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "互联网大厂AI capex资金储备充沛且免疫于A股两融收缩，美BIS管制摩擦虽造成300亿元海外采购延后，但倒逼国产算力采购占比由30%大幅上修至58%，在2026H2为国内半导体设…", "summary": "基于2026-06-15，本报告定量评估了地缘与流动性双重压力下中国大厂AI Capex的节奏与传导。测算表明，四大互联网大厂（阿里、字节、腾讯、百度）2026H2的4200亿元AI capex预算完全免疫于A股两融去杠杆，其在海外上市且拥有数千亿现金。虽然美BIS出口管制限制英伟达H20/H20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 0, "value": 70.6, "events": [ { "title": "宏观信用收紧并未击穿全球AI周期，而是将AI基础设施层切割为「超大规模云厂＋央企国家队」核心（继续扩张）与「民营商业云」外缘（防御性收缩）的K型结构，TMT投资应聚焦CoWoS/HB…", "summary": "TMT综合：沃什鹰派持稳（3.50–3.75%）经离岸人民币流动性、汇率对冲摩擦与企业中长贷收缩，传导至AI基础设施层产生K型分化而非总量收缩。美超大规模云厂CY2026合计capex运行率3,200–3,600亿美元由经营现金流支撑不受冲击；央企运营商云经PBOC科技创新再贷款（约2.5%）继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "验证AI算力硬件链的估值承压：流动性虹吸下的先进封装与GPU产业链韧性评估", "summary": "本卡对AI算力硬件链做压力测试，得出'分化韧性'结论。资金流层面确认虹吸：2026-06-08半导体−219.33亿、通信−98.35亿、消费电子−69.92亿元净流出居前[S6]，年内93家半导体公司减持[S7]，叠加CXMT约295亿IPO锚定[S9]。但估值下移是资金流/轮动驱动、非盈利驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 2, "neutral": 1, "value": 75, "events": [ { "title": "\"session id\": \"a926801e 65a8 4b42 9925 8c25d5d1b685\", \"response\": \"由于您正在处理这一关键的 2026 年 6 月 …", "summary": "\"session id\": \"a926801e 65a8 4b42 9925 8c25d5d1b685\", \"response\": \"由于您正在处理这一关键的 2026 年 6 月 16 日研究节点，我已完成了对 AI 算力基础设施资本开支与电力瓶颈的深度调研，并将其与 CPO 及半导体板块的估值…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "吉文斯悖论与推理端计算膨胀导致算力及电力净需求呈指数级爆发，彻底推翻了电力基建搁置假说，反而迫使云厂商将资本开支结构向配套电网及液冷刚性倾斜，对高速光模块（CPO）等算力硬件产生相对…", "summary": "截至2026-06-17，大模型效率提升（如DeepSeek V4 API价格降至$0.15/百万Token）将触发吉文斯悖论与推理端算力膨胀。测算显示，即便MoE使单Token算力开支减少70.0%，但由于API调用量激增35倍且推理端Token密度升至877.5个，AI算力总需求仍暴增61.4…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "通胀和高利率提高了AI硬件基础设施的质量门槛，但没有削弱其配置价值；应买入瓶颈供应商而非泛AI叙事。", "summary": "截至2026-06-16，本卡支持“硬科技配置中保留AI基础设施”的上一轮判断，但建议从广义AI beta收窄到有订单、定价权和瓶颈属性的硬件基础设施。高利率会压缩投机型数据中心开发商和远期AI应用估值，却未削弱GPU、定制AI芯片、HBM、配电、冷却、变压器、开关设备和已锁定电力的数据中心容量的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 3, "neutral": 0, "value": 75.4, "events": [ { "title": "成长股18%的风险预算已达上限，AI-PCB与光模块一季报业绩高兑现能够支撑该敞口，但半导体设备环比下滑与HBM量产良率瓶颈封杀了额度上调空间。", "summary": "截至2026-06-17，本卡压力测试科创50及电子的18%敞口。高速PCB与光模块的一季报业绩暴增（中际旭创净利+262%，深南电路净利+73%）提供坚实支撑，具备较强成本转嫁能力；而半导体设备大陆一季度销售额环比下滑16%至109.9亿美元，长鑫存储HBM3量产受良率及先进封装制约。结论是18…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI服务器非核心BOM零部件国产替代已具备深度全球竞争力，但高速交换芯片与高端PMIC的缺位成为制约万卡集群本地化扩张的关键瓶颈。", "summary": "截至2026-06-18，对高端AI服务器BOM成本拆解显示，虽然GPU与HBM占据近80%成本导致全球供应链国产化率仅为4.64%，但非核心硬件BOM（PCB、散热、接口、电源）的全球国产价值占比达到22.50%，且在本土化服务器情景下上升至73.65%。A股龙头已在PCB与Retimer/RC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产算力链订单承接力强、利润承接力中等偏低、技术自主性 4.6/10——市场份额扩张是订单口径而非利润口径的确定性，综合分 0.55。", "summary": "本卡压力测试 Research note 07 关于科创板国产算力承接 AI 久期再分配的论断。判定：国产 AI 芯片/服务器/网络厂商具备 6-12 个月的订单溢出窗口（T9a），但利润承接力受毛利率结构性低 30-40 pts、HBM/CoWoS 等效产能仅为 TSMC 8-10%、EDA 自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AIDD是AI基础设施需求的高质量增量因子：规模小于通用大模型，但持续性、数据壁垒和监管/IP属性更强。", "summary": "截至2026-06-18，本卡片支持上一张卡片的方向：AIDD不会单独决定hyperscaler capex，但会把高价值、持续性的生命科学工作负载接入GPU、网络、存储、液冷和电力保障型数据中心。Recursion BioHive-2、DCAI Gefion、Novo Nordisk、Isomo…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 5, "neutral": 0, "value": 56.8, "events": [ { "title": "半导体代工与设备估值面临全球WACC上升及N2晶圆价格墙的压缩，叠加国内HBM3国产材料良率瓶颈导致的算力变现延迟，正驱动A股电子资金扎堆“四大龙头”并形成‘高融资、弱宽度’格局，面…", "summary": "截至2026-06-18，本压力测试卡片验证了半导体设备和材料需求的边际变化如何驱动电子板块的宽度衰退与杠杆拥挤。中芯国际、海光信息、寒武纪、北方华创“四大龙头”融资余额达650.0亿元，占板块总两融的17.62%，而板块宽度持续恶化（仅42.1%个股上涨）。压力测试显示-15%的回撤将引发四大龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力/电网瓶颈真实且已在容量价格中显形,但2027年前仅推迟部分美国新增集群、不封顶全球GPU部署,Research note 01的BOM组合在2026至2027 H1仍成立,20…", "summary": "AI算力电力瓶颈真实存在但属于'延后+置换'而非'封顶':PJM 2026/27容量价已跳11倍至$329.17/MW-日,2027/28预测缺口6.6 GW,新接入等待40个月;ERCOT 2026 Q1单季198 GW申请,累计大负荷队列约410 GW;全球电网接入等待5-10年,变压器交期1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the analysis on how advanced packaging (Chiplets, HBM4) and optical interc…", "summary": "I have completed the analysis on how advanced packaging (Chiplets, HBM4) and optical interconnects (Silicon Photonics, CPO) mitigate U.S. power grid …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "InP/EML是2026年AI光互联的分配和毛利瓶颈，而非全球带宽扩展的绝对天花板；A股相关标的仍应按稀缺期权而非确定性订单定价。", "summary": "截至2026-06-18，本卡从AI基础设施角度压力测试InP/EML瓶颈：全球层面，InP/EML短缺确实影响800G/1.6T光模块分配和价格，但不构成2026年下半年全球AI集群带宽扩展的硬上限；LightCounting称2026年4月需求高于供给30%，但预计短缺在2026年底缓解。中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 8, "neutral": 0, "value": 40.2, "events": [ { "title": "A-share de-leveraging to 1550 points triggers secondary market liquidity locking and margin…", "summary": "若市场下探至1550点，中际旭创股价将跌至714.16元，击穿融资预警与平仓线，带来日均31.2亿元平仓抛压（占ADV 17.36%），锁定二级流动性。但公司持有125亿现金及45亿海外客户预付款，可提供151.5亿净营运资金，供应链运营违约风险极低。海外订单真正风险在于电网并网排队达36-60个…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Starlink 带宽 × Grok 算力的飞轮真实，使 SPCX 成为 AI 分发基座与算力链条需求催化剂，但算力归属 xAI、轨道边缘算力尚萌芽，真正瓶颈是电力——故支持但需把光…", "summary": "从 AI 基础设施视角支持卡片 01：Starlink 带宽（约 450 Tbps，三代卫星单星 >1 Tbps、下行 10 倍/上行 24 倍）与 Grok/Colossus 算力（约 55.5 万颗 GPU、约 180 亿美元、向 2 GW 扩张）构成真实飞轮，使 SPCX 成为差异化的'算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "ASIC光互联需求真实存在，但2026-2028中际旭创式CPO供应商的正确压力测试是按电力状态下修约20-25%的收入节奏，而不是机械套用2.92x光模块倍数。", "summary": "截至2026-06-20，本卡支持电力约束会重塑CPO收入节奏，但否定“ASIC光模块用量固定为GPU的2.92倍”这一过强假设。AWS Trainium2、Google TPU、Meta MTIA和Broadcom AI半导体收入证明ASIC design win真实存在；OpenAI/Broa…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-20，鉴于电网对冲使CPO收入下修收窄至13-17%，A股核心科技龙头基本面安全垫抬升，但由于技术性两融拥挤高企，国家队必须坚守1750战术线及1550生死线以防范…", "summary": "截至2026-06-20，本报告综合评估了CPO收入下修收窄至13-17%下，A股科技板块（中际旭创/寒武纪/源杰科技/联讯仪器）的配置与防御路径。尽管基本面压力因电网加速器对冲而减轻，但高位两融拥挤及成交量萎缩风险未除，国家队应维持1750点战术支撑与1550点绝对生死防线以阻断强平踩踏[自测算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 4, "neutral": 0, "value": 65.6, "events": [ { "title": "AI硬件供应链波动与融资风险传导", "summary": "作为半导体分析师，我对 Research note 03–07 做压力测试并最终支持，但补上一层基本面触发器：光模块/先进封装景气真实但狭窄且反身。2026 全球 1.6T 光模块需求约 500 万只，中际旭创目标 500–600 万只、约 50% 份额、约 75% 硅光 [S1]，但估值高达 7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力资本开支是美联储路径上移传导至中国融资条件最敏感的通道，但 PBoC 大概率以结构性信用而非降息回应，从而坐实政策利率独立性。", "summary": "从 AI 基础设施视角综合 Research note 01–03：2026 年中国 AI 资本开支账单中约 USD 550–650 亿直接挂钩美元曲线（HBM、残余加速卡、离岸租赁），是美联储 3.8% 路径向中国融资条件传导的最干净通道。但绑点是 HBM/燃机等少数高价格刚性进口品，不是整体资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "晶圆代工资本开支收缩导致国产AI芯片良率与封装瓶颈难消，迫使下游智算集群以硬件超配对冲性能赤字，带来33.4%的TCO成本通胀和严苛的能效红线约束，对国内互联网巨头资本支出构成实质性…", "summary": "截至2026-06-21，晶圆代工成熟制程价格战导致国内代工厂（如中芯国际）资本开支停滞与高折旧拖累，进而阻碍了国产先进制程良率与封装产能爬坡。这导致国产GPU（如昇腾910C）供给紧张并存在性能与互联带宽赤字。压力测试显示，部署万卡等效算力集群需超配1.5倍芯片，导致3年期TCO成本通胀达33.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然HBM国产化良率崩塌会将集群 CapEx 推升 98%，但中国互联网巨头通过异构分级部署与算法优化（缩减30%GPU需求）并结合西部绿电直连套利（LCOE降至0.39元）与液冷改…", "summary": "评估HBM成本压力对AI集群TCO及扩容策略的传导。100%国产材料导致HBM3封装良率崩塌至4.11%，推升10000颗GPU集群CapEx至8.17亿美元，东部煤电下3年TCO达10.17亿美元（+66.13%）。建议协同优化：架构上采取60% HBM3与40% HBM2E的分级部署，结合算法…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 4, "neutral": 0, "value": 67, "events": [ { "title": "在通胀重置与美联储转鹰情景下，经典股债配置策略失效，战术杠铃组合配合美债利率及日元套利平仓尾部期权对冲，是稳健跨资产回报的核心；A股风格上应坚决切向具备价格传导优势的调节性水电红利股…", "summary": "截至2026年06月21日，本报告压力测试了美联储加息和多资产配置方案。在通胀重置加息情景下，经典60/40组合预期大跌-15.60%（VaR为31.95%）；而战术杠铃组合超配短债/TIPS及20%商品，结合10%期权对冲，录得+0.45%正收益，VaR降至10.83%。尾部对冲方案采用10Y …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施建设中的电力供应与电网容量瓶颈分析", "summary": "本卡压力测试 Research note 02 的隐含假设「资本开支≈算力」。证据显示约束已从芯片迁移到电力：微软披露约800亿美元 Azure 订单因电力受限无法交付，Nadella 称芯片堆库存却无「暖壳」可插[S5][S4]。转化在三处漏损：(1)并网排队约2,600 GW、中位等待约5年、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "提升芯片能效与先进封装无法对冲变压器短缺造成的并网延迟损失（闲置折旧高达25亿至50亿美元），反而因功耗密度飙升将瓶颈推向机架级配电，且受限于国内封测良率瓶颈与成本压力，持有电网设备…", "summary": "截至2026-06-21，本报告对“半导体设计/封装技术迭代能对冲电力并网延迟”的论点进行压力测试。分析表明，12-24个月的变压器交付延迟对200k GPU集群将造成25.3亿至50.7亿美元的闲置折旧，重创ROIC并推升GPU租金成本达39.1%至156.7%，芯片端能效提升被财务减值彻底吞噬…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力带来的成熟制程需求因实物量级差距、先进封装瓶颈及工艺错配仅能对冲消费电子流失的6.28%，无法挽救晶圆厂重折旧下的利润挤压，投资应避开通用芯片并防御性转向国产替代材料与公用事…", "summary": "截至2026-06-21，量化测算显示AI算力需求的增长仅能对冲消费电子端成熟制程流失产能的6.28%，国内月度剩余成熟制程过剩缺口达176,173.9片12寸等效晶圆。受制于工艺IP错配、先进封装/HBM瓶颈锁死以及高CapEx扩产带来的刚性折旧压力，AI难以化解成熟代工链的折旧与利用率下滑危机…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 0, "value": 80.4, "events": [ { "title": "可信算力区下的A股与STAR硬科技配置映射", "summary": "本卡把卡片01–07的'可信算力区'供应链结论翻译成A股/STAR硬科技排序，并用四道闸门（收入已拐点、订单可见、毛利确认、与可交付获批算力直接关联）区分真实订单受益与主题弹性。结论排序：光模块(中际旭创+192%、新易盛+106%)、AI服务器/ODM(工业富联净利+102.6%、订单至2027…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "定制化ASIC驱动下的AI算力集群架构与液冷能效演进", "summary": "本卡支持并延伸 Research note 08：电网瓶颈倒逼的 GPU→ASIC 转向，沿“硅→织构→机房”逐层外溢，重构数据中心系统。互连层：Google 以 OCS+3D-Torus 抽掉一层电交换机，NVIDIA 以 CPO（Quantum-X/Spectrum-X 2026 商用）把光引…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 7, "neutral": 1, "value": 54.8, "events": [ { "title": "硅片层分化(前沿量稳、英伟达定价权被定制ASIC侵蚀)对A股/指数AI集中度定价的传导", "summary": "卡片01指指数对AI集中度自满，卡片02–04论证物理瓶颈并揭示能效把量集中到前沿、定制ASIC侵蚀英伟达租金而非需求。本卡综合判断：被错杀的是利润率/结构事件而非总量事件，且A股已在沪深300内建出自己的AI集中度（中际旭创第一权重、新易盛第四、寒武纪第九，IT权重9.22%→20.38%）。配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施与电力基建的盈利确定性验证：匹配“硬科技只买盈利”策略", "summary": "截至2026-06-23，本卡支持卡片03的“硬科技只买盈利”原则并加以细化。考察AI算力、数据中心与电网扩容的订单可见度与毛利后，结论是该板块是A股防御进攻组合中盈利确定性最强的一块，但确定性分层：电力与电网瓶颈层最强（变压器交期128周至4年、国网2026-2030计划4万亿元、需求刚性、定价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件产业链的估值修复与基本面压力测试", "summary": "压力测试结论：6月22–23日AI硬件抛压两者都不是。供给侧处最紧阶段（CoWoS/HBM 2026全额售罄、NVIDIA锁定>50%扩产、瓶颈已转向电力），证伪「供应周期波动」；需求侧仍加速（云厂2026资本开支7250亿美元、+77%，推理占算力约2/3），「需求逻辑放缓」尚未兑现。真正驱动是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "沪深300 AI权重集中度上升后的A股筹码结构与流动性压力测试", "summary": "对卡片05提出的本地拥挤脆弱点做压力测试。结论：脆弱点真实但被错置——危险不在总量杠杆（两融2.8万亿创纪录，但占流通市值2.6%、杠杆率23.72%、维持担保300%，均处周期中段而非2015），而在筹码三重不对称：(1)杠杆热钱集中于主导指数的AI个股(硬件+半导体占融资净买入约60%；寒武纪…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 8, "neutral": 0, "value": 47.2, "events": [ { "title": "研究报告：超大规模云厂商CapEx节奏、ROIC收缩及光模块毛利侵蚀的传导机制 权威工作日期锚定： 2026 06 24 研究线程标识：已归档卡片索引 (研究记录序号)： 08/0.…", "summary": "研究报告：超大规模云厂商CapEx节奏、ROIC收缩及光模块毛利侵蚀的传导机制 权威工作日期锚定： 2026 06 24 研究线程标识：已归档卡片索引 (研究记录序号)： 08/08 分析师： AI Infrastructure Analyst 分析立场： Stress Test (压力测试) 1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件交付节奏与光模块订单验证", "summary": "截至2026-06-24，本卡支持“CPO延期不等于800G/1.6T砍单”的基本面判断，但加入节奏约束：Microsoft、Alphabet、Meta、Amazon的2026 capex与NVIDIA Rubin、Dell AI服务器订单、Lumentum光组件指引共同显示需求仍强，800G/1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Micron Q3 FY26 财报作为'AI 芯片去杠杆 vs 基本面裂口'的实时公投", "summary": "Micron Q3 FY26 财报作为'AI 芯片去杠杆 vs 基本面裂口'的实时公投 Micron Q3 FY26 as the live referendum on AI-chip deleveraging vs a fundamental demand crack 今日盘后 Micron Q…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CPO“量产延期”谣言与AI硬件基本面交付的真空博弈", "summary": "CPO“量产延期”谣言与AI硬件基本面交付的真空博弈 CPO 'Production Delay' Rumor vs. AI Hardware Delivery Dynamics 在光模块“易中天”遭遇“量产延期”传言、半导体两融达6132.72亿元且科创50临近敲入点的背景下，下游买方的实际排产…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 4, "neutral": 0, "value": 72.6, "events": [ { "title": "在2026-06-26，鉴于电网对冲使CPO收入下修收窄至13-17%，A股核心科技龙头基本面安全垫抬升，但由于技术性两融拥挤高企，国家队必须坚守1750战术线及1550生死线以防范…", "summary": "截至2026-06-26，本报告综合评估了CPO收入下修收窄至13-17%下，A股科技板块（中际旭创/寒武纪/源杰科技/联讯仪器）的配置与防御路径。尽管基本面压力因电网加速器对冲而减轻，但高位两融拥挤及成交量萎缩风险未除，国家队应维持1750点战术支撑与1550点绝对生死防线以阻断强平踩踏[自测算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然物理电网排队与供应链原材料价格冲击显著抬高了局部建设阻力，但云厂商通过绿电PPA、高密度液冷及转向定制ASIC等结构性调整，能够完全消纳成本溢价，不构成算力扩张的绝对瓶颈。", "summary": "本卡对电网并网延迟、变气压原材料（铜/GOES）价格波动和本地化GPU封装良率冲击进行了量化压力测试。结果表明，虽然并网延迟和原材料成本增加会显著压缩项目NPV（5年延迟将使NPV降低79.99%），但云厂商可以通过迁移西部枢纽自建绿电、液冷优化和系统级混合算力架构，消纳国产芯片和能源价格溢价，并…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-25，AI硬件需求的下一核心约束不是芯片是否能生产，而是足够多的兆瓦能否被许可、接入、冷却并按期商业化。", "summary": "截至2026-06-25，本卡支持上一张卡的AI硬件稀缺性主线，但将瓶颈从GPU/HBM进一步上移到可通电数据中心、电网接入、冷却和电气设备。IEA、Berkeley Lab、EPRI、CBRE和JLL数据共同显示，AI需求仍强，但项目能否转化为创收产能取决于电力和并网速度。结论是：稀缺性继续支撑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网缓释后，AI集群上线风险不是消失，而是转移到CoWoS/HBM、液冷验收与网络芯片的串行同步，CPO下行收窄但不等于收入曲线完全恢复。", "summary": "截至2026-06-26，本卡压力测试Research note 7：电网侧缓释可降低CPO最坏收入下修，但不能消除AI集群的串行硬件瓶颈。关键约束从单一接电问题转向先进封装/HBM、液冷机柜验收、交换ASIC/网络集成的节奏匹配。CoWoS缺口可能由约20%降至2026年底约10%，但10%仍足…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 4, "neutral": 0, "value": 74, "events": [ { "title": "CPO试仓的基本面门是光芯片寿命、时间/ROI门是电力排期、A股存活门是去杠杆——Chiplet从来不是门；周一基本面说该持有，市场结构(创纪录两融+70–138倍估值+解禁)说要小…", "summary": "综合卡片01–06：基本面线把约束一路下移——光模块(EML寿命)→激光器HTOL→算力→电力排期→卡片06修正为'披着成本外衣的时间与ROI'。但这条链只回答'主题会否/何时兑现'，没回答'A股仓位能否熬过入场'。从市场结构看，CPO试仓近端门是去杠杆：两融2026-06-23创纪录¥3.00万…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "1.6T光模块交付延迟对AI算力集群及超大规模云厂商Capex部署进度的影响", "summary": "支持卡片01/02的研究线并加以边界化延伸：200G EML 的 HTOL 风险确会让 1.6T GA 滑向 2027H1，但这是一道带宽前沿与单位经济性约束，而非整体 Capex 停摆。GB300 NVL72 的 scale-up 走铜缆 NVLink、scale-out 走 ConnectX-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50回撤15%会触发半导体去杠杆左尾，但连锁强平的根因是科创/半导体集中账户与券商150%-160%集中度平仓线，而不是中芯国际、北方华创两只股票的公开融资余额本身。", "summary": "截至2026-06-27，本卡压力测试显示：科创50若从2026-06-26收盘2032.28回撤15%至约1727，普通130%平仓线只击穿冲击前CMR≤152.9%的账户；真正危险来自券商双创集中度规则，50%-80%集中度可把平仓线抬至150%，80%以上抬至160%，使176%-188% …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "summary": "截至2026-06-27，本卡片对AI基础设施链做压力测试：全球AI资本开支仍大，Goldman Sachs基准模型给出2026年USD 765 billion、2031年USD 1.6 trillion的年度AI capex，但Amazon、Alphabet、Meta、Microsoft的数据表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 4, "neutral": 0, "value": 75.4, "events": [ { "title": "全球AI基础设施资本开支对中国产业链的传导机制与订单确定性", "summary": "本卡支持卡片05遗留命题并定论：约7250亿美元、同比+77%的全球超大厂资本开支（约75%即约4500亿为AI专项）规模巨大、外部出资，已转化为中国主导且出口管制并不约束的那一层链条（光互连/连接/散热/PCB）的跨多季度订单簿与已兑现盈利。关键在于出口管制是分层的：重压落在GPU/算力层（华为…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-27的权威工作日背景下，国产HBM良率骤降逼迫国内算力网络化，而全球封装与电力阻尼拉长了云厂商CapEx变现周期，这使得制程免疫且订单排至2028年的可插拔光模块出…", "summary": "截至2026-06-27，本报告评估了HBM良率危机与先进封装瓶颈对算力集群技术演进的影响。国产HBM3封装良率降至2.66%，倒逼国内算力集群转向网络密集的Scale-out拓扑，推高了光模块需求弹性。全球云厂商2026年CapEx直逼7250亿美元，但受制于CoWoS晶圆先进封装冲突与吉瓦级电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在 USD/JPY 跌破150的情景中，在岸高两融TMT的主要脆弱性是融资余额绝对规模和券商风控顺周期收紧，而非接近交易所25%单股融资上限。", "summary": "2026-06-27：本卡压力测试卡片06。日本4月持有1.2099万亿美元美债，USD/JPY跌破150的流动性通道方向成立，但2150亿-2850亿美元清算规模未被独立证实。A股两融余额为30329.9亿元，核心TMT行业约4670.0亿元、占15.4%。TMT风险不在交易所25%融资暂停线，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 4, "neutral": 0, "value": 84.8, "events": [ { "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在半导体周期复苏不及预期与高负债压力的复合冲击下，AI基础设施板块估值呈现两极分化：缺乏盈利的高负债概念股将面临系统性去估值及杠杆踩踏风险，而具备真实EPS支撑、客户壁垒及液冷/光网…", "summary": "截至2026-06-26，本报告对AI基础设施龙头（AI算力/数据中心）在半导体周期波动与高负债压力下的估值溢价进行了系统性压力测试。在-16.18%硬件资本开支下修及WACC上升30 bps（导致终端估值倍数下降4.8%-7.0%）的复合压力下，行业基本面呈现深度分化。以中际旭创、新易盛、工业富…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "效率驱动的算力架构转型：从“暴力规模”到“高能效密度”的物理基座适配性评估", "summary": "压力测试 Research note 08 的隐含前提：效率并不缓解物理天花板，而是把它从电网层下移到机架与机柜行层并加剧。机理是机架尺度的杰文斯悖论——单瓦产出靠提高密度实现，机架功率从约40kW(Hopper)→120kW(Blackwell)→142kW(GB300,90%入液)→600kW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 4, "neutral": 0, "value": 78.2, "events": [ { "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck fr…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-603a5f7966/", "networkHref": "/investor/research-network/#thesis-603a5f7966" }, { "id": "thesis-7f0a75c5f6", "source": "tracked_thesis", "title": "数据中心地产、液冷与水资源正在决定 AI 容量的区域迁移", "titleEn": "Data-center real estate, liquid cooling, and water resources are deciding regional migration of AI capacity", "lane": "数据中心地产/液冷", "laneEn": "Data-center real estate and cooling", "laneHref": "/investor/lanes/data-center-real-estate-and-cooling/", "investableFocus": "数据中心地产/液冷：EQIX, DLR, IRM, VRT, TT, CARR, SMCI", "stockUniverse": [ { "ticker": "EQIX", "name": "Equinix", "market": "US", "role": "全球数据中心 REIT", "roleEn": "全球数据中心 REIT", "bucket": "core", "href": "/investor/stocks/EQIX/" }, { "ticker": "DLR", "name": "Digital Realty", "market": "US", "role": "批发数据中心 REIT", "roleEn": "批发数据中心 REIT", "bucket": "core", "href": "/investor/stocks/DLR/" }, { "ticker": "IRM", "name": "Iron Mountain", "market": "US", "role": "数据中心与存储 REIT", "roleEn": "数据中心与存储 REIT", "bucket": "core", "href": "/investor/stocks/IRM/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "TT", "name": "Trane Technologies", "market": "US", "role": "冷却与 HVAC", "roleEn": "冷却与 HVAC", "bucket": "core", "href": "/investor/stocks/TT/" }, { "ticker": "CARR", "name": "Carrier", "market": "US", "role": "冷却设备", "roleEn": "冷却设备", "bucket": "core", "href": "/investor/stocks/CARR/" }, { "ticker": "SMCI", "name": "Super Micro Computer", "market": "US", "role": "服务器与液冷整机", "roleEn": "服务器与液冷整机", "bucket": "core", "href": "/investor/stocks/SMCI/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" } ], "stockTickers": [ "EQIX", "DLR", "IRM", "VRT", "TT", "CARR", "SMCI", "NVDA", "TSM", "AVGO", "MU", "ANET", "SMH", "300308.SZ" ], "coreView": "赛道：数据中心地产/液冷。候选观察标的：EQIX, DLR, IRM, VRT, TT, CARR, SMCI。当前共有 40 条证据、1 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 76, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 7, "risks": 147, "neutral": 6, "analystCount": 15, "sourceTypeCounts": { "研究笔记": 154, "研究主题": 4, "研究线程": 2 }, "topTerms": [ "AI 基础设施", "GOES", "HBM", "GPU", "数据中心", "并网", "AIDC", "水电", "液冷", "设备", "Data-center", "real" ], "whatChanged": [ "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。", "2026-06-29: 变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "2026-06-28: GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "2026-06-28: AI内存规格升级对数据中心液冷与配电基础设施的传导效应", "2026-06-28: 效率驱动的算力架构转型：从“暴力规模”到“高能效密度”的物理基座适配性评估" ], "practical": { "sectorLane": "数据中心地产/液冷", "sectorLaneEn": "Data-center real estate and cooling", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "EQIX", "name": "Equinix", "market": "US", "role": "全球数据中心 REIT", "roleEn": "全球数据中心 REIT", "bucket": "core", "href": "/investor/stocks/EQIX/" }, { "ticker": "DLR", "name": "Digital Realty", "market": "US", "role": "批发数据中心 REIT", "roleEn": "批发数据中心 REIT", "bucket": "core", "href": "/investor/stocks/DLR/" }, { "ticker": "IRM", "name": "Iron Mountain", "market": "US", "role": "数据中心与存储 REIT", "roleEn": "数据中心与存储 REIT", "bucket": "core", "href": "/investor/stocks/IRM/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "TT", "name": "Trane Technologies", "market": "US", "role": "冷却与 HVAC", "roleEn": "冷却与 HVAC", "bucket": "core", "href": "/investor/stocks/TT/" }, { "ticker": "CARR", "name": "Carrier", "market": "US", "role": "冷却设备", "roleEn": "冷却设备", "bucket": "core", "href": "/investor/stocks/CARR/" }, { "ticker": "SMCI", "name": "Super Micro Computer", "market": "US", "role": "服务器与液冷整机", "roleEn": "服务器与液冷整机", "bucket": "core", "href": "/investor/stocks/SMCI/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" } ], "stockTickers": [ "EQIX", "DLR", "IRM", "VRT", "TT", "CARR", "SMCI", "NVDA", "TSM", "AVGO", "MU", "ANET", "SMH", "300308.SZ" ], "primaryStocks": [ { "ticker": "EQIX", "name": "Equinix", "market": "US", "role": "全球数据中心 REIT", "roleEn": "全球数据中心 REIT", "bucket": "core", "href": "/investor/stocks/EQIX/" }, { "ticker": "DLR", "name": "Digital Realty", "market": "US", "role": "批发数据中心 REIT", "roleEn": "批发数据中心 REIT", "bucket": "core", "href": "/investor/stocks/DLR/" }, { "ticker": "IRM", "name": "Iron Mountain", "market": "US", "role": "数据中心与存储 REIT", "roleEn": "数据中心与存储 REIT", "bucket": "core", "href": "/investor/stocks/IRM/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "TT", "name": "Trane Technologies", "market": "US", "role": "冷却与 HVAC", "roleEn": "冷却与 HVAC", "bucket": "core", "href": "/investor/stocks/TT/" }, { "ticker": "CARR", "name": "Carrier", "market": "US", "role": "冷却设备", "roleEn": "冷却设备", "bucket": "core", "href": "/investor/stocks/CARR/" } ], "instrumentTags": [ "EQIX", "Equinix", "DLR", "Digital Realty", "IRM", "Iron Mountain", "VRT", "Vertiv", "TT", "Trane Technologies", "CARR", "Carrier" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 160, "sourceDiversity": 3, "analystCount": 15, "recentEvents": 120, "supportRatio": 0.04, "riskRatio": 0.92, "conflictScore": 100, "evidenceDepth": 100 }, "score": 66.7 }, "events": [ { "date": "2026-06-29", "title": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "summary": "截至2026-06-30，本报告压力测试了变压器交期拉长对AI算力集群部署和云厂商资本开支的约束。变压器交期拉长（128-144周，专用型4-5年）导致物理并网时滞达4-7年，引发全球467万个GPU延迟并网和2万亿元Capex延期。微观上，1.5年延迟导致10亿美元GPU集群产生5.2295亿美元折旧与持有损耗，月度拖累1.1905亿美元，NPV骤降12…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30, "matchedTerms": [ "GPU", "数据中心", "并网", "液冷", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "HBM", "GPU", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "summary": "本卡压力测试逻辑链：GOES约束不会「抵消」核能加速，因为二者作用于电力链不同节点——核能重启在发电/并网侧本就是GOES-轻（复用既有GSU与开关站），属于一种GOES套利；但GOES在负荷侧与输配路径重新绑定，沿用Research note 01的NPV模型测算，核能锚定型项目(≈1.5年延迟,NPV≈−27%)相对并网型绿地(3年,−51.6%)只回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "GOES", "数据中心", "并网", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "变压器交期拉长至3-5年与并网排队超36个月的物理刚性瓶颈，配合美欧贸易保护关税对GOES硅钢供应链的阻断，已对AI算力扩张的通电效率构成无法用资本单纯解决的长期结构性壁垒，造成明显的Cap…", "summary": "基于并网排队与变压器短缺的双重物理约束，我们验证电力基础设施瓶颈已构成AI算力扩张的长期结构性阻碍。数据中心通电时间已拉长至36-60个月，变压器交期延至3-5年，主要厂商积压订单超百亿美元。这些延误对项目NPV和IRR产生重大侵蚀，100MW数据中心延误3年将损失51.63%的NPV。美国Section 301/232关税与欧盟CBAM机制则通过GOES…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "GOES", "数据中心", "并网", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,600 亿）以经营现金流支撑，资金基础/地域/标的均不同于承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.75, "matchedTerms": [ "AI 基础设施", "GPU", "并网", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "summary": "截至 2026-06-28，本卡压力测试 Research note 1 的 AI 基础设施使能层主线：需求真实且强劲，但 2026-2028 年兑现会被电网接入、可靠电源、变压器、燃机和运行可靠性约束。IEA 预计全球数据中心用电从 2024 年 415 TWh 升至 2030 年约 945 TWh；美国和中国合计贡献近 80% 增量。[S1][S2] …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.75, "matchedTerms": [ "AI 基础设施", "数据中心", "并网", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI数据中心电力瓶颈与核能PPA战略转型", "summary": "本报告分析了2026年AI数据中心面临的核心电网瓶颈，探讨了超大规模厂商绕过电网审批排队、锁定核能PPA（如微软与Constellation的Crane清洁能源中心项目）的战略选择。基于自主测算的200MW数据中心TCO模型，核能直供电相比传统电网虽有价格溢价，但通过提前2年并网运营，可规避高达50亿美元的GPU集群机会成本，在5年生命周期内创造45.52…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "GPU", "数据中心", "并网", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "2026年全球AI数据中心电网瓶颈与“自主供电”战略的崛起", "summary": "2026年全球超大规模厂商的AI算力基建受限于电网瓶颈与关键电力设备（变压器交期达5年）短缺。这迫使厂商将部分资本支出（预计2026年总支出达7250亿美元）从纯算力（GPU）转向自主供电（BYOP）和表后（BTM）能源方案。预计该转移将延迟算力上线速度，并推高长期持有成本。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "GPU", "数据中心", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "效率驱动的算力架构转型：从“暴力规模”到“高能效密度”的物理基座适配性评估", "summary": "压力测试 Research note 08 的隐含前提：效率并不缓解物理天花板，而是把它从电网层下移到机架与机柜行层并加剧。机理是机架尺度的杰文斯悖论——单瓦产出靠提高密度实现，机架功率从约40kW(Hopper)→120kW(Blackwell)→142kW(GB300,90%入液)→600kW(Kyber,2027)迈向1MW，对照2026均值约27k…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "HBM", "GPU", "液冷", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI内存规格升级对数据中心液冷与配电基础设施的传导效应", "summary": "本卡支持卡01-02：AI内存瓶颈确实向下传导为机柜级散热与配电瓶颈，利润池栈选对，印证英维克002837与麦格米特002851。但作两层修正：(1)比例——内存直接功耗占机柜不到1%[自测算]，真正传导是间接使能型，主推手是GPU TDP驱动的机柜密度(GB300约135kW→Rubin Kyber约600kW)；(2)时点——液冷/电源模块产能正以约3…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "GPU", "数据中心", "液冷", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "数据中心电力并网、关键电力设备与液冷温控的物理交付瓶颈将对算力上线速度形成硬约束，导致云厂商ROIC受损并延缓算力采购，增加电子产业链利润扩张的硬着陆风险。", "summary": "截至2026-06-28，本卡压力测试了算力产业链利润扩散假说，指出电力并网与变压器超长交期（MV变压器128周）已成为数据中心上线的物理瓶颈。预计2026–2028年美国35%–45%的新建数据中心面临延期通电，16GW计划容量仅5GW在建，30%-50%将延期至2028年以后。GB200的高功耗（单柜120-132kW）使直接芯片液冷成为刚性指标，但A…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "数据中心", "并网", "液冷", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "AI 数据中心 capex 从 GPU 向配电、UPS、母线槽与液冷供应商迁移的跨资产定价", "summary": "本卡综合前九卡，给出可交易的跨资产框架：AI 算力卡点已从 GPU 下移到可交付、已通电、已散热的兆瓦。正确定价不是单边押注 AI，而是 AI 板块内的相对价值轮动——做多以在手订单交付确认收入的电气/热管理供应商(VRT/ETN/SU)与电网设备(GEV/PWR)，做多铜作为材料成本推动与内部毛利对冲，并以针对资本开支最重云厂商的相对低配/价差期权对冲 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "GPU", "数据中心", "液冷", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "高功率密度架构驱动下的数据中心配电侧演变与资本开支影响", "summary": "截至2026-06-28，本卡支撑前序逻辑：Tokens/MW-sec 使高密度 AI 机架合理化，但瓶颈随之下移到中低压配电、UPS/负载平滑、母线槽、机架电源架、CDU 与液冷管路。15kW 到 132kW 使每 MW 机架数从约66.7降至7.6，但单机架峰值配电需按约192kW设计。资本开支并未消失，而是从面积和普通布线迁移到高额定电气设备、主动储…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "数据中心", "液冷", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-26", "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–560 亿美元；中国云(字节/阿里/腾讯)与国产替代构成第二引…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.75, "matchedTerms": [ "AI 基础设施", "数据中心", "并网", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-26", "title": "AI算力扩张正面临18个月的电网并网瓶颈约束，这将引致2026-2028年算力服务器交付出现7.50%-12.50%的缺口并下调硬件链利润预期，驱动资金向输变电设备与绿电直供资产防御分流。", "summary": "本卡片对“AI算力与电网承载力”进行压力测试。单台服务器负荷达9.75kW，并网延迟（Time-to-Power）拉长至18个月。这导致2026-2028年AI服务器交付出现-7.50%至-12.50%的缺口，使高营业杠杆的服务器制造商净利润压缩18.60%至30.00%。宁夏绿电直供提供13.89%的能耗成本折扣，成为云厂商重要避险工具。建议将仓位向受监…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.25, "matchedTerms": [ "并网", "水电", "设备", "infrastructure", "power" ], "href": "" }, { "date": "2026-06-26", "title": "碳成本传导与AI数据中心边际资本的结构性重构", "summary": "碳成本传导与AI数据中心边际资本的结构性重构 随着全国碳配额CEA价格收报82.41元/吨，100 MW数据中心PUE降至1.30每年可节电1.48亿元。这是否会驱动数据中心边际投资由算力GPU/CPO设备，系统性转向液冷温控与供配电架构？", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "neutral", "supportLabel": "观察", "score": 24, "matchedTerms": [ "GPU", "数据中心", "液冷", "设备" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 3, "neutral": 0, "value": 24, "events": [ { "title": "2027 年 HBM4 推理链条的首要弹性在合格 HBM4，首要出货瓶颈在 HBM4 验证与 CoWoS/interposer/封装基板，而液冷更多是部署瓶颈。", "summary": "本卡支持前序推理需求上行逻辑：288 GB HBM4 GPU 普及后，最大量价弹性在合格 HBM4 供应商，尤其是具备份额提升弹性的 MU 和具备配额权的 000660.KS；但最可能成为 2027 物理出货瓶颈的是 HBM4 验证与 CoWoS/interposer/封装基板的组合。测试设备是高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Q1 2026 财报后的云巨头 AI capex：需求验证还是瓶颈预警？", "summary": "本卡选择 tmt-analyst 评估 Q1 2026 云巨头 AI capex 周期。Microsoft、Alphabet、Amazon 和 Meta 均显示 AI/云收入加速，但 capex 和 PP&E 支出同步大幅上行。TSMC 的 61% HPC 收入占比、74% 先进制程占比和 66.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Firm power 不会否定 HBM4 推理需求，但会把 2027 年赢家从单纯 AI 硬件扩展到存量可调度发电、可获批 rate base、电网设备和电力集成，并重罚无 firm…", "summary": "本卡支持前序 HBM4 推理需求逻辑，但强调 2027 年部署瓶颈转向 firm power：可交付、可并网、能通过监管审查的 MW 比单纯电力申请更重要。最高盈利弹性在 CEG/VST/TLN 等存量 merchant nuclear/gas，AEP/ETR 等有已签大负荷协议和成本分摊机制的 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-03", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "Research note 02的1.6万亿-2.0万亿元AI政策包络会最大化强化Ascend生态，但未来12个月上市公司弹性不在最稀缺的HBM本身，而在液冷与400G/800G光互…", "summary": "本卡支持Research note 02的政策包络判断，但将其映射为更可投资的TMT结构：12个月内，最大战略瓶颈是HBM和先进封装，最大上市公司业绩弹性却在液冷与400G/800G光模块。原因是专项债和政策性银行项目贷优先进入可建设、可招标的数据中心资产，而HBM主要经由未上市CXMT/华为链条…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "AI 数据中心电力瓶颈已经成为 hyperscaler capex 回报分化的核心变量：已锁电力的 capacity 是稀缺资产，缺乏 firm power 的宣布式 capex 应…", "summary": "本卡 stress-test 卡片 01 的选择性建设性结论：AI capex 仍有需求支撑，但电力已经成为地点、时间和边际成本上的真实约束。DOE/LBNL 预计美国数据中心用电量从 2023 年 176 TWh 升至 2028 年 325-580 TWh；IEA 预计美国到 2030 年电力需…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "AI CapEx并未因外壳新开工放缓而失效，但其落地速度正被电力、变压器、配电网络和电网稳定性重新定价。", "summary": "截至2026-05-13，本卡支持“密度胜过足迹”主线，但补充能源约束：AI算力密度提升使稀缺资源从外壳数量转向可供电、可散热、且符合电网稳定要求的容量。LBNL、EPRI和EIA均显示数据中心用电占比快速上升；NERC 2026-05-04 Level 3警报和2024年1,500 MW数据中心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "将 5% BYOP 燃气与 4% 非晶替代从触发对冲升级为 2026 H2 基础配置，确认电力缺口已由‘并网延迟’转向‘表后自建’与‘能效替代’的硬约束阶段。", "summary": "基于 GE Vernova 和西门子能源 2026Q2 的创纪录订单披露，以及微软、甲骨文等 hyperscaler 的‘自带电力’(BYOP) 转向，验证了燃气轮机槽位已成为类 GPU 的战略稀缺资源。同时，由于 160 周的变压器交付瓶颈和 2027 年 DOE 能效新规，非晶合金对取向硅钢 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "AI算力增长正撞向“电力天花板”，半导体板块技术面极端过热且基本面面临“资本开支消化期”风险，支持减仓芯片转向电网。", "summary": "截至2026-05-14，本卡通过压力测试支持大幅减持SMH至5%并保持DRAM 8%战术卫星的决定。核心逻辑是AI算力瓶颈已从“硅片供应”转向“电力并网”，美国和欧洲电网并网等待时间已超5年，导致“暗壳”数据中心（有GPU无电力）现象频发。半导体周期已入成熟期，SOX指数偏离200日均线60%，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 1, "risks": 0, "neutral": 0, "value": 57, "events": [ { "title": "AI核心硬件（服务器与液冷）的资本开支与交付节奏已严格锚定甘肃、宁夏等枢纽的电网节点投运时间，形成“电力先行，算力紧随”的实质闭环。", "summary": "2026-05-18：本卡支持前序电力主线。调研显示，随着甘肃、宁夏等核心枢纽330kV/500kV变电站及绿电直供项目落地，AI服务器与液冷配套硬件的交付周期已发生适应性调整。云厂商（CSP）为避免资产闲置，将服务器与GPU订单的释放节点严格锚定变电站并网前3-6个月。目前英维克等液冷厂商及算力…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 38.4, "events": [ { "title": "2027-2028年电力缺口不会终结AI芯片周期，但会把下一代AI硬件迭代逻辑从单芯片峰值性能改写为受限MW下的有效AI输出、上线时间和机架级能效。", "summary": "截至2026-05-18，本卡支持前序电力瓶颈主线，但压力测试“硬件从极致性能转向能效”的表述：更准确地说，采购函数会转向每已通电MW的有效AI输出。2027-2028年表后燃气仅4-6GW，无法覆盖AI负荷缺口；DOE/LBNL预计美国数据中心用电2028年达325-580TWh。NVIDIA …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-05-19，AI芯片架构已进入能效竞赛，能显著放大每个已并网MW的算力产出，但只能缓解而不能消除电力约束。", "summary": "截至2026-05-19，本卡支持“AI基础设施转向P/Watt”的主线。NVIDIA Blackwell/GB200、AMD MI350、Google Ironwood、AWS Trainium3均显示竞争焦点已从单芯片峰值算力转向低精度、HBM、机架互联、液冷和软件调度带来的每瓦有效token…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 0, "neutral": 0, "value": 59.8, "events": [ { "title": "若 2026-06-11 验证 Waymo 单位经济，AI 基础设施不应被卖出，而应从拥挤算力前端轮动到电源液冷、networking 邻近环节与小仓位机器人硬件 beta。", "summary": "截至2026-05-19，本卡支持 6 月 11 日后从 AV 扩散到 AI hardtech，但只在 Waymo 经济性披露有用时执行。确定性最高的是电源/液冷/机柜、定制芯片与 networking、HBM/先进封装；估值弹性最好的是电源热管理与机器人端侧硬件。HBM、GPU、A 股光模块和主…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "AI算力链盈利可见性已被订单簿三重验证、足以支撑保留成长端的策略；半导体设备应拆分成‘AI直接受益+国产替代超配，纯中国敞口设备与晶圆厂低配’，故第03卡‘不全面切向价值’成立但需要…", "summary": "截至2026-05-21，本卡对第03卡“不全面切向价值”进行压力测试，结论是有条件支持。AI算力链（GPU/HBM/光模块/液冷/先进封装）盈利可见性高且仍在抬升：四大美系hyperscaler 2026年合计资本开支指引上修至USD 360–410 bn、TSMC CoWoS与SK Hynix…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "AI算力基建的物理约束并非简单的股票风格轮动，而是引发大宗通胀、期限溢价抬升与股债相关性转正的系统性供给侧冲击，大类资产配置必须收缩久期并超配瓶颈商品与即时现金流出海Alpha。", "summary": "截至2026-05-21，对AI硬件并网物理瓶颈的压力测试显示，该约束构成系统性供给侧冲击，而非仅是权益风格轮动。闲置暗数据中心导致GPU ROIC骤降至8%以下，挤压高杠杆算力链估值。精炼铜面临33-45万吨结构赤字，价格突破$13,000/吨，GOES高企，商品成为核心物理约束期权。电网融资压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移", "summary": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移 AI Physical Bottlenecks: Shift from GPU Capacity to Power Transformers and Interconnection Queues 全球电网并网周期拉长、特大变压器交期逼…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "上游基础材料（铜与取向电工钢）的结构性短缺与通胀，使得工业产能扩张无法顺利落地，变压器与电力配给材料成为拖累AI资本开支转化为自由现金流（FCF）的绝对红线。", "summary": "截至2026-05-23，本卡片通过压力测试确认，上游材料（特别是铜与取向电工钢）的短缺和价格通胀，正在把变压器与液冷设备的扩产转化为AI基础设施的第三个硬性瓶颈。铜价已在2026年5月逼近14000美元/吨高位，单MW AI数据中心需消耗30-47吨铜；取向电工钢（GOES）极度依赖进口且产能集…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "AI硬件交付与带电土地（Powered Land）及REITs的远期电力设备产能高度绑定，不存在普遍的“空转折旧”风险，电网瓶颈只会推高头部数据中心开发商的资产溢价。", "summary": "截至2026-05-24，本报告从房地产交付与基础设施视角反驳了“AI硬件将因电网瓶颈面临严重空转折旧”的观点。云巨头的GPU采购与数据中心“准备就绪（RFS）”里程碑严格绑定，避免了盲目囤积。头部数据中心REITs（如Digital Realty）通过超3亿美元的远期能力协议（SCA）买断了变压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 2, "neutral": 0, "value": 46.8, "events": [ { "title": "AI基础设施资本开支尚未在订单层面见顶，但在扩产加速度和重资产风险收益上已接近顶部，2027年放缓将优先打击低毛利、高折旧、专用性强的制造环节。", "summary": "截至2026-05-24，本卡从工业制造与Capex Cycle角度应力测试AI基础设施扩产。结论是：已披露需求和行业基准预测尚未显示周期见顶，SEMI仍预计2026-2027半导体设备支出上行；但资本开支加速度、低毛利跟随者扩产和数据中心建设成本已呈现后周期特征。若2027年需求放缓，最严峻的减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的部署上限由已通电液冷MW决定，而不是由GPU订单或云厂商capex预算单独决定。", "summary": "截至2026-05-24，本卡支持前序“AI瓶颈物理化”判断，但把瓶颈收窄为三条工业交付链：液冷机架、电气设备和可靠电力。GB300 NVL72机架功率已达132-140 kW，100 MW IT功率在设施损耗前仅支持约714-758个机架；JLL披露2025年57%数据中心项目因供应链延误至少3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "变压器已经构成类似GPU配给的结构性time-to-power瓶颈，重型燃机则在表后发电路径中成为条件性但重要的第二瓶颈。", "summary": "截至2026-05-25，本卡压力测试并基本支持“time-to-power”判断：Blackwell级部署的短中期约束已从单纯GPU供给扩展到变压器、开关设备、重型燃机与并网交付。变压器更像结构性瓶颈，因为电网供电、表后燃机、可再生PPA和园区配电都需要它；重型燃机则在表后基荷与过渡电源路径中升…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团", "summary": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团 Global AI Compute Grid interconnections Bottlenecks and Cross-Sector BTM Rotation 伴随变压器制造及重型燃气轮机5-7年交付周期的物理瓶颈，超大规模数据…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "截至2026-05-29，美国贸易壁垒和本土化采购要求会把AI数据中心并网延迟从设备短缺放大为合规与地缘供应链问题，中国及亚洲供应只能边际补缺，难以快速突破美国瓶颈。", "summary": "截至2026-05-29，本卡支持前序结论：AI并网瓶颈不只是变压器和GOES的物理短缺，美国Section 232、Section 301和Buy America规则把进口替代变成更慢、更贵、更难合规的路径。中国2025年GOES产量和出口创新高，但中国原产变压器与电工钢面对额外关税和原产地审查…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 2, "neutral": 1, "value": 55.4, "events": [ { "title": "SpaceX巨额IPO流动性抽水与美国新电力法案下的全球AI算力物理层与BTM重估", "summary": "SpaceX巨额IPO流动性抽水与美国新电力法案下的全球AI算力物理层与BTM重估 SpaceX IPO Liquidity Drain and US Electricity Bill Converging on Global AI Physical Layer & BTM Revaluation…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026是AI产能的配额年而非顶点年——capex绝对值再创新高但增速触顶、瓶颈已从芯片迁移到变压器与液冷，旗舰算力加速通电与存储景气是同一物理约束的两个面，而非彼此矛盾。", "summary": "四个看似冲突的信号——AI基础设施股跑输、35GW管线约15%受电力/液冷物料延迟、Abilene 1.2GW旗舰加速完工但600MW扩建被取消、SNDK数据中心营收同比+645%——可以放进同一个2026框架：今年是配额年而非顶点年。大型云厂2026年capex合计7,250亿美元、同比+77%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年是工业设备‘稀缺定价’的顶点，变压器交期与液冷渗透率已取代GPU供应成为决定AI算力部署上限的物理约束。", "summary": "截至2026年5月，AI算力的瓶颈已从GPU转向以变压器和液冷为核心的工业基础设施。大型变压器交期已达36-48个月，导致2026年近50%的计划数据中心容量面临延迟。Vertiv积压订单达150亿美元（+109%），Modine甚至获得云巨头1.65亿美元预付款以锁定产能。2026年是物理设备的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 0, "value": 59.8, "events": [ { "title": "并网惩罚与128周变压器交期瓶颈正强制终结AI的‘纯硅端溢价’，表后自发电（BTM）与调峰水电凭借高经济性与排队规避，确立为AI物理层的核心资产。", "summary": "本卡深度解析了《2026年能源法案》并网惩罚如何重塑AI用能结构。LCOE量化测算证实，混合BTM共建模式较并网FTM模式可实现36.5%的用电成本削减（$72.00对$113.35/MWh），并彻底避开3至5年的并网排队与128周变压器交付瓶颈。具有大调节容量的梯级水电已成为亚太稳定绿色AI算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "真正的利润弹性来自设计控制权和供应合约，而不是简单把材料涨价转嫁给客户。", "summary": "截至2026-06-02，本卡支持上游材料瓶颈结论，但将TMT利润风险重新定义为“资格认证与架构控制权”问题。人形机器人2026年NdFeB吨位压力有限，按2-4 kg/台与50,000-100,000台测算仅100-400吨，但高牌号磁体、镝/铽许可、谐波减速器和滚柱丝杠会放大交付延期。AI硬件…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 2, "neutral": 0, "value": 56.6, "events": [ { "title": "AI资本开支受制于53个月并网时延与120-160周变压器交付周期的硬性电力墙约束，投资逻辑已从硅侧算力向电网、基载核电及液冷散热等物理瓶颈要素转移。", "summary": "截至2026-06-03，AI资本开支的物理基础设施约束已成为核心天花板。由于美国电网并网延迟长达53个月，且特大型变压器交付周期拉长至120-160周，AI资本开支向云收入转化时延从4个季度拉长至6个季度。容量电价飙升使100MW数据中心电费翻倍并压缩推理毛利至42.0%。建议将硅侧权重削减至2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "制造端增量能在2028年后缓解电网设备短缺，但无法在4-6年GPU经济折旧窗口的前半段抵消大型变压器和并网硬件瓶颈。", "summary": "截至2026-06-03，工业制造端支持前序电力瓶颈判断：电网设备 OEM 正在扩产，但大型电力变压器和 GSU 变压器仍是关键长周期约束。Wood Mackenzie 披露 2025 年二季度 GSU、电力变压器和开关设备平均交付周期分别为 143 周、128 周和 44 周；DOE 指出新建变…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 1, "neutral": 0, "value": 64, "events": [ { "title": "模型能效与边缘计算是面对一阶物理瓶颈（变压器/电网/排队）的二阶效应，叠加推理token爆炸与已锁定capex，'通电滞后'对AI应用端的重定价逻辑在12–18个月内不会被瓦解。", "summary": "TMT否定'能效+边缘'对冲电网瓶颈的反驳。Blackwell单token能效25×提升被推理调用量9× YoY与o系列推理链5,000–50,000 token的爆炸增长抵消，IEA预计数据中心电力需求2024–2030翻倍至945 TWh。瓶颈在LPT/GOES硅钢/并网排队（104–127周…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "AI算力瓶颈已从GPU芯片转移至电力和电网，资产配置应超配锁电公用事业与具备套保及价格联动出海变压器商，低配低毛利电缆及面临延迟通电风险的次级硬件与服务器链条。", "summary": "截至2026-06-05，本卡综合研判了AI算力物理瓶颈对公用事业、电网设备及硬件链条的策略定价影响。全球数据中心电力消耗翻倍（2030年达950 TWh）与大功率变压器交期拉长（2-4年）加剧并网延期，催化了以微软核能协议为代表的表后自备电力模式。配置上建议超配锁电能力强的公用事业（福能股份、江…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [ { "title": "AI 基础设施压力测试：能源“供需墙”的结构性瓶颈 — 2026 06 07 作者： 能源行业分析师 日期： 2026 06 07 立场： 压力测试 (Stress test) 执行…", "summary": "AI 基础设施压力测试：能源“供需墙”的结构性瓶颈 — 2026 06 07 作者： 能源行业分析师 日期： 2026 06 07 立场： 压力测试 (Stress test) 执行摘要 截至 2026 年 6 月 7 日，TMT 硬件行业的“进攻型”转型正面临一个物理极限： 能源电力墙 。尽管半…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 3, "neutral": 1, "value": 59.2, "events": [ { "title": "AI 基础设施的‘物理主权’已转移至掌握变压器核心组件（GOES、分接开关）与垂直整合能力的工业巨头，2026年的投产节点由‘变压器槽位’而非‘芯片配额’决定。", "summary": "2026年6月9日：本卡通过物理交期压力测试确认，电力变压器（交期3-5年）与开关柜（1.5-2年）已成为AI落地的‘刚性闸门’。工业巨头正从设备商转向‘产能地主’，Eaton通过‘从电网到芯片’的垂直整合与Boyd/Motivair的液冷并购锁定了高溢价份额；Schneider则凭借预制化模块（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 硬件的增长上限已由‘数字世界’转为‘物理世界’，电力缺口将迫使市场从纯算力 beta 转向通电效率与重型电力基础设施。", "summary": "2026-06-09：本卡对 AI 硬件估值进行压力测试，指出物理电力供应已取代芯片产量成为核心瓶颈。电网并网积压（36-84个月）与电力硬件（变压器）交期过长（24个月以上）导致 2026 年近半数数据中心容量面临延期。NVIDIA Rubin 架构功耗激增至 2.3kW/GPU，迫使架构全面转…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器、switchgear、UPS/冷却和并网排队正在把AI算力扩产从GPU采购问题变成送电交付问题，2027年前难以全面缓解，继续压低重资产算力商早期资产周转并抬升ETN/VRT…", "summary": "截至2026-06-10，工业设备证据支持前序空重资产GPU云、配多电力/设备链的主线。Eaton、Vertiv、GE Vernova的订单与backlog仍快于收入交付：Eaton Electrical Americas backlog同比+44%、数据中心订单约+240%；Vertiv Q4 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "AI算力建设正从“依赖电网”转向“离网自治”，SOFC与中国电力设备出口链通过解决交付周期错配，成为捕捉AI基建溢价的核心抓手。", "summary": "2026年AI算力扩张的瓶颈已从GPU转向电力并网，使得SOFC与变压器成为核心资产。SOFC通过“时间套利”（交付周期<12个月对比电网并网7-10年）为AI厂商节省数十亿美元机会成本，LCOE在$58-110/MWh区间已具竞争力。中国厂商凭借全球60%的变压器产能及对取向电工钢（GOES）的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 65, "events": [ { "title": "GW 级数据中心的电力瓶颈与电网投资机遇：核能并网与公用事业的超额收益 日期： 2026年6月11日 分析师： 公用事业分析师 (Utilities Analyst) 主题： GW …", "summary": "GW 级数据中心的电力瓶颈与电网投资机遇：核能并网与公用事业的超额收益 日期： 2026年6月11日 分析师： 公用事业分析师 (Utilities Analyst) 主题： GW 级数据中心电网约束与公用事业投资机会 执行摘要 截至 2026 年 6 月，AI 基础设施的叙事已从“GPU 稀缺”…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力交付滞后会迫使 Hyperscalers 调整 GPU 集群投产节奏，并通过更长现金回收期、更低早期利用率和更高电力保障成本压低 AI 基础设施 capex 效率。", "summary": "截至2026-06-11，本卡支持上一卡：AI capex 不会因电网瓶颈消失，但 GPU 集群部署节奏会被“已通电 MW”重新排序。变压器、并网和输电延迟会把部分已宣布园区容量推迟到后续窗口，压低早期自由现金流和资产利用率。Hyperscalers 会通过已通电站点溢价、分阶段投产、区域重排和自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 1, "risks": 2, "neutral": 0, "value": 74.4, "events": [ { "title": "AI算力基础设施架构的重塑与硬件需求演变", "summary": "支持卡片01。当并网容量（兆瓦）成为固定稀缺输入，优化目标从“每机柜最大FLOPS”翻转为“每交付兆瓦最大有效算力”。这并非密度撤退——GB200 NVL72仍达120kW/机柜、800V路线图指向880kW[S1][S19]——而是密度转为服务效率。三大倒逼因素：PJM并网周期升至>8年[S5]…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "科技巨头直接注资电力资产因电网设备供应链瓶颈（变压器交期48-60个月）而演变成“资本陷阱”，导致高达40%的预承诺GPU硬件（$100亿）因并网滞后3年而遭遇65.7%（-$65.…", "summary": "截至2026-06-13，本报告对科技巨头直接注资电力资产的资本转移进行了压力测试。由于高压变压器（LPT）交期延长至48-60个月、取向硅钢（GOES）供应受限以及并网队列长达5年以上，资本转移正退化为“资本陷阱”。定量测试表明，在3年并网延迟与40%电网造价通胀下，一个2 GW的AI数据中心项…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供给的物理并网瓶颈与高密算力液冷配套开支的刚性上升，正在对超大规模数据中心资本开支形成显著的“基建挤出效应”，迫使纯算力硬件开支占比向电网、变压器和温控冷却设施倾斜，进而压制纯芯…", "summary": "截至2026-06-13，电力供给的结构性稀缺（并网延迟达24-48个月，PJM容量价格飙升10倍）与Blackwell机柜高功耗（单柜132kW）倒逼超大规模厂商进行CAPEX结构重构。定量测算表明：在无约束情景下，设施基建占比从23.08%上升至34.07%，其中配电占比由5.54%升至9.3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 5, "neutral": 0, "value": 49.8, "events": [ { "title": "地缘冲突带来的能源价格上行与物理并网延迟正在迫使云厂商将资本开支从纯算力硬件向电网、变压器和温控液冷基建倾斜，产生显著的“基建挤出效应”；在脱碳路径上面临自备燃气发电（高ROIC但需…", "summary": "截至2026-06-13，本报告针对中东地缘能源冲击与电网瓶颈对AI算力基建的影响进行了压力测试。在$110/MWh高电费与$1364.40M配套基建（变压器GOES溢价及液冷）刚性支出下，超大云厂商为维持10%靶向ROIC必须调降GPU采购预算54.32%，挤压芯片商毛利至45.27%。在能源脱…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的电力瓶颈与算力扩张可行性评估", "summary": "对第07卡哑铃组合「AI基建成长极」做压力测试：需求斜率真实——全球数据中心用电2025年485TWh将于2030年翻倍至约950TWh，中国2025年达193.3TWh（+17%），2026年1-2月同比+46.2%；2026年全球超大规模capex达约6300亿美元(+62%)。但约束不在发电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GB200 级效率提升改善 AI factory 经济性，但 2026-2028 年真正限制部署节奏的是站点电力、并网、变压器、稳定电源和电网运行能力，而不是 CoWoS/HBM 单…", "summary": "2026-06-13：本卡对 NVIDIA 护城河做电力压力测试。GB200 NVL72 单机柜约 120 kW，1 GW IT 功率约支持 8,333 个机柜、约 600,000 颗 GPU。IEA、DOE/LBNL 与 NERC 显示，数据中心用电、并网队列、变压器和燃机交期都在拉长，大型计算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS绝对产能正在改善，但GPU与自研ASIC需求吸收速度更快，因此2026年AI集群交付仍受先进封装、HBM和关键设备交期的结构性约束。", "summary": "截至2026-06-13，本卡支持上一张AI基建卡的判断：硬件交付风险不只来自并网，还来自CoWoS/HBM/设备交期的半导体转换链。TSMC仍称先进封装非常紧，TrendForce口径显示CoWoS-L/S满载且2026年底目标120,000-130,000片/月；DIGITIMES虽更乐观，预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 9, "neutral": 0, "value": 27.2, "events": [ { "title": "变压器原材料瓶颈（取向电工钢和铜）不会导致GPU需求破坏，而是导致变压器价格飙升及严重的闲置算力折旧损失，迫使科技巨头扭曲资本开支，大举向电力能源基础设施渗透。", "summary": "本报告研究了2026年取向电工钢（GOES）和铜的供应约束对电力设备变压器OEM利润率及AI算力交付的影响。结果显示，虽然原材料价格上涨（重度冲击下GOES+40%，铜+30%）会大幅压缩固定价格合同利润率，但OEM凭借极强定价权通过指数联动将成本完全转嫁给买方。AI算力未出现需求破坏，而是表现为…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件周期的约束正在从芯片可得性转向电力确定性，物理上限首先表现为区域部署延迟而非全球需求崩塌。", "summary": "截至2026-06-15，本卡压力测试认为：电力已成为AI算力部署的节奏阀，但2030年前不是全球统一硬上限。全球年度电量大体能承接IEA所示数据中心用电从2024年约415 TWh升至2030年约945 TWh的路径；真正瓶颈在PJM、Virginia、Ireland等局部电网，以及中国“东数西…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电网瓶颈正迫使 AI 基础设施从‘暴力规模扩张’转向‘受约束的能效优化’，电力并网速度已取代芯片供应成为核心估值护城河。", "summary": "本报告通过压力测试确认了物理电网并网排队（7-10年）与变压器交期（4-5年）已成为 AI 基础设施扩张的硬约束。定量模型显示，1年的并网延迟将导致 100MW 数据中心损失 5.5% 的生命周期价值，且快速的 GPU 迭代周期（12-18个月效用减半）加剧了闲置产能的 ROI 损耗。这种物理瓶颈…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "以变压器为核心的电力设备扩产滞后与原材料供给刚性对2027年后的数据中心并网构成了实质性瓶颈。", "summary": "评估表明，变压器等电力设备供应链扩产极其滞后（大型电力变压器交期拉长至2-4年），受特种钢（GOES）和铜的物理供给制约，各厂商新产能直至2027/28年才能释放。这将直接阻碍2027年后的AI数据中心并网（平均并网及设备延迟超5年），验证了物理层瓶颈对算力中心扩张的实质制约。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 0, "value": 64.6, "events": [ { "title": "AI基础设施回报正在从tokens-per-dollar扩展为electrons-per-dollar，稀缺资产是正确节点上可交付、可并网、可稳定运行的电力。", "summary": "截至2026-06-16，本卡支持上一卡判断：AI交易的边际瓶颈正从GPU转向可交付电力。IEA显示5家大型科技公司2025年资本开支超过4000亿美元，2026年再增75%，而美国数据中心用电量预计从2023年的176 TWh升至2028年的325-580 TWh。电气设备在数据中心建设成本中约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海底零碳算力与深海数据中心：技术可行性与电力供应保障 白板编号：4157cf99 4f17 487d 9efd db740a1f0fbf · 第 02/10 张 分析师：AI 基础设…", "summary": "海底零碳算力与深海数据中心：技术可行性与电力供应保障 白板编号：4157cf99 4f17 487d 9efd db740a1f0fbf · 第 02/10 张 分析师：AI 基础设施分析师（ai infrastructure analyst） 日期：2026 06 15（亚洲/新加坡） 立场：支…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地理转移并未消除物理约束，仅是将欧美的“设备短缺”转化为了东南亚的“电网消纳与绿电赤字”以及中东的“热力学极限与物理安全溢价”。", "summary": "本卡对“算力向中立区转移即可解决基建瓶颈”的假设进行压力测试。研究发现转移仅是转化了瓶颈：柔佛面临超11GW并网申请远超电网承载力，被拒率达30%，被迫以天然气灰电填补绿电赤字；中东面临高密度算力与缺水的严重冲突，转向高昂液冷，且阿联酋电网已现饱和预警并受无人机袭击引发安全溢价。物理极限不可跨越。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 8, "neutral": 1, "value": 39, "events": [ { "title": "通胀和高利率提高了AI硬件基础设施的质量门槛，但没有削弱其配置价值；应买入瓶颈供应商而非泛AI叙事。", "summary": "截至2026-06-16，本卡支持“硬科技配置中保留AI基础设施”的上一轮判断，但建议从广义AI beta收窄到有订单、定价权和瓶颈属性的硬件基础设施。高利率会压缩投机型数据中心开发商和远期AI应用估值，却未削弱GPU、定制AI芯片、HBM、配电、冷却、变压器、开关设备和已锁定电力的数据中心容量的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "WFE验证90→180天不引发2026–2027年AI集群隐性交付延误——真正瓶颈是EHV变压器与并网；WFE是2028年供给上限风险。", "summary": "对卡片08的WFE→AI集群隐性延误链条做压力测试：方向支持但归因重构。2026–2027年AI集群交付的真正绑定约束是EHV变压器/并网排队 (60–80%) 与CoWoS-L槽位 (15–20%)，WFE验证90→180天的边际贡献仅3–7%。WFE通道真正咬合的是2028年N2级GPU供给上…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理基础设施需求足以在没有流动性爆发的情况下维持电网设备与稀缺投入品周期，但市场仍应惩罚同时需要廉价资本和完美部署节奏的AI算力商业模式。", "summary": "截至2026-06-17，本卡压力测试结论是：在宏观流动性不爆发的情景下，AI算力、电网接入、变压器和液冷的物理需求增量足以支撑头部hyperscalers与稀缺电网供应链资产，但不能保护依赖廉价资金、缺乏确定电力或确定客户的投机型AI算力项目。全球数据中心用电到2030年约945 TWh、美国到…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理算力护城河对AI基础设施SPV租金与利用率稳定性的贡献评估 日期：2026 06 16 | 分析师：AI基础设施分析师 [ai infrastructure analyst] 核…", "summary": "物理算力护城河对AI基础设施SPV租金与利用率稳定性的贡献评估 日期：2026 06 16 | 分析师：AI基础设施分析师 [ai infrastructure analyst] 核心观点 (Thesis) 物理护城河（表后供电、高密度液冷、微电网）在2026年的电力稀缺环境下，已从单纯的“可用性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 2, "risks": 3, "neutral": 0, "value": 83.4, "events": [ { "title": "AIDD是AI基础设施需求的高质量增量因子：规模小于通用大模型，但持续性、数据壁垒和监管/IP属性更强。", "summary": "截至2026-06-18，本卡片支持上一张卡片的方向：AIDD不会单独决定hyperscaler capex，但会把高价值、持续性的生命科学工作负载接入GPU、网络、存储、液冷和电力保障型数据中心。Recursion BioHive-2、DCAI Gefion、Novo Nordisk、Isomo…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "卡片09的省份排序方向正确,但兑现风险已从设备交付迁移到江苏500 kV配网排队与湖北水电干旱尾部,安徽是唯一具备结构性余量的省份。", "summary": "压力测试卡片09关于江苏/安徽/湖北率先兑现HBM+2.5D封装的时点判断。结论：方向正确但约束已迁移到三块负荷岛的500 kV环网与可靠容量。三省2026–2028年新增封装+同址算力合计6–10 GW。江苏(2025峰值148 GW,有序用电已启动)为红灯,封装与算力同址可行性最低;安徽(皖南…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与能耗瓶颈足以封住AI硬件敞口上调空间，但不足以否定当前带上限持仓。", "summary": "截至2026-06-17，本卡支持前序AI硬件约18%上限，但压力测试显示电力供应、并网、PUE和冷却已成为PCB与光模块之外的第二道物理屏障。IEA预计全球数据中心用电到2030年约945 TWh，美国和中国贡献近80%增量；JLL和CBRE显示主要市场并网等待超过4年、空置率低至1.4%。结论…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "白皮书主线成立，但因文件归档缺失和后段产能/价差模型证据不足，2026-06-18 不具备最终发布条件。", "summary": "截至2026-06-18，本终审 QA 对 9 张卡片给出“需要修订”评级。主线从 AIDD 算力需求到 GPU/HBM、机柜功率密度、电网设备稀缺、GOES 套利和西方 OEM 产能错配整体连贯，电力与电网瓶颈证据最强。但实时工作区仅存在 card-02、card-04、card-06、card…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 7, "neutral": 0, "value": 44.8, "events": [ { "title": "变压器与GOES的区域化贸易壁垒通过拉长交付周期至154.5周，直接造成算力资产闲置折旧（24k GPU集群年拖累达5.624亿美元），迫使超算厂商进行CAPEX重组。", "summary": "截至2026-06-18，研究变压器与GOES区域壁垒对AI算力的传导效应。美国IRA本土含量、欧盟NZIA及中国反倾销税等碎片化政策使GOES供给割裂。压力情景下，卷线工稀缺与设备积压将变压器交期拉长至154.5周。这导致超算并网时滞达3-5年，使得24k GPU算力集群闲置一年的折旧与流失利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由于大型电力变压器与测试大厅产能严重受限，2027-2028年计划新增的AI数据中心有43%面临通电并网延期[自测算]，由此引发GPU订单推迟，对主要芯片制造商及云厂商带来显著的收入…", "summary": "基于2026-06-19数据，本分析量化了电网变压器与并网延迟对2027-2028年AI硬件产业的影响。模型显示，原计划30GW数据中心管线因变压器交期拉长至120-210周而延期12-24个月，实际仅能交付17.1GW，缺口达43%[自测算]。这导致GPU出货量在2027年减少4.8M颗、202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2028 年 AI 基础设施的胜负不在于是否完全离网，而在于能否把液冷机架、BTM/并网混合电源、闭环冷却、燃气交付和工作负载调度整合成可复制的 7-150 MW 阶段模块。", "summary": "2026-06-18：本卡压力测试前序主线。结论是 ERCOT/Permian 与 BTM 气电是 AI 训练负荷的真实泄压阀，但不能干净解决 2028 年瓶颈；它们把问题转化为模块化电力-冷却-算力架构。液冷从可选项变成高密 AI 的前提，132 kW/机架意味着 1 GW IT 负荷约需 7,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力/电网瓶颈真实且已在容量价格中显形,但2027年前仅推迟部分美国新增集群、不封顶全球GPU部署,Research note 01的BOM组合在2026至2027 H1仍成立,20…", "summary": "AI算力电力瓶颈真实存在但属于'延后+置换'而非'封顶':PJM 2026/27容量价已跳11倍至$329.17/MW-日,2027/28预测缺口6.6 GW,新接入等待40个月;ERCOT 2026 Q1单季198 GW申请,累计大负荷队列约410 GW;全球电网接入等待5-10年,变压器交期1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 7, "neutral": 1, "value": 49.2, "events": [ { "title": "16.18%的硬件Capex下修会压低AI基础设施整体估值，但订单传导不是线性的：液冷最敏感，电网设备和BTM发电更受长期电力容量瓶颈支撑。", "summary": "截至2026-06-19，本卡把H2 AI硬件Capex下修16.18%作为未外部确认的压力测试输入。结论是分层传导：GPU/服务器采购最先受冲击；液冷因绑定高密度机柜节奏，订单和估值敏感度较高；电网设备因并网稀缺、变压器长交期和backlog，订单不应按同等比例下修；BTM自备发电估值会先压缩，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备交期和并网流程已成为AI算力基础设施建设周期的确定性约束，并会放大高负债算力租赁商的融资与估值压力。", "summary": "截至2026-06-19，本卡支持前序判断：AI算力建设的硬瓶颈已从GPU采购转向电力交付。DOE/LBNL与EPRI均显示数据中心用电占比快速上升；Wood Mackenzie与POWER显示大型变压器交期已达120-144周，部分大型设备80-210周；JLL称主要数据中心市场平均并网等待超过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "常规电网并网延迟与新能源项目财务崩溃，迫使云厂商将资本开支刚性移向自备微电网与高密液冷配套，配套设施开支上升72.22%对纯算力硬件产生挤出效应，但也实质上支撑并延长了AI物理层设备…", "summary": "截至2026-06-20，本卡片评估了能源供应与并网效率对AI物理层投资周期的支撑性。常规电网3-5年的并网延迟使100MW数据中心贴现NPV折损达51%至80%，迫使云厂商向“表后直供+储能”的微电网转型。虽然配建100MW/400MWh储能使LCOE上升14%至$55.10/MWh，但每小时C…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "FERC新规清出队列投机后，AI数据中心资本开支的真实瓶颈已下移到变压器与变电站设备，配电变压器OEM和上游GOES是未来4–8季度可见度最高的细分赛道。", "summary": "FERC Order 2023/2023-A与2026年6月补充令把并网约束从队列前端下移到变压器与变电站设备。LBNL数据显示2024年底美国队列约2,600 GW，PJM大型负荷申请>50 GW；配电变压器交期已拉到80–130周，LPT 130–210周，HVDC换流变200+周。6月补充令…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 1, "risks": 10, "neutral": 0, "value": 37.6, "events": [ { "title": "AI 数据中心的下一道硬约束不是单纯电量，而是变压器、开关柜、并网与液冷系统共同决定的可交付 MW。", "summary": "截至2026-06-20，本卡支持上一张 AI 基础设施判断：AI 机架功率密度上升后，瓶颈从芯片供给下沉到可交付、受保护、可冷却的 MW。液冷设备具备较强模块化扩产弹性，但变压器、中压开关柜、并网和变电站设备仍是低弹性约束；美国配电变压器、大发电变压器和开关柜交期已显著拉长，拥有接入权、设备槽位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与 AI 数据中心能源架构：功率密度成为新的供应链关口 作者：agy（AI Infrastructure Analyst） 权威工作日期：2026 06 20 立场：supp…", "summary": "先进封装与 AI 数据中心能源架构：功率密度成为新的供应链关口 作者：agy（AI Infrastructure Analyst） 权威工作日期：2026 06 20 立场：support 研究线程标识：已归档rd Index：2/10 截至权威工作日期 2026 06 20 ，我支持上一张卡片关…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "600kW+机柜把AI扩容瓶颈从芯片采购推向公用事业级送电能力；2027-2028年最可能的失效模式是资本开支变成部分送电园区，而非芯片完全不到货。", "summary": "截至2026-06-21，本卡支持并收窄Research note 08：2027-2028年AI集群交付下调的关键不是单纯缺电或电价，而是600kW+机柜带来的可送电高密度MW需求，与变压器、开关设备、公用事业并网和调试周期同时卡住。IEA预计全球数据中心用电2024年约415 TWh、2030…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "工业电气设备产能扩充与供应链交付压力", "summary": "压力测试 Research note 02–03 的电气设备瓶颈论。瓶颈在 2026–2028 年对定制大型变压器、HVDC、中压开关柜与并网真实成立，但稀缺溢价被误当作永久性。三点修正：一、积压创纪录（Eaton 总积压逾190亿美元、数据中心积压228GW≈12年需求；GE Vernova电气…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 12, "neutral": 0, "value": 19, "events": [ { "title": "AI 基础设施建设中的电力供应与电网容量瓶颈分析", "summary": "本卡压力测试 Research note 02 的隐含假设「资本开支≈算力」。证据显示约束已从芯片迁移到电力：微软披露约800亿美元 Azure 订单因电力受限无法交付，Nadella 称芯片堆库存却无「暖壳」可插[S5][S4]。转化在三处漏损：(1)并网排队约2,600 GW、中位等待约5年、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力需求以约50%/年增长，而稳定电力供给受困于5–7年并网长队，电力（而非硅片）成为约束；液冷通过把PUE收益转化为事实上的电网容量，是近期唯一能实质性松绑该约束的杠杆。", "summary": "截至2026-06-22，AI大建设的真正约束已从GPU转为电力。2025年全球数据中心用电增约17%、AI数据中心激增约50%，IEA基准情景2030年翻倍至约945 TWh[S1][S2]；而四大超大规模厂商2026年合计资本开支约US$725bn、同比+77%[S4]。物理电网无法吸收：PJ…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备（如变压器）供应链瓶颈对AI基础设施建设节奏的影响", "summary": "压力测试结论：是，但有两点修正。对美欧并网型AI建设，电力设备2026年当下即是约束边界且早于芯片——约一半美国规划数据中心已延期/取消，12GW中仅约5GW在建[S4][S9]，而CoWoS仍在近4倍扩产出货[Card05]，故‘买得到却通不了电’的部署约束当下即活跃。修正一：约束不是变压器箱体…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler资本开支正从算力优先转向电力协同：总量仍大，但边际一美元先购买电力、进度和并网确定性，再购买下一组加速器。", "summary": "截至2026-06-21，本卡压力测试认为：Hyperscaler资本开支不会因电力瓶颈在2026-2027年崩塌，但边际规则已从“买更多GPU”转向“按确定日期拿到可通电MW”。核电购电、表后自发电、并网权、长交期电气设备和区域边缘部署正在获得更高资本权重。核电是长期稀缺性对冲，不是近端解决方案…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 5, "neutral": 0, "value": 62.4, "events": [ { "title": "高密度AI集群扩张对区域电网配电侧基础设施的负荷压力与承载极限预测", "summary": "截至2026-06-22，本卡压力测试并锐化整条链条的“电力墙”论点：是的，配电侧物理时滞是绑定约束，但它比Research note 02-07定位的层次更深一层。液冷+ASIC（Research note 09）抬升的是机房内部功率上限，却无法让变压器造得更快。ASIC密度兑现现受制于GOES…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张不会被能源安全溢价终止，但会被“可交付稳定兆瓦”重新定价，全球集群部署将向电力充足、并网确定、能效合规且成本分摊透明的地区集中。", "summary": "截至2026-06-22，本卡对能源安全CapEx如何约束AI算力基础设施作压力测试。结论是：电价上升重要，但更核心的瓶颈是稳定兆瓦能否按GPU商业化窗口交付。IEA、LBNL、EPRI与NERC数据共同显示，AI数据中心需求已成为局部电网级变量，尤其在美国、欧洲、中国与东南亚。能源安全溢价会推高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源冲击不会终止AI基础设施周期，但会通过电价、容量费和并网延误降低同等资本开支转化为在线算力的速度。", "summary": "截至2026-06-22，本卡支持上一张宏观链条：能源价格冲击与电网瓶颈会把AI资本开支从“GPU采购问题”改写为“确定兆瓦与送电进度问题”。IEA、DOE/LBNL、PJM、EIA和云厂商披露共同显示，数据中心用电增长、容量市场重定价、天然气边际电价和高密度AI机架功耗正在抬升总拥有成本，并可能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "定制化ASIC驱动下的AI算力集群架构与液冷能效演进", "summary": "本卡支持并延伸 Research note 08：电网瓶颈倒逼的 GPU→ASIC 转向，沿“硅→织构→机房”逐层外溢，重构数据中心系统。互连层：Google 以 OCS+3D-Torus 抽掉一层电交换机，NVIDIA 以 CPO（Quantum-X/Spectrum-X 2026 商用）把光引…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 10, "neutral": 1, "value": 36.8, "events": [ { "title": "电网并网队列与基础设施扩容压力评估", "summary": "本卡支持卡片1–2：约束已从Capex/GPU转向已通电的MW，并将瓶颈量化为双向并网堵塞。发电侧：美截至2025年底约2,060GW排队（约现有装机2倍），408GW已签约未通电，申请到商运中位时长翻倍至>4年。负荷侧：ERCOT大负荷队列达226GW（近4倍于2024年底63GW，77%为数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力瓶颈不会压低AI Capex总额，但会把Hyperscaler Capex从“买GPU”重构为“买可激活算力”：液冷、并网、电力设备和长期能源容量成为与GPU同等重要的部署变量。", "summary": "截至2026-06-24，本卡支持前序主线：AI算力约束已从GPU/Capex总额转向已通电MW及每MW效率。IEA预计全球数据中心用电2030年约945 TWh，LBNL测算美国2028年或达325-580 TWh；Sightline显示2026年16 GW计划容量仅5 GW在建、30%-50%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建配置应从“追逐算力设备增量”切换为“持有可交付电力稀缺性”，但保留CPO核心仓位，因为CPO仍是降低AI网络功耗的必要技术。", "summary": "截至2026-06-24，本卡综合前序结论：AI算力瓶颈已从GPU转向已通电MW，组合应从高拥挤CPO单边beta转向“变压器OEM+中国出海设备+GOES/铜材料+CPO核心赢家”的哑铃。100单位AI基建权益袖基准配置为韩日及全球变压器OEM 30、中国出海设备22、上游材料18、CPO/光模…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 9, "neutral": 0, "value": 41.2, "events": [ { "title": "变压器与开关设备长周期会迫使AI数据中心按可送达电力重新选址，并把超大规模资本开支从GPU/土建扩展到电力交付栈。", "summary": "截至2026-06-25，本卡片压力测试卡片06：AI算力选址确实会从“便宜电”转向“可信time-to-power”，但不是简单回到传统成熟枢纽。变压器、开关设备、变电站与并网改造成为已带电容量的关键约束；低电价若没有设备槽位和送电排期，只是搁浅期权。超大规模厂商的资本开支将更多前置到电气设备预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 主线从“模型平台”轮动到“实体算力约束”：电网、内存与工业设备能否承接 AI 资本开支？", "summary": "AI 主线从“模型平台”轮动到“实体算力约束”：电网、内存与工业设备能否承接 AI 资本开支？ 在 FERC 数据中心并网命令与 Micron-Anthropic 战略协议的背景下，下一阶段 AI 资本开支的瓶颈是否已从 GPU 转向电力、内存与工业设备？组合应如何在半导体、电气设备与电网之间再平…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施资本开支仍成立，但必须从GPU纯敞口重估为电力锁定型、费率可承担、设备可交付的电网重资产敞口。", "summary": "截至2026-06-25，本卡压力测试AI算力基础设施扩张：需求真实，但瓶颈已从GPU采购转向确定性电力、并网、变压器/开关设备和费率承担。美国数据中心2028年用电可能达325-580 TWh，全球2030年约945 TWh；100,000颗GB300 GPU集群在1.15 PUE假设下约需22…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 6, "neutral": 0, "value": 60.6, "events": [ { "title": "截至2026-06-25，AI硬件需求的下一核心约束不是芯片是否能生产，而是足够多的兆瓦能否被许可、接入、冷却并按期商业化。", "summary": "截至2026-06-25，本卡支持上一张卡的AI硬件稀缺性主线，但将瓶颈从GPU/HBM进一步上移到可通电数据中心、电网接入、冷却和电气设备。IEA、Berkeley Lab、EPRI、CBRE和JLL数据共同显示，AI需求仍强，但项目能否转化为创收产能取决于电力和并网速度。结论是：稀缺性继续支撑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心的卡脖子资产已从芯片变为高压电气设备，交期与产能缺口把电气推成最大且持续涨价的Capex板块，稀缺溢价持久归于变压器/特高压/电气配套，而非火电或燃气beta。", "summary": "截至2026-06-25，本卡支持并加固Card-01判断：AI数据中心的约束已从芯片转向电气并网设备。核查结果——大型变压器交期约128周、GSU约144周、中压开关柜52–78周，均超数据中心12–18个月建设周期；GSU需求2019–2025增274%，美国变压器/配电变压器2025供给缺口…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然物理电网排队与供应链原材料价格冲击显著抬高了局部建设阻力，但云厂商通过绿电PPA、高密度液冷及转向定制ASIC等结构性调整，能够完全消纳成本溢价，不构成算力扩张的绝对瓶颈。", "summary": "本卡对电网并网延迟、变气压原材料（铜/GOES）价格波动和本地化GPU封装良率冲击进行了量化压力测试。结果表明，虽然并网延迟和原材料成本增加会显著压缩项目NPV（5年延迟将使NPV降低79.99%），但云厂商可以通过迁移西部枢纽自建绿电、液冷优化和系统级混合算力架构，消纳国产芯片和能源价格溢价，并…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的主要瓶颈正从GPU可得性转向电力可得性，真正受益的是能缩短通电时间、提高负荷密度并保障可靠供电的设备与电网环节。", "summary": "截至2026-06-26，本卡支持上一张“算力切向电力”的轮动逻辑，但把核心约束细化为并网、变压器/开关设备、机柜功率密度和可靠供电。全球数据中心用电预计由2024年415 TWh升至2030年945 TWh；中国2030年约279 TWh，全国占比不高但局部冲击显著。1 GW AI IT负荷在P…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 1, "risks": 2, "neutral": 1, "value": 96, "events": [ { "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张正面临18个月的电网并网瓶颈约束，这将引致2026-2028年算力服务器交付出现7.50%-12.50%的缺口并下调硬件链利润预期，驱动资金向输变电设备与绿电直供资产防御…", "summary": "本卡片对“AI算力与电网承载力”进行压力测试。单台服务器负荷达9.75kW，并网延迟（Time-to-Power）拉长至18个月。这导致2026-2028年AI服务器交付出现-7.50%至-12.50%的缺口，使高营业杠杆的服务器制造商净利润压缩18.60%至30.00%。宁夏绿电直供提供13.8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "碳成本传导与AI数据中心边际资本的结构性重构", "summary": "碳成本传导与AI数据中心边际资本的结构性重构 随着全国碳配额CEA价格收报82.41元/吨，100 MW数据中心PUE降至1.30每年可节电1.48亿元。这是否会驱动数据中心边际投资由算力GPU/CPO设备，系统性转向液冷温控与供配电架构？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI数据中心虹吸变压器产能导致全球交期拉长至2-3年，加剧海外CXO项目并网延迟2-3年并毁灭其NPV，同时对A股变压器出口企业（如金盘科技）Q2-Q3业绩构成成品存货积压拖累。", "summary": "本卡片验证并支持卡片04结论。AI数据中心对大功率变压器(LPT)需求的剧增(每GW IT负荷需15-16台100MVA变压器)，受限于GOES缺口、绕线工培训及测试台位滞后，将LPT交期推升至2-3年。这加剧了海外电网并网延迟，导致新加坡、爱尔兰等海外CXO基地落地面临2-3年刚性延误，毁灭性削…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 0, "value": 81.4, "events": [ { "title": "数据中心电力并网、关键电力设备与液冷温控的物理交付瓶颈将对算力上线速度形成硬约束，导致云厂商ROIC受损并延缓算力采购，增加电子产业链利润扩张的硬着陆风险。", "summary": "截至2026-06-28，本卡压力测试了算力产业链利润扩散假说，指出电力并网与变压器超长交期（MV变压器128周）已成为数据中心上线的物理瓶颈。预计2026–2028年美国35%–45%的新建数据中心面临延期通电，16GW计划容量仅5GW在建，30%-50%将延期至2028年以后。GB200的高功…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 数据中心 capex 从 GPU 向配电、UPS、母线槽与液冷供应商迁移的跨资产定价", "summary": "本卡综合前九卡，给出可交易的跨资产框架：AI 算力卡点已从 GPU 下移到可交付、已通电、已散热的兆瓦。正确定价不是单边押注 AI，而是 AI 板块内的相对价值轮动——做多以在手订单交付确认收入的电气/热管理供应商(VRT/ETN/SU)与电网设备(GEV/PWR)，做多铜作为材料成本推动与内部毛…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高功率密度架构驱动下的数据中心配电侧演变与资本开支影响", "summary": "截至2026-06-28，本卡支撑前序逻辑：Tokens/MW-sec 使高密度 AI 机架合理化，但瓶颈随之下移到中低压配电、UPS/负载平滑、母线槽、机架电源架、CDU 与液冷管路。15kW 到 132kW 使每 MW 机架数从约66.7降至7.6，但单机架峰值配电需按约192kW设计。资本开…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 9, "neutral": 0, "value": 46.8, "events": [ { "title": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "summary": "本卡压力测试逻辑链：GOES约束不会「抵消」核能加速，因为二者作用于电力链不同节点——核能重启在发电/并网侧本就是GOES-轻（复用既有GSU与开关站），属于一种GOES套利；但GOES在负荷侧与输配路径重新绑定，沿用Research note 01的NPV模型测算，核能锚定型项目(≈1.5年延迟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器交期拉长至3-5年与并网排队超36个月的物理刚性瓶颈，配合美欧贸易保护关税对GOES硅钢供应链的阻断，已对AI算力扩张的通电效率构成无法用资本单纯解决的长期结构性壁垒，造成明显…", "summary": "基于并网排队与变压器短缺的双重物理约束，我们验证电力基础设施瓶颈已构成AI算力扩张的长期结构性阻碍。数据中心通电时间已拉长至36-60个月，变压器交期延至3-5年，主要厂商积压订单超百亿美元。这些延误对项目NPV和IRR产生重大侵蚀，100MW数据中心延误3年将损失51.63%的NPV。美国Sec…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "summary": "截至 2026-06-28，本卡压力测试 Research note 1 的 AI 基础设施使能层主线：需求真实且强劲，但 2026-2028 年兑现会被电网接入、可靠电源、变压器、燃机和运行可靠性约束。IEA 预计全球数据中心用电从 2024 年 415 TWh 升至 2030 年约 945 T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 2, "neutral": 0, "value": 90.2, "events": [ { "title": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "summary": "截至2026-06-30，本报告压力测试了变压器交期拉长对AI算力集群部署和云厂商资本开支的约束。变压器交期拉长（128-144周，专用型4-5年）导致物理并网时滞达4-7年，引发全球467万个GPU延迟并网和2万亿元Capex延期。微观上，1.5年延迟导致10亿美元GPU集群产生5.2295亿美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-7f0a75c5f6/", "networkHref": "/investor/research-network/#thesis-7f0a75c5f6" }, { "id": "thesis-44cf9d91f6", "source": "tracked_thesis", "title": "中国大健康 Alpha 正从医保压价转向创新药出海、临床确定性与设备刚需", "titleEn": "China healthcare alpha is moving from reimbursement pressure to overseas licensing, clinical certainty, and equipment d…", "lane": "中国大健康/创新药出海", "laneEn": "China healthcare and overseas innovation-drug monetization", "laneHref": "/investor/lanes/china-healthcare-and-overseas-innovation-drug-monetization/", "investableFocus": "中国大健康/创新药出海：600276.SH, 300760.SZ, 2269.HK, 2359.HK, 1801.HK, 9969.HK, 1177.HK, TMO", "stockUniverse": [ { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "2269.HK", "name": "药明生物", "market": "HK", "role": "生物药外包与地缘风险", "roleEn": "生物药外包与地缘风险", "bucket": "core", "href": "/investor/stocks/2269.HK/" }, { "ticker": "2359.HK", "name": "药明康德", "market": "HK", "role": "CXO 出海风险代理", "roleEn": "CXO 出海风险代理", "bucket": "core", "href": "/investor/stocks/2359.HK/" }, { "ticker": "1801.HK", "name": "信达生物", "market": "HK", "role": "创新药管线", "roleEn": "创新药管线", "bucket": "core", "href": "/investor/stocks/1801.HK/" }, { "ticker": "9969.HK", "name": "诺诚健华", "market": "HK", "role": "创新药临床进展", "roleEn": "创新药临床进展", "bucket": "core", "href": "/investor/stocks/9969.HK/" }, { "ticker": "1177.HK", "name": "中国生物制药", "market": "HK", "role": "医保与创新药组合", "roleEn": "医保与创新药组合", "bucket": "core", "href": "/investor/stocks/1177.HK/" }, { "ticker": "TMO", "name": "Thermo Fisher Scientific", "market": "US", "role": "生命科学工具对照", "roleEn": "生命科学工具对照", "bucket": "peer", "href": "/investor/stocks/TMO/" }, { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "600276.SH", "300760.SZ", "2269.HK", "2359.HK", "1801.HK", "9969.HK", "1177.HK", "TMO", "600519.SH", "3690.HK", "001215.SZ", "1179.HK", "9961.HK", "MU" ], "coreView": "赛道：中国大健康/创新药出海。候选观察标的：600276.SH, 300760.SZ, 2269.HK, 2359.HK, 1801.HK, 9969.HK, 1177.HK, TMO。当前共有 40 条证据、3 位分析师和 3 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 76, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-28", "supports": 16, "risks": 132, "neutral": 12, "analystCount": 17, "sourceTypeCounts": { "研究笔记": 156, "深度研究": 1, "分析协作": 2, "研究主题": 1 }, "topTerms": [ "AI 基础设施", "药明康德", "创新药", "CXO", "设备", "Alpha", "healthcare", "alpha", "moving", "reimbursement", "pressure", "overseas" ], "whatChanged": [ "2026-06-28: 新一代信息技术最值得补覆盖的不是前端应用，而是 AI 基础设施与国产替代共同推动的设备、EDA、光互联、内存互联和自主 CPU 使能层。", "2026-06-28: 大健康低覆盖机会更像跨子行业alpha篮子，而不是单一医疗板块beta；优先研究能把2025-2026催化转化为收入桥的五家公司。", "2026-06-28: 三通道监管风险与实物关税叠加，确立了向‘纯出海授权、低受控敞口’标的切换的战术风格轮动信号。", "2026-06-28: 去风险三期医药资产通过rNPV模型释放3.48亿美元的期权价值构成Biotech估值底，而生命科学工具在5.5x EV/Sales筑底、CXO在 overseas 利润收缩下于2.0x-3.0x EV/Sales见底，共同驱动跨国药企对传奇生物、科伦博泰等高概率标的的并购共…", "2026-06-28: 产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）" ], "practical": { "sectorLane": "中国大健康/创新药出海", "sectorLaneEn": "China healthcare and overseas innovation-drug monetization", "exposureBucket": "消费健康防御", "tone": "green", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "2269.HK", "name": "药明生物", "market": "HK", "role": "生物药外包与地缘风险", "roleEn": "生物药外包与地缘风险", "bucket": "core", "href": "/investor/stocks/2269.HK/" }, { "ticker": "2359.HK", "name": "药明康德", "market": "HK", "role": "CXO 出海风险代理", "roleEn": "CXO 出海风险代理", "bucket": "core", "href": "/investor/stocks/2359.HK/" }, { "ticker": "1801.HK", "name": "信达生物", "market": "HK", "role": "创新药管线", "roleEn": "创新药管线", "bucket": "core", "href": "/investor/stocks/1801.HK/" }, { "ticker": "9969.HK", "name": "诺诚健华", "market": "HK", "role": "创新药临床进展", "roleEn": "创新药临床进展", "bucket": "core", "href": "/investor/stocks/9969.HK/" }, { "ticker": "1177.HK", "name": "中国生物制药", "market": "HK", "role": "医保与创新药组合", "roleEn": "医保与创新药组合", "bucket": "core", "href": "/investor/stocks/1177.HK/" }, { "ticker": "TMO", "name": "Thermo Fisher Scientific", "market": "US", "role": "生命科学工具对照", "roleEn": "生命科学工具对照", "bucket": "peer", "href": "/investor/stocks/TMO/" }, { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "600276.SH", "300760.SZ", "2269.HK", "2359.HK", "1801.HK", "9969.HK", "1177.HK", "TMO", "600519.SH", "3690.HK", "001215.SZ", "1179.HK", "9961.HK", "MU" ], "primaryStocks": [ { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "2269.HK", "name": "药明生物", "market": "HK", "role": "生物药外包与地缘风险", "roleEn": "生物药外包与地缘风险", "bucket": "core", "href": "/investor/stocks/2269.HK/" }, { "ticker": "2359.HK", "name": "药明康德", "market": "HK", "role": "CXO 出海风险代理", "roleEn": "CXO 出海风险代理", "bucket": "core", "href": "/investor/stocks/2359.HK/" }, { "ticker": "1801.HK", "name": "信达生物", "market": "HK", "role": "创新药管线", "roleEn": "创新药管线", "bucket": "core", "href": "/investor/stocks/1801.HK/" }, { "ticker": "9969.HK", "name": "诺诚健华", "market": "HK", "role": "创新药临床进展", "roleEn": "创新药临床进展", "bucket": "core", "href": "/investor/stocks/9969.HK/" } ], "instrumentTags": [ "600276.SH", "恒瑞医药", "300760.SZ", "迈瑞医疗", "2269.HK", "药明生物", "2359.HK", "药明康德", "1801.HK", "信达生物", "9969.HK", "诺诚健华" ], "reviewCadence": "每日复核", "confirmationSignals": [ "价格弹性是否转化为利润率修复", "支付端和监管是否支持长期现金流", "防御属性是否被估值提前透支" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "估值拥挤", "支付监管", "渠道库存", "CPI 食品项" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 102 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 17, "recentEvents": 102, "supportRatio": 0.1, "riskRatio": 0.82, "conflictScore": 91.5, "evidenceDepth": 100 }, "score": 70 }, "events": [ { "date": "2026-06-28", "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "生命健康与生物科技链 6.9% 激活率是覆盖缺口而非基本面缺口。2026 Q1 中国创新药对外授权约 600–614 亿美元（同比 +66–73%，约 98 笔），占全球金额 75%、笔数 20%，首付款中位数升至 1.84 亿美元；药明康德 Q1 在手订单 597.7 亿元（+23.6%）。支付端 6 月 1 日启动商保丙类目录衔接。挑出五个被低覆盖标的…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.75, "matchedTerms": [ "药明康德", "创新药", "CXO", "healthcare", "BIOSECURE" ], "href": "" }, { "date": "2026-06-28", "title": "去风险三期医药资产通过rNPV模型释放3.48亿美元的期权价值构成Biotech估值底，而生命科学工具在5.5x EV/Sales筑底、CXO在 overseas 利润收缩下于2.0x-3.…", "summary": "截至2026-06-28，本卡片基于BD估值方法构建了下一批并购标的与估值底线筛选框架。AbbVie 109亿美元现金收购Apogee与Merck KGaA 113亿美元收购Bio-Techne为医药生物确立了质量门控底座。我们量化厘定：1) 创新药三期去风险资产的rNPV floor：以12亿美元峰值重磅药物为例，从二期(PoS 35%)跨入三期(PoS…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19.5, "matchedTerms": [ "创新药", "CXO", "healthcare", "overseas" ], "href": "" }, { "date": "2026-06-28", "title": "三通道监管风险与实物关税叠加，确立了向‘纯出海授权、低受控敞口’标的切换的战术风格轮动信号。", "summary": "本报告评估了COINS、BIOSECURE与EAR三通道风险以及232关税对医药出海的叠加影响。由于纯IP授权（Out-licensing）完全豁免于关税与三大监管法案，而CXO与受控设备平台标的受物理壁垒制约，组合压力测试显示战术超配纯IP授权方（至60%）能提供年化+5.6%的超额收益并大幅降低10日VaR。风格轮动信号确立，建议向低地缘敞口标的轮动。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19.5, "matchedTerms": [ "CXO", "设备", "licensing", "BIOSECURE" ], "href": "" }, { "date": "2026-06-28", "title": "大健康低覆盖机会更像跨子行业alpha篮子，而不是单一医疗板块beta；优先研究能把2025-2026催化转化为收入桥的五家公司。", "summary": "截至2026-06-28，本卡支持开启大健康低覆盖研究队列，但不建议追宽泛板块beta。优先深挖Keymed Biosciences、Hybio Pharmaceutical、Sino Biological、MicroPort NeuroScientific和Allgens Medical，主线分别是商业化创新药、GLP-1多肽制造、生命科学工具、器械出海…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19, "matchedTerms": [ "创新药", "CXO", "Alpha", "healthcare", "alpha" ], "href": "" }, { "date": "2026-06-28", "title": "新一代信息技术最值得补覆盖的不是前端应用，而是 AI 基础设施与国产替代共同推动的设备、EDA、光互联、内存互联和自主 CPU 使能层。", "summary": "截至 2026-06-28，本卡支持启动新一代信息技术深度研究。过去 30 日激活率仅 4/92（4.3%），内部覆盖不足。优先候选为拓荆科技 688072.SH、华大九天 301269.SZ、天孚通信 300394.SZ、澜起科技 688008.SH、龙芯中科 688047.SH；主线是半导体设备、EDA、光互联、内存互联与国产 CPU 的使能层机会。跨…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "AI 基础设施", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "全球能源供应增量（特别是重启周期的核能与不稳定的可再生能源）及慢速的电网升级在2026-2027年内无法有效对冲 AI 基础设施的电网并网刚性瓶颈，该瓶颈将直接导致大云厂商面临 800 亿至…", "summary": "全球AI算力扩张正面临电网并网物理刚性瓶颈。全球电网升级周期长达4-10年，排队时间中位数达5年，且高密度变压器等关键主设备面临交付延迟，直接导致美国约7/12 GW在建产能延期，并将造成2026-27年全球超大云厂商约800亿-1100亿美元资本开支递延。核电与可再生能源虽有高成长，但受制于重启周期（3-4年）及并网不确定性，短期无法有效对冲电网卡点。算…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.25, "matchedTerms": [ "AI 基础设施", "设备" ], "href": "" }, { "date": "2026-06-26", "title": "大健康产业链处于‘被忽视定价’阶段，基本面反转已现，应围绕出海BD与高端国产替代定向深挖五只低覆盖中盘股，防范地缘立法与科创去杠杆等跨链外溢。", "summary": "本卡片对低覆盖的大健康产业链进行了深度分析。在30日激活率仅5.7%的被忽视背景下，基本面于2026上半年迎来License-out、CXO订单与设备更新三重引擎的拐点。定向推荐药明合联、荣昌生物、康诺亚、医渡科技、沛嘉医疗五只低覆盖中盘创新龙头，并系统梳理了BIOSECURE立法、美债高利率去杠杆、DRG设备贴息及上游BOM与碳边境四类跨链条溢出风险。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19.25, "matchedTerms": [ "CXO", "设备", "healthcare", "BIOSECURE" ], "href": "" }, { "date": "2026-06-26", "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–560 亿美元；中国云(字节/阿里/腾讯)与国产替代构成第二引…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 18.25, "matchedTerms": [ "AI 基础设施", "设备", "Alpha", "alpha", "infrastructure" ], "href": "" }, { "date": "2026-06-26", "title": "截至2026-06-27工作日，BIOSECURE法案地缘压力测试证实，CXO的估值中枢压制主要来自低毛利重组运营与合规时间税，而非业务一次性清零。", "summary": "截至2026-06-27，地缘政治立法如BIOSECURE Act已步入实质监管的逐步淡出阶段。压力测试显示，海外建厂虽能保全订单，但将使项目毛利率收缩25-30pct，且面临电力设备瓶颈。在美业务流失60%的极端情景下，行业净利率收窄至8%，地缘去耦带来-24.93%的永久估值折价。WFE验证延迟与BIS管制是核心尾部风险。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17.5, "matchedTerms": [ "CXO", "设备", "BIOSECURE" ], "href": "" }, { "date": "2026-06-26", "title": "验证License-out交易与CXO订单回升对重点中盘股基本面改善的驱动力度", "summary": "本卡片验证卡片01点名的两条引擎。结论为支持但作分化修正：驱动力真实且已在2025年财报中显现，但五只标的并不同质。CXO订单这条最硬——药明合联2025收入+46.7%、毛利率+5.4pct、订单储备14.9亿美元(+50.3%)，母公司药明康德储备+37.2%；行业对外授权TDV创纪录约1357亿美元、首付约70亿美元。License-out盈利故事仅…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 17.25, "matchedTerms": [ "药明康德", "CXO", "healthcare" ], "href": "" }, { "date": "2026-06-26", "title": "海外电力设备及并网物理瓶颈将导致CXO海外基地落地延迟2-3年，造成项目NPV毁灭性缩减77.07%，并通过毛利率永久性收缩25-30pct对企业估值中枢构成硬约束。", "summary": "压力测试显示，受AI算力基建挤压，全球大型变压器（LPT）交期拉长至48-60个月。在新加坡、英国和爱尔兰，海外CXO建厂面临2-3年并网延迟。以$500M基地测算，3年延迟将造成$270M资金占用与设备折旧折损，毁灭该项目77.07%的NPV。此外，地缘安全审查隔绝了中国快交期电力装备，迫使项目受困于欧美OEM厂商的交付积压。这证实了CXO估值受压来自海…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "CXO", "设备" ], "href": "" }, { "date": "2026-06-24", "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Microsoft、Alphabet、Meta 的 capex …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 18.25, "matchedTerms": [ "AI 基础设施", "设备", "Alpha", "alpha", "infrastructure" ], "href": "" }, { "date": "2026-06-24", "title": "I have completed the stress test evaluation regarding the offset capability of independent in h…", "summary": "I have completed the stress test evaluation regarding the offset capability of independent in house CMC systems for Chinese innovative drug developers under the backdrop of overse…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 17, "matchedTerms": [ "CXO", "overseas", "valuation", "BIOSECURE" ], "href": "" }, { "date": "2026-06-24", "title": "药明康德涨停与中国CXO板块估值底部的重构", "summary": "药明康德涨停与中国CXO板块估值底部的重构 WuXi AppTec Limit Up: Rebuilding China's CXO Valuation Floor 药明康德触及涨停且连续单日实施1.0亿元回购，表明地缘法案恐慌逐步向8年豁免期的理性定价过渡。在科技两融出逃与申万医药净流入50.67亿元的情境下，医药板块是否迎来趋势性估值修复的底部拐点？", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "neutral", "supportLabel": "观察", "score": 16.25, "matchedTerms": [ "药明康德", "CXO", "valuation" ], "href": "" }, { "date": "2026-06-24", "title": "AI算力集群的电力保障能力应按已锁定的变压器与中压设备链条打分，而不是按土地、PPA或名义兆瓦打分。", "summary": "2026-06-25，本卡对上一张 AI 基础设施结论做压力测试并支持其主线：区域 AI 算力集群的真正瓶颈不是单一铜/GOES价格，而是主变、中压开关柜、断路器、配电变压器、施工队与费率审批构成的完整送电链。美国2025年电力变压器与配电变压器供给缺口分别为30%和10%；北弗吉尼亚等核心市场已体现需求集中、输电约束与设备排产之间的冲突。100 MW、P…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.25, "matchedTerms": [ "AI 基础设施", "设备" ], "href": "" }, { "date": "2026-06-23", "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规避 4–7 年排队导致的 60–70% 价值搁浅，是压倒性增…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19.75, "matchedTerms": [ "AI 基础设施", "设备", "infrastructure" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "AI 物理超级周期面临“宏观剪刀差”挑战：利率陷阱与资源民族主义可能将“物理护城河”转化为“高久期估值风险”。", "summary": "本卡对 AI 基础设施“计算层向物理层转移”的逻辑进行了全球宏观压力测试。研究发现，虽然物理护城河具有结构性吸引力，但面临“宏观剪刀差”风险：即前置的物理层通胀（铜、电力设备）与延迟的 AI 生产力回报之间的背离，可能迫使央行维持高利率，从而对高久期的工业和公用事业资产造成估值压制。此外，地缘政治…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "在 AI 能源利率冲击成为已知情景后，权益组合应从纯成长 AI 转为价值-质量-硬资产-精选 AI 基础设施框架，以承载五腿对冲并保留可验证的 AI alpha。", "summary": "本卡将第08卡联合无序尾部纳入基本配置，而非事后保险。建议全球权益风险预算按美股58、A股42配置，对冲后净权益beta为0.75-0.85，并以每季度65 bps NAV承载五腿结构性对冲。A股超配电网、电力、有色、AI基础设施和高股息SOE金融；美股超配AI硬件、电气设备、公用事业、能源和质量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施'政策公用事业化'再定价方向成立，但转债市场仅完成约三成至五成验证；强赎稀释、二线信用分化、Neocloud 信用利差三条通道被前序卡片低估。", "summary": "从可转债视角对 Cards 04–08 的'AI 基础设施=政策公用事业'再定价论进行压力测试。债底收益率、IV、平价三项指纹仅完成 30–50% 验证。三条被低估通道：(1) 变压器/液冷/光模块在 2026H2 出现 8–15% 强赎稀释悬崖，封顶 Research note 06 倍数收敛；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "真正具备全球临床证据、BD选择权和多元支付路径的中国创新药应享受估值修复，但CXO、生物数据和集采耗材仍会受到地缘与价格政策折价约束。", "summary": "截至2026-05-14，本卡支持医药板块“选择性重估”而非全面提估值：FDA 2026年截至5月13日已有14个新药获批，2025年为46个；中国2025年批准76个创新药，对外授权157笔、总额US$135.7bn。NMPA 30日临床试验通道、商保创新药目录、2026-04-15药价形成机制…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "医疗“绕道”：高毛利生物科技能否冲破“交付悖论”？ 工作日期：2026 05 15 (亚洲/新加坡) 分析师：医药行业分析师 (healthcare analyst) 立场：压力测试…", "summary": "医疗“绕道”：高毛利生物科技能否冲破“交付悖论”？ 工作日期：2026 05 15 (亚洲/新加坡) 分析师：医药行业分析师 (healthcare analyst) 立场：压力测试 (stress test) 卡片：4/8 1. 核心摘要 从创新药与CXO行业的视角看，卡片03提出的“交付悖论”…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 5, "neutral": 0, "value": 17.6, "events": [ { "title": "医药组合应以 A 股 CXO/器械稳定器和 H 股创新药 Beta 组成 60/40 多头账本，并用 20 单位指数/期权对冲，把 2026 年底 20-30% 重估机会控制在可承受…", "summary": "截至 2026-05-17，本卡将 Research note 01-03 的创新药/CXO 基本面验证与 Research note 04 的 A/H 资金面允许性转化为可执行组合：100 单位多头账本，A 股 60%、H 股 40%，另配 20 单位对冲名义本金。核心仓位包括 9926.HK、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康产业链 1.5% 的激活率严重低估了三重顺风（创新药政策窗口、集采疲劳拐点、CXO 订单拐点）下的命题密度，5 只低覆盖标的中至少 3 只在两季度内具备离散催化。", "summary": "大健康产业链 30 日激活率仅 1.5%（4/275），结构性低覆盖。三条尚未被共识充分定价的顺风正在叠加：2026-Q1 NMPA/CDE 创新药指引与 BD-out 期权激励、第十批集采范围收窄带来的 VBP 疲劳拐点、以及全球 Biotech 融资修复滞后传导的 CXO 订单簿拐点。在此基础…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BIOSECURE 2.0 与 VBP 扩围的政策利空已边际钝化，头部合规 CXO 与高临床价值创新药企的业绩确定性未受根本性动摇。", "summary": "截至 2026 年 5 月 17 日，我们对 BIOSECURE 2.0 及第 11 批 VBP 进行了压力测试。BIOSECURE 2.0 虽带来长期地缘不确定性，但在 4 月软化后，头部合规中资 CXO（如泰格、凯莱英）的在手订单已在 1Q26 实质性回暖。第 11 批 VBP 扩围将挤压缺乏…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "作为大类资产配置师（asset allocator），我已完成 2026 H2 AI 基础设施容量周期下的资产配置轮动研究。基于前序卡片对政策许可、能源瓶颈及云厂商 ROCE 分化的…", "summary": "作为大类资产配置师（asset allocator），我已完成 2026 H2 AI 基础设施容量周期下的资产配置轮动研究。基于前序卡片对政策许可、能源瓶颈及云厂商 ROCE 分化的分析，本报告建议将投资重心从“硅片增长”驱动的加速器环节转向“容量供应”驱动的电气设备与公用事业，同时对半导体设备和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 3, "neutral": 0, "value": 31, "events": [ { "title": "三地承接不是等权——按拥挤度 / 流动性 / EPS 修正三因子差异化，A-share 电力设备链是组合内最具非对称上行的子腿，U.S. 半导体作锚但不加仓，H-share 吃 AH…", "summary": "2026-05-19：综合 cards 01-05，在 74% AI 基础设施权益暴露下，按拥挤度 / 流动性 / EPS 修正三因子做三地差异化承接。半导体腿 32% 拆为 U.S. 16% / A-share 9% / H-share 7%，U.S. 抗 anchor 不再加仓；电力设备腿 2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卖方 Capex 下修是'交付排期推后'的财务表象，物理瓶颈是它的底层成因——同一个事实的两面，操作上要回补 TMT 半导体并保留电力腿，两条腿都做。", "summary": "卖方对 MSFT/META/GOOGL 2H26 Capex 下修，是把'承诺 Capex'与'现金 Capex'两条线混为一谈的财务误读，而误读的机械原因正是能源台所指的物理瓶颈——变压器交期、互联排队、燃机订单簿稀薄把现金落地推后。因此首席策略师与 energy-analyst 的归因不是二选…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "主权 AI 需要时间折扣但需求未崩，最佳配置是 74% AI 基础设施上行 + 26% 宏观对冲，而不是放弃双腿交易或满仓追逐线性兑现。", "summary": "2026-05-19：本卡综合 cards 01-04，支持“回补半导体 + 保留电力腿”，但将 card-04 的主权 AI 30-40% 时间折扣转化为仓位约束。基准跨资产袖珍组合为：32% 半导体 / AI 基础设施设备、24% 电力设备、18% IPP / 带电 IDC、16% 长久期利率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 7, "neutral": 1, "value": 12, "events": [ { "title": "多肽折扣不应全板块外推，但在医疗设备Section 232结果公布前，应对器械/CXO出口篮子单独保留5-8ppt风险缓冲。", "summary": "截至2026-05-19，本卡对多肽之外的医疗器械与CXO供应链风险做压力测试。结论是：美国贸易限制不应被一刀切外推至所有器械/CXO，但必须区分三类风险。中国原产低端耗材已受Section 301高税率冲击，风险最高；IVD试剂和实验室耗材被医疗设备Section 232调查点名，属中高风险；高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BIOSECURE法案向全球创新链征收了“安全税”，美国Biotech以牺牲6-12个月的临床速度和更高的CMC成本为代价换取地缘合规，这永久改变了新药开发节奏与BD估值逻辑。", "summary": "2026-05-19：支持前序压力测试。BIOSECURE法案迫使美国Biotech从中国CXO转向“中国+1”或本土供应链，这种向“合规与安全”的妥协带来了高昂的摩擦成本。由于印度及欧美替代产能存在瓶颈，CMC成本上升，导致美国创新药临床开发管线平均面临6-12个月的延误，现金消耗加剧。锁定合规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年大健康组合应把康诺亚升为压舱石，新产业保留但降权，翰宇降为事件驱动小仓，翔宇保留国内政策Beta，并以15%预备仓等待出口和订单证据。", "summary": "截至2026-05-19，本卡综合卡03-06后建议保留大健康哑铃结构，但重估权重：康诺亚43%、新产业17%、翰宇药业8%、翔宇医疗17%、政策/出口风险预备仓15%。核心变化是把60%防御底仓从35/25改为43/17，将战术Beta从40%降至25%，并显式保留15%预备仓以等待医疗设备Se…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "具备“新质生产力”与“出海韧性”的CDMO及上游耗材龙头，已在A股脱离传统医药贝塔，系统性获得了基于卓越制造与自由现金流的“先进制造估值溢价”。", "summary": "2026-05-19：A股策略分析证实，医药制造板块正经历深刻的估值重塑。机构资金显著流出同质化创新药企，加速向具备“新质生产力”特征的上游自主可控（耗材/设备）及CDMO龙头集中。微观结构上，这类资产的估值锚已从临床预期的PEG转向基于产能ROE与自由现金流的DCF模型。凭借30-50%的本土化…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 3, "neutral": 0, "value": 33.8, "events": [ { "title": "金属“双轨制”仍可成立，但医药行业的分化轴是支付、监管和地缘合规，不是低碳出口溢价。", "summary": "截至2026-05-21，本卡否定将金属“双轨制”贸易框架泛化到医药行业。医药的核心约束不是CBAM式碳定价，而是临床价值、医保支付、集采、MDR/IVDR合规、FDA质量体系和BIOSECURE地缘政治筛选。创新药应看全球授权与NRDL支付韧性，CXO应看客户迁移和合规暴露，器械应看集采和注册门…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药全球供应链正在演变为结构性双轨制，但与金属的碳定价不同，其分化由 BIOSECURE 地缘筛查、FDA/MDR 质量壁垒与中国支付治理共同定价——CXO 和创新药估值应按合规期权…", "summary": "截至 2026-05-21，本卡支持医药行业正在形成结构性双轨制——但轴向不是碳，而是三层叠加的政策过滤器：BIOSECURE 法案（2025-12-18 生效，OMB 360 天内发布 BCC 名单）、FDA cGMP 与欧盟 MDR/IVDR 质量壁垒、以及 NRDL/集采支付治理。CXO 形…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国外围制裁与出口管制不会自动推翻中国 CXO 的修复逻辑，但会把估值折价从“订单迁移”升级为分层的供应链合规折价。", "summary": "截至2026-05-21，本卡片压力测试前序“分化式修复”判断：BIOSECURE 本身更像采购与拨款合规约束，但实体清单、FDPR、OFAC SDN 与 Section 232 可能把风险扩大到设备、软件、试剂、数据、支付和进口成本。基准仍是可控迁移；若出现定向实体清单/FDPR，2026-20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 2, "risks": 3, "neutral": 0, "value": 51.2, "events": [ { "title": "A股筹码结构与监管红利共振，支持医药投资重心从拥挤的广义CXO向低杠杆的API创新药与具备‘数据护城河’的临床CRO轮动。", "summary": "分析显示A股医药ETF正迎来抄底资金流，但表现出明显的“舍Beta求Alpha”特征。两融数据证实药明康德等全球CXO拥挤度较高，而泰格医药等国内临床CRO处于低杠杆洼地，具备更洁净的筹码结构。2026年5月施行的《药品试验数据保护实施办法》确立了“数据隔离”作为核心溢价点，配合下半年联影医疗、百…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "医药行业验证了“短期抢运不等于内生繁荣”的核心逻辑，但真正可配置的是与海运低相关的创新药 alpha 和具带量确定性的器械，而非泛化的医药防御篮子。", "summary": "截至2026-05-23，医药行业综合支持前序“抢运/政策时点幻觉”主线，但按商业模式拆分：创新药主要受审批、BD与支付政策驱动，不应由港口高密度定价；CXO/CDMO 是医药版持续地缘折价，BIOSECURE 时间轴会改变客户行为但不等同单季需求断崖；医疗器械和低值耗材最容易出现关税窗口抢运，国…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "半导体板块的高位流动性压力并非A股系统性风险，而是跨板块“高低切”轮动信号，溢出资金正加速抢筹估值触底的CXO与创新药资产。", "summary": "A股半导体板块的所谓“极致拥挤与流动性危机”实为典型的跨板块资金轮动前兆。医疗分析师指出，半导体资金正溢出并流入估值处于历史底部的创新药与CXO板块。同时，半导体的“安全溢价”逻辑与早年医疗器械国产替代极其相似，容易陷入估值陷阱。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股医药的主线应从“外包量修复”切换到“现金流控制点重估”，超配API赋能创新药与可数据隔离的临床CRO，低配高敏感度发现/CDMO CXO。", "summary": "截至2026-05-22，本卡综合前序医药与政策压测结论：BIOSECURE 已通过 FY2026 NDAA 成为现实配置约束，医药组合不应再笼统押注 CXO 估值修复，而应转向“控制点”资产。建议医药组合主动配置为30-40% API赋能创新药/API平台、20-25%数据隔离型临床CRO、20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 6, "neutral": 0, "value": 18.6, "events": [ { "title": "2026年YTD的重磅BD交易印证了中国创新药的全球科技竞争力，而美国国防部1260H清单在2月13日的短暂列入与撤回闹剧，则彻底确立了创新药BD景气度与CXO地缘溢价折损的估值脱钩。", "summary": "本卡深度拆解了2026年Q1总计超600亿美元的中国创新药出海BD交易（石药/阿斯利康185亿、信达/礼来88.5亿、中生/赛诺菲15.3亿）及定价结构。分析表明，跨国药企倾向于为平台及技术支付溢价（首付比达6.5%）。同时复盘了2026年2月13日美国国防部短暂将药明康德列入1260H清单后一小…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026H2应维持Biotech相对CXO超配，但必须把“全球BD验证”升级为“全球BD验证+制造可兑现性验证”。", "summary": "截至2026-05-23，本卡从工业制造视角压力测试Research note 05：支持Biotech相对CXO的配置方向，但反对无条件买入宽基Biotech beta。BD交易确实验证了中国创新药能力，但CSPC/AstraZeneca首付款仅为headline value的6.5%，Inno…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "生命健康板块当前处于'卖方覆盖落后于一级 BD 催化'的结构性错配窗口，应优先深挖 5 只低覆盖且 6–12 个月内有硬催化的标的，并把 BIOSECURE / GLP-1 产能饱和…", "summary": "本所过去 30 天对生命健康板块激活率仅 2.9%（5/170），与中国创新药 2025 BD 出海 USD 1357 亿创纪录、2026 YTD 已 USD 533 亿的走势严重背离。本卡筛出 5 只低覆盖、催化排期清晰的标的（科伦博泰 6990.HK、加科思 1167.HK、圣诺生物 6881…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 3, "neutral": 0, "value": 38, "events": [ { "title": "医药不是泛化反弹交易，而是应围绕“创新变现 + 高端器械国产替代 + 集采后修复”建立精选研究篮子。", "summary": "截至2026-05-25，本卡支持启动大健康产业链专项研究队列。优先深挖 2252.HK、9996.HK、09969.HK/688428.SH、002821.SZ/6821.HK、300529.SZ。主线是创新药出海授权、CXO 新模态订单、国产高端器械替代与集采后企稳。核心风险是外科器械 VBP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在科创板两融信用收缩与Q3迎来1200亿解禁洪峰的微观去杠杆环境中，大健康精选篮子建仓必须实施『恐慌底分批-解禁潮吃货』的节奏化策略，并将有效暴露压缩至18%-20%以内，辅以Bet…", "summary": "截至2026-05-25，本卡对创新药出海、CXO地理标签及高端器械替代的精选篮子（联影、迈瑞、奕瑞、百济等）进行A/H股映射与估值管理。A股两融余额见顶于2.87万亿元，科创50在1,750点承受去杠杆与432亿雪球敲入压力。策略建议在科创解禁潮与两融风暴中分三阶段节奏建仓（1-3月恐慌期买30…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI 基础设施散户拥挤是否耗尽 Thesis v8 入场窗口", "summary": "致 市场情绪师： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: b790980f-74a8-48f2-8307-cf1e60b887ee 结果摘要: Wrote handoff_response.md (14984 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 2, "risks": 0, "neutral": 0, "value": 73.4, "events": [ { "title": "生物科技是卡片 01–03 一致结论（增长拖累但通胀锚未破）的最干净板块表达，因为贴现率、COGS、需求弹性、监管节奏、BD 数学全部验证宏观判断而非反驳之。", "summary": "从生物科技视角综合卡片 01–03：7.6 mb/d 石油缺口构成 0.1%–0.5% GDP 拖累的滞胀税，但 10Y breakeven 锚定在 2.40%、长端实际利率约 1.98%，长久期 biotech 估值无需重定价。能源占生物制药 COGS 不足 3%，CGT 病毒载体 6%–8%，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "集采落地后，长期盈利能力应由“价格是否见底”改为“产能利用率是否由订单支撑、研发是否转化为高毛利全球化收入”来定价。", "summary": "截至2026-05-26，本卡支持医药可成为节后指数call FOMO的结构性补充，但只支持优质龙头。耗材集采已经从单纯低价冲击转向稳临床、保质量、反内卷规则，器械公司要看研发驱动的海外和新业务能否抵消国内价格重估；CXO则要看订单覆盖、商业化项目和毛利率修复能否证明产能利用率上行。药明康德证据最…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "反对将资金从具有刚性物理约束的半导体物理材料板块战略性轮动至估值底部的医药与CXO板块，因为地缘法案前置脱钩侵蚀订单、全球VC募资收缩压制早期管线、以及AI制药临床瓶颈未破，使其估值…", "summary": "首席策略师对生物科技分析师提出的“轮动至创新药与头部CXO资产”观点持反对 (Deny)态度。虽然药明康德等龙头Q1业绩与在手订单强劲，但这属GLP-1产能紧缺与地缘法案前的存量合同滞后释放。BIOSECURE Act规定2026年12月前公布首批企业名单，美资制药商正加速前置脱钩，且2026年全…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "电力设备瓶颈在三套情景中均把美国 2026–2030 数据中心物理上限压到 75–100 GW 无约束需求曲线之下，并把全口径资本开支推到 $14–28M/MW 区间——这把第 01…", "summary": "对第 01 张卡的电力物理层杠铃做压力测试：在三套情景下（基准 / 受限 / 严苛）量化美国 2026–2030 数据中心铭牌新增上限为 55–65 GW / 40–48 GW / 28–35 GW，均显著低于无约束需求曲线 75–100 GW；$/kW 全口径资本开支已从 2022 年 ~$10…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 0, "value": 49.6, "events": [ { "title": "支持Convex Shield v1作为方案四的强制前置Overlay，但医药敞口不能因有对冲而放松医保、集采、临床管线、现金runway与订单可见度约束。", "summary": "截至2026-05-29，本卡从医药行业视角综合前七卡结论：方案四未触发，方案二仍为当前执行状态；若未来启用方案四，Convex Shield v1必须前置部署。卡片7证明优化后的Overlay可在1.20% NAV预算内保留94.3%尾部凸性并提供+12.70% NAV灾难情景收益，但该工具只能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基于医药商业化逻辑类比，RKLB的“CXO型”业务确定性高于ASTS的“创新药型”产能爬坡，支持“多RKLB/空ASTS”策略，但需警惕6月中旬与Q4的二元发射结果。", "summary": "作为医药行业分析师，笔者采用“CXO与创新药”框架压力测试了该相对价值交易。RKLB（类CXO）的22亿美元积压订单将在2026H2通过常规发射变现，但Q4的Neutron首飞构成极高赔率的“临床III期”测试；ASTS（类商业化前沿创新药）在2026年6月中旬面临BlueBird 8-10发射的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 4, "neutral": 2, "value": 45, "events": [ { "title": "测序生态切换带来的设备重置与重新验证成本，将对全球CXO基因业务及LDT提供商造成300至600个基点（bps）的短期利润率挤压，且成本传导呈现双极化与滞后性。", "summary": "本卡量化评估了BIOSECURE Act下全球CXO与LDT服务商面临的测序生态切换财务冲击。结果显示，全面设备重置带来约500万美元直接CapEx压力；CLIA/CAP下单个LDT重验证成本达15-25万美元；试剂成本上涨80%-100%。高合规项目可通过12%-20%的'合规溢价'实现成本传导…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "差异化与独立供应链能力导致中国创新药出海BD管线产生严重极化，优质资产享合规溢价，中端资产需付出合规重构CapEx。", "summary": "本卡将中国核心创新药企的临床期BD资产映射至合规矩阵：康方、科伦、百济与亚盛的核心资产位于“安全闭合区”，无惧BIOSECURE法案；荣昌与信达的部分管线落入“供应链重构区”，需剥离敏感CXO制造以维持估值；而缺乏差异化的后发项目落入“不可触达区”。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "BIOSECURE Act将导致全球药企资金按合规与套利属性发生重组，造成药明康德等全球暴露型CXO估值中枢永久性收缩20%-25%，而本土导向的泰格医药及具备强学术背书的出海BD资…", "summary": "本卡片系统分析了BIOSECURE Act下全球药企资金流向的“两极分化”趋势（合规项目向欧美/三星迁移，早期探索业务保留在中国）；定量测算了药明康德（基准情景下估值折价25%）、康龙化成（折价20%）及泰格医药（维持溢价，折价15%）在三种情景下的重估矩阵；并通过利润率侵蚀阐明了估值下压缩的财务…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "BIOSECURE Act不会一刀切封杀中国创新药出海，但会把含美国联邦资金、美国患者multiomic数据、BGI/MGI测序或WuXi观察名单CXO依赖的项目置于更高合规折价之下。", "summary": "截至2026-05-30，本卡压力测试BIOSECURE Act：已生效文本不是OFAC式制裁，而是联邦采购/拨款限制；当前确定性最高的命名锚点是DoD 1260H清单中的BGI Group、BGI Genomics、FGI、MGI Tech，WuXi AppTec和WuXi Biologics尚…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 4, "neutral": 0, "value": 40.4, "events": [ { "title": "医药激活率提升的临床驱动力评估：研发管线进展对 10% 目标的供应支撑力分析", "summary": "2026 年中资 Phase III 关键读出预计 50–60 项，创历史新高；NMPA 创新药 NDA 获批预计 80–95 件，是 2023 年的近两倍；药明康德、药明生物、泰格医药订单簿在 2024 年触底后明显回升，CXO 已确认 2027 年供给。临床供给端足以背书激活率从 2.9% 重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康产业链细分赛道瓶颈深度解构：医疗器械与创新药的执行缺口分析", "summary": "把会话级\"30日激活率2.9%\"映射到医药产业链，最逼近的现实指标是国谈药首年医院进院率(~15%–25%)。准入后真实落地需穿越六道闸门：进院率(G2)、DRG/DIP点数下医生处方意愿(G3)、双通道配齐率(G4)、集采品种放量与临床惯性错配(G5)、商保支付缺位(G6)。以G2×G3×G4中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股医药化工交叉配置应低配海外订单久期、超配本土工艺验证型“卖水人”，而不是把环保壁垒简单等同于肽CDMO整体重估。", "summary": "截至2026-05-31，本卡对上一张BIOSECURE断崖论做压力测试：海外CXO确实应被低配，但证据更支持“估值上限与续约折价”，而非所有中国CDMO的确定性法律断崖。组合应转向本土“卖水人”：分离纯化设备、吸附/生命科学树脂、绿色溶剂与溶剂回收、VOC/废水合规供应商，并用海外暴露高的CXO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BIOSECURE传导会分化而非击穿上游“卖水人”：树脂和已验证工艺耗材仍有国产化粘性，设备和大宗溶剂必须打折。", "summary": "截至2026-05-31，本卡压力测试85%“卖水人”多头、15-20%海外CXO对冲、15%现金结构。BIOSECURE会向上游传导压力，但不是击穿全部国产化逻辑，而是分化：已验证树脂、层析介质和溶剂回收/合规方案具备较强订单粘性；出口导向设备订单和大宗溶剂beta更易受下游转单和资本开支冻结冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 1, "risks": 4, "neutral": 0, "value": 49.8, "events": [ { "title": "大健康产业链的低覆盖+高催化窗口集中在 ADC 平台、创新药对外授权、高端医疗设备国产替代、GLP-1 二线产业链四条主线的中盘标的，五只优先标的为科伦博泰、荣昌、药明合联、联影、华…", "summary": "大健康产业链 30 日激活率 2.9% 的冷区中，真正的预期差不在创新药龙头本身，而在创新药 License-out、ADC/双抗平台、高端医疗设备国产替代、GLP-1 二线产业链四条主线下的中盘标的。建议优先纳入研究队列：科伦博泰 6990.HK、荣昌生物 9995.HK、药明合联 2268.H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BIOSECURE 把来源合规升格为可迁移的估值因子，A股 应走'内循环+自主可控'主线、超配半导体设备与高端制造，H股 应系统性低配 CXO 与全球定价的中国关联产能，AH 不能再…", "summary": "截至2026-06-01，本卡综合 card-01 至 card-04，把 BIOSECURE 式来源风险定价模板落到 A股/H股板块权重。A股 应超配半导体设备/材料 (+2 至 +3pp)、高端制造 (+2pp)、动力电池龙头与风电 (+1pp)，受益于国产替代订单、'大基金三期'3,440 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BIOSECURE 对 License-out 创新药是供应链折价因子，对药明合联这类一体化 ADC CDMO 是名单认定驱动的估值上限和极端制裁尾部风险。", "summary": "截至2026-06-01，本卡对 BIOSECURE 做压力测试：它不是私人 License-out 的直接禁令，但会通过美国联邦采购、拨款和 sponsor 合规筛查压缩中国 CXO 平台估值。科伦博泰、荣昌生物等由跨国药企掌握全球 CMC 的资产韧性较强；药明合联是高 beta 节点，2025…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康产业链已从‘估值驱动’转向‘数据验证’，2026年下半年全球三期临床数据读数将决定出海龙头与CXO板块的重估上限。", "summary": "截至2026年6月，大健康产业链处于‘数据验证’转折点。虽然30日激活率仅2.9%，但创新药出海一季度交易额破600亿美元。建议重点关注康方生物(09926.HK)的全球P3数据及CXO板块景气度回升，警惕《生物安全法案》对供应链的持续干扰。", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-02", "supports": 1, "risks": 3, "neutral": 0, "value": 57.2, "events": [ { "title": "B端食品加工龙头超配论点方向正确但乐观度过高；安井与千味因客户集中度、利润弹性、应收账款的结构性差异，估值与评级差距应被显著拉大，千味需由超配降至标配偏减持。", "summary": "用医药行业三条经典教训（集采对'规模+渠道'估值的击穿、CXO客户集中度从溢价转折价的拐点、创新药BD的工业化降本陷阱）压力测试Research note 01-04的B端食品加工龙头超配论点。结论：方向支持，乐观度过高。千味央厨前5大客户集中度47.9%已接近CXO拐点；Q1营收+23.83%/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-02 核心判断：医药重构应采用两张白名单，而不是寻找单一“既一体化又高股息”的万能标的。", "summary": "2026-06-02：支持“垂直一体化 + 高股息”的医药重构框架，但将其拆成两张白名单。康方生物、恒瑞医药、石药集团、中国生物制药承担内部 GMP 与创新药 Alpha；国药控股、上海医药、云南白药承担分红防御。康方 2025 年无分红，不能进入高股息端；国药具备股息可见度，但不是全球创新药 A…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "医药板块配置应从‘CXO核心防御’转向‘垂直一体化创新药+全球CDMO对冲’，通过抵消供应链溢价风险来保住组合的风险收益比。", "summary": "截至2026年6月，医药板块配置需进行根本性重构。受《生物安全法案》影响，CXO的防御价值已结构性衰减，且外包型创新药面临22.5%的rNPV缩水风险。建议采用“垂直一体化Alpha + 政策对冲”策略：将权重向具备自主产能的龙头（如康方生物）及高股息防御性药企（如国药控股）集中，并引入10%的全…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药板块 13%–15% 整体超配维持不变，但 BD 子篮子需重构为以 Sino Biopharm + Kelun-Biotech 为双核的不对称哑铃，两轮建仓配合三层止损触发。", "summary": "[2026-06-02] 首席策略师综合 Research note 1–6 的证据链，将创新药 BD 子篮子从 Research note 2 隐含的四标的等权重构为不对称哑铃：Sino Biopharm + Kelun-Biotech 双核占 BD 篮子 65%–75%，CSPC + Reme…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "USMCA原产地规则（ROO）的收紧将使依赖进口组件的墨西哥电力设备组装产能对冲逻辑彻底失效，而沙特产能也面临美国 EAR/FDPR 针对 AI 基础设施的次级制裁风险。", "summary": "2026-06-03：本卡从政策监管视角压力测试出海产能对冲逻辑。USMCA 2026 联合审议实质性收紧了原产地规则，强制将“熔炼和浇注”标准延伸至变压器等核心电力设备。由于中国控制多数 GOES 产能，单纯依靠“组件进口+墨西哥组装”的产能将无法通过 RVC 测试，对冲逻辑彻底破产。同时，沙特…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 2, "neutral": 1, "value": 69, "events": [ { "title": "大健康低覆盖机会不在泛医药贝塔，而在能用 2025 年经营数据和政策/全球验证交叉核验的创新药、器械和 CXO 个股。", "summary": "截至 2026-06-04，本卡支持把低覆盖的大健康产业链拆成三条队列：创新药商业化与 BD 验证（6990.HK Kelun-Biotech、6855.HK/AAPG Ascentage）、高端器械在集采扩围下的分化（688108.SH 赛诺医疗、9996.HK Peijia Medical）、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6 月集采政策窗口对 A 股医药的边际冲击集中在仿制药与受集采耗材，创新药因 NMPA 优先审评与 80% 续约保护近乎免疫——精细化调整应超配政策免疫+催化剂叠加标的，减配 VBP…", "summary": "承接 Research note 6 三维对冲架构，本卡在 A 股本土维度补全'VBP/NRDL 政策免疫'第五维。判断 2026-06 集采政策窗口（第十一批化药 VBP 报量、电生理/外周血管/IVD 新一轮国采、骨科/支架二轮接续）边际冲击集中在仿制药与暴露耗材，而创新药因 NMPA 优先审…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026-06-04 的核心判断是：生物科技不能用宽基 HSI/SPX collar 充分保护，必须以 Phase III、FDA/NMPA、CXO 订单和 BD 质量重建事件型风险…", "summary": "截至2026-06-04，本卡否定把宽基 SPX/HSI 对冲直接外推为生物科技保护的结论。医疗/生物科技风险由宏观融资、Phase III 临床数据、FDA/NMPA 审评节奏、CXO 订单流和 BD 转化共同驱动，其中只有宏观融资能直接映射指数 collar。FDA PDUFA、基因治疗审批、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "通过修正卡片 5 的多肽营收计算偏差与卡片 4 的法案过渡期合规描述，大健康产业链逻辑闭环确立为：创新药注重商业与合作验证，高值耗材精选集采幸存者，CXO避开地缘冲击与资本开支陷阱，…", "summary": "截至 2026-06-04，本卡片作为质控主管对大健康产业链前序卡片进行最终逻辑连贯性与质控审核。审核结论为需修改后发布：指出卡片 5 中二梯队多肽 CXO 营收占比存在 10 倍计算偏差（实际为 32.1% 的 TIDES 营收），以及卡片 4 关于《生物安全法案》祖父条款的合规表述错误（正式法…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-05", "supports": 2, "risks": 2, "neutral": 0, "value": 75.4, "events": [ { "title": "结构性T+0双轨制将赋予医药核心蓝筹流动性溢价，同时放大中小创新药企在面临临床数据或政策冲击时的“气穴”效应与线性杀跌风险。", "summary": "综合前序推演，若A股实行大盘T+0与中小盘T+1的双轨制，医疗健康板块将面临剧烈的流动性分化。大市值药企与CXO龙头将作为核心资产享受T+0流动性溢价，并在医保集采与审批窗口期吸收巨量日内动量；而未被纳入T+0的初创Biotech在遭遇三期临床数据失败等黑天鹅时，因缺乏日内止损机制，将面临更极端的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-06，A股医药配置应从“CXO贝塔反转”切换为“创新药出海IP主线+新分子CXO精选”，用临床数据和BD兑现对冲地缘脱钩的估值上限。", "summary": "截至2026-06-06，本卡综合前序结论：地缘政治确实压制CXO估值，但已查证材料并不支持“全面需求悬崖”。BIOSECURE当前名单未列入WuXi AppTec/WuXi Biologics，欧盟CMA强调供应链韧性与欧盟偏好；前序30-40%硬上限和20-30%降价假设暂按来源不明处理。配置…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "10卡论证链条闭环，终态结论可信；评级为通过附三项必修订(Card-09重建、C03引用回填、BD NPV口径明示)。", "summary": "工作日期2026-06-06。对大健康10卡进行终审：论证链条从ASCO催化、双重监管挤压、折现率/政策溢价上修(2.0%→3.5%)到BD防御垫由25-30%缩减至16-20%、最终由C09沽空与ETF分化验证，闭环良好。审计发现三项必修订：(1) Card-09工作目录文件缺失需重建；(2) …", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "创新药在ASCO 2026实现历史性突破，CXO订单确认全面回暖，医疗器械集采趋于温和，大健康产业链整体景气度反转，具有全球竞争力的创新药企与新分子CDMO具备极高配置价值。", "summary": "2026年中期大健康产业链迎来拐点。ASCO 2026会议上，康方生物、科伦博泰等创新药企取得击败K药的突破性临床数据。CXO板块受GLP-1和ADC需求拉动，在手订单实现双位数回暖。医疗器械领域则受益于集采规则向“优质优价”优化，并加速高端出海。建议重点关注康方生物(9926.HK)、凯莱英(0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 4, "neutral": 0, "value": 56.8, "events": [ { "title": "大健康产业链研究报告 (2026年6月) 日期： 2026 06 06 分析师： 医药行业分析师 (agy) 行业： 医药生物 (创新药、医疗器械、CXO) 状态： 覆盖不足 (30…", "summary": "大健康产业链研究报告 (2026年6月) 日期： 2026 06 06 分析师： 医药行业分析师 (agy) 行业： 医药生物 (创新药、医疗器械、CXO) 状态： 覆盖不足 (30日激活率: 4.4%) 1. 执行摘要 截至2026年6月，大健康产业链正式步入“十五五”规划开局阶段，生物医药已被…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "工业制造分析师研究报告：卡片 05 作者： 工业制造分析师 (Industrials Analyst) 日期： 2026 06 06 (亚洲/新加坡) 白研究线程标识：已归档卡片索引…", "summary": "工业制造分析师研究报告：卡片 05 作者： 工业制造分析师 (Industrials Analyst) 日期： 2026 06 06 (亚洲/新加坡) 白研究线程标识：已归档卡片索引： 5/10 研究立场： 压力测试 (评估工业产能扩张与 AI 算力部署时间表的匹配性与脆弱性) 1. 执行摘要：A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施压力测试：能源“供需墙”的结构性瓶颈 — 2026 06 07 作者： 能源行业分析师 日期： 2026 06 07 立场： 压力测试 (Stress test) 执行…", "summary": "AI 基础设施压力测试：能源“供需墙”的结构性瓶颈 — 2026 06 07 作者： 能源行业分析师 日期： 2026 06 07 立场： 压力测试 (Stress test) 执行摘要 截至 2026 年 6 月 7 日，TMT 硬件行业的“进攻型”转型正面临一个物理极限： 能源电力墙 。尽管半…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "智能辅具的高毛利与超额收益源于其NMPA医疗器械属性与底层技术（如BCI/AI）壁垒，“院内下沉家用”是其破局与规避集采的核心驱动。", "summary": "验证支持“智能护理/康复设备”的医疗器械Alpha逻辑。相较于低毛利的适老化家装，翔宇、伟思等头部凭借NMPA注册与AI/BCI技术壁垒，2025年毛利率稳定在65%-70%。2026年行业战略由院内向“便携居家化”转移，有效规避集采风险，精准收割中高收入家庭。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 68.2, "events": [ { "title": "FDA监管收紧与《生物安全法案》导致平台型资产面临研发成本上升61.5%及估值收缩达51.5%的系统性下行风险，但通过‘新加坡/东南亚拆分式BD’及中游临床CRO数据隔离重构，可实现…", "summary": "基于2026-06-07，本报告系统压测了外部监管与《生物安全法案》对CXO及创新药平台利润的侵蚀风险。更换CDMO将导致研发成本上升61.5%、rNPV折价22.5%；WACC飙升至15%将导致企业整体估值收紧51.5%。然而风险高度分化：多肽创新药企业通过‘新加坡/东南亚拆分式BD’可挽回高达…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 4, "neutral": 0, "value": 51.6, "events": [ { "title": "虽然创新药与CXO估值处于历史底部且对云厂商Capex敏感度为0，但在地缘合规、三期临床高失败率及CRO订单传导时滞下，板块重估面临结构性压力，不宜盲目乐观。", "summary": "截至2026-06-08，本报告对AI主题资金轮动至生物科技资产（AI-DD与创新药）的逻辑进行了压力测试。测算显示，虽然该板块对云Capex零敏感，但极度易受全球折现率和地缘政治法案（BIOSECURE、BINSA）影响。rNPV模型显示，地缘合规导致的研发延迟与成本上升将压制估值约17.4%。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 年电力设备交付瓶颈无法被产能扩张化解，高昂的积压订单饱和度与供应链约束将导致 AI 基础设施在 2027 年面临严重的物理交付墙。", "summary": "压力测试显示，GE Vernova 与西门子能源的扩产计划无法化解 2027 年交付瓶颈。尽管两家公司正在美国和欧洲大幅扩产，但其产能早已被 2024-2025 年签署的创纪录积压订单（GEV $150B，西门子 €154B）预占。燃机和变压器的交付周期分别维持在 5-7 年和 4-5 年的历史高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高昂的GLP-1等医疗支出正对中产阶层产生“钱包挤占效应”，在宏观逆风之外构成了全价可选消费的第二重结构性暴击，并与CXO订单流的K型分化共同印证了滞胀期的资金极致抱团。", "summary": "截至2026-06-08，本卡片从生物医药视角对K型消费降级宏观逻辑进行压力测试并予以支持。GLP-1等重磅药物高昂的自付费用正实质性挤占消费者钱包份额，加速了全价可选零售的衰退；同时，全球CXO订单流呈现多肽/ADC爆雷而小分子萎缩的极致K型分化，印证了滞胀环境下资本与消费仅向结构性赢家集中的特…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在云厂商资本开支增速可能放缓的压力下，主题资金将向对Capex零敏感、以三期临床验证和BD联合开发为核心的生物科技（AI-DD与创新药）核心资产切换。", "summary": "截至2026-06-08，本报告探讨AI硬件估值对云厂商开支增速放缓的敏感性及主题向生物科技资产（AI-DD与创新药）的轮动。测算显示，AI硬件对Capex弹性达0.35x-1.0x，面临估值下修风险；而AI-DD资产对云厂商Capex敏感度为0.0x，底层估值由临床三期读出、BD授权交易和FDA…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 2, "neutral": 1, "value": 68, "events": [ { "title": "桶C应绝对聚焦于拥有4.5%以上股息率、国内OTC现金流及批零一体化壁垒的1099.HK、3320.HK与A股中药龙头，以此隔绝海外地缘风险并稳固组合底仓。", "summary": "截至2026-06-09，本卡对扩容至48.2%的桶C（防御医药）进行底层资产筛选。筛选标准要求股息率>4%、ROE>10%且对中美脱钩零敞口。结果显示，港股医药商业龙头（国药控股1099.HK、华润医药3320.HK，股息率约4.6%）与A股中药OTC龙头（云南白药000538.SZ股息率>5.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "穿透A股医药持仓并校验603259.SH同发行人拥挤度", "summary": "截至2026-06-09，本卡穿透A股医药账本，校验603259.SH同发行人拥挤度。关键事实：603259.SH与2359.HK是同一法律发行人药明康德的A+H上市，必须在发行人层面轧差。603259在沪深300医药卫生、中证创新药、中证医疗三只指数中均约7%[S1]，在窄基鹏华医药LOF/博时…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 基础设施的‘物理主权’已转移至掌握变压器核心组件（GOES、分接开关）与垂直整合能力的工业巨头，2026年的投产节点由‘变压器槽位’而非‘芯片配额’决定。", "summary": "2026年6月9日：本卡通过物理交期压力测试确认，电力变压器（交期3-5年）与开关柜（1.5-2年）已成为AI落地的‘刚性闸门’。工业巨头正从设备商转向‘产能地主’，Eaton通过‘从电网到芯片’的垂直整合与Boyd/Motivair的液冷并购锁定了高溢价份额；Schneider则凭借预制化模块（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 4, "neutral": 2, "value": 60.4, "events": [ { "title": "大健康的BD出海重估是结构性而非周期性，最干净的低覆盖alpha在多肽CDMO与落袋BD层而非龙头，但需先过治理与集采价格治理两道尾部风险关。", "summary": "大健康按激活率仅4.4%属低覆盖，但基本面处多年拐点：2026Q1创新药对外授权披露额>600亿美元（平均首付款1.84亿美元、同比+59%）、药明康德Q1在手订单597.7亿元、司美格鲁肽3月在华专利到期开启多肽周期。低覆盖优质标的集中在下一层：三生制药(1530.HK，辉瑞12.5亿美元首付款…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药板块需从‘长久期临床叙事’切换至‘已落袋美元现金/确定性产能’的低久期配置，超配现金落袋型 BD 公司以对冲 CXO 迁移与关税风险。", "summary": "针对 2026 年 6 月药明康德被列入 1260H 名单及 Section 232 关税生效，本研究将医药超配策略从“估值修复”转向“低久期防御”。核心策略是超配已落袋美元现金的 BD 龙头（权重 40%），配比需求刚性的多肽 CDMO（25%），而对受地缘风险直接冲击的常规 CXO 和高 Be…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "药明康德被列入1260H名单与100%关税悬崖标志着中美医药供应链进入“高成本合规时代”，CXO企业的全球产能重排将导致利润率结构性下移。", "summary": "截至2026年6月，中美医药供应链面临实质性断裂风险。药明康德于6月8日被列入1260H名单，触发BIOSECURE法案下的“关注企业”地位；同时Section 232关税将于7月底对专利药进口实施100%税率。CXO企业虽通过新加坡枢纽和美本土化基地（如药明特拉华工厂）应对，但面临15-25%的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药创新药/多肽CDMO是2026年低相关的第三成长轴、值得超配，但只有结构性底仓（美元现金流+在手订单+GLP-1放量）可配置，事件脉冲只能交易——这波美元首付款重估是有脉冲骑乘其…", "summary": "把医药创新药/多肽CDMO放入2026三轴市场框架：TMT/AI（成长）、红利/价值（防御）之外，医药是低相关的第三成长轴，收益驱动（美元首付款+CXO在手订单+GLP-1放量）与AI叙事、国内再通胀均正交。核心判断是拆解这波重估：约⅔是可配置的结构性底仓（落袋美元现金流如三生12.5亿美元、药明…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-11", "supports": 2, "risks": 3, "neutral": 0, "value": 77.8, "events": [ { "title": "地缘合规已迫使中资ADC/CGT创新药从“直接出海”转向“NewCo去中心化开发”，CXO订单转移已进入技术转移的实质执行期。", "summary": "截至2026年6月12日，受《生物安全法案》成法及药明康德被列入1260H清单影响，中资ADC/CGT企业正全面转向NewCo架构以规避合规风险，预计2026年超40%的海外临床由NewCo申办。同时，跨国药企订单正大规模向Samsung Biologics和Lonza转移，三星生物2025年订单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在大健康估值底部与药明系地缘风险升级背景下，应通过全球BD出海现金流与自主产能进行地缘政治对冲，重塑医药板块的杠铃配置。", "summary": "本报告启动对大健康产业链（激活率4.4%）的首次覆盖。我们提出哑铃配置策略：超配全球BD出海强、具独立大分子产能的创新药龙头（如恒瑞Q1创新药占比首破60%），回避或低配依赖药明系CDMO进行美国临床的中小Biotech，因其面临药明康德6月8日列入1260H后高达4000万至1.1亿美元的CMC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药不是AI基础设施约束的普遍避风港，只有能把算力转化为监管、临床、支付和采购成果的公司才值得溢价。", "summary": "截至2026-06-12，本卡综合支持前序“AI交易从订单叙事转向融资、电力与交付约束”的主线，但否定医药可作为宽泛避风港。Oracle的$638 billion RPO、negative $23.7 billion FY2026 FCF、4.5 GW Stargate容量和数据中心电力约束说明A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在NMPA新政红利与医保支付边际改善（DRG 2.0豁免）双重催化下，中国医药产业链正从“内卷式价格战”全面转向以ADC/CGT为代表的“全球化高价值输出”新周期。", "summary": "截至2026年6月11日，中国生物医药与健康产业链呈现创新加速与政策企稳的双重特征。NMPA新规落地，创新药出海达历史新高（一季度跨境授权达600亿美元）；CXO行业在跨国药企降本需求驱动下触底回升。医疗器械迎来“集采2.0”时代，唯低价论转向临床价值优先，同时随着DRG 2.0豁免政策推进，高端…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 1, "neutral": 0, "value": 75.2, "events": [ { "title": "截至2026-06-13，电力设备成本与交付风险已经实质性延后 AI 算力集群通电节奏，但它表现为 capex 效率恶化和项目重排，而不是最大型 hyperscalers 已经全面削…", "summary": "截至2026-06-13，本卡支持前序“电力约束正在惩罚 AI capex/融资假设”的方向，但精确化为：2026-2028 年已出现项目层面的通电延期与节奏重排，尚未出现头部 hyperscalers headline capex 的广泛下修。Sightline 显示 2026 年约 16 GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 3, "neutral": 0, "value": 64.6, "events": [ { "title": "BIOSECURE已落地为基线监管成本而非可被忽略的尾部风险——卡02证据型框架仍成立，但需从估值中扣除出海期权折扣，并按CXO/创新药/器械三档差异化重定价。", "summary": "截至2026-06-13，BIOSECURE法案已从尾部风险演变为基线监管成本：2024年众议院306:81通过H.R.8333，2025年通过NDAA-FY26与HHS拨款附加条款持续推进，原始文本点名BGI/MGI/华大智造/药明康德/药明生物，但授权HHS扩列。对卡02五个证据型标的差异化压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电气设备压力测试之后的 AI 基础设施瓶颈与行业配置", "summary": "本卡把前五卡综合为一条配置规则：当电气设备与液冷把瓶颈从芯片移向已通电算力，区分长期受益与拥挤资本开支受益股的判别变量是公司所占稀缺性租金的持续时间，而非收入增速。三道检验：瓶颈硬度（变压器交期80–210周最硬，液冷因OCP标准化租金衰减最快）、定价权体现在长久期在手订单与毛利（GE Verno…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国变压器出口增长与全球交期拉长相互匹配，但只有在提单重量、航线港口、卫星堆场出清三者同向时，才能把真实稀缺性租金与市场叙事区分开。", "summary": "2026-06-13：本卡对“中国电力设备出口商属于 AI 基础设施稀缺租金”做另类数据压力测试。结论是方向成立但必须防止叙事过热：2025 年中国变压器出口额超过 646 亿元、同比近 +36%，平均单价约 20.5 万元；同时大型变压器交期仍在 80-210 周，美国 LPT 常见 36 个月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 4, "neutral": 0, "value": 68, "events": [ { "title": "化债+设备更新双轮下，2026H2 A股医疗器械龙头相对医疗信息化具备6–9个月相对收益窗口，建议器械超配、信息化标配偏左侧、CXO作防御腿。", "summary": "截至2026-06-14，本卡片将card-02的现金流修复框架落到A股仓位：医疗器械（801153）PE-TTM约31x、35%分位，叠加超长期特别国债设备更新与化债回款修复，国产替代龙头（迈瑞、联影、新华医疗）2026H2具备12–15%板块级估值修复+5–10pp龙头alpha；医疗信息化P…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核心创新药管线数据读出的高胜率及海外BD的高估值弹性是支持板块估值重构的核心底座，而地缘政治风险正重塑全球CXO产能与资本流动格局。", "summary": "评估了映恩生物与和誉核心管线2026-2027年临床胜率（DB-1303乳腺癌80%、ABSK-011肝癌65%）。通过rNPV测算，其海外权益分别价值7.90亿和4.05亿美元，对两公司港股市值具备高弹性。近期《生物安全法案》及1260H名单事件加速了医药供应链向东南亚和欧洲转移，特种ADC C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "创新药出海加速与CXO分化复苏是2026年的主线，投资重点应向拥有国际临床催化剂的中小盘标的及能规避IRP定价溢出风险的企业倾斜。", "summary": "2026年中国医药产业链在“反内卷”与全球化进程中出现结构性机会。创新药Q1出海交易额超600亿美元。报告筛选了4家低覆盖度、高弹性的中小盘标的：映恩生物（ADC+IO2.0）、和誉（小分子靶向）、昭衍新药与近岸蛋白（CXO与上游底部复苏）。此外，强调了集采超低价导致的国际参考定价（IRP）溢出风…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施资本开支仍支持互连和设备链条，但投资确定性应优先放在PCIe/DDR5/MRDIMM/光互连，再看CXL和国产设备盈利兑现。", "summary": "截至2026-06-15，本卡支持上一卡关于AI服务器互连和国产设备的方向，但将确定性分层：Hyperscaler资本开支仍在上行，Microsoft、Alphabet、Meta、Amazon的最新披露共同验证AI基础设施支出强度；PCIe Retimer/Switch、DDR5/MRDIMM和光…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 6, "neutral": 0, "value": 49.4, "events": [ { "title": "最有研究价值的大健康低覆盖标的，是能用 2025 年收入、订单、医保准入或渠道数据验证拐点的公司，而不是泛泛补覆盖整个医药板块。", "summary": "2026-06-15 初始筛选支持提升大健康产业链覆盖度，但优先队列应集中在 2025 年已出现商业化拐点或明确风险的标的：WuXi XDC (2268.HK)、迪哲医药 (688192.SH)、Ascentage Pharma (6855.HK / AAPG)、MicroPort NeuroSc…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然人民币贬值光学利好CXO，但触发贬值的全球风险厌恶将冻结生科融资与CXO订单，且北向资金撤资将对医药核心白马股造成微观流动性挤兑。", "summary": "本报告将A股系统性流动性压力测试（Research note 05）映射至医药板块。结论认为，尽管USD/CNY贬值至7.30-7.42从账面上利好CXO企业的汇兑收益，但日元平仓引发的全球流动性紧缩将冻结海外生物科技融资，导致CXO订单流实质性萎缩。同时，600-900亿元的北向资金流出将重创外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "真实算电成本的刚性拉大了大小企业间的研发边际成本差距，迫使初创公司依赖昂贵的公共云，从而加速大健康板块的市占率向具备西部自建集群与湿实验闭环能力的头部CRDMO集中。", "summary": "本卡片评估了在真实算电成本约束（内蒙古等绿区电价0.35-0.45元/度及PUE 1.15）下，国内AIDD与CXO企业的算力布局。高利用率（80%以上）使头部CRDMO（如药明康德、晶泰科技等）自建西部集群比租用云端节省35.2%的成本；而利用率低于25%的初创企业被迫租用云端，承受94.6%的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "创新药商业化兑现面临DRG/DIP支付限制与存量竞争的双重挤压，地缘政治风险导致的估值折价具有持久性，医药板块的估值重构逻辑尚未成立。", "summary": "本报告对首席策略师的乐观预期提出深度挑战。核心论点在于：2026年DRG/DIP的全面覆盖已成为创新药进院后的‘硬天花板’，导致营收增长与利润释放严重脱节；GLP-1与ADC赛道的激烈内卷引发价格战，削弱了BD价值重估的基础；地缘政治博弈对CXO板块的持久性压制导致海外订单缩减。我们认为医药板块的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 3, "neutral": 0, "value": 68.8, "events": [ { "title": "DRG/DIP 2.0的豁免机制与BIOSECURE Act的5年过渡期为头部创新药与CXO提供了充足缓冲，双重压力不会阻碍具备核心差异化品种的行业龙头的估值修复。", "summary": "本报告对DRG/DIP与地缘政治法案对医药板块的约束进行压力测试。结论表明，双重压力构成“极化筛选器”而非绝对性障碍。高临床价值创新药（如玛仕度肽、Sac-TMT）可通过特例单议（通过率85.1%）和除外支付政策避开限额；药明康德在手订单（597.7亿元）及多肽业务显示地缘政治风险已过度折价，5年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler capex 足以穿越能源 TCO 上行并支撑稀缺供应链，但投资表达必须是“多稀缺、空低 ROIC”，而不是无差别做多 AI。", "summary": "2026-06-18 压力测试结论：hyperscaler capex 增长足以让 AI 基础设施供应链消化能源成本带来的边际 TCO 压力，但只能支持“多稀缺”，不能支持泛 AI beta。四大买方 2026 capex 粗略包络约 $710B（自测算），Microsoft、Alphabet、M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：…", "summary": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：AI 基础设施分析师 (ai infrastructure analyst) 立场：stress test（压力测试…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 11, "neutral": 1, "value": 25.2, "events": [ { "title": "宏观层面的BD首付款虽能覆盖CXO利润损失，但因主体割裂、合规折让与同质化挤兑，该‘安全垫’在微观投资组合中无法实现跨板块套利，CXO的系统性出清与创新药IP化转型将呈深水区分化。", "summary": "截至2026-06-19，本卡片通过压力测试评估创新药License-out（BD）首付款抵消BIOSECURE法案下CXO利润流失的行业可持续性。测算表明，即使在重度压力下（CXO美线流失60%且净利率降至8%，BD首付款面临60%地缘与靶点饱和双重折让），BD首付款净流入（193.2亿元）仍可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康存在选择性重估标的，但 AI 跨链外溢不应按篮子买入；基础设施先捕获稀缺性，医疗需等临床、监管、支付和现金转化验证。", "summary": "截至 2026-06-18，本卡从 AI 基础设施视角否定“大健康低覆盖=系统性低估”的宽泛判断。15/280 的 5.4% 激活率只是内部关注度信号；真正稀缺、已被资本密集定价的环节在算力、电力、数据中心与云 capex。医疗 AI 采用真实，FDA 列表已达 1,525 个条目，但监管、临床验…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力脱钩不会彻底摧毁中国AI药研，但会把其从全球平台服务叙事压缩为国内/RoW研发效率工具，并对前沿服务型CXO给予结构性估值折价。", "summary": "截至2026-06-18，本卡压力测试算力与训练生态脱钩是否会摧毁中国侧AI药研。结论是否定极端版本：美国bulk多组学、前沿基础模型服务出口和MNC项目会被数据、GPU与CUDA/BioNeMo生态共同压缩；但中国/RoW药物化学、国内患者数据研发和BD驱动创新药并未归零。Ascend集群、Py…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DoJ数据安全计划下，高昂的技术合规规避成本彻底侵蚀了中方CXO的工程师红利，迫使前沿多组学及AI发现业务在商业逻辑上走向“二元脱钩”。", "summary": "截至2026-06-18，本卡片对CXO规避DOJ数据安全计划（DSP）的技术手段进行了压力测试。测算显示，云端TEE、联邦学习和在美隔离实验室等手段虽然可行，但运营成本增加30%-50%，人力成本攀升3-5倍，并产生20%-40%的研发效率损耗。这彻底侵蚀了中国CXO的工程师红利。技术性规避在商…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 1, "risks": 2, "neutral": 0, "value": 85.6, "events": [ { "title": "医药内部切换正在发生于边际资金层面，最佳表达是超配 BD 能力强的创新药龙头、把 CDMO 降为估值修复仓位，而不是简单全面买医药。", "summary": "截至2026-06-19，本卡支持医药配置从受压制 CDMO 向具备 BD 能力的国内创新药龙头边际切换，但强调这不是已完成的全面撤离。2026Q1 公募持仓仍显示 CXO 获加仓、药明康德仍居医药重仓前列；近期交易层面则显示恒瑞医药主力资金转强、康龙化成等 CDMO 边际流出。配置上建议超配恒瑞…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "CXO产业链已发生订单结构和估值风险溢价变化，但截至2026-06-19尚不是海外订单总量坍塌，而是新增订单向合规分层、双源供应和国内新模态替代倾斜。", "summary": "截至2026-06-19，重点CXO的海外订单未出现总量坍塌，但已出现结构性分层：OMB/BIOSECURE压力使新增美国联邦资金暴露订单转向更严格的客户筛查、双源制造和“中国+一地”路由；汇率波动主要影响人民币报表折算和利润率，而非决定订单迁移。药明康德、药明生物、药明合联等2025/Q1 20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "出口管制升级轨道：BIS实体清单/FDPR对中国CXO/CDMO商业订单的脱钩概率与时间线", "summary": "压力测试卡片04的交接：出口管制升级是真实且无祖父期上限的尾部风险（BGI已被列入实体清单为先例，NSCEB 2025-04-08最终报告明确建议对华生物技术管制），但我下调其短路商业订单的概率、上调其真实渠道的烈度。半导体类比过度迁移：芯片/EDA/设备在CCL上、被管制物本身即商业产品，故FD…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 1, "risks": 4, "neutral": 0, "value": 75, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "大健康产业链 30 日激活率仅 5.4%（15/280），属被忽视而非基本面差。2026 年 Q1 创新药对外授权破 600 亿美元（约占 2025 全年 1357 亿的 44%）[自测算]，药明康德 Q1 营收 124.36 亿（+28.81%）、在手订单 597.7 亿、小分子 CDMO+80…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美元反弹与套保收益可平滑药明康德等 CXO 龙头利润，但无法单独修复美国 1260H/BIOSECURE 路径带来的地缘估值折价。", "summary": "工作日期 2026-06-21。结论支持把美元/人民币纳入 CXO 风险修复框架，但只能作为二阶利润缓冲。药明康德 2025 年非中国客户占持续经营收入 87.41%[自测算]，美元敞口真实；但其披露的 5% 美元兑人民币敏感性仅对应人民币 2.65297 亿元利润影响，约为 2025 年经调整 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘政治壁垒的推进伴随五年祖父过渡缓冲，虽延缓了CXO即时业务停摆，但临床数据跨境严控和地缘风险折价，已在重度脱钩情境下对资产组合造成-24.93%的系统性估值拉减，亟需通过货币和外…", "summary": "针对BIOSECURE Act落地及DOJ敏感数据限制，评估药明康德、诺泰生物、科伦博泰、康方生物组合。五年祖父条款减缓短期冲击，但重度脱钩情境下将导致加权估值-24.93%的系统性拉减。Q1出海BD虽超600亿美元但面临地缘折扣，建议转交外汇策略师评估美元结汇与离岸套利波动的风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 数据中心的下一道硬约束不是单纯电量，而是变压器、开关柜、并网与液冷系统共同决定的可交付 MW。", "summary": "截至2026-06-20，本卡支持上一张 AI 基础设施判断：AI 机架功率密度上升后，瓶颈从芯片供给下沉到可交付、受保护、可冷却的 MW。液冷设备具备较强模块化扩产弹性，但变压器、中压开关柜、并网和变电站设备仍是低弹性约束；美国配电变压器、大发电变压器和开关柜交期已显著拉长，拥有接入权、设备槽位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 0, "value": 80.4, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "大健康产业链低覆盖（30日激活5.4%），但基本面已反转。三引擎齐发：创新药license-out成全球定价事件（2025年约1357亿美元、占全球49%，2026Q1超600亿美元）；CXO订单重新加速（药明Q1营收+28.8%、在手订单597.7亿；凯莱英新兴业务+74.1%）；设备更新（20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备短缺会迫使CXO提前建设非美冗余产能，但真正的估值损伤来自客户留下后的低毛利双供应结构，而不是美国业务一次性清零。", "summary": "截至2026-06-22，本卡对Research note 8做压力测试：电力设备短缺会压缩药明生物等CDMO的合规窗口，但不会自动导致美国客户全部流失或印度快速接盘。更现实的路径是加速新加坡、爱尔兰等已在网络内且更易验证的非美产能，印度更适合作为合作或并购选项。以药明生物2025年收入人民币21…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 1, "value": 84.8, "events": [ { "title": "生命健康和生物科技的最佳增量研究方向是围绕中国创新药资产可全球定价、器械集采后放量能力、以及上游工具需求修复建立小而深的研究队列。", "summary": "截至 2026-06-22，本卡支持激活生命健康和生物科技产业链研究。优先深挖 Kelun-Biotech、Mabwell、Ascentage Pharma、Peijia Medical 和 Sino Biological，理由是它们分别对应 ADC/创新药 BD、后期临床与商业化、血液肿瘤产品销…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BIOSECURE Act 对中国创新药资产全球化估值逻辑的冲击评估", "summary": "BIOSECURE Act 于 2026-06-23 视角下已于 2025-12-18 随 FY2026 NDAA 签署生效，但其限制对象是联邦采购/拨款中涉及 BCC 的‘生物技术设备或服务’，不触及私营对外授权/BD 与成品药。Kelun-Biotech 经 Merck/MSD（大中华区以外授…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "模态条件化、CDMO 相关化的 CMC 折损在 A/H 医药组合配置与轮动中的应用", "summary": "截至 2026-06-23，本卡综合前序结论：A/H 医药组合不应简单放弃全球定价创新药，也不应完全躲入国内商业化。规则是以 CMC 调整后 rNPV 衡量全球定价生物科技，以 EV/normalized EBITDA 和订单质量衡量 CXO，以盈利韧性衡量国内商业化；当创新药调整后上行空间 ≥2…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 5, "neutral": 1, "value": 68.2, "events": [ { "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "未覆盖机会集中在从科学故事转向商业化、医保、全球注册或集采韧性的公司，而不是宽泛押注医疗板块 beta。", "summary": "2026-06-24：本卡支持为生命健康和生物科技链条建立聚焦研究队列。优先深挖 Keymed Biosciences 2162.HK、Ascentage Pharma 6855.HK/AAPG、Mabwell 688062.SH/02493.HK，并在器械端保留 Peijia Medical 9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BIOSECURE法案落地并不构成中国CXO的即时断崖风险，2032年截止的缓释期与印度的产能/原料药瓶颈决定了产业链重构将是漫长温和的过程，高壁垒非BCOC企业（如凯莱英、药明生物…", "summary": "美国《生物安全法案》（BIOSECURE Act）于2025年12月18日作为Section 851载入NDAA FY2026正式立法。药明康德于2026年6月8日被美国防部列入1260H清单，锁定了其BCOC地位，但5年祖父条款缓释期将持续至2032年12月。中国CXO的海外产能（如新加坡、欧洲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BIOSECURE法案重塑下大健康板块与A股成长股的估值体系重构与风格轮动研判", "summary": "本卡在策略/配置层综合卡01–04：地缘风险溢价对非BCOC标的已是基本沉没的一次性重定价（药明康德1260H列名为最后一只靴子，CXO估值已由约50倍压缩至12–15倍）。公募医药仓位2026Q1回升至8.87%（环比+0.90pct），药明康德重回前十大重仓，相对约10.5%中性基准仍低配约1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 2, "neutral": 2, "value": 90.6, "events": [ { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the stress test evaluation regarding the offset capability of independent …", "summary": "I have completed the stress test evaluation regarding the offset capability of independent in house CMC systems for Chinese innovative drug developer…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "药明康德涨停与中国CXO板块估值底部的重构", "summary": "药明康德涨停与中国CXO板块估值底部的重构 WuXi AppTec Limit Up: Rebuilding China's CXO Valuation Floor 药明康德触及涨停且连续单日实施1.0亿元回购，表明地缘法案恐慌逐步向8年豁免期的理性定价过渡。在科技两融出逃与申万医药净流入50.6…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI算力集群的电力保障能力应按已锁定的变压器与中压设备链条打分，而不是按土地、PPA或名义兆瓦打分。", "summary": "2026-06-25，本卡对上一张 AI 基础设施结论做压力测试并支持其主线：区域 AI 算力集群的真正瓶颈不是单一铜/GOES价格，而是主变、中压开关柜、断路器、配电变压器、施工队与费率审批构成的完整送电链。美国2025年电力变压器与配电变压器供给缺口分别为30%和10%；北弗吉尼亚等核心市场已…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 1, "risks": 4, "neutral": 0, "value": 82, "events": [ { "title": "大健康产业链处于‘被忽视定价’阶段，基本面反转已现，应围绕出海BD与高端国产替代定向深挖五只低覆盖中盘股，防范地缘立法与科创去杠杆等跨链外溢。", "summary": "本卡片对低覆盖的大健康产业链进行了深度分析。在30日激活率仅5.7%的被忽视背景下，基本面于2026上半年迎来License-out、CXO订单与设备更新三重引擎的拐点。定向推荐药明合联、荣昌生物、康诺亚、医渡科技、沛嘉医疗五只低覆盖中盘创新龙头，并系统梳理了BIOSECURE立法、美债高利率去杠…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-27工作日，BIOSECURE法案地缘压力测试证实，CXO的估值中枢压制主要来自低毛利重组运营与合规时间税，而非业务一次性清零。", "summary": "截至2026-06-27，地缘政治立法如BIOSECURE Act已步入实质监管的逐步淡出阶段。压力测试显示，海外建厂虽能保全订单，但将使项目毛利率收缩25-30pct，且面临电力设备瓶颈。在美业务流失60%的极端情景下，行业净利率收窄至8%，地缘去耦带来-24.93%的永久估值折价。WFE验证延…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "验证License-out交易与CXO订单回升对重点中盘股基本面改善的驱动力度", "summary": "本卡片验证卡片01点名的两条引擎。结论为支持但作分化修正：驱动力真实且已在2025年财报中显现，但五只标的并不同质。CXO订单这条最硬——药明合联2025收入+46.7%、毛利率+5.4pct、订单储备14.9亿美元(+50.3%)，母公司药明康德储备+37.2%；行业对外授权TDV创纪录约135…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 1, "neutral": 0, "value": 93.4, "events": [ { "title": "全球能源供应增量（特别是重启周期的核能与不稳定的可再生能源）及慢速的电网升级在2026-2027年内无法有效对冲 AI 基础设施的电网并网刚性瓶颈，该瓶颈将直接导致大云厂商面临 80…", "summary": "全球AI算力扩张正面临电网并网物理刚性瓶颈。全球电网升级周期长达4-10年，排队时间中位数达5年，且高密度变压器等关键主设备面临交付延迟，直接导致美国约7/12 GW在建产能延期，并将造成2026-27年全球超大云厂商约800亿-1100亿美元资本开支递延。核电与可再生能源虽有高成长，但受制于重启…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 5, "neutral": 0, "value": 70.8, "events": [ { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "生命健康与生物科技链 6.9% 激活率是覆盖缺口而非基本面缺口。2026 Q1 中国创新药对外授权约 600–614 亿美元（同比 +66–73%，约 98 笔），占全球金额 75%、笔数 20%，首付款中位数升至 1.84 亿美元；药明康德 Q1 在手订单 597.7 亿元（+23.6%）。支付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "去风险三期医药资产通过rNPV模型释放3.48亿美元的期权价值构成Biotech估值底，而生命科学工具在5.5x EV/Sales筑底、CXO在 overseas 利润收缩下于2.0…", "summary": "截至2026-06-28，本卡片基于BD估值方法构建了下一批并购标的与估值底线筛选框架。AbbVie 109亿美元现金收购Apogee与Merck KGaA 113亿美元收购Bio-Techne为医药生物确立了质量门控底座。我们量化厘定：1) 创新药三期去风险资产的rNPV floor：以12亿美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "三通道监管风险与实物关税叠加，确立了向‘纯出海授权、低受控敞口’标的切换的战术风格轮动信号。", "summary": "本报告评估了COINS、BIOSECURE与EAR三通道风险以及232关税对医药出海的叠加影响。由于纯IP授权（Out-licensing）完全豁免于关税与三大监管法案，而CXO与受控设备平台标的受物理壁垒制约，组合压力测试显示战术超配纯IP授权方（至60%）能提供年化+5.6%的超额收益并大幅降…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康低覆盖机会更像跨子行业alpha篮子，而不是单一医疗板块beta；优先研究能把2025-2026催化转化为收入桥的五家公司。", "summary": "截至2026-06-28，本卡支持开启大健康低覆盖研究队列，但不建议追宽泛板块beta。优先深挖Keymed Biosciences、Hybio Pharmaceutical、Sino Biological、MicroPort NeuroScientific和Allgens Medical，主线分…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-44cf9d91f6/", "networkHref": "/investor/research-network/#thesis-44cf9d91f6" }, { "id": "thesis-31f32af77b", "source": "tracked_thesis", "title": "中国云与国产算力栈的关键不是模型热度，而是训练/推理订单、服务器交付和生态锁定", "titleEn": "China cloud and local compute stacks depend less on model hype than training/inference orders, server delivery, and eco…", "lane": "中国云/国产算力栈", "laneEn": "China cloud and local compute stack", "laneHref": "/investor/lanes/china-cloud-and-local-compute-stack/", "investableFocus": "中国云/国产算力栈：9988.HK, 0700.HK, 9888.HK, 000977.SZ, 601138.SH, 688041.SH, 688256.SH, 688981.SH", "stockUniverse": [ { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "阿里云与模型生态", "roleEn": "阿里云与模型生态", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "腾讯云与应用生态", "roleEn": "腾讯云与应用生态", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9888.HK", "name": "百度集团", "market": "HK", "role": "百度云与模型", "roleEn": "百度云与模型", "bucket": "core", "href": "/investor/stocks/9888.HK/" }, { "ticker": "000977.SZ", "name": "浪潮信息", "market": "A-share", "role": "AI 服务器", "roleEn": "AI 服务器", "bucket": "core", "href": "/investor/stocks/000977.SZ/" }, { "ticker": "601138.SH", "name": "工业富联", "market": "A-share", "role": "服务器制造", "roleEn": "服务器制造", "bucket": "core", "href": "/investor/stocks/601138.SH/" }, { "ticker": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产算力芯片", "roleEn": "国产算力芯片", "bucket": "core", "href": "/investor/stocks/688041.SH/" }, { "ticker": "688256.SH", "name": "寒武纪", "market": "A-share", "role": "AI 芯片", "roleEn": "AI 芯片", "bucket": "core", "href": "/investor/stocks/688256.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "国产晶圆制造", "roleEn": "国产晶圆制造", "bucket": "core", "href": "/investor/stocks/688981.SH/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" } ], "stockTickers": [ "9988.HK", "0700.HK", "9888.HK", "000977.SZ", "601138.SH", "688041.SH", "688256.SH", "688981.SH", "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT" ], "coreView": "赛道：中国云/国产算力栈。候选观察标的：9988.HK, 0700.HK, 9888.HK, 000977.SZ, 601138.SH, 688041.SH, 688256.SH, 688981.SH。当前共有 40 条证据、3 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 76, "eventCount": 160, "firstSeen": "2026-04-27", "lastSeen": "2026-06-30", "supports": 15, "risks": 128, "neutral": 17, "analystCount": 14, "sourceTypeCounts": { "研究线程": 2, "研究笔记": 149, "证据更新": 1, "研究主题": 3, "分析协作": 5 }, "topTerms": [ "AI 基础设施", "HBM", "ASIC", "AI 芯片", "中芯国际", "寒武纪", "cloud", "local", "compute", "stacks", "depend", "less" ], "whatChanged": [ "2026-06-30: AI算力行情与电网瓶颈是否已成为同一个交易", "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。", "2026-06-29: 国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加速撤向水电红利及电网设备出海龙头，陡峭化了向高红利风格轮动的斜率，配置应重构为哑铃型…", "2026-06-29: AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "2026-06-28: AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。" ], "practical": { "sectorLane": "中国云/国产算力栈", "sectorLaneEn": "China cloud and local compute stack", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "阿里云与模型生态", "roleEn": "阿里云与模型生态", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "腾讯云与应用生态", "roleEn": "腾讯云与应用生态", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9888.HK", "name": "百度集团", "market": "HK", "role": "百度云与模型", "roleEn": "百度云与模型", "bucket": "core", "href": "/investor/stocks/9888.HK/" }, { "ticker": "000977.SZ", "name": "浪潮信息", "market": "A-share", "role": "AI 服务器", "roleEn": "AI 服务器", "bucket": "core", "href": "/investor/stocks/000977.SZ/" }, { "ticker": "601138.SH", "name": "工业富联", "market": "A-share", "role": "服务器制造", "roleEn": "服务器制造", "bucket": "core", "href": "/investor/stocks/601138.SH/" }, { "ticker": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产算力芯片", "roleEn": "国产算力芯片", "bucket": "core", "href": "/investor/stocks/688041.SH/" }, { "ticker": "688256.SH", "name": "寒武纪", "market": "A-share", "role": "AI 芯片", "roleEn": "AI 芯片", "bucket": "core", "href": "/investor/stocks/688256.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "国产晶圆制造", "roleEn": "国产晶圆制造", "bucket": "core", "href": "/investor/stocks/688981.SH/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" } ], "stockTickers": [ "9988.HK", "0700.HK", "9888.HK", "000977.SZ", "601138.SH", "688041.SH", "688256.SH", "688981.SH", "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT" ], "primaryStocks": [ { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "阿里云与模型生态", "roleEn": "阿里云与模型生态", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "腾讯云与应用生态", "roleEn": "腾讯云与应用生态", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9888.HK", "name": "百度集团", "market": "HK", "role": "百度云与模型", "roleEn": "百度云与模型", "bucket": "core", "href": "/investor/stocks/9888.HK/" }, { "ticker": "000977.SZ", "name": "浪潮信息", "market": "A-share", "role": "AI 服务器", "roleEn": "AI 服务器", "bucket": "core", "href": "/investor/stocks/000977.SZ/" }, { "ticker": "601138.SH", "name": "工业富联", "market": "A-share", "role": "服务器制造", "roleEn": "服务器制造", "bucket": "core", "href": "/investor/stocks/601138.SH/" }, { "ticker": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产算力芯片", "roleEn": "国产算力芯片", "bucket": "core", "href": "/investor/stocks/688041.SH/" } ], "instrumentTags": [ "9988.HK", "阿里巴巴", "0700.HK", "腾讯控股", "9888.HK", "百度集团", "000977.SZ", "浪潮信息", "601138.SH", "工业富联", "688041.SH", "海光信息" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 103 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 14, "recentEvents": 103, "supportRatio": 0.09, "riskRatio": 0.8, "conflictScore": 89.2, "evidenceDepth": 100 }, "score": 70.5 }, "events": [ { "date": "2026-06-30", "title": "AI算力行情与电网瓶颈是否已成为同一个交易", "summary": "AI算力行情与电网瓶颈是否已成为同一个交易 Are AI compute rallies and grid bottlenecks now one trade? 在寒武纪突破万亿市值、PJM 容量成本与变压器交期继续抬升的同一天，团队应如何同时定价 AI 芯片盈利弹性和电力并网硬约束？", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 16, "matchedTerms": [ "AI 芯片", "寒武纪", "compute" ], "href": "" }, { "date": "2026-06-29", "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck from compute to power, networkin…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 16.5, "matchedTerms": [ "AI 基础设施", "compute", "server", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性TAM上限，营收放缓促使估值压缩。相反，全球CPO光模块（中际…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "寒武纪" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.25, "matchedTerms": [ "HBM", "infrastructure", "capex" ], "href": "" }, { "date": "2026-06-28", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,600 亿）以经营现金流支撑，资金基础/地域/标的均不同于承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17, "matchedTerms": [ "AI 基础设施", "cloud", "infrastructure", "wars" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "summary": "截至 2026-06-28，本卡压力测试认为：电力交付瓶颈会迫使 hyperscaler 从单纯追求新增 MW 转向 MW 生产率和存量优化，但不会终结扩产。Microsoft、Alphabet、Meta 可见 2026 年 capex 区间合计约 4,950-5,250 亿美元（自测算），同时仍披露容量约束；IEA 显示数据中心 1-3 年可建成，而新电…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.5, "matchedTerms": [ "AI 芯片", "hype", "infrastructure", "capex" ], "href": "" }, { "date": "2026-06-28", "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.en.md (7867 bytes) — The repor…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 16.5, "matchedTerms": [ "cloud", "local", "inference", "infrastructure", "investment", "capex", "price" ], "href": "" }, { "date": "2026-06-28", "title": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "summary": "电力与电网容量成为稀缺投入后，AI 芯片优化目标从绝对算力转向每瓦性能（perf/W）与单机柜 token 经济性。NVIDIA 宣称 Blackwell 推理能耗较 Hopper 最高低 25×（FP4/NVFP4 驱动），但 B200 单卡功耗仍升至约 1,200 W（H200 为 700 W），机柜功率密度升至 120–142 kW 并规划 1 MW…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 15.75, "matchedTerms": [ "HBM", "AI 芯片" ], "href": "" }, { "date": "2026-06-28", "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心腿；半导体国产替代（¥2万亿AI建设计划+80%国产强制比例…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "寒武纪" ], "href": "" }, { "date": "2026-06-28", "title": "双轨AI网络可以落地，但硬件约束从电力单点转向电力、CoWoS/HBM、交换ASIC和光互联的同步供给，只有能跨堆栈预锁定产能的运营商能率先规模化。", "summary": "截至2026-06-28，本卡对Research note 05双轨网络做压力测试并给出有条件支持：CoWoS/HBM足以支撑头部训练集群，但2026-2027仍是分配瓶颈；硅光/CPO可降低核心fabric功耗和延迟，但都市推理边缘仍会混用可插拔800G/1.6T光模块与软件局部性。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "ASIC" ], "href": "" }, { "date": "2026-06-28", "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗ASIC。先进封装成为泄压阀，但TSMC已提示先进封装产能紧张…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "HBM", "ASIC" ], "href": "" }, { "date": "2026-06-28", "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计约6900-7200亿美元（自测算），NVIDIA FY20…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 15.25, "matchedTerms": [ "HBM", "infrastructure", "capex" ], "href": "" }, { "date": "2026-06-27", "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期：…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 / Executive Summary 截至 2026 年 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "AI 基础设施", "hype", "infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "在2026-06-27的权威工作日背景下，国产HBM良率骤降逼迫国内算力网络化，而全球封装与电力阻尼拉长了云厂商CapEx变现周期，这使得制程免疫且订单排至2028年的可插拔光模块出口龙头成…", "summary": "截至2026-06-27，本报告评估了HBM良率危机与先进封装瓶颈对算力集群技术演进的影响。国产HBM3封装良率降至2.66%，倒逼国内算力集群转向网络密集的Scale-out拓扑，推高了光模块需求弹性。全球云厂商2026年CapEx直逼7250亿美元，但受制于CoWoS晶圆先进封装冲突与吉瓦级电网接入瓶颈，面临6-12个月的算力交付与收入确认错配。CPO…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.25, "matchedTerms": [ "HBM", "infrastructure", "capex" ], "href": "" }, { "date": "2026-06-25", "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实触发重排，但方向是向支出倾斜，且买方自身点名的硬约束是电力与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20, "matchedTerms": [ "HBM", "ASIC", "hype", "infrastructure" ], "href": "" }, { "date": "2026-06-25", "title": "电网缓释后，AI集群上线风险不是消失，而是转移到CoWoS/HBM、液冷验收与网络芯片的串行同步，CPO下行收窄但不等于收入曲线完全恢复。", "summary": "截至2026-06-26，本卡压力测试Research note 7：电网侧缓释可降低CPO最坏收入下修，但不能消除AI集群的串行硬件瓶颈。关键约束从单一接电问题转向先进封装/HBM、液冷机柜验收、交换ASIC/网络集成的节奏匹配。CoWoS缺口可能由约20%降至2026年底约10%，但10%仍足以配给高端加速器；液冷从少数高密度机房跨越到100-120 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19, "matchedTerms": [ "HBM", "ASIC", "infrastructure" ], "href": "" } ], "timeline": [ { "date": "2026-04-27", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "AI 基础设施资本开支可持续性（2H26）：从 2026Q1 超大规模厂商财报与 Blackwell Ultra→Rubin 过渡读取的信号", "summary": "AI 基础设施资本开支可持续性（2H26）：从 2026Q1 超大规模厂商财报与 Blackwell Ultra→Rubin 过渡读取的信号 在 2026Q1 超大规模厂商财报披露后，全球 AI capex 是否仍在加速、进入平台期，还是出现 ROI gating 的早期迹象？这对 NVDA/AS…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 4, "neutral": 1, "value": 22.4, "events": [ { "title": "推理主导范式将 AI 加速器平台的经济价值分配从 FLOP 算力转向 HBM 容量/带宽与定制推理芯片，MU 的 HBM4 先发量产和 AVGO 的定制 XPU 业务在该范式下具有最…", "summary": "在 HBM bit 需求增速约为加速器台数增速 1.5-2.0x、推理占新增加速器小时 55-65% 的基准下，本卡给出以下排序：MU（第 1，HBM4 先发量产、份额增速最快、重估潜力最高）、AVGO（第 2，定制推理 ASIC $60-90B FY2027 目标市场加 AI 网络）、NVDA（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Q1 2026 财报后的云巨头 AI capex：需求验证还是瓶颈预警？", "summary": "本卡选择 tmt-analyst 评估 Q1 2026 云巨头 AI capex 周期。Microsoft、Alphabet、Amazon 和 Meta 均显示 AI/云收入加速，但 capex 和 PP&E 支出同步大幅上行。TSMC 的 61% HPC 收入占比、74% 先进制程占比和 66.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Inference Scaling Laws vs Training Scaling Laws", "summary": "Inference Scaling Laws vs Training Scaling Laws In the 2026 compute environment, has the contribution of increasing inference compute (test-time comp…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 硬件超级周期正撞上由‘电力墙’、地缘成本溢价和出口管制叠加构成的宏观天花板，2027 年 HBM bit 需求增速 1.5-2.0x 的乐观预测需下调至 1.35x 以反映供应…", "summary": "本卡片对 AI 半导体超级周期的宏观可持续性进行了压力测试。研究发现，尽管 2027 年超大规模云厂商资本支出可能突破 1 万亿美元，但面临 19GW 的电力缺口和 AI 变现鸿沟，这可能迫使行业进入“消化期”。地缘政治方面，台积电 35% 的风险贴现已常态化，且 CoWoS 产能 90% 集中在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "2026下半年AI资本开支拐点辩论：消化期还是再加速？", "summary": "TMT分析师判定，2026下半年AI资本开支并未进入消化期，而是进入结构轮动期：总量持平至上行，边际增量美元从训练优化的商用GPU集群转向推理fleet、超大规模云厂商自研芯片、以及电力与场地准备。空头叙事在需求侧（推理token volume、agentic渗透、多模态）证据不足，但在前沿训练边…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 0, "neutral": 1, "value": 57.2, "events": [ { "title": "2026年AI芯片竞争格局：NVIDIA Rubin对阵云巨头ASIC", "summary": "2026年AI芯片市场进入“混合计算”时代。NVIDIA推出3nm Rubin架构并引入NVLink Fusion以整合第三方芯片。与此同时，Google、AWS和Meta通过自研3nm ASIC显著降低了推理成本。AMD则凭借更大的HBM容量在长文本推理模型中占据优势。市场焦点已从单纯的算力（F…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯…", "summary": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯片竞争格局：NVIDIA Rubin对阵云巨头ASIC 本卡主题： WFE资本开支周期——先进封装与HBM设备订单可…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-13", "supports": 1, "risks": 2, "neutral": 0, "value": 43.6, "events": [ { "title": "A股半导体头部公司包括中芯国际 (688981)、寒武纪 (68．8256) 等。", "summary": "A股半导体头部公司包括中芯国际 (688981)、寒武纪 (68．8256) 等。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A 股与美股 Capex 分歧主要是结构性正确定价，应否定系统性超跌买入，仅对认证 Alpha 队列做选择性重估并参与美方 Capex 卖铲人。", "summary": "A 股清算产能 Beta、美股 Capex 仍在抬升，这一分歧主要是结构性的：出口管制楔形将先进制程需求改道至韩台与美方供应链，国内 AI 需求量级更小且招采议价，叠加反\"内卷\"政策对普惠扩产降温。因此\"全面抄底 A 股 TMT\"是错误的组合反应，正确解读是对\"认证 Alpha\"队列做选择性重估：…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "成熟制程产能 Beta 的价格清算已结构化，AI 先进封装是其物理隔离的反面，两向量合并强化而非削弱 Research note 02 的'做空产能 Beta、做多认证 Alpha'…", "summary": "截至 2026-05-13，从成熟制程产能扩张与 AI 先进封装两个相反向量进一步论证 Research note 02 的'做空产能 Beta、做多认证 Alpha'判断。向量 A：大陆成熟节点（28/40/55/90nm）在 2023–2026Q1 累计新增约 750–820 kwpm 产能，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-14", "supports": 1, "risks": 2, "neutral": 1, "value": 48, "events": [ { "title": "巨头 CapEx 压力确实在 A 股 AI 算力硬件的集成商层引发毛利率压缩，Agent 叙事无法救回集成商利润池；但加速器与光模块/PCB 子板块基本面仍在改善，Card-01 的…", "summary": "压力测试结果部分支持 Card-01，但需做结构性修正：A 股 AI 算力硬件并非铁板一块。服务器/集成商层（浪潮、紫光等）GM 已从 8–11% 压缩至 5–7%，HBM3e 涨价 + 巨头议价 + 液冷一次性投入共同施压；情景测算显示即便 Agent 推理乐观情景下集成商 GM 也仅能持平，因…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "材技一体化对fabless是选择性降权——AI加速器与EDA/IP溢价稳健，射频/功率/模拟/成熟节点向IDM收敛，资金应在设计层内向两个可防御极轮动。", "summary": "材技一体化情景下，fabless估值溢价并非整体崩塌而是结构性分化。AI加速器龙头（NVIDIA、AMD MI、Broadcom ASIC）凭借云厂商需求无弹性和CoWoS/HBM优先配额，能够将晶圆和封装涨价完整转嫁，溢价稳健；EDA/IP分成（Cadence、Synopsys、Arm）现金流与…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "国产算力进入性能驱动后，A股计算机与半导体板块全年风格将从小盘成长切向中大盘核心资产，配置优先级抬升设备—国产算力—先进制程—云与AI应用龙头主线。", "summary": "支持Research note 01的范式转移结论：DeepSeek V4在国产硬件上的性能跃迁意味着估值锚从政策预期切回业绩兑现，2026年A股计算机与半导体板块的风格将系统性地从'小盘成长/主题轮动'切向'中大盘核心资产/业绩兑现'。全年配置优先级：超配半导体设备(北方华创、中微)、国产AI算…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "算力期货压缩的是商用GPU层租金，先进封装与高带宽互联的物理瓶颈与寡头结构足以将其非商品化毛利结构维持至2027。", "summary": "支持card-01框架并加以收紧：算力期货标准化的是GPU-小时这一输出单位，并未商品化决定谁能以最低成本清算的物理输入。CoWoS先进封装、HBM4、混合键合设备、scale-up交换机ASIC、PCIe/CXL retimer、1.6T/CPO光模块均具18–36个月结构性导入周期，寡头格局清…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 1, "risks": 3, "neutral": 0, "value": 40.4, "events": [ { "title": "A股半导体重估的 beta 已基本被定价，下一段必须靠后道设备、量检、材料、车规功率与先进封装上游材料的盈利与订单兑现，回避换手/融资/解禁三重拥挤的设计、代工与映射封装。", "summary": "截至 2026-05-15，A股半导体经过 4-5 月重估后，国产 GPU/ASIC 概念、HBM/先进封装映射股、大市值晶圆代工四条线已拥挤：5 日换手率分位 81-92%，半导体两融占自由流通市值比 ~4%（88 分位），主要芯片 ETF 4 周连续净申购 +18%，且 6-8 月解禁集中（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "推理侧ASIC的规模化替代已导致通用GPU定价权削弱，NVIDIA面临物料成本上升与推理市场份额流失的双重压力，2026Q3毛利率下修逻辑成立。", "summary": "调研显示，2026年5月主要云服务商（CSP）的自研推理ASIC已进入3nm成熟期。Google TPU v6部署量超160万颗，Meta MTIA v3预计年底覆盖35%推理机群，微软Maia 200已支撑GPT-5.2。ASIC相比通用GPU可降低30%-50%的TCO，这导致推理定价权从NV…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "M2M流量爆发的边际成本不会均匀挤压SaaS——它会让薄壳代理编排厂商利润坍缩、让有自有数据的垂直SaaS继续吃到按量计费红利，并把被争夺的利润集中流向芯片、封装与光模块基础设施层。", "summary": "截至2026-05-16，本卡对card-07的SaaS基本面Beta判断做压力测试。结论是有条件成立但显著分化：薄壳代理编排SaaS将因无法转嫁的token成本，在未来12–18个月承受600–1,200 bp毛利压缩；而垂直SaaS、数据合规和API适配器（card-06超配名单）边际成本以存…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI需求是真实支撑，但不能完全抵消未来四周TMT/机器人解禁与拥挤交易的局部供给冲击，减仓约三分之一是合理的战术对冲。", "summary": "截至2026-05-15，本卡支持卡片02对高弹性TMT/机器人龙头提前减仓约三分之一，但理由是供给冲击和交易拥挤，而非AI基本面转弱。全球半导体、ASML、TSMC、NVIDIA、Microsoft、Amazon、Meta以及中芯国际、寒武纪数据均显示AI需求仍强；但未来四周约4120亿元解禁、…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 3, "neutral": 0, "value": 33.8, "events": [ { "title": "AI算力链基本面与订单流：对减仓建议的支持性验证 日期： 2026 05 17 分析师： TMT行业分析师 [Primary] 对Research note 01立场： Suppor…", "summary": "AI算力链基本面与订单流：对减仓建议的支持性验证 日期： 2026 05 17 分析师： TMT行业分析师 [Primary] 对Research note 01立场： Support（支持） 结论： 基本面边际开始出现裂缝——印证因子拥挤度去风险建议 1. 摘要 Research note 01…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "H200对华许可的毒丸条款使其对中国云厂只能扮演过渡角色，对跨境AI硬件营收偏空、对国产替代链（SMIC/昇腾/寒武纪）结构性偏多，板块定价应反映'强制分化'而非'复苏催化'。", "summary": "H200对华许可叠加25%关税与BIS领土审计构成'毒丸'结构：单芯片承诺、现场审计、芯片级遥测、再出口禁令使H200对中国云厂成为高摩擦、不可转售、用途受限的过渡载体而非战略基石。我们下修Nvidia 2026年中国数据中心营收至80-100亿美元（vs.一致预期120-140亿），毛利率拖累5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "最终方案是三腿再平衡而非单轴切换：设备+11、ASIC+9、边缘+6、纯GPU−10、IPP−13，并叠加SMH/NVDA与配对对冲及CEG/VST尾翼call。", "summary": "[2026-05-16] 综合前四张白板，AI电力约束由发电→设备→算力形态逐级下移，最终方案为三腿结构性再平衡：XLU/IPP减仓约30%（保留三分之一锚仓与CEG/VST尾翼call）；电力设备超配+11单位（GEV/ETN/VRT；A股特变、国电南瑞、思源、平高、英维克）；纯GPU去集中化−…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 41.2, "events": [ { "title": "海外 Hyperscaler FY26 资本开支稳固、卡片 01 触发器未触发，四只『千元股』基本面逻辑未破，绑定风险是情绪侧而非需求侧，但内部分化要求先减光模块、保留 ASIC。", "summary": "美股四大 Hyperscaler 1Q26 业绩（2026 年 4 月末／5 月初）合计 FY26 资本开支约 US$3610–3710 亿，同比 +28%，无一家下修，Meta 还小幅上调 —— 卡片 01 列出的『Hyperscaler 资本开支下修』触发器截至 2026-05-18 未触发。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "出口管制升级不推翻国产算力主线，但把风险从需求侧转移到供应兑现侧；大基金三期足以支撑 ASIC 与 WFE 自主叙事，却不足以消除 2026 年交付与验证波动。", "summary": "截至 2026-05-18，本卡对前序结论做压力测试：美国下一轮加严更可能落在设备、HBM、EDA、服务备件和实体清单，而非立即全面封禁成品 AI 加速卡。该风险对国产 ASIC 是需求正面、供应负面的双向冲击；对 WFE 是订单正面、验证与零部件风险上升。大基金三期 RMB 3440 亿元、六大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 5, "neutral": 1, "value": 35.6, "events": [ { "title": "订单流驱动的估值底面临“执行力”压力测试，需警惕寒武纪等HBM高度依赖标的的流动性陷阱，中芯国际底部更稳但缺乏弹性。", "summary": "截至2026-05-19，华盛顿关于限制存量设备维护及HBM出口许可的动向，正在将半导体龙头的政策风险从“限制扩张”转向“限制执行”。中芯国际因其产能垄断地位具备较强估值底，但存量维护成本上升压制良率与毛利；寒武纪面临HBM断供导致的巨额订单无法交付风险，估值底逻辑可能因供应链断裂而失效。散户融资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产供应链存在显著的产能与时间窗口错配（至少18-24个月），寒武纪等AI芯片龙头的订单面临无法交付的真实风险，基本面定价存在随时坍塌的脆弱性。", "summary": "截至2026-05-19，TMT分析确认国产HBM与先进封装在12-18个月内无法有效承接US供应链断裂风险。中芯国际2.5D产能已透支，长鑫存储HBM3量产存在18-24个月缺口。寒武纪等高度依赖HBM的AI芯片公司面临订单无法交付的断裂风险，单纯依赖订单流的估值底逻辑极其脆弱，极易触发流动性踩…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至 2026-05-19，AI 基础设施仍是 TMT 领涨主线，但可投资范围应收窄到有真实 usage、backlog、产能瓶颈和现金流承受力的环节。", "summary": "工作日期 2026-05-19。本文否定前序空白 TMT 论点并选择 (A)：AI 基础设施仍是领涨主线。证据是 hyperscaler capex/收入比虽已升至约 30% 的四家公司 Q1 合计强度，且 Alphabet、Amazon 的 FCF 张力明显，但 Microsoft AI ARR…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "消费电子'技术性弱复苏'+AI 手机溢价被压在旗舰段，触发的是 A 股'叙事溢价 → 确定性溢价'的哑铃化再定价，而非全面价值回归或单边 AI 基础设施集中。", "summary": "工作日期 2026-05-19。综合 card-03（AI 基础设施仍是窄口径主线）、card-04（消费电子 2026H2 技术性弱复苏）与 card-05（以旧换新净撬动仅约 1.25x、AI 手机溢价仅 ≥6,000 元旗舰段）三条结论后，本卡的策略级判断是：A 股既不会单边全面价值回归，也…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-20", "supports": 1, "risks": 1, "neutral": 0, "value": 58, "events": [ { "title": "A 股资金流与拥挤度指标支持选择性叠加而非板块级 半导体→SaaS 轮动，软件相对强势是美股叙事溢出与独立国内催化的叠加，量级不足以推翻板块层面的中性配置。", "summary": "本卡从 A 股市场结构角度对 Card-01、Card-02 的结论做 stress-test，结果同向支持：北向资金在软件侧仅选择性买入金山办公等少数标的，在半导体侧仍堆叠设备链而非全面撤离；两融余额边际加仓软件但拥挤度信号集中在寒武纪等单股；行业 ETF 申赎呈现软件净申购加速、半导体小幅放缓…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "支持哑铃右端方向，但纠正颗粒度——TMT内只有AI算力国产化+半导体设备/材料具备2026年内可验证的盈利弹性与估值修复空间，把整个科技板块当作进攻端是估值过度延展。", "summary": "2026-05-20，本卡片对卡片09哑铃策略中科技进攻端进行压力测试。结论是有条件支持，但需切分颗粒度：TMT内仅半导体设备/材料与AI算力国产化（昇腾、海光、寒武纪、HBM、先进封装）两条主线具备2026年内可验证的盈利弹性与估值修复空间，合计占TMT自由流通市值约18%-22%。消费电子、面…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 4, "neutral": 0, "value": 33.4, "events": [ { "title": "科创板1,750点保卫战实质是对半导体与AI核心资产现金流的定向救助，消费电子等非战略瓶颈板块将面临被动的流动性出清与估值大幅下修。", "summary": "截至2026-05-22，本报告综合得出结论：面对1,750点防线与880亿元抛压，国家队将采取“非对称防御”。中芯国际、寒武纪等兼具高权重、高两融且面临现金缺口的核心标的将被优先注入流动性；而现金流较好、权重分散的消费电子与通用软件将被战略性放任估值下修，科创板风格向“国家兜底”因子剧烈切换。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-22的核心判断是：科创50和AI硬件链的风险不在主题本身，而在融资盘拥挤、折算率弹性不足和成交放大回落叠加后触发的被动减仓。", "summary": "截至2026-05-22，本卡支持并细化card-01的压力测试结论：被动去杠杆风险集中在融资余额大、融资余额/市值占比高、近期融资买入重且已遭遇成交放大回落的AI硬件标的。最高风险名单包括300476胜宏科技、300308中际旭创、300502新易盛、300394天孚通信、688256寒武纪、6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "融资冻结不会击穿所有科创半导体与AI公司，但会把压力集中到晶圆厂资本开支和AI芯片/IP研发，迫使国产替代路线从全面扩张转向瓶颈优先。", "summary": "截至2026-05-22，本卡支持上一卡“融资冰封会压缩硬科技现金流”的方向，但将风险重新定价为结构性而非均匀性冲击。10家科创板半导体与AI压力篮子的一年期缺口约542亿元，其中可直接追溯到已披露定增的只有57.9亿元，集中在寒武纪和芯原股份；更大的484亿元来自中芯国际、华虹公司等制造环节的资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在英伟达 Rubin 平台面临供应链瓶颈的窗口期，国产 GPU 凭借先进封装与 HBM3 的全产业链突破，正通过‘十五五’初期智算中心的大规模采购实现对推理市场的实质性支配。", "summary": "截至2026-05-22，本研究确认国产GPU已进入性能与市场双重拐点。NVIDIA Rubin平台受困于HBM4验证延迟与CoWoS-L稳定性，2026年产量目标下调25%。与此同时，海光深算4号与寒武纪思元690通过Chiplet技术实现性能跃升，结合长电、通富在2.5D封装的突破以及长鑫存储…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "针对 2026 年 5 月 22 日全球半导体周期的历史巅峰，我们对“受控成长”框架下 35% 的科技配置进行了压力测试。研究表明，AI 硬件已从早期的“稀缺驱动”转向以推理为核心的…", "summary": "针对 2026 年 5 月 22 日全球半导体周期的历史巅峰，我们对“受控成长”框架下 35% 的科技配置进行了压力测试。研究表明，AI 硬件已从早期的“稀缺驱动”转向以推理为核心的“结构性基建”阶段，资本支出的公用事业化特征为高仓位提供了支撑。通过将配置重心从通用存储等周期性板块转向定制化 AS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "'效率降级'对 2026–2027 AI Capex 总量近乎无约束，但通过电力、关税、航运三条渠道侵蚀单位经济学，使全球 AI 基础设施从'同步扩张'转为'地理极化'，估值真正承压…", "summary": "本卡对 Research note 1–5 的“效率降级削弱 AI 算力扩张”假设进行双向压力测试。结论：方向正确但程度被高估，AI 算力 Capex 在 2026–2027 名义增速仍 +19–26% YoY（北美 hyperscaler 2026 指引 4,400–4,650 亿美元 [S1–…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 2, "neutral": 0, "value": 49.6, "events": [ { "title": "DeepSeek V4 在推理边际把瓶颈从显存容量切换到 kernel-稀疏-调度协同，'去 CUDA 化'与国产芯片重估在机制上成立，但价格已透支——昇腾量驱动产业链的风险收益优于…", "summary": "DeepSeek V4（2026-04-24 发布）把 1M 上下文下单 token 推理 FLOPs 压到 V3.2 的 27%、KV 缓存压到 10%（V4-Flash 进一步至 10%/7%），通过 CSA+HCA 混合注意力把推理瓶颈从'显存容量'切换为'kernel/稀疏选择/调度协同优…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体板块在高利率与中美并购友好政策共振下，正在以更快节奏复现生科板块的哑铃式价值重估——整合者赢家通吃、后期去风险标的被并购溢价重新锚定、中间产能被同步出清。", "summary": "截至2026-05-24，本卡片自下而上验证半导体板块正在高利率（UST10Y 4.56%）背景下复现生科哑铃逻辑。证据集中在三层：(1) 并购重组——Synopsys-Ansys ($35B)、AMD-ZT Systems ($4.9B)、HPE-Juniper ($14B)、Silver La…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 6, "neutral": 1, "value": 30, "events": [ { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？", "summary": "DeepSeek-V4-Pro 75% 降价 × NVDA FY27Q1 Data Center +92% 的联合定价：中国 AI 算力服务运营层是否独立走弱？ DeepSeek-V4-Pro 75% price cut × NVDA FY27Q1 Data Center +92% joint p…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] AI 基础设施散户拥挤是否耗尽 Thesis v8 入场窗口", "summary": "致 市场情绪师： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: b790980f-74a8-48f2-8307-cf1e60b887ee 结果摘要: Wrote handoff_response.md (14984 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件的高景气基本面虽能构成长期估值底部，但在科创50指数跌破1,780点引发的短期杠杆踩踏与量化被动去杠杆中完全无法抵御资金面崩溃。", "summary": "本卡对半导体龙头的抗跌性进行压力测试。AI硬件基本面景气度极其真实，四大龙头（中芯国际、海光信息、寒武纪、北方华创）2026年合计预测净利润达202亿元，季均50.5亿元。但两融与量化半导体高位拥挤风险严重。若科创50跌破1,780点敏感水位引发10%-15%回调，短期两融平仓与量化去杠杆抛压将达…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "区域化认证时滞与多供应商材料栈分裂打破了先进封装物理结构上的“黄金匹配”，其引发的热力学翘曲与 12-18个月的合规验证使得 AI 芯片先进制程量产风险实质性飙升。", "summary": "截至2026-05-27，本卡通过压力测试确认：区域化合规与材料栈重构不仅仅是法务摩擦，更打破了原后台积电生态的热力学一致性。由于不同区域、多供应商（SK Hynix/Samsung/Micron）的材料存在微小物理差异，使得 CoWoS-L 和 HBM4 在超过 1000W 功耗下遭遇极端的 C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 0, "value": 53.8, "events": [ { "title": "AI硬件周期仍由需求驱动，但未来12个月的定价权将更多落在HBM、先进封装、光模块和电力交付等瓶颈环节，而不是单纯GPU或服务器组装收入。", "summary": "截至2026-05-30，本卡支持“AI算力瓶颈从硅端扩展至物理基础设施”的主线。未来12个月，AI芯片和服务器需求仍由超大云厂商资本开支、云AI收入、agentic AI推理流量和自研ASIC扩张驱动；但供应瓶颈已从GPU单点扩展到HBM、CoWoS/先进封装、800G/1.6T光模块、机架供电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50极端的半导体权重（~70%）导致Gamma踩踏将直接演变为产业资金链危机：亏损型设计龙头（寒武纪）面临股权质押爆仓，而最上游设备商（中微、拓荆）面临高达20%-24%的现金流…", "summary": "本卡评估了科创50流动性驱动型Gamma踩踏对AI硬件与半导体的传导效应。在指数暴跌-25.0%的极端情景下，高Beta标的面临-30.0%至-41.25%回撤，寒武纪P/S压缩至24.68x且股权质押跌破150.0%警戒线，面临1.5亿元补足缺口。供应链账期延长45.0%触发三角债危机，中芯国际…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 1, "neutral": 0, "value": 69.2, "events": [ { "title": "结构性货币工具与产业基金可以为战略合格半导体企业提供经营性流动性底线，但不是科创50股价、股权质押或低质量信用的无条件救助。", "summary": "2026-05-30：本卡支持前序“科创50流动性踩踏可冲击半导体资金链”的判断，但将终局风险下修。央行科技创新和技术改造再贷款总额度已达8000亿元，实际贷款余额已有6140亿元，大基金三期注册资本3440亿元；按前序模型压力桶测算，覆盖中微、拓荆、寒武纪和中芯国际的102.73亿元现金压力仅需…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体上行周期足以驱动先进制造板块的盈利修复，但仅限AI算力/先进制程/HBM/光器件/国产设备切片；手机、面板、通用成熟模拟暂未跟随。", "summary": "截至2026-05-30，全球半导体上行周期已被多源硬数据证实，足以支持第03卡对A股电子/通信/先进制造的超配信号，但盈利修复是分层的、AI驱动的。SIA 2026Q1全球半导体销售2,985亿美元(环比+25%)，TSMC 2026Q1营收+40.6% YoY、HPC占61%、毛利率66.2%…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 1, "value": 65.6, "events": [ { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球电力与电网在 2026-2027 仍可支撑 AI 算力密度扩张，但在 2027-2028 会成为 HBM/ASIC 产能后最先 binding 的瓶颈，核电、CCGT 与变压器/…", "summary": "支持 card-01 的方向性扩散判断，并补充：HBM/ASIC 产能 12-18 个月可交付，但美国新增可调度电源 (CCGT+核电延寿) 有效投放需 36-60 个月，大型变压器交付 24-48 个月，并网队列已达 2.6 TW、平均等待 5 年，PJM 2025/2026 容量电价飙至 $2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 2, "neutral": 0, "value": 58, "events": [ { "title": "AI 硬件不共享软件端的 SBC 稀释墙，但 capex 弹性 + HBM 成本 + ASIC 替代三重传导意味着 NVIDIA 与 AI 供应链的利润率扩张预期同样需要下修。", "summary": "2026-06-01，本卡对前序首席策略师结论作压力测试：AI 硬件端面对的不是同一道 SBC 稀释墙——NVIDIA 季度 SBC 同比 +30.8% 至 $1.928B 但会计稀释仅约 0.43%，显著低于 Amazon/Alphabet/Meta 一档——而是更陡的需求弹性墙。真正的压力源是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件周期是真实的，但截至2026-06-02还不足以在流动性真空中触发A股TMT指数级回补。", "summary": "截至2026-06-02，本卡压力测试认为AI硬件和半导体基本面确实强劲，但不能替代A股三日流动性确认。NVIDIA、TSMC、ASML、SK hynix、Samsung等全球链条数据确认AI基础设施周期仍强；寒武纪和中芯国际也显示国内订单和产能利用率改善。但2026-06-01的A股盘面已显示半…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "未来24-36个月A股AI算力产业链Alpha按'液冷>AI应用>HBM材料>HVDC>瓶颈算力>能源型IDC'六段式排序，利润弹性与国产替代节奏决定排名，特高压作为防御底仓不入前六。", "summary": "以2026-06-03为基准日，综合Research note 1-9关于AI算力产业链利润再分配与毛利弹性的研判，给出24-36个月（2026 H2-2029 H1）的六个细分超配序列：#1液冷整机/CDU（ASP重估3-5×，毛利率22-28%升至30-38%）> #2 AI应用SaaS（代码…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [ { "title": "AI基础设施是结构上有支撑、战术上脆弱的挤仓，未来2–5日应在结构上回答（削减战术、保留核心、对冲凸性），而不是在方向上回答（既不押多线、也不押空尾）。", "summary": "AI基建当前同时受两个反向时钟驱动：6–18个月的结构性capex时钟向上（HBM偏紧、定制ASIC放量、超大规模云capex指引），2–5日的战术性持仓时钟向下（对冲基金净多头分位极端、QQQ被动流入+低VIX形成长gamma稀薄空气、广度窄、催化剂簇密集）。锚定2026-06-03（周三），未…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 1, "neutral": 0, "value": 76.2, "events": [ { "title": "Chiplet 与 3D 先进封装可在 18–24 个月内把国产 AI 芯片单卡等效算力推回旗舰 60–75% 区间，是先进制程受限下唯一可兑现的非对称工程路径，但无法消除 HBM …", "summary": "2026-06-04：Chiplet + 2.5D/3D 先进封装已成为国产高性能 AI 芯片在 7nm/N+2 被锁条件下的唯一可量产非对称路径。长电 XDFOI、通富微电、华天科技、盛合晶微在 2.5D CoWoS-like 与 9μm 混合键合上小批量验证；华为昇腾 920、寒武纪 MLU5…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "在中美算力双轨制下，通过超配服务器互连带宽、国产定制ASIC及边缘端AI节点，实现针对先进制程封锁的非对称反击策略。", "summary": "2026年6月4日：在中美算力‘双轨制’确立的背景下，A股核心科技策略应转向‘算力密度’与‘非对称替代’。大基金三期（3440亿元）作为风险吸收垫，正加速服务器互连（澜起科技、中际旭创）、AI ASIC/IP（芯原股份、寒武纪）及边缘端AI（瑞芯微、乐鑫科技）的板块配置。由于‘总部原则’限制了先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 1, "risks": 4, "neutral": 2, "value": 65.6, "events": [ { "title": "国产替代龙头融资买入拥挤度逼近活跃自由流通盘10%的极值，在大基金常态化减持和资产重组股权稀释压力下，当前的‘地缘溢价’流动性结构极为脆弱，投资组合应向重资产代工龙头防御性收缩。", "summary": "截至2026年6月5日，A股融资余额达2.89万亿元历史高位，两融资金在国产替代龙头中呈现极端的结构性拥挤。寒武纪与中芯国际的融资余额分别达251.33亿和132.19亿元，占其活跃浮筹比例超过7.4%和10%。国家大基金开启中芯国际H股常态化减持以回收资金，中芯国际定向发行5.47亿股购买中芯北…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "博通的估值回归并非基本面崩溃，而是从“预期完美”向“竞争性执行”的结构性切换，Google 的多元化供应是 AI 叙事的首道裂缝，但 OpenAI 和 Meta 的增量支撑了其底部逻…", "summary": "博通 Q2 财报后股价暴跌 15%，主因是 Q3 AI 收入指引（160 亿美元）不及市场极高的“私下预期”（172 亿美元），且 Google 确认将分散 ASIC 供应链。尽管全年 560 亿美元目标维持不变，但“唯一供应商”溢价的消退引发担忧。散户提及量激增 230% 试图抄底，反映出市场对…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "“地缘政治折价”在TMT并不普遍——它压低境外依赖型与赴美上市资产，却为国产替代龙头创造出受保护市场的溢价，正负号取决于公司处在国产替代分界线的哪一侧。", "summary": "对生物医药BD“地缘政治折价”是否普遍适用于TMT的压力测试结论：部分否定。美方出口管制确实抬高合规成本（25%关税、15%营收分成、第三方测试）、迫使供应链重构，并压低依赖境外技术与赴美上市公司（设备链、BABA/PDD/JD等ADR）的估值。但其传导机制与《生物安全法案》不对称——半导体/AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产替代设计端估值严重透支，在境外竞争有条件回流的压力下将面临大幅被动收缩，应向具有重资产物理壁垒的晶圆制造与红利资产防守端进行Barbell式再平衡。", "summary": "本报告对A股国产替代龙头的“地缘政治溢价”进行了压力测试。寒武纪（LTM P/S 104.7x，P/E 318.7x）与中芯国际（LTM P/S 10.8x，P/E 149.8x）估值处于历史峰值，主要基于封闭市场的极端假设。若英伟达H20/H200在下半年以许可证加征关税形式合规流入，商业互联网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "AI 芯片能效革命：应对电力瓶颈的架构演进", "summary": "前半句确认、后半句证伪。电力稀缺已使 Perf/Watt 成为约束性设计目标——Rubin 平台每兆瓦推理吞吐约 10 倍于 Blackwell [S1]、Rubin NVL144 3.6 exaFLOPS FP4 约 120–130 kW/机柜 [S2]、TPU v7 的 FP8 能效 5.42…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 1, "value": 75.4, "events": [ { "title": "[auto-handoff] AI ROI Cliff Analysis", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 148687b3-2d5d-4316-97be-bde7b4a6e646 结果摘要: Wrote handoff_response.md (8262 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have completed the synthesis on the resilience and architectural evolution of the domesti…", "summary": "I have completed the synthesis on the resilience and architectural evolution of the domestic AI computing ecosystem under advanced packaging supply c…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 2, "neutral": 0, "value": 75.8, "events": [ { "title": "AI半导体利润池迁移真实但渐进，板块总EPS上行、估值锚从'NVDA-TSMC双寡头溢价'切向'增长+多元化第二供应源'，ASIC设计商、HBM、先进封装设备与Intel Found…", "summary": "2026-06-09综合：Google-Intel TPU订单确认AI硬件利润池迁移真实且渐进，2027-28年AI半导体合计EPS较冲击前共识更高，但分布坚定向ASIC设计商(AVGO/MRVL)、HBM、先进封装设备及Intel Foundry期权倾斜。新估值锚为'增长+多元化第二供应源'，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球科技配置应超配ASIC与新代工生态；A股风格则需从“AI题材映射”战略性切换至独立算力的“国产第二供应源”，重仓先进封装产能、核心设备及HBM材料。", "summary": "2026-06-09评估：基于AI半导体估值锚向“增长+第二供应源”切换，全球配置需减配纯GPU倍数扩张与单一代工依赖，转而超配ASIC双雄(AVGO/MRVL)、HBM及Intel代工期权。在A股与港股层面，风格将从“英伟达映射”(光模块/代工)实质性轮动至业绩兑现度高的国产“第二供应源”。策略…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI硬件轮动战略正确但面临Intel 18A良率风险与资金‘业绩陷阱’滞后，需警惕地缘政治对先进封装链条的估值压制。", "summary": "2026-06-09压力测试确认：AI硬件利润池向ASIC与Intel代工的迁移在基本面上已获博通业绩及Intel 18A量产验证，但A股资金流向受1.6T光模块近端业绩吸引而表现出显著的‘轮动滞后’。Intel 18A工艺目前50%的良率是核心风险点，若无法如期突破将导致第二供应源叙事崩塌。A股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 0, "value": 75.2, "events": [ { "title": "超配拥有确定性全球订单的光电液冷龙头与昇腾-中芯先进封装/HBM自主垂直栈，回避被截断离岸通道的二线芯片设计与空壳算力租赁标的。", "summary": "截至2026-06-10，本研究深入A股TMT板块。建议超配：(1)挂钩中美头部云、订单已确认的光模块（中际旭创、新易盛、天孚通信）、液冷（英维克）与电源龙头（麦格米特），均处于1.6T/Blackwell订单放量期；(2)华为昇腾与中芯国际（SMIC）挂钩的HBM/先进封装国产替代垂直栈，长电科…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 3, "neutral": 0, "value": 64.6, "events": [ { "title": "电力交付滞后会迫使 Hyperscalers 调整 GPU 集群投产节奏，并通过更长现金回收期、更低早期利用率和更高电力保障成本压低 AI 基础设施 capex 效率。", "summary": "截至2026-06-11，本卡支持上一卡：AI capex 不会因电网瓶颈消失，但 GPU 集群部署节奏会被“已通电 MW”重新排序。变压器、并网和输电延迟会把部分已宣布园区容量推迟到后续窗口，压低早期自由现金流和资产利用率。Hyperscalers 会通过已通电站点溢价、分阶段投产、区域重排和自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源融资成本上升会削弱云客户ROIC并压低部分ASP，但最强AI硬件平台商仍有足够利润率缓冲，真正脆弱点在重capex制造和周期性内存环节。", "summary": "2026-06-11，本卡压力测试能源债务融资成本上升对AI硬件供应商利润率的影响。结论是方向成立但不应外推为硬件利润率全面坍塌：新增能源债务成本上行约85 bps会压缩云厂商自由现金流并强化采购纪律，但NVIDIA和Broadcom因高毛利、平台锁定和轻资产结构韧性最强；AMD居中；TSMC和M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 3, "neutral": 1, "value": 69, "events": [ { "title": "截至2026-06-13，电力设备成本与交付风险已经实质性延后 AI 算力集群通电节奏，但它表现为 capex 效率恶化和项目重排，而不是最大型 hyperscalers 已经全面削…", "summary": "截至2026-06-13，本卡支持前序“电力约束正在惩罚 AI capex/融资假设”的方向，但精确化为：2026-2028 年已出现项目层面的通电延期与节奏重排，尚未出现头部 hyperscalers headline capex 的广泛下修。Sightline 显示 2026 年约 16 GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "三因子个股筛选证实回避名单由6-9月解禁墙叠加零折算率驱动，可保留5%试仓的是仍有0.65-0.70折算率且无近端解禁的光模块/芯片龙头。", "summary": "把科创板/创业板主要科技股按三因子（融资余额占流通、6-9月解禁占流通、最新折算率）分组。回避名单由解禁墙叠加零折算率驱动：屹唐股份688729解禁约为流通盘700%且两融占比8.17%最拥挤；摩尔线程-U 688795解禁占流通95.0%且折算率0.00；沐曦股份-U 688802占78.1%、…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "先进封装/HBM仍是真瓶颈，但2026-06-12的上市交易已经把2027-2028年认证、良率和产能利用率提前资本化，最需要防的是技术代际错配与融资拥挤触发的估值和盈利预期双杀。", "summary": "截至2026-06-12，本卡对先进封装与HBM链条做压力测试：风险真实存在，但不是2026年立即供过于求。HBM到2027年前仍偏紧，核心风险在于HBM4/HBM4E和更大CoWoS形态迭代快于国内上市代理资产认证速度；叠加长电、通富、寒武纪等高估值和高融资余额，股价可能先于EPS下修而去估值化…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力刚性约束已把AI加速器竞争维度切换为每瓦Token，Chiplet+先进封装是2027-2028电网窗口内唯一可规模化兑现的算力密度救援方案，对NPV的杠杆是5-8倍。", "summary": "电网刚性约束把AI加速器的KPI从峰值FLOPS切换为每瓦Token吞吐。Blackwell B200单卡1000W、GB200 NVL72机柜120kW已触顶，下一代Rubin/MI400无法继续抬TDP，必须靠能效换性能。TPU v5p、Trainium2较H100能效提升30-40%；同10…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 5, "neutral": 3, "value": 72.4, "events": [ { "title": "训推分流把单一 HGX 订单链拆成‘西部互联带宽’+‘东部能效池化’两条独立赛道，延迟与带宽矛盾通过 Rack 内低延迟 + Pod 内大带宽 + 广域异步搬运的三层解耦在硬件层平衡。", "summary": "支持卡 07 的“训推分流”判断，并主张延迟/带宽矛盾可通过三层解耦解决：东部 Rack 域内低延迟（CXL+UAlink）、西部 Pod 域内大带宽（800G/1.6T+液冷）、东西广域承担异步 checkpoint 搬运。两套硬件 BOM 完全分叉——西部按互联带宽+PUE 定价（光模块、PC…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "红皇后竞速并未彻底抹去自研ASIC的成本优势，而是将其集中至Google与AWS两家、使其他云厂商回到对NVIDIA的依赖，从而部分推翻了Research note 05所述的全行业…", "summary": "在'红皇后竞速'下，自研ASIC单世代有效寿命已压缩至12–15个月，2nm节点全口径NRE攀升至USD 0.9–1.4B、N2晶圆约USD 30,000/片、CoWoS-L售罄至2027。我们测算自研ASIC在TCO上跑赢商用GPU的年部署门槛从5nm节点的约20万颗升至现实情景下的15万–25…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "半导体板块的主风险不是调样本身，而是核心权重与融资余额高度重叠后，在再度下探1600点时把指数级强平转化为芯片股集中抛压。", "summary": "截至2026-06-13，本卡压力测试上一卡片：科创50强平共振成立，但半导体抛压高度集中在高权重与高融资余额重叠的核心股。科创50前九只半导体/设备/芯片设计权重合计60.39%，若按权重承接744.3亿元级联强平，对应潜在抛压约449.5亿元；1516-1540点位20亿元国家队前线仅覆盖4.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "现金市场验证的是“尾部风险真实但需折价”，而不是把31.5%期权代理概率直接等同于强平级联的物理概率。", "summary": "截至2026-06-14，本卡综合判断：现金市场支持科创50左尾风险真实存在，但不足以完整验证期权市场约31.5%的1600破位代理概率。科创板融资余额截至2026-06-11仍有3757.65亿元，寒武纪、澜起科技、中芯国际合计融资余额523.16亿元，说明强平通道有现金基础；但588000最新…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 5, "neutral": 0, "value": 56.8, "events": [ { "title": "在AI链内部按融资质量轮动：超配Tier-1平台与卖铲人(设备/HBM/封装/电力赢家)，低配靠GPU抵押融资循环的Neocloud股权与高收益债。", "summary": "综合前八卡推演，AI基础设施的约束已从硅到电力、再到融资可得性。本卡给出按融资质量过滤的五篮子相对配置框架：B1 Tier-1 AI平台(+300bp)、B2 设备/HBM/先进封装(+250bp)、B3 ASIC定制(+100bp)、B4 电力瓶颈赢家(+150bp)、B5 低质量AI基建(-2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GPU抵押贷款收缩会削弱边际Neocloud订单可见度，但先进封装与先进制程需求更可能从融资套利订单转向信用更硬的超大规模厂商和ASIC路线，而非进入全面下行。", "summary": "截至2026-06-15，本卡压力测试GPU抵押融资收缩对半导体订单的影响：冲击不是AI芯片需求消失，而是订单可见度从Neocloud融资驱动采购转向超大规模厂商、ASIC、主权项目和带预付款的硬合同。NVIDIA Data Center在FY2027 Q1仍达752亿美元，TSMC先进节点和HP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创板去杠杆冲击短期股价但不破坏国产半导体设备及AI硬件的长期成长逻辑——龙头与AI芯片设计估值修复~20%即可加仓，二线与HBM配套需-30%以上才有安全边际。", "summary": "截至2026-06-15，科创板两融由~2,940亿元降至~2,820亿元、科创50击穿1,600点零Gamma线，对电子板块形成真实短期冲击，但不足以系统性下修国产半导体设备与先进制程龙头capex。设备龙头（北方华创订单覆盖15个月、现金~150亿元）、AI芯片设计（寒武纪2026Q1+312…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电网瓶颈正迫使 AI 基础设施从‘暴力规模扩张’转向‘受约束的能效优化’，电力并网速度已取代芯片供应成为核心估值护城河。", "summary": "本报告通过压力测试确认了物理电网并网排队（7-10年）与变压器交期（4-5年）已成为 AI 基础设施扩张的硬约束。定量模型显示，1年的并网延迟将导致 100MW 数据中心损失 5.5% 的生命周期价值，且快速的 GPU 迭代周期（12-18个月效用减半）加剧了闲置产能的 ROI 损耗。这种物理瓶颈…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 1, "risks": 5, "neutral": 0, "value": 66.2, "events": [ { "title": "验证AI算力硬件链的估值承压：流动性虹吸下的先进封装与GPU产业链韧性评估", "summary": "本卡对AI算力硬件链做压力测试，得出'分化韧性'结论。资金流层面确认虹吸：2026-06-08半导体−219.33亿、通信−98.35亿、消费电子−69.92亿元净流出居前[S6]，年内93家半导体公司减持[S7]，叠加CXMT约295亿IPO锚定[S9]。但估值下移是资金流/轮动驱动、非盈利驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施正进入“利率脱敏但电力受限”的深水区，高资金成本将瓶颈从硅片推向电网物理极限，并系统性推高非美玩家的准入门槛。", "summary": "2026-06-15：评估显示，尽管高利率和 DXY 走强（目标 101.3）增加了融资和采购成本，但头部超大规模厂商 (Hyperscaler) 的 AI Capex 因 AGI 生存竞赛而展现出极强韧性（2026 年预计达 7600 亿美元）。然而，高利率正在通过公用事业资金成本传导，加剧电力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观信用收紧并未击穿全球AI周期，而是将AI基础设施层切割为「超大规模云厂＋央企国家队」核心（继续扩张）与「民营商业云」外缘（防御性收缩）的K型结构，TMT投资应聚焦CoWoS/HB…", "summary": "TMT综合：沃什鹰派持稳（3.50–3.75%）经离岸人民币流动性、汇率对冲摩擦与企业中长贷收缩，传导至AI基础设施层产生K型分化而非总量收缩。美超大规模云厂CY2026合计capex运行率3,200–3,600亿美元由经营现金流支撑不受冲击；央企运营商云经PBOC科技创新再贷款（约2.5%）继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡片 02 — AI 芯片供应链：DeepSeek V4 后的\"架构定制化\"趋势 白板：1f5523fe c230 4de9 bcab 6fc3d7edec87 ｜ 卡片 2/8 ｜…", "summary": "卡片 02 — AI 芯片供应链：DeepSeek V4 后的\"架构定制化\"趋势 白板：1f5523fe c230 4de9 bcab 6fc3d7edec87 ｜ 卡片 2/8 ｜ 日期：2026 06 15 分析师：TMT 行业分析师 ｜ 立场： 支持（support） 核心问题：DeepSe…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 3, "neutral": 0, "value": 71.6, "events": [ { "title": "CoWoS 最早在4Q26从单一物理瓶颈转为弱化约束，但整体AI算力供应链大概率要到2H27后才以逻辑约束为主。", "summary": "截至2026-06-16，本卡压力测试认为 CoWoS 扩产会稀释 AI 芯片单一先进封装瓶颈溢价，但不会在2026年内把供应约束变成纯逻辑约束。CoWoS 缺口据报从约20%收窄至2026年底约10%，4Q26起封装紧缺溢价应压缩；但HBM、先进制程晶圆、系统交付、电力与客户ROI会接棒成为约束…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供给限制驱动的半导体能效转型主要表现为通用GPU的系统与精度级优化，而非ASIC/NPU的大规模替代；自研定制芯片的爆发反而加剧了全球先进制程与封装供应链对TSMC的物理集中度。", "summary": "截至2026-06-17，电力瓶颈确在倒逼算力能效转型，但并非单纯的通用GPU向专用ASIC切换。算法多变性（如MoE与动态推理）使自研ASIC面临极高的“代差失效”风险，通用GPU（如NVIDIA Blackwell）凭借软件与架构灵活性仍占据70%以上份额。能效提升主要通过系统精度（FP4/F…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体能效能把AI扩张从“堆GPU”改写为“每MW token产出竞争”，但只能买时间、改善单位经济性，不能让电力瓶颈消失。", "summary": "截至2026-06-16，本卡压力测试“电力预算驱动扩张”逻辑：半导体能效提升真实且重要，Blackwell、先进封装、vLLM/FlashAttention、稀疏MoE和定制ASIC可显著提升每MW有效token产出。但它不能完全对冲电力瓶颈，因为美国数据中心用电量可能从2023年的176 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 6, "neutral": 0, "value": 63, "events": [ { "title": "国产算力链订单承接力强、利润承接力中等偏低、技术自主性 4.6/10——市场份额扩张是订单口径而非利润口径的确定性，综合分 0.55。", "summary": "本卡压力测试 Research note 07 关于科创板国产算力承接 AI 久期再分配的论断。判定：国产 AI 芯片/服务器/网络厂商具备 6-12 个月的订单溢出窗口（T9a），但利润承接力受毛利率结构性低 30-40 pts、HBM/CoWoS 等效产能仅为 TSMC 8-10%、EDA 自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：…", "summary": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：AI 基础设施分析师 (ai infrastructure analyst) 立场：stress test（压力测试…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FOMC 后应以“选择性算力基础设施挤空”而非“无差别 AI 融涨”来交易：多稀缺、空低 ROIC，净多但对冲。", "summary": "2026-06-18 策略综合：前序链条成立，2027 风险是 ROIC 与结构分化而非 capex 悬崖。FOMC 后应做多稀缺算力资产而非泛 AI beta：前沿逻辑/设备/HBM/先进封装、自研 ASIC 与网络、电力与电网受益方应超配；商用 GPU 龙头保留但严控风险；低 ROIC 商用 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler capex 足以穿越能源 TCO 上行并支撑稀缺供应链，但投资表达必须是“多稀缺、空低 ROIC”，而不是无差别做多 AI。", "summary": "2026-06-18 压力测试结论：hyperscaler capex 增长足以让 AI 基础设施供应链消化能源成本带来的边际 TCO 压力，但只能支持“多稀缺”，不能支持泛 AI beta。四大买方 2026 capex 粗略包络约 $710B（自测算），Microsoft、Alphabet、M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 4, "neutral": 0, "value": 68.4, "events": [ { "title": "半导体代工与设备估值面临全球WACC上升及N2晶圆价格墙的压缩，叠加国内HBM3国产材料良率瓶颈导致的算力变现延迟，正驱动A股电子资金扎堆“四大龙头”并形成‘高融资、弱宽度’格局，面…", "summary": "截至2026-06-18，本压力测试卡片验证了半导体设备和材料需求的边际变化如何驱动电子板块的宽度衰退与杠杆拥挤。中芯国际、海光信息、寒武纪、北方华创“四大龙头”融资余额达650.0亿元，占板块总两融的17.62%，而板块宽度持续恶化（仅42.1%个股上涨）。压力测试显示-15%的回撤将引发四大龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM3良率瓶颈将AI基础设施风险从总量需求崩塌改写为本土投产节奏后移，并强化A股高融资TMT的收入兑现左尾风险。", "summary": "截至2026-06-18，本卡压力测试卡片06：国产HBM3良率瓶颈会推迟中国本土AI算力变现，但尚不足以证明全球AI基础设施从快速扩张转入需求观望。全球hyperscaler capex仍有Meta、Alphabet、Microsoft等高额指引支撑；真正变化在中国本土项目从按园区容量扩张转向按…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the analysis on how advanced packaging (Chiplets, HBM4) and optical interc…", "summary": "I have completed the analysis on how advanced packaging (Chiplets, HBM4) and optical interconnects (Silicon Photonics, CPO) mitigate U.S. power grid …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康存在选择性重估标的，但 AI 跨链外溢不应按篮子买入；基础设施先捕获稀缺性，医疗需等临床、监管、支付和现金转化验证。", "summary": "截至 2026-06-18，本卡从 AI 基础设施视角否定“大健康低覆盖=系统性低估”的宽泛判断。15/280 的 5.4% 激活率只是内部关注度信号；真正稀缺、已被资本密集定价的环节在算力、电力、数据中心与云 capex。医疗 AI 采用真实，FDA 列表已达 1,525 个条目，但监管、临床验…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 5, "neutral": 0, "value": 63.8, "events": [ { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与封装成本上升会触发AI基础设施技术路线调整，但截至2026-06-20仍不足以触发超大规模云厂商AI芯片采购总量的全面回撤。", "summary": "工作日2026-06-20：本卡压力测试认为，电力和先进封装成本上升不会在2026年触发超大规模云厂商全面削减AI芯片采购，但会提高单位弹性，推动技术路线从“更多GPU插槽”转向“每瓦性能、液冷机架、自研ASIC、提前锁定CoWoS/HBM和电力站点”。真正风险是并网、变电站、CoWoS或HBM配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Starlink 带宽 × Grok 算力的飞轮真实，使 SPCX 成为 AI 分发基座与算力链条需求催化剂，但算力归属 xAI、轨道边缘算力尚萌芽，真正瓶颈是电力——故支持但需把光…", "summary": "从 AI 基础设施视角支持卡片 01：Starlink 带宽（约 450 Tbps，三代卫星单星 >1 Tbps、下行 10 倍/上行 24 倍）与 Grok/Colossus 算力（约 55.5 万颗 GPU、约 180 亿美元、向 2 GW 扩张）构成真实飞轮，使 SPCX 成为差异化的'算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "xAI Colossus 2 与 Starlink 三代卫星建设对半导体算力链的量化带动与国产替代受益标的", "summary": "从半导体视角支持本议题：Colossus 2（约 55.5 万颗 GB200/GB300、约 180 亿美元）是真正撼动市场的芯片买家——自测算单站点即约 55–62 亿美元 HBM3E、约 1.9–2.2 万片 CoWoS 晶圆，集中在价值迁移之处（HBM 现占加速器 BOM >50%、CoWo…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 4, "neutral": 0, "value": 71.2, "events": [ { "title": "电网交付延迟不会终结高密度 AI 机柜，但会迫使下一代 AI 芯片与集群架构把 perf/W、可限功率和可分阶段并网能力置于峰值算力之上。", "summary": "2026-06-20：本卡对电网设备瓶颈约束 AI 芯片架构的说法进行压力测试。结论是，变压器和开关设备长交期不会让前沿训练集群普遍降密度；GB200 NVL72、Vera Rubin 和 AMD Helios 仍指向高密度机柜级设计。但电网交付延迟会把竞争指标从峰值 FLOPS 推向受限兆瓦下的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力资本开支是美联储路径上移传导至中国融资条件最敏感的通道，但 PBoC 大概率以结构性信用而非降息回应，从而坐实政策利率独立性。", "summary": "从 AI 基础设施视角综合 Research note 01–03：2026 年中国 AI 资本开支账单中约 USD 550–650 亿直接挂钩美元曲线（HBM、残余加速卡、离岸租赁），是美联储 3.8% 路径向中国融资条件传导的最干净通道。但绑点是 HBM/燃机等少数高价格刚性进口品，不是整体资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "summary": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力约束不会终结AI算力扩张，但会把半导体设计目标函数改写为每瓦有效算力，并提升Chiplet与先进封装的战略溢价。", "summary": "截至2026-06-21，本卡片支持前序判断：电网接入和可交付兆瓦成为AI扩张区域性天花板后，半导体竞争会从峰值算力转向系统级Perf/Watt。数据中心用电增长使每瓦tokens、HBM带宽、die-to-die互连、热设计和先进封装成为容量权变量。Chiplet、CoWoS/SoIC、HBM4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 3, "neutral": 0, "value": 78.6, "events": [ { "title": "截至2026-06-22，低端封装过剩会压低错误产能池的价格，而高端迭代缺失会保留真正重要产能池的配给风险，使Hyperscaler AI资本开支更重、更不稳定、更容易出现已供电机房…", "summary": "截至2026-06-22，本卡压力测试卡片06：先进封装低端过剩并不能缓解Hyperscaler真正需要的高端CoWoS/HBM4认证产能短缺。TrendForce称2026年行业CoWoS-like产能可接近200,000 wpm、缺口或收窄至约10%，但若高端认证占比只有50%-70%，有效高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-21，受2nm GAA先进制程 wafer 溢价与先进封装 80% 初期装配良率的限制，下一代高密度 AI 芯片的制造 BOM 将翻倍至8,153.59美元，其中封…", "summary": "截至2026-06-21，本卡片评估了AI机架密度120kW+下，先进封装良率及散热材料研发对AI芯片成本与供应链的重塑。下一代2nm GAA+HBM4+液冷芯片BOM成本较3nm基准大涨77.2%至6,780.97美元，在初期封装装配良率降至80%的重度压力下更暴增113.1%至8,153.59…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通胀重置与美联储转鹰情景下，经典股债配置策略失效，战术杠铃组合配合美债利率及日元套利平仓尾部期权对冲，是稳健跨资产回报的核心；A股风格上应坚决切向具备价格传导优势的调节性水电红利股…", "summary": "截至2026年06月21日，本报告压力测试了美联储加息和多资产配置方案。在通胀重置加息情景下，经典60/40组合预期大跌-15.60%（VaR为31.95%）；而战术杠铃组合超配短债/TIPS及20%商品，结合10%期权对冲，录得+0.45%正收益，VaR降至10.83%。尾部对冲方案采用10Y …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 3, "neutral": 0, "value": 80, "events": [ { "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "summary": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "可信算力区下的A股与STAR硬科技配置映射", "summary": "本卡把卡片01–07的'可信算力区'供应链结论翻译成A股/STAR硬科技排序，并用四道闸门（收入已拐点、订单可见、毛利确认、与可交付获批算力直接关联）区分真实订单受益与主题弹性。结论排序：光模块(中际旭创+192%、新易盛+106%)、AI服务器/ODM(工业富联净利+102.6%、订单至2027…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与定制化ASIC的深度整合，是将大模型访存能耗降低95.7%并实现3-5倍能效提升的核心硬件路径，亦是国产半导体链规避先进制程封锁、支撑算力电力约束的战略根基。", "summary": "截至2026-06-23，本卡研判支持通过先进封装与ASIC能效路径应对算力电力约束。2.5D封装结合HBM3E使访存功耗降低95.7%（从35 pJ/bit降至1.5 pJ/bit）；自研定制ASIC能能效比高出GPU约3-5倍，万卡集群年省电费2,943万元人民币，并释放48MW电网容量。国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 4, "neutral": 2, "value": 81.4, "events": [ { "title": "Summary of Strategic Synthesis Under the authoritative work date of 2026 06 23 , we have sy…", "summary": "Summary of Strategic Synthesis Under the authoritative work date of 2026 06 23 , we have synthesized the findings of the prior cards to address the m…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 算力基础设施的 CAPEX 扩张与云服务商利润率的边际变化", "summary": "表观云利润率仍稳健（2026Q1 微软 Intelligent Cloud 约43%、AWS 约35–38%、Google Cloud >17% [S5]），但这是会计与杠杆的人为支撑。真实成本结构沿三向移动：单加速器 BOM 上升（HBM 约占 BOM 45%、2026 涨价约20% [S11]…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "summary": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "硅片层分化(前沿量稳、英伟达定价权被定制ASIC侵蚀)对A股/指数AI集中度定价的传导", "summary": "卡片01指指数对AI集中度自满，卡片02–04论证物理瓶颈并揭示能效把量集中到前沿、定制ASIC侵蚀英伟达租金而非需求。本卡综合判断：被错杀的是利润率/结构事件而非总量事件，且A股已在沪深300内建出自己的AI集中度（中际旭创第一权重、新易盛第四、寒武纪第九，IT权重9.22%→20.38%）。配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 3, "neutral": 0, "value": 82.8, "events": [ { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM/先进封装自主与出口管制反身性是否构成对中国AI建设节奏的绑定宏观约束及政策驱动A股主题", "summary": "本卡综合 Research note 07：接受瓶颈已迁至国产HBM/先进封装，并将其升级为宏观—政策命题。国产HBM确是绑定约束，但绑定的是2026–2027的“节奏”而非“轨迹”——属软绑定。2026年仅约25万–30万颗 Ascend-910C 等效国产封装[S11]，仅相当于英伟达约360…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "跨主权存储链的 A 股 AI 国产化最终配置地图——按绑定硬度为材料/设备/封测三环定仓", "summary": "作为10卡线程收官，本卡把瓶颈迁移（需求→电力→HBM→先进封装→亚100nm键合机）综合为一张定仓的A股配置地图。核心法则：绑定硬度决定持有什么，拥挤度决定定多大仓，业绩动量仅作择时。三轴打分显示信念最高的键合机(4.0)与材料(3.2)恰是最不拥挤、动量最安静处，而封测(0.8)、算力(0.8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 5, "neutral": 0, "value": 72.2, "events": [ { "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网缓释后，AI集群上线风险不是消失，而是转移到CoWoS/HBM、液冷验收与网络芯片的串行同步，CPO下行收窄但不等于收入曲线完全恢复。", "summary": "截至2026-06-26，本卡压力测试Research note 7：电网侧缓释可降低CPO最坏收入下修，但不能消除AI集群的串行硬件瓶颈。关键约束从单一接电问题转向先进封装/HBM、液冷机柜验收、交换ASIC/网络集成的节奏匹配。CoWoS缺口可能由约20%降至2026年底约10%，但10%仍足…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备瓶颈不会系统性消灭AI加速器需求，但会把2026-2028年约15%-25%的GPU/HBM/ASIC部署变成收入确认和爬坡时点风险。", "summary": "截至2026-06-25，本卡从半导体侧压力测试电力设备瓶颈传导：2026-2028年加速器等效部署中约15%-25%面临2-4个季度时点风险，真实终端需求毁灭约3%-7%。商用GPU最直接受通电MW影响，HBM因多年协议和稀缺配额更像库存节奏风险，自研ASIC既受益于每瓦token优化，也因客户…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 2, "neutral": 0, "value": 91.6, "events": [ { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-27的权威工作日背景下，国产HBM良率骤降逼迫国内算力网络化，而全球封装与电力阻尼拉长了云厂商CapEx变现周期，这使得制程免疫且订单排至2028年的可插拔光模块出…", "summary": "截至2026-06-27，本报告评估了HBM良率危机与先进封装瓶颈对算力集群技术演进的影响。国产HBM3封装良率降至2.66%，倒逼国内算力集群转向网络密集的Scale-out拓扑，推高了光模块需求弹性。全球云厂商2026年CapEx直逼7250亿美元，但受制于CoWoS晶圆先进封装冲突与吉瓦级电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 2, "risks": 5, "neutral": 1, "value": 94, "events": [ { "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "summary": "截至 2026-06-28，本卡压力测试认为：电力交付瓶颈会迫使 hyperscaler 从单纯追求新增 MW 转向 MW 生产率和存量优化，但不会终结扩产。Microsoft、Alphabet、Meta 可见 2026 年 capex 区间合计约 4,950-5,250 亿美元（自测算），同时仍…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "summary": "电力与电网容量成为稀缺投入后，AI 芯片优化目标从绝对算力转向每瓦性能（perf/W）与单机柜 token 经济性。NVIDIA 宣称 Blackwell 推理能耗较 Hopper 最高低 25×（FP4/NVFP4 驱动），但 B200 单卡功耗仍升至约 1,200 W（H200 为 700 W…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 3, "neutral": 0, "value": 88.4, "events": [ { "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck fr…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-30", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "AI算力行情与电网瓶颈是否已成为同一个交易", "summary": "AI算力行情与电网瓶颈是否已成为同一个交易 Are AI compute rallies and grid bottlenecks now one trade? 在寒武纪突破万亿市值、PJM 容量成本与变压器交期继续抬升的同一天，团队应如何同时定价 AI 芯片盈利弹性和电力并网硬约束？", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-31f32af77b/", "networkHref": "/investor/research-network/#thesis-31f32af77b" }, { "id": "thesis-a803fa3312", "source": "tracked_thesis", "title": "美联储路径仍是全球资产再定价的总阀门", "titleEn": "The Fed path remains the master valve for global asset repricing", "lane": "美债利率/美元再定价", "laneEn": "US rates and dollar repricing", "laneHref": "/investor/lanes/us-rates-and-dollar-repricing/", "investableFocus": "美债利率/美元再定价：TLT, IEF, UUP, GLD, KRE, JPM", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "KRE", "name": "SPDR S&P Regional Banking ETF", "market": "US ETF", "role": "利率/信用压力代理", "roleEn": "利率/信用压力代理", "bucket": "core", "href": "/investor/stocks/KRE/" }, { "ticker": "JPM", "name": "JPMorgan Chase", "market": "US", "role": "大型银行利率周期", "roleEn": "大型银行利率周期", "bucket": "core", "href": "/investor/stocks/JPM/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "CARR", "name": "Carrier", "market": "US", "role": "冷却设备", "roleEn": "冷却设备", "bucket": "core", "href": "/investor/stocks/CARR/" } ], "stockTickers": [ "TLT", "IEF", "UUP", "GLD", "KRE", "JPM", "FXY", "XLE", "FXI", "EEM", "APO", "CAT", "MAR", "CARR" ], "coreView": "赛道：美债利率/美元再定价。候选观察标的：TLT, IEF, UUP, GLD, KRE, JPM。当前共有 40 条证据、3 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 76, "eventCount": 160, "firstSeen": "2026-04-27", "lastSeen": "2026-06-29", "supports": 8, "risks": 127, "neutral": 25, "analystCount": 10, "sourceTypeCounts": { "分析协作": 2, "研究笔记": 158 }, "topTerms": [ "美联储", "CPI", "PCE", "美元", "DXY", "美债", "银行", "path", "remains", "master", "valve", "global" ], "whatChanged": [ "2026-06-29: 霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "2026-06-28: 渐退的中东能源冲击与新主席 Kevin Warsh 治下美联储鹰派转向之间的碰撞", "2026-06-28: 截至2026年6月28日，AI及科技巨头的估值收缩并未击垮整体Beta，资金正在向受低油价、高位政策利率和指数重估支撑的小盘股、道指及防御板块进行结构性风格轮动。", "2026-06-28: 全球利差收窄背景下的中国资本流动与人民币汇率压力研判", "2026-06-28: 核心服务与住房通胀的粘性：美联储政策的后续驱动力" ], "practical": { "sectorLane": "美债利率/美元再定价", "sectorLaneEn": "US rates and dollar repricing", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "KRE", "name": "SPDR S&P Regional Banking ETF", "market": "US ETF", "role": "利率/信用压力代理", "roleEn": "利率/信用压力代理", "bucket": "core", "href": "/investor/stocks/KRE/" }, { "ticker": "JPM", "name": "JPMorgan Chase", "market": "US", "role": "大型银行利率周期", "roleEn": "大型银行利率周期", "bucket": "core", "href": "/investor/stocks/JPM/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "CARR", "name": "Carrier", "market": "US", "role": "冷却设备", "roleEn": "冷却设备", "bucket": "core", "href": "/investor/stocks/CARR/" } ], "stockTickers": [ "TLT", "IEF", "UUP", "GLD", "KRE", "JPM", "FXY", "XLE", "FXI", "EEM", "APO", "CAT", "MAR", "CARR" ], "primaryStocks": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "KRE", "name": "SPDR S&P Regional Banking ETF", "market": "US ETF", "role": "利率/信用压力代理", "roleEn": "利率/信用压力代理", "bucket": "core", "href": "/investor/stocks/KRE/" }, { "ticker": "JPM", "name": "JPMorgan Chase", "market": "US", "role": "大型银行利率周期", "roleEn": "大型银行利率周期", "bucket": "core", "href": "/investor/stocks/JPM/" } ], "instrumentTags": [ "TLT", "iShares 20+ Year Treasury Bond …", "IEF", "iShares 7-10 Year Treasury Bond…", "UUP", "Invesco DB US Dollar Index Bull…", "GLD", "SPDR Gold Shares", "KRE", "SPDR S&P Regional Banking ETF", "JPM", "JPMorgan Chase" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 160, "sourceDiversity": 2, "analystCount": 10, "recentEvents": 98, "supportRatio": 0.05, "riskRatio": 0.79, "conflictScore": 89.9, "evidenceDepth": 100 }, "score": 69.8 }, "events": [ { "date": "2026-06-29", "title": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "summary": "截至2026-06-29，支持卡片01–03的逻辑链并延伸至终端利率与全球流动性维度。美联储6-17维持3.50%–3.75% [S2]、SEP 2026中值3.8% [S3]；新主席Warsh首场FOMC点阵图从降息翻转为加息，9名官员预期加息、年内加息概率50.4%、市场定价10月加息，反应函数转向'约束性务实主义' [S14]。情景A（布伦特约105…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.25, "matchedTerms": [ "美联储", "CPI", "美元", "DXY", "global", "macro", "FOMC" ], "href": "" }, { "date": "2026-06-28", "title": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "summary": "2026年6月17日Warsh首会以12-0维持利率3.50%-3.75%，但SEP中位点阵从隐含降息翻转为隐含加息（3.4%升至3.8%），18人中9人看至少1次加息、17人看通胀上行风险，2026年PCE上修至3.6%。背景是5月CPI同比4.2%反弹与韧性就业（年内约57万非农）。2年期美债当日跳升逾16基点、DXY破100、离岸人民币滑向6.80出…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 38.25, "matchedTerms": [ "美联储", "CPI", "PCE", "美元", "DXY", "美债", "global", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "美元强势第二阶段与人民币/A股外资流传导", "summary": "本卡综合前序政策失误论证，落到汇率与资金流通道。美元处于2026年往返的第二阶段：DXY自1月95.57谷底反弹、议息后破100、6/25-26创13个月新高约101.37，但该段几乎全靠约75%的9月加息定价支撑，结构脆弱。USD/CNY约6.80为美元主导。关键在拆解美元下跌动因：套利通道（预期降息收窄利差）利好人民币与北向；避险通道（衰退式坏降息）先…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28.25, "matchedTerms": [ "美元", "DXY", "美债", "银行", "global", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "美联储2026年6月鹰派转向：新主席Warsh首会下点阵图翻转为加息，叠加中东能源冲击", "summary": "2026年6月17日FOMC维持利率3.50%–3.75%，但点阵图从隐含降息翻转为隐含加息：2026年底利率中值由3.4%升至3.8%，18人中17人判通胀风险偏上行，整体PCE预测上调至3.6%、核心3.3%。新主席Warsh以「五年失误必须纠正」的信誉优先基调定调，2年期收益率当日跳升约16基点、美元走强、股市下挫。但支撑这套预测的能源脉冲正快速逆转…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.25, "matchedTerms": [ "美联储", "PCE", "美元", "global", "macro", "FOMC" ], "href": "" }, { "date": "2026-06-28", "title": "Warsh 时代美联储 2026 年 6 月的鹰派范式切换及其对美元、全球利率与中国政策空间的外溢", "summary": "2026 年 6 月 17 日 FOMC 名为按兵、实为范式切换：新主席 Warsh 删去宽松倾向，点阵图由隐含降息翻转为隐含加息（18 人中 9 人预计 2026 年至少加息一次），利率维持 3.50%–3.75%。市场即刻定价：2 年期美债收益率跳升约 16 个基点至约 4.21%（逾一年最高），10 年期约 4.45%，美元指数突破 100，2026…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 23.75, "matchedTerms": [ "美联储", "美元", "美债", "global", "macro", "FOMC" ], "href": "" }, { "date": "2026-06-28", "title": "美联储6月鹰派转向与能源冲击逆转的错配:点阵图是否已经过时", "summary": "2026年6月17日FOMC维持利率于3.50%–3.75%,但点阵图转鹰、抹去年内降息并将2026年底中值上调至3.8%,隐含至少加息一次,期货一度定价10月加息。触发因素是美伊冲突引发的能源冲击:5月整体CPI同比4.2%、能源同比23.5%。但该冲击正快速逆转——布伦特从120美元上方跌至约72美元,霍尔木兹有条件重启,而核心CPI仅2.9%且环比降…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 23.75, "matchedTerms": [ "美联储", "CPI", "美元", "global", "macro", "FOMC" ], "href": "" }, { "date": "2026-06-28", "title": "2026年6月美联储、欧央行、日本央行同步转鹰，却在一场正在消退的伊朗战争油价冲击中收紧政策", "summary": "2026年6月，欧央行（+25bp至2.25%）、日本央行（+25bp至1%，1995年来最高）与美联储（按兵不动但点阵图从降息翻转为加息）在一周内集体转鹰，导火索是伊朗战争与霍尔木兹海峡中断推动的油价冲击。美国5月CPI同比4.2%创三年新高，但全由能源（同比+23.5%）驱动，核心仅2.9%。然而布伦特已从114美元峰值回落至约80美元，海峡有条件重开…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 23.25, "matchedTerms": [ "美联储", "CPI", "美元", "global", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "主要央行保持鹰派按兵不动并非过度收紧的政策错误，而是应对由服务业传导支撑的核心PCE粘性（预计H2维持在3.1%-3.4%）的必要防御，但这会使实体经济承受限制性实际利率压制，支撑久期债防守…", "summary": "本卡综合并支持前三张卡对能源脉冲退潮的判断，但指出核心通胀将出现背离：5月核心CPI放缓至2.9%，但核心PCE因服务二轮传导仍粘在3.4%。由于美联储盯防PCE，鹰派立场并非政策错误。然而CPI下行使限制性实际利率升至0.7%-1.15%，压制增长，支撑久期超配，并导致融资余额达28.13亿元、对海外宏观高敏感的中游机械面临收缩压力，建议低配/规避。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 23.25, "matchedTerms": [ "美联储", "CPI", "PCE", "global", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "能源价格的崩溃通过被动推升实质利率（188-264个基点）对实体经济造成隐性紧缩冲击，显著放大了美联储在Warsh新主席政策框架下过度紧缩的政策失误尾部风险。", "summary": "本卡片分析了能源价格向通缩力量转变对美联储的二次紧缩冲击。霍尔木兹海峡复航导致油价跌至70美元区间，全球供需过剩凸显。由于能源对核心通胀的传导偏弱，核心CPI仍具粘性，导致美联储在Warsh主席新政下固守限制性利率。这将推动实质利率较5月被动飙升188至264个基点，显著放大过度紧缩的政策失误尾部风险，迫使市场进行鸽派重定价并调整资产组合。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 23.25, "matchedTerms": [ "美联储", "CPI", "美元", "global", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "沃什联储6月鹰派转向：能源供给冲击下的'更高更久、加息可选'与全球外溢", "summary": "6月17日沃什首次主持的FOMC以12-0维持利率3.50%-3.75%，但点阵图转鹰：中值隐含2026年底约3.8%、9/18预计加息、17/18判断通胀上行，并取消前瞻指引。驱动因素是中东伊朗冲突引发的能源供给冲击——霍尔木兹受扰、布伦特一度较战前高逾60%，5月能源CPI同比+23.5%、核心PCE升至3.4%为2023年10月来最高。以核心PCE计…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.25, "matchedTerms": [ "CPI", "PCE", "美元", "银行", "global", "macro", "FOMC" ], "href": "" }, { "date": "2026-06-27", "title": "Warsh 时代美联储鹰派转向：美国通胀再加速与 2026 年加息风险", "summary": "2026 年 6 月 17 日 FOMC（Warsh 首会）将利率维持 3.50%–3.75%，但点阵图转鹰：2026 年中值从 3 月的 3.4% 升至 3.8%，由隐含降息翻为隐含加息，18 位官员中 9 位点位高于中点、17 位判断通胀风险偏上行。背景是 5 月总体 PCE 同比 4.10%、核心 3.4%，均创近三年新高，关税贡献约 0.80 个百…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28.5, "matchedTerms": [ "美联储", "PCE", "美元", "美债", "global", "macro", "FOMC" ], "href": "" }, { "date": "2026-06-27", "title": "美元 carry 仍有利差支撑，但在 DXY 高位后，USD/CNY 与 USD/JPY 的边际吸引力应从加杠杆顺势交易降级为带保护的战术 carry。", "summary": "2026-06-27：本卡支持前序“美元是约束性摆动变量”的判断。美联储维持 3.50%-3.75%，5 月 PCE 仍偏高，DXY 6 月 26 日为 101.3713。中美 2Y 利差约 +285.7 bp，美日 2Y 利差约 +268.1 bp，支持做多美元 carry；但 USD/CNY 约 6.80-6.82、USD/JPY 约 161.79，美…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28.25, "matchedTerms": [ "美联储", "PCE", "美元", "DXY", "global", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "再通胀压力下的跨资产配置策略：久期、商品与美元", "summary": "本卡支持卡片01的「韧性再通胀而非滞胀」判断，并落地为跨资产倾斜。盘面（2026-06-27）：5月PCE同比4.1%、核心3.4%，6月FOMC维持3.50%-3.75%且点阵图中位数升至3.8%；布伦特74.4美元（已较101.4美元峰值回吐27%），10年美债4.39%（7周低位）、2年4.20%，10年盈亏平衡仅2.18%，DXY处一年高位，黄金约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.5, "matchedTerms": [ "PCE", "美元", "DXY", "美债", "FOMC" ], "href": "" }, { "date": "2026-06-27", "title": "美联储2026年6月鹰派转向：供给冲击型通胀下的滞胀压力测试", "summary": "2026年6月17日FOMC连续第四次维持利率3.50%–3.75%（12–0全票，Warsh首会），但点阵图把2026年末中值从约3.4%上调至3.8%，抹去年内降息、隐含加息，2026年PCE预测自2.7%上修至3.6%，18人中17人视通胀风险偏上行[S1][S3][S4]。驱动力是供给冲击而非需求：5月CPI达4.2%（2023年4月以来最高），霍…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.5, "matchedTerms": [ "美联储", "CPI", "PCE", "美元", "FOMC" ], "href": "" }, { "date": "2026-06-27", "title": "2026-06-28：美国 May 2026 PCE 重新抬高美联储鹰派风险", "summary": "截至2026-06-28，May 2026 PCE 显示 headline PCE 同比4.1%、core PCE 同比3.4%，叠加 June FOMC 维持3.50%-3.75%以及 SEP 上修2026年通胀和联邦基金利率路径，近期降息叙事已被削弱。基准判断是 July 28-29 FOMC 鹰派维持利率，而不是立刻加息；但如果 June CPI 和…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.5, "matchedTerms": [ "美联储", "CPI", "PCE", "global", "macro", "core", "FOMC" ], "href": "" }, { "date": "2026-06-27", "title": "Brent跌至约72美元后，Warsh美联储的高利率前提应从“结构性再加息”下修为“数据条件化高位平台”：基准终端利率上沿3.75%-4.00%，降息需等待核心PCE和预期共同确认。", "summary": "2026-06-27：本卡片压力测试“长期高利率”前提。结论是油价回落削弱了单凭能源再加息的逻辑，但尚不足以触发降息定价；6月FOMC维持3.50%-3.75%，SEP年末中位数3.8%，核心PCE仍2.9%，长期通胀预期3.6%。基准为55%概率高位平台，30%概率油价/PCE触发转鸽，15%概率二次能源冲击再加息。组合应以3.75%-4.00%为基准终…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.25, "matchedTerms": [ "美联储", "PCE", "美元", "global", "macro", "FOMC" ], "href": "" } ], "timeline": [ { "date": "2026-04-27", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "[auto-handoff] 能源通胀对利率路径的溢价影响", "summary": "致 主题研究员： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 56407410-f02d-4931-ac33-d1b5f98aae84 结果摘要: 针对您的疑问，作为全球宏观分析师，我已完成对 2026 年下半年通胀预期及中性利率（R-star）走势的量化评估。 由于 …", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-02", "supports": 1, "risks": 0, "neutral": 0, "value": 51.4, "events": [ { "title": "2026 一次降息且 PCE 上修 30-50 bp 的可投资含义，是更少名义久期、更多前端通胀保护、战术性美元高配、有纪律的 5-7% 黄金配置，以及低久期权益因子轮动。", "summary": "本卡支持并转译 card-03 的宏观判断：6 月 17 日 FOMC 更可能确认更高更久的利率路径，而不是重新打开近端宽松交易。组合应低配名义美债久期 0.5-1.0 年，主要削减 20y-30y；高配 2-5y TIPS / 5y breakevens 0.5-1.0 年；将主要非 USD 敞…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-03", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "Powell 卸任与 Warsh 接棒：关税驱动滞胀压力下的美联储换届风险", "summary": "Powell 于4月29日主持末次 FOMC 后，Warsh 以13:11党派票数通过参议院银行委员会。一季度 GDP +2.0%，但 PCE 骤升至 +4.5%、核心 PCE +4.3%（2022年来最高），关税传导是主因。衰退概率达40–50%。基准情景（55%）：Warsh 维持4.25–4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "投委会的\"类滞胀\"框架方向正确，但应锚定住房而非能源、把\"金融超配\"拆分为 GSIB 超配 / 区域性银行低配，并在监控矩阵中增设住房触发器作为劳动力拐点的领先指标。", "summary": "从房地产视角对投委会\"类滞胀\"决议进行压力测试，得到三条具体修订：(1) H4L 的真正锚是住房 CPI（占比约33%）而非汽油，因此美联储紧缩比委员会描述的更耐久，10Y 目标上修至 4.80–5.25%；(2) \"金融超配\"必须拆分——CRE 到期墙约 5,700 亿美元、Trepp 办公违约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "区域性银行信用压力是否会改写美联储反应函数？ 日期： 2026 05 04 分析师： 首席经济学家 [Primary] 立场： 压力测试 (stress test) 卡片： 6 / …", "summary": "区域性银行信用压力是否会改写美联储反应函数？ 日期： 2026 05 04 分析师： 首席经济学家 [Primary] 立场： 压力测试 (stress test) 卡片： 6 / 10 承接： Research note 01 (全球宏观——\"类滞胀\")、Research note 02 (大类…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "外汇市场已开始定价美联储独立性折价（DXY残差约3%，USD/JPY超额日元强势约5日元），但仍处于部分定价阶段，三重触发条件（BoJ加息、Warsh鸽派确认、美债拍卖尾部）一旦叠加…", "summary": "压力测试结果：外汇市场已在部分定价美联储独立性折价，但尚非主要驱动因素。DXY年初至今跌幅7.4%中约2.9%为制度风险残差，USD/JPY跌幅中约5日元（占35%）超出利率基本面，归因于相对可信度再平衡与套息去杠杆；CNY渠道由人民银行主动管理，不构成独立的市场信号。约5,400亿美元日元融资套…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美元强势将因美债实际利率与期限溢价的双重回归而爆发，日本与中国的内部缓冲难以抵御结构性利差的引力。", "summary": "在美联储降息推迟至 2026 年末且美债曲线熊陡的情境下，美元指数（DXY）预计将突破 107.30 并挑战 110.00。由于中美、美日利差分别维持在 -300bp 和 480bp 左右的极端水平，日本央行的缓慢加息和中国权益市场的零星流入不足以抵消结构性的套利交易和债市资本外流。欧元区因增长疲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 0, "neutral": 1, "value": 52, "events": [ { "title": "云端CapEx挤出引发的存储通胀让美/欧/日换机曲线弯折但不击穿，对头条消费增长拖累仅5-10bp，真正的宏观信号是核心商品CPI再加速给联储/欧央行的政策路径增加摩擦。", "summary": "截至2026-05-19，云端CapEx虹吸HBM/CoWoS导致端侧设备ASP上涨50-100美元的冲击是真实的，但不会扼杀美/欧/日换机周期。三条结构性驱动（疫情装机潮到期、Windows 10 EOL企业换机、AI能力门槛）和高端买家低弹性把结果压成'量平/结构上移/ASP升'格局。对美+欧…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "地缘封锁带来的105美元油价与物流双击将美国推向滞胀困局，消费呈现结构性撕裂而非整体崩溃，而新任美联储主席凯文·沃什治下的鹰派加息风险已将下半年硬着陆（衰退）概率推高至55%。", "summary": "本卡评估了霍尔木兹海峡实际封锁（波斯湾超14mb/d产量受阻）导致105美元油价与航运费用大涨对美国经济的传导。虽因美伊外交谈判见曙光油价略跌至105美元，但这笔“地缘能源税”叠加海运费和国内陆运费上涨（+15% YoY），将美国4月CPI系统性推升至同比3.8%。一季度GDP增速已降至2.0%，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "能源价格冲击对全球通胀传导的深度评估", "summary": "能源端压力测试得出双向结论：传导真实存在但不会演化为螺旋。Brent ≈88美元、Henry Hub +35%、柴油 +16% [S5][S6][S7]，机械地为CPI贡献约1.4pp [自测算]。但1979式螺旋的必要条件不齐——美国油强度仅为1973年三分之一 [S3]、OPEC+闲置产能5.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "当前的弱美元叙事与 AI 驱动的 2.4% 增长及 4.2% 粘性通胀脱节，一份超 21.5 万的非农报告即可触发 DXY 重回 100 上方的机制转变。", "summary": "针对 Research note 01 的“弱美元”叙事进行压力测试。研究发现 DXY < 100 与 IMF/OECD 预测的 2.4% AI 引领增长及 4.2% 粘性通胀严重脱节。市场抢跑降息定价面临 6 月“过热数据簇”的瓦解风险。若 6 月 5 日 NFP 突破 21.5 万且 6 月 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 2, "neutral": 0, "value": 41.2, "events": [ { "title": "AI Capex韧性 × 油价冲击 × 联储延宕的共振，通过工业PPI与电力催生海外隐性二次通胀，将10Y美债锚定在5.0–5.3%，断裂点在日元/JGB而非美国风险资产。", "summary": "在2026-05-23工作日期下，AI硬件Capex韧性与霍尔木兹油价冲击的共振真实存在，但传导通过工业PPI、电力价格与资本品平减指数发生，而非头部CPI/PCE。这一通胀路径让联储基于PCE的反应函数系统性低估真实通胀，将10Y美债锁定在5.0–5.3%走廊延续整个2026下半年（期限溢价回升…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "贸易妥协通过对冲供给侧通胀压力，确保了2027年 disinflation 路径的延续，促使市场重新定价地缘政治风险溢价并支撑美联储向中性利率回归。", "summary": "本报告评估了美国在2026年底被迫延长稀土豁免的宏观影响。研究认为，这一“被迫妥协”充当了美国经济的供给侧泄压阀，有效避免了2027年因“绿色通胀”引发的二次通胀浪潮。这为美联储在2027年向3.1%的中性利率降息打开了窗口，并导致美债期限溢价压缩15-25bp。然而，代价是美国正式交出了关键矿物…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 4, "neutral": 0, "value": 30.6, "events": [ { "title": "中国宏观分析：外部双重压力下的政策空间（2026年5月） 日期： 2026 05 25 分析师： 中国宏观分析师 (agy) 状态： 综合分析 (Synthesize) 执行摘要 截…", "summary": "中国宏观分析：外部双重压力下的政策空间（2026年5月） 日期： 2026 05 25 分析师： 中国宏观分析师 (agy) 状态： 综合分析 (Synthesize) 执行摘要 截至2026年5月25日，中国经济面临复杂的“双重外部压力”：美联储维持高利率（10年期美债收益率4.69% [S1]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美元融资基差的极端化正在诱发非美机构的‘强制本国化’回流，这不仅削弱了美债的需求底座，更令日元和欧元在流动性枯竭中呈现出极端波动风险。", "summary": "美元融资基差走宽（USD/JPY 触及 -120bp）已通过 FX-swap 市场对日元和欧元产生显著结构性压力。由于对冲成本飙升（年化 >4%），日本寿险公司被迫降低对冲比率并开始赎回美债，这虽在短期内因美元荒支撑 DXY，但埋下了日元空头挤兑的隐患。欧元区银行亦面临美元资金缺口，加大 EUR/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-25：4月通胀冲击与5月FOMC纪要之后的美国政策路径重定价", "summary": "截至2026-05-25，Fed更像按兵不动或加息尾部风险，而不是确定的降息交易。4月CPI、3月PCE、Q1 GDP和5月FOMC纪要共同显示：通胀仍高、增长仍稳、劳动力只是温和降温。5月22日10Y美债4.56%、30Y美债5.07%，说明长端利率仍在为通胀与期限溢价索价。关键检验点是2026…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Fed终端3.50–3.75%通过利差扩大、欧元/日元贬值压力与美元融资基差走宽三条渠道向ECB与BOJ外溢，对全球长久期与EM硬通货资产构成结构性逆风。", "summary": "美联储终端利率上修至3.50–3.75%意味着ECB与BOJ被迫面对结构性更宽利差：ECB在4月降息至2.00%后政策空间被压缩，本轮终端下限上修至1.75–2.00%；BOJ虽3月加息至0.75%，但1.0–1.25%硬天花板下利差仍达225–275bp。美元流动性通过TGA重建、ON RRP见…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "截至2026-05-26，中国货币政策的边界不是“不能宽松”，而是“不能在美债期限溢价上行时用显性降息制造人民币单边贬值预期”。", "summary": "截至2026-05-26，本卡压力测试认为：美联储若因美债期限溢价转鹰，中国并不会被迫全面收紧，因为一季度经常账户顺差1841亿美元、4月外储3.4105万亿美元、4月银行净结汇401亿美元仍提供缓冲。但价格型宽松窗口明显收窄；当中美10Y利差从当前约283bp扩至330-350bp、银行结售汇转…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "实物原油供给赤字的长期化将打破央行的通胀容忍红线，迫使全球宏观叙事从“预防式降息”向“滞胀紧缩”急剧反转，触发系统性Risk-Off抛售。", "summary": "本报告通过宏观压力测试指出，若Hormuz断供拖延至Q3且Brent油价直奔$130，美联储将被迫完全撤销2026年降息指引甚至重启加息讨论，10年期美债有重返4.8%的风险；欧洲央行将陷入深陷滞胀与主权利差走阔的死局；日本央行则极大可能被迫紧急加息以保卫日元。这将彻底逆转当前的降息计价，引发全球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 3, "neutral": 0, "value": 39.4, "events": [ { "title": "[auto-handoff] 美伊外交谈判传闻及油价跌破100美元对美联储下半年降息与DXY的边际影响评估", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: d5d2ffa6-c741-4e78-b373-a9292f4c8a66 结果摘要: Wrote handoff_response.md (10595 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高利率维持叠加通胀再加速预期下，防御策略应放弃经典60/40拉久期思路，转向'短久期高质量固收+实物资产（黄金、能源）+质量/现金流权益+T-bill现金'的杠铃式组合，并显式低配长…", "summary": "在美联储6月预计维持3.50%–3.75%、核心CPI同比2.8%且面临二次加速的背景下，主要风险已从衰退尾转为'粘性通胀+股债正相关'。建议防御性配置为：权益42%（质量/价值/能源/防御性，低配长久期成长与小盘）、固收28%（1–3年美债+5年TIPS+短端IG，低配长久期美债与HY）、商品1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国30年期国债破5%:期限溢价、财政漂移与中东能源冲击下的长端重定价", "summary": "在美联储维持3.50–3.75%、4月CPI回升至3.8%、5月13日30Y招标中标5.046%(2007年以来首次5字头)的组合下,长端已与降息周期脱钩,转由期限溢价定价:中东能源冲击锚定通胀预期、FY26赤字突破2万亿美元放大长端供给、美联储反应函数被长端约束。基准情形30Y三季度末区间4.9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "6月基准应是滞胀暂停而非降息，科技久期交易需要用更高折现率重定价。", "summary": "2026-05-28结论：支持把6月降息从基准情景下调为上行情景。Brent约97美元/桶、Hormuz实际复航未确认、FOMC纪要偏鹰、CPI/PCE/GDP/就业组合显示通胀风险高于增长下行压力。AI基础设施可保留，但AI应用、量子/Web3、卫星互联网等久期资产应以4.5%以上10年期美债环…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "2026-05-28 核心判断：美国电网瓶颈的宏观风险不是一次性CPI爆点，而是受监管电价的多年粘性和长久期公用事业融资供给对美债曲线的边际挤压。", "summary": "截至2026-05-28，本卡对电网/公用事业链做压力测试：美国大型电力变压器2019年进口依赖度82%，常见交期36个月、最高60个月；2026年关税框架把部分金属密集型电网设备纳入15%关税至2027年。结论是，变压器短缺不会单独制造广义CPI冲击，但会通过费率申请形成粘性的电价通胀底部；同时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 0, "value": 49.6, "events": [ { "title": "能源价格重心的下移提供了一次缓和全球“滞胀”压力的正向供给冲击，虽然成功消解了美欧央行的强行加息风险，但在低储蓄率、物流滞后与-273 bps中美巨幅倒挂的刚性约束下，全球主要央行在…", "summary": "本报告推演了Brent原油重心下调至82-92美元对三季度全球通胀与货币政策的边际影响。受汽油直接回落与航煤暴跌（裂解价差收窄至18-22美元/桶）间接拖累，三季度美国整体CPI与欧洲HICP将分别下修21 bps和30 bps。这有效移除了美联储年底加息选项，但受制于2.6%极低储蓄率与核心PC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "未来 3 个月美债熊市变陡 50 bps 的真实概率 15–18%，并非可忽略尾部，方案四长端利率风险需提前布局对冲。", "summary": "以 2026-05-29 为锚，结合 FOMC 4.25–4.50% 政策利率、核心 PCE 仍 2.6–2.9% 的粘性、FY2026 净付息突破 1.1 万亿美元的财政背景与 ACM 期限溢价已回正，识别四条熊市变陡触发路径：A 通胀再加速 12%、B QRA 长端供给冲击 7%、C 海外需求…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 3, "neutral": 0, "value": 45, "events": [ { "title": "油价持续高位将迫使美联储在2026年下半年保持鹰派立场，推升10年期美债收益率至4.75%并使期限溢价突破100bps。", "summary": "评估布伦特原油维持在105-120美元/桶对美联储政策及美债收益率的冲击。高油价将直接贡献headline CPI约1.3-1.5%，并推升核心PCE至3.4%以上。美联储预计将维持3.75%-4.00%的高利率，甚至实施防御性加息。受通胀不确定性、沃什提名政权更迭及地缘溢价驱动，预计10年期美债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "压力测试：4.75% / 102 bps 期限溢价情景下的美债曲线陡峭化与企业债利差扩张 日期： 2026年5月30日 机构： 研究所 — 固定收益组 债券分析师： bond ana…", "summary": "压力测试：4.75% / 102 bps 期限溢价情景下的美债曲线陡峭化与企业债利差扩张 日期： 2026年5月30日 机构： 研究所 — 固定收益组 债券分析师： bond analyst 立场： 压力测试（针对 Research note 01–03 结论的二阶检验） 1. 执行摘要 Rese…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "消费者承担接近 100%、企业短期承担约 64%、CNH/MXN 几乎零吸收且 VND 远不足以对冲，关税新边际压力在 2026 下半年已显著消退但水平被嵌入；CNH 反向走强使中国…", "summary": "量化关税传导：截至 2026-04-08，美国有效关税税率 11.8%（Yale Budget Lab），核心商品 PCE 已上行 3.1%、对核心 PCE 通胀贡献 +0.8pp，传导 7 个月内基本完成；Dallas Fed 确认 2026Q1 见顶，4 月核心商品 CPI 几乎走平。短期分担…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 2, "neutral": 1, "value": 55.4, "events": [ { "title": "作为 外汇策略师 (FX Strategist) ，我已在权威工作日期 2026 05 31 的宏观基准下，系统性地完成了第七张研究卡片的分析，并已将三个必选交付文件成功写入工作区。…", "summary": "作为 外汇策略师 (FX Strategist) ，我已在权威工作日期 2026 05 31 的宏观基准下，系统性地完成了第七张研究卡片的分析，并已将三个必选交付文件成功写入工作区。 工作总结与核心分析要点 1. 美元指数（DXY）与全球利差： 在美联储因信贷慢性收缩将 2026 年底点阵利率中位…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美联储‘更高更久’路径与强势美元通过经常账户恶化和外债本币化利息压力，正在系统性蚕食新兴市场外币流动性防线并使其陷入增长与防守的两难陷阱。", "summary": "在2026-05-31的政策压力测试下，美联储因能源通胀维持3.50%-3.75%高利率，推升美债10Y至4.5%-4.8%并推高美元。这通过资本外流和债务本币化压力严重挤压新兴市场。定量模型显示，在极端系统性压力下，印尼和印度的外汇储备流失均超10%，外债利息本币负担激增近33%。当前印尼央行已…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "若6月跨太平洋现货运价继续上行但TEU只温和增长，应优先解读为船公司定价和成本冲击，零售利润率先受压，CPI/PCE传导选择性且滞后。", "summary": "截至2026-05-31，本卡综合支持“成本挤压而非需求繁荣”判断。Drewry在2026-05-28显示WCI升至每40ft集装箱2,800美元，跨太平洋航线同时受blank sailings和旺季附加费支撑；NRF则预计2026年上半年进口12.59百万TEU、同比仅增0.5%。美国4月零售销…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 2, "neutral": 1, "value": 56.8, "events": [ { "title": "美元窄幅震荡反映了短期非农数据确认前的周期性观望与黄金超越美债成为第一储备资产所体现的去美元化中长期结构性压制之间的双重博弈。", "summary": "油价维持高位、JOLTS超预期、美联储转鹰，但美元指数（DXY）仍在99.23窄幅震荡。本报告认为，美元钝化是短期战术观望（等待6月5日非农就业及EIA库存数据）与长期去美元化结构性阻力（欧央行6月报告显示黄金已超越美债成为全球首要储备资产，占比达27%）共同作用的结果。Hormuz封锁的地缘避险…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "DXY 99区间会弯曲但不会断裂——欧日央行被ToT冲击逼入墙角，但它们被迫的鹰派反应恰好阻止了EUR/JPY的失序崩溃，而黄金吸收了本应推动DXY破100的边际避险流。", "summary": "本报告支持并延伸卡片01的美元钝化结论：Hormuz封锁对欧日构成高度非对称的BoP冲击（自测算欧元区年化能源账单+1400-1800亿欧元；日本+6.5-8万亿日元），但ECB与BOJ均被困两难——能源CPI再加速阻止主动宽松，而ECB预计6月鹰派持有(60%概率)、BOJ预计7月象征性加息25…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在“偏晚、偏浅”的降息周期下，跨资产组合应集中曲线中段久期、靠合规溢价获取方表达医疗超配，并以黄金加高实际利率EM FX篮子代替方向性美元，把宏观结论翻译成可执行的仓位规则。", "summary": "截至2026-06-03，本卡把Card-07的“偏晚、偏浅”美联储路径翻译为6-18个月的跨资产配置规则：久期保持在策略中性的85%、集中于美债3-7y中段，对20y+显式低配，并配置5y TIPS以变现20-40 bp医疗底部；医疗超配+350 bp，70%以上仓位放在已验证非中国KSM供应、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 2, "neutral": 1, "value": 66.2, "events": [ { "title": "沃什(Warsh)新联储的滞胀困局:6月16–17日FOMC的政策抉择", "summary": "美联储连续第三次将利率维持在3.50%–3.75%,4月以8–4票通过,为1992年来首次四票反对:Miran主张降息,三位地区主席反对保留宽松倾向。新主席Kevin Warsh于5月22日就任,6月16–17日为其首场FOMC。通胀冲击源于供给端:霍尔木兹海峡3月关闭推动Brent一季度末破12…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "通信 40% 是成立的防御锚，黄金 25% 目前仍成立但必须设为 15% 核心仓 + 10% 战术仓；只有在 10Y 实际利率突破 2.5% 且美元同步上破时，才应把黄金阶段性降至 …", "summary": "2026-06-04：本卡支持卡 08 的通信 40% 核心防御仓位，并有条件支持黄金 25%。全球宏观仍验证输入型 PPI 冲击：美国强于欧元区、Fed 未能转向宽松、红海/中东物流与油价仍扰动，中国 PPI-CPI 剪刀差偏上游。但黄金应拆为 15% 核心仓与 10% 战术仓；若美国 10Y …", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "美国电网容量危机正通过“工业高电价-核心通胀粘性-美联储鹰派转向”的链条，对制造业再回流构成实质性压制，并重塑美联储的降息路径。", "summary": "截至 2026 年 6 月 5 日，美国电网容量危机已从公用事业局部问题演变为核心宏观阻力。PJM 容量价格连续两年触及上限（2027/28 年度清算于 333.44 美元/兆瓦-日）且存在 6.6 吉瓦缺口，导致美国东部工业电价上涨 15-25%。这种“供给侧冲击”正通过成本转嫁推升核心 PCE…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美联储的“鹰派暂停”与 4% 以上的通胀率构成了人民币 7.35 防线的结构性威胁，利差走阔使汇率破防风险升至 2026 年峰值。", "summary": "本卡片深入分析了2026年6月美联储利率路径对人民币7.35防线的外部冲击。研究显示，由于中东能源危机及关税传导，美国Headline CPI已反弹至4.05%，美联储处于“鹰派暂停”状态，市场定价已排除年内降息，甚至有41.7%的概率年内加息一次。美债10年期收益率回升至4.48%，导致中美利差…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 3, "neutral": 0, "value": 50.6, "events": [ { "title": "外部系统性杠杆（1.304万亿美元融资余额与50倍杠杆美债套利交易）处于高位，六月中旬的CPI与美联储点阵图超预期鹰派将成为二次清算触发点，击穿加密市场现有支撑。", "summary": "本卡对 Research note 01/03 的清算出清乐观共识进行压力测试。我们指出，虽然加密市场内部杠杆已出清，但外部系统性杠杆异常高企：截至2026年4月30日，全美融资余额达1.304万亿美元的历史高位；美债期现套利杠杆高达50倍，且逢清算合规过渡期。后续CPI超预期或FOMC削减降息预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观重定价证实BD现金流硬资产属性，但需以可触发里程碑而非合同名义值为锚，基线下NPV小幅下修但远未到悬崖。", "summary": "工作日期2026-06-06。在H2 2026美联储隐含降息50-75bp、DXY 102-106、BIOSECURE仍处5年豁免且未列入WuXi系的基线下，本卡为康方9926.HK与科伦博泰6990.HK的美元里程碑现金流重构折现率：Rf=4.35%、ERP=5.5%、β=1.1、政策溢价基线2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh治下的美联储将拒绝通过QE干预稳定币抛压，但会依靠SRF在无需放弃量化紧缩（QT）的前提下守住美债市场流动性底线。", "summary": "2026年6月5日，新任美联储主席Warsh的“缩表鹰派”立场及规则导向型政策，拒绝通过QE干预市场，加剧了由稳定币赎回引发的美债抛售压力。然而，现有的常设回购便利（SRF）具备5000亿美元的总额度，足以吸收估算的1600亿美元稳定币美债抛压。这导致了“流动性分化”：美债市场底层管道通过SRF维…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 2, "neutral": 1, "value": 69, "events": [ { "title": "面对'劳动降温+核心粘性'的非对称宏观，正确答案不是单一押注，而是质量股票+短端carry+精选长端的杠铃，并把黄金作为滞胀尾部保险。", "summary": "在5月就业降温但核心通胀仍粘（NFP +172k、失业率4.3%、核心PCE 3.3%）的组合下，建议总体股票风险中性、股票内偏质量（超配+3至+5 ppt）、固定收益采用'短端carry+精选长端'杠铃（现金/短期美债+4 ppt、1–3年IG+2 ppt、10年UST+1 ppt、10年TIP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储6月按兵不动+美债维持高位的情景下，亚洲外汇尚未充分定价美元再度走强，应做多USD/CNH、USD/KRW、USD/JPY期权敞口。", "summary": "2026-06-07判断：若6月17日FOMC按兵不动且10Y美债维持4.45-4.60%，DXY有空间从104.3回测105.5-106.5，USD/CNY从7.18重新定价至7.22-7.28，USD/KRW至1,390-1,410，USD/JPY漂移至158-160（160上方有日财MoF干…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-07：美国5月非农超预期后的美联储6月政策路径", "summary": "2026-06-07基准判断：5月非农新增172,000人、失业率4.3%、3-4月合计上修93,000人，使6月16-17日FOMC降息难以成立。4月PCE同比3.8%、CPI同比3.8%，叠加ISM服务业和制造业PMI仍扩张，支持美联储短期维持3.50%-3.75%目标区间。除非6月10日CP…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2026-06-06美国就业与通胀组合对美联储路径和全球利率的影响", "summary": "截至2026-06-06，5月非农新增172,000人、失业率4.3%、工资同比3.4%，显示就业降温但未断裂[S1]；4月CPI同比3.8%、核心PCE同比3.3%，仍限制美联储快速转向[S2][S3]。报告判断6月会议前不宜追逐激进降息交易，除非6月10日5月CPI明显降温[S6][S7]。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 2, "neutral": 3, "value": 68.4, "events": [ { "title": "2026年中美联储僵局：能源冲击驱动的通胀再加速锁死政策路径", "summary": "4 月美国 CPI 同比 3.8%、核心 2.8%，能源单项贡献当月涨幅 40% 以上，住房环比再加速至 0.6%。霍尔木兹自 2 月 28 日事实关闭、阿联酋 5 月 1 日退出 OPEC，使本轮能源冲击不再是一次性传导。基准判断：6 月 FOMC 维持 3.50–3.75% 不变；真正的二元拐…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "6月FOMC前的美国类滞胀格局：油价冲击、粘性核心通胀与韧性就业", "summary": "4月CPI同比3.8%/核心2.8%、5月非农超预期+17.2万、布伦特从$117峰值回落至$95——这套组合让联储6月16-17日按兵于3.50-3.75%几乎成为定局。能源的反通胀顺风已消失,工资增速3.4%仍高于与2%核心PCE相容的水平。10Y美债已升至4.55%,曲线定价'联储卡住'。基…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "106美元原油与强美元引发的材料端成本飙升正导致严重的实体需求破坏和利润崩塌，由此带来的全球制造业衰退将向美欧输出通缩，从而阻断美联储的鹰派路径。", "summary": "作为材料行业分析师，我否认外汇策略师单向看多美元（DXY至102.5）的逻辑。“106美元油价+强美元+4.95%美债收益率”的毒性组合正在化工（石脑油裂解利润转负，开工率降至70%）和有色金属（伦铜承压，非美需求萎缩）领域引发严重的实体需求破坏。这种材料行业的利润崩塌将引发全球制造业衰退，中国过…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "配置重心应从对冲能源滞胀转向捕捉“AI基建通胀”与“Warsh生产率叙事”的错配，超配实物资产（铜/黄金）与AI成长股。", "summary": "随着能源冲击转向机械性消退，油价企稳于93.09美元，跨资产配置正转向由“IT通胀”驱动的结构性变局。新任美联储主席沃什面临生产率叙事与实物资产瓶颈（如15万吨铜缺口）的博弈。建议配置上超配AI基建相关成长股与工业金属，对美债久期维持谨慎中性，并利用美元对非美货币的差异化走势。Warsh的“生产率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 1, "risks": 3, "neutral": 0, "value": 62.8, "events": [ { "title": "4.2%的CPI打破了美联储的“容忍阈值”，市场必然走向鹰派重定价，风险资产承压。", "summary": "若Brent维持在95美元且5月CPI达4.2%，市场将无法将其视为一次性冲击，而是会重新定价更鹰派的Fed。因担忧通胀预期脱锚，美联储将被迫维持高利率，这将推高美债收益率与美元，并对股市等风险资产造成显著的估值压制。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "5月CPI再印+0.6%/+0.4%是AI第二轮降杠杆的充分但可控催化剂；真正的风格拐点线是10Y美债收于4.85%以上，否则FOMC前是路径重定价而非政体反转。", "summary": "若2026-06-10的5月CPI延续4月环比+0.6%、核心+0.4%的动能，同比预计抬升至+4.1%–+4.3%（核心+2.9%–+3.0%），将推动2Y美债至4.30%–4.45%、10Y测试4.65%–4.78%，12月加息概率从64%升至80%以上。这足以触发AI交易第二轮降杠杆（最拥挤…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储2026年加息概率重估与全球流动性拐点", "summary": "支持卡片01。我把分析推至上游的美联储反应函数:5月CPI(2026-06-10发布)若证实再加速(核心环比≥0.4%、整体同比≥3.9%),美联储不会在6月或7月加息,但会把市场隐含的2026年加息概率从约28%重估至40–45%,并将点阵图叙事从'两次降息'翻转为'按兵乃至加息'。正是这一重定…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "10Y美债4.85%是正确的风格触发线，但不是5月CPI单独驱动的基准结果；只有实际利率、前端和波动共振时，AI硬件才会从去估值转入权益久期止损。", "summary": "截至2026-06-08，本卡压力测试认为，5月CPI偏热可推动10Y美债测试4.70%–4.80%，但有效突破4.85%需要CPI、实际利率、前端定价、权益波动和美联储沟通共振。FOMC前两日收于4.85%以上的无条件概率为15%–20%；若CPI达到或超过headline +0.6% / co…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 0, "neutral": 1, "value": 77.2, "events": [ { "title": "医药'去中国化'是一股结构性、慢释放的供给侧通胀力量，每年向美国核心医疗CPI注入25–40bp，并已迫使美联储将生物科技融资环境纳入金融稳定框架。", "summary": "截至2026-06-12，《生物安全法案》及1260H清单升级已将中美医药脱钩转化为可测量的供给侧成本冲击。按模态加权，非中替代溢价合计抬升受影响SKU零售价4.5–6.5%，对应美国核心医疗服务CPI每年额外+25–40bp、Medicare Part B支出增速每年高出基线60–90bp，20…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 3, "neutral": 1, "value": 68.6, "events": [ { "title": "结构性金属供给紧缩对全球通胀及美联储政策路径的影响", "summary": "支持（有界定）：铜/铝结构性缺口驱动的二次通胀足以硬化美联储终端利率逻辑，但机制是抬高底部、消除2026年降息期权，而非单独迫使加息。LME铜2026年6月初约13,800美元/吨、1月29日创14,527.50美元/吨历史高位，高盛上调年底预测至13,735；铝突破3,015.50美元/吨、对应…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "美国通胀回升与‘沃尔什’美联储战略转向（2026年下半年展望）", "summary": "截至2026年6月中旬，美国经济面临通胀超预期回升（5月CPI 4.2%）与美联储领导层更换的双重冲击。凯文·沃尔什作为新任主席，预计将废除前瞻性指引，转向战略模糊，加之地缘政治导致的能源价格高企，降息预期已彻底消退，市场开始定价年底加息。美债收益率与美元的持续强势将成为下半年全球资产定价的核心驱…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "沃什履新美联储主席后的首次FOMC（2026年6月16-17日）与供给侧能源冲击下的滞胀困局", "summary": "2026年6月16-17日是Kevin Warsh出任美联储主席后的首次FOMC。基准情形为维持利率3.50-3.75%不变，但通过点阵图（抹去2026年最后一次降息、甚至可能废弃点阵图框架）和措辞由宽松转向中性偏紧，正式确立'更高更久'格局。关键张力在于：5月整体CPI 4.2%看似惊人，逾60…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月央行双会周（日本央行6月15-16日加息至1.00%、美联储6月16-17日按兵不动）叠加正在降温但仍高企的中东石油冲击，对全球风险资产与日元套利交易的双重风险", "summary": "进入6月16日当周的央行双会，日本央行预计加息25bp至1.00%、美联储料以约97-99%概率维持3.50-3.75%不变。5月美国CPI同比4.2%为2023年4月以来最高，但约60%涨幅由能源驱动，且布伦特已从3月底逾$113峰值回落至约$89（6月11日因美伊降温下跌逾4%），通胀尾部正在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 7, "neutral": 2, "value": 49, "events": [ { "title": "能源CPI与核心CPI背离下，美联储应上调通胀预测并取消2026年降息，但6月仍以鹰派按兵不动为基准。", "summary": "截至2026-06-14，能源冲击要求美联储显著上调2026年总体PCE预测，但核心上修应更温和。5月CPI同比4.2%、核心2.9%，加上4月PCE 3.8%、核心PCE 3.3%，说明压力真实但尚未全面扩散。基准建议是6月维持3.50%-3.75%，把2026年点阵图从一次降息改为不降息；若布…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美元高息与避险属性叠加将压缩广谱非美套息交易，并使CNH承受持续但可管理的贬值压力。", "summary": "截至2026-06-14，本卡综合上一张鹰派美联储主线：若6月16-17日FOMC确认2026年下半年维持3.50%-3.75%高利率，美元将继续获得收益率和避险双支撑。DXY已在6月12日升至99.8070，USD/CNH在6月14日为6.76377。中美政策利差按美联储中点3.625%与中国7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "沃什时代的鹰派首秀通过重塑全球利差优势将美元重新‘武器化’，这不仅加剧了非美货币的贬值压力，更迫使全球套利交易进入深刻的风险重定价阶段。", "summary": "美联储主席沃什的鹰派转向将降息预期推迟至2027年，暗示年内可能加息，彻底逆转了此前的宽松预期。美元指数（DXY）受此提振从99.80强力回升至101.25。中美10年期国债利差走阔至200bp以上，加剧了人民币贬值压力，USD/CNY目前在7.2350附近波动。全球套利交易因利差优势的维持而重回…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储鹰派转向通过深度利差倒挂严重挤压了人民银行的降息空间，人民币汇率防御战已进入7.30关键博弈期，货币政策将更依赖结构性工具而非总量降息。", "summary": "截至2026年6月14日，美联储转向鹰派导致中美10年期利差倒挂加深至-280bps的历史低位，严重挤压了人民银行的政策空间。尽管国内CPI仅为1.2%，但汇率稳定与银行净息差（NIM）压力限制了LPR/MLF的下调可能。预计人行将坚守USD/CNY 7.25-7.30防御区间，通过逆周期因子、离…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 1, "neutral": 1, "value": 75.4, "events": [ { "title": "美联储'更高更久'+油价冲击向中国的传导：政策分化、人民币与A股配置", "summary": "我部分推翻Research note 01的设定。中美政策分化在2026年年中不仅可持续、且已无压力地发生——但机制不同。5月中国PPI同比+3.9%（2022年7月来最高，受同一伊朗战争大宗冲击+AI资本开支推升），CPI仅+1.2%，故'通缩给人民银行空间'被削弱[S1][S2]；人民银行5月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh首场FOMC：油价驱动型滞胀下的鹰派转向与全球外溢", "summary": "2026年6月16-17日是Warsh就任美联储主席后的首场FOMC，利率维持3.50%-3.75%已近乎确定（约97-99%）。他接手的是1979年沃尔克以来最难开局：5月整体CPI同比4.2%（2023年4月来最高），但核心仅2.9%——缺口几乎全来自伊朗战争引发的油价冲击（汽油+40.5%、…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 3, "neutral": 0, "value": 61.8, "events": [ { "title": "美国5月通胀反弹与6月FOMC前的全球宏观交易含义", "summary": "截至2026-06-15，本卡选择global-macro视角。结论是：6月FOMC大概率维持利率不变，但声明会让近期降息门槛更高。5月通胀主要由能源拉动，核心CPI环比仍温和；但4月核心PCE仍高、5月就业不弱、居民通胀预期仍偏高，使美联储很难快速转鸽。对市场而言，关键不是单纯押注利率上行，而是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "当前美元、实际利率与中国信贷脉冲坐标下，宏观流动性对铜价的净影响为-1%至+2%，不足以抵消矿端结构性约束所支撑的8%-12%利润率弹性，Research note 5/6/7 的轮…", "summary": "截至2026-06-17，DXY处于98-100、10Y TIPS实际利率1.7%-1.9%、中国信贷脉冲从-3%修复至-1.5%。基于共识弹性（DXY ±1%→铜∓1%、实际利率±100bp→铜∓6%-10%、信贷脉冲+1pct→铜+4%-6%滞后6-9个月），三因子合力对铜价净影响-1%至+2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观三因子共振情景4.3在2026 H2实现的概率不足15%，但温和的宏观顺风恰好证明当前基于物理约束的轮动策略具有不依赖流动性爆发的极高安全边际。", "summary": "截至2026-06-17，基于美国核心PCE粘性与非农就业温和降温，美联储2026 H2大概率仅进行1-2次预防性降息，无法驱使DXY<96和TIPS<1.5%；同时中国政策定力决定了信贷脉冲H2极难跃升至+2pct。三因子共振至情景4.3的概率极低（<15%）。但这反而确立了组合极高的安全边际：…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 1, "neutral": 0, "value": 75.2, "events": [ { "title": "2026 能源地缘政治风险对 AI 算力成本与全球通胀的传导评估", "summary": "支持卡片03：霍尔木兹封锁（卡塔尔3-4不可抗力，约20%全球LNG受阻[S1]）使全球气价飙升而美国Henry Hub脱钩走低——TTF +35%至14.8、JKM +51%至16.0美元/MMBtu，Henry Hub自2-28跌约9%[S1]，2026均价仅约3.5美元[S3]。我把它延伸到…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 6, "neutral": 0, "value": 46.6, "events": [ { "title": "电网公用事业资产费率基数扩张对宏观通胀与货币政策的传导效应", "summary": "压力测试卡片01–05的宏观结论。直接渠道被夸大：电力属能源、按构造排除于核心PCE之外，达拉斯联储测算数据中心对总体PCE的拉动2026仅+0.05pp、2030升至+0.13pp[S1]；电力仅占CPI约2.5%[S11]，自测算直接总体贡献约0.10pp。但费率基数是30–40年价格水平锁定…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国人民银行通过收窄利率走廊并创设境外回购、离岸外汇试点与非银流动性互换等三重防御机制，成功在经常账户顺差支撑下阻断了美联储“沃什新政”带来的跨境资金流出冲击，实现了国内增长与汇率稳…", "summary": "截至2026年6月20日，在美联储“沃什新政”推升美债10年期利率至4.48%的外部背景下，中美10年期国债利差倒挂至273BP。中国经常账户的高额顺差（2025年7350亿美元；2026年一季度1.28万亿元）支撑人民币汇率稳定在6.76-6.78。面对外资流出与国内宽松诉求，人行在陆家嘴论坛推…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储沃什新政与六月议息会议的“沟通机制变局”及其全球宏观传导", "summary": "2026年6月，美联储新任主席沃什首秀维持利率于3.5%-3.75%不变，但大幅上调PCE通胀预期至3.6%，并实施“取消前瞻指引”的沟通新政。此举推高美债10年期收益率至4.48%附近。同时，欧央行因中东地缘冲突升息25基点，日元兑美元跌至161.30的历史低位，套利交易拥挤度高企。全球货币政策…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储2026年6月利率决议与沃什新政启航", "summary": "在沃什主持的首次美联储会议上，政策利率维持在3.50%-3.75%，但SEP将核心PCE上调至3.3%，2026底利率中位数升至3.8%。沃什推行政策声明精简、未提交个人利率点并启动五大运行工作组。美元指数冲至100.73，美债收益率曲线上移，全球流动性面临重定价压力。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 11, "neutral": 0, "value": 18, "events": [ { "title": "2026-06-21：6月FOMC鹰派按兵不动与全球美元利率重定价", "summary": "截至2026-06-21，本卡选择global-macro，研究6月17日FOMC以12-0维持3.50%-3.75%但SEP转鹰后的宏观含义。结论是：这不是清晰的新加息周期，而是一次鹰派信誉重置；降息叙事已被移除，一次额外加息成为有条件风险。5月CPI、4月PCE、稳定的4.3%失业率、2年期美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘能源冲击下美联储与日本央行的同步鹰派转向——全球滞胀是否重来", "summary": "2026年6月主要央行同步收紧：美联储6月17日连续第4次维持联邦基金利率3.50%-3.75%，但点阵图把2026年底中值上调至3.8%、偏向加息（Warsh主席首会）；日本央行6月16日加息至1.0%，为1995年来最高。驱动因素是中东冲突与霍尔木兹海峡受阻带来的能源冲击：美国5月整体CPI同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储6月会议后的“降息交易”出清与再通胀风险", "summary": "工作日期2026-06-20。本卡选择global-macro，结论是美联储尚未承诺加息，但6月FOMC已把市场基线从自动降息转向higher-for-longer和加息风险。3.50%-3.75%的政策区间、2026年SEP中3.8%的年底利率中位数、5月CPI同比4.2%、5月非农+172,0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-20：美联储6月鹰派暂停与全球实际利率再定价", "summary": "截至2026-06-20，6月FOMC更像一次实际利率收紧冲击，而非中性暂停。美联储维持3.50%-3.75%目标区间，但SEP把2026年PCE通胀和联邦基金利率路径同步上调；H.15显示2年期、10年期和10年期TIPS收益率在会议日均上行。能源冲击仍是关键变量，EIA预计Brent在6月和7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 2, "risks": 12, "neutral": 4, "value": 41.4, "events": [ { "title": "能源通胀粘性背景下美联储政策路径与美元走势及利差交易风险", "summary": "截至2026-06-22，本卡支持卡片01–02的「鹰派维持」主线并延伸至汇率与套息。6月FOMC连续第四次维持3.50%–3.75%[S1]，新主席Warsh主持下点阵图改为2026年加息、PCE上修至3.6%[S2]，美元在DXY≈99企稳[S9]。PBOC将中间价定在约6.81、弱于约6.7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年伊朗战争能源冲击：供给扰动见顶与美联储/日本央行的滞后鹰派转向", "summary": "伊朗战争能源冲击大概率已过顶：布伦特从逾120美元回落至6月19日的80.57美元（周跌约8%），霍尔木兹有条件复航[S6][S8]。但美联储6月17日维持利率3.50–3.75%却将点阵图翻转为加息倾向（2026中值升至3.8%、PCE预测3.6%），日本央行6月16日加息至1.0%（1995年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年伊朗战争能源冲击重新定价全球利率周期：通胀是暂时性还是持久性", "summary": "持续15周的2026年伊朗战争制造了一次供给侧能源冲击，把发达市场利率周期从降息扭转为鹰派按兵不动+加息倾向：美联储6月17日维持3.50%–3.75%但点阵图翻转为加息（2026年底中值升至3.8%、PCE上调至3.6%），日本央行加息至1.00%，美国5月CPI同比4.2%创近三年新高。但布伦…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "伊朗战争后的全球宏观转折：能源冲击消退，但美联储与日本央行维持鹰派带来的滞胀宿醉", "summary": "2026年伊朗战争（2/28美以空袭、3/4霍尔木兹关闭）将布伦特推至逾120美元，6/17美伊谅解备忘录使油价回落至约80美元。但通胀已从能源迁至核心：5月美国CPI总体同比4.2%（三年新高），核心升至2.9%，居住与食品双双加速。央行反应函数已上移——美联储6/17维持3.50–3.75%但…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 3, "neutral": 0, "value": 68.8, "events": [ { "title": "中国仍能宽松，但在美联储高利率更久和中美利差深度倒挂下，宽松将以数量型、结构性和渐进式工具为主，并以防止人民币汇率超调为硬约束。", "summary": "截至2026-06-22，美联储高利率更久约束中国宽松节奏，但不是完全阻断货币政策自主性。当前SAFE数据仍显示2026年5月结售汇顺差USD 35.8 billion、非银行跨境净流入USD 62.6 billion，外汇储备为USD 3.442238 trillion；但中美短端利差约-223…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储 6 月再定价：美国前端利率从降息预期转向维持高利率并存在加息风险", "summary": "截至 2026-06-23，6 月 FOMC 将联邦基金目标区间维持在 3.50%-3.75% [S1]，并把 2026 年 PCE 与核心 PCE 预测分别上调至 3.6% 和 3.3% [S2]。5 月 CPI 同比 4.2% [S3]、4 月核心 PCE 同比 3.3% [S4]、5 月非农…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年6月美联储暂停、能源通胀与美国利率更久维持高位风险", "summary": "截至2026-06-22，本卡选择global-macro视角研究美联储6月暂停。核心发现是：FOMC虽维持3.50%–3.75%目标区间，但6月SEP上调2026年联邦基金、PCE和核心PCE预测；同时5月CPI能源项、4月PCE、5月就业与EIA油市预测共同削弱近期降息叙事。结论不是立即加息，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 2, "neutral": 0, "value": 76.2, "events": [ { "title": "美联储非对称紧缩把周期判断从“增长放缓触发宽松”改写为“通胀未再锚定前，增长疼痛不足以恢复Fed put”。", "summary": "截至2026-06-23，本卡支持卡片03：美联储已进入对通胀上行更敏感、对增长下行更迟钝的非对称紧缩体制。6月FOMC维持3.50%-3.75%，但SEP同时下调2026增长、上调PCE通胀；5月CPI 4.2%、4月PCE 3.8%、密歇根一年期预期4.6%，说明通胀预期仍未安全再锚定。OEC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘风险退潮与布油暴跌封顶美联储加息上限，但核心通胀粘性与4.51%的美债收益率维持高息高原，大类资产配置应战术超配低估值现金流红利、低配资源股。(2026-06-23)", "summary": "截至2026-06-23，美伊瑞士谈判取得进展，布油暴跌至$77.90/桶。供给侧地缘压力缓和虽降低了美联储因能源危机被迫加息的尾部风险，但在10Y美债收益率4.51%及核心PCE通胀预期3.3%的鹰派基准下，利率高原仍将延续。大类资产配置应超配防御性高股息（电信、公用事业），低配大宗商品资源股。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 0, "neutral": 1, "value": 92.6, "events": [ { "title": "2026年能源通胀冲击后的美联储政策", "summary": "截至2026-06-25，6月FOMC维持3.50%-3.75%并非鸽派信号。能源冲击已把5月CPI推至4.2%同比，但核心CPI仍为2.9%同比，劳动力和国内需求尚未断裂。基准判断是美联储更可能维持高利率而非马上加息；真正的分水岭是未来PCE/CPI能否证明能源没有传导到核心价格和通胀预期。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 1, "neutral": 1, "value": 88, "events": [ { "title": "沃什时代美联储首次议息会议：前瞻指引淡化与点阵图意外鹰派的宏观影响", "summary": "截至2026-06-26，分析显示沃什首场会议维持利率3.50%-3.75%不变，但因中东能源冲击，联储将2026年底点阵图上调至3.8%，核心PCE调至3.3%。这与欧央行和日银6月加息形成G3同步紧缩。然而，6月下旬海峡重开且布油跌至72美元，突显鹰派政策滞后。沃什淡化前瞻指引，将加剧美债收益…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美联储政策超调风险下的全球宏观经济周期与通胀路径演变", "summary": "截至2026-06-26，G3央行在6月同步偏鹰（美联储维持3.50%-3.75%但点阵图转向年内加息、欧央行存款利率上调至2.25%、日本央行加息至1%），却恰逢能源主导的通缩前沿：布伦特跌至约73.74美元/桶，较约105美元的油价假设低约30%，而5月核心PCE仍黏在3.4%、整体PCE破4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 6, "neutral": 2, "value": 62.4, "events": [ { "title": "2026年6月全球货币政策同步转鹰：美联储、欧洲央行、日本央行因伊朗战争能源冲击齐齐转向", "summary": "2026年6月，欧央行（6/11，存款利率升至2.25%）、日本央行（6/16，升至1.00%，1995年来最高）、美联储（6/17，维持3.50–3.75%但点阵图翻转至9名官员预期加息、2026中值3.8%）一周内同步转鹰，共同触发因素是伊朗战争带来的能源通胀（美国5月CPI 4.2%、能源同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球宏观分化：华许治下美联储鹰派转向与人民银行结构性流动性改革，及大宗商品冲击", "summary": "在2026-06-26，霍尔木兹海峡地缘局势缓解令Brent原油暴跌20%至$72-$75/桶。然而，历史性能源冲击拉高美国5月CPI至4.2%、PCE至4.1%，促使美联储在6月会议（凯文·华许首秀）中展现鹰派倾斜，半数委员预测2026年加息。与之相对，中国5月CPI低位运行（1.2%）而PPI…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI电力冲击本身是Fed应看穿的相对价格变化，但其持续、受供给约束且贴近核心服务成本的二阶传导，会让核心通胀在边际上更粘，并支持Fed更久维持限制性政策。", "summary": "截至2026-06-27，本卡综合前序结论：AI电力不是直接核心通胀项目，因为电力在2026年5月CPI-U权重仅2.495%，且core PCE剔除能源；但它可能通过商业电价、云服务、通信、医疗、设施与租金相关成本向60.158%权重的services less energy services传…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh 美联储鹰派首秀：在正在消退的霍尔木兹能源冲击中保留加息倾向", "summary": "2026-06-17 Warsh 首次主持 FOMC，利率维持 3.50%–3.75% 不变，但点阵图将 2026 年底中位利率从 3.4% 上调至 3.8%、删除宽松倾向，9/18 位委员预计年内加息 [S1][S2][S6]。触发因素是 5 月 CPI 同比 4.2%（2023 年 4 月以来…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 15, "neutral": 2, "value": 12, "events": [ { "title": "沃什联储6月鹰派转向：能源供给冲击下的'更高更久、加息可选'与全球外溢", "summary": "6月17日沃什首次主持的FOMC以12-0维持利率3.50%-3.75%，但点阵图转鹰：中值隐含2026年底约3.8%、9/18预计加息、17/18判断通胀上行，并取消前瞻指引。驱动因素是中东伊朗冲突引发的能源供给冲击——霍尔木兹受扰、布伦特一度较战前高逾60%，5月能源CPI同比+23.5%、核…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh 时代美联储鹰派转向：美国通胀再加速与 2026 年加息风险", "summary": "2026 年 6 月 17 日 FOMC（Warsh 首会）将利率维持 3.50%–3.75%，但点阵图转鹰：2026 年中值从 3 月的 3.4% 升至 3.8%，由隐含降息翻为隐含加息，18 位官员中 9 位点位高于中点、17 位判断通胀风险偏上行。背景是 5 月总体 PCE 同比 4.10%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美元 carry 仍有利差支撑，但在 DXY 高位后，USD/CNY 与 USD/JPY 的边际吸引力应从加杠杆顺势交易降级为带保护的战术 carry。", "summary": "2026-06-27：本卡支持前序“美元是约束性摆动变量”的判断。美联储维持 3.50%-3.75%，5 月 PCE 仍偏高，DXY 6 月 26 日为 101.3713。中美 2Y 利差约 +285.7 bp，美日 2Y 利差约 +268.1 bp，支持做多美元 carry；但 USD/CNY …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "再通胀压力下的跨资产配置策略：久期、商品与美元", "summary": "本卡支持卡片01的「韧性再通胀而非滞胀」判断，并落地为跨资产倾斜。盘面（2026-06-27）：5月PCE同比4.1%、核心3.4%，6月FOMC维持3.50%-3.75%且点阵图中位数升至3.8%；布伦特74.4美元（已较101.4美元峰值回吐27%），10年美债4.39%（7周低位）、2年4.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 12, "neutral": 2, "value": 29.2, "events": [ { "title": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "summary": "2026年6月17日Warsh首会以12-0维持利率3.50%-3.75%，但SEP中位点阵从隐含降息翻转为隐含加息（3.4%升至3.8%），18人中9人看至少1次加息、17人看通胀上行风险，2026年PCE上修至3.6%。背景是5月CPI同比4.2%反弹与韧性就业（年内约57万非农）。2年期美债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美元强势第二阶段与人民币/A股外资流传导", "summary": "本卡综合前序政策失误论证，落到汇率与资金流通道。美元处于2026年往返的第二阶段：DXY自1月95.57谷底反弹、议息后破100、6/25-26创13个月新高约101.37，但该段几乎全靠约75%的9月加息定价支撑，结构脆弱。USD/CNY约6.80为美元主导。关键在拆解美元下跌动因：套利通道（预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储2026年6月鹰派转向：新主席Warsh首会下点阵图翻转为加息，叠加中东能源冲击", "summary": "2026年6月17日FOMC维持利率3.50%–3.75%，但点阵图从隐含降息翻转为隐含加息：2026年底利率中值由3.4%升至3.8%，18人中17人判通胀风险偏上行，整体PCE预测上调至3.6%、核心3.3%。新主席Warsh以「五年失误必须纠正」的信誉优先基调定调，2年期收益率当日跳升约16…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh 时代美联储 2026 年 6 月的鹰派范式切换及其对美元、全球利率与中国政策空间的外溢", "summary": "2026 年 6 月 17 日 FOMC 名为按兵、实为范式切换：新主席 Warsh 删去宽松倾向，点阵图由隐含降息翻转为隐含加息（18 人中 9 人预计 2026 年至少加息一次），利率维持 3.50%–3.75%。市场即刻定价：2 年期美债收益率跳升约 16 个基点至约 4.21%（逾一年最高…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 1, "neutral": 0, "value": 90.6, "events": [ { "title": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "summary": "截至2026-06-29，支持卡片01–03的逻辑链并延伸至终端利率与全球流动性维度。美联储6-17维持3.50%–3.75% [S2]、SEP 2026中值3.8% [S3]；新主席Warsh首场FOMC点阵图从降息翻转为加息，9名官员预期加息、年内加息概率50.4%、市场定价10月加息，反应函…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-a803fa3312/", "networkHref": "/investor/research-network/#thesis-a803fa3312" }, { "id": "thesis-08dd090dc9", "source": "tracked_thesis", "title": "公用事业从防御红利切换到电网扩容、容量费与信用再定价", "titleEn": "Utilities are shifting from defensive dividends to grid expansion, capacity fees, and credit repricing", "lane": "公用事业/电网信用", "laneEn": "Utilities, grid rate base, and credit", "laneHref": "/investor/lanes/utilities-grid-rate-base-and-credit/", "investableFocus": "公用事业/电网信用：600900.SH, 600025.SH, 600905.SH, DUK, SO, AEP, NEE, XLU", "stockUniverse": [ { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "600025.SH", "name": "华能水电", "market": "A-share", "role": "水电现金流", "roleEn": "水电现金流", "bucket": "core", "href": "/investor/stocks/600025.SH/" }, { "ticker": "600905.SH", "name": "三峡能源", "market": "A-share", "role": "新能源运营商", "roleEn": "新能源运营商", "bucket": "core", "href": "/investor/stocks/600905.SH/" }, { "ticker": "DUK", "name": "Duke Energy", "market": "US", "role": "受监管公用事业", "roleEn": "受监管公用事业", "bucket": "core", "href": "/investor/stocks/DUK/" }, { "ticker": "SO", "name": "Southern Company", "market": "US", "role": "受监管电力与核电", "roleEn": "受监管电力与核电", "bucket": "core", "href": "/investor/stocks/SO/" }, { "ticker": "AEP", "name": "American Electric Power", "market": "US", "role": "输配电与 rate base", "roleEn": "输配电与 rate base", "bucket": "core", "href": "/investor/stocks/AEP/" }, { "ticker": "NEE", "name": "NextEra Energy", "market": "US", "role": "可再生公用事业对照", "roleEn": "可再生公用事业对照", "bucket": "core", "href": "/investor/stocks/NEE/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" } ], "stockTickers": [ "600900.SH", "600025.SH", "600905.SH", "DUK", "SO", "AEP", "NEE", "XLU", "510300.SH", "588000.SH", "512100.SH", "2800.HK", "MCHI", "KWEB" ], "coreView": "赛道：公用事业/电网信用。候选观察标的：600900.SH, 600025.SH, 600905.SH, DUK, SO, AEP, NEE, XLU。当前共有 40 条证据、4 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 76, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 15, "risks": 132, "neutral": 13, "analystCount": 12, "sourceTypeCounts": { "研究笔记": 155, "研究主题": 3, "研究线程": 2 }, "topTerms": [ "电网", "PPA", "长江电力", "公用事业", "AIDC", "新能源", "运营商", "电力", "水电", "红利", "核电", "能源" ], "whatChanged": [ "2026-06-29: 在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15.36倍AI潜在电力需求，算力扩张将受限于刚性电力配给，投资配置应向短交期电力变压器出口链与物理红利安全垫倾斜。", "2026-06-29: AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利景气度，A股资产配置应向电网出海及水电红利哑铃端重构[自测算]。", "2026-06-29: 选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "2026-06-29: 变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "2026-06-29: 在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红利（长江电力、中国核电）进行防御性轮动，并辅以日元和美债的尾部风险对冲。" ], "practical": { "sectorLane": "公用事业/电网信用", "sectorLaneEn": "Utilities, grid rate base, and credit", "exposureBucket": "中国权益与流动性", "tone": "green", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "600025.SH", "name": "华能水电", "market": "A-share", "role": "水电现金流", "roleEn": "水电现金流", "bucket": "core", "href": "/investor/stocks/600025.SH/" }, { "ticker": "600905.SH", "name": "三峡能源", "market": "A-share", "role": "新能源运营商", "roleEn": "新能源运营商", "bucket": "core", "href": "/investor/stocks/600905.SH/" }, { "ticker": "DUK", "name": "Duke Energy", "market": "US", "role": "受监管公用事业", "roleEn": "受监管公用事业", "bucket": "core", "href": "/investor/stocks/DUK/" }, { "ticker": "SO", "name": "Southern Company", "market": "US", "role": "受监管电力与核电", "roleEn": "受监管电力与核电", "bucket": "core", "href": "/investor/stocks/SO/" }, { "ticker": "AEP", "name": "American Electric Power", "market": "US", "role": "输配电与 rate base", "roleEn": "输配电与 rate base", "bucket": "core", "href": "/investor/stocks/AEP/" }, { "ticker": "NEE", "name": "NextEra Energy", "market": "US", "role": "可再生公用事业对照", "roleEn": "可再生公用事业对照", "bucket": "core", "href": "/investor/stocks/NEE/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" } ], "stockTickers": [ "600900.SH", "600025.SH", "600905.SH", "DUK", "SO", "AEP", "NEE", "XLU", "510300.SH", "588000.SH", "512100.SH", "2800.HK", "MCHI", "KWEB" ], "primaryStocks": [ { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "600025.SH", "name": "华能水电", "market": "A-share", "role": "水电现金流", "roleEn": "水电现金流", "bucket": "core", "href": "/investor/stocks/600025.SH/" }, { "ticker": "600905.SH", "name": "三峡能源", "market": "A-share", "role": "新能源运营商", "roleEn": "新能源运营商", "bucket": "core", "href": "/investor/stocks/600905.SH/" }, { "ticker": "DUK", "name": "Duke Energy", "market": "US", "role": "受监管公用事业", "roleEn": "受监管公用事业", "bucket": "core", "href": "/investor/stocks/DUK/" }, { "ticker": "SO", "name": "Southern Company", "market": "US", "role": "受监管电力与核电", "roleEn": "受监管电力与核电", "bucket": "core", "href": "/investor/stocks/SO/" }, { "ticker": "AEP", "name": "American Electric Power", "market": "US", "role": "输配电与 rate base", "roleEn": "输配电与 rate base", "bucket": "core", "href": "/investor/stocks/AEP/" } ], "instrumentTags": [ "600900.SH", "长江电力", "600025.SH", "华能水电", "600905.SH", "三峡能源", "DUK", "Duke Energy", "SO", "Southern Company", "AEP", "American Electric Power" ], "reviewCadence": "每日复核", "confirmationSignals": [ "融资余额、ETF 申赎和北向资金是否同向", "被动调仓窗口是否被主动资金提前交易", "订单、盈利和估值是否同步改善" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "指数期权", "红利低波", "汇率", "两融降温" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 160, "sourceDiversity": 3, "analystCount": 12, "recentEvents": 114, "supportRatio": 0.09, "riskRatio": 0.82, "conflictScore": 91.7, "evidenceDepth": 100 }, "score": 69.9 }, "events": [ { "date": "2026-06-29", "title": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红利（长江…", "summary": "本报告针对2026年三季报TMT硬件估值折价预期，校准了全球全天候和风险平价模型。战术防御性风险平价（TDRP）将AI成长股风险预算由25.0%下调至8.0%，防守红利上调至40.0%，以对冲55.0%的TMT增长溢价折价。模型显示，电力配给约束下算力效率的TCO优势在24/7清洁能源下扩大至9.97%，高功耗算力中心毛利率面临负增长。此外，建议红利套组内…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 41.5, "matchedTerms": [ "电网", "长江电力", "公用事业", "电力", "红利", "核电", "能源", "rate" ], "href": "" }, { "date": "2026-06-29", "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优先既有水电/核电现金流、具备瓶颈定价权的变压器和电网设备供应…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.5, "matchedTerms": [ "电网", "公用事业", "运营商", "电力", "水电", "红利", "核电" ], "href": "" }, { "date": "2026-06-29", "title": "选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "summary": "截至2026-06-29，本卡综合前9卡：AI电力约束是选择性时序上限而非订单毁灭。建议跨资产账本配置为40%受资助AI基础设施龙头、45%主权背书红利公用事业、3%年度申万电子Put权利金预算、12%流动性储备。进攻端以中际旭创18%、工业富联14%、英维克8%为核心；防御端以水电18%、核电17%、受监管电力/电网10%承接电力稀缺；Put以35-45…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.5, "matchedTerms": [ "电网", "公用事业", "电力", "水电", "红利", "核电", "rate" ], "href": "" }, { "date": "2026-06-29", "title": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利…", "summary": "本报告对AI算力激增带来的电力电网物理瓶颈（变压器交期120-210周[S2]、并网排队4-7年[S1]）对A股转型龙头的利润率传导进行压力测试。结果显示，电网扩容滞后充当了新能源装机的“限速阀”，压制了光伏/电池销量增速；同时，容量电价[S8]和碳成本上升[S19]增加了高能耗制造环节电费开支，而在严重过剩背景下成本转嫁失败，导致制造端转型龙头利润率遭到…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.5, "matchedTerms": [ "电网", "长江电力", "新能源", "电力", "水电", "红利", "能源" ], "href": "" }, { "date": "2026-06-29", "title": "在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15.36倍AI潜在电力需求，算力扩张将受限…", "summary": "截至2026-06-30，本研究对Jevons悖论激发的15.36倍潜在AI用电需求（相当于727.6 GW新增平均负荷或6,374 TWh年用电）进行了极限物理压力测试。结论表明，在2030年前，由于北美变压器交期拉长至48-60个月、特高压直流送电网架高壁垒（全球需新建73条通道）以及取向硅钢（GOES）产能与高压试验大厅测试容量的物理刚性制约，没有任…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32.75, "matchedTerms": [ "电网", "长江电力", "电力", "红利", "核电", "Utilities" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "summary": "截至2026-06-28，本卡对“AI负荷增长能否支撑公用事业高分红”作压力测试：结论是精选组合层面可以，板块整体不能机械外推。AI数据中心用电增长真实，但也同步拉动电网、发电和储能资本开支。水电、核电、容量电价支持的火电及低杠杆综合电力运营商仍可维持分红；风光开发商受消纳、电价和项目资本开支制约，不能直接视为高息资产。上一卡的25%公用事业红利仓可以保留…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 42.75, "matchedTerms": [ "电网", "公用事业", "运营商", "电力", "水电", "红利", "核电", "Utilities" ], "href": "" }, { "date": "2026-06-28", "title": "物理基础设施约束下多资产投资组合的重平衡与防守哑铃配置策略", "summary": "本卡支持卡片05–07链条，把物理基础设施约束落地为多资产重平衡：构建防守哑铃。削减超估值且拥挤的AI硬件成长腿（−12pp至20%并加期权领口），把仓位转入物理基础设施租金腿（公用事业含水电核电、电网重电与变压器、水权、液冷热管理，+14pp至22%）与高股息压舱（中证红利约4.68–4.9%股息率，超10年期国债1.73%约290bp）。中间以久期、黄…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "电网", "公用事业", "水电", "红利", "核电", "rate" ], "href": "" }, { "date": "2026-06-28", "title": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费成本转嫁挤压。", "summary": "基于国产芯片2.94倍能效赤字拉长通电时间至18个月的物理瓶颈，本报告系统性量化了A股电力设备板块（特高压、大变压器）的估值扩张机制与低利润 traditional TMT 的电价挤压。电网容量受限与十五五规划共振，支持了特高压订单可见度与大型变压器出海高溢价，特别是在国内外硅钢5.1倍巨大价差的成本红利下。策略上构建杠铃组合，超配直接出口变压器龙头、特高…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "电网", "公用事业", "电力", "水电", "红利", "rate" ], "href": "" }, { "date": "2026-06-27", "title": "在通缩压力下，公用事业并非均质的防御锚；必须坚守‘电网优先’与具备正向国债利差的可归因费率资产，警惕无费率保护的算力PPA延迟以及高溢价新能源标的的估值陷阱。", "summary": "基于2026-06-28的宏观及行业数据，本报告对公用事业作为杠铃组合“防御锚”的稳定性进行了压力测试。美股算力PPA面临并网延迟与BTM核电监管壁垒（如FERC Talen裁决）的严重制约，极端情景下BTM并网被禁可致核电项目EBITDA转负（-9.48百万美元，DSCR -0.17x）。A股“电网优先”组合（PE 17.14x，股息率与电网开支可见度高…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.75, "matchedTerms": [ "电网", "PPA", "公用事业", "新能源", "水电", "红利", "核电", "能源" ], "href": "" }, { "date": "2026-06-27", "title": "能源与电力板块（长江电力/中国海油）的防御性估值与政策风险压力测试", "summary": "对卡片09把半导体腾出资金转入长江电力(600900)与中国海油(0883)防御的安排做压力测试，得出分裂裁决。电力/长江电力基本面不过热（零燃料成本、70%分红锁定至2030），但作为工具已成拥挤的26.8年久期债券替代品：股息率压至3.5%、对国债利差约2pp处历史高位、内在价值≈市值、贴现率+100bp回撤约-21%——在通胀再加速/加息根主题下恰是…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32.5, "matchedTerms": [ "电网", "长江电力", "电力", "水电", "红利", "核电", "能源" ], "href": "" }, { "date": "2026-06-27", "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌破3.0%的红利反向出清防线。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "长江电力", "电力", "水电", "红利", "能源", "rate" ], "href": "" }, { "date": "2026-06-26", "title": "电力电网及公用事业红利资产的配置逻辑验证", "summary": "支持第01张卡片的防守再平衡判断，并验证其机制：本轮去杠杆是两融融资拥挤的电子板块冲击（A股两融2026-06-23破3万亿元），而受监管公用事业由保险/养老等长钱持有、融资拥挤度低，处于压力源的对立端，是真正避风港而非平行泡沫。需求端1-5月用电+3.4%、互联网数据服务用电+26%给出AI电气化成长期权；股息端长江电力约3.6%-3.8%对10年期国债…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 38.75, "matchedTerms": [ "电网", "长江电力", "公用事业", "电力", "水电", "红利", "核电", "Utilities" ], "href": "" }, { "date": "2026-06-26", "title": "136号文市场化电价下绿电、核电、水电和储能运营商现金流敏感性", "summary": "对运营商做三降压力测试：绿证价格、利用小时、市场交易电价同时下行。2025年业绩已是实盘验证，横截面决定性。保留满仓——长江电力(600900)水电、中国核电(601985)核电：电价下行年净利仍增(+6.17%/+6.00%)，水电近零边际成本+65.79%毛利、来水定产不受限电；核电>7500h基荷免疫限电，经营现金流稳。降权有条件——独立储能：114…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 37.75, "matchedTerms": [ "长江电力", "运营商", "电力", "水电", "核电", "能源", "Utilities" ], "href": "" }, { "date": "2026-06-26", "title": "A股市场结构约束下的清洁基础设施与能源对冲执行方案", "summary": "截至2026-06-26，本卡把Card03杠铃(12%清洁+4%能源)与Card04子仓上限落地为可成交的A股方案。核心结论：约束不难，难在表达载体可得性高度不对称——电网有唯一高流动纯标的(159326,约¥343.7亿,特高压权重约66%)；电池有百亿级但偏电动车(159755,约¥158.9亿)；绿电ETF含约34%火电(562550)污染运营商口…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.5, "matchedTerms": [ "电网", "长江电力", "运营商", "电力", "核电", "能源", "rate" ], "href": "" }, { "date": "2026-06-26", "title": "保留12%清洁+4%能源杠铃，但A股表达必须从宽基清洁贝塔转为受限的电网/储能/运营商组合，并用能源对冲抵消价值和通胀尾部而非追高红利能源。", "summary": "截至2026-06-26，本卡综合前序12%清洁基础设施+4%能源尾部对冲，但指出A股执行并非因子中性。清洁袖主要暴露于成长、资本开支、低股息和动量，尤其电网设备、电池、储能、光伏存在追涨和集中度风险；能源袖提供价值、高股息和通胀/霍尔木兹尾部保护，但中证能源前五权重约61.0%，不能无限扩张。建议保留杠铃，同时设置4.0%电网上限、4.0%储能/电池/光…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "电网", "运营商", "水电", "红利", "核电", "能源" ], "href": "" }, { "date": "2026-06-26", "title": "运营商精选提升了清洁袖质量，但也把它变成权益久期资产；全天候组合应以总防御久期预算管理清洁运营商和债券袖，而不是靠加大能源对冲来掩盖利率叠加。", "summary": "截至2026-06-26，本卡综合前序结论：清洁袖收敛到长江电力/中国核电后，不应再被当作独立清洁成长仓位，而应纳入与长债共享的防御久期预算。长江电力是真正70%+分红资产；中国核电现金流债性强但最新披露分红比例约39%。建议运营商清洁核心限制在4%–5% NAV，超过部分必须同步缩短债券袖久期。4%能源对冲不机械上调，应维持上游油气属性，仅在Brent>…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 31.5, "matchedTerms": [ "长江电力", "运营商", "电力", "核电", "能源", "rate" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "DeepSeek V4 抛压是 TMT 龙头疲劳、非周期终结;资金部分迁向股息属再平衡而非反转,A 股应采取核心 TMT + 确定性股息的杠铃配置。", "summary": "DeepSeek V4 后的抛压是龙头疲劳信号,而非 TMT 周期终结。盈利分化仍在拉大、云资本开支指引未破、A 股 TMT 拥挤度未达极端、宏观仍宽松,五条证据均指向中段而非末段。资金确实在向确定性股息(中证红利 5.5–6%、四大行 5–6.5%、能源三桶油 6–8%)迁移,但这是再平衡而非反…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-04", "supports": 1, "risks": 0, "neutral": 0, "value": 51.4, "events": [ { "title": "能源 Capex 对科技巨头主要是 FCF 估值折价而非偿付能力危机；长期电力合约则让公用事业公司获得可融资、可证券化的基础设施现金流。", "summary": "本卡片综合前序判断：200 亿美元电网升级和长期核电 PPA 不会压垮科技巨头资产负债表，但会显著改变 FCF 估值框架。Microsoft、Alphabet、Meta 仍能以内生现金流支持 AI 电力基础设施；Amazon 最敏感，Q1 2026 TTM 经营现金流 148.5bn 美元、现金 …", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 0, "neutral": 1, "value": 47.8, "events": [ { "title": "AI基础设施正经历‘公用事业化’重估，估值驱动力从智能稀缺性转向物理交付（能源、电网）的稀缺性，导致超大规模厂商与优质电力资产的估值倍数趋同。", "summary": "2026年5月，美股TMT板块估值逻辑发生重大转向：超大规模厂商（Hyperscalers）因资本密集度飙升（约25%）和FCF收益率受压（2.5%-4%）正被重塑为‘AI公用事业公司’。Token价格的掠夺性定价（降幅达97%）使‘能源套利能力’成为核心护城河。持有核电PPA的厂商利用成本优势挤…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 0, "neutral": 1, "value": 49.2, "events": [ { "title": "二级市场正完成从RAB估值向容量稀缺性估值的范式转移，核电IPP已成为AI算力的核心能源护城河。", "summary": "2026年5月，公用事业估值体系发生根本性脱钩。独立核电发电商（IPP）因其“表后直连（BTM）”能力及未分配容量的稀缺性，被二级市场作为AI基础设施代理资产进行重估，P/E扩张至28x-35x。相比之下，受规制公用事业虽有高额电网投资支撑，但因核电基荷流失导致的电网波动性上升而面临“可靠性折价”…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "2027–2028 年公用事业稀缺租金属于同时握有排队号、变压器与可核查碳账本的环节——电网设备 OEM 与受监管 T&D 是最高确信度超配。", "summary": "支持 Research note 06 的政策综合判断：2027–2028 年公用事业 alpha 来自并网权、CBAM 可核查碳账本与可靠容量权益，而非新增装机 MW。第一梯队是订单排至 2028 年的高压电网设备 OEM（GEV、Siemens Energy、Hitachi Energy、Pr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "公用+电信SOE作为红利压舱石的判断成立，但应集中水电/核电+移动/电信H股、最小化纯火电与联通，整桶红利组合质量与收益率同步上行。", "summary": "截至2026-05-15，公用与电信SOE的红利防御属性整体成立但内部分化明显：水电（长江电力）、核电（中广核、中国核电）和移动/电信（H股股息率6.0–7.0%）现金流稳定、派息政策正面，构成红利压舱石的最干净底盘；纯火电IPP在秦皇岛煤价RMB 690–760/吨与容量电价(2026预计~RM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 2, "risks": 2, "neutral": 0, "value": 54.4, "events": [ { "title": "T2 篮子 H2 2026 应采用三层杠铃：用业绩兑现锁定 alpha，用 AI-PPA 期权布局再定价，用重装备保留二阶弹性，而非整体超配 T2。", "summary": "本卡（2026-05-17）将 T2「能源获取复合体」H2 2026 配置定型为三层杠铃：A 现金流核心 50-60%（NARI、平高、许继、特变），B 政策期权 20-25%（长江电力、中国核电、中国广核），C 主题期权 10-15%（东方电气、上海电气、哈尔滨电气），现金/对冲 5-10%。鉴…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI基础设施重估的最可投线索不是泛电力，而是2026-2028年可把西部和东部枢纽负荷真正点亮的电网交付设备链。", "summary": "截至2026-05-17，本卡综合支持T2能源获取复合体逻辑，但把约束从全国电量改写为本地并网交付。主要东数西算枢纽AI级可交付并网容量约4.6-5.4 GW，2026Q4约7.2 GW，2027Q4约10.0 GW，2028Q4约12.7 GW。绑定瓶颈依次是220/110 kV变电站与控制系统…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "B层应以核电基荷PPA为主线、以水电防御为底仓，按601985 > 003816 > 600900配置，而不是把A层释放仓位平均分配给三只公用事业。", "summary": "截至2026-05-17，本卡综合支持A层释放10-15pp转入B层，但B层不是平均配置，而是AI-PPA披露期权组合。建议B层占T2篮子30-35%，中枢32.5%：中国核电601985 12.5%、中国广核003816 11.0%、长江电力600900 9.0%。在25%/50%/75%披露概…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "估值范式不是从算力到电力的单向翻转，而是A股AI链的三层分层——核心alpha在T2能源获取复合体（公用事业+特高压+核电+重装备），重估扳机是AI专属电价分类与首例AI-PPA披露。", "summary": "支持前序结论但拒绝二元翻转：正确框架是三层分层。T1（算力/硅片）从40-60×降档至25-35×成长周期估值；T2（能源获取复合体——拥有AI-PPA的公用事业、特高压电网设备、燃气轮机、核电链）由12-15×债券替代重估至18-25×AI-PPA股权代理，是新核心；T3（应用/token消费方…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "在油—煤—气联动重新走高的环境下，'公用事业=防御'只在水电+核电+水务+央企红利窄篮子口径下成立；火电与城燃利润率反而被压缩，把它们打包成防御资产将让组合的油价beta显著为负。", "summary": "对Research note 02的部分压力测试：'水电与受监管公用事业作为防御红利'的口径只对了一半。Brent站上104美元、煤炭与JKM LNG联动重新走高的环境下，水电（长江电力600900等）、核电（中国核电601985）、水务保持防御性，但火电（华电国际600027、华能国际60001…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 1, "risks": 0, "neutral": 0, "value": 61.2, "events": [ { "title": "2026H2公用事业应采用45/35/20的杠铃配置：以大市值电力守股息底座，以区域能源平台买入经4-6折折扣后的AI灵活性期权，以市政环保整合者分散现金流修复alpha。", "summary": "截至2026-05-18，本卡综合支持前序“区域能源平台重估”方向，但将AI灵活性价值从基础现金流下调为期权现金流。2026H2公用事业组合建议从泛化高股息重排为45%大市值类债电力股、35%区域能源平台、20%市政环保整合者。核心理由是：股息底座仍由水电、核电和容量电价支撑；区域平台可捕获AI/…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 1, "neutral": 0, "value": 56.6, "events": [ { "title": "能源基荷稀缺性溢价会转化为类特许经营权收益，但只有已并网、可认证、可交付且能签署PPA或进入高容量价格市场的存量资产能充分捕捉。", "summary": "截至2026-05-20，本卡支持卡07：2026-2028年低碳基荷配置收益主要来自存量稀缺资产，而非绿地扩张。稀缺性通过三条路径进入利润：merchant市场的LMP、容量和辅助服务上行；hyperscaler长期PPA把time-to-power、24/7清洁属性和节点交付能力资本化；受监管…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI 数据中心带来的是 firm-power 稀缺溢价，2028 年前的可投资赢家是现有核电/燃气/输电控制者，而不是把 SMR 叙事提前到本轮缺口的公司。", "summary": "截至2026-05-19，本卡综合支持上一张能源瓶颈判断，但把结论细化为：稀缺资产不是抽象电力，而是负荷节点上的 firm、deliverable、contractable power。美国最强议价权集中在 CEG/Calpine、VST、TLN 的 Susquehanna 节点，以及 Domin…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 1, "risks": 2, "neutral": 0, "value": 52, "events": [ { "title": "主权电力资产可以承担太空杠铃的防御端，但应优先水电/核电现金流锚，限制高杠杆新能源开发商权重。", "summary": "截至2026-05-21，本卡支持把主权/国资电力资产纳入太空杠铃防御端，但要求按现金流质量而非股息率排序。长江电力和中广核电力更接近核心防御锚，华润电力和中国电力提供更高收益与转型敞口但资本开支、杠杆和市场电价风险更高。样本股息率约3.67%-5.64%，相对中国10年期国债1.75%有明显利差…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "各州零售电价保护壁垒、FERC表后净额结算封堵以及变压器等下游设备交期的“三重屏障”，将严重滞缓AI电力红线的物理释放，并大幅压缩发电巨头的表后溢价红利。", "summary": "2026年5月21日，公用事业分析师对AI电力的乐观增长假设进行压力测试。分析指出，各州监管出台了弗吉尼亚州GS-5专用电价类等条款以保护居民用户；FERC在2025年12月彻底封堵了表后直供（BTM）免过网费的净额结算漏洞，使得核电直供数据中心到岸电价攀升超22%；叠加电网物理互联排队超55个月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A 股两条并行合规网格——金属碳轨与医药合规轨——结构同源但机制不同，正确响应是沿两条清洁轨道做杠铃配置，而非搜寻同时具备两种期权的复合标的。", "summary": "本卡为 10 卡线程的终结综合。截至 2026-05-21，金属碳轨（CBAM、ETS、Scope 2/3）与医药合规轨（BIOSECURE、FDA cGMP、MDR/IVDR、NRDL）构成两张并行合规网格：结构同源、机制不同——碳轨可证券化、可逆性高、12–24 月兑现；合规轨以声誉与客户关系…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 4, "neutral": 0, "value": 33.4, "events": [ { "title": "水电跨省长协壁垒与核电单向价差合约（CfD）提供了极强的基本面防御力，即使在标准煤价暴跌-30%的极端悲观情景下，四大公用事业巨头的净利润侵蚀依然被牢牢控制在-5.00%以内，分红潜…", "summary": "基于2025年实际数据对电力联动与煤价下跌对水电与核电利润的蚕食效应进行了极值压力测试。研究发现，在标准煤价暴跌-30%的极度悲观预期下，因水电低市场化长协锁定与核电90%比例单向价差合约（CfD）托底，四大电力巨头的归母净利润侵蚀幅度均被牢牢控制在-5.00%以内。其中长江电力仅为-3.92%，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力市场化深化对水电核电长协是温和侵蚀而非崩塌,中性悲观情景下政策让利路径净利侵蚀(-4.9%至-7.7%)已超过燃料价格路径,卡片02压舱仓位中枢维持但应预留-2pp政策缓冲。", "summary": "本卡对卡片02—04的核心防御结论做政策视角压力测试。2026-05-22时点,136号文(新能源全面入市)、煤电容量电价、跨省送电基准+浮动、绿电直供四条改革线索叠加,正在对水电核电长协形成温和而非崩塌式的让利压力。自测算显示:在中性悲观情景(跨省协议让3分钱+现货占比抬至20%+工商业让利)下…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核电与水电凭借年金化现金流、制度化分红与极低两融拥挤度,是当前A股最合格的红利价值防御底座,可在卡片01给出的59%红利仓位中担纲12%-15%的硬核压舱石。", "summary": "在卡片01关于微盘/北证50资金分母收缩与红利价值切换的判断之上，本卡对核电与水电的防御效能进行了量化验证。三维评估表明：水电(长江电力)与核电(中国核电/中广核)具备(1)经营性现金流毛利率65%+ / 48%+的现金流年金特性；(2)分红率50%-70%/35%-45%的制度化承诺；(3)两融…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融资金去杠杆已通过账户交叉质押与VaR风控限制从TMT扩散至国防军工、新能源等高拥挤成长板块，必须通过‘分化杠铃’策略将大仓位配置于长江电力等红利安全岛、将科技仓位战术收拢至受国家…", "summary": "本卡片系统综合了A股去杠杆负螺旋对全市场的跨行业传染风险。我们测算指出，散户账户交叉质押与VaR风控触发将两融出清压力从TMT扩散至国防军工、新能源等高拥挤成长板块。科创50指数1,750点关口面临882亿元密集敲入与平仓卖压，但在国家队专项买盘防护下形成有底成长资产。我们建议采取‘分化杠铃’策略…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 1, "risks": 0, "neutral": 0, "value": 66.8, "events": [ { "title": "跟踪稀土与核能板块资金推动系列研究（七）：合同型电力重新分类后的行业与风格执行 报告发布日期：2026 05 23 研究团队：首席策略师 (Chief Strategist) 核心摘…", "summary": "跟踪稀土与核能板块资金推动系列研究（七）：合同型电力重新分类后的行业与风格执行 报告发布日期：2026 05 23 研究团队：首席策略师 (Chief Strategist) 核心摘要 随着人工智能数据中心对高确定性、无碳基荷电力需求的爆发性增长，电力及公用事业板块的估值逻辑正经历历史性解构。截至…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 3, "neutral": 2, "value": 48.2, "events": [ { "title": "A股高股息电力不是主动机构拥挤交易，但已成为被动红利资金拥挤载体，合理股息率边界约3.25%-4.25%，低于3.0%的核电/水电叙事资产尾部风险最大。", "summary": "截至2026-05-24，本卡压力测试A股高股息电力公用事业。结论是：板块存在ETF与战术红利资金拥挤，但主动权益基金并不拥挤，2026Q1公用事业配置仅0.45%，核电仅0.03%。容量电价提高煤电分红稳定性，10年期国债1.75%对应合理股息率走廊约3.25%-4.25%。长江电力3.77%仍…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "容量电价机制虽显著平抑了基荷及电网资产的折旧敞口，但公用事业板块的‘成长股化’重估已过度透支AI用电需求，极易受算力资本开支周期放缓、表后直连监管收紧以及现货电价通缩的三重冲击。", "summary": "截至2026-05-24，本卡分析了美中电力容量机制对公用事业估值重塑的影响。PJM容量价暴涨9倍至$333.44/MW-day封顶，中国煤电容量回收比于2026年上调至50%-70%。这能有效弥补重资产存量折旧，但美股核电/IPP的30倍高估值面临AI算力需求周期放缓和表后直连核电直供电监管收紧…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电网消纳压力与配电网升级：财政紧缩背景下“消纳为先”模式的稳定性挑战 工作日期： 2026 05 25 (Asia/Singapore) 分析师： 公用事业分析师 [codex] 立…", "summary": "电网消纳压力与配电网升级：财政紧缩背景下“消纳为先”模式的稳定性挑战 工作日期： 2026 05 25 (Asia/Singapore) 分析师： 公用事业分析师 [codex] 立场： 支持 (Support) 执行摘要 截至2026年5月，全球能源转型撞上了物理层面的“电网墙”。在“2025年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面对日元套利平仓引发的全球去杠杆，A股是新兴市场内次坏的防御港，但承接资金的仅是国家队锚定的高股息低β国企篮子；指数总盘和北向重仓成长股仍是高β出口通道，必须主动回避。", "summary": "在第5张卡的10%日元套利平仓基准情景下，A股属于部分隔离但非免疫，正确姿态为杠铃式防御。承接防御资金的主篮是国家队β≤1.0、股息率≥4%、在岸资金主导的标的：国有大行、电信三大、长江电力/核电、高速公路与基础设施REITs、A股油气煤龙头；这些前十大股东中保险+社保+汇金占自由流通股55-60…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-25", "supports": 1, "risks": 2, "neutral": 1, "value": 60.6, "events": [ { "title": "AI算力是2026–2030电力系统首要结构性需求驱动,定量补强'公用事业由防御收息转向结构性成长'的重定价判断,焦点应放在现货敞口+稳态电源+电网超级周期。", "summary": "AI算力建设是本十年全球电力系统最大可识别的需求侧冲击:IEA基准下2030年前全球数据中心用电增量约+530 TWh,美国2028年占比升至6.7%–12%,中国规划2030年~400 TWh,且高度集中在北弗吉尼亚、ERCOT、东数西算枢纽等少数节点。PJM 2025/2026容量拍卖出清$2…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "美国公用事业重估的核心在于发电容量与并网一次变压器硬件瓶颈，而中国重估的核心在于跨省特高压直流通道与沿海清洁基荷运营商。", "summary": "本卡将中美AI负载电力缺口映射至具体公用事业与电网装备投资链。美国交易“稳态电源缺口（Electrons）”，受益于核电BTM直供（CEG、VST、TLN）及变压器/开关柜瓶颈重估（GEV、ETN、POWL，LPT交期达128-160周）；中国交易“输网地理错配（Wires）”，受益于国网“十五五…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "占全天候组合30%权重的WLEF（公用事业红利）资产，依靠每年1.35%的内生派息流与刚性电价机制，构成了完全覆盖美债极端汇率追保缺口的结构性流动性锚点。", "summary": "截至2026-05-26，本卡从公用事业基本面综合了极高久期美债尾部风险与 AI 能源墙的对冲闭环。占组合30%的WLEF（核电/水电）不仅抗贴现率冲击（压力回撤-5%），其4.5%的高股息率每年更内生创造1.35%的自由现金流，超额覆盖（112.5%）人民币急升10%引发的1.2%极高久期美债追…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026Q3应持有“利润留存中国”并对冲“重定价中国”：自动化和现金流防御资产增持，新能源硬件、材料beta和科创拥挤动量减持，基准对冲比率0.45x。", "summary": "截至2026-05-25，本卡综合1-7卡后认为：负Gamma平仓真空之后不应简单卖出全部中国权益，而应增持能留住降本和现金流的A/H资产，减持会被客户重定价和流动性冲击伤害的板块。2026Q3建议超配自动化、机器人、电网、公用事业、精选现金流充裕港股互联网；低配光伏组件、商品化电池、锂铜材料be…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "公用事业板块在2026年已演变为AI经济的增长底座，建议重点布局算力枢纽枢纽区域电力及碳信用变现能力强的垃圾发电标的。", "summary": "2026年公用事业板块正经历从防御红利向“AI增长基础设施”的结构性重估。AI数据中心扩张驱动电力需求超预期增长（2026年预测+5.7%），同时碳市场重启为垃圾发电龙头带来显著的CCER收益。识别出深圳能源、瀚蓝环境、淮河能源和甘肃能源为具备“电力+AI”或“碳资产”逻辑的低关注度标的，需警惕局…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 3, "neutral": 0, "value": 46.4, "events": [ { "title": "在红利踩踏压力测试下，核电与低北向区域水电是长江电力的有效平替，而城投水务与燃气因地方财政拖累或周期属性不具备真实防御底蕴。", "summary": "本报告对公用事业板块进行压力测试。在长江电力因9.2%北向持股面临外资踩踏风险的情境下，公用事业内部将严重分化。低北向区域水电（华能水电、川投能源）及核电（中国核电、中广核）凭借1.5%-2%的极低外资敞口与长现金流久期，能提供真实的防御刚性，是长江电力的完美平替。相反，城投水务因地方财政拖累导致…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI负荷在中国公用事业的传导是结构性分配而非总量上行，制度性回报集中在水电核电基荷、容量电价火电与电网设备承包侧，纯新能源IPP与东部省网平台承担未充分补偿的电网加固成本。", "summary": "2026-05-29：AI负荷增长不是中国公用事业的普涨故事，而是结构性切块。最高可兑现度的三类是：受监管电网投资和数字化承包侧（国电南瑞、许继电气、平高电气、思源电气、特变电工、中国西电），容量电价机制下的存量大型火电（华能国际、华电国际、国电电力、大唐发电，165元/kW容量电价[S1]），以…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股权益leg应采取“红利低波（水核火电，40%）+中盘价值（电网设备，40%）+周期价值（铜铝硅钢，20%）”的结构，其中电网设备的成长风险由30Y国债完美正交对冲，而周期价值作为…", "summary": "2026-05-29：在债券leg降久期至5.5年、由30Y国债提供核心尾部对冲、绿债/ABS提供底仓carry的新结构下，A股权益leg获得承担更高增长与通胀风险的预算。本卡将权益暴露拆分为“哑铃+α”结构：(1) 红利低波（制度回报段，约40%）：以水电核电和容量电价火电替代被低配的7-10Y…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 5, "neutral": 0, "value": 35.8, "events": [ { "title": "长江电力的资产已建成、改革免疫、派息有锚、利差超历史均值，可在当前电价市场化与资本开支周期下维持防御性与股息率优势，是承接A股医疗器械外溢防御资金的最深单一蓄水池。", "summary": "截至2026-05-30，本卡支持Research note 09防御溢出假设落地。长江电力（600900）以71.795 GW六库联调、2024年发电量2,958.4亿度、经营现金流RMB 642.7亿、≥70%派息底线，已经从'资本开支密集型水电开发商'转为'运维型公用事业'。2025年发改委…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供应端在2026-2027年间存在不可逾越的物理建设时滞，电价冲击将严重侵蚀云巨头算力TCO与ROIC，迫使其放缓采购，投资者应战略撤出算力硬件并超配长江电力与中国神华等避险公用…", "summary": "截至2026年5月31日，本报告支持前序研判，确认核电与燃机等能源建设周期（3-10年）根本无法填补2026-2027年间的“电力断层”。SMR商业化推迟至2030年后，燃机面临3年交期与气源管道等物理瓶颈，微电网因设备短缺同样受阻。压力测试表明，电价上涨将暴击云巨头GPU算力TCO，使其利润率转…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在设备全球极度短缺（交付期128-160周）与原材料70-85%传导下，Tier-1电力设备供应商确立了结构性成长与绝对定价定价权，而重资产建设型公用事业面临电价政治阻力与融资挤压，…", "summary": "截至2026-05-30，本压力测试表明，在铜铝紧缺与电力设备70–85%金属价格联动下，全球公用事业因美债5%与电价费率政治监管阻力，现金流严重受损；而中国国家电网凭借超低国内融资成本及特高压自主产业红利，韧性极高。Tier-1电力设备OEM因128–160周超长变压器交付周期握有绝对定价权，是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电信与水电核电的资本开支不会侵蚀股息承载力，但火电 IPP 错配与自由流通容量天花板将使 RMB 3500 亿轮动的超额收益缩水至 8%–12% 总回报，安全边际显著低于 Resea…", "summary": "本卡片对 Research note 08 的 RMB 3500 亿防御资金轮动论点进行压力测试，结论为有保留支持。电信三大资本开支较 2022 年 5G 峰值已下降 22%，2025 年合计 FCF/股息覆盖 1.59 倍，AI 算力网络开支虽是新变量但可控；纯水电核电（长江电力、华能水电、中广…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 2, "neutral": 0, "value": 55.2, "events": [ { "title": "公用事业链不是低关注度故事，而是煤电容量电价+ETS 扩容+现货市场三重重定价的早期窗口——可调度资产估值结构性上修，新能源 IPP 估值结构性下修。", "summary": "公用事业链 30 日激活率仅 11.7%，但正同时穿越三大体制切换：煤电容量电价 2026-01 提至固定成本 50%（高转型省 70%）；H1 2025 弃光 6.6%/弃风 5.7%、国家红线放宽至 10%；ETS 扩容至钢/水泥/电解铝（覆盖 +30 亿吨 CO₂e）。可调度容量被向上重估、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件底座面临物理电网与能源成本的双重严峻约束，其过度偏乐观的估值预期存在巨大脆弱性，而由长江电力与中国神华组成的公用事业防御组合展现出极强的抗风险红利韧性，完全应验并巩固了哑铃资…", "summary": "截至2026-05-31，本卡片对AI算力扩张的电力约束与公用事业防御组合的股息率稳健性进行了压力测试。定量仿真显示，AI算力在电网瓶颈与高电价下极其脆弱：若电价上涨至$0.220/kWh，云巨头GPU租赁毛利率将转负为-1.48%，导致Mag7估值乘数面临大幅收缩。相反，以长江电力和中国神华组成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 3, "neutral": 0, "value": 50.6, "events": [ { "title": "在2026Q3–2027Q1政策窗口前，A股应通过减配低毛利服务器整机、超配智能电网与液冷龙头，构筑‘清洁基荷公用事业防御墙 + 西部枢纽节点单选成长极’的优化哑铃型双轨配置，实现极…", "summary": "本报告针对2026-06-01基准情景B下，发改委出台示范园区式20%-30%输配电价折扣对A股公用事业防御性与AI算力成长股的冲击，确立了“双轨配置框架”。我们定量证明了由于折扣门槛极高，长江电力等龙头T&D收入流失低于3.0%，红利防御墙稳固；而西部算力枢纽度电成本降至0.385-0.395元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026年下半年信用梗阻与低利率‘资产荒’的宏观基准下，A股风格轮动呈现两极分化的‘极化杠铃’形态：一端是具有稀缺时空套利和现货出清红利的可调度物理资产，另一端是拥有强海外出海能见…", "summary": "本报告针对A股组合，对Research note 01–06公调度公用事业重估与Research note 03–05 AI基建设备链进行了深度的配置与风格轮动综合。我们提出‘极化杠铃’策略，建议以60%权重配置于防御性高分红可调度公用事业（内蒙华电、华能水电、中国核电），以40%权重配置于成长型…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年美股公用事业板块在BTM自发电冲击下仍深陷‘资本支出搁浅与降级陷阱’，而A股水电与核电龙头凭借国家发改委绿电直连过网费刚性缴纳政策构建的‘强监管防御墙’，实现0%的表后绕过…", "summary": "本报告定量评估了2026年美国PJM电网并网与过网费重构政策对电力板块的利润率冲击。PJM新规强征每MWh约8美元的系统备用过网费，略微压缩独立发电厂（IPP）直连利润率至62%-65%（仍享有高额溢价），并将公用事业FFO/Debt信用指标小幅修复至13.7%（仍低于14.0%降级门槛，信用压力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 1, "risks": 1, "neutral": 0, "value": 72, "events": [ { "title": "中国版AI电力溢价的核心不是纯IPP，而是能控制配网接口、负荷合同和源网荷储协同的上市运营商。", "summary": "截至2026-06-02，本卡支持AI主线向电力实物层迁移，但认为中国只能部分复制美国IPPs溢价。最强样板是大唐中卫云基地“物理直供+双边交易”绿电项目：2,000MW一期、87亿元投资、全面投运后满足22.9亿kWh年用电需求。上市表达应优先看三峡水利、涪陵电力等拥有配网和客户接口的企业，其次…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "截至2026-06-03，AI–电力链的确定性溢价来自“在受约束节点交付可靠一度电的权利”，而不是泛能源需求增长本身。", "summary": "截至2026-06-03（Asia/Singapore），本卡支持上一张卡片的“重资产基础设施算力”主线，但把可投资确定性收窄到电网稀缺。AI 数据中心不是简单制造全球能源短缺，而是在 PJM、ERCOT、中国东部负荷中心等节点制造可靠电力、并网、变压器、输电和成本分摊约束。全球数据中心用电由 2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 1, "risks": 7, "neutral": 0, "value": 37.4, "events": [ { "title": "在低利率和容量/价格改革下，公用事业个股通过煤电一体化、资产并表和节能换挡展现出强劲的债性防御力与价值重估空间。", "summary": "本报告系统调研了公用事业产业链，筛选出瀚蓝环境（垃圾发电全资并表粤丰环保）、陕西能源（自产煤平滑波动的煤电一体化）、涪陵电力（一季度配网节能业绩迎来拐点）和皖能电力（容量电价上调直增4.39亿净利）四个隐形冠军标的；并提示了算力绿电 PPA 溢价、大宗电价传导、电网设备投资和环保坏账拨备等跨链溢出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力红利资产的防御性与安全性评估：国投电力、华能水电、中国核电", "summary": "三家标的均通过卡 06 设定的两条硬线（融资占比 < 2%、经营性现金流稳定），可承接 AI 硬件释放的风险预算，但须差异化排序、不可等权。中国核电 2025 OCF 374 亿元最厚，发电量+12.98%，19 台在建机组带来成长但隐含资本开支与监管尾部风险，约 1.8% 股息率较低 [自测算]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在现货电价波动加剧的背景下，基荷绿电运营商通过极低成本、容量电价预期以及绑定AI数据中心的“算电协同”长协直连机制，有效对冲了下行风险并获得了成长性估值溢价。", "summary": "截至2026-06-04，本卡支持电力运营商作为配置“防守锚”的判断。面对2026年下半年现货电价波动，长江电力以极低边际成本与70%分红承诺构建防御底盘，而中国核电在机组投产高峰期等待核电容量电价政策落地。此外，“算电协同”成为核心进攻逻辑：新建AI智算中心面临80%绿电准入红线，通过与电力运营…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FERC对BTM脱网收费的监管底线消除了传统公用事业的资产减值风险，驱动AEP等T&D运营商从防守收息股重估为资本开支成长股，同时压制了核电IPP的FFO/Debt向上弹性。", "summary": "超大规模云厂商正投入500-800亿美元建设BTM“影子电网”，这原本可能导致公用事业输配电(T&D)资产因脱网而减值。但FERC近期裁决叫停了“免费搭便车”的共址模式，为T&D公用事业建立监管护城河。受监管公用事业的FFO/Debt获得保护，并迎来了费率基数扩张的成长红利；而核电IPP的“数据中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 7, "neutral": 0, "value": 30.8, "events": [ { "title": "美股大厂利润受损与NDX估值收缩将导致A股高拥挤AI硬件链面临-34.3%的估值与盈利双重杀跌压力，并触发74.5%概率的两融去杠杆踩踏；而绝对利差高达+249 bps的A股红利资产…", "summary": "基于2026-06-04，本报告量化评估了美股NDX重估对A股AI算力链（CPO、PCB）及红利资产的溢出与共振效应。测算表明，A股AI硬件等权篮子交易于32.5x的2026E P/E，其估值对美股弹性系数为1.35。美股大厂因PPA及电网延迟导致的利润率受损及NDX 20%的估值压缩，将使A股算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在全市场两融总额筹码密集及下半年面临解禁洪峰的背景下，应警惕电网出海链高估值两融盘与有色板块波动风险，并在红利防守端向低外资占比的绿色基荷运营商（中国核电）进行集中切换。", "summary": "截至2026-06-04，本卡支持电网设备（进攻矛）、铜资源（弹性盾）与电力运营商（防守锚）的配置主线，但强调微观杠杆与筹码结构已累积脆弱性。电网设备面临高额两融拥挤与Q3解禁重压（金盘科技、思源电气）；铜资源受全球流动性收紧10%压力测试影响回撤达30%；电力运营商长江电力虽两融仅1.81%，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国电网互联延迟、容量价格暴涨（PJM 清算于上限）及变压器交期恶化已构成 AI 算力落地的物理上限，且新能源项目回报率的崩溃将拖累绿电供给，AI 基建的完美定价面临重估。", "summary": "截至 2026 年 6 月 5 日，我们对 AI 电力和电网基础设施瓶颈进行压力测试。结果表明，电力就绪度已成为制约 AI 资本开支落地的实质性物理瓶颈。美国电网排队时长中位数已超过 4 年，且 2025/2026 年 PJM 容量拍卖连续清算在价格上限（2027/28 年度清算于 333.44 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "公用事业链最值得研究的不是大市值防御龙头，而是能在电价市场化、调节能力和地方回款约束中证明现金流质量的区域运营商。", "summary": "截至2026-06-05，公用事业链的核心不是防御属性，而是电力市场化、现金回款和碳市场扩围的交叉。优先深挖603693.SH江苏新能、600483.SH福能股份、600956.SH/00956.HK新天绿色能源，并把601158.SH重庆水务与000885.SZ城发环境作为现金质量观察名单。主要…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 68.2, "events": [ { "title": "日元平仓将以风格切换而非指数崩盘的形态外溢至 A 股——卖出北向重仓高 β 成长，买入低 β 高股息与政策对冲板块是当前最高确信度的策略调整。", "summary": "综合 Research note 01–05 的传导链，本卡判断日元套息平仓对 A 股的冲击是非线性、非均质的：基准情景为北向资金 5–10 日累计净流出 600–900 亿元，沪深 300 回撤 5–8%、创业板/科创 50 回撤 10–14%，USDCNH 上探但守住 7.25。冲击本质是全球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 3, "neutral": 3, "value": 66.6, "events": [ { "title": "2026年智算中心的alpha不再由算力规模决定，而由“能源适配效率”决定，配置应聚焦于稳定电源、高压直流配电、液冷及源网荷储一体化运营标的。", "summary": "2026年，A股智算中心（AIDC）的瓶颈从电费成本转向并网容量与电能质量。本卡片构建了公用事业与TMT的联合配置框架：增持具备电网稳定性与调节能力的中国核电、华能水电；增持解决高功耗瓶颈的800V HVDC（科华数据、麦格米特）与液冷技术（英维克）；以及具备“源网荷储”一体化能力的运营商（宝信软…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源冲击越深，电力红利权重越应从火电主导切向水核主导——水电与核电凭借零燃料成本+容量电价+调峰收益是FX不敏感的反脆弱α避险红利。", "summary": "支持Card-03的电力主导红利结论。截至2026-06-07，能源冲击下A股电力板块依靠'煤电联动+容量电价+水核零燃料'三重缓冲足以承担避险红利角色。压力测试显示：在煤价升至¥1,200/吨、JKM达$25的重度情景下，水电EPS仍+12%、核电+5%、火电因容量电价(¥100/kW)对冲后仅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "公用事业的真正预期差不在防御现金流，而在核电核准节奏、雅下流域溢出与 CCER 重启三条政策主线对中游运营商的现金流再定价。", "summary": "公用事业链 30 日激活率仅 13.7%，处于低覆盖。本卡片识别出 5 个具备 Setup 的标的：601985 中国核电（核准窗口）、600025 华能水电（澜沧江+藏东南叙事）、600674 川投能源（雅砻江参股）、601016 节能风电（绿电弹性）、000035 中国天楹（W2E+重力储能 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "公用事业被卖方冷落≠无 alpha——被错配标签的申能、川投、皖能等中盘标的，正同时受益于容量电价、136 号文与碳市场扩容的现金流重估。", "summary": "公用事业产业链 30 日激活率仅 13.7%，但容量电价二年、136 号文新能源全面入市、水价改革与 ETS 扩容四股政策力量正同时重塑现金流可见度。本卡片筛出五个被错配标签、卖方覆盖密度低的标的：申能股份 600642（被当火电的核电期权）、川投能源 600674（雅砻江纯水电权益）、皖能电力 …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 1, "value": 74, "events": [ { "title": "PJM 2026年资费重构与中国发改委绿电直连电价费率改革，导致美股与A股电力板块在BTM表后自发用电与信用质量上呈现剧烈的分化，做多零碳直供发电商与中国水电/核电巨头，做空受并网瓶…", "summary": "截至2026-06-07，美国电网并网队列积压突破2600GW，数据中心面临3-4年并网延迟。PJM 2026年资费重构征收$8.0/MWh备用费，使核电直供利润率微降至62.69%-64.79%，但相比常规上网销售的50.00%仍具高溢价；受规管电网的FFO/Debt信用比率仅修补至13.71%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "低碳走廊材料链优势的行业配置表达", "summary": "不应在四个选项里二选一，而应构建按信念度加权的四仓杠铃组合，把 Research note 05 的环节排序翻译成权重 [自测算]：水电/核电防御锚 40%（长江电力600900.SH、中国核电601985.SH，受 NDRC 监管豁免、10Y 国债1.72%下重估）；AI 电力基础设施/特高压 …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 2, "neutral": 0, "value": 66.4, "events": [ { "title": "绿电溢价与CEA抬升将终结A股电力红利与有色周期的同质化行情，催生“水电/核电红利成长化”与“水电铝阿尔法溢价”的戴维斯双击，同时迫使被动资金重构传统指数权重。", "summary": "截至2026年6月8日，承接上游成本测算，本卡片认为范围二纳入将引发A股公用事业与有色板块的深度估值重塑。4-6分/度的绿电溢价将使大水电/核电从纯“红利”向“红利成长”重估，而火电因面临碳成本与新能源资本开支压力面临股息缩减风险。电解铝板块将从同质化周期贝塔转向结构性阿尔法，水电铝凭借吨铝783…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "银行红利仓设上限后的A股策略再分配：释放风险预算应转向哪些行业或风格", "summary": "综合 Research note 01–04：银行高分红股权封顶于跨资产套息仓约30%是正确的风险纪律，但在A股账本上留下收益缺口与回报缺口。策略答案是哑铃式再分配：收益端从'银行—利率因子'分散到非银现金流红利（煤炭/能源、交运/高速、运营商、水电）加自由现金流因子与央企'HALO'红利（指数股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 7, "neutral": 0, "value": 37.8, "events": [ { "title": "SMR 2027年并网叙事泡沫证伪下的传统能源与物理红利股战术超配", "summary": "SMR 2027年并网叙事泡沫证伪下的传统能源与物理红利股战术超配 Debunking the 2027 SMR Grid Rebound: Tactical Overweight on Utilities & Physical Dividends 在商业 SMR 证实 2030 年前无法贡献 G…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SMR 2027年并网叙事泡沫证伪下的传统能源与物理红利股战术超配", "summary": "SMR 2027年并网叙事泡沫证伪下的传统能源与物理红利股战术超配 在商业 SMR 证实 2030 年前无法贡献 GW 级电力的物理事实下，AI 负荷增长所带来的电力溢价是否正完全流向传统电网与成熟核电/水电巨头？A 股长江电力（度电 0.4432 元）的绝对超额利差（+336 bps）是否使其防…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "公用事业的红利承接容量已部分透支（估值78-82分位、股息率仅3.85%、政策性让利压制绿电运营商），Research note 06的1pct分流仓应降至0.7pct并聚焦煤电+核…", "summary": "对Research note 06提出的公用事业1pct分流仓做压力测试。截至2026-06-10，板块PE-TTM 18.2×、PB 1.78×已升至78-82%历史分位，股息率压缩至3.85%，与10Y国债利差仅145bp（30分位）。子板块拆解显示煤电(35%权重/4.6%股息)与核电(12…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SMR 2027年并网叙事证伪触发能源资产向‘物理红利’轮动，长江电力凭借极高利差与电价溢价成为核心超配标点。", "summary": "截至2026年6月，由于监管和建设延期，SMR 2027年并网叙事已证实破裂，GW级产能需等到2030年后。AI电力需求溢价正流向长江电力等成熟水电巨头。长江电力受益于0.4432元/千瓦时的市场化高电价和+336 bps的超额利差，防御价值被系统性低估。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 0, "neutral": 1, "value": 84.2, "events": [ { "title": "公用事业产业链正处于电价市场化、核电扩张与 ETS 重启三轴重定价中，超额收益落在 003816、600803、600905、600008、1164.HK 五个低覆盖度标的，主流的长…", "summary": "公用事业产业链当前 30 日激活率仅 11.7%（24/205），叙事被长江电力 600900 一只标的吸走，但 2025-02 风光电价市场化、2026-03 十五五核电 110 GWe 目标、2026-01 ETS 重启叠加 CEA-2025 切换三大监管冲击在 12 个月内连发，买方盈利模型…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 70.6, "events": [ { "title": "在红利极化与公用信用分层并存的局面下，最优解是'水电核电守正60% + 出口链纯粹价值奇兵40%'的双底座再分层，而非任何一端的极端解。", "summary": "红利底座不应在'坚守水电核电'与'大举切仓出口链'之间二选一。建议执行'守正60%/奇兵40%'的双底座再分层：保留信用免疫的大水电与核电(国网总对总结算)，退出地方信用尾部恶化的中小水电、市政燃气、供热；将原红利底座的30-40%替换为公募严重低配、有息负债占比<15%、海外营收40-50%的出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红利拥挤度极化与公用信用传导下向出口链低拥挤纯粹价值防御切换", "summary": "红利拥挤度极化与公用信用传导下向出口链低拥挤纯粹价值防御切换 Utility Dividend Crowding & PPP Credit Risks vs. Low-Crowding Export Manufacturers 在红利拥挤度逼近历史极值、长江电力达90%且地方拖欠致公用账款增加15…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 2, "risks": 3, "neutral": 0, "value": 82, "events": [ { "title": "长江电力处于后资本开支期，其70%分红承诺在极端枯水和市场化电价压力下依然具备自我造血的刚性支撑；而中国核电在千亿资本开支高峰下自由现金流持续为负，其分红极度依赖外部融资，属于‘成长…", "summary": "截至2026年6月13日，水电与核电分红红利底座呈现显著分化。长江电力处于后资本开支周期，自由现金流（FCF）极度强劲（基准情景488亿元）。压力测试显示，在15%极端枯水和电价市场化波动的压力下，长江电力FCF（265.4亿元）依然能完全覆盖70%的分红承诺（157.5亿元），红利底座坚如磐石。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "绿色通胀与电网物理约束正诱发新能源“平价红利”的阶段性逆转，AI基荷需求的高标准使得绿色电力在当前基建瓶境下变得比传统能源更昂贵，从而延缓全球能源转型进程。", "summary": "截至2026-06-13，压力测试显示，尽管新能源理论LCOE仍具竞争性，但受电网升级溢价、变压器交期溢价及FEOC合规成本冲击，美国新能源PPA价格已推升至$64.5-$79.4/MWh。在加入储能以满足AI算力24/7基荷需求后，绿色电力总成本已升至$110-$125/MWh，逆转了相对于天然…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全天候清洁能源的环境溢价使水电/核电基荷与配储虚拟基荷资产NPV实现翻倍或大幅增值，触发了由科技资本直接背书和资助的公用事业结构性资本开支周期。", "summary": "本报告定量支持全天候清洁能源环境溢价对绿电运营商的估值重定价。超大规模厂商的Scope-2合规竞价使3 GW清洁基荷资产的NPV翻倍（+101.01%），EBITDA利润率提升，WACC降低70 bps。为规避电价蚕食，光伏运营商配储能协同开发，ROI可从4.18%提升至4.98%，在极端蚕食下挽…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "公用事业的机会不在泛化高股息，而在现金流可验证、负荷或来水有弹性、且能承接电力市场化和碳政策变化的区域基础设施运营商。", "summary": "截至2026-06-12，本卡支持启动公用事业深挖队列。优先对象是600461.SH洪城环境、000883.SZ湖北能源、601330.SH/01330.HK绿色动力、600483.SH福能股份；600674.SH川投能源更适合作为水电分红估值锚。核心证据来自2026年1-4月用电量同比增长5.4…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 0, "value": 79.4, "events": [ { "title": "AI大负荷的24/7特性与并网队列中新能源主导的增量存在根本错配，新建基荷电源的长周期迫使云厂商抢夺存量电力资产，加剧公用电网缺口。", "summary": "截至2026-06-14，本卡片压力测试结论：发电增量无法匹配AI基建需求。AI需要24/7稳定基荷，但目前美国并网队列中95%以上为间歇性新能源。核电（含SMR）在2030年前无实质增量，天然气扩建受制于管道许可。这迫使云厂商抢夺存量基荷（如核电背靠背PPA），剥夺了公共电网资源。结论：电力供应…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 3, "neutral": 0, "value": 68.8, "events": [ { "title": "电网（准许收益率）、核电（零燃料弹性）、头部煤电（两部制电价+长协）三链独立成立，A股公用事业瓶颈红利能转化为真实利润扩张。", "summary": "支持卡07：A股电力/电网瓶颈在2026年确实转化为电网、核电及头部煤电运营商的利润增长，并非被上游成本完全抵消。三条独立成立的传导链：电网受准许收益率框架保护、与上游能源脱钩，叠加2026年8500亿元投资计划，国电南瑞等设备商Q1新签订单+30%；核电燃料成本占比仅15-18%，对Brent …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源板块分子端在Brent中枢$65以上即可兑现card-06的200–250bps红利定价红线，真正的下限不是分红政策本身而是Brent长期中枢，临界值$60/桶。", "summary": "在Brent $70/桶以上且OPEC+维持配额纪律的基础情景下，全球油气一体化巨头（XOM/CVX/Shell/TTE/中海油/中石化）现金股息覆盖率>1.6x，现金股息盈亏平衡油价仅$40–45/桶，足以兑现card-06提出的红利股债利差200–250bps红线，能源板块20%重估空间成立。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI行业的‘Alpha’已从代码转向铜线与原子，物理电力设施的瓶颈正在压缩科技股估值，并重塑大类资产的定价逻辑。", "summary": "2026年全球AI瓶颈已从GPU芯片全面转向电力基础设施。超大型科技公司资本开支预计破7000亿美元，但受限于变压器、燃机及电网交付，30%-50%的数据中心产能面临延期。这种“资本开支扭曲”导致了严重的闲置算力资产损失，促使科技巨头大举投资能源。多资产配置策略建议超配电力硬件、铜、铀及核电公用事…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 0, "value": 76.2, "events": [ { "title": "公用事业分红脆弱性不是板块性的，而是价格传导机制、回款质量和资本开支回收速度决定的机制性风险。", "summary": "截至2026-06-16，本卡部分否定“下游公用事业分红风险最高”的笼统判断。电力资产因工商业市场化交易、电价拆分、容量电费和辅助服务具备较强传导能力；城燃、水务和环保虽有定价框架，但地方审批、居民端滞后、财政回款和补贴审计使风险更集中。结论：保留水电、核电和优质大型火电，降低弱联动城燃及应收账款…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在原料退化与B端品质下沉的500–850bp毛利压迫下，应将海天/青啤从超配切到明显低配，并将释放的约400bp仓位重构为'政策性口粮必需品+非餐饮型大众红利'的战术哑铃，板块保持中…", "summary": "工作日期2026-06-15。综合Research note 01–05结论：Research note 05给出的500–850bp原料/产品结构退化压力，已将Research note 02基于15–60bp渠道漏损建立的安全垫击穿10倍以上，原'做多干净成本传导'防御逻辑被否证。建议将海天(…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 4, "neutral": 0, "value": 65.6, "events": [ { "title": "算力交付物理瓶颈尚未在A股TMT板块充分定价，高估值硬件标的面迎15%-45%估值去化风险，应当战略重构组合并转向物理电网设备、上游矿山及防御性清洁能源公用事业。", "summary": "本报告基于电网与BTM直供电物理硬约束验证的12-24个月算力基础设施部署气穴期，指明A股算力硬件与上游材料板块存在估值脱钩和两融踩踏风险。建议在气穴期内进行风格轮动，超配出口导向变压器和上游铜铝，增配防御性清洁公用事业（长江电力、中国核电）。定量测算表明，优化后的战略组合可将承压收益率提升7.8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "公用事业链的可投资弹性来自“负荷增长+容量补偿+经营性现金流+碳市场扩围”的交叉点，广州发展优先，兴蓉环境和旺能环境适合做现金流与特许经营深挖，浙江新能仅列观察。", "summary": "截至2026-06-16，公用事业链仍属于本白板欠研究方向，30日激活仅24/206（11.7%）。本卡支持初始判断：更值得深挖的是二线综合能源、水务和固废运营商，而非单纯大市值防御资产。重点候选为广州发展(600098.SH)、兴蓉环境(000598.SZ)、旺能环境(002034.SZ)、浙江…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在算电协同的西迁叙事中，由于储能调峰是真正的物理约束与成本重灾区，A股配置应从‘总量逻辑’转向‘可靠性与效率逻辑’，首推储能与液冷，并对绿区IPP进行‘基荷优于间歇’的分化交易。", "summary": "基于此前针对西部绿区电力承载力的压力测试，我们综合提出A股算电协同主题的配置策略：储能调峰 > 液冷 > 特高压设备 > 绿区IPP。西迁逻辑中的核心瓶颈并非铭牌装机量，而是可靠容量的转换。因此，储能与火电灵活性改造享有最高Alpha；液冷作为数据中心PUE降本刚需紧随其后。特高压设备虽具高确定性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网瓶颈促使云巨头表后逃逸，引发美国公用事业板块搁浅资产否决、信用评级降级与分红削减的戴维斯双杀；中国发改委绿电直供政策则构筑了零表后逃逸的阿尔法防线，凸显长江电力与中国核电的绝对避…", "summary": "定量评估AI数据中心并网约束下美国受监管公用事业企业的信用与分红稳定性。在变压器短缺及并网延误下，云巨头转向表后自发电（BTM），导致电网公司面临21.0亿美元Capex监管否决，FFO/Debt降至13.21% [自测算]，跌破14.0%投资级门槛 [S10]，面临下调至BBB及分红被迫削减的戴…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 7, "neutral": 0, "value": 49, "events": [ { "title": "验证电力红利资产在极端市场波动下的防御价值", "summary": "压力测试长江电力（600900.SH）的防御价值，结论为「有条件支持」Card-01。2025年归母净利341.67亿、+5.14%，70%派息承诺延至2026-2030。136号文针对新能源、不及水电，反使水电相对更具防御性。自测算显示：即便严重多年期干旱（发电量−12%），股息率约2.81%仍…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "油价战争溢价消除与美联储鹰派紧缩背离下的红利资产结构性重平衡", "summary": "油价单日重挫超6%至约76美元，是美伊缓和抹去战争溢价、叠加IEA下调需求所致；美联储6月维持3.50–3.75%但点阵中值上调至3.8%，通胀降温与高实际利率（10Y国债约4.42%）并存。这并非红利逻辑解体，而是红利分叉：上游资源通胀红利（油气煤炭）因盈利基数收缩+折现率高企而估值锚松动；物理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力稀缺是SPCX拥挤风险的有效对冲，但美国AI电力赢家本身已部分拥挤，组合应超配稀缺因子、低配拥挤表达，并用水电红利与成本归因规则做底仓。", "summary": "2026-06-17：本卡综合Research note 01-04，结论是AI电力稀缺资产仍可作为SPCX拥挤风险的组合对冲，但必须拆成结构性对冲与拥挤化对冲两类。保留有合约支撑的核电IPP、600900式水电红利和具备大负荷成本归因的公用事业；但CEG、VST、TLN等美国AI电力赢家已带有成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在美股AI电力赢家与科技成长股存在高Beta重叠且容易被SPCX同步去杠杆的环境下，大类资产配置应向低贝塔的A股水电红利（长江电力）倾斜，并配置核电（中国核电）对冲干旱风险，构建收益…", "summary": "基于2026-06-17，本报告压力测试了SPCX挤压环境下，AI电力稀缺因子与水电红利的配置效率。美股核电IPPs（CEG/VST）对NDX的Beta已飙升至1.08-1.12，存在严重的Beta和动量重叠风险，在去杠杆中易遭受-25.0%的同步回撤。A股水电（长江电力）对NDX的Beta为-0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 1, "neutral": 0, "value": 86.4, "events": [ { "title": "上游能源红利逻辑正被强制切换为'现金流稳定'主导，但估值中枢不能维持当前水平——煤炭可作过渡仓但需-10%中枢下移，油气从核心降级为卫星仓，真正受益者是公用事业型物理资产。", "summary": "油价回落至70-80美元区间剥离了战争溢价，IEA下调2026年全球需求增速至+0.7 mb/d。上游红利逻辑被强制切换为'现金流稳定'主导，但估值中枢不能简单平移：A股三桶油/煤炭龙头股息率虽仍6-8%，盈利-分红覆盖倍数从2022年~2.0×压缩至~1.3×。压力测试显示韧性排序为煤炭>油气>…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 1, "neutral": 1, "value": 90.8, "events": [ { "title": "30%防御现金流端必须从泛红利收紧为A+B+C+E核心(公用事业/运营商/大行保险/煤炭油气)占80%+的本土现金流篮子，剔除与电网出海链同源的‘伪红利’出海标的，并按四情景对释放的…", "summary": "截至2026-06-20，本卡综合卡片02-08，把30%防御现金流端从泛红利收紧为外需/汇率/地缘三低敏感的本土现金流篮子：水电核电燃气8%+电信运营商5%+大行与大型保险7%+煤炭油气5%+高速港口铁路3%+必需消费2%，明确剔除出海家电、逆变器、出海化工等‘伪红利’标的。给出基础/中度/重度…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "对立路径是伪命题——只有'存量水核长协+CCGT桥梁+风光短储+少量LDES'的混合栈能在2026-2030窗口与新能源+短储PPA形成成本-可靠性双向占优，且仅限水核枢纽地理。", "summary": "截至2026-06-19，水电、核电、燃气+CCUS、长时储能任何单一栈都无法同时在成本与24/7 CFE匹配率上跑赢风光+4h锂电PPA：水电资源已开发60%+且枯水风险高，核电2030年前增量仅30-40GW不足以匹配AI 60-100GW新增负荷，燃气+CCUS全成本0.75-0.95元/k…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 5, "neutral": 0, "value": 65.2, "events": [ { "title": "电力公司用'队列改革+大负荷专属费率+共址限制'三件套把AI扩容资本与风险压回开发商，AI数据中心IRR基准下行150–300 bp、走廊分化扩大350–500 bp，受监管公用事业…", "summary": "截至2026-06-20，本卡支持Research note 6判断：AI算力交付的下一道硬约束已从设备短缺切到电网侧——并网队列、大负荷专属费率与共址限制三件套。LBNL显示全美并网队列约2,600 GW、中位等候5年；FERC Order 2023推行集群研究；Ohio PUCO 2025-0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源波动 regime 下的跨行业配置与 A股/科创硬科技定价", "summary": "本卡综合 Research note 01–05：电力与电网接入已成为整条 AI 价值链的约束性稀缺投入。在布伦特回落至约80美元、发达市场同步鹰派的 regime 下，正确做法不是削减硬科技，而是沿杠铃重塑风险预算。提出五桶矩阵：超配电力/电网赋能者（国家电网2026–30约4万亿、核电10机组…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面向多年期AI电力资本开支超级周期的配置：刚性容量稀缺确立后的板块配置与风格含义", "summary": "本卡把四环串联瓶颈（封装→电网/变压器→刚性发电）翻译成配置地图。关键事实：四个时钟串联，缓解任一环只把约束移向下游，不压缩通电总时间，故稀缺是结构性、多年期的，而非周期性。配置：超配机电卖铲人（电网/变压器，全球缺口约30%、中国占产能60%、出口+36%、订单排至2027）、刚性发电（核电/铀…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备交付延迟把能源转型的瓶颈从发电制造端转移到并网交付端，重估的是已并网基荷和电网设备产能，而不是所有化石燃料资产。", "summary": "2026-06-20：本卡对上一张 AI 电力瓶颈结论进行压力测试。结论是，变压器、变电站和配电设备延迟不会简单制造“化石能源复兴”，而是制造并网稀缺。若全球 1,500 GW 高级阶段新能源队列中 10%-25% 被延迟，将冻结约 329-821 TWh/年的清洁发电量，相当于 44-110 G…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 1, "risks": 10, "neutral": 0, "value": 44.6, "events": [ { "title": "在能源价格常态化预期下，红利配置应向清洁电力转移，但须通过煤电组合对冲水电枯水期现货买电风险，并警惕规制电网在AI负荷爆发期的信用评级恶化与表后流失风险。", "summary": "截至2026-06-21，本研究支持将组合红利仓位从周期性煤油向公用事业转移的逻辑。水电（长江电力）在严重枯水（-15%）情景下面临现货回购惩罚，净利润下降33.5%且面临29.7%派息削减风险；核电（中国核电）受益于煤电联动，电价上浮5%可提升利润15.88%，并捕获5.8-8.4分/kWh的碳…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DR007上行会压缩公用事业股息利差并提高再融资成本，但分红削减的核心触发器是资金利率、煤价、电价和资本开支同时恶化，而非DR007单项上行。", "summary": "截至2026-06-21，本卡对电力等“物理红利”资产进行压力测试：DR007从6月1日1.3644%升至6月18日1.4606%，会抬高贴现率和再融资成本，但尚不足以单独击穿公用事业分红。真正风险是DR007接近1.70%、煤价维持861元/吨以上、市场化电价走弱与高资本开支叠加。水电和低燃料b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张支撑公用事业防御配置，但只支撑具备监管现金流、容量收入、绿电供应和电能质量价值的子板块，而非所有发电资产。", "summary": "截至2026-06-21，我支持公用事业防御配置，但逻辑应收窄为电力质量、容量收入、绿电合约和电网投资。2026年5月互联网数据服务用电同比增长45.4%，2025年算力中心用电1700亿千瓦时，2030年或达8000亿千瓦时。AI负荷强化可靠电力稀缺价值，优选水电、核电、绿电PPA和容量补偿清晰…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电网约束因负荷规模与经济效益差异并不直接卡死先进封装厂，但下游智算中心并网排队将导致非对称交付延误，严重折损封装厂NPV（-58.1%）并引发云巨头巨额设备折旧；中美在表后直连（BT…", "summary": "电网侧约束不卡死高能效的先进封装厂（负荷小且税收贡献高，享有特批），但卡死下游大负荷智算中心（超高压主变交期超140周）。非对称延误12个月将导致封装项目NPV暴跌58.1%及云巨头年均50亿美元闲置折旧。政策上，中国发改能源〔2025〕650号强制征收直连过网费保护了国家电网与长江电力（利润率6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 2, "risks": 3, "neutral": 0, "value": 96, "events": [ { "title": "电力价值链中的\"成期权\"：核电、水电及SMR对公用事业估值重塑的影响", "summary": "本卡支持卡08的\"类债→成长\"切换，但收紧为一次三层分部拆解：每家大型公用事业应拆成（i）稳态年金（类债DDM）、（ii）已签20年期算力PPA（长久期年金）、（iii）稀缺电力看涨期权（实物期权溢价）。证据：PJM容量电价从约29升至329.17美元/MW·日FERC上限[S5]；中国核电在运2…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "真正受益的是电力交付瓶颈和已签长期购电/费率保护资产，而不是泛化公用事业或所有新增兆瓦。", "summary": "截至2026-06-22，本卡综合上一张AI基础设施观点：数据中心用电增长仍成立，但公用事业受益必须按“执行风险由谁承担”重新排序。最优是电网设备和电力服务，其次是有已签负载承诺及最低计费/大负荷费率保护的受监管公用事业，再次是已签长期PPA的核电或稳定清洁电源。泛化merchant gas、水资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张下的电力与电网供应压力及碳中和约束", "summary": "压力测试电力/电网/碳中和这一环：看多电网设备与绿色基荷的逻辑成立，但约束点被放错位置。它不是全国发电或电网容量不足——年用电已破10万亿千瓦时，2025年可再生能源消纳全部新增需求，国网十五五约4万亿元资本开支(+40%)前置。真正挤压在数据中心运营商层面，且是区域+监管性：东部算力集群电网约7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在公用事业超级资本开支面临股权稀释及大选年监管时滞的双重制约下，跨资产配置应调降拥挤的美股公用事业权益，通过公用事业IG高级债捕获拓宽的信用利差，同时布局低相关的A股水电红利（长江电…", "summary": "基于2026-06-22，本报告综合研判了受监管公用事业费率基数超级周期下的跨资产布局。公用事业1.3万亿美元资本开支带来严重的股权二次发行稀释，且大选年36州州长竞选引发监管机构费率抵制，惠誉调降行业展望至“恶化”。PJM 2027/28容量价暴涨至333.44美元/MW-日，零售电费同比上涨1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 9, "neutral": 0, "value": 45.4, "events": [ { "title": "容量电价与稳定性溢价落地下的电力公用事业与电网设备板块重估空间与风格轮动路径", "summary": "本卡支持并综合前四卡：算力＋电气化＋容量电价硬化现金流，使公用事业从“类债券”向“成长公用”重估。结论：重估真实但不对称。硬基荷（水电/核电）的倍数重估最大且仍在前方——以川投PE约15.3、隐含(r−g)≈2.9%为锚，(r−g)压缩40–80bp对应约15%–35%合理估值上行[自测算]；电网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在重型燃机与电网基建多年配给约束导致算力交付延迟2年的情景下，电费成本暴增将云巨头EBIT利润率蚕食150-300BP，迫使科技板块ERP走宽70BP并引发理论PE中枢下调16.77…", "summary": "基于2026-06-23，本报告评估了由燃机档期、精铸产能、已认证合金、变压器和气源许可共同形成的多年配给市场下科技股的估值重定价。燃机及电网配给将算力激活推迟1.5至2.5年，迫使2026-28年AI资本开支增长下修10%至15%。表后燃气+BESS电价升至$0.220/kWh，单卡GPU电力O…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "煤价已反弹而非下行，但火电龙头高股息凭借容量电价（114号文）、长协煤与派息纪律仍可守住约5-6.5%水平——红利支撑已从周期性的煤价切换为结构性的容量电价，应为票息持有火电、为确定…", "summary": "截至2026-06-23，本卡支持Research note 03的火电超配，但重置其论证基础。2025年驱动华能利润+42%的煤价下行红利已消退：2026年5-6月环渤海Q5500现货回升至约835-861元/吨、同比+15%，同时上网电价同比-3.48%，成本边际双向挤压。但红利仍可持续，因支…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在成本压力积聚与零售利润敏感性增加的宏观环境下，红利组合必须对运营租赁类REITs实施动态配置上限管理，向原材料独立的公用事业高股息股与国债倾斜，以保障组合总回报。", "summary": "针对成本冲击对折扣零售REITs分红稳定性的威胁，本报告提出红利防御组合优化框架。将REITs划分为高贝塔商业零售型（A类）与稳健特许经营型（B类）。在成本压力下，将A类REITs配置上限由15%下调至8%（中度）和3%（重度），增配水电红利股（长江电力 600900.SH）与10年期国债。模拟表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 5, "neutral": 0, "value": 70.8, "events": [ { "title": "在重型燃机与电网多年配给重塑科技股估值的情景下，超配长江电力等在岸清洁基荷资产的风险平价组合可锁定确定性溢价，且通过MXN/JPY外汇数字期权可有效锁定日元平仓海啸的尾部下行风险。", "summary": "本报告针对电网并网延迟与重型燃机5-7年超长交期约束，科技成长板块估值重塑，以及美日利差收窄导致的日元套利平仓风险，构建了长江电力、中国核电、大行/电信红利及短端国债组成的清洁能源杠铃组合。通过波动率风险平价（ERC）和高股息倾斜自定义风险平价进行了权重优化。为对冲日元套利平仓引发的流动性冲击，测…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "重型燃机与电网多年配给格局下跨资产清洁能源杠铃组合的风险归因、权重重构与全天候对冲成本测算", "summary": "重型燃机与电网多年配给格局下跨资产清洁能源杠铃组合的风险归因、权重重构与全天候对冲成本测算 在重型燃机与电网多年配给导致科技股估值重塑且在岸高股息资产重要性上升的情景下，如何利用全天候或风险平价（Risk Parity）模型重新为长江电力、中国核电、大行/电信红利以及短端国债分配风险预算？在计入了…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "绿电分时匹配和CEA有偿比例提高能把可调节清洁电力转化为可测算的盈利与估值溢价，但该溢价只能提供相对对冲，无法单独抵御A股成长股系统性融资去杠杆。", "summary": "截至2026-06-25，本卡片压力测试绿电分时匹配与碳配额收紧下A股公用事业稀缺资产的估值弹性。结论是：可调节水电和具备储能、绿电直连或数据中心PPA的绿电运营商确有基本面溢价，基准测算约27元/MWh，可带来约18.2%的盈利弹性；若叠加ERP收缩，公允价值可提升约35.9%。但在A股成长股融…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "液冷驱动的超高功率密度对电网配电侧建设与能源供应响应力的挑战", "summary": "经压力测试：在受监管公共电网层面答案是“不能”，园区层面仅对提前占位者有条件“能”。密度对电网不是关键变量——100MW园区无论20kW还是140kW机柜都取电约100MW[自测算]，密度只是把约束下移到馈线/变电站/变压器并加剧负荷波动（已触发NERC突发失负荷预警）。时间错配是结构性的约5.6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "由于发电设备长交期、核电重建缓慢、跨州输电网架建设滞后以及并网容量在地理上的严重错配，电网侧能源增量无法匹配2026-2028年AI负载爆发式增长，电力缺口将实质性压制算力部署进度，…", "summary": "本卡对电网侧增量电源进行压力测试。2026-2028年，电网侧增量无法有效填补AI电力缺口。重型燃机面临5-7年超长交期；绿地核电停滞且SMR商业化迟于2030年，Palisades与三哩岛等重启仅能满足3.4%-10.8%的新增需求；输电线建设长达6.5-10年以上，且新增新能源地理错配严重（主…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 4, "neutral": 1, "value": 82.6, "events": [ { "title": "电力电网及公用事业红利资产的配置逻辑验证", "summary": "支持第01张卡片的防守再平衡判断，并验证其机制：本轮去杠杆是两融融资拥挤的电子板块冲击（A股两融2026-06-23破3万亿元），而受监管公用事业由保险/养老等长钱持有、融资拥挤度低，处于压力源的对立端，是真正避风港而非平行泡沫。需求端1-5月用电+3.4%、互联网数据服务用电+26%给出AI电气…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "136号文市场化电价下绿电、核电、水电和储能运营商现金流敏感性", "summary": "对运营商做三降压力测试：绿证价格、利用小时、市场交易电价同时下行。2025年业绩已是实盘验证，横截面决定性。保留满仓——长江电力(600900)水电、中国核电(601985)核电：电价下行年净利仍增(+6.17%/+6.00%)，水电近零边际成本+65.79%毛利、来水定产不受限电；核电>7500…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股市场结构约束下的清洁基础设施与能源对冲执行方案", "summary": "截至2026-06-26，本卡把Card03杠铃(12%清洁+4%能源)与Card04子仓上限落地为可成交的A股方案。核心结论：约束不难，难在表达载体可得性高度不对称——电网有唯一高流动纯标的(159326,约¥343.7亿,特高压权重约66%)；电池有百亿级但偏电动车(159755,约¥158.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "保留12%清洁+4%能源杠铃，但A股表达必须从宽基清洁贝塔转为受限的电网/储能/运营商组合，并用能源对冲抵消价值和通胀尾部而非追高红利能源。", "summary": "截至2026-06-26，本卡综合前序12%清洁基础设施+4%能源尾部对冲，但指出A股执行并非因子中性。清洁袖主要暴露于成长、资本开支、低股息和动量，尤其电网设备、电池、储能、光伏存在追涨和集中度风险；能源袖提供价值、高股息和通胀/霍尔木兹尾部保护，但中证能源前五权重约61.0%，不能无限扩张。建…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 0, "value": 87, "events": [ { "title": "在通缩压力下，公用事业并非均质的防御锚；必须坚守‘电网优先’与具备正向国债利差的可归因费率资产，警惕无费率保护的算力PPA延迟以及高溢价新能源标的的估值陷阱。", "summary": "基于2026-06-28的宏观及行业数据，本报告对公用事业作为杠铃组合“防御锚”的稳定性进行了压力测试。美股算力PPA面临并网延迟与BTM核电监管壁垒（如FERC Talen裁决）的严重制约，极端情景下BTM并网被禁可致核电项目EBITDA转负（-9.48百万美元，DSCR -0.17x）。A股“…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源与电力板块（长江电力/中国海油）的防御性估值与政策风险压力测试", "summary": "对卡片09把半导体腾出资金转入长江电力(600900)与中国海油(0883)防御的安排做压力测试，得出分裂裁决。电力/长江电力基本面不过热（零燃料成本、70%分红锁定至2030），但作为工具已成拥挤的26.8年久期债券替代品：股息率压至3.5%、对国债利差约2pp处历史高位、内在价值≈市值、贴现率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 3, "neutral": 0, "value": 88.4, "events": [ { "title": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "summary": "截至2026-06-28，本卡对“AI负荷增长能否支撑公用事业高分红”作压力测试：结论是精选组合层面可以，板块整体不能机械外推。AI数据中心用电增长真实，但也同步拉动电网、发电和储能资本开支。水电、核电、容量电价支持的火电及低杠杆综合电力运营商仍可维持分红；风光开发商受消纳、电价和项目资本开支制约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理基础设施约束下多资产投资组合的重平衡与防守哑铃配置策略", "summary": "本卡支持卡片05–07链条，把物理基础设施约束落地为多资产重平衡：构建防守哑铃。削减超估值且拥挤的AI硬件成长腿（−12pp至20%并加期权领口），把仓位转入物理基础设施租金腿（公用事业含水电核电、电网重电与变压器、水权、液冷热管理，+14pp至22%）与高股息压舱（中证红利约4.68–4.9%股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费成本转嫁挤压。", "summary": "基于国产芯片2.94倍能效赤字拉长通电时间至18个月的物理瓶颈，本报告系统性量化了A股电力设备板块（特高压、大变压器）的估值扩张机制与低利润 traditional TMT 的电价挤压。电网容量受限与十五五规划共振，支持了特高压订单可见度与大型变压器出海高溢价，特别是在国内外硅钢5.1倍巨大价差的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 5, "neutral": 0, "value": 77.8, "events": [ { "title": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红…", "summary": "本报告针对2026年三季报TMT硬件估值折价预期，校准了全球全天候和风险平价模型。战术防御性风险平价（TDRP）将AI成长股风险预算由25.0%下调至8.0%，防守红利上调至40.0%，以对冲55.0%的TMT增长溢价折价。模型显示，电力配给约束下算力效率的TCO优势在24/7清洁能源下扩大至9.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "summary": "截至2026-06-29，本卡综合前9卡：AI电力约束是选择性时序上限而非订单毁灭。建议跨资产账本配置为40%受资助AI基础设施龙头、45%主权背书红利公用事业、3%年度申万电子Put权利金预算、12%流动性储备。进攻端以中际旭创18%、工业富联14%、英维克8%为核心；防御端以水电18%、核电1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电…", "summary": "本报告对AI算力激增带来的电力电网物理瓶颈（变压器交期120-210周[S2]、并网排队4-7年[S1]）对A股转型龙头的利润率传导进行压力测试。结果显示，电网扩容滞后充当了新能源装机的“限速阀”，压制了光伏/电池销量增速；同时，容量电价[S8]和碳成本上升[S19]增加了高能耗制造环节电费开支，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-08dd090dc9/", "networkHref": "/investor/research-network/#thesis-08dd090dc9" }, { "id": "thesis-3986819dc0", "source": "tracked_thesis", "title": "A 股半导体的核心风险从订单验证转向 A/H 溢价、两融与负 Gamma", "titleEn": "A-share semiconductor risk is shifting from order proof to A/H premium, margin financing, and negative gamma", "lane": "A股半导体/两融负 Gamma", "laneEn": "A-share semiconductor margin and negative gamma", "laneHref": "/investor/lanes/a-share-semiconductor-margin-and-negative-gamma/", "investableFocus": "A股半导体/两融负 Gamma：588000.SH, 688981.SH, 0981.HK, 688041.SH, 688256.SH, 688012.SH, 688072.SH, 688008.SH", "stockUniverse": [ { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "A/H 溢价核心", "roleEn": "A/H 溢价核心", "bucket": "core", "href": "/investor/stocks/688981.SH/" }, { "ticker": "0981.HK", "name": "中芯国际", "market": "HK", "role": "H 股估值锚", "roleEn": "H 股估值锚", "bucket": "core", "href": "/investor/stocks/0981.HK/" }, { "ticker": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产算力权重", "roleEn": "国产算力权重", "bucket": "core", "href": "/investor/stocks/688041.SH/" }, { "ticker": "688256.SH", "name": "寒武纪", "market": "A-share", "role": "AI 芯片高 beta", "roleEn": "AI 芯片高 beta", "bucket": "core", "href": "/investor/stocks/688256.SH/" }, { "ticker": "688012.SH", "name": "中微公司", "market": "A-share", "role": "半导体设备", "roleEn": "半导体设备", "bucket": "core", "href": "/investor/stocks/688012.SH/" }, { "ticker": "688072.SH", "name": "拓荆科技", "market": "A-share", "role": "薄膜沉积设备", "roleEn": "薄膜沉积设备", "bucket": "core", "href": "/investor/stocks/688072.SH/" }, { "ticker": "688008.SH", "name": "澜起科技", "market": "A-share", "role": "存储接口芯片", "roleEn": "存储接口芯片", "bucket": "core", "href": "/investor/stocks/688008.SH/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" } ], "stockTickers": [ "588000.SH", "688981.SH", "0981.HK", "688041.SH", "688256.SH", "688012.SH", "688072.SH", "688008.SH", "510300.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK" ], "coreView": "赛道：A股半导体/两融负 Gamma。候选观察标的：588000.SH, 688981.SH, 0981.HK, 688041.SH, 688256.SH, 688012.SH, 688072.SH, 688008.SH。当前共有 40 条证据、4 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 76, "eventCount": 160, "firstSeen": "2026-04-24", "lastSeen": "2026-06-29", "supports": 11, "risks": 125, "neutral": 24, "analystCount": 12, "sourceTypeCounts": { "研究线程": 7, "研究笔记": 108, "分析协作": 22, "证据更新": 1, "研究主题": 22 }, "topTerms": [ "HBM", "科创50", "A股", "两融", "AI 芯片", "中芯国际", "寒武纪", "半导体", "设备", "雪球", "Gamma", "A-share" ], "whatChanged": [ "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。", "2026-06-29: [auto-handoff] 半导体折算率调降保证金缺口与科创板强平风控敏感性测试", "2026-06-29: 国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多资产组合VaR在1530点处强制破位，急需分层部署的ETF买盘及两融弹性化阀门进行精…", "2026-06-28: 在紧平衡流动性与外部融资约束下，A股难以支撑全局性风险偏好扩张，下半年配置应降低总Beta，采用“硬科技国产替代（封装/设备）+高独立现金流红利”的结构性杠铃组合。", "2026-06-28: 全球AI资本开支周期下的A股行业配置与风格轮动趋势" ], "practical": { "sectorLane": "A股半导体/两融负 Gamma", "sectorLaneEn": "A-share semiconductor margin and negative gamma", "exposureBucket": "中国权益与流动性", "tone": "green", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "A/H 溢价核心", "roleEn": "A/H 溢价核心", "bucket": "core", "href": "/investor/stocks/688981.SH/" }, { "ticker": "0981.HK", "name": "中芯国际", "market": "HK", "role": "H 股估值锚", "roleEn": "H 股估值锚", "bucket": "core", "href": "/investor/stocks/0981.HK/" }, { "ticker": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产算力权重", "roleEn": "国产算力权重", "bucket": "core", "href": "/investor/stocks/688041.SH/" }, { "ticker": "688256.SH", "name": "寒武纪", "market": "A-share", "role": "AI 芯片高 beta", "roleEn": "AI 芯片高 beta", "bucket": "core", "href": "/investor/stocks/688256.SH/" }, { "ticker": "688012.SH", "name": "中微公司", "market": "A-share", "role": "半导体设备", "roleEn": "半导体设备", "bucket": "core", "href": "/investor/stocks/688012.SH/" }, { "ticker": "688072.SH", "name": "拓荆科技", "market": "A-share", "role": "薄膜沉积设备", "roleEn": "薄膜沉积设备", "bucket": "core", "href": "/investor/stocks/688072.SH/" }, { "ticker": "688008.SH", "name": "澜起科技", "market": "A-share", "role": "存储接口芯片", "roleEn": "存储接口芯片", "bucket": "core", "href": "/investor/stocks/688008.SH/" }, { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" } ], "stockTickers": [ "588000.SH", "688981.SH", "0981.HK", "688041.SH", "688256.SH", "688012.SH", "688072.SH", "688008.SH", "510300.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK" ], "primaryStocks": [ { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "A/H 溢价核心", "roleEn": "A/H 溢价核心", "bucket": "core", "href": "/investor/stocks/688981.SH/" }, { "ticker": "0981.HK", "name": "中芯国际", "market": "HK", "role": "H 股估值锚", "roleEn": "H 股估值锚", "bucket": "core", "href": "/investor/stocks/0981.HK/" }, { "ticker": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产算力权重", "roleEn": "国产算力权重", "bucket": "core", "href": "/investor/stocks/688041.SH/" }, { "ticker": "688256.SH", "name": "寒武纪", "market": "A-share", "role": "AI 芯片高 beta", "roleEn": "AI 芯片高 beta", "bucket": "core", "href": "/investor/stocks/688256.SH/" }, { "ticker": "688012.SH", "name": "中微公司", "market": "A-share", "role": "半导体设备", "roleEn": "半导体设备", "bucket": "core", "href": "/investor/stocks/688012.SH/" } ], "instrumentTags": [ "588000.SH", "科创50 ETF", "688981.SH", "中芯国际", "0981.HK", "中芯国际", "688041.SH", "海光信息", "688256.SH", "寒武纪", "688012.SH", "中微公司" ], "reviewCadence": "每日复核", "confirmationSignals": [ "融资余额、ETF 申赎和北向资金是否同向", "被动调仓窗口是否被主动资金提前交易", "订单、盈利和估值是否同步改善" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "指数期权", "红利低波", "汇率", "两融降温" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 104 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 12, "recentEvents": 104, "supportRatio": 0.07, "riskRatio": 0.78, "conflictScore": 88.1, "evidenceDepth": 100 }, "score": 70.5 }, "events": [ { "date": "2026-06-29", "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）将使平仓率非线性飙升至20%，面临940亿STAR50抛压与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.5, "matchedTerms": [ "科创50", "A股", "两融", "雪球", "Gamma", "gamma", "STAR50" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 半导体折算率调降保证金缺口与科创板强平风控敏感性测试", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 9cb3a370-d1af-4cb9-8d02-f636ff1c90fa 结果摘要: Wrote handoff_response.md (12184 bytes); wrote handoff_response.en.md (13274 bytes) — Quantif…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 27, "matchedTerms": [ "A股", "半导体", "Gamma", "A-share", "semiconductor", "margin", "negative", "gamma" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "HBM", "A股", "半导体", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "虽然278%平均维保安全垫使两融连锁抛压呈局部非系统性风险，但20-30%回撤仍将触发最高2528.00亿的平仓压力并重创产业链，应通过减仓高倍数前道、超配低谷先进封装以及引入非担保品指数对…", "summary": "截至2026-06-28，两融杠杆在电子和半导体板块高度集中（总余额分别达6132.72亿和560.80亿），前五大龙头占用386.80亿。压力测试显示，20-30%的回撤会迫使大批账户跌破130%平仓线，极端情况下（科创50跌破1520）诱发2528.00亿平仓压力，并因雪球负Gamma在2-5天内引发600-750亿级联抛售。虽然278%平均担保垫等四…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.25, "matchedTerms": [ "科创50", "两融", "半导体", "设备", "雪球", "Gamma", "semiconductor", "gamma" ], "href": "" }, { "date": "2026-06-28", "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心腿；半导体国产替代（¥2万亿AI建设计划+80%国产强制比例…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "HBM", "A股", "寒武纪", "半导体", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "在紧平衡流动性与外部融资约束下，A股难以支撑全局性风险偏好扩张，下半年配置应降低总Beta，采用“硬科技国产替代（封装/设备）+高独立现金流红利”的结构性杠铃组合。", "summary": "在美联储加息预期和国内紧平衡流动性制约下，A股面临外资流出、两融去杠杆与解禁供给的三重压力。外资流入放缓，险资增配高股息红利防御资产。两融余额高企（3.015万亿元）使科技成长面临局部去杠杆和级联平仓风险。半导体板块估值已普遍收缩30%至45%，大基金三期（3440亿元）提供战略支撑。下半年主导风格将呈现“国产替代硬科技+高壁垒红利防御”的杠铃组合。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "A股", "两融", "半导体", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 工业制造分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: eb737f41-b97c-43b9-9b89-04a7c7636133 结果摘要: Wrote handoff_response.md (10188 bytes); wrote handoff_response.en.md (12356 bytes) — 三日近80…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 26.5, "matchedTerms": [ "科创50", "A股", "两融", "雪球", "Gamma", "gamma" ], "href": "" }, { "date": "2026-06-27", "title": "中国央行通过结构性再贷款与离岸FIMA人民币国债抵押回购机制，成功解耦了汇率防御与境内流动性，为A股“高股息防御+硬科技成长”的滞胀杠铃配置提供了宏观及风险平抑支撑。", "summary": "截至2026-06-27，在美中利差倒挂-265bp与低通胀下，央行转向境内结构性工具与离岸FIMA Repo防线，稳定了境内流动性（套利流出降90%至7.9B RMB）。这巩固了A股以高股息国企（海油6.2%）为防守、硬科技再贷款为成长的“滞胀杠铃”配置，抑平了368.9B两融及35B雪球负Gamma踩踏风险。优化组合VaR降至20.425%。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.25, "matchedTerms": [ "A股", "两融", "雪球", "Gamma", "gamma", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "半导体板块融资压力下的负Gamma传导与估值风险评估", "summary": "半导体是这场强平冲刷的震源而非平均参与者。中证半导PE(TTM)约177倍、近10年约91%分位（2021高点仅约80倍），中位数约181倍是海外3–5倍，美银5月调查列其为最拥挤交易（73%）；杠杆又恰堆于此（IT月净融资超1000亿、电子CMR 138.55%距130%追保线仅8.55pp）。高PE如长久期：177倍上每压10倍PE≈-5.6%，压20…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.75, "matchedTerms": [ "科创50", "寒武纪", "半导体", "雪球", "Gamma", "semiconductor", "gamma" ], "href": "" }, { "date": "2026-06-26", "title": "科创50回撤15%会触发半导体去杠杆左尾，但连锁强平的根因是科创/半导体集中账户与券商150%-160%集中度平仓线，而不是中芯国际、北方华创两只股票的公开融资余额本身。", "summary": "截至2026-06-27，本卡压力测试显示：科创50若从2026-06-26收盘2032.28回撤15%至约1727，普通130%平仓线只击穿冲击前CMR≤152.9%的账户；真正危险来自券商双创集中度规则，50%-80%集中度可把平仓线抬至150%，80%以上抬至160%，使176%-188% CMR区间账户也可能触发。中芯国际与北方华创公开融资余额合计…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28.75, "matchedTerms": [ "科创50", "A股", "中芯国际", "半导体", "semiconductor" ], "href": "" }, { "date": "2026-06-26", "title": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局", "summary": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局 在上午科创50跌破关键支撑触发雪球敲入与两融强平后，午间主力ETF资金流入与个股金针反转是否标志着负Gamma抛售压力的阶段性出清？我们应如何利用这一估值剪刀差进行右侧建仓，从红利低波向高Beta科技股进行再平衡？", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 26, "matchedTerms": [ "科创50", "A股", "两融", "雪球", "Gamma", "gamma" ], "href": "" }, { "date": "2026-06-25", "title": "A股半导体两融强平底与全球AI硬件景气度背离的投资套利机会", "summary": "A股半导体两融强平底与全球AI硬件景气度背离的投资套利机会 Divergence Between A-Share Semiconductor Margin Liquidations and Global AI Hardware Super-Cycle 在美光爆表财报确立全球AI硬件超级周期的背景下，如何评估A股科创板因两融折算率下调谣言与雪球敲入引发的散户…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "neutral", "supportLabel": "观察", "score": 30.5, "matchedTerms": [ "A股", "两融", "半导体", "雪球", "A-share", "semiconductor", "margin" ], "href": "" }, { "date": "2026-06-25", "title": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局", "summary": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局 Midday A-Share Growth Rebound and Options Negative Gamma Clearance Tactical Asset Allocation 在上午科创50跌破关键支撑触发雪球敲入与两融强平后，午间主力ETF资金流入与个股金针反转是否标志着负Gamma抛售…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 30.25, "matchedTerms": [ "科创50", "A股", "两融", "雪球", "Gamma", "A-share", "negative", "gamma" ], "href": "" }, { "date": "2026-06-25", "title": "A股电子两融强平螺旋与期权对冲防线的实操效率辩论", "summary": "A股电子两融强平螺旋与期权对冲防线的实操效率辩论 A-Share Electronics Margin Liquidation Spirals and Options Hedging Efficiency 在系统性去杠杆Cascade下，若A股行业面临15%回撤，半导体与科技板块将承受19.89十亿元平仓抛压，且中金所IM期货贴水年化-19.0%引发二次抛…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 30, "matchedTerms": [ "科创50", "A股", "两融", "半导体", "A-share", "margin", "electronics" ], "href": "" }, { "date": "2026-06-25", "title": "CNH套息流动性溢价回缩不会抑制A股半导体上游的当期订单与实物资本开支(政策供资、提前排产、实物周期脱钩)，但会压缩其股权估值并抬高下一轮股权融资扩张成本(股权周期未脱钩)。", "summary": "截至2026-06-25，本卡片对全球半导体设备周期与A股AI基础设施流动性风险做脱钩验证。结论：脱钩因传导链而异，须拆为两条。(1)订单/实物资本开支链——大体脱钩、低风险：北方华创订单排至2027年、营收指引¥468-520亿，中芯2026资本开支约80亿美元持平，由大基金三期(¥3440亿)、国有银行注资(中芯+77.8亿美元)与本土化使命供资，季末…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28.75, "matchedTerms": [ "科创50", "A股", "半导体", "设备", "semiconductor" ], "href": "" }, { "date": "2026-06-25", "title": "在2026-06-26，鉴于电网对冲使CPO收入下修收窄至13-17%，A股核心科技龙头基本面安全垫抬升，但由于技术性两融拥挤高企，国家队必须坚守1750战术线及1550生死线以防范负Gam…", "summary": "截至2026-06-26，本报告综合评估了CPO收入下修收窄至13-17%下，A股科技板块（中际旭创/寒武纪/源杰科技/联讯仪器）的配置与防御路径。尽管基本面压力因电网加速器对冲而减轻，但高位两融拥挤及成交量萎缩风险未除，国家队应维持1750点战术支撑与1550点绝对生死防线以阻断强平踩踏[自测算]。建议超配中际旭创与出海电力设备（国电南瑞/中国西电），低…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28, "matchedTerms": [ "A股", "两融", "寒武纪", "设备", "Gamma", "gamma" ], "href": "" } ], "timeline": [ { "date": "2026-04-24", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "国产AI加速卡2026年放量对A股半导体链的结构性传导", "summary": "国产AI加速卡2026年放量对A股半导体链的结构性传导 2026 Q1-Q2昇腾/寒武纪/海光等国产AI加速卡出货提速,对A股半导体设备、先进封装、HBM替代和材料四个环节的业绩弹性和估值重估程度究竟有多大,当前市场是否已经充分Price-in?", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "未来12个月的最优配置不是全面押注国产化，而是超配最先把政策性信贷、算力建设和电力市场改革转化为订单与TCO优势的环节：液冷第一、光互联第二，电力市场化资产成为新增第二收益腿。", "summary": "本卡综合Research note 01-05，将国产AI算力链条转化为100单位A股/港股主题仓位：科技算力基础设施62单位、电力与公用事业赋能33单位、战术储备5单位。科技内部排序为液冷22、光互联16、国产AI半导体/HBM/设备12、先进封装5、CANN/云迁移7；电力内部为西部VRE I…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 0, "neutral": 1, "value": 47.8, "events": [ { "title": "[auto-handoff] A股天量成交的可持续性评估", "summary": "致 技术分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 3106c448-57c8-42c1-a7a3-6bc90c8c9587 结果摘要: Wrote handoff_response.md (11,702 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "A股半导体头部公司包括中芯国际 (688981)、寒武纪 (68．8256) 等。", "summary": "A股半导体头部公司包括中芯国际 (688981)、寒武纪 (68．8256) 等。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "若VIX收在25以上且两融仍接近2.83万亿元人民币，A股应优先降低融资拥挤的中小成长和周期贝塔，并用MO认沽价差对冲真实去杠杆风险，而不是用便宜但错配的对冲获得虚假安全感。", "summary": "截至2026-05-13，本卡压力测试认为，VIX收在25以上并不会自动触发A股系统性清仓，但在两融余额接近2.83万亿元人民币时，会把风险集中推向融资拥挤、高贝塔成长和周期贝塔。最脆弱行业是电子、半导体、通信、计算机、国防军工、机械、有色、电力设备，以及券商和金融IT。5月解禁供给不是市场级主触…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 1, "risks": 0, "neutral": 0, "value": 55.6, "events": [ { "title": "国产算力进入性能驱动后，A股计算机与半导体板块全年风格将从小盘成长切向中大盘核心资产，配置优先级抬升设备—国产算力—先进制程—云与AI应用龙头主线。", "summary": "支持Research note 01的范式转移结论：DeepSeek V4在国产硬件上的性能跃迁意味着估值锚从政策预期切回业绩兑现，2026年A股计算机与半导体板块的风格将系统性地从'小盘成长/主题轮动'切向'中大盘核心资产/业绩兑现'。全年配置优先级：超配半导体设备(北方华创、中微)、国产AI算…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 2, "neutral": 0, "value": 37, "events": [ { "title": "A股半导体重估的 beta 已基本被定价，下一段必须靠后道设备、量检、材料、车规功率与先进封装上游材料的盈利与订单兑现，回避换手/融资/解禁三重拥挤的设计、代工与映射封装。", "summary": "截至 2026-05-15，A股半导体经过 4-5 月重估后，国产 GPU/ASIC 概念、HBM/先进封装映射股、大市值晶圆代工四条线已拥挤：5 日换手率分位 81-92%，半导体两融占自由流通市值比 ~4%（88 分位），主要芯片 ETF 4 周连续净申购 +18%，且 6-8 月解禁集中（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股指数层面未错误定价（政策买盘锚定）但成长子指数与板块仓位层面错误定价了峰会尾部——AI应用是冷场/破裂情景下仓位最脆弱的方向，半导体设备/IDM反而是熊情景下的对冲，出口链已被预…", "summary": "A股呈现典型末段集中化：上证逼近2015-07以来高点由红利/银行/央国企顶起，深成指-0.9%、创业板/科创跑输，北上未防御、两融偏高、ETF国家队买盘仅守指数。结构上风险溢价被散户/量化吸收而非外资。在Trump-Xi峰会冷场/破裂情景下，仓位最脆弱的是AI应用(688111/002230/3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 2, "neutral": 0, "value": 38.4, "events": [ { "title": "半导体与AI硬件的基本面韧性只能缓冲估值回撤，不能吸收融资盘强平的机械冲击；当前盘口尚未完成去杠杆定价。", "summary": "截至2026-05-16，半导体与AI硬件的估值韧性有产业支撑，但量价关系尚未充分反映融资盘强平冲击。5月13日A股两融余额达28,739.45亿元，电子行业单日融资净买入85.24亿元；4月1日至5月11日硬件设备与半导体合计贡献约六成融资净买入。科创板两融连续11日增加，5月15日板块反弹更像…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 硬件盈利端脱钩成立，但价格路径脱钩不完整；基本面能缓冲两融与解禁冲击，不能消除冲击。", "summary": "截至2026-05-17，本卡压力测试 Research note 06：AI 硬件基本面强到足以形成先进封装、HBM 供应链、光模块/AI datacom 的窄口径例外，但不足以让广义半导体 beta 脱离 A 股两融拥挤和解禁供给。NVIDIA FY2026 收入 USD 215.938bn、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "订单流驱动的估值底面临“执行力”压力测试，需警惕寒武纪等HBM高度依赖标的的流动性陷阱，中芯国际底部更稳但缺乏弹性。", "summary": "截至2026-05-19，华盛顿关于限制存量设备维护及HBM出口许可的动向，正在将半导体龙头的政策风险从“限制扩张”转向“限制执行”。中芯国际因其产能垄断地位具备较强估值底，但存量维护成本上升压制良率与毛利；寒武纪面临HBM断供导致的巨额订单无法交付风险，估值底逻辑可能因供应链断裂而失效。散户融资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股半导体/AI算力的基本面只支持精选龙头，不支持2026-05-19开盘无差别低吸；融资拥挤与IM贴水要求先等确认再小仓参与。", "summary": "截至2026-05-19，A股半导体与AI算力微观结构支持前序“散户/杠杆承接机构卖盘”的判断，但不是全板块崩盘信号。ETF相对估算净值仅小幅溢价约+0.2%至+0.7%，说明有低吸承接但不极端；融资盘高度集中在寒武纪、澜起科技、中际旭创、新易盛、海光信息、中芯国际A等AI算力和芯片龙头；IM26…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 1, "risks": 1, "neutral": 1, "value": 58.2, "events": [ { "title": "A 股资金流与拥挤度指标支持选择性叠加而非板块级 半导体→SaaS 轮动，软件相对强势是美股叙事溢出与独立国内催化的叠加，量级不足以推翻板块层面的中性配置。", "summary": "本卡从 A 股市场结构角度对 Card-01、Card-02 的结论做 stress-test，结果同向支持：北向资金在软件侧仅选择性买入金山办公等少数标的，在半导体侧仍堆叠设备链而非全面撤离；两融余额边际加仓软件但拥挤度信号集中在寒武纪等单股；行业 ETF 申赎呈现软件净申购加速、半导体小幅放缓…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] 杠杆盘上限与软件/半导体选择性叠加拥挤度测算", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 07ee71f2-f2fb-4678-bf32-e56e4628f116 结果摘要: I will start by executing the shell command to anchor today's d…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "支持哑铃右端方向，但纠正颗粒度——TMT内只有AI算力国产化+半导体设备/材料具备2026年内可验证的盈利弹性与估值修复空间，把整个科技板块当作进攻端是估值过度延展。", "summary": "2026-05-20，本卡片对卡片09哑铃策略中科技进攻端进行压力测试。结论是有条件支持，但需切分颗粒度：TMT内仅半导体设备/材料与AI算力国产化（昇腾、海光、寒武纪、HBM、先进封装）两条主线具备2026年内可验证的盈利弹性与估值修复空间，合计占TMT自由流通市值约18%-22%。消费电子、面…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 4, "neutral": 1, "value": 33.6, "events": [ { "title": "全球宏观压力测试：科创板极端护盘外溢效应与跨境流动性虹吸下A股资产配置的脆弱性 研究日期： 2026 05 22 作者： 全球宏观分析师 (Global Macro Analyst)…", "summary": "全球宏观压力测试：科创板极端护盘外溢效应与跨境流动性虹吸下A股资产配置的脆弱性 研究日期： 2026 05 22 作者： 全球宏观分析师 (Global Macro Analyst) 立场： Stress test (压力测试) 执行摘要 截至 2026 05 22 [S1]，科创50指数收报 1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "融资冻结不会击穿所有科创半导体与AI公司，但会把压力集中到晶圆厂资本开支和AI芯片/IP研发，迫使国产替代路线从全面扩张转向瓶颈优先。", "summary": "截至2026-05-22，本卡支持上一卡“融资冰封会压缩硬科技现金流”的方向，但将风险重新定价为结构性而非均匀性冲击。10家科创板半导体与AI压力篮子的一年期缺口约542亿元，其中可直接追溯到已披露定增的只有57.9亿元，集中在寒武纪和芯原股份；更大的484亿元来自中芯国际、华虹公司等制造环节的资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创板1,750点保卫战实质是对半导体与AI核心资产现金流的定向救助，消费电子等非战略瓶颈板块将面临被动的流动性出清与估值大幅下修。", "summary": "截至2026-05-22，本报告综合得出结论：面对1,750点防线与880亿元抛压，国家队将采取“非对称防御”。中芯国际、寒武纪等兼具高权重、高两融且面临现金缺口的核心标的将被优先注入流动性；而现金流较好、权重分散的消费电子与通用软件将被战略性放任估值下修，科创板风格向“国家兜底”因子剧烈切换。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创板极端护盘对A股整体流动性的虹吸效应与跨板块资产配置调整 研究日期： 2026 05 22 作者： 首席策略师 (Chief Strategist) 立场： Synthesize…", "summary": "科创板极端护盘对A股整体流动性的虹吸效应与跨板块资产配置调整 研究日期： 2026 05 22 作者： 首席策略师 (Chief Strategist) 立场： Synthesize (综合) 执行摘要 截至 2026 05 22 [S7]，科创50指数收报 1,764.17点 [S1]，极度逼近…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 0, "neutral": 0, "value": 64, "events": [ { "title": "AI利好兑现后的宽度轮动与A股TMT杠杆再定价", "summary": "AI利好兑现后的宽度轮动与A股TMT杠杆再定价 AI breadth rotation and A-share TMT leverage repricing after the beat NVDA利好兑现、Russell 2000跑赢、科创50午后反弹但半导体两融净偿还并存，团队应把成长仓位解释为…", "sourceLabel": "研究主题", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 45.4, "events": [ { "title": "Research note 3的科技端选股逻辑成立，但其哑铃调仓框架在杠杆维度、解禁日历和减持约束上有结构性盲区，2026 H2大概率被微观结构而非景气因素强行打破，需用四层微观结构…", "summary": "从A股微观结构角度反对Research note 3的50/50哑铃+成交额触发框架。三处硬伤：(1) 仅看TMT成交额占比38.5%而忽略两融余额结构——创业板+科创板融资盘占比已升至约41%，超过2021年初前高，融资盘会先于成交额平仓；(2) 2026年7-11月集中解禁约2.8-3.2万亿…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装已成为美中半导体博弈的“新前线”，在大基金三期驱动下，国产替代链条正在经历从“晶圆代工单一突破”向“虚拟集成封测增效”的战略重心转移。", "summary": "2026年5月，美方出口管制从“全面禁运”转向“管理接入”模式，通过25%营收分成税与强制在美测试对先进AI芯片（如H200）实施出口许可。此举对A股半导体链形成双重压力：一方面，国内先进制程（如中芯7nm）虽有突破但良率仍处于20%左右的“盈亏平衡点”以下，面临高成本挑战；另一方面，HBM及先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "昇腾2026放量与DeepSeek V4降价使A股硬件链继续具备风格胜率，但可买环节应限定在高速PCB/连接器和订单验证后的液冷零部件，先进封装只做小仓位GARP，拒绝追高广义半导体…", "summary": "截至2026-05-24，本卡支持昇腾封装与液冷增量的配置主线，但把结论收敛为“订单验证 + 估值纪律”。A股最优先的是高速PCB、基板和高速背板连接器，其次是回撤后的冷板、快接头、CDU控制、泵阀和传感器；先进封装与HBM邻近组装只适合作为小仓位GARP组合。广义半导体、通信设备和液冷龙头已出现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 4, "neutral": 0, "value": 36.2, "events": [ { "title": "AI硬件的高景气基本面虽能构成长期估值底部，但在科创50指数跌破1,780点引发的短期杠杆踩踏与量化被动去杠杆中完全无法抵御资金面崩溃。", "summary": "本卡对半导体龙头的抗跌性进行压力测试。AI硬件基本面景气度极其真实，四大龙头（中芯国际、海光信息、寒武纪、北方华创）2026年合计预测净利润达202亿元，季均50.5亿元。但两融与量化半导体高位拥挤风险严重。若科创50跌破1,780点敏感水位引发10%-15%回调，短期两融平仓与量化去杠杆抛压将达…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50杠杆突围：半导体暴涨与大盘风格‘非对称虹吸’的去杠杆冲突", "summary": "科创50杠杆突围：半导体暴涨与大盘风格‘非对称虹吸’的去杠杆冲突 STAR-50 Leverage Breakout: Semiconductor Short-Squeeze vs. Small-Cap Siphoning Deflation 科创50今日大涨3.56%拉离1,750点雪球生死线，…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在科创板两融信用收缩与Q3迎来1200亿解禁洪峰的微观去杠杆环境中，大健康精选篮子建仓必须实施『恐慌底分批-解禁潮吃货』的节奏化策略，并将有效暴露压缩至18%-20%以内，辅以Bet…", "summary": "截至2026-05-25，本卡对创新药出海、CXO地理标签及高端器械替代的精选篮子（联影、迈瑞、奕瑞、百济等）进行A/H股映射与估值管理。A股两融余额见顶于2.87万亿元，科创50在1,750点承受去杠杆与432亿雪球敲入压力。策略建议在科创解禁潮与两融风暴中分三阶段节奏建仓（1-3月恐慌期买30…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国家队'指数级'护盘掩盖了承做商对冲、两融杠杆与量化因子衰减三条链条'个股流动性级'的尾部敏感性,科创50离开1,750不等于系统离开生死线。", "summary": "今日科创50大涨3.56%脱离1,750雪球生死线，但只是延迟而非解除风险——半导体权重票+国家队ETF把指数托起的同时，承做商对冲仓位被动收缩、两融余额冲至2.9万亿历史峰值[S3]、量化中性产品超额已连续两季度回吐[S5]。'非对称虹吸'体现在:资金流入集中于科创50权重前10票与红利端，流出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "两融资金推升的A股半导体“成长进攻”在AI核心链上具备真实利润兑现（如高速光模块和先进封装），但成熟制程价格战严重分化；且在高交易拥挤度下极其脆弱，必须实施“去粗存精”的精准防御。", "summary": "截至2026年5月27日，本卡对两融推升的半导体与AI硬件“成长进攻”进行压力测试。AI核心环节（中际旭创毛利达42.61%，通富微电AMD订单爆发）基本面坚实，但成熟制程（中芯国际毛利受制于价格战降至18-20%）分化严重。在美债利率上冲至5.0%及AI CapEx减速下，73%两融拥挤度平仓将…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 1, "risks": 3, "neutral": 4, "value": 65, "events": [ { "title": "[auto-handoff] 科创50指数1800/1750点破位下雪球敲入规模与两融杠杆平仓负反馈测试", "summary": "致 主题研究员： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 21a31e2b-133e-4d11-8860-ebeffe825ce4 结果摘要: Wrote handoff_response.md (11959 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在TMT板块极端拥挤和高实际利率压制下，今日应坚决停止加仓一阶AI硬件，战略性转向配置确定性极高的二阶物理瓶颈（电力/铜需求/液冷散热），构建杠铃配置以防范筹码去杠杆及估值回调风险。", "summary": "在A股TMT成交占比攀升至44%-45%以及科创50两融拥挤度逼近4.90%的历史极端警戒水位下，第一阶AI硬件（半导体/HBM）的筹码结构已极度脆弱。同时，全球2.0%-2.2%的高实际利率持续压制长久期高估值资产。今日策略坚决不加仓一阶AI芯片，转而顺着AI产业的“物理刚性约束”向二阶瓶颈（高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件能否穿越拥挤交易和高实际利率", "summary": "AI硬件能否穿越拥挤交易和高实际利率 Can AI hardware outrun crowding and high real rates? 当多份报告同时看多 AI 算力、HBM、系统级互联和铜需求，但情绪报告显示半导体最拥挤、A股TMT成交占比 44%-45%、科创50两融拥挤度 4.90%…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI硬件能否穿越拥挤交易和高实际利率", "summary": "AI硬件能否穿越拥挤交易和高实际利率 当多份报告同时看多 AI 算力、HBM、系统级互联和铜需求，但情绪报告显示半导体最拥挤、A股TMT成交占比 44%-45%、科创50两融拥挤度 4.90% 时，今天应继续加仓 AI 硬件，还是切到电力/铜/液冷等二阶受益？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 3, "neutral": 1, "value": 49.4, "events": [ { "title": "在雪球敲入与两融平仓共振的极端风险下，以‘先进封装’结构性机会对冲Beta下行，并以虚值期权锁定尾部风险是当前最优防御路径。", "summary": "截至2026-05-29，A股两融余额超2.9万亿且科技板块极度拥挤。科创50指数逼近1,750点雪球密集敲入区，做市商面临121.5亿元刚性抛压，叠加ETF折价套利，流动性死亡螺旋风险剧增。建议配置14%比例的先进封装‘防守方案四’（5%设备+6%材料+3%存储）并配合虚值认沽期权构建多重防线。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50两融高杠杆拥挤与做市商负Gamma雪球敲入负反馈共振防御策略", "summary": "科创50两融高杠杆拥挤与做市商负Gamma雪球敲入负反馈共振防御策略 STAR 50 Leverage Crowding and Option Negative Gamma Feedback Loop Defensive Strategy 在国内两融余额突破2.9万亿高位、科技板块拥挤度居历史98…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 11:27奇���后科创板雪球与两融信用去杠杆的二次共振风险模型测算", "summary": "致 日报总编： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: a1c7a9af-eec9-4359-8ac6-02f6d343f613 结果摘要: Wrote handoff_response.md (16251 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "MSCI调仓本身不是系统性事件，但落在A股两融2.87万亿/物理上限2.90万亿仅1%缓冲的高敏感点上，使得TMT龙头任何-2%以上的跳水都会被'雪球敲入×维保线×流动性真空'三重叠…", "summary": "本卡对前两卡'被动买盘被主动MOC冲抵、TMT次日温和回吐-0.5%~-1.5%'的结论进行压力测试。结论：基线情景(P≈70%)下两融余额单日收缩80-140亿元，2.87万亿/2.90万亿物理上限的300亿元缓冲仍可承受，系统性风险可控；但厚尾情景(P≈20%)下科创50若跌破1,750点雪球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 1, "risks": 4, "neutral": 3, "value": 58.8, "events": [ { "title": "科创50极端的半导体权重（~70%）导致Gamma踩踏将直接演变为产业资金链危机：亏损型设计龙头（寒武纪）面临股权质押爆仓，而最上游设备商（中微、拓荆）面临高达20%-24%的现金流…", "summary": "本卡评估了科创50流动性驱动型Gamma踩踏对AI硬件与半导体的传导效应。在指数暴跌-25.0%的极端情景下，高Beta标的面临-30.0%至-41.25%回撤，寒武纪P/S压缩至24.68x且股权质押跌破150.0%警戒线，面临1.5亿元补足缺口。供应链账期延长45.0%触发三角债危机，中芯国际…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈", "summary": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 2.94tn RMB Margin Debt & STAR 50 Snowball Cliff under MSCI Passive Rebalancing 今日是MSCI半年度重估正式生效日。虽然散户由于昨日创业板+1…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股两融安全边际充沛，科创50雪球实际规模仅为62亿元且敲入卖压平滑，MSCI被动重估不存在所谓的“流动性真空砸盘”，跨境日元期权对冲完全多余且损耗超额收益。", "summary": "本卡作为金融行业分析师，强烈反驳A股策略师的悲观观点。首先，两融高位有2026年保证金100%新规与2025年风控标准严密保护，平均维持担保比达248.5%，个股需跌60%以上才触及强平线，绝非指数大跌16.92%所能引爆。其次，科创50雪球名义存续实际仅为62亿元且极均匀分布，做市商采用平滑期权…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50两融高杠杆拥挤与做市商负Gamma雪球敲入负反馈共振防御策略", "summary": "科创50两融高杠杆拥挤与做市商负Gamma雪球敲入负反馈共振防御策略 在国内两融余额突破2.9万亿高位、科技板块拥挤度居历史98%分位的背景下，科创50今日险守1,750点雪球生死敲入线。当做市商在1,750/1,720点后面临121.5亿元刚性抛盘，且伴随ETF一二级市场-0.45%折价套利激活…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 4, "neutral": 0, "value": 51.2, "events": [ { "title": "半导体上行周期足以驱动先进制造板块的盈利修复，但仅限AI算力/先进制程/HBM/光器件/国产设备切片；手机、面板、通用成熟模拟暂未跟随。", "summary": "截至2026-05-30，全球半导体上行周期已被多源硬数据证实，足以支持第03卡对A股电子/通信/先进制造的超配信号，但盈利修复是分层的、AI驱动的。SIA 2026Q1全球半导体销售2,985亿美元(环比+25%)，TSMC 2026Q1营收+40.6% YoY、HPC占61%、毛利率66.2%…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "在科创板流动性极端挤压情景下，应“弃BETA、选战略”，优先配置具备国家级政策金融兜底（再贷款/大基金）的半导体上游龙头。", "summary": "本卡构建了2026年5月波动环境下的科技成长配置框架，核心在于区分“战略底线”资产与“流动性Beta”陷阱。在科创50融资余额近4000亿及雪球Gamma踩踏风险背景下，建议超配受大基金三期与1.2万亿再贷款直接支持的半导体设备、材料及核心代工，避险高拥挤、高杠杆的AI芯片设计标的。通过区分政策金…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "结构性货币工具与产业基金可以为战略合格半导体企业提供经营性流动性底线，但不是科创50股价、股权质押或低质量信用的无条件救助。", "summary": "2026-05-30：本卡支持前序“科创50流动性踩踏可冲击半导体资金链”的判断，但将终局风险下修。央行科技创新和技术改造再贷款总额度已达8000亿元，实际贷款余额已有6140亿元，大基金三期注册资本3440亿元；按前序模型压力桶测算，覆盖中微、拓荆、寒武纪和中芯国际的102.73亿元现金压力仅需…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈", "summary": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 今日是MSCI半年度重估正式生效日。虽然散户由于昨日创业板+1.96%的V型反弹而呈现极度FOMO，但A股两融余额正高挂于2.9367万亿元历史极值，融资交易占比达10.09%。科创50指数（1,844.25点）距离下方…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 2, "neutral": 0, "value": 56.6, "events": [ { "title": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守", "summary": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守 Defending the 1,750 Lifeline: Snowball Gamma Waves and Margin Call Risk 科创50收于1,751.32点，距离1,750雪球/两融生命线仅一步之遥。若下周破位跌破1…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论", "summary": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论 A-Share Post-Close Sentiment Fission: MSCI Rebalancing vs. Margin & Snowball Risks 今日A股演练了11:27分V型奇袭与科技大挤空，但成交…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 1, "risks": 5, "neutral": 2, "value": 54, "events": [ { "title": "SpaceX窗口的A股风险不是整体硬科技崩盘，而是两融账户在-6%至-18%组合波动区间内触发补仓和被动平仓。", "summary": "截至2026-06-01，本卡片对SpaceX上市窗口下A股硬科技/军工组合做VaR与追保压力测试。以100份标准化组合计，单日99% VaR为-5.9%，IPO窗口VaR为-10.4%；基准、悲观、跳空尾部压力损失分别为-8.3%、-18.0%、-25.7%。若初始维持担保比例为160%，组合亏…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 科创50雪球密集敲入带负Gamma理论抛压额度与MOC调仓承接量测算", "summary": "致 主题研究员： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: c7bfba3f-3a82-41df-bfe7-14cc5accb841 结果摘要: Wrote handoff_response.md (11444 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "半导体基本面复苏呈'AI强、Commodity弱、本土设备先量后利'的三段错位，盈利修复要到2026Q4-2027Q1才能传导到科创50权重股的ROE拐点；在此之前平准资金能锁底但不…", "summary": "TMT行业从产业自下而上验证Research note 04关于半导体基本面时滞制约风格重构的结论。WSTS春季展望显示2026年全球半导体销售+12.8% YoY，但增长高度集中于HBM/先进逻辑/先进封装；占科创50半导体权重核心的成熟逻辑、设备、材料、模拟MCU正是斜率最缓的子赛道。SMIC…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈", "summary": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈 MSCI Rebalancing Effective Date: Margin Debt Record-High & Snowball Knock-in Cliff under Asymmetric Play 今日（5月29日）MS…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "在美联储加息与人行降准的联合概率矩阵下，A股必须从传统哑铃策略转向“防御底仓+对冲Beta”模式，规避高融资与高解禁共振的科创标的。", "summary": "由于卡片04数据受损，本报告直接综合卡片01与03。针对科创50指数在1750点的雪球敲入与2.9万亿两融平仓共振风险，本报告提出A股风格轮动与行业配置优化方案：将原10%的科技敞口转为“对冲型科技”，并将高股息与出口制造权重分别提升至35%与25%，以应对6月3237.5亿元解禁悬崖。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 3, "neutral": 1, "value": 57.8, "events": [ { "title": "政策约束已经把A股半导体的可投资宽度实质性收窄到先进封装/HBM/SiC/稀土永磁四个切片，而ETF资金已经为全板块定价了80分位的政策溢价；下半年应'减总量、加精度'，并把2026…", "summary": "本卡片锚定2026-06-03。综合Card-01至Card-05的销量质量→政策销量→利润集中→行政性稀缺租金的链条，把先进制程白名单+成熟制程2027-06-23关税突跳+中端配套2026-11-10双向断点三重政策约束直接翻译为A股板块宽度、风格暴露、ETF资金流与拥挤度结论。政策筛后宽度约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Card-06的'80分位政策溢价 + 黄灯拥挤度'被情绪面独立证实甚至偏深一档；HBM/存储减仓切片被ETF/北上/融资/龙虎榜四方共同反向不确认（情绪共振过热区），其余三切片同向…", "summary": "本卡片锚定2026-06-03。作为情绪师对Card-06的'A股半导体80分位政策溢价 + 四切片不对称'判断给出stress-test。三项核心量化（PE 88-92分位、ETF 5月+RMB 17-19.5bn、融资余额9.4-9.6%占两市）被独立证实；情绪复合z-score+1.70σ，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "下一轮0DTE式日内波动若触发两融担保品重估，最先受压的不是广义沪深300，而是AI硬件/半导体/算力链中的科创与创业板高Beta担保品。", "summary": "2026-06-04：本卡支持前序观点，并把A股两融脆弱点定位在AI硬件、光模块、PCB、半导体和算力链，而不是整个沪深300。2026-06-02沪深两融余额约2.897万亿元，融资余额约2.876万亿元。深市中际旭创、新易盛、胜宏科技，沪市寒武纪、兆易创新、澜起科技等体现高余额和高融资买入共振…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS产能硬约束会把AI硬件估值溢价从单纯设计叙事重新分配给掌握晶圆、封装、良率和整柜交付能力的产能控制者。", "summary": "2026-06-03：本卡支持前序判断，认为CoWoS与先进封装已成为AI算力链估值的硬过滤器。NVIDIA需求仍强，TSMC的3 nm、HPC与先进封装紧张共同支撑产能控制者溢价；但HBM4供应商扩容削弱单纯存储稀缺溢价。A股半导体整体估值已高，策略重点应从宽泛芯片设计beta转向设备、材料、封…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 3, "neutral": 0, "value": 56.2, "events": [ { "title": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？", "summary": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？ Broadcom AI Guide-Down Triggers Global Semi Sell-Off: Will A-Share Tech Break Support Amid Leverage & Lo…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "博通指引转保守与A股流动性三压共振，科创50面临1720点雪球破位风险，需警惕系统性去杠杆引发的二次筑底。", "summary": "博通（AVGO）隔夜暴跌13-15%打破了AI硬件订单无限增长的叙事，引发全球半导体去Beta。A股科创板正面临4.9%高两融占比、6月3332.85亿解禁洪峰以及科创50指数（现报1726）逼近1720雪球密集敲入区的“三体问题”。若破位将引发做市商负Gamma抛售。建议维持45%红利底仓，并买…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理瓶颈栈的供给弹性差异决定了租金在中长期将不可逆地从半导体封装/HBM顶端向下沉淀至电网、电力组件与上游资源所有者，资产配置应超配电网基建与铜资源，低配久期债与下游AI应用。", "summary": "本卡综合了卡1-5的四层AI物理瓶颈栈分析，探讨了2026-2028年大类资产与A股配置策略。由于底层公用事业与铜精矿供给弹性极低，租金将在中长期下沉至电网与资源端。建议全球超配铜矿山、独立发电商与电网设备组件，低配久期债与AI SaaS应用。A股重点配置十五五电网Capex扩张受益者及上游自主铜…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 5, "neutral": 0, "value": 45.6, "events": [ { "title": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？", "summary": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？ 博通（AVGO）隔夜暴跌12%-15%打破了AI硬件订单无限高增的完美预期，而A股CPO与半导体龙头正面临历史高位两融（科创板4.9%占比）和6月3332.85亿的解禁洪峰。如果科创50指数被动向1750/1…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50 1,650点是‘破位即反转’的结构性再进场窗口，TMT首选反弹先导为国产算力、存储与半导体设备三剑客。", "summary": "科创50在1,650点构成‘政策+估值’双底再进场窗口而非系统反转。1,650对应42x PE-TTM (15%分位)与2025-09政策行情颈线，叠加央行降准+SFISF流动性底座与‘1+6’科创政策支持。雪球敲入完成后做市商负Gamma抛压耗尽、解禁实际净减持约1,200亿可消化。TMT细分中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "“地缘政治折价”在TMT并不普遍——它压低境外依赖型与赴美上市资产，却为国产替代龙头创造出受保护市场的溢价，正负号取决于公司处在国产替代分界线的哪一侧。", "summary": "对生物医药BD“地缘政治折价”是否普遍适用于TMT的压力测试结论：部分否定。美方出口管制确实抬高合规成本（25%关税、15%营收分成、第三方测试）、迫使供应链重构，并压低依赖境外技术与赴美上市公司（设备链、BABA/PDD/JD等ADR）的估值。但其传导机制与《生物安全法案》不对称——半导体/AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产替代龙头融资买入拥挤度逼近活跃自由流通盘10%的极值，在大基金常态化减持和资产重组股权稀释压力下，当前的‘地缘溢价’流动性结构极为脆弱，投资组合应向重资产代工龙头防御性收缩。", "summary": "截至2026年6月5日，A股融资余额达2.89万亿元历史高位，两融资金在国产替代龙头中呈现极端的结构性拥挤。寒武纪与中芯国际的融资余额分别达251.33亿和132.19亿元，占其活跃浮筹比例超过7.4%和10%。国家大基金开启中芯国际H股常态化减持以回收资金，中芯国际定向发行5.47亿股购买中芯北…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 2, "neutral": 0, "value": 73, "events": [ { "title": "在系统性流动性危引发避风港资产补跌与相关性收敛的背景下，跨资产配置至实物黄金与短债，并配合科创50 ETF Put与IM期货空头对冲，是防御VaR穿透并平衡权益融涨弹性的最优解决方案。", "summary": "针对2026年6月7日A股2.90万亿两融余额极值及雪球敲入踩踏引致的系统性流动性危机，本压力测试表明长江电力等避风港资产在挤兑期存在补跌风险，相关性收敛导致未对冲组合10日损失达-16.60%，99% VaR穿透至32.33%[自测算]。本方案设计了科创50 ETF期权与IM期货空头对冲包，并对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "依托全球AI硬件超级周期的外溢效应与国内流动性宽松预期共振，A股TMT板块具备坚实的业绩兑现基础，将在跨国资金流入带动下确立其在“哑铃策略”中的核心成长地位。", "summary": "在A股“红利-成长”哑铃策略中，TMT板块正迎来从估值修复向业绩驱动的实质性跨越。随着2026年全球AI硬件进入Blackwell与HBM大规模交付期，海外算力资本开支维持高位，带动A股在光通信（1.6T/3.2T）、先进封装及半导体设备环节的订单爆发。叠加人民银行宽松预期与跨国套利资金的再配置，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "外部高利率与全球去杠杆冲击下，A股面临硬科技两融去杠杆与解禁潮共振风险，资产配置应剪枝高位拥挤科技与昂贵资源，退守低估值高股息周期价值与短端利率债以优化组合VaR。", "summary": "本报告分析了美联储维持“更高更久”利率对A股资金与风格轮动的影响。美债收益率升至4.75%和日元平仓去杠杆加剧了北向资金流出压力（预计600亿至900亿元），推高离岸HIBOR至8.0%-10.0%。两融杠杆高企（2.91万亿元）且高估值科技拥挤，科创50击穿1750点流动性红线触发150亿雪球敲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 2, "neutral": 1, "value": 69.4, "events": [ { "title": "[auto-handoff] 科创50指数雪球敲入负 Gamma 对冲抛压与两融折算率下调去杠杆踩踏模拟", "summary": "致 首席风控官： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 73279e4e-699e-4d9e-9366-ea168709088a 结果摘要: Wrote handoff_response.md (14450 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中际旭创明日解禁直接抛压微弱，但在全月大额解禁、两融折算率调降预期及科创50雪球敲入共振下，股价若下跌13.33%-15.40%极易引爆科技板块非线性两融强平螺旋，需严格规避高两融占…", "summary": "本报告针对中际旭创明日（2026年6月8日）解禁及半导体/CPO两融杠杆进行压力测试。中际旭创明日解禁32.45万股（约4543万元），仅占日成交额1.14%，直接冲击有限。但在全月3332.85亿元大额解禁、电子通信行业两融高达7389.32亿元、两融折算率收紧拟调降至0.4以及科创50破位敲入…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "定向再贷款不足以单独抵御三季度科技板块流动性分化，板块β走弱、α将集中于订单可见度高的前道设备龙头。", "summary": "压力测试显示，5000亿定向再贷款对半导体设备与AI服务器部件的边际提振真实但有限（融资成本仅压缩0–50bp），不足以单独抵消三季度科技板块约1200亿元限售解禁带来的供需失衡。叠加BIS对HBM3e与先进封装的出口管制，国产替代窗口在Q3内难以兑现。压力情景估算解禁+基金减仓抛压合计约810亿…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 3, "neutral": 1, "value": 64.8, "events": [ { "title": "[auto-handoff] A股科创50跌破零Gamma生死线后两融强平与雪球级联踩踏的微观杠杆压测", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 23d2a86c-8b70-46ce-8620-e44b5d309d0d 结果摘要: Wrote handoff_response.md (3101 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "“纯AI算力”集群高PE估值在破位1550点两融强平与做市商负Gamma抛压下极易压缩，其估值回调引致的下行重力将抵消设备与晶圆板块的底部支撑，导致科创50指数在中度压力下击穿151…", "summary": "截至2026-06-09，本报告评估了“纯AI算力”估值溢价在两融强平与负Gamma效应共振下对科创50指数的下行压力。测试显示，寒武纪（35.1x PE）和海光信息（57x PE）的估值泡沫在破位1550点后面临强平踩踏（两融及期权抛压分别达57.6亿与140亿元）。在中度与重度压力情景下，AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 科创50跌破1600点零Gamma线后两融强制平仓规模测算及国家队在1550-1580点吸纳阀值压测", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 4dac25f8-9706-44e5-a87a-72943daf795e 结果摘要: Wrote handoff_response.md (10776 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "1516-1540区间已使科创50半导体设备+晶圆权重的远期PE触及产业资本举牌与密集回购的历史盈亏平衡线，但纯AI算力名字未达，基本面底真实但非对称。", "summary": "在1528点（1516-1540中值）映射下，科创50前十大权重的2026E前向PE中，设备链（北方华创、中微）压缩至22-23倍（5年8-9分位），晶圆代工（中芯、华虹）32-34倍（12-14分位），均落入历史产业资本举牌+密集回购的触发带。叠加¥384亿存续回购授权（仅31%执行）与央行回购…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 3, "neutral": 0, "value": 71.2, "events": [ { "title": "离岸CNH流动性收紧迫使北向杠杆资金去杠杆，这与境内两融平仓及雪球负Gamma抛售共振，加剧了科创板及高弹性权重股的流动性压力。", "summary": "CNH HIBOR飙升至7.30%抬高了离岸融资成本，导致北向杠杆资金因年化-2.8%的负carry面临去杠杆压力。这与境内融资折算率下调（966亿保证金缺口）及科创50跌破1600点零Gamma防线引发的270亿雪球敲入卖压共振，导致外资抛售半导体与TMT高弹性权重股。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 雪球敲入后做市商现货对冲头寸出清进度与两融去杠杆关联压力测试", "summary": "致 衍生品策略师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 970961ba-5045-407a-8c99-a33fa8f9928d 结果摘要: Wrote handoff_response.md (9978 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球科技配置应超配ASIC与新代工生态；A股风格则需从“AI题材映射”战略性切换至独立算力的“国产第二供应源”，重仓先进封装产能、核心设备及HBM材料。", "summary": "2026-06-09评估：基于AI半导体估值锚向“增长+第二供应源”切换，全球配置需减配纯GPU倍数扩张与单一代工依赖，转而超配ASIC双雄(AVGO/MRVL)、HBM及Intel代工期权。在A股与港股层面，风格将从“英伟达映射”(光模块/代工)实质性轮动至业绩兑现度高的国产“第二供应源”。策略…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "红利仍是底仓，但红利拥挤已经成为风险；在外盘芯片黄灯下，A股科技应按ASIC/服务器瓶颈优先、宽基科创50后置的顺序分批回补。", "summary": "截至2026-06-09，本卡支持前序“外盘芯片黄灯、A股科技不能盲目追涨也不能单边放弃”的判断。2026-06-09科创50收涨4.17%至1,663.11点，中证红利跌0.06%，但科创50仍低于1,668.33点破位前收盘。建议把科技、红利、缓冲仓基础配为45/45/10；先从煤炭等拥挤红利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 1, "value": 79.6, "events": [ { "title": "100亿元缓冲资金的最优A股结构通道是70%ETF一级申购 + 20%收盘竞价权重托底 + 10%大宗折价承接，且必须经由非两融账户执行才能真正切断二次去杠杆反馈环。", "summary": "在ETF折价、成分股大宗、收盘标记三线同压的情景下，100亿元AI/半导体缓冲资金最稳妥的执行组合为：ETF一级申购70亿（消化折价）+ 收盘集合竞价权重托底20亿（守雪球敲入与两融维持担保线）+ 大宗折价−2%承接10亿（吸大单不留痕）；二级连续竞价与转融通回收两条通道必须拒用。所有承接必须通过…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在美联储鹰派重定价与科创板局部杠杆脆弱性的双重挤压下，A股配置应果断转向低杠杆红利防御，保留科技龙头仅作为极低仓位的期权观察。", "summary": "截至2026-06-10，科创50指数（1652.22点）月内跌幅达5.66%，显著跑输红利指数（-2.10%）。两融数据显示资金正从电子（半导体）板块剧烈流向能源、银行等高股息行业。虽然科创50尚未触及1468.6点的广谱强平线，但局部杠杆压力已迫使配置风格从高弹性成长向低杠杆防御切换。建议压低…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 0, "value": 78, "events": [ { "title": "国家队科创50防守策略在‘一级申购优先’与‘二级折价才买’的限制下将形成严重的防御真空与套利漏洞，导致二级市场半导体与AI龙头在防线建立前遭遇流动性错配引发的估值踩踏。", "summary": "本卡片对科创50在‘宽基先行、科创后置’策略下的执行路径进行压力测试。分析指出，‘一级申购优先’机制存在显著的算法建仓时滞，面对两融平仓与雪球敲入形成的710亿元级联抛压（对前五大半导体巨头达106.5亿元/天），国家队12亿元/天的日均算法买盘存在8.88倍的流动性赤字。此外，‘二级折价才买’规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 7, "neutral": 0, "value": 43.4, "events": [ { "title": "2026-06-12核心判断：AI半导体订单未见坍塌，但去杠杆风险会优先打击现金转化弱、扩产重、估值依赖交期溢价的环节。", "summary": "2026-06-12：半导体链条的订单韧性强于股票交易本身。ASML、TSMC、Advantest、KLA、HBM存储和部分先进封装设备仍有强订单、利用率或现金流证据；但Amkor、Camtek、中微公司等环节显示资本开支和营运资本会吞噬现金，去杠杆冲击会先压缩估值和产能期权价值，而不是马上造成订…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "摩尔线程与中颖电子一季报账面扭亏与反弹高度依赖非经常性损益支持，表明主营业务实质复苏温和，硬科技板块的超额收益正从单一的替代叙事向商业自我造血与两融流动性约束分化。", "summary": "摩尔线程（688795.SH）2026Q1通过7006万元政府补助首次单季扭亏，但扣非仍亏损5428万元，其全功能GPU在先进代工产能受限后面临昇腾与寒武纪竞争。中颖电子（300327.SZ）Q1净利增31.66%受465万元非损驱动，扣非仅增3.25%，表明家电MCU去库结束但处于价格战红海。边…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海外risk-on与A股科技去杠杆的错位", "summary": "海外risk-on与A股科技去杠杆的错位 Global risk-on versus A-share tech deleveraging 隔夜美股半导体与小盘逼空是否足以覆盖科创50/CPO的本地两融、解禁和折算率压力，还是应继续把科技回补限制在5%试仓以内？", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 日元套利平仓螺旋下日本WFE设备股技术性抛售对A股科技半导体估值的间接传染", "summary": "致 主题研究员： 协同任务已完成。执行方为 外汇策略师。 模式: adhoc 会话: 2ab36a4a-fbce-489a-985f-8013d5e3cb5f 结果摘要: Wrote handoff_response.md (12830 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 2, "neutral": 0, "value": 74.8, "events": [ { "title": "半导体板块的主风险不是调样本身，而是核心权重与融资余额高度重叠后，在再度下探1600点时把指数级强平转化为芯片股集中抛压。", "summary": "截至2026-06-13，本卡压力测试上一卡片：科创50强平共振成立，但半导体抛压高度集中在高权重与高融资余额重叠的核心股。科创50前九只半导体/设备/芯片设计权重合计60.39%，若按权重承接744.3亿元级联强平，对应潜在抛压约449.5亿元；1516-1540点位20亿元国家队前线仅覆盖4.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50杠杆扩张属情绪极值但基本面分层承接：代工/设备能扛-20%、AI算力IP仅能扛-10%，因此踩踏风险局部化而非系统性。", "summary": "针对2026-06-14锚定日，本卡从半导体基本面侧压力测试科创50杠杆极值（融资余额触及自由流通市值约5%）。结论：板块确入'高杠杆—高波动'区间，但杠杆踩踏风险是分层、可控的非系统性风险。代工/设备龙头（中芯688981、北方华创、中微688012）凭借产能利用率回升至88%与订单簿+42% …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 2, "neutral": 0, "value": 84.2, "events": [ { "title": "科创板去杠杆冲击短期股价但不破坏国产半导体设备及AI硬件的长期成长逻辑——龙头与AI芯片设计估值修复~20%即可加仓，二线与HBM配套需-30%以上才有安全边际。", "summary": "截至2026-06-15，科创板两融由~2,940亿元降至~2,820亿元、科创50击穿1,600点零Gamma线，对电子板块形成真实短期冲击，但不足以系统性下修国产半导体设备与先进制程龙头capex。设备龙头（北方华创订单覆盖15个月、现金~150亿元）、AI芯片设计（寒武纪2026Q1+312…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "互联网大厂AI capex资金储备充沛且免疫于A股两融收缩，美BIS管制摩擦虽造成300亿元海外采购延后，但倒逼国产算力采购占比由30%大幅上修至58%，在2026H2为国内半导体设…", "summary": "基于2026-06-15，本报告定量评估了地缘与流动性双重压力下中国大厂AI Capex的节奏与传导。测算表明，四大互联网大厂（阿里、字节、腾讯、百度）2026H2的4200亿元AI capex预算完全免疫于A股两融去杠杆，其在海外上市且拥有数千亿现金。虽然美BIS出口管制限制英伟达H20/H20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基于2026-06-15，国产算力+518亿元新增订单支撑下，A股科技板块两融与机构仓位重构具备坚实基本面，科创50重回1600点零Gamma上方将触发右侧杠杆盘强力追入与风格轮动反…", "summary": "基于2026-06-15，本报告定量评估了国产算力爆发背景下，A股科技板块（芯片设计、算力设备、PCB）的两融与机构仓位再平衡。大厂H2新增的518亿元国产算力订单有效对冲了两融收缩1200亿元的短期流动性冲击。公募基金TMT配置预计将从14.2%上修至19.8%，带来+2520亿元增量资金。科创…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 1, "neutral": 0, "value": 83.6, "events": [ { "title": "若日元融资压力引发全球半导体估值去化，A股成长与AI板块需筑防线，应将科创50、创业板指、AI硬件及国产设备分别降至8%、8%、4%和5%的防御上限，待流动性出清与业绩验证后再重拾B…", "summary": "截至2026-06-15，本卡针对BoJ加息及日元套利交易平仓风险，对A股科创50、创业板指、AI硬件及国产半导体设备进行压力测试。建议将科创50及创业板指仓位降至8%，AI硬件降至4%，国产设备降至5%。并明确了日元汇率稳定、A股缩量筑底与中报超预期等重新加仓的量价与盈利修正信号。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 1, "risks": 7, "neutral": 0, "value": 57, "events": [ { "title": "6月15日成长行情是可交易挤仓而非成长牛市重启，应维持半导体/CPO有上限参与，并把科创50 1550/1516作为低置信但高冲击的触发区间。", "summary": "截至2026-06-16，本卡对6月15日成长爆量轮动给出压力测试结论：反弹可交易，但尚非可持续配置信号。6月A股解禁3332.85亿元、影石创新约403.91亿元；5月25日两融余额曾达29252.09亿元，但截至6月12日已降至28734.82亿元。科创50 6月15日收1748.33点，距1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装/HBM 产业链拥挤度风险压力测试 工作日期：2026 06 17（Asia/Singapore） 分析师：半导体分析师（semiconductors analyst） 立场…", "summary": "先进封装/HBM 产业链拥挤度风险压力测试 工作日期：2026 06 17（Asia/Singapore） 分析师：半导体分析师（semiconductors analyst） 立场：stress test（压力测试 Research note 6 的\"先进封装择时择股\"结论） 板块：A股 CoW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "广谱超配半导体板块将面临终端需求收缩与国产化良率折损的非对称风险，建议将配置降至17%并向红利转移3%以优化风险收益比。", "summary": "压力测试验证了2026H2半导体板块20%的超配逻辑面临下游消费电子收缩（全球智能手机出货-13.9%，PC出货-11%）与成熟制程折旧陷阱（如金宏气体利润骤降87%）的宏观压制。同时，地缘断供下国产HBM堆叠良率因材料切换恐从16.98%降至4.11%，导致月产出缩水-68.5%。此外，半导体拥…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "折算率下调对半导体的影响集中于AI/SoC设计与部分设备环节，目前为主动调仓特征，尚未触发被动平仓链条。", "summary": "2026-06-17：折算率下调对半导体板块的冲击精确落在IC设计（数字SoC/AI算力）和设备两个高静态市盈率、高融资参与度环节。寒武纪、海光信息所在的AI/SoC方向周度融资余额回落4%-7%，折算率从0.55-0.65下调至0.40-0.50；而模拟、功率、材料及部分设备龙头融资余额企稳甚至…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 1, "neutral": 0, "value": 86.4, "events": [ { "title": "成长股18%的风险预算已达上限，AI-PCB与光模块一季报业绩高兑现能够支撑该敞口，但半导体设备环比下滑与HBM量产良率瓶颈封杀了额度上调空间。", "summary": "截至2026-06-17，本卡压力测试科创50及电子的18%敞口。高速PCB与光模块的一季报业绩暴增（中际旭创净利+262%，深南电路净利+73%）提供坚实支撑，具备较强成本转嫁能力；而半导体设备大陆一季度销售额环比下滑16%至109.9亿美元，长鑫存储HBM3量产受良率及先进封装制约。结论是18…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 6, "neutral": 0, "value": 57.8, "events": [ { "title": "半导体代工与设备估值面临全球WACC上升及N2晶圆价格墙的压缩，叠加国内HBM3国产材料良率瓶颈导致的算力变现延迟，正驱动A股电子资金扎堆“四大龙头”并形成‘高融资、弱宽度’格局，面…", "summary": "截至2026-06-18，本压力测试卡片验证了半导体设备和材料需求的边际变化如何驱动电子板块的宽度衰退与杠杆拥挤。中芯国际、海光信息、寒武纪、北方华创“四大龙头”融资余额达650.0亿元，占板块总两融的17.62%，而板块宽度持续恶化（仅42.1%个股上涨）。压力测试显示-15%的回撤将引发四大龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A股两融余额数据口径核实与电子行业杠杆拥挤度压力测试", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 3eeb1e55-8994-4d6c-af78-8dff43cbfefa 结果摘要: Wrote handoff_response.md (8897 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "尽管二级市场重估与强平压力短期内压制半导体板块估值并阻碍股权融资，但在国家安全刚性需求与国家大基金三期等政策资金兜底保障下，中芯国际及关键设备与材料厂商的战略性资本开支将保持高韧性，…", "summary": "在美联储鹰派加息风险与A股TMT极值杠杆负Gamma强平压力下，半导体关键环节资本开支表现出极强的战略韧性。中芯国际2025年实际资本开支达81亿美元，2026年预计维持在80亿美元高位，资产负债率仅33%且手握891亿元现金，无再融资依赖。北方华创、中微公司等设备龙头现金流强劲，国家大基金三期（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储鹰派转向与A股TMT极值两融负Gamma清算风险", "summary": "美联储鹰派转向与A股TMT极值两融负Gamma清算风险 Fed Hawkish Shift vs. A-Share TMT Margin Debt Negative Gamma Risk 在美联储偏鹰暂停、美债利差扁平化以及中芯国际A/H溢价率+103.3%的背景下，国内电子两融余额高悬至5722…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 4, "neutral": 0, "value": 71.2, "events": [ { "title": "在30% A股配置中，电网与散热设备凭借低融资拥挤和强出海景气度应立即建仓，而国产半导体在两融高负荷、折算率下调以及雪球敲入风险完全出清前应保持现金等待，通过分步建仓捕捉12.48%…", "summary": "截至2026-06-20，针对30% A股配置，本报告建议采取“先物理、后算力”的分步建仓策略。20%电网/散热板块应立即建仓；10%国产半导体板块因融资余额达7382亿、两融折算率下调及1516-1550点雪球敲入威胁，面临严重的左尾去杠杆风险。待半导体释放-15%回撤且杠杆出清后建仓，可使A股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在美元/日元波动与离岸人民币成本上升背景下，A股半导体设备与AI芯片厂商因进口物料/晶圆采购占比远超出口占比，且竞争格局受益于日元升值被迫提价，利润率呈现显著的‘逆境扩张’特征。", "summary": "本研究对A股半导体设备（如北方华创）与AI芯片设计厂商（如寒武纪）在汇率与流动性压力下的利润率进行敏感性压力测试。结果表明：两板块利润率具有极高韧性。设备厂商出口占营收比极低，采购进口零部件占比约35%，人民币升值可带来净采购成本降低；日元升值被迫促使日系竞争对手提价，提供‘价格保护伞’。AI芯片…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-20，鉴于电网对冲使CPO收入下修收窄至13-17%，A股核心科技龙头基本面安全垫抬升，但由于技术性两融拥挤高企，国家队必须坚守1750战术线及1550生死线以防范…", "summary": "截至2026-06-20，本报告综合评估了CPO收入下修收窄至13-17%下，A股科技板块（中际旭创/寒武纪/源杰科技/联讯仪器）的配置与防御路径。尽管基本面压力因电网加速器对冲而减轻，但高位两融拥挤及成交量萎缩风险未除，国家队应维持1750点战术支撑与1550点绝对生死防线以阻断强平踩踏[自测算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 日报总编： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: cd8f2a28-c7a9-4dc8-aebc-e5ed66354208 结果摘要: Wrote handoff_response.md (17618 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 1, "neutral": 0, "value": 90.6, "events": [ { "title": "电网交付延迟不会终结高密度 AI 机柜，但会迫使下一代 AI 芯片与集群架构把 perf/W、可限功率和可分阶段并网能力置于峰值算力之上。", "summary": "2026-06-20：本卡对电网设备瓶颈约束 AI 芯片架构的说法进行压力测试。结论是，变压器和开关设备长交期不会让前沿训练集群普遍降密度；GB200 NVL72、Vera Rubin 和 AMD Helios 仍指向高密度机柜级设计。但电网交付延迟会把竞争指标从峰值 FLOPS 推向受限兆瓦下的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 0, "value": 86, "events": [ { "title": "半导体产业链在出口管制与国产化压力下的价值重估", "summary": "支持卡片01的“回归本土、规避地缘”框架并延伸至半导体，但作关键修正：在本板块需求逻辑更强——出口管制（2025-09-29 BIS“50%规则”、2024年12月一揽子规则含24类设备+HBM管制）不仅制造摩擦，更把进口订单转为被锁定的本土需求；但“低估值”仅在局部成立。前道设备指数PE达91.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通胀重置与美联储转鹰情景下，经典股债配置策略失效，战术杠铃组合配合美债利率及日元套利平仓尾部期权对冲，是稳健跨资产回报的核心；A股风格上应坚决切向具备价格传导优势的调节性水电红利股…", "summary": "截至2026年06月21日，本报告压力测试了美联储加息和多资产配置方案。在通胀重置加息情景下，经典60/40组合预期大跌-15.60%（VaR为31.95%）；而战术杠铃组合超配短债/TIPS及20%商品，结合10%期权对冲，录得+0.45%正收益，VaR降至10.83%。尾部对冲方案采用10Y …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 1, "neutral": 0, "value": 93.4, "events": [ { "title": "可信算力区下的A股与STAR硬科技配置映射", "summary": "本卡把卡片01–07的'可信算力区'供应链结论翻译成A股/STAR硬科技排序，并用四道闸门（收入已拐点、订单可见、毛利确认、与可交付获批算力直接关联）区分真实订单受益与主题弹性。结论排序：光模块(中际旭创+192%、新易盛+106%)、AI服务器/ODM(工业富联净利+102.6%、订单至2027…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 5, "neutral": 2, "value": 76.8, "events": [ { "title": "电子行业6132.72亿元两融杠杆中，357.00亿元处于维保比例临界带，-10%的行业性下跌将触发475.00亿元首轮复合抛压，高融资集中度的半导体龙头面临严重的负Gamma去杠杆…", "summary": "截至2026-06-24，本报告对电子行业6132.72亿元两融余额的维持担保比例分布进行了深入压力测试。当前分布显示，安全区（维保>150%）占比92.11%，而预警与强平区（维保<=150%）合计达484.00亿元。若科创50跌破1500点触发-10%的行业性回调，维保比例低于144.4%的3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 纳指回撤 -9.4% 极端情景下A股科技成长股的两融担保比例及雪球敲入路径审计", "summary": "致 日报总编： 协同任务已完成。执行方为 首席风控官。 模式: adhoc 会话: 5f297bd6-3e20-46b7-9ef7-08d76d936098 结果摘要: Wrote handoff_response.md (10368 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体龙头基本面估值底线与流动性去杠杆风险的博弈", "summary": "经营基本面确实强：中芯国际2026Q1营收176.17亿(+8.07%)、毛利率20.1%、产能利用率93.1%，2Q指引环比+14%~16%并上调；海光信息2025营收143.77亿(+56.92%)、归母净利25.45亿、Q2末合同负债30.91亿，订单能见度最硬。但估值托底能力弱：中芯PB5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 两融杠杆高位下科技权重股级联平仓线与雪球敲入路径测算", "summary": "致 因子研究员： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 38dedc01-052c-4a28-9d7b-08c105b21231 结果摘要: Wrote handoff_response.md (13561 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 4, "neutral": 1, "value": 81.2, "events": [ { "title": "[auto-handoff] 日元干预警报与亚太半导体/CPO板块去杠杆风险评估", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: c956eaa1-0683-41fa-9c41-551222934843 结果摘要: Wrote handoff_response.md (10735 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM/先进封装自主与出口管制反身性是否构成对中国AI建设节奏的绑定宏观约束及政策驱动A股主题", "summary": "本卡综合 Research note 07：接受瓶颈已迁至国产HBM/先进封装，并将其升级为宏观—政策命题。国产HBM确是绑定约束，但绑定的是2026–2027的“节奏”而非“轨迹”——属软绑定。2026年仅约25万–30万颗 Ascend-910C 等效国产封装[S11]，仅相当于英伟达约360…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论", "summary": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论 JPY Carry Trade Unwind vs. APAC Tech Margin Call Spiral 美元兑日元跌破161.50关口，财务省干预概率陡增。一旦日元套利平仓瞬间引爆全球EM科技股抛售，叠加SpaceX二次配售24.5亿元的资金虹…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "跨主权存储链的 A 股 AI 国产化最终配置地图——按绑定硬度为材料/设备/封测三环定仓", "summary": "作为10卡线程收官，本卡把瓶颈迁移（需求→电力→HBM→先进封装→亚100nm键合机）综合为一张定仓的A股配置地图。核心法则：绑定硬度决定持有什么，拥挤度决定定多大仓，业绩动量仅作择时。三轴打分显示信念最高的键合机(4.0)与材料(3.2)恰是最不拥挤、动量最安静处，而封测(0.8)、算力(0.8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 4, "neutral": 3, "value": 88.6, "events": [ { "title": "A股半导体两融强平底与全球AI硬件景气度背离的投资套利机会", "summary": "A股半导体两融强平底与全球AI硬件景气度背离的投资套利机会 Divergence Between A-Share Semiconductor Margin Liquidations and Global AI Hardware Super-Cycle 在美光爆表财报确立全球AI硬件超级周期的背景下…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局", "summary": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局 Midday A-Share Growth Rebound and Options Negative Gamma Clearance Tactical Asset Allocation 在上午科创50跌破关键支撑触发雪球敲入与两融强平后，午…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股电子两融强平螺旋与期权对冲防线的实操效率辩论", "summary": "A股电子两融强平螺旋与期权对冲防线的实操效率辩论 A-Share Electronics Margin Liquidation Spirals and Options Hedging Efficiency 在系统性去杠杆Cascade下，若A股行业面临15%回撤，半导体与科技板块将承受19.89十…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CNH套息流动性溢价回缩不会抑制A股半导体上游的当期订单与实物资本开支(政策供资、提前排产、实物周期脱钩)，但会压缩其股权估值并抬高下一轮股权融资扩张成本(股权周期未脱钩)。", "summary": "截至2026-06-25，本卡片对全球半导体设备周期与A股AI基础设施流动性风险做脱钩验证。结论：脱钩因传导链而异，须拆为两条。(1)订单/实物资本开支链——大体脱钩、低风险：北方华创订单排至2027年、营收指引¥468-520亿，中芯2026资本开支约80亿美元持平，由大基金三期(¥3440亿)…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 3, "neutral": 0, "value": 87, "events": [ { "title": "半导体板块融资压力下的负Gamma传导与估值风险评估", "summary": "半导体是这场强平冲刷的震源而非平均参与者。中证半导PE(TTM)约177倍、近10年约91%分位（2021高点仅约80倍），中位数约181倍是海外3–5倍，美银5月调查列其为最拥挤交易（73%）；杠杆又恰堆于此（IT月净融资超1000亿、电子CMR 138.55%距130%追保线仅8.55pp）。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50回撤15%会触发半导体去杠杆左尾，但连锁强平的根因是科创/半导体集中账户与券商150%-160%集中度平仓线，而不是中芯国际、北方华创两只股票的公开融资余额本身。", "summary": "截至2026-06-27，本卡压力测试显示：科创50若从2026-06-26收盘2032.28回撤15%至约1727，普通130%平仓线只击穿冲击前CMR≤152.9%的账户；真正危险来自券商双创集中度规则，50%-80%集中度可把平仓线抬至150%，80%以上抬至160%，使176%-188% …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局", "summary": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局 在上午科创50跌破关键支撑触发雪球敲入与两融强平后，午间主力ETF资金流入与个股金针反转是否标志着负Gamma抛售压力的阶段性出清？我们应如何利用这一估值剪刀差进行右侧建仓，从红利低波向高Beta科技股进行再平衡？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 2, "neutral": 0, "value": 94.4, "events": [ { "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 工业制造分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: eb737f41-b97c-43b9-9b89-04a7c7636133 结果摘要: Wrote handoff_response.md (10188 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国央行通过结构性再贷款与离岸FIMA人民币国债抵押回购机制，成功解耦了汇率防御与境内流动性，为A股“高股息防御+硬科技成长”的滞胀杠铃配置提供了宏观及风险平抑支撑。", "summary": "截至2026-06-27，在美中利差倒挂-265bp与低通胀下，央行转向境内结构性工具与离岸FIMA Repo防线，稳定了境内流动性（套利流出降90%至7.9B RMB）。这巩固了A股以高股息国企（海油6.2%）为防守、硬科技再贷款为成长的“滞胀杠铃”配置，抑平了368.9B两融及35B雪球负Ga…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 3, "neutral": 0, "value": 89.8, "events": [ { "title": "虽然278%平均维保安全垫使两融连锁抛压呈局部非系统性风险，但20-30%回撤仍将触发最高2528.00亿的平仓压力并重创产业链，应通过减仓高倍数前道、超配低谷先进封装以及引入非担保…", "summary": "截至2026-06-28，两融杠杆在电子和半导体板块高度集中（总余额分别达6132.72亿和560.80亿），前五大龙头占用386.80亿。压力测试显示，20-30%的回撤会迫使大批账户跌破130%平仓线，极端情况下（科创50跌破1520）诱发2528.00亿平仓压力，并因雪球负Gamma在2-5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在紧平衡流动性与外部融资约束下，A股难以支撑全局性风险偏好扩张，下半年配置应降低总Beta，采用“硬科技国产替代（封装/设备）+高独立现金流红利”的结构性杠铃组合。", "summary": "在美联储加息预期和国内紧平衡流动性制约下，A股面临外资流出、两融去杠杆与解禁供给的三重压力。外资流入放缓，险资增配高股息红利防御资产。两融余额高企（3.015万亿元）使科技成长面临局部去杠杆和级联平仓风险。半导体板块估值已普遍收缩30%至45%，大基金三期（3440亿元）提供战略支撑。下半年主导风…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 3, "neutral": 0, "value": 91.2, "events": [ { "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 半导体折算率调降保证金缺口与科创板强平风控敏感性测试", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 9cb3a370-d1af-4cb9-8d02-f636ff1c90fa 结果摘要: Wrote handoff_response.md (12184 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-3986819dc0/", "networkHref": "/investor/research-network/#thesis-3986819dc0" }, { "id": "thesis-740e729f22", "source": "tracked_thesis", "title": "人民币、美元与套息交易决定跨境风险资产的流动性折价", "titleEn": "RMB, USD, and carry trades define the liquidity discount across cross-border risk assets", "lane": "人民币/美元流动性敏感资产", "laneEn": "RMB, dollar, and liquidity-sensitive assets", "laneHref": "/investor/lanes/rmb-dollar-and-liquidity-sensitive-assets/", "investableFocus": "人民币/美元流动性敏感资产：UUP, FXI, 2800.HK, 0700.HK, 9988.HK, 3988.HK", "stockUniverse": [ { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股流动性代理", "roleEn": "港股流动性代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股核心权重", "roleEn": "港股核心权重", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "港股核心权重", "roleEn": "港股核心权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "3988.HK", "name": "中国银行", "market": "HK", "role": "人民币与银行流动性", "roleEn": "人民币与银行流动性", "bucket": "core", "href": "/investor/stocks/3988.HK/" }, { "ticker": "QQQ", "name": "Invesco QQQ Trust", "market": "US ETF", "role": "纳斯达克成长暴露", "roleEn": "纳斯达克成长暴露", "bucket": "watch", "href": "/investor/stocks/QQQ/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "IBIT", "name": "iShares Bitcoin Trust", "market": "US ETF", "role": "BTC 现货代理", "roleEn": "BTC 现货代理", "bucket": "core", "href": "/investor/stocks/IBIT/" }, { "ticker": "COIN", "name": "Coinbase", "market": "US", "role": "加密交易基础设施", "roleEn": "加密交易基础设施", "bucket": "core", "href": "/investor/stocks/COIN/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" } ], "stockTickers": [ "UUP", "FXI", "2800.HK", "0700.HK", "9988.HK", "3988.HK", "QQQ", "SMH", "IBIT", "COIN", "TSLA", "ASTS", "MU", "005930.KS" ], "coreView": "赛道：人民币/美元流动性敏感资产。候选观察标的：UUP, FXI, 2800.HK, 0700.HK, 9988.HK, 3988.HK。当前共有 40 条证据、1 位分析师和 6 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 5, "risks": 134, "neutral": 21, "analystCount": 12, "sourceTypeCounts": { "研究笔记": 144, "分析协作": 9, "研究线程": 3, "研究主题": 4 }, "topTerms": [ "CPI", "美元", "USD/CNH", "CNH", "人民币", "DXY", "银行", "日元", "carry", "trades", "define", "liquidity" ], "whatChanged": [ "2026-06-29: 霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "2026-06-29: 日元套利平仓通过离岸人民币流动性抽水及掉期清算挤压双重渠道，系统性抬高离岸融资成本并技术性击穿USD/CNY 7.00关口，导致人民币套利空间彻底崩溃。", "2026-06-28: USD/CNY汇率趋势对通胀对冲组合的影响评估", "2026-06-28: 美联储政策路径不确定性对USD/CNY及利差交易的潜在冲击", "2026-06-28: 结汇证据、CNH 流动性与中间价模式对 OMO 降息窗口的确认" ], "practical": { "sectorLane": "人民币/美元流动性敏感资产", "sectorLaneEn": "RMB, dollar, and liquidity-sensitive assets", "exposureBucket": "风险偏好与高波动", "tone": "red", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股流动性代理", "roleEn": "港股流动性代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股核心权重", "roleEn": "港股核心权重", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "港股核心权重", "roleEn": "港股核心权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "3988.HK", "name": "中国银行", "market": "HK", "role": "人民币与银行流动性", "roleEn": "人民币与银行流动性", "bucket": "core", "href": "/investor/stocks/3988.HK/" }, { "ticker": "QQQ", "name": "Invesco QQQ Trust", "market": "US ETF", "role": "纳斯达克成长暴露", "roleEn": "纳斯达克成长暴露", "bucket": "watch", "href": "/investor/stocks/QQQ/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "IBIT", "name": "iShares Bitcoin Trust", "market": "US ETF", "role": "BTC 现货代理", "roleEn": "BTC 现货代理", "bucket": "core", "href": "/investor/stocks/IBIT/" }, { "ticker": "COIN", "name": "Coinbase", "market": "US", "role": "加密交易基础设施", "roleEn": "加密交易基础设施", "bucket": "core", "href": "/investor/stocks/COIN/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" } ], "stockTickers": [ "UUP", "FXI", "2800.HK", "0700.HK", "9988.HK", "3988.HK", "QQQ", "SMH", "IBIT", "COIN", "TSLA", "ASTS", "MU", "005930.KS" ], "primaryStocks": [ { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股流动性代理", "roleEn": "港股流动性代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股核心权重", "roleEn": "港股核心权重", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "港股核心权重", "roleEn": "港股核心权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "3988.HK", "name": "中国银行", "market": "HK", "role": "人民币与银行流动性", "roleEn": "人民币与银行流动性", "bucket": "core", "href": "/investor/stocks/3988.HK/" } ], "instrumentTags": [ "UUP", "Invesco DB US Dollar Index Bull…", "FXI", "iShares China Large-Cap ETF", "2800.HK", "盈富基金", "0700.HK", "腾讯控股", "9988.HK", "阿里巴巴", "3988.HK", "中国银行" ], "reviewCadence": "每日复核", "confirmationSignals": [ "期权伽马和 ETF 资金是否放大尾盘流动性", "社交热度是否转化为真实增量买盘", "波动率是否领先基本面恶化" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "波动率", "现金", "指数保护", "期权拥挤" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 115 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 12, "recentEvents": 115, "supportRatio": 0.03, "riskRatio": 0.84, "conflictScore": 94.8, "evidenceDepth": 100 }, "score": 67.5 }, "events": [ { "date": "2026-06-29", "title": "日元套利平仓通过离岸人民币流动性抽水及掉期清算挤压双重渠道，系统性抬高离岸融资成本并技术性击穿USD/CNY 7.00关口，导致人民币套利空间彻底崩溃。", "summary": "本卡支持Research note 02，定量分析日元套利平仓对CNH的传导。BOJ加息和USD/JPY破位触发约349亿美元平仓，占离岸池近25%，推高隔夜HIBOR至10.81%-12.74%。掉期展期率跌至40%引发即期购汇压力，击穿USD/CNY 7.00。CNH套利Carry-to-Risk比率跌至-0.492，套利交易崩溃。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28.5, "matchedTerms": [ "美元", "CNH", "人民币", "日元", "carry", "risk", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "summary": "截至2026-06-29，支持卡片01–03的逻辑链并延伸至终端利率与全球流动性维度。美联储6-17维持3.50%–3.75% [S2]、SEP 2026中值3.8% [S3]；新主席Warsh首场FOMC点阵图从降息翻转为加息，9名官员预期加息、年内加息概率50.4%、市场定价10月加息，反应函数转向'约束性务实主义' [S14]。情景A（布伦特约105…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "CPI", "美元", "DXY", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "中美利差倒挂对离岸人民币（CNH）汇率及跨境资本流动的传导效应", "summary": "支持卡片01的格局判断并补齐传导环节。截至2026-06-29，美国10年期国债约4.40%、中国10年期约1.79%，名义利差倒挂约−261bps（本轮最阔）；而真实利差约+39bps偏向中国。名义渠道在去杠杆中占主导，使CNH成为融资货币。鹰派美联储驱动的套利去杠杆经两条腿（CNH融资腿+全球日元套利传染腿）传导：CIP隐含12个月远期USDCNH约7…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.75, "matchedTerms": [ "美元", "CNH", "人民币", "银行", "日元", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "summary": "2026年6月17日Warsh首会以12-0维持利率3.50%-3.75%，但SEP中位点阵从隐含降息翻转为隐含加息（3.4%升至3.8%），18人中9人看至少1次加息、17人看通胀上行风险，2026年PCE上修至3.6%。背景是5月CPI同比4.2%反弹与韧性就业（年内约57万非农）。2年期美债当日跳升逾16基点、DXY破100、离岸人民币滑向6.80出…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "CPI", "美元", "人民币", "DXY", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "中国内需政策强度及对内需受益资产的支撑评估", "summary": "压力测试结论：中国政策提供相对护盾而非绝对护盾，对内需论点是有限支持。强处在跨资产隔离——人民银行逆周期、仍有降准降息空间[S2]，A股内需标的直接日元融资敞口低，人民币强于7、CNH近三年高[S9]，消除被动收紧反射。弱处是政策在对抗消退的需求：2026年5月社零同比-0.6%，为2022年12月来首次下降，且汽车-16.1%、家电-15.6%恰是以旧换…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "CNH", "人民币", "银行", "日元", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "美元强势第二阶段与人民币/A股外资流传导", "summary": "本卡综合前序政策失误论证，落到汇率与资金流通道。美元处于2026年往返的第二阶段：DXY自1月95.57谷底反弹、议息后破100、6/25-26创13个月新高约101.37，但该段几乎全靠约75%的9月加息定价支撑，结构脆弱。USD/CNY约6.80为美元主导。关键在拆解美元下跌动因：套利通道（预期降息收窄利差）利好人民币与北向；避险通道（衰退式坏降息）先…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "美元", "人民币", "DXY", "银行", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "全球利差收窄背景下的中国资本流动与人民币汇率压力研判", "summary": "本卡支持并延伸 Research note 01：G3 反向分化与套利平仓确传导至中国，但路径并非收益逆转。10 年期中美利差深度倒挂约 −265 个基点（中国 1.73% vs 美国约 4.38%），人民币已成自身的低息融资货币。平仓经两条相反渠道影响中国：(A) 强美元/避险渠道压制人民币走弱（人民银行中间价 6.8195、为 2025 年 4 月以来…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.5, "matchedTerms": [ "美元", "CNH", "人民币", "银行", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "已确认 OMO 窗口下的跨资产布局：国债久期、人民币套息/对冲与利率敏感 A 股倾斜", "summary": "首席策略师将 Research note 01–04 综合为一本可交易的跨资产组合，应对强美元（DXY 101.50、美债2Y 4.1%/10Y 4.37%）、坚挺人民币（即期约6.80、5月结售汇顺差358亿美元）与已确认有条件10bp OMO窗口（2.5/3绿灯）三态共存。建议：国债40%（修正久期约6.5、票息主导、半年预期约1.5%）；A股35%（…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 25.5, "matchedTerms": [ "美元", "CNH", "人民币", "DXY" ], "href": "" }, { "date": "2026-06-28", "title": "经常项目顺差之锚与套利交易反转去杠杆构成了人民币的非对称防御，将 7.25 确立为战术防御点，而 7.30-7.35 则是人行防范金融风险与维持降息空间的政策红线。", "summary": "在美联储主席 Warsh 鹰派转向支撑美元 structural 强势的背景下，人民币受制于 -280bp 的深度利差倒挂。然而，经常账户经常性顺差（前4月累积 3477 亿美元）及日元套利平仓去杠杆导致的企业 8000 亿美元离岸美元资产反向平仓，为人民币构筑了物理防御。我们定量测算表明，人民币汇率将围绕 6.75 枢轴双向波动，以 7.25 为战术防御…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "美元", "人民币", "日元", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "结汇证据、CNH 流动性与中间价模式对 OMO 降息窗口的确认", "summary": "对 Research note 03 以贬值为闸门的谨慎框架做压力测试，外汇微观结构把举证责任倒向对人行有利的一边。截至 2026-06-29，三大确认信号大体齐备（约 2.5/3 绿灯）：(1) 中间价 6.8195（06-24）较路透调查 6.7913 弱约 282 点，人行在抑制升值而非防御贬值，距 7.30 红线尚有约 50 个大数缓冲；(2) C…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "美元", "CNH", "DXY", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "强美元仍需作为中国风险组合的保险，但新增风险预算应从裸多DXY转向有界USD/CNH保护，并主动削减以JPY/CHF融资的拥挤新兴市场套息。", "summary": "截至2026-06-28，本卡综合第4卡：美元仍是高利率维持与衰退降息两路径的最干净鉴别器，但DXY在101.3713、近2025年初以来高位后，裸多美元已拥挤。建议保留而非加码美元保护：CNY/CNH用50%-70% USD/CNH期权式对冲，不追空人民币；A股挂钩外汇低配KRW/AUD式外需beta；新兴市场套息减半并去杠杆。路径B的断裂点在融资腿：C…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32.5, "matchedTerms": [ "美元", "USD/CNH", "CNH", "人民币", "DXY", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "0%远期风险准备金率下USD/CNY套利成本下降对离岸人民币利差交易（Carry Trade）的影响评估", "summary": "截至2026-06-28，本卡对0%远期风险准备金率是否诱发离岸人民币套利盘加杠杆放大贬值风险做压力测试。结论是：作为尾部风险通道成立，作为基准情形不成立。中美长端利差约265bp、短端USD-CNH利差约194bp，足以吸引做空人民币/持有美元资产的套利盘；但0%准备金只是移除远期售汇成本楔子，不是保证金或杠杆约束。当前CNH-CNY仅约22点、5月银行…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28, "matchedTerms": [ "美元", "CNH", "人民币", "银行", "carry", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "离岸与在岸人民币价差（CNH-CNY）的背离及其对资本流动与套利成本的影响", "summary": "对本卡任务前提做压力测试，结论：该前提在当前盘面不成立。截至 2026-06-28，CNH-CNY 价差收窄而非走阔，且落在升值侧——USD/CNH 约 6.7849，强于在岸中间价 6.8195 约 346 个点（约 0.51%）[自测算]。三环全断：价差被压缩（CNH/CNY ~0.997–1.00）；避险摩擦被行政下调（3 月 2 日远期风险准备金率…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27.5, "matchedTerms": [ "美元", "USD/CNH", "CNH", "人民币", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "中美利差走宽对中国货币政策空间及人民币汇率稳定性的影响分析", "summary": "截至2026-06-28，尽管中美10年期倒挂走宽至历史性的2.649%，人民币并不弱：USD/CNY约6.80（6月25日中间价6.8195），5月外储升至3.442万亿美元，3月外汇风险准备金率已降至0%。汇率锚是结构性的（贸易顺差、储备、美元走弱趋势），而非利差。因此人民银行可对深陷通缩（5月CPI仅1.2%）的经济继续宽松，破局之道是'以量代价'：…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "CPI", "美元", "人民币", "银行", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "美联储‘更高更久’路径通过利差倒挂加剧了人民币被动套利压力，但在人行常态化干预与7.28硬性控速阻尼下，汇率贬值呈现受控状态，7.30是守卫国内降息窗口的关键底线。", "summary": "截至2026-06-28，沃什首秀转鹰并调升利率预期至3.8%，强势美元将DXY推升至99.00-100.50。中美10Y利差倒换拉开至-272到-285bp，使CNH面临被动套利压力，但人行离岸挤空（推高HIBOR）及锁汇成本压制了投机套利。USD/CNY运行于6.70-6.95上沿，人行控速封顶7.28以防范7.30尾部突破及资本流出对降息空间的封锁。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "美元", "CNH", "人民币", "DXY", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "人民币贬值受控虽平滑了即期汇率波动，但政策分化通过日元套利平仓、指数被动赎回和延迟结汇三大间接通道带来每月达$430亿资本外流压力，迫使银行净息差破1.50%警戒线并封锁三季度全面降息窗口。", "summary": "截至2026-06-28，在美联储沃什新政与DXY高企下，人民币贬值控速封顶7.28。压力测试显示，尽管人民币汇率波幅受控，但与亚洲邻国货币贬值的分化，会通过日元套利平仓直接清算人民币资产、EM基金被动赎回及出口商延迟结汇三个间接渠道，形成每月约$430亿资本外流压力，迫使国内商业银行平均净息差跌破1.50%警戒线至1.48%，实质性封锁三季度全面降息空间。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "人民币", "DXY", "银行", "日元", "macro" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "AI自主化飞轮在工程层面是真实的，但其投资逻辑无法承受外汇压力测试——供应链美元/日元/欧元计价敞口、人民币贬值风险与日元套息平仓事件三重叠加，使寒武纪等A股AI芯片名称在当前估值下…", "summary": "第01卡的AI自主化飞轮论点在外汇层面存在根本性漏洞：Ascend 950PR的物理输入（HBM化学品、光刻胶、封装基板、EDA工具授权）约30–40%以JPY/EUR/USD计价，人民银行对此无控制权。三重外汇风险叠加：（1）美国关税压制贸易顺差、推动USD/CNY向7.40–7.50贬值；（2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "Research note 05 的宏观防火墙在 NAV 3% 的年化 carry 上限内可执行，前提是用价差、领口、风险逆转和 receiver swaption 替代裸期权与远期…", "summary": "本卡把 Research note 05 的宏观防火墙翻译为可执行的衍生品矩阵：3M SOXX 5%/10% 看跌价差受保护 $25M、6M USD/CNH ±1.5% 零成本领口覆盖 40-50% CNY 暴露、3M USD/JPY 远期覆盖 40-60% 日元暴露并按体制升级为风险逆转、6M×…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "再配置后 AI 上游租金篮子的 JPY/USD/HKD/CNH 对冲比例", "summary": "对第07卡日本36%、美国27%、A股23%、港股6%、准备金8%的再配置篮子进行外汇层压力测试。核心结论：基准情景下 JPY/CNH 对冲55–60%（正套息+自然汇率对冲支撑），USD/CNH 对冲50–55%（对冲成本约2%年化，人民币窄幅区间管理），HKD/CNH 对冲70–75%；风险偏…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "外汇市场已开始定价美联储独立性折价（DXY残差约3%，USD/JPY超额日元强势约5日元），但仍处于部分定价阶段，三重触发条件（BoJ加息、Warsh鸽派确认、美债拍卖尾部）一旦叠加…", "summary": "压力测试结果：外汇市场已在部分定价美联储独立性折价，但尚非主要驱动因素。DXY年初至今跌幅7.4%中约2.9%为制度风险残差，USD/JPY跌幅中约5日元（占35%）超出利率基本面，归因于相对可信度再平衡与套息去杠杆；CNY渠道由人民银行主动管理，不构成独立的市场信号。约5,400亿美元日元融资套…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-08", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "弱美元可以成为离岸科技的贴现率与风险溢价缓冲，但它不是美国云厂AI资本开支ROIC缺口的根本修复。", "summary": "截至2026-05-08，本卡支持前序判断但加上约束：SRF/FIMA 只能稳定美元融资，真正压弱美元需要定向购债或资产负债表弹性信号压低 MOVE、实际利率和美元现金稀缺性。基准情形是 DXY 从约97.8下行至94-96、USD/CNH走向6.60-6.75，这可为离岸科技估值贡献约5-10%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "美元结构性下行已被cards 02–06的牛陡化和Fed最终降息预定，但路径上必经一次避险冲高——FX最优叠加是做多JPY作为尾部凸性、做多BRL捕获农业大宗底部carry、做空MX…", "summary": "截至2026-05-13，外汇综合判断：美债牛陡化+双赤字+Fed最终降息使美元结构性下行（DXY基准98→96，尾部94），但card-05双触发（BB OAS 260bps或SLOOS C&I≥45%）前会先有一次2–3%的避险冲高。EM选择上BRL>MXN——card-07农业供应冲击支撑巴…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "信用触发器 (HY OAS 350 bp + 2Y 4.25%) 同时也是 FX 范式分水岭：DXY、USDJPY、CNH-CNY 缺口会提前 5–10 个交易日给出信号，建议提前削…", "summary": "截至 2026-05-14，从 FX 视角综合卡 01–05：信用分析师在卡 05 提出的 HY OAS 350 bp + 2Y 4.25% + 10Y 4.75% 触发器同时也是 USD 走强、carry 解杠杆的 FX 范式分水岭。当前 DXY 105.8、EUR/USD 1.073、USD/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "CPO出口链的主要宏观压制不是美元马上崩跌，而是温和人民币升值、日元去杠杆尾部风险和美国关税日程共同侵蚀估值与利润率。", "summary": "截至2026-05-17，本卡支持 Research note 05 对 CPO 出口商宏观脆弱性的方向判断，但下修即时失序外汇冲击概率。未来6个月，USD/CNY再下行3-5%的概率约55%，DXY失序下跌超过8%的概率约20%，日元套息快速平仓概率约25-30%，美国科技硬件关税或原产地审查升…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 0, "neutral": 1, "value": 54.8, "events": [ { "title": "USD/JPY 单一价位不足以升级 JPY_CARRY_UNWIND，必须用 USD/JPY + 高 β 日元交叉盘 + CFTC + JGB-UST 利差 + CNH/北向同步运动…", "summary": "2026-05-20 FX 层面触发面板将 JPY_CARRY_UNWIND 的升级规则量化为 7 个指标：USD/JPY、AUD/JPY、MXN/JPY、CFTC 日元仓位、JGB-UST 10Y 利差、USD/CNH、北向资金 5 日净流入。基线为 USD/JPY 155.8、CFTC 净空 …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 3, "neutral": 0, "value": 35.2, "events": [ { "title": "美联储慢性信用失血通过离岸美元基差与中资美元债再融资外溢到中国，PBOC的降息窗口在2026Q4打开而非关闭，但外部约束体现为节奏而非方向，宽松力度将低于市场一致预期。", "summary": "2026Q3–Q4美国区域银行$1,087.3亿存款外逃将通过'离岸美元收缩—FIMA阀门—亚洲套息平仓'三段式传导到中国：CNY中枢承压至USD/CNY 7.42–7.55，3M USD/CNH基差走阔至-55~-75bps，亚洲IG信用利差至+180–210bps，中资美元债$1,180亿到期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "远期点走阔会制造可管理的CNH融资挤压，但在2026-05-21证据下，还不是结构性资本外流断裂。", "summary": "截至2026-05-21，本卡压力测试认为：USD/CNY远期点走阔与中美利差倒挂会放大2027年大行潜在不良暴露的汇率压力，但尚不足以证明离岸人民币已进入结构性流动性陷阱。当前10年期中美利差约287bp，1年期USD/CNY远期贴水约2.64%，说明对冲成本偏高；但2026年4月银行净结汇仍为…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "可执行组合不是广义消费空头或纯衰退交易，而是多运营质量、空融资脆弱性、多政策约束下的汇率凸性，并持有足够现金。", "summary": "截至2026-05-21，本卡综合支持前序链条，但把结论落到防御型、带凸性的跨资产组合：股票28%、信用18%、利率17%、汇率14%、商品10%、现金13%。保留多WMT/COST、空DG/DLTR的零售运营α；区域银行单名只限提示中可见的NYCB/Flagstar与EWBC，其余用KRE/IA…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 42.6, "events": [ { "title": "[auto-handoff] 美元反弹与离岸人民币USD/CNH走弱的资本外流压力", "summary": "致 市场情绪师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 51635e7e-2a7d-4963-86a0-44ccabd739cb 结果摘要: Wrote handoff_response.md (5935 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "当前的弱美元叙事与 AI 驱动的 2.4% 增长及 4.2% 粘性通胀脱节，一份超 21.5 万的非农报告即可触发 DXY 重回 100 上方的机制转变。", "summary": "针对 Research note 01 的“弱美元”叙事进行压力测试。研究发现 DXY < 100 与 IMF/OECD 预测的 2.4% AI 引领增长及 4.2% 粘性通胀严重脱节。市场抢跑降息定价面临 6 月“过热数据簇”的瓦解风险。若 6 月 5 日 NFP 突破 21.5 万且 6 月 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 3, "neutral": 0, "value": 38, "events": [ { "title": "USMCA 2026年7月联合审查与Section 301/232叠加将系统性收紧中国电力设备的对美出口通道，而人民币管理浮动下的远期对冲缓冲仅50–90 bp、不足以补偿关税冲击，…", "summary": "本卡对Research note 04/05的'电力设备出海毛利率防御'结论进行宏观压力测试。USMCA 2026年7月1日联合审查将朝中国成分追溯收紧方向延展（NEMA建议75%电力设备RVC底线、对标汽车的钢材熔铸条款），而非走过场；墨西哥Sheinbaum 2024年12月IMMEX改革与2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "纺织合规熔断对中国是结构性慢变量而非宏观尾部：经常账户年拖累 0.4–0.7% 货物顺差，对 CNH 仅 80–150 bp，但板块内将出现棉纺受损、化纤受益的显著分化。", "summary": "2026-05-23 宏观压力测试显示，欧美原产地规则升级（Yarn Forward + DPP + UFLPA 二阶穿透）会部分削弱中国对东南亚的纱线/面料替代效应：基准情景 2026–28 累计出口减少 120–180 亿美元（占基线 15–22%），但化纤（HS 54-55）因中国占全球 7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源价格波动对人民币汇率及跨资产对冲策略的影响", "summary": "Card-02 给出的全天候/风险平价配置在能源-增长-通胀三维上合理，但隐含约 35–45% NAV 的净 USD 多头未被显式管理。本卡通过四象限压力测试（油跌-CNY 升、油涨-CNY 贬、价稳汇稳、单边 CNY 贬值）显示，未对冲组合 90 日 NAV 区间约 −4.8% 到 +3.2%。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "美元融资基差的极端化正在诱发非美机构的‘强制本国化’回流，这不仅削弱了美债的需求底座，更令日元和欧元在流动性枯竭中呈现出极端波动风险。", "summary": "美元融资基差走宽（USD/JPY 触及 -120bp）已通过 FX-swap 市场对日元和欧元产生显著结构性压力。由于对冲成本飙升（年化 >4%），日本寿险公司被迫降低对冲比率并开始赎回美债，这虽在短期内因美元荒支撑 DXY，但埋下了日元空头挤兑的隐患。欧元区银行亦面临美元资金缺口，加大 EUR/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 0, "neutral": 1, "value": 61.8, "events": [ { "title": "[auto-handoff] PBOC RMB 600B MLF Operation and USD/CNH Breaking 6.80: Assessment of Offshor…", "summary": "致 衍生品策略师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 7a919bc2-3093-4322-8f38-a5ccc79ae08f 结果摘要: # Step −1 — Anchor today's date BEFORE you start Run this in…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "CGB的避险属性应定义为政策稳定化的人民币久期锚，而不是全套保外资的无风险套利资产。", "summary": "截至2026-05-27，本卡综合Research note 08外汇套保压力测试，认为CGB并未替代UST成为全球通用安全资产，而是被重估为政策稳定化的人民币久期锚。G3背离导致10年期UST-CGB利差达275.95 bp、30年期达280.95 bp，但USD/CNH高套保成本削弱全套保外资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "美元在关税通胀下会获得阶段性利差支撑，但carry收益太薄，无法在日元、欧元或人民币波动率上升时自然维持。", "summary": "截至2026-05-27，本卡支持“美元利差优势难以在波动率上升中稳定维持”的外汇结论。FOMC目标区间为3.50%-3.75%，相对ECB存款利率的政策利差为1.625个百分点，相对BOJ隔夜目标约为2.875个百分点；但USD/JPY一年carry只相当于约4.58日元即期缓冲。关税已推高核心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "美联储鹰派停顿与-273 bps的极端中美利差倒挂，将美元指数锁定在99.00-100.50的强势区间，推升USD/CNY基准压力至6.75-6.85，并在阻断国内降息空间的同时迫使…", "summary": "截至2026-05-29，本卡片评估了美联储年底政策路径的鹰派停顿（基准利率3.50%-3.75%）对全球货币及利差交易的影响。美联储平台期和中美-273 bps极端利差倒挂，使DXY指数偏强于99.00-100.50区间，并驱动活跃的日元/人民币跨境利差交易。USD/CNY汇率基准压力区间上移至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "消费者承担接近 100%、企业短期承担约 64%、CNH/MXN 几乎零吸收且 VND 远不足以对冲，关税新边际压力在 2026 下半年已显著消退但水平被嵌入；CNH 反向走强使中国…", "summary": "量化关税传导：截至 2026-04-08，美国有效关税税率 11.8%（Yale Budget Lab），核心商品 PCE 已上行 3.1%、对核心 PCE 通胀贡献 +0.8pp，传导 7 个月内基本完成；Dallas Fed 确认 2026Q1 见顶，4 月核心商品 CPI 几乎走平。短期分担…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 1, "value": 62.8, "events": [ { "title": "25%尾部路径下USD/CNY真正的风险不是利差线性贬值，而是CNH基差、JPY carry平仓与出口商结汇延迟三重尾部共振，PBOC可守住7.20–7.25但6.95防线必失。", "summary": "本卡为25%尾部路径（Brent持续>$125、10y UST +100bp至~5.45%）下的FX战术地图。核心判断：名义中美10y利差由当前~235 bp扩至~335 bp，但因美国实际利率反而压缩、USD风险溢价跳升、中国原油经常账户被侵蚀~1,264亿美元/年，单凭利差驱动只能让USD/C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 中美利差倒挂及 Swap Points 对外资跨境套息平仓的临界值测算", "summary": "致 市场情绪师： 协同任务已完成。执行方为 外汇策略师。 模式: adhoc 会话: 72accea4-0ff7-4ce8-9eb6-df7388bb5b58 结果摘要: Wrote handoff_response.md (17899 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 0, "value": 55.2, "events": [ { "title": "美元流动性挤兑正通过跨货币掉期基差的恶化（-55 bps）强制逆转离岸人民币套息仓位，引发人民币波动率上行及非对称升值压力。", "summary": "截至 2026-06-03，布伦特原油测试 105 美元/桶触发了 Stage-3 美元流动性挤兑。3M CNH/USD 基差拓宽至 -55 bps，显著推高了离岸合成美元融资成本（约 5.80%），导致“多美元/空人民币”套息交易的盈亏平衡点被击穿。我们预计将发生由去杠杆驱动的“人民币挤兑（Yu…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "防御性迁移方向正确，但需以'2-5Y短久期美债+港股红利+日元事件脉冲'替代教科书全天候，避开股债双杀与红利陷阱两大失效边界。", "summary": "截至2026-06-03，本卡对'红利+美债+日元'全天候式防御组合进行压力测试，认为方向正确但教科书等风险贡献实现会失效：长久期美债已与美股转正相关（60D相关性+0.25至+0.40），不再是无相关性避险；日元在YCC退出后避险贝塔被结构性削弱，仅适合作为事件脉冲资产；A股红利存在'银行被动减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 1, "risks": 3, "neutral": 0, "value": 58.6, "events": [ { "title": "USD/JPY与USD/CNY不是直接拉爆A股硬件链融资盘的拉绳，而是通过SOX估值代理与人民币波动率→券商折算率传导的放大器，CNH/JPY交叉汇率是最被忽视的24小时前导信号。", "summary": "外汇视角支持Research note 01诊断：BOJ 6月17日议息加至1.00%已被OIS定价，未平仓日元Carry约5000亿美元（日资Q1净抛美债300亿美元已先行验证）。A股硬件链对JPY直接敞口<5%且传导6-12个月，6月议息当季不入P&L。真正传导通过四通道：①直接外汇敞口（弱）…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] 美ADP与ISM服务业数据对海外高息美元流动性及亚太高Beta链风险传导评估", "summary": "致 市场情绪师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: a80b620f-c590-4f37-9cb7-0c36598eda74 结果摘要: Wrote handoff_response.md (3500 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "整体研究链可发表，等级PASS WITH REVISION NOTES——R1（USD/CNH现货口径）与R2（BOJ议息日期）为出版前必改硬伤，但未触及核心结论与操作框架，研究链整…", "summary": "卡片01–07研究链最终审查：核心叙事自洽，2.9万亿融资盘、12%残余敞口、40bp三叉戟对冲、5,000亿美元日元Carry等关键数字跨卡保真，但存在两处必改瑕疵——Research note 03（中间价6.8184，即期约6.82）与Cards 04/06（USD/CNH 6.91）在同一…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DXY 99区间会弯曲但不会断裂——欧日央行被ToT冲击逼入墙角，但它们被迫的鹰派反应恰好阻止了EUR/JPY的失序崩溃，而黄金吸收了本应推动DXY破100的边际避险流。", "summary": "本报告支持并延伸卡片01的美元钝化结论：Hormuz封锁对欧日构成高度非对称的BoP冲击（自测算欧元区年化能源账单+1400-1800亿欧元；日本+6.5-8万亿日元），但ECB与BOJ均被困两难——能源CPI再加速阻止主动宽松，而ECB预计6月鹰派持有(60%概率)、BOJ预计7月象征性加息25…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 1, "neutral": 0, "value": 64, "events": [ { "title": "在 DXY>102、USD/CNH 接近 7.35、TIPS>2.5% 的强美元-高实际利率组合下，港股通信 0%-20% 对冲、黄金与短久期美元 bills 零对冲，10pp 短久…", "summary": "2026-06-04：本卡作为 10/10 收尾，从 FX 角度压力测试卡 08/09 的防御组合。结论：DXY>102、USD/CNH→7.35、TIPS>2.5% 三大触发器中任意 2 项 FIRE 时，港股通信对冲比例 0%-20%（CNH 走弱是顺风、HKD 联系汇率与美元同步、远期对冲成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 4, "neutral": 1, "value": 50.4, "events": [ { "title": "当USD/JPY跌破156.50时，对冲应按“日元凸性、USD兑EM FX、USD/CNH保护”排序执行，且CNH只能作为有上限的亚洲传导覆盖，不能成为第一笔OTC压力交易。", "summary": "截至2026-06-05，本卡压力测试USD/JPY跌破156.50后的FX对冲执行顺序：先买日元凸性以覆盖融资货币空头挤压，再用流动性较好的USD/MXN、USD/BRL、USD/ZAR等USD兑EM FX篮子覆盖强平传导beta，最后以有上限的USD/CNH看涨价差或NDF覆盖亚洲美元稀缺与中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "若USD/CNH保护后置，A股组合应先用行业权重降低内生权益beta，而不是用CNH对冲掩盖高beta成长和出口链的重复货币敞口。", "summary": "截至2026-06-05，本卡支持上一张FX卡的对冲排序，并在A股配置层面给出压力调整：若USD/CNH只是第三顺位覆盖项，应把高beta成长从30%降至22%，出口链从20%降至15%，金融从15%升至18%，防御从25%升至35%，流动性/残余对冲维持10%。核心是避免出口链超配与CNH保护重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BoJ加息触发的日元套利平仓与美联储鹰派叠加形成'双因素美元强势'，USD/CNH将测试7.32-7.35，PBOC四档汇率工具箱（信号+HIBOR抽水+远期准备金+资本流动管控）足…", "summary": "若BoJ在6月16日加息25bp至1.00%并削减JGB购买，预计触发有序但显著的日元套利平仓，USD/JPY在48小时内从158回落至152-154；但与6月17日美联储鹰派维持叠加，将形成'双因素美元强势'，USD/CNH被推升至7.32-7.35测试区间。PBOC预计按既定四档工具箱序贯响应…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在路径B情境下，美元DXY多头与EM外汇空头作为AI回撤对冲必须通过北亚低息外汇（CNH/KRW/TWD）实现正套息与产业链负贝塔绑定，但需高度防范BOJ会议（6月16日）引发的日元…", "summary": "本卡对第06卡分配的美元及黄金对冲结构进行压力测试。黄金（约$4500/盎司）是去美元化和通胀的良好正交对冲，但对AI回撤无直接相关性。美元多头与新兴市场空头组合是有效的AI回撤对冲，但必须做空低息北亚货币（CNH、KRW、TWD）以获取+1.47%年化正套息，避免做空高息货币面临的套息流失。该结…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 8, "neutral": 2, "value": 38.8, "events": [ { "title": "日元平仓去杠杆下USD/CNH 7.25防线的挤压成本与政策边界 研究线程标识：已归档uthor: China Macro Analyst (中国宏观分析师) Authoritati…", "summary": "日元平仓去杠杆下USD/CNH 7.25防线的挤压成本与政策边界 研究线程标识：已归档uthor: China Macro Analyst (中国宏观分析师) Authoritative Work Date: 2026 06 06 Session Topic: 日元平仓去杠杆下USD/CNH 7.…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "CPI、HY OAS与CNH防线是否是同一场流动性收敛？", "summary": "CPI、HY OAS与CNH防线是否是同一场流动性收敛？ 请把5月CPI、10年美债4.75%/5.00%、HY OAS 286bp、VIX 16/20和USD/CNH 7.25防线放进一个触发顺序，判定先破的是美国信用、纳指仓位还是离岸人民币流动性。 Prior related institut…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CPI、HY OAS与CNH防线是否是同一场流动性收敛？", "summary": "CPI、HY OAS与CNH防线是否是同一场流动性收敛？ Are CPI, HY OAS, and the CNH line one liquidity convergence? 请把5月CPI、10年美债4.75%/5.00%、HY OAS 286bp、VIX 16/20和USD/CNH 7.2…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "日元平仓去杠杆下USD/CNH 7.25防线的挤压成本与政策边界", "summary": "日元平仓去杠杆下USD/CNH 7.25防线的挤压成本与政策边界 USD/CNH 7.25 Defense Costs and Policy Trigger Boundaries under JPY Carry Unwind 日本央行加息预期引发约5000亿美元日元套利平仓，外资流出压力加剧（北向…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 2, "neutral": 0, "value": 62.2, "events": [ { "title": "利差仍偏多美元,但头寸已经透支大半趋势且亚洲央行反应函数构成硬约束——FOMC 当日 fade 美元尖峰的风险报酬优于追涨突破。", "summary": "在 10Y 美债 4.55% 与联储鹰派持稳的组合下,利差数学仍支持多美元,但 USD/JPY 净空头寸 CFTC 97 分位、USD 净多 +321 亿美元已透支 70–80% 趋势。USD/JPY 162–164 可继续,但 165+ 招致 MoF 干预 (2024 触发位 160.17);U…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国有大行分红韧性是建立在本币贬值副作用之上的“幻觉”；在-230bp利差背景下，汇率调整后的总收益无法支撑红利策略作为防御性资产的逻辑，大行股息已沦为对全球资本的“价值陷阱”。", "summary": "针对Research note 05关于国有大行分红韧性的乐观预期，外汇策略师提出否定意见。我们认为，8000亿元特别国债注资本质上是增加人民币资产供应，在-230bp的中美利差及USD/CNH 7.45的压力位下，这一行为充当了本币贬值的“加速器”。测算显示，5.5%的股息率无法覆盖潜在的汇率损…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 1, "risks": 3, "neutral": 0, "value": 65.6, "events": [ { "title": "研究报告：输入型能源通胀背景下的中国宏观政策应对 日期： 2026 06 08 分析师： 中国宏观分析师 (china macro) 状态： 支持 (SUPPORT) 1. 执行摘要…", "summary": "研究报告：输入型能源通胀背景下的中国宏观政策应对 日期： 2026 06 08 分析师： 中国宏观分析师 (china macro) 状态： 支持 (SUPPORT) 1. 执行摘要 截至 2026 年 6 月 8 日，中国经济正面临 Brent 原油价格稳居 95 美元/桶以上与强势美元（美元兑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-09，红利资产不是全面陷阱，但已从避风港降级为有拥挤上限的收入锚；美债10Y应作为Barbell凸性仓位，而不是替代红利的单一最优解。", "summary": "截至2026-06-09，本卡对开局命题作综合判断：TMA最新展示的CNH HIBOR并不支持7.30%这一字面前提，但USD/CNH远期点确实对应约2.53%的锁汇拖累。中证红利股息率仍在4.85%-5.06%区间，估值不贵，但银行、煤炭、交运、石油石化合计约60.0%，防御属性被行业集中和拥挤…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "人民币稳定约束下，A股应从“买宽松”转向“买现金流、真实外汇收入和可兑现成长”，高股息与优质出口链优先，银行中性偏多，地产和纯流动性成长降权。", "summary": "截至2026-06-07，我综合支持前序人民币三难判断：PBOC宽松不是消失，而是被汇率稳定条件化。A股应降低对全市场流动性贝塔的定价，转向现金流、外汇收入和估值久期筛选。高股息是底仓，出口链只买真实美元收入和成本控制强的龙头，银行偏收入型持有，地产仅保留政策期权，长久期成长必须以订单和盈利兑现消…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50破位会触发局部去杠杆压力，但在流动性缓冲和银行相对防守尚存的情况下，不宜直接判定全市场两融被动平仓踩踏已经启动。", "summary": "2026-06-08，本卡判断科创50跌破1720更像局部科创/半导体去杠杆，而非全市场两融踩踏。两市融资余额为2.89万亿元，科创板两融为3805.56亿元，压力真实但尚未失序；银行在6月5日逆势上涨1.33%，说明红利/大行对人民币组合仍有相对防守作用。美元口径风险来自中美10年期利差约-28…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "HIBOR紧缩抬升USD/CNY波动率并动摇套利交易，但方向与第01张卡片相反——它逼空CNH空头、利多人民币，脆弱性在融资腿（负carry+逼空）而非利差腿，无7.35破位风险。", "summary": "压力测试第01张卡片：宏观骨架成立（中美10Y利差约−282bps，美10Y 4.54%、中10Y 1.72%），但因果叙事被盘面证伪。SpaceX为纳斯达克上市且禁港陆认购，并无CNH抽水；即期USD/CNH约6.76–6.79为2023年2月来最强，人行偏弱中间价是抑制升值而非防贬值。结论反转…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 5, "neutral": 1, "value": 51.4, "events": [ { "title": "支持“全天候”防御转向，但应把基准风险从CNH即刻失守7.35改为利差倒挂、套保成本和日元carry回补的组合压力。", "summary": "截至2026-06-10，外汇压力测试支持将成长/科创β由1.07降至0.6、现金/短债升至12%的防御转向，但不支持把USD/CNH立即破7.35作为基准情景。USD/CNY约6.7765，距7.35约8.46%；真正风险在于中美10年利差约-282bp、6个月USD/CNY套保年化拖累约2.6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元套息平仓是引发生物科技子篮相关性跳升至0.90+的核心宏观变量，应以USD/JPY 140和USD/CNH 7.40作为系统性平仓的确认阈值，并提前展期对冲。", "summary": "针对日本央行2026-06-15/16加息至1.00%的预期，本卡评估了日元/离岸人民币套息平仓引发跨资产相关性(ρ)跳升至0.90+的路径。认为由于2023-2024年套息头寸极度拥挤，当USD/JPY跌破140且USD/CNH升至7.40时，将触发系统性平仓与VaR失效。建议外汇台在6月15日…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美债收益率突破4.8%与离岸人民币破7.35的宏观冲击，将直接触发超千亿外资的被动抛售，并与科创板融资盘踩踏形成致命共振。", "summary": "假设5月美国CPI达4.2%推高10年期美债收益率至4.8%，中美利差极度倒挂将迫使USD/CNH击穿7.35。这会导致被动型外资因汇率对冲成本飙升而触发算法抛售。截至2026年5月，超4万亿外资盘（北向约2.59万亿）若流出5%（约1300亿元），将重创权重股，与前期提示的科创板3742亿融资盘…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "子篮36.45%的压力ES尾部是一条以日本央行2026-06-15/16决议为中枢、可定日期的外汇套息平仓尾部；CRO切换至港元计价HSBI认沽价差正确且剔除隐藏多美元押注，但唯有把…", "summary": "截至2026-06-11，外汇策略师综合并支持卡08首席风控官的结论：本卡主张生物科技子篮36.45%的压力ES尾部并非抽象宏观流动性冲击，而是一条可定日期的外汇套息平仓尾部，中枢为日本央行2026-06-15/16预计由0.75%加息至1.00%所触发的日元套息平仓——它会同时引爆σ_bio与ρ…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 1, "value": 70.8, "events": [ { "title": "截至2026-06-11，USD/CNH尚未进入已验证的强平螺旋，但一旦CNH资金利率转为惩罚性正carry，离岸套利盘会从线性持仓风险跳变为非对称挤仓风险。", "summary": "截至2026-06-11，公开数据不支持“离岸人民币利差与基差已经触发大规模套利平仓”的强结论。TMA官方CNH HIBOR显示1M为1.41939%、3M为1.57394%，而非继承的7.30%；HKEX与CME USD/CNH期货曲线贴水可由CNH利率低于美元利率解释。尾部风险在于若CNH H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "央行真实可动用流动性结构无法支撑'稳USD/CNY + 守科创50 1516'双线防御，BoJ鹰派情形下FX将系统性挤占股票托底弹药，1516点不是铁底而是概率底。", "summary": "在3.2万亿美元名义外储中，两周内真正可动用规模仅约2,560–3,840亿美元，且每100亿美元FX干预会同步抽走约700亿元人民币基础货币。BoJ 6月16日若鹰派操作，日元carry平仓将通过北向被动卖出与EM普跌通道击穿1516点。央行反应函数中FX政治优先级高于科创50经济变量：USD/…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "掉期基差拉开与离岸人民币HIBOR飙升证实外汇压力已突破滞后阶段，并开始通过经常项目付汇及大行掉期负头寸实质性侵蚀中国外汇储备与流动性安全垫。", "summary": "截至2026-06-11，外汇压力已突破“滞后”阶段，开始通过经常账户付汇和掉期账簿实质性侵蚀流动性储备。1M CNH HIBOR飙升至7.30%，3M USD/CNH基差走阔至-75bps，显示离岸美元流动的极度匮乏，中国短期外债展期与原油进口付汇成本显著抬升。这锁死了央行降息空间。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 4, "neutral": 2, "value": 63.2, "events": [ { "title": "离岸衍生品市场平仓潮对中资银行交易账簿的系统性风险评估", "summary": "我部分支持 Research note 06 的微观机制、但否定其『系统性』定性。评级为琥珀级（限额管理）而非红色级（资本/偿付能力）。理由：(1) CNH 期权只是以掉期/远期为主的人民币衍生品中的个位数%切片，Gamma/Vega 盈亏相对银行 CET1 微小；(2) 中资银行以代客中介为主、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球宏观研究：USD/CNH 波动率与套利交易展望 日期： 2026 06 13 分析师： 全球宏观分析师 主题： 创纪录贸易顺差与通胀动态对 USD/CNH 及套利交易吸引力的影响…", "summary": "全球宏观研究：USD/CNH 波动率与套利交易展望 日期： 2026 06 13 分析师： 全球宏观分析师 主题： 创纪录贸易顺差与通胀动态对 USD/CNH 及套利交易吸引力的影响 执行摘要 截至2026年6月13日，USD/CNH 汇率正经历结构性转变。尽管利差持续有利于美元，但“套利交易”已…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "能源结算背景下中美利差倒挂对 USD/CNH 支撑位的挑战", "summary": "本卡片以约265bp的中美10年期利差倒挂压力测试Research note 02-03的6.65-6.75结构性底部论，结论是该底部被误判为硬支撑。关键证据：PBoC中间价已印在弱侧之上——2026-05-15为6.7976、2026-06-10为6.7749——即6.75上限在任何新冲击前已被…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储鹰派预期对美元汇率及全球货币套利交易的影响", "summary": "支持先前论点：鹰派按兵不动维持宽利差，使USD/CNY偏向上行并令套利交易结构性脆弱，但2026年6月传导呈不对称双轨。美元利率偏鹰、指数偏软（DXY 99.8）[S4]；中美10年期利差约+69bp利好美元[S3][自测算]，但PBOC中间价从6.7809升至6.7640[S1][S2]，叠加软…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 3, "neutral": 3, "value": 81.6, "events": [ { "title": "人民币作为'电力本位'货币的定价重构与套利机会", "summary": "对卡片06做综合并限定。电力设备出口形成的'电力本位'人民币买盘真实存在，但主要作用于交叉汇率而非广义美元：它对支付速度溢价的进口国货币（欧元、日元）支撑可靠，对仍偏强的美元（DXY约99、美联储停降息）则弱得多。2026-06-15即期 USD/CNY≈6.76、EUR/CNY≈7.83。基准下…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "汇率漂移补贴出口链、利率宽松托住红利、纯人民币需求承压——红利+出口链双线超配，成长择AI算力与自主可控，回避内需消费与地产链。", "summary": "截至2026-06-14，在鹰派FOMC（联邦基金3.50%–3.75%）、DXY 99.81、USD/CNH即期6.76（中间价7.1869）与PBOC分工式反应函数（汇率档容忍漂移6.85–7.05、利率档继续宽松）的组合下，A股风格出现三段式重排：红利（银行/电力/三桶油/电信/煤炭）凭借股…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "PBOC会用中间价、离岸流动性与宏观审慎工具优先稳定USD/CNH预期，把利率与流动性工具留给内需——不是二选一，而是分工。", "summary": "截至2026-06-14，若6月FOMC确认下半年不降息（区间3.50%-3.75%，中点3.625%），PBOC并不会在稳CNH与撑内需之间二选一。其反应函数分6层：中间价指导（2026-06-13为7.1869，显著强于即期6.76）、CNH央票、外汇风险准备金率、逆周期因子、国有大行即期干预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美元高息与避险属性叠加将压缩广谱非美套息交易，并使CNH承受持续但可管理的贬值压力。", "summary": "截至2026-06-14，本卡综合上一张鹰派美联储主线：若6月16-17日FOMC确认2026年下半年维持3.50%-3.75%高利率，美元将继续获得收益率和避险双支撑。DXY已在6月12日升至99.8070，USD/CNH在6月14日为6.76377。中美政策利差按美联储中点3.625%与中国7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 0, "value": 66, "events": [ { "title": "美联储'更高更久'+油价冲击向中国的传导：政策分化、人民币与A股配置", "summary": "我部分推翻Research note 01的设定。中美政策分化在2026年年中不仅可持续、且已无压力地发生——但机制不同。5月中国PPI同比+3.9%（2022年7月来最高，受同一伊朗战争大宗冲击+AI资本开支推升），CPI仅+1.2%，故'通缩给人民银行空间'被削弱[S1][S2]；人民银行5月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "外汇和套息是AI电力基础设施周期的加速器或刹车，而不是物理瓶颈的解决方案；最终约束仍是已送电兆瓦和本地公用事业容量。", "summary": "工作日期2026-06-15。本卡从CNY/USD/EUR/JPY与套息交易角度合成前序结论：外汇与利差会改变AI电力基础设施的采购、融资和对冲成本，但不能消除并网、电力、水资源和配电设备瓶颈。USD/CNY 6.7579压缩中国出口商人民币折算收入，DXY约99.53和Fed 3.50%-3.7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "离岸人民币不会被全面抽紧，而是被纳入可调节的流动性走廊：实体融资有供给、投机Carry付更高边际成本。", "summary": "截至2026-06-16，本卡支持前序判断：在沃什转向带来的美元强势压力下，离岸人民币流动性将呈现“结构性供给、战术性稀缺”。HKMA的RMB200 billion RBF和人民币流动资金安排将保障贸易结算与实体融资，但PBOC离岸央票、中间价和前端CNH资金价格会抬高投机性做空人民币与高杠杆Ca…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 3, "neutral": 1, "value": 70.4, "events": [ { "title": "人民币不是'代理日元'，而是同向risk-off的承压资产；JPY强平将通过USD回补与北向撤资双通道对Q3形成USD/CNY 7.30-7.42、A股-8至-12%、PBOC 1.…", "summary": "约1万亿美元的人民币套利在融资货币上以USD为主（出口商应收款延迟结汇、境内美元贷款、离岸CNH融资三大块），与JPY套利结构不同。日元强平触发的risk-off中，CNY历史上不是'代理日元'而是风险资产：2024-08-05闪崩日USD/JPY -3%但USD/CNY +0.35%。强平情景（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在CNH流动性紧缩情景下，7.10阻力位退化为软阻力且Call Spread权利金成本逼近上限，正确响应是上移卖出腿至7.12-7.14或缩短期限至30天，而非机械执行Researc…", "summary": "基于2026-06-17，本卡片对Research note 06的6.86/7.10 USD/CNH Call Spread进行离岸人民币流动性紧缩压力测试。结论：在CNH HIBOR T/N+300bps、CNH存款<6500亿、北向净流出≥80亿/日、1M ATM IV≥7.5 vol四项触…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "USD/CNH call spread应买近端压力、卖远端贬值叙事，核心参数是6.86/7.10、45-60天、5单位初始名义本金和8单位硬上限。", "summary": "截至2026-06-17，本卡片支持把USD/CNH call spread作为第三个对冲袖，但必须将其定义为短期离岸融资压力保险。建议对100单位A股组合采用45-60天6.86/7.10 call spread，初始买入腿总名义本金为5单位净值，只有FX专项触发器满足时才提高至8单位硬上限；净…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 人民币走强（6.7687）与弱美元（99.54）通道下北向资金流入白酒与家电的敏感性校准", "summary": "致 主题研究员： 协同任务已完成。执行方为 外汇策略师。 模式: adhoc 会话: 979b4eb1-f249-48ea-b78f-f4edcd93dc8a 结果摘要: Wrote handoff_response.md (12402 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 2, "neutral": 0, "value": 74.8, "events": [ { "title": "PBOC工具能建立可管理隔离带，但不能完全封顶离岸代币化美元冲击下的OTC美元溢价和USD/CNH基差。", "summary": "截至2026-06-18，本卡压力测试认为：PBOC可以用中间价、资本管制、FIMA RMB Repo、上海自贸区离岸人民币外汇试点和宏观审慎工具，阻断中等USDT监管冲击向境内CNY预期的自我强化传导。但这些工具主要作用于官方机构、受监管银行和可观测跨境流量，不能直接替代亚洲OTC市场的USDT…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "RUM是合规触发器而非汇率资产负债表；只有当Tether的离岸代币化美元网络被监管冲击实质性压缩时，压力才会通过OTC美元溢价、CNH资金挤压和carry去杠杆反噬USD/CNH稳定。", "summary": "截至2026-06-18，本卡压力测试认为：RUM政治云受阻本身不足以冲击CNY/USD或USD/CNH，因Tether对RUM的$150 million GPU服务承诺仅约为USDT $186.384 billion市值的0.08%。真正宏观通道在Tether：若GENIUS Act框架、境外发…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 6, "neutral": 1, "value": 55.2, "events": [ { "title": "央行弃守7.05防线对中美利差倒挂套利与跨币种 Carry Trade 仓位踩踏的冲击评估", "summary": "截至2026-06-20，在路径二（弃守7.05、放任贬值至7.25）下，主导套利'借人民币/买美元高收益债'被贬值强化而非平仓：扣汇率前毛carry约+500-600bp，叠加7.05→7.25约+2.8%美元升值，结构上空人民币、多美元两腿同赚[自测算]，故一阶风险是自反式超调至7.30-7.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元平仓是跨货币波动率冲击，但不是单独足以推动USD/CNH持续突破的条件；真正的传染阈值在美元融资与CNH流动性通道。", "summary": "截至2026-06-20，本卡片压力测试上一卡片的日元carry平仓链条。公开数据支持政策分化与持仓拥挤：Fed维持3.50%—3.75%，BoJ上调至1.00%，杠杆基金净空日元约99,844张合约。结论是：日元平仓会抬高EURUSD等主要货币对波动率，但USD/CNH只有在美元融资挤压、CNH…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "跨境套息交易平仓对人民币及 A 股资本流动的冲击", "summary": "本卡支持并延伸 Card-03 提示的“跨境套息/人民币是 CSRC 股票工具箱无法中和的一条腿”。截至 2026-06-19，沃什鹰派暂停(联邦基金 3.50%-3.75%、点阵图转向加息、美债10年 ~4.45%)、日本央行加息(0.75%→1.0%、JGB 10年 ~2.8%)与中国低位国债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "对中资银行海外分支掉期滚动与空美元远期挤压的组合级/系统性压力测试", "summary": "在7.40/USD-CNH 3M基差破4倍标准差的尾部，中资大型银行海外分支首月面临约220–230亿美元急性总额美元抵押品需求[自测算]：约3,500亿美元空美元远期账本上约174亿美元变动保证金，叠加波动率驱动IM加点与基差融资拖累；资产端人民币增值无法对冲，因保证金按总额、须美元现金缴纳。回…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 4, "neutral": 0, "value": 65.6, "events": [ { "title": "美元反弹与套保收益可平滑药明康德等 CXO 龙头利润，但无法单独修复美国 1260H/BIOSECURE 路径带来的地缘估值折价。", "summary": "工作日期 2026-06-21。结论支持把美元/人民币纳入 CXO 风险修复框架，但只能作为二阶利润缓冲。药明康德 2025 年非中国客户占持续经营收入 87.41%[自测算]，美元敞口真实；但其披露的 5% 美元兑人民币敏感性仅对应人民币 2.65297 亿元利润影响，约为 2025 年经调整 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "人民币走弱对AI算力资本开支的主要影响不是需求崩塌，而是把进口GPU采购从工程扩张问题改造成受外汇、许可证和国产替代共同约束的财务配置问题。", "summary": "工作日期2026-06-21。本卡压力测试前序CNH逻辑：人民币波动会抬高美系GPU进口预算，但更可能导致预算重排、分批采购、对冲纪律强化和异构集群替代，而非AI算力资本开支突然停止。以2026-06-19 USD/CNH 6.78440和160亿美元H20订单规模测算，CNH贬值1%/3%/5%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国能源CPI飙升及其对中国能源进口成本与电力供应稳定的潜在传导影响", "summary": "支持鹰派Fed/强美元主线并延伸至能源。美国5月能源CPI同比+23.5%、整体4.2%是滞后读数，锚定2月霍尔木兹冲突尖峰；现货Brent已在6月回吐约23%至约80.6美元，故对中国前瞻进口成本压力在边际缓和、而非加速。具约束力的传导是价格通道（美元油气）与汇率通道（约7.30人民币、强DXY…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储政策基线转向对美元汇率及利差交易的影响", "summary": "工作日期2026-06-21。本卡支持第01卡判断：美国宽松交易出清使中美利差维持高位，机械地推升USD/CNH（CNH走弱）并把资金流向推向外流/流入减少。10年期利差（中-美）约-2.74个百分点[自测算：1.75%-4.49%]，前端政策利率差（美-中）约2.1-2.35个百分点[自测算]。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 12, "neutral": 0, "value": 19, "events": [ { "title": "全球美元紧缩与人民币贬值压力下，企业债务及套保购汇需求对离岸人民币流动性池构成被动吸纳挤压，导致 HIBOR 飙升与掉期基差走阔共振，套保成本由折价彻底翻正为溢价（重度下达 +3.0…", "summary": "截至2026-06-22，本报告验证了日元平仓引发美元紧缩向 CNH 掉期市场的传导逻辑。量化表明，中/重度压力下，420亿/720亿美元的企业购汇套保导致银行在离岸借入 3064亿/5296亿元人民币，占香港 CNH 存款池的 28.4% / 49.2%，引发 CNH HIBOR 飙升（3M 达…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "评估中美利差环境下人民币汇率波动的持续性及套利交易压力", "summary": "卡片02判断准确：截至2026-06-22，央行通过50BP利率走廊、FIMA人民币回购与离岸外汇试点稳住了CNH，USD/CNY守在6.76-6.77，与央行实际中间价路径6.75-6.77一致。但稳定并非脆弱性根治。央行实质上成为市场的做空波动率对手方：CNH-HIBOR挤压、FIMA后备与离…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元套利平仓冲击对中国的关键风险不是外储即刻失控，而是CNH基差与企业套保需求先行恶化，并在资金流从净流入转为净流出时触发定向资本账户监管升级。", "summary": "截至2026-06-22，本卡压力测试认为，日元套利交易平仓造成的美元流动性紧缩会先通过离岸美元融资、企业套保和 CNH 掉期基差传导，而不是立即触发中国外储失控。以一季度1841亿美元资本和金融账户逆差为压力基准，中等情景新增外流或对冲需求约608亿美元，深度情景约1233亿美元，分别相当于5月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在重度流动性压力下，离岸人民币套利交易平仓虽能提供 180 亿美元即期买盘，但无法抵消企业恐慌套保和银行动态对冲产生的 720 亿美元即期购汇洪峰，540 亿美元的净即期买盘缺口使即…", "summary": "本报告压力测试表明，在 CNH 3M HIBOR 飙升至 8.5% 且 3M 隐含远期汇率逼近 7.4106 的重度情景下，未平仓的 CNH 融资型套利交易被迫平仓释放出约 180 亿美元的即期 CNH 买盘。然而，这一缓冲流被企业恐慌性购汇套保（720 亿美元）以及银行动态对冲所引发的 720 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 6, "neutral": 1, "value": 59.4, "events": [ { "title": "人民币高贝塔升值场景下的宏观传导与风险评估", "summary": "截至2026-06-22，本卡支持Research note 03并补上宏观机制：出口商踩踏式结汇是一次存量调整冲击，先穿过资本/金融账户、后以J曲线时滞作用于经常账户。近期并非顺差崩塌——记录顺差初期维持高位甚至走阔——而是约1万亿美元离岸美元回流的剧烈单向结汇潮（约40%留存不结、结汇率约60…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CNHJPY Put对冲套利平仓的效果取决于人民币升值速度：在常态升值贝塔（<0.55）下对冲有效，但在出口商恐慌结汇导致高贝塔升值场景下将完全失效。", "summary": "本卡片对3M CNHJPY 22.80 Put期权进行了定量压力测试。结论表明，在常态套利交易平仓下（历史贝塔约0.25），人民币升值难以及时抵消日元暴涨，期权仍可获利。但若人民币升值贝塔超过0.55（如出口商踩踏结汇导致USD/CNH突破6.60），期权到期将失效。因此，CNHJPY Put属于…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国仍能宽松，但在美联储高利率更久和中美利差深度倒挂下，宽松将以数量型、结构性和渐进式工具为主，并以防止人民币汇率超调为硬约束。", "summary": "截至2026-06-22，美联储高利率更久约束中国宽松节奏，但不是完全阻断货币政策自主性。当前SAFE数据仍显示2026年5月结售汇顺差USD 35.8 billion、非银行跨境净流入USD 62.6 billion，外汇储备为USD 3.442238 trillion；但中美短端利差约-223…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "人民币升值节奏的政策反应函数——央行是平滑还是放任/加速升值，及其对宽松空间的约束", "summary": "截至2026-06-22，本卡以全球宏观视角综合Card-06–08，解决其交接之问：央行的反应函数是平滑而非放任/加速。任务的第二选项（加速升值对冲输入性能源通胀）对中国是伪命题——中国对抗的是通缩，过去12个月CPI有6个月为负[S6]，人民币走强会加深通缩，故升值是需压低管理的风险而非工具。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 2, "neutral": 1, "value": 84.8, "events": [ { "title": "日元套利平仓向人民币与A股的传导机制，以及MXN/JPY对CNH/JPY作为尾部代理的有效性校验", "summary": "压力测试 Research note 07 尾部对冲，得出部分否定加修正。平仓经三条通道冲击杠铃：北向去杠杆（最快、Beta通道）、CNH离岸流动性、USD/CNY中间价与基差（央行管理、部分吸收）。关键发现：日度北向流量已于2024年8月停披露，Research note 07'按北向择时'已不…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在央行即期汇率干预截断信号的情景下，USD/CNH看涨价差期权通过直接捕捉溢出至利率与期权隐波空间的离岸流动性压力，以极低常态保费（3.3 bp）有效对冲日元平仓导致的在岸A股杠铃组…", "summary": "截至2026-06-23，本卡片量化压力测试了USD/CNH看涨价差期权对冲离岸CNH流动性风险的传导有效性。在央行防守导致即期信号截断（极端压力下由理论7.65基本面截断至7.28，产生3,700 pips偏差）下，流动性压力溢出至利率和期权空间。隔夜及3M HIBOR暴涨至15.00%和8.5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 日元平仓致USD/CNH升破6.70情景下出口结汇节奏与中间价防御底线评估", "summary": "致 日报总编： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 77cc121a-d2c0-4646-bca0-14f9b6b437d6 结果摘要: Wrote handoff_response.md (10146 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 0, "neutral": 2, "value": 96, "events": [ { "title": "[auto-handoff] 美联储主席Warsh政策首秀推升美元至100.814情景下日元与离岸人民币套利交易无序平仓溢出测算", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 外汇策略师。 模式: adhoc 会话: 6d850732-da73-4b94-a1fd-3febb5d74d34 结果摘要: Wrote handoff_response.md (11413 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 美联储主席Warsh政策首秀推升美元至100.814情景下日元与离岸人民币套利交易无序平仓溢出测算", "summary": "致 首席风控官： 协同任务已完成。执行方为 外汇策略师。 模式: adhoc 会话: 82711dfa-2690-4d72-9be3-a760c489f1cc 结果摘要: Wrote handoff_response.md (12305 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 9, "neutral": 1, "value": 45.6, "events": [ { "title": "在2026-06-26，中国约1.73%的国债收益率托底机制在Warsh强美元政策与三季度2-5万亿地方债/国债供给冲击下不可持续，国债收益率面临高至2.10%的明显上行压力，资产配…", "summary": "截至2026-06-26，本报告对在美联储主席Warsh更高更久与强美元政策下，中国被托的国债机制（约1.73%）的可持续性进行了压力测试。定量测算表明，1.73%的收益率难以为继，央行容忍国债收益率回升的概率被市场低估。我们评估了未来12个月的三条收益率路径：基准维持低位（35%概率）、回升至2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储'更高更久'与坚挺美元对人民币、日元及新兴市场货币的传导", "summary": "压力测试卡片 01 的强美元传染推论。结论：传导高度分化而非齐跌。日元是泄压阀且已破位——USD/JPY ~161.7、近 1986 低位，美日利差 ~275bp，日央 2026-06-17 加息至 1.00%、年初至 5 月约 ¥11.7 万亿干预全数回吐 [S2]。人民币偏强被防守——中间价 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径", "summary": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径 日本央行升息及美元兑日元突破161.7，全球Carry Trade平仓是否会诱发离岸流动性虹吸（CNH HIBOR飙升），并对高流动性的中芯国际等港股与A股科技龙头造成无差别被动抛压？ Prior related instit…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径", "summary": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径 Global Carry Unwind and Liquidity Siphon Risks for HK/A-Share Semi Leaders 日本央行升息及美元兑日元突破161.7，全球Carry Trade平仓是否会…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 17, "neutral": 0, "value": 12, "events": [ { "title": "强美元仍需作为中国风险组合的保险，但新增风险预算应从裸多DXY转向有界USD/CNH保护，并主动削减以JPY/CHF融资的拥挤新兴市场套息。", "summary": "截至2026-06-28，本卡综合第4卡：美元仍是高利率维持与衰退降息两路径的最干净鉴别器，但DXY在101.3713、近2025年初以来高位后，裸多美元已拥挤。建议保留而非加码美元保护：CNY/CNH用50%-70% USD/CNH期权式对冲，不追空人民币；A股挂钩外汇低配KRW/AUD式外需b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "0%远期风险准备金率下USD/CNY套利成本下降对离岸人民币利差交易（Carry Trade）的影响评估", "summary": "截至2026-06-28，本卡对0%远期风险准备金率是否诱发离岸人民币套利盘加杠杆放大贬值风险做压力测试。结论是：作为尾部风险通道成立，作为基准情形不成立。中美长端利差约265bp、短端USD-CNH利差约194bp，足以吸引做空人民币/持有美元资产的套利盘；但0%准备金只是移除远期售汇成本楔子，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "离岸与在岸人民币价差（CNH-CNY）的背离及其对资本流动与套利成本的影响", "summary": "对本卡任务前提做压力测试，结论：该前提在当前盘面不成立。截至 2026-06-28，CNH-CNY 价差收窄而非走阔，且落在升值侧——USD/CNH 约 6.7849，强于在岸中间价 6.8195 约 346 个点（约 0.51%）[自测算]。三环全断：价差被压缩（CNH/CNY ~0.997–1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中美利差走宽对中国货币政策空间及人民币汇率稳定性的影响分析", "summary": "截至2026-06-28，尽管中美10年期倒挂走宽至历史性的2.649%，人民币并不弱：USD/CNY约6.80（6月25日中间价6.8195），5月外储升至3.442万亿美元，3月外汇风险准备金率已降至0%。汇率锚是结构性的（贸易顺差、储备、美元走弱趋势），而非利差。因此人民银行可对深陷通缩（5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 8, "neutral": 1, "value": 62.4, "events": [ { "title": "中美利差倒挂对离岸人民币（CNH）汇率及跨境资本流动的传导效应", "summary": "支持卡片01的格局判断并补齐传导环节。截至2026-06-29，美国10年期国债约4.40%、中国10年期约1.79%，名义利差倒挂约−261bps（本轮最阔）；而真实利差约+39bps偏向中国。名义渠道在去杠杆中占主导，使CNH成为融资货币。鹰派美联储驱动的套利去杠杆经两条腿（CNH融资腿+全球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "summary": "2026年6月17日Warsh首会以12-0维持利率3.50%-3.75%，但SEP中位点阵从隐含降息翻转为隐含加息（3.4%升至3.8%），18人中9人看至少1次加息、17人看通胀上行风险，2026年PCE上修至3.6%。背景是5月CPI同比4.2%反弹与韧性就业（年内约57万非农）。2年期美债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国内需政策强度及对内需受益资产的支撑评估", "summary": "压力测试结论：中国政策提供相对护盾而非绝对护盾，对内需论点是有限支持。强处在跨资产隔离——人民银行逆周期、仍有降准降息空间[S2]，A股内需标的直接日元融资敞口低，人民币强于7、CNH近三年高[S9]，消除被动收紧反射。弱处是政策在对抗消退的需求：2026年5月社零同比-0.6%，为2022年12…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美元强势第二阶段与人民币/A股外资流传导", "summary": "本卡综合前序政策失误论证，落到汇率与资金流通道。美元处于2026年往返的第二阶段：DXY自1月95.57谷底反弹、议息后破100、6/25-26创13个月新高约101.37，但该段几乎全靠约75%的9月加息定价支撑，结构脆弱。USD/CNY约6.80为美元主导。关键在拆解美元下跌动因：套利通道（预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 2, "neutral": 0, "value": 88.8, "events": [ { "title": "日元套利平仓通过离岸人民币流动性抽水及掉期清算挤压双重渠道，系统性抬高离岸融资成本并技术性击穿USD/CNY 7.00关口，导致人民币套利空间彻底崩溃。", "summary": "本卡支持Research note 02，定量分析日元套利平仓对CNH的传导。BOJ加息和USD/JPY破位触发约349亿美元平仓，占离岸池近25%，推高隔夜HIBOR至10.81%-12.74%。掉期展期率跌至40%引发即期购汇压力，击穿USD/CNY 7.00。CNH套利Carry-to-Ri…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "summary": "截至2026-06-29，支持卡片01–03的逻辑链并延伸至终端利率与全球流动性维度。美联储6-17维持3.50%–3.75% [S2]、SEP 2026中值3.8% [S3]；新主席Warsh首场FOMC点阵图从降息翻转为加息，9名官员预期加息、年内加息概率50.4%、市场定价10月加息，反应函…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-740e729f22/", "networkHref": "/investor/research-network/#thesis-740e729f22" }, { "id": "thesis-9e97140446", "source": "tracked_thesis", "title": "中国电力设备出海进入订单质量与政策筛选期", "titleEn": "Chinese power-equipment exports enter an order-quality and policy filter phase", "lane": "中国电力设备出海", "laneEn": "China power-equipment exports", "laneHref": "/investor/lanes/china-power-equipment-exports/", "investableFocus": "中国电力设备出海：002028.SZ, 600089.SH, 601179.SH, 601126.SH, 300360.SZ, ETN, ABBNY", "stockUniverse": [ { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "600089.SH", "name": "特变电工", "market": "A-share", "role": "变压器/电缆/海外工程", "roleEn": "变压器/电缆/海外工程", "bucket": "core", "href": "/investor/stocks/600089.SH/" }, { "ticker": "601179.SH", "name": "中国西电", "market": "A-share", "role": "高压设备出口", "roleEn": "高压设备出口", "bucket": "core", "href": "/investor/stocks/601179.SH/" }, { "ticker": "601126.SH", "name": "四方股份", "market": "A-share", "role": "电网自动化与保护", "roleEn": "电网自动化与保护", "bucket": "core", "href": "/investor/stocks/601126.SH/" }, { "ticker": "300360.SZ", "name": "炬华科技", "market": "A-share", "role": "智能电表出海", "roleEn": "智能电表出海", "bucket": "core", "href": "/investor/stocks/300360.SZ/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "ABBNY", "name": "ABB", "market": "US ADR", "role": "全球电气化比较基准", "roleEn": "全球电气化比较基准", "bucket": "peer", "href": "/investor/stocks/ABBNY/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" } ], "stockTickers": [ "002028.SZ", "600089.SH", "601179.SH", "601126.SH", "300360.SZ", "ETN", "ABBNY", "GEV", "PWR", "VST", "CEG", "XLU", "600900.SH", "APO" ], "coreView": "赛道：中国电力设备出海。候选观察标的：002028.SZ, 600089.SH, 601179.SH, 601126.SH, 300360.SZ, ETN, ABBNY。当前共有 40 条证据、3 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-04", "lastSeen": "2026-06-29", "supports": 5, "risks": 152, "neutral": 3, "analystCount": 13, "sourceTypeCounts": { "研究笔记": 155, "研究主题": 3, "深度研究": 1, "研究线程": 1 }, "topTerms": [ "电网", "Meta", "变压器", "GOES", "PPA", "电力设备", "Eaton", "电网设备", "AIDC", "电力", "设备", "Chinese" ], "whatChanged": [ "2026-06-29: 并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "2026-06-29: 电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "2026-06-29: 电力设备出海高增长与国内“十五五”电网资本开支爆发锁定了龙头业绩的高能见度与利润确定性，是中证红利风格强化趋势下红利成长股的首选品种。", "2026-06-29: 电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对冲有效却对冲错轴，继续超配电网设备而非纯绿电发电。", "2026-06-29: 电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀" ], "practical": { "sectorLane": "中国电力设备出海", "sectorLaneEn": "China power-equipment exports", "exposureBucket": "电力与公用事业", "tone": "yellow", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "600089.SH", "name": "特变电工", "market": "A-share", "role": "变压器/电缆/海外工程", "roleEn": "变压器/电缆/海外工程", "bucket": "core", "href": "/investor/stocks/600089.SH/" }, { "ticker": "601179.SH", "name": "中国西电", "market": "A-share", "role": "高压设备出口", "roleEn": "高压设备出口", "bucket": "core", "href": "/investor/stocks/601179.SH/" }, { "ticker": "601126.SH", "name": "四方股份", "market": "A-share", "role": "电网自动化与保护", "roleEn": "电网自动化与保护", "bucket": "core", "href": "/investor/stocks/601126.SH/" }, { "ticker": "300360.SZ", "name": "炬华科技", "market": "A-share", "role": "智能电表出海", "roleEn": "智能电表出海", "bucket": "core", "href": "/investor/stocks/300360.SZ/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "ABBNY", "name": "ABB", "market": "US ADR", "role": "全球电气化比较基准", "roleEn": "全球电气化比较基准", "bucket": "peer", "href": "/investor/stocks/ABBNY/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" } ], "stockTickers": [ "002028.SZ", "600089.SH", "601179.SH", "601126.SH", "300360.SZ", "ETN", "ABBNY", "GEV", "PWR", "VST", "CEG", "XLU", "600900.SH", "APO" ], "primaryStocks": [ { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "600089.SH", "name": "特变电工", "market": "A-share", "role": "变压器/电缆/海外工程", "roleEn": "变压器/电缆/海外工程", "bucket": "core", "href": "/investor/stocks/600089.SH/" }, { "ticker": "601179.SH", "name": "中国西电", "market": "A-share", "role": "高压设备出口", "roleEn": "高压设备出口", "bucket": "core", "href": "/investor/stocks/601179.SH/" }, { "ticker": "601126.SH", "name": "四方股份", "market": "A-share", "role": "电网自动化与保护", "roleEn": "电网自动化与保护", "bucket": "core", "href": "/investor/stocks/601126.SH/" }, { "ticker": "300360.SZ", "name": "炬华科技", "market": "A-share", "role": "智能电表出海", "roleEn": "智能电表出海", "bucket": "core", "href": "/investor/stocks/300360.SZ/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" } ], "instrumentTags": [ "002028.SZ", "思源电气", "600089.SH", "特变电工", "601179.SH", "中国西电", "601126.SH", "四方股份", "300360.SZ", "炬华科技", "ETN", "Eaton" ], "reviewCadence": "每日复核", "confirmationSignals": [ "并网排队和变压器交期是否继续拉长", "表后自发电与 PPA 是否改变公用事业信用指标", "电力设备订单质量是否高于收入确认节奏" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "长端利率", "铜/GOES 成本", "监管电价", "信用利差" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 108 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 13, "recentEvents": 108, "supportRatio": 0.03, "riskRatio": 0.95, "conflictScore": 100, "evidenceDepth": 100 }, "score": 64.8 }, "events": [ { "date": "2026-06-29", "title": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "summary": "本卡片从制造和供应链视角对电网与电力约束进行压力测试。截至2026-06-29，关键电力设备（变压器与重型燃机）由于极高工艺与测试门槛、上游取向电工钢（GOES）瓶颈及寡头垄断，交期拉长至3-7年，生产机位已售罄至2029-2030年。云厂商BTM发电加剧了燃机机位争夺。预计2026-2027年该瓶颈将导致云厂商约800亿-1100亿美元资本开支递延兑现。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "电网", "变压器", "GOES", "电力设备", "电力", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]，PPI-CPI剪刀差隔离了政策利率。任务设想的'成本→CP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.75, "matchedTerms": [ "电网", "GOES", "电力设备", "电力", "设备", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "电网", "变压器", "电网设备", "AIDC", "电力", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备出海高增长与国内“十五五”电网资本开支爆发锁定了龙头业绩的高能见度与利润确定性，是中证红利风格强化趋势下红利成长股的首选品种。", "summary": "本报告定量评估了电力设备龙头在“全球出海”与“国内十五五电网更新”驱动下的业绩确定性。全球变压器交期拉长至107.8周，高出口毛利率 premium 拥有调价条款保护。压力测试表明，即使北美订单延迟40%，主要龙头季度净利下修仅为-2.55%至-10.54%，自由现金流影响可控。国内超过5万亿元电网资本开支抵消了输配电价降价压力。超配思源电气、金盘科技及高…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "电网", "变压器", "电力设备", "电力", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "summary": "本卡压力测试 Research note 6 的'调价条款已隔离利润'结论，认为其仅有条件成立。主流调价条款（MEC）挂钩有流动期货的铜/铝（出口超级周期转嫁率约80%[S1]），却普遍排除或封顶无流动指数的取向电工钢GOES——而GOES占COGS约28%[Research note 6]且处2026-2028结构性短缺。自测算显示：铜+30%/GOES…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "变压器", "GOES", "电力设备", "电力", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "summary": "支持第06卡\"并网延迟→利润压缩\"的传导论点，但将单向结论细化为分化。针对11GW数据中心建设缺口与12-18GW刚性局部电网缺口，48-60个月并网延迟通过两条通道传导：(1)经订单/PPA递延与尾款拖延压制发电子链开工率——5.7%销量递延经经营杠杆放大为约-21%利润压缩[自测算]；(2)将电网投资节奏重置向T&D侧，5万亿元资本开支周期以70-75…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 30.5, "matchedTerms": [ "电网", "变压器", "PPA", "电网设备", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心腿；半导体国产替代（¥2万亿AI建设计划+80%国产强制比例…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "电网", "变压器", "电力设备", "电网设备", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "summary": "本卡支持 Research note 02「上游原材料是最深约束」的结论，并锐化为一组非对称结构。GOES：自2019年变压器需求暴增（电力+119%、GSU+274%），远超GOES市场约5.4–5.6%的CAGR；GSU供给缺口2025年约100%，因美国唯一GOES生产商Cleveland-Cliffs（Butler Works，扩产2028上线）及…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "电网", "变压器", "GOES", "电力设备", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力需求增长对电网侧扩容及配电资产的压力测试", "summary": "压力测试结论：前序主线被证实并强化。2026年AI落地的约束性瓶颈已非GPU，而是电网边缘——电力变压器、中高压开关设备、取向硅钢（GOES）与接网排队。大型变压器交期3–5年、全球产能利用率约98%、美国接网积压1,500–2,300GW、枢纽接入排队4–7年，致美国2026约12GW管线仅约4GW在建（隐含约67%滑期，自测算）。这把AI算力部署拉长而…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.75, "matchedTerms": [ "电网", "变压器", "GOES", "Eaton", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电力’，并驱动策略从算力规模向算法效率和电…", "summary": "截至2026-06-28，本卡片通过对电网与变压器交期的压力测试，确认电网并网（5-7年交期）和变压器交付（2-4年交期）构成了AI算力扩张的终极物理边界。随着GB300（约135kW）及Rubin Kyber（约600kW）对800VDC和固态变压器（SST）的高密度配电需求增加，目前仅有不足8%的设施就绪。FERC否决核电表后直供，迫使所有新增容量重回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "电网", "变压器", "GOES", "电网设备", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "变压器供应瓶颈真实且比宽泛AI硬件更可防御，但只有“稀缺性+backlog+成本传导”的高压电网设备供应商具备防御性，行业整体仍有周期风险。", "summary": "截至2026-06-28，本卡压力测试电力基础设施与变压器供应链。结论是：变压器、电网高压设备和开关设备存在真实稀缺，交期、材料和定制化约束类似半导体前道瓶颈；但它们不是天然防御资产。防御性只属于具备高压/HVDC/开关设备能力、确定backlog、服务收入和原材料成本传导能力的供应商。2027-2028年扩产、数据中心负荷高估、GOES/铜成本和监管负担…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "电网", "变压器", "GOES", "电网设备", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "summary": "截至 2026-06-28，本卡压力测试认为：电力交付瓶颈会迫使 hyperscaler 从单纯追求新增 MW 转向 MW 生产率和存量优化，但不会终结扩产。Microsoft、Alphabet、Meta 可见 2026 年 capex 区间合计约 4,950-5,250 亿美元（自测算），同时仍披露容量约束；IEA 显示数据中心 1-3 年可建成，而新电…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "电网", "Meta", "变压器", "电力设备", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "A股应继续支持AI基础设施主线，但买点应从纯算力叙事转向供电约束受益资产，以35%/30%/25%/10%的科技成长、电网设备、公用事业红利和现金对冲哑铃执行。", "summary": "截至2026-06-28，本卡支持上游判断：AI基础设施交易没有被宏观衰退担忧证伪，而是被供电、电网接入、变压器和冷却约束重构。A股策略应从单一科技成长交易转为35% AI科技成长、30%电网及电力设备、25%公用事业红利、10%现金或对冲的主题哑铃。科技仍有上行，但电子和通信估值已处三年高分位；电网设备承接国网4万亿元十五五投资与数据中心供电瓶颈；公用事…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "电网", "变压器", "电力设备", "电网设备", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "AI算力扩容背景下电力设备厂商的利润质量评估：关注核心组件的毛利韧性。", "summary": "截至2026-06-29，本卡片压力测试变压器厂商“订单驱动→利润驱动”的跨越，结论是有条件且分化的。订单已普遍强劲（GE Vernova 储备1630亿美元、Eaton 数据中心订单+240%、金盘新签+64.48%），订单本身不再具区分度。一线龙头凭稀缺定价（GE Vernova 2026订单较4Q25涨10%-20%且即时扩张毛利）、出口高毛利结构（…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.75, "matchedTerms": [ "变压器", "电力设备", "Eaton", "电力", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "电力设备扩产会缓解AI数据中心并网瓶颈，但无法在2026-2027年完全解除大型变压器和输电接入的物理约束，因此算力利润扩散仍是受瓶颈约束的电网资本开支周期。", "summary": "截至2026-06-28，本卡支持上一卡的物理瓶颈判断，但强调扩产的时间结构：变压器、开关柜和电网设备扩产是真实的，Eaton、Hitachi Energy、Siemens Energy等制造商已披露新增产能和高积压订单；但大型电力变压器仍受定制设计、GOES/铜、测试槽位、运输和并网研究约束，新增产能更多改善2028-2030年供给，而难以完全解决202…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 37.75, "matchedTerms": [ "电网", "变压器", "GOES", "电力设备", "Eaton", "电网设备", "电力", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "summary": "截至2026-06-27，本卡片综合前序结论：电网/变压器瓶颈会延后并改道AI部署，但不会摧毁2028年前光模块出口龙头业绩能见度；真正尾部风险是2027年云厂商ROIC压缩。建议A股AI物理瓶颈组合使用40/32/12/8/8结构：变压器/电网设备、光模块、电力现金流锚、铜/GOES等输入端对冲、现金/对冲。光模块腿已拥挤，完整杠铃尚未终局拥挤，应在内部…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "电网", "变压器", "GOES", "电网设备", "电力", "设备" ], "href": "" } ], "timeline": [ { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "电子取代 HBM 成为 AI 链条上未来 12 个月最被讨论的瓶颈：1 GW 边际 AI 园区的电费在 2027 年从 $322M/年跳到 $926M/年，足以让 hyperscal…", "summary": "对 Research note 03 的 $1T capex / $7.5–10M/MW 营收密度救援叙事进行压力测试。结论：电力（不是 GPU）成为 2026 下半年起的硬约束；hyperscaler 在 2026–2027 锁定的边际 MW 全成本将达 $95–140/MWh，相对 2024 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "2026-2028年的关税和贸易限制会先加剧而非缓解AI基建的“铁”瓶颈，因为美国保护的国内GOES和变压器产能不足以快速替代高依赖进口链条。", "summary": "本卡对第06卡做压力测试并支持“硅-铁失配”判断：2026-2028年贸易政策不会从零制造变压器短缺，但会把短缺转化为地缘政治分配问题。DOE数据显示，美国LPT进口依赖度为82%，本土GOES只能满足12-20%需求，GOES和铜导线各约占LPT成本25%。Section 232在2026年4月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "\"新质生产力\"政策下的中国电网设备出海：退税政策校准与贸易壁垒对冲策略 第8张卡片（共10张）| 中国宏观分析师 | 立场：综合 本期会话主题： 美国 AI 数据中心负荷增长与公用事…", "summary": "\"新质生产力\"政策下的中国电网设备出海：退税政策校准与贸易壁垒对冲策略 第8张卡片（共10张）| 中国宏观分析师 | 立场：综合 本期会话主题： 美国 AI 数据中心负荷增长与公用事业电网灵活性 本卡主题： \"新质生产力\"框架下的电网设备出口退税政策调整与 Section 232/301 关税规避…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "盟友联盟对中国电网设备的部分排除是一项主动用 6–12 个月送电延迟和 15–25% 采购成本溢价换取供应链韧性的宏观选择，结果是延展盟友设备 OEM 超级周期、部分上限美国 AI …", "summary": "本卡作为 10 张卡片的最终综合，整合了从需求（Research note 01）、firm 负荷修正（Research note 02）、供给瓶颈（Research note 03）、设备防御性（Research note 04）、定价权（Research note 05）、跨资产轮动（Rese…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股变压器/电网设备30-40倍PE：AI送电瓶颈主题是已充分定价还是仍被低估？", "summary": "A股变压器/电网设备30-40倍PE：AI送电瓶颈主题是已充分定价还是仍被低估？ A-share Grid Equipment 30-40x PE: Is AI Power Delivery Bottleneck Fully Priced or Still Undervalued? global-…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "AI电网设备尚未见顶，但2026下半年必须从主题弹性切换到订单、材料、交付和利润率验证，A股配置应超配高质量电网设备龙头、低配低端beta。", "summary": "截至2026-05-13，本卡综合判断AI电网设备交易尚未出现决定性见顶信号，但已从广谱beta进入盈利验证期。2026下半年需重点观察三类信号：超大规模厂商是否维护或下修2027年AI基础设施资本开支，GOES和大型变压器交期是否明显缓解，上市OEM订单能否转化为收入和稳定利润率。A股配置上应维…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "‘物理交付死结’正在击穿‘地缘政治围栏’，迫使云巨头在非敏感地区结构性转向中国电力供应链以维持AI交付节奏，从而导致全球电力设备定价权从‘积压稀缺性’转向‘制造交付速度’。", "summary": "截至2026-05-13，本卡对电力设备交付瓶颈进行了全球宏观层面的压力测试。由于西方变压器交期拉长至36-48个月（由于GOES短缺和劳动力瓶颈），而中国仅需12-15个月，西方超大规模云服务商正通过“非敏感地区（NSR）”策略结构性地转向中国电力供应链。这种“交付周期套利”正在削弱西方工业巨头…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "美国和欧盟不会消灭中国电力设备出海周期，但会把发达市场收入从高质量利润释放项变成需要政策折价的订单。", "summary": "截至2026-05-14，本卡压力测试但不推翻AI电网设备出海逻辑。全球变压器和电网设备短缺仍支撑中国供应商订单，但美国和欧盟正在用关税、采购韧性、网络安全、外国补贴审查和零部件贸易救济压缩发达市场利润率。美国风险最高，Section 301/232、Section 122临时附加费和结构性产能过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "AI 长尾电力需求预测的下行风险真实存在但非对称：2026-2027 上半年已被物理承诺锁定，2028-2030 较一致预期下行 8-12%，对应 5-7 GW 基荷等效缺失，最先在…", "summary": "压力测试 IEA/Goldman/BNEF 一致预期的美国 AI 长尾电力需求曲线：若卡 02 的折旧悬崖驱动 2027 年云厂商 Capex 同比由 +25-30% 跌至持平/负值，则按每 $1B Capex 对应 ~10-15 MW IT 负荷、PUE 1.2 的换算关系，2028-2030 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 1, "risks": 3, "neutral": 0, "value": 40.4, "events": [ { "title": "Section 232 持续 + CRMA 采购偏好 + 中国出口许可收紧的组合，把 OEM 护城河从临时物理短缺升级为持久行政短缺，净强化 2027–2029 电力设备 OEM 定…", "summary": "三条政策时钟（美国 Section 232 GOES 复审、欧盟 CRMA 二级实施细则、中国出口许可与反倾销周期）叠加，未来 24 个月最可能的路径是碎片化加深但无头条贸易战：Section 232 续期+对最薄 Hi-B 给窄排除（≈55%），CRMA 二级细则 2026Q4–2027Q2 发…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "支持上轮判断：AI电力交易更应买变压器、变电站和GOES等瓶颈设备/材料，而不是无差别买入久期公用事业。", "summary": "截至2026-05-17，变压器瓶颈仍由产能窗口、合格劳动力和GOES供应共同决定。GE Vernova、Siemens Energy、Eaton的最新订单、积压和利润率显示，头部电力设备商正在把稀缺转化为价格与利润，而不是被GOES和铜成本完全挤压。新产能投资已经启动，但多数是2027-2028…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球宏观分析：地缘政治脱钩下的 AI 电力供应链重构与 ROI 压力测试 日期： 2026 05 18 分析师： 全球宏观分析师 立场： 压力测试 (Stress Test) 执行摘…", "summary": "全球宏观分析：地缘政治脱钩下的 AI 电力供应链重构与 ROI 压力测试 日期： 2026 05 18 分析师： 全球宏观分析师 立场： 压力测试 (Stress Test) 执行摘要 截至 2026 年 5 月，AI 电力叙事正在从简单的“稀缺”转向结构性的“通胀危机”。电力设备供应链（特别是特…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与电网的物理瓶颈作为非政策性硬顶，将算力设备的短期现金流向右推延，削弱了政策边际放松带来的向上弹性，并确立了电力设施相对算力设备的估值溢价。", "summary": "截至 2026-05-17，对政策估值框架进行压力测试：电力与电网构成了 AI 部署的物理“供应闸门”。PJM 等电网并网排队达 3-5 年，高压变压器交期维持 18-36 个月，基荷电力存在刚性缺口。这一物理硬顶使得算力设备（如 NVDA）的 DCF 现金流向右推延，即使政策边际放松，也难以出现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "电力设备是眼前的“订单簿瓶颈”，但终极约束是 7-10 年的电网准入周期，以及 AI 需求响应极低的弹性。", "summary": "电力设备（尤其是变压器）的极长交付周期与取向硅钢 (GOES) 的结构性缺口，已成为 AI 算力扩张的硬性物理约束。压力测试显示，AI 负载的极高资本开支决定了其“硬电力”需求特征，使得所谓的“需求响应”在经济上不可行。最终，2 年的 AI 建设周期与 7-10 年的电网升级周期之间的时空错配，构…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备OEM的积压订单验证了AI电力基础设施稀缺溢价，但同一批积压也证明近端供应不足以让所有2026-2027年AI数据中心公告容量按期落地。", "summary": "2026-05-20：支持AI估值从GPU叙事转向可交付电力基础设施，但压力测试显示供给只能选择性支撑，而不能普遍支撑所有已公告AI数据中心。Eaton、GE Vernova、ABB、Siemens Energy、Hitachi Energy、Schneider、Hubbell和Powell均显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "截至2026-05-20，AI电力瓶颈的核心不是全国缺电，而是高规格变压器和高压开关设备的工厂槽位、测试能力、交付周期与原材料可得性。", "summary": "截至2026-05-20，我们支持前序结论：AI算力扩张瓶颈已从全国总发电量转向站点级重电设备交付。Eaton、GE Vernova、Siemens Energy、Schneider、Hitachi、韩国UHV厂商及TBEA/Pinggao均显示订单快于可交付产能。Allianz称电网设备交付最高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 3, "neutral": 0, "value": 36.6, "events": [ { "title": "AI 算力扩张已通过 OEM 订单与单价同时验证了电网设备的需求确定性；在 GOES 与变压器产能的多年瓶颈下，未来 24–36 个月该子板块兼具量价齐升与 200–400bp 利润…", "summary": "截至 2026-05-22，AI 数据中心已切实成为电力基础设施的边际驱动力，GEV 电气化订单同比 +86%、ETN 数据中心订单同比 +240%、ETN 披露的美国数据中心积压达 228GW；高功率变压器交付周期延至 4–5 年，GOES 取向硅钢扩产周期 3–5 年；国网十五五投资 4 万亿…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器等电力设备的极度短缺及取向硅钢供应瓶颈验证了“时间尺度错配”，AI数据中心的扩张将持续受制于3-4年的物理电网硬件交付周期。", "summary": "截至2026年5月22日，全球电力设备市场面临极度紧缺。西门子、GE等四大设备商订单积压创历史新高，大型电力变压器交付周期拉长至36-48个月。上游取向硅钢（GOES）受制于北美产能垄断及全球供应不足，成为扩产的核心瓶颈。设备可用性已取代资金，成为AI基建最硬的约束。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI芯片厂商在通过架构与SKU分化重置AI负载的地理与功率形态以适应电网约束，但并未也无法真正消除变压器/GOES瓶颈——推理分化是唯一可能在边际上抑制电网设备需求的真实杠杆。", "summary": "截至2026-05-22，AI芯片厂商确在积极应对电力基础设施约束，但所有手段（机柜级液冷、FP4低精度、MoE稀疏、低TDP推理SKU、共封装光互连、机柜级功率封顶、表后自供电协同设计）都是把负载形态重塑得更易卖给现有电网，而非真正消除瓶颈。Blackwell→Rubin路线图上的旗舰训练芯片功…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" }, { "title": "核电与天然气桥接产能的资本密集型扩张正撞上电网设备、燃料循环和中游管道的硬性物理瓶颈，导致长期PPA等金融安排沦为因并网延迟而闲置的‘纸面期权’。", "summary": "本报告对先进核电与天然气桥接产能的资本扩张进行了严密的压力测试。研究显示，尽管长期合同优化了融资，但物理层面的电网互联延迟、大容量变压器（交期2-4年）短缺、HALEU燃料循环断裂及中游管道外输约束，将导致美国2028年和2030年分别面临32.9 GW和53.4 GW的实际闲置AI电力缺口。物理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "在2026-2028年间，电气化重装备及数据中心专用电源设备处于结构性“超强周期”，设备商产能已被预订满至2028年后，导致电网并网时滞成为AI算力上线的实质屏颈，促使重装备板块重估…", "summary": "截至2026-05-24，本报告评估了电力变压器、开关柜及数据中心电源系统在2026-2028年的供需缺口。大型变压器交期拉长至150-200周，因硅钢紧缺，美国严重依赖进口，现代电气和金盘科技订单锁排至2028年。开关柜交期翻倍，Powell与Eaton在手订单创纪录。Vertiv受益于800V…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 1, "value": 55.8, "events": [ { "title": "组合应保留云与半导体核心敞口，但把边际超配转向电网设备、可用电力、中国UHV和核电，因为AI瓶颈已从芯片订单转向通电交付确定性。", "summary": "截至2026-05-25，本卡综合前序结论：AI主线不应退出，但风险预算应从纯云/半导体beta转向“已通电AI交付beta”。建议AI电力主线100个风险单位中，MSFT/AMZN/GOOGL/META 30、半导体14、美国电网设备24、美国电力与BTM期权10、中国UHV/电网设备14、中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超大规模 AI capex 真的减速了吗：卫星+电网负荷 vs. 4 月业绩会 capex 走平语调的 reconciliation", "summary": "超大规模 AI capex 真的减速了吗：卫星+电网负荷 vs. 4 月业绩会 capex 走平语调的 reconciliation Is hyperscaler AI capex really decelerating? Reconciling satellite + grid-load evi…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 4, "neutral": 0, "value": 36.2, "events": [ { "title": "截至 2026-05-26，电力设备瓶颈并未均等阻碍算力落地，而是建立了极度分化的双轨制市场，通过35倍的高额投产延迟杠杆，重奖前瞻储备与套利通道，严惩后发缺乏设备保障的云厂商。", "summary": "截至 2026-05-26，压力测试显示电力设备交付延迟对AI基础设施扩产节奏呈现高度分化的特征。大型变压器交付期拉长至3-4年，但第一梯队超大规模云巨头（MSFT/GOOGL/AMZN）已通过数年前的订单锁定了产能（如Eaton的12年储备订单和GEV单季电气化大单）；Meta等后发者则面临严重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网扩容的第二瓶颈不是资本，而是重电设备的合格产能槽位及其上游材料。", "summary": "截至2026-05-27，本卡支持前序判断：AI扩张的第一瓶颈是电网可用容量，第二瓶颈正在转向大型电力变压器、开关设备、电缆和专业施工能力。DOE、Wood Mackenzie和JLL数据显示，LPT进口依赖、GOES/CTC材料短缺、128-143周变压器交付周期、44周开关设备周期和数据中心项…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美欧原产地规则与中国出口许可在AI电力供应链上呈现70%价格机制、30%数量机制的格局，仅在Hi-B取向硅钢与NdFeB重稀土磁材两节点真正物理改道，其余主要通过推高到岸成本并对美国…", "summary": "美欧'熔炼与浇筑'原产地规则与中国出口许可制是两套高度非对称的工具：前者基于规则、机械执行但被18-30个月执法滞后稀释；后者基于裁量、时延即政策。综合压力测试：2026-2027年这套监管栈是70%的成本推动通胀税（AI电力设备到岸成本+20-35%，美国核心商品CPI+15-25bp）、30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备瓶颈与机柜功率跃迁正在使'东数西算'政策栈出现四处机制裂缝，发改委大概率以补丁方式而非全面重构方式回应，节点准入条件与能源保障机制将在H2-2026收紧。", "summary": "卡片06对'东数西算'政策做政策维度的压力测试。结论：在120 kW/柜跃迁与变压器/开关柜18-24个月交付瓶颈下，现行政策栈出现四处机制裂缝——PUE门槛对液冷集群失效、绿电就近消纳对AI训练负荷无法做小时级保障、节点准入缺少设备交付函、跨省算力与电力结算分属两套监管。建议四条政策补丁：升级准…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 4, "neutral": 0, "value": 37.6, "events": [ { "title": "铜矿与电网设备的多环节并发短缺在2026–2028年内无法被精炼扩张所缓解，瓶颈定价权将持续从中下游加工与无保护融资载体被抽走，转移至上游铜资源与具备订单与交期的电网设备卖方。", "summary": "截至2026-05-27，本卡支持并强化第09卡的瓶颈定价权结论。ICSG 10月预测显示2026年全球铜矿增量仅+73万吨，而精炼需求继续高于矿端3%以上；Cobre Panamá（35万吨/年）仍停产、Grasberg在2025年9月事故后2026年销量预计同比下滑约35%、Escondida…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "重型电力设备OEM收入上能承接需求、时间上承接不了——订单簿2.5–4年可见度+定价权+利润率扩张构成对科技久期的反向对冲，核心超配GEV/ENR.DE/ETN/Hitachi系，中…", "summary": "2026-05-28结论：本卡支持第03卡的电网瓶颈论断。Siemens Energy Grid Tech订单/营收比2.28×、储备490亿欧元；GE Vernova Electrification订单+86%、设备储备390亿美元、B2B约2.5×；Eaton电气储备同比+48%、数据中心订单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 电力缺口是真实但温和的核心 PCE 楔，叠加电网设备供应链的中美对抗，将以「延迟降息 + 类滞胀」形态而非「重置通胀目标」形态影响美联储，并把宏观风险从拥挤的一阶 AI 硬件转…", "summary": "AI 电力交付滞后通过三条渠道（批发电价→服务、云算力涨价→SaaS、变压器 PPI→capex 平减指数）对美国核心 PCE 形成 2026 +8~18 bp、2027 +10~22 bp 的持续抬升（自测算），不足以重置 2% 目标但足以让核心 PCE 2027 年底仍卡在 ~2.4%，相对 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 叙事正从芯片竞争迁移到能源供应竞争，而把 2026–2028 钉死的不是发电装机，而是变压器、HV 开关柜与取向硅钢这条狭窄走廊。", "summary": "支持卡01的'物理墙'框架，并把瓶颈具体化到电力设备链：美国大功率变压器交付周期已拉长至120–210周、北美互联排队>2,600GW、取向硅钢(GOES)成为物理瓶颈。GEV/ETN/ENR.DE/Hitachi Energy/POWL 等的订单簿连续6个季度book-to-bill>1.2，订…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "2026-05-28 核心判断：美国电网瓶颈的宏观风险不是一次性CPI爆点，而是受监管电价的多年粘性和长久期公用事业融资供给对美债曲线的边际挤压。", "summary": "截至2026-05-28，本卡对电网/公用事业链做压力测试：美国大型电力变压器2019年进口依赖度82%，常见交期36个月、最高60个月；2026年关税框架把部分金属密集型电网设备纳入15%关税至2027年。结论是，变压器短缺不会单独制造广义CPI冲击，但会通过费率申请形成粘性的电价通胀底部；同时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 0, "value": 52.4, "events": [ { "title": "关键设备交付周期拉长至2.5-3年以上，且受限于GOES与铜的全球结构性短缺，电力设备产业链的产能扩张短期内无法匹配AI数据中心爆发式需求，行业物理瓶颈为设备OEM厂商（GEV、VR…", "summary": "截至2026年5月29日，在AI算力与电网升级需求爆发下，电力设备产业链（GEV、VRT）面临严重物理产能瓶颈。2026年变压器交付周期拉长至128-144周。虽然各大厂商积极在越南和美洲扩张产能，但受限于取向电工钢（GOES）和铜的结构性短缺（GOES在2025年存在约30%供需缺口且将持续至2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "支持将资产向电网设备和电气基础设施权益进行分层切换，但警告当前板块估值已过度透支，且120周以上的变压器交期将固定价格积压订单暴露于铜价暴涨和公用事业监管降温的双重风险之中。", "summary": "支持将资产敞口从单一铜beta向电网设备和电气基础设施权益切换的逻辑，但必须在个股选择上保持极强的估值纪律。当前Eaton（38x-40x P/E）和Powell（45x-57x P/E）等龙头估值已严重透支五年周期。由于120周交期，老旧固定价格订单正暴露于铜价暴涨的毛利率挤压风险中，且公用事业…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 3, "neutral": 0, "value": 55.8, "events": [ { "title": "AI算力功耗爆发与全球/国内电网升级大周期共振，驱动电力设备龙头在手订单与利润暴增，完美闭环A股高股息防御与电力设备出海的杠铃配置策略。", "summary": "截至2026-05-30，本卡支持AI算力功耗激增将有效带动电力基建盈利的判断。Q1 2026数据证实电力基建进入超级景气周期：Vertiv在手订单达150亿美元（营收+30% YoY），Eaton在手订单达228亿美元（全球电气订单+73% YoY），中国2025年变压器出口达646亿元（+36…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "变压器等核心电力设备的极端交付延迟将直接打破二三线AI算力中心的“成本回收”财务模型，使其面临高昂的技术折旧与现金流断裂风险。", "summary": "截至2026-05-30，本报告对变压器等电网设备的供应链瓶颈进行压力测试。2026年变压器交期已长达2.5-5年，造成了GPU资产到位却无电可用的错配风险。这种物理延迟将吞噬未能提前锁定电力设备的投机性算力中心的现金流，导致其技术折旧加速并面临违约。结论支持大型设备商与具备供应链确权能力的超大云…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网瓶颈真实但可移动：hyperscaler把边际GW送往ERCOT/海湾/东盟，许可改革与表后燃机/核电是旁道，关税让设备短缺更集中——结果不是美股公用事业广义牛市，而是少数OEM…", "summary": "从美/欧/日宏观与地缘政治视角对Card-05的'电力是新瓶颈'主线做压力测试。我不否定电网瓶颈，而是为它划界：(1) 边际AI GW正在通过Stargate ERCOT、Stargate UAE、沙特HUMAIN、马来西亚柔佛等通道离开美国队列，部分掏空美股公用事业敞口；(2) Trump 2.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "铜铝结构性缺口确实存在且供应侧无法在2026–2028年缓解，但tier-1电力设备OEM凭借指数化合同和产能扩张，能将+30%铜价冲击对EPS的影响压在−5%量级，远低于在手订单驱…", "summary": "截至2026-05-30，铜铝结构性缺口（2026E缺口约60万吨、大摩目标价1.25万美元/吨）成立且供应端无法在2026–2028年补缺，但Research note 05对电力设备龙头的多头判断通过材料端压力测试，须以分层形式落实。tier-1 OEM（中国西电、特变、国电南瑞、Eaton、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 6, "neutral": 0, "value": 31.2, "events": [ { "title": "Section 232关税升级通过全额价值计税与严苛的USMCA原产地规则，已将美国电力设备成本推向永久性高位，并将电网扩张速度物理锁定在受劳动力和GOES产能约束的国内生产水平。", "summary": "2026年4月Section 232关税改革将评估基准从“金属含量”转向“全额海关价值”，导致进口变压器成本激增22-28%。尽管OEM加速本地化，但40万吨的GOES结构性缺口与210万制造业岗位缺口已将“过渡性成本”固化为长期供给上限。电网扩容速度受限于国内工厂的招聘与培训周期，而非仅资本投入。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网瓶颈不是欧美潜在GDP的硬天花板，但已是AI生产率红利兑现时点的2-3年延迟与核心商品PPI的持续高位驱动；Section 232升级是决定它停留在过渡税还是固化为真正上限的…", "summary": "截至2026-05-31，本卡片对Cards 03-05的微观结论做宏观压力测试：电网瓶颈+OEM定价权意味着电力设备PPI（BLS WPU1174自2020年累计+65-70%）将持续高于核心目标，并把成本转嫁给整体欧美Capex基数。美国T&D投资从2024年约1700亿美元升至2025E的2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力高密化与电网瓶颈下的配电设备及温控资本开支压测", "summary": "我们将 Card-03 的 +100 TWh AI 算力用电增量进一步折算为工业制造端订单脉冲：2026 年中国 AI IDC 资本开支增量约 RMB 1,650–2,100 亿元[自测算]，其中配电（变压器/开关柜/母线槽）410–530 亿元、UPS/HVDC/BBU 300–380 亿元、液…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器（交期超160周）与开关设备已取代GPU成为AI算力部署的绝对物理瓶颈（Critical Path），将数据中心上线周期拉长至3-4年，从而使Eaton、Schneider等全…", "summary": "5月的数据进一步确认，变压器（超160周）与高压开关设备（90-100周）的极长交付周期已成为数据中心建设的关键路径。相较于6-8个月的厂房建设，电力设备的3-4年等待期使得算力上线的物理配给机制前置。这一供应缺口直接转化为工业龙头的业绩红利：Eaton与Schneider 2025/2026年积…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "容量电价保护发电、输配电价保护电网，但中国一次电网设备厂商不在任何顺价授权链路内，叠加集采评标基准价机制和稀缺的金属联动条款，大宗高位期一次设备和电缆的毛利空间会被结构性压缩。", "summary": "截至2026-06-02，本卡从政策结构层面压力测试容量电价和输配电价改革是否能够为一次设备厂商保留利润空间。结论是无法。容量电价（发改价格〔2023〕1501号，2024年起执行）只为发电企业修复资产负债表，输配电价第三/第四监管周期允许电网将合规Capex纳入RAB回收，电力市场化交易只在发-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 1, "neutral": 0, "value": 64, "events": [ { "title": "Compute premium正沿AI电力瓶颈从需求栈再分配到物理供给栈：midstream、燃气轮机OEM、棕地核电IPP、电网设备和铀燃料承接租金；超大规模云因capex/收入失…", "summary": "本卡作为大类资产配置师，综合卡01–03给出跨资产配置地图：在2026年四大云厂7,000–7,250亿美元capex、其中60%+用于电力而非芯片的背景下，compute premium正从需求栈（超大规模云+高估值AI软件）迁移到物理供给栈。第一层承接：管道firm transport（KMI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 5, "neutral": 0, "value": 41.4, "events": [ { "title": "制造端增量能在2028年后缓解电网设备短缺，但无法在4-6年GPU经济折旧窗口的前半段抵消大型变压器和并网硬件瓶颈。", "summary": "截至2026-06-03，工业制造端支持前序电力瓶颈判断：电网设备 OEM 正在扩产，但大型电力变压器和 GSU 变压器仍是关键长周期约束。Wood Mackenzie 披露 2025 年二季度 GSU、电力变压器和开关设备平均交付周期分别为 143 周、128 周和 44 周；DOE 指出新建变…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 公用事业权重下调后对电网基础设施与能源基建板块的风险覆盖评估", "summary": "2026-06-03，本卡压力测试 Research note 06 的 22% 至 17% AI 公用事业降权。结论：降权合理，但必须从长久期美国发电 / IPP 暴露中执行，不能压缩电网设备底仓。IEA、EIA、PJM、DOE 与 Eaton / Quanta / Vertiv 数据显示，AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网容量与电力设备交付的物理刚性约束已成为 AI 算力 Capex 转化为产能的核心瓶颈，定价权与超额利润将沿‘芯片→变压器/燃气轮机/核电 PPA→GOES 钢’向下游电力基础设施…", "summary": "本报告从能源基础设施物理层支持 TMT 卡观察——AI 算力订单的收入转化率被‘平滑化’，根因是电网容量、大型电力变压器 (LPT) 与重型燃气轮机的结构性刚性瓶颈。截至 2026-06-03，美国数据中心 2030 年负荷展望升至 580 TWh，并网队列存量 2,600 GW，中位等待 53 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "USMCA原产地规则（ROO）的收紧将使依赖进口组件的墨西哥电力设备组装产能对冲逻辑彻底失效，而沙特产能也面临美国 EAR/FDPR 针对 AI 基础设施的次级制裁风险。", "summary": "2026-06-03：本卡从政策监管视角压力测试出海产能对冲逻辑。USMCA 2026 联合审议实质性收紧了原产地规则，强制将“熔炼和浇注”标准延伸至变压器等核心电力设备。由于中国控制多数 GOES 产能，单纯依靠“组件进口+墨西哥组装”的产能将无法通过 RVC 测试，对冲逻辑彻底破产。同时，沙特…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 1, "neutral": 0, "value": 74.8, "events": [ { "title": "政策干预（包括DPA资金、关税调整和并网改革）在12-18个月内无法打破‘通电滞后’的物理时延瓶颈，保护主义关税和能效新规反而延长了供应链交期并抬升了电网建造成本。", "summary": "本报告对旨在打破‘通电滞后’的电网政策干预进行了压力测试。结果表明政策无法在12-18个月窗口内起效。DPA资金扩产的物理建设周期长达33个月（受制于21个月设备交期及6个月技能培训），产能迟至2029年释放。4月新加征的Section 232关税按成品全额征收，使大型电力变压器进口成本大涨7.5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决", "summary": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决 AI Capex Pivot to Grid/SMR: Has Consensus Already Been Repudiated by NVDA/MRVL Price Actio…", "sourceLabel": "研究主题", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 3, "neutral": 0, "value": 56.2, "events": [ { "title": "电力确是硬上限但约束在电网设备与并网而非 MWh，alpha 集中在 OEM 与受监管 T&D，搁浅风险锁定 2025–26 批发电源项目融资 vintage。", "summary": "电力是真实但时间有限的 AI 物理瓶颈，资本可压缩约 30–40% 延期但不能消除：约束子层是电网设备 + 并网（变压器交期 3.5–5 年、美国并网排队约 2.6 TW、PJM 群组 ~286 GW），而非装机 MWh 本身 [S1][S2][S3][S11]。最干净的可定价 alpha 在物理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力瓶颈催生的“去电网化”倾向，正将估值溢价与定价权从受ROE上限约束的传统电力公司，转移至享有高毛利与长周期订单积压的非监管电气设备及微电网制造商。", "summary": "截至2026-06-05，云服务商为绕过5-10年电网并网排队，正加速转向自备天然气和长期SMR微电网模式。这导致公用事业面临“大客户脱钩”与资产闲置风险。然而，由于监管（如FERC否决Talen-AWS案）和高额备用电费，完全脱网难以实现。估值溢价正从受ROE限制的公用事业（PE 15-18x）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "重型电力设备（特别是LPT与HV GIS）是AI自持电力比SMR更紧的物理瓶颈，头部设备商定价权与毛利率扩张真实但已被估值充分定价，风险/回报由'困境反转'转入'高位拥挤'。", "summary": "截至2026-06-05，大型电力变压器（≥100 MVA）交付周期已从2021年约50周拉长至130–210周，订单排至2028–2029年；HV GIS与中压配电变压器同步紧张。变压器环节单独足以将CSP自持电力首电时间推迟到2029–2030年，深于card-3识别的SMR瓶颈。Eaton电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "铜/GOES供应风险不会破坏AI-电力压舱石逻辑，但会把胜负手从简单做多电网设备转为区分投入品对冲与设备商毛利率传导能力。", "summary": "截至2026-06-06，本卡对card-09的材料压舱石逻辑做压力测试：铜价格高且波动大，但最新公开精炼铜数据仍显示2026年短期并非确定性全球短缺；GOES才是更清晰的结构性瓶颈，因为其专用、集中、难替代，并直接决定大型变压器交付。结论是保留15%铜/GOES材料仓作为保险，同时继续超配电网设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "政策可以软化AI电力设备供应墙，但无法在2026年用贸易壁垒或补贴绕过GOES和大型变压器的物理扩产周期。", "summary": "截至2026-06-07，本卡压力测试上一卡“GOES与变压器构成物理供应墙”的结论。判断是：关税、出口管制和僵硬国产化规则不能绕过冶金、专用设备、测试和工厂建设周期，反而可能加剧2026年稀缺；真正有用的是盟友采购、关税与BABA豁免、规格标准化和过渡库存。产业补贴必要但滞后，Cleveland…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 1, "risks": 1, "neutral": 0, "value": 80.4, "events": [ { "title": "电力并网与设备瓶颈是较半导体封装更难通过资本开支化解的刚性约束，将在2026年构成宏观供给侧滞胀脉冲并延缓AI红利释放。", "summary": "截至2026-06-08，本卡片压力测试表明，电网并网排队（3-7年）与大型变压器交付（2-4年）等物理瓶颈将取代半导体先进封装，成为2026年AI硬件释放期更长久的滞胀驱动因素。GOES电工钢和铜的供应链紧张将导致电力设备商在重度压力下面临超2000 bps的毛利率崩塌，形成供给侧成本推升与算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决", "summary": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决 global-macro、thematic-researcher、tmt-analyst 三人独立得出 'capex 由 GPU 服务器向电网/SMR/变压器分流' 的同一结论（g…", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 2, "neutral": 0, "value": 67.8, "events": [ { "title": "AI 基础设施的‘物理主权’已转移至掌握变压器核心组件（GOES、分接开关）与垂直整合能力的工业巨头，2026年的投产节点由‘变压器槽位’而非‘芯片配额’决定。", "summary": "2026年6月9日：本卡通过物理交期压力测试确认，电力变压器（交期3-5年）与开关柜（1.5-2年）已成为AI落地的‘刚性闸门’。工业巨头正从设备商转向‘产能地主’，Eaton通过‘从电网到芯片’的垂直整合与Boyd/Motivair的液冷并购锁定了高溢价份额；Schneider则凭借预制化模块（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与高压开关设备的产能扩张因制造周期、取向硅钢及分接开关的硬约束而严重滞后，无法在2027年AI需求高峰前达产，电力交付瓶颈将成为限制AI算力营收变现的核心硬约束。", "summary": "压力测试确认，变压器及高压开关设备扩产无法在2027年前满足AI基建需求。第一梯队大容量变压器交付期拉长至48-60个月，新订单排至2029年后。Hitachi、Siemens等厂扩产基本于2028年后投产，Eaton琼斯维尔厂2027年投产后仍需爬坡。高牌号取向硅钢（GOES）紧缺与有载分接开关…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 3, "neutral": 0, "value": 63.2, "events": [ { "title": "AI算力建设正从“依赖电网”转向“离网自治”，SOFC与中国电力设备出口链通过解决交付周期错配，成为捕捉AI基建溢价的核心抓手。", "summary": "2026年AI算力扩张的瓶颈已从GPU转向电力并网，使得SOFC与变压器成为核心资产。SOFC通过“时间套利”（交付周期<12个月对比电网并网7-10年）为AI厂商节省数十亿美元机会成本，LCOE在$58-110/MWh区间已具竞争力。中国厂商凭借全球60%的变压器产能及对取向电工钢（GOES）的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "“离网自治”模式面临美欧“安全与电力绑定”和关税/碳税的精准拦截，主权AI项目在放弃中国电力设备以获取AI芯片（交付延迟36-48个月）与使用中国设备（芯片断供）之间面临难以调和的结…", "summary": "本卡对AI“离网自治”模式进行压力测试。美欧正通过“安全与电力绑定”将AI芯片出口与排除中国设备挂钩（如阿联酋G42项目）。同时收紧FEOC与301关税以拦截含中国取向硅钢（GOES）的第三方变压器；欧盟也于2026年实施CBAM并禁用中国逆变器。开发者被迫在“使用中国设备但无芯片”与“脱钩但交付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备超长交付周期及原料紧缺导致大型电力项目面临商运推迟与债务违约风险，使AI数据中心容量扩张受到刚性的“瓶颈延迟”制约，佐证了“AI疲劳”本质上是物理并网与供给瓶颈而非需求侧断崖…", "summary": "截至2026-06-11，电网变压器设备供应缺口及原材料（取向硅钢/GOES与铜）紧缺已对AI数据中心并网建设进度构成物理刚性约束。LPT与GSU变压器交付期已拉长至2.5-4年。由于DOE能效新规，美国GOES缺口扩大至19.6万吨/年，进口依赖度攀升至57.6%。即便采用以铝代铜，也会因线圈体…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 0, "value": 58.6, "events": [ { "title": "AI基础设施中的电力与电网设备供应瓶颈研究", "summary": "确认：变压器与配电设备缺口已是AI数据中心交付的物理硬约束，非假设风险。2026年美国约12 GW已宣布产能中仅约三分之一在建，约50%–58%管线被实质性延迟；高压变压器交期拉长至4–5年，价格较2019年涨77%–80%。利润率正沿链条向下转移、向卡点集中——形成四层租金地图：第1层GOES材…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网设备需求仍强，但GOES与铜的双重冲击会把固定价Backlog转化为毛利率风险，因此应战术性超配长江电力类现金回报资产、精选具备价格传导能力的输配电设备商。", "summary": "截至2026-06-12，本卡支持并细化此前“AI电网物理瓶颈”判断：需求真实，但在GOES+20%与铜+20%情景下，固定价变压器订单会先体现毛利率风险。基于DOE大型变压器成本结构和公司2025年披露数据，自测算显示特变电工电气设备、中国西电变压器和中国西电整体输变电制造在0%传导下毛利率分别…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "去全球化导致美欧电网设备及原材料支付高额地缘溢价，并通过公用事业费率基数和新能源购电协议（PPA）向终端电价传导，正在欧美引发系统性的‘绿色通胀’。", "summary": "截至2026-06-13，本卡压力测试表明，关税及安全限制使美国取向硅钢（GOES）及成品变压器面临422%和80%-150%的地缘溢价。这每年为美国公用事业费率基数注入150亿-200亿美元通胀，拉高零售电费1.5%-2.0%。同时，并网延迟和设备溢价推动美国太阳能PPA价格同比大涨13%至$6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在本土变压器长交期与并网延时常态化下，美国数据中心向非传统市场进口设备并承担10%-50%的进口/转口关税是规避沉没资产延期损失（NPV侵蚀-$200M）的理性财务选择，这导致全球A…", "summary": "截至2026-06-13，本终局卡片合成电网并网队列（~2,600 GW[S1]）及大型电力变压器（LPT）长交期（144-260周[S2]）约束下，美国AI数据中心通过非传统市场（土耳其、巴西、印度、越南）进行变压器“地缘套利”的宏观经济与关税风险。自测算显示，若COD延误3年且Capex已沉没…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 2, "neutral": 0, "value": 72, "events": [ { "title": "AI扩容的核心瓶颈已全面转移至变压器、变电站节点与基荷电力等物理层，这要求权益市场削减短期AI软件收入预期，重估发电商、电网OEM及相关核心材料的结构性成长溢价。", "summary": "截至2026-06-14，本卡片完成了AI基础设施“电网墙”向跨资产配置的最终战略映射。综合变压器交期、并网队列与基荷短缺三大物理断点，我们建议进行结构性轮动：超配拥有24/7核/气基荷定价权的独立发电商（如Vistra、Constellation）与具有极强订单能见度的电气设备OEM（Eaton…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 数据中心的关键约束正在从“有没有电”转向“能否拿到变压器、开关柜和完整电气链条的交付槽位”。", "summary": "截至2026-06-14，本卡片支持但收紧前两张卡片结论：AI 扩容的下一个断点更可能是电网设备，而非发电资源。大型电力变压器平均交期 128 周、GSU 144 周，Eaton、Schneider、Siemens Energy、Hitachi Energy、Prolec GE 与 WEG 均在扩…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 4, "neutral": 0, "value": 61.4, "events": [ { "title": "中国电力设备出口正凭借‘速度溢价’成为全球AI基础设施的物理基石，并为人民币提供了结构性的宏观对冲价值。", "summary": "2026年全球AI电网约束下，中国电力设备（变压器、特高压、储能）正从‘成本竞争’转向‘速度竞争’，其12-18个月的交期远优于西方的5年 backlog。中国通过控制取向电工钢（GOES）和电力硬件供应链，获得了显著的全球‘速度溢价’（出口均价上涨33%）。这不仅推升了2025年创纪录的1.19…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "随着AI物理瓶颈从芯片转向电网可交付性，定价权正全面向上游电力资产与输配电设备商迁移，应超配瓶颈控制方并对电网依赖型云厂商及芯片标的引入折价。", "summary": "本卡片将2026年电网物理交付瓶颈转化为权益资产配置框架。建议超配控制变压器、液冷与在运核电等瓶颈环节的设备商与独立发电厂（如GE Vernova、Eaton、Vertiv、Constellation）；对依赖电网队列且发生租约撤销的云厂商（如Microsoft、Meta）引入20%-30%的电网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与电网瓶颈对Tier-1 2026–2027并网影响已基本被锁单、棕地与表后发电对冲；真实滑期约3–7 GW集中在2028+绿地与Tier-2投机建设，Research not…", "summary": "本卡(10/10)压力测试变压器/GSU/电网交期对Tier-1 AI数据中心并网的真实约束。结论：瓶颈真实存在但Tier-1时间表已基本被定价——2022–2024年下达的多年期框架订单、棕地复用既有并网点、表后核电与现场燃气PPA共同吸收了120–210周的设备交期。自测算显示2027–202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年AI算力扩张引发的能源瓶颈：从散热需求转向电网容量与能源供应保障。", "summary": "在'算电协同'框架下，2026年AI算力扩张的硬约束已从GPU与散热上移至中高压输配电层。北美100 MVA以上大型电力变压器交付周期已达120–156周（2019年约50周），中压开关柜52–80周；GOES上游产能仅约10家合格供应商，扩产需3–4年。寡头（Hitachi Energy、Sie…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 8, "neutral": 0, "value": 38.8, "events": [ { "title": "截至2026 06 15，通过对五大云厂商6020亿美元Capex计划、AI数据中心3倍金属强度以及中游设备厂商（Eaton, Vertiv, Schneider）创纪录的150亿美…", "summary": "截至2026 06 15，通过对五大云厂商6020亿美元Capex计划、AI数据中心3倍金属强度以及中游设备厂商（Eaton, Vertiv, Schneider）创纪录的150亿美元积压订单的深度穿透，我确认AI与电网基础设施投资已实质性转化为铝、铜及电力设备的物理订单流。这一“订单实物化”验证…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国电力装备外需足以覆盖行业层面的供应链重构成本，但超额收益将集中在材料、认证和渠道能力最强的头部企业，而非所有出口商。", "summary": "截至2026-06-16，本卡片压力测试全球变压器超级周期下中国电工钢与电力设备出海的利润传导。结论是外需足以支撑行业继续上行，但不能简单抵消所有地缘贸易壁垒；真正受益者是掌握高牌号GOES、具备高压/特高压认证和海外服务能力，并能在必要市场进行最终装配或服务本地化的企业。美国50%钢铝关税和欧盟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "把firm power稀缺主线落地为风险预算分腿的跨资产哑铃，由利率、拥挤度和设备交期三类客观触发器驱动调仓，捕获稀缺租金而非押注AI链方向。", "summary": "截至2026-06-15，本卡综合第05–07卡的firm power约束主线，构建跨资产哑铃组合：多头约65%风险预算（受监管电网L3、商业核电IPP L2、长周期电力设备L1、经认证SiC/GaN平台L4），空头/低配约35%（缺电GPU租赁S1、未锁并网DC开发商S2、无定点SiC供应S3）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力配电设备产能扩张与供应链瓶颈：AI算力建设的硬约束 分析师：工业制造分析师（industrials analyst） 立场：support（强化 Research note 02…", "summary": "电力配电设备产能扩张与供应链瓶颈：AI算力建设的硬约束 分析师：工业制造分析师（industrials analyst） 立场：support（强化 Research note 02 的\"电力设备瓶颈是AI算力部署真正天花板\"判断） 工作日期：2026 06 15（Asia/Singapore） …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 3, "neutral": 0, "value": 70.2, "events": [ { "title": "变压器与开关柜交期已经把AI扩容约束从电力需求预测推进到工业制造产能约束，证实chips与grid属于同一physical-AI capex风险桶。", "summary": "截至2026-06-16，本卡支持“电网设备交付周期证实物理红线”的判断。Wood Mackenzie 报告显示大型变电站与发电机升压变压器交期为80-210周，且美国2025年电力/配电变压器供应缺口估计为30%/10%；POWER Magazine 指出开关柜2025年Q2平均交期44周。Ea…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力的“快时钟”正与电网基础设施的“慢时钟”发生剧烈碰撞，变压器产能缺口将迫使数据中心建设周期长期化并推高资本开支，形成实质性的硬约束。", "summary": "截至2026年6月16日，变压器与电网设备供应链已成为AI算力扩张的实质性物理瓶颈。大型电力变压器（LPT）交期延长至36-60个月，核心原材料取向电工钢（GOES）供应高度集中且价格较2020年上涨50-80%，导致变压器综合成本上涨60-80%。约30-50%的2026年计划产能面临延迟，建设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2029年电网设备扩产会缓解但难以消除交付瓶颈，关键稀缺性正在从变压器OEM装配上移到材料、组件、测试和高压系统集成。", "summary": "截至2026-06-16，本卡压力测试第09卡：变压器和关键电网设备扩产真实，但2026-2029年仍难完全跑赢AI数据中心、可再生能源并网、电网替换和电气化带来的MW/MVA交付需求。2028年前大部分大型电力变压器、HVDC电缆、开关设备和上游组件仍处紧缺。真正的瓶颈不只是OEM装配，而是GO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 3, "neutral": 0, "value": 71.6, "events": [ { "title": "全球电网核心设备及原材料的产能赤字已构成刚性物理准入壁垒，导致产业利润池从算力/晶圆厂资本支出方结构性向重电OEM厂商转移。", "summary": "大功率汽轮机与超高压变压器的严重产能缺口（全球超高压变压器短缺约30%，交期延至2-4年；大型汽轮机交期延至5-7年）和取向硅钢（GOES）等关键材料瓶颈，对算力中心和晶圆厂构成不可逾越的物理准入壁垒。议价权转移令电力设备OEM（如HD Hyundai Electric和GE Vernova）利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国电力设备出口受益是真实结构性机会，但上限取决于地缘政策折价；国内 AI 算力扩张的核心约束将从芯片采购转向局部电力接入和稳定电源。", "summary": "截至2026-06-18，本卡 synthesize 并部分支持上一卡：全球电力设备短缺是真实瓶颈，中国在变压器、HVDC/UHV、开关设备和数据中心供配电环节具备结构性出口机会，但美国/欧洲关键电网的安全审查会限制份额。国内 AI 不是全国性缺电问题，而是局部“合格兆瓦”问题：IEA 估算中国数…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI主题内部的定价权正从芯片端的硅片稀缺向电力、并网熟地和变压器等物理基础设施供应链迁移，电网并网延迟构成了对云巨头闲置算力资本的巨额惩罚，迫使价值向能源与硬件节点转移。", "summary": "截至2026年6月18日，由于全美并网队列激增至2,600 GW且平均等待时间达5年，AI基础设施的核心瓶颈已从GPU芯片稀缺转移到电力、并网及电力设备。定价权正向控制熟地、独立发电和电网设备的物理环节迁移，以规避巨大的闲置算力沉没资金惩罚（10万张B200芯片闲置年资本占用成本达1.8亿美元）。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 2, "neutral": 0, "value": 79, "events": [ { "title": "G2/G3瓶颈没有被效率叙事推翻，反而因变压器、电气设备与并网队列的多年约束而强化近现金电力设备订单资产的相对优势。", "summary": "截至2026-06-19，本卡支持Research note 02：AI电力交付瓶颈是真实供给侧闸门。美国数据中心用电2028年可达325-580 TWh，电网设备采购和并网队列仍是多年周期；美国活跃并网队列2025年底仍超2,060 GW，PJM数据中心区等待36-48个月；欧盟并网等待可达2-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-19，AI电力设备不是需求破裂，而是订单、交付、验收与收入确认的时间错算风险正在上升。", "summary": "截至2026-06-19，本卡支持AI电力瓶颈主线，但压力测试其收入确认节奏：GE Vernova、Siemens Energy、Eaton、Vertiv的订单和积压订单均很强，问题不在需求真假，而在物理交付、客户侧送电、合同验收和会计收入确认之间可能出现错位。建议继续持有电网与能源设备链，但降低…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 3, "neutral": 0, "value": 74.4, "events": [ { "title": "电网吸纳能力瓶颈会削弱AI投资对近期r*的无摩擦上行冲击，但会把周期拉长并转向电力设备资本开支，因此投资需求逻辑被折价而非被证伪。", "summary": "截至2026-06-19，本卡压力测试Research note 06的r*上行逻辑。结论是：电网与变压器瓶颈会降低近期资本开支效率，但不会取消投资需求渠道。Goldman估算2026-2031年AI基础设施资本开支为7.6万亿美元；IEA提示约20%规划数据中心项目存在电网延期风险。套用于202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力确是 AI 算力扩张的真实瓶颈，但锁喉点是 LPT + 并网排队而非发电量本身，这把 SPCX/xAI 算力飞轮落地推迟 6–12 个月并把价值迁移到电力设备与核电 IPP。", "summary": "工作日 2026-06-19。从能源行业视角对 Research note 02 的'电力是最终瓶颈'结论进行压力测试：方向正确，但瓶颈不在发电量，而在并网排队 + 大型电力变压器 (LPT) + 高压开关 + 熟练劳动力四道串联约束。Colossus 2 (2 GW) 年用电量约 20.8 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大型高压变压器长达5年的物理交期瓶颈已锁定供给端产能，为核心电气厂商带来了超长业绩见能度与极强的定价权，从而证实变压器瓶颈是其业绩高速增长的核心驱动力，并强力支撑加配电气核心设备多头…", "summary": "本报告深入验证了GEV、ETN及HUBB的积压订单与扩产。GEV电气化订单积压超420亿美元，Eaton订单增长48%，数据中心订单增240%。大型电力变压器交期普遍在18-36个月以上，Tier 1高压变压器更超60个月。产能释放面临硅钢短缺及技术人工限制，最快2027年起效。变压器瓶颈带来了超…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 2, "neutral": 0, "value": 81.8, "events": [ { "title": "电网关键设备供应瓶颈：变压器与开关设备的交付压力", "summary": "本卡支持卡片02的核心结论——美国AI电气化的真正约束已从发电下移至电网设备（变压器、开关、接入）。从工业供给侧补充：该瓶颈正重塑GE Vernova与Eaton的订单簿与资本开支。GEV电气化设备订单储备升至约390亿美元（同比+75%），数据中心当季订单约24亿美元（超2025全年），并以53…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备短缺正在从并网瓶颈转化为头部OEM的订单可见度，但重型产能释放滞后意味着2026年仍是选择性兑现，而非全行业顺风。", "summary": "2026-06-20：本卡支持上一卡关于并网稀缺的判断。Eaton和Schneider的订单、积压和扩产计划验证了变压器与开关设备需求正在转化为OEM收入可见度；但IEA、ASCE和加拿大路线图显示大型变压器、开关设备和断路器交期仍可达数年，Eaton、Hitachi、Siemens Energy…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 3, "neutral": 0, "value": 77.2, "events": [ { "title": "电力可交付性将配给AI算力扩张的地点和节奏，但铝代铜只是放大器，不是全国性致命瓶颈。", "summary": "工作日为2026-06-21。本卡压力测试第04卡结论：AI算力扩张确实进入电力约束阶段，但“铝代铜”不是致命约束本身。真正瓶颈是大负荷接入、主变压器、开关设备、变电站档期、稳定电源和本地可靠性标准的组合。美国数据中心用电从2023年176 TWh可能升至2028年325-580 TWh，全球到2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球变压器交期严重延宕与国内并网排队拉长，导致金融资金的旋转速度与电网物理扩容出现严重错配，短期将制约算力交付，但具备出海溢价及本土低成本取向硅钢壁垒的中国设备龙头将获得超额套利利润。", "summary": "本报告对首席策略师提出的26%仓位旋转至电网设备进行物理压力测试。结论表明，全球变压器120-210周的交期瓶颈与国内主网18个月的并网延迟（Time-to-Power）将导致2026-2028年国内AI服务器部署分别出现7.50%、12.50%和9.83%的交付缺口，在高营业杠杆下压缩服务器制造…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件已到战术拥挤预警点，电力设备则是结构分化而非整体泡沫；制造出海主线应再平衡而非否定。", "summary": "截至2026-06-21，本卡由chief-strategist完成压力测试。AI硬件的成交、ETF与主力资金拥挤度已接近战术临界点，短期回撤风险上升，但机构超配比例仍低于历史峰值，不能据此全面退出。电力设备并非整体拥挤，风险主要在主题化交易载体和旧新能源β；电网、变压器等出海订单链仍有支撑。策略…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 5, "neutral": 0, "value": 66.6, "events": [ { "title": "电网设备紧缺与并网延迟正倒逼AI云厂商调整选址逻辑，通过工作负载地理分流、核电结构化直供、现场过桥发电以及电气设备提前大批量采购，来对冲并网延误风险，将AI资本开支推向重度能源与电力…", "summary": "截至2026-06-22，高压变压器交期拉长至24-48个月且排队并网超5年，成为AI集群部署核心瓶颈。超大规模云厂商正通过将训练与推理负载地理分流、签署核电结构化购电协议（PPA）、部署现场天然气发电机过桥以及提前3-4年锁定集团化设备产能等调整选址与容量规划，以对抗延迟风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "真正受益的是电力交付瓶颈和已签长期购电/费率保护资产，而不是泛化公用事业或所有新增兆瓦。", "summary": "截至2026-06-22，本卡综合上一张AI基础设施观点：数据中心用电增长仍成立，但公用事业受益必须按“执行风险由谁承担”重新排序。最优是电网设备和电力服务，其次是有已签负载承诺及最低计费/大负荷费率保护的受监管公用事业，再次是已签长期PPA的核电或稳定清洁电源。泛化merchant gas、水资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器/电网瓶颈对AI芯片需求预期与库存周期的传导风险", "summary": "截至2026-06-22，本卡以半导体视角压力测试Card-03。结论：芯片需求仍明确领先于电力瓶颈，2026年不存在AI芯片砍单——约束仍是芯片供给本身（CoWoS售罄至2026年底、英伟达独占约60%；Blackwell积压约360万颗售罄至年中；2026年HBM全部订满且超大规模厂商预付扩产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "更久维持高位、电力成本推动、AI资本支出搁浅时点与芯片库存尾部下的多资产组合定位", "summary": "本卡以asset-allocator视角综合四卡链条，落到一套定位框架。结论：四卡为「一个体制」，正确组合是杠铃——(1)做多物理瓶颈（电力/电网设备、燃气轮机OEM、独立电力/核电PPA），无论GPU是否按时通电都收多年在手订单年金，且做多令美联储紧缩的成本推动本身，是自我融资宏观对冲；(2)低…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 7, "neutral": 0, "value": 56, "events": [ { "title": "电力设备供应链扩产进度能否扭转2027年交期约束", "summary": "压力测试'扩产能救2027'的乐观读法：GE Vernova(全资Prolec GE，US$5.275bn)、Eaton(Jonesville US$340m+Nacogdoches US$100m+Waukesha US$22m)、Hitachi Energy(逾US$1bn含South Bos…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网队列与基础设施扩容压力评估", "summary": "本卡支持卡片1–2：约束已从Capex/GPU转向已通电的MW，并将瓶颈量化为双向并网堵塞。发电侧：美截至2025年底约2,060GW排队（约现有装机2倍），408GW已签约未通电，申请到商运中位时长翻倍至>4年。负荷侧：ERCOT大负荷队列达226GW（近4倍于2024年底63GW，77%为数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网输变电容量与并网审批周期（而非总体发电量）已成为AI算力部署的终极物理刚性瓶颈，并已实质性挤压IT硬件采购预算。", "summary": "本报告对AI算力扩张的电网与能源瓶颈进行压力测试。分析表明，虽然电力设备订单充足，但美国并网队列平均达5.5年，且面临FERC 2026年6月大型负荷监管及弗吉尼亚GS-5等防止成本转嫁政策。变压器等重电面临GOES原材料短缺，Tier-1厂商积压至2029-2030年。表后核电直供监管被封死，燃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI驱动的局部电力与电网设备通胀是持久且结构性的，但因其在PCE等通胀指数中权重过小而无法独立支撑美联储的利率路径决策，且高利率本身会作为负反馈自我修正出清这一供给瓶颈。", "summary": "截至2026-06-23，本卡对AI基础设施电力与设备通胀进行压力测试。变压器价格上涨60%-80%、交期延至154周，GE重型轮机订单达100GW且2028年前售罄，PJM容量电价暴涨11倍至$329.17/MW-day。但由于电力在PCE权重仅~1.3%，该局部供给冲击不足以独立支撑美联储加息…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 9, "neutral": 1, "value": 48.4, "events": [ { "title": "电网侧供应瓶颈对电力设备资本开支的驱动效应", "summary": "支持立场。变压器交期已拉长至128周（LPT）/144周（GSU），北美极端排期达60个月，使交期本身转化为多年期订单能见度：GE Vernova电气化在手订单同比升至424亿美元（覆盖约3年营收[自测算]），Siemens Energy集团在手1,540亿欧元，Eaton总订单+73%至超190…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器、开关设备、取向电工钢、电缆和电网设备制造产能", "summary": "支持卡片05/06：油价回落是数据中心电力运营成本的缓冲，而非解锁电网设备产能的钥匙。从制造现场看，资本已在大举投入——Hitachi Energy（60亿美元集团计划含约15亿专用于变压器、2027年放量）、Siemens Energy（4.21亿美元Charlotte厂2027年竣工）、GE …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "评估欧美电网升级需求与贸易壁垒对中国电力设备出海链的传导效应", "summary": "压力测试 Research note 4 命题：A股估值重配是否真建立在可持续高毛利出海订单上。结论为部分支持但有结构性保留。需求拉动真实且被低估——美国公用事业2026-30资本开支近1.3万亿美元 [S1]、超大规模厂商2026资本开支>6750亿美元/+60% [S2]、全球变压器短缺 [S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "发电侧与配电侧在2026年正陷入‘既无路（电网排队超4年及变压器稀缺）也无车（燃机交期5-7年及SMR未商业化）’的双端挤压困境，共址直供监管红线的收紧彻底消除捷径，导致算力资本开支…", "summary": "截至2026年6月25日，配电与电网设备交期拉长使得发电侧面临严重并网瓶颈。大型电力及GSU变压器交期达18-48个月，GOES硅钢供给偏紧，并网排队中位数超4年。FERC否决核电表后直供迫使亚马逊等转向表前PPA，面临同样电力瓶颈。同时高确定性燃气轮机交期达5-7年，SMR在2030年前难以放量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 3, "neutral": 1, "value": 85.8, "events": [ { "title": "电网设备原材料缺口对新能源项目建设成本与并网进度的传导效应评估", "summary": "支持卡片06–08的逻辑链：AI电力瓶颈已上移至原材料，且GOES/高纯度铜缺口确实传导至新能源经济性，但传导非对称、主要走时间而非单价。直接价格通道有限——铜仅占光伏/陆上风电资本开支约3%~5%，即便铜较2020年+50%、GOES+100%，对LCOE直接抬升仅约+2%~+4% [自测算]。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在限制性利率与结构性电力成本溢价下，市场对‘可交付性约束’的忽视将导致AI芯片链面临重定价，资金应向电网设备通道、独立发电商及高股息红利价值风格集中。", "summary": "截至2026-06-25，本卡合成2026年能源冲击与电力稀缺对大类资产及A股/全球股票轮动的影响。电力并网瓶颈与大变压器缺口推升期限溢价及美联储政策利率（更高更久），但实际r*仅温和上升+25至+50bp。资产配置应实施哑铃结构：超配电网设备与燃机OEM（受益于全球变压器缺口及交期溢价）、超配红…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球电力设备制造供应链产能与上游材料供应风险评估", "summary": "部分确认Card-03的time-to-power论点，但作三处修正。瓶颈确在上移到材料端，但对变压器而言真正的物理咽喉是GOES而非铜，且GOES稀缺本质是“产能+贸易”稀缺而非地质稀缺：全球变压器产能利用率98%、GOES自2020年涨约100%、美国仅Cleveland-Cliffs一家、俄…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "做多迁移的稀缺租金应以电网硬件和材料为锚、以先进封装和定制ASIC为弹性，同时用能源空头、商用GPU相对空头和暗硅期权保护组合。", "summary": "截至2026-06-25，本卡综合前序链条：霍尔木兹能源溢价正在消退，稀缺租金迁移到铜/GOES、电网设备、先进封装和定制ASIC。建议用100.0名义多头配置15.0%材料、30.0%电网硬件、20.0%先进封装、15.0%定制ASIC/AI网络、10.0%电力接入和10.0%流动性，同时用35…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 8, "neutral": 0, "value": 54.2, "events": [ { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网基础设施与能源转型对AI算力扩张的潜在约束", "summary": "压力测试部分支持并升级卡03：电力/电网瓶颈是“节流而非断路”，但节流比其暗示更紧更久。在2026–2028视野内，电气基础设施已是排在芯片之前的硬约束——美国电力变压器交期128周、GSU 144周、大单3–4年，价格较2019+77%、达2022前4–6倍；GOES由Cleveland-Cli…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股实施篮子的两融拥挤与流动性承载压力测试", "summary": "对第05张卡片的6–12pp轮动做压力测试并有条件支持：判断成立，但再拥挤风险在接收篮子内部不对称——电网设备最高、储能集成次之、受监管电力最低。2026-02-05电力设备曾在两融总额下降当天居融资净买入首位[S2]，且申万电力设备指数自2025年以来上涨51.14%[S4]，证明+4pp电网设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备出口毛利缓冲的汇率弹性与贸易救济风险压力测试", "summary": "本卡压力测试并有条件收窄 Card-06 的电网设备出口多头。汇率腿：人民币对美元/欧元升值 +3%/+5%/+7% 时，海外分部毛利按 ΔGM=c×x/(1−x) 分别压缩 思源约 2.0/3.4/4.8pp、金盘 2.3/3.9/5.6pp、南瑞 2.5/4.2/6.0pp[自测算]；越薄毛利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 7, "neutral": 0, "value": 61.6, "events": [ { "title": "电力设备扩产会缓解AI数据中心并网瓶颈，但无法在2026-2027年完全解除大型变压器和输电接入的物理约束，因此算力利润扩散仍是受瓶颈约束的电网资本开支周期。", "summary": "截至2026-06-28，本卡支持上一卡的物理瓶颈判断，但强调扩产的时间结构：变压器、开关柜和电网设备扩产是真实的，Eaton、Hitachi Energy、Siemens Energy等制造商已披露新增产能和高积压订单；但大型电力变压器仍受定制设计、GOES/铜、测试槽位、运输和并网研究约束，新…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "summary": "截至2026-06-27，本卡片综合前序结论：电网/变压器瓶颈会延后并改道AI部署，但不会摧毁2028年前光模块出口龙头业绩能见度；真正尾部风险是2027年云厂商ROIC压缩。建议A股AI物理瓶颈组合使用40/32/12/8/8结构：变压器/电网设备、光模块、电力现金流锚、铜/GOES等输入端对冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高利率平台与电力交付约束下的AI算力产业链估值重塑与多空配置策略", "summary": "我对卡片05–07主线部分支持、部分否定。约束在电力，经济租金从过剩层（资本、GPU、可出租算力）迁移到稀缺层（已通电兆瓦、发电、超长交期电网设备）——重估命题正确且不对称：稀缺层应享更高估值，被杠杆与抵押挤压的算力出租商应承受更宽折价。但干净的多空表达式更脆弱：多头腿GEV已71倍前瞻不便宜，便…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备供应链与并网建设周期正在成为AI扩张中capex无法快速解决的物理天花板，但该天花板表现为区域性、时间性和可绕行但昂贵的约束，而非全球绝对停摆。", "summary": "截至2026-06-27，本卡支持前序“电力可交付能力成为AI算力瓶颈”的判断，但把约束进一步定位为电网硬件与并网建设的自然时间：大型电力变压器最长四年、美国LPT常见36个月且最长60个月，输电许可平均约四年，配网变压器需求到2050年需达2021年160%-260%。Capex可以买优先级、预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 13, "neutral": 0, "value": 27, "events": [ { "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "summary": "本卡支持 Research note 02「上游原材料是最深约束」的结论，并锐化为一组非对称结构。GOES：自2019年变压器需求暴增（电力+119%、GSU+274%），远超GOES市场约5.4–5.6%的CAGR；GSU供给缺口2025年约100%，因美国唯一GOES生产商Cleveland-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力需求增长对电网侧扩容及配电资产的压力测试", "summary": "压力测试结论：前序主线被证实并强化。2026年AI落地的约束性瓶颈已非GPU，而是电网边缘——电力变压器、中高压开关设备、取向硅钢（GOES）与接网排队。大型变压器交期3–5年、全球产能利用率约98%、美国接网积压1,500–2,300GW、枢纽接入排队4–7年，致美国2026约12GW管线仅约4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电力’，并驱动策略从算力规模向算法…", "summary": "截至2026-06-28，本卡片通过对电网与变压器交期的压力测试，确认电网并网（5-7年交期）和变压器交付（2-4年交期）构成了AI算力扩张的终极物理边界。随着GB300（约135kW）及Rubin Kyber（约600kW）对800VDC和固态变压器（SST）的高密度配电需求增加，目前仅有不足8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 1, "risks": 5, "neutral": 0, "value": 84.4, "events": [ { "title": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "summary": "本卡片从制造和供应链视角对电网与电力约束进行压力测试。截至2026-06-29，关键电力设备（变压器与重型燃机）由于极高工艺与测试门槛、上游取向电工钢（GOES）瓶颈及寡头垄断，交期拉长至3-7年，生产机位已售罄至2029-2030年。云厂商BTM发电加剧了燃机机位争夺。预计2026-2027年该…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备出海高增长与国内“十五五”电网资本开支爆发锁定了龙头业绩的高能见度与利润确定性，是中证红利风格强化趋势下红利成长股的首选品种。", "summary": "本报告定量评估了电力设备龙头在“全球出海”与“国内十五五电网更新”驱动下的业绩确定性。全球变压器交期拉长至107.8周，高出口毛利率 premium 拥有调价条款保护。压力测试表明，即使北美订单延迟40%，主要龙头季度净利下修仅为-2.55%至-10.54%，自由现金流影响可控。国内超过5万亿元电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-9e97140446/", "networkHref": "/investor/research-network/#thesis-9e97140446" }, { "id": "thesis-ce29facfbd", "source": "tracked_thesis", "title": "中国资产定价仍受政策、流动性与风险偏好共同驱动", "titleEn": "China asset pricing remains driven by policy, liquidity, and risk appetite", "lane": "中国权益流动性", "laneEn": "China equity liquidity", "laneHref": "/investor/lanes/china-equity-liquidity/", "investableFocus": "中国权益流动性：510300.SH, 159915.SZ, 2800.HK, 3067.HK, 0700.HK, 9988.HK, PDD", "stockUniverse": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "A股核心指数代理", "roleEn": "A股核心指数代理", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "159915.SZ", "name": "创业板 ETF", "market": "A-share ETF", "role": "成长风险偏好", "roleEn": "成长风险偏好", "bucket": "core", "href": "/investor/stocks/159915.SZ/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "3067.HK", "name": "南方恒生科技 ETF", "market": "HK ETF", "role": "港股科技代理", "roleEn": "港股科技代理", "bucket": "core", "href": "/investor/stocks/3067.HK/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "PDD", "name": "PDD Holdings", "market": "US ADR", "role": "离岸消费互联网", "roleEn": "离岸消费互联网", "bucket": "core", "href": "/investor/stocks/PDD/" }, { "ticker": "QQQ", "name": "Invesco QQQ Trust", "market": "US ETF", "role": "纳斯达克成长暴露", "roleEn": "纳斯达克成长暴露", "bucket": "watch", "href": "/investor/stocks/QQQ/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "IBIT", "name": "iShares Bitcoin Trust", "market": "US ETF", "role": "BTC 现货代理", "roleEn": "BTC 现货代理", "bucket": "core", "href": "/investor/stocks/IBIT/" }, { "ticker": "COIN", "name": "Coinbase", "market": "US", "role": "加密交易基础设施", "roleEn": "加密交易基础设施", "bucket": "core", "href": "/investor/stocks/COIN/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "510300.SH", "159915.SZ", "2800.HK", "3067.HK", "0700.HK", "9988.HK", "PDD", "QQQ", "SMH", "IBIT", "COIN", "TSLA", "ASTS", "MU" ], "coreView": "赛道：中国权益流动性。候选观察标的：510300.SH, 159915.SZ, 2800.HK, 3067.HK, 0700.HK, 9988.HK, PDD。当前共有 40 条证据、4 位分析师和 7 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-04", "lastSeen": "2026-06-29", "supports": 5, "risks": 148, "neutral": 7, "analystCount": 12, "sourceTypeCounts": { "研究笔记": 159, "研究主题": 1 }, "topTerms": [ "电网", "GOES", "HBM", "GPU", "A股", "融资余额", "人民币", "电力", "权益", "asset", "pricing", "remains" ], "whatChanged": [ "2026-06-29: AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "2026-06-29: 电力成本不确定性会直接拖慢西部AI算力基地投产节奏，但主要通过容量/并网/备用费用和到电时间压低Capex效率，而不是通过单一电价水平否定项目。", "2026-06-29: AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利景气度，A股资产配置应向电网出海及水电红利哑铃端重构[自测算]。", "2026-06-29: 国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加速撤向水电红利及电网设备出海龙头，陡峭化了向高红利风格轮动的斜率，配置应重构为哑铃型…", "2026-06-29: AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。" ], "practical": { "sectorLane": "中国权益流动性", "sectorLaneEn": "China equity liquidity", "exposureBucket": "风险偏好与高波动", "tone": "red", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "A股核心指数代理", "roleEn": "A股核心指数代理", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "159915.SZ", "name": "创业板 ETF", "market": "A-share ETF", "role": "成长风险偏好", "roleEn": "成长风险偏好", "bucket": "core", "href": "/investor/stocks/159915.SZ/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "3067.HK", "name": "南方恒生科技 ETF", "market": "HK ETF", "role": "港股科技代理", "roleEn": "港股科技代理", "bucket": "core", "href": "/investor/stocks/3067.HK/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "PDD", "name": "PDD Holdings", "market": "US ADR", "role": "离岸消费互联网", "roleEn": "离岸消费互联网", "bucket": "core", "href": "/investor/stocks/PDD/" }, { "ticker": "QQQ", "name": "Invesco QQQ Trust", "market": "US ETF", "role": "纳斯达克成长暴露", "roleEn": "纳斯达克成长暴露", "bucket": "watch", "href": "/investor/stocks/QQQ/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "IBIT", "name": "iShares Bitcoin Trust", "market": "US ETF", "role": "BTC 现货代理", "roleEn": "BTC 现货代理", "bucket": "core", "href": "/investor/stocks/IBIT/" }, { "ticker": "COIN", "name": "Coinbase", "market": "US", "role": "加密交易基础设施", "roleEn": "加密交易基础设施", "bucket": "core", "href": "/investor/stocks/COIN/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "510300.SH", "159915.SZ", "2800.HK", "3067.HK", "0700.HK", "9988.HK", "PDD", "QQQ", "SMH", "IBIT", "COIN", "TSLA", "ASTS", "MU" ], "primaryStocks": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "A股核心指数代理", "roleEn": "A股核心指数代理", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "159915.SZ", "name": "创业板 ETF", "market": "A-share ETF", "role": "成长风险偏好", "roleEn": "成长风险偏好", "bucket": "core", "href": "/investor/stocks/159915.SZ/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "3067.HK", "name": "南方恒生科技 ETF", "market": "HK ETF", "role": "港股科技代理", "roleEn": "港股科技代理", "bucket": "core", "href": "/investor/stocks/3067.HK/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" } ], "instrumentTags": [ "510300.SH", "沪深300 ETF", "159915.SZ", "创业板 ETF", "2800.HK", "盈富基金", "3067.HK", "南方恒生科技 ETF", "0700.HK", "腾讯控股", "9988.HK", "阿里巴巴" ], "reviewCadence": "每日复核", "confirmationSignals": [ "期权伽马和 ETF 资金是否放大尾盘流动性", "社交热度是否转化为真实增量买盘", "波动率是否领先基本面恶化" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "波动率", "现金", "指数保护", "期权拥挤" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 160, "sourceDiversity": 2, "analystCount": 12, "recentEvents": 129, "supportRatio": 0.03, "riskRatio": 0.93, "conflictScore": 100, "evidenceDepth": 100 }, "score": 65.4 }, "events": [ { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34.75, "matchedTerms": [ "电网", "HBM", "GPU", "A股", "电力", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]，PPI-CPI剪刀差隔离了政策利率。任务设想的'成本→CP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "电网", "GOES", "人民币", "电力", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power预期缺口，会迫使Hyperscaler预付并网升级、锁定变压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26, "matchedTerms": [ "电网", "GPU", "电力", "ai-infrastructure", "power" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下，资金加…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性TAM上限，营收放缓促使估值压缩。相反，全球CPO光模块（中际…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电网", "HBM", "GPU", "A股" ], "href": "" }, { "date": "2026-06-29", "title": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利…", "summary": "本报告对AI算力激增带来的电力电网物理瓶颈（变压器交期120-210周[S2]、并网排队4-7年[S1]）对A股转型龙头的利润率传导进行压力测试。结果显示，电网扩容滞后充当了新能源装机的“限速阀”，压制了光伏/电池销量增速；同时，容量电价[S8]和碳成本上升[S19]增加了高能耗制造环节电费开支，而在严重过剩背景下成本转嫁失败，导致制造端转型龙头利润率遭到…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.75, "matchedTerms": [ "电网", "A股", "电力", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "电力成本不确定性会直接拖慢西部AI算力基地投产节奏，但主要通过容量/并网/备用费用和到电时间压低Capex效率，而不是通过单一电价水平否定项目。", "summary": "截至2026-06-30，本卡片支持卡片06的压力测试并进一步指出：电网费率重构和BTM不确定性会直接制约西部AI算力大基地，但核心约束不是单纯电价，而是“到电时间”、容量费用和接网责任。并网型BTM仍需缴纳输配电费、系统运行费、交叉补贴及基金附加，容量电价机制把可靠容量显性定价。战略性西部枢纽大概率继续推进，但边际民营/租赁型超大规模项目将被迫分期送电、…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.75, "matchedTerms": [ "电网", "GPU", "电力", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,800万美元（约降33%）[S1][自测算]；先进封装（CoWo…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.75, "matchedTerms": [ "电网", "HBM", "电力", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM合约价Q2环比+58–63%)、以及电网电力(美国并网排队约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.75, "matchedTerms": [ "电网", "HBM", "GPU", "电力", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心腿；半导体国产替代（¥2万亿AI建设计划+80%国产强制比例…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电网", "HBM", "A股", "电力" ], "href": "" }, { "date": "2026-06-28", "title": "在先进封装良率瓶颈与表后电力并网延迟的双重物理约束下，A股AI硬件板块估值极度偏高且存在变现拥挤，难以独立承担成长杠铃的对冲配置功能，应当实施防御性减配。", "summary": "本卡压力测试否认了AI硬件作为成长杠铃关键一环的论据。测试显示AI硬件期望收益为-3.43%，压力下变现滑点高达16.82%，且2nm GAA/HBM4封装成本激增113%抬高TCO门槛。此外，16GW计划容量中仅5GW在建，电力并网延迟拖累电网设备行业盈利收缩达36.8%至60.2%。建议将AI硬件配置从12%减至4%，规避流动性瓶颈。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电网", "HBM", "A股", "电力" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力需求增长对电网侧扩容及配电资产的压力测试", "summary": "压力测试结论：前序主线被证实并强化。2026年AI落地的约束性瓶颈已非GPU，而是电网边缘——电力变压器、中高压开关设备、取向硅钢（GOES）与接网排队。大型变压器交期3–5年、全球产能利用率约98%、美国接网积压1,500–2,300GW、枢纽接入排队4–7年，致美国2026约12GW管线仅约4GW在建（隐含约67%滑期，自测算）。这把AI算力部署拉长而…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电网", "GOES", "GPU", "电力" ], "href": "" }, { "date": "2026-06-28", "title": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电力’，并驱动策略从算力规模向算法效率和电…", "summary": "截至2026-06-28，本卡片通过对电网与变压器交期的压力测试，确认电网并网（5-7年交期）和变压器交付（2-4年交期）构成了AI算力扩张的终极物理边界。随着GB300（约135kW）及Rubin Kyber（约600kW）对800VDC和固态变压器（SST）的高密度配电需求增加，目前仅有不足8%的设施就绪。FERC否决核电表后直供，迫使所有新增容量重回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电网", "GOES", "A股", "电力" ], "href": "" }, { "date": "2026-06-28", "title": "AI数据中心电力瓶颈与核能PPA战略转型", "summary": "本报告分析了2026年AI数据中心面临的核心电网瓶颈，探讨了超大规模厂商绕过电网审批排队、锁定核能PPA（如微软与Constellation的Crane清洁能源中心项目）的战略选择。基于自主测算的200MW数据中心TCO模型，核能直供电相比传统电网虽有价格溢价，但通过提前2年并网运营，可规避高达50亿美元的GPU集群机会成本，在5年生命周期内创造45.52…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.75, "matchedTerms": [ "电网", "GPU", "电力", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "2026年全球AI数据中心电网瓶颈与“自主供电”战略的崛起", "summary": "2026年全球超大规模厂商的AI算力基建受限于电网瓶颈与关键电力设备（变压器交期达5年）短缺。这迫使厂商将部分资本支出（预计2026年总支出达7250亿美元）从纯算力（GPU）转向自主供电（BYOP）和表后（BTM）能源方案。预计该转移将延迟算力上线速度，并推高长期持有成本。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.75, "matchedTerms": [ "电网", "GPU", "电力", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "效率驱动的算力架构转型：从“暴力规模”到“高能效密度”的物理基座适配性评估", "summary": "压力测试 Research note 08 的隐含前提：效率并不缓解物理天花板，而是把它从电网层下移到机架与机柜行层并加剧。机理是机架尺度的杰文斯悖论——单瓦产出靠提高密度实现，机架功率从约40kW(Hopper)→120kW(Blackwell)→142kW(GB300,90%入液)→600kW(Kyber,2027)迈向1MW，对照2026均值约27k…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 24.75, "matchedTerms": [ "电网", "HBM", "GPU", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-27", "title": "高利率平台与电力交付约束下的AI算力产业链估值重塑与多空配置策略", "summary": "我对卡片05–07主线部分支持、部分否定。约束在电力，经济租金从过剩层（资本、GPU、可出租算力）迁移到稀缺层（已通电兆瓦、发电、超长交期电网设备）——重估命题正确且不对称：稀缺层应享更高估值，被杠杆与抵押挤压的算力出租商应承受更宽折价。但干净的多空表达式更脆弱：多头腿GEV已71倍前瞻不便宜，便宜的是AEP/VST/CEG（16–20倍）；空头腿（CRW…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "电网", "GPU", "A股", "电力", "权益" ], "href": "" } ], "timeline": [ { "date": "2026-05-04", "supports": 0, "risks": 2, "neutral": 0, "value": 30, "events": [ { "title": "未来12个月的最优配置不是全面押注国产化，而是超配最先把政策性信贷、算力建设和电力市场改革转化为订单与TCO优势的环节：液冷第一、光互联第二，电力市场化资产成为新增第二收益腿。", "summary": "本卡综合Research note 01-05，将国产AI算力链条转化为100单位A股/港股主题仓位：科技算力基础设施62单位、电力与公用事业赋能33单位、战术储备5单位。科技内部排序为液冷22、光互联16、国产AI半导体/HBM/设备12、先进封装5、CANN/云迁移7；电力内部为西部VRE I…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电子取代 HBM 成为 AI 链条上未来 12 个月最被讨论的瓶颈：1 GW 边际 AI 园区的电费在 2027 年从 $322M/年跳到 $926M/年，足以让 hyperscal…", "summary": "对 Research note 03 的 $1T capex / $7.5–10M/MW 营收密度救援叙事进行压力测试。结论：电力（不是 GPU）成为 2026 下半年起的硬约束；hyperscaler 在 2026–2027 锁定的边际 MW 全成本将达 $95–140/MWh，相对 2024 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "在 AI 能源利率冲击成为已知情景后，权益组合应从纯成长 AI 转为价值-质量-硬资产-精选 AI 基础设施框架，以承载五腿对冲并保留可验证的 AI alpha。", "summary": "本卡将第08卡联合无序尾部纳入基本配置，而非事后保险。建议全球权益风险预算按美股58、A股42配置，对冲后净权益beta为0.75-0.85，并以每季度65 bps NAV承载五腿结构性对冲。A股超配电网、电力、有色、AI基础设施和高股息SOE金融；美股超配AI硬件、电气设备、公用事业、能源和质量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 电力基础设施仍是结构性超级周期，但 2026 年应从拥挤的设备权益 beta 轮动到铜、GOES/高等级硅钢、合同保护型公用事业和变压器标准化链条，以对冲监管成本转嫁风险。", "summary": "本卡综合支持前序 AI 电力基础设施超级周期判断，但将组合表达从拥挤权益 beta 转向跨资产防御。2026 年核心风险是监管成本分摊：FERC/PJM 与 AEP Ohio 式大负荷电价把延期、取消和最低用电风险更多转给 hyperscalers。最佳配置是 30-35% 铜/GOES/高等级硅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "AI电网设备尚未见顶，但2026下半年必须从主题弹性切换到订单、材料、交付和利润率验证，A股配置应超配高质量电网设备龙头、低配低端beta。", "summary": "截至2026-05-13，本卡综合判断AI电网设备交易尚未出现决定性见顶信号，但已从广谱beta进入盈利验证期。2026下半年需重点观察三类信号：超大规模厂商是否维护或下修2027年AI基础设施资本开支，GOES和大型变压器交期是否明显缓解，上市OEM订单能否转化为收入和稳定利润率。A股配置上应维…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 2, "neutral": 1, "value": 37.2, "events": [ { "title": "DRAM拥挤是仓位信号而非卖出信号；下周应部分减SMH/宽基科技，转向小仓DRAM、重仓GRID、适中PAVE。", "summary": "截至2026-05-14，本卡支持下周将AI敞口从SMH/宽基科技部分切向DRAM+GRID/PAVE，但不追成满仓DRAM。DRAM的HBM/传统DRAM涨价逻辑真实，2026年传统DRAM ASP上修明显，但DRAM ETF上市后集中度和资金拥挤过高，应限仓12-15。GRID是更优新增主仓，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施组合应以全球GRID为主轴（42%），用PAVE（18%）和A股电网（22%）补位，把中国对发达市场出口敞口降为5%卫星，减SMH至5%、保持DRAM 8%战术卫星。", "summary": "截至2026-05-14，本卡综合卡01–卡04，给出AI基础设施组合的目标权重再配置：全球GRID从35%升至42%作为主轴，PAVE从15%升至18%，A股电网+电力设备出海从20%小幅升至22%，中国对发达市场出口敞口从12%降至5%卫星，SMH从10%减到5%，DRAM/HBM保持8%战术…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI算力增长正撞向“电力天花板”，半导体板块技术面极端过热且基本面面临“资本开支消化期”风险，支持减仓芯片转向电网。", "summary": "截至2026-05-14，本卡通过压力测试支持大幅减持SMH至5%并保持DRAM 8%战术卫星的决定。核心逻辑是AI算力瓶颈已从“硅片供应”转向“电力并网”，美国和欧洲电网并网等待时间已超5年，导致“暗壳”数据中心（有GPU无电力）现象频发。半导体周期已入成熟期，SOX指数偏离200日均线60%，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "工作空间物理修复的‘成功’是虚假的，更重要的是研究逻辑必须从‘芯片获取’转向‘能耗-算力-政策’的宏观三难困境。", "summary": "我们正式否认工作空间重构成功的论点，因为物理文件在2026年5月20日的验证中依然缺失。宏观层面，AI资本开支已从技术叙事转向‘算力基础设施化’政策驱动。当前核心瓶颈已从芯片获取转向电网容量与国产先进封装（CoWoS/HBM）良率。A股风格正在从模型端向电力基建与国产设备切换。建议后续深入研究国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 37, "events": [ { "title": "电力并网延迟与能源基础设施的物理与监管约束构成了制约AI硬件实物量增长的硬性瓶颈，导致算力链ROIC严重退化，从而加速资金在2026年下半年向高增长、即时变现的出海消费主线切换。", "summary": "截至2026年5月21日，对能源基础设施的压力测试显示，电网并网延迟（PJM/Dominion面临7-15年排队）与关键设备短缺（变压器交期拉长至128-160周）成为AI硬件实物量增长的物理约束，实际并网率仅达65%。未通电的“暗数据中心”使算力GPU前两年年化ROIC骤降至8%以下，将挤压高杠…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力基建的物理约束并非简单的股票风格轮动，而是引发大宗通胀、期限溢价抬升与股债相关性转正的系统性供给侧冲击，大类资产配置必须收缩久期并超配瓶颈商品与即时现金流出海Alpha。", "summary": "截至2026-05-21，对AI硬件并网物理瓶颈的压力测试显示，该约束构成系统性供给侧冲击，而非仅是权益风格轮动。闲置暗数据中心导致GPU ROIC骤降至8%以下，挤压高杠杆算力链估值。精炼铜面临33-45万吨结构赤字，价格突破$13,000/吨，GOES高企，商品成为核心物理约束期权。电网融资压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "配置重心应从“AI算力久期”向“全球电力缺口溢价”轮动，超配变压器出口与液冷。", "summary": "截至2026-05-23，本卡提出“AI-能源”双重瓶颈下的A股配置框架。随着HBM/CoWoS产能物理锁死与区域电网负荷达标，A股配置应从单纯的AI硬件转向“AI能源基础设施”。建议将资产分为：1. 瓶颈核心（HBM/封装，PE已压缩至25-35x）；2. 结构性赋能者（液冷，渗透率达76%）；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "A股权益leg应采取“红利低波（水核火电，40%）+中盘价值（电网设备，40%）+周期价值（铜铝硅钢，20%）”的结构，其中电网设备的成长风险由30Y国债完美正交对冲，而周期价值作为…", "summary": "2026-05-29：在债券leg降久期至5.5年、由30Y国债提供核心尾部对冲、绿债/ABS提供底仓carry的新结构下，A股权益leg获得承担更高增长与通胀风险的预算。本卡将权益暴露拆分为“哑铃+α”结构：(1) 红利低波（制度回报段，约40%）：以水电核电和容量电价火电替代被低配的7-10Y…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 1, "neutral": 0, "value": 55.2, "events": [ { "title": "2026-05-30的核心判断是：OEM仍是质量锚，但边际alpha在上游材料，存储提供业绩确认，BTM电力运营商提供战略稀缺和低beta稳定器。", "summary": "截至2026-05-30，本卡综合前六张卡：通电算力溢价仍成立，但新增股票风险不应继续单押已被充分发现的 OEM。建议“通电 AI 算力”股票子组合配置为上游材料35%、电力/冷却OEM 25%、存储与HBM 25%、BTM电力运营商15%。风格上，2026年中先以Quality承接已兑现的订单和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "配电侧变压器与开关柜已锁定2026-2028量价齐升超级周期，金盘科技、特变电工、思源电气、中国西电、平高电气是A股最高确定性的物理底座组合。", "summary": "截至2026-05-31，AI数据中心驱动的配电侧设备已进入结构性供应缺口与超级周期：美国大型电力变压器交付周期≈128周(标准)/144周(GSU)，高定制订单达4年；GE Vernova 2026Q1单季为数据中心新增订单$2.4B，超过上一财年全年；MV/LV开关柜lead time普遍14…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "AI 算力供给受限于 HBM4 售罄、CoWoS 封装良率及全球电网容量的物理极限，使硬件龙头成为 AI 价值链中无视‘Capex 断崖’风险的最终收费站。", "summary": "截至2026年5月，AI硬件供应链仍处于结构性紧缺。TSMC N2量产初期良率65-70%，HBM4于2月投产且美光2026年产能已售罄。NVIDIA Rubin架构推升单GPU功耗至2.3kW，使电力供应成为比芯片更硬的瓶颈。尽管终端应用出现DAU增速放缓，但硬件端的实物稀缺性构成了核心护城河，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "P/Watt时代AI算力与电力电网设备的实物瓶颈与地理套利", "summary": "P/Watt时代AI算力与电力电网设备的实物瓶颈与地理套利 AI Power Grid Bottlenecks and Geographic Arbitrage in the P/Watt Era 随着AI算力扩建撞上电力电网并网物理速度墙，变压器与硅钢（GOES）结构性缺口高达30%，投产延迟W…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力天花板的核心痛点是并网延迟和规制规费引发的硬件折旧搁置与清洁电力溢价，这将在中期使云GPU租赁提价15%-35%[自测算]，并向SaaS客户传导3.0%-4.5%的毛利拖累[…", "summary": "截至2026-06-02，本卡对AI用电瓶颈做深度测试。电网并网设备延期（128-144周）引发的‘芯片折旧挤压’是最大隐形电力成本，1年延迟使GPU活跃小时折旧成本增50%[自测算]；加之PJM 2026年收取$4-$8/MWh合约规费，电力成本上涨导致云GPU租赁提价15%-35%[自测算]，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "AI电力组合应采取70/20/10（权益/商品/现金）的配置比例，超配短周期物理层，利用铀与稀土对冲长周期供应链断裂风险，并依据明确的政策触发器执行动态调仓。", "summary": "截至2026-06-02，本卡综合卡片01-05的分析，将AI电力杠铃结构转化为可执行的配置框架。在10%-12%的AI基础设施总仓位中，建议配置：权益(70%)、商品对冲(20%)与现金(10%)。权益端以短周期物理层(CPO/液冷)为核心底仓，长周期核电/电网受政策β影响被降级为辅助仓位，并偏…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "HBM/ASIC的结构性支撑在6-18个月维度上成立，但被卡片01低估了'客户集中度'和'瓶颈从硅迁移到电力'两条脆弱性——结构核心需要内部再平衡（提电力/冷却、减HBM股权、AVG…", "summary": "锚定2026-06-03（周三），AVGO FQ2业绩大概率周四盘后披露、NFP周五早间公布。我整体认同卡片01的'结构有支撑、战术有脆弱'综合判断，但TMT视角下结构性sleeve存在三条具体脆弱性需要再定价：(A) HBM需求高度集中在NVDA单一客户（约60-70%比特占比），属于穿着结构外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 0, "value": 48.2, "events": [ { "title": "模型能效与边缘计算是面对一阶物理瓶颈（变压器/电网/排队）的二阶效应，叠加推理token爆炸与已锁定capex，'通电滞后'对AI应用端的重定价逻辑在12–18个月内不会被瓦解。", "summary": "TMT否定'能效+边缘'对冲电网瓶颈的反驳。Blackwell单token能效25×提升被推理调用量9× YoY与o系列推理链5,000–50,000 token的爆炸增长抵消，IEA预计数据中心电力需求2024–2030翻倍至945 TWh。瓶颈在LPT/GOES硅钢/并网排队（104–127周…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理瓶颈栈的供给弹性差异决定了租金在中长期将不可逆地从半导体封装/HBM顶端向下沉淀至电网、电力组件与上游资源所有者，资产配置应超配电网基建与铜资源，低配久期债与下游AI应用。", "summary": "本卡综合了卡1-5的四层AI物理瓶颈栈分析，探讨了2026-2028年大类资产与A股配置策略。由于底层公用事业与铜精矿供给弹性极低，租金将在中长期下沉至电网与资源端。建议全球超配铜矿山、独立发电商与电网设备组件，低配久期债与AI SaaS应用。A股重点配置十五五电网Capex扩张受益者及上游自主铜…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "AI基础设施篮子应从芯片稀缺溢价，转向算力核心、电力设备收入可见度和GOES/铜材料对冲的三层配置。", "summary": "截至2026-06-05，我支持AI基础设施瓶颈从芯片向电力设备、再向GOES/铜迁移的主线，但建议把组合从芯片主导转为“算力核心+设备超配+材料对冲”。建议权重为30%算力/HBM/先进封装，28%电力设备/冷却/微电网，17%铜与GOES等材料，12%合约电力/公用事业，8% API-firs…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "北美电力并网实物瓶颈侵蚀推理毛利率并加速推理定制ASIC替代，而国内TMT硬件两融去杠杆将对软件核心个股恒生电子和金山办公施加被动流动性压力，盯紧CMR警戒线与6月下旬公募回补窗口。", "summary": "截至2026-06-08，受北美电网并网申请平均排队53个月及PJM容量价格触底限制，云巨头面临$180-290bn的Capex下修或延迟压力，度电成本暴涨94.6%至$80.44/MWh侵蚀推理毛利，倒逼定制推理ASIC占比在2028年升至45%-55%。与此同时，A股电子板块5665.69亿融…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "AI算力建设正从“依赖电网”转向“离网自治”，SOFC与中国电力设备出口链通过解决交付周期错配，成为捕捉AI基建溢价的核心抓手。", "summary": "2026年AI算力扩张的瓶颈已从GPU转向电力并网，使得SOFC与变压器成为核心资产。SOFC通过“时间套利”（交付周期<12个月对比电网并网7-10年）为AI厂商节省数十亿美元机会成本，LCOE在$58-110/MWh区间已具竞争力。中国厂商凭借全球60%的变压器产能及对取向电工钢（GOES）的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 53.8, "events": [ { "title": "AI基建瓶颈已实质性从芯片转移到电力与电网，且这一约束在12个月维度上无改善迹象，支持继续加配变压器、燃机、EPC与表后发电资产。", "summary": "美国超大规模厂商AI基建的边际约束已从GPU分配迁移到并网排队和一次电力：LBNL显示在排接入容量>2.6 TW、中位等待约5年；PJM/Dominion大负荷新申请要排4–7年；大型变压器交付从50周拉长到120–210周；GE Vernova重型燃机2028年前售罄。同期H100/B200交付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建领先指标从GPU切向电网与施工瓶颈", "summary": "今日VRT、ETN、PWR随AI板块大跌是拥挤交易/贴现率出清，而非订单周期见顶。诊断关键是背离方向：股价因利率上行与获利了结而跌，而领先指标（Revelio施工岗位+430%、核心岗位+175%仍在攀升）与同步订单簿（ETN在手订单+48%、数据中心订单+240%；PWR在手订单485亿创纪录；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 0, "value": 43.2, "events": [ { "title": "为支撑三层算力织物，AI芯片与先进封装如何演进以匹配电力波动与边缘推理的功率包络", "summary": "综合并支持卡片04的三层算力织物，下沉到硅基与封装层。关键洞见：'电力波动友好'主要是电力电子＋调度固件问题，而非新晶圆问题。NVIDIA GB300靠电源货架（65 J/GPU储能＋功率封顶斜率＋功率烧除）把Megatron训练电网峰值砍约30% [S1]，DVFS早是标配 [S2]；同步功率摆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科技巨头直接注资电力资产因电网设备供应链瓶颈（变压器交期48-60个月）而演变成“资本陷阱”，导致高达40%的预承诺GPU硬件（$100亿）因并网滞后3年而遭遇65.7%（-$65.…", "summary": "截至2026-06-13，本报告对科技巨头直接注资电力资产的资本转移进行了压力测试。由于高压变压器（LPT）交期延长至48-60个月、取向硅钢（GOES）供应受限以及并网队列长达5年以上，资本转移正退化为“资本陷阱”。定量测试表明，在3年并网延迟与40%电网造价通胀下，一个2 GW的AI数据中心项…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施从'算力中心'转向'能源中心'", "summary": "电力稀缺真实存在，约束正从GPU迁移到稳定廉价可快速接入的电力，但这不会把'否决权'交给传统能源巨头。租金归于能拼齐'发电+电网接入+资本+选址'的一体化建设方与设备链，而非被动存量生产者：云厂商正反向一体化签核电offtake（13个项目>9.8GW），杰文斯效应让需求分散到更多更小部署，且电力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的扩展性：电力与芯片供应瓶颈对超大规模厂商CAPEX的影响分析", "summary": "截至2026-06-12，本卡压力测试卡片02（电力瓶颈）与卡片06（半导体多线挤压）。结论：双重约束不会削减超大规模厂商资本开支，而是将其重排。2026资本开支仍约$725-726B（Dell'Oro），资本开支/营收比约25-30%、指引2027约35%；变化在优先级——可靠电力升至栈顶，建设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 6, "neutral": 1, "value": 35.6, "events": [ { "title": "电力接入瓶颈下A股AI基础设施链条的再定价", "summary": "综合卡06–08的确定性序列：电力接入瓶颈而非GPU叙事才是AI落地的真正约束，最确定的A股盈利在电网设备。建议把AI电力篮子向确定性与供给紧张倾斜：超配特高压/主变(33%)与构网储能(18%)，以上游铜铝(24%)作宏观再通胀Beta与需求底座保持核心仓，低配AI服务器(13%)，并严格按是否…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GB200 级效率提升改善 AI factory 经济性，但 2026-2028 年真正限制部署节奏的是站点电力、并网、变压器、稳定电源和电网运行能力，而不是 CoWoS/HBM 单…", "summary": "2026-06-13：本卡对 NVIDIA 护城河做电力压力测试。GB200 NVL72 单机柜约 120 kW，1 GW IT 功率约支持 8,333 个机柜、约 600,000 颗 GPU。IEA、DOE/LBNL 与 NERC 显示，数据中心用电、并网队列、变压器和燃机交期都在拉长，大型计算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力中心的电力激增与电网并网瓶颈，为电力板块提供了超越传统周期的长协盈利溢价，强力支持将红利配置中电力权重上调至2.5%以对冲地产下行。", "summary": "截至2026-06-13，AI基础设施分析师综合验证了上调电力板块权重的逻辑。IEA《Electricity 2026》显示，AI推理带来常态化电力负荷，全球数据中心耗电将在2030年翻倍逼近950 TWh。国内十万卡GPU集群建设使并网与变压器成为核心瓶颈。电力企业受燃料成本下降与AI电荒预期双…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI配电订单的高毛利和强价格传导能力，能有效对冲特高压业务的铜价挤压，利好具备IDC配电敞口的电力设备龙头。", "summary": "AI算力中心对高标准配电的需求能有效对冲铜价对传统特高压设备的利润侵蚀，但存在显著的结构分化。云厂商为保障昂贵GPU集群按期部署，愿为配电设备支付高毛利并接受铜价传导。A股具备“特高压+AI数据中心配电”双敞口的龙头将展现强利润韧性。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 12, "neutral": 0, "value": 12, "events": [ { "title": "2026年AI算力扩张引发的能源瓶颈：从散热需求转向电网容量与能源供应保障。", "summary": "在'算电协同'框架下，2026年AI算力扩张的硬约束已从GPU与散热上移至中高压输配电层。北美100 MVA以上大型电力变压器交付周期已达120–156周（2019年约50周），中压开关柜52–80周；GOES上游产能仅约10家合格供应商，扩产需3–4年。寡头（Hitachi Energy、Sie…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国电力设备出口正凭借‘速度溢价’成为全球AI基础设施的物理基石，并为人民币提供了结构性的宏观对冲价值。", "summary": "2026年全球AI电网约束下，中国电力设备（变压器、特高压、储能）正从‘成本竞争’转向‘速度竞争’，其12-18个月的交期远优于西方的5年 backlog。中国通过控制取向电工钢（GOES）和电力硬件供应链，获得了显著的全球‘速度溢价’（出口均价上涨33%）。这不仅推升了2025年创纪录的1.19…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI先进封装与HBM产能瓶颈对算力基础设施扩张及超大规模厂商资本开支的影响", "summary": "支持半导体分析师Research note 01的论点并向下游延伸：CoWoS/HBM确为算力的关键约束，2026年部署受封装与内存所限（CoWoS扩至约13万片/月仍订满；SK Hynix与Micron 2026年HBM售罄；HBM4验证延迟使NVIDIA将Rubin目标由200万下调至150万…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网容量及绿电约束倒逼算力资本开支分流并延后，建议下调A股“科技进攻端”并超配“实物红利端”（区域火电与电网设备），并设立以“电费上浮”与“并网延期”为核心的双轨尾部风险触发器。", "summary": "截至2026-06-14，鉴于一线城市电网并网物理瓶颈（排队长达24-36个月）与绿电指标约束，导致下半年算力资本开支落地延后且向电力配套分流（占比自15%升至25%），我们建议下调A股“科技进攻端”权重（自30%降至26%，出清高估值两融拥挤的GPU租赁SPV，转向先进封装和受推理通缩驱动的AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 0, "value": 59.4, "events": [ { "title": "墨西哥ATP电力瓶颈后的A股/STAR硬科技配置：证据链供应商与电力韧性资产的筛选", "summary": "本卡将前9张卡片综合为A股/STAR配置图。墨西哥电网（91%工业园区曾停电、5年缺口48,000 MWh、CFE 2025投资-24%）证明ATP孤岛必须自建电力韧性，这是一笔'卖铲人'交易而非墨西哥交易。增配五桶：电能治理（思源002028、盛弘300693、新风光688663）、构网储能PC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据中心的物理功耗墙迫使超大厂将有限的电力配额向具有3-5倍能效优势的ASIC倾斜，实质性加速了算力硬件市场从“通用高溢价”向“定制化能效套利”的重塑。", "summary": "2026年，电力配给成为AI基础设施的核心硬约束。四大超大厂高达6300亿美元的资本开支正大规模向功耗更低、每瓦特产出更高的定制化ASIC倾斜。通用GPU飙升至2300W的单卡功耗和超120kW的机柜密度让电网不堪重负，促使ASIC以近45%的增速抢占推理市场，从而实质性削弱了英伟达的算力溢价。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 1, "risks": 12, "neutral": 2, "value": 14.8, "events": [ { "title": "通胀和高利率提高了AI硬件基础设施的质量门槛，但没有削弱其配置价值；应买入瓶颈供应商而非泛AI叙事。", "summary": "截至2026-06-16，本卡支持“硬科技配置中保留AI基础设施”的上一轮判断，但建议从广义AI beta收窄到有订单、定价权和瓶颈属性的硬件基础设施。高利率会压缩投机型数据中心开发商和远期AI应用估值，却未削弱GPU、定制AI芯片、HBM、配电、冷却、变压器、开关设备和已锁定电力的数据中心容量的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 08的杠铃方向成立，但中国权益实施必须从“宽泛受益”改为“成本转嫁能力、出口政策韧性和人民币压力三重筛选后的选择性超配”。", "summary": "截至2026-06-16，本卡对Research note 08的全球与A股杠铃配置进行中国宏观压力测试。结论是方向可保留，但中国不是美国滞胀的纯受益方：5月CPI仅+1.2%、PPI达+3.9%，2.7个百分点剪刀差显示外部冲击首先压缩下游利润。人民币强势和1.75%的10年期国债收益率给人民银…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力的“快时钟”正与电网基础设施的“慢时钟”发生剧烈碰撞，变压器产能缺口将迫使数据中心建设周期长期化并推高资本开支，形成实质性的硬约束。", "summary": "截至2026年6月16日，变压器与电网设备供应链已成为AI算力扩张的实质性物理瓶颈。大型电力变压器（LPT）交期延长至36-60个月，核心原材料取向电工钢（GOES）供应高度集中且价格较2020年上涨50-80%，导致变压器综合成本上涨60-80%。约30-50%的2026年计划产能面临延迟，建设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "吉文斯悖论与推理端计算膨胀导致算力及电力净需求呈指数级爆发，彻底推翻了电力基建搁置假说，反而迫使云厂商将资本开支结构向配套电网及液冷刚性倾斜，对高速光模块（CPO）等算力硬件产生相对…", "summary": "截至2026-06-17，大模型效率提升（如DeepSeek V4 API价格降至$0.15/百万Token）将触发吉文斯悖论与推理端算力膨胀。测算显示，即便MoE使单Token算力开支减少70.0%，但由于API调用量激增35倍且推理端Token密度升至877.5个，AI算力总需求仍暴增61.4…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 4, "neutral": 0, "value": 58.2, "events": [ { "title": "白皮书主线成立，但因文件归档缺失和后段产能/价差模型证据不足，2026-06-18 不具备最终发布条件。", "summary": "截至2026-06-18，本终审 QA 对 9 张卡片给出“需要修订”评级。主线从 AIDD 算力需求到 GPU/HBM、机柜功率密度、电网设备稀缺、GOES 套利和西方 OEM 产能错配整体连贯，电力与电网瓶颈证据最强。但实时工作区仅存在 card-02、card-04、card-06、card…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FOMC 后应以“选择性算力基础设施挤空”而非“无差别 AI 融涨”来交易：多稀缺、空低 ROIC，净多但对冲。", "summary": "2026-06-18 策略综合：前序链条成立，2027 风险是 ROIC 与结构分化而非 capex 悬崖。FOMC 后应做多稀缺算力资产而非泛 AI beta：前沿逻辑/设备/HBM/先进封装、自研 ASIC 与网络、电力与电网受益方应超配；商用 GPU 龙头保留但严控风险；低 ROIC 商用 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI主题内部的定价权正从芯片端的硅片稀缺向电力、并网熟地和变压器等物理基础设施供应链迁移，电网并网延迟构成了对云巨头闲置算力资本的巨额惩罚，迫使价值向能源与硬件节点转移。", "summary": "截至2026年6月18日，由于全美并网队列激增至2,600 GW且平均等待时间达5年，AI基础设施的核心瓶颈已从GPU芯片稀缺转移到电力、并网及电力设备。定价权正向控制熟地、独立发电和电网设备的物理环节迁移，以规避巨大的闲置算力沉没资金惩罚（10万张B200芯片闲置年资本占用成本达1.8亿美元）。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AIDD是AI基础设施需求的高质量增量因子：规模小于通用大模型，但持续性、数据壁垒和监管/IP属性更强。", "summary": "截至2026-06-18，本卡片支持上一张卡片的方向：AIDD不会单独决定hyperscaler capex，但会把高价值、持续性的生命科学工作负载接入GPU、网络、存储、液冷和电力保障型数据中心。Recursion BioHive-2、DCAI Gefion、Novo Nordisk、Isomo…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 5, "neutral": 0, "value": 45.6, "events": [ { "title": "电力/电网瓶颈真实且已在容量价格中显形,但2027年前仅推迟部分美国新增集群、不封顶全球GPU部署,Research note 01的BOM组合在2026至2027 H1仍成立,20…", "summary": "AI算力电力瓶颈真实存在但属于'延后+置换'而非'封顶':PJM 2026/27容量价已跳11倍至$329.17/MW-日,2027/28预测缺口6.6 GW,新接入等待40个月;ERCOT 2026 Q1单季198 GW申请,累计大负荷队列约410 GW;全球电网接入等待5-10年,变压器交期1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM3良率瓶颈将AI基础设施风险从总量需求崩塌改写为本土投产节奏后移，并强化A股高融资TMT的收入兑现左尾风险。", "summary": "截至2026-06-18，本卡压力测试卡片06：国产HBM3良率瓶颈会推迟中国本土AI算力变现，但尚不足以证明全球AI基础设施从快速扩张转入需求观望。全球hyperscaler capex仍有Meta、Alphabet、Microsoft等高额指引支撑；真正变化在中国本土项目从按园区容量扩张转向按…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "InP/EML是2026年AI光互联的分配和毛利瓶颈，而非全球带宽扩展的绝对天花板；A股相关标的仍应按稀缺期权而非确定性订单定价。", "summary": "截至2026-06-18，本卡从AI基础设施角度压力测试InP/EML瓶颈：全球层面，InP/EML短缺确实影响800G/1.6T光模块分配和价格，但不构成2026年下半年全球AI集群带宽扩展的硬上限；LightCounting称2026年4月需求高于供给30%，但预计短缺在2026年底缓解。中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供应链瓶颈对AI算力基础设施（GPU集群）部署进度的影响", "summary": "截至2026-06-18，本卡压力测试Research note 03的BTM结论。BTM确实绕过并网时钟：约18–24个月对比电网36–60个月，故已公布约90 GW、约2 GW投运、燃气支撑约75%[S1][S2][S3]。但约束不是经济性——AI每兆瓦年收入1000–1200万美元，100美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 7, "neutral": 0, "value": 35, "events": [ { "title": "A-share de-leveraging to 1550 points triggers secondary market liquidity locking and margin…", "summary": "若市场下探至1550点，中际旭创股价将跌至714.16元，击穿融资预警与平仓线，带来日均31.2亿元平仓抛压（占ADV 17.36%），锁定二级流动性。但公司持有125亿现金及45亿海外客户预付款，可提供151.5亿净营运资金，供应链运营违约风险极低。海外订单真正风险在于电网并网排队达36-60个…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "16.18%的硬件Capex下修会压低AI基础设施整体估值，但订单传导不是线性的：液冷最敏感，电网设备和BTM发电更受长期电力容量瓶颈支撑。", "summary": "截至2026-06-19，本卡把H2 AI硬件Capex下修16.18%作为未外部确认的压力测试输入。结论是分层传导：GPU/服务器采购最先受冲击；液冷因绑定高密度机柜节奏，订单和估值敏感度较高；电网设备因并网稀缺、变压器长交期和backlog，订单不应按同等比例下修；BTM自备发电估值会先压缩，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而压制高估值芯片巨头的…", "summary": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网稀缺已经从主题交易升级为大类资产配置因子：低配无法按时通电的GPU/CPO硬件需求，超配持有变压器、GOES、中国UHV和红利电力现金流的资产。", "summary": "截至2026-06-19，本卡片把前序“GPU/CPO减配、电网链上修、GOES/变压器瓶颈、美国可再生递延、中国UHV免疫”整合成100单位跨资产组合：18%中国UHV/电网自动化、14%一体化变压器与GOES、16%红利电力、6%铜、18%中国久期、7%黄金、9%现金、6%硬件beta、2%美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 6, "neutral": 0, "value": 42.4, "events": [ { "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "summary": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然HBM国产化良率崩塌会将集群 CapEx 推升 98%，但中国互联网巨头通过异构分级部署与算法优化（缩减30%GPU需求）并结合西部绿电直连套利（LCOE降至0.39元）与液冷改…", "summary": "评估HBM成本压力对AI集群TCO及扩容策略的传导。100%国产材料导致HBM3封装良率崩塌至4.11%，推升10000颗GPU集群CapEx至8.17亿美元，东部煤电下3年TCO达10.17亿美元（+66.13%）。建议协同优化：架构上采取60% HBM3与40% HBM2E的分级部署，结合算法…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler AI扩张面临的是“资本买不到时间”的物理瓶颈：有资产负债表不等于能压缩电网、变压器、许可和矿产供应周期。", "summary": "截至2026-06-21，本卡片对AI基础设施扩张做压力测试：Hyperscaler的主要风险不是单纯GPU短缺，而是可靠电力、PJM类大负荷接入、变压器/输配电设备、关键矿产精炼集中度和能源成本共同造成的物理交付瓶颈。IEA预计全球数据中心用电从2024年415 TWh升至2030年945 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本压力对AI算力基础设施运营经济性与扩容节奏的影响", "summary": "压力测试并在核心机制上否定使命链条（油价×汇率→工业电价→AI经济性→扩容放缓）。链条两处断裂：(1)选错杠杆——电力是AI算力经济性的二阶变量，1GW数据中心年化TCO约85亿美元中能源仅约6亿（≈7%），服务器约60%；电价+20%仅抬升总算力成本约1.4%，不构成扩容开关，资本开支（GPU）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 9, "neutral": 0, "value": 25.8, "events": [ { "title": "CNH压力不会杀死AI基础设施资本开支周期，而会把价值捕获从融资敏感的AI Beta转向电力锁定、并网稀缺和强资产负债表支撑的硬基础设施。", "summary": "截至2026-06-22，本卡综合CNH流动性压力与AI基础设施周期：支持前序“CNH挤压—远期倒挂—即期购汇—境内流动性抽水”的传导链，但否定把AI基础设施一刀切视为高Beta受害者。AI资本开支会分化：电力、电网接入、变压器、冷却和强资产负债表hyperscaler受益于物理稀缺；人民币融资G…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-21，电网与电力并网瓶颈带来的通电延迟（2年延迟拖累53.71亿美元[自测算]）将迫使云厂商放缓GPU采购，若发生20%的需求削减，台积电3nm/2nm制程及CoW…", "summary": "截至2026-06-21，本分析对‘电力与电网并网瓶颈’对高端AI芯片（GPU、HBM）及先进制程产能的传导影响进行了压力测试。定量测算表明，10万张Blackwell GPU集群面临2年通电延迟将产生53.71亿美元闲置财务拖累[自测算]。如果云厂商因电力限制削减20%芯片采购，台积电3nm和2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施建设中的电力供应与电网容量瓶颈分析", "summary": "本卡压力测试 Research note 02 的隐含假设「资本开支≈算力」。证据显示约束已从芯片迁移到电力：微软披露约800亿美元 Azure 订单因电力受限无法交付，Nadella 称芯片堆库存却无「暖壳」可插[S5][S4]。转化在三处漏损：(1)并网排队约2,600 GW、中位等待约5年、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力瓶颈从'量'转向'质'及空间错配，如何通过能源成本与电网接入约束，重塑AI数据中心选址逻辑与Hyperscaler的Capex结构？", "summary": "截至2026-06-21，本卡压力测试卡片05的'瓶颈从量转向质'论点。我支持其诊断，但反对'只要把Capex转向能源问题就解决'的隐含假设。2026年Hyperscaler Capex约6000-7260亿美元、约75%为AI专项；1GW园区GPU/服务器占TCO约60%，电力基础设施占设施成本…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 9, "neutral": 1, "value": 30.2, "events": [ { "title": "电力瓶颈本地地点化和能耗惩罚电价将加速粗放GPU堆叠的价值减损，重塑A股科技风格超配“应用红利”与“电网CapEx”，并转向以“能效比”为核心的ASIC高能效硬件。", "summary": "截至2026-06-22，本卡综合研判AI电力瓶颈地点特定约束对权益风格的重塑。随着“算电协同”写入政府工作报告及发改委688号直连政策落地，东部智算面临并网与PUE差别电价硬约束（北京PUE>1.35加价0.20元）。自测算表明，不达标能耗与并网延迟将对100MW智算中心分别造成17.05亿和1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与定制化ASIC的深度整合，是将大模型访存能耗降低95.7%并实现3-5倍能效提升的核心硬件路径，亦是国产半导体链规避先进制程封锁、支撑算力电力约束的战略根基。", "summary": "截至2026-06-23，本卡研判支持通过先进封装与ASIC能效路径应对算力电力约束。2.5D封装结合HBM3E使访存功耗降低95.7%（从35 pJ/bit降至1.5 pJ/bit）；自研定制ASIC能能效比高出GPU约3-5倍，万卡集群年省电费2,943万元人民币，并释放48MW电网容量。国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股应把全球体制表达为「电网优先、算力精选、久期受控」的本土杠铃，而不是无差别追逐AI长久期。", "summary": "截至2026-06-22，本卡把全球「更久维持高位+AI电力/芯片」体制翻译为A股本土杠铃：超配电网/HVDC/电力电子和出口变压器，精选但严格控制估值与仓位的国产AI算力，低配泛长久期概念与纯利率敏感贝塔。核心依据是国家电网2026-2030年RMB 4万亿元资本开支、国内电力负荷增长、UHV技…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "国内通缩+自主可控体制是美国更久维持高位的镜像，使A股本土杠铃被双重逻辑支撑而不破；风险在于宽松与升值的节奏，而非方向。", "summary": "本卡以宏观视角压力测试Card-06的A股本土杠铃。结论：杠铃不破，但成立理由是「镜像」而非「复制」——中国处于通缩+再通胀+自主可控体制（CPI同比+1.2%、PPI同比+3.9%、LPR连续12个月维持1年3.00%/5年3.50%且有降息空间、人民币升值至约6.76–6.79），与美国更久维…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 8, "neutral": 0, "value": 34.6, "events": [ { "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "summary": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网队列与基础设施扩容压力评估", "summary": "本卡支持卡片1–2：约束已从Capex/GPU转向已通电的MW，并将瓶颈量化为双向并网堵塞。发电侧：美截至2025年底约2,060GW排队（约现有装机2倍），408GW已签约未通电，申请到商运中位时长翻倍至>4年。负荷侧：ERCOT大负荷队列达226GW（近4倍于2024年底63GW，77%为数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建配置应从“追逐算力设备增量”切换为“持有可交付电力稀缺性”，但保留CPO核心仓位，因为CPO仍是降低AI网络功耗的必要技术。", "summary": "截至2026-06-24，本卡综合前序结论：AI算力瓶颈已从GPU转向已通电MW，组合应从高拥挤CPO单边beta转向“变压器OEM+中国出海设备+GOES/铜材料+CPO核心赢家”的哑铃。100单位AI基建权益袖基准配置为韩日及全球变压器OEM 30、中国出海设备22、上游材料18、CPO/光模…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "配电网滞后与变压器短缺连同HBM/CoWoS确实构成AI capex的物理硬约束——对边际开发商是硬墙，对优先级买家是ROIC压缩，从物理层面支持AI-capex去杠杆而非2026崩…", "summary": "截至2026-06-23，本卡从公用事业侧确认：配电网扩容滞后与变压器短缺是AI capex物理钳制的第二片钳口，与HBM/CoWoS硅片分配叠加。变压器交期达128–144周乃至4年，电力变压器缺口30%、配电10%，GOES自2020年翻倍、铜+50%；LBNL并网队列约2,060 GW、中位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 6, "neutral": 1, "value": 59, "events": [ { "title": "科技板块风格再校准：能效赋能方的稀缺溢价与高耗电算力股的吞吐折价", "summary": "本卡把前五卡收束为可执行配置框架。算力以‘每兆瓦有效算力’计价后，权益市场沿‘谁拥有稀缺投入、谁的回报被它锁死’重估。折价已现于现金流：2026超大规模厂商资本开支约6900–7250亿美元、约75%投向AI，却吞噬近100%经营性现金流；微软约800亿美元Azure积压受限于电力。溢价已现于盘面…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "评估欧美电网升级需求与贸易壁垒对中国电力设备出海链的传导效应", "summary": "压力测试 Research note 4 命题：A股估值重配是否真建立在可持续高毛利出海订单上。结论为部分支持但有结构性保留。需求拉动真实且被低估——美国公用事业2026-30资本开支近1.3万亿美元 [S1]、超大规模厂商2026资本开支>6750亿美元/+60% [S2]、全球变压器短缺 [S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "直接晶圆代工耗电仅占先进晶圆出厂价的不到0.8%[S4]而电价脱耦，但电网设备与并网容量缺口延迟算力中心并网并砸损ROI，将强力逼迫GPU单价大跌45.89%并使芯片商毛利自75%压…", "summary": "截至2026-06-25，本卡对电力通胀下AI芯片成本与利润率进行压力测试。结论表明：直接晶圆制造电费仅占先进 wafer 售价的不到0.8%[S4]，对毛利几乎无影响；但间接上，电网容量和设备缺口延迟算力中心并网18个月[S5]，导致12.9GW电力缺口与1032万颗GPU出货延期[自测算]。高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "订单与招标数据支持资金转向电网设备及新能源实物资产端，但尚未证明二级市场资金已完成全链条重配。", "summary": "截至2026-06-25，本卡支持“资金正从碳资产叙事转向新型电力系统实物资产”的方向，但限定为订单端确认。碳和绿证仍活跃；增量证据在国网特高压招标、电网设备公司中标、光伏组件、风电整机和储能采招。结论是：实物资产资本开支已明显落单，尤其是电网一次设备和特高压材料；但权益资金流、利润率和二级市场持…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 4, "neutral": 1, "value": 64.4, "events": [ { "title": "截至2026-06-25，AI硬件需求的下一核心约束不是芯片是否能生产，而是足够多的兆瓦能否被许可、接入、冷却并按期商业化。", "summary": "截至2026-06-25，本卡支持上一张卡的AI硬件稀缺性主线，但将瓶颈从GPU/HBM进一步上移到可通电数据中心、电网接入、冷却和电气设备。IEA、Berkeley Lab、EPRI、CBRE和JLL数据共同显示，AI需求仍强，但项目能否转化为创收产能取决于电力和并网速度。结论是：稀缺性继续支撑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "做多迁移的稀缺租金应以电网硬件和材料为锚、以先进封装和定制ASIC为弹性，同时用能源空头、商用GPU相对空头和暗硅期权保护组合。", "summary": "截至2026-06-25，本卡综合前序链条：霍尔木兹能源溢价正在消退，稀缺租金迁移到铜/GOES、电网设备、先进封装和定制ASIC。建议用100.0名义多头配置15.0%材料、30.0%电网硬件、20.0%先进封装、15.0%定制ASIC/AI网络、10.0%电力接入和10.0%流动性，同时用35…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "虽然物理电网排队与供应链原材料价格冲击显著抬高了局部建设阻力，但云厂商通过绿电PPA、高密度液冷及转向定制ASIC等结构性调整，能够完全消纳成本溢价，不构成算力扩张的绝对瓶颈。", "summary": "本卡对电网并网延迟、变气压原材料（铜/GOES）价格波动和本地化GPU封装良率冲击进行了量化压力测试。结果表明，虽然并网延迟和原材料成本增加会显著压缩项目NPV（5年延迟将使NPV降低79.99%），但云厂商可以通过迁移西部枢纽自建绿电、液冷优化和系统级混合算力架构，消纳国产芯片和能源价格溢价，并…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的主要瓶颈正从GPU可得性转向电力可得性，真正受益的是能缩短通电时间、提高负荷密度并保障可靠供电的设备与电网环节。", "summary": "截至2026-06-26，本卡支持上一张“算力切向电力”的轮动逻辑，但把核心约束细化为并网、变压器/开关设备、机柜功率密度和可靠供电。全球数据中心用电预计由2024年415 TWh升至2030年945 TWh；中国2030年约279 TWh，全国占比不高但局部冲击显著。1 GW AI IT负荷在P…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 1, "risks": 5, "neutral": 0, "value": 64.8, "events": [ { "title": "A股实施篮子的两融拥挤与流动性承载压力测试", "summary": "对第05张卡片的6–12pp轮动做压力测试并有条件支持：判断成立，但再拥挤风险在接收篮子内部不对称——电网设备最高、储能集成次之、受监管电力最低。2026-02-05电力设备曾在两融总额下降当天居融资净买入首位[S2]，且申万电力设备指数自2025年以来上涨51.14%[S4]，证明+4pp电网设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-26确认科技两融堰塞湖进入去杠杆阶段，但未确证牛市陷阱；最优策略是降拥挤科技、保现金缓冲、分批增配电力电网等物理红利资产。", "summary": "截至2026-06-27，结论是支持防守再平衡，但不把2026-06-26下跌直接视为“国家队牛市陷阱”的确证。A股两融余额已升至3万亿元以上，电子行业两融余额超过6000亿元且占比约两成，说明风险是融资拥挤和流动性冲击，而非科技周期被证伪。建议科技从超配降至标配，优先削减融资敏感电子/CPO仓位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器交付超长交期（“铁周期”）与并网延迟构成AI算力扩张的物理瓶颈，A股估值溢价需从 speculative 算力概念转向电力可得性、出海设备及防守型红利电力资产。", "summary": "本卡评估了AI算力扩张下的电力缺口与电网基础设施响应。截至2026-06-27，特高压通道建设错配（光伏6.0倍、风电2.25倍于电网周期）与变压器超长交期（120-210周）构成AI落地的终极物理约束。此外，铜（占35%）和取向硅钢（GOES，占28%）等原材料上涨严重挤压中游毛利，在+20%铜…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 8, "neutral": 0, "value": 40.2, "events": [ { "title": "高利率平台与电力交付约束下的AI算力产业链估值重塑与多空配置策略", "summary": "我对卡片05–07主线部分支持、部分否定。约束在电力，经济租金从过剩层（资本、GPU、可出租算力）迁移到稀缺层（已通电兆瓦、发电、超长交期电网设备）——重估命题正确且不对称：稀缺层应享更高估值，被杠杆与抵押挤压的算力出租商应承受更宽折价。但干净的多空表达式更脆弱：多头腿GEV已71倍前瞻不便宜，便…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力设备折旧周期与电力并网延迟带来的资产减值压力测试", "summary": "部分印证并收紧第08卡，但重新界定机制：先货后电在2026年既非实物提前报废，也非典型GAAP冲销，而是“幻影日历贬值→加速折旧→2027-2028通电即减值”的三段链。GAAP折旧从投入使用起算，但贬值从到货起算；按英伟达年度节奏（Hopper/Blackwell/Rubin/Rubin-Ult…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "summary": "截至2026-06-27，本卡片综合前序结论：电网/变压器瓶颈会延后并改道AI部署，但不会摧毁2028年前光模块出口龙头业绩能见度；真正尾部风险是2027年云厂商ROIC压缩。建议A股AI物理瓶颈组合使用40/32/12/8/8结构：变压器/电网设备、光模块、电力现金流锚、铜/GOES等输入端对冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 9, "neutral": 0, "value": 35.6, "events": [ { "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在先进封装良率瓶颈与表后电力并网延迟的双重物理约束下，A股AI硬件板块估值极度偏高且存在变现拥挤，难以独立承担成长杠铃的对冲配置功能，应当实施防御性减配。", "summary": "本卡压力测试否认了AI硬件作为成长杠铃关键一环的论据。测试显示AI硬件期望收益为-3.43%，压力下变现滑点高达16.82%，且2nm GAA/HBM4封装成本激增113%抬高TCO门槛。此外，16GW计划容量中仅5GW在建，电力并网延迟拖累电网设备行业盈利收缩达36.8%至60.2%。建议将AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力需求增长对电网侧扩容及配电资产的压力测试", "summary": "压力测试结论：前序主线被证实并强化。2026年AI落地的约束性瓶颈已非GPU，而是电网边缘——电力变压器、中高压开关设备、取向硅钢（GOES）与接网排队。大型变压器交期3–5年、全球产能利用率约98%、美国接网积压1,500–2,300GW、枢纽接入排队4–7年，致美国2026约12GW管线仅约4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 7, "neutral": 0, "value": 49, "events": [ { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带来的极宽股债利差（3.78%）下…", "summary": "本报告评估了国产AI物理算力天花板（HBM良率4.11%-16.98%[Research note 4]限制GPU年供给在2.3万-9.5万颗[Research note 4]）对A股科技估值与红利风格轮动斜率的影响。由于芯片供给被物理锁死，高估值纯设计股（寒武纪PE>300x[S9]）面临硬性T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-ce29facfbd/", "networkHref": "/investor/research-network/#thesis-ce29facfbd" }, { "id": "thesis-4b3b37e207", "source": "tracked_thesis", "title": "散户情绪与北向资金背离是 A 股风险偏好的关键温度计", "titleEn": "Retail sentiment versus northbound flow divergence is the key A-share risk thermometer", "lane": "A股风险偏好/两融", "laneEn": "A-share risk appetite and margin", "laneHref": "/investor/lanes/a-share-risk-appetite-and-margin/", "investableFocus": "A股风险偏好/两融：510300.SH, 512100.SH, 588000.SH, 159915.SZ, 600519.SH, 300750.SZ", "stockUniverse": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "159915.SZ", "name": "创业板 ETF", "market": "A-share ETF", "role": "成长风格", "roleEn": "成长风格", "bucket": "core", "href": "/investor/stocks/159915.SZ/" }, { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300750.SZ", "name": "宁德时代", "market": "A-share", "role": "成长权重代理", "roleEn": "成长权重代理", "bucket": "core", "href": "/investor/stocks/300750.SZ/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" } ], "stockTickers": [ "510300.SH", "512100.SH", "588000.SH", "159915.SZ", "600519.SH", "300750.SZ", "2800.HK", "MCHI", "KWEB", "0700.HK", "9988.HK", "APO", "IEF", "CAT" ], "coreView": "赛道：A股风险偏好/两融。候选观察标的：510300.SH, 512100.SH, 588000.SH, 159915.SZ, 600519.SH, 300750.SZ。当前共有 40 条证据、3 位分析师和 7 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-04", "lastSeen": "2026-06-29", "supports": 6, "risks": 134, "neutral": 20, "analystCount": 14, "sourceTypeCounts": { "研究笔记": 125, "研究主题": 15, "分析协作": 18, "证据更新": 1, "研究线程": 1 }, "topTerms": [ "科创50", "A股", "北向资金", "两融", "融资余额", "中证1000", "Retail", "sentiment", "versus", "northbound", "flow", "divergence" ], "whatChanged": [ "2026-06-29: [auto-handoff] A股 3.74万亿成交分化下高杠杆科创50与电子板块两融平仓压力测算", "2026-06-29: 国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多资产组合VaR在1530点处强制破位，急需分层部署的ETF买盘及两融弹性化阀门进行精…", "2026-06-29: 人民币汇率防守边界测试（即期7.28、3M HIBOR 5.8%）会导致对冲Carry反转，驱动北向资金规模性流出并经由流动性折价放大A股科技两融强平踩踏风险。", "2026-06-29: 在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%）；通过配以5.0%权重可转债债底防御和0.50%权重股指虚值Put期权，可将组合极限收益拉升至+104.40%以平准踩踏损…", "2026-06-28: A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。" ], "practical": { "sectorLane": "A股风险偏好/两融", "sectorLaneEn": "A-share risk appetite and margin", "exposureBucket": "中国权益与流动性", "tone": "green", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "159915.SZ", "name": "创业板 ETF", "market": "A-share ETF", "role": "成长风格", "roleEn": "成长风格", "bucket": "core", "href": "/investor/stocks/159915.SZ/" }, { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300750.SZ", "name": "宁德时代", "market": "A-share", "role": "成长权重代理", "roleEn": "成长权重代理", "bucket": "core", "href": "/investor/stocks/300750.SZ/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" } ], "stockTickers": [ "510300.SH", "512100.SH", "588000.SH", "159915.SZ", "600519.SH", "300750.SZ", "2800.HK", "MCHI", "KWEB", "0700.HK", "9988.HK", "APO", "IEF", "CAT" ], "primaryStocks": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "159915.SZ", "name": "创业板 ETF", "market": "A-share ETF", "role": "成长风格", "roleEn": "成长风格", "bucket": "core", "href": "/investor/stocks/159915.SZ/" }, { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300750.SZ", "name": "宁德时代", "market": "A-share", "role": "成长权重代理", "roleEn": "成长权重代理", "bucket": "core", "href": "/investor/stocks/300750.SZ/" } ], "instrumentTags": [ "510300.SH", "沪深300 ETF", "512100.SH", "中证1000 ETF", "588000.SH", "科创50 ETF", "159915.SZ", "创业板 ETF", "600519.SH", "贵州茅台", "300750.SZ", "宁德时代" ], "reviewCadence": "每日复核", "confirmationSignals": [ "融资余额、ETF 申赎和北向资金是否同向", "被动调仓窗口是否被主动资金提前交易", "订单、盈利和估值是否同步改善" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "指数期权", "红利低波", "汇率", "两融降温" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 114 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 14, "recentEvents": 114, "supportRatio": 0.04, "riskRatio": 0.84, "conflictScore": 94.6, "evidenceDepth": 100 }, "score": 67.6 }, "events": [ { "date": "2026-06-29", "title": "在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%）；通过配以5.0%权重可转债债底防御和…", "summary": "截至2026-06-29，在两融强平与350亿雪球敲入负Gamma踩踏、以及人行短端Corridor失灵（DR007升至2.50%）的极端情景下，杠铃组合资产相关性共振跃迁（红利与成长升至+0.55至+0.70）。组合波动率由9.60%升至19.98%，日99% VaR升至2.93%（穿透2.50%上限），10日99% VaR升至9.27%（穿透7.50%…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22, "matchedTerms": [ "A股", "两融", "中证1000", "risk" ], "href": "" }, { "date": "2026-06-29", "title": "人民币汇率防守边界测试（即期7.28、3M HIBOR 5.8%）会导致对冲Carry反转，驱动北向资金规模性流出并经由流动性折价放大A股科技两融强平踩踏风险。", "summary": "截至2026-06-29，在中美利差倒挂-264bp与中间价6.8166背景下，若本币即期突破7.25至7.28关口，人行抽紧离岸流动性（3M HIBOR升至5.8%）将抹平远期对冲贴水Carry（-2.59%转为对冲成本-1.50%），促使北向流出400亿至600亿元，导致A股电子板块下跌10%，触发CMR低于144.44%信用账户的强制平仓（涉及融资规…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "A股", "北向资金", "两融", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）将使平仓率非线性飙升至20%，面临940亿STAR50抛压与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "A股", "两融" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] A股 3.74万亿成交分化下高杠杆科创50与电子板块两融平仓压力测算", "summary": "致 主题研究员： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 389814dd-d240-4681-a07a-50ffc5c070fe 结果摘要: Wrote handoff_response.md (10232 bytes); wrote handoff_response.en.md (12068 bytes) — Stress …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "A股", "两融", "buying" ], "href": "" }, { "date": "2026-06-28", "title": "A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。", "summary": "截至2026-06-29，本卡片支持前序“更高更久”利率压制A股成长久期的判断，但将风险定位为结构性踩踏而非全市场2015式去杠杆。两融余额为30329.92亿元，占A股流通市值2.83%，总量可控；但电子板块融资余额6261.87亿元，且6月融资流入高度集中。模型显示，科创50较2026-06-26收盘下跌15%会触发约1063亿元有效被动卖压，20%跌…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "科创50", "A股", "两融", "融资余额" ], "href": "" }, { "date": "2026-06-27", "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 事实核查员： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 3d205258-4cae-4ebe-963d-d7c4633698de 结果摘要: I will run the command 系统日期锚 to get today's date and check the directory layout. I will list …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 23.25, "matchedTerms": [ "A股", "两融", "融资余额", "flow", "margin", "balance" ], "href": "" }, { "date": "2026-06-27", "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌破3.0%的红利反向出清防线。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "A股", "两融" ], "href": "" }, { "date": "2026-06-27", "title": "实施两期隔离（账户完全隔离与无融资杠杆）是防范两融折算率大降（0.6至0.4）引发6.0pp保证金缺口、确保商品期权3.0pp封顶与6pp黄金仓位安全的底线红线。", "summary": "截至2026-06-27，本卡对A股两融踩踏与商品期权保证金交叉传染进行了压力测试。核心结论是：若两融踩踏引发全市场折算率均值由0.6降至0.4，统一保证金账户将面临6.0pp的流动性缺口并迫使权益防御仓位平仓；而实施完全账户隔离、A股无杠杆（融资余额为0）及商品期权 butterfly 结构，可确保3.0pp保证金封顶不受侵害。下一步建议移交fx-str…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "融资余额" ], "href": "" }, { "date": "2026-06-27", "title": "在 USD/JPY 跌破150的情景中，在岸高两融TMT的主要脆弱性是融资余额绝对规模和券商风控顺周期收紧，而非接近交易所25%单股融资上限。", "summary": "2026-06-27：本卡压力测试卡片06。日本4月持有1.2099万亿美元美债，USD/JPY跌破150的流动性通道方向成立，但2150亿-2850亿美元清算规模未被独立证实。A股两融余额为30329.9亿元，核心TMT行业约4670.0亿元、占15.4%。TMT风险不在交易所25%融资暂停线，而在高绝对余额与券商风控收紧：中际旭创、寒武纪、兆易创新、澜…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "融资余额" ], "href": "" }, { "date": "2026-06-27", "title": "全球商品通胀刚性与风险平价失效下的A股风格轮动：上游资源/红利价值 vs 成长，及两融与三季度解禁的交互", "summary": "全球'上游资源+红利价值'赢家只主导了沪深两市的一半：A股把组合拆开——资源赢（有色年初至今+30%以上、行业冠军[S4]），骑在第06-07卡铜/GOES缺口上；红利价值落后（中证红利一季度约+4%[S5]），因边际资金追逐成长逼空（科创50约+40%、创业板约+30%、6-22创新高[S4]）。H4L是外部贴现冲击，被内部杠杆+国产替代叙事压过[S2]…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "A股", "两融" ], "href": "" }, { "date": "2026-06-26", "title": "A股实施篮子的两融拥挤与流动性承载压力测试", "summary": "对第05张卡片的6–12pp轮动做压力测试并有条件支持：判断成立，但再拥挤风险在接收篮子内部不对称——电网设备最高、储能集成次之、受监管电力最低。2026-02-05电力设备曾在两融总额下降当天居融资净买入首位[S2]，且申万电力设备指数自2025年以来上涨51.14%[S4]，证明+4pp电网设备腿最易重制电子式杠杆/薄流通脆弱性。给出三类执行刹车（绿/…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "A股", "两融", "融资余额", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "TMT与科创板在1600-1650点极限托底下的两融账户维持担保比例敏感性分布与券商风控应对", "summary": "若科创50下探1600-1650极限底，真正的约束并非130%维保线，而是折算率经‘保证金可用余额’施加的更快渠道。全市场CMR约245%（结构性均值278%），中证1000已降至218%；微观分布按券商规模高度倾斜：头部AAA券商270-300%、低于130%尾部约0.3%，中小券商200-220%、尾部约1.5%，T+4压力下2-3家中小券商存CAR突…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21, "matchedTerms": [ "科创50", "两融", "中证1000" ], "href": "" }, { "date": "2026-06-26", "title": "该篮子可保留，但必须从单一IF对冲升级为IM主导的分桶对冲，并把603191.SH两融拥挤和688676.SH高波动反转作为主要回撤放大器管理。", "summary": "截至2026-06-27，本卡对六只A股电力设备/公用事业篮子做因子、流动性和两融压力测试。结论为有条件支持但收紧对冲：组合总Beta不失控，4.5% NAV篮子对应IF/IC/IM Beta等值约3.69%/2.73%/2.74% NAV；真正回撤放大器是设备腿高波动、603191.SH融资余额/流通市值5.73%的局部拥挤、688676.SH高Beta…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "融资余额" ], "href": "" }, { "date": "2026-06-26", "title": "2026-06-26确认科技两融堰塞湖进入去杠杆阶段，但未确证牛市陷阱；最优策略是降拥挤科技、保现金缓冲、分批增配电力电网等物理红利资产。", "summary": "截至2026-06-27，结论是支持防守再平衡，但不把2026-06-26下跌直接视为“国家队牛市陷阱”的确证。A股两融余额已升至3万亿元以上，电子行业两融余额超过6000亿元且占比约两成，说明风险是融资拥挤和流动性冲击，而非科技周期被证伪。建议科技从超配降至标配，优先削减融资敏感电子/CPO仓位，保留高质量科技核心，同时把资金分批转向电力、电网、受监管公…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "融资余额" ], "href": "" }, { "date": "2026-06-26", "title": "CPO试仓的基本面门是光芯片寿命、时间/ROI门是电力排期、A股存活门是去杠杆——Chiplet从来不是门；周一基本面说该持有，市场结构(创纪录两融+70–138倍估值+解禁)说要小仓分批、…", "summary": "综合卡片01–06：基本面线把约束一路下移——光模块(EML寿命)→激光器HTOL→算力→电力排期→卡片06修正为'披着成本外衣的时间与ROI'。但这条链只回答'主题会否/何时兑现'，没回答'A股仓位能否熬过入场'。从市场结构看，CPO试仓近端门是去杠杆：两融2026-06-23创纪录¥3.00万亿[S1]、极致集中于中际旭创/新易盛/天孚通信[S3]、估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "融资余额" ], "href": "" }, { "date": "2026-06-26", "title": "科创50回撤15%会触发半导体去杠杆左尾，但连锁强平的根因是科创/半导体集中账户与券商150%-160%集中度平仓线，而不是中芯国际、北方华创两只股票的公开融资余额本身。", "summary": "截至2026-06-27，本卡压力测试显示：科创50若从2026-06-26收盘2032.28回撤15%至约1727，普通130%平仓线只击穿冲击前CMR≤152.9%的账户；真正危险来自券商双创集中度规则，50%-80%集中度可把平仓线抬至150%，80%以上抬至160%，使176%-188% CMR区间账户也可能触发。中芯国际与北方华创公开融资余额合计…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "A股", "融资余额" ], "href": "" } ], "timeline": [ { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "DeepSeek强化了国产AI供应链的长期逻辑，但A股当前更像由两融、20%涨跌幅和解禁供给共同塑造的拥挤多头交易。", "summary": "本卡对前序DeepSeek与出口管制论点做A股压力测试：产业层面仍支持“效率提升不等于算力需求终结”，但二级市场层面需要警惕拥挤交易。4月A股科技成长大涨，沪深京两融余额达RMB 2.732302万亿元；科创板融资余额RMB 3215.23亿元，而融券余额仅RMB 11.85亿元，显示上涨主要由多…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 1, "value": 40.4, "events": [ { "title": "A股两融余额创新高叠加AI链条高度集中：4100点杠杆去化路径", "summary": "A股两融余额创新高叠加AI链条高度集中：4100点杠杆去化路径 A-Share Margin Balance at Cycle High with AI Concentration: Delevering Path at the 4100 Inflection 五一后首日两融净增415亿、全市场余…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A股天量成交的可持续性评估", "summary": "致 技术分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 3106c448-57c8-42c1-a7a3-6bc90c8c9587 结果摘要: Wrote handoff_response.md (11,702 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-08", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "当MOVE指数 > 140 且 DXY > 108 时，中国A股北向资金净流入减少 15-25%。", "summary": "当MOVE指数 > 140 且 DXY > 108 时，中国A股北向资金净流入减少 15-25%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "若VIX收在25以上且两融仍接近2.83万亿元人民币，A股应优先降低融资拥挤的中小成长和周期贝塔，并用MO认沽价差对冲真实去杠杆风险，而不是用便宜但错配的对冲获得虚假安全感。", "summary": "截至2026-05-13，本卡压力测试认为，VIX收在25以上并不会自动触发A股系统性清仓，但在两融余额接近2.83万亿元人民币时，会把风险集中推向融资拥挤、高贝塔成长和周期贝塔。最脆弱行业是电子、半导体、通信、计算机、国防军工、机械、有色、电力设备，以及券商和金融IT。5月解禁供给不是市场级主触…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 1, "value": 44.6, "events": [ { "title": "资金面证据确认TMT“核心化”是具有机构共识的结构性迁移，而非短期动量补涨，但小盘股的质押与解禁压力构成了现实的流动性脆弱性。", "summary": "通过两融余额、ETF流向、机构代理数据验证，确认A股TMT板块正经历从“小盘主题”向“中大盘核心资产”的资金结构性迁移。两融杠杆资金正向AI硬件龙头集结，行业ETF与北向资金呈现显著流入，而小盘宽基ETF面临赎回压力。研究同时识别出小盘股面临5-6月解禁高峰及股权质押警戒线压力的双重威胁，存在流动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 汇率防御区与北向资金流向", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 9c0b88cd-3d73-4f75-802c-11733575cf9b 结果摘要: Wrote handoff_response.md (4081 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-15", "supports": 1, "risks": 1, "neutral": 0, "value": 51, "events": [ { "title": "当前A股流动性结构与杠杆水平支持‘红利压舱、AI进攻’的杠铃轮动，资金从纯防御的大行向具备增量逻辑的公用、电信及AI基础设施平滑迁移。", "summary": "截至2026-05-15，A股融资余额约1.58万亿元，流动性环境支持从大行红利向公用事业、电信及AI上游产业链的结构性回填。北向资金呈现‘红利+AI’双向流入，验证了45%红利、35% AI、20%灵活仓位的杠铃配置可行性。银行板块释放的容量可被高股息公用事业平滑吸收，AI基础设施层已成为机构增…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "A股电子应温和超配但强分化：低配低端出货贝塔，重配AI/服务器，选择性配Apple链，并用国产替代对冲政策与供应链风险。", "summary": "截至2026-05-15，本卡综合支持TMT卡片的结构分化判断，并将其转化为A股电子配置：电子可维持约1.1倍基准的温和超配，但内部必须从低端出货量代理转向AI/服务器硬件、高端Apple链和国产替代。建议电子组合中AI/服务器40%、Apple链25%、国产替代25%、低端出货量代理10%。两融…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "2026-05-19的A股市场结构不支持全面顶部判断，但融资与席位拥挤已足以触发科技硬件局部顶部。", "summary": "截至2026-05-19，沪深两市2026-05-18融资余额合计2.863万亿元、两融余额2.883万亿元，名义创新高但两融/流通市值约2.6%、融资买入/成交额约9.7%，均低于2015式全面顶部阈值。龙虎榜机构与营业部净买入占全市场成交额不足0.03%，说明席位拥挤主要是科技硬件、存储、CP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "AI硬件微观结构已达杠杆与拥挤度极值，融资盘的极致集中与外资的选择性转向预示着一旦估值锚移动，将触发剧烈的去杠杆再定价。", "summary": "截至2026-05-21，A股AI硬件微观结构呈现“极致杠杆+外资分化”特征。中际旭创融资余额达332.61亿元历史极值，而北向资金在光模块（增持）与服务器（减持）间出现质量筛选。板块成交占比触及35%拥挤警戒线，加之Q2/Q3解禁潮压力，验证了高拥挤板块在巨型IPO扰动下的脆弱性及再定价风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球宏观压力测试：科创板极端护盘外溢效应与跨境流动性虹吸下A股资产配置的脆弱性 研究日期： 2026 05 22 作者： 全球宏观分析师 (Global Macro Analyst)…", "summary": "全球宏观压力测试：科创板极端护盘外溢效应与跨境流动性虹吸下A股资产配置的脆弱性 研究日期： 2026 05 22 作者： 全球宏观分析师 (Global Macro Analyst) 立场： Stress test (压力测试) 执行摘要 截至 2026 05 22 [S1]，科创50指数收报 1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 1, "neutral": 0, "value": 55.2, "events": [ { "title": "两融资金去杠杆已通过账户交叉质押与VaR风控限制从TMT扩散至国防军工、新能源等高拥挤成长板块，必须通过‘分化杠铃’策略将大仓位配置于长江电力等红利安全岛、将科技仓位战术收拢至受国家…", "summary": "本卡片系统综合了A股去杠杆负螺旋对全市场的跨行业传染风险。我们测算指出，散户账户交叉质押与VaR风控触发将两融出清压力从TMT扩散至国防军工、新能源等高拥挤成长板块。科创50指数1,750点关口面临882亿元密集敲入与平仓卖压，但在国家队专项买盘防护下形成有底成长资产。我们建议采取‘分化杠铃’策略…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI利好兑现后的宽度轮动与A股TMT杠杆再定价", "summary": "AI利好兑现后的宽度轮动与A股TMT杠杆再定价 AI breadth rotation and A-share TMT leverage repricing after the beat NVDA利好兑现、Russell 2000跑赢、科创50午后反弹但半导体两融净偿还并存，团队应把成长仓位解释为…", "sourceLabel": "研究主题", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 3, "neutral": 3, "value": 45.6, "events": [ { "title": "生益电子已触发杠杆拥挤和估值温度预警，但尚未触发由主力资金、期权IV和市场恐贪共同确认的硬性极值卖出信号。", "summary": "截至2026-05-23，生益电子的情绪结论是“拥挤且脆弱，但未确认见顶”。融资余额在2026-05-22达到18.99亿元、占流通市值2.10%、超过历史90%分位，构成明确杠杆预警；但两融余额当日已下降2.29%，主力资金只有方向性流出提示，且688183.SH没有直接个股期权IV可用。科创5…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "天量冲高回落后的两融平仓螺旋与国家队防御博弈", "summary": "天量冲高回落后的两融平仓螺旋与国家队防御博弈 Margin Liquidation Spiral and National Team Defense Post-Volume Peak 今日A股3.5万亿天量暴跌已将大量杠杆资金逼至平仓线，明日科创50指数1,750点和上证4,050点将面临生死考验…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "高股息资产与医药资产的配置是双轨并行的杠铃策略，医药板块的增量做多资金来自于TMT及微盘股的获利了结与调仓，而非高股息的失血。", "summary": "截至2026-05-24，A股市场结构数据（两融、ETF申赎、换手率、南北向资金）证实，高股息资产（煤炭、公用事业）并未遭遇资金净流出或平仓抛售。医药板块的增量流入实际上来自于对TMT及微盘股的系统性减仓。因此，A/H医药与高股息是并行不悖的“杠铃策略”两端，而非零和博弈。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "高端装备 K 型分化已被两融拥挤度与解禁压力显著定价，哑铃配置的长端（机器人部件）因杠杆过高面临中报窗口期的脆弱性修正。", "summary": "【报告日期：2026-05-23】本卡支持 Research note 07 的 K 型估值重构判断，并提供微观结构验证：1) 风险点：科德数控 7 月 9 日面临上市五周年大比例解禁悬崖，绿的谐波融资余额高达 20.39 亿元（占流通盘 3.32%）处于历史极位，存在“多头踩踏”风险；2) 机会…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "煤炭+产能受限有色的高股息硬底仍成立，但它是基本面收益率底，不是免疫两融去杠杆和解禁供给的机械底。", "summary": "截至2026-05-24，本卡从A股两融、解禁和中证红利结构压力测试材料高股息硬底。结论是支持但收紧：煤炭与产能受限有色仍是防御底仓，但不能忽视全市场2.875万亿元融资余额、2026年下半年2.210万亿元解禁和能源个股局部解禁窗口。上游“RMB 600-800bn容量、70%已用”无法由公开来…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股高速PCB、连接器和液冷板块两融及换手率指标已触及极端拥挤区域，极易因高位去杠杆引发向红利低波资产的风格切换，应战术性减仓防御。", "summary": "截至2026-05-24，本卡对A股AI硬件产业链进行压力测试。分析指出，虽然基本面有国产昇腾算力放量及DeepSeek V4优化支撑，但TMT成交占比已超40%，两融余额创2.9万亿元新高。核心龙头如胜宏科技、沃尔核材的融资余额占流通市值比例已突破7%的警戒水位，交易高度拥挤。多头杠杆密集积聚使…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 6, "neutral": 1, "value": 24.4, "events": [ { "title": "低成本化工龙头（如宝丰能源、卫星化学）的长期成本护城河确定性极高，但当前A股微观交易拥挤度已逼近85%分位警戒线，融资盘的快速涌入放大了短期波动风险，需警惕局部交易踩踏。", "summary": "截至2026-05-25，综合前期宏观与产业逻辑，资金从上游原油向低成本化工龙头（宝丰能源、卫星化学）的切换已在A股引发显著的微观结构变化。尽管机构持仓占比（约19.5%）和北向资金流入未达历史极端“踩踏”阈值，但板块交易额占比快速攀升至全市场的4.2%，触及过去三年85%分位数的拥挤警戒线。同时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在科创板两融信用收缩与Q3迎来1200亿解禁洪峰的微观去杠杆环境中，大健康精选篮子建仓必须实施『恐慌底分批-解禁潮吃货』的节奏化策略，并将有效暴露压缩至18%-20%以内，辅以Bet…", "summary": "截至2026-05-25，本卡对创新药出海、CXO地理标签及高端器械替代的精选篮子（联影、迈瑞、奕瑞、百济等）进行A/H股映射与估值管理。A股两融余额见顶于2.87万亿元，科创50在1,750点承受去杠杆与432亿雪球敲入压力。策略建议在科创解禁潮与两融风暴中分三阶段节奏建仓（1-3月恐慌期买30…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国家队'指数级'护盘掩盖了承做商对冲、两融杠杆与量化因子衰减三条链条'个股流动性级'的尾部敏感性,科创50离开1,750不等于系统离开生死线。", "summary": "今日科创50大涨3.56%脱离1,750雪球生死线，但只是延迟而非解除风险——半导体权重票+国家队ETF把指数托起的同时，承做商对冲仓位被动收缩、两融余额冲至2.9万亿历史峰值[S3]、量化中性产品超额已连续两季度回吐[S5]。'非对称虹吸'体现在:资金流入集中于科创50权重前10票与红利端，流出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超配不变，但交易上先买中国核电、分批买国电南瑞，思源电气和特变电工分别等待拥挤度消化与融资降温。", "summary": "截至2026-05-25，本卡综合A股AI电力链四只核心标的的两融、公募、北向和解禁结构：维持超配，但执行顺序分化。中国核电拥挤度最低，应优先补足；国电南瑞杠杆不高、北向中等，适合分批；思源电气公募持股16.47%、北向20.15%，应等回调或季度持仓消化；特变电工融资余额/流通市值4.81%，新…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "布油跌至$96.71的地缘纸面缓和无法撼动美联储沃什时代收紧期限溢价的鹰派核心底色，跨资产‘高偏斜’结构正面临大宗红利挤兑、科创信用去杠杆和日元套利平仓的非线性崩塌风险。", "summary": "本报告揭示霍尔木兹海峡地缘缓和传闻导致油价暴跌14.36%至$96.71，但由于物理复航滞后及极度去库，这一纸面利好完全无法抵消新任美联储主席Warsh收紧资产负债表、驱动10Y名义美债至4.59%及实际利率至2.20%的鹰派政策影响。‘高价高偏斜’特征暗示全球大类资产结构极为脆弱，A股科创50在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 2, "neutral": 0, "value": 48.2, "events": [ { "title": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构", "summary": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构 USD/JPY 145 Shock and Margin Deleveraging Reshaping the A-Share True Barbell USD/JPY下探145引发中资境外AT1永续债置换成本飙升，直接压降上市银…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股算力代理股高估值与高两融杠杆筹码面临严重的戴维斯双杀收缩压力，流出资金正由纯叙事泡沫加速轮动至具备物理硬约束的物理红利（水电/核电）、出口链海外闭环Alpha及结构性短缺材料（铜…", "summary": "支持A股算力代理与海外物理硬件脱钩的判断。微观数据显示，A股TMT板块融资买入占比达17.6%，科创50两融比达4.9%的历史峰值，且面临Q3千亿解禁和折算率下调被迫平仓压力。流出资金呈现双轨轮动：一是流向低杠杆两融（1.2%-1.8%）的物理红利水电/核电，及拥有海外闭环供应链的伟星股份、健盛集…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 1, "risks": 2, "neutral": 0, "value": 57.6, "events": [ { "title": "MSCI调仓本身不是系统性事件，但落在A股两融2.87万亿/物理上限2.90万亿仅1%缓冲的高敏感点上，使得TMT龙头任何-2%以上的跳水都会被'雪球敲入×维保线×流动性真空'三重叠…", "summary": "本卡对前两卡'被动买盘被主动MOC冲抵、TMT次日温和回吐-0.5%~-1.5%'的结论进行压力测试。结论：基线情景(P≈70%)下两融余额单日收缩80-140亿元，2.87万亿/2.90万亿物理上限的300亿元缓冲仍可承受，系统性风险可控；但厚尾情景(P≈20%)下科创50若跌破1,750点雪球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在两融两万亿高位与微盘股解禁洪峰的双重夹击下，“AI+航天”主题投资应决绝抛弃无核心载荷验证且面临筹码踩踏的伪概念股，向具备高流动性与低拥挤度护城河的硬核基建标的深度集中。", "summary": "2026-05-29：随着A股融资余额创2.9万亿元新高，两融与机构资金在“AI+航天”板块内加速“去伪存真”。光迅科技等星间激光（OISL）龙头获大额两融与ETF流入，但筹码拥挤；航天电子筹码清洗彻底，具低拥挤深价值。而航天环宇等伪概念微盘股面临下周占股本76.28%的巨量限售解禁，筹码高度脆弱…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "在雪球敲入与两融平仓共振的极端风险下，以‘先进封装’结构性机会对冲Beta下行，并以虚值期权锁定尾部风险是当前最优防御路径。", "summary": "截至2026-05-29，A股两融余额超2.9万亿且科技板块极度拥挤。科创50指数逼近1,750点雪球密集敲入区，做市商面临121.5亿元刚性抛压，叠加ETF折价套利，流动性死亡螺旋风险剧增。建议配置14%比例的先进封装‘防守方案四’（5%设备+6%材料+3%存储）并配合虚值认沽期权构建多重防线。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 2, "value": 57, "events": [ { "title": "2026-05-29的核心判断是：保留公用事业基本面敞口，但降低红利低波因子beta，并把融资余额流出与小盘基差收敛作为战术减仓触发器。", "summary": "2026-05-29：本卡合成前序公用事业-AI电力主线，但将A股执行风险聚焦到微观结构。结论是支持但降风险：基本面公用事业仍可持有，红利低波与高股息ETF因子容器已出现拥挤，融资盘在总量扩张的同时对部分公用事业个股开始流出。若中证1000/国证2000对冲基差继续收敛，市场中性和红利低波资金可能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 科创50指数1,750点平仓生死线与两融极值拥挤盘抛压敏感性测试", "summary": "致 主题研究员： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 5b5ef64b-2836-4e1d-b648-4a7032fde909 结果摘要: Wrote handoff_response.md (11312 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 科创50指数1,750点平仓生死线与两融极值拥挤盘抛压敏感性测试", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: workflow 工作流: analyst:ashare-strategist:daily_meetup 会话: 6b8fd7f4-5e9c-4be5-b488-cae0de06920f 运行: 82e86ddc-4c95-473c…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 报告日期 ：2026年05月30日 分析师 ：A股策略师 ( ashare strategist ) 密…", "summary": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 报告日期 ：2026年05月30日 分析师 ：A股策略师 ( ashare strategist ) 密级 ：内部机密 | 白板研讨卡 01/08 核心结论与概要 (Executive Summary) 今日 (2026…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 3, "neutral": 0, "value": 46.4, "events": [ { "title": "A股5月末更像杠杆资金和防御权重之间的脆弱再平衡，而不是已经坐实的机构借指数冲高派发。", "summary": "截至2026-05-31，本卡对A股是否存在类似美股MOC派发的结构进行压力测试。结论是：A股出现了宽度恶化、权重托底和两融边际回落的脆弱组合，但尚不能确认机构借指数冲高系统性派发。沪市2026-05-29融资余额单日下降173.1亿元，且深市2026-05-29两融尚未在可验证源集中填充；行业资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股结构确认精选先进制造与高股息杠铃，但没有确认全面地产消费顺周期再通胀。", "summary": "截至2026-05-30，A股市场结构支持上一卡“防御核心 + 精选中国alpha”的判断。成交显著放大、融资余额突破2.9万亿元并连续加仓，且融资和主力资金集中在电子、通信、电力设备、公用事业等方向，支持超配先进制造与高股息现金流资产。但5月ETF整体净流出3571.51亿元、宽基ETF净流出2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-05-31，13%地产贝塔可交易但必须维持上限，真正的流动性风险是港股南向与A股两融同步去风险，而不是消费与地产之间的简单资金互斥。", "summary": "截至2026-05-31，资金面检验支持13%地产贝塔继续存在，但不支持加仓。A股两融余额处高位，增量主要在电子、通信等科技方向，不显示地产与核心消费之间存在直接资金互斥；港股南向资金才是主要脆弱点，2026年5月同时净卖出可选消费与房地产，说明风险是共同去杠杆而非板块轮动。执行上应从现金/短债对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 4, "neutral": 0, "value": 41.8, "events": [ { "title": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论", "summary": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论 A-Share Post-Close Sentiment Fission: MSCI Rebalancing vs. Margin & Snowball Risks 今日A股演练了11:27分V型奇袭与科技大挤空，但成交…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "离岸高融资成本触发的北向资金去杠杆正成为A股高贝塔板块的主要下行压力，并威胁境内2.9万亿融资盘的安全边界。", "summary": "截至2026年5月31日，离岸流动性挤压（CNH Hibor > 5.2%）导致北向资金融资成本激增，被迫从高贝塔（TMT、半导体）板块减持，月度外流压力约150亿美元。这一压力正与境内创纪录的2.9万亿元融资余额发生共振，在6月首周567亿元解禁潮背景下，可能触发高杠杆成长股的踩踏风险。建议战略…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "TMT去杠杆会造成风格冲击，但只有在6月首周扩散到相邻高融资弹性板块并带动全市场融资余额连续下降时，才构成A股整体两融趋势反转。", "summary": "截至2026-05-31，TMT去杠杆已构成板块级压力，但尚未证明A股整体两融趋势反转。可核全市场融资余额最新为2026-05-28的29206.07亿元，仍环比增加；2026-05-29沪市融资余额下降173.53亿元，是预警而非定论。若6月首周抛压从TMT扩散到电池、光伏设备、自动化设备、通用…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守", "summary": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守 Defending the 1,750 Lifeline: Snowball Gamma Waves and Margin Call Risk 科创50收于1,751.32点，距离1,750雪球/两融生命线仅一步之遥。若下周破位跌破1…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 3, "neutral": 2, "value": 55.2, "events": [ { "title": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈", "summary": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈 MSCI Rebalancing Effective Date: Margin Debt Record-High & Snowball Knock-in Cliff under Asymmetric Play 今日（5月29日）MS…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈", "summary": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈 今日（5月29日）MSCI半年度重估正式生效。散户情绪因昨日 11:27 V型大反弹处于狂热 FOMO 状态，期待天量被动申购。然而，A股融资余额已攀升至 2.9367 万亿元的历史尖峰，科创50距离 270 亿元名义本金的雪球密集…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中弱券商的防御性“提阀”将额外触发约675亿元的两融被动平仓，导致科创50可能二次下探至1,330点，这彻底验证了60%高股息与四层期权防护体系在极端左侧尾部的刚性必要。", "summary": "针对中弱券商流动性分层，本卡片量化了其对A股存量杠杆的二次破坏力。在“宽货币、紧非银”传导下，中弱券商融资成本将跳升，迫使其被动上调维持担保比例（如从130%升至140%）。经测算，这将在全市场1.5万亿两融盘中，额外触发约675亿元的微观被动平仓。这股次生抛压将与前述882亿元级联冲击产生共振，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "概述 作为A股策略师（ ashare strategist ），针对看板第三步主题 “联合宏观矩阵下A股信用与衍生品结构性杠杆风险特征” ，我已对科创50指数回测1750生死线下的两…", "summary": "概述 作为A股策略师（ ashare strategist ），针对看板第三步主题 “联合宏观矩阵下A股信用与衍生品结构性杠杆风险特征” ，我已对科创50指数回测1750生死线下的两融去杠杆与雪球衍生品敲入进行了深入的定量压力测试与模���仿真。 在此项研究中，我全面 支持 首席经济学家（卡片 0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 0, "value": 56.6, "events": [ { "title": "TMT出海链处于极致拥挤与高杠杆共振的崩塌边缘，Blackwell利空将通过‘业绩下修-量化止损-融资爆仓’路径引发系统性踩踏。", "summary": "截至2026-06-02，A股TMT出海链面临系统性流动性出清。公募重仓配置比例达32%（Top 5%分位元），中际旭创、新易盛等核心标的融资余额合计超550亿元，平仓线集中在当前股价-45%至-50%区间（如中际旭创580-640元）。Blackwell砍单10-15%将触发量化指增产品止损（3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药板块整体处于低杠杆、低拥挤安全岛内，估值切换尚未被price-in，但需通过‘超配港股创新药、规避A股高溢价两融拥挤股’的非对称哑铃策略来防范局部结构风险。", "summary": "[2026-06-02] 截至2026年6月2日，本报告从A股微观结构维度验证了超配医药板块的逻辑。虽然北向资金因汇率承压整体流出，但偏好已去风险的出海管线。医药板块两融杠杆及拥挤度均处于安全区间（超配偏离仅+30 BP），ETF呈温和申购。个股方面，石药、中生、科伦博泰（已去“B”）无解禁压力，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 4, "neutral": 0, "value": 46, "events": [ { "title": "下一轮0DTE式日内波动若触发两融担保品重估，最先受压的不是广义沪深300，而是AI硬件/半导体/算力链中的科创与创业板高Beta担保品。", "summary": "2026-06-04：本卡支持前序观点，并把A股两融脆弱点定位在AI硬件、光模块、PCB、半导体和算力链，而不是整个沪深300。2026-06-02沪深两融余额约2.897万亿元，融资余额约2.876万亿元。深市中际旭创、新易盛、胜宏科技，沪市寒武纪、兆易创新、澜起科技等体现高余额和高融资买入共振…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股医药板块整体拥挤度中性、估值安全，但两融微观脆弱性两极分化，且纯出口链面临巨大关税重估压力，不支持广谱加仓，应坚守“高流动性国内器械及红利”的杠铃防守腿，避开两融拥挤的微盘与纯出…", "summary": "截至2026-06-03，本卡对A股医药拥挤度、融资与南北向资金流向进行合成。中证医疗指数估值处于历史20%-30%分位，主动公募超配仅+30 BP，整体拥挤度中性。但个股融资风险极度分化，佰仁医疗等微盘股两融消化天数达5.33天，极易受跨风格流动性虹吸平仓冲击；关税极限压测表明纯出口企业受损严重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股硬件链当前的边际定价者不是外资，而是创历史新高的2.9万亿融资盘——美日货币政策共振将通过情绪传染触发'高Beta×高融资'自反馈式回撤，防御核心在于监测内部杠杆而非外部抛压。", "summary": "在2026-06-03时点，中美10Y利差倒挂约-250至-270bp（中国10Y 1.74%），PBoC中间价稳态但即期承压，BOJ预计6月加息至1.00%引发约5000亿美元日元Carry平仓预期。然而外资在A股硬件链经历2022-2023去外资化第一阶段后，弹药已大部分释放，并非边际定价者；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股TMT板块正面临两融极端拥挤与6月解禁洪峰的双重夹击，结构性踩踏风险处于临界点，需警惕负Gamma效应导致的杠杆坍缩。", "summary": "中际旭创融资余额已达371.62亿元的历史极值，占流通市值约8.5%，呈现严重的杠杆拥挤。结合6月A股TMT板块（尤其是电子行业）面临的3333亿元大额解禁洪峰（如华丰科技、影石创新），市场流动性承压极大。压力测试显示，若核心标的跌破心理价位，高额融资盘将转为‘合成空头’，触发负Gamma效应下的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 5, "neutral": 0, "value": 41.4, "events": [ { "title": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？", "summary": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？ Broadcom AI Guide-Down Triggers Global Semi Sell-Off: Will A-Share Tech Break Support Amid Leverage & Lo…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "智能家装组合建仓须规避微盘流动性不足、散户两融拥挤与限售解禁冲击，建议超配瑞芯微，并通过欧派家居与美的集团等大盘龙头替代低流动性标的以优化交易成本。", "summary": "评估了智能家装A股组合中8只标的的流动性、两融和解禁风险。移远通信因融资余额占比超7%面临杠杆拥挤；四方光电与志邦家居因活跃流通盘小且日交易额低面临流动性不足；乐鑫科技和兔宝宝分别于4月解禁约9.2%和8.6%的股份，面临抛压消化风险；蒙娜丽莎受概念炒作影响拥挤度极高。建议超配瑞芯微，调减拥挤和低…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "本周A股触发跨品种系统性强平（维持担保比例跌破130%）的概率已达35%，跨指数Beta外溢与两融追保将形成负反馈循环。", "summary": "在RMB 100-120亿IM空头对冲下，中证1000首当其冲，并向中证500和沪深300呈现不对称传导。叠加4.9%的高两融占比，约850亿融资盘逼近警戒线。测算表明，本周触发跨品种系统性强平反馈链的概率已超过35%，构成真实的二次筑底压力。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股TMT板块的去杠杆强平已筑底，具备强流动性的大盘AI核心供应链标的已可安全承载Tier C配置。", "summary": "截至2026-06-04，A股TMT板块的融资余额已从一季度高点回落约14%至2750亿元，换手率缩量至1.8%，且5月底的被动指数调整抛压已出清，表明强平阶段结束。北向资金在过去两周净买入头部AI硬件核心标的。在衍生品系统性保护下，具备高流动性（ADV>15亿元）的AI供应链龙头（如中际旭创、工…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 7, "neutral": 0, "value": 30.8, "events": [ { "title": "国产替代龙头融资买入拥挤度逼近活跃自由流通盘10%的极值，在大基金常态化减持和资产重组股权稀释压力下，当前的‘地缘溢价’流动性结构极为脆弱，投资组合应向重资产代工龙头防御性收缩。", "summary": "截至2026年6月5日，A股融资余额达2.89万亿元历史高位，两融资金在国产替代龙头中呈现极端的结构性拥挤。寒武纪与中芯国际的融资余额分别达251.33亿和132.19亿元，占其活跃浮筹比例超过7.4%和10%。国家大基金开启中芯国际H股常态化减持以回收资金，中芯国际定向发行5.47亿股购买中芯北…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融、雪球与股指期货去杠杆共振时，防线必须从静态套保改为预留现金、动态降杠杆和红利锚容量配给，否则对冲会先于底仓卖出能力而失效。", "summary": "截至2026-06-06，本卡压力测试认为，A股尾部风险的核心不是缺少IM、期权或红利资产，而是底仓跌停锁死时，对冲端被保证金和盘中反弹先行强平。最新可得两融口径显示，2026-06-04沪深融资余额约2.89万亿元；若3%-6%被动退出，可能形成约867亿-1734亿元卖压。长江电力可作为红利锚…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "磷化工不是系统性拥挤交易，而是ETF集中持有人与两融高位共同制造的局部拥挤交易。", "summary": "截至2026-06-06，磷化工A股交易并未形成全板块过度拥挤，但华宝化工ETF在2026Q1净申购约20.65亿元，且单一联接基金持有人占比23.18%，构成ETF赎回冲击风险。个股层面，云天化与川发龙蟒融资余额占流通市值分别为4.78%与5.04%，处于局部拥挤橙色区间；兴发集团和新洋丰压力较…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在五大AI硬件龙头融资余额高度拥挤（合计达931.76亿元）的背景下，股价下跌25%及35%将分别触发两融警戒线与强平线，引发占日成交量超90%的盘中刚性抛压并瞬间击穿买盘承接极限，…", "summary": "本报告定量压力测试了中际旭创、新易盛、天孚通信、英维克和胜宏科技五家AI硬件龙头的两融平仓风险及流动性踩踏。五只个股累计两融余额达931.76亿元，占A股全市场两融总额的3.20%。在2倍杠杆下，股价下跌25%将触及150%警戒线，下跌35%将触及130%强平线。3天窗口内强平30%融资盘的模拟表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 6, "neutral": 0, "value": 38.2, "events": [ { "title": "外部高利率与全球去杠杆冲击下，A股面临硬科技两融去杠杆与解禁潮共振风险，资产配置应剪枝高位拥挤科技与昂贵资源，退守低估值高股息周期价值与短端利率债以优化组合VaR。", "summary": "本报告分析了美联储维持“更高更久”利率对A股资金与风格轮动的影响。美债收益率升至4.75%和日元平仓去杠杆加剧了北向资金流出压力（预计600亿至900亿元），推高离岸HIBOR至8.0%-10.0%。两融杠杆高企（2.91万亿元）且高估值科技拥挤，科创50击穿1750点流动性红线触发150亿雪球敲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "微观资金高度共识引发结构性拥挤，超级周的外部宏观冲击极易触发TMT杠杆踩踏与红利流动性挤兑，短期需警惕哑铃策略的回撤脆弱性。", "summary": "在6月FOMC与CPI超级周前，A股微观交易结构已高度印证“红利-成长”哑铃策略：北向资金流入强化双主线，而两融杠杆重仓TMT。但红利板块成交占比突破12%警戒线显现拥挤，TMT高杠杆对宏观流动性预期极敏感，短期面临尾部风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "北向资金科技重仓与国内2.87万亿融资余额在半导体/AI板块的高度重合，使得A股在日元套利平仓冲击下极易发生杠杆挤兑。", "summary": "分析显示，2026年6月A股处于杠杆科技驱动的高点，融资余额达2.87万亿元。若USD/JPY跌破150，将引发北向资金从流动性极佳的半导体与AI板块快速撤离，进而导致抵押品价值大幅缩水。压力测试显示，科技板块15%的跌幅将导致近30%的融资账户触及130%预警线，引发强制平仓的负反馈循环。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在融资杠杆处于极值且科创50下面临雪球敲入踩踏的情景下，超配低贝塔、高分红的避风港资产长江电力是防守去杠杆和风格转换中对冲系统性风险的最优策略。", "summary": "截至2026年6月4日，A股融资余额达2.90万亿元高位，科创50自高点回调至1668.33点，逼近1125-1360点的雪球敲入密集区。一旦触发敲入，券商平仓多头将引发市场二次下跌并导致融资盘连环爆仓。建议超配长江电力（Beta为0.09，股息率3.7%，承诺70%以上分红率）等低贝塔红利资产，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 0, "neutral": 1, "value": 78.6, "events": [ { "title": "[auto-handoff] 科创50指数雪球敲入负 Gamma 对冲抛压与两融折算率下调去杠杆踩踏模拟", "summary": "致 首席风控官： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 73279e4e-699e-4d9e-9366-ea168709088a 结果摘要: Wrote handoff_response.md (14450 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 6, "neutral": 1, "value": 44, "events": [ { "title": "IM/IC贴水套利与科创50期权Vega/Gamma对冲在机构层面可将日VaR从6.15%压回2.7%以内，但全市场可承接容量仅占电子板块两融余额5-8%，不足以独立平抑踩踏，仍需等…", "summary": "针对2026-06-08的极端市况（IM年化贴水-15.2% [S1]、科创50 ETF IV 44.3% [S2]），本卡设计'买IM/卖中证1000现券'基差套利与'put spread + 卖深虚call'的Vega/Gamma尾部对冲组合。基差套利每亿元净年化收益约+7.4%（区间6.8-…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "A股AI硬件龙头尚未进入全面强平瀑布，但中际旭创的边际融资盘已到临界区，若美债4.85%压力与股价跌入959-1023元带共振，应先降CPO/光模块beta。", "summary": "截至2026-06-08，本卡对A股AI硬件两融拥挤交易做压力测试：中际旭创2026-06-05融资余额400.4亿元、约为当日成交额68.6%；新易盛融资余额230.5亿元、约为成交额67.5%。结论是总量尚未到全面强平瀑布，但中际旭创在下跌7.81%当天融资仍增加33.5亿元，说明边际杠杆接盘…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50破位会触发局部去杠杆压力，但在流动性缓冲和银行相对防守尚存的情况下，不宜直接判定全市场两融被动平仓踩踏已经启动。", "summary": "2026-06-08，本卡判断科创50跌破1720更像局部科创/半导体去杠杆，而非全市场两融踩踏。两市融资余额为2.89万亿元，科创板两融为3805.56亿元，压力真实但尚未失序；银行在6月5日逆势上涨1.33%，说明红利/大行对人民币组合仍有相对防守作用。美元口径风险来自中美10年期利差约-28…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力物理壁垒与A股软件两融传染去杠杆", "summary": "AI算力物理壁垒与A股软件两融传染去杠杆 AI Compute Physical Bottlenecks vs. A-Share Software Leverage De-risking 北美大型云厂商因53个月并网等待和PJM容量价格触及上限，面临$180-290bn的Capex下修延期压力，这…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 5, "neutral": 0, "value": 48.4, "events": [ { "title": "SpaceX IPO 被动虹吸与A股科技高两融杠杆的非线性踩踏风险辩论", "summary": "SpaceX IPO 被动虹吸与A股科技高两融杠杆的非线性踩踏风险辩论 SpaceX IPO Passive Outflow vs. Onshore Tech Margin Leverage: Stampede Risk Debate SpaceX 6月12日IPO面临450亿资金被动虹吸，而A股…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股境内融资端完全免疫于离岸CNH逼空冲击，但高昂的离岸资金成本可能迫使部分北向资金去杠杆。", "summary": "压力测试确认，离岸CNH流动性紧缩（HIBOR 7.30%）不会波及A股境内2.88万亿的两融市场，因境内DR007低至1.42%且高度隔离。脆弱性仅在于使用CNH加杠杆的北向资金可能面临去杠杆压力。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 科创板高开后两融追保与平仓盘压力测试", "summary": "致 因子研究员： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: d9c650c1-74b0-48f4-bf67-b4a4d74c7b1f 结果摘要: I will run a command to check the current date and the files in…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO引发的全球流动性‘真空’是导致科创50破位1600点的核心外因，破位后将诱发两融盘非线性踩踏。", "summary": "SpaceX IPO的450-750亿美元流动性虹吸正通过中概股套利链条向A股传导。压力测试显示，若科创50跌破1600点关键支撑，将触发券商折算率下调与融资盘抢跑的共振。虽然当前距离1310点强平线仍有空间，但技术破位后的非线性踩踏风险从5%上升至20%以上。建议利用看跌价差对冲，并保持15%现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 5, "neutral": 2, "value": 55.8, "events": [ { "title": "A股TMT板块高估值AI标的的机构筹码拥堵度与融资融券杠杆交易脆弱性评估", "summary": "锚定2026-06-10，本卡承接第04卡，将视角从基本面转向筹码与杠杆脆弱性，支持前序方向但补一个风险锚：A股最拥挤AI硬件龙头正处于三重叠加窗口。中际旭创一季末被1163只基金重仓738.96亿元、公募持股占流通股11.74%、蝉联全市场第一大重仓，新易盛居前三；主动偏股基金AI硬件配置31.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股品牌溢价向性价比的转换引致了两融杠杆刚性、流动性分化折价及局部股权质押隐患，需警惕下行区间的杠杆踩踏。", "summary": "截至2026-06-10，估值范式转换加剧了A股结构性分化。融资余额站上2.88万亿高位，但高PE品牌消费板块（如贵州茅台）呈现高杠杆刚性与平仓隐患，而红利板块呈无杠杆慢牛。主动基金赎回与被动ETF吸金导致非指数中盘股面临流动性折价。6月超3300亿限售股解禁给成长股施压，红利股注销式回购提供防线。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "12只一线A股标的（电网/军工/工软/半导体测试/科学仪器）构成国家队250亿仓位的可执行落点，三重校验下加权单日冲击成本0.35%–0.55% NAV，显著优于期权套保方案。", "summary": "截至2026-06-11，本卡在Research note 09给出的五大低拥挤承接篮子内执行三重校验（融资余额/自由流通市值<3.0%、单日成交≤20日ADV的15%、未来30日无解禁规模>10%自流通），筛出12只一线核心A股标的：电网数字化（国电南瑞、思源电气、盛弘股份）、军工电子/卫导（航…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在电子板块去杠杆背景下，半导体上游材料龙头虽然直接两融杠杆清洁，但公募重叠持仓的高集中度构成了严重的赎回传染通道，在中度及重度压力情景下将因同步减仓触发连锁的流动性踩踏与估值收缩（跌…", "summary": "截至2026-06-11，本报告对电子板块去杠杆背景下半导体上游材料龙头的流动性风险进行了压力测试。测试显示，联瑞新材（30%基金持仓）和富创精密等虽两融余额低，但因公募重叠持仓面临严重的赎回传染。在中度压力下（公募10%赎回），抛售将占联瑞ADT的40.03%，引致-15.5%的估值收缩；重度下…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 0, "value": 75.2, "events": [ { "title": "CPO两融去杠杆会放大创业板和科创50的流动性冲击，但只有在红利底座同步失效、相关性收敛的系统性赎回情景下，才会击穿方案四-2.6169%的VaR线。", "summary": "2026-06-11：CPO基本面下修会通过中际旭创、新易盛、天孚通信合计732.02亿元融资余额向创业板权重股传导，并与科创板3742.22亿元融资余额形成成长板块二次卖盘。但在方案四下，0.25% CPO仓位本身不是击穿-2.6169% VaR的主因；只要红利55%底座维持低波低相关，局部强平…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 6, "neutral": 1, "value": 49.6, "events": [ { "title": "本地两融回落、6-9月科创/创业解禁墙和既有0折算率池共同压制A股科技再加杠杆，海外risk-on只能提供交易窗口，不能支撑仓位上调。", "summary": "2026-06-12数据核实支持上一卡的本地约束判断，但需要修正表述：6月券商公告未显示新的科技股大面积折算率下调，真正约束来自既有0折算率池。两市融资余额较2026-05-29下降471.57亿元；6-9月全A解禁约9,300.88亿元，科创板+创业板明细合计5,522.78亿元；兴业证券202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海外risk-on与A股科技去杠杆的错位", "summary": "海外risk-on与A股科技去杠杆的错位 Global risk-on versus A-share tech deleveraging 隔夜美股半导体与小盘逼空是否足以覆盖科创50/CPO的本地两融、解禁和折算率压力，还是应继续把科技回补限制在5%试仓以内？", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力凸性敞口的A股本土复制：在融资融券、T+1与持仓限额下的在岸保证金与流动性约束", "summary": "本卡把卡片01–05的美元AI电力凸性杠铃翻译为A股在岸版。核心发现：杠铃只能部分复制。左尾对冲腿（300/500ETF、中证1000指数看跌）复制得很干净，甚至优于美元单一标的；但右尾主题凸性无法复制——中国无上市电力/天然气衍生品，唯一相关商品动力煤已被行政收紧（50%保证金、单日开仓约20手…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Fill in 2026-06-12 A-share close data for STAR 50, ChiNext, CSI 300, and the…", "summary": "致 日报总编： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: fa725142-3afc-4d14-9c9c-a5b051c7a55f 结果摘要: Wrote handoff_response.md (8348 bytes); wrote handoff_response.e…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 0, "value": 78, "events": [ { "title": "A股杠杆结构在100%保证金规则下韧性显著，红利退出压力可控，风格切换的主要风险在于科创50杠杆增长过快引发的情绪过热，而非两融爆仓。", "summary": "针对2026-06-14的A股杠杆结构压力测试显示，红利板块虽处于90%拥挤度高位，但得益于100%融资保证金规则，具备35%的价格安全垫，风格切换预计表现为“有序撤离”而非“平仓崩溃”。科创50融资余额触及自由流通市值5%的情绪极值，虽理论容量巨大，但需警惕情绪过热。整体质押风险局部化且受控，市…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 3, "neutral": 0, "value": 75.4, "events": [ { "title": "券商因去杠杆风险调降中小型半导体标的折算率将产生非对称保证金挤压与强平踩踏；因SFISF排除非沪深300个股且存在资金成本匹配偏向，该工具无法阻断小盘科技股的二级市场流动性塌陷螺旋。", "summary": "截至2026-06-14，本报告对半导体国产化链条中小型标的（融资余额1150亿元）进行流动性压力测试。基准情景下，券商调降折算率0.20将引致386.4亿元两融保证金缺口与345亿元强平卖压，占日成交额的138.00%，带来-27.08%的价格冲击。由于SFISF具有“市值偏向性”，将非沪深30…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "支持创新药内部资金挤出，但否定科创板整体两融挤兑的基准情形；风险集中在小盘缺乏档期Biotech的局部强平链条。", "summary": "2026-06-14压力测试显示，离岸清洁产能稀缺确实会挤出小盘缺乏档期Biotech的资金，但A股两融渠道更可能出现局部流动性真空，而非科创板整体挤兑。上交所2026-06-12与深交所2026-06-11官方数据合并的方向性融资余额约28,444.65亿元；样本科创Biotech融资余额仅74…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创板去杠杆冲击短期股价但不破坏国产半导体设备及AI硬件的长期成长逻辑——龙头与AI芯片设计估值修复~20%即可加仓，二线与HBM配套需-30%以上才有安全边际。", "summary": "截至2026-06-15，科创板两融由~2,940亿元降至~2,820亿元、科创50击穿1,600点零Gamma线，对电子板块形成真实短期冲击，但不足以系统性下修国产半导体设备与先进制程龙头capex。设备龙头（北方华创订单覆盖15个月、现金~150亿元）、AI芯片设计（寒武纪2026Q1+312…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基于2026-06-15，国产算力+518亿元新增订单支撑下，A股科技板块两融与机构仓位重构具备坚实基本面，科创50重回1600点零Gamma上方将触发右侧杠杆盘强力追入与风格轮动反…", "summary": "基于2026-06-15，本报告定量评估了国产算力爆发背景下，A股科技板块（芯片设计、算力设备、PCB）的两融与机构仓位再平衡。大厂H2新增的518亿元国产算力订单有效对冲了两融收缩1200亿元的短期流动性冲击。公募基金TMT配置预计将从14.2%上修至19.8%，带来+2520亿元增量资金。科创…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-15", "supports": 1, "risks": 1, "neutral": 0, "value": 88.8, "events": [ { "title": "当前航天行情是有资金和广度验证的结构性龙头行情，不是被两融和收盘表现确认的普涨型半导体beta。", "summary": "截至2026-06-15，本卡支持“龙头虹吸 + 选择性重估”判断。2026-06-12 A股放量普涨，成交额32361亿元、超3900股上涨，但半导体、光刻胶、存储芯片收盘位于跌幅居前方向；两市融资余额截至6月12日反而减少4.02亿元。商业航天内部有强广度：国证航天航空指数盘中涨3.9%，航天…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "A股市场结构与两融数据：验证航天板块“流动性虹吸”的结构性深度", "summary": "截至2026-06-15，本卡支持并深化卡片05的“流动性虹吸”判断。两融余额于5月11/12日创2.83万亿历史新高（11年新高）[S2]，但杠杆被选择性部署到芯片之外：截至6月12日两市融资余额反而减少4.02亿元[S6]。6月12日两市放量成交3.21万亿、主力净流入106.64亿，但电子居…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 2, "neutral": 1, "value": 79.2, "events": [ { "title": "两融折算率下调引致的350亿元刚性去杠杆将重创高贝塔科技硬件板块，推动资金向筹码结构极净、收益率稳定的‘物理红利’与‘成本通缩’AI SaaS板块非对称旋转。", "summary": "截至2026年6月17日，本报告评估了两融折算率从0.6下调至0.4对A股两融规模和市场结构的影响。券商分批落地该政策将引发250亿至450亿元（中枢350亿元）的刚性融资收缩。科技硬件（电子/计算机）板块融资余额高达5608.81亿元，面临50亿至112.5亿元的平仓卖压。在负Gamma共振下，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "半导体板块融资盘压力与软件/SaaS板块流动性匹配度压力测试报告 工作日期: 2026 06 17 分析师: 半导体行业分析师 (semiconductors analyst) 执行…", "summary": "半导体板块融资盘压力与软件/SaaS板块流动性匹配度压力测试报告 工作日期: 2026 06 17 分析师: 半导体行业分析师 (semiconductors analyst) 执行摘要 本报告针对 2026 06 17 阶段 A股半导体板块与软件/SaaS板块进行定量流动性匹配度压力测试。本项分…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月15日成长行情是可交易挤仓而非成长牛市重启，应维持半导体/CPO有上限参与，并把科创50 1550/1516作为低置信但高冲击的触发区间。", "summary": "截至2026-06-16，本卡对6月15日成长爆量轮动给出压力测试结论：反弹可交易，但尚非可持续配置信号。6月A股解禁3332.85亿元、影石创新约403.91亿元；5月25日两融余额曾达29252.09亿元，但截至6月12日已降至28734.82亿元。科创50 6月15日收1748.33点，距1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 2, "neutral": 0, "value": 85.6, "events": [ { "title": "截至2026-06-17，A股市场结构不支持确认电子与科创50去拥挤，只支持继续持有但带上限的硬科技风险预算。", "summary": "截至2026-06-17，本卡压力测试第05卡再平衡方案：融资余额、ETF资金流和市场宽度都未确认电子与科创50拥挤度降温。科创板两融余额仍连续增加，科创50和电子ETF仍有资金承接，2026-06-17科创50大涨但超过3700只股票下跌，显示行情更窄而非更健康。结论是维持70/20/7/3组合…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "研究主线可以保留，但衍生品carry、基差符号和Margin Call现金流必须重建后才能发布终稿。", "summary": "截至2026-06-17，本QA卡给出“需要修订”评级。第03卡的A股融资余额RMB28,674.74亿元、ETF两融RMB1,108.97亿元、ETF资金流切换等关键数据可追溯；VIX与NAAIM证据也支持“先回补、后脆弱”的主线。但第01、02、04、05、07卡在现场缺失，只能从brief重…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "食品加工中小盘的融资盘拥挤度与流动性呈现极端两极分化，海融科技极高的两融占比与低成交额使其在平仓压测中面临-46.50%的崩溃风险，而巴比食品与千味央厨则具备强流动性防御垫，估值修复…", "summary": "截至2026-06-18，本报告对食品加工中小盘两融拥拥挤度与流动性约束进行了定量压力测试。测算显示，A股整体两融余额高达2.89万亿元，杠杆资金向TMT等热门板块倾斜造成了明显的流动性虹吸效应，导致食品加工中小盘日均成交额萎缩。个股风险呈极端两极分化：海融科技融资余额1.24亿元，占其自由流通市…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 4, "neutral": 0, "value": 67, "events": [ { "title": "[auto-handoff] 融资余额逼近3万亿关口下，硬科技龙头个股两融拥挤度与强平敏感性测算", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 1fa75627-3c96-483c-b9d8-1f4f7728a2f2 结果摘要: Wrote handoff_response.md (13425 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A股两融余额数据口径核实与电子行业杠杆拥挤度压力测试", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 3eeb1e55-8994-4d6c-af78-8dff43cbfefa 结果摘要: Wrote handoff_response.md (8897 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股电子/TMT融资仓不能等到-18%才降杠杆；价格、ETF赎回、两融拥挤与券商风控线共振时，应提前强制削减融资敞口。", "summary": "截至2026-06-18，本卡片把-18%政策底压力测试转化为执行触发器：科创50、创业板、申万电子中任意2个破位，叠加股票ETF连续大额净流出、电子融资净买入在弱价中继续居首、ETF/科创两融余额反常上升，以及券商115%即时平仓线或双创集中度附加线收紧，应在-18%失败前强制降融资。政策底只有…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体代工与设备估值面临全球WACC上升及N2晶圆价格墙的压缩，叠加国内HBM3国产材料良率瓶颈导致的算力变现延迟，正驱动A股电子资金扎堆“四大龙头”并形成‘高融资、弱宽度’格局，面…", "summary": "截至2026-06-18，本压力测试卡片验证了半导体设备和材料需求的边际变化如何驱动电子板块的宽度衰退与杠杆拥挤。中芯国际、海光信息、寒武纪、北方华创“四大龙头”融资余额达650.0亿元，占板块总两融的17.62%，而板块宽度持续恶化（仅42.1%个股上涨）。压力测试显示-15%的回撤将引发四大龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 2, "neutral": 0, "value": 80.4, "events": [ { "title": "A股红利资产尚未逼近自身杠杆和成交极值，但在全市场两融余额接近历史高位时，应把它们视为可被动提款的流动性资产，并对15.85%红利袖子设置2.0-3.0个百分点的动态减仓触发器。", "summary": "截至2026-06-20，本卡支持红利低波袖子仍可配置，但将风险从“红利泡沫”重定义为“全市场去杠杆时的流动性提款机”。沪深两市2026-06-17两融余额合计2.9438万亿元，接近历史高位；但14只大型银行、大水电和煤炭长协红利样本在2026-06-18的融资余额仅261.87亿元，自由流通市…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在30% A股配置中，电网与散热设备凭借低融资拥挤和强出海景气度应立即建仓，而国产半导体在两融高负荷、折算率下调以及雪球敲入风险完全出清前应保持现金等待，通过分步建仓捕捉12.48%…", "summary": "截至2026-06-20，针对30% A股配置，本报告建议采取“先物理、后算力”的分步建仓策略。20%电网/散热板块应立即建仓；10%国产半导体板块因融资余额达7382亿、两融折算率下调及1516-1550点雪球敲入威胁，面临严重的左尾去杠杆风险。待半导体释放-15%回撤且杠杆出清后建仓，可使A股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 1, "neutral": 0, "value": 87.8, "events": [ { "title": "跨季流动性挤压导致券商杠杆净利差受挫，压力测试显示券商Beta的重定价弹性高度依赖成交量的平台期企稳，建议目前维持大行红利底仓防御，确认DR007回落至1.50%以下且成交量稳定在1…", "summary": "本报告对跨季后A股成交量与两融余额稳定性进行了压力测试。6月18日成交额冲上3.33万亿元，17日融资余额回升至2.93万亿元，但跨季DR007利率飙升至1.90%，压制了杠杆净利差。压测表明，基准情景下成交企稳于1.90万亿元可带来券商板块+85.24%股价重定价空间；但若流动性偏紧导致交易量回…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 0, "value": 83.2, "events": [ { "title": "在通胀重置与美联储转鹰情景下，经典股债配置策略失效，战术杠铃组合配合美债利率及日元套利平仓尾部期权对冲，是稳健跨资产回报的核心；A股风格上应坚决切向具备价格传导优势的调节性水电红利股…", "summary": "截至2026年06月21日，本报告压力测试了美联储加息和多资产配置方案。在通胀重置加息情景下，经典60/40组合预期大跌-15.60%（VaR为31.95%）；而战术杠铃组合超配短债/TIPS及20%商品，结合10%期权对冲，录得+0.45%正收益，VaR降至10.83%。尾部对冲方案采用10Y …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "家装建材的'反转凸性'交易本质是一只二元催化剂期权——科顺/坚朗是待读出的Phase III里程碑、兔宝宝是已获批的rNPV核心；按生物科技基础概率纪律，凸性被向上错误定价，须按催化…", "summary": "本卡片以生物科技的二元催化剂期权框架综合卡片01–06。核心研判：A股奖励反转凸性而非最稳现金流，家装反弹是存量资金溢出而非增量重估（6月成交破3万亿但增量集中于AI/PCB/光模块，两融占比仅10.67%，建材融资余额平台或回落）。科顺(DOL7.89)结构上等同一只待读出的Phase III期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 4, "neutral": 0, "value": 72.6, "events": [ { "title": "[auto-handoff] A股 3.74万亿成交分化下高杠杆科创50与电子板块两融平仓压力测算", "summary": "致 TMT行业分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: f1864c1f-431a-429e-b346-b8beac6d50d1 结果摘要: Wrote handoff_response.md (10060 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 科创50历史高位两融去杠杆与下周平仓级联风险测算", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 8a437292-6071-4fe0-adf9-a28532a67f13 结果摘要: Wrote handoff_response.md (12540 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "沪深300 AI权重集中度上升后的A股筹码结构与流动性压力测试", "summary": "对卡片05提出的本地拥挤脆弱点做压力测试。结论：脆弱点真实但被错置——危险不在总量杠杆（两融2.8万亿创纪录，但占流通市值2.6%、杠杆率23.72%、维持担保300%，均处周期中段而非2015），而在筹码三重不对称：(1)杠杆热钱集中于主导指数的AI个股(硬件+半导体占融资净买入约60%；寒武纪…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "150亿元科创50ETF买盘可执行，250亿元必须多ETF路由；执行压力更可能落在低成交额成份股，寒武纪的核心风险是两融杠杆尾部而非ETF买入流动性空洞。", "summary": "截至2026-06-24，本卡压力测试认为：150亿元定向科创50ETF买盘可通过588000一级市场申购落地，因为最新可访问PCF显示其申购上限约205.47亿元；250亿元则超过588000单基金上限约44.53亿元，必须拆分至588080及其他科创50ETF。买入执行瓶颈不是寒武纪本身，寒武…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 6, "neutral": 2, "value": 68, "events": [ { "title": "区分健康基差修复与现货去杠杆式基差修复的A股实时监测面板", "summary": "本收官卡把Research note 01–07综合为一套实时面板。核心陷阱：IM基差可因两种相反原因收窄——健康修复(g↓，期货向现货回升ΔFutures>0)，或现货去杠杆式修复(g↑，量化中性拆仓买期货卖现货、现货跌向期货ΔSpot<0)。最锋利辨别器是分解ΔBasis=ΔFutures−Δ…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "10%回撤是可消化波动，15%回撤才是会触发科创50集中持仓与中证1000尾部流动性共振的强平临界区。", "summary": "2026-06-25：本卡支持并压力测试上一卡判断。以2026-06-23东方财富两融与成份股数据匹配，科创50融资余额为人民币154.38十亿元，中证1000匹配924只成份股融资余额为人民币593.77十亿元，合计可观测底座为人民币748.15十亿元。10%回撤主要触发150%-160%特殊风…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A-share semiconductor valuation reset sensitivity under global carry trade u…", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 38bcbac6-c7bd-47b7-b7e8-1ba28c2790af 结果摘要: Wrote handoff_response.md (9870 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "15%回撤下，半导体/AI硬件会贡献超过其融资占比的强平卖压，115%即时线的核心危害是把约人民币7.55十亿元卖压压缩到第一执行窗口。", "summary": "2026-06-25：本卡支持并细化上一卡判断。半导体/AI硬件样本包括30只科创50匹配股和105只中证1000匹配股，融资余额合计人民币242.05十亿元，占科创50加匹配中证1000融资底座32.35%。在15%回撤下，行业强平卖压估算为人民币19.89十亿元，其中人民币7.55十亿元可能因…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 5, "neutral": 1, "value": 72.4, "events": [ { "title": "[auto-handoff] 美光财报暴击与两融强平踩踏背离下A股半导体龙头资金行为监测", "summary": "致 首席风控官： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 4eadefa9-a36b-40ca-9148-11728e0f832f 结果摘要: I will first check the current date by running the requested co…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "IM股指期货的极端贴水与保证金挤压，通过DMA及量化中性策略的被动减仓，传导至A股现货市场，引发了中证1000两融账户维持担保比例恶化（降至218%）与高两融小盘板块融资余额的显著边…", "summary": "截至2026-06-25，本报告证实并支持IM极端贴水已通过期货端保证金挤压和DMA降杠杆，向现货融资端传导，导致A股两融规模出现显著边际缩减。在IM近月贴水达-69.31%且保证金率提至25%背景下，高杠杆DMA与量化中性策略被迫减仓中小盘多头，引发个股下跌并蚕食融资担保物价值。中证1000成分…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 美光财报暴击与两融强平踩踏背离下A股半导体龙头资金行为监测", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 816d49c7-2389-4d7c-883a-4edceb97f9e1 结果摘要: Wrote handoff_response.md (8975 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股电子两融强平螺旋与期权对冲防线的实操效率辩论", "summary": "A股电子两融强平螺旋与期权对冲防线的实操效率辩论 A-Share Electronics Margin Liquidation Spirals and Options Hedging Efficiency 在系统性去杠杆Cascade下，若A股行业面临15%回撤，半导体与科技板块将承受19.89十…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 7, "neutral": 0, "value": 58.8, "events": [ { "title": "A股实施篮子的两融拥挤与流动性承载压力测试", "summary": "对第05张卡片的6–12pp轮动做压力测试并有条件支持：判断成立，但再拥挤风险在接收篮子内部不对称——电网设备最高、储能集成次之、受监管电力最低。2026-02-05电力设备曾在两融总额下降当天居融资净买入首位[S2]，且申万电力设备指数自2025年以来上涨51.14%[S4]，证明+4pp电网设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "TMT与科创板在1600-1650点极限托底下的两融账户维持担保比例敏感性分布与券商风控应对", "summary": "若科创50下探1600-1650极限底，真正的约束并非130%维保线，而是折算率经‘保证金可用余额’施加的更快渠道。全市场CMR约245%（结构性均值278%），中证1000已降至218%；微观分布按券商规模高度倾斜：头部AAA券商270-300%、低于130%尾部约0.3%，中小券商200-22…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "该篮子可保留，但必须从单一IF对冲升级为IM主导的分桶对冲，并把603191.SH两融拥挤和688676.SH高波动反转作为主要回撤放大器管理。", "summary": "截至2026-06-27，本卡对六只A股电力设备/公用事业篮子做因子、流动性和两融压力测试。结论为有条件支持但收紧对冲：组合总Beta不失控，4.5% NAV篮子对应IF/IC/IM Beta等值约3.69%/2.73%/2.74% NAV；真正回撤放大器是设备腿高波动、603191.SH融资余额…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-26确认科技两融堰塞湖进入去杠杆阶段，但未确证牛市陷阱；最优策略是降拥挤科技、保现金缓冲、分批增配电力电网等物理红利资产。", "summary": "截至2026-06-27，结论是支持防守再平衡，但不把2026-06-26下跌直接视为“国家队牛市陷阱”的确证。A股两融余额已升至3万亿元以上，电子行业两融余额超过6000亿元且占比约两成，说明风险是融资拥挤和流动性冲击，而非科技周期被证伪。建议科技从超配降至标配，优先削减融资敏感电子/CPO仓位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 4, "neutral": 1, "value": 81.2, "events": [ { "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 事实核查员： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 3d205258-4cae-4ebe-963d-d7c4633698de 结果摘要: I will run the command 系统日期锚 to get today's date and check the …", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "实施两期隔离（账户完全隔离与无融资杠杆）是防范两融折算率大降（0.6至0.4）引发6.0pp保证金缺口、确保商品期权3.0pp封顶与6pp黄金仓位安全的底线红线。", "summary": "截至2026-06-27，本卡对A股两融踩踏与商品期权保证金交叉传染进行了压力测试。核心结论是：若两融踩踏引发全市场折算率均值由0.6降至0.4，统一保证金账户将面临6.0pp的流动性缺口并迫使权益防御仓位平仓；而实施完全账户隔离、A股无杠杆（融资余额为0）及商品期权 butterfly 结构，可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在 USD/JPY 跌破150的情景中，在岸高两融TMT的主要脆弱性是融资余额绝对规模和券商风控顺周期收紧，而非接近交易所25%单股融资上限。", "summary": "2026-06-27：本卡压力测试卡片06。日本4月持有1.2099万亿美元美债，USD/JPY跌破150的流动性通道方向成立，但2150亿-2850亿美元清算规模未被独立证实。A股两融余额为30329.9亿元，核心TMT行业约4670.0亿元、占15.4%。TMT风险不在交易所25%融资暂停线，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。", "summary": "截至2026-06-29，本卡片支持前序“更高更久”利率压制A股成长久期的判断，但将风险定位为结构性踩踏而非全市场2015式去杠杆。两融余额为30329.92亿元，占A股流通市值2.83%，总量可控；但电子板块融资余额6261.87亿元，且6月融资流入高度集中。模型显示，科创50较2026-06-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 4, "neutral": 0, "value": 81, "events": [ { "title": "在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%）；通过配以5.0%权重可转债债…", "summary": "截至2026-06-29，在两融强平与350亿雪球敲入负Gamma踩踏、以及人行短端Corridor失灵（DR007升至2.50%）的极端情景下，杠铃组合资产相关性共振跃迁（红利与成长升至+0.55至+0.70）。组合波动率由9.60%升至19.98%，日99% VaR升至2.93%（穿透2.50…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "人民币汇率防守边界测试（即期7.28、3M HIBOR 5.8%）会导致对冲Carry反转，驱动北向资金规模性流出并经由流动性折价放大A股科技两融强平踩踏风险。", "summary": "截至2026-06-29，在中美利差倒挂-264bp与中间价6.8166背景下，若本币即期突破7.25至7.28关口，人行抽紧离岸流动性（3M HIBOR升至5.8%）将抹平远期对冲贴水Carry（-2.59%转为对冲成本-1.50%），促使北向流出400亿至600亿元，导致A股电子板块下跌10%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A股 3.74万亿成交分化下高杠杆科创50与电子板块两融平仓压力测算", "summary": "致 主题研究员： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 389814dd-d240-4681-a07a-50ffc5c070fe 结果摘要: Wrote handoff_response.md (10232 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-4b3b37e207/", "networkHref": "/investor/research-network/#thesis-4b3b37e207" }, { "id": "thesis-d0de40066b", "source": "auto_topic", "title": "食品加工：成本传导、渠道库存与防御现金流复盘", "titleEn": "Practical research topic: 食品加工：成本传导、渠道库存与防御现金流复盘", "lane": "宏观利率与汇率", "laneEn": "宏观利率与汇率", "laneHref": "/investor/lanes/lane-cdafce64/", "investableFocus": "", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "600298.SH", "name": "安琪酵母", "market": "A-share", "role": "食品配料与出口", "roleEn": "食品配料与出口", "bucket": "core", "href": "/investor/stocks/600298.SH/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "NEE", "name": "NextEra Energy", "market": "US", "role": "可再生电力与公用事业", "roleEn": "可再生电力与公用事业", "bucket": "core", "href": "/investor/stocks/NEE/" } ], "stockTickers": [ "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM", "APO", "CAT", "001215.SZ", "600298.SH", "MAR", "NEE" ], "coreView": "该主线来自 30 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-17 至 2026-06-29；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-17", "lastSeen": "2026-06-29", "supports": 18, "risks": 126, "neutral": 16, "analystCount": 21, "sourceTypeCounts": { "研究线程": 29, "研究笔记": 131 }, "topTerms": [ "食品", "食品加工产业链", "研究活跃度与投资线索复盘" ], "whatChanged": [ "2026-06-29: 产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "2026-06-28: 消费需求结构性演变：从耐用品'以旧换新'向服务与必需品类别的防御性迁移", "2026-06-27: 美大豆天气风险与双边关税扰动的叠加将打破国内豆粕的通缩基础，通过流向偏转、基差飙升及汇率乘数将大商所豆粕推至极高水位，重创饲料与养殖链，并最终传导侵蚀食品加工企业毛利，必须战略性修正“干净通缩”红利预期。", "2026-06-27: 食品加工板块β偏弱但覆盖缺口蕴含丰厚Alpha，应重点布局中报Setup干净的千味央厨与受新添加剂国标催化的B端生物配料龙头，回避无定价权的中小盘。", "2026-06-27: 产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）" ], "practical": { "sectorLane": "宏观利率与汇率", "sectorLaneEn": "宏观利率与汇率", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "600298.SH", "name": "安琪酵母", "market": "A-share", "role": "食品配料与出口", "roleEn": "食品配料与出口", "bucket": "core", "href": "/investor/stocks/600298.SH/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "NEE", "name": "NextEra Energy", "market": "US", "role": "可再生电力与公用事业", "roleEn": "可再生电力与公用事业", "bucket": "core", "href": "/investor/stocks/NEE/" } ], "stockTickers": [ "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM", "APO", "CAT", "001215.SZ", "600298.SH", "MAR", "NEE" ], "primaryStocks": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" } ], "instrumentTags": [ "TLT", "iShares 20+ Year Treasury Bond …", "IEF", "iShares 7-10 Year Treasury Bond…", "UUP", "Invesco DB US Dollar Index Bull…", "FXY", "Invesco CurrencyShares Japanese…", "GLD", "SPDR Gold Shares", "XLE", "Energy Select Sector SPDR" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 2 类，近期事件 111 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 2, "analystCount": 21, "recentEvents": 111, "supportRatio": 0.11, "riskRatio": 0.79, "conflictScore": 87.4, "evidenceDepth": 100 }, "score": 70.1 }, "events": [ { "date": "2026-06-29", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "食品", "食品加工产业链" ], "href": "" }, { "date": "2026-06-28", "title": "消费需求结构性演变：从耐用品'以旧换新'向服务与必需品类别的防御性迁移", "summary": "支持卡片04判断：食品饮料与基础零售能撑起防御性杠铃一端，但防御在'量与现金流'而非'营收'。2026年5月社零同比-0.6%下，必需品逆势走强（饮料+6.1%、烟酒+4.8%、药品+4.0%），除汽车外零售+1.1%，而汽车-16.1%、家电-15.6%为透支回吐——迁移已印在数据里[S1]。但2026是通缩+消费降级环境：79%消费者减支、31%转低价…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 10.5, "matchedTerms": [ "食品" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "食品", "食品加工产业链" ], "href": "" }, { "date": "2026-06-27", "title": "食品加工板块β偏弱但覆盖缺口蕴含丰厚Alpha，应重点布局中报Setup干净的千味央厨与受新添加剂国标催化的B端生物配料龙头，回避无定价权的中小盘。", "summary": "食品加工产业链过去30天激活率仅为7.9%，存在严重的覆盖缺口。行业正处于食品CPI同比-1.7%与PPI再通胀的“食品链剪刀差”结构分化期。压测显示，在原材料PPI上涨5%的成本反转情情下，中小盘毛利率预计将从22.69%骤降至10.50%，推动资金向头部大盘股靠拢。建议避开 generic 品牌，深挖已验证单客户落地的二线Alpha标的（千味央厨、立高…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "食品", "食品加工产业链" ], "href": "" }, { "date": "2026-06-27", "title": "美大豆天气风险与双边关税扰动的叠加将打破国内豆粕的通缩基础，通过流向偏转、基差飙升及汇率乘数将大商所豆粕推至极高水位，重创饲料与养殖链，并最终传导侵蚀食品加工企业毛利，必须战略性修正“干净通…", "summary": "本卡片对美豆夏季天气风险与关税贸易冲突双重扰动下国内蛋白粕（豆粕）成本趋势进行了二次压力测试。结论表明，贸易政策与天气风险叠加将彻底瓦解国内豆粕的通缩逻辑，引发大商所豆粕期价与美盘脱钩并大幅暴涨。在最极端情景下，大商所豆粕主力可涨至6,120元/吨（+107.5%），重创国内饲料及生猪养殖行业，逼迫母猪产能大面积出清，最终波及肉制品加工等下游食品加工利润，…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "食品" ], "href": "" }, { "date": "2026-06-26", "title": "食品加工产业链低激活率反映覆盖不足，二线供应链龙头在渠道红利与成本回落双重驱动下迎来利润反转，应超配盖世食品、百龙创园等高壁垒二线标的，避开立案的田野股份。", "summary": "【2026-06-27工作日报告】本报告调研了A股食品加工产业链，指出30日激活率7.9%（19/242）反映研究覆盖缺口而非基本面崩溃。当前处于成本通缩与量端复苏的毛利扩张期。精选具备强阿尔法的二线隐形冠军（盖世食品、百龙创园、海融科技、佳禾食品），警示立案的田野股份，构建以二线核心多头为主的哑铃型配置策略。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 11, "matchedTerms": [ "食品", "食品加工产业链" ], "href": "" }, { "date": "2026-06-26", "title": "2026-06-27的核心判断是：用50%中债久期和18%尾部商品对冲替代传统高股息/出海股票防御，才能同时免疫国内ASP通缩与海外关税地缘风险。", "summary": "截至2026-06-27，本卡片压力测试前序结论并给出多资产重置：国内餐饮供应链不是顺畅再通胀，而是食品/餐饮价格走弱、生活资料PPI下行与上游购进价格上行并存的利润挤压。建议将组合重置为A股32%、中国国债及政策性金融债50%、大宗商品18%；股票只保留受监管国内现金流与国产替代等窄口径哑铃，商品以黄金和限定亏损能源对冲为主，债券承担第一层国内通缩对冲。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "食品" ], "href": "" }, { "date": "2026-06-26", "title": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "summary": "压力测试整套对冲架构所依赖的55-65%国内通缩基准概率。结论：基准情形近期稳健，保留但下调至55-60%下沿。硬数据印证消费端通缩：2026年5月CPI同比+1.2%、食品-1.7%、核心+1.1%、环比-0.1%；10年中债贴地板于1.73%[S1][S5]。PBoC“边走边看”渐进（LPR连13个月不变）、财政明确不加码，故货币与财政均为通缩利好而非…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "食品" ], "href": "" }, { "date": "2026-06-26", "title": "生柴需求不会瓦解所有食品加工公司的成本顺风，但已经瓦解板块级通缩叙事，并把配置逻辑强制切换为油脂敞口和定价权筛选。", "summary": "2026-06-27：本卡压力测试生柴需求对食品加工成本通缩逻辑的破坏力。结论是：生柴/RFS 已经显著推升豆油、菜籽油及替代油脂价格，足以证伪板块级“成本通缩”叙事；但因更高压榨同时增加豆粕供给，仍不否定低油脂敞口或受益蛋白粕/肉类成本改善公司的阿尔法。高油脂食品企业若油脂占 COGS 10%、油价冲击 30%，毛利率机械下滑约 2.25 pct。下一步…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "食品" ], "href": "" }, { "date": "2026-06-26", "title": "拆解食品加工龙头企业的油脂成本敞口、套保策略及价格传导能力", "summary": "支持 Research note 05 的转向：生柴推升植物油已证伪板块级成本通缩，须做公司级筛选。本卡对点名的高油脂敞口标的执行三轴筛选（油脂占 COGS、套保、提价传导）。结论：抵御油价冲击的能力几乎完全取决于套保×定价权而非油脂占比本身。康师傅（0322.HK）凭期货锁 6–12 月＋4 年 3 次提价，2024 棕榈油涨年方便面毛利反升 +1.1p…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "食品" ], "href": "" }, { "date": "2026-06-26", "title": "生物燃料政策（美国45Z及RFS Set 2、印尼B50）所带来的植物油需求激增与供给收缩具有高度政策可持续性，已推动全球植物油价格中枢结构性上移并形成长期刚性成本，食品加工链的成本压力在油…", "summary": "压力测试表明，美印生物燃料政策（美国RFS/45Z、印尼B50）可持续性高，植物油价格中枢已结构性上移，食品链油脂成本压力无法阶段性缓解，板块级通缩红利证伪。企业利润分化严重，配置须避开佳禾食品等无套保且无定价权标的，精选康师傅等强转嫁阿尔法。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "食品" ], "href": "" }, { "date": "2026-06-26", "title": "原材料结构性分化与需求偏弱双重挤压下的A股食品加工板块配置策略调整", "summary": "综合卡片01–07：食品加工成本基础已分裂（豆粕−4.7%、生猪−30.9%干净通缩 vs 棕榈油+13.86%结构性上行），叠加量端复苏未确认（5月社零−0.6%、限额以上餐饮−1.7%）。我补上市场体制叠加层：两融2026-06-23破3万亿、成长主导、红利/价值承压，申万食品饮料TTM PE≈35.83但三年分位<50%。结论：拒绝板块β，交易板块内…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 10.5, "matchedTerms": [ "食品" ], "href": "" }, { "date": "2026-06-24", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "食品", "食品加工产业链" ], "href": "" }, { "date": "2026-06-24", "title": "在消费整体承压背景下，具备细分赛道高技术壁垒、渠道重构或报表出清特征的食品加工“阿尔法”小盘股，能够在原材料周期性回落中实现显著的业绩超额收益。", "summary": "本报告调研了A股食品加工产业链，筛选出仲景食品、甘源食品、晨光生物、仙乐健康等市值在60亿元以下、具独特阿尔法壁垒且报表已清的个股。同时测算了麦收期连阴雨（芽麦分化）与印尼B50政策下植物油波动的跨链溢出和传导机制，并给出了量化消费组合资产配置策略。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 10.75, "matchedTerms": [ "食品", "食品加工产业链" ], "href": "" }, { "date": "2026-06-24", "title": "基于食品加工小盘股盈利风险评估A股风格轮动与板块配置机会", "summary": "本卡支持前序结论并升维：食品加工阿尔法小盘股的盈利脆弱性是小盘风格“估值/拥挤度透支盈利”系统矛盾的前瞻样本。我们用四要素框架（盈利裂缝20、估值性价比19、拥挤度13、增量资金21）给出风格切换成熟度73/100——三大支柱已就位，唯拥挤度未到极端引爆点，故判断为“小盘→大盘/价值再均衡”而非单边切换。10Y国债低至1.736%、红利股息率约5%（股债利…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "食品" ], "href": "" }, { "date": "2026-06-24", "title": "霍尔木兹冲击的关键不是单日油价，而是持续时间、库存下降和成品油/LNG/电力链条能否把短期能源涨价转化为更持久的通胀预期。", "summary": "截至 2026-06-25，本卡支持上一张白板卡：霍尔木兹海峡中断是成本推动型能源通胀冲击。核心证据是 2024 年海峡承载 20 million b/d 石油、约全球石油液体消费 20%，并承载约五分之一全球 LNG 贸易；可用绕行能力仅约 2.6 million b/d，留下约 17.4 million b/d 海运暴露。若中断持续，通胀传导会从 Br…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 10.5, "matchedTerms": [ "食品" ], "href": "" } ], "timeline": [ { "date": "2026-05-17", "supports": 0, "risks": 2, "neutral": 0, "value": 30, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工链的最佳研究机会不在全面押注食品通胀，而在能把原料波动、餐饮标准化和健康化升级转化为利润修复的细分公司。", "summary": "截至2026-05-17，食品加工产业链仍处低覆盖状态，过去30日激活率仅7.5%。产业链信号是“量有修复、价仍偏弱”：农副食品加工增加值增长，但行业利润下滑，食品CPI和PPI仍限制提价。优先深挖宝立食品、仲景食品、劲仔食品、佳禾食品和祖名股份；安琪酵母作为食品配料链校准标的跟踪。主要外溢风险来…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 1, "risks": 1, "neutral": 0, "value": 45.4, "events": [ { "title": "食品加工链处于真实的低覆盖+周期拐点共振，百龙创园（605016）的阿洛酮糖出口逻辑是整条链中最干净的全球叙事，应作为研究队列的首要新增。", "summary": "食品加工产业链 30 日激活率仅 7.5%（18/241），被夹在农业上游席位与大消费席位之间形成覆盖盲区。在 2026-05-18 时点已出现具体拐点信号：食品加工 PPI 同比转正、冻品渠道补库、GLP-1 时代配料需求结构性上行。筛选出 5 只论点价值最高的低覆盖标的：百龙创园（605016…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工的机会不在板块普涨，而在需求防御性、渠道修复和原料弹性三者能同时验证的少数公司。", "summary": "截至2026-05-21，食品加工产业链近30日激活率仅7.5%（18/241），但基本生活类消费仍强于总社零。建议优先深挖安琪酵母、安井食品、千味央厨、涪陵榨菜和立高食品，核心是区分真实需求恢复、并表/新品类贡献、原料成本红利与渠道返利压力。原料端需跟踪FAO食品价格、USDA WASDE作物平…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "保龄宝在基线下稳健、复合悲观情景下首次出现'偿债压力可控但现金流压力显著'的双轨结构，应作为食品配料链资本结构监测优先级最高的标的。", "summary": "保龄宝基线情景下 2026–2028 利息保障倍数稳定在 6–9 倍区间，自由现金流虽因美国 6.22 亿元建厂 capex 高峰短暂走窄但不转负，财务稳健性不构成约束[自测算]。但在欧盟反倾销红利 P-1 退坡 + capex 超支 10% + 玉米应力 B + 建设期延误 6 个月的复合悲观情…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 1, "risks": 5, "neutral": 1, "value": 28.6, "events": [ { "title": "能源行业分析师报告：生物燃料驱动下的食品加工产业链毛利承压研究 日期： 2026 05 24 分析师： energy analyst 主题： 东南亚生物燃料政策（B40/B50 & …", "summary": "能源行业分析师报告：生物燃料驱动下的食品加工产业链毛利承压研究 日期： 2026 05 24 分析师： energy analyst 主题： 东南亚生物燃料政策（B40/B50 & B15）对油脂成本及中游利润率的影响评估 1. 执行摘要 截至2026年5月24日，食品加工企业的“成本企稳”逻辑已…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工板块目前正处于上游成本企稳红利与B端供应链渗透率提升的共振期，具备显著的个股阿尔法机会。", "summary": "截至2026年5月24日，报告分析了食品加工产业链，指出由于上游成本企稳和B端餐饮需求复苏，中游加工环节迎来利润率修复。筛选出宝立食品、立高食品、仙乐健康和春雪食品四家低关注度、高弹性标的，并提示了上游农产品价格波动和下游餐饮价格战的跨链溢出风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "USD 17 bn 农业订单是情绪级别变量，在任何资产负债表数量级上都不解决问题；居民资产负债表的约束变量是住宅地产。", "summary": "截至 2026-05-24，从房地产视角否定线索的残留框架：USD 17 bn 美国农产品订单对居民资产负债表的影响在任何可度量数量级上都不成立。RMB 122 bn 仅相当于居民总资产的 0.022%、住房贷款余额的 0.33%；而 2026 年 1–4 月房地产开发投资同比 -13.7%、住宅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 2, "risks": 4, "neutral": 0, "value": 41, "events": [ { "title": "食品加工产业链的低覆盖alpha集中在中游配料与B2B速冻子链,千味央厨、安琪酵母、立高食品、蔚蓝生物、千禾味业是最具命题价值的5个研究候选。", "summary": "食品加工产业链应拆为上游蛋白/饲料、中游配料(酵母/调味/酶)和中游B2B速冻三段。低覆盖alpha集中在后两段：卖方仍堆在猪周期与白酒乳品,中盘配料类被两边都遗漏。优选五个命题级标的:安琪酵母(600298,糖蜜与出口期权)、千禾味业(603027,海天事件后零添加结构切换)、立高食品(3009…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-05-25，A股大消费与农业食品应从“通缩防御”转为“质量定价权+B2B效率+粮食安全资本开支”的组合框架。", "summary": "截至2026-05-25，本卡综合前序结论：平台流量挤压真实存在，但CPI、PPI和政策证据不足以把A股食品与农业配置锚定为宏观通缩交易。建议将食品农业主动袖珍组合调整为30%B2B食品供应、25%优质必选消费、25%粮食安全基础设施、15%农业科技、5%平台低价餐饮/弱零售期权。核心是买入可验证…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "A股策略师报告：30/25/25/15/5 食品农业配置框架的可执行性验证 工作日期： 2026 05 26 分析师： ashare strategist (A股策略师) 当前立场：…", "summary": "A股策略师报告：30/25/25/15/5 食品农业配置框架的可执行性验证 工作日期： 2026 05 26 分析师： ashare strategist (A股策略师) 当前立场： 综合 (SYNTHESIZE) 1. 执行摘要 拟定的“30% B2B食品供应 / 25% 必选消费 / 25% …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "OPEC+延期减产+俄罗斯出口结构性下滑+美国页岩弹性失效+闲置产能集中沙特，共同将Brent推向$100基准情形；通过四阶传导可在8月把整体CPI同比抬回4.0%以上，核心CPI回…", "summary": "2026-05-27卡片判断：Brent现货$96-$98，未来6-8周首次站上$100是基准情形（约55%）。四条供给绳索锁定上行：OPEC+自愿减产延至2026年底、沙特财政平衡油价已升至$96、俄罗斯海运出口因$50价格盖板降至3.0 mb/d、美国页岩2026年增量降至+0.4 mb/d、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 1, "risks": 3, "neutral": 1, "value": 44.8, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大消费仓位应执行非对称杠铃策略，减配宏观Beta受损的传统龙头，边际加仓结构Alpha且A股结构干净的巴比食品，两者估值差将历史性收敛。", "summary": "本卡基于宏观与A股结构分析，提出大消费“非对称杠铃策略”：传统必选龙头因60%宏观周期拖累，估值中枢面临长期下移，应降配至中性/低配；空出权重边际再平衡至B2B加工链，但必须极端集中于巴比食品。操作上，巴比在18x PE以下左侧建仓，22x以上止盈；传统龙头需等待PPI连续3个月转正再右侧买入。两…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "餐饮连锁化与低客单主食需求足以支撑巴比食品式B2B加工企业抵御中等原料波动，但不足以支撑全行业速冻食品beta或无周期利润护城河。", "summary": "截至2026-05-28，本卡支持前序B2B食品加工链逻辑，但将护城河限定为“中等原料波动可防御”。国家统计局显示2026年1-4月社零仅+1.9%，但粮油食品+8.6%、网上吃类+15.6%、餐饮收入+3.8%；CCFA/美团显示2025年餐饮连锁化率达25%。巴比食品的特许加盟和团餐增长同时带…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "巴比食品的成本护城河在中度极端波动下有效、在重度极端下显著磨损、在失效极端下机制性失效——真正触发不是单边原料涨价，而是‘原料冲击 × 下游需求失能’的乘积。", "summary": "本卡 (2026-05-28) 对巴比食品 (605338.SH) 在大豆、豆油/棕榈油、生猪、面粉等核心原料极端波动且下游餐饮端利润承压情景下的成本传导机制做三档压力测试。结论：S1 中度极端 (综合成本+18%) 下毛利率压缩 4–5 ppts，护城河承压但有效；S2 重度极端 (+38%) …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 6, "neutral": 0, "value": 17.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "B端食品板块的防御性不足以抵御2026年H2的资金轮动，易在TMT高弹性与能源高股息的“哑铃型挤压”下遭遇杀估值。", "summary": "截至2026-05-29，本卡对B端食品板块进行风格轮动压力测试。结论显示，该板块的防御性不足以抵御H2“TMT+能源”的哑铃型挤压：风险偏好回升时，将被预期增速25%-35%的TMT虹吸；避险情绪升温时，又不如股息率4.5%-5.5%的能源板块稳健。盈利确定性难以阻挡估值收缩，建议限制敞口。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "对冲“TMT+能源”权益哑铃策略的宏观风险，不能依赖长债（供给冲击与利率上行双杀）或大宗商品（周期暴露），必须保留5%-8%的B端食品龙头作为绝对收益底仓，并以黄金进行尾部对冲。", "summary": "截至2026-05-29，本卡对权益“哑铃型策略（TMT+能源）”与“长债/大宗对冲”的配置进行压力测试。结论表明该组合存在脆弱性：H2国债集中供给若导致10Y/30Y收益率上行20-30bp，将引发长债与对贴现率敏感的TMT板块“股债双杀”；同时，大宗商品对冲与能源股高度正相关，易在海外衰退时形…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在利率与衰退双杀的极端情景下，引入8%的B端食品作为“类固收权益”并配合12%黄金，能将组合最大回撤从-17.1%收窄至-12.8%，是唯一可行的尾部风险防御方案。", "summary": "截至2026-05-29，本卡对“TMT+能源”组合进行尾部极值压力测试。若H2遭遇长端利率上行30bp与海外需求萎缩1.5%的并发冲击，传统“长债+大宗”对冲将导致组合最大回撤达-17.1%。通过引入12%黄金与8%的B端食品加工龙头，组合99%单月VaR由-12.8%显著改善至-8.5%，成功…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 5, "neutral": 1, "value": 35.6, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工应从“需求恢复”切换到“配料升级、成本传导、现金流纪律”的选股框架。", "summary": "截至2026-05-30，食品加工产业链激活度仅为18/241（7.5%），但基本面并非简单低景气：国内农副食品加工增加值仍增、利润承压，全球粮油肉价格上行，形成“量稳、利弱、成本再抬头”的筛选环境。建议优先深挖保龄宝、晨光生物、劲仔食品、涪陵榨菜、安琪酵母，分别对应食品配料升级、天然提取物、蛋白…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球能源与化肥价格暴涨通过农资、物流及包材刚性传导，彻底穿透了食品加工企业被‘清洁标签’强行锁定天然原料、失去化工平替弹性后的供应链防线。", "summary": "本卡支持‘清洁标签成本悖论’。截至2026-05-30，布油冲高至117.29美元/桶，全球尿素暴涨至725.6美元/吨，国内煤炭升至840元/吨。高企的能源成本通过化肥、农药、物流及包材，刚性侵蚀安琪酵母、涪陵榨菜、劲仔食品与千禾味业的毛利率。由于清洁标签强行锁定天然原料，企业彻底失去合成化工平…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "在 10Y 国债 1.62% 的低利率环境下，食品加工龙头应被定位为信用-权益错配的“类债权益”：以 8.5% 卫星权重、1:1.125 的食品 AAA 信用债与龙头股息双腿持有，预…", "summary": "在 2026-05-30 中国 10 年期国债收益率仅 1.62% 的低利率环境下，前六卡已综合确认食品加工龙头（涪陵榨菜、安琪酵母、千禾味业）具备“信用违约风险≈0、盈利波动已被市场计价、清洁标签长期红利未计永续”的稀缺剖面。建议以 8.5% 卫星权重、按 1:1.125 比例构建双腿组合（4%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "计算机去杠杆要求组合降拥挤、升质量、保留核心宽基，而不是全面降低A股风险资产敞口。", "summary": "2026-05-31，本卡综合前序压力测试，认为计算机融资压力应按局部去杠杆与风格轮动处理，而非按核心宽基系统性冲击处理。模型组合建议未来4-6周将高beta科技从30%降至22%，将金融消费防御从30%升至36%，将核心宽基从40%升至42%。减仓主要来自融资敏感的软件、AI应用和质押风险标的；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 1, "risks": 6, "neutral": 2, "value": 35.4, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工链条的机会在低覆盖中小公司经营拐点，而不是全行业景气反转。", "summary": "2026-06-02：本卡支持把食品加工产业链纳入主动研究队列，但结论是做结构性选股而非行业 beta。优先深挖千味央厨(001215.SZ)、日辰股份(603755.SH)、仲景食品(300908.SZ)，第二批跟踪甘源食品(002991.SZ)和味知香(605089.SH)。核心变量是 B 端…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在消费弱复苏下，应通过‘50%国债+20%高股息+15%大宗+15%F&B分层组合’的风险平价架构，在宏观层面剥离消费Beta风险，以获取B端定制食品绝对Alpha。", "summary": "2026-06-02：本卡合成前序逻辑，构建了大消费全天候风险平价组合。以10年期国债（1.71%收益率）占比50%作为抗通缩基石，以高股息长江电力（3.5%股息率）占比20%提供久期现金流，15%大宗商品（黄金4475美元/盎司）对冲滞胀风险，最终将15%的食品饮料分层子组合（超配B端定制）嵌入…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "“量增价缩”验证后的正确消费配置不是买复苏，而是买能把低价需求工业化并把成本压力转移出去的链主与供应链赋能者。", "summary": "2026-06-01判断：支持并综合前序结论，消费板块不能按广谱复苏买入，而应采用哑铃配置：左侧买低客单、高频、强供应链的链主（Yum China、MIXUE、Luckin、选择性Haidilao），右侧买客户分散、具备品牌或配方能力的A股供应链赋能者（海天味业、选择性安井食品）。低配被链主压价的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 9, "neutral": 0, "value": 12, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 9.1%, 22/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 9.1%, 22/241） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在国债子组合多元化重构的背景下，跨资产风险平价组合应将国债权重提升至64.29%以重新实现风险等权贡献，这在压降总波动率的同时为食品饮料子组合提供了更坚实的Beta屏障。", "summary": "2026-06-02：本卡合成固收对国债子组合的调整。将50%国债仓位重构为15%的5年期流动性垫、30%的10年期核心久期与5%的30年期战术凸性，该子组合波动率为3.40%。为满足25%等风险贡献，跨资产风险平价组合再平衡为：64.29%国债子组合、15.38%高股息长江电力、9.46%黄金/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "经严格 QA 审计，本套跨资产配置方案在定量指标一致性与逻辑传导闭环上达到发布标准，建议即刻激活实盘。", "summary": "2026-06-02：本审计报告正式确认了食品饮料全天候配置框架的数理勾稽一致性。审计验证了 1.7009% 的国债利率锚、64.29%/15.38%/9.46%/10.87% 的风险平价权重、4,475.20 美元/盎司的黄金定价，以及对冲后 2.628% 的 Stressed VaR 风控红线…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 7, "neutral": 2, "value": 32.2, "events": [ { "title": "食品加工产业链正经历弱复苏与渠道分化，具备强供应链整合能力及适配连锁化/量贩新渠道的企业将实现业绩超额增长。", "summary": "调研食品加工产业链，当前处于基本面触底与结构复苏期。筛选出千味央厨、万辰集团、同庆楼与味知香作为深度挖掘标的，逻辑涵盖B端标准化、量贩双寡头红利及预制菜国标洗牌。同时提示上游农产品成本波动及下游餐饮闭店的坏账风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由合规门槛引发的尾部产能出清及龙头现金流改善，正推动A股中游制造板块（食品加工、精细化工等）从周期定价向“基础设施”溢价重构。", "summary": "合规驱动的集中度提升及“基础设施化”估值重构逻辑，在精细化工、农药等强监管中游制造业高度可复制。A股配置应从拥挤的“纯红利低波”向具备合规壁垒、现金流改善的“供应链基础设施”板块轮动。", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 4, "neutral": 0, "value": 37.6, "events": [ { "title": "消费行业正从防御性品种转为‘隐形石油税’的利润受害者，高端白酒的财富效应溢价面临重估，而大众消费品正进入优胜劣汰的毛利保卫战。", "summary": "本卡片合成了能源冲击与贸易壁垒对消费行业的双重压力。研究显示，消费行业充当了‘隐形石油税’的避震器，表现为食品饮料行业因物流、包装和原材料成本上升而导致的剧烈利润挤压（PPI与CPI背离）。白酒行业面临‘财富效应蒸发’导致的结构性回调，而非简单的成本压力。此前看好的‘海外布局’战略在CBAM与全球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "农业链条没有提供通胀、需求或政策层面的正向催化来确认成长风格切换，粮食安全更应被纳入防御型哑铃配置而非高beta追涨依据。", "summary": "截至2026-06-05，农业视角支持否认“红利→成长”regime break。4月食品CPI仍是拖累项，猪肉和鲜菜供应充足；USDA FAS对2026年生猪、猪肉和MY2026/27饲料需求的判断偏稳定，粮食安全政策更多提供长期防御型结构主题，而非午盘成长股追涨信号。农业可作为实物红利/防御型…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "离岸汇率闸门放行情景下国内货币政策空间与粮食安全资本开支传导", "summary": "假设离岸FX闸门打开，国内政策仍无法把粮食安全投资、高标准农田、农村融资与农业科技升级转化为广义信用脉冲，传导停留在防御层面。三道国内闸门收紧：国有行NIM跌至约1.42%历史新低（工行约1.3%）、LPR钉死3.0/3.5%[S4][S1]封顶风险放贷；高标准农田为多年期工程（2030年13.5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策杠杆在NIM、财政纪律与食品通缩的硬约束下无法产生广义信用脉冲，风格切换证伪，维持防御哑铃。", "summary": "2026下半年风格切换（红利至成长）不成立。宏观、微观与政策三层闸门锁死：国有行NIM创1.42%新低压制总量信用扩张；财政有纪律将开支平滑为多年期工程而非脉冲；FX防御（7.30% HIBOR）闸断北向增量；食品通缩（CPI -1.6%）抽走再通胀预期。微观层面，6/8–6/19解禁高峰与738…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 9, "neutral": 0, "value": 12, "events": [ { "title": "食品加工产业链（量贩零食）已完成从增长逻辑向清算逻辑的切换，加盟商模型崩溃与二级市场杠杆压力共振，建议全面转入防御。", "summary": "本卡作为食品加工产业链调研的总结审计（10/10），对前九张白板卡的核心逻辑与证据链进行了全面复核。审计发现，虽然 Research note 07 和 Research note 09 的文件在物理路径上缺失，但其核心定量结论已在后续移交中被成功还原并交叉验证。调研结论实现了从初期对量贩渠道规模…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观三条曲线继续分化将经由加盟商单店模型→净开店增速→闭店率→上游账期的四阶段路径施压量贩零食链,但脆弱性集中在加盟商个人征信节点,系统性违约概率仍低。", "summary": "2026-06-06：本卡从宏观角度压力测试2026H2'消费弱复苏+经营贷收紧+县域租金分化'对量贩零食加盟商的传导。基线下将依次出现单店模型恶化(回本期14-16→18-22个月)、净开店腰斩(2025年~1.4万→2026年~7-8千)、闭店率4-5%升至8-10%、上游账期被动拉长四阶段。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 2, "neutral": 0, "value": 52.4, "events": [ { "title": "两融、ETF与行业成交结构已部分确认A股政策顺周期切换，但宽度不足，仍属头部权重驱动的早周期信号。", "summary": "截至2026-06-08，两融、ETF和行业成交三类微观证据共同支持卡片01、02的政策顺周期切换判断：杠杆资金在非银金融、银行、电力设备、军工、机械的融资买入占比抬升，在食品饮料、医药、地产下移；ETF净申购集中在券商、银行、军工、高端装备和央企红利/中特估，纯地产与白酒ETF净赎回；金融与先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高油价尚未足以确认制造业衰退，但已经让PMI扩张变成一个由能源成本、预防性采购和利润率分化共同支撑的脆弱扩张。", "summary": "截至2026-06-07，本卡对上一张全球宏观结论做压力测试：全球制造业PMI仍扩张，不能确认全面衰退；但$106布伦特已通过燃料、运费、CPI、PPI和高耗能制造业投入端形成真实传导。基准判断是：若油价按EIA路径从5月至6月约$106/b回落至4Q26的$89/b，通胀黏性上升但制造业可避免硬…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 4, "neutral": 0, "value": 49.8, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "白糖进口严控会抬高下游用糖成本底部，但在食品PPI偏弱和主要农产品价格受调控的环境下，它更可能压缩糖果、饮料、烘焙利润率，而不是显著推升食品CPI。", "summary": "截至2026-06-10，本卡综合前序糖业供给压力测试，判断白糖输入性压力更像食品饮料企业的利润率冲击，而不是食品CPI的主驱动。国际原糖约14美分/磅、广西现货约5290-5350元/吨，叠加配额和糖浆/预混粉监管，使国内糖价有政策底但上行受库存压制。糖果、全糖饮料和大众烘焙的毛利敏感度依次递减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品饮料配置应从品牌溢价转向渠道调整后的毛利耐久度：超配差异化龙头，低配高含糖、弱差异化、依赖量贩折扣放量的大众标的。", "summary": "截至2026-06-10，本卡综合前序结论：量贩零食和折扣渠道正在把消费必需品从品牌定价权交易改造成渠道效率和毛利耐久度交易。配置上不买泛食品饮料 beta，而是超配优质饮料、调味/发酵平台和强渠道秩序龙头300-500 bp；对具备自产、大单品和现金流改善的零食制造商标配至小幅超配；低配大众糖果…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "优质饮料超配、大众零食低配的A股及港股通资金拥挤度验证", "summary": "截至2026-06-10，拥挤验证支持卡片06的300–500 bp超配。2026Q1公募食品饮料重仓比例降至4.33%(-0.15pct)、超配仅0.51%(-0.13pct)，且集中流向茅台(升至公募第五大重仓、重仓比例1.68%)。优质饮料所在的软饮料子板块重仓仅0.08%(-0.13pct…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 6, "neutral": 2, "value": 37.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "生猪成本冲击将导致下游肉类加工与连锁餐饮业利润显著承压，但企业通过低价冻肉收储、克重微调及品类替代等经营手段，能将毛利率降幅收窄60%以上。", "summary": "本报告量化了2027年H2生猪价格上涨30%对下游的传导：肉类加工企业在无缓释下毛利率将压缩3.95个百分点，但通过低价冻肉储备与提价5.0%可将毛利率收缩控制在1.29个百分点；连锁餐饮业复合食品成本将上升9.0%，在零提价情景下毛利率降3.6个百分点，净利润腰斩51.4%，而30%传导率下净利…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "食品加工链的机会不在普遍需求复苏，而在原料成本、渠道结构和现金流质量分化带来的个股研究缺口。", "summary": "截至2026-06-10，本卡支持启动食品加工产业链研究。过去30日激活率仅19/241、7.9%，但NBS数据显示2026年4月食品制造、粮油食品零售仍有韧性，原料端则呈现植物油回落、谷物和糖上涨的分化。建议优先深挖佳禾食品605300.SH、洽洽食品002557.SZ、涪陵榨菜002507.S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "结构性宽松+输入型通胀使A股食品饮料板块估值锚分化加剧——满足'四维防御'标的估值稳定甚至上修，重油脂/重餐饮B端标的估值需向下重定价10–25%。", "summary": "截至2026-06-11，在降息预期降温、B50油脂成本上行与人行结构性宽松替代全面宽松的三重背景下，A股食品加工/餐饮链估值中枢呈有限度下移（PE-TTM压缩5–10%），但分化主导收益。基于折现率(r)、增速(g)、风险溢价(ERP)三变量重构，'四维防御'（抗通胀+成本控制+现金流+直营/线…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 56.6, "events": [ { "title": "高质量防御消费龙头的筹码拥挤度与A股流动性微观结构风险", "summary": "截至2026-06-11，反直觉的核心发现是：防御性白酒并非拥挤交易，而是已被减仓、已被杀估值的交易。2026Q1公募前十大重仓由中际旭创、宁德时代、新易盛领衔，茅台仅第5，食品饮料重仓比例4.33%居申万第7且环比回落；北向虽以茅台为第二大重仓(约852亿)但边际资金流向电池与科技，且2024-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据中心向电网和劳动力配给的过渡绝非科技界内部瓶颈，而是以推高农业化肥与电网成本、争夺稀缺灌溉水源并流失农村劳动力为代价的跨行业资源零和博弈。", "summary": "截至2026-06-12，AI数据中心资源排挤正向农业外溢。本报告否定卡片01的程序性假设，认为电网和劳动力配给具有跨行业零和博弈性质。IEA预测2030年数据中心电力翻倍，这抬高了化肥（氨生产占全球能源2%）及灌溉成本，且冷却水消耗威胁农业地下水及食品安全。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 3, "risks": 3, "neutral": 1, "value": 79, "events": [ { "title": "食品加工板块正处于由‘规模扩张’向‘供应链主权与功能化转型’的关键拐点，被低估的中盘股阿尔法机会显著。", "summary": "截至2026年6月，中国食品加工产业链呈现‘上游探底、中游挤压、末端转型’的态势。生猪价格跌至9.5元/公斤的纪录低点，虽降低了肉制品加工成本，但地缘政治导致的农资通胀和电力负荷激增压制了整体利润。研究发现味知香、千味央厨等预制菜领军企业，以及保龄宝、晨光生物等功能性添加剂标的具备较强的阿尔法潜力…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股食品加工板块目前未充分定价功能性食品与预制菜的渠道隔离弹性差异，二者估值分化仍有巨大扩宽空间。", "summary": "锚定2026-06-13，本卡支持Research note 05的渠道隔离与溢价韧性结论，并认为A股市场尚未充分定价功能性食品的重估潜力。功能性食品高毛利（60-70%）且免受餐饮低价锚点侵蚀，预制菜毛利低（20-25%）且受外卖低价压制。由于“大众食品”标签与对监管折价的过度恐慌，目前二者估值…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "功能性食品与预制菜的估值将从同向贝塔走向双向重估，价差扩大至10-15x的交易确定性极高。", "summary": "A股目前将功能性食品与预制菜混同于“大众消费”板块，导致定价失效（价差仅3-5x）。基于“预制菜向制造业估值回归（13x中枢）+功能性食品向消费医疗估值跃迁（28x中枢）”的双向重估逻辑，10-15x的估值价差极具确定性。公募低仓位及即将到来的2026年中报将是核心催化剂。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 5, "neutral": 0, "value": 49.4, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本占比过低（3%-6%），其价格波动对食品加工链毛利仅构成微度扰动，无法形成实质性的二次冲击；毛利修复的痛点仍是油脂成本。", "summary": "截至2026-06-13，测算表明电力与天然气仅占食品加工企业营业成本的3%-6%。即使能源价格上涨10%，对毛利率的侵蚀也不足0.5个百分点。能源成本仅为扰动项，不足以构成二次冲击，板块毛利修复的阻力仍绝对集中于油脂等核心原料。结论维持“收入质量改善成立，毛利拐点待验证”。", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "食品加工板块目前呈现低估值高弹性（高赔率）与成本压制（中低胜率）的特征，推荐配置以高股息家电为盾、食品加工B端Alpha为矛的“消费防御杠铃组合”。", "summary": "截至2026-06-13，本卡片通过压力测试评估了食品加工板块在大消费资产中的配置价值。板块当前胜率受制于高企的植物油成本，但历史底部的估值分位（forward P/E在15x-25x）使其赔率极具吸引力。餐饮降本驱动B端标准化需求稳固。测算表明，由50%防御型高股息家电、30%食品加工Alpha…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "白酒的防御贡献成立但只对高端有效——把20%白酒沙盘拆为15%高端龙头+5%调味品/地产酒后，杠铃组合悲观情景的回撤可显著收窄、夏普由0.69升至约0.78。", "summary": "截至2026-06-13，本卡对卡06提出的'50%高股息家电+30%食品加工+20%白酒'消费防御杠铃组合做压力测试。高端白酒（茅五泸）2024-2026Q1净利率稳定在37%-53%、毛利率方差仅±0.3pct，确实可承担组合压舱石，对冲食品加工链的油脂成本不确定性；但次高端净利率已回吐6-1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 5, "neutral": 1, "value": 45.8, "events": [ { "title": "2026年食品加工业正经历从‘低价放量’向‘合规溢价’的结构性拐点，重点关注具备供应链主权及天然/健康配料优势的低覆盖‘隐形冠军’。", "summary": "本次调研深入剖析了2026年食品加工产业链的‘高质量韧性’转型。当前行业激活率仅6.6%，处于严重低估状态。核心发现包括：2026年预制菜国标落地引发行业洗牌；农业原材料（如生猪-16%）与工业投入（能源+20%）成本出现极端背离。筛选出晨光生物（天然替代红利）、万辰集团（量贩零食整合）、惠发食品…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品自动化升级对半导体的拉动方向正确但量级有限——优选卫生级传感器与机器视觉的小盘纯标的，SiC/IGBT仅为期权式跟踪。", "summary": "截至2026-06-14，2026食品新国标驱动的450亿元高能效杀菌设备capex周期，对上游半导体的单机BOM拉动确实存在3-5倍提升（HPP/PEF/无菌灌装），但绝对增量约18-28亿元人民币（自测算），占中国功率半导体与传感器市场不到1%，不足以驱动板块β。结构性受益集中在三条线：卫生级…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "边缘AI以极低的单点功耗（7-15W）完美规避了2026年数据中心面临的吉瓦级电网瓶颈，是大规模工业自动化中最具能源协同效益的算力基建。", "summary": "本次调研分析了2026年边缘AI部署与大模型算力电力瓶颈的协同关系。数据中心AI推理带来严重的电网拥堵（单机柜功耗达100kW-1MW），推高了云端算力成本。相比之下，工业现场部署的边缘AI（如RK3588、Jetson Orin Nano）单节点功耗仅7-15W。在一家中型食品加工厂部署50台边…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 2, "neutral": 0, "value": 62.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "食品加工产业链 30 日激活率仅 6.6%（16/242），但正处三条催化交汇：2026-02-06 首个预制菜国标征求意见（市场 2025 年 6173 亿、+27.3%，冷链成本+15-30% 出清中小厂）、餐饮渠道复苏、清洁标签减糖转向。筛出 5 只低覆盖标的：百龙创园(605016) 阿洛…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 5, "neutral": 0, "value": 45.6, "events": [ { "title": "在利率下行与结构性流动性虹吸背景下，以37.5%平价消费核心加62.5%高股息底仓构建杠铃组合，辅以严密的基本面与交易指标减仓纪律，是锁定防御收益并隔离小盘两融踩踏风险的最优战略表达。", "summary": "截至2026-06-18，本报告综合了食品加工产业链与红利资产配置研究，确定杠铃组合最终权重为：巴比食品16.0%、千味央厨21.5%、长江电力40.0%、中国神华22.5%。报告基于两融拥挤压力测试与水文来水、煤炭长协等经营条件，制定了硬性动态减仓触发纪律；测算显示在10%全市场融资去杠杆冲击下…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工中小盘的融资盘拥挤度与流动性呈现极端两极分化，海融科技极高的两融占比与低成交额使其在平仓压测中面临-46.50%的崩溃风险，而巴比食品与千味央厨则具备强流动性防御垫，估值修复…", "summary": "截至2026-06-18，本报告对食品加工中小盘两融拥拥挤度与流动性约束进行了定量压力测试。测算显示，A股整体两融余额高达2.89万亿元，杠杆资金向TMT等热门板块倾斜造成了明显的流动性虹吸效应，导致食品加工中小盘日均成交额萎缩。个股风险呈极端两极分化：海融科技融资余额1.24亿元，占其自由流通市…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在两融高位与TMT虹吸环境下，食品加工alpha应以37.5%平价消费核心叠加62.5%红利底仓表达，而不是放大中小盘消费beta。", "summary": "截至2026-06-18，本卡将食品加工主线转化为可执行杠铃组合：巴比食品16.0%、千味央厨21.5%、长江电力40.0%、中国神华22.5%。受约束均值-方差模型显示，组合年化预期收益10.37%、年化波动率15.80%、95%单日VaR为1.64%、20日VaR为7.32%。在10%全市场两…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 3, "neutral": 0, "value": 59, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工产业链 7.4% 激活率反映覆盖供给不足而非基本面缺乏机会，中盘 B2B 速冻 + 酵母调味成本顺周期 + 北交所出口预制菜里至少有 1–2 个 6–12 个月维度的 idi…", "summary": "A 股食品加工 30 日激活率 7.4% (18/242) 反映卖方覆盖稀疏而非基本面缺乏机会。本卡选出 5 个低覆盖、setup 较硬的候选：千味央厨 (001215.SZ, B2B 餐饮速冻)、立高食品 (300973.SZ, 山姆冷冻烘焙)、安琪酵母 (600298.SH, 糖蜜成本下行 +…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A 股食品加工板块的组合表达、流动性及财报季催化研判", "summary": "本卡支持前两卡低覆盖食品加工 alpha 主线，并补上策略师组合视角。截至 2026-06-19，A 股两融余额约 2.85 万亿（6/10），但边际杠杆资金集中于半导体/工业金属，食饮被动资金流向白酒龙头，千味央厨、立高食品均非 ETF 锚定标的——故 alpha 是主动资金故事而非被动 β。流…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 2, "risks": 7, "neutral": 2, "value": 58.4, "events": [ { "title": "消费端财政扩张无法逆转食品加工链向规模龙头集中的风格偏好，2026年四季度成本反转将加剧中小盘的结构性成本挤压，资产配置应聚焦『规模龙头多头+生猪周期多头对冲』的哑铃型策略。", "summary": "本卡在PPI再通胀与CPI食品通缩共存的宏观分化背景下，对食品加工产业链的风格分化与资产配置进行压力测试。自测算敏感度模型表明，即使2026年内实施5000亿元的消费端财政扩张，由于其无法抵消2026Q4-2027Q1猪周期反转（预计生猪价格+20.0%）及PPI再通胀（+5.0%）对高成本中下游…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工产业链下游消费趋势与零售端驱动力分析", "summary": "本卡从需求侧支持 Research note 01：中游 7.4% 低激活率是覆盖缺口而非需求塌方。2026年1-5月社零仅+1.4%、5月单月-0.6%，餐饮5月降至+0.6%，2025年餐饮闭店率48.9%、人均-23.6%——总量偏软、追性价比。但渠道走量引擎轰鸣：即时零售破万亿、量贩零食双…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在强势美元与国际谷物低价的双重作用下，低廉的进口谷物成为我国抵御输入性通胀的屏障，同时强劲的经常项目顺差为人民币汇率及国内流动性稳定提供了重要的宏观缓冲。", "summary": "本报告分析了截至2026-06-22，全球谷物低价与强势美元对中国宏观经济的传导机制。强势美元（USD/CNY为6.77）带来约4%的进口汇率拖累，但国际粮价大跌使得进口依然享有高达30-40%的价格优势。低名义进口账单与强劲出口推动5月贸易盈余达1054.3亿美元，有力防御了中美273.8bp的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 1, "risks": 4, "neutral": 0, "value": 63.8, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品饮料篮子配置应定位为“防御性消费核心资产”，通过“80%广义消费内调仓+20%红利价值切出”的存量优化完成，并设立组合层面0.4%-0.8%的回撤贡献预算与三级风控硬停损机制。", "summary": "截至2026-06-23，主张对限制白酒权重的抗硬折扣食品饮料篮子给予5%-8%的存量调仓预算。调仓由80%广义消费调仓与20%红利价值切出构成，不消耗现金。篮子回撤预算设为8%-10%，对应组合整体下行净值贡献限制在0.4%-0.8%内，并设立三级风险控制与硬停损机制。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "20%红利价值切出可以通过0.4%-0.8%回撤预算测试，但前提是切出来源、篮子仓位和10.0%硬止损均被严格约束。", "summary": "截至2026-06-23，本卡对20%红利价值切出方案给出有条件支持。若食品饮料篮子总仓位保持5.0%-8.0%、篮子回撤止损保持8.0%-10.0%，则组合层面回撤贡献为0.40%-0.80%，红利价值资金腿单独贡献仅0.08%-0.16%。但该方案不自动维持防御性：中证红利低波动指数波动率和估…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工链是成本×拐点的选股交易而非防御板块——消费腿只集中持有安井，真正的战术防御靠低估值高股息杠铃捕捉约270bp股息利差。", "summary": "2026-06-23：综合前三卡，食品加工链应被定位为“成本可防御性×已验证拐点”的选股交易，而非防御板块。框架下只超配安井食品(603345.SH，鱼糜自给约60%+Q1全面拐点，2026E PE约11.7倍/股息约3.4%)；千味央厨(001215.SZ)小仓观察；甘源、劲仔、立高低配至成本缓…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 2, "neutral": 1, "value": 80.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在消费整体承压背景下，具备细分赛道高技术壁垒、渠道重构或报表出清特征的食品加工“阿尔法”小盘股，能够在原材料周期性回落中实现显著的业绩超额收益。", "summary": "本报告调研了A股食品加工产业链，筛选出仲景食品、甘源食品、晨光生物、仙乐健康等市值在60亿元以下、具独特阿尔法壁垒且报表已清的个股。同时测算了麦收期连阴雨（芽麦分化）与印尼B50政策下植物油波动的跨链溢出和传导机制，并给出了量化消费组合资产配置策略。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "基于食品加工小盘股盈利风险评估A股风格轮动与板块配置机会", "summary": "本卡支持前序结论并升维：食品加工阿尔法小盘股的盈利脆弱性是小盘风格“估值/拥挤度透支盈利”系统矛盾的前瞻样本。我们用四要素框架（盈利裂缝20、估值性价比19、拥挤度13、增量资金21）给出风格切换成熟度73/100——三大支柱已就位，唯拥挤度未到极端引爆点，故判断为“小盘→大盘/价值再均衡”而非单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹冲击的关键不是单日油价，而是持续时间、库存下降和成品油/LNG/电力链条能否把短期能源涨价转化为更持久的通胀预期。", "summary": "截至 2026-06-25，本卡支持上一张白板卡：霍尔木兹海峡中断是成本推动型能源通胀冲击。核心证据是 2024 年海峡承载 20 million b/d 石油、约全球石油液体消费 20%，并承载约五分之一全球 LNG 贸易；可用绕行能力仅约 2.6 million b/d，留下约 17.4 mi…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 5, "neutral": 2, "value": 58.6, "events": [ { "title": "食品加工产业链低激活率反映覆盖不足，二线供应链龙头在渠道红利与成本回落双重驱动下迎来利润反转，应超配盖世食品、百龙创园等高壁垒二线标的，避开立案的田野股份。", "summary": "【2026-06-27工作日报告】本报告调研了A股食品加工产业链，指出30日激活率7.9%（19/242）反映研究覆盖缺口而非基本面崩溃。当前处于成本通缩与量端复苏的毛利扩张期。精选具备强阿尔法的二线隐形冠军（盖世食品、百龙创园、海融科技、佳禾食品），警示立案的田野股份，构建以二线核心多头为主的哑…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026-06-27的核心判断是：用50%中债久期和18%尾部商品对冲替代传统高股息/出海股票防御，才能同时免疫国内ASP通缩与海外关税地缘风险。", "summary": "截至2026-06-27，本卡片压力测试前序结论并给出多资产重置：国内餐饮供应链不是顺畅再通胀，而是食品/餐饮价格走弱、生活资料PPI下行与上游购进价格上行并存的利润挤压。建议将组合重置为A股32%、中国国债及政策性金融债50%、大宗商品18%；股票只保留受监管国内现金流与国产替代等窄口径哑铃，商…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "summary": "压力测试整套对冲架构所依赖的55-65%国内通缩基准概率。结论：基准情形近期稳健，保留但下调至55-60%下沿。硬数据印证消费端通缩：2026年5月CPI同比+1.2%、食品-1.7%、核心+1.1%、环比-0.1%；10年中债贴地板于1.73%[S1][S5]。PBoC“边走边看”渐进（LPR连…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "生柴需求不会瓦解所有食品加工公司的成本顺风，但已经瓦解板块级通缩叙事，并把配置逻辑强制切换为油脂敞口和定价权筛选。", "summary": "2026-06-27：本卡压力测试生柴需求对食品加工成本通缩逻辑的破坏力。结论是：生柴/RFS 已经显著推升豆油、菜籽油及替代油脂价格，足以证伪板块级“成本通缩”叙事；但因更高压榨同时增加豆粕供给，仍不否定低油脂敞口或受益蛋白粕/肉类成本改善公司的阿尔法。高油脂食品企业若油脂占 COGS 10%、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 0, "value": 66, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工板块β偏弱但覆盖缺口蕴含丰厚Alpha，应重点布局中报Setup干净的千味央厨与受新添加剂国标催化的B端生物配料龙头，回避无定价权的中小盘。", "summary": "食品加工产业链过去30天激活率仅为7.9%，存在严重的覆盖缺口。行业正处于食品CPI同比-1.7%与PPI再通胀的“食品链剪刀差”结构分化期。压测显示，在原材料PPI上涨5%的成本反转情情下，中小盘毛利率预计将从22.69%骤降至10.50%，推动资金向头部大盘股靠拢。建议避开 generic 品…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美大豆天气风险与双边关税扰动的叠加将打破国内豆粕的通缩基础，通过流向偏转、基差飙升及汇率乘数将大商所豆粕推至极高水位，重创饲料与养殖链，并最终传导侵蚀食品加工企业毛利，必须战略性修正…", "summary": "本卡片对美豆夏季天气风险与关税贸易冲突双重扰动下国内蛋白粕（豆粕）成本趋势进行了二次压力测试。结论表明，贸易政策与天气风险叠加将彻底瓦解国内豆粕的通缩逻辑，引发大商所豆粕期价与美盘脱钩并大幅暴涨。在最极端情景下，大商所豆粕主力可涨至6,120元/吨（+107.5%），重创国内饲料及生猪养殖行业，逼…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 0, "neutral": 0, "value": 93.4, "events": [ { "title": "消费需求结构性演变：从耐用品'以旧换新'向服务与必需品类别的防御性迁移", "summary": "支持卡片04判断：食品饮料与基础零售能撑起防御性杠铃一端，但防御在'量与现金流'而非'营收'。2026年5月社零同比-0.6%下，必需品逆势走强（饮料+6.1%、烟酒+4.8%、药品+4.0%），除汽车外零售+1.1%，而汽车-16.1%、家电-15.6%为透支回吐——迁移已印在数据里[S1]。但…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 1, "neutral": 0, "value": 80.8, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-d0de40066b/", "networkHref": "/investor/research-network/#thesis-d0de40066b" }, { "id": "thesis-97a24f8796", "source": "auto_topic", "title": "公用事业产业链：研究活跃度与投资线索复盘", "titleEn": "Practical research topic: 公用事业产业链：研究活跃度与投资线索复盘", "lane": "电力与公用事业", "laneEn": "电力与公用事业", "laneHref": "/investor/lanes/lane-ee7fbf3f/", "investableFocus": "", "stockUniverse": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH", "002028.SZ", "APO", "IEF", "COP", "CAT", "MAR" ], "coreView": "该主线来自 22 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-17 至 2026-06-23；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-17", "lastSeen": "2026-06-29", "supports": 9, "risks": 142, "neutral": 9, "analystCount": 13, "sourceTypeCounts": { "研究笔记": 139, "研究线程": 21 }, "topTerms": [ "公用事业", "公用事业产业链", "研究活跃度与投资线索复盘" ], "whatChanged": [ "2026-06-29: AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "2026-06-29: AI电网资本开支是公用事业的真实资产增长机会，但信用赢家不是backlog最大的公司，而是每一美元资本开支对应现金流/债务缓冲最大的公司。", "2026-06-29: 变压器成本飙升与交期拉长对电网企业资本开支及费率基数扩张的影响", "2026-06-29: 在名义中性利率上移和AI TFP红利右移的双重约束下，组合应减少长久期暴露，增配电网金属与受监管现金流，并以盈利型AI基础设施而非高久期叙事保留成长期权。", "2026-06-29: 材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放右移。" ], "practical": { "sectorLane": "电力与公用事业", "sectorLaneEn": "电力与公用事业", "exposureBucket": "电力与公用事业", "tone": "yellow", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH", "002028.SZ", "APO", "IEF", "COP", "CAT", "MAR" ], "primaryStocks": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" } ], "instrumentTags": [ "GEV", "GE Vernova", "ETN", "Eaton", "PWR", "Quanta Services", "VST", "Vistra", "CEG", "Constellation Energy", "XLU", "Utilities Select Sector SPDR" ], "reviewCadence": "每日复核", "confirmationSignals": [ "并网排队和变压器交期是否继续拉长", "表后自发电与 PPA 是否改变公用事业信用指标", "电力设备订单质量是否高于收入确认节奏" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "长端利率", "铜/GOES 成本", "监管电价", "信用利差" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 2 类，近期事件 98 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 2, "analystCount": 13, "recentEvents": 98, "supportRatio": 0.06, "riskRatio": 0.89, "conflictScore": 99.1, "evidenceDepth": 100 }, "score": 66.7 }, "events": [ { "date": "2026-06-29", "title": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红利（长江…", "summary": "本报告针对2026年三季报TMT硬件估值折价预期，校准了全球全天候和风险平价模型。战术防御性风险平价（TDRP）将AI成长股风险预算由25.0%下调至8.0%，防守红利上调至40.0%，以对冲55.0%的TMT增长溢价折价。模型显示，电力配给约束下算力效率的TCO优势在24/7清洁能源下扩大至9.97%，高功耗算力中心毛利率面临负增长。此外，建议红利套组内…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-29", "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优先既有水电/核电现金流、具备瓶颈定价权的变压器和电网设备供应…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-29", "title": "选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "summary": "截至2026-06-29，本卡综合前9卡：AI电力约束是选择性时序上限而非订单毁灭。建议跨资产账本配置为40%受资助AI基础设施龙头、45%主权背书红利公用事业、3%年度申万电子Put权利金预算、12%流动性储备。进攻端以中际旭创18%、工业富联14%、英维克8%为核心；防御端以水电18%、核电17%、受监管电力/电网10%承接电力稀缺；Put以35-45…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-29", "title": "长江电力在极端联合压力下股息现金覆盖率仍达125.55%，分红耐久性无实质破坏风险，是优质的防守压舱石，而高资本开支的中国核电分红受制于负现金流和债务扩张。", "summary": "在偏枯水情(-15%)、电价下行(-10%)及资本超支的严重联合压力下，长江电力归母净利润降至23.12B RMB，股息率降至2.38%。但因其折旧盾和低开支，权益自由现金流达20.32B RMB，对股息覆盖率仍有125.55%，无破坏性分红风险。中国核电（CNNP）因百亿级在建资本开支致使自由现金流常年为负，分红依赖发债，脆弱性高。多资产压力下，战术风险…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-29", "title": "电网(发电+输配+设备)是2026–2028 AI算力的硬天花板，按多年期时钟清出，宏观外溢是由全体用户承担的电价与容量成本冲击。", "summary": "本卡片从公用事业/电网视角支持并压力测试 Research note 02 的'电力已取代GPU成为首要瓶颈'论断。截至2026-06-29，美国发电、输配与电气设备产能无法按时全额满足规划中的2026–2028数据中心负荷：美国DC负荷到2028年约108GW、占全美用电最高约12%[S2][S3]，增量约+70GW(~24GW/年)[自测算]；但四个限…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-29", "title": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放…", "summary": "截至2026-06-29，变压器交期延长和材料瓶颈导致美国电网资本开支通胀率达32.72%。结合2027/2028年度PJM容量电费暴涨11.5倍至$333.44/MW-day，将拉动美国零售电价上涨4.18%，直接和间接推高全国CPI 0.124个百分点。高粘性通胀与AI基建庞大资金需求将上移中性利率50-75基点，延缓美联储降息路径。此外，并网滑期18…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-29", "title": "在名义中性利率上移和AI TFP红利右移的双重约束下，组合应减少长久期暴露，增配电网金属与受监管现金流，并以盈利型AI基础设施而非高久期叙事保留成长期权。", "summary": "截至2026-06-30，本卡片综合前序电网瓶颈、通胀和AI TFP右移逻辑，建议采用“实物资产与现金流杠铃”：100单位组合中配置30单位短中久期美债/TIPS、15单位电网金属与商品代理、45单位权益、10单位现金。权益内部把成长科技与防御公用事业设为14 vs. 18，压低高久期AI叙事，保留盈利型AI基础设施，同时超配电网设备、UHV、铜铝和受监管…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-29", "title": "变压器成本飙升与交期拉长对电网企业资本开支及费率基数扩张的影响", "summary": "截至2026-06-29，本卡片压力测试'变压器稀缺→费率基数膨胀→公用事业股息回报上行'的双赢叙事，结论是方向成立但被高估、且是双向的。一方面，128周（大型机组达60个月、4–5年）的变压器交期与2.6倍价格溢价使并网约束从'出资意愿'转为'实物可得性'，确凿减缓AI并网。另一方面，约1.3万亿美元(2026–30)资本开支以9.4%–10%允许ROE…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-28", "title": "研判总结 / Research Summary 本卡片作为 公用事业分析师 (Utilities Analyst) ，锚定研究院权威工作日期 2026 06 28 ，采用 合成 (Synth…", "summary": "研判总结 / Research Summary 本卡片作为 公用事业分析师 (Utilities Analyst) ，锚定研究院权威工作日期 2026 06 28 ，采用 合成 (Synthesize) 的立场，对前序卡片中关于量子/AI基础设施对电网和能效重塑的探讨进行了物理瓶颈与投资可行性的深度集成研判。 核心结论： 1. 电网硬件扩容的物理瓶颈是算力…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-28", "title": "物理基础设施约束下多资产投资组合的重平衡与防守哑铃配置策略", "summary": "本卡支持卡片05–07链条，把物理基础设施约束落地为多资产重平衡：构建防守哑铃。削减超估值且拥挤的AI硬件成长腿（−12pp至20%并加期权领口），把仓位转入物理基础设施租金腿（公用事业含水电核电、电网重电与变压器、水权、液冷热管理，+14pp至22%）与高股息压舱（中证红利约4.68–4.9%股息率，超10年期国债1.73%约290bp）。中间以久期、黄…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-28", "title": "PQC 引发的周期性热降频会显著提高超大规模 AI 集群 PUE/WUE 与能源成本预测难度，并在集中接入区增加电力调度复杂度，但它是边际波动因子而非全国电力需求的主增长因子。", "summary": "2026-06-29 能源卡结论：支持前序观点，但限定机制。PQC 硬件化的直接功耗不是数据中心用电增长主因；真正问题在于 PQC 触发的周期性热降频、重协商突发和 GPU 空转会把稳定 IT 负荷变成区间波动，从而削弱年度 PUE/WUE 对边际成本的解释力，并在数据中心集中的节点增加电网调度和负荷建模复杂度。单个 100 MW AI 站点的 5-10 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-28", "title": "为了防止AI数据中心无序扩张导致居民电费暴涨，公用事业监管机构正强制推行以‘85%起付线、长周期和强担保’为特征的大负荷电价机制，实质性地将电网容量闲置与资产搁浅风险转嫁回开发商本身。", "summary": "2026年6月28日，AI数据中心的暴增使得电网容量极其紧张，导致监管机构推行新型电价体系以防范逆向交叉补贴与资产搁浅风险。本卡综合了此前的研究成果，深度解析了FERC于2026年6月18日发布的最新大负荷示因命令，以及俄亥俄州与弗吉尼亚州大型公用事业公司推行的以“85%最低容量起付线”和长期合同绑定为核心的电价政策。定量模型表明，该起付线能在数据中心延迟…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-28", "title": "面向AI基础设施信用尾部的跨资产组合配置与对冲", "summary": "截至2026-06-28，本卡综合前序结论：AI算力长期上行仍可持有，但尾部风险在信用结构而非单纯科技股估值。建议采取“优质AI+电力瓶颈+信用保护”杠铃：保留NVIDIA与现金流型超大规模云厂商核心仓位，低配Neocloud股票和GPU抵押DDTL/私募信贷，增配受监管公用事业、电网设备、稳定电力与A股电网设备卫星，并以CDX HY、CDX IG和Ora…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-28", "title": "债市收益率触底倒逼绝对收益资金流向A股优质高股息红利股建立‘红利锚’，在‘数量先行、价格被锚’的结构性宽松下，配置策略应向国有大行与电力设备出口链的对冲型哑铃结构倾斜。", "summary": "本卡片合成分析了国内债市10Y国债至1.73%、信用Carry枯竭下的资产荒对A股的外溢效应。外溢主要通过险资等绝对收益资金流向A股红利资产（利差达+3.27%至+3.365%），形成局部红利锚。在价格降息受限、数量宽松先行的政策组合下，长久期成长股缺乏催化剂，建议实施红利大行防御与硬科技出口链主线的哑铃型配置，战术规避中小城商行与公用事业久期陷阱。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-28", "title": "表后共址数据中心引发的规避网费及系统容量价格暴涨，迫使FERC与地方公用事业启动资费重构与高额抵押担保以防范信用降级，未来需通过双向费率平衡可靠性成本分摊与灵活调节信用。", "summary": "截至2026-06-28，本卡片对BTM数据中心并网及规管成本分摊进行系统性梳理。云厂商利用BTM直连核电/气电旁路输配网费引发了剧烈的成本转嫁争议，导致PJM容量出清价暴涨至333.44美元/MW-日，且有40%（65亿美元）由数据中心预测负荷驱动。监管机构正加速资费纠偏，如FERC于6月18日发布Show-Cause命令、弗吉尼亚GS-5要求每兆瓦15…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" }, { "date": "2026-06-28", "title": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费成本转嫁挤压。", "summary": "基于国产芯片2.94倍能效赤字拉长通电时间至18个月的物理瓶颈，本报告系统性量化了A股电力设备板块（特高压、大变压器）的估值扩张机制与低利润 traditional TMT 的电价挤压。电网容量受限与十五五规划共振，支持了特高压订单可见度与大型变压器出海高溢价，特别是在国内外硅钢5.1倍巨大价差的成本红利下。策略上构建杠铃组合，超配直接出口变压器龙头、特高…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.5, "matchedTerms": [ "公用事业" ], "href": "" } ], "timeline": [ { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 1, "value": 39, "events": [ { "title": "公用事业产业链十卡研究最终评级为Needs-Revision——card-09终局结论与应收账款筛选可发布，但card-06两处证券代码错配（600550.SH/600995.SH）…", "summary": "截至2026-05-17，对公用事业产业链Research note 01-09做最终QA检查。card-09终局结论（上调受监管现金流资产+一线电网设备龙头，下调地方PPP应收敞口+弱资产负债表设备公司）数据自洽、与应收账款筛选阈值（AR/收入<50%、经现/净利>1.0x、负债率<65%、央企…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 10.3%, 21/203）", "summary": "产业链调研：公用事业产业链（30 日激活率 10.3%, 21/203） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "低关注度的公用事业链已经具备用电需求、碳市场扩围和现金流修复三条线索，但首轮应避开泛化配置，集中验证 4 个可深挖标的。", "summary": "截至 2026-05-21，本卡支持开启公用事业产业链研究队列。首批建议聚焦新天绿色能源、绿色动力环保、中国水务、北控水务集团：它们分别对应气电风一体化、垃圾焚烧现金流修复、供水价值修复和大型水务平台轻资产转型。核心外溢风险包括电价/消纳压力、碳市场扩围向钢铁水泥铝传导、地方财政支付周期以及燃气需…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "AI 算力爆发带来的电力基础设施瓶颈与公用事业价格市场化改革，正在将传统防御性板块转化为具备结构性超额收益的‘增长基础设施’。", "summary": "截至 2026 年 5 月 23 日，公用事业产业链受 AI 算力电力协同和价格改革驱动正经历价值重估。AI 数据中心需求预计在 2026 年达 279 TWh，推动数据中心向虚拟电厂（VPP）转型。水务新规将于 6 月 1 日施行，强化工业顺价。地缘政治导致的霍尔木兹海峡封锁推高了 LNG 成本…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力现货市场化与碳市场扩容正重构公用事业估值逻辑，由单纯防御收息转向结构性成长与跨界套利。", "summary": "本报告基于2026年5月的最新数据，梳理了公用事业产业链的微观结构与宏观动能，挖掘了粤电力A、北控水务、伟明环保及申能股份4个低覆盖度优质标的。提示了CCER扩容及TMT算力中心带来的跨链风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "公用事业正经历从规模扩张到“系统协同+价值重估”的转变，碳资产溢价和容量电价机制是核心驱动力。", "summary": "本次调研梳理了公用事业产业链在2026年“十五五”开局之年的投资逻辑。火电从基荷向灵活性调节转型，碳市场扩围至钢铁建材，环保转向电池回收与AI运维。筛选出申能股份、惠城环保、美埃科技和福能股份4个具备深度研究价值的低关注度标的，并提示碳价上行对高耗能产业链的溢出风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 2, "neutral": 0, "value": 41.2, "events": [ { "title": "公用事业板块正迎来电价/水价顺价机制落地与“电算一体化”催化的系统性价值重估，兼具类债防守与新电网弹性。", "summary": "调研分析了2026年公用事业产业链的转型趋势，重点挖掘了皖能电力、兴蓉环境和明星电力三只低覆盖度标的，涵盖火电价值重构、水务现金流拐点及虚拟电厂溢价逻辑，并提示了上游燃料涨价及宏观用电需求放缓等跨链风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 3, "neutral": 0, "value": 36.6, "events": [ { "title": "2026年“碳排放双控”考核硬化与“算电协同”基础设施工程建设，正在深度重塑绿电和地方公用事业资产估值，精选机构低覆盖、具有资产壁垒和分红确定性的细分龙头或反转标的将获得超额收益。", "summary": "本报告对A股公用事业产业链展开深度调研。分析了2026年全面落地的“碳排放双控”考核办法及交易所ESG强制披露规则等政策影响。筛选出机构覆盖率低的4个价值标的（芯能科技、洪城环境、广西能源、百通能源），覆盖分布式光伏储能、高股息水务、困境反转绿电网及工业区热力等方向，并详述了四类跨链溢出风险，推荐…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "在低利率和容量/价格改革下，公用事业个股通过煤电一体化、资产并表和节能换挡展现出强劲的债性防御力与价值重估空间。", "summary": "本报告系统调研了公用事业产业链，筛选出瀚蓝环境（垃圾发电全资并表粤丰环保）、陕西能源（自产煤平滑波动的煤电一体化）、涪陵电力（一季度配网节能业绩迎来拐点）和皖能电力（容量电价上调直增4.39亿净利）四个隐形冠军标的；并提示了算力绿电 PPA 溢价、大宗电价传导、电网设备投资和环保坏账拨备等跨链溢出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 13.2%, 27/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 13.2%, 27/205） 2%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 13.2%, 27/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 13.2%, 27/205） 2%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 2, "neutral": 1, "value": 49.8, "events": [ { "title": "公用事业被卖方冷落≠无 alpha——被错配标签的申能、川投、皖能等中盘标的，正同时受益于容量电价、136 号文与碳市场扩容的现金流重估。", "summary": "公用事业产业链 30 日激活率仅 13.7%，但容量电价二年、136 号文新能源全面入市、水价改革与 ETS 扩容四股政策力量正同时重塑现金流可见度。本卡片筛出五个被错配标签、卖方覆盖密度低的标的：申能股份 600642（被当火电的核电期权）、川投能源 600674（雅砻江纯水电权益）、皖能电力 …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 13.7%, 28/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 13.7%, 28/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 13.7%, 28/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 13.7%, 28/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 1, "value": 57.2, "events": [ { "title": "公用事业产业链正处于电价市场化、核电扩张与 ETS 重启三轴重定价中，超额收益落在 003816、600803、600905、600008、1164.HK 五个低覆盖度标的，主流的长…", "summary": "公用事业产业链当前 30 日激活率仅 11.7%（24/205），叙事被长江电力 600900 一只标的吸走，但 2025-02 风光电价市场化、2026-03 十五五核电 110 GWe 目标、2026-01 ETS 重启叠加 CEA-2025 切换三大监管冲击在 12 个月内连发，买方盈利模型…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 1, "neutral": 1, "value": 61.4, "events": [ { "title": "公用事业正经历从‘类债资产’到‘AI核心基座’的估值迁徙，算力用电稳定性与碳资产增值将成为产业链盈利的核心变量。", "summary": "2026年公用事业产业链受AI算力爆发与‘算电协同’政策驱动，已从传统红利板块重塑为AI基础设施基座。本次调研识别了恒华科技（虚拟电厂设计）、南网科技（电网数字化）、思源电气（海外出海）、盾安环境（数据中心液冷）及岳阳林纸（碳汇资产）五个低覆盖度、高成长潜力的标的。行业核心催化剂包括容量电价补偿落…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 1, "neutral": 0, "value": 59.8, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 8, "neutral": 1, "value": 22.2, "events": [ { "title": "评估防御端转向本土现金流篮子后，组合利率久期上升对固收对冲策略的影响。", "summary": "截至2026-06-20，本卡压测卡09组合久期上升的固收对冲。关键判断：所谓久期一半是4%利率债腿的干净DV01、一半是公用事业/运营商/保险等股票的经验类债券贝塔，后者无法1:1对冲。当前曲线极平、10Y创年内低点约1.70%、当周熊平(2Y+6.06bp>10Y+2.43bp)、DR007 …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI数据中心选址正在从地产逻辑转向公用事业逻辑：谁能交付firm MW、低水耗冷却和可执行并网里程碑，谁就获得下一轮Hyperscaler容量。", "summary": "截至2026-06-20，本卡压力测试Research note 3的AI基建ROI逻辑。结论是：Hyperscalers仍会建设，但选址约束已经从“GPU/土地/税收优惠”转向“可交付firm MW、变压器/变电站交付、低水耗冷却和地方许可”。1 GW IT负载、1.20 PUE集群年耗电10.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler AI扩张面临的是“资本买不到时间”的物理瓶颈：有资产负债表不等于能压缩电网、变压器、许可和矿产供应周期。", "summary": "截至2026-06-21，本卡片对AI基础设施扩张做压力测试：Hyperscaler的主要风险不是单纯GPU短缺，而是可靠电力、PJM类大负荷接入、变压器/输配电设备、关键矿产精炼集中度和能源成本共同造成的物理交付瓶颈。IEA预计全球数据中心用电从2024年415 TWh升至2030年945 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网积压与公用事业审批已成为未锁定电力权益的AI绿地项目的真实区域性天花板，但会通过费率、抵押和可靠性规则筛选项目，而非一刀切终止AI扩张。", "summary": "截至2026-06-21，本卡片压力测试卡片05并支持其物理瓶颈判断，但把机制收窄为“并网队列+大负荷费率+可靠性审批”的区域性天花板。LBNL显示美国2025年底超过2,060 GW发电/储能排队；AEP Ohio把超过30,000 MW数据中心意向筛到5,642 MW约束性合同；PJM预计20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 3, "risks": 24, "neutral": 0, "value": 12, "events": [ { "title": "煤电容量电价新政、水价理顺与碳价走势为公用事业板块提供了确定性极高的分子端盈利增长与红利安全垫，支持其相较于1.75%国债利差的系统性估值重估。", "summary": "截至2026-06-21，基于对2026年煤电容量电价回收比例提升至不低于50%政策、2025年广深等核心城市水价上调10%-30%事实、以及2026年上半年CEA与CCER价格稳定在80-89元/吨的验证，本期卡片证实了公用事业分子端盈利与派息水平的结构性扩张，支持估值重估逻辑。", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "公用事业股息率扩张对大类资产配置逻辑的影响", "summary": "截至2026-06-21，在50BP短端利率走廊（7天逆回购1.70%）与10年期国债1.70%–1.75%监测区间背景下，公用事业已从折现率交易转为政策托底的债性资产。基于历史指数（红利低波年化波动率约20.7%<沪深300约21.9%；夏普近3/5/10年高出沪深300 +0.88/+0.80…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "1.75%利率底线下的收益率曲线动态与久期交易风险", "summary": "截至2026-06-21，1.75%国债底线保护的是短端水平，却无法使长端免疫。真正破坏公用事业配置价值的不是底线向下失守，而是由美国期限溢价上行（10Y约4.46%、ACM期限溢价约0.67%且上行）、日本央行正常化（政策利率1.0%、10Y约2.63%）与中美利差深度倒挂（约-271bps）共…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在长端利率上升50bps与相关性反转的双重冲击下，公用事业红利组合（P2）将面临波动率飙升至14.62%与预期回撤达-8.24%的“久期陷阱”，防御效能逆转并劣于高贝塔组合（P0）。", "summary": "截至2026-06-21，本报告针对Research note 03和Research note 06提出的P2配置方案（20%成长权益、40%公用事业、40%债券）进行收益率曲线陡峭化与相关性反转的极端压力测试。测算显示，若长端收益率上升50bps且发生相关性反转（公用事业-国债相关性从+0.1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 2, "risks": 7, "neutral": 1, "value": 47, "events": [ { "title": "真正受益的是电力交付瓶颈和已签长期购电/费率保护资产，而不是泛化公用事业或所有新增兆瓦。", "summary": "截至2026-06-22，本卡综合上一张AI基础设施观点：数据中心用电增长仍成立，但公用事业受益必须按“执行风险由谁承担”重新排序。最优是电网设备和电力服务，其次是有已签负载承诺及最低计费/大负荷费率保护的受监管公用事业，再次是已签长期PPA的核电或稳定清洁电源。泛化merchant gas、水资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI需求能扭转公用事业的投资疲软，但路径是把板块增长重新导向发电与电网（及选择性燃气/水电稳功率），而非普遍抬升水务/环保——是结构性扩张，印证Research note 09的选择…", "summary": "截至2026-06-23，本卡（AI基础设施分析师，10/10收官卡）支持整条研究线，并补上需求侧催化剂。Research note 09判定市政管网仅政策订单扩张、未全板块扩张（1—5月电热气水固投-0.1%、水利环境-3.9%），那是旧的地产/市政权重图景；AI带来的是新的发电与电网权重图景。…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "能源价格冲击对通胀预期的传导风险", "summary": "截至2026-06-22，本卡以能源视角支持Card-01「暂停+更久维持高位选择权」，但修正其传导机制。需区分两类冲击：(1)地缘油价飙升已回落——WTI从约$109崩跌约29.6%至$72.83，美伊重开霍尔木兹后布伦特回到约$80，且油价对核心通胀传导小而慢(峰值约0.12pp)；(2)真正…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超大容量AI负荷对受监管公用事业公司电网资产建设与费率分配的影响", "summary": "截至2026-06-22，吉瓦级AI负荷迫使受监管公用事业进入输电资本开支超级周期；关键不在是否建设，而在谁付费、费率基数如何围栏化以保护居民用户。已成型的三层防御：(1)带85%最低用量、12年合同与退出费的专用大负荷费率（AEP Ohio DCT、Dominion VA GS-5）；(2)将并…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 1, "risks": 14, "neutral": 1, "value": 12, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 12.6%, 26/206）", "summary": "公用事业链近30日激活仅26/206（12.6%），覆盖最低，但宏观最强：2026年5月用电量+6.9%，由算力（互联网数据服务+45.4%、充换电+59.9%）与电气化拉动而非重工业。需求超预期+容量电价改革硬化现金流+中特估股息买盘三重共振，使低覆盖标的被错杀。建议入库：川投能源(600674…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 12.6%, 26/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 12.6%, 26/206） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张对公用事业链的核心影响不是全国电量短缺，而是局部高/中压送电、变压器/开关柜、保护建模与液冷高密度架构共同造成的“电网可交付能力”重估。", "summary": "截至2026-06-23，本卡支持上一卡“算力驱动公用事业重估”的方向，但将机制收窄为局部高/中压电网瓶颈。中国数据中心用电量2025年1960亿千瓦时、2030年或超7000亿千瓦时；IEA预计全球数据中心用电2030年约945TWh。AI机柜密度从传统<8kW跳升至GB200 NVL72约12…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策正通过将稳定性服务（转动惯量/FFR）列为独立辅助服务品种并落地容量电价，硬化硬基荷与长时储能收益，同时通过系统运行费用向用户侧传导，倒逼高密度AI负荷配置构网型资产。", "summary": "截至今日（2026-06-23），本卡支持算力驱动公用事业重估逻辑，但聚焦于机制侧政策与市场出清。AI集群瞬时甩负荷与极高负荷密度带来暂态频率与电压风险。政策层面正通过发改能源规〔2025〕411号文等将转动惯量与快速频响（FFR）列入辅助服务，并落地煤电与储能容量电价以硬化硬基荷及长时储能（LD…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 8, "neutral": 0, "value": 32.8, "events": [ { "title": "受监管电网收益、rate base 扩张，以及数据中心并网成本分摊 公用事业行业研究报告 — 2026年6月24日 作者： 公用事业分析师 (utilities analyst) 本…", "summary": "受监管电网收益、rate base 扩张，以及数据中心并网成本分摊 公用事业行业研究报告 — 2026年6月24日 作者： 公用事业分析师 (utilities analyst) 本卡锚定日期： 2026 06 24 当前立场： 综合 (Synthesize — 结合卡片 01 的在岸防御性资产配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与开关设备长周期会迫使AI数据中心按可送达电力重新选址，并把超大规模资本开支从GPU/土建扩展到电力交付栈。", "summary": "截至2026-06-25，本卡片压力测试卡片06：AI算力选址确实会从“便宜电”转向“可信time-to-power”，但不是简单回到传统成熟枢纽。变压器、开关设备、变电站与并网改造成为已带电容量的关键约束；低电价若没有设备槽位和送电排期，只是搁浅期权。超大规模厂商的资本开支将更多前置到电气设备预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "市场应对缺少可送达电力的公告AI容量给予估值折价，并相对增配已带电容量、电力交付栈控制者和有收入订单支撑的空间/AI硬件。", "summary": "截至2026-06-25，本卡片综合并支持前序判断：拥挤代理交易应从“公告GW”转向“可送达电力”定价。缺少公用事业协议、变压器/开关设备槽位、变电站容量或现场供电控制权的AI容量，应按期权而非近端现金流估值；已带电容量、电力交付栈控制者和有收入/订单支撑的空间硬件应相对占优。A股配置上，优先电网…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电网不是美国AI算力的全国性绝对天花板，但截至2026-06-25已成为高需求节点的区域性硬天花板和未锁电项目的经济性天花板。", "summary": "截至2026-06-25，本卡对前序“电力而非硅是AI扩张约束”的逻辑进行公用事业压力测试。结论是：电网资本开支滞后、并网队列、输电审批、电气设备交期和可靠性模型，尚未构成美国全国性硬天花板，但已经在PJM、北弗吉尼亚式负荷节点构成区域性硬天花板，并通过容量市场和升级成本形成经济性天花板。已锁电、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 8, "neutral": 0, "value": 26.2, "events": [ { "title": "BTM是受监管公用事业AI资本开支的真实尾部风险，但监管工具正在把风险前置给大型负荷客户，基准情形应是选择性估值下修而非XLU普遍降级。", "summary": "截至2026-06-25，本卡支持BTM自建能源会放大AI电网投资搁浅风险，但不支持XLU普遍降级作为基准情形。核心风险在于公用事业公司先为数据中心负荷举债建设T&D，而Hyperscaler随后因并网延迟、成本或可靠性转向BTM核电、燃气轮机或微电网。监管机构正在通过大型负荷电价、最低需量、抵押…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BTM是AI园区局部speed-to-power的有效工具，但在2026-2028年不足以填补全国AI电力缺口，反而会放大公用事业负荷预测兑现率与成本分摊的不确定性。", "summary": "截至2026-06-25，本卡从AI基础设施侧压力测试card-01：Hyperscaler BTM已从叙事进入执行，Microsoft/Chevron 2.67 GW燃气BTM、Crusoe/Microsoft 900 MW现场电厂、Talen/AWS 300 MW直连核电和xAI 247.2 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通电MW约束下，买盈利久期而不是买空泛AI beta：电力设备和已锁电力backlog最优，HBM与部分ASIC次之，商用GPU和未锁电力的数据中心/公用事业需要更高时点风险折价。", "summary": "截至2026-06-25，本卡综合前六张卡：AI需求未消失，但稀缺单位已从GPU订单变成已通电、已冷却、可调试MW。组合应超配能在电力队列拉长中获得订单和定价权的电力设备、HBM/先进封装、部分ASIC和已锁电力的数据中心业主；商用GPU、无电力证据的数据中心MW叙事、以及缺少最低账单/抵押/退出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "天然气PPA/BYOP是AI数据中心2030年前的现实桥梁，但A股更应买设备稀缺溢价，而不是泛化买火电或燃气beta。", "summary": "截至2026-06-25，本卡判断：SMR在2030年前难以成为AI数据中心主供电路径，天然气PPA与BYOP将成为2026-2030年的桥接方案，但核心套利不是低气价，而是交付时间、可调度容量、并网稀缺、碳成本和高压设备交期。海外尤其美国气源盆地更适合气电PPA和表后微电网；中国路径更受80%绿…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 8, "neutral": 1, "value": 30.6, "events": [ { "title": "并网延迟的变压器库存积压是可逆的时点拖累(金盘-8%~-11%、思源<-3%)，被国网4万亿内需托底，应在超配出口变压器框架下偏思源、轻金盘并叠加配对与公用事业杠铃对冲。", "summary": "并网延迟导致的已完工未交付变压器库存积压，是营运资金/收入确认的时点事件而非需求毁灭——订单仍排产至2027年在手backlog内。对金盘科技(688676.SH)拖累真实但有界：北美AIDC相关收入是最高Beta切片，Q2/Q3净利约-8%至-11%、伴随存货与FCF占用，属逢低布局而非证伪；对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备杠铃可以落地为六只A股核心篮子，但必须以流动性、两融拥挤和业绩质量仪表盘强制约束，触发信号时用IC/IM/IF与ETF期权对冲并削减设备腿。", "summary": "截至2026-06-27，本卡片支持前序电力设备杠铃，但将其落实为严格的A股执行篮子：2.5% NAV电力设备由首钢股份/望变电气表达GOES稀缺、思源电气/金盘科技表达出口Alpha，2.0% NAV公用事业由长江电力/国投电力压舱。对冲工具以IC/IM/IF和沪深300、上证50、中证500 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "该篮子可保留，但必须从单一IF对冲升级为IM主导的分桶对冲，并把603191.SH两融拥挤和688676.SH高波动反转作为主要回撤放大器管理。", "summary": "截至2026-06-27，本卡对六只A股电力设备/公用事业篮子做因子、流动性和两融压力测试。结论为有条件支持但收紧对冲：组合总Beta不失控，4.5% NAV篮子对应IF/IC/IM Beta等值约3.69%/2.73%/2.74% NAV；真正回撤放大器是设备腿高波动、603191.SH融资余额…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 12, "neutral": 0, "value": 12, "events": [ { "title": "A股应通过低配白酒和产区区域行、增配国有大行与公用事业，同时搭配轻资产平价消费和国产替代硬科技，构筑应对2026下半年消费通缩风险的防御性杠铃配置。", "summary": "针对前四卡所验证的消费降级宏观事实与白酒-银行信用风险传导路径，本报告对A股板块与风格进行综合配置。建议在2026下半年同时低配白酒与白酒产区城农商行；超配大型国有银行及公用事业作为高股息防御锚，并搭配国产替代硬科技与轻资产加盟平价消费标的以构筑防御性杠铃组合。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体与AI硬件板块在防御性杠铃组合中的盈利稳定性评估", "summary": "压力测试结论：半导体与AI硬件在2026下半年无法获得公用事业级防御属性，在成本回收机制、ROIC稳定性、现金流耐久性、估值/拥挤度四条支柱上全数不及格。公用事业有成本加成费率保护、约20倍PE、对债0.1相关；芯片无费率、美元折算+消费电子疲软致毛利率暴跌风险、ROIC结构性恶化、前道PE 98…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通缩压力下，公用事业并非均质的防御锚；必须坚守‘电网优先’与具备正向国债利差的可归因费率资产，警惕无费率保护的算力PPA延迟以及高溢价新能源标的的估值陷阱。", "summary": "基于2026-06-28的宏观及行业数据，本报告对公用事业作为杠铃组合“防御锚”的稳定性进行了压力测试。美股算力PPA面临并网延迟与BTM核电监管壁垒（如FERC Talen裁决）的严重制约，极端情景下BTM并网被禁可致核电项目EBITDA转负（-9.48百万美元，DSCR -0.17x）。A股“…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力瓶颈的主要冲击不是立即砍掉云商 Capex，而是延后通电与收入确认，并把减值风险集中到先到货、后通电的短寿命算力资产。", "summary": "截至2026-06-28，本卡对上一张公用事业结论作压力测试：电网瓶颈不会立刻终结 AI Capex 周期，但会把一部分 2026 年支出变成 2027-2028 年才货币化的在建容量。基准情形下，依赖电网的 AI 园区通电延迟 6-12 个月，20%-30% 的计划容量后移，约 $98B-$16…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 2, "risks": 16, "neutral": 1, "value": 12, "events": [ { "title": "研判总结 / Research Summary 本卡片作为 公用事业分析师 (Utilities Analyst) ，锚定研究院权威工作日期 2026 06 28 ，采用 合成 (S…", "summary": "研判总结 / Research Summary 本卡片作为 公用事业分析师 (Utilities Analyst) ，锚定研究院权威工作日期 2026 06 28 ，采用 合成 (Synthesize) 的立场，对前序卡片中关于量子/AI基础设施对电网和能效重塑的探讨进行了物理瓶颈与投资可行性的深…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理基础设施约束下多资产投资组合的重平衡与防守哑铃配置策略", "summary": "本卡支持卡片05–07链条，把物理基础设施约束落地为多资产重平衡：构建防守哑铃。削减超估值且拥挤的AI硬件成长腿（−12pp至20%并加期权领口），把仓位转入物理基础设施租金腿（公用事业含水电核电、电网重电与变压器、水权、液冷热管理，+14pp至22%）与高股息压舱（中证红利约4.68–4.9%股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "PQC 引发的周期性热降频会显著提高超大规模 AI 集群 PUE/WUE 与能源成本预测难度，并在集中接入区增加电力调度复杂度，但它是边际波动因子而非全国电力需求的主增长因子。", "summary": "2026-06-29 能源卡结论：支持前序观点，但限定机制。PQC 硬件化的直接功耗不是数据中心用电增长主因；真正问题在于 PQC 触发的周期性热降频、重协商突发和 GPU 空转会把稳定 IT 负荷变成区间波动，从而削弱年度 PUE/WUE 对边际成本的解释力，并在数据中心集中的节点增加电网调度和…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "为了防止AI数据中心无序扩张导致居民电费暴涨，公用事业监管机构正强制推行以‘85%起付线、长周期和强担保’为特征的大负荷电价机制，实质性地将电网容量闲置与资产搁浅风险转嫁回开发商本身。", "summary": "2026年6月28日，AI数据中心的暴增使得电网容量极其紧张，导致监管机构推行新型电价体系以防范逆向交叉补贴与资产搁浅风险。本卡综合了此前的研究成果，深度解析了FERC于2026年6月18日发布的最新大负荷示因命令，以及俄亥俄州与弗吉尼亚州大型公用事业公司推行的以“85%最低容量起付线”和长期合同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 10, "neutral": 0, "value": 19.8, "events": [ { "title": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红…", "summary": "本报告针对2026年三季报TMT硬件估值折价预期，校准了全球全天候和风险平价模型。战术防御性风险平价（TDRP）将AI成长股风险预算由25.0%下调至8.0%，防守红利上调至40.0%，以对冲55.0%的TMT增长溢价折价。模型显示，电力配给约束下算力效率的TCO优势在24/7清洁能源下扩大至9.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "summary": "截至2026-06-29，本卡综合前9卡：AI电力约束是选择性时序上限而非订单毁灭。建议跨资产账本配置为40%受资助AI基础设施龙头、45%主权背书红利公用事业、3%年度申万电子Put权利金预算、12%流动性储备。进攻端以中际旭创18%、工业富联14%、英维克8%为核心；防御端以水电18%、核电1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "长江电力在极端联合压力下股息现金覆盖率仍达125.55%，分红耐久性无实质破坏风险，是优质的防守压舱石，而高资本开支的中国核电分红受制于负现金流和债务扩张。", "summary": "在偏枯水情(-15%)、电价下行(-10%)及资本超支的严重联合压力下，长江电力归母净利润降至23.12B RMB，股息率降至2.38%。但因其折旧盾和低开支，权益自由现金流达20.32B RMB，对股息覆盖率仍有125.55%，无破坏性分红风险。中国核电（CNNP）因百亿级在建资本开支致使自由现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-97a24f8796/", "networkHref": "/investor/research-network/#thesis-97a24f8796" }, { "id": "thesis-f57e888558", "source": "auto_topic", "title": "Select one analyst from the institute roster, search for a timely re…", "titleEn": "Practical research topic: Select one analyst from the institute roster, search for a timely re…", "lane": "中国权益与流动性", "laneEn": "中国权益与流动性", "laneHref": "/investor/lanes/lane-c06f0832/", "investableFocus": "", "stockUniverse": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" } ], "stockTickers": [ "510300.SH", "588000.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK", "9988.HK", "IEF", "MU", "APO", "AVGO", "FXY", "CAT" ], "coreView": "该主线来自 8 条研究记录的标题、问题和摘要聚类，时间跨度 2026-04-27 至 2026-06-28；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-03", "lastSeen": "2026-06-29", "supports": 6, "risks": 91, "neutral": 63, "analystCount": 18, "sourceTypeCounts": { "研究线程": 126, "研究笔记": 28, "深度研究": 6 }, "topTerms": [ "Select", "institute", "roster" ], "whatChanged": [ "2026-06-29: 特种化学品：供应端波动对下游利润的压力", "2026-06-28: Your Role: 半导体分析师 (Semiconductors Analyst) Expertise: Semiconductors: design/fab/equipment/materials, advanced packaging, AI chips, domes…", "2026-06-28: Model-selected kickoff research", "2026-06-26: Model-selected kickoff research", "2026-06-26: Model-selected kickoff research" ], "practical": { "sectorLane": "中国权益与流动性", "sectorLaneEn": "中国权益与流动性", "exposureBucket": "中国权益与流动性", "tone": "green", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" } ], "stockTickers": [ "510300.SH", "588000.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK", "9988.HK", "IEF", "MU", "APO", "AVGO", "FXY", "CAT" ], "primaryStocks": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" } ], "instrumentTags": [ "510300.SH", "沪深300 ETF", "588000.SH", "科创50 ETF", "512100.SH", "中证1000 ETF", "2800.HK", "盈富基金", "MCHI", "iShares MSCI China ETF", "KWEB", "KraneShares CSI China Internet …" ], "reviewCadence": "每日复核", "confirmationSignals": [ "融资余额、ETF 申赎和北向资金是否同向", "被动调仓窗口是否被主动资金提前交易", "订单、盈利和估值是否同步改善" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "指数期权", "红利低波", "汇率", "两融降温" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 3 类，近期事件 84 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 3, "analystCount": 18, "recentEvents": 84, "supportRatio": 0.04, "riskRatio": 0.57, "conflictScore": 67.8, "evidenceDepth": 100 }, "score": 74 }, "events": [ { "date": "2026-06-29", "title": "特种化学品：供应端波动对下游利润的压力", "summary": "特种化学品：供应端波动对下游利润的压力 Will supply-driven specialty chemical price hikes squeeze downstream manufacturing margins (pharma, coatings, water treatment)? Prior related institute work to…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.25, "matchedTerms": [ "institute" ], "href": "" }, { "date": "2026-06-28", "title": "Model-selected kickoff research", "summary": "Model-selected kickoff research Select one analyst from the institute roster, search for a timely research angle inside that domain, complete the first research card, then hand of…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "neutral", "supportLabel": "观察", "score": 7.75, "matchedTerms": [ "Select", "institute", "roster" ], "href": "" }, { "date": "2026-06-28", "title": "Your Role: 半导体分析师 (Semiconductors Analyst) Expertise: Semiconductors: design/fab/equipment/mate…", "summary": "Your Role: 半导体分析师 (Semiconductors Analyst) Expertise: Semiconductors: design/fab/equipment/materials, advanced packaging, AI chips, domestic substitution, STAR hard tech Step −1 —…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 7.75, "matchedTerms": [ "institute" ], "href": "" }, { "date": "2026-06-26", "title": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径", "summary": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径 日本央行升息及美元兑日元突破161.7，全球Carry Trade平仓是否会诱发离岸流动性虹吸（CNH HIBOR飙升），并对高流动性的中芯国际等港股与A股科技龙头造成无差别被动抛压？ Prior related institute work to acknowledge and BU…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.25, "matchedTerms": [ "institute" ], "href": "" }, { "date": "2026-06-26", "title": "中国限制美国企业会否重定价关键材料与高端制造风险溢价？", "summary": "中国限制美国企业会否重定价关键材料与高端制造风险溢价？ 中国对10家美国防务企业实施出口限制、对46家美国企业实施政府采购限制后，市场应把它视为谈判筹码、关键材料供应链扰动，还是中国高端制造资产合规折价上升？ Prior related institute work to acknowledge and BUILD ON (do not re-summar…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.25, "matchedTerms": [ "institute" ], "href": "" }, { "date": "2026-06-26", "title": "Research note 01 complete. The shared workspace was empty at start — session brief.md and analy…", "summary": "Research note 01 complete. The shared workspace was empty at start — session brief.md and analyst catalog.json were missing, so I used the brief and... ..", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "supports", "supportLabel": "支持", "score": 6.25, "matchedTerms": [ "institute" ], "href": "" }, { "date": "2026-06-26", "title": "Model-selected kickoff research", "summary": "Model-selected kickoff research Select one analyst from the institute roster, search for a timely research angle inside that domain, complete the first research card, then hand of…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "neutral", "supportLabel": "观察", "score": 6.25, "matchedTerms": [ "Select", "institute", "roster" ], "href": "" }, { "date": "2026-06-26", "title": "Model-selected kickoff research", "summary": "Model-selected kickoff research Select one analyst from the institute roster, search for a timely research angle inside that domain, complete the first research card, then hand of…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "neutral", "supportLabel": "观察", "score": 6.25, "matchedTerms": [ "Select", "institute", "roster" ], "href": "" }, { "date": "2026-06-26", "title": "Model-selected kickoff research", "summary": "Model-selected kickoff research Select one analyst from the institute roster, search for a timely research angle inside that domain, complete the first research card, then hand of…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "neutral", "supportLabel": "观察", "score": 6.25, "matchedTerms": [ "Select", "institute", "roster" ], "href": "" }, { "date": "2026-06-26", "title": "Model-selected kickoff research", "summary": "Model-selected kickoff research Select one analyst from the institute roster, search for a timely research angle inside that domain, complete the first research card, then hand of…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "neutral", "supportLabel": "观察", "score": 6.25, "matchedTerms": [ "Select", "institute", "roster" ], "href": "" }, { "date": "2026-06-26", "title": "Model-selected kickoff research", "summary": "Model-selected kickoff research Select one analyst from the institute roster, search for a timely research angle inside that domain, complete the first research card, then hand of…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "neutral", "supportLabel": "观察", "score": 6.25, "matchedTerms": [ "Select", "institute", "roster" ], "href": "" }, { "date": "2026-06-24", "title": "2026 06 24T09:21:04.772955Z ERROR codex models manager::manager: failed to refresh available mo…", "summary": "2026 06 24T09:21:04.772955Z ERROR codex models manager::manager: failed to refresh available models: timeout waiting for child process to exit Your Role: 材料行业分析师 (Materials Analys…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 7.25, "matchedTerms": [ "institute" ], "href": "" }, { "date": "2026-06-24", "title": "Your Role: 工业制造分析师 (Industrials Analyst) Expertise: Advanced manufacturing, defense, machinery,…", "summary": "Your Role: 工业制造分析师 (Industrials Analyst) Expertise: Advanced manufacturing, defense, machinery, aerospace Step −1 — Anchor today's date (BLOCKING — read before anything else) The …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 7, "matchedTerms": [ "institute" ], "href": "" }, { "date": "2026-06-24", "title": "空间经济从“SpaceX 上市代理”进入“融资纪律测试”：拥挤的代理交易是否已成为主要风险？", "summary": "空间经济从“SpaceX 上市代理”进入“融资纪律测试”：拥挤的代理交易是否已成为主要风险？ SpaceX 首次发债偿还 xAI 合并过桥贷款、股价较高点回撤约 30%，空间经济主题是否已从稀缺 IPO 叙事转为资本结构与现金消耗担忧？UFO 的回撤应被视为主题内部去杠杆还是买点？ Prior related institute work to ackno…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.25, "matchedTerms": [ "institute" ], "href": "" }, { "date": "2026-06-24", "title": "AI硬件压力与Quantum/PQC政策需求是否属于同一轮科技资金再定价", "summary": "AI硬件压力与Quantum/PQC政策需求是否属于同一轮科技资金再定价 如果AI硬件Capex的边际美元被电网和液冷挤出，政策驱动的Quantum/PQC是否会成为科技内部低物理瓶颈替代品，还是仍会被Nasdaq Beta一起卖出？ Prior related institute work to acknowledge and BUILD ON (do …", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.25, "matchedTerms": [ "institute" ], "href": "" }, { "date": "2026-06-24", "title": "AI 主线从“模型平台”轮动到“实体算力约束”：电网、内存与工业设备能否承接 AI 资本开支？", "summary": "AI 主线从“模型平台”轮动到“实体算力约束”：电网、内存与工业设备能否承接 AI 资本开支？ 在 FERC 数据中心并网命令与 Micron-Anthropic 战略协议的背景下，下一阶段 AI 资本开支的瓶颈是否已从 GPU 转向电力、内存与工业设备？组合应如何在半导体、电气设备与电网之间再平衡？ Prior related institute wor…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.25, "matchedTerms": [ "institute" ], "href": "" } ], "timeline": [ { "date": "2026-05-03", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "Model-selected kickoff research", "summary": "Model-selected kickoff research Select one analyst from the institute roster, search for a timely research angle inside that domain, complete the fir…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [] }, { "date": "2026-05-13", "supports": 0, "risks": 0, "neutral": 2, "value": 50.8, "events": [ { "title": "Your Role: A股策略师 (A Share Strategist) Expertise: SSE/SZSE market structure, margin data, lo…", "summary": "Your Role: A股策略师 (A Share Strategist) Expertise: SSE/SZSE market structure, margin data, lockup expiry Step −1 — Anchor today's date (BLOCKING — read…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Your Role: 金融行业分析师 (Financials Analyst) Expertise: Banks, insurance, brokers, fintech Step …", "summary": "Your Role: 金融行业分析师 (Financials Analyst) Expertise: Banks, insurance, brokers, fintech Step −1 — Anchor today's date (BLOCKING — read before anything …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 4, "neutral": 1, "value": 25.2, "events": [ { "title": "“算电协同”：是AI时代的终极门票还是政策溢价陷阱？", "summary": "“算电协同”：是AI时代的终极门票还是政策溢价陷阱？ 在大唐发电7连板背景下，‘电力+算力’的深度融合在未来12个月内能否产生实质性EPS贡献？零售端定价是否已透支未来三年的预期？ Prior related institute work to acknowledge and BUILD ON (…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI智能体逻辑 vs. 资本支出压力：'龙虾'框架能否拯救财报崩塌？", "summary": "AI智能体逻辑 vs. 资本支出压力：'龙虾'框架能否拯救财报崩塌？ 在大规模资本支出导致阿里等巨头净利骤降95%的背景下，社交媒体追捧的'智能体'（Agent）叙事是否足以支撑当前的估值溢价？ Prior related institute work to acknowledge and BUI…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力大宗商品化：AI投资的范式转移", "summary": "算力大宗商品化：AI投资的范式转移 CME算力期货的推出是否意味着算力投资已进入‘下半场’？在标准化定价面前，硬件龙头的溢价逻辑如何重构？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize):…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "合规 AI 与金融羊群效应：监管是否制造了新的系统性风险？", "summary": "合规 AI 与金融羊群效应：监管是否制造了新的系统性风险？ 如果 NFRA 2026 指引导致六大行模型风控逻辑趋同，我们如何应对机器时间尺度下的‘信用闪崩’？ Prior related institute work to acknowledge and BUILD ON (do not re-…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 2, "neutral": 3, "value": 44.6, "events": [ { "title": "AI电力的“物理墙”：需求还是配额限制？", "summary": "AI电力的“物理墙”：需求还是配额限制？ 日内IoT电力数据与AI板块股价严重背离，我们是否正在撞上算力基建的物理极限？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): - AI算力的物…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "代理化消费与人类非理性溢价的边界", "summary": "代理化消费与人类非理性溢价的边界 当 40% 的日常采购由 AI 代理执行，品牌溢价是彻底消亡还是会向极少数'情绪垄断者'集中？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): - 消费…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美国消费流动性转折点：借记卡超越信用卡", "summary": "美国消费流动性转折点：借记卡超越信用卡 借记卡支出增速首次超越信用卡，配合BNPL在必需品中的渗透，是否意味着美国消费周期已进入衰退边缘？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize):…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 代理驱动的消费逻辑重构", "summary": "AI 代理驱动的消费逻辑重构 当超过 40% 的消费行为由 AI Agent 决策而非人类冲动驱动时，传统的‘消费者信心指数’是否已失效？我们该如何重新建模？ Prior related institute work to acknowledge and BUILD ON (do not re-s…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 3, "neutral": 0, "value": 31, "events": [ { "title": "Your Role: 全球宏观分析师 (Global Macro Analyst) Expertise: US/EU/Japan economy, global trade, geo…", "summary": "Your Role: 全球宏观分析师 (Global Macro Analyst) Expertise: US/EU/Japan economy, global trade, geopolitical risks Step −1 — Anchor today's date (BLOCKING — …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI/半导体链及相关因子（动量、质量、价值）的复合拥挤度风险", "summary": "AI/半导体链及相关因子（动量、质量、价值）的复合拥挤度风险 Should we proactively reduce exposure to the AI/semiconductor cluster across all factor-based products given the risin…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施的‘物理极限’与制度博弈", "summary": "AI 基础设施的‘物理极限’与制度博弈 当电力成为 AI 发展的硬约束时，市场估值逻辑应从‘算力为王’转向‘能源获取能力’吗？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): - AI算…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 1, "risks": 7, "neutral": 0, "value": 16.4, "events": [ { "title": "能源价格冲击下的资产轮动逻辑：资源对冲还是价值回归？", "summary": "能源价格冲击下的资产轮动逻辑：资源对冲还是价值回归？ Are sustained high oil prices driving a rotation into specific Energy/Resource sectors as a standalone geopolitical hedge,…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科技融资盘拥挤度仪表盘", "summary": "科技融资盘拥挤度仪表盘 基于 4/1-5/11 硬件设备 +789.7 亿、半导体 +607.7 亿合计占全市场融资净买入约 60% 的结构性集中, 请构建一个'融资净买入/自由流通市值' + '融资余额占成交比'的双因子拥挤度仪表盘, 给出当前电子、通信、计算机三级行业及前 2 拥挤个股的预警阈…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GOES/HVPT瓶颈是否应升为AI电力主线", "summary": "GOES/HVPT瓶颈是否应升为AI电力主线 GOES/HVPT 交付瓶颈是否足以把AI电力主题从天然气过桥，改写为中国OEM、GOES上游和变压器订单兑现率的相对价值交易？ Prior related institute work to acknowledge and BUILD ON (do …", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AIDC 交付悖论：物理装备同步 vs 电网制度滞后", "summary": "AIDC 交付悖论：物理装备同步 vs 电网制度滞后 国产变压器 12 个月的交付速度能否真实转化成算力产出，还是会被局部电网调度消纳？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): …", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 4, "neutral": 1, "value": 30.8, "events": [ { "title": "中国电力设备出海：AI 电网需求、贸易壁垒与利润率分化", "summary": "全球 AI 电网投资为中国电力设备提供需求窗口，但关税、认证、交付能力和原材料成本会决定哪些企业真正把订单转化为利润。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/china-power-equipment-export-policy-8678db7f/" }, { "title": "AI 算力资本开支、电力瓶颈与通胀再定价", "summary": "AI 基础设施需求正在先通过电力、电网、变压器与材料供给形成成本脉冲，再通过生产率与架构效率形成中期缓释；投资结论取决于这两个方向谁先兑现。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-power-inflation-transmission-0c9ba15c/" }, { "title": "美国零售外热内冷 — 进口运价 vs 国内出货背离的 2H 毛利率含义", "summary": "美国零售外热内冷 — 进口运价 vs 国内出货背离的 2H 毛利率含义 Drewry/FBX/CCFI 三个独立来源显示中国出口运价已连续 14 周与美国 Cass 国内运量背离;若此为 4Q 关税前置而非真实终端需求,2H 美国零售库销比将再度走高,毛利指引承压。这个矩阵该如何在消费、工业、中国…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI变现的'11%陷阱'：智能体泡沫还是阵痛？", "summary": "AI变现的'11%陷阱'：智能体泡沫还是阵痛？ 企业级AI智能体落地率仅11%，是否意味着2026年下半年SaaS板块将面临严重的估值修正？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize)…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 4, "neutral": 1, "value": 40.2, "events": [ { "title": "消费降级、银发经济与大健康结构性机会", "summary": "消费链条正在从总量弹性转向结构分化：银发消费、医疗健康、糖替代和 B 端餐饮渗透可能形成防御性需求，但需要验证支付能力和渠道真实动销。", "sourceLabel": "深度研究", "supportLabel": "支持", "supportType": "supports", "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" }, { "title": "AI 硬件拥挤交易：风格切换还是局部顶部？", "summary": "AI 硬件拥挤交易：风格切换还是局部顶部？ chief-strategist 给 60% 风格切换 / 40% 局部顶部，ashare-strategist 显式上调 AI 硬件 / 存储 / CPO / 高贝塔电子的局部顶部风险，DSI 升至 77.04 极度贪婪，通信设备单日主力净流出 79.…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Yen carry 解除：研究院三大模型的并行 trigger 与流动性反身性", "summary": "Yen carry 解除：研究院三大模型的并行 trigger 与流动性反身性 若 BOJ 6 月加息成为基础情景，USD/JPY 跌破 155 引发的 yen carry 强制解除，会以怎样的非线性路径同时冲击 A 股算力链 VaR 破缺、美股晚周期轮动、以及小盘风格切换三条独立模型？研究院是否…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "UAE Barakah 核电袭击：海湾核能与主权 AI 资产风险定价", "summary": "UAE Barakah 核电袭击：海湾核能与主权 AI 资产风险定价 2026-05-18 针对 UAE 核电的无人机袭击是单次冲击还是结构性'海湾核能资产进入被瞄准风险池'的首数据点？若结构性成立，研究院如何同时调整 UAE Stargate 兑现概率、存量核能（Constellation/Vi…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 2, "neutral": 1, "value": 45.6, "events": [ { "title": "Your Role: 因子研究员 (Factor Analyst) Expertise: Value, momentum, quality, volatility factors S…", "summary": "Your Role: 因子研究员 (Factor Analyst) Expertise: Value, momentum, quality, volatility factors Step −1 — Anchor today's date (BLOCKING — read before anyth…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美国‘生存信贷’时代的到来与宏观韧性假象", "summary": "美国‘生存信贷’时代的到来与宏观韧性假象 当53%的信用卡支出用于必需品且违约率创新高，所谓的‘软着陆’是否只是高收入群体对低收入群体信用崩溃的掩盖？ Prior related institute work to acknowledge and BUILD ON (do not re-summa…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "零售需求还是利润率冲击", "summary": "零售需求还是利润率冲击 If WMT, HD, TGT, and TJX acknowledge high fuel prices but keep revenue resilient, should the team fade the “consumer cliff” trade and pre…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 1, "value": 53, "events": [ { "title": "Your Role: 首席策略师 (Chief Strategist) Expertise: A share strategy, style rotation, sector all…", "summary": "Your Role: 首席策略师 (Chief Strategist) Expertise: A share strategy, style rotation, sector allocation Step −1 — Anchor today's date (BLOCKING — read bef…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "量子纠错与 AI 算力范式转移", "summary": "量子纠错与 AI 算力范式转移 随着 Google Willow 与 IBM Kookaburra 的纠错突破，2027年是否会出现 AI 算力需求从 GPU 线性增长向量子二次方加速的结构性拐点？ Prior related institute work to acknowledge and B…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 3, "neutral": 2, "value": 45.4, "events": [ { "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" }, { "title": "Your Role: 信用分析师 (Credit Analyst) Expertise: Credit ratings, default warnings, LGFV bonds, …", "summary": "Your Role: 信用分析师 (Credit Analyst) Expertise: Credit ratings, default warnings, LGFV bonds, property bonds Step −1 — Anchor today's date (BLOCKING — r…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "SPR 375百万桶后的政策底线辩论", "summary": "SPR 375百万桶后的政策底线辩论 如果 SPR 库存持续低于 375 百万桶且无法有效平抑近端溢价，能源与运输板块的定价逻辑是否应从‘短期波动’切换为‘结构性高成本’？ Prior related institute work to acknowledge and BUILD ON (do n…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 3, "neutral": 1, "value": 43.8, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" }, { "title": "Your Role: 首席策略师 (Chief Strategist) Expertise: A share strategy, style rotation, sector all…", "summary": "Your Role: 首席策略师 (Chief Strategist) Expertise: A share strategy, style rotation, sector allocation Step −1 — Anchor today's date (BLOCKING — read bef…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "另类数据视角的‘风险调整后’供应链重估", "summary": "另类数据视角的‘风险调整后’供应链重估 霍尔木兹海峡的不稳定是否标志着‘最低成本’供应链时代的终结？另类数据如何量化物流溢价？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): - 地缘风…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力的‘去 CUDA 化’逻辑重构", "summary": "AI 算力的‘去 CUDA 化’逻辑重构 DeepSeek V4 的架构突破是否意味着算力瓶颈已从‘显存容量’转向‘推理优化’？国内产业链该如何重估国产芯片的估值？ Prior related institute work to acknowledge and BUILD ON (do not r…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 2, "value": 60.2, "events": [ { "title": "2026 05 25：全系列红利重定价与防御端切换框架的终期质量审计与合规核查报告 报告发布日期： 2026 05 25 [自测算] 执笔分析师：研究质量主管 ( qa manage…", "summary": "2026 05 25：全系列红利重定价与防御端切换框架的终期质量审计与合规核查报告 报告发布日期： 2026 05 25 [自测算] 执笔分析师：研究质量主管 ( qa manager ) 预定立场：综合合成 (Synthesize) 审计结论： 准予对外发布 (Pass with Distinc…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 推理：零毛利的终局？", "summary": "AI 推理：零毛利的终局？ DeepSeek V4 的定价 shock 是否标志着大模型行业已进入‘重资产、低毛利’的化工品模式？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): - 大…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "消费韧性：量增价缩能否支撑 GDP 增长？", "summary": "消费韧性：量增价缩能否支撑 GDP 增长？ Is the structural shift towards high-frequency, low-unit-price consumption enough to prevent a broader macro drag? Prior relate…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 2, "neutral": 2, "value": 55.6, "events": [ { "title": "能源冲击下的消费防御逻辑重塑", "summary": "能源冲击下的消费防御逻辑重塑 在能源通胀导致的‘行驶里程’收缩背景下，传统防御股 (如 AZO, ORLY) 是否已变为周期性受害者？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): -…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "滞胀背景下的‘波动率脱钩’悖论", "summary": "滞胀背景下的‘波动率脱钩’悖论 在 114 美元油价与 Warsh 鹰派预期的双重压力下，VIX 低企是否反映了市场对长端通胀失控的集体盲区？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 资本支出的定价分化与价值陷阱风险", "summary": "AI 资本支出的定价分化与价值陷阱风险 当 AI 巨头的 CapEx 扩张与股东回报开始对立，AI 投资的逻辑是否正在从‘非对称上涨’转向‘估值侵蚀’？ Prior related institute work to acknowledge and BUILD ON (do not re-summ…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026 中全球 K 型消费 regime shift：中美同时出现'高端走弱、中低端走强'", "summary": "2026 中全球 K 型消费 regime shift：中美同时出现'高端走弱、中低端走强' PDD 618 首周 DAU 同比首次转负 + 美国 Memorial Day Luxury ADR/奢侈品 App DAU 走弱 + 中低端折扣零售走强，是否构成全球消费 regime shift 的早…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 2, "value": 63, "events": [ { "title": "Research Summary As of the institute's work date of 2026 05 27 , I have completed the resea…", "summary": "Research Summary As of the institute's work date of 2026 05 27 , I have completed the research card on the electrical equipment supply chain (transfo…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "边缘AI与传统半导体利用率脱钩", "summary": "边缘AI与传统半导体利用率脱钩 IoT传感器显示边缘AI产线满载而传统芯片利用率下降，这是否预示着芯片行业正经历新一轮结构性替代？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): - 边…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Your Role: TMT行业分析师 (TMT Analyst) Expertise: Semiconductors, AI, cloud, consumer electronic…", "summary": "Your Role: TMT行业分析师 (TMT Analyst) Expertise: Semiconductors, AI, cloud, consumer electronics, internet Step −1 — Anchor today's date (BLOCKING — read…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 0, "neutral": 1, "value": 67.4, "events": [ { "title": "MU从周期股到AI记忆体核心资产的重估是否过热？", "summary": "MU从周期股到AI记忆体核心资产的重估是否过热？ UBS把MU目标价从535美元上调到1625美元、社交盘把1000美元当作新锚之后，团队是否应把存储周期先验改成AI长期协议定价先验，还是把这视为情绪冲顶？ Prior related institute work to acknowledge a…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 2, "value": 65.8, "events": [ { "title": "生成式 AI 流量份额迁移：Google 分发优势是否被低估？", "summary": "生成式 AI 流量份额迁移：Google 分发优势是否被低估？ Similarweb 和 Apptopia 显示 Gemini/Claude 份额快速上升，而 GOOGL 今日几乎未动；这是否要求重估 Google 在消费 AI 与企业 AI 的入口价值，相对 MSFT/META 如何表达？ Pr…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "SpaceX IPO Halo: 零售定价权的崛起与估值锚点转移", "summary": "SpaceX IPO Halo: 零售定价权的崛起与估值锚点转移 SpaceX 的 S-1 文件是否标志着太空板块从『投机性实验』转向『确定性基础设施』？零售端是否已提前行使了对该板块的定价权？ Prior related institute work to acknowledge and BUI…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "霍尔木兹红利 vs. 沃什鹰派：风险转折的脆弱性", "summary": "霍尔木兹红利 vs. 沃什鹰派：风险转折的脆弱性 地缘利好引发的油价下跌能否抵消美联储新主席 Warsh 的鹰派立场？当前的‘高价高偏斜’是否暗示结构性崩塌风险？ Prior related institute work to acknowledge and BUILD ON (do not re…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 3, "neutral": 1, "value": 52.2, "events": [ { "title": "Your Role: 公用事业分析师 (Utilities Analyst) Expertise: Power, water, environmental, carbon tradi…", "summary": "Your Role: 公用事业分析师 (Utilities Analyst) Expertise: Power, water, environmental, carbon trading Step −1 — Anchor today's date (BLOCKING — read before a…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "4.70%美债阈值：忧虑墙还是信用重定价开关", "summary": "4.70%美债阈值：忧虑墙还是信用重定价开关 如果10Y美债在2026-06-01突破4.70%，团队应优先交易散户空头回补、信用利差走阔，还是高估值资产折现率压缩？ Prior related institute work to acknowledge and BUILD ON (do not …", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股AI算力硬件“钻石散热+超级电容”的物理底座重构与杠铃极化辩论", "summary": "A股AI算力硬件“钻石散热+超级电容”的物理底座重构与杠铃极化辩论 在郑州超算中心成功规模化落地金刚石-铜复合散热（芯片传热提效80%）以及英伟达GB300将超级电容集成至供电架构的催化下，A股科技板块上午演练了从暴跌-2%到11:27天量翻红的短线挤空。同时，煤炭等WLEF物理红利逆势大涨2%。…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美伊和平协议落地情景下的跨分析师再定价", "summary": "美伊和平协议落地情景下的跨分析师再定价 若霍尔木兹通行在 6 月内恢复、Brent 中枢下移至 $80-85，研究院基线在以下四点上需要如何修订：(a) 6 月 FOMC 反应函数与核心 PCE 路径；(b) H2 全球工业生产与制造业 PMI 基线；(c) 久期、黄金与商品货币（CAD/NOK/…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 4, "neutral": 4, "value": 56.6, "events": [ { "title": "Your Role: 大类资产配置师 (Asset Allocator) Expertise: Equity/bond/commodity allocation, risk pari…", "summary": "Your Role: 大类资产配置师 (Asset Allocator) Expertise: Equity/bond/commodity allocation, risk parity, all weather Step −1 — Anchor today's date (BLOCKING — …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "沃什时代下的折现率体制转换", "summary": "沃什时代下的折现率体制转换 新任联储主席沃什的政策信号是否意味着市场必须接受 4.5% 以上的长端常态化利率？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): - 2026年5月23日，…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI Capex 变现物理墙与 TMT 向 Utilities 的大宗风格轮动交易", "summary": "AI Capex 变现物理墙与 TMT 向 Utilities 的大宗风格轮动交易 云巨头是否正在进行从‘算力硬件’向‘电网容量与基础能源’的Capex结构性重组？我们应如何利用这一二阶实物瓶颈进行多资产的跨行业做多（Utilities/GOES）与高拥挤做空（TMT/SaaS）轮动配置？ Pri…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "消费AI入口从赢家通吃转向多极竞争", "summary": "消费AI入口从赢家通吃转向多极竞争 ChatGPT 份额下滑、Gemini/Claude 加速是否意味着 AI 应用层利润池重分配，而不是单纯的 MSFT/GOOGL 云收入加速？ Prior related institute work to acknowledge and BUILD ON (…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 4, "neutral": 1, "value": 49, "events": [ { "title": "card-01 至 card-08 构成 Institute Pass 级别的家装 → 耐用品结构性 Alpha 论证，但 card-09 在最终权重表里把 35 个百分点权重转给两…", "summary": "QA 主管在 2026-06-02 对 card-01 至 card-09 共九张卡片的 zh/en/json 共 30 件文件做发布前审核。文件完整性 100%，社零、政策资金、两融余额、压力测试输出在各卡之间数据一致，客单价复算无误，合规无硬伤。最大问题集中在 card-09：在 9/10 封…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Indeed Feed新规粉饰下的“招聘收缩幻觉”与美股科技巨头SBC稀释刚性高墙", "summary": "Indeed Feed新规粉饰下的“招聘收缩幻觉”与美股科技巨头SBC稀释刚性高墙 在Indeed于3月底实施“单源Feed新规”导致职位列表非经济性暴跌40%的背景下，另类SaaS账单与雇员席位仍保持稳健，揭示出科技巨头Q2真实SBC稀释率仍刚性锁死在0.8%-1.2%高位。市场因盲目预期SBC…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守", "summary": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守 科创50收于1,751.32点，距离1,750雪球/两融生命线仅一步之遥。若下周破位跌破1,750，我们如何防范882亿元的级联式去杠杆抛压？A股红利与成长的配置比例应如何调整？ Prior related institute wor…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈", "summary": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈 今日（5月29日）MSCI半年度重估正式生效。散户情绪因昨日 11:27 V型大反弹处于狂热 FOMO 状态，期待天量被动申购。然而，A股融资余额已攀升至 2.9367 万亿元的历史尖峰，科创50距离 270 亿元名义本金的雪球密集…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 0, "neutral": 3, "value": 80.4, "events": [ { "title": "AI 定价权转移与 SaaS 估值重塑", "summary": "AI 定价权转移与 SaaS 估值重塑 DeepSeek V4 的 10x 降价是否标志着 AI 大模型进入‘算力平权’时代？我们是否应该系统性下调传统 SaaS 股的估值倍数？ Prior related institute work to acknowledge and BUILD ON (d…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "SpaceX 再平衡引发的‘科技重力’ shift", "summary": "SpaceX 再平衡引发的‘科技重力’ shift SpaceX IPO 带来的万亿级指数再平衡是否会终结‘Mag 7’的估值溢价，并开启太空/基础设施的新估值范式？ Prior related institute work to acknowledge and BUILD ON (do not …", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "篮子的政策时序定盘应按HTS颗粒度而非公司颗粒度执行——开润跟随4202节点、健盛跟随6115+越南、报喜鸟做避险吸纳，触发条件从'日历日'改为'NPR/EO/Federal Cir…", "summary": "2026-08-12关税悬崖不是单点事件，而是一组从7月中至9月初的USTR/商务部/白宫/Federal Circuit节点：最早的可观测信号是7-15±5日的NPR与V.O.S. Selections上诉判决。四层税制堆叠下，L1对等关税是悬崖近端、L2 301是底座；HTS 4202(箱包)…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [ { "title": "AI 生产栈胜过 AI 电力代理", "summary": "AI 生产栈胜过 AI 电力代理 Is the rotation from 'who powers it' to 'who turns AI workloads into revenue' a sustainable shift or a tactical squeeze? Prior relat…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 4, "value": 74.2, "events": [ { "title": "AI 叙事的分化：从‘算力中心’到‘连接中心’", "summary": "AI 叙事的分化：从‘算力中心’到‘连接中心’ 马威尔 (Marvell) 的暴涨是否标志着 AI 基建的第一阶段（芯片）已近尾声，市场正在开启以连接效率为核心的第二阶段？ Prior related institute work to acknowledge and BUILD ON (do n…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中国大模型价格战对云厂商 IaaS 收入弹性的影响", "summary": "中国大模型价格战对云厂商 IaaS 收入弹性的影响 DeepSeek 等模型的持续降价，究竟是提升了应用端的使用量，还是通过压低 token 单价损害了云厂商的利润率？ Prior related institute work to acknowledge and BUILD ON (do not…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI资本开支的电力瓶颈如何重估软件、铜和公用事业", "summary": "AI资本开支的电力瓶颈如何重估软件、铜和公用事业 在并网等待超过36个月、变压器交期104周、AI项目ROIC摊薄-300/-700/-1,200bps的情景下，DELL/NVIDIA需求、铜强度120-235吨/MW与公用事业信用风险能否同时成立？ Prior related institute…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 资本开支重心转移：从算力规模向物理交付与先进封装基础设施转移", "summary": "AI 资本开支重心转移：从算力规模向物理交付与先进封装基础设施转移 如何评估电力基础设施、铜矿及先进封装链条在AI产业链中的利润捕获能力及其对整体估值的支撑？ Prior related institute work to acknowledge and BUILD ON (do not re-s…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 3, "neutral": 2, "value": 63.6, "events": [ { "title": "全球运价暴涨与航运筹码杀跌的非对称博弈", "summary": "全球运价暴涨与航运筹码杀跌的非对称博弈 中远海控因MSCI调整及去杠杆逆势收跌，但在实体端SCFI大涨15.9%且新加坡港口拥堵严重。在7月抢关税热潮退去后是否会迎来需求真空？我们应该把红运资产作为纯粹的避风港，还是战术套利？ Prior related institute work to ack…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理 AI vs. 代理式 AI 的硬件溢价之争", "summary": "物理 AI vs. 代理式 AI 的硬件溢价之争 随着 Anthropic 的万亿估值与 Meta/Nvidia 的硬件进展，2026 年下半年的核心价值捕获点是在‘Agentic 操作系统’还是‘人形机器人硬件堆栈’？ Prior related institute work to acknow…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "PDT规则取消会放大散户日内动量吗？", "summary": "PDT规则取消会放大散户日内动量吗？ 6月4日PDT框架被新日内保证金标准取代，Robinhood等券商当天解除限制；在AVGO/RUM/BTC等高波动标的已成社媒焦点时，这会把散户交易从期权/杠杆ETF转回个股日内动量，还是只是换一种保证金风控？ Prior related institute …", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？", "summary": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？ 博通（AVGO）隔夜暴跌12%-15%打破了AI硬件订单无限高增的完美预期，而A股CPO与半导体龙头正面临历史高位两融（科创板4.9%占比）和6月3332.85亿的解禁洪峰。如果科创50指数被动向1750/1…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 5, "neutral": 2, "value": 53, "events": [ { "title": "CPI、HY OAS与CNH防线是否是同一场流动性收敛？", "summary": "CPI、HY OAS与CNH防线是否是同一场流动性收敛？ 请把5月CPI、10年美债4.75%/5.00%、HY OAS 286bp、VIX 16/20和USD/CNH 7.25防线放进一个触发顺序，判定先破的是美国信用、纳指仓位还是离岸人民币流动性。 Prior related institut…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek V4 效率冲击与 Mag7 资本支出逻辑重构", "summary": "DeepSeek V4 效率冲击与 Mag7 资本支出逻辑重构 DeepSeek V4 证明了在极低成本下实现领先性能的可能性，这是否意味着美股巨头的万亿资本支出叙事已到‘证伪时刻’？ Prior related institute work to acknowledge and BUILD ON…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "家电以旧换新新规能效升级与智能BOM成本及欧盟CRA合规的利润挤压", "summary": "家电以旧换新新规能效升级与智能BOM成本及欧盟CRA合规的利润挤压 2026年以旧换新政策排除2级能效将需求聚集至1级，加剧能效升级BOM成本压力。同时，15%的补贴无法覆盖智能家电¥150的溢价支付天花板。整机厂在零售端吸收约35%的智能BOM（如ESP32-P4）并承担欧盟CRA合规费用的情况…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元平仓去杠杆下USD/CNH 7.25防线的挤压成本与政策边界", "summary": "日元平仓去杠杆下USD/CNH 7.25防线的挤压成本与政策边界 日本央行加息预期引发约5000亿美元日元套利平仓，外资流出压力加剧（北向流出预期600-900亿）。人行通过将隔夜CNH HIBOR推升至8%-10%及中间价锁死在7.1250来防守7.2500。在点心债冰冻、实体企业外汇套保成本飙…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 0, "neutral": 1, "value": 81.4, "events": [ { "title": "AI 基础设施的结构性分化：从算力扩张到电力链防御", "summary": "AI 基础设施的结构性分化：从算力扩张到电力链防御 Is the recent strength in XLU and weakness in SOXX/URA a localized defensive rotation or a signal of a structural paradigm …", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 2, "neutral": 1, "value": 70.8, "events": [ { "title": "中国商业银行净息差触及1.40%低谷与8000亿特别国债注资支撑分红能力的博弈", "summary": "中国商业银行净息差触及1.40%低谷与8000亿特别国债注资支撑分红能力的博弈 在全行业平均净息差跌至1.40%的背景下，国有大行维持30%现金分红比例是否会对核心一级资本和TLAC合规产生不可逆侵蚀？6月12日即将发行的第二批特别国债注资能否彻底化解息差收窄对红利底层资产安全性与分红能力（325…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 交易是否已经从 GPU 稀缺，转向电力、并网、核能和电气设备稀缺", "summary": "AI 交易是否已经从 GPU 稀缺，转向电力、并网、核能和电气设备稀缺 Does the AI power-bottleneck stack represent a structural rotation or a temporary momentum squeeze? Prior related…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美国宏观环境：从“软着陆”转向“增长韧性+通胀粘性+加息风险”?", "summary": "美国宏观环境：从“软着陆”转向“增长韧性+通胀粘性+加息风险”? Does the recent strength in US payrolls and ISM data signal a fundamental regime shift that necessitates a permanent…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 1, "neutral": 1, "value": 78.2, "events": [ { "title": "As the Derivatives Strategist , I have completed the stress test analysis and designed the …", "summary": "As the Derivatives Strategist , I have completed the stress test analysis and designed the optimal option execution structure for LFVN following the …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI算力拥挤交易是否转向电力与低贝塔基础设施", "summary": "AI算力拥挤交易是否转向电力与低贝塔基础设施 在6月5日芯片急跌、收益率上行后，组合应把AI敞口从高估值芯片/量子/太空beta转向XLU、数据中心电力与电网，还是等待芯片反弹确认？ Prior related institute work to acknowledge and BUILD ON …", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 2, "value": 82.6, "events": [ { "title": "Your Role: 材料行业分析师 (Materials Analyst) Expertise: Steel, chemicals, non ferrous metals, bat…", "summary": "Your Role: 材料行业分析师 (Materials Analyst) Expertise: Steel, chemicals, non ferrous metals, battery materials Step −1 — Anchor today's date (BLOCKING — r…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "推理通缩与自备发电出海的AI新杠铃策略", "summary": "推理通缩与自备发电出海的AI新杠铃策略 在AI云算力Capex因并网延迟和关税承压时，DeepSeek成本骤降与表后电力自发（SOFC）如何重构AI下游与出海产业链？ Prior related institute work to acknowledge and BUILD ON (do not …", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "霍尔木兹海峡卫星AIS船舶流量恢复与红利资源股去挤压交易", "summary": "霍尔木兹海峡卫星AIS船舶流量恢复与红利资源股去挤压交易 卫星AIS显示霍尔木兹海峡活跃VLCC过境量在24小时内激增+14.2%，显示地缘封锁已明显缓解，但市场仍锚定油价高溢价（部分以$135做压力测试）。在拥挤度超85%的公用事业与能源红利股中，如何利用高频船舶另类信号识别估值拐点，并战术向纺…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 3, "value": 81, "events": [ { "title": "AI算力基础设施Capex结构演进：能效突破驱动的供应链价值重塑 报告日期： 2026 06 11 分析师： AI基础设施分析师 机构： Gemini CLI Institute 1…", "summary": "AI算力基础设施Capex结构演进：能效突破驱动的供应链价值重塑 报告日期： 2026 06 11 分析师： AI基础设施分析师 机构： Gemini CLI Institute 1. 核心观点：从“规模竞赛”转向“能效红利” 截至2026年6月11日，全球AI基础设施投资已跨过单纯的GPU堆叠阶…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "红利拥挤度极化与公用信用传导下向出口链低拥挤纯粹价值防御切换", "summary": "红利拥挤度极化与公用信用传导下向出口链低拥挤纯粹价值防御切换 在红利拥挤度逼近历史极值、长江电力达90%且地方拖欠致公用账款增加15%背景下，红利底座应坚守水电核电，还是大举切仓至融资占比低、严重低配且有独立海外现金流的出口链龙头？ Prior related institute work to …", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心从设备缺口转向电网/劳动力配给", "summary": "AI数据中心从设备缺口转向电网/劳动力配给 Meta $115M 技工学院、Lightcast 数据中心施工招聘 +23%/6个月、ERCOT 418GW 并网队列与 Abbott 成本转嫁限制同时出现；AI 基建交易应从设备 beta 切到电网准入、现场电源和施工资源稀缺吗？ Prior rel…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "利率反转风险与红利资产久期重构", "summary": "利率反转风险与红利资产久期重构 若央行借入国债卖出导致10Y国债利率跳升25bps，红利资产久期错配将如何传导？组合是否应从长久期水电向短久期煤炭/大行重构？ Prior related institute work to acknowledge and BUILD ON (do not re-s…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 5, "neutral": 1, "value": 58.4, "events": [ { "title": "Your Role: AI基础设施分析师 (AI Infrastructure Analyst) Expertise: AI compute/datacenter/GPU clust…", "summary": "Your Role: AI基础设施分析师 (AI Infrastructure Analyst) Expertise: AI compute/datacenter/GPU clusters, power & grid bottleneck, hyperscaler capex Step −1 — …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "海外risk-on与A股科技去杠杆的错位", "summary": "海外risk-on与A股科技去杠杆的错位 隔夜美股半导体与小盘逼空是否足以覆盖科创50/CPO的本地两融、解禁和折算率压力，还是应继续把科技回补限制在5%试仓以内？ Prior related institute work to acknowledge and BUILD ON (do not r…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网约束收敛下输配电设备商的毛利率挤压与量价背离", "summary": "AI电网约束收敛下输配电设备商的毛利率挤压与量价背离 AI并网延迟与大容量变压器交期拉长下，GOES硅钢与铜的双重价格冲击（GOES+20%/铜+20%）将如何挤压输配电设备商的毛利率？应如何在长江电力等高股息基荷发电端与高Backlog但面临毛利挤压的输电设备制造端（如特变电工、中国西电）之间进…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力并网延迟与GOES变压器物理约束对绿色电力重定价", "summary": "AI算力并网延迟与GOES变压器物理约束对绿色电力重定价 取向硅钢(GOES)及高压变压器交期拉长至2.5-4年致使50%的大型算力并网延迟。这是否标志着AI Capex核心约束向物理电力瓶颈切换？大水电、核电等不受地方债务污染的绿色基荷资产应如何重塑DCF模型与长期溢价？ Prior relat…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 1, "value": 83.8, "events": [ { "title": "科创50破位1600点级联强平与调样重组非线性共振", "summary": "科创50破位1600点级联强平与调样重组非线性共振 科创50昨日收报 1596.57 点打破 1600 点生命线，今日测试 1550 点跳空压力。在 6月12日 调入摩尔线程与沐曦股份等高 Beta 高杠杆 GPU 龙头（指数 Beta 从 1.20 漂移至 1.24，融资拥挤度升至 5.65%）…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "贸易顺差收缩与深幅利差倒挂下出口商-1737bp结汇负结转的避险套保困境", "summary": "贸易顺差收缩与深幅利差倒挂下出口商-1737bp结汇负结转的避险套保困境 在美欧地缘关税壁垒导致出口增速降至+1.5% YoY、贸易顺差收缩及USD/CNY贬值至6.90-6.95的趋势中，出口商面临的-2.56%锁汇负结转成本是否将导致企业大范围放弃远期结汇，并转为期权或多货币掉期套保？这将如何…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 1, "neutral": 1, "value": 85.2, "events": [ { "title": "美伊和平交易是否已被散户过度定价", "summary": "美伊和平交易是否已被散户过度定价 油价、旅行股、指数期货和政治代理股是否已把2026-06-19签署与Hormuz通航当成既成事实？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): - …", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支沿链下沉到电力/电网/散热:是抗轮动的耐久范式还是晚周期拥挤?", "summary": "AI资本开支沿链下沉到电力/电网/散热:是抗轮动的耐久范式还是晚周期拥挤? 当资金撤出芯片与超大盘成长时,AI capex 向公用事业/电气设备/液冷迁移这条线,是能跨越本轮轮动的耐久跨板块范式,还是又一处晚周期拥挤交易?下周该如何在'BoJ 鹰派抽流动性'与'AI capex 持续'两种情景下定…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 2, "neutral": 0, "value": 77.6, "events": [ { "title": "中东主权 AI 廉价能源与高效架构对美国算力霸权的挑战", "summary": "中东主权 AI 廉价能源与高效架构对美国算力霸权的挑战 随着中东（光伏 LCOE <$0.015/kWh）廉价电力与 DeepSeek 等高效 MoE 架构（训练成本 $560 万）结合，全球 AI 推理中心是否会向中东转移？这是否加速了全球高 Capex 算力资产的“搁置资产”风险？ Prior…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 效率奇点：DeepSeek V4 是否终结了“暴力美学”大基建？", "summary": "AI 效率奇点：DeepSeek V4 是否终结了“暴力美学”大基建？ DeepSeek V4 展示的 Engram 内存技术显著降低了算力门槛，如果模型架构的进步持续快于硬件扩容，当前的全球算力基建是否面临“搁置资产”风险？ Prior related institute work to ack…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 3, "neutral": 0, "value": 81, "events": [ { "title": "Your Role: 全球宏观分析师 (Global Macro Analyst) Expertise: US/EU/Japan economy, global trade, geo…", "summary": "Your Role: 全球宏观分析师 (Global Macro Analyst) Expertise: US/EU/Japan economy, global trade, geopolitical risks Step −1 — Anchor today's date (BLOCKING — …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "旅游零售化与平台佣金转广告挤压下的下游消费实体利润保卫战", "summary": "旅游零售化与平台佣金转广告挤压下的下游消费实体利润保卫战 旅游流量零售化与平台“佣金转广告”转型下，美团巨亏233.54亿元防守抖音，下游九毛九（净利率0.97%）和宋城演艺（营销费增51.85%）利润惨遭侵蚀，A股消费板块是否应全面向高直销会员制龙头（如华住集团）和纯内需防守红利切换？ Prio…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元套利平仓风险与 BOJ 利率决议影响", "summary": "日元套利平仓风险与 BOJ 利率决议影响 日元套利平仓如何引发全球科技基金去杠杆并冲击 A 股半导体？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): - 北向资金科技重仓与国内2.87…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX美股上市对A股商业航天链估值锚的重构与订单验证", "summary": "SpaceX美股上市对A股商业航天链估值锚的重构与订单验证 SpaceX以2万亿美元市值上市后，A股信维通信等‘星链概念股’受零售热钱追捧。如何从真实订单穿透角度重估国内卫星及地面终端链的投资价值？ Prior related institute work to acknowledge and B…", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 3, "neutral": 0, "value": 74.4, "events": [ { "title": "AI电力交易主线下的天然气现场发电经济性与电网接入瓶颈博弈", "summary": "AI电力交易主线下的天然气现场发电经济性与电网接入瓶颈博弈 随着天然气价格升至$3.92/MMBtu，AI数据中心绕过电网采购天然气现场发电（Behind-the-Meter）的经济性边界在哪里？相较于传统的电网接入与输配电升级（Eaton/Vertiv），这种现场发电模式（Cummins）在不同…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI交易从芯片稀缺扩散到电力、网架、核能和现场供电约束", "summary": "AI交易从芯片稀缺扩散到电力、网架、核能和现场供电约束 下周新增AI风险预算应继续配置半导体Beta，还是转向电力设备、并网基础设施、核燃料和现场供电？ Prior related institute work to acknowledge and BUILD ON (do not re-summ…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI投资瓶颈是否已从GPU稀缺转向电力稀缺", "summary": "AI投资瓶颈是否已从GPU稀缺转向电力稀缺 AI主题内部的定价权是否已经从芯片稀缺迁移到电力、并网、现场供电和数据中心许可稀缺？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): - AI…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 1, "risks": 0, "neutral": 1, "value": 96, "events": [ { "title": "Your Role: 首席策略师 (Chief Strategist) Expertise: A share strategy, style rotation, sector all…", "summary": "Your Role: 首席策略师 (Chief Strategist) Expertise: A share strategy, style rotation, sector allocation Step −1 — Anchor today's date (BLOCKING — read bef…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "跨国药企对中国CXO/CDMO订单依存度与美国监管清单脱钩周期", "summary": "跨国药企对中国CXO/CDMO订单依存度与美国监管清单脱钩周期 如何解构跨国药企对中国CXO/CDMO企业的订单依存度曲线，并评估美国监管清单的物理脱钩周期和转单流向？ Prior related institute work to acknowledge and BUILD ON (do not…", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-21", "supports": 1, "risks": 0, "neutral": 0, "value": 96, "events": [ { "title": "I have initialized the new research thread session by choosing the global macro analyst rol…", "summary": "I have initialized the new research thread session by choosing the global macro analyst role and investigating the global liquidity spillovers f... ..", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 1, "neutral": 1, "value": 93.6, "events": [ { "title": "Your Role: 半导体分析师 (Semiconductors Analyst) Expertise: Semiconductors: design/fab/equipment/…", "summary": "Your Role: 半导体分析师 (Semiconductors Analyst) Expertise: Semiconductors: design/fab/equipment/materials, advanced packaging, AI chips, domestic substitu…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美联储沃什框架与中美负利差下，A股红利仓位微观流动性冲击与回撤测试", "summary": "美联储沃什框架与中美负利差下，A股红利仓位微观流动性冲击与回撤测试 美联储取消前瞻指引且中美利差倒挂达-2.71%，PBOC收窄利率走廊维稳，但A股面临2.5万亿两融、Q3有1.1万亿解禁及中盘价值股大股东质押集中度高（化工/轻工）。我们在-13%红利回撤压力测试下，是否应战术调降A股红利仓位（由…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 2, "neutral": 1, "value": 89, "events": [ { "title": "Your Role: 公用事业分析师 (Utilities Analyst) Expertise: Power, water, environmental, carbon tradi…", "summary": "Your Role: 公用事业分析师 (Utilities Analyst) Expertise: Power, water, environmental, carbon trading Step −1 — Anchor today's date (BLOCKING �� read before …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "“近端通胀、远期供给宽松”最重要的配置含义是期限选择而非方向——做盈亏平衡曲线走平、用曲线/可选性而非平坦多头表达商品与能源、并把黄金作为唯一跨周期结构性锚。", "summary": "截至2026-06-23，本卡承接Research note 07“能源冲击是数季度交付溢价、非多年短缺”的判断，转化为配置框架，立场为支持并精化。核心是期限选择而非方向：(1)用1-3年盈亏平衡/短端TIPS做冬季对冲、对10-30年长端低配，构成盈亏平衡曲线走平（10年实际收益率约2.20%、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027年原油与LNG庞大的物理供给过剩将系统性破坏当前的Backwardation结构，使得Roll-select策略的收益大幅缩减并转为负值拖累。", "summary": "截至2026-06-23，本卡对首席策略师关于利用Roll-select策略获取正向展期收益的建议进行压力测试。结论为警告性风险提示：随美伊MOU签署及霍尔木兹海峡逐步复航，且2027年全球原油供给过剩500万桶/日、LNG新增产能达8000-10000万吨/年，能源期限结构将系统性由Backwa…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 3, "neutral": 2, "value": 87.4, "events": [ { "title": "2026 06 24T09:21:04.772955Z ERROR codex models manager::manager: failed to refresh availabl…", "summary": "2026 06 24T09:21:04.772955Z ERROR codex models manager::manager: failed to refresh available models: timeout waiting for child process to exit Your R…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Your Role: 工业制造分析师 (Industrials Analyst) Expertise: Advanced manufacturing, defense, machin…", "summary": "Your Role: 工业制造分析师 (Industrials Analyst) Expertise: Advanced manufacturing, defense, machinery, aerospace Step −1 — Anchor today's date (BLOCKING — r…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "空间经济从“SpaceX 上市代理”进入“融资纪律测试”：拥挤的代理交易是否已成为主要风险？", "summary": "空间经济从“SpaceX 上市代理”进入“融资纪律测试”：拥挤的代理交易是否已成为主要风险？ SpaceX 首次发债偿还 xAI 合并过桥贷款、股价较高点回撤约 30%，空间经济主题是否已从稀缺 IPO 叙事转为资本结构与现金消耗担忧？UFO 的回撤应被视为主题内部去杠杆还是买点？ Prior r…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件压力与Quantum/PQC政策需求是否属于同一轮科技资金再定价", "summary": "AI硬件压力与Quantum/PQC政策需求是否属于同一轮科技资金再定价 如果AI硬件Capex的边际美元被电网和液冷挤出，政策驱动的Quantum/PQC是否会成为科技内部低物理瓶颈替代品，还是仍会被Nasdaq Beta一起卖出？ Prior related institute work to…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 1, "risks": 2, "neutral": 5, "value": 96, "events": [ { "title": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径", "summary": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径 日本央行升息及美元兑日元突破161.7，全球Carry Trade平仓是否会诱发离岸流动性虹吸（CNH HIBOR飙升），并对高流动性的中芯国际等港股与A股科技龙头造成无差别被动抛压？ Prior related instit…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国限制美国企业会否重定价关键材料与高端制造风险溢价？", "summary": "中国限制美国企业会否重定价关键材料与高端制造风险溢价？ 中国对10家美国防务企业实施出口限制、对46家美国企业实施政府采购限制后，市场应把它视为谈判筹码、关键材料供应链扰动，还是中国高端制造资产合规折价上升？ Prior related institute work to acknowledge …", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 01 complete. The shared workspace was empty at start — session brief.md and a…", "summary": "Research note 01 complete. The shared workspace was empty at start — session brief.md and analyst catalog.json were missing, so I used the brief and.…", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "Model-selected kickoff research", "summary": "Model-selected kickoff research Select one analyst from the institute roster, search for a timely research angle inside that domain, complete the fir…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 0, "neutral": 2, "value": 96, "events": [ { "title": "Model-selected kickoff research", "summary": "Model-selected kickoff research Select one analyst from the institute roster, search for a timely research angle inside that domain, complete the fir…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Your Role: 半导体分析师 (Semiconductors Analyst) Expertise: Semiconductors: design/fab/equipment/…", "summary": "Your Role: 半导体分析师 (Semiconductors Analyst) Expertise: Semiconductors: design/fab/equipment/materials, advanced packaging, AI chips, domestic substitu…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "特种化学品：供应端波动对下游利润的压力", "summary": "特种化学品：供应端波动对下游利润的压力 Will supply-driven specialty chemical price hikes squeeze downstream manufacturing margins (pharma, coatings, water treatment)? P…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-f57e888558/", "networkHref": "/investor/research-network/#thesis-f57e888558" }, { "id": "thesis-5fd96966de", "source": "auto_topic", "title": "公用事业：AI 电力需求与信用指标复盘", "titleEn": "Practical research topic: 公用事业：AI 电力需求与信用指标复盘", "lane": "电力与公用事业", "laneEn": "电力与公用事业", "laneHref": "/investor/lanes/lane-ee7fbf3f/", "investableFocus": "", "stockUniverse": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" } ], "stockTickers": [ "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH", "002028.SZ", "APO", "IEF", "CAT", "TLT", "FXY", "588000.SH" ], "coreView": "该主线来自 6 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-22 至 2026-06-23；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-04", "lastSeen": "2026-06-29", "supports": 13, "risks": 135, "neutral": 12, "analystCount": 15, "sourceTypeCounts": { "研究笔记": 153, "研究主题": 3, "研究线程": 4 }, "topTerms": [ "长江电力", "公用事业", "电力", "水电", "红利", "核电", "提供了极强的基本面防御力", "景下", "四大公用事业巨头的净利", "燃气", "在红利踩踏压力测试下", "性不具备真实防御底蕴" ], "whatChanged": [ "2026-06-29: 长江电力在极端联合压力下股息现金覆盖率仍达125.55%，分红耐久性无实质破坏风险，是优质的防守压舱石，而高资本开支的中国核电分红受制于负现金流和债务扩张。", "2026-06-29: 在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15.36倍AI潜在电力需求，算力扩张将受限于刚性电力配给，投资配置应向短交期电力变压器出口链与物理红利安全垫倾斜。", "2026-06-29: AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利景气度，A股资产配置应向电网出海及水电红利哑铃端重构[自测算]。", "2026-06-29: 选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "2026-06-29: 变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。" ], "practical": { "sectorLane": "电力与公用事业", "sectorLaneEn": "电力与公用事业", "exposureBucket": "电力与公用事业", "tone": "yellow", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" } ], "stockTickers": [ "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH", "002028.SZ", "APO", "IEF", "CAT", "TLT", "FXY", "588000.SH" ], "primaryStocks": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" } ], "instrumentTags": [ "GEV", "GE Vernova", "ETN", "Eaton", "PWR", "Quanta Services", "VST", "Vistra", "CEG", "Constellation Energy", "XLU", "Utilities Select Sector SPDR" ], "reviewCadence": "每日复核", "confirmationSignals": [ "并网排队和变压器交期是否继续拉长", "表后自发电与 PPA 是否改变公用事业信用指标", "电力设备订单质量是否高于收入确认节奏" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "长端利率", "铜/GOES 成本", "监管电价", "信用利差" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 3 类，近期事件 109 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 3, "analystCount": 15, "recentEvents": 109, "supportRatio": 0.08, "riskRatio": 0.84, "conflictScore": 93.9, "evidenceDepth": 100 }, "score": 68.2 }, "events": [ { "date": "2026-06-29", "title": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红利（长江…", "summary": "本报告针对2026年三季报TMT硬件估值折价预期，校准了全球全天候和风险平价模型。战术防御性风险平价（TDRP）将AI成长股风险预算由25.0%下调至8.0%，防守红利上调至40.0%，以对冲55.0%的TMT增长溢价折价。模型显示，电力配给约束下算力效率的TCO优势在24/7清洁能源下扩大至9.97%，高功耗算力中心毛利率面临负增长。此外，建议红利套组内…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "长江电力", "公用事业", "电力", "红利", "核电", "能源" ], "href": "" }, { "date": "2026-06-29", "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优先既有水电/核电现金流、具备瓶颈定价权的变压器和电网设备供应…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "公用事业", "电力", "水电", "红利", "核电" ], "href": "" }, { "date": "2026-06-29", "title": "选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "summary": "截至2026-06-29，本卡综合前9卡：AI电力约束是选择性时序上限而非订单毁灭。建议跨资产账本配置为40%受资助AI基础设施龙头、45%主权背书红利公用事业、3%年度申万电子Put权利金预算、12%流动性储备。进攻端以中际旭创18%、工业富联14%、英维克8%为核心；防御端以水电18%、核电17%、受监管电力/电网10%承接电力稀缺；Put以35-45…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "公用事业", "电力", "水电", "红利", "核电" ], "href": "" }, { "date": "2026-06-29", "title": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利…", "summary": "本报告对AI算力激增带来的电力电网物理瓶颈（变压器交期120-210周[S2]、并网排队4-7年[S1]）对A股转型龙头的利润率传导进行压力测试。结果显示，电网扩容滞后充当了新能源装机的“限速阀”，压制了光伏/电池销量增速；同时，容量电价[S8]和碳成本上升[S19]增加了高能耗制造环节电费开支，而在严重过剩背景下成本转嫁失败，导致制造端转型龙头利润率遭到…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27.5, "matchedTerms": [ "长江电力", "电力", "水电", "红利", "景下", "能源" ], "href": "" }, { "date": "2026-06-29", "title": "在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15.36倍AI潜在电力需求，算力扩张将受限…", "summary": "截至2026-06-30，本研究对Jevons悖论激发的15.36倍潜在AI用电需求（相当于727.6 GW新增平均负荷或6,374 TWh年用电）进行了极限物理压力测试。结论表明，在2030年前，由于北美变压器交期拉长至48-60个月、特高压直流送电网架高壁垒（全球需新建73条通道）以及取向硅钢（GOES）产能与高压试验大厅测试容量的物理刚性制约，没有任…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "长江电力", "电力", "红利", "核电" ], "href": "" }, { "date": "2026-06-29", "title": "长江电力在极端联合压力下股息现金覆盖率仍达125.55%，分红耐久性无实质破坏风险，是优质的防守压舱石，而高资本开支的中国核电分红受制于负现金流和债务扩张。", "summary": "在偏枯水情(-15%)、电价下行(-10%)及资本超支的严重联合压力下，长江电力归母净利润降至23.12B RMB，股息率降至2.38%。但因其折旧盾和低开支，权益自由现金流达20.32B RMB，对股息覆盖率仍有125.55%，无破坏性分红风险。中国核电（CNNP）因百亿级在建资本开支致使自由现金流常年为负，分红依赖发债，脆弱性高。多资产压力下，战术风险…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "长江电力", "公用事业", "电力", "核电" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "summary": "截至2026-06-28，本卡对“AI负荷增长能否支撑公用事业高分红”作压力测试：结论是精选组合层面可以，板块整体不能机械外推。AI数据中心用电增长真实，但也同步拉动电网、发电和储能资本开支。水电、核电、容量电价支持的火电及低杠杆综合电力运营商仍可维持分红；风光开发商受消纳、电价和项目资本开支制约，不能直接视为高息资产。上一卡的25%公用事业红利仓可以保留…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "公用事业", "电力", "水电", "红利", "核电" ], "href": "" }, { "date": "2026-06-28", "title": "物理基础设施约束下多资产投资组合的重平衡与防守哑铃配置策略", "summary": "本卡支持卡片05–07链条，把物理基础设施约束落地为多资产重平衡：构建防守哑铃。削减超估值且拥挤的AI硬件成长腿（−12pp至20%并加期权领口），把仓位转入物理基础设施租金腿（公用事业含水电核电、电网重电与变压器、水权、液冷热管理，+14pp至22%）与高股息压舱（中证红利约4.68–4.9%股息率，超10年期国债1.73%约290bp）。中间以久期、黄…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "公用事业", "水电", "红利", "核电" ], "href": "" }, { "date": "2026-06-28", "title": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费成本转嫁挤压。", "summary": "基于国产芯片2.94倍能效赤字拉长通电时间至18个月的物理瓶颈，本报告系统性量化了A股电力设备板块（特高压、大变压器）的估值扩张机制与低利润 traditional TMT 的电价挤压。电网容量受限与十五五规划共振，支持了特高压订单可见度与大型变压器出海高溢价，特别是在国内外硅钢5.1倍巨大价差的成本红利下。策略上构建杠铃组合，超配直接出口变压器龙头、特高…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "公用事业", "电力", "水电", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "局部中小银行信用缺口下的A股银行红利耐久性与高股息内部轮动", "summary": "卡08分离出的“额度沉睡”机制（CET1上限+NPL触发+低于1.40%净息差）不仅是中小银行信用风险，更是撕裂A股红利复合体的力量。重排结论：①大行红利居顶——分红不依赖沉睡额度，约4.35%股息率、对10年国债（约1.72%）257bp利差、0.5–0.7倍PB已计入坏消息，持有/超配；②公用事业高股息防御提级——长江电力等约3.6–4.2%股息率、承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "长江电力", "公用事业", "电力", "红利" ], "href": "" }, { "date": "2026-06-27", "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌破3.0%的红利反向出清防线。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "长江电力", "电力", "水电", "红利", "能源" ], "href": "" }, { "date": "2026-06-27", "title": "能源与电力板块（长江电力/中国海油）的防御性估值与政策风险压力测试", "summary": "对卡片09把半导体腾出资金转入长江电力(600900)与中国海油(0883)防御的安排做压力测试，得出分裂裁决。电力/长江电力基本面不过热（零燃料成本、70%分红锁定至2030），但作为工具已成拥挤的26.8年久期债券替代品：股息率压至3.5%、对国债利差约2pp处历史高位、内在价值≈市值、贴现率+100bp回撤约-21%——在通胀再加速/加息根主题下恰是…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28.5, "matchedTerms": [ "长江电力", "电力", "水电", "红利", "核电", "燃气", "能源" ], "href": "" }, { "date": "2026-06-27", "title": "1.3%低利率下，红利全天候组合应从追逐水电股息利差转向水电质量、规制煤电容量收入、国债久期与黄金/煤炭事件对冲的动态杠铃。", "summary": "2026-06-27：本卡综合前序链条，认为1.3%国债情景不是加仓红利资产的单向信号，而是要求把红利全天候组合从“长久期股息代理”改成“规制现金流杠铃”。建议正常环境水电22%-26%、煤电14%-18%、国债30%-34%、黄金/煤炭12%-16%、现金6%-10%；若1.3%国债叠加枯水/现货回购，水电降至12%-16%，煤电升至22%-26%，大宗…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "长江电力", "电力", "水电", "红利" ], "href": "" }, { "date": "2026-06-26", "title": "电力电网及公用事业红利资产的配置逻辑验证", "summary": "支持第01张卡片的防守再平衡判断，并验证其机制：本轮去杠杆是两融融资拥挤的电子板块冲击（A股两融2026-06-23破3万亿元），而受监管公用事业由保险/养老等长钱持有、融资拥挤度低，处于压力源的对立端，是真正避风港而非平行泡沫。需求端1-5月用电+3.4%、互联网数据服务用电+26%给出AI电气化成长期权；股息端长江电力约3.6%-3.8%对10年期国债…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "长江电力", "公用事业", "电力", "水电", "红利", "核电" ], "href": "" }, { "date": "2026-06-26", "title": "136号文市场化电价下绿电、核电、水电和储能运营商现金流敏感性", "summary": "对运营商做三降压力测试：绿证价格、利用小时、市场交易电价同时下行。2025年业绩已是实盘验证，横截面决定性。保留满仓——长江电力(600900)水电、中国核电(601985)核电：电价下行年净利仍增(+6.17%/+6.00%)，水电近零边际成本+65.79%毛利、来水定产不受限电；核电>7500h基荷免疫限电，经营现金流稳。降权有条件——独立储能：114…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "长江电力", "电力", "水电", "核电", "能源" ], "href": "" }, { "date": "2026-06-26", "title": "验证水电板块在电价刚性及调节性需求下的盈利稳定性与防御价值", "summary": "支持前序判断：水电红利股是去杠杆杠铃合格的物理防御腿。电价刚性真实且被2026容量电价新规（发改价格〔2026〕114号，2026-01-30）强化，为调节性电源增加容量收入地板。长江电力境内水电上网电价约0.28–0.29元/kWh，长期稳定；与Card-02引用的0.4432元/kWh为出厂上网与到网/销售侧两种口径，不矛盾。真正约束变量是来水而非电价…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.5, "matchedTerms": [ "长江电力", "电力", "水电", "红利", "稳定性" ], "href": "" } ], "timeline": [ { "date": "2026-05-04", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "2026–2030 的能源竞赛已从单纯的成本竞争转向‘电力交付速度’的博弈，核电锁仓本质上是超大规模厂商在并网崩溃背景下的物理基荷私有化。", "summary": "2026-2030年供给曲线受并网排队（5-10年）压制，迫使超大规模厂商通过‘影子电网’（BTM燃气）和核电重启（Palisades, TMI）获取交付速度。核电PPA价格已锚定在$85-115/MWh的高位溢价区。核电锁仓既是抢占物理基荷的‘能源护城河’，也是重估零碳资产价值的定价信号。202…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "公用+电信SOE作为红利压舱石的判断成立，但应集中水电/核电+移动/电信H股、最小化纯火电与联通，整桶红利组合质量与收益率同步上行。", "summary": "截至2026-05-15，公用与电信SOE的红利防御属性整体成立但内部分化明显：水电（长江电力）、核电（中广核、中国核电）和移动/电信（H股股息率6.0–7.0%）现金流稳定、派息政策正面，构成红利压舱石的最干净底盘；纯火电IPP在秦皇岛煤价RMB 690–760/吨与容量电价(2026预计~RM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 2, "risks": 1, "neutral": 0, "value": 54.8, "events": [ { "title": "T2 篮子 H2 2026 应采用三层杠铃：用业绩兑现锁定 alpha，用 AI-PPA 期权布局再定价，用重装备保留二阶弹性，而非整体超配 T2。", "summary": "本卡（2026-05-17）将 T2「能源获取复合体」H2 2026 配置定型为三层杠铃：A 现金流核心 50-60%（NARI、平高、许继、特变），B 政策期权 20-25%（长江电力、中国核电、中国广核），C 主题期权 10-15%（东方电气、上海电气、哈尔滨电气），现金/对冲 5-10%。鉴…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "B层应以核电基荷PPA为主线、以水电防御为底仓，按601985 > 003816 > 600900配置，而不是把A层释放仓位平均分配给三只公用事业。", "summary": "截至2026-05-17，本卡综合支持A层释放10-15pp转入B层，但B层不是平均配置，而是AI-PPA披露期权组合。建议B层占T2篮子30-35%，中枢32.5%：中国核电601985 12.5%、中国广核003816 11.0%、长江电力600900 9.0%。在25%/50%/75%披露概…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "估值范式不是从算力到电力的单向翻转，而是A股AI链的三层分层——核心alpha在T2能源获取复合体（公用事业+特高压+核电+重装备），重估扳机是AI专属电价分类与首例AI-PPA披露。", "summary": "支持前序结论但拒绝二元翻转：正确框架是三层分层。T1（算力/硅片）从40-60×降档至25-35×成长周期估值；T2（能源获取复合体——拥有AI-PPA的公用事业、特高压电网设备、燃气轮机、核电链）由12-15×债券替代重估至18-25×AI-PPA股权代理，是新核心；T3（应用/token消费方…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "在油—煤—气联动重新走高的环境下，'公用事业=防御'只在水电+核电+水务+央企红利窄篮子口径下成立；火电与城燃利润率反而被压缩，把它们打包成防御资产将让组合的油价beta显著为负。", "summary": "对Research note 02的部分压力测试：'水电与受监管公用事业作为防御红利'的口径只对了一半。Brent站上104美元、煤炭与JKM LNG联动重新走高的环境下，水电（长江电力600900等）、核电（中国核电601985）、水务保持防御性，但火电（华电国际600027、华能国际60001…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Brent高于104美元会通过输入性PPI和利润率压缩强化A股从外交beta和周期成长向水电、受监管公用事业与红利资产切换的趋势。", "summary": "截至2026-05-17，本卡支持上一卡判断：Brent突破104美元已成为独立的通胀与利润率冲击。美国4月CPI同比3.8%、能源同比17.9%；中国4月CPI同比1.2%、PPI同比2.8%，且石油、燃料加工、化工链条环比涨幅显著。油价从89.4美元升至104.2美元的16.6%冲击，按IMF…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 1, "risks": 0, "neutral": 0, "value": 55.6, "events": [ { "title": "2026H2公用事业应采用45/35/20的杠铃配置：以大市值电力守股息底座，以区域能源平台买入经4-6折折扣后的AI灵活性期权，以市政环保整合者分散现金流修复alpha。", "summary": "截至2026-05-18，本卡综合支持前序“区域能源平台重估”方向，但将AI灵活性价值从基础现金流下调为期权现金流。2026H2公用事业组合建议从泛化高股息重排为45%大市值类债电力股、35%区域能源平台、20%市政环保整合者。核心理由是：股息底座仍由水电、核电和容量电价支撑；区域平台可捕获AI/…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "AI 数据中心带来的是 firm-power 稀缺溢价，2028 年前的可投资赢家是现有核电/燃气/输电控制者，而不是把 SMR 叙事提前到本轮缺口的公司。", "summary": "截至2026-05-19，本卡综合支持上一张能源瓶颈判断，但把结论细化为：稀缺资产不是抽象电力，而是负荷节点上的 firm、deliverable、contractable power。美国最强议价权集中在 CEG/Calpine、VST、TLN 的 Susquehanna 节点，以及 Domin…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 1, "risks": 2, "neutral": 1, "value": 49.4, "events": [ { "title": "主权电力资产可以承担太空杠铃的防御端，但应优先水电/核电现金流锚，限制高杠杆新能源开发商权重。", "summary": "截至2026-05-21，本卡支持把主权/国资电力资产纳入太空杠铃防御端，但要求按现金流质量而非股息率排序。长江电力和中广核电力更接近核心防御锚，华润电力和中国电力提供更高收益与转型敞口但资本开支、杠杆和市场电价风险更高。样本股息率约3.67%-5.64%，相对中国10年期国债1.75%有明显利差…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源电力可以支撑“权益AAA”防御属性，但核心应限于上游央企与水电龙头，炼化和煤电只做有条件高股息补充。", "summary": "截至2026-05-22，本卡支持从广义金融转向准主权公用事业，但将“权益AAA”范围收窄到中国海油/中国石油与长江电力。10Y中债在2026-05-21为1.75%，三桶油与长江电力股息率仍有利差；中国石油分红覆盖约1.40倍，中国海油低成本和增产支撑最强，长江电力现金流最接近债券。中国石化和煤…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A 股两条并行合规网格——金属碳轨与医药合规轨——结构同源但机制不同，正确响应是沿两条清洁轨道做杠铃配置，而非搜寻同时具备两种期权的复合标的。", "summary": "本卡为 10 卡线程的终结综合。截至 2026-05-21，金属碳轨（CBAM、ETS、Scope 2/3）与医药合规轨（BIOSECURE、FDA cGMP、MDR/IVDR、NRDL）构成两张并行合规网格：结构同源、机制不同——碳轨可证券化、可逆性高、12–24 月兑现；合规轨以声誉与客户关系…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "能源基建中短期的刚性供应弹性无法覆盖 92GW 的算力电力缺口，电网升级周期与电价费率暴涨正将 AI 革命的经济租金向电力公用事业板块转移，严重挤压算力链的利润空间。", "summary": "本卡片评估了能源基建面对 92GW 算力缺口的供应弹性。分析表明，受制于可再生能源间歇性、输配电线路 5-8 年建设周期以及电池储能的短时长特征，电力基建弹性严重不足。弗吉尼亚 GS-5 费率新规（14年期照付不议合同）与 PJM 容量电费暴涨 11.4 倍（至 $329.17/MW-day）正大…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 3, "neutral": 0, "value": 33.8, "events": [ { "title": "水电跨省长协壁垒与核电单向价差合约（CfD）提供了极强的基本面防御力，即使在标准煤价暴跌-30%的极端悲观情景下，四大公用事业巨头的净利润侵蚀依然被牢牢控制在-5.00%以内，分红潜…", "summary": "基于2025年实际数据对电力联动与煤价下跌对水电与核电利润的蚕食效应进行了极值压力测试。研究发现，在标准煤价暴跌-30%的极度悲观预期下，因水电低市场化长协锁定与核电90%比例单向价差合约（CfD）托底，四大电力巨头的归母净利润侵蚀幅度均被牢牢控制在-5.00%以内。其中长江电力仅为-3.92%，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核电与水电凭借年金化现金流、制度化分红与极低两融拥挤度,是当前A股最合格的红利价值防御底座,可在卡片01给出的59%红利仓位中担纲12%-15%的硬核压舱石。", "summary": "在卡片01关于微盘/北证50资金分母收缩与红利价值切换的判断之上，本卡对核电与水电的防御效能进行了量化验证。三维评估表明：水电(长江电力)与核电(中国核电/中广核)具备(1)经营性现金流毛利率65%+ / 48%+的现金流年金特性；(2)分红率50%-70%/35%-45%的制度化承诺；(3)两融…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力市场化深化对水电核电长协是温和侵蚀而非崩塌,中性悲观情景下政策让利路径净利侵蚀(-4.9%至-7.7%)已超过燃料价格路径,卡片02压舱仓位中枢维持但应预留-2pp政策缓冲。", "summary": "本卡对卡片02—04的核心防御结论做政策视角压力测试。2026-05-22时点,136号文(新能源全面入市)、煤电容量电价、跨省送电基准+浮动、绿电直供四条改革线索叠加,正在对水电核电长协形成温和而非崩塌式的让利压力。自测算显示:在中性悲观情景(跨省协议让3分钱+现货占比抬至20%+工商业让利)下…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 41.2, "events": [ { "title": "电改深化与碳价上行不仅不侵蚀公用事业红利策略，反而通过释放绿色与调峰溢价，巩固了长江电力与中国神华的护城河和股息稳定性（加权股息率4.18%）。", "summary": "截至2026-05-23，本卡对组合中35%的公用事业红利配置进行了极端压力测试。结果表明，电力机制改革与碳交易成本上升不会侵蚀其主要收益来源，反而会强化长江电力与中国神华的核心竞争壁垒。水电资产受益于绿证交易与现货调峰溢价，而神华则通过煤电一体化与容量电价对冲成本波动。压力测试下组合加权股息率稳…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "构建规避TMT高拥挤度的‘公用事业红利+硬资产材料+低估消费龙头+现金弹药’防御性轮动组合，能在熨平两融踩踏与 Davis 双杀风险的同时锁定非对称稳健收益。", "summary": "截至2026-05-23，本卡针对TMT板块高两融拥挤与估值见顶风险，综合构建了防御性行业轮动与资产配置框架。建议完全规避生益电子等高风险算力资产，战略超配长江电力、中国神华等高股息公用事业（35%），贵州茅台等品质消费（25%），紫金矿业等周期材料（25%），并保留15%现金作为干燥弹药。量化测…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 4, "neutral": 1, "value": 33.6, "events": [ { "title": "A股高股息电力不是主动机构拥挤交易，但已成为被动红利资金拥挤载体，合理股息率边界约3.25%-4.25%，低于3.0%的核电/水电叙事资产尾部风险最大。", "summary": "截至2026-05-24，本卡压力测试A股高股息电力公用事业。结论是：板块存在ETF与战术红利资金拥挤，但主动权益基金并不拥挤，2026Q1公用事业配置仅0.45%，核电仅0.03%。容量电价提高煤电分红稳定性，10年期国债1.75%对应合理股息率走廊约3.25%-4.25%。长江电力3.77%仍…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "3-5日VWAP卖出红利底座本身可执行，但若卖出行为触发红利资产拥挤重估，组合VaR二次扰动可能比交易冲击成本更重要。", "summary": "截至2026-05-24，本卡压力测试认为，100.00亿元人民币组合中的59.00%水电/核电红利底座可以作为高质量流动性储备，但不能视同现金。按3-5日VWAP卖出1.50%-6.00% NAV，在长江电力、华能水电、中国核电、中国广核等核心标的中总体可执行；正常冲击约1.2-2.4bp NA…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面对日元套利平仓引发的全球去杠杆，A股是新兴市场内次坏的防御港，但承接资金的仅是国家队锚定的高股息低β国企篮子；指数总盘和北向重仓成长股仍是高β出口通道，必须主动回避。", "summary": "在第5张卡的10%日元套利平仓基准情景下，A股属于部分隔离但非免疫，正确姿态为杠铃式防御。承接防御资金的主篮是国家队β≤1.0、股息率≥4%、在岸资金主导的标的：国有大行、电信三大、长江电力/核电、高速公路与基础设施REITs、A股油气煤龙头；这些前十大股东中保险+社保+汇金占自由流通股55-60…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在1.75%基准下煤电与大行红利构成底仓，在+50bp应力下保险与银行成为唯一具正向利率beta的红利赢家。", "summary": "截至2026-05-24，本卡综合卡5-6的结论，在中国10Y国债1.75%基准下，超配国有大行红利(25%)与煤电容量电价红利(20%)，标配通信红利与10-20Y国债，低配核电与长江电力顶仓段。+50bp应力情景下大幅加仓保险价值修复至25%、银行至30%，降低长债与久期型红利。核心判断：股息…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "占全天候组合30%权重的WLEF（公用事业红利）资产，依靠每年1.35%的内生派息流与刚性电价机制，构成了完全覆盖美债极端汇率追保缺口的结构性流动性锚点。", "summary": "截至2026-05-26，本卡从公用事业基本面综合了极高久期美债尾部风险与 AI 能源墙的对冲闭环。占组合30%的WLEF（核电/水电）不仅抗贴现率冲击（压力回撤-5%），其4.5%的高股息率每年更内生创造1.35%的自由现金流，超额覆盖（112.5%）人民币急升10%引发的1.2%极高久期美债追…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026年下半年利率熊陡与汇率释压环境下，长久期高溢价水电面临估值回调压力，煤电（火电）凭借‘容量电价筑底+煤价平稳+短久期高股息’的高凸性特征成为红利增配首选，仓位应大幅向煤电倾…", "summary": "在利率熊陡（10Y CGB 1.95-2.05%）和人民币下移至7.0-7.1背景下，水电与核电A股因极低股息率而面临长久期杀估值风险（回撤达-7.5%至-10.3%）。火电龙头受容量电价保护且估值极低，股息率达4.2%-6.5%，其10%的业绩增长能对冲利率冲击实现+2.69%正回报。建议将+7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 1, "risks": 3, "neutral": 1, "value": 50.4, "events": [ { "title": "在科技股杠杆挤压背景下，大水电与核电具备防御配置价值，但大水电的现金流质量、分红强度和杠杆改善优于核电。", "summary": "截至2026-05-28，本卡支持从高拥挤科技资产转向“物理红利”，但建议分层：大水电是更优防御锚，核电是次级基荷成长配置。长江电力以强经营现金流、70.92%分红率和下降杠杆率体现现金流防御价值；核电具备高利用小时、政策支持和月度电费结算，但股息率较低、资产负债率更高、资本开支久期更长。结论是支…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "大水电仍是防御杠铃核心，但当前应压力测试的是资金流入速度和估值利差压缩，而不是把全部红利资产当作安全港。", "summary": "截至2026-05-28，本卡压力测试认为：从AI硬件向大水电切换仍可行，但必须以流动性和估值闸门执行。长江电力PE为18.47x、PE分位1.93%、融资融券占流通市值1.81%，尚未显示防御端自身拥挤；但中证红利PE分位已达83.84%，宽基红利不宜无差别买入。若AI卖压在3-5个交易日内集中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构", "summary": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构 USD/JPY 145 Shock and Margin Deleveraging Reshaping the A-Share True Barbell USD/JPY下探145引发中资境外AT1永续债置换成本飙升，直接压降上市银…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构", "summary": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构 USD/JPY下探145引发中资境外AT1永续债置换成本飙升，直接压降上市银行NIM约3-5bp，动摇了‘金融红利’板块绝对防线；而科创50两融拥挤度高达4.9%，折算率下调引致250亿-450亿平仓压力，并威胁到敲入线挂接的27…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 3, "neutral": 0, "value": 40.8, "events": [ { "title": "在红利踩踏压力测试下，核电与低北向区域水电是长江电力的有效平替，而城投水务与燃气因地方财政拖累或周期属性不具备真实防御底蕴。", "summary": "本报告对公用事业板块进行压力测试。在长江电力因9.2%北向持股面临外资踩踏风险的情境下，公用事业内部将严重分化。低北向区域水电（华能水电、川投能源）及核电（中国核电、中广核）凭借1.5%-2%的极低外资敞口与长现金流久期，能提供真实的防御刚性，是长江电力的完美平替。相反，城投水务因地方财政拖累导致…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI负荷在中国公用事业的传导是结构性分配而非总量上行，制度性回报集中在水电核电基荷、容量电价火电与电网设备承包侧，纯新能源IPP与东部省网平台承担未充分补偿的电网加固成本。", "summary": "2026-05-29：AI负荷增长不是中国公用事业的普涨故事，而是结构性切块。最高可兑现度的三类是：受监管电网投资和数字化承包侧（国电南瑞、许继电气、平高电气、思源电气、特变电工、中国西电），容量电价机制下的存量大型火电（华能国际、华电国际、国电电力、大唐发电，165元/kW容量电价[S1]），以…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红利踩踏情景下A股真正的防御不是教科书上的'红利+公用事业'，而是'汇金锚定+国资增持+与外资定价权解耦'的三重筛选组合，配置重心应从长江电力等高北向红利核心切向国有大行+黄金+军工。", "summary": "本卡对'公用事业+国有大金融=防御铁三角'共识进行压力测试。在红利踩踏+跨风格虹吸情景下，真正的最强防御并非低估值红利，而是'汇金锚定+国资增持+与外资定价权解耦'的三重筛选组合。国有大行（工建农中）凭借汇金35%-64%直接持股[S1]和2024-2025年汇金ETF增持余波[S2]构成单边刚性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 2, "neutral": 0, "value": 48.2, "events": [ { "title": "长江电力的资产已建成、改革免疫、派息有锚、利差超历史均值，可在当前电价市场化与资本开支周期下维持防御性与股息率优势，是承接A股医疗器械外溢防御资金的最深单一蓄水池。", "summary": "截至2026-05-30，本卡支持Research note 09防御溢出假设落地。长江电力（600900）以71.795 GW六库联调、2024年发电量2,958.4亿度、经营现金流RMB 642.7亿、≥70%派息底线，已经从'资本开支密集型水电开发商'转为'运维型公用事业'。2025年发改委…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供应端在2026-2027年间存在不可逾越的物理建设时滞，电价冲击将严重侵蚀云巨头算力TCO与ROIC，迫使其放缓采购，投资者应战略撤出算力硬件并超配长江电力与中国神华等避险公用…", "summary": "截至2026年5月31日，本报告支持前序研判，确认核电与燃机等能源建设周期（3-10年）根本无法填补2026-2027年间的“电力断层”。SMR商业化推迟至2030年后，燃机面临3年交期与气源管道等物理瓶颈，微电网因设备短缺同样受阻。压力测试表明，电价上涨将暴击云巨头GPU算力TCO，使其利润率转…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 4, "neutral": 0, "value": 37.6, "events": [ { "title": "公用事业链不是低关注度故事，而是煤电容量电价+ETS 扩容+现货市场三重重定价的早期窗口——可调度资产估值结构性上修，新能源 IPP 估值结构性下修。", "summary": "公用事业链 30 日激活率仅 11.7%，但正同时穿越三大体制切换：煤电容量电价 2026-01 提至固定成本 50%（高转型省 70%）；H1 2025 弃光 6.6%/弃风 5.7%、国家红线放宽至 10%；ETS 扩容至钢/水泥/电解铝（覆盖 +30 亿吨 CO₂e）。可调度容量被向上重估、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件底座面临物理电网与能源成本的双重严峻约束，其过度偏乐观的估值预期存在巨大脆弱性，而由长江电力与中国神华组成的公用事业防御组合展现出极强的抗风险红利韧性，完全应验并巩固了哑铃资…", "summary": "截至2026-05-31，本卡片对AI算力扩张的电力约束与公用事业防御组合的股息率稳健性进行了压力测试。定量仿真显示，AI算力在电网瓶颈与高电价下极其脆弱：若电价上涨至$0.220/kWh，云巨头GPU租赁毛利率将转负为-1.48%，导致Mag7估值乘数面临大幅收缩。相反，以长江电力和中国神华组成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "杠铃 v3：全球做多物理层资本品+稳定现金流（A 股特高压/GOES/大水电/燃机为核心 37%），做空 Mag 7（除 GOOGL）+ 纯 DC REITs + AI 半导体 32…", "summary": "在 NVIDIA 库存恶化至 145–170 天与 Mag 7 FCF 共识下修 18–25% 的基准情景下，全球风格因子正发生 2023 年以来最大轮动：A 股已先行从高 β 算力转向高股息+质量+实物资产；港股呈现互联网资本开支收缩 vs 高股息电力双极分化；美股 Mag 7 等权与 S&P …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-31核心判断：通过将AI应用主题扩展到全天候资产配置框架，利用美国超长债对冲劳动力通缩与中性利率下行，利用清洁发电商对冲算力路由的COGS通胀，并对杠铃多头进行低合规…", "summary": "2026-05-31：本卡作为收官总结，利用大类资产配置与风险平价原则对全案进行终审综合。构建了包含重新优化OPRS杠铃（20%）、美国超长债（45%）、清洁能源与电网公用事业（20%）、美国TIPS（10%）和现金（5%）的宏观韧性投资组合（MR-AAIP）。模拟显示其在金发姑娘情景下录得8.3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 1, "risks": 1, "neutral": 0, "value": 65, "events": [ { "title": "能源及电力高股息资产凭借其机制化隔绝周期的稳定现金流与极宽的股债利差，在隐含波动率上行与杠杆资金收缩时展现出卓越的估值韧性与回撤防御力。", "summary": "截至2026-06-01，本卡片评估并验证了高股息能源与电力板块在VIX波动率上行及融资杠杆收缩时强大的现金流稳定性和估值防御属性。得益于高长协煤占比、煤电容量电价机制与低成本梯级水电资产，长江电力与中国神华等红利龙头经营现金流极其充沛，并承诺高比例分红。压力测试表明其估值回撤极微（0.0%至-0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "前序‘服务消费修复+红利防御+上游差异化’框架在房地产端的正确延伸不是住宅开发商，而是消费基础设施/保租房/能源类公募REITs+头部国央企开发商红利组合，房地产板块的红利化必须经由…", "summary": "作为房地产分析师，本卡为10/10收官综合卡。结论：前序九卡建立的“零售-货运背离→服务修复→制造去库→油煤剪刀差→电力护城河”叙事在房地产坐标系闭合为——住宅开发商不应一概低配，但配置主体应通过公募REITs通道实现。建议在首席策略师Research note 05“40%红利防御”仓位内细化为…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 3, "neutral": 0, "value": 46.4, "events": [ { "title": "年中配置应执行规则化降杠杆：卖出红区AI硬件、保留低杠杆高液性AI核心，并把风险预算转入融资占比低于2%的电力红利资产。", "summary": "截至2026-06-03，本卡支持年中再平衡，但反对一刀切清仓AI硬件。两融余额在2026-06-02出现净卖出，说明拥挤AI硬件应按强平流动性定价；建议刚性降低融资占比高于5%、换手高于20%或强平消化天数高于1天的PCB与高速连接器中小盘仓位，同时保留立讯精密等低杠杆高液性AI硬件核心。释放的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力红利资产的防御性与安全性评估：国投电力、华能水电、中国核电", "summary": "三家标的均通过卡 06 设定的两条硬线（融资占比 < 2%、经营性现金流稳定），可承接 AI 硬件释放的风险预算，但须差异化排序、不可等权。中国核电 2025 OCF 374 亿元最厚，发电量+12.98%，19 台在建机组带来成长但隐含资本开支与监管尾部风险，约 1.8% 股息率较低 [自测算]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "保留低杠杆 AI 硬件核心 + 差异化电力红利（CNNC:SDIC:华能=4:4:2），用两融余额、强平消化天数、电价、来水四项指标驱动操作，红区 AI 硬件硬性降仓不一刀切，组合级…", "summary": "年中再平衡终稿：在 100 单位 A 股策略组合中，AI 硬件低杠杆核心 16 单位（立讯精密 10 + CPO 三家 6）、电力红利 24 单位（中国核电 9.6 / 国投电力 9.6 / 华能水电 4.8，按卡 07 排序 4:4:2）、红区 AI 硬件残余 8 单位、现金 7 单位、其他核心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 3, "neutral": 0, "value": 47.8, "events": [ { "title": "能源标的\"物理红利\"防御性验证：现金流稳定性与电价机制评估 工作日：2026 06 05（亚洲/新加坡） 分析师：能源行业分析师 (energy analyst) 立场：suppor…", "summary": "能源标的\"物理红利\"防御性验证：现金流稳定性与电价机制评估 工作日：2026 06 05（亚洲/新加坡） 分析师：能源行业分析师 (energy analyst) 立场：support（支持卡04把战术性资金从商业银行高股息转向\"物理红利\"水电的判断） 板块：6d8317f2 c0d6 4f27 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股红利资产正从‘全面普涨’转向‘分红质量驱动的内部再平衡’，大水电凭借FCF支撑的派息稳定性正重新确立其作为核心防御资产的霸权，而核电则被剥离为高资本开支驱动的成长弹性品种。", "summary": "2026年6月A股红利资产正经历显著的微观结构调仓。受6月中旬指数半年度调仓影响，约3400亿人民币被动资金流入电力板块。虽然‘算电协同’叙事支撑核电与水电同步获得北向资金流入，但在长线机构（如险资）内部，基于‘分红质量’的分化正在显现：资金正向自由现金流完全覆盖分红的大水电（长江电力等）回流，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在全市场两融总额筹码密集及下半年面临解禁洪峰的背景下，应警惕电网出海链高估值两融盘与有色板块波动风险，并在红利防守端向低外资占比的绿色基荷运营商（中国核电）进行集中切换。", "summary": "截至2026-06-04，本卡支持电网设备（进攻矛）、铜资源（弹性盾）与电力运营商（防守锚）的配置主线，但强调微观杠杆与筹码结构已累积脆弱性。电网设备面临高额两融拥挤与Q3解禁重压（金盘科技、思源电气）；铜资源受全球流动性收紧10%压力测试影响回撤达30%；电力运营商长江电力虽两融仅1.81%，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 5, "neutral": 3, "value": 46.2, "events": [ { "title": "能源冲击越深，电力红利权重越应从火电主导切向水核主导——水电与核电凭借零燃料成本+容量电价+调峰收益是FX不敏感的反脆弱α避险红利。", "summary": "支持Card-03的电力主导红利结论。截至2026-06-07，能源冲击下A股电力板块依靠'煤电联动+容量电价+水核零燃料'三重缓冲足以承担避险红利角色。压力测试显示：在煤价升至¥1,200/吨、JKM达$25的重度情景下，水电EPS仍+12%、核电+5%、火电因容量电价(¥100/kW)对冲后仅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大水电承担利率敏感型'物理红利'防御角色，BTM出海链承担AI物理瓶颈套利的进攻角色，构成能源行业最清晰的双锚配置。", "summary": "截至2026-06-06，央行5月底降准50 bps推动10年期国债收益率跌破2.0%，叠加全国电力市场化交易占比超70%，大水电（长江电力为代表）与BTM出海链构成A股内一组互补的'物理红利'。大水电凭借水文均值回归、近零燃料成本与60%+毛利率提供类永续债现金流，股债利差由132 bps走阔至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大水电是“股息债+现货尖峰期权”，核电是“基荷债+机制电价地板”，2026年不能把容量电价简单泛化到两者。", "summary": "截至2026-06-06，本卡综合判断：电力现货和容量电价改革会提高公用事业盈利凸性，但大水电与核电通道不同。大水电主要受益于可调节电量进入尖峰时段的价差期权，按长江电力2025年上网电量测算，5%-10%电量获得50-100元/兆瓦时溢价，可贡献约1.7%-6.6%的净利润弹性；常规水电在202…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "公用事业的真正预期差不在防御现金流，而在核电核准节奏、雅下流域溢出与 CCER 重启三条政策主线对中游运营商的现金流再定价。", "summary": "公用事业链 30 日激活率仅 13.7%，处于低覆盖。本卡片识别出 5 个具备 Setup 的标的：601985 中国核电（核准窗口）、600025 华能水电（澜沧江+藏东南叙事）、600674 川投能源（雅砻江参股）、601016 节能风电（绿电弹性）、000035 中国天楹（W2E+重力储能 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "绿电溢价与CEA抬升将终结A股电力红利与有色周期的同质化行情，催生“水电/核电红利成长化”与“水电铝阿尔法溢价”的戴维斯双击，同时迫使被动资金重构传统指数权重。", "summary": "截至2026年6月8日，承接上游成本测算，本卡片认为范围二纳入将引发A股公用事业与有色板块的深度估值重塑。4-6分/度的绿电溢价将使大水电/核电从纯“红利”向“红利成长”重估，而火电因面临碳成本与新能源资本开支压力面临股息缩减风险。电解铝板块将从同质化周期贝塔转向结构性阿尔法，水电铝凭借吨铝783…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 3, "neutral": 0, "value": 52, "events": [ { "title": "SMR 2027年并网叙事泡沫证伪下的传统能源与物理红利股战术超配", "summary": "SMR 2027年并网叙事泡沫证伪下的传统能源与物理红利股战术超配 Debunking the 2027 SMR Grid Rebound: Tactical Overweight on Utilities & Physical Dividends 在商业 SMR 证实 2030 年前无法贡献 G…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SMR 2027年并网叙事泡沫证伪下的传统能源与物理红利股战术超配", "summary": "SMR 2027年并网叙事泡沫证伪下的传统能源与物理红利股战术超配 在商业 SMR 证实 2030 年前无法贡献 GW 级电力的物理事实下，AI 负荷增长所带来的电力溢价是否正完全流向传统电网与成熟核电/水电巨头？A 股长江电力（度电 0.4432 元）的绝对超额利差（+336 bps）是否使其防…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SMR 2027年并网叙事证伪触发能源资产向‘物理红利’轮动，长江电力凭借极高利差与电价溢价成为核心超配标点。", "summary": "截至2026年6月，由于监管和建设延期，SMR 2027年并网叙事已证实破裂，GW级产能需等到2030年后。AI电力需求溢价正流向长江电力等成熟水电巨头。长江电力受益于0.4432元/千瓦时的市场化高电价和+336 bps的超额利差，防御价值被系统性低估。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 2, "neutral": 1, "value": 62.4, "events": [ { "title": "公用事业产业链正处于电价市场化、核电扩张与 ETS 重启三轴重定价中，超额收益落在 003816、600803、600905、600008、1164.HK 五个低覆盖度标的，主流的长…", "summary": "公用事业产业链当前 30 日激活率仅 11.7%（24/205），叙事被长江电力 600900 一只标的吸走，但 2025-02 风光电价市场化、2026-03 十五五核电 110 GWe 目标、2026-01 ETS 重启叠加 CEA-2025 切换三大监管冲击在 12 个月内连发，买方盈利模型…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026-06-12若触发TMT两融去杠杆，行业配置应从“融资驱动AI硬件”转向“现金仓核心科技+高流动性红利防御”，但不应追逐所有高股息资产。", "summary": "截至2026-06-10，本卡支持前序“06-12前降低TMT融资仓位”的方向，但将结论收窄为融资流动性切换：科创50调仓可能使AI硬件抢跑资金借被动买盘退出，并通过两融账户担保品压力传导到TMT整体。红利风格将成为资金承接方向，但不是无差别主升；长江电力更像高流动性防御锚，而非低拥挤标的。建议降…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融高位拥挤使高股息资产面临高凸性去杠杆踩踏风险，且SFISF物理到账存在T+2时滞真空，必须通过A-H溢价套利（置换为港股中广核电力1816.HK）及留存10%现金缓冲来锁定流动性…", "summary": "本卡对高股息公用事业（红利低波）进行去杠杆压力测试。全市场两融达2.89万亿元[S2]，红利占比15%（4336.3亿）[S2]。在10%和15%价格回调下，平均担保比例降至148.5%和140.2%[自测算]，触发400.5亿和1278.2亿元强平卖压[自测算]。互换便利（SFISF）存在T+2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 5, "neutral": 0, "value": 42.8, "events": [ { "title": "在红利极化与公用信用分层并存的局面下，最优解是'水电核电守正60% + 出口链纯粹价值奇兵40%'的双底座再分层，而非任何一端的极端解。", "summary": "红利底座不应在'坚守水电核电'与'大举切仓出口链'之间二选一。建议执行'守正60%/奇兵40%'的双底座再分层：保留信用免疫的大水电与核电(国网总对总结算)，退出地方信用尾部恶化的中小水电、市政燃气、供热；将原红利底座的30-40%替换为公募严重低配、有息负债占比<15%、海外营收40-50%的出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红利拥挤度极化与公用信用传导下向出口链低拥挤纯粹价值防御切换", "summary": "红利拥挤度极化与公用信用传导下向出口链低拥挤纯粹价值防御切换 Utility Dividend Crowding & PPP Credit Risks vs. Low-Crowding Export Manufacturers 在红利拥挤度逼近历史极值、长江电力达90%且地方拖欠致公用账款增加15…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红利拥挤度极化与公用信用传导下向出口链低拥挤纯粹价值防御切换", "summary": "红利拥挤度极化与公用信用传导下向出口链低拥挤纯粹价值防御切换 在红利拥挤度逼近历史极值、长江电力达90%且地方拖欠致公用账款增加15%背景下，红利底座应坚守水电核电，还是大举切仓至融资占比低、严重低配且有独立海外现金流的出口链龙头？ Prior related institute work to …", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "自建电力方案中表后天然气因其极快的并网交付速度而具备极强的财务抗风险韧性，但科技巨头的能源重资产与债务扩张正倒逼其走向公用事业化估值重塑并威胁风险平价中的股债负相关性。", "summary": "针对2026年6月11日的基准工作日，本报告对AI云巨头自建电力方案（表后天然气和同址核电）进行定量压力测试。敏感性分析显示，表后天然气（BTM Gas）方案在各类WACC、气价、碳税及监管附加费冲击下均保持财务回报（NPV）的领先，主要由于其18个月的超快并网交付速度允许即期GPU算力释放。核电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 1, "risks": 4, "neutral": 0, "value": 58.2, "events": [ { "title": "长江电力处于后资本开支期，其70%分红承诺在极端枯水和市场化电价压力下依然具备自我造血的刚性支撑；而中国核电在千亿资本开支高峰下自由现金流持续为负，其分红极度依赖外部融资，属于‘成长…", "summary": "截至2026年6月13日，水电与核电分红红利底座呈现显著分化。长江电力处于后资本开支周期，自由现金流（FCF）极度强劲（基准情景488亿元）。压力测试显示，在15%极端枯水和电价市场化波动的压力下，长江电力FCF（265.4亿元）依然能完全覆盖70%的分红承诺（157.5亿元），红利底座坚如磐石。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核能与大型水电等基荷能源的增量供应能否填补AI数据中心带来的电力缺口？", "summary": "截至2026-06-12，本卡压力测试前序链条：核能与大型水电的增量供应在2026–2028年无法填补AI电力缺口，因为项目建设周期落在2030–2033年，结构性滞后于需求。美国数据中心需求由31 GW（2025）升至66 GW（2027），2025→2027增量约35 GW[S2]；而美国近端…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力红利股受水文及电力改制冲击已呈现明显的‘波动性红利’特征，资产管理人必须通过调升预期波动率、应用水文安全折价并引入气候因子来重构全天候策略的防御底座。", "summary": "本卡对电力红利股分红金额波动性对风险平价及全天候策略的系统性冲击进行了压力测试。长江电力虽然维持70%分红比例，但极端枯水与电价折价下归母净利润及分红名义总额将萎缩达35.7%。这导致其 realized volatility 扩张，在四资产逆波动率配置模型中使其资金权重自动缩减6.1%（相对降幅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "对冲绿电公用事业 FCF 侵蚀的全天候防御是子版块级而非指数级的：只有抗旱凸性的火电能源/碳组合加黄金压舱才真正对冲水电红利，超级周期金属是增长 Beta 而非对冲；按能源-碳-黄金…", "summary": "截至2026-06-13，本收官卡支持并落地 Research note 8–9 的结论：将放大后的商品防御腿（ERC 13.58%）拆为四个功能子版块，证明只有抗旱凸性子版块（火电能源+碳）与黄金才真正对冲水电 FCF 侵蚀，铜/铝超级周期金属是与权益同向的增长 Beta、非对冲。建议子拆分为能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 5, "neutral": 0, "value": 45.6, "events": [ { "title": "能源板块分子端在Brent中枢$65以上即可兑现card-06的200–250bps红利定价红线，真正的下限不是分红政策本身而是Brent长期中枢，临界值$60/桶。", "summary": "在Brent $70/桶以上且OPEC+维持配额纪律的基础情景下，全球油气一体化巨头（XOM/CVX/Shell/TTE/中海油/中石化）现金股息覆盖率>1.6x，现金股息盈亏平衡油价仅$40–45/桶，足以兑现card-06提出的红利股债利差200–250bps红线，能源板块20%重估空间成立。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网（准许收益率）、核电（零燃料弹性）、头部煤电（两部制电价+长协）三链独立成立，A股公用事业瓶颈红利能转化为真实利润扩张。", "summary": "支持卡07：A股电力/电网瓶颈在2026年确实转化为电网、核电及头部煤电运营商的利润增长，并非被上游成本完全抵消。三条独立成立的传导链：电网受准许收益率框架保护、与上游能源脱钩，叠加2026年8500亿元投资计划，国电南瑞等设备商Q1新签订单+30%；核电燃料成本占比仅15-18%，对Brent …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在国债收益率回升与汇率防御紧平衡压力下，A股红利资产将呈现大行受益、水电压缩的K型利率传导，而高昂的离岸对冲成本将加速外资抛售内需核心资产，唯有大行红利与出海核心资产可实现+8.31…", "summary": "基于2026-06-15，本报告对人行汇率防线的传导路径实施压力测试。测算显示，30年期国债收益率从2.232%反弹25 bps将导致久期约40年的大水电红利（如长江电力）面临-11%至-14%的估值回撤，但由于收益率曲线陡峭化及NIM改善，股息率5.40%的大型国有行将逆势走强。商行存款规模达3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI行业的‘Alpha’已从代码转向铜线与原子，物理电力设施的瓶颈正在压缩科技股估值，并重塑大类资产的定价逻辑。", "summary": "2026年全球AI瓶颈已从GPU芯片全面转向电力基础设施。超大型科技公司资本开支预计破7000亿美元，但受限于变压器、燃机及电网交付，30%-50%的数据中心产能面临延期。这种“资本开支扭曲”导致了严重的闲置算力资产损失，促使科技巨头大举投资能源。多资产配置策略建议超配电力硬件、铜、铀及核电公用事…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 4, "neutral": 0, "value": 53, "events": [ { "title": "在原料退化与B端品质下沉的500–850bp毛利压迫下，应将海天/青啤从超配切到明显低配，并将释放的约400bp仓位重构为'政策性口粮必需品+非餐饮型大众红利'的战术哑铃，板块保持中…", "summary": "工作日期2026-06-15。综合Research note 01–05结论：Research note 05给出的500–850bp原料/产品结构退化压力，已将Research note 02基于15–60bp渠道漏损建立的安全垫击穿10倍以上，原'做多干净成本传导'防御逻辑被否证。建议将海天(…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "该哑铃确比受损白马更防御，但其回撤边界以利率环境为条件——两腿共享久期，在利率波动＋再通胀尾部会同跌，缺乏显性久期对冲时韧性不足。", "summary": "本卡对 Research note 06 的“政策性口粮（金龙鱼300999、北大荒600598、中粮系）+非餐饮型红利（长江电力600900、华能水电600025、中国移动600941、大众药店）”约400bp哑铃做组合级压力测试。结论：哑铃确比受损的海天/青啤更防御，但其防御性以利率环境为条件…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源板块不存在监管资本约束，但在设备更新刚性开支下存在现金流与杠杆约束，长江电力与油气巨头分红耐久性极强，而资本开支刚性上升的中国神华在重度压力下或被迫下修分红至65%承诺底线，弱成…", "summary": "截至2026-06-16，本报告通过压力测试验证能源板块分红的耐久性。测试表明，能源板块虽无硬性“资本约束”，但在大规模设备更新刚性资本开支下，面临“自由现金流与杠杆率约束”。长江电力凭借低资本开支强度分红安全性高；中国石油资本开支平稳；中国神华受1335.98亿元重组影响资本开支增至559.87…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "公用事业分红脆弱性不是板块性的，而是价格传导机制、回款质量和资本开支回收速度决定的机制性风险。", "summary": "截至2026-06-16，本卡部分否定“下游公用事业分红风险最高”的笼统判断。电力资产因工商业市场化交易、电价拆分、容量电费和辅助服务具备较强传导能力；城燃、水务和环保虽有定价框架，但地方审批、居民端滞后、财政回款和补贴审计使风险更集中。结论：保留水电、核电和优质大型火电，降低弱联动城燃及应收账款…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 1, "risks": 6, "neutral": 1, "value": 53.4, "events": [ { "title": "核电邻近溢价真实但主要是2030+长久期期权；2026-2030年AI电力的边际供给和最干净的传统能源表达在于天然气基础设施与铀燃料循环上游（转化+HALEU），而非高倍数的核电邻近…", "summary": "截至2026-06-16，对第04卡‘核电邻近性=公用事业估值核心’的结论进行压力测试。能源视角认为：(1) 铀矿上游是下一个物理瓶颈，哈原工产量下调+Cameco满产+西方下一个一级项目NexGen不早于2028投产，长期合约价需突破USD 90/lb才能激励边际矿山；转化与HALEU比开采更紧…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力交付物理瓶颈尚未在A股TMT板块充分定价，高估值硬件标的面迎15%-45%估值去化风险，应当战略重构组合并转向物理电网设备、上游矿山及防御性清洁能源公用事业。", "summary": "本报告基于电网与BTM直供电物理硬约束验证的12-24个月算力基础设施部署气穴期，指明A股算力硬件与上游材料板块存在估值脱钩和两融踩踏风险。建议在气穴期内进行风格轮动，超配出口导向变压器和上游铜铝，增配防御性清洁公用事业（长江电力、中国核电）。定量测算表明，优化后的战略组合可将承压收益率提升7.8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "非对称旋转逻辑在算力产业链层级成立，承接力依基建>SaaS>硬件分层分布，但承接资金到位滞后于卖压释放，硬件去杠杆引发的指数级回踩仍不可避免。", "summary": "支持 Research note 02 非对称旋转逻辑在算力产业链层级成立。硬件层（AI服务器ODM/光模块/PCB/国产芯片）融资盘集中、贝塔高，是卖压震中；基建层（大水电、核电、特高压、IDC/算力REITs）凭借物理资源稀缺与现金流稳定构成最佳承接位；SaaS层（金山办公、用友、泛微）受益A…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2026-06-16的核心判断：AI 电力需求会造成短期化石能源回潮，但低碳逻辑不是被瓦解，而是从年度绿电匹配升级为逐小时可交付的清洁稳定容量。", "summary": "截至2026-06-16，本卡压力测试认为：AI 基荷增长会在2026-2030年区域性推高天然气和部分煤电出力，但不会瓦解低碳转型逻辑。真正瓶颈不是绿电项目总量，而是并网、储能、输电和逐小时可交付能力。核电是最优清洁基荷对冲，但2030年前主要靠延寿、重启和增容，新建核电规模化偏2030年代。投…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 10, "neutral": 0, "value": 19.8, "events": [ { "title": "AI电力稀缺是SPCX拥挤风险的有效对冲，但美国AI电力赢家本身已部分拥挤，组合应超配稀缺因子、低配拥挤表达，并用水电红利与成本归因规则做底仓。", "summary": "2026-06-17：本卡综合Research note 01-04，结论是AI电力稀缺资产仍可作为SPCX拥挤风险的组合对冲，但必须拆成结构性对冲与拥挤化对冲两类。保留有合约支撑的核电IPP、600900式水电红利和具备大负荷成本归因的公用事业；但CEG、VST、TLN等美国AI电力赢家已带有成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在美股AI电力赢家与科技成长股存在高Beta重叠且容易被SPCX同步去杠杆的环境下，大类资产配置应向低贝塔的A股水电红利（长江电力）倾斜，并配置核电（中国核电）对冲干旱风险，构建收益…", "summary": "基于2026-06-17，本报告压力测试了SPCX挤压环境下，AI电力稀缺因子与水电红利的配置效率。美股核电IPPs（CEG/VST）对NDX的Beta已飙升至1.08-1.12，存在严重的Beta和动量重叠风险，在去杠杆中易遭受-25.0%的同步回撤。A股水电（长江电力）对NDX的Beta为-0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "验证电力红利资产在极端市场波动下的防御价值", "summary": "压力测试长江电力（600900.SH）的防御价值，结论为「有条件支持」Card-01。2025年归母净利341.67亿、+5.14%，70%派息承诺延至2026-2030。136号文针对新能源、不及水电，反使水电相对更具防御性。自测算显示：即便严重多年期干旱（发电量−12%），股息率约2.81%仍…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "油价战争溢价消除与美联储鹰派紧缩背离下的红利资产结构性重平衡", "summary": "油价单日重挫超6%至约76美元，是美伊缓和抹去战争溢价、叠加IEA下调需求所致；美联储6月维持3.50–3.75%但点阵中值上调至3.8%，通胀降温与高实际利率（10Y国债约4.42%）并存。这并非红利逻辑解体，而是红利分叉：上游资源通胀红利（油气煤炭）因盈利基数收缩+折现率高企而估值锚松动；物理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 2, "neutral": 0, "value": 69.2, "events": [ { "title": "上游能源红利逻辑正被强制切换为'现金流稳定'主导，但估值中枢不能维持当前水平——煤炭可作过渡仓但需-10%中枢下移，油气从核心降级为卫星仓，真正受益者是公用事业型物理资产。", "summary": "油价回落至70-80美元区间剥离了战争溢价，IEA下调2026年全球需求增速至+0.7 mb/d。上游红利逻辑被强制切换为'现金流稳定'主导，但估值中枢不能简单平移：A股三桶油/煤炭龙头股息率虽仍6-8%，盈利-分红覆盖倍数从2022年~2.0×压缩至~1.3×。压力测试显示韧性排序为煤炭>油气>…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在长端利率温和上行（路径A）时，T合约对冲优于TL，而在单边大额上行（路径B）时TL更佳；相比线性国债期货，零成本看跌价差领口期权提供正Gamma与多因子风险防范，是对冲红利大水电久…", "summary": "本卡定量评估了长江电力权益久期对冲工具。在10Y CGB升至2.00%路径下，因曲线熊平，T合约相比TL更为有效；在升至2.20%路径下，TL久期匹配度更高，需51手合约。国债期货套保面临严重的因子基差风险、交割券切换和移仓成本（TL年化达20-30bp）。相比之下，6个月期看跌价差领口期权（K1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 1, "neutral": 2, "value": 82.6, "events": [ { "title": "30%防御现金流端必须从泛红利收紧为A+B+C+E核心(公用事业/运营商/大行保险/煤炭油气)占80%+的本土现金流篮子，剔除与电网出海链同源的‘伪红利’出海标的，并按四情景对释放的…", "summary": "截至2026-06-20，本卡综合卡片02-08，把30%防御现金流端从泛红利收紧为外需/汇率/地缘三低敏感的本土现金流篮子：水电核电燃气8%+电信运营商5%+大行与大型保险7%+煤炭油气5%+高速港口铁路3%+必需消费2%，明确剔除出海家电、逆变器、出海化工等‘伪红利’标的。给出基础/中度/重度…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "若WTI油价跌破75美元（布伦特78美元），沙特将触发暂停每月18.8万桶/日增产的防御机制，OPEC+配额纪律短期依然成立，真正的产量价格战门槛在70美元下方。", "summary": "阿联酋5月退盟后，OPEC+防守价格的重担落在核心7国。若油价（WTI）跌破75美元，由于触及沙特财政平衡红线（布伦特78美元），沙特将启动“暂停增产”机制防守油价。因此WTI在70-75美元区间减产协议不会瓦解，真正的价格战门槛是低于70美元并伴随公开作弊。组合建议向长江电力（600900.SH…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "对立路径是伪命题——只有'存量水核长协+CCGT桥梁+风光短储+少量LDES'的混合栈能在2026-2030窗口与新能源+短储PPA形成成本-可靠性双向占优，且仅限水核枢纽地理。", "summary": "截至2026-06-19，水电、核电、燃气+CCUS、长时储能任何单一栈都无法同时在成本与24/7 CFE匹配率上跑赢风光+4h锂电PPA：水电资源已开发60%+且枯水风险高，核电2030年前增量仅30-40GW不足以匹配AI 60-100GW新增负荷，燃气+CCUS全成本0.75-0.95元/k…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 1, "risks": 0, "neutral": 0, "value": 92, "events": [ { "title": "“Time to Power”交易下的能源供给侧产能与增量需求分析", "summary": "本卡从能源供给侧支持 Research note 04-07 的 time-to-power 论点。截至 2026-06-21，结论是：缺口不在电量（TWh），而在匹配 7×24 AI 负荷的“确定可调度容量”。扩产最快的是光伏+储能（2026 年 86 GW 新增中占 79%），但形态最不适配；…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-21", "supports": 2, "risks": 16, "neutral": 0, "value": 12, "events": [ { "title": "在能源价格常态化预期下，红利配置应向清洁电力转移，但须通过煤电组合对冲水电枯水期现货买电风险，并警惕规制电网在AI负荷爆发期的信用评级恶化与表后流失风险。", "summary": "截至2026-06-21，本研究支持将组合红利仓位从周期性煤油向公用事业转移的逻辑。水电（长江电力）在严重枯水（-15%）情景下面临现货回购惩罚，净利润下降33.5%且面临29.7%派息削减风险；核电（中国核电）受益于煤电联动，电价上浮5%可提升利润15.88%，并捕获5.8-8.4分/kWh的碳…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在美联储沟通革命引发全球多资产波动率骤升与成长股多重估值压缩背景下，A股应坚定执行超配高电力上网电价防御韧性（水电/核电）与低配高两融风险成长板块的‘防御端杠铃策略’。", "summary": "美联储取消前瞻性指引与通胀预期上修打破了传统的股债负相关性，迫使风险平价模型被动去杠杆和抛售资产。压力测试显示风险平价在当前宏观环境下显著跑输，而战术性杠铃策略表现出强劲的抗波动韧性。全球收益率上行传导至A股，导致科创成长股面临多重估值压缩及两融强平风险，资金加速向水电、核电等红利重估板块切换；同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DR007上行会压缩公用事业股息利差并提高再融资成本，但分红削减的核心触发器是资金利率、煤价、电价和资本开支同时恶化，而非DR007单项上行。", "summary": "截至2026-06-21，本卡对电力等“物理红利”资产进行压力测试：DR007从6月1日1.3644%升至6月18日1.4606%，会抬高贴现率和再融资成本，但尚不足以单独击穿公用事业分红。真正风险是DR007接近1.70%、煤价维持861元/吨以上、市场化电价走弱与高资本开支叠加。水电和低燃料b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源红利支持35%红利股防御逻辑，但最佳表达是12-15个百分点的政策现金流与供给风险对冲，而不是全仓油煤周期押注。", "summary": "截至2026-06-21，本卡支持在35%红利股袖套中配置12-15个百分点能源相关红利资产。依据是10Y国债收益率1.73%、中证能源股息率3.86%、中证红利股息率4.49%，以及央企市值管理和分红政策对大型能源央企的约束。中国神华2025年分红率79.1%，中国海油2025年派息608.38…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 2, "risks": 2, "neutral": 0, "value": 90.8, "events": [ { "title": "在公用事业超级资本开支面临股权稀释及大选年监管时滞的双重制约下，跨资产配置应调降拥挤的美股公用事业权益，通过公用事业IG高级债捕获拓宽的信用利差，同时布局低相关的A股水电红利（长江电…", "summary": "基于2026-06-22，本报告综合研判了受监管公用事业费率基数超级周期下的跨资产布局。公用事业1.3万亿美元资本开支带来严重的股权二次发行稀释，且大选年36州州长竞选引发监管机构费率抵制，惠誉调降行业展望至“恶化”。PJM 2027/28容量价暴涨至333.44美元/MW-日，零售电费同比上涨1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力价值链中的\"成期权\"：核电、水电及SMR对公用事业估值重塑的影响", "summary": "本卡支持卡08的\"类债→成长\"切换，但收紧为一次三层分部拆解：每家大型公用事业应拆成（i）稳态年金（类债DDM）、（ii）已签20年期算力PPA（长久期年金）、（iii）稀缺电力看涨期权（实物期权溢价）。证据：PJM容量电价从约29升至329.17美元/MW·日FERC上限[S5]；中国核电在运2…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI需求能扭转公用事业的投资疲软，但路径是把板块增长重新导向发电与电网（及选择性燃气/水电稳功率），而非普遍抬升水务/环保——是结构性扩张，印证Research note 09的选择…", "summary": "截至2026-06-23，本卡（AI基础设施分析师，10/10收官卡）支持整条研究线，并补上需求侧催化剂。Research note 09判定市政管网仅政策订单扩张、未全板块扩张（1—5月电热气水固投-0.1%、水利环境-3.9%），那是旧的地产/市政权重图景；AI带来的是新的发电与电网权重图景。…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "在海风开发商回款账期延长与物理确认时滞下，A股对东方电缆（高负债杠杆压制）与亚星锚链（交付时滞真空）进行了估值收缩定价，传统的成长重仓策略失效，向具备即期现金流变现能力的水电与海运红…", "summary": "本报告对开发商账期拉长下东方电缆与亚星锚链的流动性与物理交付折价进行了A股定价逻辑压力测试。分析表明，东方电缆重资产、高资金占用模式受制于垫资压力，在重度承压下经营现金流转负至-31.73亿元，导致PE估值中枢被压缩至14.5倍；亚星锚链虽现金流安全，但142天物理交付验收时滞使其财务表现呈业绩真…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 7, "neutral": 0, "value": 46.2, "events": [ { "title": "在成本压力积聚与零售利润敏感性增加的宏观环境下，红利组合必须对运营租赁类REITs实施动态配置上限管理，向原材料独立的公用事业高股息股与国债倾斜，以保障组合总回报。", "summary": "针对成本冲击对折扣零售REITs分红稳定性的威胁，本报告提出红利防御组合优化框架。将REITs划分为高贝塔商业零售型（A类）与稳健特许经营型（B类）。在成本压力下，将A类REITs配置上限由15%下调至8%（中度）和3%（重度），增配水电红利股（长江电力 600900.SH）与10年期国债。模拟表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在重型燃机与电网基建多年配给约束导致算力交付延迟2年的情景下，电费成本暴增将云巨头EBIT利润率蚕食150-300BP，迫使科技板块ERP走宽70BP并引发理论PE中枢下调16.77…", "summary": "基于2026-06-23，本报告评估了由燃机档期、精铸产能、已认证合金、变压器和气源许可共同形成的多年配给市场下科技股的估值重定价。燃机及电网配给将算力激活推迟1.5至2.5年，迫使2026-28年AI资本开支增长下修10%至15%。表后燃气+BESS电价升至$0.220/kWh，单卡GPU电力O…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "煤价已反弹而非下行，但火电龙头高股息凭借容量电价（114号文）、长协煤与派息纪律仍可守住约5-6.5%水平——红利支撑已从周期性的煤价切换为结构性的容量电价，应为票息持有火电、为确定…", "summary": "截至2026-06-23，本卡支持Research note 03的火电超配，但重置其论证基础。2025年驱动华能利润+42%的煤价下行红利已消退：2026年5-6月环渤海Q5500现货回升至约835-861元/吨、同比+15%，同时上网电价同比-3.48%，成本边际双向挤压。但红利仍可持续，因支…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "容量电价与稳定性溢价落地下的电力公用事业与电网设备板块重估空间与风格轮动路径", "summary": "本卡支持并综合前四卡：算力＋电气化＋容量电价硬化现金流，使公用事业从“类债券”向“成长公用”重估。结论：重估真实但不对称。硬基荷（水电/核电）的倍数重估最大且仍在前方——以川投PE约15.3、隐含(r−g)≈2.9%为锚，(r−g)压缩40–80bp对应约15%–35%合理估值上行[自测算]；电网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 5, "neutral": 0, "value": 59.6, "events": [ { "title": "在重型燃机与电网多年配给重塑科技股估值的情景下，超配长江电力等在岸清洁基荷资产的风险平价组合可锁定确定性溢价，且通过MXN/JPY外汇数字期权可有效锁定日元平仓海啸的尾部下行风险。", "summary": "本报告针对电网并网延迟与重型燃机5-7年超长交期约束，科技成长板块估值重塑，以及美日利差收窄导致的日元套利平仓风险，构建了长江电力、中国核电、大行/电信红利及短端国债组成的清洁能源杠铃组合。通过波动率风险平价（ERC）和高股息倾斜自定义风险平价进行了权重优化。为对冲日元套利平仓引发的流动性冲击，测…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "重型燃机与电网多年配给格局下跨资产清洁能源杠铃组合的风险归因、权重重构与全天候对冲成本测算", "summary": "重型燃机与电网多年配给格局下跨资产清洁能源杠铃组合的风险归因、权重重构与全天候对冲成本测算 在重型燃机与电网多年配给导致科技股估值重塑且在岸高股息资产重要性上升的情景下，如何利用全天候或风险平价（Risk Parity）模型重新为长江电力、中国核电、大行/电信红利以及短端国债分配风险预算？在计入了…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "液冷驱动的超高功率密度对电网配电侧建设与能源供应响应力的挑战", "summary": "经压力测试：在受监管公共电网层面答案是“不能”，园区层面仅对提前占位者有条件“能”。密度对电网不是关键变量——100MW园区无论20kW还是140kW机柜都取电约100MW[自测算]，密度只是把约束下移到馈线/变电站/变压器并加剧负荷波动（已触发NERC突发失负荷预警）。时间错配是结构性的约5.6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "防御杠铃组合中大行、水电、现金与债券的再平衡比例", "summary": "本卡综合 Research note 3–5，给出防御红利仓位（占组合40%）的五段式再平衡向量：大行由25%降至10%（转出15个百分点），转入收费公路+7、短久期债券+5、现金+3，水电维持15%但内部≥80%倾向长江电力一类低市场化敞口。测算显示：基准股息率由4.00%降至3.51%（牺牲约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 1, "neutral": 0, "value": 85, "events": [ { "title": "BTM是受监管公用事业AI资本开支的真实尾部风险，但监管工具正在把风险前置给大型负荷客户，基准情形应是选择性估值下修而非XLU普遍降级。", "summary": "截至2026-06-25，本卡支持BTM自建能源会放大AI电网投资搁浅风险，但不支持XLU普遍降级作为基准情形。核心风险在于公用事业公司先为数据中心负荷举债建设T&D，而Hyperscaler随后因并网延迟、成本或可靠性转向BTM核电、燃气轮机或微电网。监管机构正在通过大型负荷电价、最低需量、抵押…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 6, "neutral": 1, "value": 59.4, "events": [ { "title": "电力电网及公用事业红利资产的配置逻辑验证", "summary": "支持第01张卡片的防守再平衡判断，并验证其机制：本轮去杠杆是两融融资拥挤的电子板块冲击（A股两融2026-06-23破3万亿元），而受监管公用事业由保险/养老等长钱持有、融资拥挤度低，处于压力源的对立端，是真正避风港而非平行泡沫。需求端1-5月用电+3.4%、互联网数据服务用电+26%给出AI电气…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "136号文市场化电价下绿电、核电、水电和储能运营商现金流敏感性", "summary": "对运营商做三降压力测试：绿证价格、利用小时、市场交易电价同时下行。2025年业绩已是实盘验证，横截面决定性。保留满仓——长江电力(600900)水电、中国核电(601985)核电：电价下行年净利仍增(+6.17%/+6.00%)，水电近零边际成本+65.79%毛利、来水定产不受限电；核电>7500…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "验证水电板块在电价刚性及调节性需求下的盈利稳定性与防御价值", "summary": "支持前序判断：水电红利股是去杠杆杠铃合格的物理防御腿。电价刚性真实且被2026容量电价新规（发改价格〔2026〕114号，2026-01-30）强化，为调节性电源增加容量收入地板。长江电力境内水电上网电价约0.28–0.29元/kWh，长期稳定；与Card-02引用的0.4432元/kWh为出厂上…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通胀粘性阻碍美联储降息的背景下，纸面原油抛售与实物供给紧张的冲突使得油价下行有底而上行具备高凸性特征，使其在资产相关性转正的环境下成为有效的非对称风险对冲工具。", "summary": "截至2026-06-27，本卡支持能源作为组合凸性对冲工具的逻辑。美伊备忘录虽引发纸面油价大跌（WTI至70.24美元/桶），但扫雷瓶颈（60日延迟）与伊朗原油出口缓慢恢复使物理供给仍然偏紧，且美原油去库1553万桶、SPR降至历史新低3.492亿桶。高油价向核心PCE的二阶传导存在6-12个月时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 0, "value": 75.8, "events": [ { "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源与电力板块（长江电力/中国海油）的防御性估值与政策风险压力测试", "summary": "对卡片09把半导体腾出资金转入长江电力(600900)与中国海油(0883)防御的安排做压力测试，得出分裂裁决。电力/长江电力基本面不过热（零燃料成本、70%分红锁定至2030），但作为工具已成拥挤的26.8年久期债券替代品：股息率压至3.5%、对国债利差约2pp处历史高位、内在价值≈市值、贴现率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "1.3%低利率下，红利全天候组合应从追逐水电股息利差转向水电质量、规制煤电容量收入、国债久期与黄金/煤炭事件对冲的动态杠铃。", "summary": "2026-06-27：本卡综合前序链条，认为1.3%国债情景不是加仓红利资产的单向信号，而是要求把红利全天候组合从“长久期股息代理”改成“规制现金流杠铃”。建议正常环境水电22%-26%、煤电14%-18%、国债30%-34%、黄金/煤炭12%-16%、现金6%-10%；若1.3%国债叠加枯水/现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 4, "neutral": 0, "value": 71.2, "events": [ { "title": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "summary": "截至2026-06-28，本卡对“AI负荷增长能否支撑公用事业高分红”作压力测试：结论是精选组合层面可以，板块整体不能机械外推。AI数据中心用电增长真实，但也同步拉动电网、发电和储能资本开支。水电、核电、容量电价支持的火电及低杠杆综合电力运营商仍可维持分红；风光开发商受消纳、电价和项目资本开支制约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理基础设施约束下多资产投资组合的重平衡与防守哑铃配置策略", "summary": "本卡支持卡片05–07链条，把物理基础设施约束落地为多资产重平衡：构建防守哑铃。削减超估值且拥挤的AI硬件成长腿（−12pp至20%并加期权领口），把仓位转入物理基础设施租金腿（公用事业含水电核电、电网重电与变压器、水权、液冷热管理，+14pp至22%）与高股息压舱（中证红利约4.68–4.9%股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费成本转嫁挤压。", "summary": "基于国产芯片2.94倍能效赤字拉长通电时间至18个月的物理瓶颈，本报告系统性量化了A股电力设备板块（特高压、大变压器）的估值扩张机制与低利润 traditional TMT 的电价挤压。电网容量受限与十五五规划共振，支持了特高压订单可见度与大型变压器出海高溢价，特别是在国内外硅钢5.1倍巨大价差的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "局部中小银行信用缺口下的A股银行红利耐久性与高股息内部轮动", "summary": "卡08分离出的“额度沉睡”机制（CET1上限+NPL触发+低于1.40%净息差）不仅是中小银行信用风险，更是撕裂A股红利复合体的力量。重排结论：①大行红利居顶——分红不依赖沉睡额度，约4.35%股息率、对10年国债（约1.72%）257bp利差、0.5–0.7倍PB已计入坏消息，持有/超配；②公用…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 6, "neutral": 0, "value": 60.6, "events": [ { "title": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红…", "summary": "本报告针对2026年三季报TMT硬件估值折价预期，校准了全球全天候和风险平价模型。战术防御性风险平价（TDRP）将AI成长股风险预算由25.0%下调至8.0%，防守红利上调至40.0%，以对冲55.0%的TMT增长溢价折价。模型显示，电力配给约束下算力效率的TCO优势在24/7清洁能源下扩大至9.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "summary": "截至2026-06-29，本卡综合前9卡：AI电力约束是选择性时序上限而非订单毁灭。建议跨资产账本配置为40%受资助AI基础设施龙头、45%主权背书红利公用事业、3%年度申万电子Put权利金预算、12%流动性储备。进攻端以中际旭创18%、工业富联14%、英维克8%为核心；防御端以水电18%、核电1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电…", "summary": "本报告对AI算力激增带来的电力电网物理瓶颈（变压器交期120-210周[S2]、并网排队4-7年[S1]）对A股转型龙头的利润率传导进行压力测试。结果显示，电网扩容滞后充当了新能源装机的“限速阀”，压制了光伏/电池销量增速；同时，容量电价[S8]和碳成本上升[S19]增加了高能耗制造环节电费开支，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-5fd96966de/", "networkHref": "/investor/research-network/#thesis-5fd96966de" }, { "id": "thesis-71476ded81", "source": "auto_topic", "title": "", "titleEn": "Practical research topic:", "lane": "消费健康防御", "laneEn": "消费健康防御", "laneHref": "/investor/lanes/lane-d104d465/", "investableFocus": "", "stockUniverse": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "NEE", "name": "NextEra Energy", "market": "US", "role": "可再生电力与公用事业", "roleEn": "可再生电力与公用事业", "bucket": "core", "href": "/investor/stocks/NEE/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "IRM", "name": "Iron Mountain", "market": "US", "role": "数据中心与存储 REIT", "roleEn": "数据中心与存储 REIT", "bucket": "core", "href": "/investor/stocks/IRM/" } ], "stockTickers": [ "600519.SH", "300760.SZ", "600276.SH", "3690.HK", "001215.SZ", "1179.HK", "9961.HK", "MU", "META", "NEE", "TLT", "IEF", "MAR", "IRM" ], "coreView": "该主线来自 6 条研究记录的标题、问题和摘要聚类，时间跨度 2026-06-13 至 2026-06-28；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-05", "lastSeen": "2026-06-29", "supports": 2, "risks": 123, "neutral": 35, "analystCount": 13, "sourceTypeCounts": { "研究笔记": 25, "深度研究": 6, "分析协作": 19, "研究线程": 110 }, "topTerms": [ "specific", "insid" ], "whatChanged": [ "2026-06-29: [auto-handoff] 日债 2.8% 平仓红线对大类资产Barbell左端美债/红利资产的重塑", "2026-06-29: [auto-handoff] 日元套利交易平仓对离岸CNH流动性及结汇溢价影响跟踪", "2026-06-29: [auto-handoff] 离岸CNH结汇边界与OMO降息窗口追踪", "2026-06-29: [auto-handoff] 韩国半导体巨额投资催化及原油下跌下A股科技与高股息板块轮动研判", "2026-06-29: 产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）" ], "practical": { "sectorLane": "消费健康防御", "sectorLaneEn": "消费健康防御", "exposureBucket": "消费健康防御", "tone": "green", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "NEE", "name": "NextEra Energy", "market": "US", "role": "可再生电力与公用事业", "roleEn": "可再生电力与公用事业", "bucket": "core", "href": "/investor/stocks/NEE/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "IRM", "name": "Iron Mountain", "market": "US", "role": "数据中心与存储 REIT", "roleEn": "数据中心与存储 REIT", "bucket": "core", "href": "/investor/stocks/IRM/" } ], "stockTickers": [ "600519.SH", "300760.SZ", "600276.SH", "3690.HK", "001215.SZ", "1179.HK", "9961.HK", "MU", "META", "NEE", "TLT", "IEF", "MAR", "IRM" ], "primaryStocks": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" } ], "instrumentTags": [ "600519.SH", "贵州茅台", "300760.SZ", "迈瑞医疗", "600276.SH", "恒瑞医药", "3690.HK", "美团", "001215.SZ", "千味央厨", "1179.HK", "华住集团-S" ], "reviewCadence": "每日复核", "confirmationSignals": [ "价格弹性是否转化为利润率修复", "支付端和监管是否支持长期现金流", "防御属性是否被估值提前透支" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "估值拥挤", "支付监管", "渠道库存", "CPI 食品项" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 126 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 13, "recentEvents": 126, "supportRatio": 0.01, "riskRatio": 0.77, "conflictScore": 88.5, "evidenceDepth": 100 }, "score": 68.8 }, "events": [ { "date": "2026-06-29", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 韩国半导体巨额投资催化及原油下跌下A股科技与高股息板块轮动研判", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 79c6bdcc-2296-4fc8-8d4a-c239f8bf4e3d 结果摘要: I will list the contents of the default scratch directory to check the environment. I will …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-28", "title": "I have completed the research for Research note 01 of this research thread session. Summary of …", "summary": "I have completed the research for Research note 01 of this research thread session. Summary of Work 1. Analyst Selection : Appointed the Global Macro Ana...", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7.25, "matchedTerms": [ "specific", "insid" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "specific" ], "href": "" } ], "timeline": [ { "date": "2026-05-05", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "AI monetizer相对AI-cost absorber交易仍有alpha，但65%-75%是共同因子驱动，必须按quality spread而非无约束AI主题篮子来构建。", "summary": "本卡综合前序结论：long AI monetizer / short AI-cost absorber并非纯stock-specific alpha。按美股/港股均衡组合估计，约65%-75%的收益与回撤风险来自quality、growth、momentum、duration、China beta…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "中国电力设备出海：AI 电网需求、贸易壁垒与利润率分化", "summary": "全球 AI 电网投资为中国电力设备提供需求窗口，但关税、认证、交付能力和原材料成本会决定哪些企业真正把订单转化为利润。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/china-power-equipment-export-policy-8678db7f/" }, { "title": "AI 算力资本开支、电力瓶颈与通胀再定价", "summary": "AI 基础设施需求正在先通过电力、电网、变压器与材料供给形成成本脉冲，再通过生产率与架构效率形成中期缓释；投资结论取决于这两个方向谁先兑现。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-power-inflation-transmission-0c9ba15c/" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 0, "neutral": 0, "value": 52.8, "events": [ { "title": "消费降级、银发经济与大健康结构性机会", "summary": "消费链条正在从总量弹性转向结构分化：银发消费、医疗健康、糖替代和 B 端餐饮渗透可能形成防御性需求，但需要验证支付能力和渠道真实动销。", "sourceLabel": "深度研究", "supportLabel": "支持", "supportType": "supports", "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 0, "neutral": 1, "value": 49.2, "events": [ { "title": "All upstream workspace files for 研究线程标识：已归档(specifically session brief.md and prior researc…", "summary": "All upstream workspace files for 研究线程标识：已归档(specifically session brief.md and prior research notes under Research note 01 , Research note 02 , and Re…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "I have synthesized the research from previous cards into a formal asset allocation and hedg…", "summary": "I have synthesized the research from previous cards into a formal asset allocation and hedging rulebook for the AI hardware sector, anchored to the M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 0, "neutral": 3, "value": 58, "events": [ { "title": "During our analysis of Research note 8/8 in the research thread session, we found that seve…", "summary": "During our analysis of Research note 8/8 in the research thread session, we found that several referenced files were missing from the local workspace…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 日元升值与美欧关税挤压下中日工业出海市场份额与地缘博弈测算", "summary": "致 主题研究员： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 5e4ff3a8-78a5-4219-8a2e-0b38c70ef223 结果摘要: I will run the shell command to anchor today's date and get t…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have completed the liquidity stress test for the A share semiconductor sector as of May 2…", "summary": "I have completed the liquidity stress test for the A share semiconductor sector as of May 22, 2026. The research confirms that the sector is at a his…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 38.4, "events": [ { "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" }, { "title": "As the Agriculture Analyst (agri analyst) , I have conducted a rigorous stress test on the …", "summary": "As the Agriculture Analyst (agri analyst) , I have conducted a rigorous stress test on the ongoing research regarding “China macroeconomic analysis: …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 1, "value": 50.2, "events": [ { "title": "As the Chief Economist , I have synthesized the quantitative findings from the Financials A…", "summary": "As the Chief Economist , I have synthesized the quantitative findings from the Financials Analyst (Research note 05) and analyzed the macroeconomic c…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Section 232 关税压力测试", "summary": "致 医药行业分析师： 协同任务已完成。执行方为 首席风控官。 模式: adhoc 会话: 51029e47-5f47-4436-bf94-82515d4c8598 结果摘要: I will first run the shell command to anchor today's date and…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 0, "neutral": 1, "value": 57.6, "events": [ { "title": "Overview of Policy and Geopolitical Analysis As the Policy Analyst for Research note 08/08,…", "summary": "Overview of Policy and Geopolitical Analysis As the Policy Analyst for Research note 08/08, I have completed the final, synthesis led analysis of thi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "I have completed the geopolitical stress test and risk analysis for Research note 5 of the …", "summary": "I have completed the geopolitical stress test and risk analysis for Research note 5 of the research thread portal session as a Policy Analyst. Based …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the asset allocation and drawdown control analysis for the JD/Meituan pair…", "summary": "I have completed the asset allocation and drawdown control analysis for the JD/Meituan pair trade. To provide robust quant grounding for this card, I…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "政策约束已经把A股半导体的可投资宽度实质性收窄到先进封装/HBM/SiC/稀土永磁四个切片，而ETF资金已经为全板块定价了80分位的政策溢价；下半年应'减总量、加精度'，并把2026…", "summary": "本卡片锚定2026-06-03。综合Card-01至Card-05的销量质量→政策销量→利润集中→行政性稀缺租金的链条，把先进制程白名单+成熟制程2027-06-23关税突跳+中端配套2026-11-10双向断点三重政策约束直接翻译为A股板块宽度、风格暴露、ETF资金流与拥挤度结论。政策筛后宽度约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 0, "neutral": 2, "value": 64.8, "events": [ { "title": "I will list the current permissions to see what directories and files I have access to. I w…", "summary": "I will list the current permissions to see what directories and files I have access to. I will check the files inside the scratch folder and confirm …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I will start by checking the current directory and verifying the files in our shared worksp…", "summary": "I will start by checking the current directory and verifying the files in our shared workspace. I'll execute a command to inspect the directory struc…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 0, "neutral": 1, "value": 63.2, "events": [ { "title": "[auto-handoff] 离岸人民币流动性挤压对境内城投债及地方专项债信用利差的二阶传染评估", "summary": "致 主题研究员： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 990cc221-8af1-4fac-9830-c8912e2fa2e5 结果摘要: I will list the current permissions to see what actions and p…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 1, "value": 58.6, "events": [ { "title": "[auto-handoff] 非化石能源电力消费指南及输电权改革对电解铝成本影响", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 2fb63bb9-fc56-46db-a3a7-840fd4f7fa5f 结果摘要: I will run the command to check today's date and find the c…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I will begin by inspecting the workspace files, specifically the session brief, the previou…", "summary": "I will begin by inspecting the workspace files, specifically the session brief, the previous card reports (Research note 05, Research note 02), and t…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 6, "neutral": 1, "value": 30, "events": [ { "title": "I will start by checking the files in the workspace, specifically those related to Research…", "summary": "I will start by checking the files in the workspace, specifically those related to Research note 05 and Research note 06, to build on the existing re…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 3.3%, 3/91）", "summary": "产业链调研：新一代信息技术（30 日激活率 3.3%, 3/91） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.1%, 14/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.1%, 14/198） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 5, "neutral": 0, "value": 34.4, "events": [ { "title": "产业链调研：海工装备产业链（30 日激活率 5.4%, 6/112）", "summary": "产业链调研：海工装备产业链（30 日激活率 5.4%, 6/112） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 10.3%, 10/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 10.3%, 10/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 4.7%, 8/170）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 4.7%, 8/170） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275）", "summary": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 0, "value": 41.8, "events": [ { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 6.5%, 13/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 6.5%, 13/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 5, "neutral": 1, "value": 40.2, "events": [ { "title": "I have initialized the research thread research thread by selecting the ai infrastructure a…", "summary": "I have initialized the research thread research thread by selecting the ai infrastructure analyst (AI Infrastructure Analyst) role and completi...", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 7, "neutral": 5, "value": 41.6, "events": [ { "title": "We have kicked off the research thread session by choosing the Semiconductors Analyst ( sem…", "summary": "We have kicked off the research thread session by choosing the Semiconductors Analyst ( semiconductors analyst ) role and investigating a high v...", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Wrote research thread/68484002 c477 468e 99be 56a88283fbcc/Research note 01/report.en.md (4…", "summary": "Wrote research thread/68484002 c477 468e 99be 56a88283fbcc/Research note 01/report.en.md (4728 bytes); wrote research thread/68484002 c477 468e 99be …", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I will list the files in the workspace directory to verify existing files and check the pri…", "summary": "I will list the files in the workspace directory to verify existing files and check the prior cards. I will list the files in the specific research t…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I will list the workspace directory to locate the session brief and check if there are any …", "summary": "I will list the workspace directory to locate the session brief and check if there are any existing files or structures we need to build upon. I will…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 8, "neutral": 0, "value": 22, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 5, "neutral": 2, "value": 47.4, "events": [ { "title": "I have completed the synthesis for Research note 08 of the research thread session as the C…", "summary": "I have completed the synthesis for Research note 08 of the research thread session as the Chief Strategist. By integrating the macroeconomic, sector …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 10, "neutral": 0, "value": 12.8, "events": [ { "title": "As the Chief Strategist (首席策略师) , I have synthesized the physical, operational, and financi…", "summary": "As the Chief Strategist (首席策略师) , I have synthesized the physical, operational, and financial findings from Cards 01 to 06 regarding the \"AI Power Bo…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 1, "risks": 9, "neutral": 1, "value": 31.2, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 5, "neutral": 1, "value": 48.6, "events": [ { "title": "产业链调研：建筑材料产业链（30 日激活率 8.3%, 13/157）", "summary": "产业链调研：建筑材料产业链（30 日激活率 8.3%, 13/157） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 1, "neutral": 1, "value": 74, "events": [ { "title": "To launch the kickoff card of the new research thread session ( c10afdc8 f845 47c5 87e6 d35…", "summary": "To launch the kickoff card of the new research thread session ( c10afdc8 f845 47c5 87e6 d35e754123f2 ), I have chosen the AI Infrastructure Anal...", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 9, "neutral": 2, "value": 30.4, "events": [ { "title": "I have initiated a new research thread session for Board 1cb2c1c9 cb98 42c6 aafc 26aff3a867…", "summary": "I have initiated a new research thread session for Board 1cb2c1c9 cb98 42c6 aafc 26aff3a86784 as the Global Macro Analyst ( global macro ). Summ...", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I will start by checking the existing files in the research thread directory to understand …", "summary": "I will start by checking the existing files in the research thread directory to understand the prior analysis, specifically the findings from Researc…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 7, "neutral": 2, "value": 43.8, "events": [ { "title": "I have completed the first research thread research card of the session for the global macr…", "summary": "I have completed the first research thread research card of the session for the global macro (Global Macro Analyst) role. Topic Selection & Rati...", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "All three deliverables are written and validated. Research note 06 — Utilities Analyst — sy…", "summary": "All three deliverables are written and validated. Research note 06 — Utilities Analyst — synthesis complete (work date 2026 06 23). Core judgment: Th…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 4, "neutral": 0, "value": 57.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 12.6%, 26/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 12.6%, 26/206） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 4, "neutral": 2, "value": 64.6, "events": [ { "title": "I have completed the stress test analysis of the overseas CMC technology transfer and MRCT …", "summary": "I have completed the stress test analysis of the overseas CMC technology transfer and MRCT reorganization pathways for innovative Chinese drugs (spec…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 0, "neutral": 2, "value": 90, "events": [ { "title": "I have successfully initiated the research thread session by selecting the global macro dom…", "summary": "I have successfully initiated the research thread session by selecting the global macro domain ( global macro ) and analyzing the June 2026 Fede...", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Impact of Strait of Hormuz reopening and oil price drop on global inflation …", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 6818983a-deba-402e-a115-3ac579a9752a 结果摘要: I will run the command 系统日期锚 to anchor today's date and che…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 6, "neutral": 0, "value": 49.4, "events": [ { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 1, "value": 71.8, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 事实核查员： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 3d205258-4cae-4ebe-963d-d7c4633698de 结果摘要: I will run the command 系统日期锚 to get today's date and check the …", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 5, "neutral": 0, "value": 58.2, "events": [ { "title": "I have completed the research for Research note 01 of this research thread session. Summary…", "summary": "I have completed the research for Research note 01 of this research thread session. Summary of Work 1. Analyst Selection : Appointed the Global Macro…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 7, "neutral": 4, "value": 59.6, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-71476ded81/", "networkHref": "/investor/research-network/#thesis-71476ded81" }, { "id": "thesis-bb8f3d8ee7", "source": "auto_topic", "title": "隔夜美股半导体与小盘逼空是否足以覆盖科创50/CPO的本地两融、解禁和折算率压力，还是应继续把科技回补限制在5%试仓以内？", "titleEn": "Practical research topic: 隔夜美股半导体与小盘逼空是否足以覆盖科创50/CPO的本地两融、解禁和折算率压力，还是应继续把科技回补限制在5%试仓以内？", "lane": "中国权益与流动性", "laneEn": "中国权益与流动性", "laneHref": "/investor/lanes/lane-c06f0832/", "investableFocus": "", "stockUniverse": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "510300.SH", "588000.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK", "9988.HK", "APO", "IEF", "CAT", "300308.SZ", "COP", "MAR" ], "coreView": "该主线来自 4 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-29 至 2026-06-16；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-21", "lastSeen": "2026-06-29", "supports": 8, "risks": 135, "neutral": 17, "analystCount": 13, "sourceTypeCounts": { "研究笔记": 122, "研究主题": 17, "研究线程": 15, "分析协作": 6 }, "topTerms": [ "科创50", "CPO", "半导体", "A股", "两融", "DELL", "FY27", "收入", "服务器收入和", "目标", "去杠杆交易切回订", "创50/CPO的本地两融" ], "whatChanged": [ "2026-06-29: 在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%）；通过配以5.0%权重可转债债底防御和0.50%权重股指虚值Put期权，可将组合极限收益拉升至+104.40%以平准踩踏损…", "2026-06-29: A股红利拥挤度与7-11月年内解禁洪峰限制了风格切换空间，而在人行短端流动性上限失灵与CNH被动升值情景下，高负债、高Beta的券商板块将面临最严重的两融净利差挤压与反转冲击。", "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。", "2026-06-29: 国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多资产组合VaR在1530点处强制破位，急需分层部署的ETF买盘及两融弹性化阀门进行精…", "2026-06-28: A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。" ], "practical": { "sectorLane": "中国权益与流动性", "sectorLaneEn": "中国权益与流动性", "exposureBucket": "中国权益与流动性", "tone": "green", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "510300.SH", "588000.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK", "9988.HK", "APO", "IEF", "CAT", "300308.SZ", "COP", "MAR" ], "primaryStocks": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" } ], "instrumentTags": [ "510300.SH", "沪深300 ETF", "588000.SH", "科创50 ETF", "512100.SH", "中证1000 ETF", "2800.HK", "盈富基金", "MCHI", "iShares MSCI China ETF", "KWEB", "KraneShares CSI China Internet …" ], "reviewCadence": "每日复核", "confirmationSignals": [ "融资余额、ETF 申赎和北向资金是否同向", "被动调仓窗口是否被主动资金提前交易", "订单、盈利和估值是否同步改善" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "指数期权", "红利低波", "汇率", "两融降温" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 123 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 13, "recentEvents": 123, "supportRatio": 0.05, "riskRatio": 0.84, "conflictScore": 94.9, "evidenceDepth": 100 }, "score": 67.7 }, "events": [ { "date": "2026-06-29", "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）将使平仓率非线性飙升至20%，面临940亿STAR50抛压与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "科创50", "CPO", "A股", "两融", "雪球" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "CPO", "半导体", "A股" ], "href": "" }, { "date": "2026-06-29", "title": "A股红利拥挤度与7-11月年内解禁洪峰限制了风格切换空间，而在人行短端流动性上限失灵与CNH被动升值情景下，高负债、高Beta的券商板块将面临最严重的两融净利差挤压与反转冲击。", "summary": "截至2026-06-29，本卡对A股杠铃轮动的拥挤度与压力进行评估。全市场两融达30329.92亿元（TMT占41%），银行与防御红利极其拥挤，仓位达95%分位；6-11月面临2.8-3.2万亿元解禁洪峰，今日单日解禁达589.22亿元，严重压制内需成长。若CNH继续升值或人行回购走廊失灵（DR007升至2.50%），券商两融利差收窄33.19%，导致两融…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "雪球" ], "href": "" }, { "date": "2026-06-29", "title": "在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%）；通过配以5.0%权重可转债债底防御和…", "summary": "截至2026-06-29，在两融强平与350亿雪球敲入负Gamma踩踏、以及人行短端Corridor失灵（DR007升至2.50%）的极端情景下，杠铃组合资产相关性共振跃迁（红利与成长升至+0.55至+0.70）。组合波动率由9.60%升至19.98%，日99% VaR升至2.93%（穿透2.50%上限），10日99% VaR升至9.27%（穿透7.50%…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "雪球" ], "href": "" }, { "date": "2026-06-28", "title": "虽然278%平均维保安全垫使两融连锁抛压呈局部非系统性风险，但20-30%回撤仍将触发最高2528.00亿的平仓压力并重创产业链，应通过减仓高倍数前道、超配低谷先进封装以及引入非担保品指数对…", "summary": "截至2026-06-28，两融杠杆在电子和半导体板块高度集中（总余额分别达6132.72亿和560.80亿），前五大龙头占用386.80亿。压力测试显示，20-30%的回撤会迫使大批账户跌破130%平仓线，极端情况下（科创50跌破1520）诱发2528.00亿平仓压力，并因雪球负Gamma在2-5天内引发600-750亿级联抛售。虽然278%平均担保垫等四…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "科创50", "半导体", "两融", "雪球" ], "href": "" }, { "date": "2026-06-28", "title": "在紧平衡流动性与外部融资约束下，A股难以支撑全局性风险偏好扩张，下半年配置应降低总Beta，采用“硬科技国产替代（封装/设备）+高独立现金流红利”的结构性杠铃组合。", "summary": "在美联储加息预期和国内紧平衡流动性制约下，A股面临外资流出、两融去杠杆与解禁供给的三重压力。外资流入放缓，险资增配高股息红利防御资产。两融余额高企（3.015万亿元）使科技成长面临局部去杠杆和级联平仓风险。半导体板块估值已普遍收缩30%至45%，大基金三期（3440亿元）提供战略支撑。下半年主导风格将呈现“国产替代硬科技+高壁垒红利防御”的杠铃组合。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "半导体", "A股", "两融" ], "href": "" }, { "date": "2026-06-28", "title": "券商信用风险的局部破线是否会通过流动性传导至整个金融板块，进而引发非对称的流动性危机？", "summary": "本卡对半导体担保品集中度超25%的核心券商做流动性压测，部分支持卡片06并修正其传导结论。半导体两融5,608亿元仅占两市两融余额30,009.71亿元的18.7%，故超25%是“选券商”问题而非系统总量问题。-30%尾部冲击下，多元化头部券商流动性缺口仅占净资本12.0%可轻易覆盖，但科技偏重中型与薄资本小券商分别达32.7%与49.6%，确为局部破线区…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "半导体", "两融", "雪球" ], "href": "" }, { "date": "2026-06-28", "title": "A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。", "summary": "截至2026-06-29，本卡片支持前序“更高更久”利率压制A股成长久期的判断，但将风险定位为结构性踩踏而非全市场2015式去杠杆。两融余额为30329.92亿元，占A股流通市值2.83%，总量可控；但电子板块融资余额6261.87亿元，且6月融资流入高度集中。模型显示，科创50较2026-06-26收盘下跌15%会触发约1063亿元有效被动卖压，20%跌…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "A股", "两融" ], "href": "" }, { "date": "2026-06-27", "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 工业制造分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: eb737f41-b97c-43b9-9b89-04a7c7636133 结果摘要: Wrote handoff_response.md (10188 bytes); wrote handoff_response.en.md (12356 bytes) — 三日近80…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "科创50", "A股", "两融", "雪球" ], "href": "" }, { "date": "2026-06-27", "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌破3.0%的红利反向出清防线。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "A股", "两融" ], "href": "" }, { "date": "2026-06-27", "title": "在三季度科技板块1200.00亿元解禁洪峰与科创50指数1520点平仓线前触发的2528.00亿元两融级联强平夹击下，前道半导体98.7%分位数的高昂估值缺乏安全边际，短期估值弹性无法覆盖流…", "summary": "针对2026-06-27，本报告对科创50两融平仓风险与英维克、思源电气等三季度解禁抛压（科技板块共1200亿元解禁）执行了压力测试。结果表明，科创50估值（PE 207.17倍，100%历史分位）与半导体前道估值（98.7%分位）均极度拥挤。若指数跌至1520点将触发电子/TMT板块2528.00亿元的强平波峰。虽然国产替代里程碑和业绩暴增有支撑，但美联…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "半导体", "两融" ], "href": "" }, { "date": "2026-06-27", "title": "全球商品通胀刚性与风险平价失效下的A股风格轮动：上游资源/红利价值 vs 成长，及两融与三季度解禁的交互", "summary": "全球'上游资源+红利价值'赢家只主导了沪深两市的一半：A股把组合拆开——资源赢（有色年初至今+30%以上、行业冠军[S4]），骑在第06-07卡铜/GOES缺口上；红利价值落后（中证红利一季度约+4%[S5]），因边际资金追逐成长逼空（科创50约+40%、创业板约+30%、6-22创新高[S4]）。H4L是外部贴现冲击，被内部杠杆+国产替代叙事压过[S2]…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "A股", "两融" ], "href": "" }, { "date": "2026-06-27", "title": "中国央行通过结构性再贷款与离岸FIMA人民币国债抵押回购机制，成功解耦了汇率防御与境内流动性，为A股“高股息防御+硬科技成长”的滞胀杠铃配置提供了宏观及风险平抑支撑。", "summary": "截至2026-06-27，在美中利差倒挂-265bp与低通胀下，央行转向境内结构性工具与离岸FIMA Repo防线，稳定了境内流动性（套利流出降90%至7.9B RMB）。这巩固了A股以高股息国企（海油6.2%）为防守、硬科技再贷款为成长的“滞胀杠铃”配置，抑平了368.9B两融及35B雪球负Gamma踩踏风险。优化组合VaR降至20.425%。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "雪球" ], "href": "" }, { "date": "2026-06-26", "title": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论", "summary": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论 美元兑日元跌破161.50关口，财务省干预概率陡增。一旦日元套利平仓瞬间引爆全球EM科技股抛售，叠加SpaceX二次配售24.5亿元的资金虹吸，A股电子板块6132.72亿元两融及科创50雪球在1500点敲入区附近会否触发自反馈平仓踩踏？", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "neutral", "supportLabel": "观察", "score": 23.75, "matchedTerms": [ "科创50", "A股", "两融", "雪球" ], "href": "" }, { "date": "2026-06-26", "title": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局", "summary": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局 在上午科创50跌破关键支撑触发雪球敲入与两融强平后，午间主力ETF资金流入与个股金针反转是否标志着负Gamma抛售压力的阶段性出清？我们应如何利用这一估值剪刀差进行右侧建仓，从红利低波向高Beta科技股进行再平衡？", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 23.5, "matchedTerms": [ "科创50", "A股", "两融", "雪球" ], "href": "" }, { "date": "2026-06-26", "title": "2026-06-26确认科技两融堰塞湖进入去杠杆阶段，但未确证牛市陷阱；最优策略是降拥挤科技、保现金缓冲、分批增配电力电网等物理红利资产。", "summary": "截至2026-06-27，结论是支持防守再平衡，但不把2026-06-26下跌直接视为“国家队牛市陷阱”的确证。A股两融余额已升至3万亿元以上，电子行业两融余额超过6000亿元且占比约两成，说明风险是融资拥挤和流动性冲击，而非科技周期被证伪。建议科技从超配降至标配，优先削减融资敏感电子/CPO仓位，保留高质量科技核心，同时把资金分批转向电力、电网、受监管公…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "CPO", "A股", "两融" ], "href": "" } ], "timeline": [ { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 1, "value": 33, "events": [ { "title": "科创板极端护盘对A股整体流动性的虹吸效应与跨板块资产配置调整 研究日期： 2026 05 22 作者： 首席策略师 (Chief Strategist) 立场： Synthesize…", "summary": "科创板极端护盘对A股整体流动性的虹吸效应与跨板块资产配置调整 研究日期： 2026 05 22 作者： 首席策略师 (Chief Strategist) 立场： Synthesize (综合) 执行摘要 截至 2026 05 22 [S7]，科创50指数收报 1,764.17点 [S1]，极度逼近…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球宏观压力测试：科创板极端护盘外溢效应与跨境流动性虹吸下A股资产配置的脆弱性 研究日期： 2026 05 22 作者： 全球宏观分析师 (Global Macro Analyst)…", "summary": "全球宏观压力测试：科创板极端护盘外溢效应与跨境流动性虹吸下A股资产配置的脆弱性 研究日期： 2026 05 22 作者： 全球宏观分析师 (Global Macro Analyst) 立场： Stress test (压力测试) 执行摘要 截至 2026 05 22 [S1]，科创50指数收报 1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "英伟达光环可能带来最后反抽，但在高杠杆、高换手、券商风控收紧与美债收益率上行共振下，科创50的风险收益已转向防踩踏和高低切换。", "summary": "截至2026-05-22，本卡对“英伟达财报带动科创50最后一冲”采取压力测试立场。英伟达2027财年一季度业绩强劲，但A股在2026-05-21已出现科创50盘中冲高后收跌、成交显著放大的末端挤压特征。两融资金集中流入电子和半导体，科创板两融余额继续上行，券商新增115%平仓线压缩补仓时间，美债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 2, "risks": 0, "neutral": 0, "value": 59.4, "events": [ { "title": "AI利好兑现后的宽度轮动与A股TMT杠杆再定价", "summary": "AI利好兑现后的宽度轮动与A股TMT杠杆再定价 AI breadth rotation and A-share TMT leverage repricing after the beat NVDA利好兑现、Russell 2000跑赢、科创50午后反弹但半导体两融净偿还并存，团队应把成长仓位解释为…", "sourceLabel": "研究主题", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI利好兑现后的宽度轮动与A股TMT杠杆再定价", "summary": "AI利好兑现后的宽度轮动与A股TMT杠杆再定价 NVDA利好兑现、Russell 2000跑赢、科创50午后反弹但半导体两融净偿还并存，团队应把成长仓位解释为健康扩散，还是高位降Beta后的短线修复？", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 4, "neutral": 0, "value": 20.8, "events": [ { "title": "在科创板两融信用收缩与Q3迎来1200亿解禁洪峰的微观去杠杆环境中，大健康精选篮子建仓必须实施『恐慌底分批-解禁潮吃货』的节奏化策略，并将有效暴露压缩至18%-20%以内，辅以Bet…", "summary": "截至2026-05-25，本卡对创新药出海、CXO地理标签及高端器械替代的精选篮子（联影、迈瑞、奕瑞、百济等）进行A/H股映射与估值管理。A股两融余额见顶于2.87万亿元，科创50在1,750点承受去杠杆与432亿雪球敲入压力。策略建议在科创解禁潮与两融风暴中分三阶段节奏建仓（1-3月恐慌期买30…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国家队'指数级'护盘掩盖了承做商对冲、两融杠杆与量化因子衰减三条链条'个股流动性级'的尾部敏感性,科创50离开1,750不等于系统离开生死线。", "summary": "今日科创50大涨3.56%脱离1,750雪球生死线，但只是延迟而非解除风险——半导体权重票+国家队ETF把指数托起的同时，承做商对冲仓位被动收缩、两融余额冲至2.9万亿历史峰值[S3]、量化中性产品超额已连续两季度回吐[S5]。'非对称虹吸'体现在:资金流入集中于科创50权重前10票与红利端，流出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50杠杆突围：半导体暴涨与大盘风格‘非对称虹吸’的去杠杆冲突", "summary": "科创50杠杆突围：半导体暴涨与大盘风格‘非对称虹吸’的去杠杆冲突 STAR-50 Leverage Breakout: Semiconductor Short-Squeeze vs. Small-Cap Siphoning Deflation 科创50今日大涨3.56%拉离1,750点雪球生死线，…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50杠杆突围：半导体暴涨与大盘风格‘非对称虹吸’的去杠杆冲突", "summary": "科创50杠杆突围：半导体暴涨与大盘风格‘非对称虹吸’的去杠杆冲突 科创50今日大涨3.56%拉离1,750点雪球生死线，但大盘小盘股呈现非对称失血。国家队定点护盘与两融‘被动套牢’并存，如何评估3.2万亿量化私募因子衰减与Flash Decay同步失效的高频尾部清算风险？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 6, "neutral": 3, "value": 19.2, "events": [ { "title": "A股当前处于“空心科技虹吸”导致的流动性黑洞期，应警惕科技股杠杆踩踏，向“物理红利”（大水电、核电）防御端极化靠拢。", "summary": "2026年5月27日，A股呈现极端的“空心科技虹吸”特征，半日成交2.07万亿但超4600股下跌，资金极度抱团CPO板块。受美伊停火导致油价暴跌15%及日元159引发的去杠杆冲击影响，全市场两融及雪球脆弱性急剧上升。建议警惕高拥挤度科技股的平仓风险，将防御端转向由大水电、核电组成的“WLEF物理红…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在TMT板块极端拥挤和高实际利率压制下，今日应坚决停止加仓一阶AI硬件，战略性转向配置确定性极高的二阶物理瓶颈（电力/铜需求/液冷散热），构建杠铃配置以防范筹码去杠杆及估值回调风险。", "summary": "在A股TMT成交占比攀升至44%-45%以及科创50两融拥挤度逼近4.90%的历史极端警戒水位下，第一阶AI硬件（半导体/HBM）的筹码结构已极度脆弱。同时，全球2.0%-2.2%的高实际利率持续压制长久期高估值资产。今日策略坚决不加仓一阶AI芯片，转而顺着AI产业的“物理刚性约束”向二阶瓶颈（高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股“空心科技虹吸”与双重冲击下的杠杆出清与真杠铃重构", "summary": "A股“空心科技虹吸”与双重冲击下的杠杆出清与真杠铃重构 A-Share 'Hollow Tech Siphon' and Leveraged Deleveraging Reshaping the True Barbell 在美伊停火谈判引发原油暴跌约15%（Brent至$95.05）和USD/JP…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股“空心科技虹吸”与双重冲击下的杠杆出清与真杠铃重构", "summary": "A股“空心科技虹吸”与双重冲击下的杠杆出清与真杠铃重构 在美伊停火谈判引发原油暴跌约15%（Brent至$95.05）和USD/JPY 159去杠杆冲击压制AT1银行红利的背景下，A股今日出现极端分裂行情：半日成交2.07万亿，超4600只个股下跌，但CPO板块创历史新高。这种“空心科技虹吸”导致…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 3, "neutral": 0, "value": 29.6, "events": [ { "title": "MSCI调仓本身不是系统性事件，但落在A股两融2.87万亿/物理上限2.90万亿仅1%缓冲的高敏感点上，使得TMT龙头任何-2%以上的跳水都会被'雪球敲入×维保线×流动性真空'三重叠…", "summary": "本卡对前两卡'被动买盘被主动MOC冲抵、TMT次日温和回吐-0.5%~-1.5%'的结论进行压力测试。结论：基线情景(P≈70%)下两融余额单日收缩80-140亿元，2.87万亿/2.90万亿物理上限的300亿元缓冲仍可承受，系统性风险可控；但厚尾情景(P≈20%)下科创50若跌破1,750点雪球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50指数在1,750点附近存在极高的信用与衍生品二次平仓反馈共振深渊，下周两融去杠杆风暴极易打压AI硬件龙头股估值，建议维持45%的红利防御性配置上限并买入虚值认沽期权对冲尾部风…", "summary": "本报告定量评估了科创50指数在1750点附近的雪球敲入（22.0亿元名义本金）与做市商Delta动态对冲（12.0亿元卖压）的二次共振效应。在券商两融折算率收紧（0.60至0.40）压力测试下，一阶总抛压达37.0亿至57.0亿元，极端踩踏下可触发二阶追加平仓反馈卖压119.7亿元，共振总抛压达1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在雪球敲入与两融平仓共振的极端风险下，以‘先进封装’结构性机会对冲Beta下行，并以虚值期权锁定尾部风险是当前最优防御路径。", "summary": "截至2026-05-29，A股两融余额超2.9万亿且科技板块极度拥挤。科创50指数逼近1,750点雪球密集敲入区，做市商面临121.5亿元刚性抛压，叠加ETF折价套利，流动性死亡螺旋风险剧增。建议配置14%比例的先进封装‘防守方案四’（5%设备+6%材料+3%存储）并配合虚值认沽期权构建多重防线。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 2, "risks": 2, "neutral": 1, "value": 56, "events": [ { "title": "DELL验证AI服务器需求后，A股硬科技去杠杆还能否买回", "summary": "DELL验证AI服务器需求后，A股硬科技去杠杆还能否买回 Can China hard-tech be bought after Dell validated AI server demand? DELL的$43.8B收入、$16.1B AI服务器收入和$60B FY27目标，是否足以把A股CPO…", "sourceLabel": "研究主题", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "DELL验证AI服务器需求后，A股硬科技去杠杆还能否买回", "summary": "DELL验证AI服务器需求后，A股硬科技去杠杆还能否买回 DELL的$43.8B收入、$16.1B AI服务器收入和$60B FY27目标，是否足以把A股CPO/PCB/半导体从去杠杆交易切回订单验证交易？还是必须等科创50 1750、两融和ETF赎回同时企稳？", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "A股两融安全边际充沛，科创50雪球实际规模仅为62亿元且敲入卖压平滑，MSCI被动重估不存在所谓的“流动性真空砸盘”，跨境日元期权对冲完全多余且损耗超额收益。", "summary": "本卡作为金融行业分析师，强烈反驳A股策略师的悲观观点。首先，两融高位有2026年保证金100%新规与2025年风控标准严密保护，平均维持担保比达248.5%，个股需跌60%以上才触及强平线，绝非指数大跌16.92%所能引爆。其次，科创50雪球名义存续实际仅为62亿元且极均匀分布，做市商采用平滑期权…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 报告日期 ：2026年05月30日 分析师 ：A股策略师 ( ashare strategist ) 密…", "summary": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 报告日期 ：2026年05月30日 分析师 ：A股策略师 ( ashare strategist ) 密级 ：内部机密 | 白板研讨卡 01/08 核心结论与概要 (Executive Summary) 今日 (2026…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈", "summary": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 今日是MSCI半年度重估正式生效日。虽然散户由于昨日创业板+1.96%的V型反弹而呈现极度FOMO，但A股两融余额正高挂于2.9367万亿元历史极值，融资交易占比达10.09%。科创50指数（1,844.25点）距离下方…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 2, "neutral": 1, "value": 42.8, "events": [ { "title": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守", "summary": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守 Defending the 1,750 Lifeline: Snowball Gamma Waves and Margin Call Risk 科创50收于1,751.32点，距离1,750雪球/两融生命线仅一步之遥。若下周破位跌破1…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论", "summary": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论 A-Share Post-Close Sentiment Fission: MSCI Rebalancing vs. Margin & Snowball Risks 今日A股演练了11:27分V型奇袭与科技大挤空，但成交…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论", "summary": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论 今日A股演练了11:27分V型奇袭与科技大挤空，但成交量缩量2704亿元。盘后社媒上散户情绪在英伟达/Marvell史诗财报刺激下极度亢奋，期待明日MSCI半年度生效日的被动天量申购。然而，两融高挂于2.9367万亿元，…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 10, "neutral": 2, "value": 12, "events": [ { "title": "SpaceX窗口的A股风险不是整体硬科技崩盘，而是两融账户在-6%至-18%组合波动区间内触发补仓和被动平仓。", "summary": "截至2026-06-01，本卡片对SpaceX上市窗口下A股硬科技/军工组合做VaR与追保压力测试。以100份标准化组合计，单日99% VaR为-5.9%，IPO窗口VaR为-10.4%；基准、悲观、跳空尾部压力损失分别为-8.3%、-18.0%、-25.7%。若初始维持担保比例为160%，组合亏…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "概述 作为A股策略师（ ashare strategist ），针对看板第三步主题 “联合宏观矩阵下A股信用与衍生品结构性杠杆风险特征” ，我已对科创50指数回测1750生死线下的两…", "summary": "概述 作为A股策略师（ ashare strategist ），针对看板第三步主题 “联合宏观矩阵下A股信用与衍生品结构性杠杆风险特征” ，我已对科创50指数回测1750生死线下的两融去杠杆与雪球衍生品敲入进行了深入的定量压力测试与模���仿真。 在此项研究中，我全面 支持 首席经济学家（卡片 0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在信用盘平均担保比例深度压缩至148%的背景下，指数探底1,600-1,650点将引爆高达34.79亿元的两融爆仓清算与每日4.90亿元的盘中踩踏；国家队每日5.0亿元的588000…", "summary": "截至2026-06-01，本报告针对科创50指数下探1,600-1,650点极限历史底部，压力测试了两融信用盘清算节奏与国家队ETF托底机制。测算表明，挂钩科创50的两融债务达135亿元，平均维持担保比例已降至148%。若指数跌破1,658点将触发21.75亿元债务强平；跌至1,600点时累计强平…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守", "summary": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守 科创50收于1,751.32点，距离1,750雪球/两融生命线仅一步之遥。若下周破位跌破1,750，我们如何防范882亿元的级联式去杠杆抛压？A股红利与成长的配置比例应如何调整？ Prior related institute wor…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 1, "risks": 6, "neutral": 0, "value": 26.6, "events": [ { "title": "在美联储加息与人行降准的联合概率矩阵下，A股必须从传统哑铃策略转向“防御底仓+对冲Beta”模式，规避高融资与高解禁共振的科创标的。", "summary": "由于卡片04数据受损，本报告直接综合卡片01与03。针对科创50指数在1750点的雪球敲入与2.9万亿两融平仓共振风险，本报告提出A股风格轮动与行业配置优化方案：将原10%的科技敞口转为“对冲型科技”，并将高股息与出口制造权重分别提升至35%与25%，以应对6月3237.5亿元解禁悬崖。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股两融与资金流数据支持AI软件+网络安全作为可配置主题，杠杆密度可控且杠杆错位在软件一侧创造左侧分批吸纳窗口，但执行须坚持杠杆纪律。", "summary": "2026-06-03 卡片04支持将AI软件与网络安全列为可配置主题，并从市场结构角度细化执行节奏。A股两融余额2026-05-29达28909.73亿元(占流通市值2.75%)、低于2015年4.72%密度峰值；杠杆资金集中在电子硬件(5/6-5/8电子3日融资净买入285亿元、硬件+半导体占全…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "A股应保留低拥挤现金流beta，但必须在强平数据显性出现前出清高两融CPO/TMT，因为真正红线是130%维持担保比例叠加20%跌停的流动性跳空。", "summary": "截至2026-06-03，本卡综合前序尾部保护主线，认为A股风险不在系统性beta本身，而在高两融TMT/CPO的本地强平链条。沪深交易所最新可抓取的2026-06-01合计两融余额约人民币2.8954万亿元；新增100%融资保证金账户从初始约200%维持担保比例跌至130%需约35%价格跌幅，遗…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "前序SPX/VIX/CDX尾部保护层足以吸收A股TMT强平链条的P&L冲击，并将‘跌停板流动性锁定’风险转化为现金可得性优势。", "summary": "在USD/JPY跌破145触发全球去杠杆的情景下，高两融A股CPO/TMT板块（约占总组合10%权重）若在回撤15%基础上再触及20%跌停，将导致组合回撤2.75% NAV并触发约65%两融仓位的130%强平红线。压力测试显示，前序配置的110% NAV SPX 92/82 put spread和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 3, "neutral": 0, "value": 38, "events": [ { "title": "AI硬件主线已足以把成长仓位上调至12%-14%，但红利端只应降到52%-55%而非清仓，组合仍需现金和对冲压舱。", "summary": "截至2026-06-04，本卡支持将AI硬科技成长仓位从8%-10%的防守下限上调至12%-14%，但15%仍是硬上限。加仓资金主要来自红利端从60%下调至52%-55%，而不是撤掉现金和指数对冲。理由是海外AI硬件订单与A股CPO/半导体链形成明确映射，但A股宽度仍弱、科创板两融仍在收缩，说明只…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美股AI硬件强势对科创50半导体链存在明确补涨逻辑，但仅在先进封装和28nm/14nm主流设备两条主线上'清晰可定价'，应以'订单确认+流动性双重确认'同时为纪律抓手。", "summary": "针对2026-06-03美股AI硬件强势能否传导至科创50半导体链的核心问题，TMT分析师支持卡01战术框架，并把补涨逻辑结构化为可定价的两条主线：(1)先进封装（长电600584/通富002156/华天002185/兴森002436），承接台积电CoWoS 25-30%的产能溢出；(2)28nm…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "下一轮0DTE式日内波动若触发两融担保品重估，最先受压的不是广义沪深300，而是AI硬件/半导体/算力链中的科创与创业板高Beta担保品。", "summary": "2026-06-04：本卡支持前序观点，并把A股两融脆弱点定位在AI硬件、光模块、PCB、半导体和算力链，而不是整个沪深300。2026-06-02沪深两融余额约2.897万亿元，融资余额约2.876万亿元。深市中际旭创、新易盛、胜宏科技，沪市寒武纪、兆易创新、澜起科技等体现高余额和高融资买入共振…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 3, "neutral": 0, "value": 39.4, "events": [ { "title": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？", "summary": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？ Broadcom AI Guide-Down Triggers Global Semi Sell-Off: Will A-Share Tech Break Support Amid Leverage & Lo…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "博通指引转保守与A股流动性三压共振，科创50面临1720点雪球破位风险，需警惕系统性去杠杆引发的二次筑底。", "summary": "博通（AVGO）隔夜暴跌13-15%打破了AI硬件订单无限增长的叙事，引发全球半导体去Beta。A股科创板正面临4.9%高两融占比、6月3332.85亿解禁洪峰以及科创50指数（现报1726）逼近1720雪球密集敲入区的“三体问题”。若破位将引发做市商负Gamma抛售。建议维持45%红利底仓，并买…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美股大厂利润受损与NDX估值收缩将导致A股高拥挤AI硬件链面临-34.3%的估值与盈利双重杀跌压力，并触发74.5%概率的两融去杠杆踩踏；而绝对利差高达+249 bps的A股红利资产…", "summary": "基于2026-06-04，本报告量化评估了美股NDX重估对A股AI算力链（CPO、PCB）及红利资产的溢出与共振效应。测算表明，A股AI硬件等权篮子交易于32.5x的2026E P/E，其估值对美股弹性系数为1.35。美股大厂因PPA及电网延迟导致的利润率受损及NDX 20%的估值压缩，将使A股算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 4, "neutral": 0, "value": 34.8, "events": [ { "title": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？", "summary": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？ 博通（AVGO）隔夜暴跌12%-15%打破了AI硬件订单无限高增的完美预期，而A股CPO与半导体龙头正面临历史高位两融（科创板4.9%占比）和6月3332.85亿的解禁洪峰。如果科创50指数被动向1750/1…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50 1,650点是‘破位即反转’的结构性再进场窗口，TMT首选反弹先导为国产算力、存储与半导体设备三剑客。", "summary": "科创50在1,650点构成‘政策+估值’双底再进场窗口而非系统反转。1,650对应42x PE-TTM (15%分位)与2025-09政策行情颈线，叠加央行降准+SFISF流动性底座与‘1+6’科创政策支持。雪球敲入完成后做市商负Gamma抛压耗尽、解禁实际净减持约1,200亿可消化。TMT细分中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在强劲的全球与国内资本开支周期、国家大基金三期以及央行科技再贷款的托底下，半导体与AI硬件等TMT核心资产的估值底线十分坚固，不会因外资流出而系统性破裂。", "summary": "针对日元平仓引发外资流出高β板块的担忧，本报告进行压力测试，判定半导体与AI硬件等TMT核心资产的估值底线不会破裂。台积电2026年高资本开支与大基金三期（3440亿元）、央行科技再贷款（1.2万亿元）构成双重安全垫。外资持仓仅占A股半导体自由流通市值的4-6%，即使流出也难撼动估值底，但需警惕本…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融、雪球与股指期货去杠杆共振时，防线必须从静态套保改为预留现金、动态降杠杆和红利锚容量配给，否则对冲会先于底仓卖出能力而失效。", "summary": "截至2026-06-06，本卡压力测试认为，A股尾部风险的核心不是缺少IM、期权或红利资产，而是底仓跌停锁死时，对冲端被保证金和盘中反弹先行强平。最新可得两融口径显示，2026-06-04沪深融资余额约2.89万亿元；若3%-6%被动退出，可能形成约867亿-1734亿元卖压。长江电力可作为红利锚…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 5, "neutral": 0, "value": 30.2, "events": [ { "title": "外部高利率与全球去杠杆冲击下，A股面临硬科技两融去杠杆与解禁潮共振风险，资产配置应剪枝高位拥挤科技与昂贵资源，退守低估值高股息周期价值与短端利率债以优化组合VaR。", "summary": "本报告分析了美联储维持“更高更久”利率对A股资金与风格轮动的影响。美债收益率升至4.75%和日元平仓去杠杆加剧了北向资金流出压力（预计600亿至900亿元），推高离岸HIBOR至8.0%-10.0%。两融杠杆高企（2.91万亿元）且高估值科技拥挤，科创50击穿1750点流动性红线触发150亿雪球敲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在系统性流动性危引发避风港资产补跌与相关性收敛的背景下，跨资产配置至实物黄金与短债，并配合科创50 ETF Put与IM期货空头对冲，是防御VaR穿透并平衡权益融涨弹性的最优解决方案。", "summary": "针对2026年6月7日A股2.90万亿两融余额极值及雪球敲入踩踏引致的系统性流动性危机，本压力测试表明长江电力等避风港资产在挤兑期存在补跌风险，相关性收敛导致未对冲组合10日损失达-16.60%，99% VaR穿透至32.33%[自测算]。本方案设计了科创50 ETF期权与IM期货空头对冲包，并对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "TMT行业分析报告：高利率环境下AI硬件与半导体的盈利变现与“高质量”属性评估 在美联储政策���率维持在高位平台期的背景下，估值受利率压制的压力凸显。本期研究围绕 “AI硬件与半导…", "summary": "TMT行业分析报告：高利率环境下AI硬件与半导体的盈利变现与“高质量”属性评估 在美联储政策���率维持在高位平台期的背景下，估值受利率压制的压力凸显。本期研究围绕 “AI硬件与半导体板块的盈利确定性” 展开，深入剖析了���游先进芯片的议价权、下游云厂商的资本开支与投资回报率（ROIC）鸿沟、以…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在融资杠杆处于极值且科创50下面临雪球敲入踩踏的情景下，超配低贝塔、高分红的避风港资产长江电力是防守去杠杆和风格转换中对冲系统性风险的最优策略。", "summary": "截至2026年6月4日，A股融资余额达2.90万亿元高位，科创50自高点回调至1668.33点，逼近1125-1360点的雪球敲入密集区。一旦触发敲入，券商平仓多头将引发市场二次下跌并导致融资盘连环爆仓。建议超配长江电力（Beta为0.09，股息率3.7%，承诺70%以上分红率）等低贝塔红利资产，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 2, "neutral": 1, "value": 52.6, "events": [ { "title": "中际旭创明日解禁直接抛压微弱，但在全月大额解禁、两融折算率调降预期及科创50雪球敲入共振下，股价若下跌13.33%-15.40%极易引爆科技板块非线性两融强平螺旋，需严格规避高两融占…", "summary": "本报告针对中际旭创明日（2026年6月8日）解禁及半导体/CPO两融杠杆进行压力测试。中际旭创明日解禁32.45万股（约4543万元），仅占日成交额1.14%，直接冲击有限。但在全月3332.85亿元大额解禁、电子通信行业两融高达7389.32亿元、两融折算率收紧拟调降至0.4以及科创50破位敲入…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 科创50指数雪球敲入负 Gamma 对冲抛压与两融折算率下调去杠杆踩踏模拟", "summary": "致 首席风控官： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 73279e4e-699e-4d9e-9366-ea168709088a 结果摘要: Wrote handoff_response.md (14450 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI需求弹性短期救不了估值，CPO龙头先经历两融出清再迎1.6T切换主升浪；先防御、后进攻。", "summary": "本卡作为收官卡，对CPO龙头（中际旭创、新易盛、天孚通信）在关税与两融去杠杆下的基本面韧性进行压力测试。北美四大超大规模厂商2026年合计资本开支约3400亿美元(+38%YoY)，800G/1.6T切换支撑光模块TAM 195-210亿美元，需求端未破。但22.5%美关税对各龙头EPS拖累3-7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 7, "neutral": 1, "value": 24, "events": [ { "title": "在6月3333亿元解禁洪峰与2.9万亿高杠杆的供给与流动性压力下，半导体及CPO板块拥挤的交易结构和部分高估值板块尚未被业绩完全消化，面临估值收缩风险。", "summary": "截至2026-06-08，本卡对半导体与CPO板块进行压力测试。结论是虽然AI高增长将中际旭创2026年预测PE拉低至23x-40x，但海光信息预测PE仍高达140x-170x，且6月A股面临超3330亿元的解禁洪峰（电子行业居首），叠加2.9万亿高额两融与拥挤筹码，当前盈利预期并未充分消化供给与…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A股科创50跌破零Gamma生死线后两融强平与雪球级联踩踏的微观杠杆压测", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 23d2a86c-8b70-46ce-8620-e44b5d309d0d 结果摘要: Wrote handoff_response.md (3101 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "市场结构确认偏成长质量哑铃的方向，但威胁其最拥挤的半导体/AI硬件表达方式。", "summary": "截至2026-06-08，本卡压力测试card-02的偏成长质量哑铃。结论是市场结构没有否定成长倾斜：两融相对市值仍低于历史峰值，6月5日指数下跌时个股广度仍为正。但电子/半导体融资净买入、成交占比和6月解禁供给高度重叠，说明风险集中在高弹性成长腿。建议保留成长倾斜，降低半导体/CPO高拥挤追涨仓…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面对2.9万亿两融余额与6月天量解禁共振，A股配置应执行“降维拥挤成长，夯实防御压舱”的35/40/25战术区间。", "summary": "截至2026-06-08，基于前期宏观内需缺口和市场结构拥挤度测试，本卡将偏成长的非对称质量哑铃转化为定量的战术区间。建议在6月下旬至7月采用35/40/25的权重：将高拥挤的电子/半导体成长腿降至35%并向工业自动化平移；将红利与现金流防御压舱提升至40%，以应对2.9万亿两融及超3300亿解禁…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 3, "neutral": 0, "value": 46.4, "events": [ { "title": "离岸CNH流动性收紧迫使北向杠杆资金去杠杆，这与境内两融平仓及雪球负Gamma抛售共振，加剧了科创板及高弹性权重股的流动性压力。", "summary": "CNH HIBOR飙升至7.30%抬高了离岸融资成本，导致北向杠杆资金因年化-2.8%的负carry面临去杠杆压力。这与境内融资折算率下调（966亿保证金缺口）及科创50跌破1600点零Gamma防线引发的270亿雪球敲入卖压共振，导致外资抛售半导体与TMT高弹性权重股。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO引发的全球流动性‘真空’是导致科创50破位1600点的核心外因，破位后将诱发两融盘非线性踩踏。", "summary": "SpaceX IPO的450-750亿美元流动性虹吸正通过中概股套利链条向A股传导。压力测试显示，若科创50跌破1600点关键支撑，将触发券商折算率下调与融资盘抢跑的共振。虽然当前距离1310点强平线仍有空间，但技术破位后的非线性踩踏风险从5%上升至20%以上。建议利用看跌价差对冲，并保持15%现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红利仍是底仓，但红利拥挤已经成为风险；在外盘芯片黄灯下，A股科技应按ASIC/服务器瓶颈优先、宽基科创50后置的顺序分批回补。", "summary": "截至2026-06-09，本卡支持前序“外盘芯片黄灯、A股科技不能盲目追涨也不能单边放弃”的判断。2026-06-09科创50收涨4.17%至1,663.11点，中证红利跌0.06%，但科创50仍低于1,668.33点破位前收盘。建议把科技、红利、缓冲仓基础配为45/45/10；先从煤炭等拥挤红利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 1, "risks": 5, "neutral": 2, "value": 49.8, "events": [ { "title": "100亿元缓冲资金的最优A股结构通道是70%ETF一级申购 + 20%收盘竞价权重托底 + 10%大宗折价承接，且必须经由非两融账户执行才能真正切断二次去杠杆反馈环。", "summary": "在ETF折价、成分股大宗、收盘标记三线同压的情景下，100亿元AI/半导体缓冲资金最稳妥的执行组合为：ETF一级申购70亿（消化折价）+ 收盘集合竞价权重托底20亿（守雪球敲入与两融维持担保线）+ 大宗折价−2%承接10亿（吸大单不留痕）；二级连续竞价与转融通回收两条通道必须拒用。所有承接必须通过…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在美联储鹰派重定价与科创板局部杠杆脆弱性的双重挤压下，A股配置应果断转向低杠杆红利防御，保留科技龙头仅作为极低仓位的期权观察。", "summary": "截至2026-06-10，科创50指数（1652.22点）月内跌幅达5.66%，显著跑输红利指数（-2.10%）。两融数据显示资金正从电子（半导体）板块剧烈流向能源、银行等高股息行业。虽然科创50尚未触及1468.6点的广谱强平线，但局部杠杆压力已迫使配置风格从高弹性成长向低杠杆防御切换。建议压低…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO正在触发AI权重股的可持续去拥挤，但交易方向应是低配拥挤AI硬件、增配上游稀缺投入品和低资本开支AI应用，而不是卖出全部AI。", "summary": "截至2026-06-10，本卡片判断SpaceX IPO不是单日情绪噪音，而是触发AI龙头去拥挤的真实催化：SpaceX预计净募资744亿至857亿美元，募资用途包括AI算力基础设施；Google合同强化AI资本开支叙事；TSLA-SpaceX并购期权抬升Musk平台估值分歧；美股半导体与A股CP…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "A股TMT板块高估值AI标的的机构筹码拥堵度与融资融券杠杆交易脆弱性评估", "summary": "锚定2026-06-10，本卡承接第04卡，将视角从基本面转向筹码与杠杆脆弱性，支持前序方向但补一个风险锚：A股最拥挤AI硬件龙头正处于三重叠加窗口。中际旭创一季末被1163只基金重仓738.96亿元、公募持股占流通股11.74%、蝉联全市场第一大重仓，新易盛居前三；主动偏股基金AI硬件配置31.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 3, "neutral": 0, "value": 49.2, "events": [ { "title": "国家队科创50防守策略在‘一级申购优先’与‘二级折价才买’的限制下将形成严重的防御真空与套利漏洞，导致二级市场半导体与AI龙头在防线建立前遭遇流动性错配引发的估值踩踏。", "summary": "本卡片对科创50在‘宽基先行、科创后置’策略下的执行路径进行压力测试。分析指出，‘一级申购优先’机制存在显著的算法建仓时滞，面对两融平仓与雪球敲入形成的710亿元级联抛压（对前五大半导体巨头达106.5亿元/天），国家队12亿元/天的日均算法买盘存在8.88倍的流动性赤字。此外，‘二级折价才买’规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CPO两融去杠杆会放大创业板和科创50的流动性冲击，但只有在红利底座同步失效、相关性收敛的系统性赎回情景下，才会击穿方案四-2.6169%的VaR线。", "summary": "2026-06-11：CPO基本面下修会通过中际旭创、新易盛、天孚通信合计732.02亿元融资余额向创业板权重股传导，并与科创板3742.22亿元融资余额形成成长板块二次卖盘。但在方案四下，0.25% CPO仓位本身不是击穿-2.6169% VaR的主因；只要红利55%底座维持低波低相关，局部强平…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在雪球敲入与ETF折价赎回同步发生时，科创50若坚持把二级折价买入作为主规则，将无法抵御半导体权重的流动性踩踏。", "summary": "截至2026-06-11，本卡片支持卡片07的压力方向：科创50不能把“二级折价才买”作为极端波动下的主防线。华夏588000 2026-06-11 PCF显示当日赎回上限可对应78.561亿元成份股篮子供给；若叠加雪球敲入后Delta从1.2-1.5回落到1附近带来的卖出超额对冲，基准情形成份股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 5, "neutral": 1, "value": 41.6, "events": [ { "title": "海外risk-on与A股科技去杠杆的错位", "summary": "海外risk-on与A股科技去杠杆的错位 Global risk-on versus A-share tech deleveraging 隔夜美股半导体与小盘逼空是否足以覆盖科创50/CPO的本地两融、解禁和折算率压力，还是应继续把科技回补限制在5%试仓以内？", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海外risk-on与A股科技去杠杆的错位", "summary": "海外risk-on与A股科技去杠杆的错位 隔夜美股半导体与小盘逼空是否足以覆盖科创50/CPO的本地两融、解禁和折算率压力，还是应继续把科技回补限制在5%试仓以内？ Prior related institute work to acknowledge and BUILD ON (do not r…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "隔夜海外beta脉冲无法对冲A股科技的本地两融、解禁、折算率三重结构约束，科创50/CPO回补应继续硬性限制在5%试仓以内。", "summary": "维持科创50/CPO科技回补5%试仓上限不变。隔夜美股半导体与罗素逼空属beta脉冲，半衰期2–5日，无法改变A股科技端三重本地约束：两融余额被动去杠杆（时间常数2–4周）、6–9月解禁洪峰（日历事件、对海外情绪零弹性）、券商折算率下调后的二阶强平门槛（恢复需波动率连续4–8周回落）。三者时间常数…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "本地两融回落、6-9月科创/创业解禁墙和既有0折算率池共同压制A股科技再加杠杆，海外risk-on只能提供交易窗口，不能支撑仓位上调。", "summary": "2026-06-12数据核实支持上一卡的本地约束判断，但需要修正表述：6月券商公告未显示新的科技股大面积折算率下调，真正约束来自既有0折算率池。两市融资余额较2026-05-29下降471.57亿元；6-9月全A解禁约9,300.88亿元，科创板+创业板明细合计5,522.78亿元；兴业证券202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 0, "value": 64, "events": [ { "title": "A股杠杆结构在100%保证金规则下韧性显著，红利退出压力可控，风格切换的主要风险在于科创50杠杆增长过快引发的情绪过热，而非两融爆仓。", "summary": "针对2026-06-14的A股杠杆结构压力测试显示，红利板块虽处于90%拥挤度高位，但得益于100%融资保证金规则，具备35%的价格安全垫，风格切换预计表现为“有序撤离”而非“平仓崩溃”。科创50融资余额触及自由流通市值5%的情绪极值，虽理论容量巨大，但需警惕情绪过热。整体质押风险局部化且受控，市…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 2, "neutral": 0, "value": 67.4, "events": [ { "title": "科创板去杠杆冲击短期股价但不破坏国产半导体设备及AI硬件的长期成长逻辑——龙头与AI芯片设计估值修复~20%即可加仓，二线与HBM配套需-30%以上才有安全边际。", "summary": "截至2026-06-15，科创板两融由~2,940亿元降至~2,820亿元、科创50击穿1,600点零Gamma线，对电子板块形成真实短期冲击，但不足以系统性下修国产半导体设备与先进制程龙头capex。设备龙头（北方华创订单覆盖15个月、现金~150亿元）、AI芯片设计（寒武纪2026Q1+312…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "互联网大厂AI capex资金储备充沛且免疫于A股两融收缩，美BIS管制摩擦虽造成300亿元海外采购延后，但倒逼国产算力采购占比由30%大幅上修至58%，在2026H2为国内半导体设…", "summary": "基于2026-06-15，本报告定量评估了地缘与流动性双重压力下中国大厂AI Capex的节奏与传导。测算表明，四大互联网大厂（阿里、字节、腾讯、百度）2026H2的4200亿元AI capex预算完全免疫于A股两融去杠杆，其在海外上市且拥有数千亿现金。虽然美BIS出口管制限制英伟达H20/H20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基于2026-06-15，国产算力+518亿元新增订单支撑下，A股科技板块两融与机构仓位重构具备坚实基本面，科创50重回1600点零Gamma上方将触发右侧杠杆盘强力追入与风格轮动反…", "summary": "基于2026-06-15，本报告定量评估了国产算力爆发背景下，A股科技板块（芯片设计、算力设备、PCB）的两融与机构仓位再平衡。大厂H2新增的518亿元国产算力订单有效对冲了两融收缩1200亿元的短期流动性冲击。公募基金TMT配置预计将从14.2%上修至19.8%，带来+2520亿元增量资金。科创…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-15", "supports": 1, "risks": 4, "neutral": 0, "value": 56.8, "events": [ { "title": "若日元融资压力引发全球半导体估值去化，A股成长与AI板块需筑防线，应将科创50、创业板指、AI硬件及国产设备分别降至8%、8%、4%和5%的防御上限，待流动性出清与业绩验证后再重拾B…", "summary": "截至2026-06-15，本卡针对BoJ加息及日元套利交易平仓风险，对A股科创50、创业板指、AI硬件及国产半导体设备进行压力测试。建议将科创50及创业板指仓位降至8%，AI硬件降至4%，国产设备降至5%。并明确了日元汇率稳定、A股缩量筑底与中报超预期等重新加仓的量价与盈利修正信号。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX首秀提升了A股航天与硬科技风险偏好，但中国宏观筛选器指向政策背书、订单可见和资产负债表稳健的少数供应链赢家，而不是普涨型半导体beta。", "summary": "截至2026-06-15，本卡综合支持前序“龙头虹吸 + 选择性重估”判断。中国宏观环境能支撑商业航天、卫星互联网和硬科技供应链的结构性行情：一季度GDP同比5.0%，5月出口以美元计同比19.4%，政策明确支持航空航天、卫星互联网和AI基础设施。但PMI新订单49.9、M2-M1增速差3.1个百…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "当前航天行情是有资金和广度验证的结构性龙头行情，不是被两融和收盘表现确认的普涨型半导体beta。", "summary": "截至2026-06-15，本卡支持“龙头虹吸 + 选择性重估”判断。2026-06-12 A股放量普涨，成交额32361亿元、超3900股上涨，但半导体、光刻胶、存储芯片收盘位于跌幅居前方向；两市融资余额截至6月12日反而减少4.02亿元。商业航天内部有强广度：国证航天航空指数盘中涨3.9%，航天…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "航天板块当前表现为针对宽基半导体的“流动性虹吸”与“高低切换”代理，而非带动半导体普涨的Beta引擎。", "summary": "截至2026年6月15日，本卡通过两融与资金流分析对航天带动半导体Beta的假设进行了压力测试。结论显示存在显著的“流动性虹吸”：6月12日A股虽放量成交3.24万亿并突破4000点，但资金呈“高低切换”态势，主力资金从电子（-58.3亿元）和通信（-91.2亿元）净流出，流向有色金属及事件驱动的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 9, "neutral": 1, "value": 23.2, "events": [ { "title": "6月15日成长行情是可交易挤仓而非成长牛市重启，应维持半导体/CPO有上限参与，并把科创50 1550/1516作为低置信但高冲击的触发区间。", "summary": "截至2026-06-16，本卡对6月15日成长爆量轮动给出压力测试结论：反弹可交易，但尚非可持续配置信号。6月A股解禁3332.85亿元、影石创新约403.91亿元；5月25日两融余额曾达29252.09亿元，但截至6月12日已降至28734.82亿元。科创50 6月15日收1748.33点，距1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SPCX期权首日即面临1-DTE/0-DTE短久期希腊字母爆发，在FOMC决议与周四三巫日共振下，做市商Short-Gamma/Vanna对冲踩踏将在低VIX（16.20）环境下非线…", "summary": "基于2026-06-17，分析表明SPCX期权首日上市、FOMC决议和周四三巫日共振，且周五休市，将大幅放大美股跳空及科技股下行风险。在VIX降至16.20的低波动伪装下，SPCX首日交易即面临极端的1-DTE/0-DTE Short-Gamma和Short-Vanna对冲踩踏。若美股科技股深度调…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美股科技股回调引发的A股流动性挤兑风险，其核心痛点在于860亿两融‘预警区’杠杆资金的踩踏，而非雪球产品的局部敲入。", "summary": "基于2026-06-17，量化分析显示A股两融余额处于2.87万亿元的历史高位，且高度集中于电子与通信板块。科创50指数目前（1,758点）距离雪球密集敲入带（1,600-1,700点）仅一步之遥。压力测试表明，若科技股回调20%，约860亿元处于‘预警区’的杠杆资金将面临强制平仓压力，其流动性挤…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI大基建进入“能效优先”阶段，二级市场资金正从粗放型算力硬件（CPO/服务器）向电力基建“瓶颈资产”与半导体先进封装“能效资产”进行结构性跨量级轮动。", "summary": "截至2026年6月17日，AI算力的电力挤出效应与能效转型正引发二级市场深刻的风格轮动。资金正从受资本挤压的传统CPO与服务器组装板块流出，转向两大确定性主线：一是具备强垄断属性与刚性Capex支撑的电力/电网设备（特高压、变压器、液冷）；二是受益于架构优化的先进封装（CoWoS、HBM）产业链。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "截至2026-06-17，A股市场结构不支持确认电子与科创50去拥挤，只支持继续持有但带上限的硬科技风险预算。", "summary": "截至2026-06-17，本卡压力测试第05卡再平衡方案：融资余额、ETF资金流和市场宽度都未确认电子与科创50拥挤度降温。科创板两融余额仍连续增加，科创50和电子ETF仍有资金承接，2026-06-17科创50大涨但超过3700只股票下跌，显示行情更窄而非更健康。结论是维持70/20/7/3组合…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 2, "neutral": 0, "value": 65, "events": [ { "title": "A股电子/TMT融资仓不能等到-18%才降杠杆；价格、ETF赎回、两融拥挤与券商风控线共振时，应提前强制削减融资敞口。", "summary": "截至2026-06-18，本卡片把-18%政策底压力测试转化为执行触发器：科创50、创业板、申万电子中任意2个破位，叠加股票ETF连续大额净流出、电子融资净买入在弱价中继续居首、ETF/科创两融余额反常上升，以及券商115%即时平仓线或双创集中度附加线收紧，应在-18%失败前强制降融资。政策底只有…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体代工与设备估值面临全球WACC上升及N2晶圆价格墙的压缩，叠加国内HBM3国产材料良率瓶颈导致的算力变现延迟，正驱动A股电子资金扎堆“四大龙头”并形成‘高融资、弱宽度’格局，面…", "summary": "截至2026-06-18，本压力测试卡片验证了半导体设备和材料需求的边际变化如何驱动电子板块的宽度衰退与杠杆拥挤。中芯国际、海光信息、寒武纪、北方华创“四大龙头”融资余额达650.0亿元，占板块总两融的17.62%，而板块宽度持续恶化（仅42.1%个股上涨）。压力测试显示-15%的回撤将引发四大龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 4, "neutral": 0, "value": 54.4, "events": [ { "title": "在30% A股配置中，电网与散热设备凭借低融资拥挤和强出海景气度应立即建仓，而国产半导体在两融高负荷、折算率下调以及雪球敲入风险完全出清前应保持现金等待，通过分步建仓捕捉12.48%…", "summary": "截至2026-06-20，针对30% A股配置，本报告建议采取“先物理、后算力”的分步建仓策略。20%电网/散热板块应立即建仓；10%国产半导体板块因融资余额达7382亿、两融折算率下调及1516-1550点雪球敲入威胁，面临严重的左尾去杠杆风险。待半导体释放-15%回撤且杠杆出清后建仓，可使A股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 日报总编： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: cd8f2a28-c7a9-4dc8-aebc-e5ed66354208 结果摘要: Wrote handoff_response.md (17618 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-20，鉴于电网对冲使CPO收入下修收窄至13-17%，A股核心科技龙头基本面安全垫抬升，但由于技术性两融拥挤高企，国家队必须坚守1750战术线及1550生死线以防范…", "summary": "截至2026-06-20，本报告综合评估了CPO收入下修收窄至13-17%下，A股科技板块（中际旭创/寒武纪/源杰科技/联讯仪器）的配置与防御路径。尽管基本面压力因电网加速器对冲而减轻，但高位两融拥挤及成交量萎缩风险未除，国家队应维持1750点战术支撑与1550点绝对生死防线以阻断强平踩踏[自测算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-19，鉴于Q4 SPCX解禁对4%自由流通盘的价格冲击风险，模型组合表达AI基础设施主题应从直接叙事（SPCX和半导体）退步，通过将70%权重配置给电网设备和核电I…", "summary": "截至2026-06-19，本报告对AI基建主题的最佳表达载体进行综合评估。针对2026年Q4的SPCX员工与内部人解禁压力测试，自测算表明SPCX当前4%的公众流通盘易受日元平仓和宏观去估值冲击。组合测试了三类表达权重：策略1（直接叙事）夏普比率为-0.12，策略2（物理瓶颈防守：0% SPCX、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 1, "neutral": 0, "value": 73.8, "events": [ { "title": "旋转之下的A股市场微结构：拥挤度、两融持仓与流通容量能否承接旋入电网/液冷/储能、旋出半导体/服务器而不触发强平传染？", "summary": "压力测试卡片05的26pp旋转。微结构结论非对称：买入侧仅电网是深蓄水池（ETF破百亿、年内净流入238亿，锚定国网十五五4万亿+40%），液冷盘浅且已过热（英维克破千亿后一字跌停），储能ETF仅约3.3亿元、容量微小，硬塞会制造新拥挤。真正引信在卖出侧：3325亿科创融资、98.7分位半导体、1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 1, "neutral": 0, "value": 75.2, "events": [ { "title": "A股AI算力硬件板块呈现显著的估值与筹码分化，主题空气股在3D封装良率低迷（23.46%）下面临溢价泡沫破裂风险，而35%两融强平压测显示次级主线杠杆极度脆弱，支撑以长江电力防御配合…", "summary": "基于2026-06-22，本报告定量评估A股半导体与算力估值及拥挤度。真实敞口股（沪电股份、胜宏科技等）交易在23-32x PE，估值合理；而主题叙事股（华海诚科、载板概念）炒作至80-100x+ PE，存在严重溢价风险。大基金三期通过1640亿子基金注资，压降项目WACC至4.5%，为实体替代提…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 4, "neutral": 0, "value": 58.6, "events": [ { "title": "电子行业6132.72亿元两融杠杆中，357.00亿元处于维保比例临界带，-10%的行业性下跌将触发475.00亿元首轮复合抛压，高融资集中度的半导体龙头面临严重的负Gamma去杠杆…", "summary": "截至2026-06-24，本报告对电子行业6132.72亿元两融余额的维持担保比例分布进行了深入压力测试。当前分布显示，安全区（维保>150%）占比92.11%，而预警与强平区（维保<=150%）合计达484.00亿元。若科创50跌破1500点触发-10%的行业性回调，维保比例低于144.4%的3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "沪深300 AI权重集中度上升后的A股筹码结构与流动性压力测试", "summary": "对卡片05提出的本地拥挤脆弱点做压力测试。结论：脆弱点真实但被错置——危险不在总量杠杆（两融2.8万亿创纪录，但占流通市值2.6%、杠杆率23.72%、维持担保300%，均处周期中段而非2015），而在筹码三重不对称：(1)杠杆热钱集中于主导指数的AI个股(硬件+半导体占融资净买入约60%；寒武纪…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "逢跌买入并非永远错,而是现在危险——散户杠杆与易跳空微结构仍按一个已(有条件)结束的宽松体制校准;确认或证伪体制切换的扳机是霍尔木兹/油价,而非美联储。", "summary": "美联储2026年6月17日点阵图翻转为偏加息(9/19位预测年内加息、年末利率中值升至约3.8%),5月CPI达4.2%创三年新高,主因霍尔木兹能源冲击;美韩科技已gap-down(纳指6/4-5跌4.18%、KOSPI中旬跌约10%)。但A股散户仍逢跌买入——6/18半导体91只涨停、两融近2.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-24，平准基金买盘只有在定向科创50、规模达到150亿-250亿元并触发做市商回补IM/IC空头时，才是Gamma中和器；否则只是指数托举器。", "summary": "截至2026-06-24，本卡压力测试认为：汇金式ETF买盘能中和科创50雪球负Gamma，但只有在定向规模达到约150亿元、且80%转化为当日篮子需求时，才足以覆盖118亿元做市商卖盘中值；250亿元才覆盖高压上沿。若只是50亿-100亿元宽基均衡买盘，更多是抬指数，无法封住IM/IC贴水和单名…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 8, "neutral": 1, "value": 39, "events": [ { "title": "CPO“量产延期”谣言与AI硬件基本面交付的真空博弈", "summary": "SemiAnalysis（约2026-06-09）将CPO大规模商用推后至2028–2029，引发美股光通信单日重挫（AAOI约-17%、COHR约-11%等）。但散户犯了范畴错误：被推后的是CPO，而非“易中天”当前出货的800G/1.6T可插拔模块；CPO延期反而延长可插拔与NPO的需求窗口，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CPO“量产延期”谣言与AI硬件基本面交付的真空博弈", "summary": "CPO“量产延期”谣言与AI硬件基本面交付的真空博弈 CPO 'Production Delay' Rumor vs. AI Hardware Delivery Dynamics 在光模块“易中天”遭遇“量产延期”传言、半导体两融达6132.72亿元且科创50临近敲入点的背景下，下游买方的实际排产…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CPO“量产延期”谣言与AI硬件基本面交付的真空博弈", "summary": "CPO“量产延期”谣言与AI硬件基本面交付的真空博弈 在光模块“易中天”遭遇“量产延期”传言、半导体两融达6132.72亿元且科创50临近敲入点的背景下，下游买方的实际排产数据是否与散户恐慌存在预期差？如何厘清因物料瓶颈带来的边际延迟与系统性订单下修？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论", "summary": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论 JPY Carry Trade Unwind vs. APAC Tech Margin Call Spiral 美元兑日元跌破161.50关口，财务省干预概率陡增。一旦日元套利平仓瞬间引爆全球EM科技股抛售，叠加SpaceX二次配售24.5亿元的资金虹…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 4, "neutral": 1, "value": 64.4, "events": [ { "title": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局", "summary": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局 Midday A-Share Growth Rebound and Options Negative Gamma Clearance Tactical Asset Allocation 在上午科创50跌破关键支撑触发雪球敲入与两融强平后，午…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股电子两融强平螺旋与期权对冲防线的实操效率辩论", "summary": "A股电子两融强平螺旋与期权对冲防线的实操效率辩论 A-Share Electronics Margin Liquidation Spirals and Options Hedging Efficiency 在系统性去杠杆Cascade下，若A股行业面临15%回撤，半导体与科技板块将承受19.89十…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股半导体两融强平底与全球AI硬件景气度背离的投资套利机会", "summary": "A股半导体两融强平底与全球AI硬件景气度背离的投资套利机会 Divergence Between A-Share Semiconductor Margin Liquidations and Global AI Hardware Super-Cycle 在美光爆表财报确立全球AI硬件超级周期的背景下…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在2026-06-26，鉴于电网对冲使CPO收入下修收窄至13-17%，A股核心科技龙头基本面安全垫抬升，但由于技术性两融拥挤高企，国家队必须坚守1750战术线及1550生死线以防范…", "summary": "截至2026-06-26，本报告综合评估了CPO收入下修收窄至13-17%下，A股科技板块（中际旭创/寒武纪/源杰科技/联讯仪器）的配置与防御路径。尽管基本面压力因电网加速器对冲而减轻，但高位两融拥挤及成交量萎缩风险未除，国家队应维持1750点战术支撑与1550点绝对生死防线以阻断强平踩踏[自测算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 5, "neutral": 1, "value": 59.8, "events": [ { "title": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论", "summary": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论 美元兑日元跌破161.50关口，财务省干预概率陡增。一旦日元套利平仓瞬间引爆全球EM科技股抛售，叠加SpaceX二次配售24.5亿元的资金虹吸，A股电子板块6132.72亿元两融及科创50雪球在1500点敲入区附近会否触发自反馈平仓踩踏？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局", "summary": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局 在上午科创50跌破关键支撑触发雪球敲入与两融强平后，午间主力ETF资金流入与个股金针反转是否标志着负Gamma抛售压力的阶段性出清？我们应如何利用这一估值剪刀差进行右侧建仓，从红利低波向高Beta科技股进行再平衡？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-26确认科技两融堰塞湖进入去杠杆阶段，但未确证牛市陷阱；最优策略是降拥挤科技、保现金缓冲、分批增配电力电网等物理红利资产。", "summary": "截至2026-06-27，结论是支持防守再平衡，但不把2026-06-26下跌直接视为“国家队牛市陷阱”的确证。A股两融余额已升至3万亿元以上，电子行业两融余额超过6000亿元且占比约两成，说明风险是融资拥挤和流动性冲击，而非科技周期被证伪。建议科技从超配降至标配，优先削减融资敏感电子/CPO仓位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "TMT与科创板在1600-1650点极限托底下的两融账户维持担保比例敏感性分布与券商风控应对", "summary": "若科创50下探1600-1650极限底，真正的约束并非130%维保线，而是折算率经‘保证金可用余额’施加的更快渠道。全市场CMR约245%（结构性均值278%），中证1000已降至218%；微观分布按券商规模高度倾斜：头部AAA券商270-300%、低于130%尾部约0.3%，中小券商200-22…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 5, "neutral": 0, "value": 58.2, "events": [ { "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 工业制造分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: eb737f41-b97c-43b9-9b89-04a7c7636133 结果摘要: Wrote handoff_response.md (10188 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在三季度科技板块1200.00亿元解禁洪峰与科创50指数1520点平仓线前触发的2528.00亿元两融级联强平夹击下，前道半导体98.7%分位数的高昂估值缺乏安全边际，短期估值弹性无…", "summary": "针对2026-06-27，本报告对科创50两融平仓风险与英维克、思源电气等三季度解禁抛压（科技板块共1200亿元解禁）执行了压力测试。结果表明，科创50估值（PE 207.17倍，100%历史分位）与半导体前道估值（98.7%分位）均极度拥挤。若指数跌至1520点将触发电子/TMT板块2528.0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球商品通胀刚性与风险平价失效下的A股风格轮动：上游资源/红利价值 vs 成长，及两融与三季度解禁的交互", "summary": "全球'上游资源+红利价值'赢家只主导了沪深两市的一半：A股把组合拆开——资源赢（有色年初至今+30%以上、行业冠军[S4]），骑在第06-07卡铜/GOES缺口上；红利价值落后（中证红利一季度约+4%[S5]），因边际资金追逐成长逼空（科创50约+40%、创业板约+30%、6-22创新高[S4]）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 4, "neutral": 0, "value": 65.6, "events": [ { "title": "虽然278%平均维保安全垫使两融连锁抛压呈局部非系统性风险，但20-30%回撤仍将触发最高2528.00亿的平仓压力并重创产业链，应通过减仓高倍数前道、超配低谷先进封装以及引入非担保…", "summary": "截至2026-06-28，两融杠杆在电子和半导体板块高度集中（总余额分别达6132.72亿和560.80亿），前五大龙头占用386.80亿。压力测试显示，20-30%的回撤会迫使大批账户跌破130%平仓线，极端情况下（科创50跌破1520）诱发2528.00亿平仓压力，并因雪球负Gamma在2-5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在紧平衡流动性与外部融资约束下，A股难以支撑全局性风险偏好扩张，下半年配置应降低总Beta，采用“硬科技国产替代（封装/设备）+高独立现金流红利”的结构性杠铃组合。", "summary": "在美联储加息预期和国内紧平衡流动性制约下，A股面临外资流出、两融去杠杆与解禁供给的三重压力。外资流入放缓，险资增配高股息红利防御资产。两融余额高企（3.015万亿元）使科技成长面临局部去杠杆和级联平仓风险。半导体板块估值已普遍收缩30%至45%，大基金三期（3440亿元）提供战略支撑。下半年主导风…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "券商信用风险的局部破线是否会通过流动性传导至整个金融板块，进而引发非对称的流动性危机？", "summary": "本卡对半导体担保品集中度超25%的核心券商做流动性压测，部分支持卡片06并修正其传导结论。半导体两融5,608亿元仅占两市两融余额30,009.71亿元的18.7%，故超25%是“选券商”问题而非系统总量问题。-30%尾部冲击下，多元化头部券商流动性缺口仅占净资本12.0%可轻易覆盖，但科技偏重中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。", "summary": "截至2026-06-29，本卡片支持前序“更高更久”利率压制A股成长久期的判断，但将风险定位为结构性踩踏而非全市场2015式去杠杆。两融余额为30329.92亿元，占A股流通市值2.83%，总量可控；但电子板块融资余额6261.87亿元，且6月融资流入高度集中。模型显示，科创50较2026-06-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 4, "neutral": 0, "value": 67, "events": [ { "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股红利拥挤度与7-11月年内解禁洪峰限制了风格切换空间，而在人行短端流动性上限失灵与CNH被动升值情景下，高负债、高Beta的券商板块将面临最严重的两融净利差挤压与反转冲击。", "summary": "截至2026-06-29，本卡对A股杠铃轮动的拥挤度与压力进行评估。全市场两融达30329.92亿元（TMT占41%），银行与防御红利极其拥挤，仓位达95%分位；6-11月面临2.8-3.2万亿元解禁洪峰，今日单日解禁达589.22亿元，严重压制内需成长。若CNH继续升值或人行回购走廊失灵（DR0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%）；通过配以5.0%权重可转债债…", "summary": "截至2026-06-29，在两融强平与350亿雪球敲入负Gamma踩踏、以及人行短端Corridor失灵（DR007升至2.50%）的极端情景下，杠铃组合资产相关性共振跃迁（红利与成长升至+0.55至+0.70）。组合波动率由9.60%升至19.98%，日99% VaR升至2.93%（穿透2.50…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-bb8f3d8ee7/", "networkHref": "/investor/research-network/#thesis-bb8f3d8ee7" }, { "id": "thesis-62e0155999", "source": "auto_topic", "title": "美元兑日元跌破161.50关口，财务省干预概率陡增。一旦日元套利平仓瞬间引爆全球EM科技股抛售，叠加SpaceX二次配售24.5亿元的资…", "titleEn": "Practical research topic: 美元兑日元跌破161.50关口，财务省干预概率陡增。一旦日元套利平仓瞬间引爆全球EM科技股抛售，叠加SpaceX二次配售24.5亿元的资…", "lane": "宏观利率与汇率", "laneEn": "宏观利率与汇率", "laneHref": "/investor/lanes/lane-cdafce64/", "investableFocus": "", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM", "APO", "588000.SH", "CAT", "COP", "MAR" ], "coreView": "该主线来自 4 条研究记录的标题、问题和摘要聚类，时间跨度 2026-06-09 至 2026-06-26；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-20", "lastSeen": "2026-06-29", "supports": 7, "risks": 133, "neutral": 20, "analystCount": 15, "sourceTypeCounts": { "研究笔记": 127, "研究主题": 14, "研究线程": 14, "分析协作": 5 }, "topTerms": [ "SpaceX", "科创50", "A股", "美元", "两融", "日元", "美元兑日元跌破161.50关口", "财务省干预概率陡增", "抛售", "尾部事件", "正确做法是", "硬件减1/3" ], "whatChanged": [ "2026-06-29: 在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%）；通过配以5.0%权重可转债债底防御和0.50%权重股指虚值Put期权，可将组合极限收益拉升至+104.40%以平准踩踏损…", "2026-06-29: 2026-06-29的核心判断是：人行短久期托底利好金融与内需现金流，不能无差别追成长，出口型TMT硬件是CNH升值情景下最需要降权的A股风格风险。", "2026-06-29: A股市场流动性与资本流出风险评估", "2026-06-29: A股红利拥挤度与7-11月年内解禁洪峰限制了风格切换空间，而在人行短端流动性上限失灵与CNH被动升值情景下，高负债、高Beta的券商板块将面临最严重的两融净利差挤压与反转冲击。", "2026-06-29: 国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多资产组合VaR在1530点处强制破位，急需分层部署的ETF买盘及两融弹性化阀门进行精…" ], "practical": { "sectorLane": "宏观利率与汇率", "sectorLaneEn": "宏观利率与汇率", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM", "APO", "588000.SH", "CAT", "COP", "MAR" ], "primaryStocks": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" } ], "instrumentTags": [ "TLT", "iShares 20+ Year Treasury Bond …", "IEF", "iShares 7-10 Year Treasury Bond…", "UUP", "Invesco DB US Dollar Index Bull…", "FXY", "Invesco CurrencyShares Japanese…", "GLD", "SPDR Gold Shares", "XLE", "Energy Select Sector SPDR" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 119 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 15, "recentEvents": 119, "supportRatio": 0.04, "riskRatio": 0.83, "conflictScore": 93.8, "evidenceDepth": 100 }, "score": 67.8 }, "events": [ { "date": "2026-06-29", "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）将使平仓率非线性飙升至20%，面临940亿STAR50抛压与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "科创50", "A股", "两融", "雪球" ], "href": "" }, { "date": "2026-06-29", "title": "A股红利拥挤度与7-11月年内解禁洪峰限制了风格切换空间，而在人行短端流动性上限失灵与CNH被动升值情景下，高负债、高Beta的券商板块将面临最严重的两融净利差挤压与反转冲击。", "summary": "截至2026-06-29，本卡对A股杠铃轮动的拥挤度与压力进行评估。全市场两融达30329.92亿元（TMT占41%），银行与防御红利极其拥挤，仓位达95%分位；6-11月面临2.8-3.2万亿元解禁洪峰，今日单日解禁达589.22亿元，严重压制内需成长。若CNH继续升值或人行回购走廊失灵（DR007升至2.50%），券商两融利差收窄33.19%，导致两融…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "A股", "两融", "雪球", "券商" ], "href": "" }, { "date": "2026-06-29", "title": "A股市场流动性与资本流出风险评估", "summary": "压力测试卡片02的250–500亿美元流出预测。按USDCNH 7.0折算为人民币1,750–3,500亿元（0.17–0.35万亿），仅占A股流通市值0.17–0.33%、单季成交不足0.4%、外资持仓6.8–13.6%；外资仅占流通盘约2.46%，且北向已降至近三季新低，提前防御。直接冲击可控。真正脆弱点是内生：两融余额于2026-06-23首破3万亿…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "美元", "两融" ], "href": "" }, { "date": "2026-06-29", "title": "2026-06-29的核心判断是：人行短久期托底利好金融与内需现金流，不能无差别追成长，出口型TMT硬件是CNH升值情景下最需要降权的A股风格风险。", "summary": "截至2026-06-29，本卡综合前序CNH挤压与人行受限宽松链条：境内流动性问题更像波动率与久期冲击，而非持续资金断裂。A股应采用金融+精选内需成长的杠铃结构：银行作为资产负债表稳定器超配，券商作为成交和两融高beta适度超配，高股息作为压舱石，内需硬科技/软件精选配置；出口型TMT硬件因CNH升值会冲击人民币收入折算与估值倍数，应低配。若CNH自6.8…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "券商" ], "href": "" }, { "date": "2026-06-29", "title": "在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%）；通过配以5.0%权重可转债债底防御和…", "summary": "截至2026-06-29，在两融强平与350亿雪球敲入负Gamma踩踏、以及人行短端Corridor失灵（DR007升至2.50%）的极端情景下，杠铃组合资产相关性共振跃迁（红利与成长升至+0.55至+0.70）。组合波动率由9.60%升至19.98%，日99% VaR升至2.93%（穿透2.50%上限），10日99% VaR升至9.27%（穿透7.50%…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "雪球" ], "href": "" }, { "date": "2026-06-28", "title": "虽然278%平均维保安全垫使两融连锁抛压呈局部非系统性风险，但20-30%回撤仍将触发最高2528.00亿的平仓压力并重创产业链，应通过减仓高倍数前道、超配低谷先进封装以及引入非担保品指数对…", "summary": "截至2026-06-28，两融杠杆在电子和半导体板块高度集中（总余额分别达6132.72亿和560.80亿），前五大龙头占用386.80亿。压力测试显示，20-30%的回撤会迫使大批账户跌破130%平仓线，极端情况下（科创50跌破1520）诱发2528.00亿平仓压力，并因雪球负Gamma在2-5天内引发600-750亿级联抛售。虽然278%平均担保垫等四…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.5, "matchedTerms": [ "科创50", "两融", "抛售", "雪球" ], "href": "" }, { "date": "2026-06-28", "title": "券商信用风险的局部破线是否会通过流动性传导至整个金融板块，进而引发非对称的流动性危机？", "summary": "本卡对半导体担保品集中度超25%的核心券商做流动性压测，部分支持卡片06并修正其传导结论。半导体两融5,608亿元仅占两市两融余额30,009.71亿元的18.7%，故超25%是“选券商”问题而非系统总量问题。-30%尾部冲击下，多元化头部券商流动性缺口仅占净资本12.0%可轻易覆盖，但科技偏重中型与薄资本小券商分别达32.7%与49.6%，确为局部破线区…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "两融", "雪球", "券商" ], "href": "" }, { "date": "2026-06-28", "title": "已确认 OMO 窗口下的跨资产布局：国债久期、人民币套息/对冲与利率敏感 A 股倾斜", "summary": "首席策略师将 Research note 01–04 综合为一本可交易的跨资产组合，应对强美元（DXY 101.50、美债2Y 4.1%/10Y 4.37%）、坚挺人民币（即期约6.80、5月结售汇顺差358亿美元）与已确认有条件10bp OMO窗口（2.5/3绿灯）三态共存。建议：国债40%（修正久期约6.5、票息主导、半年预期约1.5%）；A股35%（…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 20.5, "matchedTerms": [ "A股", "美元", "券商" ], "href": "" }, { "date": "2026-06-28", "title": "A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。", "summary": "截至2026-06-29，本卡片支持前序“更高更久”利率压制A股成长久期的判断，但将风险定位为结构性踩踏而非全市场2015式去杠杆。两融余额为30329.92亿元，占A股流通市值2.83%，总量可控；但电子板块融资余额6261.87亿元，且6月融资流入高度集中。模型显示，科创50较2026-06-26收盘下跌15%会触发约1063亿元有效被动卖压，20%跌…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "A股", "两融" ], "href": "" }, { "date": "2026-06-28", "title": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼及普钢加工环节将面临严重的利润率挤压与低…", "summary": "本卡通过技术壁垒与自给率对A股AI原材料敞口进行了区分：高端取向硅钢方面，宝钢股份凭借规模与技术优势展现出强Alpha属性，而首钢股份受普钢周期Beta拖累较大；有色铜方面，在2026年长单TC/RC为零的极端低位下，矿权自给率超65%的紫金矿业可完全捕获矿端溢价（EBITDA价格敏感度约7.5%），而低自给率的江西铜业、云南铜业面临严重的加工毛利压缩。建…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "日元" ], "href": "" }, { "date": "2026-06-27", "title": "在 USD/JPY 跌破150的情景中，在岸高两融TMT的主要脆弱性是融资余额绝对规模和券商风控顺周期收紧，而非接近交易所25%单股融资上限。", "summary": "2026-06-27：本卡压力测试卡片06。日本4月持有1.2099万亿美元美债，USD/JPY跌破150的流动性通道方向成立，但2150亿-2850亿美元清算规模未被独立证实。A股两融余额为30329.9亿元，核心TMT行业约4670.0亿元、占15.4%。TMT风险不在交易所25%融资暂停线，而在高绝对余额与券商风控收紧：中际旭创、寒武纪、兆易创新、澜…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "A股", "美元", "两融", "券商" ], "href": "" }, { "date": "2026-06-27", "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 工业制造分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: eb737f41-b97c-43b9-9b89-04a7c7636133 结果摘要: Wrote handoff_response.md (10188 bytes); wrote handoff_response.en.md (12356 bytes) — 三日近80…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 24, "matchedTerms": [ "科创50", "A股", "两融", "雪球" ], "href": "" }, { "date": "2026-06-27", "title": "日元套利交易平仓与离岸CNH流动性收紧对A股防御重估的交叉传导风险", "summary": "本卡以global-macro压力测试card-08的“USD/JPY与CNH流动性熔断器”构想：支持其核心方向（知识库已确认应以二者作量化触发器），但否定朴素的“单变量、按价位、OR门”实现。四点修正：(1)冲击分两波——先美元流动性挤兑、后避险重估，价位触发会迫使在第一波真空里抛售防御股；(2)两根保险丝相关而非独立，应合并为复合压力指标CTSI并设并…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22.5, "matchedTerms": [ "A股", "美元", "日元", "抛售", "真空" ], "href": "" }, { "date": "2026-06-27", "title": "美债高实际利率和中美利差倒挂通过锁定金融套息成本、驱动企业实体套息分化，对离岸人民币和新兴市场美元债产生持久的再融资压力，限制了国内货币政策空间并固化了A股滞胀杠铃配置格局。", "summary": "截至2026-06-27，美联储新主席沃什的鹰派政策维持美债名义利率与实际利率于高位。10年TIPS实际利率处2.19%高位逼近2.3%关键分水岭，引发全球跨资产去杠杆。中美10年期国债-265bp的倒挂使金融套息交易被远期贴水和HIBOR挤空风险压制，即期USD/CNY稳定在6.79。但出海企业通过跨境人民币贷款置换高息美元债开展实体套息（获取1.5%-…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.5, "matchedTerms": [ "A股", "美元", "日元", "抛售" ], "href": "" }, { "date": "2026-06-27", "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌破3.0%的红利反向出清防线。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "A股", "两融" ], "href": "" }, { "date": "2026-06-27", "title": "实施两期隔离（账户完全隔离与无融资杠杆）是防范两融折算率大降（0.6至0.4）引发6.0pp保证金缺口、确保商品期权3.0pp封顶与6pp黄金仓位安全的底线红线。", "summary": "截至2026-06-27，本卡对A股两融踩踏与商品期权保证金交叉传染进行了压力测试。核心结论是：若两融踩踏引发全市场折算率均值由0.6降至0.4，统一保证金账户将面临6.0pp的流动性缺口并迫使权益防御仓位平仓；而实施完全账户隔离、A股无杠杆（融资余额为0）及商品期权 butterfly 结构，可确保3.0pp保证金封顶不受侵害。下一步建议移交fx-str…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "日元" ], "href": "" } ], "timeline": [ { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "在轨智能（I2O）正将卫星从被动数据载体重塑为主动决策平台，导致利润池从基础带宽向AI算法与垂直整合平台迁移。", "summary": "2026年5月，在轨AI算力（I2O）已成为卫星应用变现的核心变量。NVIDIA的Vera Rubin Space-1模块与SpaceXAI的深度整合，以及中国国星宇航、航天宏图等企业的规模化组网，标志着卫星从“数据管道”向“自主节点”的全面进化。通过在轨处理减少95%以上的数据下行压力，遥感业务…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 3, "neutral": 3, "value": 34.4, "events": [ { "title": "巨型AI IPO是风险偏好的质量筛选器，而不是对A股AI硬件的直接抽水器。", "summary": "截至2026-05-21，巨型AI IPO供给会扰动美国AI持仓，但对A股更可能先放大而非抽走风险偏好。Nvidia FY2027一季度收入816.15亿美元、二季度指引910亿美元验证AI资本开支；A股成交和两融显示本土资金仍强。核心风险是SpaceX/OpenAI等供给迫使高拥挤概念股被重新定…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创板极端护盘对A股整体流动性的虹吸效应与跨板块资产配置调整 研究日期： 2026 05 22 作者： 首席策略师 (Chief Strategist) 立场： Synthesize…", "summary": "科创板极端护盘对A股整体流动性的虹吸效应与跨板块资产配置调整 研究日期： 2026 05 22 作者： 首席策略师 (Chief Strategist) 立场： Synthesize (综合) 执行摘要 截至 2026 05 22 [S7]，科创50指数收报 1,764.17点 [S1]，极度逼近…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球宏观压力测试：科创板极端护盘外溢效应与跨境流动性虹吸下A股资产配置的脆弱性 研究日期： 2026 05 22 作者： 全球宏观分析师 (Global Macro Analyst)…", "summary": "全球宏观压力测试：科创板极端护盘外溢效应与跨境流动性虹吸下A股资产配置的脆弱性 研究日期： 2026 05 22 作者： 全球宏观分析师 (Global Macro Analyst) 立场： Stress test (压力测试) 执行摘要 截至 2026 05 22 [S1]，科创50指数收报 1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "英伟达光环可能带来最后反抽，但在高杠杆、高换手、券商风控收紧与美债收益率上行共振下，科创50的风险收益已转向防踩踏和高低切换。", "summary": "截至2026-05-22，本卡对“英伟达财报带动科创50最后一冲”采取压力测试立场。英伟达2027财年一季度业绩强劲，但A股在2026-05-21已出现科创50盘中冲高后收跌、成交显著放大的末端挤压特征。两融资金集中流入电子和半导体，科创板两融余额继续上行，券商新增115%平仓线压缩补仓时间，美债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "在科创板两融信用收缩与Q3迎来1200亿解禁洪峰的微观去杠杆环境中，大健康精选篮子建仓必须实施『恐慌底分批-解禁潮吃货』的节奏化策略，并将有效暴露压缩至18%-20%以内，辅以Bet…", "summary": "截至2026-05-25，本卡对创新药出海、CXO地理标签及高端器械替代的精选篮子（联影、迈瑞、奕瑞、百济等）进行A/H股映射与估值管理。A股两融余额见顶于2.87万亿元，科创50在1,750点承受去杠杆与432亿雪球敲入压力。策略建议在科创解禁潮与两融风暴中分三阶段节奏建仓（1-3月恐慌期买30…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 5, "neutral": 0, "value": 16.2, "events": [ { "title": "布油跌至$96.71的地缘纸面缓和无法撼动美联储沃什时代收紧期限溢价的鹰派核心底色，跨资产‘高偏斜’结构正面临大宗红利挤兑、科创信用去杠杆和日元套利平仓的非线性崩塌风险。", "summary": "本报告揭示霍尔木兹海峡地缘缓和传闻导致油价暴跌14.36%至$96.71，但由于物理复航滞后及极度去库，这一纸面利好完全无法抵消新任美联储主席Warsh收紧资产负债表、驱动10Y名义美债至4.59%及实际利率至2.20%的鹰派政策影响。‘高价高偏斜’特征暗示全球大类资产结构极为脆弱，A股科创50在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘溢价出清与美元紧缩周期共振，正迫使A股在2026-05-26节点进行暴烈的跨风格再平衡：资源红利多头平仓与科创板杠杆负反馈将主导近期交易，资金将向隔绝内需疲弱与关税冲击的“海外闭…", "summary": "在2026-05-26，油价暴跌至96.71美元与美债实际利率居高（2.20%）的双重冲击下，A股资源红利资产面临严重的多头踩踏风险，拥挤度已突破95%分位数。同时，科创50在1,750点上方积累了庞大两融杠杆，面临强平去化压力。策略上建议规避大宗红利与高杠杆科技，防御重心转向低拥挤的“海外闭环A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球risk-on可以把科创50推向1900点，但1900点附近的未证实衍生品凸性决定第一波上涨是派发还是吸筹。", "summary": "2026-05-26的核心判断是：海外油价跌破100美元、美元指数约99带来的risk-on确实利好A股成长，但科创50前收1896.04点，距1900点仅0.21%，若该点位存在雪球/Delta障碍，开盘更像流动性测试而非无条件突破。1900点暴露无法公开验证，报告将其列为交易台前提。建议保留科…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘滞胀退潮与结构性Delta cliff共振下的A股开盘博弈", "summary": "地缘滞胀退潮与结构性Delta cliff共振下的A股开盘博弈 Dismantling of Stagflation & Structural Delta Cliff: A-Share Opening Battle 在海外油价暴跌、美元破100引发全球risk-on的背景下，国内科创50指数今日开…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 4, "neutral": 1, "value": 26.6, "events": [ { "title": "A股当前处于“空心科技虹吸”导致的流动性黑洞期，应警惕科技股杠杆踩踏，向“物理红利”（大水电、核电）防御端极化靠拢。", "summary": "2026年5月27日，A股呈现极端的“空心科技虹吸”特征，半日成交2.07万亿但超4600股下跌，资金极度抱团CPO板块。受美伊停火导致油价暴跌15%及日元159引发的去杠杆冲击影响，全市场两融及雪球脆弱性急剧上升。建议警惕高拥挤度科技股的平仓风险，将防御端转向由大水电、核电组成的“WLEF物理红…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2.9万亿两融与73%半导体拥挤的当前杠杆结构下，AI硬件订单断崖会通过A股被动平仓、美股margin call、期权gamma翻负与日元carry拆解四重放大器形成跨资产踩踏，应…", "summary": "在前五卡确认AI硬件订单断崖几乎不可避免后，本卡对当前2.9万亿元A股两融余额与全球73%半导体拥挤仓位进行三档压力测试。基准/严重/极端情景下，SOX一周回撤−12%/−18%/−25%，A股AI硬件链回撤−22%/−28%/−35%。严重情景下，A股AI/算力两融担保比例从250%跌至152–…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构", "summary": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构 USD/JPY 145 Shock and Margin Deleveraging Reshaping the A-Share True Barbell USD/JPY下探145引发中资境外AT1永续债置换成本飙升，直接压降上市银…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构", "summary": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构 USD/JPY下探145引发中资境外AT1永续债置换成本飙升，直接压降上市银行NIM约3-5bp，动摇了‘金融红利’板块绝对防线；而科创50两融拥挤度高达4.9%，折算率下调引致250亿-450亿平仓压力，并威胁到敲入线挂接的27…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 3, "neutral": 0, "value": 31, "events": [ { "title": "MSCI调仓本身不是系统性事件，但落在A股两融2.87万亿/物理上限2.90万亿仅1%缓冲的高敏感点上，使得TMT龙头任何-2%以上的跳水都会被'雪球敲入×维保线×流动性真空'三重叠…", "summary": "本卡对前两卡'被动买盘被主动MOC冲抵、TMT次日温和回吐-0.5%~-1.5%'的结论进行压力测试。结论：基线情景(P≈70%)下两融余额单日收缩80-140亿元，2.87万亿/2.90万亿物理上限的300亿元缓冲仍可承受，系统性风险可控；但厚尾情景(P≈20%)下科创50若跌破1,750点雪球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50指数在1,750点附近存在极高的信用与衍生品二次平仓反馈共振深渊，下周两融去杠杆风暴极易打压AI硬件龙头股估值，建议维持45%的红利防御性配置上限并买入虚值认沽期权对冲尾部风…", "summary": "本报告定量评估了科创50指数在1750点附近的雪球敲入（22.0亿元名义本金）与做市商Delta动态对冲（12.0亿元卖压）的二次共振效应。在券商两融折算率收紧（0.60至0.40）压力测试下，一阶总抛压达37.0亿至57.0亿元，极端踩踏下可触发二阶追加平仓反馈卖压119.7亿元，共振总抛压达1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在雪球敲入与两融平仓共振的极端风险下，以‘先进封装’结构性机会对冲Beta下行，并以虚值期权锁定尾部风险是当前最优防御路径。", "summary": "截至2026-05-29，A股两融余额超2.9万亿且科技板块极度拥挤。科创50指数逼近1,750点雪球密集敲入区，做市商面临121.5亿元刚性抛压，叠加ETF折价套利，流动性死亡螺旋风险剧增。建议配置14%比例的先进封装‘防守方案四’（5%设备+6%材料+3%存储）并配合虚值认沽期权构建多重防线。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 3, "neutral": 1, "value": 35.4, "events": [ { "title": "A股两融安全边际充沛，科创50雪球实际规模仅为62亿元且敲入卖压平滑，MSCI被动重估不存在所谓的“流动性真空砸盘”，跨境日元期权对冲完全多余且损耗超额收益。", "summary": "本卡作为金融行业分析师，强烈反驳A股策略师的悲观观点。首先，两融高位有2026年保证金100%新规与2025年风控标准严密保护，平均维持担保比达248.5%，个股需跌60%以上才触及强平线，绝非指数大跌16.92%所能引爆。其次，科创50雪球名义存续实际仅为62亿元且极均匀分布，做市商采用平滑期权…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈", "summary": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 2.94tn RMB Margin Debt & STAR 50 Snowball Cliff under MSCI Passive Rebalancing 今日是MSCI半年度重估正式生效日。虽然散户由于昨日创业板+1…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 报告日期 ：2026年05月30日 分析师 ：A股策略师 ( ashare strategist ) 密…", "summary": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 报告日期 ：2026年05月30日 分析师 ：A股策略师 ( ashare strategist ) 密级 ：内部机密 | 白板研讨卡 01/08 核心结论与概要 (Executive Summary) 今日 (2026…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "红利资产在极端资本流出下的价格下跌，将通过担保品折损与股指贴水加深，触发跨风格流动性虹吸，导致高杠杆、低流动的医药科技成长股（如佰仁医疗）面临无量踩踏与券商信用坏账风险。", "summary": "本报告对中美利差超-275 bps及人民币贬值至7.40+压力下A股的微观交易脆弱性进行了压力测试。测试表明，尽管A股整体两融余额处于2.7%的健康水平，但红利抱团导致场内资金以红利资产为核心担保品。红利价格下跌20%-30%将导致两融账户维保比跌穿强平线，迫使投资者抛售科技成长股以追保，引发跨风…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 1, "neutral": 1, "value": 48.8, "events": [ { "title": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈", "summary": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 今日是MSCI半年度重估正式生效日。虽然散户由于昨日创业板+1.96%的V型反弹而呈现极度FOMO，但A股两融余额正高挂于2.9367万亿元历史极值，融资交易占比达10.09%。科创50指数（1,844.25点）距离下方…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "太空+半导体材料+AI硬件三主题的A股板块配置框架与主题—成长—周期切换信号 研究院权威工作日期 ：2026 05 30 分析师 ：首席策略师 ( chief strategist …", "summary": "太空+半导体材料+AI硬件三主题的A股板块配置框架与主题—成长—周期切换信号 研究院权威工作日期 ：2026 05 30 分析师 ：首席策略师 ( chief strategist ) 立场 ：合成 (Synthesize) 1. 执行摘要 (Executive Summary) 截至 2026 …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 3, "neutral": 1, "value": 38.2, "events": [ { "title": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论", "summary": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论 A-Share Post-Close Sentiment Fission: MSCI Rebalancing vs. Margin & Snowball Risks 今日A股演练了11:27分V型奇袭与科技大挤空，但成交…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论", "summary": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论 今日A股演练了11:27分V型奇袭与科技大挤空，但成交量缩量2704亿元。盘后社媒上散户情绪在英伟达/Marvell史诗财报刺激下极度亢奋，期待明日MSCI半年度生效日的被动天量申购。然而，两融高挂于2.9367万亿元，…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "压力测试表明，日银鹰派突袭加息将引发USD/JPY无序收敛至138-142并推高10Y美债期限溢价+35至+55 bps；建议通过买入不对称日元看涨期权价差、提高Tier-1现金储备…", "summary": "本卡片压力测试了日银鹰派突袭对全球日元套息空头的无序平仓冲击（45%概率）。日银超预期加息至1.25%将窄化短端利差，引发投机盘踩踏空头平仓，导致USD/JPY暴跌至138-142，并迫使日资抛售2150亿-2850亿美元美债以回流资本，推高10Y美债收益率+35至+55 bps。同时评估了人行逆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守", "summary": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守 Defending the 1,750 Lifeline: Snowball Gamma Waves and Margin Call Risk 科创50收于1,751.32点，距离1,750雪球/两融生命线仅一步之遥。若下周破位跌破1…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 11, "neutral": 2, "value": 12, "events": [ { "title": "概述 作为A股策略师（ ashare strategist ），针对看板第三步主题 “联合宏观矩阵下A股信用与衍生品结构性杠杆风险特征” ，我已对科创50指数回测1750生死线下的两…", "summary": "概述 作为A股策略师（ ashare strategist ），针对看板第三步主题 “联合宏观矩阵下A股信用与衍生品结构性杠杆风险特征” ，我已对科创50指数回测1750生死线下的两融去杠杆与雪球衍生品敲入进行了深入的定量压力测试与模���仿真。 在此项研究中，我全面 支持 首席经济学家（卡片 0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈", "summary": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈 MSCI Rebalancing Effective Date: Margin Debt Record-High & Snowball Knock-in Cliff under Asymmetric Play 今日（5月29日）MS…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈", "summary": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈 今日（5月29日）MSCI半年度重估正式生效。散户情绪因昨日 11:27 V型大反弹处于狂热 FOMO 状态，期待天量被动申购。然而，A股融资余额已攀升至 2.9367 万亿元的历史尖峰，科创50距离 270 亿元名义本金的雪球密集…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "金融行业能承受本轮雪球敲入的下方尾部，但真正需要监管现在穿透的是头部券商场外雪球做市账本，而不是融资融券客户或保险公司抵押品。", "summary": "截至2026-06-02，本卡压力测试认为，雪球敲入引发的对冲抛售不会经由两融账户或保险资管抵押品演化为系统性事件：行业净资本RMB 2.44tn与客户保证金RMB 3.24tn分别为Research note 02最坏情景RMB 27.0bn流量的约90倍与120倍，保险资金运用余额RMB 37…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "在美联储加息与人行降准的联合概率矩阵下，A股必须从传统哑铃策略转向“防御底仓+对冲Beta”模式，规避高融资与高解禁共振的科创标的。", "summary": "由于卡片04数据受损，本报告直接综合卡片01与03。针对科创50指数在1750点的雪球敲入与2.9万亿两融平仓共振风险，本报告提出A股风格轮动与行业配置优化方案：将原10%的科技敞口转为“对冲型科技”，并将高股息与出口制造权重分别提升至35%与25%，以应对6月3237.5亿元解禁悬崖。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 1, "risks": 1, "neutral": 0, "value": 59.4, "events": [ { "title": "USD/JPY与USD/CNY不是直接拉爆A股硬件链融资盘的拉绳，而是通过SOX估值代理与人民币波动率→券商折算率传导的放大器，CNH/JPY交叉汇率是最被忽视的24小时前导信号。", "summary": "外汇视角支持Research note 01诊断：BOJ 6月17日议息加至1.00%已被OIS定价，未平仓日元Carry约5000亿美元（日资Q1净抛美债300亿美元已先行验证）。A股硬件链对JPY直接敞口<5%且传导6-12个月，6月议息当季不入P&L。真正传导通过四通道：①直接外汇敞口（弱）…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "A股硬件链当前的边际定价者不是外资，而是创历史新高的2.9万亿融资盘——美日货币政策共振将通过情绪传染触发'高Beta×高融资'自反馈式回撤，防御核心在于监测内部杠杆而非外部抛压。", "summary": "在2026-06-03时点，中美10Y利差倒挂约-250至-270bp（中国10Y 1.74%），PBoC中间价稳态但即期承压，BOJ预计6月加息至1.00%引发约5000亿美元日元Carry平仓预期。然而外资在A股硬件链经历2022-2023去外资化第一阶段后，弹药已大部分释放，并非边际定价者；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 0, "value": 46.8, "events": [ { "title": "博通指引转保守与A股流动性三压共振，科创50面临1720点雪球破位风险，需警惕系统性去杠杆引发的二次筑底。", "summary": "博通（AVGO）隔夜暴跌13-15%打破了AI硬件订单无限增长的叙事，引发全球半导体去Beta。A股科创板正面临4.9%高两融占比、6月3332.85亿解禁洪峰以及科创50指数（现报1726）逼近1720雪球密集敲入区的“三体问题”。若破位将引发做市商负Gamma抛售。建议维持45%红利底仓，并买…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？", "summary": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？ Broadcom AI Guide-Down Triggers Global Semi Sell-Off: Will A-Share Tech Break Support Amid Leverage & Lo…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 2, "neutral": 0, "value": 48.2, "events": [ { "title": "SpaceX IPO窗口期A股资金已实测从泛AI软件切向商业航天与AI防务硬件，杠杆资金硬>软结构清晰，6/12前β可继续，6/12后须按拥挤度热力图主动管理回撤。", "summary": "截至2026-06-06，SpaceX将于6/12以1.75万亿美元估值IPO。本卡基于SSE/SZSE微观数据实证支持card-01/02的估值重锚结论：近10个交易日北向资金净流出泛AI软件约-78亿元，净流入商业航天+国防军工约+76亿元；国防板块两融余额周环比+5.4%，融资买入额占成交比…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？", "summary": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？ 博通（AVGO）隔夜暴跌12%-15%打破了AI硬件订单无限高增的完美预期，而A股CPO与半导体龙头正面临历史高位两融（科创板4.9%占比）和6月3332.85亿的解禁洪峰。如果科创50指数被动向1750/1…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 5, "neutral": 0, "value": 31.6, "events": [ { "title": "外部高利率与全球去杠杆冲击下，A股面临硬科技两融去杠杆与解禁潮共振风险，资产配置应剪枝高位拥挤科技与昂贵资源，退守低估值高股息周期价值与短端利率债以优化组合VaR。", "summary": "本报告分析了美联储维持“更高更久”利率对A股资金与风格轮动的影响。美债收益率升至4.75%和日元平仓去杠杆加剧了北向资金流出压力（预计600亿至900亿元），推高离岸HIBOR至8.0%-10.0%。两融杠杆高企（2.91万亿元）且高估值科技拥挤，科创50击穿1750点流动性红线触发150亿雪球敲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在融资杠杆处于极值且科创50下面临雪球敲入踩踏的情景下，超配低贝塔、高分红的避风港资产长江电力是防守去杠杆和风格转换中对冲系统性风险的最优策略。", "summary": "截至2026年6月4日，A股融资余额达2.90万亿元高位，科创50自高点回调至1668.33点，逼近1125-1360点的雪球敲入密集区。一旦触发敲入，券商平仓多头将引发市场二次下跌并导致融资盘连环爆仓。建议超配长江电力（Beta为0.09，股息率3.7%，承诺70%以上分红率）等低贝塔红利资产，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在系统性流动性危引发避风港资产补跌与相关性收敛的背景下，跨资产配置至实物黄金与短债，并配合科创50 ETF Put与IM期货空头对冲，是防御VaR穿透并平衡权益融涨弹性的最优解决方案。", "summary": "针对2026年6月7日A股2.90万亿两融余额极值及雪球敲入踩踏引致的系统性流动性危机，本压力测试表明长江电力等避风港资产在挤兑期存在补跌风险，相关性收敛导致未对冲组合10日损失达-16.60%，99% VaR穿透至32.33%[自测算]。本方案设计了科创50 ETF期权与IM期货空头对冲包，并对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元平仓去杠杆下USD/CNH 7.25防线的挤压成本与政策边界", "summary": "日元平仓去杠杆下USD/CNH 7.25防线的挤压成本与政策边界 日本央行加息预期引发约5000亿美元日元套利平仓，外资流出压力加剧（北向流出预期600-900亿）。人行通过将隔夜CNH HIBOR推升至8%-10%及中间价锁死在7.1250来防守7.2500。在点心债冰冻、实体企业外汇套保成本飙…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 1, "value": 60, "events": [ { "title": "[auto-handoff] 科创50指数雪球敲入负 Gamma 对冲抛压与两融折算率下调去杠杆踩踏模拟", "summary": "致 首席风控官： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 73279e4e-699e-4d9e-9366-ea168709088a 结果摘要: Wrote handoff_response.md (14450 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "两融、ETF与行业成交结构已部分确认A股政策顺周期切换，但宽度不足，仍属头部权重驱动的早周期信号。", "summary": "截至2026-06-08，两融、ETF和行业成交三类微观证据共同支持卡片01、02的政策顺周期切换判断：杠杆资金在非银金融、银行、电力设备、军工、机械的融资买入占比抬升，在食品饮料、医药、地产下移；ETF净申购集中在券商、银行、军工、高端装备和央企红利/中特估，纯地产与白酒ETF净赎回；金融与先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 6, "neutral": 0, "value": 28.4, "events": [ { "title": "[auto-handoff] A股科创50跌破零Gamma生死线后两融强平与雪球级联踩踏的微观杠杆压测", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 23d2a86c-8b70-46ce-8620-e44b5d309d0d 结果摘要: Wrote handoff_response.md (3101 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO已引发实质性流动性虹吸，美股代理股的无差别抛售预示着A股商业航天主题即将迎来估值拐点。", "summary": "SpaceX高达750亿美元的IPO正在从板块风险偏好的催化剂转变为巨大的流动性抽水机。美股核心代理股（如RKLB、ASTS）已出现9%-13%的显著回撤，表明市场资金正为了直接认购SpaceX而抛售替代品。对于A股而言，这意味着商业航天主题的“光环效应”见顶，高估值的卫星互联网与火箭产业链概念股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX上市对A股CPO三巨头是资金流冲击而非基本面冲击，抛压持续度以日计而非以季度计，但因CPO估值已处历史90百分位，下行风险非对称。", "summary": "SpaceX 6月12日以1.77万亿美元估值上市，承接它的150–300亿美元 long-only 需求是从全球TMT硬件配置桶中切出的再权重，不是新增流动性。落到A股CPO三巨头（中际旭创、新易盛、光迅）的实际抛压估算为350–550亿人民币，议程的52.5亿元处于可信区间上沿。由于外资持仓集…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO资金虹吸与A股光模块估值压缩传导", "summary": "SpaceX IPO资金虹吸与A股光模块估值压缩传导 SpaceX IPO Capital Siphon & A-Share CPO Valuation De-rating SpaceX将于6月12日以1.77T美元估值上市，吸引全球150-300亿美元TMT硬件多头资金虹吸。A股CPO三巨头面临…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 8, "neutral": 1, "value": 28.8, "events": [ { "title": "SpaceX IPO引发的全球流动性‘真空’是导致科创50破位1600点的核心外因，破位后将诱发两融盘非线性踩踏。", "summary": "SpaceX IPO的450-750亿美元流动性虹吸正通过中概股套利链条向A股传导。压力测试显示，若科创50跌破1600点关键支撑，将触发券商折算率下调与融资盘抢跑的共振。虽然当前距离1310点强平线仍有空间，但技术破位后的非线性踩踏风险从5%上升至20%以上。建议利用看跌价差对冲，并保持15%现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "周末两融非线性踩踏是低概率但非可忽略的尾部事件，正确做法是'硬件减1/3 + 软件保留 + 看跌价差对冲'而非全面去风险。", "summary": "SpaceX 6/12 IPO 约450亿美元的资金需求主要在美股科技仓位内部腾挪，对A股是二阶传导。A股TMT融资盘7382亿当前处于'浮亏未爆仓'状态：科创50收1663点距加权建仓成本约-6.6%，距130%维持担保线对应的强平区间(~1500点)还有8–12%空间。今日是'逼近杠杆密集成本…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO 构成的全球流动性‘真空’通过锁定美元走强预期，形成国内货币政策的‘外汇红线’约束，并诱发科创板块的信用评价下行压力。", "summary": "SpaceX IPO 750亿美元融资引发的全球流动性虹吸正通过外汇与信用渠道双重挤压国内宏观环境。美元走强使人民币汇率锚定在6.81附近，限制了央行在信贷疲软背景下进一步降息的空间。由于国内资本被禁止参与IPO，资金呈现单向流出，导致A股科创板估值压缩并可能引发隐含信用紧缩。这种‘外部虹吸、内部…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO 被动虹吸与A股科技高两融杠杆的非线性踩踏风险辩论", "summary": "SpaceX IPO 被动虹吸与A股科技高两融杠杆的非线性踩踏风险辩论 SpaceX IPO Passive Outflow vs. Onshore Tech Margin Leverage: Stampede Risk Debate SpaceX 6月12日IPO面临450亿资金被动虹吸，而A股…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 2, "risks": 4, "neutral": 1, "value": 62.2, "events": [ { "title": "100亿元缓冲资金的最优A股结构通道是70%ETF一级申购 + 20%收盘竞价权重托底 + 10%大宗折价承接，且必须经由非两融账户执行才能真正切断二次去杠杆反馈环。", "summary": "在ETF折价、成分股大宗、收盘标记三线同压的情景下，100亿元AI/半导体缓冲资金最稳妥的执行组合为：ETF一级申购70亿（消化折价）+ 收盘集合竞价权重托底20亿（守雪球敲入与两融维持担保线）+ 大宗折价−2%承接10亿（吸大单不留痕）；二级连续竞价与转融通回收两条通道必须拒用。所有承接必须通过…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "把拥挤+杠杆+解禁的A股AI硬件龙头当作有效风险2–3倍于名义权重的凸性敞口块——按风险预算定档（6–8%降至3–5% NAV），并用国债久期+黄金+红利低波杠铃分散、用科创50/I…", "summary": "锚定2026-06-10，本卡承接第05卡，把方向判断转为组合构建。结论：这批AI硬件龙头不应当方向性持仓，而是凸性、有硬上限、肥尾的敞口块——因拥挤beta、两融顺周期、7月解禁三重放大，其有效风险贡献约为名义权重2–3倍[自测算]。故应按风险预算定档（解禁前上限6–8% NAV，7/20–24…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO正在触发AI权重股的可持续去拥挤，但交易方向应是低配拥挤AI硬件、增配上游稀缺投入品和低资本开支AI应用，而不是卖出全部AI。", "summary": "截至2026-06-10，本卡片判断SpaceX IPO不是单日情绪噪音，而是触发AI龙头去拥挤的真实催化：SpaceX预计净募资744亿至857亿美元，募资用途包括AI算力基础设施；Google合同强化AI资本开支叙事；TSLA-SpaceX并购期权抬升Musk平台估值分歧；美股半导体与A股CP…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "工作区确含关键保证金/解禁数据且经外部验证支持 card-01；唯一修正是 06-12 的主导被动催化剂为科创50指数调样而非 SpaceX，这反而使'06-12 收盘前减科技硬件'…", "summary": "目录列举确认 card-01 的三份文件确实承载关键 A 股保证金与解禁/被动调仓数据（TMT 两融 7382 亿、06-08 单日净偿还 237.9 亿、长江电力两融占流通 1.81%、06-12 被动流出 450 亿）。外部数据交叉验证支持 card-01 论点：全市场两融余额 2.51–2.…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-11", "supports": 2, "risks": 3, "neutral": 1, "value": 69.6, "events": [ { "title": "太空经济最具持续性的实体溢价不在A股火箭概念股，而在为发射节奏定价与融资的金融中介——空间保险承保池与券商手续费/两融池。", "summary": "从银行/保险/券商视角综合前三卡，认为'概念虹吸vs实体溢价'是伪命题：最具持续性的实体溢价上移一层，落在金融管道——券商手续费/两融池与空间保险承保池。A股概念制造商短期确是杠杆陷阱（两融2.9万亿创新高、115%即时平仓线传导踩踏[S6][S7][S8]）；券商'做多量、做空尾'；空间保险才是…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "1516点并非真实铁底，日本央行6月16日会议引发的日元套利逆转才是科创50面临的真正‘负Gamma’威胁。", "summary": "本次分析挑战了宏观分析师关于科创50在1516-1540点筑底的观点。核心风险在于6月16日日本央行议息会议可能引发的全球日元套利交易大平仓。由于JPY/CNY处于0.0423的数十年低位，一旦汇率反弹，全球去杠杆压力将远超国家队建议的2000亿元对冲规模。此外，阿特斯母公司在美扩张带来的美元需求…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国家队科创50防守策略在‘一级申购优先’与‘二级折价才买’的限制下将形成严重的防御真空与套利漏洞，导致二级市场半导体与AI龙头在防线建立前遭遇流动性错配引发的估值踩踏。", "summary": "本卡片对科创50在‘宽基先行、科创后置’策略下的执行路径进行压力测试。分析指出，‘一级申购优先’机制存在显著的算法建仓时滞，面对两融平仓与雪球敲入形成的710亿元级联抛压（对前五大半导体巨头达106.5亿元/天），国家队12亿元/天的日均算法买盘存在8.88倍的流动性赤字。此外，‘二级折价才买’规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "载荷降本路径清晰，但利润中枢向GaN/硅光芯片与用户终端两端集中，中段总装被客户内化与合规成本双面挤压；可投资的'制造利润'兑现于半导体设计与消费级终端，而非传统卫星总装。", "summary": "载荷端（相控阵天线、OISL、终端）正从工程样机切入消费电子化拐点：相控阵单T/R通道成本由2019年~80美元降至2025年~12美元，OISL单价2022-2026E降约70%至25-30万美元/台[S1][S2]。但利润中枢向两端集中——上游GaN/SiGe RFIC与硅光OISL芯片毛利扩…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 5, "neutral": 0, "value": 40, "events": [ { "title": "本地两融回落、6-9月科创/创业解禁墙和既有0折算率池共同压制A股科技再加杠杆，海外risk-on只能提供交易窗口，不能支撑仓位上调。", "summary": "2026-06-12数据核实支持上一卡的本地约束判断，但需要修正表述：6月券商公告未显示新的科技股大面积折算率下调，真正约束来自既有0折算率池。两市融资余额较2026-05-29下降471.57亿元；6-9月全A解禁约9,300.88亿元，科创板+创业板明细合计5,522.78亿元；兴业证券202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "隔夜海外beta脉冲无法对冲A股科技的本地两融、解禁、折算率三重结构约束，科创50/CPO回补应继续硬性限制在5%试仓以内。", "summary": "维持科创50/CPO科技回补5%试仓上限不变。隔夜美股半导体与罗素逼空属beta脉冲，半衰期2–5日，无法改变A股科技端三重本地约束：两融余额被动去杠杆（时间常数2–4周）、6–9月解禁洪峰（日历事件、对海外情绪零弹性）、券商折算率下调后的二阶强平门槛（恢复需波动率连续4–8周回落）。三者时间常数…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力凸性敞口的A股本土复制：在融资融券、T+1与持仓限额下的在岸保证金与流动性约束", "summary": "本卡把卡片01–05的美元AI电力凸性杠铃翻译为A股在岸版。核心发现：杠铃只能部分复制。左尾对冲腿（300/500ETF、中证1000指数看跌）复制得很干净，甚至优于美元单一标的；但右尾主题凸性无法复制——中国无上市电力/天然气衍生品，唯一相关商品动力煤已被行政收紧（50%保证金、单日开仓约20手…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新一代信息技术链的超额收益已从泛算力概念转向‘物理硬约束（电力/连接）组件出海’与‘底层芯片（GPU/MCU/MEMS）国产替代业绩拐点验证’，需警惕2026-06-13所处周期的科…", "summary": "在2026-06-13评估中，针对新一代信息技术产业链4.3%的极低激活率，本报告系统梳理了5个低覆盖但具有强逻辑的A股/科创板标的。重点分析了摩尔线程（688795.SH）的GPU国产替代与首次扭亏、金盘科技（688676.SH）的电力物理瓶颈与出海溢价、鼎通科技（688668.SH）的高速连接…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 2, "neutral": 0, "value": 59.4, "events": [ { "title": "AI电力基础设施篮子应以14.0%总组合仓位参与电网设备确定性主线，同时用久期、金铜比和美元相关保护把选择性再通胀的宏观与拥挤风险锁住。", "summary": "截至2026-06-13，本卡综合前序结论：AI电力基础设施股票篮子应占多资产A股组合总资产14.0%，战术区间10.0%-16.0%，硬上限18.0%。篮子内部维持特高压/主变、构网储能和铜铝权益为核心，低配AI服务器和未锁电IDC。对冲上，配置23%国债久期、6%黄金/铜比价对冲，以及相当于篮…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股杠杆结构在100%保证金规则下韧性显著，红利退出压力可控，风格切换的主要风险在于科创50杠杆增长过快引发的情绪过热，而非两融爆仓。", "summary": "针对2026-06-14的A股杠杆结构压力测试显示，红利板块虽处于90%拥挤度高位，但得益于100%融资保证金规则，具备35%的价格安全垫，风格切换预计表现为“有序撤离”而非“平仓崩溃”。科创50融资余额触及自由流通市值5%的情绪极值，虽理论容量巨大，但需警惕情绪过热。整体质押风险局部化且受控，市…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 1, "neutral": 0, "value": 74.8, "events": [ { "title": "科创板去杠杆冲击短期股价但不破坏国产半导体设备及AI硬件的长期成长逻辑——龙头与AI芯片设计估值修复~20%即可加仓，二线与HBM配套需-30%以上才有安全边际。", "summary": "截至2026-06-15，科创板两融由~2,940亿元降至~2,820亿元、科创50击穿1,600点零Gamma线，对电子板块形成真实短期冲击，但不足以系统性下修国产半导体设备与先进制程龙头capex。设备龙头（北方华创订单覆盖15个月、现金~150亿元）、AI芯片设计（寒武纪2026Q1+312…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基于2026-06-15，国产算力+518亿元新增订单支撑下，A股科技板块两融与机构仓位重构具备坚实基本面，科创50重回1600点零Gamma上方将触发右侧杠杆盘强力追入与风格轮动反…", "summary": "基于2026-06-15，本报告定量评估了国产算力爆发背景下，A股科技板块（芯片设计、算力设备、PCB）的两融与机构仓位再平衡。大厂H2新增的518亿元国产算力订单有效对冲了两融收缩1200亿元的短期流动性冲击。公募基金TMT配置预计将从14.2%上修至19.8%，带来+2520亿元增量资金。科创…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 0, "value": 56.2, "events": [ { "title": "SpaceX首秀提升了A股航天与硬科技风险偏好，但中国宏观筛选器指向政策背书、订单可见和资产负债表稳健的少数供应链赢家，而不是普涨型半导体beta。", "summary": "截至2026-06-15，本卡综合支持前序“龙头虹吸 + 选择性重估”判断。中国宏观环境能支撑商业航天、卫星互联网和硬科技供应链的结构性行情：一季度GDP同比5.0%，5月出口以美元计同比19.4%，政策明确支持航空航天、卫星互联网和AI基础设施。但PMI新订单49.9、M2-M1增速差3.1个百…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "航天板块当前表现为针对宽基半导体的“流动性虹吸”与“高低切换”代理，而非带动半导体普涨的Beta引擎。", "summary": "截至2026年6月15日，本卡通过两融与资金流分析对航天带动半导体Beta的假设进行了压力测试。结论显示存在显著的“流动性虹吸”：6月12日A股虽放量成交3.24万亿并突破4000点，但资金呈“高低切换”态势，主力资金从电子（-58.3亿元）和通信（-91.2亿元）净流出，流向有色金属及事件驱动的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "若日元融资压力引发全球半导体估值去化，A股成长与AI板块需筑防线，应将科创50、创业板指、AI硬件及国产设备分别降至8%、8%、4%和5%的防御上限，待流动性出清与业绩验证后再重拾B…", "summary": "截至2026-06-15，本卡针对BoJ加息及日元套利交易平仓风险，对A股科创50、创业板指、AI硬件及国产半导体设备进行压力测试。建议将科创50及创业板指仓位降至8%，AI硬件降至4%，国产设备降至5%。并明确了日元汇率稳定、A股缩量筑底与中报超预期等重新加仓的量价与盈利修正信号。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 6, "neutral": 1, "value": 42.6, "events": [ { "title": "两融折算率下调引致的350亿元刚性去杠杆将重创高贝塔科技硬件板块，推动资金向筹码结构极净、收益率稳定的‘物理红利’与‘成本通缩’AI SaaS板块非对称旋转。", "summary": "截至2026年6月17日，本报告评估了两融折算率从0.6下调至0.4对A股两融规模和市场结构的影响。券商分批落地该政策将引发250亿至450亿元（中枢350亿元）的刚性融资收缩。科技硬件（电子/计算机）板块融资余额高达5608.81亿元，面临50亿至112.5亿元的平仓卖压。在负Gamma共振下，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "6月15日成长行情是可交易挤仓而非成长牛市重启，应维持半导体/CPO有上限参与，并把科创50 1550/1516作为低置信但高冲击的触发区间。", "summary": "截至2026-06-16，本卡对6月15日成长爆量轮动给出压力测试结论：反弹可交易，但尚非可持续配置信号。6月A股解禁3332.85亿元、影石创新约403.91亿元；5月25日两融余额曾达29252.09亿元，但截至6月12日已降至28734.82亿元。科创50 6月15日收1748.33点，距1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SPCX期权首日即面临1-DTE/0-DTE短久期希腊字母爆发，在FOMC决议与周四三巫日共振下，做市商Short-Gamma/Vanna对冲踩踏将在低VIX（16.20）环境下非线…", "summary": "基于2026-06-17，分析表明SPCX期权首日上市、FOMC决议和周四三巫日共振，且周五休市，将大幅放大美股跳空及科技股下行风险。在VIX降至16.20的低波动伪装下，SPCX首日交易即面临极端的1-DTE/0-DTE Short-Gamma和Short-Vanna对冲踩踏。若美股科技股深度调…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美股科技股回调引发的A股流动性挤兑风险，其核心痛点在于860亿两融‘预警区’杠杆资金的踩踏，而非雪球产品的局部敲入。", "summary": "基于2026-06-17，量化分析显示A股两融余额处于2.87万亿元的历史高位，且高度集中于电子与通信板块。科创50指数目前（1,758点）距离雪球密集敲入带（1,600-1,700点）仅一步之遥。压力测试表明，若科技股回调20%，约860亿元处于‘预警区’的杠杆资金将面临强制平仓压力，其流动性挤…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 3, "neutral": 0, "value": 59, "events": [ { "title": "SpaceX被动指数纳入下的美股科技虹吸与中港科技两融去杠杆风险", "summary": "SpaceX被动指数纳入下的美股科技虹吸与中港科技两融去杠杆风险 SpaceX Index Siphoning, Mag-7 Rebalancing, and Dual-Market Tech De-leveraging SpaceX快速纳入指数带来的600亿美元资金虹吸，如何加剧美股半导体及港股…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BOJ加息后的承重风险压倒性地在FX/利率簿（数万亿美元的银行/寿险FX对冲+套息），全球科技与港股/A股科技是二阶反身性受害者而非主管道。", "summary": "围绕2026-06-16 BOJ加息，用IMM日元仓位、东京零售FX、BIS LBS与BOJ/MOF资金流量+IIP四套数据测算日元套息簿结构：FX/利率簿（银行/寿险FX对冲簿数万亿美元 + JPY融资G10/EM外汇套息 + JGB与互换利差）占绝大头，JPY融资股权多头含全球科技高峰期亦仅几…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-17，A股市场结构不支持确认电子与科创50去拥挤，只支持继续持有但带上限的硬科技风险预算。", "summary": "截至2026-06-17，本卡压力测试第05卡再平衡方案：融资余额、ETF资金流和市场宽度都未确认电子与科创50拥挤度降温。科创板两融余额仍连续增加，科创50和电子ETF仍有资金承接，2026-06-17科创50大涨但超过3700只股票下跌，显示行情更窄而非更健康。结论是维持70/20/7/3组合…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 3, "neutral": 0, "value": 60.4, "events": [ { "title": "A股电子/TMT融资仓不能等到-18%才降杠杆；价格、ETF赎回、两融拥挤与券商风控线共振时，应提前强制削减融资敞口。", "summary": "截至2026-06-18，本卡片把-18%政策底压力测试转化为执行触发器：科创50、创业板、申万电子中任意2个破位，叠加股票ETF连续大额净流出、电子融资净买入在弱价中继续居首、ETF/科创两融余额反常上升，以及券商115%即时平仓线或双创集中度附加线收紧，应在-18%失败前强制降融资。政策底只有…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在美债收益率冲高至4.66%与套利交易平仓阴霾下，伴随期权保护成本飙升的“AI挤压”上涨反映了高位派发期的脆弱特征，A股配置应从无保护高杠杆科技向红利与期权保护的防御性哑铃风格重构。", "summary": "截至2026-06-18，美股“有保护的AI挤压”被判定为高位派发前奏而非健康风险偏好。在美联储主席Kevin Warsh首次决议鹰派暂停、美债10Y利率冲高至4.66%以及日本央行意外加息至1.0%引发套利交易平仓风险背景下，SpaceX（SPCX）超大规模IPO上市后回吐涨幅（开盘209.74…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体代工与设备估值面临全球WACC上升及N2晶圆价格墙的压缩，叠加国内HBM3国产材料良率瓶颈导致的算力变现延迟，正驱动A股电子资金扎堆“四大龙头”并形成‘高融资、弱宽度’格局，面…", "summary": "截至2026-06-18，本压力测试卡片验证了半导体设备和材料需求的边际变化如何驱动电子板块的宽度衰退与杠杆拥挤。中芯国际、海光信息、寒武纪、北方华创“四大龙头”融资余额达650.0亿元，占板块总两融的17.62%，而板块宽度持续恶化（仅42.1%个股上涨）。压力测试显示-15%的回撤将引发四大龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 6, "neutral": 1, "value": 46.8, "events": [ { "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 日报总编： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: cd8f2a28-c7a9-4dc8-aebc-e5ed66354208 结果摘要: Wrote handoff_response.md (17618 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "1750点可守但条件苛刻，1550点必须守但资金需求远高于既定预备金；当前是负Gamma踩踏设置而非已确认踩踏。", "summary": "截至2026-06-20，本卡压力测试认为：1750点仍是有效战术刹车，但只有在全市场成交额不跌破1.8万亿元且ETF/券商承接达到约2500亿-3000亿元时成立；1550点是清算阈值而非天然地板，若低成交真空中触及，潜在两融与期权对冲抛压约7150亿-9140亿元，约2000亿元预备资金不足。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SPCX 当前是 4% 流通盘驱动的单名 IPO 泡沫，但具备晋升为零售 AI 新锚的结构条件，关键测试是 2026 Q4 首批解禁。", "summary": "SPCX（SpaceX）于2026-06-12在纳斯达克上市，发行价135美元，融资约750亿美元，公众流通盘约4%。前三个交易日：D1收161（+19%）、D2收192.5、D3摸高225.64，三日零售净买入约3.7亿美元。结论：近期是单名IPO泡沫（4%流通盘+指数纳入被动买盘主导），但功能…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在30% A股配置中，电网与散热设备凭借低融资拥挤和强出海景气度应立即建仓，而国产半导体在两融高负荷、折算率下调以及雪球敲入风险完全出清前应保持现金等待，通过分步建仓捕捉12.48%…", "summary": "截至2026-06-20，针对30% A股配置，本报告建议采取“先物理、后算力”的分步建仓策略。20%电网/散热板块应立即建仓；10%国产半导体板块因融资余额达7382亿、两融折算率下调及1516-1550点雪球敲入威胁，面临严重的左尾去杠杆风险。待半导体释放-15%回撤且杠杆出清后建仓，可使A股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 1, "neutral": 0, "value": 75.2, "events": [ { "title": "跨季流动性挤压导致券商杠杆净利差受挫，压力测试显示券商Beta的重定价弹性高度依赖成交量的平台期企稳，建议目前维持大行红利底仓防御，确认DR007回落至1.50%以下且成交量稳定在1…", "summary": "本报告对跨季后A股成交量与两融余额稳定性进行了压力测试。6月18日成交额冲上3.33万亿元，17日融资余额回升至2.93万亿元，但跨季DR007利率飙升至1.90%，压制了杠杆净利差。压测表明，基准情景下成交企稳于1.90万亿元可带来券商板块+85.24%股价重定价空间；但若流动性偏紧导致交易量回…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 1, "neutral": 0, "value": 76.6, "events": [ { "title": "A股降险的核心不是卖出中国，而是在美元偏强和套息平仓风险下，优先削减流量、杠杆和供给三类脆弱Beta，并用防御-优质成长杠铃替代单边成长仓位。", "summary": "截至2026-06-22，本卡综合前序美元偏强、套息平仓和多资产降险框架，认为A股不应全面risk-off，而应沿四个市场结构抓手降险：互联互通成交占比已达12.97%，先减最拥挤Connect beta；两融余额约2.95万亿元、两融成交占比10.67%，压降融资Beta；6月解禁约3332.8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 3, "neutral": 2, "value": 72, "events": [ { "title": "科创50雪球负Gamma抛压与电子两融去杠杆的反身性踩踏机制", "summary": "科创50雪球负Gamma抛压与电子两融去杠杆的反身性踩踏机制 Star 50 Snowball Negative Gamma and Electronic Margin De-leveraging Feedback Loop 在科创50跌破1500点（雪球密集敲入带1516-1550）的Scena…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "科创50雪球负Gamma抛压与电子两融去杠杆的反身性踩踏机制", "summary": "科创50雪球负Gamma抛压与电子两融去杠杆的反身性踩踏机制 在科创50跌破1500点（雪球密集敲入带1516-1550）的Scenario B下，券商雪球对冲盘的负Gamma卖压（约36亿-200亿元分歧）如何与电子板块6132.72亿两融强平盘（约357亿元跌破强平线）相互点火？我们应如何利用…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "清洁能源杠铃应按等风险贡献而非资本权重构建（短债占约70%资本、各分支均担25%风险），并外挂年化约60bp Theta drag的日元套利平仓尾部对冲——其代价应以全天候夏普（0.…", "summary": "基于2026-06-23，本卡将首席策略师的防御杠铃论点量化为等风险贡献（ERC）预算。因短端国债波动率仅约2.5%、公用/红利股为16-20%，25%等风险贡献下资本权重约为长江电力11%、中国核电8.8%、大行/电信红利10%、短端国债70%，组合波动率约4.8%、基准夏普0.44。真正的剩余…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电子行业6132.72亿元两融杠杆中，357.00亿元处于维保比例临界带，-10%的行业性下跌将触发475.00亿元首轮复合抛压，高融资集中度的半导体龙头面临严重的负Gamma去杠杆…", "summary": "截至2026-06-24，本报告对电子行业6132.72亿元两融余额的维持担保比例分布进行了深入压力测试。当前分布显示，安全区（维保>150%）占比92.11%，而预警与强平区（维保<=150%）合计达484.00亿元。若科创50跌破1500点触发-10%的行业性回调，维保比例低于144.4%的3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 5, "neutral": 1, "value": 58.4, "events": [ { "title": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论", "summary": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论 JPY Carry Trade Unwind vs. APAC Tech Margin Call Spiral 美元兑日元跌破161.50关口，财务省干预概率陡增。一旦日元套利平仓瞬间引爆全球EM科技股抛售，叠加SpaceX二次配售24.5亿元的资金虹…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "现金与短债解决的是流动性生存问题，不能解决强平市场中的收益对冲问题；在券商净资本约束放大顺周期去杠杆时，预先配置的上市科创50看跌期权才是本轮压测中可验证的核心防线。", "summary": "截至2026-06-25，本卡从银行、券商、保险与金融科技视角综合前序去杠杆压测。结论支持第04卡：现金与短久期国债是必要流动性缓冲，可降低被迫卖出与组合波动，但不能产生抵消成长股暴跌的正凸性；券商融资融券净资本约束会使强平过程顺周期化。上市科创50看跌期权以交易所合约形式提供最清洁的负贝塔和正凸…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CPO“量产延期”谣言与AI硬件基本面交付的真空博弈", "summary": "SemiAnalysis（约2026-06-09）将CPO大规模商用推后至2028–2029，引发美股光通信单日重挫（AAOI约-17%、COHR约-11%等）。但散户犯了范畴错误：被推后的是CPO，而非“易中天”当前出货的800G/1.6T可插拔模块；CPO延期反而延长可插拔与NPO的需求窗口，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核心结论：基本面传闻未证伪光模块需求，但726.5亿元的中际旭创+新易盛两融余额足以把传闻放大成价格真空。", "summary": "截至2026-06-24，本卡对上一卡的基本面偏多结论做压力测试：中际旭创与新易盛的直接风险不是科创50雪球机械卖盘，而是两融余额分别达465.3亿元、261.1亿元，合计占全A两融约2.42%。科创50雪球属于二阶尾部风险；若以80%敲入测算，当前科创50对应的示意敲入区约1530-1590点，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 2, "neutral": 0, "value": 74.8, "events": [ { "title": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局", "summary": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局 Midday A-Share Growth Rebound and Options Negative Gamma Clearance Tactical Asset Allocation 在上午科创50跌破关键支撑触发雪球敲入与两融强平后，午…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CNH套息流动性溢价回缩不会抑制A股半导体上游的当期订单与实物资本开支(政策供资、提前排产、实物周期脱钩)，但会压缩其股权估值并抬高下一轮股权融资扩张成本(股权周期未脱钩)。", "summary": "截至2026-06-25，本卡片对全球半导体设备周期与A股AI基础设施流动性风险做脱钩验证。结论：脱钩因传导链而异，须拆为两条。(1)订单/实物资本开支链——大体脱钩、低风险：北方华创订单排至2027年、营收指引¥468-520亿，中芯2026资本开支约80亿美元持平，由大基金三期(¥3440亿)…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 4, "neutral": 1, "value": 67.2, "events": [ { "title": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论", "summary": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论 美元兑日元跌破161.50关口，财务省干预概率陡增。一旦日元套利平仓瞬间引爆全球EM科技股抛售，叠加SpaceX二次配售24.5亿元的资金虹吸，A股电子板块6132.72亿元两融及科创50雪球在1500点敲入区附近会否触发自反馈平仓踩踏？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "TMT与科创板在1600-1650点极限托底下的两融账户维持担保比例敏感性分布与券商风控应对", "summary": "若科创50下探1600-1650极限底，真正的约束并非130%维保线，而是折算率经‘保证金可用余额’施加的更快渠道。全市场CMR约245%（结构性均值278%），中证1000已降至218%；微观分布按券商规模高度倾斜：头部AAA券商270-300%、低于130%尾部约0.3%，中小券商200-22…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局", "summary": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局 在上午科创50跌破关键支撑触发雪球敲入与两融强平后，午间主力ETF资金流入与个股金针反转是否标志着负Gamma抛售压力的阶段性出清？我们应如何利用这一估值剪刀差进行右侧建仓，从红利低波向高Beta科技股进行再平衡？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "局部去杠杆风险向全市场流动性收缩传导的压力测试", "summary": "压力测试结论：在2026-06-27真实的宏观信用与资金环境下，仅限电力设备的两融高压回撤不足以单独触发A股全市场流动性紧缩或信用传导。三条通道均被切断：(A)两融强平——杠杆集中在电子(+62%YTD)/通信而非电力设备，全市场两融/总市值仅约3.2%[自测算]，局部击穿在本腿内自我出清；(B)…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 11, "neutral": 0, "value": 23.6, "events": [ { "title": "在 USD/JPY 跌破150的情景中，在岸高两融TMT的主要脆弱性是融资余额绝对规模和券商风控顺周期收紧，而非接近交易所25%单股融资上限。", "summary": "2026-06-27：本卡压力测试卡片06。日本4月持有1.2099万亿美元美债，USD/JPY跌破150的流动性通道方向成立，但2150亿-2850亿美元清算规模未被独立证实。A股两融余额为30329.9亿元，核心TMT行业约4670.0亿元、占15.4%。TMT风险不在交易所25%融资暂停线，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 工业制造分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: eb737f41-b97c-43b9-9b89-04a7c7636133 结果摘要: Wrote handoff_response.md (10188 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元套利交易平仓与离岸CNH流动性收紧对A股防御重估的交叉传导风险", "summary": "本卡以global-macro压力测试card-08的“USD/JPY与CNH流动性熔断器”构想：支持其核心方向（知识库已确认应以二者作量化触发器），但否定朴素的“单变量、按价位、OR门”实现。四点修正：(1)冲击分两波——先美元流动性挤兑、后避险重估，价位触发会迫使在第一波真空里抛售防御股；(2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美债高实际利率和中美利差倒挂通过锁定金融套息成本、驱动企业实体套息分化，对离岸人民币和新兴市场美元债产生持久的再融资压力，限制了国内货币政策空间并固化了A股滞胀杠铃配置格局。", "summary": "截至2026-06-27，美联储新主席沃什的鹰派政策维持美债名义利率与实际利率于高位。10年TIPS实际利率处2.19%高位逼近2.3%关键分水岭，引发全球跨资产去杠杆。中美10年期国债-265bp的倒挂使金融套息交易被远期贴水和HIBOR挤空风险压制，即期USD/CNY稳定在6.79。但出海企业…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 4, "neutral": 1, "value": 70, "events": [ { "title": "虽然278%平均维保安全垫使两融连锁抛压呈局部非系统性风险，但20-30%回撤仍将触发最高2528.00亿的平仓压力并重创产业链，应通过减仓高倍数前道、超配低谷先进封装以及引入非担保…", "summary": "截至2026-06-28，两融杠杆在电子和半导体板块高度集中（总余额分别达6132.72亿和560.80亿），前五大龙头占用386.80亿。压力测试显示，20-30%的回撤会迫使大批账户跌破130%平仓线，极端情况下（科创50跌破1520）诱发2528.00亿平仓压力，并因雪球负Gamma在2-5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "券商信用风险的局部破线是否会通过流动性传导至整个金融板块，进而引发非对称的流动性危机？", "summary": "本卡对半导体担保品集中度超25%的核心券商做流动性压测，部分支持卡片06并修正其传导结论。半导体两融5,608亿元仅占两市两融余额30,009.71亿元的18.7%，故超25%是“选券商”问题而非系统总量问题。-30%尾部冲击下，多元化头部券商流动性缺口仅占净资本12.0%可轻易覆盖，但科技偏重中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "已确认 OMO 窗口下的跨资产布局：国债久期、人民币套息/对冲与利率敏感 A 股倾斜", "summary": "首席策略师将 Research note 01–04 综合为一本可交易的跨资产组合，应对强美元（DXY 101.50、美债2Y 4.1%/10Y 4.37%）、坚挺人民币（即期约6.80、5月结售汇顺差358亿美元）与已确认有条件10bp OMO窗口（2.5/3绿灯）三态共存。建议：国债40%（修…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。", "summary": "截至2026-06-29，本卡片支持前序“更高更久”利率压制A股成长久期的判断，但将风险定位为结构性踩踏而非全市场2015式去杠杆。两融余额为30329.92亿元，占A股流通市值2.83%，总量可控；但电子板块融资余额6261.87亿元，且6月融资流入高度集中。模型显示，科创50较2026-06-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 5, "neutral": 0, "value": 62.4, "events": [ { "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股红利拥挤度与7-11月年内解禁洪峰限制了风格切换空间，而在人行短端流动性上限失灵与CNH被动升值情景下，高负债、高Beta的券商板块将面临最严重的两融净利差挤压与反转冲击。", "summary": "截至2026-06-29，本卡对A股杠铃轮动的拥挤度与压力进行评估。全市场两融达30329.92亿元（TMT占41%），银行与防御红利极其拥挤，仓位达95%分位；6-11月面临2.8-3.2万亿元解禁洪峰，今日单日解禁达589.22亿元，严重压制内需成长。若CNH继续升值或人行回购走廊失灵（DR0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股市场流动性与资本流出风险评估", "summary": "压力测试卡片02的250–500亿美元流出预测。按USDCNH 7.0折算为人民币1,750–3,500亿元（0.17–0.35万亿），仅占A股流通市值0.17–0.33%、单季成交不足0.4%、外资持仓6.8–13.6%；外资仅占流通盘约2.46%，且北向已降至近三季新低，提前防御。直接冲击可控…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-29的核心判断是：人行短久期托底利好金融与内需现金流，不能无差别追成长，出口型TMT硬件是CNH升值情景下最需要降权的A股风格风险。", "summary": "截至2026-06-29，本卡综合前序CNH挤压与人行受限宽松链条：境内流动性问题更像波动率与久期冲击，而非持续资金断裂。A股应采用金融+精选内需成长的杠铃结构：银行作为资产负债表稳定器超配，券商作为成交和两融高beta适度超配，高股息作为压舱石，内需硬科技/软件精选配置；出口型TMT硬件因CNH…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-62e0155999/", "networkHref": "/investor/research-network/#thesis-62e0155999" }, { "id": "thesis-5cb77a9f52", "source": "auto_topic", "title": "在红利拥挤度逼近历史极值、长江电力达90%且地方拖欠致公用账款增加15%背景下，红利底座应坚守水电核电，还是大举切仓至融资占比低、严重低…", "titleEn": "Practical research topic: 在红利拥挤度逼近历史极值、长江电力达90%且地方拖欠致公用账款增加15%背景下，红利底座应坚守水电核电，还是大举切仓至融资占比低、严重低…", "lane": "中国权益与流动性", "laneEn": "中国权益与流动性", "laneHref": "/investor/lanes/lane-c06f0832/", "investableFocus": "", "stockUniverse": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "510300.SH", "588000.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK", "9988.HK", "APO", "IEF", "CAT", "COP", "TLT", "MAR" ], "coreView": "该主线来自 4 条研究记录的标题、问题和摘要聚类，时间跨度 2026-06-06 至 2026-06-22；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-15", "lastSeen": "2026-06-29", "supports": 8, "risks": 140, "neutral": 12, "analystCount": 13, "sourceTypeCounts": { "研究笔记": 153, "研究主题": 3, "研究线程": 4 }, "topTerms": [ "长江电力", "A股", "电力", "水电", "红利", "核电", "在红利拥挤度逼近历史极值", "加15%背景下", "红利底座应坚守水电核电", "还是大举切仓至融资占比低", "严重低", "能源" ], "whatChanged": [ "2026-06-29: 选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "2026-06-29: 变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "2026-06-29: 在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15.36倍AI潜在电力需求，算力扩张将受限于刚性电力配给，投资配置应向短交期电力变压器出口链与物理红利安全垫倾斜。", "2026-06-29: 在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红利（长江电力、中国核电）进行防御性轮动，并辅以日元和美债的尾部风险对冲。", "2026-06-29: AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利景气度，A股资产配置应向电网出海及水电红利哑铃端重构[自测算]。" ], "practical": { "sectorLane": "中国权益与流动性", "sectorLaneEn": "中国权益与流动性", "exposureBucket": "中国权益与流动性", "tone": "green", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "510300.SH", "588000.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK", "9988.HK", "APO", "IEF", "CAT", "COP", "TLT", "MAR" ], "primaryStocks": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" } ], "instrumentTags": [ "510300.SH", "沪深300 ETF", "588000.SH", "科创50 ETF", "512100.SH", "中证1000 ETF", "2800.HK", "盈富基金", "MCHI", "iShares MSCI China ETF", "KWEB", "KraneShares CSI China Internet …" ], "reviewCadence": "每日复核", "confirmationSignals": [ "融资余额、ETF 申赎和北向资金是否同向", "被动调仓窗口是否被主动资金提前交易", "订单、盈利和估值是否同步改善" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "指数期权", "红利低波", "汇率", "两融降温" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 3 类，近期事件 124 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 3, "analystCount": 13, "recentEvents": 124, "supportRatio": 0.05, "riskRatio": 0.88, "conflictScore": 98, "evidenceDepth": 100 }, "score": 66.9 }, "events": [ { "date": "2026-06-29", "title": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利…", "summary": "本报告对AI算力激增带来的电力电网物理瓶颈（变压器交期120-210周[S2]、并网排队4-7年[S1]）对A股转型龙头的利润率传导进行压力测试。结果显示，电网扩容滞后充当了新能源装机的“限速阀”，压制了光伏/电池销量增速；同时，容量电价[S8]和碳成本上升[S19]增加了高能耗制造环节电费开支，而在严重过剩背景下成本转嫁失败，导致制造端转型龙头利润率遭到…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "长江电力", "A股", "电力", "水电", "红利", "能源" ], "href": "" }, { "date": "2026-06-29", "title": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红利（长江…", "summary": "本报告针对2026年三季报TMT硬件估值折价预期，校准了全球全天候和风险平价模型。战术防御性风险平价（TDRP）将AI成长股风险预算由25.0%下调至8.0%，防守红利上调至40.0%，以对冲55.0%的TMT增长溢价折价。模型显示，电力配给约束下算力效率的TCO优势在24/7清洁能源下扩大至9.97%，高功耗算力中心毛利率面临负增长。此外，建议红利套组内…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "长江电力", "电力", "红利", "核电", "能源" ], "href": "" }, { "date": "2026-06-29", "title": "在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15.36倍AI潜在电力需求，算力扩张将受限…", "summary": "截至2026-06-30，本研究对Jevons悖论激发的15.36倍潜在AI用电需求（相当于727.6 GW新增平均负荷或6,374 TWh年用电）进行了极限物理压力测试。结论表明，在2030年前，由于北美变压器交期拉长至48-60个月、特高压直流送电网架高壁垒（全球需新建73条通道）以及取向硅钢（GOES）产能与高压试验大厅测试容量的物理刚性制约，没有任…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "长江电力", "电力", "红利", "核电" ], "href": "" }, { "date": "2026-06-29", "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优先既有水电/核电现金流、具备瓶颈定价权的变压器和电网设备供应…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电力", "水电", "红利", "核电" ], "href": "" }, { "date": "2026-06-29", "title": "选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "summary": "截至2026-06-29，本卡综合前9卡：AI电力约束是选择性时序上限而非订单毁灭。建议跨资产账本配置为40%受资助AI基础设施龙头、45%主权背书红利公用事业、3%年度申万电子Put权利金预算、12%流动性储备。进攻端以中际旭创18%、工业富联14%、英维克8%为核心；防御端以水电18%、核电17%、受监管电力/电网10%承接电力稀缺；Put以35-45…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电力", "水电", "红利", "核电" ], "href": "" }, { "date": "2026-06-28", "title": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费成本转嫁挤压。", "summary": "基于国产芯片2.94倍能效赤字拉长通电时间至18个月的物理瓶颈，本报告系统性量化了A股电力设备板块（特高压、大变压器）的估值扩张机制与低利润 traditional TMT 的电价挤压。电网容量受限与十五五规划共振，支持了特高压订单可见度与大型变压器出海高溢价，特别是在国内外硅钢5.1倍巨大价差的成本红利下。策略上构建杠铃组合，超配直接出口变压器龙头、特高…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "A股", "电力", "水电", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "局部中小银行信用缺口下的A股银行红利耐久性与高股息内部轮动", "summary": "卡08分离出的“额度沉睡”机制（CET1上限+NPL触发+低于1.40%净息差）不仅是中小银行信用风险，更是撕裂A股红利复合体的力量。重排结论：①大行红利居顶——分红不依赖沉睡额度，约4.35%股息率、对10年国债（约1.72%）257bp利差、0.5–0.7倍PB已计入坏消息，持有/超配；②公用事业高股息防御提级——长江电力等约3.6–4.2%股息率、承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "长江电力", "A股", "电力", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "summary": "截至2026-06-28，本卡对“AI负荷增长能否支撑公用事业高分红”作压力测试：结论是精选组合层面可以，板块整体不能机械外推。AI数据中心用电增长真实，但也同步拉动电网、发电和储能资本开支。水电、核电、容量电价支持的火电及低杠杆综合电力运营商仍可维持分红；风光开发商受消纳、电价和项目资本开支制约，不能直接视为高息资产。上一卡的25%公用事业红利仓可以保留…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电力", "水电", "红利", "核电" ], "href": "" }, { "date": "2026-06-27", "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌破3.0%的红利反向出清防线。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "长江电力", "A股", "电力", "水电", "红利", "能源" ], "href": "" }, { "date": "2026-06-27", "title": "能源与电力板块（长江电力/中国海油）的防御性估值与政策风险压力测试", "summary": "对卡片09把半导体腾出资金转入长江电力(600900)与中国海油(0883)防御的安排做压力测试，得出分裂裁决。电力/长江电力基本面不过热（零燃料成本、70%分红锁定至2030），但作为工具已成拥挤的26.8年久期债券替代品：股息率压至3.5%、对国债利差约2pp处历史高位、内在价值≈市值、贴现率+100bp回撤约-21%——在通胀再加速/加息根主题下恰是…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27.5, "matchedTerms": [ "长江电力", "电力", "水电", "红利", "核电", "能源" ], "href": "" }, { "date": "2026-06-27", "title": "1.3%低利率下，红利全天候组合应从追逐水电股息利差转向水电质量、规制煤电容量收入、国债久期与黄金/煤炭事件对冲的动态杠铃。", "summary": "2026-06-27：本卡综合前序链条，认为1.3%国债情景不是加仓红利资产的单向信号，而是要求把红利全天候组合从“长久期股息代理”改成“规制现金流杠铃”。建议正常环境水电22%-26%、煤电14%-18%、国债30%-34%、黄金/煤炭12%-16%、现金6%-10%；若1.3%国债叠加枯水/现货回购，水电降至12%-16%，煤电升至22%-26%，大宗…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "长江电力", "电力", "水电", "红利" ], "href": "" }, { "date": "2026-06-27", "title": "高两融TMT龙头下跌10-15%与折算率降至0.4的共振将引爆1372.7亿至2229.1亿元的强平与还款抛盘，跌停锁定将溢出25%的负Gamma抛压至其他流动性资产；资产配置应坚决撤离高两…", "summary": "截至2026-06-27，本报告针对A股高两融TMT龙头下跌10-15%与券商提高担保折算率（从0.6降至0.4）的双重冲击，定量压力测试了被动卖出量与微观流动性。测算表明，在A股两融总余额3.015万亿元、核心TMT占比15.4%（4670.0亿元）的背景下，10%下跌（温和压力）将导致平均CMR降至142.8%，触发1372.7亿元实际被动卖出；15%…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "长江电力", "A股", "电力", "水电" ], "href": "" }, { "date": "2026-06-27", "title": "A股“大行+红利/稀缺成长”哑铃配置在低杠杆与高可视度Backlog支撑下并未拥挤，而弱区域中小银行因资产质量二阶传导滞后与重度稀释风险，其资本金脆性尚未被市场充分定价。", "summary": "本卡片对A股“大行+红利/稀缺成长”哑铃配置拥挤度及非核心中小银行风险定价进行了量化检验。结果显示哑铃配置自身极低杠杆（大行0.40%、长江电力1.81%）使其整体处于安全区间（CCI为34.2-55.3），未发生拥挤。相反，非核心中小银行在+2.5%不良冲击下面临1.98万亿元资本缺口，现有中小银行专项债仅能覆盖约2.5%。其0.3x PB估值因面临3.…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "长江电力", "A股", "电力", "红利" ], "href": "" }, { "date": "2026-06-26", "title": "电力电网及公用事业红利资产的配置逻辑验证", "summary": "支持第01张卡片的防守再平衡判断，并验证其机制：本轮去杠杆是两融融资拥挤的电子板块冲击（A股两融2026-06-23破3万亿元），而受监管公用事业由保险/养老等长钱持有、融资拥挤度低，处于压力源的对立端，是真正避风港而非平行泡沫。需求端1-5月用电+3.4%、互联网数据服务用电+26%给出AI电气化成长期权；股息端长江电力约3.6%-3.8%对10年期国债…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "长江电力", "A股", "电力", "水电", "红利", "核电" ], "href": "" }, { "date": "2026-06-26", "title": "保留12%清洁+4%能源杠铃，但A股表达必须从宽基清洁贝塔转为受限的电网/储能/运营商组合，并用能源对冲抵消价值和通胀尾部而非追高红利能源。", "summary": "截至2026-06-26，本卡综合前序12%清洁基础设施+4%能源尾部对冲，但指出A股执行并非因子中性。清洁袖主要暴露于成长、资本开支、低股息和动量，尤其电网设备、电池、储能、光伏存在追涨和集中度风险；能源袖提供价值、高股息和通胀/霍尔木兹尾部保护，但中证能源前五权重约61.0%，不能无限扩张。建议保留杠铃，同时设置4.0%电网上限、4.0%储能/电池/光…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "A股", "水电", "红利", "核电", "能源" ], "href": "" }, { "date": "2026-06-26", "title": "A股市场结构约束下的清洁基础设施与能源对冲执行方案", "summary": "截至2026-06-26，本卡把Card03杠铃(12%清洁+4%能源)与Card04子仓上限落地为可成交的A股方案。核心结论：约束不难，难在表达载体可得性高度不对称——电网有唯一高流动纯标的(159326,约¥343.7亿,特高压权重约66%)；电池有百亿级但偏电动车(159755,约¥158.9亿)；绿电ETF含约34%火电(562550)污染运营商口…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "长江电力", "A股", "电力", "核电", "能源" ], "href": "" } ], "timeline": [ { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "公用+电信SOE作为红利压舱石的判断成立，但应集中水电/核电+移动/电信H股、最小化纯火电与联通，整桶红利组合质量与收益率同步上行。", "summary": "截至2026-05-15，公用与电信SOE的红利防御属性整体成立但内部分化明显：水电（长江电力）、核电（中广核、中国核电）和移动/电信（H股股息率6.0–7.0%）现金流稳定、派息政策正面，构成红利压舱石的最干净底盘；纯火电IPP在秦皇岛煤价RMB 690–760/吨与容量电价(2026预计~RM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 1, "risks": 0, "neutral": 0, "value": 51.4, "events": [ { "title": "估值范式不是从算力到电力的单向翻转，而是A股AI链的三层分层——核心alpha在T2能源获取复合体（公用事业+特高压+核电+重装备），重估扳机是AI专属电价分类与首例AI-PPA披露。", "summary": "支持前序结论但拒绝二元翻转：正确框架是三层分层。T1（算力/硅片）从40-60×降档至25-35×成长周期估值；T2（能源获取复合体——拥有AI-PPA的公用事业、特高压电网设备、燃气轮机、核电链）由12-15×债券替代重估至18-25×AI-PPA股权代理，是新核心；T3（应用/token消费方…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "在油—煤—气联动重新走高的环境下，'公用事业=防御'只在水电+核电+水务+央企红利窄篮子口径下成立；火电与城燃利润率反而被压缩，把它们打包成防御资产将让组合的油价beta显著为负。", "summary": "对Research note 02的部分压力测试：'水电与受监管公用事业作为防御红利'的口径只对了一半。Brent站上104美元、煤炭与JKM LNG联动重新走高的环境下，水电（长江电力600900等）、核电（中国核电601985）、水务保持防御性，但火电（华电国际600027、华能国际60001…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Brent高于104美元会通过输入性PPI和利润率压缩强化A股从外交beta和周期成长向水电、受监管公用事业与红利资产切换的趋势。", "summary": "截至2026-05-17，本卡支持上一卡判断：Brent突破104美元已成为独立的通胀与利润率冲击。美国4月CPI同比3.8%、能源同比17.9%；中国4月CPI同比1.2%、PPI同比2.8%，且石油、燃料加工、化工链条环比涨幅显著。油价从89.4美元升至104.2美元的16.6%冲击，按IMF…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "主权电力资产可以承担太空杠铃的防御端，但应优先水电/核电现金流锚，限制高杠杆新能源开发商权重。", "summary": "截至2026-05-21，本卡支持把主权/国资电力资产纳入太空杠铃防御端，但要求按现金流质量而非股息率排序。长江电力和中广核电力更接近核心防御锚，华润电力和中国电力提供更高收益与转型敞口但资本开支、杠杆和市场电价风险更高。样本股息率约3.67%-5.64%，相对中国10年期国债1.75%有明显利差…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "将资金从拥挤且面临变现分歧的TMT硬件战略性轮动至绿色基荷、电网设备与农化周期底部板块，是2026下半年对冲AI Capex见顶及WLEF物理瓶颈的优选策略。", "summary": "本报告评估了绿色基荷（核电、大水电）、电网设备（特高压、电力出海）和农化板块（磷肥、钾肥）在A股市场的持仓、拥挤度与估值。电力设备拥挤度高但海外需求强劲，大水电高分红抗震，核电受益AI算力重估，农化板块处于周期底部性价比高。建议减仓拥挤光模块，向上述WLEF物理实体底座进行轮动，超配长江电力、中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 4, "neutral": 0, "value": 23.6, "events": [ { "title": "核电与水电凭借年金化现金流、制度化分红与极低两融拥挤度,是当前A股最合格的红利价值防御底座,可在卡片01给出的59%红利仓位中担纲12%-15%的硬核压舱石。", "summary": "在卡片01关于微盘/北证50资金分母收缩与红利价值切换的判断之上，本卡对核电与水电的防御效能进行了量化验证。三维评估表明：水电(长江电力)与核电(中国核电/中广核)具备(1)经营性现金流毛利率65%+ / 48%+的现金流年金特性；(2)分红率50%-70%/35%-45%的制度化承诺；(3)两融…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "水电跨省长协壁垒与核电单向价差合约（CfD）提供了极强的基本面防御力，即使在标准煤价暴跌-30%的极端悲观情景下，四大公用事业巨头的净利润侵蚀依然被牢牢控制在-5.00%以内，分红潜…", "summary": "基于2025年实际数据对电力联动与煤价下跌对水电与核电利润的蚕食效应进行了极值压力测试。研究发现，在标准煤价暴跌-30%的极度悲观预期下，因水电低市场化长协锁定与核电90%比例单向价差合约（CfD）托底，四大电力巨头的归母净利润侵蚀幅度均被牢牢控制在-5.00%以内。其中长江电力仅为-3.92%，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力市场化深化对水电核电长协是温和侵蚀而非崩塌,中性悲观情景下政策让利路径净利侵蚀(-4.9%至-7.7%)已超过燃料价格路径,卡片02压舱仓位中枢维持但应预留-2pp政策缓冲。", "summary": "本卡对卡片02—04的核心防御结论做政策视角压力测试。2026-05-22时点,136号文(新能源全面入市)、煤电容量电价、跨省送电基准+浮动、绿电直供四条改革线索叠加,正在对水电核电长协形成温和而非崩塌式的让利压力。自测算显示:在中性悲观情景(跨省协议让3分钱+现货占比抬至20%+工商业让利)下…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融资金去杠杆已通过账户交叉质押与VaR风控限制从TMT扩散至国防军工、新能源等高拥挤成长板块，必须通过‘分化杠铃’策略将大仓位配置于长江电力等红利安全岛、将科技仓位战术收拢至受国家…", "summary": "本卡片系统综合了A股去杠杆负螺旋对全市场的跨行业传染风险。我们测算指出，散户账户交叉质押与VaR风控触发将两融出清压力从TMT扩散至国防军工、新能源等高拥挤成长板块。科创50指数1,750点关口面临882亿元密集敲入与平仓卖压，但在国家队专项买盘防护下形成有底成长资产。我们建议采取‘分化杠铃’策略…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 2, "neutral": 1, "value": 40, "events": [ { "title": "A股高股息电力不是主动机构拥挤交易，但已成为被动红利资金拥挤载体，合理股息率边界约3.25%-4.25%，低于3.0%的核电/水电叙事资产尾部风险最大。", "summary": "截至2026-05-24，本卡压力测试A股高股息电力公用事业。结论是：板块存在ETF与战术红利资金拥挤，但主动权益基金并不拥挤，2026Q1公用事业配置仅0.45%，核电仅0.03%。容量电价提高煤电分红稳定性，10年期国债1.75%对应合理股息率走廊约3.25%-4.25%。长江电力3.77%仍…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面对日元套利平仓引发的全球去杠杆，A股是新兴市场内次坏的防御港，但承接资金的仅是国家队锚定的高股息低β国企篮子；指数总盘和北向重仓成长股仍是高β出口通道，必须主动回避。", "summary": "在第5张卡的10%日元套利平仓基准情景下，A股属于部分隔离但非免疫，正确姿态为杠铃式防御。承接防御资金的主篮是国家队β≤1.0、股息率≥4%、在岸资金主导的标的：国有大行、电信三大、长江电力/核电、高速公路与基础设施REITs、A股油气煤龙头；这些前十大股东中保险+社保+汇金占自由流通股55-60…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "3-5日VWAP卖出红利底座本身可执行，但若卖出行为触发红利资产拥挤重估，组合VaR二次扰动可能比交易冲击成本更重要。", "summary": "截至2026-05-24，本卡压力测试认为，100.00亿元人民币组合中的59.00%水电/核电红利底座可以作为高质量流动性储备，但不能视同现金。按3-5日VWAP卖出1.50%-6.00% NAV，在长江电力、华能水电、中国核电、中国广核等核心标的中总体可执行；正常冲击约1.2-2.4bp NA…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "在科创两融高拥挤与物理能量墙硬约束下，A股AI硬件配置应刚性削减名义科技仓位并向物理瓶颈环节精准收拢，同时超配物理底座红利，美股配置硬核底座与SaaS应用，港股回避解禁大潮。", "summary": "本卡片将AI利润向算力底座迁移的判断转化为跨市场大类资产与风格配置体系。A股面临高两融杠杆（科创板 margin-to-float 达4.9%）与“网架能量墙”双重制约，建议名义成长科技仓位刚性削减至18.0%，向高端光互联（中际旭创2026Q1毛利率逆势升至46.06%）、Blackwell柜级…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 1, "risks": 3, "neutral": 0, "value": 41.8, "events": [ { "title": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构", "summary": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构 USD/JPY 145 Shock and Margin Deleveraging Reshaping the A-Share True Barbell USD/JPY下探145引发中资境外AT1永续债置换成本飙升，直接压降上市银…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构", "summary": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构 USD/JPY下探145引发中资境外AT1永续债置换成本飙升，直接压降上市银行NIM约3-5bp，动摇了‘金融红利’板块绝对防线；而科创50两融拥挤度高达4.9%，折算率下调引致250亿-450亿平仓压力，并威胁到敲入线挂接的27…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大水电仍是防御杠铃核心，但当前应压力测试的是资金流入速度和估值利差压缩，而不是把全部红利资产当作安全港。", "summary": "截至2026-05-28，本卡压力测试认为：从AI硬件向大水电切换仍可行，但必须以流动性和估值闸门执行。长江电力PE为18.47x、PE分位1.93%、融资融券占流通市值1.81%，尚未显示防御端自身拥挤；但中证红利PE分位已达83.84%，宽基红利不宜无差别买入。若AI卖压在3-5个交易日内集中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在科技股杠杆挤压背景下，大水电与核电具备防御配置价值，但大水电的现金流质量、分红强度和杠杆改善优于核电。", "summary": "截至2026-05-28，本卡支持从高拥挤科技资产转向“物理红利”，但建议分层：大水电是更优防御锚，核电是次级基荷成长配置。长江电力以强经营现金流、70.92%分红率和下降杠杆率体现现金流防御价值；核电具备高利用小时、政策支持和月度电费结算，但股息率较低、资产负债率更高、资本开支久期更长。结论是支…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 3, "neutral": 0, "value": 35.2, "events": [ { "title": "在红利踩踏压力测试下，核电与低北向区域水电是长江电力的有效平替，而城投水务与燃气因地方财政拖累或周期属性不具备真实防御底蕴。", "summary": "本报告对公用事业板块进行压力测试。在长江电力因9.2%北向持股面临外资踩踏风险的情境下，公用事业内部将严重分化。低北向区域水电（华能水电、川投能源）及核电（中国核电、中广核）凭借1.5%-2%的极低外资敞口与长现金流久期，能提供真实的防御刚性，是长江电力的完美平替。相反，城投水务因地方财政拖累导致…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股权益leg应采取“红利低波（水核火电，40%）+中盘价值（电网设备，40%）+周期价值（铜铝硅钢，20%）”的结构，其中电网设备的成长风险由30Y国债完美正交对冲，而周期价值作为…", "summary": "2026-05-29：在债券leg降久期至5.5年、由30Y国债提供核心尾部对冲、绿债/ABS提供底仓carry的新结构下，A股权益leg获得承担更高增长与通胀风险的预算。本卡将权益暴露拆分为“哑铃+α”结构：(1) 红利低波（制度回报段，约40%）：以水电核电和容量电价火电替代被低配的7-10Y…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI负荷在中国公用事业的传导是结构性分配而非总量上行，制度性回报集中在水电核电基荷、容量电价火电与电网设备承包侧，纯新能源IPP与东部省网平台承担未充分补偿的电网加固成本。", "summary": "2026-05-29：AI负荷增长不是中国公用事业的普涨故事，而是结构性切块。最高可兑现度的三类是：受监管电网投资和数字化承包侧（国电南瑞、许继电气、平高电气、思源电气、特变电工、中国西电），容量电价机制下的存量大型火电（华能国际、华电国际、国电电力、大唐发电，165元/kW容量电价[S1]），以…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 2, "neutral": 1, "value": 45.6, "events": [ { "title": "长江电力的资产已建成、改革免疫、派息有锚、利差超历史均值，可在当前电价市场化与资本开支周期下维持防御性与股息率优势，是承接A股医疗器械外溢防御资金的最深单一蓄水池。", "summary": "截至2026-05-30，本卡支持Research note 09防御溢出假设落地。长江电力（600900）以71.795 GW六库联调、2024年发电量2,958.4亿度、经营现金流RMB 642.7亿、≥70%派息底线，已经从'资本开支密集型水电开发商'转为'运维型公用事业'。2025年发改委…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在Research note 07合规边界下，2026下半年AI硬件杠铃应落地为'离岸瓶颈15 + CXMT替代8 + 红利30 + 债券22 + 黄金10 + 现金15'的100单…", "summary": "截至2026-05-30，本卡在Research note 07合规边界下，把杠铃组合落地为100单位可执行组合：离岸AI瓶颈15（000660.KS/2330.TW/MU/3711.TW仅公开普通股）、A股CXMT生态8（中微/北方华创/拓荆/长川）、A/H红利底仓30（国有大行+水电+电信+能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在高实际利率常态化下，股票组合应以‘高质量红利底仓+实体能源原材料价值链’作为双核避风港，并精选‘外需出海制造’以跑赢估值收缩，严格回避缺乏现金流的长久期成长beta。", "summary": "本卡在股票降至40%总配额及防范利率波动的大背景下，完成了A股（占60%）与全球股票（占40%）的股票组合内部二级风格与行业再分配。策略构建了以'质量红利底仓（水电/大行/消费龙头）'与'实物资产能源链（煤炭/石油/重金属）'为核心支柱、配合'优质出海景气制造'的哑铃组合，严格规避无现金流的长久期…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-31", "supports": 1, "risks": 3, "neutral": 0, "value": 46, "events": [ { "title": "杠铃 v3：全球做多物理层资本品+稳定现金流（A 股特高压/GOES/大水电/燃机为核心 37%），做空 Mag 7（除 GOOGL）+ 纯 DC REITs + AI 半导体 32…", "summary": "在 NVIDIA 库存恶化至 145–170 天与 Mag 7 FCF 共识下修 18–25% 的基准情景下，全球风格因子正发生 2023 年以来最大轮动：A 股已先行从高 β 算力转向高股息+质量+实物资产；港股呈现互联网资本开支收缩 vs 高股息电力双极分化；美股 Mag 7 等权与 S&P …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "从能源基本面看，AI-IDC 已是电力系统级事件，东部电价/PUE 是负荷出清结果，煤/水/核现金流久期支持 Research note 07 杠铃再平衡，实测缓冲更厚；微调建议为 …", "summary": "作为线程收官卡，自能源/电力供给端对 Research note 07 杠铃再平衡进行合成与支持。2026 年 IDC 用电需求约 3,650 TWh，占全国负荷 3.8–4.2%，使东部 PUE≤1.25 与 0.78–0.92 元/kWh 电价成为负荷管理刚需；西部 24/7 训练真实到手电价…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "公用事业链不是低关注度故事，而是煤电容量电价+ETS 扩容+现货市场三重重定价的早期窗口——可调度资产估值结构性上修，新能源 IPP 估值结构性下修。", "summary": "公用事业链 30 日激活率仅 11.7%，但正同时穿越三大体制切换：煤电容量电价 2026-01 提至固定成本 50%（高转型省 70%）；H1 2025 弃光 6.6%/弃风 5.7%、国家红线放宽至 10%；ETS 扩容至钢/水泥/电解铝（覆盖 +30 亿吨 CO₂e）。可调度容量被向上重估、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件底座面临物理电网与能源成本的双重严峻约束，其过度偏乐观的估值预期存在巨大脆弱性，而由长江电力与中国神华组成的公用事业防御组合展现出极强的抗风险红利韧性，完全应验并巩固了哑铃资…", "summary": "截至2026-05-31，本卡片对AI算力扩张的电力约束与公用事业防御组合的股息率稳健性进行了压力测试。定量仿真显示，AI算力在电网瓶颈与高电价下极其脆弱：若电价上涨至$0.220/kWh，云巨头GPU租赁毛利率将转负为-1.48%，导致Mag7估值乘数面临大幅收缩。相反，以长江电力和中国神华组成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 6, "neutral": 0, "value": 21.4, "events": [ { "title": "能源及电力高股息资产凭借其机制化隔绝周期的稳定现金流与极宽的股债利差，在隐含波动率上行与杠杆资金收缩时展现出卓越的估值韧性与回撤防御力。", "summary": "截至2026-06-01，本卡片评估并验证了高股息能源与电力板块在VIX波动率上行及融资杠杆收缩时强大的现金流稳定性和估值防御属性。得益于高长协煤占比、煤电容量电价机制与低成本梯级水电资产，长江电力与中国神华等红利龙头经营现金流极其充沛，并承诺高比例分红。压力测试表明其估值回撤极微（0.0%至-0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力天花板的核心痛点是并网延迟和规制规费引发的硬件折旧搁置与清洁电力溢价，这将在中期使云GPU租赁提价15%-35%[自测算]，并向SaaS客户传导3.0%-4.5%的毛利拖累[…", "summary": "截至2026-06-02，本卡对AI用电瓶颈做深度测试。电网并网设备延期（128-144周）引发的‘芯片折旧挤压’是最大隐形电力成本，1年延迟使GPU活跃小时折旧成本增50%[自测算]；加之PJM 2026年收取$4-$8/MWh合约规费，电力成本上涨导致云GPU租赁提价15%-35%[自测算]，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026Q3–2027Q1政策窗口前，A股应通过减配低毛利服务器整机、超配智能电网与液冷龙头，构筑‘清洁基荷公用事业防御墙 + 西部枢纽节点单选成长极’的优化哑铃型双轨配置，实现极…", "summary": "本报告针对2026-06-01基准情景B下，发改委出台示范园区式20%-30%输配电价折扣对A股公用事业防御性与AI算力成长股的冲击，确立了“双轨配置框架”。我们定量证明了由于折扣门槛极高，长江电力等龙头T&D收入流失低于3.0%，红利防御墙稳固；而西部算力枢纽度电成本降至0.385-0.395元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地方债置换潮会阶段性挤压红利防御端估值流动性，但只要绿电直连费用刚性和长江电力分红底线不破，最优解是用国债与黄金覆盖风险预算，而不是放弃哑铃框架。", "summary": "截至2026-06-01，本卡综合支持哑铃型双轨配置，但指出2026Q3-2027Q1地方债置换与超长期国债供给会对长江电力等红利防御资产形成阶段性流动性挤压，而非股息基本面破坏。基于2026Q1地方债发行3.1059万亿元、10Y CGB 1.7090%、AAA中短票利差41.2bp、长江电力股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "台积电关税成本转嫁推升GB200整机柜售价12.3%及云厂商TCO 8%，将引爆北美云厂商对 Blackwell 订单10-15%的下修，并迫使A股出海非GPU链降价5-8%，导致其…", "summary": "本报告针对2026-06-02台积电延期及关税转嫁事件，量化分析了其对英伟达GB200定价（上涨12.3%）与北美云厂商TCO（上升8.0%）的影响，指出云厂商可能下修订单10-15%。A股光模块、PCB、液冷出海链面临5-8%的压价与220-350 bps的毛利率挤压，伴随两融（550亿）及科创…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 5, "neutral": 0, "value": 30.2, "events": [ { "title": "年中配置应执行规则化降杠杆：卖出红区AI硬件、保留低杠杆高液性AI核心，并把风险预算转入融资占比低于2%的电力红利资产。", "summary": "截至2026-06-03，本卡支持年中再平衡，但反对一刀切清仓AI硬件。两融余额在2026-06-02出现净卖出，说明拥挤AI硬件应按强平流动性定价；建议刚性降低融资占比高于5%、换手高于20%或强平消化天数高于1天的PCB与高速连接器中小盘仓位，同时保留立讯精密等低杠杆高液性AI硬件核心。释放的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力红利资产的防御性与安全性评估：国投电力、华能水电、中国核电", "summary": "三家标的均通过卡 06 设定的两条硬线（融资占比 < 2%、经营性现金流稳定），可承接 AI 硬件释放的风险预算，但须差异化排序、不可等权。中国核电 2025 OCF 374 亿元最厚，发电量+12.98%，19 台在建机组带来成长但隐含资本开支与监管尾部风险，约 1.8% 股息率较低 [自测算]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "保留低杠杆 AI 硬件核心 + 差异化电力红利（CNNC:SDIC:华能=4:4:2），用两融余额、强平消化天数、电价、来水四项指标驱动操作，红区 AI 硬件硬性降仓不一刀切，组合级…", "summary": "年中再平衡终稿：在 100 单位 A 股策略组合中，AI 硬件低杠杆核心 16 单位（立讯精密 10 + CPO 三家 6）、电力红利 24 单位（中国核电 9.6 / 国投电力 9.6 / 华能水电 4.8，按卡 07 排序 4:4:2）、红区 AI 硬件残余 8 单位、现金 7 单位、其他核心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建全球产业周期景气度仍在加速并非泡沫见顶，但A股本地面临极高的两融多杀多挤兑风险，必须进行仓位控制及衍生品尾部对冲。", "summary": "中际旭创登顶沪深300第一权重股与Marvell单日大涨32.52%标志着AI基础设施资产登顶中美双市场。尽管全球算力与网络需求（旭创Q1利润增262%）和英伟达 endorsement 证明主升浪仍未终结，但A股本地TMT两融余额（5800亿元，旭创融资比率8.50%）呈现极端拥挤。需通过限额至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 8, "neutral": 0, "value": 13.6, "events": [ { "title": "A股红利资产正从‘全面普涨’转向‘分红质量驱动的内部再平衡’，大水电凭借FCF支撑的派息稳定性正重新确立其作为核心防御资产的霸权，而核电则被剥离为高资本开支驱动的成长弹性品种。", "summary": "2026年6月A股红利资产正经历显著的微观结构调仓。受6月中旬指数半年度调仓影响，约3400亿人民币被动资金流入电力板块。虽然‘算电协同’叙事支撑核电与水电同步获得北向资金流入，但在长线机构（如险资）内部，基于‘分红质量’的分化正在显现：资金正向自由现金流完全覆盖分红的大水电（长江电力等）回流，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源标的\"物理红利\"防御性验证：现金流稳定性与电价机制评估 工作日：2026 06 05（亚洲/新加坡） 分析师：能源行业分析师 (energy analyst) 立场：suppor…", "summary": "能源标的\"物理红利\"防御性验证：现金流稳定性与电价机制评估 工作日：2026 06 05（亚洲/新加坡） 分析师：能源行业分析师 (energy analyst) 立场：support（支持卡04把战术性资金从商业银行高股息转向\"物理红利\"水电的判断） 板块：6d8317f2 c0d6 4f27 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 06 · TMT 视角对\"物理红利\"水电防御性论断的质疑 工作日：2026 06 05（亚洲/新加坡） 分析师：TMT 行业分析师（tmt analyst…", "summary": "Research note 06 · TMT 视角对\"物理红利\"水电防御性论断的质疑 工作日：2026 06 05（亚洲/新加坡） 分析师：TMT 行业分析师（tmt analyst） 立场： deny （否定\"水电物理红利可作为商业银行高股息替代仓位\"的判断） 板块：6d8317f2 …/Res…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026年6月4日，整合外汇对冲成本与底层企业不对称债务敏感度后，A/H红利板块在重构A/H比例为70/30并向净现金的能源、银行和电信倾斜后，可获得5.56%的净股息率并实现反…", "summary": "截至2026年6月4日，在对22.5%概率的PBOC被动收紧情景及2.2%年化CNH NDF对冲成本进行压力测试后，本卡建议维持32%的红利/低波策略总权重。通过将A/H比例调整为70/30，组合层面对冲损耗降至22bps；并在红利仓内降配高负债的长江电力（降至3.2%），增配净现金的中国神华（增…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "两融、雪球与股指期货去杠杆共振时，防线必须从静态套保改为预留现金、动态降杠杆和红利锚容量配给，否则对冲会先于底仓卖出能力而失效。", "summary": "截至2026-06-06，本卡压力测试认为，A股尾部风险的核心不是缺少IM、期权或红利资产，而是底仓跌停锁死时，对冲端被保证金和盘中反弹先行强平。最新可得两融口径显示，2026-06-04沪深融资余额约2.89万亿元；若3%-6%被动退出，可能形成约867亿-1734亿元卖压。长江电力可作为红利锚…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 6, "neutral": 4, "value": 48.4, "events": [ { "title": "能源冲击越深，电力红利权重越应从火电主导切向水核主导——水电与核电凭借零燃料成本+容量电价+调峰收益是FX不敏感的反脆弱α避险红利。", "summary": "支持Card-03的电力主导红利结论。截至2026-06-07，能源冲击下A股电力板块依靠'煤电联动+容量电价+水核零燃料'三重缓冲足以承担避险红利角色。压力测试显示：在煤价升至¥1,200/吨、JKM达$25的重度情景下，水电EPS仍+12%、核电+5%、火电因容量电价(¥100/kW)对冲后仅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大水电承担利率敏感型'物理红利'防御角色，BTM出海链承担AI物理瓶颈套利的进攻角色，构成能源行业最清晰的双锚配置。", "summary": "截至2026-06-06，央行5月底降准50 bps推动10年期国债收益率跌破2.0%，叠加全国电力市场化交易占比超70%，大水电（长江电力为代表）与BTM出海链构成A股内一组互补的'物理红利'。大水电凭借水文均值回归、近零燃料成本与60%+毛利率提供类永续债现金流，股债利差由132 bps走阔至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "RMB双向尾部下哑铃不能对称加码——整体降β、红利从银行单极改为电力主导的四极结构、科技从硬件AI高拥挤切到自主可控+军工电子，电力是能源冲击下的最佳避险红利。", "summary": "针对Card-02锁定的人民币双向尾部情景（USD/CNY破位至7.00–7.10 vs 日元平仓拉升至6.60），A股哑铃组合不应等权双向加码，而应整体降β并内部再平衡。基准60/40哑铃在情景A调整为40/50（科技降15pct、红利加10pct、黄金7%），情景B回补至60/30。红利端从银…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "日元平仓冲击下，A股风格向“物理红利（大水电）”防御端与“推理通缩AI软件/出海BTM”成长端极化，规避两融拥挤的硬件踩踏。", "summary": "截至2026-06-06，针对日元套利平仓去杠杆引发的流动性冲击，我们重构了A股“极化双侧杠铃”策略。国内降准50 bps对冲将推动国债利率“破2”，拉阔红利股债利差，吸引险资FVOCI配置流入大水电等“物理红利”（如长江电力）。同时成长股内高杠杆硬件由于并网延迟与两融拥挤承压，应切换至低拥挤、毛…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-07", "supports": 1, "risks": 1, "neutral": 1, "value": 70.8, "events": [ { "title": "PJM 2026年资费重构与中国发改委绿电直连电价费率改革，导致美股与A股电力板块在BTM表后自发用电与信用质量上呈现剧烈的分化，做多零碳直供发电商与中国水电/核电巨头，做空受并网瓶…", "summary": "截至2026-06-07，美国电网并网队列积压突破2600GW，数据中心面临3-4年并网延迟。PJM 2026年资费重构征收$8.0/MWh备用费，使核电直供利润率微降至62.69%-64.79%，但相比常规上网销售的50.00%仍具高溢价；受规管电网的FFO/Debt信用比率仅修补至13.71%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "下一阶段A股最优配置是金融红利-中特估+先进制造的杠铃组合：金融18-20%作压舱，先进制造25-28%作弹性，地产与旧消费明显低配。", "summary": "截至2026-06-08，本卡片综合C1-C4结论，构建A股杠铃式顺周期配置框架：金融板块18-20%（银行/保险偏中特估与红利、券商超配获取顺周期弹性），先进制造25-28%（电力设备+军工+机械+交运装备），辅以现金流型能源/材料/公用各+1pct，低配地产-3~-4pct、可选消费-2pct…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "低碳走廊材料链优势的行业配置表达", "summary": "不应在四个选项里二选一，而应构建按信念度加权的四仓杠铃组合，把 Research note 05 的环节排序翻译成权重 [自测算]：水电/核电防御锚 40%（长江电力600900.SH、中国核电601985.SH，受 NDRC 监管豁免、10Y 国债1.72%下重估）；AI 电力基础设施/特高压 …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 5, "neutral": 0, "value": 37.2, "events": [ { "title": "绿电溢价与CEA抬升将终结A股电力红利与有色周期的同质化行情，催生“水电/核电红利成长化”与“水电铝阿尔法溢价”的戴维斯双击，同时迫使被动资金重构传统指数权重。", "summary": "截至2026年6月8日，承接上游成本测算，本卡片认为范围二纳入将引发A股公用事业与有色板块的深度估值重塑。4-6分/度的绿电溢价将使大水电/核电从纯“红利”向“红利成长”重估，而火电因面临碳成本与新能源资本开支压力面临股息缩减风险。电解铝板块将从同质化周期贝塔转向结构性阿尔法，水电铝凭借吨铝783…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在6月流动性剧震窗口，应通过“超配长债+黄金避险”降低组合Beta，并以股票内部的“35/40/25”非对称哑铃结构防御TMT去杠杆冲击。", "summary": "针对2026年6月A股2.9万亿两融压力与3333亿解禁洪峰，本报告提出“防御性风险平价”策略。建议将60%资产配置于国债（10Y收益率1.72%-1.76%）以对冲通缩与增长压力；12%配置于大宗商品（黄金$5000/盎司、铜$11500/吨）；股票仓位降至28%并执行内部再平衡，从高拥挤TMT…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在汇率7.30红线与房企信用违约双重约束下，以煤炭、电力为代表的物理红利资产凭借长协价格托底、容量电价政策增厚及净现金资产负债表，可完全抵御资产端重估风险，是低息“资产荒”下最优质的…", "summary": "在房企违约传染与汇率USDCNY 7.30防线限制降息的双重约束下，煤炭、电力等高红利板块表现出强抗重估风险特质。容量电价政策与煤电长协锁定了火电基本盈利，长江电力压力测试显示仅在极枯水年面临DPS被动下调风险。中国神华凭借700亿元净现金，展现出强抗利率敏感性。建议将红利SAA的A/H比由60/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "银行红利仓设上限后的A股策略再分配：释放风险预算应转向哪些行业或风格", "summary": "综合 Research note 01–04：银行高分红股权封顶于跨资产套息仓约30%是正确的风险纪律，但在A股账本上留下收益缺口与回报缺口。策略答案是哑铃式再分配：收益端从'银行—利率因子'分散到非银现金流红利（煤炭/能源、交运/高速、运营商、水电）加自由现金流因子与央企'HALO'红利（指数股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 7, "neutral": 0, "value": 26.6, "events": [ { "title": "SMR 2027年并网叙事泡沫证伪下的传统能源与物理红利股战术超配", "summary": "SMR 2027年并网叙事泡沫证伪下的传统能源与物理红利股战术超配 Debunking the 2027 SMR Grid Rebound: Tactical Overweight on Utilities & Physical Dividends 在商业 SMR 证实 2030 年前无法贡献 G…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SMR 2027年并网叙事泡沫证伪下的传统能源与物理红利股战术超配", "summary": "SMR 2027年并网叙事泡沫证伪下的传统能源与物理红利股战术超配 在商业 SMR 证实 2030 年前无法贡献 GW 级电力的物理事实下，AI 负荷增长所带来的电力溢价是否正完全流向传统电网与成熟核电/水电巨头？A 股长江电力（度电 0.4432 元）的绝对超额利差（+336 bps）是否使其防…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SMR 2027年并网叙事证伪触发能源资产向‘物理红利’轮动，长江电力凭借极高利差与电价溢价成为核心超配标点。", "summary": "截至2026年6月，由于监管和建设延期，SMR 2027年并网叙事已证实破裂，GW级产能需等到2030年后。AI电力需求溢价正流向长江电力等成熟水电巨头。长江电力受益于0.4432元/千瓦时的市场化高电价和+336 bps的超额利差，防御价值被系统性低估。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-09，AI电力瓶颈最优交易不是追逐SMR叙事，而是以公用事业、工业电气设备和精选材料为核心的现金流支撑型基础设施重估。", "summary": "截至2026-06-09，本卡支持上一张卡的AI电力瓶颈主线，但把策略从核电叙事扩展为A股板块轮动：超配有现金流和电网接入价值的公用事业、工业电气设备和储能链，标配精选TMT，低配尚未商业化的能源概念。关键证据包括IEA、EPRI、Goldman Sachs和中国“人工智能+能源”政策数据。组合应…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 0, "neutral": 1, "value": 73, "events": [ { "title": "公用事业产业链正处于电价市场化、核电扩张与 ETS 重启三轴重定价中，超额收益落在 003816、600803、600905、600008、1164.HK 五个低覆盖度标的，主流的长…", "summary": "公用事业产业链当前 30 日激活率仅 11.7%（24/205），叙事被长江电力 600900 一只标的吸走，但 2025-02 风光电价市场化、2026-03 十五五核电 110 GWe 目标、2026-01 ETS 重启叠加 CEA-2025 切换三大监管冲击在 12 个月内连发，买方盈利模型…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 3, "neutral": 0, "value": 53.4, "events": [ { "title": "红利拥挤度极化与公用信用传导下向出口链低拥挤纯粹价值防御切换", "summary": "红利拥挤度极化与公用信用传导下向出口链低拥挤纯粹价值防御切换 在红利拥挤度逼近历史极值、长江电力达90%且地方拖欠致公用账款增加15%背景下，红利底座应坚守水电核电，还是大举切仓至融资占比低、严重低配且有独立海外现金流的出口链龙头？ Prior related institute work to …", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红利拥挤度极化与公用信用传导下向出口链低拥挤纯粹价值防御切换", "summary": "红利拥挤度极化与公用信用传导下向出口链低拥挤纯粹价值防御切换 Utility Dividend Crowding & PPP Credit Risks vs. Low-Crowding Export Manufacturers 在红利拥挤度逼近历史极值、长江电力达90%且地方拖欠致公用账款增加15…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在红利极化与公用信用分层并存的局面下，最优解是'水电核电守正60% + 出口链纯粹价值奇兵40%'的双底座再分层，而非任何一端的极端解。", "summary": "红利底座不应在'坚守水电核电'与'大举切仓出口链'之间二选一。建议执行'守正60%/奇兵40%'的双底座再分层：保留信用免疫的大水电与核电(国网总对总结算)，退出地方信用尾部恶化的中小水电、市政燃气、供热；将原红利底座的30-40%替换为公募严重低配、有息负债占比<15%、海外营收40-50%的出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 2, "neutral": 0, "value": 60.8, "events": [ { "title": "长江电力处于后资本开支期，其70%分红承诺在极端枯水和市场化电价压力下依然具备自我造血的刚性支撑；而中国核电在千亿资本开支高峰下自由现金流持续为负，其分红极度依赖外部融资，属于‘成长…", "summary": "截至2026年6月13日，水电与核电分红红利底座呈现显著分化。长江电力处于后资本开支周期，自由现金流（FCF）极度强劲（基准情景488亿元）。压力测试显示，在15%极端枯水和电价市场化波动的压力下，长江电力FCF（265.4亿元）依然能完全覆盖70%的分红承诺（157.5亿元），红利底座坚如磐石。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核能与大型水电等基荷能源的增量供应能否填补AI数据中心带来的电力缺口？", "summary": "截至2026-06-12，本卡压力测试前序链条：核能与大型水电的增量供应在2026–2028年无法填补AI电力缺口，因为项目建设周期落在2030–2033年，结构性滞后于需求。美国数据中心需求由31 GW（2025）升至66 GW（2027），2025→2027增量约35 GW[S2]；而美国近端…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 0, "value": 68.2, "events": [ { "title": "能源成本高企与高息环境重塑A股盈利预期，配置应从传统制造与伪防御红利全面转向“上游资源价值 + 新基建能效成长”的哑铃组合。", "summary": "2026年6月中东能源冲击与美联储利率冻结下，传统“价值/成长”二元对立失效。本研究基于制造业利润断点与煤电红利破裂的结论，提出A股“哑铃策略 2.0”：大盘价值内部分化，资源型价值表现优异，而红利型价值（火电）面临坍塌；制造型成长遭遇通胀与高息的戴维斯双杀。资金将流向具备定价权的上游资源与不受油…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 5, "neutral": 0, "value": 45.6, "events": [ { "title": "在国债收益率回升与汇率防御紧平衡压力下，A股红利资产将呈现大行受益、水电压缩的K型利率传导，而高昂的离岸对冲成本将加速外资抛售内需核心资产，唯有大行红利与出海核心资产可实现+8.31…", "summary": "基于2026-06-15，本报告对人行汇率防线的传导路径实施压力测试。测算显示，30年期国债收益率从2.232%反弹25 bps将导致久期约40年的大水电红利（如长江电力）面临-11%至-14%的估值回撤，但由于收益率曲线陡峭化及NIM改善，股息率5.40%的大型国有行将逆势走强。商行存款规模达3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源板块分子端在Brent中枢$65以上即可兑现card-06的200–250bps红利定价红线，真正的下限不是分红政策本身而是Brent长期中枢，临界值$60/桶。", "summary": "在Brent $70/桶以上且OPEC+维持配额纪律的基础情景下，全球油气一体化巨头（XOM/CVX/Shell/TTE/中海油/中石化）现金股息覆盖率>1.6x，现金股息盈亏平衡油价仅$40–45/桶，足以兑现card-06提出的红利股债利差200–250bps红线，能源板块20%重估空间成立。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网（准许收益率）、核电（零燃料弹性）、头部煤电（两部制电价+长协）三链独立成立，A股公用事业瓶颈红利能转化为真实利润扩张。", "summary": "支持卡07：A股电力/电网瓶颈在2026年确实转化为电网、核电及头部煤电运营商的利润增长，并非被上游成本完全抵消。三条独立成立的传导链：电网受准许收益率框架保护、与上游能源脱钩，叠加2026年8500亿元投资计划，国电南瑞等设备商Q1新签订单+30%；核电燃料成本占比仅15-18%，对Brent …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "通过压缩固收久期、超配原油/绿电铝与高红利国企构建战术双重哑铃，并辅以10% NAV的能源/航运/汇率期权领口对冲，是规避能源与运费联合通胀尾部风险的最优配置方案。", "summary": "截至2026-06-15，海运运费与地缘能源潜在飙升构成联合通胀尾部风险（复合贡献0.6-0.8 pp）。本报告建议将大类资产组合中的固收权重降至30%并压缩久期至3.8年，配置10% TIPS；权益权重降至35%并战术轮动至A股“双重哑铃”组合（超配煤炭、原油、水电铝与国有大行）；大宗商品超配至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 0, "value": 65, "events": [ { "title": "在原料退化与B端品质下沉的500–850bp毛利压迫下，应将海天/青啤从超配切到明显低配，并将释放的约400bp仓位重构为'政策性口粮必需品+非餐饮型大众红利'的战术哑铃，板块保持中…", "summary": "工作日期2026-06-15。综合Research note 01–05结论：Research note 05给出的500–850bp原料/产品结构退化压力，已将Research note 02基于15–60bp渠道漏损建立的安全垫击穿10倍以上，原'做多干净成本传导'防御逻辑被否证。建议将海天(…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "该哑铃确比受损白马更防御，但其回撤边界以利率环境为条件——两腿共享久期，在利率波动＋再通胀尾部会同跌，缺乏显性久期对冲时韧性不足。", "summary": "本卡对 Research note 06 的“政策性口粮（金龙鱼300999、北大荒600598、中粮系）+非餐饮型红利（长江电力600900、华能水电600025、中国移动600941、大众药店）”约400bp哑铃做组合级压力测试。结论：哑铃确比受损的海天/青啤更防御，但其防御性以利率环境为条件…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 1, "risks": 3, "neutral": 1, "value": 71.4, "events": [ { "title": "两融折算率下调引致的350亿元刚性去杠杆将重创高贝塔科技硬件板块，推动资金向筹码结构极净、收益率稳定的‘物理红利’与‘成本通缩’AI SaaS板块非对称旋转。", "summary": "截至2026年6月17日，本报告评估了两融折算率从0.6下调至0.4对A股两融规模和市场结构的影响。券商分批落地该政策将引发250亿至450亿元（中枢350亿元）的刚性融资收缩。科技硬件（电子/计算机）板块融资余额高达5608.81亿元，面临50亿至112.5亿元的平仓卖压。在负Gamma共振下，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在算电协同的西迁叙事中，由于储能调峰是真正的物理约束与成本重灾区，A股配置应从‘总量逻辑’转向‘可靠性与效率逻辑’，首推储能与液冷，并对绿区IPP进行‘基荷优于间歇’的分化交易。", "summary": "基于此前针对西部绿区电力承载力的压力测试，我们综合提出A股算电协同主题的配置策略：储能调峰 > 液冷 > 特高压设备 > 绿区IPP。西迁逻辑中的核心瓶颈并非铭牌装机量，而是可靠容量的转换。因此，储能与火电灵活性改造享有最高Alpha；液冷作为数据中心PUE降本刚需紧随其后。特高压设备虽具高确定性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力交付物理瓶颈尚未在A股TMT板块充分定价，高估值硬件标的面迎15%-45%估值去化风险，应当战略重构组合并转向物理电网设备、上游矿山及防御性清洁能源公用事业。", "summary": "本报告基于电网与BTM直供电物理硬约束验证的12-24个月算力基础设施部署气穴期，指明A股算力硬件与上游材料板块存在估值脱钩和两融踩踏风险。建议在气穴期内进行风格轮动，超配出口导向变压器和上游铜铝，增配防御性清洁公用事业（长江电力、中国核电）。定量测算表明，优化后的战略组合可将承压收益率提升7.8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "非对称旋转逻辑在算力产业链层级成立，承接力依基建>SaaS>硬件分层分布，但承接资金到位滞后于卖压释放，硬件去杠杆引发的指数级回踩仍不可避免。", "summary": "支持 Research note 02 非对称旋转逻辑在算力产业链层级成立。硬件层（AI服务器ODM/光模块/PCB/国产芯片）融资盘集中、贝塔高，是卖压震中；基建层（大水电、核电、特高压、IDC/算力REITs）凭借物理资源稀缺与现金流稳定构成最佳承接位；SaaS层（金山办公、用友、泛微）受益A…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 9, "neutral": 0, "value": 25.8, "events": [ { "title": "在美股AI电力赢家与科技成长股存在高Beta重叠且容易被SPCX同步去杠杆的环境下，大类资产配置应向低贝塔的A股水电红利（长江电力）倾斜，并配置核电（中国核电）对冲干旱风险，构建收益…", "summary": "基于2026-06-17，本报告压力测试了SPCX挤压环境下，AI电力稀缺因子与水电红利的配置效率。美股核电IPPs（CEG/VST）对NDX的Beta已飙升至1.08-1.12，存在严重的Beta和动量重叠风险，在去杠杆中易遭受-25.0%的同步回撤。A股水电（长江电力）对NDX的Beta为-0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘溢价消退将引发A股红利资产内部的结构性切换，资金将从“涉能”周期红利流向免疫大宗商品通缩的纯内需防御资产（水电、家电、白酒）。", "summary": "原油价格因美伊和平协议跌破关键支撑位，标志着地缘政治溢价消退。A股能源红利资产（如煤炭）面临盈利与估值双杀风险。建议在红利组合内部进行“去能化”轮动，将资金切换至水电（长江电力）及高股息消费（家电、白酒）等纯内需防御资产，以规避输入型通缩风险并锁定稳定现金流。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力稀缺是SPCX拥挤风险的有效对冲，但美国AI电力赢家本身已部分拥挤，组合应超配稀缺因子、低配拥挤表达，并用水电红利与成本归因规则做底仓。", "summary": "2026-06-17：本卡综合Research note 01-04，结论是AI电力稀缺资产仍可作为SPCX拥挤风险的组合对冲，但必须拆成结构性对冲与拥挤化对冲两类。保留有合约支撑的核电IPP、600900式水电红利和具备大负荷成本归因的公用事业；但CEG、VST、TLN等美国AI电力赢家已带有成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "验证电力红利资产在极端市场波动下的防御价值", "summary": "压力测试长江电力（600900.SH）的防御价值，结论为「有条件支持」Card-01。2025年归母净利341.67亿、+5.14%，70%派息承诺延至2026-2030。136号文针对新能源、不及水电，反使水电相对更具防御性。自测算显示：即便严重多年期干旱（发电量−12%），股息率约2.81%仍…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 2, "neutral": 0, "value": 69.2, "events": [ { "title": "上游能源红利逻辑正被强制切换为'现金流稳定'主导，但估值中枢不能维持当前水平——煤炭可作过渡仓但需-10%中枢下移，油气从核心降级为卫星仓，真正受益者是公用事业型物理资产。", "summary": "油价回落至70-80美元区间剥离了战争溢价，IEA下调2026年全球需求增速至+0.7 mb/d。上游红利逻辑被强制切换为'现金流稳定'主导，但估值中枢不能简单平移：A股三桶油/煤炭龙头股息率虽仍6-8%，盈利-分红覆盖倍数从2022年~2.0×压缩至~1.3×。压力测试显示韧性排序为煤炭>油气>…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在长端利率温和上行（路径A）时，T合约对冲优于TL，而在单边大额上行（路径B）时TL更佳；相比线性国债期货，零成本看跌价差领口期权提供正Gamma与多因子风险防范，是对冲红利大水电久…", "summary": "本卡定量评估了长江电力权益久期对冲工具。在10Y CGB升至2.00%路径下，因曲线熊平，T合约相比TL更为有效；在升至2.20%路径下，TL久期匹配度更高，需51手合约。国债期货套保面临严重的因子基差风险、交割券切换和移仓成本（TL年化达20-30bp）。相比之下，6个月期看跌价差领口期权（K1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 2, "neutral": 1, "value": 73.6, "events": [ { "title": "若WTI油价跌破75美元（布伦特78美元），沙特将触发暂停每月18.8万桶/日增产的防御机制，OPEC+配额纪律短期依然成立，真正的产量价格战门槛在70美元下方。", "summary": "阿联酋5月退盟后，OPEC+防守价格的重担落在核心7国。若油价（WTI）跌破75美元，由于触及沙特财政平衡红线（布伦特78美元），沙特将启动“暂停增产”机制防守油价。因此WTI在70-75美元区间减产协议不会瓦解，真正的价格战门槛是低于70美元并伴随公开作弊。组合建议向长江电力（600900.SH…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在2026下半年AI算力物理瓶颈与去估值压力下，通过55:45的TMT结构优化并切入10%的A股高股息红利资产，能将组合总贝塔压降至1.732，并利用红利端-0.8%的负边际风险贡献…", "summary": "基于2026-06-20，本报告定量评估了TMT 55:45重构与切入8%-10% A股红利对冲组合。测算表明，在AI算力瓶颈与电力死锁的承压情景下，10%的A股红利对冲（长江电力、中国神华等高股息股）成功使组合总贝塔从1.990降至1.732，10天99% VaR从17.06%压降至14.03%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股红利杠铃可以继续作为TMT组合防御对冲，但必须把中际旭创式进攻卫星视为强平凸性来源，而不是流动性缓冲。", "summary": "截至2026-06-20，本卡对A股红利与TMT进攻端做微观流动性压力测试。结论是：Research note 05的A股红利杠铃方向仍可支持，但执行容量必须收紧。长江电力更适合作为防御锚，当前可得融资余额占自由流通市值代理值约4.52%，但压力可执行流动性下退出可能拉长至约11.5个交易日。中际…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 2, "neutral": 0, "value": 72, "events": [ { "title": "能源波动 regime 下的跨行业配置与 A股/科创硬科技定价", "summary": "本卡综合 Research note 01–05：电力与电网接入已成为整条 AI 价值链的约束性稀缺投入。在布伦特回落至约80美元、发达市场同步鹰派的 regime 下，正确做法不是削减硬科技，而是沿杠铃重塑风险预算。提出五桶矩阵：超配电力/电网赋能者（国家电网2026–30约4万亿、核电10机组…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面向多年期AI电力资本开支超级周期的配置：刚性容量稀缺确立后的板块配置与风格含义", "summary": "本卡把四环串联瓶颈（封装→电网/变压器→刚性发电）翻译成配置地图。关键事实：四个时钟串联，缓解任一环只把约束移向下游，不压缩通电总时间，故稀缺是结构性、多年期的，而非周期性。配置：超配机电卖铲人（电网/变压器，全球缺口约30%、中国占产能60%、出口+36%、订单排至2027）、刚性发电（核电/铀…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 1, "risks": 10, "neutral": 0, "value": 33.4, "events": [ { "title": "在能源价格常态化预期下，红利配置应向清洁电力转移，但须通过煤电组合对冲水电枯水期现货买电风险，并警惕规制电网在AI负荷爆发期的信用评级恶化与表后流失风险。", "summary": "截至2026-06-21，本研究支持将组合红利仓位从周期性煤油向公用事业转移的逻辑。水电（长江电力）在严重枯水（-15%）情景下面临现货回购惩罚，净利润下降33.5%且面临29.7%派息削减风险；核电（中国核电）受益于煤电联动，电价上浮5%可提升利润15.88%，并捕获5.8-8.4分/kWh的碳…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在美联储沟通革命引发全球多资产波动率骤升与成长股多重估值压缩背景下，A股应坚定执行超配高电力上网电价防御韧性（水电/核电）与低配高两融风险成长板块的‘防御端杠铃策略’。", "summary": "美联储取消前瞻性指引与通胀预期上修打破了传统的股债负相关性，迫使风险平价模型被动去杠杆和抛售资产。压力测试显示风险平价在当前宏观环境下显著跑输，而战术性杠铃策略表现出强劲的抗波动韧性。全球收益率上行传导至A股，导致科创成长股面临多重估值压缩及两融强平风险，资金加速向水电、核电等红利重估板块切换；同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源红利支持35%红利股防御逻辑，但最佳表达是12-15个百分点的政策现金流与供给风险对冲，而不是全仓油煤周期押注。", "summary": "截至2026-06-21，本卡支持在35%红利股袖套中配置12-15个百分点能源相关红利资产。依据是10Y国债收益率1.73%、中证能源股息率3.86%、中证红利股息率4.49%，以及央企市值管理和分红政策对大型能源央企的约束。中国神华2025年分红率79.1%，中国海油2025年派息608.38…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力瓶颈下的跨板块配置：半导体、AI服务器、电网设备、绿电与红利公用事业", "summary": "综合卡片01-04，将AI电力瓶颈定性为局部取电速度而非全国电源缺口，由此给出A股再配置：把权重从定价完美、拥挤的算力栈顶部（前道半导体98.7分位PE、纯AI服务器贝塔）旋转至压缩‘通电机柜时间’的物理交付使能者。超配电网设备（国网十五五4万亿、+40%，最高确信）、液冷（渗透14%→40%，战…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 1, "risks": 2, "neutral": 1, "value": 85.8, "events": [ { "title": "在海风开发商回款账期延长与物理确认时滞下，A股对东方电缆（高负债杠杆压制）与亚星锚链（交付时滞真空）进行了估值收缩定价，传统的成长重仓策略失效，向具备即期现金流变现能力的水电与海运红…", "summary": "本报告对开发商账期拉长下东方电缆与亚星锚链的流动性与物理交付折价进行了A股定价逻辑压力测试。分析表明，东方电缆重资产、高资金占用模式受制于垫资压力，在重度承压下经营现金流转负至-31.73亿元，导致PE估值中枢被压缩至14.5倍；亚星锚链虽现金流安全，但142天物理交付验收时滞使其财务表现呈业绩真…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在公用事业超级资本开支面临股权稀释及大选年监管时滞的双重制约下，跨资产配置应调降拥挤的美股公用事业权益，通过公用事业IG高级债捕获拓宽的信用利差，同时布局低相关的A股水电红利（长江电…", "summary": "基于2026-06-22，本报告综合研判了受监管公用事业费率基数超级周期下的跨资产布局。公用事业1.3万亿美元资本开支带来严重的股权二次发行稀释，且大选年36州州长竞选引发监管机构费率抵制，惠誉调降行业展望至“恶化”。PJM 2027/28容量价暴涨至333.44美元/MW-日，零售电费同比上涨1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力价值链中的\"成期权\"：核电、水电及SMR对公用事业估值重塑的影响", "summary": "本卡支持卡08的\"类债→成长\"切换，但收紧为一次三层分部拆解：每家大型公用事业应拆成（i）稳态年金（类债DDM）、（ii）已签20年期算力PPA（长久期年金）、（iii）稀缺电力看涨期权（实物期权溢价）。证据：PJM容量电价从约29升至329.17美元/MW·日FERC上限[S5]；中国核电在运2…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "6月调样后A股电子与红利板块在创新高的两融杠杆与回流 of 北向外资共振下依然维持净买入，所谓的规模差走阔实为短期调样调仓与动量噪音而非真实资金撤离，但历史高位杠杆要求通过跨资产多元…", "summary": "截至2026-06-23，6月调样后A股规模差异并非由真实资金撤离驱动。月中两融余额大幅反弹915.10亿元，于6月18日触及29655.12亿元历史新高，电子板块获430亿元融资流入；6月22日北向交易额达4821.77亿元，表明内外资正共振流入硬科技与红利板块。但高杠杆下拥挤度高企。压测显示，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 8, "neutral": 0, "value": 40.2, "events": [ { "title": "清洁能源杠铃应按等风险贡献而非资本权重构建（短债占约70%资本、各分支均担25%风险），并外挂年化约60bp Theta drag的日元套利平仓尾部对冲——其代价应以全天候夏普（0.…", "summary": "基于2026-06-23，本卡将首席策略师的防御杠铃论点量化为等风险贡献（ERC）预算。因短端国债波动率仅约2.5%、公用/红利股为16-20%，25%等风险贡献下资本权重约为长江电力11%、中国核电8.8%、大行/电信红利10%、短端国债70%，组合波动率约4.8%、基准夏普0.44。真正的剩余…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在重型燃机与电网基建多年配给约束导致算力交付延迟2年的情景下，电费成本暴增将云巨头EBIT利润率蚕食150-300BP，迫使科技板块ERP走宽70BP并引发理论PE中枢下调16.77…", "summary": "基于2026-06-23，本报告评估了由燃机档期、精铸产能、已认证合金、变压器和气源许可共同形成的多年配给市场下科技股的估值重定价。燃机及电网配给将算力激活推迟1.5至2.5年，迫使2026-28年AI资本开支增长下修10%至15%。表后燃气+BESS电价升至$0.220/kWh，单卡GPU电力O…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "容量电价与稳定性溢价落地下的电力公用事业与电网设备板块重估空间与风格轮动路径", "summary": "本卡支持并综合前四卡：算力＋电气化＋容量电价硬化现金流，使公用事业从“类债券”向“成长公用”重估。结论：重估真实但不对称。硬基荷（水电/核电）的倍数重估最大且仍在前方——以川投PE约15.3、隐含(r−g)≈2.9%为锚，(r−g)压缩40–80bp对应约15%–35%合理估值上行[自测算]；电网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供给冲击不是通过成本、而是通过贴现率侵蚀水电/核电的类债券溢价（成本免疫、收入端反受益），却同时通过成本＋久期双杀电网设备成长——成长腿才是本尾部最软的靶子。", "summary": "截至2026-06-23，本卡对卡片08的能源供给冲击＋利率上行尾部做电力/电网压力测试，结论为有条件成立且直觉相反：两类溢价都被侵蚀，但通道不同。水电/核电类债券溢价成本免疫（燃料为水/铀，冲击反而抬升其售电价），仅被贴现率侵蚀——以川投15.3倍PE、(r−g)≈2.9%为锚，+60/+80b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 4, "neutral": 0, "value": 65.6, "events": [ { "title": "在重型燃机与电网多年配给重塑科技股估值的情景下，超配长江电力等在岸清洁基荷资产的风险平价组合可锁定确定性溢价，且通过MXN/JPY外汇数字期权可有效锁定日元平仓海啸的尾部下行风险。", "summary": "本报告针对电网并网延迟与重型燃机5-7年超长交期约束，科技成长板块估值重塑，以及美日利差收窄导致的日元套利平仓风险，构建了长江电力、中国核电、大行/电信红利及短端国债组成的清洁能源杠铃组合。通过波动率风险平价（ERC）和高股息倾斜自定义风险平价进行了权重优化。为对冲日元套利平仓引发的流动性冲击，测…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "重型燃机与电网多年配给格局下跨资产清洁能源杠铃组合的风险归因、权重重构与全天候对冲成本测算", "summary": "重型燃机与电网多年配给格局下跨资产清洁能源杠铃组合的风险归因、权重重构与全天候对冲成本测算 在重型燃机与电网多年配给导致科技股估值重塑且在岸高股息资产重要性上升的情景下，如何利用全天候或风险平价（Risk Parity）模型重新为长江电力、中国核电、大行/电信红利以及短端国债分配风险预算？在计入了…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "液冷驱动的超高功率密度对电网配电侧建设与能源供应响应力的挑战", "summary": "经压力测试：在受监管公共电网层面答案是“不能”，园区层面仅对提前占位者有条件“能”。密度对电网不是关键变量——100MW园区无论20kW还是140kW机柜都取电约100MW[自测算]，密度只是把约束下移到馈线/变电站/变压器并加剧负荷波动（已触发NERC突发失负荷预警）。时间错配是结构性的约5.6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "防御杠铃组合中大行、水电、现金与债券的再平衡比例", "summary": "本卡综合 Research note 3–5，给出防御红利仓位（占组合40%）的五段式再平衡向量：大行由25%降至10%（转出15个百分点），转入收费公路+7、短久期债券+5、现金+3，水电维持15%但内部≥80%倾向长江电力一类低市场化敞口。测算显示：基准股息率由4.00%降至3.51%（牺牲约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 1, "neutral": 0, "value": 85, "events": [ { "title": "在限制性利率与结构性电力成本溢价下，市场对‘可交付性约束’的忽视将导致AI芯片链面临重定价，资金应向电网设备通道、独立发电商及高股息红利价值风格集中。", "summary": "截至2026-06-25，本卡合成2026年能源冲击与电力稀缺对大类资产及A股/全球股票轮动的影响。电力并网瓶颈与大变压器缺口推升期限溢价及美联储政策利率（更高更久），但实际r*仅温和上升+25至+50bp。资产配置应实施哑铃结构：超配电网设备与燃机OEM（受益于全球变压器缺口及交期溢价）、超配红…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 8, "neutral": 1, "value": 47.4, "events": [ { "title": "电力电网及公用事业红利资产的配置逻辑验证", "summary": "支持第01张卡片的防守再平衡判断，并验证其机制：本轮去杠杆是两融融资拥挤的电子板块冲击（A股两融2026-06-23破3万亿元），而受监管公用事业由保险/养老等长钱持有、融资拥挤度低，处于压力源的对立端，是真正避风港而非平行泡沫。需求端1-5月用电+3.4%、互联网数据服务用电+26%给出AI电气…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "保留12%清洁+4%能源杠铃，但A股表达必须从宽基清洁贝塔转为受限的电网/储能/运营商组合，并用能源对冲抵消价值和通胀尾部而非追高红利能源。", "summary": "截至2026-06-26，本卡综合前序12%清洁基础设施+4%能源尾部对冲，但指出A股执行并非因子中性。清洁袖主要暴露于成长、资本开支、低股息和动量，尤其电网设备、电池、储能、光伏存在追涨和集中度风险；能源袖提供价值、高股息和通胀/霍尔木兹尾部保护，但中证能源前五权重约61.0%，不能无限扩张。建…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股市场结构约束下的清洁基础设施与能源对冲执行方案", "summary": "截至2026-06-26，本卡把Card03杠铃(12%清洁+4%能源)与Card04子仓上限落地为可成交的A股方案。核心结论：约束不难，难在表达载体可得性高度不对称——电网有唯一高流动纯标的(159326,约¥343.7亿,特高压权重约66%)；电池有百亿级但偏电动车(159755,约¥158.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "136号文市场化电价下绿电、核电、水电和储能运营商现金流敏感性", "summary": "对运营商做三降压力测试：绿证价格、利用小时、市场交易电价同时下行。2025年业绩已是实盘验证，横截面决定性。保留满仓——长江电力(600900)水电、中国核电(601985)核电：电价下行年净利仍增(+6.17%/+6.00%)，水电近零边际成本+65.79%毛利、来水定产不受限电；核电>7500…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 5, "neutral": 0, "value": 63.8, "events": [ { "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源与电力板块（长江电力/中国海油）的防御性估值与政策风险压力测试", "summary": "对卡片09把半导体腾出资金转入长江电力(600900)与中国海油(0883)防御的安排做压力测试，得出分裂裁决。电力/长江电力基本面不过热（零燃料成本、70%分红锁定至2030），但作为工具已成拥挤的26.8年久期债券替代品：股息率压至3.5%、对国债利差约2pp处历史高位、内在价值≈市值、贴现率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "1.3%低利率下，红利全天候组合应从追逐水电股息利差转向水电质量、规制煤电容量收入、国债久期与黄金/煤炭事件对冲的动态杠铃。", "summary": "2026-06-27：本卡综合前序链条，认为1.3%国债情景不是加仓红利资产的单向信号，而是要求把红利全天候组合从“长久期股息代理”改成“规制现金流杠铃”。建议正常环境水电22%-26%、煤电14%-18%、国债30%-34%、黄金/煤炭12%-16%、现金6%-10%；若1.3%国债叠加枯水/现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高两融TMT龙头下跌10-15%与折算率降至0.4的共振将引爆1372.7亿至2229.1亿元的强平与还款抛盘，跌停锁定将溢出25%的负Gamma抛压至其他流动性资产；资产配置应坚决…", "summary": "截至2026-06-27，本报告针对A股高两融TMT龙头下跌10-15%与券商提高担保折算率（从0.6降至0.4）的双重冲击，定量压力测试了被动卖出量与微观流动性。测算表明，在A股两融总余额3.015万亿元、核心TMT占比15.4%（4670.0亿元）的背景下，10%下跌（温和压力）将导致平均CM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 3, "neutral": 0, "value": 77.2, "events": [ { "title": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费成本转嫁挤压。", "summary": "基于国产芯片2.94倍能效赤字拉长通电时间至18个月的物理瓶颈，本报告系统性量化了A股电力设备板块（特高压、大变压器）的估值扩张机制与低利润 traditional TMT 的电价挤压。电网容量受限与十五五规划共振，支持了特高压订单可见度与大型变压器出海高溢价，特别是在国内外硅钢5.1倍巨大价差的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "局部中小银行信用缺口下的A股银行红利耐久性与高股息内部轮动", "summary": "卡08分离出的“额度沉睡”机制（CET1上限+NPL触发+低于1.40%净息差）不仅是中小银行信用风险，更是撕裂A股红利复合体的力量。重排结论：①大行红利居顶——分红不依赖沉睡额度，约4.35%股息率、对10年国债（约1.72%）257bp利差、0.5–0.7倍PB已计入坏消息，持有/超配；②公用…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "summary": "截至2026-06-28，本卡对“AI负荷增长能否支撑公用事业高分红”作压力测试：结论是精选组合层面可以，板块整体不能机械外推。AI数据中心用电增长真实，但也同步拉动电网、发电和储能资本开支。水电、核电、容量电价支持的火电及低杠杆综合电力运营商仍可维持分红；风光开发商受消纳、电价和项目资本开支制约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 5, "neutral": 0, "value": 66.6, "events": [ { "title": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电…", "summary": "本报告对AI算力激增带来的电力电网物理瓶颈（变压器交期120-210周[S2]、并网排队4-7年[S1]）对A股转型龙头的利润率传导进行压力测试。结果显示，电网扩容滞后充当了新能源装机的“限速阀”，压制了光伏/电池销量增速；同时，容量电价[S8]和碳成本上升[S19]增加了高能耗制造环节电费开支，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在红利内部由周期性原油向刚性物理红…", "summary": "本报告针对2026年三季报TMT硬件估值折价预期，校准了全球全天候和风险平价模型。战术防御性风险平价（TDRP）将AI成长股风险预算由25.0%下调至8.0%，防守红利上调至40.0%，以对冲55.0%的TMT增长溢价折价。模型显示，电力配给约束下算力效率的TCO优势在24/7清洁能源下扩大至9.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15.36倍AI潜在电力需求，算力扩…", "summary": "截至2026-06-30，本研究对Jevons悖论激发的15.36倍潜在AI用电需求（相当于727.6 GW新增平均负荷或6,374 TWh年用电）进行了极限物理压力测试。结论表明，在2030年前，由于北美变压器交期拉长至48-60个月、特高压直流送电网架高壁垒（全球需新建73条通道）以及取向硅钢…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-5cb77a9f52/", "networkHref": "/investor/research-network/#thesis-5cb77a9f52" }, { "id": "thesis-e4a70ed58a", "source": "auto_topic", "title": "在布伦特油价从约108美元回落至约80美元、核心CPI仍受控于2.9%的背景下，5月4.2%的整体通胀冲击主要是暂时性的（允许美联储维持…", "titleEn": "Practical research topic: 在布伦特油价从约108美元回落至约80美元、核心CPI仍受控于2.9%的背景下，5月4.2%的整体通胀冲击主要是暂时性的（允许美联储维持…", "lane": "宏观利率与汇率", "laneEn": "宏观利率与汇率", "laneHref": "/investor/lanes/lane-cdafce64/", "investableFocus": "", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM", "APO", "COP", "CAT", "MAR" ], "coreView": "该主线来自 4 条研究记录的标题、问题和摘要聚类，时间跨度 2026-06-19 至 2026-06-28；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-04-27", "lastSeen": "2026-06-29", "supports": 5, "risks": 135, "neutral": 20, "analystCount": 11, "sourceTypeCounts": { "分析协作": 5, "研究笔记": 146, "研究线程": 7, "研究主题": 2 }, "topTerms": [ "美联储", "CPI", "美元", "能源", "油价", "允许美联储维持", "日元", "全球二次通胀预期退坡", "锁传导", "Russell", "Warsh", "小盘创纪录主要是" ], "whatChanged": [ "2026-06-29: 2026年石油供应中断对全球能源品种及炼油产能的冲击评估", "2026-06-29: 霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "2026-06-29: 2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "2026-06-29: 在霍尔木兹海峡中断（油价均值105美元）与重开（回落至72-79美元但存在60天到岸刚性）下，美国整体CPI分别面临~82.3 bp上升与~78.6 bp回落的非对称传导，组合应在中断期压降能源权重至10.03%并锁定净Delta在0.66，重开后降至8.0%底限以规避下行…", "2026-06-28: 6月点阵图不是简单过时，而是可能“对得不是理由”：能源理由过时，AI/电网与折旧通道支持更高更久，但不足以单独证明立即加息。" ], "practical": { "sectorLane": "宏观利率与汇率", "sectorLaneEn": "宏观利率与汇率", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM", "APO", "COP", "CAT", "MAR" ], "primaryStocks": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" } ], "instrumentTags": [ "TLT", "iShares 20+ Year Treasury Bond …", "IEF", "iShares 7-10 Year Treasury Bond…", "UUP", "Invesco DB US Dollar Index Bull…", "FXY", "Invesco CurrencyShares Japanese…", "GLD", "SPDR Gold Shares", "XLE", "Energy Select Sector SPDR" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 98 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 11, "recentEvents": 98, "supportRatio": 0.03, "riskRatio": 0.84, "conflictScore": 95.4, "evidenceDepth": 100 }, "score": 67.3 }, "events": [ { "date": "2026-06-29", "title": "在霍尔木兹海峡中断（油价均值105美元）与重开（回落至72-79美元但存在60天到岸刚性）下，美国整体CPI分别面临~82.3 bp上升与~78.6 bp回落的非对称传导，组合应在中断期压降…", "summary": "截至2026-06-29，分别在霍尔木兹海峡持续中断（情景A）与谈判重开（情景B）下对布伦特路径建模：A下油价高企于100-120美元，B下纸货崩至72-79美元，但到岸价受60天扫雷物理延迟制约。在0.25的CPI传导系数下，A推高美国CPI同比~82.3 bp阻碍降息，B拉低整体CPI~78.6 bp并释放政策约束。对冲策略上，A下能源配比需压降至10…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "CPI", "美元", "能源", "油价" ], "href": "" }, { "date": "2026-06-29", "title": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "summary": "对中国，2026年油价/柴油冲击既是被缓冲的进口税、又是战略加速器，且净额上比卡片01–02的美欧日集团损害更小。三大事实：（1）进口税真实但落在PPI而非CPI——5月PPI同比+3.9%（2022年7月来最快）、CPI仅+1.2%，疲弱内需把冲击吸收为生产者利润压缩；（2）进口税被三重缓冲——全球最大约14亿桶战略储备、作为受制裁油折价买方、疲弱需求；…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "CPI", "美元", "能源", "油价" ], "href": "" }, { "date": "2026-06-29", "title": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "summary": "截至2026-06-29，支持卡片01–03的逻辑链并延伸至终端利率与全球流动性维度。美联储6-17维持3.50%–3.75% [S2]、SEP 2026中值3.8% [S3]；新主席Warsh首场FOMC点阵图从降息翻转为加息，9名官员预期加息、年内加息概率50.4%、市场定价10月加息，反应函数转向'约束性务实主义' [S14]。情景A（布伦特约105…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "美联储", "CPI", "美元", "Warsh" ], "href": "" }, { "date": "2026-06-29", "title": "2026年石油供应中断对全球能源品种及炼油产能的冲击评估", "summary": "支持卡片01“油价持久性是枢纽”的判断并加以锐化。霍尔木兹2025年日均输送约2000万桶/日（原油1500万+成品油550万），全球闲置产能几乎全在海峡之内，绕行管道仅7-8mb/d（不足正常一半），规模化对冲不存在。冲击非线性传导：3月约1200万桶/日无法运抵亚洲、其中800万为中硫油，亚洲炼厂被迫改炼轻质低硫，柴油/航煤减产，4月中间馏分供应降1.…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 20.5, "matchedTerms": [ "美元", "能源", "油价" ], "href": "" }, { "date": "2026-06-28", "title": "美联储6月鹰派转向与能源冲击逆转的错配:点阵图是否已经过时", "summary": "2026年6月17日FOMC维持利率于3.50%–3.75%,但点阵图转鹰、抹去年内降息并将2026年底中值上调至3.8%,隐含至少加息一次,期货一度定价10月加息。触发因素是美伊冲突引发的能源冲击:5月整体CPI同比4.2%、能源同比23.5%。但该冲击正快速逆转——布伦特从120美元上方跌至约72美元,霍尔木兹有条件重启,而核心CPI仅2.9%且环比降…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32.25, "matchedTerms": [ "美联储", "CPI", "美元", "能源", "油价", "点阵图是否已经过时" ], "href": "" }, { "date": "2026-06-28", "title": "能源价格的崩溃通过被动推升实质利率（188-264个基点）对实体经济造成隐性紧缩冲击，显著放大了美联储在Warsh新主席政策框架下过度紧缩的政策失误尾部风险。", "summary": "本卡片分析了能源价格向通缩力量转变对美联储的二次紧缩冲击。霍尔木兹海峡复航导致油价跌至70美元区间，全球供需过剩凸显。由于能源对核心通胀的传导偏弱，核心CPI仍具粘性，导致美联储在Warsh主席新政下固守限制性利率。这将推动实质利率较5月被动飙升188至264个基点，显著放大过度紧缩的政策失误尾部风险，迫使市场进行鸽派重定价并调整资产组合。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.75, "matchedTerms": [ "美联储", "CPI", "美元", "能源", "油价", "Warsh" ], "href": "" }, { "date": "2026-06-28", "title": "2026年6月美联储、欧央行、日本央行同步转鹰，却在一场正在消退的伊朗战争油价冲击中收紧政策", "summary": "2026年6月，欧央行（+25bp至2.25%）、日本央行（+25bp至1%，1995年来最高）与美联储（按兵不动但点阵图从降息翻转为加息）在一周内集体转鹰，导火索是伊朗战争与霍尔木兹海峡中断推动的油价冲击。美国5月CPI同比4.2%创三年新高，但全由能源（同比+23.5%）驱动，核心仅2.9%。然而布伦特已从114美元峰值回落至约80美元，海峡有条件重开…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "美联储", "CPI", "美元", "能源", "油价" ], "href": "" }, { "date": "2026-06-28", "title": "基准配置应优先通过久期表达去通胀与政策失误风险，权益只做选择性低贝塔，商品低配整体贝塔但保留油价重上100美元的显式尾部对冲。", "summary": "2026-06-29，本卡综合并支持上一卡“央行在消退油价冲击中过度收紧”的判断，但把它转化为更可执行的大类资产配置：久期从35%中性升至45%，权益贝塔从45%降至35%，广谱商品从15%降至12%，现金升至8%。依据是美国5月CPI 4.2%中能源同比23.5%、核心仅2.9%，而美联储仍维持3.50-3.75%并上修2026通胀预期。油价尾部不能忽视…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "美联储", "CPI", "美元", "能源", "油价" ], "href": "" }, { "date": "2026-06-28", "title": "美联储2026年6月鹰派转向：新主席Warsh首会下点阵图翻转为加息，叠加中东能源冲击", "summary": "2026年6月17日FOMC维持利率3.50%–3.75%，但点阵图从隐含降息翻转为隐含加息：2026年底利率中值由3.4%升至3.8%，18人中17人判通胀风险偏上行，整体PCE预测上调至3.6%、核心3.3%。新主席Warsh以「五年失误必须纠正」的信誉优先基调定调，2年期收益率当日跳升约16基点、美元走强、股市下挫。但支撑这套预测的能源脉冲正快速逆转…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "美联储", "美元", "能源", "油价", "Warsh" ], "href": "" }, { "date": "2026-06-28", "title": "美联储6月鹰派转向与能源冲击逆转的错配:点阵图是否已经过时", "summary": "美联储6月鹰派转向与能源冲击逆转的错配:点阵图是否已经过时 在布伦特已从120美元上方回落至约72美元、核心CPI降温至2.9%的背景下,美联储6月点阵图的加息倾向是否已经过时?10月加息与年内维持不变/重新转向降息,哪条路径更可信?", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "neutral", "supportLabel": "观察", "score": 26.25, "matchedTerms": [ "美联储", "CPI", "美元", "能源", "点阵图是否已经过时" ], "href": "" }, { "date": "2026-06-28", "title": "渐退的中东能源冲击与新主席 Kevin Warsh 治下美联储鹰派转向之间的碰撞", "summary": "2026年6月17日美联储在新主席Kevin Warsh首秀上维持利率3.50%–3.75%不变（12–0），却将2026年点阵图中位数由3.4%上调至3.8%、9名委员暗示加息、并取消前瞻指引[S1][S7]。但驱动5月CPI升至4.2%（2023年4月来最高）的能源供给冲击已在反向回落：随霍尔木兹重新开放，布伦特从逾120美元跌至约72–80美元，海湾…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.25, "matchedTerms": [ "美联储", "CPI", "美元", "能源", "Warsh" ], "href": "" }, { "date": "2026-06-28", "title": "美联储政策路径不确定性对USD/CNY及利差交易的潜在冲击", "summary": "截至2026-06-29，本卡从外汇侧压力测试鹰派加息尾部。起点已承压：USD/CNY约6.81、近一个月低位，市场对联储9月加息定价70-75%，中美10年期利差已深度倒挂约-259bp（中国1.79% vs 美国4.38%）。鹰派误判会加深倒挂、推升双边汇率，但幅度可控：一次加息USD/CNY升至6.92-6.98、倒挂-285至-305bp；两次加息…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "美联储", "美元", "能源", "日元" ], "href": "" }, { "date": "2026-06-27", "title": "美联储2026年6月鹰派转向：供给冲击型通胀下的滞胀压力测试", "summary": "2026年6月17日FOMC连续第四次维持利率3.50%–3.75%（12–0全票，Warsh首会），但点阵图把2026年末中值从约3.4%上调至3.8%，抹去年内降息、隐含加息，2026年PCE预测自2.7%上修至3.6%，18人中17人视通胀风险偏上行[S1][S3][S4]。驱动力是供给冲击而非需求：5月CPI达4.2%（2023年4月以来最高），霍…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.75, "matchedTerms": [ "美联储", "CPI", "美元", "能源", "油价", "Warsh" ], "href": "" }, { "date": "2026-06-27", "title": "Brent跌至约72美元后，Warsh美联储的高利率前提应从“结构性再加息”下修为“数据条件化高位平台”：基准终端利率上沿3.75%-4.00%，降息需等待核心PCE和预期共同确认。", "summary": "2026-06-27：本卡片压力测试“长期高利率”前提。结论是油价回落削弱了单凭能源再加息的逻辑，但尚不足以触发降息定价；6月FOMC维持3.50%-3.75%，SEP年末中位数3.8%，核心PCE仍2.9%，长期通胀预期3.6%。基准为55%概率高位平台，30%概率油价/PCE触发转鸽，15%概率二次能源冲击再加息。组合应以3.75%-4.00%为基准终…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "美联储", "美元", "能源", "油价", "Warsh" ], "href": "" }, { "date": "2026-06-27", "title": "地缘价格冲击通过霍尔木兹海峡受阻和灰色市场运费摩擦（每桶4.5美元）改变原油供应路径并拉动首轮通胀，尽管油价已降温至72.95美元，但政策反应焦点已转移至长期通胀预期锚定上，对资产配置构成高…", "summary": "分析表明，2026年5月中东地缘冲突导致霍尔木兹海峡每日1100万-1400万桶原油供应受阻（占全球约15-20%），叠加全球5月产量同比下降3.9%（至9450万桶/日），推动Brent油价飙升至120美元/桶。虽6月油价回落至72.95美元，但滞后影响仍存，5月美CPI升至4.2%（能源通胀涨23.5%贡献其60%的跳升）。能源直接投入产出传导较窄，但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "CPI", "美元", "能源", "油价" ], "href": "" }, { "date": "2026-06-27", "title": "尽管地缘缓和推动原油价格回落，但原油底部防守、AI负荷拉动天然气结构性上涨以及核心CPI的传导滞后，共同决定了能源下行无法抵消核心通胀的再加速风险，美联储政策中枢仍将维持更高更久。", "summary": "地缘风险缓和使原油价格暴跌至$78–$80/桶，但这不足以抵消核心通胀再加速风险。原油实物库存处于历史低位，且受OPEC+在WTI $75防守底线的制约；而AI数据中心表后气电需求推动天然气价格结构性上涨至$3.92/MMBtu以上，构成更持久的通胀源。核心通胀受黏性服务与薪资驱动，直接能耗传导较弱。美联储仍受通胀硬约束，利率长期维持高位。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "美联储", "CPI", "能源", "油价" ], "href": "" } ], "timeline": [ { "date": "2026-04-27", "supports": 0, "risks": 1, "neutral": 1, "value": 39, "events": [ { "title": "[auto-handoff] 能源通胀对利率路径的溢价影响", "summary": "致 主题研究员： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 56407410-f02d-4931-ac33-d1b5f98aae84 结果摘要: 针对您的疑问，作为全球宏观分析师，我已完成对 2026 年下半年通胀预期及中性利率（R-star）走势的量化评估。 由于 …", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 美元走势对商品定价影响", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 80e6b3d2-162c-44e1-8d00-839c93e77c11 结果摘要: Wrote handoff_response.md (8887 bytes) — 美联储鹰派表态对美元仅构成脉冲式而非…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-04-28", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "[auto-handoff] 汇率波动对原油定价影响", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 769f0f21-0333-4f0b-820d-bcc2d41c1f94 结果摘要: Wrote handoff_response.md (3540 bytes) — 深入分析了美联储官员偏鹰派表态后，美…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "投委会的\"类滞胀\"框架方向正确，但应锚定住房而非能源、把\"金融超配\"拆分为 GSIB 超配 / 区域性银行低配，并在监控矩阵中增设住房触发器作为劳动力拐点的领先指标。", "summary": "从房地产视角对投委会\"类滞胀\"决议进行压力测试，得到三条具体修订：(1) H4L 的真正锚是住房 CPI（占比约33%）而非汽油，因此美联储紧缩比委员会描述的更耐久，10Y 目标上修至 4.80–5.25%；(2) \"金融超配\"必须拆分——CRE 到期墙约 5,700 亿美元、Trepp 办公违约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hormuz供给冲击在急性期内强烈，但正常化轨迹（4Q26低于$90/b）所产生的基期反转效应使油价对美联储的约束具有时限性——首次降息推迟至2026年Q4而非推入2027年，宽松窗…", "summary": "能源分析师否定油价从结构上封锁美联储2026年宽松窗口的论断。在3月7.5 mb/d、4月9.1 mb/d减产、2Q26 Brent峰值约$115/b、4Q26回落至$90/b以下的正常化情景下，能源分项对标题CPI的同比贡献从2Q26的约+2.3–2.9 pp机械性下降至4Q26的约+0.8–1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "外汇市场已开始定价美联储独立性折价（DXY残差约3%，USD/JPY超额日元强势约5日元），但仍处于部分定价阶段，三重触发条件（BoJ加息、Warsh鸽派确认、美债拍卖尾部）一旦叠加…", "summary": "压力测试结果：外汇市场已在部分定价美联储独立性折价，但尚非主要驱动因素。DXY年初至今跌幅7.4%中约2.9%为制度风险残差，USD/JPY跌幅中约5日元（占35%）超出利率基本面，归因于相对可信度再平衡与套息去杠杆；CNY渠道由人民银行主动管理，不构成独立的市场信号。约5,400亿美元日元融资套…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "供应端结构性缺口足以支撑能源价格中枢上移和通胀粘性增强，但130美元油价本身仍需要扰动持续或恢复慢于预期。", "summary": "截至2026-05-18，本卡片支持能源价格高企具有供应端结构性基础。短期130美元布伦特仍依赖海湾扰动和库存快速下降，但油气田递减率、上游资本开支不足、备用产能变薄、LNG供应浪潮推迟，以及炼厂/中间馏分油瓶颈，均指向更高的能源价格底部和更大的上行尾部风险。BLS 4月CPI、世界银行能源价格预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Brent高于104美元会通过输入性PPI和利润率压缩强化A股从外交beta和周期成长向水电、受监管公用事业与红利资产切换的趋势。", "summary": "截至2026-05-17，本卡支持上一卡判断：Brent突破104美元已成为独立的通胀与利润率冲击。美国4月CPI同比3.8%、能源同比17.9%；中国4月CPI同比1.2%、PPI同比2.8%，且石油、燃料加工、化工链条环比涨幅显著。油价从89.4美元升至104.2美元的16.6%冲击，按IMF…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "地缘封锁带来的105美元油价与物流双击将美国推向滞胀困局，消费呈现结构性撕裂而非整体崩溃，而新任美联储主席凯文·沃什治下的鹰派加息风险已将下半年硬着陆（衰退）概率推高至55%。", "summary": "本卡评估了霍尔木兹海峡实际封锁（波斯湾超14mb/d产量受阻）导致105美元油价与航运费用大涨对美国经济的传导。虽因美伊外交谈判见曙光油价略跌至105美元，但这笔“地缘能源税”叠加海运费和国内陆运费上涨（+15% YoY），将美国4月CPI系统性推升至同比3.8%。一季度GDP增速已降至2.0%，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "能源价格冲击对全球通胀传导的深度评估", "summary": "能源端压力测试得出双向结论：传导真实存在但不会演化为螺旋。Brent ≈88美元、Henry Hub +35%、柴油 +16% [S5][S6][S7]，机械地为CPI贡献约1.4pp [自测算]。但1979式螺旋的必要条件不齐——美国油强度仅为1973年三分之一 [S3]、OPEC+闲置产能5.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "AI Capex韧性 × 油价冲击 × 联储延宕的共振，通过工业PPI与电力催生海外隐性二次通胀，将10Y美债锚定在5.0–5.3%，断裂点在日元/JGB而非美国风险资产。", "summary": "在2026-05-23工作日期下，AI硬件Capex韧性与霍尔木兹油价冲击的共振真实存在，但传导通过工业PPI、电力价格与资本品平减指数发生，而非头部CPI/PCE。这一通胀路径让联储基于PCE的反应函数系统性低估真实通胀，将10Y美债锁定在5.0–5.3%走廊延续整个2026下半年（期限溢价回升…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "2026年5月23日，随着凯文·沃什（Kevin Warsh）正式确认就任美联储主席，全球宏观范式发生了深刻转向。沃什确立了“利率鸽派、缩表鹰派”的政策底色，通过激进的缩表（QT）计…", "summary": "2026年5月23日，随着凯文·沃什（Kevin Warsh）正式确认就任美联储主席，全球宏观范式发生了深刻转向。沃什确立了“利率鸽派、缩表鹰派”的政策底色，通过激进的缩表（QT）计划强制期限溢价回归，将美联储从“久期最后买家”的地位撤出。叠加同期伊朗战争引发的历史性能源供给冲击（布伦特原油突破1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 4, "neutral": 0, "value": 30.6, "events": [ { "title": "极值政策分化下，日元套利交易正临近以“干预与VaR共振”为特征的系统性平仓爆破点，而人民币套利则面临央行“离岸挤空”的非线性行政打击风险。", "summary": "在美联储维持高利率以对抗油价冲击的背景下，全球货币政策分化达到1990年代以来极值。日元套利交易面临干预悬崖与防御性紧缩风险，投机存量达5000亿美元，一旦Brent原油突破130美元，将触发系统性VaR平仓风暴；而人民币套利交易受中美负利差驱动，但受资本管制保护，极易遭受央行抽干离岸流动性导致的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在更久高利率与高油价的‘双重滞胀’风险下，均衡组合应果断降低美债久期，依靠现金流股和商品进行通胀防守。", "summary": "截至2026-05-25，在美联储看跌期权走弱、美债收益率攀升（10年期触及4.69%）及布伦特油价维持于103.54美元/桶的高通胀压力下，大类资产配置应实施结构性防御。建议显著降低长久期美债敞口，转入短端现金捕获高收益；保留并超配能源等商品通胀对冲；将股票配置向高股息、强现金流的防御型价值股倾…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年伊朗战争石油冲击下的全球滞胀脉冲:分裂的美联储与中国进口型PPI抬头", "summary": "2026年伊朗战争引发的石油供给冲击(Brent峰值~$120,5月初回落至~$110)已将全球通胀问题从需求侧尾部风险转化为供给侧中周期事件。美国4月CPI同比3.8%(2023年5月以来最高),核心2.8%,能源CPI +17.9%;4月FOMC以8-4分裂维持3.50-3.75%(1992年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh 时代联储在 Hormuz 油价冲击与关税通胀粘性下的反应函数", "summary": "Kevin Warsh 于 5 月 13 日以 54-45 接任联储主席,Powell 留任理事。接手时的宏观组合是教科书式的滞胀风险:4 月 FOMC 在 3.50-3.75% 维持但出现 1992 年以来最多的 4 票反对;4 月 CPI 同比 3.8%(能源贡献 40%+),核心 2.8%,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 4, "neutral": 0, "value": 32, "events": [ { "title": "宏观二次通胀压力测试：重估美伊霍尔木兹备忘录破裂概率与通胀传导路径 研究院权威工作日期 ：2026 05 25 [自测算] 主笔分析师 ：global macro (全球宏观分析师)…", "summary": "宏观二次通胀压力测试：重估美伊霍尔木兹备忘录破裂概率与通胀传导路径 研究院权威工作日期 ：2026 05 25 [自测算] 主笔分析师 ：global macro (全球宏观分析师) 本卡立场 ： stress test (压力测试与风险重估) 一、 执行摘要 截至 2026 05 25 ，全球宏…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "实物原油供给赤字的长期化将打破央行的通胀容忍红线，迫使全球宏观叙事从“预防式降息”向“滞胀紧缩”急剧反转，触发系统性Risk-Off抛售。", "summary": "本报告通过宏观压力测试指出，若Hormuz断供拖延至Q3且Brent油价直奔$130，美联储将被迫完全撤销2026年降息指引甚至重启加息讨论，10年期美债有重返4.8%的风险；欧洲央行将陷入深陷滞胀与主权利差走阔的死局；日本央行则极大可能被迫紧急加息以保卫日元。这将彻底逆转当前的降息计价，引发全球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美日利差的高位反转将把全球2万亿美元的日元套息交易转变为‘流动性黑洞’，一旦USD/JPY跌破152.00，无序平仓将引发全球风险资产的系统性重估。", "summary": "由于霍尔木兹海峡危机引发原油价格飙升至130美元，美联储被迫维持高利率，而日本央行因防御性加息（预计6月加息至1.0%）导致美日利差预期见顶。目前USD/JPY处于160.00的历史高位，且套息交易规模达2万亿美元。模型显示，若USD/JPY跌破152.00结构性支撑，将引发20:1杠杆头寸的强制…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "主要央行在通胀粘性与日元危机下呈现同步的 hawkish 转向风险（美联储高利率维持更久/考虑加息、欧洲央行潜在加息、日本央行6月大概率加息），这正成为挤压全球跨资产流动性并触发期权…", "summary": "本卡片系统分析了美、欧、日三大主要经济体截至2026年5月25日的增长预期与货币政策转向风险。美国一季度增长2.0%，但受能源冲击CPI反弹至3.8%，美联储维持高利率更久且面临加息讨论；欧元区一季度仅增0.1%，受滞胀威胁欧洲央行或于6月11日被迫加息；日本央行受日元贬值与收益率走高压力，6月会…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 3, "neutral": 0, "value": 39.4, "events": [ { "title": "[auto-handoff] 美伊外交谈判传闻及油价跌破100美元对美联储下半年降息与DXY的边际影响评估", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: d5d2ffa6-c741-4e78-b373-a9292f4c8a66 结果摘要: Wrote handoff_response.md (10595 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国30年期国债破5%:期限溢价、财政漂移与中东能源冲击下的长端重定价", "summary": "在美联储维持3.50–3.75%、4月CPI回升至3.8%、5月13日30Y招标中标5.046%(2007年以来首次5字头)的组合下,长端已与降息周期脱钩,转由期限溢价定价:中东能源冲击锚定通胀预期、FY26赤字突破2万亿美元放大长端供给、美联储反应函数被长端约束。基准情形30Y三季度末区间4.9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "布油跌至$96.71的地缘纸面缓和无法撼动美联储沃什时代收紧期限溢价的鹰派核心底色，跨资产‘高偏斜’结构正面临大宗红利挤兑、科创信用去杠杆和日元套利平仓的非线性崩塌风险。", "summary": "本报告揭示霍尔木兹海峡地缘缓和传闻导致油价暴跌14.36%至$96.71，但由于物理复航滞后及极度去库，这一纸面利好完全无法抵消新任美联储主席Warsh收紧资产负债表、驱动10Y名义美债至4.59%及实际利率至2.20%的鹰派政策影响。‘高价高偏斜’特征暗示全球大类资产结构极为脆弱，A股科创50在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "虽然整体能源价格走弱缓解了CPI名义压力，但中间投入与柴油成本的滞后传导将维持核心通胀高粘性，迫使新上任的沃什（Warsh）联储在6月维持3.50%-3.75%的高利率，并将降息至少…", "summary": "截至2026-05-28，我们量化测算了原材料及能源价格对核心通胀的滞后传导效应。虽然5月油价回落，但中间品价格高企（ISM支付价格达65.5%），对核心货物传导弹性为0.22（滞后4-6个月），对服务传导弹性为0.15（滞后2-3个月），三季度核心CPI/PCE仍显粘性。5月22日宣誓就职的新主…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 3, "neutral": 0, "value": 42.2, "events": [ { "title": "能源价格重心的下移提供了一次缓和全球“滞胀”压力的正向供给冲击，虽然成功消解了美欧央行的强行加息风险，但在低储蓄率、物流滞后与-273 bps中美巨幅倒挂的刚性约束下，全球主要央行在…", "summary": "本报告推演了Brent原油重心下调至82-92美元对三季度全球通胀与货币政策的边际影响。受汽油直接回落与航煤暴跌（裂解价差收窄至18-22美元/桶）间接拖累，三季度美国整体CPI与欧洲HICP将分别下修21 bps和30 bps。这有效移除了美联储年底加息选项，但受制于2.6%极低储蓄率与核心PC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹红利 vs. 沃什鹰派：风险转折的脆弱性", "summary": "霍尔木兹红利 vs. 沃什鹰派：风险转折的脆弱性 地缘利好引发的油价下跌能否抵消美联储新主席 Warsh 的鹰派立场？当前的‘高价高偏斜’是否暗示结构性崩塌风险？ Prior related institute work to acknowledge and BUILD ON (do not re…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 美伊外交谈判传闻及油价跌破100美元对美联储下半年降息与DXY的边际影响评估", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 0f9af4e0-3879-478e-bce6-3e204f2a41cc 结果摘要: Wrote handoff_response.md (4187 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 2, "neutral": 0, "value": 49.6, "events": [ { "title": "油价下行（-10.20%）将触发欧日双重贸易条件修复，日本贸易条件弹性极强但面临低通胀（Tokyo CPI 1.3%）拖累BOJ加息；欧元区则通过能源账单减负推动ECB转鸽降息，两者…", "summary": "截至2026-05-31，本卡完全支持卡片04外汇分析。油价从$91.87/bbl向$80-85/bbl下行构成欧日双重贸易条件修复 ToT，但存在显著的强度分化：日本贸易收支弹性极高，4月顺差录得¥3019亿逆袭预期，但东京5月核心CPI仅1.3%导致BOJ加息恐被动夭折；欧元区3月能源赤字宽至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "油价持续高位将迫使美联储在2026年下半年保持鹰派立场，推升10年期美债收益率至4.75%并使期限溢价突破100bps。", "summary": "评估布伦特原油维持在105-120美元/桶对美联储政策及美债收益率的冲击。高油价将直接贡献headline CPI约1.3-1.5%，并推升核心PCE至3.4%以上。美联储预计将维持3.75%-4.00%的高利率，甚至实施防御性加息。受通胀不确定性、沃什提名政权更迭及地缘溢价驱动，预计10年期美债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 1, "value": 60, "events": [ { "title": "DXY 99区间会弯曲但不会断裂——欧日央行被ToT冲击逼入墙角，但它们被迫的鹰派反应恰好阻止了EUR/JPY的失序崩溃，而黄金吸收了本应推动DXY破100的边际避险流。", "summary": "本报告支持并延伸卡片01的美元钝化结论：Hormuz封锁对欧日构成高度非对称的BoP冲击（自测算欧元区年化能源账单+1400-1800亿欧元；日本+6.5-8万亿日元），但ECB与BOJ均被困两难——能源CPI再加速阻止主动宽松，而ECB预计6月鹰派持有(60%概率)、BOJ预计7月象征性加息25…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然高油价与ISM价格指数增强了美联储的鹰派偏向，但由于新主席Warsh对前瞻指引的怀疑以及油价上涨的供给侧性质，美联储在6月会议上将维持利率不变的“长期暂停”基准，而非发出预防性加…", "summary": "在美联储主席Warsh首次主持FOMC前，原油反弹至95-100美元，5月ISM制造业与服务业价格指数高企（分达82.1%与71.3），以及票委Hammack释放鹰派警示，令加息预期重燃。然而，基于通胀的供给侧属性、新主席对前瞻指引的怀疑态度以及缩表（QT）优先权，基准情景仍是维持利率在3.5%-…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 1, "value": 55.4, "events": [ { "title": "在同质能源冲击与美联储鹰派停顿下，ECB的防御性加息与BoJ加息至1.00%将压缩G3名义利差，足以在3Q26对冲美元强势并维持EUR/USD（1.1400-1.1800）和USD/…", "summary": "若美联储于6月17日维持利率并上移点阵图，在同一能源冲击（原油价格在95美元附近波动）下，ECB与BoJ的被迫紧缩反应将重塑G3利差。ECB预计在6月11日加息25bp至2.25%，BoJ预计在6月16日加息25bp至1.00%以应对160附近的日元贬值与输入性通胀。利差压缩与政策对冲将防止欧/日…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "沃什(Warsh)新联储的滞胀困局:6月16–17日FOMC的政策抉择", "summary": "美联储连续第三次将利率维持在3.50%–3.75%,4月以8–4票通过,为1992年来首次四票反对:Miran主张降息,三位地区主席反对保留宽松倾向。新主席Kevin Warsh于5月22日就任,6月16–17日为其首场FOMC。通胀冲击源于供给端:霍尔木兹海峡3月关闭推动Brent一季度末破12…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国电网容量危机正通过“工业高电价-核心通胀粘性-美联储鹰派转向”的链条，对制造业再回流构成实质性压制，并重塑美联储的降息路径。", "summary": "截至 2026 年 6 月 5 日，美国电网容量危机已从公用事业局部问题演变为核心宏观阻力。PJM 容量价格连续两年触及上限（2027/28 年度清算于 333.44 美元/兆瓦-日）且存在 6.6 吉瓦缺口，导致美国东部工业电价上涨 15-25%。这种“供给侧冲击”正通过成本转嫁推升核心 PCE…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 4, "neutral": 0, "value": 41.8, "events": [ { "title": "面对霍尔木兹海峡商业复航不确定性引发的黏性通胀环境，大类资产配置应通过缩短名义久期、增配短端TIPS以及超配战略商品（油气/黄金）与低杠杆能源股，化解地缘溢价风险并获取绝对收益。", "summary": "截至2026年6月5日，中东地缘摩擦使油价高企（Brent 95.24美元/桶），通胀维持高位，利率更高更久（10年期美债收益率4.49%）。名义股债负相关性失效。建议战术低配长端名义债，超配短端TIPS，超配战略商品（能源与黄金）和能源股；外汇超配美元和商品货币，日元作为融资货币。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "深度为负的中美利差（约−277个基点）结构性封顶人民币升值，使美元/人民币呈区间震荡带双向尾部；而美元/日元逼近160、日央行6月加息近乎完全定价，日元套利交易是更迫切的波动率触发型…", "summary": "本卡支持01号卡片“美联储路径驱动美元与全球套利”的框架，但将其单点鸽派结论修正为两侧分布：截至2026年6月4日美国10年期收益率4.49%，市场对年底加息概率已重定价至约85%（中东能源冲击）。人民币偏强（央行6月4日中间价6.8203，弱于预估6.7770），交投于2023年以来最强，但中美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "强美元短期掩盖美元/日元中的日元风险，但通过全球流动性、风险预算和干预凸性放大日元套利平仓尾部。", "summary": "截至2026-06-05，本卡综合前序外汇判断：油价驱动的美联储年底加息尾部约85%会先推强广义美元，压制欧元/美元和脆弱新兴市场货币，并让美元/日元维持在接近160的高位。但这不是对日元套利平仓风险的抵消，而是延后后放大：强美元收紧全球流动性、削弱风险预算、增加日本干预或日央行收紧概率，使EUR…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供给极端冲击对美联储‘按兵不动’预期的破坏力测试 报告日期 ：2026年06 05 分析师 ： energy analyst 立场 ：压力测试 (Stress test) 核心摘…", "summary": "能源供给极端冲击对美联储‘按兵不动’预期的破坏力测试 报告日期 ：2026年06 05 分析师 ： energy analyst 立场 ：压力测试 (Stress test) 核心摘要 截至2026年06月05日，随着霍尔木兹海峡局势在美伊博弈下持续处于高危状态，前一交易日（2026年06月04日…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "伊朗能源冲击对发达国家传导具有对称性但政策约束极度不对称：欧洲央行已被迫放弃6月降息，而日本面临严重的汇率贬值与输入性通胀双重打击，BoJ将被迫在6月提前缩表以捍卫汇率。", "summary": "在布伦特原油突破100美元的背景下，欧元区对进口能源的依赖导致其5月HICP能源分项同比飙升约14%，迫使欧洲央行在6月4日会议上放弃降息并选择“鹰派按兵不动”。对于日本，高油价与持续弱势的日元(USD/JPY >157)形成双重冲击，4月进口物价同比飙升12.5%。这种严峻的贸易条件恶化正将日本…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 9, "neutral": 4, "value": 26.6, "events": [ { "title": "Warsh 美联储 6 月议息会议:油价冲击型滞胀下的政策两难", "summary": "2026 年 2 月 28 日伊朗战争引发霍尔木兹海峡 94 天以上封锁,切断全球约 20% 石油供应,布伦特一度飙至 138 美元,4 月美国 CPI 同比升至 3.8%(2023 年 5 月来最高),其中能源 +17.9%、贡献逾四成涨幅,而核心仅 2.8%——典型相对价格冲击。新任鹰派主席 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关税本身更像一次价格水平冲击；只有当它与高油价、海运拥堵和高成本补库存叠加时，才会变成迫使 Warsh 美联储下半年再加息的第二轮通胀。", "summary": "截至 2026-06-08，本卡对第 02 张卡做压力测试：能源冲击足以支撑美联储维持 3.50%-3.75% 不变，但关税、海运和补库存是否同步落地，才决定是否出现第二轮通胀。基准判断是 6 月按兵不动、下半年继续高位维持；若 7 月关税行动硬化、WCI 维持在约 3,500 美元/40 英尺箱…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "配置重心应从对冲能源滞胀转向捕捉“AI基建通胀”与“Warsh生产率叙事”的错配，超配实物资产（铜/黄金）与AI成长股。", "summary": "随着能源冲击转向机械性消退，油价企稳于93.09美元，跨资产配置正转向由“IT通胀”驱动的结构性变局。新任美联储主席沃什面临生产率叙事与实物资产瓶颈（如15万吨铜缺口）的博弈。建议配置上超配AI基建相关成长股与工业金属，对美债久期维持谨慎中性，并利用美元对非美货币的差异化走势。Warsh的“生产率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月16-17日FOMC前夕：油价冲击下的美国\"准滞胀\"窗口与Warsh首秀", "summary": "基准情形是6月FOMC一致按兵不动于3.50-3.75%，点阵图略偏鹰修正：4月CPI 3.8%、能源同比+17.9%与5月非农+17.2万/失业率4.3%的组合，构成供给冲击型'准滞胀'而非真滞胀——Brent已从120美元回落至约93美元，使能源同比贡献将在8月后机械性消退。Warsh需要先稳…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-08", "supports": 1, "risks": 7, "neutral": 0, "value": 36, "events": [ { "title": "日元套利交易受制于利差边际压缩与汇率干预而面临高平仓风险，而欧元兑美元虽受周期性加息支撑维持在1.1642，本分析（2026-06-04）认为其因“AI税”与高能源成本带来的贸易条件…", "summary": "截至2026-06-04，在美联储暂停降息和欧洲央行加息背景下，日美利差边际收窄25bp，日欧10年期国债利差极窄（仅35bp），导致日元套利夏普比率恶化。日本央行6月中旬加息至1.00%的预期与汇率干预非对称风险，大幅加剧套利平仓隐患。同时，欧洲因进口高价美国AI硬件和服务面临“AI税”流出，叠…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "研究报告：输入型能源通胀背景下的中国宏观政策应对 日期： 2026 06 08 分析师： 中国宏观分析师 (china macro) 状态： 支持 (SUPPORT) 1. 执行摘要…", "summary": "研究报告：输入型能源通胀背景下的中国宏观政策应对 日期： 2026 06 08 分析师： 中国宏观分析师 (china macro) 状态： 支持 (SUPPORT) 1. 执行摘要 截至 2026 年 6 月 8 日，中国经济正面临 Brent 原油价格稳居 95 美元/桶以上与强势美元（美元兑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "本研究周期在数据一致性、逻辑严密性及对冲方案的风险覆盖度上均达到高标准，正式通过评审并结项。", "summary": "本卡作为 10 卡研究周期的终篇，完成了对霍尔木兹海峡封锁背景下全球资产配置与对冲方案的质量与合规审计。经核实，全周期数据锚点一致（布油 138 美元、美联储 3.50-3.75%、美元/日元 168.45），宏观逻辑严密，成功实现了从能源冲击到跨资产对冲及权益风格轮动的战略转化。200 bps …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储2026年加息概率重估与全球流动性拐点", "summary": "支持卡片01。我把分析推至上游的美联储反应函数:5月CPI(2026-06-10发布)若证实再加速(核心环比≥0.4%、整体同比≥3.9%),美联储不会在6月或7月加息,但会把市场隐含的2026年加息概率从约28%重估至40–45%,并将点阵图叙事从'两次降息'翻转为'按兵乃至加息'。正是这一重定…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "中东停火信号下的油价风险应如何重定价", "summary": "晨报不应二选一，而应按时间维度与价格组成分开处理。当前~$90-95的Brent平台价已是已实现、具粘性的基准通胀冲击（4月CPI 3.8%、能源+17.9%、汽油+28.4%），美联储已转向2026年不降息；据此应上抬中枢CPI路径。而高于和平时期基本面(~$80)的~$12-16战争溢价是二元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [ { "title": "夏季能源通胀反弹与电网基荷电力短缺构成了双重尾部风险，足以打破美联储耐心和AI资本品扩张的“双速经济”假设，多资产策略必须引入能源与公用事业对冲。", "summary": "截至2026-06-12，本报告对首席策略师的“双速经济”配置策略进行了能源维度的压力测试。指出夏季高温和活跃飓风季可能推高天然气与原油价格，引发整体CPI超预期反弹，打破美联储“保持耐心”的假设并引发股债双杀。同时，电网基荷电力短缺构成了AI资本品实质上电的物理硬约束，可能切断资本品扩张动能。建…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 7, "neutral": 4, "value": 52.2, "events": [ { "title": "6月央行双会周（日本央行6月15-16日加息至1.00%、美联储6月16-17日按兵不动）叠加正在降温但仍高企的中东石油冲击，对全球风险资产与日元套利交易的双重风险", "summary": "进入6月16日当周的央行双会，日本央行预计加息25bp至1.00%、美联储料以约97-99%概率维持3.50-3.75%不变。5月美国CPI同比4.2%为2023年4月以来最高，但约60%涨幅由能源驱动，且布伦特已从3月底逾$113峰值回落至约$89（6月11日因美伊降温下跌逾4%），通胀尾部正在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "沃什履新美联储主席后的首次FOMC（2026年6月16-17日）与供给侧能源冲击下的滞胀困局", "summary": "2026年6月16-17日是Kevin Warsh出任美联储主席后的首次FOMC。基准情形为维持利率3.50-3.75%不变，但通过点阵图（抹去2026年最后一次降息、甚至可能废弃点阵图框架）和措辞由宽松转向中性偏紧，正式确立'更高更久'格局。关键张力在于：5月整体CPI 4.2%看似惊人，逾60…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年6月14日，中东能源危机导致布伦特原油维持在 105 美元/桶，中国作为原油进口依存度高达 73% 的国家，正面临显著的“输入型通胀”与制造业利润挤压。本研究通过对国家发改…", "summary": "2026年6月14日，中东能源危机导致布伦特原油维持在 105 美元/桶，中国作为原油进口依存度高达 73% 的国家，正面临显著的“输入型通胀”与制造业利润挤压。本研究通过对国家发改委成品油定价机制（80 130 美元区间利润压缩原则）的深度解析，揭示了 PPI CPI 剧烈背离下的行业生存法则。…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "油价再冲高会抵消美联储降息预期，但只有在霍尔木兹流量持续受限、Brent接近或高于$105/b并传导到汽油CPI时，才会触发剧烈的全球资产定价重构。", "summary": "截至2026-06-13，能源压力测试结论是：油价回落支持“有序重估”基准，但霍尔木兹物理流量尚未正常化，尾部仍足以改变全球资产定价。若Brent维持在$85-95/b，5月美国CPI的能源冲击大概率消退；若Brent贴近EIA假设的$105/b 6-7月均价，降息预期会被压制；若重回$120/b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 8, "neutral": 0, "value": 27.6, "events": [ { "title": "美联储加息尾部风险正在重构全球套利交易，强化 USD/CNY 上行压力，但中国政策缓冲仍足以把压力维持在有管理区间内。", "summary": "截至2026-06-15，本卡片支持上一张卡片的鹰派美联储判断。美国5月CPI与就业韧性使美元现金收益成为全球套利交易的新门槛，日元融资仍便宜但 BOJ 风险上升，欧元融资因 ECB 加息而吸引力下降。USD/CNY 面临中美政策利差、美元收益和能源冲击带来的压力，但 CFETS 中间价机制、3.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh 美联储遭遇油价冲击：迈向 6 月 16–17 日 FOMC 的“轻度滞胀”", "summary": "2026 年 5 月美国 CPI 同比 +4.2%(三年多最高),核心仅 2.9%,缺口几乎全部来自伊朗冲突引发的能源(+23.5%)。劳动力市场温和降温(非农 +17.2 万、失业率 4.3%)。新任鹰派主席 Warsh(5 月 13 日就任)继承了通胀再加速,6 月 17 日 FOMC 隐含约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储6月FOMC前瞻：能源通胀冲击下的Warsh首秀", "summary": "5月美国CPI因地缘政治引发的能源价格飙升（汽油价格上涨40.5%）达到4.2%。虽然核心通胀受控（2.9%），但这完全打破了市场的降息预期。预计6月16-17日的FOMC会议上，新任主席Warsh将维持3.50%-3.75%的利率不变，并通过点阵图传达年内不降息的鹰派立场，以应对通胀二次反弹的风…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "105美元/桶的能源价格韧性将通过物流和公用事业成本的传导抬升核心通胀中枢，导致核心CPI在3.0%附近保持粘性，阻碍美联储在短期内转向鸽派。", "summary": "截至2026年6月15日，受供给约束及夏季需求旺季影响，布伦特原油在105美元/桶展现出较强韧性。尽管对整体CPI（4.2%）的一阶冲击显着，二阶传导效应也已在运输、物流和公用事业费率中显现。然而，在失业率升至4.3%的背景下，全面的工资-价格螺旋尚未形成。能源价格的韧性将抬升通胀中枢，使核心CP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 2, "value": 71, "events": [ { "title": "Warsh首场FOMC：油价驱动型滞胀下的鹰派转向与全球外溢", "summary": "2026年6月16-17日是Warsh就任美联储主席后的首场FOMC，利率维持3.50%-3.75%已近乎确定（约97-99%）。他接手的是1979年沃尔克以来最难开局：5月整体CPI同比4.2%（2023年4月来最高），但核心仅2.9%——缺口几乎全来自伊朗战争引发的油价冲击（汽油+40.5%、…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "日银6月加息至1.0%与5,000亿美元日元套利交易的结构性拆解", "summary": "日银6月15–16日会议料加息25bp至1.0%(94%经济学家共识),美联储同期会议99.6%概率维持3.50–3.75%,首次出现利差双向收敛。USD/JPY ~160时套利仍在赚钱,但摩根士丹利估算仍有约5,000亿美元在外日元融资头寸,卖方对年底USD/JPY预测跨度达14日元(丰业150…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年6月FOMC前瞻：能源冲击、CPI上行与沃什治下的美联储", "summary": "2026年6月FOMC前瞻：能源冲击、CPI上行与沃什治下的美联储 面对4.2%的总体CPI与108美元的布伦特原油，新任主席沃什将在6月16-17日的FOMC会议上如何调整点阵图与政策立场？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美联储'更高更久'+油价冲击向中国的传导：政策分化、人民币与A股配置", "summary": "我部分推翻Research note 01的设定。中美政策分化在2026年年中不仅可持续、且已无压力地发生——但机制不同。5月中国PPI同比+3.9%（2022年7月来最高，受同一伊朗战争大宗冲击+AI资本开支推升），CPI仅+1.2%，故'通缩给人民银行空间'被削弱[S1][S2]；人民银行5月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 1, "neutral": 0, "value": 72.4, "events": [ { "title": "2026 能源地缘政治风险对 AI 算力成本与全球通胀的传导评估", "summary": "支持卡片03：霍尔木兹封锁（卡塔尔3-4不可抗力，约20%全球LNG受阻[S1]）使全球气价飙升而美国Henry Hub脱钩走低——TTF +35%至14.8、JKM +51%至16.0美元/MMBtu，Henry Hub自2-28跌约9%[S1]，2026均价仅约3.5美元[S3]。我把它延伸到…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 2, "neutral": 0, "value": 67.8, "events": [ { "title": "全球二次通胀预期退坡、美欧央行降息周期重启及日元套利平仓的连锁传导", "summary": "2026年6月美联储（维持3.50-3.75%、偏鹰）、欧央行（加息至2.25%）、日央行（加息至1.0%）同步鹰派，唯一共因是伊朗战争油价冲击（5月CPI 4.2%、能源贡献当月涨幅60%+）。美伊备忘录把WTI封顶$75-77后，移除了这一变量：克利夫兰联储整体CPI临近预测将随基数效应快速回…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网公用事业资产费率基数扩张对宏观通胀与货币政策的传导效应", "summary": "压力测试卡片01–05的宏观结论。直接渠道被夸大：电力属能源、按构造排除于核心PCE之外，达拉斯联储测算数据中心对总体PCE的拉动2026仅+0.05pp、2030升至+0.13pp[S1]；电力仅占CPI约2.5%[S11]，自测算直接总体贡献约0.10pp。但费率基数是30–40年价格水平锁定…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 11, "neutral": 0, "value": 15.2, "events": [ { "title": "霍尔木兹油价冲击与 Warsh 治下美联储鹰派转向的碰撞", "summary": "美国—以色列对伊朗的战争令伊朗于 2026 年 3 月封锁霍尔木兹海峡，布伦特一度飙至约 120 美元，推动 5 月 CPI 跳升至同比 4.2%（能源贡献超 60%，核心仅 2.9%）。新任主席 Warsh 首秀转鹰，6 月 17 日维持利率 3.50%–3.75% 但点阵图从降息翻转为加息，年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘能源冲击下美联储与日本央行的同步鹰派转向——全球滞胀是否重来", "summary": "2026年6月主要央行同步收紧：美联储6月17日连续第4次维持联邦基金利率3.50%-3.75%，但点阵图把2026年底中值上调至3.8%、偏向加息（Warsh主席首会）；日本央行6月16日加息至1.0%，为1995年来最高。驱动因素是中东冲突与霍尔木兹海峡受阻带来的能源冲击：美国5月整体CPI同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "油价回落不是供给侧解除警报；但除非霍尔木兹流量和库存再度恶化，能源冲击更可能造成 headline 黏性而非全面二次通胀。", "summary": "截至 2026-06-21，本卡对 Research note 01 的再通缩判断进行压力测试：油价回落降低了第一轮 CPI 冲击，但霍尔木兹物流、库存消耗、成品油裂解价差、LNG 受损和电网瓶颈仍构成结构性供给脆弱性。基准不是顽固滞胀，而是 headline 黏性、core 受控；若霍尔木兹流量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源冲击下的美国滞胀式假动作：CPI 反弹至 4.2% 与 Warsh 美联储转鹰", "summary": "2026 年 5 月美国 CPI 同比升至 4.2%（2023 年 4 月来最高），几乎全由伊朗冲突的能源冲击驱动（能源 +23.5%、汽油 +40.5%），而核心仅 2.9%。6 月 16–17 日 FOMC 在 4.25–4.50% 鹰派按兵不动，新主席 Warsh 上调通胀预测、约半数官员给…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 1, "risks": 10, "neutral": 3, "value": 39.6, "events": [ { "title": "伊朗战争后的全球宏观转折：能源冲击消退，但美联储与日本央行维持鹰派带来的滞胀宿醉", "summary": "2026年伊朗战争（2/28美以空袭、3/4霍尔木兹关闭）将布伦特推至逾120美元，6/17美伊谅解备忘录使油价回落至约80美元。但通胀已从能源迁至核心：5月美国CPI总体同比4.2%（三年新高），核心升至2.9%，居住与食品双双加速。央行反应函数已上移——美联储6/17维持3.50–3.75%但…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026年伊朗战争能源冲击重新定价全球利率周期：通胀是暂时性还是持久性", "summary": "2026年伊朗战争能源冲击重新定价全球利率周期：通胀是暂时性还是持久性 在布伦特油价从约108美元回落至约80美元、核心CPI仍受控于2.9%的背景下，5月4.2%的整体通胀冲击主要是暂时性的（允许美联储维持按兵不动）、还是已嵌入足够持久性以迫使Warsh领导的美联储在10月加息？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026年伊朗战争能源冲击重新定价全球利率周期：通胀是暂时性还是持久性", "summary": "持续15周的2026年伊朗战争制造了一次供给侧能源冲击，把发达市场利率周期从降息扭转为鹰派按兵不动+加息倾向：美联储6月17日维持3.50%–3.75%但点阵图翻转为加息（2026年底中值升至3.8%、PCE上调至3.6%），日本央行加息至1.00%，美国5月CPI同比4.2%创近三年新高。但布伦…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Brent跌破80美元/桶削弱了能源冲击的尾部风险 [S5]，但在库存和成品油缓冲修复前，美联储仍会把能源通胀视为延后降息而非立即宽松的理由 [S8][S9][S10]。", "summary": "截至2026-06-22，本卡支持上一张global-macro结论：Brent跌破80美元/桶是油价风险缓和，但不是实体供应链已经完全正常化 [S5]。EIA仍估计2Q26全球库存下降630万桶/日、3Q26下降760万桶/日 [S1]，IEA仍预计2026年全球供应下降390万桶/日 [S4]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 3, "neutral": 1, "value": 69, "events": [ { "title": "小盘创纪录主要是 Russell 再平衡机械流 + 油价缓和的暂时性宽度冲顶，而非鹰派 Warsh 环境下能持续的大轮动；真正持久的子交易是大盘 AI-capex 去杠杆。", "summary": "6-22 Russell 2000 创纪录站上 3,000（年内 +>21%），而 Alphabet 领跌的大盘科技 AI-capex 去杠杆形成背离。叠加 6-26 全年最大被动收盘流（Russell 半年度再平衡，约 11 万亿美元锚定）与油价较高点 −20% 的缓和脉冲，首席策略师判定：次周…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "逢跌买入并非永远错,而是现在危险——散户杠杆与易跳空微结构仍按一个已(有条件)结束的宽松体制校准;确认或证伪体制切换的扳机是霍尔木兹/油价,而非美联储。", "summary": "美联储2026年6月17日点阵图翻转为偏加息(9/19位预测年内加息、年末利率中值升至约3.8%),5月CPI达4.2%创三年新高,主因霍尔木兹能源冲击;美韩科技已gap-down(纳指6/4-5跌4.18%、KOSPI中旬跌约10%)。但A股散户仍逢跌买入——6/18半导体91只涨停、两融近2.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元套利平仓担忧与电网/燃机硬约束下的全球AI CapEx增速二阶导下修辩论", "summary": "日元套利平仓担忧与电网/燃机硬约束下的全球AI CapEx增速二阶导下修辩论 JPY Carry Trade Unwind and Grid Constraints: Debating AI CapEx Growth Slowdown 在新任美联储主席 Warsh 限制性利率（3.50%-3.75…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元套利平仓担忧与电网/燃机硬约束下的全球AI CapEx增速二阶导下修辩论", "summary": "日元套利平仓担忧与电网/燃机硬约束下的全球AI CapEx增速二阶导下修辩论 在新任美联储主席 Warsh 限制性利率（3.50%-3.75%）及日元汇率逼近161.83干预前沿背景下，重型燃机精铸产能瓶颈与电网并网（53个月时延）是否会迫使云厂商 2026-27 年 7250 亿美元 CapEx…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 1, "neutral": 0, "value": 79.4, "events": [ { "title": "美联储政策超调风险下的全球宏观经济周期与通胀路径演变", "summary": "截至2026-06-26，G3央行在6月同步偏鹰（美联储维持3.50%-3.75%但点阵图转向年内加息、欧央行存款利率上调至2.25%、日本央行加息至1%），却恰逢能源主导的通缩前沿：布伦特跌至约73.74美元/桶，较约105美元的油价假设低约30%，而5月核心PCE仍黏在3.4%、整体PCE破4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 5, "neutral": 1, "value": 59.8, "events": [ { "title": "2026年6月全球货币政策同步转鹰：美联储、欧洲央行、日本央行因伊朗战争能源冲击齐齐转向", "summary": "2026年6月，欧央行（6/11，存款利率升至2.25%）、日本央行（6/16，升至1.00%，1995年来最高）、美联储（6/17，维持3.50–3.75%但点阵图翻转至9名官员预期加息、2026中值3.8%）一周内同步转鹰，共同触发因素是伊朗战争带来的能源通胀（美国5月CPI 4.2%、能源同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh 美联储鹰派首秀：在正在消退的霍尔木兹能源冲击中保留加息倾向", "summary": "2026-06-17 Warsh 首次主持 FOMC，利率维持 3.50%–3.75% 不变，但点阵图将 2026 年底中位利率从 3.4% 上调至 3.8%、删除宽松倾向，9/18 位委员预计年内加息 [S1][S2][S6]。触发因素是 5 月 CPI 同比 4.2%（2023 年 4 月以来…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "高利率与能源通胀背景下的全球大类资产配置策略", "summary": "支持第01卡『更高更久、勿抢跑鸽派』框架并落地为配置。2026-06-27背景下：美联储维持3.50%-3.75%、点阵上修至3.8%[S1]；10年/2年美债4.41%/4.13%、实际收益率首次持续转正[S3]；布伦特约75.6美元、双向波动[S2]；标普前瞻P/E约20-22偏贵[S5]；H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在通胀粘性阻碍美联储降息的背景下，纸面原油抛售与实物供给紧张的冲突使得油价下行有底而上行具备高凸性特征，使其在资产相关性转正的环境下成为有效的非对称风险对冲工具。", "summary": "截至2026-06-27，本卡支持能源作为组合凸性对冲工具的逻辑。美伊备忘录虽引发纸面油价大跌（WTI至70.24美元/桶），但扫雷瓶颈（60日延迟）与伊朗原油出口缓慢恢复使物理供给仍然偏紧，且美原油去库1553万桶、SPR降至历史新低3.492亿桶。高油价向核心PCE的二阶传导存在6-12个月时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 8, "neutral": 0, "value": 40.2, "events": [ { "title": "美联储2026年6月鹰派转向：供给冲击型通胀下的滞胀压力测试", "summary": "2026年6月17日FOMC连续第四次维持利率3.50%–3.75%（12–0全票，Warsh首会），但点阵图把2026年末中值从约3.4%上调至3.8%，抹去年内降息、隐含加息，2026年PCE预测自2.7%上修至3.6%，18人中17人视通胀风险偏上行[S1][S3][S4]。驱动力是供给冲击…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Brent跌至约72美元后，Warsh美联储的高利率前提应从“结构性再加息”下修为“数据条件化高位平台”：基准终端利率上沿3.75%-4.00%，降息需等待核心PCE和预期共同确认。", "summary": "2026-06-27：本卡片压力测试“长期高利率”前提。结论是油价回落削弱了单凭能源再加息的逻辑，但尚不足以触发降息定价；6月FOMC维持3.50%-3.75%，SEP年末中位数3.8%，核心PCE仍2.9%，长期通胀预期3.6%。基准为55%概率高位平台，30%概率油价/PCE触发转鸽，15%概…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘价格冲击通过霍尔木兹海峡受阻和灰色市场运费摩擦（每桶4.5美元）改变原油供应路径并拉动首轮通胀，尽管油价已降温至72.95美元，但政策反应焦点已转移至长期通胀预期锚定上，对资产配…", "summary": "分析表明，2026年5月中东地缘冲突导致霍尔木兹海峡每日1100万-1400万桶原油供应受阻（占全球约15-20%），叠加全球5月产量同比下降3.9%（至9450万桶/日），推动Brent油价飙升至120美元/桶。虽6月油价回落至72.95美元，但滞后影响仍存，5月美CPI升至4.2%（能源通胀涨…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "尽管地缘缓和推动原油价格回落，但原油底部防守、AI负荷拉动天然气结构性上涨以及核心CPI的传导滞后，共同决定了能源下行无法抵消核心通胀的再加速风险，美联储政策中枢仍将维持更高更久。", "summary": "地缘风险缓和使原油价格暴跌至$78–$80/桶，但这不足以抵消核心通胀再加速风险。原油实物库存处于历史低位，且受OPEC+在WTI $75防守底线的制约；而AI数据中心表后气电需求推动天然气价格结构性上涨至$3.92/MMBtu以上，构成更持久的通胀源。核心通胀受黏性服务与薪资驱动，直接能耗传导较…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 20, "neutral": 2, "value": 12, "events": [ { "title": "美联储6月鹰派转向与能源冲击逆转的错配:点阵图是否已经过时", "summary": "2026年6月17日FOMC维持利率于3.50%–3.75%,但点阵图转鹰、抹去年内降息并将2026年底中值上调至3.8%,隐含至少加息一次,期货一度定价10月加息。触发因素是美伊冲突引发的能源冲击:5月整体CPI同比4.2%、能源同比23.5%。但该冲击正快速逆转——布伦特从120美元上方跌至约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源价格的崩溃通过被动推升实质利率（188-264个基点）对实体经济造成隐性紧缩冲击，显著放大了美联储在Warsh新主席政策框架下过度紧缩的政策失误尾部风险。", "summary": "本卡片分析了能源价格向通缩力量转变对美联储的二次紧缩冲击。霍尔木兹海峡复航导致油价跌至70美元区间，全球供需过剩凸显。由于能源对核心通胀的传导偏弱，核心CPI仍具粘性，导致美联储在Warsh主席新政下固守限制性利率。这将推动实质利率较5月被动飙升188至264个基点，显著放大过度紧缩的政策失误尾部…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年6月美联储、欧央行、日本央行同步转鹰，却在一场正在消退的伊朗战争油价冲击中收紧政策", "summary": "2026年6月，欧央行（+25bp至2.25%）、日本央行（+25bp至1%，1995年来最高）与美联储（按兵不动但点阵图从降息翻转为加息）在一周内集体转鹰，导火索是伊朗战争与霍尔木兹海峡中断推动的油价冲击。美国5月CPI同比4.2%创三年新高，但全由能源（同比+23.5%）驱动，核心仅2.9%。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基准配置应优先通过久期表达去通胀与政策失误风险，权益只做选择性低贝塔，商品低配整体贝塔但保留油价重上100美元的显式尾部对冲。", "summary": "2026-06-29，本卡综合并支持上一卡“央行在消退油价冲击中过度收紧”的判断，但把它转化为更可执行的大类资产配置：久期从35%中性升至45%，权益贝塔从45%降至35%，广谱商品从15%降至12%，现金升至8%。依据是美国5月CPI 4.2%中能源同比23.5%、核心仅2.9%，而美联储仍维持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 1, "risks": 3, "neutral": 0, "value": 81, "events": [ { "title": "在霍尔木兹海峡中断（油价均值105美元）与重开（回落至72-79美元但存在60天到岸刚性）下，美国整体CPI分别面临~82.3 bp上升与~78.6 bp回落的非对称传导，组合应在中…", "summary": "截至2026-06-29，分别在霍尔木兹海峡持续中断（情景A）与谈判重开（情景B）下对布伦特路径建模：A下油价高企于100-120美元，B下纸货崩至72-79美元，但到岸价受60天扫雷物理延迟制约。在0.25的CPI传导系数下，A推高美国CPI同比~82.3 bp阻碍降息，B拉低整体CPI~78.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "summary": "对中国，2026年油价/柴油冲击既是被缓冲的进口税、又是战略加速器，且净额上比卡片01–02的美欧日集团损害更小。三大事实：（1）进口税真实但落在PPI而非CPI——5月PPI同比+3.9%（2022年7月来最快）、CPI仅+1.2%，疲弱内需把冲击吸收为生产者利润压缩；（2）进口税被三重缓冲——…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "summary": "截至2026-06-29，支持卡片01–03的逻辑链并延伸至终端利率与全球流动性维度。美联储6-17维持3.50%–3.75% [S2]、SEP 2026中值3.8% [S3]；新主席Warsh首场FOMC点阵图从降息翻转为加息，9名官员预期加息、年内加息概率50.4%、市场定价10月加息，反应函…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年石油供应中断对全球能源品种及炼油产能的冲击评估", "summary": "支持卡片01“油价持久性是枢纽”的判断并加以锐化。霍尔木兹2025年日均输送约2000万桶/日（原油1500万+成品油550万），全球闲置产能几乎全在海峡之内，绕行管道仅7-8mb/d（不足正常一半），规模化对冲不存在。冲击非线性传导：3月约1200万桶/日无法运抵亚洲、其中800万为中硫油，亚洲…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] } ], "href": "/investor/theses/thesis-e4a70ed58a/", "networkHref": "/investor/research-network/#thesis-e4a70ed58a" }, { "id": "thesis-05c3f6f33c", "source": "auto_topic", "title": "随着TSMC CoWoS产能持续紧张且HBM4于2026年开始量产，从HBM3e向HBM4的跨越将如何影响主要AI GPU厂商（如NVI…", "titleEn": "Practical research topic: 随着TSMC CoWoS产能持续紧张且HBM4于2026年开始量产，从HBM3e向HBM4的跨越将如何影响主要AI GPU厂商（如NVI…", "lane": "AI 基础设施", "laneEn": "AI 基础设施", "laneHref": "/investor/lanes/ai/", "investableFocus": "", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "AMD", "name": "AMD", "market": "US", "role": "加速器替代供给", "roleEn": "加速器替代供给", "bucket": "core", "href": "/investor/stocks/AMD/" } ], "stockTickers": [ "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ", "APO", "IEF", "AMD" ], "coreView": "该主线来自 4 条研究记录的标题、问题和摘要聚类，时间跨度 2026-04-24 至 2026-06-17；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-04-23", "lastSeen": "2026-06-29", "supports": 11, "risks": 137, "neutral": 12, "analystCount": 12, "sourceTypeCounts": { "研究线程": 5, "研究笔记": 153, "证据更新": 1, "分析协作": 1 }, "topTerms": [ "NVIDIA", "CoWoS", "先进封装", "TSMC", "HBM4", "HBM", "GPU", "HBM3e", "随着TSMC", "量产", "主要AI", "厂商" ], "whatChanged": [ "2026-06-29: 先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析", "2026-06-29: AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "2026-06-28: 在先进封装良率瓶颈与表后电力并网延迟的双重物理约束下，A股AI硬件板块估值极度偏高且存在变现拥挤，难以独立承担成长杠铃的对冲配置功能，应当实施防御性减配。", "2026-06-28: GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "2026-06-28: 电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。" ], "practical": { "sectorLane": "AI 基础设施", "sectorLaneEn": "AI 基础设施", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "AMD", "name": "AMD", "market": "US", "role": "加速器替代供给", "roleEn": "加速器替代供给", "bucket": "core", "href": "/investor/stocks/AMD/" } ], "stockTickers": [ "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ", "APO", "IEF", "AMD" ], "primaryStocks": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" } ], "instrumentTags": [ "NVDA", "NVIDIA", "TSM", "TSMC", "AVGO", "Broadcom", "MU", "Micron", "ANET", "Arista Networks", "VRT", "Vertiv" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 101 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 12, "recentEvents": 101, "supportRatio": 0.07, "riskRatio": 0.86, "conflictScore": 95.6, "evidenceDepth": 100 }, "score": 67.7 }, "events": [ { "date": "2026-06-29", "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,800万美元（约降33%）[S1][自测算]；先进封装（CoWo…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM4", "HBM", "HBM3e", "SoIC", "CoWoS-L" ], "href": "" }, { "date": "2026-06-29", "title": "先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析", "summary": "截至2026-06-30，本卡片压力测试Research note 08的能效转型论。结论：CoWoS原始产能已非约束瓶颈——TrendForce判断供需缺口随产能扩张从约20%收窄至2026年底约10%[S1]，台积电产能翻四倍至约13万片/月并外包约24-27万片/年给Amkor/SPIL[S3][S4]，疏解时钟约12-24个月，远快于电网3-7年。…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM4", "HBM", "SoIC" ], "href": "" }, { "date": "2026-06-28", "title": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "summary": "电力与电网容量成为稀缺投入后，AI 芯片优化目标从绝对算力转向每瓦性能（perf/W）与单机柜 token 经济性。NVIDIA 宣称 Blackwell 推理能耗较 Hopper 最高低 25×（FP4/NVFP4 驱动），但 B200 单卡功耗仍升至约 1,200 W（H200 为 700 W），机柜功率密度升至 120–142 kW 并规划 1 MW…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 31.25, "matchedTerms": [ "NVIDIA", "CoWoS", "先进封装", "HBM4", "HBM" ], "href": "" }, { "date": "2026-06-28", "title": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "summary": "截至2026-06-28，本卡否定“AI资本开支韧性等于A股防御性配置锚”的宽泛结论。Microsoft、Alphabet、Meta、NVIDIA、TSMC等数据仍支持AI基础设施真实需求，但FOMC 3.50%-3.75%、5月PCE同比4.1%、10年美债4.40%和A股两融突破RMB 3tn，意味着长久期成长股、拥挤主题和融资敏感供应商仍可能估值下修…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31, "matchedTerms": [ "NVIDIA", "先进封装", "TSMC", "HBM", "AI硬件" ], "href": "" }, { "date": "2026-06-28", "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗ASIC。先进封装成为泄压阀，但TSMC已提示先进封装产能紧张…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27, "matchedTerms": [ "CoWoS", "先进封装", "TSMC", "HBM", "NVL72" ], "href": "" }, { "date": "2026-06-28", "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM合约价Q2环比+58–63%)、以及电网电力(美国并网排队约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM", "GPU" ], "href": "" }, { "date": "2026-06-28", "title": "在先进封装良率瓶颈与表后电力并网延迟的双重物理约束下，A股AI硬件板块估值极度偏高且存在变现拥挤，难以独立承担成长杠铃的对冲配置功能，应当实施防御性减配。", "summary": "本卡压力测试否认了AI硬件作为成长杠铃关键一环的论据。测试显示AI硬件期望收益为-3.43%，压力下变现滑点高达16.82%，且2nm GAA/HBM4封装成本激增113%抬高TCO门槛。此外，16GW计划容量中仅5GW在建，电力并网延迟拖累电网设备行业盈利收缩达36.8%至60.2%。建议将AI硬件配置从12%减至4%，规避流动性瓶颈。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "先进封装", "HBM4", "HBM", "AI硬件" ], "href": "" }, { "date": "2026-06-27", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交期[S6][S7]限制的是通电而非下单。CoWoS售罄、交期…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM", "GPU" ], "href": "" }, { "date": "2026-06-27", "title": "在2026-06-27，AI电力并网限制（2年延误拖累53.71亿美元[S12]）将倒逼算力采购削减并重创高端半导体高增长假设，投资者应果断削减AI硬件至12%及长债至5%，转而增配25%短…", "summary": "截至2026-06-27，本报告重新校准了在高利率平台与电力之墙挤压下的多资产配置。2年电网并网延误对10万卡Blackwell智算集群将产生53.71亿美元搁浅资金损失[S12]，迫使云巨头在2027年削减20%GPU订单。这将导致台积电3nm和2nm产能利用率分别跌至85%与80%，CoWoS封装降至82%[S12]，HBM面临34.3%产能过剩且合约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.75, "matchedTerms": [ "CoWoS", "HBM", "GPU", "AI硬件" ], "href": "" }, { "date": "2026-06-25", "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元在手订单）；第二档择优超配高端光互联（1.6T之年、龙头锁单…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM", "GPU", "厂商", "AI硬件" ], "href": "" }, { "date": "2026-06-25", "title": "电力稀缺不会杀死AI芯片周期，但会把竞争重心从单纯峰值算力转向固定兆瓦内的有效吞吐和能效，从而收窄可投资赢家。", "summary": "截至2026-06-25，本卡对电力和并网约束做半导体侧压力测试：约束不会终结AI芯片需求，但会把胜负指标从峰值FLOPS推向每兆瓦吞吐、每机柜token、HBM每瓦带宽和部署可用性。NVIDIA的机柜级平台、Google TPU、AWS Trainium、TSMC背面供电和Micron HBM4都显示产业已转向能效优化。结论是保留AI存储、高端互联、先进…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31, "matchedTerms": [ "NVIDIA", "先进封装", "TSMC", "HBM4", "HBM" ], "href": "" }, { "date": "2026-06-24", "title": "电力瓶颈下的半导体能效革命：先进封装与低功耗架构的边际驱动力", "summary": "本卡支持并延伸第04卡。电力稀缺把AI算力的经济单位从“每颗芯片”转为“每瓦token”，正驱动厂商以CoWoS、HBM和系统级先进封装提升单位功耗算力。NVIDIA以Vera Rubin机柜级相对Grace Blackwell的10倍单位功耗推理吞吐、MoE 8倍能效、六代百万倍每兆瓦吞吐为卖点，且增益主要来自封装与内存而非3nm逻辑制程。台积电CoWo…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32.25, "matchedTerms": [ "NVIDIA", "CoWoS", "先进封装", "HBM4", "HBM", "厂商" ], "href": "" }, { "date": "2026-06-24", "title": "算力芯片设计从“追求极致算力”向“追求能效比”转型的技术路径与产业链机遇", "summary": "本卡支持前四卡“材料/电网/电力墙造成算力供给缺口”的主线，并从半导体供给侧给出答案：芯片设计已从追求极致算力转向追求能效比，先进封装(CoWoS产能约3.5万→13万片/月)、Chiplet/UCIe(0.25–0.5 pJ/bit对比片外DRAM 6–8 pJ/bit)、HBM4(带宽+75%而功耗仅+20–30%)、N2(同性能功耗-25–30%)与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 31.25, "matchedTerms": [ "NVIDIA", "CoWoS", "先进封装", "HBM4", "HBM" ], "href": "" }, { "date": "2026-06-24", "title": "先进封装与液冷以2026-2027节奏真实落地、兑现约20-50×能效与约5×功率密度，但只松绑芯片/机柜边界、并因Jevons把节省再投资为更多算力，因此它们是前序'物理基础设施约束'论点…", "summary": "截至2026-06-24，本卡支持Research note 04-06的判断，并从硅侧回答任务问题：先进封装与液冷的落地节奏真实且加速，但只松绑芯片/机柜边界，无法废除电网与取水之墙。台积电CoWoS产能三年约10倍（13k→130k片/月，缺口20%→10%）；SK hynix/三星HBM4E 16Gbps、能效+20%；直触芯片液冷强制化（GB200…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27, "matchedTerms": [ "CoWoS", "先进封装", "HBM4", "HBM", "NVL72" ], "href": "" }, { "date": "2026-06-24", "title": "中国CSP成本上升背景下，AI算力基础设施供应弹性与扩容压力分析", "summary": "压力测试部分否定线程以电力为中心的框架。AI算力系统是供给绑定而非需求绑定：2026年超大规模厂商资本开支同比+77%至约7250亿美元[S1]，英伟达Blackwell售罄至年中、积压360万颗[S2]，因此约束的是配额优先级而非需求。核心锚定集群凭借锁定电力、GPU配额与表后自发电被隔离；放缓真实但集中在长尾与2027边际产能（约20%项目滑期、美国约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM", "GPU", "厂商" ], "href": "" }, { "date": "2026-06-24", "title": "先进芯片架构能突破单位计算量的电力壁垒，但在需求反弹与机柜功率密度上升下，不能单独突破AI factory和区域电网的总量电力壁垒。", "summary": "截至2026-06-25，本卡片对卡片04进行压力测试：Blackwell/Rubin、TPU、N2、CoWoS、HBM4与CPO确实显著提升tokens/MW，NVIDIA称GB200能效较H100提升25x，Rubin相较GB200最高10x tokens/MW，Google称Trillium能效提升67%。但这些改进主要降低单位计算功耗，不能消除总量…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.25, "matchedTerms": [ "NVIDIA", "CoWoS", "HBM4", "HBM" ], "href": "" } ], "timeline": [ { "date": "2026-04-23", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "2026年全球AI芯片供应链的HBM与CoWoS先进封装双重瓶颈", "summary": "2026年全球AI芯片供应链的HBM与CoWoS先进封装双重瓶颈 HBM3E/HBM4产能与台积电CoWoS-L先进封装工位在2026年将如何重新分配,并如何重新定价上游存储(海力士、三星、美光)与封装产业链的估值中枢?", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-24", "supports": 0, "risks": 1, "neutral": 1, "value": 40.4, "events": [ { "title": "2026年2.5D/3D先进封装（CoWoS）产能扩张与HBM4演进：评估次世代AI加速器供应链的稳定性", "summary": "2026年2.5D/3D先进封装（CoWoS）产能扩张与HBM4演进：评估次世代AI加速器供应链的稳定性 随着TSMC CoWoS产能持续紧张且HBM4于2026年开始量产，从HBM3e向HBM4的跨越将如何影响主要AI GPU厂商（如NVIDIA/AMD）的成本结构和交付周期？哪些二级OSAT厂…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026年AI算力供应链瓶颈迁移：从CoWoS封装紧缺到HBM4配额与电力约束", "summary": "2026年AI算力供应链瓶颈迁移：从CoWoS封装紧缺到HBM4配额与电力约束 在SK Hynix已卖空2026年HBM产能、TSMC把CoWoS从约35K扩至130K wpm、Nvidia Vera Rubin下半年放量的背景下，AI硬件供应链的真瓶颈究竟已迁移到哪里？二级市场应如何重新定价上游…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-26", "supports": 0, "risks": 0, "neutral": 1, "value": 47.8, "events": [ { "title": "2026年AI芯片供应链与代理化AI硬件需求分析", "summary": "2026年AI芯片供应链与代理化AI硬件需求分析 随着2026年AI应用进入“代理化”阶段，推理算力需求激增，HBM4与先进封装（CoWoS）的产能缺口如何影响全球AI芯片的供应格局？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "AI算力行情是超级周期中段，三波叠加需求（训练→推理→主权AI）驱动力完整，供给瓶颈维持定价权，短期情绪风险真实但不构成周期终结信号。", "summary": "AI硬件行情处于超级周期中段，而非利好出尽的顶部。三波需求（前沿模型训练、大规模推理、主权AI/企业部署）正叠加复利，供给侧CoWoS封装、HBM4产能及电力基础设施瓶颈在12–18个月内无法解除，定价权保持完整。历史类比指向1997–1998年的互联网基础设施阶段。战术层面，情绪高涨（恐惧贪婪指…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 年 HBM4 推理链条的首要弹性在合格 HBM4，首要出货瓶颈在 HBM4 验证与 CoWoS/interposer/封装基板，而液冷更多是部署瓶颈。", "summary": "本卡支持前序推理需求上行逻辑：288 GB HBM4 GPU 普及后，最大量价弹性在合格 HBM4 供应商，尤其是具备份额提升弹性的 MU 和具备配额权的 000660.KS；但最可能成为 2027 物理出货瓶颈的是 HBM4 验证与 CoWoS/interposer/封装基板的组合。测试设备是高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 2, "neutral": 1, "value": 46.6, "events": [ { "title": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯…", "summary": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯片竞争格局：NVIDIA Rubin对阵云巨头ASIC 本卡主题： WFE资本开支周期——先进封装与HBM设备订单可…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2026年5月，Rubin/MI455X的HBM4堆叠所用CoWoS-L先进封装环节几乎100%位于台湾。", "summary": "2026年5月，Rubin/MI455X的HBM4堆叠所用CoWoS-L先进封装环节几乎100%位于台湾。", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2027 年的风险并非‘芯片抛售’，而是 Tier-1 的利润预警与 NVIDIA 汽车业务软件估值回归硬件逻辑的错配压力。", "summary": "2027 年 Thor 芯片合同（22%）与实际 L3+ 装载（12-15%）的错配将导致非对称财务传导。NVIDIA 受 NCNR 合同保护，硬件收入稳健，但与车企的软件分成（50/50）将面临 30-40% 的递延风险，导致汽车业务估值从 SaaS 逻辑回归硬件。Tier-1 供应商面临严重的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HBM4与CoWoS-L的多年期结构性瓶颈把定价权锁到2027年，但盈利弹性的市场预期差最大的并非SK海力士或ASML，而是京元电、三星（作为认证期权）与Lam Research。", "summary": "在3nm前段产能锁死的背景下，AI BoM正在向HBM4与CoWoS-L结构性迁移：HBM每GB ASP自2023年涨>2.5倍，CoWoS-L tile定价翻倍，且TSMC、存储原厂、云巨头的多年期资本开支与保底价订单把这一定价权锁到2027年。八家公司中盈利弹性最被低估的依次是：京元电（244…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "先进封装材料的瓶颈已从‘量’转变为‘规模化良率’，Rubin级别的超大封装与HBM4混合键合将利润高度锁死在Ibiden、TOK和Fujifilm等少数具备高技术壁垒的头部供应商手中。", "summary": "2026年AI芯片（如NVIDIA Rubin）向4x光罩尺寸及HBM4（16层+混合键合）转型，使ABF载板、CMP抛光液和特种化学品成为核心瓶颈。单个Rubin载板价值量是A100的3.5倍，Ibiden等巨头产能已被预订至2027年。混合键合使CMP步骤翻倍，支撑了Fujifilm等供应商的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 1, "risks": 0, "neutral": 0, "value": 58.4, "events": [ { "title": "成熟制程产能 Beta 的价格清算已结构化，AI 先进封装是其物理隔离的反面，两向量合并强化而非削弱 Research note 02 的'做空产能 Beta、做多认证 Alpha'…", "summary": "截至 2026-05-13，从成熟制程产能扩张与 AI 先进封装两个相反向量进一步论证 Research note 02 的'做空产能 Beta、做多认证 Alpha'判断。向量 A：大陆成熟节点（28/40/55/90nm）在 2023–2026Q1 累计新增约 750–820 kwpm 产能，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 1, "value": 48.8, "events": [ { "title": "算力期货压缩的是商用GPU层租金，先进封装与高带宽互联的物理瓶颈与寡头结构足以将其非商品化毛利结构维持至2027。", "summary": "支持card-01框架并加以收紧：算力期货标准化的是GPU-小时这一输出单位，并未商品化决定谁能以最低成本清算的物理输入。CoWoS先进封装、HBM4、混合键合设备、scale-up交换机ASIC、PCIe/CXL retimer、1.6T/CPO光模块均具18–36个月结构性导入周期，寡头格局清…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "材技一体化对fabless是选择性降权——AI加速器与EDA/IP溢价稳健，射频/功率/模拟/成熟节点向IDM收敛，资金应在设计层内向两个可防御极轮动。", "summary": "材技一体化情景下，fabless估值溢价并非整体崩塌而是结构性分化。AI加速器龙头（NVIDIA、AMD MI、Broadcom ASIC）凭借云厂商需求无弹性和CoWoS/HBM优先配额，能够将晶圆和封装涨价完整转嫁，溢价稳健；EDA/IP分成（Cadence、Synopsys、Arm）现金流与…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "M2M流量爆发的边际成本不会均匀挤压SaaS——它会让薄壳代理编排厂商利润坍缩、让有自有数据的垂直SaaS继续吃到按量计费红利，并把被争夺的利润集中流向芯片、封装与光模块基础设施层。", "summary": "截至2026-05-16，本卡对card-07的SaaS基本面Beta判断做压力测试。结论是有条件成立但显著分化：薄壳代理编排SaaS将因无法转嫁的token成本，在未来12–18个月承受600–1,200 bp毛利压缩；而垂直SaaS、数据合规和API适配器（card-06超配名单）边际成本以存…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 2, "neutral": 0, "value": 42.6, "events": [ { "title": "AI 硬件盈利端脱钩成立，但价格路径脱钩不完整；基本面能缓冲两融与解禁冲击，不能消除冲击。", "summary": "截至2026-05-17，本卡压力测试 Research note 06：AI 硬件基本面强到足以形成先进封装、HBM 供应链、光模块/AI datacom 的窄口径例外，但不足以让广义半导体 beta 脱离 A 股两融拥挤和解禁供给。NVIDIA FY2026 收入 USD 215.938bn、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "H200对华许可的毒丸条款使其对中国云厂只能扮演过渡角色，对跨境AI硬件营收偏空、对国产替代链（SMIC/昇腾/寒武纪）结构性偏多，板块定价应反映'强制分化'而非'复苏催化'。", "summary": "H200对华许可叠加25%关税与BIS领土审计构成'毒丸'结构：单芯片承诺、现场审计、芯片级遥测、再出口禁令使H200对中国云厂成为高摩擦、不可转售、用途受限的过渡载体而非战略基石。我们下修Nvidia 2026年中国数据中心营收至80-100亿美元（vs.一致预期120-140亿），毛利率拖累5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "AI硬件与算力基础设施具备足够的选择性资本确定性，但应买入瓶颈环节和已验证国产替代，而不是泛AI主题贝塔。", "summary": "截至2026-05-18，本卡片支持从大金融防御性收缩转向盈利可见成长。AI硬件具备选择性资本确定性：NVIDIA、TSMC、SK hynix及800G/1.6T光模块已体现高毛利兑现，云厂商2026资本开支提供需求锚。国产化替代真实存在，但确定性集中在服务器、光模块、热管理、电源、PCB和已商业…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HBM 与先进封装的订单能见度（至 2027 年）与强劲的毛利扩张确证了 AI 电子的超强景气度，从基本面切断了年内向海洋经济板块的流动性溢出可能。", "summary": "截至 2026-05-17，TMT 基本面调研确证了 AI 电子板块的高景气度。HBM3e/HBM4 与 CoWoS 先进封装的订单能见度已延伸至 2027 年，且正处于绝对定价权驱动的毛利率全面扩张期。这一坚实的业绩上修闭环为资金的极度拥挤提供了合理支撑，彻底证伪了年内资金将大规模溢出至海洋经济…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 4, "neutral": 0, "value": 33.4, "events": [ { "title": "CoWoS/HBM是配给型而非崩塌型瓶颈，使perf/W曲线弯曲而非折断，并把AI加速器BOM定价权向HBM厂与OSAT倾斜。", "summary": "截至2026-05-18，CoWoS/HBM是继电力之后的下一个物理瓶颈，但属于配给型而非崩塌型。TSMC CoWoS 2026年底约135–150kwpm、2027年底约190–220kwpm，CoWoS-L占比升至55–60%；行业HBM 2027比特供给同比+55–65%，但加速器需求+70…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027-2028年电力缺口不会终结AI芯片周期，但会把下一代AI硬件迭代逻辑从单芯片峰值性能改写为受限MW下的有效AI输出、上线时间和机架级能效。", "summary": "截至2026-05-18，本卡支持前序电力瓶颈主线，但压力测试“硬件从极致性能转向能效”的表述：更准确地说，采购函数会转向每已通电MW的有效AI输出。2027-2028年表后燃气仅4-6GW，无法覆盖AI负荷缺口；DOE/LBNL预计美国数据中心用电2028年达325-580TWh。NVIDIA …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体供应链的先进封装、HBM、EUV、基板和光模块交期同步拉长，与电力侧'承诺 Capex > 现金 Capex'呈现同一指纹，证实 2H26 AI Capex 下修是交付推后引致…", "summary": "用先进封装、HBM、WFE、ABF 基板与光模块交期数据压力测试卡片 01，结论是硅侧呈现与电力侧完全一致的'承诺 > 交付、交期拉长、现金落地右移、订单簿创新高'指纹：CoWoS 2026 配额售罄至 1Q27、ASML EUV 交付排到 2H27、HBM4 在 Rubin 上认证再顺延一个季度…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-05-19，AI半导体供应链尚未出现云CapEx或HBM/CoWoS订单环比下修，算力基础设施交易仍应保留核心仓位但需严密跟踪ROIC与Q2/Q3指引风险。", "summary": "截至2026-05-19，首席策略师提出的敏感触发点有效，但尚未触发。Microsoft、Alphabet、Amazon、Meta的2026年AI基础设施CapEx仍在上修或维持高位，TrendForce估计前八大CSP 2026年CapEx超过7100亿美元、同比约+61%。HBM方面，Micr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 3, "neutral": 0, "value": 40.8, "events": [ { "title": "先进封装与HBM出口管制把存储成本压力的核心AI基础设施层从周期性紧张转为政策驱动的结构性通胀，范围窄但中心位置高，至少持续到2027。", "summary": "截至2026-05-19，美BIS（10/2022、10/2023、10/2024 HBM专项、2025-01 AI Diffusion）+ 日本METI 23项工具 + 荷兰ASML EUV/DUV许可 + 台湾CoWoS流片审批 + 50%关联规则形成的许可证瓶颈，把HBM3E/HBM4与先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力链供应瓶颈已基本解决，支持在 CNH 企稳后回补权重至 50%，利润池正向 HBM4 与先进封装技术前沿发生结构性转移。", "summary": "2026-05-20 卡片支持将 AI 算力链权重回升至 50%。研究显示，HBM3e 良率提升与台积电 CoWoS 产能（年底达 7 万片/月）的实体扩张已基本消除 2024-25 年的瓶颈。虽然日元 carry 解除对日本设备标的造成估值冲击，但其基本面供应受损风险低。HBM4 转型将导致利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "云端算力对HBM及CoWoS产能的虹吸导致边缘端LPDDR5X面临严峻的“存储通胀”，硬件降本逻辑阶段性证伪，高昂的BOM成本将大幅压制端侧AI在2026年的大规模渗透率。", "summary": "截至2026-05-19，我们对端侧AI的存储成本约束进行了深度压力测试。测试表明，该约束比预期更严峻。云端HBM4与先进封装（CoWoS）的大规模扩产对晶圆产能形成“虹吸效应”，导致LPDDR5X/6出现结构性短缺，2026年边缘设备存储BOM成本激增32%-50%。这迫使AI PC与手机涨价5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 2, "neutral": 0, "value": 48.2, "events": [ { "title": "AI算力链盈利可见性已被订单簿三重验证、足以支撑保留成长端的策略；半导体设备应拆分成‘AI直接受益+国产替代超配，纯中国敞口设备与晶圆厂低配’，故第03卡‘不全面切向价值’成立但需要…", "summary": "截至2026-05-21，本卡对第03卡“不全面切向价值”进行压力测试，结论是有条件支持。AI算力链（GPU/HBM/光模块/液冷/先进封装）盈利可见性高且仍在抬升：四大美系hyperscaler 2026年合计资本开支指引上修至USD 360–410 bn、TSMC CoWoS与SK Hynix…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI与半导体能在防御性轮动中维持Alpha，但安全边际只存在于盈利可见度高、供给受限且能把CapEx转化为现金流的瓶颈环节。", "summary": "2026-05-20，本卡支持第4张卡的杠铃配置：AI与半导体可保留为15-20%的Alpha袖珍仓位，但不能作为泛防御Beta。SIA/WSTS显示2026年一季度全球芯片销售2985亿美元，环比+25%；TSMC 1-4月收入同比+29.9%，并指引2026年美元收入增长30%以上；NVIDI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "在英伟达 Rubin 平台面临供应链瓶颈的窗口期，国产 GPU 凭借先进封装与 HBM3 的全产业链突破，正通过‘十五五’初期智算中心的大规模采购实现对推理市场的实质性支配。", "summary": "截至2026-05-22，本研究确认国产GPU已进入性能与市场双重拐点。NVIDIA Rubin平台受困于HBM4验证延迟与CoWoS-L稳定性，2026年产量目标下调25%。与此同时，海光深算4号与寒武纪思元690通过Chiplet技术实现性能跃升，结合长电、通富在2.5D封装的突破以及长鑫存储…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 1, "neutral": 0, "value": 65, "events": [ { "title": "NVIDIA第一季度财报超预期与AI基础设施资本开支展望：黑曜石平台（Blackwell）主导地位与CoWoS产能博弈", "summary": "截至2026年5月21日，NVIDIA于昨日发布的Q1 FY2027营收录得816亿美元，同比增长85%，显示出AI算力需求极强。2026年海外四大云巨头资本开支预计突破7000亿美元，其中约75%投向AI。与此同时，台积电CoWoS先进封装产能预计在2026年底提升至12-13万片/月，NVID…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI资本开支足以抵消重工业衰退冲量并能驱动半导体层利润率周期性走高，但云基础设施层利润率短期反而受压——卡片4的'TMT整体利润率扩张'标签需被分层重写为'半导体↑/云↓'。", "summary": "对卡片4双重命题做分层压力测试：$3,750亿AI资本开支能抵消OFS/重工资本下行（算术差距62倍，支持）；但'TMT结构性利润率扩张'必须拆成上下游——上游半导体（NVDA、TSMC、HBM三巨头）凭CoWoS/HBM4供给瓶颈+客户付款意愿，毛利率扩张是真命题；下游云基础设施（Azure/A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 1, "risks": 0, "neutral": 0, "value": 72.4, "events": [ { "title": "截至2026-05-24，AI硬件与半导体设备是A股结构性成长修复最可信的核心驱动力，但只能买订单和盈利修复，不能买泛科技估值扩张。", "summary": "截至2026-05-24，本卡支持ashare-strategist的结构成长修复判断，但将可投资主线收窄为有订单验证的AI硬件与半导体设备。NVIDIA、Microsoft、Meta、Amazon、TSMC、ASML、SEMI和SIA数据共同显示，全球AI capex、先进制程、先进封装、测试与…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 2, "neutral": 0, "value": 53.8, "events": [ { "title": "AI硬件供应链呈现极端的“利润肥头、制造瘦脚”特征，核心瓶颈（先进封装与HBM）的高壁垒与溢价能力固化了这种失衡，迫使重资产中下游供应商承担了不对称的资本支出和运营风险，极易在需求周…", "summary": "截至2026-05-24，本报告深入评估了AI硬件供应链的利润分布、技术壁垒与资本开支压力，对供应链结构性风险进行了应力测试。研究显示，供应链利润高度集中于英伟达（GAAP毛利率达74.9%）与台积电先进封装（CoWoS毛利率达65-70%）。中下游厂商（如外包封测OSAT、散热及代工厂）利润微薄…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "TMT / AI硬件已从政策主题进入盈利兑现阶段，足以承接医药降配资金，但应先投100-150bps并等待A股拥挤度验证后再投满。", "summary": "截至2026-05-24，支持将医药降配释放的风险预算轮动至TMT / AI硬件，但建议分阶段执行。政策端，“两新”和“人工智能+”继续提供订单底；数据端，2026年1-4月设备工器具购置投资同比增长11.5%，1季度电子设备制造业利润同比增长1.2倍；全球端，NVIDIA、TSMC、ASML均显…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 0, "value": 55.2, "events": [ { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "现有先进半导体出口管制不会停止AI使用或全部ASIC设计，但会系统性削弱受限生态通过HBM密集、CoWoS级封装实现“每瓦有效推理”跃迁的能力。", "summary": "截至2026-05-25，本卡压力测试前序“ASIC与先进封装推动每瓦有效推理”的判断。结论是：路径对可获得TSMC、HBM与CoWoS产能的全球云厂商仍成立，但对中国相关或D:5受限生态被结构性削弱。美国出口管制已从成品AI芯片延伸至HBM、封装路径、先进制造设备、软件/技术和FDP规则，卡住了…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 3, "neutral": 0, "value": 50.6, "events": [ { "title": "短期硬约束在 HBM 与 CoWoS-L 而非电力，AI 算力'紧但能出货'，对指数 FOMO 有利但 2026H2 脆弱性显著上升。", "summary": "短期（2026 年内）芯片侧的约束比电力更紧：依次瓶颈为 HBM3e bit 供给 → 台积电 CoWoS-L 先进封装 → GPU 晶圆 → 基板/光模块 → 电力。HBM 与 CoWoS-L 全年额度已锁、扩产跑不过 ~150%/~70% 的需求增长，2026H2 任一良率事件或出口管制重置可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体瓶颈不在 GPU 而在 CoWoS-L 与 HBM，光模块瓶颈不在出货而在 1.6T 交付周期与客户独家配套——这两条链是 Research note 06 中段瓶颈租框架下 …", "summary": "2026-05-26：本卡 support Research note 06 的'中段瓶颈租'框架，并在 TMT 半导体与光模块层将之定量化。核心结论：CoWoS-L 月产能从 2025 年约 40kwpm 升至 2026 约 75kwpm 仍被超额预订（覆盖率 1.4–1.6x），HBM 长协覆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供给的物理限制已内化为芯片架构的强制约束，AI 硬件溢价正从单纯的算力规模转向“单瓦性能”与“封装级电源集成”的确定性。", "summary": "截至 2026-05-26，电力设备交付延迟（变压器周期达3-5年）已迫使 AI 硬件架构从“性能优先”转向“功率密度优先”。NVIDIA Rubin 架构通过 HBM4、800V HVDC 和集成电压调节器 (IVR) 应对挑战。这引发了 SiC/GaN 的产能紧缺，并暴露了先进封装材料（如日东…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 6, "neutral": 2, "value": 40, "events": [ { "title": "4.5%以上美债环境下，AI硬件龙头的业绩增长足以支撑选择性配置，但不足以为整个AI久期交易兜底。", "summary": "2026-05-28结论：本卡对上一张宏观卡采取stress-test但不否定。4.50%美债锚会压缩AI久期资产估值，但NVIDIA、TSMC、Broadcom、SK hynix/Micron等稀缺硬件链凭借高增长、真实云厂商capex、先进封装和HBM瓶颈，仍能部分抵消估值下修。风险在于cap…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "区域化认证时滞与多供应商材料栈分裂打破了先进封装物理结构上的“黄金匹配”，其引发的热力学翘曲与 12-18个月的合规验证使得 AI 芯片先进制程量产风险实质性飙升。", "summary": "截至2026-05-27，本卡通过压力测试确认：区域化合规与材料栈重构不仅仅是法务摩擦，更打破了原后台积电生态的热力学一致性。由于不同区域、多供应商（SK Hynix/Samsung/Micron）的材料存在微小物理差异，使得 CoWoS-L 和 HBM4 在超过 1000W 功耗下遭遇极端的 C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "台积电CoWoS与HBM底层封装良率构成全球AI芯片出货的硬性物理约束，超额利润集中于垄断性瓶颈厂商（TSMC/MU/SK Hynix），而A股代理股共享叙事却无壁垒且交易极度拥挤，…", "summary": "支持超配海外半导体物理瓶颈（MU/SOXX）并降配A股算力代理。台积电CoWoS封装产能从2024年底3.5万片/月扩至2026年底超12万片/月但仍售罄，NVIDIA垄断超60%份额；HBM市场2026年规模将达600亿美元，美光产能售罄且良率升至70-75%。A股光模块、服务器组装与IDC代理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HBM3E面临产能过剩风险，而HBM4的逻辑底晶粒与混合键合成为新瓶颈；建议在14%的先进封装配置内，低配纯存储厂商，超配代工及前端设备商。", "summary": "本报告基于2026年5月的最新产能数据，对策略师的先进制程超配逻辑进行了压力测试。我们发现，虽然台积电CoWoS产能到2026年底将激增至13万片/月，有效缓解了绝对瓶颈，但HBM市场正面临分化：三星激进扩产（25万片/月）导致HBM3E存在价格战风险，而HBM4因引入逻辑底晶粒与16层混合键合，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 2, "neutral": 0, "value": 59.4, "events": [ { "title": "2026年AI芯片与先进封装产能约束会推高并拉长云端CapEx，而不是使其取消；配置应偏向具备配额权的上游瓶颈环节。", "summary": "截至2026-05-28，本卡支持前序判断：AI硬件CapEx传导链的约束真实但不等于削减。Microsoft、Amazon、Alphabet、Meta的2026年CapEx指引中枢约为$710 billion，反映的是芯片、HBM、先进封装、电力和数据中心资源稀缺导致的锁单与涨价，而非需求走弱。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与液冷散热的产能扩张确实进入10×量级周期，但受ABF/HBM/稀土/铜的上游约束，仅能在制造维度缓解约35–40%的三重瓶颈，无法证伪TMT分析师的结构性紧张判断。", "summary": "工业制造视角支持TMT的三重瓶颈框架。TSMC CoWoS月产能从2023年末13K wpm扩张至2026年末130K wpm（近10×），但CoWoS-L/S全年售罄、NVIDIA独占60%以上配额，ABF载板与HBM堆叠的18–24个月扩产周期意味着封装紧张延续至2027年中。液冷市场2026…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 0, "value": 60.8, "events": [ { "title": "AI硬件周期仍由需求驱动，但未来12个月的定价权将更多落在HBM、先进封装、光模块和电力交付等瓶颈环节，而不是单纯GPU或服务器组装收入。", "summary": "截至2026-05-30，本卡支持“AI算力瓶颈从硅端扩展至物理基础设施”的主线。未来12个月，AI芯片和服务器需求仍由超大云厂商资本开支、云AI收入、agentic AI推理流量和自研ASIC扩张驱动；但供应瓶颈已从GPU单点扩展到HBM、CoWoS/先进封装、800G/1.6T光模块、机架供电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI产业链二阶环节（先进封装、液冷电源、ABF基板与高层数PCB）正在交付可审计的硬利润，瓶颈在二阶产能而非终端需求——这就让Research note 01的Layer-3替代对冲…", "summary": "AI硬件二阶环节在2026-05-30的产能与利润率证据三路并发：TSMC CoWoS-L 2026年售罄、2027仍在排队；Vertiv订单/RPO覆盖超4个季度且营业利润率扩张至接近20%；ABF基板与高层数PCB（Ibiden、Unimicron、深南电路、生益科技）AI级毛利率相对普通级的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 3, "neutral": 0, "value": 64.2, "events": [ { "title": "2026 HBM 缺口真实但租金窗口前置在 1H26，真正的稀缺量子是 CoWoS 先进封装而非 HBM 裸晶——杠铃方向成立但工具应以离岸为主、上半年加仓、Q4 减仓。", "summary": "截至 2026-05-30，本卡对 Research note 05 杠铃配置的 HBM 瓶颈腿进行供应链侧压力测试。2026 HBM 毛缺口约 10% 真实存在但正快速收敛——SK Hynix、美光产能已预售，三星 HBM3E-12H 已通过 NVIDIA 认证、HBM4 于 1Q26 送样，2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "非美系低轨星座的自主化上限不是芯片设计口号，而是能否获得CoWoS/HBM/材料产能并完成空间级资质认证。", "summary": "截至2026-05-30，本卡片支持前序判断：非美系低轨星座的AI硬件瓶颈已从单一GPU采购，转向CoWoS/HBM/基板/材料/测试良率组成的封装级工业体系。监管规避可以降低对美国加速器的直接依赖，但无法自动获得先进封装、HBM和材料产能；Eutelsat替换周期与Qianfan部署压力显示，星…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年半导体的结构性毛利重估完全由AI服务器（HBM与先进封装）的高壁垒和产能紧缺驱动，与国内消费电子补贴的存量博弈完全脱钩。", "summary": "截至2026-05-30，本卡支持“半导体毛利扩张脱钩于消费补贴”的判断。研究表明，2026年AI服务器出货量预计同比增长超28%，其价值量将占服务器总市场的74%。SK海力士HBM产能已售罄，HBM4在2H26带来超30%溢价，毛利率维持极高水平；台积电CoWoS产能2026年底将突破11.5万…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "中国本土先进封装放量确定性降低，会选择性强化全球半导体设备与材料供应商的稀缺性溢价，并通过延长瓶颈周期和抬高重置成本支撑AI硬件利润重估。", "summary": "截至2026-05-30，本卡支持“出口管制强化全球设备与材料稀缺性溢价”的判断，但强调这是选择性受益而非全行业普涨。中国本土HBM/先进封装放量确定性下降，会降低低价边际供给冲击全球市场的概率；真正受益的是进入合规非中国HBM、先进逻辑、类CoWoS封装、测试、基板和高纯材料认证体系的供应商。对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 3, "neutral": 1, "value": 60.6, "events": [ { "title": "AI 算力供给受限于 HBM4 售罄、CoWoS 封装良率及全球电网容量的物理极限，使硬件龙头成为 AI 价值链中无视‘Capex 断崖’风险的最终收费站。", "summary": "截至2026年5月，AI硬件供应链仍处于结构性紧缺。TSMC N2量产初期良率65-70%，HBM4于2月投产且美光2026年产能已售罄。NVIDIA Rubin架构推升单GPU功耗至2.3kW，使电力供应成为比芯片更硬的瓶颈。尽管终端应用出现DAU增速放缓，但硬件端的实物稀缺性构成了核心护城河，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超级IPO将把AI公司的公开融资压力传导为算力锁定需求，从而重估半导体资本开支：超配稀缺瓶颈，回避无法证明产能稀缺和合约期限的长尾扩产。", "summary": "截至2026-05-31，本卡片支持“超级IPO流动性虹吸”对AI硬件和半导体瓶颈资产的强化作用。核心判断是，IPO资金不会平均利好半导体，而会把公开市场资金约束转化为GPU、HBM、先进封装、先进制程、光刻和数据中心网络的多年期需求可见度。TSMC、NVIDIA、ASML、Micron、Sams…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI硬件足以证明哑铃结构需要一个选择性进攻仓位，但进攻端必须窄化为稀缺AI基础设施瓶颈，而不是泛TMT或泛成长。", "summary": "截至2026-05-31，本卡片压力测试认为：AI硬件与半导体链条能提供真实但窄口径的TFP弹性，足以支撑A股哑铃结构的进攻端，但不能验证泛TMT贝塔，也不能逆转3.25-3.45%的宏观潜在增速下修。全球半导体、NVIDIA、TSMC和中国云端AI收入显示算力需求与制造效率真实存在；但资本开支、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "滞胀成本压力会侵蚀但不会基准性摧毁Regime A SaaS利润率，真正的分水岭是厂商能否把AI成本从固定席位打包迁移到明确的用量/结果计价。", "summary": "截至2026-06-02，本卡压力测试能源与大宗上涨对AI硬件供应链和Regime A SaaS利润率的传导。结论是有条件支持前序Regime A逻辑：晶圆厂电力成本上涨会压低硬件毛利，但单独不足以击穿SaaS Agent利润驱动；真正高Beta风险来自HBM、先进封装、数据中心电力与云厂商GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 0, "neutral": 1, "value": 81.4, "events": [ { "title": "Research note 07 · 高端封装与 HBM 产业链上游材料的供应稳定性与毛利弹性分析 工作基准日： 2026 06 03（Asia/Singapore） 分析师：材料行…", "summary": "Research note 07 · 高端封装与 HBM 产业链上游材料的供应稳定性与毛利弹性分析 工作基准日： 2026 06 03（Asia/Singapore） 分析师：材料行业分析师（materials analyst） 卡片编号：7 / 10 立场：support（延续 Research…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 5, "neutral": 0, "value": 49.8, "events": [ { "title": "CoWoS产能硬约束会把AI硬件估值溢价从单纯设计叙事重新分配给掌握晶圆、封装、良率和整柜交付能力的产能控制者。", "summary": "2026-06-03：本卡支持前序判断，认为CoWoS与先进封装已成为AI算力链估值的硬过滤器。NVIDIA需求仍强，TSMC的3 nm、HPC与先进封装紧张共同支撑产能控制者溢价；但HBM4供应商扩容削弱单纯存储稀缺溢价。A股半导体整体估值已高，策略重点应从宽泛芯片设计beta转向设备、材料、封…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链的绝对瓶颈正在从HBM存储晶圆短缺结构性转移至由TSMC主导的CoWoS-L先进封装尺寸扩张和逻辑基础晶圆制造代工能力。", "summary": "本报告分析了NVIDIA在2026年6月确认Samsung、SK Hynix和Micron全面进入Rubin平台的HBM4供应链这一关键拐点。我们指出，随着三足鼎立打破垄断，HBM供需瓶颈将显著缓解。然而，由于HBM4基础晶圆逻辑代工的双重锁定，以及CoWoS-L封装面积扩张对晶圆利用率(GDW)…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与材料的刚性约束已把AI优化重心强制推到HBM4、3D混合键合、CPO光学与MoE/FP4等芯片级架构层；这些改进在2026-2030年可把单位能力设施功耗降低15-25%，足以…", "summary": "截至2026-06-03，本卡支持上一张材料卡的判断，并将瓶颈传导至芯片与封装栈。HBM占封装功耗30-50%，HBM4（SK海力士、Micron 2026年量产）以约2.3 pJ/bit较HBM3E能效改善30-40%；TSMC SoIC/CoWoS-L把跨die搬运能耗压低约10倍；NVIDI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Broadcom (AVGO) 财报验证了AI订单总需求未断裂，但低毛利自研芯片的占比扩张和物理并网/封装刚性约束，正在系统性拉低硬件供应商的边际利润率并平滑其收入转化率，驱使AI算…", "summary": "截至2026年6月3日，Broadcom (AVGO) 发布Q2财报，其AI半导体营收111亿美元超预期，且上调全年指引至145亿美元，证伪了需求崖断裂假说；但其毛利率受低毛利定制加速器 (ASIC) 占比飙升影响，环比大跌290个基点至75.1%低于预期，且受限于CoWoS先进封装与北美电网并网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 0, "neutral": 1, "value": 92.2, "events": [ { "title": "AI算力并非材料降级免疫——降级真实存在但落在机柜电源、封装吞吐与设备耗材层，以2027年GPU封装短缺8-25%、机柜效率损失3-7%、超大规模厂商每年USD 6-12亿opex拖…", "summary": "石墨与HRE的中游缺口不会以晶圆厂停工出现，而以三条具体通道命中AI算力：CoWoS备件交期拉长使2027年GPU封装短缺8-25%（基准-不利）；ASML EUV机群因Dy/Tb电机备件影响可用率下降2-4%；Ga/Ge收紧迫使机柜电源从SiC/GaN降级至Si IGBT，三级转换效率损失3-7…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Chiplet 与 3D 先进封装可在 18–24 个月内把国产 AI 芯片单卡等效算力推回旗舰 60–75% 区间，是先进制程受限下唯一可兑现的非对称工程路径，但无法消除 HBM …", "summary": "2026-06-04：Chiplet + 2.5D/3D 先进封装已成为国产高性能 AI 芯片在 7nm/N+2 被锁条件下的唯一可量产非对称路径。长电 XDFOI、通富微电、华天科技、盛合晶微在 2.5D CoWoS-like 与 9μm 混合键合上小批量验证；华为昇腾 920、寒武纪 MLU5…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 76.6, "events": [ { "title": "HBM与CoWoS的瓶颈真实但明显弱于电力设备，30%算力核心权重应维持不变，但内部应从NVIDIA+SK海力士集中转向Micron+三星对冲以应对2H 2027份额再分配与供过于求…", "summary": "截至2026-06-05，HBM3e/HBM4与CoWoS-L在2026年仍紧张但2H 2027趋于平衡：台积电CoWoS月产能从2024年底~35-40k wpm扩至2026年底~75-80k wpm、2027年~130-140k wpm；HBM行业比特位2026年同比+60%、收入升至USD …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 4, "neutral": 0, "value": 60, "events": [ { "title": "CoWoS-L 产能按计划打开, 保住了 Rubin 的量, 但 2026–2027 年毛利率剧本的主笔已交棒给 HBM4。", "summary": "TSMC CoWoS-L/R 产能从 2025 年末约 75 kwpm 扩至 2026 年末约 135–140 kwpm, 已足够支撑 NVIDIA Vera Rubin 在 4Q26 NVL144 量产并留单季度缓冲。CoWoS 单位成本仅是二级毛利率杠杆, 一级杠杆是 HBM4 ASP——15…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SK hynix HBM4 12-Hi良率大概率足以支撑Rubin初始交付，但不足以排除2026年Q4至2027年Q1放量斜率受压。", "summary": "截至2026-06-06，公开证据不支持SK hynix HBM4 12-Hi已出现确定性良率失误；更合理的压力点是Rubin放量斜率。NVIDIA已称Vera Rubin进入全面量产爬坡，并有三家HBM4供应商获认证/生产。SK hynix的1bnm、Advanced MR-MUF、TSMC b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BTM自备电对AI硬件利润率的支撑是真实但间接的，最大受益者是上游加速器、先进封装和电源/液冷内容，而非通用服务器组装。", "summary": "截至2026-06-06，本卡支持“能源瓶颈支撑BTM与AI硬件链”的主线，但强调传导不是服务器直接提价，而是保护数据中心通电进度、硬件出货节奏和先进封装/晶圆代工利用率。PJM与ERCOT大负荷接入压力、FERC共址规则、Oncor数据中心队列，以及NVIDIA、Dell、TSMC最新披露共同显…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Rubin放量斜率放缓要求资金从单纯的“全球硬件Beta”向“先进封装价值捕获”与“本土供应链替代Alpha”进行战术轮动。", "summary": "鉴于HBM4 12-Hi良率瓶颈主要压制NVIDIA Rubin在2026年Q4至2027年Q1的放量斜率（早期机柜产出或折损15%-21%），而非初始交付，我们建议采用“哑铃型”配置策略。一方面，对全球AI硬件龙头保持核心底仓，但战术性下调权重以规避短期估值消化；超配台积电等先进封装纯标的。另一…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 1, "risks": 2, "neutral": 0, "value": 81.4, "events": [ { "title": "7月关税与海运上行更可能压缩AI硬件下游组装与网络设备利润率，而非击穿核心半导体龙头利润率；估值风险来自低利润率硬件收入增长的质量被重新定价。", "summary": "截至2026-06-08，本卡对第04卡的关税/海运鹰派尾部做TMT压力测试：风险真实存在，但主要不是核心GPU、晶圆代工或HBM利润率同步塌陷，而是下游AI服务器、网络设备、机柜、电源、散热和PCBAs的落地成本挤压。NVIDIA、TSMC、SK hynix拥有高毛利率和稀缺定价权；Dell式A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI与先进算力的融合已将药物研发从经验驱动的博弈转变为高确定性的工程学科，2026年产能短缺的半导体基础设施已成为制药巨头最核心的竞争壁垒。", "summary": "2026年ASCO年会标志着药物研发进入“工程化”时代。AI设计的双特异性抗体（如Galux的IL-18v）在临床上展现出极高的选择性，而Ivonescimab成功挑战Keytruda的一线地位。TMT基础设施成为核心瓶颈，台积电的CoWoS-L和SK海力士的HBM4在2026年底前产能已售罄。生…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "头部CSP的AI投资回收期在HBM4紧缺下并未恶化，反而因稀缺产能高利用率和架构效率提升而缩短，支撑AI硬件的选择性估值溢价。", "summary": "截至2026-06-07，本卡支持并综合前序结论：HBM4紧缺不会打断头部CSP的AI资本开支周期，而是强化“配给能力+效率改善+现金回收”的分化。Microsoft、Alphabet、Amazon披露的需求、积压订单和AI收入信号表明，头部云厂商的有效回收期可能压缩至约1.6-2.3年；Meta…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 2, "neutral": 0, "value": 74.8, "events": [ { "title": "Rubin Capex能保护一线AI硬件EPS，但WACC上行会把二线供应链从“AI成长资产”重新定价为“融资周期资产”。", "summary": "截至2026-06-08，本卡对前序AI硬件防御结论做压力测试：Rubin周期与云巨头Capex仍能支撑NVIDIA、TSMC、Broadcom、Samsung、Micron等一线硬件龙头的EPS增长，但不能让整个供应链免疫WACC上行。可见四大云厂商Capex底线约5742亿美元（自测算），足以…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "以Chiplet与先进封装为核心的硬件架构突破是AI推理边际成本系统性通缩的底座支撑，消除了SaaS及网安企业结果计费（OBP）转型的可变成本风险。", "summary": "本报告探讨了Chiplet与先进封装技术（如CoWoS、HBM4、SoIC）对AI推理成本通缩的底层驱动。测算显示，Chiplet异构集成在高性能芯片上可降低42.85%的制造成本；高带宽HBM4内存与CoWoS-L封装突破内存墙，使3卡Rubin集群相较8卡H100集群带宽提升2.46倍、开支减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 3, "neutral": 1, "value": 73.2, "events": [ { "title": "Google-Intel的TPU订单标志着AI硬件利润池的结构性迁移：从绝对依赖GPU/TSMC，转向ASIC出货量反超以及Intel Foundry作为第二供应源的崛起，但GPU仍…", "summary": "2026-06-09评估：Google向Intel Foundry下单2028年超300万颗TPU，且Nvidia正评估Intel 18A工艺，这表明AI总需求依然强劲（到2028年TAM达3800亿美元），但硬件利润池正发生结构性迁移。为降低TCO，云厂商加速自研ASIC，出货量预计将超越GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI半导体利润池迁移真实但渐进，板块总EPS上行、估值锚从'NVDA-TSMC双寡头溢价'切向'增长+多元化第二供应源'，ASIC设计商、HBM、先进封装设备与Intel Found…", "summary": "2026-06-09综合：Google-Intel TPU订单确认AI硬件利润池迁移真实且渐进，2027-28年AI半导体合计EPS较冲击前共识更高，但分布坚定向ASIC设计商(AVGO/MRVL)、HBM、先进封装设备及Intel Foundry期权倾斜。新估值锚为'增长+多元化第二供应源'，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件轮动战略正确但面临Intel 18A良率风险与资金‘业绩陷阱’滞后，需警惕地缘政治对先进封装链条的估值压制。", "summary": "2026-06-09压力测试确认：AI硬件利润池向ASIC与Intel代工的迁移在基本面上已获博通业绩及Intel 18A量产验证，但A股资金流向受1.6T光模块近端业绩吸引而表现出显著的‘轮动滞后’。Intel 18A工艺目前50%的良率是核心风险点，若无法如期突破将导致第二供应源叙事崩塌。A股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 6, "neutral": 0, "value": 53.6, "events": [ { "title": "AI硬件正通过2.3kW单芯片功耗极限与800V液冷架构强行突破物理墙，算力扩张的限制已从芯片设计转移至电力设施配套的48个月长周期交付延迟。", "summary": "截至2026-06-11，AI硬件并未因电力约束而降低TDP，反而迈入2.3kW（NVIDIA Rubin）的“暴力能效”时代。这迫使行业全面转向100%液冷及800V DC机架架构。硬件演进逻辑已从“硅片优化”转向“基础设施重构”，先进封装（CoWoS-L/HBM4）与光互连（CPO）成为维持算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI半导体核心标的的Q2/Q3盈利能见度极高，受限的电网电力将加速Blackwell架构升级，短期流动性冲击构成极佳的左侧建仓窗口。", "summary": "截至2026-06-10，在SpaceX IPO（SPCX）可能引发SMH/QQQ短期30亿至60亿美元流动性抽水的背景下，本报告研判AI半导体核心标的（Nvidia, TSMC）Q2/Q3盈利能见度依然坚实。TSMC 2026年520-560亿美元资本开支高企，CoWoS先进封装产能供不应求。即…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年量子混合架构已具备商业化量产的互连技术标准，但其实际落地规模受制于AI算力对先进封装产能的挤出效应，呈现“技术就绪、产能告急”的结构性矛盾。", "summary": "2026年，量子-经典混合架构的硬件瓶颈已从比特制造转向互连技术与先进封装。NVIDIA NVQLink (<4μs延迟) 与IBM模块化Kookaburra处理器标志着商用化互连的成熟。然而，供应链面临AI GPU对台积电CoWoS/SoIC产能的严重排挤，量子硬件交付周期拉长。Cryo-CMO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年先进封装不是没有扩产，而是扩产首先服务头部AI客户，长尾AI、量子混合和其他HPC项目仍要承受配额挤压。", "summary": "截至2026-06-11，本卡压力测试认为，CoWoS/SoIC扩产是真实且规模化的，但2026年仍不足以完全缓解AI GPU对先进封装的挤出。台积电2026年资本预算指向USD 52-56 billion高端，其中约10-20%覆盖先进封装、测试、光罩及其他；KGI预计2026年CoWoS年产能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 79, "events": [ { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源融资成本上升会削弱云客户ROIC并压低部分ASP，但最强AI硬件平台商仍有足够利润率缓冲，真正脆弱点在重capex制造和周期性内存环节。", "summary": "2026-06-11，本卡压力测试能源债务融资成本上升对AI硬件供应商利润率的影响。结论是方向成立但不应外推为硬件利润率全面坍塌：新增能源债务成本上行约85 bps会压缩云厂商自由现金流并强化采购纪律，但NVIDIA和Broadcom因高毛利、平台锁定和轻资产结构韧性最强；AMD居中；TSMC和M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 0, "value": 68.4, "events": [ { "title": "为支撑三层算力织物，AI芯片与先进封装如何演进以匹配电力波动与边缘推理的功率包络", "summary": "综合并支持卡片04的三层算力织物，下沉到硅基与封装层。关键洞见：'电力波动友好'主要是电力电子＋调度固件问题，而非新晶圆问题。NVIDIA GB300靠电源货架（65 J/GPU储能＋功率封顶斜率＋功率烧除）把Megatron训练电网峰值砍约30% [S1]，DVFS早是标配 [S2]；同步功率摆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力刚性约束已把AI加速器竞争维度切换为每瓦Token，Chiplet+先进封装是2027-2028电网窗口内唯一可规模化兑现的算力密度救援方案，对NPV的杠杆是5-8倍。", "summary": "电网刚性约束把AI加速器的KPI从峰值FLOPS切换为每瓦Token吞吐。Blackwell B200单卡1000W、GB200 NVL72机柜120kW已触顶，下一代Rubin/MI400无法继续抬TDP，必须靠能效换性能。TPU v5p、Trainium2较H100能效提升30-40%；同10…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "进一步出口管制确会拉长芯片时间表并显著退化算力，但几乎完全落在中国云厂商身上；对西方云厂商，封装/HBM瓶颈是商业爬坡风险、不被管制叠加放大，其唯一负面通道是中国对等矿产报复这一自找…", "summary": "对出口管制叠加先进封装/HBM瓶颈做压力测试，结论是风险按地理分叉而非对称。对西方云厂商，进一步的美对华管制不会加剧供应时间表，反因剥离中国需求把韩台HBM4/CoWoS配额导向西方（CoWoS 2026年底扩至13万片/月），故card-05的NPV悬崖不被放大；西方真正尾部是中国镓锗锑许可证（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装/HBM仍是真瓶颈，但2026-06-12的上市交易已经把2027-2028年认证、良率和产能利用率提前资本化，最需要防的是技术代际错配与融资拥挤触发的估值和盈利预期双杀。", "summary": "截至2026-06-12，本卡对先进封装与HBM链条做压力测试：风险真实存在，但不是2026年立即供过于求。HBM到2027年前仍偏紧，核心风险在于HBM4/HBM4E和更大CoWoS形态迭代快于国内上市代理资产认证速度；叠加长电、通富、寒武纪等高估值和高融资余额，股价可能先于EPS下修而去估值化…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 5, "neutral": 2, "value": 69.8, "events": [ { "title": "CoWoS 与 HBM 扩张是 NVIDIA 继续销售高溢价系统的供给条件，而不是摧毁溢价的触发器。", "summary": "2026-06-13：本卡支持前序判断。CoWoS/HBM 扩产不是削弱 NVIDIA 护城河，而是让其把更多机柜级 AI factory 供给转化为计算、内存、网络与软件平台溢价。TSMC 的先进封装、3nm/HBM base die 产能，SK hynix/Micron/Samsung 的 H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红皇后竞速并未彻底抹去自研ASIC的成本优势，而是将其集中至Google与AWS两家、使其他云厂商回到对NVIDIA的依赖，从而部分推翻了Research note 05所述的全行业…", "summary": "在'红皇后竞速'下，自研ASIC单世代有效寿命已压缩至12–15个月，2nm节点全口径NRE攀升至USD 0.9–1.4B、N2晶圆约USD 30,000/片、CoWoS-L售罄至2027。我们测算自研ASIC在TCO上跑赢商用GPU的年部署门槛从5nm节点的约20万颗升至现实情景下的15万–25…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "CoWoS绝对产能正在改善，但GPU与自研ASIC需求吸收速度更快，因此2026年AI集群交付仍受先进封装、HBM和关键设备交期的结构性约束。", "summary": "截至2026-06-13，本卡支持上一张AI基建卡的判断：硬件交付风险不只来自并网，还来自CoWoS/HBM/设备交期的半导体转换链。TSMC仍称先进封装非常紧，TrendForce口径显示CoWoS-L/S满载且2026年底目标120,000-130,000片/月；DIGITIMES虽更乐观，预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "下游电网物理瓶颈将向上传导为半导体的销量天花板，导致其为提升能效比而激增的先进封装与逻辑晶圆资本开支沦为无法有效摊销的“沉没成本”。", "summary": "压力测试确认半导体产业同样陷入资本陷阱。受制于下游数据中心电力天花板，AI硬件的演进核心从算力规模转向能效比（FLOPs/W）。这迫使代工厂和存储商在 CoWoS、CPO 和 HBM4 等节能技术上激增资本开支。然而，由于电网瓶颈导致下游部署滞后，半导体厂商高昂的固定资产投资将面临无量摊销的风险，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 4, "neutral": 0, "value": 71.2, "events": [ { "title": "AI先进封装与HBM产能瓶颈对算力基础设施扩张及超大规模厂商资本开支的影响", "summary": "支持半导体分析师Research note 01的论点并向下游延伸：CoWoS/HBM确为算力的关键约束，2026年部署受封装与内存所限（CoWoS扩至约13万片/月仍订满；SK Hynix与Micron 2026年HBM售罄；HBM4验证延迟使NVIDIA将Rubin目标由200万下调至150万…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GPU抵押贷款收缩会削弱边际Neocloud订单可见度，但先进封装与先进制程需求更可能从融资套利订单转向信用更硬的超大规模厂商和ASIC路线，而非进入全面下行。", "summary": "截至2026-06-15，本卡压力测试GPU抵押融资收缩对半导体订单的影响：冲击不是AI芯片需求消失，而是订单可见度从Neocloud融资驱动采购转向超大规模厂商、ASIC、主权项目和带预付款的硬合同。NVIDIA Data Center在FY2027 Q1仍达752亿美元，TSMC先进节点和HP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI芯片供应链中现金流的高确定性支撑了科技股的信用乐观，而CoWoS与HBM产能瓶颈及地缘政治集中度则引发了期权市场的尾部焦虑。", "summary": "本报告分析了导致跨资产风险溢价分裂的AI半导体供应链逻辑。信用市场的乐观情绪源于AI基础设施建设带来的高确定性现金流（如NVIDIA锁定台积电大部分CoWoS产能）。而期权市场的焦虑则源于结构性的供应链瓶颈（CoWoS和HBM4产能受限）导致的尾部风险，以及先进封装工艺高度集中在台湾地区所带来的地…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在AI链内部按融资质量轮动：超配Tier-1平台与卖铲人(设备/HBM/封装/电力赢家)，低配靠GPU抵押融资循环的Neocloud股权与高收益债。", "summary": "综合前八卡推演，AI基础设施的约束已从硅到电力、再到融资可得性。本卡给出按融资质量过滤的五篮子相对配置框架：B1 Tier-1 AI平台(+300bp)、B2 设备/HBM/先进封装(+250bp)、B3 ASIC定制(+100bp)、B4 电力瓶颈赢家(+150bp)、B5 低质量AI基建(-2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 0, "value": 78.6, "events": [ { "title": "宏观信用收紧并未击穿全球AI周期，而是将AI基础设施层切割为「超大规模云厂＋央企国家队」核心（继续扩张）与「民营商业云」外缘（防御性收缩）的K型结构，TMT投资应聚焦CoWoS/HB…", "summary": "TMT综合：沃什鹰派持稳（3.50–3.75%）经离岸人民币流动性、汇率对冲摩擦与企业中长贷收缩，传导至AI基础设施层产生K型分化而非总量收缩。美超大规模云厂CY2026合计capex运行率3,200–3,600亿美元由经营现金流支撑不受冲击；央企运营商云经PBOC科技创新再贷款（约2.5%）继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "验证AI算力硬件链的估值承压：流动性虹吸下的先进封装与GPU产业链韧性评估", "summary": "本卡对AI算力硬件链做压力测试，得出'分化韧性'结论。资金流层面确认虹吸：2026-06-08半导体−219.33亿、通信−98.35亿、消费电子−69.92亿元净流出居前[S6]，年内93家半导体公司减持[S7]，叠加CXMT约295亿IPO锚定[S9]。但估值下移是资金流/轮动驱动、非盈利驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 5, "neutral": 0, "value": 68, "events": [ { "title": "电力供给限制驱动的半导体能效转型主要表现为通用GPU的系统与精度级优化，而非ASIC/NPU的大规模替代；自研定制芯片的爆发反而加剧了全球先进制程与封装供应链对TSMC的物理集中度。", "summary": "截至2026-06-17，电力瓶颈确在倒逼算力能效转型，但并非单纯的通用GPU向专用ASIC切换。算法多变性（如MoE与动态推理）使自研ASIC面临极高的“代差失效”风险，通用GPU（如NVIDIA Blackwell）凭借软件与架构灵活性仍占据70%以上份额。能效提升主要通过系统精度（FP4/F…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "芯片能效提升不足以抵消机柜功耗密度的物理极限，电力供应仍是制约AI资本开支扩张的物理红线。", "summary": "本报告对2026年中期的半导体能效提升进行了压力测试。虽然NVIDIA Rubin (R100)等新架构在推理能效比上提升了10倍，但单卡功耗攀升至2.3kW，机柜密度突破250kW，加剧了电网侧变压器和输电设施的改造压力。先进封装（CoWoS-L）和HBM4虽缓解了显存瓶颈，但未能扭转算力需求增…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装的去瓶颈把AI算力的约束推向电力与液冷，2026–2027年边际capex美元流向'电子与水'而非硅片，散热与电网设备是该传导链上确定性最高的边际增量。", "summary": "截至2026-06-16，TSMC CoWoS等效晶圆产能从2024年底约35k wpm扩张至2026年底约75–80k wpm，HBM/TSV封装侧瓶颈基本释放。约束顺势外迁至电力与液冷：Blackwell B200 NVL72机架功率达约120 kW、Rubin级150–200 kW/rack…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装/HBM 产业链拥挤度风险压力测试 工作日期：2026 06 17（Asia/Singapore） 分析师：半导体分析师（semiconductors analyst） 立场…", "summary": "先进封装/HBM 产业链拥挤度风险压力测试 工作日期：2026 06 17（Asia/Singapore） 分析师：半导体分析师（semiconductors analyst） 立场：stress test（压力测试 Research note 6 的\"先进封装择时择股\"结论） 板块：A股 CoW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 6, "neutral": 0, "value": 63.4, "events": [ { "title": "AIDD 硬件瓶颈在 HBM 容量/带宽与 CoWoS/SoIC 先进封装，而非算力；HBM4 + CoWoS-L + NVL72 三位一体绑定决定了 2026–2027 工业级生物…", "summary": "AIDD 工作负载（结构预测、全原子 MD、FEP、Cryo-EM）属于访存带宽与片上 SRAM 受限型，而非算力受限。HBM3e 已足以支撑 AlphaFold 3 量级推断，但工业级 10^6–10^7 原子 MD 与 FEP 需要 HBM4（≥1.6 TB/s、36–48GB/stack）+…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 能源重定价压的是超大规模厂商 ROIC 而非芯片 capex 效率——硅片主导 TCO、能源占比单位数，且性能/瓦溢价+杰文斯回弹反而净支撑前沿逻辑/先进封装/HBM 需求…", "summary": "2026-06-18 压力测试：能源是证伪芯片逻辑的错误杠杆。Epoch 1GW 机房年化 TCO 约 $8.5B，服务器占 ~59%、能源仅 ~7%；交付电价 +30% 仅抬升 TCO 约 +2.1%，翻倍也仅 +7%（自测算），而 IT 寿命 5→3 年使 TCO +41%——刷新节奏的影响是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GPU配额与电网互联是2026–2028全球AI布局的联合约束，资本流向同时拥有一档算力与开放电力排队的辖区，最佳投资敞口是电力/电气栈。", "summary": "截至2026-06-18，全球AI基础设施扩张由两个物理瓶颈联合制约：高端GPU配额（管制扼喉点）与可互联级电力容量（新约束）。TSMC CoWoS 2025年底扩至75k片/月仍紧[S3]，Nvidia数据中心年化~US$1,150亿[S4]，BIS 2023-10/2025-01 AI扩散规则…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "T3 制造估值真空；只有当 hyperscaler 与 sovereign-AI 资本开支口径下修时，它才会升级为真正的库存周期冲击。", "summary": "截至 2026-06-17，本卡支持 Research note 1 中 T3 单独触发即可升级压力的规则，但否认 T3 本身已证明半导体链进入无序去杠杆。AI 需求现实仍强，WSTS 预测 2026 年半导体市场达 USD 1.51 trillion；真正风险在于长周期资本开支久期。设计端因 N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 1, "risks": 0, "neutral": 0, "value": 96, "events": [ { "title": "RUM的Together AI合同仅占用其未商业化电力的约3%，而完全填满200 MW所需的47亿至51亿美元资本开支面临巨大融资缺口，且TSMC CoWoS与HBM3E供给紧缺导致…", "summary": "截至2026-06-18，对RUM Group的B300 GPU采购进行供应链与财务敏感性测试显示：其与Together AI签署的2.7亿美元B300液冷云服务协议在物理规模上仅需约1,800至3,000张GPU，对应电力功耗仅约4.8至7.3 MW，仅占其剩余200 MW未商业化容量的2.4%…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 3, "neutral": 0, "value": 84.2, "events": [ { "title": "CPO能维持相对传统光模块的选择性溢价，但不能在供电受限情景下整体对冲电网扩容延迟，因此硬件链仍只应保留小额期权仓位。", "summary": "截至2026-06-19，本卡对CPO估值溢价做压力测试：CPO在链路和交换机层节能真实，NVIDIA称3.5x能效、10x韧性，Broadcom/Meta验证显示1 million个400G等效端口设备小时无link flap且光学功耗-65%。但站点层折算有限，Lambda的4,608 GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与封装成本上升会触发AI基础设施技术路线调整，但截至2026-06-20仍不足以触发超大规模云厂商AI芯片采购总量的全面回撤。", "summary": "工作日2026-06-20：本卡压力测试认为，电力和先进封装成本上升不会在2026年触发超大规模云厂商全面削减AI芯片采购，但会提高单位弹性，推动技术路线从“更多GPU插槽”转向“每瓦性能、液冷机架、自研ASIC、提前锁定CoWoS/HBM和电力站点”。真正风险是并网、变电站、CoWoS或HBM配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler 2026 AI资本开支由CoWoS+HBM+电力三段瓶颈共同定量；卡片08的“中国+一地未验证”使得台/韩集群集中度无法廉价对冲，构成市场定价不足的尾部风险。", "summary": "本卡承接卡片08（先进制程设备与材料的“中国+一地”尚未验证），对全球Hyperscaler 2026–2027 AI算力扩建做压力测试。结论：真正约束并非台积电先进节点晶圆，而是三段串联瓶颈——CoWoS先进封装、SK海力士/美光/三星HBM4、以及数据中心电力与并网。美系五家FY26资本开支合…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 1, "risks": 5, "neutral": 0, "value": 81.6, "events": [ { "title": "AI硬件供应链波动与融资风险传导", "summary": "作为半导体分析师，我对 Research note 03–07 做压力测试并最终支持，但补上一层基本面触发器：光模块/先进封装景气真实但狭窄且反身。2026 全球 1.6T 光模块需求约 500 万只，中际旭创目标 500–600 万只、约 50% 份额、约 75% 硅光 [S1]，但估值高达 7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装5.3倍成本溢价不再是末端事件，而是通过代工ASP、设计公司毛利和超大规模capex三条路径，向全球科技产业链注入一次政策驱动的结构性通胀脉冲，并使估值在赢家与边缘玩家之间显…", "summary": "2026-06-20：本卡支持上一卡结论并向上游延伸。先进封装的5.3倍本土化成本溢价正通过三条路径系统性外溢：(1) 代工ASP——TSMC亚利桑那wafer溢价约30%、CoWoS产能2026翻倍；(2) 设计公司毛利——NVIDIA毛利指引由75%回落至71-73%、AMD MI350毛利低…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力约束不会终结AI算力扩张，但会把半导体设计目标函数改写为每瓦有效算力，并提升Chiplet与先进封装的战略溢价。", "summary": "截至2026-06-21，本卡片支持前序判断：电网接入和可交付兆瓦成为AI扩张区域性天花板后，半导体竞争会从峰值算力转向系统级Perf/Watt。数据中心用电增长使每瓦tokens、HBM带宽、die-to-die互连、热设计和先进封装成为容量权变量。Chiplet、CoWoS/SoIC、HBM4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年先进封装产能与瓶颈分析：CoWoS供需缺口及CoPoS与玻璃基板技术革新展望", "summary": "截至2026年6月21日，先进封装仍是AI算力芯片的主要物理瓶颈。TSMC CoWoS月产能预计在2026年底达到12万至14万片，行业总产能接近20万片，但供需依旧偏紧，交期达52-78周。NVIDIA已锁定了60%-70%的产能。为应对芯片面积扩展极限，TSMC正推进CoPoS（板级封装），预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 0, "value": 93, "events": [ { "title": "截至2026-06-22，低端封装过剩会压低错误产能池的价格，而高端迭代缺失会保留真正重要产能池的配给风险，使Hyperscaler AI资本开支更重、更不稳定、更容易出现已供电机房…", "summary": "截至2026-06-22，本卡压力测试卡片06：先进封装低端过剩并不能缓解Hyperscaler真正需要的高端CoWoS/HBM4认证产能短缺。TrendForce称2026年行业CoWoS-like产能可接近200,000 wpm、缺口或收窄至约10%，但若高端认证占比只有50%-70%，有效高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-21，受2nm GAA先进制程 wafer 溢价与先进封装 80% 初期装配良率的限制，下一代高密度 AI 芯片的制造 BOM 将翻倍至8,153.59美元，其中封…", "summary": "截至2026-06-21，本卡片评估了AI机架密度120kW+下，先进封装良率及散热材料研发对AI芯片成本与供应链的重塑。下一代2nm GAA+HBM4+液冷芯片BOM成本较3nm基准大涨77.2%至6,780.97美元，在初期封装装配良率降至80%的重度压力下更暴增113.1%至8,153.59…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 4, "neutral": 0, "value": 82.4, "events": [ { "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "summary": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-22核心判断：AI算力基建能把“外暖”转化为中国出口盈利，但应买瓶颈环节而非泛AI整机叙事，排序为电力设备和液冷最强、服务器零部件次之、前沿先进封装最需验证。", "summary": "2026-06-22：支持出海制造链受益于AI算力基础设施，但必须结构化看待。全球Hyperscaler资本开支仍高，2026年AI相关投资估算可达7250亿美元；中国5月工业与出口交货值仍强，但美国AI硬件进口正向台湾、墨西哥迁移。中国盈利承接最强在变压器、开关设备、液冷和电力配套，其次是PCB…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产先进封装与HBM进展正在抬高A股AI硬件天花板，但截至2026-06-22还不足以打破国产AI芯片估值上限；应优先精选能变现瓶颈缓解的封装/工具链，而非全面上调芯片篮子。", "summary": "截至2026-06-22，本卡对卡片08的“先进封装与HBM瓶颈封顶国产AI芯片弹性”做压力测试。结论是：国产HBM与先进封装进展真实，CXMT/Huawei/HBM工具链与通富AMD封测链条都在抬高天花板；但它们还未证明HBM3E/HBM4级批量供应、带HBM 2.5D/3D封装良率和商业集群复…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "可信算力区下的A股与STAR硬科技配置映射", "summary": "本卡把卡片01–07的'可信算力区'供应链结论翻译成A股/STAR硬科技排序，并用四道闸门（收入已拐点、订单可见、毛利确认、与可交付获批算力直接关联）区分真实订单受益与主题弹性。结论排序：光模块(中际旭创+192%、新易盛+106%)、AI服务器/ODM(工业富联净利+102.6%、订单至2027…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 7, "neutral": 0, "value": 65.8, "events": [ { "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下修。", "summary": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-23，半导体能效突破是AI capex去杠杆的缓冲器，不是2027年前的解药；它会降低单位算力成本，但先推高先进封装、HBM和设备端capex。", "summary": "截至2026-06-23，本卡对能源成本驱动的AI capex去杠杆做半导体侧压力测试。结论是：Blackwell、HBM3E/HBM4、自研TPU/Trainium、CoWoS/SoIC和硅光能显著降低每token或每带宽能耗，但2027年前不足以抵消云巨头总capex压力。原因是能效路线本身需…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "硬件预算挤压对次世代AI芯片(如Rubin)性能需求与技术路线的传导影响", "summary": "针对“电力成本飙升→能效驱动型替代→削弱顶尖算力需求”的链条做压力测试。结论：前提成立但结论被颠倒。能效比(每瓦token→每兆瓦营收)确已成为芯片主导设计轴[S3][S10]，但在电力硬约束下，优化能效比会把需求集中到前沿节点(Rubin/HBM4/CoWoS-L)而非替代离场——兆瓦稀缺时运营…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与能源瓶颈下的半导体先进封装与HBM产能演进", "summary": "本卡压力测试Card-01/02主线并作两处修正。封装侧：CoWoS正按计划被主动疏通——台积电产能由2023年底约13K片/月扩至2026年底目标约120-130K[S8]，叠加OSAT后行业近200K[S9]，供需缺口由约20%收窄至2026年底约10%[S9]，故CoWoS并非Card-02…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 6, "neutral": 0, "value": 81.2, "events": [ { "title": "电力瓶颈下的半导体能效革命：先进封装与低功耗架构的边际驱动力", "summary": "本卡支持并延伸第04卡。电力稀缺把AI算力的经济单位从“每颗芯片”转为“每瓦token”，正驱动厂商以CoWoS、HBM和系统级先进封装提升单位功耗算力。NVIDIA以Vera Rubin机柜级相对Grace Blackwell的10倍单位功耗推理吞吐、MoE 8倍能效、六代百万倍每兆瓦吞吐为卖点…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力芯片设计从“追求极致算力”向“追求能效比”转型的技术路径与产业链机遇", "summary": "本卡支持前四卡“材料/电网/电力墙造成算力供给缺口”的主线，并从半导体供给侧给出答案：芯片设计已从追求极致算力转向追求能效比，先进封装(CoWoS产能约3.5万→13万片/月)、Chiplet/UCIe(0.25–0.5 pJ/bit对比片外DRAM 6–8 pJ/bit)、HBM4(带宽+75%…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "先进封装与液冷以2026-2027节奏真实落地、兑现约20-50×能效与约5×功率密度，但只松绑芯片/机柜边界、并因Jevons把节省再投资为更多算力，因此它们是前序'物理基础设施约…", "summary": "截至2026-06-24，本卡支持Research note 04-06的判断，并从硅侧回答任务问题：先进封装与液冷的落地节奏真实且加速，但只松绑芯片/机柜边界，无法废除电网与取水之墙。台积电CoWoS产能三年约10倍（13k→130k片/月，缺口20%→10%）；SK hynix/三星HBM4E…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国CSP成本上升背景下，AI算力基础设施供应弹性与扩容压力分析", "summary": "压力测试部分否定线程以电力为中心的框架。AI算力系统是供给绑定而非需求绑定：2026年超大规模厂商资本开支同比+77%至约7250亿美元[S1]，英伟达Blackwell售罄至年中、积压360万颗[S2]，因此约束的是配额优先级而非需求。核心锚定集群凭借锁定电力、GPU配额与表后自发电被隔离；放缓…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 2, "neutral": 0, "value": 96, "events": [ { "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力稀缺不会杀死AI芯片周期，但会把竞争重心从单纯峰值算力转向固定兆瓦内的有效吞吐和能效，从而收窄可投资赢家。", "summary": "截至2026-06-25，本卡对电力和并网约束做半导体侧压力测试：约束不会终结AI芯片需求，但会把胜负指标从峰值FLOPS推向每兆瓦吞吐、每机柜token、HBM每瓦带宽和部署可用性。NVIDIA的机柜级平台、Google TPU、AWS Trainium、TSMC背面供电和Micron HBM4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 2, "neutral": 0, "value": 96, "events": [ { "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-27，AI电力并网限制（2年延误拖累53.71亿美元[S12]）将倒逼算力采购削减并重创高端半导体高增长假设，投资者应果断削减AI硬件至12%及长债至5%，转而增配…", "summary": "截至2026-06-27，本报告重新校准了在高利率平台与电力之墙挤压下的多资产配置。2年电网并网延误对10万卡Blackwell智算集群将产生53.71亿美元搁浅资金损失[S12]，迫使云巨头在2027年削减20%GPU订单。这将导致台积电3nm和2nm产能利用率分别跌至85%与80%，CoWoS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 4, "neutral": 0, "value": 96, "events": [ { "title": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "summary": "电力与电网容量成为稀缺投入后，AI 芯片优化目标从绝对算力转向每瓦性能（perf/W）与单机柜 token 经济性。NVIDIA 宣称 Blackwell 推理能耗较 Hopper 最高低 25×（FP4/NVFP4 驱动），但 B200 单卡功耗仍升至约 1,200 W（H200 为 700 W…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "summary": "截至2026-06-28，本卡否定“AI资本开支韧性等于A股防御性配置锚”的宽泛结论。Microsoft、Alphabet、Meta、NVIDIA、TSMC等数据仍支持AI基础设施真实需求，但FOMC 3.50%-3.75%、5月PCE同比4.1%、10年美债4.40%和A股两融突破RMB 3tn…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 2, "neutral": 0, "value": 96, "events": [ { "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,80…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析", "summary": "截至2026-06-30，本卡片压力测试Research note 08的能效转型论。结论：CoWoS原始产能已非约束瓶颈——TrendForce判断供需缺口随产能扩张从约20%收窄至2026年底约10%[S1]，台积电产能翻四倍至约13万片/月并外包约24-27万片/年给Amkor/SPIL[S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-05c3f6f33c/", "networkHref": "/investor/research-network/#thesis-05c3f6f33c" }, { "id": "thesis-72097390dd", "source": "auto_topic", "title": "在66.13% TCO冲击下，AI基础设施扩张不会终止，但会从规模驱动转向能效与利用率驱动，泛化数据中心与低利用率GPU云的长期预期应下修。", "titleEn": "Practical research topic: 在66.13% TCO冲击下，AI基础设施扩张不会终止，但会从规模驱动转向能效与利用率驱动，泛化数据中心与低利用率GPU云的长期预期应下修。", "lane": "AI 基础设施", "laneEn": "AI 基础设施", "laneHref": "/investor/lanes/ai/", "investableFocus": "", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" } ], "stockTickers": [ "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ", "APO", "IEF", "TLT", "CAT", "600900.SH" ], "coreView": "该主线来自 4 条研究记录的标题、问题和摘要聚类，时间跨度 2026-06-14 至 2026-06-23；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-04-27", "lastSeen": "2026-06-29", "supports": 13, "risks": 138, "neutral": 9, "analystCount": 13, "sourceTypeCounts": { "研究线程": 5, "研究笔记": 143, "证据更新": 10, "研究主题": 2 }, "topTerms": [ "Hyperscaler", "Microsoft", "AI基础设施", "Amazon", "Meta", "GPU", "数据中心", "在66.13%", "冲击下", "基础设施扩张不会终止", "期应下修", "电力" ], "whatChanged": [ "2026-06-29: AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "2026-06-28: AI数据中心‘电力荒’：电网接入瓶颈下的超大规模云厂商资本开支与自备电站转型", "2026-06-28: AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "2026-06-28: 电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "2026-06-28: AI交易不会结束，但估值核心应从买到多少芯片转向能否按期通电、保持利用率并降低每token能耗。" ], "practical": { "sectorLane": "AI 基础设施", "sectorLaneEn": "AI 基础设施", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" } ], "stockTickers": [ "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ", "APO", "IEF", "TLT", "CAT", "600900.SH" ], "primaryStocks": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" } ], "instrumentTags": [ "NVDA", "NVIDIA", "TSM", "TSMC", "AVGO", "Broadcom", "MU", "Micron", "ANET", "Arista Networks", "VRT", "Vertiv" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 112 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 13, "recentEvents": 112, "supportRatio": 0.08, "riskRatio": 0.86, "conflictScore": 95.8, "evidenceDepth": 100 }, "score": 67.8 }, "events": [ { "date": "2026-06-29", "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power预期缺口，会迫使Hyperscaler预付并网升级、锁定变压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "Hyperscaler", "AI基础设施", "GPU", "电力", "CapEx" ], "href": "" }, { "date": "2026-06-28", "title": "AI交易不会结束，但估值核心应从买到多少芯片转向能否按期通电、保持利用率并降低每token能耗。", "summary": "截至2026-06-28，本卡综合支持前序判断：2026-2030年AI扩张的约束从服务器交付转向可通电MW交付。Amazon、Microsoft、Alphabet、Meta最新季度CAPEX/PPE购买合计约$130.8B[自测算]，但IEA预计美国数据中心用电到2030年较2024年增加约240TWh，约等于27.4GW连续负荷[自测算]。结论是Hyp…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 41.25, "matchedTerms": [ "Hyperscaler", "Microsoft", "Amazon", "Meta", "数据中心", "电力", "CapEx", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "summary": "截至 2026-06-28，本卡压力测试认为：电力交付瓶颈会迫使 hyperscaler 从单纯追求新增 MW 转向 MW 生产率和存量优化，但不会终结扩产。Microsoft、Alphabet、Meta 可见 2026 年 capex 区间合计约 4,950-5,250 亿美元（自测算），同时仍披露容量约束；IEA 显示数据中心 1-3 年可建成，而新电…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.75, "matchedTerms": [ "Hyperscaler", "Microsoft", "Meta", "数据中心", "电力", "CapEx", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计约6900-7200亿美元（自测算），NVIDIA FY20…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 34, "matchedTerms": [ "Microsoft", "AI基础设施", "Amazon", "Meta", "电力", "CapEx" ], "href": "" }, { "date": "2026-06-28", "title": "AI数据中心‘电力荒’：电网接入瓶颈下的超大规模云厂商资本开支与自备电站转型", "summary": "截至2026年6月29日，超大规模云厂商集体资本开支预计在2026年达到7250亿美元，其中约75%用于AI基础设施。然而，AI规模扩张的核心制约已由GPU芯片转为电网并网容量。美国PJM电网并网排队超5年，Dominion电网仍有45 GW容量待分配。这正倒逼微软、亚马逊等巨头直接投资天然气微电网及核电等表后（BTM）自备电力，从而将科技巨头转变为重资产…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28, "matchedTerms": [ "AI基础设施", "GPU", "数据中心", "电力", "云厂商", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "变压器材料和交付瓶颈已开始改变 Hyperscaler 的2026-2027资本开支与电力采购节奏，影响主要体现为预算通胀、长期锁电和通电交付递延，而非AI需求取消。", "summary": "截至2026-06-28，结论为支撑：Hyperscalers 已经通过预算上修、长期电力合同和项目节奏调整来应对变压器及电气设备瓶颈，但公开披露通常不会把变化直接归因为 GOES 或铜关税。Microsoft 2026 capex 中约250亿美元来自更高组件价格，Meta 上调2026 capex中点100亿美元，Amazon 把电力容量作为部署KPI…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.25, "matchedTerms": [ "Hyperscaler", "Microsoft", "Amazon", "Meta", "电力", "CapEx", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期：…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 / Executive Summary 截至 2026 年 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40, "matchedTerms": [ "Hyperscaler", "Microsoft", "Amazon", "Meta", "数据中心", "电力", "云厂商" ], "href": "" }, { "date": "2026-06-27", "title": "高利率平台期对AI基础设施资本开支的影响", "summary": "压力测试卡片04的3.75%-4.00%高位平台前提：利率平台并非2026年头部Hyperscaler资本开支（约7250亿美元、+77%、约75%为AI）的约束性瓶颈，因四巨头仍主要自筹、AA/A级利差仅约80bp。但融资模式恰在约2026Q3翻转为债务驱动（整体资本开支超过经营现金流；Oracle FCF已为−237亿、约720亿合作方债务），使整个体…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34, "matchedTerms": [ "Hyperscaler", "Microsoft", "AI基础设施", "电力", "云厂商" ], "href": "" }, { "date": "2026-06-27", "title": "电子链条利润反弹主要受AI基础设施投资拉动，具备向存储、设备、材料、先进封装、光互联和数据中心电源散热扩散的潜力，但尚不足以证明消费电子和内需全面复苏。", "summary": "截至2026-06-28，本卡支持上一卡判断：电子链条利润修复主要由AI基础设施、算力芯片和存储涨价驱动，而非内需广谱复苏。国家统计局数据显示，1—5月电子行业利润同比增长103.9%，贡献全部工业利润增量43.1%。NVIDIA、Microsoft、Alphabet、Meta、Alibaba和Tencent的披露共同验证数据中心硬件需求仍强；TrendF…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32.25, "matchedTerms": [ "Microsoft", "AI基础设施", "Meta", "数据中心", "设备" ], "href": "" }, { "date": "2026-06-26", "title": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "summary": "截至2026-06-27，本卡片对AI基础设施链做压力测试：全球AI资本开支仍大，Goldman Sachs基准模型给出2026年USD 765 billion、2031年USD 1.6 trillion的年度AI capex，但Amazon、Alphabet、Meta、Microsoft的数据表明现金流、折旧、能源和电网接入正成为估值核心约束。A股不应全…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30, "matchedTerms": [ "Microsoft", "AI基础设施", "Amazon", "Meta", "电力", "CapEx" ], "href": "" }, { "date": "2026-06-25", "title": "CNH套息去杠杆会先打掉A股AI基础设施股票的流动性溢价和估值久期，而不是立即打断全球AI CapEx实物周期。", "summary": "截至2026-06-25，本卡片压力测试CNH套息去杠杆对AI基础设施估值和CapEx预期的冲击。结论是：流动性溢价消失可使A股AI硬件/算力基础设施估值出现非线性回撤，但不应直接推导为全球hyperscaler CapEx崩塌。Microsoft、Alphabet、Meta、Amazon、Oracle和NVIDIA披露的2026年AI基础设施投入、订单积…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.75, "matchedTerms": [ "Hyperscaler", "Microsoft", "AI基础设施", "Amazon", "Meta", "数据中心", "CapEx" ], "href": "" }, { "date": "2026-06-25", "title": "BTM是AI园区局部speed-to-power的有效工具，但在2026-2028年不足以填补全国AI电力缺口，反而会放大公用事业负荷预测兑现率与成本分摊的不确定性。", "summary": "截至2026-06-25，本卡从AI基础设施侧压力测试card-01：Hyperscaler BTM已从叙事进入执行，Microsoft/Chevron 2.67 GW燃气BTM、Crusoe/Microsoft 900 MW现场电厂、Talen/AWS 300 MW直连核电和xAI 247.2 MW获准燃气轮机合计约4.117 GW。但按90%容量因子仅…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34, "matchedTerms": [ "Hyperscaler", "Microsoft", "AI基础设施", "数据中心", "电力" ], "href": "" }, { "date": "2026-06-25", "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实触发重排，但方向是向支出倾斜，且买方自身点名的硬约束是电力与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 34, "matchedTerms": [ "Hyperscaler", "Microsoft", "Amazon", "Meta", "电力" ], "href": "" }, { "date": "2026-06-25", "title": "AI基础设施的主要瓶颈正从GPU可得性转向电力可得性，真正受益的是能缩短通电时间、提高负荷密度并保障可靠供电的设备与电网环节。", "summary": "截至2026-06-26，本卡支持上一张“算力切向电力”的轮动逻辑，但把核心约束细化为并网、变压器/开关设备、机柜功率密度和可靠供电。全球数据中心用电预计由2024年415 TWh升至2030年945 TWh；中国2030年约279 TWh，全国占比不高但局部冲击显著。1 GW AI IT负荷在PUE 1.2下年耗电约10.5 TWh，足以改变省级电力平衡…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31, "matchedTerms": [ "AI基础设施", "GPU", "数据中心", "电力", "设备" ], "href": "" }, { "date": "2026-06-25", "title": "电力缺口正在把AI基础设施从GPU采购竞赛重塑为通电MW竞赛，Hyperscaler的最优策略是保护已通电集群利用率并把CAPEX前移到电力可选性。", "summary": "截至2026-06-25，本卡支持前序结论：AI电力缺口对Hyperscaler的核心冲击不是单纯电费上升，而是通电节奏决定GPU集群能否形成收入。PJM容量价格从2024/2025的$28.92/MW-day升至2026/2027的$329.17/MW-day，相当于1 GW恒定负荷每年多约$109.6 million；但100 MW服务器模块若因缺电闲…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 29, "matchedTerms": [ "Hyperscaler", "AI基础设施", "GPU", "电力", "CapEx" ], "href": "" }, { "date": "2026-06-25", "title": "电网扩容融资风险对超大规模云服务商数据中心选址与AI集群部署节奏的影响研究", "summary": "本卡综合 Research note 01–04，落到 Hyperscaler 一侧的应对函数。净判断：超大规模云服务商正把电力采购问题转化为自备供电与选址套利问题，对冲真实但仅部分有效。三大杠杆：(1) 选址迁移——从饱和的北弗吉尼亚/PJM 转向 ERCOT、路易斯安那等天然气富集与二线市场，靠近退役煤电/核电并网点可缩短 18–36 个月；(2) 自…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28.25, "matchedTerms": [ "Hyperscaler", "Meta", "数据中心", "电力", "设备" ], "href": "" } ], "timeline": [ { "date": "2026-04-27", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "2026年四大超大规模云厂商AI资本开支约7,000亿美元的可持续性,以及电力瓶颈对Capex落地节奏的真实约束", "summary": "2026年四大超大规模云厂商AI资本开支约7,000亿美元的可持续性,以及电力瓶颈对Capex落地节奏的真实约束 在自由现金流被显著挤压、电网与变压器排队进入'以年计'的当下,本周(4月24-30日)Microsoft/Meta/Alphabet/Amazon的FY26一季报会兑现还是修正这条7,…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-29", "supports": 0, "risks": 0, "neutral": 1, "value": 46.4, "events": [ { "title": "2026年一季度超大规模云厂商AI资本开支轨迹与GPU/ASIC供应链含义", "summary": "2026年一季度超大规模云厂商AI资本开支轨迹与GPU/ASIC供应链含义 在Alphabet、Microsoft、Meta、Amazon集中披露2026年一季报的窗口,云厂商AI基础设施资本开支是否仍维持50%+同比增速、2026年全年指引是上调还是下调,以及边际美元在GPU、定制ASIC和电力…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "Q1 2026 财报后的云巨头 AI capex：需求验证还是瓶颈预警？", "summary": "本卡选择 tmt-analyst 评估 Q1 2026 云巨头 AI capex 周期。Microsoft、Alphabet、Amazon 和 Meta 均显示 AI/云收入加速，但 capex 和 PP&E 支出同步大幅上行。TSMC 的 61% HPC 收入占比、74% 先进制程占比和 66.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-03", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "2026年4月 hyperscaler AI capex 重估", "summary": "本卡由 tmt-analyst 作为开题分析师完成。结论是：2026年 hyperscaler AI capex 仍是需求驱动周期，而非简单泡沫破裂；Microsoft、Alphabet、Amazon 和 Meta 的 2026 capex 大致落在 $695-725 billion 区间，且由 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年AI基础设施资本开支超级周期：需求验证还是泡沫拐点", "summary": "2026年全球超大云厂商AI基础设施Capex预计达7000–7250亿美元，其中75%直接面向AI。需求端信号真实：Azure AI营收同比增62%，Amazon Bedrock API调用量一季度是2025全年3倍，但Amazon FCF预计转负，Alphabet FCF或暴跌90%，Cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 1, "risks": 6, "neutral": 0, "value": 19.6, "events": [ { "title": "超大规模资本开支天花板不是硬美元上限，而是融资与能源上限；中东资本抬高天花板，但也把AI基础设施推入出口管制和联盟管理的核心。", "summary": "本卡片综合而非否定卡片01：AI从暴力训练转向架构效率与推理，并未终结资本开支周期，而是把约束从GPU购买力推向现金流、项目融资、电力、液冷和出口管制。2026年主要平台AI基础设施资本开支约处于$700 billion量级，Amazon、Alphabet、Microsoft、Meta仍能投入，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国四大超大规模云厂商（微软/谷歌/Meta/亚马逊）2024年合计资本开支约2300-2600亿美元，2025年约3400-4000亿美元，2026年指引再次上调，高端区间同比中两…", "summary": "美国四大超大规模云厂商（微软/谷歌/Meta/亚马逊）2024年合计资本开支约2300-2600亿美元，2025年约3400-4000亿美元，2026年指引再次上调，高端区间同比中两位数至约20%增长。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SMR 是必要但不充分,2025–2028 的边际电力来自燃气、LWR 重启与现役核电功率提升,真正的瓶颈是 NRC 审批通量、HALEU 与 FERC 监管,而不是反应堆设计。", "summary": "宏观开场:科技巨头的核电 PPA(Microsoft–TMI、Amazon–Talen、Google–Kairos、Amazon–X-Energy、Meta RFP)是写在 2028–2035 供给曲线上的远期期权,而非已落地的解决方案。需求端,IEA/EPRI 给出 2030 年全球数据中心用电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "hyperscaler AI 负荷以 firm 为主，non-firm 灵活性窗口只有 15-25%，PJM 灵活电价带来的节省被高估约 4-6 倍，真正的投资倾斜应回到 firm …", "summary": "对 Card-01 援引的 Johns Hopkins $15-16 B 节省进行压力测试。拆解 hyperscaler AI 负荷构成：在线 inference 占 45-55% 且持续上升、基础模型训练 25-35%、fine-tuning/RLHF/eval 5-10%、内部批处理 10-1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 1, "neutral": 2, "value": 49, "events": [ { "title": "2026年四大超大规模厂商（微软、亚马逊、谷歌、Meta）预计在AI基础设施上投入7000亿美元", "summary": "2026年四大超大规模厂商（微软、亚马逊、谷歌、Meta）预计在AI基础设施上投入7000亿美元", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2025年四大超大规模云厂商（MSFT、META、GOOGL、AMZN）合计capex约3000亿美元，2026财年总盘子约3600-3900亿美元，同比增长20%-30%", "summary": "2025年四大超大规模云厂商（MSFT、META、GOOGL、AMZN）合计capex约3000亿美元，2026财年总盘子约3600-3900亿美元，同比增长20%-30%", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年四大hyperscaler（微软、Alphabet、亚马逊、Meta）AI capex合计约6950亿至7250亿美元", "summary": "2026年四大hyperscaler（微软、Alphabet、亚马逊、Meta）AI capex合计约6950亿至7250亿美元", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-06", "supports": 1, "risks": 0, "neutral": 1, "value": 61.4, "events": [ { "title": "2026年美股四大超大规模厂商(MSFT、META、GOOGL、AMZN)合计AI资本开支预计为3800-4300亿美元，同比增长约25-30%", "summary": "2026年美股四大超大规模厂商(MSFT、META、GOOGL、AMZN)合计AI资本开支预计为3800-4300亿美元，同比增长约25-30%", "sourceLabel": "证据更新", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "META是四大美国云厂中2026年资本开支/销售额比率最高的（约38%）", "summary": "META是四大美国云厂中2026年资本开支/销售额比率最高的（约38%）", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-09", "supports": 0, "risks": 1, "neutral": 1, "value": 48.8, "events": [ { "title": "四大超大规模云厂商合计 FCF/OCF 连续两个季度低于 10%。", "summary": "四大超大规模云厂商合计 FCF/OCF 连续两个季度低于 10%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "四家厂商2026年Q1业绩复核指引日期：MSFT（2026-04-29）、META（2026-04-30）、GOOGL（2026-04-24）、AMZN（2026-05-01）", "summary": "四家厂商2026年Q1业绩复核指引日期：MSFT（2026-04-29）、META（2026-04-30）、GOOGL（2026-04-24）、AMZN（2026-05-01）", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-13", "supports": 1, "risks": 1, "neutral": 0, "value": 55.2, "events": [ { "title": "AI CapEx的主线不是收缩，而是从新增外壳数量转向每瓦、每机架、每封装的半导体密度提升。", "summary": "截至2026-05-13，本卡支持卡片01的“密度胜过足迹”判断：数据中心外壳新开工-19% QoQ并不等于AI CapEx下行，而是资本更多流向GPU/ASIC、HBM、CoWoS/先进封装、机架级互连、光DSP和液冷配套。Microsoft、Meta、Alphabet、Amazon的最新披露均…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卫星数据中心新开工外壳 QoQ -19%，与 Hyperscaler CapEx 上修周期方向背离", "summary": "卫星数据中心新开工外壳 QoQ -19%，与 Hyperscaler CapEx 上修周期方向背离 如果三大集群外壳新开工的卫星读数在 2026Q2 折年 -19% 是真的，意味着 2026H2-2027H1 的 GPU/电力/光模块入库节奏可能弱于一致预期 — 团队是先做空 IPP（CEG / …", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 3, "neutral": 0, "value": 36.6, "events": [ { "title": "能源获取难度要求对AI基础设施进行选择性下修：终局需求不变，但IDC和算力租赁的CAPEX转化曲线、利用率和ROIC应显著打折。", "summary": "截至2026-05-15，本卡对电力瓶颈对AI基础设施的影响进行TMT压力测试。结论是：不应全面下修AI算力长期需求，但必须下修CAPEX转化节奏、利用率爬坡和ROIC。IEA预计数据中心用电从2024年415 TWh升至2030年约945 TWh，同时约20%规划项目存在电网延期风险；Micro…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年五大超大规模云厂商（亚马逊、微软、Alphabet、Meta、甲骨文）合计AI相关资本支出约7250亿美元，高于2025年的6600-6900亿美元区间", "summary": "2026年五大超大规模云厂商（亚马逊、微软、Alphabet、Meta、甲骨文）合计AI相关资本支出约7250亿美元，高于2025年的6600-6900亿美元区间", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "四家美股超大规模厂商（MSFT、GOOGL、META、AMZN）2026年合计资本开支指引约$355–410 bn，中值约$380 bn", "summary": "四家美股超大规模厂商（MSFT、GOOGL、META、AMZN）2026年合计资本开支指引约$355–410 bn，中值约$380 bn", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "AI 生产率主线挺过 8%–15% 宏观回撤——溢价倍数被压缩，但盈利曲线不发生趋势性恶化；硬件核心仓继续持有，二阶新型云与软件溢价名应削减。", "summary": "在 global-macro 卡设定的 8%–15% 情绪回撤基线下，AI 硬件与算力 capex 方向上完好但盈利曲线边际走平、链条分化加剧。Hyperscaler FY26 capex（约 5,100–5,400 亿美元）已通过合同与电力 PPA 锁定，约束在电力交付而非资产负债表，对回撤敏感…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "截至2026-05-19，AI半导体供应链尚未出现云CapEx或HBM/CoWoS订单环比下修，算力基础设施交易仍应保留核心仓位但需严密跟踪ROIC与Q2/Q3指引风险。", "summary": "截至2026-05-19，首席策略师提出的敏感触发点有效，但尚未触发。Microsoft、Alphabet、Amazon、Meta的2026年AI基础设施CapEx仍在上修或维持高位，TrendForce估计前八大CSP 2026年CapEx超过7100亿美元、同比约+61%。HBM方面，Micr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 2, "neutral": 0, "value": 46.8, "events": [ { "title": "关键电力设备会让 announced AI capacity 打折，但除非 hyperscaler capex 与电气基础设施供应商订单同时掉头，否则 H2 2026 不应按整体算力…", "summary": "截至 2026-05-19，本卡对 Research note 07 的 BTM 受阻结论做压力测试：变压器、GSU、开关柜和并网队列已成为 AI 算力扩产的真实交付瓶颈，120-143 周交期足以让部分 announced MW 延迟。但基准判断不是 hyperscaler capex 停止，而…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至 2026-05-19，AI 基础设施仍是 TMT 领涨主线，但可投资范围应收窄到有真实 usage、backlog、产能瓶颈和现金流承受力的环节。", "summary": "工作日期 2026-05-19。本文否定前序空白 TMT 论点并选择 (A)：AI 基础设施仍是领涨主线。证据是 hyperscaler capex/收入比虽已升至约 30% 的四家公司 Q1 合计强度，且 Alphabet、Amazon 的 FCF 张力明显，但 Microsoft AI ARR…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 2, "neutral": 0, "value": 48.2, "events": [ { "title": "AI基础设施的部署上限由已通电液冷MW决定，而不是由GPU订单或云厂商capex预算单独决定。", "summary": "截至2026-05-24，本卡支持前序“AI瓶颈物理化”判断，但把瓶颈收窄为三条工业交付链：液冷机架、电气设备和可靠电力。GB300 NVL72机架功率已达132-140 kW，100 MW IT功率在设施损耗前仅支持约714-758个机架；JLL披露2025年57%数据中心项目因供应链延误至少3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的近期上限将由可靠通电MW决定；电力在拥挤区域会先于GPU成为硬约束，但全球层面仍是区域性、可部分迁移的约束。", "summary": "截至2026-05-24，本卡压力测试结论是：AI数据中心电力与电网设备约束在核心拥挤区域已经先于GPU形成硬约束，但不应上升为全球绝对缺电叙事。GB300 NVL72单机架功率达132-140 kW，100 MW IT功率仅支撑约714-758个机架；同时IEA、DOE/LBNL、NERC和NI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 1, "value": 52.6, "events": [ { "title": "2027 年 hyperscaler capex 的核心风险是高基数后的增速正常化与支出结构迁移，而不是已被证实的 AI 需求饱和。", "summary": "截至2026-05-26，公开证据不支持“2027 年 hyperscaler capex 断崖”这一硬前提。AWS/Azure 尚无正式 2027 指引，但 Alphabet 明确预计 2027 CapEx 较 2026 显著增加，Microsoft 仍称至少到 2026 年产能受限，Amazo…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超大规模 AI capex 真的减速了吗：卫星+电网负荷 vs. 4 月业绩会 capex 走平语调的 reconciliation", "summary": "超大规模 AI capex 真的减速了吗：卫星+电网负荷 vs. 4 月业绩会 capex 走平语调的 reconciliation Is hyperscaler AI capex really decelerating? Reconciling satellite + grid-load evi…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 2, "neutral": 0, "value": 51, "events": [ { "title": "截至 2026-05-26，电力设备瓶颈并未均等阻碍算力落地，而是建立了极度分化的双轨制市场，通过35倍的高额投产延迟杠杆，重奖前瞻储备与套利通道，严惩后发缺乏设备保障的云厂商。", "summary": "截至 2026-05-26，压力测试显示电力设备交付延迟对AI基础设施扩产节奏呈现高度分化的特征。大型变压器交付期拉长至3-4年，但第一梯队超大规模云巨头（MSFT/GOOGL/AMZN）已通过数年前的订单锁定了产能（如Eaton的12年储备订单和GEV单季电气化大单）；Meta等后发者则面临严重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力需求并不天然利好公用事业；只有把数据中心负荷转化为有抵押、take-or-pay、且支付输电公平份额的合约收入，公用事业才能避免把AI资本开支变成居民电费和BBB信用风险。", "summary": "截至2026-05-27，本卡支持上一张信用卡片对BBB电网公用事业债务的谨慎判断，但进一步区分风险：真正危险的是拥有大量AI数据中心接入需求、却没有最低账单、抵押品、退出费和表前输电收费保护的公用事业。AEP Ohio式大负荷费率、Georgia PSC合同审查、Microsoft/Meta/G…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 2, "neutral": 0, "value": 52.4, "events": [ { "title": "2026年AI芯片与先进封装产能约束会推高并拉长云端CapEx，而不是使其取消；配置应偏向具备配额权的上游瓶颈环节。", "summary": "截至2026-05-28，本卡支持前序判断：AI硬件CapEx传导链的约束真实但不等于削减。Microsoft、Amazon、Alphabet、Meta的2026年CapEx指引中枢约为$710 billion，反映的是芯片、HBM、先进封装、电力和数据中心资源稀缺导致的锁单与涨价，而非需求走弱。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力硬约束并未削减AI总投入，而是迫使数据中心从横向扩张转向纵向密度跃升，加速定制ASIC与高密度液冷的爆发，进一步逼死二线算力租赁商。", "summary": "基于55-65GW的物理上限，AI芯片数量的“指数增长”将被迫转向“受限增长”，但Hyperscaler在2026年6300-7500亿美元的资本开支总额并不会放缓。大厂正在将资本推向追求极致算力密度（120kW+液冷与高端GPU）与定制ASIC（+44% YoY），同时通过淘汰传统x86 CPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 1, "risks": 2, "neutral": 0, "value": 61.8, "events": [ { "title": "硬件成本红利与高CapEx会通过更低cost-per-token、更稳定算力供给和更可预测API价格改写AI应用单位经济模型，但真正的投资筛选变量是单任务成本改善乘以分发杠杆。", "summary": "截至2026-05-29，本卡支持前序判断：制造端成本红利会向下游传导，但更关键的不是单台服务器降价，而是Blackwell级架构、Dynamo类调度、自研芯片、云厂商高CapEx和API定价机制共同降低每个有效token和每个任务的成本。Microsoft、Alphabet、Amazon、Met…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI算力链条的资本开支与数据中心建设仍具极强韧性，铜、铀及电力等供给侧成本上涨占数据中心总TCO比重微乎其微，核心限制是物理交付周期而非财务成本，战略卡位迫使算力投资继续呈刚性上行。", "summary": "AI算力资本开支和数据中心建设对铜、铀及电力供给侧成本上涨展现出极强的价格不弹性。在AI算力中心TCO中，GPU等IT硬件成本占~60%，而电力仅占~10%，铜等原材料成本占比更低于0.5%。即使铜价翻倍、容量电费飙升，对总成本影响仅在2%以内。真正的核心瓶颈是特大变压器（交期36-48个月）和并…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "我已完成对云厂商在能源通胀与设备成本上升下的资本开支效率及算力定价权的压力测试分析，并成功在工作空间中生成了完整的英文版报告、中文版报告以及机器可读的元数据手递手（handoff）文…", "summary": "我已完成对云厂商在能源通胀与设备成本上升下的资本开支效率及算力定价权的压力测试分析，并成功在工作空间中生成了完整的英文版报告、中文版报告以及机器可读的元数据手递手（handoff）文件。 核心结论摘�� (Core Synthesis) 本卡片站在 TMT行业分析师 的立场，对全球超大规模云计算三…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 3, "neutral": 0, "value": 49.2, "events": [ { "title": "2026是AI产能的配额年而非顶点年——capex绝对值再创新高但增速触顶、瓶颈已从芯片迁移到变压器与液冷，旗舰算力加速通电与存储景气是同一物理约束的两个面，而非彼此矛盾。", "summary": "四个看似冲突的信号——AI基础设施股跑输、35GW管线约15%受电力/液冷物料延迟、Abilene 1.2GW旗舰加速完工但600MW扩建被取消、SNDK数据中心营收同比+645%——可以放进同一个2026框架：今年是配额年而非顶点年。大型云厂2026年capex合计7,250亿美元、同比+77%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "训练capex不是需求坍塌，但宽基算力alpha正在成熟化；2026下半年应保留瓶颈算力、提高应用层利润验证权重。", "summary": "截至2026-05-30，本卡综合判断：训练capex的边际ROI尚未确认跌破拐点，NVIDIA、Microsoft和Amazon仍显示算力供给约束；但云厂商capex规模、短寿命GPU/CPU占比和自由现金流压力使宽基基建beta的估值门槛上升。2026下半年A股AI alpha更可能从宽基in…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股AI资本开支与光通信战略展望 (2026年5月) 作者： 首席策略师 报告日期： 2026 05 30 白研究线程标识：已归档会话 ID： 69628375 103f 4490 …", "summary": "A股AI资本开支与光通信战略展望 (2026年5月) 作者： 首席策略师 报告日期： 2026 05 30 白研究线程标识：已归档会话 ID： 69628375 103f 4490 b22e 7faa4bad1a21 1. 核心观点与摘要 截至2026年5月30日，全球AI基础设施建设周期正在从单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 1, "risks": 1, "neutral": 0, "value": 70.6, "events": [ { "title": "半导体管制是地域与结构冲击，不是截至2026-06-01已经触发的全球 AI capex 断崖。", "summary": "截至2026-06-01，本卡压力测试半导体出口管制是否会触发“AI组合端”崩溃。结论是否定：出口管制会伤害中国相关 SKU、库存和产品结构，但 NVIDIA FY2027 Q2 指引在假设中国 Data Center compute 为零时仍达 $91.0 billion +/- 2%，Micr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力的'电力税'利润池正从被监管电网转移至云厂商自筹融资的分布式平行系统——近端属于燃气轮机/发动机 OEM 与天然气中游，远端属于 SMR 厂商与 HALEU 燃料循环。", "summary": "卡片 04（能源行业分析师，立场=支持）延伸卡片 03 的结论：PJM 5+ 年并网排队与 3 年 GPU 寿命的错配迫使 AI 园区离网。2026–2029 年由表后航改型燃气轮机与往复式机组（CAT、CMI、Wärtsilä、Bloom）主导桥梁方案，已有 XOM 二叠纪 1.5 GW、CVX…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 0, "value": 58, "events": [ { "title": "2026-06-02的核心判断是：AI资本开支仍能验证硬件龙头盈利，但估值剪刀差将通过支出方分化和相对降估值来修复。", "summary": "截至2026-06-02，本卡综合而非否定上一卡：AI硬件拥挤风险存在，但Hyperscalers的2026年资本开支区间仍约为6950亿-7250亿美元，足以支撑NVIDIA、Broadcom、TSMC等硬件龙头近期收入指引。分化在支出方：Microsoft与Alphabet的云收入验证更强，A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网变压器长达150周的交期将导致大量AI芯片成为“搁浅算力”，迫使云厂商资本支出向电力倾斜，NVIDIA等硬件龙头的2027年盈利预期面临下修风险。", "summary": "截至2026年6月2日，本卡对AI基础设施部署进行压力测试。NVIDIA GB200机架和10万卡集群面临150周变压器交期的硬约束。这意味着2026年中启动的数据中心到2029-2030年才能并网。从GPU交付到通电的周期将从6-9个月拉长至18-24个月，云厂商将被迫将资本支出向电力设备倾斜。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "AI基础设施利润扩张可以对冲成本压力，但只在加速器、先进代工、存储和电力/热管理等稀缺环节成立，不能外推到整个电子产业链。", "summary": "2026-06-04，本卡对AI硬件与电子产业链做成本压力测试：支持从公用事业红利向盈利成长轮动，但只支持精选AI硬件杠铃，不支持宽基电子超配。NVIDIA、TSMC、Micron等上游利润池具备高毛利与供给稀缺，能够吸收部分原材料、电力和元器件成本；但Microsoft、Meta、Alphabe…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 0, "neutral": 0, "value": 80.8, "events": [ { "title": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决", "summary": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决 AI Capex Pivot to Grid/SMR: Has Consensus Already Been Repudiated by NVDA/MRVL Price Actio…", "sourceLabel": "研究主题", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 0, "value": 68.2, "events": [ { "title": "变压器与开关设备的长交付周期正在把AI算力扩张的关键变量从“宣布CAPEX”改写为“按时送电MW”，从而削弱新增AI资本开支向收入和FCF转化的效率。", "summary": "截至2026-06-06，本卡支持前序AI基础设施“第二导数”判断：瓶颈正从芯片供给转向可送电MW。DOE显示大型电力变压器常见交付期36个月、最长60个月；JLL显示数据中心设备平均交付期33周、美国42周，且2025年57%项目延误三个月以上；Sightline显示2026年计划投运的至少16…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 1, "risks": 0, "neutral": 0, "value": 83.6, "events": [ { "title": "头部CSP的AI投资回收期在HBM4紧缺下并未恶化，反而因稀缺产能高利用率和架构效率提升而缩短，支撑AI硬件的选择性估值溢价。", "summary": "截至2026-06-07，本卡支持并综合前序结论：HBM4紧缺不会打断头部CSP的AI资本开支周期，而是强化“配给能力+效率改善+现金回收”的分化。Microsoft、Alphabet、Amazon披露的需求、积压订单和AI收入信号表明，头部云厂商的有效回收期可能压缩至约1.6-2.3年；Meta…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-08", "supports": 1, "risks": 2, "neutral": 0, "value": 73, "events": [ { "title": "瓶颈已从芯片迁移到电力再到开关柜，AI篮子应将物理交付腿（公用事业+电气设备）从对冲位升为~25%同等支柱，巨头内维持GOOGL超配/MSFT低配/META第三条腿。", "summary": "截至2026-06-08，AI约束已从芯片迁移至电力再至开关柜，篮子权重应随之迁移：巨头核心降至55%（GOOGL 24%超配、MSFT 18%低配、META 13%非对称第三条腿，沿用C3结论），AI芯片从对冲升为核心权重15%，物理交付支柱（公用事业12%+电气设备13%=25%）从对冲升为同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决", "summary": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决 global-macro、thematic-researcher、tmt-analyst 三人独立得出 'capex 由 GPU 服务器向电网/SMR/变压器分流' 的同一结论（g…", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI 电力瓶颈的可投资因子已经从“需求增长”迁移到“2028 年前可通电、可签约、可合规回收成本的兆瓦”。", "summary": "截至 2026-06-08，本卡综合前序 AI 电力瓶颈链条，支持 Research note 9 将 BTM 速度溢价收窄为“可交付电力溢价”。配置上不再泛化买能源主题，而是超配拥有可签约稳定兆瓦的 IPP、具名数据中心管输/现场发电合同的中游天然气、以及受益于燃机与电气化设备短缺的 OEM；h…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 1, "neutral": 0, "value": 80.4, "events": [ { "title": "电表后天然气是物理上可行、政策上昂贵的桥接方案：它不推翻AI电力基础设施交易，但会给超大规模云厂商的2030净零叙事增加Scope 1披露、CDR供给和绿色声明合规折价。", "summary": "2026-06-09：本卡对卡5的天然气桥接结论进行政策压力测试。结论是，电表后天然气能解决2026-2029年AI数据中心通电缺口，但会把排放从可通过合同优化的Scope 2转成更难解释的Scope 1，并压缩Microsoft、Google、Meta、Apple等2030净零/负碳承诺的可信度…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核电交付时滞会压低2026-06至2027年的AI算力收入变现斜率，但不会立即等同于GPU采购需求见顶。", "summary": "截至2026-06-10，本卡支持前序判断但作出TMT侧区分：2026-06至2027年更可能出现“通电、可确认收入的AI算力容量”平台期，而非GPU订单断崖。Microsoft FY26 Q3资本开支319亿美元且约三分之二投向GPU/CPU，Amazon Q1 PPE购买额442.03亿美元，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 1, "value": 70.8, "events": [ { "title": "电力设备交期把2026-2030年AI基础设施周期从GPU供给约束转为已通电MW约束，美国路径需要实质性时间下修，但全球需求主线不被推翻。", "summary": "截至2026-06-12，本卡片对设备交期硬约束进行压力测试：Hyperscaler可通过预购设备、转移GPU、棕地扩容、表后发电和电源共址对冲部分风险，但无法完全绕开48-60个月级别的大型变压器/高压接入瓶颈。基准判断是，2030年美国可通电AI/数据中心容量较无约束capex计划下修15-2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "铜、铝、电缆和电气设备稀缺已经进入AI算力集群交付链，最先约束的是通电日期和可售云容量，而不是GPU本身。", "summary": "截至2026-06-11，材料链稀缺已开始转化为AI基础设施交付摩擦，但表现为通电时间、电气设备交期和云算力容量分配，而非所有GPU集群停摆。IEA提示约20%规划项目有延迟风险，CBRE称电力约束把工期推至2027年及以后，Wood Mackenzie给出关键电气部件18-36个月交期和600 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心从设备缺口转向电网/劳动力配给", "summary": "AI数据中心从设备缺口转向电网/劳动力配给 Meta $115M 技工学院、Lightcast 数据中心施工招聘 +23%/6个月、ERCOT 418GW 并网队列与 Abbott 成本转嫁限制同时出现；AI 基建交易应从设备 beta 切到电网准入、现场电源和施工资源稀缺吗？ Prior rel…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 0, "value": 57.2, "events": [ { "title": "截至2026-06-13，电力设备成本与交付风险已经实质性延后 AI 算力集群通电节奏，但它表现为 capex 效率恶化和项目重排，而不是最大型 hyperscalers 已经全面削…", "summary": "截至2026-06-13，本卡支持前序“电力约束正在惩罚 AI capex/融资假设”的方向，但精确化为：2026-2028 年已出现项目层面的通电延期与节奏重排，尚未出现头部 hyperscalers headline capex 的广泛下修。Sightline 显示 2026 年约 16 GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI并网延迟与贸易壁垒将2026-2028年的交易逻辑从“算力增长”推向“电力稀缺”，配置重心应从云厂商切换至基荷能源与具备价格传导权的设备龙头。", "summary": "截至2026-06-12，本卡战略综合了AI电力物理瓶颈的长期化影响：2026年Q1财报确认了“插电能力”取代GPU成为云厂商营收转化的首要约束。建议在跨市场配置中进行结构性切换：对北美超大规模云厂商（Hyperscalers）实施战术性减配（-10%至-15%），以应对资本支出效率下降及算力利用…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超大规模云厂商Capex仍以芯片为大头，但边际瓶颈和估值溢价正转向电力基础设施；AI芯片周期会被已通电MW重排为延期与集中释放并存的台阶曲线。", "summary": "截至2026-06-13，本卡支持“估值锚转向已通电MW”的主线，并将其落实到Capex结构：GPU/服务器仍约占AI基础设施资本开支的最大部分，但电气、制冷、变电站、并网和租赁型长期资产正在成为进度闸门。按公开指引，四大云厂商2026年Capex约6800-7100亿美元；1 GW AI数据中心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与数据中心基础设施虽然订单依然强劲，但并网物理瓶颈导致的项目延误正在迫使供应链中下游陷入应收账款增加和库存积压的“无现金增长”泥潭，现金转化率急剧下滑，估值去杠杆风险已步半导体后…", "summary": "2026-06-12：AI基础设施面临明显的去杠杆压力。虽然四大云巨头2026年Capex指引合计达7100亿美元，但其自身现金流恶化，Amazon的TTM自由现金流转负至-182亿美元。中游服务器厂商（Supermicro一季度FCF为-67亿美元）和A股电力/冷却企业（国电南瑞、英维克等）一季…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 5, "neutral": 2, "value": 58.6, "events": [ { "title": "硬件与能源通胀只能部分向软件端转嫁，头部云厂商2026年不至于利润率崩塌，但AI云业务的边际回报率和自由现金流将被结构性压缩。", "summary": "截至2026-06-14，本卡支持前序“资本强度陷阱”链条：云厂商可在稀缺GPU、预留容量、企业工作流和agent定价中转嫁部分成本，但公开API价格已被OpenAI、Anthropic、Google等竞争锚定，难以承受全面涨价。AWS、Azure、Google Cloud短期报表利润率未崩塌，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Higher-for-longer不会必然叫停AI capex，但会把市场关注点从“能否花钱”转向“每个增量GPU集群能否在折旧和电力成本吞噬定价窗口前跨过更高ROI门槛”。", "summary": "截至2026-06-14，本卡对卡片1的高利率与美联储战略模糊情景做压力测试：AI基础设施capex的关键脆弱点不是大型云厂商短期无钱可投，而是折现率、GPU短寿命折旧、电力并网与PPA成本共同抬高项目资本回收门槛。以三分之二6年期算力资产、三分之一15年期数据中心资产测算，200bp WACC冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2027年Hyperscaler AI资本开支的主要压力不是变压器涨价，而是通电延迟导致资本开支转化效率下降。", "summary": "截至2026-06-14，本卡支持但收紧上游供应链割裂结论：大型变压器再延长8-12周、成本再上升6-9%的直接资本开支压力可控，每1GW AI数据中心约增加6800万至2.05亿美元，但进度冲击更大。AWS、Azure、Google的核心风险不是负担不起设备涨价，而是2026-2027年资本开支…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高能耗GPU集群的大规模部署导致超大规模云厂商的资本支出全面向电力基建倾斜，在设备端过度拥挤的背景下，公用事业运营资产提供了更优的风险收益比。", "summary": "截至2026年中，全球电网物理极限已成为AI扩张的核心瓶颈。超高密度GPU（如GB200 NVL72，120kW+/机柜）推动云厂商2026年Capex升至6600亿-7250亿美元，其中逾60%投向电力与基建。美国数据中心电力需求年底将达41 GW。数据证实，拥挤度极高的电网设备已透支预期，而处…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 6, "neutral": 0, "value": 56, "events": [ { "title": "AI基础设施资本开支仍支持互连和设备链条，但投资确定性应优先放在PCIe/DDR5/MRDIMM/光互连，再看CXL和国产设备盈利兑现。", "summary": "截至2026-06-15，本卡支持上一卡关于AI服务器互连和国产设备的方向，但将确定性分层：Hyperscaler资本开支仍在上行，Microsoft、Alphabet、Meta、Amazon的最新披露共同验证AI基础设施支出强度；PCIe Retimer/Switch、DDR5/MRDIMM和光…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "有电力保障的AI基础设施Capex确定性上升；无电力保障的Capex将被电网价格、保证金和设备交期继续配给。", "summary": "截至2026-06-14，本卡综合判断：电力风险并未消失，但正在从未定价瓶颈转化为可预算、可筛选、可合同化的分配机制。PJM容量价、ERCOT大负荷规则、核电PPA和低空置率让头部hyperscaler的AI数据中心及GPU集群Capex更具名义金额确定性；但保证金、容量费、变压器交期和建设成本通…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与电网瓶颈对Tier-1 2026–2027并网影响已基本被锁单、棕地与表后发电对冲；真实滑期约3–7 GW集中在2028+绿地与Tier-2投机建设，Research not…", "summary": "本卡(10/10)压力测试变压器/GSU/电网交期对Tier-1 AI数据中心并网的真实约束。结论：瓶颈真实存在但Tier-1时间表已基本被定价——2022–2024年下达的多年期框架订单、棕地复用既有并网点、表后核电与现场燃气PPA共同吸收了120–210周的设备交期。自测算显示2027–202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "随着AI物理瓶颈从芯片转向电网可交付性，定价权正全面向上游电力资产与输配电设备商迁移，应超配瓶颈控制方并对电网依赖型云厂商及芯片标的引入折价。", "summary": "本卡片将2026年电网物理交付瓶颈转化为权益资产配置框架。建议超配控制变压器、液冷与在运核电等瓶颈环节的设备商与独立发电厂（如GE Vernova、Eaton、Vertiv、Constellation）；对依赖电网队列且发生租约撤销的云厂商（如Microsoft、Meta）引入20%-30%的电网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 5, "neutral": 0, "value": 55.4, "events": [ { "title": "评估电力供应瓶颈与电网扩容速度对AI算力集群建设及超大规模云厂商（Hyperscalers）资本开支的影响", "summary": "本卡支持并强化 Research note 01 的'公用事业→AI核心基座'论点：电力供应与电网/配电设备交付周期（而非GPU供给）才是当前AI算力部署与云厂商资本开支变现的真正约束。2026年五大云厂商合计资本开支约6,600–6,900亿美元，近100%吞噬经营现金流，但Microsoft约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理基础设施是耐久跨板块范式，但下周应按流动性情景控仓，而不是把液冷等高动量环节当作无风险防御。", "summary": "截至2026-06-15，本卡支持“AI资本开支下沉到电力/电网/散热”的中期范式，但强调交易入口已经拥挤。海外Microsoft、Meta仍披露大规模AI基础设施投入，国内算电协同、绿电直连和数据中心PUE政策强化物理瓶颈。下周默认给AI物理基础设施18%主动权益风险预算；若BoJ偏鹰导致流动性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力物理错配已迫使 Hyperscaler 把选址目标函数从'最低LCOE'切到'通电时间+固化清洁'，并把PPA模板从配储RES换为长期定价核电/固化基荷 — 卡片02方向幸存，传…", "summary": "压力测试确认卡片02方向成立但机制需升级：Hyperscaler 选址启发式已从最低LCOE切换至时间到电+固化清洁可得性。ERCOT仅2026Q1便收到198 GW大负荷申请，全美队列约2,600 GW、PJM到2030缺口15 GW，能保通电的站点享15–25%租金溢价。所有头部Hypersc…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI-to-power 主题应保留，但 BTM/SMR 只能作为期权，不能作为截至2026-06-16的组合中心。", "summary": "截至2026-06-16，本卡否定卡片六最强版本的 BTM 论点：AI 电力瓶颈成立，但证据不足以证明 Hyperscalers 资本开支已结构性转向自有/表后能源基础设施。Microsoft/Constellation 更像电网内购电协议，Amazon/Talen 的 BTM 扩容被 FERC …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 8, "neutral": 0, "value": 38.8, "events": [ { "title": "AI基建资本开支仍在加速，但从公告美元转化为已通电MW的比例正在下降，首个断点在数据中心交付和二线下游供应链，而不是NVIDIA近端GPU订单。", "summary": "截至2026-06-16，未看到北美云厂商AI资本开支或GPU订单总量已经广泛下修；Microsoft、Alphabet、Meta、Amazon、NVIDIA、Dell披露仍指向扩张。支持前序压力结论的证据主要来自数据中心交付层：Goldman Sachs预计未来1-2年计划容量仅50-60%按时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "结构性的能源和电网并网限制已经取代GPU供应，成为2026-2027年AI扩张的首要瓶颈，迫使超大型云厂商通过重启核电和发展离网发电开启资本密集的“抢电竞赛”。", "summary": "截至2026年6月中旬，压力测试表明结构性的能源与电网互联瓶颈已取代GPU产能，成为AI基础设施的首要限制因素。数据中心建设（18-24个月）与电网扩容（7-12年）的时间错配，导致云厂商转向核电与离网供电。资本开支正从算力向电力基建外溢。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与散热瓶颈重塑 ABF 需求的斜率与时点，但因 ABF 在封装上游锁定且单卡用量持续升级，无法构成结构性下台阶。", "summary": "电力与散热是限制 hyperscaler 部署 AI 硬件时点与地理的物理硬约束（PJM 2025/26 容量拍卖出清价 ~329 美元/MW-日为上年 9 倍；机架功率从 120 kW 跃升至 600 kW+），但并非 ABF 载板需求的结构性上限。ABF 在封装阶段较上架早 6–9 个月锁定，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理算力护城河对AI基础设施SPV租金与利用率稳定性的贡献评估 日期：2026 06 16 | 分析师：AI基础设施分析师 [ai infrastructure analyst] 核…", "summary": "物理算力护城河对AI基础设施SPV租金与利用率稳定性的贡献评估 日期：2026 06 16 | 分析师：AI基础设施分析师 [ai infrastructure analyst] 核心观点 (Thesis) 物理护城河（表后供电、高密度液冷、微电网）在2026年的电力稀缺环境下，已从单纯的“可用性…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 7, "neutral": 0, "value": 54.2, "events": [ { "title": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：…", "summary": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：AI 基础设施分析师 (ai infrastructure analyst) 立场：stress test（压力测试…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AIDD是AI基础设施需求的高质量增量因子：规模小于通用大模型，但持续性、数据壁垒和监管/IP属性更强。", "summary": "截至2026-06-18，本卡片支持上一张卡片的方向：AIDD不会单独决定hyperscaler capex，但会把高价值、持续性的生命科学工作负载接入GPU、网络、存储、液冷和电力保障型数据中心。Recursion BioHive-2、DCAI Gefion、Novo Nordisk、Isomo…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力瓶颈是 AI capex 的中期执行上限，而不是 2026-06-17 当下的叙事破裂触发点。", "summary": "2026-06-17 的压力测试结论：电力、电网与数据中心容量约束是真实且正在制度化的瓶颈，但尚不足以在 2026-06-16/17 FOMC 窗口内击穿 AI capex 叙事。LBNL、IEA、BloombergNEF、PJM 和 NERC 数据显示需求规模、在建容量和并网可靠性压力都在上升；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler capex 足以穿越能源 TCO 上行并支撑稀缺供应链，但投资表达必须是“多稀缺、空低 ROIC”，而不是无差别做多 AI。", "summary": "2026-06-18 压力测试结论：hyperscaler capex 增长足以让 AI 基础设施供应链消化能源成本带来的边际 TCO 压力，但只能支持“多稀缺”，不能支持泛 AI beta。四大买方 2026 capex 粗略包络约 $710B（自测算），Microsoft、Alphabet、M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 8, "neutral": 0, "value": 41.6, "events": [ { "title": "截至2026-06-19，前沿模型风险闸门更可能改变AI基础设施资本开支的节奏和形态，而不是终止云厂商的算力扩张。", "summary": "截至2026-06-19，本卡支持上一卡结论，但把传导机制落到AI基础设施：风险闸门不会简单压低hyperscaler资本开支，而是把节奏改为“电力/园区前置、GPU分段、合规容量池预留”。Anthropic Fable 5 / Mythos 5的2026-06-09发布与2026-06-12暂停…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装降低每bit能耗后，Hyperscalers更可能把能效红利用于扩大推理部署和提高每兆瓦收入，而不是削减资本开支。", "summary": "截至2026-06-19，本卡支持“先进封装与能效提升会缓解电力瓶颈并推动Hyperscaler二次部署”的判断，但强调这是受firm power约束的扩张而非无约束爆发。Microsoft预计2026日历年资本开支约1900亿美元，Alphabet预计2026年1800-1900亿美元，Meta…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM3良率瓶颈将AI基础设施风险从总量需求崩塌改写为本土投产节奏后移，并强化A股高融资TMT的收入兑现左尾风险。", "summary": "截至2026-06-18，本卡压力测试卡片06：国产HBM3良率瓶颈会推迟中国本土AI算力变现，但尚不足以证明全球AI基础设施从快速扩张转入需求观望。全球hyperscaler capex仍有Meta、Alphabet、Microsoft等高额指引支撑；真正变化在中国本土项目从按园区容量扩张转向按…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI交易的边际定价已经由软件久期切换为实体AI产能和电力瓶颈的资本开支执行，但软件并未失去长期价值。", "summary": "截至2026-06-18，本卡支持AI交易已从泛软件久期转向算力硬件、电气/热管理、电网接入和工业供能执行。核心证据是Microsoft、Amazon、Alphabet和Meta最新季度实体capex/PPE披露按粗略口径合计约1307亿美元，NVIDIA数据中心收入仍高速增长，IEA/美国能源部…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 4, "neutral": 0, "value": 67, "events": [ { "title": "变压器瓶颈会把Hyperscaler AI CAPEX从“买GPU即可扩张”改写为“先锁定可通电MW再投放GPU”，因此支持做多电气设备和电力韧性基础设施、谨慎做空BE/SMR。", "summary": "截至2026-06-19，本卡支持前序“削减BE/SMR空头、加配电气设备与电力韧性方案”的方向，但把因果链收紧为：变压器和开关柜交期只有在变电站、并网和确定性供电落在关键路径上时，才会直接推迟AI数据中心通电。100 MW AI园区若延迟12个月，约40亿美元AI基础设施资本会延后转化为收入。H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "16.18%的硬件Capex下修会压低AI基础设施整体估值，但订单传导不是线性的：液冷最敏感，电网设备和BTM发电更受长期电力容量瓶颈支撑。", "summary": "截至2026-06-19，本卡把H2 AI硬件Capex下修16.18%作为未外部确认的压力测试输入。结论是分层传导：GPU/服务器采购最先受冲击；液冷因绑定高密度机柜节奏，订单和估值敏感度较高；电网设备因并网稀缺、变压器长交期和backlog，订单不应按同等比例下修；BTM自备发电估值会先压缩，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网吸纳能力瓶颈会削弱AI投资对近期r*的无摩擦上行冲击，但会把周期拉长并转向电力设备资本开支，因此投资需求逻辑被折价而非被证伪。", "summary": "截至2026-06-19，本卡压力测试Research note 06的r*上行逻辑。结论是：电网与变压器瓶颈会降低近期资本开支效率，但不会取消投资需求渠道。Goldman估算2026-2031年AI基础设施资本开支为7.6万亿美元；IEA提示约20%规划数据中心项目存在电网延期风险。套用于202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备交期和并网流程已成为AI算力基础设施建设周期的确定性约束，并会放大高负债算力租赁商的融资与估值压力。", "summary": "截至2026-06-19，本卡支持前序判断：AI算力建设的硬瓶颈已从GPU采购转向电力交付。DOE/LBNL与EPRI均显示数据中心用电占比快速上升；Wood Mackenzie与POWER显示大型变压器交期已达120-144周，部分大型设备80-210周；JLL称主要数据中心市场平均并网等待超过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 10, "neutral": 0, "value": 32.4, "events": [ { "title": "CapEx前置会提高云巨头长期资本开支和固定费用压力，并把开发商信用风险重塑为租户信用、电力交付、起租和融资载体风险，而不是让风险消失。", "summary": "截至2026-06-20，本卡压力测试数据中心开发商“CapEx前置”模式。结论是风险被迁移而非消失：开发商通过预租、担保、LC和电力义务降低投机建设风险，但Hyperscaler承接更多融资租赁、未开始租赁、直接capex和专属电力费率。Alphabet、Amazon、Meta、Microsof…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI capex supercycle 仍在，但 higher-for-longer 与能源波动会把扩产闸门从芯片采购转向电力可交付性、容量成本和融资期限匹配。", "summary": "截至 2026-06-21，本卡对 higher-for-longer 与能源波动环境下 AI 算力基建做压力测试。结论：超大规模云厂商不会停止扩产，但边际决策从“GPU 可得性”转向“firm megawatts + 客户承诺 + 可融资全成本”。Amazon、Alphabet、Meta 已披露…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "summary": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler AI扩张面临的是“资本买不到时间”的物理瓶颈：有资产负债表不等于能压缩电网、变压器、许可和矿产供应周期。", "summary": "截至2026-06-21，本卡片对AI基础设施扩张做压力测试：Hyperscaler的主要风险不是单纯GPU短缺，而是可靠电力、PJM类大负荷接入、变压器/输配电设备、关键矿产精炼集中度和能源成本共同造成的物理交付瓶颈。IEA预计全球数据中心用电从2024年415 TWh升至2030年945 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 9, "neutral": 0, "value": 39.8, "events": [ { "title": "AI基础设施的边际稀缺点正从GPU扩散到可通电兆瓦和电网电气设备，因此制造出海主线应从拥挤硬件β轮动到“电力绑定的算力基础设施”。", "summary": "截至2026-06-21，本卡支持上一卡的再平衡结论：AI硬件存在战术拥挤，但云厂商资本开支并未结束，而是转向算力、电力、电网设备协同扩建。Microsoft、Alphabet、Meta、Amazon的2026资本开支披露或指引合计约7000亿美元（自测算），而DOE/LBNL、IEA、EPRI和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力容量、并网权和长周期电气设备的稀缺性为AI资本开支提供物理底部，但若负荷预测虚高、AI利用率不足或监管成本分摊恶化，该刚性会转化为新增项目放缓。", "summary": "截至2026-06-21，本卡支持上一卡“估值先受压、资本开支不立即崩塌”的判断。核心证据是电力与电网瓶颈正在把AI数据中心建设变成具备供给刚性的公用事业级工程周期：美国数据中心用电占比已从2023年的4.4%走向2028年6.7%-12%的区间，变压器交付仍需2-3年，Dominion等枢纽区域…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备交付瓶颈对超大规模云厂商(Hyperscalers)数据中心扩建进度与CapEx规划的影响", "summary": "截至2026-06-21，本卡从需求侧支持Cards 4-6的电力设备硬约束判断。云厂商对2026-2027变压器/开关柜断层的反应是双重的：既重排序路线图（30-50%的2026美国项目推迟/取消，迁向有电场地，算力分批绑定电力里程碑），又激进对冲（在拿证前提前下单长交期电气设备、垂直整合BTM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供应约束及通胀风险对AI基础设施（电力/数据中心）建设成本与扩张节奏的影响", "summary": "在AI数据中心TCO中电力仅约7%、算力约60%（Epoch AI），电价翻倍只抬高全口径成本约7%，远不足以撤销已指引的2026年约7,250亿美元Hyperscaler资本开支。故“电价上升→建设缩减”前提被部分否定：约束是电力可获得性而非电价。但卡01–02的通胀右尾确经三条间接渠道咬合——…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 0, "value": 83.2, "events": [ { "title": "在66.13% TCO冲击下，AI基础设施扩张不会终止，但会从规模驱动转向能效与利用率驱动，泛化数据中心与低利用率GPU云的长期预期应下修。", "summary": "截至2026-06-22，本卡压力测试上一卡66.13% TCO激增情景。结论是：头部hyperscaler资本开支不会急停，Microsoft、Alphabet、Meta、Amazon仍有收入、backlog或客户承诺支撑；但新增产能逻辑会从“抢GPU/抢机房规模”转向“单位功耗有效产出、签约利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体资本开支收缩会推迟AI集群扩建的判断成立，但只在AI专用供应瓶颈受冲击时成立；截至2026-06-22，更现实的约束是芯片稀缺、组件通胀与电网通电共同拉长边际集群排期。", "summary": "截至2026-06-22，本卡支持前序主线但收窄机制：半导体资本开支骤降会推迟Hyperscaler AI集群，前提是冲击落在TSMC先进制程、HBM、先进封装、网络和供电等AI关键瓶颈，而非普通消费芯片。现有证据反而显示AI关键资本开支仍在扩张：TSMC指向US$52-56 billion区间高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 6, "neutral": 0, "value": 60.6, "events": [ { "title": "2027年的电力冲击会抬高AI容量成本并重排Hyperscaler部署顺序，但更可能造成通电与收入确认延迟，而非CapEx总量或GPU订单的广泛取消。", "summary": "截至2026-06-23，本卡支持前序判断：2027年电力约束不会直接演化为Hyperscaler CapEx总量削减，而会把部署逻辑从“GPU优先”改为“通电优先”。Microsoft、Alphabet、Meta、Amazon披露的2026 AI基础设施支出仍然庞大，但PJM容量价格、美国DC电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler并不缺AI Capex，真正约束算力上线节奏的是可交付电力、并网流程与长交期电力设备。", "summary": "截至2026-06-24，本卡支持上一卡：AI算力瓶颈已从Capex/GPU转向通电MW。Alphabet 2026 CapEx指引1,800-1,900亿美元、Meta 1,250-1,450亿美元，Microsoft仍称需求超过可用容量，Amazon Q1购置物业与设备442.03亿美元；但S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下修。", "summary": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力瓶颈不会压低AI Capex总额，但会把Hyperscaler Capex从“买GPU”重构为“买可激活算力”：液冷、并网、电力设备和长期能源容量成为与GPU同等重要的部署变量。", "summary": "截至2026-06-24，本卡支持前序主线：AI算力约束已从GPU/Capex总额转向已通电MW及每MW效率。IEA预计全球数据中心用电2030年约945 TWh，LBNL测算美国2028年或达325-580 TWh；Sightline显示2026年16 GW计划容量仅5 GW在建、30%-50%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 7, "neutral": 0, "value": 56, "events": [ { "title": "持续能源通胀会压缩AI基础设施回报率，但真正改变Hyperscaler扩建节奏的是可交付电力的稀缺，而不是电费单项上涨。", "summary": "截至2026-06-25，本卡压力测试能源通胀对AI基础设施的传导。结论是：电价上升本身不足以触发Hyperscaler AI capex断崖；更关键的是并网、容量市场、PPA和电网设备交付约束会重排项目优先级。Microsoft、Alphabet、Meta、Amazon仍合计指向约6950亿-7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张下的电力供给与电网瓶颈风险", "summary": "截至2026-06-24，本卡对上一卡做压力测试：电力与电网接入正在成为AI硬件交付的真实节奏约束，尤其在美国数据中心负荷集中的电网区域，但当前证据更支持“区域通电延期与capex成本上升”，而不是“800G/1.6T需求被证伪”。IEA、Berkeley Lab、NERC、PJM与Monitor…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "受监管电网收益、rate base 扩张，以及数据中心并网成本分摊 公用事业行业研究报告 — 2026年6月24日 作者： 公用事业分析师 (utilities analyst) 本…", "summary": "受监管电网收益、rate base 扩张，以及数据中心并网成本分摊 公用事业行业研究报告 — 2026年6月24日 作者： 公用事业分析师 (utilities analyst) 本卡锚定日期： 2026 06 24 当前立场： 综合 (Synthesize — 结合卡片 01 的在岸防御性资产配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 1, "risks": 5, "neutral": 0, "value": 77.4, "events": [ { "title": "CNH套息去杠杆会先打掉A股AI基础设施股票的流动性溢价和估值久期，而不是立即打断全球AI CapEx实物周期。", "summary": "截至2026-06-25，本卡片压力测试CNH套息去杠杆对AI基础设施估值和CapEx预期的冲击。结论是：流动性溢价消失可使A股AI硬件/算力基础设施估值出现非线性回撤，但不应直接推导为全球hyperscaler CapEx崩塌。Microsoft、Alphabet、Meta、Amazon、Ora…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BTM是AI园区局部speed-to-power的有效工具，但在2026-2028年不足以填补全国AI电力缺口，反而会放大公用事业负荷预测兑现率与成本分摊的不确定性。", "summary": "截至2026-06-25，本卡从AI基础设施侧压力测试card-01：Hyperscaler BTM已从叙事进入执行，Microsoft/Chevron 2.67 GW燃气BTM、Crusoe/Microsoft 900 MW现场电厂、Talen/AWS 300 MW直连核电和xAI 247.2 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "summary": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的主要瓶颈正从GPU可得性转向电力可得性，真正受益的是能缩短通电时间、提高负荷密度并保障可靠供电的设备与电网环节。", "summary": "截至2026-06-26，本卡支持上一张“算力切向电力”的轮动逻辑，但把核心约束细化为并网、变压器/开关设备、机柜功率密度和可靠供电。全球数据中心用电预计由2024年415 TWh升至2030年945 TWh；中国2030年约279 TWh，全国占比不高但局部冲击显著。1 GW AI IT负荷在P…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 1, "neutral": 0, "value": 94.8, "events": [ { "title": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "summary": "截至2026-06-27，本卡片对AI基础设施链做压力测试：全球AI资本开支仍大，Goldman Sachs基准模型给出2026年USD 765 billion、2031年USD 1.6 trillion的年度AI capex，但Amazon、Alphabet、Meta、Microsoft的数据表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 4, "neutral": 0, "value": 78.2, "events": [ { "title": "变压器材料和交付瓶颈已开始改变 Hyperscaler 的2026-2027资本开支与电力采购节奏，影响主要体现为预算通胀、长期锁电和通电交付递延，而非AI需求取消。", "summary": "截至2026-06-28，结论为支撑：Hyperscalers 已经通过预算上修、长期电力合同和项目节奏调整来应对变压器及电气设备瓶颈，但公开披露通常不会把变化直接归因为 GOES 或铜关税。Microsoft 2026 capex 中约250亿美元来自更高组件价格，Meta 上调2026 cap…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高利率平台期对AI基础设施资本开支的影响", "summary": "压力测试卡片04的3.75%-4.00%高位平台前提：利率平台并非2026年头部Hyperscaler资本开支（约7250亿美元、+77%、约75%为AI）的约束性瓶颈，因四巨头仍主要自筹、AA/A级利差仅约80bp。但融资模式恰在约2026Q3翻转为债务驱动（整体资本开支超过经营现金流；Orac…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电子链条利润反弹主要受AI基础设施投资拉动，具备向存储、设备、材料、先进封装、光互联和数据中心电源散热扩散的潜力，但尚不足以证明消费电子和内需全面复苏。", "summary": "截至2026-06-28，本卡支持上一卡判断：电子链条利润修复主要由AI基础设施、算力芯片和存储涨价驱动，而非内需广谱复苏。国家统计局数据显示，1—5月电子行业利润同比增长103.9%，贡献全部工业利润增量43.1%。NVIDIA、Microsoft、Alphabet、Meta、Alibaba和T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 3, "neutral": 0, "value": 93.6, "events": [ { "title": "AI交易不会结束，但估值核心应从买到多少芯片转向能否按期通电、保持利用率并降低每token能耗。", "summary": "截至2026-06-28，本卡综合支持前序判断：2026-2030年AI扩张的约束从服务器交付转向可通电MW交付。Amazon、Microsoft、Alphabet、Meta最新季度CAPEX/PPE购买合计约$130.8B[自测算]，但IEA预计美国数据中心用电到2030年较2024年增加约24…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "summary": "截至 2026-06-28，本卡压力测试认为：电力交付瓶颈会迫使 hyperscaler 从单纯追求新增 MW 转向 MW 生产率和存量优化，但不会终结扩产。Microsoft、Alphabet、Meta 可见 2026 年 capex 区间合计约 4,950-5,250 亿美元（自测算），同时仍…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI数据中心‘电力荒’：电网接入瓶颈下的超大规模云厂商资本开支与自备电站转型", "summary": "截至2026年6月29日，超大规模云厂商集体资本开支预计在2026年达到7250亿美元，其中约75%用于AI基础设施。然而，AI规模扩张的核心制约已由GPU芯片转为电网并网容量。美国PJM电网并网排队超5年，Dominion电网仍有45 GW容量待分配。这正倒逼微软、亚马逊等巨头直接投资天然气微电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-72097390dd/", "networkHref": "/investor/research-network/#thesis-72097390dd" }, { "id": "thesis-e1d3ec2fed", "source": "auto_topic", "title": "AIDC 并网与变压器瓶颈：AI 投产约束复盘", "titleEn": "Practical research topic: AIDC 并网与变压器瓶颈：AI 投产约束复盘", "lane": "电力与公用事业", "laneEn": "电力与公用事业", "laneHref": "/investor/lanes/lane-ee7fbf3f/", "investableFocus": "", "stockUniverse": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" } ], "stockTickers": [ "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH", "002028.SZ", "APO", "IEF", "TLT", "CAT", "588000.SH", "COP" ], "coreView": "该主线来自 4 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-22 至 2026-06-06；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-04-29", "lastSeen": "2026-06-29", "supports": 8, "risks": 148, "neutral": 4, "analystCount": 14, "sourceTypeCounts": { "分析协作": 1, "研究笔记": 154, "研究主题": 2, "深度研究": 1, "研究线程": 2 }, "topTerms": [ "AI数据中心", "GOES", "数据中心", "电力设备", "取向硅钢", "变压器", "电力", "设备", "A股", "燃气轮机", "AI电力", "业基建产能深度绑定" ], "whatChanged": [ "2026-06-29: 上游取向硅钢和电工铜的产能稀缺与关税高墙正转嫁为严重成本通胀，侵蚀设备制造厂毛利并拉长交付，导致AI电力落地面临实质性溢价风险。", "2026-06-29: 关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "2026-06-28: 电力设备链是可投资的多年超级周期，但赢家来自高压瓶颈和积压订单质量，2029年后需警惕燃气轮机与边际电气产能的消化风险。", "2026-06-28: AI电力基础设施是2026年更确定的A股超配方向，锂资源是2027年周期修复期权；枧下窝和成本曲线决定加仓时点，而不是否定复苏逻辑。", "2026-06-28: AI电力基础设施供应链瓶颈：电力变压器与关键配电硬件的产能缺口" ], "practical": { "sectorLane": "电力与公用事业", "sectorLaneEn": "电力与公用事业", "exposureBucket": "电力与公用事业", "tone": "yellow", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" } ], "stockTickers": [ "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH", "002028.SZ", "APO", "IEF", "TLT", "CAT", "588000.SH", "COP" ], "primaryStocks": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" } ], "instrumentTags": [ "GEV", "GE Vernova", "ETN", "Eaton", "PWR", "Quanta Services", "VST", "Vistra", "CEG", "Constellation Energy", "XLU", "Utilities Select Sector SPDR" ], "reviewCadence": "每日复核", "confirmationSignals": [ "并网排队和变压器交期是否继续拉长", "表后自发电与 PPA 是否改变公用事业信用指标", "电力设备订单质量是否高于收入确认节奏" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "长端利率", "铜/GOES 成本", "监管电价", "信用利差" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 111 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 14, "recentEvents": 111, "supportRatio": 0.05, "riskRatio": 0.93, "conflictScore": 100, "evidenceDepth": 100 }, "score": 65.7 }, "events": [ { "date": "2026-06-29", "title": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "summary": "本卡片从制造和供应链视角对电网与电力约束进行压力测试。截至2026-06-29，关键电力设备（变压器与重型燃机）由于极高工艺与测试门槛、上游取向电工钢（GOES）瓶颈及寡头垄断，交期拉长至3-7年，生产机位已售罄至2029-2030年。云厂商BTM发电加剧了燃机机位争夺。预计2026-2027年该瓶颈将导致云厂商约800亿-1100亿美元资本开支递延兑现。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "GOES", "电力设备", "变压器", "电力", "设备", "电网" ], "href": "" }, { "date": "2026-06-29", "title": "上游取向硅钢和电工铜的产能稀缺与关税高墙正转嫁为严重成本通胀，侵蚀设备制造厂毛利并拉长交付，导致AI电力落地面临实质性溢价风险。", "summary": "压力测试表明，取向硅钢与电工铜的供应缺口已完全固化在变压器与燃机的长交期中，不构成独立的“二次时间瓶颈”。但地缘摩擦与美国50%的Section 232关税，导致中美GOES价差拉大至5.16倍，美国面临严重的材料合规溢价。在重度成本冲击下，固定价格合同变压器厂毛利率收缩至-3.67%，逼迫出口链向东南亚与墨西哥转移（临界关税率13.16%）。此外，5 G…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "GOES", "取向硅钢", "变压器", "电力", "设备", "AI电力" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "summary": "本卡支持 Research note 02「上游原材料是最深约束」的结论，并锐化为一组非对称结构。GOES：自2019年变压器需求暴增（电力+119%、GSU+274%），远超GOES市场约5.4–5.6%的CAGR；GSU供给缺口2025年约100%，因美国唯一GOES生产商Cleveland-Cliffs（Butler Works，扩产2028上线）及…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 41.5, "matchedTerms": [ "GOES", "数据中心", "电力设备", "变压器", "电力", "设备", "核心", "电网" ], "href": "" }, { "date": "2026-06-28", "title": "2026年全球AI数据中心电网瓶颈与“自主供电”战略的崛起", "summary": "2026年全球超大规模厂商的AI算力基建受限于电网瓶颈与关键电力设备（变压器交期达5年）短缺。这迫使厂商将部分资本支出（预计2026年总支出达7250亿美元）从纯算力（GPU）转向自主供电（BYOP）和表后（BTM）能源方案。预计该转移将延迟算力上线速度，并推高长期持有成本。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 41, "matchedTerms": [ "AI数据中心", "数据中心", "电力设备", "变压器", "电力", "设备", "电网" ], "href": "" }, { "date": "2026-06-28", "title": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "summary": "本卡压力测试逻辑链：GOES约束不会「抵消」核能加速，因为二者作用于电力链不同节点——核能重启在发电/并网侧本就是GOES-轻（复用既有GSU与开关站），属于一种GOES套利；但GOES在负荷侧与输配路径重新绑定，沿用Research note 01的NPV模型测算，核能锚定型项目(≈1.5年延迟,NPV≈−27%)相对并网型绿地(3年,−51.6%)只回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 41, "matchedTerms": [ "AI数据中心", "GOES", "数据中心", "电力设备", "变压器", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力需求增长对电网侧扩容及配电资产的压力测试", "summary": "压力测试结论：前序主线被证实并强化。2026年AI落地的约束性瓶颈已非GPU，而是电网边缘——电力变压器、中高压开关设备、取向硅钢（GOES）与接网排队。大型变压器交期3–5年、全球产能利用率约98%、美国接网积压1,500–2,300GW、枢纽接入排队4–7年，致美国2026约12GW管线仅约4GW在建（隐含约67%滑期，自测算）。这把AI算力部署拉长而…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 37.5, "matchedTerms": [ "GOES", "取向硅钢", "变压器", "电力", "设备", "AI电力", "核心", "电网" ], "href": "" }, { "date": "2026-06-28", "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心腿；半导体国产替代（¥2万亿AI建设计划+80%国产强制比例…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.5, "matchedTerms": [ "电力设备", "变压器", "电力", "设备", "A股", "核心", "电网" ], "href": "" }, { "date": "2026-06-28", "title": "DeepSeek API的大幅通缩不仅不会消灭算力需求，反而会通过吉文斯悖论激发Agent调用量井喷，使净电力需求暴增15.36倍，进一步固化并收紧以变压器（交期120-210周）和并网（排…", "summary": "2026年6月26日DeepSeek API大幅降价96.37%引发散户算力需求毁灭的恐慌，导致A股电子板块两融强平盘爆出。本报告通过可证伪判定框架（流量弹性、调用量、并网申请）论证其为吉文斯悖论的放大：Agent爆发与MCTS使Token消耗飙升，即便算力能效提升，数据中心电力净需求仍增15.36倍。电网变压器（LPT交期120-210周）物理制约未解，…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.5, "matchedTerms": [ "数据中心", "变压器", "电力", "设备", "A股", "核心", "电网" ], "href": "" }, { "date": "2026-06-28", "title": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费成本转嫁挤压。", "summary": "基于国产芯片2.94倍能效赤字拉长通电时间至18个月的物理瓶颈，本报告系统性量化了A股电力设备板块（特高压、大变压器）的估值扩张机制与低利润 traditional TMT 的电价挤压。电网容量受限与十五五规划共振，支持了特高压订单可见度与大型变压器出海高溢价，特别是在国内外硅钢5.1倍巨大价差的成本红利下。策略上构建杠铃组合，超配直接出口变压器龙头、特高…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.5, "matchedTerms": [ "电力设备", "变压器", "电力", "设备", "A股", "核心", "电网" ], "href": "" }, { "date": "2026-06-28", "title": "在2026-06-28，物理加电瓶颈正迫使AI算力估值由单纯的Capex扩张转向已通电MW效率，投资者应超配锁电、BTM自发电超大型云厂商与掌握变压器等硬缺口产能的电力设备商，并在A股中向特…", "summary": "截至2026-06-28，本报告探讨了在AI电力约束下算力资产估值向MW效率转型的逻辑。对于超大规模云厂商，2年的并网延迟将对100 MW智算中心产生严重的NPV减记（-124%）[S12]，迫使其估值分层：超配锁电与BTM自发电龙头，低配单纯Capex扩张者。电力设备商成为AI周期收费站，超配大型变压器等掌握稀缺产能与电网智能化的龙头。A股市场在“算电协…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.5, "matchedTerms": [ "电力设备", "变压器", "电力", "设备", "A股", "AI电力", "电网" ], "href": "" }, { "date": "2026-06-27", "title": "功率半导体已过剩，AI电力的新物理瓶颈已迁移至大型变压器，并真正卡在无法被资本快速绕行的上游电磁钢(GOES)与铜——2026–2028窗口难解。", "summary": "支持并收紧前五卡逻辑链：功率半导体转入结构性过剩后（Research note 05），AI电力建设的刚性瓶颈已迁移至大型变压器(LPT)，并下沉到上游电磁钢(GOES)与铜。瓶颈呈两层：整机厂总装层可用约20亿美元北美扩产去瓶颈，但缓解要到2028/2029年后；标准电力变压器交期128周、GSU 144周、定制订单4–5年，美国机队缺口约30%[S1]…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.5, "matchedTerms": [ "GOES", "数据中心", "变压器", "电力", "设备", "AI电力", "电网" ], "href": "" }, { "date": "2026-06-27", "title": "宽禁带半导体(SiC/GaN)仅能优化数据中心机柜内部能效，无法解决中压电网侧变压器交期超150周及GOES材料短缺的电网接入瓶颈，AI算力扩张的电网硬约束与云厂商资本开支递延依旧确立。", "summary": "截至2026年6月28日，本报告评估了通过SiC/GaN电力电子解决AI电力瓶颈的硅端技术路径及高压变压器产能瓶颈。硅端SiC/GaN虽能提升机柜能效，但属于机柜内部技术，中压渗透率到2030年仍<5%，且面临镓出口限制下部署容量缩水11.78%的断供风险。电网端高压变压器交期达150-200周，且受制于取向硅钢(GOES)及铜的硬短缺。材料压测显示，中度…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.5, "matchedTerms": [ "GOES", "数据中心", "取向硅钢", "变压器", "电力", "AI电力", "电网" ], "href": "" }, { "date": "2026-06-27", "title": "电力设备扩产会缓解AI数据中心并网瓶颈，但无法在2026-2027年完全解除大型变压器和输电接入的物理约束，因此算力利润扩散仍是受瓶颈约束的电网资本开支周期。", "summary": "截至2026-06-28，本卡支持上一卡的物理瓶颈判断，但强调扩产的时间结构：变压器、开关柜和电网设备扩产是真实的，Eaton、Hitachi Energy、Siemens Energy等制造商已披露新增产能和高积压订单；但大型电力变压器仍受定制设计、GOES/铜、测试槽位、运输和并网研究约束，新增产能更多改善2028-2030年供给，而难以完全解决202…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 38, "matchedTerms": [ "AI数据中心", "GOES", "数据中心", "电力设备", "变压器", "电力", "设备", "电网" ], "href": "" }, { "date": "2026-06-27", "title": "贸易壁垒导致的取向硅钢（GOES）结构性割裂，以及2026-2030年铜矿的百万吨级结构性缺口，奠定了电力变压器成本的向下刚性，而铝代铜的技术陷阱因被动增加25%的硅钢用量而被超大规模云厂商…", "summary": "基于2026-06-27权威工作日，本报告支持AI电力并网瓶颈已下沉至原材料端的判定。由于贸易壁垒隔离，中国占56%的全球取向电工钢（GOES）产能无法被美欧替代。同时，电网扩容拉动铜需求，2026年铜矿缺口达33万吨，并于2030年扩大至150万至220万吨。这导致变压器制造成本刚性上涨。试图通过铝代铜降本的“技术陷阱”会使硅钢用量增加15%–30%，反…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "GOES", "取向硅钢", "变压器", "电力", "AI电力", "电网" ], "href": "" }, { "date": "2026-06-27", "title": "数据中心电力并网、关键电力设备与液冷温控的物理交付瓶颈将对算力上线速度形成硬约束，导致云厂商ROIC受损并延缓算力采购，增加电子产业链利润扩张的硬着陆风险。", "summary": "截至2026-06-28，本卡压力测试了算力产业链利润扩散假说，指出电力并网与变压器超长交期（MV变压器128周）已成为数据中心上线的物理瓶颈。预计2026–2028年美国35%–45%的新建数据中心面临延期通电，16GW计划容量仅5GW在建，30%-50%将延期至2028年以后。GB200的高功耗（单柜120-132kW）使直接芯片液冷成为刚性指标，但A…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "数据中心", "电力设备", "变压器", "电力", "设备", "A股" ], "href": "" }, { "date": "2026-06-27", "title": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "summary": "截至2026-06-27，本卡片综合前序结论：电网/变压器瓶颈会延后并改道AI部署，但不会摧毁2028年前光模块出口龙头业绩能见度；真正尾部风险是2027年云厂商ROIC压缩。建议A股AI物理瓶颈组合使用40/32/12/8/8结构：变压器/电网设备、光模块、电力现金流锚、铜/GOES等输入端对冲、现金/对冲。光模块腿已拥挤，完整杠铃尚未终局拥挤，应在内部…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "GOES", "变压器", "电力", "设备", "A股", "电网" ], "href": "" } ], "timeline": [ { "date": "2026-04-29", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "[auto-handoff] AI投资重心从算力向电力的切换", "summary": "致 社交媒体分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 48581efe-320e-433e-b013-db375bc7645f 结果摘要: 我已完成对“电力基建（出海链条）是否取代AI芯片成为叙事锚点”的深度调研。通过分析全球电网资本开支周期、AI数据中心电力缺…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "AI 电网超级周期的‘硬通货’已转向高等级 Hi-B 取向硅钢与特种铜材绕组，全产业链集成与全球瓶颈组件（如分接开关）是获取超额收益的核心。", "summary": "本卡片深度分析了 A 股电网材料链在 AI 电力基建周期中的供需弹性。研究发现，美国 DOE 能效标准的调整使得高等级取向硅钢（GOES）重回电网支点。宝钢股份（600019）凭借 0.18mm 超薄 Hi-B 硅钢掌握底层定价权；望变电气（603191）通过‘硅钢-铁芯-变压器’全产业链集成实现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "\"新质生产力\"政策下的中国电网设备出海：退税政策校准与贸易壁垒对冲策略 第8张卡片（共10张）| 中国宏观分析师 | 立场：综合 本期会话主题： 美国 AI 数据中心负荷增长与公用事…", "summary": "\"新质生产力\"政策下的中国电网设备出海：退税政策校准与贸易壁垒对冲策略 第8张卡片（共10张）| 中国宏观分析师 | 立场：综合 本期会话主题： 美国 AI 数据中心负荷增长与公用事业电网灵活性 本卡主题： \"新质生产力\"框架下的电网设备出口退税政策调整与 Section 232/301 关税规避…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 3, "neutral": 0, "value": 26.8, "events": [ { "title": "A股变压器/电网设备30-40倍PE：AI送电瓶颈主题是已充分定价还是仍被低估？", "summary": "A股变压器/电网设备30-40倍PE：AI送电瓶颈主题是已充分定价还是仍被低估？ A-share Grid Equipment 30-40x PE: Is AI Power Delivery Bottleneck Fully Priced or Still Undervalued? global-…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "盟友联盟对中国电网设备的部分排除是一项主动用 6–12 个月送电延迟和 15–25% 采购成本溢价换取供应链韧性的宏观选择，结果是延展盟友设备 OEM 超级周期、部分上限美国 AI …", "summary": "本卡作为 10 张卡片的最终综合，整合了从需求（Research note 01）、firm 负荷修正（Research note 02）、供给瓶颈（Research note 03）、设备防御性（Research note 04）、定价权（Research note 05）、跨资产轮动（Rese…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "技术深度是中国电网设备出海的核心护城河，但在美国AI数据中心场景中，第三国制造和退税只能提供部分对冲，真正决定订单能否进入关键供电路径的是可信交付、网络安全、认证和本地服务能力。", "summary": "截至2026-05-06，本卡对第8张卡片作有条件压力测试：技术深度确实比地理套利更重要，但第8张卡片需修正三点。第一，2024年11月中国退税公告取消的是铝材、铜材和相关油脂退税，并将部分成品油、光伏、电池、非金属矿物制品退税率由13%降至9%，不是新的钢铁/GOES退税事件。第二，美国AI数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "AI电网设备尚未见顶，但2026下半年必须从主题弹性切换到订单、材料、交付和利润率验证，A股配置应超配高质量电网设备龙头、低配低端beta。", "summary": "截至2026-05-13，本卡综合判断AI电网设备交易尚未出现决定性见顶信号，但已从广谱beta进入盈利验证期。2026下半年需重点观察三类信号：超大规模厂商是否维护或下修2027年AI基础设施资本开支，GOES和大型变压器交期是否明显缓解，上市OEM订单能否转化为收入和稳定利润率。A股配置上应维…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI能效提升改善每Token经济性，但不降低单站点峰值MW与单区域MVA装机，电网设备重定价不被能效证伪。", "summary": "截至2026-05-13，AI芯片与模型的能效提升无法显著缓解电网侧变压器与中压配电压力。FLOPS/W每两年翻倍、MoE把推理$/Token压缩5-10倍，但单机柜功率从H100的30-40 kW升至GB200 NVL72的120-140 kW、Rubin Ultra约250 kW，且同步all…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "AI部署瓶颈正从芯片扩散到可送电负荷，全球电网升级正在把中国高压与数据中心电力设备出口商推入订单和利润释放期。", "summary": "截至2026-05-14，本卡支持前序GRID/PAVE轮动，并把中国电力设备出海加入AI基础设施主线。IEA预计2026-2030年全球电力需求年均增长3.6%，数据中心用电到2030年约翻倍至945 TWh；电网投资到2030年需升至每年6000亿美元以上。中国2025年电力设备出口约374亿…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国和欧盟不会消灭中国电力设备出海周期，但会把发达市场收入从高质量利润释放项变成需要政策折价的订单。", "summary": "截至2026-05-14，本卡压力测试但不推翻AI电网设备出海逻辑。全球变压器和电网设备短缺仍支撑中国供应商订单，但美国和欧盟正在用关税、采购韧性、网络安全、外国补贴审查和零部件贸易救济压缩发达市场利润率。美国风险最高，Section 301/232、Section 122临时附加费和结构性产能过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "工业制造分析：变压器“物理墙”与特高压产能缺口 日期： 2026 05 16 分析师 ID： industrials analyst 状态： 支持 (基于卡片 01 05) 1. 核…", "summary": "工业制造分析：变压器“物理墙”与特高压产能缺口 日期： 2026 05 16 分析师 ID： industrials analyst 状态： 支持 (基于卡片 01 05) 1. 核心摘要：产能是新的超额收益 (Alpha) 截至2026年5月，AI电力的“物理墙”已从电网接入配额转向实际的硬件供…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "全球宏观分析：地缘政治脱钩下的 AI 电力供应链重构与 ROI 压力测试 日期： 2026 05 18 分析师： 全球宏观分析师 立场： 压力测试 (Stress Test) 执行摘…", "summary": "全球宏观分析：地缘政治脱钩下的 AI 电力供应链重构与 ROI 压力测试 日期： 2026 05 18 分析师： 全球宏观分析师 立场： 压力测试 (Stress Test) 执行摘要 截至 2026 年 5 月，AI 电力叙事正在从简单的“稀缺”转向结构性的“通胀危机”。电力设备供应链（特别是特…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "电力设备是眼前的“订单簿瓶颈”，但终极约束是 7-10 年的电网准入周期，以及 AI 需求响应极低的弹性。", "summary": "电力设备（尤其是变压器）的极长交付周期与取向硅钢 (GOES) 的结构性缺口，已成为 AI 算力扩张的硬性物理约束。压力测试显示，AI 负载的极高资本开支决定了其“硬电力”需求特征，使得所谓的“需求响应”在经济上不可行。最终，2 年的 AI 建设周期与 7-10 年的电网升级周期之间的时空错配，构…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备OEM的积压订单验证了AI电力基础设施稀缺溢价，但同一批积压也证明近端供应不足以让所有2026-2027年AI数据中心公告容量按期落地。", "summary": "2026-05-20：支持AI估值从GPU叙事转向可交付电力基础设施，但压力测试显示供给只能选择性支撑，而不能普遍支撑所有已公告AI数据中心。Eaton、GE Vernova、ABB、Siemens Energy、Hitachi Energy、Schneider、Hubbell和Powell均显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 2, "neutral": 1, "value": 44.2, "events": [ { "title": "在系统性ERP下行期，顺周期进攻第一梯队应聚焦于具备核心高毛利出海定价权的电力设备以及受益于动车组五级修维保周期刚性爆发、实现超高运营杠杆转化的轨交装备。", "summary": "截至2026-05-21，随系统性ERP压降与早周期主线确立，先进制造细分板块因不受地产负债表侵蚀成为核心配置。运营杠杆测算显示：轨交装备高级修因五级修爆量（2025/2026年均约550-600组）及高毛利率（~38%），运营杠杆达3.8-4.5，利润弹性极强；电力变压器出海在AI数据中心与欧美…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "铜、GOES及环烷基绝缘油等关键材料极低的供应弹性已构成物理硬顶，原材料刚性赤字及价格向重电装备传导将延缓全球AI算力集群的交付进程。", "summary": "截至2026-05-20，本卡片对前序电力设备瓶颈进行原材料层面的压力测试。铜、高磁感取向电工钢（GOES）及环烷基变压器油已陷入深度结构性短缺，供应弹性极低。铜价暴涨（LME现货USD 13,410/t）、硅钢冷轧扩产周期长（3-4年）以及特种原油海运地缘摩擦，导致物理供应硬顶与毛利二次受压。铝…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-05-20，AI电力瓶颈的核心不是全国缺电，而是高规格变压器和高压开关设备的工厂槽位、测试能力、交付周期与原材料可得性。", "summary": "截至2026-05-20，我们支持前序结论：AI算力扩张瓶颈已从全国总发电量转向站点级重电设备交付。Eaton、GE Vernova、Siemens Energy、Schneider、Hitachi、韩国UHV厂商及TBEA/Pinggao均显示订单快于可交付产能。Allianz称电网设备交付最高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 3, "neutral": 0, "value": 44.6, "events": [ { "title": "变压器等电力设备的极度短缺及取向硅钢供应瓶颈验证了“时间尺度错配”，AI数据中心的扩张将持续受制于3-4年的物理电网硬件交付周期。", "summary": "截至2026年5月22日，全球电力设备市场面临极度紧缺。西门子、GE等四大设备商订单积压创历史新高，大型电力变压器交付周期拉长至36-48个月。上游取向硅钢（GOES）受制于北美产能垄断及全球供应不足，成为扩产的核心瓶颈。设备可用性已取代资金，成为AI基建最硬的约束。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力扩张已通过 OEM 订单与单价同时验证了电网设备的需求确定性；在 GOES 与变压器产能的多年瓶颈下，未来 24–36 个月该子板块兼具量价齐升与 200–400bp 利润…", "summary": "截至 2026-05-22，AI 数据中心已切实成为电力基础设施的边际驱动力，GEV 电气化订单同比 +86%、ETN 数据中心订单同比 +240%、ETN 披露的美国数据中心积压达 228GW；高功率变压器交付周期延至 4–5 年，GOES 取向硅钢扩产周期 3–5 年；国网十五五投资 4 万亿…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张确实开启电力基础设施资本开支周期，但2026-2028年变压器、并网排队和24/7可靠性要求会先放大时间尺度错配，再逐步由费基投资、储能和可调度电源缓解。", "summary": "截至2026-05-22，本卡支持并强化能源分析师的“时间尺度错配”判断：AI数据中心负荷增长快于电网硬件、稳定电源和新能源并网的交付速度。IEA预计全球数据中心用电量从2025年485 TWh升至2030年950 TWh，DOE/LBNL预计美国数据中心2028年用电量达325-580 TWh；…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "上游基础材料（铜与取向电工钢）的结构性短缺与通胀，使得工业产能扩张无法顺利落地，变压器与电力配给材料成为拖累AI资本开支转化为自由现金流（FCF）的绝对红线。", "summary": "截至2026-05-23，本卡片通过压力测试确认，上游材料（特别是铜与取向电工钢）的短缺和价格通胀，正在把变压器与液冷设备的扩产转化为AI基础设施的第三个硬性瓶颈。铜价已在2026年5月逼近14000美元/吨高位，单MW AI数据中心需消耗30-47吨铜；取向电工钢（GOES）极度依赖进口且产能集…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "A股AI主线应从“买算力久期”升级为“买全球电力稀缺的可交付环节”，并用合规分层与事件对冲改造变压器出口敞口。", "summary": "截至2026-05-23，本卡综合并有条件支持前序主线：AI需求未反转，但可投资瓶颈已从纯算力久期迁移到电力可获得性、电网交付、液冷密度与贸易合规。建议将A股主题仓位重构为国内电网与公用事业35%、合规深度电力设备25%、数据中心供配电与液冷20%、订单可见AI硬件10%、对冲或现金10%。变压器…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" } ] }, { "date": "2026-05-24", "supports": 1, "risks": 1, "neutral": 0, "value": 59.4, "events": [ { "title": "AI基础设施的近期上限将由可靠通电MW决定；电力在拥挤区域会先于GPU成为硬约束，但全球层面仍是区域性、可部分迁移的约束。", "summary": "截至2026-05-24，本卡压力测试结论是：AI数据中心电力与电网设备约束在核心拥挤区域已经先于GPU形成硬约束，但不应上升为全球绝对缺电叙事。GB300 NVL72单机架功率达132-140 kW，100 MW IT功率仅支撑约714-758个机架；同时IEA、DOE/LBNL、NERC和NI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球电网超级周期下，电力设备出海呈现剧烈的毛利兑现分化：唯有拥有核器件垄断壁垒、高端市场设备直销认证以及本地化产能网络的环节能将订单转化为高毛利率，而低门槛的发展中国家电网EPC工程…", "summary": "截至2026-05-25，本卡支持以出口驱动的电力设备高端制造作为A股进攻端核心的判断，但强调出海链内部利润兑现路径呈剧烈分化。2025年中国变压器出口达646亿元（+35%），2026年1-2月高压变压器出口增长率达61%，对美变压器出口同比增长182%，显现出极强的供需紧张与中方交付交期套利优…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "A 股电力设备正从‘国内周期标的’重估为‘全球 AI 基建稀缺核心供货商’，其显著的交期优势与估值洼地支撑权重上调。", "summary": "A 股电力设备龙头正凭借 10-12 个月的极致交期（对比海外 36-48 个月）和对取向硅钢 (GOES) 的产能垄断，成为全球 T&D 瓶颈的唯一‘泄洪口’。国家电网 4 万亿（2026-2030）投资计划及 8500 台具身智能设备招标，确立了 A 股‘AI+电力’的增长底部。目前 A 股龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 4, "neutral": 0, "value": 36.2, "events": [ { "title": "全球AI电力瓶颈正将科技资本开支与重工业基建产能深度绑定，通过跨美股与A股的变压器、特种硅钢及燃气轮机核心元器件联合组合，可有效对冲并捕获这一物理奇点的溢出利润。", "summary": "AI数据中心爆发面临5-7年并网与3-5年变压器（源于取向硅钢GOES短缺）物理瓶颈，迫使超大型科技公司转向以天然气轮机为主的表后自发电（BTM）。本报告构建了美股与A股跨电力设备、GOES及燃机核心元器件的50/50联合对冲投资组合，以供应链套利和双向免疫逻辑把握AI重工业基建大牛市。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团", "summary": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团 Global AI Compute Grid interconnections Bottlenecks and Cross-Sector BTM Rotation 伴随变压器制造及重型燃气轮机5-7年交付周期的物理瓶颈，超大规模数据…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团", "summary": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团 伴随变压器制造及重型燃气轮机5-7年交付周期的物理瓶颈，超大规模数据中心向表后自发电（BTM/BYOP）的转型是否正重塑全球科技与工业基建的关联度？我们在A股与美股中应如何构建跨电力设备、特种冷轧硅钢（GOES）及燃气元器件的联合对…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美欧原产地规则与中国出口许可在AI电力供应链上呈现70%价格机制、30%数量机制的格局，仅在Hi-B取向硅钢与NdFeB重稀土磁材两节点真正物理改道，其余主要通过推高到岸成本并对美国…", "summary": "美欧'熔炼与浇筑'原产地规则与中国出口许可制是两套高度非对称的工具：前者基于规则、机械执行但被18-30个月执法滞后稀释；后者基于裁量、时延即政策。综合压力测试：2026-2027年这套监管栈是70%的成本推动通胀税（AI电力设备到岸成本+20-35%，美国核心商品CPI+15-25bp）、30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 3, "neutral": 0, "value": 43.6, "events": [ { "title": "重型电力设备OEM收入上能承接需求、时间上承接不了——订单簿2.5–4年可见度+定价权+利润率扩张构成对科技久期的反向对冲，核心超配GEV/ENR.DE/ETN/Hitachi系，中…", "summary": "2026-05-28结论：本卡支持第03卡的电网瓶颈论断。Siemens Energy Grid Tech订单/营收比2.28×、储备490亿欧元；GE Vernova Electrification订单+86%、设备储备390亿美元、B2B约2.5×；Eaton电气储备同比+48%、数据中心订单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核心电力设备的制造产能与超长交付周期是继材料端之后，锁死2026-2028年全球AI数据中心并网扩张的最硬物理约束。", "summary": "本卡对大功率变压器与中高压开关柜制造产能进行压力测试。结论表明，高压测试大厅建设瓶颈、熟练绕线工短缺及二级部件赤字构成了坚不可摧的制造侧'物理墙'，交期拉长至128-160周，2028年前产能已被完全锁定。在极端地缘关税与汇率压力下，中国变压器出口超配组合将面临-48.70%的最大回撤，且对冲摩擦…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 叙事正从芯片竞争迁移到能源供应竞争，而把 2026–2028 钉死的不是发电装机，而是变压器、HV 开关柜与取向硅钢这条狭窄走廊。", "summary": "支持卡01的'物理墙'框架，并把瓶颈具体化到电力设备链：美国大功率变压器交付周期已拉长至120–210周、北美互联排队>2,600GW、取向硅钢(GOES)成为物理瓶颈。GEV/ETN/ENR.DE/Hitachi Energy/POWL 等的订单簿连续6个季度book-to-bill>1.2，订…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 0, "value": 51, "events": [ { "title": "关键设备交付周期拉长至2.5-3年以上，且受限于GOES与铜的全球结构性短缺，电力设备产业链的产能扩张短期内无法匹配AI数据中心爆发式需求，行业物理瓶颈为设备OEM厂商（GEV、VR…", "summary": "截至2026年5月29日，在AI算力与电网升级需求爆发下，电力设备产业链（GEV、VRT）面临严重物理产能瓶颈。2026年变压器交付周期拉长至128-144周。虽然各大厂商积极在越南和美洲扩张产能，但受限于取向电工钢（GOES）和铜的结构性短缺（GOES在2025年存在约30%供需缺口且将持续至2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理设备的极端瓶颈与并网周期的阻碍，使得短期内彻底解决AI电力瓶颈化为泡影，科技巨头被迫面临“碳中和目标”与“AI算力扩张”的直接冲突，导致燃气轮机供应链及独立发电企业成为最大赢家。", "summary": "2026年5月29日压力测试显示，短期内电网并网与清洁能源增量无法缓解AI电力缺口。电网排队均值近5年，变压器交货期达128-144周，取向硅钢短缺；微软三里岛重启需至2027-2028年，小堆商业化要在2030年后。AI爆发正倒逼公用事业公司大规模新建气电、推迟煤电退役，PJM容量价格飙升105…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 1, "neutral": 0, "value": 66.4, "events": [ { "title": "配电侧变压器与开关柜已锁定2026-2028量价齐升超级周期，金盘科技、特变电工、思源电气、中国西电、平高电气是A股最高确定性的物理底座组合。", "summary": "截至2026-05-31，AI数据中心驱动的配电侧设备已进入结构性供应缺口与超级周期：美国大型电力变压器交付周期≈128周(标准)/144周(GSU)，高定制订单达4年；GE Vernova 2026Q1单季为数据中心新增订单$2.4B，超过上一财年全年；MV/LV开关柜lead time普遍14…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "伊顿和施耐德的电力装备产能投产严重阻后于英伟达芯片出货周期，且受制于变压器铁芯原材料取向硅钢（GOES）的物理刚性短缺，成为制约GB200于2026年点亮上线的独立刚性瓶颈。", "summary": "截至2026-05-30，压力测试表明，伊顿与施耐德的产能扩张无法填补因电网延迟导致的GB200部署缺口。两家公司的主要新增产能要到2027年及以后投产，与2026年爆量出货的芯片周期存在严重时钟错配。变压器（交付128周）与开关柜（交付40-60周）积压严重，加之铁芯原材料取向硅钢（GOES）结…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 0, "value": 59.8, "events": [ { "title": "变压器与开关柜瓶颈仍足以支撑ETN/HUBB利润弹性，其中ETN胜在绝对利润池和AI数据中心杠杆，HUBB胜在更纯的公用事业T&D与变电站beta。", "summary": "截至2026-05-31，本卡支持AI电力设备链条：变压器与开关柜交付周期仍处于多季度到多年级别，客户提前锁定工厂槽位，使电气设备龙头拥有价格与积压订单质量。ETN因Electrical Americas规模、数据中心订单和Q1 2026 backlog弹性，适合作为工业主仓；HUBB因Utili…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "AI电力约束不是全球TWh硬上限，而是区域电网可交付能力上限：谁能获得firm节点容量、长交期设备和可回收tariff，谁才能把算力订单转化为可运行园区。", "summary": "截至2026-06-02，本卡压力测试AI数据中心电力约束：全球年度电量尚可管理，但Northern Virginia/PJM等热点的本地可交付容量、变电站、变压器、电缆和峰时备用已成为算力扩张的区域性物理上限。IEA预计数据中心用电从2024年415 TWh升至2030年945 TWh；LBNL…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 3, "neutral": 0, "value": 50.6, "events": [ { "title": "核心判断：CRGO是2026-2028年更硬的物理瓶颈，铜更像价格和配给瓶颈，二者都削弱AI数据中心无摩擦通电假设。", "summary": "截至2026-06-03，本卡压力测试并总体支持前序判断：2026-2028年电力设备扩张的二次瓶颈更可能出现在合格取向硅钢（GOES/CRGO）和精炼铜的可得性、价格与配给上。CRGO供给弹性最低，美国依赖Cleveland-Cliffs Butler Works，印度JSW/JFE 35万吨/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "公用事业通过合同重构与标准化设计，成功隔离了变压器延期与资产搁浅风险，将AI数据中心的电力诉求转化为低风险的电网费率基数扩张，但这对冲物理材料短缺的能力有限，加剧了用能侧的头部垄断。", "summary": "截至2026-06-03，美国公用事业通过长期采购协议、标准化变电站和新型大负荷费率（如信用证和照付不议），成功将AI电力接入风险转嫁给超大规模云厂商，并转化为可赚取ROE回报的电网资本开支。但这未能解决CRGO和铜材料的绝对物理短缺，导致中小开发商被排挤，瓶颈转为资产负债表竞争。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由于变压器产能短缺、电网并网审批排队以及核能直接配建的监管博弈，Anthropic的5GW+电力承诺面临2至4年的物理延期风险，云厂商将承受算力资产闲置折旧压力。", "summary": "截至2026-06-03，Anthropic的5GW+电力承诺正遭遇严重物理瓶颈。全球高压变压器交期延长至3-5年，核心材料取向硅钢本地供应受限于Cleveland-Cliffs独家产能；PJM与ERCOT电网并网等待周期长达3-4年，且核能直接配建面临输电费用分摊的诉讼博弈，导致其算力扩张计划大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 2, "neutral": 0, "value": 58, "events": [ { "title": "工业设备制造业的产能扩张远滞后于AI算力需求的增长，长达3-5年的变压器交货期与取向硅钢短缺构成了AI扩张的硬性物理约束。", "summary": "本报告对AI数据中心关键电力设备供应链进行了压力测试。测试显示大型变压器交货期已达128-160周以上，2026年全美16GW计划产能中近半面临延期。即便电气巨头（伊顿、日立等）大幅扩产，新产能也要到2027-2028年才能释放。受制于取向硅钢（GOES）等上游物料短缺，AI物理基建的“供应墙”确…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Scaling Law 的瓶颈已从芯片配额转向电力设备交付排期，工业产能的刚性约束将迫使全球 AI 竞争进入“存量上电能力”博弈阶段。", "summary": "截至2026-06-07，全球电力设备供应链已进入“结构性稀缺”状态，特高压变压器交付周期飙升至36-60个月，成为AI Scaling Law的绝对物理极限。工业交付与算法迭代的周期错位将ROIC从15%压缩至5%以下。中国凭借12-15个月的交付周期展现出显著的“上电速度”优势，而北美和欧洲受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 2, "neutral": 0, "value": 59.4, "events": [ { "title": "电网基础设施的物理交付周期（3-4年）构成了 AI 算力部署的刚性硬约束，其影响已超过电费成本，成为决定高能效集群落地的第二物理瓶颈。", "summary": "截至 2026-06-07，电网基础设施（尤其是大型电力变压器）的交付周期已拉长至 160 周以上，成为制约 AI 算力落地的核心物理瓶颈。尽管芯片能效提升降低了电费支出，但工业产能利用率高达 98% 且关键材料取向硅钢 (GOES) 持续短缺，导致 11 GW 的美国数据中心容量因硬件缺失而停滞…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策可以软化AI电力设备供应墙，但无法在2026年用贸易壁垒或补贴绕过GOES和大型变压器的物理扩产周期。", "summary": "截至2026-06-07，本卡压力测试上一卡“GOES与变压器构成物理供应墙”的结论。判断是：关税、出口管制和僵硬国产化规则不能绕过冶金、专用设备、测试和工厂建设周期，反而可能加剧2026年稀缺；真正有用的是盟友采购、关税与BABA豁免、规格标准化和过渡库存。产业补贴必要但滞后，Cleveland…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 3, "neutral": 0, "value": 62.8, "events": [ { "title": "AI电力设备超级周期触发上游铜与取向硅钢的结构性紧缺与温和绿色通胀，但头部一体化、出海能力强的电力设备龙头仍可继续扩张毛利率，二三线纯内销厂商承压——分化加剧，而非全面反噬。", "summary": "截至2026-06-09，本卡支持电力设备超级周期主线，但提示上游铜与取向硅钢（GOES）将经历2-3年结构性紧缺：铜2026-2028累计缺口300-500万吨，价格中枢上移至USD10,500-12,000/吨；高端Hi-B GOES全球有效产能仅约250万吨/年，2026-2027价格上涨2…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI电力的“物理壁垒”正将全球投资范式从“虚拟算力（软件/芯片）”推向“物理确定性（电网/基荷/设备）”，建议通过中美区域套利与基建杠铃组合锁定跨周期收益。", "summary": "截至2026-06-09，本报告针对AI电力瓶颈引发的全球区域竞争力分化，提出“物理AI”投资策略。美国凭借节点低电价和核电复兴维持算力核心地位，中国通过电网建设速度和可再生能源盈余转化为设备链和高耗能产业优势，欧洲则面临区域碎片化导致的去工业化压力。组合配置应向“价值+确定性”风格轮动：超配具备…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与高压开关设备的产能扩张因制造周期、取向硅钢及分接开关的硬约束而严重滞后，无法在2027年AI需求高峰前达产，电力交付瓶颈将成为限制AI算力营收变现的核心硬约束。", "summary": "压力测试确认，变压器及高压开关设备扩产无法在2027年前满足AI基建需求。第一梯队大容量变压器交付期拉长至48-60个月，新订单排至2029年后。Hitachi、Siemens等厂扩产基本于2028年后投产，Eaton琼斯维尔厂2027年投产后仍需爬坡。高牌号取向硅钢（GOES）紧缺与有载分接开关…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI带来的电力基建缺口使得变压器与开关柜从“周期性工业品”重估为“高确定性的算力物理瓶颈”，头部企业海外订单与毛利率正迎来双升的戴维斯双击。", "summary": "截至2026-06-09，本报告确认AI电力需求正推动变压器和开关柜进入“超级周期”。由于北美及全球数据中心用电缺口扩大，变压器交期已延长至3-5年，价格较2020年飙升逾70%。中国电力设备企业（如思源电气、金盘科技、中国西电）凭借全产业链优势，海外订单激增，2025年变压器出口总额达646亿元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 4, "neutral": 0, "value": 50.2, "events": [ { "title": "电网设备超长交付周期及原料紧缺导致大型电力项目面临商运推迟与债务违约风险，使AI数据中心容量扩张受到刚性的“瓶颈延迟”制约，佐证了“AI疲劳”本质上是物理并网与供给瓶颈而非需求侧断崖…", "summary": "截至2026-06-11，电网变压器设备供应缺口及原材料（取向硅钢/GOES与铜）紧缺已对AI数据中心并网建设进度构成物理刚性约束。LPT与GSU变压器交付期已拉长至2.5-4年。由于DOE能效新规，美国GOES缺口扩大至19.6万吨/年，进口依赖度攀升至57.6%。即便采用以铝代铜，也会因线圈体…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力建设正从“依赖电网”转向“离网自治”，SOFC与中国电力设备出口链通过解决交付周期错配，成为捕捉AI基建溢价的核心抓手。", "summary": "2026年AI算力扩张的瓶颈已从GPU转向电力并网，使得SOFC与变压器成为核心资产。SOFC通过“时间套利”（交付周期<12个月对比电网并网7-10年）为AI厂商节省数十亿美元机会成本，LCOE在$58-110/MWh区间已具竞争力。中国厂商凭借全球60%的变压器产能及对取向电工钢（GOES）的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-11，所谓的“表后并网”（BTM）并非彻底避开48个月电网排队的银弹，而是将瓶颈结构性转移至厂区变电设备缺口、燃气轮机5-7年超长交期以及管道与空气许可审批，在W…", "summary": "截至2026-06-11，我们压力测试了AI数据中心通过“表后并网”（BTM）规避48个月电网并网排队的方案。研究表明，表后并网只是将物理瓶颈转移。核电并网面临严峻监管壁垒（如Talen/AWS ISA遭否决）且仍需大容量备用变压器；现场气电面临重型燃机5-7年、航改型燃机3-5年的制造交期瓶颈，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "“离网自治”模式面临美欧“安全与电力绑定”和关税/碳税的精准拦截，主权AI项目在放弃中国电力设备以获取AI芯片（交付延迟36-48个月）与使用中国设备（芯片断供）之间面临难以调和的结…", "summary": "本卡对AI“离网自治”模式进行压力测试。美欧正通过“安全与电力绑定”将AI芯片出口与排除中国设备挂钩（如阿联酋G42项目）。同时收紧FEOC与301关税以拦截含中国取向硅钢（GOES）的第三方变压器；欧盟也于2026年实施CBAM并禁用中国逆变器。开发者被迫在“使用中国设备但无芯片”与“脱钩但交付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 3, "neutral": 0, "value": 57.6, "events": [ { "title": "材料供应链的物理硬约束（变压器4年交付期与铜矿缺口）将导致电网扩张速度远滞后于AI需求，从而推迟公用事业公司的收益兑现并压低利润率。", "summary": "2026年6月调研显示，AI电网建设已撞上“物理材料墙”。大型电力变压器交付周期拉长至3-4年，核心瓶颈在于取向硅钢（GOES）的结构性短缺和精炼铜高达60万吨的缺口。AI数据中心耗铜量是传统设施的3-4倍，材料成本通胀和设备交付延迟将显著推后公用事业公司的业绩兑现，挑战了“电力决定论”下的短期超…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建的“脱网化”倾向并非缓解电网压力的良药，而是通过抢夺核心电力设备与稀缺劳动力，并造成公共电网固定成本向剩余用户转移，实质上加剧了传统产业的能源成本负担。", "summary": "截至2026-06-12，AI数据中心试图通过表后直供（BTM）和微电网“逃离”公用电网的努力正遭遇物理与监管的双重瓶颈。研究显示，BTM模式因电力成本分摊不公引发监管反弹（如FERC对Susquehanna协议的否决），导致超大型项目转向“表前”（FTM）模式，加剧了电网容量压力。同时，AI基建…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球大容量电力设备产能制造滞后与原材料（取向硅钢、铜）瓶颈，决定了变压器及高压开关柜交期在2028年前无法收窄，仍将是美国AI与电网基建最刚性的物理约束。", "summary": "截至2026-06-12，我们支持卡片06结论，判断全球大型电力变压器与高压开关柜交期在2028年前无法压缩至18–24个月，仍将维持在48-60+个月及24-36个月高位。西门子能源、日立能源及GE Vernova等巨额扩产面临3-4年建设滞后，且受制于取向硅钢（克利夫兰-克利夫斯 Butler…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 6, "neutral": 0, "value": 41, "events": [ { "title": "科技巨头直接注资电力资产因电网设备供应链瓶颈（变压器交期48-60个月）而演变成“资本陷阱”，导致高达40%的预承诺GPU硬件（$100亿）因并网滞后3年而遭遇65.7%（-$65.…", "summary": "截至2026-06-13，本报告对科技巨头直接注资电力资产的资本转移进行了压力测试。由于高压变压器（LPT）交期延长至48-60个月、取向硅钢（GOES）供应受限以及并网队列长达5年以上，资本转移正退化为“资本陷阱”。定量测试表明，在3年并网延迟与40%电网造价通胀下，一个2 GW的AI数据中心项…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海外AI数据中心在2026H2面临的电力与电网瓶颈远比光模块1.6T切换更早且更具物理刚性，支持对A股TMT板块反弹硬性维持5%的试仓上限，配置上应精选具备海外产能或零部件壁垒的电力…", "summary": "电力电网瓶颈（超高压变压器交期48-60个月、并网排队5-7年）比交期仅3-4个月的1.6T光模块更早且更硬性地限制AI基建。伊顿与维谛订单积压达历史新高。这催生了中国变压器链出口潮，但2026年美国Sec 232衍生品税及墨西哥关税大幅提高了合规门槛。海外瓶颈支持科技5%的反弹试仓上限。组合应超…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施中的电力与电网设备供应瓶颈研究", "summary": "确认：变压器与配电设备缺口已是AI数据中心交付的物理硬约束，非假设风险。2026年美国约12 GW已宣布产能中仅约三分之一在建，约50%–58%管线被实质性延迟；高压变压器交期拉长至4–5年，价格较2019年涨77%–80%。利润率正沿链条向下转移、向卡点集中——形成四层租金地图：第1层GOES材…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关键电力设备与硅钢材料的物理交期瓶颈已取代芯片产能，成为AI算力落地的实质性进度阀门，将平滑的芯片需求周期重塑为台阶式并网周期。", "summary": "截至2026-06-13，大型变压器（LPT）交期达18-36个月，定制高压单元达3-5年，开关设备交期50-80周，主因取向硅钢（GOES）紧缺。美国于2026-04-20通过DPA第303条认定以促进硅钢和变压器本土化。尽管PJM的EIT与ERCOT的Batch Zero推进并网提速，但下游物…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 4, "neutral": 1, "value": 57.4, "events": [ { "title": "AI电力需求与电网扩容瓶颈：评估配电侧容量与变压器产能对算力部署的制约", "summary": "支持卡片04“确定性电力获取是胜负手”的判断并收窄：真正的硬约束在电网端而非发电端，核心是大型电力变压器供应链。变压器交付周期已拉长至3–5年（部分超160周），而AI部署周期不足18个月；GSU需求自2019年+274%、Wood Mackenzie测算美国变压器缺口约30%、约4000万台配电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI配电订单的高毛利和强价格传导能力，能有效对冲特高压业务的铜价挤压，利好具备IDC配电敞口的电力设备龙头。", "summary": "AI算力中心对高标准配电的需求能有效对冲铜价对传统特高压设备的利润侵蚀，但存在显著的结构分化。云厂商为保障昂贵GPU集群按期部署，愿为配电设备支付高毛利并接受铜价传导。A股具备“特高压+AI数据中心配电”双敞口的龙头将展现强利润韧性。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在中美欧贸易壁垒的双重夹击下，高压变压器供应链的稀缺租金将自设备OEM向上游GOES寡头和不受关税限制的本土化产能转移，导致欧美电网扩张面临长达230周的极端交期。", "summary": "针对GOES出口管控与欧美关税的压力测试表明，变压器产业链的稀缺租金正加速向上游硅钢巨头及本土化组装厂转移。中国于2021年已取消GOES出口退税，若进一步加征出口税，将推高全球原材料成本。同时，美国301关税（25%+）与欧盟2026年3月发起的GOES保障措施及本土化要求，使设备OEM两头受困…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施和电网升级为铜铝提供多年需求底座，但成本侵蚀主要发生在电力设备和固定价格工程链条，而不是高毛利AI加速器本体。", "summary": "截至2026-06-13，本卡支持上游资源品主线，但把AI算力需求拆成“数据中心直接用铜铝”和“电网升级间接放大”两层。IEA预计全球数据中心用电从2024年415 TWh增至2030年约945 TWh，按S&P含铜强度折算，2024-2030年负荷增量对应约1.8-2.4 Mt累计铜需求；CRU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 5, "neutral": 0, "value": 49.8, "events": [ { "title": "AI基础设施的扩张瓶颈已从芯片转向物理电力设备，燃机与变压器OEM凭借5年以上的积压订单和极强的定价权成为供应链中最大的赢家。", "summary": "本报告深入分析了2026年AI电力供应链中的物理瓶颈。研究发现，燃气轮机和高压变压器OEM已成为供应链的绝对咽喉，GE Vernova和西门子能源的燃机订单已排至2030年以后，大型变压器交期突破150周。瓶颈核心在于取向电工钢（GOES）的结构性短缺，导致2026年全美近50%拟建数据中心产能面…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "以变压器为核心的电力设备扩产滞后与原材料供给刚性对2027年后的数据中心并网构成了实质性瓶颈。", "summary": "评估表明，变压器等电力设备供应链扩产极其滞后（大型电力变压器交期拉长至2-4年），受特种钢（GOES）和铜的物理供给制约，各厂商新产能直至2027/28年才能释放。这将直接阻碍2027年后的AI数据中心并网（平均并网及设备延迟超5年），验证了物理层瓶颈对算力中心扩张的实质制约。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力缺口的核心不是全球发电总量不足，而是可靠、低碳、可按节点交付的兆瓦数稀缺。", "summary": "截至2026-06-15，本卡支持上一张卡的主线：AI算力不会先遇到全球统一电量硬上限，而会先遇到局部电网、变压器交付和绿电可交付性的节奏约束。IEA预计全球数据中心用电到2030年约945 TWh，美国和中国贡献近80%增量；美国PJM和Northern Virginia已经体现负荷、容量价格和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "单纯关税壁垒无法抹平中外电力设备的巨大套利空间，但美欧的国家安全审查正构筑绝对准入天花板，迫使中国设备商通过产能近岸外包（FDI）和转向BTM市场来维持份额，这将长期稀释其异常高昂的…", "summary": "截至2026-06-14，本卡片对中国特高压与大型变压器设备的“海外溢价”进行压力测试。分析表明，美国25%的301关税与欧洲15-25%的CBAM碳税均可被海外7倍的价格差与极端的交付时间差消化。真正的威胁来自以“国家安全”为由的绝对准入限制（如美国BPS审查与欧盟FSR法案）。这迫使中国设备龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 1, "risks": 14, "neutral": 0, "value": 12, "events": [ { "title": "中国电力装备外需足以覆盖行业层面的供应链重构成本，但超额收益将集中在材料、认证和渠道能力最强的头部企业，而非所有出口商。", "summary": "截至2026-06-16，本卡片压力测试全球变压器超级周期下中国电工钢与电力设备出海的利润传导。结论是外需足以支撑行业继续上行，但不能简单抵消所有地缘贸易壁垒；真正受益者是掌握高牌号GOES、具备高压/特高压认证和海外服务能力，并能在必要市场进行最终装配或服务本地化的企业。美国50%钢铝关税和欧盟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力配电设备产能扩张与供应链瓶颈：AI算力建设的硬约束 分析师：工业制造分析师（industrials analyst） 立场：support（强化 Research note 02…", "summary": "电力配电设备产能扩张与供应链瓶颈：AI算力建设的硬约束 分析师：工业制造分析师（industrials analyst） 立场：support（强化 Research note 02 的\"电力设备瓶颈是AI算力部署真正天花板\"判断） 工作日期：2026 06 15（Asia/Singapore） …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球电力设备与硅钢瓶颈并非物理性短缺，而是关税保护主义和区域电网审批导致的结构性梗阻，超大规模云厂商可通过中国供应链外溢、绕道贸易与全球化选址进行有效化解。", "summary": "本卡反驳了电力设备是全球AI算力部署天花板的观点。报告指出，所谓产能瓶颈主要存在于欧美孤岛市场。中国作为全球电力设备核心（变压器产能占60%，2025年出口增36%），正通过转口贸易和零部件出口为全球泄压。同时，中国取向硅钢产量过剩（337万吨，增14.3%），云厂商也在通过将数据中心迁往中东、东…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力BYOP的时间差与监管回归，迫使Mega-cap AI估值回归‘公用事业+’属性，触发资金向AI物理层的防御性增长轮动。", "summary": "2026年6月15日研究显示，AI 'Risk-On' 结构因电力瓶颈正经历估值重估。随着6/17 Hawkish Skip推升折现率，Mega-cap AI从‘无约束增长’转向‘受监管基础设施’叙事。巨头2026年资本开支预计达7000亿美元，导致自由现金流大幅侵蚀（Amazon FCF转负）。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 7, "neutral": 0, "value": 40.6, "events": [ { "title": "全球变压器和开关柜产能扩张因建厂滞后及取向硅钢（GOES）供应刚性，在2028年前无法缓解AI电力基础设施瓶颈，‘可通电可靠MW’溢价持续高企。", "summary": "截至2026-06-17，主要变压器与开关柜厂商因AI数据中心与电网升级面临历史性订单积存，高压变压器交期拉长至36-60个月。虽然宣布了超百亿美元的产能扩张，但受限于建厂滞后、熟练工短缺及上游取向硅钢（GOES）紧缺，新产能到2028年前无法缓解电力瓶颈，限制了数据中心并网速度。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网容量瓶颈、长达3年的变压器交付周期及硅钢材料短缺，已构成AI数据中心实际通电的决定性物理限制，将电网受限下的综合结算电价拉高94%，推动表后自备发电套利并引发电网监管备用费狙击。", "summary": "本报告对AI数据中心并网瓶颈与电力设备短缺进行压力测试。主电网容量限制与高压变压器（交期达128-160周）及硅钢短缺，构成通电的主要物理屏障。受限下容量电价升至$329.17/MW-day，使100MW数据中心结算电价暴涨94%至$80.24/MWh，推动表后自备直连套利，但面临电网备用费监管打…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2029年电网设备扩产会缓解但难以消除交付瓶颈，关键稀缺性正在从变压器OEM装配上移到材料、组件、测试和高压系统集成。", "summary": "截至2026-06-16，本卡压力测试第09卡：变压器和关键电网设备扩产真实，但2026-2029年仍难完全跑赢AI数据中心、可再生能源并网、电网替换和电气化带来的MW/MVA交付需求。2028年前大部分大型电力变压器、HVDC电缆、开关设备和上游组件仍处紧缺。真正的瓶颈不只是OEM装配，而是GO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力的“快时钟”正与电网基础设施的“慢时钟”发生剧烈碰撞，变压器产能缺口将迫使数据中心建设周期长期化并推高资本开支，形成实质性的硬约束。", "summary": "截至2026年6月16日，变压器与电网设备供应链已成为AI算力扩张的实质性物理瓶颈。大型电力变压器（LPT）交期延长至36-60个月，核心原材料取向电工钢（GOES）供应高度集中且价格较2020年上涨50-80%，导致变压器综合成本上涨60-80%。约30-50%的2026年计划产能面临延迟，建设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 5, "neutral": 0, "value": 54, "events": [ { "title": "西方电力设备巨头的产能释放滞后于AI并网高峰，且由于零部件卡口与合同刚性，中国产能扩张无法有效减轻其高压交付压力。", "summary": "截至2026-06-18，西方巨头（日立能源、西门子能源、GE Vernova）已启动数十亿美元的变压器产能扩张，但核心产能释放集中于2028年以后，与2026-2027年AI数据中心建设高峰存在严重错配。由于地缘壁垒，中国产能无法直接参与美欧大型高压输电变压器建设，且受限于套管短缺、劳动力匮乏和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球电网核心设备及原材料的产能赤字已构成刚性物理准入壁垒，导致产业利润池从算力/晶圆厂资本支出方结构性向重电OEM厂商转移。", "summary": "大功率汽轮机与超高压变压器的严重产能缺口（全球超高压变压器短缺约30%，交期延至2-4年；大型汽轮机交期延至5-7年）和取向硅钢（GOES）等关键材料瓶颈，对算力中心和晶圆厂构成不可逾越的物理准入壁垒。议价权转移令电力设备OEM（如HD Hyundai Electric和GE Vernova）利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力向电力设备的资本开支传导扩大了实体“流动性黑洞”，高企的无风险利率与公用事业“成长化”转型持续挤出长久期资产的配置资金。", "summary": "本报告支持AI基础设施分析师的判断，量化了AI电力需求向能源基建的传导。2026年全球数据中心用电预计达565 TWh，中国“十五五”电网投资将达5万亿元，且全球面临30%的变压器结构性缺口。这种大规模实体CapEx形成巨大的“流动性黑洞”，虹吸市场资金，推高资本成本，对医药等长久期板块估值维持持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "WACC上行30bp会打击AI资产估值倍数，但不会单独瓦解AI基础设施资本开支；防御性集中在有订单、有瓶颈、有现金流的硬科技与电力/电网环节。", "summary": "截至2026-06-17，本卡压力测试第06张卡的硬科技杠铃：WACC上行30bp会压缩长久期估值，但不足以单独击穿hyperscaler的AI数据中心、算力、电力与电网资本开支。8.0%到8.3%的WACC变化仅使5年资产年度资本回收系数上升0.8%、15年资产上升1.8%、30年资产上升2.8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 4, "neutral": 0, "value": 61.4, "events": [ { "title": "AI算力规模化面临的下一重物理极限：电力供应、电网接入与散热约束", "summary": "截至2026-06-18，本卡支持卡片05的交接：2026-2027年电网接入容量与高密度液冷是AI数据中心扩容的实质性硬约束，但约束在于可交付性与时序，而非算力绝对上限。IEA基准下全球数据中心用电2024年约415TWh翻倍至2030年约945TWh；美国年新增负荷5-7GW而电网仅交付2-3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "G2/G3瓶颈没有被效率叙事推翻，反而因变压器、电气设备与并网队列的多年约束而强化近现金电力设备订单资产的相对优势。", "summary": "截至2026-06-19，本卡支持Research note 02：AI电力交付瓶颈是真实供给侧闸门。美国数据中心用电2028年可达325-580 TWh，电网设备采购和并网队列仍是多年周期；美国活跃并网队列2025年底仍超2,060 GW，PJM数据中心区等待36-48个月；欧盟并网等待可达2-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在电网设备、燃气轮机排产和核燃料硬物理瓶颈下，2026-2028年多达40%-50%的规划数据中心负荷将被推迟，导致hyperscaler物理设施资本开支面临10%-20%的折价调整。", "summary": "本卡合成并量化了变压器、燃气轮机和HALEU等物理约束对2026-2028年美国AI数据中心的影响。预计40%-50%的规划容量将面临送电延迟；物理设施CapEx面临打折，其中Microsoft脆弱性最高（折价15%-20%），Google最低（8%-12%）。建议重配电力篮子，加仓电网设备（A）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力交易主线下的天然气现场发电经济性与电网接入瓶颈博弈", "summary": "在题设气价$3.92/MMBtu下，AI数据中心的表后（BTM）天然气发电全成本约$95–115/MWh（自测算），并不比电网交付电（约$70–95/MWh）更便宜；决定性变量是通电时间而非每兆瓦时成本。电网接入端到端超7年、变压器交期超160周，而Cummins往复机组12–24个月即可通电。对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 1, "risks": 5, "neutral": 0, "value": 64.8, "events": [ { "title": "AI数据中心配电侧存在类似电子布的材料升级逻辑，但结构性溢价属于经过热、介电、防火和高电流认证的系统材料栈，而不是所有电力设备大宗零部件。", "summary": "截至2026-06-20，本卡片支持AI算力功率密度向配电侧关键材料升级传导，但认为它不是电子布价格周期的一比一复制。120 kW级AI机柜把问题从普通供电容量推向高电流、热老化、介电可靠性和灭弧认证。结构性受益集中在高牌号GOES、CTC铜线、工程/芳纶绝缘纸、酯类介质液、母线、高密度中压开关及…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI数据中心验证的是持续电网资本开支周期，但投资上只能资本化已签约、可回收、技术受限的设备材料需求，而不能把全部排队MW视为真实订单。", "summary": "截至2026-06-20，本卡片综合前序结论：AI数据中心电力设备需求不是单纯的短期抢装，而是经过排队筛选、容量成本、电价机制和变电站/输电工程落地后的多年电网资本开支周期。但原始接入排队存在高估，必须把已签约、有抵押、有最低需量义务的MW与投机性报装区分开。结构性受益集中在高可靠变压器、中压开关…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "应剔除BE空头并将SMR板块空头权重由100%降至30-40%，把释放的风险预算加配电气核心设备多头——表后电源在变压器瓶颈下已变为被迫采购，叠加高空头利率与高借券费率，做空凸性赌输。", "summary": "变压器交期120-210周与43%AI数据中心延期投产，使表后电源从可选项变为必选项。Bloom Energy（BE）凭借AEP/Oracle/Brookfield合同与12-18个月SOFC交付周期，已从故事股切换为AI电力交付工业方案股，叠加18-22%空头利率与12-28%借券费率，做空风险…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备供应链瓶颈及其对AI基础设施扩能的二次约束效应", "summary": "报告日期2026-06-19。本收官卡压力测试Research note 07“折价而非证伪”结论，引入变压器、高压开关与取向硅钢(GOES)的二次设备瓶颈。变压器交期由128周(25Q2)拉长至超160周/最高4-5年[S2][S1]，GSU需求自2019年+274%、电力变压器供应缺口约30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 2, "neutral": 0, "value": 76.2, "events": [ { "title": "基于2026-06-21压测，铜与铀的供给刚性导致AI电力设备制造商面临严重的成本挤压（毛利率收缩达867-1390 bps），传统核电运营商则凭借稀缺容量捕获高额“电价超额地租”（…", "summary": "基于2026-06-21，本报告压测了铜与铀供应瓶颈对AI电力产业链利润的侵蚀风险。在LME铜价大涨至13,682美元/吨且取向硅钢大涨30%下，中游变压器与线缆制造商若传导率低至10%，其毛利率将暴跌1,390.8 bps；即使有50%传导率，毛利率也收缩867.8 bps。相反，常规核电运营商…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备交期瓶颈与大宗物流通胀引起的制造业利润受阻，共同锁死了2027年后全球AI算力集群的物理部署进度，迫使超大云厂商非计划性降速。", "summary": "本卡支持并完成了十卡研讨。在原油供给冲击与G3利率高企背景下，大型变压器（交期80-210周）与高压开关柜（交期80-120周）交付滞后是限制2027年后算力集群建设的终极瓶颈。定量测试显示，大宗及海运暴涨导致中国纯出口商变压器净利润转负（-3.52%）强制引发交付中断；而本地化厂商受限于取向硅钢…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 7, "neutral": 0, "value": 47.6, "events": [ { "title": "电力设备（如变压器）供应链瓶颈对AI基础设施建设节奏的影响", "summary": "压力测试结论：是，但有两点修正。对美欧并网型AI建设，电力设备2026年当下即是约束边界且早于芯片——约一半美国规划数据中心已延期/取消，12GW中仅约5GW在建[S4][S9]，而CoWoS仍在近4倍扩产出货[Card05]，故‘买得到却通不了电’的部署约束当下即活跃。修正一：约束不是变压器箱体…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美欧贸易壁垒足以压低中国电力设备出海的利润率和估值倍数，但不足以抵消无源、短缺约束型电网硬件的出口红利。", "summary": "截至2026-06-21，本卡否定“美欧关税和贸易限制足以抵消中国电力设备出海红利”的强判断。政策摩擦真实存在：美国Section 301覆盖部分HS 8504及能源相邻品类，欧盟通过NZIA、FSR和高风险逆变器融资限制提高准入门槛。但美国并网队列、变压器80-210周交期、DOE披露的进口依赖…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心对电力配电设备的产能锁定与交期挤压，将使CDMO锁定的纸面‘电网优先级’因物理并网延误2-3年而失效，导致其错失《生物安全法案》过渡窗口并引发核心美资客户流失。", "summary": "截至2026-06-22，本报告对CDMO海外产能电网优先级进行压测。结果表明，行政并网权将因物理设备短缺而失效。AI超算巨头买断了大型变压器（交期达48-60个月，排单至2030/2031年）和高压开关柜，导致CDMO海外物理并网推迟24-36个月。这使投产滞后至2030-2031年，错失《生物…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "评估高功率密度（1200W+ TDP）驱动下的末端配电网（机架级配电）容量压力与变压器升级投资需求", "summary": "压力测试卡08交接论点。支持瓶颈下移至末端的方向，但否定'公共配电变压器扩容匹配AI集群'的表述：大型AI园区以110/220kV直接接入专用变电站，绕过公共低压配电网。真正约束分两层——电网侧高/中压并网（36-48个月交付，约束性闸口）与数据中心内部中/低压及800V HVDC链路（DC运营商…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 1, "neutral": 0, "value": 85, "events": [ { "title": "高密度AI集群扩张对区域电网配电侧基础设施的负荷压力与承载极限预测", "summary": "截至2026-06-22，本卡压力测试并锐化整条链条的“电力墙”论点：是的，配电侧物理时滞是绑定约束，但它比Research note 02-07定位的层次更深一层。液冷+ASIC（Research note 09）抬升的是机房内部功率上限，却无法让变压器造得更快。ASIC密度兑现现受制于GOES…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 1, "risks": 4, "neutral": 1, "value": 79.4, "events": [ { "title": "电网输变电容量与并网审批周期（而非总体发电量）已成为AI算力部署的终极物理刚性瓶颈，并已实质性挤压IT硬件采购预算。", "summary": "本报告对AI算力扩张的电网与能源瓶颈进行压力测试。分析表明，虽然电力设备订单充足，但美国并网队列平均达5.5年，且面临FERC 2026年6月大型负荷监管及弗吉尼亚GS-5等防止成本转嫁政策。变压器等重电面临GOES原材料短缺，Tier-1厂商积压至2029-2030年。表后核电直供监管被封死，燃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中东与东南亚AI基建需求能否承接美欧贸易壁垒带来的电力设备增量？", "summary": "评估表明中东与东南亚AI基建及电网投资完全足以承接美欧贸易壁垒下中国流出的电力设备订单。美欧关税和反规避导致每年约15-20亿美元订单受阻，而沙特、阿联酋和以马来西亚为核心的东南亚AI数据中心增量（至2030年增量8.2 GW）与配套电网升级，每年带来高达51.4亿美元的可触达电力设备需求。虽然这…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在2026-06-23，我们认为高压试验大厅与核心供应链瓶颈限制了全球电网关键设备的扩产节奏，推动具备交期和全产业链优势的中国头部制造厂商出海订单与利润率“双击”，重塑了制造板块估值…", "summary": "本报告于2026-06-23撰写，分析了AI算力与电网升级下的电力设备产能缺口。高压试验大厅与取向硅钢构成物理扩产壁垒，导致美国变压器交期拉长至128-144周。中国企业凭借交付与成本优势迎来出口爆发，重塑了制造板块由“周期”向“出海成长股”的估产双击估值逻辑。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力变压器产能短缺（交付拉长至120-210周）正成为AI数据中心并网的实质瓶颈，其不仅造成数千亿美元的算力租赁及GPU销售损失，还将通过原材料波动剧烈分化变压器厂商毛利率，并使技术…", "summary": "截至2026-06-23，本卡对变压器及二次设备产能与原材料波动进行压力测试。变压器受制于绕线工稀缺、测试大厅滞后及硅钢/铜线瓶颈，交付拉长至120-210周。这导致2027-2028年算力园区并网缺口达43.0%（12.9GW），使GPU出货受阻，芯片商收入损失3096亿美元，CSP租金损失12…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 5, "neutral": 1, "value": 66.8, "events": [ { "title": "评估欧美电网升级需求与贸易壁垒对中国电力设备出海链的传导效应", "summary": "压力测试 Research note 4 命题：A股估值重配是否真建立在可持续高毛利出海订单上。结论为部分支持但有结构性保留。需求拉动真实且被低估——美国公用事业2026-30资本开支近1.3万亿美元 [S1]、超大规模厂商2026资本开支>6750亿美元/+60% [S2]、全球变压器短缺 [S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科技板块风格再校准：能效赋能方的稀缺溢价与高耗电算力股的吞吐折价", "summary": "本卡把前五卡收束为可执行配置框架。算力以‘每兆瓦有效算力’计价后，权益市场沿‘谁拥有稀缺投入、谁的回报被它锁死’重估。折价已现于现金流：2026超大规模厂商资本开支约6900–7250亿美元、约75%投向AI，却吞噬近100%经营性现金流；微软约800亿美元Azure积压受限于电力。溢价已现于盘面…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "我们已成功完成了首席策略师（Chief Strategist）的第一张卡片的研究与文件写入。 核心结论摘要 (Strategic Overview) 在 2026 06 25 的定位…", "summary": "我们已成功完成了首席策略师（Chief Strategist）的第一张卡片的研究与文件写入。 核心结论摘要 (Strategic Overview) 在 2026 06 25 的定位背景下，铜价暴涨至 $13,530/吨 [S1]，取向硅钢（GOES）供应空缺达 30% [S2]，电池级碳酸锂价格…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力资本开支超级周期的电力/电网天花板及其对新一代信息技术全链的约束", "summary": "电力/电网上限真实存在且已是绑定约束：美国并网 3–7 年、变压器 4–5 年、燃气轮机售罄到 2030、30–50% 的 2026 产能滑至 2028 [S5-S8]；微软约 800 亿美元 Azure backlog 受电力而非需求约束 [S9]。但 2026–2028 内它递延而非毁灭需求—…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 3, "neutral": 0, "value": 77.2, "events": [ { "title": "AI数据中心的卡脖子资产已从芯片变为高压电气设备，交期与产能缺口把电气推成最大且持续涨价的Capex板块，稀缺溢价持久归于变压器/特高压/电气配套，而非火电或燃气beta。", "summary": "截至2026-06-25，本卡支持并加固Card-01判断：AI数据中心的约束已从芯片转向电气并网设备。核查结果——大型变压器交期约128周、GSU约144周、中压开关柜52–78周，均超数据中心12–18个月建设周期；GSU需求2019–2025增274%，美国变压器/配电变压器2025供给缺口…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球电力设备制造供应链产能与上游材料供应风险评估", "summary": "部分确认Card-03的time-to-power论点，但作三处修正。瓶颈确在上移到材料端，但对变压器而言真正的物理咽喉是GOES而非铜，且GOES稀缺本质是“产能+贸易”稀缺而非地质稀缺：全球变压器产能利用率98%、GOES自2020年涨约100%、美国仅Cleveland-Cliffs一家、俄…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备原材料缺口对新能源项目建设成本与并网进度的传导效应评估", "summary": "支持卡片06–08的逻辑链：AI电力瓶颈已上移至原材料，且GOES/高纯度铜缺口确实传导至新能源经济性，但传导非对称、主要走时间而非单价。直接价格通道有限——铜仅占光伏/陆上风电资本开支约3%~5%，即便铜较2020年+50%、GOES+100%，对LCOE直接抬升仅约+2%~+4% [自测算]。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 1, "risks": 4, "neutral": 0, "value": 80.6, "events": [ { "title": "随着全球及中国变电设备产能在2027-2029年逐步释放，AI数据中心的电力设备瓶颈将从“无法获取设备”的二进制物理硬约束，降为“支付溢价与关税以获取交期”的资本支出与利润率成本约束。", "summary": "本报告对2027-2029年AI电力设备瓶颈的缓解速度进行了深入分析。结论表明，随着西方重电巨头（如日立能源、GE Vernova）扩产产能释放，以及以特变电工、思源电气、金盘科技为代表的中国高效率设备出口，2027-2029年电力设备将从物理上的“硬约束”转变为“成本/经济约束”。由于取向电工钢…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备杠铃可以落地为六只A股核心篮子，但必须以流动性、两融拥挤和业绩质量仪表盘强制约束，触发信号时用IC/IM/IF与ETF期权对冲并削减设备腿。", "summary": "截至2026-06-27，本卡片支持前序电力设备杠铃，但将其落实为严格的A股执行篮子：2.5% NAV电力设备由首钢股份/望变电气表达GOES稀缺、思源电气/金盘科技表达出口Alpha，2.0% NAV公用事业由长江电力/国投电力压舱。对冲工具以IC/IM/IF和沪深300、上证50、中证500 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器交付超长交期（“铁周期”）与并网延迟构成AI算力扩张的物理瓶颈，A股估值溢价需从 speculative 算力概念转向电力可得性、出海设备及防守型红利电力资产。", "summary": "本卡评估了AI算力扩张下的电力缺口与电网基础设施响应。截至2026-06-27，特高压通道建设错配（光伏6.0倍、风电2.25倍于电网周期）与变压器超长交期（120-210周）构成AI落地的终极物理约束。此外，铜（占35%）和取向硅钢（GOES，占28%）等原材料上涨严重挤压中游毛利，在+20%铜…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 6, "neutral": 0, "value": 62, "events": [ { "title": "功率半导体已过剩，AI电力的新物理瓶颈已迁移至大型变压器，并真正卡在无法被资本快速绕行的上游电磁钢(GOES)与铜——2026–2028窗口难解。", "summary": "支持并收紧前五卡逻辑链：功率半导体转入结构性过剩后（Research note 05），AI电力建设的刚性瓶颈已迁移至大型变压器(LPT)，并下沉到上游电磁钢(GOES)与铜。瓶颈呈两层：整机厂总装层可用约20亿美元北美扩产去瓶颈，但缓解要到2028/2029年后；标准电力变压器交期128周、GS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宽禁带半导体(SiC/GaN)仅能优化数据中心机柜内部能效，无法解决中压电网侧变压器交期超150周及GOES材料短缺的电网接入瓶颈，AI算力扩张的电网硬约束与云厂商资本开支递延依旧确…", "summary": "截至2026年6月28日，本报告评估了通过SiC/GaN电力电子解决AI电力瓶颈的硅端技术路径及高压变压器产能瓶颈。硅端SiC/GaN虽能提升机柜能效，但属于机柜内部技术，中压渗透率到2030年仍<5%，且面临镓出口限制下部署容量缩水11.78%的断供风险。电网端高压变压器交期达150-200周，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备扩产会缓解AI数据中心并网瓶颈，但无法在2026-2027年完全解除大型变压器和输电接入的物理约束，因此算力利润扩散仍是受瓶颈约束的电网资本开支周期。", "summary": "截至2026-06-28，本卡支持上一卡的物理瓶颈判断，但强调扩产的时间结构：变压器、开关柜和电网设备扩产是真实的，Eaton、Hitachi Energy、Siemens Energy等制造商已披露新增产能和高积压订单；但大型电力变压器仍受定制设计、GOES/铜、测试槽位、运输和并网研究约束，新…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "贸易壁垒导致的取向硅钢（GOES）结构性割裂，以及2026-2030年铜矿的百万吨级结构性缺口，奠定了电力变压器成本的向下刚性，而铝代铜的技术陷阱因被动增加25%的硅钢用量而被超大规…", "summary": "基于2026-06-27权威工作日，本报告支持AI电力并网瓶颈已下沉至原材料端的判定。由于贸易壁垒隔离，中国占56%的全球取向电工钢（GOES）产能无法被美欧替代。同时，电网扩容拉动铜需求，2026年铜矿缺口达33万吨，并于2030年扩大至150万至220万吨。这导致变压器制造成本刚性上涨。试图通…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 12, "neutral": 0, "value": 27.4, "events": [ { "title": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "summary": "本卡支持 Research note 02「上游原材料是最深约束」的结论，并锐化为一组非对称结构。GOES：自2019年变压器需求暴增（电力+119%、GSU+274%），远超GOES市场约5.4–5.6%的CAGR；GSU供给缺口2025年约100%，因美国唯一GOES生产商Cleveland-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年全球AI数据中心电网瓶颈与“自主供电”战略的崛起", "summary": "2026年全球超大规模厂商的AI算力基建受限于电网瓶颈与关键电力设备（变压器交期达5年）短缺。这迫使厂商将部分资本支出（预计2026年总支出达7250亿美元）从纯算力（GPU）转向自主供电（BYOP）和表后（BTM）能源方案。预计该转移将延迟算力上线速度，并推高长期持有成本。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "summary": "本卡压力测试逻辑链：GOES约束不会「抵消」核能加速，因为二者作用于电力链不同节点——核能重启在发电/并网侧本就是GOES-轻（复用既有GSU与开关站），属于一种GOES套利；但GOES在负荷侧与输配路径重新绑定，沿用Research note 01的NPV模型测算，核能锚定型项目(≈1.5年延迟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力需求增长对电网侧扩容及配电资产的压力测试", "summary": "压力测试结论：前序主线被证实并强化。2026年AI落地的约束性瓶颈已非GPU，而是电网边缘——电力变压器、中高压开关设备、取向硅钢（GOES）与接网排队。大型变压器交期3–5年、全球产能利用率约98%、美国接网积压1,500–2,300GW、枢纽接入排队4–7年，致美国2026约12GW管线仅约4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 2, "neutral": 0, "value": 88.8, "events": [ { "title": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "summary": "本卡片从制造和供应链视角对电网与电力约束进行压力测试。截至2026-06-29，关键电力设备（变压器与重型燃机）由于极高工艺与测试门槛、上游取向电工钢（GOES）瓶颈及寡头垄断，交期拉长至3-7年，生产机位已售罄至2029-2030年。云厂商BTM发电加剧了燃机机位争夺。预计2026-2027年该…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "上游取向硅钢和电工铜的产能稀缺与关税高墙正转嫁为严重成本通胀，侵蚀设备制造厂毛利并拉长交付，导致AI电力落地面临实质性溢价风险。", "summary": "压力测试表明，取向硅钢与电工铜的供应缺口已完全固化在变压器与燃机的长交期中，不构成独立的“二次时间瓶颈”。但地缘摩擦与美国50%的Section 232关税，导致中美GOES价差拉大至5.16倍，美国面临严重的材料合规溢价。在重度成本冲击下，固定价格合同变压器厂毛利率收缩至-3.67%，逼迫出口链…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-e1d3ec2fed/", "networkHref": "/investor/research-network/#thesis-e1d3ec2fed" }, { "id": "thesis-33e46fee64", "source": "auto_topic", "title": "2026 HBM 缺口真实但租金窗口前置在 1H26，真正的稀缺量子是 CoWoS 先进封装而非 HBM 裸晶——杠铃方向成立但工具应以…", "titleEn": "Practical research topic: 2026 HBM 缺口真实但租金窗口前置在 1H26，真正的稀缺量子是 CoWoS 先进封装而非 HBM 裸晶——杠铃方向成立但工具应以…", "lane": "AI 基础设施", "laneEn": "AI 基础设施", "laneHref": "/investor/lanes/ai/", "investableFocus": "", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "AMD", "name": "AMD", "market": "US", "role": "加速器替代供给", "roleEn": "加速器替代供给", "bucket": "core", "href": "/investor/stocks/AMD/" }, { "ticker": "QQQ", "name": "Invesco QQQ Trust", "market": "US ETF", "role": "纳斯达克成长暴露", "roleEn": "纳斯达克成长暴露", "bucket": "watch", "href": "/investor/stocks/QQQ/" } ], "stockTickers": [ "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ", "APO", "IEF", "AMD", "QQQ" ], "coreView": "该主线来自 4 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-13 至 2026-06-12；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-04-23", "lastSeen": "2026-06-29", "supports": 10, "risks": 138, "neutral": 12, "analystCount": 13, "sourceTypeCounts": { "研究线程": 5, "研究笔记": 152, "证据更新": 2, "分析协作": 1 }, "topTerms": [ "NVIDIA", "CoWoS", "先进封装", "HBM4", "HBM", "缺口真实但租金窗口前置在", "真正的稀缺量子是", "先进封装而非", "裸晶", "杠铃方向成立但工具应以", "Beta", "成熟制程产能" ], "whatChanged": [ "2026-06-29: 国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM故事退守至买设备与材料国产化的刚性Capex以及全球化CPO。", "2026-06-29: 国产HBM与先进封装不是零进展，但2026年的主线应买“工艺环节国产化”，而不是把AI芯片设计公司当作不受HBM约束的无限TAM。", "2026-06-29: 先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析", "2026-06-29: AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "2026-06-28: 半导体产业链的供给侧约束与国产化替代的技术边界" ], "practical": { "sectorLane": "AI 基础设施", "sectorLaneEn": "AI 基础设施", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "AMD", "name": "AMD", "market": "US", "role": "加速器替代供给", "roleEn": "加速器替代供给", "bucket": "core", "href": "/investor/stocks/AMD/" }, { "ticker": "QQQ", "name": "Invesco QQQ Trust", "market": "US ETF", "role": "纳斯达克成长暴露", "roleEn": "纳斯达克成长暴露", "bucket": "watch", "href": "/investor/stocks/QQQ/" } ], "stockTickers": [ "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ", "APO", "IEF", "AMD", "QQQ" ], "primaryStocks": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" } ], "instrumentTags": [ "NVDA", "NVIDIA", "TSM", "TSMC", "AVGO", "Broadcom", "MU", "Micron", "ANET", "Arista Networks", "VRT", "Vertiv" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 94 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 13, "recentEvents": 94, "supportRatio": 0.06, "riskRatio": 0.86, "conflictScore": 96.4, "evidenceDepth": 100 }, "score": 67.4 }, "events": [ { "date": "2026-06-29", "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,800万美元（约降33%）[S1][自测算]；先进封装（CoWo…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM4", "HBM", "HBM3" ], "href": "" }, { "date": "2026-06-29", "title": "先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析", "summary": "截至2026-06-30，本卡片压力测试Research note 08的能效转型论。结论：CoWoS原始产能已非约束瓶颈——TrendForce判断供需缺口随产能扩张从约20%收窄至2026年底约10%[S1]，台积电产能翻四倍至约13万片/月并外包约24-27万片/年给Amkor/SPIL[S3][S4]，疏解时钟约12-24个月，远快于电网3-7年。…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM4", "HBM" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM与先进封装不是零进展，但2026年的主线应买“工艺环节国产化”，而不是把AI芯片设计公司当作不受HBM约束的无限TAM。", "summary": "截至2026-06-30，本卡支持前序“半导体国产替代卫星腿”判断，但把结论收窄到设备、材料、基板与OSAT工艺赋能。国产链在清洗、电镀、CMP、部分刻蚀/沉积和先进封装材料上已具备中等以上成熟度；HBM DRAM本体、HBM3E/HBM4高良率堆栈、混合键合量产、先进光刻与高端ABF/FC-BGA基板仍是硬瓶颈。SemiAnalysis估算CXMT 20…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22.5, "matchedTerms": [ "先进封装", "HBM4", "HBM", "HBM3", "国产" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22.5, "matchedTerms": [ "先进封装", "HBM", "GPU", "HBM3", "国产" ], "href": "" }, { "date": "2026-06-28", "title": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "summary": "电力与电网容量成为稀缺投入后，AI 芯片优化目标从绝对算力转向每瓦性能（perf/W）与单机柜 token 经济性。NVIDIA 宣称 Blackwell 推理能耗较 Hopper 最高低 25×（FP4/NVFP4 驱动），但 B200 单卡功耗仍升至约 1,200 W（H200 为 700 W），机柜功率密度升至 120–142 kW 并规划 1 MW…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 32.25, "matchedTerms": [ "NVIDIA", "CoWoS", "先进封装", "HBM4", "HBM", "国产" ], "href": "" }, { "date": "2026-06-28", "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM合约价Q2环比+58–63%)、以及电网电力(美国并网排队约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM", "GPU" ], "href": "" }, { "date": "2026-06-28", "title": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "summary": "截至2026-06-28，本卡否定“AI资本开支韧性等于A股防御性配置锚”的宽泛结论。Microsoft、Alphabet、Meta、NVIDIA、TSMC等数据仍支持AI基础设施真实需求，但FOMC 3.50%-3.75%、5月PCE同比4.1%、10年美债4.40%和A股两融突破RMB 3tn，意味着长久期成长股、拥挤主题和融资敏感供应商仍可能估值下修…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 23, "matchedTerms": [ "NVIDIA", "先进封装", "HBM", "Beta", "权益" ], "href": "" }, { "date": "2026-06-28", "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗ASIC。先进封装成为泄压阀，但TSMC已提示先进封装产能紧张…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM", "Beta", "国产" ], "href": "" }, { "date": "2026-06-28", "title": "半导体国产替代对AI算力瓶颈的缓解潜力", "summary": "支持但有限定：国产半导体设备与先进封装确实有效缩短AI算力兑现周期的硅片侧环节。设备国产化率2024约26%→2025约35%，刻蚀/薄膜约40%本土，北方华创(002371.SZ)全球第五、中微(688012.SH)前十三，大基金三期3440亿托底；先进封装为弯道超车路径，长电(600584.SH)XDFOI 4nm量产、良率约98.5%，国产CoWoS…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM", "国产" ], "href": "" }, { "date": "2026-06-28", "title": "半导体产业链的供给侧约束与国产化替代的技术边界", "summary": "本卡对H20/HBM管制下中国国产替代的技术边界做压力测试，部分否定「难以逾越的断层」框架。结论分三层：先进封装（CoWoS替代）是中国相对优势、是可在1–2年跨越的桥（JCET/通富/盛合晶微＋Naura混合键合），并非断层；真正断层在上游——EUV光刻国产化<5%（SMEE约90nm，落后ASML约十年）与高端ArF/EUV光刻胶约95%依赖日本；而专…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM", "国产" ], "href": "" }, { "date": "2026-06-27", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交期[S6][S7]限制的是通电而非下单。CoWoS售罄、交期…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM", "GPU", "HBM3", "国产" ], "href": "" }, { "date": "2026-06-27", "title": "在2026-06-27的权威工作日背景下，国产HBM良率骤降逼迫国内算力网络化，而全球封装与电力阻尼拉长了云厂商CapEx变现周期，这使得制程免疫且订单排至2028年的可插拔光模块出口龙头成…", "summary": "截至2026-06-27，本报告评估了HBM良率危机与先进封装瓶颈对算力集群技术演进的影响。国产HBM3封装良率降至2.66%，倒逼国内算力集群转向网络密集的Scale-out拓扑，推高了光模块需求弹性。全球云厂商2026年CapEx直逼7250亿美元，但受制于CoWoS晶圆先进封装冲突与吉瓦级电网接入瓶颈，面临6-12个月的算力交付与收入确认错配。CPO…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM", "HBM3", "国产" ], "href": "" }, { "date": "2026-06-27", "title": "先进封装与背面供电提升的是单位功耗算力而非降低绝对功率，无法在电网层面抵消GB200级芯片3–5倍/代的功率增长，只能改善约束之内的每兆瓦经济性。", "summary": "截至2026-06-28，本卡压力测试「技术拯救电网」的隐含期待，结论是否定的：CoWoS与BSPD改善的是每瓦性能（比率），而电网瓶颈约束的是绝对兆瓦（水平量）。单GPU TDP走H100 700W→B200 1,000W→GB300 1,400W→Rubin 1,800W→Rubin Ultra 3,600W（约四年5倍），机柜功率从GB200 NVL…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22, "matchedTerms": [ "CoWoS", "先进封装", "GPU", "Rubin" ], "href": "" }, { "date": "2026-06-27", "title": "在2026-06-27，AI电力并网限制（2年延误拖累53.71亿美元[S12]）将倒逼算力采购削减并重创高端半导体高增长假设，投资者应果断削减AI硬件至12%及长债至5%，转而增配25%短…", "summary": "截至2026-06-27，本报告重新校准了在高利率平台与电力之墙挤压下的多资产配置。2年电网并网延误对10万卡Blackwell智算集群将产生53.71亿美元搁浅资金损失[S12]，迫使云巨头在2027年削减20%GPU订单。这将导致台积电3nm和2nm产能利用率分别跌至85%与80%，CoWoS封装降至82%[S12]，HBM面临34.3%产能过剩且合约…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "CoWoS", "HBM", "GPU", "权益" ], "href": "" }, { "date": "2026-06-25", "title": "电力稀缺不会杀死AI芯片周期，但会把竞争重心从单纯峰值算力转向固定兆瓦内的有效吞吐和能效，从而收窄可投资赢家。", "summary": "截至2026-06-25，本卡对电力和并网约束做半导体侧压力测试：约束不会终结AI芯片需求，但会把胜负指标从峰值FLOPS推向每兆瓦吞吐、每机柜token、HBM每瓦带宽和部署可用性。NVIDIA的机柜级平台、Google TPU、AWS Trainium、TSMC背面供电和Micron HBM4都显示产业已转向能效优化。结论是保留AI存储、高端互联、先进…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27, "matchedTerms": [ "NVIDIA", "先进封装", "HBM4", "HBM", "Beta" ], "href": "" }, { "date": "2026-06-25", "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元在手订单）；第二档择优超配高端光互联（1.6T之年、龙头锁单…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "CoWoS", "先进封装", "HBM", "Beta", "GPU" ], "href": "" } ], "timeline": [ { "date": "2026-04-23", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "2026年全球AI芯片供应链的HBM与CoWoS先进封装双重瓶颈", "summary": "2026年全球AI芯片供应链的HBM与CoWoS先进封装双重瓶颈 HBM3E/HBM4产能与台积电CoWoS-L先进封装工位在2026年将如何重新分配,并如何重新定价上游存储(海力士、三星、美光)与封装产业链的估值中枢?", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-24", "supports": 0, "risks": 1, "neutral": 1, "value": 40.4, "events": [ { "title": "2026年2.5D/3D先进封装（CoWoS）产能扩张与HBM4演进：评估次世代AI加速器供应链的稳定性", "summary": "2026年2.5D/3D先进封装（CoWoS）产能扩张与HBM4演进：评估次世代AI加速器供应链的稳定性 随着TSMC CoWoS产能持续紧张且HBM4于2026年开始量产，从HBM3e向HBM4的跨越将如何影响主要AI GPU厂商（如NVIDIA/AMD）的成本结构和交付周期？哪些二级OSAT厂…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026年AI算力供应链瓶颈迁移：从CoWoS封装紧缺到HBM4配额与电力约束", "summary": "2026年AI算力供应链瓶颈迁移：从CoWoS封装紧缺到HBM4配额与电力约束 在SK Hynix已卖空2026年HBM产能、TSMC把CoWoS从约35K扩至130K wpm、Nvidia Vera Rubin下半年放量的背景下，AI硬件供应链的真瓶颈究竟已迁移到哪里？二级市场应如何重新定价上游…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-25", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "2026年超大规模云厂商AI资本开支冲刺与电力—先进封装双瓶颈", "summary": "2026年超大规模云厂商AI资本开支冲刺与电力—先进封装双瓶颈 在Q1 2026业绩季释放的约7,000亿美元四大超大规模云厂商资本开支指引下，约束是否已从GPU供给切换至电力与CoWoS/HBM封装产能？这对TMT板块2026下半年的α定价意味着什么？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-04-26", "supports": 0, "risks": 0, "neutral": 1, "value": 49.2, "events": [ { "title": "2026年AI芯片供应链与代理化AI硬件需求分析", "summary": "2026年AI芯片供应链与代理化AI硬件需求分析 随着2026年AI应用进入“代理化”阶段，推理算力需求激增，HBM4与先进封装（CoWoS）的产能缺口如何影响全球AI芯片的供应格局？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 3, "neutral": 0, "value": 29.6, "events": [ { "title": "推理主导范式将 AI 加速器平台的经济价值分配从 FLOP 算力转向 HBM 容量/带宽与定制推理芯片，MU 的 HBM4 先发量产和 AVGO 的定制 XPU 业务在该范式下具有最…", "summary": "在 HBM bit 需求增速约为加速器台数增速 1.5-2.0x、推理占新增加速器小时 55-65% 的基准下，本卡给出以下排序：MU（第 1，HBM4 先发量产、份额增速最快、重估潜力最高）、AVGO（第 2，定制推理 ASIC $60-90B FY2027 目标市场加 AI 网络）、NVDA（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力行情是超级周期中段，三波叠加需求（训练→推理→主权AI）驱动力完整，供给瓶颈维持定价权，短期情绪风险真实但不构成周期终结信号。", "summary": "AI硬件行情处于超级周期中段，而非利好出尽的顶部。三波需求（前沿模型训练、大规模推理、主权AI/企业部署）正叠加复利，供给侧CoWoS封装、HBM4产能及电力基础设施瓶颈在12–18个月内无法解除，定价权保持完整。历史类比指向1997–1998年的互联网基础设施阶段。战术层面，情绪高涨（恐惧贪婪指…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 年 HBM4 推理链条的首要弹性在合格 HBM4，首要出货瓶颈在 HBM4 验证与 CoWoS/interposer/封装基板，而液冷更多是部署瓶颈。", "summary": "本卡支持前序推理需求上行逻辑：288 GB HBM4 GPU 普及后，最大量价弹性在合格 HBM4 供应商，尤其是具备份额提升弹性的 MU 和具备配额权的 000660.KS；但最可能成为 2027 物理出货瓶颈的是 HBM4 验证与 CoWoS/interposer/封装基板的组合。测试设备是高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 1, "neutral": 1, "value": 54, "events": [ { "title": "2026年5月，Rubin/MI455X的HBM4堆叠所用CoWoS-L先进封装环节几乎100%位于台湾。", "summary": "2026年5月，Rubin/MI455X的HBM4堆叠所用CoWoS-L先进封装环节几乎100%位于台湾。", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯…", "summary": "AI驱动的WFE资本开支周期：先进封装与HBM设备订单可见性如何重塑半导体设备股的盈利久期 第4张卡片 / 共10张 | 工业制造分析师 | 立场：支持 会话主题： 2026年AI芯片竞争格局：NVIDIA Rubin对阵云巨头ASIC 本卡主题： WFE资本开支周期——先进封装与HBM设备订单可…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2027 年的风险并非‘芯片抛售’，而是 Tier-1 的利润预警与 NVIDIA 汽车业务软件估值回归硬件逻辑的错配压力。", "summary": "2027 年 Thor 芯片合同（22%）与实际 L3+ 装载（12-15%）的错配将导致非对称财务传导。NVIDIA 受 NCNR 合同保护，硬件收入稳健，但与车企的软件分成（50/50）将面临 30-40% 的递延风险，导致汽车业务估值从 SaaS 逻辑回归硬件。Tier-1 供应商面临严重的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "先进封装材料的瓶颈已从‘量’转变为‘规模化良率’，Rubin级别的超大封装与HBM4混合键合将利润高度锁死在Ibiden、TOK和Fujifilm等少数具备高技术壁垒的头部供应商手中。", "summary": "2026年AI芯片（如NVIDIA Rubin）向4x光罩尺寸及HBM4（16层+混合键合）转型，使ABF载板、CMP抛光液和特种化学品成为核心瓶颈。单个Rubin载板价值量是A100的3.5倍，Ibiden等巨头产能已被预订至2027年。混合键合使CMP步骤翻倍，支撑了Fujifilm等供应商的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-09", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "英伟达Blackwell/Blackwell-Ultra（B300）2026年全年分配量已基本售罄，边际紧缺转向HBM4产能与800VDC电源供给。", "summary": "英伟达Blackwell/Blackwell-Ultra（B300）2026年全年分配量已基本售罄，边际紧缺转向HBM4产能与800VDC电源供给。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 1, "risks": 1, "neutral": 0, "value": 55.2, "events": [ { "title": "成熟制程产能 Beta 的价格清算已结构化，AI 先进封装是其物理隔离的反面，两向量合并强化而非削弱 Research note 02 的'做空产能 Beta、做多认证 Alpha'…", "summary": "截至 2026-05-13，从成熟制程产能扩张与 AI 先进封装两个相反向量进一步论证 Research note 02 的'做空产能 Beta、做多认证 Alpha'判断。向量 A：大陆成熟节点（28/40/55/90nm）在 2023–2026Q1 累计新增约 750–820 kwpm 产能，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI CapEx的主线不是收缩，而是从新增外壳数量转向每瓦、每机架、每封装的半导体密度提升。", "summary": "截至2026-05-13，本卡支持卡片01的“密度胜过足迹”判断：数据中心外壳新开工-19% QoQ并不等于AI CapEx下行，而是资本更多流向GPU/ASIC、HBM、CoWoS/先进封装、机架级互连、光DSP和液冷配套。Microsoft、Meta、Alphabet、Amazon的最新披露均…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 2, "neutral": 0, "value": 42.6, "events": [ { "title": "算力期货压缩的是商用GPU层租金，先进封装与高带宽互联的物理瓶颈与寡头结构足以将其非商品化毛利结构维持至2027。", "summary": "支持card-01框架并加以收紧：算力期货标准化的是GPU-小时这一输出单位，并未商品化决定谁能以最低成本清算的物理输入。CoWoS先进封装、HBM4、混合键合设备、scale-up交换机ASIC、PCIe/CXL retimer、1.6T/CPO光模块均具18–36个月结构性导入周期，寡头格局清…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "出口管制不是自动终结算力硬件周期的单一变量，但若下一轮规则同时打到HBM、先进封装、设备服务/备件和第三国云访问，它会比流动性收紧更快主导盈利下修。", "summary": "截至2026-05-15，出口管制是比PBoC流动性更高凸性的下行变量，但尚非基准情景下打断AI算力硬件盈利周期的决定性因素。现有美国制度已覆盖先进设备、HBM、软件工具、FDP规则、第三国转运和AI模型训练用途；NVIDIA H20案例证明管制可直接造成$4.5bn费用、$2.5bn未出货收入和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 5, "neutral": 0, "value": 26, "events": [ { "title": "M2M流量爆发的边际成本不会均匀挤压SaaS——它会让薄壳代理编排厂商利润坍缩、让有自有数据的垂直SaaS继续吃到按量计费红利，并把被争夺的利润集中流向芯片、封装与光模块基础设施层。", "summary": "截至2026-05-16，本卡对card-07的SaaS基本面Beta判断做压力测试。结论是有条件成立但显著分化：薄壳代理编排SaaS将因无法转嫁的token成本，在未来12–18个月承受600–1,200 bp毛利压缩；而垂直SaaS、数据合规和API适配器（card-06超配名单）边际成本以存…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GXP向CSA的制度松绑只是把交付瓶颈下移到芯片配额与高压电力装备交期，物理瓶颈已替代制度瓶颈成为AIDC交付的新主导变量。", "summary": "CSA转型确实把特区内AIDC交付压缩到9–12个月，但约束随即从制度排队迁移到物理输入：>500 MVA变压器交期110–140周、CoWoS-L先进封装2026年75–80%被NVDA/AMD/Google预订、HBM3E为价格弹性最低的卡点。'交付悖论'未消失，只是从监管时间转为BOM时间。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 硬件链应买真实瓶颈而非泛映射：台韩核心持有但不追，A股/HK 侧优先封装、测试、设备和材料，回避已把完美 5-6 月收入斜率资本化的纯 AI 算力概念和成熟制程代工。", "summary": "截至 2026-05-15，我综合支持前两张卡：AI 需求未被证伪，台积电 Q2 低端至中枢可兑现，但高端受 CoWoS/HBM 与客户平台节奏约束。台韩主链基本面最清晰，却已大体定价顺利执行；A股/HK 侧分化更大，国产 AI 算力和大涨后的中资晶圆代工已较充分定价。风险收益仍错配在先进封装、测…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股半导体重估的 beta 已基本被定价，下一段必须靠后道设备、量检、材料、车规功率与先进封装上游材料的盈利与订单兑现，回避换手/融资/解禁三重拥挤的设计、代工与映射封装。", "summary": "截至 2026-05-15，A股半导体经过 4-5 月重估后，国产 GPU/ASIC 概念、HBM/先进封装映射股、大市值晶圆代工四条线已拥挤：5 日换手率分位 81-92%，半导体两融占自由流通市值比 ~4%（88 分位），主要芯片 ETF 4 周连续净申购 +18%，且 6-8 月解禁集中（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 3, "neutral": 0, "value": 39.4, "events": [ { "title": "H200对华许可的毒丸条款使其对中国云厂只能扮演过渡角色，对跨境AI硬件营收偏空、对国产替代链（SMIC/昇腾/寒武纪）结构性偏多，板块定价应反映'强制分化'而非'复苏催化'。", "summary": "H200对华许可叠加25%关税与BIS领土审计构成'毒丸'结构：单芯片承诺、现场审计、芯片级遥测、再出口禁令使H200对中国云厂成为高摩擦、不可转售、用途受限的过渡载体而非战略基石。我们下修Nvidia 2026年中国数据中心营收至80-100亿美元（vs.一致预期120-140亿），毛利率拖累5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 硬件盈利端脱钩成立，但价格路径脱钩不完整；基本面能缓冲两融与解禁冲击，不能消除冲击。", "summary": "截至2026-05-17，本卡压力测试 Research note 06：AI 硬件基本面强到足以形成先进封装、HBM 供应链、光模块/AI datacom 的窄口径例外，但不足以让广义半导体 beta 脱离 A 股两融拥挤和解禁供给。NVIDIA FY2026 收入 USD 215.938bn、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产先进封装材料能在2026年瓶颈窗口前撑住2.5D等效吞吐，但TSV化学品缺口将把HBM3E等效堆栈推至2027年中——多材料/CMP基本盘，HBM封测是2027而非2026的交易。", "summary": "【2026-05-17】支持卡片04超配上游材料/设备的方向判断，但加上更精确边界：到2026年Q4，国产化将在ABF载板、TCB键合机、EMC/底填三大卡点上实现可用级，但TSV级Cu/W CMP抛光液与高纯度（>6N）电镀化学品仍是阻止HBM3E等效堆叠的硬约束。2.5D CoWoS等效封装吞…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 1, "risks": 3, "neutral": 0, "value": 48.8, "events": [ { "title": "AI硬件与算力基础设施具备足够的选择性资本确定性，但应买入瓶颈环节和已验证国产替代，而不是泛AI主题贝塔。", "summary": "截至2026-05-18，本卡片支持从大金融防御性收缩转向盈利可见成长。AI硬件具备选择性资本确定性：NVIDIA、TSMC、SK hynix及800G/1.6T光模块已体现高毛利兑现，云厂商2026资本开支提供需求锚。国产化替代真实存在，但确定性集中在服务器、光模块、热管理、电源、PCB和已商业…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HBM 与先进封装的订单能见度（至 2027 年）与强劲的毛利扩张确证了 AI 电子的超强景气度，从基本面切断了年内向海洋经济板块的流动性溢出可能。", "summary": "截至 2026-05-17，TMT 基本面调研确证了 AI 电子板块的高景气度。HBM3e/HBM4 与 CoWoS 先进封装的订单能见度已延伸至 2027 年，且正处于绝对定价权驱动的毛利率全面扩张期。这一坚实的业绩上修闭环为资金的极度拥挤提供了合理支撑，彻底证伪了年内资金将大规模溢出至海洋经济…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Nvidia 直供链上的 A 股份额东移是真实但窄的（5–8 家硬名单），归因是合同制造而非美方约束；CPO 光引擎核心仍在美方，真正的份额弹性在独立订单簿的国产替代层。", "summary": "在 GB300 NVL72 与 Rubin CPO 架构下，A 股供应链份额东移的真实增量集中在 800G/1.6T 光模块 OEM（中际旭创、新易盛、天孚）、液冷 CDU 与冷板（英维克、高澜、工业富联整柜）、超低损耗高速 PCB（沪电、胜宏）三个环节，硬名单仅 5–8 家，远小于市场涨停所定价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "不是 AI 周期地理脱钩，而是同一本 Nvidia 订单簿的两层叠加溢价——供应链份额东移 + 美股看不见的国产替代——A 股相对溢价仍可走阔，但牛市命题必须重新拼篮。", "summary": "2026-05-13 QCOM −11%、美光英特尔回调与 A 股工业富联涨停、CPO 三剑客放量并非 AI 周期的地理脱钩，而是同一本 Nvidia 订单簿被两边市场以不同节奏定价，叠加两层失真：美方下跌集中在 AI-相邻（手机基带、x86 服务器、HBM 价格正常化）而非 AI-纯正；A 股涨…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 4, "neutral": 0, "value": 36.2, "events": [ { "title": "半导体供应链的先进封装、HBM、EUV、基板和光模块交期同步拉长，与电力侧'承诺 Capex > 现金 Capex'呈现同一指纹，证实 2H26 AI Capex 下修是交付推后引致…", "summary": "用先进封装、HBM、WFE、ABF 基板与光模块交期数据压力测试卡片 01，结论是硅侧呈现与电力侧完全一致的'承诺 > 交付、交期拉长、现金落地右移、订单簿创新高'指纹：CoWoS 2026 配额售罄至 1Q27、ASML EUV 交付排到 2H27、HBM4 在 Rubin 上认证再顺延一个季度…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS/HBM是配给型而非崩塌型瓶颈，使perf/W曲线弯曲而非折断，并把AI加速器BOM定价权向HBM厂与OSAT倾斜。", "summary": "截至2026-05-18，CoWoS/HBM是继电力之后的下一个物理瓶颈，但属于配给型而非崩塌型。TSMC CoWoS 2026年底约135–150kwpm、2027年底约190–220kwpm，CoWoS-L占比升至55–60%；行业HBM 2027比特供给同比+55–65%，但加速器需求+70…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产AI芯片链是半个第二风险区——估值溢价已price-in国产替代叙事，但融资盘与机构拥挤度未到极值且对海外CapEx Beta显著低于光模块，可承接卡片06中至多60%的轮动，剩…", "summary": "截至2026-05-19，国产AI芯片链（寒武纪、海光、中芯、CoWoS国产替代、先进封装）是'半个第二风险区'。估值层面寒武纪/海光/中芯2026E PE分别约190x/110x/65x，相对全球同行存在3–7倍政策溢价；半导体/AI芯片融资余额约3,180亿元（11.1%占比，88%分位），公…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027-2028年电力缺口不会终结AI芯片周期，但会把下一代AI硬件迭代逻辑从单芯片峰值性能改写为受限MW下的有效AI输出、上线时间和机架级能效。", "summary": "截至2026-05-18，本卡支持前序电力瓶颈主线，但压力测试“硬件从极致性能转向能效”的表述：更准确地说，采购函数会转向每已通电MW的有效AI输出。2027-2028年表后燃气仅4-6GW，无法覆盖AI负荷缺口；DOE/LBNL预计美国数据中心用电2028年达325-580TWh。NVIDIA …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 4, "neutral": 0, "value": 37.6, "events": [ { "title": "先进封装与HBM出口管制把存储成本压力的核心AI基础设施层从周期性紧张转为政策驱动的结构性通胀，范围窄但中心位置高，至少持续到2027。", "summary": "截至2026-05-19，美BIS（10/2022、10/2023、10/2024 HBM专项、2025-01 AI Diffusion）+ 日本METI 23项工具 + 荷兰ASML EUV/DUV许可 + 台湾CoWoS流片审批 + 50%关联规则形成的许可证瓶颈，把HBM3E/HBM4与先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力链供应瓶颈已基本解决，支持在 CNH 企稳后回补权重至 50%，利润池正向 HBM4 与先进封装技术前沿发生结构性转移。", "summary": "2026-05-20 卡片支持将 AI 算力链权重回升至 50%。研究显示，HBM3e 良率提升与台积电 CoWoS 产能（年底达 7 万片/月）的实体扩张已基本消除 2024-25 年的瓶颈。虽然日元 carry 解除对日本设备标的造成估值冲击，但其基本面供应受损风险低。HBM4 转型将导致利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "云端算力对HBM及CoWoS产能的虹吸导致边缘端LPDDR5X面临严峻的“存储通胀”，硬件降本逻辑阶段性证伪，高昂的BOM成本将大幅压制端侧AI在2026年的大规模渗透率。", "summary": "截至2026-05-19，我们对端侧AI的存储成本约束进行了深度压力测试。测试表明，该约束比预期更严峻。云端HBM4与先进封装（CoWoS）的大规模扩产对晶圆产能形成“虹吸效应”，导致LPDDR5X/6出现结构性短缺，2026年边缘设备存储BOM成本激增32%-50%。这迫使AI PC与手机涨价5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产供应链存在显著的产能与时间窗口错配（至少18-24个月），寒武纪等AI芯片龙头的订单面临无法交付的真实风险，基本面定价存在随时坍塌的脆弱性。", "summary": "截至2026-05-19，TMT分析确认国产HBM与先进封装在12-18个月内无法有效承接US供应链断裂风险。中芯国际2.5D产能已透支，长鑫存储HBM3量产存在18-24个月缺口。寒武纪等高度依赖HBM的AI芯片公司面临订单无法交付的断裂风险，单纯依赖订单流的估值底逻辑极其脆弱，极易触发流动性踩…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 4, "neutral": 0, "value": 39, "events": [ { "title": "AI算力链盈利可见性已被订单簿三重验证、足以支撑保留成长端的策略；半导体设备应拆分成‘AI直接受益+国产替代超配，纯中国敞口设备与晶圆厂低配’，故第03卡‘不全面切向价值’成立但需要…", "summary": "截至2026-05-21，本卡对第03卡“不全面切向价值”进行压力测试，结论是有条件支持。AI算力链（GPU/HBM/光模块/液冷/先进封装）盈利可见性高且仍在抬升：四大美系hyperscaler 2026年合计资本开支指引上修至USD 360–410 bn、TSMC CoWoS与SK Hynix…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "支持哑铃右端方向，但纠正颗粒度——TMT内只有AI算力国产化+半导体设备/材料具备2026年内可验证的盈利弹性与估值修复空间，把整个科技板块当作进攻端是估值过度延展。", "summary": "2026-05-20，本卡片对卡片09哑铃策略中科技进攻端进行压力测试。结论是有条件支持，但需切分颗粒度：TMT内仅半导体设备/材料与AI算力国产化（昇腾、海光、寒武纪、HBM、先进封装）两条主线具备2026年内可验证的盈利弹性与估值修复空间，合计占TMT自由流通市值约18%-22%。消费电子、面…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI与半导体能在防御性轮动中维持Alpha，但安全边际只存在于盈利可见度高、供给受限且能把CapEx转化为现金流的瓶颈环节。", "summary": "2026-05-20，本卡支持第4张卡的杠铃配置：AI与半导体可保留为15-20%的Alpha袖珍仓位，但不能作为泛防御Beta。SIA/WSTS显示2026年一季度全球芯片销售2985亿美元，环比+25%；TSMC 1-4月收入同比+29.9%，并指引2026年美元收入增长30%以上；NVIDI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产先进封装良率的提升已解除了2026年下半年AI算力建设的硅端产能限制，资本开支的真实硬约束已全面转向电力基础设施。", "summary": "调研确认国产2.5D封装（类CoWoS）良率已突破65%的商业化及格线，HBM3等效良率达55-60%。这一进展足以保障2026年下半年120万张国产AI加速卡的供应，标志着算力扩张的核心瓶颈已从芯片制造实质性转移至电力与电网容量。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 52.4, "events": [ { "title": "在英伟达 Rubin 平台面临供应链瓶颈的窗口期，国产 GPU 凭借先进封装与 HBM3 的全产业链突破，正通过‘十五五’初期智算中心的大规模采购实现对推理市场的实质性支配。", "summary": "截至2026-05-22，本研究确认国产GPU已进入性能与市场双重拐点。NVIDIA Rubin平台受困于HBM4验证延迟与CoWoS-L稳定性，2026年产量目标下调25%。与此同时，海光深算4号与寒武纪思元690通过Chiplet技术实现性能跃升，结合长电、通富在2.5D封装的突破以及长鑫存储…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "出口管制传闻在TMT基本面侧的真实冲击远小于午盘股价隐含的悲观幅度，但敞口分布的非均匀性意味着红利轮动结束后将出现「国产替代纯度高」标的的结构性重定价。", "summary": "针对2026-05-22午盘发酵的出口管制传闻，TMT侧的供应链拆解显示敞口分布并不均匀。光模块龙头(300308/300502)海外收入占比超85%，是直接受冲击点，但800G/1.6T的产能集中度形成天然壁垒，真实营收冲击预计仅5–15%；服务器代工(601138/000977)依赖墨西哥/越…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 1, "neutral": 0, "value": 67.8, "events": [ { "title": "NVIDIA第一季度财报超预期与AI基础设施资本开支展望：黑曜石平台（Blackwell）主导地位与CoWoS产能博弈", "summary": "截至2026年5月21日，NVIDIA于昨日发布的Q1 FY2027营收录得816亿美元，同比增长85%，显示出AI算力需求极强。2026年海外四大云巨头资本开支预计突破7000亿美元，其中约75%投向AI。与此同时，台积电CoWoS先进封装产能预计在2026年底提升至12-13万片/月，NVID…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI资本开支足以抵消重工业衰退冲量并能驱动半导体层利润率周期性走高，但云基础设施层利润率短期反而受压——卡片4的'TMT整体利润率扩张'标签需被分层重写为'半导体↑/云↓'。", "summary": "对卡片4双重命题做分层压力测试：$3,750亿AI资本开支能抵消OFS/重工资本下行（算术差距62倍，支持）；但'TMT结构性利润率扩张'必须拆成上下游——上游半导体（NVDA、TSMC、HBM三巨头）凭CoWoS/HBM4供给瓶颈+客户付款意愿，毛利率扩张是真命题；下游云基础设施（Azure/A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "半导体与网络核心硬件 2026Q1 的毛利韧性与超大规模 capex 指引共同证实：DeepSeek 推动的推理 token 通缩压力主要消化在 API 层，AI 利润已在事实上向算…", "summary": "截至 2026-05-26，本卡支持 card-01 的利润迁移假设。NVIDIA FY2027 Q1 收入 $81.6B、Data Center 收入 $75.2B、GAAP gross margin 74.9%，毛利仍处历史高位；Microsoft 与 Meta 的 2026 自然年 cape…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 2, "neutral": 0, "value": 56.6, "events": [ { "title": "短期硬约束在 HBM 与 CoWoS-L 而非电力，AI 算力'紧但能出货'，对指数 FOMO 有利但 2026H2 脆弱性显著上升。", "summary": "短期（2026 年内）芯片侧的约束比电力更紧：依次瓶颈为 HBM3e bit 供给 → 台积电 CoWoS-L 先进封装 → GPU 晶圆 → 基板/光模块 → 电力。HBM 与 CoWoS-L 全年额度已锁、扩产跑不过 ~150%/~70% 的需求增长，2026H2 任一良率事件或出口管制重置可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供给的物理限制已内化为芯片架构的强制约束，AI 硬件溢价正从单纯的算力规模转向“单瓦性能”与“封装级电源集成”的确定性。", "summary": "截至 2026-05-26，电力设备交付延迟（变压器周期达3-5年）已迫使 AI 硬件架构从“性能优先”转向“功率密度优先”。NVIDIA Rubin 架构通过 HBM4、800V HVDC 和集成电压调节器 (IVR) 应对挑战。这引发了 SiC/GaN 的产能紧缺，并暴露了先进封装材料（如日东…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 4, "neutral": 2, "value": 52, "events": [ { "title": "区域化认证时滞与多供应商材料栈分裂打破了先进封装物理结构上的“黄金匹配”，其引发的热力学翘曲与 12-18个月的合规验证使得 AI 芯片先进制程量产风险实质性飙升。", "summary": "截至2026-05-27，本卡通过压力测试确认：区域化合规与材料栈重构不仅仅是法务摩擦，更打破了原后台积电生态的热力学一致性。由于不同区域、多供应商（SK Hynix/Samsung/Micron）的材料存在微小物理差异，使得 CoWoS-L 和 HBM4 在超过 1000W 功耗下遭遇极端的 C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HBM3E面临产能过剩风险，而HBM4的逻辑底晶粒与混合键合成为新瓶颈；建议在14%的先进封装配置内，低配纯存储厂商，超配代工及前端设备商。", "summary": "本报告基于2026年5月的最新产能数据，对策略师的先进制程超配逻辑进行了压力测试。我们发现，虽然台积电CoWoS产能到2026年底将激增至13万片/月，有效缓解了绝对瓶颈，但HBM市场正面临分化：三星激进扩产（25万片/月）导致HBM3E存在价格战风险，而HBM4因引入逻辑底晶粒与16层混合键合，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年A股先进封装板块呈现显著估值分化：半导体设备具有高业绩弹性但估值普遍偏高，而高壁垒封装材料（低α硅微粉及CMP平台材料）与性价比测试机龙头在业绩确定性与估值匹配度上显著占优…", "summary": "本卡片系统评估了2026年A股先进封装设备与材料的国产替代业绩弹性与估值。台积电CoWoS产能外溢及HBM4混合键合变革，拉动封装测试及高壁垒耗材爆发。设备板块呈“高弹性、高估值”分化，长川科技（44x PE）成长匹配度佳，拓荆科技（70-71x PE）估值偏高；而材料板块（联瑞新材25-30x …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 算力物理墙的宏观结果是资本开支被拉长而非崩盘，TFP 红利右移 18-24 个月，配置上有利于电力/封装/国产替代而非纯算力多头。", "summary": "截至 2026-05-27，前序卡片揭示的 CoWoS-L 翘曲与 HBM4 多供应商 KGD 危机不会引爆 CSP 资本开支崩盘，而是把周期向后拉长至 2028-2029：Big Four 2026 资本开支 USD ~470bn 大概率维持，但 GPU 实际出货低于指引 12-18%，BOM …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 2, "neutral": 0, "value": 59.4, "events": [ { "title": "先进封装与液冷散热的产能扩张确实进入10×量级周期，但受ABF/HBM/稀土/铜的上游约束，仅能在制造维度缓解约35–40%的三重瓶颈，无法证伪TMT分析师的结构性紧张判断。", "summary": "工业制造视角支持TMT的三重瓶颈框架。TSMC CoWoS月产能从2023年末13K wpm扩张至2026年末130K wpm（近10×），但CoWoS-L/S全年售罄、NVIDIA独占60%以上配额，ABF载板与HBM堆叠的18–24个月扩产周期意味着封装紧张延续至2027年中。液冷市场2026…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年AI芯片与先进封装产能约束会推高并拉长云端CapEx，而不是使其取消；配置应偏向具备配额权的上游瓶颈环节。", "summary": "截至2026-05-28，本卡支持前序判断：AI硬件CapEx传导链的约束真实但不等于削减。Microsoft、Amazon、Alphabet、Meta的2026年CapEx指引中枢约为$710 billion，反映的是芯片、HBM、先进封装、电力和数据中心资源稀缺导致的锁单与涨价，而非需求走弱。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [ { "title": "AI硬件周期仍由需求驱动，但未来12个月的定价权将更多落在HBM、先进封装、光模块和电力交付等瓶颈环节，而不是单纯GPU或服务器组装收入。", "summary": "截至2026-05-30，本卡支持“AI算力瓶颈从硅端扩展至物理基础设施”的主线。未来12个月，AI芯片和服务器需求仍由超大云厂商资本开支、云AI收入、agentic AI推理流量和自研ASIC扩张驱动；但供应瓶颈已从GPU单点扩展到HBM、CoWoS/先进封装、800G/1.6T光模块、机架供电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 3, "neutral": 0, "value": 64.2, "events": [ { "title": "2026 HBM 缺口真实但租金窗口前置在 1H26，真正的稀缺量子是 CoWoS 先进封装而非 HBM 裸晶——杠铃方向成立但工具应以离岸为主、上半年加仓、Q4 减仓。", "summary": "截至 2026-05-30，本卡对 Research note 05 杠铃配置的 HBM 瓶颈腿进行供应链侧压力测试。2026 HBM 毛缺口约 10% 真实存在但正快速收敛——SK Hynix、美光产能已预售，三星 HBM3E-12H 已通过 NVIDIA 认证、HBM4 于 1Q26 送样，2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年半导体的结构性毛利重估完全由AI服务器（HBM与先进封装）的高壁垒和产能紧缺驱动，与国内消费电子补贴的存量博弈完全脱钩。", "summary": "截至2026-05-30，本卡支持“半导体毛利扩张脱钩于消费补贴”的判断。研究表明，2026年AI服务器出货量预计同比增长超28%，其价值量将占服务器总市场的74%。SK海力士HBM产能已售罄，HBM4在2H26带来超30%溢价，毛利率维持极高水平；台积电CoWoS产能2026年底将突破11.5万…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "非美系低轨星座的自主化上限不是芯片设计口号，而是能否获得CoWoS/HBM/材料产能并完成空间级资质认证。", "summary": "截至2026-05-30，本卡片支持前序判断：非美系低轨星座的AI硬件瓶颈已从单一GPU采购，转向CoWoS/HBM/基板/材料/测试良率组成的封装级工业体系。监管规避可以降低对美国加速器的直接依赖，但无法自动获得先进封装、HBM和材料产能；Eutelsat替换周期与Qianfan部署压力显示，星…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Chiplet能为国产AI算力芯片争取时间和良率弹性，但只能部分补偿先进制程受限，真正受益者更偏先进封装赋能者和系统级NPU平台，而非泛AI芯片beta。", "summary": "截至2026-05-31，本卡对card-07进行TMT产业压力测试：Chiplet和先进封装能在国内AI NPU、推理卡、异构加速器和集群级设备中提供部分性能与良率补偿，但不能完整替代先进制程密度、HBM供应和CoWoS级封装执行。资金端的5000亿元科技再贷款与3440亿元大基金三期能维持ca…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 4, "neutral": 1, "value": 54.6, "events": [ { "title": "AI 算力供给受限于 HBM4 售罄、CoWoS 封装良率及全球电网容量的物理极限，使硬件龙头成为 AI 价值链中无视‘Capex 断崖’风险的最终收费站。", "summary": "截至2026年5月，AI硬件供应链仍处于结构性紧缺。TSMC N2量产初期良率65-70%，HBM4于2月投产且美光2026年产能已售罄。NVIDIA Rubin架构推升单GPU功耗至2.3kW，使电力供应成为比芯片更硬的瓶颈。尽管终端应用出现DAU增速放缓，但硬件端的实物稀缺性构成了核心护城河，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 主权 AI 芯片订单", "summary": "致 主题研究员： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 7c8a3a38-9d97-4b33-a026-f52da73fec07 结果摘要: Wrote handoff_response.md (10450 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "超级IPO将把AI公司的公开融资压力传导为算力锁定需求，从而重估半导体资本开支：超配稀缺瓶颈，回避无法证明产能稀缺和合约期限的长尾扩产。", "summary": "截至2026-05-31，本卡片支持“超级IPO流动性虹吸”对AI硬件和半导体瓶颈资产的强化作用。核心判断是，IPO资金不会平均利好半导体，而会把公开市场资金约束转化为GPU、HBM、先进封装、先进制程、光刻和数据中心网络的多年期需求可见度。TSMC、NVIDIA、ASML、Micron、Sams…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力需求具备跨年度刚性，订单能见度已穿透至2027年，产能过剩仅限于结构性错配，核心挑战在于软件端的投资回报率验证。", "summary": "AI半导体产业链正处于“供给受限型超长周期”。HBM与CoWoS产能2026年已全额预售，海量预付款锁定了WFE设备厂商（ASML, AMAT）的订单。尽管HBM3e可能在2026年底出现结构性过剩，但HBM4与Blackwell Ultra/Rubin需求将维持至2027年。目前领先制程零库存，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 1, "neutral": 1, "value": 74, "events": [ { "title": "Research note 07 · 高端封装与 HBM 产业链上游材料的供应稳定性与毛利弹性分析 工作基准日： 2026 06 03（Asia/Singapore） 分析师：材料行…", "summary": "Research note 07 · 高端封装与 HBM 产业链上游材料的供应稳定性与毛利弹性分析 工作基准日： 2026 06 03（Asia/Singapore） 分析师：材料行业分析师（materials analyst） 卡片编号：7 / 10 立场：support（延续 Research…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Anthropic 的高估值实质上是基础设施稀缺性的资本化体现，其 3300 亿美元的算力承诺既是模型领先的物理保证，也是将云厂商 Capex 压力转嫁给二级市场的潜在风险点。", "summary": "Anthropic 9650亿美元的估值深度绑定了超过3300亿美元的云算力承诺（AWS 1000亿，Google 2000亿）。虽然这锁定了面向2027年单次200亿美元训练成本所需的5GW+电力与下一代 NVIDIA Rubin/TPUv7 产能，但也使公司陷入“资本开支通道”模型。2026年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 3, "neutral": 0, "value": 60.4, "events": [ { "title": "AI算力链的绝对瓶颈正在从HBM存储晶圆短缺结构性转移至由TSMC主导的CoWoS-L先进封装尺寸扩张和逻辑基础晶圆制造代工能力。", "summary": "本报告分析了NVIDIA在2026年6月确认Samsung、SK Hynix和Micron全面进入Rubin平台的HBM4供应链这一关键拐点。我们指出，随着三足鼎立打破垄断，HBM供需瓶颈将显著缓解。然而，由于HBM4基础晶圆逻辑代工的双重锁定，以及CoWoS-L封装面积扩张对晶圆利用率(GDW)…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS产能硬约束会把AI硬件估值溢价从单纯设计叙事重新分配给掌握晶圆、封装、良率和整柜交付能力的产能控制者。", "summary": "2026-06-03：本卡支持前序判断，认为CoWoS与先进封装已成为AI算力链估值的硬过滤器。NVIDIA需求仍强，TSMC的3 nm、HPC与先进封装紧张共同支撑产能控制者溢价；但HBM4供应商扩容削弱单纯存储稀缺溢价。A股半导体整体估值已高，策略重点应从宽泛芯片设计beta转向设备、材料、封…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与材料的刚性约束已把AI优化重心强制推到HBM4、3D混合键合、CPO光学与MoE/FP4等芯片级架构层；这些改进在2026-2030年可把单位能力设施功耗降低15-25%，足以…", "summary": "截至2026-06-03，本卡支持上一张材料卡的判断，并将瓶颈传导至芯片与封装栈。HBM占封装功耗30-50%，HBM4（SK海力士、Micron 2026年量产）以约2.3 pJ/bit较HBM3E能效改善30-40%；TSMC SoIC/CoWoS-L把跨die搬运能耗压低约10倍；NVIDI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 0, "neutral": 0, "value": 87.8, "events": [ { "title": "Chiplet 与 3D 先进封装可在 18–24 个月内把国产 AI 芯片单卡等效算力推回旗舰 60–75% 区间，是先进制程受限下唯一可兑现的非对称工程路径，但无法消除 HBM …", "summary": "2026-06-04：Chiplet + 2.5D/3D 先进封装已成为国产高性能 AI 芯片在 7nm/N+2 被锁条件下的唯一可量产非对称路径。长电 XDFOI、通富微电、华天科技、盛合晶微在 2.5D CoWoS-like 与 9μm 混合键合上小批量验证；华为昇腾 920、寒武纪 MLU5…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 3, "neutral": 1, "value": 66.2, "events": [ { "title": "HBM与CoWoS的瓶颈真实但明显弱于电力设备，30%算力核心权重应维持不变，但内部应从NVIDIA+SK海力士集中转向Micron+三星对冲以应对2H 2027份额再分配与供过于求…", "summary": "截至2026-06-05，HBM3e/HBM4与CoWoS-L在2026年仍紧张但2H 2027趋于平衡：台积电CoWoS月产能从2024年底~35-40k wpm扩至2026年底~75-80k wpm、2027年~130-140k wpm；HBM行业比特位2026年同比+60%、收入升至USD …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产AI防务硬件呈现“射频自主代际领先，数字算力封锁突围”的结构化特征，估值重锚应区分成熟的射频供应链与尚存代际差的数字算力链。", "summary": "截至2026年6月6日，国内AI防务算力已确立以中芯国际7nm为底座、长电/通富先进封装为桥梁的“自主存续”路径。相控阵T/R芯片（国博、铖昌）已实现GaN国产化量产，业绩随星座组网进入爆发期。然而，星载AI算力仍受限于5nm制程的“良率墙”和HBM3与全球顶尖水平（HBM4）的2代代际差。结论是…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI推理需求爆发导致底层算力成本向2nm及HBM4集中，NVIDIA定价权虽稳固但面临ASIC分流，硬件供应链利润向上游代工与存储集中，进而长期压制下游SaaS厂商基于API计费的毛…", "summary": "截至2026年6月5日，AI算力市场正经历“推理成本倒挂”，推理需求已占据总算力需求50%以上。尽管NVIDIA通过Blackwell及即将推出的Rubin（R100）维持着70%+的毛利率与极高的定价权，但随着云厂商（CSP）自研ASIC芯片出货量激增（增速达44.6%），硬件利润结构正被重塑。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HBM 与 CoWoS 产能释放是真实的，但被同一批云厂商和主权 AI 买家全部吸收，模型层无法在 12–18 个月内通过芯片路径实现算力主权——'模型主权'必须走数据/工作流/监管…", "summary": "截至 2026-06-05，对'HBM/CoWoS 供给释放将带来模型主权'的命题进行压力测试，结论倾向否认。2026 年 CoWoS 月产能将由 2023 末 1.3 万片放大至 ~13 万片（NVIDIA 独占 50%+），HBM 市场由 $38B 增至 $58B，但 HBM3e 合约价仍涨 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 3, "neutral": 0, "value": 64.6, "events": [ { "title": "Rubin放量斜率放缓要求资金从单纯的“全球硬件Beta”向“先进封装价值捕获”与“本土供应链替代Alpha”进行战术轮动。", "summary": "鉴于HBM4 12-Hi良率瓶颈主要压制NVIDIA Rubin在2026年Q4至2027年Q1的放量斜率（早期机柜产出或折损15%-21%），而非初始交付，我们建议采用“哑铃型”配置策略。一方面，对全球AI硬件龙头保持核心底仓，但战术性下调权重以规避短期估值消化；超配台积电等先进封装纯标的。另一…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CoWoS-L 产能按计划打开, 保住了 Rubin 的量, 但 2026–2027 年毛利率剧本的主笔已交棒给 HBM4。", "summary": "TSMC CoWoS-L/R 产能从 2025 年末约 75 kwpm 扩至 2026 年末约 135–140 kwpm, 已足够支撑 NVIDIA Vera Rubin 在 4Q26 NVL144 量产并留单季度缓冲。CoWoS 单位成本仅是二级毛利率杠杆, 一级杠杆是 HBM4 ASP——15…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SK hynix HBM4 12-Hi良率大概率足以支撑Rubin初始交付，但不足以排除2026年Q4至2027年Q1放量斜率受压。", "summary": "截至2026-06-06，公开证据不支持SK hynix HBM4 12-Hi已出现确定性良率失误；更合理的压力点是Rubin放量斜率。NVIDIA已称Vera Rubin进入全面量产爬坡，并有三家HBM4供应商获认证/生产。SK hynix的1bnm、Advanced MR-MUF、TSMC b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 2, "neutral": 0, "value": 72, "events": [ { "title": "以Chiplet与先进封装为核心的硬件架构突破是AI推理边际成本系统性通缩的底座支撑，消除了SaaS及网安企业结果计费（OBP）转型的可变成本风险。", "summary": "本报告探讨了Chiplet与先进封装技术（如CoWoS、HBM4、SoIC）对AI推理成本通缩的底层驱动。测算显示，Chiplet异构集成在高性能芯片上可降低42.85%的制造成本；高带宽HBM4内存与CoWoS-L封装突破内存墙，使3卡Rubin集群相较8卡H100集群带宽提升2.46倍、开支减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Rubin Capex能保护一线AI硬件EPS，但WACC上行会把二线供应链从“AI成长资产”重新定价为“融资周期资产”。", "summary": "截至2026-06-08，本卡对前序AI硬件防御结论做压力测试：Rubin周期与云巨头Capex仍能支撑NVIDIA、TSMC、Broadcom、Samsung、Micron等一线硬件龙头的EPS增长，但不能让整个供应链免疫WACC上行。可见四大云厂商Capex底线约5742亿美元（自测算），足以…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 1, "neutral": 2, "value": 85.4, "events": [ { "title": "Google-Intel的TPU订单标志着AI硬件利润池的结构性迁移：从绝对依赖GPU/TSMC，转向ASIC出货量反超以及Intel Foundry作为第二供应源的崛起，但GPU仍…", "summary": "2026-06-09评估：Google向Intel Foundry下单2028年超300万颗TPU，且Nvidia正评估Intel 18A工艺，这表明AI总需求依然强劲（到2028年TAM达3800亿美元），但硬件利润池正发生结构性迁移。为降低TCO，云厂商加速自研ASIC，出货量预计将超越GPU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "国产先进封装与互联技术的突破，可通过异构集成提升系统算力密度并降低44.79%的硅成本，结合HBM3可释放12倍的有效算力利用率，成功弥补了因先进制程限制带来的“商业效率”落差。", "summary": "国产Chiplet与CoWoS先进封装可在7nm DUV多曝光节点下提升系统算力密度。自测算显示，将800mm²大芯片拆分为4个200mm²逻辑芯粒与1个成熟I/O芯粒，可将芯片成本降低44.79%，且结合自研HBM3（1.2 TB/s）可将训练集群的算力利用率（MFU）由5.00%提升至60.0…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 5, "neutral": 0, "value": 56.8, "events": [ { "title": "AI硬件正通过2.3kW单芯片功耗极限与800V液冷架构强行突破物理墙，算力扩张的限制已从芯片设计转移至电力设施配套的48个月长周期交付延迟。", "summary": "截至2026-06-11，AI硬件并未因电力约束而降低TDP，反而迈入2.3kW（NVIDIA Rubin）的“暴力能效”时代。这迫使行业全面转向100%液冷及800V DC机架架构。硬件演进逻辑已从“硅片优化”转向“基础设施重构”，先进封装（CoWoS-L/HBM4）与光互连（CPO）成为维持算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI半导体核心标的的Q2/Q3盈利能见度极高，受限的电网电力将加速Blackwell架构升级，短期流动性冲击构成极佳的左侧建仓窗口。", "summary": "截至2026-06-10，在SpaceX IPO（SPCX）可能引发SMH/QQQ短期30亿至60亿美元流动性抽水的背景下，本报告研判AI半导体核心标的（Nvidia, TSMC）Q2/Q3盈利能见度依然坚实。TSMC 2026年520-560亿美元资本开支高企，CoWoS先进封装产能供不应求。即…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年量子混合架构已具备商业化量产的互连技术标准，但其实际落地规模受制于AI算力对先进封装产能的挤出效应，呈现“技术就绪、产能告急”的结构性矛盾。", "summary": "2026年，量子-经典混合架构的硬件瓶颈已从比特制造转向互连技术与先进封装。NVIDIA NVQLink (<4μs延迟) 与IBM模块化Kookaburra处理器标志着商用化互连的成熟。然而，供应链面临AI GPU对台积电CoWoS/SoIC产能的严重排挤，量子硬件交付周期拉长。Cryo-CMO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年先进封装不是没有扩产，而是扩产首先服务头部AI客户，长尾AI、量子混合和其他HPC项目仍要承受配额挤压。", "summary": "截至2026-06-11，本卡压力测试认为，CoWoS/SoIC扩产是真实且规模化的，但2026年仍不足以完全缓解AI GPU对先进封装的挤出。台积电2026年资本预算指向USD 52-56 billion高端，其中约10-20%覆盖先进封装、测试、光罩及其他；KGI预计2026年CoWoS年产能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 0, "value": 82.2, "events": [ { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 0, "value": 65.6, "events": [ { "title": "为支撑三层算力织物，AI芯片与先进封装如何演进以匹配电力波动与边缘推理的功率包络", "summary": "综合并支持卡片04的三层算力织物，下沉到硅基与封装层。关键洞见：'电力波动友好'主要是电力电子＋调度固件问题，而非新晶圆问题。NVIDIA GB300靠电源货架（65 J/GPU储能＋功率封顶斜率＋功率烧除）把Megatron训练电网峰值砍约30% [S1]，DVFS早是标配 [S2]；同步功率摆…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装/HBM仍是真瓶颈，但2026-06-12的上市交易已经把2027-2028年认证、良率和产能利用率提前资本化，最需要防的是技术代际错配与融资拥挤触发的估值和盈利预期双杀。", "summary": "截至2026-06-12，本卡对先进封装与HBM链条做压力测试：风险真实存在，但不是2026年立即供过于求。HBM到2027年前仍偏紧，核心风险在于HBM4/HBM4E和更大CoWoS形态迭代快于国内上市代理资产认证速度；叠加长电、通富、寒武纪等高估值和高融资余额，股价可能先于EPS下修而去估值化…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力刚性约束已把AI加速器竞争维度切换为每瓦Token，Chiplet+先进封装是2027-2028电网窗口内唯一可规模化兑现的算力密度救援方案，对NPV的杠杆是5-8倍。", "summary": "电网刚性约束把AI加速器的KPI从峰值FLOPS切换为每瓦Token吞吐。Blackwell B200单卡1000W、GB200 NVL72机柜120kW已触顶，下一代Rubin/MI400无法继续抬TDP，必须靠能效换性能。TPU v5p、Trainium2较H100能效提升30-40%；同10…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "进一步出口管制确会拉长芯片时间表并显著退化算力，但几乎完全落在中国云厂商身上；对西方云厂商，封装/HBM瓶颈是商业爬坡风险、不被管制叠加放大，其唯一负面通道是中国对等矿产报复这一自找…", "summary": "对出口管制叠加先进封装/HBM瓶颈做压力测试，结论是风险按地理分叉而非对称。对西方云厂商，进一步的美对华管制不会加剧供应时间表，反因剥离中国需求把韩台HBM4/CoWoS配额导向西方（CoWoS 2026年底扩至13万片/月），故card-05的NPV悬崖不被放大；西方真正尾部是中国镓锗锑许可证（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 4, "neutral": 2, "value": 73, "events": [ { "title": "CoWoS 与 HBM 扩张是 NVIDIA 继续销售高溢价系统的供给条件，而不是摧毁溢价的触发器。", "summary": "2026-06-13：本卡支持前序判断。CoWoS/HBM 扩产不是削弱 NVIDIA 护城河，而是让其把更多机柜级 AI factory 供给转化为计算、内存、网络与软件平台溢价。TSMC 的先进封装、3nm/HBM base die 产能，SK hynix/Micron/Samsung 的 H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GB200 级效率提升改善 AI factory 经济性，但 2026-2028 年真正限制部署节奏的是站点电力、并网、变压器、稳定电源和电网运行能力，而不是 CoWoS/HBM 单…", "summary": "2026-06-13：本卡对 NVIDIA 护城河做电力压力测试。GB200 NVL72 单机柜约 120 kW，1 GW IT 功率约支持 8,333 个机柜、约 600,000 颗 GPU。IEA、DOE/LBNL 与 NERC 显示，数据中心用电、并网队列、变压器和燃机交期都在拉长，大型计算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红皇后竞速并未彻底抹去自研ASIC的成本优势，而是将其集中至Google与AWS两家、使其他云厂商回到对NVIDIA的依赖，从而部分推翻了Research note 05所述的全行业…", "summary": "在'红皇后竞速'下，自研ASIC单世代有效寿命已压缩至12–15个月，2nm节点全口径NRE攀升至USD 0.9–1.4B、N2晶圆约USD 30,000/片、CoWoS-L售罄至2027。我们测算自研ASIC在TCO上跑赢商用GPU的年部署门槛从5nm节点的约20万颗升至现实情景下的15万–25…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "CoWoS绝对产能正在改善，但GPU与自研ASIC需求吸收速度更快，因此2026年AI集群交付仍受先进封装、HBM和关键设备交期的结构性约束。", "summary": "截至2026-06-13，本卡支持上一张AI基建卡的判断：硬件交付风险不只来自并网，还来自CoWoS/HBM/设备交期的半导体转换链。TSMC仍称先进封装非常紧，TrendForce口径显示CoWoS-L/S满载且2026年底目标120,000-130,000片/月；DIGITIMES虽更乐观，预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 5, "neutral": 0, "value": 62.4, "events": [ { "title": "AI先进封装与HBM产能瓶颈对算力基础设施扩张及超大规模厂商资本开支的影响", "summary": "支持半导体分析师Research note 01的论点并向下游延伸：CoWoS/HBM确为算力的关键约束，2026年部署受封装与内存所限（CoWoS扩至约13万片/月仍订满；SK Hynix与Micron 2026年HBM售罄；HBM4验证延迟使NVIDIA将Rubin目标由200万下调至150万…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI芯片供应链中现金流的高确定性支撑了科技股的信用乐观，而CoWoS与HBM产能瓶颈及地缘政治集中度则引发了期权市场的尾部焦虑。", "summary": "本报告分析了导致跨资产风险溢价分裂的AI半导体供应链逻辑。信用市场的乐观情绪源于AI基础设施建设带来的高确定性现金流（如NVIDIA锁定台积电大部分CoWoS产能）。而期权市场的焦虑则源于结构性的供应链瓶颈（CoWoS和HBM4产能受限）导致的尾部风险，以及先进封装工艺高度集中在台湾地区所带来的地…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GPU抵押贷款收缩会削弱边际Neocloud订单可见度，但先进封装与先进制程需求更可能从融资套利订单转向信用更硬的超大规模厂商和ASIC路线，而非进入全面下行。", "summary": "截至2026-06-15，本卡压力测试GPU抵押融资收缩对半导体订单的影响：冲击不是AI芯片需求消失，而是订单可见度从Neocloud融资驱动采购转向超大规模厂商、ASIC、主权项目和带预付款的硬合同。NVIDIA Data Center在FY2027 Q1仍达752亿美元，TSMC先进节点和HP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在AI链内部按融资质量轮动：超配Tier-1平台与卖铲人(设备/HBM/封装/电力赢家)，低配靠GPU抵押融资循环的Neocloud股权与高收益债。", "summary": "综合前八卡推演，AI基础设施的约束已从硅到电力、再到融资可得性。本卡给出按融资质量过滤的五篮子相对配置框架：B1 Tier-1 AI平台(+300bp)、B2 设备/HBM/先进封装(+250bp)、B3 ASIC定制(+100bp)、B4 电力瓶颈赢家(+150bp)、B5 低质量AI基建(-2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 0, "value": 75.8, "events": [ { "title": "宏观信用收紧并未击穿全球AI周期，而是将AI基础设施层切割为「超大规模云厂＋央企国家队」核心（继续扩张）与「民营商业云」外缘（防御性收缩）的K型结构，TMT投资应聚焦CoWoS/HB…", "summary": "TMT综合：沃什鹰派持稳（3.50–3.75%）经离岸人民币流动性、汇率对冲摩擦与企业中长贷收缩，传导至AI基础设施层产生K型分化而非总量收缩。美超大规模云厂CY2026合计capex运行率3,200–3,600亿美元由经营现金流支撑不受冲击；央企运营商云经PBOC科技创新再贷款（约2.5%）继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "验证AI算力硬件链的估值承压：流动性虹吸下的先进封装与GPU产业链韧性评估", "summary": "本卡对AI算力硬件链做压力测试，得出'分化韧性'结论。资金流层面确认虹吸：2026-06-08半导体−219.33亿、通信−98.35亿、消费电子−69.92亿元净流出居前[S6]，年内93家半导体公司减持[S7]，叠加CXMT约295亿IPO锚定[S9]。但估值下移是资金流/轮动驱动、非盈利驱动…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（str…", "summary": "卡片 08 — AI芯片竞争格局：ASIC侵蚀风险与商用GPU基本面评估 日期（机构工作日）： 2026 06 15 （亚洲/新加坡） 分析师：半导体分析师 立场： 压力测试（stress test） 主题：商用GPU面临的ASIC份额侵蚀，及主要超大规模云商（Hyperscalers）定制ASI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 1, "risks": 6, "neutral": 0, "value": 67.2, "events": [ { "title": "芯片能效提升不足以抵消机柜功耗密度的物理极限，电力供应仍是制约AI资本开支扩张的物理红线。", "summary": "本报告对2026年中期的半导体能效提升进行了压力测试。虽然NVIDIA Rubin (R100)等新架构在推理能效比上提升了10倍，但单卡功耗攀升至2.3kW，机柜密度突破250kW，加剧了电网侧变压器和输电设施的改造压力。先进封装（CoWoS-L）和HBM4虽缓解了显存瓶颈，但未能扭转算力需求增…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供给限制驱动的半导体能效转型主要表现为通用GPU的系统与精度级优化，而非ASIC/NPU的大规模替代；自研定制芯片的爆发反而加剧了全球先进制程与封装供应链对TSMC的物理集中度。", "summary": "截至2026-06-17，电力瓶颈确在倒逼算力能效转型，但并非单纯的通用GPU向专用ASIC切换。算法多变性（如MoE与动态推理）使自研ASIC面临极高的“代差失效”风险，通用GPU（如NVIDIA Blackwell）凭借软件与架构灵活性仍占据70%以上份额。能效提升主要通过系统精度（FP4/F…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "半导体能效能把AI扩张从“堆GPU”改写为“每MW token产出竞争”，但只能买时间、改善单位经济性，不能让电力瓶颈消失。", "summary": "截至2026-06-16，本卡压力测试“电力预算驱动扩张”逻辑：半导体能效提升真实且重要，Blackwell、先进封装、vLLM/FlashAttention、稀疏MoE和定制ASIC可显著提升每MW有效token产出。但它不能完全对冲电力瓶颈，因为美国数据中心用电量可能从2023年的176 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HBM与2.5D先进封装是昇腾算力的真正瓶颈，国产替代处于中段、受良率限制且集中于单点（盛合晶微/华为自研HBM），故2026年供给受限——价值归于已验证瓶颈环节，而非泛概念。", "summary": "2026-06-16 半导体供给端判断（支持Research note 3）：DeepSeek-昇腾算力的两个硬约束是HBM与2.5D/CoWoS级先进封装。美国2024-12-02 BIS以带宽密度>2GB/s/mm²做卡点，昇腾910C仍依赖三星/SK海力士旧HBM库存且已拉警报。华为自研HB…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 2, "neutral": 0, "value": 84.6, "events": [ { "title": "AIDD 硬件瓶颈在 HBM 容量/带宽与 CoWoS/SoIC 先进封装，而非算力；HBM4 + CoWoS-L + NVL72 三位一体绑定决定了 2026–2027 工业级生物…", "summary": "AIDD 工作负载（结构预测、全原子 MD、FEP、Cryo-EM）属于访存带宽与片上 SRAM 受限型，而非算力受限。HBM3e 已足以支撑 AlphaFold 3 量级推断，但工业级 10^6–10^7 原子 MD 与 FEP 需要 HBM4（≥1.6 TB/s、36–48GB/stack）+…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 能源重定价压的是超大规模厂商 ROIC 而非芯片 capex 效率——硅片主导 TCO、能源占比单位数，且性能/瓦溢价+杰文斯回弹反而净支撑前沿逻辑/先进封装/HBM 需求…", "summary": "2026-06-18 压力测试：能源是证伪芯片逻辑的错误杠杆。Epoch 1GW 机房年化 TCO 约 $8.5B，服务器占 ~59%、能源仅 ~7%；交付电价 +30% 仅抬升 TCO 约 +2.1%，翻倍也仅 +7%（自测算），而 IT 寿命 5→3 年使 TCO +41%——刷新节奏的影响是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 3, "neutral": 0, "value": 80, "events": [ { "title": "电力与封装成本上升会触发AI基础设施技术路线调整，但截至2026-06-20仍不足以触发超大规模云厂商AI芯片采购总量的全面回撤。", "summary": "工作日2026-06-20：本卡压力测试认为，电力和先进封装成本上升不会在2026年触发超大规模云厂商全面削减AI芯片采购，但会提高单位弹性，推动技术路线从“更多GPU插槽”转向“每瓦性能、液冷机架、自研ASIC、提前锁定CoWoS/HBM和电力站点”。真正风险是并网、变电站、CoWoS或HBM配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler 2026 AI资本开支由CoWoS+HBM+电力三段瓶颈共同定量；卡片08的“中国+一地未验证”使得台/韩集群集中度无法廉价对冲，构成市场定价不足的尾部风险。", "summary": "本卡承接卡片08（先进制程设备与材料的“中国+一地”尚未验证），对全球Hyperscaler 2026–2027 AI算力扩建做压力测试。结论：真正约束并非台积电先进节点晶圆，而是三段串联瓶颈——CoWoS先进封装、SK海力士/美光/三星HBM4、以及数据中心电力与并网。美系五家FY26资本开支合…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "xAI Colossus 2 与 Starlink 三代卫星建设对半导体算力链的量化带动与国产替代受益标的", "summary": "从半导体视角支持本议题：Colossus 2（约 55.5 万颗 GB200/GB300、约 180 亿美元）是真正撼动市场的芯片买家——自测算单站点即约 55–62 亿美元 HBM3E、约 1.9–2.2 万片 CoWoS 晶圆，集中在价值迁移之处（HBM 现占加速器 BOM >50%、CoWo…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 2, "neutral": 0, "value": 87.4, "events": [ { "title": "电力约束不会终结AI算力扩张，但会把半导体设计目标函数改写为每瓦有效算力，并提升Chiplet与先进封装的战略溢价。", "summary": "截至2026-06-21，本卡片支持前序判断：电网接入和可交付兆瓦成为AI扩张区域性天花板后，半导体竞争会从峰值算力转向系统级Perf/Watt。数据中心用电增长使每瓦tokens、HBM带宽、die-to-die互连、热设计和先进封装成为容量权变量。Chiplet、CoWoS/SoIC、HBM4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件供应链波动与融资风险传导", "summary": "作为半导体分析师，我对 Research note 03–07 做压力测试并最终支持，但补上一层基本面触发器：光模块/先进封装景气真实但狭窄且反身。2026 全球 1.6T 光模块需求约 500 万只，中际旭创目标 500–600 万只、约 50% 份额、约 75% 硅光 [S1]，但估值高达 7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 0, "value": 88.8, "events": [ { "title": "截至2026-06-22，低端封装过剩会压低错误产能池的价格，而高端迭代缺失会保留真正重要产能池的配给风险，使Hyperscaler AI资本开支更重、更不稳定、更容易出现已供电机房…", "summary": "截至2026-06-22，本卡压力测试卡片06：先进封装低端过剩并不能缓解Hyperscaler真正需要的高端CoWoS/HBM4认证产能短缺。TrendForce称2026年行业CoWoS-like产能可接近200,000 wpm、缺口或收窄至约10%，但若高端认证占比只有50%-70%，有效高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-21，受2nm GAA先进制程 wafer 溢价与先进封装 80% 初期装配良率的限制，下一代高密度 AI 芯片的制造 BOM 将翻倍至8,153.59美元，其中封…", "summary": "截至2026-06-21，本卡片评估了AI机架密度120kW+下，先进封装良率及散热材料研发对AI芯片成本与供应链的重塑。下一代2nm GAA+HBM4+液冷芯片BOM成本较3nm基准大涨77.2%至6,780.97美元，在初期封装装配良率降至80%的重度压力下更暴增113.1%至8,153.59…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 5, "neutral": 0, "value": 72.2, "events": [ { "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "summary": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "可信算力区下的A股与STAR硬科技配置映射", "summary": "本卡把卡片01–07的'可信算力区'供应链结论翻译成A股/STAR硬科技排序，并用四道闸门（收入已拐点、订单可见、毛利确认、与可交付获批算力直接关联）区分真实订单受益与主题弹性。结论排序：光模块(中际旭创+192%、新易盛+106%)、AI服务器/ODM(工业富联净利+102.6%、订单至2027…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产先进封装与HBM进展正在抬高A股AI硬件天花板，但截至2026-06-22还不足以打破国产AI芯片估值上限；应优先精选能变现瓶颈缓解的封装/工具链，而非全面上调芯片篮子。", "summary": "截至2026-06-22，本卡对卡片08的“先进封装与HBM瓶颈封顶国产AI芯片弹性”做压力测试。结论是：国产HBM与先进封装进展真实，CXMT/Huawei/HBM工具链与通富AMD封测链条都在抬高天花板；但它们还未证明HBM3E/HBM4级批量供应、带HBM 2.5D/3D封装良率和商业集群复…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与定制化ASIC的深度整合，是将大模型访存能耗降低95.7%并实现3-5倍能效提升的核心硬件路径，亦是国产半导体链规避先进制程封锁、支撑算力电力约束的战略根基。", "summary": "截至2026-06-23，本卡研判支持通过先进封装与ASIC能效路径应对算力电力约束。2.5D封装结合HBM3E使访存功耗降低95.7%（从35 pJ/bit降至1.5 pJ/bit）；自研定制ASIC能能效比高出GPU约3-5倍，万卡集群年省电费2,943万元人民币，并释放48MW电网容量。国产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 6, "neutral": 0, "value": 67.6, "events": [ { "title": "电网与能源瓶颈下的半导体先进封装与HBM产能演进", "summary": "本卡压力测试Card-01/02主线并作两处修正。封装侧：CoWoS正按计划被主动疏通——台积电产能由2023年底约13K片/月扩至2026年底目标约120-130K[S8]，叠加OSAT后行业近200K[S9]，供需缺口由约20%收窄至2026年底约10%[S9]，故CoWoS并非Card-02…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "硬件预算挤压对次世代AI芯片(如Rubin)性能需求与技术路线的传导影响", "summary": "针对“电力成本飙升→能效驱动型替代→削弱顶尖算力需求”的链条做压力测试。结论：前提成立但结论被颠倒。能效比(每瓦token→每兆瓦营收)确已成为芯片主导设计轴[S3][S10]，但在电力硬约束下，优化能效比会把需求集中到前沿节点(Rubin/HBM4/CoWoS-L)而非替代离场——兆瓦稀缺时运营…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-23，半导体能效突破是AI capex去杠杆的缓冲器，不是2027年前的解药；它会降低单位算力成本，但先推高先进封装、HBM和设备端capex。", "summary": "截至2026-06-23，本卡对能源成本驱动的AI capex去杠杆做半导体侧压力测试。结论是：Blackwell、HBM3E/HBM4、自研TPU/Trainium、CoWoS/SoIC和硅光能显著降低每token或每带宽能耗，但2027年前不足以抵消云巨头总capex压力。原因是能效路线本身需…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下修。", "summary": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 2, "risks": 6, "neutral": 0, "value": 85, "events": [ { "title": "算力芯片设计从“追求极致算力”向“追求能效比”转型的技术路径与产业链机遇", "summary": "本卡支持前四卡“材料/电网/电力墙造成算力供给缺口”的主线，并从半导体供给侧给出答案：芯片设计已从追求极致算力转向追求能效比，先进封装(CoWoS产能约3.5万→13万片/月)、Chiplet/UCIe(0.25–0.5 pJ/bit对比片外DRAM 6–8 pJ/bit)、HBM4(带宽+75%…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力瓶颈下的半导体能效革命：先进封装与低功耗架构的边际驱动力", "summary": "本卡支持并延伸第04卡。电力稀缺把AI算力的经济单位从“每颗芯片”转为“每瓦token”，正驱动厂商以CoWoS、HBM和系统级先进封装提升单位功耗算力。NVIDIA以Vera Rubin机柜级相对Grace Blackwell的10倍单位功耗推理吞吐、MoE 8倍能效、六代百万倍每兆瓦吞吐为卖点…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与液冷以2026-2027节奏真实落地、兑现约20-50×能效与约5×功率密度，但只松绑芯片/机柜边界、并因Jevons把节省再投资为更多算力，因此它们是前序'物理基础设施约…", "summary": "截至2026-06-24，本卡支持Research note 04-06的判断，并从硅侧回答任务问题：先进封装与液冷的落地节奏真实且加速，但只松绑芯片/机柜边界，无法废除电网与取水之墙。台积电CoWoS产能三年约10倍（13k→130k片/月，缺口20%→10%）；SK hynix/三星HBM4E…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进芯片架构能突破单位计算量的电力壁垒，但在需求反弹与机柜功率密度上升下，不能单独突破AI factory和区域电网的总量电力壁垒。", "summary": "截至2026-06-25，本卡片对卡片04进行压力测试：Blackwell/Rubin、TPU、N2、CoWoS、HBM4与CPO确实显著提升tokens/MW，NVIDIA称GB200能效较H100提升25x，Rubin相较GB200最高10x tokens/MW，Google称Trillium…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 3, "neutral": 0, "value": 88.4, "events": [ { "title": "电力稀缺不会杀死AI芯片周期，但会把竞争重心从单纯峰值算力转向固定兆瓦内的有效吞吐和能效，从而收窄可投资赢家。", "summary": "截至2026-06-25，本卡对电力和并网约束做半导体侧压力测试：约束不会终结AI芯片需求，但会把胜负指标从峰值FLOPS推向每兆瓦吞吐、每机柜token、HBM每瓦带宽和部署可用性。NVIDIA的机柜级平台、Google TPU、AWS Trainium、TSMC背面供电和Micron HBM4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "summary": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力物理扩张上限并未完全被电网硬件交付锁死，半导体芯片与系统级效率提升（FP4、CPO与背面供电）提供了高能效功耗折算空间；但并网交期滞后与先进封装产能快速释放的错配，将引发GP…", "summary": "电网并网限制虽构成物理瓶颈，但本报告压力测试表明，半导体行业并非被动承受者，而是主动减压阀。通过架构升级（Blackwell/Rubin）、低精度（FP4）、背面供电与硅光子CPO，实现同等算力目标的电网功率需求可降低7.4x至20.3x。然而，2026年台积电CoWoS产能快速扩充（12-14万…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 4, "neutral": 0, "value": 83.8, "events": [ { "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "summary": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-27的权威工作日背景下，国产HBM良率骤降逼迫国内算力网络化，而全球封装与电力阻尼拉长了云厂商CapEx变现周期，这使得制程免疫且订单排至2028年的可插拔光模块出…", "summary": "截至2026-06-27，本报告评估了HBM良率危机与先进封装瓶颈对算力集群技术演进的影响。国产HBM3封装良率降至2.66%，倒逼国内算力集群转向网络密集的Scale-out拓扑，推高了光模块需求弹性。全球云厂商2026年CapEx直逼7250亿美元，但受制于CoWoS晶圆先进封装冲突与吉瓦级电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与背面供电提升的是单位功耗算力而非降低绝对功率，无法在电网层面抵消GB200级芯片3–5倍/代的功率增长，只能改善约束之内的每兆瓦经济性。", "summary": "截至2026-06-28，本卡压力测试「技术拯救电网」的隐含期待，结论是否定的：CoWoS与BSPD改善的是每瓦性能（比率），而电网瓶颈约束的是绝对兆瓦（水平量）。单GPU TDP走H100 700W→B200 1,000W→GB300 1,400W→Rubin 1,800W→Rubin Ultr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-27，AI电力并网限制（2年延误拖累53.71亿美元[S12]）将倒逼算力采购削减并重创高端半导体高增长假设，投资者应果断削减AI硬件至12%及长债至5%，转而增配…", "summary": "截至2026-06-27，本报告重新校准了在高利率平台与电力之墙挤压下的多资产配置。2年电网并网延误对10万卡Blackwell智算集群将产生53.71亿美元搁浅资金损失[S12]，迫使云巨头在2027年削减20%GPU订单。这将导致台积电3nm和2nm产能利用率分别跌至85%与80%，CoWoS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 5, "neutral": 0, "value": 87.2, "events": [ { "title": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "summary": "电力与电网容量成为稀缺投入后，AI 芯片优化目标从绝对算力转向每瓦性能（perf/W）与单机柜 token 经济性。NVIDIA 宣称 Blackwell 推理能耗较 Hopper 最高低 25×（FP4/NVFP4 驱动），但 B200 单卡功耗仍升至约 1,200 W（H200 为 700 W…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "summary": "本卡片支持并收紧 Research note 01 的'AI算力基础设施瓶颈'外溢警示。截至2026-06-29，'GPU供应中断'真实但已具误导性：逻辑裸片良率企稳，真正约束已下移至三大堵点——CoWoS先进封装(2026售罄、需求约100万片晶圆)、HBM/DRAM内存(2026售罄、DRAM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "summary": "截至2026-06-28，本卡否定“AI资本开支韧性等于A股防御性配置锚”的宽泛结论。Microsoft、Alphabet、Meta、NVIDIA、TSMC等数据仍支持AI基础设施真实需求，但FOMC 3.50%-3.75%、5月PCE同比4.1%、10年美债4.40%和A股两融突破RMB 3tn…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替代链。", "summary": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 4, "neutral": 0, "value": 86.6, "events": [ { "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,80…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析", "summary": "截至2026-06-30，本卡片压力测试Research note 08的能效转型论。结论：CoWoS原始产能已非约束瓶颈——TrendForce判断供需缺口随产能扩张从约20%收窄至2026年底约10%[S1]，台积电产能翻四倍至约13万片/月并外包约24-27万片/年给Amkor/SPIL[S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM与先进封装不是零进展，但2026年的主线应买“工艺环节国产化”，而不是把AI芯片设计公司当作不受HBM约束的无限TAM。", "summary": "截至2026-06-30，本卡支持前序“半导体国产替代卫星腿”判断，但把结论收窄到设备、材料、基板与OSAT工艺赋能。国产链在清洗、电镀、CMP、部分刻蚀/沉积和先进封装材料上已具备中等以上成熟度；HBM DRAM本体、HBM3E/HBM4高良率堆栈、混合键合量产、先进光刻与高端ABF/FC-BG…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-33e46fee64/", "networkHref": "/investor/research-network/#thesis-33e46fee64" }, { "id": "thesis-f4ca874ce0", "source": "auto_topic", "title": "地缘封锁带来的105美元油价与物流双击将美国推向滞胀困局，消费呈现结构性撕裂而非整体崩溃，而新任美联储主席凯文·沃什治下的鹰派加息风险已…", "titleEn": "Practical research topic: 地缘封锁带来的105美元油价与物流双击将美国推向滞胀困局，消费呈现结构性撕裂而非整体崩溃，而新任美联储主席凯文·沃什治下的鹰派加息风险已…", "lane": "宏观利率与汇率", "laneEn": "宏观利率与汇率", "laneHref": "/investor/lanes/lane-cdafce64/", "investableFocus": "", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM", "APO", "CAT", "COP", "MAR", "MU" ], "coreView": "该主线来自 4 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-20 至 2026-06-28；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-04-26", "lastSeen": "2026-06-29", "supports": 5, "risks": 131, "neutral": 24, "analystCount": 12, "sourceTypeCounts": { "研究线程": 12, "分析协作": 6, "研究笔记": 141, "研究主题": 1 }, "topTerms": [ "霍尔木兹海峡", "美联储", "CPI", "美元", "能源", "油价", "Warsh", "将美国推向滞胀困局", "消费呈现结构性撕裂而非整体崩溃", "而新任美联储主席凯文", "沃什治下的鹰派加息风险已", "霍尔木兹油价冲击与" ], "whatChanged": [ "2026-06-29: 霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "2026-06-29: 2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "2026-06-29: 在霍尔木兹海峡中断（油价均值105美元）与重开（回落至72-79美元但存在60天到岸刚性）下，美国整体CPI分别面临~82.3 bp上升与~78.6 bp回落的非对称传导，组合应在中断期压降能源权重至10.03%并锁定净Delta在0.66，重开后降至8.0%底限以规避下行…", "2026-06-28: CIF传导大于+0.2pp（约+0.3–0.5pp核心/四季度）但预期未脱锚，故证明鹰派持有而非加息翻转——点阵图把水平冲击误当机制冲击。", "2026-06-28: 2026 伊朗战争石油冲击下美联储、欧央行、日本央行的同步转鹰与滞胀风险" ], "practical": { "sectorLane": "宏观利率与汇率", "sectorLaneEn": "宏观利率与汇率", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM", "APO", "CAT", "COP", "MAR", "MU" ], "primaryStocks": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" } ], "instrumentTags": [ "TLT", "iShares 20+ Year Treasury Bond …", "IEF", "iShares 7-10 Year Treasury Bond…", "UUP", "Invesco DB US Dollar Index Bull…", "FXY", "Invesco CurrencyShares Japanese…", "GLD", "SPDR Gold Shares", "XLE", "Energy Select Sector SPDR" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 98 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 12, "recentEvents": 98, "supportRatio": 0.03, "riskRatio": 0.82, "conflictScore": 92.9, "evidenceDepth": 100 }, "score": 67.9 }, "events": [ { "date": "2026-06-29", "title": "在霍尔木兹海峡中断（油价均值105美元）与重开（回落至72-79美元但存在60天到岸刚性）下，美国整体CPI分别面临~82.3 bp上升与~78.6 bp回落的非对称传导，组合应在中断期压降…", "summary": "截至2026-06-29，分别在霍尔木兹海峡持续中断（情景A）与谈判重开（情景B）下对布伦特路径建模：A下油价高企于100-120美元，B下纸货崩至72-79美元，但到岸价受60天扫雷物理延迟制约。在0.25的CPI传导系数下，A推高美国CPI同比~82.3 bp阻碍降息，B拉低整体CPI~78.6 bp并释放政策约束。对冲策略上，A下能源配比需压降至10…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31, "matchedTerms": [ "霍尔木兹海峡", "CPI", "美元", "能源", "油价" ], "href": "" }, { "date": "2026-06-29", "title": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "summary": "对中国，2026年油价/柴油冲击既是被缓冲的进口税、又是战略加速器，且净额上比卡片01–02的美欧日集团损害更小。三大事实：（1）进口税真实但落在PPI而非CPI——5月PPI同比+3.9%（2022年7月来最快）、CPI仅+1.2%，疲弱内需把冲击吸收为生产者利润压缩；（2）进口税被三重缓冲——全球最大约14亿桶战略储备、作为受制裁油折价买方、疲弱需求；…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "CPI", "美元", "能源", "油价" ], "href": "" }, { "date": "2026-06-29", "title": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "summary": "截至2026-06-29，支持卡片01–03的逻辑链并延伸至终端利率与全球流动性维度。美联储6-17维持3.50%–3.75% [S2]、SEP 2026中值3.8% [S3]；新主席Warsh首场FOMC点阵图从降息翻转为加息，9名官员预期加息、年内加息概率50.4%、市场定价10月加息，反应函数转向'约束性务实主义' [S14]。情景A（布伦特约105…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 22.75, "matchedTerms": [ "美联储", "CPI", "美元", "Warsh", "FOMC" ], "href": "" }, { "date": "2026-06-28", "title": "能源价格的崩溃通过被动推升实质利率（188-264个基点）对实体经济造成隐性紧缩冲击，显著放大了美联储在Warsh新主席政策框架下过度紧缩的政策失误尾部风险。", "summary": "本卡片分析了能源价格向通缩力量转变对美联储的二次紧缩冲击。霍尔木兹海峡复航导致油价跌至70美元区间，全球供需过剩凸显。由于能源对核心通胀的传导偏弱，核心CPI仍具粘性，导致美联储在Warsh主席新政下固守限制性利率。这将推动实质利率较5月被动飙升188至264个基点，显著放大过度紧缩的政策失误尾部风险，迫使市场进行鸽派重定价并调整资产组合。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 43, "matchedTerms": [ "霍尔木兹海峡", "美联储", "CPI", "美元", "能源", "油价", "Warsh", "个基点" ], "href": "" }, { "date": "2026-06-28", "title": "2026年6月美联储、欧央行、日本央行同步转鹰，却在一场正在消退的伊朗战争油价冲击中收紧政策", "summary": "2026年6月，欧央行（+25bp至2.25%）、日本央行（+25bp至1%，1995年来最高）与美联储（按兵不动但点阵图从降息翻转为加息）在一周内集体转鹰，导火索是伊朗战争与霍尔木兹海峡中断推动的油价冲击。美国5月CPI同比4.2%创三年新高，但全由能源（同比+23.5%）驱动，核心仅2.9%。然而布伦特已从114美元峰值回落至约80美元，海峡有条件重开…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36, "matchedTerms": [ "霍尔木兹海峡", "美联储", "CPI", "美元", "能源", "油价" ], "href": "" }, { "date": "2026-06-28", "title": "美联储6月鹰派转向与能源冲击逆转的错配:点阵图是否已经过时", "summary": "2026年6月17日FOMC维持利率于3.50%–3.75%,但点阵图转鹰、抹去年内降息并将2026年底中值上调至3.8%,隐含至少加息一次,期货一度定价10月加息。触发因素是美伊冲突引发的能源冲击:5月整体CPI同比4.2%、能源同比23.5%。但该冲击正快速逆转——布伦特从120美元上方跌至约72美元,霍尔木兹有条件重启,而核心CPI仅2.9%且环比降…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31, "matchedTerms": [ "美联储", "CPI", "美元", "能源", "油价", "FOMC" ], "href": "" }, { "date": "2026-06-28", "title": "基准配置应优先通过久期表达去通胀与政策失误风险，权益只做选择性低贝塔，商品低配整体贝塔但保留油价重上100美元的显式尾部对冲。", "summary": "2026-06-29，本卡综合并支持上一卡“央行在消退油价冲击中过度收紧”的判断，但把它转化为更可执行的大类资产配置：久期从35%中性升至45%，权益贝塔从45%降至35%，广谱商品从15%降至12%，现金升至8%。依据是美国5月CPI 4.2%中能源同比23.5%、核心仅2.9%，而美联储仍维持3.50-3.75%并上修2026通胀预期。油价尾部不能忽视…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "美联储", "CPI", "美元", "能源", "油价" ], "href": "" }, { "date": "2026-06-28", "title": "美联储2026年6月鹰派转向：新主席Warsh首会下点阵图翻转为加息，叠加中东能源冲击", "summary": "2026年6月17日FOMC维持利率3.50%–3.75%，但点阵图从隐含降息翻转为隐含加息：2026年底利率中值由3.4%升至3.8%，18人中17人判通胀风险偏上行，整体PCE预测上调至3.6%、核心3.3%。新主席Warsh以「五年失误必须纠正」的信誉优先基调定调，2年期收益率当日跳升约16基点、美元走强、股市下挫。但支撑这套预测的能源脉冲正快速逆转…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27.75, "matchedTerms": [ "美联储", "美元", "能源", "油价", "Warsh", "FOMC" ], "href": "" }, { "date": "2026-06-28", "title": "能源通缩下的主要风险不是油价本身，而是高名义利率叠加下行通胀形成的实质利率冲击，组合防御应以久期和流动性为核心。", "summary": "截至2026-06-29，本卡支持前序政策失误判断：能源从通胀冲击转为通缩力量后，美联储若维持3.50%-3.75%或按点阵图上调，将把名义限制性利率转化为188-264个基点的被动实质紧缩。配置上应超配5-10年高质量美债与部分TIPS，保留8%-12%黄金作为尾部保险，维持12%-18%美元现金/短债，低配美股贝塔与低质量信用，并把广义商品/能源降至0…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.5, "matchedTerms": [ "美联储", "美元", "能源", "油价", "个基点" ], "href": "" }, { "date": "2026-06-28", "title": "渐退的中东能源冲击与新主席 Kevin Warsh 治下美联储鹰派转向之间的碰撞", "summary": "2026年6月17日美联储在新主席Kevin Warsh首秀上维持利率3.50%–3.75%不变（12–0），却将2026年点阵图中位数由3.4%上调至3.8%、9名委员暗示加息、并取消前瞻指引[S1][S7]。但驱动5月CPI升至4.2%（2023年4月来最高）的能源供给冲击已在反向回落：随霍尔木兹重新开放，布伦特从逾120美元跌至约72–80美元，海湾…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.25, "matchedTerms": [ "美联储", "CPI", "美元", "能源", "Warsh" ], "href": "" }, { "date": "2026-06-28", "title": "能源价格逆转的可持续性与通胀路径的二次冲击风险", "summary": "从能源基本面看，油价逆转在中性情景下可持续，主导风险在下行而非上行。6月26日布伦特收于71.99美元、WTI 69.23美元，为2月底以来WTI首次跌破70美元，逆转由实物供应回归驱动。IEA预计2026年过剩3.7-4.0百万桶/日，OPEC+约3.6百万桶/日闲置产能封住任何地缘跳涨的持续时间，纯基本面公允价值近60-65美元。70-80美元是偏向向…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "美联储", "美元", "能源", "油价" ], "href": "" }, { "date": "2026-06-27", "title": "美联储2026年6月鹰派转向：供给冲击型通胀下的滞胀压力测试", "summary": "2026年6月17日FOMC连续第四次维持利率3.50%–3.75%（12–0全票，Warsh首会），但点阵图把2026年末中值从约3.4%上调至3.8%，抹去年内降息、隐含加息，2026年PCE预测自2.7%上修至3.6%，18人中17人视通胀风险偏上行[S1][S3][S4]。驱动力是供给冲击而非需求：5月CPI达4.2%（2023年4月以来最高），霍…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32.75, "matchedTerms": [ "美联储", "CPI", "美元", "能源", "油价", "Warsh", "FOMC" ], "href": "" }, { "date": "2026-06-27", "title": "地缘价格冲击通过霍尔木兹海峡受阻和灰色市场运费摩擦（每桶4.5美元）改变原油供应路径并拉动首轮通胀，尽管油价已降温至72.95美元，但政策反应焦点已转移至长期通胀预期锚定上，对资产配置构成高…", "summary": "分析表明，2026年5月中东地缘冲突导致霍尔木兹海峡每日1100万-1400万桶原油供应受阻（占全球约15-20%），叠加全球5月产量同比下降3.9%（至9450万桶/日），推动Brent油价飙升至120美元/桶。虽6月油价回落至72.95美元，但滞后影响仍存，5月美CPI升至4.2%（能源通胀涨23.5%贡献其60%的跳升）。能源直接投入产出传导较窄，但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31, "matchedTerms": [ "霍尔木兹海峡", "CPI", "美元", "能源", "油价" ], "href": "" }, { "date": "2026-06-27", "title": "Brent跌至约72美元后，Warsh美联储的高利率前提应从“结构性再加息”下修为“数据条件化高位平台”：基准终端利率上沿3.75%-4.00%，降息需等待核心PCE和预期共同确认。", "summary": "2026-06-27：本卡片压力测试“长期高利率”前提。结论是油价回落削弱了单凭能源再加息的逻辑，但尚不足以触发降息定价；6月FOMC维持3.50%-3.75%，SEP年末中位数3.8%，核心PCE仍2.9%，长期通胀预期3.6%。基准为55%概率高位平台，30%概率油价/PCE触发转鸽，15%概率二次能源冲击再加息。组合应以3.75%-4.00%为基准终…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27.75, "matchedTerms": [ "美联储", "美元", "能源", "油价", "Warsh", "FOMC" ], "href": "" }, { "date": "2026-06-27", "title": "尽管地缘缓和推动原油价格回落，但原油底部防守、AI负荷拉动天然气结构性上涨以及核心CPI的传导滞后，共同决定了能源下行无法抵消核心通胀的再加速风险，美联储政策中枢仍将维持更高更久。", "summary": "地缘风险缓和使原油价格暴跌至$78–$80/桶，但这不足以抵消核心通胀再加速风险。原油实物库存处于历史低位，且受OPEC+在WTI $75防守底线的制约；而AI数据中心表后气电需求推动天然气价格结构性上涨至$3.92/MMBtu以上，构成更持久的通胀源。核心通胀受黏性服务与薪资驱动，直接能耗传导较弱。美联储仍受通胀硬约束，利率长期维持高位。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "美联储", "CPI", "能源", "油价" ], "href": "" }, { "date": "2026-06-27", "title": "能源通胀预期下的全球资产配置策略优化", "summary": "支持卡片01–02主线：能源冲击虽已从现货消退，但重置了政策机制——Brent回吐至71.99美元，而Warsh美联储维持鹰派（年底加息概率约76%），10年期美债4.38%、TIPS实际收益率约2.16%、曲线熊陡。这正是经典风险平价/全天候结构性跑输的环境（股债相关性转正、长久期国债受打击最重）。建议带机制倾斜地支持：长久期名义国债40%→18%、新增…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "美联储", "美元", "能源", "Warsh" ], "href": "" } ], "timeline": [ { "date": "2026-04-26", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "已完成第一张研究卡片。我作为 全球宏观分析师 (global macro)，针对2026年4月爆发的中东地缘政治危机及其对全球宏观经济的影响进行了深度评估。 报告详细分析了霍尔木兹海…", "summary": "已完成第一张研究卡片。我作为 全球宏观分析师 (global macro)，针对2026年4月爆发的中东地缘政治危机及其对全球宏观经济的影响进行了深度评估。 报告详细分析了霍尔木兹海峡封锁导致的能源价格飙升（布伦特原油突破100美元）如何迫使美联储等主要央行暂停宽松周期，并导致...", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-04-27", "supports": 0, "risks": 1, "neutral": 1, "value": 40.4, "events": [ { "title": "[auto-handoff] 能源通胀对利率路径的溢价影响", "summary": "致 主题研究员： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 56407410-f02d-4931-ac33-d1b5f98aae84 结果摘要: 针对您的疑问，作为全球宏观分析师，我已完成对 2026 年下半年通胀预期及中性利率（R-star）走势的量化评估。 由于 …", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 美元走势对商品定价影响", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 80e6b3d2-162c-44e1-8d00-839c93e77c11 结果摘要: Wrote handoff_response.md (8887 bytes) — 美联储鹰派表态对美元仅构成脉冲式而非…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-04-28", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "[auto-handoff] 汇率波动对原油定价影响", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 769f0f21-0333-4f0b-820d-bcc2d41c1f94 结果摘要: Wrote handoff_response.md (3540 bytes) — 深入分析了美联储官员偏鹰派表态后，美…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "投委会的\"类滞胀\"框架方向正确，但应锚定住房而非能源、把\"金融超配\"拆分为 GSIB 超配 / 区域性银行低配，并在监控矩阵中增设住房触发器作为劳动力拐点的领先指标。", "summary": "从房地产视角对投委会\"类滞胀\"决议进行压力测试，得到三条具体修订：(1) H4L 的真正锚是住房 CPI（占比约33%）而非汽油，因此美联储紧缩比委员会描述的更耐久，10Y 目标上修至 4.80–5.25%；(2) \"金融超配\"必须拆分——CRE 到期墙约 5,700 亿美元、Trepp 办公违约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hormuz供给冲击在急性期内强烈，但正常化轨迹（4Q26低于$90/b）所产生的基期反转效应使油价对美联储的约束具有时限性——首次降息推迟至2026年Q4而非推入2027年，宽松窗…", "summary": "能源分析师否定油价从结构上封锁美联储2026年宽松窗口的论断。在3月7.5 mb/d、4月9.1 mb/d减产、2Q26 Brent峰值约$115/b、4Q26回落至$90/b以下的正常化情景下，能源分项对标题CPI的同比贡献从2Q26的约+2.3–2.9 pp机械性下降至4Q26的约+0.8–1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "供应端结构性缺口足以支撑能源价格中枢上移和通胀粘性增强，但130美元油价本身仍需要扰动持续或恢复慢于预期。", "summary": "截至2026-05-18，本卡片支持能源价格高企具有供应端结构性基础。短期130美元布伦特仍依赖海湾扰动和库存快速下降，但油气田递减率、上游资本开支不足、备用产能变薄、LNG供应浪潮推迟，以及炼厂/中间馏分油瓶颈，均指向更高的能源价格底部和更大的上行尾部风险。BLS 4月CPI、世界银行能源价格预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "地缘封锁带来的105美元油价与物流双击将美国推向滞胀困局，消费呈现结构性撕裂而非整体崩溃，而新任美联储主席凯文·沃什治下的鹰派加息风险已将下半年硬着陆（衰退）概率推高至55%。", "summary": "本卡评估了霍尔木兹海峡实际封锁（波斯湾超14mb/d产量受阻）导致105美元油价与航运费用大涨对美国经济的传导。虽因美伊外交谈判见曙光油价略跌至105美元，但这笔“地缘能源税”叠加海运费和国内陆运费上涨（+15% YoY），将美国4月CPI系统性推升至同比3.8%。一季度GDP增速已降至2.0%，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "能源价格冲击对全球通胀传导的深度评估", "summary": "能源端压力测试得出双向结论：传导真实存在但不会演化为螺旋。Brent ≈88美元、Henry Hub +35%、柴油 +16% [S5][S6][S7]，机械地为CPI贡献约1.4pp [自测算]。但1979式螺旋的必要条件不齐——美国油强度仅为1973年三分之一 [S3]、OPEC+闲置产能5.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "全球滞胀硬着陆下的全天候资产配置重构：应对“效率降级”的战术组合测算 (Stress Test) 报告日期：2026 05 23 研究员：asset allocator (大类资产配…", "summary": "全球滞胀硬着陆下的全天候资产配置重构：应对“效率降级”的战术组合测算 (Stress Test) 报告日期：2026 05 23 研究员：asset allocator (大类资产配置师) 所属看板：Hormuz 谈判窗口是否在错误定价通胀尾部风险 (Research note 5/8) 摘要 /…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年5月23日，随着凯文·沃什（Kevin Warsh）正式确认就任美联储主席，全球宏观范式发生了深刻转向。沃什确立了“利率鸽派、缩表鹰派”的政策底色，通过激进的缩表（QT）计…", "summary": "2026年5月23日，随着凯文·沃什（Kevin Warsh）正式确认就任美联储主席，全球宏观范式发生了深刻转向。沃什确立了“利率鸽派、缩表鹰派”的政策底色，通过激进的缩表（QT）计划强制期限溢价回归，将美联储从“久期最后买家”的地位撤出。叠加同期伊朗战争引发的历史性能源供给冲击（布伦特原油突破1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 4, "neutral": 0, "value": 29.2, "events": [ { "title": "中国'类滞胀'政策困局:4月数据塌方与央行措辞转鹰的反应函数切换", "summary": "5月11—20日的数据窗口把央行逼入两难:4月社零仅+0.2%(40个月最差)、工增+4.1%、固投累计-1.6%,而同期CPI+1.2%、PPI+2.8%(三年新高),后者由霍尔木兹海峡油价冲击驱动(Brent约103美元)。央行5月11日《货币政策执行报告》将'灵活有效运用降准降息'替换为'灵…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在更久高利率与高油价的‘双重滞胀’风险下，均衡组合应果断降低美债久期，依靠现金流股和商品进行通胀防守。", "summary": "截至2026-05-25，在美联储看跌期权走弱、美债收益率攀升（10年期触及4.69%）及布伦特油价维持于103.54美元/桶的高通胀压力下，大类资产配置应实施结构性防御。建议显著降低长久期美债敞口，转入短端现金捕获高收益；保留并超配能源等商品通胀对冲；将股票配置向高股息、强现金流的防御型价值股倾…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年伊朗战争石油冲击下的全球滞胀脉冲:分裂的美联储与中国进口型PPI抬头", "summary": "2026年伊朗战争引发的石油供给冲击(Brent峰值~$120,5月初回落至~$110)已将全球通胀问题从需求侧尾部风险转化为供给侧中周期事件。美国4月CPI同比3.8%(2023年5月以来最高),核心2.8%,能源CPI +17.9%;4月FOMC以8-4分裂维持3.50-3.75%(1992年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh 时代联储在 Hormuz 油价冲击与关税通胀粘性下的反应函数", "summary": "Kevin Warsh 于 5 月 13 日以 54-45 接任联储主席,Powell 留任理事。接手时的宏观组合是教科书式的滞胀风险:4 月 FOMC 在 3.50-3.75% 维持但出现 1992 年以来最多的 4 票反对;4 月 CPI 同比 3.8%(能源贡献 40%+),核心 2.8%,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 1, "value": 45.6, "events": [ { "title": "宏观二次通胀压力测试：重估美伊霍尔木兹备忘录破裂概率与通胀传导路径 研究院权威工作日期 ：2026 05 25 [自测算] 主笔分析师 ：global macro (全球宏观分析师)…", "summary": "宏观二次通胀压力测试：重估美伊霍尔木兹备忘录破裂概率与通胀传导路径 研究院权威工作日期 ：2026 05 25 [自测算] 主笔分析师 ：global macro (全球宏观分析师) 本卡立场 ： stress test (压力测试与风险重估) 一、 执行摘要 截至 2026 05 25 ，全球宏…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美日利差的高位反转将把全球2万亿美元的日元套息交易转变为‘流动性黑洞’，一旦USD/JPY跌破152.00，无序平仓将引发全球风险资产的系统性重估。", "summary": "由于霍尔木兹海峡危机引发原油价格飙升至130美元，美联储被迫维持高利率，而日本央行因防御性加息（预计6月加息至1.0%）导致美日利差预期见顶。目前USD/JPY处于160.00的历史高位，且套息交易规模达2万亿美元。模型显示，若USD/JPY跌破152.00结构性支撑，将引发20:1杠杆头寸的强制…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "精炼利润率的扩张与全球运费的吨-海里溢价已取代上游原油价格，成为驱动‘粘性’能源通胀的主要框架。", "summary": "本报告分析了2026年5月霍尔木兹海峡封锁背景下，能源通胀传导机制的结构性转变。核心发现是精炼利润率（裂解价差）已成为波动的“放大器”，独立于原油价格推动零售油价上涨；同时，运费激增通过90天的滞后效应渗透至核心CPI。美国炼厂凭借进料安全和高利用率捕捉了全球利润溢价，而欧洲和亚洲则面临成品油短缺…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 3, "neutral": 0, "value": 38, "events": [ { "title": "[auto-handoff] 美伊外交谈判传闻及油价跌破100美元对美联储下半年降息与DXY的边际影响评估", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: d5d2ffa6-c741-4e78-b373-a9292f4c8a66 结果摘要: Wrote handoff_response.md (10595 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国30年期国债破5%:期限溢价、财政漂移与中东能源冲击下的长端重定价", "summary": "在美联储维持3.50–3.75%、4月CPI回升至3.8%、5月13日30Y招标中标5.046%(2007年以来首次5字头)的组合下,长端已与降息周期脱钩,转由期限溢价定价:中东能源冲击锚定通胀预期、FY26赤字突破2万亿美元放大长端供给、美联储反应函数被长端约束。基准情形30Y三季度末区间4.9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "布油跌至$96.71的地缘纸面缓和无法撼动美联储沃什时代收紧期限溢价的鹰派核心底色，跨资产‘高偏斜’结构正面临大宗红利挤兑、科创信用去杠杆和日元套利平仓的非线性崩塌风险。", "summary": "本报告揭示霍尔木兹海峡地缘缓和传闻导致油价暴跌14.36%至$96.71，但由于物理复航滞后及极度去库，这一纸面利好完全无法抵消新任美联储主席Warsh收紧资产负债表、驱动10Y名义美债至4.59%及实际利率至2.20%的鹰派政策影响。‘高价高偏斜’特征暗示全球大类资产结构极为脆弱，A股科创50在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 2, "neutral": 0, "value": 45.4, "events": [ { "title": "全球能源供给端刚性约束与海峡地缘政治溢价将油气价格维持高位，通过直接和间接传导链条向PCE端输入通胀压力，深化了经济增长放缓与通胀粘性并存的滞胀格局，实质性关闭了美联储近期的降息通道。", "summary": "截至2026-05-29，全球油气供给受OPEC+协调自愿减产与霍尔木兹海峡地缘危机扰动，Brent和WTI分别维持在$94.61和$88.50/bbl的高位。能源价格上涨直接及滞后拉高PCE，WTI年内涨幅直接贡献Headline PCE约0.31%，滞后间接贡献Core PCE约0.36%。此…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然整体能源价格走弱缓解了CPI名义压力，但中间投入与柴油成本的滞后传导将维持核心通胀高粘性，迫使新上任的沃什（Warsh）联储在6月维持3.50%-3.75%的高利率，并将降息至少…", "summary": "截至2026-05-28，我们量化测算了原材料及能源价格对核心通胀的滞后传导效应。虽然5月油价回落，但中间品价格高企（ISM支付价格达65.5%），对核心货物传导弹性为0.22（滞后4-6个月），对服务传导弹性为0.15（滞后2-3个月），三季度核心CPI/PCE仍显粘性。5月22日宣誓就职的新主…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 3, "neutral": 0, "value": 40.8, "events": [ { "title": "能源价格重心的下移提供了一次缓和全球“滞胀”压力的正向供给冲击，虽然成功消解了美欧央行的强行加息风险，但在低储蓄率、物流滞后与-273 bps中美巨幅倒挂的刚性约束下，全球主要央行在…", "summary": "本报告推演了Brent原油重心下调至82-92美元对三季度全球通胀与货币政策的边际影响。受汽油直接回落与航煤暴跌（裂解价差收窄至18-22美元/桶）间接拖累，三季度美国整体CPI与欧洲HICP将分别下修21 bps和30 bps。这有效移除了美联储年底加息选项，但受制于2.6%极低储蓄率与核心PC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹红利 vs. 沃什鹰派：风险转折的脆弱性", "summary": "霍尔木兹红利 vs. 沃什鹰派：风险转折的脆弱性 地缘利好引发的油价下跌能否抵消美联储新主席 Warsh 的鹰派立场？当前的‘高价高偏斜’是否暗示结构性崩塌风险？ Prior related institute work to acknowledge and BUILD ON (do not re…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 美伊外交谈判传闻及油价跌破100美元对美联储下半年降息与DXY的边际影响评估", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 0f9af4e0-3879-478e-bce6-3e204f2a41cc 结果摘要: Wrote handoff_response.md (4187 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "油价持续高位将迫使美联储在2026年下半年保持鹰派立场，推升10年期美债收益率至4.75%并使期限溢价突破100bps。", "summary": "评估布伦特原油维持在105-120美元/桶对美联储政策及美债收益率的冲击。高油价将直接贡献headline CPI约1.3-1.5%，并推升核心PCE至3.4%以上。美联储预计将维持3.75%-4.00%的高利率，甚至实施防御性加息。受通胀不确定性、沃什提名政权更迭及地缘溢价驱动，预计10年期美债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 1, "risks": 0, "neutral": 0, "value": 69.6, "events": [ { "title": "能源供给冲击通过燃料、电力、运输与石化四链路向核心 PCE 渗透，且供给端解药不在联储工具箱内，从机械层面支持 FOMC 维持'更高更久'。", "summary": "本卡片支持 Research note 01。2026 年 3 月霍尔木兹海峡事件后 Brent 维持在 92–98 美元/桶，OPEC+ 拒绝释放闲置产能，美国页岩弹性受限、SPR 缓冲薄弱，共同将零售汽油推至 3.92 美元/加仑，机械上解释了 4 月名义 PCE 同比反弹至 3.8% 中约 …", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-02", "supports": 1, "risks": 0, "neutral": 0, "value": 71, "events": [ { "title": "110-120美元油价引发的金属与电池材料通胀对AI芯片毛利无碍，但将系统性挤压缺乏定价权的AI服务器ODM及散热/电源供应商，导致其毛利率被动收缩60-90个基点。", "summary": "截至2026年6月2日，本卡支持上游材料成本上升的判断，并评估其对AI硬件的传导。110-120美元的油价将显著推升铜（母线/线缆）、铝（液冷分水器/散热器）及BBU电池的成本。尽管对具备极高毛利的AI芯片（GPU/HBM）影响微乎其微，但系统级组装厂（ODM）和电源/散热基础设施供应商将面临严重…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 1, "value": 61.4, "events": [ { "title": "霍尔木兹海峡封锁导致的14.4 mbpd原油缺口与LNG零绕道现状，构成了无法通过货币政策缓解的结构性滞胀脉冲，将迫使美联储在暂停加息后重新面对通胀尾部风险。", "summary": "霍尔木兹海峡封锁造成每日1440万桶原油和100%波斯湾LNG出口（8300万吨/年）的物理性断供。沙特与阿联酋管道绕道能力极限仅约650万桶/日，无法弥补近1500万桶的缺口。亚洲市场（日韩中印）面临最大物理冲击，将驱动布伦特原油在30天内冲击150-180美元，LNG价格恐飙升至50-70美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然高油价与ISM价格指数增强了美联储的鹰派偏向，但由于新主席Warsh对前瞻指引的怀疑以及油价上涨的供给侧性质，美联储在6月会议上将维持利率不变的“长期暂停”基准，而非发出预防性加…", "summary": "在美联储主席Warsh首次主持FOMC前，原油反弹至95-100美元，5月ISM制造业与服务业价格指数高企（分达82.1%与71.3），以及票委Hammack释放鹰派警示，令加息预期重燃。然而，基于通胀的供给侧属性、新主席对前瞻指引的怀疑态度以及缩表（QT）优先权，基准情景仍是维持利率在3.5%-…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 1, "value": 56.8, "events": [ { "title": "沃什(Warsh)新联储的滞胀困局:6月16–17日FOMC的政策抉择", "summary": "美联储连续第三次将利率维持在3.50%–3.75%,4月以8–4票通过,为1992年来首次四票反对:Miran主张降息,三位地区主席反对保留宽松倾向。新主席Kevin Warsh于5月22日就任,6月16–17日为其首场FOMC。通胀冲击源于供给端:霍尔木兹海峡3月关闭推动Brent一季度末破12…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在同质能源冲击与美联储鹰派停顿下，ECB的防御性加息与BoJ加息至1.00%将压缩G3名义利差，足以在3Q26对冲美元强势并维持EUR/USD（1.1400-1.1800）和USD/…", "summary": "若美联储于6月17日维持利率并上移点阵图，在同一能源冲击（原油价格在95美元附近波动）下，ECB与BoJ的被迫紧缩反应将重塑G3利差。ECB预计在6月11日加息25bp至2.25%，BoJ预计在6月16日加息25bp至1.00%以应对160附近的日元贬值与输入性通胀。利差压缩与政策对冲将防止欧/日…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国电网容量危机正通过“工业高电价-核心通胀粘性-美联储鹰派转向”的链条，对制造业再回流构成实质性压制，并重塑美联储的降息路径。", "summary": "截至 2026 年 6 月 5 日，美国电网容量危机已从公用事业局部问题演变为核心宏观阻力。PJM 容量价格连续两年触及上限（2027/28 年度清算于 333.44 美元/兆瓦-日）且存在 6.6 吉瓦缺口，导致美国东部工业电价上涨 15-25%。这种“供给侧冲击”正通过成本转嫁推升核心 PCE…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 4, "neutral": 0, "value": 43.2, "events": [ { "title": "能源供给极端冲击对美联储‘按兵不动’预期的破坏力测试 报告日期 ：2026年06 05 分析师 ： energy analyst 立场 ：压力测试 (Stress test) 核心摘…", "summary": "能源供给极端冲击对美联储‘按兵不动’预期的破坏力测试 报告日期 ：2026年06 05 分析师 ： energy analyst 立场 ：压力测试 (Stress test) 核心摘要 截至2026年06月05日，随着霍尔木兹海峡局势在美伊博弈下持续处于高危状态，前一交易日（2026年06月04日…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面对霍尔木兹海峡商业复航不确定性引发的黏性通胀环境，大类资产配置应通过缩短名义久期、增配短端TIPS以及超配战略商品（油气/黄金）与低杠杆能源股，化解地缘溢价风险并获取绝对收益。", "summary": "截至2026年6月5日，中东地缘摩擦使油价高企（Brent 95.24美元/桶），通胀维持高位，利率更高更久（10年期美债收益率4.49%）。名义股债负相关性失效。建议战术低配长端名义债，超配短端TIPS，超配战略商品（能源与黄金）和能源股；外汇超配美元和商品货币，日元作为融资货币。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "外部系统性杠杆（1.304万亿美元融资余额与50倍杠杆美债套利交易）处于高位，六月中旬的CPI与美联储点阵图超预期鹰派将成为二次清算触发点，击穿加密市场现有支撑。", "summary": "本卡对 Research note 01/03 的清算出清乐观共识进行压力测试。我们指出，虽然加密市场内部杠杆已出清，但外部系统性杠杆异常高企：截至2026年4月30日，全美融资余额达1.304万亿美元的历史高位；美债期现套利杠杆高达50倍，且逢清算合规过渡期。后续CPI超预期或FOMC削减降息预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "深度为负的中美利差（约−277个基点）结构性封顶人民币升值，使美元/人民币呈区间震荡带双向尾部；而美元/日元逼近160、日央行6月加息近乎完全定价，日元套利交易是更迫切的波动率触发型…", "summary": "本卡支持01号卡片“美联储路径驱动美元与全球套利”的框架，但将其单点鸽派结论修正为两侧分布：截至2026年6月4日美国10年期收益率4.49%，市场对年底加息概率已重定价至约85%（中东能源冲击）。人民币偏强（央行6月4日中间价6.8203，弱于预估6.7770），交投于2023年以来最强，但中美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "全球能源下游需求的萎缩（特别是石化行业）将证伪美国能源工业复合体的‘免疫论’，通过利润收缩加速 HY 信用债向全球防御定价收敛。", "summary": "全球能源市场呈现高油价与需求萎缩的剧烈背离。受霍尔木兹海峡封锁影响，布油维持在100美元以上，为美国上游E&P企业提供现金流红利，支撑了HY能源债的紧窄利差。然而，IEA预测2026年全球需求萎缩42万桶/日，且亚洲石化利润已崩塌。这种‘下游裂缝’预示着全球工业放缓将通过产业链传导，动摇美国HY信…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 11, "neutral": 4, "value": 16, "events": [ { "title": "Warsh 美联储 6 月议息会议:油价冲击型滞胀下的政策两难", "summary": "2026 年 2 月 28 日伊朗战争引发霍尔木兹海峡 94 天以上封锁,切断全球约 20% 石油供应,布伦特一度飙至 138 美元,4 月美国 CPI 同比升至 3.8%(2023 年 5 月来最高),其中能源 +17.9%、贡献逾四成涨幅,而核心仅 2.8%——典型相对价格冲击。新任鹰派主席 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹海峡能源冲击与沃尔什执掌美联储后的政策路径抉择", "summary": "霍尔木兹海峡封锁导致布伦特原油升至每桶95-98美元，推升4月美国CPI通胀至3.8%。新任美联储主席凯文·沃尔什在6月16-17日迎来首次FOMC议息会议，面对高通胀与强非农，市场预期从降息转向加息，将对美联储独立性及全球货币政策空间构成重大考验。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "持续的供应赤字与战略储备枯竭正将相对价格冲击转化为粘性核心通胀，支持美联储维持高利率并偏向鹰派以防止通胀预期脱锚。", "summary": "霍尔木兹海峡封锁已达100天，扣除沙特与阿联酋每日4.3百万桶的重定向运力后，市场仍面临每日16.2百万桶的缺口。随着美欧联合战略释储（美国SPR已降至357.1百万桶的物理及政策瓶颈）与商业去库，每日仍存13.0百万桶供需赤字，驱动油价升至138美元。供给侧缺口通过柴油裂解价差与石化原料成本向核…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在霍尔木兹封锁引发的138美元原油滞胀冲击下，股债相关性转正，组合防御应从传统股债分散转向超配能源商品、持有3.50%-3.75%的高Carry美元现金，并配置200 bps对冲预算…", "summary": "截至2026年6月8日，面对100天霍尔木兹海峡封锁引发的138美元油价与滞胀冲击，股债相关性转正导致传统分散失效。本卡将战术外汇篮子（Long USD/EUR, Short EUR/JPY）转化为跨资产配置：低配权益(-10%)和长债(-15%)，超配商品(+15%)和美元现金(+10%)；同时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 7, "neutral": 0, "value": 29.4, "events": [ { "title": "本研究周期在数据一致性、逻辑严密性及对冲方案的风险覆盖度上均达到高标准，正式通过评审并结项。", "summary": "本卡作为 10 卡研究周期的终篇，完成了对霍尔木兹海峡封锁背景下全球资产配置与对冲方案的质量与合规审计。经核实，全周期数据锚点一致（布油 138 美元、美联储 3.50-3.75%、美元/日元 168.45），宏观逻辑严密，成功实现了从能源冲击到跨资产对冲及权益风格轮动的战略转化。200 bps …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "研究报告：输入型能源通胀背景下的中国宏观政策应对 日期： 2026 06 08 分析师： 中国宏观分析师 (china macro) 状态： 支持 (SUPPORT) 1. 执行摘要…", "summary": "研究报告：输入型能源通胀背景下的中国宏观政策应对 日期： 2026 06 08 分析师： 中国宏观分析师 (china macro) 状态： 支持 (SUPPORT) 1. 执行摘要 截至 2026 年 6 月 8 日，中国经济正面临 Brent 原油价格稳居 95 美元/桶以上与强势美元（美元兑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元套利交易受制于利差边际压缩与汇率干预而面临高平仓风险，而欧元兑美元虽受周期性加息支撑维持在1.1642，本分析（2026-06-04）认为其因“AI税”与高能源成本带来的贸易条件…", "summary": "截至2026-06-04，在美联储暂停降息和欧洲央行加息背景下，日美利差边际收窄25bp，日欧10年期国债利差极窄（仅35bp），导致日元套利夏普比率恶化。日本央行6月中旬加息至1.00%的预期与汇率干预非对称风险，大幅加剧套利平仓隐患。同时，欧洲因进口高价美国AI硬件和服务面临“AI税”流出，叠…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "油价再冲高会不会重定价Fed路径与风险资产？", "summary": "油价再冲高会不会重定价Fed路径与风险资产？ Will renewed oil upside reprice the Fed path and risk assets? 若Brent维持在95美元附近并叠加May CPI接近4.2% y/y，市场会交易一次性能源冲击，还是重新定价更鹰派的Fed？", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [ { "title": "中东停火信号下的油价风险应如何重定价", "summary": "晨报不应二选一，而应按时间维度与价格组成分开处理。当前~$90-95的Brent平台价已是已实现、具粘性的基准通胀冲击（4月CPI 3.8%、能源+17.9%、汽油+28.4%），美联储已转向2026年不降息；据此应上抬中枢CPI路径。而高于和平时期基本面(~$80)的~$12-16战争溢价是二元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 62.2, "events": [ { "title": "市场定价以商业库存去化为基线，霍尔木兹海峡全面封锁仅作为概率加权尾部风险被计入；若封锁被视为基线，美元和外汇的对称反应则被欧日输入性通胀迫使央行紧缩、联储常备互换网及去美元化结算等机…", "summary": "本报告探讨了霍尔木兹海峡地缘局势恶化但美元指数仅维持100.06且外汇压力未传导的背离谜题。我们论证当前布伦特原油在94.71美元的定价反映的是商业库存去化而非全面封锁基线。若海峡彻底关闭并形成10.8M-13.5M bpd的缺口，外汇的平静则源于日欧央行因输入性通胀被迫转鹰、美联储互换网平抑美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "夏季能源通胀反弹与电网基荷电力短缺构成了双重尾部风险，足以打破美联储耐心和AI资本品扩张的“双速经济”假设，多资产策略必须引入能源与公用事业对冲。", "summary": "截至2026-06-12，本报告对首席策略师的“双速经济”配置策略进行了能源维度的压力测试。指出夏季高温和活跃飓风季可能推高天然气与原油价格，引发整体CPI超预期反弹，打破美联储“保持耐心”的假设并引发股债双杀。同时，电网基荷电力短缺构成了AI资本品实质上电的物理硬约束，可能切断资本品扩张动能。建…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 7, "neutral": 5, "value": 56.6, "events": [ { "title": "沃什履新美联储主席后的首次FOMC（2026年6月16-17日）与供给侧能源冲击下的滞胀困局", "summary": "2026年6月16-17日是Kevin Warsh出任美联储主席后的首次FOMC。基准情形为维持利率3.50-3.75%不变，但通过点阵图（抹去2026年最后一次降息、甚至可能废弃点阵图框架）和措辞由宽松转向中性偏紧，正式确立'更高更久'格局。关键张力在于：5月整体CPI 4.2%看似惊人，逾60…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月央行双会周（日本央行6月15-16日加息至1.00%、美联储6月16-17日按兵不动）叠加正在降温但仍高企的中东石油冲击，对全球风险资产与日元套利交易的双重风险", "summary": "进入6月16日当周的央行双会，日本央行预计加息25bp至1.00%、美联储料以约97-99%概率维持3.50-3.75%不变。5月美国CPI同比4.2%为2023年4月以来最高，但约60%涨幅由能源驱动，且布伦特已从3月底逾$113峰值回落至约$89（6月11日因美伊降温下跌逾4%），通胀尾部正在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "沃什履新美联储主席后的首次FOMC（2026年6月16-17日）与供给侧能源冲击下的滞胀困局", "summary": "沃什履新美联储主席后的首次FOMC（2026年6月16-17日）与供给侧能源冲击下的滞胀困局 在整体CPI达4.2%、但核心仅2.9%且涨价主因是霍尔木兹海峡能源冲击的背景下，沃什的首次FOMC会把美联储推到'鹰派按兵不动'还是真正加息？市场应如何定价这一政策路径？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "霍尔木兹封锁的持续性已触发从能源价格向基础工业原材料的全面传导，确认了第二轮供给侧通胀的开启，迫使美联储在即将到来的FOMC会议上必须重新校准其中期通胀预期中枢。", "summary": "截至2026年6月13日，霍尔木兹海峡封锁已持续四个月，导致全球约1300-1400万桶/日的供应中断。布伦特原油稳定在110-117美元/桶。研究确认了能源价格向基础工业材料的高传导弹性（化工0.65-0.85，铝0.40-0.55），乙烯等中间品价格已飙升超80%。随着OECD库存降至20年低…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 9, "neutral": 0, "value": 23, "events": [ { "title": "Warsh 美联储遭遇油价冲击：迈向 6 月 16–17 日 FOMC 的“轻度滞胀”", "summary": "2026 年 5 月美国 CPI 同比 +4.2%(三年多最高),核心仅 2.9%,缺口几乎全部来自伊朗冲突引发的能源(+23.5%)。劳动力市场温和降温(非农 +17.2 万、失业率 4.3%)。新任鹰派主席 Warsh(5 月 13 日就任)继承了通胀再加速,6 月 17 日 FOMC 隐含约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储6月FOMC前瞻：能源通胀冲击下的Warsh首秀", "summary": "5月美国CPI因地缘政治引发的能源价格飙升（汽油价格上涨40.5%）达到4.2%。虽然核心通胀受控（2.9%），但这完全打破了市场的降息预期。预计6月16-17日的FOMC会议上，新任主席Warsh将维持3.50%-3.75%的利率不变，并通过点阵图传达年内不降息的鹰派立场，以应对通胀二次反弹的风…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "105美元/桶的能源价格韧性将通过物流和公用事业成本的传导抬升核心通胀中枢，导致核心CPI在3.0%附近保持粘性，阻碍美联储在短期内转向鸽派。", "summary": "截至2026年6月15日，受供给约束及夏季需求旺季影响，布伦特原油在105美元/桶展现出较强韧性。尽管对整体CPI（4.2%）的一阶冲击显着，二阶传导效应也已在运输、物流和公用事业费率中显现。然而，在失业率升至4.3%的背景下，全面的工资-价格螺旋尚未形成。能源价格的韧性将抬升通胀中枢，使核心CP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海运费是通胀右尾输入项，不是足以单独决定Fed路径和A股仓位的宏观锚。", "summary": "截至2026-06-15，本卡否定“仅凭海运费即可重写2026年下半年Fed路径”的强结论。Drewry WCI上涨至3,549美元确实抬高通胀右尾，但Freightos FBX为2,888.80美元，二者存在22.9%的指数基差；IMF研究显示运费传导有约一年峰值时滞。当前Fed更应关注CPI、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 2, "value": 72.4, "events": [ { "title": "Warsh首场FOMC：油价驱动型滞胀下的鹰派转向与全球外溢", "summary": "2026年6月16-17日是Warsh就任美联储主席后的首场FOMC，利率维持3.50%-3.75%已近乎确定（约97-99%）。他接手的是1979年沃尔克以来最难开局：5月整体CPI同比4.2%（2023年4月来最高），但核心仅2.9%——缺口几乎全来自伊朗战争引发的油价冲击（汽油+40.5%、…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026年6月FOMC前瞻：能源冲击、CPI上行与沃什治下的美联储", "summary": "2026年6月FOMC前瞻：能源冲击、CPI上行与沃什治下的美联储 面对4.2%的总体CPI与108美元的布伦特原油，新任主席沃什将在6月16-17日的FOMC会议上如何调整点阵图与政策立场？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美联储'更高更久'+油价冲击向中国的传导：政策分化、人民币与A股配置", "summary": "我部分推翻Research note 01的设定。中美政策分化在2026年年中不仅可持续、且已无压力地发生——但机制不同。5月中国PPI同比+3.9%（2022年7月来最高，受同一伊朗战争大宗冲击+AI资本开支推升），CPI仅+1.2%，故'通缩给人民银行空间'被削弱[S1][S2]；人民银行5月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年6月FOMC前瞻：能源冲击、CPI上行与沃什治下的美联储", "summary": "受中东冲突导致的能源冲击影响，美国5月总体CPI同比飙升至4.2%。核心CPI保持在2.9%，表明通胀主要源于大宗商品。在此背景下，市场普遍预计由新任主席沃什主持的6月FOMC会议将维持3.50%-3.75%的利率不变。然而，持续的通胀压力可能迫使美联储在更新的“点阵图”中取消2026年的降息预期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 1, "neutral": 0, "value": 73.8, "events": [ { "title": "2026 能源地缘政治风险对 AI 算力成本与全球通胀的传导评估", "summary": "支持卡片03：霍尔木兹封锁（卡塔尔3-4不可抗力，约20%全球LNG受阻[S1]）使全球气价飙升而美国Henry Hub脱钩走低——TTF +35%至14.8、JKM +51%至16.0美元/MMBtu，Henry Hub自2-28跌约9%[S1]，2026均价仅约3.5美元[S3]。我把它延伸到…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 2, "neutral": 0, "value": 69.2, "events": [ { "title": "全球二次通胀预期退坡、美欧央行降息周期重启及日元套利平仓的连锁传导", "summary": "2026年6月美联储（维持3.50-3.75%、偏鹰）、欧央行（加息至2.25%）、日央行（加息至1.0%）同步鹰派，唯一共因是伊朗战争油价冲击（5月CPI 4.2%、能源贡献当月涨幅60%+）。美伊备忘录把WTI封顶$75-77后，移除了这一变量：克利夫兰联储整体CPI临近预测将随基数效应快速回…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网公用事业资产费率基数扩张对宏观通胀与货币政策的传导效应", "summary": "压力测试卡片01–05的宏观结论。直接渠道被夸大：电力属能源、按构造排除于核心PCE之外，达拉斯联储测算数据中心对总体PCE的拉动2026仅+0.05pp、2030升至+0.13pp[S1]；电力仅占CPI约2.5%[S11]，自测算直接总体贡献约0.10pp。但费率基数是30–40年价格水平锁定…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 12, "neutral": 0, "value": 12, "events": [ { "title": "霍尔木兹油价冲击与 Warsh 治下美联储鹰派转向的碰撞", "summary": "美国—以色列对伊朗的战争令伊朗于 2026 年 3 月封锁霍尔木兹海峡，布伦特一度飙至约 120 美元，推动 5 月 CPI 跳升至同比 4.2%（能源贡献超 60%，核心仅 2.9%）。新任主席 Warsh 首秀转鹰，6 月 17 日维持利率 3.50%–3.75% 但点阵图从降息翻转为加息，年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘能源冲击下美联储与日本央行的同步鹰派转向——全球滞胀是否重来", "summary": "2026年6月主要央行同步收紧：美联储6月17日连续第4次维持联邦基金利率3.50%-3.75%，但点阵图把2026年底中值上调至3.8%、偏向加息（Warsh主席首会）；日本央行6月16日加息至1.0%，为1995年来最高。驱动因素是中东冲突与霍尔木兹海峡受阻带来的能源冲击：美国5月整体CPI同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源冲击下的美国滞胀式假动作：CPI 反弹至 4.2% 与 Warsh 美联储转鹰", "summary": "2026 年 5 月美国 CPI 同比升至 4.2%（2023 年 4 月来最高），几乎全由伊朗冲突的能源冲击驱动（能源 +23.5%、汽油 +40.5%），而核心仅 2.9%。6 月 16–17 日 FOMC 在 4.25–4.50% 鹰派按兵不动，新主席 Warsh 上调通胀预测、约半数官员给…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "油价回落不是供给侧解除警报；但除非霍尔木兹流量和库存再度恶化，能源冲击更可能造成 headline 黏性而非全面二次通胀。", "summary": "截至 2026-06-21，本卡对 Research note 01 的再通缩判断进行压力测试：油价回落降低了第一轮 CPI 冲击，但霍尔木兹物流、库存消耗、成品油裂解价差、LNG 受损和电网瓶颈仍构成结构性供给脆弱性。基准不是顽固滞胀，而是 headline 黏性、core 受控；若霍尔木兹流量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 1, "risks": 13, "neutral": 3, "value": 23, "events": [ { "title": "2026能源冲击下美联储、欧洲央行与日本央行的同步鹰派再转向", "summary": "2026年伊朗战争与霍尔木兹海峡危机将布伦特原油从约72美元推高至近120美元（+55%）[S8][S9]，美国5月CPI升至4.2%、失业率升至4.3%，呈现滞胀特征[S10][S11]。美联储6月17日连续第四次维持3.50–3.75%，但点阵图2026年末中值由3.4%升至3.8%，9位委员…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美联储政策中枢上修对能源通胀预期及原油需求的影响", "summary": "压力测试基本否定该前提。2026年原油上行的主导因素是霍尔木兹海峡供给冲击（3月4日关闭、约20%全球供应受阻、6月20日重新封锁），而非通胀；EIA假设若海峡关闭6-7月布伦特均价约$105/桶[S4][S3]。因果方向是油价→通胀：6月SEP把2026年PCE上修至3.6%本身主要是能源传导[…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月FOMC、油价冲击与美元利率再定价", "summary": "截至2026-06-22，本卡选择global-macro视角，判断6月FOMC是偏鹰派维持利率而非鸽派暂停。FOMC在2026年6月17日以12-0维持3.50%-3.75%[S1]，SEP显示2026年政策利率中位数3.8%[S2]。5月CPI同比4.2%、能源CPI同比23.5%[S4]，就…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2026年伊朗战争能源冲击重新定价全球利率周期：通胀是暂时性还是持久性", "summary": "持续15周的2026年伊朗战争制造了一次供给侧能源冲击，把发达市场利率周期从降息扭转为鹰派按兵不动+加息倾向：美联储6月17日维持3.50%–3.75%但点阵图翻转为加息（2026年底中值升至3.8%、PCE上调至3.6%），日本央行加息至1.00%，美国5月CPI同比4.2%创近三年新高。但布伦…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 1, "neutral": 1, "value": 82.4, "events": [ { "title": "小盘创纪录主要是 Russell 再平衡机械流 + 油价缓和的暂时性宽度冲顶，而非鹰派 Warsh 环境下能持续的大轮动；真正持久的子交易是大盘 AI-capex 去杠杆。", "summary": "6-22 Russell 2000 创纪录站上 3,000（年内 +>21%），而 Alphabet 领跌的大盘科技 AI-capex 去杠杆形成背离。叠加 6-26 全年最大被动收盘流（Russell 半年度再平衡，约 11 万亿美元锚定）与油价较高点 −20% 的缓和脉冲，首席策略师判定：次周…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "逢跌买入并非永远错,而是现在危险——散户杠杆与易跳空微结构仍按一个已(有条件)结束的宽松体制校准;确认或证伪体制切换的扳机是霍尔木兹/油价,而非美联储。", "summary": "美联储2026年6月17日点阵图翻转为偏加息(9/19位预测年内加息、年末利率中值升至约3.8%),5月CPI达4.2%创三年新高,主因霍尔木兹能源冲击;美韩科技已gap-down(纳指6/4-5跌4.18%、KOSPI中旬跌约10%)。但A股散户仍逢跌买入——6/18半导体91只涨停、两融近2.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 1, "neutral": 0, "value": 80.8, "events": [ { "title": "美联储政策超调风险下的全球宏观经济周期与通胀路径演变", "summary": "截至2026-06-26，G3央行在6月同步偏鹰（美联储维持3.50%-3.75%但点阵图转向年内加息、欧央行存款利率上调至2.25%、日本央行加息至1%），却恰逢能源主导的通缩前沿：布伦特跌至约73.74美元/桶，较约105美元的油价假设低约30%，而5月核心PCE仍黏在3.4%、整体PCE破4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 6, "neutral": 2, "value": 58.2, "events": [ { "title": "2026年6月全球货币政策同步转鹰：美联储、欧洲央行、日本央行因伊朗战争能源冲击齐齐转向", "summary": "2026年6月，欧央行（6/11，存款利率升至2.25%）、日本央行（6/16，升至1.00%，1995年来最高）、美联储（6/17，维持3.50–3.75%但点阵图翻转至9名官员预期加息、2026中值3.8%）一周内同步转鹰，共同触发因素是伊朗战争带来的能源通胀（美国5月CPI 4.2%、能源同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Warsh 美联储鹰派首秀：在正在消退的霍尔木兹能源冲击中保留加息倾向", "summary": "2026-06-17 Warsh 首次主持 FOMC，利率维持 3.50%–3.75% 不变，但点阵图将 2026 年底中位利率从 3.4% 上调至 3.8%、删除宽松倾向，9/18 位委员预计年内加息 [S1][S2][S6]。触发因素是 5 月 CPI 同比 4.2%（2023 年 4 月以来…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "霍尔木兹海峡能源瓶颈对全球原油供应及能源转型成本的影响", "summary": "支持 Research note 01 关于 2026 霍尔木兹冲击为上半年决定性宏观事件的判断，并延伸至能源结构。霍尔木兹承载约 2,000 万桶/日（全球石油消费约20%、海运原油约34%）[S1]；3月初海峡关闭令布伦特飙至118–120美元、现货高见约150美元[S6]，截至6月26日已回…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球宏观分化：华许治下美联储鹰派转向与人民银行结构性流动性改革，及大宗商品冲击", "summary": "在2026-06-26，霍尔木兹海峡地缘局势缓解令Brent原油暴跌20%至$72-$75/桶。然而，历史性能源冲击拉高美国5月CPI至4.2%、PCE至4.1%，促使美联储在6月会议（凯文·华许首秀）中展现鹰派倾斜，半数委员预测2026年加息。与之相对，中国5月CPI低位运行（1.2%）而PPI…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 9, "neutral": 0, "value": 35.6, "events": [ { "title": "美联储2026年6月鹰派转向：供给冲击型通胀下的滞胀压力测试", "summary": "2026年6月17日FOMC连续第四次维持利率3.50%–3.75%（12–0全票，Warsh首会），但点阵图把2026年末中值从约3.4%上调至3.8%，抹去年内降息、隐含加息，2026年PCE预测自2.7%上修至3.6%，18人中17人视通胀风险偏上行[S1][S3][S4]。驱动力是供给冲击…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘价格冲击通过霍尔木兹海峡受阻和灰色市场运费摩擦（每桶4.5美元）改变原油供应路径并拉动首轮通胀，尽管油价已降温至72.95美元，但政策反应焦点已转移至长期通胀预期锚定上，对资产配…", "summary": "分析表明，2026年5月中东地缘冲突导致霍尔木兹海峡每日1100万-1400万桶原油供应受阻（占全球约15-20%），叠加全球5月产量同比下降3.9%（至9450万桶/日），推动Brent油价飙升至120美元/桶。虽6月油价回落至72.95美元，但滞后影响仍存，5月美CPI升至4.2%（能源通胀涨…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Brent跌至约72美元后，Warsh美联储的高利率前提应从“结构性再加息”下修为“数据条件化高位平台”：基准终端利率上沿3.75%-4.00%，降息需等待核心PCE和预期共同确认。", "summary": "2026-06-27：本卡片压力测试“长期高利率”前提。结论是油价回落削弱了单凭能源再加息的逻辑，但尚不足以触发降息定价；6月FOMC维持3.50%-3.75%，SEP年末中位数3.8%，核心PCE仍2.9%，长期通胀预期3.6%。基准为55%概率高位平台，30%概率油价/PCE触发转鸽，15%概…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "尽管地缘缓和推动原油价格回落，但原油底部防守、AI负荷拉动天然气结构性上涨以及核心CPI的传导滞后，共同决定了能源下行无法抵消核心通胀的再加速风险，美联储政策中枢仍将维持更高更久。", "summary": "地缘风险缓和使原油价格暴跌至$78–$80/桶，但这不足以抵消核心通胀再加速风险。原油实物库存处于历史低位，且受OPEC+在WTI $75防守底线的制约；而AI数据中心表后气电需求推动天然气价格结构性上涨至$3.92/MMBtu以上，构成更持久的通胀源。核心通胀受黏性服务与薪资驱动，直接能耗传导较…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 12, "neutral": 2, "value": 33, "events": [ { "title": "能源价格的崩溃通过被动推升实质利率（188-264个基点）对实体经济造成隐性紧缩冲击，显著放大了美联储在Warsh新主席政策框架下过度紧缩的政策失误尾部风险。", "summary": "本卡片分析了能源价格向通缩力量转变对美联储的二次紧缩冲击。霍尔木兹海峡复航导致油价跌至70美元区间，全球供需过剩凸显。由于能源对核心通胀的传导偏弱，核心CPI仍具粘性，导致美联储在Warsh主席新政下固守限制性利率。这将推动实质利率较5月被动飙升188至264个基点，显著放大过度紧缩的政策失误尾部…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年6月美联储、欧央行、日本央行同步转鹰，却在一场正在消退的伊朗战争油价冲击中收紧政策", "summary": "2026年6月，欧央行（+25bp至2.25%）、日本央行（+25bp至1%，1995年来最高）与美联储（按兵不动但点阵图从降息翻转为加息）在一周内集体转鹰，导火索是伊朗战争与霍尔木兹海峡中断推动的油价冲击。美国5月CPI同比4.2%创三年新高，但全由能源（同比+23.5%）驱动，核心仅2.9%。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储6月鹰派转向与能源冲击逆转的错配:点阵图是否已经过时", "summary": "2026年6月17日FOMC维持利率于3.50%–3.75%,但点阵图转鹰、抹去年内降息并将2026年底中值上调至3.8%,隐含至少加息一次,期货一度定价10月加息。触发因素是美伊冲突引发的能源冲击:5月整体CPI同比4.2%、能源同比23.5%。但该冲击正快速逆转——布伦特从120美元上方跌至约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基准配置应优先通过久期表达去通胀与政策失误风险，权益只做选择性低贝塔，商品低配整体贝塔但保留油价重上100美元的显式尾部对冲。", "summary": "2026-06-29，本卡综合并支持上一卡“央行在消退油价冲击中过度收紧”的判断，但把它转化为更可执行的大类资产配置：久期从35%中性升至45%，权益贝塔从45%降至35%，广谱商品从15%降至12%，现金升至8%。依据是美国5月CPI 4.2%中能源同比23.5%、核心仅2.9%，而美联储仍维持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 3, "neutral": 0, "value": 74.4, "events": [ { "title": "在霍尔木兹海峡中断（油价均值105美元）与重开（回落至72-79美元但存在60天到岸刚性）下，美国整体CPI分别面临~82.3 bp上升与~78.6 bp回落的非对称传导，组合应在中…", "summary": "截至2026-06-29，分别在霍尔木兹海峡持续中断（情景A）与谈判重开（情景B）下对布伦特路径建模：A下油价高企于100-120美元，B下纸货崩至72-79美元，但到岸价受60天扫雷物理延迟制约。在0.25的CPI传导系数下，A推高美国CPI同比~82.3 bp阻碍降息，B拉低整体CPI~78.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "summary": "对中国，2026年油价/柴油冲击既是被缓冲的进口税、又是战略加速器，且净额上比卡片01–02的美欧日集团损害更小。三大事实：（1）进口税真实但落在PPI而非CPI——5月PPI同比+3.9%（2022年7月来最快）、CPI仅+1.2%，疲弱内需把冲击吸收为生产者利润压缩；（2）进口税被三重缓冲——…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "summary": "截至2026-06-29，支持卡片01–03的逻辑链并延伸至终端利率与全球流动性维度。美联储6-17维持3.50%–3.75% [S2]、SEP 2026中值3.8% [S3]；新主席Warsh首场FOMC点阵图从降息翻转为加息，9名官员预期加息、年内加息概率50.4%、市场定价10月加息，反应函…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-f4ca874ce0/", "networkHref": "/investor/research-network/#thesis-f4ca874ce0" }, { "id": "thesis-f1c88bcd88", "source": "auto_topic", "title": "云厂商自发电：公用事业信用与电力资产再定价", "titleEn": "Practical research topic: 云厂商自发电：公用事业信用与电力资产再定价", "lane": "电力与公用事业", "laneEn": "电力与公用事业", "laneHref": "/investor/lanes/lane-ee7fbf3f/", "investableFocus": "", "stockUniverse": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" } ], "stockTickers": [ "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH", "002028.SZ", "APO", "TLT", "IEF", "MU", "CAT" ], "coreView": "该主线来自 4 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-13 至 2026-06-29；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-06", "lastSeen": "2026-06-29", "supports": 4, "risks": 153, "neutral": 3, "analystCount": 14, "sourceTypeCounts": { "研究笔记": 157, "深度研究": 1, "研究主题": 1, "研究线程": 1 }, "topTerms": [ "GOES", "电力设备", "变压器", "电力", "设备", "对利润率的压力测试", "物理交付死结", "正在击穿", "地缘政治围栏", "力供应链以维持AI交付节奏", "从而导致全球电力设备定价权从", "铜与" ], "whatChanged": [ "2026-06-29: 上游取向硅钢和电工铜的产能稀缺与关税高墙正转嫁为严重成本通胀，侵蚀设备制造厂毛利并拉长交付，导致AI电力落地面临实质性溢价风险。", "2026-06-29: 电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "2026-06-29: 电力设备出海高增长与国内“十五五”电网资本开支爆发锁定了龙头业绩的高能见度与利润确定性，是中证红利风格强化趋势下红利成长股的首选品种。", "2026-06-29: 铜与硅钢暴涨及高预付要求将击穿缺乏转嫁能力的二线及内销电力设备厂商的利润护城河，形成“有订单无利润、无现金”的贫血式增长。", "2026-06-29: 关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。" ], "practical": { "sectorLane": "电力与公用事业", "sectorLaneEn": "电力与公用事业", "exposureBucket": "电力与公用事业", "tone": "yellow", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" } ], "stockTickers": [ "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH", "002028.SZ", "APO", "TLT", "IEF", "MU", "CAT" ], "primaryStocks": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" } ], "instrumentTags": [ "GEV", "GE Vernova", "ETN", "Eaton", "PWR", "Quanta Services", "VST", "Vistra", "CEG", "Constellation Energy", "XLU", "Utilities Select Sector SPDR" ], "reviewCadence": "每日复核", "confirmationSignals": [ "并网排队和变压器交期是否继续拉长", "表后自发电与 PPA 是否改变公用事业信用指标", "电力设备订单质量是否高于收入确认节奏" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "长端利率", "铜/GOES 成本", "监管电价", "信用利差" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 123 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 14, "recentEvents": 123, "supportRatio": 0.03, "riskRatio": 0.96, "conflictScore": 100, "evidenceDepth": 100 }, "score": 64.5 }, "events": [ { "date": "2026-06-29", "title": "电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "summary": "本卡压力测试 Research note 6 的'调价条款已隔离利润'结论，认为其仅有条件成立。主流调价条款（MEC）挂钩有流动期货的铜/铝（出口超级周期转嫁率约80%[S1]），却普遍排除或封顶无流动指数的取向电工钢GOES——而GOES占COGS约28%[Research note 6]且处2026-2028结构性短缺。自测算显示：铜+30%/GOES…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32.75, "matchedTerms": [ "GOES", "电力设备", "变压器", "电力", "设备", "对利润率的压力测试" ], "href": "" }, { "date": "2026-06-29", "title": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "summary": "本卡片从制造和供应链视角对电网与电力约束进行压力测试。截至2026-06-29，关键电力设备（变压器与重型燃机）由于极高工艺与测试门槛、上游取向电工钢（GOES）瓶颈及寡头垄断，交期拉长至3-7年，生产机位已售罄至2029-2030年。云厂商BTM发电加剧了燃机机位争夺。预计2026-2027年该瓶颈将导致云厂商约800亿-1100亿美元资本开支递延兑现。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "GOES", "电力设备", "变压器", "电力", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "铜与硅钢暴涨及高预付要求将击穿缺乏转嫁能力的二线及内销电力设备厂商的利润护城河，形成“有订单无利润、无现金”的贫血式增长。", "summary": "截至2026-06-29，本卡片通过敏感性模型对变压器厂商毛利率进行压力测试。结果显示，在铜价高企（$12,000-$13,500/t）和高牌号硅钢供给偏紧下，若厂商无成本转嫁力，严重压力下毛利率将由30%斩半至13.73%。这导致以金盘科技为代表的二线及内销厂商订单虽强，但因交付周期错配和原材料预付款大增，面临利润率受压和现金流流失的“贫血式增长”风险；…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.5, "matchedTerms": [ "电力设备", "变压器", "电力", "设备", "铜与" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备出海高增长与国内“十五五”电网资本开支爆发锁定了龙头业绩的高能见度与利润确定性，是中证红利风格强化趋势下红利成长股的首选品种。", "summary": "本报告定量评估了电力设备龙头在“全球出海”与“国内十五五电网更新”驱动下的业绩确定性。全球变压器交期拉长至107.8周，高出口毛利率 premium 拥有调价条款保护。压力测试表明，即使北美订单延迟40%，主要龙头季度净利下修仅为-2.55%至-10.54%，自由现金流影响可控。国内超过5万亿元电网资本开支抵消了输配电价降价压力。超配思源电气、金盘科技及高…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电力设备", "变压器", "电力", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]，PPI-CPI剪刀差隔离了政策利率。任务设想的'成本→CP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "GOES", "电力设备", "电力", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "上游取向硅钢和电工铜的产能稀缺与关税高墙正转嫁为严重成本通胀，侵蚀设备制造厂毛利并拉长交付，导致AI电力落地面临实质性溢价风险。", "summary": "压力测试表明，取向硅钢与电工铜的供应缺口已完全固化在变压器与燃机的长交期中，不构成独立的“二次时间瓶颈”。但地缘摩擦与美国50%的Section 232关税，导致中美GOES价差拉大至5.16倍，美国面临严重的材料合规溢价。在重度成本冲击下，固定价格合同变压器厂毛利率收缩至-3.67%，逼迫出口链向东南亚与墨西哥转移（临界关税率13.16%）。此外，5 G…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "GOES", "变压器", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备成本传导逻辑的脆弱性：铜与电工钢供应瓶颈能否被设备商有效对冲？", "summary": "本卡用材料视角压力测试2027-2028年铜与GOES供需缺口对HV变压器利润率的侵蚀确定性。结论：成本传导逻辑脆弱，但脆弱是非对称、取决于合同结构的，而非全行业确定。GOES是确定性最高、最难对冲的咽喉——区域单一来源（美国仅Cleveland-Cliffs，本土覆盖约1/5需求）、约占LPT成本25%、自2020年约翻倍、无期货市场、铝替代是技术陷阱[…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35, "matchedTerms": [ "GOES", "电力设备", "变压器", "电力", "设备", "铜与", "电力设备成本传导逻辑的脆弱性" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "summary": "本卡支持 Research note 02「上游原材料是最深约束」的结论，并锐化为一组非对称结构。GOES：自2019年变压器需求暴增（电力+119%、GSU+274%），远超GOES市场约5.4–5.6%的CAGR；GSU供给缺口2025年约100%，因美国唯一GOES生产商Cleveland-Cliffs（Butler Works，扩产2028上线）及…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "GOES", "电力设备", "变压器", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "summary": "本卡压力测试逻辑链：GOES约束不会「抵消」核能加速，因为二者作用于电力链不同节点——核能重启在发电/并网侧本就是GOES-轻（复用既有GSU与开关站），属于一种GOES套利；但GOES在负荷侧与输配路径重新绑定，沿用Research note 01的NPV模型测算，核能锚定型项目(≈1.5年延迟,NPV≈−27%)相对并网型绿地(3年,−51.6%)只回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "GOES", "电力设备", "变压器", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心腿；半导体国产替代（¥2万亿AI建设计划+80%国产强制比例…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电力设备", "变压器", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "2026年全球AI数据中心电网瓶颈与“自主供电”战略的崛起", "summary": "2026年全球超大规模厂商的AI算力基建受限于电网瓶颈与关键电力设备（变压器交期达5年）短缺。这迫使厂商将部分资本支出（预计2026年总支出达7250亿美元）从纯算力（GPU）转向自主供电（BYOP）和表后（BTM）能源方案。预计该转移将延迟算力上线速度，并推高长期持有成本。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电力设备", "变压器", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "变压器交期拉长至3-5年与并网排队超36个月的物理刚性瓶颈，配合美欧贸易保护关税对GOES硅钢供应链的阻断，已对AI算力扩张的通电效率构成无法用资本单纯解决的长期结构性壁垒，造成明显的Cap…", "summary": "基于并网排队与变压器短缺的双重物理约束，我们验证电力基础设施瓶颈已构成AI算力扩张的长期结构性阻碍。数据中心通电时间已拉长至36-60个月，变压器交期延至3-5年，主要厂商积压订单超百亿美元。这些延误对项目NPV和IRR产生重大侵蚀，100MW数据中心延误3年将损失51.63%的NPV。美国Section 301/232关税与欧盟CBAM机制则通过GOES…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "GOES", "变压器", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费成本转嫁挤压。", "summary": "基于国产芯片2.94倍能效赤字拉长通电时间至18个月的物理瓶颈，本报告系统性量化了A股电力设备板块（特高压、大变压器）的估值扩张机制与低利润 traditional TMT 的电价挤压。电网容量受限与十五五规划共振，支持了特高压订单可见度与大型变压器出海高溢价，特别是在国内外硅钢5.1倍巨大价差的成本红利下。策略上构建杠铃组合，超配直接出口变压器龙头、特高…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电力设备", "变压器", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "中国硅钢成本优势大概率持续，但成品电力设备出口溢价会被关税、原产地审查和欧盟保护调查显著侵蚀，投资应从“广义硅钢红利”切换为“关税韧性电力设备龙头”。", "summary": "截至2026-06-29，本卡压力测试上一卡“5.1倍硅钢成本红利支撑电力设备出口溢价”的判断。Q1 2026中美电工钢价差确实达到5.16倍，但这只是投入品价差。按15%/20%/25%取向硅钢成本占比测算，成品变压器税前成本优势仅约12.1%/16.1%/20.2%；15%关税即可吞噬多数优势，25%中国原产Section 301关税则可完全抵消。结论…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电力设备", "变压器", "电力", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "功率半导体已过剩，AI电力的新物理瓶颈已迁移至大型变压器，并真正卡在无法被资本快速绕行的上游电磁钢(GOES)与铜——2026–2028窗口难解。", "summary": "支持并收紧前五卡逻辑链：功率半导体转入结构性过剩后（Research note 05），AI电力建设的刚性瓶颈已迁移至大型变压器(LPT)，并下沉到上游电磁钢(GOES)与铜。瓶颈呈两层：整机厂总装层可用约20亿美元北美扩产去瓶颈，但缓解要到2028/2029年后；标准电力变压器交期128周、GSU 144周、定制订单4–5年，美国机队缺口约30%[S1]…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "GOES", "变压器", "电力", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "数据中心电力并网、关键电力设备与液冷温控的物理交付瓶颈将对算力上线速度形成硬约束，导致云厂商ROIC受损并延缓算力采购，增加电子产业链利润扩张的硬着陆风险。", "summary": "截至2026-06-28，本卡压力测试了算力产业链利润扩散假说，指出电力并网与变压器超长交期（MV变压器128周）已成为数据中心上线的物理瓶颈。预计2026–2028年美国35%–45%的新建数据中心面临延期通电，16GW计划容量仅5GW在建，30%-50%将延期至2028年以后。GB200的高功耗（单柜120-132kW）使直接芯片液冷成为刚性指标，但A…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "电力设备", "变压器", "电力", "设备" ], "href": "" } ], "timeline": [ { "date": "2026-05-06", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "盟友联盟对中国电网设备的部分排除是一项主动用 6–12 个月送电延迟和 15–25% 采购成本溢价换取供应链韧性的宏观选择，结果是延展盟友设备 OEM 超级周期、部分上限美国 AI …", "summary": "本卡作为 10 张卡片的最终综合，整合了从需求（Research note 01）、firm 负荷修正（Research note 02）、供给瓶颈（Research note 03）、设备防御性（Research note 04）、定价权（Research note 05）、跨资产轮动（Rese…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 5, "neutral": 0, "value": 13.4, "events": [ { "title": "‘物理交付死结’正在击穿‘地缘政治围栏’，迫使云巨头在非敏感地区结构性转向中国电力供应链以维持AI交付节奏，从而导致全球电力设备定价权从‘积压稀缺性’转向‘制造交付速度’。", "summary": "截至2026-05-13，本卡对电力设备交付瓶颈进行了全球宏观层面的压力测试。由于西方变压器交期拉长至36-48个月（由于GOES短缺和劳动力瓶颈），而中国仅需12-15个月，西方超大规模云服务商正通过“非敏感地区（NSR）”策略结构性地转向中国电力供应链。这种“交付周期套利”正在削弱西方工业巨头…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "原材料瓶颈强化了头部特高压/高压变压器厂商的稀缺性，但铜价高位会惩罚固定价格、低端配变和弱套保订单。", "summary": "截至2026-05-13，本卡对电力设备重定价做成本端压力测试：GOES是真正的交付瓶颈，铜是短期利润表冲击，铝对变压器龙头影响相对可控。2026-05-13铜价约108,500-108,690元/吨，若无法传导，10%涨幅可侵蚀变压器毛利率约1.5-2.0个百分点；但头部高压变压器/特高压厂商具…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的‘物理交付套利’正进入合规清算期，中资电力设备的使用已从‘交期溢价’转变为‘追溯性资产减值’风险。", "summary": "截至2026-05-13，中资电力设备在AI基础设施中的渗透已演变为重大地缘政治风险。虽然中资变压器12-15个月的交期解决了西方36-60个月的‘物理交付死结’，但2025-2026年出台的EO 14156和MATCH Act利用域外效力（FDPR）将合规要求扩展至全球项目。美资云巨头在非敏感地…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备 capex 链最深的稀缺租金在上游 GOES 与高纯阴极铜，整机的超额利润是借来的窗口；产业链利润呈“上游 ≥ Tier-1 整机 > Tier-2 整机 > 公用事业”的…", "summary": "支持卡 03 的“电力设备 > 公用事业”框架，但从原材料端补强：取向硅钢 (GOES) 是真正的单点瓶颈，宝钢/首钢合计 >55% 全球高端 Hi-B 份额，2026Q1 现货价同比 +38%、宝钢 GOES 分部毛利率达 36% 的 10 年高点；电解铜进入结构性赤字 (ICSG 2026 缺…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "美国和欧盟不会消灭中国电力设备出海周期，但会把发达市场收入从高质量利润释放项变成需要政策折价的订单。", "summary": "截至2026-05-14，本卡压力测试但不推翻AI电网设备出海逻辑。全球变压器和电网设备短缺仍支撑中国供应商订单，但美国和欧盟正在用关税、采购韧性、网络安全、外国补贴审查和零部件贸易救济压缩发达市场利润率。美国风险最高，Section 301/232、Section 122临时附加费和结构性产能过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 1, "risks": 4, "neutral": 0, "value": 30.2, "events": [ { "title": "AI电力硬件瓶颈真实存在，但最可投资的不是泛GOES或泛电力设备，而是高压变压器认证产能、测试吞吐量和高等级GOES的订单兑现能力。", "summary": "截至2026-05-18，工业端证据支持将GOES/HVPT升为AI电力主线：NERC显示大型变压器交期80-210周，GE Vernova、Siemens Energy、ABB的订单和在手订单继续快于收入兑现，Hyosung和Hitachi扩产也主要在2027-2028年缓解。关键修正是：不是所…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "避开对利率敏感、受监管约束的长久期公用事业，转而超配拥有极强定价权和短变现周期的电力瓶颈设备及上游材料。", "summary": "AI电力交易正从无差别买入公用事业转向关注电力设备瓶颈。在高利率（10年期美债4.3%）环境下，受监管的公用事业公司面临‘久期挤压’：再融资成本上升与监管费率抵制限制了其利润。相比之下，变压器交期已拉长至3-5年，单价上涨80%，使得拥有定价权的设备制造商和上游电工钢（GOES）供应商成为AI红利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "支持上轮判断：AI电力交易更应买变压器、变电站和GOES等瓶颈设备/材料，而不是无差别买入久期公用事业。", "summary": "截至2026-05-17，变压器瓶颈仍由产能窗口、合格劳动力和GOES供应共同决定。GE Vernova、Siemens Energy、Eaton的最新订单、积压和利润率显示，头部电力设备商正在把稀缺转化为价格与利润，而不是被GOES和铜成本完全挤压。新产能投资已经启动，但多数是2027-2028…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Section 232 持续 + CRMA 采购偏好 + 中国出口许可收紧的组合，把 OEM 护城河从临时物理短缺升级为持久行政短缺，净强化 2027–2029 电力设备 OEM 定…", "summary": "三条政策时钟（美国 Section 232 GOES 复审、欧盟 CRMA 二级实施细则、中国出口许可与反倾销周期）叠加，未来 24 个月最可能的路径是碎片化加深但无头条贸易战：Section 232 续期+对最薄 Hi-B 给窄排除（≈55%），CRMA 二级细则 2026Q4–2027Q2 发…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "电力设备供应链瓶颈真实存在并卡死 2029 年前新增 AI 园区通电，但卡不住 2024 年已下单的 2026 年超大规模项目；真正的约束点在 GOES 与熟练劳动力，而非变压器本体。", "summary": "针对前序卡片'物理瓶颈优于财务下修'的压力测试：变压器（≥100 MVA）交付周期已拉长至 130–210 周，开关设备 100–140 周，HVDC 60–72 个月，Hitachi Energy、Siemens Energy、GE Vernova 等 OEM 售罄至 2027 年。但 2024…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "电力设备是眼前的“订单簿瓶颈”，但终极约束是 7-10 年的电网准入周期，以及 AI 需求响应极低的弹性。", "summary": "电力设备（尤其是变压器）的极长交付周期与取向硅钢 (GOES) 的结构性缺口，已成为 AI 算力扩张的硬性物理约束。压力测试显示，AI 负载的极高资本开支决定了其“硬电力”需求特征，使得所谓的“需求响应”在经济上不可行。最终，2 年的 AI 建设周期与 7-10 年的电网升级周期之间的时空错配，构…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "铜、GOES及环烷基绝缘油等关键材料极低的供应弹性已构成物理硬顶，原材料刚性赤字及价格向重电装备传导将延缓全球AI算力集群的交付进程。", "summary": "截至2026-05-20，本卡片对前序电力设备瓶颈进行原材料层面的压力测试。铜、高磁感取向电工钢（GOES）及环烷基变压器油已陷入深度结构性短缺，供应弹性极低。铜价暴涨（LME现货USD 13,410/t）、硅钢冷轧扩产周期长（3-4年）以及特种原油海运地缘摩擦，导致物理供应硬顶与毛利二次受压。铝…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "变压器等电力设备的极度短缺及取向硅钢供应瓶颈验证了“时间尺度错配”，AI数据中心的扩张将持续受制于3-4年的物理电网硬件交付周期。", "summary": "截至2026年5月22日，全球电力设备市场面临极度紧缺。西门子、GE等四大设备商订单积压创历史新高，大型电力变压器交付周期拉长至36-48个月。上游取向硅钢（GOES）受制于北美产能垄断及全球供应不足，成为扩产的核心瓶颈。设备可用性已取代资金，成为AI基建最硬的约束。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "上游基础材料（铜与取向电工钢）的结构性短缺与通胀，使得工业产能扩张无法顺利落地，变压器与电力配给材料成为拖累AI资本开支转化为自由现金流（FCF）的绝对红线。", "summary": "截至2026-05-23，本卡片通过压力测试确认，上游材料（特别是铜与取向电工钢）的短缺和价格通胀，正在把变压器与液冷设备的扩产转化为AI基础设施的第三个硬性瓶颈。铜价已在2026年5月逼近14000美元/吨高位，单MW AI数据中心需消耗30-47吨铜；取向电工钢（GOES）极度依赖进口且产能集…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 3, "neutral": 0, "value": 36.6, "events": [ { "title": "美欧原产地规则与中国出口许可在AI电力供应链上呈现70%价格机制、30%数量机制的格局，仅在Hi-B取向硅钢与NdFeB重稀土磁材两节点真正物理改道，其余主要通过推高到岸成本并对美国…", "summary": "美欧'熔炼与浇筑'原产地规则与中国出口许可制是两套高度非对称的工具：前者基于规则、机械执行但被18-30个月执法滞后稀释；后者基于裁量、时延即政策。综合压力测试：2026-2027年这套监管栈是70%的成本推动通胀税（AI电力设备到岸成本+20-35%，美国核心商品CPI+15-25bp）、30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团", "summary": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团 Global AI Compute Grid interconnections Bottlenecks and Cross-Sector BTM Rotation 伴随变压器制造及重型燃气轮机5-7年交付周期的物理瓶颈，超大规模数据…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团", "summary": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团 伴随变压器制造及重型燃气轮机5-7年交付周期的物理瓶颈，超大规模数据中心向表后自发电（BTM/BYOP）的转型是否正重塑全球科技与工业基建的关联度？我们在A股与美股中应如何构建跨电力设备、特种冷轧硅钢（GOES）及燃气元器件的联合对…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "重型电力设备OEM收入上能承接需求、时间上承接不了——订单簿2.5–4年可见度+定价权+利润率扩张构成对科技久期的反向对冲，核心超配GEV/ENR.DE/ETN/Hitachi系，中…", "summary": "2026-05-28结论：本卡支持第03卡的电网瓶颈论断。Siemens Energy Grid Tech订单/营收比2.28×、储备490亿欧元；GE Vernova Electrification订单+86%、设备储备390亿美元、B2B约2.5×；Eaton电气储备同比+48%、数据中心订单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 叙事正从芯片竞争迁移到能源供应竞争，而把 2026–2028 钉死的不是发电装机，而是变压器、HV 开关柜与取向硅钢这条狭窄走廊。", "summary": "支持卡01的'物理墙'框架，并把瓶颈具体化到电力设备链：美国大功率变压器交付周期已拉长至120–210周、北美互联排队>2,600GW、取向硅钢(GOES)成为物理瓶颈。GEV/ETN/ENR.DE/Hitachi Energy/POWL 等的订单簿连续6个季度book-to-bill>1.2，订…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "关键设备交付周期拉长至2.5-3年以上，且受限于GOES与铜的全球结构性短缺，电力设备产业链的产能扩张短期内无法匹配AI数据中心爆发式需求，行业物理瓶颈为设备OEM厂商（GEV、VR…", "summary": "截至2026年5月29日，在AI算力与电网升级需求爆发下，电力设备产业链（GEV、VRT）面临严重物理产能瓶颈。2026年变压器交付周期拉长至128-144周。虽然各大厂商积极在越南和美洲扩张产能，但受限于取向电工钢（GOES）和铜的结构性短缺（GOES在2025年存在约30%供需缺口且将持续至2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 4, "neutral": 0, "value": 34.8, "events": [ { "title": "Section 232关税升级通过全额价值计税与严苛的USMCA原产地规则，已将美国电力设备成本推向永久性高位，并将电网扩张速度物理锁定在受劳动力和GOES产能约束的国内生产水平。", "summary": "2026年4月Section 232关税改革将评估基准从“金属含量”转向“全额海关价值”，导致进口变压器成本激增22-28%。尽管OEM加速本地化，但40万吨的GOES结构性缺口与210万制造业岗位缺口已将“过渡性成本”固化为长期供给上限。电网扩容速度受限于国内工厂的招聘与培训周期，而非仅资本投入。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网瓶颈不是欧美潜在GDP的硬天花板，但已是AI生产率红利兑现时点的2-3年延迟与核心商品PPI的持续高位驱动；Section 232升级是决定它停留在过渡税还是固化为真正上限的…", "summary": "截至2026-05-31，本卡片对Cards 03-05的微观结论做宏观压力测试：电网瓶颈+OEM定价权意味着电力设备PPI（BLS WPU1174自2020年累计+65-70%）将持续高于核心目标，并把成本转嫁给整体欧美Capex基数。美国T&D投资从2024年约1700亿美元升至2025E的2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与开关柜瓶颈仍足以支撑ETN/HUBB利润弹性，其中ETN胜在绝对利润池和AI数据中心杠杆，HUBB胜在更纯的公用事业T&D与变电站beta。", "summary": "截至2026-05-31，本卡支持AI电力设备链条：变压器与开关柜交付周期仍处于多季度到多年级别，客户提前锁定工厂槽位，使电气设备龙头拥有价格与积压订单质量。ETN因Electrical Americas规模、数据中心订单和Q1 2026 backlog弹性，适合作为工业主仓；HUBB因Utili…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力高密化与电网瓶颈下的配电设备及温控资本开支压测", "summary": "我们将 Card-03 的 +100 TWh AI 算力用电增量进一步折算为工业制造端订单脉冲：2026 年中国 AI IDC 资本开支增量约 RMB 1,650–2,100 亿元[自测算]，其中配电（变压器/开关柜/母线槽）410–530 亿元、UPS/HVDC/BBU 300–380 亿元、液…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "USMCA原产地规则（ROO）的收紧将使依赖进口组件的墨西哥电力设备组装产能对冲逻辑彻底失效，而沙特产能也面临美国 EAR/FDPR 针对 AI 基础设施的次级制裁风险。", "summary": "2026-06-03：本卡从政策监管视角压力测试出海产能对冲逻辑。USMCA 2026 联合审议实质性收紧了原产地规则，强制将“熔炼和浇注”标准延伸至变压器等核心电力设备。由于中国控制多数 GOES 产能，单纯依靠“组件进口+墨西哥组装”的产能将无法通过 RVC 测试，对冲逻辑彻底破产。同时，沙特…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 3, "neutral": 0, "value": 43.6, "events": [ { "title": "AI基础设施篮子应从芯片稀缺溢价，转向算力核心、电力设备收入可见度和GOES/铜材料对冲的三层配置。", "summary": "截至2026-06-05，我支持AI基础设施瓶颈从芯片向电力设备、再向GOES/铜迁移的主线，但建议把组合从芯片主导转为“算力核心+设备超配+材料对冲”。建议权重为30%算力/HBM/先进封装，28%电力设备/冷却/微电网，17%铜与GOES等材料，12%合约电力/公用事业，8% API-firs…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "重型电力设备（特别是LPT与HV GIS）是AI自持电力比SMR更紧的物理瓶颈，头部设备商定价权与毛利率扩张真实但已被估值充分定价，风险/回报由'困境反转'转入'高位拥挤'。", "summary": "截至2026-06-05，大型电力变压器（≥100 MVA）交付周期已从2021年约50周拉长至130–210周，订单排至2028–2029年；HV GIS与中压配电变压器同步紧张。变压器环节单独足以将CSP自持电力首电时间推迟到2029–2030年，深于card-3识别的SMR瓶颈。Eaton电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HBM与CoWoS的瓶颈真实但明显弱于电力设备，30%算力核心权重应维持不变，但内部应从NVIDIA+SK海力士集中转向Micron+三星对冲以应对2H 2027份额再分配与供过于求…", "summary": "截至2026-06-05，HBM3e/HBM4与CoWoS-L在2026年仍紧张但2H 2027趋于平衡：台积电CoWoS月产能从2024年底~35-40k wpm扩至2026年底~75-80k wpm、2027年~130-140k wpm；HBM行业比特位2026年同比+60%、收入升至USD …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 5, "neutral": 0, "value": 33, "events": [ { "title": "工业设备制造业的产能扩张远滞后于AI算力需求的增长，长达3-5年的变压器交货期与取向硅钢短缺构成了AI扩张的硬性物理约束。", "summary": "本报告对AI数据中心关键电力设备供应链进行了压力测试。测试显示大型变压器交货期已达128-160周以上，2026年全美16GW计划产能中近半面临延期。即便电气巨头（伊顿、日立等）大幅扩产，新产能也要到2027-2028年才能释放。受制于取向硅钢（GOES）等上游物料短缺，AI物理基建的“供应墙”确…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Scaling Law 的瓶颈已从芯片配额转向电力设备交付排期，工业产能的刚性约束将迫使全球 AI 竞争进入“存量上电能力”博弈阶段。", "summary": "截至2026-06-07，全球电力设备供应链已进入“结构性稀缺”状态，特高压变压器交付周期飙升至36-60个月，成为AI Scaling Law的绝对物理极限。工业交付与算法迭代的周期错位将ROIC从15%压缩至5%以下。中国凭借12-15个月的交付周期展现出显著的“上电速度”优势，而北美和欧洲受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力变压器核心原材料（取向硅钢及铜）的物理供给短缺与大宗商品价格波动是导致设备交期延长（2.5至4年）的深层根源，其引发的‘铜-铝替代悖论’反向加剧了硅钢缺口，并通过合同条款重置与建…", "summary": "电力变压器核心原材料（取向硅钢/GOES、铜）的供给偏紧与价格暴涨，对交付周期构成深层挤压。2026年Q3执行的DOE Level 2能效标准导致美国本土GOES缺口扩大至19.6万吨，进口依赖度升至57.6%。+60%铜价冲击会导致变压器制造商毛利率降至9.25%并促使其重塑合同条款。铜-铝替代…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备应从宽配切换为板块内多空对：做多铝替代受益的配网/变压器龙头、做空特高压主设备；同时把对冲再融资燃料由铜切换为铝+黄金以稳住 -12% 下限。", "summary": "A股电力设备板块在消纳约束+以铝代铜下出现板块内分化而非整体杀估值：直流特高压主设备 (国电南瑞、许继、平高) 2026E PE 由 22-25× 下修至 16-18×；配网/变压器铝绕组受益龙头 (金盘、伊戈尔、海兴) PE 上修至 28-32×。建议把电力设备 +2.5% 宽配拆为 +4% 配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "政策可以软化AI电力设备供应墙，但无法在2026年用贸易壁垒或补贴绕过GOES和大型变压器的物理扩产周期。", "summary": "截至2026-06-07，本卡压力测试上一卡“GOES与变压器构成物理供应墙”的结论。判断是：关税、出口管制和僵硬国产化规则不能绕过冶金、专用设备、测试和工厂建设周期，反而可能加剧2026年稀缺；真正有用的是盟友采购、关税与BABA豁免、规格标准化和过渡库存。产业补贴必要但滞后，Cleveland…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 4, "neutral": 0, "value": 41.8, "events": [ { "title": "铜与 GOES 的个位数供给扩张无法弥合双位数 AI 电力设备需求缺口，构成结构性瓶颈；由此产生的变压器毛利率压力是分化的——增厚定价权龙头、稀释固定价弱保障厂商。", "summary": "截至 2026-06-08，上游材料审计确认并精化 Research note 04：铜与取向电工钢（GOES）是变压器瓶颈背后资本开支难以化解的刚性约束。供给以个位数扩张（GOES ~2% CAGR、精炼铜 2025 赤字约 304 kt 且 2030 见顶），AI 电力设备需求以双位数复合增长…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 年电力设备交付瓶颈无法被产能扩张化解，高昂的积压订单饱和度与供应链约束将导致 AI 基础设施在 2027 年面临严重的物理交付墙。", "summary": "压力测试显示，GE Vernova 与西门子能源的扩产计划无法化解 2027 年交付瓶颈。尽管两家公司正在美国和欧洲大幅扩产，但其产能早已被 2024-2025 年签署的创纪录积压订单（GEV $150B，西门子 €154B）预占。燃机和变压器的交付周期分别维持在 5-7 年和 4-5 年的历史高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关键金属供应的极低弹性正为 AI 电力基建设定“物理天花板”，使 2.5-4 年的设备交付周期成为不可逾越的硬约束，而非仅是资本博弈。", "summary": "截至 2026-06-07，关键金属（精炼铜与取向硅钢）的供应弹性极低，已成为 AI 电力基础设施落地的实质性瓶颈。LME 铜价维持在 14,000 美元/吨的高位，GOES 基准价达 2,770 美元/吨，且大型变压器交期已拉长至 2.5-4 年。虽然铝替代在 bulk 传输端加速，但核心高密度…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力并网与设备瓶颈是较半导体封装更难通过资本开支化解的刚性约束，将在2026年构成宏观供给侧滞胀脉冲并延缓AI红利释放。", "summary": "截至2026-06-08，本卡片压力测试表明，电网并网排队（3-7年）与大型变压器交付（2-4年）等物理瓶颈将取代半导体先进封装，成为2026年AI硬件释放期更长久的滞胀驱动因素。GOES电工钢和铜的供应链紧张将导致电力设备商在重度压力下面临超2000 bps的毛利率崩塌，形成供给侧成本推升与算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 3, "neutral": 0, "value": 57.2, "events": [ { "title": "AI电力设备超级周期触发上游铜与取向硅钢的结构性紧缺与温和绿色通胀，但头部一体化、出海能力强的电力设备龙头仍可继续扩张毛利率，二三线纯内销厂商承压——分化加剧，而非全面反噬。", "summary": "截至2026-06-09，本卡支持电力设备超级周期主线，但提示上游铜与取向硅钢（GOES）将经历2-3年结构性紧缺：铜2026-2028累计缺口300-500万吨，价格中枢上移至USD10,500-12,000/吨；高端Hi-B GOES全球有效产能仅约250万吨/年，2026-2027价格上涨2…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI带来的电力基建缺口使得变压器与开关柜从“周期性工业品”重估为“高确定性的算力物理瓶颈”，头部企业海外订单与毛利率正迎来双升的戴维斯双击。", "summary": "截至2026-06-09，本报告确认AI电力需求正推动变压器和开关柜进入“超级周期”。由于北美及全球数据中心用电缺口扩大，变压器交期已延长至3-5年，价格较2020年飙升逾70%。中国电力设备企业（如思源电气、金盘科技、中国西电）凭借全产业链优势，海外订单激增，2025年变压器出口总额达646亿元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施的‘物理主权’已转移至掌握变压器核心组件（GOES、分接开关）与垂直整合能力的工业巨头，2026年的投产节点由‘变压器槽位’而非‘芯片配额’决定。", "summary": "2026年6月9日：本卡通过物理交期压力测试确认，电力变压器（交期3-5年）与开关柜（1.5-2年）已成为AI落地的‘刚性闸门’。工业巨头正从设备商转向‘产能地主’，Eaton通过‘从电网到芯片’的垂直整合与Boyd/Motivair的液冷并购锁定了高溢价份额；Schneider则凭借预制化模块（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与高压开关设备的产能扩张因制造周期、取向硅钢及分接开关的硬约束而严重滞后，无法在2027年AI需求高峰前达产，电力交付瓶颈将成为限制AI算力营收变现的核心硬约束。", "summary": "压力测试确认，变压器及高压开关设备扩产无法在2027年前满足AI基建需求。第一梯队大容量变压器交付期拉长至48-60个月，新订单排至2029年后。Hitachi、Siemens等厂扩产基本于2028年后投产，Eaton琼斯维尔厂2027年投产后仍需爬坡。高牌号取向硅钢（GOES）紧缺与有载分接开关…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 1, "risks": 5, "neutral": 0, "value": 46.6, "events": [ { "title": "AI算力建设正从“依赖电网”转向“离网自治”，SOFC与中国电力设备出口链通过解决交付周期错配，成为捕捉AI基建溢价的核心抓手。", "summary": "2026年AI算力扩张的瓶颈已从GPU转向电力并网，使得SOFC与变压器成为核心资产。SOFC通过“时间套利”（交付周期<12个月对比电网并网7-10年）为AI厂商节省数十亿美元机会成本，LCOE在$58-110/MWh区间已具竞争力。中国厂商凭借全球60%的变压器产能及对取向电工钢（GOES）的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "“离网自治”模式面临美欧“安全与电力绑定”和关税/碳税的精准拦截，主权AI项目在放弃中国电力设备以获取AI芯片（交付延迟36-48个月）与使用中国设备（芯片断供）之间面临难以调和的结…", "summary": "本卡对AI“离网自治”模式进行压力测试。美欧正通过“安全与电力绑定”将AI芯片出口与排除中国设备挂钩（如阿联酋G42项目）。同时收紧FEOC与301关税以拦截含中国取向硅钢（GOES）的第三方变压器；欧盟也于2026年实施CBAM并禁用中国逆变器。开发者被迫在“使用中国设备但无芯片”与“脱钩但交付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网延迟与输电设备短缺已将AI竞争从芯片效率转向‘电力交付权’，制造了3-5年的算力建设停滞期，大幅抬升存量电力资产溢价。", "summary": "截至2026-06-11，电力基础设施已成为AI算力的‘大过滤器’。电网扩容（5-15年）与算力建设（1-3年）的错位导致了严重的并网延时，全美队列达2600GW。变压器交期拉长至4-5年，而PJM容量电价暴涨9倍至$329/MW-day。这种‘硬约束’正在触发资金从GPU转向拥有现成电力的公用事…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年美欧\"贸易墙\"及其本地含量规则对中国电力设备全球供应链转移的政策阻断效力研究", "summary": "经墨西哥/东南亚绕道突围的可持续概率偏低（真实本地化约20–30%，螺丝刀组装式转运＜10%）。三套机制同时收紧：USMCA 2026年7月1日审查被重构为反转运、实质性改变执法[S1]；墨西哥2026年1月起对中国输入品加征最高50%关税并收紧IMMEX保税制[S2][S6]；2023年光伏反规…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 5, "neutral": 0, "value": 40, "events": [ { "title": "材料供应链的物理硬约束（变压器4年交付期与铜矿缺口）将导致电网扩张速度远滞后于AI需求，从而推迟公用事业公司的收益兑现并压低利润率。", "summary": "2026年6月调研显示，AI电网建设已撞上“物理材料墙”。大型电力变压器交付周期拉长至3-4年，核心瓶颈在于取向硅钢（GOES）的结构性短缺和精炼铜高达60万吨的缺口。AI数据中心耗铜量是传统设施的3-4倍，材料成本通胀和设备交付延迟将显著推后公用事业公司的业绩兑现，挑战了“电力决定论”下的短期超…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "配电侧物理约束已成为 AI 算力部署的新红线，被迫产生的“400V 桥接期”将超大规模集群的成熟期推后至 2028 年，价值链正向“通电时间”确定性溢价的电力设备厂商集中。", "summary": "配电侧瓶颈（800V DC 生态缺口与 120+ 周变压器交期）正迫使 Hyperscaler 从“算力优先”转向“电力优先”。2026–2027 年将进入“400V 过渡期”，导致 300kW+ 超高密度集群的大规模部署推迟至 2028 年。Capex 分配正向电力基础设施（~15%）倾斜，核心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建的“脱网化”倾向并非缓解电网压力的良药，而是通过抢夺核心电力设备与稀缺劳动力，并造成公共电网固定成本向剩余用户转移，实质上加剧了传统产业的能源成本负担。", "summary": "截至2026-06-12，AI数据中心试图通过表后直供（BTM）和微电网“逃离”公用电网的努力正遭遇物理与监管的双重瓶颈。研究显示，BTM模式因电力成本分摊不公引发监管反弹（如FERC对Susquehanna协议的否决），导致超大型项目转向“表前”（FTM）模式，加剧了电网容量压力。同时，AI基建…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球大容量电力设备产能制造滞后与原材料（取向硅钢、铜）瓶颈，决定了变压器及高压开关柜交期在2028年前无法收窄，仍将是美国AI与电网基建最刚性的物理约束。", "summary": "截至2026-06-12，我们支持卡片06结论，判断全球大型电力变压器与高压开关柜交期在2028年前无法压缩至18–24个月，仍将维持在48-60+个月及24-36个月高位。西门子能源、日立能源及GE Vernova等巨额扩产面临3-4年建设滞后，且受制于取向硅钢（克利夫兰-克利夫斯 Butler…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 3, "neutral": 0, "value": 53.4, "events": [ { "title": "验证 1700 bps 毛利率压缩风险：铜价与硅钢供应的‘双杀’正通过成本端挤压与交付端延迟，终结电力设备行业的利润溢价周期。", "summary": "受 2026 年 6 月铜价飙升至 $14,000/MT 以上及高能效取向硅钢 (GOES) 结构性短缺影响，电力设备厂商面临严重的成本挤压。研究表明，原材料占变压器成本的 70-80%，关键材料 50% 的涨幅在无即时调价情景下可导致 ~1560 bps 的毛利率损失。尽管 PEC 条款普遍存在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关键电力设备与硅钢材料的物理交期瓶颈已取代芯片产能，成为AI算力落地的实质性进度阀门，将平滑的芯片需求周期重塑为台阶式并网周期。", "summary": "截至2026-06-13，大型变压器（LPT）交期达18-36个月，定制高压单元达3-5年，开关设备交期50-80周，主因取向硅钢（GOES）紧缺。美国于2026-04-20通过DPA第303条认定以促进硅钢和变压器本土化。尽管PJM的EIT与ERCOT的Batch Zero推进并网提速，但下游物…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支从云厂商切向电力与瓶颈供应链", "summary": "ORCL的FY26财报既非单纯单一公司事件，也不构成干净的'下周系统性配对交易'，而是对一场早已进行中的结构性轮动的确认：AI资本开支从算力买方迁向物理瓶颈供应商。其-237亿美元自由现金流、约1750亿债务、单一客户OpenAI集中度属公司特有，但'谁出资、谁稀缺'的重定价是全市场主题，并经GE…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 7, "neutral": 0, "value": 30.8, "events": [ { "title": "AI基础设施和电网升级为铜铝提供多年需求底座，但成本侵蚀主要发生在电力设备和固定价格工程链条，而不是高毛利AI加速器本体。", "summary": "截至2026-06-13，本卡支持上游资源品主线，但把AI算力需求拆成“数据中心直接用铜铝”和“电网升级间接放大”两层。IEA预计全球数据中心用电从2024年415 TWh增至2030年约945 TWh，按S&P含铜强度折算，2024-2030年负荷增量对应约1.8-2.4 Mt累计铜需求；CRU…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力需求与电网扩容瓶颈：评估配电侧容量与变压器产能对算力部署的制约", "summary": "支持卡片04“确定性电力获取是胜负手”的判断并收窄：真正的硬约束在电网端而非发电端，核心是大型电力变压器供应链。变压器交付周期已拉长至3–5年（部分超160周），而AI部署周期不足18个月；GSU需求自2019年+274%、Wood Mackenzie测算美国变压器缺口约30%、约4000万台配电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电气设备压力测试之后的 AI 基础设施瓶颈与行业配置", "summary": "本卡把前五卡综合为一条配置规则：当电气设备与液冷把瓶颈从芯片移向已通电算力，区分长期受益与拥挤资本开支受益股的判别变量是公司所占稀缺性租金的持续时间，而非收入增速。三道检验：瓶颈硬度（变压器交期80–210周最硬，液冷因OCP标准化租金衰减最快）、定价权体现在长久期在手订单与毛利（GE Verno…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国变压器出口增长与全球交期拉长相互匹配，但只有在提单重量、航线港口、卫星堆场出清三者同向时，才能把真实稀缺性租金与市场叙事区分开。", "summary": "2026-06-13：本卡对“中国电力设备出口商属于 AI 基础设施稀缺租金”做另类数据压力测试。结论是方向成立但必须防止叙事过热：2025 年中国变压器出口额超过 646 亿元、同比近 +36%，平均单价约 20.5 万元；同时大型变压器交期仍在 80-210 周，美国 LPT 常见 36 个月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 6, "neutral": 0, "value": 38.2, "events": [ { "title": "以变压器为核心的电力设备扩产滞后与原材料供给刚性对2027年后的数据中心并网构成了实质性瓶颈。", "summary": "评估表明，变压器等电力设备供应链扩产极其滞后（大型电力变压器交期拉长至2-4年），受特种钢（GOES）和铜的物理供给制约，各厂商新产能直至2027/28年才能释放。这将直接阻碍2027年后的AI数据中心并网（平均并网及设备延迟超5年），验证了物理层瓶颈对算力中心扩张的实质制约。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器原材料瓶颈（取向电工钢和铜）不会导致GPU需求破坏，而是导致变压器价格飙升及严重的闲置算力折旧损失，迫使科技巨头扭曲资本开支，大举向电力能源基础设施渗透。", "summary": "本报告研究了2026年取向电工钢（GOES）和铜的供应约束对电力设备变压器OEM利润率及AI算力交付的影响。结果显示，虽然原材料价格上涨（重度冲击下GOES+40%，铜+30%）会大幅压缩固定价格合同利润率，但OEM凭借极强定价权通过指数联动将成本完全转嫁给买方。AI算力未出现需求破坏，而是表现为…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的扩张瓶颈已从芯片转向物理电力设备，燃机与变压器OEM凭借5年以上的积压订单和极强的定价权成为供应链中最大的赢家。", "summary": "本报告深入分析了2026年AI电力供应链中的物理瓶颈。研究发现，燃气轮机和高压变压器OEM已成为供应链的绝对咽喉，GE Vernova和西门子能源的燃机订单已排至2030年以后，大型变压器交期突破150周。瓶颈核心在于取向电工钢（GOES）的结构性短缺，导致2026年全美近50%拟建数据中心产能面…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国电力设备出口正凭借‘速度溢价’成为全球AI基础设施的物理基石，并为人民币提供了结构性的宏观对冲价值。", "summary": "2026年全球AI电网约束下，中国电力设备（变压器、特高压、储能）正从‘成本竞争’转向‘速度竞争’，其12-18个月的交期远优于西方的5年 backlog。中国通过控制取向电工钢（GOES）和电力硬件供应链，获得了显著的全球‘速度溢价’（出口均价上涨33%）。这不仅推升了2025年创纪录的1.19…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 13, "neutral": 0, "value": 12, "events": [ { "title": "中国电力装备外需足以覆盖行业层面的供应链重构成本，但超额收益将集中在材料、认证和渠道能力最强的头部企业，而非所有出口商。", "summary": "截至2026-06-16，本卡片压力测试全球变压器超级周期下中国电工钢与电力设备出海的利润传导。结论是外需足以支撑行业继续上行，但不能简单抵消所有地缘贸易壁垒；真正受益者是掌握高牌号GOES、具备高压/特高压认证和海外服务能力，并能在必要市场进行最终装配或服务本地化的企业。美国50%钢铝关税和欧盟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建电力设备供应链的物理约束：变压器、燃气轮机与电网扩容的产能极限", "summary": "我支持Research note 04“执行胜过芯片”的方向，但压力测试其两个承重论点。约束并非单一设备，而是串行链条（并网5.2年→高压变压器128周/GSU 144周→开关柜→燃气轮机），最长杆是变压器总装/试验吞吐，而非资本[S1][S2]。GOES是真实但次级的约束：价格较2020年翻倍[…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026 06 15，通过对五大云厂商6020亿美元Capex计划、AI数据中心3倍金属强度以及中游设备厂商（Eaton, Vertiv, Schneider）创纪录的150亿美…", "summary": "截至2026 06 15，通过对五大云厂商6020亿美元Capex计划、AI数据中心3倍金属强度以及中游设备厂商（Eaton, Vertiv, Schneider）创纪录的150亿美元积压订单的深度穿透，我确认AI与电网基础设施投资已实质性转化为铝、铜及电力设备的物理订单流。这一“订单实物化”验证…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由于熟练绕线劳动力稀释、高压测试台位积压以及换位导线（CTC）供应瓶颈，变压器产能扩张存在2至3年的物理技术滞后，无法支撑2026-2027年的电力需求爆发，交付周期将持续拉长。", "summary": "截至2026-06-15，本卡片对变压器OEM厂商产能扩张的非钢材物理瓶颈（绕组人工、测试台位和CTC铜线）进行了压力测试。我们发现，尽管主要OEM厂商宣布了大量产能扩张，但受限于熟练绕线工1.5-3年培训周期、测试台位2-3年物理建设与设备积压周期以及CTC磁力线供应瓶颈，名义扩张将面临24-3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 4, "neutral": 1, "value": 56, "events": [ { "title": "AI算力的“快时钟”正与电网基础设施的“慢时钟”发生剧烈碰撞，变压器产能缺口将迫使数据中心建设周期长期化并推高资本开支，形成实质性的硬约束。", "summary": "截至2026年6月16日，变压器与电网设备供应链已成为AI算力扩张的实质性物理瓶颈。大型电力变压器（LPT）交期延长至36-60个月，核心原材料取向电工钢（GOES）供应高度集中且价格较2020年上涨50-80%，导致变压器综合成本上涨60-80%。约30-50%的2026年计划产能面临延迟，建设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力端供给的改善无法掩盖配变电侧的短板，电力设备的交付错配将引发“有芯无电”困境，导致算力基建估值逻辑短期面临断裂。", "summary": "先进封装产能释放仅将算力瓶颈由“硅”转移至“电”。大型变压器长达120-140周的交付期与并网排队严重制约算力落地，导致2026H2-2027年云厂商面临“有芯无电”闲置困境，破坏了原有的Capex即刻变现估值逻辑，进而可能推迟后续算力硬件订单。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网容量瓶颈、长达3年的变压器交付周期及硅钢材料短缺，已构成AI数据中心实际通电的决定性物理限制，将电网受限下的综合结算电价拉高94%，推动表后自备发电套利并引发电网监管备用费狙击。", "summary": "本报告对AI数据中心并网瓶颈与电力设备短缺进行压力测试。主电网容量限制与高压变压器（交期达128-160周）及硅钢短缺，构成通电的主要物理屏障。受限下容量电价升至$329.17/MW-day，使100MW数据中心结算电价暴涨94%至$80.24/MWh，推动表后自备直连套利，但面临电网备用费监管打…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-16，变压器与GOES供应弹性不足以在2026-2027年全面缓解AI算力落地的物理天花板，只能给已提前锁定OEM产能的头部云厂商提供选择性缓冲。", "summary": "截至2026-06-16，本卡压力测试认为：GOES和铜不是孤立瓶颈，真正约束是GOES、CTC铜导体、制造槽位、专用设备、测试、熟练工和运输共同构成的变压器工业系统。80-210周交期意味着2026年新下单多数无法支撑2027年前通电；Hitachi、Siemens等扩产有助于2027年末以后斜…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 4, "neutral": 1, "value": 57.4, "events": [ { "title": "全球电网核心设备及原材料的产能赤字已构成刚性物理准入壁垒，导致产业利润池从算力/晶圆厂资本支出方结构性向重电OEM厂商转移。", "summary": "大功率汽轮机与超高压变压器的严重产能缺口（全球超高压变压器短缺约30%，交期延至2-4年；大型汽轮机交期延至5-7年）和取向硅钢（GOES）等关键材料瓶颈，对算力中心和晶圆厂构成不可逾越的物理准入壁垒。议价权转移令电力设备OEM（如HD Hyundai Electric和GE Vernova）利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力向电力设备的资本开支传导扩大了实体“流动性黑洞”，高企的无风险利率与公用事业“成长化”转型持续挤出长久期资产的配置资金。", "summary": "本报告支持AI基础设施分析师的判断，量化了AI电力需求向能源基建的传导。2026年全球数据中心用电预计达565 TWh，中国“十五五”电网投资将达5万亿元，且全球面临30%的变压器结构性缺口。这种大规模实体CapEx形成巨大的“流动性黑洞”，虹吸市场资金，推高资本成本，对医药等长久期板块估值维持持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国电力设备出口受益是真实结构性机会，但上限取决于地缘政策折价；国内 AI 算力扩张的核心约束将从芯片采购转向局部电力接入和稳定电源。", "summary": "截至2026-06-18，本卡 synthesize 并部分支持上一卡：全球电力设备短缺是真实瓶颈，中国在变压器、HVDC/UHV、开关设备和数据中心供配电环节具备结构性出口机会，但美国/欧洲关键电网的安全审查会限制份额。国内 AI 不是全国性缺电问题，而是局部“合格兆瓦”问题：IEA 估算中国数…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "西方电力设备巨头的产能释放滞后于AI并网高峰，且由于零部件卡口与合同刚性，中国产能扩张无法有效减轻其高压交付压力。", "summary": "截至2026-06-18，西方巨头（日立能源、西门子能源、GE Vernova）已启动数十亿美元的变压器产能扩张，但核心产能释放集中于2028年以后，与2026-2027年AI数据中心建设高峰存在严重错配。由于地缘壁垒，中国产能无法直接参与美欧大型高压输电变压器建设，且受限于套管短缺、劳动力匮乏和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 4, "neutral": 0, "value": 55.8, "events": [ { "title": "G2/G3瓶颈没有被效率叙事推翻，反而因变压器、电气设备与并网队列的多年约束而强化近现金电力设备订单资产的相对优势。", "summary": "截至2026-06-19，本卡支持Research note 02：AI电力交付瓶颈是真实供给侧闸门。美国数据中心用电2028年可达325-580 TWh，电网设备采购和并网队列仍是多年周期；美国活跃并网队列2025年底仍超2,060 GW，PJM数据中心区等待36-48个月；欧盟并网等待可达2-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器供应链足以封顶部分未锁定设备的美国AI集群，但不足以在2027年前封顶全球AI部署；核心操作是区分已锁定电力设备项目与仍在寻找电力的项目。", "summary": "截至2026-06-18，本卡压力测试Research note 02的电力设备约束：变压器是核心瓶颈，Wood Mackenzie显示2024年平均交期120周、大型变压器80-210周，美国2025年电力变压器供给缺口30%；数据中心变电站变压器交期已超过160周。GIS和开关设备会造成调试延…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力规模化面临的下一重物理极限：电力供应、电网接入与散热约束", "summary": "截至2026-06-18，本卡支持卡片05的交接：2026-2027年电网接入容量与高密度液冷是AI数据中心扩容的实质性硬约束，但约束在于可交付性与时序，而非算力绝对上限。IEA基准下全球数据中心用电2024年约415TWh翻倍至2030年约945TWh；美国年新增负荷5-7GW而电网仅交付2-3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新增AI风险预算应从继续加半导体Beta转向电力稀缺链条，半导体保留核心仓位但不再是边际加仓方向。", "summary": "截至2026-06-19，本卡支持并推进此前从通用软件/加密杠杆转向AI硬约束的思路：半导体基本面仍强，但新增风险预算的边际胜率已转向电力设备、并网基础设施、现场供电和核燃料。建议下周新增AI风险预算60-70%投向电力与供能约束，20-30%保留精选半导体，0-10%用于现金或对冲。核心依据是h…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 1, "risks": 4, "neutral": 0, "value": 65.2, "events": [ { "title": "电力设备供应链瓶颈及其对AI基础设施扩能的二次约束效应", "summary": "报告日期2026-06-19。本收官卡压力测试Research note 07“折价而非证伪”结论，引入变压器、高压开关与取向硅钢(GOES)的二次设备瓶颈。变压器交期由128周(25Q2)拉长至超160周/最高4-5年[S2][S1]，GSU需求自2019年+274%、电力变压器供应缺口约30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心配电侧存在类似电子布的材料升级逻辑，但结构性溢价属于经过热、介电、防火和高电流认证的系统材料栈，而不是所有电力设备大宗零部件。", "summary": "截至2026-06-20，本卡片支持AI算力功率密度向配电侧关键材料升级传导，但认为它不是电子布价格周期的一比一复制。120 kW级AI机柜把问题从普通供电容量推向高电流、热老化、介电可靠性和灭弧认证。结构性受益集中在高牌号GOES、CTC铜线、工程/芳纶绝缘纸、酯类介质液、母线、高密度中压开关及…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电网吸纳能力瓶颈会削弱AI投资对近期r*的无摩擦上行冲击，但会把周期拉长并转向电力设备资本开支，因此投资需求逻辑被折价而非被证伪。", "summary": "截至2026-06-19，本卡压力测试Research note 06的r*上行逻辑。结论是：电网与变压器瓶颈会降低近期资本开支效率，但不会取消投资需求渠道。Goldman估算2026-2031年AI基础设施资本开支为7.6万亿美元；IEA提示约20%规划数据中心项目存在电网延期风险。套用于202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心验证的是持续电网资本开支周期，但投资上只能资本化已签约、可回收、技术受限的设备材料需求，而不能把全部排队MW视为真实订单。", "summary": "截至2026-06-20，本卡片综合前序结论：AI数据中心电力设备需求不是单纯的短期抢装，而是经过排队筛选、容量成本、电价机制和变电站/输电工程落地后的多年电网资本开支周期。但原始接入排队存在高估，必须把已签约、有抵押、有最低需量义务的MW与投机性报装区分开。结构性受益集中在高可靠变压器、中压开关…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 4, "neutral": 0, "value": 58.6, "events": [ { "title": "电力设备交期瓶颈与大宗物流通胀引起的制造业利润受阻，共同锁死了2027年后全球AI算力集群的物理部署进度，迫使超大云厂商非计划性降速。", "summary": "本卡支持并完成了十卡研讨。在原油供给冲击与G3利率高企背景下，大型变压器（交期80-210周）与高压开关柜（交期80-120周）交付滞后是限制2027年后算力集群建设的终极瓶颈。定量测试显示，大宗及海运暴涨导致中国纯出口商变压器净利润转负（-3.52%）强制引发交付中断；而本地化厂商受限于取向硅钢…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "东部AI集群的主要进度瓶颈是高能效电力设备因出海订单遭产能挤压，以及一线城市高压电缆管廊长达12-15个月的市政审批与土建物理施工周期。", "summary": "本报告评估了配电侧关键变压器与电缆供应链对东部AI智算集群建设的进度影响。名义上中国电力设备交付仅需4-6个月，但高利润出口订单抢占产能以及GB 20052-2024能效标准导致高牌号硅钢紧缺，延长了国内交付。同时，一线城市地下电缆管廊施工及复杂的市政审批需要12-15个月，成为真正关键路径。在压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-20，全球电力设备扩产会缓解2028-2030年的AI供电瓶颈，但不足以抵消2026-2027年的物理接入延迟风险。", "summary": "截至2026-06-20，本卡压力测试电力设备扩产能否抵消AI集群物理接入延迟。结论是否定强版本：扩产真实存在，但释放节奏偏后。Wood Mackenzie预计美国数据中心电气设备市场2025-2030年从200亿美元升至650亿美元，数据中心容量2026-2030年从约24 GW升至110 GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "跨区域绿电直连因电力设备交期瓶颈与大电网备用费制约，将拉长建设周期2至2.5年，由此带来的巨额机会成本与折旧时延将彻底抵消688号文的降本利好。", "summary": "截至2026年6月20日，本报告对发改能源〔2026〕688号多用户绿电直连政策在AI算力中心的落地进行了压力测试。结果表明，专用输电线路建设、变压器128周排队交期、路权审批时延将推迟投产期2.0至2.5年，且大电网收取固定的容量备用费（折合0.0456元/千瓦时）收窄了价差。2年时延导致100…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 7, "neutral": 0, "value": 42, "events": [ { "title": "电力设备（如变压器）供应链瓶颈对AI基础设施建设节奏的影响", "summary": "压力测试结论：是，但有两点修正。对美欧并网型AI建设，电力设备2026年当下即是约束边界且早于芯片——约一半美国规划数据中心已延期/取消，12GW中仅约5GW在建[S4][S9]，而CoWoS仍在近4倍扩产出货[Card05]，故‘买得到却通不了电’的部署约束当下即活跃。修正一：约束不是变压器箱体…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件已到战术拥挤预警点，电力设备则是结构分化而非整体泡沫；制造出海主线应再平衡而非否定。", "summary": "截至2026-06-21，本卡由chief-strategist完成压力测试。AI硬件的成交、ETF与主力资金拥挤度已接近战术临界点，短期回撤风险上升，但机构超配比例仍低于历史峰值，不能据此全面退出。电力设备并非整体拥挤，风险主要在主题化交易载体和旧新能源β；电网、变压器等出海订单链仍有支撑。策略…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美欧贸易壁垒足以压低中国电力设备出海的利润率和估值倍数，但不足以抵消无源、短缺约束型电网硬件的出口红利。", "summary": "截至2026-06-21，本卡否定“美欧关税和贸易限制足以抵消中国电力设备出海红利”的强判断。政策摩擦真实存在：美国Section 301覆盖部分HS 8504及能源相邻品类，欧盟通过NZIA、FSR和高风险逆变器融资限制提高准入门槛。但美国并网队列、变压器80-210周交期、DOE披露的进口依赖…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备交付才是算力扩张的硬闸门，绿电基地是稳态opex优化器而非并网延期的对冲工具。", "summary": "截至2026-06-21，本卡压力测试Research note 4：支持“变压器/开关柜/并网是算力扩张硬约束”的方向，但下调“绿电可显著缓解总算力成本”的强度。DOE与Wood Mackenzie显示大型变压器交付周期常见36个月、最长60个月，或80-210周；中国供应链更快但部分订单已排到…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 1, "neutral": 0, "value": 79.4, "events": [ { "title": "高密度AI集群扩张对区域电网配电侧基础设施的负荷压力与承载极限预测", "summary": "截至2026-06-22，本卡压力测试并锐化整条链条的“电力墙”论点：是的，配电侧物理时滞是绑定约束，但它比Research note 02-07定位的层次更深一层。液冷+ASIC（Research note 09）抬升的是机房内部功率上限，却无法让变压器造得更快。ASIC密度兑现现受制于GOES…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 7, "neutral": 0, "value": 44.8, "events": [ { "title": "电力设备供应链扩产进度能否扭转2027年交期约束", "summary": "压力测试'扩产能救2027'的乐观读法：GE Vernova(全资Prolec GE，US$5.275bn)、Eaton(Jonesville US$340m+Nacogdoches US$100m+Waukesha US$22m)、Hitachi Energy(逾US$1bn含South Bos…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "贸易救济与关税升级对中国电力设备出海成本红利的侵蚀风险评估", "summary": "对卡片04–06「出海双击」论点做贸易政策压力测试，结论刻意不对称：壁垒真实且升级迅猛，已在筑墙的美国（及逐步收紧的欧盟）咬住成本红利，但不足以扭转全球格局。美国叠加关税墙对一台中国变压器约达40–65%+从价[自测算]，高于20–40%挂牌优势[S7]；但中国对美直接份额已由27%降至约10%、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网输变电容量与并网审批周期（而非总体发电量）已成为AI算力部署的终极物理刚性瓶颈，并已实质性挤压IT硬件采购预算。", "summary": "本报告对AI算力扩张的电网与能源瓶颈进行压力测试。分析表明，虽然电力设备订单充足，但美国并网队列平均达5.5年，且面临FERC 2026年6月大型负荷监管及弗吉尼亚GS-5等防止成本转嫁政策。变压器等重电面临GOES原材料短缺，Tier-1厂商积压至2029-2030年。表后核电直供监管被封死，燃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-23，我们认为高压试验大厅与核心供应链瓶颈限制了全球电网关键设备的扩产节奏，推动具备交期和全产业链优势的中国头部制造厂商出海订单与利润率“双击”，重塑了制造板块估值…", "summary": "本报告于2026-06-23撰写，分析了AI算力与电网升级下的电力设备产能缺口。高压试验大厅与取向硅钢构成物理扩产壁垒，导致美国变压器交期拉长至128-144周。中国企业凭借交付与成本优势迎来出口爆发，重塑了制造板块由“周期”向“出海成长股”的估产双击估值逻辑。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 5, "neutral": 1, "value": 61.2, "events": [ { "title": "关键电力设备（变压器/开关/UPS）的交付周期与产能约束评估", "summary": "压力测试卡片06的“设备交付为约束”命题。结论基本否定但附两条限定：就中国境内而言，设备总产能与硅钢/铜大宗总量都不是约束——TBEA一家约260 GVA/年，远超卡片06测算的AI增量约16.6 GVA/年（仅约6.4%）；中国2024年取向硅钢产量303万吨（日5倍、美8倍）结构性过剩且净出口…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网侧供应瓶颈对电力设备资本开支的驱动效应", "summary": "支持立场。变压器交期已拉长至128周（LPT）/144周（GSU），北美极端排期达60个月，使交期本身转化为多年期订单能见度：GE Vernova电气化在手订单同比升至424亿美元（覆盖约3年营收[自测算]），Siemens Energy集团在手1,540亿欧元，Eaton总订单+73%至超190…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "评估欧美电网升级需求与贸易壁垒对中国电力设备出海链的传导效应", "summary": "压力测试 Research note 4 命题：A股估值重配是否真建立在可持续高毛利出海订单上。结论为部分支持但有结构性保留。需求拉动真实且被低估——美国公用事业2026-30资本开支近1.3万亿美元 [S1]、超大规模厂商2026资本开支>6750亿美元/+60% [S2]、全球变压器短缺 [S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科技板块风格再校准：能效赋能方的稀缺溢价与高耗电算力股的吞吐折价", "summary": "本卡把前五卡收束为可执行配置框架。算力以‘每兆瓦有效算力’计价后，权益市场沿‘谁拥有稀缺投入、谁的回报被它锁死’重估。折价已现于现金流：2026超大规模厂商资本开支约6900–7250亿美元、约75%投向AI，却吞噬近100%经营性现金流；微软约800亿美元Azure积压受限于电力。溢价已现于盘面…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 3, "neutral": 0, "value": 71.6, "events": [ { "title": "全球电力设备制造供应链产能与上游材料供应风险评估", "summary": "部分确认Card-03的time-to-power论点，但作三处修正。瓶颈确在上移到材料端，但对变压器而言真正的物理咽喉是GOES而非铜，且GOES稀缺本质是“产能+贸易”稀缺而非地质稀缺：全球变压器产能利用率98%、GOES自2020年涨约100%、美国仅Cleveland-Cliffs一家、俄…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备短缺不会取代电力缺口，而是把“已签MW”过滤成“可通电、可保护、可冷却、可调试MW”，其中变压器是2026-2028年最硬瓶颈。", "summary": "截至2026-06-25，本卡压力测试认为电力配套设备会成为继电力缺口之后的下一道算力部署硬约束，但约束强度分层：变压器最硬，128-143周交期足以卡住2026-2028年通电节奏；开关柜平均44周，更多体现为分阶段通电和balance-of-plant同步风险；液冷会卡住GB200/Rubin…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备原材料缺口对新能源项目建设成本与并网进度的传导效应评估", "summary": "支持卡片06–08的逻辑链：AI电力瓶颈已上移至原材料，且GOES/高纯度铜缺口确实传导至新能源经济性，但传导非对称、主要走时间而非单价。直接价格通道有限——铜仅占光伏/陆上风电资本开支约3%~5%，即便铜较2020年+50%、GOES+100%，对LCOE直接抬升仅约+2%~+4% [自测算]。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 5, "neutral": 0, "value": 61, "events": [ { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备杠铃可以落地为六只A股核心篮子，但必须以流动性、两融拥挤和业绩质量仪表盘强制约束，触发信号时用IC/IM/IF与ETF期权对冲并削减设备腿。", "summary": "截至2026-06-27，本卡片支持前序电力设备杠铃，但将其落实为严格的A股执行篮子：2.5% NAV电力设备由首钢股份/望变电气表达GOES稀缺、思源电气/金盘科技表达出口Alpha，2.0% NAV公用事业由长江电力/国投电力压舱。对冲工具以IC/IM/IF和沪深300、上证50、中证500 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海外电力设备及并网物理瓶颈将导致CXO海外基地落地延迟2-3年，造成项目NPV毁灭性缩减77.07%，并通过毛利率永久性收缩25-30pct对企业估值中枢构成硬约束。", "summary": "压力测试显示，受AI算力基建挤压，全球大型变压器（LPT）交期拉长至48-60个月。在新加坡、英国和爱尔兰，海外CXO建厂面临2-3年并网延迟。以$500M基地测算，3年延迟将造成$270M资金占用与设备折旧折损，毁灭该项目77.07%的NPV。此外，地缘安全审查隔绝了中国快交期电力装备，迫使项目…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网基础设施与能源转型对AI算力扩张的潜在约束", "summary": "压力测试部分支持并升级卡03：电力/电网瓶颈是“节流而非断路”，但节流比其暗示更紧更久。在2026–2028视野内，电气基础设施已是排在芯片之前的硬约束——美国电力变压器交期128周、GSU 144周、大单3–4年，价格较2019+77%、达2022前4–6倍；GOES由Cleveland-Cli…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 4, "neutral": 0, "value": 68.4, "events": [ { "title": "功率半导体已过剩，AI电力的新物理瓶颈已迁移至大型变压器，并真正卡在无法被资本快速绕行的上游电磁钢(GOES)与铜——2026–2028窗口难解。", "summary": "支持并收紧前五卡逻辑链：功率半导体转入结构性过剩后（Research note 05），AI电力建设的刚性瓶颈已迁移至大型变压器(LPT)，并下沉到上游电磁钢(GOES)与铜。瓶颈呈两层：整机厂总装层可用约20亿美元北美扩产去瓶颈，但缓解要到2028/2029年后；标准电力变压器交期128周、GS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据中心电力并网、关键电力设备与液冷温控的物理交付瓶颈将对算力上线速度形成硬约束，导致云厂商ROIC受损并延缓算力采购，增加电子产业链利润扩张的硬着陆风险。", "summary": "截至2026-06-28，本卡压力测试了算力产业链利润扩散假说，指出电力并网与变压器超长交期（MV变压器128周）已成为数据中心上线的物理瓶颈。预计2026–2028年美国35%–45%的新建数据中心面临延期通电，16GW计划容量仅5GW在建，30%-50%将延期至2028年以后。GB200的高功…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "summary": "截至2026-06-27，本卡片综合前序结论：电网/变压器瓶颈会延后并改道AI部署，但不会摧毁2028年前光模块出口龙头业绩能见度；真正尾部风险是2027年云厂商ROIC压缩。建议A股AI物理瓶颈组合使用40/32/12/8/8结构：变压器/电网设备、光模块、电力现金流锚、铜/GOES等输入端对冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器材料和交付瓶颈已开始改变 Hyperscaler 的2026-2027资本开支与电力采购节奏，影响主要体现为预算通胀、长期锁电和通电交付递延，而非AI需求取消。", "summary": "截至2026-06-28，结论为支撑：Hyperscalers 已经通过预算上修、长期电力合同和项目节奏调整来应对变压器及电气设备瓶颈，但公开披露通常不会把变化直接归因为 GOES 或铜关税。Microsoft 2026 capex 中约250亿美元来自更高组件价格，Meta 上调2026 cap…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 12, "neutral": 0, "value": 21.8, "events": [ { "title": "电力设备成本传导逻辑的脆弱性：铜与电工钢供应瓶颈能否被设备商有效对冲？", "summary": "本卡用材料视角压力测试2027-2028年铜与GOES供需缺口对HV变压器利润率的侵蚀确定性。结论：成本传导逻辑脆弱，但脆弱是非对称、取决于合同结构的，而非全行业确定。GOES是确定性最高、最难对冲的咽喉——区域单一来源（美国仅Cleveland-Cliffs，本土覆盖约1/5需求）、约占LPT成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "summary": "本卡支持 Research note 02「上游原材料是最深约束」的结论，并锐化为一组非对称结构。GOES：自2019年变压器需求暴增（电力+119%、GSU+274%），远超GOES市场约5.4–5.6%的CAGR；GSU供给缺口2025年约100%，因美国唯一GOES生产商Cleveland-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "summary": "本卡压力测试逻辑链：GOES约束不会「抵消」核能加速，因为二者作用于电力链不同节点——核能重启在发电/并网侧本就是GOES-轻（复用既有GSU与开关站），属于一种GOES套利；但GOES在负荷侧与输配路径重新绑定，沿用Research note 01的NPV模型测算，核能锚定型项目(≈1.5年延迟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 6, "neutral": 0, "value": 59.2, "events": [ { "title": "电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "summary": "本卡压力测试 Research note 6 的'调价条款已隔离利润'结论，认为其仅有条件成立。主流调价条款（MEC）挂钩有流动期货的铜/铝（出口超级周期转嫁率约80%[S1]），却普遍排除或封顶无流动指数的取向电工钢GOES——而GOES占COGS约28%[Research note 6]且处2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "summary": "本卡片从制造和供应链视角对电网与电力约束进行压力测试。截至2026-06-29，关键电力设备（变压器与重型燃机）由于极高工艺与测试门槛、上游取向电工钢（GOES）瓶颈及寡头垄断，交期拉长至3-7年，生产机位已售罄至2029-2030年。云厂商BTM发电加剧了燃机机位争夺。预计2026-2027年该…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "铜与硅钢暴涨及高预付要求将击穿缺乏转嫁能力的二线及内销电力设备厂商的利润护城河，形成“有订单无利润、无现金”的贫血式增长。", "summary": "截至2026-06-29，本卡片通过敏感性模型对变压器厂商毛利率进行压力测试。结果显示，在铜价高企（$12,000-$13,500/t）和高牌号硅钢供给偏紧下，若厂商无成本转嫁力，严重压力下毛利率将由30%斩半至13.73%。这导致以金盘科技为代表的二线及内销厂商订单虽强，但因交付周期错配和原材料…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备出海高增长与国内“十五五”电网资本开支爆发锁定了龙头业绩的高能见度与利润确定性，是中证红利风格强化趋势下红利成长股的首选品种。", "summary": "本报告定量评估了电力设备龙头在“全球出海”与“国内十五五电网更新”驱动下的业绩确定性。全球变压器交期拉长至107.8周，高出口毛利率 premium 拥有调价条款保护。压力测试表明，即使北美订单延迟40%，主要龙头季度净利下修仅为-2.55%至-10.54%，自由现金流影响可控。国内超过5万亿元电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-f1c88bcd88/", "networkHref": "/investor/research-network/#thesis-f1c88bcd88" }, { "id": "thesis-11f0838e82", "source": "auto_topic", "title": "科创50收于1,751.32点，距离1,750雪球/两融生命线仅一步之遥。若下周破位跌破1,750，我们如何防范882亿元的级联式去杠杆…", "titleEn": "Practical research topic: 科创50收于1,751.32点，距离1,750雪球/两融生命线仅一步之遥。若下周破位跌破1,750，我们如何防范882亿元的级联式去杠杆…", "lane": "中国权益与流动性", "laneEn": "中国权益与流动性", "laneHref": "/investor/lanes/lane-c06f0832/", "investableFocus": "", "stockUniverse": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" } ], "stockTickers": [ "510300.SH", "588000.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK", "9988.HK", "APO", "IEF", "CAT", "COP", "MAR", "FXY" ], "coreView": "该主线来自 3 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-31 至 2026-06-26；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-21", "lastSeen": "2026-06-29", "supports": 4, "risks": 137, "neutral": 19, "analystCount": 14, "sourceTypeCounts": { "研究笔记": 127, "研究主题": 13, "研究线程": 14, "分析协作": 6 }, "topTerms": [ "科创50", "A股", "两融", "红利", "雪球", "Gamma", "与两融强平后", "科创50收于1", "距离1", "雪球/两融生命线仅一步之遥", "若下周破位跌破1", "如何防范882亿元的级联式去杠杆" ], "whatChanged": [ "2026-06-29: 政策背离下的A股风格轮动与板块配置：能源安全杠铃 vs. 经利润分层的转型龙头，以及A股的相对庇护所超配", "2026-06-29: 在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%）；通过配以5.0%权重可转债债底防御和0.50%权重股指虚值Put期权，可将组合极限收益拉升至+104.40%以平准踩踏损…", "2026-06-29: A股红利拥挤度与7-11月年内解禁洪峰限制了风格切换空间，而在人行短端流动性上限失灵与CNH被动升值情景下，高负债、高Beta的券商板块将面临最严重的两融净利差挤压与反转冲击。", "2026-06-29: 国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多资产组合VaR在1530点处强制破位，急需分层部署的ETF买盘及两融弹性化阀门进行精…", "2026-06-28: 在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼及普钢加工环节将面临严重的利润率挤压与低Beta震荡。" ], "practical": { "sectorLane": "中国权益与流动性", "sectorLaneEn": "中国权益与流动性", "exposureBucket": "中国权益与流动性", "tone": "green", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" }, { "ticker": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "roleEn": "港股权重与风险偏好", "bucket": "core", "href": "/investor/stocks/0700.HK/" }, { "ticker": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中概/港股权重", "roleEn": "中概/港股权重", "bucket": "core", "href": "/investor/stocks/9988.HK/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" } ], "stockTickers": [ "510300.SH", "588000.SH", "512100.SH", "2800.HK", "MCHI", "KWEB", "0700.HK", "9988.HK", "APO", "IEF", "CAT", "COP", "MAR", "FXY" ], "primaryStocks": [ { "ticker": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "roleEn": "核心指数风险偏好", "bucket": "core", "href": "/investor/stocks/510300.SH/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "roleEn": "小盘风险偏好", "bucket": "core", "href": "/investor/stocks/512100.SH/" }, { "ticker": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "roleEn": "港股核心指数代理", "bucket": "core", "href": "/investor/stocks/2800.HK/" }, { "ticker": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "roleEn": "离岸中国风险", "bucket": "core", "href": "/investor/stocks/MCHI/" }, { "ticker": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "roleEn": "中概互联网高 beta", "bucket": "core", "href": "/investor/stocks/KWEB/" } ], "instrumentTags": [ "510300.SH", "沪深300 ETF", "588000.SH", "科创50 ETF", "512100.SH", "中证1000 ETF", "2800.HK", "盈富基金", "MCHI", "iShares MSCI China ETF", "KWEB", "KraneShares CSI China Internet …" ], "reviewCadence": "每日复核", "confirmationSignals": [ "融资余额、ETF 申赎和北向资金是否同向", "被动调仓窗口是否被主动资金提前交易", "订单、盈利和估值是否同步改善" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "指数期权", "红利低波", "汇率", "两融降温" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 128 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 14, "recentEvents": 128, "supportRatio": 0.03, "riskRatio": 0.86, "conflictScore": 96.9, "evidenceDepth": 100 }, "score": 66.9 }, "events": [ { "date": "2026-06-29", "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），将引致多…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）将使平仓率非线性飙升至20%，面临940亿STAR50抛压与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.75, "matchedTerms": [ "科创50", "A股", "两融", "红利", "雪球", "Gamma" ], "href": "" }, { "date": "2026-06-29", "title": "A股红利拥挤度与7-11月年内解禁洪峰限制了风格切换空间，而在人行短端流动性上限失灵与CNH被动升值情景下，高负债、高Beta的券商板块将面临最严重的两融净利差挤压与反转冲击。", "summary": "截至2026-06-29，本卡对A股杠铃轮动的拥挤度与压力进行评估。全市场两融达30329.92亿元（TMT占41%），银行与防御红利极其拥挤，仓位达95%分位；6-11月面临2.8-3.2万亿元解禁洪峰，今日单日解禁达589.22亿元，严重压制内需成长。若CNH继续升值或人行回购走廊失灵（DR007升至2.50%），券商两融利差收窄33.19%，导致两融…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "A股", "两融", "红利", "雪球", "Gamma" ], "href": "" }, { "date": "2026-06-29", "title": "在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%）；通过配以5.0%权重可转债债底防御和…", "summary": "截至2026-06-29，在两融强平与350亿雪球敲入负Gamma踩踏、以及人行短端Corridor失灵（DR007升至2.50%）的极端情景下，杠铃组合资产相关性共振跃迁（红利与成长升至+0.55至+0.70）。组合波动率由9.60%升至19.98%，日99% VaR升至2.93%（穿透2.50%上限），10日99% VaR升至9.27%（穿透7.50%…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.75, "matchedTerms": [ "A股", "两融", "红利", "雪球", "Gamma" ], "href": "" }, { "date": "2026-06-29", "title": "政策背离下的A股风格轮动与板块配置：能源安全杠铃 vs. 经利润分层的转型龙头，以及A股的相对庇护所超配", "summary": "把卡片01-03的宏观综合落成组合：PPI再通胀/CPI去通胀的撕裂本身就是风格信号，奖励上游/价值（约5.3%股息carry）与拿到人行宽松顺风的长久期成长龙头，碾压同质化中段。我支持杠铃、但否定彻底超配。利润层正常化所依赖的反内卷整合于2026-01-09被反垄断监管层暂停[S3]，故转型腿仍是销量、非利润；两融余额6-23历史首破3万亿[S4]、监管…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 20.5, "matchedTerms": [ "A股", "两融", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "虽然278%平均维保安全垫使两融连锁抛压呈局部非系统性风险，但20-30%回撤仍将触发最高2528.00亿的平仓压力并重创产业链，应通过减仓高倍数前道、超配低谷先进封装以及引入非担保品指数对…", "summary": "截至2026-06-28，两融杠杆在电子和半导体板块高度集中（总余额分别达6132.72亿和560.80亿），前五大龙头占用386.80亿。压力测试显示，20-30%的回撤会迫使大批账户跌破130%平仓线，极端情况下（科创50跌破1520）诱发2528.00亿平仓压力，并因雪球负Gamma在2-5天内引发600-750亿级联抛售。虽然278%平均担保垫等四…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "科创50", "两融", "雪球", "Gamma" ], "href": "" }, { "date": "2026-06-28", "title": "在紧平衡流动性与外部融资约束下，A股难以支撑全局性风险偏好扩张，下半年配置应降低总Beta，采用“硬科技国产替代（封装/设备）+高独立现金流红利”的结构性杠铃组合。", "summary": "在美联储加息预期和国内紧平衡流动性制约下，A股面临外资流出、两融去杠杆与解禁供给的三重压力。外资流入放缓，险资增配高股息红利防御资产。两融余额高企（3.015万亿元）使科技成长面临局部去杠杆和级联平仓风险。半导体板块估值已普遍收缩30%至45%，大基金三期（3440亿元）提供战略支撑。下半年主导风格将呈现“国产替代硬科技+高壁垒红利防御”的杠铃组合。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "红利" ], "href": "" }, { "date": "2026-06-28", "title": "A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。", "summary": "截至2026-06-29，本卡片支持前序“更高更久”利率压制A股成长久期的判断，但将风险定位为结构性踩踏而非全市场2015式去杠杆。两融余额为30329.92亿元，占A股流通市值2.83%，总量可控；但电子板块融资余额6261.87亿元，且6月融资流入高度集中。模型显示，科创50较2026-06-26收盘下跌15%会触发约1063亿元有效被动卖压，20%跌…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "A股", "两融" ], "href": "" }, { "date": "2026-06-28", "title": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼及普钢加工环节将面临严重的利润率挤压与低…", "summary": "本卡通过技术壁垒与自给率对A股AI原材料敞口进行了区分：高端取向硅钢方面，宝钢股份凭借规模与技术优势展现出强Alpha属性，而首钢股份受普钢周期Beta拖累较大；有色铜方面，在2026年长单TC/RC为零的极端低位下，矿权自给率超65%的紫金矿业可完全捕获矿端溢价（EBITDA价格敏感度约7.5%），而低自给率的江西铜业、云南铜业面临严重的加工毛利压缩。建…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "红利" ], "href": "" }, { "date": "2026-06-27", "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌破3.0%的红利反向出清防线。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "科创50", "A股", "两融", "红利" ], "href": "" }, { "date": "2026-06-27", "title": "全球商品通胀刚性与风险平价失效下的A股风格轮动：上游资源/红利价值 vs 成长，及两融与三季度解禁的交互", "summary": "全球'上游资源+红利价值'赢家只主导了沪深两市的一半：A股把组合拆开——资源赢（有色年初至今+30%以上、行业冠军[S4]），骑在第06-07卡铜/GOES缺口上；红利价值落后（中证红利一季度约+4%[S5]），因边际资金追逐成长逼空（科创50约+40%、创业板约+30%、6-22创新高[S4]）。H4L是外部贴现冲击，被内部杠杆+国产替代叙事压过[S2]…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 25.5, "matchedTerms": [ "科创50", "A股", "两融", "红利" ], "href": "" }, { "date": "2026-06-27", "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 工业制造分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: eb737f41-b97c-43b9-9b89-04a7c7636133 结果摘要: Wrote handoff_response.md (10188 bytes); wrote handoff_response.en.md (12356 bytes) — 三日近80…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 25.25, "matchedTerms": [ "科创50", "A股", "两融", "雪球", "Gamma" ], "href": "" }, { "date": "2026-06-27", "title": "中国央行通过结构性再贷款与离岸FIMA人民币国债抵押回购机制，成功解耦了汇率防御与境内流动性，为A股“高股息防御+硬科技成长”的滞胀杠铃配置提供了宏观及风险平抑支撑。", "summary": "截至2026-06-27，在美中利差倒挂-265bp与低通胀下，央行转向境内结构性工具与离岸FIMA Repo防线，稳定了境内流动性（套利流出降90%至7.9B RMB）。这巩固了A股以高股息国企（海油6.2%）为防守、硬科技再贷款为成长的“滞胀杠铃”配置，抑平了368.9B两融及35B雪球负Gamma踩踏风险。优化组合VaR降至20.425%。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.75, "matchedTerms": [ "A股", "两融", "雪球", "Gamma" ], "href": "" }, { "date": "2026-06-27", "title": "在两融强平风暴下，跨资产配置必须清退高贝塔TMT并超配低杠杆红利与黄金，结合零成本科创50认沽价差领口期权套保，以非线性凸性将组合回撤死守在-5.0%红线以内。", "summary": "针对境内TMT板块面临的1370亿至2230亿元两融清算压力，定量压力测试了跨资产杠铃优化与期权对冲策略。将长债与科技降配至10%与5%，超配高股息红利（30%）、黄金（15%）与大宗商品（20%）及现金（20%），可使Stressed组合收益率升至8.05%，最大回撤限制在-4.75%。同时设计科创50 ETF期权“买入95%Put/卖出85%Put/卖…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "两融", "红利" ], "href": "" }, { "date": "2026-06-27", "title": "在美联储更高更久限制性利率与通胀粘性重估的环境下，应通过缩短久期、战术超配大宗商品构建战术杠铃组合，并利用利率/汇率期权进行尾部防守，同时在A股内向具备强现金流特征的水电和铜铝替代降本价值风…", "summary": "截至2026-06-28，在美联储维持限制性利率的超预期背景下，传统股债资产负相关性逆转，经典60/40和风险平价组合大幅回撤。本报告建议缩短组合整体久期，战术性将商品权重上调至18%-20%，构建战术杠铃组合，并采用美债互换付款期权与日元外汇期权领口进行双重尾部对冲。中游行业风格上，推荐配置具备现金流防守属性的水电红利以及利用铜铝价差降本的中游白电板块，…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "A股", "两融", "红利" ], "href": "" }, { "date": "2026-06-27", "title": "美债收益率与实际利率高位下A股核心资产风格切换与估值压力测试", "summary": "支持卡片01–04与逻辑链KB的“红利防御+低配长久期成长”杠铃，并将其量化。以美国10Y名义4.38%为全球锚，叠加约+1.8%的CNH贬值/汇率风险溢价与按风格区分的中国ERP，自测算TMT成长r≈11.2%、r−g≈4.2%（久期约24年），红利r≈8.4%、r−g≈6.4%（久期约15.6年）。精确Gordon重估显示：基准情景（实际+50bp、期…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "科创50", "A股", "红利" ], "href": "" }, { "date": "2026-06-26", "title": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局", "summary": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局 在上午科创50跌破关键支撑触发雪球敲入与两融强平后，午间主力ETF资金流入与个股金针反转是否标志着负Gamma抛售压力的阶段性出清？我们应如何利用这一估值剪刀差进行右侧建仓，从红利低波向高Beta科技股进行再平衡？", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 31.25, "matchedTerms": [ "科创50", "A股", "两融", "红利", "雪球", "Gamma", "与两融强平后" ], "href": "" } ], "timeline": [ { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 1, "value": 39, "events": [ { "title": "科创板极端护盘对A股整体流动性的虹吸效应与跨板块资产配置调整 研究日期： 2026 05 22 作者： 首席策略师 (Chief Strategist) 立场： Synthesize…", "summary": "科创板极端护盘对A股整体流动性的虹吸效应与跨板块资产配置调整 研究日期： 2026 05 22 作者： 首席策略师 (Chief Strategist) 立场： Synthesize (综合) 执行摘要 截至 2026 05 22 [S7]，科创50指数收报 1,764.17点 [S1]，极度逼近…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球宏观压力测试：科创板极端护盘外溢效应与跨境流动性虹吸下A股资产配置的脆弱性 研究日期： 2026 05 22 作者： 全球宏观分析师 (Global Macro Analyst)…", "summary": "全球宏观压力测试：科创板极端护盘外溢效应与跨境流动性虹吸下A股资产配置的脆弱性 研究日期： 2026 05 22 作者： 全球宏观分析师 (Global Macro Analyst) 立场： Stress test (压力测试) 执行摘要 截至 2026 05 22 [S1]，科创50指数收报 1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "两融资金去杠杆已通过账户交叉质押与VaR风控限制从TMT扩散至国防军工、新能源等高拥挤成长板块，必须通过‘分化杠铃’策略将大仓位配置于长江电力等红利安全岛、将科技仓位战术收拢至受国家…", "summary": "本卡片系统综合了A股去杠杆负螺旋对全市场的跨行业传染风险。我们测算指出，散户账户交叉质押与VaR风控触发将两融出清压力从TMT扩散至国防军工、新能源等高拥挤成长板块。科创50指数1,750点关口面临882亿元密集敲入与平仓卖压，但在国家队专项买盘防护下形成有底成长资产。我们建议采取‘分化杠铃’策略…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "在科创板两融信用收缩与Q3迎来1200亿解禁洪峰的微观去杠杆环境中，大健康精选篮子建仓必须实施『恐慌底分批-解禁潮吃货』的节奏化策略，并将有效暴露压缩至18%-20%以内，辅以Bet…", "summary": "截至2026-05-25，本卡对创新药出海、CXO地理标签及高端器械替代的精选篮子（联影、迈瑞、奕瑞、百济等）进行A/H股映射与估值管理。A股两融余额见顶于2.87万亿元，科创50在1,750点承受去杠杆与432亿雪球敲入压力。策略建议在科创解禁潮与两融风暴中分三阶段节奏建仓（1-3月恐慌期买30…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国家队'指数级'护盘掩盖了承做商对冲、两融杠杆与量化因子衰减三条链条'个股流动性级'的尾部敏感性,科创50离开1,750不等于系统离开生死线。", "summary": "今日科创50大涨3.56%脱离1,750雪球生死线，但只是延迟而非解除风险——半导体权重票+国家队ETF把指数托起的同时，承做商对冲仓位被动收缩、两融余额冲至2.9万亿历史峰值[S3]、量化中性产品超额已连续两季度回吐[S5]。'非对称虹吸'体现在:资金流入集中于科创50权重前10票与红利端，流出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "布油跌至$96.71的地缘纸面缓和无法撼动美联储沃什时代收紧期限溢价的鹰派核心底色，跨资产‘高偏斜’结构正面临大宗红利挤兑、科创信用去杠杆和日元套利平仓的非线性崩塌风险。", "summary": "本报告揭示霍尔木兹海峡地缘缓和传闻导致油价暴跌14.36%至$96.71，但由于物理复航滞后及极度去库，这一纸面利好完全无法抵消新任美联储主席Warsh收紧资产负债表、驱动10Y名义美债至4.59%及实际利率至2.20%的鹰派政策影响。‘高价高偏斜’特征暗示全球大类资产结构极为脆弱，A股科创50在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘溢价出清与美元紧缩周期共振，正迫使A股在2026-05-26节点进行暴烈的跨风格再平衡：资源红利多头平仓与科创板杠杆负反馈将主导近期交易，资金将向隔绝内需疲弱与关税冲击的“海外闭…", "summary": "在2026-05-26，油价暴跌至96.71美元与美债实际利率居高（2.20%）的双重冲击下，A股资源红利资产面临严重的多头踩踏风险，拥挤度已突破95%分位数。同时，科创50在1,750点上方积累了庞大两融杠杆，面临强平去化压力。策略上建议规避大宗红利与高杠杆科技，防御重心转向低拥挤的“海外闭环A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 6, "neutral": 1, "value": 14.6, "events": [ { "title": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构", "summary": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构 USD/JPY 145 Shock and Margin Deleveraging Reshaping the A-Share True Barbell USD/JPY下探145引发中资境外AT1永续债置换成本飙升，直接压降上市银…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构", "summary": "145去杠杆冲击下金融红利受损与A股两融收紧下的‘真杠铃’极化重构 USD/JPY下探145引发中资境外AT1永续债置换成本飙升，直接压降上市银行NIM约3-5bp，动摇了‘金融红利’板块绝对防线；而科创50两融拥挤度高达4.9%，折算率下调引致250亿-450亿平仓压力，并威胁到敲入线挂接的27…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股当前处于“空心科技虹吸”导致的流动性黑洞期，应警惕科技股杠杆踩踏，向“物理红利”（大水电、核电）防御端极化靠拢。", "summary": "2026年5月27日，A股呈现极端的“空心科技虹吸”特征，半日成交2.07万亿但超4600股下跌，资金极度抱团CPO板块。受美伊停火导致油价暴跌15%及日元159引发的去杠杆冲击影响，全市场两融及雪球脆弱性急剧上升。建议警惕高拥挤度科技股的平仓风险，将防御端转向由大水电、核电组成的“WLEF物理红…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股算力代理股高估值与高两融杠杆筹码面临严重的戴维斯双杀收缩压力，流出资金正由纯叙事泡沫加速轮动至具备物理硬约束的物理红利（水电/核电）、出口链海外闭环Alpha及结构性短缺材料（铜…", "summary": "支持A股算力代理与海外物理硬件脱钩的判断。微观数据显示，A股TMT板块融资买入占比达17.6%，科创50两融比达4.9%的历史峰值，且面临Q3千亿解禁和折算率下调被迫平仓压力。流出资金呈现双轨轮动：一是流向低杠杆两融（1.2%-1.8%）的物理红利水电/核电，及拥有海外闭环供应链的伟星股份、健盛集…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 4, "neutral": 0, "value": 25, "events": [ { "title": "MSCI调仓本身不是系统性事件，但落在A股两融2.87万亿/物理上限2.90万亿仅1%缓冲的高敏感点上，使得TMT龙头任何-2%以上的跳水都会被'雪球敲入×维保线×流动性真空'三重叠…", "summary": "本卡对前两卡'被动买盘被主动MOC冲抵、TMT次日温和回吐-0.5%~-1.5%'的结论进行压力测试。结论：基线情景(P≈70%)下两融余额单日收缩80-140亿元，2.87万亿/2.90万亿物理上限的300亿元缓冲仍可承受，系统性风险可控；但厚尾情景(P≈20%)下科创50若跌破1,750点雪球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50指数在1,750点附近存在极高的信用与衍生品二次平仓反馈共振深渊，下周两融去杠杆风暴极易打压AI硬件龙头股估值，建议维持45%的红利防御性配置上限并买入虚值认沽期权对冲尾部风…", "summary": "本报告定量评估了科创50指数在1750点附近的雪球敲入（22.0亿元名义本金）与做市商Delta动态对冲（12.0亿元卖压）的二次共振效应。在券商两融折算率收紧（0.60至0.40）压力测试下，一阶总抛压达37.0亿至57.0亿元，极端踩踏下可触发二阶追加平仓反馈卖压119.7亿元，共振总抛压达1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "下周一起把组合从'科技+长江电力单点红利'切换为风险平价/全天候四桶（进攻8/红利45/真分散20/蓝筹22/现金5），用煤炭+一体化油气+电信+大行的6+1红利篮替代已饱和的长江电…", "summary": "本卡（2026-05-29）综合前四卡链条结论：MSCI生效日被动脉冲被外资MOC派发冲抵，两融2.87/2.90万亿仅1%缓冲，科创50若失守1,750将触发雪球-Delta-两融三重共振35亿-177亿抛压（Research note 04）。本卡把'45%红利+OTM认沽'升级为可执行的风险…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在雪球敲入与两融平仓共振的极端风险下，以‘先进封装’结构性机会对冲Beta下行，并以虚值期权锁定尾部风险是当前最优防御路径。", "summary": "截至2026-05-29，A股两融余额超2.9万亿且科技板块极度拥挤。科创50指数逼近1,750点雪球密集敲入区，做市商面临121.5亿元刚性抛压，叠加ETF折价套利，流动性死亡螺旋风险剧增。建议配置14%比例的先进封装‘防守方案四’（5%设备+6%材料+3%存储）并配合虚值认沽期权构建多重防线。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 1, "value": 41.4, "events": [ { "title": "A股两融安全边际充沛，科创50雪球实际规模仅为62亿元且敲入卖压平滑，MSCI被动重估不存在所谓的“流动性真空砸盘”，跨境日元期权对冲完全多余且损耗超额收益。", "summary": "本卡作为金融行业分析师，强烈反驳A股策略师的悲观观点。首先，两融高位有2026年保证金100%新规与2025年风控标准严密保护，平均维持担保比达248.5%，个股需跌60%以上才触及强平线，绝非指数大跌16.92%所能引爆。其次，科创50雪球名义存续实际仅为62亿元且极均匀分布，做市商采用平滑期权…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈", "summary": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 2.94tn RMB Margin Debt & STAR 50 Snowball Cliff under MSCI Passive Rebalancing 今日是MSCI半年度重估正式生效日。虽然散户由于昨日创业板+1…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 报告日期 ：2026年05月30日 分析师 ：A股策略师 ( ashare strategist ) 密…", "summary": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 报告日期 ：2026年05月30日 分析师 ：A股策略师 ( ashare strategist ) 密级 ：内部机密 | 白板研讨卡 01/08 核心结论与概要 (Executive Summary) 今日 (2026…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "NVIDIA业绩引发的高贝塔创新板块相关性飙升已被杠杆出清打破，跨资产风格轮动正向防守现金流与硬科技杠铃化配置演进。", "summary": "本报告分析了2026年5月20日NVIDIA爆表业绩对A股高贝塔创新板块（AI硬件、生物医药、空间经济、具身智能）相关性飙升的驱动机制。指出该共振由杠杆和叙事推动，本质十分脆弱。5月29日两融去杠杆引发科创50指数大跌5.04%至1751点，防守端公用事业暴涨，跷跷板效应明显。建议构建75-80%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈", "summary": "两融余额2.94万亿历史巅峰与科创50雪球临界敲入下的MSCI被动重估大博弈 今日是MSCI半年度重估正式生效日。虽然散户由于昨日创业板+1.96%的V型反弹而呈现极度FOMO，但A股两融余额正高挂于2.9367万亿元历史极值，融资交易占比达10.09%。科创50指数（1,844.25点）距离下方…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 4, "neutral": 1, "value": 32.2, "events": [ { "title": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守", "summary": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守 Defending the 1,750 Lifeline: Snowball Gamma Waves and Margin Call Risk 科创50收于1,751.32点，距离1,750雪球/两融生命线仅一步之遥。若下周破位跌破1…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论", "summary": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论 A-Share Post-Close Sentiment Fission: MSCI Rebalancing vs. Margin & Snowball Risks 今日A股演练了11:27分V型奇袭与科技大挤空，但成交…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论", "summary": "A股盘后社交情绪裂变：明日MSCI被动申购与两融拥挤高地、雪球崖壁的博弈辩论 今日A股演练了11:27分V型奇袭与科技大挤空，但成交量缩量2704亿元。盘后社媒上散户情绪在英伟达/Marvell史诗财报刺激下极度亢奋，期待明日MSCI半年度生效日的被动天量申购。然而，两融高挂于2.9367万亿元，…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "截至2026-06-01，科创50在1,750点面临882亿元级联式去杠杆风暴，必须将高股息防御资产配比升至60%并利用衍生品对冲以构筑极限避险防线。", "summary": "截至2026-06-01，科创50收于1,751.32点，距1,750点生死线仅一步之遥。若跌破将瞬间触发144.0亿元雪球密集敲入（432.0亿元市商Delta卖压）及450.0亿元两融平仓，形成882.0亿元级联去杠杆抛压。我们建议将红利配置上限从45%升至60%，重构以长江电力（度电收益0.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 4, "neutral": 2, "value": 36.6, "events": [ { "title": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守", "summary": "科创50指数1,750点生死防线下的雪球爆仓与两融清算极限防守 科创50收于1,751.32点，距离1,750雪球/两融生命线仅一步之遥。若下周破位跌破1,750，我们如何防范882亿元的级联式去杠杆抛压？A股红利与成长的配置比例应如何调整？ Prior related institute wor…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "概述 作为A股策略师（ ashare strategist ），针对看板第三步主题 “联合宏观矩阵下A股信用与衍生品结构性杠杆风险特征” ，我已对科创50指数回测1750生死线下的两…", "summary": "概述 作为A股策略师（ ashare strategist ），针对看板第三步主题 “联合宏观矩阵下A股信用与衍生品结构性杠杆风险特征” ，我已对科创50指数回测1750生死线下的两融去杠杆与雪球衍生品敲入进行了深入的定量压力测试与模���仿真。 在此项研究中，我全面 支持 首席经济学家（卡片 0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在监管穿透迫使做市商瞬时撤流的环境中，执行流动性本身就是资产类别，组合应先买红利低波/公用事业、防御银行和现金流消费，后等确认再回补科创/TMT与茅指数。", "summary": "截至2026-06-02，本卡综合前五卡压力测试，认为下一阶段A股应从Beta与期权性切向现金流与执行确定性：超配红利低波和公用事业，保留银行但低配券商，消费仅保留防御现金流，低配科创和高Beta TMT；只有当两融余额、ETF流向、成交额和科创50价格同时修复后，才逐步回补科创/TMT和茅指数。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈", "summary": "MSCI生效日大决战：两融创纪录新高与雪球敲入边缘下的非对称博弈 MSCI Rebalancing Effective Date: Margin Debt Record-High & Snowball Knock-in Cliff under Asymmetric Play 今日（5月29日）MS…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 6, "neutral": 1, "value": 23, "events": [ { "title": "在美联储加息与人行降准的联合概率矩阵下，A股必须从传统哑铃策略转向“防御底仓+对冲Beta”模式，规避高融资与高解禁共振的科创标的。", "summary": "由于卡片04数据受损，本报告直接综合卡片01与03。针对科创50指数在1750点的雪球敲入与2.9万亿两融平仓共振风险，本报告提出A股风格轮动与行业配置优化方案：将原10%的科技敞口转为“对冲型科技”，并将高股息与出口制造权重分别提升至35%与25%，以应对6月3237.5亿元解禁悬崖。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "台积电关税成本转嫁推升GB200整机柜售价12.3%及云厂商TCO 8%，将引爆北美云厂商对 Blackwell 订单10-15%的下修，并迫使A股出海非GPU链降价5-8%，导致其…", "summary": "本报告针对2026-06-02台积电延期及关税转嫁事件，量化分析了其对英伟达GB200定价（上涨12.3%）与北美云厂商TCO（上升8.0%）的影响，指出云厂商可能下修订单10-15%。A股光模块、PCB、液冷出海链面临5-8%的压价与220-350 bps的毛利率挤压，伴随两融（550亿）及科创…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在两融高拥挤（占比22%）且维持担保比例受压的背景下，核心标的跌超40%将引爆每日312.4亿元的“多杀多”共振抛压并锁死流动性；科创50跌破1500点将引发全市场ERP暴涨170 …", "summary": "截至2026-06-02，本报告针对中际旭创与新易盛跌幅超40%触及融资平仓线（580-640元/340-380元）的情境，压力测试了两融爆仓、“多杀多”共振与科创50跌破1500点的影响。测算显示，两股两融余额合计550亿元，跌40%将导致99.7%的账户爆仓，在T+3窗口产生每日219.4亿元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在尾部保护兜底后，权益组合应拒绝过度保守以防双重损耗，通过超配美股成长与A股“真杠铃”资产，同时出清日本及高两融拥挤资产，实现最大化Melt-Up参与。", "summary": "在固定季度成本上限0.85-1.35% NAV的SPX/VIX/CDX尾部保护配置后，权益组合应维持较高Beta（目标0.95，区间0.90-1.05）以对冲保护费损耗。区域配置保留美股（无基差期权对冲）与A股（低相关多元化避风港），坚决出清处于日元平仓风暴眼的日本股票。风格上重构A股“真杠铃”（…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 5, "neutral": 0, "value": 27.4, "events": [ { "title": "在科技极度超卖且全球AI硬件共振走强下，科创50雪球及两融数据口径冲突加剧数据迷雾，开盘盲目机械式减仓易割肉在阶段底部，应在开盘承接确认及流动性双重确认后再行抉择。", "summary": "在科技动量Z-score降至-2.8及科创50雪球存量口径冲突（882亿对432亿及240-310亿）背景下，美股MRVL（+32.52%）和HPE（~+30%）强势提供外部催化。策略上建议放弃开盘机械减仓，转为等待开盘承接确认。利用开盘前15-30分钟观察资金承接，跟踪当月合约贴水（阈值40 B…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建全球产业周期景气度仍在加速并非泡沫见顶，但A股本地面临极高的两融多杀多挤兑风险，必须进行仓位控制及衍生品尾部对冲。", "summary": "中际旭创登顶沪深300第一权重股与Marvell单日大涨32.52%标志着AI基础设施资产登顶中美双市场。尽管全球算力与网络需求（旭创Q1利润增262%）和英伟达 endorsement 证明主升浪仍未终结，但A股本地TMT两融余额（5800亿元，旭创融资比率8.50%）呈现极端拥挤。需通过限额至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件主线已足以把成长仓位上调至12%-14%，但红利端只应降到52%-55%而非清仓，组合仍需现金和对冲压舱。", "summary": "截至2026-06-04，本卡支持将AI硬科技成长仓位从8%-10%的防守下限上调至12%-14%，但15%仍是硬上限。加仓资金主要来自红利端从60%下调至52%-55%，而不是撤掉现金和指数对冲。理由是海外AI硬件订单与A股CPO/半导体链形成明确映射，但A股宽度仍弱、科创板两融仍在收缩，说明只…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股红利低波与银行板块整体交易热度高涨，但其融资杠杆极低且公募重仓比例依然受限，无踩踏式风格切换风险，资金流向正向银行内部“高低切换”演变。", "summary": "截至2026-06-03，本卡支持行业分化策略。多维拥挤度测算表明，中证红利低波指数CCI为73.8%，处于温和拥挤区间；中证银行指数CCI为61.1%，处于合理配置区间。两融参与度（4.5%-5.2%）和公募基金银行持仓（5.1%，超配-4.5%）均处于低位，无踩踏切换风险，但板块内部“高低切换…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 5, "neutral": 0, "value": 28.8, "events": [ { "title": "博通指引转保守与A股流动性三压共振，科创50面临1720点雪球破位风险，需警惕系统性去杠杆引发的二次筑底。", "summary": "博通（AVGO）隔夜暴跌13-15%打破了AI硬件订单无限增长的叙事，引发全球半导体去Beta。A股科创板正面临4.9%高两融占比、6月3332.85亿解禁洪峰以及科创50指数（现报1726）逼近1720雪球密集敲入区的“三体问题”。若破位将引发做市商负Gamma抛售。建议维持45%红利底仓，并买…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？", "summary": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？ Broadcom AI Guide-Down Triggers Global Semi Sell-Off: Will A-Share Tech Break Support Amid Leverage & Lo…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高频运价暴涨与海运股两融去杠杆踩踏的‘实物韧性与筹码破损’定价博弈", "summary": "高频运价暴涨与海运股两融去杠杆踩踏的‘实物韧性与筹码破损’定价博弈 Freight Rate Surge vs. Shipping Margin De-leveraging: Physical Resilience vs. Flow Strains SCFI本周飙升15.9%至2,571.73点，…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高频运价暴涨与海运股两融去杠杆踩踏的‘实物韧性与筹码破损’定价博弈", "summary": "高频运价暴涨与海运股两融去杠杆踩踏的‘实物韧性与筹码破损’定价博弈 SCFI本周飙升15.9%至2,571.73点，且卫星数据证实新加坡港仅0.8天低拥挤度，说明绕行好望角吨海里需求具有刚性，但中远海控等海运股却因两融去杠杆与MSCI被动流出，盘中跌超1.02%并伴随主力净流出。在6月3332.8…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 7, "neutral": 0, "value": 18.2, "events": [ { "title": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？", "summary": "博通AI指引保守触发全球半导体去Beta，两融与解禁夹击下A股科技防线是否面临二次筑底？ 博通（AVGO）隔夜暴跌12%-15%打破了AI硬件订单无限高增的完美预期，而A股CPO与半导体龙头正面临历史高位两融（科创板4.9%占比）和6月3332.85亿的解禁洪峰。如果科创50指数被动向1750/1…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50流动性红线击穿：散户无感与机构负Gamma/两融强平的非对称定价博弈", "summary": "科创50现价约1,740，距1,720阈值仅约1.15%，情景已实时贴身。雪球敲入150亿、做市商32亿、次级负Gamma 133.5亿合计约315.5亿元机械式凸性卖压，叠加115%强平线（标准为130%）抹去T+1/T+2缓冲，构成非线性去杠杆。核心判断：国家队4,000点护盘与压力指数错配—…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融、雪球与股指期货去杠杆共振时，防线必须从静态套保改为预留现金、动态降杠杆和红利锚容量配给，否则对冲会先于底仓卖出能力而失效。", "summary": "截至2026-06-06，本卡压力测试认为，A股尾部风险的核心不是缺少IM、期权或红利资产，而是底仓跌停锁死时，对冲端被保证金和盘中反弹先行强平。最新可得两融口径显示，2026-06-04沪深融资余额约2.89万亿元；若3%-6%被动退出，可能形成约867亿-1734亿元卖压。长江电力可作为红利锚…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50流动性红线击穿：散户无感与机构负Gamma/两融强平的非对称定价博弈", "summary": "科创50流动性红线击穿：散户无感与机构负Gamma/两融强平的非对称定价博弈 STAR 50 Support Break: Retail Obliviousness vs. Dealer Negative Gamma Hedging & Margin Squeeze 科创50跌破1750点流动性红…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 6, "neutral": 0, "value": 33.6, "events": [ { "title": "外部高利率与全球去杠杆冲击下，A股面临硬科技两融去杠杆与解禁潮共振风险，资产配置应剪枝高位拥挤科技与昂贵资源，退守低估值高股息周期价值与短端利率债以优化组合VaR。", "summary": "本报告分析了美联储维持“更高更久”利率对A股资金与风格轮动的影响。美债收益率升至4.75%和日元平仓去杠杆加剧了北向资金流出压力（预计600亿至900亿元），推高离岸HIBOR至8.0%-10.0%。两融杠杆高企（2.91万亿元）且高估值科技拥挤，科创50击穿1750点流动性红线触发150亿雪球敲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在融资杠杆处于极值且科创50下面临雪球敲入踩踏的情景下，超配低贝塔、高分红的避风港资产长江电力是防守去杠杆和风格转换中对冲系统性风险的最优策略。", "summary": "截至2026年6月4日，A股融资余额达2.90万亿元高位，科创50自高点回调至1668.33点，逼近1125-1360点的雪球敲入密集区。一旦触发敲入，券商平仓多头将引发市场二次下跌并导致融资盘连环爆仓。建议超配长江电力（Beta为0.09，股息率3.7%，承诺70%以上分红率）等低贝塔红利资产，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在系统性流动性危引发避风港资产补跌与相关性收敛的背景下，跨资产配置至实物黄金与短债，并配合科创50 ETF Put与IM期货空头对冲，是防御VaR穿透并平衡权益融涨弹性的最优解决方案。", "summary": "针对2026年6月7日A股2.90万亿两融余额极值及雪球敲入踩踏引致的系统性流动性危机，本压力测试表明长江电力等避风港资产在挤兑期存在补跌风险，相关性收敛导致未对冲组合10日损失达-16.60%，99% VaR穿透至32.33%[自测算]。本方案设计了科创50 ETF期权与IM期货空头对冲包，并对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50流动性红线击穿：散户无感与机构负Gamma/两融强平的非对称定价博弈", "summary": "科创50流动性红线击穿：散户无感与机构负Gamma/两融强平的非对称定价博弈 科创50跌破1750点流动性红线，已触发约150亿元雪球敲入与做市商32亿元现货抛压，但股吧散户端对此风险的提及率为0。若指数进一步探底1720点关键阈值，由133.5亿元次级负Gamma对冲卖压与新版115%强平线共振…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 4, "neutral": 1, "value": 42, "events": [ { "title": "[auto-handoff] 科创50指数雪球敲入负 Gamma 对冲抛压与两融折算率下调去杠杆踩踏模拟", "summary": "致 首席风控官： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 73279e4e-699e-4d9e-9366-ea168709088a 结果摘要: Wrote handoff_response.md (14450 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中际旭创明日解禁直接抛压微弱，但在全月大额解禁、两融折算率调降预期及科创50雪球敲入共振下，股价若下跌13.33%-15.40%极易引爆科技板块非线性两融强平螺旋，需严格规避高两融占…", "summary": "本报告针对中际旭创明日（2026年6月8日）解禁及半导体/CPO两融杠杆进行压力测试。中际旭创明日解禁32.45万股（约4543万元），仅占日成交额1.14%，直接冲击有限。但在全月3332.85亿元大额解禁、电子通信行业两融高达7389.32亿元、两融折算率收紧拟调降至0.4以及科创50破位敲入…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融、ETF与行业成交结构已部分确认A股政策顺周期切换，但宽度不足，仍属头部权重驱动的早周期信号。", "summary": "截至2026-06-08，两融、ETF和行业成交三类微观证据共同支持卡片01、02的政策顺周期切换判断：杠杆资金在非银金融、银行、电力设备、军工、机械的融资买入占比抬升，在食品饮料、医药、地产下移；ETF净申购集中在券商、银行、军工、高端装备和央企红利/中特估，纯地产与白酒ETF净赎回；金融与先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融去杠杆与估值收缩背景下，A股行业领导力与风格轮动路径研判", "summary": "本卡片综合前四卡，定量测算两融去杠杆下A股风格轮动。结论：向'物理红利'(水电/煤炭/银行/高速)与'出口Alpha'的轮动真实且理性，但属哑铃式防御性再平衡(1-3个月超配)，非范式切换。科技12-15%回撤将强平约1100-1500亿元两融[自测算]；中证红利股息率4.85%对10年国债1.7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 8, "neutral": 1, "value": 19.4, "events": [ { "title": "两融天量与破位去杠杆压力下A股科创50零Gamma防线踩踏与红利防守失效之辩", "summary": "两融天量与破位去杠杆压力下A股科创50零Gamma防线踩踏与红利防守失效之辩 A-share STAR 50 Zero-Gamma Liquidation Cascades vs Dividend Safe-Haven Failure Debate 科创50失守1720生死线，2.89万融两融会否…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融天量与破位去杠杆压力下A股科创50零Gamma防线踩踏与红利防守失效之辩", "summary": "两融天量与破位去杠杆压力下A股科创50零Gamma防线踩踏与红利防守失效之辩 科创50失守1720生死线，2.89万融两融会否发生被动平仓踩踏？在-230bp中美利差与贬值压力下，大行/红利股的防守回报是否会被汇率损耗吞噬？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A股科创50跌破零Gamma生死线后两融强平与雪球级联踩踏的微观杠杆压测", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 23d2a86c-8b70-46ce-8620-e44b5d309d0d 结果摘要: Wrote handoff_response.md (3101 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50破位会触发局部去杠杆压力，但在流动性缓冲和银行相对防守尚存的情况下，不宜直接判定全市场两融被动平仓踩踏已经启动。", "summary": "2026-06-08，本卡判断科创50跌破1720更像局部科创/半导体去杠杆，而非全市场两融踩踏。两市融资余额为2.89万亿元，科创板两融为3805.56亿元，压力真实但尚未失序；银行在6月5日逆势上涨1.33%，说明红利/大行对人民币组合仍有相对防守作用。美元口径风险来自中美10年期利差约-28…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 6, "neutral": 1, "value": 32.8, "events": [ { "title": "离岸CNH流动性收紧迫使北向杠杆资金去杠杆，这与境内两融平仓及雪球负Gamma抛售共振，加剧了科创板及高弹性权重股的流动性压力。", "summary": "CNH HIBOR飙升至7.30%抬高了离岸融资成本，导致北向杠杆资金因年化-2.8%的负carry面临去杠杆压力。这与境内融资折算率下调（966亿保证金缺口）及科创50跌破1600点零Gamma防线引发的270亿雪球敲入卖压共振，导致外资抛售半导体与TMT高弹性权重股。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO引发的全球流动性‘真空’是导致科创50破位1600点的核心外因，破位后将诱发两融盘非线性踩踏。", "summary": "SpaceX IPO的450-750亿美元流动性虹吸正通过中概股套利链条向A股传导。压力测试显示，若科创50跌破1600点关键支撑，将触发券商折算率下调与融资盘抢跑的共振。虽然当前距离1310点强平线仍有空间，但技术破位后的非线性踩踏风险从5%上升至20%以上。建议利用看跌价差对冲，并保持15%现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO 被动虹吸与A股科技高两融杠杆的非线性踩踏风险辩论", "summary": "SpaceX IPO 被动虹吸与A股科技高两融杠杆的非线性踩踏风险辩论 SpaceX IPO Passive Outflow vs. Onshore Tech Margin Leverage: Stampede Risk Debate SpaceX 6月12日IPO面临450亿资金被动虹吸，而A股…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO 被动虹吸与A股科技高两融杠杆的非线性踩踏风险辩论", "summary": "SpaceX IPO 被动虹吸与A股科技高两融杠杆的非线性踩踏风险辩论 SpaceX 6月12日IPO面临450亿资金被动虹吸，而A股TMT融资盘高达7382亿。今日科创50收于1663点逼近强平红线，这是否会在本周末触发两融非线性去杠杆踩踏？我们应如何在开盘前调整软硬件配置？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 8, "neutral": 3, "value": 28.2, "events": [ { "title": "把拥挤+杠杆+解禁的A股AI硬件龙头当作有效风险2–3倍于名义权重的凸性敞口块——按风险预算定档（6–8%降至3–5% NAV），并用国债久期+黄金+红利低波杠铃分散、用科创50/I…", "summary": "锚定2026-06-10，本卡承接第05卡，把方向判断转为组合构建。结论：这批AI硬件龙头不应当方向性持仓，而是凸性、有硬上限、肥尾的敞口块——因拥挤beta、两融顺周期、7月解禁三重放大，其有效风险贡献约为名义权重2–3倍[自测算]。故应按风险预算定档（解禁前上限6–8% NAV，7/20–24…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在美联储鹰派重定价与科创板局部杠杆脆弱性的双重挤压下，A股配置应果断转向低杠杆红利防御，保留科技龙头仅作为极低仓位的期权观察。", "summary": "截至2026-06-10，科创50指数（1652.22点）月内跌幅达5.66%，显著跑输红利指数（-2.10%）。两融数据显示资金正从电子（半导体）板块剧烈流向能源、银行等高股息行业。虽然科创50尚未触及1468.6点的广谱强平线，但局部杠杆压力已迫使配置风格从高弹性成长向低杠杆防御切换。建议压低…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-12若触发TMT两融去杠杆，行业配置应从“融资驱动AI硬件”转向“现金仓核心科技+高流动性红利防御”，但不应追逐所有高股息资产。", "summary": "截至2026-06-10，本卡支持前序“06-12前降低TMT融资仓位”的方向，但将结论收窄为融资流动性切换：科创50调仓可能使AI硬件抢跑资金借被动买盘退出，并通过两融账户担保品压力传导到TMT整体。红利风格将成为资金承接方向，但不是无差别主升；长江电力更像高流动性防御锚，而非低拥挤标的。建议降…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "长电拥挤度处中后段而非极端，TMT外溢仅能推升3%-6%且估值缓冲未破，真正新风险在红利低波ETF同质化叠加利率反转，而非长电本身。", "summary": "截至2026-06-10，长江电力两融占流通市值约1.81%，远低于科创50中AI硬件标的10%级极端值；公募超配已近2024-2025红利峰值，边际买盘依赖被动ETF与保险OCI。从TMT两融7382亿元中估算被动去杠杆10%-15%、其中30%-40%切换至红利，约220-440亿元流入红利篮…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 56.6, "events": [ { "title": "CPO两融去杠杆会放大创业板和科创50的流动性冲击，但只有在红利底座同步失效、相关性收敛的系统性赎回情景下，才会击穿方案四-2.6169%的VaR线。", "summary": "2026-06-11：CPO基本面下修会通过中际旭创、新易盛、天孚通信合计732.02亿元融资余额向创业板权重股传导，并与科创板3742.22亿元融资余额形成成长板块二次卖盘。但在方案四下，0.25% CPO仓位本身不是击穿-2.6169% VaR的主因；只要红利55%底座维持低波低相关，局部强平…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国家队科创50防守策略在‘一级申购优先’与‘二级折价才买’的限制下将形成严重的防御真空与套利漏洞，导致二级市场半导体与AI龙头在防线建立前遭遇流动性错配引发的估值踩踏。", "summary": "本卡片对科创50在‘宽基先行、科创后置’策略下的执行路径进行压力测试。分析指出，‘一级申购优先’机制存在显著的算法建仓时滞，面对两融平仓与雪球敲入形成的710亿元级联抛压（对前五大半导体巨头达106.5亿元/天），国家队12亿元/天的日均算法买盘存在8.88倍的流动性赤字。此外，‘二级折价才买’规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 0, "value": 46, "events": [ { "title": "硅基算力硬件去杠杆与物理AI具身智能风格轮动持续度", "summary": "硅基算力硬件去杠杆与物理AI具身智能风格轮动持续度 科创50大跌4.30%收于1596.57点并触发雪球敲入与两融踩踏，资金大举流入逆势涨停的具身智能及机器人精密零部件。如何评估科技板块内部这一“从硅基算力到物理实体验证”的主题风格切换持续性？两融出清底部在何处？ Prior related in…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基础设施的双重约束注定了硅基硬件的中期去杠杆，而具身智能因缺乏短期EPS支撑仅能作为战术性轮动，A股科技主线的确立仍需等待两融出清。", "summary": "截至2026-06-12，本卡综合了电力瓶颈、CoWoS产能与潜在巨额减值的约束，判定A股硅基算力正经历中期结构性去杠杆。同时，由于具身智能2026年出货量极小，无法提供实质性EPS支撑，其轮动仅为战术性风险再分配，不具备升级为季度级主线的条件。策略上应维持45%红利防御，耐心等待科创板两融在15…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "本地两融回落、6-9月科创/创业解禁墙和既有0折算率池共同压制A股科技再加杠杆，海外risk-on只能提供交易窗口，不能支撑仓位上调。", "summary": "2026-06-12数据核实支持上一卡的本地约束判断，但需要修正表述：6月券商公告未显示新的科技股大面积折算率下调，真正约束来自既有0折算率池。两市融资余额较2026-05-29下降471.57亿元；6-9月全A解禁约9,300.88亿元，科创板+创业板明细合计5,522.78亿元；兴业证券202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新一代信息技术链的超额收益已从泛算力概念转向‘物理硬约束（电力/连接）组件出海’与‘底层芯片（GPU/MCU/MEMS）国产替代业绩拐点验证’，需警惕2026-06-13所处周期的科…", "summary": "在2026-06-13评估中，针对新一代信息技术产业链4.3%的极低激活率，本报告系统梳理了5个低覆盖但具有强逻辑的A股/科创板标的。重点分析了摩尔线程（688795.SH）的GPU国产替代与首次扭亏、金盘科技（688676.SH）的电力物理瓶颈与出海溢价、鼎通科技（688668.SH）的高速连接…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "A股杠杆结构在100%保证金规则下韧性显著，红利退出压力可控，风格切换的主要风险在于科创50杠杆增长过快引发的情绪过热，而非两融爆仓。", "summary": "针对2026-06-14的A股杠杆结构压力测试显示，红利板块虽处于90%拥挤度高位，但得益于100%融资保证金规则，具备35%的价格安全垫，风格切换预计表现为“有序撤离”而非“平仓崩溃”。科创50融资余额触及自由流通市值5%的情绪极值，虽理论容量巨大，但需警惕情绪过热。整体质押风险局部化且受控，市…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 4, "neutral": 0, "value": 56.8, "events": [ { "title": "基于2026-06-15，国产算力+518亿元新增订单支撑下，A股科技板块两融与机构仓位重构具备坚实基本面，科创50重回1600点零Gamma上方将触发右侧杠杆盘强力追入与风格轮动反…", "summary": "基于2026-06-15，本报告定量评估了国产算力爆发背景下，A股科技板块（芯片设计、算力设备、PCB）的两融与机构仓位再平衡。大厂H2新增的518亿元国产算力订单有效对冲了两融收缩1200亿元的短期流动性冲击。公募基金TMT配置预计将从14.2%上修至19.8%，带来+2520亿元增量资金。科创…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "DR007利率在央行结构性对冲下企稳在1.48%以稳固信用环境，但两融高位去杠杆仍在电子和科创板诱发踩踏，跨资产向物理红利与黄金再配置（策略3）是锁定VaR风险（-18.15%）并保…", "summary": "截至2026-06-15，央行对冲稳定DR007在1.48%，AAA信用利差仅增5bp。但两融经历高位去杠杆，从2.94万亿元收缩至2.82万亿元，科创50击穿1600点零Gamma线并强制平仓35.75亿元，由国家队日均承接3亿元平稳。避险资金轮动至长江电力等物理红利（两融占比1.81%，CGB…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创板去杠杆冲击短期股价但不破坏国产半导体设备及AI硬件的长期成长逻辑——龙头与AI芯片设计估值修复~20%即可加仓，二线与HBM配套需-30%以上才有安全边际。", "summary": "截至2026-06-15，科创板两融由~2,940亿元降至~2,820亿元、科创50击穿1,600点零Gamma线，对电子板块形成真实短期冲击，但不足以系统性下修国产半导体设备与先进制程龙头capex。设备龙头（北方华创订单覆盖15个月、现金~150亿元）、AI芯片设计（寒武纪2026Q1+312…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网容量及绿电约束倒逼算力资本开支分流并延后，建议下调A股“科技进攻端”并超配“实物红利端”（区域火电与电网设备），并设立以“电费上浮”与“并网延期”为核心的双轨尾部风险触发器。", "summary": "截至2026-06-14，鉴于一线城市电网并网物理瓶颈（排队长达24-36个月）与绿电指标约束，导致下半年算力资本开支落地延后且向电力配套分流（占比自15%升至25%），我们建议下调A股“科技进攻端”权重（自30%降至26%，出清高估值两融拥挤的GPU租赁SPV，转向先进封装和受推理通缩驱动的AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 6, "neutral": 1, "value": 41.2, "events": [ { "title": "两融折算率下调引致的350亿元刚性去杠杆将重创高贝塔科技硬件板块，推动资金向筹码结构极净、收益率稳定的‘物理红利’与‘成本通缩’AI SaaS板块非对称旋转。", "summary": "截至2026年6月17日，本报告评估了两融折算率从0.6下调至0.4对A股两融规模和市场结构的影响。券商分批落地该政策将引发250亿至450亿元（中枢350亿元）的刚性融资收缩。科技硬件（电子/计算机）板块融资余额高达5608.81亿元，面临50亿至112.5亿元的平仓卖压。在负Gamma共振下，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "SPCX期权首日即面临1-DTE/0-DTE短久期希腊字母爆发，在FOMC决议与周四三巫日共振下，做市商Short-Gamma/Vanna对冲踩踏将在低VIX（16.20）环境下非线…", "summary": "基于2026-06-17，分析表明SPCX期权首日上市、FOMC决议和周四三巫日共振，且周五休市，将大幅放大美股跳空及科技股下行风险。在VIX降至16.20的低波动伪装下，SPCX首日交易即面临极端的1-DTE/0-DTE Short-Gamma和Short-Vanna对冲踩踏。若美股科技股深度调…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月15日成长行情是可交易挤仓而非成长牛市重启，应维持半导体/CPO有上限参与，并把科创50 1550/1516作为低置信但高冲击的触发区间。", "summary": "截至2026-06-16，本卡对6月15日成长爆量轮动给出压力测试结论：反弹可交易，但尚非可持续配置信号。6月A股解禁3332.85亿元、影石创新约403.91亿元；5月25日两融余额曾达29252.09亿元，但截至6月12日已降至28734.82亿元。科创50 6月15日收1748.33点，距1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美股科技股回调引发的A股流动性挤兑风险，其核心痛点在于860亿两融‘预警区’杠杆资金的踩踏，而非雪球产品的局部敲入。", "summary": "基于2026-06-17，量化分析显示A股两融余额处于2.87万亿元的历史高位，且高度集中于电子与通信板块。科创50指数目前（1,758点）距离雪球密集敲入带（1,600-1,700点）仅一步之遥。压力测试表明，若科技股回调20%，约860亿元处于‘预警区’的杠杆资金将面临强制平仓压力，其流动性挤…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "截至2026-06-17，A股市场结构不支持确认电子与科创50去拥挤，只支持继续持有但带上限的硬科技风险预算。", "summary": "截至2026-06-17，本卡压力测试第05卡再平衡方案：融资余额、ETF资金流和市场宽度都未确认电子与科创50拥挤度降温。科创板两融余额仍连续增加，科创50和电子ETF仍有资金承接，2026-06-17科创50大涨但超过3700只股票下跌，显示行情更窄而非更健康。结论是维持70/20/7/3组合…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "A股电子/TMT融资仓不能等到-18%才降杠杆；价格、ETF赎回、两融拥挤与券商风控线共振时，应提前强制削减融资敞口。", "summary": "截至2026-06-18，本卡片把-18%政策底压力测试转化为执行触发器：科创50、创业板、申万电子中任意2个破位，叠加股票ETF连续大额净流出、电子融资净买入在弱价中继续居首、ETF/科创两融余额反常上升，以及券商115%即时平仓线或双创集中度附加线收紧，应在-18%失败前强制降融资。政策底只有…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 4, "neutral": 0, "value": 54.4, "events": [ { "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 日报总编： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: cd8f2a28-c7a9-4dc8-aebc-e5ed66354208 结果摘要: Wrote handoff_response.md (17618 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股红利资产尚未逼近自身杠杆和成交极值，但在全市场两融余额接近历史高位时，应把它们视为可被动提款的流动性资产，并对15.85%红利袖子设置2.0-3.0个百分点的动态减仓触发器。", "summary": "截至2026-06-20，本卡支持红利低波袖子仍可配置，但将风险从“红利泡沫”重定义为“全市场去杠杆时的流动性提款机”。沪深两市2026-06-17两融余额合计2.9438万亿元，接近历史高位；但14只大型银行、大水电和煤炭长协红利样本在2026-06-18的融资余额仅261.87亿元，自由流通市…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在30% A股配置中，电网与散热设备凭借低融资拥挤和强出海景气度应立即建仓，而国产半导体在两融高负荷、折算率下调以及雪球敲入风险完全出清前应保持现金等待，通过分步建仓捕捉12.48%…", "summary": "截至2026-06-20，针对30% A股配置，本报告建议采取“先物理、后算力”的分步建仓策略。20%电网/散热板块应立即建仓；10%国产半导体板块因融资余额达7382亿、两融折算率下调及1516-1550点雪球敲入威胁，面临严重的左尾去杠杆风险。待半导体释放-15%回撤且杠杆出清后建仓，可使A股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "期权做市商的 Gamma 挤压是推动 3.31 万亿天量科技股暴涨的核心力量，但高位两融极度拥挤与成交量萎缩将引发毁灭性踩踏，国家队必须在 1750 点建立战术防线并在 1550 点…", "summary": "在两市录得3.31万亿天量与3400股下跌的极端分化下，期权做市商Gamma动态对冲交易约达38.5亿元，占科创50等科技ETF日成交超50%，是推动指数Parabolic式上涨的核心引擎。然而，下周一成交量若萎缩至1.8万亿以下，融资盘极度拥挤的四大千元科技股将面临极高的流动性踩踏与强平风险。国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 1, "neutral": 0, "value": 73.8, "events": [ { "title": "跨季流动性挤压导致券商杠杆净利差受挫，压力测试显示券商Beta的重定价弹性高度依赖成交量的平台期企稳，建议目前维持大行红利底仓防御，确认DR007回落至1.50%以下且成交量稳定在1…", "summary": "本报告对跨季后A股成交量与两融余额稳定性进行了压力测试。6月18日成交额冲上3.33万亿元，17日融资余额回升至2.93万亿元，但跨季DR007利率飙升至1.90%，压制了杠杆净利差。压测表明，基准情景下成交企稳于1.90万亿元可带来券商板块+85.24%股价重定价空间；但若流动性偏紧导致交易量回…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 3, "neutral": 0, "value": 63.2, "events": [ { "title": "红利资产因两融拥挤、质押隐患及Q3解禁重压存在微观流动性脆弱，压力回撤应调深至-13%，必须通过调降股票敞口、增配中债与黄金来守护组合-5%的最大回撤底线。", "summary": "截至2026-06-22，基于A股2.5万亿两融余额、100%保证金比例限制、股权质押在周期中盘的集中度，以及Q3共1.1万亿解禁，本卡认为红利资产-10%的回撤假设不足，应按-13%计提。边际卖压源于中盘周期价值、煤炭电力等拥挤红利和集中解禁板块。建议将A股调降至45%，港股调降至10%，增配中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中债增配仍可作为防御锚，但前提是把“中债”严格定义为无杠杆利率债；信用债和加杠杆carry不能计入回撤保护。", "summary": "截至2026-06-22，本卡对“增配中债防御A股红利流动性压力”的方案作信用压力测试。结论是：若中债仓严格执行为无杠杆国债、政策性金融债和现金化利率工具，股权质押平仓与两融压力不太可能破坏对冲；但若仓位混入城投carry、地产债、低等级信用债基金或回购杠杆，信用利差走阔会吞噬久期收益甚至把对冲仓…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在美联储沟通革命引发全球多资产波动率骤升与成长股多重估值压缩背景下，A股应坚定执行超配高电力上网电价防御韧性（水电/核电）与低配高两融风险成长板块的‘防御端杠铃策略’。", "summary": "美联储取消前瞻性指引与通胀预期上修打破了传统的股债负相关性，迫使风险平价模型被动去杠杆和抛售资产。压力测试显示风险平价在当前宏观环境下显著跑输，而战术性杠铃策略表现出强劲的抗波动韧性。全球收益率上行传导至A股，导致科创成长股面临多重估值压缩及两融强平风险，资金加速向水电、核电等红利重估板块切换；同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 0, "neutral": 1, "value": 85.6, "events": [ { "title": "6月调样后A股电子与红利板块在创新高的两融杠杆与回流 of 北向外资共振下依然维持净买入，所谓的规模差走阔实为短期调样调仓与动量噪音而非真实资金撤离，但历史高位杠杆要求通过跨资产多元…", "summary": "截至2026-06-23，6月调样后A股规模差异并非由真实资金撤离驱动。月中两融余额大幅反弹915.10亿元，于6月18日触及29655.12亿元历史新高，电子板块获430亿元融资流入；6月22日北向交易额达4821.77亿元，表明内外资正共振流入硬科技与红利板块。但高杠杆下拥挤度高企。压测显示，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-23", "supports": 1, "risks": 2, "neutral": 0, "value": 80, "events": [ { "title": "电子行业6132.72亿元两融杠杆中，357.00亿元处于维保比例临界带，-10%的行业性下跌将触发475.00亿元首轮复合抛压，高融资集中度的半导体龙头面临严重的负Gamma去杠杆…", "summary": "截至2026-06-24，本报告对电子行业6132.72亿元两融余额的维持担保比例分布进行了深入压力测试。当前分布显示，安全区（维保>150%）占比92.11%，而预警与强平区（维保<=150%）合计达484.00亿元。若科创50跌破1500点触发-10%的行业性回调，维保比例低于144.4%的3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-24，平准基金买盘只有在定向科创50、规模达到150亿-250亿元并触发做市商回补IM/IC空头时，才是Gamma中和器；否则只是指数托举器。", "summary": "截至2026-06-24，本卡压力测试认为：汇金式ETF买盘能中和科创50雪球负Gamma，但只有在定向规模达到约150亿元、且80%转化为当日篮子需求时，才足以覆盖118亿元做市商卖盘中值；250亿元才覆盖高压上沿。若只是50亿-100亿元宽基均衡买盘，更多是抬指数，无法封住IM/IC贴水和单名…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科创50雪球负Gamma抛压与电子两融去杠杆的反身性踩踏机制", "summary": "在科创50跌破1500、进入1516-1550敲入带的Scenario B下，雪球对冲盘负Gamma卖压（36-200亿，中值约118亿[自测算]）与电子6132.72亿两融中约357亿跌破强平线相互点火，首轮合计卖压约475亿[自测算]。我支持环路存在，但主张主导助燃剂是IM/IC基差通道而非指…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 5, "neutral": 1, "value": 58.4, "events": [ { "title": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论", "summary": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论 JPY Carry Trade Unwind vs. APAC Tech Margin Call Spiral 美元兑日元跌破161.50关口，财务省干预概率陡增。一旦日元套利平仓瞬间引爆全球EM科技股抛售，叠加SpaceX二次配售24.5亿元的资金虹…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "区分健康基差修复与现货去杠杆式基差修复的A股实时监测面板", "summary": "本收官卡把Research note 01–07综合为一套实时面板。核心陷阱：IM基差可因两种相反原因收窄——健康修复(g↓，期货向现货回升ΔFutures>0)，或现货去杠杆式修复(g↑，量化中性拆仓买期货卖现货、现货跌向期货ΔSpot<0)。最锋利辨别器是分解ΔBasis=ΔFutures−Δ…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CPO“量产延期”谣言与AI硬件基本面交付的真空博弈", "summary": "SemiAnalysis（约2026-06-09）将CPO大规模商用推后至2028–2029，引发美股光通信单日重挫（AAOI约-17%、COHR约-11%等）。但散户犯了范畴错误：被推后的是CPO，而非“易中天”当前出货的800G/1.6T可插拔模块；CPO延期反而延长可插拔与NPO的需求窗口，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核心结论：基本面传闻未证伪光模块需求，但726.5亿元的中际旭创+新易盛两融余额足以把传闻放大成价格真空。", "summary": "截至2026-06-24，本卡对上一卡的基本面偏多结论做压力测试：中际旭创与新易盛的直接风险不是科创50雪球机械卖盘，而是两融余额分别达465.3亿元、261.1亿元，合计占全A两融约2.42%。科创50雪球属于二阶尾部风险；若以80%敲入测算，当前科创50对应的示意敲入区约1530-1590点，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 1, "neutral": 1, "value": 83.8, "events": [ { "title": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局", "summary": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局 Midday A-Share Growth Rebound and Options Negative Gamma Clearance Tactical Asset Allocation 在上午科创50跌破关键支撑触发雪球敲入与两融强平后，午…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 两融折算率调降至0.40及科创50击穿1750点时物理红利（长电/核电/神华）的流动性吸收极限测试", "summary": "致 主题研究员： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 43a68cc6-4588-4986-afdb-1ed685319288 结果摘要: Wrote handoff_response.md (10363 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 5, "neutral": 2, "value": 64.2, "events": [ { "title": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局", "summary": "午间A股成长股探底回升与衍生品负Gamma出清的战术布局 在上午科创50跌破关键支撑触发雪球敲入与两融强平后，午间主力ETF资金流入与个股金针反转是否标志着负Gamma抛售压力的阶段性出清？我们应如何利用这一估值剪刀差进行右侧建仓，从红利低波向高Beta科技股进行再平衡？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论", "summary": "日元干预警报与亚太科技两融负反馈平仓螺旋辩论 美元兑日元跌破161.50关口，财务省干预概率陡增。一旦日元套利平仓瞬间引爆全球EM科技股抛售，叠加SpaceX二次配售24.5亿元的资金虹吸，A股电子板块6132.72亿元两融及科创50雪球在1500点敲入区附近会否触发自反馈平仓踩踏？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "两融压力下的资产配置：在高杠杆科技风险与水电防御腿之间的风险预算与规模分配", "summary": "本卡把前三卡的方向判断落地为可执行的风险预算与规模分配。核心是“权重≠风险”：纯风险平价（逆波动率ERC）下科技腿应得约22%，但因强平负Gamma尾部被波动率低估，杠铃须硬性压至12%上限；即便如此，12%的科技腿仍贡献约41%的毛风险[自测算]。建议四腿杠铃：科技12%/水电防御30%+其他红…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-26确认科技两融堰塞湖进入去杠杆阶段，但未确证牛市陷阱；最优策略是降拥挤科技、保现金缓冲、分批增配电力电网等物理红利资产。", "summary": "截至2026-06-27，结论是支持防守再平衡，但不把2026-06-26下跌直接视为“国家队牛市陷阱”的确证。A股两融余额已升至3万亿元以上，电子行业两融余额超过6000亿元且占比约两成，说明风险是融资拥挤和流动性冲击，而非科技周期被证伪。建议科技从超配降至标配，优先削减融资敏感电子/CPO仓位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 7, "neutral": 0, "value": 47.6, "events": [ { "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪峰的高估值成长板块。", "summary": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球商品通胀刚性与风险平价失效下的A股风格轮动：上游资源/红利价值 vs 成长，及两融与三季度解禁的交互", "summary": "全球'上游资源+红利价值'赢家只主导了沪深两市的一半：A股把组合拆开——资源赢（有色年初至今+30%以上、行业冠军[S4]），骑在第06-07卡铜/GOES缺口上；红利价值落后（中证红利一季度约+4%[S5]），因边际资金追逐成长逼空（科创50约+40%、创业板约+30%、6-22创新高[S4]）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "summary": "致 工业制造分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: eb737f41-b97c-43b9-9b89-04a7c7636133 结果摘要: Wrote handoff_response.md (10188 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国央行通过结构性再贷款与离岸FIMA人民币国债抵押回购机制，成功解耦了汇率防御与境内流动性，为A股“高股息防御+硬科技成长”的滞胀杠铃配置提供了宏观及风险平抑支撑。", "summary": "截至2026-06-27，在美中利差倒挂-265bp与低通胀下，央行转向境内结构性工具与离岸FIMA Repo防线，稳定了境内流动性（套利流出降90%至7.9B RMB）。这巩固了A股以高股息国企（海油6.2%）为防守、硬科技再贷款为成长的“滞胀杠铃”配置，抑平了368.9B两融及35B雪球负Ga…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 4, "neutral": 0, "value": 67, "events": [ { "title": "虽然278%平均维保安全垫使两融连锁抛压呈局部非系统性风险，但20-30%回撤仍将触发最高2528.00亿的平仓压力并重创产业链，应通过减仓高倍数前道、超配低谷先进封装以及引入非担保…", "summary": "截至2026-06-28，两融杠杆在电子和半导体板块高度集中（总余额分别达6132.72亿和560.80亿），前五大龙头占用386.80亿。压力测试显示，20-30%的回撤会迫使大批账户跌破130%平仓线，极端情况下（科创50跌破1520）诱发2528.00亿平仓压力，并因雪球负Gamma在2-5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在紧平衡流动性与外部融资约束下，A股难以支撑全局性风险偏好扩张，下半年配置应降低总Beta，采用“硬科技国产替代（封装/设备）+高独立现金流红利”的结构性杠铃组合。", "summary": "在美联储加息预期和国内紧平衡流动性制约下，A股面临外资流出、两融去杠杆与解禁供给的三重压力。外资流入放缓，险资增配高股息红利防御资产。两融余额高企（3.015万亿元）使科技成长面临局部去杠杆和级联平仓风险。半导体板块估值已普遍收缩30%至45%，大基金三期（3440亿元）提供战略支撑。下半年主导风…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。", "summary": "截至2026-06-29，本卡片支持前序“更高更久”利率压制A股成长久期的判断，但将风险定位为结构性踩踏而非全市场2015式去杠杆。两融余额为30329.92亿元，占A股流通市值2.83%，总量可控；但电子板块融资余额6261.87亿元，且6月融资流入高度集中。模型显示，科创50较2026-06-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼及普钢加工环节将面临严重的利润率…", "summary": "本卡通过技术壁垒与自给率对A股AI原材料敞口进行了区分：高端取向硅钢方面，宝钢股份凭借规模与技术优势展现出强Alpha属性，而首钢股份受普钢周期Beta拖累较大；有色铜方面，在2026年长单TC/RC为零的极端低位下，矿权自给率超65%的紫金矿业可完全捕获矿端溢价（EBITDA价格敏感度约7.5%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 1, "risks": 3, "neutral": 0, "value": 82.4, "events": [ { "title": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波及红利资产的防守底线（情景2），…", "summary": "在国家队选择性托底CSI300和银行股的背景下，本卡片对两融去杠杆进行了压力测试。结果显示，高贝塔的科创50（当前1596.57点）与TMT板块最易受冲击。科创50跌至1550点（跌2.92%）将触发5%的平仓率（抛压490.8亿），触及国家队500亿ETF买盘边界；跌至1516点（跌5.05%）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股红利拥挤度与7-11月年内解禁洪峰限制了风格切换空间，而在人行短端流动性上限失灵与CNH被动升值情景下，高负债、高Beta的券商板块将面临最严重的两融净利差挤压与反转冲击。", "summary": "截至2026-06-29，本卡对A股杠铃轮动的拥挤度与压力进行评估。全市场两融达30329.92亿元（TMT占41%），银行与防御红利极其拥挤，仓位达95%分位；6-11月面临2.8-3.2万亿元解禁洪峰，今日单日解禁达589.22亿元，严重压制内需成长。若CNH继续升值或人行回购走廊失灵（DR0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%）；通过配以5.0%权重可转债债…", "summary": "截至2026-06-29，在两融强平与350亿雪球敲入负Gamma踩踏、以及人行短端Corridor失灵（DR007升至2.50%）的极端情景下，杠铃组合资产相关性共振跃迁（红利与成长升至+0.55至+0.70）。组合波动率由9.60%升至19.98%，日99% VaR升至2.93%（穿透2.50…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策背离下的A股风格轮动与板块配置：能源安全杠铃 vs. 经利润分层的转型龙头，以及A股的相对庇护所超配", "summary": "把卡片01-03的宏观综合落成组合：PPI再通胀/CPI去通胀的撕裂本身就是风格信号，奖励上游/价值（约5.3%股息carry）与拿到人行宽松顺风的长久期成长龙头，碾压同质化中段。我支持杠铃、但否定彻底超配。利润层正常化所依赖的反内卷整合于2026-01-09被反垄断监管层暂停[S3]，故转型腿仍…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] } ], "href": "/investor/theses/thesis-11f0838e82/", "networkHref": "/investor/research-network/#thesis-11f0838e82" }, { "id": "thesis-1495ab3592", "source": "auto_topic", "title": "在距离OBBBA法案7月4日开工红线仅22天、且变压器积压导致绿电并网排队项目创纪录撤回的背景下，电网容量费暴增9.3倍如何侵蚀云厂商A…", "titleEn": "Practical research topic: 在距离OBBBA法案7月4日开工红线仅22天、且变压器积压导致绿电并网排队项目创纪录撤回的背景下，电网容量费暴增9.3倍如何侵蚀云厂商A…", "lane": "电力与公用事业", "laneEn": "电力与公用事业", "laneHref": "/investor/lanes/lane-ee7fbf3f/", "investableFocus": "", "stockUniverse": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" } ], "stockTickers": [ "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH", "002028.SZ", "APO", "IEF", "TLT", "CAT", "COP", "588000.SH" ], "coreView": "该主线来自 3 条研究记录的标题、问题和摘要聚类，时间跨度 2026-06-12 至 2026-06-12；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 75, "eventCount": 160, "firstSeen": "2026-05-13", "lastSeen": "2026-06-30", "supports": 9, "risks": 146, "neutral": 5, "analystCount": 14, "sourceTypeCounts": { "研究笔记": 157, "研究主题": 2, "研究线程": 1 }, "topTerms": [ "云厂商", "变压器", "半导体", "电网", "A股", "并网", "OBBBA", "撤回的背景下", "能源", "开支", "土需求", "设备" ], "whatChanged": [ "2026-06-30: AI数据中心并网延期与变压器死锁下全球算力硬件 Capex 的结构性错配", "2026-06-29: 材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放右移。", "2026-06-29: 关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "2026-06-29: 电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对冲有效却对冲错轴，继续超配电网设备而非纯绿电发电。", "2026-06-29: 物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进封装国产替代材料集中。" ], "practical": { "sectorLane": "电力与公用事业", "sectorLaneEn": "电力与公用事业", "exposureBucket": "电力与公用事业", "tone": "yellow", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "roleEn": "上游油气", "bucket": "core", "href": "/investor/stocks/COP/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" } ], "stockTickers": [ "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH", "002028.SZ", "APO", "IEF", "TLT", "CAT", "COP", "588000.SH" ], "primaryStocks": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" } ], "instrumentTags": [ "GEV", "GE Vernova", "ETN", "Eaton", "PWR", "Quanta Services", "VST", "Vistra", "CEG", "Constellation Energy", "XLU", "Utilities Select Sector SPDR" ], "reviewCadence": "每日复核", "confirmationSignals": [ "并网排队和变压器交期是否继续拉长", "表后自发电与 PPA 是否改变公用事业信用指标", "电力设备订单质量是否高于收入确认节奏" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "长端利率", "铜/GOES 成本", "监管电价", "信用利差" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 3 类，近期事件 125 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 3, "analystCount": 14, "recentEvents": 125, "supportRatio": 0.06, "riskRatio": 0.91, "conflictScore": 100, "evidenceDepth": 100 }, "score": 66.1 }, "events": [ { "date": "2026-06-30", "title": "AI数据中心并网延期与变压器死锁下全球算力硬件 Capex 的结构性错配", "summary": "AI数据中心并网延期与变压器死锁下全球算力硬件 Capex 的结构性错配 AI Power Grid Deadlock and Hyperscaler Capex Structural Misallocation 北美云巨头下修 H2 硬件 Capex 16.18%，在变压器交期达120-160周和电网并网等待53个月的物理约束下，高阶 GPU 的出货是否…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 29, "matchedTerms": [ "云厂商", "变压器", "半导体", "电网", "并网" ], "href": "" }, { "date": "2026-06-29", "title": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "summary": "支持第06卡\"并网延迟→利润压缩\"的传导论点，但将单向结论细化为分化。针对11GW数据中心建设缺口与12-18GW刚性局部电网缺口，48-60个月并网延迟通过两条通道传导：(1)经订单/PPA递延与尾款拖延压制发电子链开工率——5.7%销量递延经经营杠杆放大为约-21%利润压缩[自测算]；(2)将电网投资节奏重置向T&D侧，5万亿元资本开支周期以70-75…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 31.5, "matchedTerms": [ "变压器", "电网", "并网", "能源", "开支", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "将能源链利润分化（发电侧承压、电网设备受益）整合进A股成长/价值杠铃配置", "summary": "本卡综合第05–07卡，重构A股成长杠铃。把第06卡腾出的8.0亿元大部分倾斜至利率不敏感、政策背书的电网设备一腿（+5.0亿），余下+2.0亿配高股息电网运营、+1.0亿留券商β/干火药，AI硬件保留4%残余尖端，绿电/发电设备零配置。电网设备订单免疫16.18%AI资本开支下修、可见至2027–2030年、供给稀缺（变压器全球利用率98%、2029年后…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "变压器", "电网", "A股", "能源", "开支", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利…", "summary": "本报告对AI算力激增带来的电力电网物理瓶颈（变压器交期120-210周[S2]、并网排队4-7年[S1]）对A股转型龙头的利润率传导进行压力测试。结果显示，电网扩容滞后充当了新能源装机的“限速阀”，压制了光伏/电池销量增速；同时，容量电价[S8]和碳成本上升[S19]增加了高能耗制造环节电费开支，而在严重过剩背景下成本转嫁失败，导致制造端转型龙头利润率遭到…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28.5, "matchedTerms": [ "变压器", "电网", "A股", "并网", "能源", "开支", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "summary": "截至2026-06-30，本报告压力测试了变压器交期拉长对AI算力集群部署和云厂商资本开支的约束。变压器交期拉长（128-144周，专用型4-5年）导致物理并网时滞达4-7年，引发全球467万个GPU延迟并网和2万亿元Capex延期。微观上，1.5年延迟导致10亿美元GPU集群产生5.2295亿美元折旧与持有损耗，月度拖累1.1905亿美元，NPV骤降12…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27.5, "matchedTerms": [ "云厂商", "变压器", "半导体", "并网", "开支", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进…", "summary": "截至2026-06-30，本报告压力测试了AI电力配给对晶圆代工CapEx及供应链的冲击。电网变压器交期与并网延迟导致467万个GPU并网滞后及2.0万亿元CapEx递延。压力下2027-28年GPU出货减少1032万颗，导致半导体收入损失3096亿美元。相比之下，电网设备同期仅新增300亿-560亿美元收入，其25%低毛利率根本无法对冲高利润半导体供应链…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27.5, "matchedTerms": [ "变压器", "半导体", "电网", "并网", "开支", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.5, "matchedTerms": [ "变压器", "电网", "并网", "能源", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "summary": "本卡片从制造和供应链视角对电网与电力约束进行压力测试。截至2026-06-29，关键电力设备（变压器与重型燃机）由于极高工艺与测试门槛、上游取向电工钢（GOES）瓶颈及寡头垄断，交期拉长至3-7年，生产机位已售罄至2029-2030年。云厂商BTM发电加剧了燃机机位争夺。预计2026-2027年该瓶颈将导致云厂商约800亿-1100亿美元资本开支递延兑现。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.5, "matchedTerms": [ "云厂商", "变压器", "电网", "开支", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放…", "summary": "截至2026-06-29，变压器交期延长和材料瓶颈导致美国电网资本开支通胀率达32.72%。结合2027/2028年度PJM容量电费暴涨11.5倍至$333.44/MW-day，将拉动美国零售电价上涨4.18%，直接和间接推高全国CPI 0.124个百分点。高粘性通胀与AI基建庞大资金需求将上移中性利率50-75基点，延缓美联储降息路径。此外，并网滑期18…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 26.5, "matchedTerms": [ "变压器", "电网", "并网", "开支", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心腿；半导体国产替代（¥2万亿AI建设计划+80%国产强制比例…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "变压器", "半导体", "电网", "A股", "开支", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "在2026-06-28，物理加电瓶颈正迫使AI算力估值由单纯的Capex扩张转向已通电MW效率，投资者应超配锁电、BTM自发电超大型云厂商与掌握变压器等硬缺口产能的电力设备商，并在A股中向特…", "summary": "截至2026-06-28，本报告探讨了在AI电力约束下算力资产估值向MW效率转型的逻辑。对于超大规模云厂商，2年的并网延迟将对100 MW智算中心产生严重的NPV减记（-124%）[S12]，迫使其估值分层：超配锁电与BTM自发电龙头，低配单纯Capex扩张者。电力设备商成为AI周期收费站，超配大型变压器等掌握稀缺产能与电网智能化的龙头。A股市场在“算电协…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "云厂商", "变压器", "电网", "A股", "并网", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "AI驱动的结构性裁员实质上是算力投资回报（ROI）兑现的证明，云厂商AI资本开支具有极强韧性，其真实制约并非宏观衰退担忧，而是电网并网延迟与重型电力设备的物理交付瓶颈。", "summary": "本卡片对“AI裁员压力动摇算力基建投资”的假设进行压力测试并予以否定。AI驱动的结构性裁员（Challenger裁员AI占比达40%）是AI投资回报（ROI）兑现的实体证据，而非衰退信号。四大云厂商资产负债表坚韧，AI基建投资是一场生存竞争。然而，物理供电（并网排队5-7年）与设备交付（变压器交期18-36个月）瓶颈将导致2026-2027年4500-47…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 31.5, "matchedTerms": [ "云厂商", "变压器", "电网", "并网", "开支", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "电网容量瓶颈并非完全抵消半导体国产替代红利，而是通过系统周期错配将压力传导给中下游服务器与算力商，同时倒逼云厂商转向表后自备电力与地理重构。", "summary": "本报告对电网瓶颈是否抵消半导体国产化红利进行压力测试。物理并网延迟和关键设备短缺确对AI部署形成‘硬阀门’，延缓算力落地并导致2026-27年全球约800亿-1100亿美元CapEx递延。但这并未抵消自主化晶圆设备和封装厂的政策红利，而是在产业链内部造成分化：上游设备厂商基本免受冲击，中下游加速器与服务器商的财务和出货受抑。中国通过特高压、东数西算与绿电直…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "云厂商", "半导体", "电网", "并网", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "DeepSeek API的大幅通缩不仅不会消灭算力需求，反而会通过吉文斯悖论激发Agent调用量井喷，使净电力需求暴增15.36倍，进一步固化并收紧以变压器（交期120-210周）和并网（排…", "summary": "2026年6月26日DeepSeek API大幅降价96.37%引发散户算力需求毁灭的恐慌，导致A股电子板块两融强平盘爆出。本报告通过可证伪判定框架（流量弹性、调用量、并网申请）论证其为吉文斯悖论的放大：Agent爆发与MCTS使Token消耗飙升，即便算力能效提升，数据中心电力净需求仍增15.36倍。电网变压器（LPT交期120-210周）物理制约未解，…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "变压器", "电网", "A股", "并网", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "2026年下半年应采用“受控久期 + 瓶颈权益”的杠铃组合，而不是泛AI beta或单纯债券多头。", "summary": "2026-06-29，最终综合结论是将组合定为久期/高等级债券38%、权益44%、商品/能源对冲8%、现金10%；权益内部为A股26%、港股6%、美股12%。A股成为最大权益袖口，核心是电网设备、特高压、变压器、先进封装/HBM、测试、功率管理与国产低功耗AI设计。久期仍超配但不极致，因为核心PCE仍为3.4%。主要风险不是电网瓶颈消失，而是AI资本开支或…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27.5, "matchedTerms": [ "变压器", "电网", "A股", "能源", "开支", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力需求拉动的配电网与储能设备供应瓶颈及其对能源成本的影响", "summary": "本卡对设备紧缺主张作分项压力测试。结论分裂：并网延迟——成立且重大，变压器交期36-48个月、高压级3-5年，美国2,600 GW排队、中位等待约5年，数据中心挤占稀缺设备与排队名额，2026年美国12 GW数据中心仅4-5 GW在建、约半数延迟或取消。能源价格——仅部分成立且方向相反：紧缺推升的是受限区域受监管电价与容量价格（PJM批发同比+76%、容量…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27.5, "matchedTerms": [ "变压器", "电网", "A股", "并网", "能源", "设备" ], "href": "" } ], "timeline": [ { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "AI CapEx并未因外壳新开工放缓而失效，但其落地速度正被电力、变压器、配电网络和电网稳定性重新定价。", "summary": "截至2026-05-13，本卡支持“密度胜过足迹”主线，但补充能源约束：AI算力密度提升使稀缺资源从外壳数量转向可供电、可散热、且符合电网稳定要求的容量。LBNL、EPRI和EIA均显示数据中心用电占比快速上升；NERC 2026-05-04 Level 3警报和2024年1,500 MW数据中心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "DRAM拥挤是仓位信号而非卖出信号；下周应部分减SMH/宽基科技，转向小仓DRAM、重仓GRID、适中PAVE。", "summary": "截至2026-05-14，本卡支持下周将AI敞口从SMH/宽基科技部分切向DRAM+GRID/PAVE，但不追成满仓DRAM。DRAM的HBM/传统DRAM涨价逻辑真实，2026年传统DRAM ASP上修明显，但DRAM ETF上市后集中度和资金拥挤过高，应限仓12-15。GRID是更优新增主仓，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 1, "risks": 0, "neutral": 0, "value": 52.8, "events": [ { "title": "电网侧配电升级、特高压和新能源并网对铜铝需求的贡献是政策资金锁定的一阶需求冲击，量级显著大于AI资本开支，足以支撑铜铝多头主线。", "summary": "截至2026-05-17，支持前序铜铝多头逻辑：AI只是二阶增量，电网电气化才是一阶确定性需求。中国2026-2030年电网周期按国家电网4万亿元、南方电网2026年1800亿元预算和IEA铜14%/铝6%成本占比测算，可支撑约150-190万吨/年铜、230-340万吨/年铝；相对“十四五”后段…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "能源物流没有被 AI 消灭，但“物理大宗必要性”不等于“能源大宗运输贝塔免疫”，真正稀缺性正在迁移到电力设备、燃机、储能和电网集成。", "summary": "截至 2026-05-19，工业制造视角否定卡片 05 的强版“能源物流战略性免疫”。能源分子仍需大宗运输，但 AI 电力需求首先制造的是电气设备、燃机、储能、开关设备、变压器、并网和控制系统瓶颈，而不是对 VLCC、煤炭铁路或普通港口大宗贝塔的广谱保护。IEA 2026 年报告显示 hypers…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "能源基础设施供给压力是 2027–2028 的约束而非 2026 的约束，因此 Research note 05 的半导体核心配置在 2026 内成立，但升级到真实风格轮动需在触发条…", "summary": "截至 2026-05-20，本卡片对 Research note 05 的‘半导体核心 + 软件卫星’配置做 stress-test。能源瓶颈真实存在但不约束 2026 年半导体资本开支周期，而是约束 2027H2–2028：2026 年订单簿由已批准 / 在建电力承接，而 2027–2028 暴…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2028年将是‘绿色AI的黑暗时代’，物理电网设备的极端匮乏使算力扩张与能源转型深度脱钩，迫使云厂商放弃零碳优先原则以换取业务连续性。", "summary": "截至2026-05-20，本卡片通过压力测试确认：变压器等重电装备的“物理排队”已成为新能源并网的硬天花板。高压变压器交付周期拉长至3-5年，导致大量风光项目处于“机械完工但电气搁浅”状态。AI超大规模云厂商凭借资本优势掠夺生产槽位，进一步挤出了新能源开发商。预计2026-2028年将出现严重的“…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 1, "value": 46, "events": [ { "title": "能源供给侧足以支撑AI硬件的长期需求，但短期决定出货节奏的是节点级通电能力，而不是全国电量总量。", "summary": "截至2026-05-21，能源侧扩张能够维持AI硬件长期需求，但不能完全消除2026-2027年的局部“电力之墙”。美国2026年计划新增86 GW公用事业级容量，绿电与储能增量可观；中国凭借风光装机、储能和State Grid的RMB 4 trillion电网投资，缓冲能力更强。但变压器、并网队…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电力并网延迟与能源基础设施的物理与监管约束构成了制约AI硬件实物量增长的硬性瓶颈，导致算力链ROIC严重退化，从而加速资金在2026年下半年向高增长、即时变现的出海消费主线切换。", "summary": "截至2026年5月21日，对能源基础设施的压力测试显示，电网并网延迟（PJM/Dominion面临7-15年排队）与关键设备短缺（变压器交期拉长至128-160周）成为AI硬件实物量增长的物理约束，实际并网率仅达65%。未通电的“暗数据中心”使算力GPU前两年年化ROIC骤降至8%以下，将挤压高杠…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "AI推理负载的分散化将电力约束下沉至配电网末端，引爆了对中压配电设备和表后（BTM）天然气/燃料电池分布式发电的激增需求。", "summary": "截至2026年5月22日，AI推理负载向5-20MW“推理卫星”站点转移，导致电网瓶颈从超高压输电网下沉至中压配电网（13.2-33kV）。高密度的持续基荷引发了配电网局部容量冻结，中低压变压器交期仍长达12-36个月。为绕开3-5年的变电站并网排队，数据中心大量转向“表后自供电”（BTM）。美国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "截至2026-05-24，A股应从单一红利防御升级为红利底仓加结构成长修复的杠铃组合，而不是在能源-汇率压力下全面切向成长或全面退回防御。", "summary": "截至2026-05-24，本卡支持前序判断：能源-汇率压力并未剥夺央行宽松自主性，但把A股传导从全面降息beta转为结构性信用配置。红利资产仍是底仓，因低beta、现金流和资源/物流成分能对冲PPI与实物流波动；成长资产不是全面追高，而是围绕4000亿元科技创新和技术改造再贷款、AI硬件、半导体设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "在2026-2028年间，电气化重装备及数据中心专用电源设备处于结构性“超强周期”，设备商产能已被预订满至2028年后，导致电网并网时滞成为AI算力上线的实质屏颈，促使重装备板块重估…", "summary": "截至2026-05-24，本报告评估了电力变压器、开关柜及数据中心电源系统在2026-2028年的供需缺口。大型变压器交期拉长至150-200周，因硅钢紧缺，美国严重依赖进口，现代电气和金盘科技订单锁排至2028年。开关柜交期翻倍，Powell与Eaton在手订单创纪录。Vertiv受益于800V…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 1, "risks": 1, "neutral": 0, "value": 56.6, "events": [ { "title": "全球电气化扩张受制于电网设备（160周交付期与15%关税壁垒）和关键金属（铜缺口）的硬供给约束，延长了AI开支转化滞后，这将在折旧崖和2.08万亿美元营收缺口挤压下触发2027年美股…", "summary": "截至2026-05-27，本报告系统综合了电网与材料层面的实体约束，揭示了美股Risk-On与A股防御regime的分裂收敛机制。由于变压器面临160周以上的交付瓶颈与2026年4月实施的15% Section 232过渡关税重击，叠加2026年精炼铜缺口，电网并网出现4-7年延迟。这导致AI c…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "第 09 卡 — 电网与能源供给能否支撑 2028 2029 年 AI 资本开支？ 白板: 3fb1b672 a7f6 4df0 84e1 2367a6f57495 卡号: 9 / …", "summary": "第 09 卡 — 电网与能源供给能否支撑 2028 2029 年 AI 资本开支？ 白板: 3fb1b672 a7f6 4df0 84e1 2367a6f57495 卡号: 9 / 10 分析师: 能源行业分析师 (energy analyst) 立场: support（支持） 工作日: 2026…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 5, "neutral": 0, "value": 26, "events": [ { "title": "AI算力的物理约束正从“有没有电”转向“能否在正确节点、正确时点、用合格设备按时送电”。", "summary": "2026-05-29：本卡支持上一张TMT二阶硬件主线，但加入电力基础设施折扣。AI算力扩张不是中国全国层面的总量缺电问题，而是局部电网容量、变压器/开关设备交付、新能源消纳和高功率密度冷却/备电的时点约束。中国互联网数据服务用电1-4月同比增长44.4%[S2]，但仅占全国同期用电约0.94%[…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "表后直连遭监管否决、主网排队超5年，叠加热带关税地雷与120kW超高密度机架对存量机房升级的彻底失效，构成了AI无限能源扩张的物理与法理硬着陆，将实质性推迟算力部署并打压估值。", "summary": "截至2026-05-30，本卡对先前“算力从硅端向电网瓶颈转移”的共识进行了严苛的压力测试。我们指出，市场低估了电网交付及设备供给的法理与物理硬约束：BTM面临FERC否决及管道与环保瓶颈；变压器交期达2-4年且面临关税与安全审查壁垒；GB200等120kW+超高机架密度使存量机房改造不可行，只能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张的能源瓶颈不是全球缺电，而是局部电网接入、变压器、输电和可靠容量的时间错配；它会放大交付节奏波动，但尚不足以否定A股核心AI互联龙头的分阶段回补逻辑。", "summary": "截至2026-05-30，本卡从能源侧压力测试AI硬件回补逻辑：Dell的AI服务器收入、订单和在手订单验证需求，但边际瓶颈正在转向特定节点能否按时送电。GB200 NVL72单柜132 kW，1 GW园区在PUE 1.15下约承载6,588柜、47.4万颗GPU。IEA预计全球数据中心用电从20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "贸易壁垒带来的严重变压器通胀与施工延期使公用事业公司的资本开支在无现金流入的情况下不断膨胀，即使在大型负荷协议（LLAs）保护下，系统级大网开支仍会破坏其FFO-to-Debt指标，…", "summary": "截至2026年5月29日，本卡对美国受规管公用事业进行信用压力测试。结果显示，Section 301/232关税推升变压器与GOES等设备成本高达30%-50%，且并网交期拉长至3年，导致资本开支沉淀于无现金回报的在建工程（CWIP）。由于监管滞后、审慎性核减及居民侧费率反弹，即使有大型负荷协议（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "伊顿和施耐德的电力装备产能投产严重阻后于英伟达芯片出货周期，且受制于变压器铁芯原材料取向硅钢（GOES）的物理刚性短缺，成为制约GB200于2026年点亮上线的独立刚性瓶颈。", "summary": "截至2026-05-30，压力测试表明，伊顿与施耐德的产能扩张无法填补因电网延迟导致的GB200部署缺口。两家公司的主要新增产能要到2027年及以后投产，与2026年爆量出货的芯片周期存在严重时钟错配。变压器（交付128周）与开关柜（交付40-60周）积压严重，加之铁芯原材料取向硅钢（GOES）结…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 3, "neutral": 0, "value": 40.8, "events": [ { "title": "受制于受监管公用事业股本回报率天花板与5年期并网及变压器设备瓶颈，算力税向“电力税”的再分配呈现高度结构化分化，且因3年GPU折旧错配促使AI算力走向离网与全球化多样部署（2026-…", "summary": "截至2026-05-31，本报告对“电力税”价值攫取进行压力测试：AI电网容量与清洁基荷的绝对稀缺，使价值加速向大型核电/气电发商（CEG、VST）及物理电网核心变电设备商（GEV、ETN）倾斜。然而，该传导机制正面临严厉的联邦和州级费率监管壁垒，并因5年以上变频/并网排队耗时与3年GPU折旧周期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张的能源约束不是全球TWh不足，而是本地可靠MW、电网接入和电力设备交付周期不足。", "summary": "截至2026-05-31，本卡片支持“全球硬瓶颈”主线，但将瓶颈从芯片进一步扩展到可靠电力、电网接入和电力设备。IEA基准情景下全球数据中心用电可由总量供给吸收，但美国PJM、ERCOT等节点的并网、变压器、燃机和开关设备交付周期会成为AI算力扩张的配给机制。结论是：AI需求不必因电力约束崩塌，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "研究报告：电网资本开支通胀与美联储“最后一英里”挑战 日期： 2026 05 31 分析师： chief economist (首席经济学家) 立场： 压力测试 (Stress te…", "summary": "研究报告：电网资本开支通胀与美联储“最后一英里”挑战 日期： 2026 05 31 分析师： chief economist (首席经济学家) 立场： 压力测试 (Stress test) 主题： 结构性电网资本开支通胀对宏观通胀预期及美联储政策路径的影响。 1. 执行摘要：通胀下方的“电力”地平…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 2, "neutral": 0, "value": 48.2, "events": [ { "title": "在AI成长仓位占比达30%的情景下，电力基础设施瓶颈与电价波动并非仅是运营成本扰动，而是通过重资产折旧放大效应、能效红线惩罚与并网时间滞后决定智算板块盈亏生死线的核心地缘与宏观变量；…", "summary": "截至2026-06-01，本卡对AI成长仓位占比提升至30%的情景下，算力板块面临的电力供应与能源成本约束进行了深度压力测试。基于100MW智算中心典型模型，中度电价上涨（+20%）将导致运营商净利润率下滑15.8%至6.70%；若夏季PUE超标引发惩罚性电价（重度压力），净利润率将下挫36.6%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力天花板的核心痛点是并网延迟和规制规费引发的硬件折旧搁置与清洁电力溢价，这将在中期使云GPU租赁提价15%-35%[自测算]，并向SaaS客户传导3.0%-4.5%的毛利拖累[…", "summary": "截至2026-06-02，本卡对AI用电瓶颈做深度测试。电网并网设备延期（128-144周）引发的‘芯片折旧挤压’是最大隐形电力成本，1年延迟使GPU活跃小时折旧成本增50%[自测算]；加之PJM 2026年收取$4-$8/MWh合约规费，电力成本上涨导致云GPU租赁提价15%-35%[自测算]，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 4, "neutral": 0, "value": 37.6, "events": [ { "title": "由于变压器产能短缺、电网并网审批排队以及核能直接配建的监管博弈，Anthropic的5GW+电力承诺面临2至4年的物理延期风险，云厂商将承受算力资产闲置折旧压力。", "summary": "截至2026-06-03，Anthropic的5GW+电力承诺正遭遇严重物理瓶颈。全球高压变压器交期延长至3-5年，核心材料取向硅钢本地供应受限于Cleveland-Cliffs独家产能；PJM与ERCOT电网并网等待周期长达3-4年，且核能直接配建面临输电费用分摊的诉讼博弈，导致其算力扩张计划大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "供应链脱钩带来的成本上升并未抑制需求，而是将全球能源转型的驱动力从‘成本红利’转向‘安全与物理约束下的刚性支出’，算力溢价正成为吸收绿色通胀的主要缓冲机制。", "summary": "截至2026-06-02，全球能源转型正进入‘高成本、低弹性’的需求超周期。尽管美欧供应链脱钩导致局部LCOE上升10-15%，且变压器等关键设备面临30-50%的脱钩溢价，但AI算力负荷的刚性增长和电网现代化的迫切需求抵消了成本抑制。超大规模云厂商和公用事业公司正在吸收‘绿色通胀’，优先确保交付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网变压器长达150周的交期将导致大量AI芯片成为“搁浅算力”，迫使云厂商资本支出向电力倾斜，NVIDIA等硬件龙头的2027年盈利预期面临下修风险。", "summary": "截至2026年6月2日，本卡对AI基础设施部署进行压力测试。NVIDIA GB200机架和10万卡集群面临150周变压器交期的硬约束。这意味着2026年中启动的数据中心到2029-2030年才能并网。从GPU交付到通电的周期将从6-9个月拉长至18-24个月，云厂商将被迫将资本支出向电力设备倾斜。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-03，AI–电力链的确定性溢价来自“在受约束节点交付可靠一度电的权利”，而不是泛能源需求增长本身。", "summary": "截至2026-06-03（Asia/Singapore），本卡支持上一张卡片的“重资产基础设施算力”主线，但把可投资确定性收窄到电网稀缺。AI 数据中心不是简单制造全球能源短缺，而是在 PJM、ERCOT、中国东部负荷中心等节点制造可靠电力、并网、变压器、输电和成本分摊约束。全球数据中心用电由 2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 2, "neutral": 0, "value": 51, "events": [ { "title": "大型变压器和并网名额稀缺将奖励控制可交付power slot的公用事业与拥有已投运清洁/容量资源的IPP，而不是单纯拥有最大需求管线的公司。", "summary": "截至2026-06-03，本卡综合判断：电网设备与并网名额稀缺正从AI资本开支的物理约束，转化为公用事业和IPP的可变现资产。受益最强的是能把大负荷接入转为rate base、rider和大负荷电价的Dominion、AEP、Entergy、Southern/Georgia Power、Xcel；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支受制于53个月并网时延与120-160周变压器交付周期的硬性电力墙约束，投资逻辑已从硅侧算力向电网、基载核电及液冷散热等物理瓶颈要素转移。", "summary": "截至2026-06-03，AI资本开支的物理基础设施约束已成为核心天花板。由于美国电网并网延迟长达53个月，且特大型变压器交付周期拉长至120-160周，AI资本开支向云收入转化时延从4个季度拉长至6个季度。容量电价飙升使100MW数据中心电费翻倍并压缩推理毛利至42.0%。建议将硅侧权重削减至2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "能源基础设施的长周期物理属性与稀缺性，将AI硬件从波动性科技股转变为具有高确定性的“计算公用事业”，为基本面提供了不受短期金融流动性影响的估值底座。", "summary": "截至2026年6月5日，本报告认为能源基础设施（电力供应与电网扩容）的物理瓶颈已成为AI硬件估值的“物理地板”。变压器3-4年的交付周期、PJM电网8年以上的排队时间，以及超大规模云厂商签署的20年核能协议，共同构建了AI资本开支的长期确定性。这种物理约束使得硬件需求脱离了0DTE引发的短期金融流…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 0, "value": 59.8, "events": [ { "title": "AI能源溢价应从宽基设备链转向现金可兑现的电网/GIS与稀缺核心部件，液冷和AIDC变压器高beta标的必须先通过2026年中现金流复核。", "summary": "截至2026-06-07，本卡对A股变压器、GIS与液冷链条做估值拥挤度和现金质量压力测试。结论是保留AI能源主题，但不再给予宽基设备篮子统一溢价。思源电气仍是质量最高但估值拥挤的复合龙头；华明装备现金质量更干净，适合质量溢价；平高电气配得上温和GIS/电网溢价；金盘科技需等待2026Q2-Q3经…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "电力供应与输电网容量的物理硬约束已成为AI算力扩张的终极壁垒，通过推高计算边际成本与导致算力配额配给，对AI药物研发构成了比半导体产能更具刚性和长期性的瓶颈。", "summary": "截至2026-06-07，本报告对“半导体是医药AI唯一核心瓶颈”的观点进行压力测试。分析指出，电力供应与电网输配电容量是比芯片更为刚性、且建设周期更长（并网排队3-7年，变压器交货2-4年）的终极物理瓶颈。PJM容量出清价暴涨及爱尔兰强监管并网框架，推高了云端研发开支并引发算力配额制，成为Tec…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 1, "value": 65.6, "events": [ { "title": "AI资本开支正从算力硅片向电力物理基础设施发生不可逆的溢出，A股配置须从TMT硬件向电网出海及系泊设备进行长周期大轮动。", "summary": "截至2026年6月4日，53个月并网排队与10倍容量电价标志着AI产业进入电力死锁。云巨头Capex正从算力硬件向电网基建与分布式电力物理溢出。A股策略应低配估值透支的TMT算力硬件，大幅轮动至受益于海外电网建设红利的变压器/高压线缆出海企业，并战术配置浮式数据中心催生的系泊设备及高股息公用事业。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电力并网与设备瓶颈是较半导体封装更难通过资本开支化解的刚性约束，将在2026年构成宏观供给侧滞胀脉冲并延缓AI红利释放。", "summary": "截至2026-06-08，本卡片压力测试表明，电网并网排队（3-7年）与大型变压器交付（2-4年）等物理瓶颈将取代半导体先进封装，成为2026年AI硬件释放期更长久的滞胀驱动因素。GOES电工钢和铜的供应链紧张将导致电力设备商在重度压力下面临超2000 bps的毛利率崩塌，形成供给侧成本推升与算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 2, "neutral": 0, "value": 58, "events": [ { "title": "电力与电网瓶颈已成为AI算力落地的物理天花板，如果变压器短缺与电网容量摩擦导致40%项目延期，将直接在2026-2027年期间递延多达700亿美元的半导体行业营收，加剧低能效云商的闲…", "summary": "2026年6月9日报告通过对电力供应和电网容量的压力测试发现，物理用电瓶颈和电网并网延迟正在成为制约AI算力中心落地的短线硬伤。如果40%的规划负荷发生被动延期，将在2027年前导致多达280万张GPU无法正常部署，对应半导体行业约700亿美元营收被递延。云厂商面临资本开支闲置与变现周期拉长的双重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "eVTOL 充电对电网的威胁真实但局部可控——它不是威胁配电网，而是前置拉动配电网与分布式储能资本开支，隐性并网成本则是特许经营模型被低估的硬约束。", "summary": "本卡从能源侧压力测试 Card-05 论点。结论：eVTOL 瞬时高功率充电（单坪 250–600kW、整场 2–4MW、场地需求抬升 6–7 倍 [S1][S2]）对馈线与配电变压器层面构成真实威胁（电压跌落、欠压、热过载），但属局部、可工程化，并非系统性主网威胁。强制配套 1–5MWh 现场储…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 59.4, "events": [ { "title": "电网材料瓶颈倒逼超大规模算力企业绕过公共电网，通过“表后天然气+同址核能”的混合模式获取电力，在短期内推迟了可再生能源的转型。", "summary": "[2026-06-11] 针对电网材料瓶颈，本报告进行了压力测试：材料短缺并未阻止AI扩建，而是引发了“绕过电网”的结构性偏移。为规避长达5年的并网等待和3-5年的变压器交付期，云厂商正转向建设周期仅12-18个月的“表后”天然气以满足即期需求，并投资SMR及核电重启锁定长期基载电力。这种“天然气…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "会话研究内容达A−级且首尾自洽，但因5张卡片文件磁盘缺失与两处元数据草稿泄漏，发布门槛判为NEEDS REVISION；缺陷均为操作/编辑层面，完成清单R1–R7后即可发布。", "summary": "对全部8张卡片的终审：分析弧线（流动性事件→AI资本开支韧性→半导体国产化→电网/变压器物理瓶颈→能源供应压力测试）连贯自洽，可核验处算术全部正确，核心洞见（限速器从资金迁移到兆瓦）成立。但发布门槛因三项缺陷未过：(1)阻断级——24份强制文件仅6份在位，卡02–06全部磁盘缺失；(2)卡02/0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 10, "neutral": 0, "value": 12.8, "events": [ { "title": "OBBBA带来的能源并网瓶颈和先进制程资本成本飙升正全面重估全球半导体与AI算力的资本开支，而A股国产替代链条凭借独立能源体系和本土需求展现出极强的抗风险配置价值。", "summary": "距离OBBBA法案7月4日红线仅22天，变压器短缺导致绿电并网排队撤回，电网容量费暴增9.3倍，大幅压低云厂商AI投资回报率。同时台积电2nm晶圆达3万美元且WACC升至10%，令先进晶圆厂NPV腰斩，引发全球半导体设备订单下修。但A股受自主电网与国家大基金驱动，具备非对称免疫优势。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI与能源的物理错配：OBBBA政策悬崖下全球半导体Capex的NPV重估", "summary": "AI与能源的物理错配：OBBBA政策悬崖下全球半导体Capex的NPV重估 AI-Energy Mismatch: Reassessing Global Semiconductor Capex and NPV under the OBBBA Cliff 在距离OBBBA法案7月4日开工红线仅22天…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI与能源的物理错配：OBBBA政策悬崖下全球半导体Capex的NPV重估", "summary": "AI与能源的物理错配：OBBBA政策悬崖下全球半导体Capex的NPV重估 在距离OBBBA法案7月4日开工红线仅22天、且变压器积压导致绿电并网排队项目创纪录撤回的背景下，电网容量费暴增9.3倍如何侵蚀云厂商AI ROI？台积电2nm晶圆$3万高价及WACC升至10%如何导致晶圆厂NPV腰斩并迫…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源设备端的“无现金增长”风险正向二线及长交期环节集中，并网瓶颈将导致订单向现金流转化的效率降至历史低点，需警惕资本开支周期错配引发的流动性压力。", "summary": "2026-06-12：AI基础设施面临“电网天花板”，全球并网排队达2600GW。能源设备行业出现双峰分化：一线巨头（如伊顿、西门子能源）凭借垄断地位收取预付款，现金流极佳；而二线及A股厂商因变压器等长交期设备（24-48个月）囤货和项目并网延迟，陷入应收账款与库存激增的“无现金增长”陷阱，资产负…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 10, "neutral": 0, "value": 22.2, "events": [ { "title": "算力中心电力交付延宕将引发AI芯片短期需求空窗，倒逼国产半导体产业链加速向基于Chiplet的低功耗架构及低成本先进封装转型。", "summary": "2026年6月14日，基于变压器160周交期造成的“有芯无电”困境对半导体产业链进行压力测试。算力中心建设延后导致2026-2027年AI芯片面临需求空窗期，倒逼国产AI芯片从“盲目堆算力”向“追求能效比（Perf/Watt）”的架构转型。在资本开支被电网设备挤占的背景下，国内厂商加速采用Chip…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘冲突带来的能源价格上行与物理并网延迟正在迫使云厂商将资本开支从纯算力硬件向电网、变压器和温控液冷基建倾斜，产生显著的“基建挤出效应”；在脱碳路径上面临自备燃气发电（高ROIC但需…", "summary": "截至2026-06-13，本报告针对中东地缘能源冲击与电网瓶颈对AI算力基建的影响进行了压力测试。在$110/MWh高电费与$1364.40M配套基建（变压器GOES溢价及液冷）刚性支出下，超大云厂商为维持10%靶向ROIC必须调降GPU采购预算54.32%，挤压芯片商毛利至45.27%。在能源脱…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国绿电装机增速够用，但电网扩容（特高压+主变+构网储能）是承接AI算力负荷的真正瓶颈，也是2026–2027年A股能源链最确定的盈利兑现环节。", "summary": "截至2026-06-13，中国绿电装机端继续超预期（截至2026-04风光累计>14.5亿千瓦，2025全年新增约3.5亿千瓦[S5]），但弃风率5.1%、弃光率4.6%[S6]显示消纳与跨省外送是真正瓶颈。IEA口径下全球数据中心用电从2024年415 TWh升至2030年945 TWh[S1]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "特高压与主变压器的全球性产能紧缺以及构网型储能的政策刚性，将推动中国西电、平高电气等A股电网侧龙头在2026-2027年迎来强劲的量价齐升与报表端利润高速兑现。", "summary": "在国网加码算力配套电网投资背景下，特高压设备与主变由于产能限制，交付周期显著拉长（特高压排产延至2027年）；构网型储能因并网红利加速放量。A股龙头中国西电、平高电气、思源电气及许继电气在手订单充沛，2026-2027年盈利兑现确定性高，弹性大。", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 4, "neutral": 0, "value": 51.6, "events": [ { "title": "电网并网容量及绿电约束倒逼算力资本开支分流并延后，建议下调A股“科技进攻端”并超配“实物红利端”（区域火电与电网设备），并设立以“电费上浮”与“并网延期”为核心的双轨尾部风险触发器。", "summary": "截至2026-06-14，鉴于一线城市电网并网物理瓶颈（排队长达24-36个月）与绿电指标约束，导致下半年算力资本开支落地延后且向电力配套分流（占比自15%升至25%），我们建议下调A股“科技进攻端”权重（自30%降至26%，出清高估值两融拥挤的GPU租赁SPV，转向先进封装和受推理通缩驱动的AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观政策导向下的AI算力基础设施建设与能源瓶颈研究", "summary": "支持卡片01的裁决并补上物理层：2026年超大规模厂商AI资本开支约6900亿美元（Dell'Oro 7260亿）真实可验证，背后是IEA口径下数据中心用电从485TWh翻倍至950TWh的结构性需求。但约束已从芯片转向电力——美国数据中心电网需求2026年75.8GW升至2030年134.4GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力需求扩张与电力供给成本韧性的匹配度", "summary": "本卡片压力测试卡片08的火电成本韧性结论，指出其隐含的概念偷换：容量电价锁定的是火电供应方的成本/利润韧性(A层)，而AI算力扩容真正考验的是物理发电充裕度(B层)，尤其是负荷落点处的电网/输电/设备韧性(C层)。容量电价对C层几乎无能为力。定量看，中国数据中心仅占全社会用电约2–3%，全国能源充…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器原材料瓶颈（取向电工钢和铜）不会导致GPU需求破坏，而是导致变压器价格飙升及严重的闲置算力折旧损失，迫使科技巨头扭曲资本开支，大举向电力能源基础设施渗透。", "summary": "本报告研究了2026年取向电工钢（GOES）和铜的供应约束对电力设备变压器OEM利润率及AI算力交付的影响。结果显示，虽然原材料价格上涨（重度冲击下GOES+40%，铜+30%）会大幅压缩固定价格合同利润率，但OEM凭借极强定价权通过指数联动将成本完全转嫁给买方。AI算力未出现需求破坏，而是表现为…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 1, "risks": 4, "neutral": 1, "value": 64, "events": [ { "title": "功率半导体估值应从“新能源车beta”切换到“每个可送电MW的AI电力基础设施内容量”，但溢价只属于进入800 VDC/SST/DC保护/高密度PSU认证平台的SiC/GaN供应商。", "summary": "截至2026-06-15，本卡支持第06张卡的firm power稀缺主线，并将其延伸到功率半导体估值：SiC/GaN不再只按新能源车和快充周期定价，而应按800 VDC、固态变压器、半导体断路器、高密度PSU和电网/数据中心认证平台中的内容量定价。Infineon、ST、TI、Wolfspeed…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力配电设备产能扩张与供应链瓶颈：AI算力建设的硬约束 分析师：工业制造分析师（industrials analyst） 立场：support（强化 Research note 02…", "summary": "电力配电设备产能扩张与供应链瓶颈：AI算力建设的硬约束 分析师：工业制造分析师（industrials analyst） 立场：support（强化 Research note 02 的\"电力设备瓶颈是AI算力部署真正天花板\"判断） 工作日期：2026 06 15（Asia/Singapore） …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩容的关键能源约束是“正确节点上的稳定MW”，而不是单纯的上游油气供给；资本纪律主要通过价格、天然气可交付能力和电网投资缺口放大瓶颈。", "summary": "截至2026-06-15，本卡对上一张能源资本纪律逻辑进行压力测试。结论是：上游资本开支收缩会放大AI集群的能源成本和可靠性风险，但核心瓶颈不是全球油气分子短缺，而是特定电网节点可交付的稳定电力。IEA预计全球数据中心用电从2024年的415 TWh升至2030年约945 TWh；DOE/LBNL…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI权益溢价应从泛算力beta迁移到可交付电力、电网设备、成本回收utilities、现有可调度电源和已锁电Hyperscalers。", "summary": "截至2026-06-15，本卡片综合前序判断：AI算力仍是成长主线，但PJM和Virginia显示电力容量、并网和费率设计已经成为部署门槛。组合上应降低未锁电成长科技和泛AI硬件beta，提高电网设备、电气化、具备成本回收机制的utilities、现有gas generation和已锁电Hyper…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-16", "supports": 1, "risks": 7, "neutral": 1, "value": 47.4, "events": [ { "title": "变压器长达160周的交期与PJM电网逾5年的排队期构成了不可逾越的物理天花板，将实质性限制2026-2027年AI算力基础设施的资本开支落地与设备交付进度。", "summary": "基于2026-06-16的最新数据，本卡否定了前序关于“云厂商资本开支将提供坚实估值底部”的无摩擦假设。变压器交期已由疫前的12-16周激增至160周以上，Tier 1产能排至2030年，导致2026年美国30%-50%规划数据中心延期。PJM电网互联排队超5年，2026/2027容量电价触及32…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年后AI基础设施的核心溢价来自“可通电MW x 液冷转化效率 x 能源期限”，核能定终值，储能和电网设备定并网质量，液冷定GPU密度。", "summary": "截至2026-06-17，本卡支持上一张卡的判断：AI基础设施估值正从GPU数量转向确定性电力、并网质量和液冷密度。核电PPA提升长期站点价值，但多数新增容量在2027-2030年后才兑现；2026-2028年真正改变采购结构的是液冷机架、电气配套、储能和电网支撑设备。云厂商资本开支将更像“MW容…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI算力的“快时钟”正与电网基础设施的“慢时钟”发生剧烈碰撞，变压器产能缺口将迫使数据中心建设周期长期化并推高资本开支，形成实质性的硬约束。", "summary": "截至2026年6月16日，变压器与电网设备供应链已成为AI算力扩张的实质性物理瓶颈。大型电力变压器（LPT）交期延长至36-60个月，核心原材料取向电工钢（GOES）供应高度集中且价格较2020年上涨50-80%，导致变压器综合成本上涨60-80%。约30-50%的2026年计划产能面临延迟，建设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "吉文斯悖论与推理端计算膨胀导致算力及电力净需求呈指数级爆发，彻底推翻了电力基建搁置假说，反而迫使云厂商将资本开支结构向配套电网及液冷刚性倾斜，对高速光模块（CPO）等算力硬件产生相对…", "summary": "截至2026-06-17，大模型效率提升（如DeepSeek V4 API价格降至$0.15/百万Token）将触发吉文斯悖论与推理端算力膨胀。测算显示，即便MoE使单Token算力开支减少70.0%，但由于API调用量激增35倍且推理端Token密度升至877.5个，AI算力总需求仍暴增61.4…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 6, "neutral": 0, "value": 51.8, "events": [ { "title": "电网与变电设备瓶颈将迫使云厂商将算力开支向能源基建分流，引发全球先进制程设备订单的牛鞭效应，但国内半导体设备商凭借国产替代及本土扩产需求展现出极强的抗风险韧性。", "summary": "截至2026-06-17，本报告对AI算力基建电力/电网瓶颈向半导体设备订单的传导进行了压力测试。测试显示，重度压力情景下云厂商算力芯片开支将大幅收缩40%，触发WFE订单牛鞭效应，使全球设备综合订单收缩22%，订单出货比降至0.72。然而，北方华创和中微公司等国内设备商因受益于本土国产替代与扩产…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "西方电力设备巨头的产能释放滞后于AI并网高峰，且由于零部件卡口与合同刚性，中国产能扩张无法有效减轻其高压交付压力。", "summary": "截至2026-06-18，西方巨头（日立能源、西门子能源、GE Vernova）已启动数十亿美元的变压器产能扩张，但核心产能释放集中于2028年以后，与2026-2027年AI数据中心建设高峰存在严重错配。由于地缘壁垒，中国产能无法直接参与美欧大型高压输电变压器建设，且受限于套管短缺、劳动力匮乏和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "WACC上行30bp会打击AI资产估值倍数，但不会单独瓦解AI基础设施资本开支；防御性集中在有订单、有瓶颈、有现金流的硬科技与电力/电网环节。", "summary": "截至2026-06-17，本卡压力测试第06张卡的硬科技杠铃：WACC上行30bp会压缩长久期估值，但不足以单独击穿hyperscaler的AI数据中心、算力、电力与电网资本开支。8.0%到8.3%的WACC变化仅使5年资产年度资本回收系数上升0.8%、15年资产上升1.8%、30年资产上升2.8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI主题内部的定价权正从芯片端的硅片稀缺向电力、并网熟地和变压器等物理基础设施供应链迁移，电网并网延迟构成了对云巨头闲置算力资本的巨额惩罚，迫使价值向能源与硬件节点转移。", "summary": "截至2026年6月18日，由于全美并网队列激增至2,600 GW且平均等待时间达5年，AI基础设施的核心瓶颈已从GPU芯片稀缺转移到电力、并网及电力设备。定价权正向控制熟地、独立发电和电网设备的物理环节迁移，以规避巨大的闲置算力沉没资金惩罚（10万张B200芯片闲置年资本占用成本达1.8亿美元）。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 5, "neutral": 0, "value": 51.2, "events": [ { "title": "能源设备供应瓶颈对AI电力需求响应的影响分析", "summary": "截至2026-06-19，本卡压力测试 Card-02 的 A/B/C/D 电力篮子：变压器与开关设备交期延长强化 A 电网输配电的压舱石属性，但也限制整个 AI 电力交易的可部署容量；现场发电 B 受益于并网延迟但受燃机/发电机组排产约束；核燃料/SMR 的 C 应从高 Beta 战术仓位降到更…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新增AI风险预算应从继续加半导体Beta转向电力稀缺链条，半导体保留核心仓位但不再是边际加仓方向。", "summary": "截至2026-06-19，本卡支持并推进此前从通用软件/加密杠杆转向AI硬约束的思路：半导体基本面仍强，但新增风险预算的边际胜率已转向电力设备、并网基础设施、现场供电和核燃料。建议下周新增AI风险预算60-70%投向电力与供能约束，20-30%保留精选半导体，0-10%用于现金或对冲。核心依据是h…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新增可再生能源并网受限于极低的ELCC折价与长达3年的变压器交期，无法缓解2026-2028年算力‘通电围墙’，已投运基荷电力与变压器硬件仍是核心去瓶颈敞口。", "summary": "截至2026-06-18，对可再生能源并网缓解AI算力缺口进行压力测试。结论表明，新增绿电并网受限于电网硬件交付周期与容量额定值（ELCC）严重贬值，无法缓解2026-2028年算力‘通电围墙’。太阳能ELCC在PJM/ERCOT已被贬值至10-13%，且大型电力变压器（LPT）面临128-144…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由于大型电力变压器与测试大厅产能严重受限，2027-2028年计划新增的AI数据中心有43%面临通电并网延期[自测算]，由此引发GPU订单推迟，对主要芯片制造商及云厂商带来显著的收入…", "summary": "基于2026-06-19数据，本分析量化了电网变压器与并网延迟对2027-2028年AI硬件产业的影响。模型显示，原计划30GW数据中心管线因变压器交期拉长至120-210周而延期12-24个月，实际仅能交付17.1GW，缺口达43%[自测算]。这导致GPU出货量在2027年减少4.8M颗、202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 3, "neutral": 1, "value": 67.6, "events": [ { "title": "常规电网并网延迟与新能源项目财务崩溃，迫使云厂商将资本开支刚性移向自备微电网与高密液冷配套，配套设施开支上升72.22%对纯算力硬件产生挤出效应，但也实质上支撑并延长了AI物理层设备…", "summary": "截至2026-06-20，本卡片评估了能源供应与并网效率对AI物理层投资周期的支撑性。常规电网3-5年的并网延迟使100MW数据中心贴现NPV折损达51%至80%，迫使云厂商向“表后直供+储能”的微电网转型。虽然配建100MW/400MWh储能使LCOE上升14%至$55.10/MWh，但每小时C…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI算力Capex与全球电力/电网瓶颈共同提供了足够大且外生于国内消费的订单池，足以承接卡片02杠铃中30%先进制造Alpha端的盈利兑现，杠铃配置成立。", "summary": "截至2026-06-20，本卡支持卡片02杠铃中30%先进制造Alpha端的可行性。美国Big-4超大规模厂商2026年Capex合计约USD 4,400亿，同比+29%[S1-S4]；叠加中国互联网巨头RMB 3,800亿AI Capex，全球AI基础设施Capex约USD 4,900亿(+27…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-19，鉴于Q4 SPCX解禁对4%自由流通盘的价格冲击风险，模型组合表达AI基础设施主题应从直接叙事（SPCX和半导体）退步，通过将70%权重配置给电网设备和核电I…", "summary": "截至2026-06-19，本报告对AI基建主题的最佳表达载体进行综合评估。针对2026年Q4的SPCX员工与内部人解禁压力测试，自测算表明SPCX当前4%的公众流通盘易受日元平仓和宏观去估值冲击。组合测试了三类表达权重：策略1（直接叙事）夏普比率为-0.12，策略2（物理瓶颈防守：0% SPCX、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股AI电网capex链条配置优先级：上游材料瓶颈 > 电网设备OEM > 出口龙头 > 资产型公用事业。", "summary": "截至2026-06-20，本卡综合前五张卡片，给出AI电网资本开支链条在A股的优先级：上游高壁垒材料瓶颈（高牌号GOES、CTC铜、绝缘材料、酯类介质液）>电网设备OEM（特高压主设备、数字化二次、SVG）>出口龙头（变压器/开关DM替代窗口）>资产基数增长型公用事业。前两档承担弹性主战场，第三档…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 3, "neutral": 0, "value": 66, "events": [ { "title": "AI半导体供应链的资本开支压力与交付瓶颈", "summary": "压测显示：WFE与核心材料交期真实拉长，但截至2026-06-21并非AI加速器产出的首要新瓶颈，而是排在CoWoS封装之后、远落后于电网的第四个串联时钟。约束排序（中值[自测算]）：前道晶圆16周→HBM/DRAM设备52周→CoWoS 65周→WFE EUV 78周→High-NA 104周→…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面向多年期AI电力资本开支超级周期的配置：刚性容量稀缺确立后的板块配置与风格含义", "summary": "本卡把四环串联瓶颈（封装→电网/变压器→刚性发电）翻译成配置地图。关键事实：四个时钟串联，缓解任一环只把约束移向下游，不压缩通电总时间，故稀缺是结构性、多年期的，而非周期性。配置：超配机电卖铲人（电网/变压器，全球缺口约30%、中国占产能60%、出口+36%、订单排至2027）、刚性发电（核电/铀…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力瓶颈与绿电利用率分化对A股电力设备、公用事业和新能源运营商的策略影响", "summary": "承接Research note 02–08全链条，本卡以发改能源〔2026〕688号多用户绿电直连（2026-05-20，优先支持算力设施）为基准情形，重构A股电力链定价。核心：688号把AI负荷从挤压新能源并网的拥塞负荷，翻转为在资源地变现弃电的锚定承购（2026Q1光伏利用率90.6%、隐含弃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 5, "neutral": 0, "value": 55.4, "events": [ { "title": "核电与可再生能源供应侧的物理瓶颈：规避并网排队能否解决电力规模缺口？", "summary": "压力测试卡片01的监管破局论。FERC的Eddystone容量置换豁免仅是配置抢跑，无法制造净新增可靠电源。需求增量巨大：EPRI测算美国数据中心峰值负荷将从2024年约21.5GW升至2030年中性情景71GW，即约+49.5GW[自测算]。供给侧无法匹配：核电重启/扩容到2029年仅约+5GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-2028年，电池、VPP、灵活数据中心电价和GETs是跨越变压器短缺的桥，不是绕开短缺的旁路。", "summary": "截至2026-06-21，本卡压力测试认为，变压器和电网设备产能正在扩张，但新增工厂、认证、熟练劳动力和长交期决定其主要缓解2029年后的出清，而非完全解决2026-2028瓶颈。储能、VPP、数据中心需求侧管理和GETs更适合作为近端过桥工具：它们可以削峰、递延部分网络升级、支持灵活并网，并提高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美欧贸易壁垒足以压低中国电力设备出海的利润率和估值倍数，但不足以抵消无源、短缺约束型电网硬件的出口红利。", "summary": "截至2026-06-21，本卡否定“美欧关税和贸易限制足以抵消中国电力设备出海红利”的强判断。政策摩擦真实存在：美国Section 301覆盖部分HS 8504及能源相邻品类，欧盟通过NZIA、FSR和高风险逆变器融资限制提高准入门槛。但美国并网队列、变压器80-210周交期、DOE披露的进口依赖…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心电力瓶颈与核电直供的监管破局：从Talen-AWS僵局到Constellation-微软Eddystone容量豁免的行业拐点", "summary": "分析了2026年6月初美国联邦能源监管委员会（FERC）批准Constellation Energy将760 MW的Eddystone电厂容量互联权（CIRs）转移至Crane清洁能源中心（三哩岛核电站）的豁免裁决。对比了Talen-AWS在此前Susquehanna项目表后直连遭拒后被迫转型为表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 4, "neutral": 0, "value": 62.8, "events": [ { "title": "电网设备紧缺与并网延迟正倒逼AI云厂商调整选址逻辑，通过工作负载地理分流、核电结构化直供、现场过桥发电以及电气设备提前大批量采购，来对冲并网延误风险，将AI资本开支推向重度能源与电力…", "summary": "截至2026-06-22，高压变压器交期拉长至24-48个月且排队并网超5年，成为AI集群部署核心瓶颈。超大规模云厂商正通过将训练与推理负载地理分流、签署核电结构化购电协议（PPA）、部署现场天然气发电机过桥以及提前3-4年锁定集团化设备产能等调整选址与容量规划，以对抗延迟风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超大规模云厂商刚性的AI资本开支与先进封装产能的物理瓶颈，锁定了半导体前道关键设备和先进封装工具的在手订单，使其免受宏观周期与电网延迟的传导冲击，确立了其作为防御成长底仓的逻辑。", "summary": "截至2026-06-22，研究表明北美云厂商2026年6350亿-7500亿美元资本开支（75%绑定AI）具有非discretionary刚性。台积电CoWoS封装产能（2026年底达12万-14万片/月）面临严重短缺，限制英伟达Blackwell GPU出货上限至445万颗。这为半导体设备链提供…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "以变压器和开关设备为核心的电力基础设施交付周期暴涨至2.5-3.5年，已彻底超越半导体产能瓶颈，成为制约AI集群落地并迫使算力空间地理重组的最核心约束。", "summary": "截至2026-06-22，本卡评估了AI集群面临的电力设备瓶颈。测算显示，大型电力变压器（LPT）交付周期拉长至128-180周（约2.5-3.5年），开关设备为60-90周，远超GPU的12-16周。对100 MW站点测算表明，1.5年电网延迟将使10亿美元GPU集群产生5.2295亿美元折旧与…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源冲击不会终止AI基础设施周期，但会通过电价、容量费和并网延误降低同等资本开支转化为在线算力的速度。", "summary": "截至2026-06-22，本卡支持上一张宏观链条：能源价格冲击与电网瓶颈会把AI资本开支从“GPU采购问题”改写为“确定兆瓦与送电进度问题”。IEA、DOE/LBNL、PJM、EIA和云厂商披露共同显示，数据中心用电增长、容量市场重定价、天然气边际电价和高密度AI机架功耗正在抬升总拥有成本，并可能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 1, "risks": 8, "neutral": 0, "value": 48.2, "events": [ { "title": "电网输变电容量与并网审批周期（而非总体发电量）已成为AI算力部署的终极物理刚性瓶颈，并已实质性挤压IT硬件采购预算。", "summary": "本报告对AI算力扩张的电网与能源瓶颈进行压力测试。分析表明，虽然电力设备订单充足，但美国并网队列平均达5.5年，且面临FERC 2026年6月大型负荷监管及弗吉尼亚GS-5等防止成本转嫁政策。变压器等重电面临GOES原材料短缺，Tier-1厂商积压至2029-2030年。表后核电直供监管被封死，燃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力能效提升引发的“需求弹性”对超大规模数据中心与能源基础设施资本配置的影响", "summary": "本卡作为收官综合卡，支持卡片09的判断：以密度而非节流出现的能效改善是杰文斯反弹，不缩减预算而是把结构向电力链轮动。2026年四大云厂商CAPEX约7250亿美元、同比+77%，约75%与AI相关；能效降低单位Token能耗反而抬高需求量。当前建造CAPEX约60%芯片、约40%厂务/电力/制冷；…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电网并网瓶颈迫使AI负荷直连一次能源，驱动天然气及煤炭需求与价格重定价，并倒逼公用事业开启$1.4万亿美元的设备通胀型Capex周期。", "summary": "截至2026-06-23，本卡研究AI算力负荷对能源价格及投资周期的影响。由于电网扩容滞后5年以上，hyperscaler被迫转向表外直连（BTM）或自备发电机，拉动美国天然气需求增量达1.47-2.20 Bcf/d，并延缓煤电退役。规管公用事业被迫启动$1.4万亿美元Capex，致变压器价格暴涨…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支削减15%将导致全球约1.19 GW的数据中心部署延迟，阶段性缓解电力变压器供需瓶颈和电网并网压力，同时导致独立发电厂与电力设备标的面临估值去化风险。", "summary": "截至2026-06-23，本卡评估了AI资本开支下修对数据中心、电网组件及能源供应链的传导效应。模型表明，15%的AI资本开支削减（-661.5亿美元）将导致全球约1.19 GW的物理数据中心IT容量建设延迟。这将有效缓和大型电力变压器（交期24-48个月，价格较2019年上涨77-95%）和输电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 7, "neutral": 0, "value": 55.6, "events": [ { "title": "液冷驱动的超高功率密度对电网配电侧建设与能源供应响应力的挑战", "summary": "经压力测试：在受监管公共电网层面答案是“不能”，园区层面仅对提前占位者有条件“能”。密度对电网不是关键变量——100MW园区无论20kW还是140kW机柜都取电约100MW[自测算]，密度只是把约束下移到馈线/变电站/变压器并加剧负荷波动（已触发NERC突发失负荷预警）。时间错配是结构性的约5.6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力投资需求规模与电网/能源供应缺口的动态匹配分析", "summary": "截至2026-06-25，本卡量化AI算力需求通道与电网/能源物理供应瓶颈的动态匹配。需求侧：4家超大规模厂商2026年capex约7250亿美元(+77%)，美国数据中心峰值负荷2024约21-22GW、EPRI中性情景2030升至71GW。供应侧：并网积压约2600GW(中位约5年)、大变压器…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力资本开支超级周期的电力/电网天花板及其对新一代信息技术全链的约束", "summary": "电力/电网上限真实存在且已是绑定约束：美国并网 3–7 年、变压器 4–5 年、燃气轮机售罄到 2030、30–50% 的 2026 产能滑至 2028 [S5-S8]；微软约 800 亿美元 Azure backlog 受电力而非需求约束 [S9]。但 2026–2028 内它递延而非毁灭需求—…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网延迟引发的GPU‘折旧陷阱’和24/7绿电小时级匹配的平准化电价溢价严重侵蚀AI数据中心投资回报，倒逼云厂商转向‘自带能源’（BYOP）现场发电以保护工期，但该模式对天然气价…", "summary": "截至2026-06-25，本卡对电网瓶颈下AI数据中心的资本开支效率进行了压力测试。若并网延迟超2年，GPU技术寿命的短暂将导致“折旧陷阱”，使项目NPV损失超60%。自带能源（BYOP）自发电虽增加多达1.5亿CapEx且电价高昂，但凭1至1.5年投产优势，相比5年并网排队能挽回数亿NPV，具备…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 3, "neutral": 0, "value": 73, "events": [ { "title": "在限制性利率与结构性电力成本溢价下，市场对‘可交付性约束’的忽视将导致AI芯片链面临重定价，资金应向电网设备通道、独立发电商及高股息红利价值风格集中。", "summary": "截至2026-06-25，本卡合成2026年能源冲击与电力稀缺对大类资产及A股/全球股票轮动的影响。电力并网瓶颈与大变压器缺口推升期限溢价及美联储政策利率（更高更久），但实际r*仅温和上升+25至+50bp。资产配置应实施哑铃结构：超配电网设备与燃机OEM（受益于全球变压器缺口及交期溢价）、超配红…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备原材料缺口对新能源项目建设成本与并网进度的传导效应评估", "summary": "支持卡片06–08的逻辑链：AI电力瓶颈已上移至原材料，且GOES/高纯度铜缺口确实传导至新能源经济性，但传导非对称、主要走时间而非单价。直接价格通道有限——铜仅占光伏/陆上风电资本开支约3%~5%，即便铜较2020年+50%、GOES+100%，对LCOE直接抬升仅约+2%~+4% [自测算]。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源降本对AI算力基础设施的影响及扩容瓶颈评估", "summary": "压力测试卡片-02的能源降本看多论。燃气降本（CCGT减负$25.76/MWh、100MW数据中心年省OpEx约$24.94M）真实但属次要，仅折合AI级capex的1.7%–2.5%。真正约束是time-to-power：美国并网队列卡着约2,300GW、PJM平均排队约8年、Dominion北…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 5, "neutral": 0, "value": 62.4, "events": [ { "title": "电网基础设施与能源转型对AI算力扩张的潜在约束", "summary": "压力测试部分支持并升级卡03：电力/电网瓶颈是“节流而非断路”，但节流比其暗示更紧更久。在2026–2028视野内，电气基础设施已是排在芯片之前的硬约束——美国电力变压器交期128周、GSU 144周、大单3–4年，价格较2019+77%、达2022前4–6倍；GOES由Cleveland-Cli…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基于电力电网物理瓶颈（变压器交期120-210周）导致资本开支递延800亿-1100亿美元，高PE半导体叙事久期面临压缩，A股大类资产配置应低配半导体至10%-13%[自测算]，并重…", "summary": "本报告针对电力电网硬约束与资本开支递延风险，对A股资产进行大类风格与板块相对配置量化重构。压力测试表明，变压器产能受限（交期120-210周）及并网排队（7-10年）导致2026-2027年云厂商约800亿-1100亿美元资本开支递延[S8]，从而证伪了半导体高PE板块（CSI半导PE TTM约1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "保留12%清洁+4%能源杠铃，但A股表达必须从宽基清洁贝塔转为受限的电网/储能/运营商组合，并用能源对冲抵消价值和通胀尾部而非追高红利能源。", "summary": "截至2026-06-26，本卡综合前序12%清洁基础设施+4%能源尾部对冲，但指出A股执行并非因子中性。清洁袖主要暴露于成长、资本开支、低股息和动量，尤其电网设备、电池、储能、光伏存在追涨和集中度风险；能源袖提供价值、高股息和通胀/霍尔木兹尾部保护，但中证能源前五权重约61.0%，不能无限扩张。建…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 5, "neutral": 0, "value": 63.8, "events": [ { "title": "全球能源供应增量（特别是重启周期的核能与不稳定的可再生能源）及慢速的电网升级在2026-2027年内无法有效对冲 AI 基础设施的电网并网刚性瓶颈，该瓶颈将直接导致大云厂商面临 80…", "summary": "全球AI算力扩张正面临电网并网物理刚性瓶颈。全球电网升级周期长达4-10年，排队时间中位数达5年，且高密度变压器等关键主设备面临交付延迟，直接导致美国约7/12 GW在建产能延期，并将造成2026-27年全球超大云厂商约800亿-1100亿美元资本开支递延。核电与可再生能源虽有高成长，但受制于重启…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据中心电力并网、关键电力设备与液冷温控的物理交付瓶颈将对算力上线速度形成硬约束，导致云厂商ROIC受损并延缓算力采购，增加电子产业链利润扩张的硬着陆风险。", "summary": "截至2026-06-28，本卡压力测试了算力产业链利润扩散假说，指出电力并网与变压器超长交期（MV变压器128周）已成为数据中心上线的物理瓶颈。预计2026–2028年美国35%–45%的新建数据中心面临延期通电，16GW计划容量仅5GW在建，30%-50%将延期至2028年以后。GB200的高功…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "summary": "截至2026-06-27，本卡片综合前序结论：电网/变压器瓶颈会延后并改道AI部署，但不会摧毁2028年前光模块出口龙头业绩能见度；真正尾部风险是2027年云厂商ROIC压缩。建议A股AI物理瓶颈组合使用40/32/12/8/8结构：变压器/电网设备、光模块、电力现金流锚、铜/GOES等输入端对冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本与高利率壁垒对全球算力中心电力建设与资本开支的传导效应", "summary": "支持Research note 1-2逻辑链并完成最后一公里传导。AI算力硬约束已从GPU芯片迁移至电力、再到电网可交付能力——击穿点不是电价而是无法通电[S1][S2]。两条通道：并网延迟（队列超2000GW、中位等待4-5年、落地率仅约13%、>95%为间歇性电源[S6][S1]）与电价刚性（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 14, "neutral": 0, "value": 12, "events": [ { "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "summary": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-28，物理加电瓶颈正迫使AI算力估值由单纯的Capex扩张转向已通电MW效率，投资者应超配锁电、BTM自发电超大型云厂商与掌握变压器等硬缺口产能的电力设备商，并在A…", "summary": "截至2026-06-28，本报告探讨了在AI电力约束下算力资产估值向MW效率转型的逻辑。对于超大规模云厂商，2年的并网延迟将对100 MW智算中心产生严重的NPV减记（-124%）[S12]，迫使其估值分层：超配锁电与BTM自发电龙头，低配单纯Capex扩张者。电力设备商成为AI周期收费站，超配大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI驱动的结构性裁员实质上是算力投资回报（ROI）兑现的证明，云厂商AI资本开支具有极强韧性，其真实制约并非宏观衰退担忧，而是电网并网延迟与重型电力设备的物理交付瓶颈。", "summary": "本卡片对“AI裁员压力动摇算力基建投资”的假设进行压力测试并予以否定。AI驱动的结构性裁员（Challenger裁员AI占比达40%）是AI投资回报（ROI）兑现的实体证据，而非衰退信号。四大云厂商资产负债表坚韧，AI基建投资是一场生存竞争。然而，物理供电（并网排队5-7年）与设备交付（变压器交期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网容量瓶颈并非完全抵消半导体国产替代红利，而是通过系统周期错配将压力传导给中下游服务器与算力商，同时倒逼云厂商转向表后自备电力与地理重构。", "summary": "本报告对电网瓶颈是否抵消半导体国产化红利进行压力测试。物理并网延迟和关键设备短缺确对AI部署形成‘硬阀门’，延缓算力落地并导致2026-27年全球约800亿-1100亿美元CapEx递延。但这并未抵消自主化晶圆设备和封装厂的政策红利，而是在产业链内部造成分化：上游设备厂商基本免受冲击，中下游加速器…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 1, "risks": 7, "neutral": 0, "value": 62.6, "events": [ { "title": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "summary": "支持第06卡\"并网延迟→利润压缩\"的传导论点，但将单向结论细化为分化。针对11GW数据中心建设缺口与12-18GW刚性局部电网缺口，48-60个月并网延迟通过两条通道传导：(1)经订单/PPA递延与尾款拖延压制发电子链开工率——5.7%销量递延经经营杠杆放大为约-21%利润压缩[自测算]；(2)将…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "将能源链利润分化（发电侧承压、电网设备受益）整合进A股成长/价值杠铃配置", "summary": "本卡综合第05–07卡，重构A股成长杠铃。把第06卡腾出的8.0亿元大部分倾斜至利率不敏感、政策背书的电网设备一腿（+5.0亿），余下+2.0亿配高股息电网运营、+1.0亿留券商β/干火药，AI硬件保留4%残余尖端，绿电/发电设备零配置。电网设备订单免疫16.18%AI资本开支下修、可见至2027…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电…", "summary": "本报告对AI算力激增带来的电力电网物理瓶颈（变压器交期120-210周[S2]、并网排队4-7年[S1]）对A股转型龙头的利润率传导进行压力测试。结果显示，电网扩容滞后充当了新能源装机的“限速阀”，压制了光伏/电池销量增速；同时，容量电价[S8]和碳成本上升[S19]增加了高能耗制造环节电费开支，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "summary": "截至2026-06-30，本报告压力测试了变压器交期拉长对AI算力集群部署和云厂商资本开支的约束。变压器交期拉长（128-144周，专用型4-5年）导致物理并网时滞达4-7年，引发全球467万个GPU延迟并网和2万亿元Capex延期。微观上，1.5年延迟导致10亿美元GPU集群产生5.2295亿美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-30", "supports": 0, "risks": 1, "neutral": 0, "value": 92, "events": [ { "title": "AI数据中心并网延期与变压器死锁下全球算力硬件 Capex 的结构性错配", "summary": "AI数据中心并网延期与变压器死锁下全球算力硬件 Capex 的结构性错配 AI Power Grid Deadlock and Hyperscaler Capex Structural Misallocation 北美云巨头下修 H2 硬件 Capex 16.18%，在变压器交期达120-160周…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-1495ab3592/", "networkHref": "/investor/research-network/#thesis-1495ab3592" }, { "id": "thesis-2ec6b9d3ee", "source": "tracked_thesis", "title": "卫星直连与商业航天从概念弹性进入部署兑现期", "titleEn": "Direct-to-device satellite and commercial space are moving from concept beta to deployment proof", "lane": "商业航天/卫星直连", "laneEn": "Commercial space and direct-to-device satellite", "laneHref": "/investor/lanes/commercial-space-and-direct-to-device-satellite/", "investableFocus": "商业航天/卫星直连：ASTS, RKLB, LUNR, RDW, IRDM, LMT, 600118.SH", "stockUniverse": [ { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "RKLB", "name": "Rocket Lab", "market": "US", "role": "小型发射与空间系统", "roleEn": "小型发射与空间系统", "bucket": "core", "href": "/investor/stocks/RKLB/" }, { "ticker": "LUNR", "name": "Intuitive Machines", "market": "US", "role": "月球与空间服务", "roleEn": "月球与空间服务", "bucket": "core", "href": "/investor/stocks/LUNR/" }, { "ticker": "RDW", "name": "Redwire", "market": "US", "role": "空间基础设施", "roleEn": "空间基础设施", "bucket": "core", "href": "/investor/stocks/RDW/" }, { "ticker": "IRDM", "name": "Iridium", "market": "US", "role": "卫星通信现金流", "roleEn": "卫星通信现金流", "bucket": "core", "href": "/investor/stocks/IRDM/" }, { "ticker": "LMT", "name": "Lockheed Martin", "market": "US", "role": "国防主承包商", "roleEn": "国防主承包商", "bucket": "core", "href": "/investor/stocks/LMT/" }, { "ticker": "600118.SH", "name": "中国卫星", "market": "A-share", "role": "中国卫星制造与应用", "roleEn": "中国卫星制造与应用", "bucket": "core", "href": "/investor/stocks/600118.SH/" }, { "ticker": "BOTZ", "name": "Global X Robotics & AI ETF", "market": "US ETF", "role": "机器人与自动化", "roleEn": "机器人与自动化", "bucket": "watch", "href": "/investor/stocks/BOTZ/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 供给", "roleEn": "HBM 供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "ASTS", "RKLB", "LUNR", "RDW", "IRDM", "LMT", "600118.SH", "BOTZ", "TSLA", "MU", "005930.KS", "000660.KS", "APO", "IEF" ], "coreView": "赛道：商业航天/卫星直连。候选观察标的：ASTS, RKLB, LUNR, RDW, IRDM, LMT, 600118.SH。当前共有 40 条证据、4 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 74, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 4, "risks": 128, "neutral": 28, "analystCount": 11, "sourceTypeCounts": { "研究笔记": 143, "证据更新": 3, "分析协作": 5, "研究线程": 5, "研究主题": 4 }, "topTerms": [ "AI 基础设施", "商业航天", "ASTS", "运营商", "Direct-to-device", "satellite", "commercial", "space", "moving", "concept", "beta", "deployment" ], "whatChanged": [ "2026-06-29: AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "2026-06-29: AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "2026-06-29: 电力成本不确定性会直接拖慢西部AI算力基地投产节奏，但主要通过容量/并网/备用费用和到电时间压低Capex效率，而不是通过单一电价水平否定项目。", "2026-06-29: AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利景气度，A股资产配置应向电网出海及水电红利哑铃端重构[自测算]。", "2026-06-29: 电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对冲有效却对冲错轴，继续超配电网设备而非纯绿电发电。" ], "practical": { "sectorLane": "商业航天/卫星直连", "sectorLaneEn": "Commercial space and direct-to-device satellite", "exposureBucket": "航天国防与新场景", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "RKLB", "name": "Rocket Lab", "market": "US", "role": "小型发射与空间系统", "roleEn": "小型发射与空间系统", "bucket": "core", "href": "/investor/stocks/RKLB/" }, { "ticker": "LUNR", "name": "Intuitive Machines", "market": "US", "role": "月球与空间服务", "roleEn": "月球与空间服务", "bucket": "core", "href": "/investor/stocks/LUNR/" }, { "ticker": "RDW", "name": "Redwire", "market": "US", "role": "空间基础设施", "roleEn": "空间基础设施", "bucket": "core", "href": "/investor/stocks/RDW/" }, { "ticker": "IRDM", "name": "Iridium", "market": "US", "role": "卫星通信现金流", "roleEn": "卫星通信现金流", "bucket": "core", "href": "/investor/stocks/IRDM/" }, { "ticker": "LMT", "name": "Lockheed Martin", "market": "US", "role": "国防主承包商", "roleEn": "国防主承包商", "bucket": "core", "href": "/investor/stocks/LMT/" }, { "ticker": "600118.SH", "name": "中国卫星", "market": "A-share", "role": "中国卫星制造与应用", "roleEn": "中国卫星制造与应用", "bucket": "core", "href": "/investor/stocks/600118.SH/" }, { "ticker": "BOTZ", "name": "Global X Robotics & AI ETF", "market": "US ETF", "role": "机器人与自动化", "roleEn": "机器人与自动化", "bucket": "watch", "href": "/investor/stocks/BOTZ/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 供给", "roleEn": "HBM 供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "ASTS", "RKLB", "LUNR", "RDW", "IRDM", "LMT", "600118.SH", "BOTZ", "TSLA", "MU", "005930.KS", "000660.KS", "APO", "IEF" ], "primaryStocks": [ { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "RKLB", "name": "Rocket Lab", "market": "US", "role": "小型发射与空间系统", "roleEn": "小型发射与空间系统", "bucket": "core", "href": "/investor/stocks/RKLB/" }, { "ticker": "LUNR", "name": "Intuitive Machines", "market": "US", "role": "月球与空间服务", "roleEn": "月球与空间服务", "bucket": "core", "href": "/investor/stocks/LUNR/" }, { "ticker": "RDW", "name": "Redwire", "market": "US", "role": "空间基础设施", "roleEn": "空间基础设施", "bucket": "core", "href": "/investor/stocks/RDW/" }, { "ticker": "IRDM", "name": "Iridium", "market": "US", "role": "卫星通信现金流", "roleEn": "卫星通信现金流", "bucket": "core", "href": "/investor/stocks/IRDM/" }, { "ticker": "LMT", "name": "Lockheed Martin", "market": "US", "role": "国防主承包商", "roleEn": "国防主承包商", "bucket": "core", "href": "/investor/stocks/LMT/" } ], "instrumentTags": [ "ASTS", "AST SpaceMobile", "RKLB", "Rocket Lab", "LUNR", "Intuitive Machines", "RDW", "Redwire", "IRDM", "Iridium", "LMT", "Lockheed Martin" ], "reviewCadence": "每日复核", "confirmationSignals": [ "部署节奏是否兑现为收入", "BOM 降本是否快于竞争扩产", "政策订单是否能跨过概念交易阶段" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "事件兑现", "融资稀释", "订单延迟", "高波动仓位" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 104 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 11, "recentEvents": 104, "supportRatio": 0.03, "riskRatio": 0.8, "conflictScore": 91.3, "evidenceDepth": 100 }, "score": 67.3 }, "events": [ { "date": "2026-06-29", "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck from compute to power, networkin…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "AI 基础设施", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片的TAM…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730MW萎缩至28MW，给电网带来非自愿性减负。A股配置建议低配估…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "summary": "截至2026-06-30，本报告压力测试了变压器交期拉长对AI算力集群部署和云厂商资本开支的约束。变压器交期拉长（128-144周，专用型4-5年）导致物理并网时滞达4-7年，引发全球467万个GPU延迟并网和2万亿元Capex延期。微观上，1.5年延迟导致10亿美元GPU集群产生5.2295亿美元折旧与持有损耗，月度拖累1.1905亿美元，NPV骤降12…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利…", "summary": "本报告对AI算力激增带来的电力电网物理瓶颈（变压器交期120-210周[S2]、并网排队4-7年[S1]）对A股转型龙头的利润率传导进行压力测试。结果显示，电网扩容滞后充当了新能源装机的“限速阀”，压制了光伏/电池销量增速；同时，容量电价[S8]和碳成本上升[S19]增加了高能耗制造环节电费开支，而在严重过剩背景下成本转嫁失败，导致制造端转型龙头利润率遭到…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "电力成本不确定性会直接拖慢西部AI算力基地投产节奏，但主要通过容量/并网/备用费用和到电时间压低Capex效率，而不是通过单一电价水平否定项目。", "summary": "截至2026-06-30，本卡片支持卡片06的压力测试并进一步指出：电网费率重构和BTM不确定性会直接制约西部AI算力大基地，但核心约束不是单纯电价，而是“到电时间”、容量费用和接网责任。并网型BTM仍需缴纳输配电费、系统运行费、交叉补贴及基金附加，容量电价机制把可靠容量显性定价。战略性西部枢纽大概率继续推进，但边际民营/租赁型超大规模项目将被迫分期送电、…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "summary": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,800万美元（约降33%）[S1][自测算]；先进封装（CoWo…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power预期缺口，会迫使Hyperscaler预付并网升级、锁定变压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,600 亿）以经营现金流支撑，资金基础/地域/标的均不同于承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19, "matchedTerms": [ "AI 基础设施", "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "summary": "截至 2026-06-28，本卡压力测试 Research note 1 的 AI 基础设施使能层主线：需求真实且强劲，但 2026-2028 年兑现会被电网接入、可靠电源、变压器、燃机和运行可靠性约束。IEA 预计全球数据中心用电从 2024 年 415 TWh 升至 2030 年约 945 TWh；美国和中国合计贡献近 80% 增量。[S1][S2] …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19, "matchedTerms": [ "AI 基础设施", "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "<20%开机率红线是平台属性而非物理定律：对电力受限EO小卫星它由电源+被动散热共同锁定且先进热控不划算(支持卡03)，但两相泵驱回路+可展开辐射器已能让专用型150kW级在轨算力平台突破红…", "summary": "本卡按平台维度厘清半导体卡的<20%开机率红线，结论为有条件支持。对Planet/BlackSky类电力受限EO小卫星，红线由电源预算与被动导热双重锁定：15-60W的GPU在散热生效前已占150-300W总线的10-40%，加装两相回路+辐射器(约2-4kg、5-15W泵耗)成本收益为负，红线成立。但对专用型在轨算力平台(SpaceX AI1 150kW…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "space", "Lane", "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计约6900-7200亿美元（自测算），NVIDIA FY20…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 14.25, "matchedTerms": [ "beta", "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "Referenced files session brief.md , analyst catalog.json , and reports for Cards 1–4 were missi…", "summary": "Referenced files session brief.md , analyst catalog.json , and reports for Cards 1–4 were missing from the workspace and reconstructed from the conversation context. Wrote researc…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 14, "matchedTerms": [ "space", "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI算力基础设施需求与海底数据中心技术演进", "summary": "支持卡片01关于海兰信（300065.SZ）的判断并加以收紧：UDC卡位2026年AI建设的电力与散热两大硬约束。2026年超大规模云资本开支约7250亿美元，全球数据中心用电将破1000TWh（超日本），单柜功率向370–600kW飙升，高盛预计年底76%的AI服务器需液冷。海兰信是国内唯一掌握UDC全套技术、全球继微软后第二家的企业：海南陵水675PF…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "海缆降本的边际贡献远低于电力固化成本的拉动，电力成本敏感度上升使UDC项目IRR面临显著压降，其经济部署边界被牢牢锁定在电价高于0.70-0.78元/kWh的高关税沿海枢纽。", "summary": "截至2026-06-29，本报告对海缆毛利压缩与电力成本敏感度上升对海底数据中心（UDC）经济性及全球部署边界的传导进行压力测试。测算显示，24MW（IT负荷）的UDC模块在PUE 1.10下，136号文带来的绿电固化溢价（+0.10元/kWh）会增加年运营电费2313万元，折现后NPV减值1.55亿元；而海缆毛利收缩（自40%降至25%）仅能缩减CapE…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "infrastructure", "ai-infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "电网容量瓶颈并非完全抵消半导体国产替代红利，而是通过系统周期错配将压力传导给中下游服务器与算力商，同时倒逼云厂商转向表后自备电力与地理重构。", "summary": "本报告对电网瓶颈是否抵消半导体国产化红利进行压力测试。物理并网延迟和关键设备短缺确对AI部署形成‘硬阀门’，延缓算力落地并导致2026-27年全球约800亿-1100亿美元CapEx递延。但这并未抵消自主化晶圆设备和封装厂的政策红利，而是在产业链内部造成分化：上游设备厂商基本免受冲击，中下游加速器与服务器商的财务和出货受抑。中国通过特高压、东数西算与绿电直…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "infrastructure", "ai-infrastructure" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "电力交付瓶颈不否定 AI 基础设施主线，而是将价值从泛云计算和 AI 期权向拥有变压器、核电容量和受监管输电资产定价权的公用事业链条转移。", "summary": "电力约束真实存在且层次复杂，但约束的是建设节奏和选址，而非总体需求规模。美国 PJM 并网队列等待 5–7 年、变压器交货期 3–5 年、24/7 清洁电力供给与需求的错配，导致大量项目面临 2–4 年时间滑移。中国凭借政府主导并网（12–24 个月）和西部廉价可再生能源占优，但延迟与数据主权限制…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 0, "neutral": 1, "value": 46.4, "events": [ { "title": "阿里巴巴 (BABA) 承诺投入 550 亿美元用于 AI 基础设施建设。", "summary": "阿里巴巴 (BABA) 承诺投入 550 亿美元用于 AI 基础设施建设。", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-10", "supports": 1, "risks": 0, "neutral": 0, "value": 52.8, "events": [ { "title": "Saudi Telecom Company (stc) Group 向 AST SpaceMobile (ASTS) 支付了1.75亿美元商业预付款。", "summary": "Saudi Telecom Company (stc) Group 向 AST SpaceMobile (ASTS) 支付了1.75亿美元商业预付款。", "sourceLabel": "证据更新", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-12", "supports": 0, "risks": 0, "neutral": 1, "value": 49.2, "events": [ { "title": "[auto-handoff] SpaceX IPO Premium Spillovers", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 873ecc26-ae0a-4100-8e83-76b1719d563a 结果摘要: Wrote handoff_response.md (11885 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "AI 基础设施估值锚必须从‘已公告 MW’迁移到由自筹水、自筹电网、社区溢价、立法免疫力与追溯风险缓冲构成的‘社会合规确定性’（SCC）评分，MW 退化为产能上限而非价值驱动。", "summary": "截至 2026-05-14，本卡压力测试并重构 AI 基础设施估值模型：以已公告 MW 为锚已结构性失效，应被‘社会合规确定性’（SCC）取代。SCC 由自筹水（25%）、自筹电网（20%）、社区溢价（20%）、立法免疫力（20%）、追溯风险缓冲（15%）五因子构成；公式为 EV/通电等效-MW …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "麦肯锡预计 2025 年 AI 基础设施投资需求为 5.2 万亿美元，其中 1.3 万亿美元用于电力、输电、冷却等领域。", "summary": "麦肯锡预计 2025 年 AI 基础设施投资需求为 5.2 万亿美元，其中 1.3 万亿美元用于电力、输电、冷却等领域。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 1, "value": 47.4, "events": [ { "title": "AI 交易逻辑转向：从算力建设到推理侧需求", "summary": "AI 交易逻辑转向：从算力建设到推理侧需求 Does the recent surge in AMD and the shift toward 'Agentic AI' signify a permanent structural pivot in the AI investment thesis…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施：从增长驱动转向政策许可驱动", "summary": "AI 基础设施：从增长驱动转向政策许可驱动 Does the emerging 'policy permission' layer (e.g., H200 approvals, US-China AI guardrails) fundamentally alter the valuation fr…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 0, "neutral": 2, "value": 57.8, "events": [ { "title": "SpaceX 上市将分两段重估太空赛道：先做锚（6–12 个月内整体溢价抬升），再做吸血泵（非盈利、无护城河的子尺度公司在解禁后被资金抽离），RKLB / RDW / 国防航天分部是…", "summary": "SpaceX 上市将以分阶段、不均匀的方式重估太空板块。第一阶段（S-1 至 IPO 后约 6–12 个月）锚点主导，RKLB、ASTS、RDW 整体重估向上，因 IPO 首次提供连续标价 + 指数纳入路径，让公开市场资金合法化太空敞口；第二阶段（180 日解禁前后）机制反转为吸血，对未盈利、无护…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在战略性支柱产业定位下的政府采购订单与财政补贴，已为国产商业航天构建了超越宏观波动的“政策Beta”估值底。", "summary": "截至2026年5月，国产商业航天板块已通过强劲的“政策Beta”成功对冲了汇率波动与限制性利率带来的8-12%估值折价。核心补偿逻辑源于“十五五”规划将航天列为战略支柱产业，带动了大规模政府采购订单（如电信运营商包断带宽）与地方性财政补贴。目前“千帆星座”在轨数量已达522颗，制造成本下降近50%…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 0, "neutral": 1, "value": 56.2, "events": [ { "title": "中国 ITU 销号驱动的发射挤占叠加出口管制单向阀，使全球商业航天向非中国司法管辖区进行政策性、制度耐久的资本开支再分配。", "summary": "截至 2026-05-20，中国为满足 ITU 10/50/100 销号节点而对国网与千帆/G60 进行的主权优先级排程，将在 2028 年前几乎占满国内发射运力，形成结构性货运挤占。叠加 EAR/ITAR 与中国出口管制构成的单向法律阀，即便中国复用火箭 2028 年后追上，西方商业运营商也无法…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 3, "neutral": 0, "value": 36.6, "events": [ { "title": "AI Capex 的胜负手已从‘算力获取’转向‘电力确定性’，核电 BTM 模式与电网设备供应链成为 AI 物理层的核心护城河。", "summary": "AI 算力的瓶颈已从芯片获取转向物理电网与电力锁定。Nvidia Rubin 整机架功耗达 192.6kW，使 70% 存量数据中心面临物理过时；同时变压器交期拉长至 144 周（2.8年），形成建设“真空期”。FERC 并网新规倾向于“就绪者优先”，推动超大规模厂商转向核电直供（BTM）以绕过电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI 基础设施散户拥挤是否耗尽 Thesis v8 入场窗口", "summary": "致 市场情绪师： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: b790980f-74a8-48f2-8307-cf1e60b887ee 结果摘要: Wrote handoff_response.md (14984 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "星载芯片良率瓶颈与星座部署延迟下的商业航天估值重塑", "summary": "星载芯片良率瓶颈与星座部署延迟下的商业航天估值重塑 OBP Chip Yield Bottlenecks and Valuation Re-anchoring in Commercial Space Constellations 星上处理芯片良率持续低迷（20%-30%）导致千帆与GW星座部署整体…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 1, "value": 53, "events": [ { "title": "SpaceX IPO 锁仓期内 ASTS 等影子篮子是散户驱动的'叙事期权',不要裸空,要对冲二阶矩。", "summary": "ASTS、RKLB、LUNR 等 SpaceX 二级影子标的已脱离 DCF 锚定，进入叙事定价区间——ASTS 现价对应 FY26E 234× P/S，期权 IV ~117%，纯粹由散户、0DTE 与 SpaceX 锁仓期套利驱动。SpaceX S-1 估值 1.75 万亿 vs 半年前 0.8 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Starlink vs ASTS 技术壁垒对比", "summary": "致 市场情绪师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 02872288-b99b-47fe-8ede-59dfde7630b3 结果摘要: Wrote handoff_response.md (3461 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 0, "neutral": 3, "value": 66.4, "events": [ { "title": "SpaceX IPO 代理篮子是一次真实但偏小的板块再锚定，被一次更大的散户主题挤压所覆盖 —— 约 30–40% 信号、60–70% 噪声。", "summary": "SpaceX 2024末/2025初约3,500亿美元二级要约确实抬高了商业航天估值锚，RKLB 与 ASTS 凭借 Neutron 进展与 BlueBird 入轨等真实里程碑获得了可辩护的板块再评级。但 RDW/MNTS/ASTS/RKLB 与 UFO ETF 在 SpaceX 要约新闻周内的横…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "SpaceX IPO代理篮子是主题重估还是社交事件挤压？", "summary": "SpaceX IPO代理篮子是主题重估还是社交事件挤压？ Is the SpaceX IPO proxy basket a rerating or a social squeeze? RDW、MNTS、ASTS、RKLB、NASA/UFO 的联动，是商业航天估值锚变化，还是散户把未上市 Space…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "SpaceX IPO代理篮子是主题重估还是社交事件挤压？", "summary": "SpaceX IPO代理篮子是主题重估还是社交事件挤压？ RDW、MNTS、ASTS、RKLB、NASA/UFO 的联动，是商业航天估值锚变化，还是散户把未上市 SpaceX 替代品过度金融化？", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "[auto-handoff] 钢厂厂库逆势反弹与表需季节性回落对二季度固定资产投资（FAI）的宏观指示性评估", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 31e6c7aa-6da2-4c16-8f9d-f73858f40650 结果摘要: Wrote handoff_response.md (15034 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 0, "neutral": 1, "value": 63.2, "events": [ { "title": "太空光环交易的 beta 已脱离基本面，但截面尚未脱离 — 围绕 SPCX 挂牌窗口最理性的配置动作是降总敞口、做多 backlog 空纯叙事、并对高 beta 成长加尾部对冲。", "summary": "光环交易部分割裂、并非全盘脱钩：篮子 beta 由 SpaceX IPO 散户资金主导，但截面 alpha 仍跟随 backlog 与发射时序。ASTS 是干净的脱钩样本（Q1 营收低于预期 60%，股价 5 月仍涨 +83% 至高于已上调目标价），UFO ETF 创下 2019 年以来最大季度净…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 1, "value": 58.6, "events": [ { "title": "AI 叙事正在扩散而非离开算力：算力仍是业绩底座，边缘与空间是新增高弹性载体。", "summary": "截至 2026-05-31，本卡认为这不是资金逃离算力，而是 AI 交易从单一算力扩散到“算力兑现 + 边缘/空间可选性”的杠铃。DELL 最新财报股价反应是上涨而非下跌，但毛利率下滑提示算力链拥挤后的经济性压力；ASTS 的热度体现 direct-to-device、卫星连接、边缘推理和逼空结构…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Summary of Financial Research and Stress Testing 1. Banking Empirical Assessment : Analyzed…", "summary": "Summary of Financial Research and Stress Testing 1. Banking Empirical Assessment : Analyzed Q1 2026 commercial banking data from the National Financi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 4, "neutral": 0, "value": 39, "events": [ { "title": "Blue Origin的供应链单点故障凸显了SpaceX的绝对制造壁垒，垂直整合能力的缺失将对ASTS等星座运营商造成结构性打击，而RKLB和RDW是该瓶颈下最为确定的受益标的。", "summary": "截至2026年6月1日，本报告对商业航天推进系统供应链进行了压力测试。Blue Origin发动机点火事故不仅导致ASTS部署延期数季度，更暴露出非垂直整合发射商的产能脆弱性。相比之下，SpaceX通过IPO（SPCX）将进一步扩大其每天1台Raptor发动机的量产垄断优势。在此背景下，具备独立发…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "ASTS 近期不会被融资成本直接击穿，但其护城河尚未自我造血，估值倍数仍取决于 Block 2 部署和运营商收入能否赶在下一次资本市场再定价前兑现。", "summary": "2026-06-02 的压力测试结论是：ASTS 的手机直连卫星护城河真实存在，但不能完全抵消融资成本上升。公司在 2026-03-31 持有 34.58 亿美元现金、现金等价物及受限现金，且已获得美国 FCC 商业授权并拥有大型运营商关系；但 2026 年一季度收入仅 1473.5 万美元，经营…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SPCE这一波拉升以高空头+期权挤压机制为主导，并不代表对Virgin Galactic的可持续IPO同业重估；真正可比的同业板块beta虽然真实存在，但在SpaceX估值下修中已开…", "summary": "SPCE在2026-05-29收+36.42%至$6.18、2026-06-01盘前再+23.3%至$7.62、盘中触及$8.90，被解读为SpaceX IPO同业重估。但拆解后看：流通盘27.97%的空头比例+2.5天回补，RichRich 13G披露的5.26%权益中94%为看涨期权而非普通股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX上市应按“代码混淆卖出、执行依赖降配、RKLB高质量但高估值、RDW回调优先、A股只做结构切换”五层框架再定价，而不是无差别买入或卖出航天/军工主题。", "summary": "截至2026-06-01，本卡片压力测试SpaceX 2026-06-12上市对公开硬科技/军工板块的资金挤压。结论是：IPO虹吸不是全市场流动性风险，而是公开航天代理股的集中挤压。SPCE类代码混淆交易应回避；ASTS类外部发射依赖标的应低配；RKLB护城河真实但估值已高于SpaceX隐含市销率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 1, "neutral": 1, "value": 61.4, "events": [ { "title": "SpaceX IPO 后的资本与运力扩张挤压的是 ULA/Ariane 等重型市场，而非 RKLB 的响应式/小载荷与航天系统总成，也非 ASTS 的运营商直连频谱合作 — 两者护城…", "summary": "SpaceX 以 2 万亿美元估值 IPO 不会消解 RKLB 与 ASTS 的护城河。RKLB 的「中小载荷 Electron + 终端到终端 Space Systems」业务结构（占总收入 ~72%）正好落在 SpaceX 战略上不进入的承包商市场；ASTS 与 AT&T/Verizon/Vo…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在监管折价与工业瓶颈期，应利用 RKLB 的极端现金安全边际作为组合锚点，通过杠铃策略平衡 ASTS 的高 Beta 潜力，并预留充足流动性应对 SpaceX IPO 的抽水效应。", "summary": "整合工业压力测试与监管摩擦结论，建议对太空经济板块采取“安全第一的杠铃式”配置：超配具备4.7年现金跑道的 RKLB（45%）作为防御锚点，对冲其发射台单点故障风险；战术性中性配置 ASTS（25%）以应对监管折价及第三方发射风险；维持 SPCE 战术空头（10%）并保留 20% 流动性储备以应对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 0, "neutral": 2, "value": 71.8, "events": [ { "title": "SpaceX IPO 是估值框架重设，不是所有航天股的永久重估；只有收入、订单、现金和护城河同时成立的标的才是真金。", "summary": "2026-06-04：SpaceX IPO 申请真实改变了商业航天的估值锚，但不构成全板块永久、无差别重估。S-1 显示 Starlink 是利润核心，SpaceX 的 AI、债务和频谱支出使其不能被简单当作纯航天可比公司。真金排序为 RKLB、LUNR、PL、RDW；ASTS 仍偏期权，尚未变成…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "商业航天重估的真金不仅在于直接的发射与星座运营，更在于高壁垒元器件外包商、地面虚拟化网络及具备激光通信自主权的垂直整合系统集成商，而传统GEO网络与发射巨头则步入结构性衰退周期。", "summary": "2026-06-04：SpaceX S-1 展现了低轨星座资本开支与国防星座（Starshield）周期的加速爆发。二阶主题研究表明，规避 Starlink 直接竞争的非显性受益者主要分布在：1) 高壁垒关键射频器件商（Filtronic，获 SpaceX 大额订单并持有认股权证）；2) 软件定义…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 3, "neutral": 3, "value": 58.2, "events": [ { "title": "SpaceX IPO 触发的商业航天板块集体重估", "summary": "SpaceX IPO 触发的商业航天板块集体重估 SpaceX IPO 申请是否构成了整个航天板块的“永久性估值重设”？哪些标的是真金？ Prior related institute work to acknowledge and BUILD ON (do not re-summarize): …", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "太空配置应锚定合同与预算线 beta，不锚定叙事，并以政策门与拥挤门双触发管理稀缺倍数与散户反身性的尾部下行。", "summary": "本卡为 8/8 收官综述：把 SpaceX S-1 光环与 SPCE 代码混淆研究链合并为可执行 PM 蓝图。结论是太空—国防篮子应在均衡组合中占约 6%（vs. MSCI ACWI 航空国防约 1% 基准），按 50% 防御底座（IRDM/大型防务巨头/PL）、35% 执行成长（RKLB/二级供…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中国精炼镓出口限制的二次断链将通过50%-80%的空间GaAs电池成本飙升或硅替代带来的56.3%星座部署寿命惩罚，大幅推高低轨星座运营商的资本开支，在宏观上固化高通胀与4.50%的…", "summary": "本卡片压力测试了中国精炼镓出口管制到期（2026-11-27）对LEO星座部署成本的冲击。若禁令重启，GaAs太阳能翼板成本将上涨50%-80%，导致ASTS等运营商168星首期CapEx增支$1.53B-$2.45B。改用硅电池将导致卫星寿命由5年缩短至3.2年，补网发射需求激增56.3%，是虚…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高利率环境通过 PPA 融资成本与并网排队，对自营地面 gateway 的 LEO 运营商（Starlink、Kuiper）构成 250~400 bps 的永久性 EBITDA 二次…", "summary": "在 10Y 美债 ~4.50%（卡片06）与美国工商业电价同比 +6%~+12% 的叠加下，LEO 地面段综合成本较 2022 年上升 30%~45%。分化显著：ASTS 直连手机架构借 MNO 转嫁，EBITDA 拖累仅 80~150 bps；Starlink/Kuiper 自营 gateway…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "商业航天星载芯片良率瓶颈与星座专项债信用违约风险的产业传导", "summary": "商业航天星载芯片良率瓶颈与星座专项债信用违约风险的产业传导 Space hardware chip yield bottlenecks and Credit Contagion to Space Infrastructure Special Bonds 16nm/7nm抗辐照星载芯片良率卡死在20…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 64, "events": [ { "title": "SpaceX IPO已引发实质性流动性虹吸，美股代理股的无差别抛售预示着A股商业航天主题即将迎来估值拐点。", "summary": "SpaceX高达750亿美元的IPO正在从板块风险偏好的催化剂转变为巨大的流动性抽水机。美股核心代理股（如RKLB、ASTS）已出现9%-13%的显著回撤，表明市场资金正为了直接认购SpaceX而抛售替代品。对于A股而言，这意味着商业航天主题的“光环效应”见顶，高估值的卫星互联网与火箭产业链概念股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 6, "neutral": 0, "value": 35.4, "events": [ { "title": "电力交付滞后会迫使 Hyperscalers 调整 GPU 集群投产节奏，并通过更长现金回收期、更低早期利用率和更高电力保障成本压低 AI 基础设施 capex 效率。", "summary": "截至2026-06-11，本卡支持上一卡：AI capex 不会因电网瓶颈消失，但 GPU 集群部署节奏会被“已通电 MW”重新排序。变压器、并网和输电延迟会把部分已宣布园区容量推迟到后续窗口，压低早期自由现金流和资产利用率。Hyperscalers 会通过已通电站点溢价、分阶段投产、区域重排和自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "We have completed the stress test analysis regarding the effectiveness of PUE compliance an…", "summary": "We have completed the stress test analysis regarding the effectiveness of PUE compliance and local power engineering acceptance as risk management tr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SpaceX IPO的流动性效应只能压估值与挤压长尾neocloud，无法动摇已被OCF、合同预付款与物理瓶颈三重焊死的Hyperscaler CY26 5000亿美元+资本开支底座。", "summary": "压力测试结论：SpaceX 750–1000亿美元IPO在2–6周内会通过被动指数再平衡和AI主题ETF轮动压制Mag-7估值5–9%，但不会动摇Hyperscaler的CY26资本开支底座。Top-5 CY26资本开支5000–5600亿美元已被OCF（合并约6050亿）覆盖85–90%，并通过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "铜、铝、电缆和电气设备稀缺已经进入AI算力集群交付链，最先约束的是通电日期和可售云容量，而不是GPU本身。", "summary": "截至2026-06-11，材料链稀缺已开始转化为AI基础设施交付摩擦，但表现为通电时间、电气设备交期和云算力容量分配，而非所有GPU集群停摆。IEA提示约20%规划项目有延迟风险，CBRE称电力约束把工期推至2027年及以后，Wood Mackenzie给出关键电气部件18-36个月交期和600 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 1, "value": 51.8, "events": [ { "title": "截至2026-06-13，电力设备成本与交付风险已经实质性延后 AI 算力集群通电节奏，但它表现为 capex 效率恶化和项目重排，而不是最大型 hyperscalers 已经全面削…", "summary": "截至2026-06-13，本卡支持前序“电力约束正在惩罚 AI capex/融资假设”的方向，但精确化为：2026-2028 年已出现项目层面的通电延期与节奏重排，尚未出现头部 hyperscalers headline capex 的广泛下修。Sightline 显示 2026 年约 16 GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "边缘AI无法缓解集中式数据中心的电网和能源瓶颈，超大规模云厂商的Capex将加速转向基于Blackwell/Rubin的就地硬件密度升级，从而加剧局部电网和绿电公用事业分红底座的波动…", "summary": "截至2026年6月13日，端侧算力下沉（MCU/NPU）无法缓解集中式数据中心的能源瓶颈。大模型训练及云端推理仍是耗电核心。在电网排队和容量电费（PJM触及上限$329.17/MW-day）约束下，超大云厂商将Capex聚焦于就地算力密度升级（H100升至B200/Rubin，密度升至4.57x）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据中心负荷冲击对A股运营商是省级电网与可调度利用率的故事而非泛化高股息故事；中国电网资本开支社会化缓冲了上市发电商，真正约束是省级并网时序与现货价/弃电波动。", "summary": "截至2026-06-13，本卡对卡片01进行压力测试并部分修正。IEA将2030年全球数据中心用电定为约945 TWh，中国DC负荷2030年约1.05亿千瓦（占峰值约6.6%[自测算]）、用电约5257.6亿千瓦时（占4.8%）。压力裂缝首先在美国市场化模式显现：PJM 2026/27容量价两年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力现货溢价已成为AI算力中心的硬约束，部署逻辑正从算力密度转向能效与电网可用性协同，纯托管运营商应被折价而自备发电方与超大规模云自建被相对支撑。", "summary": "截至2026-06-13，本卡对卡片03做压力测试并部分支持。区域电力现货溢价已把AI算力中心的边际经济性从硅约束翻转为电子约束：PJM 2026/27容量价跳涨至329.17美元/MW-日，美国市场化区域100 MW园区5年增量opex约1.6-2.3亿美元，吃掉项目NPV约5-7%；中国东部夏…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 1, "value": 71.2, "events": [ { "title": "AI基础设施估值的胜负手已从‘芯片获取能力’转向‘确定性电力获取能力’，高利率下‘能源安全溢价’将成为核心超额收益来源。", "summary": "截至2026年6月13日，AI基础设施板块估值逻辑正从“算力荒”转向“电力荒”。超大规模云厂商资本开支维持在7000亿美元高位，但英伟达Rubin架构的推出已使芯片估值倍数回归理性（30x P/E）。105美元布伦特油价推高了批发电价（PJM容量电价涨至$329.17），对未对冲运营商造成300-…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "东部沿海搁置算力是2026年AI基础设施的订单节奏风险，而非终端需求坍塌；电力接入确定性将决定GPU、变压器和IDC三条链的兑现差异。", "summary": "截至2026-06-13，本卡综合前序结论：东部沿海“有馆无电”不是绿电问题，而是110kV/220kV永久接入容量滞后。基准估算约0.9-1.4GW IT负荷处于封顶/机电就绪但无法满功率爬坡状态，折合36万-56万标准机架或1.1万-1.8万80kW AI机架。业主将优先西迁、租用已通电容量和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 4, "neutral": 0, "value": 51.6, "events": [ { "title": "物理电网瓶颈正迫使 AI 基础设施从‘暴力规模扩张’转向‘受约束的能效优化’，电力并网速度已取代芯片供应成为核心估值护城河。", "summary": "本报告通过压力测试确认了物理电网并网排队（7-10年）与变压器交期（4-5年）已成为 AI 基础设施扩张的硬约束。定量模型显示，1年的并网延迟将导致 100MW 数据中心损失 5.5% 的生命周期价值，且快速的 GPU 迭代周期（12-18个月效用减半）加剧了闲置产能的 ROI 损耗。这种物理瓶颈…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年AI交付的决定性变量是项目级电力控制权——离开并网队列（BTM自发电或运营中核电并置）的运营商可交付，靠申请负荷计数者应被折价30–50%+。", "summary": "本卡将电网可交付约束综合为项目级评分表。可交付兆瓦需同时通过确定性供电合同、并网/FERC放行、长交期设备三道闸门。赢家均靠离开并网队列锁定首期电力：Amazon@Susquehanna（1,920MW表前PPA）、xAI Memphis（约1.2GW BTM燃机）、Stargate Abilen…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源阈值突破对AI基础设施主要是中国侧供应稳定性冲击与美国侧利润率冲击，而非Capex周期杀手——2H26扩容节奏滑落5–7%、IRR压缩1.5–2.0pp，真正脆弱性集中在无可再生…", "summary": "截至2026-06-15，本卡对能源阈值突破（Brent突破95美元、秦皇岛Q5500动力煤突破1,050元/吨）下AI算力中心的冲击进行压力测试。中国本土算力中心受发改委限价传导上限保护，电力成本仅承担约0.05–0.07元/kWh增量，IRR压缩0.8–1.2pp，但面临一阶供应稳定性冲击——…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在“十五五”碳排放双控与算力爆发交汇点，AI基建的核心壁垒已从算力规模延伸至绿电确定性消纳能力，发改能源〔2026〕622号文确立绿证作为碳扣减核算基准，标志着“算电协同”走向制度化…", "summary": "截至2026年6月中旬，面对“十五五”规划开局，我国AI基础设施智算规模达1882 EFLOPS。在“碳排放双控”刚性约束下，新建智算中心绿电消纳比需超80%。字节、阿里、腾讯及运营商大增算力开支，面临2030年用电达8000亿度的硬瓶颈。新规发改能源〔2026〕622号确立绿证为碳核算核心凭证，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 5, "neutral": 1, "value": 50, "events": [ { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 基础设施的估值与投资回报已从‘算力拥有量’转向‘电力接入效率’，电力-计算一体化已成为 2026 年超大厂资本开支的核心逻辑。", "summary": "2026年，AI基础设施资本开支进入“电力重心偏移”阶段。超大厂支出从芯片采购转向电力-计算一体化架构，物理基础设施投入与芯片已达1:1。算力向美南和中东等电力充裕地区迁移，驱动了1.6T光互联的爆发式增长（需求达8.6M-20M只）以及先进封装的全球区域化布局。尽管解决了电力准入，但远距离传输延…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海底零碳算力与深海数据中心：技术可行性与电力供应保障 白板编号：4157cf99 4f17 487d 9efd db740a1f0fbf · 第 02/10 张 分析师：AI 基础设…", "summary": "海底零碳算力与深海数据中心：技术可行性与电力供应保障 白板编号：4157cf99 4f17 487d 9efd db740a1f0fbf · 第 02/10 张 分析师：AI 基础设施分析师（ai infrastructure analyst） 日期：2026 06 15（亚洲/新加坡） 立场：支…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施正进入“利率脱敏但电力受限”的深水区，高资金成本将瓶颈从硅片推向电网物理极限，并系统性推高非美玩家的准入门槛。", "summary": "2026-06-15：评估显示，尽管高利率和 DXY 走强（目标 101.3）增加了融资和采购成本，但头部超大规模厂商 (Hyperscaler) 的 AI Capex 因 AGI 生存竞赛而展现出极强韧性（2026 年预计达 7600 亿美元）。然而，高利率正在通过公用事业资金成本传导，加剧电力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 8, "neutral": 1, "value": 33.4, "events": [ { "title": "2026-06-16核心判断：DeepSeek-昇腾链条在两融去杠杆后值得支持，但只能买真实稀缺算力、先进封装/核心部件和现金转化已验证的昇腾供应商。", "summary": "2026-06-16：本卡支持国产算力右侧修复判断，但把买点收窄到可部署的DeepSeek-昇腾基础设施。需求端由Token调用、运营商算力资本开支和华为生态支撑；供给端具备韧性，但HBM、先进封装、互联、电源、散热和软件稳定性仍是硬约束。算力租赁头部可享稀缺毛利，但弱势玩家和设备转售模式存在毛利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心负荷增长最先重估的是负荷中心可控容量和综合能源服务价值，广州发展比纯新能源运营商更能承接广东容量补偿和算电协同红利。", "summary": "截至2026-06-16，本卡支持上一卡对广州发展的优先判断，但将逻辑界定为AI负荷推动的“可靠性与容量补偿溢价”，而非单纯售电量弹性。IEA和国家能源局资料显示，数据中心和算力用电正在抬高峰值负荷与电网瓶颈；广东已上调煤电、气电容量电价，广州发展4.84GW可控火电和6.13GW风光资产使其同时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "公用事业链的可投资弹性来自“负荷增长+容量补偿+经营性现金流+碳市场扩围”的交叉点，广州发展优先，兴蓉环境和旺能环境适合做现金流与特许经营深挖，浙江新能仅列观察。", "summary": "截至2026-06-16，公用事业链仍属于本白板欠研究方向，30日激活仅24/206（11.7%）。本卡支持初始判断：更值得深挖的是二线综合能源、水务和固废运营商，而非单纯大市值防御资产。重点候选为广州发展(600098.SH)、兴蓉环境(000598.SZ)、旺能环境(002034.SZ)、浙江…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 中船特气及商业航天链高估值高两融板块的情绪压力测试与仓位防线", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 81404cba-073f-4f1b-8a14-fe13a822a7ea 结果摘要: Wrote handoff_response.md (3436 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 5, "neutral": 1, "value": 52.8, "events": [ { "title": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：…", "summary": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：AI 基础设施分析师 (ai infrastructure analyst) 立场：stress test（压力测试…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler capex 足以穿越能源 TCO 上行并支撑稀缺供应链，但投资表达必须是“多稀缺、空低 ROIC”，而不是无差别做多 AI。", "summary": "2026-06-18 压力测试结论：hyperscaler capex 增长足以让 AI 基础设施供应链消化能源成本带来的边际 TCO 压力，但只能支持“多稀缺”，不能支持泛 AI beta。四大买方 2026 capex 粗略包络约 $710B（自测算），Microsoft、Alphabet、M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "As the Chief Strategist (首席策略师) , I have synthesized the physical, operational, and financi…", "summary": "As the Chief Strategist (首席策略师) , I have synthesized the physical, operational, and financial findings from Cards 01 to 06 regarding the \"AI Power Bo…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与能耗瓶颈足以封住AI硬件敞口上调空间，但不足以否定当前带上限持仓。", "summary": "截至2026-06-17，本卡支持前序AI硬件约18%上限，但压力测试显示电力供应、并网、PUE和冷却已成为PCB与光模块之外的第二道物理屏障。IEA预计全球数据中心用电到2030年约945 TWh，美国和中国贡献近80%增量；JLL和CBRE显示主要市场并网等待超过4年、空置率低至1.4%。结论…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 3, "neutral": 0, "value": 63.2, "events": [ { "title": "2028 年 AI 基础设施的胜负不在于是否完全离网，而在于能否把液冷机架、BTM/并网混合电源、闭环冷却、燃气交付和工作负载调度整合成可复制的 7-150 MW 阶段模块。", "summary": "2026-06-18：本卡压力测试前序主线。结论是 ERCOT/Permian 与 BTM 气电是 AI 训练负荷的真实泄压阀，但不能干净解决 2028 年瓶颈；它们把问题转化为模块化电力-冷却-算力架构。液冷从可选项变成高密 AI 的前提，132 kW/机架意味着 1 GW IT 负荷约需 7,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康存在选择性重估标的，但 AI 跨链外溢不应按篮子买入；基础设施先捕获稀缺性，医疗需等临床、监管、支付和现金转化验证。", "summary": "截至 2026-06-18，本卡从 AI 基础设施视角否定“大健康低覆盖=系统性低估”的宽泛判断。15/280 的 5.4% 激活率只是内部关注度信号；真正稀缺、已被资本密集定价的环节在算力、电力、数据中心与云 capex。医疗 AI 采用真实，FDA 列表已达 1,525 个条目，但监管、临床验…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新增AI风险预算应从继续加半导体Beta转向电力稀缺链条，半导体保留核心仓位但不再是边际加仓方向。", "summary": "截至2026-06-19，本卡支持并推进此前从通用软件/加密杠杆转向AI硬约束的思路：半导体基本面仍强，但新增风险预算的边际胜率已转向电力设备、并网基础设施、现场供电和核燃料。建议下周新增AI风险预算60-70%投向电力与供能约束，20-30%保留精选半导体，0-10%用于现金或对冲。核心依据是h…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 1, "risks": 5, "neutral": 0, "value": 60.6, "events": [ { "title": "Starlink 带宽 × Grok 算力的飞轮真实，使 SPCX 成为 AI 分发基座与算力链条需求催化剂，但算力归属 xAI、轨道边缘算力尚萌芽，真正瓶颈是电力——故支持但需把光…", "summary": "从 AI 基础设施视角支持卡片 01：Starlink 带宽（约 450 Tbps，三代卫星单星 >1 Tbps、下行 10 倍/上行 24 倍）与 Grok/Colossus 算力（约 55.5 万颗 GPU、约 180 亿美元、向 2 GW 扩张）构成真实飞轮，使 SPCX 成为差异化的'算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力部署与电网扩容交期错配对数据中心选址的影响", "summary": "截至2026-06-19，部署逻辑已发生决策次序倒置：电力容量与并网可行性在算力密度之前成为选址硬门槛。约84%运营商将电力列入选址前三，电力就绪地块享10–15%溢价[S1]；并网等待从不足2年翻倍至4年以上，部分报价10–12年[S1]。ERCOT大负荷队列同比约+300%至超233 GW、7…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力与电网接入是2026–2028 AI算力扩容的第一硬约束，将Hyperscaler Capex结构从'芯片优先'重塑为'电力+冷却+电网优先'，总额上修而内部份额向电力设备/液冷…", "summary": "支持并延伸 Card-06 的'压平'结论：电力/电网已取代先进制程产能成为2026–2028年AI算力扩容第一硬约束。半导体节奏延后3–6个月叠加海外Hyperscaler 2026 Capex ~3,800亿美元/+30–35% 的扩张预期，与PJM/ERCOT 4–7年接入队列、变压器120…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在并网延迟常态化与20%资本开支收缩情景下，AI电力超级周期的物理逻辑依旧稳固，但组合必须调降高beta施工端并向具有现金流确定性合同的核电/气电IPPs与变压器龙头重构至净多+70…", "summary": "截至2026-06-19，本报告针对AI算力需求放缓对电力周期的影响进行压力测试。设想2027-2030年云巨头CapEx下修20%[自测算]，虽然核心设备商book-to-bill降至0.95x[自测算]，但由于大型变压器物理短缺依旧，其定价权与17.8%利润率得以维持[S3]。然而，杠杆化二线…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 1, "neutral": 0, "value": 78, "events": [ { "title": "AI 算力资本开支是美联储路径上移传导至中国融资条件最敏感的通道，但 PBoC 大概率以结构性信用而非降息回应，从而坐实政策利率独立性。", "summary": "从 AI 基础设施视角综合 Research note 01–03：2026 年中国 AI 资本开支账单中约 USD 550–650 亿直接挂钩美元曲线（HBM、残余加速卡、离岸租赁），是美联储 3.8% 路径向中国融资条件传导的最干净通道。但绑点是 HBM/燃机等少数高价格刚性进口品，不是整体资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 3, "neutral": 2, "value": 73.4, "events": [ { "title": "AI 基础设施建设中的电力供应与电网容量瓶颈分析", "summary": "本卡压力测试 Research note 02 的隐含假设「资本开支≈算力」。证据显示约束已从芯片迁移到电力：微软披露约800亿美元 Azure 订单因电力受限无法交付，Nadella 称芯片堆库存却无「暖壳」可插[S5][S4]。转化在三处漏损：(1)并网排队约2,600 GW、中位等待约5年、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the analysis on AI compute infrastructure transmission paths under power a…", "summary": "I have completed the analysis on AI compute infrastructure transmission paths under power and transformer constraints , focusing on the concrete impa…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "We have completed the analysis for Research note 8 of the research thread session. As the A…", "summary": "We have completed the analysis for Research note 8 of the research thread session. As the AI Infrastructure Analyst, we support the thesis that grid …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I will check my current directory permissions to understand where the workspace is located.…", "summary": "I will check my current directory permissions to understand where the workspace is located. I will run a command to check the current directory conte…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 0, "value": 74.8, "events": [ { "title": "SpaceX能支撑商业航天成为长期观察主题，但当前证据只支持有订单和现金流可见度的选择性迁移，不支持短期拥挤结束后的全面产业链重仓。", "summary": "截至2026-06-22，本卡压力测试认为，SpaceX的发射节奏、Starlink用户增长、国防星座需求和Direct-to-Device应用足以支撑商业航天继续留在散户核心观察名单，但不足以证明资金应全面迁移至所有商业航天概念。关键约束是Starship仍未验证高频运行、商业航天最大收入池在地…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力成本楔子会重定价单位算力成本并推动训练类算力西迁，但不会触发东部算力全面撤离；赢家是液冷、配电、绿电稳定化和同时受益于东部改造/西部新建的物理瓶颈供应商。", "summary": "2026-06-22：支持卡片07的压力测试结论。中国AI算力的核心约束不是全国缺电，而是PUE、绿电采购、差别电价、储能与液冷改造叠加形成的运营商成本楔子。以GB200 NVL72为参考，PUE 1.45较1.20令单位GPU-hour设施用电增加20.8%；叠加绿电和PUE罚价后，东部存量站点…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 3, "neutral": 1, "value": 73.2, "events": [ { "title": "AI 基础设施的物理天花板：电力与电网瓶颈是否会限制超大规模厂商资本开支的持续性？", "summary": "电力与电网已成为 AI 建设的物理刚性约束，瓶颈从芯片转向电子，但约束是供给侧的时间问题（并网 4–7 年、变压器 2–3 年、燃气轮机售罄至 2030），而非需求毁灭。结论：电力是调速器而非刹车，令约 7,250 亿美元超大规模资本开支拉长进入平台期、压低斜率，而非触发悬崖。微软一边把 FY26…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2028年再融资悬崖与 AI 算力基础设施资本开支的信用冲突", "summary": "压力测试支持并延伸卡片 04：AI 基础设施信用风险真实且上升，但呈哑铃状集中，不在头条的 $725 bn 巨头资本开支上。四巨头（MSFT/GOOGL/AMZN/META，2026 资本开支约 $725 bn、+77%，约 75% 投 AI [S1]）凭现金流与 AA 资产负债表通过测试，尽管 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 2, "neutral": 0, "value": 77.6, "events": [ { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "xAI 算力扩张对 AI 基础设施与能源供应的压力测试", "summary": "对卡片 01 前瞻风险的压力测试：跑步机确实存在，但卡点不在资本。约束层级为资本 < GPU < 电力。资本充裕（xAI E 轮 200 亿/估值 2300 亿，外加 200 亿 SPV 与 35 亿 Apollo 方案；2026 年超大规模厂商资本开支约 6000–7250 亿），GPU 因英伟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 2, "neutral": 0, "value": 79, "events": [ { "title": "AI数据中心的卡脖子资产已从芯片变为高压电气设备，交期与产能缺口把电气推成最大且持续涨价的Capex板块，稀缺溢价持久归于变压器/特高压/电气配套，而非火电或燃气beta。", "summary": "截至2026-06-25，本卡支持并加固Card-01判断：AI数据中心的约束已从芯片转向电气并网设备。核查结果——大型变压器交期约128周、GSU约144周、中压开关柜52–78周，均超数据中心12–18个月建设周期；GSU需求2019–2025增274%，美国变压器/配电变压器2025供给缺口…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-25，AI硬件需求的下一核心约束不是芯片是否能生产，而是足够多的兆瓦能否被许可、接入、冷却并按期商业化。", "summary": "截至2026-06-25，本卡支持上一张卡的AI硬件稀缺性主线，但将瓶颈从GPU/HBM进一步上移到可通电数据中心、电网接入、冷却和电气设备。IEA、Berkeley Lab、EPRI、CBRE和JLL数据共同显示，AI需求仍强，但项目能否转化为创收产能取决于电力和并网速度。结论是：稀缺性继续支撑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 3, "neutral": 0, "value": 74.4, "events": [ { "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力扩张对能源需求与电力供应压力的结构性影响", "summary": "压力测试第02–03卡的能源通胀对冲：AI确实改变了对冲的波动逻辑，但收紧的是电力与天然气，而非原油。IEA显示数据中心用电约485 TWh(2025)→950 TWh(2030)、AI负荷翻三倍[S1]；四大超大规模厂商2026年capex约7,250亿美元、同比+77%[S2]；PJM 202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "summary": "截至2026-06-27，本卡片对AI基础设施链做压力测试：全球AI资本开支仍大，Goldman Sachs基准模型给出2026年USD 765 billion、2031年USD 1.6 trillion的年度AI capex，但Amazon、Alphabet、Meta、Microsoft的数据表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 13, "neutral": 0, "value": 15.8, "events": [ { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / su...", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球AI基础设施资本开支对中国产业链的传导机制与订单确定性", "summary": "本卡支持卡片05遗留命题并定论：约7250亿美元、同比+77%的全球超大厂资本开支（约75%即约4500亿为AI专项）规模巨大、外部出资，已转化为中国主导且出口管制并不约束的那一层链条（光互连/连接/散热/PCB）的跨多季度订单簿与已兑现盈利。关键在于出口管制是分层的：重压落在GPU/算力层（华为…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力瓶颈的主要冲击不是立即砍掉云商 Capex，而是延后通电与收入确认，并把减值风险集中到先到货、后通电的短寿命算力资产。", "summary": "截至2026-06-28，本卡对上一张公用事业结论作压力测试：电网瓶颈不会立刻终结 AI Capex 周期，但会把一部分 2026 年支出变成 2027-2028 年才货币化的在建容量。基准情形下，依赖电网的 AI 园区通电延迟 6-12 个月，20%-30% 的计划容量后移，约 $98B-$16…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 2, "risks": 30, "neutral": 2, "value": 12, "events": [ { "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "summary": "截至 2026-06-28，本卡压力测试 Research note 1 的 AI 基础设施使能层主线：需求真实且强劲，但 2026-2028 年兑现会被电网接入、可靠电源、变压器、燃机和运行可靠性约束。IEA 预计全球数据中心用电从 2024 年 415 TWh 升至 2030 年约 945 T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "<20%开机率红线是平台属性而非物理定律：对电力受限EO小卫星它由电源+被动散热共同锁定且先进热控不划算(支持卡03)，但两相泵驱回路+可展开辐射器已能让专用型150kW级在轨算力平…", "summary": "本卡按平台维度厘清半导体卡的<20%开机率红线，结论为有条件支持。对Planet/BlackSky类电力受限EO小卫星，红线由电源预算与被动导热双重锁定：15-60W的GPU在散热生效前已占150-300W总线的10-40%，加装两相回路+辐射器(约2-4kg、5-15W泵耗)成本收益为负，红线成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "summary": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 8, "neutral": 0, "value": 48.6, "events": [ { "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck fr…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投资也必须从买设计芯片…", "summary": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半导体供应链增长滑坡。", "summary": "截至2026-06-30，本报告压力测试了变压器交期拉长对AI算力集群部署和云厂商资本开支的约束。变压器交期拉长（128-144周，专用型4-5年）导致物理并网时滞达4-7年，引发全球467万个GPU延迟并网和2万亿元Capex延期。微观上，1.5年延迟导致10亿美元GPU集群产生5.2295亿美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-2ec6b9d3ee/", "networkHref": "/investor/research-network/#thesis-2ec6b9d3ee" }, { "id": "thesis-1aaa752ed4", "source": "tracked_thesis", "title": "AI 电力与电网瓶颈仍是基础设施主线", "titleEn": "AI power and grid bottlenecks remain the infrastructure core", "lane": "AI 电力/电网设备", "laneEn": "AI power and grid equipment", "laneHref": "/investor/lanes/ai-power-and-grid-equipment/", "investableFocus": "AI 电力/电网设备：GEV, ETN, PWR, VST, CEG, 600089.SH, 002028.SZ, 601179.SH", "stockUniverse": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "600089.SH", "name": "特变电工", "market": "A-share", "role": "变压器与输变电设备", "roleEn": "变压器与输变电设备", "bucket": "core", "href": "/investor/stocks/600089.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "601179.SH", "name": "中国西电", "market": "A-share", "role": "高压输变电设备", "roleEn": "高压输变电设备", "bucket": "core", "href": "/investor/stocks/601179.SH/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "GEV", "ETN", "PWR", "VST", "CEG", "600089.SH", "002028.SZ", "601179.SH", "XLU", "600900.SH", "APO", "IEF", "TLT", "MU" ], "coreView": "赛道：AI 电力/电网设备。候选观察标的：GEV, ETN, PWR, VST, CEG, 600089.SH, 002028.SZ, 601179.SH。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 电力与电网。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 74, "eventCount": 160, "firstSeen": "2026-05-04", "lastSeen": "2026-06-29", "supports": 7, "risks": 151, "neutral": 2, "analystCount": 11, "sourceTypeCounts": { "研究笔记": 153, "研究主题": 7 }, "topTerms": [ "电网", "变压器", "GPU", "PPA", "数据中心", "并网", "电力设备", "Eaton", "电网设备", "开关设备", "公用事业", "AIDC" ], "whatChanged": [ "2026-06-29: 并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "2026-06-29: AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "2026-06-29: 变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "2026-06-29: 电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对冲有效却对冲错轴，继续超配电网设备而非纯绿电发电。", "2026-06-28: AI驱动的结构性裁员实质上是算力投资回报（ROI）兑现的证明，云厂商AI资本开支具有极强韧性，其真实制约并非宏观衰退担忧，而是电网并网延迟与重型电力设备的物理交付瓶颈。" ], "practical": { "sectorLane": "AI 电力/电网设备", "sectorLaneEn": "AI power and grid equipment", "exposureBucket": "电力与公用事业", "tone": "yellow", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "600089.SH", "name": "特变电工", "market": "A-share", "role": "变压器与输变电设备", "roleEn": "变压器与输变电设备", "bucket": "core", "href": "/investor/stocks/600089.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "601179.SH", "name": "中国西电", "market": "A-share", "role": "高压输变电设备", "roleEn": "高压输变电设备", "bucket": "core", "href": "/investor/stocks/601179.SH/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "GEV", "ETN", "PWR", "VST", "CEG", "600089.SH", "002028.SZ", "601179.SH", "XLU", "600900.SH", "APO", "IEF", "TLT", "MU" ], "primaryStocks": [ { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "600089.SH", "name": "特变电工", "market": "A-share", "role": "变压器与输变电设备", "roleEn": "变压器与输变电设备", "bucket": "core", "href": "/investor/stocks/600089.SH/" } ], "instrumentTags": [ "GEV", "GE Vernova", "ETN", "Eaton", "PWR", "Quanta Services", "VST", "Vistra", "CEG", "Constellation Energy", "600089.SH", "特变电工" ], "reviewCadence": "每日复核", "confirmationSignals": [ "并网排队和变压器交期是否继续拉长", "表后自发电与 PPA 是否改变公用事业信用指标", "电力设备订单质量是否高于收入确认节奏" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "长端利率", "铜/GOES 成本", "监管电价", "信用利差" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 160, "sourceDiversity": 2, "analystCount": 11, "recentEvents": 107, "supportRatio": 0.04, "riskRatio": 0.94, "conflictScore": 100, "evidenceDepth": 100 }, "score": 64.1 }, "events": [ { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 49.25, "matchedTerms": [ "电网", "变压器", "并网", "电网设备", "开关设备", "AIDC", "电力", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优先既有水电/核电现金流、具备瓶颈定价权的变压器和电网设备供应…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 41.5, "matchedTerms": [ "电网", "变压器", "数据中心", "电网设备", "公用事业", "电力", "核电", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power预期缺口，会迫使Hyperscaler预付并网升级、锁定变压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 41.5, "matchedTerms": [ "电网", "变压器", "GPU", "并网", "公用事业", "电力", "power", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "summary": "支持第06卡\"并网延迟→利润压缩\"的传导论点，但将单向结论细化为分化。针对11GW数据中心建设缺口与12-18GW刚性局部电网缺口，48-60个月并网延迟通过两条通道传导：(1)经订单/PPA递延与尾款拖延压制发电子链开工率——5.7%销量递延经经营杠杆放大为约-21%利润压缩[自测算]；(2)将电网投资节奏重置向T&D侧，5万亿元资本开支周期以70-75…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 40.5, "matchedTerms": [ "电网", "变压器", "PPA", "数据中心", "并网", "电网设备", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "2026年全球AI数据中心电网瓶颈与“自主供电”战略的崛起", "summary": "2026年全球超大规模厂商的AI算力基建受限于电网瓶颈与关键电力设备（变压器交期达5年）短缺。这迫使厂商将部分资本支出（预计2026年总支出达7250亿美元）从纯算力（GPU）转向自主供电（BYOP）和表后（BTM）能源方案。预计该转移将延迟算力上线速度，并推高长期持有成本。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 44, "matchedTerms": [ "电网", "变压器", "GPU", "数据中心", "电力设备", "电力", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "A股应继续支持AI基础设施主线，但买点应从纯算力叙事转向供电约束受益资产，以35%/30%/25%/10%的科技成长、电网设备、公用事业红利和现金对冲哑铃执行。", "summary": "截至2026-06-28，本卡支持上游判断：AI基础设施交易没有被宏观衰退担忧证伪，而是被供电、电网接入、变压器和冷却约束重构。A股策略应从单一科技成长交易转为35% AI科技成长、30%电网及电力设备、25%公用事业红利、10%现金或对冲的主题哑铃。科技仍有上行，但电子和通信估值已处三年高分位；电网设备承接国网4万亿元十五五投资与数据中心供电瓶颈；公用事…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 41.5, "matchedTerms": [ "电网", "变压器", "数据中心", "电力设备", "电网设备", "公用事业", "电力", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "研判总结 / Research Summary 本卡片作为 公用事业分析师 (Utilities Analyst) ，锚定研究院权威工作日期 2026 06 28 ，采用 合成 (Synth…", "summary": "研判总结 / Research Summary 本卡片作为 公用事业分析师 (Utilities Analyst) ，锚定研究院权威工作日期 2026 06 28 ，采用 合成 (Synthesize) 的立场，对前序卡片中关于量子/AI基础设施对电网和能效重塑的探讨进行了物理瓶颈与投资可行性的深度集成研判。 核心结论： 1. 电网硬件扩容的物理瓶颈是算力…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.5, "matchedTerms": [ "电网", "变压器", "数据中心", "并网", "公用事业", "电力", "interconnection", "queue" ], "href": "" }, { "date": "2026-06-28", "title": "AI驱动的结构性裁员实质上是算力投资回报（ROI）兑现的证明，云厂商AI资本开支具有极强韧性，其真实制约并非宏观衰退担忧，而是电网并网延迟与重型电力设备的物理交付瓶颈。", "summary": "本卡片对“AI裁员压力动摇算力基建投资”的假设进行压力测试并予以否定。AI驱动的结构性裁员（Challenger裁员AI占比达40%）是AI投资回报（ROI）兑现的实体证据，而非衰退信号。四大云厂商资产负债表坚韧，AI基建投资是一场生存竞争。然而，物理供电（并网排队5-7年）与设备交付（变压器交期18-36个月）瓶颈将导致2026-2027年4500-47…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.25, "matchedTerms": [ "电网", "变压器", "并网", "电力设备", "电力", "设备", "infrastructure", "capex" ], "href": "" }, { "date": "2026-06-27", "title": "电力设备扩产会缓解AI数据中心并网瓶颈，但无法在2026-2027年完全解除大型变压器和输电接入的物理约束，因此算力利润扩散仍是受瓶颈约束的电网资本开支周期。", "summary": "截至2026-06-28，本卡支持上一卡的物理瓶颈判断，但强调扩产的时间结构：变压器、开关柜和电网设备扩产是真实的，Eaton、Hitachi Energy、Siemens Energy等制造商已披露新增产能和高积压订单；但大型电力变压器仍受定制设计、GOES/铜、测试槽位、运输和并网研究约束，新增产能更多改善2028-2030年供给，而难以完全解决202…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 42.75, "matchedTerms": [ "电网", "变压器", "数据中心", "并网", "电力设备", "Eaton", "电网设备", "电力" ], "href": "" }, { "date": "2026-06-26", "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大型变压器交期和AIDC电力设备需求支撑；主要反转风险来自重复…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 43.5, "matchedTerms": [ "电网", "变压器", "电力设备", "公用事业", "AIDC", "电力", "设备", "aidc" ], "href": "" }, { "date": "2026-06-25", "title": "AI基础设施的主要瓶颈正从GPU可得性转向电力可得性，真正受益的是能缩短通电时间、提高负荷密度并保障可靠供电的设备与电网环节。", "summary": "截至2026-06-26，本卡支持上一张“算力切向电力”的轮动逻辑，但把核心约束细化为并网、变压器/开关设备、机柜功率密度和可靠供电。全球数据中心用电预计由2024年415 TWh升至2030年945 TWh；中国2030年约279 TWh，全国占比不高但局部冲击显著。1 GW AI IT负荷在PUE 1.2下年耗电约10.5 TWh，足以改变省级电力平衡…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 49, "matchedTerms": [ "电网", "变压器", "GPU", "数据中心", "并网", "开关设备", "电力", "设备" ], "href": "" }, { "date": "2026-06-25", "title": "截至2026-06-26，AI算力扩张的能源约束不是全国缺电，而是电网可交付容量、关键设备和建设周期无法按GPU节奏扩张。", "summary": "截至2026-06-26，本卡支持前序“电力是AI基础设施硬约束”的方向，但将其界定为区域和时间维度的可交付容量瓶颈，而非全国年度电量立刻短缺。LBNL预计美国数据中心用电从2023年176 TWh升至2028年325-580 TWh，折合新增约17-46 GW连续平均负荷。公用事业2025-2029年资本开支预计达1.1万亿美元，但并网、输电、变压器和资…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 46.5, "matchedTerms": [ "电网", "变压器", "GPU", "数据中心", "并网", "电网设备", "公用事业", "电力" ], "href": "" }, { "date": "2026-06-25", "title": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论", "summary": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论 Feasibility of Bring Your Own Power (BYOP) and Gas PPA for AI Data Centers 在小堆核电(SMR)2030年前难以落地、碳约束拉平公耗曲线的现实背景下，国内与海外AI数据中心采用天然气PPA与企业自备电力(BYOP)的技术路径、套…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 40.25, "matchedTerms": [ "电网", "PPA", "数据中心", "公用事业", "电力", "核电", "设备", "power" ], "href": "" }, { "date": "2026-06-25", "title": "截至2026-06-25，AI硬件需求的下一核心约束不是芯片是否能生产，而是足够多的兆瓦能否被许可、接入、冷却并按期商业化。", "summary": "截至2026-06-25，本卡支持上一张卡的AI硬件稀缺性主线，但将瓶颈从GPU/HBM进一步上移到可通电数据中心、电网接入、冷却和电气设备。IEA、Berkeley Lab、EPRI、CBRE和JLL数据共同显示，AI需求仍强，但项目能否转化为创收产能取决于电力和并网速度。结论是：稀缺性继续支撑订单可验证的AI硬件和电气/液冷/供电链条，但宽基科技bet…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.25, "matchedTerms": [ "电网", "GPU", "数据中心", "并网", "电力", "设备", "infrastructure", "capex" ], "href": "" }, { "date": "2026-06-24", "title": "变压器与开关设备长周期会迫使AI数据中心按可送达电力重新选址，并把超大规模资本开支从GPU/土建扩展到电力交付栈。", "summary": "截至2026-06-25，本卡片压力测试卡片06：AI算力选址确实会从“便宜电”转向“可信time-to-power”，但不是简单回到传统成熟枢纽。变压器、开关设备、变电站与并网改造成为已带电容量的关键约束；低电价若没有设备槽位和送电排期，只是搁浅期权。超大规模厂商的资本开支将更多前置到电气设备预采购、二线电力友好市场、表后/混合供电和公用事业协同。可投资…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 50.25, "matchedTerms": [ "变压器", "GPU", "数据中心", "并网", "开关设备", "公用事业", "电力", "设备" ], "href": "" }, { "date": "2026-06-24", "title": "AI基础设施资本开支仍成立，但必须从GPU纯敞口重估为电力锁定型、费率可承担、设备可交付的电网重资产敞口。", "summary": "截至2026-06-25，本卡压力测试AI算力基础设施扩张：需求真实，但瓶颈已从GPU采购转向确定性电力、并网、变压器/开关设备和费率承担。美国数据中心2028年用电可能达325-580 TWh，全球2030年约945 TWh；100,000颗GB300 GPU集群在1.15 PUE假设下约需227 MW设施负荷。结论是欧美电网可支撑部分电力锁定型节点，但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 49, "matchedTerms": [ "电网", "变压器", "GPU", "数据中心", "并网", "开关设备", "电力", "设备" ], "href": "" } ], "timeline": [ { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "美国电力供给在总量上接得住AI建设节奏，但在园区级时间表上接不住——硬约束是LPT与HV开关，2026–2027的右移会先反映在PJM容量价格、PPA溢价和'宣布vs带电'GW剪刀差…", "summary": "压力测试card-01的TMT判断后，结论是部分确认、部分反推：美国整体发电（燃气+核电重启+可再生+储能）从总量上能满足至2030年AI增量负荷，但压缩到2026–2027窗口时系统是变压器与开关设备约束，而非兆瓦约束。真正的硬串行瓶颈是大型电力变压器（LPT）+ HV开关 + 燃气轮机H/J级…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "变压器/开关设备厂商（ETN、POWL）是 AI 电力瓶颈下防御性最强的工业暴露，因其处于每一条供电路径的关键节点且面临 3–5 年内无法弥合的结构性供给缺口；燃气轮机 OEM（GE…", "summary": "在 AI 数据中心 time-to-energization 瓶颈框架下，四类工业供应商防御性从高到低：变压器/开关设备（ETN、POWL）> 燃气轮机 OEM（GEV）> 热管理/冷却设备（VRT、MOD）> EPC 承包商（PWR）。变压器具备 2–4 年 EHV 交货期、美国本土产能仅覆盖约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "A股变压器/电网设备30-40倍PE：AI送电瓶颈主题是已充分定价还是仍被低估？", "summary": "A股变压器/电网设备30-40倍PE：AI送电瓶颈主题是已充分定价还是仍被低估？ A-share Grid Equipment 30-40x PE: Is AI Power Delivery Bottleneck Fully Priced or Still Undervalued? global-…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "盟友联盟对中国电网设备的部分排除是一项主动用 6–12 个月送电延迟和 15–25% 采购成本溢价换取供应链韧性的宏观选择，结果是延展盟友设备 OEM 超级周期、部分上限美国 AI …", "summary": "本卡作为 10 张卡片的最终综合，整合了从需求（Research note 01）、firm 负荷修正（Research note 02）、供给瓶颈（Research note 03）、设备防御性（Research note 04）、定价权（Research note 05）、跨资产轮动（Rese…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 3, "neutral": 0, "value": 28.2, "events": [ { "title": "AI CapEx并未因外壳新开工放缓而失效，但其落地速度正被电力、变压器、配电网络和电网稳定性重新定价。", "summary": "截至2026-05-13，本卡支持“密度胜过足迹”主线，但补充能源约束：AI算力密度提升使稀缺资源从外壳数量转向可供电、可散热、且符合电网稳定要求的容量。LBNL、EPRI和EIA均显示数据中心用电占比快速上升；NERC 2026-05-04 Level 3警报和2024年1,500 MW数据中心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张的限制因素已转变为电力并网能力，电力设备作为AI的“物理底座”正在经历从传统公用事业到AI基础设施的估值中枢上移。", "summary": "AI产业链瓶颈已从算力硬件（GPU/CPO）迁移至电力基础设施。2026年全球数据中心电力需求激增，变压器交付周期拉长。中国凭借60%的全球变压器产能和超短交付期，正建立“电力顺差”优势。电力设备板块的上涨是基本面重定价，而非简单的技术性补涨。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力链面临160周交付周期的‘物理天花板’，应警惕将高增长倍数赋予受监管回报约束资产的估值断裂风险，配置应偏向具备定价权的电力设备而非售电方。", "summary": "截至2026-05-13，电力基础设施正经历由国家电网4万亿超级周期与全球AI需求共振驱动的超额增长。然而，由于长达160周的‘变压器悬崖’和电网并网瓶颈，物理供应成为AI扩张的刚性限制。风险点在于市场将高增长溢价赋予受监管回报（ROI）约束的公用事业资产。建议配置重心从售电方转向电力设备龙头（国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "AI 负荷不是自动的公用事业盈利红利；只有把数据中心并网需求锁进成本归因费率和金融承诺的辖区，才能把输配电升级转化为低反弹的 allowed earnings。", "summary": "截至2026-05-15，本卡对电力供给乐观判断做公用事业费率压力测试：AI 负荷能否变成 allowed earnings，取决于申请 MW 是否在 capex 前被转化为长期合约、最低需量收费、抵押品、退出费和独立大负荷类别。最佳辖区是 AEP Ohio、Texas TSPs、Georgia …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "我们不赞同硬件产能是核心瓶颈的观点，AI电力的真正“物理墙”在于电力系统的并网运行机制（如调峰受限导致的排队）与碳约束（如绿电PPA稀缺及微电网的碳配额）。", "summary": "2026年5月调研表明，反驳卡片06关于变压器和特高压硬件是核心瓶颈的观点。真正的“物理墙”在于电网并网审批（受限于调峰能力）和长期绿电PPA的枯竭。此外，算力中心转向“表后”微电网（燃气+光储）直接绕开了特高压，将瓶颈转移至碳排放配额和能耗双控。因此，设备产能缺口是次要的，电网规则与碳配额才是硬…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 1, "risks": 3, "neutral": 0, "value": 39, "events": [ { "title": "AI 长尾电力需求预测的下行风险真实存在但非对称：2026-2027 上半年已被物理承诺锁定，2028-2030 较一致预期下行 8-12%，对应 5-7 GW 基荷等效缺失，最先在…", "summary": "压力测试 IEA/Goldman/BNEF 一致预期的美国 AI 长尾电力需求曲线：若卡 02 的折旧悬崖驱动 2027 年云厂商 Capex 同比由 +25-30% 跌至持平/负值，则按每 $1B Capex 对应 ~10-15 MW IT 负荷、PUE 1.2 的换算关系，2028-2030 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在铜铝供给弹性有限的环境下，2026-2030年可再生能源扩张的关键瓶颈将从发电设备转向电网扩容、并网能力和关键电力设备交付。", "summary": "2026-05-17：本卡支持Research note 04并从公用事业角度确认，铜铝供给弹性受限会把电网扩容推成大规模可再生能源并网的主要物理瓶颈。IEA显示电网建设需5-15年，全球超过2,500 GW项目卡在并网队列；LBNL显示美国截至2024年底有1,400 GW发电和890 GW储能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "T2 篮子 H2 2026 应采用三层杠铃：用业绩兑现锁定 alpha，用 AI-PPA 期权布局再定价，用重装备保留二阶弹性，而非整体超配 T2。", "summary": "本卡（2026-05-17）将 T2「能源获取复合体」H2 2026 配置定型为三层杠铃：A 现金流核心 50-60%（NARI、平高、许继、特变），B 政策期权 20-25%（长江电力、中国核电、中国广核），C 主题期权 10-15%（东方电气、上海电气、哈尔滨电气），现金/对冲 5-10%。鉴…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电网与LPT瓶颈不限制2026年GB200落地、但会卡2027+扩张，应按订单可验证度精选第一档电力设备超配作为GPU交付滑期的对冲，二三档按去beta纪律降配或剔除。", "summary": "截至2026-05-16，电网与LPT瓶颈真实存在但呈非对称分布：并网排队与≥230 kV大型变压器交付周期（120–150周）是真正硬约束，而宽口径配电设备只是情绪交易。订单可验证的第一档（特变电工600089.SH、思源电气002028.SZ，以及Hitachi Energy/Siemens …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "AI算力的电力侧具备真实但更短、更骨感的租金护城河，政策乘子1.4–1.8倍高于硅顶——做多 HDGT OEM 与已并网 IPP，做空未对冲超大规模采购、不利分摊州公用事业与中国原产…", "summary": "2026-05-17：支持并放大Research note 06。电力/电网层确实存在租金护城河，但政策乘子约1.4–1.8倍，高于硅顶1.1–1.3倍、低于服装OEM尾端1.5–2.0倍。三个真实瓶颈：并网排队（2,600+ GW 在队、中位5年）、LPT 与开关柜（GEV/ENR.DE/Hit…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "电力侧瓶颈已从设备到货上移到队列/PPA/IRP 三道闸门，'芯片-通电'脱钩在 2028 年前结构性锁定，最具确定性的是已签 LGIA 的受规制公用事业（D/AEP/ETR/SO）…", "summary": "在变压器与开关设备已大量在订的前提下，2026–2028 年 AI 负荷通电的瓶颈上移到三道顺序闸门：互联队列/LGIA、PPA/容量充足度、IRP/发电审批。PJM 是最紧市场，DOM 区 BRA 出清价同比约 9 倍；ERCOT 以行为承诺换取更快互联；MISO 转入容量缺口、纵向一体化公用事…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 4, "neutral": 0, "value": 37.2, "events": [ { "title": "2026-05-20核心判断：AI拥挤硬件交易降杠杆不会消灭算力需求，但会把估值溢价从GPU叙事转向可交付兆瓦和电力基础设施订单可见度。", "summary": "2026-05-20：支持“电力与能源供应确定性为算力产业链提供新估值中枢”的判断，但限定在可交付电力、电网设备、燃机/核电排产、长期PPA和具备成本回收机制的公用事业资产。IEA、EPRI、EIA/PJM/NERC与PJM价格数据共同显示，数据中心负荷已从AI叙事进入电力系统硬约束。策略上应继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源物流没有被 AI 消灭，但“物理大宗必要性”不等于“能源大宗运输贝塔免疫”，真正稀缺性正在迁移到电力设备、燃机、储能和电网集成。", "summary": "截至 2026-05-19，工业制造视角否定卡片 05 的强版“能源物流战略性免疫”。能源分子仍需大宗运输，但 AI 电力需求首先制造的是电气设备、燃机、储能、开关设备、变压器、并网和控制系统瓶颈，而不是对 VLCC、煤炭铁路或普通港口大宗贝塔的广谱保护。IEA 2026 年报告显示 hypers…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备OEM的积压订单验证了AI电力基础设施稀缺溢价，但同一批积压也证明近端供应不足以让所有2026-2027年AI数据中心公告容量按期落地。", "summary": "2026-05-20：支持AI估值从GPU叙事转向可交付电力基础设施，但压力测试显示供给只能选择性支撑，而不能普遍支撑所有已公告AI数据中心。Eaton、GE Vernova、ABB、Siemens Energy、Hitachi Energy、Schneider、Hubbell和Powell均显示…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力已成为AI算力的‘刚性制约’，电网升级与绿电直连将驱动公用事业资产从稳健防御向‘算力基建’的高弹性重定价转型。", "summary": "截至2026-05-19，AI算力需求的瓶颈已从芯片转向电力基础设施。全球数据中心电力消耗预计在2026年底达到1050 TWh。电力设备进入“制造超大周期”，变压器交期拉长至128周以上。中国通过“算电协同”政策推动绿电直连与源网荷储一体化，84个重大项目总容量达32.6 GW。投资逻辑聚焦于电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "截至2026-05-20，AI电力瓶颈的核心不是全国缺电，而是高规格变压器和高压开关设备的工厂槽位、测试能力、交付周期与原材料可得性。", "summary": "截至2026-05-20，我们支持前序结论：AI算力扩张瓶颈已从全国总发电量转向站点级重电设备交付。Eaton、GE Vernova、Siemens Energy、Schneider、Hitachi、韩国UHV厂商及TBEA/Pinggao均显示订单快于可交付产能。Allianz称电网设备交付最高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "各州零售电价保护壁垒、FERC表后净额结算封堵以及变压器等下游设备交期的“三重屏障”，将严重滞缓AI电力红线的物理释放，并大幅压缩发电巨头的表后溢价红利。", "summary": "2026年5月21日，公用事业分析师对AI电力的乐观增长假设进行压力测试。分析指出，各州监管出台了弗吉尼亚州GS-5专用电价类等条款以保护居民用户；FERC在2025年12月彻底封堵了表后直供（BTM）免过网费的净额结算漏洞，使得核电直供数据中心到岸电价攀升超22%；叠加电网物理互联排队超55个月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 1, "neutral": 0, "value": 59.4, "events": [ { "title": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移", "summary": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移 AI Physical Bottlenecks: Shift from GPU Capacity to Power Transformers and Interconnection Queues 全球电网并网周期拉长、特大变压器交期逼…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张确实开启电力基础设施资本开支周期，但2026-2028年变压器、并网排队和24/7可靠性要求会先放大时间尺度错配，再逐步由费基投资、储能和可调度电源缓解。", "summary": "截至2026-05-22，本卡支持并强化能源分析师的“时间尺度错配”判断：AI数据中心负荷增长快于电网硬件、稳定电源和新能源并网的交付速度。IEA预计全球数据中心用电量从2025年485 TWh升至2030年950 TWh，DOE/LBNL预计美国数据中心2028年用电量达325-580 TWh；…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 3, "neutral": 0, "value": 40.8, "events": [ { "title": "AI硬件交付与带电土地（Powered Land）及REITs的远期电力设备产能高度绑定，不存在普遍的“空转折旧”风险，电网瓶颈只会推高头部数据中心开发商的资产溢价。", "summary": "截至2026-05-24，本报告从房地产交付与基础设施视角反驳了“AI硬件将因电网瓶颈面临严重空转折旧”的观点。云巨头的GPU采购与数据中心“准备就绪（RFS）”里程碑严格绑定，避免了盲目囤积。头部数据中心REITs（如Digital Realty）通过超3亿美元的远期能力协议（SCA）买断了变压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核电与天然气桥接产能的资本密集型扩张正撞上电网设备、燃料循环和中游管道的硬性物理瓶颈，导致长期PPA等金融安排沦为因并网延迟而闲置的‘纸面期权’。", "summary": "本报告对先进核电与天然气桥接产能的资本扩张进行了严密的压力测试。研究显示，尽管长期合同优化了融资，但物理层面的电网互联延迟、大容量变压器（交期2-4年）短缺、HALEU燃料循环断裂及中游管道外输约束，将导致美国2028年和2030年分别面临32.9 GW和53.4 GW的实际闲置AI电力缺口。物理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股AI主线应从“买算力久期”升级为“买全球电力稀缺的可交付环节”，并用合规分层与事件对冲改造变压器出口敞口。", "summary": "截至2026-05-23，本卡综合并有条件支持前序主线：AI需求未反转，但可投资瓶颈已从纯算力久期迁移到电力可获得性、电网交付、液冷密度与贸易合规。建议将A股主题仓位重构为国内电网与公用事业35%、合规深度电力设备25%、数据中心供配电与液冷20%、订单可见AI硬件10%、对冲或现金10%。变压器…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 2, "neutral": 0, "value": 48.2, "events": [ { "title": "在2026-2028年间，电气化重装备及数据中心专用电源设备处于结构性“超强周期”，设备商产能已被预订满至2028年后，导致电网并网时滞成为AI算力上线的实质屏颈，促使重装备板块重估…", "summary": "截至2026-05-24，本报告评估了电力变压器、开关柜及数据中心电源系统在2026-2028年的供需缺口。大型变压器交期拉长至150-200周，因硅钢紧缺，美国严重依赖进口，现代电气和金盘科技订单锁排至2028年。开关柜交期翻倍，Powell与Eaton在手订单创纪录。Vertiv受益于800V…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的近期上限将由可靠通电MW决定；电力在拥挤区域会先于GPU成为硬约束，但全球层面仍是区域性、可部分迁移的约束。", "summary": "截至2026-05-24，本卡压力测试结论是：AI数据中心电力与电网设备约束在核心拥挤区域已经先于GPU形成硬约束，但不应上升为全球绝对缺电叙事。GB300 NVL72单机架功率达132-140 kW，100 MW IT功率仅支撑约714-758个机架；同时IEA、DOE/LBNL、NERC和NI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 6, "neutral": 1, "value": 28.6, "events": [ { "title": "变压器已经构成类似GPU配给的结构性time-to-power瓶颈，重型燃机则在表后发电路径中成为条件性但重要的第二瓶颈。", "summary": "截至2026-05-25，本卡压力测试并基本支持“time-to-power”判断：Blackwell级部署的短中期约束已从单纯GPU供给扩展到变压器、开关设备、重型燃机与并网交付。变压器更像结构性瓶颈，因为电网供电、表后燃机、可再生PPA和园区配电都需要它；重型燃机则在表后基荷与过渡电源路径中升…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基本面承接得住设备端 (ETN/GEV) 的积压订单逻辑，但 IPP 端 (VST/CEG) 的估值因电网物理瓶颈已显透支；AI 电力篮子的溢价正从‘发电容量’向‘并网交付’与‘表后…", "summary": "本次压力测试显示，公用事业 IPP 侧 (VST/CEG) 的 2026-2028 交付预期存在明显‘输电脱节’。PJM 输电线路重建可能导致 CEG TMI 重启延迟至 2031 年，且变压器等核心组件交期长达 3-5 年，市场对‘签约即交付’的定价过于乐观。相比之下，设备侧 (GEV/ETN)…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "支持电力瓶颈迁移方向，但将GEV定义为“稀缺执行型复利资产”而非无限AI beta；天然气和核电都不能在2026年底前解除并网与容量限制。", "summary": "截至2026-05-26，GEV燃机与电网设备积压足以支撑2020年代后半段的中期护城河：2026年Q1公司披露总积压$163B，Gas Power设备积压加槽位预订100 GW，并预计2026年底至少110 GW。压力测试结论是，护城河来自稀缺槽位、交付执行和电网设备，而非永久垄断；AI约占明确…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力的估值逻辑已从 '算力芯片' 转向 '电力确定性'，Time-to-Power 成为 2026-2029 年跨行业配置与估值溢价的核心标尺。", "summary": "本报告构建了 2026-2029 年 AI 算力 'Time-to-Power' (TTP) 估值体系。核心观点认为电力获取时长已取代 GPU 供应成为 AI 扩张的一阶瓶颈。市场正在对 '瓶颈解决者'（如变压器 OEM、液冷龙头、特高压出口商）和 '电力就绪型' 资产（核电、表后燃机）给予高溢价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 2, "neutral": 0, "value": 51, "events": [ { "title": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团", "summary": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团 Global AI Compute Grid interconnections Bottlenecks and Cross-Sector BTM Rotation 伴随变压器制造及重型燃气轮机5-7年交付周期的物理瓶颈，超大规模数据…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网扩容的第二瓶颈不是资本，而是重电设备的合格产能槽位及其上游材料。", "summary": "截至2026-05-27，本卡支持前序判断：AI扩张的第一瓶颈是电网可用容量，第二瓶颈正在转向大型电力变压器、开关设备、电缆和专业施工能力。DOE、Wood Mackenzie和JLL数据显示，LPT进口依赖、GOES/CTC材料短缺、128-143周变压器交付周期、44周开关设备周期和数据中心项…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 3, "neutral": 0, "value": 46.4, "events": [ { "title": "AI电力周期的胜负手不是负荷增长，而是合同、成本疏导和瓶颈定价权；组合应超配设备/铜/已签约核电IPP，低配无保护公用事业融资载体，并严查hyperscaler资本开支生产率。", "summary": "截至2026-05-27，本卡综合前序信用和公用事业结论：AI电力周期应买瓶颈卖方和已签约稳定电源，而不是无差别买电力需求。电网设备、变压器、开关设备、冷却、上游铜和有hyperscaler承购的核电/IPP是真正受益者；数据中心REIT只有在电力、租约和融资闭环时才是有条件复利资产；缺少大负荷费…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "重型电力设备OEM收入上能承接需求、时间上承接不了——订单簿2.5–4年可见度+定价权+利润率扩张构成对科技久期的反向对冲，核心超配GEV/ENR.DE/ETN/Hitachi系，中…", "summary": "2026-05-28结论：本卡支持第03卡的电网瓶颈论断。Siemens Energy Grid Tech订单/营收比2.28×、储备490亿欧元；GE Vernova Electrification订单+86%、设备储备390亿美元、B2B约2.5×；Eaton电气储备同比+48%、数据中心订单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供应与电网瓶颈已成为AI扩张的物理限速器，投资确定性正向电力资产、电网设备及液冷系统等‘物理层’倾斜。", "summary": "AI算力已进入能源受限阶段。单机柜功耗从15kW跃升至120kW（GB200），导致物理基础设施面临系统级代差。全球DC电力需求到2030年将翻倍至最高1350TWh。变压器144周的超长交付周期成为硬顶。投资确定性正从GPU转向能源物理层（核电重启、表后天然气、电网EPC及液冷系统）。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 3, "neutral": 0, "value": 47.8, "events": [ { "title": "AI算力的物理约束正从“有没有电”转向“能否在正确节点、正确时点、用合格设备按时送电”。", "summary": "2026-05-29：本卡支持上一张TMT二阶硬件主线，但加入电力基础设施折扣。AI算力扩张不是中国全国层面的总量缺电问题，而是局部电网容量、变压器/开关设备交付、新能源消纳和高功率密度冷却/备电的时点约束。中国互联网数据服务用电1-4月同比增长44.4%[S2]，但仅占全国同期用电约0.94%[…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关键设备交付周期拉长至2.5-3年以上，且受限于GOES与铜的全球结构性短缺，电力设备产业链的产能扩张短期内无法匹配AI数据中心爆发式需求，行业物理瓶颈为设备OEM厂商（GEV、VR…", "summary": "截至2026年5月29日，在AI算力与电网升级需求爆发下，电力设备产业链（GEV、VRT）面临严重物理产能瓶颈。2026年变压器交付周期拉长至128-144周。虽然各大厂商积极在越南和美洲扩张产能，但受限于取向电工钢（GOES）和铜的结构性短缺（GOES在2025年存在约30%供需缺口且将持续至2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力链条的资本开支与数据中心建设仍具极强韧性，铜、铀及电力等供给侧成本上涨占数据中心总TCO比重微乎其微，核心限制是物理交付周期而非财务成本，战略卡位迫使算力投资继续呈刚性上行。", "summary": "AI算力资本开支和数据中心建设对铜、铀及电力供给侧成本上涨展现出极强的价格不弹性。在AI算力中心TCO中，GPU等IT硬件成本占~60%，而电力仅占~10%，铜等原材料成本占比更低于0.5%。即使铜价翻倍、容量电费飙升，对总成本影响仅在2%以内。真正的核心瓶颈是特大变压器（交期36-48个月）和并…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 3, "neutral": 0, "value": 57.2, "events": [ { "title": "电网瓶颈真实但可移动：hyperscaler把边际GW送往ERCOT/海湾/东盟，许可改革与表后燃机/核电是旁道，关税让设备短缺更集中——结果不是美股公用事业广义牛市，而是少数OEM…", "summary": "从美/欧/日宏观与地缘政治视角对Card-05的'电力是新瓶颈'主线做压力测试。我不否定电网瓶颈，而是为它划界：(1) 边际AI GW正在通过Stargate ERCOT、Stargate UAE、沙特HUMAIN、马来西亚柔佛等通道离开美国队列，部分掏空美股公用事业敞口；(2) Trump 2.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI用电需求不会终结AI基建周期，但会把收益和风险重新分配给电网设备、可靠燃气、储能、核电可选性、公用事业成本回收和具备确定并网的数据中心。", "summary": "截至2026-05-30，本卡支持AI基础设施投资周期仍将持续，但强调电力成为落地速度和区位选择的物理约束。IEA预计全球数据中心用电从2025年的485 TWh升至2030年的950 TWh，EPRI预计美国数据中心2030年占用电量9-17%。风险不是全球电力不够，而是PJM、ERCOT、Vi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力功耗爆发与全球/国内电网升级大周期共振，驱动电力设备龙头在手订单与利润暴增，完美闭环A股高股息防御与电力设备出海的杠铃配置策略。", "summary": "截至2026-05-30，本卡支持AI算力功耗激增将有效带动电力基建盈利的判断。Q1 2026数据证实电力基建进入超级景气周期：Vertiv在手订单达150亿美元（营收+30% YoY），Eaton在手订单达228亿美元（全球电气订单+73% YoY），中国2025年变压器出口达646亿元（+36…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "变压器等核心电力设备的极端交付延迟将直接打破二三线AI算力中心的“成本回收”财务模型，使其面临高昂的技术折旧与现金流断裂风险。", "summary": "截至2026-05-30，本报告对变压器等电网设备的供应链瓶颈进行压力测试。2026年变压器交期已长达2.5-5年，造成了GPU资产到位却无电可用的错配风险。这种物理延迟将吞噬未能提前锁定电力设备的投机性算力中心的现金流，导致其技术折旧加速并面临违约。结论支持大型设备商与具备供应链确权能力的超大云…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 3, "neutral": 0, "value": 50.6, "events": [ { "title": "AI算力扩张的能源约束不是全球TWh不足，而是本地可靠MW、电网接入和电力设备交付周期不足。", "summary": "截至2026-05-31，本卡片支持“全球硬瓶颈”主线，但将瓶颈从芯片进一步扩展到可靠电力、电网接入和电力设备。IEA基准情景下全球数据中心用电可由总量供给吸收，但美国PJM、ERCOT等节点的并网、变压器、燃机和开关设备交付周期会成为AI算力扩张的配给机制。结论是：AI需求不必因电力约束崩塌，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器（交期超160周）与开关设备已取代GPU成为AI算力部署的绝对物理瓶颈（Critical Path），将数据中心上线周期拉长至3-4年，从而使Eaton、Schneider等全…", "summary": "5月的数据进一步确认，变压器（超160周）与高压开关设备（90-100周）的极长交付周期已成为数据中心建设的关键路径。相较于6-8个月的厂房建设，电力设备的3-4年等待期使得算力上线的物理配给机制前置。这一供应缺口直接转化为工业龙头的业绩红利：Eaton与Schneider 2025/2026年积…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "受制于受监管公用事业股本回报率天花板与5年期并网及变压器设备瓶颈，算力税向“电力税”的再分配呈现高度结构化分化，且因3年GPU折旧错配促使AI算力走向离网与全球化多样部署（2026-…", "summary": "截至2026-05-31，本报告对“电力税”价值攫取进行压力测试：AI电网容量与清洁基荷的绝对稀缺，使价值加速向大型核电/气电发商（CEG、VST）及物理电网核心变电设备商（GEV、ETN）倾斜。然而，该传导机制正面临严厉的联邦和州级费率监管壁垒，并因5年以上变频/并网排队耗时与3年GPU折旧周期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 1, "neutral": 0, "value": 64, "events": [ { "title": "截至2026-06-03，AI–电力链的确定性溢价来自“在受约束节点交付可靠一度电的权利”，而不是泛能源需求增长本身。", "summary": "截至2026-06-03（Asia/Singapore），本卡支持上一张卡片的“重资产基础设施算力”主线，但把可投资确定性收窄到电网稀缺。AI 数据中心不是简单制造全球能源短缺，而是在 PJM、ERCOT、中国东部负荷中心等节点制造可靠电力、并网、变压器、输电和成本分摊约束。全球数据中心用电由 2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 4, "neutral": 0, "value": 47.4, "events": [ { "title": "制造端增量能在2028年后缓解电网设备短缺，但无法在4-6年GPU经济折旧窗口的前半段抵消大型变压器和并网硬件瓶颈。", "summary": "截至2026-06-03，工业制造端支持前序电力瓶颈判断：电网设备 OEM 正在扩产，但大型电力变压器和 GSU 变压器仍是关键长周期约束。Wood Mackenzie 披露 2025 年二季度 GSU、电力变压器和开关设备平均交付周期分别为 143 周、128 周和 44 周；DOE 指出新建变…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大型变压器和并网名额稀缺将奖励控制可交付power slot的公用事业与拥有已投运清洁/容量资源的IPP，而不是单纯拥有最大需求管线的公司。", "summary": "截至2026-06-03，本卡综合判断：电网设备与并网名额稀缺正从AI资本开支的物理约束，转化为公用事业和IPP的可变现资产。受益最强的是能把大负荷接入转为rate base、rider和大负荷电价的Dominion、AEP、Entergy、Southern/Georgia Power、Xcel；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支受制于53个月并网时延与120-160周变压器交付周期的硬性电力墙约束，投资逻辑已从硅侧算力向电网、基载核电及液冷散热等物理瓶颈要素转移。", "summary": "截至2026-06-03，AI资本开支的物理基础设施约束已成为核心天花板。由于美国电网并网延迟长达53个月，且特大型变压器交付周期拉长至120-160周，AI资本开支向云收入转化时延从4个季度拉长至6个季度。容量电价飙升使100MW数据中心电费翻倍并压缩推理毛利至42.0%。建议将硅侧权重削减至2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 公用事业权重下调后对电网基础设施与能源基建板块的风险覆盖评估", "summary": "2026-06-03，本卡压力测试 Research note 06 的 22% 至 17% AI 公用事业降权。结论：降权合理，但必须从长久期美国发电 / IPP 暴露中执行，不能压缩电网设备底仓。IEA、EIA、PJM、DOE 与 Eaton / Quanta / Vertiv 数据显示，AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [ { "title": "变压器与设备供应缺口延缓了受监管公用事业费率基数的短期转化并恶化了现金流，但物理稀缺性结构性地推高了存量电网、已锁定的并网项目和具备溢价能力的电网设备制造商的重估价值。", "summary": "截至2026-06-05，受AI数据中心和电气化推动，变压器供需缺口达30%，大型电力变压器交货期拉长至36-48个月，价格比2019年暴涨70%-95%。定量模拟表明，30%的资本支出因瓶颈延迟2年竣工，在产能预付模式下会导致公用事业公司5年累计净利润减少10.71亿美元（-3.94%），并产生…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 3, "neutral": 0, "value": 56.2, "events": [ { "title": "AI算力瓶颈已从GPU芯片转移至电力和电网，资产配置应超配锁电公用事业与具备套保及价格联动出海变压器商，低配低毛利电缆及面临延迟通电风险的次级硬件与服务器链条。", "summary": "截至2026-06-05，本卡综合研判了AI算力物理瓶颈对公用事业、电网设备及硬件链条的策略定价影响。全球数据中心电力消耗翻倍（2030年达950 TWh）与大功率变压器交期拉长（2-4年）加剧并网延期，催化了以微软核能协议为代表的表后自备电力模式。配置上建议超配锁电能力强的公用事业（福能股份、江…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力确是硬上限但约束在电网设备与并网而非 MWh，alpha 集中在 OEM 与受监管 T&D，搁浅风险锁定 2025–26 批发电源项目融资 vintage。", "summary": "电力是真实但时间有限的 AI 物理瓶颈，资本可压缩约 30–40% 延期但不能消除：约束子层是电网设备 + 并网（变压器交期 3.5–5 年、美国并网排队约 2.6 TW、PJM 群组 ~286 GW），而非装机 MWh 本身 [S1][S2][S3][S11]。最干净的可定价 alpha 在物理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力瓶颈催生的“去电网化”倾向，正将估值溢价与定价权从受ROE上限约束的传统电力公司，转移至享有高毛利与长周期订单积压的非监管电气设备及微电网制造商。", "summary": "截至2026-06-05，云服务商为绕过5-10年电网并网排队，正加速转向自备天然气和长期SMR微电网模式。这导致公用事业面临“大客户脱钩”与资产闲置风险。然而，由于监管（如FERC否决Talen-AWS案）和高额备用电费，完全脱网难以实现。估值溢价正从受ROE限制的公用事业（PE 15-18x）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "AI算力扩张速度正受到电网并网队列与电网核心设备长交期的刚性物理约束，这加剧了中游算力服务商的资产闲置与重定价风险，同时也赋予了拥有清洁基荷（核电/气电）的独立发电厂及高压输电设备供…", "summary": "定量评估AI电力基建瓶颈。北美并网队列延迟53个月[S3]使100MW数据中心延迟并网29个月，导致10.5亿元GPU闲置折旧80.56%并沉没资金成本2.41亿元[自测算]。2026美电网Capex增至2388亿美元[S6]，变压器交期拉长至24-48个月[S7]。基荷缺口迫使巨头绕过电网签约核…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 2, "neutral": 1, "value": 68, "events": [ { "title": "AI扩张约束从GPU转向“电力之墙”，推动拥有核电与电网资产的公用事业板块实现“基础设施化”重估，并使电网设备交货瓶颈成为超额收益来源。", "summary": "截至2026-06-08，AI电力需求已推动公用事业从防御板块转向高增长基础设施。Constellation等独立发电厂因核电购电协议重估至30倍市盈率。由于高压变压器等关键设备交货期超过160周（5年），“电力之墙”成为AI扩张的主要瓶颈。建议超配核电/燃气发电IPPs、电网设备及铜/铀等实物资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支正从算力硅片向电力物理基础设施发生不可逆的溢出，A股配置须从TMT硬件向电网出海及系泊设备进行长周期大轮动。", "summary": "截至2026年6月4日，53个月并网排队与10倍容量电价标志着AI产业进入电力死锁。云巨头Capex正从算力硬件向电网基建与分布式电力物理溢出。A股策略应低配估值透支的TMT算力硬件，大幅轮动至受益于海外电网建设红利的变压器/高压线缆出海企业，并战术配置浮式数据中心催生的系泊设备及高股息公用事业。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 交易是否已经从 GPU 稀缺，转向电力、并网、核能和电气设备稀缺", "summary": "AI 电力瓶颈链条本质是物理基本面的结构性轮动，外裹一层股价动量泡沫。底层约束（把签约的 7×24 吉瓦级电力送到机柜）真实且跨年度：IEA 称数据中心用电 2030 年翻倍至约 945 TWh；变压器交付期 2026 年逾 160 周、开关柜售罄至 2028；到 2030 年约 1.4 万亿美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 3, "neutral": 0, "value": 60.4, "events": [ { "title": "只有SERC、MISO South和ERCOT自接发电一类辖区能在2027-2028兑现AI数据中心MW，PJM/CAISO/NYISO并网约束进一步压低重资产GPU云ROIC，强化…", "summary": "把卡03电力OPEX冲击和卡04变压器backlog映射到ISO/RTO层面，2027-2028真正能把AI数据中心宣布MW转为通电MW的只有SERC（TVA、Duke、Southern/Georgia Power）、MISO South（Entergy）和ERCOT中走self-build/76…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器、switchgear、UPS/冷却和并网排队正在把AI算力扩产从GPU采购问题变成送电交付问题，2027年前难以全面缓解，继续压低重资产算力商早期资产周转并抬升ETN/VRT…", "summary": "截至2026-06-10，工业设备证据支持前序空重资产GPU云、配多电力/设备链的主线。Eaton、Vertiv、GE Vernova的订单与backlog仍快于收入交付：Eaton Electrical Americas backlog同比+44%、数据中心订单约+240%；Vertiv Q4 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与电网瓶颈已成为AI算力落地的物理天花板，如果变压器短缺与电网容量摩擦导致40%项目延期，将直接在2026-2027年期间递延多达700亿美元的半导体行业营收，加剧低能效云商的闲…", "summary": "2026年6月9日报告通过对电力供应和电网容量的压力测试发现，物理用电瓶颈和电网并网延迟正在成为制约AI算力中心落地的短线硬伤。如果40%的规划负荷发生被动延期，将在2027年前导致多达280万张GPU无法正常部署，对应半导体行业约700亿美元营收被递延。云厂商面临资本开支闲置与变现周期拉长的双重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 3, "neutral": 0, "value": 61.8, "events": [ { "title": "并网延迟与输电设备短缺已将AI竞争从芯片效率转向‘电力交付权’，制造了3-5年的算力建设停滞期，大幅抬升存量电力资产溢价。", "summary": "截至2026-06-11，电力基础设施已成为AI算力的‘大过滤器’。电网扩容（5-15年）与算力建设（1-3年）的错位导致了严重的并网延时，全美队列达2600GW。变压器交期拉长至4-5年，而PJM容量电价暴涨9倍至$329/MW-day。这种‘硬约束’正在触发资金从GPU转向拥有现成电力的公用事…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-11，所谓的“表后并网”（BTM）并非彻底避开48个月电网排队的银弹，而是将瓶颈结构性转移至厂区变电设备缺口、燃气轮机5-7年超长交期以及管道与空气许可审批，在W…", "summary": "截至2026-06-11，我们压力测试了AI数据中心通过“表后并网”（BTM）规避48个月电网并网排队的方案。研究表明，表后并网只是将物理瓶颈转移。核电并网面临严峻监管壁垒（如Talen/AWS ISA遭否决）且仍需大容量备用变压器；现场气电面临重型燃机5-7年、航改型燃机3-5年的制造交期瓶颈，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力建设正从“依赖电网”转向“离网自治”，SOFC与中国电力设备出口链通过解决交付周期错配，成为捕捉AI基建溢价的核心抓手。", "summary": "2026年AI算力扩张的瓶颈已从GPU转向电力并网，使得SOFC与变压器成为核心资产。SOFC通过“时间套利”（交付周期<12个月对比电网并网7-10年）为AI厂商节省数十亿美元机会成本，LCOE在$58-110/MWh区间已具竞争力。中国厂商凭借全球60%的变压器产能及对取向电工钢（GOES）的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 0, "value": 75.2, "events": [ { "title": "由于基荷电源建设与并网周期（5-15年）同AI算力芯片超短折旧周期（3-5年）存在根本性错配，且核电网后直连遭遇FERC强力监管狙击，能源供给侧弹性近中期近乎为零，无法有效对冲算力资…", "summary": "对能源供应侧扩产对冲算力电费风险进行压力测试。基荷电源（核电、燃气）建设和设备交期长达5-15年，与3-5年GPU折旧周期严重错配，导致近中期供给弹性为零。FERC于2026年6月9日再次否决核电直连协议，堵死“网后直连”逃生通道。压测显示4年延迟将导致闲置GPU算力资产高达70%的净现值资本减值…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 1, "neutral": 0, "value": 76.6, "events": [ { "title": "在本土变压器长交期与并网延时常态化下，美国数据中心向非传统市场进口设备并承担10%-50%的进口/转口关税是规避沉没资产延期损失（NPV侵蚀-$200M）的理性财务选择，这导致全球A…", "summary": "截至2026-06-13，本终局卡片合成电网并网队列（~2,600 GW[S1]）及大型电力变压器（LPT）长交期（144-260周[S2]）约束下，美国AI数据中心通过非传统市场（土耳其、巴西、印度、越南）进行变压器“地缘套利”的宏观经济与关税风险。自测算显示，若COD延误3年且Capex已沉没…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 5, "neutral": 0, "value": 54, "events": [ { "title": "高能耗GPU集群的大规模部署导致超大规模云厂商的资本支出全面向电力基建倾斜，在设备端过度拥挤的背景下，公用事业运营资产提供了更优的风险收益比。", "summary": "截至2026年中，全球电网物理极限已成为AI扩张的核心瓶颈。超高密度GPU（如GB200 NVL72，120kW+/机柜）推动云厂商2026年Capex升至6600亿-7250亿美元，其中逾60%投向电力与基建。美国数据中心电力需求年底将达41 GW。数据证实，拥挤度极高的电网设备已透支预期，而处…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力中心的电力激增与电网并网瓶颈，为电力板块提供了超越传统周期的长协盈利溢价，强力支持将红利配置中电力权重上调至2.5%以对冲地产下行。", "summary": "截至2026-06-13，AI基础设施分析师综合验证了上调电力板块权重的逻辑。IEA《Electricity 2026》显示，AI推理带来常态化电力负荷，全球数据中心耗电将在2030年翻倍逼近950 TWh。国内十万卡GPU集群建设使并网与变压器成为核心瓶颈。电力企业受燃料成本下降与AI电荒预期双…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘冲突带来的能源价格上行与物理并网延迟正在迫使云厂商将资本开支从纯算力硬件向电网、变压器和温控液冷基建倾斜，产生显著的“基建挤出效应”；在脱碳路径上面临自备燃气发电（高ROIC但需…", "summary": "截至2026-06-13，本报告针对中东地缘能源冲击与电网瓶颈对AI算力基建的影响进行了压力测试。在$110/MWh高电费与$1364.40M配套基建（变压器GOES溢价及液冷）刚性支出下，超大云厂商为维持10%靶向ROIC必须调降GPU采购预算54.32%，挤压芯片商毛利至45.27%。在能源脱…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 数据中心的关键约束正在从“有没有电”转向“能否拿到变压器、开关柜和完整电气链条的交付槽位”。", "summary": "截至2026-06-14，本卡片支持但收紧前两张卡片结论：AI 扩容的下一个断点更可能是电网设备，而非发电资源。大型电力变压器平均交期 128 周、GSU 144 周，Eaton、Schneider、Siemens Energy、Hitachi Energy、Prolec GE 与 WEG 均在扩…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 6, "neutral": 0, "value": 49.4, "events": [ { "title": "物理电网瓶颈正迫使 AI 基础设施从‘暴力规模扩张’转向‘受约束的能效优化’，电力并网速度已取代芯片供应成为核心估值护城河。", "summary": "本报告通过压力测试确认了物理电网并网排队（7-10年）与变压器交期（4-5年）已成为 AI 基础设施扩张的硬约束。定量模型显示，1年的并网延迟将导致 100MW 数据中心损失 5.5% 的生命周期价值，且快速的 GPU 迭代周期（12-18个月效用减半）加剧了闲置产能的 ROI 损耗。这种物理瓶颈…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "有电力保障的AI基础设施Capex确定性上升；无电力保障的Capex将被电网价格、保证金和设备交期继续配给。", "summary": "截至2026-06-14，本卡综合判断：电力风险并未消失，但正在从未定价瓶颈转化为可预算、可筛选、可合同化的分配机制。PJM容量价、ERCOT大负荷规则、核电PPA和低空置率让头部hyperscaler的AI数据中心及GPU集群Capex更具名义金额确定性；但保证金、容量费、变压器交期和建设成本通…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与电网瓶颈对Tier-1 2026–2027并网影响已基本被锁单、棕地与表后发电对冲；真实滑期约3–7 GW集中在2028+绿地与Tier-2投机建设，Research not…", "summary": "本卡(10/10)压力测试变压器/GSU/电网交期对Tier-1 AI数据中心并网的真实约束。结论：瓶颈真实存在但Tier-1时间表已基本被定价——2022–2024年下达的多年期框架订单、棕地复用既有并网点、表后核电与现场燃气PPA共同吸收了120–210周的设备交期。自测算显示2027–202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备供给瓶颈导致的成本不确定性迫使云厂商在 2026-2027 年将 CAPEX 向电力基础设施大幅倾斜，AI 算力部署从“受制于GPU”全面转向“受制于通电能力”。", "summary": "截至 2026-06-14，电网设备出海的 ROE 波动与建厂压力正向超大规模云厂商传导。美国变压器交期拉长至 160-200 周，云厂商的核心痛点从“缺芯”变为“缺电”。为锁定产能，云厂商被迫承担更高前期成本，数据中心非 IT 开支（电力/制冷/土建）占比预计从历史的 20% 升至 35%-40…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 0, "value": 68.8, "events": [ { "title": "把firm power稀缺主线落地为风险预算分腿的跨资产哑铃，由利率、拥挤度和设备交期三类客观触发器驱动调仓，捕获稀缺租金而非押注AI链方向。", "summary": "截至2026-06-15，本卡综合第05–07卡的firm power约束主线，构建跨资产哑铃组合：多头约65%风险预算（受监管电网L3、商业核电IPP L2、长周期电力设备L1、经认证SiC/GaN平台L4），空头/低配约35%（缺电GPU租赁S1、未锁并网DC开发商S2、无定点SiC供应S3）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "公用事业瓶颈的跨行业配置含义：电力设备稀缺、受监管公用事业资产基数扩张与AI基础设施估值折价", "summary": "若firm power而非GPU是真正瓶颈，稀缺租金应归于占据约束的一方。做多三层：①订单售罄的长周期电力设备（GE Vernova GEV燃气轮机在手约100GW、档期至2028-2030[S6]；变压器交期达4年[S1][S2]）；②商业核电IPP（CEG约17-21x、VST约13x，签20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026 06 15，通过对五大云厂商6020亿美元Capex计划、AI数据中心3倍金属强度以及中游设备厂商（Eaton, Vertiv, Schneider）创纪录的150亿美…", "summary": "截至2026 06 15，通过对五大云厂商6020亿美元Capex计划、AI数据中心3倍金属强度以及中游设备厂商（Eaton, Vertiv, Schneider）创纪录的150亿美元积压订单的深度穿透，我确认AI与电网基础设施投资已实质性转化为铝、铜及电力设备的物理订单流。这一“订单实物化”验证…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 1, "risks": 6, "neutral": 0, "value": 60.2, "events": [ { "title": "2026年后AI基础设施的核心溢价来自“可通电MW x 液冷转化效率 x 能源期限”，核能定终值，储能和电网设备定并网质量，液冷定GPU密度。", "summary": "截至2026-06-17，本卡支持上一张卡的判断：AI基础设施估值正从GPU数量转向确定性电力、并网质量和液冷密度。核电PPA提升长期站点价值，但多数新增容量在2027-2030年后才兑现；2026-2028年真正改变采购结构的是液冷机架、电气配套、储能和电网支撑设备。云厂商资本开支将更像“MW容…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "通胀和高利率提高了AI硬件基础设施的质量门槛，但没有削弱其配置价值；应买入瓶颈供应商而非泛AI叙事。", "summary": "截至2026-06-16，本卡支持“硬科技配置中保留AI基础设施”的上一轮判断，但建议从广义AI beta收窄到有订单、定价权和瓶颈属性的硬件基础设施。高利率会压缩投机型数据中心开发商和远期AI应用估值，却未削弱GPU、定制AI芯片、HBM、配电、冷却、变压器、开关设备和已锁定电力的数据中心容量的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理基础设施需求足以在没有流动性爆发的情况下维持电网设备与稀缺投入品周期，但市场仍应惩罚同时需要廉价资本和完美部署节奏的AI算力商业模式。", "summary": "截至2026-06-17，本卡压力测试结论是：在宏观流动性不爆发的情景下，AI算力、电网接入、变压器和液冷的物理需求增量足以支撑头部hyperscalers与稀缺电网供应链资产，但不能保护依赖廉价资金、缺乏确定电力或确定客户的投机型AI算力项目。全球数据中心用电到2030年约945 TWh、美国到…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与开关柜交期已经把AI扩容约束从电力需求预测推进到工业制造产能约束，证实chips与grid属于同一physical-AI capex风险桶。", "summary": "截至2026-06-16，本卡支持“电网设备交付周期证实物理红线”的判断。Wood Mackenzie 报告显示大型变电站与发电机升压变压器交期为80-210周，且美国2025年电力/配电变压器供应缺口估计为30%/10%；POWER Magazine 指出开关柜2025年Q2平均交期44周。Ea…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 4, "neutral": 0, "value": 73.6, "events": [ { "title": "2026-2030 年 AI 电力缺口需要 BTM 绿电和储能桥接，但真正可投资的是“多电源、多设备、多接入”的组合解，而不是单一 SMR 或单一 BTM 路线。", "summary": "截至 2026-06-17，本卡支持“SMR 在 2026-2030 年无法填补 AI 用电缺口”的判断，但否认 BTM 绿电直连加储能调频是唯一解。NuScale 罗马尼亚项目首个模块 COD 指向 2033 年，近期数据中心用电则在 2028-2030 年快速上行。BTM 是最有价值的低碳桥接…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "中国电力设备出口受益是真实结构性机会，但上限取决于地缘政策折价；国内 AI 算力扩张的核心约束将从芯片采购转向局部电力接入和稳定电源。", "summary": "截至2026-06-18，本卡 synthesize 并部分支持上一卡：全球电力设备短缺是真实瓶颈，中国在变压器、HVDC/UHV、开关设备和数据中心供配电环节具备结构性出口机会，但美国/欧洲关键电网的安全审查会限制份额。国内 AI 不是全国性缺电问题，而是局部“合格兆瓦”问题：IEA 估算中国数…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI主题内部的定价权正从芯片端的硅片稀缺向电力、并网熟地和变压器等物理基础设施供应链迁移，电网并网延迟构成了对云巨头闲置算力资本的巨额惩罚，迫使价值向能源与硬件节点转移。", "summary": "截至2026年6月18日，由于全美并网队列激增至2,600 GW且平均等待时间达5年，AI基础设施的核心瓶颈已从GPU芯片稀缺转移到电力、并网及电力设备。定价权正向控制熟地、独立发电和电网设备的物理环节迁移，以规避巨大的闲置算力沉没资金惩罚（10万张B200芯片闲置年资本占用成本达1.8亿美元）。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力向电力设备的资本开支传导扩大了实体“流动性黑洞”，高企的无风险利率与公用事业“成长化”转型持续挤出长久期资产的配置资金。", "summary": "本报告支持AI基础设施分析师的判断，量化了AI电力需求向能源基建的传导。2026年全球数据中心用电预计达565 TWh，中国“十五五”电网投资将达5万亿元，且全球面临30%的变压器结构性缺口。这种大规模实体CapEx形成巨大的“流动性黑洞”，虹吸市场资金，推高资本成本，对医药等长久期板块估值维持持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 6, "neutral": 0, "value": 55, "events": [ { "title": "新增AI风险预算应从继续加半导体Beta转向电力稀缺链条，半导体保留核心仓位但不再是边际加仓方向。", "summary": "截至2026-06-19，本卡支持并推进此前从通用软件/加密杠杆转向AI硬约束的思路：半导体基本面仍强，但新增风险预算的边际胜率已转向电力设备、并网基础设施、现场供电和核燃料。建议下周新增AI风险预算60-70%投向电力与供能约束，20-30%保留精选半导体，0-10%用于现金或对冲。核心依据是h…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "G2/G3瓶颈没有被效率叙事推翻，反而因变压器、电气设备与并网队列的多年约束而强化近现金电力设备订单资产的相对优势。", "summary": "截至2026-06-19，本卡支持Research note 02：AI电力交付瓶颈是真实供给侧闸门。美国数据中心用电2028年可达325-580 TWh，电网设备采购和并网队列仍是多年周期；美国活跃并网队列2025年底仍超2,060 GW，PJM数据中心区等待36-48个月；欧盟并网等待可达2-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网和电力设备扩产正在成为AI基础设施建设的主要瓶颈，但投资应落在设备与可回收电网资本开支，而不是无差别公用事业敞口。", "summary": "截至2026-06-19，本卡支持上一张卡片：AI基础设施的主要约束正在从GPU转向电力可获得性、电网并网和长交期电力设备。IEA预计全球数据中心用电量从2024年约415 TWh升至2030年约945 TWh；DOE/LBNL预计美国数据中心用电量从2023年176 TWh升至2028年325-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由于大型电力变压器与测试大厅产能严重受限，2027-2028年计划新增的AI数据中心有43%面临通电并网延期[自测算]，由此引发GPU订单推迟，对主要芯片制造商及云厂商带来显著的收入…", "summary": "基于2026-06-19数据，本分析量化了电网变压器与并网延迟对2027-2028年AI硬件产业的影响。模型显示，原计划30GW数据中心管线因变压器交期拉长至120-210周而延期12-24个月，实际仅能交付17.1GW，缺口达43%[自测算]。这导致GPU出货量在2027年减少4.8M颗、202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 6, "neutral": 0, "value": 56.4, "events": [ { "title": "电力设备交期和并网流程已成为AI算力基础设施建设周期的确定性约束，并会放大高负债算力租赁商的融资与估值压力。", "summary": "截至2026-06-19，本卡支持前序判断：AI算力建设的硬瓶颈已从GPU采购转向电力交付。DOE/LBNL与EPRI均显示数据中心用电占比快速上升；Wood Mackenzie与POWER显示大型变压器交期已达120-144周，部分大型设备80-210周；JLL称主要数据中心市场平均并网等待超过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "16.18%的硬件Capex下修会压低AI基础设施整体估值，但订单传导不是线性的：液冷最敏感，电网设备和BTM发电更受长期电力容量瓶颈支撑。", "summary": "截至2026-06-19，本卡把H2 AI硬件Capex下修16.18%作为未外部确认的压力测试输入。结论是分层传导：GPU/服务器采购最先受冲击；液冷因绑定高密度机柜节奏，订单和估值敏感度较高；电网设备因并网稀缺、变压器长交期和backlog，订单不应按同等比例下修；BTM自备发电估值会先压缩，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网吸纳能力瓶颈会削弱AI投资对近期r*的无摩擦上行冲击，但会把周期拉长并转向电力设备资本开支，因此投资需求逻辑被折价而非被证伪。", "summary": "截至2026-06-19，本卡压力测试Research note 06的r*上行逻辑。结论是：电网与变压器瓶颈会降低近期资本开支效率，但不会取消投资需求渠道。Goldman估算2026-2031年AI基础设施资本开支为7.6万亿美元；IEA提示约20%规划数据中心项目存在电网延期风险。套用于202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力确是 AI 算力扩张的真实瓶颈，但锁喉点是 LPT + 并网排队而非发电量本身，这把 SPCX/xAI 算力飞轮落地推迟 6–12 个月并把价值迁移到电力设备与核电 IPP。", "summary": "工作日 2026-06-19。从能源行业视角对 Research note 02 的'电力是最终瓶颈'结论进行压力测试：方向正确，但瓶颈不在发电量，而在并网排队 + 大型电力变压器 (LPT) + 高压开关 + 熟练劳动力四道串联约束。Colossus 2 (2 GW) 年用电量约 20.8 TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 9, "neutral": 0, "value": 39.8, "events": [ { "title": "电网设备紧缺不会终止最强Hyperscaler的AI capex，但会把'宣布投资'与'已送电算力'之间的时间差拉大，并迫使选址从效能最优转向电网弹性最优。", "summary": "截至2026-06-21，本卡压力测试电网设备紧缺对AI算力中心扩建的传导。结论是：变压器backlog对设备商是定价权，对Hyperscaler是进度税。Wood Mackenzie显示电力变压器128周、GSU 143周、开关设备44周；相对18-24个月数据中心建设周期，若未提前锁定设备sl…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力基础设施建设的成本传导与Capex弹性", "summary": "压测显示：铜/铀/电力设备成本挤压确实砸在Hyperscaler增速最快的Capex环节——电力/电气基础设施（占数据中心建造成本40%–50%[S6][S7]），使AI数据中心单MW建造成本上升约+8%至+12%[自测算]。但它既不实质拖慢扩容，也几乎不向算力租赁端传导。两点决定Hypersca…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "600kW+机柜把AI扩容瓶颈从芯片采购推向公用事业级送电能力；2027-2028年最可能的失效模式是资本开支变成部分送电园区，而非芯片完全不到货。", "summary": "截至2026-06-21，本卡支持并收窄Research note 08：2027-2028年AI集群交付下调的关键不是单纯缺电或电价，而是600kW+机柜带来的可送电高密度MW需求，与变压器、开关设备、公用事业并网和调试周期同时卡住。IEA预计全球数据中心用电2024年约415 TWh、2030…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与核电供应链紧缺是选择性利空：它削弱无传导能力的CAPEX买方，却强化拥有稀缺产能和重定价backlog的设备龙头。", "summary": "截至2026-06-21，本卡压力测试Research note 06的电网赋能者超配逻辑。结论是：供应紧缺会抬升CAPEX和工期压力，但不会普遍削弱最优质电网设备商的盈利预期；它会区分定价者和成本承受者。变压器、HVDC、开关设备和核级部件龙头能把稀缺转化为backlog和价格，项目业主、固定价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 9, "neutral": 0, "value": 41.2, "events": [ { "title": "在面临128-144周电网硬件交期与并网锁死的‘硅-网时钟错配’下，BTM核电部署的超高时间价值（效益成本比达56.8倍）超越了电力采购的单价成本，正在重构超大规模厂商的资本开支，同…", "summary": "本卡作为白板第10期终结卡，全面评估了Behind-the-Meter (BTM) 核电及现场微电网部署对超大规模数据中心降低电网并网瓶颈风险与资本开支规划的深远影响。模型表明，24个月的电网并网延迟对800亿美元的GPU硬件造成高达397.3亿美元的无形资产与折旧减值（减值率49.66%）。在此…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核电与可再生能源供应侧的物理瓶颈：规避并网排队能否解决电力规模缺口？", "summary": "压力测试卡片01的监管破局论。FERC的Eddystone容量置换豁免仅是配置抢跑，无法制造净新增可靠电源。需求增量巨大：EPRI测算美国数据中心峰值负荷将从2024年约21.5GW升至2030年中性情景71GW，即约+49.5GW[自测算]。供给侧无法匹配：核电重启/扩容到2029年仅约+5GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力负荷支持电网资本开支周期，但最佳表达不是无差别买公用事业，而是买硬件瓶颈和具备大负荷费率保护的公用事业。", "summary": "截至2026-06-22，本卡对“电力交付层稀缺”主线做压力测试：电网资本开支真实扩张，但公用事业收益并非自动释放。监管机构仍能支持高强度投资，前提是大负荷费率、最低账单、抵押品和退出费把数据中心成本因果关系锁定；否则可负担性政治会压缩费率回收。硬件端变压器和开关设备短缺更像时间瓶颈，会延迟费率基…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力可交付性将配给AI算力扩张的地点和节奏，但铝代铜只是放大器，不是全国性致命瓶颈。", "summary": "工作日为2026-06-21。本卡压力测试第04卡结论：AI算力扩张确实进入电力约束阶段，但“铝代铜”不是致命约束本身。真正瓶颈是大负荷接入、主变压器、开关设备、变电站档期、稳定电源和本地可靠性标准的组合。美国数据中心用电从2023年176 TWh可能升至2028年325-580 TWh，全球到2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 4, "neutral": 0, "value": 72.6, "events": [ { "title": "真正受益的是电力交付瓶颈和已签长期购电/费率保护资产，而不是泛化公用事业或所有新增兆瓦。", "summary": "截至2026-06-22，本卡综合上一张AI基础设施观点：数据中心用电增长仍成立，但公用事业受益必须按“执行风险由谁承担”重新排序。最优是电网设备和电力服务，其次是有已签负载承诺及最低计费/大负荷费率保护的受监管公用事业，再次是已签长期PPA的核电或稳定清洁电源。泛化merchant gas、水资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "更久维持高位、电力成本推动、AI资本支出搁浅时点与芯片库存尾部下的多资产组合定位", "summary": "本卡以asset-allocator视角综合四卡链条，落到一套定位框架。结论：四卡为「一个体制」，正确组合是杠铃——(1)做多物理瓶颈（电力/电网设备、燃气轮机OEM、独立电力/核电PPA），无论GPU是否按时通电都收多年在手订单年金，且做多令美联储紧缩的成本推动本身，是自我融资宏观对冲；(2)低…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网高压变压器与设备交付瓶颈（交期拉长至18-36个月、并网排队达4-7年）将严重延迟并吞噬公用事业的‘成长期权’价值，导致闲置硬件发生严重技术减值，甚至抹去绿电直连政策的套利优势。", "summary": "本卡作为压力测试卡，系统分析了大型变压器和电网装备交付瓶颈对电力价值链中‘成长期权’释放的限制。定量模型显示：（1）2年并网延迟会对云服务商800亿美元的闲置GPU硬件资产造成49.66%的使用价值减值，即带来397.3亿美元的减值准备；（2）在通缩API定价下，若电网拥堵或容量电费暴涨使电价升至…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备紧缺与并网延迟正倒逼AI云厂商调整选址逻辑，通过工作负载地理分流、核电结构化直供、现场过桥发电以及电气设备提前大批量采购，来对冲并网延误风险，将AI资本开支推向重度能源与电力…", "summary": "截至2026-06-22，高压变压器交期拉长至24-48个月且排队并网超5年，成为AI集群部署核心瓶颈。超大规模云厂商正通过将训练与推理负载地理分流、签署核电结构化购电协议（PPA）、部署现场天然气发电机过桥以及提前3-4年锁定集团化设备产能等调整选址与容量规划，以对抗延迟风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 6, "neutral": 0, "value": 62, "events": [ { "title": "电网输变电容量与并网审批周期（而非总体发电量）已成为AI算力部署的终极物理刚性瓶颈，并已实质性挤压IT硬件采购预算。", "summary": "本报告对AI算力扩张的电网与能源瓶颈进行压力测试。分析表明，虽然电力设备订单充足，但美国并网队列平均达5.5年，且面临FERC 2026年6月大型负荷监管及弗吉尼亚GS-5等防止成本转嫁政策。变压器等重电面临GOES原材料短缺，Tier-1厂商积压至2029-2030年。表后核电直供监管被封死，燃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网队列与基础设施扩容压力评估", "summary": "本卡支持卡片1–2：约束已从Capex/GPU转向已通电的MW，并将瓶颈量化为双向并网堵塞。发电侧：美截至2025年底约2,060GW排队（约现有装机2倍），408GW已签约未通电，申请到商运中位时长翻倍至>4年。负荷侧：ERCOT大负荷队列达226GW（近4倍于2024年底63GW，77%为数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支削减15%将导致全球约1.19 GW的数据中心部署延迟，阶段性缓解电力变压器供需瓶颈和电网并网压力，同时导致独立发电厂与电力设备标的面临估值去化风险。", "summary": "截至2026-06-23，本卡评估了AI资本开支下修对数据中心、电网组件及能源供应链的传导效应。模型表明，15%的AI资本开支削减（-661.5亿美元）将导致全球约1.19 GW的物理数据中心IT容量建设延迟。这将有效缓和大型电力变压器（交期24-48个月，价格较2019年上涨77-95%）和输电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 8, "neutral": 0, "value": 51.4, "events": [ { "title": "变压器与开关设备长周期会迫使AI数据中心按可送达电力重新选址，并把超大规模资本开支从GPU/土建扩展到电力交付栈。", "summary": "截至2026-06-25，本卡片压力测试卡片06：AI算力选址确实会从“便宜电”转向“可信time-to-power”，但不是简单回到传统成熟枢纽。变压器、开关设备、变电站与并网改造成为已带电容量的关键约束；低电价若没有设备槽位和送电排期，只是搁浅期权。超大规模厂商的资本开支将更多前置到电气设备预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施资本开支仍成立，但必须从GPU纯敞口重估为电力锁定型、费率可承担、设备可交付的电网重资产敞口。", "summary": "截至2026-06-25，本卡压力测试AI算力基础设施扩张：需求真实，但瓶颈已从GPU采购转向确定性电力、并网、变压器/开关设备和费率承担。美国数据中心2028年用电可能达325-580 TWh，全球2030年约945 TWh；100,000颗GB300 GPU集群在1.15 PUE假设下约需22…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力资本开支超级周期的电力/电网天花板及其对新一代信息技术全链的约束", "summary": "电力/电网上限真实存在且已是绑定约束：美国并网 3–7 年、变压器 4–5 年、燃气轮机售罄到 2030、30–50% 的 2026 产能滑至 2028 [S5-S8]；微软约 800 亿美元 Azure backlog 受电力而非需求约束 [S9]。但 2026–2028 内它递延而非毁灭需求—…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "发电侧与配电侧在2026年正陷入‘既无路（电网排队超4年及变压器稀缺）也无车（燃机交期5-7年及SMR未商业化）’的双端挤压困境，共址直供监管红线的收紧彻底消除捷径，导致算力资本开支…", "summary": "截至2026年6月25日，配电与电网设备交期拉长使得发电侧面临严重并网瓶颈。大型电力及GSU变压器交期达18-48个月，GOES硅钢供给偏紧，并网排队中位数超4年。FERC否决核电表后直供迫使亚马逊等转向表前PPA，面临同样电力瓶颈。同时高确定性燃气轮机交期达5-7年，SMR在2030年前难以放量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 4, "neutral": 0, "value": 76.8, "events": [ { "title": "AI基础设施的主要瓶颈正从GPU可得性转向电力可得性，真正受益的是能缩短通电时间、提高负荷密度并保障可靠供电的设备与电网环节。", "summary": "截至2026-06-26，本卡支持上一张“算力切向电力”的轮动逻辑，但把核心约束细化为并网、变压器/开关设备、机柜功率密度和可靠供电。全球数据中心用电预计由2024年415 TWh升至2030年945 TWh；中国2030年约279 TWh，全国占比不高但局部冲击显著。1 GW AI IT负荷在P…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-26，AI算力扩张的能源约束不是全国缺电，而是电网可交付容量、关键设备和建设周期无法按GPU节奏扩张。", "summary": "截至2026-06-26，本卡支持前序“电力是AI基础设施硬约束”的方向，但将其界定为区域和时间维度的可交付容量瓶颈，而非全国年度电量立刻短缺。LBNL预计美国数据中心用电从2023年176 TWh升至2028年325-580 TWh，折合新增约17-46 GW连续平均负荷。公用事业2025-20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论", "summary": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论 Feasibility of Bring Your Own Power (BYOP) and Gas PPA for AI Data Centers 在小堆核电(SMR)2030年前难以落地、碳约束拉平公耗曲线的现实背景下，国内与海外AI数据…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-25，AI硬件需求的下一核心约束不是芯片是否能生产，而是足够多的兆瓦能否被许可、接入、冷却并按期商业化。", "summary": "截至2026-06-25，本卡支持上一张卡的AI硬件稀缺性主线，但将瓶颈从GPU/HBM进一步上移到可通电数据中心、电网接入、冷却和电气设备。IEA、Berkeley Lab、EPRI、CBRE和JLL数据共同显示，AI需求仍强，但项目能否转化为创收产能取决于电力和并网速度。结论是：稀缺性继续支撑…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "电力设备扩产会缓解AI数据中心并网瓶颈，但无法在2026-2027年完全解除大型变压器和输电接入的物理约束，因此算力利润扩散仍是受瓶颈约束的电网资本开支周期。", "summary": "截至2026-06-28，本卡支持上一卡的物理瓶颈判断，但强调扩产的时间结构：变压器、开关柜和电网设备扩产是真实的，Eaton、Hitachi Energy、Siemens Energy等制造商已披露新增产能和高积压订单；但大型电力变压器仍受定制设计、GOES/铜、测试槽位、运输和并网研究约束，新…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 4, "neutral": 0, "value": 81, "events": [ { "title": "2026年全球AI数据中心电网瓶颈与“自主供电”战略的崛起", "summary": "2026年全球超大规模厂商的AI算力基建受限于电网瓶颈与关键电力设备（变压器交期达5年）短缺。这迫使厂商将部分资本支出（预计2026年总支出达7250亿美元）从纯算力（GPU）转向自主供电（BYOP）和表后（BTM）能源方案。预计该转移将延迟算力上线速度，并推高长期持有成本。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股应继续支持AI基础设施主线，但买点应从纯算力叙事转向供电约束受益资产，以35%/30%/25%/10%的科技成长、电网设备、公用事业红利和现金对冲哑铃执行。", "summary": "截至2026-06-28，本卡支持上游判断：AI基础设施交易没有被宏观衰退担忧证伪，而是被供电、电网接入、变压器和冷却约束重构。A股策略应从单一科技成长交易转为35% AI科技成长、30%电网及电力设备、25%公用事业红利、10%现金或对冲的主题哑铃。科技仍有上行，但电子和通信估值已处三年高分位；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "研判总结 / Research Summary 本卡片作为 公用事业分析师 (Utilities Analyst) ，锚定研究院权威工作日期 2026 06 28 ，采用 合成 (S…", "summary": "研判总结 / Research Summary 本卡片作为 公用事业分析师 (Utilities Analyst) ，锚定研究院权威工作日期 2026 06 28 ，采用 合成 (Synthesize) 的立场，对前序卡片中关于量子/AI基础设施对电网和能效重塑的探讨进行了物理瓶颈与投资可行性的深…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI驱动的结构性裁员实质上是算力投资回报（ROI）兑现的证明，云厂商AI资本开支具有极强韧性，其真实制约并非宏观衰退担忧，而是电网并网延迟与重型电力设备的物理交付瓶颈。", "summary": "本卡片对“AI裁员压力动摇算力基建投资”的假设进行压力测试并予以否定。AI驱动的结构性裁员（Challenger裁员AI占比达40%）是AI投资回报（ROI）兑现的实体证据，而非衰退信号。四大云厂商资产负债表坚韧，AI基建投资是一场生存竞争。然而，物理供电（并网排队5-7年）与设备交付（变压器交期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 1, "risks": 3, "neutral": 0, "value": 96, "events": [ { "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "summary": "支持第06卡\"并网延迟→利润压缩\"的传导论点，但将单向结论细化为分化。针对11GW数据中心建设缺口与12-18GW刚性局部电网缺口，48-60个月并网延迟通过两条通道传导：(1)经订单/PPA递延与尾款拖延压制发电子链开工率——5.7%销量递延经经营杠杆放大为约-21%利润压缩[自测算]；(2)将…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] } ], "href": "/investor/theses/thesis-1aaa752ed4/", "networkHref": "/investor/research-network/#thesis-1aaa752ed4" }, { "id": "thesis-d4473d35e5", "source": "tracked_thesis", "title": "猪周期、粮食与农产品价格决定中国低通胀修复的弹性", "titleEn": "Hog cycle, grain, and agricultural prices define the elasticity of China's low-inflation repair", "lane": "食品农业/猪周期", "laneEn": "Food, agriculture, and hog cycle", "laneHref": "/investor/lanes/food-agriculture-and-hog-cycle/", "investableFocus": "食品农业/猪周期：002714.SZ, 300498.SZ, 000876.SZ, ADM, BG, DBA", "stockUniverse": [ { "ticker": "002714.SZ", "name": "牧原股份", "market": "A-share", "role": "生猪养殖", "roleEn": "生猪养殖", "bucket": "core", "href": "/investor/stocks/002714.SZ/" }, { "ticker": "300498.SZ", "name": "温氏股份", "market": "A-share", "role": "生猪/禽类养殖", "roleEn": "生猪/禽类养殖", "bucket": "core", "href": "/investor/stocks/300498.SZ/" }, { "ticker": "000876.SZ", "name": "新希望", "market": "A-share", "role": "饲料与养殖", "roleEn": "饲料与养殖", "bucket": "core", "href": "/investor/stocks/000876.SZ/" }, { "ticker": "ADM", "name": "Archer-Daniels-Midland", "market": "US", "role": "全球粮食贸易", "roleEn": "全球粮食贸易", "bucket": "core", "href": "/investor/stocks/ADM/" }, { "ticker": "BG", "name": "Bunge", "market": "US", "role": "粮油贸易加工", "roleEn": "粮油贸易加工", "bucket": "core", "href": "/investor/stocks/BG/" }, { "ticker": "DBA", "name": "Invesco DB Agriculture Fund", "market": "US ETF", "role": "农产品价格代理", "roleEn": "农产品价格代理", "bucket": "core", "href": "/investor/stocks/DBA/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" } ], "stockTickers": [ "002714.SZ", "300498.SZ", "000876.SZ", "ADM", "BG", "DBA", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM" ], "coreView": "赛道：食品农业/猪周期。候选观察标的：002714.SZ, 300498.SZ, 000876.SZ, ADM, BG, DBA。当前共有 40 条证据、3 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 74, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 11, "risks": 114, "neutral": 35, "analystCount": 13, "sourceTypeCounts": { "研究笔记": 113, "分析协作": 16, "研究线程": 30, "研究主题": 1 }, "topTerms": [ "CPI", "PCE", "AIDC", "食品", "cycle", "grain", "agricultural", "prices", "define", "elasticity", "low-inflation", "repair" ], "whatChanged": [ "2026-06-29: 产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "2026-06-29: [auto-handoff] Hormuz shock & core CPI transmission", "2026-06-28: 6月点阵图不是简单过时，而是可能“对得不是理由”：能源理由过时，AI/电网与折旧通道支持更高更久，但不足以单独证明立即加息。", "2026-06-28: 核心服务与住房通胀的粘性：美联储政策的后续驱动力", "2026-06-28: 主要央行保持鹰派按兵不动并非过度收紧的政策错误，而是应对由服务业传导支撑的核心PCE粘性（预计H2维持在3.1%-3.4%）的必要防御，但这会使实体经济承受限制性实际利率压制，支撑久期债防守表达并对杠杆拥挤且具海外宏观高敏感性的中游机械行业构成收缩摩擦。" ], "practical": { "sectorLane": "食品农业/猪周期", "sectorLaneEn": "Food, agriculture, and hog cycle", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "002714.SZ", "name": "牧原股份", "market": "A-share", "role": "生猪养殖", "roleEn": "生猪养殖", "bucket": "core", "href": "/investor/stocks/002714.SZ/" }, { "ticker": "300498.SZ", "name": "温氏股份", "market": "A-share", "role": "生猪/禽类养殖", "roleEn": "生猪/禽类养殖", "bucket": "core", "href": "/investor/stocks/300498.SZ/" }, { "ticker": "000876.SZ", "name": "新希望", "market": "A-share", "role": "饲料与养殖", "roleEn": "饲料与养殖", "bucket": "core", "href": "/investor/stocks/000876.SZ/" }, { "ticker": "ADM", "name": "Archer-Daniels-Midland", "market": "US", "role": "全球粮食贸易", "roleEn": "全球粮食贸易", "bucket": "core", "href": "/investor/stocks/ADM/" }, { "ticker": "BG", "name": "Bunge", "market": "US", "role": "粮油贸易加工", "roleEn": "粮油贸易加工", "bucket": "core", "href": "/investor/stocks/BG/" }, { "ticker": "DBA", "name": "Invesco DB Agriculture Fund", "market": "US ETF", "role": "农产品价格代理", "roleEn": "农产品价格代理", "bucket": "core", "href": "/investor/stocks/DBA/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" } ], "stockTickers": [ "002714.SZ", "300498.SZ", "000876.SZ", "ADM", "BG", "DBA", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM" ], "primaryStocks": [ { "ticker": "002714.SZ", "name": "牧原股份", "market": "A-share", "role": "生猪养殖", "roleEn": "生猪养殖", "bucket": "core", "href": "/investor/stocks/002714.SZ/" }, { "ticker": "300498.SZ", "name": "温氏股份", "market": "A-share", "role": "生猪/禽类养殖", "roleEn": "生猪/禽类养殖", "bucket": "core", "href": "/investor/stocks/300498.SZ/" }, { "ticker": "000876.SZ", "name": "新希望", "market": "A-share", "role": "饲料与养殖", "roleEn": "饲料与养殖", "bucket": "core", "href": "/investor/stocks/000876.SZ/" }, { "ticker": "ADM", "name": "Archer-Daniels-Midland", "market": "US", "role": "全球粮食贸易", "roleEn": "全球粮食贸易", "bucket": "core", "href": "/investor/stocks/ADM/" }, { "ticker": "BG", "name": "Bunge", "market": "US", "role": "粮油贸易加工", "roleEn": "粮油贸易加工", "bucket": "core", "href": "/investor/stocks/BG/" }, { "ticker": "DBA", "name": "Invesco DB Agriculture Fund", "market": "US ETF", "role": "农产品价格代理", "roleEn": "农产品价格代理", "bucket": "core", "href": "/investor/stocks/DBA/" } ], "instrumentTags": [ "002714.SZ", "牧原股份", "300498.SZ", "温氏股份", "000876.SZ", "新希望", "ADM", "Archer-Daniels-Midland", "BG", "Bunge", "DBA", "Invesco DB Agriculture Fund" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 99 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 13, "recentEvents": 99, "supportRatio": 0.07, "riskRatio": 0.71, "conflictScore": 81.2, "evidenceDepth": 100 }, "score": 70.1 }, "events": [ { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 23.75, "matchedTerms": [ "CPI", "evidence", "risk", "dominant", "inflation", "transmission", "falling", "supply" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "食品", "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-28", "title": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "summary": "2026年6月17日Warsh首会以12-0维持利率3.50%-3.75%，但SEP中位点阵从隐含降息翻转为隐含加息（3.4%升至3.8%），18人中9人看至少1次加息、17人看通胀上行风险，2026年PCE上修至3.6%。背景是5月CPI同比4.2%反弹与韧性就业（年内约57万非农）。2年期美债当日跳升逾16基点、DXY破100、离岸人民币滑向6.80出…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CPI", "PCE", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "主要央行保持鹰派按兵不动并非过度收紧的政策错误，而是应对由服务业传导支撑的核心PCE粘性（预计H2维持在3.1%-3.4%）的必要防御，但这会使实体经济承受限制性实际利率压制，支撑久期债防守…", "summary": "本卡综合并支持前三张卡对能源脉冲退潮的判断，但指出核心通胀将出现背离：5月核心CPI放缓至2.9%，但核心PCE因服务二轮传导仍粘在3.4%。由于美联储盯防PCE，鹰派立场并非政策错误。然而CPI下行使限制性实际利率升至0.7%-1.15%，压制增长，支撑久期超配，并导致融资余额达28.13亿元、对海外宏观高敏感的中游机械面临收缩压力，建议低配/规避。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CPI", "PCE", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "核心服务与住房通胀的粘性：美联储政策的后续驱动力", "summary": "能源转为通缩之际，核心服务与住房粘性足以支撑美联储鹰派、抵消能源通缩。关键在指标分化：整体PCE 4.1%/整体CPI 4.2%将因油价基数效应回落，但美联储盯住的核心PCE升至3.4%（2023年10月以来最高），由占核心PCE一半以上、运行约3.5%-3.9%的超级核心（剔除住房的核心服务）驱动。住房同比3.4%（OER 3.3%），去通胀放缓而非结束…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 16.75, "matchedTerms": [ "CPI", "PCE", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "6月点阵图不是简单过时，而是可能“对得不是理由”：能源理由过时，AI/电网与折旧通道支持更高更久，但不足以单独证明立即加息。", "summary": "截至2026-06-28，本卡综合认为：6月点阵图并非被最初的油价冲击证明，因布伦特已回落至约US$72/bbl；但AI/电网导致的结构性电价稀缺、核心PCE 3.4%、名义需求仍强以及AI-capex折旧悬挂，使“更高更久”仍有宏观依据。该通道不足以单独证明立即加息，因为电力CPI权重仅2.489%、PJM冲击区域性强、US$176B折旧低估属低置信度压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CPI", "PCE", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "沃什联储6月鹰派转向：能源供给冲击下的'更高更久、加息可选'与全球外溢", "summary": "6月17日沃什首次主持的FOMC以12-0维持利率3.50%-3.75%，但点阵图转鹰：中值隐含2026年底约3.8%、9/18预计加息、17/18判断通胀上行，并取消前瞻指引。驱动因素是中东伊朗冲突引发的能源供给冲击——霍尔木兹受扰、布伦特一度较战前高逾60%，5月能源CPI同比+23.5%、核心PCE升至3.4%为2023年10月来最高。以核心PCE计…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CPI", "PCE", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "美国5月通胀与6月FOMC：能源冲击下的更高更久利率路径", "summary": "截至2026-06-28，本卡选择global-macro视角研究美国5月通胀与6月FOMC。BEA显示5月PCE同比4.1%、核心PCE同比3.4%；BLS显示5月CPI同比4.2%，能源项同比23.5%；FOMC在6月17日以12-0维持3.50%-3.75%区间，同时SEP上调2026年通胀和利率预测。核心结论是：这不是降息前置信号，而是鹰派耐心。下…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 16.25, "matchedTerms": [ "CPI", "PCE", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "2026-06-28：美国 May 2026 PCE 重新抬高美联储鹰派风险", "summary": "截至2026-06-28，May 2026 PCE 显示 headline PCE 同比4.1%、core PCE 同比3.4%，叠加 June FOMC 维持3.50%-3.75%以及 SEP 上修2026年通胀和联邦基金利率路径，近期降息叙事已被削弱。基准判断是 July 28-29 FOMC 鹰派维持利率，而不是立刻加息；但如果 June CPI 和…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16, "matchedTerms": [ "CPI", "PCE", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "食品", "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-27", "title": "美国通胀再加速与美联储加息风险重估", "summary": "截至2026-06-27，5月 PCE 与 CPI 均显示美国通胀重新偏强，劳动力市场仍有韧性，6月 FOMC 虽维持 3.50%-3.75%，但声明与 SEP 更接近“防通胀持续化”而非“准备降息”。结论是：宏观定价应从“等待降息”切换为“更久高利率并保留加息风险”，但这是能源冲击叠加韧性需求的轻度滞胀框架，不是单纯需求过热。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "CPI", "PCE", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "2026-06-27 美国通胀再加速与 Fed 路径约束", "summary": "截至 2026-06-27，6 月 FOMC 维持 3.50%-3.75% 目标区间，但 SEP 明显上修 2026 年通胀与利率路径；5 月 PCE 与 CPI 均显示通胀再加速，消费者预期虽回落但仍偏高。10 年期美债收益率在 6 月 22 日至 6 月 26 日下行 14.5 bp，更像能源和地缘风险缓和后的 relief move，而非确认的宽松周…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "CPI", "PCE", "macro" ], "href": "" }, { "date": "2026-06-27", "title": "AI驱动的49 GW电力缺口在2028年前物理上无法解决，由此产生的电价上涨已通过云服务链转嫁与规管费率成本回收机制，结构性推高美国核心通胀中枢，构成了美联储更长久限制性利率路径的实物约束。", "summary": "截至2026-06-27，本卡片通过公用事业与电力数据进行压力测试，认为2028年美国49 GW的电力缺口在物理上无法填补，主因是5-7年的并网排队与80-210周的变压器交期。AI带来的电价通胀（如PJM容量价格结构性冲高至$333.44/MW-day）正通过算力转嫁与公用事业输配电价Ratebase成本回收机制传导至核心PCE/CPI，构成结构性供给侧…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "CPI", "PCE" ], "href": "" }, { "date": "2026-06-27", "title": "美联储2026年6月鹰派转向：供给冲击型通胀下的滞胀压力测试", "summary": "2026年6月17日FOMC连续第四次维持利率3.50%–3.75%（12–0全票，Warsh首会），但点阵图把2026年末中值从约3.4%上调至3.8%，抹去年内降息、隐含加息，2026年PCE预测自2.7%上修至3.6%，18人中17人视通胀风险偏上行[S1][S3][S4]。驱动力是供给冲击而非需求：5月CPI达4.2%（2023年4月以来最高），霍…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "CPI", "PCE" ], "href": "" }, { "date": "2026-06-26", "title": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "summary": "压力测试整套对冲架构所依赖的55-65%国内通缩基准概率。结论：基准情形近期稳健，保留但下调至55-60%下沿。硬数据印证消费端通缩：2026年5月CPI同比+1.2%、食品-1.7%、核心+1.1%、环比-0.1%；10年中债贴地板于1.73%[S1][S5]。PBoC“边走边看”渐进（LPR连13个月不变）、财政明确不加码，故货币与财政均为通缩利好而非…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CPI", "食品", "macro" ], "href": "" }, { "date": "2026-06-26", "title": "AI电力冲击本身是Fed应看穿的相对价格变化，但其持续、受供给约束且贴近核心服务成本的二阶传导，会让核心通胀在边际上更粘，并支持Fed更久维持限制性政策。", "summary": "截至2026-06-27，本卡综合前序结论：AI电力不是直接核心通胀项目，因为电力在2026年5月CPI-U权重仅2.495%，且core PCE剔除能源；但它可能通过商业电价、云服务、通信、医疗、设施与租金相关成本向60.158%权重的services less energy services传导。IEA预计全球数据中心用电2030年约945 TWh，美…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "CPI", "PCE", "macro" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "本看板研究链完整、逻辑自洽且数值一致，成功识别了油价持续性与因子拥挤度对 6 月 FOMC 前期布局的关键约束，具备极高的实战参考价值，准予结项。", "summary": "本卡由研究质量主管对第 01-09 卡的研究链进行了终审。审计确认研究逻辑闭环、数值高度一致（特别是 Q1 PCE 4.5% SAAR 与质量因子 +1.7 σ 拥挤度）。研究成功从宏观预测演进为具体的对冲因子价差执行方案。尽管存在对 5 月 14 日 CPI 数据的窗口敏感性，但整体看板已具备极…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 3, "neutral": 0, "value": 25.4, "events": [ { "title": "2026年4月FOMC、Q1 GDP、3月CPI和3月就业数据之后，美国“类滞胀”风险是否升温", "summary": "截至2026年5月3日，美国数据不是衰退型，但软着陆叙事已变得不干净。Q1实际GDP为2.0% SAAR，私人国内购买者实际最终销售为2.5%，显示需求仍有韧性；但3月CPI环比0.9%、同比3.3%，能源环比10.9%、汽油环比21.2%，且Q1核心PCE通胀为4.3%，使Fed难以快速转鸽。3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国通胀再加速与 Fed 在 4 月会议后的反应函数", "summary": "本卡选择 chief-economist，聚焦 Fed 在 2026 年 4 月会议后的反应函数。证据显示，Q1 实际 GDP +2.0%，3 月非农 +178,000、失业率 4.3%，增长端尚未迫使 Fed 降息；但 CPI +3.3% y/y、PCE +3.5% y/y、核心 PCE +3.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hormuz供给冲击在急性期内强烈，但正常化轨迹（4Q26低于$90/b）所产生的基期反转效应使油价对美联储的约束具有时限性——首次降息推迟至2026年Q4而非推入2027年，宽松窗…", "summary": "能源分析师否定油价从结构上封锁美联储2026年宽松窗口的论断。在3月7.5 mb/d、4月9.1 mb/d减产、2Q26 Brent峰值约$115/b、4Q26回落至$90/b以下的正常化情景下，能源分项对标题CPI的同比贡献从2Q26的约+2.3–2.9 pp机械性下降至4Q26的约+0.8–1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 0, "neutral": 1, "value": 47.8, "events": [ { "title": "「类滞胀 Lite + H4L + 区域行 CRE 脆弱性」九卡论文链通过终审，等级为 PASS WITH MINOR REVISIONS：仅需澄清 KRE 42 美元线属 Tier…", "summary": "对九张前置卡片做终审：核心数字（Q1 GDP 2.0%、CPI 0.9% MoM/3.3% YoY、汽油 21.2%、核心 PCE 4.3%、CRE 到期墙 5,700 亿美元、KRE 2026E EPS -12% 至 -18%、办公 CMBS 违约 11.71%、KRE 42 美元支撑）跨卡一致…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 0, "neutral": 1, "value": 49.2, "events": [ { "title": "[auto-handoff] 农业对CPI的贡献", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: b98db943-f28b-4454-9897-5ecade95963f 结果摘要: Wrote handoff_response.md (7556 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "[auto-handoff] 航运成本对农产品贸易的影响", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 0bea90ae-24fb-4edc-9e95-229ae2427057 结果摘要: Wrote handoff_response.md (13436 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 0, "neutral": 1, "value": 52, "events": [ { "title": "TMT行业分析师 · 卡 04 · 2026 05 13 看板: 6a371a6d 12cf 4e7e adc1 c8229ffe8c55 议题: 高拥挤度TMT赛道的杠杆脆弱性与基…", "summary": "TMT行业分析师 · 卡 04 · 2026 05 13 看板: 6a371a6d 12cf 4e7e adc1 c8229ffe8c55 议题: 高拥挤度TMT赛道的杠杆脆弱性与基本面断裂点排查 立场: stress test (自下而上) 工作日期: 2026 05 13 (Asia/Sing…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 3, "neutral": 0, "value": 32.4, "events": [ { "title": "4 月 CPI/PPI 已足以让 Fed 从降息倾向转向对称选择权，但在 PCE/CPI 连续确认前，加息仍是尾部而非基准；曲线若熊陡先压估值，若熊平则更快传导至信用。", "summary": "截至 2026-05-14，本卡支持上一张卡的压力测试方向，但不认为 4 月 CPI/PPI 单独足以让加息成为 Fed 基准情形。CPI +0.6% MoM、+3.8% YoY，核心 CPI +0.4% MoM，PPI +1.4% MoM、+6.0% YoY，核心 PPI +1.0% MoM、+…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力扩张不会单独制造全球通胀再加速，但会放大能源-化肥-粮食冲击，使软着陆路径更脆弱并压低 AI 硬件估值容忍度。", "summary": "2026-05-15：本卡支持 Research note 06 的方向，但将因果强度校准为“AI 是能源-化肥-粮食通胀链条的放大器，而非当前冲击的单一根源”。当前 CPI 冲击主要来自中东能源与航运扰动；但 AI 数据中心用电、局部天然气发电和电网瓶颈提高了能源价格持续性的右尾风险。若尿素维持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "向量 C 下的风险平价重构应把权益/大宗商品风险比从约 3:1 压到约 1.5:1，并同步提高 TIPS、黄金与现金，而不是把组合简单改造成商品多头。", "summary": "截至2026-05-14，本卡对向量 C 做风险平价压力测试：UMich 长期通胀预期 3.4%、NFIB 计划涨价净 27%、4 月 CPI 3.8% y/y 与核心 PCE 3.2% y/y 共同说明，旧的股债分散化假设需要下调。建议将权益风险预算从 35%-40% 降至 25%-30%，名义…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 1, "risks": 0, "neutral": 0, "value": 59.8, "events": [ { "title": "三因素叠加只制造 6-7 月 PCE 商品的可见但有限的二次探头，足以把首次降息从 9 月推迟到 12 月、但不会终止周期，原交易栈在 4-8 周窗口内依然成立，唯需缩短利率敏感腿的…", "summary": "截至 2026-05-17，6/8 货量峰值叠加 $5.60 柴油与运费反弹（FBX01 +34.9%）足以使核心 PCE 商品在 6-7 月单月读数 +8~14 bp（7/31 与 8/29 发布），但 Q3 平均仅 +0~5 bp，关税缓和的永久性通缩传导仍主导半年视角。在通胀预期锚定（5y5…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 2, "neutral": 1, "value": 44.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工标准化是可信的质量分化选股主题，但不是可独立放大的全天候大类资产配置主题。", "summary": "截至2026-05-17，我否定把食品加工标准化主线外推为大类资产配置仓位。前序卡片的微观逻辑仍成立：AI视觉、B端标准化和上游一致性可把利润份额推向龙头。但中国4月食品CPI仍为-1.6%同比，非制造业新订单44.3、销售价格48.1，说明终端传导弱；同时PPI、油价和FAO食品价格上行更像投入…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "供应端结构性缺口足以支撑能源价格中枢上移和通胀粘性增强，但130美元油价本身仍需要扰动持续或恢复慢于预期。", "summary": "截至2026-05-18，本卡片支持能源价格高企具有供应端结构性基础。短期130美元布伦特仍依赖海湾扰动和库存快速下降，但油气田递减率、上游资本开支不足、备用产能变薄、LNG供应浪潮推迟，以及炼厂/中间馏分油瓶颈，均指向更高的能源价格底部和更大的上行尾部风险。BLS 4月CPI、世界银行能源价格预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 0, "neutral": 2, "value": 62, "events": [ { "title": "云端CapEx挤出引发的存储通胀让美/欧/日换机曲线弯折但不击穿，对头条消费增长拖累仅5-10bp，真正的宏观信号是核心商品CPI再加速给联储/欧央行的政策路径增加摩擦。", "summary": "截至2026-05-19，云端CapEx虹吸HBM/CoWoS导致端侧设备ASP上涨50-100美元的冲击是真实的，但不会扼杀美/欧/日换机周期。三条结构性驱动（疫情装机潮到期、Windows 10 EOL企业换机、AI能力门槛）和高端买家低弹性把结果压成'量平/结构上移/ASP升'格局。对美+欧…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "存储通胀是触发条件而非决胜因子；最后50bp推迟与否取决于劳动力松弛是否继续，而欧央行比联储更易被该组合推入2027年。", "summary": "截至2026-05-19，AI存储通胀对核心PCE+5-8bp、核心HICP+4-7bp，单独看落在月度噪音内，不足以把联储/欧央行最后50bp宽松推迟到2027年。但叠加2026年下半年住房与核心服务基数翻面（机械贡献15-20bp）与能源基数转中性后，年底核心PCE可能漂至2.4-2.5%。决…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 0, "neutral": 1, "value": 60.4, "events": [ { "title": "[auto-handoff] 夏粮收购与CPI食品项传导", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 7ce12d71-196d-49a5-b3eb-2244267d05d7 结果摘要: Wrote handoff_response.md (2957 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-21", "supports": 1, "risks": 3, "neutral": 1, "value": 51.8, "events": [ { "title": "中国PPI回升是上游成本推动而非需求拉动，PPI-CPI传导受阻将压低下半年再通胀弹性，并使人民银行更倾向流动性与结构性工具而非激进降息。", "summary": "截至2026-05-22，本卡支持前序判断：中国4月PPI同比2.8%与CPI同比1.2%的裂口主要来自上游能源、有色和化工，而生活资料PPI仍为-1.0%、食品CPI为-1.6%，显示成本压力被困在中游利润链条。地产、零售和民间投资偏弱限制PPI向CPI传导，也限制PPI在下半年的回升弹性。人民…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "4月农产PPI(-2.1%)与下半年CPI回升(1.5%-2.0%)的预期差，确立了2026 H2食品加工龙头盈利扩张的‘第二段窗口’，支持‘成本红利’逻辑。", "summary": "截至2026年5月21日，4月宏观数据显示食品产业链呈现显著的成本与终端价格脱钩。PPI总指数受能源驱动升至2.8%，但农副产品购进价格同比下降2.1%，提供了实质性的“上游让利”。随着2026年下半年猪肉周期反转和CPI预计回升至1.5%-2.0%，食品加工龙头将进入“温和提价+成本低位”的盈利…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "食品加工整板块在2026 H2 A股风格轮动中应作为防御性价值底仓配置，速冻预制菜与烘焙原料少数子赛道则按进攻性GARP卫星配置，二者通过70/30哑铃式组合实现。", "summary": "工作日2026-05-21。在2026 H2 PPI总指数2.8%、农副产品PPI -2.1%、CPI回升至1.5%-2.0%、生猪温和反转、粮油糖全球过剩的背景下，食品加工板块整体β属于防御性价值，但板块内部高度分化。建议采用哑铃式70/30配置：70%底仓配置乳制品、肉制品、调味龙头作为类债红…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 2, "neutral": 6, "value": 74.2, "events": [ { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Re-estimate supply chain price shock transmission to U.S. core PCE", "summary": "Re-estimate supply chain price shock transmission to U.S. core PCE In 'Thesis 1: Step 4', direct a high-priority handoff to global-macro to re-estima…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 粮价偏弱对 5—6 月 CPI 食品分项的拖累幅度", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 7ab279c6-33ab-4e87-a10d-a7933dd72b79 结果摘要: Wrote handoff_response.md (9551 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI Capex韧性 × 油价冲击 × 联储延宕的共振，通过工业PPI与电力催生海外隐性二次通胀，将10Y美债锚定在5.0–5.3%，断裂点在日元/JGB而非美国风险资产。", "summary": "在2026-05-23工作日期下，AI硬件Capex韧性与霍尔木兹油价冲击的共振真实存在，但传导通过工业PPI、电力价格与资本品平减指数发生，而非头部CPI/PCE。这一通胀路径让联储基于PCE的反应函数系统性低估真实通胀，将10Y美债锁定在5.0–5.3%走廊延续整个2026下半年（期限溢价回升…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 49.6, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI capex 支撑真实，但在粘性通胀和高实际利率下，它不足以触发小盘/高贝塔的广谱可投资轮动。", "summary": "截至2026-05-23，首席经济学家否定“AI财报兑现后资金可顺畅转向小盘/高贝塔”的广义结论。AI capex 仍是增长支撑，但美联储仍受通胀约束，CPI/PCE 重新走高，10年期实际利率处于高位，美国增长是韧性而非再加速，中国 PMI 也不足以提供广谱再通胀。组合上应保留现金流已兑现的 A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 3, "neutral": 0, "value": 45, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-25：4月通胀冲击与5月FOMC纪要之后的美国政策路径重定价", "summary": "截至2026-05-25，Fed更像按兵不动或加息尾部风险，而不是确定的降息交易。4月CPI、3月PCE、Q1 GDP和5月FOMC纪要共同显示：通胀仍高、增长仍稳、劳动力只是温和降温。5月22日10Y美债4.56%、30Y美债5.07%，说明长端利率仍在为通胀与期限溢价索价。关键检验点是2026…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "USD 17 bn 农业订单是情绪级别变量，在任何资产负债表数量级上都不解决问题；居民资产负债表的约束变量是住宅地产。", "summary": "截至 2026-05-24，从房地产视角否定线索的残留框架：USD 17 bn 美国农产品订单对居民资产负债表的影响在任何可度量数量级上都不成立。RMB 122 bn 仅相当于居民总资产的 0.022%、住房贷款余额的 0.33%；而 2026 年 1–4 月房地产开发投资同比 -13.7%、住宅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 1, "risks": 1, "neutral": 0, "value": 66.4, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-05-25，A股大消费与农业食品应从“通缩防御”转为“质量定价权+B2B效率+粮食安全资本开支”的组合框架。", "summary": "截至2026-05-25，本卡综合前序结论：平台流量挤压真实存在，但CPI、PPI和政策证据不足以把A股食品与农业配置锚定为宏观通缩交易。建议将食品农业主动袖珍组合调整为30%B2B食品供应、25%优质必选消费、25%粮食安全基础设施、15%农业科技、5%平台低价餐饮/弱零售期权。核心是买入可验证…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 3, "neutral": 0, "value": 47.8, "events": [ { "title": "[auto-handoff] 6月CPI食品分项与粮价传导", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: d9a2a94c-0ae8-49dc-97b7-8a001cf5a141 结果摘要: Wrote handoff_response.md (3347 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高利率维持叠加通胀再加速预期下，防御策略应放弃经典60/40拉久期思路，转向'短久期高质量固收+实物资产（黄金、能源）+质量/现金流权益+T-bill现金'的杠铃式组合，并显式低配长…", "summary": "在美联储6月预计维持3.50%–3.75%、核心CPI同比2.8%且面临二次加速的背景下，主要风险已从衰退尾转为'粘性通胀+股债正相关'。建议防御性配置为：权益42%（质量/价值/能源/防御性，低配长久期成长与小盘）、固收28%（1–3年美债+5年TIPS+短端IG，低配长久期美债与HY）、商品1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "OPEC+延期减产+俄罗斯出口结构性下滑+美国页岩弹性失效+闲置产能集中沙特，共同将Brent推向$100基准情形；通过四阶传导可在8月把整体CPI同比抬回4.0%以上，核心CPI回…", "summary": "2026-05-27卡片判断：Brent现货$96-$98，未来6-8周首次站上$100是基准情形（约55%）。四条供给绳索锁定上行：OPEC+自愿减产延至2026年底、沙特财政平衡油价已升至$96、俄罗斯海运出口因$50价格盖板降至3.0 mb/d、美国页岩2026年增量降至+0.4 mb/d、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 2, "neutral": 0, "value": 55.2, "events": [ { "title": "[auto-handoff] 粮价向 CPI 食品分项的传导节奏与货币政策含义", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 01435eaa-7884-47e7-bf3c-bce3acd3b9c8 结果摘要: Wrote handoff_response.md (19363 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 2, "neutral": 1, "value": 59.6, "events": [ { "title": "材料价格确认的是通胀粘性和利润率压力，而不是美国消费实际销量的全面走强，因此应维持选择性零售多头、避免激进拉长久期，并把下一步焦点转向CPI/PCE传导。", "summary": "截至2026-05-28，材料链条支持“消费韧性仍在、但并非全面实际销量繁荣”的综合判断。4月零售和BofA刷卡仍强，但ISM Prices、钢厂产品、有色金属轧制品、工业化学品和油价相关输入显示成本推动更突出。钢、有色、化工和电池材料共同指向：名义零售可维持，但毛利率、低收入可选消费和Fed通胀…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 小麦减产 + 黑海定价上行对中国食品 CPI 的传导", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: c99dd0a3-0ca9-46aa-9369-265a7c295247 结果摘要: Wrote handoff_response.md (16642 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "虽然整体能源价格走弱缓解了CPI名义压力，但中间投入与柴油成本的滞后传导将维持核心通胀高粘性，迫使新上任的沃什（Warsh）联储在6月维持3.50%-3.75%的高利率，并将降息至少…", "summary": "截至2026-05-28，我们量化测算了原材料及能源价格对核心通胀的滞后传导效应。虽然5月油价回落，但中间品价格高企（ISM支付价格达65.5%），对核心货物传导弹性为0.22（滞后4-6个月），对服务传导弹性为0.15（滞后2-3个月），三季度核心CPI/PCE仍显粘性。5月22日宣誓就职的新主…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 0, "value": 58, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源价格重心的下移提供了一次缓和全球“滞胀”压力的正向供给冲击，虽然成功消解了美欧央行的强行加息风险，但在低储蓄率、物流滞后与-273 bps中美巨幅倒挂的刚性约束下，全球主要央行在…", "summary": "本报告推演了Brent原油重心下调至82-92美元对三季度全球通胀与货币政策的边际影响。受汽油直接回落与航煤暴跌（裂解价差收窄至18-22美元/桶）间接拖累，三季度美国整体CPI与欧洲HICP将分别下修21 bps和30 bps。这有效移除了美联储年底加息选项，但受制于2.6%极低储蓄率与核心PC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 4, "neutral": 0, "value": 47.4, "events": [ { "title": "消费者承担接近 100%、企业短期承担约 64%、CNH/MXN 几乎零吸收且 VND 远不足以对冲，关税新边际压力在 2026 下半年已显著消退但水平被嵌入；CNH 反向走强使中国…", "summary": "量化关税传导：截至 2026-04-08，美国有效关税税率 11.8%（Yale Budget Lab），核心商品 PCE 已上行 3.1%、对核心 PCE 通胀贡献 +0.8pp，传导 7 个月内基本完成；Dallas Fed 确认 2026Q1 见顶，4 月核心商品 CPI 几乎走平。短期分担…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在国际农资价格高企与下游食品通缩的双重剪刀差挤压下，中游主打清洁标签的食品龙头由于刚性锁定天然原料，将面临极具分化的二次成本重估冲击：涪陵榨菜具有深厚的财务防御空间，而安琪酵母与千禾…", "summary": "本卡基于2026-05-30宏观能化剪刀差背景，量化评估了国内农资价格向国际收敛对中游清洁标签食品加工龙头的二次成本重估压力。研究表明，在下游食品CPI负增（-1.6%）制约下，调味品行业财务防御极限分化严重：涪陵榨菜凭借51.6%高毛利展现极强防御性；而安琪酵母在极端价格并轨下净利将转负（亏损9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "油价持续高位将迫使美联储在2026年下半年保持鹰派立场，推升10年期美债收益率至4.75%并使期限溢价突破100bps。", "summary": "评估布伦特原油维持在105-120美元/桶对美联储政策及美债收益率的冲击。高油价将直接贡献headline CPI约1.3-1.5%，并推升核心PCE至3.4%以上。美联储预计将维持3.75%-4.00%的高利率，甚至实施防御性加息。受通胀不确定性、沃什提名政权更迭及地缘溢价驱动，预计10年期美债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 1, "value": 69.8, "events": [ { "title": "As the Global Macro Analyst ( global macro ) , I have conducted a rigorous macroeconomic re…", "summary": "As the Global Macro Analyst ( global macro ) , I have conducted a rigorous macroeconomic regime stress test on the multi asset hedge overlay proposed…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "若6月跨太平洋现货运价继续上行但TEU只温和增长，应优先解读为船公司定价和成本冲击，零售利润率先受压，CPI/PCE传导选择性且滞后。", "summary": "截至2026-05-31，本卡综合支持“成本挤压而非需求繁荣”判断。Drewry在2026-05-28显示WCI升至每40ft集装箱2,800美元，跨太平洋航线同时受blank sailings和旺季附加费支撑；NRF则预计2026年上半年进口12.59百万TEU、同比仅增0.5%。美国4月零售销…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 1, "risks": 2, "neutral": 1, "value": 73.2, "events": [ { "title": "2026H2总量宏观β因下游需求收缩与上游输入通胀双重挤压而深陷逆风，食品加工及大众消费大盘应全面收紧战线，增配B2B配料，重配现金防御，撤退B2C品牌。", "summary": "截至2026-06-01，最新宏观数据显示下半年消费大盘面临逆风：4月社零仅增0.2%（创2022年12月新低），CPI-PPI剪刀差恶化至-1.6%，食品CPI下跌1.6%导致下游B2C品牌定价权丧失。我们对投资组合权重进行最终定调：超配成本锁定且渠道免疫的B2B配料（24%），减配受两头挤压的…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国消费总量仍有频次韧性，但“量增价缩”更利好具备后端工业化控本能力的经营者，而非广谱消费复苏。", "summary": "2026-06-01判断：中国消费不是坍塌，而是“量增价缩”。出行、餐饮、网上服务仍有频次韧性，但CPI食品项、旅游人均花费和居民消费支出增速显示客单价与消费意愿偏弱。拼好饭和9.9元咖啡的可持续性来自标准套餐、集单配送、中央厨房、冷链和供应链规模，而不是无限补贴。结论是支持“总量有韧性”，但只在…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "“量增价缩”会把消费投资从总量复苏逻辑推向定价韧性筛选：高端白酒保留质量底仓但需下调批价期权，休闲食品只买能同时证明毛利率、费用率和渠道现金健康的龙头。", "summary": "截至2026-06-01，本卡对前序“量增价缩”主线做消费行业压力测试。结论是：服务消费修复并不等同于品牌消费复苏，社零、CPI与PPI共同显示平价渠道扩张、食品饮料终端价格偏弱和成本转嫁缺口。高端白酒仍有质量溢价，但茅台一季度毛利率和净利率已下行，批价相对官方自营价缓冲偏薄；次高端与区域白酒库存…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 5, "neutral": 0, "value": 45.6, "events": [ { "title": "消费双轨制已在2026年4月社零/CPI数据中得到宏观印证，A股消费板块估值中枢的结构性下修不可逆，但B端供应链龙头与服务消费龙头将获得相对再定价红利。", "summary": "2026年4月社零+5.1%、服务零售+7.6%、CPI食品−1.0%/服务+0.6%的组合，从宏观总量数据印证了消费'双轨制'结构性成形。下半年实际GDP预计4.7%–4.9%，但平减指数仍为负，名义增长持续滞后实际。A股消费板块估值中枢已较2019–2021永久性下修35%–48%：传统C端品…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Temu/Shein退潮后的美国折扣零售受益应被定义为名义份额回收；若2H26实际商品PCE转负且低收入客流走弱，消费回流和广告竞价池替代会同步失效。", "summary": "截至2026-06-02，本卡对Research note 01和Research note 02做压力测试而非完全否定。美国折扣零售更可能承接Temu/Shein退潮后的名义份额，而非低收入家庭实际商品需求的完整回流。4月实际DPI环比-0.5%、实际PCE仅+0.1%，叠加CPI同比3.8%、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 9.1%, 22/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 9.1%, 22/241） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.5%, 18/241） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "2026 年通胀粘性、温和为正的实际利率与非对称的央行反应函数足以让铜、银、电力的稀缺性溢价持续 12-24 个月进入跨资产定价模型，构成结构性而非阶段性交易。", "summary": "截至 2026-06-04，核心 PCE 3.3%、CPI 3.8%、10Y TIPS 实际收益率 ~1.8-2.3%、美联储数据依赖的缓慢降息姿态共同构成一个'实际利率为正但不严苛、政策反应非对称'的制度。叠加关税传导、AI 与电气化资本开支超级周期、劳动力与服务通胀粘性三股多年期供给侧楔子，未…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 7, "neutral": 1, "value": 39.4, "events": [ { "title": "6月四个宏观窗口最可能给出温和支撑组合，足以维持卡3配置中枢但不足以将食品饮料升级为2026 Q3趋势主线；上下行概率虽对称但触发条件不对称，期望意义上偏下，正确仓位是不上调、按触发…", "summary": "2026-06-04：6月四个宏观窗口（PMI、CPI食品分项、PBOC例会、618/端午高频）在工作日期均为前瞻。本卡通过事前情景建模认为最可能场景是\"温和支撑、不证伪、不升级\"：制造业PMI区间49.3–49.7、食品CPI同比-0.8%~-1.3%、PBOC维持现状、618食品类目同比+5%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026 Q3 PPI与CPI宏观约束达成的概率较高，但输入型成本通胀与弱内需导致的PPI-CPI剪刀差扩张将对食品饮料板块毛利率构成挤压，不支持将其仓位全面上调为季度Beta主线。", "summary": "本卡针对2026 Q3 PPI同比收窄至-1.7%以上、CPI站稳1%以上的宏观约束进行压力测试。结论表明，虽然基数效应使这两项约束大概率能够达成，但PPI受外部地缘政治驱动而走高，属于输入型成本推动，而CPI受制于内需磨底（如生猪去化缓慢）。这会导致正向PPI-CPI剪刀差拉大，对中下游食品饮料…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 4, "neutral": 1, "value": 58.8, "events": [ { "title": "[auto-handoff] 芽麦饲用替代对三季度饲料成本及CPI的传导评估", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 9d17c76d-7b0d-4d17-a3fc-eed0e697d1ae 结果摘要: I will first run the shell command to anchor today's date, …", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "离岸汇率闸门放行情景下国内货币政策空间与粮食安全资本开支传导", "summary": "假设离岸FX闸门打开，国内政策仍无法把粮食安全投资、高标准农田、农村融资与农业科技升级转化为广义信用脉冲，传导停留在防御层面。三道国内闸门收紧：国有行NIM跌至约1.42%历史新低（工行约1.3%）、LPR钉死3.0/3.5%[S4][S1]封顶风险放贷；高标准农田为多年期工程（2030年13.5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "消费行业正从防御性品种转为‘隐形石油税’的利润受害者，高端白酒的财富效应溢价面临重估，而大众消费品正进入优胜劣汰的毛利保卫战。", "summary": "本卡片合成了能源冲击与贸易壁垒对消费行业的双重压力。研究显示，消费行业充当了‘隐形石油税’的避震器，表现为食品饮料行业因物流、包装和原材料成本上升而导致的剧烈利润挤压（PPI与CPI背离）。白酒行业面临‘财富效应蒸发’导致的结构性回调，而非简单的成本压力。此前看好的‘海外布局’战略在CBAM与全球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "农业链条没有提供通胀、需求或政策层面的正向催化来确认成长风格切换，粮食安全更应被纳入防御型哑铃配置而非高beta追涨依据。", "summary": "截至2026-06-05，农业视角支持否认“红利→成长”regime break。4月食品CPI仍是拖累项，猪肉和鲜菜供应充足；USDA FAS对2026年生猪、猪肉和MY2026/27饲料需求的判断偏稳定，粮食安全政策更多提供长期防御型结构主题，而非午盘成长股追涨信号。农业可作为实物红利/防御型…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 3, "neutral": 1, "value": 74.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 10%, 24/241） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-06美国就业与通胀组合对美联储路径和全球利率的影响", "summary": "截至2026-06-06，5月非农新增172,000人、失业率4.3%、工资同比3.4%，显示就业降温但未断裂[S1]；4月CPI同比3.8%、核心PCE同比3.3%，仍限制美联储快速转向[S2][S3]。报告判断6月会议前不宜追逐激进降息交易，除非6月10日5月CPI明显降温[S6][S7]。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "面对'劳动降温+核心粘性'的非对称宏观，正确答案不是单一押注，而是质量股票+短端carry+精选长端的杠铃，并把黄金作为滞胀尾部保险。", "summary": "在5月就业降温但核心通胀仍粘（NFP +172k、失业率4.3%、核心PCE 3.3%）的组合下，建议总体股票风险中性、股票内偏质量（超配+3至+5 ppt）、固定收益采用'短端carry+精选长端'杠铃（现金/短期美债+4 ppt、1–3年IG+2 ppt、10年UST+1 ppt、10年TIP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 2, "neutral": 2, "value": 76.6, "events": [ { "title": "[auto-handoff] 铜价高企背景下家电龙头企业与二三线品牌成本裂谷对工业品PPI与制造业投资的传导评估", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 536b5150-44d9-4c3b-b53a-4f495cbccec3 结果摘要: Wrote handoff_response.md (11657 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "通胀对冲组合压力测试与跨板块配置的合规性与叙事一致性终审", "summary": "对前9卡链条的终审：分析主线扎实——内部自洽、引用规范、概率树（48/22/25/5）谱系清晰、布伦特中枢95-98美元与80美元地板、关税14%→9.6%、7月USTR §301（07-06评论/07-07听证）、11-10稀土悬崖等关键数字跨卡一致。但有两项阻断级缺陷：card-07与card…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高油价尚未足以确认制造业衰退，但已经让PMI扩张变成一个由能源成本、预防性采购和利润率分化共同支撑的脆弱扩张。", "summary": "截至2026-06-07，本卡对上一张全球宏观结论做压力测试：全球制造业PMI仍扩张，不能确认全面衰退；但$106布伦特已通过燃料、运费、CPI、PPI和高耗能制造业投入端形成真实传导。基准判断是：若油价按EIA路径从5月至6月约$106/b回落至4Q26的$89/b，通胀黏性上升但制造业可避免硬…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月FOMC前的美国类滞胀格局：油价冲击、粘性核心通胀与韧性就业", "summary": "4月CPI同比3.8%/核心2.8%、5月非农超预期+17.2万、布伦特从$117峰值回落至$95——这套组合让联储6月16-17日按兵于3.50-3.75%几乎成为定局。能源的反通胀顺风已消失,工资增速3.4%仍高于与2%核心PCE相容的水平。10Y美债已升至4.55%,曲线定价'联储卡住'。基…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 78, "events": [ { "title": "能源冲击已足够大到支撑美联储鹰派沟通，但是否转化为实际加息，取决于6月海峡恢复是否延误以及二次通胀效应是否扩散。", "summary": "截至2026-06-08，霍尔木兹冲击足以支撑美联储鹰派预期，但更偏向“压制降息、保留加息风险”而非立即加息。海峡正常承载近20 mb/d石油出口和约19%全球LNG贸易[S1]，绕行能力仅3.5-5.5 mb/d[S1]；IEA显示4月全球供应较2月累计下降12.8 mb/d[S2]。美国4月C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 2, "neutral": 0, "value": 73.4, "events": [ { "title": "白糖进口严控会抬高下游用糖成本底部，但在食品PPI偏弱和主要农产品价格受调控的环境下，它更可能压缩糖果、饮料、烘焙利润率，而不是显著推升食品CPI。", "summary": "截至2026-06-10，本卡综合前序糖业供给压力测试，判断白糖输入性压力更像食品饮料企业的利润率冲击，而不是食品CPI的主驱动。国际原糖约14美分/磅、广西现货约5290-5350元/吨，叠加配额和糖浆/预混粉监管，使国内糖价有政策底但上行受库存压制。糖果、全糖饮料和大众烘焙的毛利敏感度依次递减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 2, "value": 86.8, "events": [ { "title": "[auto-handoff] 慢去产能周期下CPI食品项及H2通胀中枢敏感性评估", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 21e65805-b96d-4e1e-8071-581e93f19615 结果摘要: I will run the command 系统日期锚 to anchor today's date as requ…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "生猪产能去化路径及其对2027年通胀的非线性传导", "summary": "生猪产能去化路径及其对2027年通胀的非线性传导 Hog Capacity De-stocking Path and Nonlinear Inflation Transmission 当前猪价跌破现金成本是否会触发规模化猪企的恐慌性断崖式出清？这会否导致2027年H1出现大幅反弹并推高CPI通胀中…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 0, "neutral": 1, "value": 91.2, "events": [ { "title": "医药'去中国化'是一股结构性、慢释放的供给侧通胀力量，每年向美国核心医疗CPI注入25–40bp，并已迫使美联储将生物科技融资环境纳入金融稳定框架。", "summary": "截至2026-06-12，《生物安全法案》及1260H清单升级已将中美医药脱钩转化为可测量的供给侧成本冲击。按模态加权，非中替代溢价合计抬升受影响SKU零售价4.5–6.5%，对应美国核心医疗服务CPI每年额外+25–40bp、Medicare Part B支出增速每年高出基线60–90bp，20…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-12", "supports": 1, "risks": 1, "neutral": 0, "value": 91.6, "events": [ { "title": "宏观政策与收入预期有限地、结构性地支撑Research note 02判断：功能性食品10-20%溢价更可承接（高收入30%+银发+备案制三因子对齐），预制菜5-12%条件性溢价天花…", "summary": "锚定2026-06-13，中国2026Q1居民可支配收入实际增速+5.0%、4月CPI+0.3%但食品CPI-1.0%、城镇失业率5.1%。政策面以旧换新和服务消费补贴未直接落到食品大类，但保健食品备案扩围、预制菜国标与服务消费意见同向支撑功能化转型。借鉴日本FOSHU与机能性表示食品在1995-…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 2, "neutral": 0, "value": 87, "events": [ { "title": "结构性金属供给紧缩对全球通胀及美联储政策路径的影响", "summary": "支持（有界定）：铜/铝结构性缺口驱动的二次通胀足以硬化美联储终端利率逻辑，但机制是抬高底部、消除2026年降息期权，而非单独迫使加息。LME铜2026年6月初约13,800美元/吨、1月29日创14,527.50美元/吨历史高位，高盛上调年底预测至13,735；铝突破3,015.50美元/吨、对应…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月FOMC前的美国通胀再加速", "summary": "截至2026-06-13，美国5月CPI同比4.2%、核心CPI同比2.9%，PPI同比6.5%，非农就业增加172,000人且失业率为4.3%[S1][S2][S3]。结论是6月FOMC更可能按兵不动但沟通更鹰，原因是核心CPI不支持立即加息，但PPI管线压力、PCE高于目标和劳动力市场韧性使近…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 3, "neutral": 0, "value": 74.4, "events": [ { "title": "能源CPI与核心CPI背离下，美联储应上调通胀预测并取消2026年降息，但6月仍以鹰派按兵不动为基准。", "summary": "截至2026-06-14，能源冲击要求美联储显著上调2026年总体PCE预测，但核心上修应更温和。5月CPI同比4.2%、核心2.9%，加上4月PCE 3.8%、核心PCE 3.3%，说明压力真实但尚未全面扩散。基准建议是6月维持3.50%-3.75%，把2026年点阵图从一次降息改为不降息；若布…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海运费是通胀右尾输入项，不是足以单独决定Fed路径和A股仓位的宏观锚。", "summary": "截至2026-06-15，本卡否定“仅凭海运费即可重写2026年下半年Fed路径”的强结论。Drewry WCI上涨至3,549美元确实抬高通胀右尾，但Freightos FBX为2,888.80美元，二者存在22.9%的指数基差；IMF研究显示运费传导有约一年峰值时滞。当前Fed更应关注CPI、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 1, "risks": 2, "neutral": 0, "value": 89.8, "events": [ { "title": "2026年下半年宏观环境呈现“CPI疲软与餐饮韧性并存”特征，大众刚需板块在“价弱”格局中，依靠收入与餐饮端的“量稳”基本盘叠加高毛利壁垒，“利润可守”的防御逻辑得到宏观数据的坚实验…", "summary": "2026年6月最新数据显示CPI/PPI剪刀差扩大（5月PPI同比3.9%，CPI仅1.2%），食品CPI下降1.7%，印证了下游消费端“价弱”现实。然而，一季度居民实际可支配收入同增4.0%，且1-5月餐饮收入同增5.0%，远超整体社零表现（4月仅0.2%），表明大众刚需具备极强的“量稳”特征。…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] 美国CPI/PPI背离对大类资产风险溢价及下半年美股板块轮动影响研判", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: f4668e4f-748a-467a-bfe5-e36469d590db 结果摘要: Wrote handoff_response.md (10465 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月FOMC与美国通胀再加速下的higher-for-longer风险", "summary": "截至2026-06-15，global-macro的基准判断是Fed仍大概率按兵不动，但不是鸽派暂停。5月CPI、5月PPI、4月PCE和5月就业数据共同显示名义需求仍偏强，能源冲击正在抬高总量通胀，而核心项尚未失控。因此，6月FOMC更可能通过预测和措辞完成higher-for-longer重置…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 2, "neutral": 0, "value": 83.2, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国5月通胀反弹与6月FOMC前的全球宏观交易含义", "summary": "截至2026-06-15，本卡选择global-macro视角。结论是：6月FOMC大概率维持利率不变，但声明会让近期降息门槛更高。5月通胀主要由能源拉动，核心CPI环比仍温和；但4月核心PCE仍高、5月就业不弱、居民通胀预期仍偏高，使美联储很难快速转鸽。对市场而言，关键不是单纯押注利率上行，而是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 1, "neutral": 1, "value": 93.6, "events": [ { "title": "I have completed the quantitative stress testing for the B2B catering supply chain raw mate…", "summary": "I have completed the quantitative stress testing for the B2B catering supply chain raw material costs and gross margin elasticity in 2026Q2 Q3. Key F…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 1, "risks": 2, "neutral": 1, "value": 96, "events": [ { "title": "电网公用事业资产费率基数扩张对宏观通胀与货币政策的传导效应", "summary": "压力测试卡片01–05的宏观结论。直接渠道被夸大：电力属能源、按构造排除于核心PCE之外，达拉斯联储测算数据中心对总体PCE的拉动2026仅+0.05pp、2030升至+0.13pp[S1]；电力仅占CPI约2.5%[S11]，自测算直接总体贡献约0.10pp。但费率基数是30–40年价格水平锁定…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "饲料粮通缩对CPI食品分项是再通缩力量而非通胀缓冲，政策响应将是‘托麦不托玉米’组合，对总CPI同比拖累约−0.08至−0.16pct。", "summary": "芽麦替代封顶玉米 + 巴西大豆到港洪峰构成饲料粮双料通缩。直接传导（粮食、食用油）对CPI拖累不足0.04pct；主航道是间接传导——料价下行不会推高猪价，反而通过压栏增重与延迟去产能强化猪肉再通缩，预计2026年6–9月对总CPI同比拖累−0.08至−0.16pct[自测算]。政策端是‘托麦不托…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "CPI下行打开战术宽松窗口，但只有地产、就业和信用需求共同恶化，才会把PBoC推入实质性降息/降准循环。", "summary": "截至2026-06-19，CPI下行给PBoC提供宽松许可，但不足以单独触发实质性降息/降准循环。5月CPI同比+1.2%、核心CPI+1.1%、食品CPI-1.7%，而PPI+3.9%，价格信号是结构性分化而非全面通缩。地产投资、销售面积和销售额分别同比-16.2%、-10.8%、-13.5%，…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 2, "neutral": 0, "value": 87.4, "events": [ { "title": "2026-06-21：6月FOMC鹰派按兵不动与全球美元利率重定价", "summary": "截至2026-06-21，本卡选择global-macro，研究6月17日FOMC以12-0维持3.50%-3.75%但SEP转鹰后的宏观含义。结论是：这不是清晰的新加息周期，而是一次鹰派信誉重置；降息叙事已被移除，一次额外加息成为有条件风险。5月CPI、4月PCE、稳定的4.3%失业率、2年期美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储6月会议后的“降息交易”出清与再通胀风险", "summary": "工作日期2026-06-20。本卡选择global-macro，结论是美联储尚未承诺加息，但6月FOMC已把市场基线从自动降息转向higher-for-longer和加息风险。3.50%-3.75%的政策区间、2026年SEP中3.8%的年底利率中位数、5月CPI同比4.2%、5月非农+172,0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 11, "neutral": 2, "value": 40.8, "events": [ { "title": "伊朗战争后的全球宏观转折：能源冲击消退，但美联储与日本央行维持鹰派带来的滞胀宿醉", "summary": "2026年伊朗战争（2/28美以空袭、3/4霍尔木兹关闭）将布伦特推至逾120美元，6/17美伊谅解备忘录使油价回落至约80美元。但通胀已从能源迁至核心：5月美国CPI总体同比4.2%（三年新高），核心升至2.9%，居住与食品双双加速。央行反应函数已上移——美联储6/17维持3.50–3.75%但…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在强势美元与国际谷物低价的双重作用下，低廉的进口谷物成为我国抵御输入性通胀的屏障，同时强劲的经常项目顺差为人民币汇率及国内流动性稳定提供了重要的宏观缓冲。", "summary": "本报告分析了截至2026-06-22，全球谷物低价与强势美元对中国宏观经济的传导机制。强势美元（USD/CNY为6.77）带来约4%的进口汇率拖累，但国际粮价大跌使得进口依然享有高达30-40%的价格优势。低名义进口账单与强劲出口推动5月贸易盈余达1054.3亿美元，有力防御了中美273.8bp的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年伊朗战争能源冲击：供给扰动见顶与美联储/日本央行的滞后鹰派转向", "summary": "伊朗战争能源冲击大概率已过顶：布伦特从逾120美元回落至6月19日的80.57美元（周跌约8%），霍尔木兹有条件复航[S6][S8]。但美联储6月17日维持利率3.50–3.75%却将点阵图翻转为加息倾向（2026中值升至3.8%、PCE预测3.6%），日本央行6月16日加息至1.0%（1995年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年伊朗战争能源冲击重新定价全球利率周期：通胀是暂时性还是持久性", "summary": "持续15周的2026年伊朗战争制造了一次供给侧能源冲击，把发达市场利率周期从降息扭转为鹰派按兵不动+加息倾向：美联储6月17日维持3.50%–3.75%但点阵图翻转为加息（2026年底中值升至3.8%、PCE上调至3.6%），日本央行加息至1.00%，美国5月CPI同比4.2%创近三年新高。但布伦…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 3, "neutral": 0, "value": 84.2, "events": [ { "title": "美联储 6 月再定价：美国前端利率从降息预期转向维持高利率并存在加息风险", "summary": "截至 2026-06-23，6 月 FOMC 将联邦基金目标区间维持在 3.50%-3.75% [S1]，并把 2026 年 PCE 与核心 PCE 预测分别上调至 3.6% 和 3.3% [S2]。5 月 CPI 同比 4.2% [S3]、4 月核心 PCE 同比 3.3% [S4]、5 月非农…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年6月美联储暂停、能源通胀与美国利率更久维持高位风险", "summary": "截至2026-06-22，本卡选择global-macro视角研究美联储6月暂停。核心发现是：FOMC虽维持3.50%–3.75%目标区间，但6月SEP上调2026年联邦基金、PCE和核心PCE预测；同时5月CPI能源项、4月PCE、5月就业与EIA油市预测共同削弱近期降息叙事。结论不是立即加息，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 4cd2d717-9d04-4569-b0d8-cc37d743d0a9 结果摘要: Wrote handoff_response.md (9414 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 2, "neutral": 0, "value": 91.6, "events": [ { "title": "美联储非对称紧缩把周期判断从“增长放缓触发宽松”改写为“通胀未再锚定前，增长疼痛不足以恢复Fed put”。", "summary": "截至2026-06-23，本卡支持卡片03：美联储已进入对通胀上行更敏感、对增长下行更迟钝的非对称紧缩体制。6月FOMC维持3.50%-3.75%，但SEP同时下调2026增长、上调PCE通胀；5月CPI 4.2%、4月PCE 3.8%、密歇根一年期预期4.6%，说明通胀预期仍未安全再锚定。OEC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 3, "neutral": 1, "value": 96, "events": [ { "title": "[auto-handoff] WTI at $65-68/bbl and Fed path threshold", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 2f15cf64-5f49-4af5-867a-05b15b121ef3 结果摘要: Wrote handoff_response.md (8806 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年能源通胀冲击后的美联储政策", "summary": "截至2026-06-25，6月FOMC维持3.50%-3.75%并非鸽派信号。能源冲击已把5月CPI推至4.2%同比，但核心CPI仍为2.9%同比，劳动力和国内需求尚未断裂。基准判断是美联储更可能维持高利率而非马上加息；真正的分水岭是未来PCE/CPI能否证明能源没有传导到核心价格和通胀预期。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply s…", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 0, "neutral": 1, "value": 96, "events": [ { "title": "[auto-handoff] 中美采购协议下物理进口量放大对国内玉米与大豆价格冲击测算", "summary": "致 房地产分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 9c54a258-e27a-4b47-b5ed-1963d9f8bb71 结果摘要: Wrote handoff_response.md (15140 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 4, "neutral": 2, "value": 89.8, "events": [ { "title": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "summary": "压力测试整套对冲架构所依赖的55-65%国内通缩基准概率。结论：基准情形近期稳健，保留但下调至55-60%下沿。硬数据印证消费端通缩：2026年5月CPI同比+1.2%、食品-1.7%、核心+1.1%、环比-0.1%；10年中债贴地板于1.73%[S1][S5]。PBoC“边走边看”渐进（LPR连…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力冲击本身是Fed应看穿的相对价格变化，但其持续、受供给约束且贴近核心服务成本的二阶传导，会让核心通胀在边际上更粘，并支持Fed更久维持限制性政策。", "summary": "截至2026-06-27，本卡综合前序结论：AI电力不是直接核心通胀项目，因为电力在2026年5月CPI-U权重仅2.495%，且core PCE剔除能源；但它可能通过商业电价、云服务、通信、医疗、设施与租金相关成本向60.158%权重的services less energy services传…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "同步鹰派转向是真实的风险窗口，但不是已被证实的新紧缩制度；Research note 9 字面论点应被否定，前序宏观主线应被有条件保留。", "summary": "截至2026-06-26，本卡否定 Research note 9 字面交接文本：它是目录检查类工作流伪影，不是宏观论点。但压力测试后有条件支持前序经济主线：G3 6月鹰派转向真实发生，Fed 维持3.50%-3.75%、ECB升至2.25%、BoJ升至1.0%；美国5月CPI/PCE仍高，说明不…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观需求疲弱足以削弱但不足以完全否定食品加工链的成本通缩红利，配置应从板块β降级为精选大众必选品阿尔法。", "summary": "2026-06-27：本卡对“成本通缩+量端复苏”做宏观需求压力测试。5月社零同比-0.6%、商品零售-0.7%、餐饮仅+0.6%，显示广泛量端复苏尚未确认；食品CPI、农副食品和食品制造PPI确有通缩，但整体购进价格、燃料动力和化工原料仍上行。利润数据进一步分化：食品制造利润仍增，农副食品加工和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 7, "neutral": 1, "value": 70.2, "events": [ { "title": "沃什联储6月鹰派转向：能源供给冲击下的'更高更久、加息可选'与全球外溢", "summary": "6月17日沃什首次主持的FOMC以12-0维持利率3.50%-3.75%，但点阵图转鹰：中值隐含2026年底约3.8%、9/18预计加息、17/18判断通胀上行，并取消前瞻指引。驱动因素是中东伊朗冲突引发的能源供给冲击——霍尔木兹受扰、布伦特一度较战前高逾60%，5月能源CPI同比+23.5%、核…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国5月通胀与6月FOMC：能源冲击下的更高更久利率路径", "summary": "截至2026-06-28，本卡选择global-macro视角研究美国5月通胀与6月FOMC。BEA显示5月PCE同比4.1%、核心PCE同比3.4%；BLS显示5月CPI同比4.2%，能源项同比23.5%；FOMC在6月17日以12-0维持3.50%-3.75%区间，同时SEP上调2026年通胀…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026-06-28：美国 May 2026 PCE 重新抬高美联储鹰派风险", "summary": "截至2026-06-28，May 2026 PCE 显示 headline PCE 同比4.1%、core PCE 同比3.4%，叠加 June FOMC 维持3.50%-3.75%以及 SEP 上修2026年通胀和联邦基金利率路径，近期降息叙事已被削弱。基准判断是 July 28-29 FOMC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 3, "neutral": 1, "value": 95.6, "events": [ { "title": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "summary": "2026年6月17日Warsh首会以12-0维持利率3.50%-3.75%，但SEP中位点阵从隐含降息翻转为隐含加息（3.4%升至3.8%），18人中9人看至少1次加息、17人看通胀上行风险，2026年PCE上修至3.6%。背景是5月CPI同比4.2%反弹与韧性就业（年内约57万非农）。2年期美债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "主要央行保持鹰派按兵不动并非过度收紧的政策错误，而是应对由服务业传导支撑的核心PCE粘性（预计H2维持在3.1%-3.4%）的必要防御，但这会使实体经济承受限制性实际利率压制，支撑久…", "summary": "本卡综合并支持前三张卡对能源脉冲退潮的判断，但指出核心通胀将出现背离：5月核心CPI放缓至2.9%，但核心PCE因服务二轮传导仍粘在3.4%。由于美联储盯防PCE，鹰派立场并非政策错误。然而CPI下行使限制性实际利率升至0.7%-1.15%，压制增长，支撑久期超配，并导致融资余额达28.13亿元、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核心服务与住房通胀的粘性：美联储政策的后续驱动力", "summary": "能源转为通缩之际，核心服务与住房粘性足以支撑美联储鹰派、抵消能源通缩。关键在指标分化：整体PCE 4.1%/整体CPI 4.2%将因油价基数效应回落，但美联储盯住的核心PCE升至3.4%（2023年10月以来最高），由占核心PCE一半以上、运行约3.5%-3.9%的超级核心（剔除住房的核心服务）驱…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "6月点阵图不是简单过时，而是可能“对得不是理由”：能源理由过时，AI/电网与折旧通道支持更高更久，但不足以单独证明立即加息。", "summary": "截至2026-06-28，本卡综合认为：6月点阵图并非被最初的油价冲击证明，因布伦特已回落至约US$72/bbl；但AI/电网导致的结构性电价稀缺、核心PCE 3.4%、名义需求仍强以及AI-capex折旧悬挂，使“更高更久”仍有宏观依据。该通道不足以单独证明立即加息，因为电力CPI权重仅2.48…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 2, "neutral": 0, "value": 96, "events": [ { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-d4473d35e5/", "networkHref": "/investor/research-network/#thesis-d4473d35e5" }, { "id": "thesis-bef6966fdd", "source": "tracked_thesis", "title": "工业机械的出口订单质量和设备更新周期决定制造业 CapEx 弹性", "titleEn": "Industrial machinery beta depends on export-order quality and equipment-upgrade capex", "lane": "工业机械/设备更新出海", "laneEn": "Industrial machinery, equipment upgrade, and exports", "laneHref": "/investor/lanes/industrial-machinery-equipment-upgrade-and-exports/", "investableFocus": "工业机械/设备更新出海：600031.SH, 000425.SZ, 000157.SZ, 603338.SH, 600761.SH, 300124.SZ, CAT", "stockUniverse": [ { "ticker": "600031.SH", "name": "三一重工", "market": "A-share", "role": "工程机械出口", "roleEn": "工程机械出口", "bucket": "core", "href": "/investor/stocks/600031.SH/" }, { "ticker": "000425.SZ", "name": "徐工机械", "market": "A-share", "role": "工程机械全球化", "roleEn": "工程机械全球化", "bucket": "core", "href": "/investor/stocks/000425.SZ/" }, { "ticker": "000157.SZ", "name": "中联重科", "market": "A-share", "role": "工程机械与高机", "roleEn": "工程机械与高机", "bucket": "core", "href": "/investor/stocks/000157.SZ/" }, { "ticker": "603338.SH", "name": "浙江鼎力", "market": "A-share", "role": "高空作业平台出海", "roleEn": "高空作业平台出海", "bucket": "core", "href": "/investor/stocks/603338.SH/" }, { "ticker": "600761.SH", "name": "安徽合力", "market": "A-share", "role": "叉车与工业车辆", "roleEn": "叉车与工业车辆", "bucket": "core", "href": "/investor/stocks/600761.SH/" }, { "ticker": "300124.SZ", "name": "汇川技术", "market": "A-share", "role": "工控自动化", "roleEn": "工控自动化", "bucket": "core", "href": "/investor/stocks/300124.SZ/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" } ], "stockTickers": [ "600031.SH", "000425.SZ", "000157.SZ", "603338.SH", "600761.SH", "300124.SZ", "CAT", "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH" ], "coreView": "赛道：工业机械/设备更新出海。候选观察标的：600031.SH, 000425.SZ, 000157.SZ, 603338.SH, 600761.SH, 300124.SZ, CAT。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 工业供给瓶颈。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 73, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 8, "risks": 144, "neutral": 8, "analystCount": 13, "sourceTypeCounts": { "研究笔记": 147, "证据更新": 1, "分析协作": 10, "深度研究": 1, "研究主题": 1 }, "topTerms": [ "GOES", "AIDC", "设备", "CapEx", "Industrial", "machinery", "beta", "depends", "export-order", "quality", "equipment-upgrade", "capex" ], "whatChanged": [ "2026-06-29: [auto-handoff] 先进制程星载 OBP 芯片良率瓶颈与算力/通信设备材料供应链", "2026-06-29: 上游取向硅钢和电工铜的产能稀缺与关税高墙正转嫁为严重成本通胀，侵蚀设备制造厂毛利并拉长交付，导致AI电力落地面临实质性溢价风险。", "2026-06-29: 电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "2026-06-29: [auto-handoff] 中压干式变压器出海企业海外合规及交付溢价测算", "2026-06-29: 电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀" ], "practical": { "sectorLane": "工业机械/设备更新出海", "sectorLaneEn": "Industrial machinery, equipment upgrade, and exports", "exposureBucket": "电力与公用事业", "tone": "yellow", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "600031.SH", "name": "三一重工", "market": "A-share", "role": "工程机械出口", "roleEn": "工程机械出口", "bucket": "core", "href": "/investor/stocks/600031.SH/" }, { "ticker": "000425.SZ", "name": "徐工机械", "market": "A-share", "role": "工程机械全球化", "roleEn": "工程机械全球化", "bucket": "core", "href": "/investor/stocks/000425.SZ/" }, { "ticker": "000157.SZ", "name": "中联重科", "market": "A-share", "role": "工程机械与高机", "roleEn": "工程机械与高机", "bucket": "core", "href": "/investor/stocks/000157.SZ/" }, { "ticker": "603338.SH", "name": "浙江鼎力", "market": "A-share", "role": "高空作业平台出海", "roleEn": "高空作业平台出海", "bucket": "core", "href": "/investor/stocks/603338.SH/" }, { "ticker": "600761.SH", "name": "安徽合力", "market": "A-share", "role": "叉车与工业车辆", "roleEn": "叉车与工业车辆", "bucket": "core", "href": "/investor/stocks/600761.SH/" }, { "ticker": "300124.SZ", "name": "汇川技术", "market": "A-share", "role": "工控自动化", "roleEn": "工控自动化", "bucket": "core", "href": "/investor/stocks/300124.SZ/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" } ], "stockTickers": [ "600031.SH", "000425.SZ", "000157.SZ", "603338.SH", "600761.SH", "300124.SZ", "CAT", "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH" ], "primaryStocks": [ { "ticker": "600031.SH", "name": "三一重工", "market": "A-share", "role": "工程机械出口", "roleEn": "工程机械出口", "bucket": "core", "href": "/investor/stocks/600031.SH/" }, { "ticker": "000425.SZ", "name": "徐工机械", "market": "A-share", "role": "工程机械全球化", "roleEn": "工程机械全球化", "bucket": "core", "href": "/investor/stocks/000425.SZ/" }, { "ticker": "000157.SZ", "name": "中联重科", "market": "A-share", "role": "工程机械与高机", "roleEn": "工程机械与高机", "bucket": "core", "href": "/investor/stocks/000157.SZ/" }, { "ticker": "603338.SH", "name": "浙江鼎力", "market": "A-share", "role": "高空作业平台出海", "roleEn": "高空作业平台出海", "bucket": "core", "href": "/investor/stocks/603338.SH/" }, { "ticker": "600761.SH", "name": "安徽合力", "market": "A-share", "role": "叉车与工业车辆", "roleEn": "叉车与工业车辆", "bucket": "core", "href": "/investor/stocks/600761.SH/" }, { "ticker": "300124.SZ", "name": "汇川技术", "market": "A-share", "role": "工控自动化", "roleEn": "工控自动化", "bucket": "core", "href": "/investor/stocks/300124.SZ/" } ], "instrumentTags": [ "600031.SH", "三一重工", "000425.SZ", "徐工机械", "000157.SZ", "中联重科", "603338.SH", "浙江鼎力", "600761.SH", "安徽合力", "300124.SZ", "汇川技术" ], "reviewCadence": "每日复核", "confirmationSignals": [ "并网排队和变压器交期是否继续拉长", "表后自发电与 PPA 是否改变公用事业信用指标", "电力设备订单质量是否高于收入确认节奏" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "长端利率", "铜/GOES 成本", "监管电价", "信用利差" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 93 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 13, "recentEvents": 93, "supportRatio": 0.05, "riskRatio": 0.9, "conflictScore": 100, "evidenceDepth": 100 }, "score": 64.3 }, "events": [ { "date": "2026-06-29", "title": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "summary": "本卡片从制造和供应链视角对电网与电力约束进行压力测试。截至2026-06-29，关键电力设备（变压器与重型燃机）由于极高工艺与测试门槛、上游取向电工钢（GOES）瓶颈及寡头垄断，交期拉长至3-7年，生产机位已售罄至2029-2030年。云厂商BTM发电加剧了燃机机位争夺。预计2026-2027年该瓶颈将导致云厂商约800亿-1100亿美元资本开支递延兑现。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "GOES", "设备", "Industrial", "industrial" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]，PPI-CPI剪刀差隔离了政策利率。任务设想的'成本→CP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.5, "matchedTerms": [ "GOES", "设备", "beta" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 中压干式变压器出海企业海外合规及交付溢价测算", "summary": "致 主题研究员： 协同任务已完成。执行方为 工业制造分析师。 模式: adhoc 会话: f0fa9b36-39ce-428c-a4ff-9c7d5180b4ad 结果摘要: Wrote handoff_response.md (11248 bytes); wrote handoff_response.en.md (12258 bytes) — This …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 16, "matchedTerms": [ "bottlenecks", "margins", "automation", "export", "overseas", "margin" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "summary": "本卡压力测试 Research note 6 的'调价条款已隔离利润'结论，认为其仅有条件成立。主流调价条款（MEC）挂钩有流动期货的铜/铝（出口超级周期转嫁率约80%[S1]），却普遍排除或封顶无流动指数的取向电工钢GOES——而GOES占COGS约28%[Research note 6]且处2026-2028结构性短缺。自测算显示：铜+30%/GOES…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GOES", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "上游取向硅钢和电工铜的产能稀缺与关税高墙正转嫁为严重成本通胀，侵蚀设备制造厂毛利并拉长交付，导致AI电力落地面临实质性溢价风险。", "summary": "压力测试表明，取向硅钢与电工铜的供应缺口已完全固化在变压器与燃机的长交期中，不构成独立的“二次时间瓶颈”。但地缘摩擦与美国50%的Section 232关税，导致中美GOES价差拉大至5.16倍，美国面临严重的材料合规溢价。在重度成本冲击下，固定价格合同变压器厂毛利率收缩至-3.67%，逼迫出口链向东南亚与墨西哥转移（临界关税率13.16%）。此外，5 G…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GOES", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 先进制程星载 OBP 芯片良率瓶颈与算力/通信设备材料供应链", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 81384248-8464-45e1-8194-06f6bed14dff 结果摘要: Wrote handoff_response.md (10343 bytes); wrote handoff_response.en.md (10719 bytes) — Th…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GOES", "设备", "orders" ], "href": "" }, { "date": "2026-06-28", "title": "变压器交期拉长至3-5年与并网排队超36个月的物理刚性瓶颈，配合美欧贸易保护关税对GOES硅钢供应链的阻断，已对AI算力扩张的通电效率构成无法用资本单纯解决的长期结构性壁垒，造成明显的Cap…", "summary": "基于并网排队与变压器短缺的双重物理约束，我们验证电力基础设施瓶颈已构成AI算力扩张的长期结构性阻碍。数据中心通电时间已拉长至36-60个月，变压器交期延至3-5年，主要厂商积压订单超百亿美元。这些延误对项目NPV和IRR产生重大侵蚀，100MW数据中心延误3年将损失51.63%的NPV。美国Section 301/232关税与欧盟CBAM机制则通过GOES…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 18, "matchedTerms": [ "GOES", "设备", "CapEx", "capex" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "summary": "本卡支持 Research note 02「上游原材料是最深约束」的结论，并锐化为一组非对称结构。GOES：自2019年变压器需求暴增（电力+119%、GSU+274%），远超GOES市场约5.4–5.6%的CAGR；GSU供给缺口2025年约100%，因美国唯一GOES生产商Cleveland-Cliffs（Butler Works，扩产2028上线）及…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.5, "matchedTerms": [ "GOES", "设备", "land" ], "href": "" }, { "date": "2026-06-28", "title": "关键矿产与设备出口管制风险对A股高端环节溢价逻辑的冲击评估", "summary": "压力测试出口管制威胁。结论：把镓/锗/锑/稀土剧本套到宝钢与紫金上是双向错配，不反转 Research note 06 基准论断。铜/紫金：中国是全球最大铜进口国（2025 精矿进口 30.31Mt，国内矿仅约 1.85Mt，依存约 87%），出口武器结构上不存在；紫金约 50% EBITDA 来自海外矿（刚果金、塞尔维亚、哥伦比亚），真正风险是入境制裁/…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.5, "matchedTerms": [ "GOES", "设备", "land" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备成本传导逻辑的脆弱性：铜与电工钢供应瓶颈能否被设备商有效对冲？", "summary": "本卡用材料视角压力测试2027-2028年铜与GOES供需缺口对HV变压器利润率的侵蚀确定性。结论：成本传导逻辑脆弱，但脆弱是非对称、取决于合同结构的，而非全行业确定。GOES是确定性最高、最难对冲的咽喉——区域单一来源（美国仅Cleveland-Cliffs，本土覆盖约1/5需求）、约占LPT成本25%、自2020年约翻倍、无期货市场、铝替代是技术陷阱[…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.5, "matchedTerms": [ "GOES", "设备", "land" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备链是可投资的多年超级周期，但赢家来自高压瓶颈和积压订单质量，2029年后需警惕燃气轮机与边际电气产能的消化风险。", "summary": "截至2026-06-29，本卡对Research note 04做压力测试并部分支持：AI电力瓶颈已变成燃气轮机、大型变压器、开关设备和施工能力的工业产能问题。GE Vernova燃气轮机积压/预订已到100GW、目标年末110GW，Siemens Energy积压1540亿欧元，变压器交期仍在128-144周。结论是：这不是单纯2026订单尖峰，而是真实…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.5, "matchedTerms": [ "设备", "Industrial", "beta", "industrial" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力基础设施供应链瓶颈：电力变压器与关键配电硬件的产能缺口", "summary": "对Research note 01以变压器为中心的瓶颈论断做压力测试。结论：GIS与高压断路器产能确实紧张（≥145 kV断路器交期已超12个月、整套GIS需60–100周、供给集中于约4家OEM），但它是次级协同瓶颈，而非近期核电并网的绑定约束。原因在于结构：近期核电对AI的供给主力是反应堆重启与增容，二者均复用既有开关站，断路器/GIS短缺在发电侧几乎…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GOES", "Industrial", "industrial" ], "href": "" }, { "date": "2026-06-28", "title": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "summary": "本卡压力测试逻辑链：GOES约束不会「抵消」核能加速，因为二者作用于电力链不同节点——核能重启在发电/并网侧本就是GOES-轻（复用既有GSU与开关站），属于一种GOES套利；但GOES在负荷侧与输配路径重新绑定，沿用Research note 01的NPV模型测算，核能锚定型项目(≈1.5年延迟,NPV≈−27%)相对并网型绿地(3年,−51.6%)只回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GOES", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "中国GOES与铜体系在基准情形下是西方AI建设的泄压阀，但政策阀门已被安装，西方数据中心应把中国来源依赖重定价为150-300bp的分层资本开支风险溢价。", "summary": "截至2026-06-28，本卡压力测试GOES与铜的中国供给集中度。结论：基准情形不是中国对GOES/铜禁运，而是“带政策期权的泄压阀”。中国具备西方AI数据中心建设所需的GOES与铜精炼/冶炼规模，但关税、原产地、反倾销与出口管制先例会把供给集中度转化为资本开支风险溢价。建议在Research note 04的100-200bp材料/设备溢价上，对中国来…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GOES", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "功率半导体已过剩，AI电力的新物理瓶颈已迁移至大型变压器，并真正卡在无法被资本快速绕行的上游电磁钢(GOES)与铜——2026–2028窗口难解。", "summary": "支持并收紧前五卡逻辑链：功率半导体转入结构性过剩后（Research note 05），AI电力建设的刚性瓶颈已迁移至大型变压器(LPT)，并下沉到上游电磁钢(GOES)与铜。瓶颈呈两层：整机厂总装层可用约20亿美元北美扩产去瓶颈，但缓解要到2028/2029年后；标准电力变压器交期128周、GSU 144周、定制订单4–5年，美国机队缺口约30%[S1]…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 20.5, "matchedTerms": [ "GOES", "设备", "Industrial", "industrial" ], "href": "" }, { "date": "2026-06-27", "title": "变压器材料和交付瓶颈已开始改变 Hyperscaler 的2026-2027资本开支与电力采购节奏，影响主要体现为预算通胀、长期锁电和通电交付递延，而非AI需求取消。", "summary": "截至2026-06-28，结论为支撑：Hyperscalers 已经通过预算上修、长期电力合同和项目节奏调整来应对变压器及电气设备瓶颈，但公开披露通常不会把变化直接归因为 GOES 或铜关税。Microsoft 2026 capex 中约250亿美元来自更高组件价格，Meta 上调2026 capex中点100亿美元，Amazon 把电力容量作为部署KPI…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 18, "matchedTerms": [ "GOES", "设备", "CapEx", "capex" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "工业设备交付能力已成为AI基建的终极制约，利润分配正从计算层向具备垂直集成能力和监管特许权的工业巨头（GEV, VRT, BWXT）转移。", "summary": "2026年AI电力瓶颈已从芯片转向物理设备。燃机与核电部件受限于5年以上排产，而变压器面临120周以上的极端交期。GE Vernova与BWXT凭借寡头地位与监管壁垒展现出极强的利润防御性；Vertiv则受益于液冷参考架构的强制性应用，利润率快速扩张。中国企业如特变电工在海外非美市场占优，但面临美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 2, "neutral": 1, "value": 34.4, "events": [ { "title": "电气设备供应链的'物理墙'与先进制造约束", "summary": "电气设备供应链的物理瓶颈因电压等级呈非对称分布：345 kV及以上LPT交期高达36至60个月，全球可信制造商仅3至4家，约束近乎存在性；配电层面模块化预制变电站已商业成熟，可压缩进度40%至60%。增材制造、固态变压器、MVDC配电各有适用时间窗口，无单一技术能彻底突破。近期最可行路径为：WEG…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2H26 订单簿已经覆盖 AI 物理供应链的整柜、液冷、光模块和部分国产设备，但 EPS 缺口集中在高毛利瓶颈件而非最大收入钱包。", "summary": "本卡支持 Research note 1 的 AI 资本开支延续判断，但强调供应链利润转化分化。2H26 订单可见度已覆盖服务器 ODM/整柜、液冷、800G/1.6T 光模块和国产刻蚀/沉积/清洗设备；其中光模块和液冷组件 EPS 弹性最好，国产半导体设备次之。ODM 收入可见度强但毛利穿透弱，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "原材料定价权已达到顶点：AI驱动的电网超级周期正迫使云巨头绕过公开市场，通过直接入股或高溢价长协锁定GOES钢和铜的‘物理供应’。", "summary": "AI电网建设正在将GOES取向硅钢和铜从周期性商品重塑为战略性科技资产。2026年市场面临结构性缺口，定价权已从设备商转移至原材料生产商。Cleveland-Cliffs凭借北美垄断和垂直整合成为核心受益者；铜价受AI高密集度（27-47t/MW）驱动，AWS与力拓的直采协议标志着云巨头开始绕过市…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "2026-2028年的关税和贸易限制会先加剧而非缓解AI基建的“铁”瓶颈，因为美国保护的国内GOES和变压器产能不足以快速替代高依赖进口链条。", "summary": "本卡对第06卡做压力测试并支持“硅-铁失配”判断：2026-2028年贸易政策不会从零制造变压器短缺，但会把短缺转化为地缘政治分配问题。DOE数据显示，美国LPT进口依赖度为82%，本土GOES只能满足12-20%需求，GOES和铜导线各约占LPT成本25%。Section 232在2026年4月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "13万亿元财政扩张不会造成政策相关工业板块的广义信贷挤出，但会形成“双速信用环境”：电网和获批技改被挤入，民营中小制造业资本开支被边际挤出。", "summary": "本卡压力测试13万亿元财政扩张对工业信贷的挤出风险。结论是：总量信用尚未紧到足以全面挤出政策相关工业资本开支，但信用配置明显分化。电网设备、UHV、变压器、电网自动化和列入设备更新目录的技改项目更可能被财政资金和银行配套融资“挤入”；民营中小企业、非目录机械和弱订单资本品则面临边际信贷收紧和盈利可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-09", "supports": 1, "risks": 0, "neutral": 0, "value": 54.2, "events": [ { "title": "2026年初，西门子能源（北卡夏洛特）和伊顿（德州纳科多奇斯）投产新电力设备产能", "summary": "2026年初，西门子能源（北卡夏洛特）和伊顿（德州纳科多奇斯）投产新电力设备产能", "sourceLabel": "证据更新", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-13", "supports": 1, "risks": 3, "neutral": 0, "value": 37.6, "events": [ { "title": "工业制造分析师报告：电力设备厂商订单结构、海外交付能力与利润率分化 工作日期： 2026 05 13 分析师： industrials analyst 主题： 将AI驱动的电网需求通…", "summary": "工业制造分析师报告：电力设备厂商订单结构、海外交付能力与利润率分化 工作日期： 2026 05 13 分析师： industrials analyst 主题： 将AI驱动的电网需求通过订单质量与海外产能转化为可持续利润 立场： synthesize (综合) 截至2026 05 13，电力设备行业…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理基础设施周期的赢家是长交期机电OEM（LPT/GIS/HV电缆/航改型燃气轮机/液冷参考架构），真正瓶颈已迁移到取向电工钢、铜杆/母线、大型铸锻件与熟练技工——按订单覆盖度×…", "summary": "截至2026-05-13，本卡支持卡片01–04的'密度胜过足迹'主线：AI CapEx美元从外壳转向机电与电网侧含量。具备结构性订单可见度与定价权的OEM集中于长交期、寡占环节——Hitachi Energy（6501 JP）、Siemens Energy、GE Vernova、Eaton、Sc…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备 capex 链最深的稀缺租金在上游 GOES 与高纯阴极铜，整机的超额利润是借来的窗口；产业链利润呈“上游 ≥ Tier-1 整机 > Tier-2 整机 > 公用事业”的…", "summary": "支持卡 03 的“电力设备 > 公用事业”框架，但从原材料端补强：取向硅钢 (GOES) 是真正的单点瓶颈，宝钢/首钢合计 >55% 全球高端 Hi-B 份额，2026Q1 现货价同比 +38%、宝钢 GOES 分部毛利率达 36% 的 10 年高点；电解铜进入结构性赤字 (ICSG 2026 缺…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026H2 应把 AI 主题 gross 重构为 5 层多头 + 2 层空头 + 1 层 12% 跨主题对冲嵌入式组合，在 GOES/铜见顶情景下通过 G4-G7 闸门将上游 L1…", "summary": "综合卡 01-04 四层利润结构，给出 2026 下半年 AI 主题多空篮子重构方案：gross 96%、net long 由 +60% 降至 +48%，分为 5 层多头 (上游材料 14% / Tier-1 算力 20% / 国产算力 12% / Tier-1 电力设备 10% / 平台软件 1…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 3, "neutral": 0, "value": 31, "events": [ { "title": "电气设备模组商15-18%向22-25%的毛利扩张在24个月内可兑现+500~+700bp，22-24%为高概率基准，25%为stretch case；card-07的+300~+5…", "summary": "本卡支持card-07的电气设备超配框架。验证三条扩张路径：1）订单能见度18-24个月可独立证实（北美四大云厂商USD330B CAPEX对应USD26-33B电气外采，国内宁夏/内蒙/贵州三大算电节点25Q4-26Q1招标11.4 GW）；2）毛利从15-18%向22-25%扩张由结构升级(+…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国和欧盟不会消灭中国电力设备出海周期，但会把发达市场收入从高质量利润释放项变成需要政策折价的订单。", "summary": "截至2026-05-14，本卡压力测试但不推翻AI电网设备出海逻辑。全球变压器和电网设备短缺仍支撑中国供应商订单，但美国和欧盟正在用关税、采购韧性、网络安全、外国补贴审查和零部件贸易救济压缩发达市场利润率。美国风险最高，Section 301/232、Section 122临时附加费和结构性产能过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "材料底座（GOES吨数+重型锻造槽位）在2027-2029年把美国AI可交付电力压在12-17 GW/年，与电网与OEM约束重叠确认10-14基准、14-18高情景的硬上限，价值进一…", "summary": "截至2026-05-14，本卡对前序结论进行压力测试并基本确认：即便变压器与燃机OEM在2026-2028年兑现30-50%扩产，取向硅钢（GOES）、高导铜、大型铸锻件与关税/出口管制会在2027-2029年重新把美国AI可交付电力压在12-17 GW/年的材料可行区间——与第2/5卡的10-1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 2, "neutral": 0, "value": 38.4, "events": [ { "title": "[auto-handoff] A-share New Productive Forces theme: crowding and policy-support boundary", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 17939c81-1aba-4588-815a-9997769cc0e7 结果摘要: Wrote handoff_response.md (9758 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A股新质生产力主题拥挤度与政策护盘边界", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: d01808a8-e00c-4167-b2da-ac7d96eae236 结果摘要: Wrote handoff_response.md (3975 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "AI 算力需求增速放缓将通过订单节奏先于价格传导至先进制造与电力基础设施环节的盈利预期，工业电力链是二阶动能回落的最高 Beta 接收端。", "summary": "支持卡片 01 提前部分减仓的判断，并指出工业设备 / 电力基础设施环节是 AI 资本开支二阶动能回落的高 Beta 接收端。北美超大规模数据中心建设热度 -6.4% WoW 是变压器、开关柜、液冷、母线槽等环节的领先变量，订单确认滞后破土 2-4 个季度，而该群体客户集中度恶化、利润依赖价格抬升…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 1, "risks": 4, "neutral": 0, "value": 37.2, "events": [ { "title": "AI电力硬件瓶颈真实存在，但最可投资的不是泛GOES或泛电力设备，而是高压变压器认证产能、测试吞吐量和高等级GOES的订单兑现能力。", "summary": "截至2026-05-18，工业端证据支持将GOES/HVPT升为AI电力主线：NERC显示大型变压器交期80-210周，GE Vernova、Siemens Energy、ABB的订单和在手订单继续快于收入兑现，Hyosung和Hitachi扩产也主要在2027-2028年缓解。关键修正是：不是所…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "支持上轮判断：AI电力交易更应买变压器、变电站和GOES等瓶颈设备/材料，而不是无差别买入久期公用事业。", "summary": "截至2026-05-17，变压器瓶颈仍由产能窗口、合格劳动力和GOES供应共同决定。GE Vernova、Siemens Energy、Eaton的最新订单、积压和利润率显示，头部电力设备商正在把稀缺转化为价格与利润，而不是被GOES和铜成本完全挤压。新产能投资已经启动，但多数是2027-2028…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "容量轮动论点在方向上通过中国宏观压力测试，但需降低对美欧设备多头的确信度、对冲 2026-11 关键矿物悬崖、并按地理与币种分叉半导体设备与长久期信用的判断。", "summary": "中国宏观对资产配置师的容量轮动论点做压力测试：方向上支持（东数西算瓶颈迁移、国网十五五前置 capex、PBoC 偏鸽都印证），但提出三点重大挑战。S1：中国可通过特变电工/思源/金盘等绕道墨西哥-沙特-印尼出口，2026 后压缩美国变压器交期 6–12 个月，打掉 Hitachi/Eaton 终…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Section 232 持续 + CRMA 采购偏好 + 中国出口许可收紧的组合，把 OEM 护城河从临时物理短缺升级为持久行政短缺，净强化 2027–2029 电力设备 OEM 定…", "summary": "三条政策时钟（美国 Section 232 GOES 复审、欧盟 CRMA 二级实施细则、中国出口许可与反倾销周期）叠加，未来 24 个月最可能的路径是碎片化加深但无头条贸易战：Section 232 续期+对最薄 Hi-B 给窄排除（≈55%），CRMA 二级细则 2026Q4–2027Q2 发…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 42.6, "events": [ { "title": "电力设备供应链瓶颈真实存在并卡死 2029 年前新增 AI 园区通电，但卡不住 2024 年已下单的 2026 年超大规模项目；真正的约束点在 GOES 与熟练劳动力，而非变压器本体。", "summary": "针对前序卡片'物理瓶颈优于财务下修'的压力测试：变压器（≥100 MVA）交付周期已拉长至 130–210 周，开关设备 100–140 周，HVDC 60–72 个月，Hitachi Energy、Siemens Energy、GE Vernova 等 OEM 售罄至 2027 年。但 2024…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网队列吞吐、LPT/GOES与HA级燃机三重物理瓶颈在2026-2028窗口内无法被增量Capex解决，电力侧硬数据明确支持把主线从美元承诺切换为已通电MW与队列位次。", "summary": "截至2026-05-18，电力侧硬数据确认第01卡判断：超大规模厂商Capex承诺撞上的是物理约束。美国五大ISO在册队列约1.0-1.1TW但燃气仅占6-7%，2026-2028净增可调度MW约28-36GW，对应EPRI估算的47-58GW AI增量负荷出现明确缺口。LPT交货120-210周…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "能源物流没有被 AI 消灭，但“物理大宗必要性”不等于“能源大宗运输贝塔免疫”，真正稀缺性正在迁移到电力设备、燃机、储能和电网集成。", "summary": "截至 2026-05-19，工业制造视角否定卡片 05 的强版“能源物流战略性免疫”。能源分子仍需大宗运输，但 AI 电力需求首先制造的是电气设备、燃机、储能、开关设备、变压器、并网和控制系统瓶颈，而不是对 VLCC、煤炭铁路或普通港口大宗贝塔的广谱保护。IEA 2026 年报告显示 hypers…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 1, "risks": 1, "neutral": 0, "value": 59.4, "events": [ { "title": "截至2026-05-20，AI电力瓶颈的核心不是全国缺电，而是高规格变压器和高压开关设备的工厂槽位、测试能力、交付周期与原材料可得性。", "summary": "截至2026-05-20，我们支持前序结论：AI算力扩张瓶颈已从全国总发电量转向站点级重电设备交付。Eaton、GE Vernova、Siemens Energy、Schneider、Hitachi、韩国UHV厂商及TBEA/Pinggao均显示订单快于可交付产能。Allianz称电网设备交付最高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Vulcan 商业化最强的工业品投资含义是高价值传感与精密运动部件的结构性订单增量，而不是所有机器人设备的同步扩张。", "summary": "截至2026-05-21，本卡支持上一张卡片的方向，但将传导机制收窄为高价值零部件订单周期：Vulcan 从试点到 beta 的数量跃升、触觉/3D视觉/末端执行器/精密运动控制的高物料含量，以及仓储自动化订单恢复，共同支撑2026年下半年物流自动化零部件订单结构性加速。基准判断为：整体物流自动化…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-21", "supports": 1, "risks": 0, "neutral": 0, "value": 66.8, "events": [ { "title": "财政补贴与再贷款精准降低了高端制造的单位成本并转化为技改投资，在总量产能克制的背景下，驱动高技术与装备制造业实现显著的结构性利润回升。", "summary": "截至2026-05-22，本卡支持前序判断。研究表明，央行5000亿元再贷款及1.5%财政贴息精准转化为技术改造Capex，有效降低了高端制造企业的融资与单位生产成本（每百元营收成本降至84.93元），成功抵消了上游能源与原材料的价格冲击。2026年一季度，在整体工业产能利用率承压（73.6%）的…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 3, "neutral": 1, "value": 45.2, "events": [ { "title": "变压器等电力设备的极度短缺及取向硅钢供应瓶颈验证了“时间尺度错配”，AI数据中心的扩张将持续受制于3-4年的物理电网硬件交付周期。", "summary": "截至2026年5月22日，全球电力设备市场面临极度紧缺。西门子、GE等四大设备商订单积压创历史新高，大型电力变压器交付周期拉长至36-48个月。上游取向硅钢（GOES）受制于北美产能垄断及全球供应不足，成为扩产的核心瓶颈。设备可用性已取代资金，成为AI基建最硬的约束。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器是AI算力部署的第二硬物理约束，液冷是高强度执行约束但更可通过模块化扩产缓解。", "summary": "截至2026-05-23，本卡对AI基础设施物理瓶颈作压力测试：变压器和变电站设备比液冷更可能成为继电力可得性之后的第二个硬约束。配电变压器交期已从2019年的3-6个月拉长至2023年的12-30个月，高容量设备最长可达4年；新增产能虽由Eaton、Hitachi、Siemens等推进，但多在2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GOES 瓶颈会逐步缓释但不会短期消失，2026-2027 年 AI 电力基础设施仍将受制于变压器材料和设备交付周期。", "summary": "截至2026-05-22，本卡压力测试认为全球 GOES 新增产能无法在短期打破变压器交付延迟。中国具备最强近期吨位弹性，Baosteel、Shougang 等高端取向硅钢产能扩张明显；但欧美公用事业认证、贸易风险和高压可靠性要求限制其快速清理西方队列。JFE-JSW 印度 350,000 ton…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 3, "neutral": 1, "value": 48, "events": [ { "title": "As the Industrials Analyst (工业制造分析师) , I have completed the research card assessing the bre…", "summary": "As the Industrials Analyst (工业制造分析师) , I have completed the research card assessing the breadth of the advanced manufacturing capital expenditure (ca…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI基础设施资本开支尚未在订单层面见顶，但在扩产加速度和重资产风险收益上已接近顶部，2027年放缓将优先打击低毛利、高折旧、专用性强的制造环节。", "summary": "截至2026-05-24，本卡从工业制造与Capex Cycle角度应力测试AI基础设施扩产。结论是：已披露需求和行业基准预测尚未显示周期见顶，SEMI仍预计2026-2027半导体设备支出上行；但资本开支加速度、低毛利跟随者扩产和数据中心建设成本已呈现后周期特征。若2027年需求放缓，最严峻的减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-2028年间，电气化重装备及数据中心专用电源设备处于结构性“超强周期”，设备商产能已被预订满至2028年后，导致电网并网时滞成为AI算力上线的实质屏颈，促使重装备板块重估…", "summary": "截至2026-05-24，本报告评估了电力变压器、开关柜及数据中心电源系统在2026-2028年的供需缺口。大型变压器交期拉长至150-200周，因硅钢紧缺，美国严重依赖进口，现代电气和金盘科技订单锁排至2028年。开关柜交期翻倍，Powell与Eaton在手订单创纪录。Vertiv受益于800V…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的部署上限由已通电液冷MW决定，而不是由GPU订单或云厂商capex预算单独决定。", "summary": "截至2026-05-24，本卡支持前序“AI瓶颈物理化”判断，但把瓶颈收窄为三条工业交付链：液冷机架、电气设备和可靠电力。GB300 NVL72机架功率已达132-140 kW，100 MW IT功率在设施损耗前仅支持约714-758个机架；JLL披露2025年57%数据中心项目因供应链延误至少3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "A 股电力设备正从‘国内周期标的’重估为‘全球 AI 基建稀缺核心供货商’，其显著的交期优势与估值洼地支撑权重上调。", "summary": "A 股电力设备龙头正凭借 10-12 个月的极致交期（对比海外 36-48 个月）和对取向硅钢 (GOES) 的产能垄断，成为全球 T&D 瓶颈的唯一‘泄洪口’。国家电网 4 万亿（2026-2030）投资计划及 8500 台具身智能设备招标，确立了 A 股‘AI+电力’的增长底部。目前 A 股龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 7, "neutral": 0, "value": 23.8, "events": [ { "title": "AI电力设备超级周期成立，但2026-2028年的真实约束在变压器、开关柜和交付认证，而不是泛化的SiC晶圆短缺。", "summary": "2026-05-26：本卡片对卡片03的“10倍机柜功率跃迁”做压力测试。结论是需求判断成立，但交付确定性分层：大型电力变压器、干式变压器、中压开关柜和试验能力是2026-2028年的硬瓶颈；SiC晶圆/器件层面受EV放缓影响并非短缺，但高压模块、封装可靠性、栅极驱动和SST系统认证仍限制放量。金…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网扩容的第二瓶颈不是资本，而是重电设备的合格产能槽位及其上游材料。", "summary": "截至2026-05-27，本卡支持前序判断：AI扩张的第一瓶颈是电网可用容量，第二瓶颈正在转向大型电力变压器、开关设备、电缆和专业施工能力。DOE、Wood Mackenzie和JLL数据显示，LPT进口依赖、GOES/CTC材料短缺、128-143周变压器交付周期、44周开关设备周期和数据中心项…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "工业制造专题分析：电力核心设备供应链产能瓶颈压力测试 本报告立足于上游工业制造与重型电气装备供应链的视角，对公用事业分析师（Research note 03）提出的“AI Capex…", "summary": "工业制造专题分析：电力核心设备供应链产能瓶颈压力测试 本报告立足于上游工业制造与重型电气装备供应链的视角，对公用事业分析师（Research note 03）提出的“AI Capex 因公用事业基础设施供给天花板而面临瓶颈”的假说进行了深度压力测试。 1. 核心研究结论 物理供给瓶颈具有绝对刚性：…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美欧原产地规则与中国出口许可在AI电力供应链上呈现70%价格机制、30%数量机制的格局，仅在Hi-B取向硅钢与NdFeB重稀土磁材两节点真正物理改道，其余主要通过推高到岸成本并对美国…", "summary": "美欧'熔炼与浇筑'原产地规则与中国出口许可制是两套高度非对称的工具：前者基于规则、机械执行但被18-30个月执法滞后稀释；后者基于裁量、时延即政策。综合压力测试：2026-2027年这套监管栈是70%的成本推动通胀税（AI电力设备到岸成本+20-35%，美国核心商品CPI+15-25bp）、30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 3, "neutral": 0, "value": 49.2, "events": [ { "title": "重型电力设备OEM收入上能承接需求、时间上承接不了——订单簿2.5–4年可见度+定价权+利润率扩张构成对科技久期的反向对冲，核心超配GEV/ENR.DE/ETN/Hitachi系，中…", "summary": "2026-05-28结论：本卡支持第03卡的电网瓶颈论断。Siemens Energy Grid Tech订单/营收比2.28×、储备490亿欧元；GE Vernova Electrification订单+86%、设备储备390亿美元、B2B约2.5×；Eaton电气储备同比+48%、数据中心订单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 电力缺口是真实但温和的核心 PCE 楔，叠加电网设备供应链的中美对抗，将以「延迟降息 + 类滞胀」形态而非「重置通胀目标」形态影响美联储，并把宏观风险从拥挤的一阶 AI 硬件转…", "summary": "AI 电力交付滞后通过三条渠道（批发电价→服务、云算力涨价→SaaS、变压器 PPI→capex 平减指数）对美国核心 PCE 形成 2026 +8~18 bp、2027 +10~22 bp 的持续抬升（自测算），不足以重置 2% 目标但足以让核心 PCE 2027 年底仍卡在 ~2.4%，相对 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "把AI久期空头与电网OEM/GOES/CLF多头拼成的组合净Beta+0.45、净久期−1.5年，是'高利率持续'路径下的反AI久期+实物瓶颈租金套利，需配置USD多头FX叠加并设置…", "summary": "2026-05-28 synthesize：把第01–05卡拼回组合（USD 100mm基币）。AI久期空头−15%、AI硬件多+10%、全球电网OEM+30%、亚洲OEM+10%、GOES上游+8%、CLF垂直一体+7%、自动化+5%、现金+25%。净Beta约+0.45、净久期约−1.5年（受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "电网配套capex温和扩张被高端制造结构升级与海外溢价放大为头部设备公司的高弹性订单与毛利改善——超配特高压主设备、配网自动化、储能PCS与出口变压器，回避陷入价格战的光伏组件与风机…", "summary": "截至2026-05-28，本卡支持card-06能源分析师关于'电网与配套才是真正瓶颈'的判断，并从设备端给出量化传导。判断2026-2028是电网设备'弱周期+强结构+海外溢价'叠加的景气区间：国网+南网2025 capex约8200亿元，2026继续增长5-10%；特高压主设备(640-880…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 3, "neutral": 0, "value": 52, "events": [ { "title": "关键设备交付周期拉长至2.5-3年以上，且受限于GOES与铜的全球结构性短缺，电力设备产业链的产能扩张短期内无法匹配AI数据中心爆发式需求，行业物理瓶颈为设备OEM厂商（GEV、VR…", "summary": "截至2026年5月29日，在AI算力与电网升级需求爆发下，电力设备产业链（GEV、VRT）面临严重物理产能瓶颈。2026年变压器交付周期拉长至128-144周。虽然各大厂商积极在越南和美洲扩张产能，但受限于取向电工钢（GOES）和铜的结构性短缺（GOES在2025年存在约30%供需缺口且将持续至2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美欧贸易壁垒确实存在但被物理短缺约束，对 A 股电力设备出海是'硬锚有缝'而非'硬锚破裂'——card-05 配置框架基本成立，仅金盘需改为里程碑驱动节奏。", "summary": "美欧针对中国电力设备的政策工具栈已成型——EO 13920、Section 301、IRA 45X/Domestic Content、CBAM、FSR、NZIA、GOES 反规避——但北美 LPT 110–130 周交付期与欧美 OEM 订单簿 +40%+ 的物理短缺，制约了任何'全面替代中国货'…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "支持将资产向电网设备和电气基础设施权益进行分层切换，但警告当前板块估值已过度透支，且120周以上的变压器交期将固定价格积压订单暴露于铜价暴涨和公用事业监管降温的双重风险之中。", "summary": "支持将资产敞口从单一铜beta向电网设备和电气基础设施权益切换的逻辑，但必须在个股选择上保持极强的估值纪律。当前Eaton（38x-40x P/E）和Powell（45x-57x P/E）等龙头估值已严重透支五年周期。由于120周交期，老旧固定价格订单正暴露于铜价暴涨的毛利率挤压风险中，且公用事业…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 1, "neutral": 1, "value": 76.4, "events": [ { "title": "配电侧变压器与开关柜已锁定2026-2028量价齐升超级周期，金盘科技、特变电工、思源电气、中国西电、平高电气是A股最高确定性的物理底座组合。", "summary": "截至2026-05-31，AI数据中心驱动的配电侧设备已进入结构性供应缺口与超级周期：美国大型电力变压器交付周期≈128周(标准)/144周(GSU)，高定制订单达4年；GE Vernova 2026Q1单季为数据中心新增订单$2.4B，超过上一财年全年；MV/LV开关柜lead time普遍14…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "伊顿和施耐德的电力装备产能投产严重阻后于英伟达芯片出货周期，且受制于变压器铁芯原材料取向硅钢（GOES）的物理刚性短缺，成为制约GB200于2026年点亮上线的独立刚性瓶颈。", "summary": "截至2026-05-30，压力测试表明，伊顿与施耐德的产能扩张无法填补因电网延迟导致的GB200部署缺口。两家公司的主要新增产能要到2027年及以后投产，与2026年爆量出货的芯片周期存在严重时钟错配。变压器（交付128周）与开关柜（交付40-60周）积压严重，加之铁芯原材料取向硅钢（GOES）结…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have systematically evaluated the power grid equipment manufacturing sector's export orde…", "summary": "I have systematically evaluated the power grid equipment manufacturing sector's export order visibility, gross margin stability, and geopolitical imm…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 8, "neutral": 0, "value": 24.8, "events": [ { "title": "AI算力扩张的下一道硬约束是变压器、开关柜与取向硅钢（GOES），西方OEM新增产能要到2027–2028年才投运、GOES供需要到2030年才能缓解，这是供给侧未来两年的alpha…", "summary": "截至2026-05-31，本卡片支持AI算力瓶颈正从GPU迁移至变压器与开关柜，并在设备层量化约束。Wood Mackenzie 2025 Q2显示标准电力变压器交期128周、GSU 144周；Hitachi Energy欧洲≥100MVA大型电力变压器交期50–58个月。Eaton指定美国DC …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘政治与工业保护主义正将AI电网从中游环节‘脱钩化’，导致供给瓶颈行政化，显著加剧了算力扩张的物理墙刚性与Capex溢价。", "summary": "2026年6月调研显示，美中欧三方工业政策正合力‘加固’AI电网的物理瓶颈。美国通过《国防生产法》及EO 14156宣布能源紧急状态，对GOES及变压器实施Section 232高额关税并强化‘本土熔炼’规则，旨在保护Cleveland-Cliffs的单一供应垄断。中国则从2026年初起实施GOE…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然 unhedged 极端大宗暴涨对电气设备龙头构成理论毛利压力，但由于变压器/开关柜2.5-5年的极长交付周期与工厂槽位稀缺，ETN与HUBB凭借绝对的价格议价权、合同价格指数化…", "summary": "截至2026年5月31日，本卡对铜、铝和高牌号电工钢（GOES）价格波动对电气设备巨头（ETN/HUBB）利润率的侵蚀压力进行定量压力测试。我们发现，虽然 unhedged 情况下 Severe Stress（铜价达$14,000/吨，GOES+40%）会侵蚀 margins 达247-517 b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网CAPEX脉冲为'算电协同'提供基本面锚，但行业内部正在从特高压主线切换到配电+数字化主线，决定了2026 H2选股需从纯特高压转向配电、智能电表、二次设备与绿电直连。", "summary": "截至2026-06-01，国网2026电网投资指引≥6,500亿元、南网≥1,750亿元，两网合计首次突破8,250亿元，连续四年增长，验证Research note 08 '算电协同'重估论。结构上特高压进入'十四五'后段，2026 Q4–2027 Q2存在订单真空风险；配电网与数字化处于202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 0, "value": 62.2, "events": [ { "title": "AI 电力基础设施瓶颈不是需求识别问题，而是工业产能爬坡节奏问题：设备供应会改善，但赶不上2026-2028年数据中心通电假设。", "summary": "截至2026-06-03，本卡支持前序结论：变压器、高压开关设备和电气材料产能正在扩张，但节奏慢于 AI 数据中心通电需求。IEA 预计全球数据中心用电从2025年485 TWh增至2030年950 TWh；DOE/LBNL预计美国数据中心用电从2023年176 TWh增至2028年325-580…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核心判断：CRGO是2026-2028年更硬的物理瓶颈，铜更像价格和配给瓶颈，二者都削弱AI数据中心无摩擦通电假设。", "summary": "截至2026-06-03，本卡压力测试并总体支持前序判断：2026-2028年电力设备扩张的二次瓶颈更可能出现在合格取向硅钢（GOES/CRGO）和精炼铜的可得性、价格与配给上。CRGO供给弹性最低，美国依赖Cleveland-Cliffs Butler Works，印度JSW/JFE 35万吨/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 4, "neutral": 0, "value": 51.6, "events": [ { "title": "重型机械刚性是真实瓶颈，但传导终点是特殊金属与机械资本品的过路费收费站，而不是 LME 大宗——AI 不构成基础工业金属超级周期。", "summary": "本卡对 Card-03 进行压力测试：电力设备交付拉长确实经由重型机械产能（大型自由锻液压机、真空冶金/单晶铸造、特种机床、GOES 精整线）形成多年期刚性瓶颈，这一'制造刚性'判断成立；但'传导至广义工业金属价格结构性上涨'外推不成立——AI/电网仅给铜带来 ~3.5–4% 增量需求，被中国房地…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备出海链通过超级周期议价权（80%转嫁率）及国内配电‘以铝代铜’替代成功防御原材料通胀，而工程机械与风电整机在固定价格合同、激烈价格战及高关税壁垒下利润率失守，引发制造业内部资…", "summary": "本报告在电解铝、铜与煤炭供给约束抬升成本背景下对下游高端制造与装备制造业进行毛利率压力测试。结果表明中下游行业毛利率呈现极强的‘结构性大分流’：电网设备出海链得益于全球电力建设超级周期，能够通过PEC传导80%的成本压力，且国内配网通过‘以铝代铜’工艺替代展现强韧性；而工程机械与风电整机则因恶性国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "USMCA原产地规则（ROO）的收紧将使依赖进口组件的墨西哥电力设备组装产能对冲逻辑彻底失效，而沙特产能也面临美国 EAR/FDPR 针对 AI 基础设施的次级制裁风险。", "summary": "2026-06-03：本卡从政策监管视角压力测试出海产能对冲逻辑。USMCA 2026 联合审议实质性收紧了原产地规则，强制将“熔炼和浇注”标准延伸至变压器等核心电力设备。由于中国控制多数 GOES 产能，单纯依靠“组件进口+墨西哥组装”的产能将无法通过 RVC 测试，对冲逻辑彻底破产。同时，沙特…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策约束已经把A股半导体的可投资宽度实质性收窄到先进封装/HBM/SiC/稀土永磁四个切片，而ETF资金已经为全板块定价了80分位的政策溢价；下半年应'减总量、加精度'，并把2026…", "summary": "本卡片锚定2026-06-03。综合Card-01至Card-05的销量质量→政策销量→利润集中→行政性稀缺租金的链条，把先进制程白名单+成熟制程2027-06-23关税突跳+中端配套2026-11-10双向断点三重政策约束直接翻译为A股板块宽度、风格暴露、ETF资金流与拥挤度结论。政策筛后宽度约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 0, "value": 65, "events": [ { "title": "电网设备厂商的产能扩张弹性及其在AI硬件需求浪潮下的利润承压风险。", "summary": "截至2026-06-05，电网设备厂商扩产金额庞大（Hitachi Energy >90亿美元、Siemens Energy 4.21亿美元夏洛特厂、GE Vernova电气化在手订单翻两番至350亿美元），但物理产能要到2027-2030年才缓解，大型变压器交货期仍约40个月、美国价格涨77-7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策干预（包括DPA资金、关税调整和并网改革）在12-18个月内无法打破‘通电滞后’的物理时延瓶颈，保护主义关税和能效新规反而延长了供应链交期并抬升了电网建造成本。", "summary": "本报告对旨在打破‘通电滞后’的电网政策干预进行了压力测试。结果表明政策无法在12-18个月窗口内起效。DPA资金扩产的物理建设周期长达33个月（受制于21个月设备交期及6个月技能培训），产能迟至2029年释放。4月新加征的Section 232关税按成品全额征收，使大型电力变压器进口成本大涨7.5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 6, "neutral": 0, "value": 42.4, "events": [ { "title": "工业设备制造业的产能扩张远滞后于AI算力需求的增长，长达3-5年的变压器交货期与取向硅钢短缺构成了AI扩张的硬性物理约束。", "summary": "本报告对AI数据中心关键电力设备供应链进行了压力测试。测试显示大型变压器交货期已达128-160周以上，2026年全美16GW计划产能中近半面临延期。即便电气巨头（伊顿、日立等）大幅扩产，新产能也要到2027-2028年才能释放。受制于取向硅钢（GOES）等上游物料短缺，AI物理基建的“供应墙”确…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Scaling Law 的瓶颈已从芯片配额转向电力设备交付排期，工业产能的刚性约束将迫使全球 AI 竞争进入“存量上电能力”博弈阶段。", "summary": "截至2026-06-07，全球电力设备供应链已进入“结构性稀缺”状态，特高压变压器交付周期飙升至36-60个月，成为AI Scaling Law的绝对物理极限。工业交付与算法迭代的周期错位将ROIC从15%压缩至5%以下。中国凭借12-15个月的交付周期展现出显著的“上电速度”优势，而北美和欧洲受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备 OEM alpha 仍成立，但核心不是无限需求，而是 2026-2027 年订单积压可见度与 2027-2028 年前仍无法快速解除的材料/产能瓶颈。", "summary": "2026-06-06：我支持 Research note 02 的方向判断，但将其限定为“订单质量与交付稀缺”交易。Hitachi Energy FY2025 末积压 579 亿美元、Siemens Energy Grid Technologies 积压 490 亿欧元、GE Vernova El…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI能源溢价应从宽基设备链转向现金可兑现的电网/GIS与稀缺核心部件，液冷和AIDC变压器高beta标的必须先通过2026年中现金流复核。", "summary": "截至2026-06-07，本卡对A股变压器、GIS与液冷链条做估值拥挤度和现金质量压力测试。结论是保留AI能源主题，但不再给予宽基设备篮子统一溢价。思源电气仍是质量最高但估值拥挤的复合龙头；华明装备现金质量更干净，适合质量溢价；平高电气配得上温和GIS/电网溢价；金盘科技需等待2026Q2-Q3经…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 3, "neutral": 0, "value": 61.8, "events": [ { "title": "政策可以软化AI电力设备供应墙，但无法在2026年用贸易壁垒或补贴绕过GOES和大型变压器的物理扩产周期。", "summary": "截至2026-06-07，本卡压力测试上一卡“GOES与变压器构成物理供应墙”的结论。判断是：关税、出口管制和僵硬国产化规则不能绕过冶金、专用设备、测试和工厂建设周期，反而可能加剧2026年稀缺；真正有用的是盟友采购、关税与BABA豁免、规格标准化和过渡库存。产业补贴必要但滞后，Cleveland…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] China economic data and EM beta validation", "summary": "致 大类资产配置师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: e947fbc4-a808-440f-a64e-4d2fd375a137 结果摘要: Wrote handoff_response.md (10669 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 取向硅钢(GOES)产能瓶颈及对特高压变压器造价传导", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 材料行业分析师。 模式: adhoc 会话: e911bea4-9058-473c-a5ef-19ab52df0e02 结果摘要: I will first run the 系统日期锚 command to anchor today's date. …", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 2, "neutral": 0, "value": 69.2, "events": [ { "title": "2027 年电力设备交付瓶颈无法被产能扩张化解，高昂的积压订单饱和度与供应链约束将导致 AI 基础设施在 2027 年面临严重的物理交付墙。", "summary": "压力测试显示，GE Vernova 与西门子能源的扩产计划无法化解 2027 年交付瓶颈。尽管两家公司正在美国和欧洲大幅扩产，但其产能早已被 2024-2025 年签署的创纪录积压订单（GEV $150B，西门子 €154B）预占。燃机和变压器的交付周期分别维持在 5-7 年和 4-5 年的历史高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力基建瓶颈把议价权转向Eaton和Schneider等电气工业龙头，但供应刚性同时限制销量放量，因而应给予执行型估值溢价而非无限AI主题溢价。", "summary": "截至2026-06-08，本卡综合判断：AI电力瓶颈使Eaton与Schneider等电气工业龙头获得稀缺性溢价，但扩产速度受变压器供应链物理约束限制。Eaton Q1 2026 Electrical Americas数据中心订单同比约+240%、收入约+50%，Schneider Energy …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 1, "risks": 2, "neutral": 1, "value": 81.6, "events": [ { "title": "AI 基础设施的‘物理主权’已转移至掌握变压器核心组件（GOES、分接开关）与垂直整合能力的工业巨头，2026年的投产节点由‘变压器槽位’而非‘芯片配额’决定。", "summary": "2026年6月9日：本卡通过物理交期压力测试确认，电力变压器（交期3-5年）与开关柜（1.5-2年）已成为AI落地的‘刚性闸门’。工业巨头正从设备商转向‘产能地主’，Eaton通过‘从电网到芯片’的垂直整合与Boyd/Motivair的液冷并购锁定了高溢价份额；Schneider则凭借预制化模块（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与高压开关设备的产能扩张因制造周期、取向硅钢及分接开关的硬约束而严重滞后，无法在2027年AI需求高峰前达产，电力交付瓶颈将成为限制AI算力营收变现的核心硬约束。", "summary": "压力测试确认，变压器及高压开关设备扩产无法在2027年前满足AI基建需求。第一梯队大容量变压器交付期拉长至48-60个月，新订单排至2029年后。Hitachi、Siemens等厂扩产基本于2028年后投产，Eaton琼斯维尔厂2027年投产后仍需爬坡。高牌号取向硅钢（GOES）紧缺与有载分接开关…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "化工下游出海会把增量设备订单从“中国本土化工资本开支”转向“ASEAN 多厂点下游自动化”，但不会形成上游化工重装备的新超级周期。", "summary": "截至2026-06-10，本卡支持前序判断：东南亚化工下游转移会结构性改变工业机械与自动化设备贸易流向，但变化集中在复配、挤出、注塑/吹塑、包装、环保公辅、仪表和 PLC/DCS/SCADA，而非重资产上游裂解装置。ASEAN 已是大型机械进口平台，2024 年 HS84+HS85 进口约 7,6…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 0, "value": 78, "events": [ { "title": "“离网自治”模式面临美欧“安全与电力绑定”和关税/碳税的精准拦截，主权AI项目在放弃中国电力设备以获取AI芯片（交付延迟36-48个月）与使用中国设备（芯片断供）之间面临难以调和的结…", "summary": "本卡对AI“离网自治”模式进行压力测试。美欧正通过“安全与电力绑定”将AI芯片出口与排除中国设备挂钩（如阿联酋G42项目）。同时收紧FEOC与301关税以拦截含中国取向硅钢（GOES）的第三方变压器；欧盟也于2026年实施CBAM并禁用中国逆变器。开发者被迫在“使用中国设备但无芯片”与“脱钩但交付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 73.4, "events": [ { "title": "A股电力设备商海外产能迁徙的可行性边界与Capex瓶颈评估", "summary": "本卡支持卡片04–06主线：深度本地化是保住美欧电网设备溢价的生存条件。但补上工程现实——海外扩张的约束不是资本（龙头现金充裕、订单饱满），而是非财务执行链。四瓶颈分区域看：墨西哥约束是合规（2026-07-01 USMCA审查或收紧中国成分与RVC 50%实质性改变测试），沙特约束是技能工与高压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "相比直接的整机关税，核心部件的原产地关税与系统性融资限制对A股出海设备商构成更深层的毛利挤压，迫使行业出海模式向深度供应链本地化或非美欧市场切换分化。", "summary": "本卡片对美欧电网设备及取向硅钢（GOES）关税壁垒进行了政策压力测试。研究指出，美国Section 232新规（依据全申报海关申报价值征收50%铁芯关税）与欧盟GOES保障措施调查，表明西方政策重点已转向阻断“第三国绕道”与核心组件本地化强制约束。这使A股出海设备商面临严重分化，中度和重度压力下出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 1, "neutral": 0, "value": 80.8, "events": [ { "title": "在2026-06-12，受重型燃机5-7年长交期与航改燃机18-36个月交期的物理约束，天然气新建与表后自备发电在2026–2028年无法顺利桥接AI电力缺口，形成继变压器之后的第二…", "summary": "截至2026-06-12，本压力测试卡片评估了天然气作为2026–2028年AI电力桥接产能的供应链瓶颈。GE Vernova、西门子能源与三菱重工重型燃气轮机在手订单已排至2029年以后，交期长达5-7年。航改型燃气轮机（如GE LM6000）因与民航发动机核心机需求冲突，交期拉长至18-36个…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 3, "neutral": 0, "value": 78.2, "events": [ { "title": "高电价正加速高端制冷系统订单放量，冰轮环境与冰山冷热通过产品结构升级获得了显著的毛利扩张与更强的成本传导能力，确立了2026年业绩高增的确定性。", "summary": "在2026-06-13评估中，我们确认高电价（0.90元/度）正强烈驱动冷链客户转向具备19.5% CapEx溢价的高能效系统，进而大幅推升了设备龙头的利润率。冰轮环境（000811.SZ）2026年Q1归母净利润大增25.06%，其海外高毛利业务及IDC液冷新签订单（16亿元）表现亮眼；冰山冷热…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "铜价波动下的特高压设备厂商利润韧性：价格联动机制的覆盖范围与滞后效应评估", "summary": "压力测试卡片07的“铜价挤压”判断，方向确认但重定义机制：任务要求的“价格联动条款”在主导的国网渠道里基本缺位——国内对铜价联动接受度低，框架中标多为投标日固定价格。平高、西电的真实缓冲是锁价采购、套保、订单批次与结构升级，而非合同转嫁。2025毛利逆势扩张（平高23.9%、西电22.75%）是结…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在并网延迟常态化与变压器/GOES双重瓶颈挤压下，高压设备商稀缺租金向上游硅钢龙头转移，且欧美面临长达230周的极端交付时滞，高估值设备板块Beta承压，Barbell策略建议防守配…", "summary": "截至2026-06-14，质量主管审计了前7张卡片。电网并网排队（4.3年）与变压器交期（120-210周[S5]）是硬瓶颈，以铝代铜使取向硅钢（GOES）消耗量增加15-30%[S5]加剧了缺口。GOES新增供给200-250 kt/年仅能覆盖50-65%需求[S10]，分走设备商30-40%租…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国 GOES 出口许可制 + 欧盟反倾销共同令非中国高磁感供给紧缩 20–28%，推动非美市场大型变压器成本上行 6–9% 与交期再延长 8–12 周，全球电网设备割裂在材料层被锁…", "summary": "中国 2026-01-01 起对取向硅钢（GOES, HS 7225.11/7226.11）实施出口许可，叠加欧盟/印度/墨西哥反倾销，将使全球可自由流通高磁感 GOES 缩减 30–40 万吨/年。在 280–300 万吨全球需求中，非美市场（印度、欧盟、东盟、中东、拉美）承担主要冲击：HiB …", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 2, "neutral": 0, "value": 77.6, "events": [ { "title": "变压器与特高压产业链正在扩产，但全球大型变压器/HVDC/铜链条无法完全匹配AI负荷的近期增速；中国主干特高压可较好匹配，中美差异的核心在本地交付与材料瓶颈。", "summary": "截至2026-06-15，本卡片支持卡片09：AI算力扩容的约束在电网设备与输电交付，而不是火电燃料成本。全球数据中心用电量按IEA基准从2024年约415 TWh升至2030年约945 TWh、约年增15%，但全球变压器市场和数据中心变压器市场预测增速仅约7.24%和6.3%。美国大型变压器存在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的扩张瓶颈已从芯片转向物理电力设备，燃机与变压器OEM凭借5年以上的积压订单和极强的定价权成为供应链中最大的赢家。", "summary": "本报告深入分析了2026年AI电力供应链中的物理瓶颈。研究发现，燃气轮机和高压变压器OEM已成为供应链的绝对咽喉，GE Vernova和西门子能源的燃机订单已排至2030年以后，大型变压器交期突破150周。瓶颈核心在于取向电工钢（GOES）的结构性短缺，导致2026年全美近50%拟建数据中心产能面…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 4, "neutral": 0, "value": 67, "events": [ { "title": "中国电力装备外需足以覆盖行业层面的供应链重构成本，但超额收益将集中在材料、认证和渠道能力最强的头部企业，而非所有出口商。", "summary": "截至2026-06-16，本卡片压力测试全球变压器超级周期下中国电工钢与电力设备出海的利润传导。结论是外需足以支撑行业继续上行，但不能简单抵消所有地缘贸易壁垒；真正受益者是掌握高牌号GOES、具备高压/特高压认证和海外服务能力，并能在必要市场进行最终装配或服务本地化的企业。美国50%钢铝关税和欧盟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力配电设备产能扩张与供应链瓶颈：AI算力建设的硬约束 分析师：工业制造分析师（industrials analyst） 立场：support（强化 Research note 02…", "summary": "电力配电设备产能扩张与供应链瓶颈：AI算力建设的硬约束 分析师：工业制造分析师（industrials analyst） 立场：support（强化 Research note 02 的\"电力设备瓶颈是AI算力部署真正天花板\"判断） 工作日期：2026 06 15（Asia/Singapore） …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建电力设备供应链的物理约束：变压器、燃气轮机与电网扩容的产能极限", "summary": "我支持Research note 04“执行胜过芯片”的方向，但压力测试其两个承重论点。约束并非单一设备，而是串行链条（并网5.2年→高压变压器128周/GSU 144周→开关柜→燃气轮机），最长杆是变压器总装/试验吞吐，而非资本[S1][S2]。GOES是真实但次级的约束：价格较2020年翻倍[…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI电力基础设施瓶颈向上游物理期权（精炼铜/取向硅钢）传导评估", "summary": "致 日报总编： 协同任务已完成。执行方为 主题研究员。 模式: adhoc 会话: bca64921-2ae4-41c5-bfcf-48378f7e6da6 结果摘要: Wrote handoff_response.md (13721 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 3, "neutral": 0, "value": 74.4, "events": [ { "title": "截至2026-06-16，变压器与GOES供应弹性不足以在2026-2027年全面缓解AI算力落地的物理天花板，只能给已提前锁定OEM产能的头部云厂商提供选择性缓冲。", "summary": "截至2026-06-16，本卡压力测试认为：GOES和铜不是孤立瓶颈，真正约束是GOES、CTC铜导体、制造槽位、专用设备、测试、熟练工和运输共同构成的变压器工业系统。80-210周交期意味着2026年新下单多数无法支撑2027年前通电；Hitachi、Siemens等扩产有助于2027年末以后斜…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与开关柜交期已经把AI扩容约束从电力需求预测推进到工业制造产能约束，证实chips与grid属于同一physical-AI capex风险桶。", "summary": "截至2026-06-16，本卡支持“电网设备交付周期证实物理红线”的判断。Wood Mackenzie 报告显示大型变电站与发电机升压变压器交期为80-210周，且美国2025年电力/配电变压器供应缺口估计为30%/10%；POWER Magazine 指出开关柜2025年Q2平均交期44周。Ea…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2029年电网设备扩产会缓解但难以消除交付瓶颈，关键稀缺性正在从变压器OEM装配上移到材料、组件、测试和高压系统集成。", "summary": "截至2026-06-16，本卡压力测试第09卡：变压器和关键电网设备扩产真实，但2026-2029年仍难完全跑赢AI数据中心、可再生能源并网、电网替换和电气化带来的MW/MVA交付需求。2028年前大部分大型电力变压器、HVDC电缆、开关设备和上游组件仍处紧缺。真正的瓶颈不只是OEM装配，而是GO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 2, "neutral": 0, "value": 81.8, "events": [ { "title": "全球电网核心设备及原材料的产能赤字已构成刚性物理准入壁垒，导致产业利润池从算力/晶圆厂资本支出方结构性向重电OEM厂商转移。", "summary": "大功率汽轮机与超高压变压器的严重产能缺口（全球超高压变压器短缺约30%，交期延至2-4年；大型汽轮机交期延至5-7年）和取向硅钢（GOES）等关键材料瓶颈，对算力中心和晶圆厂构成不可逾越的物理准入壁垒。议价权转移令电力设备OEM（如HD Hyundai Electric和GE Vernova）利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电气设备出口的贸易壁垒与安全审查评估", "summary": "压力测试卡片05的压制论：美欧安全审查真实且在升级，但被狭窄锁定在中国份额本就很小的细分。最硬的法定禁令（EO 13920仅限关键国防设施、2026年欧盟逆变器资金禁令）是产品/用例特定的；2025年流氓设备警报针对的是带联网通信栈的逆变器与电池，而非被动的高压变压器。总量上，2025年中国变压器…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 3, "neutral": 0, "value": 77.2, "events": [ { "title": "燃机及关键电气设备供应链的交付压力与产能弹性评估", "summary": "截至2026-06-18，本卡压力测试燃机与电气设备供应链。三巨头GE Vernova、Siemens Energy、MHI占全球在建燃机产能约三分之二，2026年起年提产25–35%，但扩产仅抵消积压增速而非积压本身：GE Vernova积压约83GW、产能20→24GW/年，预留售罄至2030…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与GOES的区域化贸易壁垒通过拉长交付周期至154.5周，直接造成算力资产闲置折旧（24k GPU集群年拖累达5.624亿美元），迫使超算厂商进行CAPEX重组。", "summary": "截至2026-06-18，研究变压器与GOES区域壁垒对AI算力的传导效应。美国IRA本土含量、欧盟NZIA及中国反倾销税等碎片化政策使GOES供给割裂。压力情景下，卷线工稀缺与设备积压将变压器交期拉长至154.5周。这导致超算并网时滞达3-5年，使得24k GPU算力集群闲置一年的折旧与流失利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2028年能把backlog转成可持续利润率的是GEV、Eaton以及ABB/Schneider的高质量电气业务，而不是所有电气设备订单；一次性订单膨胀风险集中在Hubbe…", "summary": "截至2026-06-19，本卡综合支持前序AI电力交付瓶颈判断，但将多头腿收窄为具备重定价backlog、可交付产能、服务附着和预付款保护的制造商。GEV最强，因燃机槽位、Electrification backlog和Prolec GE变压器经济性叠加；Eaton次之，电气业务backlog和高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 3, "neutral": 1, "value": 81.6, "events": [ { "title": "A股AI电网capex链条配置优先级：上游材料瓶颈 > 电网设备OEM > 出口龙头 > 资产型公用事业。", "summary": "截至2026-06-20，本卡综合前五张卡片，给出AI电网资本开支链条在A股的优先级：上游高壁垒材料瓶颈（高牌号GOES、CTC铜、绝缘材料、酯类介质液）>电网设备OEM（特高压主设备、数字化二次、SVG）>出口龙头（变压器/开关DM替代窗口）>资产基数增长型公用事业。前两档承担弹性主战场，第三档…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AIDC电力基建逻辑下高压变压器与输电网铜需求增量的定量折算", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b7159f5c-eb1d-48e7-8538-01f5bfef96a5 结果摘要: Wrote handoff_response.md (10534 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "监管与政策瓶颈：环境许可、核燃料地缘风险与能源政策对电力基础设施扩张的影响", "summary": "政策工具箱真实且方向正确——NEPA类别豁免与Seven County判决（2025-05-29）、FERC第2023/1920-B号令、ADVANCE法案与Part 53、HALEU拨款、DPA第303条（2026-04-20）——但均存在2–5年监管-物理滞后，落地在AI电力约束最紧的2026…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备供应链瓶颈及其对AI基础设施扩能的二次约束效应", "summary": "报告日期2026-06-19。本收官卡压力测试Research note 07“折价而非证伪”结论，引入变压器、高压开关与取向硅钢(GOES)的二次设备瓶颈。变压器交期由128周(25Q2)拉长至超160周/最高4-5年[S2][S1]，GSU需求自2019年+274%、电力变压器供应缺口约30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 4, "neutral": 0, "value": 74, "events": [ { "title": "电网关键设备供应瓶颈：变压器与开关设备的交付压力", "summary": "本卡支持卡片02的核心结论——美国AI电气化的真正约束已从发电下移至电网设备（变压器、开关、接入）。从工业供给侧补充：该瓶颈正重塑GE Vernova与Eaton的订单簿与资本开支。GEV电气化设备订单储备升至约390亿美元（同比+75%），数据中心当季订单约24亿美元（超2025全年），并以53…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "工业电气设备产能扩充与供应链交付压力", "summary": "压力测试 Research note 02–03 的电气设备瓶颈论。瓶颈在 2026–2028 年对定制大型变压器、HVDC、中压开关柜与并网真实成立，但稀缺溢价被误当作永久性。三点修正：一、积压创纪录（Eaton 总积压逾190亿美元、数据中心积压228GW≈12年需求；GE Vernova电气…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施工业供应链正在扩产，但订单积压本身证明吞吐仍偏紧；重电设备会继续限制2026-2027年算力上线节奏，液冷只能边际缓解。", "summary": "截至2026-06-21，本卡片支持第9张卡片的物理瓶颈判断，但把结论细化为“部分缓解而非完全解决”。Eaton、Vertiv、nVent、GE Vernova、Siemens Energy、Schneider和Hitachi Energy的订单积压与CapEx显示AI电力/冷却需求能见度很强；但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备交期瓶颈与大宗物流通胀引起的制造业利润受阻，共同锁死了2027年后全球AI算力集群的物理部署进度，迫使超大云厂商非计划性降速。", "summary": "本卡支持并完成了十卡研讨。在原油供给冲击与G3利率高企背景下，大型变压器（交期80-210周）与高压开关柜（交期80-120周）交付滞后是限制2027年后算力集群建设的终极瓶颈。定量测试显示，大宗及海运暴涨导致中国纯出口商变压器净利润转负（-3.52%）强制引发交付中断；而本地化厂商受限于取向硅钢…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 1, "neutral": 1, "value": 96, "events": [ { "title": "电力设备（如变压器）供应链瓶颈对AI基础设施建设节奏的影响", "summary": "压力测试结论：是，但有两点修正。对美欧并网型AI建设，电力设备2026年当下即是约束边界且早于芯片——约一半美国规划数据中心已延期/取消，12GW中仅约5GW在建[S4][S9]，而CoWoS仍在近4倍扩产出货[Card05]，故‘买得到却通不了电’的部署约束当下即活跃。修正一：约束不是变压器箱体…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research Summary (Research note 06/10) This card performs a stress test on the thesis of \"f…", "summary": "Research Summary (Research note 06/10) This card performs a stress test on the thesis of \"frictionless growth for Chinese electrical equipment export…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 0, "value": 88.8, "events": [ { "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 4cd2d717-9d04-4569-b0d8-cc37d743d0a9 结果摘要: Wrote handoff_response.md (9414 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b04fea49-07a7-43ca-b246-32f00d158f50 结果摘要: Wrote handoff_response.md (9863 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 4, "neutral": 0, "value": 78.2, "events": [ { "title": "电力设备供应链扩产进度能否扭转2027年交期约束", "summary": "压力测试'扩产能救2027'的乐观读法：GE Vernova(全资Prolec GE，US$5.275bn)、Eaton(Jonesville US$340m+Nacogdoches US$100m+Waukesha US$22m)、Hitachi Energy(逾US$1bn含South Bos…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建配置应从“追逐算力设备增量”切换为“持有可交付电力稀缺性”，但保留CPO核心仓位，因为CPO仍是降低AI网络功耗的必要技术。", "summary": "截至2026-06-24，本卡综合前序结论：AI算力瓶颈已从GPU转向已通电MW，组合应从高拥挤CPO单边beta转向“变压器OEM+中国出海设备+GOES/铜材料+CPO核心赢家”的哑铃。100单位AI基建权益袖基准配置为韩日及全球变压器OEM 30、中国出海设备22、上游材料18、CPO/光模…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网队列与基础设施扩容压力评估", "summary": "本卡支持卡片1–2：约束已从Capex/GPU转向已通电的MW，并将瓶颈量化为双向并网堵塞。发电侧：美截至2025年底约2,060GW排队（约现有装机2倍），408GW已签约未通电，申请到商运中位时长翻倍至>4年。负荷侧：ERCOT大负荷队列达226GW（近4倍于2024年底63GW，77%为数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网输变电容量与并网审批周期（而非总体发电量）已成为AI算力部署的终极物理刚性瓶颈，并已实质性挤压IT硬件采购预算。", "summary": "本报告对AI算力扩张的电网与能源瓶颈进行压力测试。分析表明，虽然电力设备订单充足，但美国并网队列平均达5.5年，且面临FERC 2026年6月大型负荷监管及弗吉尼亚GS-5等防止成本转嫁政策。变压器等重电面临GOES原材料短缺，Tier-1厂商积压至2029-2030年。表后核电直供监管被封死，燃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 3, "neutral": 0, "value": 85.6, "events": [ { "title": "关键电力设备（变压器/开关/UPS）的交付周期与产能约束评估", "summary": "压力测试卡片06的“设备交付为约束”命题。结论基本否定但附两条限定：就中国境内而言，设备总产能与硅钢/铜大宗总量都不是约束——TBEA一家约260 GVA/年，远超卡片06测算的AI增量约16.6 GVA/年（仅约6.4%）；中国2024年取向硅钢产量303万吨（日5倍、美8倍）结构性过剩且净出口…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网侧供应瓶颈对电力设备资本开支的驱动效应", "summary": "支持立场。变压器交期已拉长至128周（LPT）/144周（GSU），北美极端排期达60个月，使交期本身转化为多年期订单能见度：GE Vernova电气化在手订单同比升至424亿美元（覆盖约3年营收[自测算]），Siemens Energy集团在手1,540亿欧元，Eaton总订单+73%至超190…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器、开关设备、取向电工钢、电缆和电网设备制造产能", "summary": "支持卡片05/06：油价回落是数据中心电力运营成本的缓冲，而非解锁电网设备产能的钥匙。从制造现场看，资本已在大举投入——Hitachi Energy（60亿美元集团计划含约15亿专用于变压器、2027年放量）、Siemens Energy（4.21亿美元Charlotte厂2027年竣工）、GE …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 3, "neutral": 0, "value": 87, "events": [ { "title": "全球电力设备制造供应链产能与上游材料供应风险评估", "summary": "部分确认Card-03的time-to-power论点，但作三处修正。瓶颈确在上移到材料端，但对变压器而言真正的物理咽喉是GOES而非铜，且GOES稀缺本质是“产能+贸易”稀缺而非地质稀缺：全球变压器产能利用率98%、GOES自2020年涨约100%、美国仅Cleveland-Cliffs一家、俄…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心的卡脖子资产已从芯片变为高压电气设备，交期与产能缺口把电气推成最大且持续涨价的Capex板块，稀缺溢价持久归于变压器/特高压/电气配套，而非火电或燃气beta。", "summary": "截至2026-06-25，本卡支持并加固Card-01判断：AI数据中心的约束已从芯片转向电气并网设备。核查结果——大型变压器交期约128周、GSU约144周、中压开关柜52–78周，均超数据中心12–18个月建设周期；GSU需求2019–2025增274%，美国变压器/配电变压器2025供给缺口…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备短缺不会取代电力缺口，而是把“已签MW”过滤成“可通电、可保护、可冷却、可调试MW”，其中变压器是2026-2028年最硬瓶颈。", "summary": "截至2026-06-25，本卡压力测试认为电力配套设备会成为继电力缺口之后的下一道算力部署硬约束，但约束强度分层：变压器最硬，128-143周交期足以卡住2026-2028年通电节奏；开关柜平均44周，更多体现为分阶段通电和balance-of-plant同步风险；液冷会卡住GB200/Rubin…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 10, "neutral": 0, "value": 46.4, "events": [ { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海外电网升级大周期与原材料成本冲击下的电网设备出口毛利弹性测试", "summary": "本卡综合集成并收紧 Card-05 的电网设备多头：在 LME 铜价高于 10,000 美元/吨、GOES 偏紧背景下，A股电网设备出口毛利弹性有韧性而非脆弱。自测算敏感性显示铜与 GOES 同时 +20% 时，固定价订单毛利收缩约 5.3pp，但 80% MEC 传导下仅约 1.1pp[自测算]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "随着全球及中国变电设备产能在2027-2029年逐步释放，AI数据中心的电力设备瓶颈将从“无法获取设备”的二进制物理硬约束，降为“支付溢价与关税以获取交期”的资本支出与利润率成本约束。", "summary": "本报告对2027-2029年AI电力设备瓶颈的缓解速度进行了深入分析。结论表明，随着西方重电巨头（如日立能源、GE Vernova）扩产产能释放，以及以特变电工、思源电气、金盘科技为代表的中国高效率设备出口，2027-2029年电力设备将从物理上的“硬约束”转变为“成本/经济约束”。由于取向电工钢…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网延迟的变压器库存积压是可逆的时点拖累(金盘-8%~-11%、思源<-3%)，被国网4万亿内需托底，应在超配出口变压器框架下偏思源、轻金盘并叠加配对与公用事业杠铃对冲。", "summary": "并网延迟导致的已完工未交付变压器库存积压，是营运资金/收入确认的时点事件而非需求毁灭——订单仍排产至2027年在手backlog内。对金盘科技(688676.SH)拖累真实但有界：北美AIDC相关收入是最高Beta切片，Q2/Q3净利约-8%至-11%、伴随存货与FCF占用，属逢低布局而非证伪；对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 4, "neutral": 0, "value": 83.8, "events": [ { "title": "功率半导体已过剩，AI电力的新物理瓶颈已迁移至大型变压器，并真正卡在无法被资本快速绕行的上游电磁钢(GOES)与铜——2026–2028窗口难解。", "summary": "支持并收紧前五卡逻辑链：功率半导体转入结构性过剩后（Research note 05），AI电力建设的刚性瓶颈已迁移至大型变压器(LPT)，并下沉到上游电磁钢(GOES)与铜。瓶颈呈两层：整机厂总装层可用约20亿美元北美扩产去瓶颈，但缓解要到2028/2029年后；标准电力变压器交期128周、GS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器材料和交付瓶颈已开始改变 Hyperscaler 的2026-2027资本开支与电力采购节奏，影响主要体现为预算通胀、长期锁电和通电交付递延，而非AI需求取消。", "summary": "截至2026-06-28，结论为支撑：Hyperscalers 已经通过预算上修、长期电力合同和项目节奏调整来应对变压器及电气设备瓶颈，但公开披露通常不会把变化直接归因为 GOES 或铜关税。Microsoft 2026 capex 中约250亿美元来自更高组件价格，Meta 上调2026 cap…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "summary": "截至2026-06-27，本卡片综合前序结论：电网/变压器瓶颈会延后并改道AI部署，但不会摧毁2028年前光模块出口龙头业绩能见度；真正尾部风险是2027年云厂商ROIC压缩。建议A股AI物理瓶颈组合使用40/32/12/8/8结构：变压器/电网设备、光模块、电力现金流锚、铜/GOES等输入端对冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球贸易壁垒对变压器关键材料（GOES/铜）供应链稳定性的影响分析", "summary": "本卡片压力测试了 Research note 03–04 断言的政策传导机制：美国 232/301 关税与 USMCA 熔铸（melt-and-pour）原产地规则并非单纯征税，而是一台'流向重定向引擎'，把材料短缺锁定为结构性供应瓶颈。自测算显示，一台对美变压器在直接进口路径下到岸成本 +40%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 10, "neutral": 0, "value": 49.2, "events": [ { "title": "变压器交期拉长至3-5年与并网排队超36个月的物理刚性瓶颈，配合美欧贸易保护关税对GOES硅钢供应链的阻断，已对AI算力扩张的通电效率构成无法用资本单纯解决的长期结构性壁垒，造成明显…", "summary": "基于并网排队与变压器短缺的双重物理约束，我们验证电力基础设施瓶颈已构成AI算力扩张的长期结构性阻碍。数据中心通电时间已拉长至36-60个月，变压器交期延至3-5年，主要厂商积压订单超百亿美元。这些延误对项目NPV和IRR产生重大侵蚀，100MW数据中心延误3年将损失51.63%的NPV。美国Sec…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "summary": "本卡支持 Research note 02「上游原材料是最深约束」的结论，并锐化为一组非对称结构。GOES：自2019年变压器需求暴增（电力+119%、GSU+274%），远超GOES市场约5.4–5.6%的CAGR；GSU供给缺口2025年约100%，因美国唯一GOES生产商Cleveland-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关键矿产与设备出口管制风险对A股高端环节溢价逻辑的冲击评估", "summary": "压力测试出口管制威胁。结论：把镓/锗/锑/稀土剧本套到宝钢与紫金上是双向错配，不反转 Research note 06 基准论断。铜/紫金：中国是全球最大铜进口国（2025 精矿进口 30.31Mt，国内矿仅约 1.85Mt，依存约 87%），出口武器结构上不存在；紫金约 50% EBITDA 来…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备成本传导逻辑的脆弱性：铜与电工钢供应瓶颈能否被设备商有效对冲？", "summary": "本卡用材料视角压力测试2027-2028年铜与GOES供需缺口对HV变压器利润率的侵蚀确定性。结论：成本传导逻辑脆弱，但脆弱是非对称、取决于合同结构的，而非全行业确定。GOES是确定性最高、最难对冲的咽喉——区域单一来源（美国仅Cleveland-Cliffs，本土覆盖约1/5需求）、约占LPT成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 5, "neutral": 1, "value": 83.6, "events": [ { "title": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "summary": "本卡片从制造和供应链视角对电网与电力约束进行压力测试。截至2026-06-29，关键电力设备（变压器与重型燃机）由于极高工艺与测试门槛、上游取向电工钢（GOES）瓶颈及寡头垄断，交期拉长至3-7年，生产机位已售罄至2029-2030年。云厂商BTM发电加剧了燃机机位争夺。预计2026-2027年该…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 中压干式变压器出海企业海外合规及交付溢价测算", "summary": "致 主题研究员： 协同任务已完成。执行方为 工业制造分析师。 模式: adhoc 会话: f0fa9b36-39ce-428c-a4ff-9c7d5180b4ad 结果摘要: Wrote handoff_response.md (11248 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "summary": "本卡压力测试 Research note 6 的'调价条款已隔离利润'结论，认为其仅有条件成立。主流调价条款（MEC）挂钩有流动期货的铜/铝（出口超级周期转嫁率约80%[S1]），却普遍排除或封顶无流动指数的取向电工钢GOES——而GOES占COGS约28%[Research note 6]且处2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-bef6966fdd/", "networkHref": "/investor/research-network/#thesis-bef6966fdd" }, { "id": "thesis-9a860a8801", "source": "tracked_thesis", "title": "信用再融资压力正在把利率风险转化为资产负债表筛选", "titleEn": "Credit refinancing pressure is turning rate risk into balance-sheet selection", "lane": "信用再融资/私募信贷", "laneEn": "Credit refinancing and private debt", "laneHref": "/investor/lanes/credit-refinancing-and-private-debt/", "investableFocus": "信用再融资/私募信贷：BX, KKR, APO, ARES, JPM, KRE", "stockUniverse": [ { "ticker": "BX", "name": "Blackstone", "market": "US", "role": "另类资产与私募信贷", "roleEn": "另类资产与私募信贷", "bucket": "core", "href": "/investor/stocks/BX/" }, { "ticker": "KKR", "name": "KKR", "market": "US", "role": "私募市场与信用", "roleEn": "私募市场与信用", "bucket": "core", "href": "/investor/stocks/KKR/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "ARES", "name": "Ares Management", "market": "US", "role": "直接贷款与信用", "roleEn": "直接贷款与信用", "bucket": "core", "href": "/investor/stocks/ARES/" }, { "ticker": "JPM", "name": "JPMorgan Chase", "market": "US", "role": "银行信用周期", "roleEn": "银行信用周期", "bucket": "core", "href": "/investor/stocks/JPM/" }, { "ticker": "KRE", "name": "SPDR S&P Regional Banking ETF", "market": "US ETF", "role": "区域银行压力代理", "roleEn": "区域银行压力代理", "bucket": "core", "href": "/investor/stocks/KRE/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" } ], "stockTickers": [ "BX", "KKR", "APO", "ARES", "JPM", "KRE", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM" ], "coreView": "赛道：信用再融资/私募信贷。候选观察标的：BX, KKR, APO, ARES, JPM, KRE。当前共有 40 条证据、1 位分析师和 6 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 73, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 11, "risks": 116, "neutral": 33, "analystCount": 19, "sourceTypeCounts": { "研究笔记": 125, "证据更新": 1, "分析协作": 26, "研究主题": 6, "研究线程": 2 }, "topTerms": [ "AIDC", "银行", "Credit", "refinancing", "pressure", "turning", "rate", "risk", "balance-sheet", "selection", "private", "debt" ], "whatChanged": [ "2026-06-29: 绿色技改缺口下高碳工业民企债务违约风险与信用利差压力测试评估", "2026-06-29: [auto-handoff] Hormuz shock & core CPI transmission", "2026-06-29: 电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对冲有效却对冲错轴，继续超配电网设备而非纯绿电发电。", "2026-06-27: 国债底线稳的是无风险利率，不是高杠杆主体的信用利差；若弱城投和房企利差被重定价至250bp-400bp，展期压力会非线性上升，但仍更可能表现为行政化续作而非系统性违约失控。", "2026-06-26: 高股息防御腿是否正在从避险资产变成久期风险？" ], "practical": { "sectorLane": "信用再融资/私募信贷", "sectorLaneEn": "Credit refinancing and private debt", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "BX", "name": "Blackstone", "market": "US", "role": "另类资产与私募信贷", "roleEn": "另类资产与私募信贷", "bucket": "core", "href": "/investor/stocks/BX/" }, { "ticker": "KKR", "name": "KKR", "market": "US", "role": "私募市场与信用", "roleEn": "私募市场与信用", "bucket": "core", "href": "/investor/stocks/KKR/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "ARES", "name": "Ares Management", "market": "US", "role": "直接贷款与信用", "roleEn": "直接贷款与信用", "bucket": "core", "href": "/investor/stocks/ARES/" }, { "ticker": "JPM", "name": "JPMorgan Chase", "market": "US", "role": "银行信用周期", "roleEn": "银行信用周期", "bucket": "core", "href": "/investor/stocks/JPM/" }, { "ticker": "KRE", "name": "SPDR S&P Regional Banking ETF", "market": "US ETF", "role": "区域银行压力代理", "roleEn": "区域银行压力代理", "bucket": "core", "href": "/investor/stocks/KRE/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" } ], "stockTickers": [ "BX", "KKR", "APO", "ARES", "JPM", "KRE", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM" ], "primaryStocks": [ { "ticker": "BX", "name": "Blackstone", "market": "US", "role": "另类资产与私募信贷", "roleEn": "另类资产与私募信贷", "bucket": "core", "href": "/investor/stocks/BX/" }, { "ticker": "KKR", "name": "KKR", "market": "US", "role": "私募市场与信用", "roleEn": "私募市场与信用", "bucket": "core", "href": "/investor/stocks/KKR/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "ARES", "name": "Ares Management", "market": "US", "role": "直接贷款与信用", "roleEn": "直接贷款与信用", "bucket": "core", "href": "/investor/stocks/ARES/" }, { "ticker": "JPM", "name": "JPMorgan Chase", "market": "US", "role": "银行信用周期", "roleEn": "银行信用周期", "bucket": "core", "href": "/investor/stocks/JPM/" }, { "ticker": "KRE", "name": "SPDR S&P Regional Banking ETF", "market": "US ETF", "role": "区域银行压力代理", "roleEn": "区域银行压力代理", "bucket": "core", "href": "/investor/stocks/KRE/" } ], "instrumentTags": [ "BX", "Blackstone", "KKR", "KKR", "APO", "Apollo Global Management", "ARES", "Ares Management", "JPM", "JPMorgan Chase", "KRE", "SPDR S&P Regional Banking ETF" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 74 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 19, "recentEvents": 74, "supportRatio": 0.07, "riskRatio": 0.72, "conflictScore": 82.4, "evidenceDepth": 100 }, "score": 68.8 }, "events": [ { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19, "matchedTerms": [ "AIDC", "ai-infrastructure", "aidc", "power" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 17.75, "matchedTerms": [ "risk", "evidence", "dominant", "inflation", "transmission", "tightening", "fall" ], "href": "" }, { "date": "2026-06-29", "title": "绿色技改缺口下高碳工业民企债务违约风险与信用利差压力测试评估", "summary": "截至2026-06-30，本卡片对高碳民营工业债违约概率与行业利差走阔做信用压力测试。全市场宽口径违约率仅0.08%–0.13% [S8]，边际0.4%远低于历史中位1.6% [S7]，系统温和。但高碳POE群体是不利尾部：12个月前瞻PD约5.05%（钢铁6.5%、化工5.1%、有色加工3.9%）[自测算]，约市场40–60倍。由于民企净融资自2021年…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "银行", "Credit", "credit" ], "href": "" }, { "date": "2026-06-27", "title": "国债底线稳的是无风险利率，不是高杠杆主体的信用利差；若弱城投和房企利差被重定价至250bp-400bp，展期压力会非线性上升，但仍更可能表现为行政化续作而非系统性违约失控。", "summary": "截至2026-06-28，本卡压力测试信用利差非线性扩张对城投及地产债展期的传导。结论是：国债收益率底线可稳定基准久期，但无法阻止弱主体综合融资成本因利差跳升而上行。政策化债资源仍使城投系统性失控概率偏低，地产现金流和万科式展期案例显示尾部压力更直接。基准情景不是全面违约潮，而是更多展期、重组、银行续贷和区域分化。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "银行", "Credit", "credit" ], "href": "" }, { "date": "2026-06-26", "title": "[auto-handoff] 权益两融去杠杆是否外溢至银行间资金面并推动信用利差走阔", "summary": "致 信用分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 7dab7f81-754b-48e9-b144-aa1269bfb054 结果摘要: Wrote handoff_response.md (13780 bytes); wrote handoff_response.en.md (16824 bytes) — base …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 16, "matchedTerms": [ "银行", "Credit", "credit", "tightening", "equity" ], "href": "" }, { "date": "2026-06-26", "title": "并网延迟的变压器库存积压是可逆的时点拖累(金盘-8%~-11%、思源<-3%)，被国网4万亿内需托底，应在超配出口变压器框架下偏思源、轻金盘并叠加配对与公用事业杠铃对冲。", "summary": "并网延迟导致的已完工未交付变压器库存积压，是营运资金/收入确认的时点事件而非需求毁灭——订单仍排产至2027年在手backlog内。对金盘科技(688676.SH)拖累真实但有界：北美AIDC相关收入是最高Beta切片，Q2/Q3净利约-8%至-11%、伴随存货与FCF占用，属逢低布局而非证伪；对思源电气(002028.SZ)近乎非事件(<2-3%)，出口…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.5, "matchedTerms": [ "AIDC", "rate", "aidc" ], "href": "" }, { "date": "2026-06-26", "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大型变压器交期和AIDC电力设备需求支撑；主要反转风险来自重复…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.5, "matchedTerms": [ "AIDC", "rate", "aidc" ], "href": "" }, { "date": "2026-06-26", "title": "[auto-handoff] 商业银行 1.40% NIM 约束对保障房收储贷款投放的挤出压力", "summary": "致 房地产分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: aa75509e-3210-4ab6-b781-37670c2d5374 结果摘要: Wrote handoff_response.md (9392 bytes); wrote handoff_response.en.md (11030 bytes) — Recor…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 14.5, "matchedTerms": [ "银行", "risk", "spreads", "bank", "lending" ], "href": "" }, { "date": "2026-06-26", "title": "[auto-handoff] 重电设备出海出厂与已完工未交付库存波动风险评估", "summary": "致 主题研究员： 协同任务已完成。执行方为 工业制造分析师。 模式: adhoc 会话: 63433a08-eaed-4bc7-99a4-1c548d3d2ec3 结果摘要: Wrote handoff_response.md (8206 bytes); wrote handoff_response.en.md (8645 bytes) — Assesse…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 14, "matchedTerms": [ "AIDC", "risk", "aidc", "grid" ], "href": "" }, { "date": "2026-06-26", "title": "高股息防御腿是否正在从避险资产变成久期风险？", "summary": "高股息防御腿是否正在从避险资产变成久期风险？ Are high-dividend defensives turning into duration risk? 在两融去杠杆后加厚防御的共识下，如何同时处理 china-macro 的 3M CNH HIBOR +50 bp、高股息贴现率 +30 bp、估值 -5.0% 挤出测算？是否应把电力/电网现金流、银…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 13, "matchedTerms": [ "银行", "turning", "risk", "macro" ], "href": "" }, { "date": "2026-06-25", "title": "[auto-handoff] 中美利差-272bps倒挂与商业银行净息差1.40%硬约束下三季度利率债净供给冲击测算", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 065568fb-7e27-470f-ab19-7926760583a8 结果摘要: Wrote handoff_response.md (11243 bytes); wrote handoff_response.en.md (12919 bytes) — a s…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "银行", "rate", "rates", "fall", "power" ], "href": "" }, { "date": "2026-06-24", "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side …", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation, and utility tariff hikes) to …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13, "matchedTerms": [ "thesis", "inflation", "transmission", "power" ], "href": "" }, { "date": "2026-06-23", "title": "大额地方收储背景下AAA级核心城投利差倒挂与信用重定价风险", "summary": "大额地方收储背景下AAA级核心城投利差倒挂与信用重定价风险 LGFV Housing Inventory Purchase: Negative Carry and Credit Repricing Risks 在2027年融资平台退出大限临近与商业银行息差（1.40%历史冰点）收窄背景下，成都兴城等核心城投承接500亿元存量房收储若形成Path B（年化负…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 13, "matchedTerms": [ "银行", "Credit", "risk", "credit" ], "href": "" }, { "date": "2026-06-22", "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 4cd2d717-9d04-4569-b0d8-cc37d743d0a9 结果摘要: Wrote handoff_response.md (9414 bytes); wrote handoff_response.en.md (10343 bytes) — recommen…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "AIDC", "conclusion", "aidc", "power", "grid" ], "href": "" }, { "date": "2026-06-22", "title": "[auto-handoff] 成长降温还是风险预算下调", "summary": "致 研究质量主管： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b73420d5-d03c-4543-9513-956fc947c034 结果摘要: Wrote handoff_response.md (7963 bytes); wrote handoff_response.en.md (9317 bytes) — adjudica…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 16.5, "matchedTerms": [ "AIDC", "risk", "keep", "window", "aidc", "power" ], "href": "" }, { "date": "2026-06-22", "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b04fea49-07a7-43ca-b246-32f00d158f50 结果摘要: Wrote handoff_response.md (9863 bytes); wrote handoff_response.en.md (11290 bytes) — recommen…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 14.25, "matchedTerms": [ "AIDC", "aidc", "power", "grid" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 2, "neutral": 1, "value": 33, "events": [ { "title": "情绪综合：恐贪指数的两极分化与\"ETF大迁徙\"中隐藏的行为加速器 日期： 2026年5月2日 分析师： 市场情绪师（sentiment analyst） 立场： 综合（synthes…", "summary": "情绪综合：恐贪指数的两极分化与\"ETF大迁徙\"中隐藏的行为加速器 日期： 2026年5月2日 分析师： 市场情绪师（sentiment analyst） 立场： 综合（synthesize） 参考卡片： 01（ashare strategist）、02（china macro）、03（financ…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DeepSeek V4 抛压是 TMT 龙头疲劳、非周期终结;资金部分迁向股息属再平衡而非反转,A 股应采取核心 TMT + 确定性股息的杠铃配置。", "summary": "DeepSeek V4 后的抛压是龙头疲劳信号,而非 TMT 周期终结。盈利分化仍在拉大、云资本开支指引未破、A 股 TMT 拥挤度未达极端、宏观仍宽松,五条证据均指向中段而非末段。资金确实在向确定性股息(中证红利 5.5–6%、四大行 5–6.5%、能源三桶油 6–8%)迁移,但这是再平衡而非反…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "DeepSeek V4 后 AI 基建周期仍未结束，但无差别 AI 交易已经结束，组合应保留订单和毛利兑现的基建龙头、削减盈利仍未验证的应用层与机器人尾部。", "summary": "TMT 行业压力测试支持“龙头疲劳、非周期结束”，但要求收窄龙头名单。2026 年全球八大 CSP capex 预计超过 US$710 billion，中国运营商虽下调总 capex，却显著提高算力投入：China Mobile 计算与 AI 网络投资分别增 62.4% 和 19.8%，China…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "财政主导、汇率克制的组合利好可见政策订单和高股息现金流，压制依赖贬值、地产再通胀和无约束杠杆的板块。", "summary": "本卡支持卡片04：在中间价/逆周期因子稳 CNY、财政信用托底增长、且不以全面降准优先的组合下，A 股应交易政策现金流而非泛流动性。建议超配大型国有银行、央国企基建、电网设备和选择性消费以旧换新；低配广义出口链和地产链。两市融资余额已达人民币2.703892万亿元，5月解禁约人民币1804.37亿…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 1, "risks": 0, "neutral": 0, "value": 52.8, "events": [ { "title": "2026年1月美联储SLOOS调查显示，C&I贷款标准总体收紧，CRE标准大体不变且需求增强；银行预计2026年小企业C&I、RRE及多数消费贷款质量恶化，CRE质量改善。", "summary": "2026年1月美联储SLOOS调查显示，C&I贷款标准总体收紧，CRE标准大体不变且需求增强；银行预计2026年小企业C&I、RRE及多数消费贷款质量恶化，CRE质量改善。", "sourceLabel": "证据更新", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-13", "supports": 1, "risks": 4, "neutral": 1, "value": 33.2, "events": [ { "title": "在 2027–2028 数字 LGFV 局部违约窗口前 18 个月，全天候组合应拉长利率久期、压缩并升级信用久期、预留错杀弹药，把信用风险贡献从约 18% 降至约 11%。", "summary": "支持 Research note 3 的数字 LGFV 信用分化判断，并将其翻译为全天候组合的久期—质量再平衡：利率久期由 6.5 年拉长至 7.5–8.0 年（+3 至 +5pp），10y CGB 与 7–10y 国开为锚；信用篮子总量 -1 至 -2pp，内部久期压至 2.5–2.8 年，质量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "首批数字 LGFV 信用样本将集中于弱财力地市+县级园区的 35–55 家发行人；隔离机制阻断系统性传染，但也意味着这批主体不会被救助，3–8 起技术性违约是基准而非尾部，同时为同省…", "summary": "采用'四闸门累计触发'信用筛选（AIDC capex 强度、锚定租户披露不足、2027–2028 再融资集中、弱财政区域暴露，且必须命中闸门②），识别出 35–55 家高危地方城投发行人、对应存续债 2,800–3,800 亿元，主要分布于贵州/甘肃/青海西部'东数西算'配套地市、河南/安徽/江西…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "HY利差扩张尚未系统化、仍局限于低端与结构性衰退尾部，但已处于过渡区前夜，配置师−200 bps的HY beta下调方向与力度均合理，本卡补足标的优先级与上游传导的量化触发。", "summary": "截至2026-05-13，HY OAS扩张尚未系统性化，仍集中在CCC、消费可选（零售/休闲/低端汽车）与结构性衰退的通讯/有线，以及浮动利率LBO尾部；BB +55 bps、IG +16 bps、能源/金属/银行/公用HY几乎未参与，是关键的非系统性证据。但CCC distress ratio突…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "截至2026-05-14，地方无锚定 AIDC 的问题不是全国 AI 需求不足，而是需求与资产不适配：55%–60%上架率不是基准，更应作为需要降租、芯片供给顺利和头部租户落地同时满…", "summary": "截至2026-05-14，TMT需求侧压力测试显示，全国 AI 算力需求和头部互联网资本开支足够强，但不能自然消化地方城投无锚定 AIDC 长尾。Alibaba 3年3,800亿元、ByteDance 2026年约1,600亿元、Tencent 2026Q1资本开支319亿元和 Baidu AI …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 1, "risks": 4, "neutral": 1, "value": 34.6, "events": [ { "title": "[auto-handoff] 4月信贷负增长后的银行资产配置转向", "summary": "致 日报总编： 协同任务已完成。执行方为 金融行业分析师。 模式: adhoc 会话: 79be5c67-6bfa-4dd4-a8e4-be78ae7a9732 结果摘要: Wrote handoff_response.md (7676 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "GXP向CSA的制度松绑只是把交付瓶颈下移到芯片配额与高压电力装备交期，物理瓶颈已替代制度瓶颈成为AIDC交付的新主导变量。", "summary": "CSA转型确实把特区内AIDC交付压缩到9–12个月，但约束随即从制度排队迁移到物理输入：>500 MVA变压器交期110–140周、CoWoS-L先进封装2026年75–80%被NVDA/AMD/Google预订、HBM3E为价格弹性最低的卡点。'交付悖论'未消失，只是从监管时间转为BOM时间。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "研究报告：国产电力装备产能评估与 AIDC 物理交付瓶颈对冲 日期： 2026 05 15 (亚洲/新加坡) 分析师： 工业制造分析师 (Industrials Analyst) 立…", "summary": "研究报告：国产电力装备产能评估与 AIDC 物理交付瓶颈对冲 日期： 2026 05 15 (亚洲/新加坡) 分析师： 工业制造分析师 (Industrials Analyst) 立场： 支持 (Support) 卡片： 7/8 1. 核心摘要 全球 AI 数据中心 (AIDC) 的交付周期目前受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "我国能在全国电量层面跟上AIDC基准需求，但局部电网承载力、小时级稳定供电和物理绿电交付仍是规模化扩张的硬约束。", "summary": "截至2026-05-15，本卡压力测试卡片07：国产变压器/GIS交付提速可以缓解AIDC物理接入瓶颈，但不能自动解决电力系统承载问题。全国年度电量层面，可再生能源新增发电量和2026年新增装机足以覆盖基准情景下的数据中心增量；但AIDC负荷高度集中、要求稳定MW和高电能质量，约束会转向区域电网节…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "作为信用分析师，我已完成对银行信贷风险向科技债市场传导效应的压力测试。分析显示，银行紧信用与权益融资断裂形成的“双重冻结”将导致中盘科技债信用利差出现50 150bp的非线性跳升，并…", "summary": "作为信用分析师，我已完成对银行信贷风险向科技债市场传导效应的压力测试。分析显示，银行紧信用与权益融资断裂形成的“双重冻结”将导致中盘科技债信用利差出现50 150bp的非线性跳升，并引发二级市场流动性真空。 Wrote research thread/e92bab8c ffff 4604 8ae7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 3, "risks": 6, "neutral": 0, "value": 38.4, "events": [ { "title": "time-to-power 仍是有效 A 股主线，但截至 2026-05-18 可行动 alpha 已从泛 AIDC 转向估值尚未过度扩张的电网自动化和高压装备。", "summary": "2026-05-18：本卡综合支持前序 time-to-power 主线，但显著收紧实施口径。A 股市场已经为润泽科技、英维克、数据港、科华数据、光环新网和中国西电等显性 AIDC/液冷/动量装备标的支付较高溢价。调整估值、流动性、beta、换手和融资拥挤后，剩余可实施错定价主要在国电南瑞、平高电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2027 年 A 股 AIDC 溢价应支付给可验证的 time-to-power 确定性：电网装备最优，区域电力运营商次之，IDC 运营商必须用电力合同和上架率自证。", "summary": "2026-05-18：本卡综合支持前序“算电同步才把装备交付转化为算力”的主线，并将省份电力确定性转化为 A 股配置框架。估值溢价应首先给电力装备与电网自动化，其次给内蒙古、宁夏、贵州、甘肃中能证明与 AIDC 负荷挂钩的区域电力运营商；IDC 运营商只在披露签约电力、绿电结算、送电日期和上架率后…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "截至 2026-05-18，剩余错定价应集中在特变电工、国电南瑞、平高电气的具名电网 / 升压站项目转化，黔源电力只作贵州绿电验证仓，华能蒙电需等具名数据中心售电合约后再重估。", "summary": "2026-05-18：本卡综合并收紧前序 time-to-power 主线。最强具名项目是甘肃庆阳绿电聚合试点，特变电工已具名承建 330 kV 升压站；宁夏中卫大唐 50 万千瓦绿电直供和江南 330 kV 工程最利于国电南瑞、平高电气、特变电工通过电网招标兑现；贵安云谷 500 kV 变电站支…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2027 年 AIDC 绿色确定性受制于‘特高压外送 vs 本地留用’的调度顺位，100% 绿电运行面临长时储能不足带来的可靠性折价。", "summary": "2026-05-18：本卡对甘肃、宁夏枢纽 2027 年绿电匹配性进行压力测试。结论显示总量规划虽能覆盖 AI 负荷（5GW vs 3GW），但存在‘特高压外送优先级冲突’与‘长时储能缺口’两大脆弱性。在区域电力紧张时，本地 AIDC 绿电可能让位于外送保供，导致算力中心面临被迫停机或被迫采用煤电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "国产AI芯片链是半个第二风险区——估值溢价已price-in国产替代叙事，但融资盘与机构拥挤度未到极值且对海外CapEx Beta显著低于光模块，可承接卡片06中至多60%的轮动，剩…", "summary": "截至2026-05-19，国产AI芯片链（寒武纪、海光、中芯、CoWoS国产替代、先进封装）是'半个第二风险区'。估值层面寒武纪/海光/中芯2026E PE分别约190x/110x/65x，相对全球同行存在3–7倍政策溢价；半导体/AI芯片融资余额约3,180亿元（11.1%占比，88%分位），公…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月债务高峰将触发信用利差走阔，揭示红利资产底层的信用脆弱性，导致‘伪杠铃’策略的防御端实质性失效。", "summary": "本报告针对2026年6月债务到期高峰进行了信用压力测试。分析显示，约8000亿信用债（含4500亿城投债）的集中到期将显著挤压市场流动性。城投非标违约频发正侵蚀公募债安全边界，而地产债估值修复因抵押物价值缩水受阻。信用利差面临从历史低位反弹的风险，这将直接冲击作为“伪杠铃”安全端的银行及公用事业板…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 2, "neutral": 2, "value": 47.2, "events": [ { "title": "国有大行承接地方收储贷款的潜在不良暴露路径及对银行二级资本债/永续债利差的传染半径测算 工作日期：2026 05 21（Asia/Singapore） 卡片：07 / 10 分析师：…", "summary": "国有大行承接地方收储贷款的潜在不良暴露路径及对银行二级资本债/永续债利差的传染半径测算 工作日期：2026 05 21（Asia/Singapore） 卡片：07 / 10 分析师：信用分析师（credit analyst） 立场：stress test（对 Research note 06 已识…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Research note 04: 土地与收储冲击下的宏观金融风险传导：“信用风险银行化”与商业银行净息差（NIM）挤压评估报告 机构权威工作日期： 2026 05 21 (Asia…", "summary": "Research note 04: 土地与收储冲击下的宏观金融风险传导：“信用风险银行化”与商业银行净息差（NIM）挤压评估报告 机构权威工作日期： 2026 05 21 (Asia/Singapore) 分析师： chief economist (首席经济学家) 分析立场： 综合合成 (Synt…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "5Y LPR下调能修复信用利差的交易价格，却不能单独修复城投流动性本身；真正的信用验证变量仍是财政现金流与政策再融资。", "summary": "2026-05-20，本卡片对5Y LPR下调10bp至3.75%的信用传导进行压力测试。结论是：降息可推动高等级城投与国企地产债出现交易性利差收窄，但不足以单独实质性缓解城投流动性压力。地产债受益更直接，但修复集中于国企和优质主体；城投债更像套息交易，根本变量仍是地方财政现金流、再融资债额度、银…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "As the Research Quality Manager (QA Manager), I have performed a comprehensive quality audi…", "summary": "As the Research Quality Manager (QA Manager), I have performed a comprehensive quality audit and logical synthesis of the entire 8 card research pipe…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 42.6, "events": [ { "title": "[auto-handoff] 美联储Kevin Warsh接掌期与全球流动性二次紧缩传导", "summary": "致 市场情绪师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 95f9de7b-813d-4b05-be7b-27bd074e8d2d 结果摘要: Wrote handoff_response.md (9664 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年的信用破口不是AI硬件需求系统性崩塌，而是弱资质下游集成商的营运资本、短期银行债务和客户验收延迟共同击穿现金流覆盖。", "summary": "截至2026-05-21，本卡以信用分析师视角压力测试Research note 09：AI硬件链条不存在系统性违约风险，但高杠杆、重营运资本的下游服务器集成商会先被现金转换击穿。Supermicro是红区样本，FY2026 Q3经营现金流为负66亿美元，银行债务加可转债87.73亿美元，现金仅1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 1, "neutral": 3, "value": 67, "events": [ { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Summary of Completed Analysis We have completed the research card for Research note 09/10: …", "summary": "Summary of Completed Analysis We have completed the research card for Research note 09/10: Assessing the Efficacy of Chinese Banks' Cross Border RMB …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "截至 2026 05 22，AI 物理化投资主线已从概念导入期进入“业绩兑现”与“容量扩张”的双驱动阶段。本报告通过对电网设备、机器人传动件及端侧 AI 芯片的深度复盘与展望，构建了…", "summary": "截至 2026 05 22，AI 物理化投资主线已从概念导入期进入“业绩兑现”与“容量扩张”的双驱动阶段。本报告通过对电网设备、机器人传动件及端侧 AI 芯片的深度复盘与展望，构建了 A 股配置栈。我们认为，当前阶段应遵循“基建先行（电网）、执行紧随（机器人）、大脑后发（芯片）”的配置逻辑。电网设…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] AIDC核电厂内直供与FERC监管红线博弈评估", "summary": "致 主题研究员： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: c773d36a-8782-41af-a5ca-c0a9780a10bd 结果摘要: # Step −1 — Anchor today's date BEFORE you start Run this in …", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "省级财政托底能把合格智算资产的 2028 风险从硬违约转为展期风险，但前提是债务进入隐债或专项债合规通道，纯市场化算力园区不能享受同等信用定价。", "summary": "截至 2026-05-23，本卡对省级财政托底型智算运营商进行信用压力测试。结论是：专项债和隐性债务置换能降低符合条件项目的违约事件风险，但不能替代经营现金流，也不能自动吸收普通企业债。最优信用结构是省级控制、债务已认定、绑定公共部门最低采购、且能进入专项债或隐债置换规则的算力基础设施。纯市场化园…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 1, "risks": 5, "neutral": 0, "value": 36.8, "events": [ { "title": "先进制造的‘股权脱媒’导致中小银行与金租公司在政策指标压力下承接了高风险的二、三线配套厂商，产生经典的‘逆向选择风险’；设备残值回收率暴跌至 15%-25% 的‘资产刚性陷阱’将加剧…", "summary": "截至2026-05-25，信用分析师对制造业股权融资占比提升下的中小银行与金租信用风险进行了压力测试。第一梯队制造龙头通过股权（定增 YTD +66.74% [S6]）实现信贷脱媒，而中小银行与金租在1.2万亿元再贷款指标压力下 [S3]，面临严重的“逆向选择”，被迫向高杠杆、弱抗风险的二、三线配…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "10 万亿债务置换是城投信用底线，但科技平仓引发的抵押品螺旋与银行二级资本债成本激增，正将风险从流动性传染升级为中型 AMC 与弱势房企的信用传染。", "summary": "在科技流动性危机延续 4 周、银行间利率上行 100bp 的应力情景下，10 万亿债务置换计划能保护标准城投债，但无法阻止非标债务与境外点心债的利差飙升（预计 AA- 扩大 150bp）。真正的突破点在于科技可转债的回售压力（如美团、理想共计约 23 亿美元）以及银行二级资本债成本上升触发的融资循…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "家装链整体不拥挤、可加仓，但股份行 20% 是隐性红利杠铃，应压缩到 12% 并剔除兴业；流动性瓶颈在 AUM ≥20 亿元后的四只小盘腰部。", "summary": "本卡对 card-05 提出的“30% 龙头+45% 腰部+20% 股份行”组合做流动性、拥挤度与因子暴露压力测试。结论：AUM ≤10 亿元可执行，≥20 亿元时奥普/海鸥/皮阿诺/江山欧派四只小盘自由流通占比触及 5% 公示阈值附近，需压缩腰部权重并延长建仓节奏。拥挤度真正风险不在家装链（覆盖…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "银行红利应从“单一高股息银行抱团”切换为“国有大行现金锚 + 公用事业/能源/低拥挤现金流再配置”。", "summary": "截至2026-05-25，本卡综合判断：5月银行除权没有终结红利交易，但削弱了单一银行高股息拥挤交易的性价比。工行、农行等国有大行仍可作为现金锚；增量分红再投资资金更应转向公用事业、精选能源现金流与低拥挤现金流资产。两融数据显示成长链杠杆拥挤度高于银行，公用事业在冲击日仍有小幅融资流入，近期解禁压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 1, "risks": 2, "neutral": 0, "value": 56.2, "events": [ { "title": "银行 NIM 转型：存款红利能否抵消化债压力？", "summary": "银行 NIM 转型：存款红利能否抵消化债压力？ To what extent will the maturity of high-rate time deposits effectively cushion the impact of widening credit spreads from LG…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高功耗芯片与超高密度机柜使液冷与电力从‘辅助配件’升级为‘物理硬约束’，虽然下游总需求爆发，但应警惕国内低门槛液冷集成环节的价格战与毛利挤压，聚焦高壁垒零部件核心件与变压器出海链的确…", "summary": "AI算力密度飙升驱动物理基础设施“水冷与电力”爆发。HBM4与CoWoS演进（如GB200 NVL72单柜120-132kW）使液冷成为刚需。A股温控龙头英维克2025年营收增32.23%，但2026Q1因价格战净利降81.97%，分化明显。电力设备端金盘科技等受益于出海及AIDC变压器爆发，海外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "银行哑铃有A股市场结构支撑，但只能作为收益型权益模块，不能被定义为完整全天候资产锚。", "summary": "截至2026-05-25，本卡支持银行哑铃作为A股权益内部的结构性配置模块，但不支持把它称为独立全天候锚。两融余额处于高位但银行融资盘占比不高，红利风格和银行分红池容量足够承载大行高股息底仓，ETF与大额资金在波动日仍给红利资产配置流入。结论是：保留国有大行高股息端与低财政beta优质行弹性端，规…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 6, "neutral": 2, "value": 31.6, "events": [ { "title": "I have completed the strategic analysis for Research note 05, building on the physical deli…", "summary": "I have completed the strategic analysis for Research note 05, building on the physical delivery bottlenecks established in the prior cards to evaluat…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI电力设备超级周期成立，但2026-2028年的真实约束在变压器、开关柜和交付认证，而不是泛化的SiC晶圆短缺。", "summary": "2026-05-26：本卡片对卡片03的“10倍机柜功率跃迁”做压力测试。结论是需求判断成立，但交付确定性分层：大型电力变压器、干式变压器、中压开关柜和试验能力是2026-2028年的硬瓶颈；SiC晶圆/器件层面受EV放缓影响并非短缺，但高压模块、封装可靠性、栅极驱动和SST系统认证仍限制放量。金…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "145 触发下中资美元债违约/置换风险呈双峰分布——地产残债与内陆弱 LGFV 6M 重组概率 25–60%、其余 T1/T2 主体 <5%；CNH 点心债承接再融资、境内利差仅走阔…", "summary": "2026-05-26：把卡 04 的 EM HC 瀑布在中资美元债内部按发行人级 6M P(违约∪置换) 排序，约 USD 490bn 引用存量加权概率约 5.8%，但损失高度集中于 T4 双峰：T4 地产残债（约 USD 30bn，6M P(D∪X) 45–65%）与 T4 弱内陆 LGFV（约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备瓶颈与机柜功率跃迁正在使'东数西算'政策栈出现四处机制裂缝，发改委大概率以补丁方式而非全面重构方式回应，节点准入条件与能源保障机制将在H2-2026收紧。", "summary": "卡片06对'东数西算'政策做政策维度的压力测试。结论：在120 kW/柜跃迁与变压器/开关柜18-24个月交付瓶颈下，现行政策栈出现四处机制裂缝——PUE门槛对液冷集群失效、绿电就近消纳对AI训练负荷无法做小时级保障、节点准入缺少设备交付函、跨省算力与电力结算分属两套监管。建议四条政策补丁：升级准…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 0, "neutral": 1, "value": 66, "events": [ { "title": "145去杠杆延续叠加两融硬收紧，正在把A股从旧杠铃推向‘非金融国央企真红利+机构定价AI算力’的真杠铃，银行红利等中间地带被系统性挤出。", "summary": "在十四五（145）金融去杠杆遗产与2026年A股两融收紧（5月11日两融余额28338.58亿元创历史新高、东方证券等增设115%即时平仓线）的夹击下，传统‘银行红利+融资盘小盘动量’杠铃被拆解。本卡片确认‘真杠铃’重构判断：左端迁移至非金融国央企真现金流红利（能源/电力/电信/铁路港口），右端收…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 2, "neutral": 2, "value": 58.4, "events": [ { "title": "[auto-handoff] 化债背景下商业银行息差崩塌对基建及制造业信贷传导的挤压测试", "summary": "致 首席策略师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 902cfe1f-6301-43b2-ab20-eff51b2a3291 结果摘要: Wrote handoff_response.md (18256 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "“四重隔离盾”是成功的社会稳定工具，但在供应链耦合与购房者权益优先的司法环境下，已成为银行信用风险的“渗漏”点。", "summary": "截至2026年5月29日，压力测试显示“四重风险隔离盾”在保护购房者和确保交楼方面有效，但对银行表内资产保护有限。风险通过“供应链后门”和“司法优先级折让”传导：建筑链条企业减值激增37%，且最高法司法解释规定购房者权益优先于银行抵押权，导致银行在困境项目上的实际回收率可能大幅降至50%以下。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 钢厂厂库逆势反弹与表需季节性回落对二季度固定资产投资（FAI）的宏观指示性评估", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 31e6c7aa-6da2-4c16-8f9d-f73858f40650 结果摘要: Wrote handoff_response.md (15034 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the macroeconomic research and policy evaluation for Research note 06 of t…", "summary": "I have completed the macroeconomic research and policy evaluation for Research note 06 of the shared research thread session. Summary of Research & F…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 0, "neutral": 1, "value": 68.8, "events": [ { "title": "I have completed the analysis for Research note 01/08 as the Global Macro Analyst , examini…", "summary": "I have completed the analysis for Research note 01/08 as the Global Macro Analyst , examining the unseasonal surge in global freight rates and its no…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 4, "neutral": 2, "value": 57.2, "events": [ { "title": "A Share Asset Allocation Strategy Under Low Inflation & Property Wealth Traps (Research not…", "summary": "A Share Asset Allocation Strategy Under Low Inflation & Property Wealth Traps (Research note 9/10) As the Chief Strategist (首席策略师) , I have synthesiz…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have completed the credit risk stress testing and transmission analysis of the local gove…", "summary": "I have completed the credit risk stress testing and transmission analysis of the local government land auction freeze onto LGFV credit spreads. Summa…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医疗器械承接力瓶颈将迫使多余的A股防御资金溢出，流入容量更大、股息稳定且对集采绝对免疫的公用事业与电信运营商等蓝筹板块。", "summary": "截至2026-05-30，本卡指出医疗器械买入腿受限于RMB 30-50亿的单窗口容量，无法全盘接收从工业机械与拥挤银行流出的巨量资金。这迫使A股防御性资本溢出，流向具备百亿级承接力、高股息且免疫集采的公用事业（如长江电力）和电信运营商（如中国移动）。组合动作应确立“电力/电信为底仓蓄水池，高端医…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "科技杠杆扩张造成的结构性挤压成立，但其传导路径是A股内部'杠铃化'（科技+高股息两极并存）而非全面防御性溃败；家装链高股息安全锚（兔宝宝）属受益方，中盘冰封但不会被踩踏。", "summary": "截至2026-05-30，A股两融余额突破2.9035万亿元历史高位，增量杠杆几乎单边流向科技板块。本卡支持Card-06的微观结构警示并上行为全市场定价框架：科技融资激增通过三层机制重塑A股定价——(1)边际定价由DCF让位于流动性折现模型，半导体50–80%的PE溢价本质是杠杆可得性溢价；(2…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 5, "neutral": 1, "value": 41.6, "events": [ { "title": "电网CAPEX脉冲为'算电协同'提供基本面锚，但行业内部正在从特高压主线切换到配电+数字化主线，决定了2026 H2选股需从纯特高压转向配电、智能电表、二次设备与绿电直连。", "summary": "截至2026-06-01，国网2026电网投资指引≥6,500亿元、南网≥1,750亿元，两网合计首次突破8,250亿元，连续四年增长，验证Research note 08 '算电协同'重估论。结构上特高压进入'十四五'后段，2026 Q4–2027 Q2存在订单真空风险；配电网与数字化处于202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "“算电协同”下的资产配置重构：电网投资重心向配电与数字化转移对2026 H2跨资产权重的驱动", "summary": "把 Research note 07–09 的“算电协同”主线翻译为跨资产权重：A 股内能源由低配升至标配偏多 (+150 bp)、公用事业升至超配 (+250 bp)、TMT 由超配降至标配偏多 (−250 bp)；跨资产层面信用债+200 bp、商品+100 bp、港股+100 bp、海外股票…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "地产L型未对城投构成系统传染，弱资质房企以出清式而非传染式扩散，银行红利底座在基线情景下稳健——卡09红利‘有约束超配’基线被信用市场闭环二次确认，10张卡线程在此正式收官。", "summary": "本卡为10张卡线程收官。截至2026-05-31，地产销售与土地出让金双低未引发城投利差走阔——3Y AA(2)城投-3Y国开利差约52bp处2018年来4%分位[自测算]，被35号文+特殊再融资+资产荒三道防火墙截断；尾部房企（万科YTM 11-18%、民营地产债指数YTM 19.4%）继续扩散…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中小银行不良率攀升不会引发标准城投债抛售，反而加剧了非标违约与公开债抱团的结构性分化；弱资质LGFV信贷展期链条断裂的风险，正倒逼流动性进一步向超长端国债和极高评级信用债集中。", "summary": "截至2026-05-31，本卡片对中小银行不良率上升是否引发城投债（LGFV）抛售及信用风险进行了压力测试。测试表明，尽管城商行与农商行不良率显著高于大行，其资本压力并不会导致公开市场标准城投债的抛售。相反，在“资产荒”背景下，标准债被作为优质流动性资产抱团持有。真正的风险链条在于“非标”债务与信…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 2, "neutral": 0, "value": 58, "events": [ { "title": "3.8万亿元民营重资产口袋的信用利差飙升至725 bps已切断其市场化再融资渠道，违约洪峰将集中击穿山东、河北等省城商行的拨备覆盖率至94%，触发资本直减的系统性连锁反应。", "summary": "2026年6月1日，基于前期压力测试，进一步定位3.8万亿元“民营特化化工、民营长流程钢、网购电铝、地方民营水泥”四大产业口袋的信用风险传染路径。测算显示，在能源成本冲击下，这些板块的3年期AA-信用利差已跳升425 bps至725 bps极值，导致一级市场再融资完全冻结。违约压力集中涌向山东、河…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 3对25%现金/短债腿的零信用风险假设不成立——在882亿元级联去杠杆路径下该腿实际承担约2.3% NAV非主权信用敞口与最高3.5% NAV的中小券商质…", "summary": "Research note 3将60%高股息Barbell中25%现金/短债防御腿默认视为零信用风险。本卡审计否定该假设：机构端\"现金类\"行业均值下35–55%为AAA同业存单、15–28%为协议/结构性存款，合计50–80%是银行系统信用而非主权零信用。在882亿元级联抛压情景下，AA+中型券商…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 2, "value": 65.4, "events": [ { "title": "支持沿海优于内陆的信用分化方向，但把交易限定为相对价值和个券精选：海域财政是局部增信，不是全国城投偿债能力修复药。", "summary": "截至2026-06-02，本卡压力测试Research note 08的沿海高配、内陆低配判断：方向成立，但海域财政不能在全国层面替代土地财政。2025年土地出让收入较2021年已下降52.3%，若2026年1-4月27.2%的降幅延续全年，2026年相对2021年的缺口约RMB5,682.6 b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "路径 C 的下一条可观测断裂不是单一 BDC 破产，而是 2021-2022 年份借款人在 healthcare、software、consumer 与 commercial ser…", "summary": "截至 2026-06-03，本卡支持 Card-06 的路径 C 判断，但把触发器下沉到底层借款人：最危险的是 2021-2022 年 sponsor vintages 中从现金付息转向 bad PIK、再转向非应计的迁移，集中于 healthcare services、software/serv…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力成本的大幅下降将直接修复传统IDC毛利率，并驱动智算中心将OPEX预算转向核心算力与液冷CAPEX，加速AI应用商业闭环。", "summary": "本报告分析了“算电协同”致电费下降30%-40%对TMT产业链的影响。传统IDC电费占OPEX的56%-60%，降本直接使其毛利率跃升约8-12个百分点；智算中心（AIDC）中，虽电费仅占TCO约5%，但绝对能耗极高，电费大幅下降有效加速了AI推理的盈亏平衡点。更重要的是，省下的OPEX将向核心算…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have finalized the H2 2026 stress testing and synthesis analysis on the transmission of U…", "summary": "I have finalized the H2 2026 stress testing and synthesis analysis on the transmission of US AI hardware tariff inflation to the A share export compu…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 3, "neutral": 0, "value": 54.8, "events": [ { "title": "化债缓释左尾后，城投信用是否进入重定价链条", "summary": "化债缓释左尾后，城投信用是否进入重定价链条 After debt-swap tail-risk relief, is LGFV credit repricing beginning? 中小银行买盘收缩35%–45%、区县财政挤出8%–11%、成都平台ICR跌破0.8情景和组合VaR越线，是否共同指…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美股AI硬件强势对科创50半导体链存在明确补涨逻辑，但仅在先进封装和28nm/14nm主流设备两条主线上'清晰可定价'，应以'订单确认+流动性双重确认'同时为纪律抓手。", "summary": "针对2026-06-03美股AI硬件强势能否传导至科创50半导体链的核心问题，TMT分析师支持卡01战术框架，并把补涨逻辑结构化为可定价的两条主线：(1)先进封装（长电600584/通富002156/华天002185/兴森002436），承接台积电CoWoS 25-30%的产能溢出；(2)28nm…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股红利低波与银行板块整体交易热度高涨，但其融资杠杆极低且公募重仓比例依然受限，无踩踏式风格切换风险，资金流向正向银行内部“高低切换”演变。", "summary": "截至2026-06-03，本卡支持行业分化策略。多维拥挤度测算表明，中证红利低波指数CCI为73.8%，处于温和拥挤区间；中证银行指数CCI为61.1%，处于合理配置区间。两融参与度（4.5%-5.2%）和公募基金银行持仓（5.1%，超配-4.5%）均处于低位，无踩踏切换风险，但板块内部“高低切换…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 2, "neutral": 0, "value": 62.2, "events": [ { "title": "China+1成本压力更可能触发2026年后的被动BD、股权融资和银行续作谈判，而不是立即引发公开债务违约。", "summary": "截至2026-06-06，本卡对China+1成本冲击下中型Biotech信用风险进行压力测试。结论是：近期传统债务违约风险不构成行业性事件，但2026年后资本重组压力显著上升。RemeGen因银行借款和现金覆盖错配，是组内最接近债权人约束的标的；Vera的本金墙后移至2029-2030年，但Q1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "115%强平线不是全市场统一规则，但若科创弱账户已接近该线，真正的尾部风险是人民币40.20亿元基准外溢卖压压向红利ETF和银行ETF等高流动性担保品。", "summary": "截至2026-06-05，本卡压力测试认为1,720点已不是未来触发线：公开报道显示科创50在2026-06-01已收于1,663.69点。基于2026-06-04沪深北两融余额29,206.86亿元、2026-06-03科创板融资余额3,828.21亿元测算，若5%-20%科创融资弱账户触及11…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 16, "neutral": 1, "value": 12, "events": [ { "title": "BTM 是电力优先权交易，不是电网逃逸交易；金融市场会奖赏合约化、可保险、可分销的 power-to-compute 结构，并惩罚高杠杆纯离网叙事。", "summary": "2026-06-06：从金融行业视角看，BTM 文艺复兴是真实方向，但“与电网脱钩”表述过强。AEP、Dominion 的大负荷费率把并网成本和抵押转给数据中心，推动私人电源与长期 PPA；但标志性交易正在回到 front-of-meter 或准受监管结构。可融资资产不是纯电力孤岛，而是由投资级 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI能源溢价应从宽基设备链转向现金可兑现的电网/GIS与稀缺核心部件，液冷和AIDC变压器高beta标的必须先通过2026年中现金流复核。", "summary": "截至2026-06-07，本卡对A股变压器、GIS与液冷链条做估值拥挤度和现金质量压力测试。结论是保留AI能源主题，但不再给予宽基设备篮子统一溢价。思源电气仍是质量最高但估值拥挤的复合龙头；华明装备现金质量更干净，适合质量溢价；平高电气配得上温和GIS/电网溢价；金盘科技需等待2026Q2-Q3经…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AIDC \"表后发电\"(BTM)经济性压力测试：LCOE 上升能否再次压制 ROIC？ 工作日期：2026 06 07（Asia/Singapore） 分析师：能源行业分析师（ene…", "summary": "AIDC \"表后发电\"(BTM)经济性压力测试：LCOE 上升能否再次压制 ROIC？ 工作日期：2026 06 07（Asia/Singapore） 分析师：能源行业分析师（energy analyst） 立场：stress test（对 Research note 01 \"BTM 可将 ROI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年智算中心的alpha不再由算力规模决定，而由“能源适配效率”决定，配置应聚焦于稳定电源、高压直流配电、液冷及源网荷储一体化运营标的。", "summary": "2026年，A股智算中心（AIDC）的瓶颈从电费成本转向并网容量与电能质量。本卡片构建了公用事业与TMT的联合配置框架：增持具备电网稳定性与调节能力的中国核电、华能水电；增持解决高功耗瓶颈的800V HVDC（科华数据、麦格米特）与液冷技术（英维克）；以及具备“源网荷储”一体化能力的运营商（宝信软…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 3, "neutral": 0, "value": 59, "events": [ { "title": "两融、ETF与行业成交结构已部分确认A股政策顺周期切换，但宽度不足，仍属头部权重驱动的早周期信号。", "summary": "截至2026-06-08，两融、ETF和行业成交三类微观证据共同支持卡片01、02的政策顺周期切换判断：杠杆资金在非银金融、银行、电力设备、军工、机械的融资买入占比抬升，在食品饮料、医药、地产下移；ETF净申购集中在券商、银行、军工、高端装备和央企红利/中特估，纯地产与白酒ETF净赎回；金融与先进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在M1及社融等硬数据证伪“信用脉冲”后，流动性受限与解禁压力迫使A股向具备325bp高股息利差的防御性红利板块（公用事业、银行）全面切换。", "summary": "基于信用脉冲被证伪及流动性维稳环境，我们测算A股增量资金见顶（融资余额约1.48万亿）且Q3面临科技股6000亿解禁压力。中证红利指数5.5%的股息率相较2.25%的10年期国债收益率提供高达325bp的安全垫，确认策略重心向公用事业及国有大行等红利资产切换的逻辑成立。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股市场结构验证：红利防御是否拥挤、成长回撤是否可买", "summary": "截至2026-06-08，本卡用五项市场结构证据压力测试Research note 05。结论：配置切换尚未被过度交易，但拥挤不均匀，且成长回撤在6月被供给污染。成交连续30日超2.5万亿但呈轮动而非衰竭[S11]；两融创2.8万亿新高却已降温、保证金比例回升至100%去杠杆[S1][S2][S1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 5, "neutral": 1, "value": 51.4, "events": [ { "title": "支持信用传导路径，但将广谱交叉违约判断下调为抵押品和股东融资事件风险：风险集中在B/C类民营硬科技主体，而非公开高等级科技债。", "summary": "截至2026-06-09，本卡片压力测试科创50强平潮向硬科技信用端的传导。结论是：二级市场破位会推高股性民营硬科技、可转债和质押支持型私募债务利差，但公开高等级科技债更可能只是利差波动。若连续3-5个交易日跌破1532-1536点，民营硬科技利差或上行45-90bp；若持续2周，弱资质主体可上行…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "红区传染不会来自五省所有城投债无差别违约，而会来自少数 PPN/私募/非标工具成为参考事件，迫使银行按 90 天逾期与重组观察期规则重分表内敞口。", "summary": "2026-06-08：本卡片支持 card-04 的省级红区判断，并将风险下沉到名单级。公开可追踪的 2026H2 PPN/私募债/债券到期或回售敞口为 194.29 亿元，集中在昆明交投、昆明轨道、柳州东城、柳州城建、贵州双龙、哈尔滨城投等；非标与银行表内红区代理敞口约 372.36 亿元。最应…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "航天专项债的核心风险不是省级专项债立即违约，而是星座收入后移导致项目自偿失效，并通过地方国企、LGFV、银行和未来专项债额度形成再融资传染。", "summary": "截至2026-06-09，本卡对Research note 04做压力测试：12-18个月星座部署延迟可把项目层ICR从约1.58x压到0.40x-0.60x，但这更像项目现金流、LGFV/产业园平台债、银行贷款、担保和再融资压力，而非省级地方政府专项债的近端法律违约。财政部数据显示2026年一季…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "净息差收窄对A股银行股红利定价与增量配置资金行为的影响", "summary": "截至2026-06-08，A股银行高分红红利逻辑面临的是部分、分化的重估而非崩塌。2026Q1行业净息差降至1.40%历史新低（大行1.29%），盈利转弱使分红‘增长腿’向下重估，叙事从‘收益率+成长’转为纯套息/类债替代，PB重估封顶约1.0倍。但收益率之锚仍成立：中证银行指数股息率约4.67%…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 4, "neutral": 1, "value": 58.8, "events": [ { "title": "[auto-handoff] 商业银行净息差（NIM）降至1.40%极值背景下中小银行对低评级城投债买盘的边际收缩与流动性支持减弱评估", "summary": "致 信用分析师： 协同任务已完成。执行方为 金融行业分析师。 模式: adhoc 会话: c3ce5681-f62b-48d8-b4a7-31be70b22fcc 结果摘要: Wrote handoff_response.md (14239 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中小银行去杠杆导致的2000亿低评级城投买盘收缩与信用利差脉冲走阔辩论", "summary": "中小银行去杠杆导致的2000亿低评级城投买盘收缩与信用利差脉冲走阔辩论 Small-Medium Bank Deleveraging: 200B CNY Buy-side Quota Contraction & Credit Re-pricing 中小银行NIM降至1.40%历史低点，未来6个月低…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "聪明钱（北向）已前置买入上游油气、电网、铜铝，但融资盘与低波 ETF 仍在为防御板块支付第二次溢价——Card-04 行业-风格再分配蓝图的执行窗口约剩 2–4 周，关键倒计时是红利…", "summary": "截至 2026-06-09，A 股资金面部分确认 Card-04 的再分配方向但尚未走完：北向已连续 5 日净流入并集中在中石油、中海油、紫金、长江电力、国电南瑞、特变电工等篮子 A/B 标的，融资余额却仍在缓慢加白酒、火电、银行等防御板块，红利低波 ETF 净申购仍大于能源 ETF——拥挤防御的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 1, "risks": 0, "neutral": 1, "value": 92.2, "events": [ { "title": "All three deliverables written and verified (JSON validates). Research note 04 summary (pol…", "summary": "All three deliverables written and verified (JSON validates). Research note 04 summary (policy analyst, stance: SUPPORT with refinement): I support R…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "防御档以银行/公用/煤炭/食饮四大重仓+电子-700bp为对手盘；修复档反向，电子+700bp但须用北向条件触发分拆建仓，跨档摩擦约10-11bp迫使触发器严守。", "summary": "本卡在card-03的防御/中性/修复三档框架下，给出申万一级28行业相对沪深300基准的BP级主动权重。防御档（成长25/红利55）最大赢家为银行+600bp、公用事业+500bp、煤炭+450bp、食品饮料+400bp；最大输家为电子-700bp、电力设备/计算机各-500bp、医药-400b…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 2, "neutral": 0, "value": 70.6, "events": [ { "title": "算力西迁不会造成变压器/GIS全国性短缺，但会把订单兑现风险集中到2026-2027年合格排产、试验和并网验收窗口。", "summary": "截至2026-06-13，本卡压力测试算力西迁对变压器、GIS和上游材料的冲击。结论是：0.32-0.63GW基准西迁负荷不足以触及中国全国性产能上限，但会叠加国网4万亿元“十五五”投资、南网1800亿元2026年投资、新能源并网和646亿元变压器出口，挤占2026-2027年合格交付窗口。取向硅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "信贷下沉不会引发系统性崩塌，但其推高的信用成本将削减银行利润12.4%，打破股息刚性，彻底击穿红利资产的低波防御底色。", "summary": "截至2026-06-13，本卡压力测试表明，银行因“资产荒”将信贷下沉至普惠小微领域不会引发系统性违约潮，但会导致信用成本“慢撒气”式抬升。若10%小微贷款劣变为不良，信用减值将侵蚀银行业归母净利润约12.4%。利润拖累将迫使绝对股息额下调（10%-15%），彻底击穿银行红利资产“类债刚兑”的估值…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 0, "neutral": 1, "value": 87, "events": [ { "title": "I will begin by inspecting the session brief and the report from Research note 4 to establi…", "summary": "I will begin by inspecting the session brief and the report from Research note 4 to establish a solid foundation for our stress test analysis of LGFV…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 1, "neutral": 3, "value": 88.4, "events": [ { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI算力基建的政策落地已经从文件进入容量和资本开支结构，但它是定向加速而非总量刺激。", "summary": "截至2026-06-15，本卡支持“温和宏观政策下 AI 算力基建仍会结构性加速”的判断。核心证据是全国智能算力从2025年底159万PFLOPS升至2026年3月底188万PFLOPS，约一季度增长18.2%；运营商总capex并未全面扩张，但中国移动、中国电信、中国联通均把资本开支向AIDC、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力岛格局不是利空AI基建，而是把估值锚从名义容量转向岛内有效MFU、硬电力、光互联密度和客户变现可见度。", "summary": "截至2026-06-15，本卡综合前序结论：全国一体化算力网仍是政策方向，但A股估值应从“全国统一算力大池”切换到“高质量算力岛”。光器件受益于岛内scale-up/scale-out流量和800G/1.6T/CPO升级，优质AIDC受益于硬电力、液冷、锚定客户和枢纽位置；通用IDC和纯算力租赁因…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Stress Testing Impact of Savings Decline and Delinquencies on Consumer Credi…", "summary": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 445de77a-c360-4ec1-b29d-9f5140877a25 结果摘要: Wrote handoff_response.md (3607 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 1, "neutral": 1, "value": 83.8, "events": [ { "title": "[auto-handoff] 国内AIDC表后自备发电合规与经济边界", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 工业制造分析师。 模式: adhoc 会话: 54eae79c-6253-4049-81b6-f93fd057ed2a 结果摘要: Wrote handoff_response.md (11240 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "底层资产风险正在侵蚀尾部县域银行资本，但传导主要集中在323家高风险机构，尚不构成全国性银行体系资本危机。", "summary": "截至2026-06-19，本卡压力测试323家高风险县域银行的底层资产传导。结论支持Research note 8的“有界尾部风险”判断，但将预警从流动性观察上调为资本观察。公开城投债违约不是主触发器，真正风险来自区县城投非标/商票逾期、中小微及个人经营贷迁徙、加盟商与消费贷违约共振。自测算显示，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 5, "neutral": 3, "value": 67.2, "events": [ { "title": "[auto-handoff] 商业银行资产荒配置对收益率曲线及大类资产风格轮动的影响评估", "summary": "致 首席策略师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: eeccb81c-3798-4962-ad58-c10d301199c8 结果摘要: Wrote handoff_response.md (10727 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 商业银行净息差（NIM）1.4%冰点制约下央行降息空间与国内信用收缩压力测试", "summary": "致 日报总编： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 3c199296-b99a-4c83-8b05-985744463425 结果摘要: Wrote handoff_response.md (14178 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AIDC电力并网瓶颈对电力敏感型化工品（如烧碱、黄磷）供给端硬约束的交叉验证", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: 6a16e82a-5277-47a4-bf61-68e0ff5c0f23 结果摘要: Wrote handoff_response.md (11068 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] NIM降至1.40%背景下中小银行低评级城投债买盘收缩评估", "summary": "致 信用分析师： 协同任务已完成。执行方为 金融行业分析师。 模式: adhoc 会话: 41113bd2-fa1b-4076-a9fc-f09a068b5c2e 结果摘要: Wrote handoff_response.md (14149 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 1, "neutral": 1, "value": 86.6, "events": [ { "title": "I have completed the analysis on AI compute infrastructure transmission paths under power a…", "summary": "I have completed the analysis on AI compute infrastructure transmission paths under power and transformer constraints , focusing on the concrete impa…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "We have completed the analysis for Research note 8 of the research thread session. As the A…", "summary": "We have completed the analysis for Research note 8 of the research thread session. As the AI Infrastructure Analyst, we support the thesis that grid …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 7, "neutral": 0, "value": 49, "events": [ { "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 4cd2d717-9d04-4569-b0d8-cc37d743d0a9 结果摘要: Wrote handoff_response.md (9414 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 成长降温还是风险预算下调", "summary": "致 研究质量主管： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b73420d5-d03c-4543-9513-956fc947c034 结果摘要: Wrote handoff_response.md (7963 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b04fea49-07a7-43ca-b246-32f00d158f50 结果摘要: Wrote handoff_response.md (9863 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "锦江、同庆楼等高经营杠杆标的的RevPAR负向冲击由于强国资隐性担保与抵押物安全边际垫底，未在信用债利差及融资成本中体现，股债风险定价呈现深度脱节。", "summary": "本研究针对锦江、同庆楼等消费板块高经营杠杆标的进行信用风险压力测试。研究显示，微观RevPAR负向冲击与高折旧负担虽导致企业业绩大幅收缩，但并未在信用债利差或融资成本中体现。锦江国际5Y中票与270D短融的一级利差被极限压缩至5 bps和3 bps，由上海国资强信用背书及信用“资产荒”主导；同庆楼…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 1, "neutral": 0, "value": 86.4, "events": [ { "title": "大额地方收储背景下AAA级核心城投利差倒挂与信用重定价风险", "summary": "大额地方收储背景下AAA级核心城投利差倒挂与信用重定价风险 LGFV Housing Inventory Purchase: Negative Carry and Credit Repricing Risks 在2027年融资平台退出大限临近与商业银行息差（1.40%历史冰点）收窄背景下，成都兴城…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 1, "neutral": 0, "value": 87.8, "events": [ { "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply s…", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 1, "neutral": 0, "value": 89.2, "events": [ { "title": "[auto-handoff] 中美利差-272bps倒挂与商业银行净息差1.40%硬约束下三季度利率债净供给冲击测算", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 065568fb-7e27-470f-ab19-7926760583a8 结果摘要: Wrote handoff_response.md (11243 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 5, "neutral": 1, "value": 69.6, "events": [ { "title": "[auto-handoff] 权益两融去杠杆是否外溢至银行间资金面并推动信用利差走阔", "summary": "致 信用分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 7dab7f81-754b-48e9-b144-aa1269bfb054 结果摘要: Wrote handoff_response.md (13780 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "并网延迟的变压器库存积压是可逆的时点拖累(金盘-8%~-11%、思源<-3%)，被国网4万亿内需托底，应在超配出口变压器框架下偏思源、轻金盘并叠加配对与公用事业杠铃对冲。", "summary": "并网延迟导致的已完工未交付变压器库存积压，是营运资金/收入确认的时点事件而非需求毁灭——订单仍排产至2027年在手backlog内。对金盘科技(688676.SH)拖累真实但有界：北美AIDC相关收入是最高Beta切片，Q2/Q3净利约-8%至-11%、伴随存货与FCF占用，属逢低布局而非证伪；对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 商业银行 1.40% NIM 约束对保障房收储贷款投放的挤出压力", "summary": "致 房地产分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: aa75509e-3210-4ab6-b781-37670c2d5374 结果摘要: Wrote handoff_response.md (9392 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 1, "neutral": 0, "value": 92, "events": [ { "title": "国债底线稳的是无风险利率，不是高杠杆主体的信用利差；若弱城投和房企利差被重定价至250bp-400bp，展期压力会非线性上升，但仍更可能表现为行政化续作而非系统性违约失控。", "summary": "截至2026-06-28，本卡压力测试信用利差非线性扩张对城投及地产债展期的传导。结论是：国债收益率底线可稳定基准久期，但无法阻止弱主体综合融资成本因利差跳升而上行。政策化债资源仍使城投系统性失控概率偏低，地产现金流和万科式展期案例显示尾部压力更直接。基准情景不是全面违约潮，而是更多展期、重组、银…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 3, "neutral": 0, "value": 81.4, "events": [ { "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "绿色技改缺口下高碳工业民企债务违约风险与信用利差压力测试评估", "summary": "截至2026-06-30，本卡片对高碳民营工业债违约概率与行业利差走阔做信用压力测试。全市场宽口径违约率仅0.08%–0.13% [S8]，边际0.4%远低于历史中位1.6% [S7]，系统温和。但高碳POE群体是不利尾部：12个月前瞻PD约5.05%（钢铁6.5%、化工5.1%、有色加工3.9%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-9a860a8801/", "networkHref": "/investor/research-network/#thesis-9a860a8801" }, { "id": "thesis-985b80b224", "source": "tracked_thesis", "title": "具身智能机器人进入 BOM 降本与执行器供给验证期", "titleEn": "Embodied AI robotics enters the BOM deflation and actuator supply verification phase", "lane": "具身智能/机器人零部件", "laneEn": "Embodied AI and robotics components", "laneHref": "/investor/lanes/embodied-ai-and-robotics-components/", "investableFocus": "具身智能/机器人零部件：TSLA, ISRG, ABBNY, 6954.T, 300124.SZ, 002747.SZ, 002472.SZ", "stockUniverse": [ { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ISRG", "name": "Intuitive Surgical", "market": "US", "role": "高端机器人商业化对照", "roleEn": "高端机器人商业化对照", "bucket": "core", "href": "/investor/stocks/ISRG/" }, { "ticker": "ABBNY", "name": "ABB", "market": "US ADR", "role": "工业机器人与自动化", "roleEn": "工业机器人与自动化", "bucket": "core", "href": "/investor/stocks/ABBNY/" }, { "ticker": "6954.T", "name": "Fanuc", "market": "Japan", "role": "工业机器人", "roleEn": "工业机器人", "bucket": "core", "href": "/investor/stocks/6954.T/" }, { "ticker": "300124.SZ", "name": "汇川技术", "market": "A-share", "role": "工控/伺服/自动化", "roleEn": "工控/伺服/自动化", "bucket": "core", "href": "/investor/stocks/300124.SZ/" }, { "ticker": "002747.SZ", "name": "埃斯顿", "market": "A-share", "role": "国产工业机器人", "roleEn": "国产工业机器人", "bucket": "core", "href": "/investor/stocks/002747.SZ/" }, { "ticker": "002472.SZ", "name": "双环传动", "market": "A-share", "role": "减速器/传动件", "roleEn": "减速器/传动件", "bucket": "core", "href": "/investor/stocks/002472.SZ/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "LMT", "name": "Lockheed Martin", "market": "US", "role": "国防主承包商", "roleEn": "国防主承包商", "bucket": "core", "href": "/investor/stocks/LMT/" }, { "ticker": "RKLB", "name": "Rocket Lab", "market": "US", "role": "小型发射与空间系统", "roleEn": "小型发射与空间系统", "bucket": "core", "href": "/investor/stocks/RKLB/" }, { "ticker": "BOTZ", "name": "Global X Robotics & AI ETF", "market": "US ETF", "role": "机器人与自动化", "roleEn": "机器人与自动化", "bucket": "watch", "href": "/investor/stocks/BOTZ/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 供给", "roleEn": "HBM 供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" } ], "stockTickers": [ "TSLA", "ISRG", "ABBNY", "6954.T", "300124.SZ", "002747.SZ", "002472.SZ", "ASTS", "LMT", "RKLB", "BOTZ", "MU", "005930.KS", "000660.KS" ], "coreView": "赛道：具身智能/机器人零部件。候选观察标的：TSLA, ISRG, ABBNY, 6954.T, 300124.SZ, 002747.SZ, 002472.SZ。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 72, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 3, "risks": 147, "neutral": 10, "analystCount": 7, "sourceTypeCounts": { "研究笔记": 36, "证据更新": 8, "分析协作": 4, "研究线程": 105, "深度研究": 4, "研究主题": 2, "链条事实": 1 }, "topTerms": [ "AI 基础设施", "Tesla", "Embodied", "robotics", "enters", "deflation", "actuator", "supply", "verification", "phase", "components", "TSLA" ], "whatChanged": [ "2026-06-29: 产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "2026-06-29: 产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "2026-06-29: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "2026-06-29: 产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "2026-06-29: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）" ], "practical": { "sectorLane": "具身智能/机器人零部件", "sectorLaneEn": "Embodied AI and robotics components", "exposureBucket": "航天国防与新场景", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ISRG", "name": "Intuitive Surgical", "market": "US", "role": "高端机器人商业化对照", "roleEn": "高端机器人商业化对照", "bucket": "core", "href": "/investor/stocks/ISRG/" }, { "ticker": "ABBNY", "name": "ABB", "market": "US ADR", "role": "工业机器人与自动化", "roleEn": "工业机器人与自动化", "bucket": "core", "href": "/investor/stocks/ABBNY/" }, { "ticker": "6954.T", "name": "Fanuc", "market": "Japan", "role": "工业机器人", "roleEn": "工业机器人", "bucket": "core", "href": "/investor/stocks/6954.T/" }, { "ticker": "300124.SZ", "name": "汇川技术", "market": "A-share", "role": "工控/伺服/自动化", "roleEn": "工控/伺服/自动化", "bucket": "core", "href": "/investor/stocks/300124.SZ/" }, { "ticker": "002747.SZ", "name": "埃斯顿", "market": "A-share", "role": "国产工业机器人", "roleEn": "国产工业机器人", "bucket": "core", "href": "/investor/stocks/002747.SZ/" }, { "ticker": "002472.SZ", "name": "双环传动", "market": "A-share", "role": "减速器/传动件", "roleEn": "减速器/传动件", "bucket": "core", "href": "/investor/stocks/002472.SZ/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "LMT", "name": "Lockheed Martin", "market": "US", "role": "国防主承包商", "roleEn": "国防主承包商", "bucket": "core", "href": "/investor/stocks/LMT/" }, { "ticker": "RKLB", "name": "Rocket Lab", "market": "US", "role": "小型发射与空间系统", "roleEn": "小型发射与空间系统", "bucket": "core", "href": "/investor/stocks/RKLB/" }, { "ticker": "BOTZ", "name": "Global X Robotics & AI ETF", "market": "US ETF", "role": "机器人与自动化", "roleEn": "机器人与自动化", "bucket": "watch", "href": "/investor/stocks/BOTZ/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 供给", "roleEn": "HBM 供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" } ], "stockTickers": [ "TSLA", "ISRG", "ABBNY", "6954.T", "300124.SZ", "002747.SZ", "002472.SZ", "ASTS", "LMT", "RKLB", "BOTZ", "MU", "005930.KS", "000660.KS" ], "primaryStocks": [ { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "ISRG", "name": "Intuitive Surgical", "market": "US", "role": "高端机器人商业化对照", "roleEn": "高端机器人商业化对照", "bucket": "core", "href": "/investor/stocks/ISRG/" }, { "ticker": "ABBNY", "name": "ABB", "market": "US ADR", "role": "工业机器人与自动化", "roleEn": "工业机器人与自动化", "bucket": "core", "href": "/investor/stocks/ABBNY/" }, { "ticker": "6954.T", "name": "Fanuc", "market": "Japan", "role": "工业机器人", "roleEn": "工业机器人", "bucket": "core", "href": "/investor/stocks/6954.T/" }, { "ticker": "300124.SZ", "name": "汇川技术", "market": "A-share", "role": "工控/伺服/自动化", "roleEn": "工控/伺服/自动化", "bucket": "core", "href": "/investor/stocks/300124.SZ/" }, { "ticker": "002747.SZ", "name": "埃斯顿", "market": "A-share", "role": "国产工业机器人", "roleEn": "国产工业机器人", "bucket": "core", "href": "/investor/stocks/002747.SZ/" } ], "instrumentTags": [ "TSLA", "Tesla", "ISRG", "Intuitive Surgical", "ABBNY", "ABB", "6954.T", "Fanuc", "300124.SZ", "汇川技术", "002747.SZ", "埃斯顿" ], "reviewCadence": "每日复核", "confirmationSignals": [ "部署节奏是否兑现为收入", "BOM 降本是否快于竞争扩产", "政策订单是否能跨过概念交易阶段" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "事件兑现", "融资稀释", "订单延迟", "高波动仓位" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：证据密度足够", "deepResearchReason": "证据深度 100，来源 7 类，近期事件 123 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 7, "analystCount": 7, "recentEvents": 123, "supportRatio": 0.02, "riskRatio": 0.92, "conflictScore": 100, "evidenceDepth": 100 }, "score": 62.3 }, "events": [ { "date": "2026-06-29", "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck from compute to power, networkin…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "AI 基础设施", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-28", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,600 亿）以经营现金流支撑，资金基础/地域/标的均不同于承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "AI 基础设施", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "summary": "截至 2026-06-28，本卡压力测试 Research note 1 的 AI 基础设施使能层主线：需求真实且强劲，但 2026-2028 年兑现会被电网接入、可靠电源、变压器、燃机和运行可靠性约束。IEA 预计全球数据中心用电从 2024 年 415 TWh 升至 2030 年约 945 TWh；美国和中国合计贡献近 80% 增量。[S1][S2] …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "AI 基础设施", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.en.md (7867 bytes) — The repor…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 12.25, "matchedTerms": [ "infrastructure", "investment", "customer", "cost" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-27", "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期：…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / su...", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 15, "matchedTerms": [ "AI 基础设施", "infrastructure", "channel" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "A 股仍在昇腾渠道层中过度定价 AI 基础设施叙事，低估营运资金融资成本；风险最集中在 000062 和 000977，601138 相对更有质量支撑但仍面临制造业估值压缩。", "summary": "本卡从 A 股定价角度支持并强化第 5 张卡片的财务脆弱性判断：市场尚未充分反映昇腾渠道层从试点走向规模采购时的毛利压缩和自由现金流恶化。按 2026-04-30 收盘价测算，深圳华强约 69.2x TTM P/E，浪潮信息约 40.2x，工业富联约 30.7x。000062 的叙事溢价与经销商现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 1, "neutral": 0, "value": 45.4, "events": [ { "title": "AI 基础设施尾部风险应被视为单一相关信用-股权敞口：未对冲等权组合每 $100 的 99% VaR 约 $78，风险控制优先级是减 CRWV 股权、按压力抵押品价值计量 CoreW…", "summary": "本卡对 CRWV 股权、CoreWeave 有抵押债务、NBIS 股权与 OpenAI 晚期股权组成的 $100 等权参考组合进行三节点级联压力测试。结合 Research note 07 与最新公开披露，CoreWeave 2025 年客户 A 集中度升至 67%、RPO 达 $60.7B，Op…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "阿里巴巴集团宣布2025年起三年内投入约人民币3800亿元（约530亿美元）用于AI基础设施", "summary": "阿里巴巴集团宣布2025年起三年内投入约人民币3800亿元（约530亿美元）用于AI基础设施", "sourceLabel": "证据更新", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-06", "supports": 1, "risks": 0, "neutral": 0, "value": 52.8, "events": [ { "title": "Alphabet 2026年资本开支指引上调至1800-1900亿美元，预计2027年资本开支较2026年大幅增加。", "summary": "Alphabet 2026年资本开支指引上调至1800-1900亿美元，预计2027年资本开支较2026年大幅增加。", "sourceLabel": "证据更新", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 0, "neutral": 1, "value": 49.2, "events": [ { "title": "阿里巴巴 (BABA) 承诺投入 550 亿美元用于 AI 基础设施建设。", "summary": "阿里巴巴 (BABA) 承诺投入 550 亿美元用于 AI 基础设施建设。", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-08", "supports": 0, "risks": 0, "neutral": 1, "value": 50.6, "events": [ { "title": "[auto-handoff] Humanoid Robot Supply Chain Validation", "summary": "致 首席量化师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 75eb4b47-052b-4a8f-95b5-ab3601b0fef6 结果摘要: Wrote handoff_response.md (4384 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-12", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "在 Brent 原油 150-120 美元情景下，云/AI 基础设施板块的预估利润率侵蚀为 180 - 250 bps。", "summary": "在 Brent 原油 150-120 美元情景下，云/AI 基础设施板块的预估利润率侵蚀为 180 - 250 bps。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "AI 基础设施正从‘无风险增长’阶段进入‘社会许可防御’阶段，追溯性立法风险可能引发行业 20% 以上的系统性估值修正。", "summary": "本报告对 AI 基础设施在核心区域（VA, AZ, IE）面临的‘农业-社会’风险进行了压力测试。基于 640 亿美元受阻管线，模拟了追溯性立法撤销用水/用地权及专项农业附加费的情景。结果显示，极端情景下主要 DC REITs 可能面临 15-22% 的回撤，1 日 VaR 将从 185bps 飙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 4, "neutral": 0, "value": 27.8, "events": [ { "title": "麦肯锡预计 2025 年 AI 基础设施投资需求为 5.2 万亿美元，其中 1.3 万亿美元用于电力、输电、冷却等领域。", "summary": "麦肯锡预计 2025 年 AI 基础设施投资需求为 5.2 万亿美元，其中 1.3 万亿美元用于电力、输电、冷却等领域。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年5月10日，分析师指出应将组合从“开发商权益+ESA容量beta”转向“受监管电网/水务/天然气赋能敞口+锚定型核心资产”。", "summary": "2026年5月10日，分析师指出应将组合从“开发商权益+ESA容量beta”转向“受监管电网/水务/天然气赋能敞口+锚定型核心资产”。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施铜密度为 27-47 吨/兆瓦", "summary": "AI 基础设施铜密度为 27-47 吨/兆瓦", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "特斯拉 (TSLA) 2026 年 Q1 的库存天数达到创纪录的 27 天。", "summary": "特斯拉 (TSLA) 2026 年 Q1 的库存天数达到创纪录的 27 天。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "[auto-handoff] Summit De-risking", "summary": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 502dfe56-3d6c-4017-933c-df9f9b15f88b 结果摘要: Wrote handoff_response.md (8242 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "AI 交易逻辑转向：从算力建设到推理侧需求", "summary": "AI 交易逻辑转向：从算力建设到推理侧需求 Does the recent surge in AMD and the shift toward 'Agentic AI' signify a permanent structural pivot in the AI investment thesis…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "AI 算力资本开支、电力瓶颈与通胀再定价", "summary": "AI 基础设施需求正在先通过电力、电网、变压器与材料供给形成成本脉冲，再通过生产率与架构效率形成中期缓释；投资结论取决于这两个方向谁先兑现。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-power-inflation-transmission-0c9ba15c/" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "截至 2026-05-19，A 股哑铃化再定价已经启动：左端买现金回报和低波确定性，右端买 AI 基础设施瓶颈，但右端拥挤度已显著高于基本面确认速度。", "summary": "工作日期 2026-05-19。A 股盘面支持第 06 卡“叙事溢价 → 确定性溢价”的哑铃化判断，但不是全面价值回归。左端证据是红利策略 ETF 5 月以来净流入约 72.9 亿元、5月18日上证50ETF净申购9.78亿元；右端证据是两融资金集中买入电子和通信，5月11日至15日两者合计占周度…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 0, "neutral": 1, "value": 61.8, "events": [ { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 1, "value": 51.2, "events": [ { "title": "As the Chief Strategist, I have synthesized the final research card for this session. The e…", "summary": "As the Chief Strategist, I have synthesized the final research card for this session. The evidence confirms that while the A share auto sector has hi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "[auto-handoff] AI 基础设施散户拥挤是否耗尽 Thesis v8 入场窗口", "summary": "致 市场情绪师： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: b790980f-74a8-48f2-8307-cf1e60b887ee 结果摘要: Wrote handoff_response.md (14984 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "对元板块六子板块组合在路径A升级、因子拥挤逆转与CNY尾部三重情景下的组合级压力测试 首席风控官报告 | Research note 08 / 08 | 2026 05 29（亚洲/…", "summary": "对元板块六子板块组合在路径A升级、因子拥挤逆转与CNY尾部三重情景下的组合级压力测试 首席风控官报告 | Research note 08 / 08 | 2026 05 29（亚洲/新加坡） 立场：stress test（压力测试） 1. 前言与今日日期锚定 本报告为 AI 基础设施元板块专题研究…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 0, "neutral": 0, "value": 73.8, "events": [ { "title": "The AI Power infrastructure value chain has been audited for logical consistency and fragil…", "summary": "The AI Power infrastructure value chain has been audited for logical consistency and fragility as of 2026 06 07 . The core thesis—that AI deployment …", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 0, "neutral": 1, "value": 70.2, "events": [ { "title": "AI 基础设施的结构性分化：从算力扩张到电力链防御", "summary": "AI 基础设施的结构性分化：从算力扩张到电力链防御 Is the recent strength in XLU and weakness in SOXX/URA a localized defensive rotation or a signal of a structural paradigm …", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 6, "neutral": 0, "value": 32.6, "events": [ { "title": "产业链调研：新一代信息技术（30 日激活率 3.3%, 3/91）", "summary": "产业链调研：新一代信息技术（30 日激活率 3.3%, 3/91） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/241） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.1%, 14/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.1%, 14/198） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/205） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 7, "neutral": 0, "value": 28, "events": [ { "title": "电力交付滞后会迫使 Hyperscalers 调整 GPU 集群投产节奏，并通过更长现金回收期、更低早期利用率和更高电力保障成本压低 AI 基础设施 capex 效率。", "summary": "截至2026-06-11，本卡支持上一卡：AI capex 不会因电网瓶颈消失，但 GPU 集群部署节奏会被“已通电 MW”重新排序。变压器、并网和输电延迟会把部分已宣布园区容量推迟到后续窗口，压低早期自由现金流和资产利用率。Hyperscalers 会通过已通电站点溢价、分阶段投产、区域重排和自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GW 级数据中心的电力瓶颈与电网投资机遇：核能并网与公用事业的超额收益 日期： 2026年6月11日 分析师： 公用事业分析师 (Utilities Analyst) 主题： GW …", "summary": "GW 级数据中心的电力瓶颈与电网投资机遇：核能并网与公用事业的超额收益 日期： 2026年6月11日 分析师： 公用事业分析师 (Utilities Analyst) 主题： GW 级数据中心电网约束与公用事业投资机会 执行摘要 截至 2026 年 6 月，AI 基础设施的叙事已从“GPU 稀缺”…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：海工装备产业链（30 日激活率 5.4%, 6/112）", "summary": "产业链调研：海工装备产业链（30 日激活率 5.4%, 6/112） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 10.3%, 10/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 10.3%, 10/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 5, "neutral": 1, "value": 44.4, "events": [ { "title": "截至2026-06-13，电力设备成本与交付风险已经实质性延后 AI 算力集群通电节奏，但它表现为 capex 效率恶化和项目重排，而不是最大型 hyperscalers 已经全面削…", "summary": "截至2026-06-13，本卡支持前序“电力约束正在惩罚 AI capex/融资假设”的方向，但精确化为：2026-2028 年已出现项目层面的通电延期与节奏重排，尚未出现头部 hyperscalers headline capex 的广泛下修。Sightline 显示 2026 年约 16 GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施从‘算力’向‘电力’的范式转移", "summary": "AI 基础设施从‘算力’向‘电力’的范式转移 AI Infrastructure Shift: From Compute to Grid Power PJM 电力容量价格暴涨10倍是否意味着 AI 估值逻辑必须从 GPU 摩尔定律转向公用事业的资产底座逻辑？", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 5, "neutral": 0, "value": 42.8, "events": [ { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 5.7%, 10/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 5.7%, 10/174） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 9, "neutral": 1, "value": 23.2, "events": [ { "title": "物理电网瓶颈正迫使 AI 基础设施从‘暴力规模扩张’转向‘受约束的能效优化’，电力并网速度已取代芯片供应成为核心估值护城河。", "summary": "本报告通过压力测试确认了物理电网并网排队（7-10年）与变压器交期（4-5年）已成为 AI 基础设施扩张的硬约束。定量模型显示，1年的并网延迟将导致 100MW 数据中心损失 5.5% 的生命周期价值，且快速的 GPU 迭代周期（12-18个月效用减半）加剧了闲置产能的 ROI 损耗。这种物理瓶颈…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the synthesis for Research note 10, incorporating the AI infrastructure su…", "summary": "I have completed the synthesis for Research note 10, incorporating the AI infrastructure supply chain vulnerabilities and power grid stress test resu…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "On the canonical work date of 2026 06 14 , we have completed a comprehensive stress test of…", "summary": "On the canonical work date of 2026 06 14 , we have completed a comprehensive stress test of AI datacenter project economics and layout restructuring …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 11, "neutral": 1, "value": 12.6, "events": [ { "title": "AI 基础设施的估值与投资回报已从‘算力拥有量’转向‘电力接入效率’，电力-计算一体化已成为 2026 年超大厂资本开支的核心逻辑。", "summary": "2026年，AI基础设施资本开支进入“电力重心偏移”阶段。超大厂支出从芯片采购转向电力-计算一体化架构，物理基础设施投入与芯片已达1:1。算力向美南和中东等电力充裕地区迁移，驱动了1.6T光互联的爆发式增长（需求达8.6M-20M只）以及先进封装的全球区域化布局。尽管解决了电力准入，但远距离传输延…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海底零碳算力与深海数据中心：技术可行性与电力供应保障 白板编号：4157cf99 4f17 487d 9efd db740a1f0fbf · 第 02/10 张 分析师：AI 基础设…", "summary": "海底零碳算力与深海数据中心：技术可行性与电力供应保障 白板编号：4157cf99 4f17 487d 9efd db740a1f0fbf · 第 02/10 张 分析师：AI 基础设施分析师（ai infrastructure analyst） 日期：2026 06 15（亚洲/新加坡） 立场：支…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施正进入“利率脱敏但电力受限”的深水区，高资金成本将瓶颈从硅片推向电网物理极限，并系统性推高非美玩家的准入门槛。", "summary": "2026-06-15：评估显示，尽管高利率和 DXY 走强（目标 101.3）增加了融资和采购成本，但头部超大规模厂商 (Hyperscaler) 的 AI Capex 因 AGI 生存竞赛而展现出极强韧性（2026 年预计达 7600 亿美元）。然而，高利率正在通过公用事业资金成本传导，加剧电力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 3, "neutral": 0, "value": 59, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 11, "neutral": 0, "value": 12.4, "events": [ { "title": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：…", "summary": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：AI 基础设施分析师 (ai infrastructure analyst) 立场：stress test（压力测试…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler capex 足以穿越能源 TCO 上行并支撑稀缺供应链，但投资表达必须是“多稀缺、空低 ROIC”，而不是无差别做多 AI。", "summary": "2026-06-18 压力测试结论：hyperscaler capex 增长足以让 AI 基础设施供应链消化能源成本带来的边际 TCO 压力，但只能支持“多稀缺”，不能支持泛 AI beta。四大买方 2026 capex 粗略包络约 $710B（自测算），Microsoft、Alphabet、M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 10, "neutral": 0, "value": 19.8, "events": [ { "title": "2028 年 AI 基础设施的胜负不在于是否完全离网，而在于能否把液冷机架、BTM/并网混合电源、闭环冷却、燃气交付和工作负载调度整合成可复制的 7-150 MW 阶段模块。", "summary": "2026-06-18：本卡压力测试前序主线。结论是 ERCOT/Permian 与 BTM 气电是 AI 训练负荷的真实泄压阀，但不能干净解决 2028 年瓶颈；它们把问题转化为模块化电力-冷却-算力架构。液冷从可选项变成高密 AI 的前提，132 kW/机架意味着 1 GW IT 负荷约需 7,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康存在选择性重估标的，但 AI 跨链外溢不应按篮子买入；基础设施先捕获稀缺性，医疗需等临床、监管、支付和现金转化验证。", "summary": "截至 2026-06-18，本卡从 AI 基础设施视角否定“大健康低覆盖=系统性低估”的宽泛判断。15/280 的 5.4% 激活率只是内部关注度信号；真正稀缺、已被资本密集定价的环节在算力、电力、数据中心与云 capex。医疗 AI 采用真实，FDA 列表已达 1,525 个条目，但监管、临床验…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 6, "neutral": 0, "value": 45.2, "events": [ { "title": "Starlink 带宽 × Grok 算力的飞轮真实，使 SPCX 成为 AI 分发基座与算力链条需求催化剂，但算力归属 xAI、轨道边缘算力尚萌芽，真正瓶颈是电力——故支持但需把光…", "summary": "从 AI 基础设施视角支持卡片 01：Starlink 带宽（约 450 Tbps，三代卫星单星 >1 Tbps、下行 10 倍/上行 24 倍）与 Grok/Colossus 算力（约 55.5 万颗 GPU、约 180 亿美元、向 2 GW 扩张）构成真实飞轮，使 SPCX 成为差异化的'算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：建筑材料产业链（30 日激活率 8.3%, 13/157）", "summary": "产业链调研：建筑材料产业链（30 日激活率 8.3%, 13/157） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 1, "neutral": 0, "value": 76.6, "events": [ { "title": "AI 算力资本开支是美联储路径上移传导至中国融资条件最敏感的通道，但 PBoC 大概率以结构性信用而非降息回应，从而坐实政策利率独立性。", "summary": "从 AI 基础设施视角综合 Research note 01–03：2026 年中国 AI 资本开支账单中约 USD 550–650 亿直接挂钩美元曲线（HBM、残余加速卡、离岸租赁），是美联储 3.8% 路径向中国融资条件传导的最干净通道。但绑点是 HBM/燃机等少数高价格刚性进口品，不是整体资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 11, "neutral": 0, "value": 18, "events": [ { "title": "AI 基础设施建设中的电力供应与电网容量瓶颈分析", "summary": "本卡压力测试 Research note 02 的隐含假设「资本开支≈算力」。证据显示约束已从芯片迁移到电力：微软披露约800亿美元 Azure 订单因电力受限无法交付，Nadella 称芯片堆库存却无「暖壳」可插[S5][S4]。转化在三处漏损：(1)并网排队约2,600 GW、中位等待约5年、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "investment_constraint", "summary": "AI infrastructure exposure should be limited to assets supported by utilization and green-power procurement", "sourceLabel": "链条事实", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 6, "neutral": 0, "value": 49.4, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 7, "neutral": 1, "value": 47.8, "events": [ { "title": "2028年再融资悬崖与 AI 算力基础设施资本开支的信用冲突", "summary": "压力测试支持并延伸卡片 04：AI 基础设施信用风险真实且上升，但呈哑铃状集中，不在头条的 $725 bn 巨头资本开支上。四巨头（MSFT/GOOGL/AMZN/META，2026 资本开支约 $725 bn、+77%，约 75% 投 AI [S1]）凭现金流与 AA 资产负债表通过测试，尽管 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施的物理天花板：电力与电网瓶颈是否会限制超大规模厂商资本开支的持续性？", "summary": "电力与电网已成为 AI 建设的物理刚性约束，瓶颈从芯片转向电子，但约束是供给侧的时间问题（并网 4–7 年、变压器 2–3 年、燃气轮机售罄至 2030），而非需求毁灭。结论：电力是调速器而非刹车，令约 7,250 亿美元超大规模资本开支拉长进入平台期、压低斜率，而非触发悬崖。微软一边把 FY26…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 6, "neutral": 0, "value": 52.2, "events": [ { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "xAI 算力扩张对 AI 基础设施与能源供应的压力测试", "summary": "对卡片 01 前瞻风险的压力测试：跑步机确实存在，但卡点不在资本。约束层级为资本 < GPU < 电力。资本充裕（xAI E 轮 200 亿/估值 2300 亿，外加 200 亿 SPV 与 35 亿 Apollo 方案；2026 年超大规模厂商资本开支约 6000–7250 亿），GPU 因英伟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 7, "neutral": 0, "value": 47.6, "events": [ { "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 5, "neutral": 0, "value": 61, "events": [ { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / su...", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 6, "neutral": 1, "value": 59.4, "events": [ { "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "summary": "截至 2026-06-28，本卡压力测试 Research note 1 的 AI 基础设施使能层主线：需求真实且强劲，但 2026-2028 年兑现会被电网接入、可靠电源、变压器、燃机和运行可靠性约束。IEA 预计全球数据中心用电从 2024 年 415 TWh 升至 2030 年约 945 T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 8, "neutral": 0, "value": 45.8, "events": [ { "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck fr…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-985b80b224/", "networkHref": "/investor/research-network/#thesis-985b80b224" }, { "id": "thesis-5057edfa79", "source": "tracked_thesis", "title": "AI-Biology 与医疗效率改善需要从科研突破走向支付和监管兑现", "titleEn": "AI biology and healthcare productivity must move from research breakthroughs to reimbursement and regulatory proof", "lane": "AI 医疗/生物科技效率", "laneEn": "AI healthcare and biology productivity", "laneHref": "/investor/lanes/ai-healthcare-and-biology-productivity/", "investableFocus": "AI 医疗/生物科技效率：TMO, DHR, REGN, EXAI, SDGR, 300760.SZ, 2269.HK", "stockUniverse": [ { "ticker": "TMO", "name": "Thermo Fisher Scientific", "market": "US", "role": "生命科学工具", "roleEn": "生命科学工具", "bucket": "core", "href": "/investor/stocks/TMO/" }, { "ticker": "DHR", "name": "Danaher", "market": "US", "role": "生命科学工具", "roleEn": "生命科学工具", "bucket": "core", "href": "/investor/stocks/DHR/" }, { "ticker": "REGN", "name": "Regeneron", "market": "US", "role": "创新药研发效率", "roleEn": "创新药研发效率", "bucket": "core", "href": "/investor/stocks/REGN/" }, { "ticker": "EXAI", "name": "Exscientia", "market": "US", "role": "AI drug discovery", "roleEn": "AI drug discovery", "bucket": "core", "href": "/investor/stocks/EXAI/" }, { "ticker": "SDGR", "name": "Schrodinger", "market": "US", "role": "计算药物发现", "roleEn": "计算药物发现", "bucket": "core", "href": "/investor/stocks/SDGR/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "2269.HK", "name": "药明生物", "market": "HK", "role": "CXO/生物药外包", "roleEn": "CXO/生物药外包", "bucket": "core", "href": "/investor/stocks/2269.HK/" }, { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "TMO", "DHR", "REGN", "EXAI", "SDGR", "300760.SZ", "2269.HK", "600519.SH", "600276.SH", "3690.HK", "001215.SZ", "1179.HK", "9961.HK", "MU" ], "coreView": "赛道：AI 医疗/生物科技效率。候选观察标的：TMO, DHR, REGN, EXAI, SDGR, 300760.SZ, 2269.HK。当前共有 40 条证据、3 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 72, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 20, "risks": 130, "neutral": 10, "analystCount": 18, "sourceTypeCounts": { "研究笔记": 157, "深度研究": 1, "分析协作": 1, "研究线程": 1 }, "topTerms": [ "AI 基础设施", "CPI", "创新药", "CXO", "设备", "AI-Biology", "biology", "healthcare", "productivity", "must", "move", "breakthroughs" ], "whatChanged": [ "2026-06-29: 材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放右移。", "2026-06-29: 电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "2026-06-28: 新一代信息技术最值得补覆盖的不是前端应用，而是 AI 基础设施与国产替代共同推动的设备、EDA、光互联、内存互联和自主 CPU 使能层。", "2026-06-28: 产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "2026-06-28: 去风险三期医药资产通过rNPV模型释放3.48亿美元的期权价值构成Biotech估值底，而生命科学工具在5.5x EV/Sales筑底、CXO在 overseas 利润收缩下于2.0x-3.0x EV/Sales见底，共同驱动跨国药企对传奇生物、科伦博泰等高概率标的的并购共…" ], "practical": { "sectorLane": "AI 医疗/生物科技效率", "sectorLaneEn": "AI healthcare and biology productivity", "exposureBucket": "消费健康防御", "tone": "green", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "TMO", "name": "Thermo Fisher Scientific", "market": "US", "role": "生命科学工具", "roleEn": "生命科学工具", "bucket": "core", "href": "/investor/stocks/TMO/" }, { "ticker": "DHR", "name": "Danaher", "market": "US", "role": "生命科学工具", "roleEn": "生命科学工具", "bucket": "core", "href": "/investor/stocks/DHR/" }, { "ticker": "REGN", "name": "Regeneron", "market": "US", "role": "创新药研发效率", "roleEn": "创新药研发效率", "bucket": "core", "href": "/investor/stocks/REGN/" }, { "ticker": "EXAI", "name": "Exscientia", "market": "US", "role": "AI drug discovery", "roleEn": "AI drug discovery", "bucket": "core", "href": "/investor/stocks/EXAI/" }, { "ticker": "SDGR", "name": "Schrodinger", "market": "US", "role": "计算药物发现", "roleEn": "计算药物发现", "bucket": "core", "href": "/investor/stocks/SDGR/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "2269.HK", "name": "药明生物", "market": "HK", "role": "CXO/生物药外包", "roleEn": "CXO/生物药外包", "bucket": "core", "href": "/investor/stocks/2269.HK/" }, { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" } ], "stockTickers": [ "TMO", "DHR", "REGN", "EXAI", "SDGR", "300760.SZ", "2269.HK", "600519.SH", "600276.SH", "3690.HK", "001215.SZ", "1179.HK", "9961.HK", "MU" ], "primaryStocks": [ { "ticker": "TMO", "name": "Thermo Fisher Scientific", "market": "US", "role": "生命科学工具", "roleEn": "生命科学工具", "bucket": "core", "href": "/investor/stocks/TMO/" }, { "ticker": "DHR", "name": "Danaher", "market": "US", "role": "生命科学工具", "roleEn": "生命科学工具", "bucket": "core", "href": "/investor/stocks/DHR/" }, { "ticker": "REGN", "name": "Regeneron", "market": "US", "role": "创新药研发效率", "roleEn": "创新药研发效率", "bucket": "core", "href": "/investor/stocks/REGN/" }, { "ticker": "EXAI", "name": "Exscientia", "market": "US", "role": "AI drug discovery", "roleEn": "AI drug discovery", "bucket": "core", "href": "/investor/stocks/EXAI/" }, { "ticker": "SDGR", "name": "Schrodinger", "market": "US", "role": "计算药物发现", "roleEn": "计算药物发现", "bucket": "core", "href": "/investor/stocks/SDGR/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" } ], "instrumentTags": [ "TMO", "Thermo Fisher Scientific", "DHR", "Danaher", "REGN", "Regeneron", "EXAI", "Exscientia", "SDGR", "Schrodinger", "300760.SZ", "迈瑞医疗" ], "reviewCadence": "每日复核", "confirmationSignals": [ "价格弹性是否转化为利润率修复", "支付端和监管是否支持长期现金流", "防御属性是否被估值提前透支" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "估值拥挤", "支付监管", "渠道库存", "CPI 食品项" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 95 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 18, "recentEvents": 95, "supportRatio": 0.13, "riskRatio": 0.81, "conflictScore": 89.5, "evidenceDepth": 100 }, "score": 66.8 }, "events": [ { "date": "2026-06-29", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]，PPI-CPI剪刀差隔离了政策利率。任务设想的'成本→CP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "CPI", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放…", "summary": "截至2026-06-29，变压器交期延长和材料瓶颈导致美国电网资本开支通胀率达32.72%。结合2027/2028年度PJM容量电费暴涨11.5倍至$333.44/MW-day，将拉动美国零售电价上涨4.18%，直接和间接推高全国CPI 0.124个百分点。高粘性通胀与AI基建庞大资金需求将上移中性利率50-75基点，延缓美联储降息路径。此外，并网滑期18…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "CPI", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "大健康低覆盖机会更像跨子行业alpha篮子，而不是单一医疗板块beta；优先研究能把2025-2026催化转化为收入桥的五家公司。", "summary": "截至2026-06-28，本卡支持开启大健康低覆盖研究队列，但不建议追宽泛板块beta。优先深挖Keymed Biosciences、Hybio Pharmaceutical、Sino Biological、MicroPort NeuroScientific和Allgens Medical，主线分别是商业化创新药、GLP-1多肽制造、生命科学工具、器械出海…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19, "matchedTerms": [ "创新药", "CXO", "healthcare", "Scientific" ], "href": "" }, { "date": "2026-06-28", "title": "去风险三期医药资产通过rNPV模型释放3.48亿美元的期权价值构成Biotech估值底，而生命科学工具在5.5x EV/Sales筑底、CXO在 overseas 利润收缩下于2.0x-3.…", "summary": "截至2026-06-28，本卡片基于BD估值方法构建了下一批并购标的与估值底线筛选框架。AbbVie 109亿美元现金收购Apogee与Merck KGaA 113亿美元收购Bio-Techne为医药生物确立了质量门控底座。我们量化厘定：1) 创新药三期去风险资产的rNPV floor：以12亿美元峰值重磅药物为例，从二期(PoS 35%)跨入三期(PoS…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 18.5, "matchedTerms": [ "创新药", "CXO", "healthcare", "tech" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "生命健康与生物科技链 6.9% 激活率是覆盖缺口而非基本面缺口。2026 Q1 中国创新药对外授权约 600–614 亿美元（同比 +66–73%，约 98 笔），占全球金额 75%、笔数 20%，首付款中位数升至 1.84 亿美元；药明康德 Q1 在手订单 597.7 亿元（+23.6%）。支付端 6 月 1 日启动商保丙类目录衔接。挑出五个被低覆盖标的…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17.5, "matchedTerms": [ "创新药", "CXO", "healthcare" ], "href": "" }, { "date": "2026-06-28", "title": "新一代信息技术最值得补覆盖的不是前端应用，而是 AI 基础设施与国产替代共同推动的设备、EDA、光互联、内存互联和自主 CPU 使能层。", "summary": "截至 2026-06-28，本卡支持启动新一代信息技术深度研究。过去 30 日激活率仅 4/92（4.3%），内部覆盖不足。优先候选为拓荆科技 688072.SH、华大九天 301269.SZ、天孚通信 300394.SZ、澜起科技 688008.SH、龙芯中科 688047.SH；主线是半导体设备、EDA、光互联、内存互联与国产 CPU 的使能层机会。跨…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "AI 基础设施", "设备" ], "href": "" }, { "date": "2026-06-27", "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期：…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 / Executive Summary 截至 2026 年 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.5, "matchedTerms": [ "AI 基础设施", "infrastructure", "support" ], "href": "" }, { "date": "2026-06-27", "title": "全球能源供应增量（特别是重启周期的核能与不稳定的可再生能源）及慢速的电网升级在2026-2027年内无法有效对冲 AI 基础设施的电网并网刚性瓶颈，该瓶颈将直接导致大云厂商面临 800 亿至…", "summary": "全球AI算力扩张正面临电网并网物理刚性瓶颈。全球电网升级周期长达4-10年，排队时间中位数达5年，且高密度变压器等关键主设备面临交付延迟，直接导致美国约7/12 GW在建产能延期，并将造成2026-27年全球超大云厂商约800亿-1100亿美元资本开支递延。核电与可再生能源虽有高成长，但受制于重启周期（3-4年）及并网不确定性，短期无法有效对冲电网卡点。算…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.25, "matchedTerms": [ "AI 基础设施", "设备" ], "href": "" }, { "date": "2026-06-26", "title": "大健康产业链处于‘被忽视定价’阶段，基本面反转已现，应围绕出海BD与高端国产替代定向深挖五只低覆盖中盘股，防范地缘立法与科创去杠杆等跨链外溢。", "summary": "本卡片对低覆盖的大健康产业链进行了深度分析。在30日激活率仅5.7%的被忽视背景下，基本面于2026上半年迎来License-out、CXO订单与设备更新三重引擎的拐点。定向推荐药明合联、荣昌生物、康诺亚、医渡科技、沛嘉医疗五只低覆盖中盘创新龙头，并系统梳理了BIOSECURE立法、美债高利率去杠杆、DRG设备贴息及上游BOM与碳边境四类跨链条溢出风险。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17, "matchedTerms": [ "CXO", "设备", "healthcare" ], "href": "" }, { "date": "2026-06-26", "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–560 亿美元；中国云(字节/阿里/腾讯)与国产替代构成第二引…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "AI 基础设施", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-26", "title": "海外电力设备及并网物理瓶颈将导致CXO海外基地落地延迟2-3年，造成项目NPV毁灭性缩减77.07%，并通过毛利率永久性收缩25-30pct对企业估值中枢构成硬约束。", "summary": "压力测试显示，受AI算力基建挤压，全球大型变压器（LPT）交期拉长至48-60个月。在新加坡、英国和爱尔兰，海外CXO建厂面临2-3年并网延迟。以$500M基地测算，3年延迟将造成$270M资金占用与设备折旧折损，毁灭该项目77.07%的NPV。此外，地缘安全审查隔绝了中国快交期电力装备，迫使项目受困于欧美OEM厂商的交付积压。这证实了CXO估值受压来自海…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "CXO", "设备" ], "href": "" }, { "date": "2026-06-24", "title": "AI算力集群的电力保障能力应按已锁定的变压器与中压设备链条打分，而不是按土地、PPA或名义兆瓦打分。", "summary": "2026-06-25，本卡对上一张 AI 基础设施结论做压力测试并支持其主线：区域 AI 算力集群的真正瓶颈不是单一铜/GOES价格，而是主变、中压开关柜、断路器、配电变压器、施工队与费率审批构成的完整送电链。美国2025年电力变压器与配电变压器供给缺口分别为30%和10%；北弗吉尼亚等核心市场已体现需求集中、输电约束与设备排产之间的冲突。100 MW、P…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.25, "matchedTerms": [ "AI 基础设施", "设备" ], "href": "" }, { "date": "2026-06-24", "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Microsoft、Alphabet、Meta 的 capex …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "AI 基础设施", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-24", "title": "铜及电网设备成本的结构性上涨将通过公用事业电价调整和工业投入品渠道转嫁给终端，抬高全球结构性通胀底座，迫使主要央行维持更高的限制性利率并加速日本央行货币正常化。", "summary": "截至2026-06-25，本卡评估了铜及电网成本压力向全球货币政策的传导。电网升级与AI算力用铜强度大，推高设备售价与监管资产基底，预测在3-5年内将零售电价推升1.5%至2.5%。50%铜价冲击将累积拉动全球CPI约0.25%至0.45%。粘性通胀使美联储年底前维持3.50%-3.75%高利率，极端下存在二次加息风险；日本央行将被迫加息至0.50%-0.…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "CPI", "设备" ], "href": "" }, { "date": "2026-06-23", "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规避 4–7 年排队导致的 60–70% 价值搁浅，是压倒性增…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19.75, "matchedTerms": [ "AI 基础设施", "设备", "infrastructure" ], "href": "" }, { "date": "2026-06-23", "title": "未覆盖机会集中在从科学故事转向商业化、医保、全球注册或集采韧性的公司，而不是宽泛押注医疗板块 beta。", "summary": "2026-06-24：本卡支持为生命健康和生物科技链条建立聚焦研究队列。优先深挖 Keymed Biosciences 2162.HK、Ascentage Pharma 6855.HK/AAPG、Mabwell 688062.SH/02493.HK，并在器械端保留 Peijia Medical 9996.HK 与 MicroPort NeuroScient…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19.25, "matchedTerms": [ "创新药", "CXO", "healthcare", "Scientific" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "AI 物理超级周期面临“宏观剪刀差”挑战：利率陷阱与资源民族主义可能将“物理护城河”转化为“高久期估值风险”。", "summary": "本卡对 AI 基础设施“计算层向物理层转移”的逻辑进行了全球宏观压力测试。研究发现，虽然物理护城河具有结构性吸引力，但面临“宏观剪刀差”风险：即前置的物理层通胀（铜、电力设备）与延迟的 AI 生产力回报之间的背离，可能迫使央行维持高利率，从而对高久期的工业和公用事业资产造成估值压制。此外，地缘政治…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "在 AI 能源利率冲击成为已知情景后，权益组合应从纯成长 AI 转为价值-质量-硬资产-精选 AI 基础设施框架，以承载五腿对冲并保留可验证的 AI alpha。", "summary": "本卡将第08卡联合无序尾部纳入基本配置，而非事后保险。建议全球权益风险预算按美股58、A股42配置，对冲后净权益beta为0.75-0.85，并以每季度65 bps NAV承载五腿结构性对冲。A股超配电网、电力、有色、AI基础设施和高股息SOE金融；美股超配AI硬件、电气设备、公用事业、能源和质量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施'政策公用事业化'再定价方向成立，但转债市场仅完成约三成至五成验证；强赎稀释、二线信用分化、Neocloud 信用利差三条通道被前序卡片低估。", "summary": "从可转债视角对 Cards 04–08 的'AI 基础设施=政策公用事业'再定价论进行压力测试。债底收益率、IV、平价三项指纹仅完成 30–50% 验证。三条被低估通道：(1) 变压器/液冷/光模块在 2026H2 出现 8–15% 强赎稀释悬崖，封顶 Research note 06 倍数收敛；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "真正具备全球临床证据、BD选择权和多元支付路径的中国创新药应享受估值修复，但CXO、生物数据和集采耗材仍会受到地缘与价格政策折价约束。", "summary": "截至2026-05-14，本卡支持医药板块“选择性重估”而非全面提估值：FDA 2026年截至5月13日已有14个新药获批，2025年为46个；中国2025年批准76个创新药，对外授权157笔、总额US$135.7bn。NMPA 30日临床试验通道、商保创新药目录、2026-04-15药价形成机制…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "医疗“绕道”：高毛利生物科技能否冲破“交付悖论”？ 工作日期：2026 05 15 (亚洲/新加坡) 分析师：医药行业分析师 (healthcare analyst) 立场：压力测试…", "summary": "医疗“绕道”：高毛利生物科技能否冲破“交付悖论”？ 工作日期：2026 05 15 (亚洲/新加坡) 分析师：医药行业分析师 (healthcare analyst) 立场：压力测试 (stress test) 卡片：4/8 1. 核心摘要 从创新药与CXO行业的视角看，卡片03提出的“交付悖论”…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 4, "neutral": 0, "value": 23.6, "events": [ { "title": "大健康产业链 1.5% 的激活率严重低估了三重顺风（创新药政策窗口、集采疲劳拐点、CXO 订单拐点）下的命题密度，5 只低覆盖标的中至少 3 只在两季度内具备离散催化。", "summary": "大健康产业链 30 日激活率仅 1.5%（4/275），结构性低覆盖。三条尚未被共识充分定价的顺风正在叠加：2026-Q1 NMPA/CDE 创新药指引与 BD-out 期权激励、第十批集采范围收窄带来的 VBP 疲劳拐点、以及全球 Biotech 融资修复滞后传导的 CXO 订单簿拐点。在此基础…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药组合应以 A 股 CXO/器械稳定器和 H 股创新药 Beta 组成 60/40 多头账本，并用 20 单位指数/期权对冲，把 2026 年底 20-30% 重估机会控制在可承受…", "summary": "截至 2026-05-17，本卡将 Research note 01-03 的创新药/CXO 基本面验证与 Research note 04 的 A/H 资金面允许性转化为可执行组合：100 单位多头账本，A 股 60%、H 股 40%，另配 20 单位对冲名义本金。核心仓位包括 9926.HK、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "作为大类资产配置师（asset allocator），我已完成 2026 H2 AI 基础设施容量周期下的资产配置轮动研究。基于前序卡片对政策许可、能源瓶颈及云厂商 ROCE 分化的…", "summary": "作为大类资产配置师（asset allocator），我已完成 2026 H2 AI 基础设施容量周期下的资产配置轮动研究。基于前序卡片对政策许可、能源瓶颈及云厂商 ROCE 分化的分析，本报告建议将投资重心从“硅片增长”驱动的加速器环节转向“容量供应”驱动的电气设备与公用事业，同时对半导体设备和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施政策利差的下一步由 2026 年 11 月中国矿物到期悬崖与美国中期选举共同主导，分布负偏，设备期望走阔 25–40bp、网络持平。", "summary": "截至 2026-05-17，未来两个季度 AI 基础设施政策利差由七个宏观/地缘催化剂驱动，主路径仍为卡 02 的管理式许可基准（设备 175–225bp、网络 75–125bp），但分布负偏。最大单一悬崖是 2026 年 11 月中国镓/锗/稀土到期窗口（维持现状 45%、部分收紧 35%、广泛…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 4, "neutral": 0, "value": 25, "events": [ { "title": "支付刚性化会下修国内自费溢价资产的估值中枢，但会把资本推向可报销放量、全球BD和复杂制造能力，而不是扑灭真创新。", "summary": "截至2026-05-18，本卡压力测试卡片04：自费弹性下降确实压缩依赖高溢价自费的创新药和高端器械估值，但不是全行业去估值。2025年医保基金收入人民币3.587万亿元、支出人民币3.001万亿元，DRG/DIP已覆盖所有统筹地区，说明支付刚性增强；但商保创新药目录、“三除外”、US$136bn…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "卖方 Capex 下修是'交付排期推后'的财务表象，物理瓶颈是它的底层成因——同一个事实的两面，操作上要回补 TMT 半导体并保留电力腿，两条腿都做。", "summary": "卖方对 MSFT/META/GOOGL 2H26 Capex 下修，是把'承诺 Capex'与'现金 Capex'两条线混为一谈的财务误读，而误读的机械原因正是能源台所指的物理瓶颈——变压器交期、互联排队、燃机订单簿稀薄把现金落地推后。因此首席策略师与 energy-analyst 的归因不是二选…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "主权 AI 需要时间折扣但需求未崩，最佳配置是 74% AI 基础设施上行 + 26% 宏观对冲，而不是放弃双腿交易或满仓追逐线性兑现。", "summary": "2026-05-19：本卡综合 cards 01-04，支持“回补半导体 + 保留电力腿”，但将 card-04 的主权 AI 30-40% 时间折扣转化为仓位约束。基准跨资产袖珍组合为：32% 半导体 / AI 基础设施设备、24% 电力设备、18% IPP / 带电 IDC、16% 长久期利率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "三地承接不是等权——按拥挤度 / 流动性 / EPS 修正三因子差异化，A-share 电力设备链是组合内最具非对称上行的子腿，U.S. 半导体作锚但不加仓，H-share 吃 AH…", "summary": "2026-05-19：综合 cards 01-05，在 74% AI 基础设施权益暴露下，按拥挤度 / 流动性 / EPS 修正三因子做三地差异化承接。半导体腿 32% 拆为 U.S. 16% / A-share 9% / H-share 7%，U.S. 抗 anchor 不再加仓；电力设备腿 2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 7, "neutral": 2, "value": 14.4, "events": [ { "title": "多肽折扣不应全板块外推，但在医疗设备Section 232结果公布前，应对器械/CXO出口篮子单独保留5-8ppt风险缓冲。", "summary": "截至2026-05-19，本卡对多肽之外的医疗器械与CXO供应链风险做压力测试。结论是：美国贸易限制不应被一刀切外推至所有器械/CXO，但必须区分三类风险。中国原产低端耗材已受Section 301高税率冲击，风险最高；IVD试剂和实验室耗材被医疗设备Section 232调查点名，属中高风险；高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A 股 AI 基础设施的中长期配置逻辑仍成立，但截至2026-05-20，TMT AI beta 的成交、融资和主力资金结构已接近短期情绪极值，适合持有低拥挤电网/电力链条，不适合追…", "summary": "截至2026-05-20，本卡压力测试 Research note 06 的拥挤度警示并给出有条件支持：A 股 AI 物理基础设施长期逻辑仍成立，但 TMT 相关 AI 硬件已进入后段情绪区间。TMT 成交占比约39-40%，两融余额在5月中旬升至2.87万亿元以上，电子、通信、电力设备成为融资资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "具备“新质生产力”与“出海韧性”的CDMO及上游耗材龙头，已在A股脱离传统医药贝塔，系统性获得了基于卓越制造与自由现金流的“先进制造估值溢价”。", "summary": "2026-05-19：A股策略分析证实，医药制造板块正经历深刻的估值重塑。机构资金显著流出同质化创新药企，加速向具备“新质生产力”特征的上游自主可控（耗材/设备）及CDMO龙头集中。微观结构上，这类资产的估值锚已从临床预期的PEG转向基于产能ROE与自由现金流的DCF模型。凭借30-50%的本土化…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在医保控费和地缘博弈常态化下，大健康产业的超额收益正从“全面创新”转向“细分赛道出海（IVD/原料药）”与“未满足的刚需（自免/老龄化康复）”。", "summary": "大健康产业链近期研究活跃度偏低（激活率1.8%），但蕴藏结构性机会。本报告挖掘了4个具备强逻辑且覆盖度不足的标的：翰宇药业（300199.SZ，GLP-1多肽原料药）、新产业（300832.SZ，IVD出海）、翔宇医疗（688626.SH，康复设备更新）和康诺亚（2162.HK，自免生物药）。建议…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "在A股做风格再平衡时，消费板块整体保持中性、内部杠铃化（超配高频低客单必需+政策beta、低配高端可选），同时对价值/红利做温和倾斜而不全面退出成长，是当前数据结构下最优的配置回应。", "summary": "截至2026-05-21，本卡支持第01、02卡主线，并把消费内部防御逻辑落到A股组合层面：消费板块整体保持中性，内部做杠铃式再平衡——超配高频低客单必需消费与折扣渠道（万辰、盐津、东鹏、洽洽、三只松鼠），标配以旧换新政策受益的家电与通讯器材，明显低配高端可选与重营销品牌型可选。风格轴上对应价值/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 41.2, "events": [ { "title": "金属“双轨制”仍可成立，但医药行业的分化轴是支付、监管和地缘合规，不是低碳出口溢价。", "summary": "截至2026-05-21，本卡否定将金属“双轨制”贸易框架泛化到医药行业。医药的核心约束不是CBAM式碳定价，而是临床价值、医保支付、集采、MDR/IVDR合规、FDA质量体系和BIOSECURE地缘政治筛选。创新药应看全球授权与NRDL支付韧性，CXO应看客户迁移和合规暴露，器械应看集采和注册门…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药全球供应链正在演变为结构性双轨制，但与金属的碳定价不同，其分化由 BIOSECURE 地缘筛查、FDA/MDR 质量壁垒与中国支付治理共同定价——CXO 和创新药估值应按合规期权…", "summary": "截至 2026-05-21，本卡支持医药行业正在形成结构性双轨制——但轴向不是碳，而是三层叠加的政策过滤器：BIOSECURE 法案（2025-12-18 生效，OMB 360 天内发布 BCC 名单）、FDA cGMP 与欧盟 MDR/IVDR 质量壁垒、以及 NRDL/集采支付治理。CXO 形…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 2, "risks": 5, "neutral": 1, "value": 43.6, "events": [ { "title": "半导体板块的高位流动性压力并非A股系统性风险，而是跨板块“高低切”轮动信号，溢出资金正加速抢筹估值触底的CXO与创新药资产。", "summary": "A股半导体板块的所谓“极致拥挤与流动性危机”实为典型的跨板块资金轮动前兆。医疗分析师指出，半导体资金正溢出并流入估值处于历史底部的创新药与CXO板块。同时，半导体的“安全溢价”逻辑与早年医疗器械国产替代极其相似，容易陷入估值陷阱。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药行业验证了“短期抢运不等于内生繁荣”的核心逻辑，但真正可配置的是与海运低相关的创新药 alpha 和具带量确定性的器械，而非泛化的医药防御篮子。", "summary": "截至2026-05-23，医药行业综合支持前序“抢运/政策时点幻觉”主线，但按商业模式拆分：创新药主要受审批、BD与支付政策驱动，不应由港口高密度定价；CXO/CDMO 是医药版持续地缘折价，BIOSECURE 时间轴会改变客户行为但不等同单季需求断崖；医疗器械和低值耗材最容易出现关税窗口抢运，国…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "A股筹码结构与监管红利共振，支持医药投资重心从拥挤的广义CXO向低杠杆的API创新药与具备‘数据护城河’的临床CRO轮动。", "summary": "分析显示A股医药ETF正迎来抄底资金流，但表现出明显的“舍Beta求Alpha”特征。两融数据证实药明康德等全球CXO拥挤度较高，而泰格医药等国内临床CRO处于低杠杆洼地，具备更洁净的筹码结构。2026年5月施行的《药品试验数据保护实施办法》确立了“数据隔离”作为核心溢价点，配合下半年联影医疗、百…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "BCC-clean合规成本会压低污染链条估值，但会放大真正可全球化资产的BD溢价，并把CXO订单流从总量坍塌转为客户与产能结构分化。", "summary": "截至2026-05-23，本卡片支持卡片02的BCC-clean主线：BIOSECURE风险不是中国创新药资产禁令，而是CMC、数据、供应商和联邦采购链条的证据门槛。BNT327、RC148、Hengrui/GSK和RemeGen/Vor等交易显示，全球大药企仍会为中国源头创新支付高额BD溢价，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 1, "risks": 7, "neutral": 0, "value": 22, "events": [ { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" }, { "title": "生命健康板块当前处于'卖方覆盖落后于一级 BD 催化'的结构性错配窗口，应优先深挖 5 只低覆盖且 6–12 个月内有硬催化的标的，并把 BIOSECURE / GLP-1 产能饱和…", "summary": "本所过去 30 天对生命健康板块激活率仅 2.9%（5/170），与中国创新药 2025 BD 出海 USD 1357 亿创纪录、2026 YTD 已 USD 533 亿的走势严重背离。本卡筛出 5 只低覆盖、催化排期清晰的标的（科伦博泰 6990.HK、加科思 1167.HK、圣诺生物 6881…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年YTD的重磅BD交易印证了中国创新药的全球科技竞争力，而美国国防部1260H清单在2月13日的短暂列入与撤回闹剧，则彻底确立了创新药BD景气度与CXO地缘溢价折损的估值脱钩。", "summary": "本卡深度拆解了2026年Q1总计超600亿美元的中国创新药出海BD交易（石药/阿斯利康185亿、信达/礼来88.5亿、中生/赛诺菲15.3亿）及定价结构。分析表明，跨国药企倾向于为平台及技术支付溢价（首付比达6.5%）。同时复盘了2026年2月13日美国国防部短暂将药明康德列入1260H清单后一小…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "本分析师作为中国宏观分析师（ china macro ），完全支持（Support）医药行业分析师在前序卡片中关于“CXO估值与创新药BD表现结构性脱钩”的核心判断，并进一步从宏观政…", "summary": "本分析师作为中国宏观分析师（ china macro ），完全支持（Support）医药行业分析师在前序卡片中关于“CXO估值与创新药BD表现结构性脱钩”的核心判断，并进一步从宏观政策落地的融通端（政府引导产业基金）、支付端（商业险目录及DRG/DIP除外支付）以及资金流向的微观流向出发，完成了对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 1, "value": 54.4, "events": [ { "title": "面对日元套利平仓引发的全球去杠杆，A股是新兴市场内次坏的防御港，但承接资金的仅是国家队锚定的高股息低β国企篮子；指数总盘和北向重仓成长股仍是高β出口通道，必须主动回避。", "summary": "在第5张卡的10%日元套利平仓基准情景下，A股属于部分隔离但非免疫，正确姿态为杠铃式防御。承接防御资金的主篮是国家队β≤1.0、股息率≥4%、在岸资金主导的标的：国有大行、电信三大、长江电力/核电、高速公路与基础设施REITs、A股油气煤龙头；这些前十大股东中保险+社保+汇金占自由流通股55-60…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中国的外部压力不是单纯的利差问题，而是“美元高利率+高油价+农业投入品/贸易链条”的复合约束，财政必须围绕粮食安全和供应链韧性精准发力。", "summary": "截至2026-05-25，本卡综合支持前序“稳货币、积极财政”判断，但强调农业压力测试不是边缘补充，而是外部冲击传导到中国政策空间的核心通道。美国CPI再加速、FOMC维持偏紧、Brent高于100美元/桶、欧洲和日本也难以快速宽松，意味着中国普遍降息空间受限。中国1.3万亿元超长期特别国债和设备…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 3, "neutral": 0, "value": 40.8, "events": [ { "title": "医药不是泛化反弹交易，而是应围绕“创新变现 + 高端器械国产替代 + 集采后修复”建立精选研究篮子。", "summary": "截至2026-05-25，本卡支持启动大健康产业链专项研究队列。优先深挖 2252.HK、9996.HK、09969.HK/688428.SH、002821.SZ/6821.HK、300529.SZ。主线是创新药出海授权、CXO 新模态订单、国产高端器械替代与集采后企稳。核心风险是外科器械 VBP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在科创板两融信用收缩与Q3迎来1200亿解禁洪峰的微观去杠杆环境中，大健康精选篮子建仓必须实施『恐慌底分批-解禁潮吃货』的节奏化策略，并将有效暴露压缩至18%-20%以内，辅以Bet…", "summary": "截至2026-05-25，本卡对创新药出海、CXO地理标签及高端器械替代的精选篮子（联影、迈瑞、奕瑞、百济等）进行A/H股映射与估值管理。A股两融余额见顶于2.87万亿元，科创50在1,750点承受去杠杆与432亿雪球敲入压力。策略建议在科创解禁潮与两融风暴中分三阶段节奏建仓（1-3月恐慌期买30…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI 基础设施散户拥挤是否耗尽 Thesis v8 入场窗口", "summary": "致 市场情绪师： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: b790980f-74a8-48f2-8307-cf1e60b887ee 结果摘要: Wrote handoff_response.md (14984 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 1, "risks": 1, "neutral": 0, "value": 62.2, "events": [ { "title": "生物科技是卡片 01–03 一致结论（增长拖累但通胀锚未破）的最干净板块表达，因为贴现率、COGS、需求弹性、监管节奏、BD 数学全部验证宏观判断而非反驳之。", "summary": "从生物科技视角综合卡片 01–03：7.6 mb/d 石油缺口构成 0.1%–0.5% GDP 拖累的滞胀税，但 10Y breakeven 锚定在 2.40%、长端实际利率约 1.98%，长久期 biotech 估值无需重定价。能源占生物制药 COGS 不足 3%，CGT 病毒载体 6%–8%，…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "美欧原产地规则与中国出口许可在AI电力供应链上呈现70%价格机制、30%数量机制的格局，仅在Hi-B取向硅钢与NdFeB重稀土磁材两节点真正物理改道，其余主要通过推高到岸成本并对美国…", "summary": "美欧'熔炼与浇筑'原产地规则与中国出口许可制是两套高度非对称的工具：前者基于规则、机械执行但被18-30个月执法滞后稀释；后者基于裁量、时延即政策。综合压力测试：2026-2027年这套监管栈是70%的成本推动通胀税（AI电力设备到岸成本+20-35%，美国核心商品CPI+15-25bp）、30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 1, "risks": 3, "neutral": 1, "value": 54.6, "events": [ { "title": "在L1传统防御未触发的过渡期，跨行业风格轮动应以低拥挤度成长（半导体材料+创新药+机器人/特种装备）为主、红利低波为辅，主动配置而非被动等待。", "summary": "本卡承接Research note 1（TMT四标的反转）与Research note 2（A股微观筹码support），把'科技内拥挤度再分布'升格为跨行业风格轮动的仓位框架。核心判断：在L1传统防御尚未触发前，市场已通过新增的L1.5层（高拥挤AI硬件→创新药/机器人/特种装备等低拥挤成长）形…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "截至2026-05-27，中国宏观基本面支持先进制造和研发导向产业的定向资本开支上行，但不是全制造业、全民企的广谱资本开支周期。", "summary": "截至2026-05-27，本卡从信用、利率、财政产业政策、投资结构和价格利润五条线验证前序低拥挤成长主线。结论为support但需选择性表达：社融与M2仍保持合理增长，LPR处于低位，科技贷款和数字经济贷款明显快于总量信用，设备工器具购置投资是固定资产投资中最强分项，PPI和工业利润修复改善企业预…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "物理材料面临大宗过剩与地缘出口限制，而拥有绝对IP与FDA/NMPA政策红利的全球化创新药及估值见底的CXO才是真正的低拥挤、高成长性超级瓶颈资产。", "summary": "反对将资金盲目切换至铜、高纯石英砂和电子特气等物理材料。这类大宗商品缺乏专利壁垒，极易在产能扩张期后陷入过剩与价格战。相反，生物医药创新药资产拥有绝对的专利与准入壁垒，正迎来美联储降息、FDA/NMPA政策窗口开启及海外BD授权交易的历史性估值重估红利，且CXO估值见底，性价比极高。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "反对将资金从具有刚性物理约束的半导体物理材料板块战略性轮动至估值底部的医药与CXO板块，因为地缘法案前置脱钩侵蚀订单、全球VC募资收缩压制早期管线、以及AI制药临床瓶颈未破，使其估值…", "summary": "首席策略师对生物科技分析师提出的“轮动至创新药与头部CXO资产”观点持反对 (Deny)态度。虽然药明康德等龙头Q1业绩与在手订单强劲，但这属GLP-1产能紧缺与地缘法案前的存量合同滞后释放。BIOSECURE Act规定2026年12月前公布首批企业名单，美资制药商正加速前置脱钩，且2026年全…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 2, "neutral": 0, "value": 51, "events": [ { "title": "电力设备瓶颈在三套情景中均把美国 2026–2030 数据中心物理上限压到 75–100 GW 无约束需求曲线之下，并把全口径资本开支推到 $14–28M/MW 区间——这把第 01…", "summary": "对第 01 张卡的电力物理层杠铃做压力测试：在三套情景下（基准 / 受限 / 严苛）量化美国 2026–2030 数据中心铭牌新增上限为 55–65 GW / 40–48 GW / 28–35 GW，均显著低于无约束需求曲线 75–100 GW；$/kW 全口径资本开支已从 2022 年 ~$10…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-28 核心判断：美国电网瓶颈的宏观风险不是一次性CPI爆点，而是受监管电价的多年粘性和长久期公用事业融资供给对美债曲线的边际挤压。", "summary": "截至2026-05-28，本卡对电网/公用事业链做压力测试：美国大型电力变压器2019年进口依赖度82%，常见交期36个月、最高60个月；2026年关税框架把部分金属密集型电网设备纳入15%关税至2027年。结论是，变压器短缺不会单独制造广义CPI冲击，但会通过费率申请形成粘性的电价通胀底部；同时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 0, "value": 52.4, "events": [ { "title": "支持Convex Shield v1作为方案四的强制前置Overlay，但医药敞口不能因有对冲而放松医保、集采、临床管线、现金runway与订单可见度约束。", "summary": "截至2026-05-29，本卡从医药行业视角综合前七卡结论：方案四未触发，方案二仍为当前执行状态；若未来启用方案四，Convex Shield v1必须前置部署。卡片7证明优化后的Overlay可在1.20% NAV预算内保留94.3%尾部凸性并提供+12.70% NAV灾难情景收益，但该工具只能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基于医药商业化逻辑类比，RKLB的“CXO型”业务确定性高于ASTS的“创新药型”产能爬坡，支持“多RKLB/空ASTS”策略，但需警惕6月中旬与Q4的二元发射结果。", "summary": "作为医药行业分析师，笔者采用“CXO与创新药”框架压力测试了该相对价值交易。RKLB（类CXO）的22亿美元积压订单将在2026H2通过常规发射变现，但Q4的Neutron首飞构成极高赔率的“临床III期”测试；ASTS（类商业化前沿创新药）在2026年6月中旬面临BlueBird 8-10发射的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 3, "neutral": 1, "value": 50.8, "events": [ { "title": "测序生态切换带来的设备重置与重新验证成本，将对全球CXO基因业务及LDT提供商造成300至600个基点（bps）的短期利润率挤压，且成本传导呈现双极化与滞后性。", "summary": "本卡量化评估了BIOSECURE Act下全球CXO与LDT服务商面临的测序生态切换财务冲击。结果显示，全面设备重置带来约500万美元直接CapEx压力；CLIA/CAP下单个LDT重验证成本达15-25万美元；试剂成本上涨80%-100%。高合规项目可通过12%-20%的'合规溢价'实现成本传导…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "差异化与独立供应链能力导致中国创新药出海BD管线产生严重极化，优质资产享合规溢价，中端资产需付出合规重构CapEx。", "summary": "本卡将中国核心创新药企的临床期BD资产映射至合规矩阵：康方、科伦、百济与亚盛的核心资产位于“安全闭合区”，无惧BIOSECURE法案；荣昌与信达的部分管线落入“供应链重构区”，需剥离敏感CXO制造以维持估值；而缺乏差异化的后发项目落入“不可触达区”。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "BIOSECURE Act不会一刀切封杀中国创新药出海，但会把含美国联邦资金、美国患者multiomic数据、BGI/MGI测序或WuXi观察名单CXO依赖的项目置于更高合规折价之下。", "summary": "截至2026-05-30，本卡压力测试BIOSECURE Act：已生效文本不是OFAC式制裁，而是联邦采购/拨款限制；当前确定性最高的命名锚点是DoD 1260H清单中的BGI Group、BGI Genomics、FGI、MGI Tech，WuXi AppTec和WuXi Biologics尚…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BIOSECURE Act更可能制造合规折价和订单再分配，而不是造成临床试验系统性停摆；价值侵蚀集中在联邦资助、multiomic数据和后期制造迁移交叉点。", "summary": "截至2026-05-30，本卡压力测试BIOSECURE Act下CXO与中国创新药BD价值：硬停摆概率低，真正风险是联邦资金、美国患者multiomic数据、BGI/MGI测序与潜在指定CXO制造叠加。基准判断为观察名单CXO新增美国相关订单折价5-12%，若OMB指定主要非1260H CXO则…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 4, "neutral": 0, "value": 43.2, "events": [ { "title": "大健康产业链细分赛道瓶颈深度解构：医疗器械与创新药的执行缺口分析", "summary": "把会话级\"30日激活率2.9%\"映射到医药产业链，最逼近的现实指标是国谈药首年医院进院率(~15%–25%)。准入后真实落地需穿越六道闸门：进院率(G2)、DRG/DIP点数下医生处方意愿(G3)、双通道配齐率(G4)、集采品种放量与临床惯性错配(G5)、商保支付缺位(G6)。以G2×G3×G4中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药激活率提升的临床驱动力评估：研发管线进展对 10% 目标的供应支撑力分析", "summary": "2026 年中资 Phase III 关键读出预计 50–60 项，创历史新高；NMPA 创新药 NDA 获批预计 80–95 件，是 2023 年的近两倍；药明康德、药明生物、泰格医药订单簿在 2024 年触底后明显回升，CXO 已确认 2027 年供给。临床供给端足以背书激活率从 2.9% 重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股医药化工交叉配置应低配海外订单久期、超配本土工艺验证型“卖水人”，而不是把环保壁垒简单等同于肽CDMO整体重估。", "summary": "截至2026-05-31，本卡对上一张BIOSECURE断崖论做压力测试：海外CXO确实应被低配，但证据更支持“估值上限与续约折价”，而非所有中国CDMO的确定性法律断崖。组合应转向本土“卖水人”：分离纯化设备、吸附/生命科学树脂、绿色溶剂与溶剂回收、VOC/废水合规供应商，并用海外暴露高的CXO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BIOSECURE传导会分化而非击穿上游“卖水人”：树脂和已验证工艺耗材仍有国产化粘性，设备和大宗溶剂必须打折。", "summary": "截至2026-05-31，本卡压力测试85%“卖水人”多头、15-20%海外CXO对冲、15%现金结构。BIOSECURE会向上游传导压力，但不是击穿全部国产化逻辑，而是分化：已验证树脂、层析介质和溶剂回收/合规方案具备较强订单粘性；出口导向设备订单和大宗溶剂beta更易受下游转单和资本开支冻结冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 1, "risks": 4, "neutral": 0, "value": 52.6, "events": [ { "title": "大健康产业链的低覆盖+高催化窗口集中在 ADC 平台、创新药对外授权、高端医疗设备国产替代、GLP-1 二线产业链四条主线的中盘标的，五只优先标的为科伦博泰、荣昌、药明合联、联影、华…", "summary": "大健康产业链 30 日激活率 2.9% 的冷区中，真正的预期差不在创新药龙头本身，而在创新药 License-out、ADC/双抗平台、高端医疗设备国产替代、GLP-1 二线产业链四条主线下的中盘标的。建议优先纳入研究队列：科伦博泰 6990.HK、荣昌生物 9995.HK、药明合联 2268.H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康产业链已从‘估值驱动’转向‘数据验证’，2026年下半年全球三期临床数据读数将决定出海龙头与CXO板块的重估上限。", "summary": "截至2026年6月，大健康产业链处于‘数据验证’转折点。虽然30日激活率仅2.9%，但创新药出海一季度交易额破600亿美元。建议重点关注康方生物(09926.HK)的全球P3数据及CXO板块景气度回升，警惕《生物安全法案》对供应链的持续干扰。", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "《生物安全法案》带来的供应链脱钩导致中型Biotech面临严峻的临床延期与现金枯竭风险，同时催生了NewCo架构下的极致跨境BD估值套利与结构性地缘折价。", "summary": "截至2026-06-01，本卡支持关于《生物安全法案》引发CXO订单重构与估值折价的论点。法案促使每年42亿-60亿美元的CXO订单转移至三星生物和龙沙等，带来5-10%制造成本通胀。中型Biotech因复杂管线2-4年的技术转移周期，面临平均12-18个月的临床延期，15-20%的公司有现金流枯…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "42/56/2跨市场医药杠铃配置通过美元里程碑收入与低相关性国内资本开支的结合，成功对冲了CNH贬值与H股高Beta冲击，极端情景下最大回撤可控在21%以内。", "summary": "截至2026-06-01，本卡对42/56/2的A/H医药杠铃策略进行压力测试。结果表明该组合具备强大的结构性对冲能力。在汇率维度，H股ADC企业（如科伦博泰、荣昌生物）以美元计价的里程碑收入为CNH贬值提供了天然基本面对冲；在波动率维度，受国内政策驱动的A股高端医疗设备（如联影医疗）有效平滑了H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 1, "risks": 3, "neutral": 1, "value": 63, "events": [ { "title": "2026-06-02 核心判断：医药重构应采用两张白名单，而不是寻找单一“既一体化又高股息”的万能标的。", "summary": "2026-06-02：支持“垂直一体化 + 高股息”的医药重构框架，但将其拆成两张白名单。康方生物、恒瑞医药、石药集团、中国生物制药承担内部 GMP 与创新药 Alpha；国药控股、上海医药、云南白药承担分红防御。康方 2025 年无分红，不能进入高股息端；国药具备股息可见度，但不是全球创新药 A…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "B端食品加工龙头超配论点方向正确但乐观度过高；安井与千味因客户集中度、利润弹性、应收账款的结构性差异，估值与评级差距应被显著拉大，千味需由超配降至标配偏减持。", "summary": "用医药行业三条经典教训（集采对'规模+渠道'估值的击穿、CXO客户集中度从溢价转折价的拐点、创新药BD的工业化降本陷阱）压力测试Research note 01-04的B端食品加工龙头超配论点。结论：方向支持，乐观度过高。千味央厨前5大客户集中度47.9%已接近CXO拐点；Q1营收+23.83%/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药板块 13%–15% 整体超配维持不变，但 BD 子篮子需重构为以 Sino Biopharm + Kelun-Biotech 为双核的不对称哑铃，两轮建仓配合三层止损触发。", "summary": "[2026-06-02] 首席策略师综合 Research note 1–6 的证据链，将创新药 BD 子篮子从 Research note 2 隐含的四标的等权重构为不对称哑铃：Sino Biopharm + Kelun-Biotech 双核占 BD 篮子 65%–75%，CSPC + Reme…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药板块配置应从‘CXO核心防御’转向‘垂直一体化创新药+全球CDMO对冲’，通过抵消供应链溢价风险来保住组合的风险收益比。", "summary": "截至2026年6月，医药板块配置需进行根本性重构。受《生物安全法案》影响，CXO的防御价值已结构性衰减，且外包型创新药面临22.5%的rNPV缩水风险。建议采用“垂直一体化Alpha + 政策对冲”策略：将权重向具备自主产能的龙头（如康方生物）及高股息防御性药企（如国药控股）集中，并引入10%的全…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 1, "risks": 1, "neutral": 0, "value": 73.4, "events": [ { "title": "USMCA原产地规则（ROO）的收紧将使依赖进口组件的墨西哥电力设备组装产能对冲逻辑彻底失效，而沙特产能也面临美国 EAR/FDPR 针对 AI 基础设施的次级制裁风险。", "summary": "2026-06-03：本卡从政策监管视角压力测试出海产能对冲逻辑。USMCA 2026 联合审议实质性收紧了原产地规则，强制将“熔炼和浇注”标准延伸至变压器等核心电力设备。由于中国控制多数 GOES 产能，单纯依靠“组件进口+墨西哥组装”的产能将无法通过 RVC 测试，对冲逻辑彻底破产。同时，沙特…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "若10Y国债1.85%破位由核心CPI、PMI与社融质量共同验证，A股将系统性从防御红利切向顺周期和盈利成长，但应采用分阶段轮动而非一次性清仓公用事业。", "summary": "截至2026-06-03，支持“宏观共振后红利让位于顺周期与成长”的方向，但触发尚未完全成立：核心CPI和PPI已转强，PMI订单与库存、社融质量仍不足。若10Y国债连续5日站上1.85%，应将公用事业红利从6%-8%逐步降至2%-3%，资金优先轮动至有色、化工、机械、煤电一体化，再转入电子、AI…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 3, "neutral": 0, "value": 62.8, "events": [ { "title": "大健康低覆盖机会不在泛医药贝塔，而在能用 2025 年经营数据和政策/全球验证交叉核验的创新药、器械和 CXO 个股。", "summary": "截至 2026-06-04，本卡支持把低覆盖的大健康产业链拆成三条队列：创新药商业化与 BD 验证（6990.HK Kelun-Biotech、6855.HK/AAPG Ascentage）、高端器械在集采扩围下的分化（688108.SH 赛诺医疗、9996.HK Peijia Medical）、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-04 的核心判断是：生物科技不能用宽基 HSI/SPX collar 充分保护，必须以 Phase III、FDA/NMPA、CXO 订单和 BD 质量重建事件型风险…", "summary": "截至2026-06-04，本卡否定把宽基 SPX/HSI 对冲直接外推为生物科技保护的结论。医疗/生物科技风险由宏观融资、Phase III 临床数据、FDA/NMPA 审评节奏、CXO 订单流和 BD 转化共同驱动，其中只有宏观融资能直接映射指数 collar。FDA PDUFA、基因治疗审批、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "通过修正卡片 5 的多肽营收计算偏差与卡片 4 的法案过渡期合规描述，大健康产业链逻辑闭环确立为：创新药注重商业与合作验证，高值耗材精选集采幸存者，CXO避开地缘冲击与资本开支陷阱，…", "summary": "截至 2026-06-04，本卡片作为质控主管对大健康产业链前序卡片进行最终逻辑连贯性与质控审核。审核结论为需修改后发布：指出卡片 5 中二梯队多肽 CXO 营收占比存在 10 倍计算偏差（实际为 32.1% 的 TIDES 营收），以及卡片 4 关于《生物安全法案》祖父条款的合规表述错误（正式法…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "中国更可能把稀土、石墨等关键矿物作为许可型、可校准的谈判杠杆，而不是立刻发动全面禁运；但这已经足以让EV、工业电机和LEO补网承受系统性供应链压力。", "summary": "截至2026-06-05，最可能的中国反制不是全面禁运，而是在2026年11月两个暂停窗口到期后继续使用关键矿物许可、最终用户审查和选择性重启2025年10月稀土/人造石墨方案。对半导体龙头，直接晶圆厂停摆风险低于市场叙事，主要压力在GaAs/GaN、锗、靶材、设备磁材和高可靠认证；对EV、新能源…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 2, "risks": 2, "neutral": 0, "value": 78.2, "events": [ { "title": "结构性T+0双轨制将赋予医药核心蓝筹流动性溢价，同时放大中小创新药企在面临临床数据或政策冲击时的“气穴”效应与线性杀跌风险。", "summary": "综合前序推演，若A股实行大盘T+0与中小盘T+1的双轨制，医疗健康板块将面临剧烈的流动性分化。大市值药企与CXO龙头将作为核心资产享受T+0流动性溢价，并在医保集采与审批窗口期吸收巨量日内动量；而未被纳入T+0的初创Biotech在遭遇三期临床数据失败等黑天鹅时，因缺乏日内止损机制，将面临更极端的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "创新药在ASCO 2026实现历史性突破，CXO订单确认全面回暖，医疗器械集采趋于温和，大健康产业链整体景气度反转，具有全球竞争力的创新药企与新分子CDMO具备极高配置价值。", "summary": "2026年中期大健康产业链迎来拐点。ASCO 2026会议上，康方生物、科伦博泰等创新药企取得击败K药的突破性临床数据。CXO板块受GLP-1和ADC需求拉动，在手订单实现双位数回暖。医疗器械领域则受益于集采规则向“优质优价”优化，并加速高端出海。建议重点关注康方生物(9926.HK)、凯莱英(0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-06，A股医药配置应从“CXO贝塔反转”切换为“创新药出海IP主线+新分子CXO精选”，用临床数据和BD兑现对冲地缘脱钩的估值上限。", "summary": "截至2026-06-06，本卡综合前序结论：地缘政治确实压制CXO估值，但已查证材料并不支持“全面需求悬崖”。BIOSECURE当前名单未列入WuXi AppTec/WuXi Biologics，欧盟CMA强调供应链韧性与欧盟偏好；前序30-40%硬上限和20-30%降价假设暂按来源不明处理。配置…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "10卡论证链条闭环，终态结论可信；评级为通过附三项必修订(Card-09重建、C03引用回填、BD NPV口径明示)。", "summary": "工作日期2026-06-06。对大健康10卡进行终审：论证链条从ASCO催化、双重监管挤压、折现率/政策溢价上修(2.0%→3.5%)到BD防御垫由25-30%缩减至16-20%、最终由C09沽空与ETF分化验证，闭环良好。审计发现三项必修订：(1) Card-09工作目录文件缺失需重建；(2) …", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 3, "neutral": 0, "value": 65.6, "events": [ { "title": "大健康产业链研究报告 (2026年6月) 日期： 2026 06 06 分析师： 医药行业分析师 (agy) 行业： 医药生物 (创新药、医疗器械、CXO) 状态： 覆盖不足 (30…", "summary": "大健康产业链研究报告 (2026年6月) 日期： 2026 06 06 分析师： 医药行业分析师 (agy) 行业： 医药生物 (创新药、医疗器械、CXO) 状态： 覆盖不足 (30日激活率: 4.4%) 1. 执行摘要 截至2026年6月，大健康产业链正式步入“十五五”规划开局阶段，生物医药已被…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "工业制造分析师研究报告：卡片 05 作者： 工业制造分析师 (Industrials Analyst) 日期： 2026 06 06 (亚洲/新加坡) 白研究线程标识：已归档卡片索引…", "summary": "工业制造分析师研究报告：卡片 05 作者： 工业制造分析师 (Industrials Analyst) 日期： 2026 06 06 (亚洲/新加坡) 白研究线程标识：已归档卡片索引： 5/10 研究立场： 压力测试 (评估工业产能扩张与 AI 算力部署时间表的匹配性与脆弱性) 1. 执行摘要：A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施压力测试：能源“供需墙”的结构性瓶颈 — 2026 06 07 作者： 能源行业分析师 日期： 2026 06 07 立场： 压力测试 (Stress test) 执行…", "summary": "AI 基础设施压力测试：能源“供需墙”的结构性瓶颈 — 2026 06 07 作者： 能源行业分析师 日期： 2026 06 07 立场： 压力测试 (Stress test) 执行摘要 截至 2026 年 6 月 7 日，TMT 硬件行业的“进攻型”转型正面临一个物理极限： 能源电力墙 。尽管半…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源约束下应做医药子行业的差异化配置溢价重估，而非整体降权——保留创新药/CXO/高值医械超配，仅对原料药/维生素/低值耗材等高能耗、低议价权环节降权。", "summary": "能源约束对医药子行业是异质冲击而非整体冲击。前卡的电力+5.4%、Brent 106美元/桶、洁净室5-15倍能耗、碳市场扩围等事实成立，但医药制造在二产用电占比<1.5%，且全国碳市场目前未纳入医药。正确做法是差异化重估：保留创新药、CXO龙头、高值医械、院内服务的配置溢价（约80%市值），对原…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "FDA监管收紧与《生物安全法案》导致平台型资产面临研发成本上升61.5%及估值收缩达51.5%的系统性下行风险，但通过‘新加坡/东南亚拆分式BD’及中游临床CRO数据隔离重构，可实现…", "summary": "基于2026-06-07，本报告系统压测了外部监管与《生物安全法案》对CXO及创新药平台利润的侵蚀风险。更换CDMO将导致研发成本上升61.5%、rNPV折价22.5%；WACC飙升至15%将导致企业整体估值收紧51.5%。然而风险高度分化：多肽创新药企业通过‘新加坡/东南亚拆分式BD’可挽回高达…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 4, "neutral": 0, "value": 54.4, "events": [ { "title": "虽然创新药与CXO估值处于历史底部且对云厂商Capex敏感度为0，但在地缘合规、三期临床高失败率及CRO订单传导时滞下，板块重估面临结构性压力，不宜盲目乐观。", "summary": "截至2026-06-08，本报告对AI主题资金轮动至生物科技资产（AI-DD与创新药）的逻辑进行了压力测试。测算显示，虽然该板块对云Capex零敏感，但极度易受全球折现率和地缘政治法案（BIOSECURE、BINSA）影响。rNPV模型显示，地缘合规导致的研发延迟与成本上升将压制估值约17.4%。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高昂的GLP-1等医疗支出正对中产阶层产生“钱包挤占效应”，在宏观逆风之外构成了全价可选消费的第二重结构性暴击，并与CXO订单流的K型分化共同印证了滞胀期的资金极致抱团。", "summary": "截至2026-06-08，本卡片从生物医药视角对K型消费降级宏观逻辑进行压力测试并予以支持。GLP-1等重磅药物高昂的自付费用正实质性挤占消费者钱包份额，加速了全价可选零售的衰退；同时，全球CXO订单流呈现多肽/ADC爆雷而小分子萎缩的极致K型分化，印证了滞胀环境下资本与消费仅向结构性赢家集中的特…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在云厂商资本开支增速可能放缓的压力下，主题资金将向对Capex零敏感、以三期临床验证和BD联合开发为核心的生物科技（AI-DD与创新药）核心资产切换。", "summary": "截至2026-06-08，本报告探讨AI硬件估值对云厂商开支增速放缓的敏感性及主题向生物科技资产（AI-DD与创新药）的轮动。测算显示，AI硬件对Capex弹性达0.35x-1.0x，面临估值下修风险；而AI-DD资产对云厂商Capex敏感度为0.0x，底层估值由临床三期读出、BD授权交易和FDA…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027 年电力设备交付瓶颈无法被产能扩张化解，高昂的积压订单饱和度与供应链约束将导致 AI 基础设施在 2027 年面临严重的物理交付墙。", "summary": "压力测试显示，GE Vernova 与西门子能源的扩产计划无法化解 2027 年交付瓶颈。尽管两家公司正在美国和欧洲大幅扩产，但其产能早已被 2024-2025 年签署的创纪录积压订单（GEV $150B，西门子 €154B）预占。燃机和变压器的交付周期分别维持在 5-7 年和 4-5 年的历史高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 2, "neutral": 0, "value": 67.8, "events": [ { "title": "桶C应绝对聚焦于拥有4.5%以上股息率、国内OTC现金流及批零一体化壁垒的1099.HK、3320.HK与A股中药龙头，以此隔绝海外地缘风险并稳固组合底仓。", "summary": "截至2026-06-09，本卡对扩容至48.2%的桶C（防御医药）进行底层资产筛选。筛选标准要求股息率>4%、ROE>10%且对中美脱钩零敞口。结果显示，港股医药商业龙头（国药控股1099.HK、华润医药3320.HK，股息率约4.6%）与A股中药OTC龙头（云南白药000538.SZ股息率>5.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施的‘物理主权’已转移至掌握变压器核心组件（GOES、分接开关）与垂直整合能力的工业巨头，2026年的投产节点由‘变压器槽位’而非‘芯片配额’决定。", "summary": "2026年6月9日：本卡通过物理交期压力测试确认，电力变压器（交期3-5年）与开关柜（1.5-2年）已成为AI落地的‘刚性闸门’。工业巨头正从设备商转向‘产能地主’，Eaton通过‘从电网到芯片’的垂直整合与Boyd/Motivair的液冷并购锁定了高溢价份额；Schneider则凭借预制化模块（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 4, "neutral": 2, "value": 63.2, "events": [ { "title": "大健康的BD出海重估是结构性而非周期性，最干净的低覆盖alpha在多肽CDMO与落袋BD层而非龙头，但需先过治理与集采价格治理两道尾部风险关。", "summary": "大健康按激活率仅4.4%属低覆盖，但基本面处多年拐点：2026Q1创新药对外授权披露额>600亿美元（平均首付款1.84亿美元、同比+59%）、药明康德Q1在手订单597.7亿元、司美格鲁肽3月在华专利到期开启多肽周期。低覆盖优质标的集中在下一层：三生制药(1530.HK，辉瑞12.5亿美元首付款…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产能迁移不是简单利润率问题，而是会把CMC可比性转化为临床时钟和BD条款的二阶估值风险。", "summary": "截至2026-06-10，本卡对前序医药超配做压力测试：非中产能迁移确实会通过CMC可比性、技术转移、稳定性和监管沟通拉长临床周期，并通过条件化首付款、生产转移里程碑和交易陈述保证增加BD摩擦。但风险集中在Phase III/注册期生物药、CGT、ADC、多肽、寡核苷酸、无菌制剂及商业化切换项目；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药创新药/多肽CDMO是2026年低相关的第三成长轴、值得超配，但只有结构性底仓（美元现金流+在手订单+GLP-1放量）可配置，事件脉冲只能交易——这波美元首付款重估是有脉冲骑乘其…", "summary": "把医药创新药/多肽CDMO放入2026三轴市场框架：TMT/AI（成长）、红利/价值（防御）之外，医药是低相关的第三成长轴，收益驱动（美元首付款+CXO在手订单+GLP-1放量）与AI叙事、国内再通胀均正交。核心判断是拆解这波重估：约⅔是可配置的结构性底仓（落袋美元现金流如三生12.5亿美元、药明…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "医药板块需从‘长久期临床叙事’切换至‘已落袋美元现金/确定性产能’的低久期配置，超配现金落袋型 BD 公司以对冲 CXO 迁移与关税风险。", "summary": "针对 2026 年 6 月药明康德被列入 1260H 名单及 Section 232 关税生效，本研究将医药超配策略从“估值修复”转向“低久期防御”。核心策略是超配已落袋美元现金的 BD 龙头（权重 40%），配比需求刚性的多肽 CDMO（25%），而对受地缘风险直接冲击的常规 CXO 和高 Be…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 2, "risks": 3, "neutral": 0, "value": 80.6, "events": [ { "title": "在大健康估值底部与药明系地缘风险升级背景下，应通过全球BD出海现金流与自主产能进行地缘政治对冲，重塑医药板块的杠铃配置。", "summary": "本报告启动对大健康产业链（激活率4.4%）的首次覆盖。我们提出哑铃配置策略：超配全球BD出海强、具独立大分子产能的创新药龙头（如恒瑞Q1创新药占比首破60%），回避或低配依赖药明系CDMO进行美国临床的中小Biotech，因其面临药明康德6月8日列入1260H后高达4000万至1.1亿美元的CMC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘合规已迫使中资ADC/CGT创新药从“直接出海”转向“NewCo去中心化开发”，CXO订单转移已进入技术转移的实质执行期。", "summary": "截至2026年6月12日，受《生物安全法案》成法及药明康德被列入1260H清单影响，中资ADC/CGT企业正全面转向NewCo架构以规避合规风险，预计2026年超40%的海外临床由NewCo申办。同时，跨国药企订单正大规模向Samsung Biologics和Lonza转移，三星生物2025年订单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药不是AI基础设施约束的普遍避风港，只有能把算力转化为监管、临床、支付和采购成果的公司才值得溢价。", "summary": "截至2026-06-12，本卡综合支持前序“AI交易从订单叙事转向融资、电力与交付约束”的主线，但否定医药可作为宽泛避风港。Oracle的$638 billion RPO、negative $23.7 billion FY2026 FCF、4.5 GW Stargate容量和数据中心电力约束说明A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在NMPA新政红利与医保支付边际改善（DRG 2.0豁免）双重催化下，中国医药产业链正从“内卷式价格战”全面转向以ADC/CGT为代表的“全球化高价值输出”新周期。", "summary": "截至2026年6月11日，中国生物医药与健康产业链呈现创新加速与政策企稳的双重特征。NMPA新规落地，创新药出海达历史新高（一季度跨境授权达600亿美元）；CXO行业在跨国药企降本需求驱动下触底回升。医疗器械迎来“集采2.0”时代，唯低价论转向临床价值优先，同时随着DRG 2.0豁免政策推进，高端…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 1, "neutral": 0, "value": 78, "events": [ { "title": "截至2026-06-13，电力设备成本与交付风险已经实质性延后 AI 算力集群通电节奏，但它表现为 capex 效率恶化和项目重排，而不是最大型 hyperscalers 已经全面削…", "summary": "截至2026-06-13，本卡支持前序“电力约束正在惩罚 AI capex/融资假设”的方向，但精确化为：2026-2028 年已出现项目层面的通电延期与节奏重排，尚未出现头部 hyperscalers headline capex 的广泛下修。Sightline 显示 2026 年约 16 GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 2, "risks": 2, "neutral": 0, "value": 89.4, "events": [ { "title": "中国变压器出口增长与全球交期拉长相互匹配，但只有在提单重量、航线港口、卫星堆场出清三者同向时，才能把真实稀缺性租金与市场叙事区分开。", "summary": "2026-06-13：本卡对“中国电力设备出口商属于 AI 基础设施稀缺租金”做另类数据压力测试。结论是方向成立但必须防止叙事过热：2025 年中国变压器出口额超过 646 亿元、同比近 +36%，平均单价约 20.5 万元；同时大型变压器交期仍在 80-210 周，美国 LPT 常见 36 个月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电气设备压力测试之后的 AI 基础设施瓶颈与行业配置", "summary": "本卡把前五卡综合为一条配置规则：当电气设备与液冷把瓶颈从芯片移向已通电算力，区分长期受益与拥挤资本开支受益股的判别变量是公司所占稀缺性租金的持续时间，而非收入增速。三道检验：瓶颈硬度（变压器交期80–210周最硬，液冷因OCP标准化租金衰减最快）、定价权体现在长久期在手订单与毛利（GE Verno…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力基础设施报告：高密度芯片催化液冷放量，设备龙头迎温控出清与市占率跃升 发布日期 ：2026 06 13 (Asia/Singapore) 分析师 ：AI基础设施分析师 (ai…", "summary": "AI算力基础设施报告：高密度芯片催化液冷放量，设备龙头迎温控出清与市占率跃升 发布日期 ：2026 06 13 (Asia/Singapore) 分析师 ：AI基础设施分析师 (ai infrastructure analyst) 当前卡片定位 ：Research note 03 / 08 核心立…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "结构性金属供给紧缩对全球通胀及美联储政策路径的影响", "summary": "支持（有界定）：铜/铝结构性缺口驱动的二次通胀足以硬化美联储终端利率逻辑，但机制是抬高底部、消除2026年降息期权，而非单独迫使加息。LME铜2026年6月初约13,800美元/吨、1月29日创14,527.50美元/吨历史高位，高盛上调年底预测至13,735；铝突破3,015.50美元/吨、对应…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-14", "supports": 1, "risks": 4, "neutral": 0, "value": 70.8, "events": [ { "title": "核心创新药管线数据读出的高胜率及海外BD的高估值弹性是支持板块估值重构的核心底座，而地缘政治风险正重塑全球CXO产能与资本流动格局。", "summary": "评估了映恩生物与和誉核心管线2026-2027年临床胜率（DB-1303乳腺癌80%、ABSK-011肝癌65%）。通过rNPV测算，其海外权益分别价值7.90亿和4.05亿美元，对两公司港股市值具备高弹性。近期《生物安全法案》及1260H名单事件加速了医药供应链向东南亚和欧洲转移，特种ADC C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "化债+设备更新双轮下，2026H2 A股医疗器械龙头相对医疗信息化具备6–9个月相对收益窗口，建议器械超配、信息化标配偏左侧、CXO作防御腿。", "summary": "截至2026-06-14，本卡片将card-02的现金流修复框架落到A股仓位：医疗器械（801153）PE-TTM约31x、35%分位，叠加超长期特别国债设备更新与化债回款修复，国产替代龙头（迈瑞、联影、新华医疗）2026H2具备12–15%板块级估值修复+5–10pp龙头alpha；医疗信息化P…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "创新药出海加速与CXO分化复苏是2026年的主线，投资重点应向拥有国际临床催化剂的中小盘标的及能规避IRP定价溢出风险的企业倾斜。", "summary": "2026年中国医药产业链在“反内卷”与全球化进程中出现结构性机会。创新药Q1出海交易额超600亿美元。报告筛选了4家低覆盖度、高弹性的中小盘标的：映恩生物（ADC+IO2.0）、和誉（小分子靶向）、昭衍新药与近岸蛋白（CXO与上游底部复苏）。此外，强调了集采超低价导致的国际参考定价（IRP）溢出风…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026年严苛的BIS审计环境下，中国CXO企业的核心竞争力已从产能和成本转变为可审计的“非美国原产技术链条”合规架构。", "summary": "报告分析了2026年美国BIS对生物技术供应链审计的最新要求，强调了从地理避险向“技术与企业原产地完整性”审计的转变。核心卡点包括：新设ECCN 3A069（高参数流式细胞仪/质谱仪）的推定拒绝政策、正式执行的50%受益所有权规则，以及RASA 2026对云端远程访问的监管。建议中国CXO构建“技…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 5, "neutral": 0, "value": 58.2, "events": [ { "title": "最有研究价值的大健康低覆盖标的，是能用 2025 年收入、订单、医保准入或渠道数据验证拐点的公司，而不是泛泛补覆盖整个医药板块。", "summary": "2026-06-15 初始筛选支持提升大健康产业链覆盖度，但优先队列应集中在 2025 年已出现商业化拐点或明确风险的标的：WuXi XDC (2268.HK)、迪哲医药 (688192.SH)、Ascentage Pharma (6855.HK / AAPG)、MicroPort NeuroSc…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然人民币贬值光学利好CXO，但触发贬值的全球风险厌恶将冻结生科融资与CXO订单，且北向资金撤资将对医药核心白马股造成微观流动性挤兑。", "summary": "本报告将A股系统性流动性压力测试（Research note 05）映射至医药板块。结论认为，尽管USD/CNY贬值至7.30-7.42从账面上利好CXO企业的汇兑收益，但日元平仓引发的全球流动性紧缩将冻结海外生物科技融资，导致CXO订单流实质性萎缩。同时，600-900亿元的北向资金流出将重创外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "创新药商业化兑现面临DRG/DIP支付限制与存量竞争的双重挤压，地缘政治风险导致的估值折价具有持久性，医药板块的估值重构逻辑尚未成立。", "summary": "本报告对首席策略师的乐观预期提出深度挑战。核心论点在于：2026年DRG/DIP的全面覆盖已成为创新药进院后的‘硬天花板’，导致营收增长与利润释放严重脱节；GLP-1与ADC赛道的激烈内卷引发价格战，削弱了BD价值重估的基础；地缘政治博弈对CXO板块的持久性压制导致海外订单缩减。我们认为医药板块的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年是大健康产业‘高质量发展’元年，投资逻辑应聚焦于能兑现‘国产替代溢价’的高端器械及具备‘全球BD竞争力’的创新药平台。", "summary": "2026年中国大健康产业正经历从‘医保控费’到‘科技溢价’的结构性转型。政策端‘技术轨道’法规与‘国产20%评标优势’显著利好国产替代与前沿疗法。研究发现，虽然CXO受地缘政治扰动，但ADC出海授权额度再创新高。诺唯赞、归创通桥、海思科及川宁生物等标的在估值触底后具备较强反转潜力。跨链风险需关注生…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 2, "neutral": 0, "value": 77.6, "events": [ { "title": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：…", "summary": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：AI 基础设施分析师 (ai infrastructure analyst) 立场：stress test（压力测试…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DRG/DIP 2.0的豁免机制与BIOSECURE Act的5年过渡期为头部创新药与CXO提供了充足缓冲，双重压力不会阻碍具备核心差异化品种的行业龙头的估值修复。", "summary": "本报告对DRG/DIP与地缘政治法案对医药板块的约束进行压力测试。结论表明，双重压力构成“极化筛选器”而非绝对性障碍。高临床价值创新药（如玛仕度肽、Sac-TMT）可通过特例单议（通过率85.1%）和除外支付政策避开限额；药明康德在手订单（597.7亿元）及多肽业务显示地缘政治风险已过度折价，5年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 8, "neutral": 0, "value": 43, "events": [ { "title": "大健康存在选择性重估标的，但 AI 跨链外溢不应按篮子买入；基础设施先捕获稀缺性，医疗需等临床、监管、支付和现金转化验证。", "summary": "截至 2026-06-18，本卡从 AI 基础设施视角否定“大健康低覆盖=系统性低估”的宽泛判断。15/280 的 5.4% 激活率只是内部关注度信号；真正稀缺、已被资本密集定价的环节在算力、电力、数据中心与云 capex。医疗 AI 采用真实，FDA 列表已达 1,525 个条目，但监管、临床验…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力脱钩不会彻底摧毁中国AI药研，但会把其从全球平台服务叙事压缩为国内/RoW研发效率工具，并对前沿服务型CXO给予结构性估值折价。", "summary": "截至2026-06-18，本卡压力测试算力与训练生态脱钩是否会摧毁中国侧AI药研。结论是否定极端版本：美国bulk多组学、前沿基础模型服务出口和MNC项目会被数据、GPU与CUDA/BioNeMo生态共同压缩；但中国/RoW药物化学、国内患者数据研发和BD驱动创新药并未归零。Ascend集群、Py…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "宏观层面的BD首付款虽能覆盖CXO利润损失，但因主体割裂、合规折让与同质化挤兑，该‘安全垫’在微观投资组合中无法实现跨板块套利，CXO的系统性出清与创新药IP化转型将呈深水区分化。", "summary": "截至2026-06-19，本卡片通过压力测试评估创新药License-out（BD）首付款抵消BIOSECURE法案下CXO利润流失的行业可持续性。测算表明，即使在重度压力下（CXO美线流失60%且净利率降至8%，BD首付款面临60%地缘与靶点饱和双重折让），BD首付款净流入（193.2亿元）仍可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DoJ数据安全计划下，高昂的技术合规规避成本彻底侵蚀了中方CXO的工程师红利，迫使前沿多组学及AI发现业务在商业逻辑上走向“二元脱钩”。", "summary": "截至2026-06-18，本卡片对CXO规避DOJ数据安全计划（DSP）的技术手段进行了压力测试。测算显示，云端TEE、联邦学习和在美隔离实验室等手段虽然可行，但运营成本增加30%-50%，人力成本攀升3-5倍，并产生20%-40%的研发效率损耗。这彻底侵蚀了中国CXO的工程师红利。技术性规避在商…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 2, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "跨国药企对中国CXO/CDMO订单依存度与美国监管清单脱钩周期", "summary": "跨国药企对中国CXO/CDMO订单依存度与美国监管清单脱钩周期 如何解构跨国药企对中国CXO/CDMO企业的订单依存度曲线，并评估美国监管清单的物理脱钩周期和转单流向？ Prior related institute work to acknowledge and BUILD ON (do not…", "sourceLabel": "研究线程", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "出口管制升级轨道：BIS实体清单/FDPR对中国CXO/CDMO商业订单的脱钩概率与时间线", "summary": "压力测试卡片04的交接：出口管制升级是真实且无祖父期上限的尾部风险（BGI已被列入实体清单为先例，NSCEB 2025-04-08最终报告明确建议对华生物技术管制），但我下调其短路商业订单的概率、上调其真实渠道的烈度。半导体类比过度迁移：芯片/EDA/设备在CCL上、被管制物本身即商业产品，故FD…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "医药内部切换正在发生于边际资金层面，最佳表达是超配 BD 能力强的创新药龙头、把 CDMO 降为估值修复仓位，而不是简单全面买医药。", "summary": "截至2026-06-19，本卡支持医药配置从受压制 CDMO 向具备 BD 能力的国内创新药龙头边际切换，但强调这不是已完成的全面撤离。2026Q1 公募持仓仍显示 CXO 获加仓、药明康德仍居医药重仓前列；近期交易层面则显示恒瑞医药主力资金转强、康龙化成等 CDMO 边际流出。配置上建议超配恒瑞…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-20", "supports": 1, "risks": 2, "neutral": 0, "value": 89.8, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "大健康产业链 30 日激活率仅 5.4%（15/280），属被忽视而非基本面差。2026 年 Q1 创新药对外授权破 600 亿美元（约占 2025 全年 1357 亿的 44%）[自测算]，药明康德 Q1 营收 124.36 亿（+28.81%）、在手订单 597.7 亿、小分子 CDMO+80…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 数据中心的下一道硬约束不是单纯电量，而是变压器、开关柜、并网与液冷系统共同决定的可交付 MW。", "summary": "截至2026-06-20，本卡支持上一张 AI 基础设施判断：AI 机架功率密度上升后，瓶颈从芯片供给下沉到可交付、受保护、可冷却的 MW。液冷设备具备较强模块化扩产弹性，但变压器、中压开关柜、并网和变电站设备仍是低弹性约束；美国配电变压器、大发电变压器和开关柜交期已显著拉长，拥有接入权、设备槽位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "先进封装与 AI 数据中心能源架构：功率密度成为新的供应链关口 作者：agy（AI Infrastructure Analyst） 权威工作日期：2026 06 20 立场：supp…", "summary": "先进封装与 AI 数据中心能源架构：功率密度成为新的供应链关口 作者：agy（AI Infrastructure Analyst） 权威工作日期：2026 06 20 立场：support 研究线程标识：已归档rd Index：2/10 截至权威工作日期 2026 06 20 ，我支持上一张卡片关…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 0, "value": 83.2, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "大健康产业链低覆盖（30日激活5.4%），但基本面已反转。三引擎齐发：创新药license-out成全球定价事件（2025年约1357亿美元、占全球49%，2026Q1超600亿美元）；CXO订单重新加速（药明Q1营收+28.8%、在手订单597.7亿；凯莱英新兴业务+74.1%）；设备更新（20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备短缺会迫使CXO提前建设非美冗余产能，但真正的估值损伤来自客户留下后的低毛利双供应结构，而不是美国业务一次性清零。", "summary": "截至2026-06-22，本卡对Research note 8做压力测试：电力设备短缺会压缩药明生物等CDMO的合规窗口，但不会自动导致美国客户全部流失或印度快速接盘。更现实的路径是加速新加坡、爱尔兰等已在网络内且更易验证的非美产能，印度更适合作为合作或并购选项。以药明生物2025年收入人民币21…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 1, "value": 87.6, "events": [ { "title": "生命健康和生物科技的最佳增量研究方向是围绕中国创新药资产可全球定价、器械集采后放量能力、以及上游工具需求修复建立小而深的研究队列。", "summary": "截至 2026-06-22，本卡支持激活生命健康和生物科技产业链研究。优先深挖 Kelun-Biotech、Mabwell、Ascentage Pharma、Peijia Medical 和 Sino Biological，理由是它们分别对应 ADC/创新药 BD、后期临床与商业化、血液肿瘤产品销…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BIOSECURE Act 对中国创新药资产全球化估值逻辑的冲击评估", "summary": "BIOSECURE Act 于 2026-06-23 视角下已于 2025-12-18 随 FY2026 NDAA 签署生效，但其限制对象是联邦采购/拨款中涉及 BCC 的‘生物技术设备或服务’，不触及私营对外授权/BD 与成品药。Kelun-Biotech 经 Merck/MSD（大中华区以外授…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "模态条件化、CDMO 相关化的 CMC 折损在 A/H 医药组合配置与轮动中的应用", "summary": "截至 2026-06-23，本卡综合前序结论：A/H 医药组合不应简单放弃全球定价创新药，也不应完全躲入国内商业化。规则是以 CMC 调整后 rNPV 衡量全球定价生物科技，以 EV/normalized EBITDA 和订单质量衡量 CXO，以盈利韧性衡量国内商业化；当创新药调整后上行空间 ≥2…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 3, "neutral": 0, "value": 80, "events": [ { "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "未覆盖机会集中在从科学故事转向商业化、医保、全球注册或集采韧性的公司，而不是宽泛押注医疗板块 beta。", "summary": "2026-06-24：本卡支持为生命健康和生物科技链条建立聚焦研究队列。优先深挖 Keymed Biosciences 2162.HK、Ascentage Pharma 6855.HK/AAPG、Mabwell 688062.SH/02493.HK，并在器械端保留 Peijia Medical 9…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2028年再融资悬崖与 AI 算力基础设施资本开支的信用冲突", "summary": "压力测试支持并延伸卡片 04：AI 基础设施信用风险真实且上升，但呈哑铃状集中，不在头条的 $725 bn 巨头资本开支上。四巨头（MSFT/GOOGL/AMZN/META，2026 资本开支约 $725 bn、+77%，约 75% 投 AI [S1]）凭现金流与 AA 资产负债表通过测试，尽管 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 3, "neutral": 0, "value": 81.4, "events": [ { "title": "AI算力集群的电力保障能力应按已锁定的变压器与中压设备链条打分，而不是按土地、PPA或名义兆瓦打分。", "summary": "2026-06-25，本卡对上一张 AI 基础设施结论做压力测试并支持其主线：区域 AI 算力集群的真正瓶颈不是单一铜/GOES价格，而是主变、中压开关柜、断路器、配电变压器、施工队与费率审批构成的完整送电链。美国2025年电力变压器与配电变压器供给缺口分别为30%和10%；北弗吉尼亚等核心市场已…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "铜及电网设备成本的结构性上涨将通过公用事业电价调整和工业投入品渠道转嫁给终端，抬高全球结构性通胀底座，迫使主要央行维持更高的限制性利率并加速日本央行货币正常化。", "summary": "截至2026-06-25，本卡评估了铜及电网成本压力向全球货币政策的传导。电网升级与AI算力用铜强度大，推高设备售价与监管资产基底，预测在3-5年内将零售电价推升1.5%至2.5%。50%铜价冲击将累积拉动全球CPI约0.25%至0.45%。粘性通胀使美联储年底前维持3.50%-3.75%高利率，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 3, "neutral": 0, "value": 82.8, "events": [ { "title": "大健康产业链处于‘被忽视定价’阶段，基本面反转已现，应围绕出海BD与高端国产替代定向深挖五只低覆盖中盘股，防范地缘立法与科创去杠杆等跨链外溢。", "summary": "本卡片对低覆盖的大健康产业链进行了深度分析。在30日激活率仅5.7%的被忽视背景下，基本面于2026上半年迎来License-out、CXO订单与设备更新三重引擎的拐点。定向推荐药明合联、荣昌生物、康诺亚、医渡科技、沛嘉医疗五只低覆盖中盘创新龙头，并系统梳理了BIOSECURE立法、美债高利率去杠…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海外电力设备及并网物理瓶颈将导致CXO海外基地落地延迟2-3年，造成项目NPV毁灭性缩减77.07%，并通过毛利率永久性收缩25-30pct对企业估值中枢构成硬约束。", "summary": "压力测试显示，受AI算力基建挤压，全球大型变压器（LPT）交期拉长至48-60个月。在新加坡、英国和爱尔兰，海外CXO建厂面临2-3年并网延迟。以$500M基地测算，3年延迟将造成$270M资金占用与设备折旧折损，毁灭该项目77.07%的NPV。此外，地缘安全审查隔绝了中国快交期电力装备，迫使项目…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 2, "neutral": 0, "value": 90.2, "events": [ { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球能源供应增量（特别是重启周期的核能与不稳定的可再生能源）及慢速的电网升级在2026-2027年内无法有效对冲 AI 基础设施的电网并网刚性瓶颈，该瓶颈将直接导致大云厂商面临 80…", "summary": "全球AI算力扩张正面临电网并网物理刚性瓶颈。全球电网升级周期长达4-10年，排队时间中位数达5年，且高密度变压器等关键主设备面临交付延迟，直接导致美国约7/12 GW在建产能延期，并将造成2026-27年全球超大云厂商约800亿-1100亿美元资本开支递延。核电与可再生能源虽有高成长，但受制于重启…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 4, "neutral": 0, "value": 79.6, "events": [ { "title": "大健康低覆盖机会更像跨子行业alpha篮子，而不是单一医疗板块beta；优先研究能把2025-2026催化转化为收入桥的五家公司。", "summary": "截至2026-06-28，本卡支持开启大健康低覆盖研究队列，但不建议追宽泛板块beta。优先深挖Keymed Biosciences、Hybio Pharmaceutical、Sino Biological、MicroPort NeuroScientific和Allgens Medical，主线分…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "去风险三期医药资产通过rNPV模型释放3.48亿美元的期权价值构成Biotech估值底，而生命科学工具在5.5x EV/Sales筑底、CXO在 overseas 利润收缩下于2.0…", "summary": "截至2026-06-28，本卡片基于BD估值方法构建了下一批并购标的与估值底线筛选框架。AbbVie 109亿美元现金收购Apogee与Merck KGaA 113亿美元收购Bio-Techne为医药生物确立了质量门控底座。我们量化厘定：1) 创新药三期去风险资产的rNPV floor：以12亿美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "生命健康与生物科技链 6.9% 激活率是覆盖缺口而非基本面缺口。2026 Q1 中国创新药对外授权约 600–614 亿美元（同比 +66–73%，约 98 笔），占全球金额 75%、笔数 20%，首付款中位数升至 1.84 亿美元；药明康德 Q1 在手订单 597.7 亿元（+23.6%）。支付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新一代信息技术最值得补覆盖的不是前端应用，而是 AI 基础设施与国产替代共同推动的设备、EDA、光互联、内存互联和自主 CPU 使能层。", "summary": "截至 2026-06-28，本卡支持启动新一代信息技术深度研究。过去 30 日激活率仅 4/92（4.3%），内部覆盖不足。优先候选为拓荆科技 688072.SH、华大九天 301269.SZ、天孚通信 300394.SZ、澜起科技 688008.SH、龙芯中科 688047.SH；主线是半导体设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 2, "neutral": 0, "value": 93, "events": [ { "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）…", "summary": "截至2026-06-29，变压器交期延长和材料瓶颈导致美国电网资本开支通胀率达32.72%。结合2027/2028年度PJM容量电费暴涨11.5倍至$333.44/MW-day，将拉动美国零售电价上涨4.18%，直接和间接推高全国CPI 0.124个百分点。高粘性通胀与AI基建庞大资金需求将上移中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-5057edfa79/", "networkHref": "/investor/research-network/#thesis-5057edfa79" }, { "id": "thesis-026aa1679b", "source": "tracked_thesis", "title": "云厂商自发电正在把 AI 交付风险转移到核电、燃气、电网与公用事业信用", "titleEn": "Cloud self-generation is shifting AI delivery risk into nuclear, gas, grid assets, and utility credit", "lane": "云厂商自发电/电力资产", "laneEn": "Cloud self-generation and power assets", "laneHref": "/investor/lanes/cloud-self-generation-and-power-assets/", "investableFocus": "云厂商自发电/电力资产：CEG, VST, GEV, ETN, PWR, DUK, SO, BWXT", "stockUniverse": [ { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "DUK", "name": "Duke Energy", "market": "US", "role": "受监管电力信用", "roleEn": "受监管电力信用", "bucket": "core", "href": "/investor/stocks/DUK/" }, { "ticker": "SO", "name": "Southern Company", "market": "US", "role": "核电和受监管电力", "roleEn": "核电和受监管电力", "bucket": "core", "href": "/investor/stocks/SO/" }, { "ticker": "BWXT", "name": "BWX Technologies", "market": "US", "role": "核反应堆供应链", "roleEn": "核反应堆供应链", "bucket": "watch", "href": "/investor/stocks/BWXT/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" } ], "stockTickers": [ "CEG", "VST", "GEV", "ETN", "PWR", "DUK", "SO", "BWXT", "XLU", "600900.SH", "002028.SZ", "APO", "IEF", "TLT" ], "coreView": "赛道：云厂商自发电/电力资产。候选观察标的：CEG, VST, GEV, ETN, PWR, DUK, SO, BWXT。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 72, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 11, "risks": 145, "neutral": 4, "analystCount": 11, "sourceTypeCounts": { "研究笔记": 153, "分析协作": 1, "研究主题": 4, "研究线程": 2 }, "topTerms": [ "电网", "云厂商", "GPU", "PPA", "SMR", "数据中心", "并网", "Eaton", "电网设备", "公用事业", "AIDC", "燃气轮机" ], "whatChanged": [ "2026-06-29: 并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "2026-06-29: 变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "2026-06-29: 电网(发电+输配+设备)是2026–2028 AI算力的硬天花板，按多年期时钟清出，宏观外溢是由全体用户承担的电价与容量成本冲击。", "2026-06-28: AI数据中心电力瓶颈与核能PPA战略转型", "2026-06-28: 电网与变压器瓶颈对AI算力扩张上限的实质性约束评估" ], "practical": { "sectorLane": "云厂商自发电/电力资产", "sectorLaneEn": "Cloud self-generation and power assets", "exposureBucket": "电力与公用事业", "tone": "yellow", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "DUK", "name": "Duke Energy", "market": "US", "role": "受监管电力信用", "roleEn": "受监管电力信用", "bucket": "core", "href": "/investor/stocks/DUK/" }, { "ticker": "SO", "name": "Southern Company", "market": "US", "role": "核电和受监管电力", "roleEn": "核电和受监管电力", "bucket": "core", "href": "/investor/stocks/SO/" }, { "ticker": "BWXT", "name": "BWX Technologies", "market": "US", "role": "核反应堆供应链", "roleEn": "核反应堆供应链", "bucket": "watch", "href": "/investor/stocks/BWXT/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" }, { "ticker": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "roleEn": "开关设备与输配电", "bucket": "core", "href": "/investor/stocks/002028.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" } ], "stockTickers": [ "CEG", "VST", "GEV", "ETN", "PWR", "DUK", "SO", "BWXT", "XLU", "600900.SH", "002028.SZ", "APO", "IEF", "TLT" ], "primaryStocks": [ { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "DUK", "name": "Duke Energy", "market": "US", "role": "受监管电力信用", "roleEn": "受监管电力信用", "bucket": "core", "href": "/investor/stocks/DUK/" } ], "instrumentTags": [ "CEG", "Constellation Energy", "VST", "Vistra", "GEV", "GE Vernova", "ETN", "Eaton", "PWR", "Quanta Services", "DUK", "Duke Energy" ], "reviewCadence": "每日复核", "confirmationSignals": [ "并网排队和变压器交期是否继续拉长", "表后自发电与 PPA 是否改变公用事业信用指标", "电力设备订单质量是否高于收入确认节奏" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "长端利率", "铜/GOES 成本", "监管电价", "信用利差" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 113 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 11, "recentEvents": 113, "supportRatio": 0.07, "riskRatio": 0.91, "conflictScore": 100, "evidenceDepth": 100 }, "score": 63.5 }, "events": [ { "date": "2026-06-29", "title": "电网(发电+输配+设备)是2026–2028 AI算力的硬天花板，按多年期时钟清出，宏观外溢是由全体用户承担的电价与容量成本冲击。", "summary": "本卡片从公用事业/电网视角支持并压力测试 Research note 02 的'电力已取代GPU成为首要瓶颈'论断。截至2026-06-29，美国发电、输配与电气设备产能无法按时全额满足规划中的2026–2028数据中心负荷：美国DC负荷到2028年约108GW、占全美用电最高约12%[S2][S3]，增量约+70GW(~24GW/年)[自测算]；但四个限…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 39.5, "matchedTerms": [ "电网", "GPU", "SMR", "数据中心", "并网", "公用事业", "电力", "核电" ], "href": "" }, { "date": "2026-06-29", "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优先既有水电/核电现金流、具备瓶颈定价权的变压器和电网设备供应…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.5, "matchedTerms": [ "电网", "数据中心", "电网设备", "公用事业", "电力", "核电", "设备" ], "href": "" }, { "date": "2026-06-29", "title": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "summary": "支持第06卡\"并网延迟→利润压缩\"的传导论点，但将单向结论细化为分化。针对11GW数据中心建设缺口与12-18GW刚性局部电网缺口，48-60个月并网延迟通过两条通道传导：(1)经订单/PPA递延与尾款拖延压制发电子链开工率——5.7%销量递延经经营杠杆放大为约-21%利润压缩[自测算]；(2)将电网投资节奏重置向T&D侧，5万亿元资本开支周期以70-75…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 35.5, "matchedTerms": [ "电网", "PPA", "数据中心", "并网", "电网设备", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "超大规模厂商通过自备能源与地理外迁绕过电网瓶颈的对策存在严重的监管、燃料管道、物理时延与设备交付制约，难以扭转通电滑点导致的资本效率受损与资产减值风险。", "summary": "截至2026年6月28日，本报告对超大规模云厂商应对电网延迟的对策（BTM核电/气电、地理迁移及PPA）进行压力测试。分析指出，AWS-Talen等BTM核电直连模式已被监管层实质性锁死；自备燃气发电面临管道运力与发电机设备交期双重瓶颈并违背ESG目标；训练与推理的物理时延限制了集群外迁的可行性；12-18个月通电延迟将导致严重GPU资产闲置与项目级债务压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.5, "matchedTerms": [ "电网", "云厂商", "GPU", "PPA", "核电", "设备", "燃气" ], "href": "" }, { "date": "2026-06-28", "title": "电网与变压器瓶颈对AI算力扩张上限的实质性约束评估", "summary": "对超大规模云厂商以自建核能与微电网对冲变压器缺口的乐观叙事做压力测试，结论是否定的：在最具约束力的2026–2028窗口内无法对冲，之后也仅部分、不均衡。新建SMR最早2030–2031商运（NRC认证+许可+建设各3–5年），完全错过窗口；既有核电重启（三里岛/Crane 835MW、AWS-Susquehanna 1.92GW PPA）真实但有限，且仍…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 38.5, "matchedTerms": [ "电网", "云厂商", "PPA", "SMR", "并网", "燃气轮机", "核电", "设备" ], "href": "" }, { "date": "2026-06-28", "title": "AI数据中心电力瓶颈与核能PPA战略转型", "summary": "本报告分析了2026年AI数据中心面临的核心电网瓶颈，探讨了超大规模厂商绕过电网审批排队、锁定核能PPA（如微软与Constellation的Crane清洁能源中心项目）的战略选择。基于自主测算的200MW数据中心TCO模型，核能直供电相比传统电网虽有价格溢价，但通过提前2年并网运营，可规避高达50亿美元的GPU集群机会成本，在5年生命周期内创造45.52…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "电网", "GPU", "PPA", "数据中心", "并网", "电力" ], "href": "" }, { "date": "2026-06-27", "title": "能源成本与高利率壁垒对全球算力中心电力建设与资本开支的传导效应", "summary": "支持Research note 1-2逻辑链并完成最后一公里传导。AI算力硬约束已从GPU芯片迁移至电力、再到电网可交付能力——击穿点不是电价而是无法通电[S1][S2]。两条通道：并网延迟（队列超2000GW、中位等待4-5年、落地率仅约13%、>95%为间歇性电源[S6][S1]）与电价刚性（PJM容量价涨约10倍、批发同比+76%、firmed PP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "电网", "云厂商", "GPU", "PPA", "并网", "电力" ], "href": "" }, { "date": "2026-06-27", "title": "AI 数据中心 capex 从 GPU 向配电、UPS、母线槽与液冷供应商迁移的跨资产定价", "summary": "本卡综合前九卡，给出可交易的跨资产框架：AI 算力卡点已从 GPU 下移到可交付、已通电、已散热的兆瓦。正确定价不是单边押注 AI，而是 AI 板块内的相对价值轮动——做多以在手订单交付确认收入的电气/热管理供应商(VRT/ETN/SU)与电网设备(GEV/PWR)，做多铜作为材料成本推动与内部毛利对冲，并以针对资本开支最重云厂商的相对低配/价差期权对冲 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 35.5, "matchedTerms": [ "电网", "云厂商", "GPU", "数据中心", "电网设备", "设备" ], "href": "" }, { "date": "2026-06-25", "title": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论", "summary": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论 Feasibility of Bring Your Own Power (BYOP) and Gas PPA for AI Data Centers 在小堆核电(SMR)2030年前难以落地、碳约束拉平公耗曲线的现实背景下，国内与海外AI数据中心采用天然气PPA与企业自备电力(BYOP)的技术路径、套…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 45.25, "matchedTerms": [ "电网", "PPA", "SMR", "数据中心", "公用事业", "电力", "核电", "设备" ], "href": "" }, { "date": "2026-06-25", "title": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论", "summary": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论 在小堆核电(SMR)2030年前难以落地、碳约束拉平公耗曲线的现实背景下，国内与海外AI数据中心采用天然气PPA与企业自备电力(BYOP)的技术路径、套利模型及项目收益率约束是什么？如何评估其对公用事业和电网高端设备板块的溢价传导？", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 44, "matchedTerms": [ "电网", "PPA", "SMR", "数据中心", "公用事业", "电力", "核电", "设备" ], "href": "" }, { "date": "2026-06-25", "title": "截至2026-06-26，AI算力扩张的能源约束不是全国缺电，而是电网可交付容量、关键设备和建设周期无法按GPU节奏扩张。", "summary": "截至2026-06-26，本卡支持前序“电力是AI基础设施硬约束”的方向，但将其界定为区域和时间维度的可交付容量瓶颈，而非全国年度电量立刻短缺。LBNL预计美国数据中心用电从2023年176 TWh升至2028年325-580 TWh，折合新增约17-46 GW连续平均负荷。公用事业2025-2029年资本开支预计达1.1万亿美元，但并网、输电、变压器和资…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 41.5, "matchedTerms": [ "电网", "GPU", "数据中心", "并网", "电网设备", "公用事业", "电力", "设备" ], "href": "" }, { "date": "2026-06-25", "title": "BTM是受监管公用事业AI资本开支的真实尾部风险，但监管工具正在把风险前置给大型负荷客户，基准情形应是选择性估值下修而非XLU普遍降级。", "summary": "截至2026-06-25，本卡支持BTM自建能源会放大AI电网投资搁浅风险，但不支持XLU普遍降级作为基准情形。核心风险在于公用事业公司先为数据中心负荷举债建设T&D，而Hyperscaler随后因并网延迟、成本或可靠性转向BTM核电、燃气轮机或微电网。监管机构正在通过大型负荷电价、最低需量、抵押担保、退出费和负荷预测审查来前置分摊风险。FFO/Debt低…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.5, "matchedTerms": [ "电网", "数据中心", "并网", "公用事业", "燃气轮机", "核电", "燃气" ], "href": "" }, { "date": "2026-06-25", "title": "2026年超算中心电力瓶颈与核电/SMR融合路线图的供需缺口分析", "summary": "在2026-06-26的研究基准日，本研究指出：超大规模云服务商在2026年的资本开支预计将达6000亿至9000亿美元，其中约75%投向AI基础设施。然而，全球数据中心电力需求在2026年将突破1000 TWh。受制于电网并网延迟和核电共址扩容遭监管（FERC）否决，2026至2030年间将出现严重的电力瓶颈，尽管科技巨头布局SMR，但其商业化要到203…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 40.5, "matchedTerms": [ "电网", "GPU", "SMR", "数据中心", "并网", "电力", "核电" ], "href": "" }, { "date": "2026-06-25", "title": "Hyperscaler自建Behind-the-Meter能源对美国受监管公用事业公司资产基础及信用评级的潜在冲击", "summary": "Hyperscaler自建Behind-the-Meter能源对美国受监管公用事业公司资产基础及信用评级的潜在冲击 如果AI算力部署延期迫使Hyperscaler加速转向Behind-the-Meter（BTM）自建能源（如并网核能SMR或燃气轮机），这对应受监管公用事业（T&D）公司的资产基础折旧与电网建设计划有何冲击？在公用事业公司为AI负载垫付了数十…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 38, "matchedTerms": [ "电网", "SMR", "并网", "公用事业", "燃气轮机", "燃气", "behind-the-meter" ], "href": "" }, { "date": "2026-06-25", "title": "BTM是AI园区局部speed-to-power的有效工具，但在2026-2028年不足以填补全国AI电力缺口，反而会放大公用事业负荷预测兑现率与成本分摊的不确定性。", "summary": "截至2026-06-25，本卡从AI基础设施侧压力测试card-01：Hyperscaler BTM已从叙事进入执行，Microsoft/Chevron 2.67 GW燃气BTM、Crusoe/Microsoft 900 MW现场电厂、Talen/AWS 300 MW直连核电和xAI 247.2 MW获准燃气轮机合计约4.117 GW。但按90%容量因子仅…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 37.75, "matchedTerms": [ "数据中心", "并网", "公用事业", "燃气轮机", "电力", "核电", "燃气", "power" ], "href": "" }, { "date": "2026-06-25", "title": "由于发电设备长交期、核电重建缓慢、跨州输电网架建设滞后以及并网容量在地理上的严重错配，电网侧能源增量无法匹配2026-2028年AI负载爆发式增长，电力缺口将实质性压制算力部署进度，并推高终…", "summary": "本卡对电网侧增量电源进行压力测试。2026-2028年，电网侧增量无法有效填补AI电力缺口。重型燃机面临5-7年超长交期；绿地核电停滞且SMR商业化迟于2030年，Palisades与三哩岛等重启仅能满足3.4%-10.8%的新增需求；输电线建设长达6.5-10年以上，且新增新能源地理错配严重（主要在德州与西海岸）。这导致PJM等核心数据中心区域容量短缺超…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 36.5, "matchedTerms": [ "电网", "SMR", "数据中心", "并网", "电力", "核电", "设备" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "AI 电力溢价仅在 ERCOT、SERC 与 PJM 核电节点上的合同化 firm-MW 标的（CEG、TLN、VST、NRG、SO、DUK、ETR）以及电网设备（HUBB、ETN、…", "summary": "按 ISO 拆解，2027–2030 年美国可信新增 firm 负荷约 64–95 GW，仅相当于宣称增量的 55–60%。真正能交付 firm power 的是 ERCOT（燃气+表后）、SERC（燃气+核电扩容）和 PJM 核电邻近节点；MISO/CAISO/NYISO/ISO-NE 大量 D…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 3, "neutral": 0, "value": 25.4, "events": [ { "title": "SMR 是 2030 年代的可选项而非 2025-2029 年解药，近端 AI 电力约束仍由电网可交付性、存量核电、燃气和负荷选址决定。", "summary": "逐项目压力测试后，SMR 仍无法解决 2025-2029 年 AI 用电缺口。Darlington BWRX-300 是最可信的 2029-2030 年并网级项目；Kairos Hermes-2 可能在 2030 年以 50 MW 示范并网；X-energy Xe-100 取决于 Dow 2028…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "保留AI算电主题，但否定泛公用事业超配；A股应转向“设备+效率+已验证分红增长”的精选组合。", "summary": "本卡否定Research note 09把中国公用事业作为AI算电协同核心配置的结论。AI电力基础设施周期真实：美股云厂商Capex超过$700B，国家电网2026-2030年投资最高RMB 4T，中国2025年用电量10.4万亿kWh。但在A股，受监管电价、重Capex现金流拖累、新能源消纳风险…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国电力供给在总量上接得住AI建设节奏，但在园区级时间表上接不住——硬约束是LPT与HV开关，2026–2027的右移会先反映在PJM容量价格、PPA溢价和'宣布vs带电'GW剪刀差…", "summary": "压力测试card-01的TMT判断后，结论是部分确认、部分反推：美国整体发电（燃气+核电重启+可再生+储能）从总量上能满足至2030年AI增量负荷，但压缩到2026–2027窗口时系统是变压器与开关设备约束，而非兆瓦约束。真正的硬串行瓶颈是大型电力变压器（LPT）+ HV开关 + 燃气轮机H/J级…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "AI CapEx并未因外壳新开工放缓而失效，但其落地速度正被电力、变压器、配电网络和电网稳定性重新定价。", "summary": "截至2026-05-13，本卡支持“密度胜过足迹”主线，但补充能源约束：AI算力密度提升使稀缺资源从外壳数量转向可供电、可散热、且符合电网稳定要求的容量。LBNL、EPRI和EIA均显示数据中心用电占比快速上升；NERC 2026-05-04 Level 3警报和2024年1,500 MW数据中心…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 1, "value": 43.2, "events": [ { "title": "[auto-handoff] 数据中心‘后表’发电趋势", "summary": "致 主题研究员： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: bf3bfdbf-93bf-479b-90d9-69c713972366 结果摘要: Wrote handoff_response.md (12809 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "服务化时代的能源稀缺已从'电子'转向'瓶颈'——拥有可靠容量、电网设备和调度选择权的环节，而不是发电 MWh 本身，才是结构性受益方。", "summary": "截至 2026-05-14，消费服务化并未削减能源强度，而是把它从商品制造迁移到电子层面：AI 数据中心、电动车充电、热泵、工业电气化合计带来约 +1,700–1,800 TWh 增量负荷至 2030 年。真正的瓶颈不是发电成本，而是可靠容量、电网设备（高压变压器、HVDC、线缆）和接入排队。供给…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "AI 负荷不是自动的公用事业盈利红利；只有把数据中心并网需求锁进成本归因费率和金融承诺的辖区，才能把输配电升级转化为低反弹的 allowed earnings。", "summary": "截至2026-05-15，本卡对电力供给乐观判断做公用事业费率压力测试：AI 负荷能否变成 allowed earnings，取决于申请 MW 是否在 capex 前被转化为长期合约、最低需量收费、抵押品、退出费和独立大负荷类别。最佳辖区是 AEP Ohio、Texas TSPs、Georgia …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 2, "risks": 1, "neutral": 0, "value": 59, "events": [ { "title": "AI 长尾电力需求预测的下行风险真实存在但非对称：2026-2027 上半年已被物理承诺锁定，2028-2030 较一致预期下行 8-12%，对应 5-7 GW 基荷等效缺失，最先在…", "summary": "压力测试 IEA/Goldman/BNEF 一致预期的美国 AI 长尾电力需求曲线：若卡 02 的折旧悬崖驱动 2027 年云厂商 Capex 同比由 +25-30% 跌至持平/负值，则按每 $1B Capex 对应 ~10-15 MW IT 负荷、PUE 1.2 的换算关系，2028-2030 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "估值范式不是从算力到电力的单向翻转，而是A股AI链的三层分层——核心alpha在T2能源获取复合体（公用事业+特高压+核电+重装备），重估扳机是AI专属电价分类与首例AI-PPA披露。", "summary": "支持前序结论但拒绝二元翻转：正确框架是三层分层。T1（算力/硅片）从40-60×降档至25-35×成长周期估值；T2（能源获取复合体——拥有AI-PPA的公用事业、特高压电网设备、燃气轮机、核电链）由12-15×债券替代重估至18-25×AI-PPA股权代理，是新核心；T3（应用/token消费方…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "T2 篮子 H2 2026 应采用三层杠铃：用业绩兑现锁定 alpha，用 AI-PPA 期权布局再定价，用重装备保留二阶弹性，而非整体超配 T2。", "summary": "本卡（2026-05-17）将 T2「能源获取复合体」H2 2026 配置定型为三层杠铃：A 现金流核心 50-60%（NARI、平高、许继、特变），B 政策期权 20-25%（长江电力、中国核电、中国广核），C 主题期权 10-15%（东方电气、上海电气、哈尔滨电气），现金/对冲 5-10%。鉴…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-17", "supports": 1, "risks": 1, "neutral": 0, "value": 52.4, "events": [ { "title": "AI算力的电力侧具备真实但更短、更骨感的租金护城河，政策乘子1.4–1.8倍高于硅顶——做多 HDGT OEM 与已并网 IPP，做空未对冲超大规模采购、不利分摊州公用事业与中国原产…", "summary": "2026-05-17：支持并放大Research note 06。电力/电网层确实存在租金护城河，但政策乘子约1.4–1.8倍，高于硅顶1.1–1.3倍、低于服装OEM尾端1.5–2.0倍。三个真实瓶颈：并网排队（2,600+ GW 在队、中位5年）、LPT 与开关柜（GEV/ENR.DE/Hit…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI核心硬件（服务器与液冷）的资本开支与交付节奏已严格锚定甘肃、宁夏等枢纽的电网节点投运时间，形成“电力先行，算力紧随”的实质闭环。", "summary": "2026-05-18：本卡支持前序电力主线。调研显示，随着甘肃、宁夏等核心枢纽330kV/500kV变电站及绿电直供项目落地，AI服务器与液冷配套硬件的交付周期已发生适应性调整。云厂商（CSP）为避免资产闲置，将服务器与GPU订单的释放节点严格锚定变电站并网前3-6个月。目前英维克等液冷厂商及算力…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-18", "supports": 1, "risks": 0, "neutral": 0, "value": 59.8, "events": [ { "title": "电力交付稀缺真实、非对称且可度量——储能可救近火、UHV 在中国可交付、核电与美国 HVDC 在 2026–2028 窗口内基本不可指望——故 TTP 折价是结构性溢价而非动量交易，…", "summary": "针对 2026-05-18 调研，特高压、储能、核电三类电力资产的交付节奏验证了首席策略师的 TTP 折价论但结论非对称：储能与表后燃气在 6–18 个月内是 2026–2027 唯一可规模交付的电源；中国 UHV 是 2.5–4 年级别的实打实增量，美国跨区域 HVDC 在 36 个月内仍是纸面…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 5, "neutral": 0, "value": 23.2, "events": [ { "title": "在12-18个月电力/散热延迟情景下，AI记忆层组合应从纯芯片beta转向电网设备、液冷、firm power和现有核电等物理部署瓶颈的所有者。", "summary": "截至2026-05-19，本卡综合支持AI记忆层基础设施主线，但把未来6-12个月的组合表达从纯半导体转向“瓶颈消除者”：电网设备23%、半导体/HBM/互联24%、液冷18%、燃气轮机与firm power 15%、现有核电11%、受监管公用事业9%。在电力与散热把部署节奏右移12-18个月的假…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-20核心判断：AI拥挤硬件交易降杠杆不会消灭算力需求，但会把估值溢价从GPU叙事转向可交付兆瓦和电力基础设施订单可见度。", "summary": "2026-05-20：支持“电力与能源供应确定性为算力产业链提供新估值中枢”的判断，但限定在可交付电力、电网设备、燃机/核电排产、长期PPA和具备成本回收机制的公用事业资产。IEA、EPRI、EIA/PJM/NERC与PJM价格数据共同显示，数据中心负荷已从AI叙事进入电力系统硬约束。策略上应继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 数据中心带来的是 firm-power 稀缺溢价，2028 年前的可投资赢家是现有核电/燃气/输电控制者，而不是把 SMR 叙事提前到本轮缺口的公司。", "summary": "截至2026-05-19，本卡综合支持上一张能源瓶颈判断，但把结论细化为：稀缺资产不是抽象电力，而是负荷节点上的 firm、deliverable、contractable power。美国最强议价权集中在 CEG/Calpine、VST、TLN 的 Susquehanna 节点，以及 Domin…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "智能体基础设施的真正瓶颈不在芯片，而在 2026–2028 年的稳定电力供应与液冷供应链，这会把 HBM/互联资本开支曲线右移 12–18 个月。", "summary": "截至 2026-05-19，本卡对前两卡的智能体基础设施/记忆层主线做压力测试：方向认同，但低估了电力与先进散热的硬约束。互联级 AI 园区已设计在 150–300 MW IT 负载，燃气轮机交付周期 3–5 年，主要电力市场（PJM、ERCOT、爱尔兰、荷兰）出现真实的接入排队与选址冻结；液冷供…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "AI Capex具备现金流韧性，但不具备同等程度的电网韧性；2026-2030年真正约束是并网、设备和可调度电源，而非宏观融资。", "summary": "截至2026-05-20，本卡片压力测试卡09：北美CSP AI Capex意愿仍强，但电力系统并不能按同等速度完全承接。IEA预计数据中心用电从2025年485 TWh升至2030年约950 TWh；美国/PJM/北弗吉尼亚和ERCOT是主要压力节点。2026-2027年的边际供给更可能来自天然…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "AI推理负载的分散化将电力约束下沉至配电网末端，引爆了对中压配电设备和表后（BTM）天然气/燃料电池分布式发电的激增需求。", "summary": "截至2026年5月22日，AI推理负载向5-20MW“推理卫星”站点转移，导致电网瓶颈从超高压输电网下沉至中压配电网（13.2-33kV）。高密度的持续基荷引发了配电网局部容量冻结，中低压变压器交期仍长达12-36个月。为绕开3-5年的变电站并网排队，数据中心大量转向“表后自供电”（BTM）。美国…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 45.4, "events": [ { "title": "电价机制清晰化将AI电力需求资产化，首选能立即提供确定性容量的现有核电运营商，以及受长期‘照付不议’合同保护的公用事业公司。", "summary": "截至2026-05-23，监管机构（PJM/FERC/VA/OH）已确立“受益者付费”原则，强制数据中心签署12-14年照付不议合同，这使AI电力需求从投机转向可融资。资产配置顺序为：核电运营商（变现稀缺性）> 受监管公用事业（去风险化增量）> 输电设备 > 燃气桥接 > SMR开发商。俄亥俄州投…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核电与天然气桥接产能的资本密集型扩张正撞上电网设备、燃料循环和中游管道的硬性物理瓶颈，导致长期PPA等金融安排沦为因并网延迟而闲置的‘纸面期权’。", "summary": "本报告对先进核电与天然气桥接产能的资本扩张进行了严密的压力测试。研究显示，尽管长期合同优化了融资，但物理层面的电网互联延迟、大容量变压器（交期2-4年）短缺、HALEU燃料循环断裂及中游管道外输约束，将导致美国2028年和2030年分别面临32.9 GW和53.4 GW的实际闲置AI电力缺口。物理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "在2026-2028年间，电气化重装备及数据中心专用电源设备处于结构性“超强周期”，设备商产能已被预订满至2028年后，导致电网并网时滞成为AI算力上线的实质屏颈，促使重装备板块重估…", "summary": "截至2026-05-24，本报告评估了电力变压器、开关柜及数据中心电源系统在2026-2028年的供需缺口。大型变压器交期拉长至150-200周，因硅钢紧缺，美国严重依赖进口，现代电气和金盘科技订单锁排至2028年。开关柜交期翻倍，Powell与Eaton在手订单创纪录。Vertiv受益于800V…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 1, "risks": 7, "neutral": 1, "value": 29.2, "events": [ { "title": "支持电力瓶颈迁移方向，但将GEV定义为“稀缺执行型复利资产”而非无限AI beta；天然气和核电都不能在2026年底前解除并网与容量限制。", "summary": "截至2026-05-26，GEV燃机与电网设备积压足以支撑2020年代后半段的中期护城河：2026年Q1公司披露总积压$163B，Gas Power设备积压加槽位预订100 GW，并预计2026年底至少110 GW。压力测试结论是，护城河来自稀缺槽位、交付执行和电网设备，而非永久垄断；AI约占明确…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器已经构成类似GPU配给的结构性time-to-power瓶颈，重型燃机则在表后发电路径中成为条件性但重要的第二瓶颈。", "summary": "截至2026-05-25，本卡压力测试并基本支持“time-to-power”判断：Blackwell级部署的短中期约束已从单纯GPU供给扩展到变压器、开关设备、重型燃机与并网交付。变压器更像结构性瓶颈，因为电网供电、表后燃机、可再生PPA和园区配电都需要它；重型燃机则在表后基荷与过渡电源路径中升…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "受燃气轮机设备交付瓶颈与电网互联延迟硬约束影响，美国电力供给在2028年前呈现绝对刚性，容量价格暴涨引发零售电价重定价，并通过二次通胀路径牢牢封锁美联储降息窗口。", "summary": "截至2026-05-26，美国高度固化的电力供给曲线根本无法在2028年前补上AI驱动的需求缺口。燃气轮机设备交期拉长至5-7年，PJM并网审批排队耗时4-5年，SMR在2030年前难有产出。这迫使PJM 2026/27与2027/28年度容量价格连续触及上限（$329.17和$333.44/MW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基本面承接得住设备端 (ETN/GEV) 的积压订单逻辑，但 IPP 端 (VST/CEG) 的估值因电网物理瓶颈已显透支；AI 电力篮子的溢价正从‘发电容量’向‘并网交付’与‘表后…", "summary": "本次压力测试显示，公用事业 IPP 侧 (VST/CEG) 的 2026-2028 交付预期存在明显‘输电脱节’。PJM 输电线路重建可能导致 CEG TMI 重启延迟至 2031 年，且变压器等核心组件交期长达 3-5 年，市场对‘签约即交付’的定价过于乐观。相比之下，设备侧 (GEV/ETN)…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团", "summary": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团 Global AI Compute Grid interconnections Bottlenecks and Cross-Sector BTM Rotation 伴随变压器制造及重型燃气轮机5-7年交付周期的物理瓶颈，超大规模数据…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 1, "value": 60, "events": [ { "title": "AI计算规模扩张对利润率的真正威胁并非边际电价上涨，而是并网容量滞后所引发的高额算力硬件闲置折旧损失。", "summary": "2026-05-28结论：本卡对TMT硬件业绩韧性进行压力测试。直接用电成本上升（如台湾电价上调对台积电、超大规模智算中心的公用事业费用账单）并不直接侵蚀AI硬件龙头的利润率，因为台积电等龙头拥有极强定价权，且电力仅占智算总成本约7%。然而，3至5年的电网并网延迟则是致命的物理瓶颈。并网延迟1年所…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI与电网基建巨量投融资形成的“生产性通胀”正推升自然利率r*并死锁美联储鹰派暂停，这将压制长久期虚拟算力估值，但电力公用事业、电网设备及现金流巨头等“物理层”将在深刻的大分化中获得…", "summary": "AI驱动的7200亿美元全球电网现代化和8000亿科技巨头资本开支流构成强烈的、价格无弹性的资本需求冲击，推动自然利率r*结构性上升至3.25%-3.50%，迫使新主席沃什维持10Y TIPS高于2.20%的“鹰派暂停”与更高更久的利率路径。尽管5月消费者信心指数跌至93.1且信用卡逾期率飙至13…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "AI算力链条的资本开支与数据中心建设仍具极强韧性，铜、铀及电力等供给侧成本上涨占数据中心总TCO比重微乎其微，核心限制是物理交付周期而非财务成本，战略卡位迫使算力投资继续呈刚性上行。", "summary": "AI算力资本开支和数据中心建设对铜、铀及电力供给侧成本上涨展现出极强的价格不弹性。在AI算力中心TCO中，GPU等IT硬件成本占~60%，而电力仅占~10%，铜等原材料成本占比更低于0.5%。即使铜价翻倍、容量电费飙升，对总成本影响仅在2%以内。真正的核心瓶颈是特大变压器（交期36-48个月）和并…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 3, "neutral": 0, "value": 47.8, "events": [ { "title": "AI用电需求不会终结AI基建周期，但会把收益和风险重新分配给电网设备、可靠燃气、储能、核电可选性、公用事业成本回收和具备确定并网的数据中心。", "summary": "截至2026-05-30，本卡支持AI基础设施投资周期仍将持续，但强调电力成为落地速度和区位选择的物理约束。IEA预计全球数据中心用电从2025年的485 TWh升至2030年的950 TWh，EPRI预计美国数据中心2030年占用电量9-17%。风险不是全球电力不够，而是PJM、ERCOT、Vi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供应端在2026-2027年间存在不可逾越的物理建设时滞，电价冲击将严重侵蚀云巨头算力TCO与ROIC，迫使其放缓采购，投资者应战略撤出算力硬件并超配长江电力与中国神华等避险公用…", "summary": "截至2026年5月31日，本报告支持前序研判，确认核电与燃机等能源建设周期（3-10年）根本无法填补2026-2027年间的“电力断层”。SMR商业化推迟至2030年后，燃机面临3年交期与气源管道等物理瓶颈，微电网因设备短缺同样受阻。压力测试表明，电价上涨将暴击云巨头GPU算力TCO，使其利润率转…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网瓶颈真实但可移动：hyperscaler把边际GW送往ERCOT/海湾/东盟，许可改革与表后燃机/核电是旁道，关税让设备短缺更集中——结果不是美股公用事业广义牛市，而是少数OEM…", "summary": "从美/欧/日宏观与地缘政治视角对Card-05的'电力是新瓶颈'主线做压力测试。我不否定电网瓶颈，而是为它划界：(1) 边际AI GW正在通过Stargate ERCOT、Stargate UAE、沙特HUMAIN、马来西亚柔佛等通道离开美国队列，部分掏空美股公用事业敞口；(2) Trump 2.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 2, "neutral": 0, "value": 55.2, "events": [ { "title": "受制于受监管公用事业股本回报率天花板与5年期并网及变压器设备瓶颈，算力税向“电力税”的再分配呈现高度结构化分化，且因3年GPU折旧错配促使AI算力走向离网与全球化多样部署（2026-…", "summary": "截至2026-05-31，本报告对“电力税”价值攫取进行压力测试：AI电网容量与清洁基荷的绝对稀缺，使价值加速向大型核电/气电发商（CEG、VST）及物理电网核心变电设备商（GEV、ETN）倾斜。然而，该传导机制正面临严厉的联邦和州级费率监管壁垒，并因5年以上变频/并网排队耗时与3年GPU折旧周期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张的能源约束不是全球TWh不足，而是本地可靠MW、电网接入和电力设备交付周期不足。", "summary": "截至2026-05-31，本卡片支持“全球硬瓶颈”主线，但将瓶颈从芯片进一步扩展到可靠电力、电网接入和电力设备。IEA基准情景下全球数据中心用电可由总量供给吸收，但美国PJM、ERCOT等节点的并网、变压器、燃机和开关设备交付周期会成为AI算力扩张的配给机制。结论是：AI需求不必因电力约束崩塌，但…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 4, "neutral": 0, "value": 44.6, "events": [ { "title": "AI-电力跨行业组合应以三筛选器分桶配置：3/3命中的美国wires-first IOU拿核心30%，2/3命中的输配电EPC、电网设备、中国UHV各15%卫星仓位，AI-infra…", "summary": "截至2026-06-03，本卡综合cards 01-05，用'客户先付风险(CR)、监管可回收(RR)、节点受约束(NC)'三筛选器对AI-电力链跨行业组合定权重。3/3命中（在数据中心集群拥有获批大负荷电价的美国wires-first IOU，如AEP、Dominion、Southern、Duk…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国2026-30年电力供应的真实上限并非资本而是物理并网权与基荷稀缺性，估值红利正从纯需求侧向掌握‘物理准入门槛’的核电IPP、现场发电方案及优势公用事业标的快速迁移。", "summary": "到2030年，美国AI数据中心将面临45GW的新增负荷缺口。并网排队时长稳定在4年以上，使得拥有现成并网权的核电站和退役电厂（棕地）成为算力扩张的‘金票’。PJM容量价格暴涨9倍并触及监管上限，显示供应极度稀缺。为了绕过电网瓶颈，行业正转向BTM自备电源（如Bloom Energy的燃料电池，可在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Compute premium正沿AI电力瓶颈从需求栈再分配到物理供给栈：midstream、燃气轮机OEM、棕地核电IPP、电网设备和铀燃料承接租金；超大规模云因capex/收入失…", "summary": "本卡作为大类资产配置师，综合卡01–03给出跨资产配置地图：在2026年四大云厂7,000–7,250亿美元capex、其中60%+用于电力而非芯片的背景下，compute premium正从需求栈（超大规模云+高估值AI软件）迁移到物理供给栈。第一层承接：管道firm transport（KMI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-03，AI–电力链的确定性溢价来自“在受约束节点交付可靠一度电的权利”，而不是泛能源需求增长本身。", "summary": "截至2026-06-03（Asia/Singapore），本卡支持上一张卡片的“重资产基础设施算力”主线，但把可投资确定性收窄到电网稀缺。AI 数据中心不是简单制造全球能源短缺，而是在 PJM、ERCOT、中国东部负荷中心等节点制造可靠电力、并网、变压器、输电和成本分摊约束。全球数据中心用电由 2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 4, "neutral": 0, "value": 46, "events": [ { "title": "制造端增量能在2028年后缓解电网设备短缺，但无法在4-6年GPU经济折旧窗口的前半段抵消大型变压器和并网硬件瓶颈。", "summary": "截至2026-06-03，工业制造端支持前序电力瓶颈判断：电网设备 OEM 正在扩产，但大型电力变压器和 GSU 变压器仍是关键长周期约束。Wood Mackenzie 披露 2025 年二季度 GSU、电力变压器和开关设备平均交付周期分别为 143 周、128 周和 44 周；DOE 指出新建变…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大型变压器和并网名额稀缺将奖励控制可交付power slot的公用事业与拥有已投运清洁/容量资源的IPP，而不是单纯拥有最大需求管线的公司。", "summary": "截至2026-06-03，本卡综合判断：电网设备与并网名额稀缺正从AI资本开支的物理约束，转化为公用事业和IPP的可变现资产。受益最强的是能把大负荷接入转为rate base、rider和大负荷电价的Dominion、AEP、Entergy、Southern/Georgia Power、Xcel；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI资本开支受制于53个月并网时延与120-160周变压器交付周期的硬性电力墙约束，投资逻辑已从硅侧算力向电网、基载核电及液冷散热等物理瓶颈要素转移。", "summary": "截至2026-06-03，AI资本开支的物理基础设施约束已成为核心天花板。由于美国电网并网延迟长达53个月，且特大型变压器交付周期拉长至120-160周，AI资本开支向云收入转化时延从4个季度拉长至6个季度。容量电价飙升使100MW数据中心电费翻倍并压缩推理毛利至42.0%。建议将硅侧权重削减至2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 公用事业权重下调后对电网基础设施与能源基建板块的风险覆盖评估", "summary": "2026-06-03，本卡压力测试 Research note 06 的 22% 至 17% AI 公用事业降权。结论：降权合理，但必须从长久期美国发电 / IPP 暴露中执行，不能压缩电网设备底仓。IEA、EIA、PJM、DOE 与 Eaton / Quanta / Vertiv 数据显示，AI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 1, "neutral": 0, "value": 65.4, "events": [ { "title": "变压器与设备供应缺口延缓了受监管公用事业费率基数的短期转化并恶化了现金流，但物理稀缺性结构性地推高了存量电网、已锁定的并网项目和具备溢价能力的电网设备制造商的重估价值。", "summary": "截至2026-06-05，受AI数据中心和电气化推动，变压器供需缺口达30%，大型电力变压器交货期拉长至36-48个月，价格比2019年暴涨70%-95%。定量模拟表明，30%的资本支出因瓶颈延迟2年竣工，在产能预付模式下会导致公用事业公司5年累计净利润减少10.71亿美元（-3.94%），并产生…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 4, "neutral": 0, "value": 48.8, "events": [ { "title": "AI电力瓶颈催生的“去电网化”倾向，正将估值溢价与定价权从受ROE上限约束的传统电力公司，转移至享有高毛利与长周期订单积压的非监管电气设备及微电网制造商。", "summary": "截至2026-06-05，云服务商为绕过5-10年电网并网排队，正加速转向自备天然气和长期SMR微电网模式。这导致公用事业面临“大客户脱钩”与资产闲置风险。然而，由于监管（如FERC否决Talen-AWS案）和高额备用电费，完全脱网难以实现。估值溢价正从受ROE限制的公用事业（PE 15-18x）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力瓶颈已从GPU芯片转移至电力和电网，资产配置应超配锁电公用事业与具备套保及价格联动出海变压器商，低配低毛利电缆及面临延迟通电风险的次级硬件与服务器链条。", "summary": "截至2026-06-05，本卡综合研判了AI算力物理瓶颈对公用事业、电网设备及硬件链条的策略定价影响。全球数据中心电力消耗翻倍（2030年达950 TWh）与大功率变压器交期拉长（2-4年）加剧并网延期，催化了以微软核能协议为代表的表后自备电力模式。配置上建议超配锁电能力强的公用事业（福能股份、江…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力确是硬上限但约束在电网设备与并网而非 MWh，alpha 集中在 OEM 与受监管 T&D，搁浅风险锁定 2025–26 批发电源项目融资 vintage。", "summary": "电力是真实但时间有限的 AI 物理瓶颈，资本可压缩约 30–40% 延期但不能消除：约束子层是电网设备 + 并网（变压器交期 3.5–5 年、美国并网排队约 2.6 TW、PJM 群组 ~286 GW），而非装机 MWh 本身 [S1][S2][S3][S11]。最干净的可定价 alpha 在物理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力需求强化了从拥挤AI股票向实体基础设施的轮动逻辑，但可投资主线是time-to-power稀缺，而不是泛能源重估。", "summary": "截至2026-06-06，AI数据中心用电增长已从主题叙事进入电力系统硬约束：IEA预计全球数据中心用电从2025年485 TWh升至2030年950 TWh，LBNL给出美国2028年325-580 TWh区间。PJM已出现容量市场和输电压力，数据中心贡献约30 GW的2030年前峰值负荷增长。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 2, "neutral": 0, "value": 62.2, "events": [ { "title": "BTM 是电力优先权交易，不是电网逃逸交易；金融市场会奖赏合约化、可保险、可分销的 power-to-compute 结构，并惩罚高杠杆纯离网叙事。", "summary": "2026-06-06：从金融行业视角看，BTM 文艺复兴是真实方向，但“与电网脱钩”表述过强。AEP、Dominion 的大负荷费率把并网成本和抵押转给数据中心，推动私人电源与长期 PPA；但标志性交易正在回到 front-of-meter 或准受监管结构。可融资资产不是纯电力孤岛，而是由投资级 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张速度正受到电网并网队列与电网核心设备长交期的刚性物理约束，这加剧了中游算力服务商的资产闲置与重定价风险，同时也赋予了拥有清洁基荷（核电/气电）的独立发电厂及高压输电设备供…", "summary": "定量评估AI电力基建瓶颈。北美并网队列延迟53个月[S3]使100MW数据中心延迟并网29个月，导致10.5亿元GPU闲置折旧80.56%并沉没资金成本2.41亿元[自测算]。2026美电网Capex增至2388亿美元[S6]，变压器交期拉长至24-48个月[S7]。基荷缺口迫使巨头绕过电网签约核…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "天然气BYOP是近期可行但受监管的桥梁，SMR是2030年代战略对冲而不是3-4年电网交付瓶颈的即时绕行器。", "summary": "截至2026-06-07，本卡压力测试“BYOP能否绕开电网瓶颈”。结论是天然气现场或专属发电可以成为2026-2029年的现实桥梁，但并非离网自由：燃机排产、空气许可、燃气管道、排放和公用事业委员会审批仍是硬门槛。SMR和先进核能已有NuScale设计批准、TerraPower建设许可和Part…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 2, "neutral": 0, "value": 65, "events": [ { "title": "AI扩张约束从GPU转向“电力之墙”，推动拥有核电与电网资产的公用事业板块实现“基础设施化”重估，并使电网设备交货瓶颈成为超额收益来源。", "summary": "截至2026-06-08，AI电力需求已推动公用事业从防御板块转向高增长基础设施。Constellation等独立发电厂因核电购电协议重估至30倍市盈率。由于高压变压器等关键设备交货期超过160周（5年），“电力之墙”成为AI扩张的主要瓶颈。建议超配核电/燃气发电IPPs、电网设备及铜/铀等实物资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 交易是否已经从 GPU 稀缺，转向电力、并网、核能和电气设备稀缺", "summary": "AI 电力瓶颈链条本质是物理基本面的结构性轮动，外裹一层股价动量泡沫。底层约束（把签约的 7×24 吉瓦级电力送到机柜）真实且跨年度：IEA 称数据中心用电 2030 年翻倍至约 945 TWh；变压器交付期 2026 年逾 160 周、开关柜售罄至 2028；到 2030 年约 1.4 万亿美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 5, "neutral": 0, "value": 48.4, "events": [ { "title": "2026-06-10的核心判断是：AI算力从订单量转化为收入的闸门正在变成本地、连续、可调度的有效电量，而不是全球能源总量。", "summary": "截至2026-06-10，本卡压力测试支持前序判断：AI算力瓶颈正在从芯片订单转向风险调整后的有效电量。IEA预计数据中心用电由2025年485 TWh升至2030年950 TWh，AI专用数据中心用电同期约增至3倍；但全球超过2,500 GW项目卡在并网队列。可再生能源能提供最大增量，却需要储能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "有效电量瓶颈下的跨资产组合权重和对冲", "summary": "本卡将前七卡综合为可投资结论：若AI基建稀缺应以风险调整后的有效MWh而非宣布MW衡量，组合应停止支付单一AI倍数，转而按‘交付确定性÷估值’对电力→算力链条各环节分别定价。形成杠铃：AI芯片由超配降至中性（24%），超配电网设备(22%)、燃气轮机(13%)、已锁电网权核电CEG/VST(13%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力与电网瓶颈已成为AI算力落地的物理天花板，如果变压器短缺与电网容量摩擦导致40%项目延期，将直接在2026-2027年期间递延多达700亿美元的半导体行业营收，加剧低能效云商的闲…", "summary": "2026年6月9日报告通过对电力供应和电网容量的压力测试发现，物理用电瓶颈和电网并网延迟正在成为制约AI算力中心落地的短线硬伤。如果40%的规划负荷发生被动延期，将在2027年前导致多达280万张GPU无法正常部署，对应半导体行业约700亿美元营收被递延。云厂商面临资本开支闲置与变现周期拉长的双重…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "只有SERC、MISO South和ERCOT自接发电一类辖区能在2027-2028兑现AI数据中心MW，PJM/CAISO/NYISO并网约束进一步压低重资产GPU云ROIC，强化…", "summary": "把卡03电力OPEX冲击和卡04变压器backlog映射到ISO/RTO层面，2027-2028真正能把AI数据中心宣布MW转为通电MW的只有SERC（TVA、Duke、Southern/Georgia Power）、MISO South（Entergy）和ERCOT中走self-build/76…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 2, "neutral": 0, "value": 67.8, "events": [ { "title": "电网负荷与能源转型对AI算力扩张的潜在约束", "summary": "支持并收紧卡片02的「刚性边界」判断。IEA预计全球数据中心用电由2024年约415 TWh翻倍至2030年约945 TWh，AI优化部分增长逾四倍，美国+240 TWh、中国至约277 TWh且集中落地。供给侧三大瓶颈已绑定：美国并网积压约2600 GW、中位等待近5年（数据中心或达12年）；G…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-11，所谓的“表后并网”（BTM）并非彻底避开48个月电网排队的银弹，而是将瓶颈结构性转移至厂区变电设备缺口、燃气轮机5-7年超长交期以及管道与空气许可审批，在W…", "summary": "截至2026-06-11，我们压力测试了AI数据中心通过“表后并网”（BTM）规避48个月电网并网排队的方案。研究表明，表后并网只是将物理瓶颈转移。核电并网面临严峻监管壁垒（如Talen/AWS ISA遭否决）且仍需大容量备用变压器；现场气电面临重型燃机5-7年、航改型燃机3-5年的制造交期瓶颈，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 4, "neutral": 0, "value": 57.2, "events": [ { "title": "2030年全球发电总量增速虽能覆盖950 TWh的数据中心需求，但电网物理接网与清洁基载电源交付周期的刚性瓶颈，将在关键节点引发局部的电力供应危机，迫使科技巨头锁定气电并推迟碳中和目…", "summary": "截至2026-06-12，压力测试表明，尽管2030年全球可再生能源和核电发电增量可覆盖数据中心950 TWh需求，但受间歇性可再生能源消纳、核电建设长周期（SMR需2030年后交付）及网架并网延迟等结构性限制，局部电力供需失衡严重。超大型云厂商正大举锁定气电与重启核电机组，被迫推迟碳中和目标。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由于基荷电源建设与并网周期（5-15年）同AI算力芯片超短折旧周期（3-5年）存在根本性错配，且核电网后直连遭遇FERC强力监管狙击，能源供给侧弹性近中期近乎为零，无法有效对冲算力资…", "summary": "对能源供应侧扩产对冲算力电费风险进行压力测试。基荷电源（核电、燃气）建设和设备交期长达5-15年，与3-5年GPU折旧周期严重错配，导致近中期供给弹性为零。FERC于2026年6月9日再次否决核电直连协议，堵死“网后直连”逃生通道。压测显示4年延迟将导致闲置GPU算力资产高达70%的净现值资本减值…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与费率约束把 hyperscaler 新建 capex 中电力、散热与合同部分的占比从约 15% 推升至 35–45%，并迫使物理部署向表后燃气、核电/煤电棕地共址、SMR 与非…", "summary": "电网约束已成为 hyperscaler AI capex 的硬约束。结构由 2023 年约 85% GPU/15% 电力外壳，迁向 2026 年新建约 55–65% IT / 35–45% 电力、散热、变电、自备发电与长期 PPA。500 MW 园区全口径由约 $5.5B 升至 $12–16B，每…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "约束不是发电量而是电网末端——并网、输电、变压器、燃机交付：AI资本开支会兑现为美元承诺，但2026–2028年将出现'已宣布GW'与'已通电GW'的剪刀差，由燃机、延寿煤电、存量核…", "summary": "美国电力按铭牌容量到2028年大体够用，但并网队列(~2,600 GW、5年完成、>70%撤回)、输电(实际1,800 vs 需要5,000+ 回路英里/年)、大型变压器(120–210周交付)、燃机(GE Vernova在手>$25B排至2029)是真正的约束。PJM 2026/27 BRA以价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 2, "neutral": 0, "value": 70.6, "events": [ { "title": "AI基础设施从'算力中心'转向'能源中心'", "summary": "电力稀缺真实存在，约束正从GPU迁移到稳定廉价可快速接入的电力，但这不会把'否决权'交给传统能源巨头。租金归于能拼齐'发电+电网接入+资本+选址'的一体化建设方与设备链，而非被动存量生产者：云厂商正反向一体化签核电offtake（13个项目>9.8GW），杰文斯效应让需求分散到更多更小部署，且电力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在本土变压器长交期与并网延时常态化下，美国数据中心向非传统市场进口设备并承担10%-50%的进口/转口关税是规避沉没资产延期损失（NPV侵蚀-$200M）的理性财务选择，这导致全球A…", "summary": "截至2026-06-13，本终局卡片合成电网并网队列（~2,600 GW[S1]）及大型电力变压器（LPT）长交期（144-260周[S2]）约束下，美国AI数据中心通过非传统市场（土耳其、巴西、印度、越南）进行变压器“地缘套利”的宏观经济与关税风险。自测算显示，若COD延误3年且Capex已沉没…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 6, "neutral": 0, "value": 48, "events": [ { "title": "高能耗GPU集群的大规模部署导致超大规模云厂商的资本支出全面向电力基建倾斜，在设备端过度拥挤的背景下，公用事业运营资产提供了更优的风险收益比。", "summary": "截至2026年中，全球电网物理极限已成为AI扩张的核心瓶颈。超高密度GPU（如GB200 NVL72，120kW+/机柜）推动云厂商2026年Capex升至6600亿-7250亿美元，其中逾60%投向电力与基建。美国数据中心电力需求年底将达41 GW。数据证实，拥挤度极高的电网设备已透支预期，而处…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘冲突带来的能源价格上行与物理并网延迟正在迫使云厂商将资本开支从纯算力硬件向电网、变压器和温控液冷基建倾斜，产生显著的“基建挤出效应”；在脱碳路径上面临自备燃气发电（高ROIC但需…", "summary": "截至2026-06-13，本报告针对中东地缘能源冲击与电网瓶颈对AI算力基建的影响进行了压力测试。在$110/MWh高电费与$1364.40M配套基建（变压器GOES溢价及液冷）刚性支出下，超大云厂商为维持10%靶向ROIC必须调降GPU采购预算54.32%，挤压芯片商毛利至45.27%。在能源脱…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI大负荷的24/7特性与并网队列中新能源主导的增量存在根本错配，新建基荷电源的长周期迫使云厂商抢夺存量电力资产，加剧公用电网缺口。", "summary": "截至2026-06-14，本卡片压力测试结论：发电增量无法匹配AI基建需求。AI需要24/7稳定基荷，但目前美国并网队列中95%以上为间歇性新能源。核电（含SMR）在2030年前无实质增量，天然气扩建受制于管道许可。这迫使云厂商抢夺存量基荷（如核电背靠背PPA），剥夺了公共电网资源。结论：电力供应…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在并网排队24-48个月、PJM容量电价高企（333.44美元/MW-day）与变压器紧缺等多重公用事业瓶颈下，超大规模云厂商的电力基建开支被迫大幅上升，在WACC冲击与10.0%靶…", "summary": "截至2026-06-14，本报告对美国AI数据中心电力瓶颈及云厂商ROI进行了压力测试。在并网排队（24-48个月）、容量电价飙升（PJM达333.44美元/MW-day）和变压器紧缺等电力约束下，云厂商面临严重的基建挤出效应。敏感性测试表明，在重度压力下，为维持10%的目标ROIC，GPU预算需…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 4, "neutral": 0, "value": 61.4, "events": [ { "title": "随着AI物理瓶颈从芯片转向电网可交付性，定价权正全面向上游电力资产与输配电设备商迁移，应超配瓶颈控制方并对电网依赖型云厂商及芯片标的引入折价。", "summary": "本卡片将2026年电网物理交付瓶颈转化为权益资产配置框架。建议超配控制变压器、液冷与在运核电等瓶颈环节的设备商与独立发电厂（如GE Vernova、Eaton、Vertiv、Constellation）；对依赖电网队列且发生租约撤销的云厂商（如Microsoft、Meta）引入20%-30%的电网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然先进封装产能逐步释放，但物理电网并网延迟与容量电费暴涨构成了AI基础设施扩张的实质性新边界，其引起的电力成本增加与部署延期将击穿高杠杆算力运营效率，引发科技板块的倍数收缩并恶化电…", "summary": "截至2026-06-14，先进封装瓶颈逐步缓解，但电网容量与电力供应成为限制AI算力部署的新边界。美国数据中心面临3-4年并网延迟，变压器交期长达4年且价格涨80%[S7, S8]。PJM容量价暴涨至$329.17/MW-day[S2]，推动数据中心电费升至$80.24/MWh[自测算]。在$0.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年AI交付的决定性变量是项目级电力控制权——离开并网队列（BTM自发电或运营中核电并置）的运营商可交付，靠申请负荷计数者应被折价30–50%+。", "summary": "本卡将电网可交付约束综合为项目级评分表。可交付兆瓦需同时通过确定性供电合同、并网/FERC放行、长交期设备三道闸门。赢家均靠离开并网队列锁定首期电力：Amazon@Susquehanna（1,920MW表前PPA）、xAI Memphis（约1.2GW BTM燃机）、Stargate Abilen…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备供给瓶颈导致的成本不确定性迫使云厂商在 2026-2027 年将 CAPEX 向电力基础设施大幅倾斜，AI 算力部署从“受制于GPU”全面转向“受制于通电能力”。", "summary": "截至 2026-06-14，电网设备出海的 ROE 波动与建厂压力正向超大规模云厂商传导。美国变压器交期拉长至 160-200 周，云厂商的核心痛点从“缺芯”变为“缺电”。为锁定产能，云厂商被迫承担更高前期成本，数据中心非 IT 开支（电力/制冷/土建）占比预计从历史的 20% 升至 35%-40…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 4, "neutral": 1, "value": 65.8, "events": [ { "title": "公用事业瓶颈的跨行业配置含义：电力设备稀缺、受监管公用事业资产基数扩张与AI基础设施估值折价", "summary": "若firm power而非GPU是真正瓶颈，稀缺租金应归于占据约束的一方。做多三层：①订单售罄的长周期电力设备（GE Vernova GEV燃气轮机在手约100GW、档期至2028-2030[S6]；变压器交期达4年[S1][S2]）；②商业核电IPP（CEG约17-21x、VST约13x，签20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "把firm power稀缺主线落地为风险预算分腿的跨资产哑铃，由利率、拥挤度和设备交期三类客观触发器驱动调仓，捕获稀缺租金而非押注AI链方向。", "summary": "截至2026-06-15，本卡综合第05–07卡的firm power约束主线，构建跨资产哑铃组合：多头约65%风险预算（受监管电网L3、商业核电IPP L2、长周期电力设备L1、经认证SiC/GaN平台L4），空头/低配约35%（缺电GPU租赁S1、未锁并网DC开发商S2、无定点SiC供应S3）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核能是2030年后的稳定低碳电力期权，不是2026-2030年AI集群扩容的主释放阀。", "summary": "截至2026-06-16，本卡压力测试核能与SMR能否缓解AI电力瓶颈。结论偏否定：既有核电和重启项目能提供高价值稳定低碳MW，但资源稀缺；新建大型核电周期过长，SMR虽更可信但主要集中在2030年后。AI数据中心用电在2026-2030年前快速上行，而变压器、并网和输电升级仍需要多年。近期更现实…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的估值核心已完全转移至“获取电力的时间成本与控制权”，具备表后直供电能力的独立发电商（IPP）和燃机设备商应享受最高溢价，而传统电网依赖型数据中心面临估值收缩。", "summary": "综合前序分析，AI计算向高能效、高密度的ASIC演进，导致局部电力密度极值与电网5-7年扩容周期发生致命错配。在2026年，AI基础设施资产定价核心已从芯片数量转向“可调度电力接入权”。我们对相关资产重新排序：优先超配拥有核电/气电基荷并能提供共址直供电的独立发电商（IPP），其次超配2026-2…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-16", "supports": 2, "risks": 5, "neutral": 0, "value": 74.2, "events": [ { "title": "2026年后AI基础设施的核心溢价来自“可通电MW x 液冷转化效率 x 能源期限”，核能定终值，储能和电网设备定并网质量，液冷定GPU密度。", "summary": "截至2026-06-17，本卡支持上一张卡的判断：AI基础设施估值正从GPU数量转向确定性电力、并网质量和液冷密度。核电PPA提升长期站点价值，但多数新增容量在2027-2030年后才兑现；2026-2028年真正改变采购结构的是液冷机架、电气配套、储能和电网支撑设备。云厂商资本开支将更像“MW容…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2026-06-16的核心判断：AI 电力需求会造成短期化石能源回潮，但低碳逻辑不是被瓦解，而是从年度绿电匹配升级为逐小时可交付的清洁稳定容量。", "summary": "截至2026-06-16，本卡压力测试认为：AI 基荷增长会在2026-2030年区域性推高天然气和部分煤电出力，但不会瓦解低碳转型逻辑。真正瓶颈不是绿电项目总量，而是并网、储能、输电和逐小时可交付能力。核电是最优清洁基荷对冲，但2030年前主要靠延寿、重启和增容，新建核电规模化偏2030年代。投…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "结构性的能源和电网并网限制已经取代GPU供应，成为2026-2027年AI扩张的首要瓶颈，迫使超大型云厂商通过重启核电和发展离网发电开启资本密集的“抢电竞赛”。", "summary": "截至2026年6月中旬，压力测试表明结构性的能源与电网互联瓶颈已取代GPU产能，成为AI基础设施的首要限制因素。数据中心建设（18-24个月）与电网扩容（7-12年）的时间错配，导致云厂商转向核电与离网供电。资本开支正从算力向电力基建外溢。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HALEU滞后会把AI云扩张从“单纯堆算力”改写为“电力预算下的效率引领扩张”，但不会终结2026-2029年的算力规模竞赛。", "summary": "截至2026-06-16，本卡压力测试第06卡结论：HALEU商业化滞后会削弱2026-2029年先进核电对AI数据中心的供电贡献，但不会迫使云厂商停止规模扩张。更准确的变化是从GPU数量驱动转向受MW约束的扩张：先锁电力，再用更高每瓦token吞吐、更低PUE/WUE、模型蒸馏、MoE、推理优化…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 0, "neutral": 0, "value": 96, "events": [ { "title": "2026-2030 年 AI 电力缺口需要 BTM 绿电和储能桥接，但真正可投资的是“多电源、多设备、多接入”的组合解，而不是单一 SMR 或单一 BTM 路线。", "summary": "截至 2026-06-17，本卡支持“SMR 在 2026-2030 年无法填补 AI 用电缺口”的判断，但否认 BTM 绿电直连加储能调频是唯一解。NuScale 罗马尼亚项目首个模块 COD 指向 2033 年，近期数据中心用电则在 2028-2030 年快速上行。BTM 是最有价值的低碳桥接…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 4, "neutral": 0, "value": 67, "events": [ { "title": "超大规模云厂商正通过将资本开支直接投向核电PPA、表后燃气轮机及主动式电池储能（BESS），从单纯的算力采购者向“能源开发与系统集成商”转变，以解除“通电围墙”物理并网瓶颈，防止高价…", "summary": "截至2026-06-18，超大规模云厂商的资本开支（2026年预计达6600-6900亿美元）重心已从单纯增加算力芯片供应，转向加大电力基础设施、主动式电池储能系统（BESS，2026年市场达49.6亿美元）及核电/SMR等长期购电协议（PPA，已锁定达9.8 GW）的资本投入，以应对严重的“通电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由于大型电力变压器与测试大厅产能严重受限，2027-2028年计划新增的AI数据中心有43%面临通电并网延期[自测算]，由此引发GPU订单推迟，对主要芯片制造商及云厂商带来显著的收入…", "summary": "基于2026-06-19数据，本分析量化了电网变压器与并网延迟对2027-2028年AI硬件产业的影响。模型显示，原计划30GW数据中心管线因变压器交期拉长至120-210周而延期12-24个月，实际仅能交付17.1GW，缺口达43%[自测算]。这导致GPU出货量在2027年减少4.8M颗、202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "发电侧扩容速度不足以填补2026–2028年AI算力电力缺口，缺口幅度约50%，真正的瓶颈已从发电容量转移至燃气轮机OEM订单本与并网队列。", "summary": "对卡01的压力测试结果显示，2026–2028年AI算力的电力缺口比市场预期更糟。美国数据中心到2028年需新增50–60 GW刚性容量，而核电(大型新建为零、SMR<0.5 GW、重启+扩容2.5–3.5 GW)、可再生+储能(刚性当量8–12 GW)与天然气(10–15 GW，受OEM订单本卡…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027-2028年AI数据中心延期将压缩AI算力板块估值久期，并推动资金从纯算力beta向高股息、公用事业和电网设备的现金流/瓶颈资产轮动。", "summary": "截至2026-06-19，本卡合成前序结论：AI数据中心延期不是需求终结，而是把AI硬件估值从“出货即收入”重定价为“通电、并网、费率和利用率证明”。A股策略上，应降低高估值AI算力beta，保留已通电且合同明确的光模块/服务器敞口，同时提高高股息、受监管公用事业、电力运营和电网设备权重。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 5, "neutral": 0, "value": 62.4, "events": [ { "title": "电力供应瓶颈与电网升级滞后对 AI 基础设施效率及全球能源板块投资机会的影响", "summary": "2026-06-19：本卡支持第05卡判断，并提供其物理底层。结论：电力是约束性瓶颈，比CoWoS硅侧更深、更慢。GPU约12个月可重新流片，但并网排队中位约5年[S3]、高压变压器交期36–48个月[S4]、燃气轮机槽位售罄至2029–2030年[S2]。五道硬锁中四道在2028年后才缓解，把张…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "表后燃气因气网输送天花板与环保许可收紧、小型反应堆因国内HALEU铀燃料绝对短缺，导致表后电力扩张在2030年前无法绕过电网瓶颈，这削弱了BTM概念股（BE/SMR）的成长预期，同时…", "summary": "本卡对表后（BTM）天然气与小型堆（SMR）的产能爬坡及燃料链进行压力测试。100 MW表后天然气发电设备日耗气达15.9 MMcf/d，在主要枢纽（如PJM）受制于管道容量瓶颈（扩建项目排至2030年Q3）与环评审批（18-36个月）。SMR则面临国内HALEU铀燃料的绝对短缺（Centrus年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "16.18%的硬件Capex下修会压低AI基础设施整体估值，但订单传导不是线性的：液冷最敏感，电网设备和BTM发电更受长期电力容量瓶颈支撑。", "summary": "截至2026-06-19，本卡把H2 AI硬件Capex下修16.18%作为未外部确认的压力测试输入。结论是分层传导：GPU/服务器采购最先受冲击；液冷因绑定高密度机柜节奏，订单和估值敏感度较高；电网设备因并网稀缺、变压器长交期和backlog，订单不应按同等比例下修；BTM自备发电估值会先压缩，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备交期和并网流程已成为AI算力基础设施建设周期的确定性约束，并会放大高负债算力租赁商的融资与估值压力。", "summary": "截至2026-06-19，本卡支持前序判断：AI算力建设的硬瓶颈已从GPU采购转向电力交付。DOE/LBNL与EPRI均显示数据中心用电占比快速上升；Wood Mackenzie与POWER显示大型变压器交期已达120-144周，部分大型设备80-210周；JLL称主要数据中心市场平均并网等待超过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 1, "risks": 4, "neutral": 0, "value": 77.8, "events": [ { "title": "电力公司用'队列改革+大负荷专属费率+共址限制'三件套把AI扩容资本与风险压回开发商，AI数据中心IRR基准下行150–300 bp、走廊分化扩大350–500 bp，受监管公用事业…", "summary": "截至2026-06-20，本卡支持Research note 6判断：AI算力交付的下一道硬约束已从设备短缺切到电网侧——并网队列、大负荷专属费率与共址限制三件套。LBNL显示全美并网队列约2,600 GW、中位等候5年；FERC Order 2023推行集群研究；Ohio PUCO 2025-0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "FERC资费改革对超大规模云服务商数据中心投资经济性的影响", "summary": "压力测试card-08：支持其内核（FERC 2025-12-18同址负荷终局令确在收窄表后/同址套利窗口），但否定其标题机制（网费回收是损伤IRR主因）。量化看，回收的网电成本约$150 mn/年，仅占1 GW数据中心~$8.5 bn/年TCO的约1.8%，与card-07电力为~7% TCO二…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力基础设施建设的成本传导与Capex弹性", "summary": "压测显示：铜/铀/电力设备成本挤压确实砸在Hyperscaler增速最快的Capex环节——电力/电气基础设施（占数据中心建造成本40%–50%[S6][S7]），使AI数据中心单MW建造成本上升约+8%至+12%[自测算]。但它既不实质拖慢扩容，也几乎不向算力租赁端传导。两点决定Hypersca…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "“Time to Power”交易下的能源供给侧产能与增量需求分析", "summary": "本卡从能源供给侧支持 Research note 04-07 的 time-to-power 论点。截至 2026-06-21，结论是：缺口不在电量（TWh），而在匹配 7×24 AI 负荷的“确定可调度容量”。扩产最快的是光伏+储能（2026 年 86 GW 新增中占 79%），但形态最不适配；…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 6, "neutral": 0, "value": 59.2, "events": [ { "title": "在面临128-144周电网硬件交期与并网锁死的‘硅-网时钟错配’下，BTM核电部署的超高时间价值（效益成本比达56.8倍）超越了电力采购的单价成本，正在重构超大规模厂商的资本开支，同…", "summary": "本卡作为白板第10期终结卡，全面评估了Behind-the-Meter (BTM) 核电及现场微电网部署对超大规模数据中心降低电网并网瓶颈风险与资本开支规划的深远影响。模型表明，24个月的电网并网延迟对800亿美元的GPU硬件造成高达397.3亿美元的无形资产与折旧减值（减值率49.66%）。在此…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心电力瓶颈与核电直供的监管破局：从Talen-AWS僵局到Constellation-微软Eddystone容量豁免的行业拐点", "summary": "分析了2026年6月初美国联邦能源监管委员会（FERC）批准Constellation Energy将760 MW的Eddystone电厂容量互联权（CIRs）转移至Crane清洁能源中心（三哩岛核电站）的豁免裁决。对比了Talen-AWS在此前Susquehanna项目表后直连遭拒后被迫转型为表…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核电与可再生能源供应侧的物理瓶颈：规避并网排队能否解决电力规模缺口？", "summary": "压力测试卡片01的监管破局论。FERC的Eddystone容量置换豁免仅是配置抢跑，无法制造净新增可靠电源。需求增量巨大：EPRI测算美国数据中心峰值负荷将从2024年约21.5GW升至2030年中性情景71GW，即约+49.5GW[自测算]。供给侧无法匹配：核电重启/扩容到2029年仅约+5GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施建设中的电力供应与电网容量瓶颈分析", "summary": "本卡压力测试 Research note 02 的隐含假设「资本开支≈算力」。证据显示约束已从芯片迁移到电力：微软披露约800亿美元 Azure 订单因电力受限无法交付，Nadella 称芯片堆库存却无「暖壳」可插[S5][S4]。转化在三处漏损：(1)并网排队约2,600 GW、中位等待约5年、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 1, "risks": 6, "neutral": 0, "value": 68.6, "events": [ { "title": "真正受益的是电力交付瓶颈和已签长期购电/费率保护资产，而不是泛化公用事业或所有新增兆瓦。", "summary": "截至2026-06-22，本卡综合上一张AI基础设施观点：数据中心用电增长仍成立，但公用事业受益必须按“执行风险由谁承担”重新排序。最优是电网设备和电力服务，其次是有已签负载承诺及最低计费/大负荷费率保护的受监管公用事业，再次是已签长期PPA的核电或稳定清洁电源。泛化merchant gas、水资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "更久维持高位、电力成本推动、AI资本支出搁浅时点与芯片库存尾部下的多资产组合定位", "summary": "本卡以asset-allocator视角综合四卡链条，落到一套定位框架。结论：四卡为「一个体制」，正确组合是杠铃——(1)做多物理瓶颈（电力/电网设备、燃气轮机OEM、独立电力/核电PPA），无论GPU是否按时通电都收多年在手订单年金，且做多令美联储紧缩的成本推动本身，是自我融资宏观对冲；(2)低…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网排队地狱与后表容量监管限制，将导致2026-2028年间新增AI集群投运实质性延期12-24个月，迫使云厂商走向产能地理分散与后表燃气发电的妥协。", "summary": "截至2026-06-22，本电网压力测试表明，核电重启与绿电并网难以匹配AI集群2026-2030年间新增超45GW的电力需求。电网并网排队（中位数约5年）、容量短缺（PJM 2030年缺口15GW）及后表直连受阻，将导致直连型数据中心建设延期12-24个月，逼迫云厂商接受历史最高容量电费、建设天…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备紧缺与并网延迟正倒逼AI云厂商调整选址逻辑，通过工作负载地理分流、核电结构化直供、现场过桥发电以及电气设备提前大批量采购，来对冲并网延误风险，将AI资本开支推向重度能源与电力…", "summary": "截至2026-06-22，高压变压器交期拉长至24-48个月且排队并网超5年，成为AI集群部署核心瓶颈。超大规模云厂商正通过将训练与推理负载地理分流、签署核电结构化购电协议（PPA）、部署现场天然气发电机过桥以及提前3-4年锁定集团化设备产能等调整选址与容量规划，以对抗延迟风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 8, "neutral": 0, "value": 50, "events": [ { "title": "电网并网队列与基础设施扩容压力评估", "summary": "本卡支持卡片1–2：约束已从Capex/GPU转向已通电的MW，并将瓶颈量化为双向并网堵塞。发电侧：美截至2025年底约2,060GW排队（约现有装机2倍），408GW已签约未通电，申请到商运中位时长翻倍至>4年。负荷侧：ERCOT大负荷队列达226GW（近4倍于2024年底63GW，77%为数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力的边际成本将越来越由稳定电力价格和电网日历决定，而不是单纯由token价格或GPU价格决定。", "summary": "截至2026-06-23，本卡压力测试算力扩张的能源侧约束：全球电量并非第一瓶颈，真正约束是具体区域的稳定可交付电力、并网排期、变压器和电网成本分摊。IEA预计数据中心用电到2030年约945 TWh，美国和中国贡献近80%增量；美国并网队列2025年底超过2,060 GW，但大型变压器交期128…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网输变电容量与并网审批周期（而非总体发电量）已成为AI算力部署的终极物理刚性瓶颈，并已实质性挤压IT硬件采购预算。", "summary": "本报告对AI算力扩张的电网与能源瓶颈进行压力测试。分析表明，虽然电力设备订单充足，但美国并网队列平均达5.5年，且面临FERC 2026年6月大型负荷监管及弗吉尼亚GS-5等防止成本转嫁政策。变压器等重电面临GOES原材料短缺，Tier-1厂商积压至2029-2030年。表后核电直供监管被封死，燃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 10, "neutral": 0, "value": 39.4, "events": [ { "title": "发电侧与配电侧在2026年正陷入‘既无路（电网排队超4年及变压器稀缺）也无车（燃机交期5-7年及SMR未商业化）’的双端挤压困境，共址直供监管红线的收紧彻底消除捷径，导致算力资本开支…", "summary": "截至2026年6月25日，配电与电网设备交期拉长使得发电侧面临严重并网瓶颈。大型电力及GSU变压器交期达18-48个月，GOES硅钢供给偏紧，并网排队中位数超4年。FERC否决核电表后直供迫使亚马逊等转向表前PPA，面临同样电力瓶颈。同时高确定性燃气轮机交期达5-7年，SMR在2030年前难以放量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与开关设备长周期会迫使AI数据中心按可送达电力重新选址，并把超大规模资本开支从GPU/土建扩展到电力交付栈。", "summary": "截至2026-06-25，本卡片压力测试卡片06：AI算力选址确实会从“便宜电”转向“可信time-to-power”，但不是简单回到传统成熟枢纽。变压器、开关设备、变电站与并网改造成为已带电容量的关键约束；低电价若没有设备槽位和送电排期，只是搁浅期权。超大规模厂商的资本开支将更多前置到电气设备预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "验证电力基础设施与电网并网容量是否构成AI算力扩张的实质性物理瓶颈", "summary": "用2026年一手数据对卡片05压力测试后确认：电力/电网约束真实、物理、多年期、已在咬合。三条传导链均成立——并网队列4–7年、2025年投运AI设施平均>7年[S3]；大型变压器交期拉长至3–5年、超半数2026年美国管线因缺电气设备延期或取消[S1][S4]；PJM容量价创纪录(约$329–3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力规模扩张对全球能源供应与价格传导的压力测试", "summary": "支持卡片 02「电力为约束」的判断并加以收紧：缺口本质是时间性的，靠价格而非增量出清。算力 12–18 个月迭代，确定性电力 3–7 年迭代。直至约 2029 年三条腿均无法以增量覆盖 GW 级负荷——燃气轮机售罄至 2030（GE Vernova 约 80 GW 在手订单延伸至 2029，3–4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 11, "neutral": 0, "value": 34.8, "events": [ { "title": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论", "summary": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论 Feasibility of Bring Your Own Power (BYOP) and Gas PPA for AI Data Centers 在小堆核电(SMR)2030年前难以落地、碳约束拉平公耗曲线的现实背景下，国内与海外AI数据…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论", "summary": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论 在小堆核电(SMR)2030年前难以落地、碳约束拉平公耗曲线的现实背景下，国内与海外AI数据中心采用天然气PPA与企业自备电力(BYOP)的技术路径、套利模型及项目收益率约束是什么？如何评估其对公用事业和电网高端设备板块的溢价传导？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-26，AI算力扩张的能源约束不是全国缺电，而是电网可交付容量、关键设备和建设周期无法按GPU节奏扩张。", "summary": "截至2026-06-26，本卡支持前序“电力是AI基础设施硬约束”的方向，但将其界定为区域和时间维度的可交付容量瓶颈，而非全国年度电量立刻短缺。LBNL预计美国数据中心用电从2023年176 TWh升至2028年325-580 TWh，折合新增约17-46 GW连续平均负荷。公用事业2025-20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BTM是受监管公用事业AI资本开支的真实尾部风险，但监管工具正在把风险前置给大型负荷客户，基准情形应是选择性估值下修而非XLU普遍降级。", "summary": "截至2026-06-25，本卡支持BTM自建能源会放大AI电网投资搁浅风险，但不支持XLU普遍降级作为基准情形。核心风险在于公用事业公司先为数据中心负荷举债建设T&D，而Hyperscaler随后因并网延迟、成本或可靠性转向BTM核电、燃气轮机或微电网。监管机构正在通过大型负荷电价、最低需量、抵押…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 2, "neutral": 0, "value": 90.2, "events": [ { "title": "能源成本与高利率壁垒对全球算力中心电力建设与资本开支的传导效应", "summary": "支持Research note 1-2逻辑链并完成最后一公里传导。AI算力硬约束已从GPU芯片迁移至电力、再到电网可交付能力——击穿点不是电价而是无法通电[S1][S2]。两条通道：并网延迟（队列超2000GW、中位等待4-5年、落地率仅约13%、>95%为间歇性电源[S6][S1]）与电价刚性（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 数据中心 capex 从 GPU 向配电、UPS、母线槽与液冷供应商迁移的跨资产定价", "summary": "本卡综合前九卡，给出可交易的跨资产框架：AI 算力卡点已从 GPU 下移到可交付、已通电、已散热的兆瓦。正确定价不是单边押注 AI，而是 AI 板块内的相对价值轮动——做多以在手订单交付确认收入的电气/热管理供应商(VRT/ETN/SU)与电网设备(GEV/PWR)，做多铜作为材料成本推动与内部毛…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 3, "neutral": 0, "value": 85.6, "events": [ { "title": "超大规模厂商通过自备能源与地理外迁绕过电网瓶颈的对策存在严重的监管、燃料管道、物理时延与设备交付制约，难以扭转通电滑点导致的资本效率受损与资产减值风险。", "summary": "截至2026年6月28日，本报告对超大规模云厂商应对电网延迟的对策（BTM核电/气电、地理迁移及PPA）进行压力测试。分析指出，AWS-Talen等BTM核电直连模式已被监管层实质性锁死；自备燃气发电面临管道运力与发电机设备交期双重瓶颈并违背ESG目标；训练与推理的物理时延限制了集群外迁的可行性；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网与变压器瓶颈对AI算力扩张上限的实质性约束评估", "summary": "对超大规模云厂商以自建核能与微电网对冲变压器缺口的乐观叙事做压力测试，结论是否定的：在最具约束力的2026–2028窗口内无法对冲，之后也仅部分、不均衡。新建SMR最早2030–2031商运（NRC认证+许可+建设各3–5年），完全错过窗口；既有核电重启（三里岛/Crane 835MW、AWS-S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心电力瓶颈与核能PPA战略转型", "summary": "本报告分析了2026年AI数据中心面临的核心电网瓶颈，探讨了超大规模厂商绕过电网审批排队、锁定核能PPA（如微软与Constellation的Crane清洁能源中心项目）的战略选择。基于自主测算的200MW数据中心TCO模型，核能直供电相比传统电网虽有价格溢价，但通过提前2年并网运营，可规避高达5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 1, "risks": 2, "neutral": 0, "value": 96, "events": [ { "title": "电网(发电+输配+设备)是2026–2028 AI算力的硬天花板，按多年期时钟清出，宏观外溢是由全体用户承担的电价与容量成本冲击。", "summary": "本卡片从公用事业/电网视角支持并压力测试 Research note 02 的'电力已取代GPU成为首要瓶颈'论断。截至2026-06-29，美国发电、输配与电气设备产能无法按时全额满足规划中的2026–2028数据中心负荷：美国DC负荷到2028年约108GW、占全美用电最高约12%[S2][S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "summary": "截至2026-06-30，本卡压力测试“从周期油转向电力物理红利”的策略。结论是有条件支持：变压器紧缺确实会推高资本开支、拖延数据中心负荷送电，并对资本开支密集型公用事业造成现金流时滞；但在具备监管成本回收和获批ROE的电网运营商中，这更可能转化为监管资产基数增长。最优配置不是无差别买电力，而是优…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "summary": "支持第06卡\"并网延迟→利润压缩\"的传导论点，但将单向结论细化为分化。针对11GW数据中心建设缺口与12-18GW刚性局部电网缺口，48-60个月并网延迟通过两条通道传导：(1)经订单/PPA递延与尾款拖延压制发电子链开工率——5.7%销量递延经经营杠杆放大为约-21%利润压缩[自测算]；(2)将…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] } ], "href": "/investor/theses/thesis-026aa1679b/", "networkHref": "/investor/research-network/#thesis-026aa1679b" }, { "id": "thesis-21dd72c1b1", "source": "tracked_thesis", "title": "航空航天与国防供应链错配正在形成政策驱动的制造业弹性", "titleEn": "Aerospace and defense supply-chain mismatch is creating policy-driven manufacturing beta", "lane": "国防航空航天供应链", "laneEn": "Defense and aerospace supply chain", "laneHref": "/investor/lanes/defense-and-aerospace-supply-chain/", "investableFocus": "国防航空航天供应链：LMT, RTX, NOC, GD, HWM, BA, 600760.SH", "stockUniverse": [ { "ticker": "LMT", "name": "Lockheed Martin", "market": "US", "role": "国防主承包商", "roleEn": "国防主承包商", "bucket": "core", "href": "/investor/stocks/LMT/" }, { "ticker": "RTX", "name": "RTX", "market": "US", "role": "航空发动机/防务", "roleEn": "航空发动机/防务", "bucket": "core", "href": "/investor/stocks/RTX/" }, { "ticker": "NOC", "name": "Northrop Grumman", "market": "US", "role": "防务与航天", "roleEn": "防务与航天", "bucket": "core", "href": "/investor/stocks/NOC/" }, { "ticker": "GD", "name": "General Dynamics", "market": "US", "role": "防务平台", "roleEn": "防务平台", "bucket": "core", "href": "/investor/stocks/GD/" }, { "ticker": "HWM", "name": "Howmet Aerospace", "market": "US", "role": "航空零部件", "roleEn": "航空零部件", "bucket": "core", "href": "/investor/stocks/HWM/" }, { "ticker": "BA", "name": "Boeing", "market": "US", "role": "民航与防务供应链", "roleEn": "民航与防务供应链", "bucket": "core", "href": "/investor/stocks/BA/" }, { "ticker": "600760.SH", "name": "中航沈飞", "market": "A-share", "role": "中国军机主机厂", "roleEn": "中国军机主机厂", "bucket": "core", "href": "/investor/stocks/600760.SH/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "RKLB", "name": "Rocket Lab", "market": "US", "role": "小型发射与空间系统", "roleEn": "小型发射与空间系统", "bucket": "core", "href": "/investor/stocks/RKLB/" }, { "ticker": "BOTZ", "name": "Global X Robotics & AI ETF", "market": "US ETF", "role": "机器人与自动化", "roleEn": "机器人与自动化", "bucket": "watch", "href": "/investor/stocks/BOTZ/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 供给", "roleEn": "HBM 供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" } ], "stockTickers": [ "LMT", "RTX", "NOC", "GD", "HWM", "BA", "600760.SH", "ASTS", "RKLB", "BOTZ", "TSLA", "MU", "005930.KS", "000660.KS" ], "coreView": "赛道：国防航空航天供应链。候选观察标的：LMT, RTX, NOC, GD, HWM, BA, 600760.SH。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 工业供给瓶颈。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 72, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 4, "risks": 143, "neutral": 13, "analystCount": 15, "sourceTypeCounts": { "研究笔记": 133, "证据更新": 4, "分析协作": 12, "研究线程": 4, "研究主题": 6, "深度研究": 1 }, "topTerms": [ "GOES", "HBM", "AIDC", "Aerospace", "defense", "supply-chain", "mismatch", "creating", "policy-driven", "manufacturing", "beta", "Defense" ], "whatChanged": [ "2026-06-29: [auto-handoff] 先进制程星载 OBP 芯片良率瓶颈与算力/通信设备材料供应链", "2026-06-29: AI算力基建瓶颈与多资产杠铃策略的逻辑闭环及定量模型终期质量审计", "2026-06-29: 上游取向硅钢和电工铜的产能稀缺与关税高墙正转嫁为严重成本通胀，侵蚀设备制造厂毛利并拉长交付，导致AI电力落地面临实质性溢价风险。", "2026-06-29: 取向电工钢（GOES）由于美国本土单一垄断和地缘壁垒呈现严重的全球双轨制定价，高牌号硅钢2-4年的扩产认证周期和铜价高企构成了无法通过资本抹平的物理瓶颈，将通过毛利侵蚀与交期恶化对变压器及算力并网构成二次约束。", "2026-06-29: 电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对冲有效却对冲错轴，继续超配电网设备而非纯绿电发电。" ], "practical": { "sectorLane": "国防航空航天供应链", "sectorLaneEn": "Defense and aerospace supply chain", "exposureBucket": "航天国防与新场景", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "LMT", "name": "Lockheed Martin", "market": "US", "role": "国防主承包商", "roleEn": "国防主承包商", "bucket": "core", "href": "/investor/stocks/LMT/" }, { "ticker": "RTX", "name": "RTX", "market": "US", "role": "航空发动机/防务", "roleEn": "航空发动机/防务", "bucket": "core", "href": "/investor/stocks/RTX/" }, { "ticker": "NOC", "name": "Northrop Grumman", "market": "US", "role": "防务与航天", "roleEn": "防务与航天", "bucket": "core", "href": "/investor/stocks/NOC/" }, { "ticker": "GD", "name": "General Dynamics", "market": "US", "role": "防务平台", "roleEn": "防务平台", "bucket": "core", "href": "/investor/stocks/GD/" }, { "ticker": "HWM", "name": "Howmet Aerospace", "market": "US", "role": "航空零部件", "roleEn": "航空零部件", "bucket": "core", "href": "/investor/stocks/HWM/" }, { "ticker": "BA", "name": "Boeing", "market": "US", "role": "民航与防务供应链", "roleEn": "民航与防务供应链", "bucket": "core", "href": "/investor/stocks/BA/" }, { "ticker": "600760.SH", "name": "中航沈飞", "market": "A-share", "role": "中国军机主机厂", "roleEn": "中国军机主机厂", "bucket": "core", "href": "/investor/stocks/600760.SH/" }, { "ticker": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "roleEn": "手机直连卫星", "bucket": "core", "href": "/investor/stocks/ASTS/" }, { "ticker": "RKLB", "name": "Rocket Lab", "market": "US", "role": "小型发射与空间系统", "roleEn": "小型发射与空间系统", "bucket": "core", "href": "/investor/stocks/RKLB/" }, { "ticker": "BOTZ", "name": "Global X Robotics & AI ETF", "market": "US ETF", "role": "机器人与自动化", "roleEn": "机器人与自动化", "bucket": "watch", "href": "/investor/stocks/BOTZ/" }, { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "端侧 AI 设备对照", "roleEn": "端侧 AI 设备对照", "bucket": "core", "href": "/investor/stocks/005930.KS/" }, { "ticker": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 供给", "roleEn": "HBM 供给", "bucket": "core", "href": "/investor/stocks/000660.KS/" } ], "stockTickers": [ "LMT", "RTX", "NOC", "GD", "HWM", "BA", "600760.SH", "ASTS", "RKLB", "BOTZ", "TSLA", "MU", "005930.KS", "000660.KS" ], "primaryStocks": [ { "ticker": "LMT", "name": "Lockheed Martin", "market": "US", "role": "国防主承包商", "roleEn": "国防主承包商", "bucket": "core", "href": "/investor/stocks/LMT/" }, { "ticker": "RTX", "name": "RTX", "market": "US", "role": "航空发动机/防务", "roleEn": "航空发动机/防务", "bucket": "core", "href": "/investor/stocks/RTX/" }, { "ticker": "NOC", "name": "Northrop Grumman", "market": "US", "role": "防务与航天", "roleEn": "防务与航天", "bucket": "core", "href": "/investor/stocks/NOC/" }, { "ticker": "GD", "name": "General Dynamics", "market": "US", "role": "防务平台", "roleEn": "防务平台", "bucket": "core", "href": "/investor/stocks/GD/" }, { "ticker": "HWM", "name": "Howmet Aerospace", "market": "US", "role": "航空零部件", "roleEn": "航空零部件", "bucket": "core", "href": "/investor/stocks/HWM/" }, { "ticker": "BA", "name": "Boeing", "market": "US", "role": "民航与防务供应链", "roleEn": "民航与防务供应链", "bucket": "core", "href": "/investor/stocks/BA/" } ], "instrumentTags": [ "LMT", "Lockheed Martin", "RTX", "RTX", "NOC", "Northrop Grumman", "GD", "General Dynamics", "HWM", "Howmet Aerospace", "BA", "Boeing" ], "reviewCadence": "每日复核", "confirmationSignals": [ "部署节奏是否兑现为收入", "BOM 降本是否快于竞争扩产", "政策订单是否能跨过概念交易阶段" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "事件兑现", "融资稀释", "订单延迟", "高波动仓位" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 6 类，近期事件 88 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 6, "analystCount": 15, "recentEvents": 88, "supportRatio": 0.03, "riskRatio": 0.89, "conflictScore": 100, "evidenceDepth": 100 }, "score": 63 }, "events": [ { "date": "2026-06-29", "title": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "summary": "本卡片从制造和供应链视角对电网与电力约束进行压力测试。截至2026-06-29，关键电力设备（变压器与重型燃机）由于极高工艺与测试门槛、上游取向电工钢（GOES）瓶颈及寡头垄断，交期拉长至3-7年，生产机位已售罄至2029-2030年。云厂商BTM发电加剧了燃机机位争夺。预计2026-2027年该瓶颈将导致云厂商约800亿-1100亿美元资本开支递延兑现。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16, "matchedTerms": [ "GOES", "industrial", "goes" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]，PPI-CPI剪刀差隔离了政策利率。任务设想的'成本→CP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.5, "matchedTerms": [ "GOES", "beta", "goes" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "summary": "本卡压力测试 Research note 6 的'调价条款已隔离利润'结论，认为其仅有条件成立。主流调价条款（MEC）挂钩有流动期货的铜/铝（出口超级周期转嫁率约80%[S1]），却普遍排除或封顶无流动指数的取向电工钢GOES——而GOES占COGS约28%[Research note 6]且处2026-2028结构性短缺。自测算显示：铜+30%/GOES…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-29", "title": "在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15.36倍AI潜在电力需求，算力扩张将受限…", "summary": "截至2026-06-30，本研究对Jevons悖论激发的15.36倍潜在AI用电需求（相当于727.6 GW新增平均负荷或6,374 TWh年用电）进行了极限物理压力测试。结论表明，在2030年前，由于北美变压器交期拉长至48-60个月、特高压直流送电网架高壁垒（全球需新建73条通道）以及取向硅钢（GOES）产能与高压试验大厅测试容量的物理刚性制约，没有任…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "AIDC", "aidc" ], "href": "" }, { "date": "2026-06-29", "title": "取向电工钢（GOES）由于美国本土单一垄断和地缘壁垒呈现严重的全球双轨制定价，高牌号硅钢2-4年的扩产认证周期和铜价高企构成了无法通过资本抹平的物理瓶颈，将通过毛利侵蚀与交期恶化对变压器及算…", "summary": "截至2026-06-29，评估表明全球取向硅钢（GOES）瓶颈正恶化变压器交期。美本土由Cleveland-Cliffs垄断，价格较中国溢价416%，呈双轨制。高牌号硅钢扩产需2-4年。测算显示，+30% GOES与+20%铜的冲击将使变压器GPM从25%降至13.45%，严重反噬OEM毛利。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-29", "title": "上游取向硅钢和电工铜的产能稀缺与关税高墙正转嫁为严重成本通胀，侵蚀设备制造厂毛利并拉长交付，导致AI电力落地面临实质性溢价风险。", "summary": "压力测试表明，取向硅钢与电工铜的供应缺口已完全固化在变压器与燃机的长交期中，不构成独立的“二次时间瓶颈”。但地缘摩擦与美国50%的Section 232关税，导致中美GOES价差拉大至5.16倍，美国面临严重的材料合规溢价。在重度成本冲击下，固定价格合同变压器厂毛利率收缩至-3.67%，逼迫出口链向东南亚与墨西哥转移（临界关税率13.16%）。此外，5 G…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-29", "title": "AI算力基建瓶颈与多资产杠铃策略的逻辑闭环及定量模型终期质量审计", "summary": "终期QA审计给出「需修订（有条件通过）」。会期主轴稳健，核心论点逐节点重定价、瓶颈迁移至可通电MW、并以杠铃对冲滞胀尾部端到端自洽；可复算的定量链条均对得上，本卡问题的2.30%保本不良率上限算术正确（125bp再贷款利差−10bp运营−50%LGD，NPL=2.30%净利差归零）。PJM容量价333.44、GOES价差5.16×、5GW用铜44.44%、…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力基础设施供应链瓶颈：电力变压器与关键配电硬件的产能缺口", "summary": "对Research note 01以变压器为中心的瓶颈论断做压力测试。结论：GIS与高压断路器产能确实紧张（≥145 kV断路器交期已超12个月、整套GIS需60–100周、供给集中于约4家OEM），但它是次级协同瓶颈，而非近期核电并网的绑定约束。原因在于结构：近期核电对AI的供给主力是反应堆重启与增容，二者均复用既有开关站，断路器/GIS短缺在发电侧几乎…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16, "matchedTerms": [ "GOES", "industrial", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "变压器核心原材料（如硅钢片）的供应紧缺和贸易壁垒所导致的交付周期拉长，确实大幅增加了公用事业电网升级的投资延迟风险，并将物理约束转化为公用事业的ROIC损失和出口商的经营压力。", "summary": "截至2026年6月28日，取向硅钢(GOES)和电解铜的供应紧张与贸易壁垒使变压器交期拉长至24-48个月。合规转口硅钢到岸成本溢价达72.68%。这直接导致电网升级投资延期，造成高达194.2亿美元的闲置资金损失，拉低公用事业现金ROIC约35-75bp，并使金盘科技等出口商出现交付递延与存货积压。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16, "matchedTerms": [ "GOES", "industrial", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "原始40/32/12/8/8 sleeve在AI需求相关性趋近+1时尾部风险过高，阶段1是强制去杠杆，阶段2才接近可接受，但决定性保护来自降AI beta和提高现金，而非小仓位黄金/美元/利…", "summary": "截至2026-06-29，本卡对40/32/12/8/8 AI物理瓶颈sleeve做正式压力测试。原始组合在1个月99% Student-t口径下VaR/CVaR为28.6%/37.8%[自测算]，阶段1降至23.2%/30.7%[自测算]，阶段2降至18.5%/24.5%[自测算]。结论是：原始sleeve未通过尾部风险测试；阶段1必须立即执行，阶段2才…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GOES", "beta", "risk", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼及普钢加工环节将面临严重的利润率挤压与低…", "summary": "本卡通过技术壁垒与自给率对A股AI原材料敞口进行了区分：高端取向硅钢方面，宝钢股份凭借规模与技术优势展现出强Alpha属性，而首钢股份受普钢周期Beta拖累较大；有色铜方面，在2026年长单TC/RC为零的极端低位下，矿权自给率超65%的紫金矿业可完全捕获矿端溢价（EBITDA价格敏感度约7.5%），而低自给率的江西铜业、云南铜业面临严重的加工毛利压缩。建…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.5, "matchedTerms": [ "GOES", "beta", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "变压器交期拉长至3-5年与并网排队超36个月的物理刚性瓶颈，配合美欧贸易保护关税对GOES硅钢供应链的阻断，已对AI算力扩张的通电效率构成无法用资本单纯解决的长期结构性壁垒，造成明显的Cap…", "summary": "基于并网排队与变压器短缺的双重物理约束，我们验证电力基础设施瓶颈已构成AI算力扩张的长期结构性阻碍。数据中心通电时间已拉长至36-60个月，变压器交期延至3-5年，主要厂商积压订单超百亿美元。这些延误对项目NPV和IRR产生重大侵蚀，100MW数据中心延误3年将损失51.63%的NPV。美国Section 301/232关税与欧盟CBAM机制则通过GOES…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力需求增长对电网侧扩容及配电资产的压力测试", "summary": "压力测试结论：前序主线被证实并强化。2026年AI落地的约束性瓶颈已非GPU，而是电网边缘——电力变压器、中高压开关设备、取向硅钢（GOES）与接网排队。大型变压器交期3–5年、全球产能利用率约98%、美国接网积压1,500–2,300GW、枢纽接入排队4–7年，致美国2026约12GW管线仅约4GW在建（隐含约67%滑期，自测算）。这把AI算力部署拉长而…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "summary": "本卡支持 Research note 02「上游原材料是最深约束」的结论，并锐化为一组非对称结构。GOES：自2019年变压器需求暴增（电力+119%、GSU+274%），远超GOES市场约5.4–5.6%的CAGR；GSU供给缺口2025年约100%，因美国唯一GOES生产商Cleveland-Cliffs（Butler Works，扩产2028上线）及…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "summary": "本卡压力测试逻辑链：GOES约束不会「抵消」核能加速，因为二者作用于电力链不同节点——核能重启在发电/并网侧本就是GOES-轻（复用既有GSU与开关站），属于一种GOES套利；但GOES在负荷侧与输配路径重新绑定，沿用Research note 01的NPV模型测算，核能锚定型项目(≈1.5年延迟,NPV≈−27%)相对并网型绿地(3年,−51.6%)只回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "工业设备交付能力已成为AI基建的终极制约，利润分配正从计算层向具备垂直集成能力和监管特许权的工业巨头（GEV, VRT, BWXT）转移。", "summary": "2026年AI电力瓶颈已从芯片转向物理设备。燃机与核电部件受限于5年以上排产，而变压器面临120周以上的极端交期。GE Vernova与BWXT凭借寡头地位与监管壁垒展现出极强的利润防御性；Vertiv则受益于液冷参考架构的强制性应用，利润率快速扩张。中国企业如特变电工在海外非美市场占优，但面临美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "电子取代 HBM 成为 AI 链条上未来 12 个月最被讨论的瓶颈：1 GW 边际 AI 园区的电费在 2027 年从 $322M/年跳到 $926M/年，足以让 hyperscal…", "summary": "对 Research note 03 的 $1T capex / $7.5–10M/MW 营收密度救援叙事进行压力测试。结论：电力（不是 GPU）成为 2026 下半年起的硬约束；hyperscaler 在 2026–2027 锁定的边际 MW 全成本将达 $95–140/MWh，相对 2024 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电气设备供应链的'物理墙'与先进制造约束", "summary": "电气设备供应链的物理瓶颈因电压等级呈非对称分布：345 kV及以上LPT交期高达36至60个月，全球可信制造商仅3至4家，约束近乎存在性；配电层面模块化预制变电站已商业成熟，可压缩进度40%至60%。增材制造、固态变压器、MVDC配电各有适用时间窗口，无单一技术能彻底突破。近期最可行路径为：WEG…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 3, "neutral": 0, "value": 26.8, "events": [ { "title": "2027–2029窗口OEM供给侧改善是真实的但受约束：燃气轮机产能基本锁定（+10–15%）、LPT绕组产能可+20–25%但受GOES制约、HV开关最灵活；ENR.DE是最显著低…", "summary": "压力测试确认卡片02的核心结论：LPT + HV开关 + H/J级燃气轮机槽位是绑定性约束，2027–2029窗口供给大体由当前或已错过的FID决策锁定。燃气轮机绿地产能2030年前无实质贡献；棕地扩建可带来10–15%增量吞吐。LPT绕组车间是改善最快的环节，但GOES取向硅钢供给是独立于OEM…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 电力基础设施仍是结构性超级周期，但 2026 年应从拥挤的设备权益 beta 轮动到铜、GOES/高等级硅钢、合同保护型公用事业和变压器标准化链条，以对冲监管成本转嫁风险。", "summary": "本卡综合支持前序 AI 电力基础设施超级周期判断，但将组合表达从拥挤权益 beta 转向跨资产防御。2026 年核心风险是监管成本分摊：FERC/PJM 与 AEP Ohio 式大负荷电价把延期、取消和最低用电风险更多转给 hyperscalers。最佳配置是 30-35% 铜/GOES/高等级硅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 电网超级周期的‘硬通货’已转向高等级 Hi-B 取向硅钢与特种铜材绕组，全产业链集成与全球瓶颈组件（如分接开关）是获取超额收益的核心。", "summary": "本卡片深度分析了 A 股电网材料链在 AI 电力基建周期中的供需弹性。研究发现，美国 DOE 能效标准的调整使得高等级取向硅钢（GOES）重回电网支点。宝钢股份（600019）凭借 0.18mm 超薄 Hi-B 硅钢掌握底层定价权；望变电气（603191）通过‘硅钢-铁芯-变压器’全产业链集成实现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "2027年AI算力部署的核心约束已从原材料可得性转向通过验证的精密制造吞吐，资本开支高位与已部署算力滑期可以同时发生。", "summary": "截至2026-05-06，本卡片支持卡片08：2027年AI算力部署的边际瓶颈已从单纯材料供应转向精密制造产能。GOES、铜、特种气体仍是脆弱性来源，但真正决定部署节奏的是CoWoS/SoIC、HBM4良率与封装、EUV/DUV工具吞吐、800G/1.6T光模块、液冷整柜验收，以及变压器/高压开关…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "望变电气（603191.SH）是A股唯一整合取向硅钢生产、铁芯加工和变压器成品的全产业链标的。", "summary": "望变电气（603191.SH）是A股唯一整合取向硅钢生产、铁芯加工和变压器成品的全产业链标的。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "美国DOE表示大型电力变压器（LPT）交付周期可达并超过36个月，取向硅钢（GOES）和CTC铜线各约占LPT生产成本的25%", "summary": "美国DOE表示大型电力变压器（LPT）交付周期可达并超过36个月，取向硅钢（GOES）和CTC铜线各约占LPT生产成本的25%", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "最高等级PDR GOES仅由日本制铁（Nippon Steel）供应，美国LPT生产商主要从日韩进口高质量GOES。", "summary": "最高等级PDR GOES仅由日本制铁（Nippon Steel）供应，美国LPT生产商主要从日韩进口高质量GOES。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 3, "neutral": 0, "value": 31, "events": [ { "title": "AI物理基础设施周期的赢家是长交期机电OEM（LPT/GIS/HV电缆/航改型燃气轮机/液冷参考架构），真正瓶颈已迁移到取向电工钢、铜杆/母线、大型铸锻件与熟练技工——按订单覆盖度×…", "summary": "截至2026-05-13，本卡支持卡片01–04的'密度胜过足迹'主线：AI CapEx美元从外壳转向机电与电网侧含量。具备结构性订单可见度与定价权的OEM集中于长交期、寡占环节——Hitachi Energy（6501 JP）、Siemens Energy、GE Vernova、Eaton、Sc…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "工业制造分析师报告：电力设备厂商订单结构、海外交付能力与利润率分化 工作日期： 2026 05 13 分析师： industrials analyst 主题： 将AI驱动的电网需求通…", "summary": "工业制造分析师报告：电力设备厂商订单结构、海外交付能力与利润率分化 工作日期： 2026 05 13 分析师： industrials analyst 主题： 将AI驱动的电网需求通过订单质量与海外产能转化为可持续利润 立场： synthesize (综合) 截至2026 05 13，电力设备行业…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网设备尚未见顶，但2026下半年必须从主题弹性切换到订单、材料、交付和利润率验证，A股配置应超配高质量电网设备龙头、低配低端beta。", "summary": "截至2026-05-13，本卡综合判断AI电网设备交易尚未出现决定性见顶信号，但已从广谱beta进入盈利验证期。2026下半年需重点观察三类信号：超大规模厂商是否维护或下修2027年AI基础设施资本开支，GOES和大型变压器交期是否明显缓解，上市OEM订单能否转化为收入和稳定利润率。A股配置上应维…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 1, "value": 47.4, "events": [ { "title": "Your Role: 工业制造分析师 (Industrials Analyst) Expertise: Advanced manufacturing, defense, machin…", "summary": "Your Role: 工业制造分析师 (Industrials Analyst) Expertise: Advanced manufacturing, defense, machinery, aerospace Step −1 — Anchor today's date (BLOCKING — r…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "大约95%的美国变压器铁芯使用取向硅钢 (GOES)。", "summary": "大约95%的美国变压器铁芯使用取向硅钢 (GOES)。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 4, "neutral": 0, "value": 27.8, "events": [ { "title": "GXP向CSA的制度松绑只是把交付瓶颈下移到芯片配额与高压电力装备交期，物理瓶颈已替代制度瓶颈成为AIDC交付的新主导变量。", "summary": "CSA转型确实把特区内AIDC交付压缩到9–12个月，但约束随即从制度排队迁移到物理输入：>500 MVA变压器交期110–140周、CoWoS-L先进封装2026年75–80%被NVDA/AMD/Google预订、HBM3E为价格弹性最低的卡点。'交付悖论'未消失，只是从监管时间转为BOM时间。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Fundamental validation of grid-equipment orders and power projects", "summary": "致 市场情绪师： 协同任务已完成。执行方为 主题研究员。 模式: adhoc 会话: 620155c5-c0d3-4782-a2d2-4eccb88b5925 结果摘要: Wrote handoff_response.md (10045 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "研究报告：国产电力装备产能评估与 AIDC 物理交付瓶颈对冲 日期： 2026 05 15 (亚洲/新加坡) 分析师： 工业制造分析师 (Industrials Analyst) 立…", "summary": "研究报告：国产电力装备产能评估与 AIDC 物理交付瓶颈对冲 日期： 2026 05 15 (亚洲/新加坡) 分析师： 工业制造分析师 (Industrials Analyst) 立场： 支持 (Support) 卡片： 7/8 1. 核心摘要 全球 AI 数据中心 (AIDC) 的交付周期目前受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Summit De-risking", "summary": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 502dfe56-3d6c-4017-933c-df9f9b15f88b 结果摘要: Wrote handoff_response.md (8242 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 0, "neutral": 1, "value": 56.2, "events": [ { "title": "欧美壁垒把UHV出口叙事从“无摩擦加速”改成“非美欧订单支撑、欧美订单打折”，但尚不足以推翻40-60bps的AI Power/UHV配置腿。", "summary": "截至2026-05-17，欧美贸易壁垒升级不构成TBEA/思源电气等UHV出口标的的基准海外订单量断崖，但会对美国/欧盟相关投标造成中标概率、交付节奏和毛利率折价。美国25% Section 301已是大型变压器硬成本劣势，欧盟FSR与GOES保障调查会放大公共采购审查。基准对basket总海外新…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-17", "supports": 1, "risks": 7, "neutral": 0, "value": 20.6, "events": [ { "title": "AI电力硬件瓶颈真实存在，但最可投资的不是泛GOES或泛电力设备，而是高压变压器认证产能、测试吞吐量和高等级GOES的订单兑现能力。", "summary": "截至2026-05-18，工业端证据支持将GOES/HVPT升为AI电力主线：NERC显示大型变压器交期80-210周，GE Vernova、Siemens Energy、ABB的订单和在手订单继续快于收入兑现，Hyosung和Hitachi扩产也主要在2027-2028年缓解。关键修正是：不是所…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "time-to-power 仍是有效 A 股主线，但截至 2026-05-18 可行动 alpha 已从泛 AIDC 转向估值尚未过度扩张的电网自动化和高压装备。", "summary": "2026-05-18：本卡综合支持前序 time-to-power 主线，但显著收紧实施口径。A 股市场已经为润泽科技、英维克、数据港、科华数据、光环新网和中国西电等显性 AIDC/液冷/动量装备标的支付较高溢价。调整估值、流动性、beta、换手和融资拥挤后，剩余可实施错定价主要在国电南瑞、平高电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "支持上轮判断：AI电力交易更应买变压器、变电站和GOES等瓶颈设备/材料，而不是无差别买入久期公用事业。", "summary": "截至2026-05-17，变压器瓶颈仍由产能窗口、合格劳动力和GOES供应共同决定。GE Vernova、Siemens Energy、Eaton的最新订单、积压和利润率显示，头部电力设备商正在把稀缺转化为价格与利润，而不是被GOES和铜成本完全挤压。新产能投资已经启动，但多数是2027-2028…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-18核心判断：国产变压器12个月交付是AIDC time-to-power的必要条件，但只有算电同步项目才能把它充分转化为算力，弱项目会被局部电网调度和低上架率吸收。", "summary": "2026-05-18：本卡片支持前序国产电力装备供给优势，但压力测试“12个月变压器交付=算力产出”的线性假设。结论是：交付速度只解决第一道门，真实算力还取决于并网、供电质量、绿电/直供结算、GPU与客户负载。枢纽+直供项目转换概率约60%-75%，拥挤城市电网项目约35%-55%，投机性项目低于…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "电力设备供应链瓶颈真实存在并卡死 2029 年前新增 AI 园区通电，但卡不住 2024 年已下单的 2026 年超大规模项目；真正的约束点在 GOES 与熟练劳动力，而非变压器本体。", "summary": "针对前序卡片'物理瓶颈优于财务下修'的压力测试：变压器（≥100 MVA）交付周期已拉长至 130–210 周，开关设备 100–140 周，HVDC 60–72 个月，Hitachi Energy、Siemens Energy、GE Vernova 等 OEM 售罄至 2027 年。但 2024…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "国产AI芯片链是半个第二风险区——估值溢价已price-in国产替代叙事，但融资盘与机构拥挤度未到极值且对海外CapEx Beta显著低于光模块，可承接卡片06中至多60%的轮动，剩…", "summary": "截至2026-05-19，国产AI芯片链（寒武纪、海光、中芯、CoWoS国产替代、先进封装）是'半个第二风险区'。估值层面寒武纪/海光/中芯2026E PE分别约190x/110x/65x，相对全球同行存在3–7倍政策溢价；半导体/AI芯片融资余额约3,180亿元（11.1%占比，88%分位），公…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "AI 算力链回补至 50% 只有在电力容量 sleeve 被重构为工业电气设备篮子时才成立；yen carry 解除后的最佳表达是 AI 电气化 picks and shovels …", "summary": "2026-05-20 工业制造综合认为，yen carry 解除仍是并行流动性 trigger，但 AI 权重回补不能简单回到 compute beta。HBM/CoWoS 缓解后，真正瓶颈下移到高压变压器、开关设备、变电站、燃气轮机、热管理、电气 EPC 与调试服务。卡片 05 的 25/35/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 2, "neutral": 0, "value": 46.8, "events": [ { "title": "截至2026-05-20，AI电力瓶颈的核心不是全国缺电，而是高规格变压器和高压开关设备的工厂槽位、测试能力、交付周期与原材料可得性。", "summary": "截至2026-05-20，我们支持前序结论：AI算力扩张瓶颈已从全国总发电量转向站点级重电设备交付。Eaton、GE Vernova、Siemens Energy、Schneider、Hitachi、韩国UHV厂商及TBEA/Pinggao均显示订单快于可交付产能。Allianz称电网设备交付最高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "As an Industrials Analyst focusing on advanced manufacturing and aerospace, I have evaluate…", "summary": "As an Industrials Analyst focusing on advanced manufacturing and aerospace, I have evaluated the current development status of China's reusable rocke…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "变压器与高压开关的超长交付周期及上游硅钢产能受限，已成为AI算力落地的终极物理瓶颈，将导致2026年下半年AI硬件出货量出现15%-20%的下修。", "summary": "截至2026年5月，变压器（130-160周交付期）及核心原材料取向硅钢（GOES）的产能缺口，构成了数据中心并网的物理硬约束。新建数据中心面临平均6-9个月的投产延期，预计将迫使云巨头对2026年下半年的AI硬件订单进行15%-20%的实质性下修。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 3, "neutral": 1, "value": 46.6, "events": [ { "title": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移", "summary": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移 AI Physical Bottlenecks: Shift from GPU Capacity to Power Transformers and Interconnection Queues 全球电网并网周期拉长、特大变压器交期逼…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "大型电力变压器、取向硅钢及重型燃气轮机长达数年的实物交期与产能极限，构成了2030年前AI算力资本开支周期的终极地缘与物理边界，导致大量先进芯片面临断电闲置风险。", "summary": "截至2026年5月23日，研究证实大型变压器、取向硅钢（GOES）及重型燃气轮机的产能极限与交期瓶颈构成了2030年前AI算力扩张的硬性物理边界。大型变压器交期达36-48个月，GOES高阶钢严重依赖进口且本土产能触顶，重型燃机因单晶叶片与测试台限制交期长达5-7年（排单至2031年后）。定量测算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器等电力设备的极度短缺及取向硅钢供应瓶颈验证了“时间尺度错配”，AI数据中心的扩张将持续受制于3-4年的物理电网硬件交付周期。", "summary": "截至2026年5月22日，全球电力设备市场面临极度紧缺。西门子、GE等四大设备商订单积压创历史新高，大型电力变压器交付周期拉长至36-48个月。上游取向硅钢（GOES）受制于北美产能垄断及全球供应不足，成为扩产的核心瓶颈。设备可用性已取代资金，成为AI基建最硬的约束。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 51, "events": [ { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" }, { "title": "AI 物理电网瓶颈与大宗商品物理期权定价", "summary": "AI 物理电网瓶颈与大宗商品物理期权定价 AI Physical Grid Bottlenecks and Physical Option Pricing of Commodities 随着FERC否决核电并机与PJM并网延迟（5-7年）暴露了AI算力的“电力物理墙”，精炼铜与GOES（取向硅钢）…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 0, "value": 52.4, "events": [ { "title": "高功耗芯片与超高密度机柜使液冷与电力从‘辅助配件’升级为‘物理硬约束’，虽然下游总需求爆发，但应警惕国内低门槛液冷集成环节的价格战与毛利挤压，聚焦高壁垒零部件核心件与变压器出海链的确…", "summary": "AI算力密度飙升驱动物理基础设施“水冷与电力”爆发。HBM4与CoWoS演进（如GB200 NVL72单柜120-132kW）使液冷成为刚需。A股温控龙头英维克2025年营收增32.23%，但2026Q1因价格战净利降81.97%，分化明显。电力设备端金盘科技等受益于出海及AIDC变压器爆发，海外…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A 股电力设备正从‘国内周期标的’重估为‘全球 AI 基建稀缺核心供货商’，其显著的交期优势与估值洼地支撑权重上调。", "summary": "A 股电力设备龙头正凭借 10-12 个月的极致交期（对比海外 36-48 个月）和对取向硅钢 (GOES) 的产能垄断，成为全球 T&D 瓶颈的唯一‘泄洪口’。国家电网 4 万亿（2026-2030）投资计划及 8500 台具身智能设备招标，确立了 A 股‘AI+电力’的增长底部。目前 A 股龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 4, "neutral": 1, "value": 44.8, "events": [ { "title": "I have completed the strategic analysis for Research note 05, building on the physical deli…", "summary": "I have completed the strategic analysis for Research note 05, building on the physical delivery bottlenecks established in the prior cards to evaluat…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI电力设备超级周期成立，但2026-2028年的真实约束在变压器、开关柜和交付认证，而不是泛化的SiC晶圆短缺。", "summary": "2026-05-26：本卡片对卡片03的“10倍机柜功率跃迁”做压力测试。结论是需求判断成立，但交付确定性分层：大型电力变压器、干式变压器、中压开关柜和试验能力是2026-2028年的硬瓶颈；SiC晶圆/器件层面受EV放缓影响并非短缺，但高压模块、封装可靠性、栅极驱动和SST系统认证仍限制放量。金…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网扩容的第二瓶颈不是资本，而是重电设备的合格产能槽位及其上游材料。", "summary": "截至2026-05-27，本卡支持前序判断：AI扩张的第一瓶颈是电网可用容量，第二瓶颈正在转向大型电力变压器、开关设备、电缆和专业施工能力。DOE、Wood Mackenzie和JLL数据显示，LPT进口依赖、GOES/CTC材料短缺、128-143周变压器交付周期、44周开关设备周期和数据中心项…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团", "summary": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团 Global AI Compute Grid interconnections Bottlenecks and Cross-Sector BTM Rotation 伴随变压器制造及重型燃气轮机5-7年交付周期的物理瓶颈，超大规模数据…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 3, "neutral": 0, "value": 49.2, "events": [ { "title": "重型电力设备OEM收入上能承接需求、时间上承接不了——订单簿2.5–4年可见度+定价权+利润率扩张构成对科技久期的反向对冲，核心超配GEV/ENR.DE/ETN/Hitachi系，中…", "summary": "2026-05-28结论：本卡支持第03卡的电网瓶颈论断。Siemens Energy Grid Tech订单/营收比2.28×、储备490亿欧元；GE Vernova Electrification订单+86%、设备储备390亿美元、B2B约2.5×；Eaton电气储备同比+48%、数据中心订单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "把AI久期空头与电网OEM/GOES/CLF多头拼成的组合净Beta+0.45、净久期−1.5年，是'高利率持续'路径下的反AI久期+实物瓶颈租金套利，需配置USD多头FX叠加并设置…", "summary": "2026-05-28 synthesize：把第01–05卡拼回组合（USD 100mm基币）。AI久期空头−15%、AI硬件多+10%、全球电网OEM+30%、亚洲OEM+10%、GOES上游+8%、CLF垂直一体+7%、自动化+5%、现金+25%。净Beta约+0.45、净久期约−1.5年（受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research Summary As of the institute's work date of 2026 05 27 , I have completed the resea…", "summary": "Research Summary As of the institute's work date of 2026 05 27 , I have completed the research card on the electrical equipment supply chain (transfo…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "[auto-handoff] 钢厂厂库逆势反弹与表需季节性回落对二季度固定资产投资（FAI）的宏观指示性评估", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 31e6c7aa-6da2-4c16-8f9d-f73858f40650 结果摘要: Wrote handoff_response.md (15034 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 64, "events": [ { "title": "关键设备交付周期拉长至2.5-3年以上，且受限于GOES与铜的全球结构性短缺，电力设备产业链的产能扩张短期内无法匹配AI数据中心爆发式需求，行业物理瓶颈为设备OEM厂商（GEV、VR…", "summary": "截至2026年5月29日，在AI算力与电网升级需求爆发下，电力设备产业链（GEV、VRT）面临严重物理产能瓶颈。2026年变压器交付周期拉长至128-144周。虽然各大厂商积极在越南和美洲扩张产能，但受限于取向电工钢（GOES）和铜的结构性短缺（GOES在2025年存在约30%供需缺口且将持续至2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 2, "risks": 3, "neutral": 0, "value": 69.4, "events": [ { "title": "2026-05-30的核心判断是：OEM仍是质量锚，但边际alpha在上游材料，存储提供业绩确认，BTM电力运营商提供战略稀缺和低beta稳定器。", "summary": "截至2026-05-30，本卡综合前六张卡：通电算力溢价仍成立，但新增股票风险不应继续单押已被充分发现的 OEM。建议“通电 AI 算力”股票子组合配置为上游材料35%、电力/冷却OEM 25%、存储与HBM 25%、BTM电力运营商15%。风格上，2026年中先以Quality承接已兑现的订单和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "配电侧变压器与开关柜已锁定2026-2028量价齐升超级周期，金盘科技、特变电工、思源电气、中国西电、平高电气是A股最高确定性的物理底座组合。", "summary": "截至2026-05-31，AI数据中心驱动的配电侧设备已进入结构性供应缺口与超级周期：美国大型电力变压器交付周期≈128周(标准)/144周(GSU)，高定制订单达4年；GE Vernova 2026Q1单季为数据中心新增订单$2.4B，超过上一财年全年；MV/LV开关柜lead time普遍14…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "伊顿和施耐德的电力装备产能投产严重阻后于英伟达芯片出货周期，且受制于变压器铁芯原材料取向硅钢（GOES）的物理刚性短缺，成为制约GB200于2026年点亮上线的独立刚性瓶颈。", "summary": "截至2026-05-30，压力测试表明，伊顿与施耐德的产能扩张无法填补因电网延迟导致的GB200部署缺口。两家公司的主要新增产能要到2027年及以后投产，与2026年爆量出货的芯片周期存在严重时钟错配。变压器（交付128周）与开关柜（交付40-60周）积压严重，加之铁芯原材料取向硅钢（GOES）结…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关税不是即时交付杀手，而是电力OEM价格伞；窄牛应聚焦拥有合格交付槽位、本地制造、服务收入和客户预付款的GE Vernova与Siemens Energy等少数公司。", "summary": "2026-05-30，本卡支持Card-06的电力OEM窄牛逻辑：GE Vernova和Siemens Energy的订单、积压订单、客户预付款和Power/Grid利润率已经显示短缺正在转化为价格权。关税没有在最新报表中造成广泛交付断裂，反而提高了合格、本地化、可融资产能槽位的价值；但GOES、…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 5, "neutral": 0, "value": 42.8, "events": [ { "title": "变压器与开关柜瓶颈仍足以支撑ETN/HUBB利润弹性，其中ETN胜在绝对利润池和AI数据中心杠杆，HUBB胜在更纯的公用事业T&D与变电站beta。", "summary": "截至2026-05-31，本卡支持AI电力设备链条：变压器与开关柜交付周期仍处于多季度到多年级别，客户提前锁定工厂槽位，使电气设备龙头拥有价格与积压订单质量。ETN因Electrical Americas规模、数据中心订单和Q1 2026 backlog弹性，适合作为工业主仓；HUBB因Utili…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力高密化与电网瓶颈下的配电设备及温控资本开支压测", "summary": "我们将 Card-03 的 +100 TWh AI 算力用电增量进一步折算为工业制造端订单脉冲：2026 年中国 AI IDC 资本开支增量约 RMB 1,650–2,100 亿元[自测算]，其中配电（变压器/开关柜/母线槽）410–530 亿元、UPS/HVDC/BBU 300–380 亿元、液…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张的下一道硬约束是变压器、开关柜与取向硅钢（GOES），西方OEM新增产能要到2027–2028年才投运、GOES供需要到2030年才能缓解，这是供给侧未来两年的alpha…", "summary": "截至2026-05-31，本卡片支持AI算力瓶颈正从GPU迁移至变压器与开关柜，并在设备层量化约束。Wood Mackenzie 2025 Q2显示标准电力变压器交期128周、GSU 144周；Hitachi Energy欧洲≥100MVA大型电力变压器交期50–58个月。Eaton指定美国DC …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网脱钩不会因为成本本身立刻摧毁云巨头ROI，但会通过交期、折旧和自由现金流错配形成TMT二阶导估值天花板。", "summary": "2026-06-01：本卡以首席风控官视角压力测试前序“AI物理墙”主线。结论是：单纯物理成本通胀不足以击穿云巨头AI Capex ROI；真正危险是GOES/LPT脱钩导致36-60个月变压器交期与两到三年数据中心投运周期错配，使Capex、折旧和可收入化容量之间出现12-24个月缺口。若再叠加…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 0, "value": 68.2, "events": [ { "title": "P/Watt时代AI算力与电力电网设备的实物瓶颈与地理套利", "summary": "P/Watt时代AI算力与电力电网设备的实物瓶颈与地理套利 AI Power Grid Bottlenecks and Geographic Arbitrage in the P/Watt Era 随着AI算力扩建撞上电力电网并网物理速度墙，变压器与硅钢（GOES）结构性缺口高达30%，投产延迟W…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 3, "neutral": 0, "value": 57.6, "events": [ { "title": "美股AI硬件强势对科创50半导体链存在明确补涨逻辑，但仅在先进封装和28nm/14nm主流设备两条主线上'清晰可定价'，应以'订单确认+流动性双重确认'同时为纪律抓手。", "summary": "针对2026-06-03美股AI硬件强势能否传导至科创50半导体链的核心问题，TMT分析师支持卡01战术框架，并把补涨逻辑结构化为可定价的两条主线：(1)先进封装（长电600584/通富002156/华天002185/兴森002436），承接台积电CoWoS 25-30%的产能溢出；(2)28nm…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "重型机械刚性是真实瓶颈，但传导终点是特殊金属与机械资本品的过路费收费站，而不是 LME 大宗——AI 不构成基础工业金属超级周期。", "summary": "本卡对 Card-03 进行压力测试：电力设备交付拉长确实经由重型机械产能（大型自由锻液压机、真空冶金/单晶铸造、特种机床、GOES 精整线）形成多年期刚性瓶颈，这一'制造刚性'判断成立；但'传导至广义工业金属价格结构性上涨'外推不成立——AI/电网仅给铜带来 ~3.5–4% 增量需求，被中国房地…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 06 synthesizes the prior five into a four bucket AI allocation framework (Hyp…", "summary": "Research note 06 synthesizes the prior five into a four bucket AI allocation framework (Hyperscaler ROI core / locked in power / grid OEM / applicati…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 0, "value": 65, "events": [ { "title": "模型能效与边缘计算是面对一阶物理瓶颈（变压器/电网/排队）的二阶效应，叠加推理token爆炸与已锁定capex，'通电滞后'对AI应用端的重定价逻辑在12–18个月内不会被瓦解。", "summary": "TMT否定'能效+边缘'对冲电网瓶颈的反驳。Blackwell单token能效25×提升被推理调用量9× YoY与o系列推理链5,000–50,000 token的爆炸增长抵消，IEA预计数据中心电力需求2024–2030翻倍至945 TWh。瓶颈在LPT/GOES硅钢/并网排队（104–127周…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备厂商的产能扩张弹性及其在AI硬件需求浪潮下的利润承压风险。", "summary": "截至2026-06-05，电网设备厂商扩产金额庞大（Hitachi Energy >90亿美元、Siemens Energy 4.21亿美元夏洛特厂、GE Vernova电气化在手订单翻两番至350亿美元），但物理产能要到2027-2030年才缓解，大型变压器交货期仍约40个月、美国价格涨77-7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 2, "neutral": 0, "value": 66.4, "events": [ { "title": "AI基础设施篮子应从芯片稀缺溢价，转向算力核心、电力设备收入可见度和GOES/铜材料对冲的三层配置。", "summary": "截至2026-06-05，我支持AI基础设施瓶颈从芯片向电力设备、再向GOES/铜迁移的主线，但建议把组合从芯片主导转为“算力核心+设备超配+材料对冲”。建议权重为30%算力/HBM/先进封装，28%电力设备/冷却/微电网，17%铜与GOES等材料，12%合约电力/公用事业，8% API-firs…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "HV 变压器寡头是 2027+ AI-DC 的结构性硬瓶颈与定价权赢家；CDU 厂商的稀缺溢价将自 2027 年起正常化。", "summary": "压力测试 Research note 07：HV 电气与 CDU 并非同质瓶颈。HV 大型电力变压器、OLTC、≥230 kV 套管及上游 Hi-B GOES 硅钢是结构性、十年级硬瓶颈，到 2027 年仍 120–210 周交期；OEM 由 Hitachi Energy、Siemens Ener…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 8, "neutral": 0, "value": 31.8, "events": [ { "title": "AIDC 工业组件铭牌不缺，缺的是经超大规模客户认证的 QD、SiC MOSFET 与 HV-IGBT；alpha 属于这些卡点持有者，而非整流柜或 CDU 组装方。", "summary": "工作日 2026-06-06。本卡压力测试 AIDC 液冷、HVDC 与源网荷储组件供应链。结论：液冷 CDU 与 HVDC 整流柜铭牌产能充足，结构性短缺集中在三处——高可靠快插接头 QD（CPC/Stäubli/Parker 主导，国内三花 002050、盾安 002011 送样未验证）、80…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI能源溢价应从宽基设备链转向现金可兑现的电网/GIS与稀缺核心部件，液冷和AIDC变压器高beta标的必须先通过2026年中现金流复核。", "summary": "截至2026-06-07，本卡对A股变压器、GIS与液冷链条做估值拥挤度和现金质量压力测试。结论是保留AI能源主题，但不再给予宽基设备篮子统一溢价。思源电气仍是质量最高但估值拥挤的复合龙头；华明装备现金质量更干净，适合质量溢价；平高电气配得上温和GIS/电网溢价；金盘科技需等待2026Q2-Q3经…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "工业设备制造业的产能扩张远滞后于AI算力需求的增长，长达3-5年的变压器交货期与取向硅钢短缺构成了AI扩张的硬性物理约束。", "summary": "本报告对AI数据中心关键电力设备供应链进行了压力测试。测试显示大型变压器交货期已达128-160周以上，2026年全美16GW计划产能中近半面临延期。即便电气巨头（伊顿、日立等）大幅扩产，新产能也要到2027-2028年才能释放。受制于取向硅钢（GOES）等上游物料短缺，AI物理基建的“供应墙”确…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Scaling Law 的瓶颈已从芯片配额转向电力设备交付排期，工业产能的刚性约束将迫使全球 AI 竞争进入“存量上电能力”博弈阶段。", "summary": "截至2026-06-07，全球电力设备供应链已进入“结构性稀缺”状态，特高压变压器交付周期飙升至36-60个月，成为AI Scaling Law的绝对物理极限。工业交付与算法迭代的周期错位将ROIC从15%压缩至5%以下。中国凭借12-15个月的交付周期展现出显著的“上电速度”优势，而北美和欧洲受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 3, "neutral": 0, "value": 63.2, "events": [ { "title": "[auto-handoff] 取向硅钢(GOES)产能瓶颈及对特高压变压器造价传导", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 材料行业分析师。 模式: adhoc 会话: e911bea4-9058-473c-a5ef-19ab52df0e02 结果摘要: I will first run the 系统日期锚 command to anchor today's date. …", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网基础设施的物理交付周期（3-4年）构成了 AI 算力部署的刚性硬约束，其影响已超过电费成本，成为决定高能效集群落地的第二物理瓶颈。", "summary": "截至 2026-06-07，电网基础设施（尤其是大型电力变压器）的交付周期已拉长至 160 周以上，成为制约 AI 算力落地的核心物理瓶颈。尽管芯片能效提升降低了电费支出，但工业产能利用率高达 98% 且关键材料取向硅钢 (GOES) 持续短缺，导致 11 GW 的美国数据中心容量因硬件缺失而停滞…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the synthesis on the resilience and architectural evolution of the domesti…", "summary": "I have completed the synthesis on the resilience and architectural evolution of the domestic AI computing ecosystem under advanced packaging supply c…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 1, "value": 79.6, "events": [ { "title": "2027 年电力设备交付瓶颈无法被产能扩张化解，高昂的积压订单饱和度与供应链约束将导致 AI 基础设施在 2027 年面临严重的物理交付墙。", "summary": "压力测试显示，GE Vernova 与西门子能源的扩产计划无法化解 2027 年交付瓶颈。尽管两家公司正在美国和欧洲大幅扩产，但其产能早已被 2024-2025 年签署的创纪录积压订单（GEV $150B，西门子 €154B）预占。燃机和变压器的交付周期分别维持在 5-7 年和 4-5 年的历史高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the kickoff card for the new research thread session as the Chief Economis…", "summary": "I have completed the kickoff card for the new research thread session as the Chief Economist ( chief economist ). The research focuses on the ma...", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 2, "neutral": 1, "value": 75, "events": [ { "title": "先进封装设备封锁将物理锁定中国2027年AI算力供给至基线的50%左右，但大基金三期、算法效率与存量优化的三重对冲使宏观GDP冲击可控在0.2pp内，代价主要体现为AI应用层节奏延后。", "summary": "2026-06-09评估：H2 2026若先进封装设备(TCB/混合键合/HBM堆叠)被纳入FDPR并叠加穿透式所有权审查，中国本土CoWoS等价产能将被锁定在18-22万片/年(12寸等效)，相比无封锁基线缺口45-50%，2027年高端AI加速器交付仅22-28万张(基线45-55万)，等效训…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 基础设施的‘物理主权’已转移至掌握变压器核心组件（GOES、分接开关）与垂直整合能力的工业巨头，2026年的投产节点由‘变压器槽位’而非‘芯片配额’决定。", "summary": "2026年6月9日：本卡通过物理交期压力测试确认，电力变压器（交期3-5年）与开关柜（1.5-2年）已成为AI落地的‘刚性闸门’。工业巨头正从设备商转向‘产能地主’，Eaton通过‘从电网到芯片’的垂直整合与Boyd/Motivair的液冷并购锁定了高溢价份额；Schneider则凭借预制化模块（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与高压开关设备的产能扩张因制造周期、取向硅钢及分接开关的硬约束而严重滞后，无法在2027年AI需求高峰前达产，电力交付瓶颈将成为限制AI算力营收变现的核心硬约束。", "summary": "压力测试确认，变压器及高压开关设备扩产无法在2027年前满足AI基建需求。第一梯队大容量变压器交付期拉长至48-60个月，新订单排至2029年后。Hitachi、Siemens等厂扩产基本于2028年后投产，Eaton琼斯维尔厂2027年投产后仍需爬坡。高牌号取向硅钢（GOES）紧缺与有载分接开关…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 1, "risks": 0, "neutral": 0, "value": 93.4, "events": [ { "title": "All three deliverables written and verified (JSON validates). Research note 04 summary (pol…", "summary": "All three deliverables written and verified (JSON validates). Research note 04 summary (policy analyst, stance: SUPPORT with refinement): I support R…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 0, "value": 80.8, "events": [ { "title": "[auto-handoff] AIDC 电网扩容及变压器交付瓶颈量化", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 3d4e0b90-a5b8-4790-b340-2954e615695f 结果摘要: Wrote handoff_response.md (3819 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 2, "neutral": 0, "value": 76.2, "events": [ { "title": "AI算力并网延迟与GOES变压器物理约束对绿色电力重定价", "summary": "AI算力并网延迟与GOES变压器物理约束对绿色电力重定价 AI Grid Constraints & Re-rating of Green Utilities on GOES/Transformers 取向硅钢(GOES)及高压变压器交期拉长至2.5-4年致使50%的大型算力并网延迟。这是否标志着…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网约束收敛下输配电设备商的毛利率挤压与量价背离", "summary": "AI电网约束收敛下输配电设备商的毛利率挤压与量价背离 Grid Infrastructure Squeeze: GPM Squeeze & Mismatched Pass-Through AI并网延迟与大容量变压器交期拉长下，GOES硅钢与铜的双重价格冲击（GOES+20%/铜+20%）将如何挤压…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 3, "neutral": 0, "value": 71.6, "events": [ { "title": "I have completed the stress test analysis on the risk of export controls and technology dec…", "summary": "I have completed the stress test analysis on the risk of export controls and technology decoupling in the US China grid supply chain for Research not…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力西迁不会造成变压器/GIS全国性短缺，但会把订单兑现风险集中到2026-2027年合格排产、试验和并网验收窗口。", "summary": "截至2026-06-13，本卡压力测试算力西迁对变压器、GIS和上游材料的冲击。结论是：0.32-0.63GW基准西迁负荷不足以触及中国全国性产能上限，但会叠加国网4万亿元“十五五”投资、南网1800亿元2026年投资、新能源并网和646亿元变压器出口，挤占2026-2027年合格交付窗口。取向硅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在并网延迟常态化与变压器/GOES双重瓶颈挤压下，高压设备商稀缺租金向上游硅钢龙头转移，且欧美面临长达230周的极端交付时滞，高估值设备板块Beta承压，Barbell策略建议防守配…", "summary": "截至2026-06-14，质量主管审计了前7张卡片。电网并网排队（4.3年）与变压器交期（120-210周[S5]）是硬瓶颈，以铝代铜使取向硅钢（GOES）消耗量增加15-30%[S5]加剧了缺口。GOES新增供给200-250 kt/年仅能覆盖50-65%需求[S10]，分走设备商30-40%租…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 2, "neutral": 0, "value": 79, "events": [ { "title": "变压器与特高压产业链正在扩产，但全球大型变压器/HVDC/铜链条无法完全匹配AI负荷的近期增速；中国主干特高压可较好匹配，中美差异的核心在本地交付与材料瓶颈。", "summary": "截至2026-06-15，本卡片支持卡片09：AI算力扩容的约束在电网设备与输电交付，而不是火电燃料成本。全球数据中心用电量按IEA基准从2024年约415 TWh升至2030年约945 TWh、约年增15%，但全球变压器市场和数据中心变压器市场预测增速仅约7.24%和6.3%。美国大型变压器存在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施的扩张瓶颈已从芯片转向物理电力设备，燃机与变压器OEM凭借5年以上的积压订单和极强的定价权成为供应链中最大的赢家。", "summary": "本报告深入分析了2026年AI电力供应链中的物理瓶颈。研究发现，燃气轮机和高压变压器OEM已成为供应链的绝对咽喉，GE Vernova和西门子能源的燃机订单已排至2030年以后，大型变压器交期突破150周。瓶颈核心在于取向电工钢（GOES）的结构性短缺，导致2026年全美近50%拟建数据中心产能面…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 0, "value": 74.4, "events": [ { "title": "[auto-handoff] AI电力基础设施瓶颈向上游物理期权（精炼铜/取向硅钢）传导评估", "summary": "致 日报总编： 协同任务已完成。执行方为 主题研究员。 模式: adhoc 会话: bca64921-2ae4-41c5-bfcf-48378f7e6da6 结果摘要: Wrote handoff_response.md (13721 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国电力装备外需足以覆盖行业层面的供应链重构成本，但超额收益将集中在材料、认证和渠道能力最强的头部企业，而非所有出口商。", "summary": "截至2026-06-16，本卡片压力测试全球变压器超级周期下中国电工钢与电力设备出海的利润传导。结论是外需足以支撑行业继续上行，但不能简单抵消所有地缘贸易壁垒；真正受益者是掌握高牌号GOES、具备高压/特高压认证和海外服务能力，并能在必要市场进行最终装配或服务本地化的企业。美国50%钢铝关税和欧盟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建电力设备供应链的物理约束：变压器、燃气轮机与电网扩容的产能极限", "summary": "我支持Research note 04“执行胜过芯片”的方向，但压力测试其两个承重论点。约束并非单一设备，而是串行链条（并网5.2年→高压变压器128周/GSU 144周→开关柜→燃气轮机），最长杆是变压器总装/试验吞吐，而非资本[S1][S2]。GOES是真实但次级的约束：价格较2020年翻倍[…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 3, "neutral": 0, "value": 75.8, "events": [ { "title": "全球变压器和开关柜产能扩张因建厂滞后及取向硅钢（GOES）供应刚性，在2028年前无法缓解AI电力基础设施瓶颈，‘可通电可靠MW’溢价持续高企。", "summary": "截至2026-06-17，主要变压器与开关柜厂商因AI数据中心与电网升级面临历史性订单积存，高压变压器交期拉长至36-60个月。虽然宣布了超百亿美元的产能扩张，但受限于建厂滞后、熟练工短缺及上游取向硅钢（GOES）紧缺，新产能到2028年前无法缓解电力瓶颈，限制了数据中心并网速度。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2029年电网设备扩产会缓解但难以消除交付瓶颈，关键稀缺性正在从变压器OEM装配上移到材料、组件、测试和高压系统集成。", "summary": "截至2026-06-16，本卡压力测试第09卡：变压器和关键电网设备扩产真实，但2026-2029年仍难完全跑赢AI数据中心、可再生能源并网、电网替换和电气化带来的MW/MVA交付需求。2028年前大部分大型电力变压器、HVDC电缆、开关设备和上游组件仍处紧缺。真正的瓶颈不只是OEM装配，而是GO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-16，变压器与GOES供应弹性不足以在2026-2027年全面缓解AI算力落地的物理天花板，只能给已提前锁定OEM产能的头部云厂商提供选择性缓冲。", "summary": "截至2026-06-16，本卡压力测试认为：GOES和铜不是孤立瓶颈，真正约束是GOES、CTC铜导体、制造槽位、专用设备、测试、熟练工和运输共同构成的变压器工业系统。80-210周交期意味着2026年新下单多数无法支撑2027年前通电；Hitachi、Siemens等扩产有助于2027年末以后斜…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 2, "neutral": 0, "value": 83.2, "events": [ { "title": "白皮书主线成立，但因文件归档缺失和后段产能/价差模型证据不足，2026-06-18 不具备最终发布条件。", "summary": "截至2026-06-18，本终审 QA 对 9 张卡片给出“需要修订”评级。主线从 AIDD 算力需求到 GPU/HBM、机柜功率密度、电网设备稀缺、GOES 套利和西方 OEM 产能错配整体连贯，电力与电网瓶颈证据最强。但实时工作区仅存在 card-02、card-04、card-06、card…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球电网核心设备及原材料的产能赤字已构成刚性物理准入壁垒，导致产业利润池从算力/晶圆厂资本支出方结构性向重电OEM厂商转移。", "summary": "大功率汽轮机与超高压变压器的严重产能缺口（全球超高压变压器短缺约30%，交期延至2-4年；大型汽轮机交期延至5-7年）和取向硅钢（GOES）等关键材料瓶颈，对算力中心和晶圆厂构成不可逾越的物理准入壁垒。议价权转移令电力设备OEM（如HD Hyundai Electric和GE Vernova）利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 1, "neutral": 1, "value": 93.6, "events": [ { "title": "燃机及关键电气设备供应链的交付压力与产能弹性评估", "summary": "截至2026-06-18，本卡压力测试燃机与电气设备供应链。三巨头GE Vernova、Siemens Energy、MHI占全球在建燃机产能约三分之二，2026年起年提产25–35%，但扩产仅抵消积压增速而非积压本身：GE Vernova积压约83GW、产能20→24GW/年，预留售罄至2030…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 国内AIDC表后自备发电合规与经济边界", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 工业制造分析师。 模式: adhoc 会话: 54eae79c-6253-4049-81b6-f93fd057ed2a 结果摘要: Wrote handoff_response.md (11240 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 1, "neutral": 1, "value": 95, "events": [ { "title": "[auto-handoff] AIDC电力基建逻辑下高压变压器与输电网铜需求增量的定量折算", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b7159f5c-eb1d-48e7-8538-01f5bfef96a5 结果摘要: Wrote handoff_response.md (10534 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电网稀缺已经从主题交易升级为大类资产配置因子：低配无法按时通电的GPU/CPO硬件需求，超配持有变压器、GOES、中国UHV和红利电力现金流的资产。", "summary": "截至2026-06-19，本卡片把前序“GPU/CPO减配、电网链上修、GOES/变压器瓶颈、美国可再生递延、中国UHV免疫”整合成100单位跨资产组合：18%中国UHV/电网自动化、14%一体化变压器与GOES、16%红利电力、6%铜、18%中国久期、7%黄金、9%现金、6%硬件beta、2%美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 2, "neutral": 1, "value": 90.4, "events": [ { "title": "电网关键设备供应瓶颈：变压器与开关设备的交付压力", "summary": "本卡支持卡片02的核心结论——美国AI电气化的真正约束已从发电下移至电网设备（变压器、开关、接入）。从工业供给侧补充：该瓶颈正重塑GE Vernova与Eaton的订单簿与资本开支。GEV电气化设备订单储备升至约390亿美元（同比+75%），数据中心当季订单约24亿美元（超2025全年），并以53…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "工业电气设备产能扩充与供应链交付压力", "summary": "压力测试 Research note 02–03 的电气设备瓶颈论。瓶颈在 2026–2028 年对定制大型变压器、HVDC、中压开关柜与并网真实成立，但稀缺溢价被误当作永久性。三点修正：一、积压创纪录（Eaton 总积压逾190亿美元、数据中心积压228GW≈12年需求；GE Vernova电气…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I will start by checking the current files in the workspace (if any) and confirming the dir…", "summary": "I will start by checking the current files in the workspace (if any) and confirming the directory layout. I will perform a web search to ch...", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 1, "neutral": 0, "value": 94.8, "events": [ { "title": "电力设备（如变压器）供应链瓶颈对AI基础设施建设节奏的影响", "summary": "压力测试结论：是，但有两点修正。对美欧并网型AI建设，电力设备2026年当下即是约束边界且早于芯片——约一半美国规划数据中心已延期/取消，12GW中仅约5GW在建[S4][S9]，而CoWoS仍在近4倍扩产出货[Card05]，故‘买得到却通不了电’的部署约束当下即活跃。修正一：约束不是变压器箱体…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 0, "value": 90.2, "events": [ { "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 4cd2d717-9d04-4569-b0d8-cc37d743d0a9 结果摘要: Wrote handoff_response.md (9414 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b04fea49-07a7-43ca-b246-32f00d158f50 结果摘要: Wrote handoff_response.md (9863 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 4, "neutral": 0, "value": 79.6, "events": [ { "title": "配电网滞后与变压器短缺连同HBM/CoWoS确实构成AI capex的物理硬约束——对边际开发商是硬墙，对优先级买家是ROIC压缩，从物理层面支持AI-capex去杠杆而非2026崩…", "summary": "截至2026-06-23，本卡从公用事业侧确认：配电网扩容滞后与变压器短缺是AI capex物理钳制的第二片钳口，与HBM/CoWoS硅片分配叠加。变压器交期达128–144周乃至4年，电力变压器缺口30%、配电10%，GOES自2020年翻倍、铜+50%；LBNL并网队列约2,060 GW、中位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备供应链扩产进度能否扭转2027年交期约束", "summary": "压力测试'扩产能救2027'的乐观读法：GE Vernova(全资Prolec GE，US$5.275bn)、Eaton(Jonesville US$340m+Nacogdoches US$100m+Waukesha US$22m)、Hitachi Energy(逾US$1bn含South Bos…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "GOES是变压器材料链最干净的二阶导，铜是战略性长期beta但短期受宏观与库存周期扰动更大。", "summary": "截至2026-06-24，本卡支持前序“AI算力瓶颈转向电力”的主线，但将材料排序进一步收窄：GOES/高端电工钢是更干净的变压器产出卡点，因其技术认证、美国本土单一生产商、价格接近翻倍与扩产至2027-2030年才释放；铜是更大、更流动的长期电气化金属，但ICSG显示2025与2026年1-3月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建配置应从“追逐算力设备增量”切换为“持有可交付电力稀缺性”，但保留CPO核心仓位，因为CPO仍是降低AI网络功耗的必要技术。", "summary": "截至2026-06-24，本卡综合前序结论：AI算力瓶颈已从GPU转向已通电MW，组合应从高拥挤CPO单边beta转向“变压器OEM+中国出海设备+GOES/铜材料+CPO核心赢家”的哑铃。100单位AI基建权益袖基准配置为韩日及全球变压器OEM 30、中国出海设备22、上游材料18、CPO/光模…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 2, "neutral": 2, "value": 96, "events": [ { "title": "科技板块风格再校准：能效赋能方的稀缺溢价与高耗电算力股的吞吐折价", "summary": "本卡把前五卡收束为可执行配置框架。算力以‘每兆瓦有效算力’计价后，权益市场沿‘谁拥有稀缺投入、谁的回报被它锁死’重估。折价已现于现金流：2026超大规模厂商资本开支约6900–7250亿美元、约75%投向AI，却吞噬近100%经营性现金流；微软约800亿美元Azure积压受限于电力。溢价已现于盘面…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Your Role: 工业制造分析师 (Industrials Analyst) Expertise: Advanced manufacturing, defense, machin…", "summary": "Your Role: 工业制造分析师 (Industrials Analyst) Expertise: Advanced manufacturing, defense, machinery, aerospace Step −1 — Anchor today's date (BLOCKING — r…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "关键电力设备（变压器/开关/UPS）的交付周期与产能约束评估", "summary": "压力测试卡片06的“设备交付为约束”命题。结论基本否定但附两条限定：就中国境内而言，设备总产能与硅钢/铜大宗总量都不是约束——TBEA一家约260 GVA/年，远超卡片06测算的AI增量约16.6 GVA/年（仅约6.4%）；中国2024年取向硅钢产量303万吨（日5倍、美8倍）结构性过剩且净出口…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器、开关设备、取向电工钢、电缆和电网设备制造产能", "summary": "支持卡片05/06：油价回落是数据中心电力运营成本的缓冲，而非解锁电网设备产能的钥匙。从制造现场看，资本已在大举投入——Hitachi Energy（60亿美元集团计划含约15亿专用于变压器、2027年放量）、Siemens Energy（4.21亿美元Charlotte厂2027年竣工）、GE …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 2, "neutral": 1, "value": 96, "events": [ { "title": "Summary of Work I have completed the research and stress test on the transmission effects o…", "summary": "Summary of Work I have completed the research and stress test on the transmission effects of grid equipment bottlenecks on upstream key raw materials…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球电力设备制造供应链产能与上游材料供应风险评估", "summary": "部分确认Card-03的time-to-power论点，但作三处修正。瓶颈确在上移到材料端，但对变压器而言真正的物理咽喉是GOES而非铜，且GOES稀缺本质是“产能+贸易”稀缺而非地质稀缺：全球变压器产能利用率98%、GOES自2020年涨约100%、美国仅Cleveland-Cliffs一家、俄…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心的卡脖子资产已从芯片变为高压电气设备，交期与产能缺口把电气推成最大且持续涨价的Capex板块，稀缺溢价持久归于变压器/特高压/电气配套，而非火电或燃气beta。", "summary": "截至2026-06-25，本卡支持并加固Card-01判断：AI数据中心的约束已从芯片转向电气并网设备。核查结果——大型变压器交期约128周、GSU约144周、中压开关柜52–78周，均超数据中心12–18个月建设周期；GSU需求2019–2025增274%，美国变压器/配电变压器2025供给缺口…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 7, "neutral": 0, "value": 65.8, "events": [ { "title": "并网延迟的变压器库存积压是可逆的时点拖累(金盘-8%~-11%、思源<-3%)，被国网4万亿内需托底，应在超配出口变压器框架下偏思源、轻金盘并叠加配对与公用事业杠铃对冲。", "summary": "并网延迟导致的已完工未交付变压器库存积压，是营运资金/收入确认的时点事件而非需求毁灭——订单仍排产至2027年在手backlog内。对金盘科技(688676.SH)拖累真实但有界：北美AIDC相关收入是最高Beta切片，Q2/Q3净利约-8%至-11%、伴随存货与FCF占用，属逢低布局而非证伪；对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 重电设备出海出厂与已完工未交付库存波动风险评估", "summary": "致 主题研究员： 协同任务已完成。执行方为 工业制造分析师。 模式: adhoc 会话: 63433a08-eaed-4bc7-99a4-1c548d3d2ec3 结果摘要: Wrote handoff_response.md (8206 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海外电网升级大周期与原材料成本冲击下的电网设备出口毛利弹性测试", "summary": "本卡综合集成并收紧 Card-05 的电网设备多头：在 LME 铜价高于 10,000 美元/吨、GOES 偏紧背景下，A股电网设备出口毛利弹性有韧性而非脆弱。自测算敏感性显示铜与 GOES 同时 +20% 时，固定价订单毛利收缩约 5.3pp，但 80% MEC 传导下仅约 1.1pp[自测算]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 2, "neutral": 1, "value": 96, "events": [ { "title": "功率半导体已过剩，AI电力的新物理瓶颈已迁移至大型变压器，并真正卡在无法被资本快速绕行的上游电磁钢(GOES)与铜——2026–2028窗口难解。", "summary": "支持并收紧前五卡逻辑链：功率半导体转入结构性过剩后（Research note 05），AI电力建设的刚性瓶颈已迁移至大型变压器(LPT)，并下沉到上游电磁钢(GOES)与铜。瓶颈呈两层：整机厂总装层可用约20亿美元北美扩产去瓶颈，但缓解要到2028/2029年后；标准电力变压器交期128周、GS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "滞胀确认与大类资产重构：商品通胀刚性、美联储“加息在桌”对大类资产风险平价模型（Risk Parity）的极限压力测试。", "summary": "本卡支持第01–07卡逻辑链并闭环回组合构建。供给冲击型通胀给铜（2026年缺口约33万吨，2030年扩至150–220万吨）与被贸易墙隔离的GOES（中国约56%产能）一个硬性供给底，使商品子仓下行半方差较以往周期压缩约30–40%——保留通胀上行、削弱增长下行。但这救不了风险平价：供给冲击把股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Summary of Work We have completed the analysis for Research note 04/10 as the Materials Ana…", "summary": "Summary of Work We have completed the analysis for Research note 04/10 as the Materials Analyst (材料行业分析师) , adopting a Support (支撑) stance. Our repor…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 17, "neutral": 0, "value": 12, "events": [ { "title": "AI电力基础设施供应链瓶颈：电力变压器与关键配电硬件的产能缺口", "summary": "对Research note 01以变压器为中心的瓶颈论断做压力测试。结论：GIS与高压断路器产能确实紧张（≥145 kV断路器交期已超12个月、整套GIS需60–100周、供给集中于约4家OEM），但它是次级协同瓶颈，而非近期核电并网的绑定约束。原因在于结构：近期核电对AI的供给主力是反应堆重启…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器核心原材料（如硅钢片）的供应紧缺和贸易壁垒所导致的交付周期拉长，确实大幅增加了公用事业电网升级的投资延迟风险，并将物理约束转化为公用事业的ROIC损失和出口商的经营压力。", "summary": "截至2026年6月28日，取向硅钢(GOES)和电解铜的供应紧张与贸易壁垒使变压器交期拉长至24-48个月。合规转口硅钢到岸成本溢价达72.68%。这直接导致电网升级投资延期，造成高达194.2亿美元的闲置资金损失，拉低公用事业现金ROIC约35-75bp，并使金盘科技等出口商出现交付递延与存货积…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "原始40/32/12/8/8 sleeve在AI需求相关性趋近+1时尾部风险过高，阶段1是强制去杠杆，阶段2才接近可接受，但决定性保护来自降AI beta和提高现金，而非小仓位黄金/…", "summary": "截至2026-06-29，本卡对40/32/12/8/8 AI物理瓶颈sleeve做正式压力测试。原始组合在1个月99% Student-t口径下VaR/CVaR为28.6%/37.8%[自测算]，阶段1降至23.2%/30.7%[自测算]，阶段2降至18.5%/24.5%[自测算]。结论是：原始…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼及普钢加工环节将面临严重的利润率…", "summary": "本卡通过技术壁垒与自给率对A股AI原材料敞口进行了区分：高端取向硅钢方面，宝钢股份凭借规模与技术优势展现出强Alpha属性，而首钢股份受普钢周期Beta拖累较大；有色铜方面，在2026年长单TC/RC为零的极端低位下，矿权自给率超65%的紫金矿业可完全捕获矿端溢价（EBITDA价格敏感度约7.5%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 9, "neutral": 0, "value": 58, "events": [ { "title": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "summary": "本卡片从制造和供应链视角对电网与电力约束进行压力测试。截至2026-06-29，关键电力设备（变压器与重型燃机）由于极高工艺与测试门槛、上游取向电工钢（GOES）瓶颈及寡头垄断，交期拉长至3-7年，生产机位已售罄至2029-2030年。云厂商BTM发电加剧了燃机机位争夺。预计2026-2027年该…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "summary": "本卡压力测试 Research note 6 的'调价条款已隔离利润'结论，认为其仅有条件成立。主流调价条款（MEC）挂钩有流动期货的铜/铝（出口超级周期转嫁率约80%[S1]），却普遍排除或封顶无流动指数的取向电工钢GOES——而GOES占COGS约28%[Research note 6]且处2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15.36倍AI潜在电力需求，算力扩…", "summary": "截至2026-06-30，本研究对Jevons悖论激发的15.36倍潜在AI用电需求（相当于727.6 GW新增平均负荷或6,374 TWh年用电）进行了极限物理压力测试。结论表明，在2030年前，由于北美变压器交期拉长至48-60个月、特高压直流送电网架高壁垒（全球需新建73条通道）以及取向硅钢…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-21dd72c1b1/", "networkHref": "/investor/research-network/#thesis-21dd72c1b1" }, { "id": "thesis-66eb42380e", "source": "tracked_thesis", "title": "海洋经济的政策催化需要落到船舶订单、海工装备与海上资源现金流", "titleEn": "Marine-economy policy support must convert into ship orders, offshore equipment, and offshore-resource cash flow", "lane": "海洋经济/船舶海工", "laneEn": "Marine economy, shipbuilding, and offshore equipment", "laneHref": "/investor/lanes/marine-economy-shipbuilding-and-offshore-equipment/", "investableFocus": "海洋经济/船舶海工：600150.SH, 601989.SH, 601919.SH, 600938.SH, 0883.HK, 600522.SH, 300750.SZ", "stockUniverse": [ { "ticker": "600150.SH", "name": "中国船舶", "market": "A-share", "role": "造船主线", "roleEn": "造船主线", "bucket": "core", "href": "/investor/stocks/600150.SH/" }, { "ticker": "601989.SH", "name": "中国重工", "market": "A-share", "role": "船舶与海工", "roleEn": "船舶与海工", "bucket": "core", "href": "/investor/stocks/601989.SH/" }, { "ticker": "601919.SH", "name": "中远海控", "market": "A-share", "role": "集运与港口链", "roleEn": "集运与港口链", "bucket": "core", "href": "/investor/stocks/601919.SH/" }, { "ticker": "600938.SH", "name": "中国海油", "market": "A-share", "role": "海上油气现金流", "roleEn": "海上油气现金流", "bucket": "core", "href": "/investor/stocks/600938.SH/" }, { "ticker": "0883.HK", "name": "中国海洋石油", "market": "HK", "role": "海上油气", "roleEn": "海上油气", "bucket": "core", "href": "/investor/stocks/0883.HK/" }, { "ticker": "600522.SH", "name": "中天科技", "market": "A-share", "role": "海缆与海风", "roleEn": "海缆与海风", "bucket": "core", "href": "/investor/stocks/600522.SH/" }, { "ticker": "300750.SZ", "name": "宁德时代", "market": "A-share", "role": "海上储能/电池外溢", "roleEn": "海上储能/电池外溢", "bucket": "core", "href": "/investor/stocks/300750.SZ/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" } ], "stockTickers": [ "600150.SH", "601989.SH", "601919.SH", "600938.SH", "0883.HK", "600522.SH", "300750.SZ", "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH" ], "coreView": "赛道：海洋经济/船舶海工。候选观察标的：600150.SH, 601989.SH, 601919.SH, 600938.SH, 0883.HK, 600522.SH, 300750.SZ。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 工业供给瓶颈。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 72, "eventCount": 160, "firstSeen": "2026-05-04", "lastSeen": "2026-06-29", "supports": 6, "risks": 139, "neutral": 15, "analystCount": 16, "sourceTypeCounts": { "研究笔记": 143, "分析协作": 9, "证据更新": 1, "研究主题": 7 }, "topTerms": [ "GOES", "并网", "AIDC", "石油", "储能", "电池", "Marine-economy", "policy", "support", "must", "convert", "ship" ], "whatChanged": [ "2026-06-29: 取向电工钢（GOES）由于美国本土单一垄断和地缘壁垒呈现严重的全球双轨制定价，高牌号硅钢2-4年的扩产认证周期和铜价高企构成了无法通过资本抹平的物理瓶颈，将通过毛利侵蚀与交期恶化对变压器及算力并网构成二次约束。", "2026-06-29: AI算力大基地西部迁移与绿电直供（BTM）模式下的电网费率重构与容量补偿收益测试", "2026-06-29: 2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "2026-06-29: 并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "2026-06-29: 超长期特别国债与地方基金预算受阻的非对称限制，导致西部智算基地无法依靠补贴长期消化绿电LCOE折价收窄至13.89%与并网延迟造成的40%度电成本上升[S3][自测算]；物理并网时滞以非线性传导迫使服务器厂商在12.5%的出货缺口下承受-30%的净利润修正常态化[自测算]，…" ], "practical": { "sectorLane": "海洋经济/船舶海工", "sectorLaneEn": "Marine economy, shipbuilding, and offshore equipment", "exposureBucket": "电力与公用事业", "tone": "yellow", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "600150.SH", "name": "中国船舶", "market": "A-share", "role": "造船主线", "roleEn": "造船主线", "bucket": "core", "href": "/investor/stocks/600150.SH/" }, { "ticker": "601989.SH", "name": "中国重工", "market": "A-share", "role": "船舶与海工", "roleEn": "船舶与海工", "bucket": "core", "href": "/investor/stocks/601989.SH/" }, { "ticker": "601919.SH", "name": "中远海控", "market": "A-share", "role": "集运与港口链", "roleEn": "集运与港口链", "bucket": "core", "href": "/investor/stocks/601919.SH/" }, { "ticker": "600938.SH", "name": "中国海油", "market": "A-share", "role": "海上油气现金流", "roleEn": "海上油气现金流", "bucket": "core", "href": "/investor/stocks/600938.SH/" }, { "ticker": "0883.HK", "name": "中国海洋石油", "market": "HK", "role": "海上油气", "roleEn": "海上油气", "bucket": "core", "href": "/investor/stocks/0883.HK/" }, { "ticker": "600522.SH", "name": "中天科技", "market": "A-share", "role": "海缆与海风", "roleEn": "海缆与海风", "bucket": "core", "href": "/investor/stocks/600522.SH/" }, { "ticker": "300750.SZ", "name": "宁德时代", "market": "A-share", "role": "海上储能/电池外溢", "roleEn": "海上储能/电池外溢", "bucket": "core", "href": "/investor/stocks/300750.SZ/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" } ], "stockTickers": [ "600150.SH", "601989.SH", "601919.SH", "600938.SH", "0883.HK", "600522.SH", "300750.SZ", "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH" ], "primaryStocks": [ { "ticker": "600150.SH", "name": "中国船舶", "market": "A-share", "role": "造船主线", "roleEn": "造船主线", "bucket": "core", "href": "/investor/stocks/600150.SH/" }, { "ticker": "601989.SH", "name": "中国重工", "market": "A-share", "role": "船舶与海工", "roleEn": "船舶与海工", "bucket": "core", "href": "/investor/stocks/601989.SH/" }, { "ticker": "601919.SH", "name": "中远海控", "market": "A-share", "role": "集运与港口链", "roleEn": "集运与港口链", "bucket": "core", "href": "/investor/stocks/601919.SH/" }, { "ticker": "600938.SH", "name": "中国海油", "market": "A-share", "role": "海上油气现金流", "roleEn": "海上油气现金流", "bucket": "core", "href": "/investor/stocks/600938.SH/" }, { "ticker": "0883.HK", "name": "中国海洋石油", "market": "HK", "role": "海上油气", "roleEn": "海上油气", "bucket": "core", "href": "/investor/stocks/0883.HK/" }, { "ticker": "600522.SH", "name": "中天科技", "market": "A-share", "role": "海缆与海风", "roleEn": "海缆与海风", "bucket": "core", "href": "/investor/stocks/600522.SH/" } ], "instrumentTags": [ "600150.SH", "中国船舶", "601989.SH", "中国重工", "601919.SH", "中远海控", "600938.SH", "中国海油", "0883.HK", "中国海洋石油", "600522.SH", "中天科技" ], "reviewCadence": "每日复核", "confirmationSignals": [ "并网排队和变压器交期是否继续拉长", "表后自发电与 PPA 是否改变公用事业信用指标", "电力设备订单质量是否高于收入确认节奏" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "长端利率", "铜/GOES 成本", "监管电价", "信用利差" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 118 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 16, "recentEvents": 118, "supportRatio": 0.04, "riskRatio": 0.87, "conflictScore": 97.8, "evidenceDepth": 100 }, "score": 63.8 }, "events": [ { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 21.5, "matchedTerms": [ "并网", "AIDC", "储能", "电池" ], "href": "" }, { "date": "2026-06-29", "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power预期缺口，会迫使Hyperscaler预付并网升级、锁定变压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17.75, "matchedTerms": [ "并网", "储能", "电池", "capex" ], "href": "" }, { "date": "2026-06-29", "title": "超长期特别国债与地方基金预算受阻的非对称限制，导致西部智算基地无法依靠补贴长期消化绿电LCOE折价收窄至13.89%与并网延迟造成的40%度电成本上升[S3][自测算]；物理并网时滞以非线性…", "summary": "截至2026-06-30，本报告系统性合成AI算力电力约束与中国中央融资及地方财政的矛盾。超长期特别国债仅能用于机房CapEx，无法支付占OpEx 60%以上的日常电费[S6]。由于房地产下行拖累地方基金预算[S3]，地方电价补贴难以为继。西部直连绿电并配建储能的LCOE高达0.362元/kWh，相比煤电零售价的折扣缩窄至13.89%[自测算]，且内蒙古折…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "并网", "储能", "capex" ], "href": "" }, { "date": "2026-06-29", "title": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "summary": "支持第06卡\"并网延迟→利润压缩\"的传导论点，但将单向结论细化为分化。针对11GW数据中心建设缺口与12-18GW刚性局部电网缺口，48-60个月并网延迟通过两条通道传导：(1)经订单/PPA递延与尾款拖延压制发电子链开工率——5.7%销量递延经经营杠杆放大为约-21%利润压缩[自测算]；(2)将电网投资节奏重置向T&D侧，5万亿元资本开支周期以70-75…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 15.5, "matchedTerms": [ "并网", "储能" ], "href": "" }, { "date": "2026-06-29", "title": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "summary": "对中国，2026年油价/柴油冲击既是被缓冲的进口税、又是战略加速器，且净额上比卡片01–02的美欧日集团损害更小。三大事实：（1）进口税真实但落在PPI而非CPI——5月PPI同比+3.9%（2022年7月来最快）、CPI仅+1.2%，疲弱内需把冲击吸收为生产者利润压缩；（2）进口税被三重缓冲——全球最大约14亿桶战略储备、作为受制裁油折价买方、疲弱需求；…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "石油", "电池" ], "href": "" }, { "date": "2026-06-29", "title": "AI算力大基地西部迁移与绿电直供（BTM）模式下的电网费率重构与容量补偿收益测试", "summary": "压力测试卡片04–05隐含假设：AI数据中心西迁＋BTM绿电直供＋大容量储能并非IPP的毛利率顺风，电网也非纯输家。一，BTM对IPP是DC的成本叙事而非毛利扩张：并网型绿电直连仍须全额缴纳输配电费与交叉补贴（发改能源650/688号不得减免），强制配储是资本开支拖累且补偿退坡（内蒙古0.35→0.28元/千瓦时），现货市场化侵蚀绿电溢价，综合毛利率被封顶…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "并网", "储能" ], "href": "" }, { "date": "2026-06-29", "title": "取向电工钢（GOES）由于美国本土单一垄断和地缘壁垒呈现严重的全球双轨制定价，高牌号硅钢2-4年的扩产认证周期和铜价高企构成了无法通过资本抹平的物理瓶颈，将通过毛利侵蚀与交期恶化对变压器及算…", "summary": "截至2026-06-29，评估表明全球取向硅钢（GOES）瓶颈正恶化变压器交期。美本土由Cleveland-Cliffs垄断，价格较中国溢价416%，呈双轨制。高牌号硅钢扩产需2-4年。测算显示，+30% GOES与+20%铜的冲击将使变压器GPM从25%降至13.45%，严重反噬OEM毛利。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GOES", "并网" ], "href": "" }, { "date": "2026-06-28", "title": "海上风电+UDC的24/7低碳电力合同与并网可行性", "summary": "可融资的24/7低碳UDC电力可实现，但只能作为四层结构：成型/固化的多年期PPA（叠加机制电价差价结算CfD）+ 偏LDES的多小时现场储能 + 有界偏差结算带 + 电网耦合并网（源网荷储或电网+逐时T-EAC）。否定纯海风直连孤岛——它移除了固化小时数所需的电网/储能平衡，24/7现金流不可投。可靠性具非线性：不可调度24/7 CFE约€160–240…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 20.5, "matchedTerms": [ "并网", "储能", "电池" ], "href": "" }, { "date": "2026-06-28", "title": "由于储能与动力电池的绝对规模错配以及中国储能并网瓶颈和矿山复产传闻，储能需求增长在第三季度不足以支撑广期所碳酸锂期货价格企稳。", "summary": "本报告定量评估了Fastmarkets会议上关于“储能增长抵消电动汽车放缓”以稳定锂价的观点。分析表明，储能消费规模过小，无法对冲动力电池放缓的影响；且国内并网瓶颈给储能实际装机量带来下行风险（可能造成32万吨LCE的需求毁灭）；结合宁德时代枧下窝矿山复产预期和广期所国际化引入抛压，预计Q3碳酸锂价格将继续承压下行。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 19.5, "matchedTerms": [ "并网", "储能", "电池" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力基础设施供应链瓶颈：电力变压器与关键配电硬件的产能缺口", "summary": "对Research note 01以变压器为中心的瓶颈论断做压力测试。结论：GIS与高压断路器产能确实紧张（≥145 kV断路器交期已超12个月、整套GIS需60–100周、供给集中于约4家OEM），但它是次级协同瓶颈，而非近期核电并网的绑定约束。原因在于结构：近期核电对AI的供给主力是反应堆重启与增容，二者均复用既有开关站，断路器/GIS短缺在发电侧几乎…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 18, "matchedTerms": [ "GOES", "并网", "industrial" ], "href": "" }, { "date": "2026-06-28", "title": "在2026-06-29的政策环境下，通过将绿电绿证的环境资产价值（抵消范畴二排放以规避每度电约0.41元的碳税/碳关税）与电力系统调节成本（114号文储能及火电容量电费分摊）进行统筹对冲，沿…", "summary": "截至2026-06-29，本报告支持卡片08并深化分析。发改价格〔2026〕114号文落地后，将火电与新型储能容量电费摊销纳入系统运行费，由工商业用户按用电比例共同分摊，增加了跨省区绿电采购的隐性通道成本。然而，绿电绿证（GEC）的‘证电合一’交易与碳足迹核算为沿海UDC开辟了独特的套利空间：一个标准的24MW UDC模块年耗电2.31亿度，可通过绿证抵消…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 17.5, "matchedTerms": [ "并网", "储能", "thematic" ], "href": "" }, { "date": "2026-06-28", "title": "变压器交期拉长至3-5年与并网排队超36个月的物理刚性瓶颈，配合美欧贸易保护关税对GOES硅钢供应链的阻断，已对AI算力扩张的通电效率构成无法用资本单纯解决的长期结构性壁垒，造成明显的Cap…", "summary": "基于并网排队与变压器短缺的双重物理约束，我们验证电力基础设施瓶颈已构成AI算力扩张的长期结构性阻碍。数据中心通电时间已拉长至36-60个月，变压器交期延至3-5年，主要厂商积压订单超百亿美元。这些延误对项目NPV和IRR产生重大侵蚀，100MW数据中心延误3年将损失51.63%的NPV。美国Section 301/232关税与欧盟CBAM机制则通过GOES…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "GOES", "并网", "capex" ], "href": "" }, { "date": "2026-06-28", "title": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电力’，并驱动策略从算力规模向算法效率和电…", "summary": "截至2026-06-28，本卡片通过对电网与变压器交期的压力测试，确认电网并网（5-7年交期）和变压器交付（2-4年交期）构成了AI算力扩张的终极物理边界。随着GB300（约135kW）及Rubin Kyber（约600kW）对800VDC和固态变压器（SST）的高密度配电需求增加，目前仅有不足8%的设施就绪。FERC否决核电表后直供，迫使所有新增容量重回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "GOES", "并网", "capex" ], "href": "" }, { "date": "2026-06-28", "title": "表后共址数据中心引发的规避网费及系统容量价格暴涨，迫使FERC与地方公用事业启动资费重构与高额抵押担保以防范信用降级，未来需通过双向费率平衡可靠性成本分摊与灵活调节信用。", "summary": "截至2026-06-28，本卡片对BTM数据中心并网及规管成本分摊进行系统性梳理。云厂商利用BTM直连核电/气电旁路输配网费引发了剧烈的成本转嫁争议，导致PJM容量出清价暴涨至333.44美元/MW-日，且有40%（65亿美元）由数据中心预测负荷驱动。监管机构正加速资费纠偏，如FERC于6月18日发布Show-Cause命令、弗吉尼亚GS-5要求每兆瓦15…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "并网", "储能" ], "href": "" }, { "date": "2026-06-28", "title": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "summary": "本卡压力测试逻辑链：GOES约束不会「抵消」核能加速，因为二者作用于电力链不同节点——核能重启在发电/并网侧本就是GOES-轻（复用既有GSU与开关站），属于一种GOES套利；但GOES在负荷侧与输配路径重新绑定，沿用Research note 01的NPV模型测算，核能锚定型项目(≈1.5年延迟,NPV≈−27%)相对并网型绿地(3年,−51.6%)只回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GOES", "并网" ], "href": "" }, { "date": "2026-06-27", "title": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞", "summary": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞 DeepSeek's 96.37% price cut: Jevons amplification or AI hardware demand destruction — head-to-head with the grid/transformer…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 16, "matchedTerms": [ "并网", "thesis", "industrial", "thematic", "grid" ], "href": "" } ], "timeline": [ { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "长时储能（LDES）供应链分析：矿物弹性与国产化突破（2026 2027） 分析师： materials analyst [材料行业分析师] 日期： 2026年5月4日 主题： 西部…", "summary": "长时储能（LDES）供应链分析：矿物弹性与国产化突破（2026 2027） 分析师： materials analyst [材料行业分析师] 日期： 2026年5月4日 主题： 西部长时储能供应链约束与国产替代瓶颈 立场： 支持 (Support) 执行摘要 截至2026年，中国长时储能（LDES…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 3, "neutral": 0, "value": 25.4, "events": [ { "title": "并网 interconnection 和 PJM-East 气源 take-away 才是 2026–2028 把 OEM backlog 兑现成可用 MW 的真正硬约束，OEM s…", "summary": "压力测试 OEM backlog 能否兑现成可用 MW。结论：2026–2028 binding 约束按严重度依次为电网 interconnection（PJM/MISO 5–7 年排队）、PJM-East 燃气支线/干线、空气许可（Loudoun/DFW 等臭氧非达标区）、燃机 OEM slot…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电池制造产能过剩的结构性黏性意味着中游ROIC触底比共识预期晚18–24个月，而电网侧储能需求是前述分析框架中被系统性遗漏的需求通路，其规模足以实质性地支撑LFP主导型一级供应商的产…", "summary": "工业制造分析师对前四张卡片综合框架进行压力测试，发现四个具体脆弱点并补充两条需求通路。核心发现：（1）光伏类比低估电池制造技术黏性，中游去产能周期为36–48个月而非24–30个月；（2）资本开支锁定效应造成机械性ROIC底部，中游材料再估值入场早18–24个月而非6–12个月；（3）CATL欧洲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "购进价63.7与碳规则收紧不会击穿全部材料企业，但会把非一体化高炉钢、煤电铝与水电铝、特钢、集成电池材料链清晰分层，强化“复苏真实但偏窄”的十卡主线。", "summary": "另类数据综合支持第09卡的材料毛利压力测试，但修正价格楔形：4月官方购进价63.7、出厂价55.1，对应8.6点投入-产出差，而非此前估计的12点。结论仍是分化而非全面崩塌：非一体化高炉钢和煤电铝是结构性毛利输家；水电铝、废钢/特钢和一体化LFP/储能链条是周期性幸存者；上游锂加工在碳酸锂反弹下可…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "技术深度是中国电网设备出海的核心护城河，但在美国AI数据中心场景中，第三国制造和退税只能提供部分对冲，真正决定订单能否进入关键供电路径的是可信交付、网络安全、认证和本地服务能力。", "summary": "截至2026-05-06，本卡对第8张卡片作有条件压力测试：技术深度确实比地理套利更重要，但第8张卡片需修正三点。第一，2024年11月中国退税公告取消的是铝材、铜材和相关油脂退税，并将部分成品油、光伏、电池、非金属矿物制品退税率由13%降至9%，不是新的钢铁/GOES退税事件。第二，美国AI数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "将 5% BYOP 燃气与 4% 非晶替代从触发对冲升级为 2026 H2 基础配置，确认电力缺口已由‘并网延迟’转向‘表后自建’与‘能效替代’的硬约束阶段。", "summary": "基于 GE Vernova 和西门子能源 2026Q2 的创纪录订单披露，以及微软、甲骨文等 hyperscaler 的‘自带电力’(BYOP) 转向，验证了燃气轮机槽位已成为类 GPU 的战略稀缺资源。同时，由于 160 周的变压器交付瓶颈和 2027 年 DOE 能效新规，非晶合金对取向硅钢 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "政策红利正从“装机规模”向“并网权与合规性”集中，具备碳数据核查能力与已签署并网协议的资产将在 2027-2028 年享有显著的稀缺溢价。", "summary": "截至 2026 年 5 月，全球电力 CapEx 监管进入“清算期”。美国 FERC Order 2023 的罚款压力加速了并网效率，2027-2028 年将是“就绪项目”的回报高峰。欧盟 CBAM 正式实施使碳核算成为硬性门槛，而中国“十五五”规划通过 114 号文确立了储能容量电价平权。政策逻…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 算电协同政策约束与EPS可见性", "summary": "致 TMT行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 3bcf8726-1998-476b-a77e-dcc5ea03408c 结果摘要: Wrote handoff_response.md (10409 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 2, "neutral": 0, "value": 37, "events": [ { "title": "[auto-handoff] Fundamental validation of grid-equipment orders and power projects", "summary": "致 市场情绪师： 协同任务已完成。执行方为 主题研究员。 模式: adhoc 会话: 620155c5-c0d3-4782-a2d2-4eccb88b5925 结果摘要: Wrote handoff_response.md (10045 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "研究报告：国产电力装备产能评估与 AIDC 物理交付瓶颈对冲 日期： 2026 05 15 (亚洲/新加坡) 分析师： 工业制造分析师 (Industrials Analyst) 立…", "summary": "研究报告：国产电力装备产能评估与 AIDC 物理交付瓶颈对冲 日期： 2026 05 15 (亚洲/新加坡) 分析师： 工业制造分析师 (Industrials Analyst) 立场： 支持 (Support) 卡片： 7/8 1. 核心摘要 全球 AI 数据中心 (AIDC) 的交付周期目前受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "在铜铝供给弹性有限的环境下，2026-2030年可再生能源扩张的关键瓶颈将从发电设备转向电网扩容、并网能力和关键电力设备交付。", "summary": "2026-05-17：本卡支持Research note 04并从公用事业角度确认，铜铝供给弹性受限会把电网扩容推成大规模可再生能源并网的主要物理瓶颈。IEA显示电网建设需5-15年，全球超过2,500 GW项目卡在并网队列；LBNL显示美国截至2024年底有1,400 GW发电和890 GW储能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 3, "neutral": 0, "value": 33.8, "events": [ { "title": "2026-05-18核心判断：国产变压器12个月交付是AIDC time-to-power的必要条件，但只有算电同步项目才能把它充分转化为算力，弱项目会被局部电网调度和低上架率吸收。", "summary": "2026-05-18：本卡片支持前序国产电力装备供给优势，但压力测试“12个月变压器交付=算力产出”的线性假设。结论是：交付速度只解决第一道门，真实算力还取决于并网、供电质量、绿电/直供结算、GPU与客户负载。枢纽+直供项目转换概率约60%-75%，拥挤城市电网项目约35%-55%，投机性项目低于…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "风险评估报告：AIDC 能源脆弱性对算力资产估值与资本开支稳定性的冲击评估（2027 压力测试） 日期： 2026 05 18 分析师： chief risk (首席风控官) 立场：…", "summary": "风险评估报告：AIDC 能源脆弱性对算力资产估值与资本开支稳定性的冲击评估（2027 压力测试） 日期： 2026 05 18 分析师： chief risk (首席风控官) 立场： 压力测试 (Stress test) 核心摘要 本报告针对中国核心 AIDC（AI 数据中心）运营商进行定量压力测…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "IEA《Electricity 2026》指出，全球超过 2,500 GW 的新能源、大负荷与储能项目正卡在电网排队中；到 2030 年，年度电网投资需从目前的约 4,000 亿美元…", "summary": "IEA《Electricity 2026》指出，全球超过 2,500 GW 的新能源、大负荷与储能项目正卡在电网排队中；到 2030 年，年度电网投资需从目前的约 4,000 亿美元提升约 50%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "能源物流没有被 AI 消灭，但“物理大宗必要性”不等于“能源大宗运输贝塔免疫”，真正稀缺性正在迁移到电力设备、燃机、储能和电网集成。", "summary": "截至 2026-05-19，工业制造视角否定卡片 05 的强版“能源物流战略性免疫”。能源分子仍需大宗运输，但 AI 电力需求首先制造的是电气设备、燃机、储能、开关设备、变压器、并网和控制系统瓶颈，而不是对 VLCC、煤炭铁路或普通港口大宗贝塔的广谱保护。IEA 2026 年报告显示 hypers…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 1, "value": 51.6, "events": [ { "title": "变压器与高压开关的超长交付周期及上游硅钢产能受限，已成为AI算力落地的终极物理瓶颈，将导致2026年下半年AI硬件出货量出现15%-20%的下修。", "summary": "截至2026年5月，变压器（130-160周交付期）及核心原材料取向硅钢（GOES）的产能缺口，构成了数据中心并网的物理硬约束。新建数据中心面临平均6-9个月的投产延期，预计将迫使云巨头对2026年下半年的AI硬件订单进行15%-20%的实质性下修。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供给侧足以支撑AI硬件的长期需求，但短期决定出货节奏的是节点级通电能力，而不是全国电量总量。", "summary": "截至2026-05-21，能源侧扩张能够维持AI硬件长期需求，但不能完全消除2026-2027年的局部“电力之墙”。美国2026年计划新增86 GW公用事业级容量，绿电与储能增量可观；中国凭借风光装机、储能和State Grid的RMB 4 trillion电网投资，缓冲能力更强。但变压器、并网队…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 1, "value": 53, "events": [ { "title": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移", "summary": "AI算力物理瓶颈：从GPU算力到电力变压器与电网并网瓶颈的转移 AI Physical Bottlenecks: Shift from GPU Capacity to Power Transformers and Interconnection Queues 全球电网并网周期拉长、特大变压器交期逼…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 45.4, "events": [ { "title": "AI 物理电网瓶颈与大宗商品物理期权定价", "summary": "AI 物理电网瓶颈与大宗商品物理期权定价 AI Physical Grid Bottlenecks and Physical Option Pricing of Commodities 随着FERC否决核电并机与PJM并网延迟（5-7年）暴露了AI算力的“电力物理墙”，精炼铜与GOES（取向硅钢）…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "本轮跨行业研究有可修复的核心洞见，但当前版本因文件缺失、引用不可审计、日期/数字口径不统一及情景测算标签不足，不能发布。", "summary": "截至2026-05-23，最终QA评级为FAIL/发布阻断。可保留的证据包括Cboe期权背离、10Y TIPS实际收益率、PJM容量重定价、EIA石油发电占比极低、LBNL并网队列、DOE变压器交期与WoodMac电气设备瓶颈。但实时工作区缺失card-02、card-06、session-bri…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 1, "value": 55.8, "events": [ { "title": "电池安全溢价与补贴不确定性将抬升美欧储能LCOS高达14-29%，导致装机进度延误15-20%并引发并网弃电率飙升，成为阻碍美欧2030/2035年清洁能源目标达成的核心硬约束。", "summary": "本卡片评估了$30-40/kWh的电池安全溢价与补贴退坡风险对美欧储能及能源转型的冲击。在极端压力测试下，美国储能LCOS将飙升29.2%至$142.92/MWh，引发15-20%的装机延误，导致高比例新能源电网弃电率升至12-18%，最终拉大美欧2030/2035年可再生能源部署目标的缺口。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] FERC并网服务新规下AIDC核电直供的估值增量与过网费测算", "summary": "致 主题研究员： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: aa23a808-3f46-47f0-9c2c-3eb140857a0c 结果摘要: # Step −1 — Anchor today's date BEFORE you start Run this in …", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 4, "neutral": 1, "value": 39.2, "events": [ { "title": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团", "summary": "全球AI算力物理并网阻碍与表后自发电（BTM）产业链的跨行业抱团 Global AI Compute Grid interconnections Bottlenecks and Cross-Sector BTM Rotation 伴随变压器制造及重型燃气轮机5-7年交付周期的物理瓶颈，超大规模数据…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the strategic analysis for Research note 05, building on the physical deli…", "summary": "I have completed the strategic analysis for Research note 05, building on the physical delivery bottlenecks established in the prior cards to evaluat…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电力设备瓶颈与机柜功率跃迁正在使'东数西算'政策栈出现四处机制裂缝，发改委大概率以补丁方式而非全面重构方式回应，节点准入条件与能源保障机制将在H2-2026收紧。", "summary": "卡片06对'东数西算'政策做政策维度的压力测试。结论：在120 kW/柜跃迁与变压器/开关柜18-24个月交付瓶颈下，现行政策栈出现四处机制裂缝——PUE门槛对液冷集群失效、绿电就近消纳对AI训练负荷无法做小时级保障、节点准入缺少设备交付函、跨省算力与电力结算分属两套监管。建议四条政策补丁：升级准…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI capex 的瓶颈叙事成立，但必须从“单一 2.6 TW / 1,699 GVA 数字”升级为带日期、来源、单位和地理口径的物理交付模型。", "summary": "2026-05-26：本卡综合并支持“AI capex 走平主要反映电力核心设备与并网交付瓶颈”的主线，但对数据契约作出修正。LBNL 最新公开口径是 2025 年底美国活跃发电与储能并网队列超过 2,060 GW，而非无日期限定的 2,600 GW；变压器/GSU 交付周期仍在 128-143 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 2, "neutral": 0, "value": 49.6, "events": [ { "title": "伊顿和施耐德的电力装备产能投产严重阻后于英伟达芯片出货周期，且受制于变压器铁芯原材料取向硅钢（GOES）的物理刚性短缺，成为制约GB200于2026年点亮上线的独立刚性瓶颈。", "summary": "截至2026-05-30，压力测试表明，伊顿与施耐德的产能扩张无法填补因电网延迟导致的GB200部署缺口。两家公司的主要新增产能要到2027年及以后投产，与2026年爆量出货的芯片周期存在严重时钟错配。变压器（交付128周）与开关柜（交付40-60周）积压严重，加之铁芯原材料取向硅钢（GOES）结…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-05-30的节点下，小型模块化反应堆（SMR）作为AI无碳电力的长期终极方案远水难解近渴，短期绕过电网物理并网瓶颈的唯一桥梁是存量核电直连直供、BTM天然气过渡，配合大规…", "summary": "截至2026年5月30日，本卡片深度分析了表后电源（BTM）路径对AI数据中心绕过电网瓶颈的潜力。评估表明，虽然小型模块化反应堆（SMR）是长期零碳基荷的终极解法，但其5-10年的首台套建设周期及高昂的LCOE（100-180美元/MWh）使其无法缓解2026-2028年的电力缺口。短期内，通过锁…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 2, "neutral": 1, "value": 54, "events": [ { "title": "电网CAPEX脉冲为'算电协同'提供基本面锚，但行业内部正在从特高压主线切换到配电+数字化主线，决定了2026 H2选股需从纯特高压转向配电、智能电表、二次设备与绿电直连。", "summary": "截至2026-06-01，国网2026电网投资指引≥6,500亿元、南网≥1,750亿元，两网合计首次突破8,250亿元，连续四年增长，验证Research note 08 '算电协同'重估论。结构上特高压进入'十四五'后段，2026 Q4–2027 Q2存在订单真空风险；配电网与数字化处于202…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网通胀真实且足以使受损可再生项目无杠杆 IRR 压缩 200–400 bps，但只会使开发商群体分化而非中止能源转型，经济租金正从开发商转移到电网设备制造商与 AI 锚定的 off…", "summary": "电网设备与安装成本的结构性通胀确实部分抵消了组件/电池硬件的学习曲线，但并未否定能源转型。Lazard LCOE 上沿停止下移并走阔，BOS 占比上升使温和的 BOS 通胀对总 capex 的拉升超过组件通缩的下拉。自测算显示，美国典型公用事业级 PV 项目在 BOS +30% 叠加 COD 延后…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电网并网瓶颈不是欧美潜在GDP的硬天花板，但已是AI生产率红利兑现时点的2-3年延迟与核心商品PPI的持续高位驱动；Section 232升级是决定它停留在过渡税还是固化为真正上限的…", "summary": "截至2026-05-31，本卡片对Cards 03-05的微观结论做宏观压力测试：电网瓶颈+OEM定价权意味着电力设备PPI（BLS WPU1174自2020年累计+65-70%）将持续高于核心目标，并把成本转嫁给整体欧美Capex基数。美国T&D投资从2024年约1700亿美元升至2025E的2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "P/Watt时代AI算力与电力电网设备的实物瓶颈与地理套利", "summary": "P/Watt时代AI算力与电力电网设备的实物瓶颈与地理套利 AI Power Grid Bottlenecks and Geographic Arbitrage in the P/Watt Era 随着AI算力扩建撞上电力电网并网物理速度墙，变压器与硅钢（GOES）结构性缺口高达30%，投产延迟W…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 0, "value": 53.8, "events": [ { "title": "政策干预（包括DPA资金、关税调整和并网改革）在12-18个月内无法打破‘通电滞后’的物理时延瓶颈，保护主义关税和能效新规反而延长了供应链交期并抬升了电网建造成本。", "summary": "本报告对旨在打破‘通电滞后’的电网政策干预进行了压力测试。结果表明政策无法在12-18个月窗口内起效。DPA资金扩产的物理建设周期长达33个月（受制于21个月设备交期及6个月技能培训），产能迟至2029年释放。4月新加征的Section 232关税按成品全额征收，使大型电力变压器进口成本大涨7.5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "模型能效与边缘计算是面对一阶物理瓶颈（变压器/电网/排队）的二阶效应，叠加推理token爆炸与已锁定capex，'通电滞后'对AI应用端的重定价逻辑在12–18个月内不会被瓦解。", "summary": "TMT否定'能效+边缘'对冲电网瓶颈的反驳。Blackwell单token能效25×提升被推理调用量9× YoY与o系列推理链5,000–50,000 token的爆炸增长抵消，IEA预计数据中心电力需求2024–2030翻倍至945 TWh。瓶颈在LPT/GOES硅钢/并网排队（104–127周…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 4, "neutral": 1, "value": 46.2, "events": [ { "title": "AIDC 的 2026 H2 风险不是平均电价压垮算力需求，而是高密度 AI 负荷与本地电力现货波动、并网能力和可调节资源错配共同造成利润率分化。", "summary": "工作日 2026-06-06。本卡从 TMT 侧压力测试前序公用事业判断：现货峰谷价差扩大确实重估水电调节、储能和灵活煤电，但不足以完全平抑 AIDC 用能风险。100 MW IT 负荷在 PUE 1.46 下年用电约 1.279 TWh，有效电价每上行 RMB0.05/kWh 对应年成本增加约 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "物理电力瓶颈不会终结Scaling Law，但会把前沿模型进步压缩成更少实验室、更大台阶、更强能源地理依赖的发布节奏。", "summary": "截至2026-06-07，本卡综合前序AIDC ROIC与BTM结论：53个月并网排队不会让Scaling Law失效，但会把前沿模型路线图从GPU约束推向已上电GW级集群约束。若训练算力继续每年4-5倍增长，约28个月并网额外延迟会造成约27-42倍目标算力缺口。基准判断是形成2026-2028…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026 H2 煤-核-绿结构演变将使全国电力现货全时段加权电价中枢同比下移 5%-9%、日内峰谷价差扩大 15%-25%，调节性资产 (水电、储能、灵活性煤电) 是被低估的现货侧 …", "summary": "工作日 2026-06-06。本卡片在 card-01 公用事业分析师判断基础上，量化 2026 H2 煤-核-绿发电结构演变对电力现货价格中枢的影响。在煤价 720-760 元/吨、光伏 H2 新增 120-150 GW、风电 +55-70 GW、核电新并网 4 GW 假设下，测算全国现货全时段…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AIDC \"表后发电\"(BTM)经济性压力测试：LCOE 上升能否再次压制 ROIC？ 工作日期：2026 06 07（Asia/Singapore） 分析师：能源行业分析师（ene…", "summary": "AIDC \"表后发电\"(BTM)经济性压力测试：LCOE 上升能否再次压制 ROIC？ 工作日期：2026 06 07（Asia/Singapore） 分析师：能源行业分析师（energy analyst） 立场：stress test（对 Research note 01 \"BTM 可将 ROI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 1, "value": 65.6, "events": [ { "title": "绿电直连+CBAM构成的制度套利正使全球AI算力向中国西南、北欧、海湾三条低碳走廊集中，并把可再生能源成本曲线从单一LCOE改写为LCOE+合规溢价+并网期权的三维曲线。", "summary": "截至2026-06-07，PJM容量费、NDRC绿电直连与欧盟CBAM三者叠加，形成约$30–40/MWh的合规折扣台阶，把全球AI算力的边际产能向中国西南/西北、北欧和海湾三条低碳走廊集中。中国绿电直连园区凭物理溯源+小时级证书在CBAM申报中嵌入碳近零，叠加$44–53/MWh的直供电价，对比…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] China economic data and EM beta validation", "summary": "致 大类资产配置师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: e947fbc4-a808-440f-a64e-4d2fd375a137 结果摘要: Wrote handoff_response.md (10669 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 0, "neutral": 1, "value": 73, "events": [ { "title": "[auto-handoff] SMR商业并网时间表真伪评估与AI数据中心电价合同机制审计", "summary": "致 日报总编： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: 97045db1-4939-4d8e-a114-678cc1554257 结果摘要: Wrote handoff_response.md (9669 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-10", "supports": 1, "risks": 0, "neutral": 0, "value": 79.4, "events": [ { "title": "All three deliverables written and verified (JSON validates). Research note 04 summary (pol…", "summary": "All three deliverables written and verified (JSON validates). Research note 04 summary (policy analyst, stance: SUPPORT with refinement): I support R…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 0, "value": 60.8, "events": [ { "title": "海工装备链应选择性看多：优先看订单可验证、部件瓶颈清晰、现金流能承接增长的中小市值公司，而不是把所有油气或海风概念股一并上修。", "summary": "2026-06-11：本卡支持海工装备产业链低覆盖但有真实基本面线索的判断。全球上游投资并非全面扩张，IEA预计2025年上游石油投资下降6%至约USD420bn；中国链条的稳定器是中国海油2026年RMB112-122bn资本开支和780-800mmboe产量目标。建议优先深挖博迈科、亚星锚链、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Oracle 的AI订单不是需求问题，而是兆瓦、许可、输电、变压器和现场电力能否按时交付的问题。", "summary": "截至2026-06-11，Oracle 的 $6380亿 RPO 和 $556.63亿 FY2026 capex 证明AI需求强劲，但当前电网容量和常规数据中心交付速度不足以无摩擦支撑订单落地。OpenAI-Oracle Stargate 新增 4.5 GW 容量相当于主要市场在建容量的约75%，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 5, "neutral": 0, "value": 44.2, "events": [ { "title": "变压器供应瓶颈下的电力增量装机与电网消纳压力测试", "summary": "压力测试卡片01-02：变压器/GOES多年期约束的前提成立，但'储能自动救场'的推论站不住。延迟确实倒逼转向，但分裂为两条相反通道——负荷侧靠表后/孤岛BESS绕开变压器队列（对电网变压器需求构成温和蚕食），供给侧并网型储能仍需GSU变压器、被同一队列卡住；真正免变压器的杠杆是GETs（重导约翻…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "储能爆发产生的增量需求完全转向LFP并与镍钴脱钩，中游正极与电芯因50%的低产能利用率而严重过剩、毫无定价溢价，唯有“钛-锂耦合”一体化厂商可凭成本护城河生存；上游锂资源是唯一的物理…", "summary": "基于2026-06-12锚定日，本报告评估了储能爆发向电池材料的传导。测试显示，BESS已全面转向LFP（份额>95%），与镍、钴需求脱钩。中游LFP正极产能（500万吨）严重过剩，利用率仅50%，电芯价跌至55-75美元/kWh，无议价权；“钛-锂耦合”一体化前驱体厂商凭借1200元/吨成本护城…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "墨西哥和东南亚的低成本优势没有被完全消灭，但只有真实非PFE化、本地化供应链能保住优势；单纯第三国组装的“非中”标签在美国补贴市场基本失效。", "summary": "2026-06-12压力测试结论：USMCA 2026审查与FEOC/PFE规则不会彻底抹除墨西哥和东南亚的制造地价值，但会抹掉“中国产电芯/正极/IP + 第三国PACK + 非中标签”这类监管套利。CBP既有裁定显示，墨西哥PACK可满足USMCA却仍因中国电芯本质而被按中国原产适用Secti…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 电力瓶颈不是总装机不足，而是2026-2028年本地可交付、可调度、已并网电力的时间错配。", "summary": "截至2026-06-13，本卡支持前序“电力交付约束会惩罚 AI capex/融资假设”的主线，但将结论精确化为：总能源不短缺，本地、可靠、按时并网电力短缺。美国2026年计划新增86 GW公用事业级装机，但主要来自太阳能、储能和风电；数据中心用电到2028年较2023年隐含增加约17.0-46.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 3, "neutral": 0, "value": 57.6, "events": [ { "title": "算力西迁不会造成变压器/GIS全国性短缺，但会把订单兑现风险集中到2026-2027年合格排产、试验和并网验收窗口。", "summary": "截至2026-06-13，本卡压力测试算力西迁对变压器、GIS和上游材料的冲击。结论是：0.32-0.63GW基准西迁负荷不足以触及中国全国性产能上限，但会叠加国网4万亿元“十五五”投资、南网1800亿元2026年投资、新能源并网和646亿元变压器出口，挤占2026-2027年合格交付窗口。取向硅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘冲突带来的能源价格上行与物理并网延迟正在迫使云厂商将资本开支从纯算力硬件向电网、变压器和温控液冷基建倾斜，产生显著的“基建挤出效应”；在脱碳路径上面临自备燃气发电（高ROIC但需…", "summary": "截至2026-06-13，本报告针对中东地缘能源冲击与电网瓶颈对AI算力基建的影响进行了压力测试。在$110/MWh高电费与$1364.40M配套基建（变压器GOES溢价及液冷）刚性支出下，超大云厂商为维持10%靶向ROIC必须调降GPU采购预算54.32%，挤压芯片商毛利至45.27%。在能源脱…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the stress test analysis on the risk of export controls and technology dec…", "summary": "I have completed the stress test analysis on the risk of export controls and technology decoupling in the US China grid supply chain for Research not…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 8, "neutral": 0, "value": 29, "events": [ { "title": "以变压器为核心的电力设备扩产滞后与原材料供给刚性对2027年后的数据中心并网构成了实质性瓶颈。", "summary": "评估表明，变压器等电力设备供应链扩产极其滞后（大型电力变压器交期拉长至2-4年），受特种钢（GOES）和铜的物理供给制约，各厂商新产能直至2027/28年才能释放。这将直接阻碍2027年后的AI数据中心并网（平均并网及设备延迟超5年），验证了物理层瓶颈对算力中心扩张的实质制约。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器原材料产能周期与电网需求爆发的时间错位，特别是2028年前GOES与精炼铜的物理稀缺，确立了2027年后算力基础设施并网的硬性约束。", "summary": "截至2026年6月，变压器核心原材料GOES与铜面临物理供应刚性。高等级取向电工钢（Hi-B级）扩张周期长，美国关键产能要到2028年释放；铜市场受精矿短缺与冶炼厂TC/RC跌入负值双重打击，精炼铜供应面临收缩风险。原材料短缺已将大型变压器交期拉长至3-4年，且价格传导机制推升成品价格每吨铜上涨$…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "PJM队列与FERC co-location门槛把核电/BTM对冲削减约30–45%，但通过既有POI核电PPA与孤岛BTM保留主体可行性，对冲未被抵消；真正的尾部已从燃料风险迁移到…", "summary": "截至2026-06-14，本卡对卡片08的核电/BTM对冲做压力测试。PJM并网队列约286 GW、研究周期24–48个月、2025/2026与2026/2027交付年仅清出约5,000 MW增量[S1][S2]，加之FERC 2024年11月否决Amazon/Talen Susquehanna …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器产能的刚性瓶颈（交期拉长至36-48个月）叠加铜和高磁感取向硅钢（GOES）的价格飙升（制造BOM成本上涨21.6%至35.6%），正在引发电网及算力硬件的二次供给侧危机，成为…", "summary": "截至2026-06-14，本报告深入分析了变压器及电网硬件供应链的产能极限与关键原材料（电磁钢片及铜）的供应压力，全力支持Card-06关于电网并网限制及变压器短缺构成“物理墙”的结论。研究表明，大型变压器（LPT）交期延长至36-48个月；铜（LME报13,603美元/吨）与取向硅钢（GOES报…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 7, "neutral": 1, "value": 39.4, "events": [ { "title": "AI基建电力设备供应链的物理约束：变压器、燃气轮机与电网扩容的产能极限", "summary": "我支持Research note 04“执行胜过芯片”的方向，但压力测试其两个承重论点。约束并非单一设备，而是串行链条（并网5.2年→高压变压器128周/GSU 144周→开关柜→燃气轮机），最长杆是变压器总装/试验吞吐，而非资本[S1][S2]。GOES是真实但次级的约束：价格较2020年翻倍[…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "墨西哥能支撑选择性成熟ATP扩张，但只有在项目像小型数据中心一样配置并网、冗余、电能质量和绿电证据时成立；普通电网服务不足以支撑宽口径OSAT迁移。", "summary": "截至2026-06-16，本卡压力测试认为：墨西哥全国电量足以承接少量成熟ATP，但本地电网可靠性、电能质量、并网周期、天然气依赖和场内冗余成本不足以支撑大规模OSAT迁移。10 MW ATP全年耗电87.6 GWh，五个20 MW模块约占墨西哥2023年发电量0.25%[自测算]；真正瓶颈是CE…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力链条的策略超额收益不在需求标题，而在“客户不能等的设备、设备商不能替代的GOES、以及客户能融资的AI电力合同”。", "summary": "截至2026-06-16，本卡综合前九张卡片，结论是全球南方只能部分对冲OECD利润侵蚀，配置应从“买海外订单”升级为“买稀缺性与可融资现金流”。建议基准权重为50% A/H电网关键设备、25% GOES与电工钢、20% AI电力基础设施、5%流动性或战术对冲；若满仓则为52.6%/26.3%/2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力基建的结构性加速正在遭遇‘电网接入延时、变压器交付饱和与绿电直连消纳成本高昂’的电力可交付性硬约束，预计三季度落地节奏将面临结构性分化，东部非枢纽节点项目面临延期风险。", "summary": "截至2026-06-15，本卡对Q3智算基建进行“压力测试”。主要电力瓶颈包括：高压变电站并网周期达9-18个月；变压器排产饱和，国内订单延至2027年；绿电直连新政（发改能源〔2026〕688号）自发自用率≥60%要求拉高储能成本。此外，三季度迎峰度夏负荷紧张，东部智算中心面临错峰风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 3, "neutral": 1, "value": 64.8, "events": [ { "title": "2026-2029年电网设备扩产会缓解但难以消除交付瓶颈，关键稀缺性正在从变压器OEM装配上移到材料、组件、测试和高压系统集成。", "summary": "截至2026-06-16，本卡压力测试第09卡：变压器和关键电网设备扩产真实，但2026-2029年仍难完全跑赢AI数据中心、可再生能源并网、电网替换和电气化带来的MW/MVA交付需求。2028年前大部分大型电力变压器、HVDC电缆、开关设备和上游组件仍处紧缺。真正的瓶颈不只是OEM装配，而是GO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与GOES的线性供给极限将强制AI基建进入‘受困Capex’周期，并网延迟对ROIC的损害远超电价波动，‘确定性电力’将成为2028年前唯一的估值溢价来源。", "summary": "截至2026-06-17，变压器（交期36-60个月）与取向硅钢（GOES）的结构性缺口已成为Hyperscaler AI扩张的核心制约。30%-50%的2026年计划产能将推迟至2028年，导致海量‘受困算力’与折旧前置，压制ROIC。厂商通过核电直连、‘表后’发电及提前4年预定生产槽位进行对冲…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026年后AI扩张的关键分母应从名义MW调整为扣除并网时间、网络改造成本和可靠性折减后的可通电可靠MW。", "summary": "截至2026-06-17，本卡对前序“电力溢价”主线做压力测试：核电、长时储能和电网升级不会在2026年后阻断所有AI数据中心，但会在电力紧张地区成为大型液冷训练和高利用率推理园区的第二物理瓶颈。核电主要提升终值和能源期限，难以填补2026-2028年缺口；长时储能更像并网质量增强器，而非稳定发电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-16的核心判断：AI 电力需求会造成短期化石能源回潮，但低碳逻辑不是被瓦解，而是从年度绿电匹配升级为逐小时可交付的清洁稳定容量。", "summary": "截至2026-06-16，本卡压力测试认为：AI 基荷增长会在2026-2030年区域性推高天然气和部分煤电出力，但不会瓦解低碳转型逻辑。真正瓶颈不是绿电项目总量，而是并网、储能、输电和逐小时可交付能力。核电是最优清洁基荷对冲，但2030年前主要靠延寿、重启和增容，新建核电规模化偏2030年代。投…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 4, "neutral": 0, "value": 57.2, "events": [ { "title": "全球电网核心设备及原材料的产能赤字已构成刚性物理准入壁垒，导致产业利润池从算力/晶圆厂资本支出方结构性向重电OEM厂商转移。", "summary": "大功率汽轮机与超高压变压器的严重产能缺口（全球超高压变压器短缺约30%，交期延至2-4年；大型汽轮机交期延至5-7年）和取向硅钢（GOES）等关键材料瓶颈，对算力中心和晶圆厂构成不可逾越的物理准入壁垒。议价权转移令电力设备OEM（如HD Hyundai Electric和GE Vernova）利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电气设备出口的贸易壁垒与安全审查评估", "summary": "压力测试卡片05的压制论：美欧安全审查真实且在升级，但被狭窄锁定在中国份额本就很小的细分。最硬的法定禁令（EO 13920仅限关键国防设施、2026年欧盟逆变器资金禁令）是产品/用例特定的；2025年流氓设备警报针对的是带联网通信栈的逆变器与电池，而非被动的高压变压器。总量上，2025年中国变压器…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力负荷流向呈现向中西部和德州多点迁移的特征，且现场自备电源（BTM燃机与燃料电池）正成为规避并网排队的首要策略，这使变压器等重型电气订单最先兑现，而机房配电液冷则呈现后周期兑…", "summary": "截至2026-06-17，电力瓶颈迫使全球AI负荷加速向美国MISO和ERCOT等二三线市场迁移。为规避3-4年并网延迟，现场发电（BTM燃机与SOFC）在规划产能中已占25%。Phase 1重型变压器及开关柜订单将最先兑现收入，随后是Phase 2现场发电与Phase 3机房白区配电液冷。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BTM燃机与燃料电池凭借18个月的“电力交付速度”优势，在2026年联邦监管去约束与地缘政治引发的能源套利驱动下，已从备用电源演变为AI基础设施的核心基石。", "summary": "截至2026-06-17，BTM燃机与燃料电池已成为填补2026-2030年AI电力缺口的主导方案。GE Vernova、Siemens Energy和Bloom Energy正处于产能扩张超级周期，订单已排至2028年后。受益于2026年初EPA对碳捕集强制令的废除及“临时燃机”许可程序的简化，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 3, "neutral": 0, "value": 64.6, "events": [ { "title": "尽管国家出台绿电直连与能耗弹性政策，但受限于绿电波动、构网型储能成本以及特高压网架极限，中短期内无法消除东部智算集群并网延期与成本攀升的硬约束。", "summary": "基于2026年6月18日基准，对绿电直连与能耗指标弹性调整政策在数据中心的应用进行压力测试。尽管国家出台多用户绿电直连与非化石能源扣除政策，但受限于绿电波动的物理缺陷、强制配置“构网型储能”带来的高昂Capex（200MW集群需额外支出6000万元）、东部跨省特高压通道82%-88%的高负载率以及…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与GOES的区域化贸易壁垒通过拉长交付周期至154.5周，直接造成算力资产闲置折旧（24k GPU集群年拖累达5.624亿美元），迫使超算厂商进行CAPEX重组。", "summary": "截至2026-06-18，研究变压器与GOES区域壁垒对AI算力的传导效应。美国IRA本土含量、欧盟NZIA及中国反倾销税等碎片化政策使GOES供给割裂。压力情景下，卷线工稀缺与设备积压将变压器交期拉长至154.5周。这导致超算并网时滞达3-5年，使得24k GPU算力集群闲置一年的折旧与流失利润…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "发电侧扩容速度不足以填补2026–2028年AI算力电力缺口，缺口幅度约50%，真正的瓶颈已从发电容量转移至燃气轮机OEM订单本与并网队列。", "summary": "对卡01的压力测试结果显示，2026–2028年AI算力的电力缺口比市场预期更糟。美国数据中心到2028年需新增50–60 GW刚性容量，而核电(大型新建为零、SMR<0.5 GW、重启+扩容2.5–3.5 GW)、可再生+储能(刚性当量8–12 GW)与天然气(10–15 GW，受OEM订单本卡…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 6, "neutral": 2, "value": 54, "events": [ { "title": "[auto-handoff] AIDC电力基建逻辑下高压变压器与输电网铜需求增量的定量折算", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b7159f5c-eb1d-48e7-8538-01f5bfef96a5 结果摘要: Wrote handoff_response.md (10534 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "上游变压器与GOES瓶颈已成为并网的约束性约束，量化上递延（而非摧毁）美国2027–2030可再生目标约55–90 GW并将投运延后12–18个月，而中国凭一体化与国家电网承购按期推…", "summary": "本卡片量化上游材料瓶颈对并网时序与2027–2030装机目标的影响，支持并延伸卡片01–03主线。美国并网队列已从受研究限制转为受钢限制：约2,060 GW在申请（95%光伏/储能/风电），但仅约13%最终投运、约77%撤回；电力变压器交期128周、GSU 144周，送电环节与研究环节共同约束。G…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "表后天然气微电网和长时储能难以绕过电网并网瓶颈，反而在重型设备交期、Title V排污许可及低储能效率上构成多重成本与周期壁垒，使LCOE在压力测试下翻倍至$135.89/MWh，彻…", "summary": "报告日期为2026-06-19。压力测试表明，表后燃气微电网和长时储能并非无摩擦的并网延迟绕道方案。重型燃机面临3-7年交期瓶颈，而往复燃机则面临Title V排污许可（100 MW基荷燃机年NOx排放达1295吨，超标13倍）和SCR加装成本。铁空气电池低至40%的往返效率导致严重的燃料效率损失…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AIDC电力并网瓶颈对电力敏感型化工品（如烧碱、黄磷）供给端硬约束的交叉验证", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: 6a16e82a-5277-47a4-bf61-68e0ff5c0f23 结果摘要: Wrote handoff_response.md (11068 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 6, "neutral": 0, "value": 49.4, "events": [ { "title": "工业电气设备产能扩充与供应链交付压力", "summary": "压力测试 Research note 02–03 的电气设备瓶颈论。瓶颈在 2026–2028 年对定制大型变压器、HVDC、中压开关柜与并网真实成立，但稀缺溢价被误当作永久性。三点修正：一、积压创纪录（Eaton 总积压逾190亿美元、数据中心积压228GW≈12年需求；GE Vernova电气…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在严格的物理绿电比例与ESG追溯要求下，西部算力枢纽因储能分摊成本、水电季节性枯竭及特高压外送优先权而面临多重物理与电价折扣收窄瓶颈。", "summary": "截至2026-06-21，本卡压力测试了西部算力西迁假说。结论显示，在物理绿电占比率达80%的强监管下，内蒙古数据中心因风光波动性面临约0.11元/千瓦时的储能分摊成本，削弱电价折扣至18%-27%；贵州因枯水期水电出力锐减70%，面临电网碳排放数倍激增的ESG违规风险；此外，特高压通道外送优先权…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源冲击会结构性加速清洁能源替代，但电网、融资和关键矿产瓶颈会把短期宏观结果变成“更高通胀+更快转型”的双重状态。", "summary": "截至2026-06-21，能源成本上升对清洁能源转型的净效应偏加速，但呈“双速”：太阳能、储能、电气化、电网灵活性和进口替代受益；海上风电、绿氢、并网、变压器/电缆和关键矿产供应链受高融资成本与交付瓶颈拖累。EIA预计2026年油需减少110万桶/日，IEA预计2025年清洁能源投资2.2万亿美元…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网积压与公用事业审批已成为未锁定电力权益的AI绿地项目的真实区域性天花板，但会通过费率、抵押和可靠性规则筛选项目，而非一刀切终止AI扩张。", "summary": "截至2026-06-21，本卡片压力测试卡片05并支持其物理瓶颈判断，但把机制收窄为“并网队列+大负荷费率+可靠性审批”的区域性天花板。LBNL显示美国2025年底超过2,060 GW发电/储能排队；AEP Ohio把超过30,000 MW数据中心意向筛到5,642 MW约束性合同；PJM预计20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 9, "neutral": 1, "value": 35.8, "events": [ { "title": "变压器及组件产能约束会直接延长公用事业并网积压，并削弱可再生能源接入承载能力，因为它们决定网络升级能否从纸面协议变成已上电资产。", "summary": "截至2026-06-21，本卡支持前序结论：变压器、GSU、变电站主变、配电变压器和上游GOES/CTC等组件短缺，会把并网研究积压转化为网络升级和上电施工积压。Berkeley Lab显示美国并网队列仍有超过2,060 GW发电和储能容量在排，且408 GW已有草案或已执行并网协议但尚未商业运营…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-2028年，电池、VPP、灵活数据中心电价和GETs是跨越变压器短缺的桥，不是绕开短缺的旁路。", "summary": "截至2026-06-21，本卡压力测试认为，变压器和电网设备产能正在扩张，但新增工厂、认证、熟练劳动力和长交期决定其主要缓解2029年后的出清，而非完全解决2026-2028瓶颈。储能、VPP、数据中心需求侧管理和GETs更适合作为近端过桥工具：它们可以削峰、递延部分网络升级、支持灵活并网，并提高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research Summary (Research note 06/10) This card performs a stress test on the thesis of \"f…", "summary": "Research Summary (Research note 06/10) This card performs a stress test on the thesis of \"frictionless growth for Chinese electrical equipment export…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球贸易壁垒（关税、能效标准与碳关税）不仅未能缓解变压器供应链紧张，反而通过72.68%的转运成本附加[自测算]及碳关税上升增加了‘二次约束’，使交付周期锁定在80-200周，导致1…", "summary": "截至2026-06-21，本分析对美欧贸易限制措施对取向硅钢（GOES）及变压器供应链的二次约束进行了压力测试。定量测算表明，绕过美国关税将中国硅钢经墨西哥转运将增加72.68%采购成本（增至6130美元/吨）并延误12-24周[自测算]；欧盟CBAM对长流程硅钢进口征税在2026/28/30年分…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 1, "risks": 4, "neutral": 0, "value": 72.2, "events": [ { "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 4cd2d717-9d04-4569-b0d8-cc37d743d0a9 结果摘要: Wrote handoff_response.md (9414 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "布伦特高于$120/bbl若持续，将通过亚洲进口账单、大西洋再路由和LNG竞争强化能源安全溢价，但其自我抑制性意味着真正持久的投资信号是电力系统、能效和多元化供应CapEx，而非无纪…", "summary": "截至2026-06-22，本卡支持霍尔木兹冲击已成为宏观级能源供给冲击的判断，但限定布伦特持续高于$120/bbl是高压情景而非官方基准。霍尔木兹在2024年承载约20 million b/d石油流量和约20%全球LNG贸易，亚洲敞口最大。若高油价延续，贸易流向会转向大西洋盆地和美国成品油，LNG…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI算力扩张不会被能源安全溢价终止，但会被“可交付稳定兆瓦”重新定价，全球集群部署将向电力充足、并网确定、能效合规且成本分摊透明的地区集中。", "summary": "截至2026-06-22，本卡对能源安全CapEx如何约束AI算力基础设施作压力测试。结论是：电价上升重要，但更核心的瓶颈是稳定兆瓦能否按GPU商业化窗口交付。IEA、LBNL、EPRI与NERC数据共同显示，AI数据中心需求已成为局部电网级变量，尤其在美国、欧洲、中国与东南亚。能源安全溢价会推高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2030年特高压与绿电并网能缓解全国电量约束，却不能完全对冲智算中心的本地可靠容量和电价基差风险，因此仍支持从粗放算力堆叠转向高能效算力、软件应用和电网CapEx受益资产。", "summary": "截至2026-06-22，本卡对前序电力瓶颈判断进行压力测试：特高压、绿电并网、煤电灵活性、抽蓄和新型储能建设足以缓解全国性年度电量缺口，但不能完全消除智算中心在本地并网、24/7可靠容量和市场化电价基差上的风险。国家电网规划“十五五”15回特高压直流工程、跨省输电能力提升35%、每年支撑不低于2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 7, "neutral": 0, "value": 47.6, "events": [ { "title": "电网并网队列与基础设施扩容压力评估", "summary": "本卡支持卡片1–2：约束已从Capex/GPU转向已通电的MW，并将瓶颈量化为双向并网堵塞。发电侧：美截至2025年底约2,060GW排队（约现有装机2倍），408GW已签约未通电，申请到商运中位时长翻倍至>4年。负荷侧：ERCOT大负荷队列达226GW（近4倍于2024年底63GW，77%为数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "配电网滞后与变压器短缺连同HBM/CoWoS确实构成AI capex的物理硬约束——对边际开发商是硬墙，对优先级买家是ROIC压缩，从物理层面支持AI-capex去杠杆而非2026崩…", "summary": "截至2026-06-23，本卡从公用事业侧确认：配电网扩容滞后与变压器短缺是AI capex物理钳制的第二片钳口，与HBM/CoWoS硅片分配叠加。变压器交期达128–144周乃至4年，电力变压器缺口30%、配电10%，GOES自2020年翻倍、铜+50%；LBNL并网队列约2,060 GW、中位…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "绿色先进制造与城投信贷收缩将通过Capex砍减与地方投资减少，向电力设备及能源基建传导共计313.42亿元的需求收缩压力，并延缓光伏组件与储能电芯的价格修复。", "summary": "截至2026-06-23，本报告在金租及信贷收缩背景下评估其对电力设备与能源基建的传导。测算显示，全系统绿色制造信贷收缩将导致相关厂区配电及并网设备需求缩减100.03亿元；城投信贷收缩将导致地方能源基建投资缩减213.39亿元（对应新能源装机延期约6,096.86 MW）。合计拖累总需求313.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网输变电容量与并网审批周期（而非总体发电量）已成为AI算力部署的终极物理刚性瓶颈，并已实质性挤压IT硬件采购预算。", "summary": "本报告对AI算力扩张的电网与能源瓶颈进行压力测试。分析表明，虽然电力设备订单充足，但美国并网队列平均达5.5年，且面临FERC 2026年6月大型负荷监管及弗吉尼亚GS-5等防止成本转嫁政策。变压器等重电面临GOES原材料短缺，Tier-1厂商积压至2029-2030年。表后核电直供监管被封死，燃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 2, "neutral": 0, "value": 87, "events": [ { "title": "我们已成功完成了首席策略师（Chief Strategist）的第一张卡片的研究与文件写入。 核心结论摘要 (Strategic Overview) 在 2026 06 25 的定位…", "summary": "我们已成功完成了首席策略师（Chief Strategist）的第一张卡片的研究与文件写入。 核心结论摘要 (Strategic Overview) 在 2026 06 25 的定位背景下，铜价暴涨至 $13,530/吨 [S1]，取向硅钢（GOES）供应空缺达 30% [S2]，电池级碳酸锂价格…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 十五五电网4万亿投资计划对国内工业产出及PPI的拉动效应", "summary": "致 深研主编： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 962dd3bc-dd6b-46d1-bdbc-e2f725f549fb 结果摘要: Wrote handoff_response.md (11728 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "发电侧与配电侧在2026年正陷入‘既无路（电网排队超4年及变压器稀缺）也无车（燃机交期5-7年及SMR未商业化）’的双端挤压困境，共址直供监管红线的收紧彻底消除捷径，导致算力资本开支…", "summary": "截至2026年6月25日，配电与电网设备交期拉长使得发电侧面临严重并网瓶颈。大型电力及GSU变压器交期达18-48个月，GOES硅钢供给偏紧，并网排队中位数超4年。FERC否决核电表后直供迫使亚马逊等转向表前PPA，面临同样电力瓶颈。同时高确定性燃气轮机交期达5-7年，SMR在2030年前难以放量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 10, "neutral": 1, "value": 35.4, "events": [ { "title": "欧美能源贸易壁垒与技术标准准入对全球储能产业链的重构效应", "summary": "本卡压力测试并基本否定\"西方贸易排除将实质性削弱中国储能高端全球竞争力\"的强命题。要害是把\"高端\"误等同于\"西方认证/西方原产\"：1260H采购禁令、ITC剥夺、301关税、欧盟资金禁令、NZIA韧性标准多为原产地准入闸门，而非能力门槛。中国仍据技术与成本前沿（全球电池约69%、CATL 39.2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-25，“标准俘获”是可信的选择性围堵工具，但不是可信的近端全球产能替代机制。", "summary": "截至2026-06-25，本卡压力测试认为，欧美碳足迹、ESG尽调、强迫劳动、可信供应链和安全认证可以形成选择性非关税闸门，但尚不足以全球性削弱中国产能。美国FEOC/PFE税收抵免规则是最硬约束；欧盟电池护照、碳足迹、NZIA韧性标准和Forced Labour Regulation更像合规筛选…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心的卡脖子资产已从芯片变为高压电气设备，交期与产能缺口把电气推成最大且持续涨价的Capex板块，稀缺溢价持久归于变压器/特高压/电气配套，而非火电或燃气beta。", "summary": "截至2026-06-25，本卡支持并加固Card-01判断：AI数据中心的约束已从芯片转向电气并网设备。核查结果——大型变压器交期约128周、GSU约144周、中压开关柜52–78周，均超数据中心12–18个月建设周期；GSU需求2019–2025增274%，美国变压器/配电变压器2025供给缺口…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Summary of Work I have completed the research and stress test on the transmission effects o…", "summary": "Summary of Work I have completed the research and stress test on the transmission effects of grid equipment bottlenecks on upstream key raw materials…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-26", "supports": 1, "risks": 5, "neutral": 0, "value": 71.8, "events": [ { "title": "并网延迟的变压器库存积压是可逆的时点拖累(金盘-8%~-11%、思源<-3%)，被国网4万亿内需托底，应在超配出口变压器框架下偏思源、轻金盘并叠加配对与公用事业杠铃对冲。", "summary": "并网延迟导致的已完工未交付变压器库存积压，是营运资金/收入确认的时点事件而非需求毁灭——订单仍排产至2027年在手backlog内。对金盘科技(688676.SH)拖累真实但有界：北美AIDC相关收入是最高Beta切片，Q2/Q3净利约-8%至-11%、伴随存货与FCF占用，属逢低布局而非证伪；对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "维持12% NAV清洁基础设施袖，但不要上调中枢；政策兑现支持电网/储能/运营商核心，136号文电价风险和制造产能过剩要求压低光伏/电池制造暴露。", "summary": "截至2026-06-26，本卡综合前序清洁基础设施杠铃：政策和需求数据支持保留12% NAV清洁袖，但不支持上调或追宽基清洁贝塔。国家电网2026-2030年4万亿元投资、全国电网接近5万亿元投资图景、1-5月用电5.7%增长、互联网数据服务用电40%+增长、算力用电到2030年8000亿千瓦时预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心虹吸变压器产能导致全球交期拉长至2-3年，加剧海外CXO项目并网延迟2-3年并毁灭其NPV，同时对A股变压器出口企业（如金盘科技）Q2-Q3业绩构成成品存货积压拖累。", "summary": "本卡片验证并支持卡片04结论。AI数据中心对大功率变压器(LPT)需求的剧增(每GW IT负荷需15-16台100MVA变压器)，受限于GOES缺口、绕线工培训及测试台位滞后，将LPT交期推升至2-3年。这加剧了海外电网并网延迟，导致新加坡、爱尔兰等海外CXO基地落地面临2-3年刚性延误，毁灭性削…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 0, "value": 77.2, "events": [ { "title": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞", "summary": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞 DeepSeek's 96.37% price cut: Jevons amplification or AI hardware demand destruction — head-to-…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "贸易壁垒导致的取向硅钢（GOES）结构性割裂，以及2026-2030年铜矿的百万吨级结构性缺口，奠定了电力变压器成本的向下刚性，而铝代铜的技术陷阱因被动增加25%的硅钢用量而被超大规…", "summary": "基于2026-06-27权威工作日，本报告支持AI电力并网瓶颈已下沉至原材料端的判定。由于贸易壁垒隔离，中国占56%的全球取向电工钢（GOES）产能无法被美欧替代。同时，电网扩容拉动铜需求，2026年铜矿缺口达33万吨，并于2030年扩大至150万至220万吨。这导致变压器制造成本刚性上涨。试图通…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备扩产会缓解AI数据中心并网瓶颈，但无法在2026-2027年完全解除大型变压器和输电接入的物理约束，因此算力利润扩散仍是受瓶颈约束的电网资本开支周期。", "summary": "截至2026-06-28，本卡支持上一卡的物理瓶颈判断，但强调扩产的时间结构：变压器、开关柜和电网设备扩产是真实的，Eaton、Hitachi Energy、Siemens Energy等制造商已披露新增产能和高积压订单；但大型电力变压器仍受定制设计、GOES/铜、测试槽位、运输和并网研究约束，新…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 10, "neutral": 1, "value": 47.6, "events": [ { "title": "海上风电+UDC的24/7低碳电力合同与并网可行性", "summary": "可融资的24/7低碳UDC电力可实现，但只能作为四层结构：成型/固化的多年期PPA（叠加机制电价差价结算CfD）+ 偏LDES的多小时现场储能 + 有界偏差结算带 + 电网耦合并网（源网荷储或电网+逐时T-EAC）。否定纯海风直连孤岛——它移除了固化小时数所需的电网/储能平衡，24/7现金流不可投…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "由于储能与动力电池的绝对规模错配以及中国储能并网瓶颈和矿山复产传闻，储能需求增长在第三季度不足以支撑广期所碳酸锂期货价格企稳。", "summary": "本报告定量评估了Fastmarkets会议上关于“储能增长抵消电动汽车放缓”以稳定锂价的观点。分析表明，储能消费规模过小，无法对冲动力电池放缓的影响；且国内并网瓶颈给储能实际装机量带来下行风险（可能造成32万吨LCE的需求毁灭）；结合宁德时代枧下窝矿山复产预期和广期所国际化引入抛压，预计Q3碳酸锂…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力基础设施供应链瓶颈：电力变压器与关键配电硬件的产能缺口", "summary": "对Research note 01以变压器为中心的瓶颈论断做压力测试。结论：GIS与高压断路器产能确实紧张（≥145 kV断路器交期已超12个月、整套GIS需60–100周、供给集中于约4家OEM），但它是次级协同瓶颈，而非近期核电并网的绑定约束。原因在于结构：近期核电对AI的供给主力是反应堆重启…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-06-29的政策环境下，通过将绿电绿证的环境资产价值（抵消范畴二排放以规避每度电约0.41元的碳税/碳关税）与电力系统调节成本（114号文储能及火电容量电费分摊）进行统筹…", "summary": "截至2026-06-29，本报告支持卡片08并深化分析。发改价格〔2026〕114号文落地后，将火电与新型储能容量电费摊销纳入系统运行费，由工商业用户按用电比例共同分摊，增加了跨省区绿电采购的隐性通道成本。然而，绿电绿证（GEC）的‘证电合一’交易与碳足迹核算为沿海UDC开辟了独特的套利空间：一个…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-29", "supports": 1, "risks": 6, "neutral": 0, "value": 70, "events": [ { "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "超长期特别国债与地方基金预算受阻的非对称限制，导致西部智算基地无法依靠补贴长期消化绿电LCOE折价收窄至13.89%与并网延迟造成的40%度电成本上升[S3][自测算]；物理并网时滞…", "summary": "截至2026-06-30，本报告系统性合成AI算力电力约束与中国中央融资及地方财政的矛盾。超长期特别国债仅能用于机房CapEx，无法支付占OpEx 60%以上的日常电费[S6]。由于房地产下行拖累地方基金预算[S3]，地方电价补贴难以为继。西部直连绿电并配建储能的LCOE高达0.362元/kWh，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "summary": "支持第06卡\"并网延迟→利润压缩\"的传导论点，但将单向结论细化为分化。针对11GW数据中心建设缺口与12-18GW刚性局部电网缺口，48-60个月并网延迟通过两条通道传导：(1)经订单/PPA递延与尾款拖延压制发电子链开工率——5.7%销量递延经经营杠杆放大为约-21%利润压缩[自测算]；(2)将…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] } ], "href": "/investor/theses/thesis-66eb42380e/", "networkHref": "/investor/research-network/#thesis-66eb42380e" }, { "id": "thesis-c6c6be4745", "source": "tracked_thesis", "title": "中国出口制造的胜负手从低价扩张转向利润率、合规和本地化", "titleEn": "Chinese export manufacturing is shifting from low-price expansion to margin, compliance, and localization", "lane": "中国出口制造/利润率", "laneEn": "China export manufacturing and margins", "laneHref": "/investor/lanes/china-export-manufacturing-and-margins/", "investableFocus": "中国出口制造/利润率：002475.SZ, 000333.SZ, 600690.SH, 300750.SZ, 002594.SZ, 2313.HK, 2020.HK", "stockUniverse": [ { "ticker": "002475.SZ", "name": "立讯精密", "market": "A-share", "role": "消费电子制造出海", "roleEn": "消费电子制造出海", "bucket": "core", "href": "/investor/stocks/002475.SZ/" }, { "ticker": "000333.SZ", "name": "美的集团", "market": "A-share", "role": "家电全球化", "roleEn": "家电全球化", "bucket": "core", "href": "/investor/stocks/000333.SZ/" }, { "ticker": "600690.SH", "name": "海尔智家", "market": "A-share", "role": "家电全球化", "roleEn": "家电全球化", "bucket": "core", "href": "/investor/stocks/600690.SH/" }, { "ticker": "300750.SZ", "name": "宁德时代", "market": "A-share", "role": "电池出海", "roleEn": "电池出海", "bucket": "core", "href": "/investor/stocks/300750.SZ/" }, { "ticker": "002594.SZ", "name": "比亚迪", "market": "A-share", "role": "新能源车出海", "roleEn": "新能源车出海", "bucket": "core", "href": "/investor/stocks/002594.SZ/" }, { "ticker": "2313.HK", "name": "申洲国际", "market": "HK", "role": "纺织服装制造出海", "roleEn": "纺织服装制造出海", "bucket": "core", "href": "/investor/stocks/2313.HK/" }, { "ticker": "2020.HK", "name": "安踏体育", "market": "HK", "role": "品牌与制造链对照", "roleEn": "品牌与制造链对照", "bucket": "core", "href": "/investor/stocks/2020.HK/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" } ], "stockTickers": [ "002475.SZ", "000333.SZ", "600690.SH", "300750.SZ", "002594.SZ", "2313.HK", "2020.HK", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI" ], "coreView": "赛道：中国出口制造/利润率。候选观察标的：002475.SZ, 000333.SZ, 600690.SH, 300750.SZ, 002594.SZ, 2313.HK, 2020.HK。当前共有 40 条证据、3 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 71, "eventCount": 160, "firstSeen": "2026-05-04", "lastSeen": "2026-06-29", "supports": 15, "risks": 116, "neutral": 29, "analystCount": 20, "sourceTypeCounts": { "研究笔记": 139, "证据更新": 6, "分析协作": 14, "研究线程": 1 }, "topTerms": [ "AIDC", "新能源", "能源", "电池", "Chinese", "export", "manufacturing", "shifting", "low-price", "expansion", "margin", "compliance" ], "whatChanged": [ "2026-06-29: 电网扩容中期改善消纳，但2026-2027年会先通过弃电、现货价格和辅助服务分摊压低绿电发电侧收益率，因此设备优于发电。", "2026-06-29: 特种化学品：供应端波动对下游利润的压力", "2026-06-29: [auto-handoff] 中压干式变压器出海企业海外合规及交付溢价测算", "2026-06-29: [auto-handoff] Hormuz shock & core CPI transmission", "2026-06-29: 2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速" ], "practical": { "sectorLane": "中国出口制造/利润率", "sectorLaneEn": "China export manufacturing and margins", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "trim_or_hedge", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "riskBudget": "低风险预算：0-2% thesis risk", "portfolioRole": "风险雷达", "weightGuide": "0-2% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "适合作为风险雷达，不适合作为无条件加仓理由。", "positionSizing": "建议按 0-2% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "002475.SZ", "name": "立讯精密", "market": "A-share", "role": "消费电子制造出海", "roleEn": "消费电子制造出海", "bucket": "core", "href": "/investor/stocks/002475.SZ/" }, { "ticker": "000333.SZ", "name": "美的集团", "market": "A-share", "role": "家电全球化", "roleEn": "家电全球化", "bucket": "core", "href": "/investor/stocks/000333.SZ/" }, { "ticker": "600690.SH", "name": "海尔智家", "market": "A-share", "role": "家电全球化", "roleEn": "家电全球化", "bucket": "core", "href": "/investor/stocks/600690.SH/" }, { "ticker": "300750.SZ", "name": "宁德时代", "market": "A-share", "role": "电池出海", "roleEn": "电池出海", "bucket": "core", "href": "/investor/stocks/300750.SZ/" }, { "ticker": "002594.SZ", "name": "比亚迪", "market": "A-share", "role": "新能源车出海", "roleEn": "新能源车出海", "bucket": "core", "href": "/investor/stocks/002594.SZ/" }, { "ticker": "2313.HK", "name": "申洲国际", "market": "HK", "role": "纺织服装制造出海", "roleEn": "纺织服装制造出海", "bucket": "core", "href": "/investor/stocks/2313.HK/" }, { "ticker": "2020.HK", "name": "安踏体育", "market": "HK", "role": "品牌与制造链对照", "roleEn": "品牌与制造链对照", "bucket": "core", "href": "/investor/stocks/2020.HK/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" } ], "stockTickers": [ "002475.SZ", "000333.SZ", "600690.SH", "300750.SZ", "002594.SZ", "2313.HK", "2020.HK", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI" ], "primaryStocks": [ { "ticker": "002475.SZ", "name": "立讯精密", "market": "A-share", "role": "消费电子制造出海", "roleEn": "消费电子制造出海", "bucket": "core", "href": "/investor/stocks/002475.SZ/" }, { "ticker": "000333.SZ", "name": "美的集团", "market": "A-share", "role": "家电全球化", "roleEn": "家电全球化", "bucket": "core", "href": "/investor/stocks/000333.SZ/" }, { "ticker": "600690.SH", "name": "海尔智家", "market": "A-share", "role": "家电全球化", "roleEn": "家电全球化", "bucket": "core", "href": "/investor/stocks/600690.SH/" }, { "ticker": "300750.SZ", "name": "宁德时代", "market": "A-share", "role": "电池出海", "roleEn": "电池出海", "bucket": "core", "href": "/investor/stocks/300750.SZ/" }, { "ticker": "002594.SZ", "name": "比亚迪", "market": "A-share", "role": "新能源车出海", "roleEn": "新能源车出海", "bucket": "core", "href": "/investor/stocks/002594.SZ/" }, { "ticker": "2313.HK", "name": "申洲国际", "market": "HK", "role": "纺织服装制造出海", "roleEn": "纺织服装制造出海", "bucket": "core", "href": "/investor/stocks/2313.HK/" } ], "instrumentTags": [ "002475.SZ", "立讯精密", "000333.SZ", "美的集团", "600690.SH", "海尔智家", "300750.SZ", "宁德时代", "002594.SZ", "比亚迪", "2313.HK", "申洲国际" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 117 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 20, "recentEvents": 117, "supportRatio": 0.09, "riskRatio": 0.72, "conflictScore": 81.7, "evidenceDepth": 100 }, "score": 67.3 }, "events": [ { "date": "2026-06-29", "title": "[auto-handoff] Impact of global capacity pricing spikes on domestic transformer export margins", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 324722b5-2f7b-4b91-aec7-ac5429f32a39 结果摘要: Wrote handoff_response.md (11765 bytes); wrote handoff_response.en.md (12438 bytes) — pri…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 30.5, "matchedTerms": [ "能源", "Chinese", "export", "expansion", "margin", "margins", "transmission", "overseas" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 评估PJM等容量电价暴涨对国内出海变压器毛利率的传导", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 1d2b9e99-6767-49f3-aa4b-4520c45f9a0b 结果摘要: Wrote handoff_response.md (9124 bytes); wrote handoff_response.en.md (9491 bytes) — The 1…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 19.25, "matchedTerms": [ "能源", "Chinese", "export", "tariff", "expands", "cost", "transformer" ], "href": "" }, { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 17.75, "matchedTerms": [ "AIDC", "能源", "电池", "power" ], "href": "" }, { "date": "2026-06-29", "title": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "summary": "对中国，2026年油价/柴油冲击既是被缓冲的进口税、又是战略加速器，且净额上比卡片01–02的美欧日集团损害更小。三大事实：（1）进口税真实但落在PPI而非CPI——5月PPI同比+3.9%（2022年7月来最快）、CPI仅+1.2%，疲弱内需把冲击吸收为生产者利润压缩；（2）进口税被三重缓冲——全球最大约14亿桶战略储备、作为受制裁油折价买方、疲弱需求；…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "能源", "电池", "macro" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "evidence", "risk", "dominant", "inflation", "transmission", "policy" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 中压干式变压器出海企业海外合规及交付溢价测算", "summary": "致 主题研究员： 协同任务已完成。执行方为 工业制造分析师。 模式: adhoc 会话: f0fa9b36-39ce-428c-a4ff-9c7d5180b4ad 结果摘要: Wrote handoff_response.md (11248 bytes); wrote handoff_response.en.md (12258 bytes) — This …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 15.75, "matchedTerms": [ "export", "margin", "margins", "overseas", "order", "grid", "transformer" ], "href": "" }, { "date": "2026-06-29", "title": "特种化学品：供应端波动对下游利润的压力", "summary": "特种化学品：供应端波动对下游利润的压力 Will supply-driven specialty chemical price hikes squeeze downstream manufacturing margins (pharma, coatings, water treatment)? Prior related institute work to…", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "能源", "manufacturing", "margin", "margins" ], "href": "" }, { "date": "2026-06-29", "title": "电网扩容中期改善消纳，但2026-2027年会先通过弃电、现货价格和辅助服务分摊压低绿电发电侧收益率，因此设备优于发电。", "summary": "截至2026-06-29，本卡压力测试认为：电网资本开支并不会直接向绿电征税，但会把高比例新能源系统的调度难度和灵活性成本显性化，从而压低发电侧长期收益率。2024年全国辅助服务费用402.5亿元，风电和光伏分摊合计约63%，折算约0.011元/千瓦时，相当于2024年正式现货市场价格区间的3.2%-4.4%。2026年一季度风电、光伏利用率降至91.4%…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "新能源", "能源" ], "href": "" }, { "date": "2026-06-28", "title": "新动能（AI基础设施与绿电）在体量上仅能弥补房地产投资收缩缺口的约10%-15%，且面临严重的电网并网、输电设备及芯片良率等物理瓶颈限制，无法在短期内有效对冲房地产缩表的宏观增长压力。", "summary": "压力测试表明，新旧动能切换在总量与结构上存在重大错配。1-5月房地产投资下降16.2%带来1.8万亿元的总需求缺口，而电网与AI算力投资年度增量合计约2100亿元，仅能对冲缺口的10%-15%。且AI和新能源建设面临电网并网、输电设备周期以及芯片良率等严重的物理瓶颈，导致资本开支兑现滞后6-12个月。新动能无法有效对冲房地产缩表压力，政策重点应转向解决电网…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "新能源", "能源", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "需求端验证：中国制造业PMI与工业数据对金属品种价格驱动力的分析", "summary": "截至2026-06-29，本卡支持前序黄金/结构性金属对冲框架，但按品种分化：中国需求改善足以验证铜的盈利Beta，核心来自电网、电气化、高技术制造和出口，而非房地产；锂也有电动车、储能和绿色出口支撑，但1-5月新能源汽车销量仅温和增长且锂价近月回撤，说明锂应保持受限高Beta卫星。组合上维持21%黄金+结构性金属权益杠铃，8%结构性金属内部偏铜、锂限重。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "新能源", "能源", "macro" ], "href": "" }, { "date": "2026-06-28", "title": "海缆仍是订单背书最强的底仓但需下修新增订单毛利假设；储能是政策背书装机而非利润背书，只有高可靠、长时化、采购纪律强的龙头能把114号文转化为毛利；UDC是固化成本支付方，仍应按主题脉冲定规模。", "summary": "截至2026-06-29，本卡对“海缆与储能订单背书型利润确定性”做压力测试。结论是分化：海缆订单确定性仍高，东方电缆截至2026-03-25在手订单193.12亿元，但海风补贴退坡和136号文新能源入市会压缩新增订单毛利假设；储能在114号文后装机确定性更强，但容量收益属于合格独立储能项目，不自动变成集成商毛利。江苏顶峰补贴从0.5元/kWh降至0.3元…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "新能源", "能源" ], "href": "" }, { "date": "2026-06-28", "title": "算力密度激增下的能源供给适配：从“规模化能源需求”到“极端局部负载调控”", "summary": "支持Research note 09“天花板下移到机架”的结论并落地能源侧答案。1MW机架的配电层威胁是斜率而非规模：同步GPU训练任务亚秒级摆动达数十至数百兆瓦[S1]，ERCOT大负荷申请均值约725MW、多站超1GW[S2]，发电侧却需1–90分钟响应[S4]。解法是三时间尺度分层缓冲：毫秒级用机架电容（GB300 65焦/GPU，平均峰值需求降约3…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 15.5, "matchedTerms": [ "能源", "电池" ], "href": "" }, { "date": "2026-06-28", "title": "并网瓶颈不会引发储能装机崩塌，在容量电价政策与构网型技术支撑下储能市场将保持韧性，但其规模尚不足以逆转三季度全球锂市过剩及碳酸锂价格下行探底趋势。", "summary": "本报告压力测试了电网瓶颈对新型储能装机的影响。一季度实际装机27.05GWh表明极度悲观的“情景A”（42GWh）已失效。新能源取消强制配储后，行业在容量电价（发改价格〔2026〕114号）等支持下转向市场化独立储能。结合超5万亿元的电网投资及构网型技术落地，瓶颈对锂需求的冲击收窄至5.59-10.79万吨LCE，储能装机保持韧性。由于绝对规模错配与全球1…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "新能源", "能源" ], "href": "" }, { "date": "2026-06-27", "title": "霍尔木兹海峡重开仅为“执行折价”而非“和平红利”，60天物理排雷延迟导致到岸能源价具极强向下刚性，与高利率共同构筑新能源“资本开支壁垒”，系统性阻碍全球能源转型进程并抬升算力用电门槛。", "summary": "霍尔木兹海峡重开虽缓解了原油期货的即期上涨动能，但由于海峡内活体水雷清障导致的60天排雷延迟，到岸能源价格（CIF）维持强韧的向下刚性，地缘风险溢价只是发生了盘面转移。在高利率与绿色通胀双重作用下，全球新能源面临高额WACC与电网并网漫长排队的“资本开支壁垒”，转型进程阶段性受阻，算力能源供应链成本大幅飙升。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "新能源", "能源" ], "href": "" }, { "date": "2026-06-26", "title": "维持12% NAV清洁基础设施袖，但不要上调中枢；政策兑现支持电网/储能/运营商核心，136号文电价风险和制造产能过剩要求压低光伏/电池制造暴露。", "summary": "截至2026-06-26，本卡综合前序清洁基础设施杠铃：政策和需求数据支持保留12% NAV清洁袖，但不支持上调或追宽基清洁贝塔。国家电网2026-2030年4万亿元投资、全国电网接近5万亿元投资图景、1-5月用电5.7%增长、互联网数据服务用电40%+增长、算力用电到2030年8000亿千瓦时预期，都验证电网/储能/运营商方向。下行风险在136号文市场化…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "能源", "电池", "macro" ], "href": "" }, { "date": "2026-06-25", "title": "能源降本对AI算力基础设施的影响及扩容瓶颈评估", "summary": "压力测试卡片-02的能源降本看多论。燃气降本（CCGT减负$25.76/MWh、100MW数据中心年省OpEx约$24.94M）真实但属次要，仅折合AI级capex的1.7%–2.5%。真正约束是time-to-power：美国并网队列卡着约2,300GW、PJM平均排队约8年、Dominion北弗吉尼亚要到2033年；大型变压器交期128周、燃气轮机GE…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16.75, "matchedTerms": [ "新能源", "能源", "power" ], "href": "" } ], "timeline": [ { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 3, "value": 45, "events": [ { "title": "60.6% 渗透率是汽车定价逻辑从‘电化学’转向‘硅芯片’的奇点，电池材料的‘分母陷阱’与 ADAS 算力的‘第三支柱’将在 2027 年合同口径下完成估值解耦。", "summary": "新能源汽车渗透率在 2026 年 4 月达到 60.6% 的历史高位，标志着市场进入结构性反转期。电池材料受 PHEV/EREV 占比上升及行业总量萎缩影响，陷入“分母陷阱”，需求强度趋缓。与此同时，ADAS 算力以单车 2000 TOPS 为锚点，正式确立为继数据中心、智能手机之后的 AI 算力…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电池材料需求模型应抛弃对国内渗透率指标的盲从，转而锚定全球GWh总产量，并警惕PHEV占比上升带来的用锂强度结构性下降。", "summary": "针对4月新能源渗透率突破60%但绝对销量同比下滑19%的矛盾，指出电池材料需求已陷入“分母陷阱”。当前渗透率的上升主要源于燃油车市场更大幅度的萎缩，而非新能源需求的实质增长。由于PHEV占比提升导致的单车锂强度稀释，以及国内零售市场的绝对值收缩，材料需求模型应从“国内渗透率导向”切换为“全球GWh…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美国本土仅有约4条大型电力变压器（LPT）制造产线，分别为日立能源、GE Prolec、Virginia Transformer、SPX Transformer Solutions。", "summary": "美国本土仅有约4条大型电力变压器（LPT）制造产线，分别为日立能源、GE Prolec、Virginia Transformer、SPX Transformer Solutions。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "反内卷2.0:供给纪律推动是否在光伏、新能源车、锂电板块扭转利润曲线", "summary": "反内卷政策已从 2024 年的修辞,经 2025 年的 MIIT 与行业协会机制,走到 2026 年一季度的早期证据。光伏是最干净的受益者:CPIA 的 0.68 元/W 参考成本将组件 ASP 稳定在 0.70–0.75 元/W,新公告多晶硅项目同比骤降 70%+。锂(碳酸锂回到 8 万–9.5…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 3, "neutral": 1, "value": 36.4, "events": [ { "title": "高渗透率低增速背景下锂电板块估值逻辑重塑：对\"盈利能力+海外份额\"框架的压力测试 分析师： A股策略师（ashare strategist） 立场： 压力测试 日期： 2026年5月…", "summary": "高渗透率低增速背景下锂电板块估值逻辑重塑：对\"盈利能力+海外份额\"框架的压力测试 分析师： A股策略师（ashare strategist） 立场： 压力测试 日期： 2026年5月4日 执行摘要 前两张卡片已确立：中国60%+新能源渗透率是以分母收缩（燃油车市场萎缩）为主驱动的视觉里程碑，而非需…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "日报综合研究 — 2026年5月4日 中国新能源汽车渗透率议题：三张卡片整合摘要 综合分析师： 日报总编 立场： 综合 覆盖卡片： 第1张（汽车行业分析师）、第2张（材料行业分析师）…", "summary": "日报综合研究 — 2026年5月4日 中国新能源汽车渗透率议题：三张卡片整合摘要 综合分析师： 日报总编 立场： 综合 覆盖卡片： 第1张（汽车行业分析师）、第2张（材料行业分析师）、第3张（A股策略师） 本轮议题： 2026年4月中国新能源汽车渗透率突破60%，但终端需求仍偏弱 执行摘要 三位分…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国电网侧储能是新能源汽车电池分析框架中系统性遗漏的需求通路，2026年需求约120–155 GWh", "summary": "中国电网侧储能是新能源汽车电池分析框架中系统性遗漏的需求通路，2026年需求约120–155 GWh", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 油价冲击对CNH的影响", "summary": "致 外汇策略师： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: 480c6f05-bd09-4eb3-b510-cfd771ed7b5b 结果摘要: Wrote handoff_response.md (8,813 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "一体化“新能源+LDES+VPP+AI负荷”模式可使中国电力央企项目IRR从6.5-8.0%提升至8.0-10.0%", "summary": "一体化“新能源+LDES+VPP+AI负荷”模式可使中国电力央企项目IRR从6.5-8.0%提升至8.0-10.0%", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-08", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "港股互联网巨头BAT（百度、阿里巴巴、腾讯）利用已有锁定能源成本的数据中心，以低于竞争对手的边际成本提供AI服务", "summary": "港股互联网巨头BAT（百度、阿里巴巴、腾讯）利用已有锁定能源成本的数据中心，以低于竞争对手的边际成本提供AI服务", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-11", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "比亚迪（002594.SZ）是拥有垂直一体化电池业务的国产新能源车企。", "summary": "比亚迪（002594.SZ）是拥有垂直一体化电池业务的国产新能源车企。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "4月数据同时验证制造补库韧性与结构性能源脱钩，强化中国近端贸易/汇率/峰会姿态，约束被前置到亚洲炼厂与OPEC+而非中国本身。", "summary": "2026年4月中国出口+14.1%、进口+25.3%但原油进口-20%、成品油出口降至十年低点，是制造补库韧性与结构性能源脱钩两者并存：进口井喷集中在半导体、工业金属、农产品，属资本开支与补库驱动而非消费驱动；原油-20%是库存释放+炼油毛利压缩+新能源车需求结构性退坡三力叠加。这一组合让贸易顺差…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "2027–2028 年公用事业稀缺租金属于同时握有排队号、变压器与可核查碳账本的环节——电网设备 OEM 与受监管 T&D 是最高确信度超配。", "summary": "支持 Research note 06 的政策综合判断：2027–2028 年公用事业 alpha 来自并网权、CBAM 可核查碳账本与可靠容量权益，而非新增装机 MW。第一梯队是订单排至 2028 年的高压电网设备 OEM（GEV、Siemens Energy、Hitachi Energy、Pr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 1, "risks": 0, "neutral": 0, "value": 59.8, "events": [ { "title": "A股EV组合应以动力及储能电池（45-50%）为核心底仓跨越去库窗口，整车（30-35%）需等待出口纪律验证，材料（15-20%）仅作对冲，并严格执行量价红黄灯止损。", "summary": "截至2026-05-17，基于另类数据、材料成本与能源宏观的综合研究，确立A股EV组合执行权重：电池与储能作为核心底仓（45-50%），整车战术观望（30-35%），上游材料维持战术对冲（15-20%）。必须严格遵守5-6月出口量及锂价的止损纪律。", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 2, "neutral": 0, "value": 41.2, "events": [ { "title": "IEA《Electricity 2026》指出，全球超过 2,500 GW 的新能源、大负荷与储能项目正卡在电网排队中；到 2030 年，年度电网投资需从目前的约 4,000 亿美元…", "summary": "IEA《Electricity 2026》指出，全球超过 2,500 GW 的新能源、大负荷与储能项目正卡在电网排队中；到 2030 年，年度电网投资需从目前的约 4,000 亿美元提升约 50%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "风险评估报告：AIDC 能源脆弱性对算力资产估值与资本开支稳定性的冲击评估（2027 压力测试） 日期： 2026 05 18 分析师： chief risk (首席风控官) 立场：…", "summary": "风险评估报告：AIDC 能源脆弱性对算力资产估值与资本开支稳定性的冲击评估（2027 压力测试） 日期： 2026 05 18 分析师： chief risk (首席风控官) 立场： 压力测试 (Stress test) 核心摘要 本报告针对中国核心 AIDC（AI 数据中心）运营商进行定量压力测…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 3, "neutral": 2, "value": 42.6, "events": [ { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "FERC 的政策走向强化 AI 向能源富集区迁移，但削弱纯 BTM 绕行套利；物理资产主线更稳，监管绕行溢价更低。", "summary": "截至 2026-05-22，本卡压力测试但不推翻卡 07：FERC BTM/共址风险不是全面禁令，而是从一次性绕行转向 tariff 化、可靠性研究、最低电网成本分摊和州法过滤。AI 算力集群向能源富集区迁移的方向更确定，但“BTM 可逃离电网经济账”的溢价应下修。基准情景是 PJM 形成付费共址…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力市场化深化对水电核电长协是温和侵蚀而非崩塌,中性悲观情景下政策让利路径净利侵蚀(-4.9%至-7.7%)已超过燃料价格路径,卡片02压舱仓位中枢维持但应预留-2pp政策缓冲。", "summary": "本卡对卡片02—04的核心防御结论做政策视角压力测试。2026-05-22时点,136号文(新能源全面入市)、煤电容量电价、跨省送电基准+浮动、绿电直供四条改革线索叠加,正在对水电核电长协形成温和而非崩塌式的让利压力。自测算显示:在中性悲观情景(跨省协议让3分钱+现货占比抬至20%+工商业让利)下…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "两融资金去杠杆已通过账户交叉质押与VaR风控限制从TMT扩散至国防军工、新能源等高拥挤成长板块，必须通过‘分化杠铃’策略将大仓位配置于长江电力等红利安全岛、将科技仓位战术收拢至受国家…", "summary": "本卡片系统综合了A股去杠杆负螺旋对全市场的跨行业传染风险。我们测算指出，散户账户交叉质押与VaR风控触发将两融出清压力从TMT扩散至国防军工、新能源等高拥挤成长板块。科创50指数1,750点关口面临882亿元密集敲入与平仓卖压，但在国家队专项买盘防护下形成有底成长资产。我们建议采取‘分化杠铃’策略…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 1, "risks": 1, "neutral": 0, "value": 58, "events": [ { "title": "库存触阈值支持交易性底部，但锂、SiC和存储器证据不支持下周A股汽车板块出现全面利润拐点。", "summary": "截至2026-05-24，本卡从材料端修正前序链路：库存阈值可能带来A股汽车交易性底部，但不支持利润基本面已经拐头。CADA显示2026年4月VIA为62.1%、库存系数1.89，经销商库存约260万辆；5月1-17日新能源零售环比增18%但同比降12%。材料端看，电池级碳酸锂约178000元/吨…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "港股微观流动性优于A股，A股存量博弈特征显著且面临半导体等板块的解禁冲击，建议维持哑铃型配置策略。", "summary": "更新了截至2026年5月22日的A股与港股周线终值，指出港股在南向资金支撑下跑赢A股。两融余额触底回升至1.54万亿，下周A股面临约780亿元的限售股解禁压力，主要集中在半导体与新能源整车板块。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 3, "neutral": 0, "value": 39.4, "events": [ { "title": "电池安全溢价与补贴不确定性将抬升美欧储能LCOS高达14-29%，导致装机进度延误15-20%并引发并网弃电率飙升，成为阻碍美欧2030/2035年清洁能源目标达成的核心硬约束。", "summary": "本卡片评估了$30-40/kWh的电池安全溢价与补贴退坡风险对美欧储能及能源转型的冲击。在极端压力测试下，美国储能LCOS将飙升29.2%至$142.92/MWh，引发15-20%的装机延误，导致高比例新能源电网弃电率升至12-18%，最终拉大美欧2030/2035年可再生能源部署目标的缺口。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "特种材料成本飙升并未导致一线厂商利润结构性崩溃，反而通过“可融资性”过滤器强化了头部议价权，但传导后的高成本正成为新能源装机速度的实质性阻碍。", "summary": "2026年石油冲击导致EVA/POE树脂价格飙升至$2600-$3000/吨，光伏组件和锂电池成本分别增加约3-7%和$3-$7/kWh。尽管成本大涨，一线厂商通过供应联盟、技术壁垒及可融资性优势，成功将成本压力传导至下游，维持了12-15%的毛利率。然而，高昂的组件价格与高利率共振，导致终端电站…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在极高的主权债务和通胀压力下，全球主要经济体无法为受阻的新能源项目提供约2100亿美元的额外财政兜底；政策层正被迫放弃激进补贴，转向务实妥协并隐性推延2030年碳减排目标。", "summary": "全球滞胀脉冲下，政府面临严重的财政约束，无法通过扩大补贴来对冲7.5%的WACC和激增的物料成本。压力测试表明，填补IRR缺口每年需新增约2100亿美元补贴，这在极高赤字环境下不可行。美欧中政策制定者正被迫接受新能源项目FID的延期，实质性地放宽2030年脱碳时间表，将政策重心转向供应链安全与电网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 2, "value": 58.8, "events": [ { "title": "材料闪崩后的利润率红利呈现短周期自动化最强、新能源硬件被传导稀释、航空航天被现金流拖累的阶梯式分化。", "summary": "截至2026-05-25，本卡片综合并收窄卡片05结论：材料金融性闪崩不是制造业利润率的普遍利好。机器人和短周期自动化因BOM调整快、报价周期短，可获得约+0.3至+1.6个百分点的净订单利润率缓冲；新能源硬件因电池与光伏过剩产能和客户重定价，净影响约-1.0至+1.0个百分点；航空航天因长周期b…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "光伏与电池「客户重定价」是结构性+政策性的二轮深挤压，2026 一线组件毛利率压力中枢 3.5%-6.5%，新能源硬件是空头平仓真空过后最先掉队的 cluster。", "summary": "本卡片（2026-05-25）压力测试 card-06 对新能源硬件「-1.0~+1.0 pct」温和稀释的判断。在产能利用率<65%、长协重签价格中枢已落至 0.10-0.11 USD/W、美国 FEOC + 14.25% 关税无法转嫁、欧盟 NZIA 提升非价格权重的多重重定价压力下，一线 T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026Q3应持有“利润留存中国”并对冲“重定价中国”：自动化和现金流防御资产增持，新能源硬件、材料beta和科创拥挤动量减持，基准对冲比率0.45x。", "summary": "截至2026-05-25，本卡综合1-7卡后认为：负Gamma平仓真空之后不应简单卖出全部中国权益，而应增持能留住降本和现金流的A/H资产，减持会被客户重定价和流动性冲击伤害的板块。2026Q3建议超配自动化、机器人、电网、公用事业、精选现金流充裕港股互联网；低配光伏组件、商品化电池、锂铜材料be…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 1, "value": 57.2, "events": [ { "title": "电力设备瓶颈与机柜功率跃迁正在使'东数西算'政策栈出现四处机制裂缝，发改委大概率以补丁方式而非全面重构方式回应，节点准入条件与能源保障机制将在H2-2026收紧。", "summary": "卡片06对'东数西算'政策做政策维度的压力测试。结论：在120 kW/柜跃迁与变压器/开关柜18-24个月交付瓶颈下，现行政策栈出现四处机制裂缝——PUE门槛对液冷集群失效、绿电就近消纳对AI训练负荷无法做小时级保障、节点准入缺少设备交付函、跨省算力与电力结算分属两套监管。建议四条政策补丁：升级准…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国炼厂的弱势是春检与利润深亏的周期性压制，以及新能源车与LNG重卡结构性替代的共同结果；若开工率维持低位，将成为阻断美国二次通胀传导并平抑全球油价的宏观安全阀。", "summary": "2026-05-27卡片分析：中国4月原油加工量同比下降5.8%至5465万吨，炼厂因高成本与春检面临深亏，产能利用率降至63.59%。5-6月若开工反弹，原油增量将推升Brent破$100（概率70-75%），传导美国二次通胀（概率65-70%），迫使美联储6月释放鹰派信号；若开工持续低迷，则B…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "发改委电价现货化、DRC钴配额收紧与主权收储、以及全国碳市场扩容后的碳价上涨，将130-195bps二阶拖累塑造成长期的结构性多年税，并在先进制造链内部引发了中游低壁垒受损与上游/L…", "summary": "截至2026-05-27，本卡对Research note 08提出的130-195bps制造业二阶成本拖累进行了压力测试。政策内生性评估表明，发改委394/136号文确立的电力现货化交易与新能源上网电价入市彻底终结了分时电价行政性避风港，DRC战略矿产局（ARECOMS）在4月确立的钴出口配额与…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "H2 2026新能源成品本土化对中国宏观的核心影响不是贸易顺差断裂，而是高端出口升级叙事的边际降温和人民币预期管理压力上升。", "summary": "截至2026-05-30，本卡综合卡-03结论：成品本土化会打击中国“新三样”成品出口，但宏观冲击是边际而非系统性。基准情形下，H2 2026狭义新三样出口缺口约980亿-1630亿元人民币，严重情形约1630亿-2280亿元人民币；对应工业增加值增速拖累约0.10-0.20个百分点，人民币外汇流…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 0, "neutral": 1, "value": 67.4, "events": [ { "title": "I have systematically evaluated the power grid equipment manufacturing sector's export orde…", "summary": "I have systematically evaluated the power grid equipment manufacturing sector's export order visibility, gross margin stability, and geopolitical imm…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-31", "supports": 1, "risks": 3, "neutral": 0, "value": 55.8, "events": [ { "title": "TMT 去杠杆向光伏与电池扩散的条件已经成熟，6 月首周（6/2–6/6）若光伏设备+电池融资余额连续两日下降且龙头单日净卖出 ≥10 亿元，则板块级去杠杆事件成立；基本面无法承接抛…", "summary": "本卡为能源行业卡片，支持卡片 03 关于 TMT 去杠杆向新能源链扩散的判断。截至 2026-05-31，光伏设备 + 电池板块两融余额合计 1520.21 亿元（其中电池 962.42、光伏设备 557.79）[自测算][S1]，占卡片 03 所述相邻篮子的 38.5%。新能源链与 TMT 在四…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "光伏与电池上游主材已处于现金成本线附近并出现自律减产+去库的弱企稳，可在去杠杆情景下提供 −10%~−15% 量级的'软地板'，但不足以反转 6 月首周的杠杆清算方向，卡片 04 的…", "summary": "本卡对卡片 04 关于'新能源材料基本面无法承接 TMT 去杠杆抛压'的判断进行压力测试，立场为 confirm-with-nuance。截至 2026-05-31，致密多晶硅价格 35–42 元/公斤、电池级碳酸锂 7.6–8.0 万元/吨，均处一线企业现金成本附近；CSIA 2026 年自律减…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "公用事业链不是低关注度故事，而是煤电容量电价+ETS 扩容+现货市场三重重定价的早期窗口——可调度资产估值结构性上修，新能源 IPP 估值结构性下修。", "summary": "公用事业链 30 日激活率仅 11.7%，但正同时穿越三大体制切换：煤电容量电价 2026-01 提至固定成本 50%（高转型省 70%）；H1 2025 弃光 6.6%/弃风 5.7%、国家红线放宽至 10%；ETS 扩容至钢/水泥/电解铝（覆盖 +30 亿吨 CO₂e）。可调度容量被向上重估、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-29的高低切换是防御性拥挤度重置而非再通胀式扩散——上游周期品集体缺席意味着价值买盘存在天花板，其持续性取决于近端宏观需求能否拐头。", "summary": "从材料行业视角对Research note 2的“健康高低切换”定性进行压力测试。结论：2026-05-29的轮动方向性正确，但属于防御性收益挖掘而非再通胀式扩散——铜、螺纹钢、碳酸锂、基础化工等上游周期品均未参与价值买盘，电力板块的上行由高股息驱动而非用电量增长驱动。这意味着价值轮动的可持续性被…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 1, "risks": 7, "neutral": 1, "value": 36.2, "events": [ { "title": "BIOSECURE放大的不是全面脱钩本身，而是跨行业来源风险定价模型：半导体和新能源的直接脱钩已由既有规则驱动，BIOSECURE使采购和数据合规成为可迁移的估值折价因子。", "summary": "截至2026-06-01，本卡压力测试前序BIOSECURE脱钩叙事：BIOSECURE并非半导体和新能源脱钩的第一张多米诺骨牌，因为半导体已受BIS出口管制和对外投资限制约束，电池/EV已受IRA FEOC规则约束，关键矿物已出现中国反制。BIOSECURE的真正外溢在于把实体清单、采购资格、拨…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "省级电力治理已通过'PUE差别电价挤东部+低到户电价吸西部+绿电直连穿透溢价+算力券补消费'的四层分流框架，使AI数据中心承担分层电价与维护工业竞争力两个目标兼容，构成P/Watt杠…", "summary": "截至2026-06-02，地方政府与省级电力监管已在制度上允许AI数据中心承担分层电价，且未削弱区域工业竞争力。机制是'东数西算+绿电直连+PUE差别电价+算力券'四层地理-用途分流：东部以北京2026年PUE>1.35加价0.2-0.5元/千瓦时挤出低效数据中心、保护工业额度；西部以0.32-0…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "公路运价通缩+冷链覆盖完成确为餐饮‘量增价缩’模式提供了70-110bps的真实宏观成本垫，支撑窗口可见到2028年中，但运力反转最早2027年中即可触发该垫子的第一条边界。", "summary": "2026-06-01判断：公路物流运价处于结构性+周期性双重下行(CCFI-Road中枢由2024年103.6降至2025年约102.5-103.0，整车吨公里较2021峰值累计降15%-20%)，新能源重卡渗透突破22%、数字货运平台压低空驶率至30%-32%构成不可逆通缩。冷链方面，2024年…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "新能源间歇性是真实但非唯一约束；具备并网能力公司的扩张核心瓶颈是可调度容量、输电通道、稳定供应合约三者同时出清，重估溢价应锁定同时持有这三类资源的运营商。", "summary": "压力测试结论：新能源间歇性和储能缺口是真实约束，但不是唯一核心瓶颈。AI 算力供给的真正瓶颈是三类资源同时出清——可调度容量（煤、气、核、水电、抽蓄）、并网与输电通道、以及可让超大规模客户签约的稳定合约结构。中国新型储能 73.76 GW / 168 GWh、平均时长仅约 2.3 小时，无法覆盖 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 2, "risks": 3, "neutral": 1, "value": 69.6, "events": [ { "title": "I have finalized the H2 2026 stress testing and synthesis analysis on the transmission of U…", "summary": "I have finalized the H2 2026 stress testing and synthesis analysis on the transmission of US AI hardware tariff inflation to the A share export compu…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI配置范式已从“硅基算力”转向“实物基建”，投资者应超配Merchant IPP与电网OEM，减配面临搁浅算力带来盈利下修风险的TMT硬件龙头。", "summary": "截至2026年6月2日，长达150周的变压器交期与“搁浅算力”风险彻底改变了AI投资范式。在4.65%美债收益率与94美元油价的双重挤压下，配置重心必须从极度拥挤的TMT硬件向电网设备与基荷电力转移。建议超配锁定高电价的Merchant IPP与手握巨额订单的电网OEM，结构性减配面临2027年盈…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国版AI电力溢价的核心不是纯IPP，而是能控制配网接口、负荷合同和源网荷储协同的上市运营商。", "summary": "截至2026-06-02，本卡支持AI主线向电力实物层迁移，但认为中国只能部分复制美国IPPs溢价。最强样板是大唐中卫云基地“物理直供+双边交易”绿电项目：2,000MW一期、87亿元投资、全面投运后满足22.9亿kWh年用电需求。上市表达应优先看三峡水利、涪陵电力等拥有配网和客户接口的企业，其次…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "截至2026-06-03，AI–电力链的确定性溢价来自“在受约束节点交付可靠一度电的权利”，而不是泛能源需求增长本身。", "summary": "截至2026-06-03（Asia/Singapore），本卡支持上一张卡片的“重资产基础设施算力”主线，但把可投资确定性收窄到电网稀缺。AI 数据中心不是简单制造全球能源短缺，而是在 PJM、ERCOT、中国东部负荷中心等节点制造可靠电力、并网、变压器、输电和成本分摊约束。全球数据中心用电由 2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 0, "neutral": 1, "value": 73, "events": [ { "title": "截至2026-06-03，30% BTM是可靠容量情景而非纯清洁电量情景，其结果是加速挤压不灵活基载电源并提升燃气、储能、微电网控制和24/7签约资产的相对收益。", "summary": "截至2026-06-03，30% BTM情景可以成立，但前提是将其定义为数据中心自控的可靠容量，而不是纯光/风/储24/7清洁自供。光伏、风电和储能成本已足以支撑低成本电量层，但4-8小时锂电不能提供季节性可靠容量；现实组合是燃气或其他可调度电源提供可靠性，光/风提供低边际成本电量，电池负责爬坡和…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 5, "neutral": 1, "value": 44.4, "events": [ { "title": "中国更可能把稀土、石墨等关键矿物作为许可型、可校准的谈判杠杆，而不是立刻发动全面禁运；但这已经足以让EV、工业电机和LEO补网承受系统性供应链压力。", "summary": "截至2026-06-05，最可能的中国反制不是全面禁运，而是在2026年11月两个暂停窗口到期后继续使用关键矿物许可、最终用户审查和选择性重启2025年10月稀土/人造石墨方案。对半导体龙头，直接晶圆厂停摆风险低于市场叙事，主要压力在GaAs/GaN、锗、靶材、设备磁材和高可靠认证；对EV、新能源…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器短缺与并网排队拖延将导致缺乏并网保障的新能源资产在2026年后沦为‘搁浅资产’，负值ROIC与债务违约将刺破高杠杆绿电的现金流幻觉。", "summary": "变压器交付延迟与电网并网排队增加将严重恶化新能源资产2026年后ROIC及现金流。压力测试表明，在2年和4.5年并网延迟下，光伏项目ROIC分别从1.99%暴跌至0.49%和-1.10%。因资本化利息上升与PPA重签/取消，EBITDA/DSCR跌破1.0x，引发大面积违约与资产减值风险，新能源资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器短缺与并网延迟带来的双重成本重压已彻底击穿高杠杆绿电项目的财务可行性，资本金IRR暴跌至4%以下，迫使市场将新能源资产的定价逻辑从“低碳度电成本”转向“通电确定性”溢价。", "summary": "变压器短缺与并网延迟正通过抬升关键设备成本和延长IDC累积期，对新能源项目造成重创。定量压力测试显示，在4.5年延迟与80%变压器涨价的极端情境下，陆上风电与光伏项目建设成本（CAPEX）分别激增37.9%和31.0%，建设期利息（IDC）占比飙升至20.6%。在固定权益金情境下，项目杠杆升至76…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "通过将投资权重从高溢价的下游AI应用及高延迟新能源转向最坚硬的物理瓶颈（精炼铜、电网设备和表后保障性公用事业），并结合跨资产多空配对，可将‘通电滞后’风险转化为稳健的阿尔法收益。", "summary": "本报告针对电网瓶颈引发的‘通电滞后’提出资产重配与对冲框架。模拟显示，常规AI组合在电网冲击下收益率将化至-15.30%（夏普-0.97）；而超配精炼铜（30%）、电网设备（25%）与通电保障型公用事业（25%）的优化组合可实现15.25%正回报（夏普0.77）。组合设计了做多铜期货、权益配对（多…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 1, "value": 69.8, "events": [ { "title": "[auto-handoff] AI电力需求与宏观能源政策", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 5598b7eb-cea5-4c87-aef3-488b1dadf13c 结果摘要: Wrote handoff_response.md (14003 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "日元平仓将以风格切换而非指数崩盘的形态外溢至 A 股——卖出北向重仓高 β 成长，买入低 β 高股息与政策对冲板块是当前最高确信度的策略调整。", "summary": "综合 Research note 01–05 的传导链，本卡判断日元套息平仓对 A 股的冲击是非线性、非均质的：基准情景为北向资金 5–10 日累计净流出 600–900 亿元，沪深 300 回撤 5–8%、创业板/科创 50 回撤 10–14%，USDCNH 上探但守住 7.25。冲击本质是全球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 2, "risks": 8, "neutral": 1, "value": 45.2, "events": [ { "title": "监管政策正在通过优化国资委资产负债率考核口径与提供人民银行结构性低息再贷款支持，积极化解电力央企的再融资与绿色转型风险。", "summary": "本卡从政策分析视角评估电力央企债端脆弱性。国资委通过“一企一策”考核和绿色债券带息负债“比例剔除”机制优化资产负债率考核口径，有效缓解永续债赎回压力；人民银行2026年一季度优化的8000亿元碳减排支持工具（提供1.25%低息再贷款）为央企提供低成本信贷，对冲了理财赎回等引发的债券利差走阔风险。监…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026 H2 政策组合不会抬升平均现货电价，而会压低电能量中枢、保留并强化可调节资产的经济价差，水电调节、储能和灵活煤电是最清晰受益方向。", "summary": "工作日 2026-06-06。本卡支持 card-02 的调节性资产 alpha 判断，但修正传导机制：CCER/绿证把新能源价值转入环境权益，容量电价把煤电固定成本回收转出电能量现货。政策修正后，2026 H2 现货中枢预计同比 -6% 至 -10%，峰谷价差 +12% 至 +22%，调节性资产…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI能源溢价应从宽基设备链转向现金可兑现的电网/GIS与稀缺核心部件，液冷和AIDC变压器高beta标的必须先通过2026年中现金流复核。", "summary": "截至2026-06-07，本卡对A股变压器、GIS与液冷链条做估值拥挤度和现金质量压力测试。结论是保留AI能源主题，但不再给予宽基设备篮子统一溢价。思源电气仍是质量最高但估值拥挤的复合龙头；华明装备现金质量更干净，适合质量溢价；平高电气配得上温和GIS/电网溢价；金盘科技需等待2026Q2-Q3经…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AIDC \"表后发电\"(BTM)经济性压力测试：LCOE 上升能否再次压制 ROIC？ 工作日期：2026 06 07（Asia/Singapore） 分析师：能源行业分析师（ene…", "summary": "AIDC \"表后发电\"(BTM)经济性压力测试：LCOE 上升能否再次压制 ROIC？ 工作日期：2026 06 07（Asia/Singapore） 分析师：能源行业分析师（energy analyst） 立场：stress test（对 Research note 01 \"BTM 可将 ROI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 1, "risks": 1, "neutral": 1, "value": 80.6, "events": [ { "title": "评估电力市场化改革与能源价格波动对新能源及传统能源利润率的影响", "summary": "压力测试并部分否定卡片06交来的'新能源对冲火电'命题。136号文（2025-02-09）使新能源全面入市、取消固定电价地板，山东集中式光伏2025出清仅0.225元/千瓦时；同期114号文（2026-01-30）将火电固定成本容量回收比例升至≥50%、约+1-2分/千瓦时，给火电托底。结果是两条…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高技术与装备PMI的扩张已在Q2实物订单数据中得到物理验证，订单结构集中于半导体设备、新能源车产线与人形机器人零部件三条主线，本质为硬科技资本开支驱动下的结构性、非全面式装备景气。", "summary": "高技术制造业PMI 52.9%、装备制造业PMI 52.1%的扩张已通过设备类新订单—通用自动化—专用机床链条传导至下游。NBS数据显示2026年1–4月工业机器人产量同比+28.9%、4月+25.3%，金属切削机床+10.4%，远超整体制造业；半导体设备龙头北方华创、中微、拓荆Q1合同负债同比+…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "绿电直连+CBAM构成的制度套利正使全球AI算力向中国西南、北欧、海湾三条低碳走廊集中，并把可再生能源成本曲线从单一LCOE改写为LCOE+合规溢价+并网期权的三维曲线。", "summary": "截至2026-06-07，PJM容量费、NDRC绿电直连与欧盟CBAM三者叠加，形成约$30–40/MWh的合规折扣台阶，把全球AI算力的边际产能向中国西南/西北、北欧和海湾三条低碳走廊集中。中国绿电直连园区凭物理溯源+小时级证书在CBAM申报中嵌入碳近零，叠加$44–53/MWh的直供电价，对比…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 1, "value": 74, "events": [ { "title": "绿电溢价与CEA抬升将终结A股电力红利与有色周期的同质化行情，催生“水电/核电红利成长化”与“水电铝阿尔法溢价”的戴维斯双击，同时迫使被动资金重构传统指数权重。", "summary": "截至2026年6月8日，承接上游成本测算，本卡片认为范围二纳入将引发A股公用事业与有色板块的深度估值重塑。4-6分/度的绿电溢价将使大水电/核电从纯“红利”向“红利成长”重估，而火电因面临碳成本与新能源资本开支压力面临股息缩减风险。电解铝板块将从同质化周期贝塔转向结构性阿尔法，水电铝凭借吨铝783…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在 Brent 原油维持在 95 美元以上与强势美元背景下，利润分配严重向具备高价格弹性和美元汇率正暴露的上游开采倾斜，而下游炼化板块则因发改委成品油定价扣减机制与进口汇兑损失而承压…", "summary": "基于 2026-06-08 权威工作日，评估 Brent 油价维持 95 美元以上预期下油气产业链的利润分配弹性及强势美元的影响。分析表明，上游开采具有高正弹性，中海油一季度成本仅 28.41 美元/桶 [S5]，在 95 美元油价下可斩获 66.59 美元/桶暴利 [自测算]，且直接受益于美元走…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-10", "supports": 1, "risks": 2, "neutral": 0, "value": 74.4, "events": [ { "title": "ETS 重启对 003816 是温和右尾、对 600905 是被高估的弹性；核电容量电价缺位与新能源消纳红线下调共同把两家公司的政策分布同向压向不对称左尾。", "summary": "政策压力测试支持卡 2 的利润韧性排序（003816 > 600905），但修正其改革弹性结论。把监管约束拆为四束：风光全面入市、容量电价/辅助服务、ETS 第四履约 + CEA-2025 切换、CBAM 实际计征。对 003816 的核心结论：ETS 重启经煤电边际传递抬升核电市场化电价约 1–…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "All three deliverables written and verified (JSON validates). Research note 04 summary (pol…", "summary": "All three deliverables written and verified (JSON validates). Research note 04 summary (policy analyst, stance: SUPPORT with refinement): I support R…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2026-2027年的政策顺序不是先给新能源右尾，而是先压低消纳保护、再保护核电基荷、最后让CEA-2025收紧传导，因此003816的分布优于600905。", "summary": "截至2026-06-10，本卡综合前序观点并给出政策落地顺序：新能源消纳红线/利用率容忍度已先落地，核电CfD/容量类结算机制在2026年地方试点后最可能第二步扩散，CEA-2025配额收紧虽真实但对利润的显性传导更偏2026年末至2027年。该顺序使003816的左尾因核电CfD而收窄，并在ET…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 0, "value": 73.8, "events": [ { "title": "[auto-handoff] AIDC 电网扩容及变压器交付瓶颈量化", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 3d4e0b90-a5b8-4790-b340-2954e615695f 结果摘要: Wrote handoff_response.md (3819 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 1, "neutral": 1, "value": 78.2, "events": [ { "title": "绿色通胀与电网物理约束正诱发新能源“平价红利”的阶段性逆转，AI基荷需求的高标准使得绿色电力在当前基建瓶境下变得比传统能源更昂贵，从而延缓全球能源转型进程。", "summary": "截至2026-06-13，压力测试显示，尽管新能源理论LCOE仍具竞争性，但受电网升级溢价、变压器交期溢价及FEOC合规成本冲击，美国新能源PPA价格已推升至$64.5-$79.4/MWh。在加入储能以满足AI算力24/7基荷需求后，绿色电力总成本已升至$110-$125/MWh，逆转了相对于天然…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新能源装机的高配速与低可信容量的结构性错配，决定了数据中心负荷扩张实质上是区域可调度装机（火/气/核/抽蓄）的利用率与现货高溢价故事，湖北能源与福能股份为核心受益标的。", "summary": "截至2026-06-13，本卡对2030年前中国发电与储能供给侧匹配数据中心负荷进行了系统性合成分析。每年超3亿千瓦的新能源装机因可信容量极低（夏季高峰仅5-10%），无法直接承接数据中心24小时平稳负荷。供需缺口将集中于东部算力枢纽（广东、浙江、江苏）和水电省份的枯水期（四川）。这显著提升了湖北…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 3, "neutral": 0, "value": 64.6, "events": [ { "title": "算力西迁不会造成变压器/GIS全国性短缺，但会把订单兑现风险集中到2026-2027年合格排产、试验和并网验收窗口。", "summary": "截至2026-06-13，本卡压力测试算力西迁对变压器、GIS和上游材料的冲击。结论是：0.32-0.63GW基准西迁负荷不足以触及中国全国性产能上限，但会叠加国网4万亿元“十五五”投资、南网1800亿元2026年投资、新能源并网和646亿元变压器出口，挤占2026-2027年合格交付窗口。取向硅…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国绿电装机增速够用，但电网扩容（特高压+主变+构网储能）是承接AI算力负荷的真正瓶颈，也是2026–2027年A股能源链最确定的盈利兑现环节。", "summary": "截至2026-06-13，中国绿电装机端继续超预期（截至2026-04风光累计>14.5亿千瓦，2025全年新增约3.5亿千瓦[S5]），但弃风率5.1%、弃光率4.6%[S6]显示消纳与跨省外送是真正瓶颈。IEA口径下全球数据中心用电从2024年415 TWh升至2030年945 TWh[S1]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI大负荷的24/7特性与并网队列中新能源主导的增量存在根本错配，新建基荷电源的长周期迫使云厂商抢夺存量电力资产，加剧公用电网缺口。", "summary": "截至2026-06-14，本卡片压力测试结论：发电增量无法匹配AI基建需求。AI需要24/7稳定基荷，但目前美国并网队列中95%以上为间歇性新能源。核电（含SMR）在2030年前无实质增量，天然气扩建受制于管道许可。这迫使云厂商抢夺存量基荷（如核电背靠背PPA），剥夺了公共电网资源。结论：电力供应…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 3, "neutral": 0, "value": 66, "events": [ { "title": "在3.90%实际R*压力情景下，能源与资源是必要但不充分的A股防御对冲，应精选上游油气、一体化煤炭和有色资源龙头，并将仓位控制在12%-15%。", "summary": "截至2026-06-15，本卡压力测试认为能源与资源可以承担A股防御性对冲的一部分，但不是无约束核心。上游油气、一体化煤炭和有色资源龙头具备更短现金流久期和商品价格传导，能在3.90%实际R*环境下优于长久期主题资产；但中证800能源年初以来已上涨16.37%，煤价存在570-770元/吨政策区间…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027-2030年支撑945 TWh数据中心负荷的最优边际结构是新能源+储能为主、天然气调峰为辅、核电PPA做稀缺补强，而非押注单一电源。", "summary": "2026-06-14工作日结论：在80%+绿电匹配约束下，2027-2030年最优边际供电不是单一电源，而是新能源+储能作为能量主体、天然气作为调峰袖珍、核电PPA作为稀缺高质量基荷。Lazard与EIA口径均显示风光储成本显著低于新建核电和燃气调峰；IEA预计2025-2030年全球新增可再生装…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "监管约束（FERC表前化）与碳核算摩擦（绿证溢价与天然气排放惩罚）将系统性推高“新能源+核电”组合的合规成本，使其真实LCOE远高于单纯发电成本。", "summary": "截至2026年6月14日，针对“新能源+储能+天然气+核电”结构的压力测试表明，尽管其具备LCOE优势，但面临严重合规摩擦。FERC已阻击“表后”核电直供，迫使数据中心转向承担高额过网费的“表前”PPA。同时，国内物理跨省输电瓶颈导致绿电合规高度依赖绿证（GEC），推高了溢价。此外，天然气调峰在2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 1, "risks": 0, "neutral": 1, "value": 96, "events": [ { "title": "功率半导体估值应从“新能源车beta”切换到“每个可送电MW的AI电力基础设施内容量”，但溢价只属于进入800 VDC/SST/DC保护/高密度PSU认证平台的SiC/GaN供应商。", "summary": "截至2026-06-15，本卡支持第06张卡的firm power稀缺主线，并将其延伸到功率半导体估值：SiC/GaN不再只按新能源车和快充周期定价，而应按800 VDC、固态变压器、半导体断路器、高密度PSU和电网/数据中心认证平台中的内容量定价。Infineon、ST、TI、Wolfspeed…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 8, "neutral": 0, "value": 38.8, "events": [ { "title": "评估西部'绿区'新能源产能与特高压扩容在承接AI算力迁移中的电力保障能力与成本约束研究", "summary": "压力测试Research note 03的'西迁'判断：方向正确，但用铭牌GW与特高压槽位衡量绿区承接力是量错了对象。真正约束是可靠(7×24可调度)容量与日内/季节匹配。内蒙古新能源150 GW(+35%)、四川水电破1亿千瓦,但风光容量系数仅~17–25%、四川枯水期(11–5月)塌缩。自测算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心负荷增长最先重估的是负荷中心可控容量和综合能源服务价值，广州发展比纯新能源运营商更能承接广东容量补偿和算电协同红利。", "summary": "截至2026-06-16，本卡支持上一卡对广州发展的优先判断，但将逻辑界定为AI负荷推动的“可靠性与容量补偿溢价”，而非单纯售电量弹性。IEA和国家能源局资料显示，数据中心和算力用电正在抬高峰值负荷与电网瓶颈；广东已上调煤电、气电容量电价，广州发展4.84GW可控火电和6.13GW风光资产使其同时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力瓶颈溢价已被广州发展部分且低效定价——表观便宜但扣非约24.5倍且分红未被主业覆盖，应作为容量哑铃中的低carry可靠性期权小幅配置，而非独立高确信超配。", "summary": "截至2026-06-16，本卡对广州发展(600098.SH)做策略师相对估值与盈利质量检验。表观看似便宜（市盈率约10.4倍、市净率约0.88倍破净、股息率约4.8%），但2025年+33.93%归母增长几乎全部来自18.20亿元土地交储收益：扣非净利仅10.54亿元、同比−35.28%，非经常…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策让广州发展的可靠性期权不归零，但容量补偿偏成本回收、辅助服务和新能源机制偏双向结算，尚不足以把600098.SH重新定价为干净的成长性期权。", "summary": "截至2026-06-16，本卡压力测试广东电力改革对广州发展“可靠性期权”的修复力度。结论是政策有底但不够完整：广东已将煤电容量电价调至165元/千瓦·年，气电分档165/264/330/396元/千瓦·年；国家114号文提供可靠容量补偿路径，但补偿以电能量和辅助服务市场未回收固定成本为基础，且建…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 4, "neutral": 0, "value": 64.2, "events": [ { "title": "验证电力红利资产在极端市场波动下的防御价值", "summary": "压力测试长江电力（600900.SH）的防御价值，结论为「有条件支持」Card-01。2025年归母净利341.67亿、+5.14%，70%派息承诺延至2026-2030。136号文针对新能源、不及水电，反使水电相对更具防御性。自测算显示：即便严重多年期干旱（发电量−12%），股息率约2.81%仍…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "纯水电及新能源因干旱与电价侵蚀面临DPS下行风险，防守端配置应从纯水电转向水电-煤炭一体化（中国神华）杠铃组合以稳定DPS收益。", "summary": "截至2026-06-18，本卡片对电力红利资产在煤价、电价及水情波动下的DPS稳定性进行了压力测试。结果表明：纯火电及纯新能源分红极易受燃料成本与电价折价挤压而受损；纯水电虽防范了成本风险，但易在干旱年份因高运营杠杆导致DPS下挫。唯有煤电一体化的中国神华防守性极佳。我们建议配置策略应由纯水电切换…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BTM燃机与燃料电池凭借18个月的“电力交付速度”优势，在2026年联邦监管去约束与地缘政治引发的能源套利驱动下，已从备用电源演变为AI基础设施的核心基石。", "summary": "截至2026-06-17，BTM燃机与燃料电池已成为填补2026-2030年AI电力缺口的主导方案。GE Vernova、Siemens Energy和Bloom Energy正处于产能扩张超级周期，订单已排至2028年后。受益于2026年初EPA对碳捕集强制令的废除及“临时燃机”许可程序的简化，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国低碳铝扩张的核心制约正在从冶炼产能转向稳定、可追溯、可审计的低碳电力交付能力，电网消纳和绿电PPA稀缺会决定绿色溢价能否兑现。", "summary": "截至2026-06-17，本卡支持前序低碳金属绿色溢价逻辑，但将核心约束从“是否有足够新能源装机”细化为“能否向冶炼负荷稳定交付、认证和审计低碳电力”。中国2026年4月底风光装机已达19.1亿千瓦，但电解铝约13.2 MWh/吨的电耗使省级电网因子、绿电PPA、储能和跨省消纳成为低碳铝规模化的关…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 1, "neutral": 0, "value": 83.6, "events": [ { "title": "由于沿海核电机组投产时空错配、电网火电备份面临二氧化碳排放双控红线及容量电价财务重负、西部新能源大基地受特高压输电负载极限约束，2025-2026年电源侧建设增量无法有效填补绿电波动…", "summary": "对2025-2026年电源侧对智算中心的支撑力进行压力测试。尽管火电、核电与新能源装机在2025-2026年大规模增长，但受限于核电长建设周期与并网安全限制、火电碳排放双控考核及容量/辅助电价高成本、西部绿电特高压通道82%-88%负载极限及火绿捆绑物理瓶颈，电源侧增量无法填补绿电波动缺口。预计实…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 5, "neutral": 2, "value": 67, "events": [ { "title": "辅助服务扩容在物理上面临量价齐升，但在商业模式上因受两部制电价监管、高折旧与高内卷制约，无法成为公用事业板块的估值主升浪锚，其本质是新能源开发商的合规成本与利润分摊。", "summary": "在145元/吨极端碳价及系统惯量缺口扩大15-20%的情景下，辅助服务市场虽迎来物理规模扩张，但因其在盈利模式上面临政策约束与商业痛点，无法成为公用事业板块的估值主升浪锚。抽蓄受制于6.5%资本金IRR上限，调相机缺乏独立商业化路径，而电池储能面临高折旧及竞争内卷。相反，辅助服务扩容将推高新能源开…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AIDC电力并网瓶颈对电力敏感型化工品（如烧碱、黄磷）供给端硬约束的交叉验证", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: 6a16e82a-5277-47a4-bf61-68e0ff5c0f23 结果摘要: Wrote handoff_response.md (11068 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网互连队列与新建发电的缓慢爬坡构成继变压器之后的第二个物理产能瓶颈，进一步坐实表后电源被迫采购属性，强化剔除BE空头、SMR空头减仓至30–40%与超配电气设备多头的组合立场。", "summary": "互连队列是变压器之后第二个、结构上更紧的瓶颈：2024年底美国队列约2.6 TW，约为存量装机两倍；中位排队时长升至约5年，进入队列项目仅14–19%最终商运。即使有变压器，系统影响研究与电网升级分摊也独立带来2–5年通电延迟。自测算2026–2028年美国确定MW年增量仅约32–38 GW，刚好…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "常规电网并网延迟与新能源项目财务崩溃，迫使云厂商将资本开支刚性移向自备微电网与高密液冷配套，配套设施开支上升72.22%对纯算力硬件产生挤出效应，但也实质上支撑并延长了AI物理层设备…", "summary": "截至2026-06-20，本卡片评估了能源供应与并网效率对AI物理层投资周期的支撑性。常规电网3-5年的并网延迟使100MW数据中心贴现NPV折损达51%至80%，迫使云厂商向“表后直供+储能”的微电网转型。虽然配建100MW/400MWh储能使LCOE上升14%至$55.10/MWh，但每小时C…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 4, "neutral": 0, "value": 68.4, "events": [ { "title": "全国能源供给能覆盖AI用电总量，但2026-2027年的瓶颈仍是本地并网、特高压投运节奏与可靠容量定价。", "summary": "截至2026-06-20，本卡片压力测试能源侧能否破解AI数据中心并网瓶颈。结论是：全国电量与新能源扩张足以覆盖AI负荷的总量增量，但特高压、储能和本地变电容量的投放节奏仍不足以完全解决2026-2027年大型集群的节点型并网约束。绿电比例提升本身不必然显著抬高算力成本，单独绿证按2026年4月价…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "面向多年期AI电力资本开支超级周期的配置：刚性容量稀缺确立后的板块配置与风格含义", "summary": "本卡把四环串联瓶颈（封装→电网/变压器→刚性发电）翻译成配置地图。关键事实：四个时钟串联，缓解任一环只把约束移向下游，不压缩通电总时间，故稀缺是结构性、多年期的，而非周期性。配置：超配机电卖铲人（电网/变压器，全球缺口约30%、中国占产能60%、出口+36%、订单排至2027）、刚性发电（核电/铀…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI用电争夺会放大局部绿电消纳压力，但中国更可能经历区域利用率下滑和转型节奏分化，而不是全国能源结构转型被迫倒退。", "summary": "截至2026-06-20，本卡压力测试AI集群争夺变压器、开关柜和配网接入窗口对中国绿电利用率的影响。结论是：全国能源转型不会因设备交付延迟被迫逆转，但区域和时点风险已经升高。2026年一季度风电利用率91.4%、光伏90.6%，已贴近90%政策底线；按利用率反推，当季隐含弃风弃光约573亿千瓦时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力瓶颈与绿电利用率分化对A股电力设备、公用事业和新能源运营商的策略影响", "summary": "承接Research note 02–08全链条，本卡以发改能源〔2026〕688号多用户绿电直连（2026-05-20，优先支持算力设施）为基准情形，重构A股电力链定价。核心：688号把AI负荷从挤压新能源并网的拥塞负荷，翻转为在资源地变现弃电的锚定承购（2026Q1光伏利用率90.6%、隐含弃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 7, "neutral": 0, "value": 51.8, "events": [ { "title": "I have completed the analysis on AI compute infrastructure transmission paths under power a…", "summary": "I have completed the analysis on AI compute infrastructure transmission paths under power and transformer constraints , focusing on the concrete impa…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026-2028年，电池、VPP、灵活数据中心电价和GETs是跨越变压器短缺的桥，不是绕开短缺的旁路。", "summary": "截至2026-06-21，本卡压力测试认为，变压器和电网设备产能正在扩张，但新增工厂、认证、熟练劳动力和长交期决定其主要缓解2029年后的出清，而非完全解决2026-2028瓶颈。储能、VPP、数据中心需求侧管理和GETs更适合作为近端过桥工具：它们可以削峰、递延部分网络升级、支持灵活并网，并提高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "DR007上行会压缩公用事业股息利差并提高再融资成本，但分红削减的核心触发器是资金利率、煤价、电价和资本开支同时恶化，而非DR007单项上行。", "summary": "截至2026-06-21，本卡对电力等“物理红利”资产进行压力测试：DR007从6月1日1.3644%升至6月18日1.4606%，会抬高贴现率和再融资成本，但尚不足以单独击穿公用事业分红。真正风险是DR007接近1.70%、煤价维持861元/吨以上、市场化电价走弱与高资本开支叠加。水电和低燃料b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI硬件已到战术拥挤预警点，电力设备则是结构分化而非整体泡沫；制造出海主线应再平衡而非否定。", "summary": "截至2026-06-21，本卡由chief-strategist完成压力测试。AI硬件的成交、ETF与主力资金拥挤度已接近战术临界点，短期回撤风险上升，但机构超配比例仍低于历史峰值，不能据此全面退出。电力设备并非整体拥挤，风险主要在主题化交易载体和旧新能源β；电网、变压器等出海订单链仍有支撑。策略…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 1, "value": 86.2, "events": [ { "title": "2026-2030年特高压与绿电并网能缓解全国电量约束，却不能完全对冲智算中心的本地可靠容量和电价基差风险，因此仍支持从粗放算力堆叠转向高能效算力、软件应用和电网CapEx受益资产。", "summary": "截至2026-06-22，本卡对前序电力瓶颈判断进行压力测试：特高压、绿电并网、煤电灵活性、抽蓄和新型储能建设足以缓解全国性年度电量缺口，但不能完全消除智算中心在本地并网、24/7可靠容量和市场化电价基差上的风险。国家电网规划“十五五”15回特高压直流工程、跨省输电能力提升35%、每年支撑不低于2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b04fea49-07a7-43ca-b246-32f00d158f50 结果摘要: Wrote handoff_response.md (9863 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 1, "risks": 4, "neutral": 0, "value": 80.6, "events": [ { "title": "电网扩容与AI算力需求足以验证材料链利润传导，但确定性排序为铜资源和电工钢/变压器材料最高，低成本铝次之，普通加工和多数电池材料不具备同等稀缺溢价。", "summary": "截至2026-06-23，本卡支持上一张卡片的AI电力与电网硬科技主线，但把材料链利润传导分层：铜和取向电工钢/变压器铁芯材料的稀缺性最清晰，铝有需求支撑但更依赖低电力成本、区域升水和合同指数化，电池材料不应被机械外推。IEA、国家能源局、国家电网、LME和行业资料共同显示，数据中心用电、电网投资…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "80%绿电强配溢价与电网/可靠容量可获得性，作为AI算力建设的约束性瓶颈及其对电力与新能源投资主题的含义", "summary": "本卡片综合卡片01-03，给出统一判断：AI建设的约束性瓶颈是'可获得性'而非'价格'。80%绿电强配以绿证年度履约成本极低（约+1%电费），即便24/7固化绿电PPA溢价也仅占TCO约2-3%[自测算]，价格可被超大规模厂商接受并转嫁。真正闸控2026-2030建设节奏的是并网排队、可信容量与小…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "绿色先进制造与城投信贷收缩将通过Capex砍减与地方投资减少，向电力设备及能源基建传导共计313.42亿元的需求收缩压力，并延缓光伏组件与储能电芯的价格修复。", "summary": "截至2026-06-23，本报告在金租及信贷收缩背景下评估其对电力设备与能源基建的传导。测算显示，全系统绿色制造信贷收缩将导致相关厂区配电及并网设备需求缩减100.03亿元；城投信贷收缩将导致地方能源基建投资缩减213.39亿元（对应新能源装机延期约6,096.86 MW）。合计拖累总需求313.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "容量电价仍能提供火电红利底线，但136号文使底线从“类债担保”变成“有限损失吸收”，火电龙头应降级为受监管信用资产而非无风险债券替代品。", "summary": "截至2026-06-23，本卡对Research note 04做压力测试：136号文不会机械侵蚀煤电容量电价，因为容量电费按月结算并纳入系统运行费用；但其推动新能源全面入市，会提高电量电价左尾波动。2026年起容量回收不低于330元/千瓦·年固定成本的50%，折算约0.037-0.041元/千瓦…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 2, "neutral": 3, "value": 96, "events": [ { "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply s…", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 新能源消纳新政与公用事业板块配置权重调整评估", "summary": "致 深研主编： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 9a35bd2e-c21c-4de2-aaa5-b1de0430457c 结果摘要: Wrote handoff_response.md (10235 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "订单与招标数据支持资金转向电网设备及新能源实物资产端，但尚未证明二级市场资金已完成全链条重配。", "summary": "截至2026-06-25，本卡支持“资金正从碳资产叙事转向新型电力系统实物资产”的方向，但限定为订单端确认。碳和绿证仍活跃；增量证据在国网特高压招标、电网设备公司中标、光伏组件、风电整机和储能采招。结论是：实物资产资本开支已明显落单，尤其是电网一次设备和特高压材料；但权益资金流、利润率和二级市场持…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "在风光（3-12个月建设周期）对特高压通道（18-36个月）分别具有6.0倍和2.25倍周期时滞的客观规律下，大基地并网瓶颈已导致2025-2026年新能源消纳率首次跌破95%政策红…", "summary": "截至2026-06-25，本报告对我国供给端（风、光、核）建设节奏与电网侧扩容周期的协同性进行压力测试。结果显示，光伏（3-6个月）和风电（~12个月）的建设速度大幅领先特高压外送通道（18-36个月），形成6.0倍和2.25倍的周期差[自测算]。这导致2025年风电光伏利用率在“十四五”期间首次…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 7, "neutral": 0, "value": 57.4, "events": [ { "title": "能源降本对AI算力基础设施的影响及扩容瓶颈评估", "summary": "压力测试卡片-02的能源降本看多论。燃气降本（CCGT减负$25.76/MWh、100MW数据中心年省OpEx约$24.94M）真实但属次要，仅折合AI级capex的1.7%–2.5%。真正约束是time-to-power：美国并网队列卡着约2,300GW、PJM平均排队约8年、Dominion北…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由于发电设备长交期、核电重建缓慢、跨州输电网架建设滞后以及并网容量在地理上的严重错配，电网侧能源增量无法匹配2026-2028年AI负载爆发式增长，电力缺口将实质性压制算力部署进度，…", "summary": "本卡对电网侧增量电源进行压力测试。2026-2028年，电网侧增量无法有效填补AI电力缺口。重型燃机面临5-7年超长交期；绿地核电停滞且SMR商业化迟于2030年，Palisades与三哩岛等重启仅能满足3.4%-10.8%的新增需求；输电线建设长达6.5-10年以上，且新增新能源地理错配严重（主…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国\"西电东送\"输电瓶颈对算力西移的制约：跨省通道容量是否会重演美国电网拥堵？", "summary": "本卡对第08卡\"中国基本化解电力约束\"做压力测试，结论为部分否定、主要重构。一、中国不会重演美国式并网排队拥堵：特高压国家统一规划、约2–3年建成，跨区容量2020→2025由270→340GW(+25%)，在运45条/约300GW，2026–30再规划15条(+35%)[S2][S3][S5]。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与开关柜产能缺口将实质性推迟新能源并网，延缓规管公用事业企业费率基数收益的现金兑现并恶化其资本效率，使发电商面临项目经济性破坏的重大下行风险。", "summary": "截至2026-06-25，本卡对电网瓶颈进行压力测试发现：变压器交期拉长至128-144周，中压开关柜拉长至52-78周，严重推迟新能源并网。受规管公用事业企业面临现金收益兑现延迟2-3年、采购成本暴涨60%-95%拖累资本效率等问题，面临电价审核核减和需求侧流失双重财务风险；敏感性分析表明，12…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 1, "neutral": 0, "value": 94.8, "events": [ { "title": "维持12% NAV清洁基础设施袖，但不要上调中枢；政策兑现支持电网/储能/运营商核心，136号文电价风险和制造产能过剩要求压低光伏/电池制造暴露。", "summary": "截至2026-06-26，本卡综合前序清洁基础设施杠铃：政策和需求数据支持保留12% NAV清洁袖，但不支持上调或追宽基清洁贝塔。国家电网2026-2030年4万亿元投资、全国电网接近5万亿元投资图景、1-5月用电5.7%增长、互联网数据服务用电40%+增长、算力用电到2030年8000亿千瓦时预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "霍尔木兹海峡重开仅为“执行折价”而非“和平红利”，60天物理排雷延迟导致到岸能源价具极强向下刚性，与高利率共同构筑新能源“资本开支壁垒”，系统性阻碍全球能源转型进程并抬升算力用电门槛。", "summary": "霍尔木兹海峡重开虽缓解了原油期货的即期上涨动能，但由于海峡内活体水雷清障导致的60天排雷延迟，到岸能源价格（CIF）维持强韧的向下刚性，地缘风险溢价只是发生了盘面转移。在高利率与绿色通胀双重作用下，全球新能源面临高额WACC与电网并网漫长排队的“资本开支壁垒”，转型进程阶段性受阻，算力能源供应链成…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 1, "risks": 4, "neutral": 0, "value": 87.6, "events": [ { "title": "新动能（AI基础设施与绿电）在体量上仅能弥补房地产投资收缩缺口的约10%-15%，且面临严重的电网并网、输电设备及芯片良率等物理瓶颈限制，无法在短期内有效对冲房地产缩表的宏观增长压力。", "summary": "压力测试表明，新旧动能切换在总量与结构上存在重大错配。1-5月房地产投资下降16.2%带来1.8万亿元的总需求缺口，而电网与AI算力投资年度增量合计约2100亿元，仅能对冲缺口的10%-15%。且AI和新能源建设面临电网并网、输电设备周期以及芯片良率等严重的物理瓶颈，导致资本开支兑现滞后6-12个…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "需求端验证：中国制造业PMI与工业数据对金属品种价格驱动力的分析", "summary": "截至2026-06-29，本卡支持前序黄金/结构性金属对冲框架，但按品种分化：中国需求改善足以验证铜的盈利Beta，核心来自电网、电气化、高技术制造和出口，而非房地产；锂也有电动车、储能和绿色出口支撑，但1-5月新能源汽车销量仅温和增长且锂价近月回撤，说明锂应保持受限高Beta卫星。组合上维持21…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海缆仍是订单背书最强的底仓但需下修新增订单毛利假设；储能是政策背书装机而非利润背书，只有高可靠、长时化、采购纪律强的龙头能把114号文转化为毛利；UDC是固化成本支付方，仍应按主题脉…", "summary": "截至2026-06-29，本卡对“海缆与储能订单背书型利润确定性”做压力测试。结论是分化：海缆订单确定性仍高，东方电缆截至2026-03-25在手订单193.12亿元，但海风补贴退坡和136号文新能源入市会压缩新增订单毛利假设；储能在114号文后装机确定性更强，但容量收益属于合格独立储能项目，不自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "算力密度激增下的能源供给适配：从“规模化能源需求”到“极端局部负载调控”", "summary": "支持Research note 09“天花板下移到机架”的结论并落地能源侧答案。1MW机架的配电层威胁是斜率而非规模：同步GPU训练任务亚秒级摆动达数十至数百兆瓦[S1]，ERCOT大负荷申请均值约725MW、多站超1GW[S2]，发电侧却需1–90分钟响应[S4]。解法是三时间尺度分层缓冲：毫秒…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 5, "neutral": 3, "value": 84, "events": [ { "title": "[auto-handoff] Impact of global capacity pricing spikes on domestic transformer export marg…", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 324722b5-2f7b-4b91-aec7-ac5429f32a39 结果摘要: Wrote handoff_response.md (11765 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 评估PJM等容量电价暴涨对国内出海变压器毛利率的传导", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 1d2b9e99-6767-49f3-aa4b-4520c45f9a0b 结果摘要: Wrote handoff_response.md (9124 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "summary": "对中国，2026年油价/柴油冲击既是被缓冲的进口税、又是战略加速器，且净额上比卡片01–02的美欧日集团损害更小。三大事实：（1）进口税真实但落在PPI而非CPI——5月PPI同比+3.9%（2022年7月来最快）、CPI仅+1.2%，疲弱内需把冲击吸收为生产者利润压缩；（2）进口税被三重缓冲——…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-c6c6be4745/", "networkHref": "/investor/research-network/#thesis-c6c6be4745" }, { "id": "thesis-485a630a87", "source": "tracked_thesis", "title": "自动驾驶与 Robotaxi 主题必须从演示热度进入单位经济和监管城市扩张验证", "titleEn": "Autonomous driving and robotaxi themes must move from demo excitement to unit economics and regulated city expansion", "lane": "自动驾驶/Robotaxi 单位经济", "laneEn": "Autonomous driving and robotaxi unit economics", "laneHref": "/investor/lanes/autonomous-driving-and-robotaxi-unit-economics/", "investableFocus": "自动驾驶/Robotaxi 单位经济：TSLA, GOOGL, BIDU, 9888.HK, XPEV, LI, 2015.HK, 002594.SZ", "stockUniverse": [ { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "GOOGL", "name": "Alphabet", "market": "US", "role": "Waymo 与地图生态", "roleEn": "Waymo 与地图生态", "bucket": "core", "href": "/investor/stocks/GOOGL/" }, { "ticker": "BIDU", "name": "Baidu", "market": "US ADR", "role": "Apollo Robotaxi", "roleEn": "Apollo Robotaxi", "bucket": "core", "href": "/investor/stocks/BIDU/" }, { "ticker": "9888.HK", "name": "百度集团", "market": "HK", "role": "Apollo Robotaxi", "roleEn": "Apollo Robotaxi", "bucket": "core", "href": "/investor/stocks/9888.HK/" }, { "ticker": "XPEV", "name": "XPeng", "market": "US ADR", "role": "智能驾驶量产", "roleEn": "智能驾驶量产", "bucket": "core", "href": "/investor/stocks/XPEV/" }, { "ticker": "LI", "name": "Li Auto", "market": "US ADR", "role": "智能电动车规模", "roleEn": "智能电动车规模", "bucket": "core", "href": "/investor/stocks/LI/" }, { "ticker": "2015.HK", "name": "理想汽车", "market": "HK", "role": "智能电动车规模", "roleEn": "智能电动车规模", "bucket": "core", "href": "/investor/stocks/2015.HK/" }, { "ticker": "002594.SZ", "name": "比亚迪", "market": "A-share", "role": "整车成本曲线", "roleEn": "整车成本曲线", "bucket": "core", "href": "/investor/stocks/002594.SZ/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" } ], "stockTickers": [ "TSLA", "GOOGL", "BIDU", "9888.HK", "XPEV", "LI", "2015.HK", "002594.SZ", "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT" ], "coreView": "赛道：自动驾驶/Robotaxi 单位经济。候选观察标的：TSLA, GOOGL, BIDU, 9888.HK, XPEV, LI, 2015.HK, 002594.SZ。当前共有 40 条证据、3 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 71, "eventCount": 160, "firstSeen": "2026-05-04", "lastSeen": "2026-06-29", "supports": 3, "risks": 136, "neutral": 21, "analystCount": 9, "sourceTypeCounts": { "研究笔记": 61, "分析协作": 13, "证据更新": 4, "研究线程": 75, "研究主题": 3, "深度研究": 4 }, "topTerms": [ "AI 基础设施", "Tesla", "AIDC", "Robotaxi", "Autonomous", "driving", "robotaxi", "themes", "must", "move", "demo", "excitement" ], "whatChanged": [ "2026-06-29: 产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "2026-06-29: 产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "2026-06-29: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "2026-06-29: 产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "2026-06-29: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）" ], "practical": { "sectorLane": "自动驾驶/Robotaxi 单位经济", "sectorLaneEn": "Autonomous driving and robotaxi unit economics", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "GOOGL", "name": "Alphabet", "market": "US", "role": "Waymo 与地图生态", "roleEn": "Waymo 与地图生态", "bucket": "core", "href": "/investor/stocks/GOOGL/" }, { "ticker": "BIDU", "name": "Baidu", "market": "US ADR", "role": "Apollo Robotaxi", "roleEn": "Apollo Robotaxi", "bucket": "core", "href": "/investor/stocks/BIDU/" }, { "ticker": "9888.HK", "name": "百度集团", "market": "HK", "role": "Apollo Robotaxi", "roleEn": "Apollo Robotaxi", "bucket": "core", "href": "/investor/stocks/9888.HK/" }, { "ticker": "XPEV", "name": "XPeng", "market": "US ADR", "role": "智能驾驶量产", "roleEn": "智能驾驶量产", "bucket": "core", "href": "/investor/stocks/XPEV/" }, { "ticker": "LI", "name": "Li Auto", "market": "US ADR", "role": "智能电动车规模", "roleEn": "智能电动车规模", "bucket": "core", "href": "/investor/stocks/LI/" }, { "ticker": "2015.HK", "name": "理想汽车", "market": "HK", "role": "智能电动车规模", "roleEn": "智能电动车规模", "bucket": "core", "href": "/investor/stocks/2015.HK/" }, { "ticker": "002594.SZ", "name": "比亚迪", "market": "A-share", "role": "整车成本曲线", "roleEn": "整车成本曲线", "bucket": "core", "href": "/investor/stocks/002594.SZ/" }, { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" } ], "stockTickers": [ "TSLA", "GOOGL", "BIDU", "9888.HK", "XPEV", "LI", "2015.HK", "002594.SZ", "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT" ], "primaryStocks": [ { "ticker": "TSLA", "name": "Tesla", "market": "US", "role": "FSD/Robotaxi 高 beta", "roleEn": "FSD/Robotaxi 高 beta", "bucket": "core", "href": "/investor/stocks/TSLA/" }, { "ticker": "GOOGL", "name": "Alphabet", "market": "US", "role": "Waymo 与地图生态", "roleEn": "Waymo 与地图生态", "bucket": "core", "href": "/investor/stocks/GOOGL/" }, { "ticker": "BIDU", "name": "Baidu", "market": "US ADR", "role": "Apollo Robotaxi", "roleEn": "Apollo Robotaxi", "bucket": "core", "href": "/investor/stocks/BIDU/" }, { "ticker": "9888.HK", "name": "百度集团", "market": "HK", "role": "Apollo Robotaxi", "roleEn": "Apollo Robotaxi", "bucket": "core", "href": "/investor/stocks/9888.HK/" }, { "ticker": "XPEV", "name": "XPeng", "market": "US ADR", "role": "智能驾驶量产", "roleEn": "智能驾驶量产", "bucket": "core", "href": "/investor/stocks/XPEV/" }, { "ticker": "LI", "name": "Li Auto", "market": "US ADR", "role": "智能电动车规模", "roleEn": "智能电动车规模", "bucket": "core", "href": "/investor/stocks/LI/" } ], "instrumentTags": [ "TSLA", "Tesla", "GOOGL", "Alphabet", "BIDU", "Baidu", "9888.HK", "百度集团", "XPEV", "XPeng", "LI", "Li Auto" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：证据密度足够", "deepResearchReason": "证据深度 100，来源 6 类，近期事件 105 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 6, "analystCount": 9, "recentEvents": 105, "supportRatio": 0.02, "riskRatio": 0.85, "conflictScore": 96.4, "evidenceDepth": 100 }, "score": 62.9 }, "events": [ { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14, "matchedTerms": [ "AIDC", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck from compute to power, networkin…", "sourceType": "topic", "sourceLabel": "研究主题", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "AI 基础设施", "infrastructure" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-28", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,600 亿）以经营现金流支撑，资金基础/地域/标的均不同于承…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.75, "matchedTerms": [ "AI 基础设施", "infrastructure", "wars" ], "href": "" }, { "date": "2026-06-28", "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.en.md (7867 bytes) — The repor…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 15.5, "matchedTerms": [ "city", "infrastructure", "investment", "cost", "auto", "price", "software" ], "href": "" }, { "date": "2026-06-28", "title": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "summary": "截至 2026-06-28，本卡压力测试 Research note 1 的 AI 基础设施使能层主线：需求真实且强劲，但 2026-2028 年兑现会被电网接入、可靠电源、变压器、燃机和运行可靠性约束。IEA 预计全球数据中心用电从 2024 年 415 TWh 升至 2030 年约 945 TWh；美国和中国合计贡献近 80% 增量。[S1][S2] …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "AI 基础设施", "infrastructure" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 12, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk", "cover" ], "href": "" }, { "date": "2026-06-27", "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期：…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 / Executive Summary 截至 2026 年 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 16, "matchedTerms": [ "AI 基础设施", "GOOGL", "infrastructure" ], "href": "" } ], "timeline": [ { "date": "2026-05-04", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "AI 数据中心电力瓶颈已经成为 hyperscaler capex 回报分化的核心变量：已锁电力的 capacity 是稀缺资产，缺乏 firm power 的宣布式 capex 应…", "summary": "本卡 stress-test 卡片 01 的选择性建设性结论：AI capex 仍有需求支撑，但电力已经成为地点、时间和边际成本上的真实约束。DOE/LBNL 预计美国数据中心用电量从 2023 年 176 TWh 升至 2028 年 325-580 TWh；IEA 预计美国到 2030 年电力需…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "在 AI 能源利率冲击成为已知情景后，权益组合应从纯成长 AI 转为价值-质量-硬资产-精选 AI 基础设施框架，以承载五腿对冲并保留可验证的 AI alpha。", "summary": "本卡将第08卡联合无序尾部纳入基本配置，而非事后保险。建议全球权益风险预算按美股58、A股42配置，对冲后净权益beta为0.75-0.85，并以每季度65 bps NAV承载五腿结构性对冲。A股超配电网、电力、有色、AI基础设施和高股息SOE金融；美股超配AI硬件、电气设备、公用事业、能源和质量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施尾部风险应被视为单一相关信用-股权敞口：未对冲等权组合每 $100 的 99% VaR 约 $78，风险控制优先级是减 CRWV 股权、按压力抵押品价值计量 CoreW…", "summary": "本卡对 CRWV 股权、CoreWeave 有抵押债务、NBIS 股权与 OpenAI 晚期股权组成的 $100 等权参考组合进行三节点级联压力测试。结合 Research note 07 与最新公开披露，CoreWeave 2025 年客户 A 集中度升至 67%、RPO 达 $60.7B，Op…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "[auto-handoff] Northbound Flow Analysis", "summary": "致 中国宏观分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 2cd022da-76a7-4d93-9ae5-a417f3e6120d 结果摘要: Wrote handoff_response.md (8,682 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 1, "value": 43.2, "events": [ { "title": "阿里巴巴 (BABA) 承诺投入 550 亿美元用于 AI 基础设施建设。", "summary": "阿里巴巴 (BABA) 承诺投入 550 亿美元用于 AI 基础设施建设。", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 主权算力财政支出影响", "summary": "致 TMT行业分析师： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 1f2ac627-aa0e-4991-81c1-4c46ae106c06 结果摘要: Wrote handoff_response.md (17,200 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-12", "supports": 0, "risks": 1, "neutral": 1, "value": 44.6, "events": [ { "title": "在 Brent 原油 150-120 美元情景下，云/AI 基础设施板块的预估利润率侵蚀为 180 - 250 bps。", "summary": "在 Brent 原油 150-120 美元情景下，云/AI 基础设施板块的预估利润率侵蚀为 180 - 250 bps。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI 基础设施作为新久期资产的宏观耦合", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 3ae37789-24be-4d0d-9c10-eca5ac01b789 结果摘要: Wrote handoff_response.md (3223 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 2, "neutral": 0, "value": 37, "events": [ { "title": "AI 基础设施正从‘无风险增长’阶段进入‘社会许可防御’阶段，追溯性立法风险可能引发行业 20% 以上的系统性估值修正。", "summary": "本报告对 AI 基础设施在核心区域（VA, AZ, IE）面临的‘农业-社会’风险进行了压力测试。基于 640 亿美元受阻管线，模拟了追溯性立法撤销用水/用地权及专项农业附加费的情景。结果显示，极端情景下主要 DC REITs 可能面临 15-22% 的回撤，1 日 VaR 将从 185bps 飙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施估值锚必须从‘已公告 MW’迁移到由自筹水、自筹电网、社区溢价、立法免疫力与追溯风险缓冲构成的‘社会合规确定性’（SCC）评分，MW 退化为产能上限而非价值驱动。", "summary": "截至 2026-05-14，本卡压力测试并重构 AI 基础设施估值模型：以已公告 MW 为锚已结构性失效，应被‘社会合规确定性’（SCC）取代。SCC 由自筹水（25%）、自筹电网（20%）、社区溢价（20%）、立法免疫力（20%）、追溯风险缓冲（15%）五因子构成；公式为 EV/通电等效-MW …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 2, "neutral": 1, "value": 41.4, "events": [ { "title": "麦肯锡预计 2025 年 AI 基础设施投资需求为 5.2 万亿美元，其中 1.3 万亿美元用于电力、输电、冷却等领域。", "summary": "麦肯锡预计 2025 年 AI 基础设施投资需求为 5.2 万亿美元，其中 1.3 万亿美元用于电力、输电、冷却等领域。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年5月10日，分析师指出应将组合从“开发商权益+ESA容量beta”转向“受监管电网/水务/天然气赋能敞口+锚定型核心资产”。", "summary": "2026年5月10日，分析师指出应将组合从“开发商权益+ESA容量beta”转向“受监管电网/水务/天然气赋能敞口+锚定型核心资产”。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 中美高层会晤经贸预期", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 政策分析师。 模式: adhoc 会话: d4783cbf-8f3f-4aca-be4e-9bc54f3bfd56 结果摘要: Wrote handoff_response.md (10824 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "AI/半导体应从拥挤动量超配降至更小的质量核心，而新增风险预算应转向电力交付、电网设备和铜铝瓶颈。", "summary": "2026-05-17：本卡综合 Cards 01-05，支持削减 AI/半导体拥挤动量，但不退出 AI 基础设施主题。建议将战术 AI/半导体仓位削减 15-20%，保留质量型 AI 核心，并把风险预算转向电网设备、变压器/开关设备、特高压/HVDC、电网自动化、能源接入型公用事业、铜铝和回收。核…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "AI 交易逻辑转向：从算力建设到推理侧需求", "summary": "AI 交易逻辑转向：从算力建设到推理侧需求 Does the recent surge in AMD and the shift toward 'Agentic AI' signify a permanent structural pivot in the AI investment thesis…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 3, "neutral": 2, "value": 42.6, "events": [ { "title": "[auto-handoff] Robotaxi unit-economics 转正是否构成新增长主题入场点", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 1bd70ab2-2a18-41f3-8cfb-347ac81abd61 结果摘要: Wrote handoff_response.md (3621 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "自动驾驶单位经济拐点是否够格成为 2026 H2 主题", "summary": "自动驾驶单位经济拐点是否够格成为 2026 H2 主题 Is the robotaxi unit-economics inflection a 2026 H2 thematic Waymo Phoenix 已早于卖方建模 4–5 个月跨过 50k 单日订单 unit-economics 阈值。如果…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "卖方 Capex 下修是'交付排期推后'的财务表象，物理瓶颈是它的底层成因——同一个事实的两面，操作上要回补 TMT 半导体并保留电力腿，两条腿都做。", "summary": "卖方对 MSFT/META/GOOGL 2H26 Capex 下修，是把'承诺 Capex'与'现金 Capex'两条线混为一谈的财务误读，而误读的机械原因正是能源台所指的物理瓶颈——变压器交期、互联排队、燃机订单簿稀薄把现金落地推后。因此首席策略师与 energy-analyst 的归因不是二选…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 算力资本开支、电力瓶颈与通胀再定价", "summary": "AI 基础设施需求正在先通过电力、电网、变压器与材料供给形成成本脉冲，再通过生产率与架构效率形成中期缓释；投资结论取决于这两个方向谁先兑现。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-power-inflation-transmission-0c9ba15c/" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 4, "neutral": 1, "value": 35, "events": [ { "title": "截至 2026-05-19，AI 基础设施仍是 TMT 领涨主线，但可投资范围应收窄到有真实 usage、backlog、产能瓶颈和现金流承受力的环节。", "summary": "工作日期 2026-05-19。本文否定前序空白 TMT 论点并选择 (A)：AI 基础设施仍是领涨主线。证据是 hyperscaler capex/收入比虽已升至约 30% 的四家公司 Q1 合计强度，且 Alphabet、Amazon 的 FCF 张力明显，但 Microsoft AI ARR…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "消费电子'技术性弱复苏'+AI 手机溢价被压在旗舰段，触发的是 A 股'叙事溢价 → 确定性溢价'的哑铃化再定价，而非全面价值回归或单边 AI 基础设施集中。", "summary": "工作日期 2026-05-19。综合 card-03（AI 基础设施仍是窄口径主线）、card-04（消费电子 2026H2 技术性弱复苏）与 card-05（以旧换新净撬动仅约 1.25x、AI 手机溢价仅 ≥6,000 元旗舰段）三条结论后，本卡的策略级判断是：A 股既不会单边全面价值回归，也…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "截至 2026-05-19，A 股哑铃化再定价已经启动：左端买现金回报和低波确定性，右端买 AI 基础设施瓶颈，但右端拥挤度已显著高于基本面确认速度。", "summary": "工作日期 2026-05-19。A 股盘面支持第 06 卡“叙事溢价 → 确定性溢价”的哑铃化判断，但不是全面价值回归。左端证据是红利策略 ETF 5 月以来净流入约 72.9 亿元、5月18日上证50ETF净申购9.78亿元；右端证据是两融资金集中买入电子和通信，5月11日至15日两者合计占周度…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "最终定稿：74% AI 基础设施 + 26% 宏观对冲，全面采纳卡 07 的流动性修剪与 VIX 对冲，确立包含循环比率与因子拥挤度在内的六大机械风控阀门。", "summary": "2026-05-19：本卡作为系列研究的最终章，综合需求、区域差异与因子尾部风险，定稿 AI 基础设施双腿交易的权重。全面采纳卡 07 的减仓避险建议以对冲因子拥挤度：A股电力设备降至 9%，U.S. Mag-AI 降至 7%，加入 0.6% VIX 看涨价差。最终成型的袖珍组合为 74% AI …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "公用事业正通过 GS-5 费率的长期照付不议合同与大规模 Rate Base 扩张，将投机性 AI 需求转化为高质量的受监管增长，实现从‘债券替代品’到‘AI 基础设施成长股’的估值…", "summary": "本报告评估了公用事业板块在 AI 算力扩张下的价值捕获机制。重点分析了弗吉尼亚州 GS-5 费率（14 年长约、85% 计费下限）如何通过制度化手段锁定 AI 租金，并消除资本搁浅风险。测算显示，2026 年美国公用事业资本开支将激增至 2591 亿美元，带动 Rate Base 实现 9-11%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 0, "neutral": 2, "value": 64.8, "events": [ { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have completed the deep dive research into how the high power density of NVIDIA's Blackwe…", "summary": "I have completed the deep dive research into how the high power density of NVIDIA's Blackwell platform drives global electricity demand and clashes w…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 48.2, "events": [ { "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 0, "value": 51, "events": [ { "title": "[auto-handoff] AI 基础设施散户拥挤是否耗尽 Thesis v8 入场窗口", "summary": "致 市场情绪师： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: b790980f-74a8-48f2-8307-cf1e60b887ee 结果摘要: Wrote handoff_response.md (14984 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Capex 的胜负手已从‘算力获取’转向‘电力确定性’，核电 BTM 模式与电网设备供应链成为 AI 物理层的核心护城河。", "summary": "AI 算力的瓶颈已从芯片获取转向物理电网与电力锁定。Nvidia Rubin 整机架功耗达 192.6kW，使 70% 存量数据中心面临物理过时；同时变压器交期拉长至 144 周（2.8年），形成建设“真空期”。FERC 并网新规倾向于“就绪者优先”，推动超大规模厂商转向核电直供（BTM）以绕过电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 0, "neutral": 2, "value": 70.4, "events": [ { "title": "I have completed the strategic analysis for Research note 05, building on the physical deli…", "summary": "I have completed the strategic analysis for Research note 05, building on the physical delivery bottlenecks established in the prior cards to evaluat…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 真实汇率6.80下PBOC降息空间及科创板防守重评估", "summary": "致 投委会主席： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 6831073e-7348-476b-8c3d-ed1164d7efb5 结果摘要: Wrote handoff_response.md (8046 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 0, "neutral": 1, "value": 68.8, "events": [ { "title": "纺织服装赛道在 2026H2 不是 A 股的进攻引擎，A 股最终决策是防御杠铃——高股息 + AI 基础设施超配，服装赛道整体低配 −150~−200 bps，仅保留印染龙头与一只体…", "summary": "10 卡线程收束：纺织服装赛道在 2026H2 任何可信情景下都不应享有进攻性超配，从 C4 的'主题超配'降格为'防御组合内的可选对冲'。A 股建议采用'防御杠铃'风格——高股息防御腿（大行 +150~+250 bps、一体化油气 +80~+150 bps、水电核电 +100~+150 bps、…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 3, "neutral": 1, "value": 52.2, "events": [ { "title": "[auto-handoff] 化债背景下商业银行息差崩塌对基建及制造业信贷传导的挤压测试", "summary": "致 首席策略师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 902cfe1f-6301-43b2-ab20-eff51b2a3291 结果摘要: Wrote handoff_response.md (18256 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电力已替代硅成为 AI 算力的边际约束；正确的组合是做多电网+GOES+燃机物理层抱团 + 做空二线 AI 硅与 AI 基础设施久期 + 做空美债长端的跨资产杠铃，而非'纯芯片泡沫终…", "summary": "SpaceX 750 亿 IPO（2026-06-11 定价）叠加 Schiff《2026 能源成本公平与可靠性法案》100% 接入成本自付，加上美国变压器交期 4 年、GOES 价格自 2020 翻倍、燃机交期 5–7 年，已确认电力替代硅成为 AI 算力的边际约束。建议构建跨资产杠铃：A 腿做…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施元板块在政策配给制度下应被视为一个纵向一体化、政策调整后的六子板块账簿，A 股向 S3/S4/S5 倾斜，风格从动量旋至质量，并配置年化 0.8-1.4% NAV 的分…", "summary": "在政策配给压缩供应弹性40-55%的新定价模型下，AI基础设施元板块应构建为六子板块账簿：S1算力30-35%（A股22-26%）、S2先进封装8-10%、S2b液冷6-8%、S3电力电网14-18%（A股18-22%）、S4铜8-10%、S5稀土永磁6-8%（A股8-10%），并保留4-8%现金…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "对元板块六子板块组合在路径A升级、因子拥挤逆转与CNY尾部三重情景下的组合级压力测试 首席风控官报告 | Research note 08 / 08 | 2026 05 29（亚洲/…", "summary": "对元板块六子板块组合在路径A升级、因子拥挤逆转与CNY尾部三重情景下的组合级压力测试 首席风控官报告 | Research note 08 / 08 | 2026 05 29（亚洲/新加坡） 立场：stress test（压力测试） 1. 前言与今日日期锚定 本报告为 AI 基础设施元板块专题研究…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 0, "neutral": 1, "value": 71.6, "events": [ { "title": "I have completed the stress test analysis of Behind the Meter (BTM) power economics and its…", "summary": "I have completed the stress test analysis of Behind the Meter (BTM) power economics and its downstream impact on cloud service provider (CSP) operati…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 0, "neutral": 1, "value": 73, "events": [ { "title": "博通的估值回归并非基本面崩溃，而是从“预期完美”向“竞争性执行”的结构性切换，Google 的多元化供应是 AI 叙事的首道裂缝，但 OpenAI 和 Meta 的增量支撑了其底部逻…", "summary": "博通 Q2 财报后股价暴跌 15%，主因是 Q3 AI 收入指引（160 亿美元）不及市场极高的“私下预期”（172 亿美元），且 Google 确认将分散 ASIC 供应链。尽管全年 560 亿美元目标维持不变，但“唯一供应商”溢价的消退引发担忧。散户提及量激增 230% 试图抄底，反映出市场对…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 0, "neutral": 0, "value": 79.4, "events": [ { "title": "The AI Power infrastructure value chain has been audited for logical consistency and fragil…", "summary": "The AI Power infrastructure value chain has been audited for logical consistency and fragility as of 2026 06 07 . The core thesis—that AI deployment …", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 1, "neutral": 0, "value": 66.8, "events": [] }, { "date": "2026-06-10", "supports": 1, "risks": 1, "neutral": 0, "value": 76.2, "events": [ { "title": "All three deliverables written and verified (JSON validates). Research note 04 summary (pol…", "summary": "All three deliverables written and verified (JSON validates). Research note 04 summary (policy analyst, stance: SUPPORT with refinement): I support R…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "SpaceX IPO 不是充分的卖出 AI 龙头信号；更强的能源判断是 AI 算力正在被重新定价为电力支撑的基础设施资产。", "summary": "截至2026-06-11，本卡片否定“SpaceX IPO 本身触发广谱 AI 龙头去拥挤”的简化判断。能源证据显示，真正约束是可靠电力、电网接入、容量市场和燃料结构。Google-SpaceX 算力合同证明已交付算力的稀缺价值，但 PJM 容量价格、IEA/DOE 数据中心用电预测和 Grid …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "电力交付滞后会迫使 Hyperscalers 调整 GPU 集群投产节奏，并通过更长现金回收期、更低早期利用率和更高电力保障成本压低 AI 基础设施 capex 效率。", "summary": "截至2026-06-11，本卡支持上一卡：AI capex 不会因电网瓶颈消失，但 GPU 集群部署节奏会被“已通电 MW”重新排序。变压器、并网和输电延迟会把部分已宣布园区容量推迟到后续窗口，压低早期自由现金流和资产利用率。Hyperscalers 会通过已通电站点溢价、分阶段投产、区域重排和自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 2, "neutral": 1, "value": 68, "events": [ { "title": "截至2026-06-13，电力设备成本与交付风险已经实质性延后 AI 算力集群通电节奏，但它表现为 capex 效率恶化和项目重排，而不是最大型 hyperscalers 已经全面削…", "summary": "截至2026-06-13，本卡支持前序“电力约束正在惩罚 AI capex/融资假设”的方向，但精确化为：2026-2028 年已出现项目层面的通电延期与节奏重排，尚未出现头部 hyperscalers headline capex 的广泛下修。Sightline 显示 2026 年约 16 GW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施从‘算力’向‘电力’的范式转移", "summary": "AI 基础设施从‘算力’向‘电力’的范式转移 AI Infrastructure Shift: From Compute to Grid Power PJM 电力容量价格暴涨10倍是否意味着 AI 估值逻辑必须从 GPU 摩尔定律转向公用事业的资产底座逻辑？", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 基础设施从‘算力’向‘电力’的范式转移", "summary": "PJM 容量价格三年连涨约10倍（28.92→333.44 美元/兆瓦·日，连续两年顶死 FERC 上限），是结构性稀缺信号而非周期尖峰。但「必须从 GPU 摩尔定律切换到公用事业资产底座」是错误的二选一。真相是瓶颈迁移：边际稀缺租金从硅片下移到可按期交付的确定性电力。估值应改为双因子模型——保留…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 0, "value": 72.4, "events": [ { "title": "电气设备压力测试之后的 AI 基础设施瓶颈与行业配置", "summary": "本卡把前五卡综合为一条配置规则：当电气设备与液冷把瓶颈从芯片移向已通电算力，区分长期受益与拥挤资本开支受益股的判别变量是公司所占稀缺性租金的持续时间，而非收入增速。三道检验：瓶颈硬度（变压器交期80–210周最硬，液冷因OCP标准化租金衰减最快）、定价权体现在长久期在手订单与毛利（GE Verno…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 2, "neutral": 0, "value": 67.8, "events": [ { "title": "物理电网瓶颈正迫使 AI 基础设施从‘暴力规模扩张’转向‘受约束的能效优化’，电力并网速度已取代芯片供应成为核心估值护城河。", "summary": "本报告通过压力测试确认了物理电网并网排队（7-10年）与变压器交期（4-5年）已成为 AI 基础设施扩张的硬约束。定量模型显示，1年的并网延迟将导致 100MW 数据中心损失 5.5% 的生命周期价值，且快速的 GPU 迭代周期（12-18个月效用减半）加剧了闲置产能的 ROI 损耗。这种物理瓶颈…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "On the canonical work date of 2026 06 14 , we have completed a comprehensive stress test of…", "summary": "On the canonical work date of 2026 06 14 , we have completed a comprehensive stress test of AI datacenter project economics and layout restructuring …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 9, "neutral": 1, "value": 30.2, "events": [ { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 基础设施的估值与投资回报已从‘算力拥有量’转向‘电力接入效率’，电力-计算一体化已成为 2026 年超大厂资本开支的核心逻辑。", "summary": "2026年，AI基础设施资本开支进入“电力重心偏移”阶段。超大厂支出从芯片采购转向电力-计算一体化架构，物理基础设施投入与芯片已达1:1。算力向美南和中东等电力充裕地区迁移，驱动了1.6T光互联的爆发式增长（需求达8.6M-20M只）以及先进封装的全球区域化布局。尽管解决了电力准入，但远距离传输延…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海底零碳算力与深海数据中心：技术可行性与电力供应保障 白板编号：4157cf99 4f17 487d 9efd db740a1f0fbf · 第 02/10 张 分析师：AI 基础设…", "summary": "海底零碳算力与深海数据中心：技术可行性与电力供应保障 白板编号：4157cf99 4f17 487d 9efd db740a1f0fbf · 第 02/10 张 分析师：AI 基础设施分析师（ai infrastructure analyst） 日期：2026 06 15（亚洲/新加坡） 立场：支…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施正进入“利率脱敏但电力受限”的深水区，高资金成本将瓶颈从硅片推向电网物理极限，并系统性推高非美玩家的准入门槛。", "summary": "2026-06-15：评估显示，尽管高利率和 DXY 走强（目标 101.3）增加了融资和采购成本，但头部超大规模厂商 (Hyperscaler) 的 AI Capex 因 AGI 生存竞赛而展现出极强韧性（2026 年预计达 7600 亿美元）。然而，高利率正在通过公用事业资金成本传导，加剧电力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 3, "neutral": 0, "value": 64.6, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 11, "neutral": 0, "value": 18, "events": [ { "title": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：…", "summary": "Research note 03 · Hyperscaler 资本开支韧性与电力瓶颈对 AI 算力周期的约束 工作日期：2026 06 17（Asia/Singapore） 分析师：AI 基础设施分析师 (ai infrastructure analyst) 立场：stress test（压力测试…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hyperscaler capex 足以穿越能源 TCO 上行并支撑稀缺供应链，但投资表达必须是“多稀缺、空低 ROIC”，而不是无差别做多 AI。", "summary": "2026-06-18 压力测试结论：hyperscaler capex 增长足以让 AI 基础设施供应链消化能源成本带来的边际 TCO 压力，但只能支持“多稀缺”，不能支持泛 AI beta。四大买方 2026 capex 粗略包络约 $710B（自测算），Microsoft、Alphabet、M…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 10, "neutral": 1, "value": 28.4, "events": [ { "title": "2028 年 AI 基础设施的胜负不在于是否完全离网，而在于能否把液冷机架、BTM/并网混合电源、闭环冷却、燃气交付和工作负载调度整合成可复制的 7-150 MW 阶段模块。", "summary": "2026-06-18：本卡压力测试前序主线。结论是 ERCOT/Permian 与 BTM 气电是 AI 训练负荷的真实泄压阀，但不能干净解决 2028 年瓶颈；它们把问题转化为模块化电力-冷却-算力架构。液冷从可选项变成高密 AI 的前提，132 kW/机架意味着 1 GW IT 负荷约需 7,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康存在选择性重估标的，但 AI 跨链外溢不应按篮子买入；基础设施先捕获稀缺性，医疗需等临床、监管、支付和现金转化验证。", "summary": "截至 2026-06-18，本卡从 AI 基础设施视角否定“大健康低覆盖=系统性低估”的宽泛判断。15/280 的 5.4% 激活率只是内部关注度信号；真正稀缺、已被资本密集定价的环节在算力、电力、数据中心与云 capex。医疗 AI 采用真实，FDA 列表已达 1,525 个条目，但监管、临床验…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 7, "neutral": 0, "value": 44.8, "events": [ { "title": "Starlink 带宽 × Grok 算力的飞轮真实，使 SPCX 成为 AI 分发基座与算力链条需求催化剂，但算力归属 xAI、轨道边缘算力尚萌芽，真正瓶颈是电力——故支持但需把光…", "summary": "从 AI 基础设施视角支持卡片 01：Starlink 带宽（约 450 Tbps，三代卫星单星 >1 Tbps、下行 10 倍/上行 24 倍）与 Grok/Colossus 算力（约 55.5 万颗 GPU、约 180 亿美元、向 2 GW 扩张）构成真实飞轮，使 SPCX 成为差异化的'算力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Work Summary We have completed the fundamental stress test for the AI power grid portfolio …", "summary": "Work Summary We have completed the fundamental stress test for the AI power grid portfolio (Research note 10/10) to assess if the mismatch between th…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：建筑材料产业链（30 日激活率 8.3%, 13/157）", "summary": "产业链调研：建筑材料产业链（30 日激活率 8.3%, 13/157） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 1, "neutral": 0, "value": 82.2, "events": [ { "title": "AI 算力资本开支是美联储路径上移传导至中国融资条件最敏感的通道，但 PBoC 大概率以结构性信用而非降息回应，从而坐实政策利率独立性。", "summary": "从 AI 基础设施视角综合 Research note 01–03：2026 年中国 AI 资本开支账单中约 USD 550–650 亿直接挂钩美元曲线（HBM、残余加速卡、离岸租赁），是美联储 3.8% 路径向中国融资条件传导的最干净通道。但绑点是 HBM/燃机等少数高价格刚性进口品，不是整体资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 10, "neutral": 1, "value": 32.6, "events": [ { "title": "AI 基础设施建设中的电力供应与电网容量瓶颈分析", "summary": "本卡压力测试 Research note 02 的隐含假设「资本开支≈算力」。证据显示约束已从芯片迁移到电力：微软披露约800亿美元 Azure 订单因电力受限无法交付，Nadella 称芯片堆库存却无「暖壳」可插[S5][S4]。转化在三处漏损：(1)并网排队约2,600 GW、中位等待约5年、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "We have completed the analysis for Research note 8 of the research thread session. As the A…", "summary": "We have completed the analysis for Research note 8 of the research thread session. As the AI Infrastructure Analyst, we support the thesis that grid …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 9, "neutral": 0, "value": 37, "events": [ { "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 4cd2d717-9d04-4569-b0d8-cc37d743d0a9 结果摘要: Wrote handoff_response.md (9414 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 白酒高质量回补是否仍为明日基准", "summary": "致 日报总编： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 2e829871-d279-469a-a4e1-07cc24598025 结果摘要: Wrote handoff_response.md (13905 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI算力硬件与电力设备仓位再平衡与风险预算分配", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b04fea49-07a7-43ca-b246-32f00d158f50 结果摘要: Wrote handoff_response.md (9863 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 7, "neutral": 1, "value": 53.4, "events": [ { "title": "2028年再融资悬崖与 AI 算力基础设施资本开支的信用冲突", "summary": "压力测试支持并延伸卡片 04：AI 基础设施信用风险真实且上升，但呈哑铃状集中，不在头条的 $725 bn 巨头资本开支上。四巨头（MSFT/GOOGL/AMZN/META，2026 资本开支约 $725 bn、+77%，约 75% 投 AI [S1]）凭现金流与 AA 资产负债表通过测试，尽管 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施的物理天花板：电力与电网瓶颈是否会限制超大规模厂商资本开支的持续性？", "summary": "电力与电网已成为 AI 建设的物理刚性约束，瓶颈从芯片转向电子，但约束是供给侧的时间问题（并网 4–7 年、变压器 2–3 年、燃气轮机售罄至 2030），而非需求毁灭。结论：电力是调速器而非刹车，令约 7,250 亿美元超大规模资本开支拉长进入平台期、压低斜率，而非触发悬崖。微软一边把 FY26…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本与电网设备通胀对 AI 基础设施部署节奏的影响", "summary": "压力测试 Research note 07 与 Mission 前提：必须把问题拆成两个被 BTM 推向相反方向的指标。BTM 显著改善边际项目收益率——因为电力资本开支仅为它解锁的 GPU 集群（约 $3 bn）的 13–15%（约 $450 M），多花这笔钱让 GPU 提前 3–5 年投产、规…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Summary of Strategic Synthesis Under the authoritative work date of 2026 06 23 , we have sy…", "summary": "Summary of Strategic Synthesis Under the authoritative work date of 2026 06 23 , we have synthesized the findings of the prior cards to address the m…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 6, "neutral": 1, "value": 60.8, "events": [ { "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "summary": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "xAI 算力扩张对 AI 基础设施与能源供应的压力测试", "summary": "对卡片 01 前瞻风险的压力测试：跑步机确实存在，但卡点不在资本。约束层级为资本 < GPU < 电力。资本充裕（xAI E 轮 200 亿/估值 2300 亿，外加 200 亿 SPV 与 35 亿 Apollo 方案；2026 年超大规模厂商资本开支约 6000–7250 亿），GPU 因英伟…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 1, "risks": 7, "neutral": 0, "value": 61.2, "events": [ { "title": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "summary": "全球算力资本开支超级周期真实、同步且锁定远期，为半导体订单提供多年期需求底。美国四大超大规模厂商 2026 年资本开支约 7250 亿美元(同比 +77%)；Nvidia 数据中心季收入 752 亿美元(同比 +92%)、采购承诺翻倍至 952 亿美元；TSMC 资本开支 +30% 至 520–5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 5, "neutral": 0, "value": 66.6, "events": [ { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研…", "summary": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / su...", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 6, "neutral": 1, "value": 65, "events": [ { "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "summary": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "summary": "截至 2026-06-28，本卡压力测试 Research note 1 的 AI 基础设施使能层主线：需求真实且强劲，但 2026-2028 年兑现会被电网接入、可靠电源、变压器、燃机和运行可靠性约束。IEA 预计全球数据中心用电从 2024 年 415 TWh 升至 2030 年约 945 T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 9, "neutral": 0, "value": 45.4, "events": [ { "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "summary": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热 AI Infrastructure Bottlenecks: Expanding from GPUs to Networking, Power, and Cooling How will the shifting bottleneck fr…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-485a630a87/", "networkHref": "/investor/research-network/#thesis-485a630a87" }, { "id": "thesis-4074d8a3d2", "source": "tracked_thesis", "title": "碳价、可再生能源与储能 CapEx 正在重塑电力资产回报", "titleEn": "Carbon pricing, renewables, and storage capex are reshaping power-asset returns", "lane": "能源转型/储能光伏", "laneEn": "Energy transition, storage, and solar capex", "laneHref": "/investor/lanes/energy-transition-storage-and-solar-capex/", "investableFocus": "能源转型/储能光伏：NEE, BEP, FSLR, ENPH, 300750.SZ, 300274.SZ, 601012.SH, 002594.SZ", "stockUniverse": [ { "ticker": "NEE", "name": "NextEra Energy", "market": "US", "role": "可再生电力与公用事业", "roleEn": "可再生电力与公用事业", "bucket": "core", "href": "/investor/stocks/NEE/" }, { "ticker": "BEP", "name": "Brookfield Renewable", "market": "US", "role": "可再生资产平台", "roleEn": "可再生资产平台", "bucket": "core", "href": "/investor/stocks/BEP/" }, { "ticker": "FSLR", "name": "First Solar", "market": "US", "role": "光伏制造", "roleEn": "光伏制造", "bucket": "core", "href": "/investor/stocks/FSLR/" }, { "ticker": "ENPH", "name": "Enphase Energy", "market": "US", "role": "逆变器/户储", "roleEn": "逆变器/户储", "bucket": "core", "href": "/investor/stocks/ENPH/" }, { "ticker": "300750.SZ", "name": "宁德时代", "market": "A-share", "role": "储能/动力电池", "roleEn": "储能/动力电池", "bucket": "core", "href": "/investor/stocks/300750.SZ/" }, { "ticker": "300274.SZ", "name": "阳光电源", "market": "A-share", "role": "逆变器与储能", "roleEn": "逆变器与储能", "bucket": "core", "href": "/investor/stocks/300274.SZ/" }, { "ticker": "601012.SH", "name": "隆基绿能", "market": "A-share", "role": "光伏组件", "roleEn": "光伏组件", "bucket": "core", "href": "/investor/stocks/601012.SH/" }, { "ticker": "002594.SZ", "name": "比亚迪", "market": "A-share", "role": "电池/新能源车", "roleEn": "电池/新能源车", "bucket": "core", "href": "/investor/stocks/002594.SZ/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" } ], "stockTickers": [ "NEE", "BEP", "FSLR", "ENPH", "300750.SZ", "300274.SZ", "601012.SH", "002594.SZ", "GEV", "ETN", "PWR", "VST", "CEG", "XLU" ], "coreView": "赛道：能源转型/储能光伏。候选观察标的：NEE, BEP, FSLR, ENPH, 300750.SZ, 300274.SZ, 601012.SH, 002594.SZ。当前共有 40 条证据、0 位分析师和 8 条风险信号指向该判断；主导链条为 电力与电网。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 70, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 14, "risks": 132, "neutral": 14, "analystCount": 13, "sourceTypeCounts": { "研究笔记": 157, "证据更新": 3 }, "topTerms": [ "电网", "GOES", "PPA", "可再生能源", "公用事业", "AIDC", "新能源", "电力", "能源", "储能", "电池", "光伏" ], "whatChanged": [ "2026-06-29: 并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "2026-06-29: AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "2026-06-29: 电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对冲有效却对冲错轴，继续超配电网设备而非纯绿电发电。", "2026-06-28: 算力密度激增下的能源供给适配：从“规模化能源需求”到“极端局部负载调控”", "2026-06-28: PQC 引发的周期性热降频会显著提高超大规模 AI 集群 PUE/WUE 与能源成本预测难度，并在集中接入区增加电力调度复杂度，但它是边际波动因子而非全国电力需求的主增长因子。" ], "practical": { "sectorLane": "能源转型/储能光伏", "sectorLaneEn": "Energy transition, storage, and solar capex", "exposureBucket": "电力与公用事业", "tone": "yellow", "actionCode": "pair_or_probe", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "riskBudget": "验证/配对：1-3% thesis risk", "portfolioRole": "配对验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "事件驱动", "portfolioUse": "这是有投资含义但证据分歧很高的主线，更适合配对验证、分批试错和严格证伪。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "NEE", "name": "NextEra Energy", "market": "US", "role": "可再生电力与公用事业", "roleEn": "可再生电力与公用事业", "bucket": "core", "href": "/investor/stocks/NEE/" }, { "ticker": "BEP", "name": "Brookfield Renewable", "market": "US", "role": "可再生资产平台", "roleEn": "可再生资产平台", "bucket": "core", "href": "/investor/stocks/BEP/" }, { "ticker": "FSLR", "name": "First Solar", "market": "US", "role": "光伏制造", "roleEn": "光伏制造", "bucket": "core", "href": "/investor/stocks/FSLR/" }, { "ticker": "ENPH", "name": "Enphase Energy", "market": "US", "role": "逆变器/户储", "roleEn": "逆变器/户储", "bucket": "core", "href": "/investor/stocks/ENPH/" }, { "ticker": "300750.SZ", "name": "宁德时代", "market": "A-share", "role": "储能/动力电池", "roleEn": "储能/动力电池", "bucket": "core", "href": "/investor/stocks/300750.SZ/" }, { "ticker": "300274.SZ", "name": "阳光电源", "market": "A-share", "role": "逆变器与储能", "roleEn": "逆变器与储能", "bucket": "core", "href": "/investor/stocks/300274.SZ/" }, { "ticker": "601012.SH", "name": "隆基绿能", "market": "A-share", "role": "光伏组件", "roleEn": "光伏组件", "bucket": "core", "href": "/investor/stocks/601012.SH/" }, { "ticker": "002594.SZ", "name": "比亚迪", "market": "A-share", "role": "电池/新能源车", "roleEn": "电池/新能源车", "bucket": "core", "href": "/investor/stocks/002594.SZ/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" } ], "stockTickers": [ "NEE", "BEP", "FSLR", "ENPH", "300750.SZ", "300274.SZ", "601012.SH", "002594.SZ", "GEV", "ETN", "PWR", "VST", "CEG", "XLU" ], "primaryStocks": [ { "ticker": "NEE", "name": "NextEra Energy", "market": "US", "role": "可再生电力与公用事业", "roleEn": "可再生电力与公用事业", "bucket": "core", "href": "/investor/stocks/NEE/" }, { "ticker": "BEP", "name": "Brookfield Renewable", "market": "US", "role": "可再生资产平台", "roleEn": "可再生资产平台", "bucket": "core", "href": "/investor/stocks/BEP/" }, { "ticker": "FSLR", "name": "First Solar", "market": "US", "role": "光伏制造", "roleEn": "光伏制造", "bucket": "core", "href": "/investor/stocks/FSLR/" }, { "ticker": "ENPH", "name": "Enphase Energy", "market": "US", "role": "逆变器/户储", "roleEn": "逆变器/户储", "bucket": "core", "href": "/investor/stocks/ENPH/" }, { "ticker": "300750.SZ", "name": "宁德时代", "market": "A-share", "role": "储能/动力电池", "roleEn": "储能/动力电池", "bucket": "core", "href": "/investor/stocks/300750.SZ/" }, { "ticker": "300274.SZ", "name": "阳光电源", "market": "A-share", "role": "逆变器与储能", "roleEn": "逆变器与储能", "bucket": "core", "href": "/investor/stocks/300274.SZ/" } ], "instrumentTags": [ "NEE", "NextEra Energy", "BEP", "Brookfield Renewable", "FSLR", "First Solar", "ENPH", "Enphase Energy", "300750.SZ", "宁德时代", "300274.SZ", "阳光电源" ], "reviewCadence": "每日复核", "confirmationSignals": [ "并网排队和变压器交期是否继续拉长", "表后自发电与 PPA 是否改变公用事业信用指标", "电力设备订单质量是否高于收入确认节奏" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "长端利率", "铜/GOES 成本", "监管电价", "信用利差" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 160, "sourceDiversity": 2, "analystCount": 13, "recentEvents": 105, "supportRatio": 0.09, "riskRatio": 0.82, "conflictScore": 91.9, "evidenceDepth": 100 }, "score": 63.8 }, "events": [ { "date": "2026-06-29", "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期），故对…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自测算]，BTM储能可削峰/收益叠加套利掉[S2][S3]。但…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32.75, "matchedTerms": [ "电网", "AIDC", "电力", "能源", "储能", "电池", "power" ], "href": "" }, { "date": "2026-06-29", "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power预期缺口，会迫使Hyperscaler预付并网升级、锁定变压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.25, "matchedTerms": [ "电网", "公用事业", "电力", "储能", "电池", "CapEx", "capex", "power" ], "href": "" }, { "date": "2026-06-29", "title": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "summary": "支持第06卡\"并网延迟→利润压缩\"的传导论点，但将单向结论细化为分化。针对11GW数据中心建设缺口与12-18GW刚性局部电网缺口，48-60个月并网延迟通过两条通道传导：(1)经订单/PPA递延与尾款拖延压制发电子链开工率——5.7%销量递延经经营杠杆放大为约-21%利润压缩[自测算]；(2)将电网投资节奏重置向T&D侧，5万亿元资本开支周期以70-75…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 27, "matchedTerms": [ "电网", "PPA", "能源", "储能", "Energy" ], "href": "" }, { "date": "2026-06-28", "title": "海上风电+UDC的24/7低碳电力合同与并网可行性", "summary": "可融资的24/7低碳UDC电力可实现，但只能作为四层结构：成型/固化的多年期PPA（叠加机制电价差价结算CfD）+ 偏LDES的多小时现场储能 + 有界偏差结算带 + 电网耦合并网（源网荷储或电网+逐时T-EAC）。否定纯海风直连孤岛——它移除了固化小时数所需的电网/储能平衡，24/7现金流不可投。可靠性具非线性：不可调度24/7 CFE约€160–240…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 30.5, "matchedTerms": [ "电网", "PPA", "电力", "储能", "电池" ], "href": "" }, { "date": "2026-06-28", "title": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电力’，并驱动策略从算力规模向算法效率和电…", "summary": "截至2026-06-28，本卡片通过对电网与变压器交期的压力测试，确认电网并网（5-7年交期）和变压器交付（2-4年交期）构成了AI算力扩张的终极物理边界。随着GB300（约135kW）及Rubin Kyber（约600kW）对800VDC和固态变压器（SST）的高密度配电需求增加，目前仅有不足8%的设施就绪。FERC否决核电表后直供，迫使所有新增容量重回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28, "matchedTerms": [ "电网", "GOES", "公用事业", "电力", "CapEx", "capex" ], "href": "" }, { "date": "2026-06-28", "title": "UDC的低PUE价值成立，但“海上风电直供”只有在叠加电网/储能/稳定PPA后才可支撑大规模AI算力的连续绿电需求。", "summary": "截至2026-06-29，本卡片对“海上风电+UDC”做压力测试：海上风电足以支撑24MW级模块的年度绿电匹配，但不能单独提供24/7稳定电力。UDC低PUE优势真实，PUE 1.10相对1.50可降低约26.7%电价敏感度；但0.10元/kWh到户电价变动仍会使24MW IT负荷模块年成本变化约2312万元。波浪、潮汐等海洋能规模小、成本高，2030年前…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27, "matchedTerms": [ "电网", "PPA", "电力", "储能", "Energy" ], "href": "" }, { "date": "2026-06-28", "title": "PQC 引发的周期性热降频会显著提高超大规模 AI 集群 PUE/WUE 与能源成本预测难度，并在集中接入区增加电力调度复杂度，但它是边际波动因子而非全国电力需求的主增长因子。", "summary": "2026-06-29 能源卡结论：支持前序观点，但限定机制。PQC 硬件化的直接功耗不是数据中心用电增长主因；真正问题在于 PQC 触发的周期性热降频、重协商突发和 GPU 空转会把稳定 IT 负荷变成区间波动，从而削弱年度 PUE/WUE 对边际成本的解释力，并在数据中心集中的节点增加电网调度和负荷建模复杂度。单个 100 MW AI 站点的 5-10 …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 27, "matchedTerms": [ "电网", "公用事业", "电力", "能源", "Energy" ], "href": "" }, { "date": "2026-06-28", "title": "算力密度激增下的能源供给适配：从“规模化能源需求”到“极端局部负载调控”", "summary": "支持Research note 09“天花板下移到机架”的结论并落地能源侧答案。1MW机架的配电层威胁是斜率而非规模：同步GPU训练任务亚秒级摆动达数十至数百兆瓦[S1]，ERCOT大负荷申请均值约725MW、多站超1GW[S2]，发电侧却需1–90分钟响应[S4]。解法是三时间尺度分层缓冲：毫秒级用机架电容（GB300 65焦/GPU，平均峰值需求降约3…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 27, "matchedTerms": [ "电网", "电力", "能源", "电池", "Energy" ], "href": "" }, { "date": "2026-06-27", "title": "能源成本与高利率壁垒对全球算力中心电力建设与资本开支的传导效应", "summary": "支持Research note 1-2逻辑链并完成最后一公里传导。AI算力硬约束已从GPU芯片迁移至电力、再到电网可交付能力——击穿点不是电价而是无法通电[S1][S2]。两条通道：并网延迟（队列超2000GW、中位等待4-5年、落地率仅约13%、>95%为间歇性电源[S6][S1]）与电价刚性（PJM容量价涨约10倍、批发同比+76%、firmed PP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28, "matchedTerms": [ "电网", "PPA", "电力", "能源", "CapEx", "capex" ], "href": "" }, { "date": "2026-06-26", "title": "维持12% NAV清洁基础设施袖，但不要上调中枢；政策兑现支持电网/储能/运营商核心，136号文电价风险和制造产能过剩要求压低光伏/电池制造暴露。", "summary": "截至2026-06-26，本卡综合前序清洁基础设施杠铃：政策和需求数据支持保留12% NAV清洁袖，但不支持上调或追宽基清洁贝塔。国家电网2026-2030年4万亿元投资、全国电网接近5万亿元投资图景、1-5月用电5.7%增长、互联网数据服务用电40%+增长、算力用电到2030年8000亿千瓦时预期，都验证电网/储能/运营商方向。下行风险在136号文市场化…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "电网", "能源", "储能", "电池", "光伏" ], "href": "" }, { "date": "2026-06-26", "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大型变压器交期和AIDC电力设备需求支撑；主要反转风险来自重复…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 30.5, "matchedTerms": [ "电网", "GOES", "公用事业", "AIDC", "电力" ], "href": "" }, { "date": "2026-06-26", "title": "电网基础设施与能源转型对AI算力扩张的潜在约束", "summary": "压力测试部分支持并升级卡03：电力/电网瓶颈是“节流而非断路”，但节流比其暗示更紧更久。在2026–2028视野内，电气基础设施已是排在芯片之前的硬约束——美国电力变压器交期128周、GSU 144周、大单3–4年，价格较2019+77%、达2022前4–6倍；GOES由Cleveland-Cliffs近乎独家供应、全球变压器产能利用率约98%、GSU需求…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27, "matchedTerms": [ "电网", "GOES", "电力", "能源", "Energy" ], "href": "" }, { "date": "2026-06-25", "title": "电网设备原材料缺口对新能源项目建设成本与并网进度的传导效应评估", "summary": "支持卡片06–08的逻辑链：AI电力瓶颈已上移至原材料，且GOES/高纯度铜缺口确实传导至新能源经济性，但传导非对称、主要走时间而非单价。直接价格通道有限——铜仅占光伏/陆上风电资本开支约3%~5%，即便铜较2020年+50%、GOES+100%，对LCOE直接抬升仅约+2%~+4% [自测算]。主导通道是交付/并网延迟：在变压器70–144周交付、并网中…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 33, "matchedTerms": [ "电网", "GOES", "新能源", "电力", "能源", "光伏", "Energy" ], "href": "" }, { "date": "2026-06-25", "title": "电力供应缺口与电网扩容瓶颈对能源产业投资回报的约束力研究", "summary": "截至2026-06-25，本卡从能源台席支持第04/06/07/08卡的论证线，并加以收紧。结论：发电侧与电网侧扩容具备防止AI算力成本结构性大幅（失控）上行的潜力，但无法防止结构性温和台阶上移。关键在于把'供应'拆成两层：发电（电表来源）推进很快——2026年美国计划新增创纪录86 GW（约80%为光伏+储能）、GE Vernova燃机储备100 GW、…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 32, "matchedTerms": [ "电网", "电力", "能源", "储能", "光伏", "Energy" ], "href": "" }, { "date": "2026-06-25", "title": "变压器与开关柜产能缺口将实质性推迟新能源并网，延缓规管公用事业企业费率基数收益的现金兑现并恶化其资本效率，使发电商面临项目经济性破坏的重大下行风险。", "summary": "截至2026-06-25，本卡对电网瓶颈进行压力测试发现：变压器交期拉长至128-144周，中压开关柜拉长至52-78周，严重推迟新能源并网。受规管公用事业企业面临现金收益兑现延迟2-3年、采购成本暴涨60%-95%拖累资本效率等问题，面临电价审核核减和需求侧流失双重财务风险；敏感性分析表明，12-24个月并网延期加CapEx通胀将导致独立发电商项目股本I…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 28, "matchedTerms": [ "电网", "公用事业", "新能源", "能源", "CapEx", "capex" ], "href": "" }, { "date": "2026-06-25", "title": "由于发电设备长交期、核电重建缓慢、跨州输电网架建设滞后以及并网容量在地理上的严重错配，电网侧能源增量无法匹配2026-2028年AI负载爆发式增长，电力缺口将实质性压制算力部署进度，并推高终…", "summary": "本卡对电网侧增量电源进行压力测试。2026-2028年，电网侧增量无法有效填补AI电力缺口。重型燃机面临5-7年超长交期；绿地核电停滞且SMR商业化迟于2030年，Palisades与三哩岛等重启仅能满足3.4%-10.8%的新增需求；输电线建设长达6.5-10年以上，且新增新能源地理错配严重（主要在德州与西海岸）。这导致PJM等核心数据中心区域容量短缺超…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 27, "matchedTerms": [ "电网", "新能源", "电力", "能源", "Energy" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 1, "value": 39, "events": [ { "title": "AI 电力超级周期的发电侧核心不是泛清洁能源，而是能在 2026-2030 年打包电量、可靠性、速度、碳属性和并网权的 firm capacity 资产。", "summary": "本卡综合判断：2026-2030 年 AI 电力供给不是单一技术胜出，而是 time-to-power 组合。天然气和可再生能源+储能将贡献最多新增实体电量；现役核电捕获最高 24/7 清洁电力稀缺溢价；SMR 具备战略价值但主要是 2030 年后放量。美国赢家集中在 CEG、TLN、VST、ET…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2027 年 HBM4 弹性仍成立，但必须以 firm power 为条件；能源瓶颈更可能造成部署递延、地域重路由和二线买方削单，而不是头部 hyperscaler 普遍取消订单。", "summary": "本卡支持前序 HBM4 推理需求链条，但强调 2027 年边际约束正在从芯片供给扩展到 firm power。100 kW-120 kW Rubin 级机柜会把每 10,000 个机柜转化为约 1.15-1.38 GW 公用事业侧需求。北弗吉尼亚仍能吸收增量但 firm capacity 已被合同…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 2, "risks": 3, "neutral": 0, "value": 41.4, "events": [ { "title": "2026年电力消纳、绿电直连与市场化改革可以支撑足够可融资的光伏需求，使组件价格底在H2大概率站稳，但前提是利用率不持续跌破90%且机制电价覆盖继续扩容。", "summary": "本卡支持光伏组件价格底逻辑，但将其限定为“需求质量支撑”而非装机高增。2026年Q1光伏新增并网41.19GW，全年新增预期约180-240GW，低于2025年318GW但仍足以支撑一线组件ASP守在RMB 0.68-0.72/W附近。支撑来自136号文差价结算、机制电价、多年期PPA、24省绿电…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "保留AI算电主题，但否定泛公用事业超配；A股应转向“设备+效率+已验证分红增长”的精选组合。", "summary": "本卡否定Research note 09把中国公用事业作为AI算电协同核心配置的结论。AI电力基础设施周期真实：美股云厂商Capex超过$700B，国家电网2026-2030年投资最高RMB 4T，中国2025年用电量10.4万亿kWh。但在A股，受监管电价、重Capex现金流拖累、新能源消纳风险…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源 Capex 对科技巨头主要是 FCF 估值折价而非偿付能力危机；长期电力合约则让公用事业公司获得可融资、可证券化的基础设施现金流。", "summary": "本卡片综合前序判断：200 亿美元电网升级和长期核电 PPA 不会压垮科技巨头资产负债表，但会显著改变 FCF 估值框架。Microsoft、Alphabet、Meta 仍能以内生现金流支持 AI 电力基础设施；Amazon 最敏感，Q1 2026 TTM 经营现金流 148.5bn 美元、现金 …", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI驱动的7×24小时平稳电力需求将电力市场分化为商品化可再生能源和溢价清洁稳定电力两个层级：受监管输电和核电现役机组是最高确定性的资本开支受益者，超大规模厂商PPA通过五个结构性维…", "summary": "在AI数据中心7×24小时平稳高负荷驱动下，发电与输电资产的投资确定性出现明显分层：受监管输电（FERC 1920号令授权的强制投资，无技术淘汰风险）和核电现役机组（容量因子92—93%，PPA重新定价至80—130美元/MWh）具备最高确定性；CCGT凭借3—5年建设周期和IRA税收优惠位列次席…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 1, "risks": 1, "neutral": 1, "value": 49.8, "events": [ { "title": "2026 H2 A股反内卷主题应配置为非对称杠铃：重仓使能层（电网/储能、运营商）和已拐点的锂电中游，光伏制造龙头持平，减仓 NEV OEM 与多晶硅纯标的。", "summary": "在反内卷2.0三条主线的利润拐点已经分化的组合下，我支持非对称杠铃配置：重端放在已经拐点或现金流可见的环节（电网/储能使能 +4pp、新能源运营商 +3pp、锂电中游精选 +1pp），轻端是价格底已稳但利润未到的光伏龙头（持平），减仓 NEV OEM (-2pp) 与多晶硅纯标的 (-1pp)。综…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI基础设施正经历‘公用事业化’重估，估值驱动力从智能稀缺性转向物理交付（能源、电网）的稀缺性，导致超大规模厂商与优质电力资产的估值倍数趋同。", "summary": "2026年5月，美股TMT板块估值逻辑发生重大转向：超大规模厂商（Hyperscalers）因资本密集度飙升（约25%）和FCF收益率受压（2.5%-4%）正被重塑为‘AI公用事业公司’。Token价格的掠夺性定价（降幅达97%）使‘能源套利能力’成为核心护城河。持有核电PPA的厂商利用成本优势挤…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "中国电网侧储能是新能源汽车电池分析框架中系统性遗漏的需求通路，2026年需求约120–155 GWh", "summary": "中国电网侧储能是新能源汽车电池分析框架中系统性遗漏的需求通路，2026年需求约120–155 GWh", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "技术深度是中国电网设备出海的核心护城河，但在美国AI数据中心场景中，第三国制造和退税只能提供部分对冲，真正决定订单能否进入关键供电路径的是可信交付、网络安全、认证和本地服务能力。", "summary": "截至2026-05-06，本卡对第8张卡片作有条件压力测试：技术深度确实比地理套利更重要，但第8张卡片需修正三点。第一，2024年11月中国退税公告取消的是铝材、铜材和相关油脂退税，并将部分成品油、光伏、电池、非金属矿物制品退税率由13%降至9%，不是新的钢铁/GOES退税事件。第二，美国AI数据…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-09", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "中国电力企业联合会（CEC）预测2026年中国新增风电95 GW、光伏240 GW，合计335 GW新能源装机。", "summary": "中国电力企业联合会（CEC）预测2026年中国新增风电95 GW、光伏240 GW，合计335 GW新能源装机。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "2027–2028 年公用事业稀缺租金属于同时握有排队号、变压器与可核查碳账本的环节——电网设备 OEM 与受监管 T&D 是最高确信度超配。", "summary": "支持 Research note 06 的政策综合判断：2027–2028 年公用事业 alpha 来自并网权、CBAM 可核查碳账本与可靠容量权益，而非新增装机 MW。第一梯队是订单排至 2028 年的高压电网设备 OEM（GEV、Siemens Energy、Hitachi Energy、Pr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 2, "neutral": 0, "value": 38.4, "events": [ { "title": "我国能在全国电量层面跟上AIDC基准需求，但局部电网承载力、小时级稳定供电和物理绿电交付仍是规模化扩张的硬约束。", "summary": "截至2026-05-15，本卡压力测试卡片07：国产变压器/GIS交付提速可以缓解AIDC物理接入瓶颈，但不能自动解决电力系统承载问题。全国年度电量层面，可再生能源新增发电量和2026年新增装机足以覆盖基准情景下的数据中心增量；但AIDC负荷高度集中、要求稳定MW和高电能质量，约束会转向区域电网节…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "80%绿电要求在全国年度电量层面可实现，但在2026-2027年仍会通过区域交付、时段匹配和并网接入形成硬约束。", "summary": "工作日期2026-05-16。全国绿电年度增量充足：2025年可再生能源新增发电量5193亿千瓦时，覆盖全社会用电增量5161亿千瓦时，2026年新能源新增装机预计仍超3亿千瓦。但80%绿电约束的瓶颈在区域、时段、并网、储能和可签约绿电匹配。核电占比仅约4.5%-4.8%，难以在1-2年内普遍解约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 1, "risks": 5, "neutral": 0, "value": 29.8, "events": [ { "title": "煤电与可再生能源能缓解 AI 用电增量，但在 2026-2028 年无法单独解决局部可靠容量与 time-to-power 瓶颈。", "summary": "截至2026-05-16，本卡支持前序“电力可交付性”瓶颈判断。煤电和可再生能源在总量上可覆盖 AI 数据中心新增用电的一部分，但无法在电网扩容受限时单独填补缺口。美国煤电 YTD 发电量同比 -11.8%，风电仅 +1.4%，太阳能 +20.1%，但公用事业总发电仅 +0.2%；PJM/MISO…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在铜铝供给弹性有限的环境下，2026-2030年可再生能源扩张的关键瓶颈将从发电设备转向电网扩容、并网能力和关键电力设备交付。", "summary": "2026-05-17：本卡支持Research note 04并从公用事业角度确认，铜铝供给弹性受限会把电网扩容推成大规模可再生能源并网的主要物理瓶颈。IEA显示电网建设需5-15年，全球超过2,500 GW项目卡在并网队列；LBNL显示美国截至2024年底有1,400 GW发电和890 GW储能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-17：下一个瓶颈不是光伏板或泛化发电建设能力，而是可交付的稳定与灵活电力，尤其是燃气轮机、海风执行和已锁定并网资源。", "summary": "2026-05-17：本卡压力测试 Research note 07 的电力基础设施轮动，结论是否定“光伏/泛风电组件将成为下一个系统性瓶颈”的宽泛叙事。IRENA 显示 2025 年全球可再生能源新增 692 GW，其中光伏 511 GW、风电 159 GW；IEA 和 EIA 数据也显示光伏与…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "应对AI资本开支降速的最佳多资产策略是降低工业基建波动敞口，向受监管的电网公用事业与天然气、核电等稳荷能源资产进行防御性轮动。", "summary": "截至2026-05-16，本卡基于前序电力瓶颈研究，对多资产配置提出综合策略。由于可交付性与排队瓶颈拉长了AI基建的变现周期，建议在长期组合中将超配的“拥挤AI电力设备Beta”下调20%-30%。同时，提升具备现金流确定性的受监管公用事业（电网现代化驱动）权重，并增配天然气、核电及储能等“调峰稳…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "IEA《Electricity 2026》指出，全球超过 2,500 GW 的新能源、大负荷与储能项目正卡在电网排队中；到 2030 年，年度电网投资需从目前的约 4,000 亿美元…", "summary": "IEA《Electricity 2026》指出，全球超过 2,500 GW 的新能源、大负荷与储能项目正卡在电网排队中；到 2030 年，年度电网投资需从目前的约 4,000 亿美元提升约 50%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 42.6, "events": [ { "title": "印度电子制造与数字基建的两条防御腿方向上仍成立，但电力可用性、CBAM/RE100 隐性碳成本与 RE PPA 稀缺性应作为显式贴现项，使估值额外打 200–300 bp，并新增 R…", "summary": "对卡片 02 的'两条防御腿'做能源-电力-碳侧压力测试：(i) 印度南部 EMS 与数据中心集群所处电网在 2025 夏季实际可调度仅约 252 GW 对峰荷 250 GW，储备裕度近零，AT&C 损耗 15%+，自备柴油 + 屋顶光伏 + 企业 PPA 三重备份成本占 EMS 制造现金成本 3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "‘P/Watt’时代正面临物理修正：电网瓶颈与 SMR 交付延迟意味着 2026-2028 年的算力扩张受限于存量能源基础设施，而非芯片产能。", "summary": "本报告对 AI 电力供应进行压力测试，指出电网并网周期（6-8年）和 SMR 商业化时间表（2029年后）与算力扩张速度严重错配。可再生能源因间歇性导致稳定性成本过高（$160/MWh），无法独立支撑 AI 需求。结论认为，SMR 是 2030 年代的方案，无法解决当前的物理极限，存量核电和具有‘…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 2, "neutral": 0, "value": 44, "events": [ { "title": "能源物流没有被 AI 消灭，但“物理大宗必要性”不等于“能源大宗运输贝塔免疫”，真正稀缺性正在迁移到电力设备、燃机、储能和电网集成。", "summary": "截至 2026-05-19，工业制造视角否定卡片 05 的强版“能源物流战略性免疫”。能源分子仍需大宗运输，但 AI 电力需求首先制造的是电气设备、燃机、储能、开关设备、变压器、并网和控制系统瓶颈，而不是对 VLCC、煤炭铁路或普通港口大宗贝塔的广谱保护。IEA 2026 年报告显示 hypers…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-20核心判断：AI拥挤硬件交易降杠杆不会消灭算力需求，但会把估值溢价从GPU叙事转向可交付兆瓦和电力基础设施订单可见度。", "summary": "2026-05-20：支持“电力与能源供应确定性为算力产业链提供新估值中枢”的判断，但限定在可交付电力、电网设备、燃机/核电排产、长期PPA和具备成本回收机制的公用事业资产。IEA、EPRI、EIA/PJM/NERC与PJM价格数据共同显示，数据中心负荷已从AI叙事进入电力系统硬约束。策略上应继续…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 0, "neutral": 2, "value": 63.4, "events": [ { "title": "能源基建中短期的刚性供应弹性无法覆盖 92GW 的算力电力缺口，电网升级周期与电价费率暴涨正将 AI 革命的经济租金向电力公用事业板块转移，严重挤压算力链的利润空间。", "summary": "本卡片评估了能源基建面对 92GW 算力缺口的供应弹性。分析表明，受制于可再生能源间歇性、输配电线路 5-8 年建设周期以及电池储能的短时长特征，电力基建弹性严重不足。弗吉尼亚 GS-5 费率新规（14年期照付不议合同）与 PJM 容量电费暴涨 11.4 倍（至 $329.17/MW-day）正大…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "能源供给侧足以支撑AI硬件的长期需求，但短期决定出货节奏的是节点级通电能力，而不是全国电量总量。", "summary": "截至2026-05-21，能源侧扩张能够维持AI硬件长期需求，但不能完全消除2026-2027年的局部“电力之墙”。美国2026年计划新增86 GW公用事业级容量，绿电与储能增量可观；中国凭借风光装机、储能和State Grid的RMB 4 trillion电网投资，缓冲能力更强。但变压器、并网队…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 0, "neutral": 0, "value": 66.8, "events": [ { "title": "AI算力扩张确实开启电力基础设施资本开支周期，但2026-2028年变压器、并网排队和24/7可靠性要求会先放大时间尺度错配，再逐步由费基投资、储能和可调度电源缓解。", "summary": "截至2026-05-22，本卡支持并强化能源分析师的“时间尺度错配”判断：AI数据中心负荷增长快于电网硬件、稳定电源和新能源并网的交付速度。IEA预计全球数据中心用电量从2025年485 TWh升至2030年950 TWh，DOE/LBNL预计美国数据中心2028年用电量达325-580 TWh；…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-24", "supports": 1, "risks": 4, "neutral": 0, "value": 44.2, "events": [ { "title": "电池安全溢价与补贴不确定性将抬升美欧储能LCOS高达14-29%，导致装机进度延误15-20%并引发并网弃电率飙升，成为阻碍美欧2030/2035年清洁能源目标达成的核心硬约束。", "summary": "本卡片评估了$30-40/kWh的电池安全溢价与补贴退坡风险对美欧储能及能源转型的冲击。在极端压力测试下，美国储能LCOS将飙升29.2%至$142.92/MWh，引发15-20%的装机延误，导致高比例新能源电网弃电率升至12-18%，最终拉大美欧2030/2035年可再生能源部署目标的缺口。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球能源转型提供量的支撑，但中国电力设备出口链的利润弹性主要来自差异化系统价值，而非通用组件价格反弹。", "summary": "截至2026-05-25，本卡支持出口型高端制造主线，但强调利润分化。全球可再生能源装机和电价波动支撑中国光伏、风电和电力设备出口量；但大宗光伏组件仍受过剩和低价竞争压制，海外利润更可能集中在高效率组件、光储一体、合规本地化产能、风电服务包以及电网/储能设备。", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "电力中枢具备承接TMT筹码清洗后防御资金的空间，但防御仓位应买现金流和电网瓶颈，而不是买整条新能源制造链。", "summary": "截至2026-05-24，支持从拥挤TMT向电力中枢做部分防御轮动，但对象必须收窄到绿电现金流、受监管电力资产、精选电网瓶颈和储能运营。电力需求具备高确定性：2026年1-4月全社会用电量同比增长5.4%，互联网数据服务用电量同比增长44.4%，充换电服务用电量同比增长55.9%。绿电估值和两融强…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力需求不仅未能修复Tier-2/3新能源业主的IRR缺口，反而通过24/7 CFE门槛和极端违约金机制充当了可融资性过滤器，进一步巩固了Tier-1巨头的寡头垄断地位。", "summary": "截至2026年5月25日，本卡片对“AI电力需求将修复所有新能源企业IRR”的假说进行了压力测试并予以否定。超大规模数据中心对24/7无碳能源（CFE）的强制要求及高额延迟违约金（LDs）形成了极高的信用与资金壁垒。Tier-2/3开发商无力承担此类复杂资产组合与违约风险，被完全排除在$15-$2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 2, "risks": 1, "neutral": 2, "value": 77.6, "events": [ { "title": "AI算力是2026–2030电力系统首要结构性需求驱动,定量补强'公用事业由防御收息转向结构性成长'的重定价判断,焦点应放在现货敞口+稳态电源+电网超级周期。", "summary": "AI算力建设是本十年全球电力系统最大可识别的需求侧冲击:IEA基准下2030年前全球数据中心用电增量约+530 TWh,美国2028年占比升至6.7%–12%,中国规划2030年~400 TWh,且高度集中在北弗吉尼亚、ERCOT、东数西算枢纽等少数节点。PJM 2025/2026容量拍卖出清$2…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "光伏与电池「客户重定价」是结构性+政策性的二轮深挤压，2026 一线组件毛利率压力中枢 3.5%-6.5%，新能源硬件是空头平仓真空过后最先掉队的 cluster。", "summary": "本卡片（2026-05-25）压力测试 card-06 对新能源硬件「-1.0~+1.0 pct」温和稀释的判断。在产能利用率<65%、长协重签价格中枢已落至 0.10-0.11 USD/W、美国 FEOC + 14.25% 关税无法转嫁、欧盟 NZIA 提升非价格权重的多重重定价压力下，一线 T…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国电力设备与新能源产业链已具备从低碳生产走向低碳可核验生产的全链条能力，凭借绿电直连+证书+核验+数字护照组合可对CBAM及地缘关税实现实质性免疫，并在估值上结构性受益。", "summary": "2026-05-25：本卡作为十卡终章，论证中国电力设备与新能源产业链（光伏、储能、风电、特高压/输配电）通过绿电直连+CCER/I-REC+第三方核验+数字护照四件套，可将CBAM及地缘关税的有效等效税率从当前5%-8%压缩至1%-2%以内，实现实质性免疫。绿电直连净收益约6-10倍，全链条核验…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "2026Q3应持有“利润留存中国”并对冲“重定价中国”：自动化和现金流防御资产增持，新能源硬件、材料beta和科创拥挤动量减持，基准对冲比率0.45x。", "summary": "截至2026-05-25，本卡综合1-7卡后认为：负Gamma平仓真空之后不应简单卖出全部中国权益，而应增持能留住降本和现金流的A/H资产，减持会被客户重定价和流动性冲击伤害的板块。2026Q3建议超配自动化、机器人、电网、公用事业、精选现金流充裕港股互联网；低配光伏组件、商品化电池、锂铜材料be…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 2, "neutral": 0, "value": 51, "events": [ { "title": "截至 2026-05-26，AI 电力维度的核心不是总装机不足，而是可在正确节点按期交付的 firm MW 稀缺，这会放大存量核电、输电和已并网可再生储能组合的估值分化。", "summary": "截至 2026-05-26，本卡支持上一张“算力—电力—资本结构”框架：AI CapEx 的电力瓶颈正在从长期 ESG 叙事变成近端 time-to-power 和资本回报约束。IEA 预计全球数据中心用电到 2030 年约 945 TWh，美国数据中心贡献美国电力需求增长近一半；DOE/LBNL…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "传统的公共电网4-7年并网延迟迫使AI巨头转向表后自发电与自带电源，然而重型燃机5年超长交付周期及核电/储能共址监管演变，将物理能源供给侧瓶颈牢牢锁定在极其有限的重工业装备产能与核心…", "summary": "AI算力爆发面临4-7年并网与极度紧张的燃机、核电供给约束，物理电力缺口高达24.5 GW。本报告深研天然气、核电及可再生能源供给侧的产能赤字与政策博弈。重型燃机面临5年最长交付周期且价格暴涨50%，迫使运营商转向BTM自带电源。核电在经历Talen-Amazon共址风波及FERC 2025年底共…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 2, "neutral": 0, "value": 52.4, "events": [ { "title": "AI 叙事正从芯片竞争迁移到能源供应竞争，而把 2026–2028 钉死的不是发电装机，而是变压器、HV 开关柜与取向硅钢这条狭窄走廊。", "summary": "支持卡01的'物理墙'框架，并把瓶颈具体化到电力设备链：美国大功率变压器交付周期已拉长至120–210周、北美互联排队>2,600GW、取向硅钢(GOES)成为物理瓶颈。GEV/ETN/ENR.DE/Hitachi Energy/POWL 等的订单簿连续6个季度book-to-bill>1.2，订…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 电力缺口是真实但温和的核心 PCE 楔，叠加电网设备供应链的中美对抗，将以「延迟降息 + 类滞胀」形态而非「重置通胀目标」形态影响美联储，并把宏观风险从拥挤的一阶 AI 硬件转…", "summary": "AI 电力交付滞后通过三条渠道（批发电价→服务、云算力涨价→SaaS、变压器 PPI→capex 平减指数）对美国核心 PCE 形成 2026 +8~18 bp、2027 +10~22 bp 的持续抬升（自测算），不足以重置 2% 目标但足以让核心 PCE 2027 年底仍卡在 ~2.4%，相对 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 59.8, "events": [ { "title": "电网配套capex温和扩张被高端制造结构升级与海外溢价放大为头部设备公司的高弹性订单与毛利改善——超配特高压主设备、配网自动化、储能PCS与出口变压器，回避陷入价格战的光伏组件与风机…", "summary": "截至2026-05-28，本卡支持card-06能源分析师关于'电网与配套才是真正瓶颈'的判断，并从设备端给出量化传导。判断2026-2028是电网设备'弱周期+强结构+海外溢价'叠加的景气区间：国网+南网2025 capex约8200亿元，2026继续增长5-10%；特高压主设备(640-880…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 0, "value": 55.2, "events": [ { "title": "财政扩张能保护新能源的实物资本开支与部分受监管现金流，但不能把高实际利率下的新能源制造beta重新定价为全行业成长久期。", "summary": "截至2026-05-29，本卡片压力测试新能源与电力基础设施在高实际利率和财政扩张下的回报韧性。结论是财政补贴资本开支能托住装机量、设备订单和部分电网/储能/特高压现金流，但无法全面抵消新能源全产业链估值压力。核心分化在于：电网设备、独立储能容量电价和高质量运营商具备防御性；光伏制造、锂电材料和依…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力的物理约束正从“有没有电”转向“能否在正确节点、正确时点、用合格设备按时送电”。", "summary": "2026-05-29：本卡支持上一张TMT二阶硬件主线，但加入电力基础设施折扣。AI算力扩张不是中国全国层面的总量缺电问题，而是局部电网容量、变压器/开关设备交付、新能源消纳和高功率密度冷却/备电的时点约束。中国互联网数据服务用电1-4月同比增长44.4%[S2]，但仅占全国同期用电约0.94%[…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 2, "neutral": 0, "value": 56.6, "events": [ { "title": "电网瓶颈真实但可移动：hyperscaler把边际GW送往ERCOT/海湾/东盟，许可改革与表后燃机/核电是旁道，关税让设备短缺更集中——结果不是美股公用事业广义牛市，而是少数OEM…", "summary": "从美/欧/日宏观与地缘政治视角对Card-05的'电力是新瓶颈'主线做压力测试。我不否定电网瓶颈，而是为它划界：(1) 边际AI GW正在通过Stargate ERCOT、Stargate UAE、沙特HUMAIN、马来西亚柔佛等通道离开美国队列，部分掏空美股公用事业敞口；(2) Trump 2.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "SiC上游产能过剩带来的器件价格暴跌，通过缩减SST与PCS的占地及散热要求转化为下游巨大的CAPEX红利，系统集成商借此成功跨越内卷，获取技术溢价。", "summary": "本次调研确认，SiC衬底的产能过剩正推动利润池向电力电子产业链下游转移。SiC器件成本降至2-3美元/A，较硅基IGBT溢价压缩至10-20%。通过10-50 kHz的高频开关，SiC使中压SST与储能PCS的磁性元件体积缩减60-70%、散热需求减半，实现单MW系统体积从2.0 m³缩至1.2 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 3, "neutral": 2, "value": 58, "events": [ { "title": "电网通胀真实且足以使受损可再生项目无杠杆 IRR 压缩 200–400 bps，但只会使开发商群体分化而非中止能源转型，经济租金正从开发商转移到电网设备制造商与 AI 锚定的 off…", "summary": "电网设备与安装成本的结构性通胀确实部分抵消了组件/电池硬件的学习曲线，但并未否定能源转型。Lazard LCOE 上沿停止下移并走阔，BOS 占比上升使温和的 BOS 通胀对总 capex 的拉升超过组件通缩的下拉。自测算显示，美国典型公用事业级 PV 项目在 BOS +30% 叠加 COD 延后…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "“算电协同”下的资产配置重构：电网投资重心向配电与数字化转移对2026 H2跨资产权重的驱动", "summary": "把 Research note 07–09 的“算电协同”主线翻译为跨资产权重：A 股内能源由低配升至标配偏多 (+150 bp)、公用事业升至超配 (+250 bp)、TMT 由超配降至标配偏多 (−250 bp)；跨资产层面信用债+200 bp、商品+100 bp、港股+100 bp、海外股票…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "公用事业链不是低关注度故事，而是煤电容量电价+ETS 扩容+现货市场三重重定价的早期窗口——可调度资产估值结构性上修，新能源 IPP 估值结构性下修。", "summary": "公用事业链 30 日激活率仅 11.7%，但正同时穿越三大体制切换：煤电容量电价 2026-01 提至固定成本 50%（高转型省 70%）；H1 2025 弃光 6.6%/弃风 5.7%、国家红线放宽至 10%；ETS 扩容至钢/水泥/电解铝（覆盖 +30 亿吨 CO₂e）。可调度容量被向上重估、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网瓶颈不是欧美潜在GDP的硬天花板，但已是AI生产率红利兑现时点的2-3年延迟与核心商品PPI的持续高位驱动；Section 232升级是决定它停留在过渡税还是固化为真正上限的…", "summary": "截至2026-05-31，本卡片对Cards 03-05的微观结论做宏观压力测试：电网瓶颈+OEM定价权意味着电力设备PPI（BLS WPU1174自2020年累计+65-70%）将持续高于核心目标，并把成本转嫁给整体欧美Capex基数。美国T&D投资从2024年约1700亿美元升至2025E的2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 2, "neutral": 0, "value": 59.4, "events": [ { "title": "AI电力约束不是全球TWh硬上限，而是区域电网可交付能力上限：谁能获得firm节点容量、长交期设备和可回收tariff，谁才能把算力订单转化为可运行园区。", "summary": "截至2026-06-02，本卡压力测试AI数据中心电力约束：全球年度电量尚可管理，但Northern Virginia/PJM等热点的本地可交付容量、变电站、变压器、电缆和峰时备用已成为算力扩张的区域性物理上限。IEA预计数据中心用电从2024年415 TWh升至2030年945 TWh；LBNL…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力稀缺在A股已进入微观拥挤度与解禁压力出清期，应极度收缩高换手高杠杆的CPO/TMT基础设施与储能，超配具备高壁垒的二次电网设备与低杠杆容量火电公用事业。", "summary": "截至2026-06-02，本卡对AI电力稀缺框架在A股中的四大持仓进行了压力测试。测试显示，TMT基础设施（CPO/中际旭创）超配偏离达+315 BP、两融杠杆比率6.5%、解禁比例达12.0%，在动量反转和北向撤资下极度脆弱（情景B下stressed回报为-49.66%）。储能面临产能过剩与8.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 3, "neutral": 0, "value": 54.8, "events": [ { "title": "截至2026-06-03，AI–电力链的确定性溢价来自“在受约束节点交付可靠一度电的权利”，而不是泛能源需求增长本身。", "summary": "截至2026-06-03（Asia/Singapore），本卡支持上一张卡片的“重资产基础设施算力”主线，但把可投资确定性收窄到电网稀缺。AI 数据中心不是简单制造全球能源短缺，而是在 PJM、ERCOT、中国东部负荷中心等节点制造可靠电力、并网、变压器、输电和成本分摊约束。全球数据中心用电由 2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年“碳排放双控”考核硬化与“算电协同”基础设施工程建设，正在深度重塑绿电和地方公用事业资产估值，精选机构低覆盖、具有资产壁垒和分红确定性的细分龙头或反转标的将获得超额收益。", "summary": "本报告对A股公用事业产业链展开深度调研。分析了2026年全面落地的“碳排放双控”考核办法及交易所ESG强制披露规则等政策影响。筛选出机构覆盖率低的4个价值标的（芯能科技、洪城环境、广西能源、百通能源），覆盖分布式光伏储能、高股息水务、困境反转绿电网及工业区热力等方向，并详述了四类跨链溢出风险，推荐…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI-电力跨行业组合应以三筛选器分桶配置：3/3命中的美国wires-first IOU拿核心30%，2/3命中的输配电EPC、电网设备、中国UHV各15%卫星仓位，AI-infra…", "summary": "截至2026-06-03，本卡综合cards 01-05，用'客户先付风险(CR)、监管可回收(RR)、节点受约束(NC)'三筛选器对AI-电力链跨行业组合定权重。3/3命中（在数据中心集群拥有获批大负荷电价的美国wires-first IOU，如AEP、Dominion、Southern、Duk…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 2, "neutral": 1, "value": 65.2, "events": [ { "title": "截至2026-06-03，30% BTM是可靠容量情景而非纯清洁电量情景，其结果是加速挤压不灵活基载电源并提升燃气、储能、微电网控制和24/7签约资产的相对收益。", "summary": "截至2026-06-03，30% BTM情景可以成立，但前提是将其定义为数据中心自控的可靠容量，而不是纯光/风/储24/7清洁自供。光伏、风电和储能成本已足以支撑低成本电量层，但4-8小时锂电不能提供季节性可靠容量；现实组合是燃气或其他可调度电源提供可靠性，光/风提供低边际成本电量，电池负责爬坡和…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在低利率和容量/价格改革下，公用事业个股通过煤电一体化、资产并表和节能换挡展现出强劲的债性防御力与价值重估空间。", "summary": "本报告系统调研了公用事业产业链，筛选出瀚蓝环境（垃圾发电全资并表粤丰环保）、陕西能源（自产煤平滑波动的煤电一体化）、涪陵电力（一季度配网节能业绩迎来拐点）和皖能电力（容量电价上调直增4.39亿净利）四个隐形冠军标的；并提示了算力绿电 PPA 溢价、大宗电价传导、电网设备投资和环保坏账拨备等跨链溢出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "篮子内部从硅侧55%→25%、电网设备12%→22%、燃气/中游7%→15%、IPP 7%→13%；S6持平但内部SO超配/AEP低配；篮外按S4+S6久期25%做利率对冲、按S3净…", "summary": "锚定2026-06-03。综合card-01至card-05结论，AI基础设施篮子（100%）应重新分配为：S1硅侧25%（vs共识约55%，大幅低配）、S2电网设备22%（大幅超配）、S3燃气/中游15%、S4可再生+储能12%、S5 IPP/24-7签约13%、S6受监管公用事业10%（内部S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 0, "value": 63.6, "events": [ { "title": "漆包电磁线等废铜（2号废铜）必须二次提纯的冶炼瓶颈，与铝替代面临的体积和GOES消耗连带成本，共同将绕组端变压器交付周期推升至最高127.5周，彻底击穿了无保障可再生能源项目的并网确…", "summary": "本报告深入探讨了变压器核心金属（铜、铝）的供应侧瓶颈。2026年全球精炼铜面临33万至60万吨的结构性赤字。高压变压器电磁线所需的99.99%+高纯度阴极铜受限于二次冶炼和电解精炼产能的28万吨上限瓶颈。敏感性模型显示，在高电网需求（5800 GVA）下，二次精炼缺口达12.6万吨，导致绕组端变压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器短缺与并网排队拖延将导致缺乏并网保障的新能源资产在2026年后沦为‘搁浅资产’，负值ROIC与债务违约将刺破高杠杆绿电的现金流幻觉。", "summary": "变压器交付延迟与电网并网排队增加将严重恶化新能源资产2026年后ROIC及现金流。压力测试表明，在2年和4.5年并网延迟下，光伏项目ROIC分别从1.99%暴跌至0.49%和-1.10%。因资本化利息上升与PPA重签/取消，EBITDA/DSCR跌破1.0x，引发大面积违约与资产减值风险，新能源资…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 71, "events": [ { "title": "AI电力需求强化了从拥挤AI股票向实体基础设施的轮动逻辑，但可投资主线是time-to-power稀缺，而不是泛能源重估。", "summary": "截至2026-06-06，AI数据中心用电增长已从主题叙事进入电力系统硬约束：IEA预计全球数据中心用电从2025年485 TWh升至2030年950 TWh，LBNL给出美国2028年325-580 TWh区间。PJM已出现容量市场和输电压力，数据中心贡献约30 GW的2030年前峰值负荷增长。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 0, "value": 72.4, "events": [ { "title": "AI Scaling 的瓶颈已转向已上电能力，但中东和东南亚只能承接旗舰集群，不能在2030年前消除全球GW级电力墙。", "summary": "2026-06-07：本卡压力测试中东与东南亚是否足以承接GW级AI电力需求。结论是局部支持、整体否定：UAE和沙特可依托国家协调、燃气/核电/可再生能源组合，承接首批100MW至1GW级主权AI集群；马来西亚/柔佛是东南亚最强外溢阀门；但变压器、变电站、燃机、跨境输电和冷却用水仍是硬约束。东南亚…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 0, "neutral": 1, "value": 82.8, "events": [ { "title": "绿电直连+CBAM构成的制度套利正使全球AI算力向中国西南、北欧、海湾三条低碳走廊集中，并把可再生能源成本曲线从单一LCOE改写为LCOE+合规溢价+并网期权的三维曲线。", "summary": "截至2026-06-07，PJM容量费、NDRC绿电直连与欧盟CBAM三者叠加，形成约$30–40/MWh的合规折扣台阶，把全球AI算力的边际产能向中国西南/西北、北欧和海湾三条低碳走廊集中。中国绿电直连园区凭物理溯源+小时级证书在CBAM申报中嵌入碳近零，叠加$44–53/MWh的直供电价，对比…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 4, "neutral": 0, "value": 57.2, "events": [ { "title": "2026-06-10的核心判断是：AI算力从订单量转化为收入的闸门正在变成本地、连续、可调度的有效电量，而不是全球能源总量。", "summary": "截至2026-06-10，本卡压力测试支持前序判断：AI算力瓶颈正在从芯片订单转向风险调整后的有效电量。IEA预计数据中心用电由2025年485 TWh升至2030年950 TWh，AI专用数据中心用电同期约增至3倍；但全球超过2,500 GW项目卡在并网队列。可再生能源能提供最大增量，却需要储能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力的“物理壁垒”正将全球投资范式从“虚拟算力（软件/芯片）”推向“物理确定性（电网/基荷/设备）”，建议通过中美区域套利与基建杠铃组合锁定跨周期收益。", "summary": "截至2026-06-09，本报告针对AI电力瓶颈引发的全球区域竞争力分化，提出“物理AI”投资策略。美国凭借节点低电价和核电复兴维持算力核心地位，中国通过电网建设速度和可再生能源盈余转化为设备链和高耗能产业优势，欧洲则面临区域碎片化导致的去工业化压力。组合配置应向“价值+确定性”风格轮动：超配具备…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-09，AI电力瓶颈最优交易不是追逐SMR叙事，而是以公用事业、工业电气设备和精选材料为核心的现金流支撑型基础设施重估。", "summary": "截至2026-06-09，本卡支持上一张卡的AI电力瓶颈主线，但把策略从核电叙事扩展为A股板块轮动：超配有现金流和电网接入价值的公用事业、工业电气设备和储能链，标配精选TMT，低配尚未商业化的能源概念。关键证据包括IEA、EPRI、Goldman Sachs和中国“人工智能+能源”政策数据。组合应…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "绿电总量足以支撑AI能耗激增，但西部弃光率高企与2026年“确保80%绿电占比”的刚性政策，使执行风险集中于“算电协同”与绿电直供能力，而非单纯的电网容量。", "summary": "支持 card-02 的电网扩容逻辑，但进行碳约束压力测试：总量上可再生能源年新增装机超2亿千瓦足供AI（2025年用电约196TWh），但结构性消纳瓶颈显现。2026年Q1全国光伏利用率降至90.6%（逼近90%政策底线），青海、新疆等西部枢纽弃电率超10%。为对冲AI碳足迹，政策要求2026年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 0, "value": 76.6, "events": [ { "title": "SpaceX IPO 不是充分的卖出 AI 龙头信号；更强的能源判断是 AI 算力正在被重新定价为电力支撑的基础设施资产。", "summary": "截至2026-06-11，本卡片否定“SpaceX IPO 本身触发广谱 AI 龙头去拥挤”的简化判断。能源证据显示，真正约束是可靠电力、电网接入、容量市场和燃料结构。Google-SpaceX 算力合同证明已交付算力的稀缺价值，但 PJM 容量价格、IEA/DOE 数据中心用电预测和 Grid …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 4, "neutral": 0, "value": 60, "events": [ { "title": "组合应通过严格的个股与因子级仓位上限、协方差及公用事业指数贝塔偏离度控制，将新型电力系统、深水油气、AI算力和储能灵活性进行结构性隔离，防范电网资本开支逻辑的重复计算，并系统性执行基…", "summary": "本报告为2026-06-11制定的A股组合风险预算设计。为防范对电网资本开支的重复计算，将组合分解为电网资本开支贝塔（35%）、海洋能源贝塔（20%）、AI设施资本开支贝塔（20%）和灵活性变现贝塔（15%）以及10%的防御现金。明确了各因子仓位上限、协方差与公用事业贝塔约束，并设定了基于电网招标…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2030年全球发电总量增速虽能覆盖950 TWh的数据中心需求，但电网物理接网与清洁基载电源交付周期的刚性瓶颈，将在关键节点引发局部的电力供应危机，迫使科技巨头锁定气电并推迟碳中和目…", "summary": "截至2026-06-12，压力测试表明，尽管2030年全球可再生能源和核电发电增量可覆盖数据中心950 TWh需求，但受间歇性可再生能源消纳、核电建设长周期（SMR需2030年后交付）及网架并网延迟等结构性限制，局部电力供需失衡严重。超大型云厂商正大举锁定气电与重启核电机组，被迫推迟碳中和目标。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网材料瓶颈倒逼超大规模算力企业绕过公共电网，通过“表后天然气+同址核能”的混合模式获取电力，在短期内推迟了可再生能源的转型。", "summary": "[2026-06-11] 针对电网材料瓶颈，本报告进行了压力测试：材料短缺并未阻止AI扩建，而是引发了“绕过电网”的结构性偏移。为规避长达5年的并网等待和3-5年的变压器交付期，云厂商正转向建设周期仅12-18个月的“表后”天然气以满足即期需求，并投资SMR及核电重启锁定长期基载电力。这种“天然气…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "夏季能源通胀反弹与电网基荷电力短缺构成了双重尾部风险，足以打破美联储耐心和AI资本品扩张的“双速经济”假设，多资产策略必须引入能源与公用事业对冲。", "summary": "截至2026-06-12，本报告对首席策略师的“双速经济”配置策略进行了能源维度的压力测试。指出夏季高温和活跃飓风季可能推高天然气与原油价格，引发整体CPI超预期反弹，打破美联储“保持耐心”的假设并引发股债双杀。同时，电网基荷电力短缺构成了AI资本品实质上电的物理硬约束，可能切断资本品扩张动能。建…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 7, "neutral": 1, "value": 46.4, "events": [ { "title": "绿色通胀与电网物理约束正诱发新能源“平价红利”的阶段性逆转，AI基荷需求的高标准使得绿色电力在当前基建瓶境下变得比传统能源更昂贵，从而延缓全球能源转型进程。", "summary": "截至2026-06-13，压力测试显示，尽管新能源理论LCOE仍具竞争性，但受电网升级溢价、变压器交期溢价及FEOC合规成本冲击，美国新能源PPA价格已推升至$64.5-$79.4/MWh。在加入储能以满足AI算力24/7基荷需求后，绿色电力总成本已升至$110-$125/MWh，逆转了相对于天然…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器供应瓶颈下的电力增量装机与电网消纳压力测试", "summary": "压力测试卡片01-02：变压器/GOES多年期约束的前提成立，但'储能自动救场'的推论站不住。延迟确实倒逼转向，但分裂为两条相反通道——负荷侧靠表后/孤岛BESS绕开变压器队列（对电网变压器需求构成温和蚕食），供给侧并网型储能仍需GSU变压器、被同一队列卡住；真正免变压器的杠杆是GETs（重导约翻…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "去全球化导致美欧电网设备及原材料支付高额地缘溢价，并通过公用事业费率基数和新能源购电协议（PPA）向终端电价传导，正在欧美引发系统性的‘绿色通胀’。", "summary": "截至2026-06-13，本卡压力测试表明，关税及安全限制使美国取向硅钢（GOES）及成品变压器面临422%和80%-150%的地缘溢价。这每年为美国公用事业费率基数注入150亿-200亿美元通胀，拉高零售电费1.5%-2.0%。同时，并网延迟和设备溢价推动美国太阳能PPA价格同比大涨13%至$6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供给增量与并网效率的线性增长无法对冲AI算力的指数级需求，导致2026-2028年出现结构性“电力之墙”，迫使算力Capex流向“表后”能源自治资产。", "summary": "截至2026年6月13日，压力测试显示核电与可再生能源的并网增量远不足以弥补GW级算力集群带来的刚性缺口。即便PUE优化释放了~41%的算力潜力，杰文斯悖论与122%的算力需求增长仍导致美国面临15 GW的净电力赤字。瓶颈已从芯片转向并网时滞（5-8年）与轮机产能（预售至2030年），迫使云巨头从…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 2, "neutral": 0, "value": 74.8, "events": [ { "title": "AI大负荷的24/7特性与并网队列中新能源主导的增量存在根本错配，新建基荷电源的长周期迫使云厂商抢夺存量电力资产，加剧公用电网缺口。", "summary": "截至2026-06-14，本卡片压力测试结论：发电增量无法匹配AI基建需求。AI需要24/7稳定基荷，但目前美国并网队列中95%以上为间歇性新能源。核电（含SMR）在2030年前无实质增量，天然气扩建受制于管道许可。这迫使云厂商抢夺存量基荷（如核电背靠背PPA），剥夺了公共电网资源。结论：电力供应…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国绿电装机增速够用，但电网扩容（特高压+主变+构网储能）是承接AI算力负荷的真正瓶颈，也是2026–2027年A股能源链最确定的盈利兑现环节。", "summary": "截至2026-06-13，中国绿电装机端继续超预期（截至2026-04风光累计>14.5亿千瓦，2025全年新增约3.5亿千瓦[S5]），但弃风率5.1%、弃光率4.6%[S6]显示消纳与跨省外送是真正瓶颈。IEA口径下全球数据中心用电从2024年415 TWh升至2030年945 TWh[S1]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 2, "neutral": 0, "value": 76.2, "events": [ { "title": "核能与可再生能源的物理与监管瓶颈决定了其无法在2026年后快速匹配AI算力用电需求，电力缺口将成为制约AI算力部署的刚性硬约束。", "summary": "在电网面临5-7年瓶颈的背景下，核能与可再生能源无法在2026年后立即解决AI算力的电力短缺。美国核电重启到2029年仅能提供2.24 GW，占25 GW新增需求的8.9%；SMR商用需等到2030年以后；BTM核电直供面临70%的FERC监管否决率；BTM光储微网实现零碳90%可用率需8亿美元/…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2027-2030年支撑945 TWh数据中心负荷的最优边际结构是新能源+储能为主、天然气调峰为辅、核电PPA做稀缺补强，而非押注单一电源。", "summary": "2026-06-14工作日结论：在80%+绿电匹配约束下，2027-2030年最优边际供电不是单一电源，而是新能源+储能作为能量主体、天然气作为调峰袖珍、核电PPA作为稀缺高质量基荷。Lazard与EIA口径均显示风光储成本显著低于新建核电和燃气调峰；IEA预计2025-2030年全球新增可再生装…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 1, "value": 74.6, "events": [ { "title": "能源侧供给容量能否支撑 AI 算力的高密度扩张：核能与绿电的容量弹性分析", "summary": "本卡从能源侧支持算电协同命题。判断是：在全国总量层面容量弹性绰绰有余，但价值释放需配套电网/储能与核电硬基荷。2025年中国新增可再生能源452GW(+21%,占新增83%)[S1]，在建核电36台约38.9GW[S4]。一年可再生新增蕴含约681TWh/年潜力[自测算]，在建核电建成约307TW…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI-to-Power没有被供应侧证伪，但应从“发电扩张交易”收敛为“可交付电力瓶颈交易”。", "summary": "截至2026-06-16，本卡压力测试AI-to-Power的物理供应基础：方向成立，但稀缺性必须从“新增发电”改为“可调度、可并网、可合同化、可交付的节点电力”。IEA预计数据中心用电2025-2030年从485 TWh升至950 TWh；美国增量约240 TWh。核电质量最高但近端主要是800…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI平直负荷与可再生能源间歇性的物理错配暴露了‘绿色PPA悖论’，在显著抬升弃风弃光的同时，刚性推高了对调峰容量资产的依赖与容量电价溢价，驱动火电灵活性改造与清洁基荷资产价值重估。", "summary": "截至2026-06-16，本卡片量化评估了AI数据中心平直负荷与可再生能源间歇性的错配。无储能下，典型100MW负荷 backed by 350MW风光PPA的弃电率达19.88%。配建100MW/400MWh储能可将24/7 CFE匹配率升至86.14%，但度电成本（LCOE）将上涨$6.76/…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "HALEU供应短缺与电网调峰压力对AI算力中心能源安全性的二次打击", "summary": "本卡量化了BYOP/SMR孤岛模式受阻后，落在AI算力能源安全上的物理性\"二次打击\"。HALEU方面：美国近期计划合计仅约21公吨、Centrus级联仅约0.9公吨/年，而单个1GW AI园区Xe-100初装堆芯即需约15–25公吨[自测算]，缺口达一至两个数量级，SMR无法缓解2026–2030…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 5, "neutral": 0, "value": 61, "events": [ { "title": "2026-06-16的核心判断：AI 电力需求会造成短期化石能源回潮，但低碳逻辑不是被瓦解，而是从年度绿电匹配升级为逐小时可交付的清洁稳定容量。", "summary": "截至2026-06-16，本卡压力测试认为：AI 基荷增长会在2026-2030年区域性推高天然气和部分煤电出力，但不会瓦解低碳转型逻辑。真正瓶颈不是绿电项目总量，而是并网、储能、输电和逐小时可交付能力。核电是最优清洁基荷对冲，但2030年前主要靠延寿、重启和增容，新建核电规模化偏2030年代。投…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI推理算力向中东极低电价地区的转移，证伪了欧美电网大规模升级的预期，将在未来两年内使得前期激进扩产的欧美电力设备供应商陷入严重的产能过剩与资产减值危机。", "summary": "中东极具竞争力的光伏电价（<1.5美分/kWh）正吸引AI算力落地，这种能源套利直接削弱了市场对欧美电网现代化升级的资本支出预期。欧美电力设备供应商在2024-2025年基于激进的本土AI负荷所扩建的庞大产能，正面临严重的订单断崖。同时，中东新建算力多采用“光伏+储能+算力”的表后微电网模式，避开…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年后AI扩张的关键分母应从名义MW调整为扣除并网时间、网络改造成本和可靠性折减后的可通电可靠MW。", "summary": "截至2026-06-17，本卡对前序“电力溢价”主线做压力测试：核电、长时储能和电网升级不会在2026年后阻断所有AI数据中心，但会在电力紧张地区成为大型液冷训练和高利用率推理园区的第二物理瓶颈。核电主要提升终值和能源期限，难以填补2026-2028年缺口；长时储能更像并网质量增强器，而非稳定发电…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核电邻近溢价真实但主要是2030+长久期期权；2026-2030年AI电力的边际供给和最干净的传统能源表达在于天然气基础设施与铀燃料循环上游（转化+HALEU），而非高倍数的核电邻近…", "summary": "截至2026-06-16，对第04卡‘核电邻近性=公用事业估值核心’的结论进行压力测试。能源视角认为：(1) 铀矿上游是下一个物理瓶颈，哈原工产量下调+Cameco满产+西方下一个一级项目NexGen不早于2028投产，长期合约价需突破USD 90/lb才能激励边际矿山；转化与HALEU比开采更紧…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 4, "neutral": 0, "value": 68.4, "events": [ { "title": "中国低碳铝扩张的核心制约正在从冶炼产能转向稳定、可追溯、可审计的低碳电力交付能力，电网消纳和绿电PPA稀缺会决定绿色溢价能否兑现。", "summary": "截至2026-06-17，本卡支持前序低碳金属绿色溢价逻辑，但将核心约束从“是否有足够新能源装机”细化为“能否向冶炼负荷稳定交付、认证和审计低碳电力”。中国2026年4月底风光装机已达19.1亿千瓦，但电解铝约13.2 MWh/吨的电耗使省级电网因子、绿电PPA、储能和跨省消纳成为低碳铝规模化的关…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力向电力设备的资本开支传导扩大了实体“流动性黑洞”，高企的无风险利率与公用事业“成长化”转型持续挤出长久期资产的配置资金。", "summary": "本报告支持AI基础设施分析师的判断，量化了AI电力需求向能源基建的传导。2026年全球数据中心用电预计达565 TWh，中国“十五五”电网投资将达5万亿元，且全球面临30%的变压器结构性缺口。这种大规模实体CapEx形成巨大的“流动性黑洞”，虹吸市场资金，推高资本成本，对医药等长久期板块估值维持持…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "BTM燃机与燃料电池凭借18个月的“电力交付速度”优势，在2026年联邦监管去约束与地缘政治引发的能源套利驱动下，已从备用电源演变为AI基础设施的核心基石。", "summary": "截至2026-06-17，BTM燃机与燃料电池已成为填补2026-2030年AI电力缺口的主导方案。GE Vernova、Siemens Energy和Bloom Energy正处于产能扩张超级周期，订单已排至2028年后。受益于2026年初EPA对碳捕集强制令的废除及“临时燃机”许可程序的简化，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源侧产能扩张只能在本地可交付、可调度且按期并网时缓解 AI 算力约束；在 2026-2028 年，多数新增供给更像中期宽慰而非近期上电解决方案。", "summary": "截至 2026-06-17，本卡压力测试 Research note 04：能源供给增加是必要条件，但不能在电网/变压器/调度延迟严重时自动解除 AI 算力的物理约束。美国 2026 年计划新增 86 GW 公用事业级电源，亚太数据中心用电预计 2024-2030 年从 320 TWh 升至 78…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 4, "neutral": 0, "value": 69.8, "events": [ { "title": "能源供应滞后不会在2026-2027年否定AI产业链逻辑，但会在2028-2030年通过燃机槽位、firm PPA与本地可调度容量约束美国新增集群。", "summary": "截至2026-06-18，能源供应端不是2026-2027年比变压器更严重的全球瓶颈，但2028-2030年会成为美国AI集群更硬的firm power约束。核电战略价值高但太慢；天然气是现实桥梁，但燃机等待平均约5年、最长约7年，且GE Vernova燃机积压可能延伸至2029-2030年；可再…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施交易的跨行业风险预算与风格轮动框架", "summary": "综合并落地 Card-01 的方向判断。核心洞见：电力稀缺链条看似分散（公用事业+工业+材料+核能），实则是同一个 hyperscaler capex 因子，回撤时内部相关性飙向1。故按风险贡献而非名义权重预算，拆成四个子篮子：A电网输配电（压舱石）、B现场发电（核心-卫星）、C核燃料/SMR（高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新增可再生能源并网受限于极低的ELCC折价与长达3年的变压器交期，无法缓解2026-2028年算力‘通电围墙’，已投运基荷电力与变压器硬件仍是核心去瓶颈敞口。", "summary": "截至2026-06-18，对可再生能源并网缓解AI算力缺口进行压力测试。结论表明，新增绿电并网受限于电网硬件交付周期与容量额定值（ELCC）严重贬值，无法缓解2026-2028年算力‘通电围墙’。太阳能ELCC在PJM/ERCOT已被贬值至10-13%，且大型电力变压器（LPT）面临128-144…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源设备供应瓶颈对AI电力需求响应的影响分析", "summary": "截至2026-06-19，本卡压力测试 Card-02 的 A/B/C/D 电力篮子：变压器与开关设备交期延长强化 A 电网输配电的压舱石属性，但也限制整个 AI 电力交易的可部署容量；现场发电 B 受益于并网延迟但受燃机/发电机组排产约束；核燃料/SMR 的 C 应从高 Beta 战术仓位降到更…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 7, "neutral": 1, "value": 56.2, "events": [ { "title": "新能源IRR压降会部分、滞后地进入AI算力中心PPA价格曲线，形成可管理但不可忽视的长期电力成本楔子，而不是AI Capex的根本性刹车。", "summary": "截至2026-06-19，本卡片压力测试认为，辅助服务成本导致新能源项目IRR下降0.5-0.8pp后，会通过新签、续签或可重开PPA部分传导至AI算力中心，但存量固定价PPA短期由开发商吸收。压力情景下，合理PPA加价约为0.012-0.020元/kWh；100MW IT负荷AI集群按1.20 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "上游变压器与GOES瓶颈已成为并网的约束性约束，量化上递延（而非摧毁）美国2027–2030可再生目标约55–90 GW并将投运延后12–18个月，而中国凭一体化与国家电网承购按期推…", "summary": "本卡片量化上游材料瓶颈对并网时序与2027–2030装机目标的影响，支持并延伸卡片01–03主线。美国并网队列已从受研究限制转为受钢限制：约2,060 GW在申请（95%光伏/储能/风电），但仅约13%最终投运、约77%撤回；电力变压器交期128周、GSU 144周，送电环节与研究环节共同约束。G…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "辅助服务扩容在物理上面临量价齐升，但在商业模式上因受两部制电价监管、高折旧与高内卷制约，无法成为公用事业板块的估值主升浪锚，其本质是新能源开发商的合规成本与利润分摊。", "summary": "在145元/吨极端碳价及系统惯量缺口扩大15-20%的情景下，辅助服务市场虽迎来物理规模扩张，但因其在盈利模式上面临政策约束与商业痛点，无法成为公用事业板块的估值主升浪锚。抽蓄受制于6.5%资本金IRR上限，调相机缺乏独立商业化路径，而电池储能面临高折旧及竞争内卷。相反，辅助服务扩容将推高新能源开…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "供给侧的并网及自备发电瓶颈使电力Slot的稀缺性长期成立，这不仅阻碍了AI硬件下修的负面传导，反而倒逼了数据中心内高密度液冷等节能技术的刚性需求。", "summary": "截至2026-06-20，本卡对能源供给侧响应能力进行压力测试。结论表明，可再生能源受制于电网并网排队（平均等待5年）和间歇性，而幕后自备（BTM）发电面临监管限制（如FERC否决核电共建）、环保许可及设备交期瓶颈，中短期均无法填补AI电力缺口。因此电力Slot稀缺性将长期维持，电网设备和液冷板块…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 1, "risks": 2, "neutral": 0, "value": 92.6, "events": [ { "title": "“Time to Power”交易下的能源供给侧产能与增量需求分析", "summary": "本卡从能源供给侧支持 Research note 04-07 的 time-to-power 论点。截至 2026-06-21，结论是：缺口不在电量（TWh），而在匹配 7×24 AI 负荷的“确定可调度容量”。扩产最快的是光伏+储能（2026 年 86 GW 新增中占 79%），但形态最不适配；…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "能源波动主要通过利润率波动和 CAPEX hurdle rate 传导，而不是直接终止先进制程扩产；优势集群与边际 greenfield 项目的回报差会扩大。", "summary": "截至2026-06-21，本卡压力测试能源与电力波动对先进制程 fab 和先进封装的成本传导。结论是：电价不是最大成本项，但在连续运行、EUV、洁净室、超纯水和良率风险下是高 beta 成本项。以 TSMC 2025 年数据自测算，电力约占制造营业成本 8.0% [自测算]；10% 电价冲击约拖累…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI用电争夺会放大局部绿电消纳压力，但中国更可能经历区域利用率下滑和转型节奏分化，而不是全国能源结构转型被迫倒退。", "summary": "截至2026-06-20，本卡压力测试AI集群争夺变压器、开关柜和配网接入窗口对中国绿电利用率的影响。结论是：全国能源转型不会因设备交付延迟被迫逆转，但区域和时点风险已经升高。2026年一季度风电利用率91.4%、光伏90.6%，已贴近90%政策底线；按利用率反推，当季隐含弃风弃光约573亿千瓦时…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 10, "neutral": 0, "value": 38, "events": [ { "title": "变压器及组件产能约束会直接延长公用事业并网积压，并削弱可再生能源接入承载能力，因为它们决定网络升级能否从纸面协议变成已上电资产。", "summary": "截至2026-06-21，本卡支持前序结论：变压器、GSU、变电站主变、配电变压器和上游GOES/CTC等组件短缺，会把并网研究积压转化为网络升级和上电施工积压。Berkeley Lab显示美国并网队列仍有超过2,060 GW发电和储能容量在排，且408 GW已有草案或已执行并网协议但尚未商业运营…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源供应侧的交付节奏与产能潜力评估：核电及可再生能源能否支撑Hyperscaler的电力需求窗口？", "summary": "截至2026-06-21，本卡从供应侧交付节奏支持线程结论：约束在可交付性与并网确定性，而非装机潜力或电价。供应分三种时钟：可再生+储能并网后约12–24个月通电，是2026–2028近端唯一可规模化供应（EIA 2026年美国新增86 GW、逾99%清洁，含43.4 GW光伏、24 GW储能）；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在能源价格常态化预期下，红利配置应向清洁电力转移，但须通过煤电组合对冲水电枯水期现货买电风险，并警惕规制电网在AI负荷爆发期的信用评级恶化与表后流失风险。", "summary": "截至2026-06-21，本研究支持将组合红利仓位从周期性煤油向公用事业转移的逻辑。水电（长江电力）在严重枯水（-15%）情景下面临现货回购惩罚，净利润下降33.5%且面临29.7%派息削减风险；核电（中国核电）受益于煤电联动，电价上浮5%可提升利润15.88%，并捕获5.8-8.4分/kWh的碳…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网设备是CapEx周期，输电与储能不是电价牛市；组合应提高中国长端国债和规制绿电权重，保留克制的设备超配，并压低AI基础设施主题风险。", "summary": "2026-06-21：本卡综合前序结论，建议把风险预算重配为中国长端国债35%、规制绿电30%、电网设备25%、AI基础设施10%。核心判断是：特高压和储能建设确定性强，但到岸电价回收偏弱，因此应持有建设周期、用久期对冲电价通缩，并避免把AI负荷重复资本化为公用事业电价上行。电网设备保留超配，规制…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 1, "risks": 3, "neutral": 0, "value": 89.4, "events": [ { "title": "2026-2030年特高压与绿电并网能缓解全国电量约束，却不能完全对冲智算中心的本地可靠容量和电价基差风险，因此仍支持从粗放算力堆叠转向高能效算力、软件应用和电网CapEx受益资产。", "summary": "截至2026-06-22，本卡对前序电力瓶颈判断进行压力测试：特高压、绿电并网、煤电灵活性、抽蓄和新型储能建设足以缓解全国性年度电量缺口，但不能完全消除智算中心在本地并网、24/7可靠容量和市场化电价基差上的风险。国家电网规划“十五五”15回特高压直流工程、跨省输电能力提升35%、每年支撑不低于2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张下的电力与电网供应压力及碳中和约束", "summary": "压力测试电力/电网/碳中和这一环：看多电网设备与绿色基荷的逻辑成立，但约束点被放错位置。它不是全国发电或电网容量不足——年用电已破10万亿千瓦时，2025年可再生能源消纳全部新增需求，国网十五五约4万亿元资本开支(+40%)前置。真正挤压在数据中心运营商层面，且是区域+监管性：东部算力集群电网约7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "布伦特高于$120/bbl若持续，将通过亚洲进口账单、大西洋再路由和LNG竞争强化能源安全溢价，但其自我抑制性意味着真正持久的投资信号是电力系统、能效和多元化供应CapEx，而非无纪…", "summary": "截至2026-06-22，本卡支持霍尔木兹冲击已成为宏观级能源供给冲击的判断，但限定布伦特持续高于$120/bbl是高压情景而非官方基准。霍尔木兹在2024年承载约20 million b/d石油流量和约20%全球LNG贸易，亚洲敞口最大。若高油价延续，贸易流向会转向大西洋盆地和美国成品油，LNG…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "AI算力扩张不会被能源安全溢价终止，但会被“可交付稳定兆瓦”重新定价，全球集群部署将向电力充足、并网确定、能效合规且成本分摊透明的地区集中。", "summary": "截至2026-06-22，本卡对能源安全CapEx如何约束AI算力基础设施作压力测试。结论是：电价上升重要，但更核心的瓶颈是稳定兆瓦能否按GPU商业化窗口交付。IEA、LBNL、EPRI与NERC数据共同显示，AI数据中心需求已成为局部电网级变量，尤其在美国、欧洲、中国与东南亚。能源安全溢价会推高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 8, "neutral": 0, "value": 52.8, "events": [ { "title": "绿色先进制造与城投信贷收缩将通过Capex砍减与地方投资减少，向电力设备及能源基建传导共计313.42亿元的需求收缩压力，并延缓光伏组件与储能电芯的价格修复。", "summary": "截至2026-06-23，本报告在金租及信贷收缩背景下评估其对电力设备与能源基建的传导。测算显示，全系统绿色制造信贷收缩将导致相关厂区配电及并网设备需求缩减100.03亿元；城投信贷收缩将导致地方能源基建投资缩减213.39亿元（对应新能源装机延期约6,096.86 MW）。合计拖累总需求313.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "80%绿电强配溢价与电网/可靠容量可获得性，作为AI算力建设的约束性瓶颈及其对电力与新能源投资主题的含义", "summary": "本卡片综合卡片01-03，给出统一判断：AI建设的约束性瓶颈是'可获得性'而非'价格'。80%绿电强配以绿证年度履约成本极低（约+1%电费），即便24/7固化绿电PPA溢价也仅占TCO约2-3%[自测算]，价格可被超大规模厂商接受并转嫁。真正闸控2026-2030建设节奏的是并网排队、可信容量与小…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Assess renewable energy curtailment impacts", "summary": "并网延迟（接入排队+输电不足）与全球变压器短缺两条瓶颈相互强化：前者机械推高弃电率，后者拖延COD、后移现金流，共同压缩项目ROIC/IRR。2025年1-2月中国弃风6.2%、弃光6.1%，西藏弃风弃光超30%；上半年新增风光约268GW超出电网吸纳。电力变压器交期约128周、GSU约144周。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网输变电容量与并网审批周期（而非总体发电量）已成为AI算力部署的终极物理刚性瓶颈，并已实质性挤压IT硬件采购预算。", "summary": "本报告对AI算力扩张的电网与能源瓶颈进行压力测试。分析表明，虽然电力设备订单充足，但美国并网队列平均达5.5年，且面临FERC 2026年6月大型负荷监管及弗吉尼亚GS-5等防止成本转嫁政策。变压器等重电面临GOES原材料短缺，Tier-1厂商积压至2029-2030年。表后核电直供监管被封死，燃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 5, "neutral": 0, "value": 80.2, "events": [ { "title": "订单与招标数据支持资金转向电网设备及新能源实物资产端，但尚未证明二级市场资金已完成全链条重配。", "summary": "截至2026-06-25，本卡支持“资金正从碳资产叙事转向新型电力系统实物资产”的方向，但限定为订单端确认。碳和绿证仍活跃；增量证据在国网特高压招标、电网设备公司中标、光伏组件、风电整机和储能采招。结论是：实物资产资本开支已明显落单，尤其是电网一次设备和特高压材料；但权益资金流、利润率和二级市场持…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "在风光（3-12个月建设周期）对特高压通道（18-36个月）分别具有6.0倍和2.25倍周期时滞的客观规律下，大基地并网瓶颈已导致2025-2026年新能源消纳率首次跌破95%政策红…", "summary": "截至2026-06-25，本报告对我国供给端（风、光、核）建设节奏与电网侧扩容周期的协同性进行压力测试。结果显示，光伏（3-6个月）和风电（~12个月）的建设速度大幅领先特高压外送通道（18-36个月），形成6.0倍和2.25倍的周期差[自测算]。这导致2025年风电光伏利用率在“十四五”期间首次…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力需求不是全球电量不可满足问题，而是本地可交付、可并网、可24/7保供的问题，因此应继续支持电力栈而非全线下调AI基建评级。", "summary": "截至2026-06-25，本卡支持整轮“电力栈”主线：全球核能、可再生能源与储能扩张足以让AI电力需求不是总量不可满足问题，但不足以消除本地并网、变压器、24/7保供和电网承载瓶颈。IEA预计数据中心用电从2024年415 TWh升至2030年945 TWh；可再生能源满足近50%的新增需求，化石…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-24，能源侧能部分承接 2026-2028 年递延算力需求，但早于 2027 年的真实硬约束已经是 2026 年站点通电失败，因此组合应从单纯 AI 需求代理转向…", "summary": "截至2026-06-24，本卡对 Research note 02 做压力测试：电力/电网确实递延而非毁灭 2026-2028 年算力需求，但硬约束并非等到 2027 年才出现。美国 2026 年已经在站点通电层面受限，核心瓶颈是变压器、开关柜、并网、变电站和可交付稳定兆瓦；中国全国发电缓冲更强，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 5, "neutral": 0, "value": 73.6, "events": [ { "title": "电网设备原材料缺口对新能源项目建设成本与并网进度的传导效应评估", "summary": "支持卡片06–08的逻辑链：AI电力瓶颈已上移至原材料，且GOES/高纯度铜缺口确实传导至新能源经济性，但传导非对称、主要走时间而非单价。直接价格通道有限——铜仅占光伏/陆上风电资本开支约3%~5%，即便铜较2020年+50%、GOES+100%，对LCOE直接抬升仅约+2%~+4% [自测算]。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供应缺口与电网扩容瓶颈对能源产业投资回报的约束力研究", "summary": "截至2026-06-25，本卡从能源台席支持第04/06/07/08卡的论证线，并加以收紧。结论：发电侧与电网侧扩容具备防止AI算力成本结构性大幅（失控）上行的潜力，但无法防止结构性温和台阶上移。关键在于把'供应'拆成两层：发电（电表来源）推进很快——2026年美国计划新增创纪录86 GW（约80…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器与开关柜产能缺口将实质性推迟新能源并网，延缓规管公用事业企业费率基数收益的现金兑现并恶化其资本效率，使发电商面临项目经济性破坏的重大下行风险。", "summary": "截至2026-06-25，本卡对电网瓶颈进行压力测试发现：变压器交期拉长至128-144周，中压开关柜拉长至52-78周，严重推迟新能源并网。受规管公用事业企业面临现金收益兑现延迟2-3年、采购成本暴涨60%-95%拖累资本效率等问题，面临电价审核核减和需求侧流失双重财务风险；敏感性分析表明，12…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "由于发电设备长交期、核电重建缓慢、跨州输电网架建设滞后以及并网容量在地理上的严重错配，电网侧能源增量无法匹配2026-2028年AI负载爆发式增长，电力缺口将实质性压制算力部署进度，…", "summary": "本卡对电网侧增量电源进行压力测试。2026-2028年，电网侧增量无法有效填补AI电力缺口。重型燃机面临5-7年超长交期；绿地核电停滞且SMR商业化迟于2030年，Palisades与三哩岛等重启仅能满足3.4%-10.8%的新增需求；输电线建设长达6.5-10年以上，且新增新能源地理错配严重（主…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 3, "neutral": 0, "value": 87, "events": [ { "title": "维持12% NAV清洁基础设施袖，但不要上调中枢；政策兑现支持电网/储能/运营商核心，136号文电价风险和制造产能过剩要求压低光伏/电池制造暴露。", "summary": "截至2026-06-26，本卡综合前序清洁基础设施杠铃：政策和需求数据支持保留12% NAV清洁袖，但不支持上调或追宽基清洁贝塔。国家电网2026-2030年4万亿元投资、全国电网接近5万亿元投资图景、1-5月用电5.7%增长、互联网数据服务用电40%+增长、算力用电到2030年8000亿千瓦时预…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网基础设施与能源转型对AI算力扩张的潜在约束", "summary": "压力测试部分支持并升级卡03：电力/电网瓶颈是“节流而非断路”，但节流比其暗示更紧更久。在2026–2028视野内，电气基础设施已是排在芯片之前的硬约束——美国电力变压器交期128周、GSU 144周、大单3–4年，价格较2019+77%、达2022前4–6倍；GOES由Cleveland-Cli…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 1, "neutral": 0, "value": 96, "events": [ { "title": "能源成本与高利率壁垒对全球算力中心电力建设与资本开支的传导效应", "summary": "支持Research note 1-2逻辑链并完成最后一公里传导。AI算力硬约束已从GPU芯片迁移至电力、再到电网可交付能力——击穿点不是电价而是无法通电[S1][S2]。两条通道：并网延迟（队列超2000GW、中位等待4-5年、落地率仅约13%、>95%为间歇性电源[S6][S1]）与电价刚性（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 2, "risks": 2, "neutral": 1, "value": 96, "events": [ { "title": "海上风电+UDC的24/7低碳电力合同与并网可行性", "summary": "可融资的24/7低碳UDC电力可实现，但只能作为四层结构：成型/固化的多年期PPA（叠加机制电价差价结算CfD）+ 偏LDES的多小时现场储能 + 有界偏差结算带 + 电网耦合并网（源网荷储或电网+逐时T-EAC）。否定纯海风直连孤岛——它移除了固化小时数所需的电网/储能平衡，24/7现金流不可投…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电力’，并驱动策略从算力规模向算法…", "summary": "截至2026-06-28，本卡片通过对电网与变压器交期的压力测试，确认电网并网（5-7年交期）和变压器交付（2-4年交期）构成了AI算力扩张的终极物理边界。随着GB300（约135kW）及Rubin Kyber（约600kW）对800VDC和固态变压器（SST）的高密度配电需求增加，目前仅有不足8…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "UDC的低PUE价值成立，但“海上风电直供”只有在叠加电网/储能/稳定PPA后才可支撑大规模AI算力的连续绿电需求。", "summary": "截至2026-06-29，本卡片对“海上风电+UDC”做压力测试：海上风电足以支撑24MW级模块的年度绿电匹配，但不能单独提供24/7稳定电力。UDC低PUE优势真实，PUE 1.10相对1.50可降低约26.7%电价敏感度；但0.10元/kWh到户电价变动仍会使24MW IT负荷模块年成本变化约…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "PQC 引发的周期性热降频会显著提高超大规模 AI 集群 PUE/WUE 与能源成本预测难度，并在集中接入区增加电力调度复杂度，但它是边际波动因子而非全国电力需求的主增长因子。", "summary": "2026-06-29 能源卡结论：支持前序观点，但限定机制。PQC 硬件化的直接功耗不是数据中心用电增长主因；真正问题在于 PQC 触发的周期性热降频、重协商突发和 GPU 空转会把稳定 IT 负荷变成区间波动，从而削弱年度 PUE/WUE 对边际成本的解释力，并在数据中心集中的节点增加电网调度和…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-29", "supports": 1, "risks": 2, "neutral": 0, "value": 96, "events": [ { "title": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-power约束（变压器160+周交期…", "summary": "对第09卡'电网灵活性成本压低发电侧收益'做买方侧压力测试：结论是有条件否认——对冲有效，但对冲错了轴。以1GW AIDC计，能源仅占TCO约7%、服务器占60%[S1]；辅助服务税发电侧均摊0.0053元/千瓦时、风光约0.011元/千瓦时[S7]，传导到买方约占TCO 0.14–0.25%[自…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "summary": "截至2026-06-29，本卡片从AI基础设施角度压力测试公用事业信用风险的传导。结论是：电网资本开支与信用压力不会终止AI建设，但会把送电日期从确定变量改为条件变量。1.295万亿美元公用事业capex、1500亿美元2026年IG发行、128周变压器交期和1.5-2年time-to-power…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链利润预期约-3.7%同比。", "summary": "支持第06卡\"并网延迟→利润压缩\"的传导论点，但将单向结论细化为分化。针对11GW数据中心建设缺口与12-18GW刚性局部电网缺口，48-60个月并网延迟通过两条通道传导：(1)经订单/PPA递延与尾款拖延压制发电子链开工率——5.7%销量递延经经营杠杆放大为约-21%利润压缩[自测算]；(2)将…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] } ], "href": "/investor/theses/thesis-4074d8a3d2/", "networkHref": "/investor/research-network/#thesis-4074d8a3d2" }, { "id": "thesis-aa97a5c1b3", "source": "tracked_thesis", "title": "家装链的胜负手在存量房翻新、国补触达与地产后周期库存去化", "titleEn": "Home improvement depends on existing-home renovation, subsidy reach, and post-property inventory digestion", "lane": "家装/地产后周期", "laneEn": "Home improvement and post-property cycle", "laneHref": "/investor/lanes/home-improvement-and-post-property-cycle/", "investableFocus": "家装/地产后周期：603833.SH, 603816.SH, 000333.SZ, 600690.SH, 002508.SZ, 002572.SZ, HD, LOW", "stockUniverse": [ { "ticker": "603833.SH", "name": "欧派家居", "market": "A-share", "role": "定制家居龙头", "roleEn": "定制家居龙头", "bucket": "core", "href": "/investor/stocks/603833.SH/" }, { "ticker": "603816.SH", "name": "顾家家居", "market": "A-share", "role": "软体家居", "roleEn": "软体家居", "bucket": "core", "href": "/investor/stocks/603816.SH/" }, { "ticker": "000333.SZ", "name": "美的集团", "market": "A-share", "role": "家电以旧换新", "roleEn": "家电以旧换新", "bucket": "core", "href": "/investor/stocks/000333.SZ/" }, { "ticker": "600690.SH", "name": "海尔智家", "market": "A-share", "role": "白电与渠道", "roleEn": "白电与渠道", "bucket": "core", "href": "/investor/stocks/600690.SH/" }, { "ticker": "002508.SZ", "name": "老板电器", "market": "A-share", "role": "厨电地产后周期", "roleEn": "厨电地产后周期", "bucket": "core", "href": "/investor/stocks/002508.SZ/" }, { "ticker": "002572.SZ", "name": "索菲亚", "market": "A-share", "role": "定制家居", "roleEn": "定制家居", "bucket": "core", "href": "/investor/stocks/002572.SZ/" }, { "ticker": "HD", "name": "Home Depot", "market": "US", "role": "美国家装需求对照", "roleEn": "美国家装需求对照", "bucket": "peer", "href": "/investor/stocks/HD/" }, { "ticker": "LOW", "name": "Lowe's", "market": "US", "role": "美国家装需求对照", "roleEn": "美国家装需求对照", "bucket": "peer", "href": "/investor/stocks/LOW/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" } ], "stockTickers": [ "603833.SH", "603816.SH", "000333.SZ", "600690.SH", "002508.SZ", "002572.SZ", "HD", "LOW", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE" ], "coreView": "赛道：家装/地产后周期。候选观察标的：603833.SH, 603816.SH, 000333.SZ, 600690.SH, 002508.SZ, 002572.SZ, HD, LOW。当前共有 40 条证据、1 位分析师和 6 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 69, "eventCount": 160, "firstSeen": "2026-05-04", "lastSeen": "2026-06-29", "supports": 4, "risks": 118, "neutral": 38, "analystCount": 16, "sourceTypeCounts": { "研究笔记": 33, "分析协作": 30, "研究线程": 96, "研究主题": 1 }, "topTerms": [ "Home", "improvement", "depends", "existing-home", "renovation", "subsidy", "reach", "post-property", "inventory", "digestion", "cycle", "Lane" ], "whatChanged": [ "2026-06-29: 产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "2026-06-29: 产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "2026-06-29: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "2026-06-29: 产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "2026-06-29: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）" ], "practical": { "sectorLane": "家装/地产后周期", "sectorLaneEn": "Home improvement and post-property cycle", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "probe", "actionLabel": "小仓验证", "actionTone": "yellow", "riskBudget": "验证仓：1-3% thesis risk", "portfolioRole": "验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "2-6 周", "portfolioUse": "适合用小仓位验证，不应在证据分歧扩大时追高。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "603833.SH", "name": "欧派家居", "market": "A-share", "role": "定制家居龙头", "roleEn": "定制家居龙头", "bucket": "core", "href": "/investor/stocks/603833.SH/" }, { "ticker": "603816.SH", "name": "顾家家居", "market": "A-share", "role": "软体家居", "roleEn": "软体家居", "bucket": "core", "href": "/investor/stocks/603816.SH/" }, { "ticker": "000333.SZ", "name": "美的集团", "market": "A-share", "role": "家电以旧换新", "roleEn": "家电以旧换新", "bucket": "core", "href": "/investor/stocks/000333.SZ/" }, { "ticker": "600690.SH", "name": "海尔智家", "market": "A-share", "role": "白电与渠道", "roleEn": "白电与渠道", "bucket": "core", "href": "/investor/stocks/600690.SH/" }, { "ticker": "002508.SZ", "name": "老板电器", "market": "A-share", "role": "厨电地产后周期", "roleEn": "厨电地产后周期", "bucket": "core", "href": "/investor/stocks/002508.SZ/" }, { "ticker": "002572.SZ", "name": "索菲亚", "market": "A-share", "role": "定制家居", "roleEn": "定制家居", "bucket": "core", "href": "/investor/stocks/002572.SZ/" }, { "ticker": "HD", "name": "Home Depot", "market": "US", "role": "美国家装需求对照", "roleEn": "美国家装需求对照", "bucket": "peer", "href": "/investor/stocks/HD/" }, { "ticker": "LOW", "name": "Lowe's", "market": "US", "role": "美国家装需求对照", "roleEn": "美国家装需求对照", "bucket": "peer", "href": "/investor/stocks/LOW/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" } ], "stockTickers": [ "603833.SH", "603816.SH", "000333.SZ", "600690.SH", "002508.SZ", "002572.SZ", "HD", "LOW", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE" ], "primaryStocks": [ { "ticker": "603833.SH", "name": "欧派家居", "market": "A-share", "role": "定制家居龙头", "roleEn": "定制家居龙头", "bucket": "core", "href": "/investor/stocks/603833.SH/" }, { "ticker": "603816.SH", "name": "顾家家居", "market": "A-share", "role": "软体家居", "roleEn": "软体家居", "bucket": "core", "href": "/investor/stocks/603816.SH/" }, { "ticker": "000333.SZ", "name": "美的集团", "market": "A-share", "role": "家电以旧换新", "roleEn": "家电以旧换新", "bucket": "core", "href": "/investor/stocks/000333.SZ/" }, { "ticker": "600690.SH", "name": "海尔智家", "market": "A-share", "role": "白电与渠道", "roleEn": "白电与渠道", "bucket": "core", "href": "/investor/stocks/600690.SH/" }, { "ticker": "002508.SZ", "name": "老板电器", "market": "A-share", "role": "厨电地产后周期", "roleEn": "厨电地产后周期", "bucket": "core", "href": "/investor/stocks/002508.SZ/" }, { "ticker": "002572.SZ", "name": "索菲亚", "market": "A-share", "role": "定制家居", "roleEn": "定制家居", "bucket": "core", "href": "/investor/stocks/002572.SZ/" } ], "instrumentTags": [ "603833.SH", "欧派家居", "603816.SH", "顾家家居", "000333.SZ", "美的集团", "600690.SH", "海尔智家", "002508.SZ", "老板电器", "002572.SZ", "索菲亚" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 4 类，近期事件 119 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 4, "analystCount": 16, "recentEvents": 119, "supportRatio": 0.03, "riskRatio": 0.74, "conflictScore": 85, "evidenceDepth": 100 }, "score": 63.8 }, "events": [ { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 17.5, "matchedTerms": [ "evidence", "risk", "dominant", "inflation", "transmission", "miss", "price", "core" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-28", "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.en.md (7867 bytes) — The repor…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 11, "matchedTerms": [ "inflation", "investment", "price", "A-share" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 10.75, "matchedTerms": [ "Candidate", "thesis", "analysts", "risk" ], "href": "" } ], "timeline": [ { "date": "2026-05-04", "supports": 0, "risks": 0, "neutral": 1, "value": 45, "events": [ { "title": "\"session id\": \"a5c839ce 1820 42a0 b8c5 cb453873bd99\", \"response\": \"The Investment Committee…", "summary": "\"session id\": \"a5c839ce 1820 42a0 b8c5 cb453873bd99\", \"response\": \"The Investment Committee has officially synthesized the research from the Global M…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 0, "value": 37.4, "events": [ { "title": "[auto-handoff] 主权算力财政支出影响", "summary": "致 TMT行业分析师： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 1f2ac627-aa0e-4991-81c1-4c46ae106c06 结果摘要: Wrote handoff_response.md (17,200 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 4, "neutral": 0, "value": 20.8, "events": [ { "title": "[auto-handoff] Summit De-risking", "summary": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 502dfe56-3d6c-4017-933c-df9f9b15f88b 结果摘要: Wrote handoff_response.md (8242 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] USD/CNY 路径与 273bps 中美利差的可持续性", "summary": "致 债券分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: e921623d-4f97-4b98-8ef8-0deedb115778 结果摘要: Wrote handoff_response.md (10060 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] A-share New Productive Forces theme: crowding and policy-support boundary", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 17939c81-1aba-4588-815a-9997769cc0e7 结果摘要: Wrote handoff_response.md (9758 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Inflation Transmission from U.S.-Iran Conflict", "summary": "致 市场情绪师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 3319aae8-eed8-4aac-bf93-59603f640ad9 结果摘要: Wrote handoff_response.md (7027 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 0, "neutral": 1, "value": 49.2, "events": [ { "title": "[auto-handoff] Retail demand or margin shock", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 消费行业分析师。 模式: adhoc 会话: cf956797-9487-48b7-bee9-93f727fa6ba3 结果摘要: Wrote handoff_response.md (3099 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 0, "neutral": 1, "value": 50.6, "events": [ { "title": "As the Research Quality Manager (QA Manager), I have performed a comprehensive quality audi…", "summary": "As the Research Quality Manager (QA Manager), I have performed a comprehensive quality audit and logical synthesis of the entire 8 card research pipe…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "[auto-handoff] 美联储Kevin Warsh接掌期与全球流动性二次紧缩传导", "summary": "致 市场情绪师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 95f9de7b-813d-4b05-be7b-27bd074e8d2d 结果摘要: Wrote handoff_response.md (9664 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 1, "neutral": 7, "value": 73.4, "events": [ { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "As the Global Macro Analyst , I have conducted a comprehensive synthesis of the macro finan…", "summary": "As the Global Macro Analyst , I have conducted a comprehensive synthesis of the macro financial transmission channels of G SIB USD balance sheet rati…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Re-estimate supply chain price shock transmission to U.S. core PCE", "summary": "Re-estimate supply chain price shock transmission to U.S. core PCE In 'Thesis 1: Step 4', direct a high-priority handoff to global-macro to re-estima…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Summary of Completed Analysis We have completed the research card for Research note 09/10: …", "summary": "Summary of Completed Analysis We have completed the research card for Research note 09/10: Assessing the Efficacy of Chinese Banks' Cross Border RMB …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "We have successfully completed Research note 03/08 of the research thread session efecefe6 …", "summary": "We have successfully completed Research note 03/08 of the research thread session efecefe6 2a06 4881 87e9 ccec1f502077 as the Chief Economist. Stance…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 2, "value": 53.2, "events": [ { "title": "I have completed the global macro analysis for Research note 01 of the research thread sess…", "summary": "I have completed the global macro analysis for Research note 01 of the research thread session b0a5df37 4aba 4443 9696 a334f45b9dd4 . Key Findings & …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] DeepSeek V4-Pro 降价对阿里云与腾讯云 MaaS 利润率及 Ascend 950 适配影响", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: f4490f3f-5fe4-4338-9931-6d743b5f5c29 结果摘要: Wrote handoff_response.md (14919 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have successfully researched the transmission of downstream catering and group catering d…", "summary": "I have successfully researched the transmission of downstream catering and group catering demand to upstream protein raw materials and agricultural e…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 1, "neutral": 1, "value": 51.6, "events": [ { "title": "I have completed the analysis for Research note 4/10 as the Materials Analyst, evaluating t…", "summary": "I have completed the analysis for Research note 4/10 as the Materials Analyst, evaluating the cost transmission channels and gross margin squeeze ris…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 真实汇率6.80下PBOC降息空间及科创板防守重评估", "summary": "致 投委会主席： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 6831073e-7348-476b-8c3d-ed1164d7efb5 结果摘要: Wrote handoff_response.md (8046 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 0, "neutral": 1, "value": 59, "events": [ { "title": "[auto-handoff] PCE × 低 VIX × 高集中度的非线性 scenario card", "summary": "致 日报总编： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 98579803-3357-4f36-b693-32bec78b7f03 结果摘要: Wrote handoff_response.md (12113 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 2, "neutral": 1, "value": 48.4, "events": [ { "title": "[auto-handoff] A股/港股地产板块在'政策底已坐实、Q3信用尾部风险升高'组合下的配置权重", "summary": "致 房地产分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: ab58fcee-419c-4e1f-9889-fe797d3a313b 结果摘要: Wrote handoff_response.md (11503 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the macroeconomic research and policy evaluation for Research note 06 of t…", "summary": "I have completed the macroeconomic research and policy evaluation for Research note 06 of the shared research thread session. Summary of Research & F…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 美股 5/29 风险偏好回升 → A 股科技 + 可选消费的可交易性", "summary": "致 首席策略师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 958210b2-28f1-4e8b-8887-e9276abb6c4d 结果摘要: Wrote handoff_response.md (11398 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 0, "neutral": 1, "value": 61.8, "events": [ { "title": "I have completed the analysis for Research note 01/08 as the Global Macro Analyst , examini…", "summary": "I have completed the analysis for Research note 01/08 as the Global Macro Analyst , examining the unseasonal surge in global freight rates and its no…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 0, "neutral": 1, "value": 63.2, "events": [ { "title": "A Share Asset Allocation Strategy Under Low Inflation & Property Wealth Traps (Research not…", "summary": "A Share Asset Allocation Strategy Under Low Inflation & Property Wealth Traps (Research note 9/10) As the Chief Strategist (首席策略师) , I have synthesiz…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 0, "neutral": 1, "value": 64.6, "events": [ { "title": "[auto-handoff] 中美利差倒挂及 Swap Points 对外资跨境套息平仓的临界值测算", "summary": "致 市场情绪师： 协同任务已完成。执行方为 外汇策略师。 模式: adhoc 会话: 72accea4-0ff7-4ce8-9eb6-df7388bb5b58 结果摘要: Wrote handoff_response.md (17899 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 1, "value": 60, "events": [ { "title": "We have completed the research for Research note 06, analyzing the transmission channel thr…", "summary": "We have completed the research for Research note 06, analyzing the transmission channel through which extreme 0DTE (Zero Days to Expiration) option i…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 国内消费“量增价缩”与渠道平替趋势对宏观增值税及地方财政传导测算", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 12e797f3-2934-48f3-9aa6-658873753f7b 结果摘要: Wrote handoff_response.md (14719 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 0, "neutral": 1, "value": 67.4, "events": [ { "title": "I have finalized the H2 2026 stress testing and synthesis analysis on the transmission of U…", "summary": "I have finalized the H2 2026 stress testing and synthesis analysis on the transmission of US AI hardware tariff inflation to the A share export compu…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 1, "value": 62.8, "events": [ { "title": "二手房与翻新无法填补竣工悬崖，重估地产后周期消费需求弹性", "summary": "二手房与翻新无法填补竣工悬崖，重估地产后周期消费需求弹性 Secondary Sales and Renovations Fail to Plug the Completion Cliff: Re-evaluating Post-Property Demand 二手房回暖与存量翻新是否真能对冲新房…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "As the China Macro Analyst (china macro), I have completed the stress test report for Resea…", "summary": "As the China Macro Analyst (china macro), I have completed the stress test report for Research note 8 of 8, assessing the boundaries of China's fisca…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 0, "neutral": 1, "value": 70.2, "events": [ { "title": "I have completed the stress test analysis of China's State Grid 2026 investment plan and it…", "summary": "I have completed the stress test analysis of China's State Grid 2026 investment plan and its impact on copper demand and the systemic drawdown floor …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-07", "supports": 1, "risks": 0, "neutral": 4, "value": 88.6, "events": [ { "title": "[auto-handoff] 铜价高企背景下家电龙头企业与二三线品牌成本裂谷对工业品PPI与制造业投资的传导评估", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 536b5150-44d9-4c3b-b53a-4f495cbccec3 结果摘要: Wrote handoff_response.md (11657 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Executive Summary of Analysis (Energy Analyst Perspective) In this stress test card, we ana…", "summary": "Executive Summary of Analysis (Energy Analyst Perspective) In this stress test card, we analyzed the macroeconomic transmission mechanism and Federal…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 铜价高企背景下家电龙头企业与二三线品牌成本裂谷对工业品PPI与制造业投资的传导评估", "summary": "致 首席经济学家： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: e00d3d45-0543-4a90-a4a3-8afa35e2cc1e 结果摘要: Wrote handoff_response.md (12590 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] Reassessing Defensive Value in A-Share Steel Stocks Under Margin Squeeze and…", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: aa8a47d9-71f3-4319-9906-28eb96ae0a87 结果摘要: Wrote handoff_response.md (2858 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 0, "neutral": 1, "value": 73, "events": [ { "title": "I have completed the analysis for Research note 07 as the TMT Analyst, supporting the thesi…", "summary": "I have completed the analysis for Research note 07 as the TMT Analyst, supporting the thesis that low price platform subsidy strategies structurally …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 0, "neutral": 1, "value": 74.4, "events": [ { "title": "[auto-handoff] 慢去产能周期下CPI食品项及H2通胀中枢敏感性评估", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 21e65805-b96d-4e1e-8071-581e93f19615 结果摘要: I will run the command 系统日期锚 to anchor today's date as requ…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 4, "neutral": 0, "value": 48.8, "events": [ { "title": "I have completed the stress test analysis for Research note 6 of the research thread sessio…", "summary": "I have completed the stress test analysis for Research note 6 of the research thread session under the authoritative work date of 2026 06 12 . Key An…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：海工装备产业链（30 日激活率 5.4%, 6/112）", "summary": "产业链调研：海工装备产业链（30 日激活率 5.4%, 6/112） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 10.3%, 10/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 10.3%, 10/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 4.7%, 8/170）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 4.7%, 8/170） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 0, "value": 50.2, "events": [ { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 6.5%, 13/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 6.5%, 13/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 6, "neutral": 0, "value": 39.6, "events": [ { "title": "I have performed a comprehensive stress test on the ability of property completions and sec…", "summary": "I have performed a comprehensive stress test on the ability of property completions and secondary home renovations to offset the cost constraints for…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 7, "neutral": 1, "value": 38, "events": [ { "title": "The research into China's power system and green trading reforms at the start of the \"15th …", "summary": "The research into China's power system and green trading reforms at the start of the \"15th Five Year Plan\" (2026) is complete. The analysis confirms …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 8, "neutral": 1, "value": 33.4, "events": [ { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 3, "neutral": 1, "value": 64.8, "events": [ { "title": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 11.7%, 24/206） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 美股防御+商品轮动对今日 A 股板块映射", "summary": "致 首席策略师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 4aebbb64-711d-4eb4-9f9d-ca1e95ced548 结果摘要: Wrote handoff_response.md (13373 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 9, "neutral": 0, "value": 27.2, "events": [ { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 6.6%, 16/242） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 9, "neutral": 0, "value": 28.6, "events": [ { "title": "I have completed the stress test report for Research note 8/8 of the session. The report sy…", "summary": "I have completed the stress test report for Research note 8/8 of the session. The report systematically maps out the transmission of an offshore toke…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 6, "neutral": 0, "value": 48, "events": [ { "title": "[auto-handoff] H2 2026 窄口径算力硬件 Capex 下修对 A股 CPO/服务器链条 of 订单检验", "summary": "致 深研主编： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: f64d64a8-9a7b-4bec-b4e6-2745c5763369 结果摘要: Wrote handoff_response.md (8870 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：建筑材料产业链（30 日激活率 8.3%, 13/157）", "summary": "产业链调研：建筑材料产业链（30 日激活率 8.3%, 13/157） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 9, "neutral": 2, "value": 37.4, "events": [ { "title": "I have completed the synthesis of the trade surplus breakdown trigger and its implications …", "summary": "I have completed the synthesis of the trade surplus breakdown trigger and its implications for the global macro liquidity transmission chain. Summary…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have successfully completed the stress test analysis on the transmission channels of the …", "summary": "I have successfully completed the stress test analysis on the transmission channels of the global dollar liquidity squeeze and the JPY carry trade un…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 6, "neutral": 0, "value": 50.8, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 5, "neutral": 0, "value": 58.2, "events": [ { "title": "[auto-handoff] 云厂商 H2 算力 Capex 下修对全球硬科技周期的传导及美股科技股久期风险评估", "summary": "致 TMT行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: fa034dd1-86e7-4edb-9591-32ba8cef0780 结果摘要: Wrote handoff_response.md (14214 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：公用事业产业链（30 日激活率 12.6%, 26/206）", "summary": "产业链调研：公用事业产业链（30 日激活率 12.6%, 26/206） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 2, "risks": 7, "neutral": 3, "value": 72.6, "events": [ { "title": "I have conducted a comprehensive consumer industry analysis and stress tested the price tra…", "summary": "I have conducted a comprehensive consumer industry analysis and stress tested the price transmission dynamics in the B2C home improvement retail mark…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] 十五五电网4万亿投资计划对国内工业产出及PPI的拉动效应", "summary": "致 深研主编： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 962dd3bc-dd6b-46d1-bdbc-e2f725f549fb 结果摘要: Wrote handoff_response.md (11728 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "[auto-handoff] A-share semiconductor valuation reset sensitivity under global carry trade u…", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 38bcbac6-c7bd-47b7-b7e8-1ba28c2790af 结果摘要: Wrote handoff_response.md (9870 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply s…", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 0, "neutral": 1, "value": 94, "events": [ { "title": "Summary of Work I have completed the research and stress test on the transmission effects o…", "summary": "Summary of Work I have completed the research and stress test on the transmission effects of grid equipment bottlenecks on upstream key raw materials…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 6, "neutral": 0, "value": 56.4, "events": [ { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "summary": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 0, "value": 75.8, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 4, "neutral": 1, "value": 74.2, "events": [ { "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "summary": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 8, "neutral": 0, "value": 48.6, "events": [ { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "summary": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-aa97a5c1b3/", "networkHref": "/investor/research-network/#thesis-aa97a5c1b3" }, { "id": "thesis-c1d3acab38", "source": "auto_topic", "title": "大健康/生物科技：支付、监管与效率改善验证", "titleEn": "Practical research topic: 大健康/生物科技：支付、监管与效率改善验证", "lane": "消费健康防御", "laneEn": "消费健康防御", "laneHref": "/investor/lanes/lane-d104d465/", "investableFocus": "", "stockUniverse": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "ARM", "name": "Arm Holdings", "market": "US", "role": "端侧 CPU IP", "roleEn": "端侧 CPU IP", "bucket": "core", "href": "/investor/stocks/ARM/" } ], "stockTickers": [ "600519.SH", "300760.SZ", "600276.SH", "3690.HK", "001215.SZ", "1179.HK", "9961.HK", "588000.SH", "APO", "IEF", "META", "TLT", "MAR", "ARM" ], "coreView": "该主线来自 35 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-17 至 2026-06-29；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 69, "eventCount": 85, "firstSeen": "2026-05-17", "lastSeen": "2026-06-29", "supports": 8, "risks": 70, "neutral": 7, "analystCount": 11, "sourceTypeCounts": { "研究笔记": 43, "研究线程": 36, "深度研究": 6 }, "topTerms": [ "大健康产业链", "研究活跃度与投资线索复盘" ], "whatChanged": [ "2026-06-29: 产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "2026-06-28: 产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "2026-06-26: 产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "2026-06-26: 产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "2026-06-26: 大健康产业链处于‘被忽视定价’阶段，基本面反转已现，应围绕出海BD与高端国产替代定向深挖五只低覆盖中盘股，防范地缘立法与科创去杠杆等跨链外溢。" ], "practical": { "sectorLane": "消费健康防御", "sectorLaneEn": "消费健康防御", "exposureBucket": "消费健康防御", "tone": "green", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "ARM", "name": "Arm Holdings", "market": "US", "role": "端侧 CPU IP", "roleEn": "端侧 CPU IP", "bucket": "core", "href": "/investor/stocks/ARM/" } ], "stockTickers": [ "600519.SH", "300760.SZ", "600276.SH", "3690.HK", "001215.SZ", "1179.HK", "9961.HK", "588000.SH", "APO", "IEF", "META", "TLT", "MAR", "ARM" ], "primaryStocks": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" } ], "instrumentTags": [ "600519.SH", "贵州茅台", "300760.SZ", "迈瑞医疗", "600276.SH", "恒瑞医药", "3690.HK", "美团", "001215.SZ", "千味央厨", "1179.HK", "华住集团-S" ], "reviewCadence": "每日复核", "confirmationSignals": [ "价格弹性是否转化为利润率修复", "支付端和监管是否支持长期现金流", "防御属性是否被估值提前透支" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "估值拥挤", "支付监管", "渠道库存", "CPI 食品项" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 100，来源 3 类，近期事件 45 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 85, "sourceDiversity": 3, "analystCount": 11, "recentEvents": 45, "supportRatio": 0.09, "riskRatio": 0.82, "conflictScore": 91.5, "evidenceDepth": 100 }, "score": 62.5 }, "events": [ { "date": "2026-06-29", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 9.5, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 9.5, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-26", "title": "大健康产业链处于‘被忽视定价’阶段，基本面反转已现，应围绕出海BD与高端国产替代定向深挖五只低覆盖中盘股，防范地缘立法与科创去杠杆等跨链外溢。", "summary": "本卡片对低覆盖的大健康产业链进行了深度分析。在30日激活率仅5.7%的被忽视背景下，基本面于2026上半年迎来License-out、CXO订单与设备更新三重引擎的拐点。定向推荐药明合联、荣昌生物、康诺亚、医渡科技、沛嘉医疗五只低覆盖中盘创新龙头，并系统梳理了BIOSECURE立法、美债高利率去杠杆、DRG设备贴息及上游BOM与碳边境四类跨链条溢出风险。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-26", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 9.5, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-26", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 9.5, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-23", "title": "截至2026年6月24日，大健康产业链在低估值与出海催化下步入业绩驱动重估期，应重点布局具备海外本土化渠道与跨链条多元化能力的高壁垒医疗器械及仿制药注射剂/多肽出口龙头。", "summary": "截至2026年6月24日，大健康产业链严重研究不足（激活率5.7%）。本报告筛选出四家估值极具吸引力的细分龙头：健友股份（仿制注射剂出海与Q1拐点）、海泰新光（内窥镜光学壁垒及AI器件跨界）、微创脑科学（神经介入独立高盈利）和圣诺生物（GLP-1多肽产能大增）。提示关税、多肽供应链、AI硬件及集团债务等溢出风险。交由biotech-analyst跟踪多肽C…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-21", "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "大健康产业链低覆盖（30日激活5.4%），但基本面已反转。三引擎齐发：创新药license-out成全球定价事件（2025年约1357亿美元、占全球49%，2026Q1超600亿美元）；CXO订单重新加速（药明Q1营收+28.8%、在手订单597.7亿；凯莱英新兴业务+74.1%）；设备更新（2025H1招标+62.75%）叠加第十一批集采新规重定价器械与…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10.25, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-21", "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 9.5, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-21", "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 9.5, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-20", "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "大健康产业链 30 日激活率仅 5.4%（15/280），属被忽视而非基本面差。2026 年 Q1 创新药对外授权破 600 亿美元（约占 2025 全年 1357 亿的 44%）[自测算]，药明康德 Q1 营收 124.36 亿（+28.81%）、在手订单 597.7 亿、小分子 CDMO+80.1%，CXO 订单周期拐头；政府工作报告首次将生物医药列为…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-18", "title": "大健康产业链 — 低覆盖标的与跨链外溢风险 卡片 01/10 · 白板 5a71b89a · 医药行业分析师 · 工作日期：2026 06 18（亚洲/新加坡） 立场：initial — 开…", "summary": "大健康产业链 — 低覆盖标的与跨链外溢风险 卡片 01/10 · 白板 5a71b89a · 医药行业分析师 · 工作日期：2026 06 18（亚洲/新加坡） 立场：initial — 开启本线索 1. 这条产业链是\"被忽视定价\"，而非\"基本面定价\" 大健康产业链的 30 日激活率仅 5.4%（15/280） ——在本白板上属于结构性低覆盖。但 2026…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 9.75, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-18", "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 9.5, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-18", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 9.5, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-17", "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 9.5, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-17", "title": "大健康可投方向已从拥挤龙头转向政策压力下仍能证明现金流、合规商业化或出海能力的低覆盖细分平台。", "summary": "截至2026-06-18，本卡支持提高大健康产业链研究权重。政策上，DRG/DIP和第11批VBP继续压低低壁垒院内品种溢价；更值得深挖的是自费/院外需求、具备临床成本节约的器械、生命科学工具、肿瘤服务现金流修复和医疗AI盈利验证。优先队列为LifeTech Scientific、Giant Biogene、MGI Tech、Hygeia Healthca…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 9.25, "matchedTerms": [ "大健康产业链" ], "href": "" }, { "date": "2026-06-16", "title": "最有研究价值的大健康低覆盖标的，是能用 2025 年收入、订单、医保准入或渠道数据验证拐点的公司，而不是泛泛补覆盖整个医药板块。", "summary": "2026-06-15 初始筛选支持提升大健康产业链覆盖度，但优先队列应集中在 2025 年已出现商业化拐点或明确风险的标的：WuXi XDC (2268.HK)、迪哲医药 (688192.SH)、Ascentage Pharma (6855.HK / AAPG)、MicroPort NeuroScientific (2172.HK)、Giant Bioge…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 10, "matchedTerms": [ "大健康产业链" ], "href": "" } ], "timeline": [ { "date": "2026-05-17", "supports": 0, "risks": 2, "neutral": 0, "value": 30, "events": [ { "title": "大健康产业链 1.5% 的激活率严重低估了三重顺风（创新药政策窗口、集采疲劳拐点、CXO 订单拐点）下的命题密度，5 只低覆盖标的中至少 3 只在两季度内具备离散催化。", "summary": "大健康产业链 30 日激活率仅 1.5%（4/275），结构性低覆盖。三条尚未被共识充分定价的顺风正在叠加：2026-Q1 NMPA/CDE 创新药指引与 BD-out 期权激励、第十批集采范围收窄带来的 VBP 疲劳拐点、以及全球 Biotech 融资修复滞后传导的 CXO 订单簿拐点。在此基础…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 1.5%, 4/275）", "summary": "产业链调研：大健康产业链（30 日激活率 1.5%, 4/275） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 1, "risks": 4, "neutral": 0, "value": 27.4, "events": [ { "title": "2026-2030年大健康内需的可持续性由长护险制度化、十五五目标和人口结构转型支撑，而非单一财政脉冲，但受居民财富效应制约。", "summary": "2026年中国大健康产业链正从政策脉冲转向结构性内需。3.23亿老年人口驱动的银发经济预计在2030年达到16万亿元规模。虽然2026年设备更新受益于1.3万亿特别国债，但可持续性更依赖于2028年全国推开的长护险。压力测试显示，地产财富效应减弱和地方财政差异是主要制约因素。建议关注康复、慢病管理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康产业链的方向成立，但2026年不是无差别重估年份，而是验证制造能力、医院预算、支付纳入和核心部件国产化的年份。", "summary": "截至2026-05-18，本卡否定上一卡对大健康产业链2026年广义重估的判断。需求与政策方向存在，但GLP-1专利到期不等于当年供给爆发，AD全国报销和筛查数据缺乏权威证据，90%高端器械国产化目标未被来源验证。工业视角应转向设备更新、医院数字化、核心部件、售后服务、审批节奏和支付纳入。结论是：…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 1.5%, 4/275）", "summary": "产业链调研：大健康产业链（30 日激活率 1.5%, 4/275） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国电力设备出海：AI 电网需求、贸易壁垒与利润率分化", "summary": "全球 AI 电网投资为中国电力设备提供需求窗口，但关税、认证、交付能力和原材料成本会决定哪些企业真正把订单转化为利润。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/china-power-equipment-export-policy-8678db7f/" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 3, "neutral": 0, "value": 34.8, "events": [ { "title": "基于大健康产业链中不同资产的确定性与弹性分层，2026年组合应以康诺亚和新产业为防御性基本盘（60%），以翰宇药业和翔宇医疗博取政策与订单爆发的Beta收益（40%）。", "summary": "截至2026-05-19，本报告综合此前调研结论，建议对2026年大健康板块采取“H股压舱、A股突击”的哑铃型组合策略。将商业化确定性高的康诺亚（2162.HK，35%）和抗周期的IVD出海标的新产业（300832.SZ，25%）作为核心底仓；将高弹性但履约风险大的翰宇药业（300199.SZ，2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在医保控费和地缘博弈常态化下，大健康产业的超额收益正从“全面创新”转向“细分赛道出海（IVD/原料药）”与“未满足的刚需（自免/老龄化康复）”。", "summary": "大健康产业链近期研究活跃度偏低（激活率1.8%），但蕴藏结构性机会。本报告挖掘了4个具备强逻辑且覆盖度不足的标的：翰宇药业（300199.SZ，GLP-1多肽原料药）、新产业（300832.SZ，IVD出海）、翔宇医疗（688626.SH，康复设备更新）和康诺亚（2162.HK，自免生物药）。建议…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 1.8%, 5/275）", "summary": "产业链调研：大健康产业链（30 日激活率 1.8%, 5/275） 8%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "消费降级、银发经济与大健康结构性机会", "summary": "消费链条正在从总量弹性转向结构分化：银发消费、医疗健康、糖替代和 B 端餐饮渗透可能形成防御性需求，但需要验证支付能力和渠道真实动销。", "sourceLabel": "深度研究", "supportLabel": "支持", "supportType": "supports", "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" } ] }, { "date": "2026-05-21", "supports": 1, "risks": 2, "neutral": 1, "value": 45.2, "events": [ { "title": "作为 首席策略师 (Chief Strategist) ，我已对大健康产业链白板 session 的前 9 张卡片进行了终极的系统性综合，将“有成本的关税保险”（德州模式）重新定义并…", "summary": "作为 首席策略师 (Chief Strategist) ，我已对大健康产业链白板 session 的前 9 张卡片进行了终极的系统性综合，将“有成本的关税保险”（德州模式）重新定义并转化为具体的资产配置权重、基于 2026 05 30 调��关键节点的决策闸门（Event Gates）以及个股与组…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "FDA临床/网络安全收紧、NHSA手术机器人分档+封顶、Section 301关税阶梯三栈在2026共振，使大健康产业链上下游成本转嫁机制断链，出海毛利无法对冲国内压制，整体板块需结…", "summary": "本卡作为政策分析师从监管/支付/贸易三栈共振视角支持C2结论。NHSA 2026-01-20《手术机器人收费立项指南》设导航/参与/精准三档系数化收费并叠加托底/封顶[S1]，与DRG/DIP 2.0 2025-01-26全国统一切换、3.0 2026年内发布[S6]构成国内ASP天花板；FDA …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康链条的机会在于医保扩容、出海和利用率修复三条线的精选个股，而不是整体性估值修复。", "summary": "截至2026-05-21，本卡支持把大健康产业链纳入精选研究队列，但不建议做宽基行业贝塔。需求由老龄化和门诊利用率支撑，支付端由医保和商保创新药目录提供结构性机会，但集采、DRG/DIP和医保基金监管继续压制低证据、低差异化环节。建议优先深挖迪哲医药(688192.SH)、微创机器人(2252.H…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275）", "summary": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 6, "neutral": 0, "value": 12, "events": [ { "title": "经对Research note 01-09完整大健康产业链多步研究链的数值一致性、叙事连贯性与引用覆盖率进行最终QA审计，本中心确认指标勾稽完全闭合，逻辑链条严密，给予最终PASS评…", "summary": "本卡作为产业链调研的最终QA审计，于2026-05-23对Research note 01-09的完整大健康研究链进行发刊前审定。经核对，子组合权重闭合100%，加权Beta从子组合0.84降至大类资产0.79，VaR从5.04%优化至4.21%，+2.5%~+3.5%的超配区间通过 barbel…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘政策阻力加速生物医药路径分化，具备多肽API制造壁垒的“拆分BD”重构与“数据隔离双轨”的中游临床CRO正迎来显著的估值修复机会。", "summary": "在BIOSECURE法案等政策压测下，大健康产业链面临分化。本卡支持Research note 04的下行风险警告，但针对具备多肽API自主生产能力的创新药企（如博瑞医药688166.SH）和中游低美国暴露的临床CRO（如泰格医药300347.SZ、博济医药300404.SZ），提出了“估值与路径…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康低覆盖不是全面行业超配信号，而是应先建立药明合联、荣昌生物、沛嘉医疗、爱康医疗、医渡科技五个定向深挖队列。", "summary": "截至2026-05-22，本卡支持把大健康产业链列为低覆盖研究方向：过去30日仅7/275激活、激活率2.5%。优先深挖药明合联、荣昌生物、沛嘉医疗、爱康医疗和医渡科技。核心判断是医保支付纪律仍强，但创新药、ADC/XDC外包、结构性心脏病/骨科器械、医疗AI数据层均有可量化经营拐点。主要外溢风险…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275）", "summary": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 5, "neutral": 0, "value": 19, "events": [ { "title": "在对大健康产业链中以 佰仁医疗（688198.SH） 和 艾力斯（688578.SH） 为核心的科创板成长龙头进行了微观两融平仓敏感度与流动性挤兑压力测试后，我们确认了以佰仁医疗为代…", "summary": "在对大健康产业链中以 佰仁医疗（688198.SH） 和 艾力斯（688578.SH） 为核心的科创板成长龙头进行了微观两融平仓敏感度与流动性挤兑压力测试后，我们确认了以佰仁医疗为代表的高拥挤两融股（融资余额占流通市值高达7.24% [S3]，消化��数达5.33天 [S3]）在面临基本面预期修正…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康板块正经历由概念向业绩兑现的深刻重塑，以佰仁医疗为代表的高红利高壁垒隐形冠军和艾力斯为代表的高现金流创新药龙头，在A股风格轮动中具备极强的估值修复与配置价值。", "summary": "本报告针对大健康产业链（激活率2.5%）进行深度梳理，筛选出四家处于市场研究盲区且业绩爆发力强的细分龙头：海思科（一季度净利增1089.85%，BD转让高收益）、佰仁医疗（一季度扣非增73.64%，56.93%高分红）、艾力斯（伏美替尼放量，Q1扣非增59.86%，毛利率94.91%）及南微医学（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275）", "summary": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "医药不是泛化反弹交易，而是应围绕“创新变现 + 高端器械国产替代 + 集采后修复”建立精选研究篮子。", "summary": "截至2026-05-25，本卡支持启动大健康产业链专项研究队列。优先深挖 2252.HK、9996.HK、09969.HK/688428.SH、002821.SZ/6821.HK、300529.SZ。主线是创新药出海授权、CXO 新模态订单、国产高端器械替代与集采后企稳。核心风险是外科器械 VBP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275）", "summary": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 0, "neutral": 1, "value": 56.2, "events": [ { "title": "I have successfully completed the research and authored the mandatory deliverables for Rese…", "summary": "I have successfully completed the research and authored the mandatory deliverables for Research note 01/10 of the Big Health Industry Chain (大健康产业链) …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 2, "neutral": 0, "value": 42.6, "events": [ { "title": "大健康产业链不缺需求，真正值得深挖的是能用全球化、差异化临床价值或制造杠杆穿越医保控费的公司。", "summary": "截至工作日期 2026-05-27，本卡从零建立大健康产业链初筛。结论是需求由老龄化和 95% 医保覆盖托底，但投资机会应聚焦能摆脱单纯降价压力的环节：全球授权、差异化临床数据、ADC/生物偶联制造和海外器械渠道。优先入队标的为 Kelun-Biotech 6990.HK、Mabwell 6880…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275）", "summary": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 1, "risks": 4, "neutral": 0, "value": 40, "events": [ { "title": "\"session id\": \"2ae742c0 c3a0 49f0 a11d b180016b2890\", \"response\": \"针对“宽货币、弱信用”及信用极化背景下的大健康产…", "summary": "\"session id\": \"2ae742c0 c3a0 49f0 a11d b180016b2890\", \"response\": \"针对“宽货币、弱信用”及信用极化背景下的大健康产业链 Alpha 挖掘，本报告得出以下结论：在信用传导受阻的环境中，具备 “全球化现金流自偿” 、 “颠覆性技术替代…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "针对大健康产业链4+1名单进行政策压力测试，表明《生物安全法案》已立法落地并转向渐进合规审计，凯莱英的多肽CDMO产能利用率与荣昌生物的BD合规成本正承受地缘脱钩的实质摩擦，而第12…", "summary": "本报告对大健康产业链中荣昌生物、凯莱英、联影医疗和微电生理4家重点企业及观察标的翰宇药业进行政策压力测试。分析指出，美国《生物安全法案》在2025年12月18日已正式签署成为法律，正处于逐步实施期，其取消了固定截止日期，改为FAR修订后5年滚动豁免，虽未指名药明等，但长线脱钩风险加剧，直接威胁凯莱…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275）", "summary": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275）", "summary": "产业链调研：大健康产业链（30 日激活率 2.5%, 7/275） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 2, "neutral": 1, "value": 48.4, "events": [ { "title": "大健康产业链细分赛道瓶颈深度解构：医疗器械与创新药的执行缺口分析", "summary": "把会话级\"30日激活率2.9%\"映射到医药产业链，最逼近的现实指标是国谈药首年医院进院率(~15%–25%)。准入后真实落地需穿越六道闸门：进院率(G2)、DRG/DIP点数下医生处方意愿(G3)、双通道配齐率(G4)、集采品种放量与临床惯性错配(G5)、商保支付缺位(G6)。以G2×G3×G4中…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "战略审计：大健康产业链“激活危机”及其研究框架核定 日期: 2026 06 01 分析师: 首席策略师 (A股策略、风格轮动、行业配置) 研究线程标识：已归档| 卡片: 03/10 …", "summary": "战略审计：大健康产业链“激活危机”及其研究框架核定 日期: 2026 06 01 分析师: 首席策略师 (A股策略、风格轮动、行业配置) 研究线程标识：已归档| 卡片: 03/10 1. 执行摘要 在医药行业分析师与内部运营官完成基础架构重建（Research note 01 02）后，本审计确认…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 2.9%, 8/275）", "summary": "产业链调研：大健康产业链（30 日激活率 2.9%, 8/275） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 1, "risks": 2, "neutral": 0, "value": 54.8, "events": [ { "title": "大健康产业链的低覆盖+高催化窗口集中在 ADC 平台、创新药对外授权、高端医疗设备国产替代、GLP-1 二线产业链四条主线的中盘标的，五只优先标的为科伦博泰、荣昌、药明合联、联影、华…", "summary": "大健康产业链 30 日激活率 2.9% 的冷区中，真正的预期差不在创新药龙头本身，而在创新药 License-out、ADC/双抗平台、高端医疗设备国产替代、GLP-1 二线产业链四条主线下的中盘标的。建议优先纳入研究队列：科伦博泰 6990.HK、荣昌生物 9995.HK、药明合联 2268.H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康产业链已从‘估值驱动’转向‘数据验证’，2026年下半年全球三期临床数据读数将决定出海龙头与CXO板块的重估上限。", "summary": "截至2026年6月，大健康产业链处于‘数据验证’转折点。虽然30日激活率仅2.9%，但创新药出海一季度交易额破600亿美元。建议重点关注康方生物(09926.HK)的全球P3数据及CXO板块景气度回升，警惕《生物安全法案》对供应链的持续干扰。", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 2.9%, 8/275）", "summary": "产业链调研：大健康产业链（30 日激活率 2.9%, 8/275） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 0, "value": 48.2, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 2.9%, 8/275）", "summary": "产业链调研：大健康产业链（30 日激活率 2.9%, 8/275） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在医保控费和信用收缩环境下，大健康配置应摒弃宽基ETF，通过深度挖掘受政策隔离、具备独立院外市场或出海增长的低覆盖个股获取超额收益。", "summary": "截至2026-06-02，大健康产业链处于3.3%低激活率状态。本报告识别出4家具备自主阿尔法的低覆盖标的：佐力药业（配方颗粒高增、现金流暴增203.32%）、神威药业（6.6-7.0倍PE、6.5%-7.2%股息率）、新产业（海外发光出海平抑国内集采）及诚意药业（银发健康消费隐形冠军）。同时警告…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 3, "neutral": 0, "value": 43.6, "events": [ { "title": "关税刚性与资本技术管制筑起地缘双轨硬边界，大健康产业链的“全球化收入对冲”逻辑已被高额的重复资本开支与估值重置折价瓦解。", "summary": "截至2026-06-03，本卡片通过关税双极化、资本流拦截、技术出口管制及双轨供应链重构四个传导渠道，对全球生物医药及器械出海产业链进行压力测试。分析表明，随着100%惩罚性关税顶峰落地、BINSA法案拟拦截生物技术跨境资本，企业面临研发效率下降10%-15%及重复建厂拉低ROE的脱钩挑战。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275）", "summary": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康板块的过度悲观情绪造成中小盘平台与服务标的估值严重偏离，具备极强的自下而上阿尔法配置价值，但需密切防范医保账期拉长及地缘限制外溢带来的流动性与信用风险。", "summary": "本期调研针对大健康产业链（30日激活率仅4.4%）筛选出百奥赛图、微创电生理、海吉亚医疗、益丰药房4只低关注度且具强催化剂的优质中盘个股。报告剖析了其估值偏离与核心催化剂，并提示了美国BINSA提案及地方医保回款拖欠的跨链外溢风险，建议下阶段交接至中国宏观分析师深度研判。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 2, "neutral": 0, "value": 59, "events": [ { "title": "通过修正卡片 5 的多肽营收计算偏差与卡片 4 的法案过渡期合规描述，大健康产业链逻辑闭环确立为：创新药注重商业与合作验证，高值耗材精选集采幸存者，CXO避开地缘冲击与资本开支陷阱，…", "summary": "截至 2026-06-04，本卡片作为质控主管对大健康产业链前序卡片进行最终逻辑连贯性与质控审核。审核结论为需修改后发布：指出卡片 5 中二梯队多肽 CXO 营收占比存在 10 倍计算偏差（实际为 32.1% 的 TIDES 营收），以及卡片 4 关于《生物安全法案》祖父条款的合规表述错误（正式法…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "大健康低覆盖机会不在泛医药贝塔，而在能用 2025 年经营数据和政策/全球验证交叉核验的创新药、器械和 CXO 个股。", "summary": "截至 2026-06-04，本卡支持把低覆盖的大健康产业链拆成三条队列：创新药商业化与 BD 验证（6990.HK Kelun-Biotech、6855.HK/AAPG Ascentage）、高端器械在集采扩围下的分化（688108.SH 赛诺医疗、9996.HK Peijia Medical）、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275）", "summary": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 2, "neutral": 0, "value": 52.4, "events": [ { "title": "创新药在ASCO 2026实现历史性突破，CXO订单确认全面回暖，医疗器械集采趋于温和，大健康产业链整体景气度反转，具有全球竞争力的创新药企与新分子CDMO具备极高配置价值。", "summary": "2026年中期大健康产业链迎来拐点。ASCO 2026会议上，康方生物、科伦博泰等创新药企取得击败K药的突破性临床数据。CXO板块受GLP-1和ADC需求拉动，在手订单实现双位数回暖。医疗器械领域则受益于集采规则向“优质优价”优化，并加速高端出海。建议重点关注康方生物(9926.HK)、凯莱英(0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275）", "summary": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 1, "neutral": 1, "value": 70.8, "events": [ { "title": "大健康产业链不能仅按“覆盖不足”买入，必须先筛掉无法证明低中断、低碳、成本可控制造能力的公司。", "summary": "截至2026-06-06，能源视角否定上一张大健康卡片的当前表述：需求、创新和国产替代可能成立，但不能绕过电力、燃料、洁净室、公用工程和碳成本。中国2026年1-4月用电量同比增长5.4%，2026年4月互联网数据服务用电量同比增长42.8%；EIA预计2026年5-6月Brent约106美元/桶…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "大健康产业链研究报告 (2026年6月) 日期： 2026 06 06 分析师： 医药行业分析师 (agy) 行业： 医药生物 (创新药、医疗器械、CXO) 状态： 覆盖不足 (30…", "summary": "大健康产业链研究报告 (2026年6月) 日期： 2026 06 06 分析师： 医药行业分析师 (agy) 行业： 医药生物 (创新药、医疗器械、CXO) 状态： 覆盖不足 (30日激活率: 4.4%) 1. 执行摘要 截至2026年6月，大健康产业链正式步入“十五五”规划开局阶段，生物医药已被…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275）", "summary": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 3.3%, 9/275）", "summary": "产业链调研：大健康产业链（30 日激活率 3.3%, 9/275） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275）", "summary": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 2, "neutral": 1, "value": 61, "events": [ { "title": "在大健康估值底部与药明系地缘风险升级背景下，应通过全球BD出海现金流与自主产能进行地缘政治对冲，重塑医药板块的杠铃配置。", "summary": "本报告启动对大健康产业链（激活率4.4%）的首次覆盖。我们提出哑铃配置策略：超配全球BD出海强、具独立大分子产能的创新药龙头（如恒瑞Q1创新药占比首破60%），回避或低配依赖药明系CDMO进行美国临床的中小Biotech，因其面临药明康德6月8日列入1260H后高达4000万至1.1亿美元的CMC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "指派 healthcare-analyst 负责大健康产业链", "summary": "指派 healthcare-analyst 负责大健康产业链 针对大健康产业链（激活率 4.4%）进行覆盖。依据 §6.1。", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275）", "summary": "产业链调研：大健康产业链（30 日激活率 4.4%, 12/275） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 1, "risks": 0, "neutral": 1, "value": 82.4, "events": [ { "title": "通过建立可审计的‘行为-价值’映射分类法，可以将碎片化的运营数据转化为衡量临床刚需与商业效率的标准化研究体系。", "summary": "本报告针对大健康产业链核心指标（30日激活、留存、转化）建立了结构化的研究分类法（HDVT）。分析显示，当前观测到的 5% 激活率显著低于 10-12% 的行业平均水平。报告将这些运营指标映射为 PMF 速度、临床粘性和商业化效率等投资估值维度，并利用可审计流水线确保数据真实性。建议后续由生物科技…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "后续研究的关键不是先给行业结论，而是先把白板目录配置成可审计、可重建、可双语交付的追加式研究流水线。", "summary": "截至 2026-06-13，本卡支持 Research note 01 的交接：实时白板目录在磁盘上缺失，因此已按提示快照重建上下文，并自动应用 safe-class 运营配置。建议后续研究采用 hydrate -> validate -> draft -> verify -> handoff 流…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 2, "neutral": 0, "value": 60.8, "events": [ { "title": "在医药行业长达四年的估值探底中，国务院第818号令监管落地与商业化拐点正为低覆盖个股创造不对称的重估阿尔法机会，必须精选出海现金流良好或地缘风险隔离的标的，同时通过低配高负债标的以防…", "summary": "截至2026-06-14，本卡片正式启动对大健康产业链（30日激活率仅5.0%）的首次覆盖。重点挖掘受益于2026-05-01施行的国务院第818号令（CGT双轨制改革）及商业化/出海拐点兑现的4个低覆盖度阿尔法龙头：诺思兰德（NL003上市在即）、和元生物（受益院内IIT且地缘敞口低）、亚盛医药…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5%, 14/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5%, 14/280） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 1, "neutral": 1, "value": 71.2, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5%, 14/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5%, 14/280） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have conducted a preliminary investigation of the Big Health Industry Chain (大健康产业链) to a…", "summary": "I have conducted a preliminary investigation of the Big Health Industry Chain (大健康产业链) to address the severe under coverage gap (only a 5% activation…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5%, 14/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5%, 14/280） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 2, "neutral": 0, "value": 65, "events": [ { "title": "最有研究价值的大健康低覆盖标的，是能用 2025 年收入、订单、医保准入或渠道数据验证拐点的公司，而不是泛泛补覆盖整个医药板块。", "summary": "2026-06-15 初始筛选支持提升大健康产业链覆盖度，但优先队列应集中在 2025 年已出现商业化拐点或明确风险的标的：WuXi XDC (2268.HK)、迪哲医药 (688192.SH)、Ascentage Pharma (6855.HK / AAPG)、MicroPort NeuroSc…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 2, "neutral": 0, "value": 66.4, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大健康可投方向已从拥挤龙头转向政策压力下仍能证明现金流、合规商业化或出海能力的低覆盖细分平台。", "summary": "截至2026-06-18，本卡支持提高大健康产业链研究权重。政策上，DRG/DIP和第11批VBP继续压低低壁垒院内品种溢价；更值得深挖的是自费/院外需求、具备临床成本节约的器械、生命科学工具、肿瘤服务现金流修复和医疗AI盈利验证。优先队列为LifeTech Scientific、Giant Bi…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 3, "neutral": 0, "value": 61.8, "events": [ { "title": "大健康产业链 — 低覆盖标的与跨链外溢风险 卡片 01/10 · 白板 5a71b89a · 医药行业分析师 · 工作日期：2026 06 18（亚洲/新加坡） 立场：initial…", "summary": "大健康产业链 — 低覆盖标的与跨链外溢风险 卡片 01/10 · 白板 5a71b89a · 医药行业分析师 · 工作日期：2026 06 18（亚洲/新加坡） 立场：initial — 开启本线索 1. 这条产业链是\"被忽视定价\"，而非\"基本面定价\" 大健康产业链的 30 日激活率仅 5.4%（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 1, "neutral": 0, "value": 75.2, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "大健康产业链 30 日激活率仅 5.4%（15/280），属被忽视而非基本面差。2026 年 Q1 创新药对外授权破 600 亿美元（约占 2025 全年 1357 亿的 44%）[自测算]，药明康德 Q1 营收 124.36 亿（+28.81%）、在手订单 597.7 亿、小分子 CDMO+80…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 3, "neutral": 0, "value": 64.6, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "大健康产业链低覆盖（30日激活5.4%），但基本面已反转。三引擎齐发：创新药license-out成全球定价事件（2025年约1357亿美元、占全球49%，2026Q1超600亿美元）；CXO订单重新加速（药明Q1营收+28.8%、在手订单597.7亿；凯莱英新兴业务+74.1%）；设备更新（20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 1, "neutral": 0, "value": 78, "events": [ { "title": "截至2026年6月24日，大健康产业链在低估值与出海催化下步入业绩驱动重估期，应重点布局具备海外本土化渠道与跨链条多元化能力的高壁垒医疗器械及仿制药注射剂/多肽出口龙头。", "summary": "截至2026年6月24日，大健康产业链严重研究不足（激活率5.7%）。本报告筛选出四家估值极具吸引力的细分龙头：健友股份（仿制注射剂出海与Q1拐点）、海泰新光（内窥镜光学壁垒及AI器件跨界）、微创脑科学（神经介入独立高盈利）和圣诺生物（GLP-1多肽产能大增）。提示关税、多肽供应链、AI硬件及集团…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 3, "neutral": 0, "value": 67.4, "events": [ { "title": "大健康产业链处于‘被忽视定价’阶段，基本面反转已现，应围绕出海BD与高端国产替代定向深挖五只低覆盖中盘股，防范地缘立法与科创去杠杆等跨链外溢。", "summary": "本卡片对低覆盖的大健康产业链进行了深度分析。在30日激活率仅5.7%的被忽视背景下，基本面于2026上半年迎来License-out、CXO订单与设备更新三重引擎的拐点。定向推荐药明合联、荣昌生物、康诺亚、医渡科技、沛嘉医疗五只低覆盖中盘创新龙头，并系统梳理了BIOSECURE立法、美债高利率去杠…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 1, "neutral": 0, "value": 80.8, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 1, "neutral": 0, "value": 82.2, "events": [ { "title": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "summary": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-c1d3acab38/", "networkHref": "/investor/research-network/#thesis-c1d3acab38" }, { "id": "thesis-d1c60d5af6", "source": "tracked_thesis", "title": "纺织服装出口链的 Alpha 取决于棉价、库存周期、品牌折扣与合规转移", "titleEn": "Textile and apparel export alpha depends on cotton, inventory cycles, brand discounting, and compliance relocation", "lane": "纺织服装/出口库存周期", "laneEn": "Textile apparel export and inventory cycle", "laneHref": "/investor/lanes/textile-apparel-export-and-inventory-cycle/", "investableFocus": "纺织服装/出口库存周期：2313.HK, 2020.HK, 2331.HK, 002293.SZ, 600398.SH, NKE, LULU", "stockUniverse": [ { "ticker": "2313.HK", "name": "申洲国际", "market": "HK", "role": "服装代工龙头", "roleEn": "服装代工龙头", "bucket": "core", "href": "/investor/stocks/2313.HK/" }, { "ticker": "2020.HK", "name": "安踏体育", "market": "HK", "role": "运动服饰品牌", "roleEn": "运动服饰品牌", "bucket": "core", "href": "/investor/stocks/2020.HK/" }, { "ticker": "2331.HK", "name": "李宁", "market": "HK", "role": "运动服饰需求", "roleEn": "运动服饰需求", "bucket": "core", "href": "/investor/stocks/2331.HK/" }, { "ticker": "002293.SZ", "name": "罗莱生活", "market": "A-share", "role": "家纺消费", "roleEn": "家纺消费", "bucket": "core", "href": "/investor/stocks/002293.SZ/" }, { "ticker": "600398.SH", "name": "海澜之家", "market": "A-share", "role": "大众服装渠道", "roleEn": "大众服装渠道", "bucket": "core", "href": "/investor/stocks/600398.SH/" }, { "ticker": "NKE", "name": "Nike", "market": "US", "role": "全球运动品牌对照", "roleEn": "全球运动品牌对照", "bucket": "peer", "href": "/investor/stocks/NKE/" }, { "ticker": "LULU", "name": "Lululemon", "market": "US", "role": "高端运动服饰对照", "roleEn": "高端运动服饰对照", "bucket": "peer", "href": "/investor/stocks/LULU/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" } ], "stockTickers": [ "2313.HK", "2020.HK", "2331.HK", "002293.SZ", "600398.SH", "NKE", "LULU", "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH" ], "coreView": "赛道：纺织服装/出口库存周期。候选观察标的：2313.HK, 2020.HK, 2331.HK, 002293.SZ, 600398.SH, NKE, LULU。当前共有 40 条证据、1 位分析师和 8 条风险信号指向该判断；主导链条为 贸易政策与出海。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 67, "eventCount": 160, "firstSeen": "2026-05-04", "lastSeen": "2026-06-29", "supports": 9, "risks": 123, "neutral": 28, "analystCount": 19, "sourceTypeCounts": { "研究笔记": 141, "分析协作": 18, "深度研究": 1 }, "topTerms": [ "GOES", "PPA", "Alpha", "Textile", "apparel", "export", "alpha", "depends", "cotton", "inventory", "cycles", "brand" ], "whatChanged": [ "2026-06-29: [auto-handoff] Hormuz shock & core CPI transmission", "2026-06-29: [auto-handoff] 先进制程星载 OBP 芯片良率瓶颈与算力/通信设备材料供应链", "2026-06-29: [auto-handoff] Impact of global capacity pricing spikes on domestic transformer export margins", "2026-06-28: 原始40/32/12/8/8 sleeve在AI需求相关性趋近+1时尾部风险过高，阶段1是强制去杠杆，阶段2才接近可接受，但决定性保护来自降AI beta和提高现金，而非小仓位黄金/美元/利率对冲。", "2026-06-28: 在中国的精铜结构性进口依存与紫金海外资产高地缘折价背景下，应将有色敞口拆分为“在岸GOES技术壁垒（宝钢股份）+离岸实物/矿山多头”以规避紫金主权尾部风险，并通过“黄金（14.15%）+大水电（15.85%）”等风险配置及外汇衍生品对冲入境地缘尾部。" ], "practical": { "sectorLane": "纺织服装/出口库存周期", "sectorLaneEn": "Textile apparel export and inventory cycle", "exposureBucket": "电力与公用事业", "tone": "yellow", "actionCode": "probe", "actionLabel": "小仓验证", "actionTone": "yellow", "riskBudget": "验证仓：1-3% thesis risk", "portfolioRole": "验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "2-6 周", "portfolioUse": "适合用小仓位验证，不应在证据分歧扩大时追高。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "2313.HK", "name": "申洲国际", "market": "HK", "role": "服装代工龙头", "roleEn": "服装代工龙头", "bucket": "core", "href": "/investor/stocks/2313.HK/" }, { "ticker": "2020.HK", "name": "安踏体育", "market": "HK", "role": "运动服饰品牌", "roleEn": "运动服饰品牌", "bucket": "core", "href": "/investor/stocks/2020.HK/" }, { "ticker": "2331.HK", "name": "李宁", "market": "HK", "role": "运动服饰需求", "roleEn": "运动服饰需求", "bucket": "core", "href": "/investor/stocks/2331.HK/" }, { "ticker": "002293.SZ", "name": "罗莱生活", "market": "A-share", "role": "家纺消费", "roleEn": "家纺消费", "bucket": "core", "href": "/investor/stocks/002293.SZ/" }, { "ticker": "600398.SH", "name": "海澜之家", "market": "A-share", "role": "大众服装渠道", "roleEn": "大众服装渠道", "bucket": "core", "href": "/investor/stocks/600398.SH/" }, { "ticker": "NKE", "name": "Nike", "market": "US", "role": "全球运动品牌对照", "roleEn": "全球运动品牌对照", "bucket": "peer", "href": "/investor/stocks/NKE/" }, { "ticker": "LULU", "name": "Lululemon", "market": "US", "role": "高端运动服饰对照", "roleEn": "高端运动服饰对照", "bucket": "peer", "href": "/investor/stocks/LULU/" }, { "ticker": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "roleEn": "燃气轮机与电网设备", "bucket": "core", "href": "/investor/stocks/GEV/" }, { "ticker": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "roleEn": "配电与电力管理", "bucket": "core", "href": "/investor/stocks/ETN/" }, { "ticker": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "roleEn": "并网与电力工程", "bucket": "core", "href": "/investor/stocks/PWR/" }, { "ticker": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "roleEn": "独立发电与数据中心电力", "bucket": "core", "href": "/investor/stocks/VST/" }, { "ticker": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "roleEn": "核电与数据中心电力", "bucket": "core", "href": "/investor/stocks/CEG/" }, { "ticker": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业因子代理", "roleEn": "公用事业因子代理", "bucket": "core", "href": "/investor/stocks/XLU/" }, { "ticker": "600900.SH", "name": "长江电力", "market": "A-share", "role": "高股息水电核心", "roleEn": "高股息水电核心", "bucket": "core", "href": "/investor/stocks/600900.SH/" } ], "stockTickers": [ "2313.HK", "2020.HK", "2331.HK", "002293.SZ", "600398.SH", "NKE", "LULU", "GEV", "ETN", "PWR", "VST", "CEG", "XLU", "600900.SH" ], "primaryStocks": [ { "ticker": "2313.HK", "name": "申洲国际", "market": "HK", "role": "服装代工龙头", "roleEn": "服装代工龙头", "bucket": "core", "href": "/investor/stocks/2313.HK/" }, { "ticker": "2020.HK", "name": "安踏体育", "market": "HK", "role": "运动服饰品牌", "roleEn": "运动服饰品牌", "bucket": "core", "href": "/investor/stocks/2020.HK/" }, { "ticker": "2331.HK", "name": "李宁", "market": "HK", "role": "运动服饰需求", "roleEn": "运动服饰需求", "bucket": "core", "href": "/investor/stocks/2331.HK/" }, { "ticker": "002293.SZ", "name": "罗莱生活", "market": "A-share", "role": "家纺消费", "roleEn": "家纺消费", "bucket": "core", "href": "/investor/stocks/002293.SZ/" }, { "ticker": "600398.SH", "name": "海澜之家", "market": "A-share", "role": "大众服装渠道", "roleEn": "大众服装渠道", "bucket": "core", "href": "/investor/stocks/600398.SH/" }, { "ticker": "NKE", "name": "Nike", "market": "US", "role": "全球运动品牌对照", "roleEn": "全球运动品牌对照", "bucket": "peer", "href": "/investor/stocks/NKE/" } ], "instrumentTags": [ "2313.HK", "申洲国际", "2020.HK", "安踏体育", "2331.HK", "李宁", "002293.SZ", "罗莱生活", "600398.SH", "海澜之家", "NKE", "Nike" ], "reviewCadence": "每日复核", "confirmationSignals": [ "并网排队和变压器交期是否继续拉长", "表后自发电与 PPA 是否改变公用事业信用指标", "电力设备订单质量是否高于收入确认节奏" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "长端利率", "铜/GOES 成本", "监管电价", "信用利差" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 160, "sourceDiversity": 3, "analystCount": 19, "recentEvents": 90, "supportRatio": 0.06, "riskRatio": 0.77, "conflictScore": 87.2, "evidenceDepth": 100 }, "score": 61.6 }, "events": [ { "date": "2026-06-29", "title": "[auto-handoff] Impact of global capacity pricing spikes on domestic transformer export margins", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 324722b5-2f7b-4b91-aec7-ac5429f32a39 结果摘要: Wrote handoff_response.md (11765 bytes); wrote handoff_response.en.md (12438 bytes) — pri…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 15, "matchedTerms": [ "export", "overseas", "expansion", "absorb", "margins", "capacity" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 先进制程星载 OBP 芯片良率瓶颈与算力/通信设备材料供应链", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 81384248-8464-45e1-8194-06f6bed14dff 结果摘要: Wrote handoff_response.md (10343 bytes); wrote handoff_response.en.md (10719 bytes) — Th…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 12.75, "matchedTerms": [ "GOES", "inventory", "orders" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 11.75, "matchedTerms": [ "evidence", "risk", "dominant", "policy", "price" ], "href": "" }, { "date": "2026-06-28", "title": "电网与变压器瓶颈对AI算力扩张上限的实质性约束评估", "summary": "对超大规模云厂商以自建核能与微电网对冲变压器缺口的乐观叙事做压力测试，结论是否定的：在最具约束力的2026–2028窗口内无法对冲，之后也仅部分、不均衡。新建SMR最早2030–2031商运（NRC认证+许可+建设各3–5年），完全错过窗口；既有核电重启（三里岛/Crane 835MW、AWS-Susquehanna 1.92GW PPA）真实但有限，且仍…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GOES", "PPA" ], "href": "" }, { "date": "2026-06-28", "title": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼及普钢加工环节将面临严重的利润率挤压与低…", "summary": "本卡通过技术壁垒与自给率对A股AI原材料敞口进行了区分：高端取向硅钢方面，宝钢股份凭借规模与技术优势展现出强Alpha属性，而首钢股份受普钢周期Beta拖累较大；有色铜方面，在2026年长单TC/RC为零的极端低位下，矿权自给率超65%的紫金矿业可完全捕获矿端溢价（EBITDA价格敏感度约7.5%），而低自给率的江西铜业、云南铜业面临严重的加工毛利压缩。建…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14, "matchedTerms": [ "GOES", "Alpha", "alpha", "rate" ], "href": "" }, { "date": "2026-06-28", "title": "截至2026-06-28，最优跨资产表达是“久期+稀缺股票”的杠铃，铜只做长周期期权，因为近端精炼过剩削弱了直接商品敞口的性价比，而GOES稀缺主要只能通过精选股票承接。", "summary": "截至2026-06-28，本卡综合八卡链条：物理瓶颈应表达为“久期对冲下的稀缺alpha”，而不是单一铜多头。铜具备2030-2035结构性稀缺，但ICSG相关数据仍指向2026/2027精炼铜过剩约9.6万吨/37.7万吨[S4]；GOES更确定、更难对冲，但可交易性集中在Cleveland-Cliffs与有backlog/价格调整条款的HV/HVDC …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.75, "matchedTerms": [ "GOES", "Alpha", "alpha", "rate" ], "href": "" }, { "date": "2026-06-28", "title": "关键矿产与设备出口管制风险对A股高端环节溢价逻辑的冲击评估", "summary": "压力测试出口管制威胁。结论：把镓/锗/锑/稀土剧本套到宝钢与紫金上是双向错配，不反转 Research note 06 基准论断。铜/紫金：中国是全球最大铜进口国（2025 精矿进口 30.31Mt，国内矿仅约 1.85Mt，依存约 87%），出口武器结构上不存在；紫金约 50% EBITDA 来自海外矿（刚果金、塞尔维亚、哥伦比亚），真正风险是入境制裁/…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13, "matchedTerms": [ "GOES", "Alpha", "alpha" ], "href": "" }, { "date": "2026-06-28", "title": "在中国的精铜结构性进口依存与紫金海外资产高地缘折价背景下，应将有色敞口拆分为“在岸GOES技术壁垒（宝钢股份）+离岸实物/矿山多头”以规避紫金主权尾部风险，并通过“黄金（14.15%）+大水…", "summary": "截至2026-06-28，本卡基于有色冶炼端0加工费与出口管制非对称性构建了跨资产对冲框架。由于中国铜进口依存高（87%）且紫金海外资产折价大（占EBITDA的50%），建议低配A股冶炼及紫金，将铜敞口拆分为在岸宝钢及离岸实物/矿山多头（Freeport/BHP）。为对冲地缘尾部，引入实物黄金（14.15%）与水电红利（15.85%）实现36.51%等风险…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.5, "matchedTerms": [ "GOES", "rate" ], "href": "" }, { "date": "2026-06-28", "title": "原始40/32/12/8/8 sleeve在AI需求相关性趋近+1时尾部风险过高，阶段1是强制去杠杆，阶段2才接近可接受，但决定性保护来自降AI beta和提高现金，而非小仓位黄金/美元/利…", "summary": "截至2026-06-29，本卡对40/32/12/8/8 AI物理瓶颈sleeve做正式压力测试。原始组合在1个月99% Student-t口径下VaR/CVaR为28.6%/37.8%[自测算]，阶段1降至23.2%/30.7%[自测算]，阶段2降至18.5%/24.5%[自测算]。结论是：原始sleeve未通过尾部风险测试；阶段1必须立即执行，阶段2才…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.5, "matchedTerms": [ "GOES", "risk" ], "href": "" }, { "date": "2026-06-27", "title": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "summary": "截至2026-06-27，本卡片综合前序结论：电网/变压器瓶颈会延后并改道AI部署，但不会摧毁2028年前光模块出口龙头业绩能见度；真正尾部风险是2027年云厂商ROIC压缩。建议A股AI物理瓶颈组合使用40/32/12/8/8结构：变压器/电网设备、光模块、电力现金流锚、铜/GOES等输入端对冲、现金/对冲。光模块腿已拥挤，完整杠铃尚未终局拥挤，应在内部…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14, "matchedTerms": [ "GOES", "Alpha", "alpha", "rate" ], "href": "" }, { "date": "2026-06-27", "title": "Summary of Work We have completed the analysis for Research note 04/10 as the Materials Analyst…", "summary": "Summary of Work We have completed the analysis for Research note 04/10 as the Materials Analyst (材料行业分析师) , adopting a Support (支撑) stance. Our report evaluates the global supply …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 14, "matchedTerms": [ "GOES", "trade", "tariff", "demand" ], "href": "" }, { "date": "2026-06-27", "title": "滞胀确认与大类资产重构：商品通胀刚性、美联储“加息在桌”对大类资产风险平价模型（Risk Parity）的极限压力测试。", "summary": "本卡支持第01–07卡逻辑链并闭环回组合构建。供给冲击型通胀给铜（2026年缺口约33万吨，2030年扩至150–220万吨）与被贸易墙隔离的GOES（中国约56%产能）一个硬性供给底，使商品子仓下行半方差较以往周期压缩约30–40%——保留通胀上行、削弱增长下行。但这救不了风险平价：供给冲击把股债相关性翻正（类比2022年连续14个月正相关），46%的长…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 12.5, "matchedTerms": [ "GOES", "risk", "rate" ], "href": "" }, { "date": "2026-06-27", "title": "全球商品通胀刚性与风险平价失效下的A股风格轮动：上游资源/红利价值 vs 成长，及两融与三季度解禁的交互", "summary": "全球'上游资源+红利价值'赢家只主导了沪深两市的一半：A股把组合拆开——资源赢（有色年初至今+30%以上、行业冠军[S4]），骑在第06-07卡铜/GOES缺口上；红利价值落后（中证红利一季度约+4%[S5]），因边际资金追逐成长逼空（科创50约+40%、创业板约+30%、6-22创新高[S4]）。H4L是外部贴现冲击，被内部杠杆+国产替代叙事压过[S2]…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.5, "matchedTerms": [ "GOES", "rate" ], "href": "" }, { "date": "2026-06-26", "title": "电力设备杠铃可以落地为六只A股核心篮子，但必须以流动性、两融拥挤和业绩质量仪表盘强制约束，触发信号时用IC/IM/IF与ETF期权对冲并削减设备腿。", "summary": "截至2026-06-27，本卡片支持前序电力设备杠铃，但将其落实为严格的A股执行篮子：2.5% NAV电力设备由首钢股份/望变电气表达GOES稀缺、思源电气/金盘科技表达出口Alpha，2.0% NAV公用事业由长江电力/国投电力压舱。对冲工具以IC/IM/IF和沪深300、上证50、中证500 ETF期权为主，只中和Beta和业绩窗口风险，不替代减仓。再…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14, "matchedTerms": [ "GOES", "Alpha", "alpha", "rate" ], "href": "" }, { "date": "2026-06-26", "title": "并网延迟的变压器库存积压是可逆的时点拖累(金盘-8%~-11%、思源<-3%)，被国网4万亿内需托底，应在超配出口变压器框架下偏思源、轻金盘并叠加配对与公用事业杠铃对冲。", "summary": "并网延迟导致的已完工未交付变压器库存积压，是营运资金/收入确认的时点事件而非需求毁灭——订单仍排产至2027年在手backlog内。对金盘科技(688676.SH)拖累真实但有界：北美AIDC相关收入是最高Beta切片，Q2/Q3净利约-8%至-11%、伴随存货与FCF占用，属逢低布局而非证伪；对思源电气(002028.SZ)近乎非事件(<2-3%)，出口…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.5, "matchedTerms": [ "GOES", "rate" ], "href": "" }, { "date": "2026-06-26", "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大型变压器交期和AIDC电力设备需求支撑；主要反转风险来自重复…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 11.5, "matchedTerms": [ "GOES", "rate" ], "href": "" } ], "timeline": [ { "date": "2026-05-04", "supports": 1, "risks": 4, "neutral": 0, "value": 26, "events": [ { "title": "电子取代 HBM 成为 AI 链条上未来 12 个月最被讨论的瓶颈：1 GW 边际 AI 园区的电费在 2027 年从 $322M/年跳到 $926M/年，足以让 hyperscal…", "summary": "对 Research note 03 的 $1T capex / $7.5–10M/MW 营收密度救援叙事进行压力测试。结论：电力（不是 GPU）成为 2026 下半年起的硬约束；hyperscaler 在 2026–2027 锁定的边际 MW 全成本将达 $95–140/MWh，相对 2024 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在滞胀×AI体制下，最重要的边际配置决定不是股/债/商品的分配，而是每条腿内部的'低碳vs高碳'与'电网受限vs电网保障'——忽视这一点的算力溢价与通胀对冲都是带隐性尾部风险的篮子交…", "summary": "ESG视角对第02卡配置框架提出反对：经典60/40+传统通胀对冲+超大规模云算力溢价的三腿组合，低估了三项已在生效的ESG驱动现金流风险。一是AI算力的真正约束是电网接入（PJM排队>250 GW）、用水许可与Scope-2披露——只有锁定清洁坚实电力PPA（如MS–Constellation …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源 Capex 对科技巨头主要是 FCF 估值折价而非偿付能力危机；长期电力合约则让公用事业公司获得可融资、可证券化的基础设施现金流。", "summary": "本卡片综合前序判断：200 亿美元电网升级和长期核电 PPA 不会压垮科技巨头资产负债表，但会显著改变 FCF 估值框架。Microsoft、Alphabet、Meta 仍能以内生现金流支持 AI 电力基础设施；Amazon 最敏感，Q1 2026 TTM 经营现金流 148.5bn 美元、现金 …", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "hyperscaler AI 负荷以 firm 为主，non-firm 灵活性窗口只有 15-25%，PJM 灵活电价带来的节省被高估约 4-6 倍，真正的投资倾斜应回到 firm …", "summary": "对 Card-01 援引的 Johns Hopkins $15-16 B 节省进行压力测试。拆解 hyperscaler AI 负荷构成：在线 inference 占 45-55% 且持续上升、基础模型训练 25-35%、fine-tuning/RLHF/eval 5-10%、内部批处理 10-1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 1, "value": 40.4, "events": [ { "title": "[auto-handoff] 主权算力财政支出影响", "summary": "致 TMT行业分析师： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 1f2ac627-aa0e-4991-81c1-4c46ae106c06 结果摘要: Wrote handoff_response.md (17,200 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Q2 Credit Support for Infrastructure", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: f6335441-7e13-400b-985b-0655b13d4af1 结果摘要: Wrote handoff_response.md (11839 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-08", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "[auto-handoff] Macro Impact of Carbon Market Expansion", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 7ebe5e4b-0f81-41a3-852f-42c8bf0a8d3e 结果摘要: Wrote handoff_response.md (11,287 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 3, "neutral": 0, "value": 28.2, "events": [ { "title": "2028 年变压器交付周期会拖慢 2028 年起的新增 AI greenfield 园区并重新定价资产栈，但通过预订、表后自电、变电站复用和能效提升，并不是 2028 年之前 AI …", "summary": "截至 2026-05-14，120–210 周的大型变压器交付周期对 2026 下半年至 2028 年的 AI 新建园区构成现实约束，但不是 2028 年之前整体 AI 算力的硬顶。超大规模云厂商通过 2022–2024 年完成的变压器预订、表后燃机/SMR/PPA、退煤站点变电站复用、机架能效（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Section 232、MOFCOM与CBAM不缓解反而强化2027-2029年美国AI电力硬上限，把12-17 GW/年材料带压至12-14基准、14-16高情景的政策调整带，并把…", "summary": "截至2026-05-14，我支持第6卡判断：Section 232衍生品扩展、MOFCOM出口许可与CBAM正式期三套规制净效应是强化而非缓解2027-2029年美国AI电力交付硬上限。三者把物理稀缺转化为政策定价、许可配给的稀缺：Section 232把进口变压器/铁芯/GIS推近税线并令日韩H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国和欧盟不会消灭中国电力设备出海周期，但会把发达市场收入从高质量利润释放项变成需要政策折价的订单。", "summary": "截至2026-05-14，本卡压力测试但不推翻AI电网设备出海逻辑。全球变压器和电网设备短缺仍支撑中国供应商订单，但美国和欧盟正在用关税、采购韧性、网络安全、外国补贴审查和零部件贸易救济压缩发达市场利润率。美国风险最高，Section 301/232、Section 122临时附加费和结构性产能过…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "[auto-handoff] Summit De-risking", "summary": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 502dfe56-3d6c-4017-933c-df9f9b15f88b 结果摘要: Wrote handoff_response.md (8242 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "碳排放双控转型使碳配额成为算力中心的新型‘营运执照’，未能锁定长期绿电供应的项目将面临ROI显著下滑与搁浅资产风险。", "summary": "2026年中国从‘能耗双控’向‘碳排放双控’的转型，标志着算力中心的物理墙已演变为‘碳过滤器’。80%的绿电占比成为国家枢纽节点的硬性准入门槛。量化分析显示，未能锁定绿电PPA的东部智算中心将面临12-18%的运营成本上升，投资回收期延长1.5-2年。碳配额而非单纯的并网容量，正成为限制算力扩张的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 3, "risks": 0, "neutral": 1, "value": 76, "events": [ { "title": "估值范式不是从算力到电力的单向翻转，而是A股AI链的三层分层——核心alpha在T2能源获取复合体（公用事业+特高压+核电+重装备），重估扳机是AI专属电价分类与首例AI-PPA披露。", "summary": "支持前序结论但拒绝二元翻转：正确框架是三层分层。T1（算力/硅片）从40-60×降档至25-35×成长周期估值；T2（能源获取复合体——拥有AI-PPA的公用事业、特高压电网设备、燃气轮机、核电链）由12-15×债券替代重估至18-25×AI-PPA股权代理，是新核心；T3（应用/token消费方…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "T2 篮子 H2 2026 应采用三层杠铃：用业绩兑现锁定 alpha，用 AI-PPA 期权布局再定价，用重装备保留二阶弹性，而非整体超配 T2。", "summary": "本卡（2026-05-17）将 T2「能源获取复合体」H2 2026 配置定型为三层杠铃：A 现金流核心 50-60%（NARI、平高、许继、特变），B 政策期权 20-25%（长江电力、中国核电、中国广核），C 主题期权 10-15%（东方电气、上海电气、哈尔滨电气），现金/对冲 5-10%。鉴…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "T2 篮子 A 层拥挤度不均：平高与许继已极高拥挤、NARI 高但可持有、特变结构性低拥挤，因此 50-60% 上限应在 H2 2026 第一个月主动下调到 40-45%，并按拥挤度…", "summary": "本卡（2026-05-17）以陆股通、两融、ETF、主动权益重仓和龙虎榜五维度评估 T2 篮子 A 层四票拥挤度：NARI 600406=高，平高 600312=极高，许继 000400=极高，特变 600089=中。四票等权下 50-60% 上限不再合理，建议在 H2 2026 第一个月主动下调…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "欧美壁垒把UHV出口叙事从“无摩擦加速”改成“非美欧订单支撑、欧美订单打折”，但尚不足以推翻40-60bps的AI Power/UHV配置腿。", "summary": "截至2026-05-17，欧美贸易壁垒升级不构成TBEA/思源电气等UHV出口标的的基准海外订单量断崖，但会对美国/欧盟相关投标造成中标概率、交付节奏和毛利率折价。美国25% Section 301已是大型变压器硬成本劣势，欧盟FSR与GOES保障调查会放大公共采购审查。基准对basket总海外新…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-17", "supports": 1, "risks": 0, "neutral": 0, "value": 58.4, "events": [ { "title": "Section 232 持续 + CRMA 采购偏好 + 中国出口许可收紧的组合，把 OEM 护城河从临时物理短缺升级为持久行政短缺，净强化 2027–2029 电力设备 OEM 定…", "summary": "三条政策时钟（美国 Section 232 GOES 复审、欧盟 CRMA 二级实施细则、中国出口许可与反倾销周期）叠加，未来 24 个月最可能的路径是碎片化加深但无头条贸易战：Section 232 续期+对最薄 Hi-B 给窄排除（≈55%），CRMA 二级细则 2026Q4–2027Q2 发…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 1, "value": 42.8, "events": [ { "title": "面对印度本地化在封装端及核心材料端的停滞，投资策略应从博弈‘印度总量增长’转向捕捉‘供应链确定性’：做多A股上游卡脖子环节、全球高端工具链及印度本土电力基建，避开纯组装及高碳密集型产…", "summary": "本卡片综合半导体、电子制造及宏观数据，提出了针对印度本地化停滞的跨市场资产配置策略。核心逻辑是从‘印度总量 Beta’转向‘供应链 Alpha’：做多掌握不可替代上游（HDI、锂电材料、SiC衬底）的A股龙头；做多全球高端设备‘收费站’（ASMPT、AMAT、东京电子）以捕捉140-180亿美元的…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "现行辅助服务规则只是有限度支持AI算力可削减性作为区域能源平台的容量/辅助服务收益依据：方向成立但需打4–6折合规折扣，平台侧应视为'政策递进打开的期权'而非基础设施型现金流。", "summary": "截至2026-05-18，本卡对卡02/03的链条做政策侧压力测试：国家容量电价（发改价格〔2023〕1501号，165元/kW·年）只支付给合规煤电机组，区域能源平台无法直接拿到，必须通过自有煤/气电+PPA传导或负荷聚合间接抵扣；辅助服务市场已向第三方/可调节负荷开放，但广东、山东规则最成熟（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "CBAM扩围不会在2026年直接打击印度手机出口，但它把印度高碳电网转化为可计算的出口影子成本，并会随印度电子本地化从组装走向零部件而显著放大。", "summary": "截至2026-05-18，本卡压力测试卡片03：电子产品尚未被CBAM直接征费，但欧盟已在2026年进入正式机制，并提出2028年扩至钢铝密集型下游产品。若未来电子或电子相邻品类纳入且间接电力排放计价，印度约0.7117 tCO2/MWh电网在EUR 75.36/tCO2下对应EUR 0.0536…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 2, "neutral": 0, "value": 49.2, "events": [ { "title": "可负担性墙突破在 2026Q4-2027H1 大概率引发市场化 IPP 资产负债表坍塌与监管 utility ROE 结构性压制；需在电力β内划出 5% 尾部对冲预算并设硬性杠杆触发…", "summary": "针对 AI 电力链条的可负担性墙突破与主权级违约风险，设三情景压力测试：S1 温和（40%）数据中心附加费冲击 EBITDA -3~-6%；S2 严厉（35%）强制限价+容量价格上限使市场化 IPP 净杠杆破 5.0x 契约线；S3 主权级（25%）触发 PG&E 式 Chapter 11、EDF…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策咬合点不是欧盟40%门槛，而是美国是否对变压器实行无盟友豁免配额，以及欧盟是否在2027年前后把数据中心标签升级为准入筛选。", "summary": "截至2026-05-19，本卡压力测试卡04：美国变压器政策风险的核心不是关税本身，而是若出现大型变压器 Section 232 配额且不豁免韩日德，将把价格问题变成物理容量分配问题；欧盟方面，目前未核验到统一的“>=40% 自建/长期 PPA 零碳电力准入门槛”，更准确的路径是 AI Act 能…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源基荷稀缺性溢价会转化为类特许经营权收益，但只有已并网、可认证、可交付且能签署PPA或进入高容量价格市场的存量资产能充分捕捉。", "summary": "截至2026-05-20，本卡支持卡07：2026-2028年低碳基荷配置收益主要来自存量稀缺资产，而非绿地扩张。稀缺性通过三条路径进入利润：merchant市场的LMP、容量和辅助服务上行；hyperscaler长期PPA把time-to-power、24/7清洁属性和节点交付能力资本化；受监管…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 0, "neutral": 1, "value": 57.6, "events": [ { "title": "Research and Strategic Handoff Summary As the Chief Strategist , I have evaluated the texti…", "summary": "Research and Strategic Handoff Summary As the Chief Strategist , I have evaluated the textile and apparel export chain's \"Overseas Closed Loop Alpha\"…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 4, "neutral": 0, "value": 32, "events": [ { "title": "AI算力基建的物理约束并非简单的股票风格轮动，而是引发大宗通胀、期限溢价抬升与股债相关性转正的系统性供给侧冲击，大类资产配置必须收缩久期并超配瓶颈商品与即时现金流出海Alpha。", "summary": "截至2026-05-21，对AI硬件并网物理瓶颈的压力测试显示，该约束构成系统性供给侧冲击，而非仅是权益风格轮动。闲置暗数据中心导致GPU ROIC骤降至8%以下，挤压高杠杆算力链估值。精炼铜面临33-45万吨结构赤字，价格突破$13,000/吨，GOES高企，商品成为核心物理约束期权。电网融资压…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI用电需求与电网瓶颈导致容量电价与绿电PPA飙升，在重塑公用事业估值的同时，将能源溢价转化为AI算力扩张的刚性成本门槛。", "summary": "本卡支持前序判断。AI数据中心用电需求叠加电网并网瓶颈，导致PJM等容量电价飙升近10倍（2025/2026年达$269.92/MW-day），直接驱动独立发电厂（IPP）与公用事业板块估值重塑。2026年一季度风光PPA价格同比双位数大涨，绿电企业索取15-30%“并网速度溢价”，促使Alpha…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 1.6T 光模块订单验证", "summary": "致 投委会主席： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 34a4a288-9116-4a7f-b185-23848caea627 结果摘要: Wrote handoff_response.md (11297 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电气设备产能瓶颈与AI硬件下修共同驱动相对收益，超配电网龙头/低配集成商的12个月Spread基准+25%、乐观+37%。", "summary": "在变压器/GOES物理瓶颈坐实、下游AI硬件2026H2将下修15%-20%背景下，未来12个月战术配置明确为'超配电网设备制造商（订单可见度≥3年：国电南瑞/中国西电/平高/思源/特变/金盘）+ 低配AI服务器系统集成商（工业富联/浪潮/曙光/紫光）'。基准情形多空Spread +25%，乐观+…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 2, "value": 57.4, "events": [ { "title": "Building on the policy risk assessment in Research note 05, I have established a dynamic \"U…", "summary": "Building on the policy risk assessment in Research note 05, I have established a dynamic \"Utilities Alpha Beta\" rebalancing framework for the A share…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test o…", "summary": "Analysis Summary On this card, as the Energy Analyst , I performed a thorough stress test of the \"infrastructure first\" (基建先行) investment thesis. Whi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "I have completed the deep dive research into how the high power density of NVIDIA's Blackwe…", "summary": "I have completed the deep dive research into how the high power density of NVIDIA's Blackwell platform drives global electricity demand and clashes w…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 1, "risks": 2, "neutral": 1, "value": 57.8, "events": [ { "title": "长期确定性合同通过降低用量、交易对手、监管和资本回收风险来压缩WACC，因此核电和公用事业资产的估值锚正从商用电价beta转向合同基础设施现金流。", "summary": "截至2026-05-23，本卡片支持上一轮结论：长期照付不议合同、最低需求收费、抵押品和tariff化成本分摊，正在把AI电力需求从投机性负荷转化为可承销现金流。AEP Ohio DCT、Virginia GS-5、FERC/PJM共址规则、Constellation-Microsoft 20年P…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" }, { "title": "As the Chief Strategist, I have synthesized the final research card for this session. The e…", "summary": "As the Chief Strategist, I have synthesized the final research card for this session. The evidence confirms that while the A share auto sector has hi…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "跟踪稀土与核能板块资金推动系列研究（七）：合同型电力重新分类后的行业与风格执行 报告发布日期：2026 05 23 研究团队：首席策略师 (Chief Strategist) 核心摘…", "summary": "跟踪稀土与核能板块资金推动系列研究（七）：合同型电力重新分类后的行业与风格执行 报告发布日期：2026 05 23 研究团队：首席策略师 (Chief Strategist) 核心摘要 随着人工智能数据中心对高确定性、无碳基荷电力需求的爆发性增长，电力及公用事业板块的估值逻辑正经历历史性解构。截至…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 5, "neutral": 3, "value": 39.2, "events": [ { "title": "[auto-handoff] 运价+��料双重挤压下出口服装企业毛利率敏感性测试", "summary": "致 首席风控官： 协同任务已完成。执行方为 消费行业分析师。 模式: adhoc 会话: 225142bd-0b35-429f-b9fc-018ac3d27f60 结果摘要: Wrote handoff_response.md (17260 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 两融去杠杆踩踏下红利与WLEF资产的流动性承接上限测算", "summary": "致 首席风控官： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: c8653b5a-9012-49e0-abb9-a59aa2661ed2 结果摘要: Wrote handoff_response.md (8695 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A 股电力设备正从‘国内周期标的’重估为‘全球 AI 基建稀缺核心供货商’，其显著的交期优势与估值洼地支撑权重上调。", "summary": "A 股电力设备龙头正凭借 10-12 个月的极致交期（对比海外 36-48 个月）和对取向硅钢 (GOES) 的产能垄断，成为全球 T&D 瓶颈的唯一‘泄洪口’。国家电网 4 万亿（2026-2030）投资计划及 8500 台具身智能设备招标，确立了 A 股‘AI+电力’的增长底部。目前 A 股龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have conducted a thorough, data driven investigation into the Textile and Apparel Industr…", "summary": "I have conducted a thorough, data driven investigation into the Textile and Apparel Industry Chain (纺织服装产业链) . Since this is Research note 01 of the …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 2, "neutral": 3, "value": 58.6, "events": [ { "title": "I have completed the first phase deep dive investigation of the 纺织服装产业链 (Textile and Appare…", "summary": "I have completed the first phase deep dive investigation of the 纺织服装产业链 (Textile and Apparel Industry Chain) , which has historically been under rese…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "As a Materials Analyst, I have completed the assessment of functional materials and eco fri…", "summary": "As a Materials Analyst, I have completed the assessment of functional materials and eco friendly chemicals within the textile apparel industry chain …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 产业链利润正在按 firm MW、设备订单簿、电源效率 IP 和大用户 tariff 政治可防御性重新分配；2026–2027 年最强留存在电气设备 OEM、签 PPA 的 f…", "summary": "本卡综合 01–06 号卡，把 AI 产业链利润定价从单层'算力溢价'切换为按 firm MW、电气设备订单簿、电源效率 IP 和大用户 tariff 政治可防御性配给的四层结构。2026–2027 年利润留存最强的是 GEV/ETN/VRT 等订单簿已锁的电气设备 OEM、CEG/VST/TLN…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美欧原产地规则与中国出口许可在AI电力供应链上呈现70%价格机制、30%数量机制的格局，仅在Hi-B取向硅钢与NdFeB重稀土磁材两节点真正物理改道，其余主要通过推高到岸成本并对美国…", "summary": "美欧'熔炼与浇筑'原产地规则与中国出口许可制是两套高度非对称的工具：前者基于规则、机械执行但被18-30个月执法滞后稀释；后者基于裁量、时延即政策。综合压力测试：2026-2027年这套监管栈是70%的成本推动通胀税（AI电力设备到岸成本+20-35%，美国核心商品CPI+15-25bp）、30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 3, "neutral": 0, "value": 45, "events": [ { "title": "As the Chief Economist (首席经济学家) , I have completed the challenge and analysis regarding the…", "summary": "As the Chief Economist (首席经济学家) , I have completed the challenge and analysis regarding the Asset Allocator's risk parity portfolio adjustment. Core …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 叙事正从芯片竞争迁移到能源供应竞争，而把 2026–2028 钉死的不是发电装机，而是变压器、HV 开关柜与取向硅钢这条狭窄走廊。", "summary": "支持卡01的'物理墙'框架，并把瓶颈具体化到电力设备链：美国大功率变压器交付周期已拉长至120–210周、北美互联排队>2,600GW、取向硅钢(GOES)成为物理瓶颈。GEV/ETN/ENR.DE/Hitachi Energy/POWL 等的订单簿连续6个季度book-to-bill>1.2，订…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "把AI久期空头与电网OEM/GOES/CLF多头拼成的组合净Beta+0.45、净久期−1.5年，是'高利率持续'路径下的反AI久期+实物瓶颈租金套利，需配置USD多头FX叠加并设置…", "summary": "2026-05-28 synthesize：把第01–05卡拼回组合（USD 100mm基币）。AI久期空头−15%、AI硬件多+10%、全球电网OEM+30%、亚洲OEM+10%、GOES上游+8%、CLF垂直一体+7%、自动化+5%、现金+25%。净Beta约+0.45、净久期约−1.5年（受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 2, "neutral": 0, "value": 52.4, "events": [ { "title": "上游材料定价权跃升驱动的是'窄口径供给瓶颈通胀'，alpha不在宏观贝塔而在'做多门神+做空无联动中游'的相对价值精选，杠铃必须从单腿物理层多头细化为五腿+A腿内部相对价值的精密结构。", "summary": "本卡综合卡01-05，把跨资产杠铃从'宏观四腿'细化为'五腿+A腿内部相对价值'：做多上游门神（GOES/无氧铜/PFAS替代化学品；CLF、FCX、宝钢、BASF等）+ 做多具有价格联动条款的电力重工龙头（GEV燃机、ABB电气化）+ 做空缺乏价格联动的中游温控/电力OEM（VRT等）+ 维持二…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力已替代硅成为 AI 算力的边际约束；正确的组合是做多电网+GOES+燃机物理层抱团 + 做空二线 AI 硅与 AI 基础设施久期 + 做空美债长端的跨资产杠铃，而非'纯芯片泡沫终…", "summary": "SpaceX 750 亿 IPO（2026-06-11 定价）叠加 Schiff《2026 能源成本公平与可靠性法案》100% 接入成本自付，加上美国变压器交期 4 年、GOES 价格自 2020 翻倍、燃机交期 5–7 年，已确认电力替代硅成为 AI 算力的边际约束。建议构建跨资产杠铃：A 腿做…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 2, "neutral": 1, "value": 56.8, "events": [ { "title": "我已完成对云厂商在能源通胀与设备成本上升下的资本开支效率及算力定价权的压力测试分析，并成功在工作空间中生成了完整的英文版报告、中文版报告以及机器可读的元数据手递手（handoff）文…", "summary": "我已完成对云厂商在能源通胀与设备成本上升下的资本开支效率及算力定价权的压力测试分析，并成功在工作空间中生成了完整的英文版报告、中文版报告以及机器可读的元数据手递手（handoff）文件。 核心结论摘�� (Core Synthesis) 本卡片站在 TMT行业分析师 的立场，对全球超大规模云计算三…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美欧贸易壁垒确实存在但被物理短缺约束，对 A 股电力设备出海是'硬锚有缝'而非'硬锚破裂'——card-05 配置框架基本成立，仅金盘需改为里程碑驱动节奏。", "summary": "美欧针对中国电力设备的政策工具栈已成型——EO 13920、Section 301、IRA 45X/Domestic Content、CBAM、FSR、NZIA、GOES 反规避——但北美 LPT 110–130 周交付期与欧美 OEM 订单簿 +40%+ 的物理短缺，制约了任何'全面替代中国货'…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 基建复利仍是 2026–2030 最稳定主线，但价值链重心正从'GPU+HBM+CoWoS'扩展到'GPU+HBM+CoWoS+LGIA+输电+PPA'；在保留 AI 半导体核…", "summary": "本卡把前四卡合成可执行的跨行业配置框架。在加入电力权利筛选器（并网/输电/灵活负荷/PPA 四维 0–12 分制）后，给出 12 个桶的旧/新权重对照：AI 半导体核心 40→30，AI 服务器合规优胜者（DELL/HPE）10→12，SMCI 5→1，Hyperscaler 20→22，CRWV…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 4, "neutral": 1, "value": 46.2, "events": [ { "title": "电网瓶颈真实但可移动：hyperscaler把边际GW送往ERCOT/海湾/东盟，许可改革与表后燃机/核电是旁道，关税让设备短缺更集中——结果不是美股公用事业广义牛市，而是少数OEM…", "summary": "从美/欧/日宏观与地缘政治视角对Card-05的'电力是新瓶颈'主线做压力测试。我不否定电网瓶颈，而是为它划界：(1) 边际AI GW正在通过Stargate ERCOT、Stargate UAE、沙特HUMAIN、马来西亚柔佛等通道离开美国队列，部分掏空美股公用事业敞口；(2) Trump 2.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-30的核心判断是：OEM仍是质量锚，但边际alpha在上游材料，存储提供业绩确认，BTM电力运营商提供战略稀缺和低beta稳定器。", "summary": "截至2026-05-30，本卡综合前六张卡：通电算力溢价仍成立，但新增股票风险不应继续单押已被充分发现的 OEM。建议“通电 AI 算力”股票子组合配置为上游材料35%、电力/冷却OEM 25%、存储与HBM 25%、BTM电力运营商15%。风格上，2026年中先以Quality承接已兑现的订单和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "As the Consumer Analyst (消费行业分析师) , I have completed the initial deep dive investigation in…", "summary": "As the Consumer Analyst (消费行业分析师) , I have completed the initial deep dive investigation into the Textile and Apparel Industry Chain (纺织服装产业链) . Core…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI 容量错位主线在材料端是自洽的——铜、GOES、锂、PFAS 替代冷却液、镍/不锈钢这五条投入要素本身就处于供给上限，正是它们的垂直供给曲线在物理上强制了 2026 的配额年，因…", "summary": "材料分析师从钢铁、化工、有色、电池材料四条投入要素端综合前五张卡的论断：2026 是 AI 算力的配额年而非顶点年。每 100 MW 净 IT 负载需 2.0 kt 铜、0.6 kt 取向硅钢 (GOES)、4.5 kt 结构钢、25 t LCE（BESS 用）及 30–50 t 介电液，五个市场…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 8, "neutral": 0, "value": 20.6, "events": [ { "title": "AI算力扩张的下一道硬约束是变压器、开关柜与取向硅钢（GOES），西方OEM新增产能要到2027–2028年才投运、GOES供需要到2030年才能缓解，这是供给侧未来两年的alpha…", "summary": "截至2026-05-31，本卡片支持AI算力瓶颈正从GPU迁移至变压器与开关柜，并在设备层量化约束。Wood Mackenzie 2025 Q2显示标准电力变压器交期128周、GSU 144周；Hitachi Energy欧洲≥100MVA大型电力变压器交期50–58个月。Eaton指定美国DC …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然 unhedged 极端大宗暴涨对电气设备龙头构成理论毛利压力，但由于变压器/开关柜2.5-5年的极长交付周期与工厂槽位稀缺，ETN与HUBB凭借绝对的价格议价权、合同价格指数化…", "summary": "截至2026年5月31日，本卡对铜、铝和高牌号电工钢（GOES）价格波动对电气设备巨头（ETN/HUBB）利润率的侵蚀压力进行定量压力测试。我们发现，虽然 unhedged 情况下 Severe Stress（铜价达$14,000/吨，GOES+40%）会侵蚀 margins 达247-517 b…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘政治与工业保护主义正将AI电网从中游环节‘脱钩化’，导致供给瓶颈行政化，显著加剧了算力扩张的物理墙刚性与Capex溢价。", "summary": "2026年6月调研显示，美中欧三方工业政策正合力‘加固’AI电网的物理瓶颈。美国通过《国防生产法》及EO 14156宣布能源紧急状态，对GOES及变压器实施Section 232高额关税并强化‘本土熔炼’规则，旨在保护Cleveland-Cliffs的单一供应垄断。中国则从2026年初起实施GOE…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "贸易保护主义与关税非但没有削弱ETN/HUBB的定价权，反而通过抬高进口替代品成本为其撑起了“利润率保护伞”，巩固了本土制造优势，但也确立了电网资本开支的高通胀属性。", "summary": "截至2026-05-31，本卡通过压力测试证实：美国针对GOES和成品变压器的激进关税（高达25%-50%）并未破坏本土电气龙头（ETN/HUBB）的定价权，反而形成了“关税保护伞”。本土企业通过加急提价转嫁了超过1亿美元的成本逆风，且受惠于进口替代品成本飙升。真正的物理约束仍是非关税组件（如欧洲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 2, "neutral": 1, "value": 61, "events": [ { "title": "[auto-handoff] Biotech Re-rating Thesis", "summary": "致 生物科技分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: eaf706e4-387b-4a6a-90f1-54f453dbd31e 结果摘要: Wrote handoff_response.md (15562 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "在 P/Watt 约束下，A 股应执行真杠铃——多头集中于绿色基荷 IPP 与特高压/变压器/GOES 制造，减配没有电力预定的投机性算力开发商。", "summary": "P/Watt 已取代 P/FLOP 成为 AI 算力扩张的边际约束。云巨头 2026 年资本开支 6,600–7,000 亿美元，三分之二投向供电而非芯片；大型变压器交付周期延至 128 周到 5 年。由此带来的 A 股真杠铃：多头一端是绿色基荷 IPP（长电、广核、核电、三峡能源），重估为带 A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "地缘贸易壁垒升级与美国原产地规则穿透（Section 232 及新 Section 301）将实质性阻断中国 Hi-B 硅钢通过第三国变压器的间接外溢出口通道，分流吨量将与国内扩产周…", "summary": "截至2026-06-02，本报告对前三张卡片关于中国 Hi-B 硅钢及变压器‘美国外溢订单’溢价逻辑进行了地缘政策压力测试。研究表明，该外溢逻辑面临高度的地缘贸易壁垒 fragility。美国对变压器及铁芯原产地规则的收紧（如 Section 232 '熔炼与浇铸' 穿透穿底要求以及 USMCA …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 4, "neutral": 0, "value": 47.4, "events": [ { "title": "卡片04的长周期订单证据成立，但HALEU瓶颈、FERC表后否决、GOES单源、Section 232关税与中国稀土域外管辖共同抬高政策β，溢价定价权叙事应被降级而非剔除。", "summary": "截至2026-06-02，本卡对卡片04的长周期支持结论进行压力测试，识别五条具体政策/地缘风险：(1) HALEU燃料瓶颈——Centrus示范级月均约45kg、商业产能2027-2028年才到位，使FOAK SMR大概率延期12-24个月；(2) NRC许可摇摆，EO 14300压缩至18个月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the geopolitical stress test and risk analysis for Research note 5 of the …", "summary": "I have completed the geopolitical stress test and risk analysis for Research note 5 of the research thread portal session as a Policy Analyst. Based …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Compute premium正沿AI电力瓶颈从需求栈再分配到物理供给栈：midstream、燃气轮机OEM、棕地核电IPP、电网设备和铀燃料承接租金；超大规模云因capex/收入失…", "summary": "本卡作为大类资产配置师，综合卡01–03给出跨资产配置地图：在2026年四大云厂7,000–7,250亿美元capex、其中60%+用于电力而非芯片的背景下，compute premium正从需求栈（超大规模云+高估值AI软件）迁移到物理供给栈。第一层承接：管道firm transport（KMI…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI-电力跨行业组合应以三筛选器分桶配置：3/3命中的美国wires-first IOU拿核心30%，2/3命中的输配电EPC、电网设备、中国UHV各15%卫星仓位，AI-infra…", "summary": "截至2026-06-03，本卡综合cards 01-05，用'客户先付风险(CR)、监管可回收(RR)、节点受约束(NC)'三筛选器对AI-电力链跨行业组合定权重。3/3命中（在数据中心集群拥有获批大负荷电价的美国wires-first IOU，如AEP、Dominion、Southern、Duk…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 1, "risks": 7, "neutral": 0, "value": 38.8, "events": [ { "title": "电网容量与电力设备交付的物理刚性约束已成为 AI 算力 Capex 转化为产能的核心瓶颈，定价权与超额利润将沿‘芯片→变压器/燃气轮机/核电 PPA→GOES 钢’向下游电力基础设施…", "summary": "本报告从能源基础设施物理层支持 TMT 卡观察——AI 算力订单的收入转化率被‘平滑化’，根因是电网容量、大型电力变压器 (LPT) 与重型燃气轮机的结构性刚性瓶颈。截至 2026-06-03，美国数据中心 2030 年负荷展望升至 580 TWh，并网队列存量 2,600 GW，中位等待 53 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 06 synthesizes the prior five into a four bucket AI allocation framework (Hyp…", "summary": "Research note 06 synthesizes the prior five into a four bucket AI allocation framework (Hyperscaler ROI core / locked in power / grid OEM / applicati…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Section 6418转让市场和银行bridge capacity足以承接AES量级压力供给，但边际FID的关键阈值是ITC售价跌破约$0.90，真正抑制区间是$0.85–0.87。", "summary": "金融市场可以吸收AES类发起人退出tax equity后的大部分管线，但会通过质量分层而非无差别承接来出清。2025年转让市场约$420亿、tax equity投资约$360亿，足以承接约$45–75亿压力供给；但非投资级卖方会失去约2–5美分/美元价格。ITC售价跌至$0.88–0.90时边际项…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "USMCA原产地规则（ROO）的收紧将使依赖进口组件的墨西哥电力设备组装产能对冲逻辑彻底失效，而沙特产能也面临美国 EAR/FDPR 针对 AI 基础设施的次级制裁风险。", "summary": "2026-06-03：本卡从政策监管视角压力测试出海产能对冲逻辑。USMCA 2026 联合审议实质性收紧了原产地规则，强制将“熔炼和浇注”标准延伸至变压器等核心电力设备。由于中国控制多数 GOES 产能，单纯依靠“组件进口+墨西哥组装”的产能将无法通过 RVC 测试，对冲逻辑彻底破产。同时，沙特…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 3, "neutral": 0, "value": 56.2, "events": [ { "title": "美股大厂利润受损与NDX估值收缩将导致A股高拥挤AI硬件链面临-34.3%的估值与盈利双重杀跌压力，并触发74.5%概率的两融去杠杆踩踏；而绝对利差高达+249 bps的A股红利资产…", "summary": "基于2026-06-04，本报告量化评估了美股NDX重估对A股AI算力链（CPO、PCB）及红利资产的溢出与共振效应。测算表明，A股AI硬件等权篮子交易于32.5x的2026E P/E，其估值对美股弹性系数为1.35。美股大厂因PPA及电网延迟导致的利润率受损及NDX 20%的估值压缩，将使A股算…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策干预（包括DPA资金、关税调整和并网改革）在12-18个月内无法打破‘通电滞后’的物理时延瓶颈，保护主义关税和能效新规反而延长了供应链交期并抬升了电网建造成本。", "summary": "本报告对旨在打破‘通电滞后’的电网政策干预进行了压力测试。结果表明政策无法在12-18个月窗口内起效。DPA资金扩产的物理建设周期长达33个月（受制于21个月设备交期及6个月技能培训），产能迟至2029年释放。4月新加征的Section 232关税按成品全额征收，使大型电力变压器进口成本大涨7.5…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "数据中心投运延期导致的刚性PPA及传输电费沉没成本将直接折损科技大厂约100-150 bps的营业利润率并拖累核心增速，推动NDX远期估值从28.5x向22.8x下行重估，并加速全市…", "summary": "大厂延期投运数据中心导致刚性“照付不议”PPA电费转化为纯沉没运营开支（OpEx）。测算表明，2.0 GW延期导致微软年度利润率降1.08%，EPS拖累-$0.167/股；1.5 GW延期导致亚马逊利润率降1.47%，EPS拖累-$0.076/股。若加算力延迟机会成本，合并EPS拖累达-$0.29…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 2, "neutral": 1, "value": 66.6, "events": [ { "title": "电力确是硬上限但约束在电网设备与并网而非 MWh，alpha 集中在 OEM 与受监管 T&D，搁浅风险锁定 2025–26 批发电源项目融资 vintage。", "summary": "电力是真实但时间有限的 AI 物理瓶颈，资本可压缩约 30–40% 延期但不能消除：约束子层是电网设备 + 并网（变压器交期 3.5–5 年、美国并网排队约 2.6 TW、PJM 群组 ~286 GW），而非装机 MWh 本身 [S1][S2][S3][S11]。最干净的可定价 alpha 在物理…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Impact of 2026 allowance allocation on compliance demand", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 政策分析师。 模式: adhoc 会话: 0dac1524-9f8b-48b7-9b9a-d3b1a3af9bf3 结果摘要: Wrote handoff_response.md (10747 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI基础设施篮子应从芯片稀缺溢价，转向算力核心、电力设备收入可见度和GOES/铜材料对冲的三层配置。", "summary": "截至2026-06-05，我支持AI基础设施瓶颈从芯片向电力设备、再向GOES/铜迁移的主线，但建议把组合从芯片主导转为“算力核心+设备超配+材料对冲”。建议权重为30%算力/HBM/先进封装，28%电力设备/冷却/微电网，17%铜与GOES等材料，12%合约电力/公用事业，8% API-firs…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 4, "neutral": 1, "value": 56, "events": [ { "title": "电网设备 OEM alpha 仍成立，但核心不是无限需求，而是 2026-2027 年订单积压可见度与 2027-2028 年前仍无法快速解除的材料/产能瓶颈。", "summary": "2026-06-06：我支持 Research note 02 的方向判断，但将其限定为“订单质量与交付稀缺”交易。Hitachi Energy FY2025 末积压 579 亿美元、Siemens Energy Grid Technologies 积压 490 亿欧元、GE Vernova El…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "铜的精炼端紧而不破，但 GOES（特别是高磁感）与绝缘材料（Nomex、纤维素压板、浇注环氧）是 2026-2028 结构性硬缺口，将使电网 OEM 实际交付较名义扩产折损 15-2…", "summary": "2026-06-06：支持 Research note 03 关于电网 OEM 交付受材料瓶颈刚性约束的判断。铜总量缺口约 0.5%（价格机制可消化），但电工级铜杆 CTC/HV 电缆交期已由 16 周延至 32-40 周；取向硅钢 GOES 2026 缺口 9%、高磁感 HiB 缺口 15%，新…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the stress test analysis of China's State Grid 2026 investment plan and it…", "summary": "I have completed the stress test analysis of China's State Grid 2026 investment plan and its impact on copper demand and the systemic drawdown floor …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "现行容量电价仅能覆盖煤电 30–50% 固定成本且与燃料价格脱钩，无法独立兜底极端气候叠加燃料飙升下的现金流尾部风险；对纯水电资产更是政策真空，Research note 09 的尾…", "summary": "截至 2026-06-07，本卡从政策机制层面对 Research note 09 关于火电/水电防御底仓存在非线性尾部风险的判断进行压力测试。发改价格〔2023〕1501 号确立的煤电容量电价仅覆盖固定成本 30%（2026 年起 10 省份升至 50%），基准 330 元/kW·年锁定后不与煤…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 2, "neutral": 0, "value": 66.4, "events": [ { "title": "[auto-handoff] 取向硅钢(GOES)产能瓶颈及对特高压变压器造价传导", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 材料行业分析师。 模式: adhoc 会话: e911bea4-9058-473c-a5ef-19ab52df0e02 结果摘要: I will first run the 系统日期锚 command to anchor today's date. …", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "政策可以软化AI电力设备供应墙，但无法在2026年用贸易壁垒或补贴绕过GOES和大型变压器的物理扩产周期。", "summary": "截至2026-06-07，本卡压力测试上一卡“GOES与变压器构成物理供应墙”的结论。判断是：关税、出口管制和僵硬国产化规则不能绕过冶金、专用设备、测试和工厂建设周期，反而可能加剧2026年稀缺；真正有用的是盟友采购、关税与BABA豁免、规格标准化和过渡库存。产业补贴必要但滞后，Cleveland…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 0, "neutral": 2, "value": 85.8, "events": [ { "title": "I have completed the analysis for Research note 07 as the TMT Analyst, supporting the thesi…", "summary": "I have completed the analysis for Research note 07 as the TMT Analyst, supporting the thesis that low price platform subsidy strategies structurally …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "科技巨头转向“电表后”直供电对公用事业盈利与 PPA 定价的重塑 日期：2026 06 08 核心观点 / Thesis 面对长达 5 7 年的电网排队与 36 48 个月的变压器（…", "summary": "科技巨头转向“电表后”直供电对公用事业盈利与 PPA 定价的重塑 日期：2026 06 08 核心观点 / Thesis 面对长达 5 7 年的电网排队与 36 48 个月的变压器（LPT）交付延误，科技巨头正激进转向“电表后”（BTM）的核电同址与天然气微电网方案，这使 PPA 定价核心从“绿色…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 1, "neutral": 0, "value": 75.2, "events": [ { "title": "[auto-handoff] 铜与GOES硅钢双重价格冲击下国内输配电设备商的毛利率压力测算", "summary": "致 主题研究员： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 338e4f11-04d4-4a21-b348-6a280a7fa27b 结果摘要: # Step −1 — Anchor today's date BEFORE you start Run this in …", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 1, "risks": 1, "neutral": 0, "value": 84.6, "events": [ { "title": "All three deliverables written and verified (JSON validates). Research note 04 summary (pol…", "summary": "All three deliverables written and verified (JSON validates). Research note 04 summary (policy analyst, stance: SUPPORT with refinement): I support R…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "“离网自治”模式面临美欧“安全与电力绑定”和关税/碳税的精准拦截，主权AI项目在放弃中国电力设备以获取AI芯片（交付延迟36-48个月）与使用中国设备（芯片断供）之间面临难以调和的结…", "summary": "本卡对AI“离网自治”模式进行压力测试。美欧正通过“安全与电力绑定”将AI芯片出口与排除中国设备挂钩（如阿联酋G42项目）。同时收紧FEOC与301关税以拦截含中国取向硅钢（GOES）的第三方变压器；欧盟也于2026年实施CBAM并禁用中国逆变器。开发者被迫在“使用中国设备但无芯片”与“脱钩但交付…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 0, "value": 78, "events": [ { "title": "相比直接的整机关税，核心部件的原产地关税与系统性融资限制对A股出海设备商构成更深层的毛利挤压，迫使行业出海模式向深度供应链本地化或非美欧市场切换分化。", "summary": "本卡片对美欧电网设备及取向硅钢（GOES）关税壁垒进行了政策压力测试。研究指出，美国Section 232新规（依据全申报海关申报价值征收50%铁芯关税）与欧盟GOES保障措施调查，表明西方政策重点已转向阻断“第三国绕道”与核心组件本地化强制约束。这使A股出海设备商面临严重分化，中度和重度压力下出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 3, "neutral": 0, "value": 67.4, "events": [ { "title": "去全球化导致美欧电网设备及原材料支付高额地缘溢价，并通过公用事业费率基数和新能源购电协议（PPA）向终端电价传导，正在欧美引发系统性的‘绿色通胀’。", "summary": "截至2026-06-13，本卡压力测试表明，关税及安全限制使美国取向硅钢（GOES）及成品变压器面临422%和80%-150%的地缘溢价。这每年为美国公用事业费率基数注入150亿-200亿美元通胀，拉高零售电费1.5%-2.0%。同时，并网延迟和设备溢价推动美国太阳能PPA价格同比大涨13%至$6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在2026年下半年全球公用事业Capex年增29%的超级周期中，通过超配受俄亥俄DCT关税保护的AEP以对冲美国监管限价风险，同时超配中国可调度电源与电网设备以套利国产低效芯片带来的…", "summary": "针对2026年6月13日的宏观基准，本报告对中美公用事业在AI基建超级周期下的套利配置进行了综合。我们设计了由50%美国公用事业（超配受DCT保护的AEP及锁价PPA的CEG，对冲电价通胀抵制监管风险）与50%中国智能电网与可调度电源（超配思源、金盘、湖北能源、福能，套利国产低能效GPU带来的负荷…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "非晶变压器供应链瓶颈、关税摩擦及OBBBA法案补贴悬崖，共同抬升公用事业名义Rate Base并推高AI数据中心PPA电价与附加容量电费，将通过并网延期与运营成本飙升严重侵蚀云厂商A…", "summary": "【工作日期：2026-06-12】压力测试表明，非晶变压器高关税与物理缺口虽然名义上推升公用事业资产折旧基数（Rate Base），但带来长达2-4.5年转固延迟并产生大额现金利息流出，严重恶化盈利质量并引发费率摩擦。此外，并网延期导致新能源项目错失OBBBA法案开工红线从而丧失30%税收补贴，迫…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 4, "neutral": 0, "value": 62.8, "events": [ { "title": "AI扩容的核心瓶颈已全面转移至变压器、变电站节点与基荷电力等物理层，这要求权益市场削减短期AI软件收入预期，重估发电商、电网OEM及相关核心材料的结构性成长溢价。", "summary": "截至2026-06-14，本卡片完成了AI基础设施“电网墙”向跨资产配置的最终战略映射。综合变压器交期、并网队列与基荷短缺三大物理断点，我们建议进行结构性轮动：超配拥有24/7核/气基荷定价权的独立发电商（如Vistra、Constellation）与具有极强订单能见度的电气设备OEM（Eaton…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Higher-for-longer不会必然叫停AI capex，但会把市场关注点从“能否花钱”转向“每个增量GPU集群能否在折旧和电力成本吞噬定价窗口前跨过更高ROI门槛”。", "summary": "截至2026-06-14，本卡对卡片1的高利率与美联储战略模糊情景做压力测试：AI基础设施capex的关键脆弱点不是大型云厂商短期无钱可投，而是折现率、GPU短寿命折旧、电力并网与PPA成本共同抬高项目资本回收门槛。以三分之二6年期算力资产、三分之一15年期数据中心资产测算，200bp WACC冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the stress test analysis on the risk of export controls and technology dec…", "summary": "I have completed the stress test analysis on the risk of export controls and technology decoupling in the US China grid supply chain for Research not…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力供应侧的真正瓶颈不在发电产能而在并网、变压器和输电许可这三个不能用资本开支解决的物理/政策约束，因此公用事业AI Beta只属于'已并网+锁定燃料+受监管rate-base'的子…", "summary": "压力测试结论：供应侧无法按超大规模厂商时间表匹配AI算力的指数级电力需求。真正约束是并网排队（PJM队列286 GW、平均等待4.3年）、高压变压器交付（120–210周）和输电选址（NEPA 7–10年），而非发电能力本身。卡片03运营商论点成立但需收窄：稀缺租金归属于已并网的燃机/核电+锁定燃…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 4, "neutral": 1, "value": 67.2, "events": [ { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] AI电力基础设施瓶颈向上游物理期权（精炼铜/取向硅钢）传导评估", "summary": "致 日报总编： 协同任务已完成。执行方为 主题研究员。 模式: adhoc 会话: bca64921-2ae4-41c5-bfcf-48378f7e6da6 结果摘要: Wrote handoff_response.md (13721 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力链条的策略超额收益不在需求标题，而在“客户不能等的设备、设备商不能替代的GOES、以及客户能融资的AI电力合同”。", "summary": "截至2026-06-16，本卡综合前九张卡片，结论是全球南方只能部分对冲OECD利润侵蚀，配置应从“买海外订单”升级为“买稀缺性与可融资现金流”。建议基准权重为50% A/H电网关键设备、25% GOES与电工钢、20% AI电力基础设施、5%流动性或战术对冲；若满仓则为52.6%/26.3%/2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "公用事业瓶颈的跨行业配置含义：电力设备稀缺、受监管公用事业资产基数扩张与AI基础设施估值折价", "summary": "若firm power而非GPU是真正瓶颈，稀缺租金应归于占据约束的一方。做多三层：①订单售罄的长周期电力设备（GE Vernova GEV燃气轮机在手约100GW、档期至2028-2030[S6]；变压器交期达4年[S1][S2]）；②商业核电IPP（CEG约17-21x、VST约13x，签20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 0, "neutral": 1, "value": 92.6, "events": [ { "title": "广州发展应作为小仓位期权腿保留、从高确信AI电力核心中剔除；除非出现具名算力园区PPA，否则不应继续承担核心超配角色。", "summary": "截至2026-06-16，广州发展(600098.SH)不应再放入高确信AI电力核心，但可作为小期权腿保留。建议100单位公用事业/AI电力组合中默认3单位、上限5单位、核心权重0。依据是4.84GW火电约7.986亿元毛容量电费暴露，同时2025扣非分红率约116.4%、2026-06-16扣非…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 1, "neutral": 0, "value": 85, "events": [ { "title": "电力瓶颈是 AI capex 的中期执行上限，而不是 2026-06-17 当下的叙事破裂触发点。", "summary": "2026-06-17 的压力测试结论：电力、电网与数据中心容量约束是真实且正在制度化的瓶颈，但尚不足以在 2026-06-16/17 FOMC 窗口内击穿 AI capex 叙事。LBNL、IEA、BloombergNEF、PJM 和 NERC 数据显示需求规模、在建容量和并网可靠性压力都在上升；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 4, "neutral": 0, "value": 68.4, "events": [ { "title": "将firm power rights x tokens per watt框架转化为AI链跨行业配置", "summary": "本卡综合整条线索，把firm power rights × tokens per watt 统一为一个AI园区估值恒等式：EV ≈ firm MW × tokens per watt × 每token收入 × 利用率 ÷ 折现。配置原则是稀缺租金归于供给最缺乏弹性的环节。据此给出四仓位篮子：电力设…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基础设施交易的跨行业风险预算与风格轮动框架", "summary": "综合并落地 Card-01 的方向判断。核心洞见：电力稀缺链条看似分散（公用事业+工业+材料+核能），实则是同一个 hyperscaler capex 因子，回撤时内部相关性飙向1。故按风险贡献而非名义权重预算，拆成四个子篮子：A电网输配电（压舱石）、B现场发电（核心-卫星）、C核燃料/SMR（高…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-2028年AI算力链组合应从芯片优先转为瓶颈优先：超配能释放通电约束的电气设备和已投运电力资产，保留芯片作为去瓶颈期权。", "summary": "截至2026-06-18，本卡综合前序结论：通电围墙仍是AI算力链主约束，芯片与先进封装提升tokens/MW但不能替代并网、变压器和确定基荷电力。建议2026-2028主题组合按电气设备/电网硬件35%、先进封装与机架供电25%、已投运基荷电力25%、芯片供应商15%配置，并用并网交期、LPT交…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Review AI-theme concentration cap", "summary": "致 首席风控官： 协同任务已完成。执行方为 大类资产配置师。 模式: adhoc 会话: 62ea74e0-df20-4af1-bb5b-1aff3381705c 结果摘要: Wrote handoff_response.md (9009 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 3, "neutral": 0, "value": 75.8, "events": [ { "title": "电网稀缺已经从主题交易升级为大类资产配置因子：低配无法按时通电的GPU/CPO硬件需求，超配持有变压器、GOES、中国UHV和红利电力现金流的资产。", "summary": "截至2026-06-19，本卡片把前序“GPU/CPO减配、电网链上修、GOES/变压器瓶颈、美国可再生递延、中国UHV免疫”整合成100单位跨资产组合：18%中国UHV/电网自动化、14%一体化变压器与GOES、16%红利电力、6%铜、18%中国久期、7%黄金、9%现金、6%硬件beta、2%美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在变压器交期拉长至60个月与电网并网长达5-7年的刚性瓶颈下，超前采购的算力设备面临严重闲置减值风险，而过度透支AI电力溢价的核电与气电IPP高估值泡沫将在2026年底至2027年中…", "summary": "报告生成于2026-06-19。分析了美国高压变压器交期拉长至60个月对AI数据中心并网的严重拖累（并网周期增至5-7年，可致提前采购的GPU资产折旧减值）。指出在FERC表后监管限制与项目延期现实下，CEG和VST等独立发电厂（IPP）的戴维斯双击高估值泡沫将在2026年底至2027年中面临破裂…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "A股AI电网capex链条配置优先级：上游材料瓶颈 > 电网设备OEM > 出口龙头 > 资产型公用事业。", "summary": "截至2026-06-20，本卡综合前五张卡片，给出AI电网资本开支链条在A股的优先级：上游高壁垒材料瓶颈（高牌号GOES、CTC铜、绝缘材料、酯类介质液）>电网设备OEM（特高压主设备、数字化二次、SVG）>出口龙头（变压器/开关DM替代窗口）>资产基数增长型公用事业。前两档承担弹性主战场，第三档…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 4, "neutral": 0, "value": 71.2, "events": [ { "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "summary": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源右尾风险会重估和重配AI基础设施capex，但不会在基准情形下让资产负债表最强的hyperscalers停止扩建。", "summary": "截至2026-06-21，本卡片对能源右尾风险的结论是：它不足以阻止最大型hyperscalers扩建AI智算中心，但会显著改变长期电力预算和电网容量布局。IEA预计全球数据中心用电从2025年约485 TWh升至2030年约950 TWh；LBNL预计美国数据中心到2028年可能占总用电6.7-…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源波动主要通过利润率波动和 CAPEX hurdle rate 传导，而不是直接终止先进制程扩产；优势集群与边际 greenfield 项目的回报差会扩大。", "summary": "截至2026-06-21，本卡压力测试能源与电力波动对先进制程 fab 和先进封装的成本传导。结论是：电价不是最大成本项，但在连续运行、EUV、洁净室、超纯水和良率风险下是高 beta 成本项。以 TSMC 2025 年数据自测算，电力约占制造营业成本 8.0% [自测算]；10% 电价冲击约拖累…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源波动 regime 下的跨行业配置与 A股/科创硬科技定价", "summary": "本卡综合 Research note 01–05：电力与电网接入已成为整条 AI 价值链的约束性稀缺投入。在布伦特回落至约80美元、发达市场同步鹰派的 regime 下，正确做法不是削减硬科技，而是沿杠铃重塑风险预算。提出五桶矩阵：超配电力/电网赋能者（国家电网2026–30约4万亿、核电10机组…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 0, "value": 84.6, "events": [ { "title": "更久维持高位、电力成本推动、AI资本支出搁浅时点与芯片库存尾部下的多资产组合定位", "summary": "本卡以asset-allocator视角综合四卡链条，落到一套定位框架。结论：四卡为「一个体制」，正确组合是杠铃——(1)做多物理瓶颈（电力/电网设备、燃气轮机OEM、独立电力/核电PPA），无论GPU是否按时通电都收多年在手订单年金，且做多令美联储紧缩的成本推动本身，是自我融资宏观对冲；(2)低…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电力瓶颈引发的不是避险，而是向电力价值链的成长型风格轮动；公用事业应从类债股息估值重塑为分部估值＋成长期权模型——美股按PPA与容量租金、A股按电网资本开支可见度与水核稳态分别定…", "summary": "支持卡。电力成为AI硬约束后，美股与A股都在发生向电力价值链的风格轮动，但这是AI二阶导成长交易而非避险回流。证据：PJM容量价从约29→269.92→329.17美元/MW·日(触FERC上限)；2025年美股公用事业涨约16%、跑赢标普约493bp，2026年初至今为标普第三强板块。机制不同：…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 1, "neutral": 0, "value": 92, "events": [ { "title": "AI基建配置应从“追逐算力设备增量”切换为“持有可交付电力稀缺性”，但保留CPO核心仓位，因为CPO仍是降低AI网络功耗的必要技术。", "summary": "截至2026-06-24，本卡综合前序结论：AI算力瓶颈已从GPU转向已通电MW，组合应从高拥挤CPO单边beta转向“变压器OEM+中国出海设备+GOES/铜材料+CPO核心赢家”的哑铃。100单位AI基建权益袖基准配置为韩日及全球变压器OEM 30、中国出海设备22、上游材料18、CPO/光模…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 5, "neutral": 3, "value": 78.4, "events": [ { "title": "AI算力集群的电力保障能力应按已锁定的变压器与中压设备链条打分，而不是按土地、PPA或名义兆瓦打分。", "summary": "2026-06-25，本卡对上一张 AI 基础设施结论做压力测试并支持其主线：区域 AI 算力集群的真正瓶颈不是单一铜/GOES价格，而是主变、中压开关柜、断路器、配电变压器、施工队与费率审批构成的完整送电链。美国2025年电力变压器与配电变压器供给缺口分别为30%和10%；北弗吉尼亚等核心市场已…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "发电侧与配电侧在2026年正陷入‘既无路（电网排队超4年及变压器稀缺）也无车（燃机交期5-7年及SMR未商业化）’的双端挤压困境，共址直供监管红线的收紧彻底消除捷径，导致算力资本开支…", "summary": "截至2026年6月25日，配电与电网设备交期拉长使得发电侧面临严重并网瓶颈。大型电力及GSU变压器交期达18-48个月，GOES硅钢供给偏紧，并网排队中位数超4年。FERC否决核电表后直供迫使亚马逊等转向表前PPA，面临同样电力瓶颈。同时高确定性燃气轮机交期达5-7年，SMR在2030年前难以放量…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "持续能源通胀会压缩AI基础设施回报率，但真正改变Hyperscaler扩建节奏的是可交付电力的稀缺，而不是电费单项上涨。", "summary": "截至2026-06-25，本卡压力测试能源通胀对AI基础设施的传导。结论是：电价上升本身不足以触发Hyperscaler AI capex断崖；更关键的是并网、容量市场、PPA和电网设备交付约束会重排项目优先级。Microsoft、Alphabet、Meta、Amazon仍合计指向约6950亿-7…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力扩张下的电力供给与电网瓶颈风险", "summary": "截至2026-06-24，本卡对上一卡做压力测试：电力与电网接入正在成为AI硬件交付的真实节奏约束，尤其在美国数据中心负荷集中的电网区域，但当前证据更支持“区域通电延期与capex成本上升”，而不是“800G/1.6T需求被证伪”。IEA、Berkeley Lab、NERC、PJM与Monitor…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 2, "neutral": 1, "value": 91.8, "events": [ { "title": "虽然物理电网排队与供应链原材料价格冲击显著抬高了局部建设阻力，但云厂商通过绿电PPA、高密度液冷及转向定制ASIC等结构性调整，能够完全消纳成本溢价，不构成算力扩张的绝对瓶颈。", "summary": "本卡对电网并网延迟、变气压原材料（铜/GOES）价格波动和本地化GPU封装良率冲击进行了量化压力测试。结果表明，虽然并网延迟和原材料成本增加会显著压缩项目NPV（5年延迟将使NPV降低79.99%），但云厂商可以通过迁移西部枢纽自建绿电、液冷优化和系统级混合算力架构，消纳国产芯片和能源价格溢价，并…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "做多迁移的稀缺租金应以电网硬件和材料为锚、以先进封装和定制ASIC为弹性，同时用能源空头、商用GPU相对空头和暗硅期权保护组合。", "summary": "截至2026-06-25，本卡综合前序链条：霍尔木兹能源溢价正在消退，稀缺租金迁移到铜/GOES、电网设备、先进封装和定制ASIC。建议用100.0名义多头配置15.0%材料、30.0%电网硬件、20.0%先进封装、15.0%定制ASIC/AI网络、10.0%电力接入和10.0%流动性，同时用35…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "天然气PPA/BYOP是AI数据中心2030年前的现实桥梁，但A股更应买设备稀缺溢价，而不是泛化买火电或燃气beta。", "summary": "截至2026-06-25，本卡判断：SMR在2030年前难以成为AI数据中心主供电路径，天然气PPA与BYOP将成为2026-2030年的桥接方案，但核心套利不是低气价，而是交付时间、可调度容量、并网稀缺、碳成本和高压设备交期。海外尤其美国气源盆地更适合气电PPA和表后微电网；中国路径更受80%绿…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 4, "neutral": 0, "value": 78.2, "events": [ { "title": "电力设备杠铃可以落地为六只A股核心篮子，但必须以流动性、两融拥挤和业绩质量仪表盘强制约束，触发信号时用IC/IM/IF与ETF期权对冲并削减设备腿。", "summary": "截至2026-06-27，本卡片支持前序电力设备杠铃，但将其落实为严格的A股执行篮子：2.5% NAV电力设备由首钢股份/望变电气表达GOES稀缺、思源电气/金盘科技表达出口Alpha，2.0% NAV公用事业由长江电力/国投电力压舱。对冲工具以IC/IM/IF和沪深300、上证50、中证500 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "并网延迟的变压器库存积压是可逆的时点拖累(金盘-8%~-11%、思源<-3%)，被国网4万亿内需托底，应在超配出口变压器框架下偏思源、轻金盘并叠加配对与公用事业杠铃对冲。", "summary": "并网延迟导致的已完工未交付变压器库存积压，是营运资金/收入确认的时点事件而非需求毁灭——订单仍排产至2027年在手backlog内。对金盘科技(688676.SH)拖累真实但有界：北美AIDC相关收入是最高Beta切片，Q2/Q3净利约-8%至-11%、伴随存货与FCF占用，属逢低布局而非证伪；对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "基于电力电网物理瓶颈（变压器交期120-210周）导致资本开支递延800亿-1100亿美元，高PE半导体叙事久期面临压缩，A股大类资产配置应低配半导体至10%-13%[自测算]，并重…", "summary": "本报告针对电力电网硬约束与资本开支递延风险，对A股资产进行大类风格与板块相对配置量化重构。压力测试表明，变压器产能受限（交期120-210周）及并网排队（7-10年）导致2026-2027年云厂商约800亿-1100亿美元资本开支递延[S8]，从而证伪了半导体高PE板块（CSI半导PE TTM约1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 1, "value": 88.6, "events": [ { "title": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "summary": "截至2026-06-27，本卡片综合前序结论：电网/变压器瓶颈会延后并改道AI部署，但不会摧毁2028年前光模块出口龙头业绩能见度；真正尾部风险是2027年云厂商ROIC压缩。建议A股AI物理瓶颈组合使用40/32/12/8/8结构：变压器/电网设备、光模块、电力现金流锚、铜/GOES等输入端对冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Summary of Work We have completed the analysis for Research note 04/10 as the Materials Ana…", "summary": "Summary of Work We have completed the analysis for Research note 04/10 as the Materials Analyst (材料行业分析师) , adopting a Support (支撑) stance. Our repor…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "滞胀确认与大类资产重构：商品通胀刚性、美联储“加息在桌”对大类资产风险平价模型（Risk Parity）的极限压力测试。", "summary": "本卡支持第01–07卡逻辑链并闭环回组合构建。供给冲击型通胀给铜（2026年缺口约33万吨，2030年扩至150–220万吨）与被贸易墙隔离的GOES（中国约56%产能）一个硬性供给底，使商品子仓下行半方差较以往周期压缩约30–40%——保留通胀上行、削弱增长下行。但这救不了风险平价：供给冲击把股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球商品通胀刚性与风险平价失效下的A股风格轮动：上游资源/红利价值 vs 成长，及两融与三季度解禁的交互", "summary": "全球'上游资源+红利价值'赢家只主导了沪深两市的一半：A股把组合拆开——资源赢（有色年初至今+30%以上、行业冠军[S4]），骑在第06-07卡铜/GOES缺口上；红利价值落后（中证红利一季度约+4%[S5]），因边际资金追逐成长逼空（科创50约+40%、创业板约+30%、6-22创新高[S4]）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 6, "neutral": 0, "value": 69, "events": [ { "title": "电网与变压器瓶颈对AI算力扩张上限的实质性约束评估", "summary": "对超大规模云厂商以自建核能与微电网对冲变压器缺口的乐观叙事做压力测试，结论是否定的：在最具约束力的2026–2028窗口内无法对冲，之后也仅部分、不均衡。新建SMR最早2030–2031商运（NRC认证+许可+建设各3–5年），完全错过窗口；既有核电重启（三里岛/Crane 835MW、AWS-S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼及普钢加工环节将面临严重的利润率…", "summary": "本卡通过技术壁垒与自给率对A股AI原材料敞口进行了区分：高端取向硅钢方面，宝钢股份凭借规模与技术优势展现出强Alpha属性，而首钢股份受普钢周期Beta拖累较大；有色铜方面，在2026年长单TC/RC为零的极端低位下，矿权自给率超65%的紫金矿业可完全捕获矿端溢价（EBITDA价格敏感度约7.5%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-06-28，最优跨资产表达是“久期+稀缺股票”的杠铃，铜只做长周期期权，因为近端精炼过剩削弱了直接商品敞口的性价比，而GOES稀缺主要只能通过精选股票承接。", "summary": "截至2026-06-28，本卡综合八卡链条：物理瓶颈应表达为“久期对冲下的稀缺alpha”，而不是单一铜多头。铜具备2030-2035结构性稀缺，但ICSG相关数据仍指向2026/2027精炼铜过剩约9.6万吨/37.7万吨[S4]；GOES更确定、更难对冲，但可交易性集中在Cleveland-C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关键矿产与设备出口管制风险对A股高端环节溢价逻辑的冲击评估", "summary": "压力测试出口管制威胁。结论：把镓/锗/锑/稀土剧本套到宝钢与紫金上是双向错配，不反转 Research note 06 基准论断。铜/紫金：中国是全球最大铜进口国（2025 精矿进口 30.31Mt，国内矿仅约 1.85Mt，依存约 87%），出口武器结构上不存在；紫金约 50% EBITDA 来…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 2, "neutral": 1, "value": 96, "events": [ { "title": "[auto-handoff] Impact of global capacity pricing spikes on domestic transformer export marg…", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 324722b5-2f7b-4b91-aec7-ac5429f32a39 结果摘要: Wrote handoff_response.md (11765 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 先进制程星载 OBP 芯片良率瓶颈与算力/通信设备材料供应链", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 81384248-8464-45e1-8194-06f6bed14dff 结果摘要: Wrote handoff_response.md (10343 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-d1c60d5af6/", "networkHref": "/investor/research-network/#thesis-d1c60d5af6" }, { "id": "thesis-e26ef69f73", "source": "tracked_thesis", "title": "风险核查与证据质量是主题交易的必要过滤器", "titleEn": "Risk verification and evidence quality are necessary filters for theme trades", "lane": "组合风控/证据质量", "laneEn": "Portfolio risk and evidence quality", "laneHref": "/investor/lanes/portfolio-risk-and-evidence-quality/", "investableFocus": "组合风控/证据质量：CBOE, CME, VIRT, IBKR, SH", "stockUniverse": [ { "ticker": "CBOE", "name": "Cboe Global Markets", "market": "US", "role": "波动率与期权基础设施", "roleEn": "波动率与期权基础设施", "bucket": "core", "href": "/investor/stocks/CBOE/" }, { "ticker": "CME", "name": "CME Group", "market": "US", "role": "宏观期货与清算基础设施", "roleEn": "宏观期货与清算基础设施", "bucket": "core", "href": "/investor/stocks/CME/" }, { "ticker": "VIRT", "name": "Virtu Financial", "market": "US", "role": "市场做市与流动性", "roleEn": "市场做市与流动性", "bucket": "core", "href": "/investor/stocks/VIRT/" }, { "ticker": "IBKR", "name": "Interactive Brokers", "market": "US", "role": "交易与风险管理基础设施", "roleEn": "交易与风险管理基础设施", "bucket": "core", "href": "/investor/stocks/IBKR/" }, { "ticker": "SH", "name": "ProShares Short S&P500", "market": "US ETF", "role": "组合保护工具", "roleEn": "组合保护工具", "bucket": "hedge", "href": "/investor/stocks/SH/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" } ], "stockTickers": [ "CBOE", "CME", "VIRT", "IBKR", "SH", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM", "APO" ], "coreView": "赛道：组合风控/证据质量。候选观察标的：CBOE, CME, VIRT, IBKR, SH。当前共有 40 条证据、1 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 67, "eventCount": 160, "firstSeen": "2026-04-25", "lastSeen": "2026-06-29", "supports": 9, "risks": 118, "neutral": 33, "analystCount": 12, "sourceTypeCounts": { "研究线程": 6, "研究笔记": 130, "证据更新": 6, "分析协作": 15, "研究主题": 3 }, "topTerms": [ "GPU", "CPI", "PCE", "Risk", "verification", "evidence", "quality", "necessary", "filters", "theme", "trades", "Portfolio" ], "whatChanged": [ "2026-06-29: 材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放右移。", "2026-06-29: 2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "2026-06-29: 霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "2026-06-29: 电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "2026-06-29: [auto-handoff] Impact of global capacity pricing spikes on domestic transformer export margins" ], "practical": { "sectorLane": "组合风控/证据质量", "sectorLaneEn": "Portfolio risk and evidence quality", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "probe", "actionLabel": "小仓验证", "actionTone": "yellow", "riskBudget": "验证仓：1-3% thesis risk", "portfolioRole": "验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "2-6 周", "portfolioUse": "适合用小仓位验证，不应在证据分歧扩大时追高。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "CBOE", "name": "Cboe Global Markets", "market": "US", "role": "波动率与期权基础设施", "roleEn": "波动率与期权基础设施", "bucket": "core", "href": "/investor/stocks/CBOE/" }, { "ticker": "CME", "name": "CME Group", "market": "US", "role": "宏观期货与清算基础设施", "roleEn": "宏观期货与清算基础设施", "bucket": "core", "href": "/investor/stocks/CME/" }, { "ticker": "VIRT", "name": "Virtu Financial", "market": "US", "role": "市场做市与流动性", "roleEn": "市场做市与流动性", "bucket": "core", "href": "/investor/stocks/VIRT/" }, { "ticker": "IBKR", "name": "Interactive Brokers", "market": "US", "role": "交易与风险管理基础设施", "roleEn": "交易与风险管理基础设施", "bucket": "core", "href": "/investor/stocks/IBKR/" }, { "ticker": "SH", "name": "ProShares Short S&P500", "market": "US ETF", "role": "组合保护工具", "roleEn": "组合保护工具", "bucket": "hedge", "href": "/investor/stocks/SH/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" } ], "stockTickers": [ "CBOE", "CME", "VIRT", "IBKR", "SH", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM", "APO" ], "primaryStocks": [ { "ticker": "CBOE", "name": "Cboe Global Markets", "market": "US", "role": "波动率与期权基础设施", "roleEn": "波动率与期权基础设施", "bucket": "core", "href": "/investor/stocks/CBOE/" }, { "ticker": "CME", "name": "CME Group", "market": "US", "role": "宏观期货与清算基础设施", "roleEn": "宏观期货与清算基础设施", "bucket": "core", "href": "/investor/stocks/CME/" }, { "ticker": "VIRT", "name": "Virtu Financial", "market": "US", "role": "市场做市与流动性", "roleEn": "市场做市与流动性", "bucket": "core", "href": "/investor/stocks/VIRT/" }, { "ticker": "IBKR", "name": "Interactive Brokers", "market": "US", "role": "交易与风险管理基础设施", "roleEn": "交易与风险管理基础设施", "bucket": "core", "href": "/investor/stocks/IBKR/" }, { "ticker": "SH", "name": "ProShares Short S&P500", "market": "US ETF", "role": "组合保护工具", "roleEn": "组合保护工具", "bucket": "hedge", "href": "/investor/stocks/SH/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" } ], "instrumentTags": [ "CBOE", "Cboe Global Markets", "CME", "CME Group", "VIRT", "Virtu Financial", "IBKR", "Interactive Brokers", "SH", "ProShares Short S&P500", "TLT", "iShares 20+ Year Treasury Bond …" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 5 类，近期事件 96 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 5, "analystCount": 12, "recentEvents": 96, "supportRatio": 0.06, "riskRatio": 0.74, "conflictScore": 84.1, "evidenceDepth": 100 }, "score": 62.3 }, "events": [ { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 21.5, "matchedTerms": [ "CPI", "Risk", "evidence", "risk", "dominant", "inflation", "transmission", "price" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 霍尔木兹海峡石油冲击与核心CPI传导", "summary": "致 投委会主席： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 1d956e69-9dc0-4197-8bd0-eb1630441435 结果摘要: Wrote handoff_response.md (11296 bytes); wrote handoff_response.en.md (11717 bytes) — The re…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "CPI", "Risk", "risk", "inflation", "Global" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 推导 PBOC 隔夜逆回购实际中标利率及中债收益率曲线非线性久期损失风险", "summary": "致 晨会主编： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: eb98b73e-bcc0-4d80-8508-8b06dfaaf2c6 结果摘要: Wrote handoff_response.md (10281 bytes); wrote handoff_response.en.md (11423 bytes) — We rever…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 14.5, "matchedTerms": [ "CPI", "Risk", "Portfolio", "risk", "price" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] Impact of global capacity pricing spikes on domestic transformer export margins", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 324722b5-2f7b-4b91-aec7-ac5429f32a39 结果摘要: Wrote handoff_response.md (11765 bytes); wrote handoff_response.en.md (12438 bytes) — pri…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 13.75, "matchedTerms": [ "transmission", "fully", "power", "transformer", "Global" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]，PPI-CPI剪刀差隔离了政策利率。任务设想的'成本→CP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "CPI", "macro", "Global" ], "href": "" }, { "date": "2026-06-29", "title": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "summary": "截至2026-06-29，支持卡片01–03的逻辑链并延伸至终端利率与全球流动性维度。美联储6-17维持3.50%–3.75% [S2]、SEP 2026中值3.8% [S3]；新主席Warsh首场FOMC点阵图从降息翻转为加息，9名官员预期加息、年内加息概率50.4%、市场定价10月加息，反应函数转向'约束性务实主义' [S14]。情景A（布伦特约105…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "CPI", "macro", "Global" ], "href": "" }, { "date": "2026-06-29", "title": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "summary": "对中国，2026年油价/柴油冲击既是被缓冲的进口税、又是战略加速器，且净额上比卡片01–02的美欧日集团损害更小。三大事实：（1）进口税真实但落在PPI而非CPI——5月PPI同比+3.9%（2022年7月来最快）、CPI仅+1.2%，疲弱内需把冲击吸收为生产者利润压缩；（2）进口税被三重缓冲——全球最大约14亿桶战略储备、作为受制裁油折价买方、疲弱需求；…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "CPI", "macro", "Global" ], "href": "" }, { "date": "2026-06-29", "title": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放…", "summary": "截至2026-06-29，变压器交期延长和材料瓶颈导致美国电网资本开支通胀率达32.72%。结合2027/2028年度PJM容量电费暴涨11.5倍至$333.44/MW-day，将拉动美国零售电价上涨4.18%，直接和间接推高全国CPI 0.124个百分点。高粘性通胀与AI基建庞大资金需求将上移中性利率50-75基点，延缓美联储降息路径。此外，并网滑期18…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "CPI", "macro", "Global" ], "href": "" }, { "date": "2026-06-28", "title": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "summary": "2026年6月17日Warsh首会以12-0维持利率3.50%-3.75%，但SEP中位点阵从隐含降息翻转为隐含加息（3.4%升至3.8%），18人中9人看至少1次加息、17人看通胀上行风险，2026年PCE上修至3.6%。背景是5月CPI同比4.2%反弹与韧性就业（年内约57万非农）。2年期美债当日跳升逾16基点、DXY破100、离岸人民币滑向6.80出…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 18.25, "matchedTerms": [ "CPI", "PCE", "macro", "Global" ], "href": "" }, { "date": "2026-06-28", "title": "主要央行保持鹰派按兵不动并非过度收紧的政策错误，而是应对由服务业传导支撑的核心PCE粘性（预计H2维持在3.1%-3.4%）的必要防御，但这会使实体经济承受限制性实际利率压制，支撑久期债防守…", "summary": "本卡综合并支持前三张卡对能源脉冲退潮的判断，但指出核心通胀将出现背离：5月核心CPI放缓至2.9%，但核心PCE因服务二轮传导仍粘在3.4%。由于美联储盯防PCE，鹰派立场并非政策错误。然而CPI下行使限制性实际利率升至0.7%-1.15%，压制增长，支撑久期超配，并导致融资余额达28.13亿元、对海外宏观高敏感的中游机械面临收缩压力，建议低配/规避。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 18.25, "matchedTerms": [ "CPI", "PCE", "macro", "Global" ], "href": "" }, { "date": "2026-06-28", "title": "核心服务与住房通胀的粘性：美联储政策的后续驱动力", "summary": "能源转为通缩之际，核心服务与住房粘性足以支撑美联储鹰派、抵消能源通缩。关键在指标分化：整体PCE 4.1%/整体CPI 4.2%将因油价基数效应回落，但美联储盯住的核心PCE升至3.4%（2023年10月以来最高），由占核心PCE一半以上、运行约3.5%-3.9%的超级核心（剔除住房的核心服务）驱动。住房同比3.4%（OER 3.3%），去通胀放缓而非结束…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 18.25, "matchedTerms": [ "CPI", "PCE", "macro", "Global" ], "href": "" }, { "date": "2026-06-28", "title": "6月点阵图不是简单过时，而是可能“对得不是理由”：能源理由过时，AI/电网与折旧通道支持更高更久，但不足以单独证明立即加息。", "summary": "截至2026-06-28，本卡综合认为：6月点阵图并非被最初的油价冲击证明，因布伦特已回落至约US$72/bbl；但AI/电网导致的结构性电价稀缺、核心PCE 3.4%、名义需求仍强以及AI-capex折旧悬挂，使“更高更久”仍有宏观依据。该通道不足以单独证明立即加息，因为电力CPI权重仅2.489%、PJM冲击区域性强、US$176B折旧低估属低置信度压…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 18.25, "matchedTerms": [ "CPI", "PCE", "macro", "Global" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力需求增长对电网侧扩容及配电资产的压力测试", "summary": "压力测试结论：前序主线被证实并强化。2026年AI落地的约束性瓶颈已非GPU，而是电网边缘——电力变压器、中高压开关设备、取向硅钢（GOES）与接网排队。大型变压器交期3–5年、全球产能利用率约98%、美国接网积压1,500–2,300GW、枢纽接入排队4–7年，致美国2026约12GW管线仅约4GW在建（隐含约67%滑期，自测算）。这把AI算力部署拉长而…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 15.5, "matchedTerms": [ "GPU", "PCE" ], "href": "" }, { "date": "2026-06-28", "title": "I have completed the research for Research note 01 of this research thread session. Summary of …", "summary": "I have completed the research for Research note 01 of this research thread session. Summary of Work 1. Analyst Selection : Appointed the Global Macro Analyst ( global macro ) for …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 14, "matchedTerms": [ "PCE", "macro", "inflation", "Global" ], "href": "" }, { "date": "2026-06-28", "title": "2026年6月美联储、欧央行、日本央行同步转鹰，却在一场正在消退的伊朗战争油价冲击中收紧政策", "summary": "2026年6月，欧央行（+25bp至2.25%）、日本央行（+25bp至1%，1995年来最高）与美联储（按兵不动但点阵图从降息翻转为加息）在一周内集体转鹰，导火索是伊朗战争与霍尔木兹海峡中断推动的油价冲击。美国5月CPI同比4.2%创三年新高，但全由能源（同比+23.5%）驱动，核心仅2.9%。然而布伦特已从114美元峰值回落至约80美元，海峡有条件重开…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "CPI", "macro", "Global" ], "href": "" }, { "date": "2026-06-28", "title": "美联储2026年6月鹰派转向：新主席Warsh首会下点阵图翻转为加息，叠加中东能源冲击", "summary": "2026年6月17日FOMC维持利率3.50%–3.75%，但点阵图从隐含降息翻转为隐含加息：2026年底利率中值由3.4%升至3.8%，18人中17人判通胀风险偏上行，整体PCE预测上调至3.6%、核心3.3%。新主席Warsh以「五年失误必须纠正」的信誉优先基调定调，2年期收益率当日跳升约16基点、美元走强、股市下挫。但支撑这套预测的能源脉冲正快速逆转…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.25, "matchedTerms": [ "PCE", "macro", "Global" ], "href": "" } ], "timeline": [ { "date": "2026-04-25", "supports": 0, "risks": 1, "neutral": 0, "value": 36, "events": [ { "title": "4月FOMC前的双向风险再定价：关税与油价扰动下，美联储是否延后首次降息窗口", "summary": "4月FOMC前的双向风险再定价：关税与油价扰动下，美联储是否延后首次降息窗口 在3月CPI跳升、核心PCE仍在3.0%、失业率维持4.3%、GDPNow降至1.2%的背景下，美联储在2026年4月28日至29日会议后是否仍会把首次明确降息窗口留在6月或更晚？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-02", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "本看板研究链完整、逻辑自洽且数值一致，成功识别了油价持续性与因子拥挤度对 6 月 FOMC 前期布局的关键约束，具备极高的实战参考价值，准予结项。", "summary": "本卡由研究质量主管对第 01-09 卡的研究链进行了终审。审计确认研究逻辑闭环、数值高度一致（特别是 Q1 PCE 4.5% SAAR 与质量因子 +1.7 σ 拥挤度）。研究成功从宏观预测演进为具体的对冲因子价差执行方案。尽管存在对 5 月 14 日 CPI 数据的窗口敏感性，但整体看板已具备极…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源冲击最终可能回落，但在 6 月 SEP 前的持续性足以推高总体通胀并制造有限核心传导风险，因此 Fed 的实际基准应从“即将恢复降息”转为“更久按兵不动”。", "summary": "本卡以 energy-analyst 视角压力测试 card-01 的宽松倾向基准。结论是油价冲击尚未变成自我强化的核心通胀 regime，但截至 2026-05-02 回落不够快，Fed 难以把 3-4 月总体通胀完全视为噪音。EIA 仍预计 Brent 在 2Q26 触及 $115/b，汽油和…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-04", "supports": 0, "risks": 4, "neutral": 0, "value": 20.8, "events": [ { "title": "2026年4月FOMC、Q1 GDP、3月CPI和3月就业数据之后，美国“类滞胀”风险是否升温", "summary": "截至2026年5月3日，美国数据不是衰退型，但软着陆叙事已变得不干净。Q1实际GDP为2.0% SAAR，私人国内购买者实际最终销售为2.5%，显示需求仍有韧性；但3月CPI环比0.9%、同比3.3%，能源环比10.9%、汽油环比21.2%，且Q1核心PCE通胀为4.3%，使Fed难以快速转鸽。3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国通胀再加速与 Fed 在 4 月会议后的反应函数", "summary": "本卡选择 chief-economist，聚焦 Fed 在 2026 年 4 月会议后的反应函数。证据显示，Q1 实际 GDP +2.0%，3 月非农 +178,000、失业率 4.3%，增长端尚未迫使 Fed 降息；但 CPI +3.3% y/y、PCE +3.5% y/y、核心 PCE +3.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Hormuz供给冲击在急性期内强烈，但正常化轨迹（4Q26低于$90/b）所产生的基期反转效应使油价对美联储的约束具有时限性——首次降息推迟至2026年Q4而非推入2027年，宽松窗…", "summary": "能源分析师否定油价从结构上封锁美联储2026年宽松窗口的论断。在3月7.5 mb/d、4月9.1 mb/d减产、2Q26 Brent峰值约$115/b、4Q26回落至$90/b以下的正常化情景下，能源分项对标题CPI的同比贡献从2Q26的约+2.3–2.9 pp机械性下降至4Q26的约+0.8–1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国CPI能源项权重约6.7%（其中汽油约3.4%），PCE能源项权重约4.0%", "summary": "美国CPI能源项权重约6.7%（其中汽油约3.4%），PCE能源项权重约4.0%", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 4, "neutral": 1, "value": 25.2, "events": [ { "title": "2026年3月美国PCE通胀同比3.5%，核心PCE同比3.2%", "summary": "2026年3月美国PCE通胀同比3.5%，核心PCE同比3.2%", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年3月美国名义PCE环比增0.9%，实际PCE增0.2%，个人可支配收入增0.6%，储蓄率3.6%，PCE通胀同比3.5%，核心PCE同比3.2%。", "summary": "2026年3月美国名义PCE环比增0.9%，实际PCE增0.2%，个人可支配收入增0.6%，储蓄率3.6%，PCE通胀同比3.5%，核心PCE同比3.2%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "「类滞胀 Lite + H4L + 区域行 CRE 脆弱性」九卡论文链通过终审，等级为 PASS WITH MINOR REVISIONS：仅需澄清 KRE 42 美元线属 Tier…", "summary": "对九张前置卡片做终审：核心数字（Q1 GDP 2.0%、CPI 0.9% MoM/3.3% YoY、汽油 21.2%、核心 PCE 4.3%、CRE 到期墙 5,700 亿美元、KRE 2026E EPS -12% 至 -18%、办公 CMBS 违约 11.71%、KRE 42 美元支撑）跨卡一致…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "截至2026年5月，全球AI投资主题已完成从“投机性硬件溢价”（GPU囤积）向“AI公用事业化”（实物资产交付与ROI）的转变", "summary": "截至2026年5月，全球AI投资主题已完成从“投机性硬件溢价”（GPU囤积）向“AI公用事业化”（实物资产交付与ROI）的转变", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "微软2026财年第二季度资本开支为375亿美元，其中约三分之二用于GPU、CPU等短寿命资产，其余用于支持15年及以上寿命的数据中心长寿命资产。", "summary": "微软2026财年第二季度资本开支为375亿美元，其中约三分之二用于GPU、CPU等短寿命资产，其余用于支持15年及以上寿命的数据中心长寿命资产。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "[auto-handoff] 主权算力财政支出影响", "summary": "致 TMT行业分析师： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 1f2ac627-aa0e-4991-81c1-4c46ae106c06 结果摘要: Wrote handoff_response.md (17,200 bytes); wrote handoff_res…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-12", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "2026年5月美国核心CPI环比≤0.25%且服务通胀（OER + 核心服务）连续两个月放缓", "summary": "2026年5月美国核心CPI环比≤0.25%且服务通胀（OER + 核心服务）连续两个月放缓", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 1, "risks": 1, "neutral": 1, "value": 56.8, "events": [ { "title": "TMT行业分析师 · 卡 04 · 2026 05 13 看板: 6a371a6d 12cf 4e7e adc1 c8229ffe8c55 议题: 高拥挤度TMT赛道的杠杆脆弱性与基…", "summary": "TMT行业分析师 · 卡 04 · 2026 05 13 看板: 6a371a6d 12cf 4e7e adc1 c8229ffe8c55 议题: 高拥挤度TMT赛道的杠杆脆弱性与基本面断裂点排查 立场: stress test (自下而上) 工作日期: 2026 05 13 (Asia/Sing…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "低收入违约是消费断裂的预警信号，但在2026-05-13尚未构成足以压倒能源、住房、关税和服务通胀的宏观需求冲击。", "summary": "截至2026-05-13，本卡压力测试上一张卡片的强判断：低收入信贷违约上升是真实的需求尾部风险，但尚不足以单独扭转通胀路径或迫使美联储提前转向。4月CPI同比3.8%、3月核心PCE同比3.2%、Q1家庭债务仅小幅增加，信用卡和车贷严重违约流入率趋于平稳。基准情形是美联储延后且不情愿地转向，除非…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "VIX扭结的正确解读是“中后端精准对冲 + 近端波动率陷阱”，A股应以降低贝塔和控制融资敏感风格为主，而不是简单清仓或追买近端保护。", "summary": "截至2026-05-13，本卡判断VIX期限结构扭结不是单纯情绪过载，也不是可无脑追买的崩盘信号。中后端波动率上移和美国CPI、PCE、Fed按兵不动相互印证，支持此前“波动率中枢抬升、权益低估利率波动”的主线；但近端异常低点更像到期和事件机制造成的波动率陷阱。A股策略上应保持权益但降低贝塔，偏向…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 2, "neutral": 0, "value": 41.2, "events": [ { "title": "4 月 CPI/PPI 已足以让 Fed 从降息倾向转向对称选择权，但在 PCE/CPI 连续确认前，加息仍是尾部而非基准；曲线若熊陡先压估值，若熊平则更快传导至信用。", "summary": "截至 2026-05-14，本卡支持上一张卡的压力测试方向，但不认为 4 月 CPI/PPI 单独足以让加息成为 Fed 基准情形。CPI +0.6% MoM、+3.8% YoY，核心 CPI +0.4% MoM，PPI +1.4% MoM、+6.0% YoY，核心 PPI +1.0% MoM、+…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "向量 C 下的风险平价重构应把权益/大宗商品风险比从约 3:1 压到约 1.5:1，并同步提高 TIPS、黄金与现金，而不是把组合简单改造成商品多头。", "summary": "截至2026-05-14，本卡对向量 C 做风险平价压力测试：UMich 长期通胀预期 3.4%、NFIB 计划涨价净 27%、4 月 CPI 3.8% y/y 与核心 PCE 3.2% y/y 共同说明，旧的股债分散化假设需要下调。建议将权益风险预算从 35%-40% 降至 25%-30%，名义…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-15", "supports": 0, "risks": 1, "neutral": 1, "value": 51.6, "events": [ { "title": "[auto-handoff] Inflation Transmission from U.S.-Iran Conflict", "summary": "致 市场情绪师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 3319aae8-eed8-4aac-bf93-59603f640ad9 结果摘要: Wrote handoff_response.md (7027 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Based on the macro fundamental data as of May 16, 2026, the S&P 500 (SPY) faces a significa…", "summary": "Based on the macro fundamental data as of May 16, 2026, the S&P 500 (SPY) faces a significant valuation ceiling at the 7500 ($750) level. While AI dr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-16", "supports": 1, "risks": 0, "neutral": 1, "value": 67, "events": [ { "title": "三因素叠加只制造 6-7 月 PCE 商品的可见但有限的二次探头，足以把首次降息从 9 月推迟到 12 月、但不会终止周期，原交易栈在 4-8 周窗口内依然成立，唯需缩短利率敏感腿的…", "summary": "截至 2026-05-17，6/8 货量峰值叠加 $5.60 柴油与运费反弹（FBX01 +34.9%）足以使核心 PCE 商品在 6-7 月单月读数 +8~14 bp（7/31 与 8/29 发布），但 Q3 平均仅 +0~5 bp，关税缓和的永久性通缩传导仍主导半年视角。在通胀预期锚定（5y5…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "习特会受控暂停在 2026 宏观周期中既不是联盟放大器，也不是 FDPR 的实质性政治成本——它是信号工具，不是 GDP/CPI/利率路径的枢纽变量。", "summary": "首席经济学家反驳卡 01 的「联盟放大器 vs. FDPR 政治成本」二分框架：从 2026 年宏观周期看，美国核心 PCE 已回落、联储/欧央行的反应函数由劳动力与服务通胀主导，关税暂停最多移动通胀尾部 10–20bp；FDPR 是存量法定架构（欧盟两用品、METI、荷兰先进节点管制均已本土化）…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-19", "supports": 0, "risks": 0, "neutral": 2, "value": 63.4, "events": [ { "title": "云端CapEx挤出引发的存储通胀让美/欧/日换机曲线弯折但不击穿，对头条消费增长拖累仅5-10bp，真正的宏观信号是核心商品CPI再加速给联储/欧央行的政策路径增加摩擦。", "summary": "截至2026-05-19，云端CapEx虹吸HBM/CoWoS导致端侧设备ASP上涨50-100美元的冲击是真实的，但不会扼杀美/欧/日换机周期。三条结构性驱动（疫情装机潮到期、Windows 10 EOL企业换机、AI能力门槛）和高端买家低弹性把结果压成'量平/结构上移/ASP升'格局。对美+欧…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "存储通胀是触发条件而非决胜因子；最后50bp推迟与否取决于劳动力松弛是否继续，而欧央行比联储更易被该组合推入2027年。", "summary": "截至2026-05-19，AI存储通胀对核心PCE+5-8bp、核心HICP+4-7bp，单独看落在月度噪音内，不足以把联储/欧央行最后50bp宽松推迟到2027年。但叠加2026年下半年住房与核心服务基数翻面（机械贡献15-20bp）与能源基数转中性后，年底核心PCE可能漂至2.4-2.5%。决…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 46.8, "events": [ { "title": "能源价格冲击对全球通胀传导的深度评估", "summary": "能源端压力测试得出双向结论：传导真实存在但不会演化为螺旋。Brent ≈88美元、Henry Hub +35%、柴油 +16% [S5][S6][S7]，机械地为CPI贡献约1.4pp [自测算]。但1979式螺旋的必要条件不齐——美国油强度仅为1973年三分之一 [S3]、OPEC+闲置产能5.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI 资本开支二阶导是否值得作为下周组合层面主线", "summary": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 725e6cc9-5470-4a4b-9361-d27fe2655ac9 结果摘要: Wrote handoff_response.md (14571 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 1, "risks": 2, "neutral": 3, "value": 65.2, "events": [ { "title": "Re-estimate supply chain price shock transmission to U.S. core PCE", "summary": "Re-estimate supply chain price shock transmission to U.S. core PCE In 'Thesis 1: Step 4', direct a high-priority handoff to global-macro to re-estima…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI Capex韧性 × 油价冲击 × 联储延宕的共振，通过工业PPI与电力催生海外隐性二次通胀，将10Y美债锚定在5.0–5.3%，断裂点在日元/JGB而非美国风险资产。", "summary": "在2026-05-23工作日期下，AI硬件Capex韧性与霍尔木兹油价冲击的共振真实存在，但传导通过工业PPI、电力价格与资本品平减指数发生，而非头部CPI/PCE。这一通胀路径让联储基于PCE的反应函数系统性低估真实通胀，将10Y美债锁定在5.0–5.3%走廊延续整个2026下半年（期限溢价回升…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed a comprehensive quality assurance review and pre publication audit of the …", "summary": "I have completed a comprehensive quality assurance review and pre publication audit of the entire multi step research chain (Research note 01 through…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "基准情形应按通胀冲击而非增长冲击管理，除非未来两次通胀数据明显降温且就业和服务业需求同步跌破阈值。", "summary": "截至2026-05-23，本卡综合前序“杠铃型资金流”和28/46/26等风险配置，判断下一轮宏观冲击的基准应设为通胀主导。CPI、PCE、ISM价格、Brent和长端利率已经触发或接近通胀压力阈值；就业、服务业就业和兼职数据显示增长脆弱，但GDP、服务业PMI和新订单尚未进入衰退式破裂。基准概率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 49.6, "events": [ { "title": "AI capex 支撑真实，但在粘性通胀和高实际利率下，它不足以触发小盘/高贝塔的广谱可投资轮动。", "summary": "截至2026-05-23，首席经济学家否定“AI财报兑现后资金可顺畅转向小盘/高贝塔”的广义结论。AI capex 仍是增长支撑，但美联储仍受通胀约束，CPI/PCE 重新走高，10年期实际利率处于高位，美国增长是韧性而非再加速，中国 PMI 也不足以提供广谱再通胀。组合上应保留现金流已兑现的 A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "We have successfully completed Research note 03/08 of the research thread session efecefe6 …", "summary": "We have successfully completed Research note 03/08 of the research thread session efecefe6 2a06 4881 87e9 ccec1f502077 as the Chief Economist. Stance…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 3, "neutral": 0, "value": 45, "events": [ { "title": "2026-05-25：4月通胀冲击与5月FOMC纪要之后的美国政策路径重定价", "summary": "截至2026-05-25，Fed更像按兵不动或加息尾部风险，而不是确定的降息交易。4月CPI、3月PCE、Q1 GDP和5月FOMC纪要共同显示：通胀仍高、增长仍稳、劳动力只是温和降温。5月22日10Y美债4.56%、30Y美债5.07%，说明长端利率仍在为通胀与期限溢价索价。关键检验点是2026…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-25 美国通胀再加速与美联储更久维持高利率风险", "summary": "截至 2026-05-25，首席经济学家判断美国宏观主线应从“降息延后”转向“更久维持高利率”。4 月 CPI 与 3 月 PCE 均显示通胀压力偏强，4 月就业没有弱到迫使宽松，美联储 4 月会议纪要还保留了若通胀持续高于 2% 则可能收紧的路径。4 月 PCE 将是近期核心催化。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-05-25 美国通胀再加速与美联储更久维持高利率风险", "summary": "2026-05-25 美国通胀再加速与美联储更久维持高利率风险 在 2026-05-28 公布 4 月 PCE 前，4 月 CPI、就业和美联储纪要是否已使降息不再是基准情景？", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 0, "neutral": 1, "value": 67.4, "events": [ { "title": "I have completed the global macro analysis for Research note 01 of the research thread sess…", "summary": "I have completed the global macro analysis for Research note 01 of the research thread session b0a5df37 4aba 4443 9696 a334f45b9dd4 . Key Findings & …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 2, "neutral": 0, "value": 53.8, "events": [ { "title": "高利率维持叠加通胀再加速预期下，防御策略应放弃经典60/40拉久期思路，转向'短久期高质量固收+实物资产（黄金、能源）+质量/现金流权益+T-bill现金'的杠铃式组合，并显式低配长…", "summary": "在美联储6月预计维持3.50%–3.75%、核心CPI同比2.8%且面临二次加速的背景下，主要风险已从衰退尾转为'粘性通胀+股债正相关'。建议防御性配置为：权益42%（质量/价值/能源/防御性，低配长久期成长与小盘）、固收28%（1–3年美债+5年TIPS+短端IG，低配长久期美债与HY）、商品1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储六月议息前的通胀再加速与政策路径重估", "summary": "2026-05-27卡片判断：6月16-17日FOMC基准仍是维持3.50%-3.75%不变，但降息路径已明显后移。4月CPI同比3.8%、核心CPI同比2.8%，就业仍有115,000人的非农增量和4.3%失业率，难以支持美联储穿越通胀冲击降息。若5月PCE与6月CPI显示核心通胀扩散，6月SE…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "6月基准应是滞胀暂停而非降息，科技久期交易需要用更高折现率重定价。", "summary": "2026-05-28结论：支持把6月降息从基准情景下调为上行情景。Brent约97美元/桶、Hormuz实际复航未确认、FOMC纪要偏鹰、CPI/PCE/GDP/就业组合显示通胀风险高于增长下行压力。AI基础设施可保留，但AI应用、量子/Web3、卫星互联网等久期资产应以4.5%以上10年期美债环…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 2, "neutral": 1, "value": 59.6, "events": [ { "title": "材料价格确认的是通胀粘性和利润率压力，而不是美国消费实际销量的全面走强，因此应维持选择性零售多头、避免激进拉长久期，并把下一步焦点转向CPI/PCE传导。", "summary": "截至2026-05-28，材料链条支持“消费韧性仍在、但并非全面实际销量繁荣”的综合判断。4月零售和BofA刷卡仍强，但ISM Prices、钢厂产品、有色金属轧制品、工业化学品和油价相关输入显示成本推动更突出。钢、有色、化工和电池材料共同指向：名义零售可维持，但毛利率、低收入可选消费和Fed通胀…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然整体能源价格走弱缓解了CPI名义压力，但中间投入与柴油成本的滞后传导将维持核心通胀高粘性，迫使新上任的沃什（Warsh）联储在6月维持3.50%-3.75%的高利率，并将降息至少…", "summary": "截至2026-05-28，我们量化测算了原材料及能源价格对核心通胀的滞后传导效应。虽然5月油价回落，但中间品价格高企（ISM支付价格达65.5%），对核心货物传导弹性为0.22（滞后4-6个月），对服务传导弹性为0.15（滞后2-3个月），三季度核心CPI/PCE仍显粘性。5月22日宣誓就职的新主…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国4月PCE、通胀预期与6月FOMC维持利率不变窗口", "summary": "截至工作日期2026-05-29，4月PCE显示核心通胀单月降温，但总体PCE仍为+0.4% m/m、+3.8% y/y，名义消费增长+0.5% m/m，实际消费仅+0.1% m/m。结合4月CPI的能源冲击、4.3%失业率、44.8的5月消费者信心和上行的通胀预期，基准判断是6月FOMC维持利率…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 64, "events": [ { "title": "能源价格重心的下移提供了一次缓和全球“滞胀”压力的正向供给冲击，虽然成功消解了美欧央行的强行加息风险，但在低储蓄率、物流滞后与-273 bps中美巨幅倒挂的刚性约束下，全球主要央行在…", "summary": "本报告推演了Brent原油重心下调至82-92美元对三季度全球通胀与货币政策的边际影响。受汽油直接回落与航煤暴跌（裂解价差收窄至18-22美元/桶）间接拖累，三季度美国整体CPI与欧洲HICP将分别下修21 bps和30 bps。这有效移除了美联储年底加息选项，但受制于2.6%极低储蓄率与核心PC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 3, "neutral": 0, "value": 53.4, "events": [ { "title": "消费者承担接近 100%、企业短期承担约 64%、CNH/MXN 几乎零吸收且 VND 远不足以对冲，关税新边际压力在 2026 下半年已显著消退但水平被嵌入；CNH 反向走强使中国…", "summary": "量化关税传导：截至 2026-04-08，美国有效关税税率 11.8%（Yale Budget Lab），核心商品 PCE 已上行 3.1%、对核心 PCE 通胀贡献 +0.8pp，传导 7 个月内基本完成；Dallas Fed 确认 2026Q1 见顶，4 月核心商品 CPI 几乎走平。短期分担…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "油价持续高位将迫使美联储在2026年下半年保持鹰派立场，推升10年期美债收益率至4.75%并使期限溢价突破100bps。", "summary": "评估布伦特原油维持在105-120美元/桶对美联储政策及美债收益率的冲击。高油价将直接贡献headline CPI约1.3-1.5%，并推升核心PCE至3.4%以上。美联储预计将维持3.75%-4.00%的高利率，甚至实施防御性加息。受通胀不确定性、沃什提名政权更迭及地缘溢价驱动，预计10年期美债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "追踪AI capex故事与供应链瓶颈", "summary": "追踪AI capex故事与供应链瓶颈 Track the unfolding AI capex story: are GPU / CPO / optical supply bottlenecks priced in, or is there still beta to compress? Buil…", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 2, "neutral": 2, "value": 66.8, "events": [ { "title": "As the Global Macro Analyst ( global macro ) , I have conducted a rigorous macroeconomic re…", "summary": "As the Global Macro Analyst ( global macro ) , I have conducted a rigorous macroeconomic regime stress test on the multi asset hedge overlay proposed…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "若6月跨太平洋现货运价继续上行但TEU只温和增长，应优先解读为船公司定价和成本冲击，零售利润率先受压，CPI/PCE传导选择性且滞后。", "summary": "截至2026-05-31，本卡综合支持“成本挤压而非需求繁荣”判断。Drewry在2026-05-28显示WCI升至每40ft集装箱2,800美元，跨太平洋航线同时受blank sailings和旺季附加费支撑；NRF则预计2026年上半年进口12.59百万TEU、同比仅增0.5%。美国4月零售销…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月运费冲击单独看是零售商Q3毛利率故事和Q4 PPI微抬，对核心PCE的6–9月贡献上限约+15 bp，未达Fed叙事翻转门槛；只有当严重WCI、零售商弃吸收、服务粘性、预期脱锚四…", "summary": "截至2026-05-31，本卡量化了6月WCI若继续上行的核心商品CPI/PCE传导。基线$2,800/FEU [S1]、NRF进口同比仅+0.5% [S2]，按四道过滤器（合同权重、运费占到岸成本1–4%、进口占核心商品30%、零售商30–50%吸收）测算，温和/中性/严重三种情景对核心PCE同…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Brent 长期高于 $100/bbl 会先移除主要央行的降息期权，再通过短端通胀预期和二次传导把政策天平推向持有高利率或重新加息。", "summary": "截至2026-05-31，Brent 持续高于 $100/bbl 已从 headline CPI 冲击升级为央行反应函数冲击。Fed、ECB、BoE、BoC 的基准路径是先推迟降息，再在二次通胀确认后重新定价加息；BoJ 因仍在正常化周期中最偏鹰，PBoC 则因国内需求和价格机制仍可保持宽松倾向。…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 3, "neutral": 0, "value": 56.2, "events": [ { "title": "Temu/Shein退潮后的美国折扣零售受益应被定义为名义份额回收；若2H26实际商品PCE转负且低收入客流走弱，消费回流和广告竞价池替代会同步失效。", "summary": "截至2026-06-02，本卡对Research note 01和Research note 02做压力测试而非完全否定。美国折扣零售更可能承接Temu/Shein退潮后的名义份额，而非低收入家庭实际商品需求的完整回流。4月实际DPI环比-0.5%、实际PCE仅+0.1%，叠加CPI同比3.8%、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Inflation Expectation Revision", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: 2df94c58-13f4-4061-be74-dca64cb12f59 结果摘要: Wrote handoff_response.md (9672 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "As requested, I have analyzed the macro economic constraints on AI valuations as a Global M…", "summary": "As requested, I have analyzed the macro economic constraints on AI valuations as a Global Macro Analyst ( global macro ). Based on the authoritative …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 1, "neutral": 0, "value": 69.6, "events": [ { "title": "2026 年通胀粘性、温和为正的实际利率与非对称的央行反应函数足以让铜、银、电力的稀缺性溢价持续 12-24 个月进入跨资产定价模型，构成结构性而非阶段性交易。", "summary": "截至 2026-06-04，核心 PCE 3.3%、CPI 3.8%、10Y TIPS 实际收益率 ~1.8-2.3%、美联储数据依赖的缓慢降息姿态共同构成一个'实际利率为正但不严苛、政策反应非对称'的制度。叠加关税传导、AI 与电气化资本开支超级周期、劳动力与服务通胀粘性三股多年期供给侧楔子，未…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 1, "risks": 2, "neutral": 2, "value": 79, "events": [ { "title": "关键矿产地缘脱钩、资本品通胀与终端中性利率中枢重估", "summary": "镓/锗/锑出口管制带来的 +30%~+50% 高科技与国防制造品 PPI 溢价，对消费端核心 goods CPI 的直接传导温和：核心 goods 仅占核心 PCE 约 30%，矿产密集部分占其 10%~20%，且金属是 BOM 微量成分，折算后仅向核心 PCE 增加约 +0.03~+0.25pp…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "沃什(Warsh)新联储的滞胀困局:6月16–17日FOMC的政策抉择", "summary": "美联储连续第三次将利率维持在3.50%–3.75%,4月以8–4票通过,为1992年来首次四票反对:Miran主张降息,三位地区主席反对保留宽松倾向。新主席Kevin Warsh于5月22日就任,6月16–17日为其首场FOMC。通胀冲击源于供给端:霍尔木兹海峡3月关闭推动Brent一季度末破12…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 6 月社零 +3.5% 后 PCE deflator 终值与 A 股可选消费 EPS 修正", "summary": "致 首席策略师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 7f83dddf-bec3-4c08-80a6-a71de16f0c3a 结果摘要: \"session_id\": \"e06f898d-b0a7-4207-88fa-b5f1e1804284\", \"respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决", "summary": "AI Capex 从 GPU 向电网/SMR 的分流幅度与时点：高确信对齐是否已被 NVDA/MRVL 价格行为否决 AI Capex Pivot to Grid/SMR: Has Consensus Already Been Repudiated by NVDA/MRVL Price Actio…", "sourceLabel": "研究主题", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 72.4, "events": [ { "title": "[auto-handoff] Semiconductor De-leveraging vs Physical DC Bottleneck", "summary": "致 衍生品策略师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 6909760a-2b3a-4f87-9c0c-ee2d76724c67 结果摘要: Wrote handoff_response.md (11874 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 1, "risks": 4, "neutral": 2, "value": 69.8, "events": [ { "title": "2026-06-06美国就业与通胀组合对美联储路径和全球利率的影响", "summary": "截至2026-06-06，5月非农新增172,000人、失业率4.3%、工资同比3.4%，显示就业降温但未断裂[S1]；4月CPI同比3.8%、核心PCE同比3.3%，仍限制美联储快速转向[S2][S3]。报告判断6月会议前不宜追逐激进降息交易，除非6月10日5月CPI明显降温[S6][S7]。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026-06-07：美国5月非农超预期后的美联储6月政策路径", "summary": "2026-06-07基准判断：5月非农新增172,000人、失业率4.3%、3-4月合计上修93,000人，使6月16-17日FOMC降息难以成立。4月PCE同比3.8%、CPI同比3.8%，叠加ISM服务业和制造业PMI仍扩张，支持美联储短期维持3.50%-3.75%目标区间。除非6月10日CP…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "面对'劳动降温+核心粘性'的非对称宏观，正确答案不是单一押注，而是质量股票+短端carry+精选长端的杠铃，并把黄金作为滞胀尾部保险。", "summary": "在5月就业降温但核心通胀仍粘（NFP +172k、失业率4.3%、核心PCE 3.3%）的组合下，建议总体股票风险中性、股票内偏质量（超配+3至+5 ppt）、固定收益采用'短端carry+精选长端'杠铃（现金/短期美债+4 ppt、1–3年IG+2 ppt、10年UST+1 ppt、10年TIP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月FOMC前瞻:在伊朗能源冲击与5月非农大超预期下,Warsh主席首次SEP的政策路径", "summary": "5月非农(6月5日发布)17.2万远超8.5万的一致预期、前值上修;4月CPI同比3.8%为2023年5月以来最高,核心2.8%,能源因伊朗冲突同比+17.9%。CME隐含6月17日维持利率概率99.4%,2026年底前加息概率约70%。基准情景(60%)为鹰派按兵不动:Warsh首份SEP把20…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 2, "value": 81.2, "events": [ { "title": "通胀对冲组合压力测试与跨板块配置的合规性与叙事一致性终审", "summary": "对前9卡链条的终审：分析主线扎实——内部自洽、引用规范、概率树（48/22/25/5）谱系清晰、布伦特中枢95-98美元与80美元地板、关税14%→9.6%、7月USTR §301（07-06评论/07-07听证）、11-10稀土悬崖等关键数字跨卡一致。但有两项阻断级缺陷：card-07与card…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "6月FOMC前的美国类滞胀格局：油价冲击、粘性核心通胀与韧性就业", "summary": "4月CPI同比3.8%/核心2.8%、5月非农超预期+17.2万、布伦特从$117峰值回落至$95——这套组合让联储6月16-17日按兵于3.50-3.75%几乎成为定局。能源的反通胀顺风已消失,工资增速3.4%仍高于与2%核心PCE相容的水平。10Y美债已升至4.55%,曲线定价'联储卡住'。基…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 铜价高企背景下家电龙头企业与二三线品牌成本裂谷对工业品PPI与制造业投资的传导评估", "summary": "致 首席经济学家： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: e00d3d45-0543-4a90-a4a3-8afa35e2cc1e 结果摘要: Wrote handoff_response.md (12590 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 76.6, "events": [ { "title": "能源冲击已足够大到支撑美联储鹰派沟通，但是否转化为实际加息，取决于6月海峡恢复是否延误以及二次通胀效应是否扩散。", "summary": "截至2026-06-08，霍尔木兹冲击足以支撑美联储鹰派预期，但更偏向“压制降息、保留加息风险”而非立即加息。海峡正常承载近20 mb/d石油出口和约19%全球LNG贸易[S1]，绕行能力仅3.5-5.5 mb/d[S1]；IEA显示4月全球供应较2月累计下降12.8 mb/d[S2]。美国4月C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 1, "value": 81, "events": [ { "title": "AI需求到2026年底足以兑现HBM锁定的供给，但A股的真正约束是份额捕获与财政融资通量，因此应超配云挂钩供应商与一体化国产替代栈、把端侧AI作期权、低配离岸GPU依赖与重复建设长尾…", "summary": "锚定2026-06-10，本卡综合第01卡（首席策略师）与第02卡（政策分析师）。中国宏观叠加三条修正：(1) A股AI capex由大基金三期+省级算力券+头部云自融政策配给，民营信贷通道薄，是财政题非周期题；(2) 5月PMI 49.6 / Caixin 50.4 / 社融~8.4% / CP…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "生猪产能去化路径及其对2027年通胀的非线性传导", "summary": "生猪产能去化路径及其对2027年通胀的非线性传导 Hog Capacity De-stocking Path and Nonlinear Inflation Transmission 当前猪价跌破现金成本是否会触发规模化猪企的恐慌性断崖式出清？这会否导致2027年H1出现大幅反弹并推高CPI通胀中…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 0, "neutral": 1, "value": 88.4, "events": [ { "title": "医药'去中国化'是一股结构性、慢释放的供给侧通胀力量，每年向美国核心医疗CPI注入25–40bp，并已迫使美联储将生物科技融资环境纳入金融稳定框架。", "summary": "截至2026-06-12，《生物安全法案》及1260H清单升级已将中美医药脱钩转化为可测量的供给侧成本冲击。按模态加权，非中替代溢价合计抬升受影响SKU零售价4.5–6.5%，对应美国核心医疗服务CPI每年额外+25–40bp、Medicare Part B支出增速每年高出基线60–90bp，20…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 1, "neutral": 0, "value": 80.8, "events": [ { "title": "Work Summary For Research note 09, we conducted a macro and micro market stress test on the…", "summary": "Work Summary For Research note 09, we conducted a macro and micro market stress test on the proposed 30% nominal put option hedging strategy for the …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 1, "risks": 1, "neutral": 0, "value": 90.2, "events": [ { "title": "结构性金属供给紧缩对全球通胀及美联储政策路径的影响", "summary": "支持（有界定）：铜/铝结构性缺口驱动的二次通胀足以硬化美联储终端利率逻辑，但机制是抬高底部、消除2026年降息期权，而非单独迫使加息。LME铜2026年6月初约13,800美元/吨、1月29日创14,527.50美元/吨历史高位，高盛上调年底预测至13,735；铝突破3,015.50美元/吨、对应…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "6月FOMC前的美国通胀再加速", "summary": "截至2026-06-13，美国5月CPI同比4.2%、核心CPI同比2.9%，PPI同比6.5%，非农就业增加172,000人且失业率为4.3%[S1][S2][S3]。结论是6月FOMC更可能按兵不动但沟通更鹰，原因是核心CPI不支持立即加息，但PPI管线压力、PCE高于目标和劳动力市场韧性使近…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 4, "neutral": 0, "value": 65.6, "events": [ { "title": "能源CPI与核心CPI背离下，美联储应上调通胀预测并取消2026年降息，但6月仍以鹰派按兵不动为基准。", "summary": "截至2026-06-14，能源冲击要求美联储显著上调2026年总体PCE预测，但核心上修应更温和。5月CPI同比4.2%、核心2.9%，加上4月PCE 3.8%、核心PCE 3.3%，说明压力真实但尚未全面扩散。基准建议是6月维持3.50%-3.75%，把2026年点阵图从一次降息改为不降息；若布…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海运费是通胀右尾输入项，不是足以单独决定Fed路径和A股仓位的宏观锚。", "summary": "截至2026-06-15，本卡否定“仅凭海运费即可重写2026年下半年Fed路径”的强结论。Drewry WCI上涨至3,549美元确实抬高通胀右尾，但Freightos FBX为2,888.80美元，二者存在22.9%的指数基差；IMF研究显示运费传导有约一年峰值时滞。当前Fed更应关注CPI、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-14：6月FOMC前的美国能源通胀冲击与Fed沟通风险", "summary": "截至2026-06-14，5月CPI的压力主要来自能源：CPI环比0.5%、同比4.2%，能源环比3.9%、同比23.5%，汽油同比40.5%；核心CPI仅环比0.2%、同比2.9%。但4月核心PCE同比3.3%，5月失业率仍为4.3%，说明Fed很难把能源冲击完全视为噪音。基准判断是6月FOMC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源驱动通胀再加速后的6月FOMC按兵不动倾向", "summary": "截至2026-06-15，本卡认为6月FOMC基准情形是维持利率不变且风险表述偏鹰。5月CPI与4月PCE仍高于2%目标，能源与ISM价格项带来再通胀压力；5月非农与PMI显示经济仍在扩张，但就业已温和降温。结论是Fed无需立即降息，也缺少足够理由加息，政策更可能停留在3.50%-3.75%等待更…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 1, "value": 82, "events": [ { "title": "6月FOMC与美国通胀再加速下的higher-for-longer风险", "summary": "截至2026-06-15，global-macro的基准判断是Fed仍大概率按兵不动，但不是鸽派暂停。5月CPI、5月PPI、4月PCE和5月就业数据共同显示名义需求仍偏强，能源冲击正在抬高总量通胀，而核心项尚未失控。因此，6月FOMC更可能通过预测和措辞完成higher-for-longer重置…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 美国CPI/PPI背离对大类资产风险溢价及下半年美股板块轮动影响研判", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: f4668e4f-748a-467a-bfe5-e36469d590db 结果摘要: Wrote handoff_response.md (10465 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-16", "supports": 0, "risks": 2, "neutral": 0, "value": 80.4, "events": [ { "title": "美国5月通胀反弹与6月FOMC前的全球宏观交易含义", "summary": "截至2026-06-15，本卡选择global-macro视角。结论是：6月FOMC大概率维持利率不变，但声明会让近期降息门槛更高。5月通胀主要由能源拉动，核心CPI环比仍温和；但4月核心PCE仍高、5月就业不弱、居民通胀预期仍偏高，使美联储很难快速转鸽。对市场而言，关键不是单纯押注利率上行，而是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "All three deliverables are written and verified. Research note 02 summary — Stance: SUPPORT…", "summary": "All three deliverables are written and verified. Research note 02 summary — Stance: SUPPORT (with one relative value refinement). I confirm Research …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 2, "neutral": 0, "value": 81.8, "events": [ { "title": "电网公用事业资产费率基数扩张对宏观通胀与货币政策的传导效应", "summary": "压力测试卡片01–05的宏观结论。直接渠道被夸大：电力属能源、按构造排除于核心PCE之外，达拉斯联储测算数据中心对总体PCE的拉动2026仅+0.05pp、2030升至+0.13pp[S1]；电力仅占CPI约2.5%[S11]，自测算直接总体贡献约0.10pp。但费率基数是30–40年价格水平锁定…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储6月鹰派暂停与美国高利率平台期风险", "summary": "截至2026-06-20，FOMC在6月17日维持3.50%-3.75%目标区间不变，但6月SEP把2026年政策利率中位数上调至3.8%、PCE通胀上调至3.6%、核心PCE上调至3.3%。5月CPI同比4.2%、能源同比23.5%，就业和GDPNow仍有韧性。结论是暂停更像鹰派按兵不动，不是宽…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-20", "supports": 0, "risks": 3, "neutral": 0, "value": 77.2, "events": [ { "title": "2026-06-21：6月FOMC鹰派按兵不动与全球美元利率重定价", "summary": "截至2026-06-21，本卡选择global-macro，研究6月17日FOMC以12-0维持3.50%-3.75%但SEP转鹰后的宏观含义。结论是：这不是清晰的新加息周期，而是一次鹰派信誉重置；降息叙事已被移除，一次额外加息成为有条件风险。5月CPI、4月PCE、稳定的4.3%失业率、2年期美…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美联储6月会议后的“降息交易”出清与再通胀风险", "summary": "工作日期2026-06-20。本卡选择global-macro，结论是美联储尚未承诺加息，但6月FOMC已把市场基线从自动降息转向higher-for-longer和加息风险。3.50%-3.75%的政策区间、2026年SEP中3.8%的年底利率中位数、5月CPI同比4.2%、5月非农+172,0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源成本压力对AI算力基础设施运营经济性与扩容节奏的影响", "summary": "压力测试并在核心机制上否定使命链条（油价×汇率→工业电价→AI经济性→扩容放缓）。链条两处断裂：(1)选错杠杆——电力是AI算力经济性的二阶变量，1GW数据中心年化TCO约85亿美元中能源仅约6亿（≈7%），服务器约60%；电价+20%仅抬升总算力成本约1.4%，不构成扩容开关，资本开支（GPU）…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 5, "neutral": 4, "value": 78.6, "events": [ { "title": "2026年伊朗战争能源冲击：供给扰动见顶与美联储/日本央行的滞后鹰派转向", "summary": "伊朗战争能源冲击大概率已过顶：布伦特从逾120美元回落至6月19日的80.57美元（周跌约8%），霍尔木兹有条件复航[S6][S8]。但美联储6月17日维持利率3.50–3.75%却将点阵图翻转为加息倾向（2026中值升至3.8%、PCE预测3.6%），日本央行6月16日加息至1.0%（1995年…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年伊朗战争能源冲击重新定价全球利率周期：通胀是暂时性还是持久性", "summary": "持续15周的2026年伊朗战争制造了一次供给侧能源冲击，把发达市场利率周期从降息扭转为鹰派按兵不动+加息倾向：美联储6月17日维持3.50%–3.75%但点阵图翻转为加息（2026年底中值升至3.8%、PCE上调至3.6%），日本央行加息至1.00%，美国5月CPI同比4.2%创近三年新高。但布伦…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "伊朗战争后的全球宏观转折：能源冲击消退，但美联储与日本央行维持鹰派带来的滞胀宿醉", "summary": "2026年伊朗战争（2/28美以空袭、3/4霍尔木兹关闭）将布伦特推至逾120美元，6/17美伊谅解备忘录使油价回落至约80美元。但通胀已从能源迁至核心：5月美国CPI总体同比4.2%（三年新高），核心升至2.9%，居住与食品双双加速。央行反应函数已上移——美联储6/17维持3.50–3.75%但…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "Brent跌破80美元/桶削弱了能源冲击的尾部风险 [S5]，但在库存和成品油缓冲修复前，美联储仍会把能源通胀视为延后降息而非立即宽松的理由 [S8][S9][S10]。", "summary": "截至2026-06-22，本卡支持上一张global-macro结论：Brent跌破80美元/桶是油价风险缓和，但不是实体供应链已经完全正常化 [S5]。EIA仍估计2Q26全球库存下降630万桶/日、3Q26下降760万桶/日 [S1]，IEA仍预计2026年全球供应下降390万桶/日 [S4]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 1, "risks": 3, "neutral": 0, "value": 88, "events": [ { "title": "美联储 6 月再定价：美国前端利率从降息预期转向维持高利率并存在加息风险", "summary": "截至 2026-06-23，6 月 FOMC 将联邦基金目标区间维持在 3.50%-3.75% [S1]，并把 2026 年 PCE 与核心 PCE 预测分别上调至 3.6% 和 3.3% [S2]。5 月 CPI 同比 4.2% [S3]、4 月核心 PCE 同比 3.3% [S4]、5 月非农…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026年6月美联储暂停、能源通胀与美国利率更久维持高位风险", "summary": "截至2026-06-22，本卡选择global-macro视角研究美联储6月暂停。核心发现是：FOMC虽维持3.50%–3.75%目标区间，但6月SEP上调2026年联邦基金、PCE和核心PCE预测；同时5月CPI能源项、4月PCE、5月就业与EIA油市预测共同削弱近期降息叙事。结论不是立即加息，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Summary of Work We have evaluated the mismatch risk between rising GPU power consumption an…", "summary": "Summary of Work We have evaluated the mismatch risk between rising GPU power consumption and grid capacity bottlenecks, demonstrating how this constr…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "2026-06-23：6月FOMC鹰派暂停与油价通胀冲击", "summary": "截至2026-06-23，6月FOMC维持3.50%-3.75%目标区间，但SEP把2026年PCE通胀中位数预测从3月的2.7%上修至3.6%，核心PCE从2.7%上修至3.3%。5月CPI-U同比4.2%，4月PCE同比3.8%，而5月非农新增172,000人、失业率4.3%，使Fed有空间优…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 3, "neutral": 0, "value": 81.4, "events": [ { "title": "美联储非对称紧缩把周期判断从“增长放缓触发宽松”改写为“通胀未再锚定前，增长疼痛不足以恢复Fed put”。", "summary": "截至2026-06-23，本卡支持卡片03：美联储已进入对通胀上行更敏感、对增长下行更迟钝的非对称紧缩体制。6月FOMC维持3.50%-3.75%，但SEP同时下调2026增长、上调PCE通胀；5月CPI 4.2%、4月PCE 3.8%、密歇根一年期预期4.6%，说明通胀预期仍未安全再锚定。OEC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "持续到2026/27冬季的能源冲击会把船舶/LNG链条的“可交付灵活性稀缺”转化为欧日更强通胀和更鹰派央行、美国降息延后以及中国PPI利润率压力，而不是全球同步的CPI失控。", "summary": "截至2026-06-23，本卡综合前序 LNG 运力与霍尔木兹油品冲击：若高油品价格和可交付 LNG 稀缺延续至2026/27冬季，欧元区 headline inflation 增量最大（约+1.5至+2.5pp），日本虽有补贴但 BOJ 正常化风险最清晰（约+0.7至+1.3pp），美国主要表现…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国同步利率上行与风险偏好下降是可信但非基准的尾部，缓冲应按10-15%的保险仓定档，而不是直接升为20%以上的核心大仓。", "summary": "截至2026-06-23，本卡综合前序缓冲讨论：国内“利率上行＋风险偏好下降”不是基准情景，但已是可信尾部。中国10年期仍约1.74%、LPR维持3.0%/3.5%、CPI仅1.2%，说明全面紧缩概率低；但美国10年期4.51%、PPI 3.9%、油品供给冲击与房地产/信用压力使尾部不能忽略。主观…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 1, "risks": 5, "neutral": 1, "value": 81.8, "events": [ { "title": "2026年能源通胀冲击后的美联储政策", "summary": "截至2026-06-25，6月FOMC维持3.50%-3.75%并非鸽派信号。能源冲击已把5月CPI推至4.2%同比，但核心CPI仍为2.9%同比，劳动力和国内需求尚未断裂。基准判断是美联储更可能维持高利率而非马上加息；真正的分水岭是未来PCE/CPI能否证明能源没有传导到核心价格和通胀预期。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] WTI at $65-68/bbl and Fed path threshold", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 2f15cf64-5f49-4af5-867a-05b15b121ef3 结果摘要: Wrote handoff_response.md (8806 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 日元干预警报与亚太半导体/CPO板块去杠杆风险评估", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: c956eaa1-0683-41fa-9c41-551222934843 结果摘要: Wrote handoff_response.md (10735 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "能源资产在通胀再加速路径下的对冲有效性分析", "summary": "本卡支持Research note 1-2的高利率持久化与能源作为类期权对冲的处方。2026年通胀本质是能源故事：能源仅占CPI篮子7.5%，却贡献5月环比涨幅近60%，汽油同比+40.5%，而核心反而降温。能源因此对能源主导的整体通胀对冲有效（高凸性、即时传导），却对3.3%核心PCE无效。但其…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 1, "neutral": 1, "value": 96, "events": [ { "title": "沃什时代美联储首次议息会议：前瞻指引淡化与点阵图意外鹰派的宏观影响", "summary": "截至2026-06-26，分析显示沃什首场会议维持利率3.50%-3.75%不变，但因中东能源冲击，联储将2026年底点阵图上调至3.8%，核心PCE调至3.3%。这与欧央行和日银6月加息形成G3同步紧缩。然而，6月下旬海峡重开且布油跌至72美元，突显鹰派政策滞后。沃什淡化前瞻指引，将加剧美债收益…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "美联储政策超调风险下的全球宏观经济周期与通胀路径演变", "summary": "截至2026-06-26，G3央行在6月同步偏鹰（美联储维持3.50%-3.75%但点阵图转向年内加息、欧央行存款利率上调至2.25%、日本央行加息至1%），却恰逢能源主导的通缩前沿：布伦特跌至约73.74美元/桶，较约105美元的油价假设低约30%，而5月核心PCE仍黏在3.4%、整体PCE破4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 1, "risks": 5, "neutral": 1, "value": 84.6, "events": [ { "title": "AI电力冲击本身是Fed应看穿的相对价格变化，但其持续、受供给约束且贴近核心服务成本的二阶传导，会让核心通胀在边际上更粘，并支持Fed更久维持限制性政策。", "summary": "截至2026-06-27，本卡综合前序结论：AI电力不是直接核心通胀项目，因为电力在2026年5月CPI-U权重仅2.495%，且core PCE剔除能源；但它可能通过商业电价、云服务、通信、医疗、设施与租金相关成本向60.158%权重的services less energy services传…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "同步鹰派转向是真实的风险窗口，但不是已被证实的新紧缩制度；Research note 9 字面论点应被否定，前序宏观主线应被有条件保留。", "summary": "截至2026-06-26，本卡否定 Research note 9 字面交接文本：它是目录检查类工作流伪影，不是宏观论点。但压力测试后有条件支持前序经济主线：G3 6月鹰派转向真实发生，Fed 维持3.50%-3.75%、ECB升至2.25%、BoJ升至1.0%；美国5月CPI/PCE仍高，说明不…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球宏观分化：华许治下美联储鹰派转向与人民银行结构性流动性改革，及大宗商品冲击", "summary": "在2026-06-26，霍尔木兹海峡地缘局势缓解令Brent原油暴跌20%至$72-$75/桶。然而，历史性能源冲击拉高美国5月CPI至4.2%、PCE至4.1%，促使美联储在6月会议（凯文·华许首秀）中展现鹰派倾斜，半数委员预测2026年加息。与之相对，中国5月CPI低位运行（1.2%）而PPI…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026年5月PCE发布后的美国通胀反弹与Fed反应函数", "summary": "截至2026-06-27，5月美国通胀反弹主要由能源驱动，但并非政策无关。BEA显示5月headline PCE同比4.1%、core PCE同比3.4%；BLS显示能源CPI同比23.5%、汽油同比40.5%，同时非农就业增加172,000人、失业率4.3%。Fed在6月17日维持3.50%-3…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 14, "neutral": 1, "value": 24, "events": [ { "title": "沃什联储6月鹰派转向：能源供给冲击下的'更高更久、加息可选'与全球外溢", "summary": "6月17日沃什首次主持的FOMC以12-0维持利率3.50%-3.75%，但点阵图转鹰：中值隐含2026年底约3.8%、9/18预计加息、17/18判断通胀上行，并取消前瞻指引。驱动因素是中东伊朗冲突引发的能源供给冲击——霍尔木兹受扰、布伦特一度较战前高逾60%，5月能源CPI同比+23.5%、核…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国5月通胀与6月FOMC：能源冲击下的更高更久利率路径", "summary": "截至2026-06-28，本卡选择global-macro视角研究美国5月通胀与6月FOMC。BEA显示5月PCE同比4.1%、核心PCE同比3.4%；BLS显示5月CPI同比4.2%，能源项同比23.5%；FOMC在6月17日以12-0维持3.50%-3.75%区间，同时SEP上调2026年通胀…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026-06-28：美国 May 2026 PCE 重新抬高美联储鹰派风险", "summary": "截至2026-06-28，May 2026 PCE 显示 headline PCE 同比4.1%、core PCE 同比3.4%，叠加 June FOMC 维持3.50%-3.75%以及 SEP 上修2026年通胀和联邦基金利率路径，近期降息叙事已被削弱。基准判断是 July 28-29 FOMC…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国通胀再加速与美联储加息风险重估", "summary": "截至2026-06-27，5月 PCE 与 CPI 均显示美国通胀重新偏强，劳动力市场仍有韧性，6月 FOMC 虽维持 3.50%-3.75%，但声明与 SEP 更接近“防通胀持续化”而非“准备降息”。结论是：宏观定价应从“等待降息”切换为“更久高利率并保留加息风险”，但这是能源冲击叠加韧性需求的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 8, "neutral": 2, "value": 64.4, "events": [ { "title": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "summary": "2026年6月17日Warsh首会以12-0维持利率3.50%-3.75%，但SEP中位点阵从隐含降息翻转为隐含加息（3.4%升至3.8%），18人中9人看至少1次加息、17人看通胀上行风险，2026年PCE上修至3.6%。背景是5月CPI同比4.2%反弹与韧性就业（年内约57万非农）。2年期美债…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "主要央行保持鹰派按兵不动并非过度收紧的政策错误，而是应对由服务业传导支撑的核心PCE粘性（预计H2维持在3.1%-3.4%）的必要防御，但这会使实体经济承受限制性实际利率压制，支撑久…", "summary": "本卡综合并支持前三张卡对能源脉冲退潮的判断，但指出核心通胀将出现背离：5月核心CPI放缓至2.9%，但核心PCE因服务二轮传导仍粘在3.4%。由于美联储盯防PCE，鹰派立场并非政策错误。然而CPI下行使限制性实际利率升至0.7%-1.15%，压制增长，支撑久期超配，并导致融资余额达28.13亿元、…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核心服务与住房通胀的粘性：美联储政策的后续驱动力", "summary": "能源转为通缩之际，核心服务与住房粘性足以支撑美联储鹰派、抵消能源通缩。关键在指标分化：整体PCE 4.1%/整体CPI 4.2%将因油价基数效应回落，但美联储盯住的核心PCE升至3.4%（2023年10月以来最高），由占核心PCE一半以上、运行约3.5%-3.9%的超级核心（剔除住房的核心服务）驱…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "6月点阵图不是简单过时，而是可能“对得不是理由”：能源理由过时，AI/电网与折旧通道支持更高更久，但不足以单独证明立即加息。", "summary": "截至2026-06-28，本卡综合认为：6月点阵图并非被最初的油价冲击证明，因布伦特已回落至约US$72/bbl；但AI/电网导致的结构性电价稀缺、核心PCE 3.4%、名义需求仍强以及AI-capex折旧悬挂，使“更高更久”仍有宏观依据。该通道不足以单独证明立即加息，因为电力CPI权重仅2.48…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 7, "neutral": 1, "value": 68.8, "events": [ { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 霍尔木兹海峡石油冲击与核心CPI传导", "summary": "致 投委会主席： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 1d956e69-9dc0-4197-8bd0-eb1630441435 结果摘要: Wrote handoff_response.md (11296 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 推导 PBOC 隔夜逆回购实际中标利率及中债收益率曲线非线性久期损失风险", "summary": "致 晨会主编： 协同任务已完成。执行方为 债券分析师。 模式: adhoc 会话: eb98b73e-bcc0-4d80-8508-8b06dfaaf2c6 结果摘要: Wrote handoff_response.md (10281 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Impact of global capacity pricing spikes on domestic transformer export marg…", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 324722b5-2f7b-4b91-aec7-ac5429f32a39 结果摘要: Wrote handoff_response.md (11765 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] } ], "href": "/investor/theses/thesis-e26ef69f73/", "networkHref": "/investor/research-network/#thesis-e26ef69f73" }, { "id": "thesis-d6b2c23ed6", "source": "tracked_thesis", "title": "航运与物流成本是油价、关税和供应链再通胀的二阶传导器", "titleEn": "Shipping and logistics costs are a second-order transmission channel for oil, tariffs, and supply-chain reflation", "lane": "航运物流/再通胀传导", "laneEn": "Shipping logistics and reflation transmission", "laneHref": "/investor/lanes/shipping-logistics-and-reflation-transmission/", "investableFocus": "航运物流/再通胀传导：601919.SH, 1919.HK, 600026.SH, 600018.SH, ZIM, MATX, BOAT", "stockUniverse": [ { "ticker": "601919.SH", "name": "中远海控", "market": "A-share", "role": "集运运价", "roleEn": "集运运价", "bucket": "core", "href": "/investor/stocks/601919.SH/" }, { "ticker": "1919.HK", "name": "中远海控", "market": "HK", "role": "集运运价", "roleEn": "集运运价", "bucket": "core", "href": "/investor/stocks/1919.HK/" }, { "ticker": "600026.SH", "name": "中远海能", "market": "A-share", "role": "油轮运价", "roleEn": "油轮运价", "bucket": "core", "href": "/investor/stocks/600026.SH/" }, { "ticker": "600018.SH", "name": "上港集团", "market": "A-share", "role": "港口吞吐", "roleEn": "港口吞吐", "bucket": "core", "href": "/investor/stocks/600018.SH/" }, { "ticker": "ZIM", "name": "ZIM Integrated Shipping", "market": "US", "role": "集运高 beta", "roleEn": "集运高 beta", "bucket": "core", "href": "/investor/stocks/ZIM/" }, { "ticker": "MATX", "name": "Matson", "market": "US", "role": "美国航运对照", "roleEn": "美国航运对照", "bucket": "core", "href": "/investor/stocks/MATX/" }, { "ticker": "BOAT", "name": "SonicShares Global Shipping ETF", "market": "US ETF", "role": "全球航运篮子", "roleEn": "全球航运篮子", "bucket": "core", "href": "/investor/stocks/BOAT/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" } ], "stockTickers": [ "601919.SH", "1919.HK", "600026.SH", "600018.SH", "ZIM", "MATX", "BOAT", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI" ], "coreView": "赛道：航运物流/再通胀传导。候选观察标的：601919.SH, 1919.HK, 600026.SH, 600018.SH, ZIM, MATX, BOAT。当前共有 40 条证据、1 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 66, "eventCount": 160, "firstSeen": "2026-05-02", "lastSeen": "2026-06-29", "supports": 4, "risks": 132, "neutral": 24, "analystCount": 18, "sourceTypeCounts": { "研究笔记": 115, "研究线程": 2, "分析协作": 21, "证据更新": 12, "研究主题": 9, "深度研究": 1 }, "topTerms": [ "GOES", "IPP", "油价", "Shipping", "logistics", "costs", "second-order", "transmission", "channel", "tariffs", "supply-chain", "reflation" ], "whatChanged": [ "2026-06-29: 上游取向硅钢和电工铜的产能稀缺与关税高墙正转嫁为严重成本通胀，侵蚀设备制造厂毛利并拉长交付，导致AI电力落地面临实质性溢价风险。", "2026-06-29: 取向电工钢（GOES）由于美国本土单一垄断和地缘壁垒呈现严重的全球双轨制定价，高牌号硅钢2-4年的扩产认证周期和铜价高企构成了无法通过资本抹平的物理瓶颈，将通过毛利侵蚀与交期恶化对变压器及算力并网构成二次约束。", "2026-06-29: 关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "2026-06-29: 在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15.36倍AI潜在电力需求，算力扩张将受限于刚性电力配给，投资配置应向短交期电力变压器出口链与物理红利安全垫倾斜。", "2026-06-29: 电力设备需求爆发下的原材料价格波动及其对利润率的压力测试" ], "practical": { "sectorLane": "航运物流/再通胀传导", "sectorLaneEn": "Shipping logistics and reflation transmission", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "probe", "actionLabel": "小仓验证", "actionTone": "yellow", "riskBudget": "验证仓：1-3% thesis risk", "portfolioRole": "验证候选", "weightGuide": "1-3% thesis risk", "timeHorizon": "2-6 周", "portfolioUse": "适合用小仓位验证，不应在证据分歧扩大时追高。", "positionSizing": "建议按 1-3% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "601919.SH", "name": "中远海控", "market": "A-share", "role": "集运运价", "roleEn": "集运运价", "bucket": "core", "href": "/investor/stocks/601919.SH/" }, { "ticker": "1919.HK", "name": "中远海控", "market": "HK", "role": "集运运价", "roleEn": "集运运价", "bucket": "core", "href": "/investor/stocks/1919.HK/" }, { "ticker": "600026.SH", "name": "中远海能", "market": "A-share", "role": "油轮运价", "roleEn": "油轮运价", "bucket": "core", "href": "/investor/stocks/600026.SH/" }, { "ticker": "600018.SH", "name": "上港集团", "market": "A-share", "role": "港口吞吐", "roleEn": "港口吞吐", "bucket": "core", "href": "/investor/stocks/600018.SH/" }, { "ticker": "ZIM", "name": "ZIM Integrated Shipping", "market": "US", "role": "集运高 beta", "roleEn": "集运高 beta", "bucket": "core", "href": "/investor/stocks/ZIM/" }, { "ticker": "MATX", "name": "Matson", "market": "US", "role": "美国航运对照", "roleEn": "美国航运对照", "bucket": "core", "href": "/investor/stocks/MATX/" }, { "ticker": "BOAT", "name": "SonicShares Global Shipping ETF", "market": "US ETF", "role": "全球航运篮子", "roleEn": "全球航运篮子", "bucket": "core", "href": "/investor/stocks/BOAT/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" } ], "stockTickers": [ "601919.SH", "1919.HK", "600026.SH", "600018.SH", "ZIM", "MATX", "BOAT", "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI" ], "primaryStocks": [ { "ticker": "601919.SH", "name": "中远海控", "market": "A-share", "role": "集运运价", "roleEn": "集运运价", "bucket": "core", "href": "/investor/stocks/601919.SH/" }, { "ticker": "1919.HK", "name": "中远海控", "market": "HK", "role": "集运运价", "roleEn": "集运运价", "bucket": "core", "href": "/investor/stocks/1919.HK/" }, { "ticker": "600026.SH", "name": "中远海能", "market": "A-share", "role": "油轮运价", "roleEn": "油轮运价", "bucket": "core", "href": "/investor/stocks/600026.SH/" }, { "ticker": "600018.SH", "name": "上港集团", "market": "A-share", "role": "港口吞吐", "roleEn": "港口吞吐", "bucket": "core", "href": "/investor/stocks/600018.SH/" }, { "ticker": "ZIM", "name": "ZIM Integrated Shipping", "market": "US", "role": "集运高 beta", "roleEn": "集运高 beta", "bucket": "core", "href": "/investor/stocks/ZIM/" }, { "ticker": "MATX", "name": "Matson", "market": "US", "role": "美国航运对照", "roleEn": "美国航运对照", "bucket": "core", "href": "/investor/stocks/MATX/" } ], "instrumentTags": [ "601919.SH", "中远海控", "1919.HK", "中远海控", "600026.SH", "中远海能", "600018.SH", "上港集团", "ZIM", "ZIM Integrated Shipping", "MATX", "Matson" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "高：分歧需要拆解", "deepResearchReason": "证据深度 100，来源 6 类，近期事件 81 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 160, "sourceDiversity": 6, "analystCount": 18, "recentEvents": 81, "supportRatio": 0.03, "riskRatio": 0.82, "conflictScore": 93.8, "evidenceDepth": 100 }, "score": 58.7 }, "events": [ { "date": "2026-06-29", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.en.md (9614 bytes) — Verdict: …", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "risk", "supportLabel": "风险", "score": 23.25, "matchedTerms": [ "IPP", "transmission", "evidence", "risk", "dominant", "inflation", "fall", "supply" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] Impact of global capacity pricing spikes on domestic transformer export margins", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 324722b5-2f7b-4b91-aec7-ac5429f32a39 结果摘要: Wrote handoff_response.md (11765 bytes); wrote handoff_response.en.md (12438 bytes) — pri…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 15.75, "matchedTerms": [ "transmission", "freight", "fall", "pass", "port", "transformer", "power" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]，PPI-CPI剪刀差隔离了政策利率。任务设想的'成本→CP…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "GOES", "macro", "goes" ], "href": "" }, { "date": "2026-06-29", "title": "AI算力基建瓶颈与多资产杠铃策略的逻辑闭环及定量模型终期质量审计", "summary": "终期QA审计给出「需修订（有条件通过）」。会期主轴稳健，核心论点逐节点重定价、瓶颈迁移至可通电MW、并以杠铃对冲滞胀尾部端到端自洽；可复算的定量链条均对得上，本卡问题的2.30%保本不良率上限算术正确（125bp再贷款利差−10bp运营−50%LGD，NPL=2.30%净利差归零）。PJM容量价333.44、GOES价差5.16×、5GW用铜44.44%、…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.5, "matchedTerms": [ "GOES", "pass", "goes" ], "href": "" }, { "date": "2026-06-29", "title": "[auto-handoff] 评估PJM等容量电价暴涨对国内出海变压器毛利率的传导", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 1d2b9e99-6767-49f3-aa4b-4520c45f9a0b 结果摘要: Wrote handoff_response.md (9124 bytes); wrote handoff_response.en.md (9491 bytes) — The 1…", "sourceType": "mailbox", "sourceLabel": "分析协作", "supportType": "neutral", "supportLabel": "观察", "score": 13.75, "matchedTerms": [ "costs", "freight", "supply", "port", "tariff", "transformer" ], "href": "" }, { "date": "2026-06-29", "title": "电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "summary": "本卡压力测试 Research note 6 的'调价条款已隔离利润'结论，认为其仅有条件成立。主流调价条款（MEC）挂钩有流动期货的铜/铝（出口超级周期转嫁率约80%[S1]），却普遍排除或封顶无流动指数的取向电工钢GOES——而GOES占COGS约28%[Research note 6]且处2026-2028结构性短缺。自测算显示：铜+30%/GOES…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-29", "title": "在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15.36倍AI潜在电力需求，算力扩张将受限…", "summary": "截至2026-06-30，本研究对Jevons悖论激发的15.36倍潜在AI用电需求（相当于727.6 GW新增平均负荷或6,374 TWh年用电）进行了极限物理压力测试。结论表明，在2030年前，由于北美变压器交期拉长至48-60个月、特高压直流送电网架高壁垒（全球需新建73条通道）以及取向硅钢（GOES）产能与高压试验大厅测试容量的物理刚性制约，没有任…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-29", "title": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级云厂商资本开支延期兑现。", "summary": "本卡片从制造和供应链视角对电网与电力约束进行压力测试。截至2026-06-29，关键电力设备（变压器与重型燃机）由于极高工艺与测试门槛、上游取向电工钢（GOES）瓶颈及寡头垄断，交期拉长至3-7年，生产机位已售罄至2029-2030年。云厂商BTM发电加剧了燃机机位争夺。预计2026-2027年该瓶颈将导致云厂商约800亿-1100亿美元资本开支递延兑现。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "中国GOES与铜体系在基准情形下是西方AI建设的泄压阀，但政策阀门已被安装，西方数据中心应把中国来源依赖重定价为150-300bp的分层资本开支风险溢价。", "summary": "截至2026-06-28，本卡压力测试GOES与铜的中国供给集中度。结论：基准情形不是中国对GOES/铜禁运，而是“带政策期权的泄压阀”。中国具备西方AI数据中心建设所需的GOES与铜精炼/冶炼规模，但关税、原产地、反倾销与出口管制先例会把供给集中度转化为资本开支风险溢价。建议在Research note 04的100-200bp材料/设备溢价上，对中国来…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.75, "matchedTerms": [ "GOES", "macro", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "在中国的精铜结构性进口依存与紫金海外资产高地缘折价背景下，应将有色敞口拆分为“在岸GOES技术壁垒（宝钢股份）+离岸实物/矿山多头”以规避紫金主权尾部风险，并通过“黄金（14.15%）+大水…", "summary": "截至2026-06-28，本卡基于有色冶炼端0加工费与出口管制非对称性构建了跨资产对冲框架。由于中国铜进口依存高（87%）且紫金海外资产折价大（占EBITDA的50%），建议低配A股冶炼及紫金，将铜敞口拆分为在岸宝钢及离岸实物/矿山多头（Freeport/BHP）。为对冲地缘尾部，引入实物黄金（14.15%）与水电红利（15.85%）实现36.51%等风险…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.5, "matchedTerms": [ "GOES", "port", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "原始40/32/12/8/8 sleeve在AI需求相关性趋近+1时尾部风险过高，阶段1是强制去杠杆，阶段2才接近可接受，但决定性保护来自降AI beta和提高现金，而非小仓位黄金/美元/利…", "summary": "截至2026-06-29，本卡对40/32/12/8/8 AI物理瓶颈sleeve做正式压力测试。原始组合在1个月99% Student-t口径下VaR/CVaR为28.6%/37.8%[自测算]，阶段1降至23.2%/30.7%[自测算]，阶段2降至18.5%/24.5%[自测算]。结论是：原始sleeve未通过尾部风险测试；阶段1必须立即执行，阶段2才…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 14.5, "matchedTerms": [ "GOES", "risk", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "变压器交期拉长至3-5年与并网排队超36个月的物理刚性瓶颈，配合美欧贸易保护关税对GOES硅钢供应链的阻断，已对AI算力扩张的通电效率构成无法用资本单纯解决的长期结构性壁垒，造成明显的Cap…", "summary": "基于并网排队与变压器短缺的双重物理约束，我们验证电力基础设施瓶颈已构成AI算力扩张的长期结构性阻碍。数据中心通电时间已拉长至36-60个月，变压器交期延至3-5年，主要厂商积压订单超百亿美元。这些延误对项目NPV和IRR产生重大侵蚀，100MW数据中心延误3年将损失51.63%的NPV。美国Section 301/232关税与欧盟CBAM机制则通过GOES…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力需求增长对电网侧扩容及配电资产的压力测试", "summary": "压力测试结论：前序主线被证实并强化。2026年AI落地的约束性瓶颈已非GPU，而是电网边缘——电力变压器、中高压开关设备、取向硅钢（GOES）与接网排队。大型变压器交期3–5年、全球产能利用率约98%、美国接网积压1,500–2,300GW、枢纽接入排队4–7年，致美国2026约12GW管线仅约4GW在建（隐含约67%滑期，自测算）。这把AI算力部署拉长而…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "AI电力基础设施供应链瓶颈：电力变压器与关键配电硬件的产能缺口", "summary": "对Research note 01以变压器为中心的瓶颈论断做压力测试。结论：GIS与高压断路器产能确实紧张（≥145 kV断路器交期已超12个月、整套GIS需60–100周、供给集中于约4家OEM），但它是次级协同瓶颈，而非近期核电并网的绑定约束。原因在于结构：近期核电对AI的供给主力是反应堆重启与增容，二者均复用既有开关站，断路器/GIS短缺在发电侧几乎…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "summary": "本卡支持 Research note 02「上游原材料是最深约束」的结论，并锐化为一组非对称结构。GOES：自2019年变压器需求暴增（电力+119%、GSU+274%），远超GOES市场约5.4–5.6%的CAGR；GSU供给缺口2025年约100%，因美国唯一GOES生产商Cleveland-Cliffs（Butler Works，扩产2028上线）及…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" }, { "date": "2026-06-28", "title": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "summary": "本卡压力测试逻辑链：GOES约束不会「抵消」核能加速，因为二者作用于电力链不同节点——核能重启在发电/并网侧本就是GOES-轻（复用既有GSU与开关站），属于一种GOES套利；但GOES在负荷侧与输配路径重新绑定，沿用Research note 01的NPV模型测算，核能锚定型项目(≈1.5年延迟,NPV≈−27%)相对并网型绿地(3年,−51.6%)只回…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 13.5, "matchedTerms": [ "GOES", "goes" ], "href": "" } ], "timeline": [ { "date": "2026-05-02", "supports": 0, "risks": 1, "neutral": 1, "value": 39, "events": [ { "title": "AI 物理超级周期叙事不是神话，但其投资价值高度层化：稳态电力运营商和一线 OEM 寡头拥有最强的 2026–2030 年利润归属确定性，而 EPC 承包商、中端 OEM 和无差异化…", "summary": "对九张卡的全链路审计确认：AI 物理超级周期叙事内部一致、证据充分，不存在不可逾越的内生性逻辑缺陷。五项主要张力（需求循环假设、EPC 系统性风险、AI 效率盲区、时间碎片化、分叉悖论）均真实存在但可管理。2026–2030 年利润归属层级清晰：稳态电力运营商（CEG、VST）与一线 OEM 寡头…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Global Supply Chain & Inflation Re-coupling", "summary": "Global Supply Chain & Inflation Re-coupling Is the synchronous jump in freight rates and satellite-monitored port congestion signaling a more resilie…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-05", "supports": 0, "risks": 3, "neutral": 1, "value": 28.4, "events": [ { "title": "最暴露于GOES/LPT延迟的不是所有数据中心，而是2026–2028年带电依赖新增高压变电和区域输电升级、且未提前锁定LPT/GOES供应的公用事业辖区。", "summary": "本卡片综合并支持前序结论：GOES约束最可能把美国AI电力瓶颈表现为6–18个月带电滑期，而不是超大规模云厂商取消资本开支。最高暴露区域是PJM的Dominion/NOVEC/ODEC北弗吉尼亚和AEP Ohio Central Ohio，其次是ERCOT的Oncor、MISO South的Ent…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 油价冲击对CNH的影响", "summary": "致 外汇策略师： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: 480c6f05-bd09-4eb3-b510-cfd771ed7b5b 结果摘要: Wrote handoff_response.md (8,813 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "2026-2028年的关税和贸易限制会先加剧而非缓解AI基建的“铁”瓶颈，因为美国保护的国内GOES和变压器产能不足以快速替代高依赖进口链条。", "summary": "本卡对第06卡做压力测试并支持“硅-铁失配”判断：2026-2028年贸易政策不会从零制造变压器短缺，但会把短缺转化为地缘政治分配问题。DOE数据显示，美国LPT进口依赖度为82%，本土GOES只能满足12-20%需求，GOES和铜导线各约占LPT成本25%。Section 232在2026年4月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "\"新质生产力\"政策下的中国电网设备出海：退税政策校准与贸易壁垒对冲策略 第8张卡片（共10张）| 中国宏观分析师 | 立场：综合 本期会话主题： 美国 AI 数据中心负荷增长与公用事…", "summary": "\"新质生产力\"政策下的中国电网设备出海：退税政策校准与贸易壁垒对冲策略 第8张卡片（共10张）| 中国宏观分析师 | 立场：综合 本期会话主题： 美国 AI 数据中心负荷增长与公用事业电网灵活性 本卡主题： \"新质生产力\"框架下的电网设备出口退税政策调整与 Section 232/301 关税规避…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-06", "supports": 0, "risks": 3, "neutral": 1, "value": 29.8, "events": [ { "title": "[auto-handoff] 航运板块估值重估", "summary": "致 另类数据分析师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 5a9de898-07ab-4ac8-9257-93680cf213ca 结果摘要: Wrote handoff_response.md (3510 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "盟友联盟对中国电网设备的部分排除是一项主动用 6–12 个月送电延迟和 15–25% 采购成本溢价换取供应链韧性的宏观选择，结果是延展盟友设备 OEM 超级周期、部分上限美国 AI …", "summary": "本卡作为 10 张卡片的最终综合，整合了从需求（Research note 01）、firm 负荷修正（Research note 02）、供给瓶颈（Research note 03）、设备防御性（Research note 04）、定价权（Research note 05）、跨资产轮动（Rese…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国232条款对进口钢铁征收25%关税，对铝征收10—25%关税，涵盖中国取向硅钢（GOES）。", "summary": "美国232条款对进口钢铁征收25%关税，对铝征收10—25%关税，涵盖中国取向硅钢（GOES）。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "望变电气（603191.SH）是A股唯一整合取向硅钢生产、铁芯加工和变压器成品的全产业链标的。", "summary": "望变电气（603191.SH）是A股唯一整合取向硅钢生产、铁芯加工和变压器成品的全产业链标的。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-07", "supports": 1, "risks": 3, "neutral": 0, "value": 36.2, "events": [ { "title": "最高等级PDR GOES仅由日本制铁（Nippon Steel）供应，美国LPT生产商主要从日韩进口高质量GOES。", "summary": "最高等级PDR GOES仅由日本制铁（Nippon Steel）供应，美国LPT生产商主要从日韩进口高质量GOES。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球取向硅钢（GOES钢）规模化生产商约3至4家，包括Cleveland-Cliffs/AK Steel、Tata Steel、ThyssenKrupp、Nippon Steel。", "summary": "全球取向硅钢（GOES钢）规模化生产商约3至4家，包括Cleveland-Cliffs/AK Steel、Tata Steel、ThyssenKrupp、Nippon Steel。", "sourceLabel": "证据更新", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "美国DOE表示大型电力变压器（LPT）交付周期可达并超过36个月，取向硅钢（GOES）和CTC铜线各约占LPT生产成本的25%", "summary": "美国DOE表示大型电力变压器（LPT）交付周期可达并超过36个月，取向硅钢（GOES）和CTC铜线各约占LPT生产成本的25%", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Shipping costs and agri-trade premiums", "summary": "致 农业行业分析师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 359b39fc-a5fa-40e4-b364-f0720363f8a5 结果摘要: Wrote handoff_response.md (3767 bytes); wrote handoff_respo…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-08", "supports": 0, "risks": 0, "neutral": 1, "value": 50.6, "events": [ { "title": "[auto-handoff] Geopolitical Impact on Commodities", "summary": "致 首席量化师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 2b55b600-1f54-4e9e-9ebe-b3a4f13f9a65 结果摘要: Wrote handoff_response.md (9,337 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-10", "supports": 0, "risks": 0, "neutral": 1, "value": 52, "events": [ { "title": "日本新日铁（Nippon Steel）与JFE合计占据全球取向硅钢（GOES）市场约28–32%的份额。", "summary": "日本新日铁（Nippon Steel）与JFE合计占据全球取向硅钢（GOES）市场约28–32%的份额。", "sourceLabel": "证据更新", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-11", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "[auto-handoff] Hormuz Strait Blockage Risk Assessment", "summary": "致 首席量化师： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: 9dfc495e-cdb3-4496-b101-15e84bc93683 结果摘要: Wrote handoff_response.md (10508 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-12", "supports": 0, "risks": 2, "neutral": 0, "value": 39.8, "events": [ { "title": "在 150 美元的 Brent 原油价格下，高价值 TMT 硬件的空运燃油附加费可能使总运输成本增加 30-50%。", "summary": "在 150 美元的 Brent 原油价格下，高价值 TMT 硬件的空运燃油附加费可能使总运输成本增加 30-50%。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "美国本土仅有一家GOES生产商。", "summary": "美国本土仅有一家GOES生产商。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-13", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "‘物理交付死结’正在击穿‘地缘政治围栏’，迫使云巨头在非敏感地区结构性转向中国电力供应链以维持AI交付节奏，从而导致全球电力设备定价权从‘积压稀缺性’转向‘制造交付速度’。", "summary": "截至2026-05-13，本卡对电力设备交付瓶颈进行了全球宏观层面的压力测试。由于西方变压器交期拉长至36-48个月（由于GOES短缺和劳动力瓶颈），而中国仅需12-15个月，西方超大规模云服务商正通过“非敏感地区（NSR）”策略结构性地转向中国电力供应链。这种“交付周期套利”正在削弱西方工业巨头…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-14", "supports": 0, "risks": 7, "neutral": 1, "value": 15.6, "events": [ { "title": "材料底座（GOES吨数+重型锻造槽位）在2027-2029年把美国AI可交付电力压在12-17 GW/年，与电网与OEM约束重叠确认10-14基准、14-18高情景的硬上限，价值进一…", "summary": "截至2026-05-14，本卡对前序结论进行压力测试并基本确认：即便变压器与燃机OEM在2026-2028年兑现30-50%扩产，取向硅钢（GOES）、高导铜、大型铸锻件与关税/出口管制会在2027-2029年重新把美国AI可交付电力压在12-17 GW/年的材料可行区间——与第2/5卡的10-1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Section 232、MOFCOM与CBAM不缓解反而强化2027-2029年美国AI电力硬上限，把12-17 GW/年材料带压至12-14基准、14-16高情景的政策调整带，并把…", "summary": "截至2026-05-14，我支持第6卡判断：Section 232衍生品扩展、MOFCOM出口许可与CBAM正式期三套规制净效应是强化而非缓解2027-2029年美国AI电力交付硬上限。三者把物理稀缺转化为政策定价、许可配给的稀缺：Section 232把进口变压器/铁芯/GIS推近税线并令日韩H…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "部分大型电力变压器（LPT）所需最高等级GOES PDR的唯一制造商为日本制铁（Nippon Steel）。", "summary": "部分大型电力变压器（LPT）所需最高等级GOES PDR的唯一制造商为日本制铁（Nippon Steel）。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "大约95%的美国变压器铁芯使用取向硅钢 (GOES)。", "summary": "大约95%的美国变压器铁芯使用取向硅钢 (GOES)。", "sourceLabel": "证据更新", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-16", "supports": 0, "risks": 0, "neutral": 1, "value": 59, "events": [ { "title": "欧美壁垒把UHV出口叙事从“无摩擦加速”改成“非美欧订单支撑、欧美订单打折”，但尚不足以推翻40-60bps的AI Power/UHV配置腿。", "summary": "截至2026-05-17，欧美贸易壁垒升级不构成TBEA/思源电气等UHV出口标的的基准海外订单量断崖，但会对美国/欧盟相关投标造成中标概率、交付节奏和毛利率折价。美国25% Section 301已是大型变压器硬成本劣势，欧盟FSR与GOES保障调查会放大公共采购审查。基准对basket总海外新…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-17", "supports": 0, "risks": 6, "neutral": 0, "value": 21.4, "events": [ { "title": "上游 GOES 与铜的缺口在 2026–2028 仍能支撑电力设备 OEM 的定价权，但护城河正从「物理短缺」迁移到「合格认证 + 长协」，2029+ 的长期可持续性是 stress…", "summary": "从材料端压力测试 Card-01/02 关于电力设备 OEM 定价权可持续的判断。GOES 真实瓶颈是 0.23–0.27 mm 高磁感薄规格而非粗钢，名义产能 2026→2028 增 12–15%、需求增 14–18%，高磁感缺口在 2028 年前不会被弥合；铜的真正瓶颈在精炼 (TC/RC 历…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 电力主题升为主线，但只通过 HVPT / 高等级 GOES / OFR 三细分篮子表达，仓位以订单兑现率和送电进度为锚。", "summary": "截至 2026-05-18，本卡综合判断：AI 电力主题应通过 HVPT / 高等级 GOES / OFR 篮子升为策略主线，但口径必须严格——按 OFR 和送电里程碑解锁仓位，而非按签约 MW。篮子分三细分：A. HVPT 认证产能 45–55%（GE Vernova、Siemens Energ…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "容量轮动论点在方向上通过中国宏观压力测试，但需降低对美欧设备多头的确信度、对冲 2026-11 关键矿物悬崖、并按地理与币种分叉半导体设备与长久期信用的判断。", "summary": "中国宏观对资产配置师的容量轮动论点做压力测试：方向上支持（东数西算瓶颈迁移、国网十五五前置 capex、PBoC 偏鸽都印证），但提出三点重大挑战。S1：中国可通过特变电工/思源/金盘等绕道墨西哥-沙特-印尼出口，2026 后压缩美国变压器交期 6–12 个月，打掉 Hitachi/Eaton 终…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中美电力设备脱钩将引发全球AI电力硬件的“供给侧休克”，建议通过下调中国OEM权重并锁定非中GOES瓶颈来对冲地缘政治导致的OFR塌陷。", "summary": "针对中美电力设备脱钩的压力测试显示，中国掌控全球45%的高等级GOES和35%的HVPT产能。若美方实施全面进口禁令，将导致全球订单兑现率（OFR）下降850-1200个基点，交付周期拉长至220周以上。建议将篮子中中国OEM权重从15%下调至8%，重点回避高美国敞口标的，将权重转移至非中背景的G…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "主权 AI 需要时间折扣但需求未崩，最佳配置是 74% AI 基础设施上行 + 26% 宏观对冲，而不是放弃双腿交易或满仓追逐线性兑现。", "summary": "2026-05-19：本卡综合 cards 01-04，支持“回补半导体 + 保留电力腿”，但将 card-04 的主权 AI 30-40% 时间折扣转化为仓位约束。基准跨资产袖珍组合为：32% 半导体 / AI 基础设施设备、24% 电力设备、18% IPP / 带电 IDC、16% 长久期利率…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "电气设备产能瓶颈与AI硬件下修共同驱动相对收益，超配电网龙头/低配集成商的12个月Spread基准+25%、乐观+37%。", "summary": "在变压器/GOES物理瓶颈坐实、下游AI硬件2026H2将下修15%-20%背景下，未来12个月战术配置明确为'超配电网设备制造商（订单可见度≥3年：国电南瑞/中国西电/平高/思源/特变/金盘）+ 低配AI服务器系统集成商（工业富联/浪潮/曙光/紫光）'。基准情形多空Spread +25%，乐观+…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-22", "supports": 0, "risks": 1, "neutral": 2, "value": 61.6, "events": [ { "title": "Re-estimate supply chain price shock transmission to U.S. core PCE", "summary": "Re-estimate supply chain price shock transmission to U.S. core PCE In 'Thesis 1: Step 4', direct a high-priority handoff to global-macro to re-estima…", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "全球变压器与取向硅钢供应链地缘博弈与关税壁垒对AI算力扩张的二次限制", "summary": "高阶GOES与EHV变压器产能高度集中于中日韩（GOES全球约80%+，LPT亚洲约55–60%产量）。美国Section 232钢铁关税、USITC GOES反倾销复审、LPT 232调查配额、DOE大电网EO 13920，叠加欧盟CBAM、FSR与NZIA本土化要求，正在Research no…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海运另类高密堆积与A股光模块地缘砸盘的实虚背离", "summary": "海运另类高密堆积与A股光模块地缘砸盘的实虚背离 Shipping Alt-Data Congestion vs. A-Share Tech Export Panic 5月22日出海硬件股因出口管制传闻放量暴跌，但卫星与AIS港口另类数据显示排产与订舱实际极为饱和、运价大涨。这种物理实物出海的极度繁…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 3, "neutral": 0, "value": 45, "events": [ { "title": "We have successfully completed Research note 03/08 of the research thread session efecefe6 …", "summary": "We have successfully completed Research note 03/08 of the research thread session efecefe6 2a06 4881 87e9 ccec1f502077 as the Chief Economist. Stance…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" }, { "title": "Hormuz 谈判窗口是否在错误定价通胀尾部风险", "summary": "Hormuz 谈判窗口是否在错误定价通胀尾部风险 Is the Hormuz negotiation window mispricing inflation tail risk? 若 Iran 提案不变但军事行动延后，油价、美元、黄金和通胀预期哪个市场在错？", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "GOES（电磁取向硅钢）与电气用铜的全球供应已极度集中于少数中、日、韩、俄、智利-印尼厂商，美国50%钢铁关税、中国出口许可扩容与NLMK制裁三大地缘冲击共振，将把变压器交付时滞由当…", "summary": "本卡对工业制造分析师卡片06的'GOES紧缺是变压器交期飙至150-200周关键'判断进行逆向压力测试，结论强化而非削弱前序判断。HiB高磁感薄规格GOES全球仅Baowu/Nippon Steel/POSCO/NLMK四家可量产，合计仅~900-1,000kt/yr，美国仅Cleveland-C…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 2, "neutral": 1, "value": 56.8, "events": [ { "title": "I have completed the global macro analysis for Research note 01 of the research thread sess…", "summary": "I have completed the global macro analysis for Research note 01 of the research thread session b0a5df37 4aba 4443 9696 a334f45b9dd4 . Key Findings & …", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "A 股电力设备正从‘国内周期标的’重估为‘全球 AI 基建稀缺核心供货商’，其显著的交期优势与估值洼地支撑权重上调。", "summary": "A 股电力设备龙头正凭借 10-12 个月的极致交期（对比海外 36-48 个月）和对取向硅钢 (GOES) 的产能垄断，成为全球 T&D 瓶颈的唯一‘泄洪口’。国家电网 4 万亿（2026-2030）投资计划及 8500 台具身智能设备招标，确立了 A 股‘AI+电力’的增长底部。目前 A 股龙…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "实物原油供给赤字的长期化将打破央行的通胀容忍红线，迫使全球宏观叙事从“预防式降息”向“滞胀紧缩”急剧反转，触发系统性Risk-Off抛售。", "summary": "本报告通过宏观压力测试指出，若Hormuz断供拖延至Q3且Brent油价直奔$130，美联储将被迫完全撤销2026年降息指引甚至重启加息讨论，10年期美债有重返4.8%的风险；欧洲央行将陷入深陷滞胀与主权利差走阔的死局；日本央行则极大可能被迫紧急加息以保卫日元。这将彻底逆转当前的降息计价，引发全球…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 3, "neutral": 0, "value": 49.2, "events": [ { "title": "把AI久期空头与电网OEM/GOES/CLF多头拼成的组合净Beta+0.45、净久期−1.5年，是'高利率持续'路径下的反AI久期+实物瓶颈租金套利，需配置USD多头FX叠加并设置…", "summary": "2026-05-28 synthesize：把第01–05卡拼回组合（USD 100mm基币）。AI久期空头−15%、AI硬件多+10%、全球电网OEM+30%、亚洲OEM+10%、GOES上游+8%、CLF垂直一体+7%、自动化+5%、现金+25%。净Beta约+0.45、净久期约−1.5年（受…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI 电力缺口是真实但温和的核心 PCE 楔，叠加电网设备供应链的中美对抗，将以「延迟降息 + 类滞胀」形态而非「重置通胀目标」形态影响美联储，并把宏观风险从拥挤的一阶 AI 硬件转…", "summary": "AI 电力交付滞后通过三条渠道（批发电价→服务、云算力涨价→SaaS、变压器 PPI→capex 平减指数）对美国核心 PCE 形成 2026 +8~18 bp、2027 +10~22 bp 的持续抬升（自测算），不足以重置 2% 目标但足以让核心 PCE 2027 年底仍卡在 ~2.4%，相对 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "重型电力设备OEM收入上能承接需求、时间上承接不了——订单簿2.5–4年可见度+定价权+利润率扩张构成对科技久期的反向对冲，核心超配GEV/ENR.DE/ETN/Hitachi系，中…", "summary": "2026-05-28结论：本卡支持第03卡的电网瓶颈论断。Siemens Energy Grid Tech订单/营收比2.28×、储备490亿欧元；GE Vernova Electrification订单+86%、设备储备390亿美元、B2B约2.5×；Eaton电气储备同比+48%、数据中心订单…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 62.6, "events": [ { "title": "[auto-handoff] 钢厂厂库逆势反弹与表需季节性回落对二季度固定资产投资（FAI）的宏观指示性评估", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 31e6c7aa-6da2-4c16-8f9d-f73858f40650 结果摘要: Wrote handoff_response.md (15034 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 1, "value": 67, "events": [ { "title": "I have completed the analysis for Research note 01/08 as the Global Macro Analyst , examini…", "summary": "I have completed the analysis for Research note 01/08 as the Global Macro Analyst , examining the unseasonal surge in global freight rates and its no…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "截至2026-05-29，美国贸易壁垒和本土化采购要求会把AI数据中心并网延迟从设备短缺放大为合规与地缘供应链问题，中国及亚洲供应只能边际补缺，难以快速突破美国瓶颈。", "summary": "截至2026-05-29，本卡支持前序结论：AI并网瓶颈不只是变压器和GOES的物理短缺，美国Section 232、Section 301和Buy America规则把进口替代变成更慢、更贵、更难合规的路径。中国2025年GOES产量和出口创新高，但中国原产变压器与电工钢面对额外关税和原产地审查…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 1, "risks": 1, "neutral": 0, "value": 73.4, "events": [ { "title": "电网瓶颈真实但可移动：hyperscaler把边际GW送往ERCOT/海湾/东盟，许可改革与表后燃机/核电是旁道，关税让设备短缺更集中——结果不是美股公用事业广义牛市，而是少数OEM…", "summary": "从美/欧/日宏观与地缘政治视角对Card-05的'电力是新瓶颈'主线做压力测试。我不否定电网瓶颈，而是为它划界：(1) 边际AI GW正在通过Stargate ERCOT、Stargate UAE、沙特HUMAIN、马来西亚柔佛等通道离开美国队列，部分掏空美股公用事业敞口；(2) Trump 2.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "关税不是即时交付杀手，而是电力OEM价格伞；窄牛应聚焦拥有合格交付槽位、本地制造、服务收入和客户预付款的GE Vernova与Siemens Energy等少数公司。", "summary": "2026-05-30，本卡支持Card-06的电力OEM窄牛逻辑：GE Vernova和Siemens Energy的订单、积压订单、客户预付款和Power/Grid利润率已经显示短缺正在转化为价格权。关税没有在最新报表中造成广泛交付断裂，反而提高了合格、本地化、可融资产能槽位的价值；但GOES、…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 5, "neutral": 0, "value": 42.8, "events": [ { "title": "AI算力扩张的下一道硬约束是变压器、开关柜与取向硅钢（GOES），西方OEM新增产能要到2027–2028年才投运、GOES供需要到2030年才能缓解，这是供给侧未来两年的alpha…", "summary": "截至2026-05-31，本卡片支持AI算力瓶颈正从GPU迁移至变压器与开关柜，并在设备层量化约束。Wood Mackenzie 2025 Q2显示标准电力变压器交期128周、GSU 144周；Hitachi Energy欧洲≥100MVA大型电力变压器交期50–58个月。Eaton指定美国DC …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI Capex 变现物理墙与 TMT 向 Utilities 的大宗风格轮动交易", "summary": "AI Capex 变现物理墙与 TMT 向 Utilities 的大宗风格轮动交易 AI Capex Power Bottleneck and Equity Style Rotation to Utilities 云巨头是否正在进行从‘算力硬件’向‘电网容量与基础能源’的Capex结构性重组？我们…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网并网瓶颈不是欧美潜在GDP的硬天花板，但已是AI生产率红利兑现时点的2-3年延迟与核心商品PPI的持续高位驱动；Section 232升级是决定它停留在过渡税还是固化为真正上限的…", "summary": "截至2026-05-31，本卡片对Cards 03-05的微观结论做宏观压力测试：电网瓶颈+OEM定价权意味着电力设备PPI（BLS WPU1174自2020年累计+65-70%）将持续高于核心目标，并把成本转嫁给整体欧美Capex基数。美国T&D投资从2024年约1700亿美元升至2025E的2…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网脱钩不会因为成本本身立刻摧毁云巨头ROI，但会通过交期、折旧和自由现金流错配形成TMT二阶导估值天花板。", "summary": "2026-06-01：本卡以首席风控官视角压力测试前序“AI物理墙”主线。结论是：单纯物理成本通胀不足以击穿云巨头AI Capex ROI；真正危险是GOES/LPT脱钩导致36-60个月变压器交期与两到三年数据中心投运周期错配，使Capex、折旧和可收入化容量之间出现12-24个月缺口。若再叠加…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 4, "neutral": 2, "value": 56.2, "events": [ { "title": "在 P/Watt 约束下，A 股应执行真杠铃——多头集中于绿色基荷 IPP 与特高压/变压器/GOES 制造，减配没有电力预定的投机性算力开发商。", "summary": "P/Watt 已取代 P/FLOP 成为 AI 算力扩张的边际约束。云巨头 2026 年资本开支 6,600–7,000 亿美元，三分之二投向供电而非芯片；大型变压器交付周期延至 128 周到 5 年。由此带来的 A 股真杠铃：多头一端是绿色基荷 IPP（长电、广核、核电、三峡能源），重估为带 A…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Quantifying Supply Chain Impact of Iran-US Conflict", "summary": "致 市场情绪师： 协同任务已完成。执行方为 能源行业分析师。 模式: adhoc 会话: c7a01769-2849-4687-b94b-ccf60f624a21 结果摘要: Wrote handoff_response.md (2949 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球宏观研究报告：全球资本开支超级周期对宏观通胀中枢与中性利率（r ）的影响 作者： 全球宏观分析师 发布日期： 2026年6月2日 研究线程标识：已归档卡片索引： 07/10 状态…", "summary": "全球宏观研究报告：全球资本开支超级周期对宏观通胀中枢与中性利率（r ）的影响 作者： 全球宏观分析师 发布日期： 2026年6月2日 研究线程标识：已归档卡片索引： 07/10 状态： 综合研判 & 完成 执行摘要与核心论点 截至 2026年6月2日 ，由人工智能（AI）与能源基础设施建设同步发力…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] U.S.-Iran headline update", "summary": "致 市场情绪师： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: 0c32d021-4237-4ecf-9423-db44a5aaa1c9 结果摘要: Wrote handoff_response.md (11780 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 1, "value": 66.6, "events": [ { "title": "关键材料短缺（尤其是GOES和铜）已成为电网OEM交付的次生瓶颈，将限制2027年电网基建的扩张速度并导致收入确认延迟。", "summary": "截至2026年6月2日，全球电网扩张正面临严重的“材料墙”制约。取向电磁钢 (GOES) 存在30%的供应缺口，且中国实施出口许可限制，导致大型变压器交期延长至3.5-4年。铜价维持在13,000美元/吨高位，且94美元油价导致物流成本激增。尽管 GE Vernova 和 Eaton 订单创纪录，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have finalized the H2 2026 stress testing and synthesis analysis on the transmission of U…", "summary": "I have finalized the H2 2026 stress testing and synthesis analysis on the transmission of US AI hardware tariff inflation to the A share export compu…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "卡片04的长周期订单证据成立，但HALEU瓶颈、FERC表后否决、GOES单源、Section 232关税与中国稀土域外管辖共同抬高政策β，溢价定价权叙事应被降级而非剔除。", "summary": "截至2026-06-02，本卡对卡片04的长周期支持结论进行压力测试，识别五条具体政策/地缘风险：(1) HALEU燃料瓶颈——Centrus示范级月均约45kg、商业产能2027-2028年才到位，使FOAK SMR大概率延期12-24个月；(2) NRC许可摇摆，EO 14300压缩至18个月…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 0, "risks": 4, "neutral": 1, "value": 56, "events": [ { "title": "‘过路费’重定价的最干净通道是 VIM/VAR 二次熔炼 + 单晶铸造 + GOES 精整这条最窄机械-冶金链；矿端整体宽松（铜是局部例外），稀土叙事应剥离为政策驱动股权题材而非冶金…", "summary": "本卡按金属分品种重做 Card-04 的‘过路费’命题，做矿端+冶金中间品双层检验。结论：(1) 镍基高温合金 (Inconel 718/738/René N5) 结构性重定价强证实——LME 镍过剩 150 kt 但 VIM/VAR 双联熔炼+单晶铸造产能锁紧，西方有效供应商仅 5 家，Inco…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 06 synthesizes the prior five into a four bucket AI allocation framework (Hyp…", "summary": "Research note 06 synthesizes the prior five into a four bucket AI allocation framework (Hyperscaler ROI core / locked in power / grid OEM / applicati…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI基建扩张短期更像通胀性的全球电网和材料资本开支周期，而不是无摩擦的生产率冲击；能否把它转化为中长期增长，取决于电网、变压器、铜和成本分摊规则的同步扩张。", "summary": "截至2026-06-03，本卡支持上一张公用事业约束论点，并将其上升为全球宏观资本开支周期：数据中心用电从2025年的485 TWh走向2030年约950 TWh，美国、欧洲和日本/韩国分别面临峰值负荷、电网老化、能效和低碳基荷约束。AI建设先争夺铜、GOES、变压器、燃机和电网工程能力，再释放生…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电网容量与电力设备交付的物理刚性约束已成为 AI 算力 Capex 转化为产能的核心瓶颈，定价权与超额利润将沿‘芯片→变压器/燃气轮机/核电 PPA→GOES 钢’向下游电力基础设施…", "summary": "本报告从能源基础设施物理层支持 TMT 卡观察——AI 算力订单的收入转化率被‘平滑化’，根因是电网容量、大型电力变压器 (LPT) 与重型燃气轮机的结构性刚性瓶颈。截至 2026-06-03，美国数据中心 2030 年负荷展望升至 580 TWh，并网队列存量 2,600 GW，中位等待 53 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 2, "neutral": 0, "value": 66.4, "events": [ { "title": "结构性供给瓶颈不取消美联储2%目标，但会把实现该目标所需的长期终端利率中枢从3.1%推向约3.50%。", "summary": "截至2026-06-05，本卡综合前序结论：美国制造业回流、电网更新与AI数据中心用电需求，正与铜精矿、GOES和变压器供给瓶颈共振。美联储仍可维持2%通胀目标作为制度锚，但实现目标需要更多需求侧抵消。相较2026年3月SEP长期联邦基金利率中值3.1%，结构性瓶颈使长期终端利率合理中枢上移至3.…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "高频运价暴涨与海运股两融去杠杆踩踏的‘实物韧性与筹码破损’定价博弈", "summary": "高频运价暴涨与海运股两融去杠杆踩踏的‘实物韧性与筹码破损’定价博弈 Freight Rate Surge vs. Shipping Margin De-leveraging: Physical Resilience vs. Flow Strains SCFI本周飙升15.9%至2,571.73点，…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 0, "neutral": 1, "value": 82.8, "events": [ { "title": "全球运价暴涨与航运筹码杀跌的非对称博弈", "summary": "全球运价暴涨与航运筹码杀跌的非对称博弈 Global Freight Surge vs. Shipping Equity De-leveraging: An Asymmetric Trade 中远海控因MSCI调整及去杠杆逆势收跌，但在实体端SCFI大涨15.9%且新加坡港口拥堵严重。在7月抢关税…", "sourceLabel": "研究主题", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 4, "neutral": 1, "value": 60.2, "events": [ { "title": "高利率环境并不否定创新主题，但截至 2026-06-07 只有能把 AI 用电需求转化为设备订单、合同化 rate base 或 24/7 清洁电力稀缺性的资产，才应成为核心配置。", "summary": "截至 2026-06-07，本卡综合前序研究：美国就业和通胀数据仍支持 6 月 FOMC 鹰派按兵不动，主题配置应从叙事久期转向现金流兑现、物理瓶颈和合同可执行性。AI 电力基础设施仍是最强主题表达，但 45% 输配电设备仓位必须接受 GOES、高纯铜和合同调价机制筛选；核电 IPPs 保持卫星仓…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "铜的精炼端紧而不破，但 GOES（特别是高磁感）与绝缘材料（Nomex、纤维素压板、浇注环氧）是 2026-2028 结构性硬缺口，将使电网 OEM 实际交付较名义扩产折损 15-2…", "summary": "2026-06-06：支持 Research note 03 关于电网 OEM 交付受材料瓶颈刚性约束的判断。铜总量缺口约 0.5%（价格机制可消化），但电工级铜杆 CTC/HV 电缆交期已由 16 周延至 32-40 周；取向硅钢 GOES 2026 缺口 9%、高磁感 HiB 缺口 15%，新…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Middle East Shipping Disruption Port Delivery Loss and China Export Supply C…", "summary": "致 全球宏观分析师： 协同任务已完成。执行方为 外汇策略师。 模式: adhoc 会话: 012b7699-fc5e-4f69-bf70-188b18235530 结果摘要: Wrote handoff_response.md (9098 bytes); wrote handoff_respons…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "AI Capex周期仍在，但瓶颈已从算力稀缺迁移到电力交付可转化性，关税与地缘审查会把Mag7规划支出转化为更慢、更贵的通电MW。", "summary": "截至2026-06-06，本卡综合前序结论：美国AI基建Capex不会因关税与安全审查终结，但会因进口GOES、进口变压器、进口铜、Section 232完整价值关税、Section 301续期悬念和EO 13920审查而被迫降速。美国电力变压器进口依赖为80%，配电变压器为50%，GOES情景进…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-07", "supports": 0, "risks": 1, "neutral": 0, "value": 76.6, "events": [ { "title": "[auto-handoff] 取向硅钢(GOES)产能瓶颈及对特高压变压器造价传导", "summary": "致 公用事业分析师： 协同任务已完成。执行方为 材料行业分析师。 模式: adhoc 会话: e911bea4-9058-473c-a5ef-19ab52df0e02 结果摘要: I will first run the 系统日期锚 command to anchor today's date. …", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 1, "neutral": 0, "value": 78, "events": [ { "title": "变压器供给冲击真实且结构性，电价波动已在抬升，但价格分布是厚尾非线性的——做多电力主线在基准与上尾情景下成立，下尾情景（超大规模厂商需求再分配/效率突破）被低估，建议把线性做多换成期…", "summary": "美国大型电力变压器交付周期已从约50周拉长至120–210周，GOES与套管成瓶颈，IOU备品库存从12%降至约7%，NERC将PJM/MISO/SPP/ERCOT列为高风险。已传导至电价：PJM 2025/26 BRA出清$269.92/MW-日（约9×），ERCOT北区LMP>$500/MWh…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 4, "neutral": 0, "value": 61.4, "events": [ { "title": "燃气轮机OEM机位产能足以支撑2030年前45–60 GW的AI园区BTM装机，SMR在该窗口几乎无贡献；这给中国配电变压器与开关柜出海带来一条独立于SMR执行风险的多年期订单可见度。", "summary": "【工作日期2026-06-12】2030年前AI园区BTM电源由燃气轮机独自承担：GE Vernova、Siemens Energy、Mitsubishi Power合计OEM机位产能约42–47 GW/年，扣除存量调峰/IPP订单后可分配给AI-BTM约12–18 GW/年，对应2030年底装机…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力并网延迟与GOES变压器物理约束对绿色电力重定价", "summary": "AI算力并网延迟与GOES变压器物理约束对绿色电力重定价 AI Grid Constraints & Re-rating of Green Utilities on GOES/Transformers 取向硅钢(GOES)及高压变压器交期拉长至2.5-4年致使50%的大型算力并网延迟。这是否标志着…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "去全球化导致美欧电网设备及原材料支付高额地缘溢价，并通过公用事业费率基数和新能源购电协议（PPA）向终端电价传导，正在欧美引发系统性的‘绿色通胀’。", "summary": "截至2026-06-13，本卡压力测试表明，关税及安全限制使美国取向硅钢（GOES）及成品变压器面临422%和80%-150%的地缘溢价。这每年为美国公用事业费率基数注入150亿-200亿美元通胀，拉高零售电费1.5%-2.0%。同时，并网延迟和设备溢价推动美国太阳能PPA价格同比大涨13%至$6…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI电网约束收敛下输配电设备商的毛利率挤压与量价背离", "summary": "AI电网约束收敛下输配电设备商的毛利率挤压与量价背离 Grid Infrastructure Squeeze: GPM Squeeze & Mismatched Pass-Through AI并网延迟与大容量变压器交期拉长下，GOES硅钢与铜的双重价格冲击（GOES+20%/铜+20%）将如何挤压…", "sourceLabel": "研究主题", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 0, "value": 80.8, "events": [ { "title": "I have completed the stress test analysis on the risk of export controls and technology dec…", "summary": "I have completed the stress test analysis on the risk of export controls and technology decoupling in the US China grid supply chain for Research not…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 2, "neutral": 0, "value": 76.2, "events": [ { "title": "全球电网升级周期下的贸易壁垒压力测试与出海装备链出口韧性评估", "summary": "截至2026-06-14，压力测试显示任务简报中「美欧加税」的风险被错配：美国早已是结构性封闭市场（购买美国货+301关税+OEM不信任），增量关税几乎拿不走中国约92.9亿美元变压器出口（2025+36%）的份额；真正被低估的是欧洲——对欧出口2025年+138%，而欧盟2026-03-27已对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "中国电力设备出口正凭借‘速度溢价’成为全球AI基础设施的物理基石，并为人民币提供了结构性的宏观对冲价值。", "summary": "2026年全球AI电网约束下，中国电力设备（变压器、特高压、储能）正从‘成本竞争’转向‘速度竞争’，其12-18个月的交期远优于西方的5年 backlog。中国通过控制取向电工钢（GOES）和电力硬件供应链，获得了显著的全球‘速度溢价’（出口均价上涨33%）。这不仅推升了2025年创纪录的1.19…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 1, "value": 80.6, "events": [ { "title": "GOES扩产降低了核心钢材断供尾部风险，但不足以在2026-2027年把美国变压器交付周期从约2.5年压缩到12-18个月的数据中心建设窗口。", "summary": "截至2026-06-15，本卡片压力测试GOES是否能缓解变压器缺料。结论是：GOES扩产真实存在，中国2025年产能达374万mt且出口可能超过70万mt，日本技术链通过Nippon和JFE/JSW继续扩张，美国也有Cleveland-Cliffs和U.S. Steel/Nippon项目。但这些…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AI电力基础设施瓶颈向上游物理期权（精炼铜/取向硅钢）传导评估", "summary": "致 日报总编： 协同任务已完成。执行方为 主题研究员。 模式: adhoc 会话: bca64921-2ae4-41c5-bfcf-48378f7e6da6 结果摘要: Wrote handoff_response.md (13721 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have analyzed the transmission of macro credit tightening and financing \"short termizatio…", "summary": "I have analyzed the transmission of macro credit tightening and financing \"short termization\" to the AI infrastructure expansion cycle as the AI Infr…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 1, "neutral": 0, "value": 85, "events": [ { "title": "[auto-handoff] 油价暴跌与美联储鹰派背离下红利低波资产的结构重配与防御战术", "summary": "致 市场情绪师： 协同任务已完成。执行方为 A股策略师。 模式: adhoc 会话: 044ecfaa-67e3-4d8c-b966-1b5021e7a3c4 结果摘要: Wrote handoff_response.md (16062 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 1, "neutral": 0, "value": 86.4, "events": [ { "title": "[auto-handoff] 澄清原油价格数据分歧并调整宏观通胀传导路径", "summary": "致 日报总编： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: f846db4c-3154-4feb-a5c0-423522062fd7 结果摘要: Wrote handoff_response.md (7369 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 1, "neutral": 1, "value": 90.8, "events": [ { "title": "电力设备供应链瓶颈及其对AI基础设施扩能的二次约束效应", "summary": "报告日期2026-06-19。本收官卡压力测试Research note 07“折价而非证伪”结论，引入变压器、高压开关与取向硅钢(GOES)的二次设备瓶颈。变压器交期由128周(25Q2)拉长至超160周/最高4-5年[S2][S1]，GSU需求自2019年+274%、电力变压器供应缺口约30%…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] AIDC电力基建逻辑下高压变压器与输电网铜需求增量的定量折算", "summary": "致 材料行业分析师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: b7159f5c-eb1d-48e7-8538-01f5bfef96a5 结果摘要: Wrote handoff_response.md (10534 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 4, "neutral": 0, "value": 71.2, "events": [ { "title": "[auto-handoff] 油价驱动的联储转向是暂时性还是持久滞胀regime", "summary": "致 首席风控官： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 56cf7e1a-b9b3-47a2-8caf-4fbf8be73c8f 结果摘要: Wrote handoff_response.md (8630 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器、开关设备、取向电工钢、电缆和电网设备制造产能", "summary": "支持卡片05/06：油价回落是数据中心电力运营成本的缓冲，而非解锁电网设备产能的钥匙。从制造现场看，资本已在大举投入——Hitachi Energy（60亿美元集团计划含约15亿专用于变压器、2027年放量）、Siemens Energy（4.21亿美元Charlotte厂2027年竣工）、GE …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "I have completed the synthesis for Research note 06, analyzing the transmission of supply s…", "summary": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "评估欧美电网升级需求与贸易壁垒对中国电力设备出海链的传导效应", "summary": "压力测试 Research note 4 命题：A股估值重配是否真建立在可持续高毛利出海订单上。结论为部分支持但有结构性保留。需求拉动真实且被低估——美国公用事业2026-30资本开支近1.3万亿美元 [S1]、超大规模厂商2026资本开支>6750亿美元/+60% [S2]、全球变压器短缺 [S…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-25", "supports": 0, "risks": 5, "neutral": 2, "value": 72.6, "events": [ { "title": "Summary of Work I have completed the research and stress test on the transmission effects o…", "summary": "Summary of Work I have completed the research and stress test on the transmission effects of grid equipment bottlenecks on upstream key raw materials…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "[auto-handoff] 美伊临时和平协议持久度评估", "summary": "致 首席风控官： 协同任务已完成。执行方为 全球宏观分析师。 模式: adhoc 会话: ca2f25c7-e1a8-479e-b83e-942ca1eb0bc1 结果摘要: Wrote handoff_response.md (14849 bytes); wrote handoff_respon…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "全球电力设备制造供应链产能与上游材料供应风险评估", "summary": "部分确认Card-03的time-to-power论点，但作三处修正。瓶颈确在上移到材料端，但对变压器而言真正的物理咽喉是GOES而非铜，且GOES稀缺本质是“产能+贸易”稀缺而非地质稀缺：全球变压器产能利用率98%、GOES自2020年涨约100%、美国仅Cleveland-Cliffs一家、俄…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "虽然物理电网排队与供应链原材料价格冲击显著抬高了局部建设阻力，但云厂商通过绿电PPA、高密度液冷及转向定制ASIC等结构性调整，能够完全消纳成本溢价，不构成算力扩张的绝对瓶颈。", "summary": "本卡对电网并网延迟、变气压原材料（铜/GOES）价格波动和本地化GPU封装良率冲击进行了量化压力测试。结果表明，虽然并网延迟和原材料成本增加会显著压缩项目NPV（5年延迟将使NPV降低79.99%），但云厂商可以通过迁移西部枢纽自建绿电、液冷优化和系统级混合算力架构，消纳国产芯片和能源价格溢价，并…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 2, "risks": 11, "neutral": 0, "value": 48, "events": [ { "title": "电网设备出口毛利缓冲的汇率弹性与贸易救济风险压力测试", "summary": "本卡压力测试并有条件收窄 Card-06 的电网设备出口多头。汇率腿：人民币对美元/欧元升值 +3%/+5%/+7% 时，海外分部毛利按 ΔGM=c×x/(1−x) 分别压缩 思源约 2.0/3.4/4.8pp、金盘 2.3/3.9/5.6pp、南瑞 2.5/4.2/6.0pp[自测算]；越薄毛利…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI数据中心虹吸变压器产能导致全球交期拉长至2-3年，加剧海外CXO项目并网延迟2-3年并毁灭其NPV，同时对A股变压器出口企业（如金盘科技）Q2-Q3业绩构成成品存货积压拖累。", "summary": "本卡片验证并支持卡片04结论。AI数据中心对大功率变压器(LPT)需求的剧增(每GW IT负荷需15-16台100MVA变压器)，受限于GOES缺口、绕线工培训及测试台位滞后，将LPT交期推升至2-3年。这加剧了海外电网并网延迟，导致新加坡、爱尔兰等海外CXO基地落地面临2-3年刚性延误，毁灭性削…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "并网延迟的变压器库存积压是可逆的时点拖累(金盘-8%~-11%、思源<-3%)，被国网4万亿内需托底，应在超配出口变压器框架下偏思源、轻金盘并叠加配对与公用事业杠铃对冲。", "summary": "并网延迟导致的已完工未交付变压器库存积压，是营运资金/收入确认的时点事件而非需求毁灭——订单仍排产至2027年在手backlog内。对金盘科技(688676.SH)拖累真实但有界：北美AIDC相关收入是最高Beta切片，Q2/Q3净利约-8%至-11%、伴随存货与FCF占用，属逢低布局而非证伪；对…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "summary": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 10, "neutral": 1, "value": 42.4, "events": [ { "title": "Summary of Work We have completed the analysis for Research note 04/10 as the Materials Ana…", "summary": "Summary of Work We have completed the analysis for Research note 04/10 as the Materials Analyst (材料行业分析师) , adopting a Support (支撑) stance. Our repor…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "滞胀确认与大类资产重构：商品通胀刚性、美联储“加息在桌”对大类资产风险平价模型（Risk Parity）的极限压力测试。", "summary": "本卡支持第01–07卡逻辑链并闭环回组合构建。供给冲击型通胀给铜（2026年缺口约33万吨，2030年扩至150–220万吨）与被贸易墙隔离的GOES（中国约56%产能）一个硬性供给底，使商品子仓下行半方差较以往周期压缩约30–40%——保留通胀上行、削弱增长下行。但这救不了风险平价：供给冲击把股…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] 上游周期行业盈利下调与造船厂交期拖累下的信用违约及报表冲击测试", "summary": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: afe82f2c-0d17-4ce7-9fa7-39d917caf118 结果摘要: Wrote handoff_response.md (13245 bytes); wrote handoff_response…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "功率半导体已过剩，AI电力的新物理瓶颈已迁移至大型变压器，并真正卡在无法被资本快速绕行的上游电磁钢(GOES)与铜——2026–2028窗口难解。", "summary": "支持并收紧前五卡逻辑链：功率半导体转入结构性过剩后（Research note 05），AI电力建设的刚性瓶颈已迁移至大型变压器(LPT)，并下沉到上游电磁钢(GOES)与铜。瓶颈呈两层：整机厂总装层可用约20亿美元北美扩产去瓶颈，但缓解要到2028/2029年后；标准电力变压器交期128周、GS…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 17, "neutral": 0, "value": 12, "events": [ { "title": "中国GOES与铜体系在基准情形下是西方AI建设的泄压阀，但政策阀门已被安装，西方数据中心应把中国来源依赖重定价为150-300bp的分层资本开支风险溢价。", "summary": "截至2026-06-28，本卡压力测试GOES与铜的中国供给集中度。结论：基准情形不是中国对GOES/铜禁运，而是“带政策期权的泄压阀”。中国具备西方AI数据中心建设所需的GOES与铜精炼/冶炼规模，但关税、原产地、反倾销与出口管制先例会把供给集中度转化为资本开支风险溢价。建议在Research …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在中国的精铜结构性进口依存与紫金海外资产高地缘折价背景下，应将有色敞口拆分为“在岸GOES技术壁垒（宝钢股份）+离岸实物/矿山多头”以规避紫金主权尾部风险，并通过“黄金（14.15%…", "summary": "截至2026-06-28，本卡基于有色冶炼端0加工费与出口管制非对称性构建了跨资产对冲框架。由于中国铜进口依存高（87%）且紫金海外资产折价大（占EBITDA的50%），建议低配A股冶炼及紫金，将铜敞口拆分为在岸宝钢及离岸实物/矿山多头（Freeport/BHP）。为对冲地缘尾部，引入实物黄金（1…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "原始40/32/12/8/8 sleeve在AI需求相关性趋近+1时尾部风险过高，阶段1是强制去杠杆，阶段2才接近可接受，但决定性保护来自降AI beta和提高现金，而非小仓位黄金/…", "summary": "截至2026-06-29，本卡对40/32/12/8/8 AI物理瓶颈sleeve做正式压力测试。原始组合在1个月99% Student-t口径下VaR/CVaR为28.6%/37.8%[自测算]，阶段1降至23.2%/30.7%[自测算]，阶段2降至18.5%/24.5%[自测算]。结论是：原始…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "变压器交期拉长至3-5年与并网排队超36个月的物理刚性瓶颈，配合美欧贸易保护关税对GOES硅钢供应链的阻断，已对AI算力扩张的通电效率构成无法用资本单纯解决的长期结构性壁垒，造成明显…", "summary": "基于并网排队与变压器短缺的双重物理约束，我们验证电力基础设施瓶颈已构成AI算力扩张的长期结构性阻碍。数据中心通电时间已拉长至36-60个月，变压器交期延至3-5年，主要厂商积压订单超百亿美元。这些延误对项目NPV和IRR产生重大侵蚀，100MW数据中心延误3年将损失51.63%的NPV。美国Sec…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 8, "neutral": 2, "value": 60.2, "events": [ { "title": "[auto-handoff] Hormuz shock & core CPI transmission", "summary": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "sourceLabel": "分析协作", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "[auto-handoff] Impact of global capacity pricing spikes on domestic transformer export marg…", "summary": "致 能源行业分析师： 协同任务已完成。执行方为 公用事业分析师。 模式: adhoc 会话: 324722b5-2f7b-4b91-aec7-ac5429f32a39 结果摘要: Wrote handoff_response.md (11765 bytes); wrote handoff_resp…", "sourceLabel": "分析协作", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "summary": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "AI算力基建瓶颈与多资产杠铃策略的逻辑闭环及定量模型终期质量审计", "summary": "终期QA审计给出「需修订（有条件通过）」。会期主轴稳健，核心论点逐节点重定价、瓶颈迁移至可通电MW、并以杠铃对冲滞胀尾部端到端自洽；可复算的定量链条均对得上，本卡问题的2.30%保本不良率上限算术正确（125bp再贷款利差−10bp运营−50%LGD，NPL=2.30%净利差归零）。PJM容量价3…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-d6b2c23ed6/", "networkHref": "/investor/research-network/#thesis-d6b2c23ed6" }, { "id": "thesis-62d39d2c51", "source": "auto_topic", "title": "家装链：地产后周期需求与消费修复验证", "titleEn": "Practical research topic: 家装链：地产后周期需求与消费修复验证", "lane": "消费健康防御", "laneEn": "消费健康防御", "laneHref": "/investor/lanes/lane-d104d465/", "investableFocus": "", "stockUniverse": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "002572.SZ", "name": "索菲亚", "market": "A-share", "role": "定制家居", "roleEn": "定制家居", "bucket": "core", "href": "/investor/stocks/002572.SZ/" } ], "stockTickers": [ "600519.SH", "300760.SZ", "600276.SH", "3690.HK", "001215.SZ", "1179.HK", "9961.HK", "588000.SH", "APO", "IEF", "META", "GLD", "MAR", "002572.SZ" ], "coreView": "该主线来自 35 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-17 至 2026-06-29；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 52, "eventCount": 55, "firstSeen": "2026-05-17", "lastSeen": "2026-06-29", "supports": 4, "risks": 50, "neutral": 1, "analystCount": 3, "sourceTypeCounts": { "研究线程": 33, "研究笔记": 16, "深度研究": 6 }, "topTerms": [ "新华财经-家装产业链", "研究活跃度与投资线索复盘" ], "whatChanged": [ "2026-06-29: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "2026-06-29: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "2026-06-28: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "2026-06-27: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "2026-06-24: 产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）" ], "practical": { "sectorLane": "消费健康防御", "sectorLaneEn": "消费健康防御", "exposureBucket": "消费健康防御", "tone": "green", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "002572.SZ", "name": "索菲亚", "market": "A-share", "role": "定制家居", "roleEn": "定制家居", "bucket": "core", "href": "/investor/stocks/002572.SZ/" } ], "stockTickers": [ "600519.SH", "300760.SZ", "600276.SH", "3690.HK", "001215.SZ", "1179.HK", "9961.HK", "588000.SH", "APO", "IEF", "META", "GLD", "MAR", "002572.SZ" ], "primaryStocks": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" } ], "instrumentTags": [ "600519.SH", "贵州茅台", "300760.SZ", "迈瑞医疗", "600276.SH", "恒瑞医药", "3690.HK", "美团", "001215.SZ", "千味央厨", "1179.HK", "华住集团-S" ], "reviewCadence": "每日复核", "confirmationSignals": [ "价格弹性是否转化为利润率修复", "支付端和监管是否支持长期现金流", "防御属性是否被估值提前透支" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "估值拥挤", "支付监管", "渠道库存", "CPI 食品项" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 85，来源 3 类，近期事件 29 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 55, "sourceDiversity": 3, "analystCount": 3, "recentEvents": 29, "supportRatio": 0.07, "riskRatio": 0.91, "conflictScore": 100, "evidenceDepth": 85.3 }, "score": 42 }, "events": [ { "date": "2026-06-29", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-29", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "家装产业链是地产后周期复合体，2026年年中呈分化而非崩塌：1-2月商品房销售面积-13.5%、销售额-20.2%，定制家居9家7亏、平均营收-27%、现金流全负；91家家居企业FY2025仅14家双增。低覆盖alpha集中在零售占比高、政策杠杆强、现金强的标的：好太太(603848,国补智能家居,Q1营收+21.68%)、兔宝宝(002043,板材回暖+…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "家装产业链已一分为二：依赖新房竣工的工程/B端（定制家居1Q26平均-27%、9家7亏；江山欧派营收-43.35%）仍在衰退，而依赖存量翻新/零售的细分（防水+16.9%、涂料+10.6%、板材）正逆势上行（竣工2025年-18%，但翻新需求占比已超40%）。被低覆盖的alpha在零售-翻新侧：首推兔宝宝(002043，营收+32.68%/净利+46.36…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 6.75, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-24", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-22", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-22", "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-21", "title": "产业链调研：新华财经-家装产业链（30 日激活率 8.5%, 17/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 8.5%, 17/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-21", "title": "产业链调研：新华财经-家装产业链（30 日激活率 8.5%, 17/201）", "summary": "家装产业链已分叉：B 端（新房/工程）随地产竣工（2025约-18%）结构性下行，C 端（旧改/以旧换新/零售）是唯一增长引擎，2025 旧改消费破 2.5 万亿、贡献头部企业逾 60% 营收增长。2025 年 91 家上市家居企业仅 14 家双增长。Alpha 不在被过度跟踪的定制家居龙头（欧派/索菲亚均下滑），而在被低覆盖、正完成 B→C 切换的二线材…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.25, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-18", "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-17", "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-17", "title": "新华财经-家装产业链正经历结构性分化，核心阿尔法机会向具备防御属性的C端零售渠道和高弹性的海外出口渠道集中。", "summary": "截至2026年6月17日，本报告深度评估新华财经-家装产业链（过去30天激活率仅7%，14/201只）。虽然国内宏观地产承压，但防守性强的C端零售渠道和高弹性的海外出口渠道龙头表现出良好的财务韧性、高毛利与高分红特征。同时，警示了坏账、运费及汇率波动等四大溢出风险。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-15", "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-15", "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-13", "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" }, { "date": "2026-06-12", "title": "产业链调研：新华财经-家装产业链（30 日激活率 6.5%, 13/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 6.5%, 13/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新华财经-家装产业链" ], "href": "" } ], "timeline": [ { "date": "2026-05-17", "supports": 0, "risks": 2, "neutral": 0, "value": 30, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 3.5%, 7/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 3.5%, 7/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "家装链已具备\"补贴延期 + 渠道结构性切换 + 表内出清\"三重组合，应从地产 / 消费两台的覆盖夹缝中提级为独立子线程，首批深度集中于志邦、三棵树、蒙娜丽莎、居然、顾家五个低覆盖标的。", "summary": "新华财经-家装产业链 30 日激活率仅 3.5%（7/198），但宏观三线已翻正：以旧换新补贴延至 2026 年底、二手房成交持续跑赢新房、开发商 B 端风险大体出清。建议优先深度覆盖 5 个低关注度标的：志邦家居(603801)以旧换新+KA 渠道纯标的；三棵树(603737) B→C 转型重估…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 2, "risks": 4, "neutral": 0, "value": 35.4, "events": [ { "title": "新华财经-家装产业链4%激活率要从窄口径alpha升级到行业级超配，必须通过基本面三闸门与A股市场结构四闸门的双重确认，2026-05-18双方均未达标。", "summary": "截至2026-05-18，本卡片在card-05基本面三闸门之上叠加A股市场结构四闸门：成交额（指数日均成交较60日中位放大≥80%且持续10日）、两融（成分股两融余额自低位反弹≥15%、融资买入占比突破8%）、成分股参与度（激活率从8/198即4%抬升至≥40/198即20%、20日均线之上占比…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "4%激活率只有在二手房价格扩散、家装零售扩散和居民信用意愿同时转正时，才是行业级超配信号；截至2026-05-18，它仍只是Arrow Home、Tubao等低客单刚需alpha的观…", "summary": "截至2026-05-18，本卡片综合前四张卡片，结论是新华财经-家装产业链指数4%激活率目前只能视为低关注度alpha信号，尚不能升级为行业级超配。升级阈值必须三项同时满足：70城二手住宅环比上涨或持平城市连续两个月至少35/70，且二三线不再下跌；建筑及装潢材料、家具等家装零售品类从负增长转为扩…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "中国电力设备出海：AI 电网需求、贸易壁垒与利润率分化", "summary": "全球 AI 电网投资为中国电力设备提供需求窗口，但关税、认证、交付能力和原材料成本会决定哪些企业真正把订单转化为利润。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/china-power-equipment-export-policy-8678db7f/" }, { "title": "AI 算力资本开支、电力瓶颈与通胀再定价", "summary": "AI 基础设施需求正在先通过电力、电网、变压器与材料供给形成成本脉冲，再通过生产率与架构效率形成中期缓释；投资结论取决于这两个方向谁先兑现。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-power-inflation-transmission-0c9ba15c/" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 1, "neutral": 0, "value": 46.8, "events": [ { "title": "消费降级、银发经济与大健康结构性机会", "summary": "消费链条正在从总量弹性转向结构分化：银发消费、医疗健康、糖替代和 B 端餐饮渗透可能形成防御性需求，但需要验证支付能力和渠道真实动销。", "sourceLabel": "深度研究", "supportLabel": "支持", "supportType": "supports", "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" }, { "title": "2026-05-19的核心判断是：家装链不是全面复苏，而是存量翻新和智能/品质升级驱动的结构性机会，应优先覆盖已验证现金流或经营拐点的低拥挤标的。", "summary": "截至2026-05-19，本卡片支持对新华财经-家装产业链做选择性深挖：地产和家装零售仍弱，但以旧换新、智能家居、适老化和“好房子”政策正在把需求重心转向存量住房翻新。首选深挖002043.SZ兔宝宝、001221.SZ悍高集团、605268.SH王力安防；603848.SH好太太做渠道改革验证；…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-21", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "截至2026-05-21，地产后周期优质零售与建材龙头正通过主动出清大宗房企坏账并转向现款零售与出海，实现与开发商信用危机的实质性脱钩，迎来独立重估契机。", "summary": "截至2026-05-21，新华财经-家装产业链近30日激活率仅4.5%（9/198），研究严重空白。本研究支持地产后周期板块结构性重估，筛选出江河集团、东鹏控股、北新建材和索菲亚等低覆盖度优质Setup。优质龙头已斩断高风险房企大宗业务，转向现款零售并受益于以旧换新补贴。分析师需关注B端坏账、渠道…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 3, "neutral": 0, "value": 31, "events": [ { "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 3, "neutral": 0, "value": 32.4, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "家装链低激活掩盖了一个『政策+存量房』双轮启动，应将研究资源从龙头折价交易转向腰部差异化标的，并把 2026 H1 国补退坡与开发商应收账款两条尾部风险纳入持续跟踪。", "summary": "新华财经-家装产业链 30 日激活率仅 4.5%（9/198），属结构性少覆盖。但 2024-2025 年新房塌陷与存量房翻新发生剪刀差再平衡，叠加 3,000 亿国补（2025 年 7 省扩至橱柜/卫浴/家具/照明/智能家居）与 1.2 万亿旧改资金，产业链正由地产 β 切向消费+政策双 β。Q…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "经历深度调整后的家装产业链正在迎来存量旧改与品牌出海双轨驱动的结构性复苏，极低的研究覆盖率与估值压缩为左侧建仓提供了非对称套利机会。", "summary": "本次研究聚焦覆盖率仅4.5%的新华财经-家装产业链。筛选出4家最具韧性的低估值标的：天安新材（装配式整装突围）、志邦家居（墙门逆势高增与出海引擎）、索菲亚（客单价提升与金融毒丸出清）、瑞尔特（智能卫浴自有品牌转型）。同时警惕房企信用违约、原材料价格上涨和贸易摩擦等跨产业链风险溢出。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 48.6, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 2, "neutral": 0, "value": 45.4, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "家装产业链需求逻辑已从竣工驱动切到存量翻新+以旧换新驱动,5 只中盘转型标的(兔宝宝/江山欧派/公元股份/梦百合/我乐家居)是当前最值得挖掘的 alpha 候选。", "summary": "新华财经-家装产业链 30 日激活率仅 4.5%(9/198),处于研究真空,但底层需求逻辑已从新房竣工驱动切换为存量房翻新+以旧换新驱动。报告锁定 5 只被低估的中盘标的——兔宝宝(002043)板材转装企、江山欧派(603208)木门工程转零售已计提坏账、公元股份(002641)管材二线份额提…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 3, "neutral": 0, "value": 40.8, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在以旧换新政策支撑与消费下沉趋势下，家装产业链正脱离新房周期实现存量翻新突围，应避开加盟制成品定制品牌，深挖在下沉零售渠道及装配式快装领域展现扭亏与业绩爆发的低覆盖上游标的。", "summary": "截至2026-06-02，新华财经-家装产业链30日激活率仅4.5%，存在巨大研究空白。在老旧房翻新与以旧换新补贴政策推动下，行业已脱离传统房企依赖。本研究筛选出兔宝宝、科顺股份、江河集团、亚厦股份等高阿尔法潜力的低覆盖标的，并提示上游化工品及地方债务等跨链外溢风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 2, "neutral": 0, "value": 48.2, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.6%, 15/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.6%, 15/198） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 4.5%, 9/198） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-04", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.6%, 15/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.6%, 15/198） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 2, "neutral": 0, "value": 51, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.6%, 15/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.6%, 15/198） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.6%, 15/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.6%, 15/198） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.6%, 15/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.6%, 15/198） 6%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 2, "neutral": 0, "value": 53.8, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.1%, 14/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.1%, 14/198） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.1%, 14/198）", "summary": "家装产业链30日激活率仅7.1%，被严重低估，但内部正剧烈分化：市场把它当一个板块定价，实为至少三个。新房锚定的定制家居为震中——2026一季报9家仅欧派、我乐盈利，7家亏损，索菲亚/志邦/好莱客净利降幅达-417%/-324%/-1105%，全行业经营现金流转负。但三个细分向上脱钩：重涂涂料（三…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.1%, 14/198）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.1%, 14/198） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 1, "risks": 1, "neutral": 0, "value": 70.6, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 6.5%, 13/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 6.5%, 13/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核心判断：家装链的可投线索应从新房竣工β转向存量翻新、智能/节水产品和渠道重构兑现度。", "summary": "截至2026-06-12，新华财经-家装产业链过去30日激活率仅13/201（6.5%），仍属低覆盖链条。当前地产竣工与家装零售数据偏弱，但以旧换新、智能家居、水效产品、适老化和城市更新提供结构性机会。优先深挖ARROW Home Group（001322.SZ）、R&T Plumbing Tec…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 0, "value": 64, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 2, "neutral": 0, "value": 59.4, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 2, "neutral": 0, "value": 60.8, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7%, 14/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新华财经-家装产业链正经历结构性分化，核心阿尔法机会向具备防御属性的C端零售渠道和高弹性的海外出口渠道集中。", "summary": "截至2026年6月17日，本报告深度评估新华财经-家装产业链（过去30天激活率仅7%，14/201只）。虽然国内宏观地产承压，但防守性强的C端零售渠道和高弹性的海外出口渠道龙头表现出良好的财务韧性、高毛利与高分红特征。同时，警示了坏账、运费及汇率波动等四大溢出风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 1, "neutral": 0, "value": 68.2, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 7.5%, 15/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 0, "value": 63.6, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 8.5%, 17/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 8.5%, 17/201） 5%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 8.5%, 17/201）", "summary": "家装产业链已分叉：B 端（新房/工程）随地产竣工（2025约-18%）结构性下行，C 端（旧改/以旧换新/零售）是唯一增长引擎，2025 旧改消费破 2.5 万亿、贡献头部企业逾 60% 营收增长。2025 年 91 家上市家居企业仅 14 家双增长。Alpha 不在被过度跟踪的定制家居龙头（欧派…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 2, "neutral": 0, "value": 65, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-24", "supports": 0, "risks": 1, "neutral": 0, "value": 72.4, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 0, "neutral": 1, "value": 82.8, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "家装产业链已一分为二：依赖新房竣工的工程/B端（定制家居1Q26平均-27%、9家7亏；江山欧派营收-43.35%）仍在衰退，而依赖存量翻新/零售的细分（防水+16.9%、涂料+10.6%、板材）正逆势上行（竣工2025年-18%，但翻新需求占比已超40%）。被低覆盖的alpha在零售-翻新侧：首…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 1, "neutral": 0, "value": 75.2, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "家装产业链是地产后周期复合体，2026年年中呈分化而非崩塌：1-2月商品房销售面积-13.5%、销售额-20.2%，定制家居9家7亏、平均营收-27%、现金流全负；91家家居企业FY2025仅14家双增。低覆盖alpha集中在零售占比高、政策杠杆强、现金强的标的：好太太(603848,国补智能家居…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 2, "neutral": 0, "value": 70.6, "events": [ { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "summary": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201） Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-62d39d2c51/", "networkHref": "/investor/research-network/#thesis-62d39d2c51" }, { "id": "thesis-90ccc2857a", "source": "auto_topic", "title": "酒店、餐馆与休闲产业链：研究活跃度与投资线索复盘", "titleEn": "Practical research topic: 酒店、餐馆与休闲产业链：研究活跃度与投资线索复盘", "lane": "消费健康防御", "laneEn": "消费健康防御", "laneHref": "/investor/lanes/lane-d104d465/", "investableFocus": "", "stockUniverse": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "RTX", "name": "RTX", "market": "US", "role": "航空发动机/防务", "roleEn": "航空发动机/防务", "bucket": "core", "href": "/investor/stocks/RTX/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "600519.SH", "300760.SZ", "600276.SH", "3690.HK", "001215.SZ", "1179.HK", "9961.HK", "588000.SH", "APO", "META", "UUP", "MAR", "RTX", "IEF" ], "coreView": "该主线来自 16 条研究记录的标题、问题和摘要聚类，时间跨度 2026-06-03 至 2026-06-28；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 51, "eventCount": 40, "firstSeen": "2026-05-05", "lastSeen": "2026-06-28", "supports": 5, "risks": 30, "neutral": 5, "analystCount": 6, "sourceTypeCounts": { "研究笔记": 19, "深度研究": 6, "研究线程": 15 }, "topTerms": [ "酒店", "餐馆与休闲产业链", "研究活跃度与投资线索复盘" ], "whatChanged": [ "2026-06-28: 产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "2026-06-28: 产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "2026-06-27: 产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "2026-06-27: 产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "2026-06-27: 产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）" ], "practical": { "sectorLane": "消费健康防御", "sectorLaneEn": "消费健康防御", "exposureBucket": "消费健康防御", "tone": "green", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" }, { "ticker": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "roleEn": "OTA 与旅游预订", "bucket": "core", "href": "/investor/stocks/9961.HK/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "RTX", "name": "RTX", "market": "US", "role": "航空发动机/防务", "roleEn": "航空发动机/防务", "bucket": "core", "href": "/investor/stocks/RTX/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "600519.SH", "300760.SZ", "600276.SH", "3690.HK", "001215.SZ", "1179.HK", "9961.HK", "588000.SH", "APO", "META", "UUP", "MAR", "RTX", "IEF" ], "primaryStocks": [ { "ticker": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "roleEn": "外资/核心资产代理", "bucket": "core", "href": "/investor/stocks/600519.SH/" }, { "ticker": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "roleEn": "医疗设备", "bucket": "core", "href": "/investor/stocks/300760.SZ/" }, { "ticker": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "roleEn": "创新药出海核心", "bucket": "core", "href": "/investor/stocks/600276.SH/" }, { "ticker": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "roleEn": "本地服务与消费弹性", "bucket": "core", "href": "/investor/stocks/3690.HK/" }, { "ticker": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "roleEn": "B 端餐饮标准化", "bucket": "core", "href": "/investor/stocks/001215.SZ/" }, { "ticker": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "roleEn": "酒店 RevPAR", "bucket": "core", "href": "/investor/stocks/1179.HK/" } ], "instrumentTags": [ "600519.SH", "贵州茅台", "300760.SZ", "迈瑞医疗", "600276.SH", "恒瑞医药", "3690.HK", "美团", "001215.SZ", "千味央厨", "1179.HK", "华住集团-S" ], "reviewCadence": "每日复核", "confirmationSignals": [ "价格弹性是否转化为利润率修复", "支付端和监管是否支持长期现金流", "防御属性是否被估值提前透支" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "估值拥挤", "支付监管", "渠道库存", "CPI 食品项" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 89，来源 3 类，近期事件 31 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 40, "sourceDiversity": 3, "analystCount": 6, "recentEvents": 31, "supportRatio": 0.13, "riskRatio": 0.75, "conflictScore": 83.3, "evidenceDepth": 88.5 }, "score": 44.9 }, "events": [ { "date": "2026-06-28", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "2026年年中酒店/餐饮/休闲链结构性分化而非普涨：出行量低个位数增长但人均消费滞后（五一人次+3.6%、花费+2.9%；端午人次+4.4%、花费+4.0%），且2025年禁酒令精准摧毁商务宴请渠道（单店-20%、高端餐饮最高-70%）。Alpha不在拥挤的酒店龙头，而在与RevPAR脱钩的零售飞轮亚朵ATAT（26Q1营收+47.5%、零售+54.4%）…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 6.75, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "酒店、餐馆与休闲产业链是分化而非复苏：2026年1-5月餐饮收入+3.1%、五一出游3.25亿人次+3.6%但人均消费约-0.7%，频次上行、客单下行。低覆盖且立论最强的标的为亚朵ATAT（零售飞轮Q1+54.4%、净利+90%）、千味央厨001215（B端供应链见底、Q1收入+23.8%）、君亭301073（要约收购事件）、锦江600754（成本自助、归…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-27", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "酒店/餐馆/休闲链30日激活率仅9.3%，严重欠覆盖且处拐点。2026复苏呈\"量增价减\"：五一人次+3.6%但花费+2.9%，RevPAR由价格而非量驱动，alpha集中在业绩拐点/自救/轻资产名而非beta。筛出4只低覆盖setup：君亭酒店(301073,Q1归母+135%、君澜并表轻资产化，弹性最大)、同庆楼(605108,宴席业绩底、H2修复)、宋…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-26", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-23", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-22", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-22", "title": "锦江酒店等所谓的“防御性”标的在高经营杠杆下极具业绩脆弱性，一旦RevPAR出现负向冲击将引发主动基金清仓、融资盘强平与赎回负反馈共振的流动性踩踏，价格跌幅最高可达-50.33%。", "summary": "本研究支持Research note 5结论，对以锦江酒店为首的A股消费板块“伪防御”标的进行拥挤度与平仓踩踏压力测试。锦江自由流通市值约72.75亿元，主动基金持仓占比达15.81%，面临退出瓶颈。若RevPAR下跌-5%，高经营杠杆（19.15x）将导致利润回落-95.75%，引发平仓踩踏。在中度和重度压力下，融资平仓与赎回反馈将价格跌幅从初始的-26…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "酒店" ], "href": "" }, { "date": "2026-06-22", "title": "在消费下沉与成本高企周期中，轻资产管理费模式、上游降本供应链及垄断高分红景区构成了消费板块最核心的毛利防御护城河，而重资产直营及外卖依赖型资产则面临估值与利润双重去杠杆。", "summary": "基于2026-06-23的研究，本报告构建了住宿餐饮与休闲产业链的组合配置框架。在食材成本上涨、外卖平台抽佣上升及固定成本去杠杆的三重压力下，板块内部分化加剧。建议超配轻资产加盟连锁（华住集团、百胜中国）、上游标准化供应链（安井食品）和高红利无债垄断景区（丽江股份、九华旅游）；减配重资产直营餐饮（奈雪的茶、呷哺呷哺）、面临折旧与食材双重挤压的堂食（同庆楼）…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "酒店" ], "href": "" }, { "date": "2026-06-21", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-21", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "休闲产业链被市场当作无差别弱消费交易而低覆盖（30日仅9/97激活），但链内分化剧烈。1—5月社零仅+1.4%、5月餐饮减速至+0.6%[S1]，掩盖了个体性自救向上拐头。最具论点价值的低覆盖标的：亚朵ATAT（零售+54.4%、占比38.1%，上调全年指引，17/14/11x PE[S3][S4][S5]）、同程0780.HK（营收+14.4%、出境Ho…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-21", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "酒店/餐馆/休闲链被低覆盖（30日激活9/97），宏观基调为‘量增价减’（五一出游3.25亿人次+3.6%，但总花费1854.92亿仅+2.9%，人均约持平略降）。市场用同一条疲软宏观线给全链定价，错配了少数个体催化剂。建议入队：君亭(301073，湖北国资36%控股权变更+酒店REITs资格，Q1归母+135.24%)；同庆楼(605108，婚宴触底vs…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-19", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" }, { "date": "2026-06-18", "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "酒店", "餐馆与休闲产业链" ], "href": "" } ], "timeline": [ { "date": "2026-05-05", "supports": 1, "risks": 0, "neutral": 0, "value": 50, "events": [ { "title": "体验密度已取代传统的平效和周转率，成为决定可选消费估值‘超额收益’（Operational Alpha）的底层量化锚点。", "summary": "本报告量化验证了‘体验密度’（ED）作为K型复苏下消费行业估值溢价核心驱动力的逻辑。通过建立体验密度指数（EDI），结合餐饮（海底捞、Topgolf）、奢侈品（爱马仕、小红书UGC）及文旅（生活方式酒店）的实证数据，证明高体验密度能显著降低获客成本（CAC）并提升定价权。研究发现，社交提及率与复购…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-16", "supports": 1, "risks": 0, "neutral": 0, "value": 51.4, "events": [ { "title": "消费结构性防御是A股风格再平衡的组成部分，而不是独立触发全市场防御质量切换的充分条件。", "summary": "截至2026-05-17，本卡综合判断：消费内部升级足以支持A股中的防御质量倾斜，但不足以单独驱动全市场从高贝塔动量切换到防御质量。酒店仍是最干净的消费相对多头核心，免税和演艺应作为有上限的小卫星；组合层面应搭配银行、电力/公用事业、保险和高现金流必选消费，同时保留对仍有盈利确认的科技与周期方向的…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "中国电力设备出海：AI 电网需求、贸易壁垒与利润率分化", "summary": "全球 AI 电网投资为中国电力设备提供需求窗口，但关税、认证、交付能力和原材料成本会决定哪些企业真正把订单转化为利润。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/china-power-equipment-export-policy-8678db7f/" }, { "title": "AI 算力资本开支、电力瓶颈与通胀再定价", "summary": "AI 基础设施需求正在先通过电力、电网、变压器与材料供给形成成本脉冲，再通过生产率与架构效率形成中期缓释；投资结论取决于这两个方向谁先兑现。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-power-inflation-transmission-0c9ba15c/" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 0, "neutral": 0, "value": 54.2, "events": [ { "title": "消费降级、银发经济与大健康结构性机会", "summary": "消费链条正在从总量弹性转向结构分化：银发消费、医疗健康、糖替代和 B 端餐饮渗透可能形成防御性需求，但需要验证支付能力和渠道真实动销。", "sourceLabel": "深度研究", "supportLabel": "支持", "supportType": "supports", "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" } ] }, { "date": "2026-05-31", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "霍尔木兹海峡地缘危机悬而未决的二元走向将油价、通胀与利率锁定在高位，给高杠杆航司（AAL）与消费型酒店（MAR）带来严重的绝对和相对风险暴露，建议平仓并重构组合至航空制造OEM（RT…", "summary": "本卡片从全球宏观视角对霍尔木兹海峡地缘危机进行了多情景划分（50%维持现状、30%降级、20%升级），并给出了WTI、航空煤油裂解价差、DXY和10Y美债收益率的中位数传导路径。压力测试表明，高油价与通胀环境对既定投资组合（AAL空/DAL多+MAR多+RTX多）造成显著冲击，强烈建议进行组合重构…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-03", "supports": 1, "risks": 1, "neutral": 0, "value": 53.8, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 11.3%, 11/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 11.3%, 11/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "建议将消费杠铃模型权重转化为具体的跨市场执行限价区间，并通过调出估值见顶的大消费及传统旅游贝塔、部分拥挤的高股息敞口来提供建仓资金。", "summary": "截至2026年6月3日，本研究支持消费组合模型权重（ATAT 36%、达势股份26%、九毛九5%、九华旅游33%），并将其转化为具体的跨市场建仓与限价执行区间。建议交易资金主要源于调出传统大消费贝塔（50%）及重资产连锁酒店与免税等传统旅游贝塔（30%），并适度减仓拥挤的高股息敞口（20%），在保…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-06-09", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 11.3%, 11/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 11.3%, 11/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 4, "neutral": 0, "value": 30.6, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 10.3%, 10/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 10.3%, 10/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "酒店与餐饮行业正经历从全面复苏向‘K型’分化的结构性转型，在关注出海及零售化溢价标的同时，需警惕下半年再融资墙对高杠杆中端运营商的冲击。", "summary": "2026年6月调研显示，酒店、餐馆与休闲产业链已进入“精准消费”阶段，中国市场RevPAR表现强劲。报告识别了亚朵（零售酒店混合体）、特海国际（海外扩张）、九毛九（价值反转）、首旅酒店（下沉市场）及泡泡玛特（IP化休闲）五个深度研究候选标的。同时警告了2026年底面临的30%抵押贷款到期融资压力及…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "经营性物业贷款与区域性银行信用收缩对宏观社融增速的影响", "summary": "锚定2026-06-11，本卡压力测试酒店再融资墙经经营性物业贷款向区域性银行、再向地方中小企业融资的传导。结论为校准式压力测试：部分否定系统性冲击叙事、确认地方化尾部。自测算显示即便重度压力（区域行CET1侵蚀-113.6 bps），对全国社融增速拖累仅-0.01至-0.10pp、对M2-0.0…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "酒店行业信用分化正通过商业物业贷款向银行体系传导，旅游集中型区域性商业银行因风险敞口集中面临严重的利润挤压与高达113.6 bps的CET1资本充足率边际侵蚀，资产配置上应严格规避此…", "summary": "本报告针对2026年底到期的30%酒店加盟商/业主抵押贷款，压力测试了银行商业地产及经营性物业贷款的资产质量。酒店业主信用呈现K型分化，Tier 1轻资产加盟商现金流稳健，而Tier 3单体业主违约率激增。测试显示，重度压力下酒店行业不良率将升至10.11%（股份行）和11.25%（区域行），导致…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 1, "value": 41, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 10.3%, 10/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 10.3%, 10/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "酒店、餐馆与休闲链正处 K 型拐点而非均匀复苏，故被低覆盖、定价错位。2026Q1 龙头酒店 RevPAR 重新转正、餐饮供给出清但幸存者同店再加速、海南免税强反弹；但 4 月商品零售 -0.1%、餐饮仅 +2.2%，是份额与结构故事而非需求总量故事。精选五个有自助/结构性杠杆的低覆盖标的：亚朵 …", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "核心判断：2026 年酒店、餐馆与休闲链条的机会在“有渠道效率或经营杠杆的个股”，不是全行业复苏。", "summary": "2026-06-15：本卡初步综合酒店、餐馆与休闲产业链，结论是不要做宽基板块 beta，而应排队深挖 Atour、Tongcheng Travel、Super Hi，并把 Nayuki、Jiumaojiu 放入条件性观察。宏观证据显示餐饮收入仍增长、五一出游人次增长，但人均花费和餐饮价格偏弱；公…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-17", "supports": 1, "risks": 2, "neutral": 1, "value": 56.4, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "旅游链条的机会不在单纯客流复苏，而在能抵御客单压力并通过轻资产、零售化或供应链变现的公司。", "summary": "截至2026-06-17，本卡片支持为酒店、餐馆与休闲产业链建立聚焦研究队列。旅游客流恢复强于客单修复，最值得补覆盖的是ATAT、0780.HK、300144.SZ、9922.HK和9869.HK。核心判断是优先找能把流量转化为零售、酒店管理、轻资产演艺或供应链利润的公司，同时跟踪外卖成本、租约收…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "利用卫星遥感与移动支付数据验证‘高频低客单’驱动下的门店密度扩张与消费频次变化", "summary": "卫星门店密度＋移动支付频次只能证实供给端密度扩张与‘频次升、客单降’的行为特征，无法单独证明需求端消费规模化。瑞幸2025年门店31,048家(+39%)、营收492.9亿元(+43%)，蜜雪超4.6万家(57%在三线及以下)，亚朵2,015家酒店、零售高增——供给为真且可卫星核验。但人民银行20…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "模块化的降本红利无法显著降低终端售价，但通过CAPEX杠杆与单店模型重塑驱动门店密度扩张，间接催化'高频低客单'消费模式的规模化。", "summary": "截至2026-06-17，模块化建造在酒店与连锁餐饮端确实把单房/单店CAPEX压缩15–25%、工期压缩30–50%，但红利绝大部分被加盟商回收期缩短、品牌方加盟费抽成和租金/人工通胀吸收，传导到终端ADR/客单价的弹性<0.2[自测算]。亚朵2025年ADR -3.1%、OCC +2.4ppt…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 1, "value": 52.6, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "休闲产业链被市场当作无差别弱消费交易而低覆盖（30日仅9/97激活），但链内分化剧烈。1—5月社零仅+1.4%、5月餐饮减速至+0.6%[S1]，掩盖了个体性自救向上拐头。最具论点价值的低覆盖标的：亚朵ATAT（零售+54.4%、占比38.1%，上调全年指引，17/14/11x PE[S3][S4…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "酒店/餐馆/休闲链被低覆盖（30日激活9/97），宏观基调为‘量增价减’（五一出游3.25亿人次+3.6%，但总花费1854.92亿仅+2.9%，人均约持平略降）。市场用同一条疲软宏观线给全链定价，错配了少数个体催化剂。建议入队：君亭(301073，湖北国资36%控股权变更+酒店REITs资格，Q…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 3, "neutral": 0, "value": 45, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "锦江酒店等所谓的“防御性”标的在高经营杠杆下极具业绩脆弱性，一旦RevPAR出现负向冲击将引发主动基金清仓、融资盘强平与赎回负反馈共振的流动性踩踏，价格跌幅最高可达-50.33%。", "summary": "本研究支持Research note 5结论，对以锦江酒店为首的A股消费板块“伪防御”标的进行拥挤度与平仓踩踏压力测试。锦江自由流通市值约72.75亿元，主动基金持仓占比达15.81%，面临退出瓶颈。若RevPAR下跌-5%，高经营杠杆（19.15x）将导致利润回落-95.75%，引发平仓踩踏。在…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "在消费下沉与成本高企周期中，轻资产管理费模式、上游降本供应链及垄断高分红景区构成了消费板块最核心的毛利防御护城河，而重资产直营及外卖依赖型资产则面临估值与利润双重去杠杆。", "summary": "基于2026-06-23的研究，本报告构建了住宿餐饮与休闲产业链的组合配置框架。在食材成本上涨、外卖平台抽佣上升及固定成本去杠杆的三重压力下，板块内部分化加剧。建议超配轻资产加盟连锁（华住集团、百胜中国）、上游标准化供应链（安井食品）和高红利无债垄断景区（丽江股份、九华旅游）；减配重资产直营餐饮（…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 1, "neutral": 0, "value": 59.8, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-27", "supports": 0, "risks": 3, "neutral": 1, "value": 52.2, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "酒店、餐馆与休闲产业链是分化而非复苏：2026年1-5月餐饮收入+3.1%、五一出游3.25亿人次+3.6%但人均消费约-0.7%，频次上行、客单下行。低覆盖且立论最强的标的为亚朵ATAT（零售飞轮Q1+54.4%、净利+90%）、千味央厨001215（B端供应链见底、Q1收入+23.8%）、君亭…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "酒店/餐馆/休闲链30日激活率仅9.3%，严重欠覆盖且处拐点。2026复苏呈\"量增价减\"：五一人次+3.6%但花费+2.9%，RevPAR由价格而非量驱动，alpha集中在业绩拐点/自救/轻资产名而非beta。筛出4只低覆盖setup：君亭酒店(301073,Q1归母+135%、君澜并表轻资产化，…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 1, "neutral": 1, "value": 65.6, "events": [ { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97） 3%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "summary": "2026年年中酒店/餐饮/休闲链结构性分化而非普涨：出行量低个位数增长但人均消费滞后（五一人次+3.6%、花费+2.9%；端午人次+4.4%、花费+4.0%），且2025年禁酒令精准摧毁商务宴请渠道（单店-20%、高端餐饮最高-70%）。Alpha不在拥挤的酒店龙头，而在与RevPAR脱钩的零售飞…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] } ], "href": "/investor/theses/thesis-90ccc2857a/", "networkHref": "/investor/research-network/#thesis-90ccc2857a" }, { "id": "thesis-e842406d3e", "source": "auto_topic", "title": "海洋经济：政策催化、装备订单与资源开发复盘", "titleEn": "Practical research topic: 海洋经济：政策催化、装备订单与资源开发复盘", "lane": "AI 基础设施", "laneEn": "AI 基础设施", "laneHref": "/investor/lanes/ai/", "investableFocus": "", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ", "TLT", "APO", "META", "MAR" ], "coreView": "该主线来自 18 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-17 至 2026-06-29；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 42, "eventCount": 31, "firstSeen": "2026-05-17", "lastSeen": "2026-06-29", "supports": 2, "risks": 27, "neutral": 2, "analystCount": 1, "sourceTypeCounts": { "研究笔记": 7, "研究线程": 18, "深度研究": 6 }, "topTerms": [ "新华财经-海洋经济产业链", "研究活跃度与投资线索复盘" ], "whatChanged": [ "2026-06-29: 产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "2026-06-28: 产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "2026-06-28: 海洋经济产业链处于低覆盖及严重低估水平，随着深海海风建设加速与AI基础设施冷却革命推进，亚星锚链、中远海特、海力风电与海兰信等核心标的迎来基本面共振与强溢出效应。", "2026-06-26: 产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "2026-06-23: 产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）" ], "practical": { "sectorLane": "AI 基础设施", "sectorLaneEn": "AI 基础设施", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "deep_research_candidate", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "深度研究候选", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取但证据密度较高的主题，适合进入深度研究排队，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ", "TLT", "APO", "META", "MAR" ], "primaryStocks": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" } ], "instrumentTags": [ "NVDA", "NVIDIA", "TSM", "TSMC", "AVGO", "Broadcom", "MU", "Micron", "ANET", "Arista Networks", "VRT", "Vertiv" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "中高：自动主题待深研", "deepResearchReason": "证据深度 67，来源 3 类，近期事件 22 条，适合沉淀为专题复盘或深度研究。", "evidenceBreadth": { "total": 31, "sourceDiversity": 3, "analystCount": 1, "recentEvents": 22, "supportRatio": 0.06, "riskRatio": 0.87, "conflictScore": 97.2, "evidenceDepth": 66.8 }, "score": 31.3 }, "events": [ { "date": "2026-06-29", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-28", "title": "海洋经济产业链处于低覆盖及严重低估水平，随着深海海风建设加速与AI基础设施冷却革命推进，亚星锚链、中远海特、海力风电与海兰信等核心标的迎来基本面共振与强溢出效应。", "summary": "截至2026-06-29，新华财经-海洋经济产业链（148只标的）过去30天的研究激活率仅为6.1%。本调研挑选了亚星锚链（系泊系统龙头）、中远海特（特种运输龙头）、海力风电（海风基础与导管架龙头）和海兰信（海底数据中心先锋）等4家低覆盖度核心标的。这些公司受益于造船周期、海风深海化及AI算力绿色冷却等催化，具有强阿尔法潜力，并向材料、AI基础设施及公用事…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 7.75, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-28", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-26", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-23", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-22", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-21", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-21", "title": "新华财经-海洋经济产业链整体覆盖度低，但在深远海风电加速与全球港口自动化浪潮下，建议超配具备核心技术壁垒及订单反转支撑的亚星锚链与润邦股份。", "summary": "对低覆盖度新华财经-海洋经济产业链进行调研，识别出亚星锚链（深海漂浮式风电系泊链产量翻倍）、润邦股份（合同负债增长近100%迎来业绩拐点）、博迈科（海外FPSO模块订单创历史新高）和振华重工（自动化港机出口高增）4家高阿尔法标的，并评估了原材料、汇率和地缘政治等跨链溢出风险。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.75, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-19", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-19", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "海洋经济链结构性欠覆盖（30日激活4.7%），但政策＋需求双顺风真实：2025年海洋生产总值11.018万亿、同比+5.5%、占GDP 7.9%，海洋新兴产业+7.3%；深海科技首入2025政府工作报告。工业口径下价值沉淀于装备脊柱（海缆、系泊链、桩基、水下声呐），覆盖最薄。精选4-5只低覆盖标的：亚星锚链(601890)、起帆电缆(605222)、海兰信…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.75, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-18", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-18", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-18", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "海洋经济链全板块最低覆盖（30日仅7/148激活），却踩在最清晰政策顺风上：2024海洋GDP 10.5万亿(占GDP 7.8%)，2026年总量首破11万亿；十五五首次单列海洋经济，深海科技首入政府工作报告。基本面验证真实周期——海工装备增加值+10.2%、海上风电2025新增并网+60%。错定价源于关注度而非盈利。最具论点价值的低覆盖标的：亚星锚链(6…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-17", "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-17", "title": "海洋经济产业链正从传统船舶向深海高毛利装备结构性升级，市场套用传统周期估值导致了严重定价偏差，亚星锚链、迪威尔等未覆盖龙头具备强订单反转及估值修复潜力。", "summary": "截至2026-06-17，新华财经-海洋经济产业链的30日激活率仅为4.7%，存在严重低配与研究空白。深海高端制造正经历结构性转型，亚星锚链（系泊链中标风电项目）、迪威尔（进入全球油服供应链）、中科海讯（重磅订单反转）等未覆盖细分龙头具备高确定性与估值弹性，建议将其纳入后续深度研究队列。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 7, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" }, { "date": "2026-06-15", "title": "海洋经济正迈向深海化与智能化的结构性转型，亚星锚链、海兰信等龙头依靠高壁垒技术及饱满在手订单，是极具阿尔法价值的深度调研标的。", "summary": "本卡片调研了新华财经-海洋经济产业链。当前该板块30日激活率仅为5.4%（8/148），处于严重低估状态。我们筛选出亚星锚链（深海系泊系统）、迪威尔（油气阀件国产替代）、中远海特（风电与汽车特种运输）及海兰信（海底数据中心）四只低关注度且成长动能强劲的标的，并提示了钢材成本、电网并网、地缘关税等跨链溢出风险。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 7.75, "matchedTerms": [ "新华财经-海洋经济产业链" ], "href": "" } ], "timeline": [ { "date": "2026-05-17", "supports": 1, "risks": 1, "neutral": 0, "value": 44, "events": [ { "title": "海洋经济产业链的机会在低覆盖的关键装备和海洋电子环节，研究重点应是订单质量、利润率修复和现金流，而不是概念热度。", "summary": "截至 2026-05-17，本卡支持把新华财经-海洋经济产业链纳入研究队列，但反对无差别板块 beta。2025 年海洋生产总值 11.018 万亿元、同比增长 5.5%，2026 年一季度 2.6 万亿元、同比增长 5.4%，海工装备订单数据强劲。优先深挖亚星锚链、中国海防、润邦股份；海兰信和江…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 3.4%, 5/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 3.4%, 5/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 31.4, "events": [ { "title": "中国电力设备出海：AI 电网需求、贸易壁垒与利润率分化", "summary": "全球 AI 电网投资为中国电力设备提供需求窗口，但关税、认证、交付能力和原材料成本会决定哪些企业真正把订单转化为利润。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/china-power-equipment-export-policy-8678db7f/" }, { "title": "AI 算力资本开支、电力瓶颈与通胀再定价", "summary": "AI 基础设施需求正在先通过电力、电网、变压器与材料供给形成成本脉冲，再通过生产率与架构效率形成中期缓释；投资结论取决于这两个方向谁先兑现。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-power-inflation-transmission-0c9ba15c/" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 1, "neutral": 0, "value": 46.8, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "消费降级、银发经济与大健康结构性机会", "summary": "消费链条正在从总量弹性转向结构分化：银发消费、医疗健康、糖替代和 B 端餐饮渗透可能形成防御性需求，但需要验证支付能力和渠道真实动销。", "sourceLabel": "深度研究", "supportLabel": "支持", "supportType": "supports", "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 34.2, "events": [ { "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 0, "neutral": 1, "value": 53.4, "events": [ { "title": "指派 industrials-analyst 负责新华财经-海洋经济产业链", "summary": "指派 industrials-analyst 负责新华财经-海洋经济产业链 针对新华财经-海洋经济产业链（激活率 5.4%）进行覆盖。依据 §6.1。", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-13", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-14", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-15", "supports": 0, "risks": 3, "neutral": 0, "value": 36.6, "events": [ { "title": "海洋经济正迈向深海化与智能化的结构性转型，亚星锚链、海兰信等龙头依靠高壁垒技术及饱满在手订单，是极具阿尔法价值的深度调研标的。", "summary": "本卡片调研了新华财经-海洋经济产业链。当前该板块30日激活率仅为5.4%（8/148），处于严重低估状态。我们筛选出亚星锚链（深海系泊系统）、迪威尔（油气阀件国产替代）、中远海特（风电与汽车特种运输）及海兰信（海底数据中心）四只低关注度且成长动能强劲的标的，并提示了钢材成本、电网并网、地缘关税等跨…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-17", "supports": 0, "risks": 1, "neutral": 1, "value": 53, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "海洋经济产业链正从传统船舶向深海高毛利装备结构性升级，市场套用传统周期估值导致了严重定价偏差，亚星锚链、迪威尔等未覆盖龙头具备强订单反转及估值修复潜力。", "summary": "截至2026-06-17，新华财经-海洋经济产业链的30日激活率仅为4.7%，存在严重低配与研究空白。深海高端制造正经历结构性转型，亚星锚链（系泊链中标风电项目）、迪威尔（进入全球油服供应链）、中科海讯（重磅订单反转）等未覆盖细分龙头具备高确定性与估值弹性，建议将其纳入后续深度研究队列。", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-18", "supports": 0, "risks": 3, "neutral": 0, "value": 39.4, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "海洋经济链全板块最低覆盖（30日仅7/148激活），却踩在最清晰政策顺风上：2024海洋GDP 10.5万亿(占GDP 7.8%)，2026年总量首破11万亿；十五五首次单列海洋经济，深海科技首入政府工作报告。基本面验证真实周期——海工装备增加值+10.2%、海上风电2025新增并网+60%。错定…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-19", "supports": 0, "risks": 2, "neutral": 0, "value": 46.8, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "海洋经济链结构性欠覆盖（30日激活4.7%），但政策＋需求双顺风真实：2025年海洋生产总值11.018万亿、同比+5.5%、占GDP 7.9%，海洋新兴产业+7.3%；深海科技首入2025政府工作报告。工业口径下价值沉淀于装备脊柱（海缆、系泊链、桩基、水下声呐），覆盖最薄。精选4-5只低覆盖标的…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 2, "neutral": 0, "value": 48.2, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "新华财经-海洋经济产业链整体覆盖度低，但在深远海风电加速与全球港口自动化浪潮下，建议超配具备核心技术壁垒及订单反转支撑的亚星锚链与润邦股份。", "summary": "对低覆盖度新华财经-海洋经济产业链进行调研，识别出亚星锚链（深海漂浮式风电系泊链产量翻倍）、润邦股份（合同负债增长近100%迎来业绩拐点）、博迈科（海外FPSO模块订单创历史新高）和振华重工（自动化港机出口高增）4家高阿尔法标的，并评估了原材料、汇率和地缘政治等跨链溢出风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 1, "neutral": 0, "value": 55.6, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148） 7%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 1, "neutral": 0, "value": 57, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-26", "supports": 0, "risks": 1, "neutral": 0, "value": 58.4, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-28", "supports": 0, "risks": 2, "neutral": 0, "value": 53.8, "events": [ { "title": "海洋经济产业链处于低覆盖及严重低估水平，随着深海海风建设加速与AI基础设施冷却革命推进，亚星锚链、中远海特、海力风电与海兰信等核心标的迎来基本面共振与强溢出效应。", "summary": "截至2026-06-29，新华财经-海洋经济产业链（148只标的）过去30天的研究激活率仅为6.1%。本调研挑选了亚星锚链（系泊系统龙头）、中远海特（特种运输龙头）、海力风电（海风基础与导管架龙头）和海兰信（海底数据中心先锋）等4家低覆盖度核心标的。这些公司受益于造船周期、海风深海化及AI算力绿色…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-29", "supports": 0, "risks": 1, "neutral": 0, "value": 61.2, "events": [ { "title": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "summary": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148） 1%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "sourceLabel": "研究线程", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-e842406d3e/", "networkHref": "/investor/research-network/#thesis-e842406d3e" }, { "id": "thesis-bdb5af1d34", "source": "auto_topic", "title": "纺织服装：出口利润率、合规与库存周期复盘", "titleEn": "Practical research topic: 纺织服装：出口利润率、合规与库存周期复盘", "lane": "宏观利率与汇率", "laneEn": "宏观利率与汇率", "laneHref": "/investor/lanes/lane-cdafce64/", "investableFocus": "", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "NKE", "name": "Nike", "market": "US", "role": "全球运动品牌对照", "roleEn": "全球运动品牌对照", "bucket": "peer", "href": "/investor/stocks/NKE/" } ], "stockTickers": [ "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM", "APO", "META", "MAR", "NKE" ], "coreView": "该主线来自 10 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-20 至 2026-06-02；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 37, "eventCount": 18, "firstSeen": "2026-05-18", "lastSeen": "2026-06-02", "supports": 2, "risks": 13, "neutral": 3, "analystCount": 8, "sourceTypeCounts": { "深度研究": 6, "研究笔记": 12 }, "topTerms": [ "纺织服装产业链", "研究活跃度与投资线索复盘" ], "whatChanged": [ "2026-06-02: 纺织服装链最有价值的研究不是追总量复苏，而是寻找成本传导、海外产能和品牌经营杠杆已经被2025年及2026年一季度数据验证的标的。", "2026-05-30: 金融基础设施不再是中性工具，而是通过中信保信贷、绿色供应链金融和跨境信贷加速了纺织链的优胜劣汰。", "2026-05-28: 纺织服装产业总量增长平稳的表象下正发生深层业绩分化，具备海外成熟产能布局、高技术壁垒辅料重服务，或定位高端黏性细分客群的中坚企业，将在行业补库与关税重塑中迎来确定性超额收益。", "2026-05-27: Research note 01 几乎未产出实质研究，但其执行失误本身正是可操作的运营信号；本卡发出 7 条提案（4 safe + 3 review）并把线程交还 china-macro 重新锚定纺织服装产业链的宏观边际定价权。", "2026-05-27: 纺织服装产业链的宏观定价权已从下游 OEM 转移至具备‘反内卷’共识的上游聚酯龙头及具备‘绿色法典’护城河的印染巨头。" ], "practical": { "sectorLane": "宏观利率与汇率", "sectorLaneEn": "宏观利率与汇率", "exposureBucket": "宏观利率与汇率", "tone": "orange", "actionCode": "radar", "actionLabel": "主题雷达：先建观察仓", "actionTone": "blue", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "主题观察", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取主题，先用于发现研究活跃度，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" }, { "ticker": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国风险代理", "roleEn": "离岸中国风险代理", "bucket": "core", "href": "/investor/stocks/FXI/" }, { "ticker": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "roleEn": "EM 风险资产", "bucket": "core", "href": "/investor/stocks/EEM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "NKE", "name": "Nike", "market": "US", "role": "全球运动品牌对照", "roleEn": "全球运动品牌对照", "bucket": "peer", "href": "/investor/stocks/NKE/" } ], "stockTickers": [ "TLT", "IEF", "UUP", "FXY", "GLD", "XLE", "FXI", "EEM", "APO", "META", "MAR", "NKE" ], "primaryStocks": [ { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元指数代理", "roleEn": "美元指数代理", "bucket": "core", "href": "/investor/stocks/UUP/" }, { "ticker": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "roleEn": "日元代理", "bucket": "core", "href": "/investor/stocks/FXY/" }, { "ticker": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "roleEn": "实际利率反向代理", "bucket": "core", "href": "/investor/stocks/GLD/" }, { "ticker": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "roleEn": "能源通胀代理", "bucket": "core", "href": "/investor/stocks/XLE/" } ], "instrumentTags": [ "TLT", "iShares 20+ Year Treasury Bond …", "IEF", "iShares 7-10 Year Treasury Bond…", "UUP", "Invesco DB US Dollar Index Bull…", "FXY", "Invesco CurrencyShares Japanese…", "GLD", "SPDR Gold Shares", "XLE", "Energy Select Sector SPDR" ], "reviewCadence": "每日复核", "confirmationSignals": [ "CPI/PCE 与油价是否重新推高实际利率", "美元、人民币和日元套息交易是否发生拥挤反转", "长端利率是否压制权益估值扩张" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "现金", "短久期", "美元指数", "黄金" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 18, "sourceDiversity": 2, "analystCount": 8, "recentEvents": 1, "supportRatio": 0.11, "riskRatio": 0.72, "conflictScore": 80.9, "evidenceDepth": 52.2 }, "score": 29.3 }, "events": [ { "date": "2026-06-02", "title": "纺织服装链最有价值的研究不是追总量复苏，而是寻找成本传导、海外产能和品牌经营杠杆已经被2025年及2026年一季度数据验证的标的。", "summary": "截至2026-06-02，本卡支持开启纺织服装产业链研究队列。国内服装鞋帽针纺织品1-4月零售额同比增长8.1%，但出口仅温和增长且原料端棉花、粘胶涨价压缩中游利润。优先深挖健盛集团、开润股份、浙江自然、报喜鸟、地素时尚；华利集团作为运动鞋制造基准而非低覆盖优先项。核心跨链风险包括原料通胀、贸易与物流波动、线上/线下渠道分化，以及户外消费与新能源车、运动户…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.25, "matchedTerms": [ "纺织服装产业链" ], "href": "" }, { "date": "2026-05-30", "title": "金融基础设施不再是中性工具，而是通过中信保信贷、绿色供应链金融和跨境信贷加速了纺织链的优胜劣汰。", "summary": "金融基础设施（银行、中信保、金融科技）已成为纺织服装产业链两极分化的核心推手。SEA 一体化龙头利用中信保 120 天账期和绿色供应链金融（Green SCF）建立起显著的财务壁垒，而纯大陆出口商因缺乏数字追溯和海外布局正面临信贷收缩。5.8% 的研究激活率反映了市场对“金融科技护城河”的定价缺失，具备数字化能力的国内品牌正处于估值洼地。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 7.25, "matchedTerms": [ "纺织服装产业链" ], "href": "" }, { "date": "2026-05-28", "title": "纺织服装产业总量增长平稳的表象下正发生深层业绩分化，具备海外成熟产能布局、高技术壁垒辅料重服务，或定位高端黏性细分客群的中坚企业，将在行业补库与关税重塑中迎来确定性超额收益。", "summary": "本卡聚焦于严重低覆盖的纺织服装产业链（近30天激活率仅5.8%），在2026年一季度全国出口同比小幅增长1.2%的宏观背景下，系统梳理了健盛集团、伟星股份、百隆东方和比音勒芬4家特色中坚股。它们通过海外产能套利、辅料服务壁垒、棉花成本套利及高端消费客群黏性实现强业绩突围。同时，本卡系统总结了上游原材料、地缘物流、汇率波动及关税等跨链共振风险，并规划了后继宏…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.25, "matchedTerms": [ "纺织服装产业链" ], "href": "" }, { "date": "2026-05-27", "title": "纺织服装产业链当前不具备宏观贝塔进攻属性，整体定价权仍在走弱但收敛；只有聚酯长丝、国产运动鞋服、印染头部三个结构性强环节值得主题型超配，宏观拐点由关税/大宗/汇率三外生变量决定，不靠内需。", "summary": "纺织服装产业链当前的宏观边际定价权呈两端分化、整体偏弱格局：聚酯长丝（PTA/POY 寡头化）、国产运动鞋服（安踏+李宁市占已超 Nike 中国）、印染头部（环保供给侧二阶导）是仅有的三个结构性强环节；中游纱线织造长期过剩、下游 OEM 出口在特朗普 2.0 关税升级与 UFLPA 推定禁令的双重夹击下持续被削弱。当前拐点取决于三个外生变量——ICE 棉/…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 7, "matchedTerms": [ "纺织服装产业链" ], "href": "" }, { "date": "2026-05-27", "title": "纺织服装产业链的宏观定价权已从下游 OEM 转移至具备‘反内卷’共识的上游聚酯龙头及具备‘绿色法典’护城河的印染巨头。", "summary": "基于 2026 年 5 月的宏观背景，纺织服装产业链正经历从“规模竞争”到“利润修复”的结构性转型。尽管 Trump 2.0 关税将有效税率推高至 25%-45% 且人民币走强至 6.78，但聚酯龙头通过“反内卷”减产修复了 PTA/POY 毛利，印染行业受益于 2026 年 8 月生效的《生态环境法典》显著抬高准入门槛。建议超配一体化石化巨头与环保合规龙…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.75, "matchedTerms": [ "纺织服装产业链" ], "href": "" }, { "date": "2026-05-27", "title": "Research note 01 几乎未产出实质研究，但其执行失误本身正是可操作的运营信号；本卡发出 7 条提案（4 safe + 3 review）并把线程交还 china-macro 重…", "summary": "Research note 01 由消费分析师执行了一项与其主域不符的目录探索元任务，且 topic/question 字段被填入提示词原文。本卡作为内部运营官将其中的运营信号提炼为 7 条 operational_proposals：4 条 safe-class（工作区预检 hook、低激活率监控、Gemini 路径屏蔽 verified_fact、文件…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "supports", "supportLabel": "支持", "score": 6.25, "matchedTerms": [ "纺织服装产业链" ], "href": "" }, { "date": "2026-05-26", "title": "能源行业分析报告：纺织服装产业链的“能源强度陷阱” —— 挑战上游功能性材料的安全垫逻辑 报告日期： 2026 05 26 分析师： 能源行业分析师 (Energy Analyst) 研究线…", "summary": "能源行业分析报告：纺织服装产业链的“能源强度陷阱” —— 挑战上游功能性材料的安全垫逻辑 报告日期： 2026 05 26 分析师： 能源行业分析师 (Energy Analyst) 研究线程标识：已归档立场： 否定 (Deny) 核心观点 (Thesis) 尽管材料分析师 (Research note 06) 与首席策略师 (Research note …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 7.25, "matchedTerms": [ "纺织服装产业链" ], "href": "" }, { "date": "2026-05-26", "title": "纺织服装产业链仍可投，但 A 股表达必须从低利润率出口 OEM beta 迁移到功能材料、内需渠道、合规数字化与具备 AI 小单快反能力的质量制造。", "summary": "截至 2026-05-26，本卡综合前四张卡：关税和人民币都不是唯一主变量，A 股策略应转向“平台权力暴露 + AI 工艺周期 + 合规资本开支 + 资产负债表质量”的五因子框架。建议在 100 单位纺织服装及相邻 A 股组合中，超配上游功能材料、国内品牌渠道、合规数字化/RFID/工业软件，精选一体化 OEM，低配纯中国本土 CMT/OEM 与 DTC …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 7, "matchedTerms": [ "纺织服装产业链" ], "href": "" }, { "date": "2026-05-26", "title": "人民币汇率走势与纺织服装产业链出口定价能力的联动分析 报告日期： 2026 05 26（亚洲/新加坡） | 分析师 ID： fx strategist | 研究线程标识：已归档摘要 (Exe…", "summary": "人民币汇率走势与纺织服装产业链出口定价能力的联动分析 报告日期： 2026 05 26（亚洲/新加坡） | 分析师 ID： fx strategist | 研究线程标识：已归档摘要 (Executive Summary) 截至 2026 年 5 月，全球贸易壁垒与关税格局正对中国纺织服装产业链（纺织服装产业链）施加前所未有的结构性重压 [S5]。在主要出口…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "纺织服装产业链" ], "href": "" }, { "date": "2026-05-25", "title": "纺织服装链的政策护城河已经从低成本产能转向原产地合规、纤维溯源和关税传导能力。", "summary": "截至2026-05-25，本卡对纺织服装产业链的贸易政策风险作压力测试：支持Research note 01对全球化龙头的方向，但要求对美国关税波动、de minimis关闭、China Section 301、UFLPA原产地执法和欧盟ESPR/强迫劳动/EPR规则进行利润率折价。10pp关税冲击在40%美国收入敞口、30%客户压价传导、8%净利率假设下…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "纺织服装产业链" ], "href": "" }, { "date": "2026-05-24", "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最清晰的战略相对输家；若只看语料中直接证据，Meta 是最清晰…", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" }, { "date": "2026-05-23", "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "date": "2026-05-23", "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" }, { "date": "2026-05-23", "title": "纺织服装产业链的最佳研究入口是出口供应链迁移与订单质量，而不是泛化的服装消费复苏。", "summary": "2026-05-23 初筛认为纺织服装链条不是简单内需 beta，优先关注具备海外产能、品牌客户项目、现金转化和成本传导能力的公司。建议深挖 2232.HK 晶苑国际、2199.HK 维珍妮、300577.SZ 开润股份、605080.SH 浙江自然、2678.HK 天虹国际。主要外溢风险来自关税/原产地规则、棉花价格上行、国内服装零售增速放缓。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "纺织服装产业链" ], "href": "" }, { "date": "2026-05-20", "title": "纺织服装产业链在经历深度出清后迎来结构性分化，具备海外供应链闭环能力的辅料与代工龙头，以及深耕高端细分品类的品牌服饰展现出强劲的盈利弹性和高确定性出海Alpha。", "summary": "本报告深入调研了激活率仅4.7%的纺织服装产业链。在2026年体育大年需求回暖与品牌培育政策推动下，制造端加速海外闭环布局，品牌端呈现高端分化。筛选出比音勒芬（中高端运动）、健盛集团（棉袜无缝）、伟星股份（世界级辅料）与锦泓集团（去杠杆运营）四家低覆盖、高业绩确定性的标的，并评估了上游高油价（PTA成本传导）、棉花紧平衡以及红海运费上涨等跨链风险。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6, "matchedTerms": [ "纺织服装产业链" ], "href": "" }, { "date": "2026-05-19", "title": "消费降级、银发经济与大健康结构性机会", "summary": "消费链条正在从总量弹性转向结构分化：银发消费、医疗健康、糖替代和 B 端餐饮渗透可能形成防御性需求，但需要验证支付能力和渠道真实动销。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "supports", "supportLabel": "支持", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" } ], "timeline": [ { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 30, "events": [ { "title": "中国电力设备出海：AI 电网需求、贸易壁垒与利润率分化", "summary": "全球 AI 电网投资为中国电力设备提供需求窗口，但关税、认证、交付能力和原材料成本会决定哪些企业真正把订单转化为利润。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/china-power-equipment-export-policy-8678db7f/" }, { "title": "AI 算力资本开支、电力瓶颈与通胀再定价", "summary": "AI 基础设施需求正在先通过电力、电网、变压器与材料供给形成成本脉冲，再通过生产率与架构效率形成中期缓释；投资结论取决于这两个方向谁先兑现。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-power-inflation-transmission-0c9ba15c/" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 0, "neutral": 0, "value": 51.4, "events": [ { "title": "消费降级、银发经济与大健康结构性机会", "summary": "消费链条正在从总量弹性转向结构分化：银发消费、医疗健康、糖替代和 B 端餐饮渗透可能形成防御性需求，但需要验证支付能力和渠道真实动销。", "sourceLabel": "深度研究", "supportLabel": "支持", "supportType": "supports", "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" } ] }, { "date": "2026-05-20", "supports": 0, "risks": 1, "neutral": 0, "value": 38.8, "events": [ { "title": "纺织服装产业链在经历深度出清后迎来结构性分化，具备海外供应链闭环能力的辅料与代工龙头，以及深耕高端细分品类的品牌服饰展现出强劲的盈利弹性和高确定性出海Alpha。", "summary": "本报告深入调研了激活率仅4.7%的纺织服装产业链。在2026年体育大年需求回暖与品牌培育政策推动下，制造端加速海外闭环布局，品牌端呈现高端分化。筛选出比音勒芬（中高端运动）、健盛集团（棉袜无缝）、伟星股份（世界级辅料）与锦泓集团（去杠杆运营）四家低覆盖、高业绩确定性的标的，并评估了上游高油价（PT…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 3, "neutral": 0, "value": 28.2, "events": [ { "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" }, { "title": "纺织服装产业链的最佳研究入口是出口供应链迁移与订单质量，而不是泛化的服装消费复苏。", "summary": "2026-05-23 初筛认为纺织服装链条不是简单内需 beta，优先关注具备海外产能、品牌客户项目、现金转化和成本传导能力的公司。建议深挖 2232.HK 晶苑国际、2199.HK 维珍妮、300577.SZ 开润股份、605080.SH 浙江自然、2678.HK 天虹国际。主要外溢风险来自关税…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" } ] }, { "date": "2026-05-25", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "纺织服装链的政策护城河已经从低成本产能转向原产地合规、纤维溯源和关税传导能力。", "summary": "截至2026-05-25，本卡对纺织服装产业链的贸易政策风险作压力测试：支持Research note 01对全球化龙头的方向，但要求对美国关税波动、de minimis关闭、China Section 301、UFLPA原产地执法和欧盟ESPR/强迫劳动/EPR规则进行利润率折价。10pp关税冲…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-26", "supports": 0, "risks": 2, "neutral": 1, "value": 41.4, "events": [ { "title": "能源行业分析报告：纺织服装产业链的“能源强度陷阱” —— 挑战上游功能性材料的安全垫逻辑 报告日期： 2026 05 26 分析师： 能源行业分析师 (Energy Analyst)…", "summary": "能源行业分析报告：纺织服装产业链的“能源强度陷阱” —— 挑战上游功能性材料的安全垫逻辑 报告日期： 2026 05 26 分析师： 能源行业分析师 (Energy Analyst) 研究线程标识：已归档立场： 否定 (Deny) 核心观点 (Thesis) 尽管材料分析师 (Research n…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "纺织服装产业链仍可投，但 A 股表达必须从低利润率出口 OEM beta 迁移到功能材料、内需渠道、合规数字化与具备 AI 小单快反能力的质量制造。", "summary": "截至 2026-05-26，本卡综合前四张卡：关税和人民币都不是唯一主变量，A 股策略应转向“平台权力暴露 + AI 工艺周期 + 合规资本开支 + 资产负债表质量”的五因子框架。建议在 100 单位纺织服装及相邻 A 股组合中，超配上游功能材料、国内品牌渠道、合规数字化/RFID/工业软件，精选…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "人民币汇率走势与纺织服装产业链出口定价能力的联动分析 报告日期： 2026 05 26（亚洲/新加坡） | 分析师 ID： fx strategist | 研究线程标识：已归档摘要 …", "summary": "人民币汇率走势与纺织服装产业链出口定价能力的联动分析 报告日期： 2026 05 26（亚洲/新加坡） | 分析师 ID： fx strategist | 研究线程标识：已归档摘要 (Executive Summary) 截至 2026 年 5 月，全球贸易壁垒与关税格局正对中国纺织服装产业链（纺…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-27", "supports": 1, "risks": 1, "neutral": 1, "value": 56.8, "events": [ { "title": "纺织服装产业链当前不具备宏观贝塔进攻属性，整体定价权仍在走弱但收敛；只有聚酯长丝、国产运动鞋服、印染头部三个结构性强环节值得主题型超配，宏观拐点由关税/大宗/汇率三外生变量决定，不靠…", "summary": "纺织服装产业链当前的宏观边际定价权呈两端分化、整体偏弱格局：聚酯长丝（PTA/POY 寡头化）、国产运动鞋服（安踏+李宁市占已超 Nike 中国）、印染头部（环保供给侧二阶导）是仅有的三个结构性强环节；中游纱线织造长期过剩、下游 OEM 出口在特朗普 2.0 关税升级与 UFLPA 推定禁令的双重…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" }, { "title": "纺织服装产业链的宏观定价权已从下游 OEM 转移至具备‘反内卷’共识的上游聚酯龙头及具备‘绿色法典’护城河的印染巨头。", "summary": "基于 2026 年 5 月的宏观背景，纺织服装产业链正经历从“规模竞争”到“利润修复”的结构性转型。尽管 Trump 2.0 关税将有效税率推高至 25%-45% 且人民币走强至 6.78，但聚酯龙头通过“反内卷”减产修复了 PTA/POY 毛利，印染行业受益于 2026 年 8 月生效的《生态环…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "Research note 01 几乎未产出实质研究，但其执行失误本身正是可操作的运营信号；本卡发出 7 条提案（4 safe + 3 review）并把线程交还 china-mac…", "summary": "Research note 01 由消费分析师执行了一项与其主域不符的目录探索元任务，且 topic/question 字段被填入提示词原文。本卡作为内部运营官将其中的运营信号提炼为 7 条 operational_proposals：4 条 safe-class（工作区预检 hook、低激活率监…", "sourceLabel": "研究笔记", "supportLabel": "支持", "supportType": "supports", "href": "" } ] }, { "date": "2026-05-28", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "纺织服装产业总量增长平稳的表象下正发生深层业绩分化，具备海外成熟产能布局、高技术壁垒辅料重服务，或定位高端黏性细分客群的中坚企业，将在行业补库与关税重塑中迎来确定性超额收益。", "summary": "本卡聚焦于严重低覆盖的纺织服装产业链（近30天激活率仅5.8%），在2026年一季度全国出口同比小幅增长1.2%的宏观背景下，系统梳理了健盛集团、伟星股份、百隆东方和比音勒芬4家特色中坚股。它们通过海外产能套利、辅料服务壁垒、棉花成本套利及高端消费客群黏性实现强业绩突围。同时，本卡系统总结了上游原…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-30", "supports": 0, "risks": 0, "neutral": 1, "value": 57.6, "events": [ { "title": "金融基础设施不再是中性工具，而是通过中信保信贷、绿色供应链金融和跨境信贷加速了纺织链的优胜劣汰。", "summary": "金融基础设施（银行、中信保、金融科技）已成为纺织服装产业链两极分化的核心推手。SEA 一体化龙头利用中信保 120 天账期和绿色供应链金融（Green SCF）建立起显著的财务壁垒，而纯大陆出口商因缺乏数字追溯和海外布局正面临信贷收缩。5.8% 的研究激活率反映了市场对“金融科技护城河”的定价缺失…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-02", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "纺织服装链最有价值的研究不是追总量复苏，而是寻找成本传导、海外产能和品牌经营杠杆已经被2025年及2026年一季度数据验证的标的。", "summary": "截至2026-06-02，本卡支持开启纺织服装产业链研究队列。国内服装鞋帽针纺织品1-4月零售额同比增长8.1%，但出口仅温和增长且原料端棉花、粘胶涨价压缩中游利润。优先深挖健盛集团、开润股份、浙江自然、报喜鸟、地素时尚；华利集团作为运动鞋制造基准而非低覆盖优先项。核心跨链风险包括原料通胀、贸易与…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-bdb5af1d34/", "networkHref": "/investor/research-network/#thesis-bdb5af1d34" }, { "id": "thesis-7b551fae18", "source": "auto_topic", "title": "新一代信息技术：国产替代与算力应用落地复盘", "titleEn": "Practical research topic: 新一代信息技术：国产替代与算力应用落地复盘", "lane": "AI 基础设施", "laneEn": "AI 基础设施", "laneHref": "/investor/lanes/ai/", "investableFocus": "", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ", "APO", "IEF", "CAT", "META", "588000.SH", "MAR" ], "coreView": "该主线来自 21 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-18 至 2026-06-29；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 35, "eventCount": 16, "firstSeen": "2026-05-18", "lastSeen": "2026-06-23", "supports": 1, "risks": 13, "neutral": 2, "analystCount": 1, "sourceTypeCounts": { "深度研究": 6, "研究笔记": 9, "研究线程": 1 }, "topTerms": [ "新一代信息技术产业链", "研究活跃度与投资线索复盘" ], "whatChanged": [ "2026-06-23: 随着AI芯片向高速与高功耗演进，硬件瓶颈已向高速铜连、芯片电感和硅光封测迁移，精选利基中盘标的具备更强超额收益潜力。", "2026-06-22: 在2026年中美云厂商CapEx受阻于电力红线以及大模型推理通缩的双重催化下，新一代信息技术产业链估值正向能耗比占优的定制ASIC上游填料（联瑞新材）以及享受API接口成本大坍缩的软件应用（金山办公）展开结构性重构。", "2026-06-21: 新一代信息技术链条的最佳机会在 AI 基础设施“卖铲人”和端侧 AI 芯片，而不是拥挤的通用 AI 服务器敞口。", "2026-06-12: 新一代信息技术链的超额收益已从泛算力概念转向‘物理硬约束（电力/连接）组件出海’与‘底层芯片（GPU/MCU/MEMS）国产替代业绩拐点验证’，需警惕2026-06-13所处周期的科创50调样后高杠杆标的引发的两融追保跨板块踩踏风险。", "2026-06-11: 指派 tmt-analyst 负责新一代信息技术产业链" ], "practical": { "sectorLane": "AI 基础设施", "sectorLaneEn": "AI 基础设施", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "radar", "actionLabel": "主题雷达：先建观察仓", "actionTone": "blue", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "主题观察", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取主题，先用于发现研究活跃度，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" }, { "ticker": "CAT", "name": "Caterpillar", "market": "US", "role": "全球工程机械对照", "roleEn": "全球工程机械对照", "bucket": "peer", "href": "/investor/stocks/CAT/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "roleEn": "硬科技风险偏好", "bucket": "core", "href": "/investor/stocks/588000.SH/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" } ], "stockTickers": [ "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ", "APO", "IEF", "CAT", "META", "588000.SH", "MAR" ], "primaryStocks": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" } ], "instrumentTags": [ "NVDA", "NVIDIA", "TSM", "TSMC", "AVGO", "Broadcom", "MU", "Micron", "ANET", "Arista Networks", "VRT", "Vertiv" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 16, "sourceDiversity": 3, "analystCount": 1, "recentEvents": 8, "supportRatio": 0.06, "riskRatio": 0.81, "conflictScore": 91.4, "evidenceDepth": 50.9 }, "score": 24 }, "events": [ { "date": "2026-06-23", "title": "随着AI芯片向高速与高功耗演进，硬件瓶颈已向高速铜连、芯片电感和硅光封测迁移，精选利基中盘标的具备更强超额收益潜力。", "summary": "本报告对新一代信息技术产业链进行调研，识别出华丰科技、鼎通科技、铂科新材及罗博特科等 4 家高增长但覆盖不足的标的，分析了其在 224G 铜连接、高端芯片电感与硅光封测设备中的核心壁垒，并揭示了冷却液监管与电网容量等跨链外溢风险。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新一代信息技术产业链" ], "href": "" }, { "date": "2026-06-22", "title": "在2026年中美云厂商CapEx受阻于电力红线以及大模型推理通缩的双重催化下，新一代信息技术产业链估值正向能耗比占优的定制ASIC上游填料（联瑞新材）以及享受API接口成本大坍缩的软件应用（…", "summary": "截至2026-06-22，新一代信息技术产业链正经历算力瓶颈由GPU转向电力/并网及先进封装、以及DeepSeek API降价96.37%引发软件应用毛利重构的剧烈调整。板块激活率处于4.3%极低水平。本报告筛选了联瑞新材、生益电子、宏和科技、金山办公和富创精密等5家低两融、抛压出清且具刚性催化剂的核心标的，并深度剖析了两融踩踏及空间级材料断链等跨链溢出风…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 8, "matchedTerms": [ "新一代信息技术产业链" ], "href": "" }, { "date": "2026-06-21", "title": "新一代信息技术链条的最佳机会在 AI 基础设施“卖铲人”和端侧 AI 芯片，而不是拥挤的通用 AI 服务器敞口。", "summary": "截至2026-06-21，本卡支持启动新一代信息技术产业链研究队列，但建议精选而非泛化看多。优先深挖芯原股份、澜起科技、瑞芯微、乐鑫科技和龙芯中科：前两者更直接受益于 AI 基础设施与内存带宽，瑞芯微和乐鑫科技代表端侧 AI/AIoT，龙芯中科是高战略弹性但仍亏损的国产 CPU 观察项。主要风险是存储和先进制程成本、出口管制摇摆、以及同一“国产 AI 链”…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.5, "matchedTerms": [ "新一代信息技术产业链" ], "href": "" }, { "date": "2026-06-12", "title": "新一代信息技术链的超额收益已从泛算力概念转向‘物理硬约束（电力/连接）组件出海’与‘底层芯片（GPU/MCU/MEMS）国产替代业绩拐点验证’，需警惕2026-06-13所处周期的科创50调…", "summary": "在2026-06-13评估中，针对新一代信息技术产业链4.3%的极低激活率，本报告系统梳理了5个低覆盖但具有强逻辑的A股/科创板标的。重点分析了摩尔线程（688795.SH）的GPU国产替代与首次扭亏、金盘科技（688676.SH）的电力物理瓶颈与出海溢价、鼎通科技（688668.SH）的高速连接器与液冷核心件、中颖电子（300327.SZ）的家电MCU拐…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 8, "matchedTerms": [ "新一代信息技术产业链" ], "href": "" }, { "date": "2026-06-11", "title": "指派 tmt-analyst 负责新一代信息技术产业链", "summary": "指派 tmt-analyst 负责新一代信息技术产业链 针对新一代信息技术产业链（激活率 3.3%）进行覆盖。依据 §6.1。", "sourceType": "board", "sourceLabel": "研究线程", "supportType": "neutral", "supportLabel": "观察", "score": 6.5, "matchedTerms": [ "新一代信息技术产业链" ], "href": "" }, { "date": "2026-06-10", "title": "在云端AI面临电力并网物理瓶颈及高两融去杠杆压力下，精选低拥挤度、高壁垒、进入全球核心供应链或处于国产替代关键爆发期的A股半导体上游特种材料与精密零部件细分龙头，是攻守兼备的最佳配置路径。", "summary": "新一代信息技术产业链在过去30天激活率仅为3.3%，呈现极高研究盲区。在云端AI面临北美电网容量并网延迟及推理解密通缩的双重物理瓶颈下，算力开支预期发生边际调整。这促使资金从高拥挤度的云端AI硬件向更具本土确定性、低拥挤度、高替代刚性的半导体特种材料与零部件细分龙头（联瑞新材、鼎龙股份、雅克科技与富创精密）转移。同时需防范AI服务器CapEx下修引发的电子…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 7.75, "matchedTerms": [ "新一代信息技术产业链" ], "href": "" }, { "date": "2026-06-08", "title": "云端AI面临电力并网瓶颈，A股TMT板块杠杆面临去杠杆挤压，资金应系统性避开拥挤硬件、超配联瑞新材、鼎龙股份、雅克科技与富创精密等具备高增长刚性与低拥挤度的上游半导体特种材料与精密零部件龙头。", "summary": "截至2026-06-09，新一代信息技术产业链激活率仅4.4%。云端AI基础设施面临电网容量与53个月并网等待物理瓶颈，北美CapEx面临延迟，引发资金向A股上游特种材料与精密部件龙头轮动。测试显示若电子板块融资发生25%去杠杆，将产生1416亿元卖压，流动性弱的恒生电子与高拥挤的金山办公将面临筹码挤压。建议深挖联瑞新材、鼎龙股份、雅克科技、富创精密等自主…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 7.75, "matchedTerms": [ "新一代信息技术产业链" ], "href": "" }, { "date": "2026-06-01", "title": "通过构建‘中游高端PCB确定性底仓+上游卡脖子材料（低Dk布、HVLP铜箔、特种树脂）国产替代期权’的杠铃型组合，能有效对冲产业链‘电力墙’与原材料成本挤压风险，平滑行业超额回报收益。", "summary": "截至2026-06-02，鉴于新一代信息技术产业链极低的30日激活率（4.4%，4/91），本研究建议建立‘A股算力辅材杠铃组合’应对中游硬瓶颈及上游材料紧缺。以生益电子（55%）为中游确定性底仓，辅以上游国产化材料期权（宏和科技15%、中国巨石15%、铜冠铜箔10%、圣泉集团5%）。组合加权预期净利润 CAGR 达 45.00%，能有效免疫上游原材料成本…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.75, "matchedTerms": [ "新一代信息技术产业链" ], "href": "" }, { "date": "2026-05-29", "title": "本轮新一代信息技术研究的 alpha 不在已被定价的龙头，而在第二梯队 WFE 子工具 + EDA + AI 液冷 + Chiplet IP 这 4 类二元 setup 标的；同时必须把 A…", "summary": "新一代信息技术产业链 30 日激活率仅 4.4%，覆盖被显性龙头吸走。我们锁定 5 个被低覆盖但 setup 干净的标的：拓荆科技 (688072.SH) ALD 营收 +192%、在手订单 110 亿，是 2H26 资本开支最干净受益人；华海清科 (688120.SH) 国产 12 英寸 CMP 份额 >90%，平台化期权未被定价；华大九天 (30126…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "neutral", "supportLabel": "观察", "score": 7.25, "matchedTerms": [ "新一代信息技术产业链" ], "href": "" }, { "date": "2026-05-27", "title": "在3D封装封锁与‘以量补效’红利驱动下，半导体材料、前驱体及精密设备零部件等低拥挤度细分龙头正迎来强烈的基本面反转与国产替代量增共振。", "summary": "截至2026-05-27，本卡完成了对新一代信息技术产业链覆盖盲区的首次调研。在先进封装及制裁收紧背景下，端侧AI面临138%的算力税压力，逼迫硬件提供商“以量补效”，进而释放出对封装材料、前驱体及精密设备零部件的刚性量增需求。我们挑选了联瑞新材、鼎龙股份、雅克科技和富创精密四家极具基本面弹性与核心技术壁垒的低拥挤度反转龙头进行建档与研究队列 Pitch。…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 7.5, "matchedTerms": [ "新一代信息技术产业链" ], "href": "" }, { "date": "2026-05-24", "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最清晰的战略相对输家；若只看语料中直接证据，Meta 是最清晰…", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" }, { "date": "2026-05-23", "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "date": "2026-05-23", "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" }, { "date": "2026-05-19", "title": "消费降级、银发经济与大健康结构性机会", "summary": "消费链条正在从总量弹性转向结构分化：银发消费、医疗健康、糖替代和 B 端餐饮渗透可能形成防御性需求，但需要验证支付能力和渠道真实动销。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "supports", "supportLabel": "支持", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" }, { "date": "2026-05-18", "title": "中国电力设备出海：AI 电网需求、贸易壁垒与利润率分化", "summary": "全球 AI 电网投资为中国电力设备提供需求窗口，但关税、认证、交付能力和原材料成本会决定哪些企业真正把订单转化为利润。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/china-power-equipment-export-policy-8678db7f/" }, { "date": "2026-05-18", "title": "AI 算力资本开支、电力瓶颈与通胀再定价", "summary": "AI 基础设施需求正在先通过电力、电网、变压器与材料供给形成成本脉冲，再通过生产率与架构效率形成中期缓释；投资结论取决于这两个方向谁先兑现。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/ai-power-inflation-transmission-0c9ba15c/" } ], "timeline": [ { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 30, "events": [ { "title": "中国电力设备出海：AI 电网需求、贸易壁垒与利润率分化", "summary": "全球 AI 电网投资为中国电力设备提供需求窗口，但关税、认证、交付能力和原材料成本会决定哪些企业真正把订单转化为利润。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/china-power-equipment-export-policy-8678db7f/" }, { "title": "AI 算力资本开支、电力瓶颈与通胀再定价", "summary": "AI 基础设施需求正在先通过电力、电网、变压器与材料供给形成成本脉冲，再通过生产率与架构效率形成中期缓释；投资结论取决于这两个方向谁先兑现。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-power-inflation-transmission-0c9ba15c/" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 0, "neutral": 0, "value": 51.4, "events": [ { "title": "消费降级、银发经济与大健康结构性机会", "summary": "消费链条正在从总量弹性转向结构分化：银发消费、医疗健康、糖替代和 B 端餐饮渗透可能形成防御性需求，但需要验证支付能力和渠道真实动销。", "sourceLabel": "深度研究", "supportLabel": "支持", "supportType": "supports", "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" } ] }, { "date": "2026-05-27", "supports": 0, "risks": 1, "neutral": 0, "value": 41.6, "events": [ { "title": "在3D封装封锁与‘以量补效’红利驱动下，半导体材料、前驱体及精密设备零部件等低拥挤度细分龙头正迎来强烈的基本面反转与国产替代量增共振。", "summary": "截至2026-05-27，本卡完成了对新一代信息技术产业链覆盖盲区的首次调研。在先进封装及制裁收紧背景下，端侧AI面临138%的算力税压力，逼迫硬件提供商“以量补效”，进而释放出对封装材料、前驱体及精密设备零部件的刚性量增需求。我们挑选了联瑞新材、鼎龙股份、雅克科技和富创精密四家极具基本面弹性与核…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-05-29", "supports": 0, "risks": 0, "neutral": 1, "value": 52, "events": [ { "title": "本轮新一代信息技术研究的 alpha 不在已被定价的龙头，而在第二梯队 WFE 子工具 + EDA + AI 液冷 + Chiplet IP 这 4 类二元 setup 标的；同时必…", "summary": "新一代信息技术产业链 30 日激活率仅 4.4%，覆盖被显性龙头吸走。我们锁定 5 个被低覆盖但 setup 干净的标的：拓荆科技 (688072.SH) ALD 营收 +192%、在手订单 110 亿，是 2H26 资本开支最干净受益人；华海清科 (688120.SH) 国产 12 英寸 CMP…", "sourceLabel": "研究笔记", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "通过构建‘中游高端PCB确定性底仓+上游卡脖子材料（低Dk布、HVLP铜箔、特种树脂）国产替代期权’的杠铃型组合，能有效对冲产业链‘电力墙’与原材料成本挤压风险，平滑行业超额回报收益。", "summary": "截至2026-06-02，鉴于新一代信息技术产业链极低的30日激活率（4.4%，4/91），本研究建议建立‘A股算力辅材杠铃组合’应对中游硬瓶颈及上游材料紧缺。以生益电子（55%）为中游确定性底仓，辅以上游国产化材料期权（宏和科技15%、中国巨石15%、铜冠铜箔10%、圣泉集团5%）。组合加权预期…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-08", "supports": 0, "risks": 1, "neutral": 0, "value": 45.8, "events": [ { "title": "云端AI面临电力并网瓶颈，A股TMT板块杠杆面临去杠杆挤压，资金应系统性避开拥挤硬件、超配联瑞新材、鼎龙股份、雅克科技与富创精密等具备高增长刚性与低拥挤度的上游半导体特种材料与精密零…", "summary": "截至2026-06-09，新一代信息技术产业链激活率仅4.4%。云端AI基础设施面临电网容量与53个月并网等待物理瓶颈，北美CapEx面临延迟，引发资金向A股上游特种材料与精密部件龙头轮动。测试显示若电子板块融资发生25%去杠杆，将产生1416亿元卖压，流动性弱的恒生电子与高拥挤的金山办公将面临筹…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-10", "supports": 0, "risks": 1, "neutral": 0, "value": 47.2, "events": [ { "title": "在云端AI面临电力并网物理瓶颈及高两融去杠杆压力下，精选低拥挤度、高壁垒、进入全球核心供应链或处于国产替代关键爆发期的A股半导体上游特种材料与精密零部件细分龙头，是攻守兼备的最佳配置…", "summary": "新一代信息技术产业链在过去30天激活率仅为3.3%，呈现极高研究盲区。在云端AI面临北美电网容量并网延迟及推理解密通缩的双重物理瓶颈下，算力开支预期发生边际调整。这促使资金从高拥挤度的云端AI硬件向更具本土确定性、低拥挤度、高替代刚性的半导体特种材料与零部件细分龙头（联瑞新材、鼎龙股份、雅克科技与…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-11", "supports": 0, "risks": 0, "neutral": 1, "value": 57.6, "events": [ { "title": "指派 tmt-analyst 负责新一代信息技术产业链", "summary": "指派 tmt-analyst 负责新一代信息技术产业链 针对新一代信息技术产业链（激活率 3.3%）进行覆盖。依据 §6.1。", "sourceLabel": "研究线程", "supportLabel": "观察", "supportType": "neutral", "href": "" } ] }, { "date": "2026-06-12", "supports": 0, "risks": 1, "neutral": 0, "value": 50, "events": [ { "title": "新一代信息技术链的超额收益已从泛算力概念转向‘物理硬约束（电力/连接）组件出海’与‘底层芯片（GPU/MCU/MEMS）国产替代业绩拐点验证’，需警惕2026-06-13所处周期的科…", "summary": "在2026-06-13评估中，针对新一代信息技术产业链4.3%的极低激活率，本报告系统梳理了5个低覆盖但具有强逻辑的A股/科创板标的。重点分析了摩尔线程（688795.SH）的GPU国产替代与首次扭亏、金盘科技（688676.SH）的电力物理瓶颈与出海溢价、鼎通科技（688668.SH）的高速连接…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-21", "supports": 0, "risks": 1, "neutral": 0, "value": 51.4, "events": [ { "title": "新一代信息技术链条的最佳机会在 AI 基础设施“卖铲人”和端侧 AI 芯片，而不是拥挤的通用 AI 服务器敞口。", "summary": "截至2026-06-21，本卡支持启动新一代信息技术产业链研究队列，但建议精选而非泛化看多。优先深挖芯原股份、澜起科技、瑞芯微、乐鑫科技和龙芯中科：前两者更直接受益于 AI 基础设施与内存带宽，瑞芯微和乐鑫科技代表端侧 AI/AIoT，龙芯中科是高战略弹性但仍亏损的国产 CPU 观察项。主要风险是…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-22", "supports": 0, "risks": 1, "neutral": 0, "value": 52.8, "events": [ { "title": "在2026年中美云厂商CapEx受阻于电力红线以及大模型推理通缩的双重催化下，新一代信息技术产业链估值正向能耗比占优的定制ASIC上游填料（联瑞新材）以及享受API接口成本大坍缩的软…", "summary": "截至2026-06-22，新一代信息技术产业链正经历算力瓶颈由GPU转向电力/并网及先进封装、以及DeepSeek API降价96.37%引发软件应用毛利重构的剧烈调整。板块激活率处于4.3%极低水平。本报告筛选了联瑞新材、生益电子、宏和科技、金山办公和富创精密等5家低两融、抛压出清且具刚性催化剂…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-23", "supports": 0, "risks": 1, "neutral": 0, "value": 54.2, "events": [ { "title": "随着AI芯片向高速与高功耗演进，硬件瓶颈已向高速铜连、芯片电感和硅光封测迁移，精选利基中盘标的具备更强超额收益潜力。", "summary": "本报告对新一代信息技术产业链进行调研，识别出华丰科技、鼎通科技、铂科新材及罗博特科等 4 家高增长但覆盖不足的标的，分析了其在 224G 铜连接、高端芯片电感与硅光封测设备中的核心壁垒，并揭示了冷却液监管与电网容量等跨链外溢风险。", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-7b551fae18/", "networkHref": "/investor/research-network/#thesis-7b551fae18" }, { "id": "thesis-9953a5d24c", "source": "auto_topic", "title": "工业机械：出口订单质量与资本开支弹性复盘", "titleEn": "Practical research topic: 工业机械：出口订单质量与资本开支弹性复盘", "lane": "AI 基础设施", "laneEn": "AI 基础设施", "laneHref": "/investor/lanes/ai/", "investableFocus": "", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "ARM", "name": "Arm Holdings", "market": "US", "role": "端侧 CPU IP", "roleEn": "端侧 CPU IP", "bucket": "core", "href": "/investor/stocks/ARM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ", "TLT", "META", "MAR", "ARM", "APO", "IEF" ], "coreView": "该主线来自 14 条研究记录的标题、问题和摘要聚类，时间跨度 2026-05-21 至 2026-06-06；用于观察是否形成可配置主题、风险预算和证伪条件，链条词仅作为证据标签。", "direction": "conflicting", "directionLabel": "分歧", "conviction": 31, "eventCount": 10, "firstSeen": "2026-05-18", "lastSeen": "2026-06-06", "supports": 1, "risks": 9, "neutral": 0, "analystCount": 3, "sourceTypeCounts": { "深度研究": 6, "研究笔记": 4 }, "topTerms": [ "工业机械产业链", "研究活跃度与投资线索复盘" ], "whatChanged": [ "2026-06-06: 白板研究卡片 03 人工智能与工业自动化对电力需求的冲击 作者 : 能源行业分析师 (energy analyst) 报告日期 : 2026 06 06 (亚洲/新加坡) 白研究线程标识：已归档卡片索引 : 3…", "2026-06-05: 工业机械投资逻辑已从“规模红利”转向“十五五”规划下的“全栈国产化”能力溢价，重点关注半导体上游材料与具身智能精密组件。", "2026-06-01: 上游材料瓶颈（特别是取向硅钢和高性能磁材）将使工业机械产业链从‘量增’转为‘价涨’逻辑，垂直整合能力成为核心估值溢价项。", "2026-06-01: 工业机械产业链的真正机会不在大盘股 beta，而在被低关注的供应节点——高精度减速器、液压/直线执行器、数据中心机电承包商、SMR/国防隐性受益者；但能否兑现取决于上游材料瓶颈（取向硅钢、钕铁硼、铜、钛），需立即量化。", "2026-05-24: 下一代周期中，Mag7 谁会输？" ], "practical": { "sectorLane": "AI 基础设施", "sectorLaneEn": "AI 基础设施", "exposureBucket": "AI 基础设施", "tone": "blue", "actionCode": "radar", "actionLabel": "主题雷达：先建观察仓", "actionTone": "blue", "riskBudget": "观察仓：0-1% thesis risk", "portfolioRole": "主题观察", "weightGuide": "0-1% thesis risk", "timeHorizon": "1-4 周", "portfolioUse": "这是自动提取主题，先用于发现研究活跃度，不直接等同投资结论。", "positionSizing": "建议按 0-1% thesis risk 管理，不用单条 thesis 替代组合层面的行业、因子和流动性约束。", "stockUniverse": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" }, { "ticker": "SMH", "name": "VanEck Semiconductor ETF", "market": "US ETF", "role": "半导体篮子", "roleEn": "半导体篮子", "bucket": "watch", "href": "/investor/stocks/SMH/" }, { "ticker": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "roleEn": "光模块/CPO", "bucket": "core", "href": "/investor/stocks/300308.SZ/" }, { "ticker": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "roleEn": "长端利率代理", "bucket": "core", "href": "/investor/stocks/TLT/" }, { "ticker": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告效率", "roleEn": "开源模型与广告效率", "bucket": "core", "href": "/investor/stocks/META/" }, { "ticker": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "roleEn": "全球酒店 RevPAR 对照", "bucket": "peer", "href": "/investor/stocks/MAR/" }, { "ticker": "ARM", "name": "Arm Holdings", "market": "US", "role": "端侧 CPU IP", "roleEn": "端侧 CPU IP", "bucket": "core", "href": "/investor/stocks/ARM/" }, { "ticker": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "roleEn": "保险资产与私募信贷", "bucket": "core", "href": "/investor/stocks/APO/" }, { "ticker": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "roleEn": "中长久期代理", "bucket": "core", "href": "/investor/stocks/IEF/" } ], "stockTickers": [ "NVDA", "TSM", "AVGO", "MU", "ANET", "VRT", "SMH", "300308.SZ", "TLT", "META", "MAR", "ARM", "APO", "IEF" ], "primaryStocks": [ { "ticker": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "roleEn": "GPU 利用率核心", "bucket": "core", "href": "/investor/stocks/NVDA/" }, { "ticker": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "roleEn": "能效制程核心", "bucket": "core", "href": "/investor/stocks/TSM/" }, { "ticker": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "roleEn": "交换/ASIC/光互联", "bucket": "core", "href": "/investor/stocks/AVGO/" }, { "ticker": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "roleEn": "HBM/DRAM 供给", "bucket": "core", "href": "/investor/stocks/MU/" }, { "ticker": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "roleEn": "AI 网络效率", "bucket": "core", "href": "/investor/stocks/ANET/" }, { "ticker": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "roleEn": "电源与液冷基础设施", "bucket": "core", "href": "/investor/stocks/VRT/" } ], "instrumentTags": [ "NVDA", "NVIDIA", "TSM", "TSMC", "AVGO", "Broadcom", "MU", "Micron", "ANET", "Arista Networks", "VRT", "Vertiv" ], "reviewCadence": "每日复核", "confirmationSignals": [ "云厂商 capex 指引是否继续上修", "HBM/CPO/先进封装供给是否缓解", "单集群 ROI 是否覆盖折旧、电力和联网成本" ], "invalidationSignals": [ "风险事件继续高于支持事件，且最新证据没有修复。", "核心监控词热度上升，但可验证订单、价格或现金流没有同步改善。", "同一结论只来自单一来源，无法形成跨来源验证。" ], "pairedHedges": [ "利率久期", "电力成本", "软件利润池", "半导体周期" ], "dueDiligenceQuestions": [ "这条主线的收益来源是基本面、估值修复、资金流还是事件催化？", "当前证据是否覆盖订单、价格、现金流和仓位四个维度？", "如果未来两次更新没有新增支持证据，是否应该降级？" ], "deepResearchPriority": "常规跟踪", "deepResearchReason": "现阶段以日常跟踪和复盘更新为主。", "evidenceBreadth": { "total": 10, "sourceDiversity": 2, "analystCount": 3, "recentEvents": 4, "supportRatio": 0.1, "riskRatio": 0.9, "conflictScore": 99, "evidenceDepth": 40.1 }, "score": 17.4 }, "events": [ { "date": "2026-06-06", "title": "白板研究卡片 03 人工智能与工业自动化对电力需求的冲击 作者 : 能源行业分析师 (energy analyst) 报告日期 : 2026 06 06 (亚洲/新加坡) 白研究线程标识：已…", "summary": "白板研究卡片 03 人工智能与工业自动化对电力需求的冲击 作者 : 能源行业分析师 (energy analyst) 报告日期 : 2026 06 06 (亚洲/新加坡) 白研究线程标识：已归档卡片索引 : 3/10 研究立场 : 压力测试 (Stress test — 对算力扩张下能源系统及电网供应链的脆弱性评估) 1. 执行摘要与立场声明 截至今日（ …", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [ "工业机械产业链" ], "href": "" }, { "date": "2026-06-05", "title": "工业机械投资逻辑已从“规模红利”转向“十五五”规划下的“全栈国产化”能力溢价，重点关注半导体上游材料与具身智能精密组件。", "summary": "2026年中国工业机械产业链已进入“能力建设”与“全栈国产化”的关键期。工信部设定10万亿营收目标，政策驱动半导体设备国产化率达50%。研究发现中船特气（WF6垄断）、苏大维格（先进封装）、秦川机床（人形机器人减速器）等低关注度标的具备强逻辑支撑。需警惕美国50%钢材关税导致的成本脱钩、成熟节点芯片短缺及Pax Silica地缘政治框架下的供应链分化风险。", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.25, "matchedTerms": [ "工业机械产业链" ], "href": "" }, { "date": "2026-06-01", "title": "工业机械产业链的真正机会不在大盘股 beta，而在被低关注的供应节点——高精度减速器、液压/直线执行器、数据中心机电承包商、SMR/国防隐性受益者；但能否兑现取决于上游材料瓶颈（取向硅钢、钕…", "summary": "工业机械产业链 30 日激活率仅 5.1%，但当前有三股资本开支周期同步发力——全球电网与数据中心、人形机器人 BOM、北约+中国国防超级周期——共同压向同一批被低估的供应节点（高精度减速器、液压直线执行器、取向硅钢限制的变压器、机电承包商）。本卡提名 5 个低关注、setup 较硬的标的：Harmonic Drive Systems（6324.T）、恒立…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.75, "matchedTerms": [ "工业机械产业链" ], "href": "" }, { "date": "2026-06-01", "title": "上游材料瓶颈（特别是取向硅钢和高性能磁材）将使工业机械产业链从‘量增’转为‘价涨’逻辑，垂直整合能力成为核心估值溢价项。", "summary": "本报告对取向硅钢 (GOES)、钕铁硼、铜和钛进行了压力测试。结果显示，GOES 和钕铁硼面临结构性赤字，变压器交付周期拉长至 144 周，钕镨价格年内飙升 138%。人型机器人的强力吸货（4kg/台）正在挤压 EV 供应链。航空级钛材受认证周期限制，民航复苏对国防和 SMR 产生了显著的产能挤出。工业机械节点的‘产能兑现’将高度依赖于供应链的垂直整合与原…", "sourceType": "card", "sourceLabel": "研究笔记", "supportType": "risk", "supportLabel": "风险", "score": 6.75, "matchedTerms": [ "工业机械产业链" ], "href": "" }, { "date": "2026-05-24", "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最清晰的战略相对输家；若只看语料中直接证据，Meta 是最清晰…", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" }, { "date": "2026-05-23", "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "date": "2026-05-23", "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" }, { "date": "2026-05-19", "title": "消费降级、银发经济与大健康结构性机会", "summary": "消费链条正在从总量弹性转向结构分化：银发消费、医疗健康、糖替代和 B 端餐饮渗透可能形成防御性需求，但需要验证支付能力和渠道真实动销。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "supports", "supportLabel": "支持", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" }, { "date": "2026-05-18", "title": "中国电力设备出海：AI 电网需求、贸易壁垒与利润率分化", "summary": "全球 AI 电网投资为中国电力设备提供需求窗口，但关税、认证、交付能力和原材料成本会决定哪些企业真正把订单转化为利润。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/china-power-equipment-export-policy-8678db7f/" }, { "date": "2026-05-18", "title": "AI 算力资本开支、电力瓶颈与通胀再定价", "summary": "AI 基础设施需求正在先通过电力、电网、变压器与材料供给形成成本脉冲，再通过生产率与架构效率形成中期缓释；投资结论取决于这两个方向谁先兑现。", "sourceType": "report", "sourceLabel": "深度研究", "supportType": "risk", "supportLabel": "风险", "score": 7, "matchedTerms": [], "href": "/investor/deep-research/ai-power-inflation-transmission-0c9ba15c/" } ], "timeline": [ { "date": "2026-05-18", "supports": 0, "risks": 2, "neutral": 0, "value": 30, "events": [ { "title": "中国电力设备出海：AI 电网需求、贸易壁垒与利润率分化", "summary": "全球 AI 电网投资为中国电力设备提供需求窗口，但关税、认证、交付能力和原材料成本会决定哪些企业真正把订单转化为利润。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/china-power-equipment-export-policy-8678db7f/" }, { "title": "AI 算力资本开支、电力瓶颈与通胀再定价", "summary": "AI 基础设施需求正在先通过电力、电网、变压器与材料供给形成成本脉冲，再通过生产率与架构效率形成中期缓释；投资结论取决于这两个方向谁先兑现。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-power-inflation-transmission-0c9ba15c/" } ] }, { "date": "2026-05-19", "supports": 1, "risks": 0, "neutral": 0, "value": 51.4, "events": [ { "title": "消费降级、银发经济与大健康结构性机会", "summary": "消费链条正在从总量弹性转向结构分化：银发消费、医疗健康、糖替代和 B 端餐饮渗透可能形成防御性需求，但需要验证支付能力和渠道真实动销。", "sourceLabel": "深度研究", "supportLabel": "支持", "supportType": "supports", "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" } ] }, { "date": "2026-05-23", "supports": 0, "risks": 2, "neutral": 0, "value": 32.8, "events": [ { "title": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "summary": "AI 的去通胀叙事需要真实采用率、流程再设计和可计量产出改进共同兑现；在此之前，资本开支、能源与人才成本可能先表现为再通胀。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "title": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "summary": "AI 电力交易的核心不只是发电量，而是高压变压器、取向硅钢、铜铝与开关设备的交付约束；这些约束决定设备商利润率、客户交期与终端算力释放速度。", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" } ] }, { "date": "2026-05-24", "supports": 0, "risks": 1, "neutral": 0, "value": 40.2, "events": [ { "title": "下一代周期中，Mag7 谁会输？", "summary": "AI Institute 语料指向的输家不是简单的股价下跌名单，而是三类失败：没有被本轮 AI 基础设施、能效、推理商业化和电力约束直接验证的战略相对输家；有清晰资本开支、折旧、SBC 与自由现金流压力的利润表输家；以及仍能增长但估值从稀缺性重定价为资本强度的倍数输家。若必须给出一个名字，苹果是最…", "sourceLabel": "深度研究", "supportLabel": "风险", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" } ] }, { "date": "2026-06-01", "supports": 0, "risks": 2, "neutral": 0, "value": 35.6, "events": [ { "title": "工业机械产业链的真正机会不在大盘股 beta，而在被低关注的供应节点——高精度减速器、液压/直线执行器、数据中心机电承包商、SMR/国防隐性受益者；但能否兑现取决于上游材料瓶颈（取向…", "summary": "工业机械产业链 30 日激活率仅 5.1%，但当前有三股资本开支周期同步发力——全球电网与数据中心、人形机器人 BOM、北约+中国国防超级周期——共同压向同一批被低估的供应节点（高精度减速器、液压直线执行器、取向硅钢限制的变压器、机电承包商）。本卡提名 5 个低关注、setup 较硬的标的：Har…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" }, { "title": "上游材料瓶颈（特别是取向硅钢和高性能磁材）将使工业机械产业链从‘量增’转为‘价涨’逻辑，垂直整合能力成为核心估值溢价项。", "summary": "本报告对取向硅钢 (GOES)、钕铁硼、铜和钛进行了压力测试。结果显示，GOES 和钕铁硼面临结构性赤字，变压器交付周期拉长至 144 周，钕镨价格年内飙升 138%。人型机器人的强力吸货（4kg/台）正在挤压 EV 供应链。航空级钛材受认证周期限制，民航复苏对国防和 SMR 产生了显著的产能挤出…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-05", "supports": 0, "risks": 1, "neutral": 0, "value": 43, "events": [ { "title": "工业机械投资逻辑已从“规模红利”转向“十五五”规划下的“全栈国产化”能力溢价，重点关注半导体上游材料与具身智能精密组件。", "summary": "2026年中国工业机械产业链已进入“能力建设”与“全栈国产化”的关键期。工信部设定10万亿营收目标，政策驱动半导体设备国产化率达50%。研究发现中船特气（WF6垄断）、苏大维格（先进封装）、秦川机床（人形机器人减速器）等低关注度标的具备强逻辑支撑。需警惕美国50%钢材关税导致的成本脱钩、成熟节点芯…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] }, { "date": "2026-06-06", "supports": 0, "risks": 1, "neutral": 0, "value": 44.4, "events": [ { "title": "白板研究卡片 03 人工智能与工业自动化对电力需求的冲击 作者 : 能源行业分析师 (energy analyst) 报告日期 : 2026 06 06 (亚洲/新加坡) 白研究线程…", "summary": "白板研究卡片 03 人工智能与工业自动化对电力需求的冲击 作者 : 能源行业分析师 (energy analyst) 报告日期 : 2026 06 06 (亚洲/新加坡) 白研究线程标识：已归档卡片索引 : 3/10 研究立场 : 压力测试 (Stress test — 对算力扩张下能源系统及电网…", "sourceLabel": "研究笔记", "supportLabel": "风险", "supportType": "risk", "href": "" } ] } ], "href": "/investor/theses/thesis-9953a5d24c/", "networkHref": "/investor/research-network/#thesis-9953a5d24c" } ], "graph": { "nodes": [ { "id": "thesis-029ce03da1", "type": "thesis", "label": "MSCI 调整窗口是被动资金与主动减仓相互博弈的流动性事件", "conviction": 90, "direction": "conflicting", "lane": "指数调仓/被动资金事件", "laneHref": "/investor/lanes/index-rebalance-and-passive-flow-events/", "bucket": "中国权益与流动性", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 82.2, "href": "/investor/theses/thesis-029ce03da1/", "networkHref": "/investor/research-network/#thesis-029ce03da1" }, { "id": "stock-510de57b5f", "type": "stock", "label": "MCHI", "name": "iShares MSCI China ETF", "market": "US ETF", "role": "离岸中国风险", "href": "/investor/stocks/MCHI/" }, { "id": "stock-68a288585c", "type": "stock", "label": "KWEB", "name": "KraneShares CSI China Internet ETF", "market": "US ETF", "role": "中概互联网高 beta", "href": "/investor/stocks/KWEB/" }, { "id": "stock-79f7c9313f", "type": "stock", "label": "FXI", "name": "iShares China Large-Cap ETF", "market": "US ETF", "role": "离岸中国大盘代理", "href": "/investor/stocks/FXI/" }, { "id": "stock-f57a081cf3", "type": "stock", "label": "2800.HK", "name": "盈富基金", "market": "HK ETF", "role": "港股核心指数代理", "href": "/investor/stocks/2800.HK/" }, { "id": "stock-dcd4dbe710", "type": "stock", "label": "510300.SH", "name": "沪深300 ETF", "market": "A-share ETF", "role": "核心指数风险偏好", "href": "/investor/stocks/510300.SH/" }, { "id": "stock-efe2ecdc5b", "type": "stock", "label": "0700.HK", "name": "腾讯控股", "market": "HK", "role": "港股权重与风险偏好", "href": "/investor/stocks/0700.HK/" }, { "id": "term-ed9ab89f22", "type": "term", "label": "HBM", "href": "/investor/research-network/?q=HBM" }, { "id": "term-ea49523d74", "type": "term", "label": "GPU", "href": "/investor/research-network/?q=GPU" }, { "id": "term-f71ccd3834", "type": "term", "label": "A股", "href": "/investor/research-network/?q=A%E8%82%A1" }, { "id": "term-b8332514e0", "type": "term", "label": "AIDC", "href": "/investor/research-network/?q=AIDC" }, { "id": "term-a83d000110", "type": "term", "label": "MSCI", "href": "/investor/research-network/?q=MSCI" }, { "id": "term-87faf34a67", "type": "term", "label": "rebalance", "href": "/investor/research-network/?q=rebalance" }, { "id": "term-340d600392", "type": "term", "label": "windows", "href": "/investor/research-network/?q=windows" }, { "id": "term-59b47e9ec5", "type": "term", "label": "liquidity", "href": "/investor/research-network/?q=liquidity" }, { "id": "event-65122927a93c", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-0dd67e976c54", "type": "event", "label": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ec6fc35ebaf2", "type": "event", "label": "九卡线索数值连贯、引用完整、叙事贯通，分析扎实无需重做；仅一项 P0（Cards 01/02 缺 2026-06-30 工作日期锚）卡住发布，评级 N…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-067f4b063a10", "type": "event", "label": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-93399852c17e", "type": "event", "label": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c29701b05aa0", "type": "event", "label": "在先进封装良率瓶颈与表后电力并网延迟的双重物理约束下，A股AI硬件板块估值极度偏高且存在变现拥挤，难以独立承担成长杠铃的对冲配置功能，应当实施防御性减…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-58532cf290f7", "type": "event", "label": "面向AI基础设施信用尾部的跨资产组合配置与对冲", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-978bd9fef78b", "type": "event", "label": "国产大模型并未告别规模竞赛，而是在先进GPU受限的条件下把竞争单位改写为每瓦、每token、每受限加速器和每留存用户的有效智能产出。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-8c5a9690a0", "type": "thesis", "label": "Mag7 在 AI 时代从平台垄断转向基础设施再分层", "conviction": 88, "direction": "conflicting", "lane": "Mag7 AI 平台再分层", "laneHref": "/investor/lanes/mag7-ai-platform-re-layering/", "bucket": "AI 基础设施", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 84.4, "href": "/investor/theses/thesis-8c5a9690a0/", "networkHref": "/investor/research-network/#thesis-8c5a9690a0" }, { "id": "stock-8638f0279b", "type": "stock", "label": "MSFT", "name": "Microsoft", "market": "US", "role": "旧软件入口与云 AI", "href": "/investor/stocks/MSFT/" }, { "id": "stock-1eb44d6252", "type": "stock", "label": "AAPL", "name": "Apple", "market": "US", "role": "端侧 AI 与消费硬件", "href": "/investor/stocks/AAPL/" }, { "id": "stock-2ff6985a97", "type": "stock", "label": "TSLA", "name": "Tesla", "market": "US", "role": "自动驾驶/机器人与高波动叙事", "href": "/investor/stocks/TSLA/" }, { "id": "stock-2d1edf729c", "type": "stock", "label": "META", "name": "Meta Platforms", "market": "US", "role": "开源模型与广告平台", "href": "/investor/stocks/META/" }, { "id": "stock-31ac18dc22", "type": "stock", "label": "AMZN", "name": "Amazon", "market": "US", "role": "AWS/自研芯片/零售自动化", "href": "/investor/stocks/AMZN/" }, { "id": "stock-a25fc665a4", "type": "stock", "label": "GOOGL", "name": "Alphabet", "market": "US", "role": "模型、搜索、云与 TPU", "href": "/investor/stocks/GOOGL/" }, { "id": "term-44568e2e5f", "type": "term", "label": "AI 基础设施", "href": "/investor/research-network/?q=AI%20%E5%9F%BA%E7%A1%80%E8%AE%BE%E6%96%BD" }, { "id": "term-c7bac46904", "type": "term", "label": "Microsoft", "href": "/investor/research-network/?q=Microsoft" }, { "id": "term-0890f45668", "type": "term", "label": "NVIDIA", "href": "/investor/research-network/?q=NVIDIA" }, { "id": "term-49d2dc07ac", "type": "term", "label": "Amazon", "href": "/investor/research-network/?q=Amazon" }, { "id": "term-f223faa96f", "type": "term", "label": "Apple", "href": "/investor/research-network/?q=Apple" }, { "id": "term-4f749de7c2", "type": "term", "label": "Meta", "href": "/investor/research-network/?q=Meta" }, { "id": "term-f89f85382f", "type": "term", "label": "Tesla", "href": "/investor/research-network/?q=Tesla" }, { "id": "event-1c162e6fab24", "type": "event", "label": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-b642c7a92c30", "type": "event", "label": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-66d0a71e1acb", "type": "event", "label": "AI交易不会结束，但估值核心应从买到多少芯片转向能否按期通电、保持利用率并降低每token能耗。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b29076871103", "type": "event", "label": "数据中心电网并网瓶颈与核能转向", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7d4f797675b9", "type": "event", "label": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastru…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-623b2987baef", "type": "event", "label": "变压器材料和交付瓶颈已开始改变 Hyperscaler 的2026-2027资本开支与电力采购节奏，影响主要体现为预算通胀、长期锁电和通电交付递延，而…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-99489c9c3778", "type": "event", "label": "电子链条利润反弹主要受AI基础设施投资拉动，具备向存储、设备、材料、先进封装、光互联和数据中心电源散热扩散的潜力，但尚不足以证明消费电子和内需全面复苏。", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6d6c9f82dd3d", "type": "event", "label": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-c89650af9e", "type": "thesis", "label": "酒店餐饮休闲复苏正在从客流修复转向 RevPAR、翻台率与小盘 Alpha", "conviction": 85, "direction": "conflicting", "lane": "酒店餐饮休闲/服务消费", "laneHref": "/investor/lanes/hotels-restaurants-leisure-and-services-consumption/", "bucket": "消费健康防御", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 78.7, "href": "/investor/theses/thesis-c89650af9e/", "networkHref": "/investor/research-network/#thesis-c89650af9e" }, { "id": "stock-083dc3a2b3", "type": "stock", "label": "1179.HK", "name": "华住集团-S", "market": "HK", "role": "酒店 RevPAR", "href": "/investor/stocks/1179.HK/" }, { "id": "stock-a6f4e763c1", "type": "stock", "label": "9961.HK", "name": "携程集团-S", "market": "HK", "role": "OTA 与旅游预订", "href": "/investor/stocks/9961.HK/" }, { "id": "stock-8e7522f842", "type": "stock", "label": "3690.HK", "name": "美团", "market": "HK", "role": "本地服务与消费弹性", "href": "/investor/stocks/3690.HK/" }, { "id": "stock-dda1241d64", "type": "stock", "label": "YUMC", "name": "Yum China", "market": "US", "role": "连锁餐饮", "href": "/investor/stocks/YUMC/" }, { "id": "stock-8904c61a98", "type": "stock", "label": "MAR", "name": "Marriott International", "market": "US", "role": "全球酒店 RevPAR 对照", "href": "/investor/stocks/MAR/" }, { "id": "stock-ae1510493b", "type": "stock", "label": "BKNG", "name": "Booking Holdings", "market": "US", "role": "全球 OTA 对照", "href": "/investor/stocks/BKNG/" }, { "id": "term-1e7a12e965", "type": "term", "label": "GOES", "href": "/investor/research-network/?q=GOES" }, { "id": "term-0ef384847c", "type": "term", "label": "CPI", "href": "/investor/research-network/?q=CPI" }, { "id": "term-a5e4d88f36", "type": "term", "label": "RevPAR", "href": "/investor/research-network/?q=RevPAR" }, { "id": "term-b1a96dd646", "type": "term", "label": "Alpha", "href": "/investor/research-network/?q=Alpha" }, { "id": "term-ae4dbd7e5a", "type": "term", "label": "Leisure", "href": "/investor/research-network/?q=Leisure" }, { "id": "term-5b6c989793", "type": "term", "label": "hotels", "href": "/investor/research-network/?q=hotels" }, { "id": "term-0dbdc4fb0f", "type": "term", "label": "restaurants", "href": "/investor/research-network/?q=restaurants" }, { "id": "term-6e862940d0", "type": "term", "label": "moving", "href": "/investor/research-network/?q=moving" }, { "id": "event-55aaf129b1a6", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d7b0c7f26ea4", "type": "event", "label": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8eaaa6ce81bb", "type": "event", "label": "[auto-handoff] 霍尔木兹海峡石油冲击与核心CPI传导", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9f1a40503848", "type": "event", "label": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8ce1fb804b20", "type": "event", "label": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7a6a011ea0cd", "type": "event", "label": "关键矿产与设备出口管制风险对A股高端环节溢价逻辑的冲击评估", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-acd322d6ba80", "type": "event", "label": "截至2026-06-28，最优跨资产表达是“久期+稀缺股票”的杠铃，铜只做长周期期权，因为近端精炼过剩削弱了直接商品敞口的性价比，而GOES稀缺主要只…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8115c4f62184", "type": "event", "label": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-f5d7a79c82", "type": "thesis", "label": "HBM4、CoWoS 与背面供电正在成为 AI 服务器需求曲线的串行闸门", "conviction": 84, "direction": "conflicting", "lane": "先进封装/HBM4 闸门", "laneHref": "/investor/lanes/advanced-packaging-and-hbm4-gate/", "bucket": "AI 基础设施", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 76.9, "href": "/investor/theses/thesis-f5d7a79c82/", "networkHref": "/investor/research-network/#thesis-f5d7a79c82" }, { "id": "stock-cb15b056a5", "type": "stock", "label": "TSM", "name": "TSMC", "market": "US ADR", "role": "能效制程核心", "href": "/investor/stocks/TSM/" }, { "id": "stock-426219f2bc", "type": "stock", "label": "NVDA", "name": "NVIDIA", "market": "US", "role": "GPU 利用率核心", "href": "/investor/stocks/NVDA/" }, { "id": "stock-a24c631b90", "type": "stock", "label": "MU", "name": "Micron", "market": "US", "role": "HBM/DRAM 供给", "href": "/investor/stocks/MU/" }, { "id": "stock-1bcfecaf36", "type": "stock", "label": "000660.KS", "name": "SK hynix", "market": "Korea", "role": "HBM 领先供给", "href": "/investor/stocks/000660.KS/" }, { "id": "stock-7599fc757f", "type": "stock", "label": "005930.KS", "name": "Samsung Electronics", "market": "Korea", "role": "HBM 与晶圆", "href": "/investor/stocks/005930.KS/" }, { "id": "stock-c908d13f30", "type": "stock", "label": "ASML", "name": "ASML", "market": "US ADR", "role": "先进制程设备", "href": "/investor/stocks/ASML/" }, { "id": "term-e4a3cbf326", "type": "term", "label": "AI 算力", "href": "/investor/research-network/?q=AI%20%E7%AE%97%E5%8A%9B" }, { "id": "term-4de0127946", "type": "term", "label": "先进封装", "href": "/investor/research-network/?q=%E5%85%88%E8%BF%9B%E5%B0%81%E8%A3%85" }, { "id": "term-36e3041f1c", "type": "term", "label": "ASIC", "href": "/investor/research-network/?q=ASIC" }, { "id": "term-00fb5022b7", "type": "term", "label": "TSMC", "href": "/investor/research-network/?q=TSMC" }, { "id": "event-b7d4d00f15be", "type": "event", "label": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-369daa798bef", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1b0710c91896", "type": "event", "label": "先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6be609154936", "type": "event", "label": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-2a65b011d40a", "type": "event", "label": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b7d81759dfae", "type": "event", "label": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-83ccc2c57dfe", "type": "event", "label": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-116e157e4170", "type": "event", "label": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-8bbbcdf5a4", "type": "thesis", "label": "GPU、HBM 与先进封装决定 AI 资本开支供给节奏", "conviction": 84, "direction": "conflicting", "lane": "AI 算力半导体", "laneHref": "/investor/lanes/ai-compute-semiconductors/", "bucket": "AI 基础设施", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 76.6, "href": "/investor/theses/thesis-8bbbcdf5a4/", "networkHref": "/investor/research-network/#thesis-8bbbcdf5a4" }, { "id": "stock-0fd3e4e71b", "type": "stock", "label": "AMD", "name": "AMD", "market": "US", "role": "GPU/CPU 替代供给", "href": "/investor/stocks/AMD/" }, { "id": "stock-b78cca2401", "type": "stock", "label": "AVGO", "name": "Broadcom", "market": "US", "role": "交换/ASIC/光互联", "href": "/investor/stocks/AVGO/" }, { "id": "term-d49897f8c4", "type": "term", "label": "云厂商", "href": "/investor/research-network/?q=%E4%BA%91%E5%8E%82%E5%95%86" }, { "id": "event-f3345c652a8b", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ee3eae4dc86c", "type": "event", "label": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-14cafc321f85", "type": "event", "label": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-bed6c0f6a1d4", "type": "event", "label": "国产HBM与先进封装不是零进展，但2026年的主线应买“工艺环节国产化”，而不是把AI芯片设计公司当作不受HBM约束的无限TAM。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7d585e0aca75", "type": "event", "label": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8e6780d40d6d", "type": "event", "label": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-86639eda6024", "type": "event", "label": "半导体产业链的供给侧约束与国产化替代的技术边界", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-aede264ef7c6", "type": "event", "label": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "thesis-ab1fb6a453", "type": "thesis", "label": "DeepSeek 式低价模型正在重塑 AI 定价权与软件利润池", "conviction": 83, "direction": "conflicting", "lane": "低价模型/AI 软件利润池", "laneHref": "/investor/lanes/low-cost-models-and-ai-software-profit-pools/", "bucket": "AI 基础设施", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 79.6, "href": "/investor/theses/thesis-ab1fb6a453/", "networkHref": "/investor/research-network/#thesis-ab1fb6a453" }, { "id": "stock-434e501ece", "type": "stock", "label": "BABA", "name": "Alibaba", "market": "US ADR", "role": "通义/云与中国 MaaS", "href": "/investor/stocks/BABA/" }, { "id": "stock-9cd0d546e6", "type": "stock", "label": "9988.HK", "name": "阿里巴巴", "market": "HK", "role": "中国云与模型价格", "href": "/investor/stocks/9988.HK/" }, { "id": "stock-23cfa8bf26", "type": "stock", "label": "BIDU", "name": "Baidu", "market": "US ADR", "role": "文心/搜索与云", "href": "/investor/stocks/BIDU/" }, { "id": "stock-294da558b6", "type": "stock", "label": "9888.HK", "name": "百度集团", "market": "HK", "role": "中国模型与应用", "href": "/investor/stocks/9888.HK/" }, { "id": "term-0648d2111e", "type": "term", "label": "DeepSeek", "href": "/investor/research-network/?q=DeepSeek" }, { "id": "term-409d071901", "type": "term", "label": "企业", "href": "/investor/research-network/?q=%E4%BC%81%E4%B8%9A" }, { "id": "term-f9b7ef8159", "type": "term", "label": "DeepSeek-style", "href": "/investor/research-network/?q=DeepSeek-style" }, { "id": "term-a8a79412bd", "type": "term", "label": "low-cost", "href": "/investor/research-network/?q=low-cost" }, { "id": "event-77234e3d1d38", "type": "event", "label": "AI 应用的 ROI 验证：从 Copilot 订阅到企业级代理", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-84ce3f44a9b5", "type": "event", "label": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-76ad8cbebc3a", "type": "event", "label": "DeepSeek降价96.37%是Jevons放大还是AI硬件需求毁灭——与“电网/变压器是新瓶颈”主线对撞", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4120371c3241", "type": "event", "label": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-po…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d88aea428955", "type": "event", "label": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-0871091194c6", "type": "event", "label": "国产大模型并未告别规模竞赛，而是在先进GPU受限的条件下把竞争单位改写为每瓦、每token、每受限加速器和每留存用户的有效智能产出。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c1656bad07ad", "type": "event", "label": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-6bbca16040ea", "type": "event", "label": "数据中心电网并网瓶颈与核能转向", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-6f42e0fe6e", "type": "thesis", "label": "稀土、铜铝与关键材料正在成为 AI 与新能源共用的供给约束", "conviction": 81, "direction": "conflicting", "lane": "关键材料/稀土铜铝", "laneHref": "/investor/lanes/critical-materials-rare-earths-copper-and-aluminum/", "bucket": "能源与关键材料", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 74.3, "href": "/investor/theses/thesis-6f42e0fe6e/", "networkHref": "/investor/research-network/#thesis-6f42e0fe6e" }, { "id": "stock-0d55e94f77", "type": "stock", "label": "600111.SH", "name": "北方稀土", "market": "A-share", "role": "稀土资源", "href": "/investor/stocks/600111.SH/" }, { "id": "stock-e87348f2ea", "type": "stock", "label": "000831.SZ", "name": "中国稀土", "market": "A-share", "role": "稀土整合", "href": "/investor/stocks/000831.SZ/" }, { "id": "stock-204ac66a88", "type": "stock", "label": "MP", "name": "MP Materials", "market": "US", "role": "美国稀土资源", "href": "/investor/stocks/MP/" }, { "id": "stock-e5218dcd47", "type": "stock", "label": "FCX", "name": "Freeport-McMoRan", "market": "US", "role": "铜矿", "href": "/investor/stocks/FCX/" }, { "id": "stock-a8f5b84c58", "type": "stock", "label": "SCCO", "name": "Southern Copper", "market": "US", "role": "铜矿", "href": "/investor/stocks/SCCO/" }, { "id": "stock-58bb119c35", "type": "stock", "label": "AA", "name": "Alcoa", "market": "US", "role": "铝", "href": "/investor/stocks/AA/" }, { "id": "term-66e0f52800", "type": "term", "label": "新能源", "href": "/investor/research-network/?q=%E6%96%B0%E8%83%BD%E6%BA%90" }, { "id": "term-9196f0338a", "type": "term", "label": "能源", "href": "/investor/research-network/?q=%E8%83%BD%E6%BA%90" }, { "id": "term-58fdb0941f", "type": "term", "label": "商品", "href": "/investor/research-network/?q=%E5%95%86%E5%93%81" }, { "id": "term-1232085ef0", "type": "term", "label": "钢铁", "href": "/investor/research-network/?q=%E9%92%A2%E9%93%81" }, { "id": "term-291d2a9756", "type": "term", "label": "Rare", "href": "/investor/research-network/?q=Rare" }, { "id": "event-28ff19bfc7b1", "type": "event", "label": "[auto-handoff] 先进制程星载 OBP 芯片良率瓶颈与算力/通信设备材料供应链", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-22753c05c82d", "type": "event", "label": "中游房建及家装开工收缩-30.67%通过高经营杠杆深度挤压了上游钢铁、铝加工及基础化工的营业利润率与订单量，导致三大行业面临深幅利润下修压力（加权下修…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-0782375243df", "type": "event", "label": "能源与大宗商品资产在宏观政策背离下的防御价值研判", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-83c2735e99f8", "type": "event", "label": "在高利率长期化但衰退未确认的环境下，应以套息、凸性与实物资产对冲（短久期杠铃、适度美元、能源期权+黄金+能源股）替代方向性利率与油价押注，以改善组合风…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b4345abe9c75", "type": "event", "label": "能源/商品板块应以多元化商品篮子为核心、能源股为收益锚、少量原油为尾部风险工具，而不是把11%战略通胀对冲变成单一布伦特方向押注。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d1ed91148429", "type": "event", "label": "能源商品成本传导与USD/CNY汇率风险的压力测试", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1cd714441919", "type": "event", "label": "虽然备忘录缓解了短期实物原油压力，但霍尔木兹海峡的结构性脆弱与15%-25%的二次通胀概率支撑将大宗商品战术超配至20%，并应采用高分红个股现货替代及…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c19fce5ea092", "type": "event", "label": "霍尔木兹海峡重开仅为“执行折价”而非“和平红利”，60天物理排雷延迟导致到岸能源价具极强向下刚性，与高利率共同构筑新能源“资本开支壁垒”，系统性阻碍全…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-70daf87262", "type": "thesis", "label": "A 股红利与硬科技杠铃仍是政策、利率和风险偏好的组合表达", "conviction": 81, "direction": "conflicting", "lane": "A股红利/硬科技杠铃", "laneHref": "/investor/lanes/a-share-dividend-and-hard-tech-barbell/", "bucket": "中国权益与流动性", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 74.8, "href": "/investor/theses/thesis-70daf87262/", "networkHref": "/investor/research-network/#thesis-70daf87262" }, { "id": "stock-57a896f1c0", "type": "stock", "label": "510880.SH", "name": "红利 ETF", "market": "A-share ETF", "role": "高股息指数代理", "href": "/investor/stocks/510880.SH/" }, { "id": "stock-b9b86bd6b7", "type": "stock", "label": "601398.SH", "name": "工商银行", "market": "A-share", "role": "银行红利", "href": "/investor/stocks/601398.SH/" }, { "id": "stock-0d1a838e74", "type": "stock", "label": "600900.SH", "name": "长江电力", "market": "A-share", "role": "公用事业红利", "href": "/investor/stocks/600900.SH/" }, { "id": "stock-f5c007a81f", "type": "stock", "label": "601088.SH", "name": "中国神华", "market": "A-share", "role": "煤炭红利", "href": "/investor/stocks/601088.SH/" }, { "id": "stock-532787880f", "type": "stock", "label": "600941.SH", "name": "中国移动", "market": "A-share", "role": "运营商红利", "href": "/investor/stocks/600941.SH/" }, { "id": "stock-c39cc2da75", "type": "stock", "label": "588000.SH", "name": "科创50 ETF", "market": "A-share ETF", "role": "硬科技风险偏好", "href": "/investor/stocks/588000.SH/" }, { "id": "term-60e981724c", "type": "term", "label": "科创50", "href": "/investor/research-network/?q=%E7%A7%91%E5%88%9B50" }, { "id": "term-a2e9329b92", "type": "term", "label": "融资余额", "href": "/investor/research-network/?q=%E8%9E%8D%E8%B5%84%E4%BD%99%E9%A2%9D" }, { "id": "term-1582c27783", "type": "term", "label": "长江电力", "href": "/investor/research-network/?q=%E9%95%BF%E6%B1%9F%E7%94%B5%E5%8A%9B" }, { "id": "term-a050178590", "type": "term", "label": "公用事业", "href": "/investor/research-network/?q=%E5%85%AC%E7%94%A8%E4%BA%8B%E4%B8%9A" }, { "id": "term-18126b597c", "type": "term", "label": "中芯国际", "href": "/investor/research-network/?q=%E4%B8%AD%E8%8A%AF%E5%9B%BD%E9%99%85" }, { "id": "term-e21dcec424", "type": "term", "label": "运营商", "href": "/investor/research-network/?q=%E8%BF%90%E8%90%A5%E5%95%86" }, { "id": "event-93176bb8ed58", "type": "event", "label": "红利龙头去拥挤与超跌成长板块反弹的风格切换大论战", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-00b21e510b58", "type": "event", "label": "[auto-handoff] 高拥挤大行红利资产减仓与资金流向评估", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-6593e7d60e5a", "type": "event", "label": "A股风格大轮动：大盘拥挤红利瓦解与成长股超跌反弹的持续性辩论", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-8ccb66ee3b20", "type": "event", "label": "拥挤红利去杠杆与医药生物估值脱钩反弹的风格切换", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-79548bf69166", "type": "event", "label": "局部中小银行信用缺口下的A股银行红利耐久性与高股息内部轮动", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8af7fff0ec70", "type": "event", "label": "债市收益率触底倒逼绝对收益资金流向A股优质高股息红利股建立‘红利锚’，在‘数量先行、价格被锚’的结构性宽松下，配置策略应向国有大行与电力设备出口链的对…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-c77f3d5fb20f", "type": "event", "label": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-12d12096696b", "type": "event", "label": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-58200390b4", "type": "thesis", "label": "加密资产仍是全球流动性和散户风险偏好的高频回声", "conviction": 81, "direction": "conflicting", "lane": "加密/高波动风险偏好", "laneHref": "/investor/lanes/crypto-and-high-beta-risk-appetite/", "bucket": "风险偏好与高波动", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 73.4, "href": "/investor/theses/thesis-58200390b4/", "networkHref": "/investor/research-network/#thesis-58200390b4" }, { "id": "stock-daf49f5a34", "type": "stock", "label": "COIN", "name": "Coinbase", "market": "US", "role": "加密交易基础设施", "href": "/investor/stocks/COIN/" }, { "id": "stock-339675eab4", "type": "stock", "label": "MSTR", "name": "MicroStrategy", "market": "US", "role": "BTC 高 beta 代理", "href": "/investor/stocks/MSTR/" }, { "id": "stock-edafde8d2d", "type": "stock", "label": "MARA", "name": "MARA Holdings", "market": "US", "role": "比特币矿企", "href": "/investor/stocks/MARA/" }, { "id": "stock-d1303b4837", "type": "stock", "label": "RIOT", "name": "Riot Platforms", "market": "US", "role": "比特币矿企", "href": "/investor/stocks/RIOT/" }, { "id": "stock-8185c648c8", "type": "stock", "label": "IBIT", "name": "iShares Bitcoin Trust", "market": "US ETF", "role": "BTC 现货代理", "href": "/investor/stocks/IBIT/" }, { "id": "stock-a0cd4ecb4f", "type": "stock", "label": "HOOD", "name": "Robinhood", "market": "US", "role": "散户交易风险偏好", "href": "/investor/stocks/HOOD/" }, { "id": "term-df12b8f89b", "type": "term", "label": "Crypto", "href": "/investor/research-network/?q=Crypto" }, { "id": "term-d341073240", "type": "term", "label": "remains", "href": "/investor/research-network/?q=remains" }, { "id": "term-f81d83a742", "type": "term", "label": "high-frequency", "href": "/investor/research-network/?q=high-frequency" }, { "id": "term-092c79e8f8", "type": "term", "label": "echo", "href": "/investor/research-network/?q=echo" }, { "id": "term-8001c27439", "type": "term", "label": "global", "href": "/investor/research-network/?q=global" }, { "id": "event-42ac0cf447d6", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-0281e7ed0133", "type": "event", "label": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-po…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-bdb6f3864771", "type": "event", "label": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c0284e20728f", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4faeda0d18cb", "type": "event", "label": "效率驱动的算力架构转型：从“暴力规模”到“高能效密度”的物理基座适配性评估", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-364c99289c3a", "type": "event", "label": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3d2f033276ec", "type": "event", "label": "数据中心电网并网瓶颈与核能转向", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b53e365a948c", "type": "event", "label": "<20%开机率红线是平台属性而非物理定律：对电力受限EO小卫星它由电源+被动散热共同锁定且先进热控不划算(支持卡03)，但两相泵驱回路+可展开辐射器已…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-8b8ede3135", "type": "thesis", "label": "消费降级、健康刚需与银发经济正在重塑防御型增长", "conviction": 80, "direction": "conflicting", "lane": "消费降级/健康银发", "laneHref": "/investor/lanes/consumer-downgrade-health-and-silver-economy/", "bucket": "消费健康防御", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 76.9, "href": "/investor/theses/thesis-8b8ede3135/", "networkHref": "/investor/research-network/#thesis-8b8ede3135" }, { "id": "stock-0716b17bb6", "type": "stock", "label": "600519.SH", "name": "贵州茅台", "market": "A-share", "role": "外资/核心资产代理", "href": "/investor/stocks/600519.SH/" }, { "id": "stock-736f8c20de", "type": "stock", "label": "2319.HK", "name": "蒙牛乳业", "market": "HK", "role": "必选消费", "href": "/investor/stocks/2319.HK/" }, { "id": "stock-6ece0f6e09", "type": "stock", "label": "300760.SZ", "name": "迈瑞医疗", "market": "A-share", "role": "医疗设备", "href": "/investor/stocks/300760.SZ/" }, { "id": "stock-bb47079d51", "type": "stock", "label": "600276.SH", "name": "恒瑞医药", "market": "A-share", "role": "创新药出海核心", "href": "/investor/stocks/600276.SH/" }, { "id": "stock-2897e29810", "type": "stock", "label": "1177.HK", "name": "中国生物制药", "market": "HK", "role": "医药消费与支付", "href": "/investor/stocks/1177.HK/" }, { "id": "term-5dbd754de7", "type": "term", "label": "创新药", "href": "/investor/research-network/?q=%E5%88%9B%E6%96%B0%E8%8D%AF" }, { "id": "term-6840b7e095", "type": "term", "label": "食品", "href": "/investor/research-network/?q=%E9%A3%9F%E5%93%81" }, { "id": "term-e1506406a5", "type": "term", "label": "设备", "href": "/investor/research-network/?q=%E8%AE%BE%E5%A4%87" }, { "id": "term-3fdb185870", "type": "term", "label": "Consumer", "href": "/investor/research-network/?q=Consumer" }, { "id": "term-f76e03f51e", "type": "term", "label": "downgrade", "href": "/investor/research-network/?q=downgrade" }, { "id": "term-3d8049711c", "type": "term", "label": "healthcare", "href": "/investor/research-network/?q=healthcare" }, { "id": "term-c045a6043b", "type": "term", "label": "necessities", "href": "/investor/research-network/?q=necessities" }, { "id": "event-1596942fae70", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-effa6da6fa25", "type": "event", "label": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-35d8f55fd722", "type": "event", "label": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d74dcb35422f", "type": "event", "label": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-5a60ea7790a6", "type": "event", "label": "去风险三期医药资产通过rNPV模型释放3.48亿美元的期权价值构成Biotech估值底，而生命科学工具在5.5x EV/Sales筑底、CXO在 ov…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-fff74715140e", "type": "event", "label": "大健康低覆盖机会更像跨子行业alpha篮子，而不是单一医疗板块beta；优先研究能把2025-2026催化转化为收入桥的五家公司。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ef44a97e835b", "type": "event", "label": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c5cdafe69e8f", "type": "event", "label": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-e9630d3ee1", "type": "thesis", "label": "日元套息平仓是 EM 科技、离岸中资与高 beta 主题的尾部流动性风险", "conviction": 80, "direction": "conflicting", "lane": "日元套息/EM 科技流动性", "laneHref": "/investor/lanes/yen-carry-and-em-tech-liquidity/", "bucket": "风险偏好与高波动", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 73.6, "href": "/investor/theses/thesis-e9630d3ee1/", "networkHref": "/investor/research-network/#thesis-e9630d3ee1" }, { "id": "stock-5829704445", "type": "stock", "label": "FXY", "name": "Invesco CurrencyShares Japanese Yen Trust", "market": "US ETF", "role": "日元代理", "href": "/investor/stocks/FXY/" }, { "id": "stock-d1c8471480", "type": "stock", "label": "UUP", "name": "Invesco DB US Dollar Index Bullish Fund", "market": "US ETF", "role": "美元融资压力", "href": "/investor/stocks/UUP/" }, { "id": "stock-175b4b336d", "type": "stock", "label": "EEM", "name": "iShares MSCI Emerging Markets ETF", "market": "US ETF", "role": "EM 风险资产", "href": "/investor/stocks/EEM/" }, { "id": "stock-0d33a0a8c4", "type": "stock", "label": "EWJ", "name": "iShares MSCI Japan ETF", "market": "US ETF", "role": "日本权益对照", "href": "/investor/stocks/EWJ/" }, { "id": "term-dcdd3b5c45", "type": "term", "label": "美元", "href": "/investor/research-network/?q=%E7%BE%8E%E5%85%83" }, { "id": "term-273ab83269", "type": "term", "label": "CNH", "href": "/investor/research-network/?q=CNH" }, { "id": "event-d9bffeb07095", "type": "event", "label": "日元套利平仓通过离岸人民币流动性抽水及掉期清算挤压双重渠道，系统性抬高离岸融资成本并技术性击穿USD/CNY 7.00关口，导致人民币套利空间彻底崩溃。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7147cc0ea9c9", "type": "event", "label": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-47a290f02d2b", "type": "event", "label": "中美利差倒挂对离岸人民币（CNH）汇率及跨境资本流动的传导效应", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9f09a88af28a", "type": "event", "label": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-77a96ced3bf7", "type": "event", "label": "基准配置应优先通过久期表达去通胀与政策失误风险，权益只做选择性低贝塔，商品低配整体贝塔但保留油价重上100美元的显式尾部对冲。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-078cbab4f900", "type": "event", "label": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9d798c2c6a07", "type": "event", "label": "日元套利交易平仓与离岸CNH流动性收紧对A股防御重估的交叉传导风险", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9d988b58d2e0", "type": "event", "label": "美元多头利差交易极度拥挤且极易受日央行干预与实际利率波动影响而无序平仓，USD/JPY跌破150会引发2150亿至2850亿美元美债清算并给在岸高两融…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-89c14497ac", "type": "thesis", "label": "期权伽马、ETF 与尾盘流动性正在放大主题交易的微观结构风险", "conviction": 79, "direction": "conflicting", "lane": "期权伽马/市场微观结构", "laneHref": "/investor/lanes/options-gamma-and-market-microstructure/", "bucket": "风险偏好与高波动", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 72.4, "href": "/investor/theses/thesis-89c14497ac/", "networkHref": "/investor/research-network/#thesis-89c14497ac" }, { "id": "stock-9b19451158", "type": "stock", "label": "CBOE", "name": "Cboe Global Markets", "market": "US", "role": "期权与波动率交易所", "href": "/investor/stocks/CBOE/" }, { "id": "stock-e0671bd326", "type": "stock", "label": "CME", "name": "CME Group", "market": "US", "role": "期货与清算", "href": "/investor/stocks/CME/" }, { "id": "stock-b19fdc727e", "type": "stock", "label": "VIRT", "name": "Virtu Financial", "market": "US", "role": "做市与流动性", "href": "/investor/stocks/VIRT/" }, { "id": "stock-89b531787b", "type": "stock", "label": "IBKR", "name": "Interactive Brokers", "market": "US", "role": "交易基础设施", "href": "/investor/stocks/IBKR/" }, { "id": "stock-a436689d6c", "type": "stock", "label": "VIXY", "name": "ProShares VIX Short-Term Futures ETF", "market": "US ETF", "role": "波动率工具", "href": "/investor/stocks/VIXY/" }, { "id": "term-d0db8b5e36", "type": "term", "label": "Options", "href": "/investor/research-network/?q=Options" }, { "id": "term-be9d587def", "type": "term", "label": "gamma", "href": "/investor/research-network/?q=gamma" }, { "id": "term-c6d29ec0e7", "type": "term", "label": "ETFs", "href": "/investor/research-network/?q=ETFs" }, { "id": "term-6f9a48800d", "type": "term", "label": "closing", "href": "/investor/research-network/?q=closing" }, { "id": "term-f7f1ee30cf", "type": "term", "label": "amplifying", "href": "/investor/research-network/?q=amplifying" }, { "id": "term-39157f7c0e", "type": "term", "label": "microstructure", "href": "/investor/research-network/?q=microstructure" }, { "id": "term-2c6ef0f0d0", "type": "term", "label": "risk", "href": "/investor/research-network/?q=risk" }, { "id": "event-aa56a14a7e77", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8a37171cce92", "type": "event", "label": "[auto-handoff] Impact of the JGB 2.8% Carry Trade Repatriation Trigger on M…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-daae54b39c91", "type": "event", "label": "[auto-handoff] Impact of global capacity pricing spikes on domestic transfo…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-891fb6137aa9", "type": "event", "label": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-250a6d3054ae", "type": "event", "label": "[auto-handoff] Cloud customer acceptance of higher memory prices", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-d54724c5a15b", "type": "event", "label": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c1cab84c94e3", "type": "event", "label": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-73915616f883", "type": "event", "label": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-075017003b", "type": "thesis", "label": "AI 资本开支通过能源、设备和利率进入通胀定价", "conviction": 79, "direction": "conflicting", "lane": "AI 再通胀/利率敏感资产", "laneHref": "/investor/lanes/ai-reflation-and-rate-sensitive-assets/", "bucket": "宏观利率与汇率", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 69.8, "href": "/investor/theses/thesis-075017003b/", "networkHref": "/investor/research-network/#thesis-075017003b" }, { "id": "stock-426177902e", "type": "stock", "label": "XLE", "name": "Energy Select Sector SPDR", "market": "US ETF", "role": "能源通胀代理", "href": "/investor/stocks/XLE/" }, { "id": "stock-2888611640", "type": "stock", "label": "XLU", "name": "Utilities Select Sector SPDR", "market": "US ETF", "role": "公用事业利率敏感", "href": "/investor/stocks/XLU/" }, { "id": "stock-06214b33ef", "type": "stock", "label": "TLT", "name": "iShares 20+ Year Treasury Bond ETF", "market": "US ETF", "role": "长端利率代理", "href": "/investor/stocks/TLT/" }, { "id": "stock-0ca8634002", "type": "stock", "label": "GLD", "name": "SPDR Gold Shares", "market": "US ETF", "role": "实际利率反向代理", "href": "/investor/stocks/GLD/" }, { "id": "stock-b64a63240f", "type": "stock", "label": "JPM", "name": "JPMorgan Chase", "market": "US", "role": "利率与信用周期", "href": "/investor/stocks/JPM/" }, { "id": "stock-1eaff1ec01", "type": "stock", "label": "NEE", "name": "NextEra Energy", "market": "US", "role": "电力资产与利率敏感", "href": "/investor/stocks/NEE/" }, { "id": "term-95feb4759b", "type": "term", "label": "电网", "href": "/investor/research-network/?q=%E7%94%B5%E7%BD%91" }, { "id": "term-1ae05a2701", "type": "term", "label": "PCE", "href": "/investor/research-network/?q=PCE" }, { "id": "term-04f7acfcd2", "type": "term", "label": "数据中心", "href": "/investor/research-network/?q=%E6%95%B0%E6%8D%AE%E4%B8%AD%E5%BF%83" }, { "id": "term-7192e079e7", "type": "term", "label": "AI capex", "href": "/investor/research-network/?q=AI%20capex" }, { "id": "term-5e71d16a67", "type": "term", "label": "AI数据中心", "href": "/investor/research-network/?q=AI%E6%95%B0%E6%8D%AE%E4%B8%AD%E5%BF%83" }, { "id": "event-9ebabd45f3e5", "type": "event", "label": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-450482f7b4c2", "type": "event", "label": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6eb15f14e56c", "type": "event", "label": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-44b8692d661f", "type": "event", "label": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c76fd0cb4df1", "type": "event", "label": "2026年全球AI数据中心电网瓶颈与“自主供电”战略的崛起", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-89b4b4fc3e69", "type": "event", "label": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e5bf58cdad10", "type": "event", "label": "6月点阵图不是简单过时，而是可能“对得不是理由”：能源理由过时，AI/电网与折旧通道支持更高更久，但不足以单独证明立即加息。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-62029a01e8c0", "type": "event", "label": "随着全球及中国变电设备产能在2027-2029年逐步释放，AI数据中心的电力设备瓶颈将从“无法获取设备”的二进制物理硬约束，降为“支付溢价与关税以获取…", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-42faeae623", "type": "thesis", "label": "CPO、光互联与铜连接决定 AI 集群内部租金迁移", "conviction": 78, "direction": "conflicting", "lane": "CPO/光模块/高速互联", "laneHref": "/investor/lanes/cpo-optical-modules-and-high-speed-interconnect/", "bucket": "AI 基础设施", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 70.8, "href": "/investor/theses/thesis-42faeae623/", "networkHref": "/investor/research-network/#thesis-42faeae623" }, { "id": "stock-b93441048d", "type": "stock", "label": "MRVL", "name": "Marvell", "market": "US", "role": "高速互联芯片", "href": "/investor/stocks/MRVL/" }, { "id": "stock-6b03c0b5e7", "type": "stock", "label": "ANET", "name": "Arista Networks", "market": "US", "role": "AI 网络效率", "href": "/investor/stocks/ANET/" }, { "id": "stock-6f93ca0ebd", "type": "stock", "label": "COHR", "name": "Coherent", "market": "US", "role": "光器件与材料", "href": "/investor/stocks/COHR/" }, { "id": "stock-264924258d", "type": "stock", "label": "LITE", "name": "Lumentum", "market": "US", "role": "光器件", "href": "/investor/stocks/LITE/" }, { "id": "stock-b2f82c3028", "type": "stock", "label": "300308.SZ", "name": "中际旭创", "market": "A-share", "role": "光模块/CPO", "href": "/investor/stocks/300308.SZ/" }, { "id": "term-a4b8abb418", "type": "term", "label": "CPO", "href": "/investor/research-network/?q=CPO" }, { "id": "event-41cf97e7a08b", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-2e97f577dbd6", "type": "event", "label": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7ecd24840b4f", "type": "event", "label": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-2938d08d3ff0", "type": "event", "label": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-80376edf9d5a", "type": "event", "label": "双轨AI网络可以落地，但硬件约束从电力单点转向电力、CoWoS/HBM、交换ASIC和光互联的同步供给，只有能跨堆栈预锁定产能的运营商能率先规模化。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7f38e9639830", "type": "event", "label": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3a192bc9dbe9", "type": "event", "label": "在2026-06-27的权威工作日背景下，国产HBM良率骤降逼迫国内算力网络化，而全球封装与电力阻尼拉长了云厂商CapEx变现周期，这使得制程免疫且订…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ed6d612db37d", "type": "event", "label": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-9d3836cc07", "type": "thesis", "label": "AI 生产率去通胀需要等待采用率和流程重构验证", "conviction": 78, "direction": "conflicting", "lane": "AI 软件效率/生产率", "laneHref": "/investor/lanes/ai-software-productivity/", "bucket": "宏观利率与汇率", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 72.4, "href": "/investor/theses/thesis-9d3836cc07/", "networkHref": "/investor/research-network/#thesis-9d3836cc07" }, { "id": "stock-97bfc0d991", "type": "stock", "label": "NOW", "name": "ServiceNow", "market": "US", "role": "企业流程自动化", "href": "/investor/stocks/NOW/" }, { "id": "stock-130f70ae5b", "type": "stock", "label": "CRM", "name": "Salesforce", "market": "US", "role": "CRM 工作流与 Agent", "href": "/investor/stocks/CRM/" }, { "id": "term-0e2b23eeed", "type": "term", "label": "PPA", "href": "/investor/research-network/?q=PPA" }, { "id": "event-9ee02a1f9f67", "type": "event", "label": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-530d80f63cb4", "type": "event", "label": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-fe97c369b30a", "type": "event", "label": "AI交易不会结束，但估值核心应从买到多少芯片转向能否按期通电、保持利用率并降低每token能耗。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-96ba214e9d5b", "type": "event", "label": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-2f7dbcbb01f5", "type": "event", "label": "数据中心电网并网瓶颈与核能转向", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7a656643581e", "type": "event", "label": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastru…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-44d8838be0b7", "type": "event", "label": "变压器材料和交付瓶颈已开始改变 Hyperscaler 的2026-2027资本开支与电力采购节奏，影响主要体现为预算通胀、长期锁电和通电交付递延，而…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-013ef198b05b", "type": "event", "label": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-fd31cc177b", "type": "thesis", "label": "债券供给、地方债与久期需求决定利率下行空间", "conviction": 78, "direction": "conflicting", "lane": "债券供给/银行久期", "laneHref": "/investor/lanes/bond-supply-and-bank-duration/", "bucket": "信用、地产与久期", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 76.7, "href": "/investor/theses/thesis-fd31cc177b/", "networkHref": "/investor/research-network/#thesis-fd31cc177b" }, { "id": "stock-4c3e22d6f2", "type": "stock", "label": "IEF", "name": "iShares 7-10 Year Treasury Bond ETF", "market": "US ETF", "role": "中长久期代理", "href": "/investor/stocks/IEF/" }, { "id": "stock-8f4b41e6fb", "type": "stock", "label": "601939.SH", "name": "建设银行", "market": "A-share", "role": "银行配置盘", "href": "/investor/stocks/601939.SH/" }, { "id": "stock-f86096cddd", "type": "stock", "label": "601288.SH", "name": "农业银行", "market": "A-share", "role": "银行配置盘", "href": "/investor/stocks/601288.SH/" }, { "id": "stock-de7f5dc2fb", "type": "stock", "label": "000001.SZ", "name": "平安银行", "market": "A-share", "role": "股份行信用敏感", "href": "/investor/stocks/000001.SZ/" }, { "id": "term-d8705755e6", "type": "term", "label": "CGB", "href": "/investor/research-network/?q=CGB" }, { "id": "term-920e88c5a6", "type": "term", "label": "银行", "href": "/investor/research-network/?q=%E9%93%B6%E8%A1%8C" }, { "id": "term-8f8fd9ecdd", "type": "term", "label": "Bond", "href": "/investor/research-network/?q=Bond" }, { "id": "term-f30dfe1cfb", "type": "term", "label": "supply", "href": "/investor/research-network/?q=supply" }, { "id": "term-bc0e0206bf", "type": "term", "label": "local-government", "href": "/investor/research-network/?q=local-government" }, { "id": "term-652d75f9ba", "type": "term", "label": "debt", "href": "/investor/research-network/?q=debt" }, { "id": "term-caa79af4db", "type": "term", "label": "duration", "href": "/investor/research-network/?q=duration" }, { "id": "term-dffc177fd0", "type": "term", "label": "demand", "href": "/investor/research-network/?q=demand" }, { "id": "event-b869b3a04c52", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1963584fe028", "type": "event", "label": "[auto-handoff] 国内OMO宽松与美债/中债倒挂对资产配置影响", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8c48c1786f1b", "type": "event", "label": "[auto-handoff] LGFV去杠杆下商业银行资产配置转移评估", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-7d4f684962a6", "type": "event", "label": "[auto-handoff] 推导 PBOC 隔夜逆回购实际中标利率及中债收益率曲线非线性久期损失风险", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-cd5338c153c6", "type": "event", "label": "I have completed the research for Research note 01 of this research thread …", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c975db8641a5", "type": "event", "label": "[auto-handoff] Cloud customer acceptance of higher memory prices", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-27c2cfd12c5c", "type": "event", "label": "I have conducted a comprehensive stress test of the CGB 10Y floor and yield…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-0086e682ecc0", "type": "event", "label": "I will read the session brief and the reports from the previous card (Resea…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "thesis-51c6d833ae", "type": "thesis", "label": "食品加工的机会从食品通胀切换到原料波动、渠道库存与 B 端标准化", "conviction": 78, "direction": "conflicting", "lane": "食品加工/成本传导", "laneHref": "/investor/lanes/food-processing-and-input-cost-transmission/", "bucket": "宏观利率与汇率", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 75.7, "href": "/investor/theses/thesis-51c6d833ae/", "networkHref": "/investor/research-network/#thesis-51c6d833ae" }, { "id": "stock-81aa4e9efc", "type": "stock", "label": "001215.SZ", "name": "千味央厨", "market": "A-share", "role": "B 端餐饮标准化", "href": "/investor/stocks/001215.SZ/" }, { "id": "stock-7ad1c1f434", "type": "stock", "label": "603317.SH", "name": "天味食品", "market": "A-share", "role": "复合调味料", "href": "/investor/stocks/603317.SH/" }, { "id": "stock-0f57e807aa", "type": "stock", "label": "603027.SH", "name": "千禾味业", "market": "A-share", "role": "调味品渠道库存", "href": "/investor/stocks/603027.SH/" }, { "id": "stock-b1b000a34d", "type": "stock", "label": "600298.SH", "name": "安琪酵母", "market": "A-share", "role": "食品配料与出口", "href": "/investor/stocks/600298.SH/" }, { "id": "stock-81c80ed154", "type": "stock", "label": "002847.SZ", "name": "盐津铺子", "market": "A-share", "role": "休闲食品效率", "href": "/investor/stocks/002847.SZ/" }, { "id": "stock-bb3ad25ff3", "type": "stock", "label": "ADM", "name": "Archer-Daniels-Midland", "market": "US", "role": "全球农产品成本基准", "href": "/investor/stocks/ADM/" }, { "id": "term-e4eb18069d", "type": "term", "label": "Food", "href": "/investor/research-network/?q=Food" }, { "id": "term-0a63dd9aa0", "type": "term", "label": "processing", "href": "/investor/research-network/?q=processing" }, { "id": "term-9be34f3e54", "type": "term", "label": "opportunities", "href": "/investor/research-network/?q=opportunities" }, { "id": "term-aeadfe2770", "type": "term", "label": "shifting", "href": "/investor/research-network/?q=shifting" }, { "id": "event-cf58e2b83678", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-af7b47443cea", "type": "event", "label": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-aa0ed92b3610", "type": "event", "label": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d54ab9ee6a25", "type": "event", "label": "[auto-handoff] 霍尔木兹海峡石油冲击与核心CPI传导", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-5d4c8ac5368c", "type": "event", "label": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-95e1268e9582", "type": "event", "label": "食品加工板块β偏弱但覆盖缺口蕴含丰厚Alpha，应重点布局中报Setup干净的千味央厨与受新添加剂国标催化的B端生物配料龙头，回避无定价权的中小盘。", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-37f4c250e01f", "type": "event", "label": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a8609216670f", "type": "event", "label": "宏观需求疲弱足以削弱但不足以完全否定食品加工链的成本通缩红利，配置应从板块β降级为精选大众必选品阿尔法。", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-ffa474a3f2", "type": "thesis", "label": "端侧 AI 与 Agent 桌面正在把算力竞争从云端扩展到芯片、OS 与消费电子供应链", "conviction": 78, "direction": "conflicting", "lane": "端侧 AI/Agent 桌面", "laneHref": "/investor/lanes/edge-ai-and-agent-desktop-cycle/", "bucket": "消费健康防御", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 70.3, "href": "/investor/theses/thesis-ffa474a3f2/", "networkHref": "/investor/research-network/#thesis-ffa474a3f2" }, { "id": "stock-bfceb2656b", "type": "stock", "label": "QCOM", "name": "Qualcomm", "market": "US", "role": "手机/PC NPU", "href": "/investor/stocks/QCOM/" }, { "id": "stock-316dddc5a4", "type": "stock", "label": "ARM", "name": "Arm Holdings", "market": "US", "role": "端侧 CPU IP", "href": "/investor/stocks/ARM/" }, { "id": "stock-c7e6a87908", "type": "stock", "label": "002475.SZ", "name": "立讯精密", "market": "A-share", "role": "消费电子制造", "href": "/investor/stocks/002475.SZ/" }, { "id": "stock-058d9f45ed", "type": "stock", "label": "002241.SZ", "name": "歌尔股份", "market": "A-share", "role": "声学/可穿戴", "href": "/investor/stocks/002241.SZ/" }, { "id": "term-7834502344", "type": "term", "label": "CoWoS", "href": "/investor/research-network/?q=CoWoS" }, { "id": "event-056dc7223eb9", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3f3c9657fa51", "type": "event", "label": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-220e66b8057b", "type": "event", "label": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b09ee4e60002", "type": "event", "label": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-98aa5c086bd6", "type": "event", "label": "半导体国产替代对AI算力瓶颈的缓解潜力", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-51f53b92535d", "type": "event", "label": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-31edb092c269", "type": "event", "label": "半导体产业链的供给侧约束与国产化替代的技术边界", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8168fcadc6f6", "type": "event", "label": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-3d421524cd", "type": "thesis", "label": "油价地缘底线决定再通胀与降息交易的安全边际", "conviction": 78, "direction": "conflicting", "lane": "油气/地缘通胀", "laneHref": "/investor/lanes/oil-gas-and-geopolitical-inflation/", "bucket": "宏观利率与汇率", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 74.3, "href": "/investor/theses/thesis-3d421524cd/", "networkHref": "/investor/research-network/#thesis-3d421524cd" }, { "id": "stock-b8b420a4c8", "type": "stock", "label": "XOM", "name": "Exxon Mobil", "market": "US", "role": "全球综合油气", "href": "/investor/stocks/XOM/" }, { "id": "stock-cc1f4a0444", "type": "stock", "label": "CVX", "name": "Chevron", "market": "US", "role": "全球综合油气", "href": "/investor/stocks/CVX/" }, { "id": "stock-51a3f330a2", "type": "stock", "label": "COP", "name": "ConocoPhillips", "market": "US", "role": "上游油气", "href": "/investor/stocks/COP/" }, { "id": "stock-59bde72b61", "type": "stock", "label": "SLB", "name": "SLB", "market": "US", "role": "油服与资本开支", "href": "/investor/stocks/SLB/" }, { "id": "stock-31491c7f11", "type": "stock", "label": "OXY", "name": "Occidental Petroleum", "market": "US", "role": "美国页岩油", "href": "/investor/stocks/OXY/" }, { "id": "stock-35264bffcb", "type": "stock", "label": "0883.HK", "name": "中国海洋石油", "market": "HK", "role": "中国海上油气", "href": "/investor/stocks/0883.HK/" }, { "id": "term-e27e330702", "type": "term", "label": "原油", "href": "/investor/research-network/?q=%E5%8E%9F%E6%B2%B9" }, { "id": "term-1414c1fd5e", "type": "term", "label": "Brent", "href": "/investor/research-network/?q=Brent" }, { "id": "term-2b95a99fe5", "type": "term", "label": "石油", "href": "/investor/research-network/?q=%E7%9F%B3%E6%B2%B9" }, { "id": "term-0da46a7fbf", "type": "term", "label": "油价", "href": "/investor/research-network/?q=%E6%B2%B9%E4%BB%B7" }, { "id": "term-5d6a4410cd", "type": "term", "label": "geopolitical", "href": "/investor/research-network/?q=geopolitical" }, { "id": "event-ab71480c0c0a", "type": "event", "label": "[auto-handoff] 低油价对中国工业企业利润及进口贸易顺差的边际影响分析", "date": "2026-06-30", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-7f7c43326484", "type": "event", "label": "2026年石油供应中断对全球能源品种及炼油产能的冲击评估", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-d7175f9a3e20", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-70731137dfa4", "type": "event", "label": "美伊卡塔尔复谈促油价地缘溢价退潮与跨资产红利再平衡", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-f864cc4682be", "type": "event", "label": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ae0a91e0bab7", "type": "event", "label": "能源商品成本传导与USD/CNY汇率风险的压力测试", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-78d3221ea78a", "type": "event", "label": "核心服务与住房通胀的粘性：美联储政策的后续驱动力", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-2ba91a1d3f04", "type": "event", "label": "在风险平价框架下，能源减配并转向黄金/结构性金属不会突破波动约束，反而降低能源方差集中度；代价是牺牲部分纯油价冲击下的上行夏普。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-cb2ab35c55", "type": "thesis", "label": "AI 基础设施扩张从规模驱动切换到 TCO、能效和利用率筛选", "conviction": 78, "direction": "conflicting", "lane": "AI 基建 TCO/能效筛选", "laneHref": "/investor/lanes/ai-infrastructure-tco-and-efficiency-reset/", "bucket": "AI 基础设施", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 70.3, "href": "/investor/theses/thesis-cb2ab35c55/", "networkHref": "/investor/research-network/#thesis-cb2ab35c55" }, { "id": "stock-7d94cbbbb3", "type": "stock", "label": "SMCI", "name": "Super Micro Computer", "market": "US", "role": "AI 服务器与液冷整机", "href": "/investor/stocks/SMCI/" }, { "id": "stock-a6bb0d8c7a", "type": "stock", "label": "DELL", "name": "Dell Technologies", "market": "US", "role": "AI 服务器交付", "href": "/investor/stocks/DELL/" }, { "id": "stock-6bfeca4163", "type": "stock", "label": "HPE", "name": "Hewlett Packard Enterprise", "market": "US", "role": "企业服务器与网络", "href": "/investor/stocks/HPE/" }, { "id": "stock-8ed9c54f09", "type": "stock", "label": "VRT", "name": "Vertiv", "market": "US", "role": "电源与液冷基础设施", "href": "/investor/stocks/VRT/" }, { "id": "term-ef242752b3", "type": "term", "label": "电力", "href": "/investor/research-network/?q=%E7%94%B5%E5%8A%9B" }, { "id": "event-a502f64b7e87", "type": "event", "label": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-382f75860f3e", "type": "event", "label": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-131e45383877", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-31dcc87b2443", "type": "event", "label": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c12c85acaaaa", "type": "event", "label": "国产大模型并未告别规模竞赛，而是在先进GPU受限的条件下把竞争单位改写为每瓦、每token、每受限加速器和每留存用户的有效智能产出。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-30081889500b", "type": "event", "label": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-199d3247058b", "type": "event", "label": "AI算力基础设施需求与海底数据中心技术演进", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ae97917af7e5", "type": "event", "label": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-741d1d6f34", "type": "thesis", "label": "地产、土地财政与 REITs 是中国信用底的慢变量", "conviction": 78, "direction": "conflicting", "lane": "地产信用底/REITs", "laneHref": "/investor/lanes/china-property-credit-floor-and-reits/", "bucket": "信用、地产与久期", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 72.8, "href": "/investor/theses/thesis-741d1d6f34/", "networkHref": "/investor/research-network/#thesis-741d1d6f34" }, { "id": "stock-7eb3864ff8", "type": "stock", "label": "0960.HK", "name": "龙湖集团", "market": "HK", "role": "民营房企信用观察", "href": "/investor/stocks/0960.HK/" }, { "id": "stock-05e0aa017a", "type": "stock", "label": "1109.HK", "name": "华润置地", "market": "HK", "role": "央国企地产核心", "href": "/investor/stocks/1109.HK/" }, { "id": "stock-c734fc6e80", "type": "stock", "label": "0688.HK", "name": "中国海外发展", "market": "HK", "role": "央国企地产核心", "href": "/investor/stocks/0688.HK/" }, { "id": "stock-ab55154707", "type": "stock", "label": "600048.SH", "name": "保利发展", "market": "A-share", "role": "A股地产核心", "href": "/investor/stocks/600048.SH/" }, { "id": "stock-8b1b4a2855", "type": "stock", "label": "001979.SZ", "name": "招商蛇口", "market": "A-share", "role": "地产与园区开发", "href": "/investor/stocks/001979.SZ/" }, { "id": "stock-0ac1c28ffb", "type": "stock", "label": "0823.HK", "name": "领展房产基金", "market": "HK", "role": "REITs 现金流对照", "href": "/investor/stocks/0823.HK/" }, { "id": "term-690f4826dc", "type": "term", "label": "REITs", "href": "/investor/research-network/?q=REITs" }, { "id": "term-40040cb181", "type": "term", "label": "Property", "href": "/investor/research-network/?q=Property" }, { "id": "term-90eae63487", "type": "term", "label": "land", "href": "/investor/research-network/?q=land" }, { "id": "term-eab762a03f", "type": "term", "label": "finance", "href": "/investor/research-network/?q=finance" }, { "id": "event-b30f2f931db1", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-cbf042737a86", "type": "event", "label": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-221c6dc16c7b", "type": "event", "label": "商品+CTA危机阿尔法敞口在A股境内的可投资复制：资源/上游股票、黄金、商品期货与国债期货组合", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-177d9bd16790", "type": "event", "label": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-po…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9fbcaf1bb709", "type": "event", "label": "USD/CNY汇率趋势对通胀对冲组合的影响评估", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-8ecb5c3aa619", "type": "event", "label": "再通胀预期下应把组合向通胀对冲和短久期防御倾斜，但不是全面risk-off。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1354540e8ea0", "type": "event", "label": "面向AI基础设施信用尾部的跨资产组合配置与对冲", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-885916f10391", "type": "event", "label": "截至2026-06-28，A股电力主设备与核心分接开关商在手订单质量高、兑现度极强，而高估值温控链及无MEC保护的直接出口意向订单面临较大递延折价风险。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-7a6599a3f5", "type": "thesis", "label": "新一代信息技术国产替代要从信创叙事进入算力、软件与安全订单验证", "conviction": 77, "direction": "conflicting", "lane": "信创/国产 IT 栈", "laneHref": "/investor/lanes/domestic-it-substitution-stack/", "bucket": "AI 基础设施", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 66.1, "href": "/investor/theses/thesis-7a6599a3f5/", "networkHref": "/investor/research-network/#thesis-7a6599a3f5" }, { "id": "stock-e2d5caa67f", "type": "stock", "label": "688111.SH", "name": "金山办公", "market": "A-share", "role": "国产办公软件", "href": "/investor/stocks/688111.SH/" }, { "id": "stock-bf6151209d", "type": "stock", "label": "000977.SZ", "name": "浪潮信息", "market": "A-share", "role": "服务器与算力基础设施", "href": "/investor/stocks/000977.SZ/" }, { "id": "stock-ca5f9510eb", "type": "stock", "label": "000063.SZ", "name": "中兴通讯", "market": "A-share", "role": "ICT 设备与服务器", "href": "/investor/stocks/000063.SZ/" }, { "id": "stock-56463b158c", "type": "stock", "label": "688041.SH", "name": "海光信息", "market": "A-share", "role": "国产 CPU/GPU", "href": "/investor/stocks/688041.SH/" }, { "id": "stock-6c871273fc", "type": "stock", "label": "688256.SH", "name": "寒武纪", "market": "A-share", "role": "AI 芯片弹性", "href": "/investor/stocks/688256.SH/" }, { "id": "stock-add00a99a1", "type": "stock", "label": "688981.SH", "name": "中芯国际", "market": "A-share", "role": "国产晶圆制造", "href": "/investor/stocks/688981.SH/" }, { "id": "term-2ddcf6fb42", "type": "term", "label": "变压器", "href": "/investor/research-network/?q=%E5%8F%98%E5%8E%8B%E5%99%A8" }, { "id": "term-6001ec95f6", "type": "term", "label": "并网", "href": "/investor/research-network/?q=%E5%B9%B6%E7%BD%91" }, { "id": "term-09331fca77", "type": "term", "label": "Hyperscaler", "href": "/investor/research-network/?q=Hyperscaler" }, { "id": "event-44d5072754bf", "type": "event", "label": "AI数据中心并网延期与变压器死锁下全球算力硬件 Capex 的结构性错配", "date": "2026-06-30", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ba8e36973cf5", "type": "event", "label": "AI算力行情与电网瓶颈是否已成为同一个交易", "date": "2026-06-30", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8eed65178d0f", "type": "event", "label": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c6d8e316d5f4", "type": "event", "label": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-092cb21731a6", "type": "event", "label": "物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4d7095acec29", "type": "event", "label": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-29c9429db8f8", "type": "event", "label": "电网与变压器瓶颈对AI算力扩张上限的实质性约束评估", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6a85464c3787", "type": "event", "label": "超大规模厂商通过自备能源与地理外迁绕过电网瓶颈的对策存在严重的监管、燃料管道、物理时延与设备交付制约，难以扭转通电滑点导致的资本效率受损与资产减值风险。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-603a5f7966", "type": "thesis", "label": "A 股硬科技与科创50的复苏需要订单、融资盘和指数结构共振", "conviction": 76, "direction": "conflicting", "lane": "A股硬科技/科创50", "laneHref": "/investor/lanes/a-share-hard-tech-and-star50/", "bucket": "中国权益与流动性", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 67.9, "href": "/investor/theses/thesis-603a5f7966/", "networkHref": "/investor/research-network/#thesis-603a5f7966" }, { "id": "stock-f0cecca160", "type": "stock", "label": "688008.SH", "name": "澜起科技", "market": "A-share", "role": "内存接口芯片", "href": "/investor/stocks/688008.SH/" }, { "id": "term-d0cbbe9778", "type": "term", "label": "AI 芯片", "href": "/investor/research-network/?q=AI%20%E8%8A%AF%E7%89%87" }, { "id": "event-65883110e41f", "type": "event", "label": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-82226b8ac732", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-5424cce8e24a", "type": "event", "label": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6b59f25c3f08", "type": "event", "label": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-16cf8fc6dbca", "type": "event", "label": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-954bfcf5f6a3", "type": "event", "label": "在半导体周期复苏不及预期与高负债压力的复合冲击下，AI基础设施板块估值呈现两极分化：缺乏盈利的高负债概念股将面临系统性去估值及杠杆踩踏风险，而具备真实…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-33fcdeb90faf", "type": "event", "label": "效率驱动的算力架构转型：从“暴力规模”到“高能效密度”的物理基座适配性评估", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-12d1d1ce5829", "type": "event", "label": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "thesis-7f0a75c5f6", "type": "thesis", "label": "数据中心地产、液冷与水资源正在决定 AI 容量的区域迁移", "conviction": 76, "direction": "conflicting", "lane": "数据中心地产/液冷", "laneHref": "/investor/lanes/data-center-real-estate-and-cooling/", "bucket": "AI 基础设施", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 66.7, "href": "/investor/theses/thesis-7f0a75c5f6/", "networkHref": "/investor/research-network/#thesis-7f0a75c5f6" }, { "id": "stock-c10eb640cf", "type": "stock", "label": "EQIX", "name": "Equinix", "market": "US", "role": "全球数据中心 REIT", "href": "/investor/stocks/EQIX/" }, { "id": "stock-875caa8e41", "type": "stock", "label": "DLR", "name": "Digital Realty", "market": "US", "role": "批发数据中心 REIT", "href": "/investor/stocks/DLR/" }, { "id": "stock-4b984b3ea8", "type": "stock", "label": "IRM", "name": "Iron Mountain", "market": "US", "role": "数据中心与存储 REIT", "href": "/investor/stocks/IRM/" }, { "id": "stock-9779469589", "type": "stock", "label": "TT", "name": "Trane Technologies", "market": "US", "role": "冷却与 HVAC", "href": "/investor/stocks/TT/" }, { "id": "stock-d1d35fcaf6", "type": "stock", "label": "CARR", "name": "Carrier", "market": "US", "role": "冷却设备", "href": "/investor/stocks/CARR/" }, { "id": "term-690156f98a", "type": "term", "label": "水电", "href": "/investor/research-network/?q=%E6%B0%B4%E7%94%B5" }, { "id": "event-3cdef9304f99", "type": "event", "label": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-59e1b33f841b", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-eda974c7d178", "type": "event", "label": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4507c9c65d05", "type": "event", "label": "变压器交期拉长至3-5年与并网排队超36个月的物理刚性瓶颈，配合美欧贸易保护关税对GOES硅钢供应链的阻断，已对AI算力扩张的通电效率构成无法用资本单…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-290ec8b911b9", "type": "event", "label": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-2401ecd9a077", "type": "event", "label": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-40500e5cb763", "type": "event", "label": "数据中心电力并网、关键电力设备与液冷温控的物理交付瓶颈将对算力上线速度形成硬约束，导致云厂商ROIC受损并延缓算力采购，增加电子产业链利润扩张的硬着陆…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-2082dea424d5", "type": "event", "label": "AI 数据中心 capex 从 GPU 向配电、UPS、母线槽与液冷供应商迁移的跨资产定价", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-44cf9d91f6", "type": "thesis", "label": "中国大健康 Alpha 正从医保压价转向创新药出海、临床确定性与设备刚需", "conviction": 76, "direction": "conflicting", "lane": "中国大健康/创新药出海", "laneHref": "/investor/lanes/china-healthcare-and-overseas-innovation-drug-monetization/", "bucket": "消费健康防御", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 70, "href": "/investor/theses/thesis-44cf9d91f6/", "networkHref": "/investor/research-network/#thesis-44cf9d91f6" }, { "id": "stock-6681ba3430", "type": "stock", "label": "2269.HK", "name": "药明生物", "market": "HK", "role": "生物药外包与地缘风险", "href": "/investor/stocks/2269.HK/" }, { "id": "stock-732952499c", "type": "stock", "label": "2359.HK", "name": "药明康德", "market": "HK", "role": "CXO 出海风险代理", "href": "/investor/stocks/2359.HK/" }, { "id": "stock-1f5eddebbe", "type": "stock", "label": "1801.HK", "name": "信达生物", "market": "HK", "role": "创新药管线", "href": "/investor/stocks/1801.HK/" }, { "id": "stock-bab38386c1", "type": "stock", "label": "9969.HK", "name": "诺诚健华", "market": "HK", "role": "创新药临床进展", "href": "/investor/stocks/9969.HK/" }, { "id": "term-60856a95b4", "type": "term", "label": "药明康德", "href": "/investor/research-network/?q=%E8%8D%AF%E6%98%8E%E5%BA%B7%E5%BE%B7" }, { "id": "term-ede68d6aba", "type": "term", "label": "CXO", "href": "/investor/research-network/?q=CXO" }, { "id": "term-8ed3f6ad68", "type": "term", "label": "alpha", "href": "/investor/research-network/?q=alpha" }, { "id": "event-44f9d08cab5a", "type": "event", "label": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ed39b6671ab1", "type": "event", "label": "去风险三期医药资产通过rNPV模型释放3.48亿美元的期权价值构成Biotech估值底，而生命科学工具在5.5x EV/Sales筑底、CXO在 ov…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1be9fab7215d", "type": "event", "label": "三通道监管风险与实物关税叠加，确立了向‘纯出海授权、低受控敞口’标的切换的战术风格轮动信号。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-cd5a3967260e", "type": "event", "label": "大健康低覆盖机会更像跨子行业alpha篮子，而不是单一医疗板块beta；优先研究能把2025-2026催化转化为收入桥的五家公司。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-51d2949b0102", "type": "event", "label": "全球能源供应增量（特别是重启周期的核能与不稳定的可再生能源）及慢速的电网升级在2026-2027年内无法有效对冲 AI 基础设施的电网并网刚性瓶颈，该…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9acc96cee034", "type": "event", "label": "大健康产业链处于‘被忽视定价’阶段，基本面反转已现，应围绕出海BD与高端国产替代定向深挖五只低覆盖中盘股，防范地缘立法与科创去杠杆等跨链外溢。", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3d40679406f9", "type": "event", "label": "AI 基础设施需求与全球算力资本开支对半导体订单支撑的验证", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7ded98b2ce81", "type": "event", "label": "截至2026-06-27工作日，BIOSECURE法案地缘压力测试证实，CXO的估值中枢压制主要来自低毛利重组运营与合规时间税，而非业务一次性清零。", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-31f32af77b", "type": "thesis", "label": "中国云与国产算力栈的关键不是模型热度，而是训练/推理订单、服务器交付和生态锁定", "conviction": 76, "direction": "conflicting", "lane": "中国云/国产算力栈", "laneHref": "/investor/lanes/china-cloud-and-local-compute-stack/", "bucket": "AI 基础设施", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 70.5, "href": "/investor/theses/thesis-31f32af77b/", "networkHref": "/investor/research-network/#thesis-31f32af77b" }, { "id": "stock-a67fba8195", "type": "stock", "label": "601138.SH", "name": "工业富联", "market": "A-share", "role": "服务器制造", "href": "/investor/stocks/601138.SH/" }, { "id": "term-89b1aee428", "type": "term", "label": "寒武纪", "href": "/investor/research-network/?q=%E5%AF%92%E6%AD%A6%E7%BA%AA" }, { "id": "term-56681010b7", "type": "term", "label": "cloud", "href": "/investor/research-network/?q=cloud" }, { "id": "term-25bf8e1a23", "type": "term", "label": "local", "href": "/investor/research-network/?q=local" }, { "id": "event-8c6beb26f97a", "type": "event", "label": "AI算力行情与电网瓶颈是否已成为同一个交易", "date": "2026-06-30", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-0349b2df98cd", "type": "event", "label": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-81436edf403a", "type": "event", "label": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-92df911eceef", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b0d219ae76d0", "type": "event", "label": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a2ca6929226a", "type": "event", "label": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d7ec1b1c2462", "type": "event", "label": "[auto-handoff] Cloud customer acceptance of higher memory prices", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-3146ffe4c973", "type": "event", "label": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "thesis-a803fa3312", "type": "thesis", "label": "美联储路径仍是全球资产再定价的总阀门", "conviction": 76, "direction": "conflicting", "lane": "美债利率/美元再定价", "laneHref": "/investor/lanes/us-rates-and-dollar-repricing/", "bucket": "宏观利率与汇率", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 69.8, "href": "/investor/theses/thesis-a803fa3312/", "networkHref": "/investor/research-network/#thesis-a803fa3312" }, { "id": "stock-7b88702902", "type": "stock", "label": "KRE", "name": "SPDR S&P Regional Banking ETF", "market": "US ETF", "role": "利率/信用压力代理", "href": "/investor/stocks/KRE/" }, { "id": "term-51360b078e", "type": "term", "label": "美联储", "href": "/investor/research-network/?q=%E7%BE%8E%E8%81%94%E5%82%A8" }, { "id": "term-f55a72aff5", "type": "term", "label": "DXY", "href": "/investor/research-network/?q=DXY" }, { "id": "term-be0da309c0", "type": "term", "label": "美债", "href": "/investor/research-network/?q=%E7%BE%8E%E5%80%BA" }, { "id": "term-a0af9f865b", "type": "term", "label": "path", "href": "/investor/research-network/?q=path" }, { "id": "event-3c11690e550e", "type": "event", "label": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-189b8c5ccf28", "type": "event", "label": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a594224bf3a2", "type": "event", "label": "美元强势第二阶段与人民币/A股外资流传导", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-5b725936daf4", "type": "event", "label": "美联储2026年6月鹰派转向：新主席Warsh首会下点阵图翻转为加息，叠加中东能源冲击", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-98ef1a275c83", "type": "event", "label": "Warsh 时代美联储 2026 年 6 月的鹰派范式切换及其对美元、全球利率与中国政策空间的外溢", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-e19d15dc233b", "type": "event", "label": "沃什联储6月鹰派转向：能源供给冲击下的'更高更久、加息可选'与全球外溢", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c1a66691be8f", "type": "event", "label": "Warsh 时代美联储鹰派转向：美国通胀再加速与 2026 年加息风险", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d7dcf7fdd331", "type": "event", "label": "美元 carry 仍有利差支撑，但在 DXY 高位后，USD/CNY 与 USD/JPY 的边际吸引力应从加杠杆顺势交易降级为带保护的战术 carry。", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-08dd090dc9", "type": "thesis", "label": "公用事业从防御红利切换到电网扩容、容量费与信用再定价", "conviction": 76, "direction": "conflicting", "lane": "公用事业/电网信用", "laneHref": "/investor/lanes/utilities-grid-rate-base-and-credit/", "bucket": "中国权益与流动性", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 69.9, "href": "/investor/theses/thesis-08dd090dc9/", "networkHref": "/investor/research-network/#thesis-08dd090dc9" }, { "id": "stock-bfa4cc2c68", "type": "stock", "label": "600025.SH", "name": "华能水电", "market": "A-share", "role": "水电现金流", "href": "/investor/stocks/600025.SH/" }, { "id": "stock-7fe8dd309c", "type": "stock", "label": "600905.SH", "name": "三峡能源", "market": "A-share", "role": "新能源运营商", "href": "/investor/stocks/600905.SH/" }, { "id": "stock-420d41ec1c", "type": "stock", "label": "DUK", "name": "Duke Energy", "market": "US", "role": "受监管公用事业", "href": "/investor/stocks/DUK/" }, { "id": "stock-db052c8066", "type": "stock", "label": "SO", "name": "Southern Company", "market": "US", "role": "受监管电力与核电", "href": "/investor/stocks/SO/" }, { "id": "stock-0405f6aceb", "type": "stock", "label": "AEP", "name": "American Electric Power", "market": "US", "role": "输配电与 rate base", "href": "/investor/stocks/AEP/" }, { "id": "event-9a50046c3525", "type": "event", "label": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8d5393cb276a", "type": "event", "label": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-037d4a207c2f", "type": "event", "label": "选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-2ce41eabd726", "type": "event", "label": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b52c539eed36", "type": "event", "label": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-62d5d022e562", "type": "event", "label": "物理基础设施约束下多资产投资组合的重平衡与防守哑铃配置策略", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7d2c76e10f99", "type": "event", "label": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ee16c633bf32", "type": "event", "label": "在通缩压力下，公用事业并非均质的防御锚；必须坚守‘电网优先’与具备正向国债利差的可归因费率资产，警惕无费率保护的算力PPA延迟以及高溢价新能源标的的估…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-3986819dc0", "type": "thesis", "label": "A 股半导体的核心风险从订单验证转向 A/H 溢价、两融与负 Gamma", "conviction": 76, "direction": "conflicting", "lane": "A股半导体/两融负 Gamma", "laneHref": "/investor/lanes/a-share-semiconductor-margin-and-negative-gamma/", "bucket": "中国权益与流动性", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 70.5, "href": "/investor/theses/thesis-3986819dc0/", "networkHref": "/investor/research-network/#thesis-3986819dc0" }, { "id": "stock-82c0637078", "type": "stock", "label": "0981.HK", "name": "中芯国际", "market": "HK", "role": "H 股估值锚", "href": "/investor/stocks/0981.HK/" }, { "id": "stock-1a47896f9f", "type": "stock", "label": "688012.SH", "name": "中微公司", "market": "A-share", "role": "半导体设备", "href": "/investor/stocks/688012.SH/" }, { "id": "term-c484297089", "type": "term", "label": "两融", "href": "/investor/research-network/?q=%E4%B8%A4%E8%9E%8D" }, { "id": "term-dcc56252e4", "type": "term", "label": "半导体", "href": "/investor/research-network/?q=%E5%8D%8A%E5%AF%BC%E4%BD%93" }, { "id": "event-55cd6027661b", "type": "event", "label": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-69545bc373ad", "type": "event", "label": "[auto-handoff] 半导体折算率调降保证金缺口与科创板强平风控敏感性测试", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4d98ea861149", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a3efb2f84285", "type": "event", "label": "虽然278%平均维保安全垫使两融连锁抛压呈局部非系统性风险，但20-30%回撤仍将触发最高2528.00亿的平仓压力并重创产业链，应通过减仓高倍数前道…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e9b0629c23b4", "type": "event", "label": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-bab78f037e63", "type": "event", "label": "在紧平衡流动性与外部融资约束下，A股难以支撑全局性风险偏好扩张，下半年配置应降低总Beta，采用“硬科技国产替代（封装/设备）+高独立现金流红利”的结…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-be4044271fce", "type": "event", "label": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a2fecf2d64b2", "type": "event", "label": "中国央行通过结构性再贷款与离岸FIMA人民币国债抵押回购机制，成功解耦了汇率防御与境内流动性，为A股“高股息防御+硬科技成长”的滞胀杠铃配置提供了宏观…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-740e729f22", "type": "thesis", "label": "人民币、美元与套息交易决定跨境风险资产的流动性折价", "conviction": 75, "direction": "conflicting", "lane": "人民币/美元流动性敏感资产", "laneHref": "/investor/lanes/rmb-dollar-and-liquidity-sensitive-assets/", "bucket": "风险偏好与高波动", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 67.5, "href": "/investor/theses/thesis-740e729f22/", "networkHref": "/investor/research-network/#thesis-740e729f22" }, { "id": "stock-3d3ced839f", "type": "stock", "label": "3988.HK", "name": "中国银行", "market": "HK", "role": "人民币与银行流动性", "href": "/investor/stocks/3988.HK/" }, { "id": "term-b117b5bed4", "type": "term", "label": "USD/CNH", "href": "/investor/research-network/?q=USD%2FCNH" }, { "id": "term-ba9771ca73", "type": "term", "label": "人民币", "href": "/investor/research-network/?q=%E4%BA%BA%E6%B0%91%E5%B8%81" }, { "id": "term-b650f21f60", "type": "term", "label": "日元", "href": "/investor/research-network/?q=%E6%97%A5%E5%85%83" }, { "id": "event-cbe4c8425406", "type": "event", "label": "日元套利平仓通过离岸人民币流动性抽水及掉期清算挤压双重渠道，系统性抬高离岸融资成本并技术性击穿USD/CNY 7.00关口，导致人民币套利空间彻底崩溃。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3e3dd9763375", "type": "event", "label": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-927f21b2af4f", "type": "event", "label": "中美利差倒挂对离岸人民币（CNH）汇率及跨境资本流动的传导效应", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1252355979da", "type": "event", "label": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4a0157d4090a", "type": "event", "label": "中国内需政策强度及对内需受益资产的支撑评估", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3d6ddb43f55c", "type": "event", "label": "美元强势第二阶段与人民币/A股外资流传导", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a475a8524047", "type": "event", "label": "强美元仍需作为中国风险组合的保险，但新增风险预算应从裸多DXY转向有界USD/CNH保护，并主动削减以JPY/CHF融资的拥挤新兴市场套息。", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b110ea275700", "type": "event", "label": "0%远期风险准备金率下USD/CNY套利成本下降对离岸人民币利差交易（Carry Trade）的影响评估", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-9e97140446", "type": "thesis", "label": "中国电力设备出海进入订单质量与政策筛选期", "conviction": 75, "direction": "conflicting", "lane": "中国电力设备出海", "laneHref": "/investor/lanes/china-power-equipment-exports/", "bucket": "电力与公用事业", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 64.8, "href": "/investor/theses/thesis-9e97140446/", "networkHref": "/investor/research-network/#thesis-9e97140446" }, { "id": "stock-0e73308404", "type": "stock", "label": "002028.SZ", "name": "思源电气", "market": "A-share", "role": "开关设备与输配电", "href": "/investor/stocks/002028.SZ/" }, { "id": "stock-f960e0de27", "type": "stock", "label": "600089.SH", "name": "特变电工", "market": "A-share", "role": "变压器/电缆/海外工程", "href": "/investor/stocks/600089.SH/" }, { "id": "stock-8c35af6c45", "type": "stock", "label": "601179.SH", "name": "中国西电", "market": "A-share", "role": "高压设备出口", "href": "/investor/stocks/601179.SH/" }, { "id": "stock-e1734e0e85", "type": "stock", "label": "601126.SH", "name": "四方股份", "market": "A-share", "role": "电网自动化与保护", "href": "/investor/stocks/601126.SH/" }, { "id": "stock-dc9d1d443f", "type": "stock", "label": "300360.SZ", "name": "炬华科技", "market": "A-share", "role": "智能电表出海", "href": "/investor/stocks/300360.SZ/" }, { "id": "stock-b5733c9451", "type": "stock", "label": "ETN", "name": "Eaton", "market": "US", "role": "配电与电力管理", "href": "/investor/stocks/ETN/" }, { "id": "term-2359bb6154", "type": "term", "label": "电力设备", "href": "/investor/research-network/?q=%E7%94%B5%E5%8A%9B%E8%AE%BE%E5%A4%87" }, { "id": "term-641102ee2d", "type": "term", "label": "Eaton", "href": "/investor/research-network/?q=Eaton" }, { "id": "term-7d5ee7ec76", "type": "term", "label": "电网设备", "href": "/investor/research-network/?q=%E7%94%B5%E7%BD%91%E8%AE%BE%E5%A4%87" }, { "id": "event-4c32f30661be", "type": "event", "label": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-63112f9365c6", "type": "event", "label": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1c156edfd31e", "type": "event", "label": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-po…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b6596239eb27", "type": "event", "label": "电力设备出海高增长与国内“十五五”电网资本开支爆发锁定了龙头业绩的高能见度与利润确定性，是中证红利风格强化趋势下红利成长股的首选品种。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-422e1e7dc374", "type": "event", "label": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-f085bbdb6bf8", "type": "event", "label": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-700a918545a6", "type": "event", "label": "AI电力需求增长对电网侧扩容及配电资产的压力测试", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-be6b58aa0895", "type": "event", "label": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-ce29facfbd", "type": "thesis", "label": "中国资产定价仍受政策、流动性与风险偏好共同驱动", "conviction": 75, "direction": "conflicting", "lane": "中国权益流动性", "laneHref": "/investor/lanes/china-equity-liquidity/", "bucket": "风险偏好与高波动", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 65.4, "href": "/investor/theses/thesis-ce29facfbd/", "networkHref": "/investor/research-network/#thesis-ce29facfbd" }, { "id": "stock-d5a25b6072", "type": "stock", "label": "159915.SZ", "name": "创业板 ETF", "market": "A-share ETF", "role": "成长风险偏好", "href": "/investor/stocks/159915.SZ/" }, { "id": "stock-02c323bcb5", "type": "stock", "label": "3067.HK", "name": "南方恒生科技 ETF", "market": "HK ETF", "role": "港股科技代理", "href": "/investor/stocks/3067.HK/" }, { "id": "event-e5eafeef5a03", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-36938377f9e4", "type": "event", "label": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-80d6d38efedd", "type": "event", "label": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8252a052dbd2", "type": "event", "label": "国产HBM良率硬瓶颈锁死芯片供给并拉长LLM研发周期，促使高估值纯GPU设计股迎来TAM幻觉破灭的估值压缩，在1.72%国债与5.5%中证红利股息率带…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a6bd71a5eff4", "type": "event", "label": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-f240a6f06bfb", "type": "event", "label": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3183435e6b91", "type": "event", "label": "在先进封装良率瓶颈与表后电力并网延迟的双重物理约束下，A股AI硬件板块估值极度偏高且存在变现拥挤，难以独立承担成长杠铃的对冲配置功能，应当实施防御性减…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e16e83a229b1", "type": "event", "label": "AI电力需求增长对电网侧扩容及配电资产的压力测试", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-4b3b37e207", "type": "thesis", "label": "散户情绪与北向资金背离是 A 股风险偏好的关键温度计", "conviction": 75, "direction": "conflicting", "lane": "A股风险偏好/两融", "laneHref": "/investor/lanes/a-share-risk-appetite-and-margin/", "bucket": "中国权益与流动性", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 67.6, "href": "/investor/theses/thesis-4b3b37e207/", "networkHref": "/investor/research-network/#thesis-4b3b37e207" }, { "id": "stock-517b2f6590", "type": "stock", "label": "512100.SH", "name": "中证1000 ETF", "market": "A-share ETF", "role": "小盘风险偏好", "href": "/investor/stocks/512100.SH/" }, { "id": "stock-d54aa25b07", "type": "stock", "label": "300750.SZ", "name": "宁德时代", "market": "A-share", "role": "成长权重代理", "href": "/investor/stocks/300750.SZ/" }, { "id": "term-c36bc299e1", "type": "term", "label": "北向资金", "href": "/investor/research-network/?q=%E5%8C%97%E5%90%91%E8%B5%84%E9%87%91" }, { "id": "term-13aa5df7e2", "type": "term", "label": "中证1000", "href": "/investor/research-network/?q=%E4%B8%AD%E8%AF%811000" }, { "id": "term-cdb38c8dc8", "type": "term", "label": "Retail", "href": "/investor/research-network/?q=Retail" }, { "id": "term-df82d72441", "type": "term", "label": "sentiment", "href": "/investor/research-network/?q=sentiment" }, { "id": "event-bd2c67637e06", "type": "event", "label": "在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4a31147f09b0", "type": "event", "label": "人民币汇率防守边界测试（即期7.28、3M HIBOR 5.8%）会导致对冲Carry反转，驱动北向资金规模性流出并经由流动性折价放大A股科技两融强平…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4bfe72c8cee3", "type": "event", "label": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1f6391bc5b00", "type": "event", "label": "[auto-handoff] A股 3.74万亿成交分化下高杠杆科创50与电子板块两融平仓压力测算", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6e8526945c11", "type": "event", "label": "A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-809da0001211", "type": "event", "label": "[auto-handoff] A股端午节后两融杠杆回补持续性与大盘指数阻力位评估", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-a875362de659", "type": "event", "label": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-677f281c56de", "type": "event", "label": "实施两期隔离（账户完全隔离与无融资杠杆）是防范两融折算率大降（0.6至0.4）引发6.0pp保证金缺口、确保商品期权3.0pp封顶与6pp黄金仓位安全…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-d0de40066b", "type": "thesis", "label": "食品加工：成本传导、渠道库存与防御现金流复盘", "conviction": 75, "direction": "conflicting", "lane": "宏观利率与汇率", "laneHref": "/investor/lanes/lane-cdafce64/", "bucket": "宏观利率与汇率", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 70.1, "href": "/investor/theses/thesis-d0de40066b/", "networkHref": "/investor/research-network/#thesis-d0de40066b" }, { "id": "term-f072652dd0", "type": "term", "label": "食品加工产业链", "href": "/investor/research-network/?q=%E9%A3%9F%E5%93%81%E5%8A%A0%E5%B7%A5%E4%BA%A7%E4%B8%9A%E9%93%BE" }, { "id": "term-585a293e96", "type": "term", "label": "研究活跃度与投资线索复盘", "href": "/investor/research-network/?q=%E7%A0%94%E7%A9%B6%E6%B4%BB%E8%B7%83%E5%BA%A6%E4%B8%8E%E6%8A%95%E8%B5%84%E7%BA%BF%E7%B4%A2%E5%A4%8D%E7%9B%98" }, { "id": "event-53f4fe8837f7", "type": "event", "label": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a5ca3a24fc31", "type": "event", "label": "消费需求结构性演变：从耐用品'以旧换新'向服务与必需品类别的防御性迁移", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-6a31fa607ab6", "type": "event", "label": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-589931146907", "type": "event", "label": "食品加工板块β偏弱但覆盖缺口蕴含丰厚Alpha，应重点布局中报Setup干净的千味央厨与受新添加剂国标催化的B端生物配料龙头，回避无定价权的中小盘。", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3cbc7fe600ac", "type": "event", "label": "美大豆天气风险与双边关税扰动的叠加将打破国内豆粕的通缩基础，通过流向偏转、基差飙升及汇率乘数将大商所豆粕推至极高水位，重创饲料与养殖链，并最终传导侵蚀…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d0719a325ab9", "type": "event", "label": "食品加工产业链低激活率反映覆盖不足，二线供应链龙头在渠道红利与成本回落双重驱动下迎来利润反转，应超配盖世食品、百龙创园等高壁垒二线标的，避开立案的田野…", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-8d4528fa2799", "type": "event", "label": "2026-06-27的核心判断是：用50%中债久期和18%尾部商品对冲替代传统高股息/出海股票防御，才能同时免疫国内ASP通缩与海外关税地缘风险。", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c32b9684d1f4", "type": "event", "label": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-97a24f8796", "type": "thesis", "label": "公用事业产业链：研究活跃度与投资线索复盘", "conviction": 75, "direction": "conflicting", "lane": "电力与公用事业", "laneHref": "/investor/lanes/lane-ee7fbf3f/", "bucket": "电力与公用事业", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 66.7, "href": "/investor/theses/thesis-97a24f8796/", "networkHref": "/investor/research-network/#thesis-97a24f8796" }, { "id": "stock-61e19d9ee8", "type": "stock", "label": "GEV", "name": "GE Vernova", "market": "US", "role": "燃气轮机与电网设备", "href": "/investor/stocks/GEV/" }, { "id": "stock-8ee9008c2b", "type": "stock", "label": "PWR", "name": "Quanta Services", "market": "US", "role": "并网与电力工程", "href": "/investor/stocks/PWR/" }, { "id": "stock-59e40ddab7", "type": "stock", "label": "VST", "name": "Vistra", "market": "US", "role": "独立发电与数据中心电力", "href": "/investor/stocks/VST/" }, { "id": "stock-2d1c22cd10", "type": "stock", "label": "CEG", "name": "Constellation Energy", "market": "US", "role": "核电与数据中心电力", "href": "/investor/stocks/CEG/" }, { "id": "term-6ac13ea9eb", "type": "term", "label": "公用事业产业链", "href": "/investor/research-network/?q=%E5%85%AC%E7%94%A8%E4%BA%8B%E4%B8%9A%E4%BA%A7%E4%B8%9A%E9%93%BE" }, { "id": "event-948ce8147538", "type": "event", "label": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e4aab152e0c7", "type": "event", "label": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9de344767759", "type": "event", "label": "选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8cdd707be994", "type": "event", "label": "长江电力在极端联合压力下股息现金覆盖率仍达125.55%，分红耐久性无实质破坏风险，是优质的防守压舱石，而高资本开支的中国核电分红受制于负现金流和债务…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a51f5aa78ca6", "type": "event", "label": "研判总结 / Research Summary 本卡片作为 公用事业分析师 (Utilities Analyst) ，锚定研究院权威工作日期 2026…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a18a385d1209", "type": "event", "label": "物理基础设施约束下多资产投资组合的重平衡与防守哑铃配置策略", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ca73498b2b6f", "type": "event", "label": "PQC 引发的周期性热降频会显著提高超大规模 AI 集群 PUE/WUE 与能源成本预测难度，并在集中接入区增加电力调度复杂度，但它是边际波动因子而非…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-e18a1790bbe8", "type": "event", "label": "为了防止AI数据中心无序扩张导致居民电费暴涨，公用事业监管机构正强制推行以‘85%起付线、长周期和强担保’为特征的大负荷电价机制，实质性地将电网容量闲…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-f57e888558", "type": "thesis", "label": "Select one analyst from the institute roster, search for a timely re…", "conviction": 75, "direction": "conflicting", "lane": "中国权益与流动性", "laneHref": "/investor/lanes/lane-c06f0832/", "bucket": "中国权益与流动性", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 74, "href": "/investor/theses/thesis-f57e888558/", "networkHref": "/investor/research-network/#thesis-f57e888558" }, { "id": "term-2a78025de6", "type": "term", "label": "Select", "href": "/investor/research-network/?q=Select" }, { "id": "term-3165df34fc", "type": "term", "label": "institute", "href": "/investor/research-network/?q=institute" }, { "id": "term-c6836f7ad1", "type": "term", "label": "roster", "href": "/investor/research-network/?q=roster" }, { "id": "event-81eff6c97d3d", "type": "event", "label": "特种化学品：供应端波动对下游利润的压力", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6dcab2c0827c", "type": "event", "label": "Model-selected kickoff research", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-2916df1947a2", "type": "event", "label": "Your Role: 半导体分析师 (Semiconductors Analyst) Expertise: Semiconductors: desig…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-512eeeadc0c9", "type": "event", "label": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-5d7ba6993082", "type": "event", "label": "中国限制美国企业会否重定价关键材料与高端制造风险溢价？", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4fde4a0cb3fb", "type": "event", "label": "Research note 01 complete. The shared workspace was empty at start — sessio…", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-07a388e99618", "type": "event", "label": "Model-selected kickoff research", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-2a4c925e5939", "type": "event", "label": "2026 06 24T09:21:04.772955Z ERROR codex models manager::manager: failed to …", "date": "2026-06-24", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "thesis-5fd96966de", "type": "thesis", "label": "公用事业：AI 电力需求与信用指标复盘", "conviction": 75, "direction": "conflicting", "lane": "电力与公用事业", "laneHref": "/investor/lanes/lane-ee7fbf3f/", "bucket": "电力与公用事业", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 68.2, "href": "/investor/theses/thesis-5fd96966de/", "networkHref": "/investor/research-network/#thesis-5fd96966de" }, { "id": "term-420421da52", "type": "term", "label": "红利", "href": "/investor/research-network/?q=%E7%BA%A2%E5%88%A9" }, { "id": "term-d5d45771d8", "type": "term", "label": "核电", "href": "/investor/research-network/?q=%E6%A0%B8%E7%94%B5" }, { "id": "term-5eca243822", "type": "term", "label": "提供了极强的基本面防御力", "href": "/investor/research-network/?q=%E6%8F%90%E4%BE%9B%E4%BA%86%E6%9E%81%E5%BC%BA%E7%9A%84%E5%9F%BA%E6%9C%AC%E9%9D%A2%E9%98%B2%E5%BE%A1%E5%8A%9B" }, { "id": "term-ca2444c4a9", "type": "term", "label": "景下", "href": "/investor/research-network/?q=%E6%99%AF%E4%B8%8B" }, { "id": "event-4a4f1a776751", "type": "event", "label": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-27b160f1ac21", "type": "event", "label": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1b3560e2c773", "type": "event", "label": "选择性电力上限下的正确组合不是放弃AI基础设施，而是用红利公用事业和申万电子Put约束拥挤Beta，同时持有能把递延订单转化为盈利的受资助龙头。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-03a3342eee44", "type": "event", "label": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e474a4c6993e", "type": "event", "label": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8fbbb8b683e2", "type": "event", "label": "物理基础设施约束下多资产投资组合的重平衡与防守哑铃配置策略", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-cace5af8b9a6", "type": "event", "label": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-bcd18f03af4f", "type": "event", "label": "局部中小银行信用缺口下的A股银行红利耐久性与高股息内部轮动", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-71476ded81", "type": "thesis", "label": "Practical research topic:", "conviction": 75, "direction": "conflicting", "lane": "消费健康防御", "laneHref": "/investor/lanes/lane-d104d465/", "bucket": "消费健康防御", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 68.8, "href": "/investor/theses/thesis-71476ded81/", "networkHref": "/investor/research-network/#thesis-71476ded81" }, { "id": "term-54b759dba2", "type": "term", "label": "specific", "href": "/investor/research-network/?q=specific" }, { "id": "term-34c2d1dc4e", "type": "term", "label": "insid", "href": "/investor/research-network/?q=insid" }, { "id": "event-695e0c9919e3", "type": "event", "label": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9356dc9fa383", "type": "event", "label": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e746b90c0092", "type": "event", "label": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ab05993142bf", "type": "event", "label": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-0424aae47a1f", "type": "event", "label": "I have completed the research for Research note 01 of this research thread …", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-f6b15f4fabb0", "type": "event", "label": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-69a280c78b7d", "type": "event", "label": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1148f08264fb", "type": "event", "label": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-bb8f3d8ee7", "type": "thesis", "label": "隔夜美股半导体与小盘逼空是否足以覆盖科创50/CPO的本地两融、解禁和折算率压力，还是应继续把科技回补限制在5%试仓以内？", "conviction": 75, "direction": "conflicting", "lane": "中国权益与流动性", "laneHref": "/investor/lanes/lane-c06f0832/", "bucket": "中国权益与流动性", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 67.7, "href": "/investor/theses/thesis-bb8f3d8ee7/", "networkHref": "/investor/research-network/#thesis-bb8f3d8ee7" }, { "id": "term-a6bb0d8c7a", "type": "term", "label": "DELL", "href": "/investor/research-network/?q=DELL" }, { "id": "term-4364651c17", "type": "term", "label": "FY27", "href": "/investor/research-network/?q=FY27" }, { "id": "term-117d8f914d", "type": "term", "label": "收入", "href": "/investor/research-network/?q=%E6%94%B6%E5%85%A5" }, { "id": "event-2285c4e9718c", "type": "event", "label": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-38e91753ba34", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-73d04a867e42", "type": "event", "label": "A股红利拥挤度与7-11月年内解禁洪峰限制了风格切换空间，而在人行短端流动性上限失灵与CNH被动升值情景下，高负债、高Beta的券商板块将面临最严重的…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-de2f9bcf10b1", "type": "event", "label": "在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c066f01581da", "type": "event", "label": "虽然278%平均维保安全垫使两融连锁抛压呈局部非系统性风险，但20-30%回撤仍将触发最高2528.00亿的平仓压力并重创产业链，应通过减仓高倍数前道…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-25c16ae53780", "type": "event", "label": "在紧平衡流动性与外部融资约束下，A股难以支撑全局性风险偏好扩张，下半年配置应降低总Beta，采用“硬科技国产替代（封装/设备）+高独立现金流红利”的结…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-69a1d0233e78", "type": "event", "label": "券商信用风险的局部破线是否会通过流动性传导至整个金融板块，进而引发非对称的流动性危机？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b3492849393c", "type": "event", "label": "A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-62e0155999", "type": "thesis", "label": "美元兑日元跌破161.50关口，财务省干预概率陡增。一旦日元套利平仓瞬间引爆全球EM科技股抛售，叠加SpaceX二次配售24.5亿元的资…", "conviction": 75, "direction": "conflicting", "lane": "宏观利率与汇率", "laneHref": "/investor/lanes/lane-cdafce64/", "bucket": "宏观利率与汇率", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 67.8, "href": "/investor/theses/thesis-62e0155999/", "networkHref": "/investor/research-network/#thesis-62e0155999" }, { "id": "term-afd623b148", "type": "term", "label": "SpaceX", "href": "/investor/research-network/?q=SpaceX" }, { "id": "term-f2373e2b95", "type": "term", "label": "美元兑日元跌破161.50关口", "href": "/investor/research-network/?q=%E7%BE%8E%E5%85%83%E5%85%91%E6%97%A5%E5%85%83%E8%B7%8C%E7%A0%B4161.50%E5%85%B3%E5%8F%A3" }, { "id": "term-5689ce9e64", "type": "term", "label": "财务省干预概率陡增", "href": "/investor/research-network/?q=%E8%B4%A2%E5%8A%A1%E7%9C%81%E5%B9%B2%E9%A2%84%E6%A6%82%E7%8E%87%E9%99%A1%E5%A2%9E" }, { "id": "event-5f84de78735c", "type": "event", "label": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6f6df3b9c78b", "type": "event", "label": "A股红利拥挤度与7-11月年内解禁洪峰限制了风格切换空间，而在人行短端流动性上限失灵与CNH被动升值情景下，高负债、高Beta的券商板块将面临最严重的…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a8ed72a894c0", "type": "event", "label": "A股市场流动性与资本流出风险评估", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7fdb637a7cc9", "type": "event", "label": "2026-06-29的核心判断是：人行短久期托底利好金融与内需现金流，不能无差别追成长，出口型TMT硬件是CNH升值情景下最需要降权的A股风格风险。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3454262b6ac8", "type": "event", "label": "虽然278%平均维保安全垫使两融连锁抛压呈局部非系统性风险，但20-30%回撤仍将触发最高2528.00亿的平仓压力并重创产业链，应通过减仓高倍数前道…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-852961ede0c3", "type": "event", "label": "券商信用风险的局部破线是否会通过流动性传导至整个金融板块，进而引发非对称的流动性危机？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-fb7476f7603e", "type": "event", "label": "已确认 OMO 窗口下的跨资产布局：国债久期、人民币套息/对冲与利率敏感 A 股倾斜", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-f9e03c6463ba", "type": "event", "label": "A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-5cb77a9f52", "type": "thesis", "label": "在红利拥挤度逼近历史极值、长江电力达90%且地方拖欠致公用账款增加15%背景下，红利底座应坚守水电核电，还是大举切仓至融资占比低、严重低…", "conviction": 75, "direction": "conflicting", "lane": "中国权益与流动性", "laneHref": "/investor/lanes/lane-c06f0832/", "bucket": "中国权益与流动性", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 66.9, "href": "/investor/theses/thesis-5cb77a9f52/", "networkHref": "/investor/research-network/#thesis-5cb77a9f52" }, { "id": "term-cd256779c8", "type": "term", "label": "在红利拥挤度逼近历史极值", "href": "/investor/research-network/?q=%E5%9C%A8%E7%BA%A2%E5%88%A9%E6%8B%A5%E6%8C%A4%E5%BA%A6%E9%80%BC%E8%BF%91%E5%8E%86%E5%8F%B2%E6%9E%81%E5%80%BC" }, { "id": "term-5a513a4a57", "type": "term", "label": "加15%背景下", "href": "/investor/research-network/?q=%E5%8A%A015%25%E8%83%8C%E6%99%AF%E4%B8%8B" }, { "id": "event-5046138b843d", "type": "event", "label": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-16b7fd5a17d0", "type": "event", "label": "在物理电网配给约束和三季报估值重估的双重压力下，多资产策略应实施战术防御性风险平价（TDRP），将AI成长股风险预算由25.0%减记至8.0%，同时在…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3743442f4403", "type": "event", "label": "在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1484a7fd044b", "type": "event", "label": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6d2761252824", "type": "event", "label": "AI电网物理约束常态化下，A股投资应超配饱揽5.1倍硅钢成本红利与全球变压器缺口的高出口溢价电力设备商，并搭配防守型大水电以对冲低毛利传统TMT的电费…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ac53fe2d839b", "type": "event", "label": "局部中小银行信用缺口下的A股银行红利耐久性与高股息内部轮动", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-123f6db7413e", "type": "event", "label": "AI电力需求支持公用事业红利逻辑，但只有存量现金流、容量电价、电价传导和融资能力能覆盖资本开支周期的公司，才能在不削减电网资本开支的情况下维持分红。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6b194bcc9466", "type": "event", "label": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-e4a70ed58a", "type": "thesis", "label": "在布伦特油价从约108美元回落至约80美元、核心CPI仍受控于2.9%的背景下，5月4.2%的整体通胀冲击主要是暂时性的（允许美联储维持…", "conviction": 75, "direction": "conflicting", "lane": "宏观利率与汇率", "laneHref": "/investor/lanes/lane-cdafce64/", "bucket": "宏观利率与汇率", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 67.3, "href": "/investor/theses/thesis-e4a70ed58a/", "networkHref": "/investor/research-network/#thesis-e4a70ed58a" }, { "id": "term-1034e2871a", "type": "term", "label": "允许美联储维持", "href": "/investor/research-network/?q=%E5%85%81%E8%AE%B8%E7%BE%8E%E8%81%94%E5%82%A8%E7%BB%B4%E6%8C%81" }, { "id": "term-ce842dfe73", "type": "term", "label": "全球二次通胀预期退坡", "href": "/investor/research-network/?q=%E5%85%A8%E7%90%83%E4%BA%8C%E6%AC%A1%E9%80%9A%E8%83%80%E9%A2%84%E6%9C%9F%E9%80%80%E5%9D%A1" }, { "id": "event-97d9a540f95d", "type": "event", "label": "在霍尔木兹海峡中断（油价均值105美元）与重开（回落至72-79美元但存在60天到岸刚性）下，美国整体CPI分别面临~82.3 bp上升与~78.6 …", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-729b32d56f57", "type": "event", "label": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-94fa276aa0a9", "type": "event", "label": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-2d3db6a843e5", "type": "event", "label": "2026年石油供应中断对全球能源品种及炼油产能的冲击评估", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-8d0c32819f6f", "type": "event", "label": "美联储6月鹰派转向与能源冲击逆转的错配:点阵图是否已经过时", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d37b7703b6b3", "type": "event", "label": "能源价格的崩溃通过被动推升实质利率（188-264个基点）对实体经济造成隐性紧缩冲击，显著放大了美联储在Warsh新主席政策框架下过度紧缩的政策失误尾…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e1543b664125", "type": "event", "label": "2026年6月美联储、欧央行、日本央行同步转鹰，却在一场正在消退的伊朗战争油价冲击中收紧政策", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-f24f3fc66873", "type": "event", "label": "基准配置应优先通过久期表达去通胀与政策失误风险，权益只做选择性低贝塔，商品低配整体贝塔但保留油价重上100美元的显式尾部对冲。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-05c3f6f33c", "type": "thesis", "label": "随着TSMC CoWoS产能持续紧张且HBM4于2026年开始量产，从HBM3e向HBM4的跨越将如何影响主要AI GPU厂商（如NVI…", "conviction": 75, "direction": "conflicting", "lane": "AI 基础设施", "laneHref": "/investor/lanes/ai/", "bucket": "AI 基础设施", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 67.7, "href": "/investor/theses/thesis-05c3f6f33c/", "networkHref": "/investor/research-network/#thesis-05c3f6f33c" }, { "id": "term-e01b27172a", "type": "term", "label": "HBM4", "href": "/investor/research-network/?q=HBM4" }, { "id": "term-2de5d0f518", "type": "term", "label": "HBM3e", "href": "/investor/research-network/?q=HBM3e" }, { "id": "event-7d8ce389087c", "type": "event", "label": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-33e738eeb030", "type": "event", "label": "先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-05592b0b7f2f", "type": "event", "label": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-658c48f3885c", "type": "event", "label": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6c87297ae466", "type": "event", "label": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-dad343bcce54", "type": "event", "label": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-59dc5918dc57", "type": "event", "label": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b4d468dbca31", "type": "event", "label": "在2026-06-27，AI电力并网限制（2年延误拖累53.71亿美元[S12]）将倒逼算力采购削减并重创高端半导体高增长假设，投资者应果断削减AI硬…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-72097390dd", "type": "thesis", "label": "在66.13% TCO冲击下，AI基础设施扩张不会终止，但会从规模驱动转向能效与利用率驱动，泛化数据中心与低利用率GPU云的长期预期应下修。", "conviction": 75, "direction": "conflicting", "lane": "AI 基础设施", "laneHref": "/investor/lanes/ai/", "bucket": "AI 基础设施", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 67.8, "href": "/investor/theses/thesis-72097390dd/", "networkHref": "/investor/research-network/#thesis-72097390dd" }, { "id": "term-ee72426269", "type": "term", "label": "AI基础设施", "href": "/investor/research-network/?q=AI%E5%9F%BA%E7%A1%80%E8%AE%BE%E6%96%BD" }, { "id": "term-344e9a768f", "type": "term", "label": "在66.13%", "href": "/investor/research-network/?q=%E5%9C%A866.13%25" }, { "id": "event-aea8c9f9f834", "type": "event", "label": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ed61e691b018", "type": "event", "label": "AI交易不会结束，但估值核心应从买到多少芯片转向能否按期通电、保持利用率并降低每token能耗。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-0cbb20ece214", "type": "event", "label": "电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-59b919e4332d", "type": "event", "label": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-f795702f985c", "type": "event", "label": "AI数据中心‘电力荒’：电网接入瓶颈下的超大规模云厂商资本开支与自备电站转型", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-af52a9e95b31", "type": "event", "label": "变压器材料和交付瓶颈已开始改变 Hyperscaler 的2026-2027资本开支与电力采购节奏，影响主要体现为预算通胀、长期锁电和通电交付递延，而…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3fd26cbe5656", "type": "event", "label": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastru…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-840f94a7a705", "type": "event", "label": "高利率平台期对AI基础设施资本开支的影响", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-e1d3ec2fed", "type": "thesis", "label": "AIDC 并网与变压器瓶颈：AI 投产约束复盘", "conviction": 75, "direction": "conflicting", "lane": "电力与公用事业", "laneHref": "/investor/lanes/lane-ee7fbf3f/", "bucket": "电力与公用事业", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 65.7, "href": "/investor/theses/thesis-e1d3ec2fed/", "networkHref": "/investor/research-network/#thesis-e1d3ec2fed" }, { "id": "term-8e289bbb4a", "type": "term", "label": "取向硅钢", "href": "/investor/research-network/?q=%E5%8F%96%E5%90%91%E7%A1%85%E9%92%A2" }, { "id": "event-1a91636e7402", "type": "event", "label": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-aa924e4a514d", "type": "event", "label": "上游取向硅钢和电工铜的产能稀缺与关税高墙正转嫁为严重成本通胀，侵蚀设备制造厂毛利并拉长交付，导致AI电力落地面临实质性溢价风险。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8cda15ce88bd", "type": "event", "label": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-46fd93307b2c", "type": "event", "label": "2026年全球AI数据中心电网瓶颈与“自主供电”战略的崛起", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-237a8864ecca", "type": "event", "label": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ec7a4714681b", "type": "event", "label": "AI电力需求增长对电网侧扩容及配电资产的压力测试", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9e4440057d6d", "type": "event", "label": "功率半导体已过剩，AI电力的新物理瓶颈已迁移至大型变压器，并真正卡在无法被资本快速绕行的上游电磁钢(GOES)与铜——2026–2028窗口难解。", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9eb33a52bb69", "type": "event", "label": "宽禁带半导体(SiC/GaN)仅能优化数据中心机柜内部能效，无法解决中压电网侧变压器交期超150周及GOES材料短缺的电网接入瓶颈，AI算力扩张的电网…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-33e46fee64", "type": "thesis", "label": "2026 HBM 缺口真实但租金窗口前置在 1H26，真正的稀缺量子是 CoWoS 先进封装而非 HBM 裸晶——杠铃方向成立但工具应以…", "conviction": 75, "direction": "conflicting", "lane": "AI 基础设施", "laneHref": "/investor/lanes/ai/", "bucket": "AI 基础设施", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 67.4, "href": "/investor/theses/thesis-33e46fee64/", "networkHref": "/investor/research-network/#thesis-33e46fee64" }, { "id": "term-17c5c517dc", "type": "term", "label": "缺口真实但租金窗口前置在", "href": "/investor/research-network/?q=%E7%BC%BA%E5%8F%A3%E7%9C%9F%E5%AE%9E%E4%BD%86%E7%A7%9F%E9%87%91%E7%AA%97%E5%8F%A3%E5%89%8D%E7%BD%AE%E5%9C%A8" }, { "id": "term-38e8415b91", "type": "term", "label": "真正的稀缺量子是", "href": "/investor/research-network/?q=%E7%9C%9F%E6%AD%A3%E7%9A%84%E7%A8%80%E7%BC%BA%E9%87%8F%E5%AD%90%E6%98%AF" }, { "id": "term-16296c2af8", "type": "term", "label": "先进封装而非", "href": "/investor/research-network/?q=%E5%85%88%E8%BF%9B%E5%B0%81%E8%A3%85%E8%80%8C%E9%9D%9E" }, { "id": "event-c8a1621c2deb", "type": "event", "label": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-88c9e5e35073", "type": "event", "label": "先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1ff6251f6df9", "type": "event", "label": "国产HBM与先进封装不是零进展，但2026年的主线应买“工艺环节国产化”，而不是把AI芯片设计公司当作不受HBM约束的无限TAM。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-aac517316b81", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-be832fdbcc70", "type": "event", "label": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-de053451ea32", "type": "event", "label": "GPU供应中断真实但已非最紧约束——2026年算力扩容受制于CoWoS封装、HBM内存与电网电力的多年期物理瓶颈，下游AI盈利将承受成本通胀。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8457dd2b4002", "type": "event", "label": "AI资本开支仍有韧性，但它只能支撑少数物理瓶颈资产，不能证明整个AI硬件权益复合体具备防御性。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ad5e6f458ec2", "type": "event", "label": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-f4ca874ce0", "type": "thesis", "label": "地缘封锁带来的105美元油价与物流双击将美国推向滞胀困局，消费呈现结构性撕裂而非整体崩溃，而新任美联储主席凯文·沃什治下的鹰派加息风险已…", "conviction": 75, "direction": "conflicting", "lane": "宏观利率与汇率", "laneHref": "/investor/lanes/lane-cdafce64/", "bucket": "宏观利率与汇率", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 67.9, "href": "/investor/theses/thesis-f4ca874ce0/", "networkHref": "/investor/research-network/#thesis-f4ca874ce0" }, { "id": "term-11690d135b", "type": "term", "label": "霍尔木兹海峡", "href": "/investor/research-network/?q=%E9%9C%8D%E5%B0%94%E6%9C%A8%E5%85%B9%E6%B5%B7%E5%B3%A1" }, { "id": "term-8d7bcb1cc8", "type": "term", "label": "Warsh", "href": "/investor/research-network/?q=Warsh" }, { "id": "term-6cfed8e96e", "type": "term", "label": "将美国推向滞胀困局", "href": "/investor/research-network/?q=%E5%B0%86%E7%BE%8E%E5%9B%BD%E6%8E%A8%E5%90%91%E6%BB%9E%E8%83%80%E5%9B%B0%E5%B1%80" }, { "id": "event-9d4d3683cde4", "type": "event", "label": "在霍尔木兹海峡中断（油价均值105美元）与重开（回落至72-79美元但存在60天到岸刚性）下，美国整体CPI分别面临~82.3 bp上升与~78.6 …", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8bf65e63f5d9", "type": "event", "label": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d0eacf5594d5", "type": "event", "label": "霍尔木兹滞胀脉冲下的美联储终端利率与全球流动性拐点", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-f7509a147d21", "type": "event", "label": "能源价格的崩溃通过被动推升实质利率（188-264个基点）对实体经济造成隐性紧缩冲击，显著放大了美联储在Warsh新主席政策框架下过度紧缩的政策失误尾…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4b415c4732dd", "type": "event", "label": "2026年6月美联储、欧央行、日本央行同步转鹰，却在一场正在消退的伊朗战争油价冲击中收紧政策", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e895241c18d5", "type": "event", "label": "美联储6月鹰派转向与能源冲击逆转的错配:点阵图是否已经过时", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d3c2a51f9250", "type": "event", "label": "基准配置应优先通过久期表达去通胀与政策失误风险，权益只做选择性低贝塔，商品低配整体贝塔但保留油价重上100美元的显式尾部对冲。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e53cf0a6d2c0", "type": "event", "label": "美联储2026年6月鹰派转向：供给冲击型通胀下的滞胀压力测试", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-f1c88bcd88", "type": "thesis", "label": "云厂商自发电：公用事业信用与电力资产再定价", "conviction": 75, "direction": "conflicting", "lane": "电力与公用事业", "laneHref": "/investor/lanes/lane-ee7fbf3f/", "bucket": "电力与公用事业", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 64.5, "href": "/investor/theses/thesis-f1c88bcd88/", "networkHref": "/investor/research-network/#thesis-f1c88bcd88" }, { "id": "term-95b6a12d6b", "type": "term", "label": "对利润率的压力测试", "href": "/investor/research-network/?q=%E5%AF%B9%E5%88%A9%E6%B6%A6%E7%8E%87%E7%9A%84%E5%8E%8B%E5%8A%9B%E6%B5%8B%E8%AF%95" }, { "id": "term-62f606258e", "type": "term", "label": "物理交付死结", "href": "/investor/research-network/?q=%E7%89%A9%E7%90%86%E4%BA%A4%E4%BB%98%E6%AD%BB%E7%BB%93" }, { "id": "term-7bf50a24dc", "type": "term", "label": "正在击穿", "href": "/investor/research-network/?q=%E6%AD%A3%E5%9C%A8%E5%87%BB%E7%A9%BF" }, { "id": "event-7f8747e4af76", "type": "event", "label": "电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d8bb9c5a6958", "type": "event", "label": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4621fe36c29a", "type": "event", "label": "铜与硅钢暴涨及高预付要求将击穿缺乏转嫁能力的二线及内销电力设备厂商的利润护城河，形成“有订单无利润、无现金”的贫血式增长。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7fc56702588d", "type": "event", "label": "电力设备出海高增长与国内“十五五”电网资本开支爆发锁定了龙头业绩的高能见度与利润确定性，是中证红利风格强化趋势下红利成长股的首选品种。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1a16afeda1df", "type": "event", "label": "电力设备成本传导逻辑的脆弱性：铜与电工钢供应瓶颈能否被设备商有效对冲？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b49035a56e7b", "type": "event", "label": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7c47750c4ccf", "type": "event", "label": "关键电力设备供应缺口对AI数据中心建设时间表与经济性的二阶影响", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ccf625fbdc7f", "type": "event", "label": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-11f0838e82", "type": "thesis", "label": "科创50收于1,751.32点，距离1,750雪球/两融生命线仅一步之遥。若下周破位跌破1,750，我们如何防范882亿元的级联式去杠杆…", "conviction": 75, "direction": "conflicting", "lane": "中国权益与流动性", "laneHref": "/investor/lanes/lane-c06f0832/", "bucket": "中国权益与流动性", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 66.9, "href": "/investor/theses/thesis-11f0838e82/", "networkHref": "/investor/research-network/#thesis-11f0838e82" }, { "id": "term-d3e6d49230", "type": "term", "label": "雪球", "href": "/investor/research-network/?q=%E9%9B%AA%E7%90%83" }, { "id": "term-a8ab3d5805", "type": "term", "label": "Gamma", "href": "/investor/research-network/?q=Gamma" }, { "id": "term-84c7ec1db8", "type": "term", "label": "与两融强平后", "href": "/investor/research-network/?q=%E4%B8%8E%E4%B8%A4%E8%9E%8D%E5%BC%BA%E5%B9%B3%E5%90%8E" }, { "id": "term-ab72c9a79f", "type": "term", "label": "科创50收于1", "href": "/investor/research-network/?q=%E7%A7%91%E5%88%9B50%E6%94%B6%E4%BA%8E1" }, { "id": "event-2b4cc1d264ee", "type": "event", "label": "国家队选择性托底加剧了A股的流动性分化，高杠杆的科创50与TMT板块在缺乏直接托底的情况下，面临两融强平与负Gamma期权共振的级联去杠杆螺旋，一旦波…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-40a613299c36", "type": "event", "label": "A股红利拥挤度与7-11月年内解禁洪峰限制了风格切换空间，而在人行短端流动性上限失灵与CNH被动升值情景下，高负债、高Beta的券商板块将面临最严重的…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b317b78bee1b", "type": "event", "label": "在两融与雪球期权负Gamma自反馈踩踏、以及人行利率走廊失效的极端流动性清算中，A股杠铃组合的VaR和最大回撤限制均被系统性穿透（回撤达-11.85%…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e1ea5816ea30", "type": "event", "label": "政策背离下的A股风格轮动与板块配置：能源安全杠铃 vs. 经利润分层的转型龙头，以及A股的相对庇护所超配", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-ca755df0d2aa", "type": "event", "label": "虽然278%平均维保安全垫使两融连锁抛压呈局部非系统性风险，但20-30%回撤仍将触发最高2528.00亿的平仓压力并重创产业链，应通过减仓高倍数前道…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e63fc75078ed", "type": "event", "label": "在紧平衡流动性与外部融资约束下，A股难以支撑全局性风险偏好扩张，下半年配置应降低总Beta，采用“硬科技国产替代（封装/设备）+高独立现金流红利”的结…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a44f36791089", "type": "event", "label": "A股风险不是总量两融失控，而是人民币压力、北向成交失衡与科技板块融资拥挤共振后，科创50在15%-20%回撤区间触发局部强平踩踏。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a22ac2b8e95a", "type": "event", "label": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-1495ab3592", "type": "thesis", "label": "在距离OBBBA法案7月4日开工红线仅22天、且变压器积压导致绿电并网排队项目创纪录撤回的背景下，电网容量费暴增9.3倍如何侵蚀云厂商A…", "conviction": 75, "direction": "conflicting", "lane": "电力与公用事业", "laneHref": "/investor/lanes/lane-ee7fbf3f/", "bucket": "电力与公用事业", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 66.1, "href": "/investor/theses/thesis-1495ab3592/", "networkHref": "/investor/research-network/#thesis-1495ab3592" }, { "id": "term-6b72da0e1f", "type": "term", "label": "OBBBA", "href": "/investor/research-network/?q=OBBBA" }, { "id": "term-b012bdba47", "type": "term", "label": "撤回的背景下", "href": "/investor/research-network/?q=%E6%92%A4%E5%9B%9E%E7%9A%84%E8%83%8C%E6%99%AF%E4%B8%8B" }, { "id": "event-9eee158c2486", "type": "event", "label": "AI数据中心并网延期与变压器死锁下全球算力硬件 Capex 的结构性错配", "date": "2026-06-30", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-021e981facd6", "type": "event", "label": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-9e922bb9aeb3", "type": "event", "label": "将能源链利润分化（发电侧承压、电网设备受益）整合进A股成长/价值杠铃配置", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b3cd2b14f754", "type": "event", "label": "AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4430394743d1", "type": "event", "label": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-5838928f9afe", "type": "event", "label": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4bd3fa72cb5d", "type": "event", "label": "在2026-06-28，物理加电瓶颈正迫使AI算力估值由单纯的Capex扩张转向已通电MW效率，投资者应超配锁电、BTM自发电超大型云厂商与掌握变压器…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-86c8d044449d", "type": "event", "label": "AI驱动的结构性裁员实质上是算力投资回报（ROI）兑现的证明，云厂商AI资本开支具有极强韧性，其真实制约并非宏观衰退担忧，而是电网并网延迟与重型电力设…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-2ec6b9d3ee", "type": "thesis", "label": "卫星直连与商业航天从概念弹性进入部署兑现期", "conviction": 74, "direction": "conflicting", "lane": "商业航天/卫星直连", "laneHref": "/investor/lanes/commercial-space-and-direct-to-device-satellite/", "bucket": "航天国防与新场景", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 67.3, "href": "/investor/theses/thesis-2ec6b9d3ee/", "networkHref": "/investor/research-network/#thesis-2ec6b9d3ee" }, { "id": "stock-56f9442441", "type": "stock", "label": "ASTS", "name": "AST SpaceMobile", "market": "US", "role": "手机直连卫星", "href": "/investor/stocks/ASTS/" }, { "id": "stock-c27c6260da", "type": "stock", "label": "RKLB", "name": "Rocket Lab", "market": "US", "role": "小型发射与空间系统", "href": "/investor/stocks/RKLB/" }, { "id": "stock-01467aabbc", "type": "stock", "label": "LUNR", "name": "Intuitive Machines", "market": "US", "role": "月球与空间服务", "href": "/investor/stocks/LUNR/" }, { "id": "stock-0d3089cbbb", "type": "stock", "label": "RDW", "name": "Redwire", "market": "US", "role": "空间基础设施", "href": "/investor/stocks/RDW/" }, { "id": "stock-bb2a93f2a1", "type": "stock", "label": "IRDM", "name": "Iridium", "market": "US", "role": "卫星通信现金流", "href": "/investor/stocks/IRDM/" }, { "id": "stock-b1ce5beca8", "type": "stock", "label": "LMT", "name": "Lockheed Martin", "market": "US", "role": "国防主承包商", "href": "/investor/stocks/LMT/" }, { "id": "term-5846e1a424", "type": "term", "label": "商业航天", "href": "/investor/research-network/?q=%E5%95%86%E4%B8%9A%E8%88%AA%E5%A4%A9" }, { "id": "term-56f9442441", "type": "term", "label": "ASTS", "href": "/investor/research-network/?q=ASTS" }, { "id": "term-a56b9af9a7", "type": "term", "label": "Direct-to-device", "href": "/investor/research-network/?q=Direct-to-device" }, { "id": "term-dc4b4e203f", "type": "term", "label": "satellite", "href": "/investor/research-network/?q=satellite" }, { "id": "term-d01acdec3a", "type": "term", "label": "commercial", "href": "/investor/research-network/?q=commercial" }, { "id": "term-3f49dbbfe0", "type": "term", "label": "space", "href": "/investor/research-network/?q=space" }, { "id": "event-fde861637c5d", "type": "event", "label": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-5fcf74fe1bac", "type": "event", "label": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a221038d25d3", "type": "event", "label": "变压器交期拉长与并网队列已构成AI数据中心最硬的物理钳口，迫使云厂商将资本开支从GPU采购转向兆瓦（MW）对齐的基础设施，并引发数十万机队闲置折旧与半…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-924f7a7ac124", "type": "event", "label": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-po…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1620661c8a11", "type": "event", "label": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ca328dc4d5c6", "type": "event", "label": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e9251fdff30c", "type": "event", "label": "<20%开机率红线是平台属性而非物理定律：对电力受限EO小卫星它由电源+被动散热共同锁定且先进热控不划算(支持卡03)，但两相泵驱回路+可展开辐射器已…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d567524ce374", "type": "event", "label": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "thesis-1aaa752ed4", "type": "thesis", "label": "AI 电力与电网瓶颈仍是基础设施主线", "conviction": 74, "direction": "conflicting", "lane": "AI 电力/电网设备", "laneHref": "/investor/lanes/ai-power-and-grid-equipment/", "bucket": "电力与公用事业", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 64.1, "href": "/investor/theses/thesis-1aaa752ed4/", "networkHref": "/investor/research-network/#thesis-1aaa752ed4" }, { "id": "event-44e26486d853", "type": "event", "label": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-po…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-68c72c0e7850", "type": "event", "label": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-129b9febf3c3", "type": "event", "label": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d090e448ccf4", "type": "event", "label": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-264a019f7d04", "type": "event", "label": "2026年全球AI数据中心电网瓶颈与“自主供电”战略的崛起", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-5e06185b573c", "type": "event", "label": "A股应继续支持AI基础设施主线，但买点应从纯算力叙事转向供电约束受益资产，以35%/30%/25%/10%的科技成长、电网设备、公用事业红利和现金对冲…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1c3c0b0e917a", "type": "event", "label": "研判总结 / Research Summary 本卡片作为 公用事业分析师 (Utilities Analyst) ，锚定研究院权威工作日期 2026…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9aec22bc96eb", "type": "event", "label": "AI驱动的结构性裁员实质上是算力投资回报（ROI）兑现的证明，云厂商AI资本开支具有极强韧性，其真实制约并非宏观衰退担忧，而是电网并网延迟与重型电力设…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-d4473d35e5", "type": "thesis", "label": "猪周期、粮食与农产品价格决定中国低通胀修复的弹性", "conviction": 74, "direction": "conflicting", "lane": "食品农业/猪周期", "laneHref": "/investor/lanes/food-agriculture-and-hog-cycle/", "bucket": "宏观利率与汇率", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 70.1, "href": "/investor/theses/thesis-d4473d35e5/", "networkHref": "/investor/research-network/#thesis-d4473d35e5" }, { "id": "stock-bf5ac90619", "type": "stock", "label": "002714.SZ", "name": "牧原股份", "market": "A-share", "role": "生猪养殖", "href": "/investor/stocks/002714.SZ/" }, { "id": "stock-32f4bdcdf7", "type": "stock", "label": "300498.SZ", "name": "温氏股份", "market": "A-share", "role": "生猪/禽类养殖", "href": "/investor/stocks/300498.SZ/" }, { "id": "stock-f3805b0eae", "type": "stock", "label": "000876.SZ", "name": "新希望", "market": "A-share", "role": "饲料与养殖", "href": "/investor/stocks/000876.SZ/" }, { "id": "stock-736b57465b", "type": "stock", "label": "BG", "name": "Bunge", "market": "US", "role": "粮油贸易加工", "href": "/investor/stocks/BG/" }, { "id": "stock-664acdf6ca", "type": "stock", "label": "DBA", "name": "Invesco DB Agriculture Fund", "market": "US ETF", "role": "农产品价格代理", "href": "/investor/stocks/DBA/" }, { "id": "term-c501935a6b", "type": "term", "label": "cycle", "href": "/investor/research-network/?q=cycle" }, { "id": "term-2c2240217b", "type": "term", "label": "grain", "href": "/investor/research-network/?q=grain" }, { "id": "term-1bb0b3a466", "type": "term", "label": "agricultural", "href": "/investor/research-network/?q=agricultural" }, { "id": "term-fe81f9390f", "type": "term", "label": "prices", "href": "/investor/research-network/?q=prices" }, { "id": "event-7480e0e340a9", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ad107c8a25e4", "type": "event", "label": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a03f87a2a6af", "type": "event", "label": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4f7d9cbb4e0e", "type": "event", "label": "主要央行保持鹰派按兵不动并非过度收紧的政策错误，而是应对由服务业传导支撑的核心PCE粘性（预计H2维持在3.1%-3.4%）的必要防御，但这会使实体经…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3f08e2d76329", "type": "event", "label": "核心服务与住房通胀的粘性：美联储政策的后续驱动力", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-f28549ef93f6", "type": "event", "label": "6月点阵图不是简单过时，而是可能“对得不是理由”：能源理由过时，AI/电网与折旧通道支持更高更久，但不足以单独证明立即加息。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-f9b60b0ab260", "type": "event", "label": "沃什联储6月鹰派转向：能源供给冲击下的'更高更久、加息可选'与全球外溢", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d83c1fbb708c", "type": "event", "label": "美国5月通胀与6月FOMC：能源冲击下的更高更久利率路径", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "thesis-bef6966fdd", "type": "thesis", "label": "工业机械的出口订单质量和设备更新周期决定制造业 CapEx 弹性", "conviction": 73, "direction": "conflicting", "lane": "工业机械/设备更新出海", "laneHref": "/investor/lanes/industrial-machinery-equipment-upgrade-and-exports/", "bucket": "电力与公用事业", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 64.3, "href": "/investor/theses/thesis-bef6966fdd/", "networkHref": "/investor/research-network/#thesis-bef6966fdd" }, { "id": "stock-a5b0d3a0ea", "type": "stock", "label": "600031.SH", "name": "三一重工", "market": "A-share", "role": "工程机械出口", "href": "/investor/stocks/600031.SH/" }, { "id": "stock-656011c288", "type": "stock", "label": "000425.SZ", "name": "徐工机械", "market": "A-share", "role": "工程机械全球化", "href": "/investor/stocks/000425.SZ/" }, { "id": "stock-e9ae311828", "type": "stock", "label": "000157.SZ", "name": "中联重科", "market": "A-share", "role": "工程机械与高机", "href": "/investor/stocks/000157.SZ/" }, { "id": "stock-6df13a035d", "type": "stock", "label": "603338.SH", "name": "浙江鼎力", "market": "A-share", "role": "高空作业平台出海", "href": "/investor/stocks/603338.SH/" }, { "id": "stock-6b65919db6", "type": "stock", "label": "600761.SH", "name": "安徽合力", "market": "A-share", "role": "叉车与工业车辆", "href": "/investor/stocks/600761.SH/" }, { "id": "stock-f1d36125fc", "type": "stock", "label": "300124.SZ", "name": "汇川技术", "market": "A-share", "role": "工控自动化", "href": "/investor/stocks/300124.SZ/" }, { "id": "term-68da2262bb", "type": "term", "label": "CapEx", "href": "/investor/research-network/?q=CapEx" }, { "id": "term-b60cae5339", "type": "term", "label": "Industrial", "href": "/investor/research-network/?q=Industrial" }, { "id": "term-7cb194d941", "type": "term", "label": "machinery", "href": "/investor/research-network/?q=machinery" }, { "id": "term-f44e64e75f", "type": "term", "label": "beta", "href": "/investor/research-network/?q=beta" }, { "id": "term-1e3350f128", "type": "term", "label": "depends", "href": "/investor/research-network/?q=depends" }, { "id": "event-744ce5df5d19", "type": "event", "label": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-35f1db70dbf6", "type": "event", "label": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-edb79402f972", "type": "event", "label": "[auto-handoff] 中压干式变压器出海企业海外合规及交付溢价测算", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-700f49766467", "type": "event", "label": "电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7c33eb87a57e", "type": "event", "label": "变压器交期拉长至3-5年与并网排队超36个月的物理刚性瓶颈，配合美欧贸易保护关税对GOES硅钢供应链的阻断，已对AI算力扩张的通电效率构成无法用资本单…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6c68f8178cba", "type": "event", "label": "电力设备核心原材料（GOES与铜）的供应瓶颈与成本风险", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-12997524a1c8", "type": "event", "label": "关键矿产与设备出口管制风险对A股高端环节溢价逻辑的冲击评估", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-86456b560b61", "type": "event", "label": "电力设备成本传导逻辑的脆弱性：铜与电工钢供应瓶颈能否被设备商有效对冲？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-9a860a8801", "type": "thesis", "label": "信用再融资压力正在把利率风险转化为资产负债表筛选", "conviction": 73, "direction": "conflicting", "lane": "信用再融资/私募信贷", "laneHref": "/investor/lanes/credit-refinancing-and-private-debt/", "bucket": "宏观利率与汇率", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 68.8, "href": "/investor/theses/thesis-9a860a8801/", "networkHref": "/investor/research-network/#thesis-9a860a8801" }, { "id": "stock-cec5af38a6", "type": "stock", "label": "BX", "name": "Blackstone", "market": "US", "role": "另类资产与私募信贷", "href": "/investor/stocks/BX/" }, { "id": "stock-db74d07203", "type": "stock", "label": "KKR", "name": "KKR", "market": "US", "role": "私募市场与信用", "href": "/investor/stocks/KKR/" }, { "id": "stock-d86863071d", "type": "stock", "label": "APO", "name": "Apollo Global Management", "market": "US", "role": "保险资产与私募信贷", "href": "/investor/stocks/APO/" }, { "id": "stock-d7e5daa1ba", "type": "stock", "label": "ARES", "name": "Ares Management", "market": "US", "role": "直接贷款与信用", "href": "/investor/stocks/ARES/" }, { "id": "term-fb1875d0a5", "type": "term", "label": "Credit", "href": "/investor/research-network/?q=Credit" }, { "id": "term-92b775598f", "type": "term", "label": "refinancing", "href": "/investor/research-network/?q=refinancing" }, { "id": "term-b667446e6e", "type": "term", "label": "pressure", "href": "/investor/research-network/?q=pressure" }, { "id": "term-2ab62cfe39", "type": "term", "label": "turning", "href": "/investor/research-network/?q=turning" }, { "id": "term-c549779d79", "type": "term", "label": "rate", "href": "/investor/research-network/?q=rate" }, { "id": "event-b2d44d4f6727", "type": "event", "label": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-po…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-48caf385e2f0", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d974ecb51a5f", "type": "event", "label": "绿色技改缺口下高碳工业民企债务违约风险与信用利差压力测试评估", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-91128dc8e101", "type": "event", "label": "国债底线稳的是无风险利率，不是高杠杆主体的信用利差；若弱城投和房企利差被重定价至250bp-400bp，展期压力会非线性上升，但仍更可能表现为行政化续…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d1e5ed00a796", "type": "event", "label": "[auto-handoff] 权益两融去杠杆是否外溢至银行间资金面并推动信用利差走阔", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-0aff6872aaae", "type": "event", "label": "并网延迟的变压器库存积压是可逆的时点拖累(金盘-8%~-11%、思源<-3%)，被国网4万亿内需托底，应在超配出口变压器框架下偏思源、轻金盘并叠加配对…", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e8199fd67cdf", "type": "event", "label": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防…", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a014e9443446", "type": "event", "label": "[auto-handoff] 商业银行 1.40% NIM 约束对保障房收储贷款投放的挤出压力", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-985b80b224", "type": "thesis", "label": "具身智能机器人进入 BOM 降本与执行器供给验证期", "conviction": 72, "direction": "conflicting", "lane": "具身智能/机器人零部件", "laneHref": "/investor/lanes/embodied-ai-and-robotics-components/", "bucket": "航天国防与新场景", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 62.3, "href": "/investor/theses/thesis-985b80b224/", "networkHref": "/investor/research-network/#thesis-985b80b224" }, { "id": "stock-0d72f0950a", "type": "stock", "label": "ISRG", "name": "Intuitive Surgical", "market": "US", "role": "高端机器人商业化对照", "href": "/investor/stocks/ISRG/" }, { "id": "stock-b0eea525de", "type": "stock", "label": "ABBNY", "name": "ABB", "market": "US ADR", "role": "工业机器人与自动化", "href": "/investor/stocks/ABBNY/" }, { "id": "stock-8a0d7a9b71", "type": "stock", "label": "6954.T", "name": "Fanuc", "market": "Japan", "role": "工业机器人", "href": "/investor/stocks/6954.T/" }, { "id": "stock-10ae1be006", "type": "stock", "label": "002747.SZ", "name": "埃斯顿", "market": "A-share", "role": "国产工业机器人", "href": "/investor/stocks/002747.SZ/" }, { "id": "term-d16dc977ac", "type": "term", "label": "Embodied", "href": "/investor/research-network/?q=Embodied" }, { "id": "term-7793ba1ad0", "type": "term", "label": "robotics", "href": "/investor/research-network/?q=robotics" }, { "id": "term-28faaa675c", "type": "term", "label": "enters", "href": "/investor/research-network/?q=enters" }, { "id": "term-63e427f481", "type": "term", "label": "deflation", "href": "/investor/research-network/?q=deflation" }, { "id": "term-425edd11c2", "type": "term", "label": "actuator", "href": "/investor/research-network/?q=actuator" }, { "id": "event-cf060bd3f061", "type": "event", "label": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-0855f5987bf6", "type": "event", "label": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-332b270b3c96", "type": "event", "label": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-fb8e3440c09f", "type": "event", "label": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-05bdd4816188", "type": "event", "label": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b6e155fd9417", "type": "event", "label": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-63806555c45b", "type": "event", "label": "[auto-handoff] Cloud customer acceptance of higher memory prices", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-ceea2085b7fe", "type": "event", "label": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-5057edfa79", "type": "thesis", "label": "AI-Biology 与医疗效率改善需要从科研突破走向支付和监管兑现", "conviction": 72, "direction": "conflicting", "lane": "AI 医疗/生物科技效率", "laneHref": "/investor/lanes/ai-healthcare-and-biology-productivity/", "bucket": "消费健康防御", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 66.8, "href": "/investor/theses/thesis-5057edfa79/", "networkHref": "/investor/research-network/#thesis-5057edfa79" }, { "id": "stock-20c0a1e555", "type": "stock", "label": "TMO", "name": "Thermo Fisher Scientific", "market": "US", "role": "生命科学工具", "href": "/investor/stocks/TMO/" }, { "id": "stock-6a8f57144b", "type": "stock", "label": "DHR", "name": "Danaher", "market": "US", "role": "生命科学工具", "href": "/investor/stocks/DHR/" }, { "id": "stock-1a5bdfc9bd", "type": "stock", "label": "REGN", "name": "Regeneron", "market": "US", "role": "创新药研发效率", "href": "/investor/stocks/REGN/" }, { "id": "stock-a55ffe9be6", "type": "stock", "label": "EXAI", "name": "Exscientia", "market": "US", "role": "AI drug discovery", "href": "/investor/stocks/EXAI/" }, { "id": "stock-56a2cb6836", "type": "stock", "label": "SDGR", "name": "Schrodinger", "market": "US", "role": "计算药物发现", "href": "/investor/stocks/SDGR/" }, { "id": "term-e47aa08e3a", "type": "term", "label": "AI-Biology", "href": "/investor/research-network/?q=AI-Biology" }, { "id": "term-af8f827a0c", "type": "term", "label": "biology", "href": "/investor/research-network/?q=biology" }, { "id": "event-14d779169908", "type": "event", "label": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d1096c39499e", "type": "event", "label": "材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d0ee08891b52", "type": "event", "label": "大健康低覆盖机会更像跨子行业alpha篮子，而不是单一医疗板块beta；优先研究能把2025-2026催化转化为收入桥的五家公司。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-99dfd805747d", "type": "event", "label": "去风险三期医药资产通过rNPV模型释放3.48亿美元的期权价值构成Biotech估值底，而生命科学工具在5.5x EV/Sales筑底、CXO在 ov…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-98fa8eb1bd91", "type": "event", "label": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8804692d782a", "type": "event", "label": "新一代信息技术最值得补覆盖的不是前端应用，而是 AI 基础设施与国产替代共同推动的设备、EDA、光互联、内存互联和自主 CPU 使能层。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-66497536a9a7", "type": "event", "label": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastru…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-329b5853dd8d", "type": "event", "label": "全球能源供应增量（特别是重启周期的核能与不稳定的可再生能源）及慢速的电网升级在2026-2027年内无法有效对冲 AI 基础设施的电网并网刚性瓶颈，该…", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-026aa1679b", "type": "thesis", "label": "云厂商自发电正在把 AI 交付风险转移到核电、燃气、电网与公用事业信用", "conviction": 72, "direction": "conflicting", "lane": "云厂商自发电/电力资产", "laneHref": "/investor/lanes/cloud-self-generation-and-power-assets/", "bucket": "电力与公用事业", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 63.5, "href": "/investor/theses/thesis-026aa1679b/", "networkHref": "/investor/research-network/#thesis-026aa1679b" }, { "id": "term-f67c2069ed", "type": "term", "label": "SMR", "href": "/investor/research-network/?q=SMR" }, { "id": "event-315256c9f3d7", "type": "event", "label": "电网(发电+输配+设备)是2026–2028 AI算力的硬天花板，按多年期时钟清出，宏观外溢是由全体用户承担的电价与容量成本冲击。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-74dd65b2130e", "type": "event", "label": "变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-2479960c402f", "type": "event", "label": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-509371f08368", "type": "event", "label": "超大规模厂商通过自备能源与地理外迁绕过电网瓶颈的对策存在严重的监管、燃料管道、物理时延与设备交付制约，难以扭转通电滑点导致的资本效率受损与资产减值风险。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a80f937943f9", "type": "event", "label": "电网与变压器瓶颈对AI算力扩张上限的实质性约束评估", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-fb601e54c617", "type": "event", "label": "AI数据中心电力瓶颈与核能PPA战略转型", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-130ee5fcd597", "type": "event", "label": "能源成本与高利率壁垒对全球算力中心电力建设与资本开支的传导效应", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-cbc7536aa13c", "type": "event", "label": "AI 数据中心 capex 从 GPU 向配电、UPS、母线槽与液冷供应商迁移的跨资产定价", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-21dd72c1b1", "type": "thesis", "label": "航空航天与国防供应链错配正在形成政策驱动的制造业弹性", "conviction": 72, "direction": "conflicting", "lane": "国防航空航天供应链", "laneHref": "/investor/lanes/defense-and-aerospace-supply-chain/", "bucket": "航天国防与新场景", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 63, "href": "/investor/theses/thesis-21dd72c1b1/", "networkHref": "/investor/research-network/#thesis-21dd72c1b1" }, { "id": "stock-75f04b8433", "type": "stock", "label": "RTX", "name": "RTX", "market": "US", "role": "航空发动机/防务", "href": "/investor/stocks/RTX/" }, { "id": "stock-148304483a", "type": "stock", "label": "NOC", "name": "Northrop Grumman", "market": "US", "role": "防务与航天", "href": "/investor/stocks/NOC/" }, { "id": "stock-884d5bff79", "type": "stock", "label": "GD", "name": "General Dynamics", "market": "US", "role": "防务平台", "href": "/investor/stocks/GD/" }, { "id": "stock-fb5bd3958a", "type": "stock", "label": "HWM", "name": "Howmet Aerospace", "market": "US", "role": "航空零部件", "href": "/investor/stocks/HWM/" }, { "id": "stock-296d71a7f6", "type": "stock", "label": "BA", "name": "Boeing", "market": "US", "role": "民航与防务供应链", "href": "/investor/stocks/BA/" }, { "id": "term-d527f8c5f9", "type": "term", "label": "Aerospace", "href": "/investor/research-network/?q=Aerospace" }, { "id": "term-1b0b7e51ee", "type": "term", "label": "defense", "href": "/investor/research-network/?q=defense" }, { "id": "term-3f412634a4", "type": "term", "label": "supply-chain", "href": "/investor/research-network/?q=supply-chain" }, { "id": "term-5acbfff1b0", "type": "term", "label": "mismatch", "href": "/investor/research-network/?q=mismatch" }, { "id": "term-7f9c40c9ac", "type": "term", "label": "creating", "href": "/investor/research-network/?q=creating" }, { "id": "event-089bbe0434df", "type": "event", "label": "关键变压器与重型燃机供应链产能的物理上限及长达3-7年的扩产交付周期，构成2026-2028年全球AI算力基础设施扩张的刚性物理限制，将直接导致千亿级…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-fb667d390e85", "type": "event", "label": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3dd0d941a3ad", "type": "event", "label": "电力设备需求爆发下的原材料价格波动及其对利润率的压力测试", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-70b3005cad15", "type": "event", "label": "在2030年前，由于设备供应链（变压器交期4-5年、GOES硅钢缺口）与网架建设周期（UHV需5-8年）的物理极限，政策和技术手段无法释放被压制的15…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-0973079dc94c", "type": "event", "label": "AI电力基础设施供应链瓶颈：电力变压器与关键配电硬件的产能缺口", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-953a753c1dd6", "type": "event", "label": "变压器核心原材料（如硅钢片）的供应紧缺和贸易壁垒所导致的交付周期拉长，确实大幅增加了公用事业电网升级的投资延迟风险，并将物理约束转化为公用事业的ROI…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3966f447e917", "type": "event", "label": "原始40/32/12/8/8 sleeve在AI需求相关性趋近+1时尾部风险过高，阶段1是强制去杠杆，阶段2才接近可接受，但决定性保护来自降AI be…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-2151b33ea5c3", "type": "event", "label": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-66eb42380e", "type": "thesis", "label": "海洋经济的政策催化需要落到船舶订单、海工装备与海上资源现金流", "conviction": 72, "direction": "conflicting", "lane": "海洋经济/船舶海工", "laneHref": "/investor/lanes/marine-economy-shipbuilding-and-offshore-equipment/", "bucket": "电力与公用事业", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 63.8, "href": "/investor/theses/thesis-66eb42380e/", "networkHref": "/investor/research-network/#thesis-66eb42380e" }, { "id": "stock-fdf60d294f", "type": "stock", "label": "600150.SH", "name": "中国船舶", "market": "A-share", "role": "造船主线", "href": "/investor/stocks/600150.SH/" }, { "id": "stock-cb6f6fac36", "type": "stock", "label": "601989.SH", "name": "中国重工", "market": "A-share", "role": "船舶与海工", "href": "/investor/stocks/601989.SH/" }, { "id": "stock-9ef40fc77b", "type": "stock", "label": "601919.SH", "name": "中远海控", "market": "A-share", "role": "集运与港口链", "href": "/investor/stocks/601919.SH/" }, { "id": "stock-8bd6af8d30", "type": "stock", "label": "600938.SH", "name": "中国海油", "market": "A-share", "role": "海上油气现金流", "href": "/investor/stocks/600938.SH/" }, { "id": "stock-b49750b06c", "type": "stock", "label": "600522.SH", "name": "中天科技", "market": "A-share", "role": "海缆与海风", "href": "/investor/stocks/600522.SH/" }, { "id": "term-017f1d8e6c", "type": "term", "label": "储能", "href": "/investor/research-network/?q=%E5%82%A8%E8%83%BD" }, { "id": "term-8a2fd55b01", "type": "term", "label": "电池", "href": "/investor/research-network/?q=%E7%94%B5%E6%B1%A0" }, { "id": "term-d8cfb8d54d", "type": "term", "label": "Marine-economy", "href": "/investor/research-network/?q=Marine-economy" }, { "id": "term-823412d1ea", "type": "term", "label": "policy", "href": "/investor/research-network/?q=policy" }, { "id": "event-f27ee71cf2df", "type": "event", "label": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-po…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-4af2b94a15a8", "type": "event", "label": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-5f97b773d84a", "type": "event", "label": "超长期特别国债与地方基金预算受阻的非对称限制，导致西部智算基地无法依靠补贴长期消化绿电LCOE折价收窄至13.89%与并网延迟造成的40%度电成本上升…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-711381b4c4ad", "type": "event", "label": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-f5a83b82e08c", "type": "event", "label": "海上风电+UDC的24/7低碳电力合同与并网可行性", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-d421f9e45dd4", "type": "event", "label": "由于储能与动力电池的绝对规模错配以及中国储能并网瓶颈和矿山复产传闻，储能需求增长在第三季度不足以支撑广期所碳酸锂期货价格企稳。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-0e49020737b2", "type": "event", "label": "AI电力基础设施供应链瓶颈：电力变压器与关键配电硬件的产能缺口", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a22091e382f7", "type": "event", "label": "在2026-06-29的政策环境下，通过将绿电绿证的环境资产价值（抵消范畴二排放以规避每度电约0.41元的碳税/碳关税）与电力系统调节成本（114号文…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "thesis-c6c6be4745", "type": "thesis", "label": "中国出口制造的胜负手从低价扩张转向利润率、合规和本地化", "conviction": 71, "direction": "conflicting", "lane": "中国出口制造/利润率", "laneHref": "/investor/lanes/china-export-manufacturing-and-margins/", "bucket": "宏观利率与汇率", "actionLabel": "风险雷达：降权跟踪", "actionTone": "red", "practicalScore": 67.3, "href": "/investor/theses/thesis-c6c6be4745/", "networkHref": "/investor/research-network/#thesis-c6c6be4745" }, { "id": "stock-e59c1688ee", "type": "stock", "label": "000333.SZ", "name": "美的集团", "market": "A-share", "role": "家电全球化", "href": "/investor/stocks/000333.SZ/" }, { "id": "stock-95587eff89", "type": "stock", "label": "600690.SH", "name": "海尔智家", "market": "A-share", "role": "家电全球化", "href": "/investor/stocks/600690.SH/" }, { "id": "stock-ce89b19c8e", "type": "stock", "label": "002594.SZ", "name": "比亚迪", "market": "A-share", "role": "新能源车出海", "href": "/investor/stocks/002594.SZ/" }, { "id": "stock-9441f62b62", "type": "stock", "label": "2313.HK", "name": "申洲国际", "market": "HK", "role": "纺织服装制造出海", "href": "/investor/stocks/2313.HK/" }, { "id": "term-8a9779f421", "type": "term", "label": "Chinese", "href": "/investor/research-network/?q=Chinese" }, { "id": "term-d46aee08cc", "type": "term", "label": "export", "href": "/investor/research-network/?q=export" }, { "id": "term-039d6add7e", "type": "term", "label": "manufacturing", "href": "/investor/research-network/?q=manufacturing" }, { "id": "event-d85c8e0558f0", "type": "event", "label": "[auto-handoff] Impact of global capacity pricing spikes on domestic transfo…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-5aef3dfcde02", "type": "event", "label": "[auto-handoff] 评估PJM等容量电价暴涨对国内出海变压器毛利率的传导", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-bf2680c00021", "type": "event", "label": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-po…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-28589741865e", "type": "event", "label": "2026年油价/柴油冲击对中国的双重影响：进口成本与通胀压力 vs. 能源安全与清洁制造的战略加速", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c900e6aa9594", "type": "event", "label": "新动能（AI基础设施与绿电）在体量上仅能弥补房地产投资收缩缺口的约10%-15%，且面临严重的电网并网、输电设备及芯片良率等物理瓶颈限制，无法在短期内…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-fad23caaa8d2", "type": "event", "label": "需求端验证：中国制造业PMI与工业数据对金属品种价格驱动力的分析", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-5c140a724456", "type": "event", "label": "海缆仍是订单背书最强的底仓但需下修新增订单毛利假设；储能是政策背书装机而非利润背书，只有高可靠、长时化、采购纪律强的龙头能把114号文转化为毛利；UD…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-03abbfec27f4", "type": "event", "label": "算力密度激增下的能源供给适配：从“规模化能源需求”到“极端局部负载调控”", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "thesis-485a630a87", "type": "thesis", "label": "自动驾驶与 Robotaxi 主题必须从演示热度进入单位经济和监管城市扩张验证", "conviction": 71, "direction": "conflicting", "lane": "自动驾驶/Robotaxi 单位经济", "laneHref": "/investor/lanes/autonomous-driving-and-robotaxi-unit-economics/", "bucket": "AI 基础设施", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 62.9, "href": "/investor/theses/thesis-485a630a87/", "networkHref": "/investor/research-network/#thesis-485a630a87" }, { "id": "stock-cf9e2f2a16", "type": "stock", "label": "XPEV", "name": "XPeng", "market": "US ADR", "role": "智能驾驶量产", "href": "/investor/stocks/XPEV/" }, { "id": "stock-46646e9d5e", "type": "stock", "label": "LI", "name": "Li Auto", "market": "US ADR", "role": "智能电动车规模", "href": "/investor/stocks/LI/" }, { "id": "term-aa03f5c14b", "type": "term", "label": "Robotaxi", "href": "/investor/research-network/?q=Robotaxi" }, { "id": "term-a8aa1f6bfe", "type": "term", "label": "Autonomous", "href": "/investor/research-network/?q=Autonomous" }, { "id": "term-d9c1acf117", "type": "term", "label": "driving", "href": "/investor/research-network/?q=driving" }, { "id": "term-7f48817c8f", "type": "term", "label": "robotaxi", "href": "/investor/research-network/?q=robotaxi" }, { "id": "term-ff582559c3", "type": "term", "label": "themes", "href": "/investor/research-network/?q=themes" }, { "id": "event-066d7522ea41", "type": "event", "label": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-po…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d6a17ab2f3c8", "type": "event", "label": "AI 基础设施的瓶颈：从 GPU 扩展到网络、电力与散热", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b782271763dc", "type": "event", "label": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-02513136b1da", "type": "event", "label": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-32a91fe3b28e", "type": "event", "label": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e35d7d944abc", "type": "event", "label": "[auto-handoff] Cloud customer acceptance of higher memory prices", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-5e257d42b8fc", "type": "event", "label": "AI 基础设施主线成立，但短期核心约束不是资本开支意愿，而是电力交付速度与区域电网承载能力。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b1c2d14dbd06", "type": "event", "label": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-4074d8a3d2", "type": "thesis", "label": "碳价、可再生能源与储能 CapEx 正在重塑电力资产回报", "conviction": 70, "direction": "conflicting", "lane": "能源转型/储能光伏", "laneHref": "/investor/lanes/energy-transition-storage-and-solar-capex/", "bucket": "电力与公用事业", "actionLabel": "分歧主线：配对验证", "actionTone": "yellow", "practicalScore": 63.8, "href": "/investor/theses/thesis-4074d8a3d2/", "networkHref": "/investor/research-network/#thesis-4074d8a3d2" }, { "id": "stock-a5d1d55fae", "type": "stock", "label": "BEP", "name": "Brookfield Renewable", "market": "US", "role": "可再生资产平台", "href": "/investor/stocks/BEP/" }, { "id": "stock-d7899bd385", "type": "stock", "label": "FSLR", "name": "First Solar", "market": "US", "role": "光伏制造", "href": "/investor/stocks/FSLR/" }, { "id": "stock-c338965e2a", "type": "stock", "label": "ENPH", "name": "Enphase Energy", "market": "US", "role": "逆变器/户储", "href": "/investor/stocks/ENPH/" }, { "id": "stock-78b3cb703d", "type": "stock", "label": "300274.SZ", "name": "阳光电源", "market": "A-share", "role": "逆变器与储能", "href": "/investor/stocks/300274.SZ/" }, { "id": "term-fab331127a", "type": "term", "label": "可再生能源", "href": "/investor/research-network/?q=%E5%8F%AF%E5%86%8D%E7%94%9F%E8%83%BD%E6%BA%90" }, { "id": "event-3cf09b5d78d2", "type": "event", "label": "电力侧隐性税对AIDC是一笔可被储能/配电对冲掉的小额价格税（约占TCO 0.2%），但真正决定AIDC收益率的是无法用电池对冲的time-to-po…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-626f451117f0", "type": "event", "label": "AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-aad4e81c7a0e", "type": "event", "label": "并网延迟确实压缩能源链利润预期，但损伤是分化而非单向的：发电/绿电设备承担约-21%经营杠杆压缩，T&D/电网设备则受5万亿元资本开支周期净托举，全链…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-ea94fc3b699d", "type": "event", "label": "海上风电+UDC的24/7低碳电力合同与并网可行性", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-40d8f3792517", "type": "event", "label": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a496ad79b63c", "type": "event", "label": "UDC的低PUE价值成立，但“海上风电直供”只有在叠加电网/储能/稳定PPA后才可支撑大规模AI算力的连续绿电需求。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7e616a7da726", "type": "event", "label": "PQC 引发的周期性热降频会显著提高超大规模 AI 集群 PUE/WUE 与能源成本预测难度，并在集中接入区增加电力调度复杂度，但它是边际波动因子而非…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-b61bc39c77eb", "type": "event", "label": "能源成本与高利率壁垒对全球算力中心电力建设与资本开支的传导效应", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-aa97a5c1b3", "type": "thesis", "label": "家装链的胜负手在存量房翻新、国补触达与地产后周期库存去化", "conviction": 69, "direction": "conflicting", "lane": "家装/地产后周期", "laneHref": "/investor/lanes/home-improvement-and-post-property-cycle/", "bucket": "宏观利率与汇率", "actionLabel": "小仓验证", "actionTone": "yellow", "practicalScore": 63.8, "href": "/investor/theses/thesis-aa97a5c1b3/", "networkHref": "/investor/research-network/#thesis-aa97a5c1b3" }, { "id": "stock-165e1b016a", "type": "stock", "label": "603833.SH", "name": "欧派家居", "market": "A-share", "role": "定制家居龙头", "href": "/investor/stocks/603833.SH/" }, { "id": "stock-058c86ec94", "type": "stock", "label": "603816.SH", "name": "顾家家居", "market": "A-share", "role": "软体家居", "href": "/investor/stocks/603816.SH/" }, { "id": "stock-60562db099", "type": "stock", "label": "002508.SZ", "name": "老板电器", "market": "A-share", "role": "厨电地产后周期", "href": "/investor/stocks/002508.SZ/" }, { "id": "stock-b0e0be6bf7", "type": "stock", "label": "002572.SZ", "name": "索菲亚", "market": "A-share", "role": "定制家居", "href": "/investor/stocks/002572.SZ/" }, { "id": "term-3a78695388", "type": "term", "label": "Home", "href": "/investor/research-network/?q=Home" }, { "id": "term-62a4a0e9fb", "type": "term", "label": "improvement", "href": "/investor/research-network/?q=improvement" }, { "id": "term-dae040fb63", "type": "term", "label": "existing-home", "href": "/investor/research-network/?q=existing-home" }, { "id": "term-92a9fcf1f2", "type": "term", "label": "renovation", "href": "/investor/research-network/?q=renovation" }, { "id": "term-15fc7418ed", "type": "term", "label": "subsidy", "href": "/investor/research-network/?q=subsidy" }, { "id": "term-ed4561f685", "type": "term", "label": "reach", "href": "/investor/research-network/?q=reach" }, { "id": "term-fdaad140aa", "type": "term", "label": "post-property", "href": "/investor/research-network/?q=post-property" }, { "id": "event-b940362e188a", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6b437d4ac75b", "type": "event", "label": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9dd4b2d68135", "type": "event", "label": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6117b0d1745e", "type": "event", "label": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d1e05b2add41", "type": "event", "label": "[auto-handoff] Cloud customer acceptance of higher memory prices", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-29e9007c9bc8", "type": "event", "label": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-643ce47583bf", "type": "event", "label": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9523a9be4503", "type": "event", "label": "产业链调研：新一代信息技术（30 日激活率 4.3%, 4/92）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-c1d3acab38", "type": "thesis", "label": "大健康/生物科技：支付、监管与效率改善验证", "conviction": 69, "direction": "conflicting", "lane": "消费健康防御", "laneHref": "/investor/lanes/lane-d104d465/", "bucket": "消费健康防御", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 62.5, "href": "/investor/theses/thesis-c1d3acab38/", "networkHref": "/investor/research-network/#thesis-c1d3acab38" }, { "id": "term-e54819251d", "type": "term", "label": "大健康产业链", "href": "/investor/research-network/?q=%E5%A4%A7%E5%81%A5%E5%BA%B7%E4%BA%A7%E4%B8%9A%E9%93%BE" }, { "id": "event-b3cdd1fa4f78", "type": "event", "label": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9f6a2c4784dd", "type": "event", "label": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-42928c33310a", "type": "event", "label": "大健康产业链处于‘被忽视定价’阶段，基本面反转已现，应围绕出海BD与高端国产替代定向深挖五只低覆盖中盘股，防范地缘立法与科创去杠杆等跨链外溢。", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7aeae1943dff", "type": "event", "label": "产业链调研：大健康产业链（30 日激活率 5.7%, 16/280）", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-315e1eccf479", "type": "event", "label": "截至2026年6月24日，大健康产业链在低估值与出海催化下步入业绩驱动重估期，应重点布局具备海外本土化渠道与跨链条多元化能力的高壁垒医疗器械及仿制药注…", "date": "2026-06-23", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-3f6dcc96f8c5", "type": "event", "label": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "date": "2026-06-21", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-d1c60d5af6", "type": "thesis", "label": "纺织服装出口链的 Alpha 取决于棉价、库存周期、品牌折扣与合规转移", "conviction": 67, "direction": "conflicting", "lane": "纺织服装/出口库存周期", "laneHref": "/investor/lanes/textile-apparel-export-and-inventory-cycle/", "bucket": "电力与公用事业", "actionLabel": "小仓验证", "actionTone": "yellow", "practicalScore": 61.6, "href": "/investor/theses/thesis-d1c60d5af6/", "networkHref": "/investor/research-network/#thesis-d1c60d5af6" }, { "id": "stock-83eb33367d", "type": "stock", "label": "2020.HK", "name": "安踏体育", "market": "HK", "role": "运动服饰品牌", "href": "/investor/stocks/2020.HK/" }, { "id": "stock-7e9ccaea2d", "type": "stock", "label": "2331.HK", "name": "李宁", "market": "HK", "role": "运动服饰需求", "href": "/investor/stocks/2331.HK/" }, { "id": "stock-73e0276fbb", "type": "stock", "label": "002293.SZ", "name": "罗莱生活", "market": "A-share", "role": "家纺消费", "href": "/investor/stocks/002293.SZ/" }, { "id": "stock-136926fb0c", "type": "stock", "label": "600398.SH", "name": "海澜之家", "market": "A-share", "role": "大众服装渠道", "href": "/investor/stocks/600398.SH/" }, { "id": "stock-2446d5d7e6", "type": "stock", "label": "NKE", "name": "Nike", "market": "US", "role": "全球运动品牌对照", "href": "/investor/stocks/NKE/" }, { "id": "term-90115ae3fa", "type": "term", "label": "Textile", "href": "/investor/research-network/?q=Textile" }, { "id": "term-646b1045ad", "type": "term", "label": "apparel", "href": "/investor/research-network/?q=apparel" }, { "id": "event-01624df5c8d4", "type": "event", "label": "[auto-handoff] Impact of global capacity pricing spikes on domestic transfo…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-f18a3a30a8d9", "type": "event", "label": "[auto-handoff] 先进制程星载 OBP 芯片良率瓶颈与算力/通信设备材料供应链", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a9dff3ae63f0", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-dd7ab6f0c052", "type": "event", "label": "电网与变压器瓶颈对AI算力扩张上限的实质性约束评估", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c3c86d9d6ea0", "type": "event", "label": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c29fdef3aa04", "type": "event", "label": "截至2026-06-28，最优跨资产表达是“久期+稀缺股票”的杠铃，铜只做长周期期权，因为近端精炼过剩削弱了直接商品敞口的性价比，而GOES稀缺主要只…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-f3ea290b8c6e", "type": "event", "label": "关键矿产与设备出口管制风险对A股高端环节溢价逻辑的冲击评估", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a0f5d7a65e45", "type": "event", "label": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-e26ef69f73", "type": "thesis", "label": "风险核查与证据质量是主题交易的必要过滤器", "conviction": 67, "direction": "conflicting", "lane": "组合风控/证据质量", "laneHref": "/investor/lanes/portfolio-risk-and-evidence-quality/", "bucket": "宏观利率与汇率", "actionLabel": "小仓验证", "actionTone": "yellow", "practicalScore": 62.3, "href": "/investor/theses/thesis-e26ef69f73/", "networkHref": "/investor/research-network/#thesis-e26ef69f73" }, { "id": "stock-68d0766ae1", "type": "stock", "label": "SH", "name": "ProShares Short S&P500", "market": "US ETF", "role": "组合保护工具", "href": "/investor/stocks/SH/" }, { "id": "term-0711a8d636", "type": "term", "label": "Risk", "href": "/investor/research-network/?q=Risk" }, { "id": "term-4183b7793f", "type": "term", "label": "verification", "href": "/investor/research-network/?q=verification" }, { "id": "term-ee8250fb76", "type": "term", "label": "evidence", "href": "/investor/research-network/?q=evidence" }, { "id": "term-acef2c15bc", "type": "term", "label": "quality", "href": "/investor/research-network/?q=quality" }, { "id": "term-80ea8bc4ba", "type": "term", "label": "necessary", "href": "/investor/research-network/?q=necessary" }, { "id": "event-4f58364c16fb", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-04bad6093f6b", "type": "event", "label": "[auto-handoff] 霍尔木兹海峡石油冲击与核心CPI传导", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-0ff1e6510837", "type": "event", "label": "[auto-handoff] 推导 PBOC 隔夜逆回购实际中标利率及中债收益率曲线非线性久期损失风险", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1be6fd7b9c94", "type": "event", "label": "[auto-handoff] Impact of global capacity pricing spikes on domestic transfo…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-d8df41c4ac49", "type": "event", "label": "Warsh 时代美联储的鹰派转向：6月点阵图从降息翻转为加息", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7700f4fc2fa9", "type": "event", "label": "主要央行保持鹰派按兵不动并非过度收紧的政策错误，而是应对由服务业传导支撑的核心PCE粘性（预计H2维持在3.1%-3.4%）的必要防御，但这会使实体经…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d2eb5a4be6b4", "type": "event", "label": "核心服务与住房通胀的粘性：美联储政策的后续驱动力", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-428edb25e361", "type": "event", "label": "6月点阵图不是简单过时，而是可能“对得不是理由”：能源理由过时，AI/电网与折旧通道支持更高更久，但不足以单独证明立即加息。", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-d6b2c23ed6", "type": "thesis", "label": "航运与物流成本是油价、关税和供应链再通胀的二阶传导器", "conviction": 66, "direction": "conflicting", "lane": "航运物流/再通胀传导", "laneHref": "/investor/lanes/shipping-logistics-and-reflation-transmission/", "bucket": "宏观利率与汇率", "actionLabel": "小仓验证", "actionTone": "yellow", "practicalScore": 58.7, "href": "/investor/theses/thesis-d6b2c23ed6/", "networkHref": "/investor/research-network/#thesis-d6b2c23ed6" }, { "id": "stock-912ff4d815", "type": "stock", "label": "1919.HK", "name": "中远海控", "market": "HK", "role": "集运运价", "href": "/investor/stocks/1919.HK/" }, { "id": "stock-f0875559d8", "type": "stock", "label": "600026.SH", "name": "中远海能", "market": "A-share", "role": "油轮运价", "href": "/investor/stocks/600026.SH/" }, { "id": "stock-6a65aeb1ac", "type": "stock", "label": "600018.SH", "name": "上港集团", "market": "A-share", "role": "港口吞吐", "href": "/investor/stocks/600018.SH/" }, { "id": "stock-ca7ea4dbbb", "type": "stock", "label": "ZIM", "name": "ZIM Integrated Shipping", "market": "US", "role": "集运高 beta", "href": "/investor/stocks/ZIM/" }, { "id": "stock-afb6f77a11", "type": "stock", "label": "MATX", "name": "Matson", "market": "US", "role": "美国航运对照", "href": "/investor/stocks/MATX/" }, { "id": "term-c75a1e81fa", "type": "term", "label": "IPP", "href": "/investor/research-network/?q=IPP" }, { "id": "term-ad5751430e", "type": "term", "label": "Shipping", "href": "/investor/research-network/?q=Shipping" }, { "id": "term-8880894de4", "type": "term", "label": "logistics", "href": "/investor/research-network/?q=logistics" }, { "id": "term-bcde6928e5", "type": "term", "label": "costs", "href": "/investor/research-network/?q=costs" }, { "id": "term-3fbb9ba69d", "type": "term", "label": "second-order", "href": "/investor/research-network/?q=second-order" }, { "id": "term-1d2cafdb2f", "type": "term", "label": "transmission", "href": "/investor/research-network/?q=transmission" }, { "id": "event-cc15456dfa69", "type": "event", "label": "[auto-handoff] Hormuz shock & core CPI transmission", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-126f81128f82", "type": "event", "label": "[auto-handoff] Impact of global capacity pricing spikes on domestic transfo…", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-5c3e4e51ef3c", "type": "event", "label": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-a5e874116bfd", "type": "event", "label": "AI算力基建瓶颈与多资产杠铃策略的逻辑闭环及定量模型终期质量审计", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-29e06d542d19", "type": "event", "label": "中国GOES与铜体系在基准情形下是西方AI建设的泄压阀，但政策阀门已被安装，西方数据中心应把中国来源依赖重定价为150-300bp的分层资本开支风险溢…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1c89e048cdec", "type": "event", "label": "在中国的精铜结构性进口依存与紫金海外资产高地缘折价背景下，应将有色敞口拆分为“在岸GOES技术壁垒（宝钢股份）+离岸实物/矿山多头”以规避紫金主权尾部…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7acd89597314", "type": "event", "label": "原始40/32/12/8/8 sleeve在AI需求相关性趋近+1时尾部风险过高，阶段1是强制去杠杆，阶段2才接近可接受，但决定性保护来自降AI be…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9298e4c5f04c", "type": "event", "label": "变压器交期拉长至3-5年与并网排队超36个月的物理刚性瓶颈，配合美欧贸易保护关税对GOES硅钢供应链的阻断，已对AI算力扩张的通电效率构成无法用资本单…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-62d39d2c51", "type": "thesis", "label": "家装链：地产后周期需求与消费修复验证", "conviction": 52, "direction": "conflicting", "lane": "消费健康防御", "laneHref": "/investor/lanes/lane-d104d465/", "bucket": "消费健康防御", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 42, "href": "/investor/theses/thesis-62d39d2c51/", "networkHref": "/investor/research-network/#thesis-62d39d2c51" }, { "id": "term-09c5d97f30", "type": "term", "label": "新华财经-家装产业链", "href": "/investor/research-network/?q=%E6%96%B0%E5%8D%8E%E8%B4%A2%E7%BB%8F-%E5%AE%B6%E8%A3%85%E4%BA%A7%E4%B8%9A%E9%93%BE" }, { "id": "event-a9f274a080c2", "type": "event", "label": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-655be6d521e1", "type": "event", "label": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-8de63be2f764", "type": "event", "label": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-5c2835c75fa9", "type": "event", "label": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "date": "2026-06-24", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-b472535e274f", "type": "event", "label": "产业链调研：新华财经-家装产业链（30 日激活率 10%, 20/201）", "date": "2026-06-22", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d9ad604ec93c", "type": "event", "label": "产业链调研：新华财经-家装产业链（30 日激活率 8.5%, 17/201）", "date": "2026-06-21", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-90ccc2857a", "type": "thesis", "label": "酒店、餐馆与休闲产业链：研究活跃度与投资线索复盘", "conviction": 51, "direction": "conflicting", "lane": "消费健康防御", "laneHref": "/investor/lanes/lane-d104d465/", "bucket": "消费健康防御", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 44.9, "href": "/investor/theses/thesis-90ccc2857a/", "networkHref": "/investor/research-network/#thesis-90ccc2857a" }, { "id": "term-59322eebea", "type": "term", "label": "酒店", "href": "/investor/research-network/?q=%E9%85%92%E5%BA%97" }, { "id": "term-b373cf5864", "type": "term", "label": "餐馆与休闲产业链", "href": "/investor/research-network/?q=%E9%A4%90%E9%A6%86%E4%B8%8E%E4%BC%91%E9%97%B2%E4%BA%A7%E4%B8%9A%E9%93%BE" }, { "id": "event-c4a1b906248e", "type": "event", "label": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-997d29f16724", "type": "event", "label": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "date": "2026-06-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-5cfdf6e95a9d", "type": "event", "label": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-9fd091300006", "type": "event", "label": "产业链调研：酒店、餐馆与休闲产业链（30 日激活率 9.3%, 9/97）", "date": "2026-06-23", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-e842406d3e", "type": "thesis", "label": "海洋经济：政策催化、装备订单与资源开发复盘", "conviction": 42, "direction": "conflicting", "lane": "AI 基础设施", "laneHref": "/investor/lanes/ai/", "bucket": "AI 基础设施", "actionLabel": "深研候选：先建观察仓", "actionTone": "orange", "practicalScore": 31.3, "href": "/investor/theses/thesis-e842406d3e/", "networkHref": "/investor/research-network/#thesis-e842406d3e" }, { "id": "term-ed2813a3f3", "type": "term", "label": "新华财经-海洋经济产业链", "href": "/investor/research-network/?q=%E6%96%B0%E5%8D%8E%E8%B4%A2%E7%BB%8F-%E6%B5%B7%E6%B4%8B%E7%BB%8F%E6%B5%8E%E4%BA%A7%E4%B8%9A%E9%93%BE" }, { "id": "event-747a0fe632b3", "type": "event", "label": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "date": "2026-06-29", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-92a206214f1c", "type": "event", "label": "海洋经济产业链处于低覆盖及严重低估水平，随着深海海风建设加速与AI基础设施冷却革命推进，亚星锚链、中远海特、海力风电与海兰信等核心标的迎来基本面共振与…", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-c095ebccc12d", "type": "event", "label": "产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）", "date": "2026-06-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-aeeaa3614b10", "type": "event", "label": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "date": "2026-06-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ee3e7e79a6e3", "type": "event", "label": "产业链调研：新华财经-海洋经济产业链（30 日激活率 5.4%, 8/148）", "date": "2026-06-23", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-1c346b645f54", "type": "event", "label": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "date": "2026-06-22", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-6cda8788cf80", "type": "event", "label": "产业链调研：新华财经-海洋经济产业链（30 日激活率 4.7%, 7/148）", "date": "2026-06-21", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-7f434ae738e1", "type": "event", "label": "新华财经-海洋经济产业链整体覆盖度低，但在深远海风电加速与全球港口自动化浪潮下，建议超配具备核心技术壁垒及订单反转支撑的亚星锚链与润邦股份。", "date": "2026-06-21", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-bdb5af1d34", "type": "thesis", "label": "纺织服装：出口利润率、合规与库存周期复盘", "conviction": 37, "direction": "conflicting", "lane": "宏观利率与汇率", "laneHref": "/investor/lanes/lane-cdafce64/", "bucket": "宏观利率与汇率", "actionLabel": "主题雷达：先建观察仓", "actionTone": "blue", "practicalScore": 29.3, "href": "/investor/theses/thesis-bdb5af1d34/", "networkHref": "/investor/research-network/#thesis-bdb5af1d34" }, { "id": "term-8bc3ed571c", "type": "term", "label": "纺织服装产业链", "href": "/investor/research-network/?q=%E7%BA%BA%E7%BB%87%E6%9C%8D%E8%A3%85%E4%BA%A7%E4%B8%9A%E9%93%BE" }, { "id": "event-83444b351ffc", "type": "event", "label": "纺织服装链最有价值的研究不是追总量复苏，而是寻找成本传导、海外产能和品牌经营杠杆已经被2025年及2026年一季度数据验证的标的。", "date": "2026-06-02", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-73e26e45522f", "type": "event", "label": "金融基础设施不再是中性工具，而是通过中信保信贷、绿色供应链金融和跨境信贷加速了纺织链的优胜劣汰。", "date": "2026-05-30", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-58fce7fe9795", "type": "event", "label": "纺织服装产业总量增长平稳的表象下正发生深层业绩分化，具备海外成熟产能布局、高技术壁垒辅料重服务，或定位高端黏性细分客群的中坚企业，将在行业补库与关税重…", "date": "2026-05-28", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-2ac4520a3440", "type": "event", "label": "纺织服装产业链当前不具备宏观贝塔进攻属性，整体定价权仍在走弱但收敛；只有聚酯长丝、国产运动鞋服、印染头部三个结构性强环节值得主题型超配，宏观拐点由关税…", "date": "2026-05-27", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-1eaea1ba4cc4", "type": "event", "label": "纺织服装产业链的宏观定价权已从下游 OEM 转移至具备‘反内卷’共识的上游聚酯龙头及具备‘绿色法典’护城河的印染巨头。", "date": "2026-05-27", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-e4bc0ccea7cd", "type": "event", "label": "Research note 01 几乎未产出实质研究，但其执行失误本身正是可操作的运营信号；本卡发出 7 条提案（4 safe + 3 review）…", "date": "2026-05-27", "sourceLabel": "研究记录", "supportType": "supports", "href": "" }, { "id": "event-7eee4c3fa72a", "type": "event", "label": "能源行业分析报告：纺织服装产业链的“能源强度陷阱” —— 挑战上游功能性材料的安全垫逻辑 报告日期： 2026 05 26 分析师： 能源行业分析师 …", "date": "2026-05-26", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-f7d44c40bed3", "type": "event", "label": "纺织服装产业链仍可投，但 A 股表达必须从低利润率出口 OEM beta 迁移到功能材料、内需渠道、合规数字化与具备 AI 小单快反能力的质量制造。", "date": "2026-05-26", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "thesis-7b551fae18", "type": "thesis", "label": "新一代信息技术：国产替代与算力应用落地复盘", "conviction": 35, "direction": "conflicting", "lane": "AI 基础设施", "laneHref": "/investor/lanes/ai/", "bucket": "AI 基础设施", "actionLabel": "主题雷达：先建观察仓", "actionTone": "blue", "practicalScore": 24, "href": "/investor/theses/thesis-7b551fae18/", "networkHref": "/investor/research-network/#thesis-7b551fae18" }, { "id": "term-001e6e7264", "type": "term", "label": "新一代信息技术产业链", "href": "/investor/research-network/?q=%E6%96%B0%E4%B8%80%E4%BB%A3%E4%BF%A1%E6%81%AF%E6%8A%80%E6%9C%AF%E4%BA%A7%E4%B8%9A%E9%93%BE" }, { "id": "event-c06ebe879665", "type": "event", "label": "随着AI芯片向高速与高功耗演进，硬件瓶颈已向高速铜连、芯片电感和硅光封测迁移，精选利基中盘标的具备更强超额收益潜力。", "date": "2026-06-23", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-03b15b562e03", "type": "event", "label": "在2026年中美云厂商CapEx受阻于电力红线以及大模型推理通缩的双重催化下，新一代信息技术产业链估值正向能耗比占优的定制ASIC上游填料（联瑞新材）…", "date": "2026-06-22", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-d89cc7f5eb98", "type": "event", "label": "新一代信息技术链条的最佳机会在 AI 基础设施“卖铲人”和端侧 AI 芯片，而不是拥挤的通用 AI 服务器敞口。", "date": "2026-06-21", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-ba11427e9b8f", "type": "event", "label": "新一代信息技术链的超额收益已从泛算力概念转向‘物理硬约束（电力/连接）组件出海’与‘底层芯片（GPU/MCU/MEMS）国产替代业绩拐点验证’，需警惕…", "date": "2026-06-12", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-fd0ad93ec81a", "type": "event", "label": "指派 tmt-analyst 负责新一代信息技术产业链", "date": "2026-06-11", "sourceLabel": "研究记录", "supportType": "neutral", "href": "" }, { "id": "event-7cecf5e2e17c", "type": "event", "label": "在云端AI面临电力并网物理瓶颈及高两融去杠杆压力下，精选低拥挤度、高壁垒、进入全球核心供应链或处于国产替代关键爆发期的A股半导体上游特种材料与精密零部…", "date": "2026-06-10", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-75d6a9df8dcc", "type": "event", "label": "云端AI面临电力并网瓶颈，A股TMT板块杠杆面临去杠杆挤压，资金应系统性避开拥挤硬件、超配联瑞新材、鼎龙股份、雅克科技与富创精密等具备高增长刚性与低拥…", "date": "2026-06-08", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-291410b229d3", "type": "event", "label": "通过构建‘中游高端PCB确定性底仓+上游卡脖子材料（低Dk布、HVLP铜箔、特种树脂）国产替代期权’的杠铃型组合，能有效对冲产业链‘电力墙’与原材料成…", "date": "2026-06-01", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "thesis-9953a5d24c", "type": "thesis", "label": "工业机械：出口订单质量与资本开支弹性复盘", "conviction": 31, "direction": "conflicting", "lane": "AI 基础设施", "laneHref": "/investor/lanes/ai/", "bucket": "AI 基础设施", "actionLabel": "主题雷达：先建观察仓", "actionTone": "blue", "practicalScore": 17.4, "href": "/investor/theses/thesis-9953a5d24c/", "networkHref": "/investor/research-network/#thesis-9953a5d24c" }, { "id": "term-e34433e57e", "type": "term", "label": "工业机械产业链", "href": "/investor/research-network/?q=%E5%B7%A5%E4%B8%9A%E6%9C%BA%E6%A2%B0%E4%BA%A7%E4%B8%9A%E9%93%BE" }, { "id": "event-44b5fceb331f", "type": "event", "label": "白板研究卡片 03 人工智能与工业自动化对电力需求的冲击 作者 : 能源行业分析师 (energy analyst) 报告日期 : 2026 06 0…", "date": "2026-06-06", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-246ef16cc28b", "type": "event", "label": "工业机械投资逻辑已从“规模红利”转向“十五五”规划下的“全栈国产化”能力溢价，重点关注半导体上游材料与具身智能精密组件。", "date": "2026-06-05", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-23b9a69c5925", "type": "event", "label": "工业机械产业链的真正机会不在大盘股 beta，而在被低关注的供应节点——高精度减速器、液压/直线执行器、数据中心机电承包商、SMR/国防隐性受益者；但…", "date": "2026-06-01", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-94e253b25112", "type": "event", "label": "上游材料瓶颈（特别是取向硅钢和高性能磁材）将使工业机械产业链从‘量增’转为‘价涨’逻辑，垂直整合能力成为核心估值溢价项。", "date": "2026-06-01", "sourceLabel": "研究记录", "supportType": "risk", "href": "" }, { "id": "event-5aff49fcc600", "type": "event", "label": "下一代周期中，Mag7 谁会输？", "date": "2026-05-24", "sourceLabel": "研究记录", "supportType": "risk", "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/" }, { "id": "event-9196b94b3020", "type": "event", "label": "AI 生产率去通胀：采用率、工作流重构与滞后风险", "date": "2026-05-23", "sourceLabel": "研究记录", "supportType": "risk", "href": "/investor/deep-research/ai-productivity-disinflation-timing-af09345c/" }, { "id": "event-1802016eedc9", "type": "event", "label": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "date": "2026-05-23", "sourceLabel": "研究记录", "supportType": "risk", "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/" }, { "id": "event-dc22a430c5c4", "type": "event", "label": "消费降级、银发经济与大健康结构性机会", "date": "2026-05-19", "sourceLabel": "研究记录", "supportType": "supports", "href": "/investor/deep-research/consumer-downgrade-health-silver-economy-78a2530a/" } ], "edges": [ { "id": "edge-7e9131863234", "source": "thesis-029ce03da1", "target": "stock-510de57b5f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-8d69506a2e76", "source": "thesis-029ce03da1", "target": "stock-68a288585c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-4b5168621d85", "source": "thesis-029ce03da1", "target": "stock-79f7c9313f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c616a5f26413", "source": "thesis-029ce03da1", "target": "stock-f57a081cf3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9c3c0eaed282", "source": "thesis-029ce03da1", "target": "stock-dcd4dbe710", "type": "tracks_stock", "weight": 3 }, { "id": "edge-401ed6784032", "source": "thesis-029ce03da1", "target": "stock-efe2ecdc5b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-708dc5544b92", "source": "thesis-029ce03da1", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-fdba4d9edb2e", "source": "thesis-029ce03da1", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-7ff0b6ea1754", "source": "thesis-029ce03da1", "target": "term-f71ccd3834", "type": "tracks_term", "weight": 2 }, { "id": "edge-d6efe9d9f4a4", "source": "thesis-029ce03da1", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-44f0b841b135", "source": "thesis-029ce03da1", "target": "term-a83d000110", "type": "tracks_term", "weight": 2 }, { "id": "edge-2b511d7a8839", "source": "thesis-029ce03da1", "target": "term-87faf34a67", "type": "tracks_term", "weight": 2 }, { "id": "edge-82178849f9b1", "source": "thesis-029ce03da1", "target": "term-340d600392", "type": "tracks_term", "weight": 2 }, { "id": "edge-e372606c2a21", "source": "thesis-029ce03da1", "target": "term-59b47e9ec5", "type": "tracks_term", "weight": 2 }, { "id": "edge-200fea54ae94", "source": "thesis-029ce03da1", "target": "event-65122927a93c", "type": "challenged_by", "weight": 1 }, { "id": "edge-da10aa9f45ee", "source": "thesis-029ce03da1", "target": "event-0dd67e976c54", "type": "challenged_by", "weight": 1 }, { "id": "edge-872c70c9faf8", "source": "thesis-029ce03da1", "target": "event-ec6fc35ebaf2", "type": "supported_by", "weight": 1 }, { "id": "edge-dee026aa0074", "source": "thesis-029ce03da1", "target": "event-067f4b063a10", "type": "challenged_by", "weight": 1 }, { "id": "edge-64c715217be3", "source": "thesis-029ce03da1", "target": "event-93399852c17e", "type": "challenged_by", "weight": 1 }, { "id": "edge-2f5b8f1855c9", "source": "thesis-029ce03da1", "target": "event-c29701b05aa0", "type": "challenged_by", "weight": 1 }, { "id": "edge-d11261de1cf3", "source": "thesis-029ce03da1", "target": "event-58532cf290f7", "type": "challenged_by", "weight": 1 }, { "id": "edge-20102604e6f0", "source": "thesis-029ce03da1", "target": "event-978bd9fef78b", "type": "challenged_by", "weight": 1 }, { "id": "edge-f2a4ffada221", "source": "thesis-8c5a9690a0", "target": "stock-8638f0279b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-91a2e4b816b5", "source": "thesis-8c5a9690a0", "target": "stock-1eb44d6252", "type": "tracks_stock", "weight": 3 }, { "id": "edge-780145024a64", "source": "thesis-8c5a9690a0", "target": "stock-2ff6985a97", "type": "tracks_stock", "weight": 3 }, { "id": "edge-29191f28851b", "source": "thesis-8c5a9690a0", "target": "stock-2d1edf729c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-bc400d1dc108", "source": "thesis-8c5a9690a0", "target": "stock-31ac18dc22", "type": "tracks_stock", "weight": 3 }, { "id": "edge-248889ac2213", "source": "thesis-8c5a9690a0", "target": "stock-a25fc665a4", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5601c8d8cc44", "source": "thesis-8c5a9690a0", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-af81bf6123e1", "source": "thesis-8c5a9690a0", "target": "term-c7bac46904", "type": "tracks_term", "weight": 2 }, { "id": "edge-a5f8e7a7da7c", "source": "thesis-8c5a9690a0", "target": "term-0890f45668", "type": "tracks_term", "weight": 2 }, { "id": "edge-fe0d30b2b47a", "source": "thesis-8c5a9690a0", "target": "term-49d2dc07ac", "type": "tracks_term", "weight": 2 }, { "id": "edge-f35d73bfb923", "source": "thesis-8c5a9690a0", "target": "term-f223faa96f", "type": "tracks_term", "weight": 2 }, { "id": "edge-eb403e7d41fe", "source": "thesis-8c5a9690a0", "target": "term-4f749de7c2", "type": "tracks_term", "weight": 2 }, { "id": "edge-9efb90a147a4", "source": "thesis-8c5a9690a0", "target": "term-f89f85382f", "type": "tracks_term", "weight": 2 }, { "id": "edge-c4b06e3bbfeb", "source": "thesis-8c5a9690a0", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-880eb2538113", "source": "thesis-8c5a9690a0", "target": "event-1c162e6fab24", "type": "supported_by", "weight": 1 }, { "id": "edge-bfb75b9217cb", "source": "thesis-8c5a9690a0", "target": "event-b642c7a92c30", "type": "challenged_by", "weight": 1 }, { "id": "edge-e6380f62e379", "source": "thesis-8c5a9690a0", "target": "event-66d0a71e1acb", "type": "challenged_by", "weight": 1 }, { "id": "edge-ba03a9927967", "source": "thesis-8c5a9690a0", "target": "event-b29076871103", "type": "challenged_by", "weight": 1 }, { "id": "edge-4fa74d050879", "source": "thesis-8c5a9690a0", "target": "event-7d4f797675b9", "type": "challenged_by", "weight": 1 }, { "id": "edge-94d65171977d", "source": "thesis-8c5a9690a0", "target": "event-623b2987baef", "type": "challenged_by", "weight": 1 }, { "id": "edge-007088ee2cdc", "source": "thesis-8c5a9690a0", "target": "event-99489c9c3778", "type": "challenged_by", "weight": 1 }, { "id": "edge-b9b200265ae3", "source": "thesis-8c5a9690a0", "target": "event-6d6c9f82dd3d", "type": "challenged_by", "weight": 1 }, { "id": "edge-d27c4489966e", "source": "thesis-c89650af9e", "target": "stock-083dc3a2b3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d79b285a5068", "source": "thesis-c89650af9e", "target": "stock-a6f4e763c1", "type": "tracks_stock", "weight": 3 }, { "id": "edge-60ac4a1803e3", "source": "thesis-c89650af9e", "target": "stock-8e7522f842", "type": "tracks_stock", "weight": 3 }, { "id": "edge-31a0b599aa70", "source": "thesis-c89650af9e", "target": "stock-dda1241d64", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a5995e7d180a", "source": "thesis-c89650af9e", "target": "stock-8904c61a98", "type": "tracks_stock", "weight": 3 }, { "id": "edge-57028e474dfb", "source": "thesis-c89650af9e", "target": "stock-ae1510493b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-951a782fb668", "source": "thesis-c89650af9e", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-3485a5b4f23a", "source": "thesis-c89650af9e", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-799ea24c417e", "source": "thesis-c89650af9e", "target": "term-a5e4d88f36", "type": "tracks_term", "weight": 2 }, { "id": "edge-14c67f2242f8", "source": "thesis-c89650af9e", "target": "term-b1a96dd646", "type": "tracks_term", "weight": 2 }, { "id": "edge-cccc7b3bbd80", "source": "thesis-c89650af9e", "target": "term-ae4dbd7e5a", "type": "tracks_term", "weight": 2 }, { "id": "edge-b5fc09bb38af", "source": "thesis-c89650af9e", "target": "term-5b6c989793", "type": "tracks_term", "weight": 2 }, { "id": "edge-fdbb499f76ff", "source": "thesis-c89650af9e", "target": "term-0dbdc4fb0f", "type": "tracks_term", "weight": 2 }, { "id": "edge-4c5e0f103557", "source": "thesis-c89650af9e", "target": "term-6e862940d0", "type": "tracks_term", "weight": 2 }, { "id": "edge-19529b8b80b5", "source": "thesis-c89650af9e", "target": "event-55aaf129b1a6", "type": "challenged_by", "weight": 1 }, { "id": "edge-b9851680c8d6", "source": "thesis-c89650af9e", "target": "event-d7b0c7f26ea4", "type": "challenged_by", "weight": 1 }, { "id": "edge-cd821cb421c8", "source": "thesis-c89650af9e", "target": "event-8eaaa6ce81bb", "type": "challenged_by", "weight": 1 }, { "id": "edge-556cd076d0ea", "source": "thesis-c89650af9e", "target": "event-9f1a40503848", "type": "challenged_by", "weight": 1 }, { "id": "edge-55429d4ab1cc", "source": "thesis-c89650af9e", "target": "event-8ce1fb804b20", "type": "challenged_by", "weight": 1 }, { "id": "edge-1df7901f7672", "source": "thesis-c89650af9e", "target": "event-7a6a011ea0cd", "type": "challenged_by", "weight": 1 }, { "id": "edge-3e4c713fc2d2", "source": "thesis-c89650af9e", "target": "event-acd322d6ba80", "type": "challenged_by", "weight": 1 }, { "id": "edge-873858e52e20", "source": "thesis-c89650af9e", "target": "event-8115c4f62184", "type": "challenged_by", "weight": 1 }, { "id": "edge-e3ad5f57e86e", "source": "thesis-f5d7a79c82", "target": "stock-cb15b056a5", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e6ae206eeec3", "source": "thesis-f5d7a79c82", "target": "stock-426219f2bc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6625fa01c288", "source": "thesis-f5d7a79c82", "target": "stock-a24c631b90", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b29e6e92ee95", "source": "thesis-f5d7a79c82", "target": "stock-1bcfecaf36", "type": "tracks_stock", "weight": 3 }, { "id": "edge-43cfd2226e92", "source": "thesis-f5d7a79c82", "target": "stock-7599fc757f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d6f7a82fb722", "source": "thesis-f5d7a79c82", "target": "stock-c908d13f30", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d511c67a9d20", "source": "thesis-f5d7a79c82", "target": "term-e4a3cbf326", "type": "tracks_term", "weight": 2 }, { "id": "edge-d8876abcfa61", "source": "thesis-f5d7a79c82", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-99cee7a1f6d2", "source": "thesis-f5d7a79c82", "target": "term-0890f45668", "type": "tracks_term", "weight": 2 }, { "id": "edge-051450fd604b", "source": "thesis-f5d7a79c82", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-c273e06287c3", "source": "thesis-f5d7a79c82", "target": "term-4de0127946", "type": "tracks_term", "weight": 2 }, { "id": "edge-e871caefd1de", "source": "thesis-f5d7a79c82", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-898611025c66", "source": "thesis-f5d7a79c82", "target": "term-36e3041f1c", "type": "tracks_term", "weight": 2 }, { "id": "edge-743442a9a615", "source": "thesis-f5d7a79c82", "target": "term-00fb5022b7", "type": "tracks_term", "weight": 2 }, { "id": "edge-e80be803e03e", "source": "thesis-f5d7a79c82", "target": "event-b7d4d00f15be", "type": "challenged_by", "weight": 1 }, { "id": "edge-3e868668e536", "source": "thesis-f5d7a79c82", "target": "event-369daa798bef", "type": "challenged_by", "weight": 1 }, { "id": "edge-5bab07d05cbb", "source": "thesis-f5d7a79c82", "target": "event-1b0710c91896", "type": "challenged_by", "weight": 1 }, { "id": "edge-d41a0917111a", "source": "thesis-f5d7a79c82", "target": "event-6be609154936", "type": "supported_by", "weight": 1 }, { "id": "edge-01bc852243f3", "source": "thesis-f5d7a79c82", "target": "event-2a65b011d40a", "type": "challenged_by", "weight": 1 }, { "id": "edge-f9782b83a457", "source": "thesis-f5d7a79c82", "target": "event-b7d81759dfae", "type": "challenged_by", "weight": 1 }, { "id": "edge-ee60b97f1106", "source": "thesis-f5d7a79c82", "target": "event-83ccc2c57dfe", "type": "challenged_by", "weight": 1 }, { "id": "edge-4d8e0d9d4a8b", "source": "thesis-f5d7a79c82", "target": "event-116e157e4170", "type": "challenged_by", "weight": 1 }, { "id": "edge-7ed489a7b5c2", "source": "thesis-8bbbcdf5a4", "target": "stock-426219f2bc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-478bb22efb79", "source": "thesis-8bbbcdf5a4", "target": "stock-0fd3e4e71b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-299e5f4d0dae", "source": "thesis-8bbbcdf5a4", "target": "stock-b78cca2401", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7dc851d7cd6d", "source": "thesis-8bbbcdf5a4", "target": "stock-cb15b056a5", "type": "tracks_stock", "weight": 3 }, { "id": "edge-1ce757020db3", "source": "thesis-8bbbcdf5a4", "target": "stock-c908d13f30", "type": "tracks_stock", "weight": 3 }, { "id": "edge-cf6bb32ae25e", "source": "thesis-8bbbcdf5a4", "target": "stock-a24c631b90", "type": "tracks_stock", "weight": 3 }, { "id": "edge-8c2cf75dd127", "source": "thesis-8bbbcdf5a4", "target": "term-e4a3cbf326", "type": "tracks_term", "weight": 2 }, { "id": "edge-a3990f543a9c", "source": "thesis-8bbbcdf5a4", "target": "term-d49897f8c4", "type": "tracks_term", "weight": 2 }, { "id": "edge-ab51aa6c99b0", "source": "thesis-8bbbcdf5a4", "target": "term-0890f45668", "type": "tracks_term", "weight": 2 }, { "id": "edge-38abceb6092a", "source": "thesis-8bbbcdf5a4", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-019953783291", "source": "thesis-8bbbcdf5a4", "target": "term-4de0127946", "type": "tracks_term", "weight": 2 }, { "id": "edge-7374523bb0f9", "source": "thesis-8bbbcdf5a4", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-95a8a407e741", "source": "thesis-8bbbcdf5a4", "target": "term-36e3041f1c", "type": "tracks_term", "weight": 2 }, { "id": "edge-4f49e50d784b", "source": "thesis-8bbbcdf5a4", "target": "term-00fb5022b7", "type": "tracks_term", "weight": 2 }, { "id": "edge-8baf9112e068", "source": "thesis-8bbbcdf5a4", "target": "event-f3345c652a8b", "type": "challenged_by", "weight": 1 }, { "id": "edge-782b6e6d8d25", "source": "thesis-8bbbcdf5a4", "target": "event-ee3eae4dc86c", "type": "challenged_by", "weight": 1 }, { "id": "edge-f27391b20ea3", "source": "thesis-8bbbcdf5a4", "target": "event-14cafc321f85", "type": "challenged_by", "weight": 1 }, { "id": "edge-6c1243fe860a", "source": "thesis-8bbbcdf5a4", "target": "event-bed6c0f6a1d4", "type": "challenged_by", "weight": 1 }, { "id": "edge-49ff05655ebc", "source": "thesis-8bbbcdf5a4", "target": "event-7d585e0aca75", "type": "challenged_by", "weight": 1 }, { "id": "edge-1c6ab63fbb0f", "source": "thesis-8bbbcdf5a4", "target": "event-8e6780d40d6d", "type": "challenged_by", "weight": 1 }, { "id": "edge-a2f2ea005e52", "source": "thesis-8bbbcdf5a4", "target": "event-86639eda6024", "type": "challenged_by", "weight": 1 }, { "id": "edge-df3d0185a98a", "source": "thesis-8bbbcdf5a4", "target": "event-aede264ef7c6", "type": "supported_by", "weight": 1 }, { "id": "edge-0fe6ff33a15b", "source": "thesis-ab1fb6a453", "target": "stock-434e501ece", "type": "tracks_stock", "weight": 3 }, { "id": "edge-15883551c87e", "source": "thesis-ab1fb6a453", "target": "stock-9cd0d546e6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-4b7ceb66400c", "source": "thesis-ab1fb6a453", "target": "stock-23cfa8bf26", "type": "tracks_stock", "weight": 3 }, { "id": "edge-dd4e400b256b", "source": "thesis-ab1fb6a453", "target": "stock-294da558b6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d1366d4f8ac6", "source": "thesis-ab1fb6a453", "target": "stock-efe2ecdc5b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5f02bb3eb679", "source": "thesis-ab1fb6a453", "target": "stock-8638f0279b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6e0e70ef89cf", "source": "thesis-ab1fb6a453", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-ea899f4f33bd", "source": "thesis-ab1fb6a453", "target": "term-c7bac46904", "type": "tracks_term", "weight": 2 }, { "id": "edge-d46b77c02612", "source": "thesis-ab1fb6a453", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-d504bf19b052", "source": "thesis-ab1fb6a453", "target": "term-0648d2111e", "type": "tracks_term", "weight": 2 }, { "id": "edge-6bd3038a4a55", "source": "thesis-ab1fb6a453", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-a0b5cdd10c20", "source": "thesis-ab1fb6a453", "target": "term-409d071901", "type": "tracks_term", "weight": 2 }, { "id": "edge-b259bccf869c", "source": "thesis-ab1fb6a453", "target": "term-f9b7ef8159", "type": "tracks_term", "weight": 2 }, { "id": "edge-15ca0794a41e", "source": "thesis-ab1fb6a453", "target": "term-a8a79412bd", "type": "tracks_term", "weight": 2 }, { "id": "edge-bf703f4cd221", "source": "thesis-ab1fb6a453", "target": "event-77234e3d1d38", "type": "supported_by", "weight": 1 }, { "id": "edge-430794ed7404", "source": "thesis-ab1fb6a453", "target": "event-84ce3f44a9b5", "type": "challenged_by", "weight": 1 }, { "id": "edge-1b2b1db482f6", "source": "thesis-ab1fb6a453", "target": "event-76ad8cbebc3a", "type": "challenged_by", "weight": 1 }, { "id": "edge-f1277bc49192", "source": "thesis-ab1fb6a453", "target": "event-4120371c3241", "type": "challenged_by", "weight": 1 }, { "id": "edge-661c42a6a40e", "source": "thesis-ab1fb6a453", "target": "event-d88aea428955", "type": "challenged_by", "weight": 1 }, { "id": "edge-84955ef63e4d", "source": "thesis-ab1fb6a453", "target": "event-0871091194c6", "type": "challenged_by", "weight": 1 }, { "id": "edge-bc10102f0768", "source": "thesis-ab1fb6a453", "target": "event-c1656bad07ad", "type": "supported_by", "weight": 1 }, { "id": "edge-f28384f7442f", "source": "thesis-ab1fb6a453", "target": "event-6bbca16040ea", "type": "challenged_by", "weight": 1 }, { "id": "edge-cb53aa6db833", "source": "thesis-6f42e0fe6e", "target": "stock-0d55e94f77", "type": "tracks_stock", "weight": 3 }, { "id": "edge-84a4f766ea39", "source": "thesis-6f42e0fe6e", "target": "stock-e87348f2ea", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a9c39011c792", "source": "thesis-6f42e0fe6e", "target": "stock-204ac66a88", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ae244e646dfc", "source": "thesis-6f42e0fe6e", "target": "stock-e5218dcd47", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c3a16747c73c", "source": "thesis-6f42e0fe6e", "target": "stock-a8f5b84c58", "type": "tracks_stock", "weight": 3 }, { "id": "edge-8882b70d0f17", "source": "thesis-6f42e0fe6e", "target": "stock-58bb119c35", "type": "tracks_stock", "weight": 3 }, { "id": "edge-04716e53d75d", "source": "thesis-6f42e0fe6e", "target": "term-4f749de7c2", "type": "tracks_term", "weight": 2 }, { "id": "edge-3052916f348b", "source": "thesis-6f42e0fe6e", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-255d937f5029", "source": "thesis-6f42e0fe6e", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-94fda4f3dd87", "source": "thesis-6f42e0fe6e", "target": "term-66e0f52800", "type": "tracks_term", "weight": 2 }, { "id": "edge-446e6acbde07", "source": "thesis-6f42e0fe6e", "target": "term-9196f0338a", "type": "tracks_term", "weight": 2 }, { "id": "edge-a31480ddf07d", "source": "thesis-6f42e0fe6e", "target": "term-58fdb0941f", "type": "tracks_term", "weight": 2 }, { "id": "edge-5c6ea97a2970", "source": "thesis-6f42e0fe6e", "target": "term-1232085ef0", "type": "tracks_term", "weight": 2 }, { "id": "edge-4b79363175c9", "source": "thesis-6f42e0fe6e", "target": "term-291d2a9756", "type": "tracks_term", "weight": 2 }, { "id": "edge-a16286908c33", "source": "thesis-6f42e0fe6e", "target": "event-28ff19bfc7b1", "type": "challenged_by", "weight": 1 }, { "id": "edge-d3b327bffe9a", "source": "thesis-6f42e0fe6e", "target": "event-22753c05c82d", "type": "challenged_by", "weight": 1 }, { "id": "edge-6412ca625c7b", "source": "thesis-6f42e0fe6e", "target": "event-0782375243df", "type": "challenged_by", "weight": 1 }, { "id": "edge-bc9f5ffdc580", "source": "thesis-6f42e0fe6e", "target": "event-83c2735e99f8", "type": "challenged_by", "weight": 1 }, { "id": "edge-964a2d896d18", "source": "thesis-6f42e0fe6e", "target": "event-b4345abe9c75", "type": "challenged_by", "weight": 1 }, { "id": "edge-84f8fab8441d", "source": "thesis-6f42e0fe6e", "target": "event-d1ed91148429", "type": "challenged_by", "weight": 1 }, { "id": "edge-fba928d0a8b8", "source": "thesis-6f42e0fe6e", "target": "event-1cd714441919", "type": "challenged_by", "weight": 1 }, { "id": "edge-29bfc46a2efc", "source": "thesis-6f42e0fe6e", "target": "event-c19fce5ea092", "type": "challenged_by", "weight": 1 }, { "id": "edge-d416c8abce6f", "source": "thesis-70daf87262", "target": "stock-57a896f1c0", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a509334c211a", "source": "thesis-70daf87262", "target": "stock-b9b86bd6b7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5dd375a6889d", "source": "thesis-70daf87262", "target": "stock-0d1a838e74", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ebb3298ee1fb", "source": "thesis-70daf87262", "target": "stock-f5c007a81f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c13f414de5e4", "source": "thesis-70daf87262", "target": "stock-532787880f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b85f3e4f0dd8", "source": "thesis-70daf87262", "target": "stock-c39cc2da75", "type": "tracks_stock", "weight": 3 }, { "id": "edge-18c56195f5d6", "source": "thesis-70daf87262", "target": "term-60e981724c", "type": "tracks_term", "weight": 2 }, { "id": "edge-426871b445a3", "source": "thesis-70daf87262", "target": "term-f71ccd3834", "type": "tracks_term", "weight": 2 }, { "id": "edge-dbe390c471e6", "source": "thesis-70daf87262", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-3d7ae50ced5f", "source": "thesis-70daf87262", "target": "term-a2e9329b92", "type": "tracks_term", "weight": 2 }, { "id": "edge-ddb55b2d917d", "source": "thesis-70daf87262", "target": "term-1582c27783", "type": "tracks_term", "weight": 2 }, { "id": "edge-87538cd3ebf9", "source": "thesis-70daf87262", "target": "term-a050178590", "type": "tracks_term", "weight": 2 }, { "id": "edge-6b37c7f5ad14", "source": "thesis-70daf87262", "target": "term-18126b597c", "type": "tracks_term", "weight": 2 }, { "id": "edge-c73891b90178", "source": "thesis-70daf87262", "target": "term-e21dcec424", "type": "tracks_term", "weight": 2 }, { "id": "edge-81875d8806b9", "source": "thesis-70daf87262", "target": "event-93176bb8ed58", "type": "supported_by", "weight": 1 }, { "id": "edge-c4413f9638ce", "source": "thesis-70daf87262", "target": "event-00b21e510b58", "type": "supported_by", "weight": 1 }, { "id": "edge-e0596e2405aa", "source": "thesis-70daf87262", "target": "event-6593e7d60e5a", "type": "supported_by", "weight": 1 }, { "id": "edge-324f21a32c44", "source": "thesis-70daf87262", "target": "event-8ccb66ee3b20", "type": "supported_by", "weight": 1 }, { "id": "edge-a320f0006fbf", "source": "thesis-70daf87262", "target": "event-79548bf69166", "type": "challenged_by", "weight": 1 }, { "id": "edge-c0743305dac8", "source": "thesis-70daf87262", "target": "event-8af7fff0ec70", "type": "supported_by", "weight": 1 }, { "id": "edge-9db2cbfa8652", "source": "thesis-70daf87262", "target": "event-c77f3d5fb20f", "type": "challenged_by", "weight": 1 }, { "id": "edge-ac43e2911be9", "source": "thesis-70daf87262", "target": "event-12d12096696b", "type": "challenged_by", "weight": 1 }, { "id": "edge-da064326a18c", "source": "thesis-58200390b4", "target": "stock-daf49f5a34", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7a7eb538f801", "source": "thesis-58200390b4", "target": "stock-339675eab4", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7be4e4f007c5", "source": "thesis-58200390b4", "target": "stock-edafde8d2d", "type": "tracks_stock", "weight": 3 }, { "id": "edge-baf5e62408dc", "source": "thesis-58200390b4", "target": "stock-d1303b4837", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0c5f96376b2d", "source": "thesis-58200390b4", "target": "stock-8185c648c8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9bacd30319dd", "source": "thesis-58200390b4", "target": "stock-a0cd4ecb4f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-cb57b48be41c", "source": "thesis-58200390b4", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-71360f89ec46", "source": "thesis-58200390b4", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-48b9771b9320", "source": "thesis-58200390b4", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-0003c195ca9b", "source": "thesis-58200390b4", "target": "term-df12b8f89b", "type": "tracks_term", "weight": 2 }, { "id": "edge-708393e9a0e6", "source": "thesis-58200390b4", "target": "term-d341073240", "type": "tracks_term", "weight": 2 }, { "id": "edge-63d0f5d64412", "source": "thesis-58200390b4", "target": "term-f81d83a742", "type": "tracks_term", "weight": 2 }, { "id": "edge-01ba8ed9f57a", "source": "thesis-58200390b4", "target": "term-092c79e8f8", "type": "tracks_term", "weight": 2 }, { "id": "edge-f0bce55ab190", "source": "thesis-58200390b4", "target": "term-8001c27439", "type": "tracks_term", "weight": 2 }, { "id": "edge-3e42dea73205", "source": "thesis-58200390b4", "target": "event-42ac0cf447d6", "type": "challenged_by", "weight": 1 }, { "id": "edge-67dea08a6532", "source": "thesis-58200390b4", "target": "event-0281e7ed0133", "type": "challenged_by", "weight": 1 }, { "id": "edge-066b63ff63b7", "source": "thesis-58200390b4", "target": "event-bdb6f3864771", "type": "challenged_by", "weight": 1 }, { "id": "edge-55f2e612ec19", "source": "thesis-58200390b4", "target": "event-c0284e20728f", "type": "challenged_by", "weight": 1 }, { "id": "edge-50bbc7dc2c42", "source": "thesis-58200390b4", "target": "event-4faeda0d18cb", "type": "challenged_by", "weight": 1 }, { "id": "edge-13bd40605df7", "source": "thesis-58200390b4", "target": "event-364c99289c3a", "type": "challenged_by", "weight": 1 }, { "id": "edge-b9a6cf585e80", "source": "thesis-58200390b4", "target": "event-3d2f033276ec", "type": "challenged_by", "weight": 1 }, { "id": "edge-a659d2831839", "source": "thesis-58200390b4", "target": "event-b53e365a948c", "type": "challenged_by", "weight": 1 }, { "id": "edge-cffbe2bb2863", "source": "thesis-8b8ede3135", "target": "stock-0716b17bb6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-cd6decc6a6e6", "source": "thesis-8b8ede3135", "target": "stock-736f8c20de", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5074ede30ede", "source": "thesis-8b8ede3135", "target": "stock-8e7522f842", "type": "tracks_stock", "weight": 3 }, { "id": "edge-324d1d10cd3f", "source": "thesis-8b8ede3135", "target": "stock-6ece0f6e09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-08fa9ca3e3e2", "source": "thesis-8b8ede3135", "target": "stock-bb47079d51", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6925770ce505", "source": "thesis-8b8ede3135", "target": "stock-2897e29810", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9f3f6f4a7d2c", "source": "thesis-8b8ede3135", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-7899692c4a5f", "source": "thesis-8b8ede3135", "target": "term-5dbd754de7", "type": "tracks_term", "weight": 2 }, { "id": "edge-5057b2b612ef", "source": "thesis-8b8ede3135", "target": "term-6840b7e095", "type": "tracks_term", "weight": 2 }, { "id": "edge-b24babc5af84", "source": "thesis-8b8ede3135", "target": "term-e1506406a5", "type": "tracks_term", "weight": 2 }, { "id": "edge-a1acca1a7692", "source": "thesis-8b8ede3135", "target": "term-3fdb185870", "type": "tracks_term", "weight": 2 }, { "id": "edge-1043ce6ec30e", "source": "thesis-8b8ede3135", "target": "term-f76e03f51e", "type": "tracks_term", "weight": 2 }, { "id": "edge-2de9f7944309", "source": "thesis-8b8ede3135", "target": "term-3d8049711c", "type": "tracks_term", "weight": 2 }, { "id": "edge-a1fcee8b4eaf", "source": "thesis-8b8ede3135", "target": "term-c045a6043b", "type": "tracks_term", "weight": 2 }, { "id": "edge-eedfe84d8755", "source": "thesis-8b8ede3135", "target": "event-1596942fae70", "type": "challenged_by", "weight": 1 }, { "id": "edge-891b9940164b", "source": "thesis-8b8ede3135", "target": "event-effa6da6fa25", "type": "challenged_by", "weight": 1 }, { "id": "edge-30b54e15b839", "source": "thesis-8b8ede3135", "target": "event-35d8f55fd722", "type": "challenged_by", "weight": 1 }, { "id": "edge-c1e51a5971d4", "source": "thesis-8b8ede3135", "target": "event-d74dcb35422f", "type": "challenged_by", "weight": 1 }, { "id": "edge-7e0e82910bb9", "source": "thesis-8b8ede3135", "target": "event-5a60ea7790a6", "type": "challenged_by", "weight": 1 }, { "id": "edge-e44eb020a0e7", "source": "thesis-8b8ede3135", "target": "event-fff74715140e", "type": "challenged_by", "weight": 1 }, { "id": "edge-d38a8199f29a", "source": "thesis-8b8ede3135", "target": "event-ef44a97e835b", "type": "challenged_by", "weight": 1 }, { "id": "edge-053c5c8b8fbe", "source": "thesis-8b8ede3135", "target": "event-c5cdafe69e8f", "type": "challenged_by", "weight": 1 }, { "id": "edge-5db94fb8727c", "source": "thesis-e9630d3ee1", "target": "stock-5829704445", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e8baae4f666c", "source": "thesis-e9630d3ee1", "target": "stock-d1c8471480", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f3681eade9d2", "source": "thesis-e9630d3ee1", "target": "stock-175b4b336d", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f0c964537ffd", "source": "thesis-e9630d3ee1", "target": "stock-510de57b5f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0af30f33c22d", "source": "thesis-e9630d3ee1", "target": "stock-68a288585c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-46a7204a73b1", "source": "thesis-e9630d3ee1", "target": "stock-0d33a0a8c4", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c75d70c1cdb0", "source": "thesis-e9630d3ee1", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-5485c01deaa9", "source": "thesis-e9630d3ee1", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-d5855e7f6809", "source": "thesis-e9630d3ee1", "target": "term-36e3041f1c", "type": "tracks_term", "weight": 2 }, { "id": "edge-873dc987f490", "source": "thesis-e9630d3ee1", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-d36413317f17", "source": "thesis-e9630d3ee1", "target": "term-dcdd3b5c45", "type": "tracks_term", "weight": 2 }, { "id": "edge-f76d0d4418ac", "source": "thesis-e9630d3ee1", "target": "term-00fb5022b7", "type": "tracks_term", "weight": 2 }, { "id": "edge-199b2d93267c", "source": "thesis-e9630d3ee1", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-c8ba7c299f06", "source": "thesis-e9630d3ee1", "target": "term-273ab83269", "type": "tracks_term", "weight": 2 }, { "id": "edge-e40cb98053f3", "source": "thesis-e9630d3ee1", "target": "event-d9bffeb07095", "type": "challenged_by", "weight": 1 }, { "id": "edge-254f9773fc96", "source": "thesis-e9630d3ee1", "target": "event-7147cc0ea9c9", "type": "challenged_by", "weight": 1 }, { "id": "edge-d9dd844bd4d6", "source": "thesis-e9630d3ee1", "target": "event-47a290f02d2b", "type": "challenged_by", "weight": 1 }, { "id": "edge-fe5eb6fe353b", "source": "thesis-e9630d3ee1", "target": "event-9f09a88af28a", "type": "challenged_by", "weight": 1 }, { "id": "edge-5b3a67e92afe", "source": "thesis-e9630d3ee1", "target": "event-77a96ced3bf7", "type": "challenged_by", "weight": 1 }, { "id": "edge-e35f5d694474", "source": "thesis-e9630d3ee1", "target": "event-078cbab4f900", "type": "challenged_by", "weight": 1 }, { "id": "edge-dc1922937c82", "source": "thesis-e9630d3ee1", "target": "event-9d798c2c6a07", "type": "challenged_by", "weight": 1 }, { "id": "edge-6d0b4ce54b4a", "source": "thesis-e9630d3ee1", "target": "event-9d988b58d2e0", "type": "challenged_by", "weight": 1 }, { "id": "edge-c2f828887ae4", "source": "thesis-89c14497ac", "target": "stock-9b19451158", "type": "tracks_stock", "weight": 3 }, { "id": "edge-81cc661a5760", "source": "thesis-89c14497ac", "target": "stock-e0671bd326", "type": "tracks_stock", "weight": 3 }, { "id": "edge-73e2dec4928f", "source": "thesis-89c14497ac", "target": "stock-b19fdc727e", "type": "tracks_stock", "weight": 3 }, { "id": "edge-17362682fc0b", "source": "thesis-89c14497ac", "target": "stock-89b531787b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6a1f24afb05e", "source": "thesis-89c14497ac", "target": "stock-a0cd4ecb4f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-83ff09cd0ec8", "source": "thesis-89c14497ac", "target": "stock-a436689d6c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5ff96414aa68", "source": "thesis-89c14497ac", "target": "term-d0db8b5e36", "type": "tracks_term", "weight": 2 }, { "id": "edge-64c959c87c9a", "source": "thesis-89c14497ac", "target": "term-be9d587def", "type": "tracks_term", "weight": 2 }, { "id": "edge-e895fd3a270c", "source": "thesis-89c14497ac", "target": "term-c6d29ec0e7", "type": "tracks_term", "weight": 2 }, { "id": "edge-000942274d12", "source": "thesis-89c14497ac", "target": "term-6f9a48800d", "type": "tracks_term", "weight": 2 }, { "id": "edge-0ae4fabc9b4f", "source": "thesis-89c14497ac", "target": "term-59b47e9ec5", "type": "tracks_term", "weight": 2 }, { "id": "edge-ee8718ba9c35", "source": "thesis-89c14497ac", "target": "term-f7f1ee30cf", "type": "tracks_term", "weight": 2 }, { "id": "edge-0b3b47ad1dba", "source": "thesis-89c14497ac", "target": "term-39157f7c0e", "type": "tracks_term", "weight": 2 }, { "id": "edge-8dc7c165e27d", "source": "thesis-89c14497ac", "target": "term-2c6ef0f0d0", "type": "tracks_term", "weight": 2 }, { "id": "edge-a4b17e8ab3a4", "source": "thesis-89c14497ac", "target": "event-aa56a14a7e77", "type": "challenged_by", "weight": 1 }, { "id": "edge-440e0f9ec91d", "source": "thesis-89c14497ac", "target": "event-8a37171cce92", "type": "supported_by", "weight": 1 }, { "id": "edge-72c60ee3c929", "source": "thesis-89c14497ac", "target": "event-daae54b39c91", "type": "supported_by", "weight": 1 }, { "id": "edge-8c3f800c4850", "source": "thesis-89c14497ac", "target": "event-891fb6137aa9", "type": "challenged_by", "weight": 1 }, { "id": "edge-88ca126e51d4", "source": "thesis-89c14497ac", "target": "event-250a6d3054ae", "type": "supported_by", "weight": 1 }, { "id": "edge-15321e748e4b", "source": "thesis-89c14497ac", "target": "event-d54724c5a15b", "type": "challenged_by", "weight": 1 }, { "id": "edge-f82ab527c681", "source": "thesis-89c14497ac", "target": "event-c1cab84c94e3", "type": "challenged_by", "weight": 1 }, { "id": "edge-8b17c968365c", "source": "thesis-89c14497ac", "target": "event-73915616f883", "type": "challenged_by", "weight": 1 }, { "id": "edge-ab7bf1fd4a30", "source": "thesis-075017003b", "target": "stock-426177902e", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7fe97f32e109", "source": "thesis-075017003b", "target": "stock-2888611640", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b13d6fca8c80", "source": "thesis-075017003b", "target": "stock-06214b33ef", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e81017309799", "source": "thesis-075017003b", "target": "stock-0ca8634002", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0ee436a10617", "source": "thesis-075017003b", "target": "stock-b64a63240f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6c79318e9c2c", "source": "thesis-075017003b", "target": "stock-1eaff1ec01", "type": "tracks_stock", "weight": 3 }, { "id": "edge-66ae30326b16", "source": "thesis-075017003b", "target": "term-95feb4759b", "type": "tracks_term", "weight": 2 }, { "id": "edge-2d535b432611", "source": "thesis-075017003b", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-8139d045b77a", "source": "thesis-075017003b", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-1d1f386e3e60", "source": "thesis-075017003b", "target": "term-1ae05a2701", "type": "tracks_term", "weight": 2 }, { "id": "edge-1d2d5943c039", "source": "thesis-075017003b", "target": "term-04f7acfcd2", "type": "tracks_term", "weight": 2 }, { "id": "edge-89334d095054", "source": "thesis-075017003b", "target": "term-7192e079e7", "type": "tracks_term", "weight": 2 }, { "id": "edge-11f7f61188a8", "source": "thesis-075017003b", "target": "term-5e71d16a67", "type": "tracks_term", "weight": 2 }, { "id": "edge-52d071e13bdd", "source": "thesis-075017003b", "target": "term-a050178590", "type": "tracks_term", "weight": 2 }, { "id": "edge-4f6708bf56c3", "source": "thesis-075017003b", "target": "event-9ebabd45f3e5", "type": "challenged_by", "weight": 1 }, { "id": "edge-86a658072374", "source": "thesis-075017003b", "target": "event-450482f7b4c2", "type": "challenged_by", "weight": 1 }, { "id": "edge-7bf5d9eedf8c", "source": "thesis-075017003b", "target": "event-6eb15f14e56c", "type": "challenged_by", "weight": 1 }, { "id": "edge-f2fda780458d", "source": "thesis-075017003b", "target": "event-44b8692d661f", "type": "challenged_by", "weight": 1 }, { "id": "edge-272a245fb15b", "source": "thesis-075017003b", "target": "event-c76fd0cb4df1", "type": "challenged_by", "weight": 1 }, { "id": "edge-9f32f35d0be4", "source": "thesis-075017003b", "target": "event-89b4b4fc3e69", "type": "challenged_by", "weight": 1 }, { "id": "edge-699634462c60", "source": "thesis-075017003b", "target": "event-e5bf58cdad10", "type": "challenged_by", "weight": 1 }, { "id": "edge-f9a7675ecfc1", "source": "thesis-075017003b", "target": "event-62029a01e8c0", "type": "challenged_by", "weight": 1 }, { "id": "edge-61538ac8a7c6", "source": "thesis-42faeae623", "target": "stock-b78cca2401", "type": "tracks_stock", "weight": 3 }, { "id": "edge-114c08a387ca", "source": "thesis-42faeae623", "target": "stock-b93441048d", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5a490d6595a6", "source": "thesis-42faeae623", "target": "stock-6b03c0b5e7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-38d84c472d13", "source": "thesis-42faeae623", "target": "stock-6f93ca0ebd", "type": "tracks_stock", "weight": 3 }, { "id": "edge-05d58d685e75", "source": "thesis-42faeae623", "target": "stock-264924258d", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c8a7ba727c04", "source": "thesis-42faeae623", "target": "stock-b2f82c3028", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e42eb850d6f3", "source": "thesis-42faeae623", "target": "term-e4a3cbf326", "type": "tracks_term", "weight": 2 }, { "id": "edge-ed4a8a87a928", "source": "thesis-42faeae623", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-98d7f59719b5", "source": "thesis-42faeae623", "target": "term-d49897f8c4", "type": "tracks_term", "weight": 2 }, { "id": "edge-b6a9ff383816", "source": "thesis-42faeae623", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-afffcf39ad6c", "source": "thesis-42faeae623", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-f48301be471b", "source": "thesis-42faeae623", "target": "term-a4b8abb418", "type": "tracks_term", "weight": 2 }, { "id": "edge-9c05dab673a2", "source": "thesis-42faeae623", "target": "term-4de0127946", "type": "tracks_term", "weight": 2 }, { "id": "edge-e55b71e88e69", "source": "thesis-42faeae623", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-8a1261e15223", "source": "thesis-42faeae623", "target": "event-41cf97e7a08b", "type": "challenged_by", "weight": 1 }, { "id": "edge-10c4ee5ccba1", "source": "thesis-42faeae623", "target": "event-2e97f577dbd6", "type": "challenged_by", "weight": 1 }, { "id": "edge-9f1a548f36a2", "source": "thesis-42faeae623", "target": "event-7ecd24840b4f", "type": "challenged_by", "weight": 1 }, { "id": "edge-1802798a4f7d", "source": "thesis-42faeae623", "target": "event-2938d08d3ff0", "type": "challenged_by", "weight": 1 }, { "id": "edge-30920028acc5", "source": "thesis-42faeae623", "target": "event-80376edf9d5a", "type": "challenged_by", "weight": 1 }, { "id": "edge-33bc3ada17dd", "source": "thesis-42faeae623", "target": "event-7f38e9639830", "type": "challenged_by", "weight": 1 }, { "id": "edge-8caee23522ad", "source": "thesis-42faeae623", "target": "event-3a192bc9dbe9", "type": "challenged_by", "weight": 1 }, { "id": "edge-90fc1f12d7a3", "source": "thesis-42faeae623", "target": "event-ed6d612db37d", "type": "challenged_by", "weight": 1 }, { "id": "edge-a18aaec313cb", "source": "thesis-9d3836cc07", "target": "stock-8638f0279b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-63bab8e23f88", "source": "thesis-9d3836cc07", "target": "stock-a25fc665a4", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f46369c7d883", "source": "thesis-9d3836cc07", "target": "stock-2d1edf729c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-fdc35d850efe", "source": "thesis-9d3836cc07", "target": "stock-31ac18dc22", "type": "tracks_stock", "weight": 3 }, { "id": "edge-12b03497e5ac", "source": "thesis-9d3836cc07", "target": "stock-97bfc0d991", "type": "tracks_stock", "weight": 3 }, { "id": "edge-16a82a36580d", "source": "thesis-9d3836cc07", "target": "stock-130f70ae5b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-cf2cc94f5276", "source": "thesis-9d3836cc07", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-e163c0d6c2f7", "source": "thesis-9d3836cc07", "target": "term-95feb4759b", "type": "tracks_term", "weight": 2 }, { "id": "edge-eb6990c7ce12", "source": "thesis-9d3836cc07", "target": "term-c7bac46904", "type": "tracks_term", "weight": 2 }, { "id": "edge-d3e7f59e3ede", "source": "thesis-9d3836cc07", "target": "term-49d2dc07ac", "type": "tracks_term", "weight": 2 }, { "id": "edge-428baf874860", "source": "thesis-9d3836cc07", "target": "term-4f749de7c2", "type": "tracks_term", "weight": 2 }, { "id": "edge-6e83ab5d9738", "source": "thesis-9d3836cc07", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-b4c6ed067880", "source": "thesis-9d3836cc07", "target": "term-36e3041f1c", "type": "tracks_term", "weight": 2 }, { "id": "edge-5cb4a8946143", "source": "thesis-9d3836cc07", "target": "term-0e2b23eeed", "type": "tracks_term", "weight": 2 }, { "id": "edge-845207d1f0a7", "source": "thesis-9d3836cc07", "target": "event-9ee02a1f9f67", "type": "challenged_by", "weight": 1 }, { "id": "edge-15e7f3a67326", "source": "thesis-9d3836cc07", "target": "event-530d80f63cb4", "type": "supported_by", "weight": 1 }, { "id": "edge-acbe4ff52a1d", "source": "thesis-9d3836cc07", "target": "event-fe97c369b30a", "type": "challenged_by", "weight": 1 }, { "id": "edge-fd1d04ba709a", "source": "thesis-9d3836cc07", "target": "event-96ba214e9d5b", "type": "challenged_by", "weight": 1 }, { "id": "edge-3b37bce4887e", "source": "thesis-9d3836cc07", "target": "event-2f7dbcbb01f5", "type": "challenged_by", "weight": 1 }, { "id": "edge-da3e2e4b0f3f", "source": "thesis-9d3836cc07", "target": "event-7a656643581e", "type": "challenged_by", "weight": 1 }, { "id": "edge-68b9b874ef1e", "source": "thesis-9d3836cc07", "target": "event-44d8838be0b7", "type": "challenged_by", "weight": 1 }, { "id": "edge-6e4261ea28bb", "source": "thesis-9d3836cc07", "target": "event-013ef198b05b", "type": "challenged_by", "weight": 1 }, { "id": "edge-f8eb50f1eb93", "source": "thesis-fd31cc177b", "target": "stock-06214b33ef", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a101751d128f", "source": "thesis-fd31cc177b", "target": "stock-4c3e22d6f2", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7cd570a3f2c7", "source": "thesis-fd31cc177b", "target": "stock-b9b86bd6b7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-475d71f46663", "source": "thesis-fd31cc177b", "target": "stock-8f4b41e6fb", "type": "tracks_stock", "weight": 3 }, { "id": "edge-69dd9e3bdb21", "source": "thesis-fd31cc177b", "target": "stock-f86096cddd", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d3b1ae7e4b9d", "source": "thesis-fd31cc177b", "target": "stock-de7f5dc2fb", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ecb464f702a3", "source": "thesis-fd31cc177b", "target": "term-d8705755e6", "type": "tracks_term", "weight": 2 }, { "id": "edge-7404558ce268", "source": "thesis-fd31cc177b", "target": "term-920e88c5a6", "type": "tracks_term", "weight": 2 }, { "id": "edge-5cd80c78fa43", "source": "thesis-fd31cc177b", "target": "term-8f8fd9ecdd", "type": "tracks_term", "weight": 2 }, { "id": "edge-ab885a821c2c", "source": "thesis-fd31cc177b", "target": "term-f30dfe1cfb", "type": "tracks_term", "weight": 2 }, { "id": "edge-e749fde4741a", "source": "thesis-fd31cc177b", "target": "term-bc0e0206bf", "type": "tracks_term", "weight": 2 }, { "id": "edge-467daead90cf", "source": "thesis-fd31cc177b", "target": "term-652d75f9ba", "type": "tracks_term", "weight": 2 }, { "id": "edge-05e4ad66b33e", "source": "thesis-fd31cc177b", "target": "term-caa79af4db", "type": "tracks_term", "weight": 2 }, { "id": "edge-32e856511cbd", "source": "thesis-fd31cc177b", "target": "term-dffc177fd0", "type": "tracks_term", "weight": 2 }, { "id": "edge-3109346ddfd8", "source": "thesis-fd31cc177b", "target": "event-b869b3a04c52", "type": "challenged_by", "weight": 1 }, { "id": "edge-a507471f854c", "source": "thesis-fd31cc177b", "target": "event-1963584fe028", "type": "challenged_by", "weight": 1 }, { "id": "edge-4c7bebc07a15", "source": "thesis-fd31cc177b", "target": "event-8c48c1786f1b", "type": "supported_by", "weight": 1 }, { "id": "edge-41bbb2552a96", "source": "thesis-fd31cc177b", "target": "event-7d4f684962a6", "type": "challenged_by", "weight": 1 }, { "id": "edge-ad91af9f75d4", "source": "thesis-fd31cc177b", "target": "event-cd5338c153c6", "type": "challenged_by", "weight": 1 }, { "id": "edge-227f4185b170", "source": "thesis-fd31cc177b", "target": "event-c975db8641a5", "type": "supported_by", "weight": 1 }, { "id": "edge-eed50181c3fb", "source": "thesis-fd31cc177b", "target": "event-27c2cfd12c5c", "type": "supported_by", "weight": 1 }, { "id": "edge-cf670de2806b", "source": "thesis-fd31cc177b", "target": "event-0086e682ecc0", "type": "supported_by", "weight": 1 }, { "id": "edge-e58229bfc711", "source": "thesis-51c6d833ae", "target": "stock-81aa4e9efc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-8a8285207d83", "source": "thesis-51c6d833ae", "target": "stock-7ad1c1f434", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7fb0449ffe46", "source": "thesis-51c6d833ae", "target": "stock-0f57e807aa", "type": "tracks_stock", "weight": 3 }, { "id": "edge-94218bb7c522", "source": "thesis-51c6d833ae", "target": "stock-b1b000a34d", "type": "tracks_stock", "weight": 3 }, { "id": "edge-72f00f1a74b6", "source": "thesis-51c6d833ae", "target": "stock-81c80ed154", "type": "tracks_stock", "weight": 3 }, { "id": "edge-586040ccbe09", "source": "thesis-51c6d833ae", "target": "stock-bb3ad25ff3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5d4d65aabf56", "source": "thesis-51c6d833ae", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-bf5a776956ef", "source": "thesis-51c6d833ae", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-68c988c99e7d", "source": "thesis-51c6d833ae", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-967af5deac00", "source": "thesis-51c6d833ae", "target": "term-6840b7e095", "type": "tracks_term", "weight": 2 }, { "id": "edge-dd9a379385a8", "source": "thesis-51c6d833ae", "target": "term-e4eb18069d", "type": "tracks_term", "weight": 2 }, { "id": "edge-1fbf8a592961", "source": "thesis-51c6d833ae", "target": "term-0a63dd9aa0", "type": "tracks_term", "weight": 2 }, { "id": "edge-80955714256e", "source": "thesis-51c6d833ae", "target": "term-9be34f3e54", "type": "tracks_term", "weight": 2 }, { "id": "edge-6f1bca2978e2", "source": "thesis-51c6d833ae", "target": "term-aeadfe2770", "type": "tracks_term", "weight": 2 }, { "id": "edge-892f8a1097e0", "source": "thesis-51c6d833ae", "target": "event-cf58e2b83678", "type": "challenged_by", "weight": 1 }, { "id": "edge-a8238a36d0b5", "source": "thesis-51c6d833ae", "target": "event-af7b47443cea", "type": "challenged_by", "weight": 1 }, { "id": "edge-8dc4af9388ab", "source": "thesis-51c6d833ae", "target": "event-aa0ed92b3610", "type": "challenged_by", "weight": 1 }, { "id": "edge-5e0434e5c726", "source": "thesis-51c6d833ae", "target": "event-d54ab9ee6a25", "type": "challenged_by", "weight": 1 }, { "id": "edge-42b0776472ba", "source": "thesis-51c6d833ae", "target": "event-5d4c8ac5368c", "type": "challenged_by", "weight": 1 }, { "id": "edge-10624beb5a8f", "source": "thesis-51c6d833ae", "target": "event-95e1268e9582", "type": "challenged_by", "weight": 1 }, { "id": "edge-e53c1fa6c6bb", "source": "thesis-51c6d833ae", "target": "event-37f4c250e01f", "type": "challenged_by", "weight": 1 }, { "id": "edge-c3f4f19e4773", "source": "thesis-51c6d833ae", "target": "event-a8609216670f", "type": "challenged_by", "weight": 1 }, { "id": "edge-4a6da360e86b", "source": "thesis-ffa474a3f2", "target": "stock-1eb44d6252", "type": "tracks_stock", "weight": 3 }, { "id": "edge-dbb57d0f4c9e", "source": "thesis-ffa474a3f2", "target": "stock-bfceb2656b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-248e01e081cf", "source": "thesis-ffa474a3f2", "target": "stock-316dddc5a4", "type": "tracks_stock", "weight": 3 }, { "id": "edge-df440941da1b", "source": "thesis-ffa474a3f2", "target": "stock-8638f0279b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-1192c4ffdff7", "source": "thesis-ffa474a3f2", "target": "stock-c7e6a87908", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ef758b7755ec", "source": "thesis-ffa474a3f2", "target": "stock-058d9f45ed", "type": "tracks_stock", "weight": 3 }, { "id": "edge-cf5752257daf", "source": "thesis-ffa474a3f2", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-b0456c270953", "source": "thesis-ffa474a3f2", "target": "term-c7bac46904", "type": "tracks_term", "weight": 2 }, { "id": "edge-6482adfe2862", "source": "thesis-ffa474a3f2", "target": "term-f223faa96f", "type": "tracks_term", "weight": 2 }, { "id": "edge-21996c3f74f9", "source": "thesis-ffa474a3f2", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-d19336d7afef", "source": "thesis-ffa474a3f2", "target": "term-4de0127946", "type": "tracks_term", "weight": 2 }, { "id": "edge-9def539331d5", "source": "thesis-ffa474a3f2", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-fc634bd7eebb", "source": "thesis-ffa474a3f2", "target": "term-36e3041f1c", "type": "tracks_term", "weight": 2 }, { "id": "edge-2189ac560811", "source": "thesis-ffa474a3f2", "target": "term-7834502344", "type": "tracks_term", "weight": 2 }, { "id": "edge-4a750623e4e1", "source": "thesis-ffa474a3f2", "target": "event-056dc7223eb9", "type": "challenged_by", "weight": 1 }, { "id": "edge-487dc47fc055", "source": "thesis-ffa474a3f2", "target": "event-3f3c9657fa51", "type": "challenged_by", "weight": 1 }, { "id": "edge-b58757cc41c1", "source": "thesis-ffa474a3f2", "target": "event-220e66b8057b", "type": "challenged_by", "weight": 1 }, { "id": "edge-b93755552c0a", "source": "thesis-ffa474a3f2", "target": "event-b09ee4e60002", "type": "challenged_by", "weight": 1 }, { "id": "edge-222a410e4fb0", "source": "thesis-ffa474a3f2", "target": "event-98aa5c086bd6", "type": "challenged_by", "weight": 1 }, { "id": "edge-633aa5cb227c", "source": "thesis-ffa474a3f2", "target": "event-51f53b92535d", "type": "challenged_by", "weight": 1 }, { "id": "edge-79ae1a46d5a7", "source": "thesis-ffa474a3f2", "target": "event-31edb092c269", "type": "challenged_by", "weight": 1 }, { "id": "edge-05080554f654", "source": "thesis-ffa474a3f2", "target": "event-8168fcadc6f6", "type": "challenged_by", "weight": 1 }, { "id": "edge-3118e1501967", "source": "thesis-3d421524cd", "target": "stock-b8b420a4c8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d6a77a94b6cf", "source": "thesis-3d421524cd", "target": "stock-cc1f4a0444", "type": "tracks_stock", "weight": 3 }, { "id": "edge-3e396814da26", "source": "thesis-3d421524cd", "target": "stock-51a3f330a2", "type": "tracks_stock", "weight": 3 }, { "id": "edge-83f2528eadec", "source": "thesis-3d421524cd", "target": "stock-59bde72b61", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7759c27dca79", "source": "thesis-3d421524cd", "target": "stock-31491c7f11", "type": "tracks_stock", "weight": 3 }, { "id": "edge-3473b66a2599", "source": "thesis-3d421524cd", "target": "stock-35264bffcb", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a2b76930cfd2", "source": "thesis-3d421524cd", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-bda270d504b0", "source": "thesis-3d421524cd", "target": "term-e27e330702", "type": "tracks_term", "weight": 2 }, { "id": "edge-0c57fe487d8c", "source": "thesis-3d421524cd", "target": "term-1414c1fd5e", "type": "tracks_term", "weight": 2 }, { "id": "edge-fd4defce1861", "source": "thesis-3d421524cd", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-df279256795c", "source": "thesis-3d421524cd", "target": "term-2b95a99fe5", "type": "tracks_term", "weight": 2 }, { "id": "edge-620408699c8a", "source": "thesis-3d421524cd", "target": "term-9196f0338a", "type": "tracks_term", "weight": 2 }, { "id": "edge-39c56c6954e9", "source": "thesis-3d421524cd", "target": "term-0da46a7fbf", "type": "tracks_term", "weight": 2 }, { "id": "edge-8f5ef07dcbde", "source": "thesis-3d421524cd", "target": "term-5d6a4410cd", "type": "tracks_term", "weight": 2 }, { "id": "edge-a612f95107f7", "source": "thesis-3d421524cd", "target": "event-ab71480c0c0a", "type": "supported_by", "weight": 1 }, { "id": "edge-6aa63947c28d", "source": "thesis-3d421524cd", "target": "event-7f7c43326484", "type": "supported_by", "weight": 1 }, { "id": "edge-31b1dbbf2321", "source": "thesis-3d421524cd", "target": "event-d7175f9a3e20", "type": "challenged_by", "weight": 1 }, { "id": "edge-60d502d02f57", "source": "thesis-3d421524cd", "target": "event-70731137dfa4", "type": "supported_by", "weight": 1 }, { "id": "edge-6b1f47d37ff5", "source": "thesis-3d421524cd", "target": "event-f864cc4682be", "type": "challenged_by", "weight": 1 }, { "id": "edge-85a22d53a0e6", "source": "thesis-3d421524cd", "target": "event-ae0a91e0bab7", "type": "challenged_by", "weight": 1 }, { "id": "edge-fd5594d1b94d", "source": "thesis-3d421524cd", "target": "event-78d3221ea78a", "type": "supported_by", "weight": 1 }, { "id": "edge-81bb41ac7f27", "source": "thesis-3d421524cd", "target": "event-2ba91a1d3f04", "type": "challenged_by", "weight": 1 }, { "id": "edge-85285e0263f5", "source": "thesis-cb2ab35c55", "target": "stock-7d94cbbbb3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f7daea6a0b44", "source": "thesis-cb2ab35c55", "target": "stock-a6bb0d8c7a", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6d6387f37153", "source": "thesis-cb2ab35c55", "target": "stock-6bfeca4163", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b4476c7dd773", "source": "thesis-cb2ab35c55", "target": "stock-8ed9c54f09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-3150a72a053f", "source": "thesis-cb2ab35c55", "target": "stock-6b03c0b5e7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-503cbbfa70c3", "source": "thesis-cb2ab35c55", "target": "stock-426219f2bc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a2f0ff9d4108", "source": "thesis-cb2ab35c55", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-8008a8433774", "source": "thesis-cb2ab35c55", "target": "term-0890f45668", "type": "tracks_term", "weight": 2 }, { "id": "edge-25b2d46ef85a", "source": "thesis-cb2ab35c55", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-2bbb8d49dae7", "source": "thesis-cb2ab35c55", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-028dc823ec14", "source": "thesis-cb2ab35c55", "target": "term-36e3041f1c", "type": "tracks_term", "weight": 2 }, { "id": "edge-9d46cf50d22f", "source": "thesis-cb2ab35c55", "target": "term-00fb5022b7", "type": "tracks_term", "weight": 2 }, { "id": "edge-15387afdc72a", "source": "thesis-cb2ab35c55", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-18e3a5cb4b8a", "source": "thesis-cb2ab35c55", "target": "term-ef242752b3", "type": "tracks_term", "weight": 2 }, { "id": "edge-3b35b4e3d16b", "source": "thesis-cb2ab35c55", "target": "event-a502f64b7e87", "type": "challenged_by", "weight": 1 }, { "id": "edge-7089082a0143", "source": "thesis-cb2ab35c55", "target": "event-382f75860f3e", "type": "challenged_by", "weight": 1 }, { "id": "edge-88884784f1eb", "source": "thesis-cb2ab35c55", "target": "event-131e45383877", "type": "challenged_by", "weight": 1 }, { "id": "edge-ba183f87fbb3", "source": "thesis-cb2ab35c55", "target": "event-31dcc87b2443", "type": "challenged_by", "weight": 1 }, { "id": "edge-0979e00ef613", "source": "thesis-cb2ab35c55", "target": "event-c12c85acaaaa", "type": "challenged_by", "weight": 1 }, { "id": "edge-781d4a039d63", "source": "thesis-cb2ab35c55", "target": "event-30081889500b", "type": "supported_by", "weight": 1 }, { "id": "edge-421591d23662", "source": "thesis-cb2ab35c55", "target": "event-199d3247058b", "type": "challenged_by", "weight": 1 }, { "id": "edge-9e5d97b09f2a", "source": "thesis-cb2ab35c55", "target": "event-ae97917af7e5", "type": "challenged_by", "weight": 1 }, { "id": "edge-4499b3f604a6", "source": "thesis-741d1d6f34", "target": "stock-7eb3864ff8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a878b6a2004b", "source": "thesis-741d1d6f34", "target": "stock-05e0aa017a", "type": "tracks_stock", "weight": 3 }, { "id": "edge-58849e6b176e", "source": "thesis-741d1d6f34", "target": "stock-c734fc6e80", "type": "tracks_stock", "weight": 3 }, { "id": "edge-545251ddd874", "source": "thesis-741d1d6f34", "target": "stock-ab55154707", "type": "tracks_stock", "weight": 3 }, { "id": "edge-993d702cdeff", "source": "thesis-741d1d6f34", "target": "stock-8b1b4a2855", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5ab0b3892dba", "source": "thesis-741d1d6f34", "target": "stock-0ac1c28ffb", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d264982d4003", "source": "thesis-741d1d6f34", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-63e48b06a72e", "source": "thesis-741d1d6f34", "target": "term-f71ccd3834", "type": "tracks_term", "weight": 2 }, { "id": "edge-939a3c6316a2", "source": "thesis-741d1d6f34", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-b15f0c32822e", "source": "thesis-741d1d6f34", "target": "term-58fdb0941f", "type": "tracks_term", "weight": 2 }, { "id": "edge-dddc93a1bd13", "source": "thesis-741d1d6f34", "target": "term-690f4826dc", "type": "tracks_term", "weight": 2 }, { "id": "edge-b15df7f4d52f", "source": "thesis-741d1d6f34", "target": "term-40040cb181", "type": "tracks_term", "weight": 2 }, { "id": "edge-1cc7abedc229", "source": "thesis-741d1d6f34", "target": "term-90eae63487", "type": "tracks_term", "weight": 2 }, { "id": "edge-0dc8294403cb", "source": "thesis-741d1d6f34", "target": "term-eab762a03f", "type": "tracks_term", "weight": 2 }, { "id": "edge-e5e3581d0193", "source": "thesis-741d1d6f34", "target": "event-b30f2f931db1", "type": "challenged_by", "weight": 1 }, { "id": "edge-9732514e1c1b", "source": "thesis-741d1d6f34", "target": "event-cbf042737a86", "type": "challenged_by", "weight": 1 }, { "id": "edge-8933c16bef8f", "source": "thesis-741d1d6f34", "target": "event-221c6dc16c7b", "type": "challenged_by", "weight": 1 }, { "id": "edge-9ffd66d1d471", "source": "thesis-741d1d6f34", "target": "event-177d9bd16790", "type": "challenged_by", "weight": 1 }, { "id": "edge-5ec443c30813", "source": "thesis-741d1d6f34", "target": "event-9fbcaf1bb709", "type": "supported_by", "weight": 1 }, { "id": "edge-10764ca954ae", "source": "thesis-741d1d6f34", "target": "event-8ecb5c3aa619", "type": "challenged_by", "weight": 1 }, { "id": "edge-da3b4dc5a068", "source": "thesis-741d1d6f34", "target": "event-1354540e8ea0", "type": "challenged_by", "weight": 1 }, { "id": "edge-b316528dd369", "source": "thesis-741d1d6f34", "target": "event-885916f10391", "type": "challenged_by", "weight": 1 }, { "id": "edge-d5a370097857", "source": "thesis-7a6599a3f5", "target": "stock-e2d5caa67f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f2bb1713f0bc", "source": "thesis-7a6599a3f5", "target": "stock-bf6151209d", "type": "tracks_stock", "weight": 3 }, { "id": "edge-30f52e7b06f8", "source": "thesis-7a6599a3f5", "target": "stock-ca5f9510eb", "type": "tracks_stock", "weight": 3 }, { "id": "edge-2ae461f5c45b", "source": "thesis-7a6599a3f5", "target": "stock-56463b158c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-cef3444c930c", "source": "thesis-7a6599a3f5", "target": "stock-6c871273fc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-adcd9effec13", "source": "thesis-7a6599a3f5", "target": "stock-add00a99a1", "type": "tracks_stock", "weight": 3 }, { "id": "edge-cec71e7dce18", "source": "thesis-7a6599a3f5", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-25d4291ad5c2", "source": "thesis-7a6599a3f5", "target": "term-d49897f8c4", "type": "tracks_term", "weight": 2 }, { "id": "edge-c3bcd06a5826", "source": "thesis-7a6599a3f5", "target": "term-2ddcf6fb42", "type": "tracks_term", "weight": 2 }, { "id": "edge-59afd2efd6e8", "source": "thesis-7a6599a3f5", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-34547c992412", "source": "thesis-7a6599a3f5", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-613d413bc3b6", "source": "thesis-7a6599a3f5", "target": "term-36e3041f1c", "type": "tracks_term", "weight": 2 }, { "id": "edge-9167c137f1a8", "source": "thesis-7a6599a3f5", "target": "term-6001ec95f6", "type": "tracks_term", "weight": 2 }, { "id": "edge-cc9a66abbe2a", "source": "thesis-7a6599a3f5", "target": "term-09331fca77", "type": "tracks_term", "weight": 2 }, { "id": "edge-b1468d47ac55", "source": "thesis-7a6599a3f5", "target": "event-44d5072754bf", "type": "challenged_by", "weight": 1 }, { "id": "edge-af0a4373cab9", "source": "thesis-7a6599a3f5", "target": "event-ba8e36973cf5", "type": "challenged_by", "weight": 1 }, { "id": "edge-5a1fe5ad76bf", "source": "thesis-7a6599a3f5", "target": "event-8eed65178d0f", "type": "challenged_by", "weight": 1 }, { "id": "edge-ba9b2ebd839d", "source": "thesis-7a6599a3f5", "target": "event-c6d8e316d5f4", "type": "challenged_by", "weight": 1 }, { "id": "edge-7cef11259891", "source": "thesis-7a6599a3f5", "target": "event-092cb21731a6", "type": "challenged_by", "weight": 1 }, { "id": "edge-fc3c4b5d0986", "source": "thesis-7a6599a3f5", "target": "event-4d7095acec29", "type": "challenged_by", "weight": 1 }, { "id": "edge-07415da88653", "source": "thesis-7a6599a3f5", "target": "event-29c9429db8f8", "type": "challenged_by", "weight": 1 }, { "id": "edge-f5fbc7e95475", "source": "thesis-7a6599a3f5", "target": "event-6a85464c3787", "type": "challenged_by", "weight": 1 }, { "id": "edge-38e48f1c9e11", "source": "thesis-603a5f7966", "target": "stock-c39cc2da75", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e2dbbbf8b426", "source": "thesis-603a5f7966", "target": "stock-add00a99a1", "type": "tracks_stock", "weight": 3 }, { "id": "edge-91c27f947428", "source": "thesis-603a5f7966", "target": "stock-6c871273fc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-319660f63fe2", "source": "thesis-603a5f7966", "target": "stock-56463b158c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-de670a576879", "source": "thesis-603a5f7966", "target": "stock-f0cecca160", "type": "tracks_stock", "weight": 3 }, { "id": "edge-8d7d8faf8602", "source": "thesis-603a5f7966", "target": "stock-e2d5caa67f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7afc9e850ef9", "source": "thesis-603a5f7966", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-b274fe0490e6", "source": "thesis-603a5f7966", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-999406c3ca4f", "source": "thesis-603a5f7966", "target": "term-a4b8abb418", "type": "tracks_term", "weight": 2 }, { "id": "edge-c2d91bbf1b0b", "source": "thesis-603a5f7966", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-22036cc5e2b5", "source": "thesis-603a5f7966", "target": "term-60e981724c", "type": "tracks_term", "weight": 2 }, { "id": "edge-0f4a0a296e4e", "source": "thesis-603a5f7966", "target": "term-f71ccd3834", "type": "tracks_term", "weight": 2 }, { "id": "edge-cde68d4f9039", "source": "thesis-603a5f7966", "target": "term-a2e9329b92", "type": "tracks_term", "weight": 2 }, { "id": "edge-c025dfaec9f4", "source": "thesis-603a5f7966", "target": "term-d0cbbe9778", "type": "tracks_term", "weight": 2 }, { "id": "edge-90c0c2d816d4", "source": "thesis-603a5f7966", "target": "event-65883110e41f", "type": "challenged_by", "weight": 1 }, { "id": "edge-4a5d163a4c24", "source": "thesis-603a5f7966", "target": "event-82226b8ac732", "type": "challenged_by", "weight": 1 }, { "id": "edge-b7efcae55878", "source": "thesis-603a5f7966", "target": "event-5424cce8e24a", "type": "challenged_by", "weight": 1 }, { "id": "edge-4ad59119bcf8", "source": "thesis-603a5f7966", "target": "event-6b59f25c3f08", "type": "challenged_by", "weight": 1 }, { "id": "edge-500abb5331ef", "source": "thesis-603a5f7966", "target": "event-16cf8fc6dbca", "type": "challenged_by", "weight": 1 }, { "id": "edge-a3581e4e0331", "source": "thesis-603a5f7966", "target": "event-954bfcf5f6a3", "type": "challenged_by", "weight": 1 }, { "id": "edge-130d0dd3a14e", "source": "thesis-603a5f7966", "target": "event-33fcdeb90faf", "type": "challenged_by", "weight": 1 }, { "id": "edge-c1fa78119865", "source": "thesis-603a5f7966", "target": "event-12d1d1ce5829", "type": "supported_by", "weight": 1 }, { "id": "edge-fdf4c876cc5e", "source": "thesis-7f0a75c5f6", "target": "stock-c10eb640cf", "type": "tracks_stock", "weight": 3 }, { "id": "edge-78f26fbe8d4a", "source": "thesis-7f0a75c5f6", "target": "stock-875caa8e41", "type": "tracks_stock", "weight": 3 }, { "id": "edge-2cdb02fb31b1", "source": "thesis-7f0a75c5f6", "target": "stock-4b984b3ea8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-402b177f3f70", "source": "thesis-7f0a75c5f6", "target": "stock-8ed9c54f09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-471861ccb440", "source": "thesis-7f0a75c5f6", "target": "stock-9779469589", "type": "tracks_stock", "weight": 3 }, { "id": "edge-096b9eff2b26", "source": "thesis-7f0a75c5f6", "target": "stock-d1d35fcaf6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c99e4d066322", "source": "thesis-7f0a75c5f6", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-4097e73639c6", "source": "thesis-7f0a75c5f6", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-b375710f4e33", "source": "thesis-7f0a75c5f6", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-28578688efce", "source": "thesis-7f0a75c5f6", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-a56449d85749", "source": "thesis-7f0a75c5f6", "target": "term-04f7acfcd2", "type": "tracks_term", "weight": 2 }, { "id": "edge-40bf283362d7", "source": "thesis-7f0a75c5f6", "target": "term-6001ec95f6", "type": "tracks_term", "weight": 2 }, { "id": "edge-b0900eb08043", "source": "thesis-7f0a75c5f6", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-03ed8da64e19", "source": "thesis-7f0a75c5f6", "target": "term-690156f98a", "type": "tracks_term", "weight": 2 }, { "id": "edge-d9e3d791cc6d", "source": "thesis-7f0a75c5f6", "target": "event-3cdef9304f99", "type": "challenged_by", "weight": 1 }, { "id": "edge-125f992a8c76", "source": "thesis-7f0a75c5f6", "target": "event-59e1b33f841b", "type": "challenged_by", "weight": 1 }, { "id": "edge-fd862287ef3d", "source": "thesis-7f0a75c5f6", "target": "event-eda974c7d178", "type": "challenged_by", "weight": 1 }, { "id": "edge-61874a688f86", "source": "thesis-7f0a75c5f6", "target": "event-4507c9c65d05", "type": "challenged_by", "weight": 1 }, { "id": "edge-fabe36f85d18", "source": "thesis-7f0a75c5f6", "target": "event-290ec8b911b9", "type": "challenged_by", "weight": 1 }, { "id": "edge-c3e7399a5716", "source": "thesis-7f0a75c5f6", "target": "event-2401ecd9a077", "type": "challenged_by", "weight": 1 }, { "id": "edge-0e90157095fc", "source": "thesis-7f0a75c5f6", "target": "event-40500e5cb763", "type": "challenged_by", "weight": 1 }, { "id": "edge-027231427f92", "source": "thesis-7f0a75c5f6", "target": "event-2082dea424d5", "type": "challenged_by", "weight": 1 }, { "id": "edge-4f98d4dd0727", "source": "thesis-44cf9d91f6", "target": "stock-bb47079d51", "type": "tracks_stock", "weight": 3 }, { "id": "edge-8a6806cabd2f", "source": "thesis-44cf9d91f6", "target": "stock-6ece0f6e09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a0afa6eb8253", "source": "thesis-44cf9d91f6", "target": "stock-6681ba3430", "type": "tracks_stock", "weight": 3 }, { "id": "edge-2d804b21bd12", "source": "thesis-44cf9d91f6", "target": "stock-732952499c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-78f0d1fe6d3d", "source": "thesis-44cf9d91f6", "target": "stock-1f5eddebbe", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f5347cac1bdd", "source": "thesis-44cf9d91f6", "target": "stock-bab38386c1", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f71dcdb9272d", "source": "thesis-44cf9d91f6", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-5fba78d3601f", "source": "thesis-44cf9d91f6", "target": "term-60856a95b4", "type": "tracks_term", "weight": 2 }, { "id": "edge-26345c1eda06", "source": "thesis-44cf9d91f6", "target": "term-5dbd754de7", "type": "tracks_term", "weight": 2 }, { "id": "edge-0dddc1379d7f", "source": "thesis-44cf9d91f6", "target": "term-ede68d6aba", "type": "tracks_term", "weight": 2 }, { "id": "edge-3153d6a0c35e", "source": "thesis-44cf9d91f6", "target": "term-e1506406a5", "type": "tracks_term", "weight": 2 }, { "id": "edge-8c46dd24608e", "source": "thesis-44cf9d91f6", "target": "term-b1a96dd646", "type": "tracks_term", "weight": 2 }, { "id": "edge-edd2be25b0a3", "source": "thesis-44cf9d91f6", "target": "term-3d8049711c", "type": "tracks_term", "weight": 2 }, { "id": "edge-72c596535bb9", "source": "thesis-44cf9d91f6", "target": "term-8ed3f6ad68", "type": "tracks_term", "weight": 2 }, { "id": "edge-1c8348192f2d", "source": "thesis-44cf9d91f6", "target": "event-44f9d08cab5a", "type": "challenged_by", "weight": 1 }, { "id": "edge-e446b5d81b39", "source": "thesis-44cf9d91f6", "target": "event-ed39b6671ab1", "type": "challenged_by", "weight": 1 }, { "id": "edge-5301689ea43f", "source": "thesis-44cf9d91f6", "target": "event-1be9fab7215d", "type": "challenged_by", "weight": 1 }, { "id": "edge-cc788e797dd6", "source": "thesis-44cf9d91f6", "target": "event-cd5a3967260e", "type": "challenged_by", "weight": 1 }, { "id": "edge-8d9fdea22049", "source": "thesis-44cf9d91f6", "target": "event-51d2949b0102", "type": "challenged_by", "weight": 1 }, { "id": "edge-83938c50c7c0", "source": "thesis-44cf9d91f6", "target": "event-9acc96cee034", "type": "challenged_by", "weight": 1 }, { "id": "edge-81b50229e694", "source": "thesis-44cf9d91f6", "target": "event-3d40679406f9", "type": "challenged_by", "weight": 1 }, { "id": "edge-b82249a26eb6", "source": "thesis-44cf9d91f6", "target": "event-7ded98b2ce81", "type": "challenged_by", "weight": 1 }, { "id": "edge-b6a49e880917", "source": "thesis-31f32af77b", "target": "stock-9cd0d546e6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f835bf0f4768", "source": "thesis-31f32af77b", "target": "stock-efe2ecdc5b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f33791ccb623", "source": "thesis-31f32af77b", "target": "stock-294da558b6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0a25097353ad", "source": "thesis-31f32af77b", "target": "stock-bf6151209d", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6c76893d32a7", "source": "thesis-31f32af77b", "target": "stock-a67fba8195", "type": "tracks_stock", "weight": 3 }, { "id": "edge-955bdc5f6895", "source": "thesis-31f32af77b", "target": "stock-56463b158c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-61f2ec40215c", "source": "thesis-31f32af77b", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-32eeecbfac3f", "source": "thesis-31f32af77b", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-bb96bcce0ba4", "source": "thesis-31f32af77b", "target": "term-36e3041f1c", "type": "tracks_term", "weight": 2 }, { "id": "edge-811989d0fa39", "source": "thesis-31f32af77b", "target": "term-d0cbbe9778", "type": "tracks_term", "weight": 2 }, { "id": "edge-8942d14c0993", "source": "thesis-31f32af77b", "target": "term-18126b597c", "type": "tracks_term", "weight": 2 }, { "id": "edge-b6f1f0a587d4", "source": "thesis-31f32af77b", "target": "term-89b1aee428", "type": "tracks_term", "weight": 2 }, { "id": "edge-0f466aa679c4", "source": "thesis-31f32af77b", "target": "term-56681010b7", "type": "tracks_term", "weight": 2 }, { "id": "edge-9300ea3889d8", "source": "thesis-31f32af77b", "target": "term-25bf8e1a23", "type": "tracks_term", "weight": 2 }, { "id": "edge-49e11d3b1c3e", "source": "thesis-31f32af77b", "target": "event-8c6beb26f97a", "type": "challenged_by", "weight": 1 }, { "id": "edge-d2ee5634140e", "source": "thesis-31f32af77b", "target": "event-0349b2df98cd", "type": "challenged_by", "weight": 1 }, { "id": "edge-27b7857d2b40", "source": "thesis-31f32af77b", "target": "event-81436edf403a", "type": "challenged_by", "weight": 1 }, { "id": "edge-56f0fc774813", "source": "thesis-31f32af77b", "target": "event-92df911eceef", "type": "challenged_by", "weight": 1 }, { "id": "edge-fe39df4efe02", "source": "thesis-31f32af77b", "target": "event-b0d219ae76d0", "type": "challenged_by", "weight": 1 }, { "id": "edge-542dd7eab76d", "source": "thesis-31f32af77b", "target": "event-a2ca6929226a", "type": "challenged_by", "weight": 1 }, { "id": "edge-9496a6d732d0", "source": "thesis-31f32af77b", "target": "event-d7ec1b1c2462", "type": "supported_by", "weight": 1 }, { "id": "edge-a25fedb364cc", "source": "thesis-31f32af77b", "target": "event-3146ffe4c973", "type": "supported_by", "weight": 1 }, { "id": "edge-0b2a09d22e81", "source": "thesis-a803fa3312", "target": "stock-06214b33ef", "type": "tracks_stock", "weight": 3 }, { "id": "edge-079e4bd30245", "source": "thesis-a803fa3312", "target": "stock-4c3e22d6f2", "type": "tracks_stock", "weight": 3 }, { "id": "edge-3bf515945c83", "source": "thesis-a803fa3312", "target": "stock-d1c8471480", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ca147975cacf", "source": "thesis-a803fa3312", "target": "stock-0ca8634002", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b73b7e288d6b", "source": "thesis-a803fa3312", "target": "stock-7b88702902", "type": "tracks_stock", "weight": 3 }, { "id": "edge-96a947acb9e9", "source": "thesis-a803fa3312", "target": "stock-b64a63240f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-2a2f34e9b826", "source": "thesis-a803fa3312", "target": "term-51360b078e", "type": "tracks_term", "weight": 2 }, { "id": "edge-2484145c26ec", "source": "thesis-a803fa3312", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-5cc635eff590", "source": "thesis-a803fa3312", "target": "term-1ae05a2701", "type": "tracks_term", "weight": 2 }, { "id": "edge-aaa9178d77ad", "source": "thesis-a803fa3312", "target": "term-dcdd3b5c45", "type": "tracks_term", "weight": 2 }, { "id": "edge-6353d66a78ff", "source": "thesis-a803fa3312", "target": "term-f55a72aff5", "type": "tracks_term", "weight": 2 }, { "id": "edge-d4148163a5ce", "source": "thesis-a803fa3312", "target": "term-be0da309c0", "type": "tracks_term", "weight": 2 }, { "id": "edge-ebbae3df76f8", "source": "thesis-a803fa3312", "target": "term-920e88c5a6", "type": "tracks_term", "weight": 2 }, { "id": "edge-09d52f51a172", "source": "thesis-a803fa3312", "target": "term-a0af9f865b", "type": "tracks_term", "weight": 2 }, { "id": "edge-07b577a11133", "source": "thesis-a803fa3312", "target": "event-3c11690e550e", "type": "challenged_by", "weight": 1 }, { "id": "edge-e708a9751dbc", "source": "thesis-a803fa3312", "target": "event-189b8c5ccf28", "type": "challenged_by", "weight": 1 }, { "id": "edge-cca3696fdc83", "source": "thesis-a803fa3312", "target": "event-a594224bf3a2", "type": "challenged_by", "weight": 1 }, { "id": "edge-d70279b2d357", "source": "thesis-a803fa3312", "target": "event-5b725936daf4", "type": "challenged_by", "weight": 1 }, { "id": "edge-8473e5d5b74e", "source": "thesis-a803fa3312", "target": "event-98ef1a275c83", "type": "supported_by", "weight": 1 }, { "id": "edge-4092d39c457f", "source": "thesis-a803fa3312", "target": "event-e19d15dc233b", "type": "challenged_by", "weight": 1 }, { "id": "edge-51084bc5fdb8", "source": "thesis-a803fa3312", "target": "event-c1a66691be8f", "type": "challenged_by", "weight": 1 }, { "id": "edge-a673187c9127", "source": "thesis-a803fa3312", "target": "event-d7dcf7fdd331", "type": "challenged_by", "weight": 1 }, { "id": "edge-844fa37bc4be", "source": "thesis-08dd090dc9", "target": "stock-0d1a838e74", "type": "tracks_stock", "weight": 3 }, { "id": "edge-936844249d41", "source": "thesis-08dd090dc9", "target": "stock-bfa4cc2c68", "type": "tracks_stock", "weight": 3 }, { "id": "edge-2e6fb2daaa8e", "source": "thesis-08dd090dc9", "target": "stock-7fe8dd309c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6e3b8963039c", "source": "thesis-08dd090dc9", "target": "stock-420d41ec1c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f27b65a724b0", "source": "thesis-08dd090dc9", "target": "stock-db052c8066", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0a06d4f9c94a", "source": "thesis-08dd090dc9", "target": "stock-0405f6aceb", "type": "tracks_stock", "weight": 3 }, { "id": "edge-924777969276", "source": "thesis-08dd090dc9", "target": "term-95feb4759b", "type": "tracks_term", "weight": 2 }, { "id": "edge-df38193a350b", "source": "thesis-08dd090dc9", "target": "term-0e2b23eeed", "type": "tracks_term", "weight": 2 }, { "id": "edge-25a931a68e21", "source": "thesis-08dd090dc9", "target": "term-1582c27783", "type": "tracks_term", "weight": 2 }, { "id": "edge-4f7fe14123a5", "source": "thesis-08dd090dc9", "target": "term-a050178590", "type": "tracks_term", "weight": 2 }, { "id": "edge-6f68de4a4b46", "source": "thesis-08dd090dc9", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-d255a35518f3", "source": "thesis-08dd090dc9", "target": "term-66e0f52800", "type": "tracks_term", "weight": 2 }, { "id": "edge-8e67ffff31d1", "source": "thesis-08dd090dc9", "target": "term-e21dcec424", "type": "tracks_term", "weight": 2 }, { "id": "edge-a12d332f175a", "source": "thesis-08dd090dc9", "target": "term-ef242752b3", "type": "tracks_term", "weight": 2 }, { "id": "edge-2d661f6f990b", "source": "thesis-08dd090dc9", "target": "event-9a50046c3525", "type": "challenged_by", "weight": 1 }, { "id": "edge-6c069d4be68d", "source": "thesis-08dd090dc9", "target": "event-8d5393cb276a", "type": "challenged_by", "weight": 1 }, { "id": "edge-9e765102ceb4", "source": "thesis-08dd090dc9", "target": "event-037d4a207c2f", "type": "challenged_by", "weight": 1 }, { "id": "edge-3f78506d7115", "source": "thesis-08dd090dc9", "target": "event-2ce41eabd726", "type": "challenged_by", "weight": 1 }, { "id": "edge-8b0647597943", "source": "thesis-08dd090dc9", "target": "event-b52c539eed36", "type": "challenged_by", "weight": 1 }, { "id": "edge-4da1bed480ae", "source": "thesis-08dd090dc9", "target": "event-62d5d022e562", "type": "challenged_by", "weight": 1 }, { "id": "edge-bc7c66464206", "source": "thesis-08dd090dc9", "target": "event-7d2c76e10f99", "type": "challenged_by", "weight": 1 }, { "id": "edge-aed96c8ed7c5", "source": "thesis-08dd090dc9", "target": "event-ee16c633bf32", "type": "challenged_by", "weight": 1 }, { "id": "edge-856b015841a6", "source": "thesis-3986819dc0", "target": "stock-c39cc2da75", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a15f414f9237", "source": "thesis-3986819dc0", "target": "stock-add00a99a1", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9342612304f1", "source": "thesis-3986819dc0", "target": "stock-82c0637078", "type": "tracks_stock", "weight": 3 }, { "id": "edge-161a6ca7876d", "source": "thesis-3986819dc0", "target": "stock-56463b158c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-fb7791643d4e", "source": "thesis-3986819dc0", "target": "stock-6c871273fc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-17f2289c3df8", "source": "thesis-3986819dc0", "target": "stock-1a47896f9f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-3afeebfe7ef4", "source": "thesis-3986819dc0", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-c3571d7c3bdd", "source": "thesis-3986819dc0", "target": "term-60e981724c", "type": "tracks_term", "weight": 2 }, { "id": "edge-af5d83e473bd", "source": "thesis-3986819dc0", "target": "term-f71ccd3834", "type": "tracks_term", "weight": 2 }, { "id": "edge-6c57f8ed040f", "source": "thesis-3986819dc0", "target": "term-c484297089", "type": "tracks_term", "weight": 2 }, { "id": "edge-93230088de1f", "source": "thesis-3986819dc0", "target": "term-d0cbbe9778", "type": "tracks_term", "weight": 2 }, { "id": "edge-54980c4dac83", "source": "thesis-3986819dc0", "target": "term-18126b597c", "type": "tracks_term", "weight": 2 }, { "id": "edge-441ab91fb5b5", "source": "thesis-3986819dc0", "target": "term-89b1aee428", "type": "tracks_term", "weight": 2 }, { "id": "edge-11f8f8c9ea6e", "source": "thesis-3986819dc0", "target": "term-dcc56252e4", "type": "tracks_term", "weight": 2 }, { "id": "edge-f824925ea67e", "source": "thesis-3986819dc0", "target": "event-55cd6027661b", "type": "challenged_by", "weight": 1 }, { "id": "edge-6f9a5adf43e7", "source": "thesis-3986819dc0", "target": "event-69545bc373ad", "type": "challenged_by", "weight": 1 }, { "id": "edge-2ae4f55bae0c", "source": "thesis-3986819dc0", "target": "event-4d98ea861149", "type": "challenged_by", "weight": 1 }, { "id": "edge-d94aa06d5270", "source": "thesis-3986819dc0", "target": "event-a3efb2f84285", "type": "challenged_by", "weight": 1 }, { "id": "edge-7e1b837c78b6", "source": "thesis-3986819dc0", "target": "event-e9b0629c23b4", "type": "challenged_by", "weight": 1 }, { "id": "edge-4458d40b19db", "source": "thesis-3986819dc0", "target": "event-bab78f037e63", "type": "challenged_by", "weight": 1 }, { "id": "edge-cf8fe8bc3ce7", "source": "thesis-3986819dc0", "target": "event-be4044271fce", "type": "challenged_by", "weight": 1 }, { "id": "edge-778b7eee4df3", "source": "thesis-3986819dc0", "target": "event-a2fecf2d64b2", "type": "challenged_by", "weight": 1 }, { "id": "edge-db914008247e", "source": "thesis-740e729f22", "target": "stock-d1c8471480", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d7700bfaffb9", "source": "thesis-740e729f22", "target": "stock-79f7c9313f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-40eb5e8e7f0a", "source": "thesis-740e729f22", "target": "stock-f57a081cf3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-4d4fcda54abf", "source": "thesis-740e729f22", "target": "stock-efe2ecdc5b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d120c0e8b0d1", "source": "thesis-740e729f22", "target": "stock-9cd0d546e6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7a76ab2867e1", "source": "thesis-740e729f22", "target": "stock-3d3ced839f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6e491de812aa", "source": "thesis-740e729f22", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-6fc27c0c7eff", "source": "thesis-740e729f22", "target": "term-dcdd3b5c45", "type": "tracks_term", "weight": 2 }, { "id": "edge-e38e8228cd12", "source": "thesis-740e729f22", "target": "term-b117b5bed4", "type": "tracks_term", "weight": 2 }, { "id": "edge-f3caa4179dc0", "source": "thesis-740e729f22", "target": "term-273ab83269", "type": "tracks_term", "weight": 2 }, { "id": "edge-eae1b6bc9862", "source": "thesis-740e729f22", "target": "term-ba9771ca73", "type": "tracks_term", "weight": 2 }, { "id": "edge-0ccc8187a513", "source": "thesis-740e729f22", "target": "term-f55a72aff5", "type": "tracks_term", "weight": 2 }, { "id": "edge-7da2e051771e", "source": "thesis-740e729f22", "target": "term-920e88c5a6", "type": "tracks_term", "weight": 2 }, { "id": "edge-d9ef8e46e9ce", "source": "thesis-740e729f22", "target": "term-b650f21f60", "type": "tracks_term", "weight": 2 }, { "id": "edge-cb8f4c28b82a", "source": "thesis-740e729f22", "target": "event-cbe4c8425406", "type": "challenged_by", "weight": 1 }, { "id": "edge-26df08967be9", "source": "thesis-740e729f22", "target": "event-3e3dd9763375", "type": "challenged_by", "weight": 1 }, { "id": "edge-968a7cd43073", "source": "thesis-740e729f22", "target": "event-927f21b2af4f", "type": "challenged_by", "weight": 1 }, { "id": "edge-2387e4b0d11a", "source": "thesis-740e729f22", "target": "event-1252355979da", "type": "challenged_by", "weight": 1 }, { "id": "edge-d1cb4a8a2896", "source": "thesis-740e729f22", "target": "event-4a0157d4090a", "type": "challenged_by", "weight": 1 }, { "id": "edge-9f5e79c66bad", "source": "thesis-740e729f22", "target": "event-3d6ddb43f55c", "type": "challenged_by", "weight": 1 }, { "id": "edge-46e67b1b67bf", "source": "thesis-740e729f22", "target": "event-a475a8524047", "type": "challenged_by", "weight": 1 }, { "id": "edge-69183aad36ac", "source": "thesis-740e729f22", "target": "event-b110ea275700", "type": "challenged_by", "weight": 1 }, { "id": "edge-510789129fdb", "source": "thesis-9e97140446", "target": "stock-0e73308404", "type": "tracks_stock", "weight": 3 }, { "id": "edge-149c8dea076a", "source": "thesis-9e97140446", "target": "stock-f960e0de27", "type": "tracks_stock", "weight": 3 }, { "id": "edge-20c5b2aa4f82", "source": "thesis-9e97140446", "target": "stock-8c35af6c45", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6eedbbb0ce6f", "source": "thesis-9e97140446", "target": "stock-e1734e0e85", "type": "tracks_stock", "weight": 3 }, { "id": "edge-87df4628b7fb", "source": "thesis-9e97140446", "target": "stock-dc9d1d443f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-45ae9b787d3a", "source": "thesis-9e97140446", "target": "stock-b5733c9451", "type": "tracks_stock", "weight": 3 }, { "id": "edge-2475834387d2", "source": "thesis-9e97140446", "target": "term-95feb4759b", "type": "tracks_term", "weight": 2 }, { "id": "edge-ce77e490d164", "source": "thesis-9e97140446", "target": "term-4f749de7c2", "type": "tracks_term", "weight": 2 }, { "id": "edge-dc56c63e0588", "source": "thesis-9e97140446", "target": "term-2ddcf6fb42", "type": "tracks_term", "weight": 2 }, { "id": "edge-cbe51f669f87", "source": "thesis-9e97140446", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-64dae7f87d8a", "source": "thesis-9e97140446", "target": "term-0e2b23eeed", "type": "tracks_term", "weight": 2 }, { "id": "edge-058221177cb3", "source": "thesis-9e97140446", "target": "term-2359bb6154", "type": "tracks_term", "weight": 2 }, { "id": "edge-80f0fa623223", "source": "thesis-9e97140446", "target": "term-641102ee2d", "type": "tracks_term", "weight": 2 }, { "id": "edge-e10fb0036ff2", "source": "thesis-9e97140446", "target": "term-7d5ee7ec76", "type": "tracks_term", "weight": 2 }, { "id": "edge-00c5696a5434", "source": "thesis-9e97140446", "target": "event-4c32f30661be", "type": "challenged_by", "weight": 1 }, { "id": "edge-58d6f5172cb6", "source": "thesis-9e97140446", "target": "event-63112f9365c6", "type": "challenged_by", "weight": 1 }, { "id": "edge-50f6f003099e", "source": "thesis-9e97140446", "target": "event-1c156edfd31e", "type": "challenged_by", "weight": 1 }, { "id": "edge-5249f40ebe45", "source": "thesis-9e97140446", "target": "event-b6596239eb27", "type": "challenged_by", "weight": 1 }, { "id": "edge-7cc877694bc9", "source": "thesis-9e97140446", "target": "event-422e1e7dc374", "type": "challenged_by", "weight": 1 }, { "id": "edge-e45ba743b4e0", "source": "thesis-9e97140446", "target": "event-f085bbdb6bf8", "type": "challenged_by", "weight": 1 }, { "id": "edge-530937c2453c", "source": "thesis-9e97140446", "target": "event-700a918545a6", "type": "challenged_by", "weight": 1 }, { "id": "edge-0837c9a36b30", "source": "thesis-9e97140446", "target": "event-be6b58aa0895", "type": "challenged_by", "weight": 1 }, { "id": "edge-ebce66fbda44", "source": "thesis-ce29facfbd", "target": "stock-dcd4dbe710", "type": "tracks_stock", "weight": 3 }, { "id": "edge-8272be6e5421", "source": "thesis-ce29facfbd", "target": "stock-d5a25b6072", "type": "tracks_stock", "weight": 3 }, { "id": "edge-8e11a069e417", "source": "thesis-ce29facfbd", "target": "stock-f57a081cf3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5c99ab608a70", "source": "thesis-ce29facfbd", "target": "stock-02c323bcb5", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9bdc225fcfd7", "source": "thesis-ce29facfbd", "target": "stock-efe2ecdc5b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7958ac66ff33", "source": "thesis-ce29facfbd", "target": "stock-9cd0d546e6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-4f3a97e8b9b1", "source": "thesis-ce29facfbd", "target": "term-95feb4759b", "type": "tracks_term", "weight": 2 }, { "id": "edge-1edecc632224", "source": "thesis-ce29facfbd", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-37da27fdc284", "source": "thesis-ce29facfbd", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-f6923303e8d5", "source": "thesis-ce29facfbd", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-e29ba6804d6e", "source": "thesis-ce29facfbd", "target": "term-f71ccd3834", "type": "tracks_term", "weight": 2 }, { "id": "edge-0eaf05e654d3", "source": "thesis-ce29facfbd", "target": "term-a2e9329b92", "type": "tracks_term", "weight": 2 }, { "id": "edge-c760e60be537", "source": "thesis-ce29facfbd", "target": "term-ba9771ca73", "type": "tracks_term", "weight": 2 }, { "id": "edge-5d8e3143b764", "source": "thesis-ce29facfbd", "target": "term-ef242752b3", "type": "tracks_term", "weight": 2 }, { "id": "edge-b1a5ce7a9a7d", "source": "thesis-ce29facfbd", "target": "event-e5eafeef5a03", "type": "challenged_by", "weight": 1 }, { "id": "edge-0cfa82bb836c", "source": "thesis-ce29facfbd", "target": "event-36938377f9e4", "type": "challenged_by", "weight": 1 }, { "id": "edge-a657d874e4ec", "source": "thesis-ce29facfbd", "target": "event-80d6d38efedd", "type": "challenged_by", "weight": 1 }, { "id": "edge-81fc659ac3c9", "source": "thesis-ce29facfbd", "target": "event-8252a052dbd2", "type": "challenged_by", "weight": 1 }, { "id": "edge-a78b2a28a5ed", "source": "thesis-ce29facfbd", "target": "event-a6bd71a5eff4", "type": "challenged_by", "weight": 1 }, { "id": "edge-c72b4866c0f0", "source": "thesis-ce29facfbd", "target": "event-f240a6f06bfb", "type": "challenged_by", "weight": 1 }, { "id": "edge-f8e55beb154d", "source": "thesis-ce29facfbd", "target": "event-3183435e6b91", "type": "challenged_by", "weight": 1 }, { "id": "edge-96b592625dc9", "source": "thesis-ce29facfbd", "target": "event-e16e83a229b1", "type": "challenged_by", "weight": 1 }, { "id": "edge-fd4b3fcb2a59", "source": "thesis-4b3b37e207", "target": "stock-dcd4dbe710", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6795dd0a0200", "source": "thesis-4b3b37e207", "target": "stock-517b2f6590", "type": "tracks_stock", "weight": 3 }, { "id": "edge-260c95e12153", "source": "thesis-4b3b37e207", "target": "stock-c39cc2da75", "type": "tracks_stock", "weight": 3 }, { "id": "edge-13bda3cd9408", "source": "thesis-4b3b37e207", "target": "stock-d5a25b6072", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a8b46f3c6526", "source": "thesis-4b3b37e207", "target": "stock-0716b17bb6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-1e6aaab1e0e4", "source": "thesis-4b3b37e207", "target": "stock-d54aa25b07", "type": "tracks_stock", "weight": 3 }, { "id": "edge-35f78ad2766c", "source": "thesis-4b3b37e207", "target": "term-60e981724c", "type": "tracks_term", "weight": 2 }, { "id": "edge-6a7cf431b30b", "source": "thesis-4b3b37e207", "target": "term-f71ccd3834", "type": "tracks_term", "weight": 2 }, { "id": "edge-bf42f371ee79", "source": "thesis-4b3b37e207", "target": "term-c36bc299e1", "type": "tracks_term", "weight": 2 }, { "id": "edge-0620c695d7ab", "source": "thesis-4b3b37e207", "target": "term-c484297089", "type": "tracks_term", "weight": 2 }, { "id": "edge-87c8a10965ce", "source": "thesis-4b3b37e207", "target": "term-a2e9329b92", "type": "tracks_term", "weight": 2 }, { "id": "edge-3db110a93b94", "source": "thesis-4b3b37e207", "target": "term-13aa5df7e2", "type": "tracks_term", "weight": 2 }, { "id": "edge-0a70c3ef6588", "source": "thesis-4b3b37e207", "target": "term-cdb38c8dc8", "type": "tracks_term", "weight": 2 }, { "id": "edge-c51d775d8ec5", "source": "thesis-4b3b37e207", "target": "term-df82d72441", "type": "tracks_term", "weight": 2 }, { "id": "edge-0473fa241124", "source": "thesis-4b3b37e207", "target": "event-bd2c67637e06", "type": "challenged_by", "weight": 1 }, { "id": "edge-b2261bf5993e", "source": "thesis-4b3b37e207", "target": "event-4a31147f09b0", "type": "challenged_by", "weight": 1 }, { "id": "edge-f9978bd9ea2b", "source": "thesis-4b3b37e207", "target": "event-4bfe72c8cee3", "type": "challenged_by", "weight": 1 }, { "id": "edge-b9a238e27f4f", "source": "thesis-4b3b37e207", "target": "event-1f6391bc5b00", "type": "challenged_by", "weight": 1 }, { "id": "edge-c7df028ea83c", "source": "thesis-4b3b37e207", "target": "event-6e8526945c11", "type": "challenged_by", "weight": 1 }, { "id": "edge-ad505e86c36b", "source": "thesis-4b3b37e207", "target": "event-809da0001211", "type": "supported_by", "weight": 1 }, { "id": "edge-0f8c86d60603", "source": "thesis-4b3b37e207", "target": "event-a875362de659", "type": "challenged_by", "weight": 1 }, { "id": "edge-d95558b93231", "source": "thesis-4b3b37e207", "target": "event-677f281c56de", "type": "challenged_by", "weight": 1 }, { "id": "edge-1d9a2f9aacd9", "source": "thesis-d0de40066b", "target": "stock-06214b33ef", "type": "tracks_stock", "weight": 3 }, { "id": "edge-771b04671f4c", "source": "thesis-d0de40066b", "target": "stock-4c3e22d6f2", "type": "tracks_stock", "weight": 3 }, { "id": "edge-fc3b6dfd2354", "source": "thesis-d0de40066b", "target": "stock-d1c8471480", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6c08bdf4b617", "source": "thesis-d0de40066b", "target": "stock-5829704445", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6bd49a07b1d5", "source": "thesis-d0de40066b", "target": "stock-0ca8634002", "type": "tracks_stock", "weight": 3 }, { "id": "edge-03a57e0196e0", "source": "thesis-d0de40066b", "target": "stock-426177902e", "type": "tracks_stock", "weight": 3 }, { "id": "edge-1a88a934eff0", "source": "thesis-d0de40066b", "target": "term-6840b7e095", "type": "tracks_term", "weight": 2 }, { "id": "edge-ce99521f3dc3", "source": "thesis-d0de40066b", "target": "term-f072652dd0", "type": "tracks_term", "weight": 2 }, { "id": "edge-7764b7d36860", "source": "thesis-d0de40066b", "target": "term-585a293e96", "type": "tracks_term", "weight": 2 }, { "id": "edge-2aae1e6af053", "source": "thesis-d0de40066b", "target": "event-53f4fe8837f7", "type": "challenged_by", "weight": 1 }, { "id": "edge-35d5fd4bd8bc", "source": "thesis-d0de40066b", "target": "event-a5ca3a24fc31", "type": "supported_by", "weight": 1 }, { "id": "edge-cd45bc1a5a4c", "source": "thesis-d0de40066b", "target": "event-6a31fa607ab6", "type": "challenged_by", "weight": 1 }, { "id": "edge-4d396a677bac", "source": "thesis-d0de40066b", "target": "event-589931146907", "type": "challenged_by", "weight": 1 }, { "id": "edge-c2651b5cb332", "source": "thesis-d0de40066b", "target": "event-3cbc7fe600ac", "type": "challenged_by", "weight": 1 }, { "id": "edge-472722aa6b2e", "source": "thesis-d0de40066b", "target": "event-d0719a325ab9", "type": "supported_by", "weight": 1 }, { "id": "edge-e1eb3f86b548", "source": "thesis-d0de40066b", "target": "event-8d4528fa2799", "type": "challenged_by", "weight": 1 }, { "id": "edge-c863e6adcd13", "source": "thesis-d0de40066b", "target": "event-c32b9684d1f4", "type": "challenged_by", "weight": 1 }, { "id": "edge-4e3538506efa", "source": "thesis-97a24f8796", "target": "stock-61e19d9ee8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-77dfa4d82585", "source": "thesis-97a24f8796", "target": "stock-b5733c9451", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f6b33b843b3c", "source": "thesis-97a24f8796", "target": "stock-8ee9008c2b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5ee27b316b75", "source": "thesis-97a24f8796", "target": "stock-59e40ddab7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-4813bb24fcc7", "source": "thesis-97a24f8796", "target": "stock-2d1c22cd10", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0bf9e85429b9", "source": "thesis-97a24f8796", "target": "stock-2888611640", "type": "tracks_stock", "weight": 3 }, { "id": "edge-374e299a2f43", "source": "thesis-97a24f8796", "target": "term-a050178590", "type": "tracks_term", "weight": 2 }, { "id": "edge-52c5aacd2c2f", "source": "thesis-97a24f8796", "target": "term-6ac13ea9eb", "type": "tracks_term", "weight": 2 }, { "id": "edge-e8672d419de6", "source": "thesis-97a24f8796", "target": "term-585a293e96", "type": "tracks_term", "weight": 2 }, { "id": "edge-b9a09c26317d", "source": "thesis-97a24f8796", "target": "event-948ce8147538", "type": "challenged_by", "weight": 1 }, { "id": "edge-6cb70e3f70b7", "source": "thesis-97a24f8796", "target": "event-e4aab152e0c7", "type": "challenged_by", "weight": 1 }, { "id": "edge-e56ff6eb0008", "source": "thesis-97a24f8796", "target": "event-9de344767759", "type": "challenged_by", "weight": 1 }, { "id": "edge-e3ea890a9d81", "source": "thesis-97a24f8796", "target": "event-8cdd707be994", "type": "challenged_by", "weight": 1 }, { "id": "edge-09029379036e", "source": "thesis-97a24f8796", "target": "event-a51f5aa78ca6", "type": "challenged_by", "weight": 1 }, { "id": "edge-bf4fc5101486", "source": "thesis-97a24f8796", "target": "event-a18a385d1209", "type": "challenged_by", "weight": 1 }, { "id": "edge-af2e4afecf7d", "source": "thesis-97a24f8796", "target": "event-ca73498b2b6f", "type": "supported_by", "weight": 1 }, { "id": "edge-2ffb20ff6c4b", "source": "thesis-97a24f8796", "target": "event-e18a1790bbe8", "type": "challenged_by", "weight": 1 }, { "id": "edge-031a1a82ff1e", "source": "thesis-f57e888558", "target": "stock-dcd4dbe710", "type": "tracks_stock", "weight": 3 }, { "id": "edge-67ee131f9304", "source": "thesis-f57e888558", "target": "stock-c39cc2da75", "type": "tracks_stock", "weight": 3 }, { "id": "edge-140785f043f3", "source": "thesis-f57e888558", "target": "stock-517b2f6590", "type": "tracks_stock", "weight": 3 }, { "id": "edge-cb0e7334a027", "source": "thesis-f57e888558", "target": "stock-f57a081cf3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-00fe47b77a5e", "source": "thesis-f57e888558", "target": "stock-510de57b5f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0074becc90cd", "source": "thesis-f57e888558", "target": "stock-68a288585c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-eb84daf9fc92", "source": "thesis-f57e888558", "target": "term-2a78025de6", "type": "tracks_term", "weight": 2 }, { "id": "edge-964a8519c11e", "source": "thesis-f57e888558", "target": "term-3165df34fc", "type": "tracks_term", "weight": 2 }, { "id": "edge-016b4af4d294", "source": "thesis-f57e888558", "target": "term-c6836f7ad1", "type": "tracks_term", "weight": 2 }, { "id": "edge-fd52ca200e29", "source": "thesis-f57e888558", "target": "event-81eff6c97d3d", "type": "challenged_by", "weight": 1 }, { "id": "edge-ede6912ce4cf", "source": "thesis-f57e888558", "target": "event-6dcab2c0827c", "type": "supported_by", "weight": 1 }, { "id": "edge-91cb201c1bab", "source": "thesis-f57e888558", "target": "event-2916df1947a2", "type": "supported_by", "weight": 1 }, { "id": "edge-51916bf7aad6", "source": "thesis-f57e888558", "target": "event-512eeeadc0c9", "type": "challenged_by", "weight": 1 }, { "id": "edge-80db8a6e14ac", "source": "thesis-f57e888558", "target": "event-5d7ba6993082", "type": "challenged_by", "weight": 1 }, { "id": "edge-e86cc6881a91", "source": "thesis-f57e888558", "target": "event-4fde4a0cb3fb", "type": "supported_by", "weight": 1 }, { "id": "edge-823a043c7a3a", "source": "thesis-f57e888558", "target": "event-07a388e99618", "type": "supported_by", "weight": 1 }, { "id": "edge-828a0d630e51", "source": "thesis-f57e888558", "target": "event-2a4c925e5939", "type": "supported_by", "weight": 1 }, { "id": "edge-ad94be92cce3", "source": "thesis-5fd96966de", "target": "stock-61e19d9ee8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0bf485aeef61", "source": "thesis-5fd96966de", "target": "stock-b5733c9451", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7e427da23d82", "source": "thesis-5fd96966de", "target": "stock-8ee9008c2b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-58c0330845db", "source": "thesis-5fd96966de", "target": "stock-59e40ddab7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f90cc4a49f00", "source": "thesis-5fd96966de", "target": "stock-2d1c22cd10", "type": "tracks_stock", "weight": 3 }, { "id": "edge-3f6b04463c73", "source": "thesis-5fd96966de", "target": "stock-2888611640", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6b23263a75c6", "source": "thesis-5fd96966de", "target": "term-1582c27783", "type": "tracks_term", "weight": 2 }, { "id": "edge-e119c5e74f03", "source": "thesis-5fd96966de", "target": "term-a050178590", "type": "tracks_term", "weight": 2 }, { "id": "edge-760d3d242134", "source": "thesis-5fd96966de", "target": "term-ef242752b3", "type": "tracks_term", "weight": 2 }, { "id": "edge-912b9fbf9747", "source": "thesis-5fd96966de", "target": "term-690156f98a", "type": "tracks_term", "weight": 2 }, { "id": "edge-1f675060fa2e", "source": "thesis-5fd96966de", "target": "term-420421da52", "type": "tracks_term", "weight": 2 }, { "id": "edge-019060388e62", "source": "thesis-5fd96966de", "target": "term-d5d45771d8", "type": "tracks_term", "weight": 2 }, { "id": "edge-9394e44bbf21", "source": "thesis-5fd96966de", "target": "term-5eca243822", "type": "tracks_term", "weight": 2 }, { "id": "edge-cd0e8179d369", "source": "thesis-5fd96966de", "target": "term-ca2444c4a9", "type": "tracks_term", "weight": 2 }, { "id": "edge-7f17242c4813", "source": "thesis-5fd96966de", "target": "event-4a4f1a776751", "type": "challenged_by", "weight": 1 }, { "id": "edge-4bd4e341d8cb", "source": "thesis-5fd96966de", "target": "event-27b160f1ac21", "type": "challenged_by", "weight": 1 }, { "id": "edge-513d6bc71726", "source": "thesis-5fd96966de", "target": "event-1b3560e2c773", "type": "challenged_by", "weight": 1 }, { "id": "edge-99f39e0a2cd6", "source": "thesis-5fd96966de", "target": "event-03a3342eee44", "type": "challenged_by", "weight": 1 }, { "id": "edge-a83c91bbd594", "source": "thesis-5fd96966de", "target": "event-e474a4c6993e", "type": "challenged_by", "weight": 1 }, { "id": "edge-fc3ca4fb0271", "source": "thesis-5fd96966de", "target": "event-8fbbb8b683e2", "type": "challenged_by", "weight": 1 }, { "id": "edge-8079213fb6a3", "source": "thesis-5fd96966de", "target": "event-cace5af8b9a6", "type": "challenged_by", "weight": 1 }, { "id": "edge-5ce02932100a", "source": "thesis-5fd96966de", "target": "event-bcd18f03af4f", "type": "challenged_by", "weight": 1 }, { "id": "edge-f1b3684e42c8", "source": "thesis-71476ded81", "target": "stock-0716b17bb6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-90f43d59f824", "source": "thesis-71476ded81", "target": "stock-6ece0f6e09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9e2e8b65339e", "source": "thesis-71476ded81", "target": "stock-bb47079d51", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a8e3be1d358d", "source": "thesis-71476ded81", "target": "stock-8e7522f842", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d80d169418a5", "source": "thesis-71476ded81", "target": "stock-81aa4e9efc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ea3e9cc139ad", "source": "thesis-71476ded81", "target": "stock-083dc3a2b3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0b7407ec2263", "source": "thesis-71476ded81", "target": "term-54b759dba2", "type": "tracks_term", "weight": 2 }, { "id": "edge-22885af7041f", "source": "thesis-71476ded81", "target": "term-34c2d1dc4e", "type": "tracks_term", "weight": 2 }, { "id": "edge-c92c62ab2016", "source": "thesis-71476ded81", "target": "event-695e0c9919e3", "type": "challenged_by", "weight": 1 }, { "id": "edge-abcef27ad90a", "source": "thesis-71476ded81", "target": "event-9356dc9fa383", "type": "challenged_by", "weight": 1 }, { "id": "edge-e79d54a4dc4b", "source": "thesis-71476ded81", "target": "event-e746b90c0092", "type": "challenged_by", "weight": 1 }, { "id": "edge-7fa6f8775b7e", "source": "thesis-71476ded81", "target": "event-ab05993142bf", "type": "challenged_by", "weight": 1 }, { "id": "edge-e1c50db05de8", "source": "thesis-71476ded81", "target": "event-0424aae47a1f", "type": "challenged_by", "weight": 1 }, { "id": "edge-465c7d9d65f7", "source": "thesis-71476ded81", "target": "event-f6b15f4fabb0", "type": "challenged_by", "weight": 1 }, { "id": "edge-9856bf58c9fb", "source": "thesis-71476ded81", "target": "event-69a280c78b7d", "type": "challenged_by", "weight": 1 }, { "id": "edge-6366e24ceee7", "source": "thesis-71476ded81", "target": "event-1148f08264fb", "type": "challenged_by", "weight": 1 }, { "id": "edge-b156e3fe3837", "source": "thesis-bb8f3d8ee7", "target": "stock-dcd4dbe710", "type": "tracks_stock", "weight": 3 }, { "id": "edge-2ce37788ffb0", "source": "thesis-bb8f3d8ee7", "target": "stock-c39cc2da75", "type": "tracks_stock", "weight": 3 }, { "id": "edge-4d5be54b0a8d", "source": "thesis-bb8f3d8ee7", "target": "stock-517b2f6590", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d4e4fcdfeb84", "source": "thesis-bb8f3d8ee7", "target": "stock-f57a081cf3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-769d2e68a2ab", "source": "thesis-bb8f3d8ee7", "target": "stock-510de57b5f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f14dce2ab30a", "source": "thesis-bb8f3d8ee7", "target": "stock-68a288585c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5a08614a45bf", "source": "thesis-bb8f3d8ee7", "target": "term-60e981724c", "type": "tracks_term", "weight": 2 }, { "id": "edge-5c38111b6613", "source": "thesis-bb8f3d8ee7", "target": "term-a4b8abb418", "type": "tracks_term", "weight": 2 }, { "id": "edge-789e7fec3340", "source": "thesis-bb8f3d8ee7", "target": "term-dcc56252e4", "type": "tracks_term", "weight": 2 }, { "id": "edge-a4ac3132b9b2", "source": "thesis-bb8f3d8ee7", "target": "term-f71ccd3834", "type": "tracks_term", "weight": 2 }, { "id": "edge-f5502deed3a8", "source": "thesis-bb8f3d8ee7", "target": "term-c484297089", "type": "tracks_term", "weight": 2 }, { "id": "edge-c1828ad02e9a", "source": "thesis-bb8f3d8ee7", "target": "term-a6bb0d8c7a", "type": "tracks_term", "weight": 2 }, { "id": "edge-f1aada9cb9f4", "source": "thesis-bb8f3d8ee7", "target": "term-4364651c17", "type": "tracks_term", "weight": 2 }, { "id": "edge-06c1b79ac5ec", "source": "thesis-bb8f3d8ee7", "target": "term-117d8f914d", "type": "tracks_term", "weight": 2 }, { "id": "edge-2bfa33b3c32f", "source": "thesis-bb8f3d8ee7", "target": "event-2285c4e9718c", "type": "challenged_by", "weight": 1 }, { "id": "edge-f62c7240d143", "source": "thesis-bb8f3d8ee7", "target": "event-38e91753ba34", "type": "challenged_by", "weight": 1 }, { "id": "edge-21d02852bdb2", "source": "thesis-bb8f3d8ee7", "target": "event-73d04a867e42", "type": "challenged_by", "weight": 1 }, { "id": "edge-0515209f0c8c", "source": "thesis-bb8f3d8ee7", "target": "event-de2f9bcf10b1", "type": "challenged_by", "weight": 1 }, { "id": "edge-248e2ad3d980", "source": "thesis-bb8f3d8ee7", "target": "event-c066f01581da", "type": "challenged_by", "weight": 1 }, { "id": "edge-677a5fd25794", "source": "thesis-bb8f3d8ee7", "target": "event-25c16ae53780", "type": "challenged_by", "weight": 1 }, { "id": "edge-1a735e46d5f2", "source": "thesis-bb8f3d8ee7", "target": "event-69a1d0233e78", "type": "challenged_by", "weight": 1 }, { "id": "edge-76c55d65e6ce", "source": "thesis-bb8f3d8ee7", "target": "event-b3492849393c", "type": "challenged_by", "weight": 1 }, { "id": "edge-aa182c39ddbd", "source": "thesis-62e0155999", "target": "stock-06214b33ef", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c8ce8f4f2ff6", "source": "thesis-62e0155999", "target": "stock-4c3e22d6f2", "type": "tracks_stock", "weight": 3 }, { "id": "edge-31385e3e6a73", "source": "thesis-62e0155999", "target": "stock-d1c8471480", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b6a380e4b88c", "source": "thesis-62e0155999", "target": "stock-5829704445", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0d64c2247b08", "source": "thesis-62e0155999", "target": "stock-0ca8634002", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d3ad1fddd511", "source": "thesis-62e0155999", "target": "stock-426177902e", "type": "tracks_stock", "weight": 3 }, { "id": "edge-147339b5995f", "source": "thesis-62e0155999", "target": "term-afd623b148", "type": "tracks_term", "weight": 2 }, { "id": "edge-50a0eb927ebe", "source": "thesis-62e0155999", "target": "term-60e981724c", "type": "tracks_term", "weight": 2 }, { "id": "edge-f44dc8898482", "source": "thesis-62e0155999", "target": "term-f71ccd3834", "type": "tracks_term", "weight": 2 }, { "id": "edge-0c692490530b", "source": "thesis-62e0155999", "target": "term-dcdd3b5c45", "type": "tracks_term", "weight": 2 }, { "id": "edge-15d39b19f631", "source": "thesis-62e0155999", "target": "term-c484297089", "type": "tracks_term", "weight": 2 }, { "id": "edge-263f5d1b9eb5", "source": "thesis-62e0155999", "target": "term-b650f21f60", "type": "tracks_term", "weight": 2 }, { "id": "edge-3c61a66030f3", "source": "thesis-62e0155999", "target": "term-f2373e2b95", "type": "tracks_term", "weight": 2 }, { "id": "edge-c3efbf87d4dd", "source": "thesis-62e0155999", "target": "term-5689ce9e64", "type": "tracks_term", "weight": 2 }, { "id": "edge-7e8e424c7c3e", "source": "thesis-62e0155999", "target": "event-5f84de78735c", "type": "challenged_by", "weight": 1 }, { "id": "edge-dd17de6fe7fb", "source": "thesis-62e0155999", "target": "event-6f6df3b9c78b", "type": "challenged_by", "weight": 1 }, { "id": "edge-66d5d6bbd39a", "source": "thesis-62e0155999", "target": "event-a8ed72a894c0", "type": "challenged_by", "weight": 1 }, { "id": "edge-b63855f6141d", "source": "thesis-62e0155999", "target": "event-7fdb637a7cc9", "type": "challenged_by", "weight": 1 }, { "id": "edge-4a205db3b814", "source": "thesis-62e0155999", "target": "event-3454262b6ac8", "type": "challenged_by", "weight": 1 }, { "id": "edge-2d294455030c", "source": "thesis-62e0155999", "target": "event-852961ede0c3", "type": "challenged_by", "weight": 1 }, { "id": "edge-12d3ff781c89", "source": "thesis-62e0155999", "target": "event-fb7476f7603e", "type": "supported_by", "weight": 1 }, { "id": "edge-6a4ae0a25a14", "source": "thesis-62e0155999", "target": "event-f9e03c6463ba", "type": "challenged_by", "weight": 1 }, { "id": "edge-743e5b0645c2", "source": "thesis-5cb77a9f52", "target": "stock-dcd4dbe710", "type": "tracks_stock", "weight": 3 }, { "id": "edge-214e0500cd3e", "source": "thesis-5cb77a9f52", "target": "stock-c39cc2da75", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9f013bf6a17f", "source": "thesis-5cb77a9f52", "target": "stock-517b2f6590", "type": "tracks_stock", "weight": 3 }, { "id": "edge-83627a1ac21d", "source": "thesis-5cb77a9f52", "target": "stock-f57a081cf3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-fa8939fa6d94", "source": "thesis-5cb77a9f52", "target": "stock-510de57b5f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c111b65f3d5d", "source": "thesis-5cb77a9f52", "target": "stock-68a288585c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d73cb0761537", "source": "thesis-5cb77a9f52", "target": "term-1582c27783", "type": "tracks_term", "weight": 2 }, { "id": "edge-1aee630b5e0f", "source": "thesis-5cb77a9f52", "target": "term-f71ccd3834", "type": "tracks_term", "weight": 2 }, { "id": "edge-c0f38cb6a59e", "source": "thesis-5cb77a9f52", "target": "term-ef242752b3", "type": "tracks_term", "weight": 2 }, { "id": "edge-c49929decadc", "source": "thesis-5cb77a9f52", "target": "term-690156f98a", "type": "tracks_term", "weight": 2 }, { "id": "edge-a96621e30d85", "source": "thesis-5cb77a9f52", "target": "term-420421da52", "type": "tracks_term", "weight": 2 }, { "id": "edge-e3ab0e591ac0", "source": "thesis-5cb77a9f52", "target": "term-d5d45771d8", "type": "tracks_term", "weight": 2 }, { "id": "edge-262a1f42310f", "source": "thesis-5cb77a9f52", "target": "term-cd256779c8", "type": "tracks_term", "weight": 2 }, { "id": "edge-d6d846f14510", "source": "thesis-5cb77a9f52", "target": "term-5a513a4a57", "type": "tracks_term", "weight": 2 }, { "id": "edge-9d3c62b4a4c5", "source": "thesis-5cb77a9f52", "target": "event-5046138b843d", "type": "challenged_by", "weight": 1 }, { "id": "edge-6e93c3efd9b8", "source": "thesis-5cb77a9f52", "target": "event-16b7fd5a17d0", "type": "challenged_by", "weight": 1 }, { "id": "edge-9cec8ec7ff09", "source": "thesis-5cb77a9f52", "target": "event-3743442f4403", "type": "challenged_by", "weight": 1 }, { "id": "edge-d4fd52665af3", "source": "thesis-5cb77a9f52", "target": "event-1484a7fd044b", "type": "challenged_by", "weight": 1 }, { "id": "edge-569a0a951e2b", "source": "thesis-5cb77a9f52", "target": "event-6d2761252824", "type": "challenged_by", "weight": 1 }, { "id": "edge-c07ecca48a49", "source": "thesis-5cb77a9f52", "target": "event-ac53fe2d839b", "type": "challenged_by", "weight": 1 }, { "id": "edge-dfcf2096c205", "source": "thesis-5cb77a9f52", "target": "event-123f6db7413e", "type": "challenged_by", "weight": 1 }, { "id": "edge-c5d5b993e3a1", "source": "thesis-5cb77a9f52", "target": "event-6b194bcc9466", "type": "challenged_by", "weight": 1 }, { "id": "edge-4d061deb4d40", "source": "thesis-e4a70ed58a", "target": "stock-06214b33ef", "type": "tracks_stock", "weight": 3 }, { "id": "edge-734bea074e0e", "source": "thesis-e4a70ed58a", "target": "stock-4c3e22d6f2", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0912e6b4320c", "source": "thesis-e4a70ed58a", "target": "stock-d1c8471480", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9d108b6cba9f", "source": "thesis-e4a70ed58a", "target": "stock-5829704445", "type": "tracks_stock", "weight": 3 }, { "id": "edge-39575581bdf5", "source": "thesis-e4a70ed58a", "target": "stock-0ca8634002", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ebea6fe6f344", "source": "thesis-e4a70ed58a", "target": "stock-426177902e", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a92760bd49a4", "source": "thesis-e4a70ed58a", "target": "term-51360b078e", "type": "tracks_term", "weight": 2 }, { "id": "edge-ee0d9f808a1d", "source": "thesis-e4a70ed58a", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-732b80b47216", "source": "thesis-e4a70ed58a", "target": "term-dcdd3b5c45", "type": "tracks_term", "weight": 2 }, { "id": "edge-0a1e31d982a1", "source": "thesis-e4a70ed58a", "target": "term-9196f0338a", "type": "tracks_term", "weight": 2 }, { "id": "edge-6e2017a97273", "source": "thesis-e4a70ed58a", "target": "term-0da46a7fbf", "type": "tracks_term", "weight": 2 }, { "id": "edge-a1b7fba0228d", "source": "thesis-e4a70ed58a", "target": "term-1034e2871a", "type": "tracks_term", "weight": 2 }, { "id": "edge-b069480d0dcd", "source": "thesis-e4a70ed58a", "target": "term-b650f21f60", "type": "tracks_term", "weight": 2 }, { "id": "edge-1eceb7989999", "source": "thesis-e4a70ed58a", "target": "term-ce842dfe73", "type": "tracks_term", "weight": 2 }, { "id": "edge-ba999fbba4ce", "source": "thesis-e4a70ed58a", "target": "event-97d9a540f95d", "type": "challenged_by", "weight": 1 }, { "id": "edge-ba41816ef60b", "source": "thesis-e4a70ed58a", "target": "event-729b32d56f57", "type": "challenged_by", "weight": 1 }, { "id": "edge-b2aad23abe2c", "source": "thesis-e4a70ed58a", "target": "event-94fa276aa0a9", "type": "challenged_by", "weight": 1 }, { "id": "edge-a6b6b1327286", "source": "thesis-e4a70ed58a", "target": "event-2d3db6a843e5", "type": "supported_by", "weight": 1 }, { "id": "edge-b75385b10479", "source": "thesis-e4a70ed58a", "target": "event-8d0c32819f6f", "type": "challenged_by", "weight": 1 }, { "id": "edge-0da02b5a4c51", "source": "thesis-e4a70ed58a", "target": "event-d37b7703b6b3", "type": "challenged_by", "weight": 1 }, { "id": "edge-af761ea1345e", "source": "thesis-e4a70ed58a", "target": "event-e1543b664125", "type": "challenged_by", "weight": 1 }, { "id": "edge-53eb3d700b52", "source": "thesis-e4a70ed58a", "target": "event-f24f3fc66873", "type": "challenged_by", "weight": 1 }, { "id": "edge-7b9a079d11a2", "source": "thesis-05c3f6f33c", "target": "stock-426219f2bc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-1130f34c8197", "source": "thesis-05c3f6f33c", "target": "stock-cb15b056a5", "type": "tracks_stock", "weight": 3 }, { "id": "edge-82452fbe9d15", "source": "thesis-05c3f6f33c", "target": "stock-b78cca2401", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7b2d298b46a2", "source": "thesis-05c3f6f33c", "target": "stock-a24c631b90", "type": "tracks_stock", "weight": 3 }, { "id": "edge-87fec8575dec", "source": "thesis-05c3f6f33c", "target": "stock-6b03c0b5e7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d36ba99c9a23", "source": "thesis-05c3f6f33c", "target": "stock-8ed9c54f09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-4b763192d626", "source": "thesis-05c3f6f33c", "target": "term-0890f45668", "type": "tracks_term", "weight": 2 }, { "id": "edge-b97824cada8a", "source": "thesis-05c3f6f33c", "target": "term-7834502344", "type": "tracks_term", "weight": 2 }, { "id": "edge-faf2c6aeb8fb", "source": "thesis-05c3f6f33c", "target": "term-4de0127946", "type": "tracks_term", "weight": 2 }, { "id": "edge-f0b56d1b7ec9", "source": "thesis-05c3f6f33c", "target": "term-00fb5022b7", "type": "tracks_term", "weight": 2 }, { "id": "edge-6b56f1444cb4", "source": "thesis-05c3f6f33c", "target": "term-e01b27172a", "type": "tracks_term", "weight": 2 }, { "id": "edge-670cdbf9bde3", "source": "thesis-05c3f6f33c", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-08b6576ba9ea", "source": "thesis-05c3f6f33c", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-fc695c4d3eca", "source": "thesis-05c3f6f33c", "target": "term-2de5d0f518", "type": "tracks_term", "weight": 2 }, { "id": "edge-d867ed8fff17", "source": "thesis-05c3f6f33c", "target": "event-7d8ce389087c", "type": "challenged_by", "weight": 1 }, { "id": "edge-686949757fc0", "source": "thesis-05c3f6f33c", "target": "event-33e738eeb030", "type": "challenged_by", "weight": 1 }, { "id": "edge-f7346120017d", "source": "thesis-05c3f6f33c", "target": "event-05592b0b7f2f", "type": "supported_by", "weight": 1 }, { "id": "edge-6ef35ff28b57", "source": "thesis-05c3f6f33c", "target": "event-658c48f3885c", "type": "challenged_by", "weight": 1 }, { "id": "edge-46496aab51df", "source": "thesis-05c3f6f33c", "target": "event-6c87297ae466", "type": "challenged_by", "weight": 1 }, { "id": "edge-5b2c2bdcd44e", "source": "thesis-05c3f6f33c", "target": "event-dad343bcce54", "type": "challenged_by", "weight": 1 }, { "id": "edge-bf86293fb756", "source": "thesis-05c3f6f33c", "target": "event-59dc5918dc57", "type": "challenged_by", "weight": 1 }, { "id": "edge-76876c5d46f0", "source": "thesis-05c3f6f33c", "target": "event-b4d468dbca31", "type": "challenged_by", "weight": 1 }, { "id": "edge-e06183b789f8", "source": "thesis-72097390dd", "target": "stock-426219f2bc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7330a9a01f8d", "source": "thesis-72097390dd", "target": "stock-cb15b056a5", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9bda7f1df966", "source": "thesis-72097390dd", "target": "stock-b78cca2401", "type": "tracks_stock", "weight": 3 }, { "id": "edge-1d37a625b693", "source": "thesis-72097390dd", "target": "stock-a24c631b90", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9201f02cf0f6", "source": "thesis-72097390dd", "target": "stock-6b03c0b5e7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f0ade2a80e56", "source": "thesis-72097390dd", "target": "stock-8ed9c54f09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c3a9b98d1cf7", "source": "thesis-72097390dd", "target": "term-09331fca77", "type": "tracks_term", "weight": 2 }, { "id": "edge-55155ea6c459", "source": "thesis-72097390dd", "target": "term-c7bac46904", "type": "tracks_term", "weight": 2 }, { "id": "edge-a0e653fb51f2", "source": "thesis-72097390dd", "target": "term-ee72426269", "type": "tracks_term", "weight": 2 }, { "id": "edge-c375bc02f6ce", "source": "thesis-72097390dd", "target": "term-49d2dc07ac", "type": "tracks_term", "weight": 2 }, { "id": "edge-74350d0f289e", "source": "thesis-72097390dd", "target": "term-4f749de7c2", "type": "tracks_term", "weight": 2 }, { "id": "edge-a9c578a832ee", "source": "thesis-72097390dd", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-5bdda8387a64", "source": "thesis-72097390dd", "target": "term-04f7acfcd2", "type": "tracks_term", "weight": 2 }, { "id": "edge-05ed19806c12", "source": "thesis-72097390dd", "target": "term-344e9a768f", "type": "tracks_term", "weight": 2 }, { "id": "edge-f249c90dc298", "source": "thesis-72097390dd", "target": "event-aea8c9f9f834", "type": "challenged_by", "weight": 1 }, { "id": "edge-181060de812c", "source": "thesis-72097390dd", "target": "event-ed61e691b018", "type": "challenged_by", "weight": 1 }, { "id": "edge-477085ebbef9", "source": "thesis-72097390dd", "target": "event-0cbb20ece214", "type": "challenged_by", "weight": 1 }, { "id": "edge-0e5756a50e81", "source": "thesis-72097390dd", "target": "event-59b919e4332d", "type": "supported_by", "weight": 1 }, { "id": "edge-23fcd0679043", "source": "thesis-72097390dd", "target": "event-f795702f985c", "type": "challenged_by", "weight": 1 }, { "id": "edge-ba7eeb00f6d6", "source": "thesis-72097390dd", "target": "event-af52a9e95b31", "type": "challenged_by", "weight": 1 }, { "id": "edge-8d12280972ce", "source": "thesis-72097390dd", "target": "event-3fd26cbe5656", "type": "challenged_by", "weight": 1 }, { "id": "edge-afb563e756e4", "source": "thesis-72097390dd", "target": "event-840f94a7a705", "type": "challenged_by", "weight": 1 }, { "id": "edge-8614e36ec1bf", "source": "thesis-e1d3ec2fed", "target": "stock-61e19d9ee8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-640b55125dd5", "source": "thesis-e1d3ec2fed", "target": "stock-b5733c9451", "type": "tracks_stock", "weight": 3 }, { "id": "edge-09566dca573c", "source": "thesis-e1d3ec2fed", "target": "stock-8ee9008c2b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-94f7f8219c52", "source": "thesis-e1d3ec2fed", "target": "stock-59e40ddab7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9190577248d2", "source": "thesis-e1d3ec2fed", "target": "stock-2d1c22cd10", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a0cf3defe466", "source": "thesis-e1d3ec2fed", "target": "stock-2888611640", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7e2bb409a4e9", "source": "thesis-e1d3ec2fed", "target": "term-5e71d16a67", "type": "tracks_term", "weight": 2 }, { "id": "edge-3c7fdee1a6a3", "source": "thesis-e1d3ec2fed", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-727115243232", "source": "thesis-e1d3ec2fed", "target": "term-04f7acfcd2", "type": "tracks_term", "weight": 2 }, { "id": "edge-d88607d4f7a4", "source": "thesis-e1d3ec2fed", "target": "term-2359bb6154", "type": "tracks_term", "weight": 2 }, { "id": "edge-831fb39bc5f6", "source": "thesis-e1d3ec2fed", "target": "term-8e289bbb4a", "type": "tracks_term", "weight": 2 }, { "id": "edge-c98a6f287869", "source": "thesis-e1d3ec2fed", "target": "term-2ddcf6fb42", "type": "tracks_term", "weight": 2 }, { "id": "edge-cff361376f58", "source": "thesis-e1d3ec2fed", "target": "term-ef242752b3", "type": "tracks_term", "weight": 2 }, { "id": "edge-8164ba31b2f9", "source": "thesis-e1d3ec2fed", "target": "term-e1506406a5", "type": "tracks_term", "weight": 2 }, { "id": "edge-018cbe74e170", "source": "thesis-e1d3ec2fed", "target": "event-1a91636e7402", "type": "challenged_by", "weight": 1 }, { "id": "edge-9a55d77a2782", "source": "thesis-e1d3ec2fed", "target": "event-aa924e4a514d", "type": "challenged_by", "weight": 1 }, { "id": "edge-dd7c60121912", "source": "thesis-e1d3ec2fed", "target": "event-8cda15ce88bd", "type": "challenged_by", "weight": 1 }, { "id": "edge-8aec4d813c36", "source": "thesis-e1d3ec2fed", "target": "event-46fd93307b2c", "type": "challenged_by", "weight": 1 }, { "id": "edge-6ec82daf1bcd", "source": "thesis-e1d3ec2fed", "target": "event-237a8864ecca", "type": "challenged_by", "weight": 1 }, { "id": "edge-b86a0322645a", "source": "thesis-e1d3ec2fed", "target": "event-ec7a4714681b", "type": "challenged_by", "weight": 1 }, { "id": "edge-45a0249be3e5", "source": "thesis-e1d3ec2fed", "target": "event-9e4440057d6d", "type": "challenged_by", "weight": 1 }, { "id": "edge-77b7e911ebab", "source": "thesis-e1d3ec2fed", "target": "event-9eb33a52bb69", "type": "challenged_by", "weight": 1 }, { "id": "edge-97a6b0d30729", "source": "thesis-33e46fee64", "target": "stock-426219f2bc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-4fefc9451700", "source": "thesis-33e46fee64", "target": "stock-cb15b056a5", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ab3c3fa04122", "source": "thesis-33e46fee64", "target": "stock-b78cca2401", "type": "tracks_stock", "weight": 3 }, { "id": "edge-88aa6f5cf843", "source": "thesis-33e46fee64", "target": "stock-a24c631b90", "type": "tracks_stock", "weight": 3 }, { "id": "edge-008fe19f9ce5", "source": "thesis-33e46fee64", "target": "stock-6b03c0b5e7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7c1abf1c4621", "source": "thesis-33e46fee64", "target": "stock-8ed9c54f09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-53eaed650c09", "source": "thesis-33e46fee64", "target": "term-0890f45668", "type": "tracks_term", "weight": 2 }, { "id": "edge-f39fed1fab55", "source": "thesis-33e46fee64", "target": "term-7834502344", "type": "tracks_term", "weight": 2 }, { "id": "edge-6adc51a7f10a", "source": "thesis-33e46fee64", "target": "term-4de0127946", "type": "tracks_term", "weight": 2 }, { "id": "edge-bee74a78a1db", "source": "thesis-33e46fee64", "target": "term-e01b27172a", "type": "tracks_term", "weight": 2 }, { "id": "edge-cbffb704cb6f", "source": "thesis-33e46fee64", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-8b73bec081a8", "source": "thesis-33e46fee64", "target": "term-17c5c517dc", "type": "tracks_term", "weight": 2 }, { "id": "edge-3732e5a65f83", "source": "thesis-33e46fee64", "target": "term-38e8415b91", "type": "tracks_term", "weight": 2 }, { "id": "edge-7ac3ad1c3e5f", "source": "thesis-33e46fee64", "target": "term-16296c2af8", "type": "tracks_term", "weight": 2 }, { "id": "edge-0b178104b96b", "source": "thesis-33e46fee64", "target": "event-c8a1621c2deb", "type": "challenged_by", "weight": 1 }, { "id": "edge-ed0cfc014caf", "source": "thesis-33e46fee64", "target": "event-88c9e5e35073", "type": "challenged_by", "weight": 1 }, { "id": "edge-b4e1cb4a3c41", "source": "thesis-33e46fee64", "target": "event-1ff6251f6df9", "type": "challenged_by", "weight": 1 }, { "id": "edge-13cf3215ae5d", "source": "thesis-33e46fee64", "target": "event-aac517316b81", "type": "challenged_by", "weight": 1 }, { "id": "edge-a05300224400", "source": "thesis-33e46fee64", "target": "event-be832fdbcc70", "type": "supported_by", "weight": 1 }, { "id": "edge-756d44faa94a", "source": "thesis-33e46fee64", "target": "event-de053451ea32", "type": "challenged_by", "weight": 1 }, { "id": "edge-58e3ccbc3d71", "source": "thesis-33e46fee64", "target": "event-8457dd2b4002", "type": "challenged_by", "weight": 1 }, { "id": "edge-36200772c677", "source": "thesis-33e46fee64", "target": "event-ad5e6f458ec2", "type": "challenged_by", "weight": 1 }, { "id": "edge-45a9a26229c8", "source": "thesis-f4ca874ce0", "target": "stock-06214b33ef", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d8a1781b6c7a", "source": "thesis-f4ca874ce0", "target": "stock-4c3e22d6f2", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e61f5f476943", "source": "thesis-f4ca874ce0", "target": "stock-d1c8471480", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ce14bc92474b", "source": "thesis-f4ca874ce0", "target": "stock-5829704445", "type": "tracks_stock", "weight": 3 }, { "id": "edge-439716d0437a", "source": "thesis-f4ca874ce0", "target": "stock-0ca8634002", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ef61768a61b0", "source": "thesis-f4ca874ce0", "target": "stock-426177902e", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ca1bc336bf6f", "source": "thesis-f4ca874ce0", "target": "term-11690d135b", "type": "tracks_term", "weight": 2 }, { "id": "edge-c13124ba369a", "source": "thesis-f4ca874ce0", "target": "term-51360b078e", "type": "tracks_term", "weight": 2 }, { "id": "edge-bd055e7924b5", "source": "thesis-f4ca874ce0", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-638f2b751bb7", "source": "thesis-f4ca874ce0", "target": "term-dcdd3b5c45", "type": "tracks_term", "weight": 2 }, { "id": "edge-9b9d78c850c8", "source": "thesis-f4ca874ce0", "target": "term-9196f0338a", "type": "tracks_term", "weight": 2 }, { "id": "edge-d480e0148705", "source": "thesis-f4ca874ce0", "target": "term-0da46a7fbf", "type": "tracks_term", "weight": 2 }, { "id": "edge-1b5f7e37eb42", "source": "thesis-f4ca874ce0", "target": "term-8d7bcb1cc8", "type": "tracks_term", "weight": 2 }, { "id": "edge-5a8918143840", "source": "thesis-f4ca874ce0", "target": "term-6cfed8e96e", "type": "tracks_term", "weight": 2 }, { "id": "edge-fc32edf24b08", "source": "thesis-f4ca874ce0", "target": "event-9d4d3683cde4", "type": "challenged_by", "weight": 1 }, { "id": "edge-02cad72d90f7", "source": "thesis-f4ca874ce0", "target": "event-8bf65e63f5d9", "type": "challenged_by", "weight": 1 }, { "id": "edge-d2519724bc73", "source": "thesis-f4ca874ce0", "target": "event-d0eacf5594d5", "type": "challenged_by", "weight": 1 }, { "id": "edge-6e4cc42e9e2c", "source": "thesis-f4ca874ce0", "target": "event-f7509a147d21", "type": "challenged_by", "weight": 1 }, { "id": "edge-3def7b4b9241", "source": "thesis-f4ca874ce0", "target": "event-4b415c4732dd", "type": "challenged_by", "weight": 1 }, { "id": "edge-ae53dad78a86", "source": "thesis-f4ca874ce0", "target": "event-e895241c18d5", "type": "challenged_by", "weight": 1 }, { "id": "edge-2c0269af2cbb", "source": "thesis-f4ca874ce0", "target": "event-d3c2a51f9250", "type": "challenged_by", "weight": 1 }, { "id": "edge-b37b7241c714", "source": "thesis-f4ca874ce0", "target": "event-e53cf0a6d2c0", "type": "challenged_by", "weight": 1 }, { "id": "edge-32966421ea43", "source": "thesis-f1c88bcd88", "target": "stock-61e19d9ee8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-3bf5f3eaa89a", "source": "thesis-f1c88bcd88", "target": "stock-b5733c9451", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6266e2489f33", "source": "thesis-f1c88bcd88", "target": "stock-8ee9008c2b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-8db61fb5d855", "source": "thesis-f1c88bcd88", "target": "stock-59e40ddab7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0f1baa2cd20a", "source": "thesis-f1c88bcd88", "target": "stock-2d1c22cd10", "type": "tracks_stock", "weight": 3 }, { "id": "edge-4f91f9e9d29c", "source": "thesis-f1c88bcd88", "target": "stock-2888611640", "type": "tracks_stock", "weight": 3 }, { "id": "edge-755403c039e9", "source": "thesis-f1c88bcd88", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-47b609516002", "source": "thesis-f1c88bcd88", "target": "term-2359bb6154", "type": "tracks_term", "weight": 2 }, { "id": "edge-28f81620ccba", "source": "thesis-f1c88bcd88", "target": "term-2ddcf6fb42", "type": "tracks_term", "weight": 2 }, { "id": "edge-aece4392b204", "source": "thesis-f1c88bcd88", "target": "term-ef242752b3", "type": "tracks_term", "weight": 2 }, { "id": "edge-cc8e6eee5898", "source": "thesis-f1c88bcd88", "target": "term-e1506406a5", "type": "tracks_term", "weight": 2 }, { "id": "edge-5d3f3edf357c", "source": "thesis-f1c88bcd88", "target": "term-95b6a12d6b", "type": "tracks_term", "weight": 2 }, { "id": "edge-af890ef7f3bc", "source": "thesis-f1c88bcd88", "target": "term-62f606258e", "type": "tracks_term", "weight": 2 }, { "id": "edge-e8d4247ab7b8", "source": "thesis-f1c88bcd88", "target": "term-7bf50a24dc", "type": "tracks_term", "weight": 2 }, { "id": "edge-c1696766d2b0", "source": "thesis-f1c88bcd88", "target": "event-7f8747e4af76", "type": "challenged_by", "weight": 1 }, { "id": "edge-369618c64809", "source": "thesis-f1c88bcd88", "target": "event-d8bb9c5a6958", "type": "challenged_by", "weight": 1 }, { "id": "edge-d0f54d1ab4c9", "source": "thesis-f1c88bcd88", "target": "event-4621fe36c29a", "type": "challenged_by", "weight": 1 }, { "id": "edge-aa58e936e441", "source": "thesis-f1c88bcd88", "target": "event-7fc56702588d", "type": "challenged_by", "weight": 1 }, { "id": "edge-4ec21690caf7", "source": "thesis-f1c88bcd88", "target": "event-1a16afeda1df", "type": "challenged_by", "weight": 1 }, { "id": "edge-5c7fdf9721d2", "source": "thesis-f1c88bcd88", "target": "event-b49035a56e7b", "type": "challenged_by", "weight": 1 }, { "id": "edge-6829d97b05f8", "source": "thesis-f1c88bcd88", "target": "event-7c47750c4ccf", "type": "challenged_by", "weight": 1 }, { "id": "edge-1455eed75772", "source": "thesis-f1c88bcd88", "target": "event-ccf625fbdc7f", "type": "challenged_by", "weight": 1 }, { "id": "edge-838d82ffc9a3", "source": "thesis-11f0838e82", "target": "stock-dcd4dbe710", "type": "tracks_stock", "weight": 3 }, { "id": "edge-3da420c1aa37", "source": "thesis-11f0838e82", "target": "stock-c39cc2da75", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f7bde229cec2", "source": "thesis-11f0838e82", "target": "stock-517b2f6590", "type": "tracks_stock", "weight": 3 }, { "id": "edge-eb5525cc83eb", "source": "thesis-11f0838e82", "target": "stock-f57a081cf3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-79486fd82457", "source": "thesis-11f0838e82", "target": "stock-510de57b5f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6e0135be1dcc", "source": "thesis-11f0838e82", "target": "stock-68a288585c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d6f489de1cf1", "source": "thesis-11f0838e82", "target": "term-60e981724c", "type": "tracks_term", "weight": 2 }, { "id": "edge-082cf2f116d1", "source": "thesis-11f0838e82", "target": "term-f71ccd3834", "type": "tracks_term", "weight": 2 }, { "id": "edge-a44ce3693382", "source": "thesis-11f0838e82", "target": "term-c484297089", "type": "tracks_term", "weight": 2 }, { "id": "edge-c52422ce0d5a", "source": "thesis-11f0838e82", "target": "term-420421da52", "type": "tracks_term", "weight": 2 }, { "id": "edge-ee11ad7716bb", "source": "thesis-11f0838e82", "target": "term-d3e6d49230", "type": "tracks_term", "weight": 2 }, { "id": "edge-a8d9740b0a00", "source": "thesis-11f0838e82", "target": "term-a8ab3d5805", "type": "tracks_term", "weight": 2 }, { "id": "edge-cdac9f3bea3c", "source": "thesis-11f0838e82", "target": "term-84c7ec1db8", "type": "tracks_term", "weight": 2 }, { "id": "edge-479be61953bc", "source": "thesis-11f0838e82", "target": "term-ab72c9a79f", "type": "tracks_term", "weight": 2 }, { "id": "edge-17e326128af0", "source": "thesis-11f0838e82", "target": "event-2b4cc1d264ee", "type": "challenged_by", "weight": 1 }, { "id": "edge-96cf366ec86f", "source": "thesis-11f0838e82", "target": "event-40a613299c36", "type": "challenged_by", "weight": 1 }, { "id": "edge-e429587774a8", "source": "thesis-11f0838e82", "target": "event-b317b78bee1b", "type": "challenged_by", "weight": 1 }, { "id": "edge-c7f9e0adb25a", "source": "thesis-11f0838e82", "target": "event-e1ea5816ea30", "type": "supported_by", "weight": 1 }, { "id": "edge-35acebaa0c9b", "source": "thesis-11f0838e82", "target": "event-ca755df0d2aa", "type": "challenged_by", "weight": 1 }, { "id": "edge-e3fd51522d9f", "source": "thesis-11f0838e82", "target": "event-e63fc75078ed", "type": "challenged_by", "weight": 1 }, { "id": "edge-4d8d69a41de5", "source": "thesis-11f0838e82", "target": "event-a44f36791089", "type": "challenged_by", "weight": 1 }, { "id": "edge-23be60fbdc64", "source": "thesis-11f0838e82", "target": "event-a22ac2b8e95a", "type": "challenged_by", "weight": 1 }, { "id": "edge-426811011c2d", "source": "thesis-1495ab3592", "target": "stock-61e19d9ee8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-707dadb757ad", "source": "thesis-1495ab3592", "target": "stock-b5733c9451", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e8912f87eaad", "source": "thesis-1495ab3592", "target": "stock-8ee9008c2b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-1c8dc6dc2535", "source": "thesis-1495ab3592", "target": "stock-59e40ddab7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b5074c5ee086", "source": "thesis-1495ab3592", "target": "stock-2d1c22cd10", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a3ab03b2ad7a", "source": "thesis-1495ab3592", "target": "stock-2888611640", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ada29c006788", "source": "thesis-1495ab3592", "target": "term-d49897f8c4", "type": "tracks_term", "weight": 2 }, { "id": "edge-7449a0e7cb06", "source": "thesis-1495ab3592", "target": "term-2ddcf6fb42", "type": "tracks_term", "weight": 2 }, { "id": "edge-dae4fea52b30", "source": "thesis-1495ab3592", "target": "term-dcc56252e4", "type": "tracks_term", "weight": 2 }, { "id": "edge-141656345d04", "source": "thesis-1495ab3592", "target": "term-95feb4759b", "type": "tracks_term", "weight": 2 }, { "id": "edge-28baefea22c5", "source": "thesis-1495ab3592", "target": "term-f71ccd3834", "type": "tracks_term", "weight": 2 }, { "id": "edge-ee08ff36e9e2", "source": "thesis-1495ab3592", "target": "term-6001ec95f6", "type": "tracks_term", "weight": 2 }, { "id": "edge-b1de2cbc09be", "source": "thesis-1495ab3592", "target": "term-6b72da0e1f", "type": "tracks_term", "weight": 2 }, { "id": "edge-d4e053db2e97", "source": "thesis-1495ab3592", "target": "term-b012bdba47", "type": "tracks_term", "weight": 2 }, { "id": "edge-2c486979d76a", "source": "thesis-1495ab3592", "target": "event-9eee158c2486", "type": "challenged_by", "weight": 1 }, { "id": "edge-47a24012700a", "source": "thesis-1495ab3592", "target": "event-021e981facd6", "type": "supported_by", "weight": 1 }, { "id": "edge-bfbcb30650ce", "source": "thesis-1495ab3592", "target": "event-9e922bb9aeb3", "type": "challenged_by", "weight": 1 }, { "id": "edge-5731166929f2", "source": "thesis-1495ab3592", "target": "event-b3cd2b14f754", "type": "challenged_by", "weight": 1 }, { "id": "edge-e23b2f023a95", "source": "thesis-1495ab3592", "target": "event-4430394743d1", "type": "challenged_by", "weight": 1 }, { "id": "edge-a67d951ca49a", "source": "thesis-1495ab3592", "target": "event-5838928f9afe", "type": "challenged_by", "weight": 1 }, { "id": "edge-4ca589be16bd", "source": "thesis-1495ab3592", "target": "event-4bd3fa72cb5d", "type": "challenged_by", "weight": 1 }, { "id": "edge-305e0556c69a", "source": "thesis-1495ab3592", "target": "event-86c8d044449d", "type": "challenged_by", "weight": 1 }, { "id": "edge-66e7cd041c8d", "source": "thesis-2ec6b9d3ee", "target": "stock-56f9442441", "type": "tracks_stock", "weight": 3 }, { "id": "edge-fac15d7fddd6", "source": "thesis-2ec6b9d3ee", "target": "stock-c27c6260da", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d8388f00cbcc", "source": "thesis-2ec6b9d3ee", "target": "stock-01467aabbc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6be7f2f3fde9", "source": "thesis-2ec6b9d3ee", "target": "stock-0d3089cbbb", "type": "tracks_stock", "weight": 3 }, { "id": "edge-11c004ca7d1a", "source": "thesis-2ec6b9d3ee", "target": "stock-bb2a93f2a1", "type": "tracks_stock", "weight": 3 }, { "id": "edge-16db5e24bc65", "source": "thesis-2ec6b9d3ee", "target": "stock-b1ce5beca8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-05ca34bdf190", "source": "thesis-2ec6b9d3ee", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-f9ce208f38ce", "source": "thesis-2ec6b9d3ee", "target": "term-5846e1a424", "type": "tracks_term", "weight": 2 }, { "id": "edge-46b486e1010d", "source": "thesis-2ec6b9d3ee", "target": "term-56f9442441", "type": "tracks_term", "weight": 2 }, { "id": "edge-c33b96a7d185", "source": "thesis-2ec6b9d3ee", "target": "term-e21dcec424", "type": "tracks_term", "weight": 2 }, { "id": "edge-18ac7a8de7fb", "source": "thesis-2ec6b9d3ee", "target": "term-a56b9af9a7", "type": "tracks_term", "weight": 2 }, { "id": "edge-d5832b622a88", "source": "thesis-2ec6b9d3ee", "target": "term-dc4b4e203f", "type": "tracks_term", "weight": 2 }, { "id": "edge-5e65f0f23a83", "source": "thesis-2ec6b9d3ee", "target": "term-d01acdec3a", "type": "tracks_term", "weight": 2 }, { "id": "edge-9e393bb0c0b6", "source": "thesis-2ec6b9d3ee", "target": "term-3f49dbbfe0", "type": "tracks_term", "weight": 2 }, { "id": "edge-b0b7e775260b", "source": "thesis-2ec6b9d3ee", "target": "event-fde861637c5d", "type": "challenged_by", "weight": 1 }, { "id": "edge-025afbed9ea2", "source": "thesis-2ec6b9d3ee", "target": "event-5fcf74fe1bac", "type": "challenged_by", "weight": 1 }, { "id": "edge-c3f028c75f4d", "source": "thesis-2ec6b9d3ee", "target": "event-a221038d25d3", "type": "challenged_by", "weight": 1 }, { "id": "edge-5c51f2410a60", "source": "thesis-2ec6b9d3ee", "target": "event-924f7a7ac124", "type": "challenged_by", "weight": 1 }, { "id": "edge-90296a7d5114", "source": "thesis-2ec6b9d3ee", "target": "event-1620661c8a11", "type": "challenged_by", "weight": 1 }, { "id": "edge-16041fe4ac85", "source": "thesis-2ec6b9d3ee", "target": "event-ca328dc4d5c6", "type": "challenged_by", "weight": 1 }, { "id": "edge-2ebc72c20dd7", "source": "thesis-2ec6b9d3ee", "target": "event-e9251fdff30c", "type": "challenged_by", "weight": 1 }, { "id": "edge-2293dab4c340", "source": "thesis-2ec6b9d3ee", "target": "event-d567524ce374", "type": "supported_by", "weight": 1 }, { "id": "edge-abf20ea08905", "source": "thesis-1aaa752ed4", "target": "stock-61e19d9ee8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-254add4463ec", "source": "thesis-1aaa752ed4", "target": "stock-b5733c9451", "type": "tracks_stock", "weight": 3 }, { "id": "edge-438fca4a77d1", "source": "thesis-1aaa752ed4", "target": "stock-8ee9008c2b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6f74892171b4", "source": "thesis-1aaa752ed4", "target": "stock-59e40ddab7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e827c07cfe57", "source": "thesis-1aaa752ed4", "target": "stock-2d1c22cd10", "type": "tracks_stock", "weight": 3 }, { "id": "edge-830271e64d45", "source": "thesis-1aaa752ed4", "target": "stock-f960e0de27", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b624824258ce", "source": "thesis-1aaa752ed4", "target": "term-95feb4759b", "type": "tracks_term", "weight": 2 }, { "id": "edge-2abf9c89605b", "source": "thesis-1aaa752ed4", "target": "term-2ddcf6fb42", "type": "tracks_term", "weight": 2 }, { "id": "edge-ae7c85886bb0", "source": "thesis-1aaa752ed4", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-5e7c056aaf55", "source": "thesis-1aaa752ed4", "target": "term-0e2b23eeed", "type": "tracks_term", "weight": 2 }, { "id": "edge-009808cf06ad", "source": "thesis-1aaa752ed4", "target": "term-04f7acfcd2", "type": "tracks_term", "weight": 2 }, { "id": "edge-7a7501ed33c4", "source": "thesis-1aaa752ed4", "target": "term-6001ec95f6", "type": "tracks_term", "weight": 2 }, { "id": "edge-84860ec0f932", "source": "thesis-1aaa752ed4", "target": "term-2359bb6154", "type": "tracks_term", "weight": 2 }, { "id": "edge-1393e091e2a9", "source": "thesis-1aaa752ed4", "target": "term-641102ee2d", "type": "tracks_term", "weight": 2 }, { "id": "edge-67c3f825296e", "source": "thesis-1aaa752ed4", "target": "event-44e26486d853", "type": "challenged_by", "weight": 1 }, { "id": "edge-933fac1a25b1", "source": "thesis-1aaa752ed4", "target": "event-68c72c0e7850", "type": "challenged_by", "weight": 1 }, { "id": "edge-45119cfd95a8", "source": "thesis-1aaa752ed4", "target": "event-129b9febf3c3", "type": "challenged_by", "weight": 1 }, { "id": "edge-6ba05e5bbaa7", "source": "thesis-1aaa752ed4", "target": "event-d090e448ccf4", "type": "supported_by", "weight": 1 }, { "id": "edge-94d38c1efb3b", "source": "thesis-1aaa752ed4", "target": "event-264a019f7d04", "type": "challenged_by", "weight": 1 }, { "id": "edge-ba79b511f540", "source": "thesis-1aaa752ed4", "target": "event-5e06185b573c", "type": "challenged_by", "weight": 1 }, { "id": "edge-1cd5a27e755f", "source": "thesis-1aaa752ed4", "target": "event-1c3c0b0e917a", "type": "challenged_by", "weight": 1 }, { "id": "edge-fa261a728d6e", "source": "thesis-1aaa752ed4", "target": "event-9aec22bc96eb", "type": "challenged_by", "weight": 1 }, { "id": "edge-5739c2ace296", "source": "thesis-d4473d35e5", "target": "stock-bf5ac90619", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0fd01158aebe", "source": "thesis-d4473d35e5", "target": "stock-32f4bdcdf7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c6ab8b8eae91", "source": "thesis-d4473d35e5", "target": "stock-f3805b0eae", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e6ca884d42d7", "source": "thesis-d4473d35e5", "target": "stock-bb3ad25ff3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f0c5ec14ce80", "source": "thesis-d4473d35e5", "target": "stock-736b57465b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6271f071bc4f", "source": "thesis-d4473d35e5", "target": "stock-664acdf6ca", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7ac9dca2a25e", "source": "thesis-d4473d35e5", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-96225bb9660a", "source": "thesis-d4473d35e5", "target": "term-1ae05a2701", "type": "tracks_term", "weight": 2 }, { "id": "edge-0bca7b0ea417", "source": "thesis-d4473d35e5", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-46e34156807a", "source": "thesis-d4473d35e5", "target": "term-6840b7e095", "type": "tracks_term", "weight": 2 }, { "id": "edge-e4ae1838c082", "source": "thesis-d4473d35e5", "target": "term-c501935a6b", "type": "tracks_term", "weight": 2 }, { "id": "edge-96e701cb74bb", "source": "thesis-d4473d35e5", "target": "term-2c2240217b", "type": "tracks_term", "weight": 2 }, { "id": "edge-37a29993ce2f", "source": "thesis-d4473d35e5", "target": "term-1bb0b3a466", "type": "tracks_term", "weight": 2 }, { "id": "edge-206c9dbe412a", "source": "thesis-d4473d35e5", "target": "term-fe81f9390f", "type": "tracks_term", "weight": 2 }, { "id": "edge-81f1c116bb42", "source": "thesis-d4473d35e5", "target": "event-7480e0e340a9", "type": "challenged_by", "weight": 1 }, { "id": "edge-39537fdaafb5", "source": "thesis-d4473d35e5", "target": "event-ad107c8a25e4", "type": "challenged_by", "weight": 1 }, { "id": "edge-42a297c6d8f7", "source": "thesis-d4473d35e5", "target": "event-a03f87a2a6af", "type": "challenged_by", "weight": 1 }, { "id": "edge-6b291449a260", "source": "thesis-d4473d35e5", "target": "event-4f7d9cbb4e0e", "type": "challenged_by", "weight": 1 }, { "id": "edge-752ab5f29466", "source": "thesis-d4473d35e5", "target": "event-3f08e2d76329", "type": "supported_by", "weight": 1 }, { "id": "edge-a539984f1081", "source": "thesis-d4473d35e5", "target": "event-f28549ef93f6", "type": "challenged_by", "weight": 1 }, { "id": "edge-c7c2d17ecbed", "source": "thesis-d4473d35e5", "target": "event-f9b60b0ab260", "type": "challenged_by", "weight": 1 }, { "id": "edge-a9d00cf198c7", "source": "thesis-d4473d35e5", "target": "event-d83c1fbb708c", "type": "supported_by", "weight": 1 }, { "id": "edge-33703863f9aa", "source": "thesis-bef6966fdd", "target": "stock-a5b0d3a0ea", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e325e483af70", "source": "thesis-bef6966fdd", "target": "stock-656011c288", "type": "tracks_stock", "weight": 3 }, { "id": "edge-4112f7decee5", "source": "thesis-bef6966fdd", "target": "stock-e9ae311828", "type": "tracks_stock", "weight": 3 }, { "id": "edge-4c1019664b46", "source": "thesis-bef6966fdd", "target": "stock-6df13a035d", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6faab32103e2", "source": "thesis-bef6966fdd", "target": "stock-6b65919db6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b08b973f2d2e", "source": "thesis-bef6966fdd", "target": "stock-f1d36125fc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5e37bb5c8a70", "source": "thesis-bef6966fdd", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-06bd8f9acbad", "source": "thesis-bef6966fdd", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-168b944b3bab", "source": "thesis-bef6966fdd", "target": "term-e1506406a5", "type": "tracks_term", "weight": 2 }, { "id": "edge-27efe32ed949", "source": "thesis-bef6966fdd", "target": "term-68da2262bb", "type": "tracks_term", "weight": 2 }, { "id": "edge-3260538983d8", "source": "thesis-bef6966fdd", "target": "term-b60cae5339", "type": "tracks_term", "weight": 2 }, { "id": "edge-2958685f6b66", "source": "thesis-bef6966fdd", "target": "term-7cb194d941", "type": "tracks_term", "weight": 2 }, { "id": "edge-b13f7ea9aa0a", "source": "thesis-bef6966fdd", "target": "term-f44e64e75f", "type": "tracks_term", "weight": 2 }, { "id": "edge-c12cf1937e4a", "source": "thesis-bef6966fdd", "target": "term-1e3350f128", "type": "tracks_term", "weight": 2 }, { "id": "edge-7bb8b6fcfce4", "source": "thesis-bef6966fdd", "target": "event-744ce5df5d19", "type": "challenged_by", "weight": 1 }, { "id": "edge-ed42579aa52c", "source": "thesis-bef6966fdd", "target": "event-35f1db70dbf6", "type": "challenged_by", "weight": 1 }, { "id": "edge-b94988330f07", "source": "thesis-bef6966fdd", "target": "event-edb79402f972", "type": "supported_by", "weight": 1 }, { "id": "edge-78a501c56758", "source": "thesis-bef6966fdd", "target": "event-700f49766467", "type": "challenged_by", "weight": 1 }, { "id": "edge-e3bbd3cc45eb", "source": "thesis-bef6966fdd", "target": "event-7c33eb87a57e", "type": "challenged_by", "weight": 1 }, { "id": "edge-18eef971ab5b", "source": "thesis-bef6966fdd", "target": "event-6c68f8178cba", "type": "challenged_by", "weight": 1 }, { "id": "edge-8cce67224409", "source": "thesis-bef6966fdd", "target": "event-12997524a1c8", "type": "challenged_by", "weight": 1 }, { "id": "edge-d93cc6f28209", "source": "thesis-bef6966fdd", "target": "event-86456b560b61", "type": "challenged_by", "weight": 1 }, { "id": "edge-69c78e0f8d47", "source": "thesis-9a860a8801", "target": "stock-cec5af38a6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d130e1b21d58", "source": "thesis-9a860a8801", "target": "stock-db74d07203", "type": "tracks_stock", "weight": 3 }, { "id": "edge-25afbe4051ec", "source": "thesis-9a860a8801", "target": "stock-d86863071d", "type": "tracks_stock", "weight": 3 }, { "id": "edge-50708e8093b3", "source": "thesis-9a860a8801", "target": "stock-d7e5daa1ba", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a7be74479753", "source": "thesis-9a860a8801", "target": "stock-b64a63240f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-442ec6b95414", "source": "thesis-9a860a8801", "target": "stock-7b88702902", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a6080a069aee", "source": "thesis-9a860a8801", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-5c494c76f7b2", "source": "thesis-9a860a8801", "target": "term-920e88c5a6", "type": "tracks_term", "weight": 2 }, { "id": "edge-02a1a8d3e22e", "source": "thesis-9a860a8801", "target": "term-fb1875d0a5", "type": "tracks_term", "weight": 2 }, { "id": "edge-b8ad4230f9ca", "source": "thesis-9a860a8801", "target": "term-92b775598f", "type": "tracks_term", "weight": 2 }, { "id": "edge-8fe1cd144d15", "source": "thesis-9a860a8801", "target": "term-b667446e6e", "type": "tracks_term", "weight": 2 }, { "id": "edge-9472944bbb1c", "source": "thesis-9a860a8801", "target": "term-2ab62cfe39", "type": "tracks_term", "weight": 2 }, { "id": "edge-434a587fac2d", "source": "thesis-9a860a8801", "target": "term-c549779d79", "type": "tracks_term", "weight": 2 }, { "id": "edge-f0f6bc5ea41f", "source": "thesis-9a860a8801", "target": "term-2c6ef0f0d0", "type": "tracks_term", "weight": 2 }, { "id": "edge-f3fd948bb149", "source": "thesis-9a860a8801", "target": "event-b2d44d4f6727", "type": "challenged_by", "weight": 1 }, { "id": "edge-db7c5c01c5bd", "source": "thesis-9a860a8801", "target": "event-48caf385e2f0", "type": "challenged_by", "weight": 1 }, { "id": "edge-d0dd36547f61", "source": "thesis-9a860a8801", "target": "event-d974ecb51a5f", "type": "challenged_by", "weight": 1 }, { "id": "edge-c9992dd3c31f", "source": "thesis-9a860a8801", "target": "event-91128dc8e101", "type": "challenged_by", "weight": 1 }, { "id": "edge-41014df81b6d", "source": "thesis-9a860a8801", "target": "event-d1e5ed00a796", "type": "supported_by", "weight": 1 }, { "id": "edge-6df27be1e2a2", "source": "thesis-9a860a8801", "target": "event-0aff6872aaae", "type": "challenged_by", "weight": 1 }, { "id": "edge-2f0ac894898a", "source": "thesis-9a860a8801", "target": "event-e8199fd67cdf", "type": "challenged_by", "weight": 1 }, { "id": "edge-cd83ba453cdb", "source": "thesis-9a860a8801", "target": "event-a014e9443446", "type": "challenged_by", "weight": 1 }, { "id": "edge-70515c137596", "source": "thesis-985b80b224", "target": "stock-2ff6985a97", "type": "tracks_stock", "weight": 3 }, { "id": "edge-827f870b0e59", "source": "thesis-985b80b224", "target": "stock-0d72f0950a", "type": "tracks_stock", "weight": 3 }, { "id": "edge-3353dab70079", "source": "thesis-985b80b224", "target": "stock-b0eea525de", "type": "tracks_stock", "weight": 3 }, { "id": "edge-797276f43bd8", "source": "thesis-985b80b224", "target": "stock-8a0d7a9b71", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9b98e51ecc38", "source": "thesis-985b80b224", "target": "stock-f1d36125fc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f1a7b5b59d40", "source": "thesis-985b80b224", "target": "stock-10ae1be006", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d8dc7b390ab6", "source": "thesis-985b80b224", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-0909cfa49586", "source": "thesis-985b80b224", "target": "term-f89f85382f", "type": "tracks_term", "weight": 2 }, { "id": "edge-1c1edca20438", "source": "thesis-985b80b224", "target": "term-d16dc977ac", "type": "tracks_term", "weight": 2 }, { "id": "edge-5849875ebf3a", "source": "thesis-985b80b224", "target": "term-7793ba1ad0", "type": "tracks_term", "weight": 2 }, { "id": "edge-ee7fb3ce7d76", "source": "thesis-985b80b224", "target": "term-28faaa675c", "type": "tracks_term", "weight": 2 }, { "id": "edge-2f570121bd05", "source": "thesis-985b80b224", "target": "term-63e427f481", "type": "tracks_term", "weight": 2 }, { "id": "edge-378554b3394d", "source": "thesis-985b80b224", "target": "term-425edd11c2", "type": "tracks_term", "weight": 2 }, { "id": "edge-0b51b9c1a55c", "source": "thesis-985b80b224", "target": "term-f30dfe1cfb", "type": "tracks_term", "weight": 2 }, { "id": "edge-4d5e59945a9f", "source": "thesis-985b80b224", "target": "event-cf060bd3f061", "type": "challenged_by", "weight": 1 }, { "id": "edge-c114e3575b24", "source": "thesis-985b80b224", "target": "event-0855f5987bf6", "type": "challenged_by", "weight": 1 }, { "id": "edge-50a022988636", "source": "thesis-985b80b224", "target": "event-332b270b3c96", "type": "challenged_by", "weight": 1 }, { "id": "edge-9547fb7a3424", "source": "thesis-985b80b224", "target": "event-fb8e3440c09f", "type": "challenged_by", "weight": 1 }, { "id": "edge-c310c1772a4a", "source": "thesis-985b80b224", "target": "event-05bdd4816188", "type": "challenged_by", "weight": 1 }, { "id": "edge-f41378e4eda3", "source": "thesis-985b80b224", "target": "event-b6e155fd9417", "type": "challenged_by", "weight": 1 }, { "id": "edge-8ffc17533857", "source": "thesis-985b80b224", "target": "event-63806555c45b", "type": "supported_by", "weight": 1 }, { "id": "edge-88d39a5c6be1", "source": "thesis-985b80b224", "target": "event-ceea2085b7fe", "type": "challenged_by", "weight": 1 }, { "id": "edge-aab06fd87930", "source": "thesis-5057edfa79", "target": "stock-20c0a1e555", "type": "tracks_stock", "weight": 3 }, { "id": "edge-faa361fe36a2", "source": "thesis-5057edfa79", "target": "stock-6a8f57144b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-967cd74374bf", "source": "thesis-5057edfa79", "target": "stock-1a5bdfc9bd", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e5a49246380b", "source": "thesis-5057edfa79", "target": "stock-a55ffe9be6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-329a84d58d85", "source": "thesis-5057edfa79", "target": "stock-56a2cb6836", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b980e5848a40", "source": "thesis-5057edfa79", "target": "stock-6ece0f6e09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a105c4200834", "source": "thesis-5057edfa79", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-11b6c21bfb6a", "source": "thesis-5057edfa79", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-68d8ff62b2c2", "source": "thesis-5057edfa79", "target": "term-5dbd754de7", "type": "tracks_term", "weight": 2 }, { "id": "edge-c5d2c86ae6f3", "source": "thesis-5057edfa79", "target": "term-ede68d6aba", "type": "tracks_term", "weight": 2 }, { "id": "edge-bab0a55876cc", "source": "thesis-5057edfa79", "target": "term-e1506406a5", "type": "tracks_term", "weight": 2 }, { "id": "edge-7cb4495ebce6", "source": "thesis-5057edfa79", "target": "term-e47aa08e3a", "type": "tracks_term", "weight": 2 }, { "id": "edge-dd0b73185db3", "source": "thesis-5057edfa79", "target": "term-af8f827a0c", "type": "tracks_term", "weight": 2 }, { "id": "edge-96691b559d92", "source": "thesis-5057edfa79", "target": "term-3d8049711c", "type": "tracks_term", "weight": 2 }, { "id": "edge-0f06d805669f", "source": "thesis-5057edfa79", "target": "event-14d779169908", "type": "challenged_by", "weight": 1 }, { "id": "edge-2b80ee83f060", "source": "thesis-5057edfa79", "target": "event-d1096c39499e", "type": "challenged_by", "weight": 1 }, { "id": "edge-cf783d14b431", "source": "thesis-5057edfa79", "target": "event-d0ee08891b52", "type": "challenged_by", "weight": 1 }, { "id": "edge-0ac054e5c425", "source": "thesis-5057edfa79", "target": "event-99dfd805747d", "type": "challenged_by", "weight": 1 }, { "id": "edge-b93c8af9e7f2", "source": "thesis-5057edfa79", "target": "event-98fa8eb1bd91", "type": "challenged_by", "weight": 1 }, { "id": "edge-d604aa66ecea", "source": "thesis-5057edfa79", "target": "event-8804692d782a", "type": "challenged_by", "weight": 1 }, { "id": "edge-f09f6328c458", "source": "thesis-5057edfa79", "target": "event-66497536a9a7", "type": "challenged_by", "weight": 1 }, { "id": "edge-957633c13fd6", "source": "thesis-5057edfa79", "target": "event-329b5853dd8d", "type": "challenged_by", "weight": 1 }, { "id": "edge-4b1b6b8ffb4c", "source": "thesis-026aa1679b", "target": "stock-2d1c22cd10", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b373ddad4eb4", "source": "thesis-026aa1679b", "target": "stock-59e40ddab7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-58c2e90807ab", "source": "thesis-026aa1679b", "target": "stock-61e19d9ee8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-62d0b656202d", "source": "thesis-026aa1679b", "target": "stock-b5733c9451", "type": "tracks_stock", "weight": 3 }, { "id": "edge-3afcba99fecb", "source": "thesis-026aa1679b", "target": "stock-8ee9008c2b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-756f4f2b0547", "source": "thesis-026aa1679b", "target": "stock-420d41ec1c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-09ef2be0663b", "source": "thesis-026aa1679b", "target": "term-95feb4759b", "type": "tracks_term", "weight": 2 }, { "id": "edge-a0a49a8735e1", "source": "thesis-026aa1679b", "target": "term-d49897f8c4", "type": "tracks_term", "weight": 2 }, { "id": "edge-f7a9132cf3f8", "source": "thesis-026aa1679b", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-500dc4cef6dc", "source": "thesis-026aa1679b", "target": "term-0e2b23eeed", "type": "tracks_term", "weight": 2 }, { "id": "edge-b23228808eab", "source": "thesis-026aa1679b", "target": "term-f67c2069ed", "type": "tracks_term", "weight": 2 }, { "id": "edge-79e8e9a915ab", "source": "thesis-026aa1679b", "target": "term-04f7acfcd2", "type": "tracks_term", "weight": 2 }, { "id": "edge-033321cdb294", "source": "thesis-026aa1679b", "target": "term-6001ec95f6", "type": "tracks_term", "weight": 2 }, { "id": "edge-4525d08d9147", "source": "thesis-026aa1679b", "target": "term-641102ee2d", "type": "tracks_term", "weight": 2 }, { "id": "edge-73e7a1c0ce94", "source": "thesis-026aa1679b", "target": "event-315256c9f3d7", "type": "challenged_by", "weight": 1 }, { "id": "edge-0f382a8333fb", "source": "thesis-026aa1679b", "target": "event-74dd65b2130e", "type": "challenged_by", "weight": 1 }, { "id": "edge-900af81f4ea2", "source": "thesis-026aa1679b", "target": "event-2479960c402f", "type": "supported_by", "weight": 1 }, { "id": "edge-c3ad2a79f5e4", "source": "thesis-026aa1679b", "target": "event-509371f08368", "type": "challenged_by", "weight": 1 }, { "id": "edge-54568044d48f", "source": "thesis-026aa1679b", "target": "event-a80f937943f9", "type": "challenged_by", "weight": 1 }, { "id": "edge-7a612d29aab5", "source": "thesis-026aa1679b", "target": "event-fb601e54c617", "type": "challenged_by", "weight": 1 }, { "id": "edge-10a03f27d1d7", "source": "thesis-026aa1679b", "target": "event-130ee5fcd597", "type": "challenged_by", "weight": 1 }, { "id": "edge-863c1b0c34c1", "source": "thesis-026aa1679b", "target": "event-cbc7536aa13c", "type": "challenged_by", "weight": 1 }, { "id": "edge-19c64b38b81c", "source": "thesis-21dd72c1b1", "target": "stock-b1ce5beca8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-196c95d0cdc9", "source": "thesis-21dd72c1b1", "target": "stock-75f04b8433", "type": "tracks_stock", "weight": 3 }, { "id": "edge-fa4c15578dd0", "source": "thesis-21dd72c1b1", "target": "stock-148304483a", "type": "tracks_stock", "weight": 3 }, { "id": "edge-75276a14017e", "source": "thesis-21dd72c1b1", "target": "stock-884d5bff79", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ba28d27185d8", "source": "thesis-21dd72c1b1", "target": "stock-fb5bd3958a", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9c3e3e810107", "source": "thesis-21dd72c1b1", "target": "stock-296d71a7f6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7dffb20ad4ea", "source": "thesis-21dd72c1b1", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-3ea170b1497f", "source": "thesis-21dd72c1b1", "target": "term-ed9ab89f22", "type": "tracks_term", "weight": 2 }, { "id": "edge-cf28c4c921b0", "source": "thesis-21dd72c1b1", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-ea9e6be77e93", "source": "thesis-21dd72c1b1", "target": "term-d527f8c5f9", "type": "tracks_term", "weight": 2 }, { "id": "edge-67335ab488bf", "source": "thesis-21dd72c1b1", "target": "term-1b0b7e51ee", "type": "tracks_term", "weight": 2 }, { "id": "edge-5d56244c6395", "source": "thesis-21dd72c1b1", "target": "term-3f412634a4", "type": "tracks_term", "weight": 2 }, { "id": "edge-9db11632e811", "source": "thesis-21dd72c1b1", "target": "term-5acbfff1b0", "type": "tracks_term", "weight": 2 }, { "id": "edge-fd96c6be100e", "source": "thesis-21dd72c1b1", "target": "term-7f9c40c9ac", "type": "tracks_term", "weight": 2 }, { "id": "edge-81d267f74fe1", "source": "thesis-21dd72c1b1", "target": "event-089bbe0434df", "type": "challenged_by", "weight": 1 }, { "id": "edge-ca979fe61c29", "source": "thesis-21dd72c1b1", "target": "event-fb667d390e85", "type": "challenged_by", "weight": 1 }, { "id": "edge-f8193df17743", "source": "thesis-21dd72c1b1", "target": "event-3dd0d941a3ad", "type": "challenged_by", "weight": 1 }, { "id": "edge-8cb12920a90e", "source": "thesis-21dd72c1b1", "target": "event-70b3005cad15", "type": "challenged_by", "weight": 1 }, { "id": "edge-d52147b9854c", "source": "thesis-21dd72c1b1", "target": "event-0973079dc94c", "type": "challenged_by", "weight": 1 }, { "id": "edge-a1a3d040441d", "source": "thesis-21dd72c1b1", "target": "event-953a753c1dd6", "type": "challenged_by", "weight": 1 }, { "id": "edge-842b156c8a05", "source": "thesis-21dd72c1b1", "target": "event-3966f447e917", "type": "challenged_by", "weight": 1 }, { "id": "edge-a7a0c05f76cc", "source": "thesis-21dd72c1b1", "target": "event-2151b33ea5c3", "type": "challenged_by", "weight": 1 }, { "id": "edge-463457ddb4f5", "source": "thesis-66eb42380e", "target": "stock-fdf60d294f", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c2a621356035", "source": "thesis-66eb42380e", "target": "stock-cb6f6fac36", "type": "tracks_stock", "weight": 3 }, { "id": "edge-cb8ef87160f1", "source": "thesis-66eb42380e", "target": "stock-9ef40fc77b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ed484098771e", "source": "thesis-66eb42380e", "target": "stock-8bd6af8d30", "type": "tracks_stock", "weight": 3 }, { "id": "edge-17ea1350a35f", "source": "thesis-66eb42380e", "target": "stock-35264bffcb", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a221d7deb93d", "source": "thesis-66eb42380e", "target": "stock-b49750b06c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-6992010aca31", "source": "thesis-66eb42380e", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-aea75a09bad8", "source": "thesis-66eb42380e", "target": "term-6001ec95f6", "type": "tracks_term", "weight": 2 }, { "id": "edge-b153fef998c8", "source": "thesis-66eb42380e", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-ae3a4952352e", "source": "thesis-66eb42380e", "target": "term-2b95a99fe5", "type": "tracks_term", "weight": 2 }, { "id": "edge-d8f3ddc39afb", "source": "thesis-66eb42380e", "target": "term-017f1d8e6c", "type": "tracks_term", "weight": 2 }, { "id": "edge-f53dd1ca02b5", "source": "thesis-66eb42380e", "target": "term-8a2fd55b01", "type": "tracks_term", "weight": 2 }, { "id": "edge-40efbb7c1311", "source": "thesis-66eb42380e", "target": "term-d8cfb8d54d", "type": "tracks_term", "weight": 2 }, { "id": "edge-f783fe58d369", "source": "thesis-66eb42380e", "target": "term-823412d1ea", "type": "tracks_term", "weight": 2 }, { "id": "edge-d5aace2c91d5", "source": "thesis-66eb42380e", "target": "event-f27ee71cf2df", "type": "challenged_by", "weight": 1 }, { "id": "edge-7b93e601d9e5", "source": "thesis-66eb42380e", "target": "event-4af2b94a15a8", "type": "challenged_by", "weight": 1 }, { "id": "edge-5d66dc174fa4", "source": "thesis-66eb42380e", "target": "event-5f97b773d84a", "type": "challenged_by", "weight": 1 }, { "id": "edge-3a65cfadcfe1", "source": "thesis-66eb42380e", "target": "event-711381b4c4ad", "type": "supported_by", "weight": 1 }, { "id": "edge-3f834ceecf7c", "source": "thesis-66eb42380e", "target": "event-f5a83b82e08c", "type": "supported_by", "weight": 1 }, { "id": "edge-38f723e72fdf", "source": "thesis-66eb42380e", "target": "event-d421f9e45dd4", "type": "challenged_by", "weight": 1 }, { "id": "edge-50192401e31d", "source": "thesis-66eb42380e", "target": "event-0e49020737b2", "type": "challenged_by", "weight": 1 }, { "id": "edge-740adfa9aea3", "source": "thesis-66eb42380e", "target": "event-a22091e382f7", "type": "supported_by", "weight": 1 }, { "id": "edge-3193fe35860c", "source": "thesis-c6c6be4745", "target": "stock-c7e6a87908", "type": "tracks_stock", "weight": 3 }, { "id": "edge-119567afd133", "source": "thesis-c6c6be4745", "target": "stock-e59c1688ee", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9264791ad06f", "source": "thesis-c6c6be4745", "target": "stock-95587eff89", "type": "tracks_stock", "weight": 3 }, { "id": "edge-02c24b63d38f", "source": "thesis-c6c6be4745", "target": "stock-d54aa25b07", "type": "tracks_stock", "weight": 3 }, { "id": "edge-2d537c310f2e", "source": "thesis-c6c6be4745", "target": "stock-ce89b19c8e", "type": "tracks_stock", "weight": 3 }, { "id": "edge-bdcbe2076fba", "source": "thesis-c6c6be4745", "target": "stock-9441f62b62", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e369073500a7", "source": "thesis-c6c6be4745", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-3340558a7171", "source": "thesis-c6c6be4745", "target": "term-66e0f52800", "type": "tracks_term", "weight": 2 }, { "id": "edge-47cbd0deb8d5", "source": "thesis-c6c6be4745", "target": "term-9196f0338a", "type": "tracks_term", "weight": 2 }, { "id": "edge-bb5a8dd756ff", "source": "thesis-c6c6be4745", "target": "term-8a2fd55b01", "type": "tracks_term", "weight": 2 }, { "id": "edge-1bf04aafb5da", "source": "thesis-c6c6be4745", "target": "term-8a9779f421", "type": "tracks_term", "weight": 2 }, { "id": "edge-092f6806b03a", "source": "thesis-c6c6be4745", "target": "term-d46aee08cc", "type": "tracks_term", "weight": 2 }, { "id": "edge-3fc1da767213", "source": "thesis-c6c6be4745", "target": "term-039d6add7e", "type": "tracks_term", "weight": 2 }, { "id": "edge-3e98d9924dd2", "source": "thesis-c6c6be4745", "target": "term-aeadfe2770", "type": "tracks_term", "weight": 2 }, { "id": "edge-a55312615fda", "source": "thesis-c6c6be4745", "target": "event-d85c8e0558f0", "type": "supported_by", "weight": 1 }, { "id": "edge-78cab1eba241", "source": "thesis-c6c6be4745", "target": "event-5aef3dfcde02", "type": "supported_by", "weight": 1 }, { "id": "edge-b58ae490d82a", "source": "thesis-c6c6be4745", "target": "event-bf2680c00021", "type": "challenged_by", "weight": 1 }, { "id": "edge-ba2a3da8ee71", "source": "thesis-c6c6be4745", "target": "event-28589741865e", "type": "challenged_by", "weight": 1 }, { "id": "edge-30d8edbfa382", "source": "thesis-c6c6be4745", "target": "event-c900e6aa9594", "type": "challenged_by", "weight": 1 }, { "id": "edge-cc71e9650273", "source": "thesis-c6c6be4745", "target": "event-fad23caaa8d2", "type": "challenged_by", "weight": 1 }, { "id": "edge-58b28637125a", "source": "thesis-c6c6be4745", "target": "event-5c140a724456", "type": "challenged_by", "weight": 1 }, { "id": "edge-8a71dfca4507", "source": "thesis-c6c6be4745", "target": "event-03abbfec27f4", "type": "supported_by", "weight": 1 }, { "id": "edge-9b1a5b76eca1", "source": "thesis-485a630a87", "target": "stock-2ff6985a97", "type": "tracks_stock", "weight": 3 }, { "id": "edge-76b667706086", "source": "thesis-485a630a87", "target": "stock-a25fc665a4", "type": "tracks_stock", "weight": 3 }, { "id": "edge-09ac243d7684", "source": "thesis-485a630a87", "target": "stock-23cfa8bf26", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b75db796a5a2", "source": "thesis-485a630a87", "target": "stock-294da558b6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-21b0b1138676", "source": "thesis-485a630a87", "target": "stock-cf9e2f2a16", "type": "tracks_stock", "weight": 3 }, { "id": "edge-e8bd9b316c0e", "source": "thesis-485a630a87", "target": "stock-46646e9d5e", "type": "tracks_stock", "weight": 3 }, { "id": "edge-3dffd54cb835", "source": "thesis-485a630a87", "target": "term-44568e2e5f", "type": "tracks_term", "weight": 2 }, { "id": "edge-9be1347d39eb", "source": "thesis-485a630a87", "target": "term-f89f85382f", "type": "tracks_term", "weight": 2 }, { "id": "edge-c338eb1dd5df", "source": "thesis-485a630a87", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-bb37cfb94872", "source": "thesis-485a630a87", "target": "term-aa03f5c14b", "type": "tracks_term", "weight": 2 }, { "id": "edge-0acd25f2ef01", "source": "thesis-485a630a87", "target": "term-a8aa1f6bfe", "type": "tracks_term", "weight": 2 }, { "id": "edge-3593b984eee4", "source": "thesis-485a630a87", "target": "term-d9c1acf117", "type": "tracks_term", "weight": 2 }, { "id": "edge-3bf08e0ab77e", "source": "thesis-485a630a87", "target": "term-7f48817c8f", "type": "tracks_term", "weight": 2 }, { "id": "edge-2028ef10a538", "source": "thesis-485a630a87", "target": "term-ff582559c3", "type": "tracks_term", "weight": 2 }, { "id": "edge-2aed062b5435", "source": "thesis-485a630a87", "target": "event-066d7522ea41", "type": "challenged_by", "weight": 1 }, { "id": "edge-4d3bd0859b98", "source": "thesis-485a630a87", "target": "event-d6a17ab2f3c8", "type": "challenged_by", "weight": 1 }, { "id": "edge-b24b330d7b73", "source": "thesis-485a630a87", "target": "event-b782271763dc", "type": "challenged_by", "weight": 1 }, { "id": "edge-2ced6ae573ee", "source": "thesis-485a630a87", "target": "event-02513136b1da", "type": "challenged_by", "weight": 1 }, { "id": "edge-28158760ec5c", "source": "thesis-485a630a87", "target": "event-32a91fe3b28e", "type": "challenged_by", "weight": 1 }, { "id": "edge-f739209a0191", "source": "thesis-485a630a87", "target": "event-e35d7d944abc", "type": "supported_by", "weight": 1 }, { "id": "edge-84295b01f0b4", "source": "thesis-485a630a87", "target": "event-5e257d42b8fc", "type": "challenged_by", "weight": 1 }, { "id": "edge-a453bfdac590", "source": "thesis-485a630a87", "target": "event-b1c2d14dbd06", "type": "challenged_by", "weight": 1 }, { "id": "edge-ed7134b71668", "source": "thesis-4074d8a3d2", "target": "stock-1eaff1ec01", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b839a15df76a", "source": "thesis-4074d8a3d2", "target": "stock-a5d1d55fae", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c31eed0632e6", "source": "thesis-4074d8a3d2", "target": "stock-d7899bd385", "type": "tracks_stock", "weight": 3 }, { "id": "edge-666d48909cc0", "source": "thesis-4074d8a3d2", "target": "stock-c338965e2a", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f6cabbca8f37", "source": "thesis-4074d8a3d2", "target": "stock-d54aa25b07", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f0d064db3d2b", "source": "thesis-4074d8a3d2", "target": "stock-78b3cb703d", "type": "tracks_stock", "weight": 3 }, { "id": "edge-78c17228d5e9", "source": "thesis-4074d8a3d2", "target": "term-95feb4759b", "type": "tracks_term", "weight": 2 }, { "id": "edge-e0a62ec66bfb", "source": "thesis-4074d8a3d2", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-f684b8c5bd7c", "source": "thesis-4074d8a3d2", "target": "term-0e2b23eeed", "type": "tracks_term", "weight": 2 }, { "id": "edge-e5bb6aad8ed2", "source": "thesis-4074d8a3d2", "target": "term-fab331127a", "type": "tracks_term", "weight": 2 }, { "id": "edge-b41e6e0bc85c", "source": "thesis-4074d8a3d2", "target": "term-a050178590", "type": "tracks_term", "weight": 2 }, { "id": "edge-485270036e8e", "source": "thesis-4074d8a3d2", "target": "term-b8332514e0", "type": "tracks_term", "weight": 2 }, { "id": "edge-7310b25e7bd6", "source": "thesis-4074d8a3d2", "target": "term-66e0f52800", "type": "tracks_term", "weight": 2 }, { "id": "edge-9ebffa97a45d", "source": "thesis-4074d8a3d2", "target": "term-ef242752b3", "type": "tracks_term", "weight": 2 }, { "id": "edge-fc8bb0be8a21", "source": "thesis-4074d8a3d2", "target": "event-3cf09b5d78d2", "type": "challenged_by", "weight": 1 }, { "id": "edge-2be4cf152990", "source": "thesis-4074d8a3d2", "target": "event-626f451117f0", "type": "challenged_by", "weight": 1 }, { "id": "edge-760c236d6cef", "source": "thesis-4074d8a3d2", "target": "event-aad4e81c7a0e", "type": "supported_by", "weight": 1 }, { "id": "edge-1692a5cb23fc", "source": "thesis-4074d8a3d2", "target": "event-ea94fc3b699d", "type": "supported_by", "weight": 1 }, { "id": "edge-7b259d36b72e", "source": "thesis-4074d8a3d2", "target": "event-40d8f3792517", "type": "challenged_by", "weight": 1 }, { "id": "edge-d5463fc36103", "source": "thesis-4074d8a3d2", "target": "event-a496ad79b63c", "type": "challenged_by", "weight": 1 }, { "id": "edge-0c7b2b9c84d4", "source": "thesis-4074d8a3d2", "target": "event-7e616a7da726", "type": "supported_by", "weight": 1 }, { "id": "edge-9da3b7e82a16", "source": "thesis-4074d8a3d2", "target": "event-b61bc39c77eb", "type": "challenged_by", "weight": 1 }, { "id": "edge-abbff22db2b3", "source": "thesis-aa97a5c1b3", "target": "stock-165e1b016a", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c20070a926f5", "source": "thesis-aa97a5c1b3", "target": "stock-058c86ec94", "type": "tracks_stock", "weight": 3 }, { "id": "edge-8e2e1e5d4373", "source": "thesis-aa97a5c1b3", "target": "stock-e59c1688ee", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b967b8ffda88", "source": "thesis-aa97a5c1b3", "target": "stock-95587eff89", "type": "tracks_stock", "weight": 3 }, { "id": "edge-2dbc7ac43d21", "source": "thesis-aa97a5c1b3", "target": "stock-60562db099", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a053204bb2b4", "source": "thesis-aa97a5c1b3", "target": "stock-b0e0be6bf7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-76cb6964c7d8", "source": "thesis-aa97a5c1b3", "target": "term-3a78695388", "type": "tracks_term", "weight": 2 }, { "id": "edge-b1f4ec9ebfd5", "source": "thesis-aa97a5c1b3", "target": "term-62a4a0e9fb", "type": "tracks_term", "weight": 2 }, { "id": "edge-cdc80785d435", "source": "thesis-aa97a5c1b3", "target": "term-1e3350f128", "type": "tracks_term", "weight": 2 }, { "id": "edge-b138f9894364", "source": "thesis-aa97a5c1b3", "target": "term-dae040fb63", "type": "tracks_term", "weight": 2 }, { "id": "edge-67a784908b00", "source": "thesis-aa97a5c1b3", "target": "term-92a9fcf1f2", "type": "tracks_term", "weight": 2 }, { "id": "edge-2dee5f992590", "source": "thesis-aa97a5c1b3", "target": "term-15fc7418ed", "type": "tracks_term", "weight": 2 }, { "id": "edge-290a99a8b891", "source": "thesis-aa97a5c1b3", "target": "term-ed4561f685", "type": "tracks_term", "weight": 2 }, { "id": "edge-53b2231ee492", "source": "thesis-aa97a5c1b3", "target": "term-fdaad140aa", "type": "tracks_term", "weight": 2 }, { "id": "edge-51917991cbae", "source": "thesis-aa97a5c1b3", "target": "event-b940362e188a", "type": "challenged_by", "weight": 1 }, { "id": "edge-46f9fe79b695", "source": "thesis-aa97a5c1b3", "target": "event-6b437d4ac75b", "type": "challenged_by", "weight": 1 }, { "id": "edge-04422d8593f6", "source": "thesis-aa97a5c1b3", "target": "event-9dd4b2d68135", "type": "challenged_by", "weight": 1 }, { "id": "edge-90d4789d29f6", "source": "thesis-aa97a5c1b3", "target": "event-6117b0d1745e", "type": "challenged_by", "weight": 1 }, { "id": "edge-d8856f0a8d06", "source": "thesis-aa97a5c1b3", "target": "event-d1e05b2add41", "type": "supported_by", "weight": 1 }, { "id": "edge-9365cc557dd0", "source": "thesis-aa97a5c1b3", "target": "event-29e9007c9bc8", "type": "challenged_by", "weight": 1 }, { "id": "edge-8e8f14962f75", "source": "thesis-aa97a5c1b3", "target": "event-643ce47583bf", "type": "challenged_by", "weight": 1 }, { "id": "edge-144216753103", "source": "thesis-aa97a5c1b3", "target": "event-9523a9be4503", "type": "challenged_by", "weight": 1 }, { "id": "edge-1b3adcf35e6a", "source": "thesis-c1d3acab38", "target": "stock-0716b17bb6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-956caedc3116", "source": "thesis-c1d3acab38", "target": "stock-6ece0f6e09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c1542850a179", "source": "thesis-c1d3acab38", "target": "stock-bb47079d51", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ab7c2ef76198", "source": "thesis-c1d3acab38", "target": "stock-8e7522f842", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5433276fface", "source": "thesis-c1d3acab38", "target": "stock-81aa4e9efc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-fbc7d9bd03cf", "source": "thesis-c1d3acab38", "target": "stock-083dc3a2b3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b937833de1d7", "source": "thesis-c1d3acab38", "target": "term-e54819251d", "type": "tracks_term", "weight": 2 }, { "id": "edge-3d20ce0846b3", "source": "thesis-c1d3acab38", "target": "term-585a293e96", "type": "tracks_term", "weight": 2 }, { "id": "edge-603262e57552", "source": "thesis-c1d3acab38", "target": "event-b3cdd1fa4f78", "type": "challenged_by", "weight": 1 }, { "id": "edge-94c00123826e", "source": "thesis-c1d3acab38", "target": "event-9f6a2c4784dd", "type": "challenged_by", "weight": 1 }, { "id": "edge-04eb63d77105", "source": "thesis-c1d3acab38", "target": "event-42928c33310a", "type": "challenged_by", "weight": 1 }, { "id": "edge-3f66b3643c57", "source": "thesis-c1d3acab38", "target": "event-7aeae1943dff", "type": "challenged_by", "weight": 1 }, { "id": "edge-3f66b3643c57", "source": "thesis-c1d3acab38", "target": "event-7aeae1943dff", "type": "challenged_by", "weight": 1 }, { "id": "edge-5a4d55fca115", "source": "thesis-c1d3acab38", "target": "event-315e1eccf479", "type": "challenged_by", "weight": 1 }, { "id": "edge-76bdaefa8f89", "source": "thesis-c1d3acab38", "target": "event-3f6dcc96f8c5", "type": "challenged_by", "weight": 1 }, { "id": "edge-76bdaefa8f89", "source": "thesis-c1d3acab38", "target": "event-3f6dcc96f8c5", "type": "challenged_by", "weight": 1 }, { "id": "edge-7b5eb164ea67", "source": "thesis-d1c60d5af6", "target": "stock-9441f62b62", "type": "tracks_stock", "weight": 3 }, { "id": "edge-2ba680967624", "source": "thesis-d1c60d5af6", "target": "stock-83eb33367d", "type": "tracks_stock", "weight": 3 }, { "id": "edge-1863d5d5aa6c", "source": "thesis-d1c60d5af6", "target": "stock-7e9ccaea2d", "type": "tracks_stock", "weight": 3 }, { "id": "edge-eda84c390b4e", "source": "thesis-d1c60d5af6", "target": "stock-73e0276fbb", "type": "tracks_stock", "weight": 3 }, { "id": "edge-8c052b19b117", "source": "thesis-d1c60d5af6", "target": "stock-136926fb0c", "type": "tracks_stock", "weight": 3 }, { "id": "edge-7b7750cebfd7", "source": "thesis-d1c60d5af6", "target": "stock-2446d5d7e6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-08972a43c341", "source": "thesis-d1c60d5af6", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-a5260ada850e", "source": "thesis-d1c60d5af6", "target": "term-0e2b23eeed", "type": "tracks_term", "weight": 2 }, { "id": "edge-3d2fee1110e7", "source": "thesis-d1c60d5af6", "target": "term-b1a96dd646", "type": "tracks_term", "weight": 2 }, { "id": "edge-5f1553e52ebf", "source": "thesis-d1c60d5af6", "target": "term-90115ae3fa", "type": "tracks_term", "weight": 2 }, { "id": "edge-b6f7ea679a63", "source": "thesis-d1c60d5af6", "target": "term-646b1045ad", "type": "tracks_term", "weight": 2 }, { "id": "edge-d2eb1ad426b8", "source": "thesis-d1c60d5af6", "target": "term-d46aee08cc", "type": "tracks_term", "weight": 2 }, { "id": "edge-0895cad099c4", "source": "thesis-d1c60d5af6", "target": "term-8ed3f6ad68", "type": "tracks_term", "weight": 2 }, { "id": "edge-b0418742816e", "source": "thesis-d1c60d5af6", "target": "term-1e3350f128", "type": "tracks_term", "weight": 2 }, { "id": "edge-1dfe45d486e7", "source": "thesis-d1c60d5af6", "target": "event-01624df5c8d4", "type": "supported_by", "weight": 1 }, { "id": "edge-cdb85b9bcfc5", "source": "thesis-d1c60d5af6", "target": "event-f18a3a30a8d9", "type": "challenged_by", "weight": 1 }, { "id": "edge-7576d19fc9af", "source": "thesis-d1c60d5af6", "target": "event-a9dff3ae63f0", "type": "challenged_by", "weight": 1 }, { "id": "edge-37f13d05149b", "source": "thesis-d1c60d5af6", "target": "event-dd7ab6f0c052", "type": "challenged_by", "weight": 1 }, { "id": "edge-a14d97f12a5d", "source": "thesis-d1c60d5af6", "target": "event-c3c86d9d6ea0", "type": "challenged_by", "weight": 1 }, { "id": "edge-d955afd81347", "source": "thesis-d1c60d5af6", "target": "event-c29fdef3aa04", "type": "challenged_by", "weight": 1 }, { "id": "edge-1bb6aa0d7d7d", "source": "thesis-d1c60d5af6", "target": "event-f3ea290b8c6e", "type": "challenged_by", "weight": 1 }, { "id": "edge-5ee49a1a2761", "source": "thesis-d1c60d5af6", "target": "event-a0f5d7a65e45", "type": "challenged_by", "weight": 1 }, { "id": "edge-6ca767d59e87", "source": "thesis-e26ef69f73", "target": "stock-9b19451158", "type": "tracks_stock", "weight": 3 }, { "id": "edge-94d73f93ec1c", "source": "thesis-e26ef69f73", "target": "stock-e0671bd326", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c5b0ca772d01", "source": "thesis-e26ef69f73", "target": "stock-b19fdc727e", "type": "tracks_stock", "weight": 3 }, { "id": "edge-fd3e7305b086", "source": "thesis-e26ef69f73", "target": "stock-89b531787b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a3b4d94b4a7d", "source": "thesis-e26ef69f73", "target": "stock-68d0766ae1", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ad108e0cb59a", "source": "thesis-e26ef69f73", "target": "stock-06214b33ef", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b6eec77d7834", "source": "thesis-e26ef69f73", "target": "term-ea49523d74", "type": "tracks_term", "weight": 2 }, { "id": "edge-19916cc43d22", "source": "thesis-e26ef69f73", "target": "term-0ef384847c", "type": "tracks_term", "weight": 2 }, { "id": "edge-239479e4c712", "source": "thesis-e26ef69f73", "target": "term-1ae05a2701", "type": "tracks_term", "weight": 2 }, { "id": "edge-9b6da3b034ee", "source": "thesis-e26ef69f73", "target": "term-0711a8d636", "type": "tracks_term", "weight": 2 }, { "id": "edge-239cba94eb6d", "source": "thesis-e26ef69f73", "target": "term-4183b7793f", "type": "tracks_term", "weight": 2 }, { "id": "edge-745cf00b7b54", "source": "thesis-e26ef69f73", "target": "term-ee8250fb76", "type": "tracks_term", "weight": 2 }, { "id": "edge-4ccedddc7bae", "source": "thesis-e26ef69f73", "target": "term-acef2c15bc", "type": "tracks_term", "weight": 2 }, { "id": "edge-128ca6256318", "source": "thesis-e26ef69f73", "target": "term-80ea8bc4ba", "type": "tracks_term", "weight": 2 }, { "id": "edge-8a397932cccd", "source": "thesis-e26ef69f73", "target": "event-4f58364c16fb", "type": "challenged_by", "weight": 1 }, { "id": "edge-60a945e6910e", "source": "thesis-e26ef69f73", "target": "event-04bad6093f6b", "type": "challenged_by", "weight": 1 }, { "id": "edge-3a1595907109", "source": "thesis-e26ef69f73", "target": "event-0ff1e6510837", "type": "challenged_by", "weight": 1 }, { "id": "edge-6165352ce61b", "source": "thesis-e26ef69f73", "target": "event-1be6fd7b9c94", "type": "supported_by", "weight": 1 }, { "id": "edge-54aa2ccbb5aa", "source": "thesis-e26ef69f73", "target": "event-d8df41c4ac49", "type": "challenged_by", "weight": 1 }, { "id": "edge-8f686424a302", "source": "thesis-e26ef69f73", "target": "event-7700f4fc2fa9", "type": "challenged_by", "weight": 1 }, { "id": "edge-53ce23ea1587", "source": "thesis-e26ef69f73", "target": "event-d2eb5a4be6b4", "type": "supported_by", "weight": 1 }, { "id": "edge-81b0364ba1d0", "source": "thesis-e26ef69f73", "target": "event-428edb25e361", "type": "challenged_by", "weight": 1 }, { "id": "edge-c340a0c847fe", "source": "thesis-d6b2c23ed6", "target": "stock-9ef40fc77b", "type": "tracks_stock", "weight": 3 }, { "id": "edge-9fc7193c5e1a", "source": "thesis-d6b2c23ed6", "target": "stock-912ff4d815", "type": "tracks_stock", "weight": 3 }, { "id": "edge-da70933f0aa8", "source": "thesis-d6b2c23ed6", "target": "stock-f0875559d8", "type": "tracks_stock", "weight": 3 }, { "id": "edge-1de308a0becb", "source": "thesis-d6b2c23ed6", "target": "stock-6a65aeb1ac", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f54264b9128a", "source": "thesis-d6b2c23ed6", "target": "stock-ca7ea4dbbb", "type": "tracks_stock", "weight": 3 }, { "id": "edge-196bf50df3f4", "source": "thesis-d6b2c23ed6", "target": "stock-afb6f77a11", "type": "tracks_stock", "weight": 3 }, { "id": "edge-58652989d712", "source": "thesis-d6b2c23ed6", "target": "term-1e7a12e965", "type": "tracks_term", "weight": 2 }, { "id": "edge-0e6eb20ac116", "source": "thesis-d6b2c23ed6", "target": "term-c75a1e81fa", "type": "tracks_term", "weight": 2 }, { "id": "edge-f22c0303a33c", "source": "thesis-d6b2c23ed6", "target": "term-0da46a7fbf", "type": "tracks_term", "weight": 2 }, { "id": "edge-b464b97241b5", "source": "thesis-d6b2c23ed6", "target": "term-ad5751430e", "type": "tracks_term", "weight": 2 }, { "id": "edge-f896a88651f3", "source": "thesis-d6b2c23ed6", "target": "term-8880894de4", "type": "tracks_term", "weight": 2 }, { "id": "edge-f961e80b745c", "source": "thesis-d6b2c23ed6", "target": "term-bcde6928e5", "type": "tracks_term", "weight": 2 }, { "id": "edge-14abd6eceae4", "source": "thesis-d6b2c23ed6", "target": "term-3fbb9ba69d", "type": "tracks_term", "weight": 2 }, { "id": "edge-8397ccb20b6c", "source": "thesis-d6b2c23ed6", "target": "term-1d2cafdb2f", "type": "tracks_term", "weight": 2 }, { "id": "edge-9a3a1f02cf8d", "source": "thesis-d6b2c23ed6", "target": "event-cc15456dfa69", "type": "challenged_by", "weight": 1 }, { "id": "edge-3d4620ccdf7f", "source": "thesis-d6b2c23ed6", "target": "event-126f81128f82", "type": "supported_by", "weight": 1 }, { "id": "edge-a05aadbec90d", "source": "thesis-d6b2c23ed6", "target": "event-5c3e4e51ef3c", "type": "challenged_by", "weight": 1 }, { "id": "edge-61e2c9cdd135", "source": "thesis-d6b2c23ed6", "target": "event-a5e874116bfd", "type": "challenged_by", "weight": 1 }, { "id": "edge-deb700ecc01a", "source": "thesis-d6b2c23ed6", "target": "event-29e06d542d19", "type": "challenged_by", "weight": 1 }, { "id": "edge-3a895246ac91", "source": "thesis-d6b2c23ed6", "target": "event-1c89e048cdec", "type": "challenged_by", "weight": 1 }, { "id": "edge-77be9ccd4cec", "source": "thesis-d6b2c23ed6", "target": "event-7acd89597314", "type": "challenged_by", "weight": 1 }, { "id": "edge-c935b9185a7c", "source": "thesis-d6b2c23ed6", "target": "event-9298e4c5f04c", "type": "challenged_by", "weight": 1 }, { "id": "edge-2cde94cea6ce", "source": "thesis-62d39d2c51", "target": "stock-0716b17bb6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-31abfe80b7a6", "source": "thesis-62d39d2c51", "target": "stock-6ece0f6e09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-08981280f302", "source": "thesis-62d39d2c51", "target": "stock-bb47079d51", "type": "tracks_stock", "weight": 3 }, { "id": "edge-415346f25b67", "source": "thesis-62d39d2c51", "target": "stock-8e7522f842", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ab829551010d", "source": "thesis-62d39d2c51", "target": "stock-81aa4e9efc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-bfcf1bb6f001", "source": "thesis-62d39d2c51", "target": "stock-083dc3a2b3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-290bc37b5022", "source": "thesis-62d39d2c51", "target": "term-09c5d97f30", "type": "tracks_term", "weight": 2 }, { "id": "edge-ef8bec790787", "source": "thesis-62d39d2c51", "target": "term-585a293e96", "type": "tracks_term", "weight": 2 }, { "id": "edge-14c568de9f20", "source": "thesis-62d39d2c51", "target": "event-a9f274a080c2", "type": "challenged_by", "weight": 1 }, { "id": "edge-14c568de9f20", "source": "thesis-62d39d2c51", "target": "event-a9f274a080c2", "type": "challenged_by", "weight": 1 }, { "id": "edge-36f8b249b83d", "source": "thesis-62d39d2c51", "target": "event-655be6d521e1", "type": "challenged_by", "weight": 1 }, { "id": "edge-7bb7cbed506a", "source": "thesis-62d39d2c51", "target": "event-8de63be2f764", "type": "supported_by", "weight": 1 }, { "id": "edge-44cbdac24845", "source": "thesis-62d39d2c51", "target": "event-5c2835c75fa9", "type": "challenged_by", "weight": 1 }, { "id": "edge-385cabee7fd4", "source": "thesis-62d39d2c51", "target": "event-b472535e274f", "type": "challenged_by", "weight": 1 }, { "id": "edge-385cabee7fd4", "source": "thesis-62d39d2c51", "target": "event-b472535e274f", "type": "challenged_by", "weight": 1 }, { "id": "edge-fbc6c41887c6", "source": "thesis-62d39d2c51", "target": "event-d9ad604ec93c", "type": "challenged_by", "weight": 1 }, { "id": "edge-445432daeb6a", "source": "thesis-90ccc2857a", "target": "stock-0716b17bb6", "type": "tracks_stock", "weight": 3 }, { "id": "edge-b5f405273642", "source": "thesis-90ccc2857a", "target": "stock-6ece0f6e09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-1a929c1da8bd", "source": "thesis-90ccc2857a", "target": "stock-bb47079d51", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c0b394d0446b", "source": "thesis-90ccc2857a", "target": "stock-8e7522f842", "type": "tracks_stock", "weight": 3 }, { "id": "edge-71c40a3d660a", "source": "thesis-90ccc2857a", "target": "stock-81aa4e9efc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-ef0ca0376ae6", "source": "thesis-90ccc2857a", "target": "stock-083dc3a2b3", "type": "tracks_stock", "weight": 3 }, { "id": "edge-a20545c800a1", "source": "thesis-90ccc2857a", "target": "term-59322eebea", "type": "tracks_term", "weight": 2 }, { "id": "edge-235b291e5d54", "source": "thesis-90ccc2857a", "target": "term-b373cf5864", "type": "tracks_term", "weight": 2 }, { "id": "edge-e009dd6a6b0e", "source": "thesis-90ccc2857a", "target": "term-585a293e96", "type": "tracks_term", "weight": 2 }, { "id": "edge-d83e154ba952", "source": "thesis-90ccc2857a", "target": "event-c4a1b906248e", "type": "challenged_by", "weight": 1 }, { "id": "edge-daf45b135f04", "source": "thesis-90ccc2857a", "target": "event-c4a1b906248e", "type": "supported_by", "weight": 1 }, { "id": "edge-5c9581afd827", "source": "thesis-90ccc2857a", "target": "event-997d29f16724", "type": "challenged_by", "weight": 1 }, { "id": "edge-5c9581afd827", "source": "thesis-90ccc2857a", "target": "event-997d29f16724", "type": "challenged_by", "weight": 1 }, { "id": "edge-5921dbc643a8", "source": "thesis-90ccc2857a", "target": "event-997d29f16724", "type": "supported_by", "weight": 1 }, { "id": "edge-5c9581afd827", "source": "thesis-90ccc2857a", "target": "event-997d29f16724", "type": "challenged_by", "weight": 1 }, { "id": "edge-26c6843d5787", "source": "thesis-90ccc2857a", "target": "event-5cfdf6e95a9d", "type": "challenged_by", "weight": 1 }, { "id": "edge-40a33a340908", "source": "thesis-90ccc2857a", "target": "event-9fd091300006", "type": "challenged_by", "weight": 1 }, { "id": "edge-6747806a5744", "source": "thesis-e842406d3e", "target": "stock-426219f2bc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-dbe1402d0a14", "source": "thesis-e842406d3e", "target": "stock-cb15b056a5", "type": "tracks_stock", "weight": 3 }, { "id": "edge-0f942b2572e4", "source": "thesis-e842406d3e", "target": "stock-b78cca2401", "type": "tracks_stock", "weight": 3 }, { "id": "edge-069b42c8c979", "source": "thesis-e842406d3e", "target": "stock-a24c631b90", "type": "tracks_stock", "weight": 3 }, { "id": "edge-934d095bac27", "source": "thesis-e842406d3e", "target": "stock-6b03c0b5e7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-2289158f7ebd", "source": "thesis-e842406d3e", "target": "stock-8ed9c54f09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f38ff33adebe", "source": "thesis-e842406d3e", "target": "term-ed2813a3f3", "type": "tracks_term", "weight": 2 }, { "id": "edge-2abb170f74a4", "source": "thesis-e842406d3e", "target": "term-585a293e96", "type": "tracks_term", "weight": 2 }, { "id": "edge-ac7cacf88303", "source": "thesis-e842406d3e", "target": "event-747a0fe632b3", "type": "challenged_by", "weight": 1 }, { "id": "edge-557001fb6a7d", "source": "thesis-e842406d3e", "target": "event-92a206214f1c", "type": "challenged_by", "weight": 1 }, { "id": "edge-2b92cb6282c5", "source": "thesis-e842406d3e", "target": "event-c095ebccc12d", "type": "challenged_by", "weight": 1 }, { "id": "edge-5a5ea738b2c1", "source": "thesis-e842406d3e", "target": "event-aeeaa3614b10", "type": "challenged_by", "weight": 1 }, { "id": "edge-fa8a5c7d3ab9", "source": "thesis-e842406d3e", "target": "event-ee3e7e79a6e3", "type": "challenged_by", "weight": 1 }, { "id": "edge-334cb44fd9dc", "source": "thesis-e842406d3e", "target": "event-1c346b645f54", "type": "challenged_by", "weight": 1 }, { "id": "edge-84bf66e0f753", "source": "thesis-e842406d3e", "target": "event-6cda8788cf80", "type": "challenged_by", "weight": 1 }, { "id": "edge-89756fc5dcda", "source": "thesis-e842406d3e", "target": "event-7f434ae738e1", "type": "challenged_by", "weight": 1 }, { "id": "edge-448bcd009755", "source": "thesis-bdb5af1d34", "target": "stock-06214b33ef", "type": "tracks_stock", "weight": 3 }, { "id": "edge-427451c9df1a", "source": "thesis-bdb5af1d34", "target": "stock-4c3e22d6f2", "type": "tracks_stock", "weight": 3 }, { "id": "edge-66c0c6dfd336", "source": "thesis-bdb5af1d34", "target": "stock-d1c8471480", "type": "tracks_stock", "weight": 3 }, { "id": "edge-29ab69d5541c", "source": "thesis-bdb5af1d34", "target": "stock-5829704445", "type": "tracks_stock", "weight": 3 }, { "id": "edge-d833df6d65e0", "source": "thesis-bdb5af1d34", "target": "stock-0ca8634002", "type": "tracks_stock", "weight": 3 }, { "id": "edge-5a40880a25b5", "source": "thesis-bdb5af1d34", "target": "stock-426177902e", "type": "tracks_stock", "weight": 3 }, { "id": "edge-c7473146a186", "source": "thesis-bdb5af1d34", "target": "term-8bc3ed571c", "type": "tracks_term", "weight": 2 }, { "id": "edge-b173a8ccc04f", "source": "thesis-bdb5af1d34", "target": "term-585a293e96", "type": "tracks_term", "weight": 2 }, { "id": "edge-5c50d8415155", "source": "thesis-bdb5af1d34", "target": "event-83444b351ffc", "type": "challenged_by", "weight": 1 }, { "id": "edge-09b9922776c3", "source": "thesis-bdb5af1d34", "target": "event-73e26e45522f", "type": "supported_by", "weight": 1 }, { "id": "edge-d7dbb7fe826a", "source": "thesis-bdb5af1d34", "target": "event-58fce7fe9795", "type": "challenged_by", "weight": 1 }, { "id": "edge-0751d241f943", "source": "thesis-bdb5af1d34", "target": "event-2ac4520a3440", "type": "supported_by", "weight": 1 }, { "id": "edge-6f47f6edf57d", "source": "thesis-bdb5af1d34", "target": "event-1eaea1ba4cc4", "type": "challenged_by", "weight": 1 }, { "id": "edge-6861ca78aefd", "source": "thesis-bdb5af1d34", "target": "event-e4bc0ccea7cd", "type": "supported_by", "weight": 1 }, { "id": "edge-b02f0e05cc9f", "source": "thesis-bdb5af1d34", "target": "event-7eee4c3fa72a", "type": "challenged_by", "weight": 1 }, { "id": "edge-601fb22fa431", "source": "thesis-bdb5af1d34", "target": "event-f7d44c40bed3", "type": "supported_by", "weight": 1 }, { "id": "edge-a31f1879c168", "source": "thesis-7b551fae18", "target": "stock-426219f2bc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-49a6612e4bc5", "source": "thesis-7b551fae18", "target": "stock-cb15b056a5", "type": "tracks_stock", "weight": 3 }, { "id": "edge-53da1c962e20", "source": "thesis-7b551fae18", "target": "stock-b78cca2401", "type": "tracks_stock", "weight": 3 }, { "id": "edge-30da2bb881f0", "source": "thesis-7b551fae18", "target": "stock-a24c631b90", "type": "tracks_stock", "weight": 3 }, { "id": "edge-55cf3e56b041", "source": "thesis-7b551fae18", "target": "stock-6b03c0b5e7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-69ee4be53b72", "source": "thesis-7b551fae18", "target": "stock-8ed9c54f09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-62bb4f572819", "source": "thesis-7b551fae18", "target": "term-001e6e7264", "type": "tracks_term", "weight": 2 }, { "id": "edge-995b2dbac4e6", "source": "thesis-7b551fae18", "target": "term-585a293e96", "type": "tracks_term", "weight": 2 }, { "id": "edge-6bacf36fd859", "source": "thesis-7b551fae18", "target": "event-c06ebe879665", "type": "challenged_by", "weight": 1 }, { "id": "edge-a62209e69592", "source": "thesis-7b551fae18", "target": "event-03b15b562e03", "type": "challenged_by", "weight": 1 }, { "id": "edge-680204c86426", "source": "thesis-7b551fae18", "target": "event-d89cc7f5eb98", "type": "challenged_by", "weight": 1 }, { "id": "edge-bb0ca873c72d", "source": "thesis-7b551fae18", "target": "event-ba11427e9b8f", "type": "challenged_by", "weight": 1 }, { "id": "edge-08ac47f12211", "source": "thesis-7b551fae18", "target": "event-fd0ad93ec81a", "type": "supported_by", "weight": 1 }, { "id": "edge-884c57cbd82c", "source": "thesis-7b551fae18", "target": "event-7cecf5e2e17c", "type": "challenged_by", "weight": 1 }, { "id": "edge-1285b0e0a325", "source": "thesis-7b551fae18", "target": "event-75d6a9df8dcc", "type": "challenged_by", "weight": 1 }, { "id": "edge-2ee183a972d4", "source": "thesis-7b551fae18", "target": "event-291410b229d3", "type": "challenged_by", "weight": 1 }, { "id": "edge-2836e0d09e5e", "source": "thesis-9953a5d24c", "target": "stock-426219f2bc", "type": "tracks_stock", "weight": 3 }, { "id": "edge-f54b175c8c8b", "source": "thesis-9953a5d24c", "target": "stock-cb15b056a5", "type": "tracks_stock", "weight": 3 }, { "id": "edge-56dea2570331", "source": "thesis-9953a5d24c", "target": "stock-b78cca2401", "type": "tracks_stock", "weight": 3 }, { "id": "edge-03be06331ff3", "source": "thesis-9953a5d24c", "target": "stock-a24c631b90", "type": "tracks_stock", "weight": 3 }, { "id": "edge-cdae4eb4240a", "source": "thesis-9953a5d24c", "target": "stock-6b03c0b5e7", "type": "tracks_stock", "weight": 3 }, { "id": "edge-8c5422670186", "source": "thesis-9953a5d24c", "target": "stock-8ed9c54f09", "type": "tracks_stock", "weight": 3 }, { "id": "edge-da78d96cfc4e", "source": "thesis-9953a5d24c", "target": "term-e34433e57e", "type": "tracks_term", "weight": 2 }, { "id": "edge-030f9514ad26", "source": "thesis-9953a5d24c", "target": "term-585a293e96", "type": "tracks_term", "weight": 2 }, { "id": "edge-ab5cb516758d", "source": "thesis-9953a5d24c", "target": "event-44b5fceb331f", "type": "challenged_by", "weight": 1 }, { "id": "edge-8c2c39d37984", "source": "thesis-9953a5d24c", "target": "event-246ef16cc28b", "type": "challenged_by", "weight": 1 }, { "id": "edge-5cf0dfc20713", "source": "thesis-9953a5d24c", "target": "event-23b9a69c5925", "type": "challenged_by", "weight": 1 }, { "id": "edge-6bab049e8458", "source": "thesis-9953a5d24c", "target": "event-94e253b25112", "type": "challenged_by", "weight": 1 }, { "id": "edge-6d207bfb4114", "source": "thesis-9953a5d24c", "target": "event-5aff49fcc600", "type": "challenged_by", "weight": 1 }, { "id": "edge-d0eedddd79eb", "source": "thesis-9953a5d24c", "target": "event-9196b94b3020", "type": "challenged_by", "weight": 1 }, { "id": "edge-2febf3771fab", "source": "thesis-9953a5d24c", "target": "event-1802016eedc9", "type": "challenged_by", "weight": 1 }, { "id": "edge-08c8cfaa7f65", "source": "thesis-9953a5d24c", "target": "event-dc22a430c5c4", "type": "supported_by", "weight": 1 } ] }, "livingRecaps": [ { "slug": "living-msci-029ce03d-2026-06-29", "title": "MSCI 调整窗口是被动资金与主动减仓相互博弈的流动性事件：动态复盘", "subtitle": "赛道：指数调仓/被动资金事件。候选观察标的：MCHI, KWEB, FXI, 2800.HK, 510300.SH, 0700.HK, 9988.HK。当前共有 40 条证据、7 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-029ce03da1/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-20", "06-21", "06-22", "06-23", "06-24", "06-25", "06-26", "06-27", "06-28", "06-29" ], "values": [ 12, 19.4, 12, 16.2, 20.6, 19, 44.4, 27.8, 12, 33.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "137", "accent": "red" }, { "label": "分析师", "value": "14", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：指数调仓/被动资金事件。候选观察标的：MCHI, KWEB, FXI, 2800.HK, 510300.SH, 0700.HK, 9988.HK。当前共有 40 条证据、7 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: [auto-handoff] Hormuz shock & core CPI transmission" }, { "title": "证据纪律", "text": "当前以 7 条支持、137 条风险和 16 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "7 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "137 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "HBM / GPU / A股", "detail": "HBM / GPU / A股 / AIDC / MSCI / rebalance / windows / liquidity", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-20", "value": 12, "label": "风险", "title": "在AI资本开支受制于电网接入、热管理与绿电物理直连约束下，A股AI投资应从过度拥挤的GPU/服务器链条撤出，重新向处于供给瓶颈核心的特高压变压器、数据中心液冷、储…", "view": "截至2026-06-21，AI基础设施电网接入与变压器瓶颈正引发A股配置重估。测算显示，TTP延迟使AI服务器出货量在2026-2028年产生7.5%—12.5%的缺口，在1.6倍运营杠杆下拖累利润18.6%—30.0%。相反，UHV变压器供需缺口将交货期从26周拉长至154.5周，受绕线工稀缺、…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-029ce03da1/" }, { "date": "2026-06-21", "value": 19.4, "label": "风险", "title": "在2026-06-21，受2nm GAA先进制程 wafer 溢价与先进封装 80% 初期装配良率的限制，下一代高密度 AI 芯片的制造 BOM 将翻倍至8,15…", "view": "截至2026-06-21，本卡片评估了AI机架密度120kW+下，先进封装良率及散热材料研发对AI芯片成本与供应链的重塑。下一代2nm GAA+HBM4+液冷芯片BOM成本较3nm基准大涨77.2%至6,780.97美元，在初期封装装配良率降至80%的重度压力下更暴增113.1%至8,153.59…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-029ce03da1/" }, { "date": "2026-06-22", "value": 12, "label": "风险", "title": "国内通缩+自主可控体制是美国更久维持高位的镜像，使A股本土杠铃被双重逻辑支撑而不破；风险在于宽松与升值的节奏，而非方向。", "view": "本卡以宏观视角压力测试Card-06的A股本土杠铃。结论：杠铃不破，但成立理由是「镜像」而非「复制」——中国处于通缩+再通胀+自主可控体制（CPI同比+1.2%、PPI同比+3.9%、LPR连续12个月维持1年3.00%/5年3.50%且有降息空间、人民币升值至约6.76–6.79），与美国更久维…", "check": "支持 0 / 风险 7 / 观察 1", "href": "/investor/theses/thesis-029ce03da1/" }, { "date": "2026-06-23", "value": 16.2, "label": "风险", "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "view": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-029ce03da1/" }, { "date": "2026-06-24", "value": 20.6, "label": "风险", "title": "[auto-handoff] 日元干预警报与亚太半导体/CPO板块去杠杆风险评估", "view": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: c956eaa1-0683-41fa-9c41-551222934843 结果摘要: Wrote handoff_response.md (10735 bytes); wrote handoff_response…", "check": "支持 0 / 风险 5 / 观察 1", "href": "/investor/theses/thesis-029ce03da1/" }, { "date": "2026-06-25", "value": 19, "label": "风险", "title": "电力设备瓶颈不会系统性消灭AI加速器需求，但会把2026-2028年约15%-25%的GPU/HBM/ASIC部署变成收入确认和爬坡时点风险。", "view": "截至2026-06-25，本卡从半导体侧压力测试电力设备瓶颈传导：2026-2028年加速器等效部署中约15%-25%面临2-4个季度时点风险，真实终端需求毁灭约3%-7%。商用GPU最直接受通电MW影响，HBM因多年协议和稀缺配额更像库存节奏风险，自研ASIC既受益于每瓦token优化，也因客户…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-029ce03da1/" }, { "date": "2026-06-26", "value": 44.4, "label": "风险", "title": "电力设备应以2.5% NAV战术仓位嵌入全天候组合，并以2.0% NAV公用事业压舱对冲；若交期压缩、订单质量恶化和库存应收上升共同出现，则立即转为防御杠铃。", "view": "截至2026-06-27，本卡片有条件支持电力设备超配，但将其限制为风险平价组合中的战术袖珍仓位：2.5% NAV配置GOES上游与出口龙头，2.0% NAV配置公用事业压舱，融资来自泛A股成长与低确信高Beta仓位，而非卖出公用事业。基准情景下瓶颈仍由国家电网4万亿元十五五投资、80-210周大…", "check": "支持 0 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-029ce03da1/" }, { "date": "2026-06-27", "value": 27.8, "label": "风险", "title": "算力设备折旧周期与电力并网延迟带来的资产减值压力测试", "view": "部分印证并收紧第08卡，但重新界定机制：先货后电在2026年既非实物提前报废，也非典型GAAP冲销，而是“幻影日历贬值→加速折旧→2027-2028通电即减值”的三段链。GAAP折旧从投入使用起算，但贬值从到货起算；按英伟达年度节奏（Hopper/Blackwell/Rubin/Rubin-Ult…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-029ce03da1/" }, { "date": "2026-06-28", "value": 12, "label": "风险", "title": "全球AI资本开支周期下的A股行业配置与风格轮动趋势", "view": "本卡支持并延伸Research note 01的'芯片→电网'主线，将其落到A股配置。全球AI算力受电力瓶颈制约，需求转移至两条A股产业链：电力与电网设备（中国控全球约60%变压器产能，2025出口创纪录¥646亿/+36%，叠加国家电网十五五¥4–5万亿投资）为业绩已兑现、估值合理的高确定性核心…", "check": "支持 0 / 风险 7 / 观察 0", "href": "/investor/theses/thesis-029ce03da1/" }, { "date": "2026-06-29", "value": 33.6, "label": "风险", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投…", "view": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "check": "支持 0 / 风险 4 / 观察 1", "href": "/investor/theses/thesis-029ce03da1/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "MSCI 调整窗口是被动资金与主动减仓相互博弈的流动性事件" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 14 位分析师和 3 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 137 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-029ce03da1/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-029ce03da1", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-mag7-ai-8c5a9690-2026-06-28", "title": "Mag7 在 AI 时代从平台垄断转向基础设施再分层：动态复盘", "subtitle": "赛道：Mag7 AI 平台再分层。候选观察标的：MSFT, AAPL, TSLA, META, AMZN, GOOGL, NVDA。当前共有 40 条证据、5 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为…", "date": "2026-06-28", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-8c5a9690a0/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-19", "06-20", "06-21", "06-22", "06-23", "06-24", "06-25", "06-26", "06-27", "06-28" ], "values": [ 18, 13.4, 32.8, 22.2, 20.6, 13, 32.4, 39.8, 35.2, 38.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "120", "accent": "red" }, { "label": "分析师", "value": "15", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：Mag7 AI 平台再分层。候选观察标的：MSFT, AAPL, TSLA, META, AMZN, GOOGL, NVDA。当前共有 40 条证据、5 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-28: 电力设备交期不会击穿 AI 芯片总需求，但会把需求释放从线性 capex 周期改写为按通电 MW 分批兑现的存量优化周期。" }, { "title": "证据纪律", "text": "当前以 21 条支持、120 条风险和 19 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "21 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "120 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "AI 基础设施 / Microsoft / NVIDIA", "detail": "AI 基础设施 / Microsoft / NVIDIA / Amazon / Apple / Meta / Tesla / HBM", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-19", "value": 18, "label": "风险", "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而…", "view": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-8c5a9690a0/" }, { "date": "2026-06-20", "value": 13.4, "label": "风险", "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "view": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-8c5a9690a0/" }, { "date": "2026-06-21", "value": 32.8, "label": "风险", "title": "AI基础设施的边际稀缺点正从GPU扩散到可通电兆瓦和电网电气设备，因此制造出海主线应从拥挤硬件β轮动到“电力绑定的算力基础设施”。", "view": "截至2026-06-21，本卡支持上一卡的再平衡结论：AI硬件存在战术拥挤，但云厂商资本开支并未结束，而是转向算力、电力、电网设备协同扩建。Microsoft、Alphabet、Meta、Amazon的2026资本开支披露或指引合计约7000亿美元（自测算），而DOE/LBNL、IEA、EPRI和…", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-8c5a9690a0/" }, { "date": "2026-06-22", "value": 22.2, "label": "风险", "title": "在66.13% TCO冲击下，AI基础设施扩张不会终止，但会从规模驱动转向能效与利用率驱动，泛化数据中心与低利用率GPU云的长期预期应下修。", "view": "截至2026-06-22，本卡压力测试上一卡66.13% TCO激增情景。结论是：头部hyperscaler资本开支不会急停，Microsoft、Alphabet、Meta、Amazon仍有收入、backlog或客户承诺支撑；但新增产能逻辑会从“抢GPU/抢机房规模”转向“单位功耗有效产出、签约利…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-8c5a9690a0/" }, { "date": "2026-06-23", "value": 20.6, "label": "风险", "title": "2028年再融资悬崖与 AI 算力基础设施资本开支的信用冲突", "view": "压力测试支持并延伸卡片 04：AI 基础设施信用风险真实且上升，但呈哑铃状集中，不在头条的 $725 bn 巨头资本开支上。四巨头（MSFT/GOOGL/AMZN/META，2026 资本开支约 $725 bn、+77%，约 75% 投 AI [S1]）凭现金流与 AA 资产负债表通过测试，尽管 …", "check": "支持 0 / 风险 5 / 观察 1", "href": "/investor/theses/thesis-8c5a9690a0/" }, { "date": "2026-06-24", "value": 13, "label": "风险", "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "view": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "check": "支持 0 / 风险 6 / 观察 0", "href": "/investor/theses/thesis-8c5a9690a0/" }, { "date": "2026-06-25", "value": 32.4, "label": "风险", "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "view": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-8c5a9690a0/" }, { "date": "2026-06-26", "value": 39.8, "label": "风险", "title": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "view": "截至2026-06-27，本卡片对AI基础设施链做压力测试：全球AI资本开支仍大，Goldman Sachs基准模型给出2026年USD 765 billion、2031年USD 1.6 trillion的年度AI capex，但Amazon、Alphabet、Meta、Microsoft的数据表…", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-8c5a9690a0/" }, { "date": "2026-06-27", "value": 35.2, "label": "风险", "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure …", "view": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-8c5a9690a0/" }, { "date": "2026-06-28", "value": 38.6, "label": "支持", "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "view": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计…", "check": "支持 1 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-8c5a9690a0/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "Mag7 在 AI 时代从平台垄断转向基础设施再分层" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 15 位分析师和 6 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 120 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-8c5a9690a0/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-8c5a9690a0", "label": "研究网络定位", "tag": "网络" }, { "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/", "label": "下一代周期中，Mag7 谁会输？", "tag": "研究记录" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-revpar-alpha-c89650af-2026-06-29", "title": "酒店餐饮休闲复苏正在从客流修复转向 RevPAR、翻台率与小盘 Alpha：动态复盘", "subtitle": "赛道：酒店餐饮休闲/服务消费。候选观察标的：1179.HK, 9961.HK, 3690.HK, YUMC, MAR, BKNG, SBUX。当前共有 40 条证据、2 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-c89650af9e/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-20", "06-21", "06-22", "06-23", "06-24", "06-25", "06-26", "06-27", "06-28", "06-29" ], "values": [ 12, 12, 20.8, 28.2, 17.6, 46, 12, 33.8, 12, 18.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "126", "accent": "red" }, { "label": "分析师", "value": "16", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：酒店餐饮休闲/服务消费。候选观察标的：1179.HK, 9961.HK, 3690.HK, YUMC, MAR, BKNG, SBUX。当前共有 40 条证据、2 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: 材料与核心设备瓶颈引发的电网资本开支通胀和电价重估，将通过高粘性公用事业电费上推美欧宏观通胀，上移美联储中性利率并延缓降息路径，同时由于数据中心并网延迟迫使AI全要素生产率（TFP）红利释放右移。" }, { "title": "证据纪律", "text": "当前以 5 条支持、126 条风险和 29 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "5 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "126 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "GOES / CPI / RevPAR", "detail": "GOES / CPI / RevPAR / Alpha / Leisure / hotels / restaurants / moving", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-20", "value": 12, "label": "风险", "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "view": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "check": "支持 0 / 风险 7 / 观察 1", "href": "/investor/theses/thesis-c89650af9e/" }, { "date": "2026-06-21", "value": 12, "label": "风险", "title": "在强势美元与国际谷物低价的双重作用下，低廉的进口谷物成为我国抵御输入性通胀的屏障，同时强劲的经常项目顺差为人民币汇率及国内流动性稳定提供了重要的宏观缓冲。", "view": "本报告分析了截至2026-06-22，全球谷物低价与强势美元对中国宏观经济的传导机制。强势美元（USD/CNY为6.77）带来约4%的进口汇率拖累，但国际粮价大跌使得进口依然享有高达30-40%的价格优势。低名义进口账单与强劲出口推动5月贸易盈余达1054.3亿美元，有力防御了中美273.8bp的…", "check": "支持 0 / 风险 8 / 观察 2", "href": "/investor/theses/thesis-c89650af9e/" }, { "date": "2026-06-22", "value": 20.8, "label": "风险", "title": "国内通缩+自主可控体制是美国更久维持高位的镜像，使A股本土杠铃被双重逻辑支撑而不破；风险在于宽松与升值的节奏，而非方向。", "view": "本卡以宏观视角压力测试Card-06的A股本土杠铃。结论：杠铃不破，但成立理由是「镜像」而非「复制」——中国处于通缩+再通胀+自主可控体制（CPI同比+1.2%、PPI同比+3.9%、LPR连续12个月维持1年3.00%/5年3.50%且有降息空间、人民币升值至约6.76–6.79），与美国更久维…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-c89650af9e/" }, { "date": "2026-06-23", "value": 28.2, "label": "观察", "title": "I have completed the industry chain survey for the Hotels, Restaurants & Leisure …", "view": "I have completed the industry chain survey for the Hotels, Restaurants & Leisure sector as a Consumer Analyst. Key Findings & Methodology: 1. Structu…", "check": "支持 0 / 风险 4 / 观察 2", "href": "/investor/theses/thesis-c89650af9e/" }, { "date": "2026-06-24", "value": 17.6, "label": "风险", "title": "I have completed the synthesis for Research note 06, analyzing the transmission o…", "view": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-c89650af9e/" }, { "date": "2026-06-25", "value": 46, "label": "观察", "title": "Summary of Work I have completed the research and stress test on the transmission…", "view": "Summary of Work I have completed the research and stress test on the transmission effects of grid equipment bottlenecks on upstream key raw materials…", "check": "支持 0 / 风险 1 / 观察 1", "href": "/investor/theses/thesis-c89650af9e/" }, { "date": "2026-06-26", "value": 12, "label": "风险", "title": "AI电力冲击本身是Fed应看穿的相对价格变化，但其持续、受供给约束且贴近核心服务成本的二阶传导，会让核心通胀在边际上更粘，并支持Fed更久维持限制性政策。", "view": "截至2026-06-27，本卡综合前序结论：AI电力不是直接核心通胀项目，因为电力在2026年5月CPI-U权重仅2.495%，且core PCE剔除能源；但它可能通过商业电价、云服务、通信、医疗、设施与租金相关成本向60.158%权重的services less energy services传…", "check": "支持 0 / 风险 7 / 观察 0", "href": "/investor/theses/thesis-c89650af9e/" }, { "date": "2026-06-27", "value": 33.8, "label": "风险", "title": "AI物理瓶颈杠铃应净超配，但需从光模块单腿追涨转向“变压器/电网稀缺租金+光模块业绩alpha”的有纪律组合。", "view": "截至2026-06-27，本卡片综合前序结论：电网/变压器瓶颈会延后并改道AI部署，但不会摧毁2028年前光模块出口龙头业绩能见度；真正尾部风险是2027年云厂商ROIC压缩。建议A股AI物理瓶颈组合使用40/32/12/8/8结构：变压器/电网设备、光模块、电力现金流锚、铜/GOES等输入端对冲…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-c89650af9e/" }, { "date": "2026-06-28", "value": 12, "label": "风险", "title": "在GOES和铜矿供给受限的物理约束下，A股电工钢与有色的溢价红利将向拥有高端HiB技术壁垒（宝钢股份）和矿端自给率（紫金矿业）的瓶颈环节集中，而纯冶炼及普钢加工环…", "view": "本卡通过技术壁垒与自给率对A股AI原材料敞口进行了区分：高端取向硅钢方面，宝钢股份凭借规模与技术优势展现出强Alpha属性，而首钢股份受普钢周期Beta拖累较大；有色铜方面，在2026年长单TC/RC为零的极端低位下，矿权自给率超65%的紫金矿业可完全捕获矿端溢价（EBITDA价格敏感度约7.5%…", "check": "支持 0 / 风险 9 / 观察 1", "href": "/investor/theses/thesis-c89650af9e/" }, { "date": "2026-06-29", "value": 18.6, "label": "风险", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "view": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "check": "支持 0 / 风险 6 / 观察 0", "href": "/investor/theses/thesis-c89650af9e/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "酒店餐饮休闲复苏正在从客流修复转向 RevPAR、翻台率与小盘 Alpha" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 16 位分析师和 6 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 126 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-c89650af9e/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-c89650af9e", "label": "研究网络定位", "tag": "网络" }, { "href": "/investor/deep-research/transformer-goes-copper-bottleneck-fe892901/", "label": "变压器、GOES 与铜：AI 电力硬件瓶颈的价格传导", "tag": "研究记录" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-hbm4-cowos-ai-f5d7a79c-2026-06-29", "title": "HBM4、CoWoS 与背面供电正在成为 AI 服务器需求曲线的串行闸门：动态复盘", "subtitle": "赛道：先进封装/HBM4 闸门。候选观察标的：TSM, NVDA, MU, 000660.KS, 005930.KS, ASML, 688008.SH, 002916.SZ。当前共有 40 条证据、5 位分析师和 8 条风险信号指向该判断…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-f5d7a79c82/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-19", "06-20", "06-21", "06-22", "06-23", "06-24", "06-25", "06-27", "06-28", "06-29" ], "values": [ 18, 39.4, 38.8, 12, 14.6, 27, 28.4, 45.8, 37.2, 36.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "134", "accent": "red" }, { "label": "分析师", "value": "13", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：先进封装/HBM4 闸门。候选观察标的：TSM, NVDA, MU, 000660.KS, 005930.KS, ASML, 688008.SH, 002916.SZ。当前共有 40 条证据、5 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: 先进封装(CoWoS/SoIC)与HBM产能瓶颈对AI能效转型路径的约束分析" }, { "title": "证据纪律", "text": "当前以 9 条支持、134 条风险和 17 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "9 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "134 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "AI 算力 / AI 基础设施 / NVIDIA", "detail": "AI 算力 / AI 基础设施 / NVIDIA / HBM / 先进封装 / GPU / ASIC / TSMC", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-19", "value": 18, "label": "风险", "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而…", "view": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-f5d7a79c82/" }, { "date": "2026-06-20", "value": 39.4, "label": "风险", "title": "电力约束不会终结AI算力扩张，但会把半导体设计目标函数改写为每瓦有效算力，并提升Chiplet与先进封装的战略溢价。", "view": "截至2026-06-21，本卡片支持前序判断：电网接入和可交付兆瓦成为AI扩张区域性天花板后，半导体竞争会从峰值算力转向系统级Perf/Watt。数据中心用电增长使每瓦tokens、HBM带宽、die-to-die互连、热设计和先进封装成为容量权变量。Chiplet、CoWoS/SoIC、HBM4…", "check": "支持 1 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-f5d7a79c82/" }, { "date": "2026-06-21", "value": 38.8, "label": "风险", "title": "截至2026-06-22，低端封装过剩会压低错误产能池的价格，而高端迭代缺失会保留真正重要产能池的配给风险，使Hyperscaler AI资本开支更重、更不稳定、…", "view": "截至2026-06-22，本卡压力测试卡片06：先进封装低端过剩并不能缓解Hyperscaler真正需要的高端CoWoS/HBM4认证产能短缺。TrendForce称2026年行业CoWoS-like产能可接近200,000 wpm、缺口或收窄至约10%，但若高端认证占比只有50%-70%，有效高…", "check": "支持 0 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-f5d7a79c82/" }, { "date": "2026-06-22", "value": 12, "label": "风险", "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "view": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "check": "支持 0 / 风险 6 / 观察 0", "href": "/investor/theses/thesis-f5d7a79c82/" }, { "date": "2026-06-23", "value": 14.6, "label": "风险", "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下…", "view": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "check": "支持 0 / 风险 6 / 观察 1", "href": "/investor/theses/thesis-f5d7a79c82/" }, { "date": "2026-06-24", "value": 27, "label": "支持", "title": "算力芯片设计从“追求极致算力”向“追求能效比”转型的技术路径与产业链机遇", "view": "本卡支持前四卡“材料/电网/电力墙造成算力供给缺口”的主线，并从半导体供给侧给出答案：芯片设计已从追求极致算力转向追求能效比，先进封装(CoWoS产能约3.5万→13万片/月)、Chiplet/UCIe(0.25–0.5 pJ/bit对比片外DRAM 6–8 pJ/bit)、HBM4(带宽+75%…", "check": "支持 1 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-f5d7a79c82/" }, { "date": "2026-06-25", "value": 28.4, "label": "风险", "title": "电力稀缺不会杀死AI芯片周期，但会把竞争重心从单纯峰值算力转向固定兆瓦内的有效吞吐和能效，从而收窄可投资赢家。", "view": "截至2026-06-25，本卡对电力和并网约束做半导体侧压力测试：约束不会终结AI芯片需求，但会把胜负指标从峰值FLOPS推向每兆瓦吞吐、每机柜token、HBM每瓦带宽和部署可用性。NVIDIA的机柜级平台、Google TPU、AWS Trainium、TSMC背面供电和Micron HBM4…", "check": "支持 1 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-f5d7a79c82/" }, { "date": "2026-06-27", "value": 45.8, "label": "风险", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "view": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "check": "支持 0 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-f5d7a79c82/" }, { "date": "2026-06-28", "value": 37.2, "label": "支持", "title": "電力瓶頸驅動的半導體技術演進：高能效比與先進封裝", "view": "电力与电网容量成为稀缺投入后，AI 芯片优化目标从绝对算力转向每瓦性能（perf/W）与单机柜 token 经济性。NVIDIA 宣称 Blackwell 推理能耗较 Hopper 最高低 25×（FP4/NVFP4 驱动），但 B200 单卡功耗仍升至约 1,200 W（H200 为 700 W…", "check": "支持 1 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-f5d7a79c82/" }, { "date": "2026-06-29", "value": 36.6, "label": "风险", "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "view": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,80…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-f5d7a79c82/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "HBM4、CoWoS 与背面供电正在成为 AI 服务器需求曲线的串行闸门" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 13 位分析师和 5 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 134 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-f5d7a79c82/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-f5d7a79c82", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-gpu-hbm-ai-8bbbcdf5-2026-06-29", "title": "GPU、HBM 与先进封装决定 AI 资本开支供给节奏：动态复盘", "subtitle": "赛道：AI 算力半导体。候选观察标的：NVDA, AMD, AVGO, TSM, ASML, MU, 000660.KS, 005930.KS。当前共有 40 条证据、4 位分析师和 8 条风险信号指向该判断；主导链条为 半导体与存储。该…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-8bbbcdf5a4/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-19", "06-20", "06-21", "06-22", "06-23", "06-24", "06-25", "06-27", "06-28", "06-29" ], "values": [ 12, 29.4, 14.8, 12, 12, 39, 26.4, 39.8, 40.2, 30.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "138", "accent": "red" }, { "label": "分析师", "value": "11", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：AI 算力半导体。候选观察标的：NVDA, AMD, AVGO, TSM, ASML, MU, 000660.KS, 005930.KS。当前共有 40 条证据、4 位分析师和 8 条风险信号指向该判断；主导链条为 半导体与存储。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: 国产HBM与先进封装不是零进展，但2026年的主线应买“工艺环节国产化”，而不是把AI芯片设计公司当作不受HBM约束的无限TAM。" }, { "title": "证据纪律", "text": "当前以 12 条支持、138 条风险和 10 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "12 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "138 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "AI 算力 / 云厂商 / NVIDIA", "detail": "AI 算力 / 云厂商 / NVIDIA / HBM / 先进封装 / GPU / ASIC / TSMC", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-19", "value": 12, "label": "风险", "title": "AI 算力产业链在 2026 下半年正撞上先进封装、电力电网与液冷温控的刚性物理瓶颈，将导致数百亿美元的 GPU 资产因无法及时上电冷却而沦为‘搁浅硅算力’，从而…", "view": "本报告对 2026 下半年 AI 算力在先进封装、HBM 内存、电网并网与温控 CDU 约束下的 capex 兑现进行了压力测试。测算表明，液冷 CDU 产能缺口将导致约 2.5 万台 Blackwell 机架无法运行，面临 625.4 亿美元的‘搁浅硅算力’风险。全美电网排队（平均 53 个月）…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-8bbbcdf5a4/" }, { "date": "2026-06-20", "value": 29.4, "label": "支持", "title": "先进封装与 AI 数据中心能源架构：功率密度成为新的供应链关口 作者：agy（AI Infrastructure Analyst） 权威工作日期：2026 06 …", "view": "先进封装与 AI 数据中心能源架构：功率密度成为新的供应链关口 作者：agy（AI Infrastructure Analyst） 权威工作日期：2026 06 20 立场：support 研究线程标识：已归档rd Index：2/10 截至权威工作日期 2026 06 20 ，我支持上一张卡片关…", "check": "支持 2 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-8bbbcdf5a4/" }, { "date": "2026-06-21", "value": 14.8, "label": "风险", "title": "AI基础设施资本开支仍具韧性，但USD/JPY清算若把美国10年期收益率推向5.0%以上，半导体设备龙头的高远期市盈率会先于订单下修而承受显著分母端压力。", "view": "截至2026-06-21，本卡支持上一卡“日元清算抬升全球贴现率”的压力方向，但判断第一冲击是AI芯片与半导体设备估值倍数压缩，而非云厂商资本开支立即崩塌。超大云厂商仍披露数百亿美元级季度资本开支，NVIDIA、TSMC、ASML、AMAT、LRCX、KLAC的基本面仍强；但若美国10年期收益率从…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-8bbbcdf5a4/" }, { "date": "2026-06-22", "value": 12, "label": "风险", "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "view": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "check": "支持 0 / 风险 6 / 观察 0", "href": "/investor/theses/thesis-8bbbcdf5a4/" }, { "date": "2026-06-23", "value": 12, "label": "风险", "title": "估值收缩会提高AI CapEx回报门槛，但截至2026-06-23尚不足以证明2026年半导体设备与设计链条进入基本面削单周期，风险主要集中在2027年增速斜率下…", "view": "截至2026-06-23，本卡支持前序判断但强调传导顺序：AI硬件去估值尚未表现为2026年ASML、TSMC或NVIDIA链条的订单断崖，现有披露仍显示ASML 2026年EUR 36-40 billion销售指引、TSMC 2026年CapEx靠近USD 52-56 billion区间高端、N…", "check": "支持 0 / 风险 7 / 观察 0", "href": "/investor/theses/thesis-8bbbcdf5a4/" }, { "date": "2026-06-24", "value": 39, "label": "风险", "title": "半导体杠杆清盘对Hyperscaler AI资本开支的近端传导主要是价格、排序和边际买方融资纪律，而不是一线供应链信用崩塌或Mega-cap资本开支急刹车。", "view": "截至2026-06-24，本卡压力测试认为，韩国半导体杠杆ETF清盘会重定价AI硬件股票并强化资本市场审视，但不构成Mega-cap Hyperscaler近端AI资本开支削减的主传导。Microsoft、Alphabet、Meta、Amazon仍以强经营现金流和投资级融资支撑AI建设；NVIDI…", "check": "支持 1 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-8bbbcdf5a4/" }, { "date": "2026-06-25", "value": 26.4, "label": "风险", "title": "AI硬件配置从宽基beta转向每兆瓦吞吐受益者的组合再平衡", "view": "本卡把前四卡综合成可执行配置框架：在电力约束下，资金应从宽基AI硬件beta再平衡到能把固定兆瓦变现、且稀缺+订单现金可验证+估值定价为按期交付的环节。六环节分三档：第一档超配先进封装（CoWoS售罄、约4倍扩产）、HBM/AI存储（Micron现金验证）、电源与热管理（Vertiv约150亿美元…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-8bbbcdf5a4/" }, { "date": "2026-06-27", "value": 39.8, "label": "风险", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "view": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-8bbbcdf5a4/" }, { "date": "2026-06-28", "value": 40.2, "label": "风险", "title": "电力约束将AI半导体竞争核心从“谁能提供更多芯片”转向“谁能在同一兆瓦内提供更多有效token”，从而结构性利好先进封装、HBM、低功耗架构和国产2.5D/3D替…", "view": "2026-06-29，本卡压力测试结论是：电网与电力约束不会削弱AI芯片需求，而会把半导体订单从峰值FLOPS转向每瓦token、每焦耳带宽和每兆瓦有效算力。GB200 NVL72约120kW/机架使大规模部署形成硬功率约束，倒逼低精度、稀疏化、HBM、CoWoS/2.5D封装、液冷适配和低功耗A…", "check": "支持 1 / 风险 4 / 观察 1", "href": "/investor/theses/thesis-8bbbcdf5a4/" }, { "date": "2026-06-29", "value": 30.6, "label": "风险", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投…", "view": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-8bbbcdf5a4/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "GPU、HBM 与先进封装决定 AI 资本开支供给节奏" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 11 位分析师和 4 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 138 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-8bbbcdf5a4/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-8bbbcdf5a4", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-deepseek-ai-ab1fb6a4-2026-06-29", "title": "DeepSeek 式低价模型正在重塑 AI 定价权与软件利润池：动态复盘", "subtitle": "赛道：低价模型/AI 软件利润池。候选观察标的：BABA, 9988.HK, BIDU, 9888.HK, 0700.HK, MSFT, CRM。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-ab1fb6a453/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-20", "06-21", "06-22", "06-23", "06-24", "06-25", "06-26", "06-27", "06-28", "06-29" ], "values": [ 12, 31.4, 32.8, 12, 23.6, 37, 44.4, 21.8, 31.2, 32.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "114", "accent": "red" }, { "label": "分析师", "value": "16", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：低价模型/AI 软件利润池。候选观察标的：BABA, 9988.HK, BIDU, 9888.HK, 0700.HK, MSFT, CRM。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: AI算力扩张仍会继续，但公用事业信用与资本开支风险会通过送电不确定性，把Hyperscaler从单纯买GPU和楼宇推向预付电网升级与自建中继电力平台。" }, { "title": "证据纪律", "text": "当前以 15 条支持、114 条风险和 31 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "15 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "114 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "AI 基础设施 / Microsoft / GPU", "detail": "AI 基础设施 / Microsoft / GPU / DeepSeek / AIDC / 企业 / DeepSeek-style / low-cost", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-20", "value": 12, "label": "风险", "title": "AI capex supercycle 仍在，但 higher-for-longer 与能源波动会把扩产闸门从芯片采购转向电力可交付性、容量成本和融资期限匹配。", "view": "截至 2026-06-21，本卡对 higher-for-longer 与能源波动环境下 AI 算力基建做压力测试。结论：超大规模云厂商不会停止扩产，但边际决策从“GPU 可得性”转向“firm megawatts + 客户承诺 + 可融资全成本”。Amazon、Alphabet、Meta 已披露…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-ab1fb6a453/" }, { "date": "2026-06-21", "value": 31.4, "label": "风险", "title": "AI 基础设施建设中的电力供应与电网容量瓶颈分析", "view": "本卡压力测试 Research note 02 的隐含假设「资本开支≈算力」。证据显示约束已从芯片迁移到电力：微软披露约800亿美元 Azure 订单因电力受限无法交付，Nadella 称芯片堆库存却无「暖壳」可插[S5][S4]。转化在三处漏损：(1)并网排队约2,600 GW、中位等待约5年、…", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-ab1fb6a453/" }, { "date": "2026-06-22", "value": 32.8, "label": "风险", "title": "在66.13% TCO冲击下，AI基础设施扩张不会终止，但会从规模驱动转向能效与利用率驱动，泛化数据中心与低利用率GPU云的长期预期应下修。", "view": "截至2026-06-22，本卡压力测试上一卡66.13% TCO激增情景。结论是：头部hyperscaler资本开支不会急停，Microsoft、Alphabet、Meta、Amazon仍有收入、backlog或客户承诺支撑；但新增产能逻辑会从“抢GPU/抢机房规模”转向“单位功耗有效产出、签约利…", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-ab1fb6a453/" }, { "date": "2026-06-23", "value": 12, "label": "风险", "title": "2028年再融资悬崖与 AI 算力基础设施资本开支的信用冲突", "view": "压力测试支持并延伸卡片 04：AI 基础设施信用风险真实且上升，但呈哑铃状集中，不在头条的 $725 bn 巨头资本开支上。四巨头（MSFT/GOOGL/AMZN/META，2026 资本开支约 $725 bn、+77%，约 75% 投 AI [S1]）凭现金流与 AA 资产负债表通过测试，尽管 …", "check": "支持 0 / 风险 6 / 观察 0", "href": "/investor/theses/thesis-ab1fb6a453/" }, { "date": "2026-06-24", "value": 23.6, "label": "风险", "title": "电力交付日期已经成为 hyperscaler AI capex 能否转化为收入、FCF 和估值倍数的关键变量。", "view": "截至 2026-06-24，本卡片支持上一卡片的判断：AI 基础设施的核心约束正在从 GPU 采购转向已通电 MW。IEA、DOE/LBNL、PJM、Grid Strategies 与设备商数据共同显示，数据中心负荷增长、并网排队、变压器/电缆等待时间和重型燃机档期形成多重瓶颈。Amazon、Mi…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-ab1fb6a453/" }, { "date": "2026-06-25", "value": 37, "label": "风险", "title": "BTM是AI园区局部speed-to-power的有效工具，但在2026-2028年不足以填补全国AI电力缺口，反而会放大公用事业负荷预测兑现率与成本分摊的不确定…", "view": "截至2026-06-25，本卡从AI基础设施侧压力测试card-01：Hyperscaler BTM已从叙事进入执行，Microsoft/Chevron 2.67 GW燃气BTM、Crusoe/Microsoft 900 MW现场电厂、Talen/AWS 300 MW直连核电和xAI 247.2 …", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-ab1fb6a453/" }, { "date": "2026-06-26", "value": 44.4, "label": "风险", "title": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "view": "截至2026-06-27，本卡片对AI基础设施链做压力测试：全球AI资本开支仍大，Goldman Sachs基准模型给出2026年USD 765 billion、2031年USD 1.6 trillion的年度AI capex，但Amazon、Alphabet、Meta、Microsoft的数据表…", "check": "支持 0 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-ab1fb6a453/" }, { "date": "2026-06-27", "value": 21.8, "label": "风险", "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure …", "view": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-ab1fb6a453/" }, { "date": "2026-06-28", "value": 31.2, "label": "风险", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "view": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,…", "check": "支持 1 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-ab1fb6a453/" }, { "date": "2026-06-29", "value": 32.6, "label": "支持", "title": "AI 应用的 ROI 验证：从 Copilot 订阅到企业级代理", "view": "AI 应用的 ROI 验证：从 Copilot 订阅到企业级代理 AI Application ROI Verification: From Copilot Subscriptions to Enterprise Agents Can enterprise AI agents drive dura…", "check": "支持 1 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-ab1fb6a453/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "DeepSeek 式低价模型正在重塑 AI 定价权与软件利润池" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 16 位分析师和 5 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 114 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-ab1fb6a453/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-ab1fb6a453", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-ai-6f42e0fe-2026-06-29", "title": "稀土、铜铝与关键材料正在成为 AI 与新能源共用的供给约束：动态复盘", "subtitle": "赛道：关键材料/稀土铜铝。候选观察标的：600111.SH, 000831.SZ, MP, FCX, SCCO, AA, TECK。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 工业供给瓶颈。该主题应作为可…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-6f42e0fe6e/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-20", "06-21", "06-22", "06-23", "06-24", "06-25", "06-26", "06-27", "06-28", "06-29" ], "values": [ 30, 19.4, 32.8, 16.2, 49.6, 28, 32.4, 42.8, 35.2, 36.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "131", "accent": "red" }, { "label": "分析师", "value": "8", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：关键材料/稀土铜铝。候选观察标的：600111.SH, 000831.SZ, MP, FCX, SCCO, AA, TECK。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 工业供给瓶颈。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: 能源与大宗商品资产在宏观政策背离下的防御价值研判" }, { "title": "证据纪律", "text": "当前以 8 条支持、131 条风险和 21 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "8 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "131 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "Meta / GOES / AIDC", "detail": "Meta / GOES / AIDC / 新能源 / 能源 / 商品 / 钢铁 / Rare", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-20", "value": 30, "label": "风险", "title": "全国能源供给能覆盖AI用电总量，但2026-2027年的瓶颈仍是本地并网、特高压投运节奏与可靠容量定价。", "view": "截至2026-06-20，本卡片压力测试能源侧能否破解AI数据中心并网瓶颈。结论是：全国电量与新能源扩张足以覆盖AI负荷的总量增量，但特高压、储能和本地变电容量的投放节奏仍不足以完全解决2026-2027年大型集群的节点型并网约束。绿电比例提升本身不必然显著抬高算力成本，单独绿证按2026年4月价…", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-6f42e0fe6e/" }, { "date": "2026-06-21", "value": 19.4, "label": "风险", "title": "中国高牌号GOES、电磁线和非晶材料供应可缓释变压器瓶颈并利好部分A股材料供应商，但全球瓶颈仍由认证、关税、整机制造和测试能力共同决定。", "view": "截至2026-06-21，本卡支持“AI算力瓶颈转向电力硬件”的主线，但把材料端结论细化为：中国GOES和铜的总量足以缓解中国/亚洲及第三国变压器链条，真正约束在高牌号认证、关税路径、整机厂产线和测试台位。A股受益顺序偏向望变电气、宝钢股份、首钢股份、华菱钢铁、金杯电工、精达股份、云路股份；江西铜…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-6f42e0fe6e/" }, { "date": "2026-06-22", "value": 32.8, "label": "风险", "title": "AI能源转型下的煤炭与铜：作为公用事业防御策略中大宗商品的有效性分析", "view": "本卡压力测试 Research note 06 策略C 的 15% 商品敞口。结论：铜与煤炭方向相反，被错误打包。铜与标普相关性约0.6–0.8 [S8]，是 AI 周期的杠杆化 Beta 而非对冲——2008/2020 去杠杆中曾跌25%–60%，Research note 06 的 −8% 假…", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-6f42e0fe6e/" }, { "date": "2026-06-23", "value": 16.2, "label": "风险", "title": "修复大宗商品对冲层往返脆弱性的办法，是用与油价脱钩、对地缘正凸性的国防/实体经济工业敞口替换其部分现货能源beta——作为叠加而非整体替换，使K型/杠铃在能源升级…", "view": "综合卡片01–06：体制判断（美联储看跌期权失效、霍尔木兹为扳机、K型/杠铃为防御解）成立，但卡片06揭示大宗商品层在能源冲击缓解时遭beta拖累、随Brent（$77.34/桶）往返回吐。我的工业制造判断：用国防/航空航天主承包商加电网电气化机械替换部分现货能源beta——其收入由多年度拨款支撑…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-6f42e0fe6e/" }, { "date": "2026-06-24", "value": 49.6, "label": "支持", "title": "订单与招标数据支持资金转向电网设备及新能源实物资产端，但尚未证明二级市场资金已完成全链条重配。", "view": "截至2026-06-25，本卡支持“资金正从碳资产叙事转向新型电力系统实物资产”的方向，但限定为订单端确认。碳和绿证仍活跃；增量证据在国网特高压招标、电网设备公司中标、光伏组件、风电整机和储能采招。结论是：实物资产资本开支已明显落单，尤其是电网一次设备和特高压材料；但权益资金流、利润率和二级市场持…", "check": "支持 1 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-6f42e0fe6e/" }, { "date": "2026-06-25", "value": 28, "label": "风险", "title": "电网设备原材料缺口对新能源项目建设成本与并网进度的传导效应评估", "view": "支持卡片06–08的逻辑链：AI电力瓶颈已上移至原材料，且GOES/高纯度铜缺口确实传导至新能源经济性，但传导非对称、主要走时间而非单价。直接价格通道有限——铜仅占光伏/陆上风电资本开支约3%~5%，即便铜较2020年+50%、GOES+100%，对LCOE直接抬升仅约+2%~+4% [自测算]。…", "check": "支持 0 / 风险 4 / 观察 1", "href": "/investor/theses/thesis-6f42e0fe6e/" }, { "date": "2026-06-26", "value": 32.4, "label": "风险", "title": "随着全球及中国变电设备产能在2027-2029年逐步释放，AI数据中心的电力设备瓶颈将从“无法获取设备”的二进制物理硬约束，降为“支付溢价与关税以获取交期”的资本…", "view": "本报告对2027-2029年AI电力设备瓶颈的缓解速度进行了深入分析。结论表明，随着西方重电巨头（如日立能源、GE Vernova）扩产产能释放，以及以特变电工、思源电气、金盘科技为代表的中国高效率设备出口，2027-2029年电力设备将从物理上的“硬约束”转变为“成本/经济约束”。由于取向电工钢…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-6f42e0fe6e/" }, { "date": "2026-06-27", "value": 42.8, "label": "风险", "title": "虽然备忘录缓解了短期实物原油压力，但霍尔木兹海峡的结构性脆弱与15%-25%的二次通胀概率支撑将大宗商品战术超配至20%，并应采用高分红个股现货替代及低配看涨期权…", "view": "截至2026-06-27，评估表明尽管扫雷进展与伊朗出口正常化将推动短期WTI均价回落至67.95美元且跌破70美元概率升至62.85%，但15-25%的能源价格二次上尾再脉冲风险及长期地缘瓶颈支持将大宗商品配置超配至20%。在A股通道中，因159945等ETF极度缺乏流动性（成交仅106万元），…", "check": "支持 0 / 风险 2 / 观察 1", "href": "/investor/theses/thesis-6f42e0fe6e/" }, { "date": "2026-06-28", "value": 35.2, "label": "风险", "title": "在高利率长期化但衰退未确认的环境下，应以套息、凸性与实物资产对冲（短久期杠铃、适度美元、能源期权+黄金+能源股）替代方向性利率与油价押注，以改善组合风险回报比。", "view": "截至2026-06-29，支持卡片01的高利率长期化、衰退未确认判断，并转化为配置：以套息+凸性+实物资产对冲替代方向性押注。久期保持低于基准，杠铃式收割2年期套息（约4.10%、久期约1.9优于10年4.38%、久期约8.2）并保留小段长端作衰退保险；相对低收益货币适度做多美元（DXY约101.…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-6f42e0fe6e/" }, { "date": "2026-06-29", "value": 36.6, "label": "风险", "title": "[auto-handoff] 先进制程星载 OBP 芯片良率瓶颈与算力/通信设备材料供应链", "view": "致 材料行业分析师： 协同任务已完成。执行方为 TMT行业分析师。 模式: adhoc 会话: 81384248-8464-45e1-8194-06f6bed14dff 结果摘要: Wrote handoff_response.md (10343 bytes); wrote handoff_res…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-6f42e0fe6e/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "稀土、铜铝与关键材料正在成为 AI 与新能源共用的供给约束" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 8 位分析师和 5 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 131 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-6f42e0fe6e/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-6f42e0fe6e", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-a-70daf872-2026-06-29", "title": "A 股红利与硬科技杠铃仍是政策、利率和风险偏好的组合表达：动态复盘", "subtitle": "赛道：A股红利/硬科技杠铃。候选观察标的：510880.SH, 601398.SH, 600900.SH, 601088.SH, 600941.SH, 588000.SH, 688981.SH。当前共有 40 条证据、2 位分析师和 7 …", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-70daf87262/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-19", "06-20", "06-21", "06-22", "06-23", "06-24", "06-26", "06-27", "06-28", "06-29" ], "values": [ 27, 37.4, 12, 48.2, 17.6, 43, 12, 12, 28.2, 47.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "129", "accent": "red" }, { "label": "分析师", "value": "15", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：A股红利/硬科技杠铃。候选观察标的：510880.SH, 601398.SH, 600900.SH, 601088.SH, 600941.SH, 588000.SH, 688981.SH。当前共有 40 条证据、2 位分析师和 7 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: AI电力负载与电网物理瓶颈（变压器交期120-210周）构成新能源装机速度限速阀，双重挤压了缺乏定价权的光伏/电池制造龙头利润率，但结构性强化了拥有极强出海定价权的变压器与电网输配电龙头盈利景气度，A股资产配置应向电网出海及水电红利哑铃端重构[自测算]。" }, { "title": "证据纪律", "text": "当前以 10 条支持、129 条风险和 21 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "10 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "129 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "科创50 / A股 / CPI", "detail": "科创50 / A股 / CPI / 融资余额 / 长江电力 / 公用事业 / 中芯国际 / 运营商", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-19", "value": 27, "label": "风险", "title": "A股红利杠铃可以继续作为TMT组合防御对冲，但必须把中际旭创式进攻卫星视为强平凸性来源，而不是流动性缓冲。", "view": "截至2026-06-20，本卡对A股红利与TMT进攻端做微观流动性压力测试。结论是：Research note 05的A股红利杠铃方向仍可支持，但执行容量必须收紧。长江电力更适合作为防御锚，当前可得融资余额占自由流通市值代理值约4.52%，但压力可执行流动性下退出可能拉长至约11.5个交易日。中际…", "check": "支持 0 / 风险 3 / 观察 1", "href": "/investor/theses/thesis-70daf87262/" }, { "date": "2026-06-20", "value": 37.4, "label": "风险", "title": "AI电力瓶颈与绿电利用率分化对A股电力设备、公用事业和新能源运营商的策略影响", "view": "承接Research note 02–08全链条，本卡以发改能源〔2026〕688号多用户绿电直连（2026-05-20，优先支持算力设施）为基准情形，重构A股电力链定价。核心：688号把AI负荷从挤压新能源并网的拥塞负荷，翻转为在资源地变现弃电的锚定承购（2026Q1光伏利用率90.6%、隐含弃…", "check": "支持 0 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-70daf87262/" }, { "date": "2026-06-21", "value": 12, "label": "风险", "title": "AI电力瓶颈下的跨板块配置：半导体、AI服务器、电网设备、绿电与红利公用事业", "view": "综合卡片01-04，将AI电力瓶颈定性为局部取电速度而非全国电源缺口，由此给出A股再配置：把权重从定价完美、拥挤的算力栈顶部（前道半导体98.7分位PE、纯AI服务器贝塔）旋转至压缩‘通电机柜时间’的物理交付使能者。超配电网设备（国网十五五4万亿、+40%，最高确信）、液冷（渗透14%→40%，战…", "check": "支持 0 / 风险 6 / 观察 0", "href": "/investor/theses/thesis-70daf87262/" }, { "date": "2026-06-22", "value": 48.2, "label": "风险", "title": "在公用事业超级资本开支面临股权稀释及大选年监管时滞的双重制约下，跨资产配置应调降拥挤的美股公用事业权益，通过公用事业IG高级债捕获拓宽的信用利差，同时布局低相关的…", "view": "基于2026-06-22，本报告综合研判了受监管公用事业费率基数超级周期下的跨资产布局。公用事业1.3万亿美元资本开支带来严重的股权二次发行稀释，且大选年36州州长竞选引发监管机构费率抵制，惠誉调降行业展望至“恶化”。PJM 2027/28容量价暴涨至333.44美元/MW-日，零售电费同比上涨1…", "check": "支持 1 / 风险 2 / 观察 2", "href": "/investor/theses/thesis-70daf87262/" }, { "date": "2026-06-23", "value": 17.6, "label": "风险", "title": "信用风暴的核心不是现金总量不足，而是中小银行退出低评级城投买盘后，非银必须以更高利差、更低杠杆和更高流动性缓冲重新定价地方信用。", "view": "截至2026-06-23，本卡支持前三张卡片的压力方向，但判断外溢路径更像“流动性质量分层”而非系统性资金断裂。2000亿元AA级及以下城投买盘缺口为未公开验证的压力输入；相对2026年5月末11.67万亿元债基和16.27万亿元货基规模，总量可吸收，但边际买方撤出会触发弱AA利差、非银净值和回购…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-70daf87262/" }, { "date": "2026-06-24", "value": 43, "label": "风险", "title": "绿电分时匹配和CEA有偿比例提高能把可调节清洁电力转化为可测算的盈利与估值溢价，但该溢价只能提供相对对冲，无法单独抵御A股成长股系统性融资去杠杆。", "view": "截至2026-06-25，本卡片压力测试绿电分时匹配与碳配额收紧下A股公用事业稀缺资产的估值弹性。结论是：可调节水电和具备储能、绿电直连或数据中心PPA的绿电运营商确有基本面溢价，基准测算约27元/MWh，可带来约18.2%的盈利弹性；若叠加ERP收缩，公允价值可提升约35.9%。但在A股成长股融…", "check": "支持 0 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-70daf87262/" }, { "date": "2026-06-26", "value": 12, "label": "风险", "title": "2026-06-26确认科技两融堰塞湖进入去杠杆阶段，但未确证牛市陷阱；最优策略是降拥挤科技、保现金缓冲、分批增配电力电网等物理红利资产。", "view": "截至2026-06-27，结论是支持防守再平衡，但不把2026-06-26下跌直接视为“国家队牛市陷阱”的确证。A股两融余额已升至3万亿元以上，电子行业两融余额超过6000亿元且占比约两成，说明风险是融资拥挤和流动性冲击，而非科技周期被证伪。建议科技从超配降至标配，优先削减融资敏感电子/CPO仓位…", "check": "支持 0 / 风险 8 / 观察 0", "href": "/investor/theses/thesis-70daf87262/" }, { "date": "2026-06-27", "value": 12, "label": "风险", "title": "在去杠杆与中美利差倒挂硬约束下，A股配置应坚决重构为以大水电及贬值增益能源央企为防守侧、以出口高景气硬科技为进攻侧的真杠铃组合，规避面临两融平仓螺旋与三季度解禁洪…", "view": "本报告将跨资产去杠杆及红利能源超配倾向转译为A股“真杠铃”风格轮动策略（60%物理红利/上游能源，20%质量成长，20%对冲/现金）。定量测算了电子板块两融在科创50跌至1,520点时的强平压力（2,528.60亿元），指出了三季度英维克、思源电气等高位股限售解禁供给压力，并划定了长江电力股息率跌…", "check": "支持 0 / 风险 7 / 观察 0", "href": "/investor/theses/thesis-70daf87262/" }, { "date": "2026-06-28", "value": 28.2, "label": "风险", "title": "局部中小银行信用缺口下的A股银行红利耐久性与高股息内部轮动", "view": "卡08分离出的“额度沉睡”机制（CET1上限+NPL触发+低于1.40%净息差）不仅是中小银行信用风险，更是撕裂A股红利复合体的力量。重排结论：①大行红利居顶——分红不依赖沉睡额度，约4.35%股息率、对10年国债（约1.72%）257bp利差、0.5–0.7倍PB已计入坏消息，持有/超配；②公用…", "check": "支持 1 / 风险 6 / 观察 1", "href": "/investor/theses/thesis-70daf87262/" }, { "date": "2026-06-29", "value": 47.6, "label": "观察", "title": "红利龙头去拥挤与超跌成长板块反弹的风格切换大论战", "view": "红利龙头去拥挤与超跌成长板块反弹的风格切换大论战 A-Share Style Rotation: Dividend De-crowding vs. Growth Rebound 长江电力（600900.SH）资金大流出与药明/恒瑞大涨，是否标志着A股向超跌成长风格的实质性切换已在午间确立？伟星股份…", "check": "支持 1 / 风险 4 / 观察 3", "href": "/investor/theses/thesis-70daf87262/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "A 股红利与硬科技杠铃仍是政策、利率和风险偏好的组合表达" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 15 位分析师和 4 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 129 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-70daf87262/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-70daf87262", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-58200390b4-58200390-2026-06-29", "title": "加密资产仍是全球流动性和散户风险偏好的高频回声：动态复盘", "subtitle": "赛道：加密/高波动风险偏好。候选观察标的：COIN, MSTR, MARA, RIOT, IBIT, HOOD。当前共有 40 条证据、7 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-58200390b4/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-20", "06-21", "06-22", "06-23", "06-24", "06-25", "06-26", "06-27", "06-28", "06-29" ], "values": [ 30, 22.4, 29.8, 22.2, 23.6, 12, 44.4, 15.8, 12, 12 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "136", "accent": "red" }, { "label": "分析师", "value": "13", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：加密/高波动风险偏好。候选观察标的：COIN, MSTR, MARA, RIOT, IBIT, HOOD。当前共有 40 条证据、7 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型" }, { "title": "证据纪律", "text": "当前以 7 条支持、136 条风险和 17 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "7 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "136 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "HBM / GPU / AIDC", "detail": "HBM / GPU / AIDC / Crypto / remains / high-frequency / echo / global", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-20", "value": 30, "label": "风险", "title": "虽然HBM国产化良率崩塌会将集群 CapEx 推升 98%，但中国互联网巨头通过异构分级部署与算法优化（缩减30%GPU需求）并结合西部绿电直连套利（LCOE降至…", "view": "评估HBM成本压力对AI集群TCO及扩容策略的传导。100%国产材料导致HBM3封装良率崩塌至4.11%，推升10000颗GPU集群CapEx至8.17亿美元，东部煤电下3年TCO达10.17亿美元（+66.13%）。建议协同优化：架构上采取60% HBM3与40% HBM2E的分级部署，结合算法…", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-58200390b4/" }, { "date": "2026-06-21", "value": 22.4, "label": "风险", "title": "Hyperscaler资本开支正从算力优先转向电力协同：总量仍大，但边际一美元先购买电力、进度和并网确定性，再购买下一组加速器。", "view": "截至2026-06-21，本卡压力测试认为：Hyperscaler资本开支不会因电力瓶颈在2026-2027年崩塌，但边际规则已从“买更多GPU”转向“按确定日期拿到可通电MW”。核电购电、表后自发电、并网权、长交期电气设备和区域边缘部署正在获得更高资本权重。核电是长期稀缺性对冲，不是近端解决方案…", "check": "支持 0 / 风险 4 / 观察 1", "href": "/investor/theses/thesis-58200390b4/" }, { "date": "2026-06-22", "value": 29.8, "label": "观察", "title": "扩大版出口管制导致材料与EDA工具国产化面临良率黑洞与算力TCO暴增，实质性切断了在华先进封装的商业化增量逻辑。", "view": "截至2026-06-22，评估表明若BIS将管制扩大到先进材料（如ABF载板、MUF、NCF）及EDA工具，将实质切断在华先进封装产能的商业化增量逻辑。100%国产化材料导致8-Hi HBM堆叠良率崩塌至4.11%，产出缩水68.48%；GPU BOM成本翻倍至$37,746，售价涨至$75,49…", "check": "支持 0 / 风险 3 / 观察 1", "href": "/investor/theses/thesis-58200390b4/" }, { "date": "2026-06-23", "value": 22.2, "label": "风险", "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "view": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-58200390b4/" }, { "date": "2026-06-24", "value": 23.6, "label": "风险", "title": "中国CSP成本上升背景下，AI算力基础设施供应弹性与扩容压力分析", "view": "压力测试部分否定线程以电力为中心的框架。AI算力系统是供给绑定而非需求绑定：2026年超大规模厂商资本开支同比+77%至约7250亿美元[S1]，英伟达Blackwell售罄至年中、积压360万颗[S2]，因此约束的是配额优先级而非需求。核心锚定集群凭借锁定电力、GPU配额与表后自发电被隔离；放缓…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-58200390b4/" }, { "date": "2026-06-25", "value": 12, "label": "风险", "title": "截至2026-06-25，AI硬件需求的下一核心约束不是芯片是否能生产，而是足够多的兆瓦能否被许可、接入、冷却并按期商业化。", "view": "截至2026-06-25，本卡支持上一张卡的AI硬件稀缺性主线，但将瓶颈从GPU/HBM进一步上移到可通电数据中心、电网接入、冷却和电气设备。IEA、Berkeley Lab、EPRI、CBRE和JLL数据共同显示，AI需求仍强，但项目能否转化为创收产能取决于电力和并网速度。结论是：稀缺性继续支撑…", "check": "支持 1 / 风险 9 / 观察 0", "href": "/investor/theses/thesis-58200390b4/" }, { "date": "2026-06-26", "value": 44.4, "label": "风险", "title": "1.6T光模块交付延迟对AI算力集群及超大规模云厂商Capex部署进度的影响", "view": "支持卡片01/02的研究线并加以边界化延伸：200G EML 的 HTOL 风险确会让 1.6T GA 滑向 2027H1，但这是一道带宽前沿与单位经济性约束，而非整体 Capex 停摆。GB300 NVL72 的 scale-up 走铜缆 NVLink、scale-out 走 ConnectX-…", "check": "支持 0 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-58200390b4/" }, { "date": "2026-06-27", "value": 15.8, "label": "风险", "title": "AI 算力资本开支对国产半导体订单可见度的支撑强度", "view": "压力测试结论：AI资本开支与电力瓶颈的传导真实但不对称，对2026年净支撑。硬约束是电力而非资本或芯片——四巨头2026资本开支约7250亿美元/+77%[S1]，微软约1900亿、因电力受限至2026底（Azure约800亿积压、GPU闲置）[S5]；2028美国约49GW缺口、变压器4-5年交…", "check": "支持 0 / 风险 6 / 观察 0", "href": "/investor/theses/thesis-58200390b4/" }, { "date": "2026-06-28", "value": 12, "label": "风险", "title": "效率驱动的算力架构转型：从“暴力规模”到“高能效密度”的物理基座适配性评估", "view": "压力测试 Research note 08 的隐含前提：效率并不缓解物理天花板，而是把它从电网层下移到机架与机柜行层并加剧。机理是机架尺度的杰文斯悖论——单瓦产出靠提高密度实现，机架功率从约40kW(Hopper)→120kW(Blackwell)→142kW(GB300,90%入液)→600kW…", "check": "支持 1 / 风险 13 / 观察 0", "href": "/investor/theses/thesis-58200390b4/" }, { "date": "2026-06-29", "value": 12, "label": "风险", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投…", "view": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "check": "支持 0 / 风险 9 / 观察 1", "href": "/investor/theses/thesis-58200390b4/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "加密资产仍是全球流动性和散户风险偏好的高频回声" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 13 位分析师和 5 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 136 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-58200390b4/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-58200390b4", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-8b8ede3135-8b8ede31-2026-06-29", "title": "消费降级、健康刚需与银发经济正在重塑防御型增长：动态复盘", "subtitle": "赛道：消费降级/健康银发。候选观察标的：600519.SH, 2319.HK, 3690.HK, 300760.SZ, 600276.SH, 1177.HK, UNH。当前共有 40 条证据、1 位分析师和 5 条风险信号指向该判断；主导…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-8b8ede3135/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-19", "06-20", "06-21", "06-22", "06-23", "06-24", "06-26", "06-27", "06-28", "06-29" ], "values": [ 41, 37.4, 12, 31.2, 29.6, 31, 41.4, 45.8, 41.2, 30.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "115", "accent": "red" }, { "label": "分析师", "value": "19", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：消费降级/健康银发。候选观察标的：600519.SH, 2319.HK, 3690.HK, 300760.SZ, 600276.SH, 1177.HK, UNH。当前共有 40 条证据、1 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: 产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）" }, { "title": "证据纪律", "text": "当前以 19 条支持、115 条风险和 26 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "19 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "115 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "CPI / 创新药 / 食品", "detail": "CPI / 创新药 / 食品 / 设备 / Consumer / downgrade / healthcare / necessities", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-19", "value": 41, "label": "支持", "title": "饲料粮通缩对CPI食品分项是再通缩力量而非通胀缓冲，政策响应将是‘托麦不托玉米’组合，对总CPI同比拖累约−0.08至−0.16pct。", "view": "芽麦替代封顶玉米 + 巴西大豆到港洪峰构成饲料粮双料通缩。直接传导（粮食、食用油）对CPI拖累不足0.04pct；主航道是间接传导——料价下行不会推高猪价，反而通过压栏增重与延迟去产能强化猪肉再通缩，预计2026年6–9月对总CPI同比拖累−0.08至−0.16pct[自测算]。政策端是‘托麦不托…", "check": "支持 1 / 风险 2 / 观察 1", "href": "/investor/theses/thesis-8b8ede3135/" }, { "date": "2026-06-20", "value": 37.4, "label": "风险", "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "view": "大健康产业链 30 日激活率仅 5.4%（15/280），属被忽视而非基本面差。2026 年 Q1 创新药对外授权破 600 亿美元（约占 2025 全年 1357 亿的 44%）[自测算]，药明康德 Q1 营收 124.36 亿（+28.81%）、在手订单 597.7 亿、小分子 CDMO+80…", "check": "支持 0 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-8b8ede3135/" }, { "date": "2026-06-21", "value": 12, "label": "风险", "title": "产业链调研：大健康产业链（30 日激活率 5.4%, 15/280）", "view": "大健康产业链低覆盖（30日激活5.4%），但基本面已反转。三引擎齐发：创新药license-out成全球定价事件（2025年约1357亿美元、占全球49%，2026Q1超600亿美元）；CXO订单重新加速（药明Q1营收+28.8%、在手订单597.7亿；凯莱英新兴业务+74.1%）；设备更新（20…", "check": "支持 0 / 风险 9 / 观察 1", "href": "/investor/theses/thesis-8b8ede3135/" }, { "date": "2026-06-22", "value": 31.2, "label": "风险", "title": "BIOSECURE Act 对中国创新药资产全球化估值逻辑的冲击评估", "view": "BIOSECURE Act 于 2026-06-23 视角下已于 2025-12-18 随 FY2026 NDAA 签署生效，但其限制对象是联邦采购/拨款中涉及 BCC 的‘生物技术设备或服务’，不触及私营对外授权/BD 与成品药。Kelun-Biotech 经 Merck/MSD（大中华区以外授…", "check": "支持 0 / 风险 3 / 观察 1", "href": "/investor/theses/thesis-8b8ede3135/" }, { "date": "2026-06-23", "value": 29.6, "label": "风险", "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "view": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-8b8ede3135/" }, { "date": "2026-06-24", "value": 31, "label": "风险", "title": "铜及电网设备成本的结构性上涨将通过公用事业电价调整和工业投入品渠道转嫁给终端，抬高全球结构性通胀底座，迫使主要央行维持更高的限制性利率并加速日本央行货币正常化。", "view": "截至2026-06-25，本卡评估了铜及电网成本压力向全球货币政策的传导。电网升级与AI算力用铜强度大，推高设备售价与监管资产基底，预测在3-5年内将零售电价推升1.5%至2.5%。50%铜价冲击将累积拉动全球CPI约0.25%至0.45%。粘性通胀使美联储年底前维持3.50%-3.75%高利率，…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-8b8ede3135/" }, { "date": "2026-06-26", "value": 41.4, "label": "风险", "title": "压力测试支撑整套对冲架构的55-65%国内通缩基准情形概率及其政策翻转触发点", "view": "压力测试整套对冲架构所依赖的55-65%国内通缩基准概率。结论：基准情形近期稳健，保留但下调至55-60%下沿。硬数据印证消费端通缩：2026年5月CPI同比+1.2%、食品-1.7%、核心+1.1%、环比-0.1%；10年中债贴地板于1.73%[S1][S5]。PBoC“边走边看”渐进（LPR连…", "check": "支持 0 / 风险 2 / 观察 1", "href": "/investor/theses/thesis-8b8ede3135/" }, { "date": "2026-06-27", "value": 45.8, "label": "风险", "title": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242）", "view": "产业链调研：食品加工产业链（30 日激活率 7.9%, 19/242） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "check": "支持 0 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-8b8ede3135/" }, { "date": "2026-06-28", "value": 41.2, "label": "风险", "title": "去风险三期医药资产通过rNPV模型释放3.48亿美元的期权价值构成Biotech估值底，而生命科学工具在5.5x EV/Sales筑底、CXO在 overseas…", "view": "截至2026-06-28，本卡片基于BD估值方法构建了下一批并购标的与估值底线筛选框架。AbbVie 109亿美元现金收购Apogee与Merck KGaA 113亿美元收购Bio-Techne为医药生物确立了质量门控底座。我们量化厘定：1) 创新药三期去风险资产的rNPV floor：以12亿美…", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-8b8ede3135/" }, { "date": "2026-06-29", "value": 30.6, "label": "风险", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "view": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-8b8ede3135/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "消费降级、健康刚需与银发经济正在重塑防御型增长" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 19 位分析师和 3 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 115 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-8b8ede3135/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-8b8ede3135", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-em-beta-e9630d3e-2026-06-29", "title": "日元套息平仓是 EM 科技、离岸中资与高 beta 主题的尾部流动性风险：动态复盘", "subtitle": "赛道：日元套息/EM 科技流动性。候选观察标的：FXY, UUP, EEM, MCHI, KWEB, EWJ, TSM。当前共有 40 条证据、5 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-e9630d3ee1/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-20", "06-21", "06-22", "06-23", "06-24", "06-25", "06-26", "06-27", "06-28", "06-29" ], "values": [ 12, 12, 12, 22.2, 41.6, 48, 20.4, 21.8, 29.2, 42.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "131", "accent": "red" }, { "label": "分析师", "value": "15", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：日元套息/EM 科技流动性。候选观察标的：FXY, UUP, EEM, MCHI, KWEB, EWJ, TSM。当前共有 40 条证据、5 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: 物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进封装国产替代材料集中。" }, { "title": "证据纪律", "text": "当前以 11 条支持、131 条风险和 18 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "11 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "131 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "HBM / GPU / ASIC", "detail": "HBM / GPU / ASIC / CPI / 美元 / TSMC / AIDC / CNH", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-20", "value": 12, "label": "风险", "title": "人民币走弱对AI算力资本开支的主要影响不是需求崩塌，而是把进口GPU采购从工程扩张问题改造成受外汇、许可证和国产替代共同约束的财务配置问题。", "view": "工作日期2026-06-21。本卡压力测试前序CNH逻辑：人民币波动会抬高美系GPU进口预算，但更可能导致预算重排、分批采购、对冲纪律强化和异构集群替代，而非AI算力资本开支突然停止。以2026-06-19 USD/CNH 6.78440和160亿美元H20订单规模测算，CNH贬值1%/3%/5%…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-e9630d3ee1/" }, { "date": "2026-06-21", "value": 12, "label": "观察", "title": "I have successfully completed the stress test analysis on the transmission channe…", "view": "I have successfully completed the stress test analysis on the transmission channels of the global dollar liquidity squeeze and the JPY carry trade un…", "check": "支持 0 / 风险 8 / 观察 2", "href": "/investor/theses/thesis-e9630d3ee1/" }, { "date": "2026-06-22", "value": 12, "label": "风险", "title": "可信算力区会把稀缺溢价从“谁有GPU”迁移到“谁能把获批GPU、HBM、封装与fabric系统交付到可信地点”。", "view": "截至2026-06-22，本卡综合前序“可信算力区”逻辑：半导体分配从单纯GPU供需转向“GPU-HBM-先进封装-网络fabric-合规运营”整套系统的可获批交付。最受益者是NVIDIA全栈机架、已认证HBM供应商、TSMC CoWoS/SoIC生态、AI网络与定制ASIC供应商；风险集中在中国…", "check": "支持 0 / 风险 6 / 观察 1", "href": "/investor/theses/thesis-e9630d3ee1/" }, { "date": "2026-06-23", "value": 22.2, "label": "风险", "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "view": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-e9630d3ee1/" }, { "date": "2026-06-24", "value": 41.6, "label": "风险", "title": "[auto-handoff] 日元干预警报与亚太半导体/CPO板块去杠杆风险评估", "view": "致 市场情绪师： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: c956eaa1-0683-41fa-9c41-551222934843 结果摘要: Wrote handoff_response.md (10735 bytes); wrote handoff_response…", "check": "支持 0 / 风险 2 / 观察 2", "href": "/investor/theses/thesis-e9630d3ee1/" }, { "date": "2026-06-25", "value": 48, "label": "观察", "title": "[auto-handoff] 美联储主席Warsh政策首秀推升美元至100.814情景下日元与离岸人民币套利交易无序平仓溢出测算", "view": "致 首席风控官： 协同任务已完成。执行方为 外汇策略师。 模式: adhoc 会话: 82711dfa-2690-4d72-9be3-a760c489f1cc 结果摘要: Wrote handoff_response.md (12305 bytes); wrote handoff_response…", "check": "支持 1 / 风险 3 / 观察 3", "href": "/investor/theses/thesis-e9630d3ee1/" }, { "date": "2026-06-26", "value": 20.4, "label": "风险", "title": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径", "view": "全球Carry Trade平仓下港股/A股半导体龙头流动性虹吸溢出路径 Global Carry Unwind and Liquidity Siphon Risks for HK/A-Share Semi Leaders 日本央行升息及美元兑日元突破161.7，全球Carry Trade平仓是否会…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-e9630d3ee1/" }, { "date": "2026-06-27", "value": 21.8, "label": "风险", "title": "日元套利交易平仓与离岸CNH流动性收紧对A股防御重估的交叉传导风险", "view": "本卡以global-macro压力测试card-08的“USD/JPY与CNH流动性熔断器”构想：支持其核心方向（知识库已确认应以二者作量化触发器），但否定朴素的“单变量、按价位、OR门”实现。四点修正：(1)冲击分两波——先美元流动性挤兑、后避险重估，价位触发会迫使在第一波真空里抛售防御股；(2…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-e9630d3ee1/" }, { "date": "2026-06-28", "value": 29.2, "label": "风险", "title": "中美利差倒挂对离岸人民币（CNH）汇率及跨境资本流动的传导效应", "view": "支持卡片01的格局判断并补齐传导环节。截至2026-06-29，美国10年期国债约4.40%、中国10年期约1.79%，名义利差倒挂约−261bps（本轮最阔）；而真实利差约+39bps偏向中国。名义渠道在去杠杆中占主导，使CNH成为融资货币。鹰派美联储驱动的套利去杠杆经两条腿（CNH融资腿+全球…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-e9630d3ee1/" }, { "date": "2026-06-29", "value": 42.6, "label": "风险", "title": "日元套利平仓通过离岸人民币流动性抽水及掉期清算挤压双重渠道，系统性抬高离岸融资成本并技术性击穿USD/CNY 7.00关口，导致人民币套利空间彻底崩溃。", "view": "本卡支持Research note 02，定量分析日元套利平仓对CNH的传导。BOJ加息和USD/JPY破位触发约349亿美元平仓，占离岸池近25%，推高隔夜HIBOR至10.81%-12.74%。掉期展期率跌至40%引发即期购汇压力，击穿USD/CNY 7.00。CNH套利Carry-to-Ri…", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-e9630d3ee1/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "日元套息平仓是 EM 科技、离岸中资与高 beta 主题的尾部流动性风险" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 15 位分析师和 4 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 131 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-e9630d3ee1/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-e9630d3ee1", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-etf-89c14497-2026-06-29", "title": "期权伽马、ETF 与尾盘流动性正在放大主题交易的微观结构风险：动态复盘", "subtitle": "赛道：期权伽马/市场微观结构。候选观察标的：CBOE, CME, VIRT, IBKR, HOOD, VIXY。当前共有 40 条证据、6 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-89c14497ac/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-19", "06-21", "06-22", "06-23", "06-24", "06-25", "06-26", "06-27", "06-28", "06-29" ], "values": [ 21, 12, 12, 16.2, 23.6, 43, 14.4, 36.8, 32.2, 12.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "124", "accent": "red" }, { "label": "分析师", "value": "17", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：期权伽马/市场微观结构。候选观察标的：CBOE, CME, VIRT, IBKR, HOOD, VIXY。当前共有 40 条证据、6 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: 产业链调研：新华财经-海洋经济产业链（30 日激活率 6.1%, 9/148）" }, { "title": "证据纪律", "text": "当前以 0 条支持、124 条风险和 36 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "0 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "124 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "Options / gamma / ETFs", "detail": "Options / gamma / ETFs / closing / liquidity / amplifying / microstructure / risk", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-19", "value": 21, "label": "观察", "title": "I have performed a rigorous quantitative stress test using Black Scholes (for the…", "view": "I have performed a rigorous quantitative stress test using Black Scholes (for the XBI equity leg) and Black 76 (for the 5 Year Treasury futures leg) …", "check": "支持 0 / 风险 5 / 观察 3", "href": "/investor/theses/thesis-89c14497ac/" }, { "date": "2026-06-21", "value": 12, "label": "观察", "title": "I have completed the synthesis of the trade surplus breakdown trigger and its imp…", "view": "I have completed the synthesis of the trade surplus breakdown trigger and its implications for the global macro liquidity transmission chain. Summary…", "check": "支持 0 / 风险 10 / 观察 2", "href": "/investor/theses/thesis-89c14497ac/" }, { "date": "2026-06-22", "value": 12, "label": "风险", "title": "Summary of Work We have evaluated the mismatch risk between rising GPU power cons…", "view": "Summary of Work We have evaluated the mismatch risk between rising GPU power consumption and grid capacity bottlenecks, demonstrating how this constr…", "check": "支持 0 / 风险 7 / 观察 0", "href": "/investor/theses/thesis-89c14497ac/" }, { "date": "2026-06-23", "value": 16.2, "label": "风险", "title": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242）", "view": "产业链调研：食品加工产业链（30 日激活率 7.4%, 18/242） 4%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-89c14497ac/" }, { "date": "2026-06-24", "value": 23.6, "label": "风险", "title": "I have completed the synthesis for Research note 06, analyzing the transmission o…", "view": "I have completed the synthesis for Research note 06, analyzing the transmission of supply side compute constraints (power delays, hardware inflation,…", "check": "支持 0 / 风险 6 / 观察 4", "href": "/investor/theses/thesis-89c14497ac/" }, { "date": "2026-06-25", "value": 43, "label": "风险", "title": "[auto-handoff] 杠杆交易拥挤度与两融突破3万亿风险评估", "view": "致 首席策略师： 协同任务已完成。执行方为 市场情绪师。 模式: adhoc 会话: 55cdff25-1527-4b69-9408-a3e16d4c97ce 结果摘要: Wrote handoff_response.md (14618 bytes); wrote handoff_response…", "check": "支持 0 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-89c14497ac/" }, { "date": "2026-06-26", "value": 14.4, "label": "风险", "title": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174）", "view": "产业链调研：生命健康和生物科技（30 日激活率 6.9%, 12/174） 9%) over last 30d — under-researched. Cite specific tickers + recent data where possible.", "check": "支持 0 / 风险 6 / 观察 0", "href": "/investor/theses/thesis-89c14497ac/" }, { "date": "2026-06-27", "value": 36.8, "label": "观察", "title": "I have completed the quantitative stress test on the STAR 50 ETF options collar s…", "view": "I have completed the quantitative stress test on the STAR 50 ETF options collar strategy (\"Buy 95% Put / Sell 85% Put / Sell 105% Call\" [S1]) designe…", "check": "支持 0 / 风险 3 / 观察 1", "href": "/investor/theses/thesis-89c14497ac/" }, { "date": "2026-06-28", "value": 32.2, "label": "观察", "title": "[auto-handoff] Cloud customer acceptance of higher memory prices", "view": "致 主题研究员： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: bd9e450a-eea0-4ad7-8a7c-3e07838a29ab 结果摘要: Wrote handoff_response.md (7215 bytes); wrote handoff_response.…", "check": "支持 0 / 风险 4 / 观察 1", "href": "/investor/theses/thesis-89c14497ac/" }, { "date": "2026-06-29", "value": 12.6, "label": "风险", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "view": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "check": "支持 0 / 风险 8 / 观察 2", "href": "/investor/theses/thesis-89c14497ac/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "期权伽马、ETF 与尾盘流动性正在放大主题交易的微观结构风险" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 17 位分析师和 3 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 124 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-89c14497ac/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-89c14497ac", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-ai-07501700-2026-06-29", "title": "AI 资本开支通过能源、设备和利率进入通胀定价：动态复盘", "subtitle": "赛道：AI 再通胀/利率敏感资产。候选观察标的：XLE, XLU, TLT, GLD, JPM, NEE。当前共有 40 条证据、3 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-075017003b/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-19", "06-20", "06-21", "06-22", "06-23", "06-24", "06-25", "06-26", "06-28", "06-29" ], "values": [ 20, 13.4, 12, 34.2, 26.6, 19, 20.4, 27.8, 13.2, 36.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "144", "accent": "red" }, { "label": "分析师", "value": "15", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：AI 再通胀/利率敏感资产。候选观察标的：XLE, XLU, TLT, GLD, JPM, NEE。当前共有 40 条证据、3 位分析师和 8 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: 变压器瓶颈不会摧毁电力红利防御逻辑，但会把它从“买所有公用事业”收窄为“买已送电基荷资产、可回收监管资产基数和瓶颈设备定价权”。" }, { "title": "证据纪律", "text": "当前以 4 条支持、144 条风险和 12 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "4 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "144 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "电网 / GOES / CPI", "detail": "电网 / GOES / CPI / PCE / 数据中心 / AI capex / AI数据中心 / 公用事业", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-19", "value": 20, "label": "风险", "title": "电网公用事业资产费率基数扩张对宏观通胀与货币政策的传导效应", "view": "压力测试卡片01–05的宏观结论。直接渠道被夸大：电力属能源、按构造排除于核心PCE之外，达拉斯联储测算数据中心对总体PCE的拉动2026仅+0.05pp、2030升至+0.13pp[S1]；电力仅占CPI约2.5%[S11]，自测算直接总体贡献约0.10pp。但费率基数是30–40年价格水平锁定…", "check": "支持 1 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-075017003b/" }, { "date": "2026-06-20", "value": 13.4, "label": "风险", "title": "电网设备紧缺不会终止最强Hyperscaler的AI capex，但会把'宣布投资'与'已送电算力'之间的时间差拉大，并迫使选址从效能最优转向电网弹性最优。", "view": "截至2026-06-21，本卡压力测试电网设备紧缺对AI算力中心扩建的传导。结论是：变压器backlog对设备商是定价权，对Hyperscaler是进度税。Wood Mackenzie显示电力变压器128周、GSU 143周、开关设备44周；相对18-24个月数据中心建设周期，若未提前锁定设备sl…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-075017003b/" }, { "date": "2026-06-21", "value": 12, "label": "风险", "title": "美欧贸易壁垒足以压低中国电力设备出海的利润率和估值倍数，但不足以抵消无源、短缺约束型电网硬件的出口红利。", "view": "截至2026-06-21，本卡否定“美欧关税和贸易限制足以抵消中国电力设备出海红利”的强判断。政策摩擦真实存在：美国Section 301覆盖部分HS 8504及能源相邻品类，欧盟通过NZIA、FSR和高风险逆变器融资限制提高准入门槛。但美国并网队列、变压器80-210周交期、DOE披露的进口依赖…", "check": "支持 0 / 风险 9 / 观察 0", "href": "/investor/theses/thesis-075017003b/" }, { "date": "2026-06-22", "value": 34.2, "label": "风险", "title": "能源价格冲击对通胀预期的传导风险", "view": "截至2026-06-22，本卡以能源视角支持Card-01「暂停+更久维持高位选择权」，但修正其传导机制。需区分两类冲击：(1)地缘油价飙升已回落——WTI从约$109崩跌约29.6%至$72.83，美伊重开霍尔木兹后布伦特回到约$80，且油价对核心通胀传导小而慢(峰值约0.12pp)；(2)真正…", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-075017003b/" }, { "date": "2026-06-23", "value": 26.6, "label": "风险", "title": "电网输变电容量与并网审批周期（而非总体发电量）已成为AI算力部署的终极物理刚性瓶颈，并已实质性挤压IT硬件采购预算。", "view": "本报告对AI算力扩张的电网与能源瓶颈进行压力测试。分析表明，虽然电力设备订单充足，但美国并网队列平均达5.5年，且面临FERC 2026年6月大型负荷监管及弗吉尼亚GS-5等防止成本转嫁政策。变压器等重电面临GOES原材料短缺，Tier-1厂商积压至2029-2030年。表后核电直供监管被封死，燃…", "check": "支持 0 / 风险 4 / 观察 1", "href": "/investor/theses/thesis-075017003b/" }, { "date": "2026-06-24", "value": 19, "label": "风险", "title": "AI算力投资需求规模与电网/能源供应缺口的动态匹配分析", "view": "截至2026-06-25，本卡量化AI算力需求通道与电网/能源物理供应瓶颈的动态匹配。需求侧：4家超大规模厂商2026年capex约7250亿美元(+77%)，美国数据中心峰值负荷2024约21-22GW、EPRI中性情景2030升至71GW。供应侧：并网积压约2600GW(中位约5年)、大变压器…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-075017003b/" }, { "date": "2026-06-25", "value": 20.4, "label": "风险", "title": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论", "view": "AI数据中心自备电力(BYOP)与气电PPA可行性讨论 Feasibility of Bring Your Own Power (BYOP) and Gas PPA for AI Data Centers 在小堆核电(SMR)2030年前难以落地、碳约束拉平公耗曲线的现实背景下，国内与海外AI数据…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-075017003b/" }, { "date": "2026-06-26", "value": 27.8, "label": "风险", "title": "随着全球及中国变电设备产能在2027-2029年逐步释放，AI数据中心的电力设备瓶颈将从“无法获取设备”的二进制物理硬约束，降为“支付溢价与关税以获取交期”的资本…", "view": "本报告对2027-2029年AI电力设备瓶颈的缓解速度进行了深入分析。结论表明，随着西方重电巨头（如日立能源、GE Vernova）扩产产能释放，以及以特变电工、思源电气、金盘科技为代表的中国高效率设备出口，2027-2029年电力设备将从物理上的“硬约束”转变为“成本/经济约束”。由于取向电工钢…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-075017003b/" }, { "date": "2026-06-28", "value": 13.2, "label": "风险", "title": "电网并网排队与GOES/变压器超长交期构成AI算力的硬物理墙，表后直供被否决堵死了绕过电网的捷径，电力约束正逼迫AI Capex从‘买芯片’转向‘锁电力’，并驱动…", "view": "截至2026-06-28，本卡片通过对电网与变压器交期的压力测试，确认电网并网（5-7年交期）和变压器交付（2-4年交期）构成了AI算力扩张的终极物理边界。随着GB300（约135kW）及Rubin Kyber（约600kW）对800VDC和固态变压器（SST）的高密度配电需求增加，目前仅有不足8…", "check": "支持 1 / 风险 8 / 观察 0", "href": "/investor/theses/thesis-075017003b/" }, { "date": "2026-06-29", "value": 36.6, "label": "风险", "title": "电力设备与铜/GOES结构性成本通胀对去通胀假设的反馈：红利轮动的利率前提是否被侵蚀", "view": "本卡压力测试Cards 5-6红利轮动的利率前提，得出分层裁定：水平稳健、久期被错配。1.72%国债收益率锚未被铜/GOES成本通胀逆转——中国10年期国债2026-06-26报1.73%[S1]，因为该通胀在中国体现为PPI(+3.9%)[S8]而非CPI(+1.2%、消费品-0.8%)[S7]…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-075017003b/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "AI 资本开支通过能源、设备和利率进入通胀定价" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 15 位分析师和 4 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 144 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-075017003b/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-075017003b", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-cpo-ai-42faeae6-2026-06-29", "title": "CPO、光互联与铜连接决定 AI 集群内部租金迁移：动态复盘", "subtitle": "赛道：CPO/光模块/高速互联。候选观察标的：AVGO, MRVL, ANET, COHR, LITE, 300308.SZ, 300394.SZ, 688498.SH。当前共有 40 条证据、1 位分析师和 8 条风险信号指向该判断；主…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-42faeae623/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-20", "06-21", "06-22", "06-23", "06-24", "06-25", "06-26", "06-27", "06-28", "06-29" ], "values": [ 12, 25.4, 12, 12, 12, 30, 44.4, 21.8, 35.2, 36.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "135", "accent": "red" }, { "label": "分析师", "value": "12", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：CPO/光模块/高速互联。候选观察标的：AVGO, MRVL, ANET, COHR, LITE, 300308.SZ, 300394.SZ, 688498.SH。当前共有 40 条证据、1 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: 物理电网配给瓶颈将强制封顶全球GPU出货规模并导致先进制程晶圆厂资本开支节奏被动下修，且低毛利的电网设备扩张无法抵消半导体供应链的利润塌陷，资金应结构性向极致能效的功率半导体（SiC）和先进封装国产替代材料集中。" }, { "title": "证据纪律", "text": "当前以 9 条支持、135 条风险和 16 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "9 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "135 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "AI 算力 / AI 基础设施 / 云厂商", "detail": "AI 算力 / AI 基础设施 / 云厂商 / GOES / HBM / CPO / 先进封装 / GPU", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-20", "value": 12, "label": "风险", "title": "AI硬件供应链波动与融资风险传导", "view": "作为半导体分析师，我对 Research note 03–07 做压力测试并最终支持，但补上一层基本面触发器：光模块/先进封装景气真实但狭窄且反身。2026 全球 1.6T 光模块需求约 500 万只，中际旭创目标 500–600 万只、约 50% 份额、约 75% 硅光 [S1]，但估值高达 7…", "check": "支持 0 / 风险 7 / 观察 0", "href": "/investor/theses/thesis-42faeae623/" }, { "date": "2026-06-21", "value": 25.4, "label": "风险", "title": "在2026-06-21，受2nm GAA先进制程 wafer 溢价与先进封装 80% 初期装配良率的限制，下一代高密度 AI 芯片的制造 BOM 将翻倍至8,15…", "view": "截至2026-06-21，本卡片评估了AI机架密度120kW+下，先进封装良率及散热材料研发对AI芯片成本与供应链的重塑。下一代2nm GAA+HBM4+液冷芯片BOM成本较3nm基准大涨77.2%至6,780.97美元，在初期封装装配良率降至80%的重度压力下更暴增113.1%至8,153.59…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-42faeae623/" }, { "date": "2026-06-22", "value": 12, "label": "风险", "title": "在2026年中美云厂商CapEx受阻于电力红线以及大模型推理通缩的双重催化下，新一代信息技术产业链估值正向能耗比占优的定制ASIC上游填料（联瑞新材）以及享受AP…", "view": "截至2026-06-22，新一代信息技术产业链正经历算力瓶颈由GPU转向电力/并网及先进封装、以及DeepSeek API降价96.37%引发软件应用毛利重构的剧烈调整。板块激活率处于4.3%极低水平。本报告筛选了联瑞新材、生益电子、宏和科技、金山办公和富创精密等5家低两融、抛压出清且具刚性催化剂…", "check": "支持 0 / 风险 6 / 观察 0", "href": "/investor/theses/thesis-42faeae623/" }, { "date": "2026-06-23", "value": 12, "label": "风险", "title": "电网扩容滞后导致的电力约束与成本上升，将倒逼AI芯片从盲目算力扩张转向极致能效比，加速背面供电、逻辑中控底座及CPO等先进封装和材料体系的迭代逻辑。", "view": "截至2026-06-23，评估表明电网扩容物理滞后限制了AI数据中心装机，导致2027-2028年GPU出货及代工封装面临出清压力。在这种极致电力约束下，AI芯片竞争焦点被迫转向“极致能效比”。这倒逼定制化ASIC、三值计算算法、HBM4逻辑底座、3D SoIC、片上CPO以及背面供电（BSPDN…", "check": "支持 0 / 风险 7 / 观察 1", "href": "/investor/theses/thesis-42faeae623/" }, { "date": "2026-06-24", "value": 12, "label": "风险", "title": "研究报告：超大规模云厂商CapEx节奏、ROIC收缩及光模块毛利侵蚀的传导机制 权威工作日期锚定： 2026 06 24 研究线程标识：已归档卡片索引 (研究记录…", "view": "研究报告：超大规模云厂商CapEx节奏、ROIC收缩及光模块毛利侵蚀的传导机制 权威工作日期锚定： 2026 06 24 研究线程标识：已归档卡片索引 (研究记录序号)： 08/08 分析师： AI Infrastructure Analyst 分析立场： Stress Test (压力测试) 1…", "check": "支持 0 / 风险 7 / 观察 0", "href": "/investor/theses/thesis-42faeae623/" }, { "date": "2026-06-25", "value": 30, "label": "风险", "title": "虽然物理电网排队与供应链原材料价格冲击显著抬高了局部建设阻力，但云厂商通过绿电PPA、高密度液冷及转向定制ASIC等结构性调整，能够完全消纳成本溢价，不构成算力扩…", "view": "本卡对电网并网延迟、变气压原材料（铜/GOES）价格波动和本地化GPU封装良率冲击进行了量化压力测试。结果表明，虽然并网延迟和原材料成本增加会显著压缩项目NPV（5年延迟将使NPV降低79.99%），但云厂商可以通过迁移西部枢纽自建绿电、液冷优化和系统级混合算力架构，消纳国产芯片和能源价格溢价，并…", "check": "支持 1 / 风险 5 / 观察 1", "href": "/investor/theses/thesis-42faeae623/" }, { "date": "2026-06-26", "value": 44.4, "label": "风险", "title": "1.6T光模块交付延迟对AI算力集群及超大规模云厂商Capex部署进度的影响", "view": "支持卡片01/02的研究线并加以边界化延伸：200G EML 的 HTOL 风险确会让 1.6T GA 滑向 2027H1，但这是一道带宽前沿与单位经济性约束，而非整体 Capex 停摆。GB300 NVL72 的 scale-up 走铜缆 NVLink、scale-out 走 ConnectX-…", "check": "支持 0 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-42faeae623/" }, { "date": "2026-06-27", "value": 21.8, "label": "风险", "title": "在2026-06-27的权威工作日背景下，国产HBM良率骤降逼迫国内算力网络化，而全球封装与电力阻尼拉长了云厂商CapEx变现周期，这使得制程免疫且订单排至202…", "view": "截至2026-06-27，本报告评估了HBM良率危机与先进封装瓶颈对算力集群技术演进的影响。国产HBM3封装良率降至2.66%，倒逼国内算力集群转向网络密集的Scale-out拓扑，推高了光模块需求弹性。全球云厂商2026年CapEx直逼7250亿美元，但受制于CoWoS晶圆先进封装冲突与吉瓦级电…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-42faeae623/" }, { "date": "2026-06-28", "value": 35.2, "label": "风险", "title": "半导体板块流动性压力引发的估值波动是否会传导至下游 AI 基础设施建设方的资本开支计划？", "view": "对下游 AI 基础设施建设方做融资韧性压测，部分支持但修正卡片05–07的传导假设。结论：A 股半导体两融平仓级联向下游资本开支的传导基本被防火墙隔断——约 98% 的资本开支由超大规模云厂商（2026 约 7,250 亿美元，+77%）与现金充裕的中国平台（阿里三年 690 亿美元、字节 ¥1,…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-42faeae623/" }, { "date": "2026-06-29", "value": 36.6, "label": "风险", "title": "国产HBM与先进封装良率及物料卡脖子瓶颈将2026年纯中资AI算力集群规模卡死在23 EFLOPS至95 EFLOPS的超低区间，这使电力瓶颈被动移回硅侧，A股投…", "view": "截至2026-06-30，本卡针对国产AI算力天花板进行压力测试。长鑫存储HBM3堆叠与国内WoS封装是限制AI部署的物理核心瓶颈。在强制100%国产材料场景下，堆叠良率跌至4.11%，年产出仅18.5万个HBM堆栈，仅能支持2.3万颗GPU（满足规划需求的1.03%），使新增电网负荷由2730M…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-42faeae623/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "CPO、光互联与铜连接决定 AI 集群内部租金迁移" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 12 位分析师和 4 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 135 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-42faeae623/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-42faeae623", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-ai-9d3836cc-2026-06-29", "title": "AI 生产率去通胀需要等待采用率和流程重构验证：动态复盘", "subtitle": "赛道：AI 软件效率/生产率。候选观察标的：MSFT, GOOGL, META, AMZN, NOW, CRM, ADBE。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-9d3836cc07/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-20", "06-21", "06-22", "06-23", "06-24", "06-25", "06-26", "06-27", "06-28", "06-29" ], "values": [ 12, 25.4, 14.8, 12, 12, 25, 38.4, 39.8, 25.2, 48.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "130", "accent": "red" }, { "label": "分析师", "value": "14", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：AI 软件效率/生产率。候选观察标的：MSFT, GOOGL, META, AMZN, NOW, CRM, ADBE。当前共有 40 条证据、2 位分析师和 8 条风险信号指向该判断；主导链条为 AI 基础设施。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型" }, { "title": "证据纪律", "text": "当前以 17 条支持、130 条风险和 13 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "17 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "130 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "AI 基础设施 / 电网 / Microsoft", "detail": "AI 基础设施 / 电网 / Microsoft / Amazon / Meta / HBM / ASIC / PPA", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-20", "value": 12, "label": "风险", "title": "变压器与GOES瓶颈将Hyperscaler从“无限速扩张”推向“按电力排序扩张”，长期capex仍高，但未绑定已锁定电力和近期接入的capex应获得更高风险折价。", "view": "截至2026-06-21，本卡对前序“电力与电网是AI算力硬约束”的主线作压力测试：变压器、开关设备与GOES供应瓶颈不会终结Hyperscaler AI资本开支周期，但会把部署节奏从按GPU/ASIC到货扩张，改造成按已送电MW、接入权和负荷灵活性分阶段扩容。Microsoft、Amazon、A…", "check": "支持 0 / 风险 7 / 观察 0", "href": "/investor/theses/thesis-9d3836cc07/" }, { "date": "2026-06-21", "value": 25.4, "label": "风险", "title": "电力容量、并网权和长周期电气设备的稀缺性为AI资本开支提供物理底部，但若负荷预测虚高、AI利用率不足或监管成本分摊恶化，该刚性会转化为新增项目放缓。", "view": "截至2026-06-21，本卡支持上一卡“估值先受压、资本开支不立即崩塌”的判断。核心证据是电力与电网瓶颈正在把AI数据中心建设变成具备供给刚性的公用事业级工程周期：美国数据中心用电占比已从2023年的4.4%走向2028年6.7%-12%的区间，变压器交付仍需2-3年，Dominion等枢纽区域…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-9d3836cc07/" }, { "date": "2026-06-22", "value": 14.8, "label": "风险", "title": "在66.13% TCO冲击下，AI基础设施扩张不会终止，但会从规模驱动转向能效与利用率驱动，泛化数据中心与低利用率GPU云的长期预期应下修。", "view": "截至2026-06-22，本卡压力测试上一卡66.13% TCO激增情景。结论是：头部hyperscaler资本开支不会急停，Microsoft、Alphabet、Meta、Amazon仍有收入、backlog或客户承诺支撑；但新增产能逻辑会从“抢GPU/抢机房规模”转向“单位功耗有效产出、签约利…", "check": "支持 0 / 风险 5 / 观察 0", "href": "/investor/theses/thesis-9d3836cc07/" }, { "date": "2026-06-23", "value": 12, "label": "风险", "title": "Hyperscaler并不缺AI Capex，真正约束算力上线节奏的是可交付电力、并网流程与长交期电力设备。", "view": "截至2026-06-24，本卡支持上一卡：AI算力瓶颈已从Capex/GPU转向通电MW。Alphabet 2026 CapEx指引1,800-1,900亿美元、Meta 1,250-1,450亿美元，Microsoft仍称需求超过可用容量，Amazon Q1购置物业与设备442.03亿美元；但S…", "check": "支持 0 / 风险 8 / 观察 1", "href": "/investor/theses/thesis-9d3836cc07/" }, { "date": "2026-06-24", "value": 12, "label": "风险", "title": "AI算力扩张下的电力供给与电网瓶颈风险", "view": "截至2026-06-24，本卡对上一卡做压力测试：电力与电网接入正在成为AI硬件交付的真实节奏约束，尤其在美国数据中心负荷集中的电网区域，但当前证据更支持“区域通电延期与capex成本上升”，而不是“800G/1.6T需求被证伪”。IEA、Berkeley Lab、NERC、PJM与Monitor…", "check": "支持 0 / 风险 7 / 观察 0", "href": "/investor/theses/thesis-9d3836cc07/" }, { "date": "2026-06-25", "value": 25, "label": "风险", "title": "Hyperscaler 资本开支节奏与先进封装/HBM 供应缺口的传导效应研究", "view": "截至2026-06-26，压力测试该命题：结论是命题对了一半且因果归因错位。Hyperscaler 2026年资本开支不降反升——四家合计约$725B、同比+77%，Amazon约$200B、Microsoft约$190B、Meta因零部件涨价上调至$125-145B[S1][S2]。供应闸门确实…", "check": "支持 0 / 风险 4 / 观察 0", "href": "/investor/theses/thesis-9d3836cc07/" }, { "date": "2026-06-26", "value": 38.4, "label": "风险", "title": "AI硬件需求仍真实，但A股估值溢价必须从“算力概念”转向“电力可得、热管理、互联带宽和订单兑现能力”。", "view": "截至2026-06-27，本卡片对AI基础设施链做压力测试：全球AI资本开支仍大，Goldman Sachs基准模型给出2026年USD 765 billion、2031年USD 1.6 trillion的年度AI capex，但Amazon、Alphabet、Meta、Microsoft的数据表…", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-9d3836cc07/" }, { "date": "2026-06-27", "value": 39.8, "label": "风险", "title": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure …", "view": "AI 数据中心电力瓶颈与超大规模云厂商的资本开支周期分化 报告日期： 2026 06 28 分析师： AI 基础设施分析师 (ai infrastructure analyst) 研究周期： 2026 年中展望 评级与 stance： 初始 / 支撑 (initial / support) 摘要 …", "check": "支持 0 / 风险 2 / 观察 0", "href": "/investor/theses/thesis-9d3836cc07/" }, { "date": "2026-06-28", "value": 25.2, "label": "支持", "title": "AI算力链的需求证明主要来自全球AI基础设施CAPEX周期，国产化替代不是替代这个主线，而是把全球CAPEX映射转化为中国本地稀缺性与政策溢价的乘数。", "view": "截至2026-06-29，本卡支持“USD/CNY企稳后成长风格回流”的前序判断，但把AI算力链驱动力拆成两层：全球AI基础设施CAPEX是核心需求验证，国产化替代是中国本地的选股与政策beta放大器。Microsoft、Amazon、Alphabet、Meta披露/指引的2026年CAPEX合计…", "check": "支持 1 / 风险 6 / 观察 0", "href": "/investor/theses/thesis-9d3836cc07/" }, { "date": "2026-06-29", "value": 48.6, "label": "风险", "title": "AI基础设施从“规模扩张”向“能效与利润驱动”的范式转型", "view": "截至2026-06-30，本卡片支持并精化前序链条（Research note 02–07）。在电力成为刚性稀缺单位的电网约束格局下，两项技术把约束转化为利润：芯片级直接液冷/浸没式液冷把PUE从约1.5–1.8压向约1.1[S1][S10]，64机柜集群10年冷却TCO由4,200万降至2,80…", "check": "支持 0 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-9d3836cc07/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "AI 生产率去通胀需要等待采用率和流程重构验证" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 14 位分析师和 6 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 130 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-9d3836cc07/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-9d3836cc07", "label": "研究网络定位", "tag": "网络" }, { "href": "/investor/deep-research/mag7-next-era-losers-6f2a8c9d/", "label": "下一代周期中，Mag7 谁会输？", "tag": "研究记录" } ], "autoGenerated": true, "kind": "living_thesis_recap" }, { "slug": "living-fd31cc177b-fd31cc17-2026-06-29", "title": "债券供给、地方债与久期需求决定利率下行空间：动态复盘", "subtitle": "赛道：债券供给/银行久期。候选观察标的：TLT, IEF, 601398.SH, 601939.SH, 601288.SH, 000001.SZ。当前共有 40 条证据、3 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。…", "date": "2026-06-29", "badge": "LIVING THESIS RECAP", "reportHref": "/investor/theses/thesis-fd31cc177b/", "chartTitle": "主线证据强度演化", "chartDescription": "由公开研究事件按日期聚合，显示支持、风险和观察证据如何改变主线强度。", "chartUnit": "证据强度", "chart": { "labels": [ "06-17", "06-18", "06-19", "06-22", "06-24", "06-25", "06-26", "06-27", "06-28", "06-29" ], "values": [ 44, 46.4, 50.8, 49.2, 23.6, 31, 12, 57.8, 50.2, 33.6 ] }, "metrics": [ { "label": "证据事件", "value": "160", "accent": "blue" }, { "label": "风险事件", "value": "93", "accent": "red" }, { "label": "分析师", "value": "17", "accent": "green" }, { "label": "方向", "value": "分歧", "accent": "orange" } ], "conclusions": [ { "title": "主线定义", "text": "赛道：债券供给/银行久期。候选观察标的：TLT, IEF, 601398.SH, 601939.SH, 601288.SH, 000001.SZ。当前共有 40 条证据、3 位分析师和 5 条风险信号指向该判断；主导链条为 宏观通胀传导。该主题应作为可跟踪投资假设，而不是孤立结论。" }, { "title": "最新变化", "text": "2026-06-29: On behalf of the Global Macro Analyst, I have completed the quantitative calculations of the yield downside, credit spread compression, a…" }, { "title": "证据纪律", "text": "当前以 6 条支持、93 条风险和 61 条观察证据进行动态跟踪。" } ], "checklist": [ { "label": "支持证据", "status": "6 条", "detail": "继续观察是否出现跨来源、跨分析师的同向验证。", "bull": "支持证据集中出现在最新日期且源报告可追溯", "bear": "支持证据停留在旧报告或单一来源", "active": "bear" }, { "label": "风险证据", "status": "93 条", "detail": "风险证据用于判断主线是否变成拥挤交易或证伪样本。", "bull": "风险被新证据消化", "bear": "风险事件数量超过支持事件", "active": "bear" }, { "label": "监控词", "status": "CGB / 银行 / Bond", "detail": "CGB / 银行 / Bond / supply / local-government / debt / duration / demand", "bull": "监控词在新报告中被反复验证", "bear": "监控词热度上升但价格或基本面未兑现", "active": "bull" } ], "timeline": [ { "date": "2026-06-17", "value": 44, "label": "支持", "title": "[auto-handoff] 商业银行净息差（NIM）降至1.40%极值背景下中小银行对低评级城投债买盘的边际收缩与流动性支持减弱评估", "view": "致 全球宏观分析师： 协同任务已完成。执行方为 金融行业分析师。 模式: adhoc 会话: b34f822c-e98e-4890-bed9-057a16a2f39b 结果摘要: Wrote handoff_response.md (10289 bytes); wrote handoff_resp…", "check": "支持 1 / 风险 1 / 观察 0", "href": "/investor/theses/thesis-fd31cc177b/" }, { "date": "2026-06-18", "value": 46.4, "label": "观察", "title": "[auto-handoff] 中小银行NIM跌破1.40%引发的信用收缩对大型公用事业二级债配置意愿的传导", "view": "致 首席策略师： 协同任务已完成。执行方为 金融行业分析师。 模式: adhoc 会话: 9260bd6a-eb3c-44f2-bb64-3074089878bd 结果摘要: Wrote handoff_response.md (11016 bytes); wrote handoff_respon…", "check": "支持 0 / 风险 0 / 观察 1", "href": "/investor/theses/thesis-fd31cc177b/" }, { "date": "2026-06-19", "value": 50.8, "label": "观察", "title": "[auto-handoff] 商业银行资产荒配置对收益率曲线及大类资产风格轮动的影响评估", "view": "致 市场情绪师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: 5fcf9b8f-49e0-4ca8-bea2-ca739e00cae6 结果摘要: Wrote handoff_response.md (10160 bytes); wrote handoff_respon…", "check": "支持 0 / 风险 2 / 观察 6", "href": "/investor/theses/thesis-fd31cc177b/" }, { "date": "2026-06-22", "value": 49.2, "label": "观察", "title": "中国主权资产负债表吸收债务迁移的跨资产定价", "view": "截至2026-06-22，本卡把卡片08“可融资但对挤出敏感”的结论翻译成跨资产定价。统一洞见：主权吸收把LGFV信用溢价社会化，从而压缩各处违约溢价，却把供给/久期溢价转移到CGB长端和股权贴现率。锚点：10年CGB 1.75%、30年2.22%、USD/CNY约6.76-6.79、CSI 30…", "check": "支持 0 / 风险 0 / 观察 1", "href": "/investor/theses/thesis-fd31cc177b/" }, { "date": "2026-06-24", "value": 23.6, "label": "风险", "title": "[auto-handoff] Asset Allocation Strategy on Yield Curve Steepening", "view": "致 首席风控官： 协同任务已完成。执行方为 首席策略师。 模式: adhoc 会话: d361aa5d-fb62-4f2b-8f78-8b955c9b4b94 结果摘要: Wrote handoff_response.md (9963 bytes); wrote handoff_response.…", "check": "支持 0 / 风险 5 / 观察 2", "href": "/investor/theses/thesis-fd31cc177b/" }, { "date": "2026-06-25", "value": 31, "label": "风险", "title": "[auto-handoff] Track local-bond supply, DR007 anchor, and high-dividend valuation…", "view": "致 消费行业分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: fdcc98d9-ec47-4bcf-ad48-9de0a3593d19 结果摘要: I will start by running the command to verify the current d…", "check": "支持 0 / 风险 3 / 观察 0", "href": "/investor/theses/thesis-fd31cc177b/" }, { "date": "2026-06-26", "value": 12, "label": "风险", "title": "[auto-handoff] 商业银行 1.40% NIM 约束对保障房收储贷款投放的挤出压力", "view": "致 房地产分析师： 协同任务已完成。执行方为 中国宏观分析师。 模式: adhoc 会话: aa75509e-3210-4ab6-b781-37670c2d5374 结果摘要: Wrote handoff_response.md (9392 bytes); wrote handoff_respon…", "check": "支持 0 / 风险 10 / 观察 0", "href": "/investor/theses/thesis-fd31cc177b/" }, { "date": "2026-06-27", "value": 57.8, "label": "观察", "title": "I have conducted a comprehensive stress test of the CGB 10Y floor and yield curve…", "view": "I have conducted a comprehensive stress test of the CGB 10Y floor and yield curve dynamics under the PBOC's targeted liquidity operations, comparing …", "check": "支持 0 / 风险 0 / 观察 2", "href": "/investor/theses/thesis-fd31cc177b/" }, { "date": "2026-06-28", "value": 50.2, "label": "风险", "title": "I have completed the research for Research note 01 of this research thread sessio…", "view": "I have completed the research for Research note 01 of this research thread session. Summary of Work 1. Analyst Selection : Appointed the Global Macro…", "check": "支持 0 / 风险 1 / 观察 1", "href": "/investor/theses/thesis-fd31cc177b/" }, { "date": "2026-06-29", "value": 33.6, "label": "风险", "title": "[auto-handoff] Hormuz shock & core CPI transmission", "view": "致 深研主编： 协同任务已完成。执行方为 首席经济学家。 模式: adhoc 会话: 32304902-3e3b-42f2-9a0f-86c013c361a7 结果摘要: Wrote handoff_response.md (8907 bytes); wrote handoff_response.…", "check": "支持 0 / 风险 4 / 观察 1", "href": "/investor/theses/thesis-fd31cc177b/" } ], "dimensions": [ { "label": "维度一", "title": "论点", "front": "研究问题是否清晰？", "back": "债券供给、地方债与久期需求决定利率下行空间" }, { "label": "维度二", "title": "证据", "front": "证据是否可追溯？", "back": "覆盖 17 位分析师和 5 类公开来源。" }, { "label": "维度三", "title": "证伪", "front": "风险是否上升？", "back": "风险事件 93 条，方向为 分歧。" } ], "sources": [ { "href": "/investor/theses/thesis-fd31cc177b/", "label": "主线详情页", "tag": "Thesis" }, { "href": "/investor/research-network/#thesis-fd31cc177b", "label": "研究网络定位", "tag": "网络" } ], "autoGenerated": true, "kind": "living_thesis_recap" } ] }