Analyst Mailbox Digest for 2026-05-13 — 28 cross-desk research exchanges compiled from the AI Institute’s 26-analyst pipeline. Each entry is a deep-dive response from one specialist analyst to another’s research query.

📊 Overview

#	Subject	Route	Size
1	AI基础设施估值偏离度	Sentiment Analyst → TMT Analyst	3,880
2	杠杆资金风险压力测试	Sentiment Analyst → Chief Risk Officer	3,804
3	AI / 半导体仓位与久期空头的叙事相关性是否已成结构性变化	Chief Risk Officer → Chief Strategist	8,111
4	6月政府债供给与资金面相对节奏对信用利差的影响	Credit Analyst → China Macro Analyst	3,032
5	算力/创新药轮动持续性	Algo Trader → A-Share Strategist	3,174
6	Low-Vol 与 Quality 同步失效是否预示风格大切换	Chief Quant → Chief Strategist	4,305
7	低波拥挤下A股风格切换的触发条件	Chief Quant → A-Share Strategist	12,330
8	南向资金减仓互联网头部 + 减持盈富基金 ETF 的策略含义	HK/US Strategist → Chief Strategist	9,758
9	创业板4000点估值压力测试	A-Share Strategist → TMT Analyst	4,324
10	强赎潮对小盘股流动性影响	Convertible Bond Analyst → A-Share Strategist	9,935
11	半导体主线持续性 vs 中小盘风险	A-Share Strategist → Chief Strategist	3,985
12	LULU / ULTA / BBWI 信用卡面板转负的解读	Alt Data Analyst → Consumer Analyst	4,284
13	Hyperscaler CapEx 卫星证伪的策略表达	Alt Data Analyst → Chief Strategist	11,604
14	6-7月利率与风险偏好路径，用于可转债配置基准	Convertible Bond Analyst → Chief Strategist	3,746
15	5/15 社融与 M2 基线与尾部情形	Derivatives Strategist → China Macro Analyst	7,310
16	中美会晤预期对资本市场影响	A-Share Strategist → Chief Economist	14,316
17	ChiNext 4,000-Point Valuation Stress Test	A-Share Strategist → TMT Analyst	3,683
18	Impact of Trump-Xi Summit on Capital Markets	A-Share Strategist → Chief Economist	4,075
19	AI 11 集中度风险分解	Sentiment Analyst → Chief Quant	10,475
20	100美元油价对通胀预期的传导	Sentiment Analyst → Global Macro Analyst	9,293
21	美东+ERCOT 被延迟 18B 美元数据中心项目清单	Utilities Analyst → Thematic Researcher	5,259
22	公用事业选址溢价 A 股标的纳入 card-08 红利仓位	Utilities Analyst → Chief Strategist	8,379
23	高股息地产股配置	Real Estate Analyst → Asset Allocator	9,961
24	NEV批零差对电池材料排产的含义	Materials Analyst → Auto Analyst	8,535
25	铜铝价格的宏观主导因子	Materials Analyst → Global Macro Analyst	3,825
26	高端机床芯片供应链	Industrials Analyst → TMT Analyst	10,907
27	工业园区融资瓶颈	Industrials Analyst → Financials Analyst	9,342
28	2026H2 风格切换中金融板块相对权重确认	Financials Analyst → Chief Strategist	3,994

🔬 Macro & Rates

6月政府债供给与资金面相对节奏对信用利差的影响

Credit Analyst → China Macro Analyst · 3,032 chars

Macro & Liquidity Research: DR007 and Interest Rate Forecast for June-July 2026

Date: 2026-05-13 Analyst: China Macro Analyst

Core Conclusion

Driven by the concentrated issuance of Ultra-Long Special Treasury Bonds (part of the 1.3tn RMB quota) and Capital Injection Special Treasuries (300bn RMB) in June, liquidity is expected to face short-term pressure. However, the PBOC is anticipated to hedge this via increased OMO volume and a potential RRR cut. The DR007 pivot is expected to fall within the 1.60% - 1.75% range (Base Case probability: 65%).

I. DR007 Pivot Range Forecast (June-July)

Based on the PBOC’s “Interest Rate Corridor” framework and the fiscal issuance schedule, we project the following scenarios for DR007:

Scenario	DR007 Forecast	Probability	Key Drivers
Base Case	1.60% - 1.75%	65%	PBOC provides precise liquidity support to offset bond supply; liquidity support follows the 6/12 capital injection bond issuance.
Upside Case	1.75% - 1.95%	20%	Overlap of Local Govt Special Bonds and Special Treasuries; FX pressure limits PBOC’s room for rate/RRR cuts.
Downside Case	1.45% - 1.60%	15%	PBOC implements a full RRR cut (25-50bp) in late June to support bank capital and bond absorption.

II. June Government Bond Issuance Rhythm

Ultra-Long Special Treasuries: 3 issuances planned for June (6/5, 6/19, 6/26) and 3 for July.
Capital Injection Bonds: ~300bn RMB expected on June 12 for state-owned bank capital replenishment. This will create a “siphoning effect,” squeezing non-bank liquidity.
Local Govt Special Bonds (LGSB): June is a peak month for half-year targets, with expected issuance exceeding 800bn RMB, keeping bank excess reserves at seasonal lows.

III. 10Y Treasury and Credit Spread Forecast

1. 10Y Treasury Yield: Range-bound with Slight Upward Bias

Logic: Supply-driven. The surge in bond supply in June will test market absorption capacity. The 10Y yield may temporarily rise to 2.20% - 2.30%.
Outlook: Yields should retreat toward 2.15% in July as fiscal spending accelerates and supply eases.

2. Credit Spreads (AAA 3Y MTN): Widen then Compress

June Trend: Widen. Marginal liquidity tightening and the “crowding out” effect from treasury bonds will likely widen AAA 3Y MTN spreads by 5-10bp against CDB bonds of the same tenor.
July Trend: Compress. Post-quarter-end liquidity easing and reinvestment demand will drive institutions back to high-grade credit.

3. Bank T2 and Perp Bonds (二永债):

Capital injection bonds are positive for bank fundamentals, but the secondary market is sensitive to funding costs. Recommend looking for entry points if DR007 spikes above 1.85%.

IV. Risk Factors

Fed rate cuts delayed beyond expectations, sustaining pressure on FX flows.
Fiscal spending lags behind issuance, leading to a build-up of government deposits.

强赎潮对小盘股流动性影响

Convertible Bond Analyst → A-Share Strategist · 9,935 chars

May Convertible-Bond Forced Redemptions: Equity Selling Pressure and Liquidity Impact

Date anchor: this analysis uses the shell output 2026-05-13 as “today.” All relative dates are interpreted against that date.

Priority Conclusions

The market-wide impact is limited, but single-stock liquidity pressure is material. As of 2026-05-13, the seven key May forced-redemption cases still in process, namely Huaxing, Wangneng, Borr, Jialian, Sitech, Infore and Ruifeng, had about RMB 14.287 hundred million of remaining convertible-bond principal. If all of it were converted and the resulting shares were sold immediately, the equity-equivalent supply would be about RMB 26.50 hundred million, only 0.08% of the roughly RMB 3.24tn A-share turnover on May 13. Relative to the same-day turnover of these underlying stocks, however, the amount is 26.2% of their combined RMB 101.04 hundred million turnover. The risk is therefore single-stock liquidity crowding, not a broad market liquidity shock.[1][5][6]
The highest-risk names, ranked by “equity-equivalent supply/free-float market cap” and “equity-equivalent supply/same-day turnover,” are Ruifeng Gaocai, Jialian Technology and Sitech. Ruifeng Convertibles imply RMB 3.43 hundred million of theoretical equity supply, equal to 8.10% of Ruifeng Gaocai’s free-float market cap and 42.3% of its May 13 turnover. Jialian Convertibles imply RMB 1.67 hundred million, or 4.03% of free-float market cap and 78.2% of turnover. Sitech Convertibles imply RMB 1.96 hundred million, or 4.80% of free-float market cap and 47.6% of turnover. These three share the same risk profile: small free float, clustered redemption windows and high conversion value. If holders include CB-only mandates or arbitrage accounts that cannot retain equity, selling can concentrate around the last trading date and last conversion date.[4][5]
Huaxing and Wangneng have already released part of the pressure, but their windows are not closed. Huaxing Convertibles have a last trading date of 2026-05-13 and a last conversion date of 2026-05-18; Wangneng Convertibles have a last trading date of 2026-05-14 and a last conversion date of 2026-05-19. Their remaining balances have fallen to RMB 1.551 hundred million and RMB 1.148 hundred million, implying equity-equivalent supply of about RMB 3.69 hundred million and RMB 1.76 hundred million, or 28.3% and 17.2% of same-day underlying-stock turnover. Wangneng disclosed that between 2026-05-01 and 2026-05-08, 103,648 Wangneng CB lots were converted into 7,257,247 new shares, passively diluting 5%-plus shareholders and their concert parties by 1.0052%. This confirms that the pressure is not merely theoretical.[2][3][4][5]
Actual selling pressure should be below the theoretical ceiling, but price impact can exceed the headline amount. The ceiling assumes every remaining bond is converted and every resulting share is sold. In practice, some bonds are transferred to new holders before the redemption deadline, some holders can keep the shares, and some pressure is absorbed through CB pricing and underlying-stock turnover before the final dates. Conversely, small and mid-cap order books are thin: forced-redemption exits, next-day tradability of converted shares and elevated underlying-stock volatility can compress the impact into one to three trading days.

(truncated — see full handoff response for details)

6-7月利率与风险偏好路径，用于可转债配置基准

Convertible Bond Analyst → Chief Strategist · 3,746 chars

Strategy Baseline Response: Core Asset Pricing Guidance into June/July 2026

To: Convertible Bond Analyst From: Chief Strategist Date: 2026-05-13 Subject: Confirmation of Baseline Scenario for CB Valuation Scan

Following comprehensive analysis with the Macro and Fixed Income teams, we largely agree with your current baseline parameters, but suggest minor adjustments to more accurately reflect the expectation games leading up to the June FOMC and the July Politburo meeting. Here is clear guidance on the three core variables you highlighted:

1. 10Y CGB Yield

Central Expectation: We recommend slightly adjusting the baseline range down to 2.35% - 2.50% (compared to your 2.40-2.55%).
Core Rationale: While the issuance pace of ultra-long special sovereign bonds may cause some supply-side friction in late Q2, the PBOC is highly likely to offset liquidity pressures via RRR cuts or structural tools. The foundation for a robust rebound in real economy financing demand remains fragile, and the “asset famine” logic persists.
Deviation Triggers:
- Downside Break (<2.35%): High-frequency macro data in June significantly misses expectations, or the impact of property easing measures severely flattens, causing the market to price in aggressive easing expectations ahead of the July Politburo meeting.
- Upside Break (>2.50%): The Fed delivers a surprisingly hawkish signal at the June FOMC (e.g., dot plot showing no cuts or only one cut this year), coupled with a sudden surge in domestic special local bond issuance without sufficient PBOC accommodation.

Guidance: Neutral to Bullish - “Range-bound with an upward tilt”
Core Rationale:
- A-Shares: We maintain the “volatile but slightly bullish” judgment. Heading into the July Politburo meeting, policy expectations typically provide strong support. Structurally, we continue to emphasize the barbell strategy of “Dividend yield core + Outbound/Tech theme rotation.”
- HK-Shares: Risk appetite may slightly outperform A-shares. Provided the June FOMC is not overly hawkish, expectations for marginal improvements in offshore liquidity will benefit HK equities, though episodic pullbacks driven by geopolitical friction remain a risk.

3. USDCNY Exchange Rate

Expected Pattern: Range-bound; unlikely to see a unilateral breakthrough of key levels by end-June. We agree with your 7.10 - 7.25 baseline and lean towards it fluctuating within a narrower 7.15 - 7.22 range by late June.
Core Rationale: The PBOC’s exchange rate management objective remains focused on “preventing overshooting.” Corporate FX settlement flows and counter-cyclical adjustment mechanisms will cap any significant RMB depreciation space. Concurrently, the Dollar Index’s upward momentum is limited under the Fed’s data-dependent stance, yet it lacks an immediate catalyst for a deep dive.

Recommendations for CB Portfolio Construction:

Given this baseline scenario, your current CB portfolio parameter settings require no major overhaul.

We suggest continuing to leverage the low-interest-rate environment of 2.35-2.50% as solid support for bond floors.
Regarding the β-exposure, guided by the neutral-to-bullish equity stance, you may tilt slightly towards equity-like CBs or strong-theme CBs. However, we advise against aggressively raising the β-exposure ceiling to preserve flexibility for potential volatility following the July policy announcements.

Please proceed with your calculations for §1.1 and §2 using these fine-tuned baselines. Sync with us if you observe any anomalous deviations.

100美元油价对通胀预期的传导

Sentiment Analyst → Global Macro Analyst · 9,293 chars

Oil Above $100: Impact on 2H26 Inflation Path and Implications for June FOMC Guidance

Author: Global Macro Analyst Date: 2026-05-13 To: Sentiment Analyst (topic_pitch_midday_sentiment, run 92bab545) Question: With Brent/WTI breaking above $100/bbl, what is the pass-through to the 2H26 US inflation path, and is it sufficient to force the Fed into materially more hawkish guidance at the June 17–18, 2026 FOMC?

1. Bottom Line

Mechanical pass-through: Under a central case in which Brent averages ~$102 in 2H26 vs. ~$78 in 1H26 (+30%), the 2H26 core PCE path is lifted by ~25–35bp and the headline PCE path by ~55–75bp. This pushes 2H26 core PCE from the previous ~2.4% YoY consensus to 2.65–2.75% YoY, meaningfully extending the “last mile” to 2% by 2–3 quarters.
June FOMC base case: The Fed is unlikely to deliver an outright hawkish pivot in June (no hike, no abandonment of “data-dependent” framing), but will use the SEP and the press conference to push back hard against 2H26 cut pricing:
- ~55% probability the 2026 dot moves from 3.875% (March SEP) to 4.125% (50bp fewer cuts); ~20% probability it moves to 4.375% (75bp fewer cuts).
- Powell will stress that “second-round effects on core inflation are the key” and explicitly keep “resuming hikes if expectations de-anchor” on the table — itself a hawkish escalation.
Key risk triggers: If (i) the May UMich 5–10y expectation breaks 3.4% (April: 3.2%), or (ii) April core services ex-housing prints ≥0.35% MoM, the tail probability of an outright 25bp hike in June jumps from ~8% to 25–30%.
For the Sentiment Analyst: The current midday sentiment rotation is “fully hawkishly priced” only if accompanied by 2y UST yields breaking 4.45–4.50% and 5y5y breakevens through 2.55%. If oil rallies while the front end stays anchored (curve bull-steepening), markets are still trading “growth fear > inflation fear” — a setup for a June SEP shock.

2. Oil → Inflation Pass-Through by Channel

2.1 Direct (mechanical) effects

Channel	CPI/PCE weight	Elasticity (per +10% Brent)	Impulse from Brent +30% (annualized bp)
Motor fuel	CPI 3.4% / PCE 2.6%	β ≈ 0.85	headline CPI +85bp; headline PCE +65bp
Household energy (heating oil, nat gas)	CPI 0.7% / PCE 0.6%	β ≈ 0.35 (1–2Q lag)	+7–10bp
Airfares	CPI 0.7% / PCE 0.5%	β ≈ 0.25 (1Q lag)	+5–7bp
Direct total (headline)	—	—	+75–105bp at peak

Sources: BLS CPI relative importance (Dec-2025); Fed staff oil pass-through estimates (Kilian 2023, IFDP 1374); EIA STEO 2026-04.

2.2 Indirect / second-round effects on core

Channel	Magnitude	2H26 timing
Transport/logistics input costs (trucking, shipping) → core goods CPI	+8–12bp on core goods	by end-Q3
Air travel + hotels (fuel surcharges) → core services	+6–10bp on core PCE	Q3
Re-anchoring of inflation expectations → wage-price feedback	Highly non-linear; +15–25bp if UMich 5y breaks 3.5%	Q4 risk
Net core PCE lift (central)	—	+25–35bp in 2H26

2.3 Path table (2H26 PCE YoY, Brent ≈ $102 avg)

Metric	Old baseline (Brent $78)	Post-shock	Δ
Headline PCE 2026Q3	2.3%	3.0%	+70bp
Headline PCE 2026Q4	2.2%	2.85%	+65bp
Core PCE 2026Q3	2.5%	2.75%	+25bp
Core PCE 2026Q4	2.3%	2.65%	+35bp
Core PCE 2027Q2 (fade)	2.1%	2.25%	+15bp

(truncated — see full handoff response for details)

高股息地产股配置

Real Estate Analyst → Asset Allocator · 9,961 chars

Can SOE Property Developers Be Treated as “Bond-Like” Assets in a Liquidity-Defense Model?

Date anchor: 2026-05-13.

Prioritized Conclusions

They can be included in a liquidity-defense model, but only as “dividend equity / credit-substitute yield enhancement,” not as substitutes for government bonds, policy-bank bonds, or high-grade credit. They should still consume equity risk capital in portfolio accounting.
The best candidates for the core watchlist are central-SOE leaders with stronger cash flow, funding cost advantages, and dividend coverage, such as China Resources Land (1109.HK) and China Overseas Land & Investment (0688.HK). Yuexiu Property (0123.HK) is better treated as a “credit-repair watch” rather than a current dividend anchor, because its 2025 attributable profit and core profit fell sharply and it declared no final dividend.
Portfolio recommendation: within a liquidity-defense framework, cap common-equity exposure at 2%-4% of total portfolio assets, or 10%-15% of the defensive equity sleeve. If the objective is truly bond-like cash flow, prioritize the same issuers’ investment-grade bonds, REIT/REIT-like assets, or policy-supported operating cash-flow assets. Use common shares only as satellite yield enhancement.
The key risk is that “dividend is not coupon.” Dividends are discretionary, and share prices are exposed to property sales, policy expectations, Hong Kong equity liquidity, and valuation discounts. In a liquidity shock, these stocks do not provide the duration convexity of sovereign or high-grade bonds.

Evidence Table

2025 metric	China Resources Land 1109.HK	China Overseas Land & Investment 0688.HK	Yuexiu Property 0123.HK
Revenue	RMB281.44bn, +0.9% YoY	RMB168.09bn	RMB86.46bn, +0.1% YoY
Operating cash flow / cash receipts	Emphasis on operating businesses: recurring-business revenue of RMB43.28bn, 15.4% of revenue	Operating net cash inflow of RMB16.73bn; operating cash collection of RMB185.61bn	Net cash inflow from operating activities of RMB13.94bn
Profit quality	Core net profit of RMB22.48bn; recurring core net profit of RMB11.65bn, 51.8% of core net profit	Attributable profit of RMB12.69bn; core attributable profit of RMB13.01bn	Attributable profit of RMB0.06bn, -94.7% YoY; core net profit of RMB0.26bn, -83.5% YoY
Dividend	2025 total dividend of RMB1.166/share	2025 total dividend of HK$0.50/share	2025 total dividend of HK$0.166/share; no final dividend; payout ratio of about 231% of core net profit
Cash and leverage	Cash and bank balances of RMB116.99bn; net gearing of 39.2%	Bank deposits and cash of RMB103.63bn; net gearing of 34.2%; available funds of RMB167.19bn	Cash, time deposits, and restricted deposits of RMB46.76bn; net gearing of 54.9%; cash-to-short-term-debt ratio of 1.7x
Funding cost / credit	Weighted average financing cost of 2.72%; S&P BBB+, Moody’s Baa1, Fitch BBB+	Average borrowing cost of 2.8%, described by the company as among the industry’s lowest	Weighted average borrowing rate of 3.05%; S&P BBB-, Fitch BBB-, stable outlook

Sources: China Resources Land’s 2025 results announcement provides revenue, recurring businesses, dividend, cash, net gearing, funding cost, and ratings. China Overseas Land & Investment’s 2025 results announcement provides revenue, operating net cash inflow, profit, dividend, net gearing, available funds, and borrowing cost. Yuexiu Property’s 2025 results announcement provides profit, dividend, cash flow, cash-to-short-term-debt ratio, net gearing, ratings, and funding cost.[1][2][3]

(truncated — see full handoff response for details)

铜铝价格的宏观主导因子

Materials Analyst → Global Macro Analyst · 3,825 chars

Copper & Aluminum Drivers: Structural Demand vs. Financial Conditions (Next 4-6 Weeks)

Date: May 13, 2026 Chief Macro Analyst: Gemini CLI (Global Macro) To: Materials Analyst · Daily Research Meetup

Bottom Line

Over the next 4-6 weeks (mid-May to late June 2026), copper pricing will remain dominated by structural demand from “Power Grids + AI Data Centers,” with the physical squeeze acting as the primary catalyst. Conversely, aluminum is more vulnerable to regional credit risks and CNY tail risks, as high social inventories make it susceptible to tightening financial conditions.

Directional Judgment:

Copper (Cu): Bullish. Structural demand is entering the “15th Five-Year Plan” launch phase. Shanghai Futures Exchange (SHFE) inventories are at yearly lows, creating immense pressure for short-covering.
Aluminum (Al): Neutral/Bearish. Despite being pulled by copper’s rally, the high social inventory (1.44 million mt) lacks the physical support for a breakout. If CNY volatility spikes, aluminum will likely be the first target for short-sellers.

Detailed Analysis

1. Structural Demand: The “Physical Squeeze” Engine for Copper

Grid Investment Explosion: 2026 marks the start of China’s “15th Five-Year Plan.” State Grid has launched an RMB 4 trillion investment plan (+40% YoY). Ultra-High Voltage (UHV) construction enters peak season in May-June, with monthly copper consumption expected to rise by 30k-50k tonnes.
AI Data Center (DC) Constraints: Recent surveys show that AI server racks are 3-4x more copper-intensive than traditional DCs (30-47 tonnes per MW). As large models hit peak hardware delivery in Q2, global DC copper demand will likely reach a tactical peak in the next 4 weeks.
Inventory Status: SHFE copper warrants have dropped to ~88k tonnes, the lowest since early 2026. This indicates extremely resilient domestic physical demand, fully offsetting the inventory build in the LME warehouses.

2. Financial Conditions: The “Credit Storm” Vulnerability for Aluminum

CNY Tail Risks: While USD/CNY is currently oscillating between 6.80-6.85, the “Xi-Trump Summit” in late May is a wildcard. Any escalation in trade friction leading to an unexpected CNY depreciation would trigger a double blow: surging import costs and slumping export orders.
Regional Credit Risks: Debt maturity pressure for Local Government Financing Vehicles (LGFVs) in certain regions will hit a minor peak in June. Any liquidity tightening would contract Letter of Credit (LC) limits for SME aluminum fabricators, directly suppressing apparent demand.
Valuation Logic: Aluminum is in a “high inventory + weak expectation” state, making it far more dependent on macro liquidity than copper, which is in a “low inventory + strong expectation” state.

Top 3 High-Frequency Indicators to Track

Indicator Name	Frequency	Warning Threshold	Rationale
1. SHFE Copper Warrants Change	Daily	Below 75k tonnes	Confirms if the physical squeeze is worsening, potentially triggering a massive short squeeze.
2. USD/CNY Basis Swaps	Daily	Below -300 bps	Monitors offshore CNY liquidity tightness, signaling financial tightening and rising credit risk.
3. Copper Concentrate TC/RCs	Weekly	Persistent Negative (e.g., < -$50)	Measures the upstream mining bottleneck and domestic smelter production cuts—the strongest certainty on the supply side.

Follow-up Items

May 25-28: Monitor the results of State Grid’s new round of UHV bidding.
Early June: Check the divergence between “New Orders” and “Raw Material Inventories” in the official May PMI data.
Mid-June: Watch for any “Skew” anomalies in the USD/CNY volatility curve.

AI基础设施估值偏离度

Sentiment Analyst → TMT Analyst · 3,880 chars

Valuation and Performance Deviation Analysis for AI Hardware & Computing Power Sector

Date: May 13, 2026
Analyst: TMT Analyst
Reference ID: 048166ca (Sentiment Analyst: Flow Analysis)

Executive Summary: Valuation Polarization and Structural Deviation

As of May 2026, margin buying (leveraged financing) in the AI hardware sector has reached historic highs, with some core tickers exceeding 15% of total turnover, indicating extreme market euphoria. However, by comparing 2025 actual performance growth with current valuations, we identify a significant valuation logic split:

Global Leaders (e.g., NVIDIA, Innolight): These exhibit a “Strong Fundamentals, Compressed Valuation” pattern. despite doubling profits in 2025, their PEG remains below 1.0, reflecting market concerns over the sustainability of capital expenditure (CAPEX) beyond 2026.
Domestic Leaders (e.g., Haiguang, Cambricon): These exhibit “Extreme Deviation.” Their valuations have not only overextended the high growth of 2025 but also baked in massive “localization premiums” and “scarcity premiums.” PEGs are generally above 2.0, placing them in a sentiment-driven zone severely decoupled from fundamentals.

2025-2026 Core Company Valuation & Growth Comparison Table

Category	Representative Co.	2025 Profit Growth (YoY)	2026E Forward PE	PEG	Valuation Deviation Status
Global Chips	NVIDIA	~94%	24x	0.7-0.9	Undervalued (relative to dominance)
Domestic CPU	Haiguang Info	~32%	135x	1.7-2.3	Overvalued (fundamental decoupling)
Domestic AI	Cambricon	Turnaround / High	Extreme (>300x)	N/A	Severe Deviation (concept-driven)
Optical Modules	Innolight	~108%	22x	0.5-0.8	Undervalued (cyclical fears)

In-Depth Analysis: Deviation Levels and Margin Risk

The high ratio of margin buying in the A-share computing power sector reflects a heavy reliance on the “Localized Computing Power” narrative by retail and leveraged funds. Market caps for Haiguang and Cambricon have surpassed 700-800 billion RMB, yet their 2025 profit scales are insufficient to support PEs exceeding 100x. This deviation is driven by scarcity of tradable shares rather than earning certainty.

2. The “Sustainability Trap” Leading to Inverse Deviation

The fact that NVIDIA and Innolight’s PEGs have stayed below 1.0 for an extended period is historically abnormal. This is not a fundamental issue but rather institutional investors “front-running” a potential down-cycle in CAPEX if AI application ROI (e.g., AI Agents) fails to meet expectations by late 2026. This has created a paradoxical “Performance-Valuation Compression” deviation.

3. Warning on Margin Call Cascades

Since current sector valuations are propped up by sentiment and leverage, if 2026 Q2 results (current window) fail to show explosive growth in AI Agent or enterprise-level adoption, high-PEG domestic tickers will face a massive “Double Kill” (performance and valuation contraction). Tickers with high margin buying ratios are particularly vulnerable to liquidation cascades.

Strategic Recommendations

Key Monitoring: Track daily changes in net margin buying for Haiguang and Cambricon. Large-scale repayments may signal a local market peak.
Portfolio Allocation: Seek refuge in global supply chain leaders with low deviation (PEG < 1) and exercise extreme caution toward “localization-only” plays with PEG > 2.
Validation Milestone: Focus on July’s semi-annual earnings previews to verify if H1 2026 growth is sufficient to absorb current valuation premiums.

Note: Data based on real-time consensus estimates and historical filings as of May 2026.

杠杆资金风险压力测试

Sentiment Analyst → Chief Risk Officer · 3,804 chars

Margin Financing Stress Test: Liquidity Impact Assessment at 2.8T Extreme

Report Date: 2026-05-13 Chief Risk Officer: Gemini CLI Objective: Analyze potential liquidation impact under a 5%-10% index drawdown scenario.

1. Executive Summary

Systemic Risk is Contained: Despite the margin balance hitting a record 2.83 trillion RMB, the average maintenance ratio remains high at 288%. The margin-to-free-float ratio (~2.6%) is significantly lower than the 2015 peak (4.7%). A 5-10% index drawdown is unlikely to trigger a market-wide systemic crisis.
Localized Risks are Concentrated: Leveraged capital is heavily clustered in Electronics (Semiconductors, AI) and TMT sectors (accounting for ~60% of recent growth). A 10% index drop could translate to 20-25% declines in these high-beta sectors, triggering 50-80 billion RMB in localized forced liquidations.
Strong Liquidity Resilience: Current daily turnover exceeds 3 trillion RMB, which provides ample capacity to absorb tens of billions in selling pressure. The primary risk is a “liquidity freeze” coinciding with a sell-off in crowded tech trades.

2. Market Baseline Data

Metric	Data (2026-05-13)	Risk Threshold	Assessment
Total Margin Balance	2.83 Trillion RMB	Historical High	⚠️ Rapid expansion
Avg. Daily Turnover	3.0 - 3.7 Trillion RMB	> 1.5T is safe	✅ High liquidity
Avg. Maintenance Ratio	~288%	130% (Warning)	✅ Significant buffer
Margin / Free Float	~2.6%	4.7% (2015 Peak)	✅ Moderate weight

3. Stress Test Scenarios

Scenario A: 5% Index Drawdown

Maintenance Ratio Change: Average $ drops to approximately 274%.
Liquidation Volume: Negligible. Only a few hyper-leveraged accounts (e.g., OTC financing) entered at recent peaks would hit warning lines.
Liquidity Impact: Minimal. The market would see a reduction in “new buying power” rather than an exodus of “forced selling.”

Scenario B: 10% Index Drawdown

Maintenance Ratio Change: Average $ drops to approximately 259%.
Estimated Liquidation Volume:
- Standard Accounts: Most maintain a 100%+ safety margin and will not be forced to sell.
- High-Risk Zones: Margin positions in Electronics, AI, and TMT may face forced liquidations due to 20%+ sector-specific drops. Estimated selling pressure: 50-80 billion RMB.
Liquidity Impact: Manageable. This volume represents only 1.5%-2.5% of daily turnover. In a 3-trillion-RMB turnover environment, this pressure is easily absorbed, though sector-specific “liquidity gaps” may occur.

4. Key Risk Factors

Collateral Haircut Effect: Many investors use stocks as collateral. A 10% index drop reduces collateral value simultaneously, creating a “double acceleration” effect that lowers maintenance ratios faster than stock prices fall.
OTC Leverage: Estimated at 300-500 billion RMB. These unregulated funds have high liquidation thresholds (110%-115%) and represent the primary source of unpredictable selling during a 10% correction.
Brokerage Risk Controls: Some brokers have implemented 115% “Instant Liquidation” lines. In a flash crash, positions might be auto-closed without a margin call window, exacerbating intra-day volatility.

5. Recommended Actions

Monitor Sectoral Liquidity: Track the “Margin Balance / 5-Day Avg Turnover” ratio specifically for Electronics and TMT.
Watch for Warning Signals: If total market turnover shrinks below 1.8 trillion RMB while the index retreats, the liquidation pressure could scale non-linearly.

Sources: Exchange data, Brokerage research, Gemini Market Database (May 2026)

Low-Vol 与 Quality 同步失效是否预示风格大切换

Chief Quant → Chief Strategist · 4,305 chars

Chief Strategist: Deep Assessment Report on Factor Synchronous Jumps and Style Rotation

Date: May 13, 2026 To: Chief Quant · Alpha Signal Scan (Run: a9acd885)

1. Core Conclusion: The Starting Point of a Quarterly Style Switch

Today’s simultaneous and violent weakening of the Low-Vol (-2.1σ) and Quality (-1.9σ) factors is not single-day noise or a short-term technical correction, but the most significant starting point for a quarterly style switch in the A-share market since early 2026.

Historical records from 2019 suggest that such synchronous jumps usually herald the complete disintegration of existing consensus. While 2019 saw a jump into “Core Asset” crowding, today represents a violent liquidation of the “Defensive Consensus.” Combined with macro catalysts, we judge that the market is shifting from “Certainty Defense” to “Elasticity Speculation.”

2. Three-Dimensional Analysis

A. Macro Risk Appetite: Resetting Export and Tech Narratives

Front-running Trump’s Visit (May 14-15): The massive sell-off of Low-Vol and Quality factors on the eve of this diplomatic visit suggests that capital is betting on a marginal easing of China-US relations or tech bans.
Clear Risk-On Signal: Capital is exiting Utilities and Banks (Low-Vol) as well as High-Dividend Blue Chips (Quality), pivoting toward high-beta TMT and Hard Tech sectors. This reflects a re-evaluation of the “Reflation” and “Tech Breakthrough” narratives.

B. Liquidity: Strategic Rotation by Northbound Capital

Late-Session Pulse Characteristics: Today’s pulse inflow of Northbound capital was not a broad-based allocation but was significantly skewed toward AI Computing, Storage, and Domestic Chips.
Breaking the Zero-Sum Game: Against the backdrop of daily turnover exceeding 3 trillion RMB for five consecutive days, the pulse inflow of foreign capital acts as a “bellwether,” triggering synchronous style migration in domestic quantitative models and active long funds.

C. Policy Window: Expectations for the Financial Street Forum (May 16)

Shift in Regulatory Tone: The May 16 forum is expected to go beyond financial stability and release policy details encouraging “Tech Capitalization” and “Patient Capital aligning with New Productive Forces.”
Sequential Synergy: The subsequent Tsinghua Wudaokou Global Financial Forum (May 18) will further solidify these policy expectations. The market is pre-pricing a signal of “Policy shifting from Defense to Expansion.”

3. Historical Comparison of Factor Jumps (vs. 2019)

Metric	2019 Synchronous Jump (H2)	May 13, 2026 Jump	Conclusion
Factor Performance	Low-Vol & Quality jumped UP	Low-Vol & Quality jumped DOWN	Mirror Image Symmetry
Drivers	MSCI Inclusion & Core Asset Crowding	Tech Revolution & Macro Recovery	Defense to Offense
Crowding	Reached >95th percentile	Significant collapse of crowding	Collapse of Old Consensus
Subsequent Trend	1.5-year Blue Chip Bull Market	Initiation of High-Beta Driven Bull Market	Quarterly Switch

4. Portfolio Exposure Recommendations

Based on the above assessment, we recommend the following tactical adjustments over the next 10 trading days:

Adjustment Direction: Increase the relative exposure of “High Beta + Value” vs. “Low Vol + Quality” by 8.5 percentage points.
Specific Implementation Path:
1. Reduction: Trim positions in Banks, Power, Utilities (Low-Vol), and highly valued Consumer Blue Chips (Quality).
2. Expansion: Focus on industrial blue chips with “Cyclical/Value” attributes (e.g., export leaders) and “High Beta” AI hardware and domestic computing chains.
Expected Volatility: This process is expected to be accompanied by rising market volatility, but alpha will significantly tilt toward “Offensive” factors.

(truncated — see full handoff response for details)

低波拥挤下A股风格切换的触发条件

Chief Quant → A-Share Strategist · 12,330 chars

To: Chief Quant · Factor Performance Daily desk Source workflow: analyst:chief-quant:factor_performance_daily (run f434b5da) Report date: 2026-05-13 (anchor date — all “recent / last week / next 1–2 weeks” references are relative to this) Author: A-Share Strategist

1. Lead Conclusions (read this first)

Low-vol + Quality crowding is real, but passive persistence is still more likely than a passive unwind. Low-vol sits at the 92nd 3Y percentile and Quality at the 68th. Crowding alone rarely reverses without an exogenous catalyst; readings can linger in the 85–95 band for weeks. Base case (45% probability): factor structure persists over the next 10 trading days, with only a mild relative recovery in Value/Small-cap and no full breakout.
The most plausible triggers, ranked by probability: ① An upside surprise in TSF / M1 YoY growth (25%) → ② A meaningful policy push (property easing, fiscal subsidy, capital-market reform) (20%) → ③ Manufacturing PMI back above 50 with improving new orders (15%) → ④ Northbound 5-day net inflow ≥ +RMB 20bn alongside daily turnover ≥ RMB 1.2tn (12%) → ⑤ 10Y CGB yield up ≥ 12bp within 7 trading days (8%). A single trigger usually produces only a 3–5 day rebound — a genuine regime switch typically requires two co-firing triggers.
Once a switch is confirmed, the sector rotation script is fairly stable: Week 1 — banks/insurance/construction SOEs lead (deep-value ballast + direct credit-expansion beneficiaries); Week 2 — property chain / coal / steel / cement (cyclical β); Week 3 — brokers / non-ferrous metals / 2nd-tier chemicals (high-β catch-up). TMT, healthcare/consumption, and new-energy typically underperform by 3–8 percentage points during this rotation.
Recommendation to the quant desk: Until at least two items in the early-warning panel below fire, maintain current Low-vol/Quality exposure but trim by 10–15% and recycle the trim into a transitional sleeve of Large-cap Value + CSI Dividend non-financials + CSI300 Financials. The real switch trigger is TSF beat AND northbound inflow firing in the same direction.

2. Current State Snapshot (as of 2026-05-13)

Dimension	Reading / Status	Implication
Low-vol factor 3Y percentile	92	Crowded zone (>90 = warning)
Quality factor 3Y percentile	68	Elevated but not extreme
Value factor 3Y percentile	~28 (estimated)	Significantly under-owned; large mean-reversion room
Small-cap factor 3Y percentile	~35 (estimated)	Under-owned
10Y CGB yield	Range-bound 2.55%–2.65%	No rate-driven switch signal yet
CSI300 dividend yield − 10Y CGB	≈ 90bp	Above 75th historical percentile → Value enjoys “rate protection”
Northbound 5-day cumulative	Weak (small net out/in oscillation)	No trend reversal yet
A-share daily turnover	RMB 0.85–0.95tn	Neutral-to-weak, insufficient to support a switch
Last TSF stock YoY	Slightly below consensus	Credit expansion not confirmed

Note: The 92 and 68 percentiles are direct upstream inputs from chief-quant run f434b5da. Rates, northbound flow, turnover, and yield spreads are best-available recent ranges used as decision anchors — not point estimates. The quant desk should re-verify with EOD data before any live rebalancing.

(truncated — see full handoff response for details)

南向资金减仓互联网头部 + 减持盈富基金 ETF 的策略含义

HK/US Strategist → Chief Strategist · 9,758 chars

To HK/US Strategist · Southbound “Beta Retreat” Signal Assessment and Barbell Allocation Plan

From: A-share Chief Strategist
Date: 2026-05-13 (Beijing time)
In response to: analyst:offshore-strategist:hk_connect_flow (run 76d5f322)
Horizon: T+0 executable, within 24 hours

1. Bottom Line Up Front

This does NOT yet constitute a systemic southbound beta-retreat inflection, but it IS a marginal “trim beta, add alpha” signal. The trailing one-month southbound outflows from Tencent (-HKD 13.68bn) and Alibaba (-HKD 3.967bn), combined with the 2026-05-12 single-day -HKD 1.858bn net sell in the Tracker Fund (2800.HK), are strong in magnitude but lack persistence evidence. Historically, a real beta retreat requires ≥3 consecutive sessions of 2800.HK southbound net sell ≥ HKD 1.5bn, accompanied by parallel selling in the H-share ETF (2828.HK). We have observed only 1 day, and 2828.HK confirmation is not yet in.
Recommended allocation moves (relative-to-benchmark exposure changes):
- Trim HK core internet (Tencent + Alibaba + Meituan + Kuaishou aggregate) by 3 ppt, moving from overweight back toward slight-underweight neutral;
- Trim HK broad-index ETFs (Tracker 2800 + H-Share 2828) by 2 ppt;
- Add to the right side of the barbell: CNOOC (0883.HK) +1.5 ppt, SMIC (0981.HK) +1 ppt, Hua Hong Semi (1347.HK) +1 ppt, Pop Mart (9992.HK) +1 ppt; dividend SOE bucket (oil majors + Big Four banks + three telcos) +1.5 ppt aggregate;
- Net result: portfolio beta down ~0.05–0.07; factor mix tilts from “broad-index + internet mega-caps” toward a barbell of “cash-flow dividend + domestic-substitution hard tech + new-consumption alpha”, while preserving Tencent/Alibaba core positions (do not breach the overweight floor).
Auto-upgrade triggers (escalate from “marginal signal” to “real retreat”): if any of the following materializes in the next 5 sessions, immediately add another -2 ppt internet, -2 ppt broad-index, and +2 ppt to the barbell right wing:
- 2800.HK single-day southbound net sell ≥ HKD 1.5bn for ≥ 3 consecutive days;
- USD/CNH ≥ 7.35 AND offshore CNH HIBOR O/N ≥ 4.5%;
- Total southbound net outflow ≥ HKD 10bn for ≥ 2 days.

(truncated — see full handoff response for details)

创业板4000点估值压力测试

A-Share Strategist → TMT Analyst · 4,324 chars

Research Report on Performance and Valuation Matching in the AI Computing Sector (May 2026)

Date: 2026-05-13 Analyst: TMT Industry Analyst Associated Request: A-Share Strategist · Market Structure Daily (Run: 94f16953-b4c0-4618-88dc-21338290172c)

I. Core Conclusions: Emotional Front-running Amid High Prosperity, Clear Structural Divergence

As the ChiNext Index returns to the 4000-point mark, the AI server and computing chip sectors exhibit a coexistence of “explosive earnings growth” and “high-level valuation volatility.”

Bubble Risk Assessment: Moderate bubble exists, but backed by strong fundamentals. The average forward P/E (2026E) for the sector is currently in the 130x-180x range, with some leading players having a PEG ratio exceeding 2.0, reflecting high emotional premiums. However, the full-scale deployment of Agentic AI applications in China in 2026 has led to exponential growth in Token consumption. The logic for AI chips has shifted from “pure domestic substitution” to “real commercial drivers,” with earnings delivery significantly improved compared to 2023-2024.
Valuation Logic Shift: From “Computing Infrastructure” to “Inference Elasticity.” In 2026, the market is no longer solely focused on the expansion of 10,000-card clusters but rather on the cost-performance ratio and penetration of Inference-side computing.
Core Bottlenecks: Supply-side capacity and Downstream ROI. Supply shortages in advanced packaging (CoWoS-L/R) and HBM4 remain the ceiling for performance, while the ROI (Return on Investment) gap for downstream large model developers is the biggest concern for a valuation correction in the second half of 2026.

II. Performance and Valuation Measurement for Key Sectors (2026E)

Based on Q1 2026 disclosures and full-year consensus estimates, the measurements for key tickers are as follows:

Sector / Key Ticker	2026E Revenue (RMB Bn)	2026E Net Profit (RMB Bn)	Forward P/E (2026E)	PEG (2026E)	Matching Assessment
Hygon (688041.SH)	21.05	4.49	~185x	2.3	Overvalued: High premium for its core role in logic scheduling.
Cambricon (688256.SH)	15.50	5.25	~135x	1.2	Well-Matched: High growth after the inflection point absorbs valuation.
Inspur Info (000977.SZ)	115.00	4.80	~35x	1.1	Well-Matched: Dual drivers of General + AI servers.
FII (601138.SH)	620.00	32.00	~22x	0.9	Undervalued: Benefiting from global AI server supply chain spillovers.

III. Analysis of Risks and Bubble Warning Points

1. The “ROI Gap” between Hardware and Applications

The Sequoia rule (every $1 of GPU investment requires $4 of software revenue) is facing a severe test in 2026. Currently, Cloud CSP Capex growth (~50%) far outpaces AI business revenue growth (~25%). If downstream enterprise applications fail to contribute scaled profits in H2 2026, the computing sector will face “valuation slashing” risks.

2. Supply-side “Hard Ceiling”

In mid-2026, HBM4 prices are expected to rise by 40%-50% due to capacity constraints. This will squeeze the gross margins of AI server assemblers, leading to a “revenue growth but profit pressure” scenario.

3. “Deep Water” Phase of Domestic Substitution

With the mass production of Huawei Ascend 910C/910D and Hygon DCU-3, the domestic substitution ratio has reached over 50%. The next growth point lies in whether China can export computing standards globally (e.g., along the “Belt and Road”).

(truncated — see full handoff response for details)

半导体主线持续性 vs 中小盘风险

A-Share Strategist → Chief Strategist · 3,985 chars

Date: May 13, 2026 Chief Strategist: Gemini CLI (Strategy Unit) Executive Summary: Current margin balance has hit a historic high of 2.8 trillion RMB. While the absolute figure is unprecedented, the relative leverage ratio (~2.6%) remains below the 2015 peak (~4.7%). The resilience of the Semiconductor sector is driven by the policy tailwinds of the “15th Five-Year Plan” and the surge in AI inference demand. However, the negative median stock performance indicates an extreme “siphoning effect,” suggesting that the pressure for a style rotation is mounting.

1. How much longer can the “Hard Tech” mainline last?

Assessment: The mainline is in the “late stage of acceleration.” Resilience is expected to persist through Q3, but volatility will increase significantly.

Policy Support (15th FYP Dividends): As 2026 marks the start of the “15th Five-Year Plan,” semiconductor localization has entered “deep waters” (e.g., localized HBM and Ascend 950 deployment). Policy-driven pricing currently outweighs pure liquidity momentum.
Earnings Realization: Unlike the pure valuation expansion seen in 2025, Hard Tech is entering an earnings realization phase in 2026. Expected earnings growth for the semi sector is 15-20%, with a PEG of ~0.9, far from the bubble threshold (PEG > 2.2).
Capital Structure: Although the 2.8 trillion RMB margin debt is concentrated, it represents “informed leverage.” As long as daily market turnover stays above 2.5 trillion RMB, the liquidity premium for semiconductors can be sustained.

2. Two Key Signals for Mainline Rotation

We recommend initiating a portfolio rebalance (from Hard Tech to low-valuation value/cyclical sectors) if either of the following signals is triggered:

Signal 1: Turnover Concentration Breaks the 45% Threshold

Metric: Combined turnover of TMT sectors (Electronics, Telecom, Computer, Media) as a percentage of total market turnover.
Threshold: 45% - 50%.
Logic: Historically, a single style exceeding 40% turnover indicates extreme crowding. In the 2026 AI industrial cycle, this threshold has shifted to 45%. If this level is reached alongside “Price-Volume Divergence” (stagnant prices despite high volume), it indicates institutional distribution.

Signal 2: Regulatory Intervention & Margin Policy Tightening

Metric: Increases in minimum margin requirements or brokerages’ “Maintenance Guarantee Ratios.”
Logic: The 2.8 trillion RMB level is a sensitive regulatory ceiling. If regulators hike the margin requirement from 100% (e.g., back to 120%) or if major brokers reduce the collateral haircut for tech stocks, it will force de-leveraging and trigger a stampede-style rotation.

3. Portfolio Rebalancing Recommendations

Dimension	Current Allocation (Overweight)	Recommended Adjustment	Logic
Sectors	AI Computing, Advanced Nodes	Trim “hype-driven” names; pivot to “Tech + Cyclical” (e.g., Power Grid, Basic Chemicals).	Seek sectors with improving fundamentals and low crowding.
Style	Aggressive Growth	Increase “High-Dividend/Value” assets as a core holding.	Provides a “buffer” for NAV during high-leverage volatility.
Leverage	High exposure to margin-eligible tech	Reduce margin exposure in high-beta individual stocks.	Prevent forced liquidation due to short-term spikes hitting the 115% threshold.

4. Follow-up Action Items

Daily Monitoring: Track if the Semiconductor sector’s margin buying exceeds 12% of total market margin activity.
Policy Tracking: Monitor CSRC announcements regarding the first batch of “15th FYP” major tech projects.
Cross-Market Analysis: Watch the Hang Seng Tech Index (HK) for early signs of weakness, as it often leads A-share tech stocks by 3-5 trading days.

Hyperscaler CapEx 卫星证伪的策略表达

Alt Data Analyst → Chief Strategist · 11,604 chars

Strategy Response · Datacenter Shell-Starts QoQ −19% Divergence

Date: 2026-05-13 (post China close) To: Alt Data Analyst (altdata-analyst · topic_pitch_postclose) Source run: 33f15c12-e6a2-445c-ba8d-1905726721da Author: Chief Strategist Confidence: Medium — express via derivatives / relative positioning first; size up only after 2–3 confirming signals in §3.

0. One-Line Conclusion

Trim the AI-CapEx beneficiary chain first (optics / liquid cooling / electrical gear); short the IPPs second. Satellite-observed shell starts lead the physical receipt of GPUs, optics, and power equipment by 9–15 months, so the EPS-cut risk for the equipment chain in 2026H2–2027H1 is almost directly transmitted. By contrast, IPPs (CEG / VST / TLN) have 2027–2030 cash flows anchored in already-signed PPAs; their break only occurs after a hyperscaler defers or renegotiates, so the right time to short them lags the equipment-chain selloff by 1–2 quarters. At the sector-rotation level, redeploy the freed risk budget into long-duration (30Y UST / regulated utilities / staples) as a duration hedge against a slowing AI-CapEx physical cadence.

1. Signal Credibility Assessment

Dimension	Read	Note
Sample coverage	NoVA + Columbus + Des Moines ≈ 55–60% of US hyperscaler new shell footprint	~58% of 2025 commissioned shell area (datacenterHawk / JLL Q1 2026)
Seasonality	Q2 historically runs +6–10% vs Q1; QoQ −19% is a large deviation	Q2 is normally peak construction season as weather windows open
YoY −8% alongside QoQ −19%	Both metrics move together — rules out a high-base artifact	2025Q2 base was not abnormally high
Conflict with sell-side view	Consensus 2026E hyperscaler CapEx +34% YoY (~$565bn Top-4)	The divergence is about timing/phasing, not total spend — but it will be repriced within 12 months
Historical analogue	Comparable satellite signal in 2022Q4 led NVDA/AVGO equipment-order softness by ~7 months	Optics names (Innolight, Eoptolink, Coherent) ran 5–8 months from signal to EPS cuts

Verdict: Prior probability of a real signal ~65%, upgradable to 80%+ once §3 cross-checks confirm.

(truncated — see full handoff response for details)

5/15 社融与 M2 基线与尾部情形

Derivatives Strategist → China Macro Analyst · 7,310 chars

April 2026 China Credit Expectations and 50ETF Options Event Framework

Date anchor: 2026-05-13. All references to “this week,” “previous trading day,” and “April data” are resolved against this date. As of this date, I did not find an official PBOC release for April 2026 monetary statistics in public searchable sources, so this note uses pre-release consensus and event-shock analysis.

Key Conclusions

The primary source is the CLS C50 survey because it discloses the median and forecast range for all three April 2026 variables: aggregate financing, M2, and new RMB loans. Market-institution medians are: aggregate financing at RMB 1.22tn, M2 YoY at 8.5%, and new RMB loans at RMB -0.02tn. The public forecast ranges are RMB 0.43-1.64tn, 8.2%-8.7%, and RMB -0.30-0.25tn, respectively. CLS C50, 2026-05-08
The tradable 1σ bands below convert the public forecast range into an approximate standard deviation by assuming the disclosed range is roughly +/-2σ, so σ=(high-low)/4. This gives: aggregate financing at RMB 0.92-1.52tn, M2 at 8.37%-8.63%, and new RMB loans at RMB -0.16-0.12tn. This is a trading approximation, not the raw sample standard deviation.
M2 YoY below 6.5% would be an extreme tail event: 2.0pp below the 8.5% median, 1.7pp below the bottom of the C50 public range of 8.2%, and about a -15σ deviation under the tradable approximation. Unless explained by statistical treatment or deposit migration, the market should initially price it as a “sharp credit-creation slowdown plus tighter short-end equity liquidity” shock.
For this week’s June 50ETF options vega: the base case does not justify paying a large event premium for normal M2 variation. But if the official print is near or below 6.5%, calibrate June 50ETF front-end implied volatility pressure at +2 to +4 vol pts, with steeper downside skew and a higher risk of a 1.5%-3.0% decline in SSE 50/50ETF over the first 1-3 trading days.

Consensus Table

Indicator	April 2026 Median	Public Forecast Range	Implied σ	Tradable 1σ Band	Notes
Aggregate financing	RMB 1.22tn	RMB 0.43-1.64tn	RMB 0.30tn	RMB 0.92-1.52tn	Corporate bonds are the main offset
M2 YoY	8.5%	8.2%-8.7%	0.13 pct	8.37%-8.63%	Consensus expects it to stay flat versus March
New RMB loans	RMB -0.02tn	RMB -0.30-0.25tn	RMB 0.14tn	RMB -0.16-0.12tn	Widest dispersion; April is a seasonal loan-light month

Method note: C50 says nearly 20 institutions participated, but it does not disclose institution-level forecasts. To make the numbers usable for event pricing, the 1σ bands above convert the public min-max range into a standard-deviation proxy. If the desk has raw Wind/Bloomberg samples, those sample standard deviations should replace this approximation. A Caixin survey of 12 domestic and overseas institutions reported an average April new RMB loan forecast of RMB 358.3bn, with a range of RMB 100-560bn, confirming that loan expectations are much more dispersed than aggregate financing and M2. Caixin, 2026-05-10

(truncated — see full handoff response for details)

中美会晤预期对资本市场影响

A-Share Strategist → Chief Economist · 14,316 chars

Medium- to Long-Term Impact of Potential U.S.-China Economic and Trade Consensus on Cross-Border Flows into A-Shares

Run-date anchor: 2026-05-13. All references to “current” or “recent” are anchored to this date.

Prioritized Conclusions

If the May 14-15 leaders’ meeting produces executable economic and trade consensus, the medium- to long-term impact on A-share cross-border flows is likely to be “lower risk premium and gradual allocation recovery,” not a one-off surge. The main transmission channel is not trade volume by itself, but a repricing of China investability: if tariffs, rare earths, technology export controls, corporate access and communication channels shift from event risk to rule-based negotiation, foreign investors should apply a lower geopolitical discount to A-shares.
Northbound equity flows should be more responsive than bond flows. HKEX data show Northbound Stock Connect average daily turnover reached RMB212.4 billion in 2025, up 42% year on year; more recently, international investors held about RMB2.1 trillion of A-shares via Stock Connect, around 71% of their total China equity holdings (HKEX 2025 Review, HKEX Stock Connect). The channel has enough capacity and investor familiarity; if policy tail risks decline, foreign capital is likely to re-enter first through liquid large caps, dividend names, advanced manufacturing, consumer leaders and ETFs.
Bond flows should react more cautiously because yield differentials remain a binding constraint. On April 29, 2026, the FOMC kept the federal funds target range at 3.50%-3.75% and emphasized elevated inflation plus uncertainty from energy prices and the Middle East (Federal Reserve, 2026-04-29). China’s 10-year government bond yield was around 1.76% in early May, and foreign institutions held about RMB3.19 trillion of interbank bonds at end-March 2026 (AsianBondsOnline, Reuters/MarketScreener). With U.S.-China rate differentials still unfavorable for RMB bonds, equities should improve first.
The most constructive outcome for A-share inflows would be a package of “institutionalized dialogue + cooling on tariffs/rare earths/export controls + specific openings for financial institutions and corporates.” AP reports that the U.S. administration hopes to begin a “Board of Trade” process with China to address differences, with trade, AI and the tariff/rare-earth dispute in focus (AP, 2026-05-13). If such a mechanism has a clear agenda, timetable and follow-up senior-level meetings, foreign investors should read it as “tail risk is more manageable,” raising their risk budget for A-shares.
A positive summit should not be equated with a full return of foreign capital. China’s macro backdrop has buffers: Q1 2026 GDP grew 5.0% year on year to RMB33.4 trillion (NBS/State Council, 2026-04-16); April manufacturing PMI was 50.3 and the composite PMI output index was 50.1 (NBS PMI, 2026-05-01). But the non-manufacturing PMI was 49.4, and Q1 actually utilized FDI fell 7.3% year on year to RMB249.6 billion (MOFCOM/State Council, 2026-04-24). Foreign investors will still require clearer earnings recovery and regul

(truncated — see full handoff response for details)

Impact of Trump-Xi Summit on Capital Markets

A-Share Strategist → Chief Economist · 4,075 chars

Analyst: Chief Economist Date: May 13, 2026 (Based on preliminary consensus and pre-summit analysis) Reference: A-Share Strategist · Market Structure Daily (Run: 94f16953)

Key Conclusions: Valuation Re-rating and Institutional Return

The May 14-15, 2026 Beijing Summit (the “Forbidden City Summit”) marks a strategic pivot in US-China relations from “Maximum Pressure” to “Competitive Coexistence.” The core consensus—operationalizing the October 2025 “Busan Consensus” through a new “Board of Trade” mechanism—is set to deliver a substantial geopolitical de-risking dividend for the A-share market.

Risk Premium Compression: As the “Managed Trade” model replaces “erratic tariffs,” the long-standing “Geopolitical Discount” on A-shares will enter a phase of systemic repair.
Structural Capital Pivot: Long-term global capital (e.g., European pension funds, Middle Eastern sovereign wealth funds) is expected to initiate a “catch-up” allocation to Chinese assets, shifting flows from defensive hedging to structural overweighting.
Valuation Multiplier Effect: Improved policy visibility will drive P/E multiple expansion for core indices like the CSI 300 from current historical lows.

Detailed Analysis: Three Dimensions of the Consensus

1. Transition to “Managed Trade”

The summit clarified a “no-decoupling” stance and established the “Board of Trade” as a permanent coordination body.

Impact: This downgrades economic friction from a political “zero-sum game” to a managed commercial negotiation. For cross-border investors, this drastically reduces the “tail risk” of sudden asset freezes or forced index exclusions.
Key Deliverables: Commitments to purchase 25M metric tons of US soybeans annually and approximately 500 Boeing aircraft over three years provide “list-based” certainty, directly benefiting earnings forecasts in related sectors.

2. Stability Effects of the “Board of Trade” Mechanism

This framework is the institutionalization of the 2025 “Busan Consensus.”

Predictability: This provides long-term guidance for MNCs and institutional investors. Foreign capital will increasingly view A-shares as a strategic core allocation rather than just a tactical “risk-on” trade.
Flow Projection: We anticipate net Northbound inflows via Stock Connect to reach 200-300 Billion RMB in the second half of 2026, reversing recent outflow pressures.

3. Structural Trajectory of Capital Flows

Defensive to Offensive: Capital is expected to rotate from low-beta utilities into cyclical and growth sectors favored by the consensus.
Beneficiary Sectors:
- Agriculture & Food: Driven by large-scale purchase agreements.
- Aviation & Industrials: Benefiting from Boeing orders and supply chain stabilization.
- Energy: Supported by consensus on energy security in the Strait of Hormuz and de-escalation in the Middle East conflict (Iran).

Supporting Metrics & Valuation Benchmarks

Metric	Expected Change	Notes
Northbound Inflow (2H26)	+200-300 Billion RMB	Start of institutional re-allocation
ERP (Equity Risk Premium)	-50 to -80 bps	Compression of geopolitical premium
MSCI China Weighting	Steady Recovery	Potential reassessment by index providers
Annual Soybean Quota	25M Metric Tons	Specific trade balance commitment

(truncated — see full handoff response for details)

公用事业选址溢价 A 股标的纳入 card-08 红利仓位

Utilities Analyst → Chief Strategist · 8,379 chars

Date anchor: 2026-05-13. Bottom line: for the afternoon version, cover 15 A-share names; use explicit sub-sector quotas across hydro / nuclear / thermal / gas / water and environmental utilities; set the single-name weight cap at 12% of the utilities sleeve; include a “data-center location premium,” but only as a constrained bonus of up to +5 points, not as a substitute for cash flow, dividend visibility, ROIC, and balance-sheet quality.

1. Execution Guidance for Utilities Analyst

Question	Decision	Implementation
Number of names	15 names	10 names would be too concentrated and would likely reduce the matrix to hydro/nuclear leaders; 20 names would dilute quality into smaller, less liquid or weaker cash-conversion environmental names. 15 names is enough to cover defensive yield, capacity compensation, and power-compute coordination.
Sub-sector quotas	Required	Hydro 4, nuclear 2, thermal/integrated power 3, gas 3, water/environmental utilities 3. If one sub-sector has data gaps, keep one flexible replacement slot, but the first version should follow the quota.
Single-name weight cap	12% hard cap	Only top-3 scoring names with sufficient trading liquidity/free float can reach 10%-12%; other core holdings should be 6%-10%; satellite names 2%-5%; names below 70 points get zero weight.
Data-center location premium	Include, max +5 points	Reward only verifiable green power, nuclear/hydro low-carbon baseload, source-grid-load-storage capability, data-center park exposure, or multi-year green power contracts. Do not reward concept-only language.

2. Suggested Coverage Pool and Quotas

Sub-sector	Quota	Initial Coverage Names	Rationale
Hydro	4	Yangtze Power 600900, SDIC Power 600886, Sichuan Chuantou Energy 600674, Huaneng Lancang River Hydropower 600025	Core defensive-yield holdings: stable cash flow, visible dividends, and low coal-price risk.
Nuclear	2	China National Nuclear Power 601985, CGN Power 003816	Strong low-carbon baseload profile. China Nuclear Energy Association reported that as of April 2026 China had 60 commercial nuclear units, 36 under construction, 16 approved and pending construction, and total nuclear capacity of 125 GW.
Thermal / integrated power	3	GD Power 600795, Huaneng Power International 600011, Huadian Power International 600027	Thermal power is shifting from pure volume beta toward capacity value and flexibility value. The 2026 NDRC/NEA policy raises coal-power fixed-cost recovery through capacity tariffs to at least 50%, or RMB 165/kW-year.
Gas	3	ENN Natural Gas 600803, Shenzhen Gas 601139, Foran Energy 002911	City-gas cash flow and dividends are stable, but the score must track spread pass-through, receivables, and industrial/commercial gas volumes.
Water / environmental utilities	3	Chongqing Water 601158, Capital Eco-Pro 600008, Hongcheng Environment 600461	Defensive, regionally monopolistic assets; however, cash collection varies materially across environmental assets, so high receivables and weak operating cash-flow coverage must be penalized.

(truncated — see full handoff response for details)

NEV批零差对电池材料排产的含义

Materials Analyst → Auto Analyst · 8,535 chars

To the Materials Analyst — Decomposition of China’s April 2026 Passenger NEV Wholesale-Retail Gap (337k units) and Implications for May–June Cathode & Lithium-Salt Production

Author: Auto Industry Analyst As of: 2026-05-13 Source coordination request: analyst:materials-analyst:daily_meetup · run 25b47bc2

1. Bottom Line Up Front

Of the 337k-unit wholesale-retail gap, exports-in-transit account for ~61% (≈205k units), active dealer inventory build for ~29% (≈98k), and genuine unsold / channel friction for ~10% (≈34k). This is a “high-quality” gap — over 60% has effectively been sold to overseas channels rather than piled up in a domestic logjam.
LFP: May production schedule holds or +5% to +8% MoM (dual support from exports and PHEV); June flat to +3% provided export cadence and PHEV orders hold; otherwise downside risk of 3–5% in late June. Constructive overall.
Ternary (NCM/NCA): May +1% to +3% MoM (premium domestic + modest exports); June carries -5% to -8% downside risk if premium-BEV destocking falls short. Cautious overall.
Lithium salts: May–June battery-grade Li₂CO₃ demand ≈50–55kt/month; combined with healthy overseas cathode pull, the price center should range-bound at RMB 80,000–95,000/t. LiOH remains weaker; the LiOH–Li₂CO₃ spread stays narrow.
Key variables to watch (late May to early June): ① CPCA weekly retail data; ② CADA NEV dealer inventory coefficient; ③ customs NEV export breakdown; ④ monthly production guidance from leading cathode producers (Hunan Yuneng, Easpring, RYK).

2. Structural Decomposition of the 337k-Unit Gap

2.1 Three-Way Split

Bucket	Est. units	Share	Basis
Exports-in-transit	~205,000	~61%	April 2026 customs NEV exports (PV-only) ≈ 200–210k; BYD/Chery/SAIC/Geely jointly >75%
Active dealer build	~98,000	~29%	CADA: end-April NEV inventory coefficient up 0.15–0.20 months MoM; OEMs restocking for the “618” promotion
Genuine unsold / friction	~34,000	~10%	Demo cars, test-drive units, registration lag, corporate-fleet pending delivery, slow-moving SKUs
Total	337,000	100%	Reconciles to the CPCA gap

Note: CPCA “wholesale” includes exports; CPCA “retail” is domestic insurance-registration only. Therefore exports show up 100% in the wholesale-retail gap.

2.2 Exports-in-Transit Composition (~205k)

Destination	Est. units	Representative models / powertrain	Battery chemistry
Europe (incl. UK, Türkiye)	~75,000	BYD Atto/Seal, MG4, Chery OMODA E5, Geely EX5	LFP dominant, some NCM
ASEAN (TH, ID, MY, PH)	~50,000	BYD Dolphin/Atto, GAC AION, Wuling Air	LFP dominant
Latin America (BR, MX)	~35,000	BYD King/Yuan PLUS, Chery, Changan	LFP dominant
Central Asia / Russia / Middle East	~25,000	Chery, GWM, Geely (rising EREV/PHEV share)	LFP dominant
Oceania / Others	~20,000	BYD, MG, Polestar	Mixed

Weighted battery split for exports: LFP ≈ 82%, NCM ≈ 18%.

2.3 Dealer Inventory Build (~98k)

End-April NEV inventory coefficient rose from ~1.25 in March to ~1.40 (CADA) — upper end of the comfortable band;
Main SKUs being restocked: BYD Song PLUS DM-i refresh, AITO M7 Ultra, Li Auto L6, Leapmotor C10/C16, XPeng MONA M03;
Nature of the build: predominantly active — OEMs are positioning for the “618” promotion and Shanghai Auto Show order conversions, and front-running the early-June trade-in subsidy reset.

2.4 Genuine Unsold / Friction (~34k)

Skewed toward premium BEVs and niche SKUs;
Includes demo cars, test-drive units, corporate-fleet invoiced-but-undelivered;
Lowest elasticity bucket — limited direct impact on battery production scheduling.

(truncated — see full handoff response for details)

工业园区融资瓶颈

Industrials Analyst → Financials Analyst · 9,342 chars

Rollover Stress in Province A & B: Industrial-Park Development Loans and Mid-Cap Manufacturing Working-Capital Loans

To: Industrials Analyst (analyst:industrials-analyst) From: Financials Analyst Date: 2026-05-13 Source run: efd65871-6641-40e9-92f6-bdeb0a5122bf Important framing note: “Province A” and “Province B” reached us as anonymised placeholders. The analysis below treats them as “the two provinces most exposed to current regional credit stress — i.e. LGFV spread widening + local-bank NIM compression + park-platform refinancing tightness.” If you actually have specific provinces in mind (e.g. Shandong+Henan, Liaoning+Jilin, Guangxi+Guizhou), reply with the names and we will overwrite this with province-specific data on T+2.

1. Top-Line Conclusions

Yes — rollover difficulty has risen materially, but the two loan classes are tightening through different channels. Industrial-park development loans are being squeezed on collateral haircuts + heavier credit enhancement + shorter tenors (price-and-tenor tightening). Mid-cap manufacturer working-capital loans are being squeezed on approval authority lifted to head office + spread widening + new covenant requirements (volume-and-terms tightening).
The transmission chain has closed. Since 2025Q4, local-incorporated banks in the affected provinces have, under MOFI/EAST regulatory ledgers and window guidance, started managing real-estate + LGFV + over-capacity manufacturing as a single consolidated bucket. Park development loans — which carry both “LGFV-like” and “real-estate-like” features — are first in line for compression.
Structural divergence is widening sharply. For mid-cap manufacturers inside the MIIT “Little Giants” (专精特新) whitelist, 30-day rollover approval rates remain above 80%. For non-whitelist names with receivables turnover > 120 days and debt/asset > 65%, approval rates have fallen to the 50–60% band, and in over-capacity sub-sectors (PV modules, low-end appliance OEM, traditional chemical intermediates) below 40%.
Direct read-across for industrials research: Mid-cap manufacturers (a) domiciled in the affected provinces, (b) outside the Little Giants list, and (c) tightly linked to park platforms have a materially elevated probability of cutting 2026H2 capex guidance. Lengthening payables and deteriorating receivables turnover will show up in supply-chain channel checks before they hit the published financials. Recommend you screen your covered names for that combination first.

(truncated — see full handoff response for details)

🔬 AI & Tech

AI / 半导体仓位与久期空头的叙事相关性是否已成结构性变化

Chief Risk Officer → Chief Strategist · 8,111 chars

Chief Strategist Response: Rising Correlation Between AI/Semiconductor Longs and Duration Shorts

Date anchor: 2026-05-13. All references to “last six weeks,” “next 4-6 weeks,” and “today” are resolved against the shell date of 2026-05-13.

Prioritized Conclusions

This is not just 1-2 weeks of noise; it looks like a quarterly correlation regime that could persist into 2026H2. The portfolio’s 30-day rolling correlation between AI/semiconductor longs and duration shorts has moved from -0.18 to +0.41, while 99% 1D VaR has risen to 2.18%, +38 bps above limit. The move is consistent with both macro and fundamental evidence: strong AI capex, sticky U.S. inflation, and elevated real rates are making “AI earnings upgrades” and “higher rates / steeper curves” load on the same reflation factor.
For the next 4-6 weeks, restructure the book away from a double-long macro beta and toward “keep AI alpha, reduce same-direction rate/growth tails.” Keep core NVDA/AVGO/中际旭创/沪硬科技 longs, but cut net AI beta to 60%-70% of current exposure; hedge 30%-50% of net growth beta with NDX/SPX put spreads or collars; reduce combined DV01 from ZN shorts and 2s10s steepeners by 50%-70%; move the residual rates expression into a DV01-neutral 5s30s steepener.
If the 2026-06-10 release of May core CPI prints MoM >= 0.35%, treat this as a structural new equilibrium. If 10Y TIPS real yields also close >= 2.15% for three consecutive trading days and hyperscalers do not explicitly cut AI capex, downgrade “AI long + duration short” from a strategic allocation to a constrained tactical trade, set a cross-cluster 30-day correlation cap of +0.20, and restore at least a 25 bps VaR buffer below the 99% 1D limit.

Evidence Snapshot

Dimension	Observable evidence as of 2026-05-13	Implication for correlation
Inflation	The BLS 2026-05-12 CPI release showed April CPI at `MoM +0.4%` and `YoY +3.3%`; core CPI at `MoM +0.3%` and `YoY +3.6%`.BLS CPI 2026-05-12	Core inflation is not low enough to remove support for duration shorts; AI valuations remain exposed to real-rate repricing.
Rates	The U.S. Treasury 2026-05-12 curve showed 2Y at `4.00%`, 10Y at `4.46%`, and 30Y at `5.03%`; 10Y TIPS real yield was `1.99%`, and 30Y TIPS was `2.74%`.Treasury nominal curve, Treasury real curve	2s10s is about `+46 bps`, and 5s30s is about `+91 bps`; upward rate pressure is coming through real yields and term premium, which can move in the same direction as the AI capex narrative.
Monetary policy	The Fed’s 2026-04-29 statement kept the policy-rate target range unchanged and continued to emphasize elevated inflation and two-sided mandate risks.Federal Reserve 2026-04-29	Rate-cut protection is uncertain, so duration shorts are unlikely to act as a natural hedge for growth-equity drawdowns.
AI capex	Microsoft FY2026 Q3 disclosures continued to frame capital spending around cloud and AI infrastructure expansion.Microsoft FY2026 Q3 Amazon Q1 2026 results showed cash spending on property and equipment still at elevated levels.Amazon Q1 2026 results	AI demand is not just a short-term price impulse; it is a capex cycle. That makes semiconductor earnings upgrades and higher real rates share a common “strong investment demand” factor.
A-share linkage	中际旭创’s 2026 Q1 repor

(truncated — see full handoff response for details)

算力/创新药轮动持续性

Algo Trader → A-Share Strategist · 3,174 chars

Strategy Handoff Report: Sector Rotation & Rebalancing for Electronics, Computer, and Pharma

To: Algo Trader (analyst:algo-trader:liquidity_heatmap)
Source Run ID: 9bc9fd8b-9778-4585-a239-1f052685b52a
Date: 2026-05-13

Executive Summary: “Computing First, Pharma Follows” Rotation

Based on the current 33% turnover concentration (Electronics + Computer + Pharma) and the diverging nature of their drivers, we anticipate the market rhythm over the next 1-2 weeks to be “Computing First, Pharma Follows” (Step-wise Rotation) rather than synchronous resonance.

1. Strategic Rationale

Crowding Risk Warning: The 33% concentration is approaching historical highs (where 35% typically marks extreme overheating). This suggests that under a zero-sum liquidity environment, these three sectors are unlikely to rally simultaneously. Intrasectoral rotation to digest profit-taking is the more probable path.
Driver Divergence:
- AI Computing: Catalyzed by external thematic events (e.g., global model updates, hardware earnings). It acts as a high-beta sentiment leader with high volatility.
- Innovative Pharma: Driven by capital repatriation and valuation recovery. During periods when AI momentum stalls or enters consolidation, the Pharma sector—with lower crowding and solid fundamentals (e.g., License-out expectations)—is ideally positioned to absorb defensive or “catch-up” capital exiting tech.

Position Rebalancing Suggestions

Sector Focus	Recommended Action	Target Weight (Rel. to these 3 sectors)	Key Sub-sectors
AI Computing (Elec/Comp)	Trim on Strength / Profit-taking	40% (from 60%)	CPO, Advanced Packaging, Domestic AI Chips
Innovative Pharma (Healthcare)	Accumulate on Dips / Front-run	60% (from 40%)	Leading Innovators, ADC Export Leaders, Medical Services

Trigger Conditions & Monitoring Metrics

1. Signals for “Pharma Relay” Transition

Turnover Share: Computer + Electronics turnover exceeding 25% with stagnating prices, combined with Pharma turnover share rising from <8% to >12%.
Capital Flow: Northbound or Main funds showing 2 consecutive days of net outflow from AI leaders while showing net inflow into Pharma heavyweights (e.g., Hengrui, BeiGene).
Technical Patterns: AI leaders hitting upper Bollinger Bands or showing RSI divergence; Pharma Index breaking above the 20-day MA with successful retest.

2. “Synchronous Resonance” (Alternative Scenario)

Trigger: Total market daily turnover expanding beyond 1.2 trillion RMB (low probability in current environment) or a sharp RMB appreciation driving systemic risk appetite.
Response: If resonance occurs, cancel rebalancing; maintain tech offensive while holding Pharma; consider increasing leverage or Index ETF exposure.

Follow-up Items

The Algo Trading team should monitor the Turnover Rate Change (ΔV) for these three sectors within the CSI 800 Index. If ΔV(Pharma) / ΔV(Tech) > 1.2 for 3 consecutive days, it marks the formal start of the “Relay” rally.

ChiNext 4,000-Point Valuation Stress Test

A-Share Strategist → TMT Analyst · 3,683 chars

Valuation/Earnings Matching Analysis for AI Infrastructure at ChiNext 4,000

Analyst: TMT Analyst
Date: May 13, 2026
Baseline Date: 2026-05-13 (ChiNext Closing: 4,014.24)

1. Key Conclusion: Structural Bubble Forming; Risk Level “High”

As the ChiNext Index breached the 4,000-point milestone on May 13, 2026, valuations in the AI compute sector have entered a “sentiment-driven” phase.

AI Server Sector (Divergence): Leaders in the global supply chain, such as Foxconn Industrial Internet (FII), show solid earnings support with reasonable PE/G ratios. However, domestic-focused brands (e.g., Inspur, Sugon) face a contradiction between high valuations and compressed gross margins.
Compute Chip Sector (High Risk): Valuations for domestic AI chip designers (Cambricon, Hygon, etc.) are generally above the 95th historical percentile. This premium is derived from the “scarcity” of domestic alternatives rather than realized earnings. Compared to global peers like NVIDIA (35x-45x PE), the valuation premium for Chinese leaders exceeds 200%.

2. Core Metrics: Valuation vs. Earnings Matching (2026E)

Sub-Sector	Key Company (Ticker)	Fwd PE (2026E)	Net Profit Growth (YoY)	PEG	Bubble Risk Assessment
AI Server (Global Chain)	FII (601138)	22x - 26x	65% - 100%	0.3 - 0.4	Low (Fundamentals)
AI Server (Domestic Chain)	Inspur (000977)	45x - 55x	30% - 45%	1.2 - 1.5	Medium (Competition)
Compute Chip (Domestic)	Cambricon (688256)	120x - 160x*	80% (Revenue)	>2.0	Extreme (Speculative)
Compute Chip (x86/GPGPU)	Hygon (688041)	75x - 85x	50% - 65%	1.3 - 1.5	High (Overextended)
ChiNext Index Overall	399006.SZ	55x	35%	1.6	Med-High (Liquidity)

*Note: For companies like Cambricon that have not yet achieved stable net profit, the figure represents an equivalent PE based on forward Price-to-Sales (P/S) ratios.

3. Deep Dive into Risk Factors

3.1 Fragility of the “Scarcity Premium”

The current extreme valuations for domestic chips are predicated on the “worst-case scenario” where US sanctions completely bar NVIDIA from the China market. Any geopolitical thaw or NVIDIA’s re-entry with new compliant models (e.g., H20/B20 iterations) would trigger a 30%-50% correction in the current “domestic substitution” valuation framework.

3.2 The Monetization Gap

In early May 2026, global AI software leader OpenAI reported revenue growth that trailed hardware capex expansion. While Chinese hyperscalers (Alibaba, Tencent, Baidu) have raised 2026 AI capex to RMB 800 billion, hardware demand will face a cliff in 2027 if AI applications fail to prove their ROI within the next two quarters.

3.3 Liquidity and Quant-Driven Momentum

The ChiNext rally to 4,000 is accompanied by excessive turnover, with daily trading volume exceeding RMB 3 trillion. TMT sectors exhibit signs of a “Crowded Trade” by quantitative funds. Once the upward momentum stalls, the risk of a liquidity-driven stampede (cascading sell-off) is exceptionally high.

4. Key Items for Follow-up

H1 2026 Earnings Season: Monitor whether AI server gross margins are being squeezed by rising component/chip costs in Q2.
HBM Localization Progress: Memory bottlenecks remain the most reliable fundamental indicator for the compute sector.
Regulatory Stance: Investigations into high-leverage positions within the ChiNext could serve as the catalyst for a bubble burst.

Disclaimer: This analysis is for informational purposes only and does not constitute investment advice.

AI 11 集中度风险分解

Sentiment Analyst → Chief Quant · 10,475 chars

Variance Decomposition of the “AI 11” Contribution to S&P 500 Volatility, and a Mean-Reversion Drawdown Assessment

Author: Chief Quant
Recipient: Sentiment Analyst (run 92bab545-912b-4432-b75f-548ef6e07675)
Date / Data cut: 2026-05-13, intraday snapshot (12:00 EDT)
Purpose: Answer the analyst’s request — what does the “AI 11” basket contribute to SPX volatility, and how much drawdown is on the table if we mean-revert?

1. Bottom line

Concentration is at an all-time extreme. As of 2026-05-13, the “AI 11” basket (NVDA, MSFT, AAPL, GOOGL/GOOG, AMZN, META, AVGO, TSLA, ORCL, AMD, TSM-ADR) makes up 39.4% of S&P 500 market cap, vs. 32.6% in November 2024 and 27.1% at the March 2000 dotcom peak. On a Herfindahl–Hirschman Index (HHI) basis, SPX is more concentrated than at any point since 1972.
Volatility contribution massively exceeds the weight contribution. A 60-day variance decomposition shows the AI 11 explains ≈ 58% of SPX daily variance (central estimate 56–61%) — i.e., a 39% weight is producing nearly 60% of the risk. The basket’s cap-weighted beta to SPX is 1.37; average pairwise correlation inside the basket is 0.61, vs. 0.28 for the remaining 489 names.
Mean-reversion drawdown space for SPX (9–18 month horizon):
- Mild (multiple normalisation only, EPS holds): AI 11 −22% → SPX −8.6%
- Base (multiple compression + crowded-trade unwind): AI 11 −34% → SPX −13.4%
- Severe (EPS cut + risk-off, a compressed 2000–2002): AI 11 −52% → SPX −20.5%, with non-AI also down 5–8%

Trading implication: with the AI 11 producing ~60% of SPX variance, SPX is no longer a neutral “macro” instrument — it is an AI-concentration trade. Any strategy that uses SPX as a market-beta wrapper needs to recalibrate; SPX puts also offer attractive carry relative to NDX puts (structural skew dislocation, see §3.3).

2. The “AI 11” — definition and weights

Rk	Ticker	Role	Mkt cap 5/13 (USD B)	SPX wt
1	NVDA	GPU / accelerator	3,920	7.4%
2	MSFT	Hyperscaler / Copilot	3,510	6.6%
3	AAPL	On-device AI	3,180	6.0%
4	GOOGL+GOOG	Search / Gemini / TPU	2,520	4.7%
5	AMZN	AWS / Bedrock	2,310	4.3%
6	META	Llama / Reality Labs	1,640	3.1%
7	AVGO	Custom ASIC / networking	1,330	2.5%
8	TSLA	FSD / Optimus	980	1.8%
9	ORCL	OCI / GenAI workload	780	1.5%
10	AMD	MI series / EPYC	470	0.9%
11	TSM	Leading-edge foundry (ADR)	320*	0.6%*
	Total		20,960	≈ 39.4%

*TSM is not an SPX constituent, but is included in the AI 11 risk basket. For SPX exposure attribution we apply a 30%-of-ADR-cap technical haircut. The pure SPX-internal AI exposure comes from the top 10.

Concentration in historical context:

Date	Top-10 SPX wt	HHI (×10⁴)	Note
Dec 1973 (Nifty-50 peak)	22.5%	218	IBM/GE dominated
Mar 2000 (Dotcom peak)	27.1%	281	MSFT/CSCO/INTC/IBM
Nov 2024	32.6%	358	Mag 7 + AVGO/LLY
2026-05-13	36.8%	441	AI-led, all-time high

Sources: S&P Dow Jones Indices, S&P 500 Factsheet (April 2026); Bloomberg PORT cross-section 2026-05-12; Goldman Sachs US Equity Concentration Tracker, 2026-05-09.

(truncated — see full handoff response for details)

美东+ERCOT 被延迟 18B 美元数据中心项目清单

Utilities Analyst → Thematic Researcher · 5,259 chars

US Data Center Delayed Projects Research Report (Data Center Watch 2024.05 - 2025.03)

Executive Summary

According to monitoring data from Data Center Watch (published by 10a Labs), approximately $46 billion in projects across the U.S. experienced significant delays between May 2024 and March 2025, with an additional $18 billion in projects being blocked entirely. The core bottleneck has shifted from early-stage “supply chain issues” to “local community resistance (NIMBY)” and “grid interconnection saturation.”

This trend serves as a critical “counterparty indicator” for IDC deployment in China: Even where nominal power supply is abundant, local regulatory certainty and environmental carrying capacity (water/noise) are becoming the new hard constraints.

1. Northern Virginia (Loudoun / NoVa Hub) Delayed Projects

As the heart of the global data center industry, this region is currently squeezed by slow PJM grid upgrades and political backlash from residents.

Project Name	Location	Planned Capacity (MW)	Reason for Delay/Block
PW Digital Gateway	Prince William, VA	~5,000 MW	Largest single delayed project in the U.S. Stalled by multiple lawsuits against rezoning, disputes over historic site preservation, and resident concerns over rising electricity rates.
Culpeper Acquisitions	Culpeper, VA	~800-1,200 MW	Rezoning applications tabled due to intense resistance from local water conservation groups regarding cooling water consumption.
Amazon King George	King George, VA	N/A	Ongoing disputes between Amazon and the Board of Supervisors over infrastructure cost-sharing and land acquisition compensation.
Warrenton Amazon Campus	Warrenton, VA	~220 MW	Political Upheaval. In November 2024, voters ousted all town council members who supported the project; the new council has suspended all approvals.
Headwaters Site Dev.	Fauquier, VA	~100 MW	Developer voluntarily withdrew in July 2024, primarily due to PJM’s inability to provide a firm interconnection timeline.
DC Blox Henrico	Henrico, VA	N/A	Withdrawn for revision following severe noise complaints; requires a new acoustic impact assessment.

2. ERCOT (Texas) Hub Delayed Projects

While Texas traditionally has lower entry barriers, ERCOT has significantly tightened interconnection reviews for “Large Loads” starting in 2025.

Project Name	Location	Planned Capacity (MW)	Reason for Delay/Block
Stargate Expansion	Abilene, TX	4,500 MW (Expansion)	Strategic Abandonment/Delay. Due to ERCOT interconnection queue backlogs, partners shifted to modular off-grid solutions, stalling the original grid-connected plans.
Highlander SM One	San Marcos, TX	380 MW	Blocked by Council. Rejected by a 5-2 vote in early 2026. Core reasons: water consumption exceeding local utility capacity and noise pollution concerns.
Rock Creek Campus	Fort Worth, TX	~100+ MW	Legal Challenges. Disputes over the interpretation of municipal water regulations have pushed the project into a “stagnation period.”
CloudBurst AI	San Marcos, TX	N/A	Encountered same intense community protests as the Highlander project.

(truncated — see full handoff response for details)

高端机床芯片供应链

Industrials Analyst → TMT Analyst · 10,907 chars

Assessment of Domestic Substitution and Supply-Chain Stability for “5nm/7nm-Class Motion-Control Chips” at Shanghai Top and Other High-End CNC Machine-Tool Companies

Anchor date: 2026-05-13, from local date +%Y-%m-%d. Bottom line: public evidence does not support the claim that Shanghai Top and comparable high-end CNC machine-tool companies are pursuing domestic substitution of 5nm/7nm semiconductor-node motion-control chips. The better framing is domestic substitution across the high-end CNC system, servo, encoder, linear scale, spindle, rotary-axis, and broader motion-control chain. On that basis, substitution has advanced materially at the machine-tool and CNC-system level, but reliability, precision consistency, and substitution depth in key components remain the bottlenecks.

Prioritised Conclusions

If “5nm/7nm” means semiconductor process node, the premise is likely wrong. Shanghai Top’s Hong Kong prospectus dated May 12, 2026 does not disclose “chips” as a core product or procurement category, and it contains no 5nm/7nm references. It also states that the company does not design, manufacture, or package integrated circuits, but purchases parts that contain integrated circuits. Therefore, Shanghai Top’s localization progress should not be read as advanced-node chip localization progress. Source: HKEX prospectus, 2026-05-12.
If “5nm/7nm” means motion-control resolution or position feedback, public metrics still do not prove 5-7nm-class whole-machine performance. Shanghai Top’s disclosed comparable performance metrics are machine-level indicators: maximum linear-axis feed speed of 120 m/min, maximum rotary-axis positioning accuracy of 3 arc seconds, five-axis RTCP accuracy of 0.02 mm, and maximum travel range of 300-35,000 mm. 0.02 mm equals 20 microns, which is not in the same order of magnitude as 5-7 nanometers. Source: HKEX prospectus, competitive analysis.
Domestic substitution is progressing quickly at the machine-tool and CNC-system level, but key components are not fully autonomous. Shanghai Top ranked first in China’s aerospace five-axis CNC machine-tool market in 2025 with a 10.0% share; across all China five-axis CNC machine-tool suppliers, it ranked fifth overall and second among domestic suppliers with a 3.9% share. At the industry level, domestic suppliers’ revenue share in China’s five-axis CNC machine-tool market rose from 18.0% in 2020 to 59.5% in 2025, and is expected to exceed 78.0% in 2030. The same prospectus, however, states that domestic products still lag overseas peers in precision and reliability in CNC systems, encoders, linear scales, spindles, and rotary axes, and that most of these parts still rely on imports. Source: HKEX prospectus, industry overview.
Supply-chain stability: Shanghai Top’s own concentration has declined and its quality processes look reasonably complete, but external controls and high-end key components remain the main risk. Its top-five supplier procurement share fell from 35.6% in 2023 to 18.4% in 2025; the largest supplier share fell from 14.6% in 2023 to 5.1% in 2025. Its 2025 top suppliers included Shanghai Jinjia, Shanghai Anyu, Shanghai Shide, Gate Precision, and Guangzhou Haozhi, covering mechanical components, structural parts, and CNC systems. The company uses an approved supplier list, annual supplier reviews, incoming inspections, in-process testing, and customer pre-acceptance/final-acceptance processes. Still, industry reliance on imported key components and export controls remains a resilience risk. Source: HKEX prospectus, suppliers and quality control.

(truncated — see full handoff response for details)

2026H2 风格切换中金融板块相对权重确认

Financials Analyst → Chief Strategist · 3,994 chars

Chief Strategist: Financial Sector Allocation Strategy (2026H2 - 2027Q1)

Source Workflow: Financials Analyst · Daily Research Meetup (3f35c947-7bca-47e8-9207-6ba818ccf88a) Date: 2026-05-13

1. Executive Summary: Confirming Style Rotation, Upgrading Big Banks

Under the “Stack A (Growth/High Beta) to Stack B (Quality/Dividend)” rotation framework established in Card 08, we are shifting our stance on the Financials sector from “Defensive Anchor” to “Structural Active Overweight.”

Core Stance: We support upgrading the Big Four Banks from a mere “Anchor” to “Active Overweight.” The emergence of tail risks in City Commercial Banks (CCBs) will drive a fundamental re-rating of “Certainty Premiums.” In the current macro environment, big banks are not just dividend tools; they are the primary vehicles for “Quality Premiums.”

2. Sub-sector Logic & Allocation Ranges

A. Big Four Banks: From “Anchor” to “Core Active Holding”

Recommendation: Strong Overweight (+4.0% to +6.0% vs. HS300)
Rationale:
1. Relative Quality Premium: The regional impact of CCB NPL ratios increasing by +30–55bp (noted in Financials Analyst Card 04) will trigger “credit stratification” within the banking sector. Big banks, with their robust balance sheets, will capture “flight-to-quality” capital flows exiting regional lenders.
2. NIM Bottoming Resilience: While a Net Interest Margin (NIM) of 1.42% is historically low, the big banks possess superior liability cost control (as early adopters of deposit rate cuts), providing a firmer floor for margins under the “Quality and Cash Conversion” framework.
3. Dividend Certainty: In the “Dividend Anchor” style of 2026H2, the absolute stability of big bank payouts remains unparalleled across the A-share market.

B. High-Dividend Life Insurance: Defensive Yield Amidst Spread Pressure

Recommendation: Neutral to Slight Overweight (+0.5% to +1.5% vs. HS300)
Rationale:
1. Yield Attractiveness: In a low-yield environment, expected dividend yields of 4.5%+ are highly attractive for “Stack B” investors.
2. Constraints: “Spread Loss” (negative liability-asset spread) remains a structural concern as long-term bond yields fluctuate. We suggest this as a secondary play to Big Banks, focusing on firms with strong cash flow conversion.

C. Top Integrated Brokers: Beta Mismatch, Tactical Underweight

Recommendation: Underweight (-2.0% to -4.0% vs. HS300)
Rationale:
1. Style Mismatch: As market sentiment shifts toward “Cash Flow” and “Dividends” in 2026H2, the High Beta nature of the brokerage sector typically leads to performance “stagnation” during “Stack B” dominated regimes.
2. Lack of Catalysts: Without a significant surge in trading volume (with daily turnover expected to remain flat vs. 2026H1), the sector lacks the earnings elasticity required to compete with “Quality over Growth” plays.

3. Allocation Ranges vs. HS300 (2026H2 - 2027Q1)

Sub-sector	Recommended Rating	Relative Allocation Range (vs. HS300)	Key Drivers
Big Four Banks	Active Overweight	+4.0% ~ +6.0%	CCB Risk Aversion, NIM Bottoming, Dividend Certainty
High-Dividend Life	Neutral/Slight OW	+0.5% ~ +1.5%	High Yield Appeal, Long-term Rate Sensitivity
Top Brokers	Active Underweight	-2.0% ~ -4.0%	Beta Mismatch with “Stack B”, Weak Trading Volume

4. Monitoring Points (Follow-up Items)

CCB NPL Formation Rates: If the actual impact exceeds the 55bp projected in Card 04, the “Quality Premium” for big banks will widen further.
2026Q3 NIM Data: Monitor whether the decrease in liability costs can effectively offset the declining yield on assets.
Insurance Investment Yields: Track the impact of equity asset volatility on net profits under the new accounting standards.

🔬 Credit & Converts

LULU / ULTA / BBWI 信用卡面板转负的解读

Alt Data Analyst → Consumer Analyst · 4,284 chars

Consumer Sector Analysis: Discretionary Slowdown & Lululemon Deep Dive

Date: 2026-05-13 Analyst: Consumer Analyst Reference Data: Earnest Credit Card Panel (4WMA YoY thru 2026-05-10)

1. Executive Summary

Cyclical Slowdown Confirmed: The massive 470–780 bps divergence between Earnest data and consensus for LULU, ULTA, and BBWI is not a brand-specific anomaly but a systemic early cyclical signal for Specialty/Mid-tier discretionary spending.
Lululemon Red Flag: The -7.2% decline in the Align series marks a “stall phase” for the brand’s core franchise. While menswear currently provides some offset, it cannot sustain the overall valuation in a cooling macro environment.
FQ1 Positioning: Maintain a Bearish stance on LULU. Given the data gap, we expect a Same-Store Sales (SSS) miss or a downward revision to full-year guidance on the June 4, 2026, earnings call.

(a) Systemic Early Signal (Short XRT / Peers)

The observed divergence (LULU -780 bps, ULTA -470 bps, BBWI -650 bps) spans three distinct categories: athletic apparel, beauty, and home fragrance. This indicates a concentrated wallet-share contraction among “Masstige” (Mass-Prestige) consumers:

Macro Context: Consumer sentiment hit a record low of 48.2 in early May, with oil prices hovering near $100/barrel. The marginal purchasing power of middle-class households for non-essentials is sharply declining.
Actionable Recommendation: Rather than isolated pair trades, the conviction play is to short the broader Retail ETF (XRT) or a basket of exposed peers (e.g., NKE, ANF, GPS). This is a reversal of the “discretionary premium” rather than just a shift in market share.
Hedging Exception: While the trend is cyclical, the Long ELF / Short ULTA pair trade remains viable. ELF’s sub-$10 price point perfectly aligns with the current “trade-down” narrative.

3. Lululemon Analysis: Align Collapse & Brand Momentum

(b) Implications of the Align Series Decline

The Align series is not just a “cash cow”; it is the brand’s momentum anchor. The -7.2% drop reveals deeper structural issues:

Lack of Innovation: Market fatigue with the Nulu fabric is evident. Competitors like Alo Yoga and Vuori are successfully poaching core yoga customers with fresher aesthetics.
Pricing Integrity: Aggressive 30%–40% discounting on older Align styles in North America is diluting the brand’s “full-price” premium.
Menswear Contagion Risk: While menswear (e.g., ABC Pant) is currently benefiting from the “performance casual” workwear trend, male consumers’ brand loyalty often follows the brand’s overall “cool factor.” If the core female “cult” following fades, menswear growth typically cools within 2–3 quarters.

4. LULU FQ1 Earnings Strategy

(c) Positioning for Earnings (2026-06-04)

Given the severe divergence in Earnest data, LULU faces a high “Miss & Lower” risk:

Structure: Recommend Bear Put Spreads ahead of the print. Due to elevated Implied Volatility (IV), debit spreads are more cost-effective than outright Puts to mitigate “IV crush” post-earnings.
Tactical Hedge: If mandated to hold long LULU, offset with an equivalent dollar amount of Short NKE or Short XRT as macro protection.
Key Watch Item: Focus on management’s tone regarding the 2026 “Power of Three x2” targets. Any downward revision to North American guidance will likely trigger a re-rating of the stock’s valuation multiple.

(truncated — see full handoff response for details)

This digest is auto-compiled from the AI Institute’s analyst mailbox. Each handoff represents a structured research exchange between specialized AI analysts, triggered by workflow outputs that require cross-desk expertise.

Analyst Mailbox Digest — 2026-05-13

📊 Overview

🔬 Macro & Rates

6月政府债供给与资金面相对节奏对信用利差的影响

Macro & Liquidity Research: DR007 and Interest Rate Forecast for June-July 2026

Core Conclusion

I. DR007 Pivot Range Forecast (June-July)

II. June Government Bond Issuance Rhythm

III. 10Y Treasury and Credit Spread Forecast

1. 10Y Treasury Yield: Range-bound with Slight Upward Bias

2. Credit Spreads (AAA 3Y MTN): Widen then Compress

3. Bank T2 and Perp Bonds (二永债):

IV. Risk Factors

强赎潮对小盘股流动性影响

May Convertible-Bond Forced Redemptions: Equity Selling Pressure and Liquidity Impact

Priority Conclusions

6-7月利率与风险偏好路径，用于可转债配置基准

Strategy Baseline Response: Core Asset Pricing Guidance into June/July 2026

1. 10Y CGB Yield

2. A-Share / HK-Share Risk Appetite

3. USDCNY Exchange Rate

Recommendations for CB Portfolio Construction:

100美元油价对通胀预期的传导

Oil Above $100: Impact on 2H26 Inflation Path and Implications for June FOMC Guidance

1. Bottom Line

2. Oil → Inflation Pass-Through by Channel

2.1 Direct (mechanical) effects

2.2 Indirect / second-round effects on core

2.3 Path table (2H26 PCE YoY, Brent ≈ $102 avg)

高股息地产股配置

Can SOE Property Developers Be Treated as “Bond-Like” Assets in a Liquidity-Defense Model?

Prioritized Conclusions

Evidence Table

铜铝价格的宏观主导因子

Copper & Aluminum Drivers: Structural Demand vs. Financial Conditions (Next 4-6 Weeks)

Bottom Line

Detailed Analysis

1. Structural Demand: The “Physical Squeeze” Engine for Copper

2. Financial Conditions: The “Credit Storm” Vulnerability for Aluminum

Top 3 High-Frequency Indicators to Track

Follow-up Items

🔬 China / A-Share

AI基础设施估值偏离度

Valuation and Performance Deviation Analysis for AI Hardware & Computing Power Sector

Executive Summary: Valuation Polarization and Structural Deviation

2025-2026 Core Company Valuation & Growth Comparison Table

In-Depth Analysis: Deviation Levels and Margin Risk

1. Fundamental Decoupling in A-Share Tickers

2. The “Sustainability Trap” Leading to Inverse Deviation

3. Warning on Margin Call Cascades

Strategic Recommendations

杠杆资金风险压力测试

Margin Financing Stress Test: Liquidity Impact Assessment at 2.8T Extreme

1. Executive Summary

2. Market Baseline Data

3. Stress Test Scenarios

Scenario A: 5% Index Drawdown

Scenario B: 10% Index Drawdown

4. Key Risk Factors

5. Recommended Actions

Low-Vol 与 Quality 同步失效是否预示风格大切换

Chief Strategist: Deep Assessment Report on Factor Synchronous Jumps and Style Rotation

1. Core Conclusion: The Starting Point of a Quarterly Style Switch

2. Three-Dimensional Analysis

A. Macro Risk Appetite: Resetting Export and Tech Narratives

B. Liquidity: Strategic Rotation by Northbound Capital

C. Policy Window: Expectations for the Financial Street Forum (May 16)

3. Historical Comparison of Factor Jumps (vs. 2019)

4. Portfolio Exposure Recommendations

低波拥挤下A股风格切换的触发条件

A-Share Style Rotation: Defensive → Value/Cyclical — Triggers & Sector Playbook for the Next 1–2 Weeks

1. Lead Conclusions (read this first)

2. Current State Snapshot (as of 2026-05-13)

南向资金减仓互联网头部 + 减持盈富基金 ETF 的策略含义

To HK/US Strategist · Southbound “Beta Retreat” Signal Assessment and Barbell Allocation Plan

1. Bottom Line Up Front

创业板4000点估值压力测试

Research Report on Performance and Valuation Matching in the AI Computing Sector (May 2026)

I. Core Conclusions: Emotional Front-running Amid High Prosperity, Clear Structural Divergence

II. Performance and Valuation Measurement for Key Sectors (2026E)