China Stocks Enter the Rotation Lure: A-share Hard Tech Still Leads, but H-share Internet Can Begin to Be Built Opportunistically

Overview

This report is a China Musings strategy/macro market commentary from Goldman Sachs centered on frequent questions raised at recent investor meetings in Asia and the Americas: divergence between A-shares and H-shares, hard-tech versus soft-tech rotation, pressure on internet profitability and conditions for recovery, whether Chinese AI is overheating, foreign capital flows and North Asia comparison, Hong Kong IPO supply, macro factors being displaced by the AI theme, and how to generate returns in China equities in 2H26. The core view is that A-shares and hard-tech still hold tactical advantages, while large-cap H-share internet names have become selectively investable after substantial de-rating, conditional on profit recovery delivering over the next several months to quarters.

Core views

Goldman argues that internal dispersion across Chinese stocks has reached historically extreme or rare levels, and the trade logic between AI infrastructure suppliers/hardware and AI capital expenditure participants/software platforms is now visible in China. Short-term preference still leans to China A-shares, especially sectors benefiting most from profit momentum and the AI hard-tech theme. But after H-share internet and soft-tech have materially underperformed, valuations may already reflect pessimistic scenarios of continuing de-rating in core businesses and AI return-on-investment; if profits improve in 2Q or 3Q, the market could shift from mixed earnings and valuation views to a business-segment additive valuation framework. Chinese AI is not viewed as a whole as a bubble, but there are valuation, concentration, and leverage risks in semiconductors and selected A-share hard-tech proxies. Outside AI, materials, capital goods, insurance, as well as shipbuilding, consumption, healthcare/biotech, housing, brokers, and the hog cycle may also offer alpha opportunities.

Analysis framework

The report uses investor feedback and FAQ synthesis, cross-market and cross-index relative return comparisons, A-H market rotation and hard-tech versus soft-tech rotation models, profit momentum and valuation analysis, cash-flow and IPO supply assessment, and top-down AI thematic market-size estimates to form a tactical view on China equity positioning for 2H26.

Methodology notes

• Market rotationA-H Market Rotation Model and Hard vs. Soft Tech Rotation Model

  Use fundamental, valuation, and liquidity factors to assess directional relative performance between A-shares and H-shares and between hard-tech and soft-tech.

  The model suggests H-share soft-tech may begin to improve relative to A-share hard-tech over the coming months, but sustained upside still requires confirmation from underlying profit recovery.

• Thematic investingChina Select AI Portfolio

  A theme portfolio covering the Chinese AI value chain.

  The portfolio includes 50 stocks across 22 AI sub-industries and is designed to give investors balanced exposure to the full Chinese AI value chain.

• Valuation and fundamentalsAI Economic Gains versus Market-Implied Return

  Compare AI-driven efficiency gains, incremental profits, and TAM with current earnings implied by AI stock prices.

  Goldman believes Chinese AI’s aggregate economic gains may be 50% to 100% above current AI stock discount rates, so it does not view Chinese AI equities as broadly in a bubble, but emphasizes localized overheating risks.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Chinese A-shares

  tactical preference/maintain overweight

  Strengths

  More favorable profit momentum, with international investors still underappreciating its diversification value, and stronger concentration in hard-tech and policy-support themes.

  Weaknesses

  Valuations of some A-share hard-tech and semiconductor proxy names have become elevated, with rising concentration and leverage risk.

  Comparison

  Still Goldman’s preferred market relative to H-shares, though soft-tech rotation could allow periodic H-share improvement.

  Risks

  Crowded AI positioning, excessive valuation premium, profit realization below expectations.

• Large-cap H-share internet/soft tech

  begin gradual allocation

  Strengths

  Clear de-rating year-to-date; valuations may already reflect pessimistic scenarios for core-business and AI investment returns, and 2Q or 3Q may bring operating profit recovery.

  Weaknesses

  MSCI China profit pressure is still largely driven by internet, and AI capital expenditures and subsidy losses continue to weigh on profits.

  Comparison

  Still behind A-share hard-tech, but the gap may narrow over the next several months.

  Risks

  Delayed profit recovery, AI capex ROI below expectations, sustained platform competition, and ongoing subsidy pressure.

• Chinese AI stocks

  high-conviction thematic allocation

  Strengths

  China represents 11% of global AI-related market capitalization and 18% of revenue, has relative strength in power, infrastructure, and physical AI layers, and AI economic gains may still be underpriced.

  Weaknesses

  Underweighted by international investors, with some links constrained by supply-chain bottlenecks and technical bottlenecks.

  Comparison

  Compared with Korean and Taiwanese AI hardware chains, Chinese AI exposure is more diversified and less closely watched.

  Risks

  Localized overheating, elevated valuations, rising concentration, leverage expansion, and AI commercialization benefits below expectations.

• Hong Kong IPOs

  alpha source but monitor supply closely

  Strengths

  Median one-month and three-month post-listing returns of 32% and 30% year-to-date make it an important alpha channel in a weak-beta environment.

  Weaknesses

  An active and prolonged IPO pipeline raises liquidity-drain concerns.

  Comparison

  Goldman expects remaining issuance to remain manageable relative to listed-company dividends and buybacks.

  Risks

  Supply of issuance or secondary offerings coming in higher than expected, weakening secondary-market liquidity, and declining IPO returns.

• South Korea and Taiwan stocks

  North Asia relative reference

  Strengths

  Strong year-to-date performance, supported by storage, contract manufacturing, and AI infrastructure.

  Weaknesses

  Directional correlation to China stocks is near historic lows, with fund flows showing more caution toward China beta risk.

  Comparison

  Hedge funds hold net risk exposure in EM Asia excluding China at historical highs, while net exposure to Chinese stocks is near the lower range.

  Risks

  If the global AI hardware cycle turns down, North Asia outperformance could reverse and shift regional risk appetite.

Key data

• STAR50 excess return versus HSTECH year-to-date68 percentage pointsSTAR50 is used as a proxy for onshore AI hard-tech, while HSTECH is mostly offshore internet/platform companies.
• ChiNext outperformance versus SSE Composite/CSI 300 year-to-date19 percentage points/17 percentage pointsShows that growth and technology-style exposure has clearly outperformed within Chinese equities.
• Recent HSTECH performanceUp 11% over the past two weeks, still down 14% year-to-dateShows that H-share soft-tech is rebounding after deep underperformance.
• MSCI China Q1 profit changedown 8% year-over-yearThe internet sector was the main drag; internet accounted for roughly 35% of index profit weight.
• Internet sector subsidy lossesCumulative losses since 2Q25 exceed RMB 180 billionSubsidy losses are a major source of pressure on internet profitability.
• Chinese AI economic gains versus current price-implied levelspotentially 50% to 100% higherSupports the view that Chinese AI is not a bubble overall.
• Share of Chinese AI in global AI stocks11% by market cap, 18% by revenueThe report says China is an important component of the global AI stock universe, though still underweighted by international investors.
• Number of Hong Kong IPOs and proceeds year-to-date100 companies, $35 billionMedian one-month and three-month post-listing returns were 32% and 30%, respectively.
• Expected remaining issuance volume this yearIPO $25 billion; follow-on offerings $45 billionGoldman believes supply pressure is manageable relative to listed-company cash returns.
• Listed-company cash returnsFY2025 around $500 billion, FY2026 around $560 billionIncludes dividends and buybacks, used for comparison against new issuance and refinancing supply pressure.
• US-China Relations Barometer39Near lows since April 2022, suggesting lower pricing for bilateral tension in the market.

Impact & implications

For portfolio construction, the report suggests avoiding simply chasing AI hardware and semiconductors that have already risen substantially. Instead, while maintaining core allocation to A-shares and hard-tech, investors should gradually build selective exposure to de-rated large-cap H-share internet names, using profit realization as the signal to add weight. The Chinese AI theme remains the core alpha source, but risk-adjusted returns should be improved through sector diversification, lower concentration, and a tilt toward profit-driven exposure. Macro and geopolitical factors are temporarily not the dominant pricing drivers; investors should focus more on AI, policy support, profit recovery, IPO supply, and changing market-capital flow structure.

Risks

• Profit recovery in H-share internet may still require several months or quarters; if 2Q or 3Q results fail to materialize, valuation repricing could stall.
• Some semiconductor and A-share hard-tech proxy names are showing rising valuation, concentration, and leverage risk, indicating localized overheating.
• AI capex for internet companies may exceed $100 billion/$120 billion; if ROI disappoints, it will continue to pressure profits and valuations.
• If Hong Kong IPO and follow-on issuance supply exceeds expectations, it could intensify liquidity concerns.
• Macroeconomic variables including weak consumption, the path toward real estate stabilization, RMB dynamics, exports, and policy stimulus pace may again become market-defining factors.
• Foreign investors are more market-neutral and alpha-oriented; without profit confirmation, beta recovery in Chinese stocks may be limited.

What to watch

• 2Q26 and 3Q26 operating profits at large Chinese internet companies, narrowing subsidy losses, and AI-related new business growth.
• Whether H-share soft-tech relative returns versus A-share hard-tech continue to improve.
• Valuations, trading concentration, and leverage metrics for Chinese AI hard-tech, semiconductors, and AI infrastructure names.
• The scale of Hong Kong IPO and follow-on issuance, cornerstone investor participation, and 1-to-3-month post-listing performance.
• Net and gross exposures and alpha/beta preference shifts of EM and hedge funds toward Chinese equities.
• Potential policy stimulus from the July Politburo meeting, signs of real-estate stabilization, and RMB and export performance.
• Progress in the Chinese AI ecosystem across EDA, lithography, advanced packaging, HBM, LLMs, robotics, and AI tokens.