Goldman Sachs maintains Overweight on A-shares and downgrades H-shares to Market Weight, emphasizing stock-picking themes over index beta

Overview

This is a China equity portfolio strategy report. Goldman Sachs believes investors should not simply buy the index in the current market, but instead seek structural opportunities around earnings delivery, policy support, and different layers of the AI value chain. The report maintains an Overweight view on A-shares, but downgrades H-shares and MSCI China to Market Weight due to delayed internet earnings recovery and the rising opportunity cost of offshore China equities.

Core views

The core views include: first, return dispersion within Chinese equities is significant, with CSI300, ChiNext, and the hard-tech AI chain clearly outperforming MSCI China, HSTECH, and soft-tech platform stocks; second, earnings delivery is the dominant factor behind global equity returns, and China’s offshore market has lagged because profits of large-cap internet stocks are under pressure; third, A-shares benefit from improving earnings, liquidity, diversification value, and hard-tech exposure; fourth, AI remains a key long-term theme, but investors should differentiate among power, infrastructure, semiconductors, models, applications, and other layers; fifth, the offshore market is better approached through selected alpha themes rather than a simple bet on index beta.

Analysis framework

The report uses top-down regional and index comparisons, earnings forecast revisions, valuation target calculations, an A-H rotation model, thematic portfolio comparisons, and layered attribution of the China AI equity universe. Its explanatory framework attributes the year-to-date return dispersion to four factors: index composition, differences between hard tech and soft tech, direction of policy support, and earnings growth delivery; it also uses multivariate regression to explain part of the excess returns of individual Chinese AI stocks.

Methodology notes

• Strategy frameworkAlpha over Beta

  When index performance is weak but stock and theme dispersion is significant, excess returns can be generated through selected themes and stocks rather than simply holding broad-based indices.

  The report argues that the offshore market still contains abundant alpha opportunities, especially in selected AI portfolios, the 15th Five-Year Plan portfolio, and shareholder return themes.

• Attribution frameworkFour-factor return dispersion framework

  Index composition, differences between hard tech and soft tech, policy support, and earnings growth delivery jointly explain return differences within Chinese equities and the AI value chain.

  The report says these factors together explain about 40% of the year-to-date excess returns of individual stocks in the China AI equity universe.

• Valuation and earningsEPS growth and fair P/E target method

  Index targets and potential upside are jointly derived using earnings growth forecasts and fair valuation multiples.

  The report lowers MSCI China FY26/FY27 EPS growth to 8%/12%, cuts fair P/E to 12x, and derives a target of 85 and 11% potential upside over 12 months.

• Thematic layeringGlobal AI five-layer thematic mapping

  More than 3,700 global AI companies are mapped into 29 industries and 5 thematic layers to compare valuations, growth, and market performance across different AI segments.

  Based on this, the report distinguishes among power, infrastructure, semiconductors, models, applications, and physical AI, and emphasizes that China’s opportunities are more concentrated in hard tech and manufacturing segments.

Asset mapping & comparison

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

• China A-shares

  Maintain Overweight; this is the report’s most preferred China market beta exposure.

  Strengths

  Improving earnings growth momentum, favorable macro and market liquidity, diversification value for international investors not yet fully priced in, and relatively high exposure to proxy assets for AI hard tech.

  Weaknesses

  Still dependent on sustained earnings delivery and policy support; small- and mid-cap long strategies are also affected by the value of financing and securities lending inventory.

  Comparison

  Year to date, A-shares have performed better than H-shares and some regional alternative markets in both absolute returns and Sharpe ratio.

  Risks

  If earnings recovery or liquidity conditions are weaker than expected, the Overweight case for A-shares could be undermined.

• H-shares and MSCI China

  Downgraded from Overweight to Market Weight; better approached through selected alpha themes.

  Strengths

  Valuation and mean reversion may still narrow the A-H performance gap in 2H26; the MSCI China target still offers 11% potential upside.

  Weaknesses

  Weak earnings from large-cap internet stocks, prolonged subsidy losses, and slower-than-expected AI monetization have delayed the earnings recovery path by 3 to 6 months.

  Comparison

  Compared with other North Asian markets, H-shares currently lack sufficient earnings growth momentum and visibility, increasing the opportunity cost of maintaining an Overweight stance.

  Risks

  If the earnings trajectory does not materialize around 3Q26, valuation re-rating and relative performance recovery may continue to be delayed.

• AI hard tech

  Strategically favored, especially in power, infrastructure, semiconductors, and physical AI.

  Strengths

  Benefiting from strong cyclical growth, policy support, technological self-reliance goals, and China’s competitive advantages in advanced manufacturing segments.

  Weaknesses

  Some semiconductor assets have already been significantly re-rated, with valuations looking high relative to international peers and long-term addressable market potential.

  Comparison

  Hard tech has significantly outperformed soft tech, and AI hard tech contributed the main part of market cap gains in Chinese AI stocks.

  Risks

  If capex, policy support, or downstream demand slow, high-valuation hard-tech segments may come under pressure.

• Internet platforms and AI applications

  Cautious in the short term, waiting for earnings recovery and narrower subsidy losses.

  Strengths

  E-commerce businesses still generate cash flow, new opportunities such as cloud and agentic AI are growing rapidly, and compressed valuations provide a basis for rebound.

  Weaknesses

  Subsidy losses in food delivery and instant retail are deeper and more prolonged, while AI monetization is slower than expected, weighing on short-term profit growth.

  Comparison

  Soft tech and the application layer are clearly lagging AI infrastructure and other hard-tech segments.

  Risks

  Continued subsidy competition, marginal regulatory tightening, or failure of new businesses to contribute profits quickly would all drag on re-rating.

• Selected AI portfolios, the 15th Five-Year Plan portfolio, and shareholder return themes

  The three alpha themes most favored by the report.

  Strengths

  Selected AI portfolios provide balanced exposure across the extended AI value chain; the 15th Five-Year Plan portfolio benefits from structural policy tailwinds; shareholder return themes deliver strong risk-adjusted returns.

  Weaknesses

  Theme performance still depends on market risk appetite, policy delivery, and stock-level earnings quality.

  Comparison

  Compared with simply holding offshore indices, these themes better align with the report’s principle of alpha over beta.

  Risks

  If themes become crowded, policy implementation disappoints, or earnings fall short, excess returns may retrace.

• Media and capital goods

  Sector allocation shifts media to neutral and upgrades capital goods.

  Strengths

  Capital goods better align with the report’s preferred hard-tech and manufacturing-chain direction.

  Weaknesses

  Media and some soft-tech areas are more affected by earnings visibility and regulatory expectations.

  Comparison

  Capital is rotating from media into capital goods, reflecting a portfolio tilt toward hard tech and policy-supported directions.

  Risks

  If soft-tech earnings reverse faster than expected or capital goods valuations expand too quickly, sector rotation may reverse.

Key data

• MSCI China year-to-date performanceMSCI China -8%, MXAPJ +27%The relative underperformance marks the largest 5-month performance gap since 2002.
• MSCI China relative to MXAPJUnderperformed by 35 percentage pointsThe report attributes this to insufficient earnings delivery and the drag from index composition.
• CSI300 relative to MSCI ChinaOutperformed by 13 percentage pointsThe A-share index has outperformed the offshore China index.
• ChiNext relative to HSTECHOutperformed by 40 percentage pointsThis reflects significant divergence between hard tech and soft tech.
• Increase in Chinese AI stock market capIncreased by US$3.8tn since the DeepSeek momentOf this, US$3.5tn came from existing companies and US$0.3tn from new IPOs.
• Hard tech contributionAI hard tech contributed 85% of market cap gainsInfrastructure generated 65 percentage points of alpha relative to applications.
• MSCI China EPS forecastFY26/FY27 cut from 12%/14% to 8%/12%This mainly reflects softer FY25 and 1Q26 earnings, as well as a lower slope of recovery in the internet sector.
• A-share EPS forecastRaised from 16% to 20% for 2026Goldman Sachs raised its forecast to reflect commodity prices, export demand, and a lower base.
• Internet subsidy lossesRmb34bn in 1Q26, cumulative Rmb180bn since 2Q25Subsidy losses in food delivery and instant retail are deeper and more persistent than previously expected.
• Valuation assumptionMSCI China fair P/E cut from 13x to 12xHigher earnings risk has led to multiple compression.
• Index targetsMSCI China target 85; CSI300 target 5,500The MSCI China target implies 11% potential upside over 12 months.
• China AI global shareAI-related market cap/revenue about 10%/16%The report believes Chinese AI stocks remain underowned by international investors.

Impact & implications

The investment implication is to reduce reliance on broad offshore China beta and shift toward A-shares, AI hard-tech value chains with higher earnings visibility, and selected thematic portfolios. If internet subsidy losses narrow, new AI-related businesses grow, and the earnings trajectory materializes around 3Q26, the relative performance of H-shares and MSCI China could improve; before then, the stronger earnings cycle in other North Asian markets raises the opportunity cost of staying overweight H-shares.

Risks

• China corporate earnings recovery is slower than expected, especially if profits of large offshore internet stocks continue to come in below consensus.
• Subsidy losses in food delivery and instant retail continue to widen or narrow more slowly than expected.
• AI monetization progress is insufficient to offset related capex and investment pressure.
• Valuations of hard-tech assets such as semiconductors are already high; if growth or policy support falls short of expectations, risk-reward may deteriorate.
• Stronger earnings momentum in other North Asian markets may continue to raise the opportunity cost of holding H-shares and MSCI China.
• The timing of A-H rotation and mean reversion is uncertain, and realization may have to wait until the 3Q26 earnings trajectory becomes clearer.
• Basket trades are constrained by market liquidity and stock borrowing conditions, so actual tradability may be lower than the theoretical allocation recommendation.

What to watch

• Whether the earnings growth trajectory of MSCI China and H-shares materializes around 3Q26.
• The upcoming earnings season in mid-to-late August, especially profit performance of internet and AI application companies.
• Whether subsidy losses in food delivery and instant retail begin to narrow.
• Whether NBS industrial profits, PPI, tier-one city property activity, and export demand can continue to support A-share earnings.
• Whether AI investment demand, supply tightness, and price re-inflation can persist.
• The direction of policy support under the 15th Five-Year Plan for hardware, semiconductors, models, and advanced manufacturing.
• Whether the A-H rotation model, valuation gaps, and regional fund flows can drive relative performance recovery in 2H26.