Goldman Sachs suggests Chinese equities are at a 'rotation temptation' juncture, with A-share hard tech remaining the preferred tactical play, while H-share internet stocks may see a gradual earnings recovery.
The primary trend in Chinese equities this year has been clear: A-shares focused on hard technology have been strong, while H-shares, particularly internet platforms, have been weak. The STAR 50 Index, representing onshore AI hard tech, has outperformed the HSTECH Index, dominated by offshore internet platforms, by 68 percentage points year-to-date. The ChiNext Index has outperformed the Shanghai Composite and the CSI 300 by 19 and 17 percentage points, respectively. This performance gap has reached extreme levels in the history of the Chinese stock market.
Goldman Sachs Asia portfolio strategists, including Kinger Lau, wrote in a China portfolio strategy research note dated July 13: "We reiterate our broad tactical preference for A-shares over H-shares and hard tech over soft tech, but we will begin to gradually allocate to some H-share soft tech/internet names after their valuation derating, anticipating an earnings recovery in the coming months."
This is not a simple binary choice between A-shares and H-shares. The core distinction in the framework is that hard tech is closer to the supply side of AI infrastructure, while soft tech and internet platforms resemble the entities undertaking AI capital expenditures. The former benefits from demand for computing power, semiconductors, equipment, and infrastructure. The latter must first prove that AI investments and subsidy spending will not continue to erode profits.
Earnings are an unavoidable hurdle for H-share internet stocks. MSCI China's first-quarter profits fell 8% year-on-year, with the internet sector being a significant drag. This sector accounts for about 35% of the index's earnings weight and has accumulated over RMB 180 billion in subsidy losses since Q2 2025, while also being a major absorber of domestic AI capital expenditures. Related AI capital expenditure this year and next could exceed $100 billion and $120 billion, respectively.
A-Shares Hard Tech Remains the Main Theme, But Rotation Pressure is Emerging
Performance divergence in Asian equities this year is not limited to mainland China. Markets in South Korea and Taiwan, driven by memory, foundries, and AI infrastructure, have seen gains that have attracted international investor attention. The divergence within mainland China is more subtle: the gap between A-share hard tech and H-share internet is larger than many overseas investors anticipated.
The HSTECH Index, dominated by offshore internet platforms, has rebounded 11% over the past two weeks but is still down 14% year-to-date. This rebound stems from several factors: improved sentiment around some internet leaders, depressed starting valuations, and continued progress in AI application and commercialization.
The model suggests that the underperformance of H-share soft tech relative to A-share hard tech may narrow over the coming months. Variables such as valuation, earnings expectations, and liquidity support this view.
However, a sustained rebound is missing one piece: profits. Valuations for some leading internet companies may already be pricing in a pessimistic scenario of continued de-rating for core businesses and negative net present value or return on investment for AI investments. For the market to re-rate these stocks, it still needs to see a recovery in underlying profits, which may take several months or even quarters.
Without Profits, H-Shares Struggle for Sustained Gains
MSCI China's recent performance has been weighed down by earnings. Six weeks ago, its rating was downgraded from overweight to neutral, with one key input being poor earnings delivery.
In 2025, index earnings grew only 7%, marking the sixth consecutive year of missing consensus expectations. By Q1 2026, earnings pressure had not eased but was instead amplified by subsidies and AI investments in the internet sector.
Subsequent profit recovery hinges on three main factors.
First, whether subsidy losses can narrow. Over the past year, internet platforms have made significant investments in areas like quick commerce, pressuring their income statements.
Second, whether new AI-related opportunities, including cloud, Agentic AI, and AI tokens, can contribute revenue.
Third, whether traditional e-commerce businesses can maintain stable cash flow. Only if these businesses continue to generate cash will AI investments not be viewed purely as a profit black hole by the market.
If operating profit inflections appear in Q2 or Q3 results, the valuation approach for internet companies may change. The market might shift from using blended earnings discounts to employing sum-of-the-parts valuations to assess core businesses, AI investments, and new business options.
Chinese AI is Not a Broad Bubble, But Some Trades Are Crowded
The answer to whether "A-share AI is overheated" is not a simple yes or no.
Since the 'DeepSeek moment', the market capitalization of China's AI-related stocks has risen. However, top-down estimates suggest the economic benefits from AI—through efficiency gains, new profits, and potential market expansion—could be 50% to 100% higher than levels reflected in current stock prices. This forms the basis for the view that "Chinese AI stocks as a whole are not considered a bubble."
The risk lies in specific pockets.
Valuations for the semiconductor sector and some A-share hard tech names are already at relatively high levels compared to their own history and global peers. Concurrently, concentration and leverage risks are rising. This means continuing with the AI theme is possible, but investors can no longer treat all AI assets as the same trade.
A more suitable approach is to control single-direction exposure, diversify across different parts of the AI value chain, and place a higher priority on profit realization capability. Pure thematic trading opportunities still exist, but the risk-reward profile is not as clear as at the start of the year.
Foreign Capital Has Not Left Mainland China, But Prefers Alpha Over Beta
A common question is whether the AI rally in South Korea and Taiwan has diverted funds from mainland Chinese stocks.
The data presents a more complex picture. The directional return correlation between mainland Chinese stocks and the South Korean and Taiwanese markets is near their respective historical lows. This indicates investors are not simply treating the three as the same "North Asia AI trade."
Hedge funds' net exposure to emerging Asia excluding mainland China is near historical highs, while their net exposure to mainland Chinese stocks is at the bottom of its range. However, total exposure is near cycle highs, suggesting capital has not fully withdrawn but is leaning more towards market-neutral and stock-specific alpha strategies.
Mutual fund allocation tells another story. Especially for emerging market funds, allocation to mainland China has risen to multi-year highs and shows a mild overweight for the first time since records began in 2011. This change may be related to China's declining weight in emerging market benchmarks.
Foreign participation is also visible in Hong Kong IPOs. Year-to-date participation by foreign cornerstone investors is near 2021 highs. Capital is still present but is not rushing to buy index beta, preferring to seek structural opportunities.
Hong Kong IPOs Are Hot, But 'Liquidity Drain' Fears May Be Overstated
Hong Kong IPOs have become an important channel for investors to capture alpha this year.
Year-to-date, 100 companies have listed on the Hong Kong Exchange, raising a total of $35 billion. The median stock price returns one month and three months post-listing are 32% and 30%, respectively.
Better-performing new listings often share common traits: larger issue size, listing as independent H-shares, and a moderate cornerstone investor shareholding ratio. A cornerstone ratio of 30% to 50% has strong explanatory power for subsequent performance.
The market worries that too many IPOs will drain liquidity. Current estimates suggest Hong Kong IPO volume for the remainder of the year is around $25 billion; including follow-on offerings, the total could be about $45 billion. In comparison, listed companies are returning significantly more cash to shareholders via dividends and buybacks—approximately $500 billion and $560 billion for fiscal 2025 and 2026, respectively.
A-share IPOs are also accelerating. Chinese DRAM maker ChangXin Memory Technologies is reportedly planning a STAR Market listing, aiming to raise $4.3 billion. If completed, it would be the largest A-share IPO since May 2022.
AI Overshadows Macro, But Macro Has Not Disappeared
The recent focus of investor discussions has clearly shifted towards AI. Questions surrounding China's AI ecosystem are more specific: progress in building an independent ecosystem, bottlenecks in EDA, lithography, advanced packaging, HBM, cost-efficiency differences in AI capital expenditure between China and the US, and competitive pressure from China on existing global players in areas like large models, robotics, and AI tokens.
China is already a significant part of the global AI equity pool. Chinese companies account for about 11% of global AI-related market capitalization and 18% of revenue. However, international investor allocation to Chinese AI stocks remains low, especially in power, infrastructure, and physical AI.
Under the OpenRouter metric, token usage for Chinese AI models is approaching half of the global total. This means China's AI story is not just about the "supply chain" but has already entered the application and consumption volume phase.
Macro topics have temporarily receded but remain on the investor checklist. Discussions include: whether the property market will stabilize by 2027, China's economic resilience amid global oil price shocks, persistently weak domestic consumption, and China's export performance.
Second Half Outlook: A-Shares Remain Preferred, H-Share Internet Warrants Gradual Observation
For the second half of 2026, the tactical market preference remains with A-shares. Two main reasons are cited: first, cyclical earnings momentum is more favorable; second, for international investors, A-shares still offer undervalued diversification value.
However, large-cap H-share internet stocks have entered an observation zone. After significant valuation derating this year, if profit recovery materializes over the next few quarters, these stocks could shift from a "value trap" to an "earnings recovery trade."
At the sector level, materials, capital goods, and insurance are listed as overweight directions. Beyond AI, several non-AI micro themes are worth watching: shipbuilding, new consumption, healthcare/biotech, the extended real estate industry chain, securities firms, and the hog cycle.
Comments