Goldman Sachs Identifies Ten Signals to Gauge the Current Sentiment of China's AI Sector

TL;DR
· Goldman Sachs has outlined ten signals to assess the progress of the correction in China's AI hard technology sector, considering dimensions such as price, valuation, leverage, investor positioning, policy, and fundamentals.
· Style reversal, reduced retail investor enthusiasm, increased buybacks, and net buying by the 'national team' indicate that some of the overheated sentiment has been released.
· However, trading volume, financing, and institutional positioning in A-shares remain highly concentrated, and the momentum for upward revisions in hard technology earnings forecasts is also beginning to slow, meaning that a correction cannot be directly equated with a bottoming out.

After a rapid rise in the first half of the year, China's AI hard technology is entering a phase of emotional reassessment.

Goldman Sachs' China strategy report released on August 9 shows that the average increase of AI hard technology representative stocks in its defined China AI stock pool was 33% in the first half of the year. With the global AI hardware trading reversing, the STAR Market, ChiNext index, and CSI 1000 have all corrected more than 20% from recent highs.

The question is whether this round of correction is merely a short-term profit-taking or signifies a deeper change in AI trading.

Goldman Sachs did not only look at the index decline but assessed the sentiment and correction cycle of China's AI hard technology through ten dimensions: earnings differentiation, market concentration, valuation, leverage, retail sentiment, institutional positioning, corporate behavior, policy, 'national team' funds, and earnings forecasts.

The ten signals point to a moderately non-extreme answer: the speculative positions, excessive valuations, and leverage that had accumulated earlier have released a considerable portion, but some concentrated risk still exists, and the momentum for fundamental upward revisions is also slowing. The most crowded phase of the market may have passed, but it cannot yet be confirmed that the adjustment is completely over.

The ten indicators compiled by Goldman Sachs show that the market breadth and concentration risk of China's AI hard technology are particularly prominent, with leverage, institutional positioning, and earnings revisions still at relatively high levels; valuation, retail sentiment, corporate behavior, and policy signals have clearly cooled. Data as of August 7, 2026.

Price Signals: Prices Begin to Cool, Concentration Remains High

Signal One: The Earnings Gap Between Soft and Hard Technology Narrows Rapidly

In the first half of the year, the biggest characteristic of AI trading in China was not a general rise in all AI assets, but rather a high concentration of funds in hard technology.

Goldman Sachs data shows that in the first six months of this year, the earnings gap between the STAR Market and the Hang Seng Tech Index once exceeded 100 percentage points, comparable to the high levels seen when market dispersion surged in early 2021. Small-cap, high-growth, and momentum styles also became the main driving forces in the market at that time.

Entering July, these styles quickly reversed. As global AI hardware trading receded, the excess returns of hard technology, small-cap, and momentum factors significantly retraced, and the performance gap of soft technology relative to hard technology has returned to near long-term average levels.

The previous leaders have retraced most of their gains since July.

The adjustment amplitude of momentum and size factors has even entered a relatively extreme range. Goldman Sachs thus judges that, from the perspective of price momentum, the phase of retraction and style reversal may be beginning to ease.

This signal leans positive: the direction that had the largest gains and the most crowded positions has experienced a significant cooling. However, a slowing down of price declines does not mean that the related stocks have completed their clearing.

Signal Two: The Gains of the STAR Market Are Still Contributed by a Few Stocks

The second signal comes from market breadth.

Goldman Sachs estimates that 90% of the gains of the STAR Market this year have come from the top 10 performing stocks. In comparison, this ratio was 23% in early 2021, and 46% for the ChiNext index in mid-2015.

Trading is also concentrating in a few directions. The technology sector, ChiNext, and the STAR Market currently account for 26%, 14%, and 6% of A-share cash trading volume, respectively, all at relatively high levels in recent years.

Meanwhile, the correlation between individual A-share stock returns remains at a near historical low. Low correlation means that the market is not primarily trading based on macro factors for broad rises or falls, but continues to chase a few micro themes like AI and technology.

This indicates that although hard technology stock prices have corrected, AI remains the most important pricing clue in the market, and returns and trading have not truly spread. Sentiment has cooled, but the trading structure remains concentrated.

About 90% of the STAR Market's gains this year have come from the top 10 performing stocks, significantly higher than the 23% in early 2021 and 46% in mid-2015 for the ChiNext index. The dominance of a few stocks in driving index gains indicates that the market breadth of AI hard technology trading remains insufficient.

Signal Three: Valuations Return to the Mean, Absolute Prices Still Not Cheap

Valuation is the third signal to assess whether sentiment has been fully released.

At the peak in June, the forward P/E ratio of the ChiNext index, calculated on a market capitalization-weighted basis, was about 26 times, while the median forward P/E ratio of the STAR Market constituent stocks was about 50 times. Following the subsequent stock price decline, the valuation multiples of A-share hard technology companies have fallen to near or below long-term average levels.

If earnings growth is taken into account, the forward PEG of the relevant companies is also at historical median or lower levels. This means that, purely from the perspective of valuation relative to its own history, some of the overheating has been digested.

However, the absolute valuation of the STAR Market remains high. After a simultaneous adjustment in the global AI sector, the valuation discount of Chinese AI concept stocks relative to overseas peers has also significantly narrowed, and the previous cushion of 'AI is cheaper in China' has thinned.

Therefore, the valuation signal does not indicate 'it is now cheap', but rather 'the time of being most expensive may have passed'. The market will rely more on earnings realization going forward, rather than continuing to depend solely on valuation expansion.

Funding Signals: Retail Investors Have Cooled, Institutions and Leverage Have Not Fully Retreated

Signal Four: Financing Balance Declines, But Leverage Concentration Hits New Highs

The financing balance is the most direct indicator of observing leverage trading in A-shares.

Goldman Sachs data shows that the financing balance of A-shares has decreased from about 3 trillion yuan to about 2.6 trillion yuan, with the proportion of free float market capitalization also falling from 6.0% to 5.5%. This indicates that some leveraged funds have exited the market.

However, whether in terms of financing balance or financing proportion, it is still above historical norms. Goldman Sachs believes that if deleveraging continues, compared to the more pronounced recent deleveraging in the South Korean and Taiwanese markets, the deleveraging cycle in A-shares may still be in its early stages.

Although the financing balance of A-shares has fallen from its peak, leverage remains concentrated in the technology sector. According to Goldman Sachs' estimates, the top 10% of stocks with the most concentrated financing balance account for about 30% of the total market financing balance, which is at a historical high, and further declines in hard technology could amplify localized volatility.

It is also important to note the distribution of leverage. According to Goldman Sachs' estimates, the top 10% of stocks with the most concentrated financing balance currently account for about 30% of the total financing balance in A-shares, setting a historical high, mainly concentrated in the AI hard technology sector.

This means that while the overall market's leverage risk may be significantly lower than in 2015, there is still structural pressure within AI hard technology. If related stocks continue to decline, the concentrated financing positions may still amplify volatility.

Signal Five: Retail Investor Risk Appetite Has Dropped from Overheated to Neutral

Goldman Sachs' fifth signal comes from retail sentiment.

Despite the continuous increase in the proportion of domestic public funds, pensions, and insurance capital, retail investors still contribute about 70% of the daily trading volume in A-shares. Therefore, changes in retail sentiment will still directly affect short-term market fluctuations.

Goldman Sachs has constructed a retail sentiment barometer that includes high-frequency indicators such as financing data, new account openings, IPO subscriptions, turnover rates, and stock allocations. This indicator has now dropped from about one standard deviation above the average of the past year a month ago to around zero standard deviation.

Goldman Sachs' A-share retail sentiment indicator has dropped from about one standard deviation above the average a month ago to around zero standard deviation, indicating that the enthusiasm for chasing prices has rapidly dissipated. However, the current risk appetite has only returned to neutral and slightly subdued, and has not entered an extremely pessimistic range.

This indicates that the risk appetite of retail investors in A-shares has fallen from an overheated state to neutral or slightly subdued.

The rapid cooling of retail sentiment is evidence that this round of adjustment has released some speculative pressure. However, zero standard deviation does not represent extreme pessimism, nor is it a 'panic bottom' in the traditional sense. It only indicates that the enthusiasm for chasing prices has faded, and it cannot yet prove that the market has completed its final round of selling.

Signal Six: Slight Reduction in Public Fund Holdings, Tech Allocation Remains at Historical Highs

Institutional investors behave more complexly than retail investors.

Goldman Sachs data shows that the total assets managed by domestic public funds have approached 40 trillion yuan, with approximately 7 trillion yuan allocated to stocks, accounting for 6.6% and 15% of the total market capitalization and free float market capitalization of A-shares, respectively.

During the market adjustment period, the cash ratio of equity public funds has increased, indicating that fund managers have moderately reduced risk. However, their allocation and overweighting of tech stocks such as semiconductors, hardware, and software remain at historical highs.

The reduction of risk in systematic strategies may be more pronounced. The cash trading and financing spreads of small and mid-cap stocks have decreased, indicating that the activity of systematic investors such as quantitative funds has weakened, possibly undergoing a deeper level of deleveraging.

There is a clear divergence within this signal: quantitative and short-term funds have contracted, but traditional institutions have not significantly loosened their core positions in tech stocks. As long as institutional allocations remain high, AI hard tech is unlikely to be defined as a fully crowded trade.

Behavior and Policy Signals: Companies Begin Buybacks, "National Team" Funds Start to Support the Market

Signal Seven: Increase in Buybacks, Decrease in Abnormal Trading Alerts

The actions of listed companies often reflect insiders' judgments on valuation and risk more than market slogans.

Goldman Sachs observes corporate behavior from three perspectives: buybacks, trading alerts, and important shareholder transactions. On a quarterly basis, A-share buybacks in the third quarter of 2026 have risen to a multi-year high, with the number and amount of announced buyback transactions increasing by 35% and 59% year-on-year, respectively.

An increase in buybacks usually indicates that management believes the company's stock price is below its intrinsic value, or at least is willing to invest cash at the current price to support shareholder returns.

Another change is that the number of abnormal stock price fluctuations and trading risk alerts issued by listed companies significantly increased in June, just before the market correction, but has since decreased sharply. The trading of important shareholders and company executives has also shifted from net reductions for most of the first half of the year to a more balanced state.

This set of signals leans towards the positive: insiders' behavior is no longer inclined towards reductions and warnings of overheating as it was in the late stages of the rise, but is gradually shifting towards buybacks and reduced net selling.

Signal Eight: Policy Tightening Risks Return to Neutral from Peak

The sentiment cycle of the Chinese stock market is often influenced by policy changes.

The deleveraging in 2015 and the regulatory tightening that began at the end of 2020 have both been important reasons for market reversals; conversely, clear policy support has repeatedly driven strong rebounds in Chinese stocks.

Goldman Sachs uses large language models to analyze public statements from regulatory agencies and policymakers, measuring the policy support and tightening risks faced by the stock market from the frequency and intensity of wording.

The model shows that concerns among decision-makers about market overheating and the resulting policy tightening risks peaked in the first quarter of 2026 and have since retreated to a more neutral range.

This means that current policies are neither clear sentiment boosters nor the main source of pressure in this round of adjustments. Compared to price, leverage, and profit factors, policy signals are temporarily closer to neutral.

Signal Nine: "National Team" Shifts from Reducing Holdings to Net Buying

The ninth signal comes from the "National Team."

Goldman Sachs estimates that the broad "National Team" currently holds about 5 trillion yuan in A-share assets, equivalent to 5% of the total market capitalization of A-shares. Historically, such funds typically buy against the trend when the market is under pressure and may reduce holdings when the market rises and valuation attractiveness declines.

Goldman Sachs' tracking data shows that after selling about 1.5 trillion yuan in A-shares over the previous six months, the "National Team" has turned to net buying of over 140 billion yuan in the past three weeks, including a small amount of STAR 50 ETF.

Goldman Sachs estimates that after selling about 1.5 trillion yuan in A-shares over the previous six months, the "National Team" has turned to net buying of over 140 billion yuan in the past three weeks, indicating a resurgence of market support. However, the report has not confirmed that the current buying scale has reached the historical threshold typically corresponding to a mid-term bottom.

The shift from selling to buying by the "National Team" indicates a change in the judgment of policy funds regarding market risks, providing a certain downward buffer for the market.

However, this does not directly confirm a mid-term bottom. Historical backtesting by Goldman Sachs shows that only when the weekly net buying scale of the "National Team" exceeds 1.5 standard deviations is it more likely to correspond to a mid-term bottom for the market. The current clearer conclusion is that support forces have reappeared, rather than the market having received unconditional support.

Fundamental Signals: AI Investment Remains Strong, Earnings Revision Momentum Begins to Peak

Signal Ten: Capital Expenditures Continue to Grow, but Marginal Changes No Longer Accelerate

The last signal, which also determines how far the AI market can go, comes from capital expenditures and earnings forecasts.

Goldman Sachs expects that the nine major super-large enterprises and cloud service providers in the U.S. and China may spend more than $900 billion on AI this year, further increasing to $1.3 trillion next year, equivalent to 1.7% and 2.3% of the GDP of the U.S. and China, respectively.

These expenditures are transmitted along the global AI industry chain, covering South Korean memory chips, Taiwanese semiconductor foundries, Japanese semiconductor materials and equipment, as well as Chinese power, infrastructure, and technology companies.

Since the beginning of this year, the capital expenditure forecasts for 2026 and 2027 of eight listed super-large enterprises in the U.S. and China have been raised by 32% and 75%, respectively, driving the earnings forecasts for China's hard tech sector up by 12% and 22% for the fiscal years 2026 and 2027, respectively.

This indicates that the rise of China's AI hard tech is not merely driven by sentiment and leverage; there is indeed support from capital expenditures and earnings growth behind it.

The upward revision of earnings forecasts for China's hard tech remains significantly higher than that for soft tech, indicating that the AI market is not entirely driven by sentiment. However, Goldman Sachs believes that the momentum for upward revisions in hard tech earnings may have peaked in the short term, while soft tech still has room for cyclical improvement in earnings revisions.

The issue is that the speed of upward revisions in capital expenditure forecasts has slowed from high levels, and the momentum for upward revisions in earnings forecasts for Chinese hard tech companies has also shown signs of peaking at least in the short term. In contrast, the momentum for earnings revisions in soft tech still has cyclical improvement potential.

Therefore, in the coming months, the key factors influencing AI sentiment will no longer just be how large the capital expenditures are, but whether capital expenditures can continue to exceed expectations and whether cloud vendors can provide clearer paths for AI monetization.

Ten Signals Point to: Overheating Has Cooled, Adjustment Not Yet Fully Ended

Putting the ten signals together, the current position of China's AI hard tech is becoming clearer.

Positive signals include: the extreme return gap between hard tech and soft tech has narrowed, momentum and small-cap styles have completed a rapid reversal, valuations have returned to long-term averages, retail sentiment has dropped to neutral, buybacks by listed companies have increased, important shareholders' reductions have become more balanced, and the "National Team" has also shifted back to net buying.

These changes indicate that the speculative positions, valuation overheating, and leverage pressures accumulated in the first half of the year have released a considerable portion.

However, cautious signals also exist: the absolute valuation of the STAR 50 remains high, the concentration of A-share tech trading and earnings is at high levels, financing balances are still above historical norms, leverage concentration has reached new highs, and public funds' allocation to tech stocks remains at historical highs.

The fundamentals are positioned between the two. Global AI capital expenditures remain strong, and hard tech earnings forecasts are also being revised upward, but the momentum for upward revisions in capital expenditures and earnings forecasts has begun to slow.

Goldman Sachs thus maintains an "overweight" judgment on A-shares and is structurally optimistic about AI hard tech in the long term, but emphasizes rotation and diversified allocation in the short term. Its recommendations include gradually increasing allocations to selected Hong Kong soft tech, policy beneficiaries, and self-controlled directions, while also focusing on stocks with upward earnings revisions, IPOs, and cash returns from dividends and buybacks.

For investors, these ten signals do not simply indicate "AI has bottomed" or "the AI market has ended," but rather serve as a thermometer of sentiment: the hottest times may have passed, the most urgent declines may also be starting to slow, but concentrated positions and earnings expectations still need to continue to digest.

In the next phase, whether China's AI trading can reheat will depend more on corporate earnings and AI monetization, rather than relying again on valuations, leverage, and chasing sentiment.