Need AI to provide answers

The volatility of global risk assets superficially stems from liquidity issues, but fundamentally, it is due to the over-reliance of risk assets on a single narrative of AI. When the pace of industrial development (especially commercialization) fails to keep up with the rhythm of the secondary market, appropriate valuation corrections can also serve as a means to alleviate risks. On Thursday evening (November 20) Beijing time, the release of U.S. non-farm payroll data and the downward revision of interest rate cut expectations by the Federal Reserve triggered a correction in the valuations of high-priced assets. The market's anxiety regarding the sustainability of North American AI infrastructure was further amplified by the delayed expectations of interest rate cuts. The broadening of commercialization scenarios by AI, hardware cost reductions, and rising financial stability risks that may force the Federal Reserve to cut rates early could break the current deadlock. Prior to this, for the A-shares, the continuous inflow of absolute return-oriented funds has enhanced the intrinsic stability of the market. Under the increasingly dominant ecosystem of left-side stable funds, A-shares/Hong Kong stocks may resemble U.S. stocks more in the future, exhibiting "sharp declines and slow recoveries." For investors needing to increase equity allocations, the current early release of risks provides an opportunity to reallocate A-shares/Hong Kong stocks at the end of the year and to layout for 2026. From a configuration perspective, the reassessment of pricing power in resources/traditional manufacturing and enterprises going overseas remains the core allocation direction, while the high-cut-low strategy may face increased difficulty in rotation timing due to overly consistent expectations.

The downward revision of interest rate cut expectations by the Federal Reserve this year triggered a correction in high asset valuations

Federal Reserve Chairman Jerome Powell stated in the interest rate meeting on October 29 that a rate cut in December is "not a done deal," leading to a continuous downward adjustment of market expectations for interest rate cuts by the Federal Reserve this year. Subsequent hawkish remarks from some Federal Reserve officials and the hawkish tone of the October meeting minutes further strengthened expectations for liquidity tightening. The S&P 500 has cumulatively fallen 4.2% since the October interest rate meeting, while the CSI 300 has dropped 5.1%, and the Hang Seng Tech Index has decreased by 11.5%. The correction in the technology sector, which has relatively high valuation positions in both China and the U.S., is more pronounced, with the STAR Market 50 and the Nasdaq Index retreating by 12.6% and 6.5%, respectively. Since the U.S. Bureau of Labor Statistics (BLS) will no longer separately release October non-farm data, the September non-farm data will be the last official reference before the December interest rate meeting. The better-than-expected new non-farm data further obscured the path for interest rate cuts (despite the unemployment rate also exceeding expectations and previous months' non-farm data being revised down), leading to adjustments in various risk assets in global markets.

Market anxiety regarding the sustainability of North American AI infrastructure has also been amplified by delayed expectations for interest rate cuts

This week, Google's release of Gemini 3 and Nvidia's disclosure of its Q3 2025 financial report can be considered better than expected, especially as Gemini 3 demonstrates that the model's multimodal capabilities continue to progress rapidly. However, these advancements have not dispelled investors' concerns about whether AI can form a commercial closed loop and whether the increasingly debt-financed investments in AI infrastructure are sustainable. Since September, the CDS (credit default swap) prices of major U.S. technology companies have significantly risen. From September until November 21, the 5-year CDS quote for Oracle, which has been the most aggressive in debt financing, has increased by 167% Microsoft, Amazon, and Google rose by 58%, 53%, and 34%, respectively. Before AI gains broader commercial value, AI infrastructure spending still relies on the operating cash flow and debt financing of major tech companies, which in turn is highly dependent on a healthy U.S. economy, advertising revenue, and corporate IT budget expenditures—variables that are all pro-cyclical. The Federal Reserve's pause in interest rate cuts has somewhat changed the market's judgment on the cost of corporate debt financing, further deepening concerns about the sustainability of AI infrastructure investments. Until AI forms ecological barriers and a business model with increasing returns to scale, the tech sector will remain highly sensitive to real economic performance and monetary policy. In the past, the tech sector's weak correlation with economic performance was precisely its advantage; however, currently, heavy asset investments are making the tech sector increasingly resemble cyclical stocks.

Three Breakthrough Points: AI Expands Commercialization Scenarios, Cost-Effective Hardware, Financial Stability Risks Rise, Forcing the Federal Reserve to Cut Rates Early

1) Improvement in model capabilities or formation of ecological barriers breaks the commercialization anxiety. The market does not deny the future commercialization space and application scenarios of AI, but rather that the speed of improvement in AI model capabilities is not keeping pace with the current guidance for AI infrastructure investment, making it difficult to quell the debate surrounding corporate earnings reports. The last round that broke the panic over computing power demand was the powerful capabilities of inference models and the sharp increase in token consumption. If the improvement in model capabilities stagnates, even impressive hardware orders or corporate earnings reports may not alleviate investors' anxiety about AI commercialization. According to the comprehensive assessment results of the Artificial Analysis AI analysis index as of November 21, the top three LLM models are Google Gemini 3 Pro (72.9 points), OpenAI GPT-5.1 (69.7 points), and Kimi K2 Thinking (67.0 points). Chinese models have quietly become the cornerstone models for many overseas AI startups, with costs significantly lower than North American competitors. If the ceiling on model capabilities does not rise quickly enough, the competitive landscape for applications will deteriorate rapidly. Therefore, the market response after the full rollout of Gemini 3 Pro is worth paying attention to, as this powerful model was only recently released. In addition to improvements in model capabilities, the enhancement of user ecological barriers also helps to form a business model with increasing returns to scale (similar to the internet). Personalized AI and Agents are possible forms, and we see Google demonstrating this potential. Alibaba's recent launch of the Qianwen APP also has the potential to become an ecological entry point. Whether WeChat will also launch an Agent in the future, and whether Apple's Apple Intelligence can successfully upgrade, are all events worth watching.

2) Reducing infrastructure costs through hardware discounts is also a way to make investments more sustainable. According to data from GPU cloud service provider Jarvislabs, conservative estimates suggest that GPU computing power accounts for nearly 50% of AIDC investment costs. This uneven distribution of costs has led to significant debate in the market regarding the depreciation of computing cards The debate will continue as long as there is no clear evidence of the economic benefits generated by hardware. After all, the explosion of LLMs has only been three years, and the deployment of servers equipped with advanced computing cards is still in the early stages of large-scale deployment. If upstream companies represented by NVIDIA start to "share profits," it could be a way to alleviate the pressure of investment in AI infrastructure, but it would also damage the profit expectations and valuations of the hardware supply chain. Currently, NVIDIA's approach to supporting the industry chain is through direct strategic investments or purchasing services, indirectly "sharing profits," but this has led to concerns in the market about the opacity of mutual financing among major tech companies.

3) Deteriorating financial conditions force the Federal Reserve to cut interest rates early. This possibility is perhaps what secondary market investors are most looking forward to in the short term. The narratives around AI, computing power dollars, cryptocurrencies, and Web 3.0—any significant volatility in asset prices could harm global financial stability, which is a risk that the current Federal Reserve cannot afford. According to the Securities Times, on November 21 local time, New York Fed President Williams also stated that given the current policy is somewhat tight, there is still room for interest rate cuts in the near term, and inflation progress has stalled, but it is expected to reach the 2% target by 2027. After Williams made the above statement, U.S. stocks, cryptocurrencies, and precious metals all saw varying degrees of recovery. As of Friday, CME showed that the market's expectation of a December interest rate cut by the Federal Reserve surged from 30% on the 19th to 71%. What investors can most look forward to in the short term may only be the Federal Reserve. However, strictly speaking, the current U.S. Financial Stress Index (FSI) has only slightly increased and is still far from the levels seen in April, remaining at the 34th percentile since 2000, which is not high.

For A-shares, stable return-oriented absolute yield, continuous capital inflow enhances the market's inherent stability.

1) The biggest difference between A-shares and U.S. stocks is that investors have a certain psychological expectation for this round of market adjustment. Over the past 10 trading days, the average daily turnover rate of A-shares has dropped to 1.88%, within the estimated calm period range of 1.7% to 1.9%; the TMT sector's trading volume share fell to 29.8% in the week of November 21, at the 33rd percentile since 2023, indicating a significant decrease in congestion; in terms of implied volatility, since the end of October, the implied volatility of S&P 500 index options has continued to rise, from 15.8% on October 27 to 26.4% on November 20, while during the same period, the implied volatility of CSI 300 and CSI 1000 index options has remained stable, oscillating slightly around 19.0% and 23.4%, respectively. In fact, A-share investors also hope that the market can reverse the current situation of a single narrative around AI and return to a diversified logic-driven approach to re-enter the market and position for next year 2) The power of left-side layout and trading in the market is continuously growing. Firstly, the ETF market has shown a counter-cyclical characteristic where the more the index declines, the more subscriptions increase. In the past two weeks, the cumulative net subscription of ETFs in the A-share market reached 25.2 billion yuan, with broad-based ETFs accounting for a cumulative net subscription of 5.59 billion yuan, and industry and thematic ETFs accumulating a net subscription of 19.61 billion yuan, clearly indicating a "buying on dips" characteristic. The Hong Kong stock ETF market has also exhibited a similar phenomenon, with net subscriptions occurring for 21 consecutive trading days as of November 20, and only one trading day (October 21) since July this year showing net outflows, with a cumulative net subscription scale of 267.7 billion yuan. Stable return-type products have become the main force for incremental capital inflows. In the first three quarters of this year, the net subscription scale of public offering "fixed income +" products was 105.5 billion yuan, 82.4 billion yuan, and 479.5 billion yuan respectively (totaling 667.4 billion yuan), especially in the third quarter, where actively managed equity products faced some redemption pressure, while fixed income + products continued to attract capital. Bank fixed income + wealth management products are also highly favored. According to the FICC team of CITIC Securities Research Department's forecast, the scale of bank "fixed income +" wealth management products is expected to grow by more than 1.4 trillion yuan throughout 2025. The increase in insurance is even more pronounced, with our estimation suggesting that theoretically, an average of 150 billion yuan needs to be invested in A-shares each month. These large-scale funds with low equity allocation ratios are becoming the main force entering the market, and they typically regard drawdown control as one of the most important strategic constraints, which also determines that mainstream products are likely to primarily adopt a left-side layout approach. In an ecosystem where incremental funds are increasingly dominated by left-side stable funds, A-shares/Hong Kong stocks may exhibit more of a "sharp drop and slow rise" pattern similar to U.S. stocks in the future. For investors needing to increase equity allocation, the current early release of risks provides an opportunity to reallocate A-shares/Hong Kong stocks and plan for 2026 by the end of the year.

Risk warning and disclaimer

The market has risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial conditions, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investing based on this is at one's own risk