Author: Shen Jianguang
Investors need to pay close attention to the high valuations of US stocks, their divergence from fundamentals, and their over-reliance on AI. If the results of AI applications fall short of expectations, a significant correction is likely next year.
On November 20th, the US stock market experienced a sharp correction. Although Nvidia announced third-quarter revenue that exceeded market expectations, the Nasdaq index fell sharply after opening higher, ultimately closing down 2.2% from the previous trading day, and the S&P 500 also fell 1.6%. In my opinion, the market correction may reflect investors' concerns about the excessively high valuations of US stocks and their divergence from economic fundamentals. At the same time, the current US economy and capital markets are highly dependent on investment and applications in artificial intelligence (AI). If its profitability falls short of expectations, investors may need to prepare for further corrections in the US stock market in 2026.
Reason 1: US Stocks Are Overvalued While the S&P 500's PE ratio of nearly 30 and the Nasdaq's PE ratio of around 40 don't appear to be at historical extremes on the surface, historical extremes often occur after significant shocks severely impact corporate earnings. For example, the 2020 pandemic caused the S&P 500's PE ratio to rise above 40 at the beginning of 2021. To avoid the impact of such extremes, many investors use the average earnings over the past ten years as a reference to calculate the PE ratio, i.e., calculating the "cyclically adjusted" PE. If this indicator is used, the current S&P 500's PE ratio has reached 40.0, second only to the period of the US tech bubble in 2001, making the expensiveness of stock valuations obvious. The high valuations of stocks correspond to the market's optimistic expectations for corporate profits, which are quite challenging to achieve. Looking at long-term PE ratios, the market expects Nasdaq's valuation to fall back to 25.6 times by 2027, meaning that the earnings of Nasdaq component stocks will grow by 61% in just over two years. In the long run, the proportion of corporate profits to gross national income in an economy is generally stable, meaning that the overall growth rate of corporate profits is roughly equivalent to nominal GDP. Even over the next two years, it will not be easy for a stock market with a market capitalization (Nasdaq's current total market capitalization is approximately $40 trillion) that already exceeds one-third of US GDP to achieve a significantly higher growth rate for the profits of its companies (despite including foreign companies and foreign revenue) than the nominal GDP growth rate of the US (and globally). Structurally, the current boom in the US stock market is primarily concentrated in AI-related stocks. Statistics show that since the launch of ChatGPT at the end of 2022, AI-related stocks have contributed 75% of the S&P 500's returns. Among them, the performance of the "Big Seven" US stocks, led by Nvidia and Microsoft, has been particularly impressive. However, even though the profit prospects of large companies are relatively stable, what is more noteworthy is that the current P/E ratio of the more than 3,000 companies remaining in the Nasdaq index after excluding the "Big Seven" (whose market capitalization accounts for about half of the Nasdaq's total market capitalization) is as high as 50, while the projected P/E ratio for 2027 is around 24. This indicates that the market expects these companies' profits to grow by more than 100% in about two years—a prospect that is likely even more difficult to achieve. Reason Two: The Performance of US Stocks Deviates from Economic Fundamentals The recent continuous rise in the US stock market is inconsistent with the fact that the US economy has slowed significantly this year, with rising inflationary pressures and increasing pressure on the job market, highlighting the risk of stagflation. This divergence between the performance of US stocks and economic fundamentals is also cause for concern. In the first half of this year, US GDP grew at an annualized rate of approximately 1.5% (third-quarter economic data has not yet been released due to the government shutdown), while the Federal Reserve's forecast for full-year economic growth is only 1.6%. Meanwhile, consistent with the economic slowdown, the US job market has cooled significantly recently. In the four months from June to September, the average monthly non-farm payrolls in the United States increased by only 43,000, significantly lower than the average of 135,000 in the 12 months prior to May of this year. While the economy is slowing, inflation in the United States has rebounded, indicating stagflation pressures. Influenced by factors such as tariffs, the year-on-year growth rate of commodity prices in the US inflation basket has recently rebounded. With the further development of the tariff inflation effect, overall US inflation may rise further this year. The Federal Reserve predicts that the full-year PCE may reach around 3%, higher than in 2024. In fact, due to the uncertain inflation outlook, changes in market expectations for Fed rate cuts are also one of the key factors in the recent market adjustment. The possibility of future inflationary pressures exceeding expectations and a tighter US liquidity environment impacting the market in 2026 cannot be ruled out. Structurally, the US economy also relies heavily on AI. The US economic growth in the first half of this year mainly relied on increased household consumption and corporate capital expenditure, while residential investment, inventory adjustments, net exports, and government spending all contributed negatively to economic growth. Considering that US consumer spending growth has also slowed compared to 2024, corporate capital expenditure, especially AI-related investments such as data centers, is currently the most crucial driver of the US economy. Harvard economist Nick Lichtenberg even claimed that without AI-related capital expenditure, US GDP growth in the first half of 2025 would only be 0.1%. Reason Three: The Impact of Artificial Intelligence on Corporate Profitability Remains Highly Uncertain. As a new generation of general-purpose technological advancement comparable to the steam engine and the internet, artificial intelligence is highly anticipated by all sectors of society, including investors, and has become a key factor supporting the US capital market and the US economy. Currently, artificial intelligence may have a positive impact on the US economy and even the global economy, but the magnitude of this impact, especially its contribution to corporate profits, remains highly uncertain. If the actual application of artificial intelligence proves to be less than expected, the impact on the market should not be underestimated. In terms of overall impact, some studies do not support the optimistic prediction that AI will completely change the world. For example, Goldman Sachs predicts that AI will increase global GDP by about 7% over the next ten years, or an average annual increase of 0.7%. Nobel laureate economist Daron Esmooglu's estimate is an order of magnitude lower, predicting that AI will boost US GDP by a total of 1.1-1.6% over the next decade, averaging about 0.1% annually. Even if AI has some impact on the overall economy, how this will affect corporate profits remains highly uncertain. On the one hand, the implementation of AI is not easy. For example, a recent systematic analysis of over 300 publicly available AI projects and announcements by MIT, along with a survey of 153 leaders, concluded that 95% of generative AI investments yielded almost no returns for companies, half of the projects failed, and only 5% achieved commercialization. Meanwhile, investors have also widely noted that OpenAI, a leader in this wave of artificial intelligence, has supported the profitability and valuation of upstream and downstream companies through external financing, despite its massive losses. For example, OpenAI has pledged to invest $1.4 trillion in data center infrastructure and cloud services over the next few years, supporting the profitability of upstream chip manufacturers and cloud service providers. At the same time, OpenAI provides some API services below cost, facilitating market expansion for downstream startups. However, if OpenAI's model becomes unsustainable, the entire industry chain could be severely impacted. On the other hand, IMF research and numerous studies indicate that the benefits of AI may be highly unevenly distributed across industries, populations, and even countries. This means that if the benefits generated by AI are excessively concentrated in a few companies, or even form higher barriers to monopoly, it could lead to greater class conflict in American society and even policy risks such as antitrust investigations, which also need to be guarded against. Reason Four: Be Wary of AI Companies Forming Alliances Across the Supply Chain Recently, the hottest AI giants in the US have shown signs of interconnecting and forming alliances across the supply chain; specifically, suppliers provide funding to customers, customers and suppliers share revenue, and they hold shares in each other. This makes it extremely difficult to assess the development of real AI demand and the independent financial situation of each company. Given the significant uncertainty surrounding the future of AI, this lack of transparency may increase the risk of market adjustments. In September of this year, Nvidia and OpenAI announced a 10 gigawatt computing power partnership, with Nvidia expected to invest $100 billion in OpenAI to provide cash for the latter to purchase computing equipment and promising to provide OpenAI with sufficient chips. This deal itself has already attracted much attention and discussion. Furthermore, OpenAI and chip giant AMD announced a massive computing infrastructure agreement, under which the two companies will collaborate to deploy a total of 6 gigawatts of computing power. As the scale of the deployment expands and other conditions are met, OpenAI can exercise its options to acquire approximately 10% of AMD's equity. In fact, Oracle, and even Microsoft and xAI, have signed similar agreements. Overall, the author believes that investors need to closely monitor the high valuations of US stocks, their divergence from fundamentals, and the vulnerability arising from over-reliance on AI. If the results of AI applications fall short of expectations, the US stock market may experience a significant correction in 2026.