CICC: How to view the trends of the US stock market and the AI hardware industry chain?
Author: CICC Research
On November 5, Eastern Time, the three major U.S. indices, the S&P 500, Nasdaq, and Dow Jones, fell by 4.4%, 5.5%, and 3.4%, respectively, marking the largest single-day decline since the tariff shock in April 2025. The AI industry chain saw significant declines, with the Nasdaq 100 dropping over 6% and the Philadelphia Semiconductor Index plummeting about 12%, both recording the largest single-day declines since March 2020.
Chart: The three major U.S. indices, S&P 500, Nasdaq, and Dow Jones, fell by 4.4%, 5.5%, and 3.4% respectively
Source: FactSet, CICC Research Department
How to view the significant decline in U.S. stocks and the subsequent trends. Our brief comments are as follows:
I. Characteristics of this decline: significant drop, broad decline, signs of panic and stampede
1)Significant drop: The Philadelphia Semiconductor and Nasdaq 100 indices both recorded their largest single-day declines since March 2020, falling by 12% and 6%, respectively.
2)Broad decline: Unlike previous instances when AI cooled down and defensive sectors rose slightly, all 11 sectors of the U.S. stock market fell yesterday, including defensive consumer staples and utilities, indicating possible signs of a stampede.
Chart: All 11 sectors of the U.S. stock market fell yesterday
Source: FactSet, CICC Research Department
3)Increased volatility and panic: The VIX rose to 28 (a level above 25 is considered dangerous, second only to the situation in Iran at the beginning of the year). The gap between the spot VIX and three-month futures widened to the largest since April 2025, which often directly reflects market panic and may even be a precursor to further pressure.
Chart: VIX rose to 28, second only to the level during the Iran situation at the beginning of the year
Source: Bloomberg, CICC Research Department
II. Reasons for this decline: AI regulation amplified by emotions
The direct trigger for this decline remains the market's panic over the potential increase in AI regulation and the halt of data center construction following a Democratic victory in the midterm elections. Newly elected Democratic Texas Governor Gina Hinojosa stated in a media interview on Thursday that "Texas will introduce laws to protect citizens from the impact of data centers" [1]; at the federal level, Sanders also reiterated calls to pause the construction of national data centers [ https://www.cnbc.com/2026/11/05/democrats-data-center-restrictions-midterms.html\]. The convergence of state and federal policy proposals has intensified market concerns about the progress and sustainability of AI infrastructure investments.
In recent days, the market has fluctuated repeatedly, with a significant drop in pre-market futures on the day of the ballot opening on the 3rd, followed by a rebound and a global surge, leading to today's decline. As we pointed out in "The Asset Implications of the Democratic 'Sweep'," many Democratic issues can have completely opposite effects on the same asset, and in a situation where it is unclear which can be advanced first, the global market "chooses" a direction to trade, where both surges and declines can find a multitude of explanations.
Essentially, the core impact of the midterm elections on asset pricing lies in whether the dominant market force aims for "correction" (inflation, Iran situation, Federal Reserve, tariffs) or is guided by "opposition for the sake of opposition" (AI, fiscal policy). The day before yesterday was trading the former, and yesterday was trading the latter, but this does not mean the market can predict which direction will develop first. For example, while concerns about AI regulation do need attention, the Democratic advantage in the Senate is still insufficient to push through federal AI regulatory legislation.
III. How to view the market at this level? Short-term focus on liquidity, stabilization requires clearing and catalysts, market focuses on gold and Hengke
Combining the characteristics of the decline analyzed above, regardless of the reasons and catalysts, and whether subsequent concerns materialize, the short-term trend is no longer dominated by fundamental logic.
► The emergence of panic and stampede, whether there will be further liquidation and liquidity risks, will take some time to manifest, as we have already seen some hedge funds announcing liquidation. The leverage level in U.S. stocks had previously risen to a historical high, with the implied leverage level of margin balances reported by FINRA reaching a historical peak of 3.4, and the ratio of bullish to bearish positions in CBOE options open interest also rising to 1.33, the highest since the end of 2018. This also means that the market is more sensitive to changes in price and volatility; once a rapid decline triggers margin calls, passive deleveraging, and adjustments in derivatives hedging, selling pressure may spread from the AI sector to other holdings, reinforcing the interplay between price declines and deleveraging.
Chart: FINRA reported margin balances implied leverage level reached a historical peak of 3.4
Source: Haver, CICC Research Department
Chart: CBOE options open interest bullish/bearish ratio also rose to 1.33
Source: Haver, CICC Research Department
Currently, the situation is still limited to the stock market. If we see high U.S. Treasury rates and a strong dollar, we need to be even more vigilant, as this indicates that the market is selling off safe assets, including U.S. Treasuries, to meet funding gaps, and liquidity risks will sharply increase.
► Currently, AI is in a transitional state, and the key issue is the lack of demand and a new narrative and catalyst for models, which will amplify credit risks and regulatory pressure risks. Additionally, high-intensity capital expenditures continuously squeeze the free cash flow of cloud vendors, leading to growing concerns about the speed of AI revenue realization and investment returns. The market's stabilization requires natural clearing or strong catalysts to reverse pessimistic expectations; from any angle, some patience is needed.
Chart: Overall U.S. stock valuations are currently not at extreme highs
Source: Bloomberg, CICC Research Department
In terms of timing, the first half of next year is a critical verification period; the market needs to see similar to the first quarter of this year where Anthropic's performance exceeded expectations and the business achieved actual breakthroughs to boost confidence. Otherwise, as market patience wears thin, the prospects for capital expenditures and returns in 2028 will further dim, and the main trading line in the market may shift to recession trades.
Chart: Previous growth expectations have been more fully priced in, with clearer pullbacks in segments, while cloud vendors are relatively resilient
Source: FactSet, CICC Research Department
Structurally, AI may still be the main line, but internally, more emphasis should be placed on profitability sustainability and certainty. Overall U.S. stock valuations are currently not at extreme highs, but there is significant differentiation within the AI industry chain. In this round of AI market, hardware infrastructure segments such as network interconnection/light modules, semiconductor manufacturing, and liquid cooling/temperature control have incorporated higher profit growth expectations, while the pricing of cloud vendors is relatively reasonable. Recent market performance also reflects this difference, with clearer pullbacks in segments where growth expectations have been more fully priced in, while cloud vendors have shown relative resilience. This also means that the pricing of cloud vendors still leaves room for subsequent profit growth, but if the profitability of segments like light modules cannot sustain high growth, there is a risk of further valuation adjustments.
► Observing the performance of the mid-term elections over the past two days, the market trading direction with the least resistance has gradually converged on Hengke and gold. 1) Hengke has a low exposure to AI, so there is not much to worry about AI risks. The expectation of a decrease in geopolitical risks leading to a drop in oil prices and inflation can also reduce the pressure for interest rate hikes. 2) Gold benefits from the rising risks of political games and can also benefit from the decrease in oil prices and inflation, which lowers the pressure for interest rate hikes, as well as the weakening of the US dollar and US Treasury yields.
However, the original issue with Hengke lies not externally but internally. The "advantage" is that it is low enough and has a small AI exposure. However, relying solely on improvements in the external environment and a rebound in capital reallocation faces resistance from fundamental pressures and the escalation of geopolitical risks. Over the past two days, during the rebound process, significant net selling from the southbound market and an increase in short selling in Hong Kong stocks indicate a widening of market divergences (《How Does the Market Find a Bottom?》). In contrast, gold faces less resistance, and various factors still provide support.
Technology Hardware: Short-term headwinds for overseas AI infrastructure, maintaining optimism for domestic computing power development during the 14th Five-Year Plan
Recently, driven by the fact that the growth rate of leading AI Lab ARR is slowing, overseas market investors have begun to worry about the unsustainable high investment returns of current AI infrastructure. They also hold a pessimistic attitude towards the expected growth of North American CSP Capex in 2028. After the mid-term elections in the United States on November 5, US stocks further reflected market concerns about the US government's increased AI regulation and the slowdown in AI infrastructure investment, with the Philadelphia Semiconductor Index dropping over 10% and the Nasdaq Index falling over 5%. Looking back at the A-shares, we see that since July, most Chinese AI hardware industry chain targets (including overseas chains and domestic chains) have seen a correction of over 50%. We believe that in the short term, the construction of North American data centers changes the slope of hardware demand but does not alter the long-term prosperity trend; domestic computing power demand may be even stronger.
From the perspective of overseas chain targets, we believe that the current valuation premium of related companies is significantly digested, and the mid-to-long-term prosperity of AI remains. The market's overselling reflects concerns about the pace of overseas AI data center construction. According to Wind data, we calculate that the current PE (TTM) of the Shenwan Communication Equipment Index is 28.92 times, which is below the 10th percentile of the past three years. We believe that the recent sharp decline in overseas AI stock prices has fully squeezed the valuation bubble. What remains to be clarified is the potential changes in the pace of overseas AI data center construction and its impact on the performance realization of core suppliers; however, after the short-term financial impact amplifies volatility, market concerns about the construction pace and profit realization have been significantly released. In terms of factual numbers, the total capital expenditure of the top four cloud vendors in the US for Q3 2026 increased by 84.2% year-on-year, while Microsoft Azure/AWS revenue also saw high growth year-on-year, indicating that demand continues to convert into revenue. Looking ahead, we believe that 2027 will still be a year of active investment for cloud vendors.
From the perspective of domestic chains, in terms of computing power chips and server systems, the related demand driven by domestic data centers remains strong. In the short term, the contradiction still lies on the supply side, and the adjustment of the development slope overseas provides a time window for domestic computing power hardware to catch up. We believe that the performance of related companies is expected to continue to be released. In terms of production factors, the core mid-to-long-term domestic substitution logic for wafer fabs, semiconductor equipment, and components remains unchanged. This year, affected by AI sentiment, the short-term price correction of the large manufacturing sector has been significant, and valuations are now attractive. We continue to be optimistic about the vigorous development of the domestic computing power industry chain during the 14th Five-Year Plan.
Risks
AI industry progress is below expectations; AI hardware domestic penetration is below expectations; policy changes.