The Zhitong Finance App learned that Shen Wan Hongyuan released a research report saying that in 2026, US stocks revolved around the two main lines of AI development and the Middle East conflict. The main index rose by about 10%. The market moved from K-type differentiation to convergence. Energy led the whole year, and the global developed funds allocated US stocks had reached the 100% quantile of history. Looking ahead to 2027, the slowdown in AI investment growth is dragging down the economy to a marginal downturn. Monetary policy is still tight but loose expectations may ferment ahead of schedule, and the risk of the political cycle will rise after the midterm elections. In terms of market pricing, the AI hardware supply shock has been gradually interpreted. Semiconductor valuations have declined from 25x to 14.58x, but the contradiction between computing power supply and demand and the risk of a negative cycle in AI financing have not been fully priced. In terms of structure, it is recommended to focus on low-beta defensive pharmaceutical consumption, AI applications and data collection topics, and policy opportunities such as new energy sources after the midterm elections.
Shen Wan Hongyuan's main views are as follows:
Since 2026, US stocks have revolved around the development and changes of AI itself and the two main lines of the Middle East conflict
1) Looking at the whole year, with the exception of the small market, market growth and market value all closed up with profit absorption valuations, and the increase in major indices was concentrated around 10%. 2) It has gone through a process of K-type differentiation to convergence, from AI computing power in the first half of the year to AI applications in the second half of the year, while energy led the year; 3) US stocks followed the rise in the Asia-Pacific market more in the first half of the year. US stocks outperformed the non-US market significantly in the second half of the year. As of 2026Q3, the allocation coefficient for global development funds to US stocks had also reached 100% of the historical quantile.
Looking ahead to 2027, US investment will drive endogenous demand to slow down, and improvements in potential demand will increase sensitivity to liquidity and policies
1) On the economic side, after the growth rate of AI investment slowed, the economy was marginally downgraded, waiting for squeezed non-AI demand to be gradually released. The US stock industry and consumer capex is already below the historical 20% quartile, and residents' credit has also declined to a low level. The marginal easing of real interest rates will help improve non-AI manufacturing and mitigate “carbon-based” consumption. 2) In terms of monetary policy, the lagging impact of AI investment on the economy still marginally supports the Federal Reserve's austerity, focusing on the risk of upward liquidity caused by changes in the monetary policy framework, while uncertainty about oil prices is still a potential constraint. Looking at the review history, it took at least 9 months for the first interest rate hike to the inflection point of liquidity expectations. It was less than 1 month in 1997. At that time, US stocks faced a financial crisis in Southeast Asia after fluctuating and rising, and the Federal Reserve then began cutting interest rates. Currently, the impact of AI on terminal service prices is biased but the total impact is limited. If the Federal Reserve emphasizes the evaluation of AI's long-term deflation trend in the future, it may lead to early fermentation of loose expectations.
The impact of political cycle risk on the market will be higher than in 2026
Under the benchmark hypothesis, the White House's substantial influence on foreign affairs, tariffs, and immigration policies is limited by congressional interference, and industrial policies (AI, medical care, new energy, etc.) exist as a basis for a game with Congress. 1) Lowering the cost of living is a potential foundation for cooperation between the two parties, but the substantial impact of differences in opinions should not be overestimated. The inflation issue is not only the reason Trump won the 2024 election, but also the reason why the Democratic Party swept Congress in the current round. The market's first reaction after this election was pricing, which also saw a rise in consumption, retail, and aviation, leading the decline in construction projects, power equipment, and semiconductors. However, historically, the industry performance on the first trading day after the midterm elections had a high probability of reversal within three months. Continuity over the next 12 months was relatively higher, but the impact was also slightly higher than the random probability. In the future, the two parties will focus on energy diversification, domestic infrastructure supply such as electricity, and housing, as well as negotiation space for external tariffs and the US-Iran conflict; 2) There is a cooperative and competitive relationship between the two parties in AI construction, and it is likely that construction costs will rise and regulations will become stricter. The two parties will agree that the current increase in electricity and hydraulic costs brought about by the construction of the data center itself will be agreed upon. However, AI regulations are likely to become stricter, and the Democratic Party is more cautious than the Republican Party on “AI frontier security” and “computing power trade.” After 1998, based on the Microsoft antitrust case, the National Assembly held several hearings, causing market sentiment to fluctuate. 3) American populism intensified, and the conflict between the two parties gradually escalated or led to an impact on market risk appetite. Currently, the division between US capital and workers' compensation has reached its greatest extent since 1960, and distribution conflicts may make it more difficult for the two parties to reconcile debt issues. After the president lost Congress in 1994, two government shutdowns in the fourth quarter of 1995 caused market turmoil. Meanwhile, in 2010, the president formed a divided National Assembly, and a game of debt ceilings and fiscal contraction throughout 2011 triggered a decline in corporate profits.
Judging from market pricing, the current supply impact of AI hardware on short-term data center construction has been gradually interpreted
As of 2026/10/30, the dynamic valuation of US semiconductor products and equipment was 14.58x (the high at the beginning of the year was 25x), and the bottom of the past cycle was about 10-12x. Profit expectations for 2027 are still marginally revised, but the growth rate is close to a standstill. Historically, stock prices have often gradually bottomed out after profit expectations declined. On the NASDAQ side, after considering the two companies Anthropic and Open AI, the NASDAQ's static PE valuation is 30.2x, which is in the historical 47.1% quantile. The 2027 dynamic PS valuation is 5.25x, which is in the historical 69.4% fraction. The profit change itself is still the key to the subsequent trend of the market.
However, the turbulence in computing power prices caused by potential supply-demand contradictions in 2027 has yet to be fully priced. Under a neutral assumption, the US computing power supply will increase by 50 GW in 2027, and the supply growth rate is higher than this year. On the demand side, Anthropic's revenue increased tenfold this year, and next year it may slow to 2 to 3 times, and demand growth will slow marginally. Qualitatively speaking, AI-deep customer token spending increased sharply in the first half of the year, and stagnated relatively in the third quarter. The increase in AI revenue depended on the spread of paying users. Token prices fell 50% this year due to advances in computing power and model competition. Currently, GPU rental prices are still strong, and the future is the core indicator for observing changes in the supply and demand pattern.
The tail risk is concerned about the macroeconomic second-order impact caused by rising financing restrictions
AI capital expenditure was mainly based on cash flow support in the third quarter to external financing. According to the current Capex pace, it is expected that the total cash flow of the North American cloud factory cannot be corrected in 2027. Currently, the bond market has begun to price the risk of project default. Compared with 2008, as of 1105, Coreweave's CDS (>1200bp) has surpassed the CDS (600bp) during the Lehman crisis, while the average CDS of AI technology companies has risen from 60 bp to the current 133 bp. At that time, the banking industry's CDS rose from a high of 60 bp to 200 bp. At the macro level, the worsening of bank credit risk in the third quarter of 2007 led to a widening of cross-industry credit spreads, and broad-based fundamentals began to weaken substantially in 2008. Currently, industrial credit risk has emerged, but BAA-level interest spreads are still narrowing. The macro impact may be less than in 2008, but the signal and logical trends are still meaningful. Focus on the formation of a possible negative circular fluctuation chain in the future: rising financing costs → cutting marginal projects → declining capital expenditure growth → declining supplier profits → falling collateral (GPU) valuations → further increases in financing costs.
In terms of pace, the short to medium term market is still in the digestion stage after the momentum is reversed. Liquidity expectations are in a transition period from austerity to easing, waiting for the emergence of positive AI externalities for a long time
US stocks entered the digestion stage after the mid-year momentum reversal in 2026. Judging from the speed of digestion, it was not enough, and the market fluctuated and settled in a way that spread widely. After a reversal in technological momentum in 2000, the S&P 500 fluctuated at a high level for 2 quarters. During this period, finance, industry, utilities, and energy remained resilient until fundamentals deteriorated and overall profits downgraded to a correction in the market. According to bottom-up expectations, the profit growth rate of the S&P 500 index will slow from 32.9% in 2026 to 16.3% in 2027. The low profit growth rate may occur in 2027Q2, so pay attention to changes in liquidity when profit expectations slow down. In the long run, AI is a productivity tool. Currently, the US Fortune 500 companies account for 0.1% of AI spending. Looking at the scenario, according to the final consumption classification of computing power disclosed by Nvidia, the proportion of internal use of CSP is 42%, the proportion of cutting-edge models is 18%, traditional enterprises are 12%, edge computing is 7%, and AI-native enterprises are 22%. Among them, enterprise AI and edge computing correspond to enterprise-level agent and robot scenarios, respectively, and there is still room for improvement.
Structurally, judging from profit growth expectations, the industries with the most obvious improvement in 2027 are healthcare and industry
The growth rate of all other industries will slow marginally next year, information technology will slow down at a high level, and communications and energy may enter a negative growth range. In terms of potential catalysts, it is recommended to look for relevant opportunities around low-beta market styles, AI themes, and post-midterm election policy preferences. 1) The low beta market style focuses on defensive medicine and consumption. As a direction where the market was low in the early stages, pharmaceuticals have now received an inflow of capital on the right side, while consumption is still flowing out marginally. Looking at the breakdown, as AI for Science accelerates, pharmaceutical companies' R&D needs are still spreading, and the life science industry is benefiting even more. The profit peak of medical equipment has passed, and it has a certain valuation and cost performance ratio. Like hospitals and health insurance companies, it depends on the Democratic Party's policy preferences after winning the election. 2) In terms of AI topics, the value of data collection in downstream scenarios may be highlighted, and application-side opportunities are more worthy of attention. In terms of penetration of AI applications, from tool payment to workflow substitution to macro-level diffusion, the US subclass of applications is entering the second phase. In the primary market, financing for enterprise process support, industrial manufacturing and operation and maintenance, and medical technology is more active. However, no company has yet replicated the ARR growth rate in the early days of the coding circuit (ARR for vertical applications was around $10 billion). Currently, the increase in the penetration rate of both physical AI and enterprise-level AI is at a stage of development where insufficient data limits the effectiveness of models. Software companies focusing on data collection and data distribution are worth paying attention to. To the C-side, the overall PB valuation of consumer electronics is at a median level. With the release of cost pressure, revenue and profit margins are expected to improve marginally next year. Furthermore, companies that control user traffic entrances (such as social networking and retail) are still expected to grow nonlinearly in AI revenue. 3) In terms of policy direction, electricity investment and new energy opportunities may once again deserve attention after the Democratic Party came to power, and defense and banking are facing headwinds. The Democratic Party advocates clean energy, and the restoration of tax credits requires the President's cooperation, making it difficult to implement them in the short term. If the Democratic Party is more proactive at regulatory levels such as fuel vehicle emission standards, energy storage, photovoltaics may usher in marginal restoration in valuation. Furthermore, investment in power grids and nuclear energy is also a long-term direction that both parties agree on.
Risk warning: 1) Be wary of the risk of excessive computing power in the US; 2) In the process of AI technology evolution, due to slow redistribution of wealth due to changes in corporate organizational forms and rising regulatory and compliance costs after increasing negative social externalities, macroeconomic demand is difficult to expand effectively; 3) The Middle East conflict has continued beyond expectations, and stagflationary pressure has increased, and US fiscal expansion is already limited. Market sell-off of the bond market has triggered a liquidity crisis; 4) Upgraded geographical protection measures trigger a liquidity crisis; 4) Upgraded geographical protection measures trigger a liquidity crisis; The rise had an impact on profit expectations.