Soochow: After the Midterm Elections, Can You Bottom-Fish the Nasdaq?

Soochow released a research report stating that it may hold a cautious attitude toward bottom-fishing the Nasdaq index. Over the next 12 months, the growth rate of US AI infrastructure investment faces downward pressure, and earnings expectations for upstream hardware companies still carry adjustment risk. An index rebound may occur, but Soochow's Chen Li prefers to place new investment opportunities in downstream applications, especially To C AI Agents that can leverage cost-effective models to reduce service costs. If AI infrastructure investment truly slows, how much impact the US economy will suffer will also gradually become clear. The data provided by chief analysts at major overseas banks differ widely. Morgan believes that 67% of US economic is related to AI, while Goldman Sachs believes that at most only 10% is contributed by AI. The coming year will be a good test of their views. Whether slower AI growth is enough to push the US quickly into recession cannot yet be concluded. The key lies in whether consumption, services, and other business investment can absorb this shift in demand, and whether financing pressure spreads to more sectors. Soochow's main views are as follows: Today's consensus and variables after the US midterm elections On November 4, 2026, the first trading day after the election results were announced, the Dow rose about 3%, the Nasdaq rose about 1.4%, and WTI crude oil futures fell more than 7%. The market expects that a Republican defeat could bring policy adjustments, and the political cost of the Iran war is also drawing more attention. The next day, the Nasdaq fell 5.5%, and the semiconductor index fell about 12%. Investors focused their discussions on how much resistance Democratic victories in Congress and state-level elections would create for data center construction. Expectations for de-escalation of the war have their basis. Voters are dissatisfied with the cost of living, and the government will face greater political difficulty in continuing to bear war expenditures and energy price increases. But how the president responds to defeat and how Congress uses appropriations and investigative powers still need to be observed. Iran also has its own military actions and negotiating demands. Even if the US increases its willingness to end the war, energy supply and transportation order may not recover immediately. The current situation in the Strait of Hormuz fully demonstrates that this is no longer something the US can decide unilaterally. Even if the US unilaterally exits the battlefield, Iran, Saudi Arabia, Israel, Palestine, and other forces may still keep the region in persistent turmoil. On AI, market concerns have already translated into projects and financing. One investment bank estimates that a potential construction pause could affect about 30% of the expected new US data center capacity in 2028. Approval delays may lead to project relocation and may also prompt companies to increase spending on energy, water resources, and community compensation. Facilities that have already obtained permits and connected to power may even become more scarce as a result. But for companies reliant on external financing, delays are indeed dangerous. The data center has not yet been put into use, but loan interest still has to be paid. On November 5, the spread on CoreWeave's five-year credit default swaps rose to about 1,289 basis points, and the stock fell about 10% that day. This set of trading data shows that investors are already demanding higher compensation for financing and debt repayment risks. Wider credit spreads may in turn further raise corporate financing costs. The entrepreneurs who have become global "heroes" in the AI industry may need to take a closer look: whether their business activities truly provide benefits to ordinary people, rather than merely raising their water, electricity, gas, and oil prices. Of course, Democrats are not opposing AI for the sake of opposing AI. If AI companies put up real money to promote local employment, no one will turn away money. But this inevitably means companies need larger-scale funding. The rates market, meanwhile, remains focused on inflation. Employment data released on November 6 was stronger than expected, with the unemployment rate falling to 4.1%. The market-implied probability of a December rate hike rose from about 70% to 76%. But if AI projects slow markedly, equipment procurement, construction activity, and funding demand will all be affected; if the war eases, the energy shock may also weaken. At the same time, wages and service prices are sticky, and falling oil prices may not be enough to make the Fed pivot quickly. Views on the US economy are therefore divided. Consumption and employment are resilient, supporting the view that the economy can withstand part of the AI investment adjustment. But shrinking stock wealth, tighter financing, and delayed hiring by companies may in turn affect household spending after a few months. The market's current concern about cooling AI investment and persistent rate-hike expectations can coexist in the short term; if the investment contraction continues to expand, the US economic outlook for 2027 will be clouded. A chance to bottom-fish the Nasdaq? Could it be AI applications Over the next 12 months, the growth rate of new US AI infrastructure investment will face greater slowdown pressure. Higher interest rates, energy costs, and changes in the political environment will raise the effective investment threshold for projects. This will affect earnings expectations for upstream hardware companies. I prefer to focus on companies that can generate downstream application revenue, especially AI Agents aimed at individual users. There is another important reason for favoring this direction: the public needs products that are reasonably priced and can reliably complete tasks, and does not necessarily need every service to call the strongest and most expensive model. We expect that as products gradually mature, To C applications will place greater emphasis on the cost-effectiveness of models, and lightweight, flexible small models will be used more widely. For application companies, how much it costs to complete each task is often more important than how many parameters the underlying model has. If users pay the same amount and the cost of model calls and task execution falls, companies have the opportunity to retain more profit and can also expand their user base by lowering prices. This is also why I believe investment opportunities may shift downstream. In the past, the market focused on continuous breakthroughs in model capabilities and how many chips and how much power those breakthroughs required. Next, part of the commercial value may come from another capability: turning existing model capabilities into services that the public is willing to buy at lower cost. AI short dramas, video production, and other auxiliary applications are similar commercial opportunities. However, different tasks have different requirements for model capabilities, and it cannot be assumed that all applications can be completed with small models. The common direction is to reduce the cost of providing services through model selection and process optimization. Therefore, we may hold a cautious attitude toward bottom-fishing the Nasdaq index. Over the next 12 months, the growth rate of US AI infrastructure investment faces downward pressure, and earnings expectations for upstream hardware companies still carry adjustment risk. An index rebound may occur, but I prefer to place new investment opportunities in downstream applications, especially To C AI Agents that can leverage cost-effective models to reduce service costs. If AI infrastructure investment truly slows, how much impact the US economy will suffer will also gradually become clear. The data provided by chief analysts at major overseas banks differ widely. Morgan believes that 67% of US economic is related to AI, while Goldman Sachs believes that at most only 10% is contributed by AI. The coming year will be a good test of their views. Whether slower AI growth is enough to push the US quickly into recession cannot yet be concluded. The key lies in whether consumption, services, and other business investment can absorb this shift in demand, and whether financing pressure spreads to more sectors. At the end of 2006, strong employment and consumption convinced investors that the housing problem could be absorbed. As it turned out, the housing market receded and dealt a severe shock to the financial system, ultimately plunging the US economy into a deep recession. These issues will add great uncertainty to the stock market in 2027.