The AI correction was already underway before the election made it worse

What fell, and how fast

Technology stocks led the market down on Thursday. The Nasdaq 100 fell by approximately 6.4% and the S&P 500 fell by approximately 4.4%, according to UBS, and the Philadelphia Semiconductor Index fell 12.3% on Friday. State Street records a similar Thursday for the broader market, with the Nasdaq Composite down 5.5% and the S&P 500 down 4.4% following the midterm results. Nordea notes the reaction was global: markets reacted to the Democratic midterm victory with a decline in technology and AI stocks. m:…1405 m:…9984 m:…2219

The selling was not confined to one day or one index. State Street notes that global semiconductors were already in a 20% drawdown before the midterm elections, and that semiconductor equities have been through meaningful corrections before, in 2022 and 2025. The correction, in other words, was underway before voters delivered the result the market spent the week repricing. m:…9984

The election was an accelerant

The midterm results gave the selloff a second push. The Democratic Party gained control of both the House of Representatives and the US Senate in the 2026 midterm elections, according to J.P. Morgan, and stock market futures plunged nearly 2% following the results. J.P. Morgan's team says Democratic gains increase the likelihood of tougher scrutiny around data centre development, electricity affordability and grid infrastructure costs. State Street puts it the same way: the election added political and regulatory uncertainty that accelerated a repricing already underway. m:…3249 m:…9984

Nordea's analysts think the market's reaction has gone too far. They describe the decline in technology and AI stocks as driven by fears of tighter regulation, and believe the current market reaction is too categorical, with the underlying fundamentals of technology and AI companies still strong. The disagreement is the story: everyone agrees the politics changed, and nobody agrees how much the politics change the economics. m:…2219

The argument underneath is about the buildout

The numbers behind the argument are large in both directions. UBS counts proposed US data centre capacity of 400 gigawatts against its own buildout assumption of 90 gigawatts by 2030. UBS still forecasts aggregate AI capital spending of $1.0 trillion in 2026, $1.4 trillion in 2027 and $1.5 trillion in 2028. UBS believes political opposition at a state and local level is unlikely to change underlying demand for data centres, and State Street notes that AI remains a strategic priority for both parties at the federal level. m:…1405 m:…9984

That gap between proposed capacity and assumed buildout is where the correction lives. If only a fraction of the proposed 400 gigawatts gets built, the companies geared to building it are worth less than the market assumed a month ago. State Street notes that Democratic control of Congress alone would not likely be enough to curtail the AI buildout in the face of a presidential veto. J.P. Morgan adds a nearer-term wrinkle: utilities with exposure to Texas and Georgia face near-term uncertainty due to AI-related power demand. m:…1405 m:…9984 m:…3249

What the big houses are telling clients

The advice converging from the big research houses is to stay in AI but spread the bet. UBS views the recent technology sector decline as a de-risking event rather than a fundamental change, and recommends reducing excessive single-stock concentration relative to technology and broad market benchmarks. It prefers diversification across foundry, compute, networking and power technologies over concentrated exposure to individual companies. State Street remains constructive on US equities with an emphasis on selectivity, and tells investors to look across the AI value chain for businesses positioned to translate structural demand into sustainable earnings. m:…1405 m:…9984

J.P. Morgan draws a narrower line: companies with operational data centres and secured power have a competitive edge in the current regulatory environment, and regulatory proposals are expected to target new data centre developments rather than existing ones. Hyperscaler investment in AI infrastructure is expected to continue despite the regulatory uncertainty, the bank says. UBS, meanwhile, maintains a relative preference for the Eurozone, emerging markets, and Asia ex-Japan. The trade, for now, is not about whether AI gets built. It is about who already owns the ground it gets built on. m:…3249 m:…1405