The race to build artificial intelligence infrastructure has turned into a test of nerve for investors. On November 5, the Nasdaq fell 5.5 percent and the S&P 500 fell 4.4 percent, wiping about $2.5 trillion from the Nasdaq-100 in a single session, with Nvidia down 9.3 percent, AMD down 17.3 percent and CoreWeave down 9.8 percent. The AI investment boom is still running, but the selloff showed how fast the market can turn on the companies that have borrowed most heavily to build it.
What set off the AI investment selloff
The immediate trigger was politics. Democratic gains in the November 3 midterm elections raised the prospect that state governments could slow data center construction, after candidates who campaigned on pausing new projects won in Texas, Ohio and Georgia. Voters in several states had also pushed back against higher power bills linked to data centers. The AI investment selloff erased the market’s best day in three months, and the damage spread to Tokyo and Seoul the next day before AI shares recovered part of their losses on Friday.
Beneath the politics sits a funding question. Bank of America strategist Sebastian Raedler warned in a client note on October 16 that AI could succeed as a technology and still disappoint investors. His comparison looked back at past capital booms, where strong early profits encouraged heavy investment, capacity grew faster than profitable demand and pricing power weakened. The AI investment race, in his reading, carries the same risk: the technology works, but the returns may not match the money poured in.
How big the AI investment buildout has become
Consensus estimates put hyperscaler capital spending over the next 12 months at about $1 trillion, up from $330 billion in early 2025. That is roughly 3 percent of United States gross domestic product, against 2.2 percent during the railroad boom of the 1870s and 1.1 percent during the technology boom of the 1990s. Amazon, Alphabet, Meta, Microsoft and SpaceX are guiding to roughly $803 billion to $828 billion of calendar-2026 capital spending between them, and most of them have signalled higher spending for 2027.
Competition is also squeezing prices. Average AI token prices have fallen by about 50 percent since the summer, as Chinese open-weight models press the model layer and labs undercut each other for customers. That pressure matters because the AI investment buildout is being financed with debt as well as equity. Two investors at Wellington Management said at an event last week that debt markets have almost entirely shut off access to money for any company that does not have substantial revenue independent of AI data centers.
AI investment winners and losers so far
The Wellington investors, Matthew Witheiler and Jeff Courey, drew a line between companies that can fund themselves and those that cannot. Frontier labs and the largest cloud providers sit on the stronger side, along with fast-growing AI startups whose revenue is rising 150 percent a year or more. On the weaker side are infrastructure investors carrying large debt loads and companies that do not yet make a profit. Anthropic filed for an initial public offering on November 2, reporting about $17 billion of revenue in the third quarter and about $895 billion of commitments for chips, cloud capacity and data centers.
The money is still flowing into the biggest names. Accel is set to lead a $5 billion round into Thinking Machines Lab at a $40 billion valuation, Business Insider reported on November 6. The lab was founded by former OpenAI chief technology officer Mira Murati, and the round values it at roughly 400 times its reported revenue. SpaceX and Apollo announced a $40 billion chip financing deal on the same day Anthropic filed. The AI investment question is no longer whether the money is there. It is whether the returns will arrive before the debt comes due.
What companies are still spending
Enterprise buyers have not pulled back. A West Monroe survey released on November 5 found that 74 percent of enterprise buyers now allocate more than 10 percent of their information technology budgets to AI, up from 23 percent in 2025. Alphabet, Meta and Amazon pointed to higher 2027 spending in late-October earnings calls, while Microsoft trimmed its 2026 outlook to about $170 billion because third-party data center capacity was delayed, not because demand fell. The AI investment spending continues, and the market will keep testing whether it pays.