Wall Street's AI trade cracked on Thursday. The Nasdaq Composite fell 5.52% to 24,315.00, its steepest one-day drop since April 2025, and closed in correction territory, more than 10% below its October 6 record. The S&P 500 lost 4.36% and the Dow fell 3.44%, and Reuters calculated the session erased roughly $3 trillion in value from S&P 500 companies. Chip and data center names took the worst of it. Nvidia fell about 9%, AMD dropped more than 14%, Micron slid 15%, CoreWeave lost almost 10%, and the Philadelphia Semiconductor index plunged 12%.
The trigger was politics. Democrats swept Tuesday's midterms, flipping both chambers of Congress and taking governor races including Texas, where governor elect Gina Hinojosa has promised to halt new data center construction. Investors read the result as a vote against the power hungry, water hungry infrastructure behind AI. UBS strategists warned that nearly 30% of expected 2028 data center additions could be delayed or cancelled, and the selling spread to Asia on Friday, when Seoul halted KOSPI trading for 20 minutes after an 8% fall and Samsung Electronics and SK hynix sank more than 9% each.
For founders the read through is capital, not capability. Model and tool releases kept shipping this week, from OpenAI's GPT-6 Cyber security model to Google's cheaper Gemini Omni 1.2 Flash video model and Thinking Machines' new Inkling 2 Small. But the cost of borrowing for anyone who builds or rents AI compute just went up, and UBS expects a chunk of future data center capacity to slip. Expect a harder funding market for compute heavy startups, closer scrutiny of where data centers get built, and a premium on teams that can show revenue without owning the infrastructure underneath it.
