Key points
- Exit polling found 56% called AI and data centers a major factor, 76% opposed a local data center.
- Roughly 45 gigawatts of data center capacity is due next year, with 10 to 15 gigawatts in Texas at risk.
- Elon Musk spent $470–500m this cycle for almost nothing, while tech gave over $800m of the top 20's $2bn.
Tech's midterm bet fails as AI backlash bites
The tech right spent hundreds of millions on the 2026 midterms and lost, and the biggest reason is AI. Data centers and artificial intelligence became unpopular in a way that cut across normal party lines, and the industry's assumption that it was too rich to need persuasion proved wrong. Ezra Klein argues the pitch itself is the problem, since it promises possible extinction and probable job loss from people who do not seem trustworthy. The result is a slower, more expensive build-out and a Democratic Party where taking on billionaires is ordinary.
Takeaways
- The tech right's AI agenda became its liability. Exit polling found 56 percent of voters called AI and data centers crucial or major to their vote, and 76 percent opposed a data center. On AI's effect, 57 percent said it would hurt society against 24 percent who said it would help.
- Data center opposition threatens gigawatts of compute. Erin Griffith reports investors are terrified, especially about Texas, where a moratorium was expected to be lifted. Roughly 45 gigawatts are due online next year, and 10 to 15 gigawatts in Texas could be at risk.
- Musk's money was hugely counterproductive. He gave roughly $470m to $500m this cycle. His Wisconsin bet on Brad Schimel lost, and allies had among the worst losses of their lives. The Texas race for Ken Paxton consumed about half his spending.
- Tech's assumption it need not persuade failed. Erin Griffith says VCs argued revealed preference settled the question because people use Facebook, Amazon, iPhones and ChatGPT. The election disproved that, and tech's belief that it was too rich and powerful to persuade anyone met its backlash.
- Billionaires spent $330m to kill a wealth tax. California's proposition on a couple of hundred of the state's wealthiest residents failed, with Gavin Newsom opposed. Sergey Brin put more than $100m into Building a Better California, and the fight consumed $330m in total.
- A slower build-out could help the AI trade. Ezra Klein's contrarian case is that constrained compute might be good for markets, could even mean the end of Oracle, and could push Anthropic and OpenAI toward diffusion and economically useful deployment instead of racing to superintelligence.
What comes next
- Ezra Klein predicts a wealth tax will be table stakes for Democratic candidates in 2028.
- Teddy Schleifer expects Elon Musk to back JD Vance in 2028, even after Musk's money did little this cycle.
- Schleifer is watching AI-safety money in the Anthropic and effective-altruist Democratic world, with Dario Amodei possibly on the 2028 donor list. The Anthropic IPO will mint billionaires.