The midterms weren’t really about data centers. An AI researcher says voters want a pause on AI itself

What happened
During the 2026 US midterm elections, voters across the country supported measures to halt both data center expansion and the development of advanced AI technologies. Several political candidates campaigned on promises to impose immediate moratoriums on data center construction and AI development, with some winning key races. Lawmakers are now being urged to enact an indefinite pause on advanced AI, including superintelligence, due to security concerns and broad bipartisan support. A new research paper proposes a technical solution to enforce this pause by restricting the production of AI chips used for training new models.
Key details
- 40 out of 41 data center-related ballot measures opposed industry expansion.
- Gina Hinojosa won the Texas governor race on a platform calling for a data center moratorium.
- A September bill proposed banning superintelligence in the US.
- 70% of Democrats support a unilateral pause by American AI companies, even if China does not pause.
- A proposed 'hardwired' pause would replace current AI chips with 'inference-only' chips that cannot train new models.
Why it matters
US lawmakers and AI companies now face strong public pressure to halt the development of advanced AI, not just limit data center growth. If Congress enacts an indefinite moratorium, AI companies could be required to stop training new models and switch to using chips that only run approved AI systems. This would limit the risks associated with advanced AI development and could affect the business models and stock prices of major AI firms, while also addressing security concerns raised by voters and experts.
