The midterms weren’t really about data centers. An AI researcher says voters want a pause on AI itself
The 2026 midterms have already been read as a revolt against data centers. But voters weren’t only objecting to higher power bills and new server farms down the road. They want a pause on AI development itself, and that is the policy the new Congress should take up.
That part of the story is familiar. Across America, shoo-in candidates were caught flat-footed by the populist backlash to data center buildouts. Greg Abbott, until now a popular governor in a state Republicans have held for 32 years, was unseated by Democrat Gina Hinojosa, who focused her campaign on an immediate moratorium on data center construction. Meanwhile, vulnerable incumbent Kelly Ayotte adopted the call for a moratorium and won her reelection bid.
Of the 41 data-center related ballot measures, 40 went against the industry. But the revolt isn’t just about data centers. A pause on AI development itself is now the policy favored by Americans of all political stripes, and Congress has a duty to produce one. We need an immediate, indefinite, and international moratorium. The fate of humanity may well depend on it. Despite competition from Chinese AI, most Americans even support American AI companies pausing unilaterally until Congress acts, including 70 percent of Democrats.
In September, many dismissed Sen. Bernie Sanders and Rep. Greg Casar’s bill to ban superintelligence as a messaging exercise. But in the midst of a growing cybersecurity crisis that experts warn could herald a similar biosecurity crisis, it is increasingly clear that politicians must act to stop AI companies from further endangering constituents.
By the time the new Congress convenes in 2027, the conversation may no longer be about whether to pause AI, but how to pause it. Rep. Yassamin Ansari spelled it out: “Top of the Democratic agenda on day one needs to be a pause on advanced AI development, regulation of this out-of-control industry, and a ban on artificial superintelligence.” And she is not alone.
This conversation is overdue. For too long, many commentators have insisted that AI is inevitable. But AI development isn’t a force of nature; it is a multi-trillion-dollar investment in what are arguably the most sophisticated physical devices in existence: AI chips. AI development is driving the data center boom, but a moratorium on new data centers isn’t enough.
I’m convinced we can stop AI, but we need to target the chips themselves. A new research paper that I co-authored with a multidisciplinary team of 26 researchers explains how. What’s more, we can do it without trusting China or tanking the U.S. economy.
The central idea of our paper is called a “hardwired” pause. The idea is to halt the production of chips used for training AI, and instead build “inference-only” chips. Whereas current AI chips can be used to train or run any AI model, these inference-only chips are designed to do one—and only one—job: Run approved AI models. This means they cannot be used to advance the frontier of AI capabilities, even if stolen, seized, or smuggled abroad.
Our plan addresses one of the primary arguments against an AI pause: That adversaries would continue development in secret. Replacing existing AI chips with inference-only chips would make this much harder. Governments would need to either hide a large quantity of existing chips or make more chips in secret. However, our analysis suggests more than 99.9 percent of existing chips could be located and governed.
Making chips in secret would be extremely difficult due to the complexity of the technology and the supply chains involved. Similarly, a hardwired pause would provide security in case a rival power suddenly decided to rip up the agreement.
Meanwhile, the economic impacts of a hardwired pause would be limited by the continued availability of existing AI models. Inference-only chips could also run these models much faster and more efficiently. While AI companies’ stock prices might decline, monetary policy could mitigate a resulting decline in spending, and this seems unlikely to trigger a broader recession under current conditions.
It is also important to weigh a moratorium’s economic impacts against the harms it would mitigate. Besides the possibility of human extinction, which experts and whistleblowers keep warning about, superhuman AI would bring unprecedented levels of unemployment and existential threats to our democracy and our way of life.
As these dangers become clearer, Americans are increasingly demanding a proportionate response. President Trump’s AI accelerationist stance may have cost Republicans in the midterms, but his upcoming meetings with President Xi Jinping present an opportunity for a handshake agreement to pause AI before something unprecedentedly catastrophic occurs. Forward-thinking leaders in both parties (and globally) will continue to benefit politically by recognizing the risks from AI and the need for urgent action.
Ultimately, no one will benefit if AI development continues unabated. Why are we still allowing AI companies to endanger everyone and everything we care about?
The only conceivable justification? If we don’t do it, someone else will. Our research undercuts this dangerous line of thinking.
The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.
This story was originally featured on Fortune.com