"We cannot sit back and allow a handful of billionaire Big Tech oligarchs to make decisions that will reshape our economy, our democracy and the future of humanity." That's Senator Bernie Sanders, announcing legislation on Tuesday that would halt all new data center construction in the United States.
The bill, introduced with Rep. Alexandria Ocasio-Cortez filing a companion version in the House, proposes a national moratorium on building new data centers. The freeze would stay in place until Congress passes federal AI legislation covering worker protections, consumer safety, environmental standards, and civil rights.
There's no sunset clause or automatic expiration. The moratorium lifts only when comprehensive AI regulation is enacted, which, given Congress's pace on tech legislation, could mean a very long wait.
What's Driving This
Sanders cited three main concerns: surging electricity demand from data centers driving up power costs for regular customers, massive water consumption for cooling, and AI's potential to eliminate "tens of millions of blue- and white-collar jobs." The announcement follows Denver's recent local moratorium on data center construction, signaling that these concerns are building at multiple levels of government.
The energy math is real. AI training runs and inference workloads are power-hungry, and utilities in data center hotspots like Northern Virginia are already struggling to keep up with demand. But a blanket federal construction ban is an extremely blunt instrument for a nuanced problem.
Practical Impact
This bill faces long odds in the current Congress. But it shifts the Overton window on data center regulation and gives cover to local governments considering their own restrictions. For AI tool users, the relevant signal is that the political backlash to AI infrastructure buildout is intensifying. If data center capacity does get constrained, whether by federal action, local moratoriums, or simple power grid limitations, that eventually shows up as higher prices or longer wait times for AI services.
The broader pattern here is clear: AI companies spent 2024 and 2025 racing to build compute capacity. 2026 is when the political system starts pushing back.