70 percent. That's how many Americans oppose AI data center construction in their communities, according to a new Gallup survey. Only 7 percent said they "strongly" support new facilities. And the survey's sharpest finding: Americans would rather live near a nuclear power plant than a data center.
That's not a fringe view. Those numbers suggest that as AI companies race to build infrastructure, they've lost the public on the question of where it goes.
The opposition has real grounding. Data centers consume significant amounts of electricity and water - a single large facility can draw millions of gallons daily for cooling and enough power to serve a small city. They generate construction jobs and local tax revenue, but relatively few permanent positions for the scale of land and resources they occupy.
The nuclear comparison stands out. Nuclear power carries decades of accumulated public anxiety - Three Mile Island, Chernobyl, ongoing debates about waste storage. For Americans to prefer nuclear facilities as neighbors over AI data centers signals that the economic case for data centers simply hasn't landed with the public.
Some of this is a visibility problem. A factory that opens in a town is legible - residents can see what it makes, who works there, how it connects to local life. Data centers are architecturally opaque: massive, windowless buildings with no obvious relationship to the product they enable. The fact that a ChatGPT query runs through a facility in someone's neighborhood isn't something most users think about, but the people living near these buildings feel the noise, the truck traffic, and the pressure on water and power grids regardless.
Tech companies have mostly responded to community opposition with economic arguments: investment, tax base, grid upgrades. Seven percent "strongly" in favor is not a gap that more economic messaging will close.
The survey arrives as the federal government is pushing to accelerate data center permits, including executive action aimed at fast-tracking approvals on public lands. That policy direction is now running directly against public opinion data showing broad resistance across demographics. For AI companies, this is a slow-building problem - one that won't affect a product launch next quarter but will increasingly shape where and how quickly new infrastructure can be built.