75% of Americans Oppose Local Data Centers. The Reason Isn't the Buildings.

Roughly three-quarters of Americans now oppose building a data center in their own community. A Gallup survey conducted March 2-18, 2026 found 71% opposed (48% strongly), and a Zvi Mowshowitz / Cremieux analysis of public polling pegs the figure closer to 75%. Both numbers land in the same neighborhood as the opposition that greeted new coal plants a generation ago.

The puzzle is that the offer on the table has not changed. Data center projects typically arrive with construction jobs, county tax revenue, and the kind of new grid capacity that state capitals are publicly begging for. The polling barely budges whether a project brings those benefits or not, which suggests the objection is pointed at something other than the physical building.

That something is the bundle of associations a data center carries: AI, big tech, big money, and a wave of industrial-scale new construction. The strongest evidence is a comparison Zvi and Cremieux draw against transmission lines. New high-voltage power lines attract a similar level of public rejection to new coal plants, even though no one sees or hears them the way they would a smokestack. What residents are rejecting is not the physical footprint. It is who is building, how, and why.

The rejection is now showing up in project tracking. NBC News, reporting on a Q1 2026 study from Data Center Watch, counted at least 75 projects blocked or delayed in the first three months of 2026, totaling roughly $130 billion in shelved value. That is the most in any quarter since Data Center Watch, a project of AI research firm 10a Labs, began tracking in 2023. Q1 2026 alone roughly matched the entire 2025 total.

Active opposition groups more than doubled to 833 across 49 states, and moratorium legislation was introduced in 14 states between January and March. A federal version filed by Sanders and Ocasio-Cortez picked up bipartisan co-sponsors, an unusual coalition given the rest of the AI policy map. The tracking is industry-adjacent rather than neutral government data, but the trend lines up with both the public polling and the on-the-ground reporting from places like Heatmap's coverage of contested projects in Virginia, Texas, and the Pacific Northwest.

None of this dismisses the legitimate local concerns driving many of these fights. Data centers draw real water in already-dry regions, push audible noise into nearby neighborhoods, and stack tax abatements that often pay back slower than the promised jobs. Where those costs land, the residents who live with them are not wrong to push back, and treating the opposition as purely symbolic would miss the people doing the actual organizing.

The question the polling opens is whether the standard concession package can ever close the gap. Concession payments, community-benefit agreements, and grid upgrades are the currency local negotiations have run on for a generation. They were built to answer the question "what does this cost us?" Data center opposition is, increasingly, an answer to a different question: "should this be built at all, by these actors, in this form?"

That second question does not have a polling-shaped resolution. It has a political one, and it is being answered state by state, county by county, project by project, on terms the deal-makers in the room did not expect to be negotiating on. The next concrete data point to watch is whether any of the 14 state moratorium bills clear a chamber before summer recess, and whether the federal bipartisan version attracts enough co-sponsors to force a hearing.