While the country counted congressional seats on Tuesday, it was also voting, town by town, on the other great construction boom of the decade: the data centre. The ballot trackers that follow the issue counted the night the same way whichever set of measures they used. Redbud Consulting’s tally, cited by Fast Company, had 45 data-centre propositions decided on Tuesday and the restrictive side ahead in 44 of them — 44 to 1. The “Data Centers on the Ballot” tracker, with 47 measures identified across 10 states, found the same 44-to-1 split among the races it could score on Wednesday. Ballotpedia’s narrower set of 41 measures in 10 states finished at 40 to 1. However the measures are counted, the shape of the result does not change: in ban votes, moratoriums, size caps, rezoning fights and deal repeals, the side trying to limit data centres won essentially everywhere. Nothing statewide was on offer — an Ohio constitutional amendment to ban facilities above 25 megawatts failed to gather the 413,487 signatures it needed by July, and its backers are aiming at 2027 — so the revolt was fought, and won, entirely at the level of the village board and the county ballot.
Ohio, with 21 measures, was the centre of it, and the development side lost all 21. Voters banned larger facilities outright in town after town, by margins that were not close: Hubbard at 76.6 percent (2,622 votes to 802), Sunbury at 79 percent, Pataskala at 74 percent, Granville above 82 percent, Trenton and Urbana around 80 and 77 percent. In Ashville, 82 to 83 percent voted to repeal the village’s own deal terms with the developer EdgeConneX for two data centres and an 800-megawatt gas plant on 195 acres. Grove City voted 55 percent to require a public vote on any development over 50 acres, 20 megawatts or 500,000 gallons of water a day; Perry Township voted by more than 87 percent to require financial guarantees from developers seeking tax breaks; Wilmington rejected all four rezonings near a planned Amazon Web Services campus by roughly 80 to 20. In Piqua, voters went further and recalled the mayor and two commissioners over a Meta data-centre fight. Kansas passed all three of its bans — Americus outright, Emporia’s “high impact” ban by 60 to 40, El Dorado’s for ten years. Maine towns approved moratoriums and an outright ban in North Berwick; California’s Alhambra backed a citywide ban with about 82 percent in early counting; Janesville, Wisconsin, voted 62.8 percent to restrict. The arguments that won were concrete and local rather than ideological. A research director at Blue Rose Research, Ali Mortell, told Time that “concrete local impacts” — higher energy bills, pollution — “were far more salient” for voters than abstract fears about artificial intelligence; the same firm found Americans would rather live near a prison, a methadone clinic or a nuclear plant than an AI data centre. NBC News exit polling found 76 percent of voters opposed to data centres in their communities. Forbes has reported that the regional grid operator’s capacity auctions have added about $16 a month to the average Ohio household’s electricity bill, with more than 60 percent of the increase attributed to projected data-centre demand.
The one exception proves what the rule was made of. In Harwood, North Dakota — population about 940 — voters rejected, 325 to 105, a measure that would have given residents a greater say over large construction projects. Harwood was not voting on a hypothetical: a $3 billion AI data centre, the size of 17 football fields, is already under construction nearby, and the developer hopes to employ 200 people — a fifth of the town. (The measures passed elsewhere generally cannot touch projects already permitted.) The industry’s answer came from Adam Kovacevich of the Chamber of Progress: “The results offer little evidence of a voter mandate to crack down on AI or data centers,” he argued, noting that Democratic gubernatorial candidates who opposed bans won at a higher rate than those who backed them. The issue has a congressional echo, though. In Texas, Governor Greg Abbott campaigned on making his state the industry’s capital and lost; Governor-elect Gina Hinojosa has pledged a moratorium on new data-centre development until the legislature deals with the water, electricity, air and property-rights impacts. Texas was expected to supply more than 41 percent of all added American capacity in 2028. The centres themselves are not going away — the demand for computing power has not voted on anything — but after Tuesday, in most of the places that were asked, the answer to building them next door is no.
The Facts
- The counts: restrictive side ahead in 44 of 45 decided propositions (Redbud Consulting via Fast Company); 40 of 41 in Ballotpedia’s set. All unofficial.
- Ohio: 21 measures, 21 losses for the development side — bans passed by 74-83% in town after town; Piqua also recalled its mayor.
- The exception: Harwood, North Dakota (pop. ~940) rejected a greater-say measure 325-105; a $3 billion project is already being built nearby.
- The winning arguments: power bills and local impacts — NBC exit polling found 76% opposed to data centres in their communities.
- Texas echo: Abbott, the industry’s champion, lost the governorship; Governor-elect Hinojosa has pledged a moratorium.
Sources: Fast Company and Stateline reporting of the Redbud Consulting and Data Centers on the Ballot trackers; Ballotpedia; the Columbus Dispatch and Cincinnati Enquirer via the trackers; Time; Forbes; NBC News exit polling; the North Dakota Secretary of State’s results for Harwood.





Comments