Data centers were some of the biggest losers in last night's US midterm elections

On Tuesday, tens of millions of Americans voted in midterm elections to deliver a sweep of generational Democratic victories in federal and state races, reflecting a leftward electoral shift that's being interpreted as a nationwide rejection of Trumpism.

But even as Republican candidates and incumbents suffered defeats in what had traditionally been considered some of the party's most entrenched strongholds, the midterms also issued a stinging rebuke to another major source of voter frustration: Data centers. At nearly every opportunity, Americans voted in favor of local ballot measures to halt, slow, or stymie the pace of data center construction in their communities.

As tracked by Data Centers on the Ballot, across the 47 local midterm elections in the American cities and townships that featured ballot measures on data center construction, voters favored restrictions on the rate of data center expansion in 38 of the 39 completed counts. Of the eight races that remain open, voters supporting additional data center restrictions are leading in all six of the races with only partial reporting. For the two remaining races, no results have yet been posted.

Here's the rundown of the types of measures voters are supporting:

• In ten of the closed races, voters approved adding megawatt caps for data center projects to their city or town charters. Typically, those towns selected 25 MW caps; for comparison, a planned OpenAI facility in Effingham County, Georgia is expected to demand 3,200 MW. Voters in favor of megawatt caps are leading in two partial counts.
• Ten of the elections included non-binding advisory questions posed by village councils and town electors, asking whether voters thought their towns should permit or restrict data center development. In nine closed races, voters voted in favor of restriction; pro-restriction voters are leading in the remaining partial count.
• In seven closed races, voters rejected rezoning measures that would permit the construction of data center sites, and voters opposing a similar rezoning decision are leading in another partial count.
• Voters in five races approved outright bans on new data center constructions, and a similar ban is leading in one race with partial reporting. In two other races, measures that would require voter approval before new data center construction can progress have either passed or are leading.
• In six other completed races, voters opted for other data center restrictions, like temporary moratoriums and additional conditions on data center tax benefits.

At time of writing, only one ballot measure that would have introduced data center restrictions failed: In Harwood, North Dakota, voters refused an amendment to the city charter that would have required Harwood to commission an independent review before approving new data center construction or expansion.

Aside from that one exception, it's not hard to suss out why so many of those measures have passed. NBC exit polls show that 56% of yesterday's midterm voters considered AI data centers either "crucial" or "a major factor" in their choice of Congress person, with data center opponents driven by concerns of environmental toll, energy costs, noise pollution, and—one must imagine—the upward-spiraling cost of RAM.