Voters in Trenton, Ohio, backed a ban on data centres above 25 megawatts by 79.7%, a result that put Amazon Web Services-linked projects at the centre of a local revolt against the physical demands of the AI boom. That vote was one of dozens of municipal and county contests that gave the 3 November US midterms an unusually concrete afterlife: while Democrats won control of the House and Senate, according to NBC News and the BBC on 4 November, voters in small jurisdictions were deciding who gets to host the warehouses of the digital economy — and who pays when public services need more money.

The results remained partial and unofficial on 4 November. But the early pattern was clear enough to make ballot measures a defining part of the election story. Local restrictions on data centres were passing by large margins, while statewide efforts to impose or protect higher taxes on the very rich produced a more divided map.

Local votes put the AI buildout on the defensive There was no statewide referendum on data centres. Ballotpedia counted 41 local measures in 10 states on the 3 November ballot, including 21 in Ohio. Twenty-eight asked voters to limit development; 13 would have facilitated it. Stateline reported on 4 November that 33 restriction measures had passed or were leading late on election night, with only one result clearly favouring the industry. The Washington Post separately counted at least 29 jurisdictions approving bans or restrictions. Ohio supplied the sharpest verdicts. Voters in Sunbury, Pataskala, Granville and Urbana approved bans or charter restrictions by margins ranging from 73.6% to 82.4%, according to the Data Centers on the Ballot tracker. In Wilmington, four rezonings tied to an Amazon-linked project were rejected by roughly 80% of voters. Ashville voters repealed an EdgeConneX agreement with 81.6% support. Grove City’s restriction was leading with 54.6% as counting continued. The measures varied in design, but their common target was the rapid arrival of facilities that promise tax revenue and construction work while raising fears over electricity demand, water use, land use and household bills. An NBC exit poll cited by Stateline found that 56% of respondents said AI and data centres were a crucial or major factor in their House vote. A September Quinnipiac poll reported by Bloomberg found 72% of Americans would oppose a data centre in their own community. Outside Ohio, the margins were similarly stark. Emporia, Kansas, approved a 10-year ban on “high-impact” data centres by 4,346 votes to 2,891, or 60%, KVOE reported on 3 November; litigation over the measure remains pending. Bloomington Township, Illinois, passed restrictions with 77.6%, while an advisory question supportive of data-centre development in Joliet Township was rejected by 78.9%. Janesville, Wisconsin, required voter approval for projects costing more than $450m, passing with 62.8%. Wells, Maine, approved a six-month moratorium with 81.9%. Alhambra, California, was approaching an 82% vote for a citywide ban. Boulder City, Nevada, was rejecting a plan to allow development on desert-tortoise preserve land by about 70.6%. The exceptions mattered. Solano County, California, rejected a proposed $5-per-square-foot annual tax on data centres by 61.5%. Harwood, North Dakota, rejected additional review and a public hearing requirement by 75.6% — the clearest industry-side result in the tracker’s early returns.

“Ballot measures to restrict or prohibit data center development would send a signal that the area is closed for business,” Dan Diorio of the Data Center Coalition told Bloomberg before the vote. That warning reflects a genuine tension. Local bans may deter investment and jobs, and UCLA political scientist Megan Mullin told Bloomberg that they can tie the hands of future local leaders.

Yet the size of Tuesday’s margins suggested that communities want binding leverage before a developer arrives, rather than assurances after the deal is signed.

Tax votes split over how far states should go

The national argument over inequality played out differently. California’s Proposition 40 was the only wealth tax proper on the ballot: a one-time 5% levy on net worth above $1bn, affecting roughly 200 billionaires, with 90% of proceeds designated for health care after federal Medicaid cuts. As of 7.23am Eastern time on 4 November, AP-sourced results showed the measure trailing, with 55.6% voting no and 44.4% yes, and about 59% of ballots counted.

The Associated Press had not called the race. California’s Legislative Analyst’s Office estimated that the measure could raise tens of billions of dollars over several years, but also said billionaire departures could reduce annual income-tax receipts by less than $1bn. The campaign was a costly test of whether taxing wealth can survive an organised counterattack.

CalMatters reported that opponents spent $316.7m, including $102m from Google co-founder Sergey Brin, who had moved out of California.