US voters backed restrictions on new AI data centers in 40 of 41 local ballot measures
During the U.S. midterm elections on November 3, 2026, there were 41 local ballot measures related to data centers across 10 states. According to Ballotpedia’s statistics and CNBC’s compiled results, voters supported restrictions on development or rejected measures favorable to development in 40 of them, with only one exception in Harwood, North Dakota.
Harwood voters rejected by about 75% (325 votes to 105 votes) an initiative that would have required additional review of industrial projects, including data centers. Ohio alone accounted for 21 measures, the most in the country, involving blanket bans, restrictions on facilities exceeding 25 megawatts, moratoriums, and zoning-related votes. Many measures passed by margins of more than 70%.
These votes mainly targeted the construction, approval and operation of new data centers or large-scale facilities, reflecting community concerns about power consumption, water use, noise and land impacts. The relevant results were based on unofficial vote counts, and some measures were advisory in nature or required further local implementation.
Original source:
Title: Voters rebuff data centers in every ballot measure except one during the 2026 midterms Publisher: CNBC Publication date: November 5, 2026 (updated) URL: https://www.cnbc.com/2026/11/04/data-center-ballot-measures-2026-midterm.html Nature of source: news report (citing Ballotpedia and others) Verification status: full text read, core figures consistent with the Ballotpedia page Supporting content: 41 measures, 10 states, 40 to 1, Harwood exception, 21 measures in Ohio, etc.
ABAB AI Insight
Data center expansion depends on local land, power and water resources, creating a direct conflict with federal-level AI promotion and local ballots. Historically, similar large-scale infrastructure, such as nuclear power plants and highways, has also often met resistance at the local level, but AI projects have seen concentrated backlash because of their high energy intensity and disputes over tax incentives.
The Harwood exception shows that small towns where projects are already under construction and are expected to bring jobs and tax revenue may prioritize economic benefits over additional procedures. This contrasts with the nationwide majority of communities that prioritize restricting new development, indicating that the distribution of benefits and the stage of the project affect voting outcomes.
The concentration in Ohio is the highest, related to the state’s earlier large sales tax abatements for Amazon, Meta and others, which led communities to perceive spillover costs. The concentration of capital by major tech companies into a small number of hyperscale projects makes local externalities easier for voters to notice and turn into voting action.
Points to watch include whether state-level legislation follows, whether existing projects are impeded, whether developers shift to other states or adjust scale, and whether power capacity and grid investment become new constraints. Local votes do not automatically end federal or private capital plans, but they increase approval and political risks.
ABAB News · Cognitive Law
Local control over land and resources has become the main political constraint on national computing power expansion.
The difference in voting outcomes between existing projects and new projects reflects the distribution effects of sunk costs and expected benefits.
Energy-consumption externalities create a structural confrontation between community interests and industrial capital paths.