A Guth Labs publication ↗
Back to News

Local voters favored data-center limits in 40 of 41 November contests

AI-generated conceptual editorial illustration: A ballot box in a small town facing stylized data-center buildings, representing local land-use decisions.
guthlabs.ai · AI-generated conceptual editorial illustration

Voters favored tighter limits on data-center development in 40 of the 41 local ballot contests tracked by Ballotpedia on November 3. The result includes approvals of restrictions and rejections of proposals that would have supported development. It is a count of measures, not a nationwide percentage of voters opposing data centers. Ballotpedia results analysis

The distinction matters because the measures asked different questions. Some concerned bans, others temporary pauses, zoning, financial guarantees or advice to local officials. Treating every outcome as the cancellation of a proposed AI facility would overstate what the tally establishes.

Forty of 41 is the election-day figure

Ballotpedia's separate total of 46 restrictive outcomes among 47 decided measures includes six votes earlier in 2026. It should not be described as the November 3 total. The exception on election day was Harwood, North Dakota, where voters rejected additional review requirements for industrial projects including data centers. Measure tracker

The November outcomes included both binding measures and nonbinding expressions of opinion. That means their immediate legal effects differ even when they point in a similar political direction. A developer needs to read the actual measure, not infer its obligations from the aggregate result.

Alhambra illustrates a binding land-use decision

Los Angeles County's published results list 19,022 votes in favor of Alhambra's Measure DB and 3,647 against, roughly 84% support, in the tally reviewed November 11. The measure asks voters to affirm the city council's existing ban by defining data centers in the municipal code and prohibiting them across all city zones. County results

This is a decision about local land use, rather than a direct referendum on whether residents should use AI. The ballot language does not distinguish a specific model developer or establish how much computing demand belongs to AI rather than other services. The totals should also be understood as a reported count during the post-election reporting period, not presented here as a certification announcement.

The example shows why project-level analysis remains necessary. A citywide prohibition, a six-month pause and a requirement for an additional vote create different constraints. A national summary can reveal a pattern without replacing the local legal documents.

The siting process is part of the infrastructure plan

For companies planning compute campuses, community approval needs to be treated as a project dependency alongside financing, equipment and power. A credible proposal should explain the relevant land use, utility arrangements and responsibilities clearly enough for residents to evaluate them.

The results alone do not tell us which concern motivated each voter, how views differ in places without a ballot measure or whether a revised project would win support. Those questions require additional evidence rather than assumptions about public sentiment.

What the tally does establish is a strong pattern among the local questions that reached voters. The next stage is implementation: determining which rules change immediately, which votes merely advise officials and what each outcome means for the projects actually affected.

Sources and citations

Revision history

Published .