A GridSight research note · 6 November 2026

Data centers get built where they're welcome.Tuesday showed what welcome means.

Forty of 41 local data center votes on November 3 went toward limits. Read closely, the results say less about whether data centers get built than about where.

The data centers this country needs will get built, in communities that want them. The November 3 votes made clear that a council's yes and a town's yes are different things, and only one of them held.

The distinction matters because grid capacity, not demand, is what limits AI growth. Power takes years to plan and connect. An approval that is reversed at the ballot box costs a project time that the buildout cannot easily make up.

What happened

40 of 41November 3 measures went toward limits on development
46 of 47local data center votes in 2026, counting six earlier in the year
77%median share for the winning side
7local officials recalled over data center decisions on November 3

Voters in 10 states decided 41 local data center measures on November 3. In 40 of them, the outcome limited development, and all six local votes held earlier in 2026 went the same way. Only two results came in under 60 percent: Sprigg Township, Ohio, which adopted zoning 53 to 47, and Grove City, Ohio, which adopted a public vote requirement for large developments 55 to 45.

The single exception was Harwood, North Dakota, where voters rejected an added review requirement for industrial projects, 75 to 25. Harwood approved Applied Digital's $3 billion campus in 2025, and construction is well underway. Residents there have lived alongside a project, and they chose not to add a new hurdle.

Every result is in our ballot tracker, measure by measure.

Official welcome proved fragile

Several results reversed decisions local officials had already made. Ashville, Ohio voters repealed the village's agreement with EdgeConneX, 82 to 18. Wilmington, Ohio voters overturned four rezonings for data center sites, each by about 80 percent. Piqua and Urbana, Ohio passed bans and recalled officials on the same night. Seven officials were recalled over data center decisions on November 3 alone, including in Mason, Michigan, Yukon, Oklahoma and San Angelo, Texas.

Most of these measures began with residents, not councils. By Ballotpedia's September count, 31 of 38 qualified through citizen petitions. An approval can now be put to a public vote, and on Tuesday the public vote usually won.

Resident welcome has to be earned

The starting point is low. A UMass Amherst poll this year found that only 11 percent of Americans would support an AI data center in their own community. Welcome rarely exists by default. It is built through what a project discloses, what it commits to in writing, and whether residents believe the benefits will reach them.

The votes also show where the buildout goes next. A city ban binds only that city. Ohio has at least 138 local moratoriums and 23 communities that have effectively banned data centers, against 257 operating facilities, by the Ohio Capital Journal's count. Our read is that the votes will not stop the buildout. They will steer it toward places where the welcome is real.

What this means for you

Developers

Site selection now includes a question no land survey answers: will residents stand behind an approval? The time to earn that is before a vote gets scheduled, not after a petition is filed.

Utilities

Large load forecasts carry a local approval risk that most models do not price. A signed agreement in a community that has not accepted the project is a softer number than it looks.

Counties and local officials

A council decision can be challenged at the ballot box, and Tuesday showed how often that challenge wins. Knowing where residents stand before you approve protects both the project and the officials who vote for it.

Investors

Community risk is now a diligence input with a track record. An approval that a referendum can reverse should be priced like one.

What to watch

Many communities pause before they vote. Our moratorium outcomes tracker shows that most pauses end with rules that let projects move forward.

Sources: Ballotpedia, Data center related ballot measures, 2026; unofficial county and municipal results as compiled by Duane Morris Government Strategies, 5 November 2026, and the GridSight ballot tracker; CNBC; Stateline; UMass Amherst; Ohio Capital Journal; North Dakota Monitor. Results are unofficial until counties certify them.

Know whether the welcome is real before it is tested.

The GridSight scores community sentiment from the public record and local search, so you can see where a project stands before a petition is filed.

Score your marketSee every result