Voters deliver a sweeping rebuke to data center expansion

Voters backed restrictions or rejected pro-development measures in all but one of 41 local ballot measures across at least 10 states, according to a Ballotpedia analysis and results compiled by CNBC.

The Washington Post separately reported that at least 38 jurisdictions passed measures limiting or blocking data centers, with many winning by wide margins.

The votes landed as companies rushed to build data centers (computing facilities that host AI workloads) for artificial intelligence. Host communities are weighing promised investment against electricity bills, water consumption, noise, land use and local control, while the industry is committing billions of dollars to expansion.

Neither tally shows that voters adopted a national rule or spoke with one voice. The counts came from separate analyses, and local ballot language varied.

What did voters decide in Ohio?

Ohio, described as a major hub for data center development, produced some of the most decisive results in the country. The reported votes were:

• In the City of Defiance, voters approved a charter amendment (a change to a city’s governing document) restricting data center construction by a 64.73% to 35.27% margin, according to unofficial county election results.
• Pataskala voters approved a data center-related charter amendment by 73.62% to 26.38%.
• Granville voters backed a similar amendment by 82.45% to 17.55%.
• In Ashville, voters rejected a data center referendum, with 81.61% against the measure and 18.39% in favor.

The margins support The Washington Post’s assessment that opposition was intense where projects appeared directly on the ballot. The measures did not impose one national rule, and the source does not provide legal text or implementation schedules. Local votes now create measurable political risk for developers.

How much development has the backlash already disrupted?

More than 145 data center projects were delayed or canceled nationwide through early August, according to S&P Global figures cited by The Washington Post. The Post reported that public opposition accounted for the vast majority of those setbacks.

Those figures establish material disruption, but the report does not provide a nationwide project total against which to measure the 145-plus figure. It also does not separate temporary delays from permanent cancellations or say whether any delayed project later restarted, leaving the amount of capital at risk unspecified.

The figures cover delays and cancellations, but the source does not quantify either outcome. The election result therefore reaches beyond symbolic votes because The Post linked public opposition to most setbacks. Public opposition is already disrupting development, not merely influencing debate.

Did opposition cross party lines?

A Washington Post-ABC News-Ipsos poll conducted before the election found that seven in 10 Americans opposed new data center construction in their communities. That included a majority of Republicans, suggesting that opposition was not confined to one party.

An election-day poll cited by The Post found that 56% of voters considered artificial intelligence and data centers a major or crucial factor in their vote. The issue still ranked last among five issues surveyed, behind government corruption, the war in Iran and the cost of living. The backlash also entered statewide campaigns, where data center development sat alongside affordability and economic growth:

• In Texas, Democrat Gina Hinojosa defeated Republican Gov. Greg Abbott after campaigning for a proposed moratorium (a temporary pause on new projects) and tougher environmental protections and protections for electricity ratepayers (customers).
• Ohio elected Democrat Amy Acton as governor. She supports a moratorium and stricter regulations.
• Georgia Gov.-elect Keisha Lance Bottoms made greater accountability for data center developers a central campaign issue.

The polls and campaigns show broad concern, but they do not prove data centers alone drove those statewide outcomes. The source gives no sample sizes, polling margins or evidence separating data center votes from other issues. Data center policy crossed party lines, although it ranked last among the five issues in the election-day poll.

What is the industry’s counterargument?

Industry representatives argue that investment can deliver substantial benefits to host communities. “Affordability, jobs, and the economy are priorities for Americans across the country,” said Dan Diorio, executive vice president of state policy and government affairs at the Data Center Coalition. He said data centers can generate jobs, local tax revenue and investment.

“The data center industry is committed to advancing America’s digital infrastructure, supporting energy reliability and affordability, and promoting long-term, responsible economic development,” Diorio said. The report offers no project-level accounting of those benefits, utility costs or offsets to local concerns, leaving the industry’s economic case untested here.

How is OpenAI responding to the vote?

OpenAI said Thursday that it would pay the full cost of powering data centers it owns and operates, solicit community input, publish construction plans and avoid using nondisclosure agreements (secrecy terms) with public officials. It also released a policy blueprint encouraging lawmakers to adopt similar standards.

OpenAI Chief Global Affairs Officer Chris Lehane told reporters, according to CNBC: “What we’re doing with this blueprint is really laying out the best lessons that we have learned and committing to applying them as they go forward.” The announcement followed Tuesday’s election results, although the blueprint had been in development since August. It built on earlier OpenAI pledges to protect ratepayers and address community concerns.

OpenAI is seeking rapid expansion to meet demand for AI services and has targeted hundreds of billions of dollars in computing infrastructure spending by 2030. That figure is a target, not a disclosed schedule of contracts or projects, and the source does not say how much spending would fall under its new cost pledge.

The power-cost commitment applies specifically to facilities OpenAI owns and operates; the source does not say it covers any other data centers used by the company. The report also gives no enforcement timetable, penalty or dispute process for the promises.

OpenAI has offered a template for local consent, but the scale of planned spending leaves that template untested. The announcement shows what the company promises; it does not show whether public officials will accept them.

Where did voters favor development?

Not every community rejected data center growth. In Harwood, North Dakota, roughly three-quarters of voters rejected proposed restrictions on industrial development, according to The Washington Post.

That vote supplies an important counterexample to a blanket reading of the election. The measure concerned industrial development rather than a specifically identified data center project, so it should not be treated as a direct referendum on data centers.

The contrast leaves two policy tracks: communities can impose limits, while others can protect development from restrictions. OpenAI’s blueprint supplies one voluntary framework for the first track, but it does not resolve the ballot measures already approved. The next test is whether disclosure, ratepayer protection and local moratoria can coexist with hundreds of billions of dollars in planned AI infrastructure spending.