Voters rejected data centers in 40 of 41 local ballot measures

For three years the story of the AI boom has been told in chips, gigawatts and capital spending. On November 3, 2026, a different group got a say: local voters. In the US midterm elections, 41 local ballot measures about data centers were on the ballot in 10 states. When the results came in, voters had gone against data centers in every one of them except a single small town in North Dakota.

The result was not a quiet local footnote. Data centers showed up at the top of the ticket too, in Senate and governor races in Ohio, Texas and Georgia. In the NBC News exit poll, 56% of voters said AI data centers were a crucial or a major factor in how they voted for Congress. This article first lays out what actually happened, based on Ballotpedia, CNBC, WIRED and the exit poll numbers reported by DCD. After that, clearly separated, comes djcroman's opinion on what it means for AI companies and for everyone who uses their products.

What was on the ballot

Ballotpedia, the nonpartisan election encyclopedia, tracked every local measure related to data centers in 2026. Its count for election day: 41 measures in 10 states. The way they were worded matters, because it rules out the easy explanation that voters simply said no to everything.

For 28 of the measures, a yes vote would limit data center development. Seventeen of those were bans. Twelve of the bans only applied above a size threshold, such as a Kansas measure that would ban data centers using more than 25 megawatts of electricity, and five would ban data centers outright. Three measures were six month moratoriums. One in Wells, Maine, would even cancel recent permits for new or expanded data centers. Others required voter approval for large projects, added public hearings, or forced developers who get tax breaks to give financial guarantees to the community.

For the other 13 measures, a yes vote would make building easier. Four of them in Wilmington, Ohio, would rezone land from residential to industrial use, which would allow a data center on those sites. Seven were non binding advisory questions in Illinois, New Jersey and Ohio asking voters to express support for data center projects.

So voters faced a real mix: sometimes yes meant stop, sometimes yes meant build. They picked the side that limits data centers almost every time.

The results: 40 of 41

According to Ballotpedia's results page and additional results gathered by CNBC, voters sided against data centers in every measure except one. The exception was Harwood, North Dakota, where voters rejected an extra layer of independent review and a public hearing for big industrial projects, data centers included.

A results check by the real estate site RealtyWire put the split in plain numbers: all 13 measures that would have eased development lost, most of them by wide margins. Of the 28 measures that would limit data centers, 24 were marked approved on the day after the election, three more showed majorities in favor but were not yet called, and only the Harwood measure failed.

The margins were not close. In Alhambra, California, 82% voted to prohibit data centers in the city. In Edgewater, Florida, 77% voted to ban new ones. In Joliet, Illinois, voters were asked whether the city should approve the construction of AI data centers, and 79% said no. In Bloomington, Illinois, 78% backed an advisory moratorium on large data centers. Morgan County in Illinois approved restrictions with 88%.

The election day votes also fit a pattern from earlier in the year. Ballotpedia counted six data center measures before November, and voters went against the industry in all six. Canterbury, New Hampshire, banned data centers with 83.6%. Monterey Park, California, approved a ban with 88.3%. Augusta Township, Michigan, rejected a data center rezoning with 89.6% against. Counting those, the year stands at 46 decisions against data centers out of 47 decided measures, with two more votes still to come in December in Clarksville, Arkansas, and Mandan, North Dakota.

Why Ohio matters most

More than half of the November measures, 21 of 41, were in Ohio. Illinois followed with 7, Maine with 4 and Kansas with 3. California, Florida, Nevada, New Jersey, North Dakota and Wisconsin had one each.

Ohio is not a random state for this fight. As CNBC points out, OpenAI announced in August that it had secured 8 gigawatts of computing capacity at the PORTS Pike Technology Campus in Ohio, planned as the largest AI data center campus in the world. SoftBank and SB Energy are developing it together with the US Department of Energy, Nvidia is backing the project, and the total cost could pass $500 billion. Capacity is supposed to come online in phases starting in 2028.

On the same night, Ohio's politics shifted around that project. Democrat Sherrod Brown won back a Senate seat after spending months calling his opponent "the face of data centers in Ohio". In the governor's race, Democrat Amy Acton beat Republican Vivek Ramaswamy. Acton supports what she calls a conditional moratorium on data centers: new projects would have to cover 100% of the utility costs they cause and be built exclusively with union labor. That does not cancel the Ohio campus, but it gives a project of that size a much more demanding partner in the state capital.

The top of the ticket

The local measures were only one layer. According to CNBC and WIRED, candidates who campaigned on pausing or regulating new data centers won key races in Ohio, Wisconsin, Georgia and Texas. In Texas, which is on track to become the largest data center market in the world by 2030, Gina Hinojosa unseated Governor Greg Abbott after blaming him for the rapid buildout across the state. In Georgia, Keisha Lance Bottoms won the governor's race, and in Kansas Cindy Holscher was close to a win. WIRED reports that all three campaigned not only for more regulation but for data center moratoriums.

Money did not save the other side. WIRED reports that a new industry super PAC called Building the Future spent $1.5 million to support Darline Graham in South Carolina, and she lost to Annie Andrews, who favors a moratorium. The same PAC put more than $1.5 million into the Kansas governor's race. CNBC's analysis of federal election filings found that crypto, AI and online gambling interests poured more than $377 million into the midterms, and that Elon Musk alone spent at least $474 million, nearly half of it in Texas, where his side lost both marquee races. His reaction on X: "Data centers in space looking pretty appealing right now."

What voters actually said

The exit poll explains why candidates leaned into the issue. NBC News asked voters how important AI data centers were to their vote for Congress. As reported by DCD:

• 23% said crucial
• 33% said a major factor
• 31% said a minor factor
• 13% said not a factor

That adds up to 56% for crucial or major, the same share CNBC reported from its own voter poll. The issue leaned strongly Democratic, but it was not purely partisan: among voters who called it crucial, 30% were Republicans. The broader numbers are even more striking. 76% of respondents said they would oppose a new data center in their own community, and 57% said AI will do more to hurt society than to help it. DCD notes that data centers still ranked behind government corruption, cost of living, immigration enforcement and the war in Iran as voting issues, so this was not a single issue election. But it was clearly one of the issues.

Why do people who have never seen a data center vote against one? The Democratic strategist Alexander McCoy told WIRED that voters link data centers and AI so closely that fixing noise, water or pollution would not end the opposition, because people also worry about what the projects enable: effects on jobs, prices and society. Sacha Haworth of the Tech Oversight Project framed it as a story about corruption, affordability and the influence of tech billionaires. Either way, the data center has become the physical face of AI, the one part of it a voter can actually see, drive past and vote on.

What happens next in Washington

The reaction came fast. Rep. Greg Casar of Texas, who chairs the Congressional Progressive Caucus, said AI company CEOs should be called to testify under oath, and he called the data center revolt "just the early tremor of a much bigger AI earthquake upending American politics." Rep. Robert Garcia of California, the presumptive next chair of the House Oversight Committee, put it in two words: "Lawyer up."

There is also a state level precedent already in place. In July, New York Governor Kathy Hochul issued an executive order for the first state moratorium on data center construction, a one year pause on permits for new data centers expected to use 50 megawatts or more. At the same time, Ballotpedia notes that 38 states had tax incentives for data centers as of April 2026. Expect that second number to come under pressure in states where the new governors ran against those incentives.

The industry is already adjusting. According to WIRED, environmental groups and big tech companies are working together in a "Responsible Data Center Development Collaborative" to write a framework for how data centers should be built.

djcroman's take: permission is the new GPU shortage

This part is opinion.

For most of the AI boom, the scarce resource was hardware. Whoever could buy the most Nvidia GPUs won. That problem is getting solved with money. The 2026 midterms show the next scarce resource, and money does not solve it as easily: local permission. A town can be outspent, but it cannot be outbid when its residents vote 79% against a project.

Our bet is that three things follow. First, the industry will start paying for its own power and grid upgrades as a matter of course, because "data centers raise your electricity bill" is the attack line that worked. Acton's condition in Ohio, 100% of utility costs, is the template. Second, new capacity will cluster in places that already said yes, which gives those few locations and their utilities a lot of pricing power. Third, the hyperscalers, which CNBC notes are already cash flow negative except for Microsoft because of AI spending, will face a slower and more expensive buildout than their capex plans assume.

For readers, that matters even if you never vote on a data center. Slower and pricier compute eventually shows up in the price of AI subscriptions, in tighter usage limits and in which regions get new features first. And if you build on AI, this is a reason to care about efficiency: smaller models, smarter caching and less wasted inference are not only cheaper, they are also easier to get permission for.

The AI industry likes to say its biggest risk is a competitor or a regulator. After November 3, the more honest answer is a town hall full of voters who were asked a simple question and gave a clear answer.

Sources

• Ballotpedia News: Voters in ten states will vote on data center related ballot measures November 3 (Nov 3, 2026)
• Ballotpedia: Data center related ballot measures, 2026 (results table)
• CNBC: Voters rebuff data centers in every ballot measure except one during the 2026 midterms (Nov 4, 2026)
• WIRED: The Results Are In: Voters Hate Data Centers (Nov 4, 2026)
• DCD: NBC exit poll, 56% say AI data centers were a crucial or major factor in Congressional vote decision (Nov 4, 2026)
• RealtyWire: Data center ballot measures, voters back limits in 37 of 41