In February, the city council in Wilmington, Ohio, rezoned about 546 acres of rural land for data center development. Residents gathered enough signatures to put all four rezonings on Tuesday's ballot. The developer's deal reportedly fell apart over the summer, and voters still turned out to overturn every one of them, by about 80%.
Wilmington's four votes were among 47 local data center measures on Tuesday's ballot, and I watched all of them come in. This fall I built DataCentersOnTheBallot.com, a tracker for every local data center measure on the ballot, with live results on election night. It got a lot more attention than I expected, with mentions in Rolling Stone, on NPR and on Bloomberg News!
Since Tuesday, the question I've heard most is some version of: what did this election mean for data centers, and for AI? Here's my short answer. Data centers lost badly, and they're going to be harder to build. But this wasn't an AI election. It was an affordability election, and data centers were the part of it voters could see.
One Out of 47
On Tuesday, voters in 10 states decided 47 local measures about data centers. Ohio alone had 21. In 46 of them, the side that wanted to restrict data centers won.
Most weren't close. The median restrict side won with about 77%, and across all 47 measures and roughly 260,000 votes, the restrict side took 73%. When a YES vote meant allowing a data center, as in Wilmington, Ohio's four rezonings and the advisory votes in five Illinois townships, the data center lost all 13 times. Wilmington wasn't the only town to overrule its own council, either. In Ashville, Ohio, voters threw out the village's deal with EdgeConneX for two data centers and a gas plant, also by about 80%.
Bigger places voted the same way. Morgan County, Illinois, voted countywide, 88% for restricting data centers on more than 10,000 ballots. Alhambra, California, a city of 80,000 outside Los Angeles, voted 83% to lock in its ban, with ballots still being counted.
The one exception was Harwood, North Dakota, where about 75% of 430 voters rejected a rule requiring an independent review and a public hearing before approving large industrial projects. Its wording turns out to matter.
The Closer the Question, the Bigger the No
Saying "data center" on the ballot mattered. In the 38 measures whose ballot language used the words, the restrict side averaged 76%. In the nine that didn't, it averaged 64%, and that group includes Harwood.
So did how specific the measure was. The restrict side's average by type of measure:
- "Do you want data centers here?" (10 advisory votes): 81%
- Votes on a specific project, like Wilmington and Ashville (6): 79%
- Blanket bans on any size (4): 78%
- Moratoriums and pauses (7, all in Maine): 74%
- Bans above a size or power threshold, mostly 25 megawatts (14): 71%
- Process rules that never mention data centers, like a citywide vote on big developments or adopting township zoning (5): 51%, or 57% without Harwood
The 47th, in Perry, Ohio, set conditions instead of a ban (a hyperscale data center seeking a tax break would have to post a bond first) and passed with 87%.
Voters say no to a data center they can picture. Ask them to approve a procedure and the vote gets close: Grove City, Ohio's public-vote requirement passed with 55%, and Sprigg Township, Ohio, where adopting zoning was the township's only way to get a say over a data center planned at an old power plant, passed it with 53% on 627 ballots, the narrowest margin of the night.
The polls show the same thing. Stockton University asked New Jersey voters about data centers three ways. Nationally, they split 40% to 40%. In New Jersey, 52% opposed them. In their own town, 65% opposed.
Did It Move Votes?
Everyone agrees voters don't want a data center next door. The NBC News exit poll found 76% opposed to one in their community, and the AP Voter Poll found about three-quarters, with majorities opposed in both red and blue states. CBS News's exit poll also landed at 76%, and it shows how far across party lines that goes: 87% of Democrats, 78% of independents and 62% of Republicans opposed a data center in their community, with majorities against in urban, suburban and rural areas alike. The argument is over whether that opposition moved votes, and the answer depends a lot on how you ask.
- NBC: 56% called AI and data centers a crucial or major factor in their vote for Congress (23% crucial, 33% major).
- AP Voter Poll: about 1 in 4 called AI data centers "crucial," well below the shares saying that about government corruption or the cost of living.
- TrueDot.ai (national, 1,952 registered voters, October 29, paid for by OpenAI): just 3% named data centers the most important issue in their vote, 14th of 15 issues.
TrueDot ran a useful experiment on this, splitting its sample in half at random. The first half was asked straight out how much candidates' positions on data centers influenced their vote, and 57% said "a lot" or "some." The second half was first asked a yes-or-no question: did candidates' positions on data centers influence your vote at all? Only the people who said yes were then asked how much. Counting everyone in that second half, just 33% said data centers influenced their vote. Same survey, same week, a 24-point gap from wording alone. When you ask people how much something mattered, many will give it some credit. When you first let them say it didn't matter, a lot of them do.
My read is that data centers were a real issue, just not the main one. A majority of voters calling something a factor is a lot. But in NBC's exit poll, government corruption (90%) and the cost of living (89%) topped the list, followed by the war in Iran (74%) and immigration enforcement (73%), with AI and data centers well behind. When TrueDot asked what Washington should prioritize in 2027, 10% picked data centers, against 38% for the economy and 27% for grocery costs.
It Was About the Bill
None of this should surprise anyone who's been watching the issue polling. In YouGov's tracking, inflation has been the top issue since 2022, and on election eve it led jobs and the economy by 19 points and immigration by 23.
So why the landslides on the local ballot? Because a data center is the one piece of the affordability squeeze that people can see and vote against. You can't vote down grocery prices. You can vote down a 200-acre campus next to the subdivision, especially if you think it will raise your power bill.
Ask voters what worries them about data centers and the bill comes first. In TrueDot's survey, higher electricity bills led at 23%, ahead of AI harms (15%), grid strain (15%) and water use (13%). In University of Houston polling, 76% of Texas voters expected a new local data center to raise their electric bill. And the bills are already a strain: roughly 80% of people who've lived in the same home for three years say their electric bill went up, and 1 in 5 Americans say they've cut back on groceries to pay it, according to Consumer Reports.
One more TrueDot number stood out to me: 28% said "politicians aren't listening" affected their vote, more than the 23% who cited the facilities themselves. A lot of this anger is about who gets a say.
Data centers were the part of the affordability election that voters could see, point to, and vote against.
The Candidates: Running Against the Server Farm
Data centers also showed up in the races at the top of the ticket, and the candidates who ran against them did well.
- Ohio: Sherrod Brown beat Jon Husted for Senate by about five points after running ads calling Husted "the face of data centers in Ohio." An NRSC memo in August reportedly called data centers "the anchor hanging around Husted's neck." Democrat Amy Acton won the governor's race.
- Texas: Gina Hinojosa beat Gov. Greg Abbott, promising to halt new data center approvals until the legislature writes protections. James Talarico beat Ken Paxton for Senate after calling for an end to "sweetheart tax deals" for Big Tech.
- Wisconsin: Republican Tom Tiffany called his opponent "Data Center David Crowley." Crowley, who backed a moratorium and ending the state's sales tax break for data centers, won the governorship.
- Georgia: Democrats won both seats on the Public Service Commission, the board that sets Georgia Power's rates, giving them a 3-2 majority for the first time in about two decades. Both winners ran on making data centers pay for their own power.
Campaigns spent more than $380 million on TV ads that mentioned data centers, and Democrats outspent Republicans more than 3-to-1, according to AdImpact. Data Center Watch counts 32 of 34 winning governor candidates as supporting stronger regulation, and 12 as supporting bans or moratoriums.
Does that mean data centers won Democrats the election? I'd be careful. In NBC's exit poll, voters who called AI and data centers a crucial or major factor backed Democrats 65% to 33%. But TrueDot found that among partisans who said data center positions pulled them toward a party, 88% named their own party. People tend to find reasons for the vote they were already going to cast. And the issue wasn't purely partisan going in. Most Republicans in CBS's exit poll opposed a local data center too, and Abbott had paused data center approvals himself, and he still lost in a wave year driven by prices, the war and corruption.
What changed is the politics. Opposing data centers started as a cause for town councils and zoning boards. After Tuesday, it's a winning message for statewide candidates, and the people who ran on it now control the governor's office in Texas, Ohio and Wisconsin and the utility regulator in Georgia.
AI Rode Along. It Didn't Drive.
What about AI itself? Voters are skeptical. In NBC's exit poll, 57% said AI will do more to hurt society than help, and 24% said help. YouGov finds the country more evenly split, 40% expecting negative effects to 35% positive. In a USC Annenberg survey, 76% said the government is doing too little to regulate AI.
Meta's president, Dina Powell McCormick, said this week that "the visible sort of manifestation of AI is a data center," and that building new ones in the U.S. is "going to be harder." I'd put it differently. The polls say data centers are the visible face of the electric bill. AI harms came second to bills in TrueDot's list of concerns, and none of the 47 local measures asked voters about AI. They asked about zoning, power and water.
AI did show up in this election in one other way: as a research tool. In polling by the National Conference on Citizenship and The Carter Center, the share of Americans who used an AI chatbot to learn about candidates rose from 14% to 25% over just two weeks in October, even though only 22% trust AI a great deal or quite a bit to give them unbiased information. Voters are wary of AI and using it anyway.
The AI number I'd keep an eye on is about the parties. Asked which party would better handle AI, 30% said Democrats, 27% said Republicans and 43% said neither. Nobody owns this issue yet. That's an opening for whichever party, and whichever company, offers a credible answer on costs and accountability first.
I'm an AI optimist, and I don't read Tuesday as a rejection of the technology. What voters object to is a deal where the benefits go somewhere else and the costs land on their utility bill. That can be fixed.
The Bill Comes Due
- Texas. By some industry estimates, Texas accounts for about 30% of the data center capacity being planned or built in the U.S., and 79% of likely Texas voters told the University of Houston they wanted limits. Hinojosa was careful with her framing on CNN: "It is not that I hate data centers. I hate the way they are taking advantage of Texas, and that stops when I'm governor." Watch whether the pause she promised is a pause or a ban.
- Georgia's new majority. The Democratic PSC will decide how much data centers pay for the power and grid upgrades they need. Watch its first rulings on large-load tariffs and Georgia Power's next rate request.
- The next round of local votes. A nearly identical 25-megawatt charter amendment passed in all nine Ohio communities that voted on it Tuesday. Expect copycats in 2027, and an attempt at a statewide measure in Ohio.
- How the industry responds. OpenAI published new commitments for the communities it builds in, and CFO Sarah Friar told CNBC that if the election means more focus on data centers and communities, OpenAI is "going to keep doing the work" to get into them and build trust. The deal voters want is already in the polls: 99% of likely Texas voters in the Houston survey supported requiring data center companies to pay for the grid upgrades their data centers need, 89% of them strongly. Only 1% opposed. (The survey set aside the small share who were unsure.) I've never seen a policy question come back that lopsided. A company that commits to that in writing and shows it on the bill would be on the side of nearly every voter.
For two years, the industry's pitch for data centers has been about jobs, tax revenue and the race with China. Tuesday's voters were asking a simpler question: who pays for the power? The companies that answer it first, and prove it, will have a much easier time getting the next one built.
Poll Vault has tracked 145 polls that mention data centers this year. See them all →
What Else We Tracked
- Three-Quarters of Americans Want More AI Regulation and Distrust Big Tech — USC Annenberg
- Americans Remain Largely Opposed to Data Centers Near Home, With Slight Thaw — Heatmap Pro / Embold Research
- MAGA Voters Back Continued AI Development but Are Wary of Data Centers Nearby — CNBC
- AI Data Center Construction Matters to About Half of Respondents but Ranks 10th — Gallup
- Supermajority Wants More AI Regulation as Americans Doubt Benefits, Fear Risks — ABC News/Washington Post/Ipsos
- Voters Oppose Local Transmission Lines for AI Data Centers and Doubt Cost Benefits — Data for Progress
- Two-Thirds of Voters Want National AI Rules; Few Rank Regulation a Priority — NetChoice / Echelon Insights
Until next time,
Alex Lundry
Due Data is powered by Poll Vault. Get the full data behind every poll mentioned above.
Want the full data behind every poll mentioned above?
Create free account →