Data center ballot measures went against the industry almost everywhere on November 3. By our tally, voters sided against data centers in 37 of the 38 local measures with votes counted, and the one exception came in a small North Dakota city.
The count combines Ballotpedia’s list of data center-related ballot measures with unofficial results compiled by Data Centers on the Ballot, a tracker run by Redbud Consulting that reads county results pages. All figures are unofficial, as of the tracker’s 10:22 a.m. ET update on Wednesday. Nothing has been certified.
For the companies building AI infrastructure, this was the first nationwide test of local opinion. Many of these measures turn on power, water and tax breaks, which are the same questions now driving utility bills and city budgets.
How we counted 41 measures
Ballotpedia tracks 50 local data center measures in 13 states this year. That breaks down as 41 on November 3, six decided earlier and three still to come in December.
Of the 41 on Tuesday’s ballot, 28 would limit data centers and 13 would help them, Ballotpedia News reported before polls closed. A “yes” on a ban and a “no” on a rezoning point the same way, so we sorted each result by which side won rather than by yes or no.
Here is the arithmetic for those 41:
- 31 finished with a complete unofficial count, and every one went against data centers.
- 6 are partly counted, with the anti-data-center side ahead in each.
- 1 went the industry’s way: Harwood, North Dakota.
- 3 have no results posted on the tracker yet.
That gives 31 + 6 + 1 + 3 = 41. Among the 38 with votes counted, 37 went against data centers. Add the six earlier votes this year, all of which went against data centers, and the total stands at 43 of 44.
All 13 measures written to help data centers are losing or have lost. Two of those, in Downe Township, New Jersey, and Boulder City, Nevada, are still partly counted.
One note on sources. Early Wednesday, Ballotpedia’s results table showed the yes and no columns reversed for Granville and Wilmington, Ohio. We used the tracker’s counts there. For Wilmington, they match Fortune’s report that about 80% of voters rejected the four rezonings.
Ohio went 21 for 21, and the rest followed
Ohio had 21 of the 41 measures, more than any other state. The anti-data-center side won or leads in all 21. Nineteen counts are complete. Oregon, which caps data centers at 25 megawatts, and Grove City are still counting.
Grove City is the closest race in the country. Its Issue 28 would require a public vote on projects above 50 acres, 20 megawatts of peak demand or 500,000 gallons of water a day. It led 54.6% to 45.4% at the last update.
Elsewhere, the margins were large:
| State | Measures on Nov 3 | Against data centers | Still out or exception |
|---|---|---|---|
| Ohio | 21 | 21 (19 complete, 2 counting) | none |
| Illinois | 7 | 6 complete | Morgan County not posted |
| Maine | 4 | 2 complete | North Berwick, Wiscasset not posted |
| Kansas | 3 | 3 (2 complete, Ottawa counting) | none |
| California, Florida, New Jersey, Nevada, Wisconsin | 1 each | 5 (Edgewater and Janesville complete; Alhambra, Downe, Boulder City counting) | none |
| North Dakota | 1 | 0 | Harwood voted for the industry |
In Madison County, Illinois, voters in four townships rejected advisory questions on allowing data centers. Chouteau Township was 83.27% against, St. Louis Public Radio’s Katie Grawitch reported. “I am so stunned by the numbers,” said Heather Groetecke, who led the signature drive.
The tracker also follows five measures that aren’t on Ballotpedia’s list, in four Maine towns and Americus, Kansas. The three counted so far all went against data centers.
Power, water and tax breaks were on the ballot
These measures were mostly about specific limits, not about AI in general. Twelve set megawatt caps, most at 25 megawatts, according to the tracker. Emporia, Kansas, banned “high impact data centers,” defined as needing more than 10 megawatts or 100,000 gallons of water a day, Route Fifty’s Chris Teale reported.
Money was the other thread. In Perry, Ohio, 87% voted to make hyperscale developers seeking tax incentives give the village financial guarantees. Thirty-eight states offered data center tax incentives as of April, Ballotpedia News noted.
Route Fifty says voters cited water supplies, power bills and secretive deals between local officials and developers. In a June Reuters/Ipsos poll cited by Newsweek, 77% of respondents worried data centers could raise their electricity rates. That fear lands on households already paying more, as our look at rising electricity bills shows.
Nine of the measures were advisory and bind no one, so a lopsided “no” in Joliet or Granite City is a message to officials, not a ban.
The industry says it pays its own way
The Data Center Coalition, the industry’s main trade group, says data centers are not the cause of higher bills. In a July 23 statement, its president Josh Levi said the industry is committed to “paying our full energy costs.”
Levi told E&E News in May that the industry’s own research found the buildout had not raised costs for other customers. He said the industry had not pushed those facts hard enough. “The absence of really leaning forward with those facts leaves a vacuum,” he said.
Backers did spend in Ohio. Build American AI ran at least $2.54 million in TV ads there, citing $40 billion in investment and more than 95,000 jobs, Fortune reported, citing DeSmog.
Local officials have made the revenue case too. Festus, Missouri, City Administrator Greg Camp called a $6 billion CRG Clayco project a once-in-a-lifetime revenue opportunity, St. Louis Public Radio reported. In April, Festus voters ousted four of the eight council members anyway.
The one industry win shows how narrow some of these fights are. In Harwood, about 75% of voters rejected a charter amendment requiring independent review of big industrial projects. Route Fifty notes it would not have touched Applied Digital’s campus, which was already approved.
Finance is paying attention. Banks and asset managers backing the buildout have started treating local opposition as a credit risk, Semafor’s J.D. Capelouto reported. For a sense of how the fights play out after a project is built, see our report on xAI’s offer to buy Southaven homes.
December votes in Arkansas and North Dakota are next
The count isn’t over. Maine towns that count their own ballots often post results a day or more later, the tracker notes, and California counts mail ballots for days. The tracker says it will update hourly through November 13, and counties certify after that.
On December 1, voters in Clarksville, Arkansas, decide whether to keep two ordinances authorizing up to $50 billion and $5 billion in industrial development bonds for data center projects, with payment-in-lieu-of-taxes deals. On December 15, Mandan, North Dakota, votes on banning data centers of 25 megawatts or more.
The biggest test may come in 2027. A campaign for an Ohio constitutional amendment banning data centers over 25 megawatts skipped this year’s July 1 signature deadline, Ballotpedia reports. The campaign needed 413,487 valid signatures this year and says it is confident of making the 2027 general election ballot.