AI
“In Tuesday's midterm election, all but one of more than 40 counties, cities, and towns across the United States that voted on measures related to local AI data centers chose to restrict, limit, or ban the facilities.”
Plain restatementIn the November 3, 2026 US midterm election, more than 40 local ballot measures concerning data centers were decided, and in all but one the outcome went against data center development.
This one checks out in substance. Independent trackers agree that data center ballot measures went against the industry almost everywhere on November 3, 2026. Ballotpedia counted 41 local measures across 10 states and found voters favored limiting data centers in 40 of them, with the only pro industry outcome in Harwood, North Dakota, a city of roughly 800 people where 75 percent of voters rejected a proposed charter amendment adding review requirements for large industrial projects. Two details in the post are looser than the evidence. About a third of those 40 outcomes were voters defeating measures that would have helped data centers, including four rezoning questions in a single Ohio city and several non binding advisory questions, rather than voters enacting new restrictions. The figure of more than 40 counts measures, not places, and the number of distinct towns and counties is lower, with Ohio alone accounting for 21 measures across 18 municipalities. All results remain unofficial until certified, and three more data center measures are scheduled for December votes in Arkansas and North Dakota.
In [drifted from the evidence:] Tuesday's midterm election, [drifted from the evidence:] all but one of more than 40 [drifted from the evidence:] counties, cities, and towns across the United States that voted on measures related to local [drifted from the evidence:] AI data centers [drifted from the evidence:] chose to restrict, limit, or ban the [drifted from the evidence:] facilities.
In [added by the neutral restatement:] the November 3, 2026 US midterm election, more than 40 local [added by the neutral restatement:] ballot measures concerning data centers [added by the neutral restatement:] were decided, and in all but one the [added by the neutral restatement:] outcome went against data center development.
Red-tinted words in the claim drifted from the evidence. Green-tinted words are what a neutral restatement needs.
The trace / claim to source
- More than 40 data center related local ballot measures were decided on November 3, 2026. Ballotpedia's verified list is 41 measures in 10 states; the tracker the post cites lists 47.
- In all but one of those contests the outcome went against data center development. Ballotpedia records 40 of 41, and the Washington Post independently counted at least 38 jurisdictions passing bans or restrictions.
- The single exception was Harwood, North Dakota, and the post describes it correctly: voters rejected, rather than approved, a measure that would have added requirements for certain industrial projects.
- The post's description of Harwood's size is in line with local reporting, which puts the city's estimated population at around 800, with 430 votes cast.
- The post's description of the Harwood project as a roughly 900-acre data center with a 2027 timeline matches company and press accounts of Applied Digital's Polaris Forge 2, which contracted more than 900 acres and is due to reach full operation in early 2027.
- Ohio was the center of gravity, with 21 of the 41 measures across 18 municipalities.
- The claim says these places "chose to restrict, limit, or ban the facilities," but by Ballotpedia's own breakdown only 27 of the 40 were voters approving something restrictive. The other 13 were voters defeating measures that would have helped data centers, including four rezoning referendums in a single Ohio city and seven non-binding questions. Rejecting a rezoning is a real setback for a project, but it is not the same act as enacting a restriction, and the claim's wording merges the two.
- At least nine of the measures were non-binding advisory questions, seven support measures that were defeated and two opposition measures that passed, which express voter sentiment without changing law. The claim presents the whole set as communities choosing to restrict or ban.
- The claim counts "more than 40 counties, cities, and towns," but the figures in the underlying trackers count measures, not jurisdictions. Ohio's 21 measures were spread across 18 municipalities and four of the measures were in Wilmington alone, so the number of distinct localities voting is lower than the number of measures. The Washington Post's independent count is phrased in jurisdictions and lands at least 38.
- Date context mismatch, mild: the post's scorecard asserts a complete picture with zero contests too close to call and all 229 precincts in, while trackers on the same day still showed six counts open on the larger 47-measure list. All results at that point, and still today, are unofficial returns awaiting certification.
- Exaggeration, scope of subject matter: the measures are data center measures, not all of them AI specific, and some reach industrial development generally. The sole exception illustrates this: Harwood's mayor said the rejected amendment was not strictly aimed at data centers and would have applied to any business coming to town.
- The exact denominator. Ballotpedia's verified list is 41, the tracker the post cites says 47, and other outlets used 44, 45 and 46. The inclusion rules that separate these lists are not fully published, so there is no single authoritative count of how many measures qualify as data center related.
- The exact number of distinct localities, as opposed to measures, that voted. No source retrieved states this figure directly, and the Post's "at least 38 jurisdictions" counts only those that passed restrictions.
- Whether any result shifts on final canvass. Several contests were decided by narrow margins, including a reported 55 to 45 charter amendment in Grove City, Ohio, and certification is still pending.
- The caption's assertion that activists in hundreds more locales are working to put data center measures on special election ballots. No source was found that quantifies this, and it remains uninvestigated.
- The claim's link to newly elected officials and future AI oversight is a forecast, not a verifiable present fact.
Ballotpedia, which tracked the measures before and after the election, reports that voters in 10 states decided 41 ballot measures affecting local data center policy on November 3, with a "yes" vote on 28 of them limiting data center development through policies such as bans or moratoriums and a "yes" on the remaining 13 facilitating development through policies like zoning. Its post-election tally states that voters favored limiting data center development in 40 of the 41 local ballot measure contests, broken into two parts: voters approved 27 measures that limited data center development or expressed opposition to it, and voters defeated 13 measures that would have favored data centers or expressed support for development. The defeated pro-development group included seven non-binding support measures, five in Illinois, one in New Jersey and one in Ohio; four rezoning measures, all in Wilmington, Ohio, that would have rezoned specific areas of land from residential to industrial use allowing data center construction; and one measure that would have added data centers as an approved use for a particular area of land in Boulder City, Nevada. The approved restriction group included three six-month moratoriums on data center development, all in Maine, and two non-binding measures expressing opposition to data center development in Illinois, along with five full data center bans: one in California, one in Florida, and three in Ohio. Ohio accounted for the largest share. Ohio residents across 18 municipalities voted on 21 ballot measures on data centers, more than half of the 41 nationwide, and at least 14 municipalities passed bans or other restrictions. On the single exception, local reporting confirms the detail. Harwood's home rule charter amendment, known as "Article X," which would have established requirements for the city to follow when handling certain industrial projects, was rejected by 75% of voters, with 430 total votes cast in a city whose estimated population is around 800. The amendment was created by members of Voices of Harwood, a group that organized in response to how the city handled the Applied Digital data center proposal. The mayor, who opposed the measure, said the amendment was not strictly aimed at data centers and covered any kind of business coming to town. Independent counts converge on the same direction with different denominators. The Washington Post reported that voters in at least 38 jurisdictions passed bans or restrictions on data centers. Fox Business reported that voters backed restrictions or rejected pro-development measures in all but one of 41 local ballot measures across 10 states, citing Ballotpedia analysis and results compiled by CNBC, and noted the Post's separate count of at least 38 jurisdictions. The tracker the post cites counts more measures: as of 12:30 p.m. Eastern on Nov. 5 it listed 47 measures with results, 41 counts complete for election night and six still being counted.
Complete reasoning
The reply is formatted for pasting into the thread where the claim is circulating.
Ask this case
Answers come only from the case file above; nothing is added.
Did more than 40 places really vote against data centers?
More than 40 ballot measures were decided, not more than 40 distinct places. Ballotpedia's verified count is 41 measures in 10 states, but Ohio alone had 21 measures across only 18 municipalities, so the number of separate towns and counties voting is lower than 40.
Is it accurate to say these communities chose to restrict, limit, or ban data centers?
Only partly. Ballotpedia found 27 of the 40 outcomes were voters actively approving restrictions like bans or moratoriums, while the other 13 were voters defeating measures that would have helped data centers, including four rezoning votes in one Ohio city and several non-binding questions. Rejecting a pro-development measure is not the same as enacting a new restriction.
What was the one exception, and what happened there?
The exception was Harwood, North Dakota, a city of about 800 people. Voters there rejected, by a 75 percent margin, a charter amendment that would have added review requirements for large industrial projects, including data centers.
Were all these measures specifically about AI data centers?
Not entirely. Some measures covered industrial development more broadly rather than data centers specifically. Harwood's mayor said the rejected amendment was not aimed strictly at data centers and would have applied to any business coming to town.
Is 41 the final, official number of measures?
No single authoritative count exists. Ballotpedia's verified list says 41, the tracker cited in the original post lists 47, and other outlets used figures of 44, 45, and 46. All results were unofficial and awaiting certification at the time of reporting, and three more measures were scheduled for December votes in Arkansas and North Dakota.