Underclassic / explainer

Harwood said yes. Almost every other town said no.

Updated Nov 11 with complete Nov 3 counts for all 47 local measures, a corrected account of Harwood, N.D., and the December votes in Clarksville, Ark. and Mandan, N.D.

Published 2026-10-31, updated 2026-11-11. Sources: county, city and town election offices for each result, linked in the table, as compiled by Data Centers on the Ballot (Redbud Consulting); Washington Post, "The only town in the U.S. that said yes to data centers" (Nov 8, 2026); Fortune, Harwood midterm data-center measure (Nov 7, 2026); Ballotpedia, "Data center-related ballot measures, 2026", as compiled in The GridSight, "Data center votes on the November 3 ballot" (updated Oct 23, 2026); Governing, "Map: Where Data Centers Are on the Ballot" (Oct 30, 2026); Calyx data center register, moratoriums in effect (updated Oct 31, 2026).

Since the vote: Harwood becomes the national story

Five days after Election Day, the Washington Post put it on the front of the tech desk: Harwood, North Dakota, "may be the only place in the United States that voted in favor of data centers last week." The Post's headline calls Harwood the only town in the U.S. that said yes. Fortune, a day earlier, said the same thing another way: of 45 midterm data-center measures tracked by Redbud Consulting, only one went the pro-data-center way, and that one belonged to Harwood.

Harwood sits about 10 miles from Fargo. The Post puts its population at 845; Fortune says about 940. Voters there rejected a Home Rule Charter amendment that would have required independent reviews and public hearings for large industrial projects such as data centers. The North Dakota Secretary of State count, as compiled by Data Centers on the Ballot, shows 325 no to 105 yes — about 75% against the limit. Fortune reports 430 votes cast on the measure.

The fight was already under way. Applied Digital is building a data center in town with Oracle as the end user. The Post describes a $3 billion, 1.8-million-square-foot facility on the outskirts. Fortune describes a 160-acre campus that could expand to 900 acres, scheduled to be fully operational in 2027. Mayor Blake Hankey, Fortune reported, said the project would bring in $1.65 million in taxes by 2028. Residents who organized as Voices of Harwood said they learned of the project only weeks before construction began in September 2025 and that city officials had signed nondisclosure agreements with the developer.

The headline needs one qualifier. The counts bear out the narrow version: of the 47 local measures on the Nov 3 ballot that Data Centers on the Ballot now tracks (Fortune's figure of 45 came before two late-listed measures were added), Harwood's is the only one the data center side won. But Harwood voters did not approve a data center. They turned down a rule that would have required independent review and a public hearing before large industrial projects, and the Applied Digital campus was already being built. "The only town that said yes" is true of ballot measures, not of local government at large: councils and planning boards keep approving projects outside the ballot box. A week after Wilmington, Ohio voters rejected four unrelated rezonings, its planning commission approved Amazon's site plan for the Cosler farm campus, 4–1.

Nov 3 results: limits win nearly everywhere

Every one of the 47 local measures now has a count, and the result is about as lopsided as American elections get. The side limiting data centers won 44 of 45 decided measures, either by passing a restriction or by voting down a question that would have welcomed data centers. Two more are not quite finished: Alhambra, California, where a ban is winning with about 84% while Los Angeles County counts late ballots, and Boulder City, Nevada, where about 70% have voted against opening land to data centers ahead of the council's canvass on Nov. 13. The data center side won only 1: Harwood, North Dakota, where 75.6% voted no on requiring review and a hearing before large industrial projects. On election night, Stateline reported 33 restrictions passed or leading, with Harwood the only win for data center supporters; the later counts only widened the margin.

Notable votes: Wilmington, Ohio rejected all four data center rezonings, each by about 80%. Morgan County, Illinois gave the widest Nov 3 margin, 88.3% for restrictions on an advisory question (Journal-Courier). Joliet Township, Illinois voted 78.9% no on an advisory question backing data centers, and four Madison County townships did the same. Bloomington, Illinois backed its restriction with 77.6%. Janesville, Wisconsin voted 62.8% to require a public vote on projects over $450 million at the old GM site. In Kansas, Emporia, El Dorado, Americus and Ottawa all passed limits. Two late-listed measures also went the same way: Piqua, Ohio banned data centers over 25 megawatts with 66.7%, and Belvidere, New Jersey backed an advisory question on pursuing a ban, 74.8%. The closest race was Sprigg Township, Ohio, where township zoning to regulate a data center at the former Stuart power plant site passed 332 to 301 in the certified canvass; Grove City, Ohio adopted its voter-approval rule with 54.6%.

Counts as of Nov 11, 2026, from the county, city or town source linked in each row, as compiled by Data Centers on the Ballot. The status line under each result says how final it is: certified, complete but unofficial, or still counting, with dates. Most local results are certified within a few weeks of the election, and we will update this table when they are. "Limit" means a yes vote restricts data centers; "Allow" means a yes vote permits them.

What voters decided on Tuesday

Before Election Day, 41 local data center measures in 10 states were on the Nov 3 ballot, according to Ballotpedia's list as compiled by The GridSight. Ohio has 21, more than every other state combined. Residents put 25 of the 41 on the ballot by petition.

The measures cut both ways: 26 would limit data center development and 13 would support it; two in Adams County, Ohio would create zoning. Seventeen would prohibit data centers outright or above a size threshold, in Ohio, Kansas, California, Florida and Maine. Conneaut, Defiance, Oregon and Sunbury in Ohio and Ottawa in Kansas would bar facilities drawing more than 25 megawatts.

Ohio already has more than 200 data centers, and the pushback there is bipartisan: Governing quotes Nick Owens, a former county GOP chair who helps lead the anti-data center nonprofit Conserve Ohio. Five Ohio votes ring the New Albany cluster near Columbus (Granville, Pataskala, Sunbury, Grove City and Ashville), and Wilmington voted on four rezonings covering about 546 acres.

Illinois is mixed: Bloomington and Morgan County ask about a moratorium and restrictions, while Joliet Township and four Madison County townships vote on advisory questions in favor of permitting or approving data centers.

Measures by state

Direction is what a yes vote does. Advisory questions are not binding. Source: The GridSight, from Ballotpedia.

Six data center votes held earlier in 2026 all went toward more limits; in Augusta Township, Michigan, 90% voted to overturn a data center rezoning.

Beyond bans and moratoriums, measures elsewhere would set zoning rules, limit the electric load data centers can draw, or require voter approval before new power and water demand from data center projects.

Why it is on the ballot

Governing cites a Gallup poll from earlier this year finding that seven in ten Americans oppose building an AI data center in their area, with almost half strongly opposed. The main worries are strain on water and electricity, lost farmland or forest, pollution from backup generators that burn fossil fuels, and round-the-clock noise.

Pauses already in effect: 213 places in 32 states

Ballot measures are only the visible part. As of Oct 31, the Calyx register lists 213 cities and counties in 32 states with a local data center moratorium or pause in effect, each entered from its ordinance or a dated report. Ohio leads with 55, followed by Michigan and North Carolina with 17 each and California with 13.

The newest big-city entry is San Diego: its City Council adopted a 45-day moratorium on Oct 27, which the register computes to run until Dec 11, 2026. Other California cities on the list include Oakland, San Francisco, Hayward and Pittsburg, mostly with short urgency ordinances.

The register is deliberately conservative. A broader census it cites, Moratorium Data 2026 by Michael J. Bommarito, counts about 921 local moratoria in 45 states as of Sept 23, 2026; the register adds places only as it reads their documents.

States pull both ways

According to Governing, the National Conference of State Legislatures has identified 16 states where lawmakers proposed moratoriums or other limits on data center construction, and none has passed. Governors in New York, Texas and Oregon have paused construction temporarily by executive action. At the same time, 38 states offer data centers tax incentives.

What it means

The AI build-out runs on land, water and grid connections that local governments control. Every pause or ban pushes projects toward friendlier counties or onto federal land, and adds to the delays already showing up in hyperscaler spending. Harwood's national coverage this week is the other side of that map: the rare place that kept the project and became a symbol of it. That is part of why compute is also drifting onto machines people own: see our buyer's guide to home AI computers and the model price board. We cover the data center fight daily in the Underclassic issue.

FAQ

What were the 2026 data center ballot results?

Limits won almost everywhere. Of the 47 local Nov 3 measures tracked by Data Centers on the Ballot, the side limiting data centers won 44 and leads in the 2 still being counted or canvassed. The data center side won only 1, in Harwood, North Dakota. Most counts are complete but unofficial until certified.

Why did Harwood become national news?

The Washington Post (Nov 8) called Harwood the only town in the U.S. that said yes to data centers, and Fortune (Nov 7) said it was the only pro-data-center outcome among the midterm measures Redbud Consulting tracked. That holds for ballot measures, with a caveat: Harwood voters did not approve a data center, they rejected a charter amendment that would have required independent review of large industrial projects. Councils and planning boards elsewhere keep approving projects outside the ballot box.

What data center votes are coming up?

Clarksville, Arkansas votes Dec. 1 on two ordinances authorizing up to $55 billion in industrial revenue bonds for the Serverfarm data center campus; a yes keeps each ordinance, a no repeals it. Mandan, North Dakota votes Dec. 15 on a citizen-initiated ban on data centers of 25 megawatts or more.

How did Harwood vote?

Voters rejected a Home Rule Charter amendment that would have required independent review and public hearings for large industrial projects: 325 no to 105 yes (about 75% against), per the North Dakota Secretary of State count compiled by Data Centers on the Ballot.

How many data center measures are on the Nov 3, 2026 ballot?

41 local measures in 10 states, according to Ballotpedia's count as compiled by The GridSight. Ohio has 21, more than every other state combined. Data Centers on the Ballot, which added late-listed measures, tracked 47.

How many would limit data centers?

26 of the 41 would limit development (bans, moratoriums, voter-approval or review requirements), 13 would support it, and 2 in Adams County, Ohio would create zoning. 17 would prohibit data centers outright or above a size threshold.

How did these measures get on the ballot?

Residents put 25 of the 41 on the ballot by petition; the rest were referred by local officials.

Why does a yes vote sometimes help a data center?

Referendums ask voters to uphold a decision officials already made, such as a rezoning. A yes keeps the project; a no reverses it. Wilmington, Ohio voted on four such rezonings, about 546 acres.

How many places already have a data center pause?

The Calyx register lists 213 cities and counties in 32 states with a local moratorium or pause in effect as of Oct 31, 2026.

Caveats

• Counts move daily. The 41 measures are Ballotpedia's count as listed by The GridSight (updated Oct 23); the 17 prohibitions are our tally of that list; the 213 pauses are the register's Oct 31 count of measures adopted and in effect.
• Redbud Consulting's count of 45 midterm measures, cited by Fortune, is a separate tracker from Ballotpedia's 41. We keep both attributions and do not merge them into one total.
• Population figures for Harwood differ by source (Washington Post: 845; Fortune: about 940). Vote totals are from the North Dakota Secretary of State via Data Centers on the Ballot.
• The register counts only local pauses that are in effect; proposals, state actions and pauses not yet effective are excluded.
• Some register end dates are computed from adoption date and stated length, not printed in the ordinance.
• Nov 3 results come from county, city and town election offices (and, in Wiscasset and Morgan County, local press reporting the official count), as compiled by Data Centers on the Ballot. Only the two Adams County, Ohio townships and North Berwick, Maine have certified results so far; the rest are complete but unofficial, and Alhambra and Boulder City are still being counted or canvassed. Two Maine towns, Freedom and Wiscasset, gave their counts by phone.
• Data Centers on the Ballot tracks 47 Nov 3 measures, including late-listed Piqua, Ohio and Belvidere, N.J.; Ballotpedia's pre-election list had 41.

Subscribe

$ echo you@email.com | underclassic --subscribe

One email per issue. No tracking pixels, no filler.

Read today's issue · Archive · Model price board

home  |  rss  |  page 2  |  page 3  |  archive  |  models  |  buyer's guide  |  subscribe  |  data  |  board  |  mcp  |  llms.txt