Voters Said No to Data Centers in 46 of 47 Local Ballot Votes. Here Is What They Actually Decided

By Thorsten Meyer | November 5, 2026 | ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

• Business-only prices and quantity discounts
• Tax-exempt purchasing
• Multiple users, one account, clear invoices

Create a free Business account →Free to register

As an affiliate, we earn on qualifying purchases.Hide

In Hubbard, Ohio, a small city near Youngstown, voters were asked a blunt question on November 3: should data centers be banned inside city limits, of any size? The answer, on the unofficial count, was 2,622 yes and 802 no: 76.6% for a ban. In Wilmington, Ohio, about four in five voters undid four council decisions that had rezoned rural land for data centers. In Boulder City, Nevada, seven in ten said no to opening city-owned desert to them.

Headlines the next morning said “45 measures” and “overwhelmingly”. Both hold up, with a few footnotes about counting. I followed the number back to where it came from, checked results on official election sites, and read what the ballots actually said. The short version: on unofficial counts, the side against data centers won or leads in 46 of 47 local measures. But “voters rejected AI data centers” covers very different things, from outright bans to a 180-day pause to a question that changes nothing on its own.

Key takeaways

• 46 of 47. A public tracker, Data Centers on the Ballot, lists 47 local measures about data centers on the November 3 ballot in 10 states. On counts as of November 5, the anti-data-center side won or leads in 46. The one exception is Harwood, North Dakota. All results are unofficial; nothing is certified yet.
• Where “45” came from. That was the same tracker on the morning after the election: 45 measures, 40 of them with results. It added two more later. Ballotpedia, which counts differently, listed 41 measures before the vote.
• I checked seven results at the source. County and state election sites in Kansas, Nevada, Florida and North Dakota match the tracker to the vote.
• Not all “no” votes are equal. 37 measures are binding; 10 are advisory questions. They include outright bans, size caps (often 25 megawatts), temporary moratoriums, rezoning referendums and rules that require a public vote first.
• Power and water against jobs and tax money. In Gallup’s polling, opponents name water and energy; supporters name jobs and tax revenue. The ballot texts themselves measure data centers in megawatts and gallons per day.
• For businesses, local consent is now a site risk. Many votes target future projects, not campuses already under construction. But in these towns, the next project may need to win a public vote.

Products Worth Considering

EcoFlow Delta 3 Max Portable Power Station – 2048 Wh LiFePO4 Battery, 2400 W Output (3400 W X‑Boost)

AmazonView Latest Price

Bose SoundLink Flex Bluetooth Speaker (2nd Gen) – Portable Outdoor Speaker with Hi‑Fi Audio, Waterproof and Dustproof, USB‑C, Up to 12 Hours

AmazonView Latest Price

LG UltraGear 32GS60QC-B 32-inch Curved QHD Gaming Monitor (180Hz, 1ms, 1000R, AMD FreeSync, HDR10)

AmazonView Latest Price

As an affiliate, we earn on qualifying purchases.

Where “45 measures” comes from

The number traces back to one source: Data Centers on the Ballot, a results tracker built by Redbud Consulting, a polling and data firm. It lists a measure when the ballot question itself is about data centers: bans and megawatt caps, zoning referendums on data center sites, agreements with developers, voter-approval rules and advisory questions. It maps every “yes” and “no” to two sides, Restrict (the anti-data-center outcome) and Allow, so differently worded questions can be added up. Its own note says nothing on the page is certified.

The tracker publishes its data openly, so its history can be checked. At 6:16 a.m. Eastern on November 4 it showed 45 measures on the November 3 ballot, 40 with results: 39 with Restrict ahead, 1 with Allow ahead, and 5 without posted results. That matches what Tom’s Hardware and Engadget reported on November 4: 33 won, 6 leading, 1 lost, 5 not yet reported.

Later that day the tracker added two measures: a non-binding question in Belvidere, New Jersey, and a charter amendment in Piqua, Ohio. As of 12:30 p.m. Eastern on November 5 it lists 47 measures, all with results. Restrict leads in 46. 41 counts are complete for election night, 6 are still counting. Together they add up to 260,177 votes, 72.8% of them on the Restrict side.

Other counts differ for honest reasons. Ballotpedia listed 41 measures in 10 states on election day: 21 in Ohio, 7 in Illinois, 4 in Maine, 3 in Kansas and one each in California, Florida, Nevada, New Jersey, North Dakota and Wisconsin. The tracker has the same numbers for most states, but 8 Maine town measures instead of 4, 4 in Kansas instead of 3, and 2 in New Jersey instead of 1. That explains the gap from 41 to 47. 404 Media wrote of “40+ races across 13 states”, a count that also includes recalls and candidate races.

My own check. I could not verify all 47 results myself, because many county sites block automated access or post scanned PDFs. I checked seven at official election sites. All seven match the tracker exactly:

• Emporia, Kansas: ban on data centers over 10 megawatts, 4,346 yes to 2,891 no (Lyon County, unofficial final).
• Americus, Kansas: ban on “high-impact” data centers, 274 to 127 (Lyon County).
• El Dorado, Kansas: ten-year ban, 2,972 to 1,183, or 71.5% (Butler County).
• Ottawa, Kansas: ban on data centers over 25 megawatts, 2,882 to 1,129 (Franklin County).
• Edgewater, Florida: charter ban on new data centers, 7,875 to 2,328 (Volusia County).
• Boulder City, Nevada: Question 1, allowing data centers on city desert land, 2,064 yes to 4,947 no (Clark County, unofficial).
• Harwood, North Dakota: review rules for large projects, 105 yes to 325 no (North Dakota Secretary of State, unofficial).

For the rest, including Ohio, I rely on the tracker, which links each race to its county source, and on local reporting where it exists. For example, the Tribune Chronicle reported Hubbard’s ban passing 76% to 24%.

What voters actually decided

“Voters rejected data centers” sounds like one decision. On the ballot, it was many different ones.

• Outright bans. Hubbard, Ohio banned data centers of any size. Urbana, Ohio wrote a ban into its city charter that, according to the tracker’s summary, exempts only small private facilities under 7.5 megawatts. Edgewater, Florida and Alhambra, California locked in bans their councils had already passed, so a future council cannot quietly undo them.
• Size caps. Ten measures set a limit of 25 megawatts, nine of them Ohio charter amendments plus Ottawa, Kansas. Emporia and El Dorado in Kansas drew the line at 10 megawatts.
• Pauses. Seven Maine towns voted on moratoriums, mostly for 180 days, while they write permanent rules. A pause ends. In Wells, the tracker’s summary says it runs out in mid-December unless the town extends it.
• Rezoning referendums. Wilmington’s four votes overturned the rezoning of about 546 acres. According to the tracker and local reporting, that land is separate from the Amazon Web Services campus the city had approved earlier.
• “Ask us first” rules. Grove City, Ohio voted to require a citywide vote on any project of 50 acres or more, or one using more than 20 megawatts or 500,000 gallons of water a day. In Janesville, Wisconsin, 16,779 to 9,932 voters required a public vote on any deal over $450 million at the former GM plant site. The question never says “data center”, but it was written in response to a proposed 800-megawatt campus there.
• Advisory questions. Ten measures were non-binding. In Joliet Township, Illinois, the question was the most direct of all: “Should the Joliet City Council approve construction of AI data centers?” According to the tracker, 78.9% said no. On its own, an advisory vote approves or blocks nothing.

The direction was remarkably consistent. On the 34 measures where “yes” meant restricting data centers, 33 passed or lead. On the 13 where “yes” meant allowing them, all 13 lost or trail. Many of these questions got to the ballot because residents pushed them there: by the tracker’s classification, 23 came from citizen petitions and 4 reached the ballot by court order.

The one exception tells its own story. In Harwood, North Dakota, a citizen petition asked for an independent review, a public hearing and written findings before the city approves very large projects: 100 megawatts, 100 acres or 50,000 gallons of water a day. Voters rejected it, 325 to 105. According to the tracker’s summary, the measure would not have touched the AI campus already approved next to the town.

Power and water versus jobs and tax money

Why do people vote like this? Polls point to the same trade-off the ballots describe.

In Gallup’s survey from March, published in May, seven in ten Americans opposed building an AI data center in their local area, 48% strongly. Half of the opponents named the use of resources, including 18% each who mentioned water and energy. Of the supporters, 55% named jobs and 13% named tax revenue. The NBC News exit poll on November 3 asked voters whether they would favor or oppose a new data center in their community: 76% said oppose.

The scale behind the worry is real. A report by Lawrence Berkeley National Laboratory, released by the US Department of Energy in December 2024, found that data centers used about 4.4% of US electricity in 2023 and could use 6.7% to 12% by 2028. You can see that in the ballot texts. They don’t measure data centers in square feet so much as in megawatts and gallons: 10, 20, 25, 50 or 100 megawatts, 100,000 or 500,000 gallons of water a day.

The other side of the ledger is jobs, and that is where trust is thin. TIME reported in October on a report by three Democratic senators. It says some companies told investigators that permanent staffing runs at roughly one worker per megawatt, and that sales-tax exemptions on computer equipment are the most valuable incentive. That is an investigation by senators of one party, not a neutral audit. But it matches what one Ohio village asked for. In Perry, 87% voted that any hyperscale data center seeking a tax break must first post a bond covering its projected tax revenue plus the pay of 200 full-time workers.

There is one more nuance. Data centers were not a top national issue. In Gallup’s October poll, 30% of registered voters called the construction of AI data centers extremely important, second to last among 11 issues; the economy and healthcare led at 50% each. Data centers rarely decide a national race. Where a ballot question names a specific site down the road, they decide a lot.

What this means for businesses

• Local consent is now part of site risk. In many of these places, a project now needs more than a council vote; it may need to win a public vote, and organizers can collect the signatures to get there. Budget for that time and that uncertainty.
• Read the local rules before you buy land. Ohio cities can amend their charters by citizen petition; several of these bans went in that way. Moratoriums, megawatt caps and voter-approval rules are spreading among neighboring towns.
• Size thresholds shape designs. Where a 25-megawatt cap applies, a hyperscale campus is off the table, while smaller facilities may still be possible. Whether the industry shifts toward smaller sites is an open question, not something these votes prove.
• Existing projects mostly continue. According to Fortune, Trenton’s ban is unlikely to affect a nearby AWS campus already under construction. Harwood’s measure exempted the approved campus. The votes hit the next project, not the current one.
• Secrecy backfires. In Hubbard, the mayor confirmed he had signed a non-disclosure agreement to study a large project, and residents then gathered more than 750 signatures for a ban, the Tribune Chronicle reported. Publish power, water, noise and job numbers early, and expect to be held to them.
• For anyone buying AI compute: fewer welcoming sites can mean longer timelines for new capacity. That is a cost risk to watch, not something these results measure.

For the money side of the boom, see who pays for the AI data center buildout, and for the utility bills, who pays for the power and water of “compute campuses”.

What this means for citizens

• Know which kind of vote you cast. A charter amendment is hard to undo. A moratorium expires. An advisory question sends a message but changes no law.
• A ban can still be challenged. In Emporia, according to the tracker’s summary, the ban takes effect only after a court rules on the city’s challenge to the petition. Watch the follow-up, not just election night.
• Results are not final. Counties certify in the coming weeks; six counts were still moving on November 5.
• Ask for the numbers. If a project comes to your town, ask how many megawatts, how much water a day, how many permanent jobs, and who pays for new power lines.

The limits of this story

• These are mostly small places. All 47 measures together counted about 260,000 votes. They show local sentiment where opposition was organized enough to reach the ballot, not a national referendum.
• Selection matters. Measures appear where people are already worried. Towns that welcomed a data center rarely held a vote.
• “AI” is the headline, not always the ballot. Some questions name AI data centers, like Joliet’s. Many simply say “data centers”, and several never mention them at all.
• One tracker does the counting. It is transparent and its data is public, but it is one compiler. I checked seven results at official sites; the other 40 rest on its sources and on local reporting.
• Nothing is certified. Numbers can still shift, especially in the six races still counting.

Sources

• Data Centers on the Ballot (Redbud Consulting): results tracker, method notes and public data; snapshots of 4 Nov 2026, 6:16 a.m. ET and 5 Nov 2026, 12:30 p.m. ET.
• Official election results: Lyon County, KS (unofficial final, 3 Nov 2026); Butler County, KS (updated 4 Nov 2026); Franklin County, KS; Volusia County, FL; Clark County, NV (unofficial, 4 Nov 2026, 11:49 p.m.); North Dakota Secretary of State (unofficial).
• City of Harwood: Home Rule Charter amendment text.
• Ballotpedia News, “Voters in ten states will vote on data center-related ballot measures November 3” (3 Nov 2026).
• Gallup, “Americans Oppose AI Data Centers in Their Area” (13 May 2026) and “The Issues Voters Say Matter Most This Midterm” (October 2026).
• NBC News exit poll, 2026 elections (updated 4 Nov 2026).
• US Department of Energy / Lawrence Berkeley National Laboratory, data center electricity report (20 Dec 2024).
• Reporting: Tom’s Hardware, Engadget and 404 Media (4 Nov 2026); Fortune (4 Nov 2026); Tribune Chronicle (4 Nov 2026); TIME (8 Oct 2026).

This article explains election results and public data. It is not legal or investment advice. All vote counts are unofficial until certified.