The Data Center Vote: Where America's AI Boom Meets Political Resistance

Data centers have become one of the loudest issues of the 2026 midterms. Campaigns have spent more than $53 million on ads that mention them, national polls show broad opposition to building them, and on Nov. 3 voters in 34 communities will decide local ballot measures on whether, where and how they get built.

The election will answer some questions about how much the issue moves voters. For anyone trying to site a data center, though, the more lasting effect may show up in county zoning codes and permitting rules, and that shift is already under way.

A Pew Research Center survey conducted Sept. 28 to Oct. 4 found that 61% of registered voters dislike elected officials who support building data centers in their area. Just 8% said they like it. Among supporters of Democratic candidates, 75% disliked those officials. Among supporters of Republican candidates, 44% did, and 14% said they liked them.

A Quinnipiac University poll released Sept. 29 asked a different question: "Would you support or oppose the building of an AI data center in your community?" Seventy-three percent of registered voters said they'd oppose it, including 85% of Democrats, 76% of independents and 55% of Republicans. Quinnipiac surveyed 1,032 registered voters Sept. 24 to 27, with a margin of error of plus or minus 3.8 percentage points.

The two polls measure different things. Quinnipiac asked about the project. Pew asked about the politicians who back it. Read together, they suggest that supporting a data center now carries political risk across party lines, though the risk is much higher with Democratic voters.

Neither poll shows how many votes the issue will actually move. Someone who doesn't want a data center down the road may still vote on inflation, health care or immigration. The polls establish vulnerability, and the vote count will have to wait for November.

The industry's pitch to communities has been the classic economic development pitch: billions in capital investment, construction activity and a big new property-tax payer. Pew's tracking suggests the public is weighing something else.

In a survey conducted July 20 to Aug. 9, half of U.S. adults said data centers are mostly bad for home energy costs, up from 38% in January. The share who called them mostly bad for the environment rose from 39% to 54%, and for the quality of life of people living nearby, from 30% to 49%.

Jobs didn't help much. Twenty-two percent said data centers are mostly good for local jobs, and 24% said mostly bad. In January, the split leaned the other way, 25% good and 15% bad.

That's a hard spot for an industry whose projects can cost billions but employ relatively few people once construction ends. A residential customer worried about a rising power bill isn't likely to be won over by the size of the capital investment. Developers will increasingly have to show specific, local benefits people can see.

Citing AdImpact data, the Cook Political Report reported Sept. 29 that campaigns and outside groups had spent more than $53 million on ads mentioning data centers in House, Senate and governor's races. Democratic advertisers accounted for more than $38 million. New York's 17th, Wisconsin's 3rd and Pennsylvania's 8th congressional districts had each seen more than $1 million.

Ad spending shows where campaigns think the issue lands. It doesn't show whether it changes anyone's vote. What it does confirm is that data center siting has moved from planning commission meetings into national campaign messaging, and that a project's political environment can now change with the next election.

People inside the industry concede they didn't see it coming. Matt Landek, CEO of L5 Mission Critical, the data center operations company Walbridge launched in September, told Location Decision earlier this month that the industry assumed "the community just will continue to not care what happens, especially as it became politicized." He doesn't think the industry got the message wrong so much as it was caught flat-footed. "I just think the overall industry … might have just been unprepared," he said. "I hope maybe after the midterm elections, people will be able to have a more honest conversation about what's happening."

The most direct read on public sentiment will come from local ballot measures. According to Ballotpedia, 38 data center measures are scheduled in 2026 across 35 jurisdictions in eight states, and 34 of them are on the Nov. 3 ballot.

Ohio has 20 of them. Illinois has seven, Wisconsin and Kansas have three each, California has two, and Michigan, Nevada and Florida have one each. The measures cover moratoriums, zoning changes, project approvals and outright bans. Thirty-one were put on the ballot by citizen petition. Nine are advisory, and the other 29 would enact, amend, approve or reject laws, zoning decisions or development agreements.

Voters decided four measures earlier this year. Each time, they sided with more restrictions, either by approving limits or rejecting a measure that would have made development easier.

A zoning referendum isn't the same as a congressional race, and the wording of each question and the project behind it will shape the results. But for a developer with a site under contract, a local vote that changes the zoning matters more than who wins the House.

Jeremy Solomon's company, Learnewable, scores U.S. counties on community sentiment and permitting conditions for energy and data center projects. When we talked with him in late September, he said data centers stood apart from everything else he tracks. "Data centers is pretty dramatic. It's changed. It's gotten worse," he said. Sentiment toward solar and wind, by comparison, has improved.

Two patterns stood out to him. Opposition grows with exposure: "The more data centers that are in a county … the more negative the sentiment is toward additional data centers," he said. "That is for real." And more populated counties are less receptive, which leaves rural places with fewer residents to object as the likeliest landing spots. Those happen to be the places developers are already heading.

AI has pushed data center developers beyond established markets into places with open land and available power. Many of those places are now where the backlash is building.

Site selection for data centers has always started with electricity, transmission, fiber, land costs, taxes and permitting timelines. Community acceptance mattered too, but it rarely sat near the top of the list. Every new moratorium or zoning limit widens the gap between the sites that work on paper and the sites a community will actually approve.

Site selectors say that's already reshaping their shortlists. "We're increasingly hearing about companies making decisions to even eliminate places that have moratoriums on data centers from their selection criteria," Taylor Stepp of Strategic Development Partners told Location Decision in an interview this month.

Stepp said checking a community's mood has gone from an afterthought to standard practice. "Three years ago, this wasn't even really on anyone's radar," he said. Now "it is a standard component of every project that we work on." The stakes for consultants are personal: "If you're in my shoes and you … provide a finalist site and the company gets embarrassed because of … out of left field negative community sentiment, you're gonna get fired." He put it bluntly. "The rejection rate because of the fear of the unknown has never been higher."

Timing makes it harder. These projects take years of utility coordination, capital commitments and approvals. Opposition that shows up after a site is chosen, or a new county board elected halfway through, can stall a project that looked sound on paper. Developers now have to ask whether a community has the power and the land, and also whether it will still support the project by the time it's ready to energize.

Three things are clear before the election. Opposition to local data center development is widespread. It extends to the officials who support it. And communities are acting on it, through zoning restrictions, moratoriums and ballot measures.

What isn't clear is how much the issue changes who wins. Answering that would take polling that tests whether data center positions move candidate preference, followed by a look at actual results. Few people in the business expect the issue to fade after Nov. 3, either. "I think this will probably go through the next presidential election," Stepp said.

For the development side, though, that question matters less than it might seem. Local restrictions change investment decisions whether or not data centers decide a single congressional race. The industry's political exposure will be measured in votes on Nov. 3, and also in the number of counties where building a data center becomes slower, more expensive or impossible. That second number is the one site selectors will be living with long after the election.

Sources: Pew Research Center (Oct. 8 midterm report and Sept. 22 short read), Quinnipiac University Poll (Sept. 29), Cook Political Report with AdImpact data (Sept. 29), Ballotpedia (Sept. 21) and Location Decision interviews with Matt Landek, Jeremy Solomon and Taylor Stepp.