AI Data Centers Become a Deciding Issue in 2026 Midterm Races

More than 54 million Americans have already voted, and AI data centers have become one of the defining issues of the November 3 midterms. Early turnout as of late October ran ahead of both 2018 and 2022 at the same point in the cycle, and forecasters expect early votes to make up two-thirds of all ballots cast.

Brad Carson, head of Public First Action, told Fast Company: "There's hardly a race in the country right now where data centers aren't playing a role in some form." That claim covers a lot of ground. But the numbers behind it are concrete.

How many measures are actually on the ballot?

Thirteen states have AI data center-related measures on the ballot across elections through the end of 2026, according to Ballotpedia. There are 49 measures in total, and 84% of them will be decided in the November midterms.

Ohio leads with 21 measures. The rest break down as follows:

• Illinois: 7
• Maine: 4
• Wisconsin: 3
• Kansas: 3
• California: 2
• Arkansas: 2
• New Hampshire: 2
• Florida, Michigan, Nevada, New Jersey, North Dakota: 1 each

Ohio's concentration is not accidental. Both the governor's race and the Senate races in Ohio, plus the Senate race in neighboring Michigan, have made data centers a central issue, the Washington Post reported.

The ballot measures are only part of the picture. In 20 of the 36 gubernatorial races this year, candidates are calling for a moratorium on AI data centers, according to Multistate, a policy tracking firm. That is more than half of all governorships on the ballot.

What do voters actually think?

The polling explains why candidates are running this way. Two-thirds of Americans oppose building new AI data centers, according to a Washington Post poll. That opposition is broad, not shallow.

Fifty-seven percent of Americans say artificial intelligence is bad for society. One-third of those respondents describe it as "very" bad. That is roughly one in five Americans overall holding the most negative view available on the survey.

A separate Washington Post-ABC News-Ipsos poll found that 67% of Americans—Republicans and Democrats alike—do not think the government is doing enough to regulate AI. Cross-party agreement on any regulatory question is rare. Its appearance here suggests the political risk of a pro-AI stance is no longer confined to one side of the aisle.

The skepticism does not stop there. A number of Republican candidates are distancing themselves from President Donald Trump's pro-AI policies, the Post reported. When members of a president's own party break with him on an issue, that usually signals polling they have seen privately.

What is not being said?

The source data describes the political backlash but says nothing about the economic counterargument—jobs, tax revenue, or local investment that data center projects typically promise. Candidates calling for moratoriums are not, at least in this record, addressing what would replace that investment. Voters on November 3 will render a verdict on the backlash; the harder question of what happens to the buildout afterward remains open.

Whatever the results, 41 ballot measures decided in a single day amount to the broadest public referendum on AI infrastructure the United States has held. The industry's local politics have now caught up with its capital spending.