1. What did voters decide on AI data centers?
Voters in at least 29 U.S. jurisdictions approved bans or restrictions on data centers in elections held on Tuesday, November 3, a result that complicates the rapid buildout of the infrastructure behind artificial intelligence. The measures differed widely, from outright prohibitions on new facilities to requirements that major projects win an additional round of public approval before construction can begin.
Taken together, the outcomes show that local frustration with large computing campuses has moved from public hearings and zoning meetings onto the ballot itself. Electricity bills, water supplies and the lack of public oversight dominated the debate in many of the communities that voted.
Not every measure carries the force of law. Some were advisory questions that signal community sentiment without legally stopping a project, so passage does not automatically halt construction. Even so, supporters of moratoriums now have election results to point to as they push for firmer rules, including possible statewide limits.
2. How many data center measures were on the ballot?
A tally of ballot questions counted 41 data-center measures on the November 3 ballots across 10 states. Of those, 28 would curb development if approved, while 13 would allow or support projects. The split matters because a defeat for one of the pro-development measures can be read as a loss for the companies and local officials that backed it, in the same way that a win for a restriction can.
The breadth of the count also helps explain why the 29 approvals are being treated as a trend rather than a scattering of local disputes. Measures appeared in several regions of the country and in communities with very different economies, which suggests the concerns are not confined to a single state or a single kind of voter.
For developers, the practical issue is predictability. A campus can take years to plan, finance and equip, and a ballot result that adds a public vote or bars a use after land has been assembled can unsettle that timetable.
3. Why did Ohio become the center of the backlash?
Ohio produced some of the clearest results of the night. All 10 charter measures backed by the grassroots group Conserve Ohio passed. Charter measures amend a municipality's founding document, which gives them more standing than an informal resolution and makes them harder for local officials to set aside.
In Hubbard, Ohio, more than three-quarters of voters supported a ban on data centers. In Granville, Trenton and Sunbury, voters backed restrictions by margins of roughly 78% to 82%. Margins of that size leave little room to describe the outcomes as the work of a narrow, motivated minority.
The Ohio results also line up with the political calendar. Ohio Governor-elect Amy Acton supports a pause on new projects and stricter rules, so the state's incoming executive and the communities that voted share a general direction, even if the details of any statewide policy remain to be written.
4. What happened in Illinois and Wisconsin?
Outside Ohio, voters in other Midwestern communities delivered similar messages. In Joliet Township, Illinois, nearly 80% of voters rejected an advisory proposal that expressed support for AI data centers. Because the question was advisory, the vote does not legally block anything, but a rejection of that scale gives opponents a strong argument in later negotiations.
In Janesville, Wisconsin, more than 60% of voters approved a requirement for another public vote before major development can proceed at a former General Motors site that has been linked to a proposed data center. The measure does not necessarily kill the project. It does ensure that residents, rather than only elected officials and developers, will have a say before large-scale work begins on the property.
The Janesville case is a reminder that many of these fights concern former industrial land. Sites that once employed large workforces can look attractive for new development, but residents increasingly want to know what they would receive in exchange, and how much power and water the new occupant would draw.
5. Why are electricity and water at the heart of the fight?
Data centers house the chips and servers used to train and run AI systems, and they are among the most power-hungry commercial buildings in operation. Because their electricity demand can be large and constant, proposed campuses have become flashpoints over utility bills and the capacity of local grids to serve homes and businesses at the same time.
Opponents raise a wider set of objections as well. Water use for cooling is a recurring concern in communities where supplies are already under strain. Noise from equipment and the scale of the buildings are other complaints, along with a more procedural grievance: that agreements with developers are sometimes reached without meaningful scrutiny from the people who live nearby.
That last point helps explain why so many of the measures focused on process rather than outright bans. Requiring an extra public vote, or tighter approval rules, speaks directly to the sense that decisions were made out of sight.
6. How are AI companies responding?
AI companies have acknowledged the backlash and are answering with more lobbying and public relations work. OpenAI has reportedly tried to recast data centers as "compute campuses," an effort to distance the facilities from the negative associations that the older term now carries in many communities.
Industry leaders have also warned about the economic stakes of a broad crackdown. BlackRock CEO Larry Fink said last month that a blanket moratorium on data centers could carry serious economic consequences, and in his remark he warned that such a pause would push the country into recession. His argument reflects a view common among investors: that spending on AI infrastructure has become a major driver of growth and that restricting it could weigh on the wider economy.
The disagreement is therefore not only about individual sites. It is about whether the pace of AI investment should be set by the companies and financiers funding it or shaped more directly by the communities asked to host the buildings.
7. What role did the elections for governor play?
The local votes coincided with statewide results that favored candidates promising tougher oversight. In Texas, Democrat Gina Hinojosa defeated Republican Gov. Greg Abbott after campaigning for a moratorium on new data-center projects. In Ohio, Governor-elect Amy Acton has likewise backed a pause and stricter rules.
Those outcomes matter because the most consequential limits on data centers are likely to come from state capitols as well as town halls. Local bans can steer individual projects elsewhere, but statewide rules on power procurement, water permits or approval procedures would apply across much larger territory and could change where companies decide to build.
For the industry, the combination of local ballot wins and sympathetic governors raises the odds that the debate will move from isolated skirmishes to formal policy.
8. What comes next for data center development?
The immediate effect of the vote will vary from place to place. Where bans passed, new proposals face a clear obstacle. Where measures were advisory, officials still decide whether to treat the result as binding in practice, and developers must judge how much political risk they are willing to carry.
The larger question is whether the results prompt stricter statewide rules. Wins by moratorium supporters could lead to such rules, and the outcomes in Texas and Ohio suggest that incoming leaders will have both the mandate and the motive to pursue them. Companies that depend on new capacity for AI training and services are likely to respond with more community engagement, more public commitments on power and water, and more lobbying.
Investors and local officials will also be watching how communities that approved restrictions handle projects already in the pipeline. Questions about whether existing agreements are grandfathered, whether land already purchased can be used another way, and whether neighboring jurisdictions will adopt similar limits are likely to shape the next round of announcements. Each answer will help show how far the election results reach in practice. Those details, more than the headline total, will decide how much new capacity actually gets built over the coming year.
What is already clear is that the politics of AI infrastructure have shifted. For voters in these 29 communities, the question was not whether artificial intelligence is useful, but who should bear its costs and who gets to decide.
9. How do ballot measures differ from zoning decisions?
Most data-center approvals have traditionally been handled through zoning boards, planning commissions and utility regulators, bodies that meet in public but often hear from a small number of residents. A ballot measure changes that dynamic by putting the question in front of every registered voter in a jurisdiction, and the result carries a legitimacy that a single hearing rarely does.
The measures voted on this week took several legal forms. Charter amendments rewrite local governing rules and are difficult to reverse without another vote. Advisory questions ask residents for their opinion without binding officials. Referendums on specific sites, like the one in Janesville, add a requirement for a further public vote before work moves forward. Each form gives opponents a different level of leverage, which is part of why the headline total of 29 does not translate into 29 identical outcomes.
For companies, the distinction affects planning. A binding ban means a site is effectively closed to the use. An advisory loss means a project may still proceed legally, but at the cost of public goodwill and with a heightened chance that elected officials will hesitate to approve the necessary permits.