Voters Say No to AI Data Centers, But the Midterm Map Still Has No Clear Winner

Updated September 17, 2026 1:23 am

Likely voters are sending a blunt message to Washington and the tech industry: they do not want more AI data centers built near them. A new New York Times/Siena University survey found 61% oppose construction of data centers to power AI, even as the issue remains far down the list of priorities heading into the midterm elections.

That combination matters because it shows a growing public backlash against the physical footprint of artificial intelligence, while also suggesting that neither Democrats nor Republicans have yet turned that backlash into a dependable political advantage.

What the new poll says about AI backlash

The New York Times and Siena University surveyed 1,503 likely voters in early September and found a strong majority opposed to building more AI data centers. Only 14% said they strongly support construction, underscoring how little enthusiasm exists for the infrastructure boom now accompanying the AI race.

The results line up with other recent surveys showing that public skepticism toward AI infrastructure is spreading faster than the technology’s consumer adoption. The numbers also suggest that opposition is not confined to one ideological camp. People are reacting to local concerns, environmental fears, and a broad sense that the benefits of AI are being pushed onto communities without enough public input.

Why are voters resisting data center expansion?

Voters who said they opposed new data centers were asked why, and the answers reveal a mix of practical and cultural resistance.

Environmental concerns came first. Water use and broader ecological impact were the leading reasons, cited by 32% of opponents. That is a significant signal in an era when hyperscale facilities are consuming huge amounts of electricity and, in some regions, large volumes of water for cooling.

Local community impact came next at 21%. That category likely reflects noise, land use, construction traffic, strain on power grids, and the sense that residents are asked to absorb the downside of projects whose benefits flow elsewhere.

Distrust of AI itself was also a major factor, mentioned by 18% of opponents. That is notable because it indicates the issue is not only about utilities and zoning. For many voters, data centers have become a physical symbol of broader discomfort with how fast AI is being deployed.

Regulation concerns and economic worries were less common but still meaningful, each cited by about one in ten opponents. Together, those responses point to a public that is uneasy both about how the industry is regulated and about who actually benefits from its growth.

Support is not evenly spread across the electorate

The partisan split in the survey was striking. Among voters who supported Donald Trump in 2024, the result was close to even, with 49% in favor of new data center construction and 45% opposed. By contrast, 74% of Kamala Harris voters said they opposed the projects, and 67% of respondents who did not vote in 2024 also opposed them.

That divide suggests Republicans may have more room to frame AI infrastructure as an economic development story, while Democrats are more likely to see it through the lens of environmental regulation, local control, and corporate power. But the survey also shows that skepticism remains broad enough to complicate the industry’s efforts to win over public opinion with promises of jobs and growth.

The survey’s overall picture is clear: people do not instinctively reject AI, but they are much less comfortable with the massive server farms and utility demands needed to support it.

How important is AI in the midterm race?

It is not very important yet, at least not compared with the economy, inflation, foreign policy, or opposition to President Donald Trump and Republicans. Across nearly every demographic in the survey, AI and data centers registered at less than 1% as the issue voters said mattered most in deciding their midterm vote.

There was one modest exception. Among voters aged 18 to 29, 3% picked AI and data centers as a top issue, a higher share than older groups but still a very small number relative to traditional election concerns. That suggests the public is aware of the issue, but not yet organizing its ballot choice around it.

In other words, the poll draws a sharp distinction between irritation and political salience. Many voters may dislike what they are seeing in their neighborhoods or reading about in the news, but that dislike has not yet displaced bread-and-butter concerns that dominate election behavior.

Who do voters trust on the issue?

Voters still trust their own side more than the other party to handle AI and data centers. Overall, 40% of respondents said they favored Democrats on the issue, while 42% chose Republicans and 18% said they did not know or refused to answer.

That near-even split reinforces the idea that the politics of AI infrastructure remain unsettled. Neither party has created a durable advantage, even though the issue is increasingly visible in local fights over permits, zoning, power lines, and utility commitments.

For Democrats, the challenge is to turn environmental and community concerns into a broader consumer and labor message without appearing anti-growth. For Republicans, the challenge is to champion development without ignoring the local opposition that often arises when projects are proposed in conservative or rural areas.

Why the poll matters beyond the midterms

The significance of the survey extends well beyond election messaging. It captures an emerging tension in the AI boom: the technology may be popular as a concept, but the infrastructure needed to support it is becoming politically toxic in many places.

That distinction is important for companies, utilities, governors, mayors, and county boards that are racing to attract investment. Data centers are not abstract software products. They require land, transmission capacity, water, and long-term coordination with local authorities. As the buildout spreads, so does the chance that opposition will harden into organized resistance.

Recent polling suggests this shift is not isolated. An NBC News survey in March found AI polling worse than Immigration and Customs Enforcement in favorability terms, while a Gallup poll in May reported that seven in ten Americans opposed local construction of AI data centers. The new Times/Siena results fit that pattern and add another clue: the public may be increasingly willing to accept AI as a general trend, but not the costs attached to its physical expansion.

What the numbers suggest for policy and politics

The survey points to a policy debate that is likely to intensify over the next year. State legislatures, regulators, and local planning boards may face more pressure to scrutinize water use, power demand, environmental reviews, and siting decisions. At the same time, lawmakers who want to attract AI investment will have to explain why residents should accept major infrastructure projects in exchange for promises that are often vague or long term.

That tension could produce several outcomes:

• Stricter permitting rules for large AI facilities
• More public hearings and environmental disclosure requirements
• Pushback from residents over grid strain and water access
• Partisan efforts to frame data centers as either economic assets or corporate overreach

It could also reshape how the tech industry talks about itself. Companies have spent years marketing AI through productivity gains, consumer convenience, and productivity boosts. The current backlash suggests they may need to spend more time explaining electricity consumption, cooling systems, and what communities gain in return.

How did AI become a local political issue?

AI became a local issue because its supporting infrastructure is impossible to hide. A chatbot may be invisible to most people, but the server campuses that power those tools are large, power-hungry, and increasingly concentrated in specific regions. That makes them easier to oppose than software alone.

Residents who live near proposed projects are often the first to notice the costs: more truck traffic during construction, concerns over groundwater, substations and transmission upgrades, and fears that industrial-scale development will alter the character of a town or county.

Those worries have helped turn data centers into a symbol of a larger question: who gets to decide how AI grows, and who bears the burden when it does?

How the issue could evolve from here

The immediate answer is that opposition may continue rising even if the issue remains low on the national agenda. That is a familiar pattern in American politics: a policy can be unpopular in the abstract, yet only become decisive once it is tied to a concrete local fight.

If AI infrastructure continues expanding at the pace many companies want, more voters are likely to encounter the consequences directly. That could elevate the issue in places where new facilities are proposed, especially in communities already uneasy about water scarcity, power reliability, or industrial development.

For now, the political map is ambiguous. Voters generally dislike AI data centers, but they have not yet chosen either party as a clear champion or antagonist. That leaves room for both sides to shape the story — and for the next round of local opposition to do much of the talking.

What the poll makes plain is that the AI boom is no longer just a Silicon Valley story. It is becoming a land-use story, a utility story, an environmental story, and increasingly, a campaign issue waiting for the right moment to break through.

Timeline of recent polling on AI data centers

For political strategists, the lesson is straightforward: the public may not rank AI as a top voting issue today, but it increasingly sees the infrastructure behind AI as something to resist. That is a warning sign for an industry that depends on scale, speed, and public tolerance to keep building.