Gallup: Americans Oppose Local AI Data Centers More Than Nuclear Plants

Illustration of cats outside a fenced cooling tower.

A building full of AI servers has managed to attract more local opposition than a nuclear power plant. Somewhere, a cooling tower is feeling unexpectedly popular.

That is the striking comparison in Gallup’s May 13 report: 71% of Americans oppose an AI data center in their area, compared with 53% opposing a nuclear energy plant. It is a substantial gap, although neither proposal receives majority approval.

What the Poll Asked

The published questionnaire and methodology describe telephone interviews with 1,000 adults across all 50 states and Washington, D.C., conducted March 2–18. The overall sampling margin is four percentage points at the 95% confidence level.

Respondents were asked separately about construction in their area. They were not offered a forced choice between a reactor and a server facility next door. Support was 27% for the AI data center and 45% for the nuclear plant; strong opposition was 48% and 34%, respectively.

So the headline-worthy finding is stronger resistance to local AI infrastructure. It does not mean a majority is enthusiastically requesting a nuclear neighbor. There is a difference between winning a popularity contest and merely losing it less badly.

Resources, Jobs and the Right Denominator

Gallup examined reasons in a separate April web survey. Half of data-center opponents mentioned resource use. Other concerns included pollution, quality of life and economic costs. Among supporters, 55% specifically mentioned jobs; that number describes reasons within the supporting group, not the share of all Americans supporting construction.

Those answers suggest that the debate reaches well beyond whether somebody enjoys a chatbot. Residents can like an online service and still ask difficult questions about the land, water and electricity needed to run it.

Advertisement

A cat would understand the practical objection. A radiator the size of a shopping mall sounds magnificent until somebody explains that you are not allowed to sleep on it.

The Electricity Demand Is Real; a Bill Still Needs Explaining

Berkeley Lab’s account of its 2024 energy report estimates that U.S. data centers used 176 terawatt-hours of electricity in 2023. Its scenarios put annual consumption at 325–580 TWh in 2028. That covers data centers overall, and the future range is a projection rather than a measured outcome.

These figures describe energy consumed over time. A gigawatt figure describes power. Keeping the units and the time period straight matters when comparing forecasts; an impressive number with the wrong label explains very little.

Large new loads make grid planning a serious local issue. But a household’s higher winter bill, by itself, cannot identify a nearby data center as the cause. A useful investigation would separate the rate charged from the electricity consumed, then examine the utility’s costs and rules for allocating infrastructure expenses.

That is a more demanding conversation than pointing at a warehouse. It is also the conversation that can establish who is actually paying for what.

Maine Shows Why the Details Matter

The political response is already taking shape, but proposed restrictions should not be confused with laws in force. On April 24, Governor Janet Mills vetoed Maine’s LD 307, the proposed temporary restriction on large data centers.

Mills said she supported a temporary pause but wanted an exemption for a redevelopment project in Jay. Her stated reasoning weighed the town’s hoped-for jobs and investment against the broader policy. Whatever one thinks of that decision, it is more specific than saying the state simply banned new construction.

The example illustrates the tension: an infrastructure project can be an unwelcome demand on resources to one community and an economic opportunity to another. Neither possibility excuses vague promises. The proposed benefits and obligations should be explicit enough for residents to scrutinize.

The Cloud Has a Street Address

Our guide to video streaming and content delivery follows the machinery behind an ordinary online activity. AI has brought another reason to look at the physical systems behind the screen.

For a local project, I would want concrete answers about the water source, power supply, noise, permanent jobs and infrastructure payments. A national popularity figure cannot settle those questions for a particular site, and a glossy promise about the future cannot replace them.

The lesson for developers is straightforward: explain the neighborhood consequences with the same enthusiasm used to explain the technology. The cloud may sound weightless, but the planning meeting takes place on the ground.

🐾 Curiosity looks good on you. Explore the Goodies, or find our illustrated books on Amazon.

Share this story

Leave a Reply

Your email address will not be published. Required fields are marked *