Most Americans oppose AI data centers in their communities, according to a new Gallup survey. The survey found that water use, energy consumption, and cost are the main reasons residents do not want the facilities built nearby.

The finding arrives at a moment when data centers have become the physical backbone of the artificial intelligence boom. Companies are building massive computing facilities to train and run AI models, and the people who live near those facilities are raising objections that local officials cannot brush aside. The Gallup results measure how widespread that resistance has become: the opposition is the majority view, not a fringe position.

The three concerns sit at the center of the conflict, and they reinforce one another.

Water is the most direct issue. Data centers need cooling, and many facilities rely on evaporative systems that draw heavily from local water supplies. In drought-prone regions, a single facility can consume a significant share of a community's water. Residents who have watched their water tables drop over years of dry weather are not eager to hand that resource to a server farm. The Gallup survey shows the concern is not limited to the arid states where data centers have clustered; it registers across the country.

Energy is the second concern. AI workloads are power-hungry, and the rapid buildout of data centers has forced utilities to expand generation and transmission capacity quickly. That power has to come from somewhere, and communities are increasingly aware that a data center can mean higher electricity rates for everyone else on the grid. The economics compound the problem: data centers employ relatively few people once built, so the jobs argument that typically accompanies economic development carries less weight.

Cost is the third concern, and it ties the first two together. Utilities do not absorb the expense of new substations, transformers, and transmission lines. They pass it along to ratepayers. When a data center is the reason a utility upgrades its network, residents end up paying for infrastructure that mostly serves a tech company's operations. The Gallup survey suggests Americans have noticed that arrangement and do not like it.

The survey's framing matters. It measured opposition to data centers in respondents' own communities, which is a different question from whether Americans support AI development in the abstract. The distinction explains much of the current conflict. Broad curiosity about AI products does not translate into enthusiasm for a windowless concrete building humming next to a residential neighborhood.

That gap is the central tension of the AI infrastructure buildout. Companies are making siting decisions at national and global scale, choosing locations based on access to power, fiber, water, and land. The people who live at those locations make decisions at the scale of a city council meeting or a county commission hearing. Those two scales are colliding, and the Gallup numbers indicate that, for now, the local scale is winning.

The survey also lands in a policy environment where data centers have become a point of contention in state capitals. Some states that once offered tax incentives to attract the facilities are reconsidering those deals. Others are weighing new disclosure requirements around water and power use. The federal government has treated data centers as critical infrastructure for national competitiveness, which sets up a potential clash between Washington's priorities and local resistance.

The opposition itself cuts across the usual political lines. Environmental groups object to the water and emissions profile of large data centers. Conservative residents object to the cost and the strain on the electric grid. That combination makes data center siting a rare issue that can draw fire from both ends of the spectrum, and it makes the politics harder for companies to manage.

None of this means the buildout will stop. The computing demands of AI are still growing, and the companies involved have deep pockets and long construction pipelines. But the Gallup survey is a warning that expansion plans will face friction at the permitting stage, in utility rate cases, and at the ballot box.

For policymakers, the message is that public support for AI infrastructure cannot be assumed. Transparency about resource use would help, as would siting decisions that avoid the most water-stressed and grid-constrained areas. The industry has begun to talk about efficiency improvements, from more powerful chips to cooling systems that use less water, but those changes will take years to reach the scale of the current buildout. In the meantime, the companies and the communities have to negotiate the terms.

For residents, the survey is a measure of how much their concerns have broken through. A decade ago, a data center proposal might have been greeted as a clean industry win worth celebrating. The Gallup numbers show that promise no longer carries the weight it once did. Residents are now asking how much water a facility will use, how much power it will draw, and what it will cost their households.

The answers to those questions will shape where AI infrastructure gets built and where it does not. The Gallup survey is not a verdict on AI itself. It is a verdict on the terms under which Americans are willing to host its machinery, and those terms have gotten stricter.