Most Americans do not want AI data centers built in their communities, according to a new Gallup survey. The poll found majority opposition to the facilities, with energy consumption, water use and the strain on local utilities driving the resistance.

The finding turns a quiet tension in the AI boom into a measurable political problem. Data centers have become the physical backbone of artificial intelligence. Every chatbot response, image generation and machine-learning training run happens in a warehouse-sized building packed with servers. Those buildings draw enormous amounts of electricity around the clock, and they need water to keep the hardware from overheating.

That makes them unwelcome neighbors in the eyes of most Americans. The Gallup results do not point to a single objection. Opposition spans three connected worries: how much power the facilities consume, how much water they use, and what they do to local utilities and ratepayers.

On energy, the concern is straightforward. A large data center can demand as much electricity as a small city, and AI workloads push that demand higher than traditional cloud computing. A new facility landing in a community forces the local grid to absorb that load. Utilities often respond by building new power plants, extending transmission lines or buying power on the open market. Those costs can drift onto residential bills.

Water is the second worry. Server cooling has traditionally relied on water, either through evaporative cooling towers or chilled-water systems. In regions facing drought or competing water demands, a data center becomes a flashpoint. Communities that spent years managing scarce water resources are not eager to hand a large share to a private company running servers.

The utility concern ties the first two together. Local power and water systems were built for residential and commercial use, not for industrial-scale computing. A data center plugging into that infrastructure raises worries about brownouts, rate increases and degraded service. The poll suggests Americans have absorbed these tradeoffs and found them unacceptable, at least when the facility is in their own backyard.

The finding is a warning for the AI industry, which has spent recent years announcing data center projects at a rapid pace. Companies need land, power and water, and they have often sought approval in smaller communities where large projects can move faster than in big cities. The Gallup results suggest that path is about to get harder. Local opposition is a known killer of infrastructure projects. Any facility that requires zoning changes, permits or utility agreements gives opponents leverage.

There is a familiar pattern here. Data centers are following the same arc that renewable energy projects traced for years: national polling showed broad support for wind and solar, but proposed installations drew fierce local resistance over sight lines, noise and land use. The public can support the idea of AI while rejecting the physical footprint required to run it.

That gap matters for policy. If opposition is broad and durable, data center development will concentrate in the few places where officials welcome it, where power is cheap and where water is available. Some states have already positioned themselves as data center destinations, offering tax breaks and streamlined permitting. The Gallup survey suggests those incentives are working against the national mood. Politicians who approve data center projects may face a backlash from constituents who feel the costs landed on them.

For utilities, the survey is a signal that customers are watching. People may accept higher bills for their own consumption, but they resent paying for someone else's data center. Utility regulators already wrestle with how to allocate grid upgrade costs when a large industrial customer arrives. The polling suggests the public will demand that data center operators bear the costs of their own demand, not the ratepayers.

The industry is not helpless here. Some data center operators have responded by building on-site solar farms, signing wind power purchase agreements and using air-cooled or recycled-water cooling systems. Those choices reduce the footprint that worries residents. But the poll indicates the default perception is negative. The burden is on the industry to demonstrate what it is doing, and to do it before asking communities for permission.

Timing also matters. The survey captures public opinion at a moment when electricity prices are a live issue for many households and water scarcity headlines are common across the West and Southwest. Those conditions make the opposition understandable.

The deeper lesson from the Gallup finding is that AI infrastructure is now a community issue, a utility issue and a political issue. Americans who use AI tools daily and expect AI to transform society still draw a line at the physical machinery humming behind their town's borders.

None of this means data centers will stop getting built. Demand for AI compute is too strong, and companies have too much invested in expansion plans. Every future project will still face a harder approval process, a more organized opposition and a more skeptical public. The industry that wants to push AI into every corner of American life now has to answer for the infrastructure in the corners where people actually live.