Most of us interact with a data center constantly without ever thinking about it. The search that settles an argument at the kitchen table, the GPS that reroutes you around an accident on the way to school, the hospital system that pulls up your daughter’s allergy history in an emergency room, the app that tells you whether the prescription is ready: All of it runs through buildings full of servers somewhere. Data centers are as integral to American life as the electrical grid itself. That’s why demand for them is climbing sharply as artificial intelligence moves from novelty to infrastructure.
But therein lies a paradox: Americans don’t want these buildings near them. A March 2026 Gallup poll found that 71% of Americans oppose building an AI data center in their local area, with 48% strongly opposed. That is higher local opposition than to nuclear power plants. Among those opposed, half pointed to the water and electricity these facilities consume; roughly a fifth mentioned quality-of-life concerns such as noise and traffic; and another fifth worried these projects would make everyday costs unaffordable.
Those worries deserve a serious answer, not a lecture about progress. Women who manage the household budget have watched utility bills climb, and they are entitled to ask who is paying for the transmission lines and generating capacity that a new campus requires. The good news is that data centers are excellent neighbors when states set the terms in advance. One of the most promising terms involves land nobody else wants.
The Environmental Protection Agency (EPA) estimates there are more than 450,000 brownfields across the country, which are properties contaminated by hazardous substances from a previous industrial life. Old foundries, shuttered steel mills, defunct chemical plants, and abandoned rail yards sit idle for decades because investigation costs and legal uncertainty with liability scare off any buyer with better options. Many of these sites are already zoned industrial, already sit near heavy transmission infrastructure built for the factory that closed, already have water and rail access, and already sit at a comfortable distance from neighborhoods. EPA has issued formal guidance on redeveloping brownfield and Superfund sites as AI data centers, along with a mapping tool that helps developers find them.
Indiana is a great case study in repurposing brownfields for AI data center projects. The Indiana Brownfields Program maintains a continually updated public inventory of contaminated sites, and its Phase I Environmental Site Assessment initiative covers the cost of professional assessments for local governments and nonprofits, which removes the single largest unknown standing between a polluted parcel and a paying tenant. Regional planners in Northwest Indiana went further this past May with a framework recommending that communities catalog brownfield sites against data center specifications, adopt model development agreements, and write performance standards directly into zoning so that developers know what is expected before they arrive.
States that follow this approach will win the projects with buy-in from the community and without sacrificing farmland. Pairing brownfield redevelopment with ratepayer protection makes the deal all the better. Indiana regulators approved a settlement with Indiana Michigan Power requiring large-load customers to sign twelve-year minimum contracts, pay for at least 80% of their contracted capacity every month whether they use it or not, post substantial collateral, and cover exit fees if they leave early. Amazon, Google, Microsoft, and the state utility consumer counselor all signed it. Legislators can add to that list by tying tax abatements to job and investment benchmarks, requiring companies to fund their own substations and transmission upgrades, and mandating public disclosure of water usage.
Handled this way, a data center cleans up the worst parcel in the county, pays its own freight, and leaves everyone’s bills lower than the alternative.