By Bill Pan

OpenAI said it will pay the full energy costs of future data centers it owns and operates, and give local residents a greater say in how projects are developed.

The ChatGPT developer unveiled the commitments on Nov. 5 as part of a broader policy framework that it said federal, state, and local governments should consider adopting for artificial intelligence infrastructure projects nationwide.

“Building more computing infrastructure capacity in the United States can help lower the cost of delivering AI, put these tools in more people’s hands, and ensure America stays at the forefront of the technology,” OpenAI said.

“The communities that host this infrastructure should have a clear say in shaping their futures, too.”

A central part of the framework, dubbed National Community Compact, addresses concerns that the rapid buildup of AI and cloud computing data centers could force other utility customers to shoulder the cost of new power plants and transmission lines.

OpenAI pledged to pay the full cost of the electricity its facilities consume and the infrastructure needed to serve them. It said it would provide “financial assurances” before major spending begins so other utility customers don’t end up paying for infrastructure if a project is delayed, scaled back, or canceled.

The company wants similar protections written into public policy. Its blueprint calls on state utility commissions to require large electricity users to cover the costs they add to the grid.

Data-center electricity use has been climbing rapidly as companies build increasingly powerful computing systems for AI. The Energy Information Administration estimated that servers accounted for about 7 percent of commercial-sector electricity consumption in 2025 and expects that share to rise substantially over coming decades.

On water consumption, OpenAI said each campus would have a water plan and would prioritize closed-loop systems—which recirculate a fixed volume of water in a sealed system—or low-water cooling systems.

The company also pledged to publish independent studies of noise generated by its facilities before operations begin and every year afterward.

Limits on Secrecy

OpenAI also said it will generally stop using nondisclosure agreements with public officials during negotiations over its data centers, with exceptions such as information involving security concerns and trade secrets.

The company’s policy blueprint goes further by calling for public disclosure of tax incentives offered to data-center developers.

OpenAI said governments should include measurable commitments and mechanisms to recover incentives if developers fail to deliver what they promised.

OpenAI also pledged to publish expected job numbers and work with unions, community colleges, and training providers to prepare local workers for construction and permanent positions.

Data Centers Face Voter Pushback

The announcement comes shortly after voters across the United States overwhelmingly backed tighter restrictions on data center development in the Nov. 3 midterm elections.

As of Nov. 5, at least 44 local measures restricting data center development have passed or are on track to pass, meaning developers will have to work around them or build elsewhere.

Opposition to rapid data center development also emerged in several state races, including Texas, one of the country’s largest and fastest-growing markets for AI infrastructure.

Texas Gov.-elect Gina Hinojosa, a Democrat, campaigned on halting new data center approvals until the state adopts additional protections governing electricity costs, water use, and the role of local communities.

According to Bloom Energy’s 2026 Data Center Power report, Texas is projected to exceed 40 gigawatts of data center capacity by 2028, accounting for nearly 30 percent of projected U.S. demand. The report estimates that total U.S. data center capacity could approach 150 gigawatts by then, nearly double its 2025 level.