On November 2, 2026, OpenAI launched the Grid Protectors Program, pledging to negotiate with utilities, regulators and policymakers so that future campuses it builds itself will voluntarily reduce electricity demand when the grid is under strain. The first site is a planned campus in Effingham County, Georgia, where an existing agreement already includes up to 1,000 megawatts of flexible demand response. This is not a completed energy-saving result; it is a building program in which the rules are agreed first and applied as each campus is built.
In its announcement, OpenAI cited a forecast that U.S. data center capacity will exceed 100 gigawatts by 2030. A campus needs steady power all year, while the grid is most strained only at certain times. What OpenAI wants is for that year-round demand to help spread the fixed costs of the grid, while yielding demand at peak times.
What 1,000 megawatts means in the Georgia agreement
According to OpenAI, 1,000 megawatts is roughly the average electricity demand of 700,000 households, more than the household demand of Washington, D.C. and Boston combined. The Effingham County campus is still in development; OpenAI has committed to pay the full infrastructure and electric-service costs required to serve the project, and only once the campus is operating can it reduce demand during tight periods. Georgia Power says this flexibility can reduce the amount of new generation it needs to build to serve future growth. The agreement is also linked to the utility's broader cost-reduction plan: Georgia Power projects about $950 million in annual customer savings beginning in 2029 from its portfolio of large-load agreements, at least $15 a month for a typical residential customer using 1,000 kilowatt-hours. Those figures belong to the utility's overall portfolio, not to the Effingham County campus alone.
The pledge has limits: only self-built campuses, with terms agreed site by site
The current limits of the program should be stated plainly:
- It covers campuses OpenAI develops directly; rented third-party compute is not within the default scope of this pledge.
- How much demand can be cut, how quickly and for how long must be negotiated for each campus; there is no single nationwide figure.
- Campuses may also use on-site generation, storage batteries and power controls to smooth demand around peaks, but that does not replace investment in new generation and transmission.
OpenAI says it will publish further commitments as arrangements are finalized. For communities, the real test will come after the campus starts operating: how often reduction requests arrive, how long each lasts, and how the campus's own compute work makes way. None of those details has been published yet.
