OpenAI's Grid Protectors starts in Georgia with 1,000 MW flexible demand

OpenAI frames the Grid Protectors Program as a national effort, but its concrete commitments so far are concentrated in Georgia. The Effingham County campus is still in development, and OpenAI says it intends for that project to serve as a pilot for the program, building on an agreement that already provides a model for future campuses. The arrangements are expected to differ by location, with additional commitments to be shared as they are finalized.

The program rests on a distinction between paying for infrastructure and shaping how much power a campus draws. Power lines, substations and other grid infrastructure are a fixed cost whether demand is low or high, and OpenAI argues that steady year-round demand from compute campuses can help spread those fixed costs across more electricity sales. The flexibility piece is meant to complement that: OpenAI says it will design facilities to reduce electricity use when needed and work with utilities to agree on how much demand can be reduced, how quickly and for how long. Depending on local needs and available equipment, it will also work with grid operators to supply power from on-site generation or storage.

OpenAI is careful to say flexibility cannot replace investment in new energy supply and transmission, and it points to long-term contracts and financial assurances as protections for other customers if a project is delayed, reduced or canceled. The Georgia Power figures give a sense of the scale the utility is attaching to this approach: it projects approximately $950 million in annual customer savings beginning in 2029 from its large-load agreements and expected growth, including Project Camellia, and estimates a benefit of at least $15 a month for a typical residential customer using 1,000 kilowatt-hours.