OpenAI launched its Grid Protectors Program on Monday, November 2, 2026, as a national effort to ensure its future data centers, which the company refers to as "compute campuses," can reduce electricity demand when local grids experience strain. The program's initial pilot will be at OpenAI's planned 3.2-gigawatt campus in Effingham County, Georgia, known as Project Camellia. Under a 25-year agreement approved in August, Georgia Power can curtail up to 1,000 megawatts from the site during periods of high demand. This capacity is roughly equal to the average electricity demand of 700,000 homes.
The company stated that this flexible demand response will allow the data center to proactively reduce its power consumption before residential customers are impacted during periods of high demand. OpenAI also committed to paying the full cost of the infrastructure and electric service required for the Effingham County project, ensuring that Georgia families will not subsidize the data center's electricity needs. Georgia Power anticipates that agreements with large users like OpenAI will contribute to approximately $950 million in annual customer savings starting in 2029.
The Grid Protectors Program comes amidst increasing scrutiny and public opposition to data center developments across the United States due to their substantial energy demands and perceived impact on local power grids and electricity bills. OpenAI's head of energy policy, John McCarrick, stated that the initiative reflects a shift in industry thinking about reliability, technological improvements, and feedback from communities. The company emphasizes that while flexible demand is a valuable resource, it does not replace the need for new power plants and transmission lines.
OpenAI plans to negotiate site-specific arrangements with utilities, regulators, and policymakers for each future campus. These agreements will detail the amount of power a campus can reduce, how quickly it can respond, and the duration of the reduction, with terms varying by location. Depending on local needs, some sites may also contribute power to the grid from on-site generators or batteries.
A five-day trial conducted in December 2025 by National Grid, Emerald AI, EPRI, Nebius, and NVIDIA demonstrated the feasibility of AI computing workloads reducing power use during simulated grid emergencies. The trial, which involved a London-area GPU cluster, showed power consumption cuts of up to 40% across more than 200 scenarios, with a 100% compliance rate. In one test, the facility reduced 30% of its load within 40 seconds following a simulated loss of a large generating plant. OpenAI intends to make power flexibility a core component of its future campus development and operations.