Brookhaven National Lab to Build AI Foundation Model for U.S. Power Grid

Brookhaven National Laboratory (BNL) is leading a new three-year research initiative to develop an AI-based foundation model designed to modernize the U.S. electric grid. The project, known as GridFM, aims to drastically accelerate how the nation plans for increasing electricity demand by automating complex infrastructure simulations.

Funded by the U.S. Department of Energy's (DOE) Office of Electricity, the project carries a total award of $14.2 million. Approximately $3.9 million of that funding is allocated directly to BNL, with the remaining balance distributed among collaborating partner institutions. The primary technical objective is the creation of a system capable of analyzing 1 billion grid activity scenarios within a 24-hour window, a scale of simulation far beyond current traditional capabilities.

The Genesis Mission

GridFM represents the second phase of the DOE's broader "Genesis Mission," an initiative focused on applying artificial intelligence to critical national infrastructure. BNL enters this phase as a primary contributor; the laboratory has already led seven Phase I projects and participated in 29 others across various scientific domains. This established track record positions the lab to transition AI research from theoretical frameworks into operational tools for the energy sector.

Impact on Grid Reliability

As national electricity demand continues to climb, traditional grid analysis methods have become a bottleneck, often proving too slow to keep pace with the need for new energy facilities and demand sources. By leveraging a foundation model to simulate billions of outcomes daily, utility companies can identify potential failures and optimization points with unprecedented speed. Catherine Jereza, Assistant Secretary for the Office of Electricity, stated that the project would provide utility companies with better tools to plan and operate the grid more quickly, reliably, and affordably.

Future Outlook

The success of GridFM could fundamentally change the timeline for integrating new energy sources into the national infrastructure, reducing the cost and risk associated with grid expansion. While the project is now moving into its three-year development cycle, the industry will be watching to see how effectively the model's simulations translate into real-world utility deployments and whether the foundation model approach can be scaled to other critical infrastructure sectors under the Genesis Mission.