DOE Awards $68 Million for AI Foundation Models in Scientific Research

The Department of Energy awarded $68 million across 43 grants to 11 multi-institution research teams building artificial intelligence foundation models for scientific discovery — an investment that pushes federal AI research beyond chatbots and into the physics lab.

What the Research Covers

Selected through competitive peer review under DOE's Funding Opportunity Announcement for Advancements in Artificial Intelligence for Science, the projects span three years with $20 million released in the first year and subsequent funding contingent on congressional appropriations.

The research falls into three categories:

• Foundation models at scale: Teams are studying how large scientific AI models improve as they grow in size and complexity, mirroring the scaling laws that transformed commercial AI but applied to physics, chemistry, and materials science.
• Privacy-preserving training: Projects developing techniques to train models on data distributed across multiple institutions without exposing sensitive research information — critical for biomedical and national security applications.
• Energy-efficient AI: Development of algorithms optimized for next-generation microtechnologies, addressing the growing electricity demands of AI compute infrastructure.

Argonne National Laboratory and Oak Ridge National Laboratory are among the national labs anchoring funded projects.

What AI Researchers Should Watch For

These awards sit within DOE's broader AI-for-science push, which includes the $320 million Genesis Mission funding the American Science Cloud and Transformational AI Models Consortium. The DOE Office of Science's ASCR program is the pipeline for follow-on solicitations as the program scales.

University researchers and national lab affiliates working at the intersection of AI and physical sciences are the primary audience. The emphasis on privacy-preserving and energy-efficient approaches also signals DOE's priorities for future funding rounds — proposals that address these constraints will have a structural advantage.

For researchers tracking federal AI funding opportunities across DOE, NSF, and NIH, in-depth analysis of the evolving landscape is available on the Granted blog.