The US Department of Energy has awarded funding to 12 Phase II projects under its Genesis Mission, launched in November 2025 to train AI agents and scientific foundation models on DOE supercomputers, cloud AI resources and federal computing. The department is providing $159 million across the 12 projects, bringing the total number of Phase II selections to 14, with six Phase I projects expected to follow.
The projects span fusion energy, through a Commonwealth Fusion Systems digital twin for the SPARC device, RNA structure work led by UC San Diego, lattice quantum chromodynamics at MIT, geothermal reservoir modeling at UC Irvine, rare-earth separations at the University of Illinois Urbana-Champaign, extreme-environment chip design at Fermilab and quantum computing error correction work led by Harvard. Oak Ridge National Laboratory is providing supercomputing capacity to six of the projects.
Cloud providers are adding capacity alongside the federal machines. Amazon Web Services has pledged up to $50 million in credits over three years, and Crusoe Cloud has committed $50 million in AI compute credits. Microsoft joined in July with $40 million in Azure credits and $20 million in engineering services, and Google offered $40 million in cloud credits the same month.
Under secretary for science Dario Gil described the awards as the step that takes demonstrated Phase II work to the next level, combining national laboratories, universities and industry partners. For the infrastructure market, the significance is structural: a federally anchored buyer of AI compute and scientific software work, funded in part by credits from the same cloud firms selling commercial capacity.