Genesis Mission Phase II Awards Tie AI to Fusion and Materials

What DOE announced

The Department of Energy has moved its Genesis Mission, its effort to build artificial intelligence into the core of American science and engineering, into a larger funding phase. The agency named a dozen Phase II awards, unveiled a four-year doctoral pathway under a Graduate Fellowship Pilot, and awarded $96 million to 67 early-career scientists 1.

Phase II is where the money gets serious. Phase I grants, nearly 300 of which were announced at the Genesis Mission Summit in July 2026, offer up to $750,000 over nine months to test whether an idea is feasible. Phase II projects can run up to $15 million over three years 1. The program launched with 26 science and technology challenges, including accelerating fusion energy research, materials discovery and drug development, and has since grown to 33 1.

Most Phase II winners are familiar names. They include national laboratories such as Argonne, Oak Ridge and Lawrence Berkeley, and universities including UC San Diego, MIT and the University of Illinois Urbana-Champaign. One private company made the list: Commonwealth Fusion Systems, the MIT spinout pursuing commercial fusion 2. Darío Gil, DOE's Under Secretary for Science, described the applicant pool as "an embarrassment of riches" 2.

Fusion and materials as test cases for AI-driven science

Commonwealth Fusion Systems is the only company among the Phase II winners 2, and fusion research was one of the program's founding challenge areas 1. Together, those facts show the scale of DOE's ambition. Its bet is that AI can shorten development cycles in fields where experiments are expensive and slow.

The materials and instrumentation side of Genesis is already producing results. Argonne's SYNAPS-I team showed a chat-driven agentic system on DOE beamlines. The system takes plain-language goals, segments images, reconstructs ptychography scans, and decides on its own to run higher-resolution scans when it spots regions of interest 5. A SLAC-led Phase I project aims to build a closed-loop "super intelligence" agent for catalysis research. Other Genesis selections fund robotics and embodied AI for autonomous discovery testbeds 5. These projects suggest the program is moving from prototypes toward funded efforts that connect AI agents directly to lab instruments 5.

Rebuilding the PhD

The education piece may prove more disruptive than the research awards. DOE posted a funding opportunity (DE-FOA-0003673) on October 8 offering up to $100 million for roughly 100 doctoral fellows, about $1 million per fellow. Letters of intent are due November 9 3. One analysis argues the program is often misread as a student fellowship, comparable to NSF's graduate program. In practice, DOE is paying institutions to redesign their programs, with students carried through those experiments 3. Committing to four years amounts to claiming that roughly two years of the current doctorate are slack: qualifying exams, rotations, teaching duties, and dissertations finished as funding runs out 3.

Harriet Kung, deputy director for science programs in DOE's Office of Science, framed the goal as retaining talent. She called the long PhD track a "leaky path" that loses exceptional domestic students. She said the aim is a model that could eventually train 100,000 domestic students 2. The program's "domestic" eligibility requirement is a significant choice in its own right 3, given how heavily U.S. graduate science depends on international students.

The cost side of the ledger

The official announcements leave out where the money comes from. Science reports that Genesis is drawing on DOE research budgets that were already tight, partly because of the cost of running large facilities 4. Some academics now feel they have to apply to Genesis just to keep supporting their students. Many also faced punishing timelines. For an earlier round, researchers had about six weeks after DOE announced $293 million in Genesis funding to come up with a project, find national-lab or industry partners, and clear their own university's internal review. One researcher called it "crazy" 4.

Reading the moment

The pieces fit together. DOE is concentrating money, talent pipelines and lab infrastructure around one thesis: AI-accelerated discovery is a national priority worth reorganizing for. The fusion and materials projects are the visible showcase. The four-year PhD is the long-term structural bet.

The risks follow from the same choices. Fast awards with short application windows favor groups that already have national-lab ties and the staff to write proposals quickly. A program funded partly by squeezing existing grants may shift research effort rather than add to it. If Phase II projects deliver results the agency can verify, especially in fusion, the approach will look justified. If they don't, the speed that defined Genesis will become the main criticism of it.