208. The Genesis Mission: How AI Supercomputing Is About to Reshape American Science and Energy

After 22 years at IBM, where he rose to senior vice president and director of IBM Research, Dr. Dario Gil now leads one of the most ambitious science and technology initiatives in a generation. As the Department of Energy's (DOE's) Under Secretary for Science and director of the Genesis Mission, Gil is orchestrating a convergence of high-performance computing, artificial intelligence (AI), and quantum computing aimed at transforming how America does science and engineering. The Genesis Mission rests on a straightforward premise: a computing revolution is underway, and the U.S. should harness it to double the productivity of its trillion-dollar-a-year research and development engine within a decade. The initiative is built on three pillars: a platform for accelerating discovery anchored in high-performance computing, AI supercomputing, and quantum computing; a portfolio of national challenges in energy, physical sciences, and national security; and a university engagement effort to rethink how future scientists and engineers are educated in the age of AI. Gil offered fusion energy as a prime example of how AI can compress timelines. By training neural networks on validated simulation data, researchers can build surrogate models that run thousands to tens of thousands of times faster, allowing engineers to iterate on reactor designs in hours rather than months. AI is also being applied to real-time plasma control through collaborative work involving Google DeepMind and Commonwealth Fusion Systems. On the grid, Gil shared two striking examples. The DOE's Office of Electricity is developing AI agents to help developers fix deficient interconnection applications—which account for 80% to 90% of submissions—potentially accelerating studies by up to a year. Meanwhile, Brookhaven National Laboratory's Grid FM emulator can speed power flow calculations by 100x, compressing what would be 20 years of conventional analysis of the Texas transmission grid into roughly two months. Gil was candid about the tension between AI as an energy solution and AI as a source of surging electricity demand, noting that planned data centers now reach gigawatt scale. The path forward, he said, involves optimizing the existing grid, accelerating nuclear energy, investing in fusion, and driving major efficiency gains in AI hardware. New supercomputing infrastructure is already being built through the Genesis Consortium, a partnership of 27 industrial players. Argonne and Oak Ridge National Laboratories are each standing up large GPU clusters this year, with a 100,000-GPU system planned for Argonne in 2027—the largest science-oriented cluster in the world. Asked what success looks like, Gil pointed to the AlphaFold story: 50 years of work produced 200,000 protein structures, then AI predicted 200 million in two years. Success, he said, will mean 50 to 100 comparable breakthroughs across all domains of science within three to five years.