At the SCSP AI+ Expo, U.S. Energy Secretary Chris Wright and NVIDIA Vice President Ian Buck announced that the Department of Energy’s Genesis Mission will combine large AI models and national — lab datasets with purpose — built supercomputers at Argonne National Laboratory to speed energy research and scientific discovery. They presented Genesis as a national effort to embed AI directly into scientific workflows, arguing the initiative matters because it seeks to translate model scale and compute into faster experiments and analysis across DOE laboratories.
Genesis, the DOE program to bring AI into scientific workflows, counts NVIDIA among its partners. Buck said, “NVIDIA is 100% committed and invested in Genesis,” and described an open-source NVIDIA model trained on 1.5 million physics papers and fine-tuned on 100,000 fusion papers to produce a specialized AI agent researchers can query to accelerate experiments and interpretation.
The technical centerpiece is two Argonne AI supercomputers. Equinox is being stood up now with 10,000 NVIDIA Grace Blackwell GPUs. A planned second system, Solstice, is intended to use 100,000 next-generation GPUs paired with NVIDIA’s Vera Rubin architecture. Buck quantified the scale, saying 100,000 of the next-generation GPUs would deliver about 5,000 exaflops—“five times larger than the entire TOP500 supercomputer list combined.
The partnership pairs the DOE’s 17 national laboratories, their scientists and national datasets with NVIDIA’s full stack — chips, algorithms, software and deployment methods — so lab-scale science can run on the same building blocks used in major AI labs. Buck stressed the goal of broad access, saying the technology and software should be “for all of world science to go get access to.” NVIDIA leaders will also take part in several SCSP panels on workforce, simulation, infrastructure and quantum leadership, signaling a broader ecosystem push.
Wright placed Genesis in the context of U.S. energy production trends, noting that over the last 20 years the country tripled oil production and doubled natural gas output while seeing little growth in electricity generation — the energy most critical for AI. He said the DOE is re-emphasizing natural gas, nuclear and coal as part of a near-term strategy, and added that three small modular reactors will go critical by July 4 of this year. The department has also established a strategic fusion office to leverage AI-accelerated science.
For builders and researchers, the concrete implications are access and scale: shared supercomputers at Argonne, open models trained on domain literature, and integrated software stacks aim to shorten research cycles in fusion, materials and other energy domains. Wright warned, however, that the pace of AI-driven science will be constrained by grid growth unless regulatory and bureaucratic barriers are addressed, making the coupling of compute scale and electricity policy a central practical challenge for deploying AI at scientific scale.
Sources
Replies (0)
No replies in this topic yet.
