Lifetime Citizen Portal Access — AI Briefings, Alerts & Unlimited Follows
Oak Ridge presenters preview 'Discovery' exascale system and applications in medicine, aviation and nuclear energy
Loading...
Summary
Presenters at Oak Ridge National Laboratory outlined plans for the next-generation exascale system 'Discovery,' describing hardware partners, claims of improved performance over Frontier, and research applications including digital human twins, flight-scale simulations and AI tools for nuclear power.
Oak Ridge National Laboratory presenters on a recent briefing outlined plans for the next-generation exascale system known as "Discovery," said it is being designed to extend the capabilities of the existing Frontier system and to support research from precision medicine to aviation and nuclear energy.
Speaker 3 said the Oak Ridge Leadership Computing Facility (OLCF) — which hosts Frontier — has delivered significant advances and that “Discovery is the next evolution of the nation's exascale supercomputing capability.” The presenters repeatedly framed Discovery as building on lessons from Frontier while improving both speed and efficiency.
Speaker 4 described the project's strategic aim: “What excites me even more about discovery is that it will set a new standard for sovereign AI and for, sovereign leadership computing.” The presenters said Discovery will pursue greater throughput and parallelism, enabling researchers to run multiple large-scale problems simultaneously.
On hardware, Speaker 3 said Discovery will feature Hewlett Packard Enterprise’s new GX platform and AMD’s newest GPUs. Speaker 2 and Speaker 3 noted that the teams worked closely with HPE and AMD on Frontier and will apply those lessons to make Discovery a better scientific instrument.
Presenters illustrated intended applications. Speaker 4 sketched a long-term vision for a complete digital twin of the human that could allow clinicians to test treatments and life choices in silico to help guide physicians’ recommendations. Speaker 5 said exascale lets researchers scale turbulent-flow simulations up to flight scale, noting that “at any given time, there are about 1,000,000 people flying on GE engines,” and described using those simulations to study how fan-blade wake interacts with aircraft wings.
A private-sector participant, Speaker 6, identified as representing Atomic Canyon, a generative AI company in the nuclear power space, said the company “would not be here without Oak Ridge National Laboratory and without access to Frontier,” and argued AI can streamline regulatory work and make design, construction and operations of nuclear plants more efficient.
Throughout the briefing, presenters emphasized energy efficiency as well as raw performance. Speaker 1 said the team is not only deploying the world’s fastest systems but also "the world's most efficient computers," and Speaker 3 added that improving efficiency increases scientific productivity.
The presenters framed Discovery as a national-competitiveness and research tool rather than a finished product; many claims about future capabilities were forward-looking and presented as the project’s intended goals rather than completed outcomes. No formal approvals or funding decisions were recorded in the transcript; the session focused on technical vision and partner collaboration.
