AMD and the United States Department of Energy have today revealed a US$1 billion (A$1.53 billion) joint development of two next-generation supercomputers at Oak Ridge National Laboratory on 27 October, designed to expand American dominance in AI and high-performance computing.
The Lux AI supercomputer and Discovery supercomputer represent a combined US$1 billion investment of private and public funding.
Lux will spin up in early 2026 under a new public-private partnership model, whilst Discovery arrives in 2028 through traditional procurement.
"We are proud and honoured to partner with the Department of Energy and Oak Ridge National Laboratory to accelerate America's foundation for science and innovation," AMD CEO Lisa Su said.
"Discovery and Lux will leverage our high-performance and AI computing silicon to tackle the most critical U.S. research priorities in science, energy, and medicine."
Lux will be the nation's first dedicated AI Factory for science, powered by AMD Instinct MI355X GPUs, EPYC CPUs, and Pensando advanced networking fabric.
Co-developed by ORNL, the chip giant, Oracle Cloud Infrastructure, and HPE, the system addresses immediate needs to expand DOE's machine learning capabilities in fusion, fission, materials discovery, quantum computing, advanced manufacturing, and grid modernisation.
"Winning the AI race requires new and creative partnerships that will bring together the brightest minds and industries American technology and science has to offer," U.S. Secretary of Energy Chris Wright said.
"That's why the Trump administration is announcing the first example of a new commonsense approach to computing partnerships with Lux."
The new framework accelerates deployment timeframes from years to months through co-investments from DOE and private players.
The approach allows for shared compute power and infrastructure, propelling technologies that underpin American science, energy, and industry.
Discovery will feature next-gen EPYC CPUs, codenamed "Venice", and Instinct MI430X GPUs - a fresh MI400 Series accelerator engineered specifically for sovereign AI and scientific workloads.
Built on HPE Cray Supercomputing GX5000, Discovery's "Bandwidth Everywhere" architecture allows science and machine learning applications to run efficiently with impressive memory capacity and network throughput compared to first-generation exascale rigs.
"The Discovery system will drive scientific innovation faster and farther than ever before," ORNL Director Stephen Streiffer said.
"Oak Ridge's prowess in supercomputing has transformed how researchers solve problems.
“With Discovery and Lux, we're accelerating the pace of Gold Standard Science at a scale that secures America's pole position in an increasingly competitive global environment.”
Both systems directly support the U.S. AI Action Plan by turbocharging AI-enabled science, strengthening national competitiveness, and building secure, sovereign AI infrastructure.
Discovery will catalyse breakthroughs in energy, biology, advanced materials, national security, and manufacturing innovation, helping engineer next-generation reactors, batteries, catalysts, semiconductors, and critical materials.
ORNL is expected to take delivery of Discovery in 2028, with user operations kicking off in 2029.
