Over the last few days, the U.S. Department of Energy (DOE) announced a couple of strategic partnerships to build no less than four powerful AI supercomputers, spread across two national laboratories. AMD and Nvidia will be powering two major U.S. government-backed AI infrastructure projects—AMD with HPE for Sovereign AI Factory supercomputers and Nvidia with Oracle for the DOE’s largest AI system yet, though Oracle will also be involved with AMD’s project as well.
AMD Will Power The DOE’s Sovereign AI Factory Supercomputer
The U.S. government selected AMD to provide the compute and networking technologies for two next-generation Sovereign AI Factory supercomputers, dubbed Lux and Discovery, which will reside at Oak Ridge National Laboratory (ORNL). Lux and Discovery are designed to expand America’s leadership in artificial intelligence (AI) and high-performance computing (HPC), and potentially “drive breakthroughs in science, energy, and national security.”
The Lux AI Supercomputer is coming first. Lux is being co-developed by ORNL, AMD, Oracle Cloud Infrastructure (OCI) and HPE and will be powered by AMD’s current Instinct MI355X GPUs, EPYC CPUs and Pensando advanced networking technologies. According to the AMD, Lux will be deployed early next year.
Discovery will come later. Discovery will be powered next-generation AMD EPYC CPUs, codenamed “Venice,” and specialized AMD Instinct MI430X GPUs. Per AMD’s release, the Instinct MI430X is a new MI400 Series accelerator engineered specifically for sovereign AI and scientific computing. ORNL is expected to take delivery of Discovery sometime in 2028, with user operations expected in 2029. Expected performance levels weren’t disclosed, but you can expect Discovery to be one of the most powerful AMD-based AI systems yet.
This deployment reinforces AMD’s position in the high-performance computing (HPC) and AI accelerator space. It also signals growing government trust in AMD’s hardware and software ecosystem for mission-critical applications. The Sovereign AI Factory could catalyze broader adoption of AMD’s Instinct accelerator platforms across defense contractors, national labs, and enterprise AI deployments, potentially expanding AMD’s data center revenue stream.
Nvidia DGX Superpod
Nvidia
Nvidia and Oracle’s DOE Partnership For Massive AI Systems
Nearly in parallel, the U.S. Department of Energy (DOE), Argonne National Laboratory, announced a partnership with Nvidia and Oracle to build its largest AI systems yet—Solstice and Equinox. This initiative will leverage Oracle Cloud Infrastructure (OCI) and be powered by Nvidia’s latest Blackwell GPUs, though as part of the partnership Oracle will also be providing the DOE with access to cloud AI computing resources that use a combination of Nvidia Hopper and Blackwell architectures. Solstice and Equinox are designed to support AI research across energy, security, and discovery science.
Solstice is the more powerful of the two systems. Solstice will feature 100,000 Nvidia Blackwell GPUs and be the largest AI supercomputer in the DOE’s lab complex; the smaller Equinox system will feature 10,000 Nvidia Blackwell GPUs. Construction at the ANL’s Argonne Leadership Computing Facility will begin immediately on Equinox, which is expected to be delivered in 2026. Nvidia didn’t say whether Blackwell GB200 Superchips or the newer GB300 Blackwell Ultra parts will power the systems.
Solstice and Equinox will connect Argonne’s experimental facilities, such as the Advanced Photon Source, and other network of scientific instruments. One of the goals of the project is to enable researchers to build agentic AI models that can reason and infer on massive amounts of data and autonomously test hypotheses across multiple scientific disciplines and areas of research.
This partnership strengthens Nvidia’s foothold in government and scientific computing, reinforcing its leadership in AI hardware and software. It also boosts Oracle’s visibility in the AI cloud market, potentially attracting enterprise and public sector clients seeking high-performance AI solutions.
These AI Systems Will Power The Next Decade Of AI Research
Both initiatives reflect a broader national strategy to secure and domestic AI infrastructure. AMD’s Sovereign AI Factory emphasizes on-prem, secure compute for sensitive applications, while Nvidia and Oracle’s DOE system highlights massively scalable AI for open scientific research.
For AMD, this is a high-profile moment to expand its influence in AI and HPC. For Nvidia, it’s a continuation of its dominance. Oracle, meanwhile, emerges as a key enabling partner for both chipmakers.