The U.S. Department of Energy is teaming up with AMD on two new supercomputers, tapping the company’s latest AI accelerators to push scientific research and high-performance computing forward. The projects underscore AMD’s momentum in data center AI and signal stiffer competition in a space long dominated by a single rival.
The first system, dubbed Lux, is slated to come online within the next six months and will be powered by AMD’s Instinct MI355X AI chips. Built in collaboration with partners including HP, Oracle, and Oak Ridge National Laboratory, Lux is described as arriving on a record deployment timeline, highlighting how quickly the platform is being readied for real-world research workloads.
A second system, called Discovery, is expected to be delivered by 2028. It will feature a custom variant of the Instinct MI430 designed for high-performance computing. Discovery has reportedly been under consideration since last year, and AMD has been selected as the primary computing provider for the project.
Together, the Department of Energy projects are valued at nearly $1 billion, with the agency also exploring additional private partnerships to expand its computational capabilities. While no single reason was cited for the choice of hardware, AMD’s strong track record in HPC at national labs likely played a role. Existing systems such as the Frontier supercomputer have given the Department extensive hands-on experience with AMD’s software and hardware stack, making the transition to newer Instinct accelerators a practical path. That said, the agency continues to collaborate broadly with industry and could incorporate other AI platforms in the future as needs evolve.
What this means:
– Faster time to science for academic and government researchers working on AI, climate modeling, materials science, energy systems, and other data-intensive fields
– A higher-profile showcase for AMD’s Instinct MI355X and forthcoming MI430-class technology in mission-critical environments
– A more competitive landscape for AI infrastructure, which can improve innovation, supply resilience, and cost efficiency for large-scale computing
With Lux nearing deployment and Discovery on the horizon, these systems are poised to become flagship platforms for the next wave of AI-driven discovery across the U.S. research ecosystem.
