U.S. DEPARTMENT OF ENERGY FORMS $1 BILLION PARTNERSHIP WITH AMD TO DEVELOP SUPERCOMPUTERS FOR AI AND RESEARCH

The U.S. Department of Energy (DOE) has announced a landmark $1 billion collaboration with Advanced Micro Devices (AMD) to design and build two next-generation supercomputers aimed at tackling some of science’s most daunting challenges—ranging from fusion energy and cancer research to national security and advanced artificial intelligence.

- 

The first system, codenamed “Lux,” is expected to come online within the next six months and will be built around AMD's latest MI355X Al-accelerated chips, coupled with high-performance CPUs and networking components from AMD's ecosystem. The second machine, known as “Discovery,” is slated for delivery in 2028 and operational readiness by 2029, leveraging the MI430 series of Al-tuned processors designed specifically for high-end computing and inference workloads.

WHAT THE PARTNERSHIP MEANS FOR U.S. COMPUTING CAPABILITIES

This collaboration signals a major upgrade to America’s scientific computing infrastructure. The DOE has explained that the pair of machines will deliver dramatically higher Al capacity than current systems—Lux alone is projected to achieve around three times the Al compute capacity of existing supercomputers according to internal lab statements. The focus spans multiple research domains: modelling fusion plasmas to advance clean energy, simulating molecular structures to accelerate drug discovery, and processing vast streams of sensor data in national defence contexts. The co-development model spreads cost, resources and access between the DOE and its industrial partners. AMD will supply hardware and accelerator technology, while Hewlett Packard Enterprise (HPE) and Oracle Cloud Infrastructure (OCI) are reported as infrastructure partners alongside DOE's Oak Ridge National Laboratory (ORNL). The combined effort underscores an emerging trend of public-private partnerships to speed innovation in high-performance computing and Al.