AMD has officially entered into a monumental $1 billion collaboration with the United States Department of Energy to conceive and construct two revolutionary artificial intelligence supercomputers, aptly named Lux and Discovery. These cutting-edge systems are being developed in partnership with industry leaders Oracle and Hewlett Packard Enterprise (HPE) and will be installed at the iconic Oak Ridge National Laboratory (ORNL) located in Oak Ridge, Tennessee. This partnership stands as a testament to the Department of Energy’s ongoing commitment to advancing the frontiers of computational research, as well as AMD’s growing influence in high-performance computing and AI-driven innovation.
The two new systems, Lux and Discovery, represent a major evolution in supercomputing design and purpose. Lux, the first of the pair to debut, is scheduled to become operational in early 2026, marking the beginning of a new era in AI-centric scientific exploration. Discovery, following a few years later in 2029, will build upon the capabilities and architectural breakthroughs pioneered by Lux, offering a more advanced platform optimized for both energy efficiency and computational throughput. Both machines will draw upon the technological foundations established by previous record-breaking systems such as Frontier, the ORNL-based supercomputer that held the title of the world’s fastest machine until the debut of El Capitan at Lawrence Livermore National Laboratory. AMD’s role in those earlier ventures demonstrates its longstanding partnership with the U.S. government in pursuing the next generation of exascale computing.
According to a detailed press release unveiling the initiative, Lux will not simply be another supercomputer—it will function as the nation’s first dedicated “AI Factory.” This term reflects a new paradigm in research infrastructure: a purpose-built environment designed explicitly to train, refine, and deploy large-scale AI foundation models. These models will directly support groundbreaking scientific efforts in domains such as energy production, materials science, and national security. Through a computational architecture specifically optimized for data-heavy and model-intensive workloads, Lux aims to accelerate AI-driven research by dramatically shortening the time from hypothesis to result, empowering scientists to pursue discoveries that would previously have required months—or even years—of computational effort.
Meanwhile, Discovery will take this model a step further through a design principle described as “Bandwidth Everywhere,” which focuses on increasing computational throughput and data transfer performance while maintaining stringent efficiency standards. This architectural philosophy ensures that Discovery will not only surpass the capabilities of its predecessors but will also deliver exceptional computing power within a comparable energy and cost footprint. The Department of Energy envisions Discovery becoming an essential national asset—driving progress in areas like advanced materials research, clean energy development, manufacturing innovation, and biological discovery. It will support projects as diverse as the design of next-generation nuclear reactors, the improvement of high-capacity batteries, and the creation of advanced catalysts and semiconductor materials critical to technological independence.
As AMD continues to deepen its partnerships with governmental and private-sector institutions, collaborations like Lux and Discovery illustrate the company’s dedication to pushing the limits of technological possibility. With the Oak Ridge National Laboratory once again serving as a hub for these supercomputing milestones, this initiative reaffirms the United States’ leadership in computational science and AI infrastructure. The convergence of AMD’s advanced chip technology, Oracle’s cloud expertise, and HPE’s enterprise solutions is poised to redefine what large-scale computing can achieve, ushering in a future where innovation is fueled by data, powered by artificial intelligence, and sustained by cutting-edge engineering.
Sourse: https://www.theverge.com/news/807483/amd-department-of-energy-announce-1-billion-ai-supercomputer-partnership
