AMD and the US Department of Energy (DOE) have announced two new supercomputers to be built at Oak Ridge National Laboratory (ORNL), designed to drive breakthroughs in artificial intelligence (AI), science, and national security. The new systems—Lux and Discovery—will form the backbone of America’s AI and high-performance computing (HPC) ambitions, supporting the US AI Action Plan by advancing secure, sovereign AI infrastructure and strengthening national competitiveness.
The Lux AI supercomputer and the Discovery supercomputer will be DOE flagship systems that combine public and private investment worth about US$1 billion. They aim to enhance the nation’s ability to conduct large-scale AI-enabled research and tackle complex challenges in areas such as energy, medicine, materials, and national security.
AMD chair and CEO Dr Lisa Su said the partnership represented a major step forward in national innovation. “Discovery and Lux will leverage AMD’s high-performance and AI computing technologies to advance the most critical US research priorities in science, energy, and medicine – demonstrating the power of public-private partnership at their best,” she said.
The Lux AI supercomputer, scheduled for deployment in early 2026, will be the first dedicated AI factory for science in the United States. Co-developed by ORNL, AMD, Oracle Cloud Infrastructure (OCI), and Hewlett Packard Enterprise (HPE), Lux will use AMD Instinct MI355X GPUs, AMD EPYC CPUs, and AMD Pensando networking technologies. The system is purpose-built to train, fine-tune, and deploy AI foundation models that accelerate scientific discovery and innovation.
Lux is designed to boost AI-driven research by handling data-intensive workloads with exceptional efficiency. It will support scientists and researchers across disciplines such as advanced manufacturing, energy research, and healthcare. The system will also strengthen US leadership in AI for science, giving the country a decisive advantage in developing frontier capabilities within a secure, domestic computing environment.
Oracle Cloud Infrastructure Executive Vice President Mahesh Thiagarajan said, “Oracle will deliver sovereign, high-performance AI infrastructure that will support the co-development of the Lux AI cluster,” underscoring Oracle’s commitment to advancing scientific AI capabilities.
Discovery: Advancing the next leap in scientific computing
Discovery, ORNL’s next-generation supercomputer, will further deepen collaboration between AMD, DOE, HPE, and the laboratory. It will use AMD’s upcoming EPYC CPUs, codenamed “Venice,” and the new AMD Instinct MI430X GPUs—part of the MI400 Series designed for sovereign AI and scientific computing. These technologies will allow Discovery to train and deploy AI models on systems built entirely in the US, ensuring data protection and supporting national competitiveness.
US Secretary of Energy Chris Wright described Discovery as part of a new era of computing partnerships. “Winning the AI race requires new and creative partnerships that bring together the brightest minds and industries American technology and science have to offer,” he said. “Working with AMD and HPE, we’re bringing new capacity online faster than ever before, turning shared innovation into national strength.”
Set to arrive in 2028 with user operations expected in 2029, Discovery will succeed Frontier, currently the world’s first exascale system. Built with HPE’s Cray Supercomputing GX5000 platform, Discovery will offer higher performance, energy efficiency, and AI capabilities while maintaining similar power costs. It will feature a “Bandwidth Everywhere” design to deliver greater memory capacity and global network bandwidth, improving the performance of both AI and scientific applications.
ORNL Director Stephen Streiffer noted that Discovery would dramatically expand the lab’s computing capabilities. “The integration of high-performance computing and AI promises breakthroughs at the accelerated speed and scale necessary for continued US leadership in an increasingly competitive global environment,” he said.
HPE President and CEO Antonio Neri added, “Our partnership with AMD is advancing the next generation of supercomputing in the AI era. This collaboration also reinforces American leadership in applying AI to science, energy, and national security.”
Once operational, Discovery will anchor the American AI Stack—a framework designed to support secure, open, and sovereign AI systems. It will be built on open-source software and open standards to ensure seamless scalability and accessibility for researchers.
Both Lux and Discovery represent key milestones in the US government’s effort to maintain leadership in AI and HPC, creating a foundation for innovation that balances national security, energy efficiency, and global competitiveness.
