The US Department of Energy (DOE) and AMD are teaming up to build two new advanced AI supercomputers at Oak Ridge National Laboratory (ORNL). This partnership marks a big step forward in scientific, energy, and national security research. With a combined investment of over $1 billion from both public and private sources, these supercomputers will form part of a secure national network designed specifically to support AI development using US-built infrastructure.

Introducing Lux AI: The Future of AI Research

Set to go online in early 2026, Lux AI will be the first of its kind in the US—a dedicated “AI Factory” for training and deploying cutting-edge AI models. This facility is a collaborative effort involving ORNL, AMD, Oracle Cloud Infrastructure, and Hewlett Packard Enterprise. Lux AI will harness AMD’s high-performance computing and AI technology to handle data-intensive tasks, accelerating research across areas like energy, materials science, and medicine.

The architecture of Lux AI is designed to support multiple research groups simultaneously while maintaining strict data security. This approach allows different teams to collaborate without risking data leaks or breaches. It reflects a broader trend among large organizations to adopt secure, shared AI infrastructure that fosters innovation while protecting sensitive information. The project highlights how coordinated efforts can align national goals in technology, energy efficiency, and data governance.

Discovery: The Next Generation Supercomputer

The second supercomputer, called Discovery, will become operational in 2028. It will serve as the DOE’s flagship at Oak Ridge, using AMD’s latest “Venice” EPYC processors and MI430X GPUs. These components are specially designed to optimize AI workloads and scientific simulations. Discovery’s “Bandwidth Everywhere” design enhances memory and network performance without increasing power consumption, enabling faster data processing and more complex modeling.

This system builds on lessons learned from Frontier, the world’s first exascale supercomputer. It’s designed for compatibility, so existing applications can transition smoothly to the new platform. The upgrades will allow scientists to run more detailed simulations and analyze larger data sets, pushing forward the frontiers of science and technology. According to U.S. Energy Secretary Chris Wright, this project exemplifies a “commonsense approach to computing partnerships” that strengthens the country through shared innovation and expertise.

Overall, the collaboration between AMD and the DOE showcases how strategic partnerships can drive technological progress. By investing in these supercomputers, the US aims to lead in AI research, energy efficiency, and scientific discovery. The new systems will support a secure, standards-based computing network built entirely within the US, reinforcing national leadership in high-performance computing.

Both projects emphasize the role of public-private collaboration in accelerating scientific breakthroughs. AMD’s CEO, Dr. Lisa Su, expressed pride in working with the DOE and ORNL to bolster America’s foundation in science and innovation. As these supercomputers come online, they are expected to significantly enhance the country’s capabilities in AI, energy, and security, positioning the US at the forefront of technological advancement.

Inspired by

• https://www.artificialintelligence-news.com/news/why-amd-work-with-the-doe-matters-for-enterprise-ai-strategy/