NEWS  /  Analysis

AMD Reaches $1 Billion AI Deal with U.S. Energy Department to Build Two Supercomputer

By  LiDan  Oct 28, 2025, 3:32 a.m. ET

The first supercomputer, based on AMD's MI355X AI chips, is scheduled to be completed and operational within the next six months. The second, more advanced supercomputer, based on AMD's MI430 series AI chips, is set to get ready for operation in 2029.

AsianFin -- Advanced Micro Devices Inc. (AMD) shares finished nearly 2.7% higher on Monday to close at a new record of $259.67 after the company reached a mega deal with the U.S. government to deploy its artificial intelligence (AI) chips.


AI Generated Image

AI Generated Image

AMD and the U.S. Department of Energy (DOE) has formed a $1 billion partnership to build two next-generation supercomputers aimed at unlocking scientific breakthroughs in issues such as nuclear power, national security, and cancer treatment. “Wining the AI race requires new and create partnerships that will bring together the brightest minds and industries American technology and science has to offer,” said U.S. Secretary of Energy Chris Wright in a statement.

With the new supercomunters, the U.S. government is working to ensure sufficient computing power to run increasingly complex experiments that require processing massive amounts of data. Wright said these systems will "supercharge" progress in areas such as nuclear and fusion energy, defense and national security technologies, and drug development.

Wright predicts that the adoption of these AI systems will provide a practical path to mastering fusion energy within the next two or three years. "My hope is in the next five or eight years, we will turn most cancers, many of which today are ultimate death sentences, into manageable conditions," Wright said.

Under the partnership, the first supercomputer, called Lux, is scheduled to be completed and operational within the next six months. The system will be built based on AMD's MI355X AI chips and will also include AMD central processing units (CPUs) and networking chips. The system is jointly developed by AMD, Hewlett Packard Enterprise (HPE), Oracle Corp.'s cloud infrastructure business, OCI, and the DOE's Oak Ridge National Laboratory (ORNL).

AMD CEO Lisa Su said Lux will boast the fastest deployment of a computer of this size she has ever seen. This is exactly the speed and agility that we want to achieve for America's AI efforts, she said.

ORNL director Stephen Streiffer said the Lux supercomputer will provide approximately three times the AI capacity of current supercomputers.

The second supercomputer, named Discovery, is poised to more advanced than Lux and be based on AMD's MI430 series AI chips that are optimized for high-performance computing. The system, co-designed by ORNL, HPE and AMD, is expected to be delivered in 2028 and ready for operation in 2029.

Su noted that the MI430 is a specialized variant of the MI400 series, combining key features of traditional supercomputer chips with the ability to run AI applications.

Reuters cited DOE officials that the department will host the supercomputers, while companies will provide the machines and cover the capital expenditures, and both parties will share the computing power. These two AMD chip-based supercomputers will be the first of these types of collaborations between DOE labs and private companies and laboratories across the country.

The partnership unveiled on Monday marks the DOE’s expansion of collaboration with chipmaking giants to boost the computing power in key sectors. Wright on May 29 a new contract with Dell Technologies to develop NERSC-10, the next flagship supercomputer at the National Energy Research Scientific Computing Center (NERSC), a user facility under the department at Berkeley Lab.  

The planned new supercomputer, a Dell system powered by Nvidia Corporation’s next-generation Vera Rubin platform, will be engineered to support large-scale high-performance computing (HPC) workloads like those in molecular dynamics, high-energy physics, and AI training and inference—and provide a robust environment for the workflows that make cutting-edge science possible.  

The new system, due in 2026, will be named after Jennifer Doudna, the Berkeley Lab-based biochemist who was awarded the 2020 Nobel Prize for Chemistry in recognition of her work on the gene-editing technology CRISPR.

Please sign in and then enter your comment