The U.S. Department of Energy is partnering with AMD (AMD.US) to build two AI supercomputers to accelerate scientific breakthroughs in nuclear energy, cancer drugs, and more.

On Monday, U.S. Energy Corp. and Advanced Micro Devices (AMD) reached a cooperation agreement worth a total of 1 billion dollars to jointly create two new generation artificial intelligence-driven supercomputers. These supercomputers will be used to tackle large scientific challenges, including nuclear energy, cancer treatment, and national security. Energy Secretary Chris Wright and AMD CEO Su Zifeng confirmed this news to the media. Wright stated that these two new supercomputing systems will provide "strong momentum for nuclear and fusion energy, defense and national security technology, as well as drug development." The global scientific community is currently trying to replicate nuclear fusion reactions that occur inside the sun, which release enormous energy through extreme high temperature and pressure. Wright said, "We have seen substantial progress, but the plasma is extremely unstable, and we are trying to recreate the core environment of the sun on Earth." With the help of AI supercomputers to drive experimental simulations, there is hope to find a feasible path for fusion energy in the next two to three years. In addition to the energy sector, Wright emphasized that the supercomputers will also be used to manage the U.S. nuclear weapons arsenal and significantly accelerate molecular-level simulations in cancer drug development. "Our hope is that within the next five to eight years, we can transform many cancers currently considered a 'death sentence' into manageable diseases." The first supercomputer, named "Lux," is expected to be up and running within six months, constructed using AMD's latest MI355X AI accelerator card, along with AMD CPU and network chips, and jointly designed and developed by AMD, Hewlett Packard Enterprise Co., Oracle Corporation cloud infrastructure, and Oak Ridge National Laboratory (ORNL). Su Zifeng said that Lux is the fastest deployment case of its kind she has seen. "This is exactly the speed and agility we hope to bring to the U.S. AI strategy." ORNL Director Stephen Streiffer stated that Lux's AI capabilities will be three times that of existing supercomputers. The second supercomputer, "Discovery," will have even higher performance and will utilize the AMD MI430 AI chip optimized for high-performance computing. It will be designed jointly by ORNL, Hewlett Packard Enterprise Co., and AMD, with a target delivery date in 2028 and operational in 2029. The MI430 is a special version of the AMD MI400 series, integrating traditional supercomputing chip features with AI computing capabilities. Streiffer mentioned that Discovery will achieve a "significant leap in computing power," but the exact scale is still difficult to estimate. The Energy Department stated that the new cluster will be hosted by the department, with enterprises responsible for providing equipment and capital expenditures, and both parties sharing computing resources. These two systems based on AMD chips are just the beginning of the U.S. promoting a public-private collaboration model, and it is expected that there will be more similar projects with private companies and national DOE laboratories in the future. At the time of writing, AMD rose by 0.54%, Hewlett Packard Enterprise Co. increased by 1.89%, and Oracle Corporation fell by 0.13% against the market trend.