The U.S. Department of Energy (DoE) has reportedly collaborated with AMD on two new supercomputer projects, utilizing Team Red's latest AI chips to address scientific challenges.
AMD to Collaborate With the U.S. DoE To Build Out Two Cutting-Edge Supercomputers, With Record Deployment Times
Based on a new report from Reuters, it seems like AMD has managed to secure a massive partnership with the U.S. DoE, which involves the construction of two new supercomputers, mainly for academic purposes. This marks a major deal for Team Red, which is currently in pursuit of having its tech stack widely adopted by customers in the industry, providing a significant challenge to NVIDIA. The report claims that the supercomputers will utilize AMD's Instinct MI355X AI chips, as well as a new MI430 variant, which we'll discuss later.
Now, in terms of the supercomputer build-out, it is claimed that the first project, called "Lux", will come online in the next six months and will employ Team Red's latest MI355X AI chips. The system will involve partners like HP, Oracle, and Oak Ridge National Laboratory (ORNL), and the deployment of Lux comes at a record time, according to CEO Lisa Su.
The second supercomputer will be called "Discovery", which is expected to be delivered by 2028, featuring a custom Instinct MI430 AI chip variant designed for high-performance computing. Interestingly, Project Discovery has been under consideration by the DoE since last year, and it appears that AMD has been chosen as the primary computing provider.
The Department of Energy will finance projects worth almost $1 billion, and it is claimed that the DoE is also seeking further private partnerships to scale up the division's computing capabilities. If you are curious about why AMD has been adopted here, instead of NVIDIA, well, the report doesn't disclose a specific reason, but a suitable bet would be how the administration is more experienced with AMD's tech stack on the HPC-level, and since supercomputers like the Frontier have been under use by the DoE, opting for AMD makes sense this way.
Of course, DoE is expanding its collaboration with the public sector; hence, we could see NVIDIA's AI hardware being adopted as well in the future. However, by the looks of it, AMD is the preferred choice for now.
Follow Wccftech on Google to get more of our news coverage in your feeds.
