AMD just locked in a $1 billion contract with the US Department of Energy to build two cutting-edge supercomputers - Lux and Discovery - marking the government's biggest bet yet on AI infrastructure. The deal, involving Oracle and Hewlett Packard Enterprise, positions AMD as the backbone of America's first dedicated AI Factory for scientific research and national security applications.
AMD just pulled off one of the biggest government AI deals of the year. The chipmaker's $1 billion partnership with the US Department of Energy to build two supercomputers represents a massive vote of confidence in AMD's ability to challenge Nvidia's AI dominance, especially in the critical national security sector. The announcement sent AMD shares up 3% in after-hours trading as investors recognized the strategic importance of this government endorsement.
The first machine, Lux, will come online in early 2026 at Oak Ridge National Laboratory in Tennessee. What makes this different from typical supercomputers is its focus - it's being billed as America's first "AI Factory" specifically designed to "train, fine-tune, and deploy AI foundation models," according to AMD's press release. That's a direct shot at China's growing AI capabilities and signals Washington's commitment to maintaining technological superiority.
Oracle and Hewlett Packard Enterprise are joining AMD as key partners, bringing their enterprise infrastructure expertise to what's essentially a national AI training ground. The collaboration builds on AMD's previous success with the Frontier supercomputer, which held the world's fastest title until Lawrence Livermore's El Capitan system took over last year. But unlike those traditional number-crunching machines, Lux is purpose-built for the AI era.
The timing isn't coincidental. As OpenAI and other AI companies push for more compute power, the government is racing to ensure it doesn't fall behind in AI development. "Lux is designed to accelerate AI-driven science through its advanced architecture, optimized for data-intensive and model-centric workloads," the announcement states. Translation: this isn't just about raw computing power - it's about training the next generation of AI models for everything from nuclear research to cybersecurity.
Discovery, the second supercomputer due in 2029, takes a different approach with what AMD calls "Bandwidth Everywhere" architecture. The company promises it will deliver "more computing output at a similar cost" compared to existing systems like Frontier. That efficiency focus reflects the growing reality that AI workloads aren't just about peak performance - they're about sustained, cost-effective processing of massive datasets.
