Why the US Energy Department is Building AI Supercomputers

Governments are shifting from traditional tech procurement to public-private partnerships that reshape how they acquire AI computing infrastructure, a trend now accelerated by the US Department of Energy’s US$1bn alliance with Advanced Micro Devices (AMD) to create two new supercomputers for scientific and medical breakthroughs.

The purpose of the partnership

Rather than purchasing technology outright, the government partners with private industry, which supplies capital and equipment and shares computing resources with public agencies.

AMD, a major challenger to Intel and Nvidia, will supply its latest AI chips for both systems.

US Energy Secretary Chris Wright wants these systems to “supercharge” innovation in nuclear power, fusion energy, advanced defence and pharmaceuticals.

On fusion power - which replicates stellar processes by merging light atoms in superheated plasma - Chris admits it’s a formidable challenge: “We’ve made great progress, but plasmas are unstable – and we need to recreate the centre of the sun on Earth,” he explains.

“We’re going to get just massively faster progress using the computation from these AI systems that I believe will have practical pathways to harness fusion energy in the next two or three years.”

This timeline is much more optimistic than the decades-long estimates typical among fusion researchers.

The supercomputers will also simulate cancer treatments on the molecular scale.

“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,” Chris says.

About Lux and Discovery

The first system, Lux, is slated to go live within six months - a pace that AMD CEO Lisa Su calls “remarkable” for a project of such magnitude.

Lux will leverage AMD’s MI355X chips, which excel at the complex mathematical operations behind AI, alongside AMD’s processors and networking hardware.

This system is a combined effort from AMD, Hewlett Packard Enterprise, Oracle Cloud Infrastructure and Oak Ridge National Laboratory in Tennessee.

Oak Ridge director Stephen Streiffer notes Lux will triple the facility’s current AI processing power. Discovery, the second machine, is even more powerful, set to use AMD’s MI430 chips tailored for high performance and intensive computation, ready in 2028 and operational by 2029.

Lisa explains the MI430 is a hybrid of classical supercomputer processors and components specialised for AI task loads - a necessary convergence because hardware for supercomputing and AI has typically served distinct functions. Streiffer expects Discovery to significantly boost the lab’s computational capabilities.

Partnership structure

The arrangement is simple: the Department of Energy provides buildings and physical infrastructure, while AMD and its partners supply equipment and bear the capital expense.

Both parties share computing time.

According to Department of Energy officials, these two AMD-powered supercomputers are just the start, with future collaborations expected between government agencies and private tech firms.