The U.S. Division of Power (DOE) has unveiled two groundbreaking AI supercomputers, Discovery and Lux, that promise to propel America’s dominance in AI, high-performance computing (HPC), and scientific discovery.
The supercomputing programs, to be housed at Oak Ridge Nationwide Laboratory (ORNL), are poised to redefine the boundaries of scientific analysis, nationwide safety, and technological innovation.
By combining unprecedented computing energy with superior AI capabilities, Discovery and Lux promise to remodel how scientists mannequin advanced programs, analyse large datasets, and resolve among the nation’s most urgent technological challenges.
Powering the subsequent period of AI science
Discovery and Lux will not be simply sooner computer systems – they symbolize a brand new paradigm in how AI and science intersect.
Designed and constructed by Hewlett Packard Enterprise (HPE) and AMD, the machines will allow scientists to speed up discovery in fields from clear power to superior manufacturing, medication, and cybersecurity.
Discovery will likely be spearheaded by HPE, whereas the AMD-led Lux system will combine Oracle Cloud Infrastructure (OCI) as a part of the Lux AI Cluster, giving researchers entry to unprecedented computing energy each on-site and within the cloud.
Commenting on the landmark AI supercomputers, Secretary of Power Chris Wright stated: “Successful the AI race requires new and inventive partnerships that can convey collectively the brightest minds and industries American know-how and science has to supply.
“That’s why the Trump administration is saying the primary instance of a brand new commonsense method to computing partnerships with Lux. We’re additionally saying, as a part of a aggressive procurement course of, Discovery.
“Working with AMD and HPE, we’re bringing new capability on-line sooner than ever earlier than, turning shared innovation into nationwide energy, and proving that America leads when private-public companions construct collectively.”
Discovery: A leap past frontier
The Discovery system will mark a big improve from ORNL’s present flagship supercomputer, Frontier, which in 2022 grew to become the primary to interrupt the exascale barrier.
Discovery, scheduled for supply in 2028, will surpass Frontier in each metric of efficiency, mixing conventional supercomputing with cutting-edge AI and quantum computing analysis.
At its core lies the brand new HPE Cray Supercomputing GX5000, powered by AMD’s upcoming EPYC™ ‘Venice’ processors and Intuition™ MI430X GPUs.
The system’s next-generation structure will allow scientists to simulate, mannequin, and analyse information at file speeds.
Paired with an revolutionary Distributed Asynchronous Object Storage (DAOS)-based platform, Discovery is predicted to ship unmatched scalability and reliability for advanced scientific workloads.
Accelerating discovery throughout disciplines
With Discovery’s AI-driven computing capabilities, researchers will be capable of generate insights as soon as thought unimaginable.
The system is predicted to play a pivotal function in remodeling nuclear power by enhancing security and decreasing prices, creating AI-powered digital twins for personalised medication, and dramatically shortening aerospace design cycles from years to months.
In essence, Discovery is constructed to make ‘Gold Normal Science’ sooner, extra accessible, and extra impactful than ever earlier than.
Lux: Increasing near-term AI capability
Arriving earlier in 2026, the Lux AI Cluster will bolster the DOE’s near-term AI capability, accelerating progress in fusion power, quantum supplies, superior manufacturing, and energy grid resilience.
The system will use AMD’s Intuition™ MI355X GPUs, EPYC™ CPUs, and Pensando™ networking know-how, all housed in HPE’s ProLiant Compute XD685 framework.
Lux will present a safe and scalable basis for AI analysis throughout the nationwide laboratory community, supporting large-scale coaching and distributed inference.
It is going to additionally function a testing floor for integrating cloud-based AI options with bodily supercomputing infrastructure – a mannequin anticipated to information the DOE’s subsequent technology of computing initiatives.
Charting the way forward for supercomputing
ORNL’s legacy in high-performance computing stretches again twenty years, with leadership-class machines reminiscent of Jaguar, Titan, Summit, and Frontier – every the world’s quickest in its period.
Discovery and Lux proceed that custom, positioning the US on the forefront of world innovation in AI supercomputers and high-performance scientific computing.
By coupling world-class {hardware} with deep experience throughout nationwide laboratories and trade companions, Discovery and Lux will act as engines of innovation for many years to return.
Because the race for AI supercomputers intensifies worldwide, the DOE’s daring funding indicators a transparent message: America intends to guide it.
