HPE and AMD’s Lux AI cluster ranks No. 14 on TOP500
Lux, an AI cluster developed by HPE and AMD for Oak Ridge National Laboratory, has entered the TOP500 ranking as the world’s 14th fastest system. The platform delivers 220.79 petaflops and is designed to give U.S. researchers cloud-like access to AI training and inferencing resources.
- Lux ranked No. 14 on the TOP500 list.
- The system delivers 220.79 petaflops of performance.
- Half of Lux’s annual 3.5 million node-hours will support the Genesis Mission.
- Lux includes AMD Instinct MI355X GPUs and HPE Juniper Networking QFX Switches.
- HPE made an in-kind investment of $37 million in the Genesis Mission.
A dedicated AI system for the Genesis Mission
HPE announced the TOP500 debut at SC26 on November 16, 2026. The company said Lux surpassed systems designed for traditional supercomputing workloads and is one of the first dedicated AI systems deployed to support the U.S. Department of Energy’s Genesis Mission.
The mission focuses on using AI to strengthen U.S. technological leadership and global competitiveness. According to HPE, it aims to accelerate breakthroughs in energy research, scientific discovery and national security through a platform combining AI, supercomputing, quantum systems and advanced scientific instruments.
Lux is intended to operate as a flexible, multi-tenant platform. That design allows researchers across the U.S. to scale training and inferencing workloads, rather than limiting the system to a single project or organization.
Hardware and networking built for large AI workloads
Lux is built on direct liquid-cooled HPE ProLiant Compute XD685 systems and HPE Juniper Networking data center networking solutions. Its platform also includes AMD Instinct MI355X GPUs, AMD EPYC CPUs and AMD Pensando DPUs.
HPE said the configuration supports large model AI training and deep learning, as well as integrated management and security features. Those include Silicon Root of Trust protection, which is available in all HPE ProLiant servers, according to the company.
The system uses HPE Juniper Networking QFX Switches that support 800G. HPE said advanced load balancing and congestion management are designed to reduce network bottlenecks and improve GPU utilization. HPE Networking Data Center Director provides intent-based deployment, management and lifecycle automation.
Lux is also the first deployment of HPE Networking solutions at a DOE national laboratory, bringing scalable, high-performance Ethernet to AI-driven science, according to HPE.
Research access and the wider HPE partnership
Half of Lux’s annual 3.5 million node-hours will be dedicated to the Genesis Mission, while the remaining half will be available for proprietary use. Oak Ridge said the system will provide researchers with additional capacity and flexibility for work involving Earth system prediction, materials development, biology and fusion energy.
“Lux will provide researchers with additional capacity and flexibility to advance innovative SI approaches and accelerate breakthroughs in critical areas, including Earth system prediction, materials development, biology, and fusion energy,” said Georgia Tourassi, associate director of computing and computational sciences at the Oak Ridge Leadership Computing Facility.
HPE has also made an in-kind investment of $37 million to advance the Genesis Mission and support research projects aimed at solving complex problems. The company said Lux was announced about a year ago as part of its broader collaboration with ORNL and AMD, which also includes Discovery, a forthcoming next-generation supercomputer intended to succeed Frontier.
For researchers, the announcement means another national laboratory resource focused specifically on AI workloads, with access divided between a federal mission and proprietary use. For HPE and AMD, Lux demonstrates how their hardware and networking platforms are being positioned for shared scientific AI infrastructure and the DOE’s broader Genesis Mission.


