AMD and Hewlett Packard Enterprise (HPE) will power two new systems at the U.S. Department of Energy’s Oak Ridge National Laboratory (ORNL) — the Discovery exascale supercomputer and the Lux AI supercomputer — representing a combined $1 billion public-private investment in U.S. sovereign AI and scientific infrastructure. The systems are central to the U.S. AI Action Plan and are designed to accelerate AI-enabled science, energy innovation, and national security through open, standards-based architectures.
The Lux AI supercomputer, co-developed by AMD, HPE, ORNL, and Oracle Cloud Infrastructure (OCI), will be the first U.S. AI Factory for science. Scheduled for deployment in early 2026, Lux will be powered by AMD Instinct MI355X GPUs, AMD EPYC CPUs, and AMD Pensando networking, with direct liquid cooling for energy efficiency. The system will operate as a sovereign, multi-tenant AI cloud platform to support training and inference across DOE labs and research institutions. OCI will provide secure, high-performance cloud integration to support AI model training and national-scale data sharing.
The Discovery supercomputer — arriving in 2028 and entering user operations in 2029 — will succeed Frontier, the world’s first exascale system, also built by HPE and AMD for ORNL. Based on the new HPE Cray Supercomputing GX5000 architecture, Discovery will feature next-generation AMD EPYC “Venice” CPUs and AMD Instinct MI430X GPUs, a new MI400-series accelerator optimized for sovereign AI and scientific computing. Discovery introduces a “Bandwidth Everywhere” design for greater node and global interconnect capacity, improved memory performance, and up to 10× higher application productivity. It will also use the DAOS-based HPE Cray Supercomputing Storage Systems K3000, offering 300% more IOPS per rack than Frontier and full liquid cooling to maintain energy efficiency.
- Lux AI Factory to deploy in early 2026 as the first U.S. AI supercomputer dedicated to science
- Discovery exascale system to arrive in 2028 with AMD EPYC “Venice” CPUs and Instinct MI430X GPUs
- $1 billion investment from DOE, AMD, HPE, and partners for sovereign AI infrastructure
- HPE Cray GX5000 introduces converged AI-HPC architecture with DAOS-based K3000 storage
- Both systems built on open standards to advance the American AI Stack
“We are proud and honored to partner with the Department of Energy and Oak Ridge National Laboratory to accelerate America’s foundation for science and innovation,” said Dr. Lisa Su, chair and CEO of AMD. “Discovery and Lux will leverage AMD’s high-performance and AI computing technologies to advance the most critical U.S. research priorities in science, energy, and medicine – demonstrating the power of public-private partnership at their best.”
🌐 Analysis: The AMD–HPE collaboration for ORNL cements U.S. leadership in sovereign AI computing and exascale science. Discovery and Lux mark a strategic evolution beyond Frontier, introducing open, federated infrastructure for AI-HPC convergence. AMD’s MI400-series accelerators, paired with HPE’s GX5000 and DAOS-optimized storage, position the U.S. AI Action Plan on a foundation competitive with NVIDIA’s DGX Cloud and Intel’s Aurora architecture, underscoring an industry-wide shift toward open, sovereign AI infrastructure.
🌐 We’re tracking the latest developments in AI infrastructure and supercomputing. Follow our ongoing coverage at: https://convergedigest.com/category/ai-infrastructure/
