HPE Powers the Next Frontier in AI and Supercomputing with “Discovery” and “Lux” for Oak Ridge National Laboratory 

/ AI, Application Development, Infrastructure

New systems built on the Cray GX5000 architecture mark a leap forward in exascale performance, AI convergence, and sovereign computing

The News

Hewlett Packard Enterprise (HPE) has been selected by the U.S. Department of Energy’s (DOE) Oak Ridge National Laboratory (ORNL) to build two next-generation systems including Discovery, a second-generation exascale supercomputer, and Lux, a dedicated AI cluster designed to democratize access to AI compute resources across U.S. research institutions.

The announcement reinforces HPE’s role in high performance computing (HPC) and AI infrastructure, following the success of Frontier, the first exascale system ever built. Discovery will use HPE’s new Cray Supercomputing GX5000 platform with unified AI and HPC architecture, while Lux will provide a multi-tenant, cloud-like environment for AI training and inference powered by AMD Instinct MI355X GPUs and EPYC CPUs.

Both systems are designed to advance U.S. capabilities in science, energy, and national security, while driving breakthroughs in precision medicine, climate modeling, quantum research, and aerospace engineering.

Analyst Take

The introduction of Discovery represents more than a generational upgrade from Frontier. It signals the evolution from exascale performance to AI-scale computing. Built on the HPE Cray Supercomputing GX5000, Discovery introduces a converged platform optimized for AI, HPC, and quantum workloads, uniting traditionally siloed compute environments under a single architecture.

This unified approach is critical as modern research workflows increasingly rely on hybrid compute paradigms. For example, AI-assisted modeling now complements physics-based simulations in materials science, while quantum-classical workflows are emerging in molecular research. By designing Discovery around convergence, HPE is building the infrastructure substrate for the next decade of computational science.

The claimed technical innovations in Discovery are substantial:

  • 25% smaller footprint per rack, enhancing data center density
  • 300% increase in IOPS via the new Cray Supercomputing Storage Systems K3000 with DAOS integration, providing object storage performance for HPC
  • Next-gen HPE Slingshot interconnect, delivering ultra-low-latency communication for hybrid AI/HPC workloads
  • Full liquid cooling using HPE’s five decades of thermal innovation, driving improved energy efficiency and operational cost reduction

Discovery could enable “a new level of converged HPC, AI, and quantum computing capabilities,” according to Bronson Messer, ORNL’s Director of Science, signaling a potential shift in research productivity and computational power.

Lux: A Sovereign AI Factory for National Innovation

Where Discovery pushes the boundaries of exascale science, Lux addresses the nation’s growing demand for AI compute accessibility. Built on HPE ProLiant XD685 systems with direct liquid cooling and powered by AMD Instinct GPUs, Lux acts as a sovereign AI factory, a term increasingly used to describe self-contained, secure AI infrastructure supporting domestic research and innovation.

The system’s cloud-like, multi-tenant design may allow U.S. researchers to train, deploy, and iterate AI models in a secure, scalable environment. By providing centralized AI resources, Lux aims to reduce the reliance on commercial cloud platforms for sensitive or mission-critical workloads.

This shift toward sovereign AI infrastructure mirrors global trends in national technology policy, where control over compute resources is becoming as important as access to data itself. Lux exemplifies how public-private partnerships can balance innovation with sovereignty, ensuring AI advancement remains aligned with public interests and national goals.

HPE and AMD: Partnership at the Core of Performance Innovation

The collaboration between HPE and AMD continues to define the state of HPC and AI performance. The inclusion of next-generation AMD EPYC “Venice” CPUs and Instinct MI430X GPUs in Discovery underscores the companies’ shared focus on heterogeneous computing, combining multiple processor types for maximum flexibility and performance.

This partnership also strengthens AMD’s role as a cornerstone provider in both supercomputing and AI acceleration, competing directly with NVIDIA’s data center dominance.

Convergence as a Strategic Imperative

From a market perspective, HPE’s move reflects a broader industry convergence between HPC and AI infrastructure. According to theCUBE Research, over 72% of enterprises view AI-HPC integration as a top strategic priority for research and innovation by 2027. The line between modeling, simulation, and inference continues to blur, requiring compute systems that can handle both deterministic and probabilistic workloads seamlessly.

By delivering Discovery and Lux as complementary systems (one exascale, one AI-native), HPE is working to bridge government, academia, and industry needs. This combination of AI and HPC expertise could become a template for future national labs and global research centers seeking to modernize compute environments for next-gen science.

Looking Ahead

Discovery and Lux together mark a new epoch in computational capability, one defined not just by speed, but by synergy. As the U.S. seeks to maintain global leadership in scientific computing, HPE’s design philosophy (merging AI, HPC, and quantum under a unified architecture) could shape how national research infrastructure evolves.

Over the next few years, expect Discovery to become a proving ground for AI-assisted scientific discovery, while Lux operationalizes AI democratization within sovereign boundaries. Together, they embody a future where AI fuels science and supercomputing powers innovation, forming a feedback loop that continuously accelerates discovery.

Key Takeaways

  • HPE builds Discovery and Lux for ORNL. Two new DOE-backed systems combine exascale performance with sovereign AI infrastructure.
  • Discovery unifies AI, HPC, and quantum. The Cray GX5000 platform introduces converged architecture and record-breaking storage performance.
  • Lux democratizes AI access. A multi-tenant cluster brings sovereign, cloud-like AI resources to researchers nationwide.
  • AMD partnership deepens. Next-gen EPYC and Instinct chips underpin both systems, reinforcing heterogeneous compute capabilities.
  • Convergence drives strategy. HPE’s integration of AI and HPC places it at the forefront of national-scale innovation infrastructure.

Author

  • Paul Nashawaty

    Paul Nashawaty, Practice Leader and Lead Principal Analyst, specializes in application modernization across build, release and operations. With a wealth of expertise in digital transformation initiatives spanning front-end and back-end systems, he also possesses comprehensive knowledge of the underlying infrastructure ecosystem crucial for supporting modernization endeavors. With over 25 years of experience, Paul has a proven track record in implementing effective go-to-market strategies, including the identification of new market channels, the growth and cultivation of partner ecosystems, and the successful execution of strategic plans resulting in positive business outcomes for his clients.

    View all posts