The U.S. Department of Energy (DOE) has joined forces with AMD to develop two cutting-edge AI supercomputers at Oak Ridge National Laboratory (ORNL). This initiative is a cornerstone of a broader national AI agenda aimed at propelling advancements in scientific research, energy innovation, and national security, while reinforcing America’s leadership in high-performance computing (HPC).
With an investment nearing $1 billion from both public and private sectors, these two supercomputers will integrate into a secure, standards-based national computing framework engineered within the United States. This collaboration exemplifies how a unified AI strategy can harmonize objectives across innovation, energy efficiency, and data governance.
Lux AI: Pioneering AI Model Training for Scientific Breakthroughs
Scheduled to become operational by early 2026, Lux AI will be America’s inaugural “AI Factory”-a dedicated platform designed to train and deploy sophisticated AI models targeting challenges in science, energy, and security domains. This system is a joint effort involving ORNL, AMD, Oracle Cloud Infrastructure, and Hewlett Packard Enterprise.
Lux AI’s architecture incorporates AMD Instinct MI355X GPUs, EPYC processors, and Pensando networking technology to manage intensive AI workloads. Its design facilitates accelerated research in critical fields such as renewable energy systems, advanced materials, and healthcare innovations. Importantly, Lux supports multi-tenant collaboration with strict data isolation, reflecting emerging best practices in managing sensitive AI operations within large organizations.
Discovery: The Next Leap in America’s Supercomputing and AI Capabilities
Following Lux, the Discovery supercomputer is slated for deployment in 2028 as DOE’s flagship HPC system at Oak Ridge. It will harness AMD’s forthcoming “Venice” EPYC CPUs and MI430X GPUs, engineered specifically for AI and scientific computing demands.
Discovery introduces a “Bandwidth Everywhere” architecture that enhances memory and network throughput without increasing power consumption. This innovation enables the system to process vast datasets and execute complex AI models efficiently, addressing the energy constraints faced by modern data centers.
Building on insights gained from Frontier-the world’s first exascale supercomputer-Discovery ensures seamless migration of existing applications, accelerating adoption and maximizing research productivity.
U.S. Energy Secretary Chris Wright emphasized the importance of inventive partnerships, stating, “Success in the AI arena depends on uniting the brightest minds and industries in American science and technology.” He highlighted that these new systems embody a pragmatic approach to collaborative computing that fortifies national innovation.
ORNL Director Stephen Streiffer remarked that Discovery will “propel scientific discovery at unprecedented speed and scale,” underscoring the transformative potential of integrating HPC with AI to shorten the path from research questions to practical solutions.
Collaborative Ecosystem: Driving Sustainable AI Innovation
Key industry players such as AMD, Hewlett Packard Enterprise (HPE), and Oracle are instrumental in the design, construction, and ongoing support of these supercomputers. Antonio Neri, HPE’s president and CEO, noted that this partnership will enable Oak Ridge to achieve “unmatched productivity and scalability.” Meanwhile, Oracle’s Executive Vice President Mahesh Thiagarajan highlighted their role in delivering “sovereign, high-performance AI infrastructure” critical to the Lux AI cluster’s co-development.
Once operational, Lux and Discovery will empower DOE researchers to run expansive AI models that deepen understanding in energy systems, biological sciences, materials engineering, and national defense. Discovery, in particular, will accelerate innovation in next-generation battery technologies, nuclear reactors, semiconductor design, and critical material development.
Implications for Enterprise Leaders: Embracing AI and HPC Synergy
For business and technology leaders, these initiatives illustrate how integrating AI strategies with HPC infrastructure can dramatically enhance research velocity, operational efficiency, and data security. Crucially, they demonstrate that substantial performance improvements can be achieved without proportionally increasing energy consumption-a vital consideration amid growing sustainability concerns.
The DOE’s collaborative model, uniting government and industry expertise, offers a blueprint for private enterprises aiming to build shared, secure AI infrastructure while retaining control over sensitive data. As AI workloads expand, organizations will need to strike a balance between computational power, energy efficiency, and governance frameworks.
The Lux and Discovery projects exemplify this balance in action: open, cooperative platforms designed to accelerate discovery at scale. They serve as a compelling example of how a visionary AI strategy can transform infrastructure investments into enduring competitive advantages.
