DOE and AMD launch $1B supercomputing effort at ORNL: Lux AI (2026) and Discovery (2028)
Advanced Micro Devices Inc. (AMD) announced a $1 billion partnership with the U.S. Department of Energy to design two new supercomputers at Oak Ridge National Laboratory (ORNL): Lux AI and Discovery. The systems are intended to support the national AI Action Plan by accelerating AI-enabled science, strengthening U.S. competitiveness, and building secure, sovereign AI infrastructure.
Lux AI is slated for early 2026. Discovery is scheduled for 2028. Development involves AMD, ORNL, Oracle Cloud Infrastructure, and Hewlett Packard Enterprise, with HPE also teaming on Discovery.
Leadership emphasized that these systems will advance priority research across science, energy, and medicine-showing public-private collaboration at its best.
Why this matters for government
- National capacity for AI and simulation: Expanded compute for climate modeling, grid reliability, advanced materials, biosurveillance, and national security use cases.
- Sovereign AI infrastructure: Emphasis on domestic facilities, data residency, and security aligns with agency directives on zero trust, privacy, and supply-chain assurance.
- Hybrid footprint: Collaboration with Oracle Cloud Infrastructure and HPE signals tighter lab-cloud integration. Agencies should expect hybrid workflows and plan for data movement, egress policies, and auditability.
- Shared national asset: Access will likely flow through competitive allocation programs. Budget planning should account for proposal development, data preparation, and staff time for joint work with ORNL.
- Interagency coordination: Results and tooling from Lux AI can inform standards, best practices, and shared services ahead of Discovery's arrival.
Actions to take this fiscal year
- Prioritize candidate workloads: Identify high-impact models and simulations that benefit from large-scale compute. Prepare sanitized datasets and clear success metrics.
- Tighten data governance: Define classification, residency, lineage, and retention policies suitable for lab environments and hybrid architectures.
- Build skills early: Upskill teams in AI engineering, HPC job schedulers, security controls, and evaluation methods. For structured options, see AI courses by job role.
- Engage with ORNL: Track calls for proposals, user program briefings, and technical overviews to position your agency's workloads. Start here: Oak Ridge National Laboratory.
- Plan hybrid routes: Map how lab resources will connect to your current cloud and on-prem environments, including identity, logging, and compliance pathways.
Timeline signals
- 2026: Lux AI deployment. Expect early science programs and limited previews before full availability-begin proposal groundwork in 2025.
- 2028: Discovery follows with larger scale. Align long-lead funding, staffing, and interagency partnerships by 2026-2027.
Policy and oversight tie-ins
- Align projects with federal AI risk management, privacy, and security guidance. Coordinate with CIO/CISO, privacy officers, and oversight bodies.
- Track national strategy updates to ensure mission fit and compliance. Reference: AI.gov.
Bottom line: Lux AI and Discovery expand national AI and HPC capacity on clear timelines. Agencies that prepare workloads, governance, and skills now will be first in line to convert this infrastructure into measurable mission outcomes.
Your membership also unlocks:
