INL and NVIDIA align AI platform to speed advanced nuclear rollout

by Clarence Oxford
Los Angeles CA (SPX) Feb 23, 2026
Idaho National Laboratory has entered a new collaboration with NVIDIA to apply artificial intelligence and accelerated computing to the full life cycle of advanced nuclear reactors, from early design through long term operation. The partners aim to shorten deployment timelines for commercial scale nuclear units and cut operating costs while maintaining high safety margins.

The work sits under the Genesis Mission, a national initiative to build an integrated scientific and computing platform to tackle high impact challenges in energy, security, and discovery science. Within this framework, the INL NVIDIA effort drives the Prometheus Grand Challenge, focused on delivering nuclear energy that is faster to deploy, safer to run, and cheaper to operate.

Prometheus is planned as a converged platform that uses AI to support design, licensing, manufacturing, construction, and operation of reactors in human in the loop workflows. Program targets include at least a twofold acceleration of project schedules and operational cost reductions of more than fifty percent by using digital tools to optimize engineering, streamline regulatory interactions, and improve plant performance.

The initiative is framed as a response to two linked priorities: harnessing AI to underpin a new industrial and scientific cycle, and meeting sharply rising electricity demand from data centers, transportation, and advanced manufacturing. The concept is that AI techniques will help move more reactors from concept to grid connection, while nuclear plants in turn supply the steady baseload power needed to run next generation AI infrastructure at scale.

INL Director John Wagner said the collaboration represents a step change in how the United States can approach deployment of abundant, reliable nuclear energy. He emphasized that applying AI to design, licensing, and operations provides a way to compress timelines that have traditionally extended over decades while preserving rigorous safety analysis and review.

NVIDIA will contribute AI infrastructure and accelerated computing platforms to run large scale simulations and enable complex workflow automation. Company executive John Josephakis noted that combining INL's long standing nuclear capabilities with NVIDIA hardware and software is intended to support faster, safer, and lower cost reactors that can supply the large amounts of energy needed for future scientific and industrial workloads.

The U.S. Department of Energy sees the effort as a model of targeted AI deployment for critical infrastructure. Rian Bahran, Deputy Assistant Secretary of Energy for Nuclear Reactors, said the public private partnership is designed to go beyond incremental efficiency gains and instead to change how nuclear plants are planned, built, and used for research and development and discovery.

Several technical thrusts define the initial scope of the collaboration. One is the development of generative AI models, high fidelity digital twins, and agent based workflows to capture reactor physics, materials behavior, and plant systems so that engineers can explore design options and operating strategies in silico before committing to hardware.

A second area centers on helping the wider nuclear industry adopt accelerated computing and AI tools. The team plans to work with vendor companies and utility operators and to engage with regulators to clarify how autonomous control features and advanced digital capabilities can fit within existing safety frameworks and future licensing approaches.

The partners will also tap Department of Energy leadership class supercomputers for computationally intensive model training and physics based simulations. At the same time, they are studying on premises NVIDIA AI systems that could support real time plant monitoring, predictive maintenance, and control within operating nuclear facilities.

Access to real world data is another key element. INL will use its historical nuclear data sets, current laboratory experiments, and on site reactors, including the Neutron Radiography Reactor and the not yet operational Microreactor Applications Research Validation and Evaluation system, to calibrate and validate digital twins and AI models against measured performance.

The project includes a major code modernization and acceleration component. Core nuclear simulation tools such as MOOSE, BISON, Griffin, and Pronghorn are being optimized to run on NVIDIA GPU architectures, with the goal of enabling higher resolution multiphysics models and faster turnaround times for safety analysis, fuel performance studies, and reactor systems evaluations.

Looking ahead, INL and NVIDIA anticipate that Prometheus could expand beyond the initial partners. The collaboration may bring in reactor developers, utilities, investors, and additional national laboratories to build out an ecosystem that supports AI enhanced nuclear deployment across different reactor technologies and project scales.