The US Department of Energy’s (DOE) Princeton Plasma Physics Laboratory (PPPL) has launched a new computing platform that pairs artificial intelligence (AI) with high performance computing with the aim of ending the bottleneck holding back fusion energy research by speeding the simulations needed to advance the field.
The project – the Simulation, Technology and Experiment Leveraging Learning-Accelerated Research enabled by AI (STELLAR-AI) – will bring together national laboratories, universities, technology companies and industry partners to build the computational foundation the fusion community needs.
It can take months to run a single high-fidelity computer simulation or to train an artificially intelligent (AI) system capable of designing an ideal fusion system using existing infrastructure. STELLAR-AI is designed to reduce that timeline by connecting computing resources directly to experimental devices, including PPPL’s National Spherical Torus Experiment-Upgrade (NSTX-U) scheduled to go live this year. This will allow researchers to analyse data as experiments occur.
“Fusion is a complex system of systems. We need AI and high performance computing to really optimise the design for economic construction and operation,” said Jonathan Menard, Deputy Director for research at PPPL. “We want to link simulation technology and experiments, in particular, NSTX-U, with AI and partnerships to get to accelerated fusion.”
STELLAR-AI will achieve this goal by integrating central processing units (CPUs), graphics processing units (GPUs) and quantum processing units (QPUs) in an ideal configuration of hardware for tackling the challenges facing private fusion companies. CPUs are standard computer chips that handle everyday computing tasks; GPUs are specialised chips that excel at the parallel calculations needed for artificial intelligence; and QPUs use the principles of quantum physics to solve certain complex problems that would take traditional computers far longer to complete.
STELLAR-AI is part of the Genesis Mission, a US effort launched by Executive Order in November 2025 to use AI to speed up scientific discovery across DOE laboratories. “The Genesis platform is an integrated, ambitious system that will bring together the various unique DOE assets: experimental and user facilities, the supercomputers, data archives and, importantly, the AI models,” said Shantenu Jha, head of PPPL’s Computational Sciences Department. While Genesis provides that broad infrastructure, STELLAR-AI contributes fusion-specific computer codes, data and scientific models back into the national system. The project also aligns with the DOE’s Fusion Science & Technology Roadmap, which calls for building an AI-Fusion Digital Convergence platform to accelerate commercialisation of a fusion power plant.
Researchers plan to use STELLAR-AI for projects that span simulation, design and real-time experiment support. One effort will create a digital twin of NSTX-U: a computer model that mirrors the physical machine so closely that scientists can test ideas virtually before running actual experiments.
Another project, called StellFoundry, uses AI to speed the design of stellarators. This requires sifting through enormous amounts of data to find the best configurations, a process that traditionally takes months or years and will greatly benefit from the STELLAR-AI platform.
PPPL says the strength of STELLAR-AI lies in its partnerships with DOE National Laboratories, AI and HPC companies, academic institutions, as well as fusion and engineering companies. The team includes capabilities from national laboratories, including PPPL and UKAEA as well as top universities such as Massachusetts Institute of Technology and University of Wisconsin-Madison.
Princeton University, which manages the laboratory for the DOE’s Office of Science, is also a key partner. Princeton will support operations, research software engineering, and user training for the STELLAR-AI infrastructure. Crucial technical support comes from tech companies such as NVIDIA, which is providing expertise to improve the performance of several critical fusion codes, and Microsoft, which will federate Azure’s leading cloud capabilities. There is also direct collaboration with the fusion industry, including Commonwealth Fusion Systems, General Atomics, Type One Energy and Realta Fusion.
