loading
loading
Off the Wire Press Releases The proposed federation links SUNRISE, UKAEA’s AI supercomputing platform, with PPPL’s STELLAR-AI to speed up fusion energy research Sept. 14, 2026 — Fusion energy researchers from the United States and the United Kingdom met online Sept. 9 and 10 to turn a shared vision into a plan: linking two of the world’s most capable fusion computing platforms so that artificial intelligence (AI) can help make fusion a reliable source of electricity. The joint workshop brought together scientists and engineers from the Princeton Plasma Physics Laboratory (PPPL), a U.S. Department of Energy (DOE) national laboratory, and the UK Atomic Energy Authority (UKAEA) to discuss a new partnership known as the SUNRISE – STELLAR-AI Federation. The federation will link PPPL’s new AI and high-performance computing platform, Simulation, Technology and Experiment Leveraging Learning-Accelerated Research (STELLAR-AI), with UKAEA’s SUNRISE, the United Kingdom’s first AI supercomputer for fusion. Linking the two systems will let researchers on both sides of the Atlantic share experimental data, train AI on the results of two different experimental fusion facilities and run their models on the machine best suited to the task. Both facilities are spherical tokamaks: machines shaped like cored apples that use magnetic fields to confine a plasma. Experiments in these machines will study whether this compact design could one day form the centerpiece of a fusion power plant that reliably produces electricity. “A fusion power plant is one of the most complex machines humanity has ever tried to build, and no single laboratory or nation will design it alone,” said Jonathan Menard, chief scientist at PPPL. “By federating the SUNRISE and STELLAR-AI computer platforms, we can train AI on data from two leading fusion facilities, test ideas across two of the most capable computing systems in the world that are designed for fusion research and accelerate the path to a compact fusion power plant.” The memorandum of understanding between UKAEA and PPPL that was signed in June gives the two organizations a framework to explore joint endeavors in advanced computing and AI. On the U.S. side, the effort is part of the DOE’s Genesis Mission, a national initiative to unite U.S. national labs, industry and academia to accelerate scientific discovery in energy, science and security. Building the Path to Commercial Fusion Power Plants “Fusion is one of the great scientific and engineering challenges of our time. To solve these challenges, fusion needs partnerships,” said Joe Milnes, executive director for engineering and computing at UKAEA. “The U.S. and UK are two global leaders in fusion research, and the federation of SUNRISE and STELLAR-AI can build on a long history of transatlantic cooperation to further advance fusion development.” Fusion, the process that powers the sun and stars, joins light atomic nuclei to release energy. Re-creating the fusion process on Earth to produce a reliable source of electricity is one of the hardest challenges in science and engineering, and it depends on modeling extremely complex systems. Both PPPL and UKAEA see AI and high-performance computing as essential tools for perfecting the systems for expediting the delivery of electricity from fusion. By coordinating priorities now, PPPL and UKAEA aim to shorten the design cycle for future fusion systems, make better use of existing experiments and give the next generation of fusion scientists a shared platform to build on. The partners plan to expand the federation over time, with the long-term goal of a shared foundation model for spherical tokamaks and, eventually, a wider range of tokamak configurations. Two Supercomputers, One Goal SUNRISE is UKAEA’s new AI supercomputer backed by a £45 million investment from the UK government. It is Phase 1 of the compute infrastructure that underpins the designation of UKAEA’s Culham Campus as the UK’s first AI Growth Zone, with further AI and high-performance computing infrastructure envisaged on the campus benefiting from its direct connection to the UK’s 400 kilovolt supergrid. STELLAR-AI is PPPL’s $13 million AI and high-performance computing platform to be operated with support from Princeton University. STELLAR-AI is designed to host AI tools and workflows as they mature. “Our goal is to let models and experiments move freely between the two systems,” said Shantenu Jha, head of computational sciences at PPPL. “We will turn a collection of supercomputers into a single engine for fusion discovery.” Both organizations operate world-leading compact fusion devices: the National Spherical Torus Experiment-Upgrade (NSTX-U) at PPPL and the Mega Amp Spherical Tokamak Upgrade (MAST-U) at UKAEA. That shared shape is what makes their data ideal for federating. Fusion models trained on a single machine can struggle to predict conditions with high enough accuracy. Quantity and quality of data are critical to building predictive AI tools. Models tested against multiple machines are better able to guide researchers and may provide the ideal designs for future projects, such as the UK’s Spherical Tokamak for Energy Production (STEP) and U.S. concepts such as PPPL’s Spherical Tokamak Advanced Reactor. Workshop participants worked through the practical steps a federation would require. Those include building shared, AI-ready data pipelines that merge NSTX-U and MAST-U results into an open format, training large fusion foundation models on the combined datasets and jointly developing digital twins of both devices. A digital twin is a detailed virtual copy of a real machine, kept in step with it through a constant feed of data, so researchers can test changes and predict how the device will behave in software before trying anything on the actual hardware. These twins would support both the design of future power plants and real-time decisions during experiments. Researchers also examined how to move computing jobs smoothly between the platforms, which use different hardware, so scientists can pick the right processor for each problem. “UKAEA and PPPL would draw on experimental data from two world-leading fusion machines, using simulation to fill the gaps and extend those datasets into regimes we have yet to explore. Together, we can develop digital twins of both machines to support the design of future fusion power plants, creating models of spherical tokamaks that are more predictive, more actionable and, ultimately, more useful to fusion engineers,” said Rob Akers, UKAEA director of computing programmes and senior fellow. “The real power will come from working as one team. Together, we can learn faster and learn more, extracting maximum insight. Computing and international collaboration will help us accelerate the journey from today’s experiments to tomorrow’s fusion power plants,” Akers said. Source: Rachel Kremen, PPPL
document fetched 2026-09-15 02:51 · canonical https://hpcwire.com/off-the-wire/pppl-links-stellar-ai-with-uk-fusion-supercomputer-under-genesis-mission