UK and US Link AI Supercomputers to Accelerate Commercial Fusion Energy

The UK Atomic Energy Authority (UKAEA) and the US Department of Energy’s Princeton Plasma Physics Laboratory (PPPL) have signed a Joint Declaration of Intent to link their AI supercomputing platforms. This partnership aims to accelerate the development of commercial fusion energy by training machine-learning models on combined datasets from both nations.

Announced during the Global Fusion Policy Summit in London on September 14, 2026, the collaboration establishes the SUNRISE–STELLAR-AI Federation. The initiative links the UKAEA's SUNRISE supercomputer—the UK's first AI system dedicated to fusion, supported by £45 million in government funding—with the STELLAR-AI platform at PPPL. The federation will specifically utilize data from two compact spherical tokamaks: the MAST Upgrade facility in the UK and the NSTX-U facility in the US.

Overcoming Data Silos

Fusion energy research has historically been hindered by data silos, where AI models trained on a single machine often fail to generalize when applied to different facilities. By federating these supercomputers, researchers can move beyond isolated datasets to create "digital twins." These virtual copies of fusion machines allow engineers to test software changes and configurations in a simulated environment before applying them to physical hardware, significantly shortening the design cycle for future power plants.

Implications for Energy Production

Commercial fusion remains one of the most complex engineering challenges in history. This transatlantic capability allows for the training of more predictive and actionable models by synthesizing high-quality data from two world-leading facilities. According to Rob Akers, Director of Computing Programmes and Senior Fellow at UKAEA, these digital twins will create models of spherical tokamaks that are more useful to fusion engineers. Jonathan Menard, Chief Scientist at PPPL, noted that federating the platforms will accelerate the path toward a compact fusion power plant.

The Path to the Grid

Moving forward, the partnership is designed to support the design of two major projects: the UK's Spherical Tokamak for Energy Production (STEP Fusion) and the US's Spherical Tokamak Advanced Reactor (STAR). By refining the predictive capabilities of AI, the federation seeks to move fusion from experimental physics to a viable, carbon-free, and sustainable energy source on the power grid.