UK Atomic Power Authority readies fusion simulation AI supercomputer

Culham Campus in Oxfordshire, the location of the UK’s first synthetic intelligence (AI) development zone, will home the UK’s first AI supercomputer.

The UK authorities is investing £45m right into a 1.4MW supercomputer named Dawn, a key first step in establishing the nation’s first AI Progress Zone on the UK Atomic Power Authority’s (UKAEA).

Unveiled within the UK’s Fusion Technique, Dawn is because of be prepared by June. The federal government claims it’s the world’s strongest AI supercomputer devoted to fusion power.

With the worldwide financial system underneath siege from the efficient closure of the Strait of Hormuz because of the US and Israeli bombing of Iran, the UK authorities is find out how to buffer the nation’s power from volatility within the oil market.

Patrick Vallance, minister for science, innovation, analysis and nuclear, mentioned: “By backing our fusion business, we’re not solely securing our future power independence, however from innovation and analysis to engineers, we’re additionally offering the expert clear power jobs of the longer term for British individuals.”

Funded by the Division of Power Safety and Internet Zero (DESNZ), Dawn is being constructed to deal with fusion power challenges in areas corresponding to plasma turbulence, supplies growth and tritium gasoline breeding. Additionally it is being positioned to strengthen important AI capabilities at Culham Campus and throughout the UK’s high-performance computing panorama, contributing to the federal government’s AI Alternatives Motion Plan and AI for Science technique.

The 6.76 Exaflops AI supercomputer includes a collaboration between AMD, DESNZ, the Division for Science, Innovation and Expertise (DSIT), Dell Applied sciences, Intel, UKAEA, the College of Cambridge, and Weka, an information platform supplier.

Dawn makes use of AMD Epyc processors and AMD Intuition graphics processor unit (GPU) acceleration, purpose-built on a Dell PowerEdge platform. Thomas Zacharia, senior vice-president of technique and growth for the general public sector at AMD, mentioned: “Fusion analysis pushes the boundaries of science and computing, demanding huge simulation, advanced modelling and superior AI to speed up progress.”

The UKAEA mentioned Dawn can be used to speed up modelling, to energy high-fidelity simulations and allow the creation of digital twins for advanced methods.

Rob Akers, UKAEA’s director for computing programmes, mentioned: “Dawn will carry that functionality to fusion by combining high-fidelity simulation with physics-informed AI to develop predictive digital twins that scale back the price, danger and time of studying that will in any other case require costly and time-consuming bodily testing.

“UKAEA is proud to be working with such a pioneering group of companions to harness AI and high-performance computing at scale to help the UK’s fusion roadmap and web zero mission,” he added.

In 2023, Dell Applied sciences, Intel, the College of Cambridge and UKAEA shared plans to make use of supercomputers and AI to advance the event of the UK’s prototype fusion energy plant design capabilities by way of the ‘industrial metaverse’.

In January 2026, £36m of presidency funding was injected into the Cambridge supercomputing centre.

Paul Calleja, director of the Cambridge Analysis Computing Service, mentioned: “Dawn is a crucial first step within the UK’s daring imaginative and prescient to strengthen its sovereign scientific computing functionality, speed up fusion analysis, and lay the foundations for the Culham AI Progress Zone.”