British Government Unveils £45 Million Sunrise Supercomputer to Accelerate National Nuclear Fusion Research Goals

Britain reveals Sunrise, the world's most powerful AI supercomputer for fusion energy, designed to fast-track commercial nuclear power by the 2040s.

By: AXL Media

Published: Mar 17, 2026, 7:53 AM EDT

Source: The information in this article was sourced from ITPro

A Virtual Testing Ground for the Apollo Era of Fusion

The UK Atomic Energy Authority is set to receive a massive infrastructure boost with the introduction of the Sunrise supercomputer, a facility designed to mirror the iterative intensity of the Apollo space program. According to Rob Akers, the director for computing programmes at UKAEA, the goal is to utilize virtual environments to test and refine complex energy systems before committing to physical construction. This £45 million investment aims to provide a 1.4 megawatt hub where AI-accelerated modeling can occur at a scale of 6.76 exaflops, effectively creating a sandbox for the next generation of clean energy.

The Strategic Intersection of AI Growth and Atomic Energy

Situated within the UK's inaugural AI Growth Zone in Oxfordshire, the Sunrise system represents a pivot toward physics-informed artificial intelligence. Thomas Zacharia, a senior vice president at AMD, noted that the integration of EPYC processors and Instinct GPU acceleration on Dell hardware is purpose-built to handle the unique demands of the fusion community. By placing this high-performance computing asset at the Culham Campus, the government is positioning the region as a central node for the Department for Energy Security and Net Zero, ensuring that academic research translates directly into industrial application.

Tackling the Turbulence of Plasma and Tritium Cycles

The technical scope of the project focuses on the most persistent hurdles in nuclear science, specifically the management of plasma turbulence and the development of specialized materials. One of the primary initiatives slated for the system is the Lithium Breeding Tritium Innovation program, which seeks to master the fuel-cycle technologies necessary for self-sufficient fusion plants. By simulating these volatile conditions in a digital environment, researchers hope to reduce the astronomical costs and physical risks associated with traditional laboratory testing, providing a more stable path toward a functional energy grid.