Simulating Fusion Energy at Exascale

Nuclear fusion promises a virtually limitless, clean and safe source of energy, but achieving it on Earth remains one of the most complex scientific and engineering challenges. Fusion reactors involve extreme conditions — ultra-hot plasmas, intense magnetic fields and materials exposed to high radiation — all interacting across a wide range of spatial and temporal scales. Understanding and controlling these processes requires advanced large-scale computer simulations.Recent advances in high-performance computing are transforming how fusion systems can be studied. The emergence of exascale supercomputers, capable of performing more than a billion billion calculations per second, enables simulations with unprecedented realism. Instead of modelling isolated aspects of a reactor, researchers can now move towards integrated simulations that combine plasma physics, materials science and engineering.This progress opens the way to the concept of a digital twin of a fusion reactor: a comprehensive virtual replica that mirrors the behaviour of a real device. Digital twins can be used to optimise reactor designs, guide experimental programmes and anticipate operational challenges before costly hardware is built. They are expected to play a key role in major international fusion projects, as well as in the rapidly growing private fusion sector.Fully exploiting these new computing capabilities also presents major challenges. Fusion simulation codes must be adapted to highly complex and heterogeneous computer architectures, often based on graphics processing units. Key issues include managing massive data volumes, minimising data movement, ensuring efficient parallelism and maintaining portability across different supercomputing platforms.By bringing together physics, computer science and engineering, exascale simulations are becoming a cornerstone of fusion energy research and an essential tool for accelerating the path towards practical fusion power.