Federico Felici and Jonas Buchli, former researchers with backgrounds at Google DeepMind and EPFL, have founded Fusionality, a startup focused on developing control systems for fusion power plants. The company, based in Lausanne, Switzerland, raised $3.7 million in a pre-seed funding round led by Founderful and Playfair. Fusionality aims to address a gap in the fusion industry where many companies build their own control systems from scratch despite significant overlap in requirements.
Fusion reactors rely on maintaining precise conditions within superheated plasma to sustain fusion reactions, a process that demands complex and rapid control. While various startups and manufacturers are developing components to convert fusion breakthroughs into grid-ready electricity, few specialize in the integrated hardware and software needed to manage the reactors themselves.
Felici and Buchli’s experience includes teaching AI to control tokamaks, donut-shaped devices used in fusion research. Felici worked at EPFL and later at Google DeepMind, where he developed simulations and machine learning tools for fusion. Although AI is expected to enhance future control systems, Felici notes it is not yet ready to fully automate reactor control but can optimize specific system components.
Currently, Fusionality is concentrating on magnetic confinement fusion, which uses strong electromagnets to contain and heat fusion fuel. The startup plans to create modular control technologies that can be adapted by various fusion companies, likened by Felici to building blocks that can be combined to meet diverse reactor designs. The company’s initial team of seven will focus on developing these foundational technologies to support the growing fusion power industry.