Federico Felici and Jonas Buchli launched Fusionality this year to design control systems and simulation environments for fusion reactors, aiming to fill a gap in the industry where few companies specialize in the complex hardware and software needed to manage these advanced energy systems. Based in Lausanne, Switzerland, the startup recently secured a $3.7 million (CHF 3 million) pre-seed investment from venture capital firms Founderful and Playfair. Felici is the company’s CEO, while Buchli serves as its Chief Technology Officer. Felici and Buchli met while teaching artificial intelligence to control an experimental tokamak, a donut-shaped machine used to study nuclear fusion. Felici was working at EPFL, a leading research institution in Switzerland that hosts a tokamak, while Buchli was employed at Google DeepMind, a company known for its work in artificial intelligence. Later, Felici joined Google DeepMind, where he helped develop simulations and machine learning tools for fusion devices. Felici pointed out that many companies in the fusion industry currently build their own control systems from scratch, even though 80% of these systems are similar across different projects. Fusionality aims to leverage this commonality by creating a standardized set of control systems and simulation environments that startups can customize for their specific reactor designs. This approach could reduce costs and development time for smaller companies entering the field. Fusion power is generated when atoms combine to release energy, a process that requires extremely high temperatures to occur in a plasma state. Maintaining the right conditions in this plasma—such as temperature, shape, and fuel levels—requires rapid decision-making, making the control systems involved both complex and critical. However, much of the underlying physics remains consistent across different reactor designs, offering opportunities for shared technological solutions. Felici acknowledged that artificial intelligence will play a growing role in future fusion control systems, but he emphasized that it is not yet capable of managing an entire reactor on its own. "I wouldn't be someone who advocates for AI to take the role of controlling the entire fusion reactor," he said. Instead, he sees AI as a tool to improve and optimize parts of the control system rather than replace human oversight entirely. Fusionality is currently focusing on magnetic confinement, a method that uses strong electromagnets to keep fusion fuel hot and dense enough to sustain a reaction. Several startups, including Commonwealth Fusion Systems and Realta Fusion, are also working on magnetic confinement devices. In the future, Fusionality plans to expand its technology to other fusion approaches as well. With its pre-seed funding, Fusionality’s seven-person team aims to develop a carefully chosen set of core technologies, which Felici compared to Lego blocks—starting with a few foundational pieces and building outward from there. The company aspires to provide the full range of technologies needed to operate a fusion reactor, supporting both current and future advancements in the field.