CHF 3m · Pre-Seed · Energy · Lausanne, Switzerland
Fusionality has raised CHF 3 million in pre-seed funding co-led by Founderful and Playfair to build operations technology for magnetic-fusion devices. The Lausanne company is spinning out of EPFL’s Swiss Plasma Center and will use the financing to assemble its founding team and begin its first customer engagements, according to regional economic-development reporting and TechCrunch.
Fusionality is not designing a reactor. It is building the measurement, simulation, control and data layer that reactor developers need to operate one. Magnetic-fusion machines must continuously measure a superheated plasma, infer its state and adjust magnets, heating and fuel systems quickly enough to keep the plasma within its target conditions. The company says its team combines plasma physics, control engineering, data infrastructure, AI and real-time hardware-software integration in one stack, drawing on experience with EPFL’s TCV tokamak and other devices (company website).
A reusable control layer for fusion builders
The commercial premise is that much of this operations stack is repeated across otherwise different fusion designs. Fusionality co-founder Federico Felici told TechCrunch that many developers still build control systems from scratch even though a large share of the underlying work is common. Fusionality plans to productise that common layer as a set of interoperable control and simulation components that customers can adapt to their own machines.
That model could let reactor companies keep their engineering teams focused on the parts of the device that differentiate them. Fusionality says its hardware and software are intended to transfer between deployments with limited customisation, so later installations should require less duplicated engineering than the first (Greater Geneva Bern area). The proposition therefore depends on making research-grade control expertise reliable and modular enough for several customers, rather than delivering a new bespoke stack each time.
The founders bring unusually direct experience to that problem. Felici previously led control work at the Swiss Plasma Center before joining Google DeepMind, while Jonas Buchli was an ETH Zurich professor and later a DeepMind senior research scientist. Their earlier collaboration used deep reinforcement learning to control EPFL’s TCV tokamak and resulted in a 2022 Nature paper, according to the company profile and regional reporting.
The seven-person company is initially concentrating on magnetic confinement and a limited set of technologies, rather than claiming to automate an entire reactor. The near-term test is practical: turn the founders’ laboratory and control-system knowledge into products that survive real customer deployments, while supporting machines whose architectures and operating constraints still differ materially. The pre-seed round gives Fusionality room to hire the specialised engineers needed for that translation.
Sources checked:
Tech.eu Funding Explorer · Fusionality · Fusionality announcement · TechCrunch · Greater Geneva Bern area
Displayed date uses 8 September 2026, the publication date of the original TechCrunch report and the matching Tech.eu event date.
