Federico Felici and Jonas Buchli founded Fusionality in Lausanne this year to address a practical gap in the nuclear fusion energy sector: the lack of ready-made, specialized tools for building the control systems that operate fusion reactors. The startup has raised a $3.7 million, or CHF 3 million, seed round led by Founderful and Playfair.
The funding comes as a new supply chain takes shape around fusion companies attempting to move from engineering experiments to generating electricity for the grid. While other companies develop reactor components or manufacture high-precision parts, Fusionality believes that relatively few organizations are focusing on the software and hardware needed to control the reactor itself.
A Common Problem Across Different Designs
Felici said that many fusion companies build their control systems from scratch, even though about 80% of every control system is effectively similar across companies. Fusionality is building a package of control systems and simulation environments around this shared portion, with startups later customizing the tools to suit their specific reactor designs.
Fusion reactors need to maintain extremely hot plasma within an appropriate shape, temperature, and fuel level. This task requires fast and complex decisions because plasma is unstable, and changing conditions can disrupt an experiment or cause it to lose the desired state. The value of control tools therefore extends beyond reducing development time; it is also connected to companies’ ability to test their systems before deploying them on costly and complex experimental devices.
Artificial Intelligence as Part of the System
The two founders met while working on teaching artificial intelligence to control an experimental tokamak, a ring-shaped reactor used by researchers to study fusion. Felici was working at EPFL in Switzerland, while Buchli was working at Google DeepMind. Felici later moved to Google DeepMind, where he developed simulation processes and machine-learning interfaces for fusion devices.
However, the company does not present artificial intelligence as a replacement for the entire control system. Felici explained that artificial intelligence will play an important role in the future, but for now it is better suited to supplementing, improving, or tuning parts of the system rather than taking control of the reactor as a whole.
What Changes in Practice?
In its initial phase, Fusionality is focusing on magnetic-confinement fusion, which uses powerful magnets to keep fusion fuel dense and hot enough to power an electricity-generating plant. Companies developing devices under this approach include Commonwealth Fusion Systems, Realta Fusion, Proxima Fusion, and Type One Energy.
The funding will be used to develop a carefully limited number of technologies, according to Felici, in the form of “building blocks” that can be assembled and expanded gradually. The company’s team currently numbers seven people, while Fusionality has not specified the first technologies it will build. The company says its ultimate ambition is to provide a broad set of tools needed to operate a fusion reactor, later expanding into approaches other than magnetic confinement.
certi.news Analysis
The move shows that the challenge facing the fusion sector is not limited to inventing a working reactor; it also includes building a repeatable tooling layer around it. If Fusionality succeeds in turning the common parts of control systems into customizable products, startups may be able to reduce duplicated engineering work. However, the source does not specify the technologies the company will launch, its availability timeline, or operational results proving its ability to work outside experimental environments; its actual commercial and technical impact therefore remains to be verified.