Fusionality raises $3.7M to build control systems for fusion reactors, founded by DeepMind alumni

The idea for Fusionality emerged after the founders repeatedly heard from fusion companies that they wanted to buy components for control systems but found no suppliers who understood their specific requirements. “Across the industry, many companies make their own control systems from scratch. But actually, 80% of each company’s control system is really exactly the same,” Felici told TechCrunch.

Felici and Buchli met while using AI to control an experimental tokamak at EPFL in Switzerland. Felici later moved to DeepMind, where he developed simulations and machine learning interfaces for fusion devices. The company is focusing on magnetic confinement reactors, a branch that uses powerful electromagnets to sustain the superheated plasma needed for fusion.

While AI will play an important role in future control systems, Felici cautioned that it is not yet ready to take full control. “I wouldn't be somebody who advocates for AI to take the role of controlling the entire fusion reactor,” he said. Instead, AI is more likely to “complement or enhance or optimize” parts of the system. Fusionality is part of a growing ecosystem of suppliers serving the fusion power industry, which includes companies like Kyoto Fusioneering developing components for next-generation reactors.

Why This Matters

The fusion industry is approaching the point where commercial reactors need reliable, standardised components. A company that can provide off-the-shelf control systems could lower costs and development timelines for multiple startups.
The founders' DeepMind and EPFL background links fusion control to advanced AI techniques, potentially accelerating progress in maintaining stable plasma conditions.
The $3.7 million pre-seed round signals early investor confidence in the fusion supply chain, an area that has historically received less attention than reactor design.

Background

Fusion power promises nearly limitless clean energy by replicating the process that powers the Sun. However, maintaining the precise conditions for sustained fusion — extreme temperatures, densities, and magnetic confinement — requires split-second adjustments. Most fusion startups today build their own control systems from scratch, a costly and time-consuming process. Fusionality aims to provide a common platform that can be customised, freeing startups to focus on their core reactor designs. The company is initially targeting magnetic confinement approaches, which include tokamaks and stellarators.

Key Perspectives

Federico Felici (CEO, Fusionality): He sees a clear market gap: fusion companies want to buy control systems but cannot find suppliers that “speak the language of fusion.” By leveraging shared physics across designs, Fusionality can deliver 80% of a control system as standard, reducing duplication of effort. Fusion startups: Many companies acknowledge the burden of building custom control software and hardware. A commercial product could accelerate their development timetables and allow them to allocate resources elsewhere. Critics/Skeptics: Felici himself notes that AI is not yet ready to fully control a reactor. Skeptics might argue that the industry is still too small and fragmented to support a dedicated supplier, or that reactor designs are too diverse for a one-size-fits-most solution. Additionally, regulatory and safety requirements for fusion plants could demand custom engineering that a standard platform cannot provide.

What to Watch

Whether Fusionality can secure pilot agreements with one or more prominent fusion startups within the next 12–18 months.
The timing of its first commercial product release and the range of reactor types it supports.
Follow-on funding rounds and whether larger investors see the fusion supply chain as a viable bet.

ZOTPAPER