Google DeepMind alumni launch Fusionality to build the software layer for fusion power

TechCrunch reports that Lausanne-based Fusionality was founded in 2026 by Federico Felici and Jonas Buchli, both of whom have links to Google DeepMind and academic fusion research. The company reportedly raised $3.7m, or CHF 3m, from Founderful and Playfair.

Fusionality is focused initially on magnetic-confinement fusion. Its stated work covers integrated measurement, control, simulation and data systems rather than asking an AI model to run an entire reactor unsupervised.

That is an important distinction. Fusion needs hardware, materials, plasma physics and control software to work together. Better software will not make the engineering problem vanish, but poor software can certainly make it harder.

Fusion experiments generate enormous volumes of measurements and require precise control of unstable plasma conditions. That creates a real software problem alongside the reactor hardware problem: systems have to observe what is happening, model it quickly and adjust the machine without introducing new risks.

Fusionality’s founders are not promising that AI alone will deliver commercial fusion. The more modest idea is to build reusable tools that help several fusion companies develop and operate their experiments. That could become valuable infrastructure if the sector moves from laboratory demonstrations towards repeatable power systems.

The company’s official description focuses on integrated measurement, control, simulation and data systems. In practice, that means connecting the sensors, models and control loops that let engineers understand what a plasma is doing and respond quickly when conditions change.

That is not as flashy as promising a magic AI reactor, but it is exactly the sort of infrastructure an experimental industry needs. The remaining question is whether Fusionality can turn a small team’s expertise into software that works across different machines and fusion approaches.

For now, the funding report is stronger than the public evidence available from the company. The draft should keep the round, founders and technical aims attributed to TechCrunch and Fusionality rather than presenting commercial fusion progress as already achieved.

That makes the company worth watching without pretending the commercial-fusion finish line has moved closer overnight. Fusionality is building tools for a difficult industry; it has not yet demonstrated a working power plant.

Our opinion

The fusion industry has enough miracle claims already. Fusionality’s software-first approach is refreshingly practical — provided the company can turn promising control tools into systems that survive contact with an actual reactor.