Fusionality, a startup founded this year by Google DeepMind researchers, has raised $3.7 million in pre-seed funding from Founderful and Playfair to develop standardized control systems for fusion power plants.
Fusionality CEO Federico Felici and CTO Jonas Buchli identified the gap while working on experimental fusion devices. At Google DeepMind, they developed AI simulations and machine learning interfaces for fusion reactors. The problem they observed: roughly 80 percent of the custom control systems built by fusion companies share identical fundamental requirements, yet each startup engineers its own solution from scratch.
"Fusion companies want to buy control system components rather than build them," Felici said. "But there's almost no one who understands the technical requirements."
The Lausanne-based company plans to sell a standardized suite of control systems and simulation environments that fusion startups can customize for their specific reactor designs. The economics are straightforward—reducing engineering redundancy accelerates time-to-market for fusion energy companies and creates a recurring revenue stream for Fusionality.
Fusion reactors maintain stability by fusing atoms at extreme temperatures inside a plasma field. The plasma is inherently unstable and requires split-second control adjustments. The physics governing plasma stability is consistent across reactor designs, which gives Fusionality a foundation for a modular product.
Felici cautioned that artificial intelligence will play a supporting role in future control systems but cannot yet manage an entire fusion reactor autonomously. The implication: Fusionality's initial product will likely combine AI with traditional control logic, not replace engineering teams.
Fusionality operates in a nascent supply chain supporting fusion energy startups. Competitors include Kyoto Fusioneering, which develops power conversion components to turn fusion breakthroughs into grid electricity.
