Fusionality has raised $3.7 million to develop reusable control systems for fusion machines, positioning itself as a provider of an adaptable operations layer for magnetic-confinement reactors.
Fusionality’s reusable operations layer
Fusionality is initially targeting magnetic-confinement systems, including tokamaks and other devices that use magnetic fields to contain hot plasma. Its stated product scope covers real-time diagnostics, plasma control, data systems and simulation environments. The company says fusion developers commonly build control systems internally even though much of the underlying functionality is repeated across reactor designs. Fusionality’s proposed difference is a reusable operations layer that can be adapted to a specific machine, rather than a reactor company starting every control stack from scratch. That positioning is the startup’s commercial claim; it has not published comparative performance data against named commercial platforms.
Founders’ research and the role of AI
The founders’ earlier research demonstrated reinforcement-learning controllers trained in simulation and deployed on TCV hardware. A Nature paper reported control of multiple plasma shapes, including elongated and ‘snowflake’ configurations, and described a controller operating 19 magnetic coils at a 10-kilohertz control rate. Fusionality says AI will complement conventional control methods rather than independently operate an entire reactor.
