Fusionality Raises $3.7M to Build Fusion Software

The big picture: Lausanne-based Fusionality, a startup developing measurement, simulation, and control systems for fusion companies, has secured $3.7 million in pre-seed funding. The round was led by Founderful and Playfair. The company’s official website is fusionality.com.

Why it matters:

• Critical Need: Over 30 private fusion companies and government projects require advanced software to convert raw plasma measurements into control signals for heating, fueling, and magnetic systems within milliseconds.
• Market Gap: Despite record private fusion investment, most capital is not directed towards reusable infrastructure, with companies typically designing control systems from scratch.
• Expertise Leverage: Fusionality’s founders bring extensive experience from Google DeepMind and EPFL, where they applied AI to control plasma in tokamak reactors.

How it works:

• AI-Driven Control: Fusionality develops specialized AI to prevent extreme plasma clouds, reaching 100 million degrees, from contacting reactor walls by precisely managing magnetic fields.
• Founders’ Background: Co-founders Federico Felici and Jonas Buchli previously conducted research at Google DeepMind and EPFL’s Swiss Plasma Center, publishing findings on deep reinforcement learning for plasma control in Nature.
• Standardized Software: The company offers a basic software layer designed for minimal adaptation, aiming to reduce onboarding costs for customers and provide a common infrastructure for the nascent fusion industry.

The catch: Fusionality faces a competitive landscape, including existing players like Next Step Fusion, which supplies plasma modeling and control software. A significant challenge also emerges from Google DeepMind’s partnership with Commonwealth Fusion Systems and the open-sourcing of the TORAX plasma simulator, potentially offering free solutions that could diminish the market for paid third-party software. The industry’s current tendency for each company to develop proprietary control systems from scratch could also hinder the widespread adoption of a standardized solution.