The Swiss startup Fusionality, founded by former Google DeepMind member, has completed a $3.7 million seed round of financing to develop the control systems and simulation tools required for nuclear fusion reactors. The company believes that there has been a long-term shortage in the fusion industry of general suppliers who understand both control engineering and fusion physics.

Aiming at the reactor control system

Fusionality was established in Lausanne, Switzerland this year, with a financing scale of 3 million Swiss francs, which is approximately 3.7 million US dollars. The investors include Founderful and Playfair. The company's tokens were issued by Feder, with Felici serving as the CEO and Jonas Buchli as the Chief Technology Officer.

The two founders stated that they have repeatedly heard the same kind of demand from fusion companies in the past: many companies wish to directly purchase control system components, but there are almost no mature suppliers on the market, especially a lack of teams that understand the operating logic of fusion equipment.

First, develop the common modules and simulation environment.

Fusion power generation relies on the stable operation of plasma, but the temperature, shape, and fuel level all need to be adjusted in an extremely short time, making the development of control systems very challenging. Fusionality believes that although the reactor designs of different companies vary, a considerable portion of the basic capabilities in control systems are common.

The company plans to first develop a set of reusable control modules and simulation environments, after which customers can make custom adjustments according to their respective reactor designs. Felici indicates that many companies in the industry still start from scratch when building control systems, but about 80% of these systems are actually quite similar to each other.

• Financing scale: $3.7 million in pre-seed financing
• Current team: 7 people
• Current Focus: Magnetic Confinement Fusion Direction

AI should be used as an auxiliary tool first.

Both founders have previously been involved in controlling experimental tokamak devices using AI. Felici once worked on related research at the École Polytechnique Fédérale de Lausanne in Switzerland, and Buchli held a position at Google DeepMind. Subsequently, Felici also joined DeepMind, where they were responsible for the simulation of fusion devices and the development of machine learning interfaces.

The company believes that AI will play an important role in future fusion control systems, but in the short term, it is not yet sufficient to independently take over the entire reactor control task. At this stage, AI is more suitable for supplementing, optimizing, or enhancing certain aspects of the system.

Fusionality is currently focusing on magnetic confinement fusion, a approach that maintains the high temperature and high density of fusion fuel through strong electromagnets. The company stated that this funding will be used to advance the development of a few core technical modules, with plans to gradually expand the product range in the future to cover more capabilities required for the operation of fusion reactors.