Former DeepMind Engineers Launch Fusionality to Speed Up Fusion Power Development

Former engineers from Google DeepMind have founded a new venture, Fusionality, aimed at providing advanced software tools that could accelerate the path to commercial fusion energy. The startup focuses on delivering sophisticated control systems and high‑fidelity simulation environments designed to help emerging fusion companies iterate more quickly and reduce the technical risk associated with bringing a working reactor to the grid.

Fusionality’s platform combines real‑time control algorithms with AI‑enhanced modeling to create digital twins of experimental reactors. By allowing engineers to test control strategies and predict plasma behavior in a virtual setting, the tools promise to shorten the design‑build‑test cycle that traditionally dominates fusion research. The company says its software can integrate data from a variety of diagnostics, offering a unified view that streamlines decision‑making during experiments.

The effort arrives at a moment when private‑sector fusion projects are gaining momentum worldwide. Startups such as Commonwealth Fusion Systems, Tokamak Energy and Helion Energy have secured sizable funding and are pursuing compact, high‑field designs. Yet all share a common bottleneck: the need for precise, reliable control of plasma dynamics and the ability to simulate complex interactions before costly hardware runs. Fusionality positions itself as a shared infrastructure layer that could be adopted across these disparate approaches.

Accelerating fusion development carries broader implications for the energy transition. If reactors can reach net‑positive output sooner, they could provide a baseload, carbon‑free power source to complement intermittent renewables. Analysts have long highlighted fusion’s potential to meet long‑term climate goals, but timelines have remained speculative. Tools that cut engineering lead times may help align commercial milestones with policy targets for decarbonization.

While the company has not disclosed detailed financial terms, it reports having secured early‑stage investment from venture firms that specialize in deep‑tech and clean‑energy ventures. Fusionality also notes collaborations with national laboratories and academic groups, leveraging existing experimental data to train its simulation models. These partnerships are intended to validate the software against real‑world reactor performance and to refine its predictive capabilities.

Looking ahead, Fusionality plans to roll out beta versions of its control suite to a select group of fusion startups later this year, with broader availability slated for early next year. Success will be measured by reductions in experimental downtime and improvements in plasma stability metrics reported by its users. If the platform proves effective, it could become a standard component of the fusion development stack, shaping how the next generation of reactors are designed, tested, and ultimately integrated into the electricity grid.