When Federico Felici and Jonas Buchli spent years coaxing superheated plasma into submission inside fusion reactors, they noticed something odd: every single fusion company was reinventing the same unglamorous wheels. The diagnostics. The control systems. The data infrastructure. All bespoke, all built from scratch, all expensive.
So the pair, both veterans of Google DeepMind's fusion experiments, decided someone should just sell those wheels. Their Lausanne-based startup Fusionality announced between September 8–10, 2026 it had raised CHF 3 million (roughly $3.7 million) in a pre-seed round led by Founderful and Playfair Capital. The pitch is straightforward: reusable operations infrastructure for magnetic-confinement fusion reactors, the kind of technology fusion developers need but would rather not build themselves.
The company declined to say what valuation investors paid. Fusionality had incorporated earlier in the year and previously received a CHF 40,000 award from Venture Kick, a Swiss startup program, according to the program's announcement.
Felici, who holds a PhD in plasma physics from EPFL and led control activities at the university's Swiss Plasma Center, brings credentials that matter in a field where most people claiming expertise have never actually run a reactor. He taught at TU Eindhoven before his stint at DeepMind. Buchli, an electrical engineer and former ETH Zürich faculty member, led robotics and reinforcement learning teams as a senior research scientist at the same AI lab.
The two met during a collaboration between EPFL and DeepMind that yielded a widely cited Nature paper in February 2022, demonstrating reinforcement learning control of EPFL's TCV tokamak. "We have spent most of our careers making plasmas in real fusion devices behave as we wanted," Felici said in a statement shortly after the funding announcement. "Now we founded Fusionality to bring that expertise" to companies racing to commercialize fusion energy.
What They're Actually Selling
Fusionality's software stack handles measurement and diagnostics, real-time plasma control, scalable data infrastructure, and simulation tools for validating code before deployment on live devices. The startup designed the platform to adapt across different fusion machines with minimal customization. TechCrunch described it as "Lego block" modular technology, though that metaphor perhaps oversimplifies the physics involved.
Initial focus targets magnetic-confinement devices like tokamaks and stellarators, with plans to extend to other fusion approaches later. "We would ideally buy many of the components to make our control system, but there isn't really anybody providing them," Felici told TechCrunch. "And especially there isn't anybody who speaks the language of fusion."
That gap appears real. Felix Neubeck of Playfair Capital called the fusion industry's operations infrastructure "the clearest gap" in its supply chain. "Federico and Jonas are among a very small number" with the expertise to address it, he said in a statement.
The capital will go toward hiring specialized engineers and securing first paying customers, the company said. Fusionality posted a "Founding Operator" role after the fundraise closed, according to a LinkedIn job listing. TechCrunch reported the company had seven employees as of early September, though startups at this stage tend to grow quickly. Fusionality has not publicly disclosed customer names or letters of intent, though sources familiar with the matter noted initial customer engagements are in the works.
A Crowded but Cash-Rich Market
Private fusion companies have attracted significant capital in recent years. According to a Fusion Industry Association report published in mid-2026, private fusion firms raised $4.48 billion in the 12-month period ending July of that year, bringing total private investment since 2021 to more than $14.24 billion across dozens of companies employing upwards of 16,000 people. Those figures suggest an industry transitioning from academic curiosity to industrial ambition, though commercial viability remains unproven.
Magnetic-confinement fusion companies that could become Fusionality customers include Commonwealth Fusion Systems, Realta Fusion, Proxima Fusion, and Type One Energy. Competition exists but appears thin. Next Step Fusion offers control software, diagnostics, and integrated modeling tools including its Fusion Twin and NSFsim products, with control demonstrations on the DIII-D tokamak. Labs and utilities also use open frameworks like MARTe, which has UK Atomic Energy Authority lineage, a supply chain note from the U.S. Department of Energy's Office of Science shows.
Paolo Ricci, director of the Swiss Plasma Center at EPFL, said the center "has decades of experience in plasma control and diagnostics. It is a great development for the field" to see that expertise commercialized, according to Tech Funding News.
Founderful, the Zurich-based pre-seed firm formerly known as Wingman Ventures, has been building out its fusion portfolio. The firm raised $85 million toward a $120 million Fund II target announced in early 2024. Playfair Capital, based in London, invests in a handful of pre-seed startups per year across the UK and Europe from its $70 million Fund III raised in 2023.
One detail worth noting: Felici told TechCrurch he does not advocate for AI to fully control fusion reactors, distinguishing the startup's approach from autonomous systems. That stance may reassure potential customers wary of handing over full control to algorithms, though it also raises questions about how much of the startup's DeepMind heritage will actually ship in the product. The line between AI-assisted and AI-controlled plasma management remains blurry, and Fusionality will need to define it more clearly as it scales.
