Sometimes, founders spend years scouring industries for a gap in the market, hunting for a problem complex enough to build a company around. Other times, the market practically knocks down their door. Federico Felici and Jonas Buchli fall squarely into the latter camp.
Over and over again, the two researchers heard the exact same lament from private nuclear fusion companies racing to commercialize clean energy.
“We would ideally buy many of the components to make our control system,” fusion developers kept telling them, “but there isn’t really anybody providing them—and especially there isn’t anybody who speaks the language of fusion.”
Recognizing a glaring bottleneck in the burgeoning commercial fusion supply chain, Felici and Buchli decided to stop solving these problems piecemeal for academic institutions and tech giants. Instead, they packed up their expertise and founded Fusionality, a Lausanne, Switzerland-based startup aiming to provide the critical nervous system for the next generation of power plants.
The company is entering the market with momentum, having recently closed a $3.7 million (CHF 3 million) pre-seed funding round backed by prominent venture firms Founderful and Playfair. With Felici stepping in as Chief Executive Officer and Buchli as Chief Technology Officer, Fusionality is poised to tackle one of the most notoriously difficult engineering challenges on Earth: keeping a man-made star stable long enough to generate electricity for the grid.
Main Facts: The Anatomy of a Fusion Bottleneck
To understand why Fusionality’s emergence is a watershed moment for the nuclear fusion industry, one must understand how a fusion reactor works—and why controlling one is akin to trying to juggle lightning bolts inside a hurricane.
Nuclear fusion generates vast amounts of energy by forcing lightweight atomic nuclei together to form heavier ones, mimicking the process that powers the sun and stars. On Earth, this is most commonly achieved by heating hydrogen isotopes to extreme temperatures—often exceeding 100 million degrees Celsius—until they transform into a state of matter called plasma.
However, superheated plasma is notoriously fickle. It writhes, twists, and attempts to escape containment, struggling against the magnetic or inertial fields designed to hold it in place. Maintaining the precise temperature, density, shape, and fuel mix required to sustain a fusion reaction demands split-second decision-making. If the plasma touches the walls of the reactor even for a fraction of a millisecond, it can damage the vessel and extinguish the reaction entirely.
Traditionally, every single experimental fusion device built over the past several decades has relied on a bespoke, custom-built control system designed completely from scratch by its own team of physicists and engineers.
According to Felici, this represents a massive duplication of engineering effort. “Across the industry, many companies make their own control systems from scratch,” he notes. “But actually, 80% of each company’s control system is really exactly the same.”
While reactor designs vary—some use magnetic fields shaped like doughnuts (tokamaks), others use stellarators, and some rely on laser inertial confinement—the foundational physics governing plasma behavior remain largely consistent. Fusionality aims to exploit this universal overlap. By building a modular suite of control hardware, software, and simulation environments, the startup can provide a standardized foundation that individual fusion companies can fine-tune to match their specific machine architecture.
Chronology: From DeepMind and EPFL to Startup Launch
The journey to Fusionality began long before the company’s incorporation this year. It is rooted in years of academic rigor and cutting-edge artificial intelligence research at some of the world’s most prestigious scientific institutions.
The Academic and Research Roots
Federico Felici and Jonas Buchli originally crossed paths while tackling the monumental challenge of teaching artificial intelligence to control experimental tokamak reactors—the donut-shaped magnetic confinement devices utilized by researchers worldwide.
- The EPFL Years: Felici was based at the École Polytechnique Fédérale de Lausanne (EPFL) in Switzerland, an institution globally renowned for its plasma physics research and home to advanced experimental tokamak facilities. There, Felici spent years wrestling with the real-time feedback loops required to stabilize high-energy plasmas.
- The Google DeepMind Connection: Concurrently, Buchli was working at Google DeepMind, applying advanced machine learning and robotics principles to complex control problems. Eventually, Felici’s path also led him to Google DeepMind, where the two researchers collaborated further. Felici focused heavily on developing sophisticated simulations and machine learning interfaces specifically tailored for fusion devices.
Spending Years Solving the Industry’s Pain Points
For years, Felici and Buchli weren’t just studying fusion control theoretically; they were actively living through the exact mechanical and software headaches that private fusion companies now face daily. They designed algorithms to predict plasma instabilities, built simulation tools to test control theories safely, and fine-tuned real-time data loops.
As the private fusion sector exploded over the last five years—attracting billions of dollars in venture capital and government backing—dozens of new startups began designing their own commercial power plants. As these companies transitioned from theoretical blueprints to physical engineering, they ran into the exact control system roadblock that Felici and Buchli had spent their careers navigating.
“We spent quite a lot of our time solving exactly the problems that our customers need to solve,” Felici explains. Recognizing that the commercial fusion market was finally reaching a critical mass, the duo realized that the window of opportunity was wide open. “It felt like the right time.”
Supporting Data: Funding, Market Landscape, and Initial Focus
Fusionality’s rapid ascent from concept to funded startup underscores the venture capital community’s growing appetite for picks-and-shovels plays within the climate tech and deep tech sectors.
Financial backing and Team Growth
- Pre-Seed Funding: $3.7 million (CHF 3 million).
- Lead Investors: Founderful and Playfair.
- Current Team Size: 7 core team members, scaling up to tackle initial technological deployments.
- Leadership Structure: Federico Felici (CEO) and Jonas Buchli (CTO).
Navigating the Fusion Supply Chain
Fusionality enters a nascent but rapidly evolving ecosystem of supply chain companies servicing the nuclear fusion industry. The market is broadly split into two categories:
- Hardware Manufacturers: Companies like Kyoto Fusioneering, which are building the specialized thermal and energy-conversion components required to transform engineering breakthroughs into electricity feeding into the public grid, alongside traditional high-precision manufacturers adapting their skills for nuclear tolerances.
- Control Systems Specialists: A sparse category, now anchored by Fusionality, focusing on the high-level software, digital twins, and hardware interfaces that manage the reactor’s core operations.
Strategic Market Focus: Magnetic Confinement
For its initial product rollout, Fusionality is narrowing its commercial crosshairs on magnetic confinement fusion—an approach that utilizes immensely powerful electromagnets to compress and heat nuclear fuel to ignition temperatures.
This is currently the most heavily funded and commercialized branch of the fusion industry. Prominent players in this space include:
- Commonwealth Fusion Systems (which recently closed massive funding rounds exceeding $1 billion)
- Proxima Fusion
- Type One Energy (backed by Bill Gates’ Breakthrough Energy Ventures)
- Realta Fusion
Felici notes that while magnetic confinement is the immediate priority, Fusionality’s underlying technological architecture possesses the scalability to eventually expand into other fusion methodologies as the broader industry matures.
Official Perspectives: The Role of AI and Modular Engineering
While artificial intelligence has transformed industries ranging from drug discovery to software development, Felici maintains a pragmatic, measured stance regarding its application inside a live, multi-million-degree nuclear reactor.
The Limits of AI in the Reactor Core
AI will unquestionably play an integral part in the future of energy production, but Felici draws a hard line when it comes to total autonomy.
“AI will play an important role in future control systems,” Felici says. However, he quickly clarifies: “I wouldn’t be somebody who advocates for AI to take the role of controlling the entire fusion reactor.”
In experimental physics, absolute reliability and deterministic safety margins are paramount. Rather than handing complete operational control over to a black-box neural network, Fusionality views AI as an optimizer rather than a master controller. Today, machine learning models are far better suited to “complement, enhance, or optimize” specific sub-systems within the reactor framework—such as predicting localized heat fluxes or forecasting minor plasma disruptions milliseconds before they cascade into a major instability.
The "Lego Block" Product Philosophy
With its newly secured $3.7 million pre-seed capital, the seven-person Fusionality team is focusing on engineering a carefully curated selection of core technologies. While Felici declines to reveal the exact proprietary specifications of their initial products, he uses an accessible analogy to describe their modular development strategy.
“We start with a few of these blocks and then we build out more,” Felici explains, comparing the components to Lego bricks. “We have the ambition to serve the really wide range of technology that you need to operate a fusion reactor.”
By offering a modular toolkit rather than a monolithic, take-it-or-leave-it software package, Fusionality gives fusion developers the flexibility to integrate pre-validated control subsystems into their unique reactor vessels, dramatically slashing development timelines and engineering overhead.
Implications: Accelerating the Timeline to Commercial Fusion
The implications of Fusionality’s business model extend far beyond a single startup’s balance sheet; they touch upon the core timeline of global decarbonization.
For decades, critics of nuclear fusion have repeated the tired adage that "fusion is thirty years away—and always will be." This persistent delay has largely been driven by the staggering complexity of scaling up experimental physics into reliable, commercial-grade engineering. Every hour a fusion startup spends writing custom control code from scratch or troubleshooting feedback latency is an hour not spent refining plasma confinement or optimizing energy capture.
By standardizing the unglamorous, repetitive 80% of fusion control systems, companies like Fusionality act as an institutional force multiplier for the entire sector.
If successful, Fusionality will achieve several critical outcomes for the industry:
- Reduced Capital Expenditure: Fusion startups will no longer need to burn millions of venture dollars building complex control infrastructure from scratch.
- Faster Iteration Cycles: Standardized simulation environments and control blocks allow engineering teams to test new hypotheses virtually and safely, accelerating the design-build-test loop.
- Enhanced Safety and Reliability: By leveraging battle-tested control modules derived from decades of combined research at EPFL and Google DeepMind, new reactor designs inherit a higher baseline of operational stability right out of the gate.
As private fusion companies inch closer to firing up net-energy-producing pilot plants later this decade, the invisible architecture keeping those artificial stars contained will matter just as much as the magnetic coils and superconducting magnets holding them together. With a clear vision, experienced founders, and fresh capital, Fusionality is quietly building the invisible bridge that could help carry nuclear fusion from the realm of theoretical physics into humanity’s everyday power grid.
