The thesis: shrink the machine, not the physics
Commonwealth Fusion Systems is the best-capitalized and most closely watched bet that fusion energy can be turned into a buildable, financeable product this decade. Its central idea is deceptively simple: the tokamak — the doughnut-shaped magnetic bottle that most of the fusion establishment considers the surest route to net energy — works better the stronger its magnetic field, and the strength of that field had been capped for decades by the limits of conventional superconductors. Break that ceiling and you can build a machine that reaches the same fusion conditions in a fraction of the volume, at a fraction of the cost, and on a private-company timeline rather than a multinational-treaty one.
That is the whole wager. Where the international ITER project pursues fusion with a reactor of enormous scale and a schedule measured in decades, CFS is trying to reach net energy in a tokamak roughly one-fortieth the volume by using a new class of magnet. If SPARC, the demonstration machine now nearing completion in Devens, Massachusetts, produces more fusion energy than it consumes in 2027 as planned, CFS will have collapsed a generational physics milestone into a startup's product roadmap — and made itself the front-runner to build the world's first commercial fusion power plant.
From an MIT student design to a $3B company
CFS's origins run through MIT's Plasma Science and Fusion Center, where professor Dennis Whyte supervised a series of student-led designs for a compact, high-field tokamak built around a then-emerging high-temperature superconductor. That work crystallized into the ARC concept — a small fusion power plant made newly plausible by magnets that had been impossible a decade earlier. Rather than pursue the idea through federal grants and the slow cadence of academic fusion, the team chose to spin it out as a private company and raise venture capital against a hard commercial goal.
The company was founded in 2018 by a group drawn from the PSFC — CEO Bob Mumgaard, chief science officer Brandon Sorbom, professors Dennis Whyte and Zach Hartwig, former PSFC deputy director Martin Greenwald, and Dan Brunner — with an initial $50 million from the Italian energy major Eni. A $115 million Series A followed in 2019 with Eni, Bill Gates's Breakthrough Energy Ventures, and Khosla Ventures. The decisive moment came in 2021: after demonstrating a record-setting 20-tesla magnet, CFS raised a $1.8 billion Series B — the largest private fusion round to that point — and began building SPARC. A further $863 million Series B2 in 2025 brought total capital raised to roughly $3 billion, backed by investors including Nvidia, Google, Temasek, and Bill Gates. The company now employs around 1,100 people.
The magnet bet: high-temperature superconductors
The technology that makes CFS different is not a new kind of fusion but a new kind of magnet. Conventional superconducting tokamaks, including ITER, use low-temperature superconductors that must be chilled to near absolute zero and that saturate at relatively modest field strengths. CFS instead builds its magnets from high-temperature superconducting (HTS) tape made of rare-earth barium copper oxide (REBCO), a material discovered in the 1980s but only recently manufacturable in wires long and capable enough to wind into a fusion-scale magnet.
REBCO carries enormous current with no resistance while operating at a more forgiving ~20 kelvin, and it tolerates far stronger fields — letting CFS produce a roughly 20-tesla toroidal field, more than double what earlier designs achieved. Because a tokamak's fusion performance scales steeply with magnetic field strength, that jump is what shrinks the whole machine: SPARC targets reactor-class conditions in a device roughly a fortieth the volume of ITER. In 2021 CFS proved the concept was real, testing a 20-tesla HTS magnet and validating the physics case for a compact, high-field path to net energy. The magnets come in three flavors — toroidal-field, poloidal-field, and central-solenoid coils — and the manufacturing know-how behind them has become an asset in its own right.
SPARC: the net-energy demonstration
SPARC is the machine that has to work first. It is a compact, high-field tokamak whose entire purpose is to demonstrate net energy from fusion — peer-reviewed design work projects a fusion gain of roughly Q ~11 and about 140 megawatts of fusion power, which would make SPARC the first magnetic-confinement device built outside a government laboratory to produce more fusion power than it consumes. Achieving that would resolve the single question that has hung over fusion for seventy years: not whether the physics works, but whether a buildable machine can cross into net gain.
Assembly at the Devens campus is now roughly 75% complete and moving fast. At CES 2026 in January, CFS installed the first of SPARC's 18 toroidal-field magnets, each about 24 tons and generating roughly 20 tesla; in May 2026 the team lowered the second 48-ton half of the vacuum vessel into Tokamak Hall, giving the tokamak its final structural shape. All 18 magnets are targeted for installation by the end of summer 2026, ahead of first plasma and a net-energy (Q>1) attempt in 2027. Remaining work includes closing the external cryostat and integrating fuel and diagnostic lines.
ARC: building the market before the machine proves itself
CFS is not waiting for SPARC to finish before commercializing. ARC — its first grid-scale plant, officially named the Fall Line Fusion Power Station — is planned for a 100-acre site at the James River Industrial Center in Chesterfield County, Virginia, on land tied to a partnership with Dominion Energy, and is designed to deliver about 400 megawatts of electricity. The company has moved aggressively to lock in demand ahead of the physics: Google signed a strategic partnership including a 200 MW power purchase agreement, and Eni committed to a more than $1 billion offtake from the first plant.
In April 2026 CFS became the first fusion company ever to file a grid-interconnection request, applying to PJM Interconnection, the largest U.S. wholesale electricity market — a procedural first that signals CFS intends to treat ARC like a real generating asset rather than a science project. Then, on June 4, 2026, the company published a five-paper ARC Physics Basis in a special issue of the Journal of Plasma Physics, the most detailed peer-reviewed case yet for a commercial-scale tokamak, projecting roughly 1.1 GW of fusion power and about 400 MW of continuous net electricity. CFS still targets first electricity from ARC in the early 2030s.
The moat: selling magnets while the reactors get built
CFS's most durable near-term advantage may not be its reactor at all but its magnet factory. Having industrialized HTS magnet production for its own machines, the company is now selling that capability to others, giving it a rare thing in fusion — a path to revenue before any plant switches on. In April 2026 it signed a long-term strategic agreement to design and manufacture HTS magnets for Realta Fusion's magnetic-mirror systems, a deal CFS calls its largest of its kind with potential multi-billion-dollar value over its life. That sits alongside a cable-licensing arrangement with Type One Energy and magnet supply to the University of Wisconsin's WHAM experiment.
The company is also de-risking the harder problems a commercial reactor must eventually solve. It is building an AI-powered digital twin of SPARC with Nvidia and Siemens to speed commissioning, and in July 2026 it became the first international partner in the UK Atomic Energy Authority's £220 million LIBRTI program, gaining access to world-leading infrastructure for testing tritium breeding — one of fusion's biggest unsolved challenges, since a commercial plant must breed its own fuel. Together, the magnet business, the digital twin, and the tritium work turn CFS from a single-shot reactor bet into a broader fusion-industrial platform.
What to watch
The near-term milestones are unusually concrete. First is whether CFS installs all 18 toroidal-field magnets by the end of summer 2026 and closes out SPARC assembly on schedule — the machine is the company's proof of everything else. The defining test comes in 2027: first plasma, and then the attempt to cross Q>1. Success there would be the most important result in the history of private fusion; a serious slip would ripple through a sector that has raised billions largely on CFS's credibility.
On the commercial side, watch the Fall Line Fusion Power Station work through PJM's interconnection study and Virginia permitting, and watch whether the HTS magnet business — anchored by the Realta deal — grows into recurring revenue that funds the roadmap. Fusion remains a field where timelines have historically slipped, and CFS still faces the hardest engineering of all between a net-energy demonstration and a plant that reliably feeds the grid, including tritium breeding and materials that survive years of neutron bombardment. But no private company has ever been closer to answering fusion's central question, and CFS has arranged its money, its magnets, and its customers so that if the physics cooperates, the plant is already half-sold.
Sources
- CFS — ARC Physics Basis published in Journal of Plasma Physics
- MIT News — MIT and CFS launch a novel approach to fusion power
- Wikipedia — Commonwealth Fusion Systems
- CFS — First fusion company to apply to PJM interconnection (ARC / Fall Line)
- CFS — $863M Series B2 round
- TechCrunch — CFS leans on magnets for near-term revenue