Nuclear fusion transitions from grand state project to venture capital industry

In three months, between June and September 2025, global private funding for nuclear fusion rose from 9.9 to 13 billion euros (approximately 11.6 to 15.2 billion dollars), according to the F4E Fusion Observatory. The cumulative funding as of September 2025 represents more than 8 times the total cumulative amount recorded in 2020. This is not a trend. It is a turning point.

Nuclear fusion has long been the promise that science made to the world and has never kept. For fifty years, it embodied the grand public project that was slow, expensive, and always a decade away — the energy of the future, and always the future. This model is being replaced, not by a sudden scientific breakthrough, but by a change in structure: venture capital has taken over from states, dozens of private companies are competing for the market, and the first commercial pilot plants are targeted between 2030 and 2035. Microsoft, Google, and Eni have signed power purchase agreements for energy that does not yet exist. The political stakes are immense: how do Western states organize their energy policy in the face of a China that invests 1.5 billion dollars per year in the public fusion sector, in parallel with this private rush?

The essentials

  • Cumulative private funding for fusion rose from 9.9 to 13 billion euros (approximately 11.6 to 15.2 billion dollars) between June and September 2025, according to the F4E Fusion Observatory (December 2025), representing more than 8 times the cumulative total of 2020
  • 77 private companies identified by the F4E Fusion Observatory are developing fusion projects worldwide, targeting commercial pilot plants between 2030 and 2035
  • China invests 1.5 billion dollars per year in fusion through the public sector, alongside the Western private movement
  • The financing model draws directly from SpaceX: milestones paid upon delivery, no open-ended subsidies
  • Microsoft, Google, and Eni have signed power purchase agreements for fusion energy — for energy not yet produced commercially

ITER symbolized a world that no longer exists

To understand the scale of the change, we must return to ITER. The international project launched in the 1980s and under construction since 2010 at Cadarache represents everything the new model rejects: full public funding, governance by consensus among 35 countries, successively delayed timelines, costs that have risen from an initial estimate of 5 billion euros to more than 20 billion, and an objective that remains merely to demonstrate scientific feasibility — not to produce commercial electricity.

ITER is not a failure. It is a scientific machine of unparalleled complexity, designed to prove that fusion can produce ten times more energy than it consumes. But its logic is that of a space program from the 1960s: states that pool risks, political timelines, a cumbersome international organization. This model produced irreplaceable fundamental knowledge. It does not produce an industry.

What has been happening since 2021 is of a different nature. Fusion is no longer trying to convince states that it is possible. It is trying to convince investors that it is profitable before its competitors do. This shift changes everything.

The SpaceX model applied to plasma physics

The reference is explicit in the sector. The model that private fusion companies claim, and that the U.S. government has adopted in its procurement processes, draws directly from NASA’s COTS (Commercial Orbital Transportation Services) program: technical milestones paid upon delivery, not open-ended subsidies. SpaceX, then Rocket Lab, then Blue Origin demonstrated that this system could reduce the cost of reaching orbit by a factor of ten in fifteen years. The analogy with Ariane 6 is enlightening: Europe maintained a national program model where the United States switched to competition subsidized by milestones.

The American MILESTONE program, launched by the Department of Energy in 2022, operates on exactly this principle. Commonwealth Fusion Systems, TAE Technologies, and several other companies have received contracts conditional on precise technical milestones. If the demonstration is successful, the money follows. If not, it does not.

Commonwealth Fusion Systems embodies the typical trajectory. Founded in 2018 by MIT researchers, the company raised 1.8 billion dollars to build SPARC, a compact tokamak using high-temperature superconducting magnets. In 2021, it demonstrated a magnetic field of 20 teslas with REBCO-based magnets (barium oxide and copper doped with rare earths), a world record and a technical milestone judged decisive by the scientific community. The next step is a demonstration of net energy gain by 2027, before a commercial pilot plant targeted for 2035.

What distinguishes this approach from the ITER model is not the technology — it is the pace. A two to three-year iteration cycle versus ten to fifteen years for a public program. And a clear decision horizon: if milestones are not met, investors exit.

Dozens of companies, radically different approaches

One of the most counterintuitive pieces of data in the sector is its technological diversity. The F4E Fusion Observatory counts 77 private fusion companies worldwide — a broader scope than that of the Fusion Industry Association, which identified 53 in its 2025 report, and which includes notably Chinese players. These companies do not all do the same thing. There are compact tokamaks (Commonwealth Fusion Systems, Tokamak Energy), inertial confinement devices (Marvel Fusion, Focused Energy), hybrid approaches (Helion, TAE Technologies), and alternative magnetic configurations such as stellarators and reversed-field configurations.

This diversity is precisely what the public model could not finance. A government program concentrates resources on the approach deemed most promising. Venture capital, on the other hand, bets simultaneously on several horses. The result is an accelerated experimentation ecosystem that recalls, in its structure, what Philippe Aghion calls “creative destruction applied to major technological challenges”: multiple approaches in competition, the most successful capturing subsequent financing, others disappearing.

Helion Energy has perhaps taken this logic the furthest. The company raised 500 million dollars in 2021 (Series E), led by Sam Altman who personally invested 375 million dollars, then signed an electricity purchase agreement with Microsoft in 2023 to deliver at least 50 megawatts of fusion energy by 2028. This agreement includes a financial penalty clause if Helion does not deliver. Microsoft is buying a promise with a penalty attached. Commonwealth Fusion Systems, for its part, signed a 200-megawatt power purchase contract with Google in June 2025, as well as a 1 billion dollar agreement with Eni in September 2025. Google is also an equity investor in TAE Technologies. Eni has taken a stake in Commonwealth Fusion Systems.

These contracts reveal a new industrial logic. Major energy buyers no longer simply wait for technology to be ready to negotiate prices. They enter into financing technological risk in exchange for priority access and guaranteed pricing. This is exactly what major data center operators do with conventional nuclear energy and renewables: secure supply before capacity exists.

China plays a different tune

While Western venture capital structures a private industry, China pursues a radically opposite strategy. The country invests approximately 1.5 billion dollars per year in fusion through the public sector, according to IEA State of Energy Innovation 2026 data. This investment is concentrated on two main programs: CFETR (China Fusion Engineering Test Reactor), a large tokamak whose construction is expected to begin in the 2030s, and EAST (Experimental Advanced Superconducting Tokamak), which has set several records for plasma confinement duration.

The logic is that of a strategic state program, consistent with Chinese industrial doctrine on critical technologies. China is not trying to beat Commonwealth Fusion Systems on the markets. It is trying to master fusion technology in a sovereign manner, independent of Western licenses and supply chains.

The paradox is obvious. The West is privatizing its fusion research at the precise moment when China is further nationalizing it. The two models could converge toward commercial plants at similar horizons. But the geopolitical implications are distinct: fusion commercialized by American and European companies remains subject to market rules, intellectual property rights, and the decisions of private investors. Fusion commercialized by the Chinese state obeys national supply priorities and strategic power considerations.

This divide echoes what occurred in semiconductors, batteries, or modular reactors. China commercialized the world’s first modular nuclear reactor precisely because it maintained a coherent state program over twenty years, whereas Western countries fragmented their efforts. The question for fusion is not whether the private model is theoretically superior to the public model. It is whether coordination between private investors and Western public policies is sufficient to maintain a technological lead against a competitor that spends continuously and according to plan.

What states are doing — and what they are not yet doing

The response of Western governments is uneven. The United States has the most coherent framework. The Department of Energy’s MILESTONE program, ARPA-E funding, and a bipartisan bill on commercial fusion (the FUSION Act) outline an ecosystem where the state plays a catalyst role: it defines milestones, it pays upon delivery, it coordinates regulation. The Nuclear Regulatory Commission has been working since 2022 on a regulatory framework specific to fusion, distinct from that of fission, which is a necessary condition for authorizing the construction of pilot plants.

Europe is lagging on this point. The relationship between ITER, Fusion for Energy (F4E), and private European companies like Tokamak Energy or Marvel Fusion remains complex. F4E, the European agency managing Euratom’s participation in ITER, has developed a Fusion Observatory to track the private ecosystem, but a financing and regulatory framework comparable to the American MILESTONE program does not yet exist at the European level. Several member states, including the United Kingdom with its STEP program (Spherical Tokamak for Energy Production) and France, have launched national initiatives, but without the coordination and critical mass of the American framework.

This regulatory lag is probably the most underestimated obstacle. A fusion pilot plant producing electricity in 2032 in the United States can only be connected to the European grid if regulation is in place. And building a regulatory framework takes time — as much time as building the plant.

2030 as a moment of truth

The commercial pilot plants targeted between 2030 and 2035 are not marketing objectives. They correspond to contractual commitments for several companies. Helion has promised 50 megawatts to Microsoft by 2028. Commonwealth Fusion Systems targets an ARC plant of 400 megawatts by 2035. Tokamak Energy targets a net gain demonstration by 2026.

These milestones will create clarity where there is none yet. The private fusion sector currently rests on anticipated confidence: electricity buyers sign contracts, investors finance funding rounds, governments allocate programs. All of this before a single commercial plant has produced a kilowatt-hour. If the first milestones are met in the next two or three years, acceleration will be massive. If several major companies miss their deadlines, market sentiment could reverse quickly.

This is precisely the difference from the ITER model: the public model absorbed delays in a political calendar. The private model, on the other hand, has creditors. Investors who have put billions on the table expect a return. This pressure is both the strength of the model — it accelerates execution — and its main risk. A wave of synchronized disappointments between 2027 and 2030 could dry up funding at the worst moment, when technical demonstrations would be closest.

The real question for energy policies is therefore not whether fusion will replace renewables or conventional nuclear in the energy mix of 2050. It is to determine what structure of public support is necessary to absorb the inevitable shocks of a sector in demonstration phase, without falling back into the open-ended subsidy model that ITER represents. NASA’s COTS program failed several times. SpaceX nearly disappeared in 2008. It was the milestone mechanism, not the absence of risk, that allowed it to hold.

Nuclear fusion does not need states to believe in it. It needs them to organize regulation, absorb the risks of initial demonstration, and coordinate power purchase policies. This work is underway in the United States. It remains to be done in Europe.

Sources

  1. Fusion Industry Association / IEA State of Energy Innovation 2026 — https://www.fusionindustryassociation.org/iea-features-fusion-in-state-of-energy-innovation-2026-report/
  2. Fusion for Energy — Fusion Observatory (F4E, European Commission)
  3. U.S. Department of Energy — MILESTONE Program (Office of Science, Fusion Energy Sciences)
  4. Nuclear Regulatory Commission — fusion regulatory framework (NRC, United States)
  5. Commonwealth Fusion Systems — SPARC technical data and fundraising
  6. Helion Energy — Microsoft power purchase agreement, 2023
  7. IEA — State of Energy Innovation 2026
  8. FIA Global Fusion Industry Report 2025 (July 2025)
  9. F4E Fusion Observatory – Global Investment in the Private Fusion Sector (December 2025)
  10. CFS – Series B (1.8 billion dollars, November 2021)
  11. CFS – Series B2 (863 million dollars, August 2025)
  12. MIT News – 20 tesla REBCO magnet demonstration (September 2021)
  13. Helion Energy – Microsoft PPA (May 2023)
  14. CFS + Eni – 1 billion dollar PPA (September 2025)
  15. CFS + Google – 200 MW PPA (June 2025)
  16. Helion Energy – Series E 500 million dollar funding (November 2021)
  17. IEA State of Energy Innovation 2026 (February 2026)
  18. ITER – initial cost and budget overruns
  19. SCSP – China’s $6.5 Billion Fusion Buildout (September 2025)