Fusion Energy Just Had Its Biggest Year and Almost Nobody Noticed
The National Ignition Facility more than doubled its fusion energy yield. China's EAST tokamak more than doubled its plasma confinement record. Private fusion companies raised a record $4.48 billion. Google and Microsoft both signed power purchase agreements for fusion plants that don't exist yet. All of this happened while AI dominated every tech headline.
Editorial
In April 2025, the National Ignition Facility at Lawrence Livermore National Laboratory fired 192 laser beams at a target the size of a peppercorn and produced 8.6 megajoules of fusion energy from 2.08 megajoules of laser input, according to LLNL. The target gain of 4.13 was nearly three times the result from the facility's first ignition shot in December 2022, which produced 3.15 megajoules. Three months earlier, China's EAST tokamak sustained high-confinement plasma for 1,066 seconds, more than doubling its own record of 403 seconds set in 2023, according to the Chinese Academy of Sciences. In May 2025, Germany's Wendelstein 7-X stellarator set a world record for sustained fusion performance, surpassing marks held by tokamaks that had stood for years, according to EUROfusion.
None of these events generated sustained mainstream coverage. In the same period, AI dominated virtually every technology headline, policy debate, and corporate earnings call. But the Fusion Industry Association's 2026 annual report tells a different story about where serious money is moving. Private fusion companies raised $4.48 billion in the 12 months ending July 2026, the highest annual total since the survey began in 2021 and a 69% increase over the previous year, according to the FIA. Cumulative private investment has reached $14.24 billion. Fifty-six companies now operate in the sector, employing over 16,000 people.
What records were actually broken?
The NIF's April 2025 shot is the headline number. The facility has now achieved ignition at least 11 times since December 2022, and the April result more than doubled the original energy yield, according to LLNL. The breakthrough was enabled by a new capsule fabrication technique using continuous gradient tungsten doping of synthetic diamond, which allowed the fuel to compress more evenly and reach temperatures exceeding 100 million degrees Celsius.
China's EAST tokamak, operated by the Chinese Academy of Sciences, sustained plasma for 1,066 seconds in January 2025, a 2.6x improvement over its 403-second record. Then in January 2026, EAST achieved something physicists had debated for decades: stable plasma densities 1.3 to 1.65 times the Greenwald limit, a theoretical boundary that was believed to constrain how much fuel a tokamak could hold. Since fusion power scales roughly with the square of density, exceeding the Greenwald limit could mean future reactors generate significantly more power than existing models predict, according to a paper published in Science Advances.
France's WEST tokamak sustained plasma for 1,337 seconds at 50 million degrees Celsius in February 2025, surpassing EAST's duration record by 25%, according to CEA-IRFM. South Korea's KSTAR held plasma at 100 million degrees for 48 seconds, up from 31 seconds in 2021, with a goal of 300 seconds by the end of 2026, according to NucNet. Germany's Wendelstein 7-X, a stellarator rather than a tokamak, achieved the highest sustained triple product for long plasma discharges ever recorded, according to EUROfusion.
Who is building fusion reactors right now?
Commonwealth Fusion Systems, based in Massachusetts, has its SPARC tokamak approximately 75% complete. The first of 18 superconducting magnets, each weighing 24 tons and generating a 20-tesla field, was installed in January 2026. All 18 are expected to be in place by the end of summer 2026, according to TechCrunch. SPARC is targeting first plasma in late 2026 and net energy gain by 2027. CFS has raised nearly $3 billion in total, including an $863 million Series B2 in August 2025 with investors including Nvidia, Google, and Breakthrough Energy Ventures.
Helion Energy, backed by OpenAI CEO Sam Altman, is constructing its Orion plant in Malaga, Washington. In June 2026, Helion became the first company in the world to receive regulatory licenses for a fusion power plant, according to GeekWire. The company raised $465 million in a Series G led by Thrive Capital in June 2026, and its Polaris prototype has demonstrated plasma temperatures of 150 million degrees Celsius, according to Helion.
General Fusion, based in British Columbia, began operating its LM26 demonstration machine in early 2025 and achieved compressional plasma heating of approximately 8.4 million degrees in June 2026 using a unique approach that requires no superconducting magnets or high-powered lasers, according to General Fusion. The company is going public via a SPAC merger targeting a Nasdaq listing under the ticker GFUZ. In Europe, Proxima Fusion raised 411 million euros in July 2026, the largest fusion investment on the continent, to build a commercial stellarator power plant, according to the Max Planck Society.
Why are Google and Microsoft buying fusion power they can't use yet?
Microsoft signed the world's first fusion power purchase agreement with Helion in May 2023, agreeing to buy 50 megawatts or more from the Orion plant by 2028, according to CNBC. In June 2025, Google signed a 200-megawatt agreement with Commonwealth Fusion Systems for the planned ARC commercial plant in Chesterfield County, Virginia, the largest corporate fusion offtake deal ever, according to CFS.
The logic is straightforward: AI is consuming electricity faster than the grid can supply it. Data center power consumption is projected to more than double from approximately 448 terawatt-hours in 2025 to nearly 980 terawatt-hours by 2030, according to TTMS. The IEA noted in its 2026 report that power demand is rising faster than the U.S. grid was designed to handle. Hyperscaler capital expenditure exceeded $400 billion in 2025 and is expected to jump another 75% in 2026, according to Morgan Stanley. Fusion offers what solar and wind cannot easily provide at data center scale: continuous, high-density, low-carbon baseload power.
The Fusion Industry Association's 2026 report identifies AI energy demand as a primary driver accelerating fusion development. Google and Microsoft are placing bets on fusion not because the technology is proven, but because they need electricity sources that do not yet exist at the scale they will require by the early 2030s.
What changed about "always 30 years away"?
A review of 45 publications spanning 1985 to 2022, published in the Journal of Fusion Energy, found that 20 years ago scientists estimated commercial fusion was 28.3 years away. That estimate has compressed to 17.8 years. In the FIA's 2026 survey, 71% of fusion companies expect the first plant to deliver commercial electricity by the 2030s, and 89% believe fusion will generate grid power by 2035.
The shift is partly regulatory. On February 26, 2026, the Nuclear Regulatory Commission proposed a rule to regulate commercial fusion under its byproduct material framework rather than as traditional nuclear reactors, according to Axios. This distinction is significant: fusion will not face the same licensing burden as fission plants, potentially cutting years off development timelines. A final rule is targeted for October 2026.
Legislation is following. The Office of Fusion Act, introduced by Senators Alex Padilla and John Cornyn in 2025, would establish a dedicated Office of Fusion within the Department of Energy and set a goal to begin construction on more than one private-sector fusion power plant by December 2028, according to Congress.gov. The Fusion Advanced Manufacturing Parity Act proposes a 25% tax credit for domestic manufacturing of fusion components.
National governments are treating fusion as a strategic race. China spends approximately $1.5 billion annually on fusion research, nearly twice the U.S. federal budget, according to ChinaTalk. In July 2025, Beijing launched China Fusion Energy Co. Ltd. with $2.1 billion in registered capital. The UK committed 2.5 billion pounds over five years for fusion research and development, according to World Nuclear News. Japan accelerated its commercialization timeline from the 2040s to the 2030s.
What are the critics saying?
John Holdren, co-director of Harvard's Belfer Center and former White House science advisor under President Obama, has called predictions of commercial fusion by 2030 or 2035 "hype at this point." The Belfer Center warns that fusion enthusiasm is "dangerous in feeding false hope that there is a technological silver bullet that will save us from challenging problems of addressing climate change," according to the Belfer Center.
The technical skepticism centers on tritium. The global civilian tritium stockpile is only 20 to 30 kilograms, and a single 1-gigawatt fusion plant would require approximately 55 kilograms per year, according to Physics World. Current global production from CANDU fission reactors yields less than 4 kilograms annually, and tritium decays at 5.5% per year. Commercial fusion depends on unproven tritium breeding blankets requiring enriched Lithium-6, which is itself scarce. The DOE published a materials roadmap in June 2026 acknowledging that fusion "hits a materials wall," with engineering gaps in reactor wall materials that must be solved by the 2030s or commercial timelines slip, according to Tech Times.
The industry's own assessment is candid about the gap. Fusion companies told the FIA they need an estimated $77 billion more to build the first generation of commercial plants, roughly eight times the total raised so far. No private company has yet demonstrated sustained net energy gain. Earth911 captured the balance: "Fusion isn't a near-term energy solution. It's not science fiction either, but it's a long-horizon, high-risk, high-reward option with unavoidable uncertainty." The $14.24 billion already invested says the money disagrees with the skeptics on timeline, even if the physics still has work to do.
