Nuclear fusion has spent seventy years as the energy source that's always thirty years away. A Boston area startup just convinced pension funds to bet four billion dollars that this time is different.
Inside a factory in Devens, Massachusetts, workers are assembling a doughnut shaped machine called SPARC, a demonstration fusion reactor that's now roughly eighty percent built.
SPARC is a spinout of MIT research, built around high temperature superconducting magnets developed with the university. The reactor uses eighteen of these magnets, each weighing about 24 tons and generating a 20 tesla magnetic field, more than enough to bend hydrogen isotopes into the kind of collision that powers stars. The company started installing them in January 2026 and expected all eighteen in place by the end of summer.
On July 30, Commonwealth Fusion Systems announced it had raised another billion dollars, bringing its total to four billion since 2018. For the first time, some of that money came from pension funds and sovereign wealth funds, not just tech billionaires and venture capital.
Only two backers agreed to be named publicly, Google and the Italian energy company Eni. The rest, described only as significant institutional investors, infrastructure funds, and industrial corporate partners, represent a kind of money that almost never touches unproven physics. Company leadership called it a deliberate signal. "Fusion takes machines, it takes capital, it takes building things," CEO Bob Mumgaard said after the announcement, adding that a billion dollars is an important marker, but that the industry will ultimately need far more than that to do fusion seriously. Total capital raised since 2018 now represents roughly 30 percent of all private fusion investment worldwide. Earlier rounds also included Nvidia and Breakthrough Energy Ventures. See Canary Media's coverage of the round.
Cumulative capital raised by Commonwealth Fusion Systems since 2018, culminating in the July 30, 2026 round that brought the total to four billion dollars. Named investors include Google and Eni.
SPARC is expected to achieve first plasma later this year, and net energy gain in 2027, producing more power from fusion than it takes to run the machine. No private company has ever crossed that line.
Scientists call this threshold breakeven, and until recently only a government lab, the National Ignition Facility in California, had ever managed it, in a 2022 experiment that took far more setup energy than its instruments measured. Commonwealth's aim is different, a continuously operating magnetic system rather than a single laser pulse, which is part of why the company treats 2027 as a genuine engineering deadline rather than a talking point. It's also why the claim invites real scrutiny.
Not everyone with a fusion background is convinced the timeline is real. Harvard's John Holdren, who has studied fusion physics since the 1960s and served as the White House's top science advisor, has warned that over hyping fusion is misallocating serious money toward ventures he believes have no realistic chance of near term success.
Holdren's specific worry isn't the physics, which he considers real progress, it's the gap between a working demonstration and an actual power plant that's cheap enough to matter, a jump the industry has underestimated before. ITER, the government led international fusion project in France, was supposed to be finished in 2016 and is now targeting full operation more than two decades past that date, a cautionary tale Commonwealth's own backers are well aware of. Read Holdren's full argument at the Belfer Center.
"Fusion takes machines, it takes capital, it takes building things." A billion dollars is an important marker, but the industry will need far more to do fusion seriously.
The physics is real progress, but the gap between a demonstration and a power plant cheap enough to matter is one the industry has underestimated before, as ITER's decades of delay show.
Commonwealth's next machine, a commercial plant called ARC, is planned for Chesterfield County, Virginia. It won't produce a single watt until the early 2030s. Google and Eni have already agreed to buy more than half of its output.
In April, Commonwealth became the first fusion company in history to file a grid interconnection application with PJM, the mid Atlantic's power market operator, a bureaucratic step that only makes sense for a company that genuinely expects to be selling electricity within the decade. Mumgaard has said the current round of institutional money could open the door to funding a string of ARC style plants beyond the first one, though he's also acknowledged the industry may eventually need as much as ten billion dollars in total to get fusion built at real scale.
Target grid connection: early 2030s.
Commonwealth isn't fusion's only bet. A separate company, Type One Energy, is building its own 400 megawatt project at a former coal plant in Tennessee, with the Tennessee Valley Authority already lined up as the buyer. Increasingly, people in the industry describe the whole race as one the United States can't afford to lose to China.
Whether any of these companies actually deliver commercial fusion on schedule remains an open question, one that has burned optimists before. What's changed is who's willing to find out. Pension funds and sovereign wealth funds don't chase moonshots for the story, they chase them when the risk finally looks survivable. Four billion dollars says a piece of the financial world now believes that moment has arrived.