FUSION TIME
Nuclear fusion has been the power source of tomorrow for so long it seems more like a cliché than a viable energy technology. But according to many engineers in the field, the future is, in fact, now.
Written by Michael Abrams
Thea Energy is looking to build fusion power plants with the help of an improved stellarator design. Photo: Thea Energy
IS FUSION FINALLY HAVING ITS MOMENT?
A quick scan of headlines from news sources as diverse as Scientific American, CNN, the Washington insider publication The Hill, and video game website GameSpot.com all ran articles in May 2025 on some aspect of nuclear fusion’s journey from long-prophesized power source to actual, existing power plants. In more technical settings, engineers and physicists are announcing breakthroughs in plasma containment, and startups are releasing press statements about their successful funding rounds. Those with a cynical mindset could well dismiss this press and venture capital interest as the next technology bubble, following on the heels of virtual reality goggles or artwork sold in the form of nonfungible tokens.
The last time fusion was featured in the pages of Mechanical Engineering, Jeffrey Winters captured that tension between optimism and cynicism. “Fusion will probably remain a futuristic-sounding technology for the time-being—a byword for something that would be world changing if we could only get it,” Winters wrote in 2019. “But the momentum in the field seems to be building, and money and talent pouring in seems to point to a breakthrough that will catch the cynics by surprise.”
A lot has happened in the six years since that article appeared. Most notably, in December 2022, researchers at Lawrence Livermore Laboratory’s National Ignition Facility managed what so many had thought impossible: They ignited a fusion event that gave off more energy than they’d put in. Since then, they’ve repeated the experiment seven more times, yielding increasingly more energy with every shot. The latest result occurred in April 2025, when they fired two megajoules worth of lasers at their fuel and produced 8.6 megajoules.
“That’s a spectacular result,” said Steve Cowley, a professor of astrophysical sciences at Princeton University, and director of the U.S. Department of Energy’s Princeton Plasma Physics Laboratory there. “They got ignition in a little peppercorn sized pellet. A megajoule is a hand grenade—this is eight and a half hand grenades that went off in there. They put two megajoules of laser light in, got eight megajoules out. Spectacular result.”
A planar coil stellarator. Video: Thea Energy
Eight and a half hand grenades exploding in an instant, as impressive as it may be, is a long, long way from contributing to our grid, to say nothing of a free-energy utopia.
“We’ve done a little bit of fusion,” Cowley said. “But there’s a big difference between doing a little bit of fusion and doing a self-sustained fusion burn in an environment that would make a reactor that would work 24/7.”
Though the results from the National Ignition Facility may be far from giving us limitless electricity, they did open up a firehose of money into the field. When Winters wrote about fusion startups in 2019, there were fewer than a dozen fusion companies developing various methods for building a practical, sustainable fusion power plant. Now there are upwards of 40, and there are billions of dollars flowing their way.
“There's an awful lot of hype behind closed doors with venture capitalists,” Cowley said. “Unlike the public sector program, where everything receives incredible scrutiny by scientists around the world, these private companies are making claims to investors that don't receive that kind of scrutiny.”
Whether we are, in fact, at the dawning of a fusion age remains an open question. The problem with determining whether fusion’s time has finally arrived is contending with fusion time itself. For decades, milestones have been met regularly enough to keep the technology’s promise alive, but the ultimate goal of an operable power plant continues to remain shimmering on the horizon. Given all the activity, engineers in the field contend that the situation is different now. But for a technology where tomorrow—as in “the power source of tomorrow”—can stretch for 70 years or more, now could mean a frustratingly long time from today.
SPARC, a type of Tokamak, is a fusion machine being built at Commonwealth's 60-acre campus in Devens, Mass. This donut-shaped apparatus can recreate the sun’s power here on Earth. Photo: Commonwealth Fusion Systems
A close-up view of Thea Energy's magnet array. Photo: Thea Energy
Proof of the possibility of fusion occurs every day. Starting at sunrise, to be exact. The sun’s core is hot enough—27 million degrees Fahrenheit—and dense enough that pairs of atomic nuclei are continually squashed together until they fuse into a brand-new nucleus. The chain of reactions starts with four hydrogen nuclei (essentially four protons) and results in one helium nucleus (made up of two protons and two neutrons). Thanks to a discrepancy in cosmic accounting, there’s less mass in the helium nucleus than there was in the original four protons, and that lost mass converts to energy to make the sun shine.
The promise that terrestrial fusion reactors would be miniature suns in a bottle is more poetry than prose. The density and pressure of the solar core—160 grams per cubic centimeter crushed by the pressure of 250 billion atmospheres—are unachievable on Earth, and its proton-based fusion reactions are too slow for a power plant. By one estimate, the sun’s power density is about the same as a compost pile.
Instead, physicists and engineers are taking a shortcut, usually starting with a plasma mix of deuterium, an isotope of hydrogen with one neutron, and tritium, an isotope with two neutrons, that provides an easier and more straightforward path to fusion. They are also replacing the crushing pressure of the sun’s core with temperatures of more than 150 million degrees Fahrenheit. Extreme temperature does the work that extreme pressure does in the sun’s core and forces the nuclei to fuse and release excess energy, but that heat also provides a powerful impetus for the plasma to fly apart.
Keeping that extraordinarily hot plasma contained and bringing the nuclei close enough to fuse has been the chief aim of fusion researchers for decades. The type of plasma container that’s had the most success is called a tokamak, which is essentially a giant magnetic bottle in the shape of a donut inside a donut inside a donut. One set of magnets puts a current in the plasma along the length of the donut, another wraps around it to twist and confine the plasma, and a third shapes and positions it. Tokamaks have been shaping plasmas since the first one, the Russian T-1, ran in 1958, and they’ve been making fusion—albeit subcritical—since the Russian T-3 managed it in 1968.
In March, Commonwealth began assembling the SPARC tokamak in Devens, Mass. The first element was a disc-shaped stainless steel cryostat base. Photo: Commonwealth Fusion Systems
Some 100-odd tokamaks have been built since then. And they’ve been colossal. JET, an experimental torus near Oxford, England, which was once the only tokamak to manage deuterium-tritium fusion, had 32 copper-wound coils, each of which weighed 12 tons. Japan’s JT-60SA, described as the largest tokamak in the world, can make a plasma of 160 cubic meters. China’s Experimental Advanced Superconducting Tokamak, or EAST, is 11 meters tall, 8 meters wide, and weighs 400 tons; it set a record in January 2025 for containing a plasma for 17 minutes.
An even more massive tokamak is under construction in southern France. When complete, the International Thermonuclear Experimental Reactor (ITER) will be 12.4 meters across and have the capacity to contain a plasma of some 840 cubic meters. If everything goes as smoothly as the forecast from ITER, it will make its first plasma somewhere in the mid-2030s and might make net positive fusion by the end of that decade.
Thus far, however, not everything has gone smoothly. The cost has already reached $22 billion versus its initial expected budget of $6 billion, and according to the original schedule, ITER should have been completed and achieving fusion already. The cost and construction overruns parallel the experience of the recently completed Vogtle nuclear reactors near Waynesboro, Ga., the end price tag of which has scared off potential investors of building new, traditionally sized reactors.
The parallels run deeper. Just as the nuclear industry has turned to smaller, modular plants to get around the challenge of building gigawatt-scale reactors, so too has the experience of ITER prompted fusion advocates to look at smaller alternatives.
“What we should be really focusing on is using everything we know to find a faster, cheaper route to fusion,” said Princeton’s Cowley. “Fusion is possible, but it’s not clear that it’s possible at a cost the consumer wants to pay.”
Thea researchers successfully tested and controlled a three-by-three array of its planar magnets in March 2025. Photo: Thea Energy
A rendering of Thea Energy's stellarator. Graphic: Thea Energy
One factor that drove up the scale—and cost—of experimental fusion reactors was the magnets used to confine the plasmas. The stronger the magnet, the better the confinement, but extremely strong magnets tended to be huge. Until roughly 15 years ago, the strongest magnets were made of niobium titanium, which required liquid helium to cool them to a temperature where the material became superconducting, but at least they could reach a magnetic field strength of 13 Tesla. (For reference, a powerful rare-earth magnet made of neodymium produces a field of about 0.1 Tesla.) Thanks to developments in high-temperature superconductors, magnets cooled by relatively inexpensive liquid nitrogen can now produce 20-Tesla fields.
Those powerful magnets have helped reshape the thinking on the design of tokamaks.
“In tokamaks, typically, a lot of stuff gets a lot better in the plasma the higher you can go in the magnetic field,” said Alex Creely, director of tokamak operations and chief engineer of ARC Conceptual Design at Commonwealth Fusion Systems. “If you can increase the magnetic field, you can make your whole machine with some combination of more fusion power and smaller.”
Commonwealth is perhaps the best-known fusion startup in the world, having spun out of plasma research at the Massachusetts Institute of Technology in 2018. The Devens, Mass., company subsequently raised around $2 billion to fund its development and demonstration work.
“Our innovation was to say, okay, we have this relatively well-known physics, we have this new material—we’re going to take that material, make it into a superconducting magnet, and then design a new tokamak around it,” Creely said. “It lets you take what would have been a very, very large machine to make net energy in a tokamak and make it much smaller.”
Commonwealth’s goal is not to do anything like basic research, but to make a power plant as quickly as possible.
“The whole point of these science experiments today is to try new crazy stuff—you want to learn stuff, and the way you learn stuff is to push the boundaries in every possible direction,” Creely said. “Whereas, in a power plant, the goal is to be the most boring possible thing. People really like boring power plants that just sit there and make electricity.”
The ARC reactor Creely is working on is named after its key attributes: affordable, robust, and compact. An interim proof-of-concept tokamak, called SPARC—adding “smallest possible” or perhaps “soonest possible” to the acronym—is designed to hold plasmas for 30 seconds and produce 140 megawatts of power, or about as much as it consumes. When the machine was announced in 2018, the completion date was 2025; at present, SPARC won’t be finished until next year, and breakeven production won’t start until the year after that.
While SPARC will be another exciting milestone, ARC is intended to be Creely’s ideal boring power plant from the get-go. It will keep a plasma going for longer, will produce more energy, and will plug right into the grid, providing 400 megawatts of net electricity.
“The plan is to do roughly 15-minute pulses with about a minute between pulses to recharge one of the magnets. You produce electricity continuously, because there’s enough thermal mass in the system that your turbine doesn’t slow down,” Creely said. “And then you just do it back-to-back-to-back-to-back.”
This superconducting planar coil three-by-three magnet array system demonstrated that small and simple electromagnets can create and control stellarator-relevant magnetic fields. Photo: Thea Energy
Tokamaks are arguably the most mature technology in the marathon to create fusion power. They have had documented success in sustaining hot plasmas and fusing nuclei together, albeit with more energy required than is produced. But the plasmas within a tokamak can be fleeting due to the current required to run through the plasma to help it keep its shape.
A different approach—one that also dates to the early 1950s—relies on magnets that wrap around the torus to provide the stabilizing twist. Conceived by famed Princeton physicist Lyman Spitzer, this machine, called a stellarator, needs to have poloidal coils that are nearly as twisty as the plasma they are shaping. The result is something that looks like a broken slinky around a Möbius strip of a rubber band—if it in fact looks like anything else on Earth.
In essence, the current is “sort of cut it up into pieces along the contours of constant current. And that gives you those shapes,” said David Gates, the chief technology officer of Thea Energy, headquartered in Kearny, N.J., which is pinning its hopes of building fusion power plants on improving stellarator design.
In Spitzer’s day, figuring out the ideal shape of those coils was an impossible task. With today’s computational power and theoretical advances in stellarator design, it’s a cinch. “Now we have gotten to the point where we can do that in seconds—minutes, maybe, worst case. And so that allows us to now not just find a solution, but to compare solutions. There’s a huge number of solutions to this problem,” Gates said.
Those solutions, however theoretically perfect, produce coils that are complicated to build and require unforgiving precision. But Gates and his colleagues at Thea have found a way to sidestep the impossible slinky.
“If I have only one set of coils, it has to look like the wiggly coils,” he said. “What I do is I add a second set of coils that face the plasma, and I adjust the currents in all those smaller coils—we call them shaping coils—to get the shape that I wanted in the first place. Now, all of our coils are flat and convex, so we can build them simply. And then, because I now have lots of adjustability in the system, I can change the currents of all the individual coils, I can reduce the precision requirement enormously.”
Two massive stellarators designed before the advent of advanced computational tools show how difficult it was to design these coils without them. Princeton Plasma Physics Laboratory sunk four years and $70 million in its NCSX stellarator that proved to be impossible to build. Germany’s Max Planck Institute for Plasma Physics stuck with its Wendelstein 7-X machine through nearly two decades of construction and one billion euros of spending. (At least it works: In May 2025, Wendelstein 7-X brought a plasma to 54 million degrees Fahrenheit and held it for 43 seconds.)
Gates, who worked at the Princeton Plasma Physics Laboratory during the NCSX years, said these difficulties inspired him to find a simpler way, one that’s faster and cheaper to construct. “It took a minute,” he said. A minute in fusion time is the 10 to 15 years he spent developing his idea.
Google signed a power purchase agreement in June 2025 for 200 megawatts of electricity from Commonwealth’s inaugural ARC power plant in Chesterfield County, Va. The company expects the facility will provide power on the grid in the early 2030s. Photo: Commonwealth Fusion
In March 2025, Thea successfully tested and controlled a three-by-three array of their planar magnets. But there’s much to do before the company attempts any plasma containment. Among other things, their researchers need to design a support structure that can handle the forces created by the magnets, calculate thermal transport and heat sources in cryogenic cooling, make a model of the plasma and calculate the actual fusion it’s going to make, and figure out what to do when a helium nucleus at extreme temperature hits a wall.
And there are more mundane issues: “Plumbing is not a small problem,” Gates said. “You have to design where the pipes go, and where the current leads go, and all the stuff you do when you design something.”
Gates is confident that his ideas will lead to fusion energy. “The physicist in me says, ‘Well, this is trivial,’” he said. “But I was confident that the engineering part of it was solvable. And it has turned out to be.”
In the meantime, Commonwealth is working to get ARC, its first commercial-scale plant, up and running in Chesterfield County, Va., in the early 2030s. If they achieve that milestone, they will have completed a utility-scale fusion power plant before the gargantuan ITER facility in France demonstrates that it is possible.
That, though, would just be the beginning. A first power plant—be it from Commonwealth or another company—will be a breakthrough, no doubt, but won’t change the world by itself.
“You need to build thousands of them,” Commonwealth’s Creely said. “One of our challenges, as a company, but also just as an industry, is that we need to design and build products that you can build many, many, many of, as well as scaling quickly enough to have a real impact.”
Will we finally get to a fusion-powered future of clean and limitless electricity? It may not be a matter of if anymore, but when.
Michael Abrams is a technology writer in Westfield, N.J.
© 2025 The American Society of Mechanical Engineers. All rights reserved.