Ever since nuclear fusion was discovered in the 1930s, scientists have wondered if we could somehow replicate and harness the phenomenon behind starlight—the smashing together of hydrogen atoms to form helium and a stupendous amount of clean energy. Fusing hydrogen would yield 200 million times more energy than simply burning it. Unlike nuclear fission, which powers the world’s 440 atomic reactors, hydrogen fusion produces no harmful radiation, only neutrons that are captured and added back to the reaction. Instead of radioactive wastes with long, lethal half-lives, fusion’s by-product is helium, the most stable element—and a year’s worth from a fusion plant wouldn’t supply a party balloon business.
Dennis Whyte’s part in the fusion quest began in graduate school, in a lab belonging to the electric utility Hydro-Québec, just outside Montreal. There he was shown a device built to replicate stellar fusion on an earthly scale. It was a doughnut-shaped hollow chamber, big enough for a lanky physicist like him to stand inside, based on a design conceived in 1950 by the future Nobel Peace Prize laureate Andrei Sakharov, who also developed hydrogen bombs for the Soviet Union. It was called a tokamak, a word derived from a Russian phrase meaning “ring-shaped chamber with magnetic coils.”
The idea is straightforward: Fill the doughnut with hydrogen gas, and then heat that gas until it turns to electrically charged plasma. In this ionic state, plasma would be held in place by magnets positioned around the tokamak. Achieving fusion on Earth without the immense pressure of a star’s interior, scientists calculated, would require temperatures nearly 10 times hotter than our sun’s center—around 100 million degrees Celsius. So the trick would be to suspend the hot plasma so perfectly in a surrounding magnetic field that it wouldn’t touch inner surfaces of the chamber. Such contact would instantly cool it, stopping the fusion reaction.
The good part about that was safety. In a failure, a fusion power plant wouldn’t melt down—just the opposite. The bad part was that gaseous plasma wasn’t very cooperative—any slight irregularity in the chamber walls could cause destabilizing turbulence. But the concept was so tantalizing that by the mid-1980s, 75 universities and governmental institutes around the world had tokamaks. If anyone could get fusion—the most energy-dense reaction in the universe—to work, the deuterium in a liter of seawater could meet one person’s electricity needs for a year. It would be, effectively, a limitless resource.
Besides turbulence, there were two other big obstacles. The magnets surrounding the plasma needed to be really powerful—meaning really big. In 1986, 35 nations representing half the world’s population—including the US, China, India, Japan, what is now the entire European Union, South Korea, and Russia—agreed to jointly build the International Thermonuclear Experimental Reactor, a $40 billion giant tokamak in southern France. Standing 100 feet tall on a 180-acre site, ITER (the acronym also formed the Latin word for “journey”) is equipped with 18 magnets weighing 360 tons apiece, made from the best superconductors then available. If it works, ITER will produce 500 megawatts of electricity—but not before 2035, if then. It’s still under construction. The second obstacle is the biggest: Many tokamaks have briefly achieved fusion, but doing so always took more energy than they produced.
After earning his doctorate in 1992, Whyte worked on an ITER prototype at San Diego’s National Fusion Facility, taught at the University of Wisconsin, and in 2006 was hired by MIT. By then, he understood how huge the stakes were, and how life-changing commercial-scale fusion energy could be—if it could be sustained, and if it could be produced affordably.
MIT had been trying since 1969. The red brick buildings of its Plasma Science and Fusion Center, where Whyte came to work, had originally housed the National Biscuit Company. PSFC’s sixth tokamak, Alcator C-Mod, built in 1991, was in Nabisco’s old Oreo cookie factory. C-Mod’s magnets were coiled with copper to serve as a conductor (think of how copper wire wrapped around a nail and connected to a battery turns it into an electromagnet). Before C-Mod was finally decommissioned, its magnetic fields, 160,000 times stronger than Earth’s, set the world record for the highest plasma pressure in a tokamak.
As Ohm’s law describes, however, metals like copper have internal resistance, so it could run for only four seconds before overheating—and needed more energy to ignite its fusion reactions than what came out of it. Like the now 160 similar tokamaks around the world, C-Mod was an interesting science experiment but mainly reinforced the joke that fusion energy was 20 years away and always would be.
Each year, Whyte had challenged PhD students in his fusion design classes to conjure something just as compact as C-Mod, one-800th the scale of ITER, that could achieve and sustain fusion—with an energy gain. But in 2013, as he neared 50, he increasingly had doubts. He’d devoted his career to the fusion dream, but unless something radically changed, he feared it wouldn’t happen in his lifetime.
The US Department of Energy decided to scale back on fusion. It informed MIT that funding for Alcator C-Mod would end in 2016. So Whyte decided he would either quit fusion and do something else or try something different to get there faster.
There was a new generation of ceramic “high-temperature” superconductors, not available when ITER’s huge magnets were being wrapped in metallic superconducting cable, which has to be chilled to 4 kelvin above absolute zero (–452.47 °F) for its resistance to current to fall to zero. Discovered accidentally in 1986 in a Swiss lab, the new ceramic superconductors still needed to be cooled to 20 K (–423.7 °F). But with far smaller power requirements, their output was so much greater that a year later the discoverers won a Nobel Prize.
The potential applications were limitless, but because ceramic is so brittle, coiling it around electromagnets wasn’t feasible. Then one day Whyte ran into research scientist Leslie Bromberg ’73, PhD ’77, in the hallway holding a fistful of what resembled unspooled tape from a VCR cassette. “What’s all that?” he asked.
“Superconducting tape, new stuff.” The filmy strips were coated with ceramic crystals of rare-earth barium copper oxide. “It’s called ReBCO,” Bromberg said.
ReBCO’s rare-earth component, yttrium, is 400 times more common than silver. Could superconducting tape, Whyte immediately wondered, be wound like copper wire to make much smaller but far more powerful magnets?
The class met in a windowless room in a former Nabisco cracker factory, surrounded by blackboards.
He assigned his 2013 fusion design class to see. If the students managed to double the strength of a magnetic field surrounding hot plasma, he knew, they might multiply fusion’s power density sixteenfold. They came up with an eye-opening design they called Vulcan. It yielded five peer-reviewed papers—but whether layers of wound ReBCO tape could stand the stress of the current needed to hold plasma suspended while being superheated to ignite a fusion reaction was unknown.
For two years, his classes refined Vulcan. By 2015, with ReBCO more consistent in quality and supply, he challenged his students—11 male and one female, including an Argentine, a Russian, and a Korean—to outdo what 35 nations had been attempting for nearly 30 years.
“Let’s see if ReBCO lets us build a 500-megawatt tokamak—the same as ITER, only way smaller.”
If superconducting tape could let them make a fusion reactor to fit the footprint of a decommissioned coal-fired plant, he told them, it could plug right into existing power lines. To then make enough carbon-free energy to stop pushing Earth’s climate past the edge, its components would have to be mass-producible, so any competent contractor could assemble and service them.
The class met in a windowless room in a former Nabisco cracker factory, surrounded by blackboards. Divided into teams, the students set about figuring out how thin-tape electromagnets could be made robust, and how to capture neutrons expelled from fusion reactions so their heat could be used for turning a turbine—and so they could be harnessed to breed more tritium for the plasma. That’s crucial, because natural tritium is exceedingly rare. Since ReBCO-wrapped magnets would be so much smaller, shrinking the dimensions of one component rippled through everything else. One team’s innovations fed another’s, and parts of the design started to link together. As excitement spread through PSFC, members of earlier classes, now postdocs or faculty members, pitched in. Whyte’s students, some with doctoral dissertations due, were putting in 50-hour weeks on this, reminding him of why he’d dreamed of fusion in first place.
Whyte looked it over for the thousandth time. He was pretty sure they hadn’t broken any laws of physics.
And then, at the semester’s end, out popped their design. Just over 10 feet in diameter, it actually looked like a prototype power plant. While ITER had massive shielding, their tokamak would be wrapped in a compact blanket containing a molten-salt mixture of lithium fluoride and beryllium fluoride to absorb the heat of the neutrons escaping from the fusion reaction. Those neutrons would also react with the lithium to breed more tritium.
The blanket’s heat would be tapped for electricity—except one-fifth of the heat energy would remain in the plasma, meaning the reaction was now heating itself and was self-sustaining, producing more energy than was needed to ignite it. Net fusion energy had been achieved.
The ReBCO magnets, although just a 40th the size of ITER’s, could deliver a magnetic field strength of 23 tesla (a hospital MRI machine typically operates at 1.5 tesla). That was more than enough to achieve a fusion reaction, yet it would require less electricity than its copper-clad C-Mod predecessor by a factor of 2,000. Everything was designed for easy maintenance, and parts could be replaced without having to dismantle the entire reactor.
Most important, the calculated energy output was more than 13 times the input.
Whyte looked it over for the thousandth time. He was pretty sure they hadn’t broken any laws of physics. He calculated the cost per watt and was astonished. Suddenly their goal wasn’t just building a much smaller ITER. It was being commercially competitive.
Stunned, he told his wife, “This can actually work.”
They called it ARC, for “affordable, robust, compact.” “Buildable in a decade,” Whyte predicted.The peer-reviewed article his 12 students published in FusionEngineering and Design estimated it would cost around $5 billion. In 2015, that wasn’t much more than the cost of a comparably sized coal-fired plant, and one-eighth ITER’s price tag.
That May, Whyte gave a keynote about ARC at a fusion engineering symposium in Austin, Texas. Four of his students attended. When he described their plan for a workable reactor by 2025, in just 10 years, conferees were astounded—everyone else was talking decades. Afterward, the MIT contingent went to lunch at Stubb’s Bar-B-Q. It was clear that with the climate eroding and the Intergovernmental Panel on Climate Change warning that yet-uninvented technologies were needed to keep temperatures from soaring into dreaded realms, they had to do this. But since the DOE had pulled its funding, how could they?
On a napkin, Whyte started listing what they’d need to do and what each step might cost. Over ribs, they crafted a proposal to spin off a startup to raise venture capital to finance a SPARC (for “soon-as-possible ARC”) demo fusion reactor to show that this could really happen. Then they’d build a commercial-scale ARC.
Forming a company would free them from academic and government funding cycles, but they were plasma physicists, most still in their 20s, without business backgrounds. Nevertheless, Whyte and Martin Greenwald, deputy director of the PSFC, agreed to join them, and in 2018 Commonwealth Fusion Systems, CFS, was born. Three of his former students would run the company, and three would remain at MIT’s Plasma Science and Fusion Center, which—in a profit-sharing agreement—would be CFS’s research arm.
They opened shop up the street, in The Engine, MIT’s “tough tech” startup incubator, and gained the attention of climate-concerned backers like Bill Gates, George Soros, and Jeff Bezos. But they weren’t the only ones competing for fusion funds, and it became a race to see who could make commercial-scale fusion first.
The CFS team may have been young, but because of its partnership with MIT and its more than a hundred experienced fusion scientists, it had a running start.
By the end of 2021, Commonwealth Fusion Systems had raised more than $2 billion and was breaking ground on 47 acres outside Boston for a commercial fusion energy campus, to build SPARC by 2025—and commercial-scale, mass-producible ARC by 2030.
Gaining and actually sustaining net energy is perpetually called fusion’s yet-unreached “holy grail,” but by September 2021, the CFS team of CEO Bob Mumgaard, SM ’15, PhD ’15 (a coauthor of the Vulcan design), chief science officer Brandon Sorbom, PhD ’17 (lead author of the 2015 fusion design class’s breakthrough paper), Whyte, and their 200 CFS colleagues were confident they could do it—if their magnets held. For three years, straight through the pandemic, they’d worked in PSFC’s West Cell laboratory, the cavernous former Oreo factory that had housed Alcator C-Mod, furiously solving problems like how to solder thin-film ReBCO tape together into a structure strong enough to withstand 40,000 amps passing through it—enough to power a small town.
The completed SPARC would have 18 magnets encircling its plasma chamber, but for this test they’d built just one. It was composed of 16 layers, each a D-shaped, 10-foot-high steel disk grooved like an LP. On one side, the grooves held tight spirals of ReBCO film, 270 kilometers in all—the distance from Boston to Albany. “Yet all that ReBCO holds just a sprinkling of rare earth,” said Sorbom. “That’s the magic of superconductors: A tiny bit of material can carry so much current. By comparison, a wind turbine’s rare-earth neodymium magnets weigh tons.”
On each disk’s flip side, the grooves channeled liquid helium to cool the superconductor for zero resistance. (The design dates to history’s first high-field magnet, built at MIT in the 1930s, which used copper conductors and water for coolant.) Each layer was built on an automated assembly line. “The idea,” said Mumgaard, “is to make 100,000 magnets a year someday. This can’t be a scientific curiosity. This needs to be an energy source.”
Although covid-19 had waned, an outbreak could foil everything, so they maintained coronavirus protocols, moving computer terminals outside beneath a tent to avoid crowding within. Others worked virtually. For a month, dozens worked eight-hour, continuous shifts. Some operated the electromagnetic coil, encased in stainless steel in the middle of the room, which over a week had to be gradually supercooled from room temperature of 298 K down to 20 K before slowly ramping up to full magnetic strength. Others constantly compared real-time data with redundant models. As the temperature dropped, the internal connections, welds, and valves contracted at different rates, so they watched for leaks.
On September 2, 2021, the Thursday before Labor Day, they started ramping up by a few kiloamps, stopping frequently to check what the current was revealing, how the cooling characteristics had changed, and how the stresses on the ReBCO coil increased as the magnetic field strengthened to record heights.
Two nights later, they cranked the amperage toward their goal: a 20-tesla magnetic field, powerful enough to lift 421 Boeing 747s or contain a continuous fusion reaction. They’d been aiming for 7:00 a.m. on Sunday, the 5th. At 3:30, the large screen in the design center showed that they’d reached 40 kiloamps, and the magnetic field had reached 19.56 tesla.
At 4:30 a.m., they were at 19.98 tesla. Things got very quiet. At 5:20 a.m., every redundant on-screen meter read 20 tesla, and nothing had leaked or exploded—except under the tent, where champagne corks were popping.
Five years earlier, on its final four-second run, C-Mod’s copper-conducting magnet had consumed 200 million watts of energy to reach 5.7 tesla. This took 30 watts—less energy by a factor of around 10 million, Whyte told reporters—to produce a magnetic field strong enough to sustain a fusion reaction. The joints that transferred current from one layer to the next actually performed better than expected. That was the biggest unknown, because there was only one way to test them: in the magnet itself. They looked spectacular.
After five hours, the team ramped down the power. “It’s a Kitty Hawk moment,” Mumgaard said.
Adapted from Hope Dies Last: Visionary People Across the World, Fighting to Find Us a Future by Alan Weisman, published by Dutton, an imprint of Penguin Random House. © 2025 by Alan Weisman.
About The Author
