The quest to build a star on Earth

The quest for fusion energy — the clean, potentially limitless source that could end mankind’s power woes — began as an answer to an old question, one we’ve been asking since we first raised our heads toward the sky.

It was the mid-19th century. Charles Darwin’s theory of natural selection had upended our notions of ourselves and our world. But the theory had a problem. How, physicist Lord Kelvin asked, could the sun have been shining for so long? Wouldn’t it have burned through its fuel well before humans had evolved as Darwin proposed?

Neither man lived to learn the astonishing answer: that inside our sun lighter elements are constantly fusing into heavier ones, liberating vast amounts of energy in the process.

“The store is well-nigh inexhaustible,” astronomer Arthur Eddington wrote in 1920, “if only it could be tapped.”

A century later, a handful of startups say we are closer than ever to making it happen.

In the next few years, these companies say, their fusion machines will produce more energy than they take to run. Soon after, they will start generating electricity for factories, data centers, steel mills and more, helping humanity take a decisive step away from fossil fuels, away from global warming and air pollution, away from powering our lives by setting tiny fires in engines and boilers and furnaces.

Big-name investors, including Bill Gates, Jeff Bezos, Vinod Khosla and Sam Altman, have staked hundreds of millions of dollars on this, fusion’s potential Kitty Hawk moment: the one that shows that the limits of our species’s mastery have once again been catapulted forward.

Today’s fusion startups aren’t just preparing for this moment in the lab. They are signing presale deals with customers, developing supply chains, cultivating a workforce, talking with regulators — all the elements that will be needed to make fusion an affordable, practical power source, not just a science experiment.

And yet, closer than ever does not necessarily mean close. Fusion’s history is a graveyard of missed deadlines and thwarted milestones, bursts of excitement followed by bruising disappointments.

The sunny view is that the startups are moving more quickly than government labs ever could. They can try, fail and try again. But the lesson from more than half a century of fusion research, said Gerald Navratil, a professor of plasma physics at Columbia University, is that failure happens in ways nobody anticipates.

“Even if the physics idea sounds tantalizing,” Navratil said, “until you actually do it for real,” in a real machine, producing real energy, “it’s just an idea.”

How hard can it be?

Creating a working star on Earth might sound flat-out impossible, had scientists not already gone so far toward doing it.

First you need to heat a puff of gas to unimaginable temperatures, over 100 million degrees Celsius. This makes the gas so hot that the electrons are ripped free from their atoms. So hot that the gas transcends gas and enters another state of matter: plasma.

With enough heat, the atoms start to fuse, something they’re extremely not eager to do. Make your plasma hold onto this heat for long enough, and at high enough pressure, and more energy comes out than you put in to heat it up.

Fusion is the opposite of the fission process that powers today’s nuclear plants. Atoms don’t split; they weld together. The basic fuel isn’t uranium, but hydrogen extracted from seawater. There’s no threat of runaway reactions, and the radioactive waste it leaves behind is less dangerous. Making it happen, and controlling it, is just much, much trickier.

“For fission, if you just pile the right kind of material in one place, it’ll get hot,” said Robert Goldston, a Princeton University professor of astrophysical sciences. “For fusion, it’s a different story.”

Once you’ve created some plasma, what’s next? The stuff wriggles and squirms like a snake of superhot Jell-O, so you have to hold it steady, otherwise it could whip out and melt your equipment. Or it might just fall apart, for as violent as the plasma is, it is also fragile: You could snuff it out by blowing on it.

Inside the sun, gravity holds the plasma together. On Earth, people use superstrong magnets or lasers.

By this point maybe you’ve done it: Atoms are fusing, high-energy particles are blasting out of the plasma. Your machine has to survive the pummeling. But it also has to put the energy to work, producing electricity, keeping the reaction going, all without disturbing your plasma, which is as precarious as a toy top spinning on a fingertip.

Racing toward the 2030s

Step inside Commonwealth Fusion Systems’ enormous new building in the Massachusetts countryside and it could be a construction site like any other: bare gray floors, taped-up plastic sheets, spiders in the corners.

Keep going, though, and you hit the wall: 8 feet of concrete, wrapped around the building’s innermost sanctum, protecting the wider world from what’s inside.

Here, in a room as airy and grand as a temple, a colossal machine will soon be placed at the altar. In a circle around its core will sit 18 giant magnets, each powerful enough to hoist an aircraft carrier. When the machine is turned on, the magnetic forces within will be as strong as 10 heavy rockets lifting off from Earth.

Only in the fusion industry would this be considered a compact machine, yet that’s what it is: a small but souped-up version of a tokamak, the doughnut-shaped fusion device that scientists have built scores of since the 1960s. (“Tokamak” is a Russian acronym.)

If there’s a big fish in the commercial fusion pond, Commonwealth is it. Since its founding in 2018, the company has raised over $2 billion, more than any other fusion startup.

The machine it’s building in Massachusetts, SPARC, is a demonstration device. Commonwealth is aiming for SPARC to produce net energy, in what it calls a commercially relevant way, in 2027. Its next machine, ARC, is the one it says will generate electricity for paying customers, in the early 2030s.

(Not coincidentally, ARC is also the name of the fusion device built by Tony Stark, the hero of the “Iron Man” comics. Among Commonwealth’s investors is a firm co-founded by Robert Downey Jr., who played Stark onscreen.)

One reason Commonwealth is confident about hitting its goals is that, in certain ways, SPARC is a “conservative” machine, even a “dull” one, said Bob Mumgaard, the company’s chief executive. Scientists have been studying and building tokamaks for so long that Commonwealth doesn’t need to reinvent the wheel except in a few key areas, Mumgaard said.

Commonwealth is putting most of its creative energy toward ARC, the machine it wants to build next. “We’re still very much learning on the steep part of the learning curve on that,” Mumgaard said.

The company’s scientists and engineers are still figuring out how to make ARC’s plasma less likely to thrash around, and how to keep parts of the machine from overheating. They are also examining how well the materials they’re using can withstand high-energy particles whipping through them, and whether they’ll need to be supplemented with materials that have yet to be invented.

More ‘shots on goal’

The knock against tokamaks, as Mumgaard acknowledges, is that they’re complicated gizmos.

Difficult to build. A pain to take apart and maintain. And expensive: According to Mumgaard, SPARC will end up costing around $1.2 billion to construct.

The biggest tokamak being built anywhere on Earth, a multinational project in France called ITER, is on track to cost tens of billions of dollars and won’t be ready for experiments until the mid-2030s.

That’s where the other side of the private fusion boom comes in. Most of today’s startups aren’t following ITER and Commonwealth and building tokamaks. They think they can do fusion more cheaply and easily using other types of machines, even if their designs vary a lot in how well scientists have gotten them to work.

“They have not been proven yet,” said Earl Marmar, a physicist at the Massachusetts Institute of Technology. “But, you know, good luck. I hope something works soon for sure.”

Type One Energy and Thea Energy are working on stellarators, which are similar to tokamaks but twisted and complexly rippled, like a doughnut as imagined by Salvador Dalí. Realta Fusion is building a reactor that the company’s co-founder, Cary Forest, calls “Tootsie Roll shaped”: a cylinder with magnets at both ends.

In an office park near Seattle, Zap Energy is making fusion devices in which filaments of plasma are, yes, zapped with electricity. Less than a mile away, Helion Energy is working on a fusion machine that shoots two rings of plasma at each other. Helion says it will start using its technology to generate electricity for Microsoft in 2028.

Across the Canadian border, near Vancouver, a company called General Fusion is aiming to squish plasma together not with fancy magnets or lasers or other exotic parts, but with pistons a bit like the ones in a car engine. The company is hoping to demonstrate the feasibility of its new machine in 2026.

“It’s kind of the Wild West right now,” said Richard Magee, vice president of physics research at the fusion company TAE Technologies, as he showed off the firm’s bus-size test reactor in Southern California. “It’s going to be really interesting to see who’s still standing in 10 years.”

When it comes to the big goal of moving humanity into a fusion-powered age, more companies could mean more “shots on goal,” as Jean Paul Allain, head of the Department of Energy’s fusion science program, puts it. When anyone scores, everybody benefits.

What worries researchers is how much some fusion startups are promising, and how soon. Even if their pilot plants are successful, there’s still a ways to go before those will be ready to meet a serious share of the globe’s electricity needs, said Steven Cowley, director of the Princeton Plasma Physics Laboratory.

“There’s an awful lot of supercharged hype,” Cowley said. “You worry about the consequences when people don’t deliver.”

This article originally appeared in The New York Times.

© 2024 The New York Times Company