The idea first lit up Dennis Whyte when he was in high school, in the remote reaches of Saskatchewan, Canada, in the 1980s. He wrote a term paper on how scientists were trying to harness fusion (the physical effect that fuels the stars) in wondrously efficient power plants on Earth. This is the ultimate clean-energy dream. It would provide massive amounts of clean electricity, with no greenhouse gases or air pollution. It would do it on a constant basis, unlike solar and wind. Whatever waste it created would be easily manageable, unlike today’s nuclear power plants. And fuel would be limitless. One of the main ingredients needed for fusion is abundant in water. Just one little gram of hydrogen fuel for a fusion reactor would provide as much power as 10 tons of coal.

Whyte got an A on that paper, but his physics teacher also wrote: “It’s too complicated.” That comment, Whyte says with a hearty laugh, “was sort of a harbinger of things to come.”

Indeed, over the next few decades, as Whyte mastered the finicky physics that fusion power would require and became a professor at MIT, the concept seemingly got no closer to becoming reality. It’s not that the science was shaky: It’s that reliably bottling up miniature stars, inside complex machines on Earth, demands otherworldly amounts of patience, not to mention billions and billions of dollars. Researchers, like Whyte, knew all too well the sardonic joke about their work: fusion is the energy source of the future, and it always will be.

That line took on an especially bitter edge one day in 2012, when the U.S. Department of Energy announced it would eliminate funding for MIT’s experimental fusion reactor. Whyte was angry about the suddenness of the news. “It was absolutely absurd — you can put that in your article — fucking absurd that happened with a program that was acknowledged to be excellent.” But above all, he was dismayed. Global warming was bearing down year after year, yet this idea that could save civilization was losing what little momentum it had.

Wendelstein 7-X fusion reactor in Germany, 2017. Photo: Picture Alliance/Getty

So Whyte thought about giving up. He looked for other things to focus on, “stuff that wasn’t as exciting, quite frankly,” but stuff that would be achievable. “Everyone understands delays in projects, and science hurdles you’ve got to overcome, but I saw fusion energy being used for something accelerating away from us,” he says. “You start getting pretty dejected when you realize, in your professional career, you’re never going to see this happen.”

As it turned out, Whyte never really walked away. Instead, he and his colleagues and graduate students at MIT’s Plasma Science and Fusion Center figured out a new angle. And last winter, MIT declared that Whyte’s lab had a fundamentally new approach to fusion and threw its weight behind their plan with an unusually public bet, spinning out a company to capitalize on it. An Italian oil company and private investors — including a firm funded by Bill Gates and Jeff Bezos — put at least $75 million into the company, known as Commonwealth Fusion Systems [CFS]. The startup intends to demonstrate the workings of fusion power by 2025.

The recent progress is remarkable, says the founder of one startup developing fusion power. “The world has been waiting for fusion for a long time.”

Real, live, economically viable power plants could then follow in the 2030s. No joke. When I ask Whyte, who is 54, to compare his level of optimism now to any other point in his career, he says, simply: “It is at the maximum.”

But it’s not just MIT. At least 10 other startups also are trying new approaches to fusion power. All of them contend that it’s no longer a tantalizingly tricky science experiment, and is becoming a matter of engineering. If even just one of these ventures can pull it off, the energy source of the future is closer than it seems.

“It’s remarkable,” says David Kingham, executive vice chairman of Tokamak Energy, a British company whose goal is to put fusion power on the grid by 2030. “The world has been waiting for fusion for a long time.”

Imagine that I told you I was developing a special machine. If I put power into it, I could get 10 times as much out. Because of the undeniable laws of physics, I could show you on paper exactly why it should be a cost-effective source of vast amounts of electricity.

Oh, here’s the catch: My paper sketch would come true — especially the part about it being cost-effective — but only if I built the machine just right. Which might require materials that haven’t been invented yet. Until I perfected that design, my machine would use up more power than it produced. And I couldn’t get close to perfecting the design without spending years and years building expensive test machines that would reveal problems that I would try to address in subsequent versions.

If it seems crazy, well, that’s the story of fusion power.

Fusion definitely works. You see it every day. Our sun and other stars blast hydrogen atoms together with such intense force that their nuclei overcome their normal inclination to repel each other. Instead they fuse, sparking a reaction that transforms the hydrogen into helium and releases cosmic amounts of energy in the process.

We also have great paper sketches for fusion power machines. Fusion happens inside stars because of the crushing pressure created by their gravity. To generate that effect inside a fusion reactor, ionized gas — which is called plasma — must be heated and compressed by man-made forces, such as an ultra-powerful magnetic field. But whatever the method, there’s just one main goal. If you get enough plasma to stay hot enough for long enough, then you can trigger so much fusion inside it that a huge multiplier effect is unlocked. At that point, the energy that is released helps keep the plasma hot, extending the reaction. And there still is plenty of energy left over to turn into electricity.

The problem is that we’re still plugging away on predecessors to the machines that could generate that effect. Ever since the 1950s, scientists have used spherical or doughnut-shaped machines called tokamaks, including the one at MIT that lost funding a few years ago, to create fusion reactions in plasmas bottled up by magnetic fields. But no one has done it long enough — while also getting it hot enough and dense enough — to really tip the balance and get it going. Heating the plasma and squeezing it in place still takes more energy than you can harvest from it.

So, that’s the name of the game in fusion: to get past that point. ITER, a mega-billion-dollar reactor being built in France by an international consortium, is designed to do it and finally prove the concept. But ITER — which is also way behind schedule and over budget — overcomes the limitations of previous tokamaks by being enormous. It’s the size of 60 soccer fields, which probably isn’t an economical setup for power plants that the world will need by the tens of thousands.

ITER (International Thermonuclear Experimental Reactor) under construction. Photo: Christophe Simon/Getty

Could you go the other direction, and instead make fusion machines much smaller, which is also to say much less expensive? That is what motivates all the fusion startups. Several have decided the answer is to use something other than a tokamak and its circular coils of magnets. They’re updating old designs, including hitting plasma with lasers, or cooking up new ones, such as compressing it with something like a particle accelerator. One startup plans to push on the material with pistons.

But Whyte and his colleagues at MIT made a different decision, one that could prove crucial to making fusion power arise sooner than people expect. Even though things looked dire a few years ago, when their fusion machine lost funding, Whyte’s team decided to double down on tokamaks. As Whyte saw it, why try to invent something totally new when you could take advantage of all those decades and billions spent researching tokamaks? Instead, they would rethink the design to make tokamaks modular and much cheaper and weave in brand-new materials that can induce and confine a fusion reaction.

After getting the news of the funding shutdown, the university, and other supporters of the program, persuaded Congress to grant a temporary reprieve. They could keep running their fusion reactor into 2016, enough time for experiments to be finished and to keep PhD students going on the research they had come to MIT to undertake. And then they dug in.

The most intriguing questions Whyte and his students were exploring had to do with how tokamaks could produce lots of electricity without being gigantic and expensive. MIT’s tokamak, which still sits in a two-story tall, garage-like room in a former Nabisco cookie warehouse, generated a magnetic field by running electricity through copper coils that surrounded a round metal chamber. In that chamber, plasma would be heated with microwaves and other methods to millions of degrees. On one of its last runs, it set a new record for plasma pressure while hitting 35 million Celsius.

Just outside the chamber, the vital measurement isn’t heat, but cold. The magnets that squeeze the plasma in place have to be kept well below 200 Celsius, or else their performance will degrade from a buildup of electrical resistance.

Particle accelerator. Photo: Monty Rakusen/Getty

It was a graduate student who suggested that the MIT team see what would happen if they made magnets out of a newly developed superconducting tape. A superconductor conducts electricity so well that it doesn’t build up electrical resistance and this new tape maintains that property, even at slightly higher temperatures than other superconductors do.

Using less energy on cooling could make a tokamak cheaper to run. But that benefit was minor compared to the other things Whyte’s group figured out. As they plotted out ways of winding the tape into coils in a tokamak, they realized this method could double the strength of the magnetic field they could exert on a plasma. Increasing the field strength is crucial because plasma is wild. It’s unstable and evasive, and only overwhelming force can keep it from spreading out and cooling too much.

Perhaps best of all: using this tape instead of rigid superconductors could make the machine 10 times smaller.

That led them to another problem with traditional tokamaks. If you need to replace parts of the machine, you have to take the whole thing apart and put it back together. That’s unacceptable for a power plant in regular use. And again, one of Whyte’s graduate students had a great idea. If you apply the superconducting tape in sections, with joints, the magnets can be snapped on and off for quick and easy repairs or upgrades.

“This was the beginning of the ‘aha!’ moment,” Whyte says. “The people who are in CFS were in that class.”

Other big ideas kept coming. One of the great things about fusion is its inherent safety. It’s impossible for this tiny star to slip out and cause trouble, because the plasma’s weird physical state can’t be sustained outside of the magnetic field. Still, the plasma does send something out that you’ve got to deal with: neutrons.

Fusion projects generally aim to fuse two forms of hydrogen: deuterium and tritium. Deuterium is readily available in seawater, but tritium is very rare, so you have to make it. (More on that in a minute.) In this version of fusion, 80 percent of the energy that is released comes out in the form of neutrons. These are subatomic particles that have no electric charge, so they’re not contained by the magnetic field. They come flying out like angry spittle.

In fusion experiments measured in seconds or less, flying neutrons aren’t a big problem. But over time, they can be nasty. These particles jump a foot and a half from the plasma and have enough energy to rearrange the atoms in the tokamak’s inner wall, eventually degrading it. What to do about that in a power plant that needs to run for long stretches?

Whyte describes the answer with a wry smile. “We turned the problem around,” he says.

In essence, the MIT plan takes a ride on the neutrons by catching them in a liquid. Neutrons wreck solid materials by scrambling the order of their atoms, but liquids are already disordered, by definition. In the design that CFS is developing, the neutrons pass through an inch or two of steel and then barrel into a liquified salt, which they essentially just heat up. Then, that molten salt can be pumped around a power station to generate electricity. By the way, there’s lithium in the molten salt, and when neutrons hit lithium, they create tritium, which you can take out and use to fuel the fusion reactor.

Thetatron Experiment designed to study the ionisation and compression produced in deuterium plasma, 1964. Photo: Fox Photos/Getty

This setup isn’t perfect, however. Blanketing the tokamak’s steel wall with molten salt will lessen, but not eliminate, the damage that the neutrons would otherwise cause to the metal. It will have to be replaced every so often. Just how often? That’s a crucial question for the cost of a power plant.

For now, Whyte says, the metal barrier should last a year or two. That’s not great, so materials that better withstand neutrons have to be developed, to extend the lifespan of that wall. That looks doable; reducing the erosion of the wall in fusion reactors is a long-standing field of research.

But the issue is nonetheless significant enough that General Fusion, the company that intends to compress plasma with pistons, plans to keep a solid metal case relatively far away. It will directly surround the plasma with liquid metal that gets pumped off to convert its heat to electricity. There will be lithium in that liquid, too, to breed tritium.

Even if the MIT team manages to extend the life of the barrier, there’s another issue: The neutron bombardment will eventually render the metal radioactive.

Is that a big problem? Well, one of the novel things about a fusion company called TAE Technologies, which has raised $600 million from Google, the late Microsoft founder Paul Allen, and other luminaries, is that it plans to fuse hydrogen protons with boron, a reasonably abundant element, because that reaction emits hardly any neutrons. TAE’s co-founder and CEO, Michl Binderbauer, says that because of its cleaner profile, hydrogen-boron fusion is “the single shining opportunity for mankind.”

But since we’re talking about fusion, of course there’s a catch. Hydrogen-boron fusion is much harder to pull off: The plasma has to get to billions of degrees, not millions. And the “reaction rate” is much lower, which means less fusion happens. TAE is going to start with deuterium-tritium fusion before trying to work its way up.

In the meantime, Whyte and just about everyone else in fusion thinks deuterium-tritium fusion is well worth gunning for. Any radioactive components in MIT’s design will be relatively small and have a short half-life. The material would be nowhere near as problematic as the stuff that comes out of nuclear power plants today. If fusion plant operators have to replace the inner wall from the reactor Whyte envisions, they’d “put it in a swimming pool for 10 years,” he says. “And then you can walk up beside it.”

Before any of that happens, CFS will try to pull off fusion’s most elusive trick: doing something ahead of schedule.

With the investment it’s raised, the company has about three years to test components of its reactor design, especially those still-unproven new magnets. Then, it will need to raise hundreds of millions to build a prototype reactor, at a location to be determined. The company has said it intends to get that reactor running by 2025. But its CEO, a former MIT graduate student named Robert Mumgaard, says it could happen even sooner.

Alas, fusion timelines still have a habit of slipping, even in private companies. Over the past few years, the defense contractor Lockheed Martin, and a few startups, said they hoped to show working prototypes by now — possibly even ones that achieved the ultimate, a net power gain. That hasn’t happened. When I asked for updates, I got some vague replies, ranging from “we are hard at work” to “the preliminary results are promising.”

If fusion power just won’t work, “I’m scared for the world,” says the CEO of Commonwealth Fusion Systems.

I got the most reassuring answer from Christofer Mowry, CEO of General Fusion.

His company said in 2017 that it hoped to get its first prototype running in three to five years. It’s really more like five years from now. But, he says, that’s because the company has needed time to raise “a few hundred million dollars,” not because fusion science is still iffy. Because so many companies are trying to make fusion power practical, and because demand for it will be so high, “I’m 100 percent confident that this is going to happen,” Mowry says. “Are we going to have commercial fusion power plants on the grid by 2030? Maybe. But it won’t be 50 years, I can tell you that.”

At CFS, Mumgaard sees parallels with the story of human flight. Before the Wright brothers finally got a plane off the ground, a lot of people tried and got kind of close. Plenty of observers assumed that meant human flight would always remain a fantasy. But all that time, through all those failures with gliders and flapping man-made wings, engineers were systematically probing aerodynamics. The Wright brothers built on that knowledge and combined it with insights about control mechanisms that they had from working with bicycles. And only then, was it obvious: yes, humans can fly.

“When you have the insight into a piece of technology and you get it over that hump, it goes,” Mumgaard says. “We’re think we’re at this point.” He refers to his company’s plan to build a prototype as the Kitty Hawk moment.

But what if that parallel breaks down? What if fusion power just won’t work, or won’t work at a cost that anyone will be willing to pay? Then “I’m scared for the world,” Mumgaard says.

And it’s hard to cheer him up on that point. None of the existing carbon-free alternatives seem suited to the scale of the climate problem. Conventional nuclear power is unpopular and expensive. There aren’t many more waterways to dam. For solar and wind to be the primary answer, you’d need epic amounts of batteries, which might be environmentally or economically prohibitive.

When I pose the same question to Whyte, I get a slightly different answer. He sounds like a person who has considered what would happen if he gave up on this dream, and then renewed it instead. “I would never say ‘if we don’t develop fusion we’re not going to make it,’” he says. “But, boy, I’ll put it the other way around: If you make fusion economical, you have given yourself an arrow in the quiver which is almost unmatched in going after this.”

“We’re giving it our best shot,” he continues. “Others are giving it their best shot.” And then he slaps his hand on the table in front of him for emphasis. “Let’s get there.”