If you have an ACS member number, please enter it here so we can link this account to your membership. (optional)

ACS values your privacy. By submitting your information, you are gaining access to C&EN and subscribing to our weekly newsletter. We use the information you provide to make your reading experience better, and we will never sell your data to third party members.


Nuclear Power

Fusion start-ups hope to revolutionize energy in the coming decades

With the help of venture capital funding and new technologies, a cadre of companies want to commercialize fusion energy in the next 20 years

by Katherine Bourzac
August 6, 2018 | A version of this story appeared in Volume 96, Issue 32
A photograph of the General Fusion reactor, showing its hedgehog piston spines.

Credit: General Fusion | The pistons surrounding General Fusion's main reactor fire once per second, pressurizing the liquid metal and plasma within.


In brief

A group of start-up companies hopes to capitalize on the potential of nuclear fusion to turn abundant fuels into carbon-neutral energy. These companies want to commercialize fusion by the 2030s to feed growing global demand for energy while addressing climate change. Meanwhile, a multi-billion-dollar international science project called ITER wants to demonstrate net energy generation from fusion by 2035, leading to commercial reactors in the 2050s or 2060s. Skeptics of the start-ups say the companies may be overlooking serious challenges, particularly with regard to finding suitable reactor materials, that the more science-focused ITER project will explore. But fusion entrepreneurs expect new technologies, including high-temperature superconductors and machine learning, will give them the edge.

Martin Greenwald stands in a windowless control room at Massachusetts Institute of Technology, holding a heavy, thick cable in his left hand. Pinched between his right index finger and thumb is a thin strip of dull-brown metal tape. That strip can carry as much electrical current as the hefty cable, he explains, and could enable scientists to build a mighty machine they’ve been dreaming about for decades.

By winding spool after spool of the yttrium barium copper oxide tape into some of the strongest magnets on Earth, scientists at the MIT Plasma Science & Fusion Center hope to build a nuclear fusion reactor. If all goes well, these ultrastrong magnets will lasso and wrangle a roughly 100 million ºC hydrogen plasma, driving its protons to fuse and releasing tremendous amounts of energy over sustainable time periods. Commonwealth Fusion Systems, a spin-off company from this MIT project, hopes to turn the research into a commercial reactor to produce carbon-free electricity.

“Fusion has the features you want in the ideal energy source,” Dennis Whyte, director of the MIT center and a cofounder of Commonwealth Systems, says. The fuels are abundant, and the process is safer than nuclear fission because there is no risk of a meltdown. Fusion reactors should use less land than renewable energy technologies such as solar panels and provide continuous power as coal plants do but without greenhouse gas emissions.

Commonwealth Fusion Systems and other fusion start-ups hope to build commercial reactors and realize this vision in the next decade or two. While these scrappy upstarts are getting on with their work, another, much larger effort is already under way.

A multinational, multi-billion-dollar, multidecade project called ITER promises to demonstrate net energy production from nuclear fusion after its reactor turns on in 2025. (Iter means “the way” in Latin; the project was originally called the International Thermonuclear Experimental Reactor.) But the ITER design is not scalable—it’s far too large and expensive to serve as a power plant on the electrical grid. Instead, it’s designed to give partner countries the research tools they need to start building practical fusion reactors sometime in 2055 at the earliest.

That’s far too late, some researchers say, especially in the face of climate change. “We need fusion energy to be deployable at a scale of tens of gigawatts at many power plants in the 2030s to tackle carbon emissions,” says David Kingham, executive vice chair of Tokamak Energy, a fusion start-up in Oxfordshire, England.

This dream of clean, abundant energy from nuclear fusion has been echoing in basement labs like the one at MIT since the 1950s. But since that time, no one has yet shown that a fusion reactor can produce more energy than it consumes—let alone run stably for years or decades.

Fusion entrepreneurs at these start-up companies say things are different now. They expect new technologies, such as high-temperature superconductors, big-data analytics, and artificial intelligence methods, will give their projects the edge by overcoming practical problems that have long plagued fusion reactors.

Also, fusion entrepreneurs and the deep-pocketed investors who are sponsoring them are seeing green. “Fusion has been undervalued by governments. It’s long term. It’s speculative. But the upside is huge,” Greenwald, the deputy director of the MIT center and a cofounder of Commonwealth Fusion Systems, says. “There’s trillions of dollars to be made. There’s trillions of watts of additional demand coming,” says Michl Binderbauer, CEO of TAE Technologies in Foothill Ranch, Calif.

Still, there are skeptics who think these start-up companies’ promises are unreasonable—at least on the aggressive time frames they’re promising. And some of the companies, skeptics say, are working on designs that physicists have deemed not ready for the grid anytime soon, while neglecting practical issues such as how to build reactors resilient enough to withstand the intense heat produced during fusion.

Fusion proponents may not have a net-energy-generating fusion reactor on Earth to point to, but they encourage skeptics to look up at the sun. Like Prometheus, “we’re bringing the fire of the stars down to Earth,” Whyte says.

Prometheus unbound

A schematic depicting the fusion reaction between deuterium and tritium.
At high enough temperatures, deuterium and tritium fuse to produce helium, a neutron, and a whole bunch of energy.
A schematic showing a few components of the ITER tokamak reactor.
Credit: ITER
The tokamak reactor at ITER will be 30 meters tall. An 830-m3 plasma will be held in a vacuum vessel via strong magnets.

The sun is a giant sphere of incandescent plasma. Thanks to its crushing gravity and intense heat—about 15 million ºC inside its core—hydrogen nuclei smash together, producing helium and tremendous amounts of energy.

But at earthly temperatures, atoms do not want to fuse. Even when zooming around in a hot plasma, stripped of their electrons, atomic nuclei just bounce off one another. To make fusion work as an energy source on Earth, researchers have to overcome that resistance by creating intense, sunlike conditions inside a reactor. Then they can use the heat produced by those reactions to generate electricity.

For nearly 70 years, researchers have been exploring how to heat and contain a plasma—typically one that consists of deuterium and tritium—for sustained periods of time. To do so, many have taken inspiration from the sun: They heat up the plasma and encourage nuclei to get close by applying tremendous pressure. When the plasma gets hot enough, deuterium and tritium overcome their initial shyness and fuse, producing helium and high-energy neutrons. Some of those neutrons in turn strike nuclei in the plasma, transferring some of their energy and encouraging more fusion reactions. Many others will strike the sides of a reactor, heating up the walls. Engineers can harness that heat to generate power­—for example, by driving turbines.

Corralling an ultrahot plasma for sustained periods of time is incredibly challenging. One false move, and the plasma starts bucking like a bronco. Instabilities can cause the plasma to crash into the walls of a multi-million-dollar reactor and damage it.

The best-performing fusion reactors built so far, called tokamaks, use magnetic fields to control and compress the plasma. But there are other ways to do it: Some reactors pulse the plasma with current to induce it to make its own magnetic field, and the National Ignition Facility at Lawrence Livermore National Laboratory compresses plasma with energetic laser blasts.

ITER is working on a tokamak design. The international effort, initiated by the U.S. and Soviet Union in 1985, consists of 35 countries banding together to mitigate some of the risks in building what many believe will be the first reactor to demonstrate net energy generation from fusion. (The only human-made fusion reactions that release more energy than they consume are hydrogen bombs, in all their frightful variations. Ivy Mike, the first hydrogen bomb, vaporized the island of Elugelab when it was detonated by the U.S. in 1952.)

ITER broke ground on a site for its reactor in southern France in 2010. The project plans to switch on its tokamak in 2025. By 2035, ITER will use 50 MW of input power to produce 500 MW of fusion power.

Once the reactor is generating net power, the team will take at least 10 years “to explore different parameters to optimize the process,” ITER Director General Bernard Bigot says. Bigot says it may not be until 2055 or 2060 that fusion is ready for commercial implementation on the power grid.

Given this multidecade timeline, ITER’s workers often liken themselves to stonemasons hewing blocks for a cathedral: They may not get to see the fruits of their efforts in their own lifetimes. To the researchers at fusion start-ups, the whole thing seems downright medieval. They don’t want to be stonemasons working on a large geopolitical science project; they want to transform the electrical grid, pronto.

It’s not just about speed. Plasma physicists worry that the fusion community has, with ITER, put all its eggs in one basket, neglecting promising new technologies because of budgetary limitations and the need to pick one can’t-fail reactor design. “We think fusion is too important for just one try,” MIT’s Greenwald says. They believe their reactor concepts have promise that ought to be explored—and they hope to demonstrate their effectiveness faster than ITER can with its reactor.

An aerial photograph of the construction of the ITER facility.
Credit: ITER
On a 180-hectare site in southern France, engineers with the ITER project are building the world’s largest tokamak.

Improving reactor designs

Greenwald is one of a group of researchers betting that they can improve on the tokamak, making more compact but still powerful reactors with the help of new superconducting materials. Tokamaks were dreamed up by nuclear physicist Andrei Sakharov in 1950. (The reactor name comes from the Russian acronym for “toroidal chamber with magnetic coil.”) These reactors use complex, expensive, and extremely powerful magnets to confine plasma into a whirling, bagel-shaped toroid.

“The tokamak has gotten right up to net energy gain but just didn’t get to the finish line,” MIT’s Whyte says. For that reason, fusion enthusiasts who are slightly more conservative favor the reactor design.

Whyte hopes to push the tokamak forward with new high-temperature superconductor materials such as yttrium barium copper oxide. These materials weren’t available when the design of ITER’s massive reactor was finalized. Today, companies including Bruker sell them in the form of skinny tapes like the one Greenwald held between his fingers in the MIT lab.

High-temperature superconductors provide much higher currents than other conductive materials used to make magnets and require less-complex cryogenic systems than low-temperature superconductors. Both features may allow researchers to build a more compact but very high field tokamak reactor.


The idea makes scientific sense but will require a lot of engineering to realize. Although these materials are on the market, no one has used them to build a large, high-field magnet like those required for tokamaks. “We have to develop the highest-performing superconducting magnets ever built” in order to meet the goals of the MIT project, Whyte says.

Commonwealth Fusion Systems, the company spun off of the MIT project, has an aggressive schedule. The company plans to build an experimental reactor based on these new magnets in three years—and they expect that reactor to produce twice as much energy as needed to run it. If all goes as planned, this first reactor will produce 100 MW, one-fifth of ITER’s expected power output, at about one sixty-fifth of the volume.

Tokamak Energy in England is also banking on new magnet materials, though its reactor has a slightly different design. Instead of encouraging plasma to form a doughnut shape, Tokamak Energy uses magnetic fields to shape plasmas into a thicker toroid that looks like an apple with the core taken out. Scientists at Oak Ridge National Laboratory experimented with this plasma shape back in the 1980s. The shape can, in theory, allow for a smaller reactor. Executive Vice Chair Kingham says the company’s reactors could be on the grid in 2030.

Meanwhile, other companies, including General Fusion of British Columbia, are working on alternative reactor designs that bear little resemblance to the tokamak. General Fusion’s technology has its roots in LINUS, a 1970s U.S. Naval Research Laboratory project.

In this reactor design, a deuterium-tritium plasma gets injected into a liquid-metal cavity encased in a steel sphere surrounded by hedgehog-spine-like pistons. Every second, before the plasma can expand and cool, the pistons fire, pushing the liquid-metal walls inward and compressing the ionized isotopes. The liquid metal and pistons in this design play the role of the tremendous magnets used in tokamaks. Neutrons produced by fusion reactions heat the liquid metal, which gets continuously pumped through a heat exchanger so that the heat can generate electricity.

The project didn’t work the first time because “the enabling technologies didn’t exist back then,” says Christofer Mowry, CEO of General Fusion. The company thinks those technologies are now in hand. “It’s only now that it can be taken off the drawing board,” Mowry says.

For example, the company has partnered with GE to use three-dimensional printing to make a complex mesh cage that spins the liquid metal around, generating complex centrifugal forces that hold it in a spherical shape with a hollow center to house the plasma. Achieving and maintaining the right shape in the liquid metal was a problem for earlier attempts at this technology—the liquid metal has to move in such a way that it pressurizes the plasma but never actually touches it.

General Fusion is also hopeful about succeeding because of data analytics techniques it has implemented with help from Microsoft. These techniques may help the company during its research phase, enabling engineers to more quickly optimize the reaction conditions. Data analytics methods such as machine learning could also assist during reactor operation, helping adjust pressure and other conditions on the fly to keep the liquid metal and plasma in the right state. By making these small corrections, the company could avoid losing control of the plasma, which would force the company to turn off the machine.

The company has demonstrated the liquid-metal cage, the pistons, the plasma injector, and other components. However, General Fusion has not yet integrated all the parts in a full reactor. The company plans to have a commercial plant in the next decade.

A photograph of a mesh cage used in General Fusion’s reactor.
Credit: General Fusion
In collaboration with GE, General Fusion is using a three-dimensional printer to make mesh cages that will spin a combination of liquid lead and lithium to create a swirling sphere with a hole inside for injected plasma. This is a plastic prototype of the cage.

Alternative fuels

General Fusion isn’t the only company with an alternative reactor design that looks beyond strong magnets. At TAE Technologies, engineers are using an arsenal of technologies to control their plasma.

Wrangling plasma in a reactor with magnetic fields is like trying to hold jelly in place with rubber bands, CEO Binderbauer says. To get better control, you can either use stronger rubber bands or try to make the plasma jelly stiffer. TAE Technologies uses four particle accelerators, small magnets, and radio-frequency pulses to steer and stiffen its plasma.

In a hangarlike building next door to a dance studio in the suburbs of Orange County, TAE Technologies engineers are running a battery of tests on their reactor. Every few minutes, a metallic clang rings out like a hammer striking an anvil. That’s the sound of 750 MW of electricity discharging in the testing room. Part of that power comes from capacitors charged by electricity off the grid. The system has to be disconnected from the grid when it’s running; otherwise, the pulses will create disruptive echoes in the region’s power lines, Binderbauer explains.

On either side of the TAE reactor is a pair of long quartz tubes. Hydrogen gas gets injected into these tubes, and metal coils wound around the tubes pump it with radio-frequency pulses generated by those clanging electrical discharges. The pulses convert the gas into plasma, and the two plasma streams rush down the tubes, colliding head-on in the center of the reactor. Magnetic fields hold the colliding plasmas steady, while beams of protons streaming out of four particle accelerators bombard the plasma. The beams are aimed so that the plasma forms a tirelike, spinning toroid.

The energy injections from TAE’s particle accelerators help make the plasma stiffer, so the magnetic fields don’t have to be so strong. The TAE reactor uses some magnets to guide the plasma jelly, but they are much smaller than those needed for a tokamak. As a result, the TAE reactor is relatively compact.

For the past nine months, TAE Technologies’ experimental reactor has been “huffing and puffing and making plasma,” Binderbauer says. Standing on a walkway above the company’s steel reactor chamber, Binderbauer points out the lines attached to 4,000 temperature, pressure, density, and other sensors. “There’s no fusion here—right now we’re studying how to hold the material together,” he explains.

TAE engineers are repeatedly pulsing the system and gathering data from those thousands of sensors. The company is collaborating with scientists at Google to develop automatic control systems driven by artificial intelligence. If the plasma starts to move in the wrong way, TAE engineers want the system to self-correct rapidly before the plasma hits the reactor wall. Artificial intelligence can make those corrections quickly, Binderbauer says.

After about a year of this data collection and analysis, TAE engineers will have gathered enough information to build their next-generation reactor, which Binderbauer expects will generate net energy from fusion. He expects the company will make that demonstration in about five or six years.

One thing TAE Technologies is not currently testing in its facility is the fuel it ultimately plans to use: a mixture of hydrogen and boron. Most companies hope to use the hydrogen isotopes deuterium and tritium, which is radioactive, as fuel. Fusion reactions involving those hydrogen isotopes take the least amount of energy to initiate, and the neutrons produced are highly energetic. These energetic neutrons can bombard atoms in the reactor walls, creating radioactive isotopes. Binderbauer wants to avoid using or producing radioactive materials.

But hydrogen-boron fuel comes with a significant caveat. Instead of operating at tens or hundreds of millions of degrees, TAE Technologies’ reactor will have to heat the proton-boron plasma to over a billion degrees to initiate fusion.

Tim Luce, the scientific director of ITER, acknowledges that radioactivity is a concern with fusion. One of ITER’s oft-criticized delays was due to the extended process of complying with France’s nuclear regulations. But Luce says fusion is hard enough without having to make a plasma that’s over a thousand times as hot as the sun. “You’ve set the goalpost—which was already high—even higher,” he says.

Others are more blunt. “Proton-boron fusion really attracts the outer fringes,” says Edward Morse, a self-described fusion-start-up skeptic and plasma physicist at the University of California, Berkeley. “The mainstream effort is deuterium and tritium. That’s how the bombs work. We know this works.”

The back side of TAE Technologies' fusion reactor.
Credit: TAE Technologies
The TAE Technologies reactor will use high-energy radio-frequency pulses, particle-accelerator beams, and magnets to control a proton-boron plasma.

Some cold water on hot fusion

Morse started working in fusion in the 1970s and 1980s, when researchers were thinking up many of the alternative designs that start-up companies are pursuing today. “I was there the first time around, so I sound like Joni Mitchell talking about Woodstock,” he says. ITER had good reasons for not picking these alternative designs, he says: The tokamak has a more solid experimental success record.

To other start-up skeptics, these companies’ alternative designs aren’t the only source of concern. The companies’ short timelines to commercialization also seem questionable. ITER’s slower pace might be needed to work out some problems these start-ups aren’t fully appreciating, the skeptics say.

One issue is heat, says Mickey Wade, director of advanced fusion systems at General Atomics. The company has been in the fusion market the longest—since the 1950s—but it currently has no plans to market fusion reactors for grid-scale power generation. The San Diego company operates the largest magnetic fusion research facility in the U.S., called DIII-D.

Unleashing the heat of fusion in a small reactor will be no small feat. “When you heat a plasma to a higher temperature than that of the sun, that energy has to dump onto something,” Wade says. Companies making compact reactors aim to generate that heat in a device significantly smaller than the ITER reactor. The smaller volume of reactor walls will have to absorb a tremendous amount of heat, which is likely to damage them.

Future fusion reactors may need to be made of heat- and radiation-resilient materials that don’t exist yet. But most people working at fusion companies and even in academic groups are plasma physicists. Wade says the field needs help from materials scientists to do experiments on what will happen to materials when they’re bombarded with uncontrollable neutrons, each carrying a potentially atom-shaking 14 MeV of energy.

There are some high-strength alloys that probably come close to being able to withstand the neutron bombardment, says Laila El-Guebaly, a materials scientist at the University of Wisconsin, Madison’s Fusion Technology Institute. But there’s nothing on Earth like a fusion reactor, so it’s hard to know for sure, and it’s difficult to test new materials under extreme conditions today. “The lack of a 14-MeV neutron source provides a huge hurdle for the development of fusion worldwide,” El-Guebaly says. ITER may offer the first chance to investigate these questions, Wade adds.

Tokamak Energy’s Kingham agrees that shielding materials from neutrons is a big challenge. Because of the geometry of the company’s spherical tokamak reactors, the magnets will be particularly vulnerable to neutron bombardment. The company is developing shielding materials to protect them, and he expects they may need to be as thick as 60 cm.

Then there’s the tritium fuel problem. The isotope has a half-life of just 12 years, so there’s not much of it on Earth. As a result, fusion facilities will have to produce the radioactive material on-site to keep running over long periods. Fusion researchers usually envision doing this inside the reactor, with the help of those wild neutrons. Part of ITER’s project is to explore materials for “breeding” tritium from lithium and energetic neutrons. General Fusion’s Mowry says his company has addressed this issue in its designs. It plans to dope the liquid-metal walls of its reactor with lithium. But no one has demonstrated lithium breeding in a full reactor yet.

Although some fusion researchers may be skeptical of the alternative approaches of these start-ups, people working on fusion are happy to have the company, as well as the vote of confidence from private investors. “These companies are complementary to ITER,” ITER Director General Bigot says. “Their investors want to make fusion a reality, and this demonstrates trust in fusion as a better energy supply for the world in the middle and long term.”

Fusion researchers have learned the hard way that they need to be skeptical. Those who’ve been working on fusion for decades have lived through multiple waves of the fusion hype cycle or have been swept away by it themselves. There are many stories of fusion researchers who made big promises—exciting their colleagues, the public, and governments—and then failed to deliver. Some of these researchers turned out to be frauds, while most simply got too carried away to notice errors in their science. It’s hard not to become entranced by the potential of fusion technology, ITER’s Luce says. “People want to see fusion work because it is as good an energy source as you can imagine,” he says.

Today, fusion researchers are quietly celebrating small victories and trying to keep their funders excited without spreading too much hype. ITER’s experimental reactor is 50% built as of January. In the U.K., Tokamak Energy researchers recently got their reactors operating at 20 million degrees—a temperature so hot it doesn’t matter much whether you use kelvin or Celsius.

“We were quite excited to get to the temperature at the center of the sun, the hottest place in the solar system,” Kingham says. “It doesn’t happen every day.” To celebrate that night, the team headed to the local pub—but nothing too rowdy. They had to be back at work the next day. And the next, and the next.

“I feel that in many ways, time is our biggest enemy,” MIT’s Whyte says. “If we want fusion to be a big factor in combating climate change—ticktock, ticktock.”

Katherine Bourzac is a freelance science writer based in San Francisco.


This story was updated on Aug. 23, 2018, to clarify the caption for the photograph of the mesh cage made by General Fusion. The cage pictured is a plastic prototype, not the final titanium part. The Fusion Sampler table was also updated to correct the status of the General Fusion pilot plant. The plant is still in the planning stages, not under construction. Also this story was updated Aug. 7, 2018, to correct a date in the “Fusion sampler” table. Commonwealth Fusion Systems plans to produce net energy from fusion by 2025, not 2021.


This article has been sent to the following recipient:

Chemistry matters. Join us to get the news you need.