Fusion Finally Crossed the Line. Here's What That Really Means.

There’s a joke that’s haunted fusion research for decades: nuclear fusion is always thirty years away, and always will be.

It persisted because it was fair. Every generation of researchers seemed to be climbing toward a summit that kept retreating. The basic idea—forcing hydrogen nuclei together to release energy, the same process powering the sun—was clear by the 1950s. Building a machine that could do it in a controlled, sustained, net-energy-positive way turned out to be one of the most difficult engineering problems in human history.

Then, in the span of about eighteen months between 2021 and 2023, two separate research teams hit milestones that looked very different from the ones before. Not “we made progress toward the threshold.” Not “our models predict we’ll get there.” Past the threshold. Numbers that don’t lie.

Understanding why these milestones matter requires understanding what fusion scientists actually mean when they talk about “ignition,” “breakeven,” and “gain”—words that get used loosely in press releases and precisely in papers, and the difference matters enormously.

What Fusion Actually Is

At the center of the sun, hydrogen atoms are squeezed together so tightly—temperatures of 15 million degrees, pressures unimaginable on Earth—that their nuclei fuse. When two hydrogen nuclei combine to form helium, a tiny bit of mass vanishes and becomes energy, governed by Einstein’s E = mc². That released energy is what makes stars shine and, in principle, what could power a civilization.

The version of this reaction most useful for Earth-based fusion uses two isotopes of hydrogen: deuterium (one proton + one neutron) and tritium (one proton + two neutrons). Fuse them together and you get helium, a free neutron, and 17.6 million electron volts of energy. Deuterium is essentially unlimited—it’s found in ordinary seawater. Tritium is rarer but can be bred from lithium, which is abundant. The fuel supply, in principle, is inexhaustible.

The challenge is confinement. To fuse, hydrogen nuclei must be moving fast enough to overcome their mutual electrostatic repulsion—the same force that keeps your hand from passing through a table. That requires temperatures of roughly 100 million degrees Celsius, about seven times hotter than the core of the sun. At those temperatures, any material you might want to use as a container instantly vaporizes.

This is the fundamental problem fusion researchers have wrestled with since the 1950s: how do you hold something 100 million degrees hot long enough and dense enough that more energy comes out than you put in?

Two Approaches, Both at a Turning Point

There are two main strategies, and both of them just had landmark results.

Inertial confinement fusion (ICF), practiced at the National Ignition Facility (NIF) in California, uses intense laser pulses to compress a tiny pellet of hydrogen fuel so rapidly that it ignites before it has time to fly apart. The pellet, a hollow sphere about the size of a pepper grain, contains a thin ice layer of frozen deuterium-tritium on the inside. The lasers hit the outer shell with 192 beams simultaneously, converting it to plasma that blows outward. By Newton’s third law, the inner fuel is driven inward at extraordinary speed—reaching temperatures and densities where fusion ignites.

Magnetic confinement fusion, practiced at tokamaks including ITER (the international experiment under construction in France) and the startup-backed SPARC device being designed in Massachusetts, uses powerful magnetic fields to hold superheated plasma in a donut-shaped magnetic bottle while it reaches fusion temperatures. The plasma never touches the walls because the magnetic field keeps it in the center.

Both approaches have the same fundamental goal—enough heat, enough density, long enough—measured by the Lawson criterion, a 1957 formulation that defines the conditions under which fusion can sustain itself.

The NIF: From Burning Plasma to Ignition

The National Ignition Facility was built in Livermore, California over a decade at a cost of roughly $3.5 billion. When it fired up around 2009, it was—and remains—the world’s largest and most energetic laser. Its 192 laser beams can deliver up to 1.9 megajoules of ultraviolet light in a pulse lasting a few nanoseconds. The building that houses it is the size of three football fields.

Progress in the 2010s was frustrating. The pellets weren’t imploding as symmetrically as models predicted. Engineering problems compounded. The fusion yields were tiny fractions of what they’d need to be. By around 2015, the project had been scaled back, recalibrated, and declared a partial failure by some critics.

Then something changed.

A series of design improvements—better capsule surface finish, new inner-shell geometries, adjustments to the laser pulse shape—began paying off. And in spring 2021, an experiment produced a result that genuinely surprised the team: the plasma was heating itself.

That might sound like a small thing. It isn’t. When a fusion plasma heats itself—when the energetic helium nuclei produced by each fusion reaction stay in the plasma long enough to heat the surrounding fuel and trigger more fusion reactions—that’s called a burning plasma. It’s a qualitative transition, the moment a reaction begins to feed itself rather than relying entirely on externally injected energy. It’s what happens inside the sun. It had never been achieved in a laboratory before.

The results were published in Nature in January 2022 by Zylstra and colleagues: burning plasma achieved, fusion self-heating measurably exceeding the mechanical work injected into the implosion. A capsule gain of 5.8—meaning the fuel pellet released 5.8 times more energy than the mechanical work that compressed it. The overall target gain—fusion yield divided by total laser energy—was still 0.72. Not breakeven. But unambiguously burning.

Six months later, a shot from August 2021 was reported in Physical Review Letters (Abu-Shawareb et al., 2022). This experiment produced 1.37 megajoules of fusion yield from 1.92 megajoules of laser input—still not target-gain breakeven, but meeting the Lawson criterion for ignition by nine different formulations. Ignition, in the rigorous physics definition, is the state where a plasma’s fusion heating can overcome all its energy losses and propagate a burn through surrounding cold fuel. The Lawson criterion is the quantitative threshold for that. They crossed it.

Then came December 5, 2022.

The Night They Made History

At 1:03 AM on December 5, 2022, NIF fired all 192 beams at a new capsule design. The shot took about 20 billionths of a second. Data collection and analysis took longer.

When the numbers came back, they showed something the facility had been built to achieve: target gain greater than one.

The laser delivered 2.05 megajoules. The target produced 3.15 megajoules of fusion energy. A gain of 1.5.

For the first time in 70 years of trying, a laboratory fusion experiment had released more energy than the driver put into the target. Energy in, more energy out. The fundamental proof of principle for controlled fusion—demonstrated.

The results were published in February 2024 in Physical Review Letters (the “Indirect Drive ICF Collaboration”), with 338 citations already. The paper is matter-of-fact in the way physics papers are. The achievement is anything but.

But Wait—About That Wall Plug

Here’s where honesty requires slowing down.

The 2.05 megajoules that hit the target represents only a fraction of the electrical energy NIF consumed to fire those lasers. The full electrical draw for a NIF shot is roughly 300 megajoules—about 150 times more than the energy delivered to the capsule. So in terms of electricity in versus fusion energy out, the ratio is still about 1:100 in the wrong direction.

This isn’t a secret or a failure—it’s a known feature of where NIF sits in the development trajectory. NIF was built as a research facility, not a prototype power plant. Its lasers are inefficient, its repetition rate is slow (a few shots per day, not per second), and it was designed to answer a physics question: can laser-driven ignition work? The answer is yes.

The path to a practical power plant requires laser systems that are much more efficient (modern concepts target ~15% wall-plug efficiency vs. NIF’s ~0.5%), much faster (10+ shots per second), and mass-producible fusion targets (NIF’s hand-crafted capsules cost thousands of dollars each). None of those are solved. All of them are being worked on.

The December 2022 result doesn’t mean fusion power plants are imminent. It means the underlying physics works. That’s the thing that was actually in question, and it’s the thing that’s now been answered.

The Other Path: Magnets and SPARC

While NIF was working on inertial confinement, a different community of researchers was redesigning the tokamak—the donut-shaped magnetic fusion device that has dominated government-funded fusion research for decades.

Tokamaks work by threading magnetic field lines through a plasma in a torus, keeping it magnetically contained while it heats up. The challenges are different from laser fusion but similarly daunting: keeping the plasma stable, preventing disruptions where the magnetic confinement suddenly fails, and heating the plasma to fusion temperatures using external radio frequency power or injected beams of fast atoms.

The biggest tokamak ever built is ITER, under construction near Cadarache in France. It’s an international collaboration—35 countries, about €20 billion, first plasma expected around 2025 with full deuterium-tritium operation in the 2030s. ITER is designed to produce 500 megawatts of fusion power from 50 megawatts of heating power—a gain of 10. It’s a crucial demonstration, but at roughly 23,000 cubic meters, it’s also enormous.

Here’s the key insight that changed the game for a smaller approach: fusion power scales steeply with magnetic field strength. Roughly, fusion power per unit volume goes as the fourth power of the magnetic field. Double the field, and you get sixteen times more power in the same volume. Or equivalently, you can achieve the same performance in a much smaller, cheaper machine.

For decades, this was theoretical. The magnets capable of reaching much higher fields than conventional superconductors existed only in small laboratory scale. Then a new class of superconducting materials—rare earth barium copper oxide, known as REBCO—began to mature. Unlike the conventional niobium-based superconductors used in most large magnets, REBCO operates at relatively “high” temperatures (around 20 Kelvin, compared to 4 Kelvin for conventional superconductors) and can carry extraordinarily high current densities even in strong magnetic fields.

In 2018, researchers at MIT’s Plasma Science and Fusion Center and a newly formed spinout called Commonwealth Fusion Systems (CFS) began a three-year program to prove REBCO could be used in a large-scale, high-field fusion magnet. The question was whether these tapes—individually thin as a credit card, individually fragile—could be assembled into a coil capable of reaching fields ten times stronger than Earth’s, surviving the enormous mechanical stresses of such a field, and doing so reliably.

The answer came in September 2021.

20 Tesla. Large Scale.

On September 5, 2021, in a test facility in Devens, Massachusetts, a magnet system called the SPARC Toroidal Field Model Coil (TFMC) was powered up.

The coil was roughly three meters across—large enough that you couldn’t easily reach across it—and weighed over ten thousand kilograms. It used 270 kilometers of REBCO tape wound into 16 pancake-shaped coil sections. And when they pushed 40,500 amperes of current through it:

20.1 Tesla.

For context: MRI machines typically operate at 1.5 to 3 Tesla. The Large Hadron Collider’s superconducting dipole magnets reach about 8.3 Tesla. The TFMC hit 20.1—the strongest magnetic field ever achieved in a large-scale superconducting fusion magnet—and held it. The mechanical stresses on the REBCO tape stacks approached a gigapascal—comparable to the pressure at the bottom of the Mariana Trench—and the structure held.

The result was published in detail in the IEEE Transactions on Applied Superconductivity in 2024. The engineering achievement is substantial: the team had to invent new ways to join the tape stacks electrically (achieving resistances of 0.5 to 2.0 nanoohms per joint), develop a new pressure-vessel style cooling scheme, and create computational models that correctly predicted the magnet’s electromagnetic behavior before it was ever tested.

What SPARC Is Designed to Do

With this magnet technology validated, Commonwealth Fusion Systems and MIT are designing SPARC: a compact tokamak that uses the new high-field magnets to achieve burning plasma performance in a machine about 1/65th the volume of ITER.

SPARC’s design parameters, published in the Journal of Plasma Physics (Creely et al., 2020), target a magnetic field of 12.2 Tesla on axis, a major radius of 1.85 meters, and plasma current of 8.7 million amperes. With conservative plasma physics assumptions, SPARC is predicted to achieve a plasma gain (Q) of about 2—the minimum required to call it net energy producing. With nominal assumptions, Q is projected at approximately 11, producing about 140 megawatts of fusion power.

For comparison: ITER, with its 35-country consortium and €20 billion budget, is designed for Q = 10. SPARC aims to match that in a machine you could fit in a large office building.

SPARC is not a power plant. It’s a burning plasma research device, designed to answer questions about how to build ARC—Commonwealth Fusion’s planned commercial fusion reactor. But the trajectory is direct: prove the magnet technology (done), build SPARC (in progress), demonstrate net energy with burning plasma physics (projected in the late 2020s), build ARC (2030s).

Commonwealth Fusion announced completion of the SPARC engineering design in 2022. They’ve broken ground on a new facility in Devens, Massachusetts. The machine is being built.

Why This Time Is Different

Fusion has announced breakthroughs before. There have been record plasma temperatures, record confinement times, record neutron yields. Each one was real and each one was also, by itself, insufficient. The joke about thirty years persisted because partial milestones have a long history.

What’s different now?

The physics question has been answered at NIF. We know that fusion ignition—sustained, self-heating fusion where more energy comes out of the target than goes into it—works in a laboratory. That was the foundational question, and it’s been experimentally answered. The remaining challenges are engineering: efficiency, repetition rate, economics.

The magnet technology that changes the economics of tokamaks is real. The TFMC wasn’t a computer model or a small-scale prototype scaled by extrapolation. It was a fusion-scale magnet hitting 20 Tesla, built with production techniques, surviving full mechanical loads. The scaling argument that lets SPARC shrink ITER’s volume by a factor of 65 is now experimentally grounded.

Private capital has arrived. Commonwealth Fusion Systems has raised over $2 billion. TAE Technologies, Helion Energy, General Fusion, Zap Energy, and others have raised hundreds of millions more. The pace of private investment in fusion accelerated by an order of magnitude in the 2020s. This is partly rational response to the milestones above—the physics and engineering risks look genuinely different after 2022—and partly investor excitement in a sector that now has clearer milestones. Either way, the money funds work that wasn’t happening before.

The energy context has changed. The 1970s oil crisis briefly accelerated fusion funding; then cheap oil blunted the urgency. Today, the pressure to decarbonize global energy by mid-century creates a different kind of urgency—one that doesn’t go away when oil prices drop. A clean, baseload, fuel-unlimited energy source fits a need that exists and isn’t going to stop existing.

What’s Still Hard (And It’s a Lot)

Honesty demands a counterweight.

NIF’s December 2022 shot was a physics triumph inside a one-of-a-kind facility built for weapons research, with laser efficiency of ~0.5% and capsules that cannot be manufactured at scale. The path from “the physics works” to “power plants producing electricity” requires laser efficiency improvements by a factor of 30 or more, target manufacturing at cents apiece instead of thousands of dollars, repetition rates of ten shots per second instead of one per day, and tritium breeding blankets that have never been tested at scale. These are large engineering challenges, not small ones.

For tokamaks, SPARC still needs to demonstrate burning plasma and net energy. The plasma physics community debates whether SPARC’s models are optimistic. Disruptions—sudden failures of magnetic confinement that can damage the machine—remain a serious concern. The materials science of components facing both neutron bombardment and plasma contact is brutal.

ITER is running late and over budget, a reminder that fusion projects of any size are complex.

None of these challenges are fundamental—none of them require overturning known physics. But “not fundamental” and “easy” are different things. The history of fusion is partly a history of engineering problems that looked tractable and turned out to be harder than expected.

A 70-Year Bet, Starting to Pay

The joke about fusion being always thirty years away will probably persist for a while, because jokes have their own half-life. But the underlying claim it makes—that there’s no real progress, that the community is chasing a forever-receding goal—has been empirically falsified.

On December 5, 2022, at 1:03 in the morning, a capsule of frozen deuterium-tritium the size of a pepper grain released more energy than the laser put into it. For a few billionths of a second, inside a tiny collapsing plasma fireball, the same process that makes the sun shine happened in a California laboratory—and this time, the energy balance was on our side.

In September 2021, in a test facility in Massachusetts, engineers switched on a magnet that hit 20 Tesla at fusion scale, vindicating a theory that compact, high-field tokamaks could achieve what ITER-class machines require orders of magnitude more material to do.

These aren’t milestones in a story that ends in disappointment. They’re the first verified paragraphs of a story whose ending depends on engineering that we know how to do, economics that are improving, and political will that’s still uncertain.

Sixty years of work by thousands of researchers built to this. The line has been crossed. The rest is engineering.

---

Key papers and sources for this piece:

Burning plasma: Zylstra AB, Hurricane OA, et al. “Burning plasma achieved in inertial fusion.” Nature 601, 542–548 (2022). DOI: 10.1038/s41586-021-04281-w. PMID: 35082418. ~499 citations.

Lawson criterion: Abu-Shawareb H, et al. (Indirect Drive ICF Collaboration). “Lawson Criterion for Ignition Exceeded in an Inertial Fusion Experiment.” Physical Review Letters 129, 075001 (2022). DOI: 10.1103/PhysRevLett.129.075001. ~465 citations.

Ignition/target gain > 1: Abu-Shawareb H, et al. “Achievement of Target Gain Larger than Unity in an Inertial Fusion Experiment.” Physical Review Letters 132, 065102 (2024). DOI: 10.1103/PhysRevLett.132.065102. ~338 citations.

SPARC overview: Creely AJ, et al. “Overview of the SPARC tokamak.” Journal of Plasma Physics 86, 865860502 (2020). DOI: 10.1017/s0022377820001257. ~406 citations.

SPARC TFMC program: Hartwig ZS, et al. “The SPARC Toroidal Field Model Coil Program.” arXiv:2308.12301 (2023). ~3 citations (preprint).

SPARC TFMC design/fabrication: Hartwig ZS, et al. “Design, Fabrication, and Assembly of the SPARC Toroidal Field Model Coil.” IEEE Transactions on Applied Superconductivity 34 (2024). DOI: 10.1109/tasc.2024.3356571. ~71 citations.