Every few months, a headline announces another “breakthrough” in nuclear fusion. A reactor in South Korea just held plasma at 100 million degrees for over a hundred seconds. A Chinese experiment blew past a density limit that physicists assumed was unbreakable. A startup in Massachusetts is assembling a machine designed to prove that fusion power is commercially viable.
So what is nuclear fusion, actually? Why has it been “30 years away” for the past 60 years? And why are scientists suddenly making real progress?
This is the explainer that cuts through the hype.
In This Article
- What nuclear fusion is and how it differs from fission
- Why fusion requires such extreme conditions
- The fuel that could power civilization for millions of years
- How tokamaks and other reactor designs work
- The 2026 breakthroughs that changed the conversation
- When (or if) fusion power will actually reach the grid
What is nuclear fusion?
Nuclear fusion is the process of forcing two light atomic nuclei together so they merge into a single heavier nucleus, releasing enormous amounts of energy in the process.
This is the same reaction that powers every star in the universe. The sun fuses about 600 million tons of hydrogen into helium every second, and the tiny fraction of mass lost in each reaction converts directly into energy according to Einstein’s famous equation, E = mc².
Think of it this way: fission (the reaction used in today’s nuclear power plants) works by splitting heavy atoms like uranium apart. Fusion does the opposite. It squeezes light atoms together. Both release energy, but fusion releases roughly four times more energy per kilogram of fuel than fission, and nearly four million times more than burning coal or oil.
Why fusion is so difficult
If fusion is the universe’s favorite energy source, why can’t we just build a miniature sun on Earth?
The problem is the Coulomb barrier. Atomic nuclei are positively charged, and positive charges repel each other. To overcome that repulsion and get nuclei close enough for the strong nuclear force to grab hold and fuse them, you need extraordinary conditions: temperatures around 100 million degrees Celsius (roughly seven times hotter than the core of the sun) and enough pressure or confinement to keep the nuclei colliding long enough to sustain the reaction.
At those temperatures, matter enters a state called plasma, where electrons are stripped from atoms entirely. Plasma is not a solid, liquid, or gas. It is a superheated soup of charged particles that behaves in ways that make it incredibly difficult to contain.
You cannot hold plasma in a physical container because no material can survive contact with something that hot. So physicists had to invent other approaches.
How fusion reactors work: the tokamak and its competitors
The most common approach to containing plasma is a device called a tokamak, a doughnut-shaped chamber that uses powerful magnetic fields to suspend plasma in mid-air, keeping it away from the chamber walls.
The concept was first developed by Soviet physicists in the 1950s. The name comes from a Russian abbreviation meaning “toroidal chamber with magnetic coils.” Inside a tokamak, superconducting magnets cooled to near absolute zero (around minus 269°C) generate fields strong enough to confine plasma heated to 100 million degrees. The temperature difference between the magnets and the plasma they contain is among the most extreme in all of engineering.
But the tokamak is not the only game in town.
Stellarators
A stellarator uses twisted magnetic coils to confine plasma without needing the electrical current that flows through a tokamak’s plasma. Germany’s Wendelstein 7-X is the world’s largest stellarator and has demonstrated increasingly long plasma pulses. Stellarators are harder to build but potentially more stable for continuous operation.
Inertial confinement
Instead of using magnets, inertial confinement fusion (ICF) uses lasers or other drivers to compress a tiny pellet of fuel so rapidly that fusion ignites before the fuel can fly apart. This is the approach used at the National Ignition Facility (NIF) in California, which made history in December 2022 by achieving ignition: the fusion reactions produced more energy than the lasers delivered to the fuel.
Compact and alternative designs
Several private companies are pursuing smaller, faster designs. Commonwealth Fusion Systems (CFS), a spinoff from MIT, is building SPARC, a compact tokamak that uses high-temperature superconducting magnets to achieve stronger magnetic fields in a smaller package. Other startups are exploring approaches ranging from magnetized plasmas to pulsed fusion.
The fuel: seawater and rocks
The most promising fusion fuel is a mixture of two hydrogen isotopes: deuterium and tritium.
Deuterium is found naturally in seawater at a concentration of about 30 grams per cubic meter. The oceans contain enough deuterium to power fusion reactors for hundreds of millions of years.
Tritium is rarer. It is radioactive with a half-life of about 12 years and exists only in trace amounts in nature. But it can be produced inside a fusion reactor itself by surrounding the chamber with lithium, a common element found in rocks and brine deposits. When neutrons from the fusion reaction strike the lithium, they breed new tritium.
A single gram of deuterium-tritium fuel releases as much energy as about 2,400 gallons of oil. A fusion power plant could run on roughly 250 kilograms of fuel per year, compared to the millions of tons of coal a fossil fuel plant burns annually.
And unlike fission, fusion produces no long-lived radioactive waste. The primary byproduct is helium, an inert gas. Some reactor components will become mildly radioactive over time from neutron bombardment, but the waste decays to safe levels within decades rather than millennia.
Why fusion has taken so long
The running joke in physics is that fusion power is always 30 years away. There are real reasons it has taken this long.
The physics is brutal. Plasma is inherently unstable. It writhes, twists, and tries to escape magnetic confinement in ways that are extraordinarily difficult to predict and control. Small instabilities can grow into disruptions that slam superheated plasma into reactor walls, damaging the machine and halting experiments.
The engineering is unprecedented. Building a device that simultaneously maintains temperatures hotter than the sun’s core and magnets colder than outer space, while withstanding intense neutron bombardment, is an engineering challenge unlike anything else humanity has attempted.
Funding has been inconsistent. In the 1970s, a U.S. government study mapped out fusion timelines at various funding levels. The conclusion: aggressive funding could deliver a demonstration reactor by 2000. The level of funding actually provided corresponded to the “fusion never” timeline on the chart. For decades, fusion research received enough money to keep labs open but not enough to build the machines needed to make real progress.
That is starting to change.
The 2026 breakthroughs that matter
Several developments in 2025 and 2026 have shifted the fusion conversation from “someday” to “maybe soon.”
South Korea’s KSTAR: 102 seconds at 100 million degrees
In February 2026, South Korea’s KSTAR tokamak sustained plasma at 100 million degrees Celsius for 102 consecutive seconds, more than doubling its previous record of 48 seconds. The International Atomic Energy Agency called it the most significant milestone in controlled fusion since NIF achieved ignition in 2022. Maintaining plasma at fusion-relevant temperatures for longer durations is essential because a commercial reactor will need to run continuously, not in brief pulses.
China’s EAST: breaking the Greenwald limit
China’s EAST tokamak achieved something physicists thought was not possible with current designs. The Greenwald limit is an empirical density ceiling: push plasma beyond it, and the reaction typically destabilizes violently, potentially damaging the reactor. EAST operated at 1.3 to 1.65 times the Greenwald limit by using electron cyclotron resonance heating during startup and carefully managing how the plasma interacts with the reactor walls. The results, published in Science Advances in January 2026, suggest a pathway to denser, more energetic plasmas in future machines.
Commonwealth Fusion Systems: assembling SPARC
CFS began physically assembling its SPARC tokamak in Devens, Massachusetts in early 2026. The first of 18 toroidal field magnets was installed in January, followed by the cryostat base in March. SPARC is designed to produce up to 140 megawatts of fusion power in 10-second bursts, demonstrating net energy gain with margin to spare. If successful, CFS plans to build a 400-megawatt commercial power plant called ARC in Virginia in the early 2030s.
ITER: slow but still pivotal
The international ITER project in southern France, funded by 35 nations at a cost exceeding €20 billion, continues construction toward its revised target of first deuterium-deuterium operation in 2035. In April 2026, the project received its fifth central solenoid magnet module and a new vacuum vessel sector. ITER remains the largest and most ambitious fusion experiment ever attempted, designed to produce 500 megawatts of fusion power from 50 megawatts of input, a tenfold energy gain (Q = 10).
NIF’s continued progress
The National Ignition Facility, which achieved ignition in December 2022, has continued refining its approach. NIF demonstrated that fusion energy output can exceed the energy delivered to the fuel, validating the physics of inertial confinement. While NIF’s laser-based approach is unlikely to become the basis for power plants (the lasers themselves consume far more energy than the fusion produces), its results have confirmed fundamental physics that benefits all fusion approaches.
Fusion vs. fission: a quick comparison
| Fusion | Fission | |
|---|---|---|
| Process | Combines light atoms | Splits heavy atoms |
| Fuel | Deuterium, tritium (from water and lithium) | Uranium, plutonium |
| Energy per kg | ~4x more than fission | High, but less than fusion |
| Waste | Helium (inert), some activated materials | Long-lived radioactive waste |
| Meltdown risk | None (reaction stops if conditions falter) | Possible (Chernobyl, Fukushima) |
| Weapons risk | Minimal | Significant (fuel can be weaponized) |
| Status | Experimental | Operational since 1956 |
One of fusion’s most appealing safety features is that the reaction is self-limiting. If the plasma cools or the magnetic field weakens, fusion simply stops. There is no chain reaction to spiral out of control, no risk of meltdown, and no possibility of a Chernobyl-style disaster.
When will fusion power reach the grid?
The honest answer: probably not before the late 2030s at the earliest, and commercial scale likely arrives in the 2040s.
Commonwealth Fusion Systems aims to have its ARC power plant producing electricity by the early 2030s. China has announced plans for a fusion engineering reactor called CFETR, with construction beginning around 2030. The UK’s STEP (Spherical Tokamak for Energy Production) program targets a prototype fusion power plant by 2040.
Private investment is accelerating the timeline. More than $7 billion in private capital has flowed into fusion startups as of 2025, compared to essentially zero a decade earlier. Companies like CFS, TAE Technologies, Helion Energy, and General Fusion are pursuing different approaches, and the competition is pushing innovation faster than government-funded projects alone ever did.
But significant challenges remain. No reactor has yet produced sustained net energy (where total energy out exceeds total energy in, including all the energy needed to run the magnets, lasers, and cooling systems). Materials that can withstand years of neutron bombardment inside a commercial reactor are still being developed. And the tritium breeding and handling systems needed for continuous operation have not been tested at scale.
Fusion will not replace fossil fuels overnight. But the progress of the past few years suggests it will not stay 30 years away forever, either.
Frequently asked questions about nuclear fusion
What is the difference between nuclear fusion and nuclear fission?
Fission splits heavy atoms (like uranium) into lighter ones, releasing energy. Fusion combines light atoms (like hydrogen isotopes) into heavier ones, also releasing energy. Fusion produces roughly four times more energy per kilogram of fuel, generates no long-lived radioactive waste, and cannot melt down.
Is nuclear fusion safe?
Yes. A fusion reaction is self-limiting: if the plasma cools or containment fails, the reaction simply stops. There is no chain reaction, no risk of meltdown, and no risk of explosion. The fuel (deuterium and tritium) is not weapons-grade, and the primary waste product is helium.
Why is nuclear fusion so hard to achieve?
Fusion requires temperatures of about 100 million degrees Celsius and sustained plasma confinement, conditions that push the boundaries of physics and engineering simultaneously. Plasma is inherently unstable and difficult to control, and no physical material can contain it directly.
When will nuclear fusion produce electricity?
The first demonstration fusion power plants are expected in the late 2030s. Commonwealth Fusion Systems, the UK STEP program, and China’s CFETR project are all targeting prototype reactors within the next 10 to 15 years. Commercial, grid-scale fusion power will likely follow in the 2040s.
How much fuel does a fusion reactor need?
Very little. A fusion power plant would consume roughly 250 kilograms of deuterium-tritium fuel per year. Deuterium is extracted from seawater (which is essentially unlimited), and tritium can be bred from lithium inside the reactor itself.
What to read next
If the physics of extreme energy interests you, explore how black holes warp space itself, or learn why entropy means everything eventually falls apart. For the technology side, see how quantum computing harnesses the weirdness of the subatomic world or how batteries store the energy we already have.