Every few months, a headline declares that nuclear fusion — the process that powers the sun — has taken a giant leap toward solving our energy problems. Then nothing seems to change. So what is actually going on? Is fusion finally close, or is it still, as the old joke goes, "thirty years away and always will be"?
The truth is more interesting than either extreme. Fusion has genuinely crossed a scientific threshold that stood for fifty years — and it remains separated from the electricity grid by engineering problems that have almost nothing to do with plasma physics. Understanding which is which is the whole trick to reading fusion news sensibly.
What Fusion Actually Is
Fusion is the opposite of the fission used in today's nuclear plants. Fission splits heavy atoms like uranium. Fusion combines light atoms — usually two isotopes of hydrogen, deuterium and tritium — into helium, releasing energy in the process. Our full fusion vs. fission comparison covers the differences in depth.
The energy comes from a small discrepancy in mass. A helium nucleus weighs slightly less than the deuterium and tritium that formed it, and that missing mass emerges as kinetic energy carried mostly by a single high-speed neutron. It is a tiny difference per reaction, but run it across enough atoms and the numbers become extraordinary.
The appeal is genuine. Deuterium can be extracted from ordinary seawater. There is no chain reaction to run away, so a meltdown in the fission sense is physically impossible — if conditions fail, the reaction simply stops. And fusion produces no long-lived, high-level radioactive waste.
The catch is that making it work on Earth is one of the hardest engineering challenges humanity has ever attempted.
Why It Is So Hard
To fuse, two nuclei must get close enough for the strong nuclear force to bind them. But nuclei are positively charged, and like charges repel — the Coulomb barrier. Overcoming it requires temperatures of roughly 100 million degrees Celsius, several times hotter than the core of the sun.
At those temperatures matter becomes plasma, a turbulent, electrically charged fourth state of matter that no physical container can touch. And heat alone isn't enough. Physicists judge a fusion attempt by the triple product — plasma density, temperature, and how long you hold it together. Push all three high enough simultaneously and the reaction sustains itself; fall short on any one and it fizzles. Decades of fusion research are essentially a fight to raise that product.
Two approaches dominate:
- Magnetic confinement uses powerful magnetic fields to trap plasma in a doughnut-shaped chamber called a tokamak, holding it for seconds or longer at relatively low density. This is the approach of ITER in France.
- Inertial confinement blasts a peppercorn-sized fuel capsule with the world's most powerful lasers, crushing it so violently that it fuses in billionths of a second, before it can fly apart. This is the approach of the US National Ignition Facility (NIF).
The Breakthrough That Mattered
In December 2022, NIF achieved something genuinely historic: ignition. For the first time, a fusion reaction produced more energy than the laser light delivered to the fuel capsule — about 3.15 megajoules out from 2.05 megajoules in.
What is less widely reported is that it was not a one-off. NIF has now reached ignition 11 times, and the yields have climbed substantially. The current record came on 7 April 2025: 8.6 MJ of fusion energy from 2.08 MJ of laser energy — a target gain of 4.13, roughly quadruple the input. A June 2026 shot produced 7.9 MJ at a gain of about 3.8, confirming that multi-megajoule results are now repeatable rather than lucky.
| NIF milestone shot | Laser energy in | Fusion energy out | Target gain |
|---|---|---|---|
| 5 December 2022 (first ignition) | 2.05 MJ | 3.15 MJ | ~1.5 |
| 12 February 2024 | 2.2 MJ | 5.2 MJ | ~2.4 |
| 7 April 2025 (record) | 2.08 MJ | 8.6 MJ | 4.13 |
| 20 June 2026 | — | 7.9 MJ | ~3.8 |
That is real, repeated scientific progress. But it comes with an asterisk that matters enormously.
The Asterisk: Three Different Ways to Count
"More energy out than in" depends entirely on where you draw the boundary.
Target gain counts only the laser energy that reaches the capsule. By this measure NIF is comfortably above break-even — gain 4.13.
Facility energy counts the electricity needed to fire those lasers, and this is where the picture changes. NIF's power conditioning system stores around 400 MJ and delivers roughly 330 MJ of electrical energy per shot to produce about 2 MJ of ultraviolet laser light — a wall-plug efficiency near 0.7%. Set the record 8.6 MJ yield against ~330 MJ drawn from the grid and the facility consumed roughly forty times more energy than the reaction released.
Grid energy would additionally count the losses in converting fusion heat into electricity, plus everything needed to run the plant.
None of this makes the achievement fake. NIF was built for nuclear-weapons stockpile stewardship, not electricity generation, and nobody designed its 1990s-era lasers for efficiency. But it does explain why a genuine scientific triumph did not change your electricity bill.
Two World Records That Measure Different Things
Fusion has a second headline number, and it comes from the magnetic side. In its final deuterium-tritium campaign, the Joint European Torus (JET) in Oxfordshire produced 69 megajoules over about five seconds — from roughly 0.2 milligrams of fuel. That is far more total energy than any NIF shot.
But JET's gain was below one: it consumed more energy than it produced. So NIF holds the record for efficiency of the reaction, and JET for total energy released and sustained power. Reading fusion news well means always asking which quantity a record refers to.
JET fired its last pulse in December 2023 and has been retired. Its data now underpins ITER.
The Engineering Wall Nobody Headlines: Tritium
Here is the problem that rarely makes the news but may matter more than plasma physics.
Deuterium is abundant. Tritium is not. It is radioactive with a half-life of about 12 years, so it decays away rather than sitting in a warehouse. Estimates of the entire global civilian inventory run to roughly 20 to 30 kilograms — for a fuel that commercial plants would consume by the tens of kilograms per year.
The intended solution is elegant: line the reactor with a lithium "breeding blanket." Each fusion reaction fires out a high-energy neutron; when that neutron strikes lithium, it produces fresh tritium. A plant would then breed its own fuel.
Elegant, but unproven at scale. A blanket must breed more tritium than the reactor burns, because tritium is lost to decay, absorbed into structural materials, and held up in processing. Only a small fraction of injected tritium actually fuses in a single pass, so nearly all of it must be recovered, purified, and re-injected. It also has to survive years of intense neutron bombardment — which demands structural alloys that, in several cases, do not yet have industrial supply chains.
As Physics World has put it, the industry has to rise to its tritium challenge. Any timeline that assumes plasma performance is the last hurdle is skipping this one.
The Private-Sector Surge
For decades fusion belonged to governments and giant facilities. That has changed sharply.
The Fusion Industry Association's 2026 survey of 56 companies found cumulative investment in private fusion had reached $14.24 billion, of which $13.26 billion came from private investors. The 12 months to July 2026 were the sector's biggest yet, drawing a record $4.48 billion.
The technical driver behind much of this is high-temperature superconducting magnets, which generate far stronger fields than the older superconductors ITER was designed around. Stronger fields mean a smaller, cheaper machine for the same plasma performance — which is what makes privately funded reactors conceivable at all. Several companies now publicly target grid electricity in the 2030s.
Competition and capital do accelerate engineering in ways slow megaprojects cannot. Whether any individual company hits its aggressive date is uncertain; that the field is moving faster than at any point in its history is not.
Where the Big Machines Actually Stand
ITER remains the largest fusion project ever attempted, and its schedule is the field's most instructive reality check. In 2024 the ITER Council adopted a revised baseline: start of research operations in 2034, full magnetic energy in 2036, and the start of deuterium-tritium operations in 2039 — a delay of several years against the previous plan, at roughly €5 billion in additional cost.
The reasoning was deliberate rather than purely a stumble: ITER chose to begin experiments with a more complete machine rather than rush an early, limited first plasma. But it is a useful corrective. Fusion timelines have slipped consistently for fifty years, and the burden of proof sits with anyone promising otherwise.
So, How Close Are We?
A realistic reading of the field in 2026:
- Scientific feasibility: settled. Fusion produces net energy at the reaction level, repeatedly.
- Engineering break-even at the facility level: not yet demonstrated by anyone, and roughly a factor of forty away at NIF specifically.
- A working demonstration power plant: plausibly in the 2030s if the private sector's magnet-led bet pays off; delays are the historical norm.
- A closed tritium fuel cycle at scale: the genuine long pole, and less advanced than plasma performance.
- Fusion as a meaningful slice of the grid: more likely the 2040s and beyond.
Fusion is almost certainly not going to help with this decade's climate targets — solar, wind, storage, and existing fission are the tools for that job, as we cover in our look at whether carbon capture can close the gap. But as clean, abundant baseload power for the second half of the century, fusion has moved from science fiction into a serious engineering race.
Frequently Asked Questions
Is nuclear fusion the same as today's nuclear power?
No. Today's reactors use fission, which splits heavy atoms like uranium and leaves long-lived, high-level radioactive waste. Fusion combines light hydrogen isotopes into helium. There is no chain reaction, so a runaway meltdown is physically impossible — if conditions fail, the plasma simply cools and the reaction stops. Fusion does make reactor materials radioactive over time through neutron bombardment, but that waste is far shorter-lived than fission's.
Did scientists really achieve fusion "net energy gain"?
Yes, with an important qualifier. Since December 2022 the National Ignition Facility has reached ignition 11 times, with a record 8.6 MJ of fusion energy from 2.08 MJ of laser energy — a target gain of 4.13. But that accounting covers only the laser light hitting the capsule. Firing those lasers draws roughly 330 MJ of grid electricity, so the facility still consumed around forty times more energy than the reaction released.
Why is tritium such a problem?
Tritium is radioactive with a roughly 12-year half-life, so it cannot be stockpiled, and the entire global civilian inventory is estimated at only about 20 to 30 kilograms. A commercial plant would burn tens of kilograms a year. The plan is for reactors to breed their own by having fusion neutrons strike a lithium blanket — but that blanket must produce more tritium than the reactor consumes, despite losses to decay, material absorption, and processing, and must survive years of neutron damage.
What's the difference between the NIF and JET records?
They measure different things. NIF holds the record for reaction efficiency: more fusion energy out than laser energy in, gain 4.13. JET holds the record for total energy from a single sustained shot — 69 megajoules over about five seconds from roughly 0.2 milligrams of fuel — but its gain was below one, meaning it consumed more than it produced. Neither has achieved net energy at the level of the whole facility.
When will fusion actually power the grid?
A demonstration plant is plausible in the 2030s, and several privately funded companies target that decade on the strength of high-temperature superconducting magnets. Fusion becoming a meaningful share of electricity supply is more realistically a 2040s-and-beyond prospect. ITER, the largest project in the field, now plans deuterium-tritium operations for 2039 — and fusion schedules have slipped consistently for fifty years.
Can fusion help fight climate change this decade?
No, and it is worth being blunt about it. Fusion will not arrive in time to affect targets for the 2020s or early 2030s. Solar, wind, batteries, grid upgrades, and existing fission are the tools that matter on that timescale. Fusion's value proposition is different: clean, dense, weather-independent baseload power for the second half of the century, in a world that will need enormous amounts of firm electricity.
Is fusion research a waste of money if it's decades away?
That depends on what you compare it to. Annual private fusion investment — $4.48 billion in the year to July 2026 — is small against global energy capital spending, and the technology's upside is a fuel source drawn largely from seawater with no long-lived waste. The reasonable criticism is not that fusion is funded, but that its timelines are routinely presented with more confidence than the engineering justifies.
The Bottom Line
Nuclear fusion is no longer a question of whether the physics works — that part is settled, and settled repeatedly. It is now a question of engineering, cost, and time.
Read the records carefully and the shape of the challenge becomes clear. NIF has quadrupled the energy out of a reaction relative to the laser energy in, which is a genuine scientific triumph. It is also still about forty times short of break-even once you count the electricity that fires the lasers, and the fuel cycle that any real plant depends on has not been closed at scale anywhere.
"Always thirty years away" is finally starting to look out of date. But the honest version of fusion optimism is specific: the physics is done, the magnets are improving fast, the money has arrived — and the boring problems of neutrons, materials, and tritium now decide the timeline.
Sources
- Achieving Fusion Ignition — National Ignition Facility, Lawrence Livermore National Laboratory
- Power Conditioning System — National Ignition Facility, Lawrence Livermore National Laboratory
- New baseline to prioritize robust start to exploitation — ITER Organization
- JET tokamak's final fusion energy record — EUROfusion
- The Global Fusion Industry in 2026 — Fusion Industry Association
- The fusion industry must rise to its tritium challenge — Physics World, IOP Publishing
Related on PrimusSource: Nuclear Fusion vs Fission: What's the Difference?, How Lasers Ignited a Star: The NIF Ignition Breakthrough Explained, The Photon: Light's Strangest Particle and more in our physics topic hub.
