Every second, the sun turns millions of tonnes of hydrogen into helium, and a tiny fraction of that mass becomes pure energy. Sunlight. Heat. All of it, from atoms joining together.
That process is called fusion. And for about fifty years, fusion power has been the world’s favourite joke: always twenty years away, always just around the corner, never here.
But on October 8, 2026, a company in Richmond, British Columbia, cleared a bar that only a few fusion efforts have ever cleared — and it did it the slow way, the cheap way, the way the experts said probably wouldn’t work. This guide explains what General Fusion actually achieved with its LM26 machine, what it doesn’t mean, and why a machine in a suburb of Vancouver might matter to your electricity bill one day.
Fusion in plain language
Fusion is what happens when two light atoms — usually forms of hydrogen called deuterium and tritium — are pushed together so hard that they join into one heavier atom. When they join, a little bit of mass disappears, and that missing mass becomes energy. A lot of energy, from a very small amount of fuel.
The fuel is effectively everywhere. Deuterium is sitting in ordinary seawater. But getting atoms to fuse takes three things at once:
- Temperature. The atoms have to be moving fast enough to overcome the force that pushes them apart. That means plasma — the fourth state of matter — at tens of millions of degrees.
- Density. There have to be enough atoms packed close together that they actually collide and fuse, instead of just flying past each other.
- Time. The plasma has to stay hot and dense long enough for a useful number of fusions to happen before it cools down or falls apart.
Those three conditions have a name: the Lawson criterion. Get all three high enough at the same time, and the plasma produces more energy than it takes to keep it going. That is the whole game — and why fusion is hard. Any one of the three is easy. All three at once is not.
Fusion power, if it works, is the prize everyone wants. No carbon dioxide while it runs. Fuel from seawater. No long-lived nuclear waste like today’s fission plants. That is why governments and investors keep coming back to it, decade after decade, even after all the missed deadlines.
Three ways to bottle a star
There are, broadly, three ways people have tried to build a fusion machine.
Magnetic confinement is the tokamak: a giant metal donut in which powerful magnets hold a ring of plasma in place, like an invisible bottle, while it is heated to extreme temperatures. It is the approach behind the biggest fusion project on Earth, the international ITER machine in France. The drawback is right there in the size: superconducting magnets chilled to near absolute zero, huge complexity, enormous cost.
Laser confinement takes a tiny pellet of fuel, about the size of a peppercorn, and hits it with the world’s most powerful lasers — all at once, from every direction. The pellet implodes in a billionth of a second, and for that instant it becomes hot and dense enough to fuse. The drawback is the scale of the machinery for the size of the bang. The lasers are the size of a stadium, the pellet is the size of a peppercorn, and you would need to do it over and over again, many times a second, to run a power plant. Nobody has done that yet.
Magnetized target fusion is the third way, and the one this story is about. Instead of a giant magnet or a giant laser, you start with a magnetized plasma and then squeeze it — mechanically, with a metal wall that closes in on it, like a piston.
Think of it this way. If you pump up a bicycle tyre, the air inside gets warm: compressing a gas heats it up. Magnetized target fusion is the same idea, taken to an extreme. The drawback is that it is the least proven of the three at scale. The squeeze has to be almost perfectly even, or the plasma wobbles, leaks heat, and cools before it can do anything useful. For twenty years, the question has been whether that squeeze could ever heat a plasma enough to matter.
The BC company
General Fusion was founded in 2002 in Richmond, British Columbia. For more than twenty years, it has worked on one idea: that fusion doesn’t have to be a giant, exotic machine. It can be practical — built from existing materials, with no superconducting magnets and no stadium-sized lasers. A liquid metal wall does three jobs at once: it protects the machine from the neutrons fusion produces, it breeds more fuel, and it captures the heat so it can make steam and spin a turbine, the way an ordinary power plant does.
In the summer of 2026, the company listed on the Nasdaq under the ticker GFUZ — the first fusion company ever to be publicly listed. The company describes itself as TIME’s No. 1 GreenTech Company of 2026; that is the company’s own claim, and TIME’s list was not independently verified for this guide. The company’s whole commercial plan is anchored by one machine: LM26, the Lawson Machine 26.
LM26 is the world’s first commercially relevant-scale magnetized target fusion demonstration machine — roughly half the diameter of a commercial-scale machine, by the company’s own design figures. It ran its first shots in 2025.
In June 2026, it produced its first real result. The machine squeezed a plasma with a lithium liner and heated its electrons to about 0.72 keV (±0.08) — roughly 8.4 million degrees Celsius — more than a threefold rise in temperature during the squeeze alone, across the first eleven compression shots. The company published the technical paper publicly and submitted it for peer review. It was a real step. And it was still a long way from the first real milestone: 1 keV, about ten million degrees.
That is the ladder this story climbs. In October, the machine took its next step up it.
What LM26 actually measured on October 8
On October 8, 2026, General Fusion announced the result jointly with the UK Atomic Energy Authority (UKAEA), Britain’s national fusion laboratory. The way they announced it matters as much as the result itself: they published the papers at the same time and submitted them for peer review.
Here is what they measured, using two different instruments:
- Thomson scattering — the gold standard for measuring plasma temperature — fired a laser through the plasma and read how the light scattered. Just before peak compression, the electrons hit approximately 1.1 keV, about 12.6 million degrees Celsius. The technique has history: UKAEA scientists and Russian researchers first used Thomson scattering to verify 1 keV electrons in a Russian tokamak in 1969. For LM26, UKAEA built a key diagnostic component, a polychromator, at its Diagnostic Innovation Centre of Excellence (DICE) at the Culham Campus in England.
- An absolute extreme ultraviolet diagnostic (AXUV) measured approximately 1.2 keV, about 13.7 million degrees, at the moment of peak compression itself. Two instruments, two different methods, the same heating trend.
And there was a third number. During the compression, the fuel ions — the actual atoms that would fuse — also heated up, to approximately 0.46 keV, up to double where they started. The squeeze took milliseconds (about three milliseconds from the start of compression to peak temperature, versus microseconds or nanoseconds for other compression approaches), and temperature, density, and magnetic field all rose together, the way they should if the plasma is staying stable under compression.
The precise figures tabulated from the papers for the October experiment (shot LMC-12) are: Thomson 1.090 ± 0.040 keV; AXUV 1.180 ± 0.065 keV; deuterium ion temperature 0.459 ± 0.033 keV. This guide, like the company’s announcement, uses the rounded figures.
The two caveats, stated plainly
This is where a lot of retellings would stop: a world first, a milestone hit. All true — but not the whole truth. Two caveats belong in the story, not in a footnote.
Caveat 1: Electrons are not the fuel. The headline number, 1.1 keV, is an electron temperature. But fusion happens between the fuel ions — the heavy atoms, the deuterium. The ions here reached about 0.46 keV. That is real heating — they roughly doubled — but it is less than half the electron temperature, and a long way from the roughly 10 keV that a burning fusion plasma needs. Heating the electrons is a necessary step. It is not the same thing as heating the fuel.
Caveat 2: Not yet peer reviewed. The papers describing these results have been submitted for peer review, but they have not been published yet. The Thomson-scattering paper, co-authored with UKAEA, has been submitted to the journal Review of Scientific Instruments. As of November 12, 2026, no published version had appeared, and the paper was available as a preprint. Until review is complete, these numbers are a company-reported result, backed by a national laboratory’s co-authorship — but not yet an independently verified finding. No other machine has reproduced them.
What comes next
The roadmap has three rungs, and the first one is done.
Rung two is 10 keV — about one hundred million degrees Celsius, ten times hotter than what LM26 just reached. Upgrades to the machine are already underway to support the higher compression ratios that step will need, and General Atomics Energy Group is advising on the diagnostic system for it. Ten keV is the temperature range where a deuterium-tritium plasma genuinely starts to burn, where fusion reactions produce enough heat to sustain themselves.
Rung three is the Lawson criterion itself: temperature, density, and confinement time, all high enough at the same time, so the plasma produces more energy than it takes to keep it going. Coverage of the announcement reported the company’s target as the end of 2028 for that milestone. Treat that date the way it was reported — as a target, from coverage of the announcement, not as a promise in the company’s own release, which says only that the program will advance to the Lawson criterion after 10 keV. Targets in fusion have slipped before. This one will be judged by results, the same way the 1 keV result was.
And after that comes the engineering: a machine that does the same thing over and over, reliably, thousands of times a day, breeding its own fuel, capturing its own heat, making steam, spinning a turbine — and doing it cheaply enough to compete. General Fusion’s stated goal is to begin operating a first-of-a-kind plant around 2035. Grid power from fusion is a 2030s prospect, not something imminent.
There are real risks on that road, and the company’s own filings name them. The peer review now underway could challenge the October results. Squeezing harder, toward 10 keV, gets more difficult, not less. And General Fusion is a development-stage company with no revenue from fusion energy yet.
Why it matters for Canada
This is a British Columbia company, built over two decades with support from the Government of Canada’s Strategic Response Fund and other programmes. The measurement that made the headline was made together with the UK’s national fusion laboratory. The machine is in Richmond, the team is Canadian, and the world is watching.
It also answers a question that has hung over this approach for twenty years: can you heat a plasma by squeezing it — slowly, mechanically, with ordinary materials — at a scale that matters? In June, the answer was: a little. In October, the answer was: yes, to 1 keV, with the ions heating too, and with the plasma staying stable while it happens. The idea that was a sketch on a whiteboard in Richmond in 2002 now has numbers attached to it, measured by two instruments, co-published with a national laboratory, and submitted for the world to check.
That is not a power plant. It is a proof that the path is real.
The bottom line
Fusion has been twenty years away for fifty years. This result doesn’t end that joke — but it puts a date, a machine, and a measurement on the next step. And for the first time in a long time, the next step is happening in British Columbia.
What to watch next: the outcome of peer review at Review of Scientific Instruments, the machine upgrades now underway for the 10 keV step, and whether the Lawson-criterion target holds. This guide will be updated when peer review concludes or a new milestone is announced.
Sources
General Fusion, “General Fusion Achieves World-First 1 keV Electron Temperature Milestone on its Path to Commercial Fusion Power,” October 8, 2026 — https://generalfusion.com/post/general-fusion-achieves-world-first-1-kev-electron-temperature-milestone-on-its-path-to-commercial-fusion-power/
General Fusion / UKAEA, “General Fusion, UKAEA Announce Magnetized Target Fusion Milestone,” October 8, 2026 — https://generalfusion.com/post/general-fusion-ukaea-announce-magnetized-target-fusion-milestone/
General Fusion, “General Fusion Achieves Compressional Plasma Heating with LM26 Magnetized Target Fusion Machine,” June 22, 2026 — https://generalfusion.com/post/general-fusion-achieves-compressional-plasma-heating-with-lm26-magnetized-target-fusion-machine/
S. J. Howard et al., “The science of compressional heating on the LM26 magnetized target fusion experiment,” arXiv:2606.23974 (June 2026 preprint, submitted for peer review) — https://arxiv.org/abs/2606.23974
Thomson-scattering paper preprint, arXiv:2610.11769 (October 9, 2026; submitted to Review of Scientific Instruments, peer review pending as of November 12, 2026) — https://arxiv.org/abs/2610.11769
Seeking Alpha, “Fusion’s Prove-It Era Just Delivered a World-First Result,” October 8, 2026 — https://seekingalpha.com/pr/20680333-fusion-s-prove-it-era-just-delivered-a-world-first-result
Attribution notes: the end-of-2028 Lawson-criterion target is from coverage of the announcement, not from General Fusion’s own release. The TIME No. 1 GreenTech Company of 2026 description is the company’s own claim. All keV figures are company-reported, co-authored with UKAEA, and pending peer review.
