A gasoline engine splits its work in two. The piston squeezes the mixture, then the spark plug lights it, and neither part has to be any good at the other’s job. A diesel skips the plug and makes the squeeze do both, which works fine in a truck. According to a British fusion company, it’s also the expensive way to do fusion, and on September 22 that company announced it’s got the squeezing half working, at least on test targets that weren’t trying to make any fusion.<\/p>\n
The company is First Light Fusion, based in Yarnton, just north of Oxford, and the machine is M3, a 44-ton bank of capacitors built around a vacuum chamber in the company’s own building. According to First Light’s announcement<\/a>, a series of shots on M3 crushed a target built from several nested shells, and did it in the controlled, staged way the company’s fusion design calls for. First Light was careful to spell out what the experiments weren’t. They didn’t ignite anything or measure fusion gain. They tested the compression step on its own.<\/p>\n Which is a smaller claim than the phrase “fusion milestone” usually comes with, and frankly a more believable one.<\/p>\n A capacitor’s basically a battery with no patience. It can’t hold much, but whatever it holds, it’ll hand over all at once. M3’s got 192 of them, arranged in six banks around a vacuum chamber about 5.6 feet across, and the whole setup sits inside an enclosure roughly 52 feet on a side. First Light’s own spec sheet<\/a> puts the machine at 44 tons, with about 9 miles of cable running through it, 300 diagnostic probes watching every shot, and enough thin plastic insulation film to cover almost four football fields.<\/p>\n On a typical shot the machine runs at 140,000 volts, which stores 1.22 megajoules of energy. That reads like a weapons-grade number. It’s about a third of a kilowatt-hour, which is what your 1,500-watt space heater burns in 14 minutes or so. The violence comes from the clock. M3 dumps that energy in about two millionths of a second, and the current peaks at 10 million amps.<\/p>\n About those 10 million amps. The pitch during construction was more than 14 million, close to 500 lightning strikes landing at once in the company’s own comparison, and you’ll still see that figure around. The spec sheet today says 10, and I can’t find anywhere that First Light explains the gap. My read is that 14 million’s what the bank does at its full 200,000-volt rating and 10 million’s what you get at the 140,000 volts it typically runs at, since peak current in a machine like this scales with the charge voltage, and 140 is 70 percent of 200. First Light hasn’t spelled that out anywhere I can see, so that’s my math, not theirs.<\/p>\n Each shot’s a bit of an event, too. First Light says M3 fires once every other day during a campaign, and the company designed and built it in 18 months for about $4.6 million, \u00a33.6 million at the exchange rate when it opened. For a 10-million-amp machine, that’s pretty cheap. The cheapness is sort of the whole argument. We’ll get to that.<\/p>\n In this business a “liner” is just a thin metal tube. Run enough current down it and the magnetic field that current makes squeezes the tube inward on itself. First Light’s standard version is aluminum, 0.63 inches across, a bit fatter than a AA battery, with a piece of acrylic (the clear plastic big aquarium walls are made of) sitting in the middle to take the hit.<\/p>\n How fast does the tube collapse? Faster than anything you’d want near you. In the platform experiments First Light’s Oliver Nash presented at an American Physical Society meeting<\/a>, a laser velocity gauge (photonic Doppler velocimetry, basically a radar gun that uses light instead of radio waves) tracked the inner wall of the tube moving inward at about 19,000 mph, call it 5.3 miles every second, by the time it had closed from a radius of 0.31 inches to 0.18. That’s the house specialty. Take a tube and make it a much smaller tube, very quickly.<\/p>\n The September shots used a more complicated target. Instead of one tube, the shells are nested, and First Light says the geometry of the nest is what turns one fairly crude electrical pulse into several shock waves arriving on a schedule, each one squeezing the inside a little harder without heating it too early. The point, in the words of Mark Thomas, First Light’s CEO, is to “simplify the machine by putting more functionality into the target.” Jeremy Chittenden, the Imperial College plasma physicist who chairs the company’s science advisory board, called it a significant step and made a point of noting that the generator doing the driving was a low-voltage one.<\/p>\n So is this fusion fuel getting squeezed to fusion conditions? Not yet. Chittenden’s wording is “materials,” the announcement says the next phase moves toward “fusion relevant fuel conditions,” and the integrated test that adds the rapid heating comes only after that. There’s no date on either step yet, and the announcement doesn’t say what hardware would deliver that second burst in a finished system. I’m not going to guess.<\/p>\n Here’s how the economics go, as First Light tells it. The machines that compress fusion fuel today have to deliver a huge burst of energy with extreme precision, and every shot beats on the hardware. The company’s estimate is that a FLARE-style compression driver (FLARE is its name for the two-step design) could come in at a tenth the cost of comparable inertial fusion systems. That’s the company’s number, not mine, and “could ultimately” is doing a fair bit of work in it. But the logic holds up. If the target handles the timing, the machine’s job shrinks to delivering a lot of current on schedule, and a machine with one job’s cheaper to build and cheaper to keep running.<\/p>\n For scale, the biggest pulsed-power machine on Earth is Z, at Sandia National Laboratories in Albuquerque. We covered Z recently<\/a>: 2,160 capacitors, 26 million amps, 20 megajoules dumped in about 100 billionths of a second. First Light knows that machine well, since it’s run its own targets on Z and set the facility’s record for pressure in quartz doing it. On a typical shot M3 stores 1.22 megajoules to Z’s 20 and peaks at 10 million amps to Z’s 26, and First Light describes it as the second-highest-current pulsed-power machine running anywhere right now. I haven’t checked every capacitor bank on Earth, so take that as the company’s claim.<\/p>\n I’d argue that comparison is the whole story. First Light built M3 for the price of a decent house in Palo Alto, and it’s now using that machine to show a fusion design doing its compression on a budget driver. The company stopped planning its own power plant a while back and now pitches FLARE as a fuel system someone else’s machine could run.<\/p>\n Still, a cheap compression stage is one stage. ITER, the giant tokamak in southern France, just took delivery of the last 440-ton slice<\/a> of a plasma chamber that’ll weigh 5,732 tons, and that project’s been collecting parts for years. A 44-ton machine in an industrial park outside Oxford crushing a tube narrower than a quarter isn’t competing with that, and from the looks of things First Light doesn’t claim it is. It’s one piece of a different approach, in the same broad family as the Canadian machine that collapses a lithium can<\/a> around its plasma. Lighting the fuel once it’s squeezed is a separate problem, and First Light’s own announcement says these shots didn’t try to do that.<\/p>\n First Light published the result on September 22, 2026, and the company says the next phase moves its nested-shell targets toward the fuel conditions fusion needs, with the heating step coming only after that.<\/p>\n","protected":false},"excerpt":{"rendered":" A gasoline engine splits its work in two. The piston squeezes the mixture, then the spark plug lights it, and … <\/p>\nSo what is M3, exactly?<\/h2>\n
What actually gets crushed?<\/h2>\n
So why would you want a weaker machine?<\/h2>\n