You’ve probably got a power bank in a drawer somewhere, the kind that tops your phone back up at 20 or 30 watts. Now picture the same basic idea, energy stored up and dumped out on command, scaled to 50 megawatts. That’s roughly what an Israeli fusion company just switched on, and it built the thing to do one very specific job: cook plasma.<\/p>\n
The company is nT-Tao, a fusion startup working out of Hod Hasharon, and it announced on September 15<\/a> that it had deployed a new pulsed-power subsystem for its C3 fusion prototype. The bank stacks 50 modules the company calls Power Electronic Generators, each rated at roughly 1 megawatt, wired into ten branches of five and designed to run anywhere from 50 to 2,000 kHz. The earlier generation used 12 of the same modules for about 10 megawatts, so 50 modules and 50 megawatts is a fivefold jump in one step. nT-Tao files all of this under its MEGA platform, short for Modular Energy Generator Architecture, which is a fancy name for a very large, very angry power bank.<\/p>\n Pulsed power, if you haven’t run into the term, just means storing energy fairly slowly and releasing it in short, violent bursts. Your camera flash does it. So does a defibrillator. nT-Tao does it at a scale meant to heat hydrogen until it stops behaving like a gas at all.<\/p>\n Fusion means forcing light atomic nuclei to merge, which releases energy, and nuclei only merge when they’re moving fast enough to overcome the fact that they repel each other. Fast, in this context, means hot. Heat any gas far enough and the electrons get stripped away from the atoms, and what’s left is plasma, a soup of charged particles that conducts electricity and responds to magnets. That’s the property everything here hangs on, because it means you can hold plasma in place with magnetic fields and pump energy into it electrically.<\/p>\n nT-Tao’s pitch is a fusion machine compact enough to ship in a container, running plasma at densities the company claims are about 1,000 times higher than conventional approaches. Heating something that dense inside a small machine takes a lot of power delivered very fast, and that’s the job the new bank exists for. The 50 megawatts describes what the electronics are rated to push in a burst, by the way, not anything the machine generates. The bank consumes power; it doesn’t make any.<\/p>\n So what makes this hard, beyond the wattage?<\/p>\n Think about pushing a kid on a swing. Time your pushes to the swing’s natural rhythm and a small effort builds into a big arc. Push at the wrong moment and the swing pushes back at you. The circuit feeding energy into a plasma works on pretty much the same principle (engineers call it resonance), except this swing changes its own rhythm while you’re mid-push. As the plasma heats up during a pulse, its electrical characteristics shift within fractions of a millisecond, and if the drive circuit doesn’t follow, the energy transfer collapses and power gets reflected back into your expensive electronics instead of landing in the plasma.<\/p>\n nT-Tao’s answer is a real-time nonlinear controller that tracks those shifts and retunes the system on the fly, so all 50 modules keep behaving like a single source. The company developed the method with researchers from Ben-Gurion University of the Negev and published it in the journal Actuators in December 2025, as Nuclear Engineering International<\/a> reported. I’ll flag one thing about that paper: the validation came from detailed time-domain simulations of laboratory plasma conditions, not from measurements on this new 50-megawatt bank. Simulation is how you build a controller, but it isn’t the same as watching it survive contact with a real shot.<\/p>\n “The hard part isn’t generating 50 MW of pulsed power,” said Boaz Weinfeld, nT-Tao’s co-founder and CTO, in the announcement, arguing the real trick is getting 50 modules to act as one precisely tuned system rather than fifty separate ones. If that framing sounds familiar, we’ve covered a tokamak controller that corrects its plasma 50 times a second<\/a>. nT-Tao’s version of the problem is the same idea squeezed into the lifetime of a single pulse.<\/p>\n Frankly, the modularity argument is what I find genuinely clever here. Building the bank out of 1-megawatt bricks means scaling up is mostly a manufacturing exercise, repeat a proven unit instead of engineering a bigger one, and manufacturing problems are arguably the kind you’d want to have in fusion, because factories are good at them.<\/p>\n So where does the machine on the receiving end actually stand? C3, the prototype this bank feeds, fired its first plasma pulses<\/a> a bit over two months after assembly started, according to the company, which announced the milestone on January 15. Its predecessor, C2-A, reached around 1.8 million degrees Fahrenheit (about one million Celsius, or roughly 100 electronvolts if you speak plasma physics). nT-Tao’s September announcement points to about 2.7 million degrees Fahrenheit (1.5 million Celsius) as already demonstrated on its machines, and the new bank exists specifically to push C3 past that.<\/p>\n Here’s the honest context you should hold next to those numbers. The big deuterium-tritium programs, ITER and the large tokamaks, aim for plasma in the neighborhood of 180 to 270 million degrees Fahrenheit (100 to 150 million Celsius), and private outfits chase the same wall from other angles, like the Canadian machine that crushes its plasma inside collapsing lithium<\/a>. nT-Tao’s position is that its high-density approach changes what a compact machine needs, and maybe it does, but the temperature gap between where the company sits and where fusion happens is still a couple orders of magnitude. No one at nT-Tao is pretending otherwise; the company itself frames the new bank as one step toward its next milestone.<\/p>\n There’s also a decent list of things nT-Tao hasn’t published yet: what that next temperature milestone actually is, how much of the 50 megawatts lands in the plasma per pulse, how long each pulse lasts, and what C3 has reached since the new hardware went in. So I can’t tell you from the outside whether the bank is earning its keep. The next set of C3 temperature numbers will answer that one way or the other.<\/p>\n The end goal, per the company, is a container-sized system putting out 20 megawatts of electricity for data centers, industrial sites and ships. On September 21, nT-Tao also signed a memorandum of understanding<\/a> with Orion Nuclear Energy Corporation, a US nuclear project developer, to scout deployment opportunities in the United States. An MOU is a framework, not a sale, so I wouldn’t hang much on it yet.<\/p>\n The hardware, though, is real, and it’s running. nT-Tao switched the 50-module bank on at its Hod Hasharon headquarters and announced the deployment on September 15, 2026, eight months after it announced C3’s first plasma.<\/p>\n","protected":false},"excerpt":{"rendered":" You’ve probably got a power bank in a drawer somewhere, the kind that tops your phone back up at 20 … <\/p>\nSo why does a fusion machine need a 50-megawatt power bank?<\/h2>\n
The plasma won’t hold still<\/h2>\n
2.7 million degrees is hot, but fusion wants a lot more<\/h2>\n