100 Million Fusion Reactions a Second, 0.14 Milliwatts of Power: The Honest Math of ENN's Hydrogen-Boron Milestone
China's ENN Group fused hydrogen and boron 100 million times a second, a first for a commercial fusion company. Run the arithmetic and the machine produced roughly 0.14 milliwatts, while a gigawatt power plant would need seven trillion times more. Both numbers are real. Only one of them is a headline.
0.14 milliwatts.
That is the fusion power produced by the milestone everyone is celebrating. One hundred million fusion reactions every second, and the energy output is roughly what a grain of rice would release as heat if you dropped it from a table. A phone charger running at five watts puts out thirty-six thousand times more power. Your body radiates about seven hundred thousand times more.
On September 28, 2026, China's ENN Group announced that its EXL-50U spherical tokamak had achieved hydrogen-boron fusion at a rate exceeding 100 million reactions per second, which the company described as the first hydrogen-boron fusion reaction achieved by a commercial fusion company on its own machine (StartupNews.fyi; Good News Network). Its chief engineer shared the milestone on social media that morning, and some outlets reported the news in the days that followed. ENN drove the rate by combining high-energy neutral beam injection with radiofrequency waves, generating a population of "fast protons" inside the plasma that act as a catalyst, colliding with boron nuclei and triggering the reaction again and again.
What makes this interesting is the reaction itself. A proton plus boron-11 yields three helium nuclei, alpha particles, and 8.7 million electron-volts of energy, with no neutrons (PMC peer-reviewed paper). That "no neutrons" clause is the entire reason anyone bothers with a fuel this difficult. Deuterium-tritium fusion, the reaction every mainstream program chases, sprays 14-MeV neutrons that batter the reactor walls, activate the surrounding structure, and demand exotic materials that do not exist yet at industrial scale. Hydrogen-boron throws out charged particles whose energy could, in principle, be converted directly into electricity, no steam turbine required, while the fuel is abundant and non-radioactive. If it worked, the plants would be simpler than D-T plants, not harder.
Now the arithmetic, with the inputs shown so you can check my work.
Each reaction releases 8.7 MeV. One electron-volt is 1.602 × 10−19 joules, so 8.7 MeV is 1.394 × 10−12 joules per reaction. At 108 reactions per second, total fusion power is 108 × 1.394 × 10−12 = 1.394 × 10−4 watts. That is 0.14 milliwatts. All of it leaves as kinetic energy in the three alpha particles, which are born inside a plasma already millions of degrees hot and contribute nothing measurable to anything.
Here is the scale problem in one table.
| Milestone | Reactions / second | Fusion power | Multiple of today |
|---|---|---|---|
| ENN EXL-50U (Sept 2026) | 108 | 0.14 mW | 1× |
| One LED nightlight (~0.5 W) | 3.6 × 1011 | 0.5 W | 3,600× |
| One phone charger (5 W) | 3.6 × 1012 | 5 W | 36,000× |
| One gigawatt power plant | 7.2 × 1020 | 1 GW | 7.2 trillion× |
Seven point two trillion is the factor by which today's reaction rate must grow to reach a gigawatt of fusion power. Put it on a growth ruler: even if the rate compounded tenfold every single year, log10(7.2 × 1012) = 12.86, so it would take thirteen years of uninterrupted tenfold annual growth.
Before dismissing the whole thing, run the second calculation, because it is the reason serious money chases this reaction anyway.
A one-gigawatt plant at 7.2 × 1020 reactions per second consumes one boron-11 atom per reaction. Boron-11 weighs 11.0093 atomic mass units, so the mass burn rate is 7.2 × 1020 × 11.0093 × 1.6605 × 10−27 kilograms, which works out to about 1.31 × 10−5 kilograms per second, or roughly 1.1 kilograms per day. Even priced at a punitive $100 per kilogram, the daily fuel bill for a gigawatt plant is $113, which is $0.0000047 per kilowatt-hour. Fuel is, for all practical purposes, free. A coal plant of the same size burns roughly seven thousand tonnes of coal a day. Hydrogen-boron fusion's entire economic problem is the machine, and coal's entire economic problem is the fuel. That asymmetry is what the milestone is really about.
And for a sense of how early "early" is: at three alphas per reaction and 108 reactions per second, the machine manufactures 3 × 108 helium atoms per second. A standard 14-liter party balloon holds about 3.8 × 1023 helium atoms. Dividing one by the other gives 1.26 × 1015 seconds, or roughly forty million years, to fill a single balloon with fusion-produced helium: a real reaction, at geological scale.
Credit where it is due: the EXL-50U has been on a genuine streak. Per ENN's own spec sheet, the machine is a spherical tokamak 12.7 by 14.9 feet, and the September result did not come out of nowhere. In April 2025 it pushed a million amps of plasma current through hydrogen-boron fuel, which ENN calls a world first for that fuel. The 1.2-tesla field milestone followed in May 2025. In December 2025 the company reported the first hydrogen-boron plasma running in H-mode, the high-confinement state every serious fusion machine chases, and by January 2026 electron temperatures had passed 180 million degrees Fahrenheit. In April 2026, ENN joined the physics activities of the international tokamak program coordinated through ITER, only the second private fusion company to do so after Commonwealth Fusion Systems (autonocion, citing ENN's spec sheet and published papers). Outside labs are treating the operation seriously. ENN has started building Helong-2, its third-generation device, and is targeting 2030 for first electricity from hydrogen-boron fusion.
Now the physics bar, stated plainly, because it is brutal. A 2026 peer-reviewed study in Frontiers in Nuclear Engineering puts the minimum ignition conditions for proton-boron-11 at ion temperatures of 230 to 250 keV, with bremsstrahlung radiation losses dominating unless the electron temperature stays well below the ion temperature (Frontiers in Nuclear Engineering). Deuterium-tritium ignites near 10 keV. Physics World puts the requirement at roughly 1.5 billion kelvin (Physics World). It must hold a plasma whose ions are vastly hotter than its electrons, which is intrinsically unstable, because electrons absorb energy more easily and equilibrate fast. ENN's trick, firing fast protons as a beam-target catalyst, sits in a class of experiments that has produced alpha particles before without any path to net energy gain; a 2023 American Physical Society poster on proton fast ignition states the finding directly, that the reported alphas "come from beam fusion reactions that do not scale to net energy gain" (APS DPP 2023 poster, Mehlhorn et al.).
One more honesty item, and it is a good one. Boron arrives at the plasma as diborane gas mixed with hydrogen, plus fine powder dropped straight in, per ENN's published papers. Diborane is pyrophoric, ignites spontaneously in moist air, and is highly toxic by inhalation, with one of the lowest occupational exposure limits of any industrial gas (NOAA CAMEO; ScienceDirect). Clean-fuel narratives survive the plasma physics. They die at the gas-handling manifold.
ENN is not alone on this fuel. TAE Technologies, founded in 1998 as Tri Alpha Energy, pursues p-B11 with a field-reversed configuration instead of a tokamak, has raised over $1 billion from backers including Google and Chevron, and is moving toward Da Vinci, a 50-megawatt-electric first-of-a-kind plant, with siting decisions targeted for the end of 2026 (OilPrice). Australia's HB11 Energy takes the laser-driven path. Neither has claimed net energy gain. Here is the honest scoreboard for the entire aneutronic fusion industry: three well-funded approaches, zero watts of net electricity, one hundred million reactions a second.
The Strongest Case Against This Article
Here is the skeptic's case at full strength, and parts of it are correct. That 100-million-reactions figure is self-reported by a company with every incentive to impress, with no peer-reviewed paper and no published independent diagnostic confirmation; as one commentator put it, this is a company grading its own homework, so keep the attribution in mind. Beam-target fusion has been producing small numbers of alpha particles for decades without ever approaching net gain, and the APS analysis says flatly that this class of reactions does not scale to it. p-B11's Lawson bar is so far above D-T's that several generations of plasma physicists have called the fuel economically hopeless, and nothing in the September announcement changes the bremsstrahlung physics. Helong-2 targets electricity by 2030 while the reaction rate sits twelve to thirteen orders of magnitude below power-plant scale, which makes the timeline a marketing artifact rather than an engineering schedule. And the "first commercial company" framing is doing quiet definitional work: university and national-lab teams measured p-B11 fusion in magnetically confined plasmas years ago, including a US-Japan team on Japan's Large Helical Device, so the novelty is the corporate letterhead, not the physics. If you believe fusion announcements should be measured in watts rather than reactions, this one measures 0.00014 of the former, and the press cycle it generated is the real product.
What This Does Not Prove
The honest bounds first. ENN's claims cover the reaction rate, machine specifications, and 2030 timeline; no independent peer review exists yet. Nobody disclosed the input wall-plug power of the neutral-beam, RF, and magnet systems, so no Q value can be computed; the heating is described here only as megawatt-scale, typical of tokamak beam systems. My $100-per-kilogram boron figure is illustrative, and the actual cost of separating boron-11 for reactor fuel is not public. That thirteen-year tenfold-growth ruler is illustrative, not a prediction; rate scaling runs into physics walls like bremsstrahlung and confinement that pure compounding ignores. Nothing here demonstrates that beam-target p-B11 can ever reach net energy gain, which is the only result that would change the economics.
What You Can Do
Here are three moves, depending on who you are.
If you invest in energy: fusion-adjacent stocks moved on this news, and the honest read is that a milestone is not progress to power. What matters is the reaction rate versus the 7 × 1020 per second a gigawatt needs. Watch for ENN's peer-reviewed papers and independent diagnostic confirmation, not press releases; a rate number without a published diagnostic is a rumor with a decimal point. If you follow the technology: the two real schedule signals are Helong-2 construction updates and TAE's Da Vinci siting decision, targeted for the end of 2026, because siting a 50-megawatt plant forces a company to commit to physics it can defend in front of regulators. And if you just want the story straight: the prize everyone is chasing is direct conversion, turning charged alpha particles into electricity without a steam turbine, plus reactor walls that never become radioactive waste. If aneutronic fusion ever works, the plants are simpler than anything in the D-T roadmap. That "if" is doing all the work, and it is a physics "if", not an engineering one.
The Bottom Line
A commercial company has now fused hydrogen and boron on its own machine at a rate no private company has reported before, and the honest arithmetic says the machine produced about 0.14 milliwatts, needs seven trillion times more for a gigawatt, and would fill one party balloon with fusion-made helium in forty million years. At scale, the fuel would cost essentially nothing, the physics bar is roughly thirty times higher than the reaction the rest of the industry is struggling with, and the boron arrives as one of industry's nastiest gases. It is simultaneously a genuine first and thirteen orders of magnitude from a power plant, and anyone telling you only one of those two things is selling something. The milestone is real, and so is the math.