On 19 June 2026, Realta Fusion, working with the University of Wisconsin–Madison, drew a few amperes of current at around a hundred volts from charged particles streaming out of the ends of their WHAM mirror device — enough, in their own words, to light a few bulbs. It is worth pausing on how modest that sounds, and how significant it actually is.
This was the first time a private fusion company has converted the kinetic energy of a fusion-relevant plasma directly into electricity, without first turning it into heat. The team has been careful to say what it is not: it is neither net-electric production nor large-scale conversion of fusion-born power. That restraint is exactly why the result deserves attention. A real current in a real circuit is a physical fact, not a press release.
I want to use the occasion to discuss something the fusion community does not talk about often enough — the rather nineteenth-century machine sitting at the heart of almost every twenty-first-century power plant we propose to build.
The boiler we rarely mention
For all the sophistication of magnetic confinement, the back end of nearly every fusion power-plant concept is a steam turbine. Neutrons heat a blanket, the blanket heats a working fluid, the fluid drives a turbine, the turbine turns a generator. Coal plants do this. Gas plants do this. Fission plants do this. Most fusion designs, quietly, do this too. We spend decades mastering plasmas at a hundred million degrees and then hand the energy to a heat engine that James Watt would recognise.
The trouble with heat engines is not engineering quality; it is thermodynamics. Any cycle that converts heat to work is bounded by the Carnot limit, which depends on the ratio of the cold-side to the hot-side temperature. A fusion blanket cannot run arbitrarily hot — its materials, its coolant chemistry, and its structural margins set a ceiling. In practice that leaves the thermal-to-electric conversion of a fusion plant somewhere in the region of thirty to forty-something percent. More than half the energy you worked so hard to liberate leaves as waste heat through the condenser, by construction. This is not a flaw to be engineered away. It is a law.
What direct energy conversion does differently
Direct energy conversion sidesteps the heat engine entirely. The idea is old and elegant: a charged particle leaving the plasma carries kinetic energy, and if you make it climb against an opposing electrostatic potential as it travels toward a collector, it arrives slowed down, having deposited its energy as electrical work in the external circuit. No boiling, no spinning mass, no Carnot ceiling.
The achievable efficiencies are high — and they were not invented this June. Richard Post pioneered the concept for mirror machines at Livermore in the 1960s and 1970s, and the experiments of Barr and Moir on the Livermore mirror programme demonstrated it decades ago: beam-based converters reached peak efficiencies of around sixty-five percent, and a true plasma direct converter tested on the Tandem Mirror Experiment recovered close to half — roughly forty-eight to fifty percent — of the end-loss energy it was handed. What Realta has added is the first demonstration of the principle on a modern, commercially-oriented mirror device. The physics is inherited; the context is new.
Why this is a mirror story, not a tokamak story
Here is the part that I find genuinely interesting, and it explains why direct conversion has always been bound up with the magnetic mirror in particular.
A tokamak confines its plasma on closed, nested magnetic surfaces. Particles that are well confined stay on those surfaces; there is no natural, directed stream of charged particles leaving the machine that you could decelerate against a grid. To extract energy you essentially have to wait for it to arrive as heat. The mirror is the opposite. A magnetic mirror confines particles in an open bottle, and it has a “loss cone” in velocity space — a population of ions and electrons that scatter into trajectories the field cannot hold, and so escape axially out of the ends in a directed flux.
For half a century that leakiness was treated as the mirror’s great weakness. Direct conversion reframes it as the mirror’s defining advantage. The particles you were “losing” arrive at the expander region as an ordered beam, which is precisely what an electrostatic collector wants. The geometry that was a liability becomes the harvesting mechanism. Realta put their converter exactly there — in the end-ring assembly, in the path of the loss-cone flux. This is why the high-field axisymmetric mirror, the family that includes WHAM and the linear-mirror architecture we are pursuing at ASPL Fusion, is the natural home for direct conversion. It is not bolted on. It falls out of the field topology.
The honest picture for deuterium–tritium
I want to be careful here, because this is where enthusiasm tends to outrun the physics.
First-generation fusion will burn deuterium and tritium, and a D–T reaction puts about eighty percent of its energy into a fast neutron and only twenty percent into a charged alpha particle. Neutrons carry no charge. You cannot decelerate a neutron against an electric field; there is nothing to grip. The neutron energy has to be caught in a blanket as heat, and that heat, if you want electricity from it, goes through a turbine. So a D–T mirror plant equipped with direct conversion is not a turbine-free plant. It is a hybrid — most of the power still flows through the thermal island, with direct conversion handling the charged fraction. Realta’s own analogy of a hybrid car powertrain is apt.
So what does direct conversion actually buy a D–T plant? Its real value is less glamorous than “no more turbines” and more important than it first appears: it recaptures the input power. A mirror is driven hard by neutral beams and radio-frequency heating, and a large fraction of that injected power flows straight back out through the loss cone. Recovering it directly — at roughly half efficiency, plausibly better with staged collectors — instead of pushing it through a thermal cycle dramatically lowers the recirculating power fraction, the tax a plant pays to keep itself running. Lower that tax, and you reach net-electric operation at a lower fusion gain than you would otherwise need. For mirror machines, whose gain has historically been their hardest number to push up, that is a meaningful relaxation of the requirement. Direct conversion is not what makes a D–T mirror work; it is what gives the design room to breathe.
The aneutronic horizon — stated carefully
Where direct conversion stops being a useful supplement and starts becoming the main event is with advanced, low-neutron fuels.
Consider the limiting cases. Proton–boron-11 fusion is essentially aneutronic: its yield emerges as three charged alpha particles, in principle almost entirely recoverable by direct conversion, with no blanket, no tritium breeding, and no neutron activation of the structure. Deuterium–helium-3 puts the great majority of its energy into charged products as well. In a plant burning fuels like these, the turbine does not merely shrink — it could in principle disappear, and with it the Carnot ceiling, the condenser, and much of the thermal balance-of-plant. The open mirror, with its directed loss-cone exhaust, is one of the few architectures in which this is even geometrically sensible.
The honesty obligation is equally important. These fuels are hard. Proton–boron requires plasma temperatures and confinement quality well beyond what D–T demands, and its reactivity is unforgiving of the Lawson product. Deuterium–helium-3 needs a supply of helium-3 that does not presently exist at scale. No one should promise an aneutronic power plant on a near-term calendar, and I will not.
At ASPL Fusion our base case is deuterium–tritium, and our staged programme is built around it. The advanced-fuel path — deuterium–helium-3 in particular — sits in our planning as a long-horizon upgrade and an upside, not as a commitment we are asking anyone to underwrite today. What I will say is that the architecture we have chosen does not foreclose that horizon. An axisymmetric linear mirror with high-temperature superconducting magnets is, by its very topology, a machine to which direct conversion can be added and in which advanced fuels remain physically conceivable. And there is a quiet internal logic worth noting: a D–T mirror that breeds a tritium surplus also, through tritium’s decay, accumulates helium-3 over time — so the fuel for the harder reaction is, in a sense, a by-product of the easier one. We treat that as a strategic option to keep open, not a claim to bank.
None of this is in competition with India’s national programme. It is complementary to it — a commercial route, on Indian soil, that explores the linear-mirror corner of the design space while the national effort pursues the toroidal mainline.
What I take from Realta’s bulbs
The thing I appreciate most about the WHAM result is its scale. A few amps, a hundred volts, a few bulbs, and a clear statement of everything it does not yet demonstrate. That is how this field should report progress — by physical quantities measured, not milestones declared. Direct energy conversion will not abolish the steam turbine for deuterium–tritium; it will trim the recirculating-power tax and ease the gain requirement that has long burdened the mirror. For the advanced fuels that may come after, it changes the question entirely, because it offers a path to electricity that never passes through a boiler at all.
The mirror’s leakiness was treated as its flaw for fifty years. It now looks like the feature that lets the geometry pay for itself. There is a lesson in that worth carrying beyond plasma physics: the property of a system you have been trying to suppress is sometimes the one you should be learning to harvest.