Can Fusion Power Be Profitable? A New Test Says What It Would Take

On December 5, 2022, 192 laser beams at the National Ignition Facility in California delivered 2.05 megajoules of energy to a peppercorn-sized fuel capsule and got 3.15 megajoules of fusion energy back, the first controlled fusion experiment to produce more energy than it absorbed. After sixty years of chasing it, fusion’s physics question had an answer. The next question is ruder: can it pay?

Investors are acting as if it can. Fusion companies raised a record $4.48 billion in the twelve months to July 2026, bringing the industry’s reported total to $14.24 billion across 56 companies, according to the Fusion Industry Association’s 2026 report. Commonwealth Fusion Systems, an MIT spinoff, has announced plans to build its first commercial plant, called ARC, in Chesterfield County, Virginia, expecting to be the first to make fusion power available at grid scale. What the industry has lacked is a shared, impartial way to ask whether any of these plants can operate at a profit.

A Control Surface Turns Physics Into Accounting

A study published last month in the Journal of Fusion Energy proposes one. Dennis Whyte, a professor of nuclear science and engineering at MIT, and Andrew Lo, a professor of finance at MIT’s Sloan School of Management, together with four colleagues, built the framework around a single physical object every fusion concept shares: a control surface, the engineered wall that completely surrounds the fusion fuel and through which every joule of fusion energy must escape. Fusion can only happen at thermonuclear temperatures, so the burning fuel must float in isolation, whether inside magnetic fields or at the focus of lasers, and the wall that catches its energy is where physics hands over to accounting. Whyte and Lo normalize every flow of money to that surface, in millions of dollars per square meter per calendar year, which makes the math independent of how big the plant is and indifferent to how it confines its plasma.

The inspiration is the field’s most trusted tool. In a 1957 paper, the British physicist John Lawson worked out the combinations of temperature, plasma density, and energy confinement time that let a fusion plasma produce net energy, a result now called the Lawson criterion. Its power is its silence: it says nothing about magnets, lasers, or reactor geometry, yet it judges them all. The plasma gain it defines, Q_p, is the ratio of fusion power produced to the external power needed to sustain the plasma.

“The Lawson Criterion describes the scientific success of energy gain from fusion plasmas, while our framework generally describes economic Q, which is the ratio of capital gained to that expended,” Whyte explains. Where Lawson needed three plasma quantities, the economic version needs ten knobs: the net price of the energy sold, the efficiency of converting fusion power into a sellable product, the fusion power density through the surface, the energy fluence the surface can absorb before failing, how long replacement takes and what it costs, the construction cost of the whole plant per square meter, the cost of consumable fuel targets, the plant’s lifetime, and the interest rate on the capital.

The bookkeeping then runs like a power plant’s inbox and outbox. Revenue arrives from selling energy. Three costs drain it: consumable targets, the recurring replacement of the energy-capturing surface, and the fixed burden of construction and financing. Because every replacement shuts the plant down, the framework solves for how much of the year a plant can actually run, and that feedback makes the economics nonlinear: push the power density up and revenue climbs, but the surface burns out faster and downtime grows. “It doesn’t matter whether the fusion power plant is a small or large, the bottom line is: In both cases you better have money coming out that exceeds the money going in, otherwise it’s not going to be around for very long,” Lo says.

Viability Starts Around Two Megawatts per Square Meter

Running the ten parameters over wide ranges produced results the authors themselves call surprises. The first is a floor: across plausible assumptions, a plant needs to push roughly 2 megawatts of fusion power through each square meter of its energy-capturing surface to break even, and in worked examples the requirement runs from 2 to 5. That inverts a piece of industry folklore holding that a gentler plant, with low power density and a cheaper wall, would be easier to make profitable. At low power density, the fixed costs of construction and financing spread over too little revenue, and the plant drowns in its own mortgage.

The second surprise concerns the wall itself. The base case assumes the surface survives about 3 megawatt-years per square meter of neutron battering, which at viable power densities means swapping it every 0.6 to 1.5 years, with each swap taking about five weeks, a rhythm borrowed from refueling outages at fission plants. Within those bounds, the model’s verdict is blunt: a surface that is cheap to replace and quick to swap beats a marvel of durability. Money spent making the wall nearly immortal would buy less viability than money spent making its replacement routine.

There’s a catch, and the authors state it themselves. The parameter the whole result leans on, the energy fluence limit of the surface, is the one with almost no direct measurement behind it; nobody has yet run a commercial-grade fusion wall long enough to know when it dies. The paper concedes that some of the most important parameters “must be adduced from distant examples,” chiefly fission experience, and that clearing Q_econ of 1 is necessary but not sufficient, since no model can capture every cost of a real project.

The referees, it should also be said, have money in the game. Whyte co-founded Commonwealth Fusion Systems, the company now planning the Virginia plant, and he and Lo co-founded Rutherford Energy Ventures, a fusion consultancy and investment advisory firm that, with Stone Mountain Capital, provided financial support acknowledged in the paper, though the authors declare no competing interests. The framework’s defense against its own conflict is structural: because the ten parameters never name a technology, tilting the math toward one company’s design would require tilting it toward all of them.

The Number That Decides Everything Is Still Unmeasured

Lo, who has spent years engineering financing models for drug development, frames the exercise as the start of a cost curve rather than a verdict. “This pattern of learning by doing exists in all deep technology sectors,” he says, pointing out that sequencing a human genome is a million times cheaper now than it was about 25 years ago, a collapse tracked in data kept by the National Human Genome Research Institute. “We’re going to see the same thing, but maybe not to the same degree, in fusion energy.” The framework is built to ride that curve: feed in cheaper financing, faster replacement crews, or hardier walls as they arrive, and the viability threshold moves.

To that end the team published a fusion economics calculator online where anyone can plug in their own ten numbers and watch a design pass or fail. Its most honest feature may be the blank it cannot fill. The fate of every design on that page, and eventually of the plant rising in Chesterfield County, turns on a quantity no laboratory has yet measured: how many megawatt-years a square meter of reactor wall can absorb before the crew comes to swap it.

FAQ