The believer who stopped promising

There is a particular way an engineer describes a thing when he has decided to stop selling it. He reaches for the smallest possible image. Asked, this winter, what was happening inside the building in Devens, Massachusetts, where his company is assembling the machine that is supposed to change the price of energy on Earth, Bob Mumgaard did not reach for the cosmos. He reached for a toy. "It's a complicated Lego set," he said, "but we have a good set of instructions and a good set of people that will put it together."

Listen to the man long enough and you notice the deflation is deliberate. The first of eighteen superconducting magnets had just gone in. Each one weighs twenty-four tons and produces a field of twenty tesla, roughly thirteen times the pull of a hospital MRI, cooled to two hundred and fifty-three degrees below zero so it can carry thirty thousand amps without melting. It is, by any honest measure, one of the most extreme objects ever fabricated by a private company. Mumgaard called it the kind of magnet "you could use to, like, lift an aircraft carrier," and then he called the whole thing a Lego set. Both are true. Only one of them is the sentence he wants you to remember.

The joke he inherited

Every fusion physicist is handed the same joke on the first day, the way a new soldier is handed a uniform that does not fit. Fusion is thirty years away, and it always will be. It is told at conferences and at dinner tables and, most cruelly, in funding meetings, and for seventy years it had the unbearable quality of being correct. The men and women who gave their careers to confining a plasma hotter than the core of the sun spent those careers explaining why the date had moved again.

Mumgaard came up inside that joke. He took a degree in engineering physics from Nebraska, then a master's and a doctorate at MIT, the doctorate in applied plasma physics, which is the academic phrase for spending your twenties trying to hold a star still. In 2018 he and a small group spun a company out of MIT's Plasma Science and Fusion Center with a bet that was less about physics than about timing: that a new high-temperature superconductor could shrink the magnet, and a smaller, stronger magnet could shrink the machine, and a smaller machine could shrink the one number the field had never managed to move — the date.

He spent a decade learning to make promises. He is spending this one learning which promise is safe to keep.

You can hear, in the early Mumgaard, a man who decided the cure for the joke was confidence. "People said it would take at least ten years" to build the magnet, he told one interviewer. "We were confident we could do it in three, and we did." He says it plainly, without the swagger you would expect, the way someone reports a fact about the weather. In September 2021 the company energised a magnet to twenty tesla and the photograph of that day became, for a while, the most-shared image in the field. Money followed the way it always does — the swagger justified. A Series B that November raised $1.8 billion. The total now is somewhere near three billion dollars, almost all of it private, from Google and Nvidia and Bill Gates's fund and an Italian oil company that decided it would rather own the thing that replaces it.

What he no longer says

Here is the tell. Watch what Mumgaard does with the word inevitable. He has said, more than once, that fusion has crossed from "this is impossible" to "this will be inevitable," and it is a lovely line, the kind a founder reaches for instinctively. But notice that inevitable is a claim with no date attached. It is the one promise that cannot be falsified by a calendar. A decade ago he was the man who said three years and meant it. Now he builds his confidence out of words that the future cannot embarrass.

The machine itself does still have dates. SPARC, the tokamak in Devens, is meant to be largely complete by the end of 2026; all eighteen magnets are due in by the end of the summer. The goal that matters — net energy gain, the moment a fusion plasma produces more power than is poured in to sustain it, the threshold physicists write as Q greater than one — is targeted for 2027. These are real dates, published and peer-reviewed, and they have so far held. What is striking is how Mumgaard arranges his sentences around them. He does not say it will work in 2027. He says: "The main argument against fusion is making it work, and that's why we're building SPARC and showing that it can work." The date is in the room. He simply declines to stand next to it.

This is not evasion, or not only. It is something closer to a man who has read his own field's history and refuses to add a chapter to it. For seventy years the harm was never the physics. The harm was the promise — the date given to a senator, the date printed in a magazine, the date that arrived and found nothing, and taught the public to file fusion alongside flying cars in the drawer of things that are coming and never come.

The Lego set in Devens

So he has built a vocabulary of the small and the concrete. "Parts are arriving and it's being assembled." "It'll go bang, bang, bang throughout the first half of this year." The aircraft-carrier magnet becomes a toy on the floor of a child's room. He has lately taken to calling fusion "no longer a science project" but "the next big thing in tech," and you can read that as hype, but it functions as the opposite — it is a way of dragging the thing down out of the heavens, where promises go to die, and into the ordinary world of supply chains and schedules and people who put pieces together according to instructions.

The company has even begun to dress its caution as machismo. When CFS filed in April to join the PJM grid operator's interconnection queue — the unglamorous paperwork by which a power plant that does not yet exist asks permission to one day sell electricity in Virginia — Mumgaard framed the filing not as a forecast but as a dare to himself. "When you're serious about building a power plant in the early 2030s, you act now," he said. "This is execution." Even the boast is about doing, not about when it will be done. The commercial plant, called ARC, is meant to put roughly 400 megawatts onto the grid in the early 2030s — enough for about 300,000 homes. Notice the phrase he uses for the horizon: not a year, but a decade, with its comfortable elastic edges.

The thing the believer cannot give up

It would be too neat to say Mumgaard has stopped believing. He has not. The belief is the engine of the whole enterprise, and it shows in the one register where he still allows himself to sound urgent — not the technology, but the reason for it. "With climate change, we're dealing with a ticking clock," he has said, "so we want to move as fast as we can." That is the sentence with feeling in it. The clock he is willing to name is the one counting down on the planet, not the one counting up to his own machine. He will commit, without flinching, to the urgency. He has simply learned to stop committing to the date that would measure whether the urgency was met.

There is a kind of person who promises in order to make the future arrive — who believes, half-magically, that the date said aloud becomes the date that comes true. Founders are nearly all this kind of person; it is most of what the job is. Mumgaard was that man, and the magnets prove he was sometimes right. What has changed is harder to photograph. Somewhere across a decade of raising money on milestones, the true believer learned the more frightening discipline: to do the thing on time and let the timing speak for itself, rather than the other way around. To put the magnet in the building and call it a Lego set. To stand near 2027 and not point at it.

Whether SPARC reaches Q greater than one next year, no one outside the plasma can honestly say, and the most credible thing about Mumgaard now is that he is among them. The old joke was that fusion is always thirty years away. The quieter thing happening in Devens is a man who spent his youth shrinking that number to three, and his middle age learning that the bravest thing he can do with a number is decline to say it out loud. The magnets keep arriving. He keeps reaching for the smallest available word. You begin to suspect that the discipline is the achievement — that the believer who stopped promising is closer to delivering than the believer who never could stop.