Microsoft has signed the world’s first-ever deal to purchase fusion energy.
The company has agreed to buy 50 megawatts of energy per year from Washington state-based Helion Energy starting in 2028.
For decades, controlled fusion power has represented a holy grail of clean energy among scientists and policymakers — power on demand, without the radioactive waste of traditional nuclear power or the planet-heating fumes of fossil fuels.
The need for clean electricity to slow the climate crisis has helped spur a wave of investment and innovation in the field, which now boasts dozens of private companies and more than $4 billion in venture capital invested in 2021 alone.
Fusion has definite benefits. The fuel “is inherently clean, inherently safe, and could one day provide vast amounts of the type of power we need to tackle the climate crisis,” Sen. Maria Cantwell (D-Wash.) said in a statement following the announcement.
The Microsoft-Helion deal “confirms that the State of Washington is the world’s leading hub for fusion energy innovation and commercialization.”
But the deal also represents a big-money bet on a bold proposition: that fusion power will be brought to market years or even decades earlier than most experts had predicted.
The news comes as fusion startups are under increasing pressure to demonstrate that their approaches are viable.
Fusion experts interviewed by The Hill declined to comment on the specifics of Helion’s approach, which have remained a closely held secret.
In the burgeoning world of fusion, the company has “been very close to the vest with their information — which is part of their advantage,” said Andrew Sowder, a fusion expert at the Electric Power Research Institute (EPRI).
Whether or not the technical elements pan out by 2028, the announcement marks “a market milestone” for the fusion industry, said Dennis Whyte, director of the Plasma Science and Fusion Center at MIT.
Whyte leads a competing, private sector-funded fusion team that is seeking a more traditional route to fusion: a stable “burn” within the donut-shaped reactor known as a tokamak.
But he said that the Microsoft-Helion deal represented good news for the industry: a strong indicator that, once other approaches were ready, buyers would be interested.
“And that is important, because it’s signaling what the world needs, what the consumer needs,” Whyte said.
Microsoft is willing to take the bet because it is an insatiable consumer of carbon-free electricity.
The tech giant is squeezed between the growing computational demands of its AI software and its commitment to power its servers with carbon-free electricity by 2030.
The company’s deal with Helion would only be a drop in the bucket of its larger power needs — 50 megawatts is only about 10 percent of the power a company data center consumes in a year.
But for Helion, it represents an infusion of capital the company can use to bring their prototype to power-plant scale.
The flip side? Helion is now financially on the hook to develop commercial fusion much more quickly than most observers had thought possible.
The power purchase agreement “is a firm contract,” a Helion spokesperson wrote to The Hill. “Pricing will be market rate. If Helion misses the deadline for putting power on the grid for Microsoft by then, Helion will face financial penalties, which signifies the strength of this agreement.”
It’s a big bet. For decades, scientists at government-sponsored labs have tried to replicate fusion by using magnetic fields to simulate the exotic physics possible within the crushing gravity of a star.
A big milestone on that road came last year, when the Lawrence Livermore National Laboratory announced that it had achieved ignition: a fusion reaction in which more energy came out than was put in.
But that was only a small step on a road to a much larger goal, upon which most fusion approaches depend: fusing gases together to create a self-sustaining reaction that consistently puts out more energy than gets put in — even when the power costs of the facility are factored in.
Conventional fusion uses magnetic fields to contain a fusion reaction that is self-sustaining, like a fire in a hearth.
Trapped in its magnetic blankets, the fusion reaction releases a flood of radioactive neutrons into the walls of its container, creating an immense release of heat — which can then be used to boil water and run an electric turbine, just like in a nuclear or natural gas plant.
That approach — which the MIT team is also using — makes a lot of sense, Sowder said. Neutrons are the easiest ways to get heat out, and thermal plants are a highly mature technology — after all, society has been building thermal power plants for more than a hundred years.
But the thermal comes with significant challenges — not least that the neutrons tend to turn their container radioactive — and despite significant lab milestones, any kind of self-sufficient reaction suitable for a commercial application is still a long way off.
Helion CEO David Kirtley told The Hill that his company will be able to produce power earlier than anyone else because the company is not concerned with creating a stable, self-sustaining reaction.
“That’s some pretty hard physics to do,” and it’s also unnecessary, he argued, as long as all you care about is generating electricity.
Helion’s alternate approach builds on one of the scientific precepts that undergirds the modern world: Maxwell’s equation, or the revelation that magnetic fields and electric currents are two sides of the same phenomenon.
Rather than a roaring fire, Helion’s sixth-generation Polaris reactor is powered by something more like a series of small explosions, or what Kirtley calls a “fusion engine”: a series of small fusion reactions in the center of magnetic field.
Each reaction takes the form of a cycle. First, a magnetic field compresses a small cloud of gas until it is forced to fuse together — causing an expanding fusion reaction.
As that powerful pulse of energy pushes out against the magnetic field that contains it, it generates electric current that powers the next compression and discharges excess electricity onto the grid.
By capturing electricity directly — with no need to boil water and spin a turbine — “we’ve been able to build fusion systems that are much, much smaller than any of the other approaches to fusion,” Kirtley said.
“And that means you can build them faster, you can learn more, and you can build power plants sooner.”
This idea of “direct energy conversion” represents “kind of a holy grail” for the fusion industry, Sowder of EPRI said.
Sowder added that this high degree of efficiency is likely essential for Helion, given the relative poverty of their chosen fuel — Helium-3, a light form of helium (from which the company takes its name) that is far-less energy dense than deuterium or tritium, which power most fusion approaches.
With less power available in the fuel, Sowder said, Helion can’t afford to waste any on boiling water.
He compared their approach to “like solar panels: you know, they take sunlight and convert it directly into electricity — which makes up for the extreme inefficiency of the overall process.”
For Helion, the infusion of cash from Microsoft will be essential to bring together the two successful lines of research the team has developed into a single functioning power plant.
Helion has shown it can achieve the necessary physics off of a single pulse and is working to tighten the cycles that will drive its engine.
“We want to get from where we are now, which is operating once every 10 minutes. And getting that up to operating on the order of once a second, and we’re now building the first machines to do those at scale,” Kirtley said.
If the physics are successful, Kirtley said, Helion’s partners in the deal — both Microsoft and national utility Constellation Energy — would be essential in the drive to turn a prototype reactor into a power plant, because those companies have far more experience building them.
That could help Helion get around a paradoxical handicap of its approach: the fact that, in comparison to more traditional forms of fusion, any power plant it builds will be so different from the thermal power plants currently dominant on the American landscape.
Most fusion “plants do use thermodynamic cycles to do power conversion — and that’s been our main meat potatoes for decades, right?” Sowder asked.
That means, for both better and worse, Helion will be starting from the ground up.
“Unless — unless again, they’re able to draw on some other technology, right? And this is where I’m in the dark,” Sowder said. “And this is the benefit of innovation, right? You take something and then you apply it in a new way. And they may have done that.”