How CFS is building a fusion factory, not just a single fusion machine

Copy LinkXLinkedInEmail

At Commonwealth Fusion Systems (CFS), we’re not just building SPARC, the machine we plan to use to demonstrate net fusion energy in 2027. We’re also building a factory — the machine that builds the machine.

And at the FusionXInvest conference this week, our Chief Science Officer Brandon Sorbom detailed how that’s happening at our magnet-making building in Devens, Massachusetts.

“Our factory now looks a lot more like an auto factory,” Sorbom said at the fusion conference at the Computer History Museum in Mountain View, California, the heart of Silicon Valley. That’s about six upgrades from the first CFS manufacturing process that really could only produce “artisanal” magnets, he said.

Magnet manufacturing is at the heart of CFS’ work to commercialize fusion energy. Our SPARC tokamak and its power plant successor, ARC, use super-powerful magnets to confine and control plasma. Key to our approach is high-temperature superconducting (HTS) tape that enables us to make stronger magnets and therefore smaller and more economical tokamaks that can meet humanity’s fast-growing power demand. Factory improvements bring higher manufacturing volume in turn brings lower product costs.

Each of our D-shaped toroidal field (TF) magnets is made of a stack of 16 “pancakes.” And each pancake is a steel plate that houses a stack of HTS tape inside a spiral channel.

CFS' toroidal field model coil on display at the Computer History Museum


Our first pancake took 30 days to make, Sorbom said. As we made our way along the learning curve, we got faster: Our 16th pancake took 12 days to make. Now we’re making a pancake per day, and CFS has now built well over half of the 288 TF pancakes we’ll need for SPARC.

“The really big thing that’s changed is the industrial scale and speed of building these pancakes,” Sorbom said in an on-stage interview with Fusion Energy Insights Chief Executive Melanie Windridge. “It’s one thing to be able to build a magnet by hand. It’s a whole different level to actually build a factory to just crank these things out.”

At the Computer History Museum, CFS also is showing a pancake from the Toroidal Field Model Coil (TFMC) demonstration program, which in 2021 proved the viability of our TF magnet design at the scale of SPARC. It was a seminal moment that meant the company could proceed with its plans to design and build SPARC.

The pop-up exhibit is a fitting display for a computer museum. Just like the transistor led to the computing revolution, HTS tape enables new components like our magnets that make fusion energy practical. And like the transistor improvements charted by Moore’s Law brought down the cost of computers, manufacturing improvements will bring down the cost of fusion.

The TFMC magnet houses 270 kilometers of HTS tape that carries the current, and obtaining the tape was another industrial success for CFS. The tape is made of a rare-earth barium copper oxide (REBCO) superconductor which is deposited on a flat metal substrate by a thin film process similar to how computer chips or LCD screens are made. It took researchers years to make a commercially useful wire — more commonly called tape since it’s flat — out of the material, and its initial supply was limited. CFS worked closely with the major HTS suppliers to scale up this supply chain to the levels required to build fusion magnets.

SPARC needs about 10,000 kilometers of HTS tape. ARC, with somewhat bigger magnets, will need even more.

Inside the magnet factory at Commonwealth Fusion Systems' headquarters in Devens, Massachusetts


“I used to get laughed at five years ago by some people when I had the audacity to suggest that we would somehow get 10,000 kilometers of tape for SPARC in 2025,” Sorbom said. But now, nearly all of that tape has now been delivered to CFS, with much of it already in the pancakes that make up the magnets.

As with magnet manufacturing, we’ll have to turn the HTS supply crank even faster as we move to put fusion power on the grid by the early 2030s with the first ARC in Chesterfield County, Virginia.

“We’re going to have to scale it up again for ARC. There are some people today who don’t believe that we’re going to be able to get to the next step of tape,” Sorbom said. “And to those people, I say, just look back over the last five years and see what we’ve done.”