Commonwealth Fusion Systems installs first Sparc reactor magnet and inks Nvidia deal

Commonwealth Fusion Systems just hit a big milestone for fusion — and pulled Nvidia deeper into the race.

Speaking at CES 2026 on Tuesday, the company said it has installed the first magnet in Sparc, its demonstration fusion reactor that it hopes to switch on next year.

It’s the first of 18 massive, D‑shaped magnets that will eventually form a doughnut around the reactor, generating an intense magnetic field to confine and compress superheated plasma. If the physics and engineering cooperate, that plasma should release more energy than it takes to heat and compress it.

After decades of promise and delay, fusion power “just around the corner” suddenly looks less like hype. CFS and a growing pack of rivals are now racing to deliver the first fusion-generated electrons to the grid in the early 2030s.

A magnet that could “lift an aircraft carrier”

The new magnet is not subtle. Installed upright on a 24‑foot‑wide, 75‑ton stainless steel ring called a cryostat (set in place last March), each Sparc magnet weighs about 24 tons.

Once the full set is installed, the magnets will produce a magnetic field of 20 tesla — roughly 13 times stronger than a typical MRI machine.

“It’s the type of magnet that you could use to, like, lift an aircraft carrier,” Bob Mumgaard, CFS co‑founder and CEO, said.

To safely carry more than 30,000 amps of current, the magnets will be cooled to -253 °C (-423 °F). Inside the magnetic doughnut, the plasma they hold in place will be burning at over 100 million degrees Celsius.

Key components for all of Sparc’s magnets are already complete, and CFS expects to install all 18 by the end of the summer, Mumgaard said. “It’ll go bang, bang, bang throughout the first half of this year as we put together this revolutionary technology.”

Nvidia, Siemens and a full‑fidelity digital twin

Building Sparc isn’t just about hardware. CFS is also betting that better simulations — and better AI — can help it move faster.

The company announced it is working with Nvidia and Siemens on a digital twin of the reactor. Siemens is providing design and manufacturing software, which will help CFS capture data and feed it into Nvidia’s Omniverse libraries.

CFS already runs extensive simulations to predict how different parts of the reactor will behave. The problem, Mumgaard said, is that they’re siloed.

“With the digital twin,” he said, “these are no longer isolated simulations that are just used for design. They’ll be alongside the physical thing the whole way through, and we’ll be constantly comparing them to each other.”

The idea is to use the twin as a live counterpart to the machine. CFS can tweak parameters, run experiments in software first and only then push changes to Sparc itself. “It will run alongside so we can learn from the machine even faster,” Mumgaard said.

Billions raised, billions still to go

All of this is expensive. CFS has raised nearly $3 billion so far, including an $863 million Series B2 round in August that drew Nvidia, Google and nearly three dozen other investors.

Sparc is a demonstration device. The follow‑up, a commercial‑scale power plant called Arc, is where CFS hopes to actually sell fusion electricity to the grid. Because Arc will be the first of its kind, the company expects it to cost another several billion dollars.

Mumgaard is betting that digital twins and AI tools will pay for themselves by shortening that journey.

“As the machine learning tools get better, as the representations get more precise, we can see it go even faster, which is good because we have an urgency for fusion to get to the grid,” he said.

If CFS can keep Sparc on schedule, install all 18 magnets by summer and make the digital twin deliver, the long‑promised age of fusion power might finally move from the lab to the grid — and Nvidia will be there in the loop, not just as a chip vendor, but as an investor and software partner.