3D-Printed Tungsten Tiles Took 74 Plasma Shots Inside a U.S. Fusion Reactor. The Real Story Is the Factory, Not the Physics.

This July, an undergraduate named Olivia Mehalick climbed into the vacuum vessel of a working fusion reactor in San Diego to help install a pair of tungsten tiles. NC State had never printed tungsten this big. Her team's largest earlier print was about 50 millimeters across. These were roughly 140 by 130.

Then the reactor fired. Seventy-four plasma discharges later, over three days, the tiles were still there.

What Got Tested

The tiles came from a three-way team: NC State University, Oak Ridge National Laboratory, and the Japanese precision-manufacturing company JEOL. JEOL printed them in Akishima, Japan, on a large-format electron-beam machine. Oak Ridge's Fusion Materials Core Program paid for the fabrication, and NC State did the finishing. From the invitation to test to installation took about eight months.

They went into the lower divertor of the DIII-D tokamak, the San Diego machine that is a Department of Energy user facility. The divertor is the exhaust zone, the part of a fusion machine that takes the worst heat and particle bombardment. If you want to know whether a material belongs in a power plant, this is where you put it.

Why Tungsten Is So Hard to Print

Tungsten melts at 3,422°C, which is why fusion designers love it. It's also brittle below a few hundred degrees, which is why it cracks when you try to build it up layer by layer with most 3D printing methods.

The team used electron-beam powder bed fusion. A beam of electrons melts tungsten powder in a vacuum, and the beam also keeps the surrounding metal above about 1,400°C for the entire build. Staying hot keeps the part out of the temperature range where tungsten shatters. Coverage of the project describes this as the trick that kept the large tiles crack-free.

After printing, the tiles shipped to North Carolina. Wire cutting shaped the mounting features, and a computer-guided grinder took roughly 250 microns off the plasma-facing surface to give it a mirror finish that fits the divertor's curve. They were mounted on copper-alloy pedestals and installed in July.

The Neighbor That Melted

Here's the part that makes this more than a lab demo. According to 3DPrinting.com, a slightly misaligned conventional tile next to the printed pair melted along its leading edge during the run and spattered liquid metal onto one of the printed tiles.

That wasn't in the plan, and it wasn't a good day for the conventional tile. But it handed the team an accidental stress test, and the printed tiles came out intact. Jonathan Coburn of Sandia, who led the testing of the novel tiles, put it plainly in VoxelMatters' coverage: "As far as we could tell in the control room, your AM tiles performed really well."

Why Printing Matters

A fusion power plant will need a lot of tiles. They'll sit in a punishing environment, wear out, and need replacing. Machining each one from a solid block of tungsten is slow and wasteful, and tungsten is miserable to machine. If you can print them to shape, in complex curved forms, the economics start to look different.

Oak Ridge's Michael Kirka framed the milestone as showing that printed tungsten tiles can go into a working tokamak and come out intact. NC State's Tim Horn said the key was finding "hardware and partners who could move fast." That's a manufacturing story as much as a physics one.

Our Take

For years, the headline question in fusion was whether you could make a plasma that produces more energy than it uses. That question isn't settled, but it's no longer the only one. Every company we cover has to answer a second set of questions about supply chains and parts: where do the magnets, the heating systems, and the wall materials come from, and can anyone make them at volume?

We've seen the same pattern in this week's other fusion stories. Proxima Fusion's deals with Japanese suppliers are about tape and gyrotrons, not plasma physics. The Commerce Department's letters of intent for Helion and Commonwealth Fusion Systems are for power chips and electronics. The tungsten test belongs on the same list. Fusion is slowly turning into a manufacturing problem, and that's a good sign: it means the physics is far enough along that the boring parts matter.

What We Cannot Conclude

• Three days and 74 shots isn't a lifetime. A power plant divertor has to survive far longer. This shows survival through a short, hard run, not through years of use.
• We don't have the full-size tile's heat load. The 2025 test on small printed buttons saw about 20 megawatts per square meter, with brief peaks above 35. Coverage of the full-size run doesn't give a comparable figure.
• The wear analysis isn't done. The tiles are being removed, imaged and surface-scanned. Detailed erosion results are still coming, and NC State plans to lead the analysis.
• DIII-D isn't a power plant. The machine runs pulses, not continuous power, and doesn't burn deuterium-tritium fuel, so the neutron damage a real plant would cause isn't part of this test.

Why It Matters

• Printed parts are now in the room. This is a full-size additively manufactured tungsten component surviving inside a working U.S. tokamak.
• The failure mode got tested by accident. A melting neighbor is the kind of event that wrecks a test, and the printed tiles shrugged it off.
• It points at the supply chain. Fusion companies will need components at volume, and a method for printing tungsten without cracks is a piece of that puzzle.
• There's a long way to go. Passing a three-day test is a first step, not a qualification.