AI and Wire Arc 3D Printing Target the Nuclear Supply Gap

The U.S. wants to build more nuclear reactors. It does not have enough domestic forging capacity to make the pressure vessels that hold them. That bottleneck is why Oak Ridge National Laboratory and Idaho National Laboratory are now teaming up to 3D print and AI-qualify nuclear pressure vessels at home.

The partnership, announced at the M2IND summit, pairs ORNL's large-format wire arc additive manufacturing with INL's Prometheus AI platform. The idea is simple: build the vessel, monitor the print in real time, and use the collected data to prove the part is safe without cutting it apart.

In July, ORNL researchers printed a small nuclear pressure vessel on the MedUSA platform. Three robotic arms melted steel wire with electric arcs into a closed, domed vessel roughly three feet by five feet. It showed that a steel alloy relevant to nuclear applications can be formed at scale. Now the work shifts to qualification.

Patxi Fernandez-Zelaia, ORNL's lead researcher, put the goal plainly: use data gathered during printing to judge whether a component is fit for extreme environments. If it works, pressure vessels could be born-qualified, meaning their manufacturing record is the proof regulators see.

This matters because forging a large pressure vessel is slow, expensive, and dependent on a tiny number of suppliers. Additive manufacturing could cut lead times and bring production back to the U.S. The same methods could later apply to reactor internals, heat exchangers, and structural parts for chemical plants or defense systems.

ORNL is already embedding 3D printing across nuclear construction. In 2021, printed fuel assembly brackets developed with the Tennessee Valley Authority and Framatome became the first 3D-printed safety-related parts installed in a commercial reactor, at TVA's Browns Ferry plant. The lab also supplied modular 3D-printed concrete formwork for Kairos Power's Hermes molten-salt reactor in Oak Ridge, the first Generation IV reactor to receive an NRC construction permit.

The effort also overlaps with a separate push at Argonne National Laboratory. Argonne recently submitted a proposal to open an ASME code pathway for laser powder bed fusion in nuclear reactor manufacturing, working with Oak Ridge, Idaho, and Los Alamos. Both efforts rely on real-time monitoring and AI analytics to build a digital record of how each part performs.

The nuclear sector has long treated 3D printing as a promising lab technology. These lab partnerships show it is now a qualification problem, not a hardware problem. The machines can print the parts. The remaining question is whether the data can convince regulators that printed vessels are just as trustworthy as forged ones.