Three former SpaceX engineers have traded rocket engines for steel beams. Their startup, 1872, officially opened Factory One in Cincinnati on July 22, and the pitch is straightforward: use AI-driven software and robots to automate steel fabrication for infrastructure that America desperately needs but increasingly lacks the welders to build.
The company’s first target is steel skids—rectangular frames that serve as movable foundations for modular buildings. The initial customer base? Companies building AI data centers and small modular nuclear reactors, two sectors racing to scale while staring down a brutal labor crunch. The American Welding Society projects the US will need 320,500 new welding professionals by 2029, a gap driven by retirements and surging demand across data centers, semiconductor fabs, and shipyards. Tighter immigration restrictions under the Trump administration have only squeezed the pipeline further.
But 1872 isn’t chasing full autonomy. CEO Dan Summers told Ars Technica the company is “building towards autonomy, but we’re not necessarily building in a dogmatic fashion towards full autonomy.” He estimates they may hit 80 percent autonomous operations and stop there, citing diminishing returns on the final stretch. That’s a pragmatic stance you don’t often hear from robotics startups, which typically pitch visions of lights-out factories. It also reflects where industrial automation has actually landed: supervised autonomy, not human-free utopias.
The timing matters more than the technology. The US hasn’t seriously invested in automating heavy fabrication the way it has in automotive assembly or electronics. Data centers and nuclear infrastructure are both bottlenecked by the same shortage of skilled welders and fitters. If 1872 can prove out even partial automation on steel skids by 2027, it would give those industries a template for scaling without waiting on a labor force that isn’t coming. Whether the economics hold beyond the prototype stage is the real question—and one Factory One will have to answer quickly.
