The Navy just put $900 million behind AI welding robots. Commercial construction has the same welder shortage HII is solving for.

Huntington Ingalls Industries, the country's largest military shipbuilder, just signed up to $900 million in production agreements to put AI-guided robots on welding and finishing work it can't staff with people. The gap it's solving — not enough certified welders, an aging workforce, and not enough young people replacing them — is the same gap structural steel erectors and mechanical subs are already living with on commercial jobs.

What did HII actually agree to?

On August 6, HII signed performance-based production agreements with two robotics companies: Path Robotics, which builds AI-guided robotic welding systems, and GrayMatter Robotics, which builds robotic sanding and finishing systems. The deal is worth up to $900 million over seven years, with Path Robotics eligible for up to $600 million of that. The work runs through HII's HYPR program and covers welding, grinding, blasting, painting, assembly, and inspection across Navy shipbuilding — aircraft carriers, submarines, destroyers, amphibious ships, future frigates, and unmanned surface vessels.

"We are committed to making generational investments to develop new shipbuilding capability and to expand capacity," said Eric Chewning, HII's executive vice president of maritime systems and corporate strategy. "Together, we are defining a new approach to robotics in shipbuilding: automation that can adapt to extreme levels of complexity, mix, and size." Path Robotics CEO Andy Lonsberry framed it as validation of a longer bet: "Physical AI can scale shipbuilding capacity in one of the world's most demanding production environments." HII says it plans to outsource more than 2.5 million hours of shipbuilding work in 2026, up 30% from 2025, as it expands a distributed manufacturing network that fabricates more ship sections outside its primary yards before final assembly.

Why is this different from a standard welding robot?

Factory welding robots have existed for decades, but they're fixed-cell machines: bolt a robotic arm to the floor, feed it identical parts, and it repeats the same weld path all day. That works for automotive assembly lines. It doesn't work for shipbuilding — or for structural steel and piping — where every joint is a slightly different fit-up on a part that isn't standardized.

Path's Obsidian model is built around machine vision that reads the actual joint in front of it and adjusts the weld path accordingly, rather than executing a pre-programmed motion. Paired with Path's Rove platform — which puts that same AI on a mobile, legged robot instead of a fixed arm — the system can move to an irregular, unfinished part instead of waiting for the part to come to it. That mobility-plus-adaptability combination is the specific thing that has kept robotic welding largely confined to repetitive factory production rather than the field-condition work that fills most of a structural or mechanical sub's welding hours.

How bad is the welder shortage this is aimed at?

That last figure matters most for the audience HII's deal doesn't directly touch: sheet metal, HVAC, and piping subs bidding commercial work, not aircraft carriers. The math is the same one driving the Navy's spending — an aging certified workforce retiring faster than apprenticeship pipelines can replace it — just with a much smaller check behind it.

Should a sub actually plan around this?

Not on the jobsite, not yet. Robotic welding's proven ground today is the shop: pre-fab structural connections, piping spools, and repetitive fabrication that gets built off-site under controlled conditions and delivered to the field. That's where fixed-cell robotic welding already works and where a mechanical or structural sub evaluating automation gets the clearest return. The $900 million HII deal is worth tracking because it's real capital, from a company with no incentive to overstate results, aimed specifically at closing the mobility gap that's kept robots off the kind of irregular, field-condition welding that fills a sub's actual backlog. If Rove or a similar mobile system holds up in a shipyard, that's the signal a jobsite-deployable version is getting closer — not a reason to budget for one this year.

Related: Autodesk and the University of Florida opened a robotics lab aimed at the same framing labor gap in July — different trade, same math on an aging, undersized workforce.

Forward this to the estimator who just lost a bid because the sub couldn't staff enough certified welders to hold the schedule.

Construction AI Brief publishes new coverage on AI's construction stakes multiple times a week. Subscribe at constructionaibrief.com.