The Navy’s AI bet to fix its submarine bottleneck

The answer to America’s submarine bottleneck, the U.S. Navy has decided, lies as much in software as it does in steel. A new multibillion-dollar facility in Cherokee, Alabama, aims to harness AI and robotics to build submarine components faster and more reliably.

The automated “factory of the future” will produce parts for the Navy’s Virginia-class attack submarines and Columbia-class ballistic missile submarines, both central to the U.S. fleet. It will cost $2.4 billion to develop.

“This factory is the first of three facilities designed to address the most critical bottlenecks in the maritime industrial base,” said John C. Phelan, secretary of the Navy, in a statement. The bottleneck is significant: a shortage of labor.

The project is a major public-private push to revive U.S. submarine manufacturing capacity through heavy automation, says Chris Power, founder and CEO of Hadrian, the company behind the Alabama facility. “In the U.S., just the submarine program alone is 70 million man hours in deficit,” Power says, noting that the gap traces back in part to the offshoring of manufacturing jobs in the 1980s and 1990s. “There aren’t that many skilled workers to hire.”

The company’s answer is to layer in automation and AI. “We have to give the American workforce superpowers of AI and robotics to allow them to be more productive,” Power explains.

He says the site will begin producing components and large subassemblies later this year, then ramp up over the following 18 to 24 months. The goal is to automate “80% to 90% of the key efforts that are really complicated.”

That productivity push comes as the U.S. faces mounting geopolitical pressure, with demand for military hardware unlikely to ease. Automation may help close the production gap. It does less to solve what happens after deployment, when equipment breaks and needs to be fixed.

That talent shortage extends beyond wartime scenarios. Cynthia Cook, senior fellow in the defense and security department at the Center for Strategic and International Studies, says the Alabama plant is part of a broader effort to rebuild a hollowed-out shipbuilding base. The U.S., she argues, can no longer rely solely on coastal shipyards. Inland factories can produce modules and components in regions where labor is more available.

There will “still be a huge need for repair and sustainment,” Power says, adding that this work will remain “manual for a while, because it’s so nuanced.” Factory automation, in his view, should handle repeatable, high-volume production, while skilled workers focus on complex, higher-value repair.

This tension is not theoretical. The USS Gerald R. Ford is currently undergoing repairs in Crete after a fire in its laundry area caused significant damage, with fixes expected to take a year or more.

Cook agrees that maintaining repair expertise will remain critical. “When a ship comes in for maintenance, you really need a lot of folks who have tacit knowledge about ships, what can go wrong and how to fix things,” she says. “A repair skill is different from a new-build skill.” She believes the system will retain enough capacity to preserve that knowledge.

Others are less certain. Automation can deepen labor challenges if it demands skills the current workforce lacks, especially without a coherent strategy for recruiting, retaining, and training workers, says Christophe Combemale, assistant research professor of engineering and public policy at Carnegie Mellon University. He is also concerned about how little visibility Washington has into the capacity and quality of U.S. manufacturing training pipelines.

At the same time, Combemale does see some upside. “Some aspects of automation actually improve the longevity of this knowledge by making tacit knowledge explicit,” he says—warning, however, that without careful planning expertise could erode. “Are you making the labor shortage worse by demanding new skills that your incumbent workforce doesn’t have?”

It’s a question that may only be answered when the system is tested—at which point it will become clear whether AI and automation have strengthened the industrial base or quietly reshaped its limits.

View the full article