The shipbuilding gap.

Those large merchant ships made up 57% of the world’s tonnage. The US delivered five, about 0.1%.

A leaked 2023 Office of Naval Intelligence slide put China’s shipyard capacity at 232 times America’s: about 23 million tons against under 100,000. The Navy called it estimates from public sources. It measures capacity, not ships delivered, and its method is not public.

China’s state shipbuilder, CSSC, once delivered more than 250 ships in a single year. CSIS counts that as more commercial tonnage than the whole US industry has built since the end of World War II.

Satellite images of Jiangnan’s base on Changxing Island show Evergreen container ships moored beside Type 055 warships. China calls this military-civil fusion (军民融合, jūnmín rónghé). Commercial orders pay for the cranes, workers and steel that naval work also uses.

Britain led for about a century. Japan took the lead in 1956, Korea in 2000 and China in 2010. Each new leader built ships in a different way. Today China, Korea and Japan build about 95% of the world’s ships.

In the 1890s Britain built about 75 to 80% of the world’s ships, two-thirds of them on the Clyde in Scotland.

Economic historian Edward Lorenz found the problem was the system, not the workers. Each craft union held rights over certain machines and materials, which protected its members from hire-and-fire cycles. Those rights made welding and prefabrication very hard to adopt. A riveter took about six months to train. A wartime welder took about two weeks.

From 1941 to 1945, 18 emergency yards built 2,710 Liberty ships, a standard cargo design. The ships were welded, not riveted, from about 250,000 prefabricated parts each.

Henry Kaiser’s yards hired people who had never seen a shipyard. Prefabrication and standard work made the jobs quick to learn. By mid-1944, 41% of the welders at his Richmond, California, yards were women.

Economist Peter Thompson went back to the yard records decades later. The yards had also roughly tripled their equipment. Counting that cuts the share of the speed-up credited to learning about in half. Some of the speed came at the cost of quality: at the worst yard, over 60% of ships had cracked hulls.

When production stopped, the yards forgot 3 to 5% of what they had learned each month. Flow and standard work without quality built in at the source led to repair work at sea.

From 1936, the US paid its yards up to half the cost difference with foreign yards. In 1981 the Reagan administration ended that subsidy. Every rival country kept paying its own.

Wages do not explain the gap that followed. In 2010 US yard labor cost about $18 an hour, against about $16 in Korea. The difference is volume. A yard that builds two ships a year cannot learn as fast as one that builds twenty. A 2005 benchmark study found US yards built few ships in modules and fitted little equipment before assembly. It told them to “become more lean.”

Post-war Japan stopped building hulls piece by piece. Its yards welded large blocks, fitted pipes and wiring inside each block, then joined the blocks in the dock. Government-planned orders kept demand steady while the yards learned. By the mid-1960s Japan built about half the world’s ships.

Korea took the method to a larger scale. In the 1990s Korea built giant docks while Japan, after government-ordered capacity cuts in 1980 and 1988, held back. Japan now builds about 13% of the world’s ships.

China paid for its rise. Economists Barwick, Kalouptsidi and Zahur measured about $91 billion in hidden subsidies from 2006 to 2013. 69% of it paid for new yards, many of them inefficient, and the returns were poor. Capacity is not capability.

Some Chinese yards now improve with each ship. Hudong-Zhonghua builds LNG carriers, a market Japan and Korea held five years ago. Chinese trade press credits digital shipbuilding (数字化造船, shùzìhuà zàochuán) and parallel production (并联式, bìngliánshì), where blocks are built at the same time instead of one after another.

In December 2024, Korea’s Hanwha bought Philly Shipyard from Norway’s Aker. The yard built most of the large US ships for domestic trade, at about 1.5 ships a year. In 2025 it announced an expansion with two new docks.

Philadelphia workers now train at Hanwha Ocean’s yard in Geoje, Korea, the first yard to deliver 200 LNG carriers. In August 2025 the yard won an order for 10 product tankers, the largest US commercial order in over 20 years. Lean readers know the pattern from NUMMI in 1984: a foreign production system brought into a struggling US plant, with its workers retrained in it.

While US port fees loomed in 2025, China’s monthly share of new orders fell below 30%. Once shipowners adjusted, it rose back to 84%. Over the full year it fell from 70% to 63%.

The Navy’s own build times keep growing. Destroyers and submarines that took 5 to 6 years in the 2000s now take 9 to 10. Virginia-class submarines arrive at about 1.2 a year, against 2.0 funded.

Each time the lead moved, it moved to a better production system: rivets to welds, welds to blocks, blocks to scale, scale to state-funded learning. Steady demand lets a yard learn, and a yard that stops building forgets. Philadelphia is now testing whether the method can come back to the US.