Japanese and Korean shipyards bet on robots to keep up with China

What the FT reported

The backdrop is simple. China dominates shipbuilding, and its main rivals are running out of people. Japanese and Korean yards have full order books for years ahead, competition is rising, and young workers avoid hard welding and painting jobs. Hence the push for smart yards, where robots take over repetitive work and people supervise them.

Both countries matter strategically to the United States: they are allies that can build complex ships, from LNG carriers to submarines, and American plans to rebuild shipbuilding capacity depend on them. So this automation race is about more than economics.

The most concrete figure in the report came from HD Hyundai Samho, which says a robot can handle up to 50 welding cells a day against about 30 for a worker. That shortens delivery by a few days. It sounds modest, but for a yard with three to four years of backlog every saved day is money.

Korea: hundreds of robots and a welding humanoid

By February 2026 HD Hyundai's shipbuilding units had 211 industrial and collaborative robots, up 64% in a year (BusinessKorea). At the Ulsan yards robots find welding points from design data and move between hull blocks on their own. By 2030 the group wants a small fully automated line, a so-called dark factory.

Hanwha Ocean has spent 160 billion won over two years and targets 70% automation by 2030. It says one operator can run up to 10 welding robots on standard work. Samsung Heavy Industries is developing AI welding robots with Rainbow Robotics and admits automation is costly and technically hard, but necessary to survive.

The part most relevant to our catalogue: in March 2026 HD Hyundai signed a deal with US-based Persona AI to develop a bipedal welding humanoid for shipyards, and HD Hyundai Samho, working with Neura Robotics and LG CNS, is assessing humanoid ROI with a rollout planned from 2027.

Japan: Fanuc, Kawasaki and a state programme

Japan's shipbuilding workforce fell from over 90,000 in 2016 to 70,300 in 2023 (OECD review). According to the FT, Fanuc has built its first two welding and painting robots designed specifically for shipyards and eventually wants as many robots in the industry as people, about 75,000, so yards can run at night.

On 6 October 2026 Kawasaki Heavy Industries showed a prototype four-legged robot, 60 by 44 cm, that holds onto a hull with electromagnets, climbs vertical and curved surfaces and decides how much welding a seam needs. Its AI and digital-twin stack is being built with NVIDIA. Trials at the Sakaide yard start in 2027, with practical use targeted for 2028.

Separately, KHI, Shin Kurushima Dockyard and Namura Shipbuilding are developing legged and wheeled robots for welding and painting hull-block exteriors, at height and on curved surfaces, under a government programme running to March 2027, with ClassNK and ABS supporting the evaluation.

China: faster and with state backing

What worries Japan and Korea most is that Chinese yards are not only cheaper but also faster at building and deploying robots. Experts disagree on whether China is already ahead, but many think any gap could close in about five years. State backing makes it easier for Chinese companies to absorb the risks of automation.

An example from a neighbouring field: in June 2026 a CNOOC subsidiary put an AI welding robot for offshore platforms into service in Tianjin. It handles steel up to 70 mm thick and parts up to 30 tonnes, and its developers claim a first-pass weld quality rate above 98%. These figures come from state media and have not been independently checked.

Why a shipyard is so hard to automate

A car plant turns out thousands of identical cars, and a robot on the line repeats one motion. Ships don't work that way: each has its own design and layout, the blocks are huge, and work happens at height, in cramped compartments, in dust and outdoors. That makes a classic robot arm hard to use, so yards are moving to mobile robots that find the seam from the drawing and to legged machines that cling to the hull itself.

The industry itself says robots won't replace people entirely. Installing equipment in narrow interior spaces still needs human hands, and Korean unions have already secured separate talks on job security.

What this means for robot buyers and makers

Shipyards are becoming one of the first heavy-industry customers for humanoids and legged robots, not as a trade-show demo but with a specific job and an ROI case. For makers this signals that demand appears where people are physically scarce and the work is dirty and dangerous. For buyers in other fields, such as steel structures, bridges and tanks, it means solutions proven in shipyards may be available within a couple of years.

If you are weighing robotic welding or inspection, run the numbers in our ROI calculator and browse legged platforms in the catalogue. If you build or integrate such robots, list your model on RoboHub for free.