Korea picked ten industries for its humanoids. Homes, care and venues are not on the list

Korea's ₩2.3 trillion humanoid plan, announced August 27, 2026, sets a target of mass-production systems for humanoids specialised to ten industries by 2030. Herald Business lists them: display, aviation, logistics, retail distribution, shipbuilding, automotive, home appliances, chemicals, hospitals and hotels. The list is a map of where Korea already has large domestic operators, and it says less about where humanoids are likely to earn money first.

Six of the ten are manufacturing: display panels, aircraft, ships, cars, appliances and chemicals. Two are the movement of goods, logistics and retail distribution. Two are services, hospitals and hotels. All ten are workplaces with an employer who can buy equipment and measure it against a labour cost, which is what the plan's other instruments — 1,080 state-purchased humanoids by 2030 and a core component localisation target of 80%, up from about 45% — are built around.

What the list leaves out

Some of the obvious omissions are handled elsewhere. Care, agriculture, construction, ports and defence are among the eight sectors in a separate ₩2.8 trillion physical AI demonstration budget announced the same day. That budget funds field trials of robots of any form, not humanoid production lines, so the split reflects which sectors the government thinks need a humanoid and which only need automation.

Elder care shows why. The care robot Korean local governments already pay for is not a humanoid. Hyodol, a talking doll for older people living alone, had more than 12,000 units in homes as of August 2025 at ₩1.6 million each, according to Rest of World, with Seoul's Guro district alone distributing 412 since 2019. The buyer is a welfare office, and the product works without arms or legs.

Two categories appear in neither budget: the household and entertainment venues. The only home-first humanoid with a public price, 1X's NEO at $20,000 or $499 a month, is a US product, and Korea's plan has no instrument aimed at domestic buyers. Performance and venue robotics sell tickets rather than replacing shifts, so they do not fit a framework measured in production lines and localisation rates.

Where the early deployments are

Logistics is the one listed industry with a commercial humanoid deployment. On September 3, 2026, CJ Logistics put two dual-arm humanoids to work inserting cushioning into boxes at Olive Young's Yangji distribution centre in Yongin, according to Seoul Economic Daily, after tests at its Gunpo fulfilment centre in 2025. Robotis supplied the hardware, Aidin Robotics the hands and RLWRLD the foundation model; picking, sorting and inspection are the planned next tasks.

Retail distribution is at the trial stage: Coupang has been testing Rainbow Robotics' wheeled, dual-arm RB-Y1 in a fulfilment centre since June. In hotels, Lotte Hotel Seoul staff wear cameras on their heads, chests and hands so that RLWRLD can collect training data; no robot is in service, and the company is aiming for deployment by 2028.

The other listed sectors are further out. Samsung Medical Center demonstrated two humanoid surgical assistants in mid-September from a consortium including Rainbow Robotics and Aidin Robotics; Korea Biomedical Review reports a first clinical trial targeted for 2029, as a Class III device. Hyundai has set 2028 for Atlas on its Georgia plant lines, and HD Hyundai is a disclosed partner of Persona AI's shipyard welding humanoid, with no contract value or unit count published.

Why it matters

The list is a production-side bet: pick the sectors big enough to justify dedicated humanoid lines, and let volume pull component costs down toward the 80% domestic target. The early evidence mostly supports the choice of sectors. The one paying deployment is in a listed industry, and the hospital, hotel, automotive and shipbuilding programmes all carry 2028 or 2029 dates that fit a 2030 horizon.

What the evidence does not support is the idea that the humanoid form is needed everywhere on the list. The first commercial job is packaging, one narrow repeated motion, done by wheeled dual-arm machines rather than bipeds. Where Korea has already bought robots at scale, in elder care, it bought a doll. The households and venues outside both budgets will be shaped by private buyers and by products priced like NEO, not by state procurement.