Samsung Electronics chairman Lee Jae-yong visited the company’s Robotics Experience business promotion office in Seocho, Seoul on 27 October, reviewing the state of humanoid development and how ready the technology is to be sold.
The date was not incidental. It marked four years since he took the chairman’s role, and the tour functions as a public progress note on a programme Samsung has been assembling quietly since the middle of the year.
A single room for a scattered effort
The office opened in July 2026 and reports directly into the Device eXperience division. Its stated job is to consolidate work that had been spread across the group: long-term strategy, core technology development and commercialisation.
The hiring pattern is the tell. Samsung brought in Lee Dong-geon, previously responsible for robot strategy at Hyundai Motor Group, and added robot control and hand-manipulation specialists from Seoul National University and Ajou University. A conglomerate that intends to buy its way into a market hires bankers. One that intends to build hires the people who have shipped the hardware.
Data before demonstrations
Samsung is building a data factory at its Gumi plant in North Gyeongsang Province to collect robot learning data, and runs research centres in the United States, China and Japan. The sequencing it has described is deployment in its own plants first, to prove capability and stability, then industrial use, then the home. By 2030 it wants its domestic and overseas production sites converted into AI-based autonomous factories.
That order is the opposite of the one most humanoid startups follow, which is a filmed demonstration followed by a pilot customer followed by a search for a use case.
Samsung already owns the first customer. It runs some of the most complex electronics assembly on earth, which is a hard environment and an honest test. It also owns Rainbow Robotics, having raised its stake to 35 per cent at the end of 2024 to become the largest shareholder, giving it a robot platform to pair with its own AI and software.
The number that decides the market
Chris Hauser, a lab director at the office, said at the Samsung AI Forum in September that the company intends to extend from industrial robots into home and service robots, and that a general-purpose robot deployed in a home or a factory needs a task success rate of 99.99 per cent to be reliable.
That standard explains the caution. A humanoid that succeeds 99 times in 100 looks impressive in a video and unusable in a workplace, because the hundredth attempt is the one that drops the part, damages the line or injures someone. Precision hand manipulation and cost are the two constraints Samsung has named, and both are engineering problems with no shortcut.
Samsung is also reviewing whether to show the technology at CES 2027 in Las Vegas in January, which would be its first public statement of intent rather than capability.
Editor’s note: This is an adapted translation of the original Gasgoo report. It has been trimmed and restructured for readability for an international business audience. The full original (in Chinese) is at https://www.gasgoo.com/apps/e060bad9eca91db7f01403fc7120041f/robot/news/70476074.