The most consequential robotics deal in Asia on Thursday was not the biggest one. In Shenzhen, a startup called Kinetix AI disclosed more than RMB 500 million — roughly $74.5 million — in Angel+ financing to build humanoid robots, dexterous hands, embodied AI models and the first-person training data to drive them. The entire stack, all at once. In Anyang, a manufacturing suburb south of Seoul, a company called AIDIN Robotics announced KRW 16 billion, about $12.1 million, to keep making one thing: the component that tells a robot how hard it is squeezing.
Only one of those companies has a shipyard to work in starting now.
AIDIN said on September 11 that it had closed a strategic round from two corporate investors. HD Hyundai Robotics, the robot arm of the shipbuilding conglomerate and the largest robot-solutions business in Korea by revenue, put in KRW 13 billion (about $9.8 million) for an equity stake it declined to size, at the investee company's request. Samsung Venture Investment supplied the remaining KRW 3 billion (about $2.3 million). Neither is a financial investor in the conventional sense, and the terms make that plain: alongside the cheque, HD Hyundai committed to co-developing products with AIDIN and to putting them into its own yards.
What a force sensor actually solves
AIDIN was founded in 2019 as a spin-out of the robotics lab at Sungkyunkwan University by Choi Hyouk-ryeol, a professor of mechanical engineering and a leading Korean researcher in force and tactile interaction, and his former student Lee Yoon-haeng, now chief executive. The company designs and manufactures capacitive six-axis force-torque sensors small enough to sit inside a robot's joints and fingertips, paired with the force-control software that interprets their output.
That pairing is the point. Plenty of firms sell sensors; far fewer sell the control layer that turns a stream of force readings into a robot that modulates its own grip. Competitiveness in physical AI is shifting from how precisely a robot can move to how well it controls what happens when it touches something.
If a robot cannot perceive and control force and contact on its own, humanoids cannot move beyond demonstrations and into the field, Lee said. AIDIN Robotics is the only company in Korea that has vertically integrated that domain, from the sensor through to force-control technology.
The case for capacitive sensing is economic as much as technical. AIDIN's patented approach, which exploits what it calls the fringe effect, claims sensors roughly one-third the size and one-tenth the price of conventional units while preserving resolution — the combination that matters when a humanoid needs dozens of them, one per joint. The design won Korea's IR52 Jang Young-sil award, and the company already exports to ten markets including Germany, France, China and Japan.
Why a shipbuilder wrote the cheque
HD Hyundai is not buying a science project. The two companies will begin by combining AIDIN's force and tactile sensors and control software with HD Hyundai Robotics' existing robot platforms, then co-develop a heavy-duty five-finger robotic hand specifically for shipbuilding and heavy industry, intended for the humanoid HD Hyundai Robotics has in development.
The nearer-term revenue is duller and probably larger. Shipyards are full of surface-treatment work — polishing, grinding and finishing hull plating and block weld seams — done by hand because the surfaces are irregular and conventional automation cannot cope. It is precisely the category force control unlocks. The partners plan to deploy at Korean yards first, then expand into broader Korean manufacturing and overseas, with U.S. shipyards named as a target.
This investment is a strategic choice to combine the capabilities of both companies and develop robot solutions based on physical AI, an HD Hyundai Robotics official said. We will continue to pursue strategic partnerships with promising technology firms to help strengthen the global competitiveness of Korea's robotics industry.
AIDIN will spend the money on manufacturing. The company is building toward annual capacity of 30,000 force-torque sensors, plus quality-verification systems and R&D headcount — the unglamorous wager that when humanoid production finally scales, the constraint will be who can actually ship parts.
The picks-and-shovels thesis, wearing work boots
The comparison with Kinetix is instructive rather than invidious. Kinetix raised six times as much to build everything, backed by Vertex Ventures, Fangguang Capital and Wanshi Capital — a bet that the winner in embodied AI will be whoever owns the full stack. AIDIN raised less to build one layer, backed by the companies that will buy it. Both theses can be right. Only one of them produces revenue before the humanoid market exists.
The distinction is the kind of capital, not the amount. A venture fund can give a robotics company money and a board seat. HD Hyundai gave AIDIN money and a deployment site — one of the largest shipbuilding operations on earth, with unstructured surfaces, real tolerances and workers who will notice immediately if the thing does not work. In contact robotics, field data is the moat: force-control know-how accumulated in production environments is what separates a demo from a product. AIDIN now has a customer, a shareholder and a test bed in the same entity.
The round also completes a striking cap table. After a KRW 4.5 billion Series A in 2022 and a KRW 15 billion Series B in 2024 that brought in POSCO Technology Investment, CJ Logistics and Samsung Next, a components supplier now counts shipbuilding, steel, logistics and consumer electronics among its shareholders — which is to say, much of its addressable market owns a piece of it.
Three things to watch. Whether the surface-treatment solution reaches production deployment at an HD Hyundai yard within a year, which is the only proof that matters. Whether the 30,000-unit line gets built on schedule, since capacity is what separates a supplier from a laboratory. And whether Korean-made sensing can hold price against Chinese component makers attacking the same bill of materials from a domestic humanoid ecosystem moving considerably faster. AIDIN has bought itself a shipyard's worth of time to find out.
“If a robot cannot perceive and control force and contact on its own, humanoids cannot move beyond demonstrations and into the field.”— Lee Yoon-haeng, Co-founder and CEO, AIDIN Robotics