[Upcoming K-Robot Era]③South Korea Sandwiched Between US and China... Lacking a Full-Stack Ecosystem

South Korea’s robot software industry is often described as being structurally “sandwiched” between the power struggles of the United States and China. The United States holds an advantage in artificial intelligence (AI) and foundational software technologies, while China, leveraging massive capital and government support, is adding software capabilities to its competitive low-cost manufacturing. In this environment, South Korea is said to remain mostly at the level of manufacturing (hardware). In particular, compared to the US and China, which have established vertically integrated physical AI systems, Korea is seen as lacking the virtuous cycle and key links that generate synergy.

On the 3rd (local time), a robot is dancing at the AI Connecting Ideas exhibition booth at the Mobile World Congress (MWC) 2026 held at Fira Gran Via exhibition center in Barcelona, Spain. 2026.3.3 Photo by Jin-Hyung Kang

원본보기 아이콘

According to the Ministry of Science and ICT’s “2024 Technology Level Assessment Results” released on the 6th, the technology gap between the US and South Korea in the ICT and software sectors stands at 1.7 years, a reduction of 0.3 years compared to the previous evaluation. However, China narrowed its gap with the US from 1.4 years to 0.9 years during the same period, accelerating its pursuit. Notably, China has overtaken South Korea in these fields since 2018, and the gap continues to widen each year.The technology assessment, conducted every two years, focuses on 11 major scientific and technological fields, including ICT and software.

China’s rapid progress was also evident in the evaluation of national strategic technologies. In AI, South Korea was at 80.6% of the US level, with a 2.1-year gap, whereas China was only 0.9 years behind the US (93.0%). In advanced robotics and manufacturing, the gap was 2.3 years between South Korea and the US, but only 1.3 years between China and the US, showing China’s superiority.

Since the evaluation is based on 2024 technology levels, some analysts note that South Korea may have narrowed the gap since then, but so has China. In its report, the Ministry of Science and ICT stated, “China is no longer just a follower; in some areas, it is already redefining global standards and markets. As the need for securing data resources, standardization, and building AI convergence infrastructure grows, South Korea must respond strategically.”

원본보기 아이콘

Physical AI, exemplified by robots and autonomous vehicles, refers to the realization of an AI agent’s (assistant’s) intentions and plans in the real world through sensors and actuators. Unlike traditional automation systems, AI agents serve as the “brains” (software) that can flexibly respond to various variables, while sensors and actuators act as the “body” (hardware) that connects to the real world. Ultimately, the key to physical AI is “interaction,” and the focus lies in building a system ecosystem and securing high-quality data to support this.

The United States boasts world-leading competitiveness in this field. For example, Tesla’s Full Self-Driving (FSD) serves not only as an autonomous driving function but also as a laboratory for physical AI. FSD does not rely on highly detailed maps of specific areas but uses camera-based visual information to make real-time decisions. This gives it the versatility to adapt to new environments. Tesla employs an end-to-end approach, in which a massive neural network processes the entire chain from input (video) to output (steering and acceleration). This is a method clearly different from Google’s Waymo, which operates based on pre-learning and updates.

The Tesla Optimus robot directly inherits and utilizes the perception, environmental understanding, and path planning software from FSD. The vast amount of data collected daily by Tesla vehicles worldwide is used to “train the brain” of these robots. Furthermore, Tesla has built its own infrastructure and ecosystem, developing everything from custom chips for AI inference to the Dojo supercomputer in-house.

At the Hyundai Motor booth at CES 2026 held in Las Vegas, USA, the robot Atlas is demonstrating part transportation. Photo by Yonhap News

원본보기 아이콘

China, which accounts for over 80% of global humanoid robot production, is also evaluated as having vertically integrated its hardware and software supply chains and established a practical robotics ecosystem. The government is providing comprehensive support, emphasizing “embodied intelligence”-a next-generation AI robot technology that learns through interaction. Not only has China moved beyond its image as a low-cost production base by internalizing core components, but Baidu’s “ERNIE Bot” and Alibaba’s AI are embedding end-to-end data into Unitree robots and others. In regions like Shenzhen, thousands of robot parts manufacturers are clustered together, enabling rapid supply.

In contrast, South Korea currently lacks its own physical AI models optimized for controlling robots’ physical movements, and the scale of data available to train AI is critically insufficient. Unlike China’s Alibaba or robot manufacturers, there is no domestic open robot operating system (OS) or standard platform readily available for anyone to use, leaving companies to fend for themselves. That said, there are some noteworthy efforts. The most prominent is Hyundai Motor Group, which is building a “mobility intelligence” system through Boston Dynamics, autonomous driving data, and its own robotics lab.

To overcome these limitations, experts suggest building a demand-driven application ecosystem. Beyond simply expanding robot production capacity, it is necessary to establish institutional foundations and demonstration environments so that robots can be applied across various industries such as manufacturing, logistics, and apparel. Eun-Kyo Cho, head of the China Research Team at the Korea Institute for Industrial Economics & Trade, analyzed that this approach could help gradually build a “K-Robot” ecosystem.

Cho stated, “The robot industry is a convergent sector that is difficult to grow at the level of a single company. The nation should provide comprehensive support at the level of the industrial ecosystem, leveraging existing strengths in semiconductors, equipment, and components,” adding, “Institutionally supporting convergent demand across industries will be a crucial national task.”

Related Articles: ①China, once obsessed with the shell, now fixated on the brain... HW to SW competitiveness shift

②MWC becomes a battleground for Chinese robots... From hardware to robotics AI

© The Asia Business Daily(www.asiae.co.kr). All rights reserved.