As China's artificial intelligence (AI) and robotics industries accelerate commercialization, South Korean companies are urged to focus on 'learning-based site design' rather than merely adopting robots.
Kim Soo-yeon, head of the EY Han Young AI Center, stated during a special lecture at the 18th Good Growth, Good Jobs Forum (2026 GGGF) held at The Plaza Hotel in Seoul on September 3, "The core competitiveness of China's AI and robotics industry lies not in the large models themselves but in the 'flywheel' structure that capitalizes on on-site data and know-how." This virtuous cycle of accumulating data and experience continuously enhances AI, which is considered a key factor in China's rapid application of AI technology in the field.
Kim shared insights from her recent observations at the World Artificial Intelligence Conference (WAIC) and the World Robot Conference (WRC 2026), highlighting the vibrant trends in China's AI agents and physical AI ecosystem.
"China is quickly pushing to apply and monetize AI agents in industrial settings," Kim noted, adding, "It is evolving into a robot-centric economy beyond simple software AI." She explained that the massive investments and numerous companies involved are driving the rapid pace of technological advancement.
According to Kim, the Chinese AI agent market is experiencing explosive growth. Investment in this market is expected to surge from 55.9 billion yuan (approximately $8.5 billion) in 2025 to 815.1 billion yuan (about $123.5 billion) by 2030, with an average annual growth rate of 70.9%.
Amid this quantitative growth, there is a strong focus on reducing reliance on foreign technology and strengthening domestic capabilities. Approximately 80-85% of AI agent projects are being developed in-house by companies. In fact, 93% of Chinese firms are utilizing domestic models, with 68% adopting a 'multi-model' approach that combines optimal AIs for different tasks. This two-track strategy employs high-performance models for complex tasks and cost-effective models for simpler ones, while critical tasks are managed through independent internal networks to enhance control.
The growth of agent AI is also rapidly expanding into the physical AI sector. In the first half of this year, China's investment in physical AI reached approximately 93.5 billion yuan (around $14 billion), a fivefold increase compared to the same period last year. The annual production of humanoid robots is expected to exceed 100,000 units, driven by significant investments from major Chinese companies, including state-owned enterprises, Ant Group, and JD.com, into the robotics AI ecosystem.
However, Kim pointed out that there are still limitations in the actual commercialization level in industrial settings. Most robots deployed in the field are traditional industrial robots, such as wheeled or quadrupedal models, rather than bipedal humanoids.
The development of sophisticated human-like robotic hands (general-purpose hands) remains in the early stages. Kim explained, "In practice, specialized end-effectors, such as suction or claw-type tools tailored for specific tasks, are primarily used. There is still a need for technological advancement and validation before humanoids can effectively complete tasks in manufacturing settings."
Kim urged South Korea to accelerate its strategic response, moving beyond being a passive consumer of technology to developing its own strategies in light of China's rapid AI and robotics advancements. "The opportunity for South Korean companies lies not in passively acquiring completed AI models or robotic products, but in strategically systematizing and designing the workplace to ensure that actual work processes and the experiences of on-site workers are continuously accumulated as data and learned automatically," she emphasized.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.