China's Physical AI Revolution: A New Era of Robot Development

In a rapidly evolving landscape, Chinese companies are pushing the boundaries of physical AI, with a focus on real-world applications and commercial value.

TL;DR
• Chinese companies are leading the development of physical AI, with a focus on real-world applications and commercial value.
• Real-world operational data is crucial in improving physical AI models.
• Humanoid robots are being commercialized in 2026, with a focus on intralogistics use cases.

Summary
The Chinese robotics market is expected to continue its growth trend, with industrial robot sales projected to increase by around 10% in 2025. This growth is driven by the country's strong demand for automation, particularly in the electric vehicle manufacturing sector.

Key drivers of this growth include the increasing adoption of collaborative robots, which offer easier installation, high flexibility, and relatively low energy consumption.
• Easing installation and high flexibility
• Relatively low energy consumption
• New applications, particularly in electric-vehicle manufacturing

The market is also witnessing a shift towards more intelligent solutions, such as semi-autonomous welding. Intense price pressure and consolidation are expected to persist, but Chinese companies are moving to the forefront of development in physical AI. Investment in early-stage robotics companies in China in 2025 was roughly five to six times higher than in Europe.

Several Chinese companies, such as X Square, Meituan, and Alibaba, have raised significant funding to develop embodied AI. X Square's WALL-A system processes visual inputs and understands causal physical relationships, enabling robots to infer hidden objects and autonomously correct errors. Chinese companies are also focusing on developing full-stack systems built around clearly defined tasks, which enables faster iteration cycles, higher reliability, and better cost/performance.

The use of high-quality, real-world data is becoming increasingly important for improving models. TARS Robotics has released an open dataset, 'World in Your Hands,' which includes high-precision 6D motion data. This data collection approach can generate up to 1.8 TB per operator, per day of high-precision 6D motion data.

The strategic positioning of some companies, such as Galaxea Dynamics, could help build a platform role in AI-enabled robot control over the medium to long term. Galaxea's plug-and-play approach enables developers to get the system running in under 30 minutes and then fine-tune it for new tasks.

Intralogistics use cases, alongside simple assembly and inspection tasks, are among the first applications that Chinese makers of humanoids aim to commercialize. The anticipated IPO of Unitree in 2026 could become one of the larger Chinese robotics tech listings in some time, with more likely to follow.

As the physical AI landscape continues to evolve, it's essential to stay ahead of the curve and understand the key drivers of this revolution.