Chinese technology companies are accelerating the development of embodied AI systems. The industry is moving beyond chatbots and text generation tools.
Companies now want artificial intelligence to control robots in real environments. This shift could reshape automation, logistics, and household services.
Alibaba recently launched its Qwen3.7-Max model. The company said the model supports advanced “tool-calling” functions. These features allow AI systems to activate software and hardware components independently.
Alibaba explained that the model can guide robots through navigation, obstacle avoidance, and task planning. The company also released several robotics-focused AI tools.
These tools include a robotic gripper agent, a navigation model, and a vision-language system. The systems are designed for physical interaction in real environments.
The company is investing heavily in embodied AI as competition intensifies across China’s technology sector.
Chinese robotics start-up Zeroth also announced a major development earlier this month. Its M1 humanoid robot integrated Tencent’s OpenClaw AI agent framework.
The system allows large language models to understand spoken instructions. Robots can then convert those commands into physical movement almost immediately.
Experts say the industry is entering a new phase.
Wu Bangyi, chief data officer at Tianyu Shuke, said previous AI development mainly focused on digital tasks. He explained that embodied AI aims to help machines understand and operate within the physical world.
He added that the goal is transforming Artificial Intelligence from cognitive intelligence into action intelligence.
Analysts believe investors are watching the sector closely. UBS analyst Carl Berrisford said autonomous agents and embodied AI could become major growth areas.
However, the industry still faces a major challenge.
Companies need massive amounts of real-world training data. Many developers struggle to collect enough high-quality information for robotics systems.
Goldman Sachs identified data shortages as a major bottleneck in a May report.
AgiBot co-founder Yao Maoqing compared robotics training data with large language models. He said GPT-5 used data equal to nearly 10 billion hours.
By comparison, the robotics sector only has about 500,000 hours of high-quality embodied AI data available.
Chinese firms are now expanding testing facilities and data collection projects to solve the issue.
X Square Robot recently partnered with home-services platform 58 Daojia. The companies launched robot-assisted cleaning services in Beijing and Shenzhen.
Under the programme, robots work alongside human cleaners inside residential homes. The project helps companies gather detailed interaction data from real environments.
Developers believe these trials will improve future embodied AI systems.
China is also increasing infrastructure investment. Nearly 30 embodied AI training centres and data facilities are planned or already operating nationwide.
The information came from the Embodied AI Development Report 2025. The report was jointly released by the China Academy of Information and Communications Technology and Tsinghua University.
The race to dominate embodied AI is becoming highly competitive. Chinese firms want AI systems that can perform physical tasks without direct human control.
The technology still faces barriers, but companies continue expanding research, testing, and commercial deployment.
Industry leaders believe embodied AI could become one of the most important sectors in global artificial intelligence development.
Post a Comment