China's embodied AI robots move from training rooms to real-world jobs

In China, graduation ceremonies are no longer limited to human students. Robots are also completing training and moving into jobs.

Thirty robots formed the first class at a robot school in Hangzhou, east China's Zhejiang Province. After more than two months of training, four have graduated and taken up formal jobs.

One now works as a museum guide in Hangzhou, another leads visitors around the robot school, while a third inspects equipment at a Nestlé production workshop in Tianjin. The fourth has traveled to the Republic of Korea to perform on the piano.

Their progress shows one of the major challenges facing embodied artificial intelligence, or AI designed to interact with the physical world. Training robots for real-world tasks requires facilities, data and testing environments that individual companies or users may not be able to build on their own.

In Xiong'an, a planned city about 105 km southwest of Beijing, robots are learning everyday tasks including collecting garbage, arranging shelves and transporting goods at a specialized training center.

Human instructors use full-size models of workplaces and public areas to train the machines. Wearing virtual-reality headsets and exoskeletons, they control robots while the machines practice tasks in simulated settings.

Training a robot for even a simple task can take considerable time. Liu Rui, an instructor with a digital urban technology subsidiary of China Xiong'an Group, said more than 1,000 repetitions were needed to teach a robot the basic action of picking up an object.

The robot then had to learn the complete sequence of picking up garbage, standing upright and placing it in a bin. Mastering those three actions took another week.

Each practice session also creates valuable training data. Instructors label the information, divide movements into smaller steps and identify objects in individual frames. This allows robots to better understand their surroundings and determine what action to take.

China had completed or put into operation more than 70 robot training facilities by the end of June, while more than 40 others were being built or planned, according to the China Academy of Information and Communications Technology.

The growing network is becoming part of China's wider AI and robotics industry. More than 6,200 AI companies were operating in the country in 2025, with the core AI industry valued at more than 1.2 trillion yuan, or about $173.9 billion, according to the Ministry of Industry and Information Technology.

Official data also showed that China developed more than 400 complete humanoid robot products in 2025, accounting for more than half of the global total. Chinese manufacturers made up 97% of global humanoid robot shipments during the first half of 2026, according to a report presented at the 2026 World Robot Conference.

Jiang Lei, chief scientist at the National and Local Co-Built Humanoid Robotics Innovation Center, said embodied AI had progressed from "zero to one," but expanding it from "one to 10" remains difficult.

He said the main challenges include setting standards, managing engineering projects and carrying out pilot-scale testing. A major issue is making robots capable of adapting to different real-world situations.

China's large manufacturing sector gives robot developers both a major market and a wide variety of real and simulated environments where machines can be trained and tested.

The expansion of robot schools and training centers is also turning training data into an important economic resource for embodied AI.

Large language models and multimodal AI systems have developed rapidly by using huge amounts of written and online data. Embodied AI is different because robots must learn how to physically interact with the world.

A factory, warehouse or restaurant may require robots to use different levels of force, accuracy and movement. Collecting enough high-quality data for all these situations remains expensive and difficult.

Industry analysts say much of this information has never been systematically collected, processed or stored.

In 2025, when China Xiong'an Group worked to connect its digital platform more closely with the physical city, several leading AI companies approached the group to buy high-quality datasets containing tens of millions of hours of data.

Cao Yue, a manager at the group's digital urban technology subsidiary, said training datasets produced at the Xiong'an robot school are used to develop AI models and are also stored in an urban data space, where they can later be shared or traded.

In this way, the robot school also serves as a "megafactory" producing the data needed to train embodied AI.

At a national pilot base for embodied AI applications in Hangzhou, companies can use computing resources, data, AI models and testing environments through a public technical platform.

Zhang Haiwei, founder of a robotics technology company in Hangzhou, said companies previously spent tens of millions of yuan just to collect training data for a single real-world application.

By joining the shared facility, he said, startups can access national-level resources in one place and reduce their research and development costs during the early testing stage.

The approach is simple: rather than requiring every company to build its own training site, data system and testing environment, shared facilities can reduce costs and help transform laboratory prototypes into robots capable of working in real-world settings.

The Human Question

As more robots enter workplaces, another question is becoming increasingly important: What will happen to human workers?

Yu Jiang, a researcher at the Chinese Academy of Sciences, said technological changes can remove some jobs while creating new ones.

He compared the transition to the arrival of automobiles, when some horse-carriage drivers eventually became automobile drivers.
Xiong Rong, a professor at Zhejiang University, said robots are likely to take on more repetitive, physically demanding and dangerous jobs, allowing people to focus more on management, technology and innovation.

Earlier this month, China added 11 occupations to its list of officially recognized professions. The new jobs include embodied AI robot application technicians and AI agent developers.

Experts said workers will need lifelong learning and new vocational skills to adjust to a changing workplace, where human judgment, empathy and creativity work alongside the precision and endurance of robots.