China's embodied intelligence market is expected to exceed one trillion yuan in 2026, with the industry's hot and cold divergence gradually returning to rationality

The embodied intelligence industry, described as "reshaping the paradigm of human production and life and the global competitive landscape," is in 2026 undergoing accelerated deployment from the "laboratory" to the "training ground" and then to the "real-world setting," while also driving rapid growth in related industries.

Wang Xin, director of the Industrial Statistics Department of the National Bureau of Statistics, said while interpreting the industrial data for the first three quarters of this year that new economic growth points such as embodied intelligence and human-machine collaboration are helping the robotics sector develop rapidly, with output of products such as robot reducers and industrial robots growing by 44.4% and 29.1%, respectively.

According to data compiled by the China Commercial Industry Research Institute, in broad terms, China's embodied intelligence market grew from 213.3 billion yuan in 2018 to 915 billion yuan in 2025, an increase of more than threefold; in 2026 it is expected to exceed one trillion yuan.

"Traditional robotics and manufacturing industries are recovering on the back of this momentum, and old and new growth drivers are accelerating their integration." The "Embodied Intelligence Development Report (2026)" (hereinafter referred to as "the report"), recently released by the Institute of Artificial Intelligence of the China Academy of Information and Communications Technology together with the Department of Electronic Engineering of Tsinghua University and the Embodied Intelligence Testing Laboratory, shows that, on the one hand, the traditional robotics industry is directly benefiting from the incremental demand brought by embodied intelligence; on the other hand, established manufacturing enterprises are accelerating their entry through various means such as acquisitions, equity investment and building their own operations.

At a symposium on embodied intelligence chaired by Liu Liehong, director of the National Data Administration, in early September, representatives from enterprises including the Institute of Automation of the Chinese Academy of Sciences, the Beijing Institute for General Artificial Intelligence, Guanglun Intelligence, JD Group, DeepWise, the Shijingshan Humanoid Robot Data Training Center, Galaxea, Zibianliang and the Beijing Humanoid Robot Innovation Center attended.

"At present, the embodied intelligence industry is booming and shows data-driven characteristics, with sustained growth in demand for high-quality, diverse and large-scale data," the National Data Administration said. It will promote the development of embodied intelligence data standards in due course, guide local data systems to carry out related work in an orderly manner, promote the healthy development of the industry, and actively support enterprises in increasing data investment to better empower the development of embodied intelligence.

Based on statistical data analysis of the embodied intelligence industry map of the Artificial Intelligence Industry Development Alliance of China, the report shows that as of October 8, 2026, there were 405 embodied intelligence-related enterprises in China, of which 180, 257 and 212 were upstream, midstream and downstream enterprises, respectively, with 191 enterprises covering more than two links of the industrial chain, accounting for 47.2%.

In terms of the regional distribution of embodied intelligence enterprises, Beijing (109), Guangdong (84), Shanghai (58), Jiangsu (45) and Zhejiang (39) are the top five provinces by scale, with their combined number of enterprises accounting for 82.7%. "Beijing has an extensive layout in areas such as algorithm models, complete machines and data collection equipment, while Guangdong and Shanghai both have many enterprises participating in complete-machine R&D and basic supporting facilities," the report said in analyzing the layout characteristics of embodied intelligence enterprises in these five provinces, adding that Jiangsu and Zhejiang have formed a clustering trend in robot products, supporting supply and industry applications, and that "the 'head-concentrated, long-tail-distributed' pattern of regional industry distribution remains significant."

As the global investment and financing market remains active and enterprises around the world accelerate their layout, capital is also tilting toward leading projects and enterprises. From the structure of financed enterprises, the report says that foreign embodied intelligence financing is more focused on billion-dollar-level leading projects, while domestic embodied intelligence investment and financing is expanding rapidly, with capital gradually concentrating in leading and mid-to-late-stage projects.

As enthusiasm continues to rise, challenges in technology, applications and homogeneous competition are also gradually bringing the industry back to rationality.

At the 2026 Artificial Intelligence Industry Conference held recently, Wei Liang, vice president of the China Academy of Information and Communications Technology, said while interpreting "embodied intelligence real-world training," one of the ten key words of the AI industry, that as real-world training goes deeper and more substantive, industrial challenges are becoming increasingly clear. These are mainly reflected in the limited ability to complete complex operational tasks, with long-horizon tasks involving temporal and spatial reasoning difficult to handle; insufficient exploration of high-value vertical scenarios, with pilots mostly concentrated on tasks that are easy to demonstrate and standardize; and a lack of large-scale product delivery capability, with difficulty generalizing single-pilot capabilities to multiple scenarios.

"Over the past year, the iteration and achievements of embodied intelligence technology and industry have advanced rapidly, but there has also been a polarization between hot and cold." The report analyzes that, on the one hand, capital market and policy attention continues to heat up, with new achievements in areas such as models, bodies and data released intensively; on the other hand, large-scale deployment and a closed commercial loop still face multiple constraints, and there remains an obvious gap between technology demonstrations and real delivery.

For example, in terms of applications, the report says that real-world deployment is far smaller than shipments, and technological maturity cannot yet meet real demand.

According to relevant statistics, among publicly disclosed winning bids related to embodied intelligence and humanoid robots in China in the first half of 2026, educational and research institutions accounted for the highest share, more than half (55.5%), while industrial and technology enterprises accounted for less than half that of educational and research institutions (21.1%), and government and public platforms accounted for about 20.6%.

In actual use, the report says, a considerable proportion of equipment is used for showroom reception or event demonstrations, and outside performance schedules there is a large amount of idle time, with some equipment powered on for fewer than 10 days per month on average.

In terms of deployment costs, the report cites data showing that complete-machine prices still lack economic usability. The price of a complete machine accounts for 40% to 60% of total cost of ownership, and actual deployment also entails multiple expenditures such as scenario modification and system integration, data collection and model training, staffing for operations and maintenance, and losses from downtime. Taking the annual comprehensive operations and maintenance cost of a single wheeled-arm or bipedal humanoid robot as an example, a price of 500,000 to 700,000 yuan is far higher than the labor cost of most positions.

The report cautions that it is necessary to recognize that the development of embodied intelligence will not be completed suddenly with a single technological breakthrough, but is more likely to show a spiral rise and gradual evolution. "The future development of embodied intelligence requires rethinking the underlying architecture of artificial intelligence at the technical level, so that models, bodies, environments and actions form a closer dynamic coupling," the report says, adding that at the industrial level it will shift from "technology-driven" to "value-driven." From the perspective of future global development priorities, the near term is a critical period for commercial validation and capability closed loops; the next three to five years will enter a stage of scale expansion and ecosystem building; and in the long term, the development focus may further shift toward shaping general and global capabilities.

The outline of the 15th Five-Year Plan proposes encouraging technological innovation in areas such as multimodality, agents, embodied intelligence and swarm intelligence, and exploring a development path for general artificial intelligence.