Why China’s humanoid robot industry is winning the early market
China’s humanoid robots captured worldwide attention with their kung fu flips during the nation’s televised Spring Festival Gala, while the Chinese phone manufacturer Honor is preparing to debut its inaugural humanoid robot at MWC in Spain. Robotics has been identified as a major focus within China’s “Made in China 2025” initiative, initially centered on factory automation rather than humanoid robotics. However, rapid advancements in multimodal AI are accelerating the development of embodied AI—autonomous machines capable of functioning in real-world environments—a strategy that officials believe could alleviate labor shortages and boost productivity significantly.
At this nascent stage of humanoid robot innovation, Chinese firms are outstripping their U.S. counterparts in both speed and volume, according to expertise from the office of Eric Schmidt. With a highly developed hardware supply chain, much of it established through the EV sector and encompassing components like sensors and batteries, combined with the world’s strongest manufacturing foundation, Chinese companies can iterate much faster than Western competitors. This results in robots that are not only more affordable but also updated and released more frequently. For instance, Unitree dispatched roughly 36 times more units last year compared to U.S. rivals Figure and Tesla.
Global shipments of humanoid robots stood at a modest 13,317 units last year, marking a small base given the industry’s expectation to nearly double shipments annually, potentially reaching 2.6 million by 2035. However, these numbers require cautious interpretation due to uncertainties about how many units represent actual commercial sales versus demonstration models or pilot projects, reflecting the early, evolving nature of the field. The leading shipment figures for 2025 are dominated by China’s Agibot and Unitree, followed by UBTech, Leju Robotics, Engine AI, and Fourier Intelligence, highlighting Beijing’s early sectoral dominance.
A notable recent transition has been a move away from “demo-driven excitement” towards “operations-driven adoption,” as explained by Galbot’s chief strategy officer. Galbot’s G1 humanoid robot performed at this year’s Spring Festival Gala alongside robots from Unitree Robotics, Noetix, and MagicLab. The growing interest revolves around whether these robots can operate reliably in real-world settings and shoulder human workloads—an interest reinforced in China by policies advocating automation upgrades and a manufacturing ecosystem that facilitates rapid iteration. While increased investment in humanoid startups has accelerated progress, enduring adoption depends on demonstrating consistent and reliable value in practical production or service environments instead of one-off showcases.
Investment continues to flow into China’s robotics sector. Unitree, valued at approximately $3 billion after its Series C round, targets up to a $7 billion valuation for a prospective IPO. Similarly, Galbot has raised over $300 million recently, reaching an estimated $3 billion valuation, marking one of the most substantial financings in China’s humanoid robotics scene. While U.S. companies also focus on moving beyond flashy demonstrations toward grounded deployments, and with some aiming to produce tens of thousands of humanoid robots in the near term, China is simultaneously developing a range of robots from affordable mass-market versions to advanced applications, applying humanoids across industrial, consumer, and rehabilitation sectors.
Still, significant challenges remain for China’s leadership in humanoid robots, particularly regarding AI systems and integrated software. The field largely hinges on vision-language-action models and “world models,” both of which are still in early development phases. Nvidia currently leads in comprehensive humanoid software stacks, and most Chinese humanoid startups use Nvidia’s Orin chips, though domestic chipmakers are working on indigenous alternatives. A core issue is enabling robot foundation models to predict forthcoming physical states in unpredictable environments, a task more complex than predicting words with language models. Due to limited training data, companies primarily depend on simulated environments to generate synthetic data, but real-world data remain critical. Consequently, while hardware has advanced to provide greater robot dexterity, software—essentially the robot’s “brain”—is still in its infancy.
Safety concerns also pose a major challenge as public acceptance could be jeopardized by any serious accidents involving humanoid robots, prompting cautious and gradual deployment strategies in China. Regulatory oversight is expected to increase alongside industry maturity. Given data limitations, demand for humanoids is projected to grow first in controlled workplaces such as industrial manufacturing, warehousing, and retail, where tasks are repetitive, shifts are long, and processes clearly defined—an ideal environment for robots to demonstrate scalable value.
Humanoid robotics development outside China involves other Asia-Pacific players as well. Japan’s ecosystem, with contributions from startups and semiconductor manufacturers, aims for humanoid mass production by 2027. Long known for pioneering projects like Honda’s Asimo and SoftBank’s Pepper, Japan is recognized for precision and advanced control, particularly expanding the use of humanoid robots in eldercare. According to Coral Capital’s CEO, Tokyo’s continued success is supported by labor shortages, cultural affinity for robots as companions rather than threats, and dominance in key parts of the robotics supply chain. Meanwhile, Hyundai Motor’s Boston Dynamics division has introduced a new Atlas humanoid designed for factory use by 2028, planning production of up to 30,000 units annually in the U.S.
Nevertheless, China’s rapid advancement in humanoid robotics is driven by a convergence of supportive government policy, industrial strategies, labor shortages, and private capital influx. China’s distinctive speed-to-scale advantage compresses the entire innovation and deployment cycle—including R&D, supply chain, manufacturing, integration, and customer adoption—into a compact and continuous loop. This allows Chinese humanoid companies to move swiftly from prototype phases to real-world operations, iterating quickly based on practical feedback, a pace that remains difficult for other countries to match.
