With the rapid advancement of artificial intelligence technology, the wave of mass production for humanoid robots is surging forward. Numerous technology companies and innovative teams are intensifying their investments, striving to transition humanoid robots from laboratory settings to the commercial market. Breakthroughs in core component localization, enhancements in whole-machine manufacturing technology, and the expansion of application scenarios have collectively established a solid foundation for the large-scale production of humanoid robots. The industry widely regards 2025 as the inaugural year for mass production of humanoid robots. To swiftly realize this consensus, various enterprises specializing in humanoid robots are tacitly accelerating their mass production and commercialization processes. Much like the poetic imagery of “a sudden spring breeze bringing thousands of pear trees into bloom,” it seems that overnight, the gates to mass production of humanoid robots have swung open.
-
Commercial Launch and Market Expansion
Recently, Fourier Intelligence announced the pre-sale of its first “Carebot” humanoid robot, the GR-3, equipped with a full sensory interaction system. This product is scheduled for market release and delivery in October 2025. According to incomplete statistics, multiple companies, including Qingxin Yichuang, Lingbao CASBOT, Songyan Power, Zhongqing Robotics, Xinghai Tu, and Unitree Technology, have begun selling their humanoid robots on online platforms such as JD.com, with prices ranging from tens of thousands to 500,000 yuan. Beyond direct sales, in July 2025, Zhiyuan Robot and Unitree Technology were awarded a procurement project for humanoid bipedal robot manufacturing services by China Mobile (Hangzhou) Information Technology Co., Ltd., spanning from 2025 to 2027, with a total budget of 124.05 million yuan. In September, following a 90.5115 million yuan equipment procurement project from MiYi (Shanghai) Automotive Technology Co., Ltd., Ubtech secured another procurement contract worth 250 million yuan from a renowned domestic enterprise for embodied intelligent humanoid robot products and solutions. This contract represents the largest globally for humanoid robots to date. Public data indicates that Ubtech’s Walker series of humanoid robots have now amassed nearly 400 million yuan in contracts.
-
Market Size Projections and Growth Trends
High-tech Robot Industry Research Institute (GGII) forecasts that the global market size for humanoid robots will reach approximately 1.017 billion USD in 2024, expanding to 15 billion USD by 2030. The compound annual growth rate from 2024 to 2030 is expected to exceed 56%, with global sales of humanoid robots rising from 11,900 units to 605,700 units. Specifically, China’s market size is projected to approach 38 billion yuan by 2030, with sales increasing from 4,000 units to 271,200 units. This growth underscores the accelerating adoption and production of humanoid robots worldwide.
Region 2024 Market Size 2030 Market Size Sales Volume (2024) Sales Volume (2030) CAGR (2024-2030) Global 1.017 billion USD 15 billion USD 11,900 units 605,700 units Over 56% China Not specified 38 billion yuan 4,000 units 271,200 units Not specified -
Cost Reduction and Localization Advances
At a recent press conference held by the State Council Information Office on the high-quality completion of the “14th Five-Year Plan,” officials from the Ministry of Industry and Information Technology stated that China has achieved full industrial chain manufacturing capabilities for humanoid robots, spanning from key chips and components to complete machines. In 2024, the localization rate of core components for domestic humanoid robots exceeded 70%, with the autonomous supply chain system significantly enhancing cost stability for large-scale production. The rising localization rate of core components, evolution in modular production technology, optimization of software and algorithms, and increased supply chain maturity have effectively controlled the production costs of humanoid robots. Previously, reliance on imports for precision reducers, servo motors, and force sensors accounted for over 50% of the total machine cost. Now, domestic manufacturers have reduced joint costs by 50% through material innovations, such as nanoscale carburized steel harmonic reducers, and process breakthroughs, while achieving precision levels approaching international top standards. For instance, Tiantai Robot’s new generation of integrated drive-control joint modules have seen prices drop to three-digit figures, with cost reductions exceeding 50% year-on-year.
Zhang Minliang, Hardware R&D Director at Kepler Humanoid Robot, remarked, “By innovating processes, we successfully shortened the grinding time for nuts from 20 hours to 2 hours. Additionally, we optimized the combination of 28 joint modules, reducing them from 14 specifications to 4 rotational and 4 linear specifications. This ‘volume aggregation for cost reduction’ strategy slashed the cost of the entire joint module suite by 50%.” A representative from Ubtech highlighted that the mass production and commercialization of humanoid robots result from synergistic advancements in technology, cost optimization, and policy support. Technologically, significant breakthroughs have been made in the coordination of the “brain,” “cerebellum,” and “limbs.” In hardware, the accelerated localization of core components continues to drive down costs. Regarding policy and ecosystem, a multi-tiered support system from central and local governments has been gradually perfected, providing a solid foundation for the R&D and industrial application of humanoid robots.
-
Persistent Technological Challenges
Despite the accelerated pace towards mass production, several hurdles remain. For example, Tesla’s Optimus project faces potential delays in its 2025 target of producing 5,000 units due to hardware issues such as overheating joint motors, insufficient load capacity of dexterous hands, and inadequate battery endurance. Furthermore, downstream OEM hardware solutions have not yet been finalized, and significant variations in customer demands lead to highly customized development models that constrain mass production efficiency. Wang Xingxing, founder of Unitree Technology, noted that while current robot hardware is largely sufficient, the primary challenge lies in the inadequate integration of AI models with hardware. For instance, using AI to control dexterous hands for more intricate tasks remains highly challenging. He also pointed out that both data and models are crucial for robot development, but current robot data collection suffers from issues like noise, quality, and ambiguous standards for high-quality data acquisition, data types, and scale. Additionally, multi-modal fusion is suboptimal, with significant difficulties in aligning robot movements with video and language models.
Mu Yao, Assistant Professor at the Artificial Intelligence Institute of Shanghai Jiao Tong University, stated, “Current humanoid robots primarily rely on visual and auditory solutions for interaction, but tactile perception, as a key sensory modality for physical contact, is essential for large-scale applications. Existing solutions struggle to replicate the fine operational perception capabilities of the human hand. The high cost of six-dimensional force sensors impedes scalable application, and the flexibility and durability of dexterous hands still require improvement.” Wang Zhenyu, Executive Deputy Director of the Comprehensive Research Support Center at the Chinese Academy of Sciences’ Institutes of Science and Development, observed that in related combat competitions, although robots can perform preset actions, they frequently exhibit issues such as “missed attacks” and “accidental rope touches,” reflecting deficiencies in environmental perception and real-time decision-making. Future technological breakthroughs may focus on developing universal spatial intelligence large models to enhance environmental understanding and autonomous decision-making, strengthening multi-modal perception fusion, and exploring new materials and high-power-density drive systems to balance performance and energy consumption.
Undeniably, humanoid robots still require further breakthroughs in core technologies, including mechanical structure, motion control, interactive operation, and intelligent decision-making. Simultaneously, the high cost of acquiring high-quality, labeled real-world multi-modal data restricts the generalization capabilities and application expansion of humanoid robots.
-
Application Scenarios and Implementation Consensus
The industry has seemingly reached a consensus on the application sequence for humanoid robots: “first industrial, then commercial services, and finally household.” This landing order results from comprehensive consideration of factors such as technological maturity, scenario complexity, and economic logic. The industrial sector, as the initial stage for commercial robot deployment, is prioritized due to its “high determinism” characteristics. Compared to industrial settings, service sectors require robots to possess preliminary environmental adaptability and human-machine interaction capabilities. In contrast, the diverse and personalized demands of household scenarios necessitate integrated functions like perception, decision-making, and interaction, making it the “ultimate scenario” for testing core robot technologies.
Currently, Ubtech has collaborated with renowned enterprises such as Dongfeng Liuzhou Auto, Geely Auto, FAW-Volkswagen Qingdao Branch, Audi FAW, BYD, BAIC New Energy, Foxconn, and SF Express, with its industrial humanoid robot Walker S series undergoing factory training. Galaxy General has achieved deployments in retail, factory sorting, and intelligent logistics scenarios, with future plans to explore consumer-facing applications. Throughout the development of humanoid robots, the reliability and safety of key technologies like motion control, endurance, and communication stability need to be enhanced through practical scenario training. At present, most companies’ humanoid robots remain at the L1 to L2 stages of the “Humanoid Robot Intelligence Grading Standards,” with a few leading firms exploring the transition from L2 to L3. A Ubtech representative emphasized that to achieve broader commercial scale, it is essential to further improve product reliability, durability, and complex environment adaptability, while continuously reducing costs and identifying essential scenarios.
-
Forming Closed-Loop Essential Scenarios
The key to commercializing humanoid robots lies in establishing closed-loop essential scenarios. On one hand, humanoid robots must identify clear and high-frequency essential scenarios across fields ranging from industrial manufacturing to healthcare, logistics, and home services. On the other hand, the connection between technology and application needs strengthening to ensure that humanoid robots can operate efficiently and stably in these scenarios, meeting genuine user needs and thereby forming a complete cycle from technology to application to service. As the Ubtech representative stated, when procuring humanoid robots, clients’ core consideration is not solely performance or price but rather stability, reliability, and overall return on investment in practical applications. Clients focus on whether the humanoid robots can effectively address specific business pain points—for example, capabilities like 7×24 continuous stable operation and autonomous battery swapping in industrial settings. Additionally, the company’s solid technical development strength, mature solutions, and reliable large-scale delivery assurance are critical decision factors.
Moreover, the development of humanoid robots requires concerted efforts from upstream and downstream enterprises, including component suppliers, system integrators, and application developers. Only by forming a complete industrial chain can the large-scale commercial application of humanoid robots be propelled. In this process, component enterprises, whole-machine manufacturers, and downstream scenario companies should foster an ecosystem characterized by “focused division of labor and collaborative攻坚.”
In summary, the mass production of humanoid robots marks a significant milestone, driven by technological progress, cost efficiencies, and strategic partnerships. However, overcoming remaining challenges in hardware, AI integration, and data quality is crucial for unlocking the full potential of humanoid robots across diverse sectors. The continued collaboration across the ecosystem will be instrumental in shaping the future landscape for humanoid robots, ensuring they evolve from novel innovations to integral components of modern industry and daily life.