US bans Chinese robots as global humanoid race heats up

By mid-2026, the US government, led by the FCC and a White House interagency body, imposed a targeted import ban on new Chinese-made humanoid robots and related technologies, citing national security and cybersecurity risks tied to potential surveillance and cyberattack capabilities embedded in devices from firms like Unitree and Agibot. This ban, which strategically exempts existing authorized models, underscores escalating geopolitical tensions as China condemned the move, warning of 'resolute countermeasures' and highlighting disruptions to global supply chains and bilateral trade relations.

China’s dominance in the global humanoid robotics market is stark, controlling approximately 85% of shipments with Unitree and Agibot alone delivering around 10,000 units in 2025, dwarfing US competitors like Tesla and Figure, which shipped only a few hundred units combined. This hardware supremacy is complemented by China’s aggressive financial momentum, exemplified by Unitree’s anticipated $619.4 million IPO, signaling a strategic push that challenges US leadership, which instead focuses on AI software and world-model startups to enhance robot utility.

The US-China robotics rivalry is intensifying as each nation leverages its unique strengths: China’s abundant energy resources and hardware innovation versus the US’s semiconductor leadership and advanced AI software development. This dynamic sets the stage for a protracted competition through 2027 and into the 2030s, with the humanoid robotics sector emerging as a critical battleground in the broader AI and technology rivalry between the two powers.

Unlike the backlash seen with restrictions on Chinese large language models, the import ban on Chinese humanoid robots has not provoked significant opposition from US companies reliant on these technologies, indicating a consensus to prioritize national security over commercial interests in this sector. This contrast highlights the unique geopolitical sensitivity surrounding physical AI devices integrated into critical American infrastructure, which the US government views as posing 'unacceptable risks' that justify stringent import controls.

By early 2026, AGIBOT had firmly transitioned from prototype to large-scale commercial deployment, achieving a milestone of 10,000 humanoid robots produced by March and surpassing(https://www.agibot.com/article/231/detail/85.html) by mid-year. Their unified “One Robotic Body with Three Intelligences” architecture integrates motion, interaction, and operational intelligence to reliably handle real-world constraints such as force and timing, enabling diverse applications across industrial handling, logistics, retail, and cleaning. This maturity is underscored by their hardware innovations like the A3 Ultra humanoid with 51 degrees of freedom and the OmniHand 3 Ultra-M’s advanced tactile sensing, reflecting a shift from capability demonstration to scalable, reliable manufacturing and ecosystem-driven deployment.

Apptronik exemplifies the fusion of advanced AI and humanoid hardware, leveraging NASA heritage and government-funded innovations such as liquid-cooled robotic actuators to enhance robot dexterity and endurance. Their Apollo humanoid robot, standing 5’8” and designed for human-centric environments, has proven its value in factories like Mercedes-Benz assembly lines and even space applications, performing tedious tasks to augment rather than replace human labor. The launch of Apollo 2 and expansion of their Robot Park training facility in Austin further highlight the critical role of integrated hardware and AI training environments in advancing practical humanoid deployments.

Industry leaders like Tesla and OpenAI are strategically positioning humanoid robots as essential tools for future infrastructure and manufacturing, with Tesla’s Optimus robot naturally extending its extensive automated factory expertise into humanoid robotics. OpenAI’s in-house robotics team, led by Aditya Ramesh, focuses on embodied AI to support skilled workers in constructing data centers and even orbital projects like Dyson swarms, reflecting a vision where robotic construction and custom chip development underpin future technological dominance. This approach underscores the enormous market potential of humanoid robots, anticipated to surpass the GPU market in scale and profitability once full-stack integration is achieved.

The integration of physical AI with humanoid robotics is poised to revolutionize manufacturing by enabling smart machines to autonomously perform complex, repetitive tasks in human-designed environments, thereby boosting productivity and addressing labor challenges. However, a critical hurdle remains in fully digitizing the physical world to capture sensory data essential for training these robots, a process currently underway through innovative methods like factory workers using goggles and gloves to generate real-world interaction datasets. Meanwhile, advances in AI-driven manipulation, exemplified by London-based Humanoid’s KinetIQ Ascend achieving near-human dexterity, and hardware breakthroughs such as AMD’s Ryzen AI Embedded X100 series, are accelerating the commercial viability of humanoid robots across industrial and consumer markets.

Labor shortages driven by aging populations and declining fertility rates are a foundational economic force propelling humanoid robotics adoption globally. With projections such as the UN’s estimate of 2.2 billion people aged 65 and above by the late 2070s and Europe facing a deficit of seven million workers by 2030, industries are urgently seeking robotic solutions to fill critical gaps, especially in healthcare and eldercare. This demographic pressure, combined with AI’s rapid advancement, creates a compelling case for humanoids to assume roles that human labor cannot sustainably meet.

The commercial viability of humanoid robots is rapidly improving due to significant cost reductions and innovative financing models. Morgan Stanley projects hardware costs dropping from $200,000 in 2024 to as low as $50,000 by 2050, while companies like Unitree already offer robots at $13,500. Meanwhile, robotics-as-a-service (RaaS) models are transforming capital expenditures into manageable operating expenses, exemplified by AGIBOT’s Sharebot rental service generating $1.5 billion in projected revenue by 2026. This shift enables businesses to integrate humanoids into existing human-centric environments without costly infrastructure redesigns, accelerating adoption across sectors.

Investment trends underscore robust financial backing and strategic supply-chain focus fueling the humanoid robotics boom. In 2025 alone, venture capital poured $21 billion into robotics, with landmark rounds like Neura Robotics’ $1.4 billion raise from Amazon, Nvidia, Bosch, and Qualcomm signaling strong market confidence. Morgan Stanley highlights that beyond robot manufacturers, foundational component suppliers—such as Baidu, NVIDIA, Samsung Electronics, AMD, and Texas Instruments—are poised to capture significant value, reflecting a broader ecosystem play as the sector undergoes a business model rerating over the next 3-5 years.

China’s aggressive $400 billion robotics investment and government-backed policies are reshaping the global humanoid robotics landscape by driving down costs and scaling production. XPENG’s dual-purpose manufacturing facility aims to produce 1,000 humanoids monthly by year-end 2026, leveraging shared supply chains with EV production to reduce labor costs by 35%. This strategy enables China to flood emerging markets like the Middle East and Southeast Asia with affordable, high-performance humanoids, compelling U.S. manufacturers to reconsider their industrial strategies amid intensifying international competition.

Humanoid robots are uniquely transforming manufacturing by enabling automation of complex, high-variability tasks in existing factory environments without costly infrastructure changes. Boston Dynamics’ Atlas robot, for example, leverages simulation-based reinforcement learning and egocentric video training to autonomously perform dexterous tasks directly from human demonstrations, allowing seamless integration into human-shaped factories. As Aya Durbin explains, these robots are designed not to replace all automation but to tackle tasks requiring flexible, adaptable solutions where traditional automation falls short.

The economic viability of humanoid robots hinges on their ability to match or surpass human labor costs and performance, as demonstrated by deployments like Aitu’s garment factory robots achieving a 97% success rate in fabric handling and paying for themselves within 18 months. Similarly, PIABOT Robotics’ G2 robot improved task efficiency by roughly one-third on automotive assembly lines, addressing acute labor shortages and high turnover in manufacturing. These successes underscore that even narrow task automation with humanoids can yield significant commercial and operational benefits.

Despite their promise, humanoid robots face persistent integration challenges, particularly in mastering reliable manipulation of diverse objects and synchronizing with complex industrial systems. China’s Ministry of Industry and Information Technology supports extensive real-world training programs to accelerate skill transfer, while companies like Siemens emphasize deep connectivity through digital twins, AI-enabled sensing, and fleet management to embed humanoids as collaborative tools rather than isolated machines. This holistic approach, including workforce training and safety culture adaptation, is critical for scaling humanoid deployment in brownfield factories.

The ongoing evolution of physical AI and digital environment digitization is rapidly advancing humanoid robots toward fully autonomous operation in dynamic manufacturing settings. As highlighted by experts, the challenge lies in digitizing tactile and visual data from scratch to train robots effectively, a process supported by workers wearing specialized gear to capture real-world interactions. While fully human-like factory robots are not yet a reality, the integration of AI decision-making with physical AI promises enormous productivity gains by taking on hard-to-staff, high-turnover jobs and enhancing overall factory efficiency.

China has emerged as the dominant force in humanoid robotics manufacturing, leveraging massive government-backed investments—such as its $400 billion push in 2026—and a sprawling ecosystem of 130-140 robotics companies with 30-40 IPOs expected in 2024. Firms like XPENG are pioneering dual-purpose factories targeting 1,000 humanoids per month, while companies like Leaderdrive and Unitree lead global shipments, with China accounting for 85% of units shipped in 2025. This manufacturing scale is complemented by pragmatic deployment strategies focusing on narrow, repeatable factory tasks, supported by extensive real-world training facilities that integrate embodied AI, enabling rapid skill transfer and operational stability on the factory floor.

The United States maintains a competitive edge primarily through leadership in AI software and world-model development, with startups focusing on the 'brain' of humanoid robots that enhance utility and adaptability. Companies like OpenAI are strategically building in-house robotics teams to advance infrastructure-supporting robots, reflecting a vision where humanoid robots play pivotal roles in constructing data centers and large-scale projects. This software supremacy is underpinned by American semiconductor dominance, which contrasts with China's manufacturing prowess, setting the stage for a complementary yet intense US-China rivalry fueled by massive venture capital and accelerated innovation timelines.

Japan and Europe offer distinct competitive advantages rooted in cultural acceptance and advanced integration of AI with manufacturing finesse. Japan benefits from a public notably less anxious about AI—only 28% compared to 50% in the US—and excels in producing durable, high-quality robotic components, with companies like Honda demonstrating technological prowess. Meanwhile, Europe, led by Germany, faces acute labor shortages projected at seven million workers by 2030, driving strategic investments from global giants such as Amazon, Nvidia, Bosch, and Qualcomm into startups like Neura Robotics. European firms emphasize AI-enabled sensing, digital twins, and fleet management to enable humanoid robots to collaborate adaptively within human environments, showcasing a focus on sophisticated software-hardware integration.

Across the globe, the humanoid robotics industry remains in an early but rapidly accelerating phase, with market forecasts projecting an explosive growth from 20,000 units shipped in 2025 to 10 million by 2035, representing an 86% CAGR. Investment strategies increasingly favor exposure to the broader supply chain—particularly foundational component suppliers like actuator and harmonic drive manufacturers—over betting on individual robot makers, many of which remain private. This ecosystem approach reflects the complex value distribution across hardware, software, manufacturing, and marketing, signaling that the humanoid robotics market could add several trillion dollars in market capitalization as robots become indispensable in sectors ranging from healthcare to law enforcement.

Humanoid robotics is poised for rapid integration into everyday environments within the next 3 to 5 years, beginning in predictable and structured settings such as factories and warehouses before expanding into logistics, healthcare, and eventually homes. Tesla exemplifies this phased adoption strategy, leveraging its evolution from a car manufacturer to a robotics powerhouse by embedding humanoid robots in its automated factories, underscoring the industrial sector as the critical launchpad for broader deployment.

The economic incentives driving humanoid robot adoption are staggering, with the market opportunity projected in the tens of trillions of dollars as robots replace a significant portion of human labor, which constitutes roughly half of global GDP. At an estimated $50,000 per unit, capable of performing tasks at a fraction of human labor costs, leading robotics firms stand to achieve multi-trillion-dollar valuations, marking physical AI and robotics as the next major investment frontier characterized by breakthrough technology and a rapidly expanding yet still nascent market.

Consumer adoption is rapidly gravitating towards rental models, as exemplified by AGIBOT’s Sharebot platform launched in late 2025, which offers humanoid robots for as low as $517 per day. This approach transforms costly capital expenditures into manageable operating expenses, broadening access and enabling use cases in controlled, short-duration scenarios like corporate events, retail promotions, and live-stream content creation where robots serve as 'cyber intelligent hosts' rather than complex industrial workers.

Each rental deployment serves a dual purpose by generating invaluable real-world operational data that accelerates robot learning and development, effectively bridging the physical AI data gap that simulations cannot address. As Li Yiyan, CEO of Sharebot, emphasizes, these rentals are pivotal in transitioning humanoid robots from mere exhibition pieces to scalable, large-scale applications, with companies investing over $100 million annually to harness such data for continuous improvement.