China, Japan, Nvidia drive humanoid robot surge

By early 2026, China has firmly established itself as the global frontrunner in humanoid robotics commercialization, driven by companies like UBTECH, Unitree, and LimX Dynamics. UBTECH marked a significant milestone by producing 1,000 units of its Walker S2 humanoid robot with ambitions to scale to 10,000 units within the year, while Unitree shipped over 5,500 humanoids in 2025 alone, generating more than $100 million in revenue and becoming the world's leading manufacturer by volume. LimX Dynamics, backed by major investors including Alibaba and Lenovo, rapidly expanded its manufacturing capabilities and set sights on international markets such as the Middle East and the U.S., showcasing China's early success in both domestic deployment and global outreach.

China's dominance in the humanoid robotics sector is underpinned by a robust manufacturing ecosystem and unparalleled supply chain advantages, largely inherited from its expansive electric vehicle industry. This strong hardware foundation, encompassing sensors and batteries, enables Chinese companies to iterate and scale production far faster than their Western counterparts. As analyst Selina Xu observes, this manufacturing prowess, combined with government initiatives like 'Made in China 2025' pivoting from factory automation toward embodied AI, has accelerated China's lead in deploying nearly 85% of the world's humanoid robots and more than half of industrial automation units in 2025.

The early humanoid robotics market is dominated by Chinese firms, with Agibot, Unitree, UBTECH, Leju Robotics, Engine AI, and Fourier Intelligence leading shipment volumes and setting the pace ahead of U.S. competitors such as Tesla and Figure AI. In 2025, Chinese companies accounted for the majority of the approximately 13,000 humanoid robots shipped globally, with Unitree alone shipping roughly 36 times more units than its American rivals. This surge reflects not only manufacturing scale but also strong government and investor support, which has fueled rapid commercialization and technological innovation, including LimX Dynamics’ development of agentic AI systems to enhance autonomous robot capabilities.

In early 2026, Japan galvanized its industrial giants—SoftBank, Sony, Honda, NEC, and others—into the Japan AI Foundation Model Development consortium, backed by a staggering ¥1 trillion ($6.3 billion) over five years, to pioneer a trillion-parameter AI model tailored specifically for Physical AI applications in robotics and machine control. This initiative is a cornerstone of Japan's broader ambition to reclaim global leadership in AI and robotics, with parallel efforts like Noetra Corp aiming to deploy 10 million AI-enabled robots by 2040, supported by ¥380 billion ($2.33 billion) and a coalition of 44 companies, underscoring a national-scale commitment to technological sovereignty and ecosystem integration.

Central to Japan’s strategy is the FRONTia Project, awarded to Noetra Corp in mid-2026, which aspires to build the world's first national AI infrastructure dedicated to physical AI, backed by an estimated $6.1 billion investment through 2030. This project leverages Japan’s manufacturing prowess and global partnerships, notably with Nvidia providing critical hardware such as 13,750 Vera CPUs and 27,500 Rubin GPUs, to develop advanced multimodal foundation models that will power robotics across industry and healthcare, thereby fostering a sovereign AI ecosystem that addresses both domestic needs and global social challenges.

Noetra’s CEO, Hironobu Tamba, candidly describes the initiative as Japan’s 'last chance' to compete in the AI race dominated by China and the U.S., emphasizing the strategic imperative of developing a reliable, secure foundational AI model that enables Japanese companies to retain control over their data and reduce dependence on foreign cloud infrastructure. This aligns with government policies under Prime Minister Sanae Takaichi’s administration, which simultaneously supports ventures like Rapidus for cutting-edge chip manufacturing and sets ambitious targets to deploy 10 million AI-enabled robots by 2040 across sectors including manufacturing, shipbuilding, and nursing, reflecting a cohesive national vision for technological self-sufficiency.

By mid-2026, Noetra had launched full-scale R&D efforts under METI’s FRONTia umbrella to develop a Japan-based multimodal foundation model for physical AI, aiming for phased releases beginning in fiscal 2026 and culminating in the ambitious 'Real-world Native AI' by 2030. This model will feature advanced reasoning, generative capabilities, and deep Japanese language understanding, supported by collaborations with research institutions like AIST and international partners, and powered by substantial infrastructure investments including Nvidia GPU systems. This public-private partnership exemplifies Japan’s strategic push to integrate trustworthy AI seamlessly into real-world environments, reinforcing its technological sovereignty and ecosystem integration goals.

Nvidia has strategically positioned itself as a foundational AI platform provider in the global robotics ecosystem by forging deep partnerships with key Japanese industrial players and government initiatives. Central to this effort is the FRONTia Project, a ¥1 trillion ($6.1 billion) national AI infrastructure led by Noetra Corp—backed by NEC, Sony, Honda, and SoftBank—where Nvidia supplies thousands of Vera CPUs and Rubin GPUs to power advanced physical AI applications. This collaboration, endorsed by Japan’s Ministry of Economy, Trade and Industry, aims to transition robots from laboratory experiments into real factory environments, marking a pivotal shift in Japan’s industrial AI capabilities.

Expanding beyond Japan, Nvidia has cemented a multi-faceted partnership with South Korea’s LG Group to co-develop next-generation physical AI humanoid robots, leveraging its Jetson Thor computing platform and Isaac GR00T foundation model. LG’s unique advantage lies in its vertically integrated group structure—spanning electronics, sensing components, batteries, and industrial software—which allows it to internally build robot bodies rather than relying on external hardware. This collaboration includes the creation of a sprawling 10,000-square-meter data factory in Seoul to collect and synthesize real-world and simulated robot motion data, accelerating robotics development through a comprehensive ecosystem that integrates Nvidia’s simulation, reasoning, and AI factory infrastructure.

Nvidia’s robotics platform strategy mirrors its dominant data center chip business by positioning itself as the indispensable 'shovel seller' in the robotics gold rush, supplying advanced, compact, and low-power Jetson Thor chips that serve as the cognitive core for humanoid and warehouse robots. By embedding its technology across a broad spectrum of manufacturers—including Fanuc, Yaskawa, Kawasaki Heavy Industries, and Fujitsu—Nvidia ensures its components become the essential building blocks for diverse robotics applications worldwide. While robotics currently represents a modest slice of Nvidia’s vast business, this foundational role in physical AI positions the company for significant growth as the ecosystem matures.

Beyond hardware, Nvidia is deepening its robotics ecosystem through full-stack software platforms like Isaac for simulation, Cosmos for synthetic data, GR00T for robot reasoning, and DSX for AI factory infrastructure, creating high switching costs and customer reliance. Collaborations with LG and startups like Foxglove demonstrate how Nvidia’s integrated approach—combining real-world data with synthetic training and agentic AI tools—accelerates physical AI development and deployment. This strategic expansion from traditional data center compute into physical automation workloads signals Nvidia’s foundational role in the emerging global physical AI ecosystem, with early commercial traction expected in the coming quarters.

In mid-2026, NEURA Robotics emerged as a dominant force in the physical AI robotics ecosystem by securing a monumental $1.4 billion Series C funding round led by Tether, with heavyweight participation from Amazon, Nvidia, Qualcomm, and the European Investment Bank, valuing the company at around $7 billion. This capital infusion is strategically aimed at accelerating the mass production and global deployment of cognitive humanoid robots, targeting millions of units by 2030. NEURA’s innovative integration of Tether’s crypto wallet and edge AI software into their robots not only enables autonomous financial transactions but also enhances on-device AI processing across sectors such as manufacturing, logistics, and healthcare, positioning the company at the forefront of the convergence between AI, robotics, and crypto infrastructure.

Delta Intelligence has demonstrated remarkable momentum by raising approximately $74 million in its sixth funding round within just six months, underscoring robust investor confidence in its native 3D full-body robot intelligence. Founder Ma Xiaojian emphasized the critical challenge of achieving full-body coordination for complex tasks like climbing ladders and opening heavy doors, which their proprietary 3D world engine and panoramic data collection device are uniquely designed to address. This approach transcends traditional 2D-based models by enabling robots to better understand and interact with complex environments, signaling a significant leap forward in embodied AI capabilities for industrial applications.

Strategic collaborations are rapidly shaping the global robotics landscape, exemplified by RLWRLD’s August 2026 partnerships with South Korean startups PhysicsSimLab, Space AI, and NdotLight to enhance the DexBench humanoid dexterity evaluation benchmark developed with Nvidia. By integrating advanced 3D simulation assets, high-precision digital twins, and specialized environments for handling delicate objects, this alliance aims to push the boundaries of robot dexterity crucial for manufacturing and logistics. RLWRLD’s robotics foundation model RLDX-1, recognized as a World Economic Forum 2026 Technology Pioneer, reflects the growing importance of human-level dexterity supported by major industry players like Lotte, SK Telecom, and CJ Logistics.

The global surge in robotics venture funding, which hit a record $55.8 billion in 2026, highlights a vibrant ecosystem fueled by both corporate giants and nimble startups across continents. Chinese firm Dexmal is targeting a $3 billion valuation backed by Alibaba and supported by China’s expansive government investment exceeding $26 billion, reflecting the country’s ambition to dominate embodied AI with over 140 humanoid robotics companies. Meanwhile, NEC’s strategic investment in U.S.-based Dexmate, creator of the humanoid robot 'VEGA' and an integrated Physical AI development platform, marks a deliberate expansion from digital AI into Physical AI robotics. NEC’s Corporate SVP Shigeki Wada underscored this move as a commitment to open innovation and cross-sector collaboration to accelerate Physical AI adoption globally, illustrating the increasingly interconnected nature of the robotics funding and partnership landscape.

By early 2026, robotics technology had made remarkable strides in agility and dexterity, exemplified by Unitree Robotics' G1 humanoid enduring 130,000 steps in subzero temperatures and ChangingTek Robotics' anatomically inspired hand capable of handling everything from tools to delicate fruits. These advances highlight a clear transition from theoretical research to robust, practical designs that can operate reliably in extreme and varied environments.

The World Humanoid Robot Games in mid-2026 dramatically showcased how far humanoid robots have evolved, with feats like the 'Superman' robot sprinting faster than Usain Bolt and Unitree’s robot leaping two meters from a standstill. This leap in dynamic physical capability—from last year’s 'janky' prototypes to machines executing complex acrobatics—signals a pivotal shift toward real-world applications where speed, agility, and precision are paramount.

Concurrently, the deployment of humanoid robots in military contexts, such as Foundation Robotics’ Phantom MK-1 units operating on Ukraine’s front lines with capabilities to handle weapons like pistols and M16 rifles, underscores a new operational phase beyond experimentation. While debates continue about the broader societal roles of such agile robots—ranging from search and rescue to security—the financial and strategic incentives clearly favor militarized uses, reflecting a pragmatic commercialization trajectory.

Beyond industrial and military spheres, robotics innovation is branching into creative and niche applications, as seen with CASBOT’s humanoid robot rock band and the autonomous drone-umbrella project Flying Umbrella 2.0. These ventures illustrate the expanding ecosystem of physical AI robotics, where technological advances in autonomy and dexterity enable diverse, unexpected real-world uses that enrich human experiences and hint at future commercial opportunities.