Plunging Auto Margins Drive China's EV Giants into Humanoid Robotics Race
Facing a severe contraction in manufacturing margins, China’s leading automakers are aggressively repurposing their autonomous driving architectures and factory floors to mass-produce humanoid robots.
The strategic pivot reached a critical juncture in September 2026, with at least 14 domestic manufacturers—ranging from EV startups to state-owned behemoths—deploying capital and production lines into the robotics sector. XPeng Inc. activated its dedicated robotics assembly line this month after securing a US$900 million funding round in August, pushing its robotics unit valuation past US$6.3 billion.
Investors are weighing this shift as traditional auto operations transition into a heavy-asset, low-yield trap. Official data for the first seven months of 2026 reveals that while Chinese auto manufacturing revenue edged up 2.7% to RMB 6.08 trillion (US$844.4 billion), total sector profits plummeted 20.4% to RMB 216.24 billion (US$30.03 billion), yielding a razor-thin 3.6% margin. Simultaneously, national auto sales contracted 3.7% to 17.6 million units, forcing automakers to seek new monetization avenues for their AI and hardware investments.
Transferring Autonomous AI Feeds Robotics Ambitions
Automakers are betting that the foundational technology of smart vehicles—perception, spatial understanding, and decision-making—translates directly to humanoid mechanics. Companies are utilizing existing supply chains, lidar sensors, and AI computing power to bypass the initial bottlenecks of robot development.
However, the transition exposes a critical data gap. While automakers possess petabytes of road-mapping and driving algorithms, humanoid robots require complex "manipulation data"—the tactile feedback and spatial physics required to grasp objects, adjust force, and navigate uneven factory floors. Millions of miles of driving data cannot automatically teach a robotic arm how to tighten a screw.
Diverging Commercialization Paths Fracture the Market
The sector is splitting into distinct operational models based on inherent corporate strengths. Software-driven entities like XPeng position themselves as physical AI companies, integrating vehicle and robot development on a shared AI foundation. The company targets mass production by late 2026 and commercial retail deployment by 2027.
Conversely, traditional manufacturing giants are utilizing their assembly lines as closed-loop data training grounds. SAIC Motor integrated its "Nengzai No. 1" humanoid into the Buick Electra E7 battery production line in March 2026, shifting from laboratory testing to real-world industrial application. Guangzhou Automobile spun off its robotics division into Weilun Technology in February, raising over RMB 100 million (US$13.8 million) in August to scale core components. Changan Automobile has set a definitive 2028 timeline for humanoid mass production.
Meanwhile, Chery Automobile is bypassing the general-purpose humanoid bottleneck by monetizing vertical applications. By July 2026, the company delivered over 2,000 units of its "Mo Jia" series—encompassing security, medical, and quadruped robots—across 60 countries, proving that immediate commercialization in specialized sectors is viable before achieving artificial general intelligence.
Lacking Manipulation Data Threatens Long-Term Viability
Despite China accounting for 84.7% of global humanoid robot shipments in 2025 with 14,400 units, industry engineers warn against equating automotive manufacturing prowess with robotics viability. The "last meter" of robotics—dynamic balance, joint articulation, and model generalization—cannot be solved purely through automotive platform standardization.
Firms lacking proprietary AI capabilities or closed-loop operational data risk turning their second growth curve into a capital-intensive research sink. The ultimate metric for this transition is not the speed of prototype unveiling, but the ability to integrate functional robotics into active industrial labor forces.