By Yuan Jingming
On a production line in Zhangjiang’s “Robot Valley”, a new kind of industrial race is under way. Every 70 seconds, a finished joint module—a sophisticated assembly of motors, reducers, and encoders—rolls off the belt. After undergoing rigorous aging and load tests, these sleek black units are packaged and rushed via high-speed logistics to assembly plants in Shanghai, Beijing, and Shenzhen.
They leave the factory as components but on arrival they become the starting point of a robotic body.
The spotlight of the robotics industry often shines on the flashy demonstrations of Original Equipment Manufacturers. Recently, Tesla’s Elon Musk teased the summer production of the Optimus Gen 3, while Figure AI showcased its robots performing autonomous household chores. Yet, behind these cinematic displays, a silent and fierce battle is raging upstream in the supply chain. In the realm of joint modules — often called the “motion hub” of robotics — a high-stakes battle centered on timing, yield rates, and massive scalability has entered fever pitch.
Shanghai-based Eyou Robot Technology, which supplies robot makers including AgiBot, expects shipments of its joints to exceed 400,000 units in 2026, according to founder Sun Zeju. For a sector that was essentially a barren wilderness when the company was founded in 2018, this represents a staggering leap.
Fortune favors the early starters
That early start may now prove decisive. “Companies founded after 2023 will struggle to achieve mass production,” Sun said, reflecting a broader reality that China’s humanoid robotics sector is entering a new phase. Analysts expect meaningful order volumes to materialise in 2026, with leading manufacturers scaling shipments rapidly. Yet key components still lag automotive-grade reliability, and while prices have fallen under intense capital investment, consistency and durability remain challenges.
A common misjudgment in the current boom, industry insiders say, is to mistake favorable timing for insight. Humanoid robots, despite their human-like form, are first and foremost machines — industrial products that must withstand repeated physical stress with precision.
Following the unveiling of Tesla’s Optimus prototype in 2022, a wave of start-ups emerged, many relying on outsourced manufacturing and rapid prototyping. But as orders begin to arrive, some are discovering that presentation slides cannot easily be converted into reliable hardware.
The integration imperative
Eyou’s strategy was shaped earlier. In 2018, Sun identified a fragmentation problem in industrial automation: motors, reducers and other components were sourced separately and assembled piecemeal, resulting in bulky designs and inconsistent performance.
Drawing an analogy with the evolution of electronics from discrete components to integrated circuits, Sun bet that robot joints would follow a similar path towards integration. Between 2018 and 2021, the company focused on developing integrated joint systems — a period that, in hindsight, built crucial technical know-how.
That experience now underpins its manufacturing ambitions. While many competitors accept defect rates of around 0.1%, Eyou is targeting 100 parts per million — or 0.01% — a threshold it argues is essential for large-scale deployment.
The manufacturing bottleneck
The stakes are high. Joints account for nearly half the cost of a humanoid robot, and their reliability directly determines whether machines can walk steadily or perform tasks without vibration. At low volumes, manual assembly can suffice. At scale, it becomes a bottleneck.
In January 2026, Eyou launched what it describes as the world’s first automated production line for robot joints. Automation, Sun said, is the only way to reduce defect rates beyond the limits of manual assembly.
The benefits are not yet fully reflected in costs, given high material and R&D expenses. But as prices fall across the industry, labour costs will loom larger, giving early adopters of automation a structural advantage.
At the same time, pricing pressure is spreading through the supply chain as robot makers push for more affordable products.
Consumer demand reshapes the market
Industry participants describe the sector’s development in phases: 2024 as the first year of mass production, 2025 as the start of commercialisation, and 2026 as the beginning of large-scale deployment.
One unexpected catalyst has been the consumer market. While many expected industrial use cases to dominate early adoption, smaller humanoid robots — typically 0.8 to 1.2 metres tall — are already appearing in roles such as reception, companionship and education.
Consumer demand is highly price-sensitive but potentially vast, making it a key driver of scale. For component suppliers, this translates into demand for millions of joints annually.
Eyou aims to surpass one million units in annual shipments within the next year — a milestone that would signal not just sales growth but the maturation of its supply chain and manufacturing systems.
Betting on uncertain technology paths
Yet uncertainty remains over the industry’s technological direction. Different applications favour different joint types — from high-precision harmonic drives in industrial settings to lower-cost planetary systems for consumer robots.
Rather than betting on a single path, Eyou is developing multiple product lines to cover a range of potential outcomes. “We don’t know which technology will ultimately dominate,” Sun said. “But all humanoid robots will need joints.”
Even as attention remains fixed on the polished demonstrations of robot makers, the decisive contest is unfolding deeper in the supply chain — where success depends less on spectacle than on who started early enough to master the complexities of manufacturing at scale.
Source:
Gasgoo Embodied AI
