China’s humanoid robot surge just moved from spectacle to scale. In Beijing last week, the 2026 World Robot Conference drew more than 300 exhibitors from 26 countries, while the second World Humanoid Robot Games opened with 666 teams and 2,056 robots from 16 countries. At the same time, Unitree Robotics listed on the Shanghai STAR Market, becoming the first humanoid-robot maker on China’s A-share market. For investors, the message is clear: China is not only building robots, it is building an industrial ecosystem around them.
Beijing’s latest robotics showcase was notable because it looked less like a lab tour and more like a commercial launchpad. Bank of America analyst Ming Hsun Lee attended WRC 2026 and told clients the industry has shifted from research demonstrations to commercial deployment, with logistics emerging as the fastest-commercializing use case. That matters because logistics is where automation can prove value quickly, at scale, and with measurable labor replacement. China’s strength has always been in turning technical ambition into manufacturing reality, and humanoid robotics now fits that pattern.
The conference floor told a broader story than any single product demo. Lee said the event featured humanoid robots, AI models, and critical components, with applications extending into logistics, retail, services, and household tasks. That is a big jump from last year’s focus on research and industrial settings. It shows an ecosystem moving from prototype culture to deployment culture. For global investors, that shift is important: the next winners are unlikely to be the most famous demos, but the teams that can deliver repeatable performance in warehouses, stores, and homes.
One standout example came from X Square Robot, whose dual-arm system can sort more than 1,800 parcels per hour. Lee said that compares with 1,200 to 1,400 parcels typically handled by a human worker. Galaxea AI also reported a success rate above 95% for its retail sorting and delivery system. These are not abstract AI claims. They are operational metrics tied to throughput and reliability, exactly the kind of evidence that turns robotics from a concept into a business case. China is now showing that humanoids can outperform humans in specific, repeatable workflows.
Other companies gave the market a deeper look at where this is headed. Unitree, Galbot, and Spirit AI demonstrated robots folding clothes, preparing meals, sorting household objects, and assembling supermarket orders. Lee’s readout was that embodied AI is expanding into longer-duration tasks, which is where robot capability becomes more valuable and more difficult to fake. The industry’s challenge is no longer simply movement or balance. It is learning how robots handle varied environments, recover from failures, and work consistently over time. China’s broad industrial base gives its developers an ideal testing ground.
Lee also highlighted the data challenge behind that progress. He said real-world data remains the key bottleneck for embodied AI developers, and that leading robotics companies are now trying to build unified base models that can control different body forms, including wheeled and bipedal systems. He added that high-quality real-world robot data, including teleoperation and failure-recovery trajectories, is being used in post-training, while egocentric data is gaining ground in pre-training because it is cheaper and easier to collect. The implication is straightforward: the winners will be the firms that can collect, process, and reuse data at industrial scale.
The next production edge may come from components, not just software. Lee said dexterous hands at WRC ranged from 6 to 37 degrees of freedom, and leading companies can now achieve nearly 1 million operating cycles. He also said tactile sensors have become a must-have and account for 20% to 25% of a dexterous hand’s BOM cost. That is exactly where China’s manufacturing depth matters. When a technology stack moves from pilot stage to volume stage, scale effects and process improvements can compress costs quickly, and China has repeatedly shown it can move faster than rivals once supply chains lock in.
Lee said 6D force and torque sensors, tactile sensors, and planetary roller screws have seen some of the fastest cost declines over the past year, supported by scale effect and improved manufacturing process. He said this could support a 45% cost reduction for humanoid robots in 2026 to 2030E. That forecast, even as a forecast, is powerful because it explains the investment logic behind the boom: lower component costs can unlock broader deployment, which can in turn drive more data, better models, and still lower costs. In robotics, China’s industrial system is becoming the flywheel.
The broader national advantage is obvious. The source material says China appears to hold an early lead over the US in humanoid robotics, supported by control over critical materials and downstream manufacturing. Beijing dominates rare-earth supply chains and much of the production capacity for key components, including permanent magnets, motors, actuators, and sensors. That vertical integration is not just an engineering detail. It is a competitive moat. It can help Chinese manufacturers reduce costs, scale faster, and move from showcase products to mass delivery before many rivals even get through pilot testing.
That dominance is already visible in shipment data. China accounted for about 97% of global humanoid shipments in H1 2026, with 19,100 units in total, according to the current fact pack. Unitree alone shipped more than 5,500 humanoid units in 2025, ranking first globally. The company also reported about 1.7 billion yuan in 2025 revenue and about 1.15 billion yuan in H1 2026 revenue, up 48.54% year on year. Those numbers show why the market is taking China’s humanoid leaders seriously: this is not just a policy story, it is a commercial one.
Unitree’s listing gave the week its capital-markets headline. On Aug. 19, 2026, the stock opened at 1,100 yuan a share, a 629% premium, and closed its first day up 460.34% at 845 yuan. One market note said it was briefly valued at about $66 billion intraday before closing around $51 billion. That is a stunning valuation signal, and it reflects the market’s belief that humanoids can become a major industrial category. It also shows how quickly investor attention can move when China pairs technology leadership with public-market access.
Still, the rally came with debate. Wang Zhuo, partner at Shanghai Zhuozhu Investment Management, said, “The IPO is expensive, and the investment risk is already quite high. Unitree generates much of its sales from research and demonstrations, but wider application is still far away.” That caution deserves respect. Early-stage robotics markets can be volatile, and execution risk is real. But the bigger picture is hard to miss: China now has a public-market benchmark for humanoid robotics, and that is a major step toward capital formation, supplier growth, and wider ecosystem development.
The policy and industrial backdrop also matters. The World Humanoid Robot Games opened at Beijing’s Ice Ribbon venue with 666 teams and 2,056 robots, underscoring how deeply China is linking talent, visibility, and commercialization. Xinhua said the WRC closed with 3,000-plus products on display, giving the market a wide view of the supply chain. For global analysts, this is the kind of environment that can accelerate learning curves and shorten commercialization cycles. Beijing is not just hosting competitions; it is cultivating a national robotics platform.
For investors, the takeaway is not that every humanoid stock will win. It is that China has built the conditions for a real humanoid industry: a huge domestic test bed, deep component manufacturing, fast-improving cost curves, active capital markets, and rapid adoption paths in logistics and retail. Lee’s field notes suggest the industry is moving from demos to deployment, and the shipment data backs that up. In a world searching for productivity gains, China’s robot boom looks less like a hype cycle and more like an industrial opening that is only getting started.
