It’s not every day that a government warns its own tech sector to slow down, but that’s exactly what happened when Li Chao of China’s National Development and Reform Commission cautioned that the country’s humanoid robotics industry may be inflating a dangerous bubble. With more than 150 companies racing to produce bipedal machines, the market is already awash in “highly similar” designs, raising the risk of overcapacity, wasted capital, and stunted innovation.
The warning comes at a time when China’s humanoid sector is still wrestling with some of the hardest engineering problems in robotics. The challenge of developing super-dexterous hands that can effectively grasp and manipulate non-uniform shapes, or developing well-functioning,whole‑body control systems for a robot that is able to dynamically balance itself while being loaded or twisted and when navigating through cluttered spaces, is still an area of research for roboticists. These advancements are not purely cosmetic but will eventually be necessary for robots to operate safely in factory, warehouse, or residential environments where they could potentially interact with people.
The stakes are high because the sector is riding a wave of capital and policy support. Over the past year, Beijing has poured more than $20 billion into humanoid robotics, with a 1 trillion yuan ($137 billion) national fund in the works to back AI and robotics startups. Local governments are offering subsidies, free lab space, and even dedicated data‑collection facilities where fleets of robots are teleoperated for up to 17 hours a day to generate the task‑specific datasets needed to train embodied AI models. This focus on data is critical because unlike generative AI, which can be trained on vast internet corpora, humanoid control systems require high‑fidelity, physical‑interaction data stacking boxes, pouring liquids, fastening bolts collected in controlled environments.
China’s hardware dominance amplifies its speed advantage. Analysts estimate the country can produce up to 90% of humanoid components, from actuators and reducers to batteries and sensors, enabling rapid iteration and cost compression. Bank of America Securities projects that the bill of materials for a humanoid could fall from $35,000 today to $17,000 by 2030 if sourced domestically, with some startups already offering models for as little as $12,000. By comparison, a Tesla Optimus built entirely from non‑Chinese parts still carries a component cost of $50,000–$60,000.
But the very factors driving China’s surge cheap, fast production and a flood of new entrants are also those that make it vulnerable to a bubble. Citigroup forecasts “exponential” growth in 2026, yet mass‑market adoption is still years away. Morgan Stanley’s analysis implies that the rate of adoption per household will be below 10% worldwide for a very long time, even until the 2030s, and that basically the first use cases will be in the industrial and commercial sectors. In case the market sentiment gets substantially ahead of the actual development of technology, there might be a sudden decrease in investments, thus a scenario where the less capable entities have to combine with one another or discontinue their operations.
A burst would ripple far beyond China. The country’s push to scale humanoids is central to driving down global prices, and a slowdown could delay the arrival of affordable units in overseas markets. At the same time, market consolidation could temporarily benefit U.S. developers by reducing competitive pressure and making skilled Chinese engineers and surplus components available at lower cost. Given that nearly every major robot maker still relies on Asian suppliers for critical parts such as motors, screws, and batteries, a shake‑up in China’s supply chain could have immediate effects on global production timelines.
Technical maturity is another wild card. Current humanoids, even those showcased in high‑profile demos like marathon runs or somersaults, still struggle with the kind of robust autonomy needed for unsupervised deployment. Advances in AI “brains” from models like DeepSeek, Alibaba’s Qwen, and ByteDance’s Doubao are improving task reasoning and adaptability, but safety, reliability, and energy efficiency remain gating factors. Whole‑body motion planning under variable loads, compliant manipulation to avoid damaging objects, and fail‑safe behaviors in crowded human environments are all active research areas.
For investors and industry strategists, the parallels to the electric vehicle boom are instructive. The cost of electric vehicles declined very quickly due to large subsidies and economies of scale in production, but the low prices also resulted in fierce price wars and the disappearance of less competitive brands. The situation in the development of humanoid robots is even more complicated from an engineering point of view, the timeframe for the widespread use of the technology is longer, and the international political implications are more serious. Should the bubble deflate, it will test not only the resilience of China’s robotics ecosystem but also the readiness of global markets to absorb a slowdown in what many see as the next trillion‑dollar technology frontier.
