On September 12, UBTECH's 10,000-unit-level super smart factory for industrial humanoid robots in Liuzhou officially commenced production. With one humanoid robot rolling off the production line every 10 minutes, the planned annual capacity exceeds 10,000 units. The entire process from production to testing, warehousing, and shipping is fully digitized, realizing the vision of "using robots to build robots."
The direct result of capacity release is the accelerated drop in prices. At the end of August, Goldman Sachs predicted that the average price of a complete humanoid robot would fall from USD 41,800 in 2025 to USD 21,300 in 2035, a nearly 49% decrease over a decade. However, the pace of price reductions in the Chinese market is much faster than overseas forecasts: the unit price of Unitree Robotics' humanoid robots has dropped from CNY 593,400 in 2023 to CNY 166,400 in 2025, a shrinkage of about 72%; in June this year, the price of its consumer-grade product R1 was further reduced to CNY 29,900, and the pre-sale price of Songyan Dynamics' small-sized product is already less than CNY 10,000. Industrial-grade models are no exception: Astribot's T1 starts at CNY 89,900.
The driving forces behind the price drop: scaled mass production dilutes fixed costs, the localization rate of core components has exceeded 75%, and the convergence of technological routes coupled with manufacturers' market positioning. The unit price of harmonic reducers has dropped from CNY 3,000-5,000 to CNY 1,500-2,000, planetary roller screws have seen a drop of over 80%, and six-axis force sensors are moving from the CNY 10,000 level to under CNY 5,000.
However, a halved price merely crosses the threshold for market entry. The core element that truly triggers the industry's scaled explosion has never been low prices, but rather the ability to "get the job done."
Zhang Shaozheng, co-founder of Yuanli Lingji, gave a counter-intuitive judgment: the cheapest robots are actually the most expensive. Let's calculate the overall ROI: a robot with a 99% success rate and one with 95% seem to differ by only 4 percentage points, but the actual cost of manual fallback is several times higher. Robots are not consumer electronics; they are means of production. A robot that breaks down every other day is meaningless, no matter how cheap it is.
There are different ways for humanoid robots to reduce prices. One relies on increasing production volume, maturing processes, and improving yield rates, which naturally brings down costs. This kind of price reduction does not affect quality; on the contrary, the more they sell, the more problems are exposed, and the more reliable the product becomes. The other is forced to cut prices by price wars, relying on downgrading specifications, reducing material usage, and skipping processes to lower prices, which will definitely sacrifice quality. Zhang Shaozheng's bottom line is very clear: Reliability cannot be traded off. Performance can have high and low tiers, but reliability does not.
The industry is shifting from a "capacity race" to a "reliability race." The industry timeline given by Zhang Shaozheng is: 2027 to 2028 will be a critical validation period, with the core focus on whether robots can stably implement standardized processes in scenarios such as industry and logistics, completing the leap from being able to demonstrate to being able to work; if the supply chain, model capabilities, and mass production progress maintain the current pace, 2028 to 2030 will become the true inflection point for cost and scale.
Returning to the industry milestone of the 10,000-unit factory commencing production, capacity release will further accelerate the price drop. Goldman Sachs' ten-year "halving" prediction is highly likely to be realized, and may even happen ahead of schedule. But ultimately, what determines the speed of humanoid robot popularization is not whether it sells for USD 20,000 or USD 30,000, but whether the robot can stably complete high-value work every day.
This article is compiled based on publicly available online information and is for reference only, not constituting investment advice.