•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
2026 is set to become a turning point for China’s humanoid-robot industry, not because of a sudden surge in scale, but because of a “harsh culling” that will force market demand expectations to be replaced by a more realistic breakdown of where the technology can actually be applied. As hundreds of automakers and Chinese startups race to launch walking and dancing robots, a key question has emerged: are today’s humanoid-robot sales truly delivering labor capacity, or are they largely investment-driven stories?
Industry analysis highlights a paradox in current orders. Many deals appear to stop at brand displays, pilot validations, or strategic cooperation agreements rather than translating into sustained production demand. According to Huxiu, these orders often do not reflect real market pull: display orders help buyers attract attention for public relations, pilot orders can be used to “buy” reassurance and avoid falling behind, and strategic contracts carry significant PR value.
What the industry needs most—repeat, recurring orders that improve labor productivity—remains difficult to achieve, leaving companies caught between delivering real value and finding a workable monetization path.
A central industrial barrier remains unresolved: manufacturers have not yet answered why many tasks that naturally call for automation still reject humanoid-shaped robots. In factory environments, being “human-like” is not the goal; usefulness is. While humanoid robots are often promoted for versatility, that same versatility can become a weakness in mass-production settings.
Financial data underscores the challenge. Huxiu cites figures from China’s High-end Robotics Industry Research Institute showing that in Q1 2026 the price of a humanoid robot dropped to 100,000 yuan, down 33% from 2025. However, this is only the basic version.
In practice, deploying robots into specific production steps increases total cost of ownership, including operation, maintenance, and other incidental costs. Huxiu reports that these additional costs can double the figure or more. A robot that can actually work in a factory could cost 500,000 to 600,000 yuan.
Using a simple depreciation-style calculation, replacing a worker earning 100,000 yuan per year with a robot costing 500,000 yuan implies a payback period of up to five years. This is a critical constraint because manufacturing equipment depreciation cycles are typically about five to six years, while technology changes rapidly.
Manufacturing buyers also require more than payback math: they need continuity, stable output, and predictable maintenance.
Reliability concerns further limit adoption. Studies cited by Huxiu indicate that the “body life” of a humanoid robot ranges only from six to eight months. When failures occur, repair time can extend to a month—an interruption that is unacceptable in production lines where losses are measured by the minute.
XPeng Motors chairman Ha Tieu-Bang is quoted as saying that the difficulty of researching and developing humanoid robots is “tens of times” higher than for cars. In the near term, he argues that deploying humanoid robots in factories is not the optimal solution, and that at the current stage they are better suited for shopping guidance, display, or companionship roles.
Automakers may have an advantage through existing distribution channels, including plants for testing and showrooms for selling. However, this does not guarantee commercial success. When a carmaker places a robot into its own production line, it cannot eliminate R&D or manufacturing costs—it only shifts them from procurement to development departments.
Huxiu notes that if a robot priced around 500,000 yuan cannot demonstrate performance superior to a human worker, plant managers have little incentive to use it unless driven by corporate pressure. There is also a risk that prioritizing internal orders could mask technological flaws, slowing the product’s maturation for real market competition.
For consumer markets, humanoid robots also face weak value propositions. A device priced at hundreds of thousands of yuan that can only walk and greet is too expensive as a toy, lacks flexibility for childcare, and takes up too much space to function well as a decorative item. As a result, Huxiu argues that the main appeal for automakers is not short-term revenue but the possibility of entering a new interaction paradigm—though this requires major trade-offs.
Redirecting billions of yuan toward other areas, such as optimizing autonomous driving algorithms or battery technology, could potentially deliver higher returns than humanoid-robot development at the current stage.
Huxiu’s forecast for the next two years points to a powerful culling across the humanoid-robot industry. Companies that rely on flashy demonstration videos but cannot convert them into orders are expected to exit. Firms that survive on demo budgets or government projects may remain, but are likely to struggle to scale.
The eventual winners may not be those with the most dazzling technology, but those that can minimize total cost of ownership, repair quickly, and turn each pilot into a sustainable repeat-purchase cycle.
In 2026, Huxiu describes the industry as a “saga of contrasts”: humanoid robot bodies carry lofty valuation figures and impressive demonstration content, while commercialization momentum stalls due to unresolved efficiency problems. The biggest purge, the analysis suggests, will occur not in laboratories but in sales channels—where a walking, dancing robot is only an entry ticket, and true market viability will be proven when vendors can persuade customers to repurchase based on real productive value.
Source: Huxiu
Premium gym chains are entering a “golden era” that is ending or already in decline, as rising operating costs collide with shifting consumer preferences toward more flexible, community-based ways to exercise. Long-term memberships are shrinking, margins are pressured by higher rents and facility expenses, and competition from smaller, more personalized…
