At the 20th Frost & Sullivan Global Growth, Innovation, and Leadership Summit alongside the 5th New Investment Conference held in Shanghai, UBTECH ROBOTICS (09880.HK) CFO and Board Secretary Zhang Ju outlined the company's strategic vision for embodied intelligent humanoid robots.
According to Zhang, commercialization will not begin with complex scenarios but will instead target simple, repetitive tasks that humans are reluctant to perform and that traditional automation cannot fully cover. This approach is driving rapid expansion across three key application domains: industrial, commercial, and household.
This strategy is underpinned by a significant shift in UBTECH ROBOTICS' revenue structure. The company's 2025 annual report shows that full-sized embodied intelligent humanoid robot products and services generated approximately 821 million yuan in revenue, a year-on-year increase of about 2,203.7%. This segment now accounts for 41.1% of total revenue, making it the company's largest income source. Zhang anticipates that the proportion of humanoid robot revenue will continue to grow rapidly in 2026.
However, Zhang refrained from labeling the current phase as "mass production." Drawing a parallel to the automotive industry, he stated that true mass production involves millions of units. Humanoid robots, he noted, are still in the small-scale production phase, measured in tens of thousands of units. This perspective draws a realistic boundary for the highly hyped industry: while production numbers and order growth are important, scaling from tens of thousands to hundreds of thousands and millions requires the collective maturity of the supply chain, application scenarios, technology, and product capabilities.
On the production side, Zhang believes that China, particularly the Pearl River Delta and Yangtze River Delta regions, already possesses a strong manufacturing and supply chain foundation. Capacity expansion, he argues, is not the most critical concern for the robotics industry. The most important question is what tasks robots will perform in factories and why companies are willing to pay for them. His answer begins with "simple, repetitive, and low-value-added" tasks such as material handling, sorting, and machine loading/unloading. These jobs involve high labor intensity and high turnover rates, representing the "capillaries" that existing automation equipment has not fully covered.
Zhang emphasized that humanoid robots are not intended to replace highly mature robotic arms or automated production lines but to fill the gaps where automation still requires significant human involvement. For example, while painting and welding in automotive factories are highly automated, assembly lines still rely on a large workforce. This highlights the presence of non-standard, fragmented, and flexible tasks in the production system that need to be reorganized. This means the value of humanoid robots lies not just in their human-like form but in their ability to enter spaces designed for humans, use existing tools, and perform tasks across multiple workstations.
Regarding the form factor—whether bipedal, wheeled, or fixed—Zhang does not advocate for a single predetermined solution. He stated that the product form should be determined by the scenario: wheeled platforms can be used on flat surfaces, bipedal forms are needed for crossing tracks or stairs, and some fixed workstations may not require movement at all. Versatility is not about having one robot do everything, but rather making economical configurations based on real needs, using a common technology base.
Commercial viability ultimately depends on the return on investment cycle. Zhang noted that the payback period for industrial humanoid robots depends on whether the factory operates on one, two, or three shifts. Under high utilization rates, customers can typically expect to recover their investment within one to two years. This assessment also explains why industrial scenarios are the first to be implemented: task boundaries are relatively clear, operating time can be quantified, and replacement costs are easier to calculate.
However, a robot's ability to operate autonomously without human remote control tests more than just the hardware. Zhang described UBTECH ROBOTICS' research and development focus as being on the "brain." According to the company's annual report, R&D expenses for 2025 were approximately 508 million yuan, and the company had accumulated 2,985 authorized patents by the end of the year. Zhang stated that about 70% of the company's R&D investment is directed toward software, including embodied intelligence large models, world models, VLA models, swarm intelligence, and task planning and execution capabilities for real-world scenarios. This is a key point he emphasizes when explaining the company's current lack of profitability: a loss does not necessarily mean the business model is invalid, but rather that the company is actively choosing to invest resources in long-term R&D.
Zhang said that if R&D investment were reduced, short-term financial performance could improve, and the company could even become profitable. However, UBTECH ROBOTICS currently prioritizes technological depth and future competitiveness, and it will seize the historic opportunity to broaden and deepen its technological and product moat.
In overseas markets, UBTECH ROBOTICS has adopted a pragmatic approach: "riding the boat to go to sea." Zhang explained that industrial humanoid robots can first enter the factories of Chinese manufacturing companies or multinational corporations within China. After validation, they can follow these clients' overseas production expansions into local factories. Compared to directly entering unfamiliar markets, this method leverages existing client relationships, process experience, and validation results, reducing the uncertainty of early international expansion. He also believes that China remains a significant global manufacturing hub, and the largest application scenarios for industrial humanoid robots will continue to be domestic for a considerable period.
If the keyword for industrial scenarios is efficiency, the starting point for household scenarios may be emotion. Responding to the question of why bringing robots into homes is more difficult, Zhang offered a different perspective. If the initial requirement is for a robot to cook, clean, and handle all household chores, the technical barrier is indeed very high. However, if the entry point is companionship, interaction, and emotional value, the path for a product to enter the home might be shorter. In his envisioned roadmap, household robots are not initially all-purpose tools but rather first establish a human-robot relationship before gradually adding practical functions.
Commercial scenarios lie between industrial and household applications. Tasks such as guiding, hosting, retail, and coffee making have both service attributes and relatively clear procedures. Zhang noted that performance is only a small part of commercial applications; in the future, robots can undertake various tasks in more service roles, providing humans with a better quality of life. This leads to UBTECH ROBOTICS' "three-pronged" strategy: the industrial sector is first to validate economic viability, the commercial sector expands interaction and services, and the household sector targets a longer-term and potentially broader consumer market.