Humanoid robot commercialization is shifting from demonstrations to real-world deployment, with four major applications poised to cross the payback-period threshold first

Overview

The report examines the path for humanoid robots to move from demonstrations of locomotion capabilities to real-world commercial adoption, focusing on the payback periods, price sensitivity, and potential market size of four near-term deployable applications in China and the United States. Its core conclusion is that the economics of inspection are already viable, while other applications still require some price reductions; however, learning curves from scaling are expected to materially ease pricing constraints within two to three years.

Core views

The report believes that the humanoid robotics industry's focus in 2026 is rapidly shifting from demonstrations of walking and locomotion capabilities to commercial adoption. The four applications most likely to be deployed first over the next five years are urban and industrial inspection, warehouse picking and material handling, last-mile home delivery, and factory material handling and inspection. These applications have already reached or are approaching human performance levels, while their tasks are relatively standardized and the labor substitution relationship is relatively clear. Although factory installation and assembly, as well as most commercial and household applications, may have larger long-term addressable markets and have already seen proof-of-concept projects, the report believes that their technological maturity and investment returns remain insufficient to support near-term scaling. The report calculates the payback period as upfront investment divided by annual net savings or net benefits after deployment and provides separate estimates for China and the United States. Based on currently available robot prices, the payback periods for the four applications range from 1.0 year for inspection in the United States to 6.7 years for factory material handling and inspection in China. Although robot prices are generally higher in the United States, labor costs are typically three to five times those in China, resulting in faster overall payback; Chinese manufacturing wages are only 23% of U.S. levels, reducing the labor substitution benefits in factory applications. Inspection has a payback period below two years in both countries because each robot can replace more inspection personnel and robot prices are relatively low. Last-mile delivery has a relatively long payback period because the report assumes that one robot must still be paired with one courier, resulting in a low substitution ratio. The U.S. inspection calculation illustrates this economic mechanism: eight robots are purchased at $75,000 each, with integration costs equal to an additional 30% of the robot price, resulting in total upfront investment of $780,000. Before deployment, there are 30 inspection personnel; after deployment, 12 operators are retained, while annual robot maintenance and electricity costs total $90,000. Annual cost savings amount to $808,560, corresponding to a 1.0-year payback period. By comparison, the Chinese factory material handling and inspection solution uses two robots priced at $65,000 each, with the additional 30% integration cost bringing upfront investment to $169,000. Annual net savings are only $25,373, resulting in a payback period of 6.7 years. The report treats 2.5 years as a typical payback-period threshold at which adoption accelerates materially. Excluding inspection, for the other three applications to reach this threshold, robot prices must decline by 30% to 50% from current levels in China and by 5% to 40% in the United States. The corresponding robot prices are approximately $30,000 to $40,000 for general applications and $17,000 for last-mile delivery in China; in the United States, prices for general applications could remain at $130,000 to $140,000, while last-mile delivery robots would need to cost approximately $46,000. The report therefore believes that three of the four applications are already attractive at current prices in the United States, whereas China is more dependent on price reductions. The price-reduction outlook is based on learning curves from scaling. Drawing on experience from the electric vehicle and battery industries, the report notes that costs typically decline by 15% to 18% each time production doubles, and cost reductions are often faster when an industry is at an extremely low production stage. Major humanoid robot manufacturers such as Agility also expect increased production to reduce costs significantly. Based on this experience, the report concludes that robot pricing is unlikely to remain a major adoption barrier over the next two to three years. However, alternative solutions still have advantages at this stage: in material handling and installation applications, the payback periods for an “AGV + collaborative robot” solution are four years in China and 2.6 years in the United States, shorter than the 6.7-year and 2.7-year payback periods for humanoid robots, respectively. The report expects humanoid robots to gradually achieve better economics only after production scales up. Real-world deployment data are used to demonstrate that the technology is approaching commercial requirements. Agility Digit achieved 98% accuracy in the GXO project, handled more than 100,000 totes, and accumulated over 65,000 operating hours. Figure 02 loaded more than 900 components in the BMW project, accumulated over 1,250 operating hours, and participated in the production of more than 30,000 BMW X3 vehicles. Agibot achieved an overall success rate above 99.5% in related tasks, with each process taking 18 to 20 seconds and throughput reaching up to 310 items per hour. UBTECH Walker S accumulated more than 10,000 operating hours at Zeekr's factory, with a failure rate below 0.5%. These cases show that simple, repetitive tasks are reaching deployable performance levels, but tasks involving 10 to 20 steps and small components still have relatively low success rates. Regarding market size, the report estimates the potential installed base for the four applications based on substitutable jobs, the addressable proportion of each application, and penetration rates. It estimates approximately 9.9 million units in China and 1.2 million units in the United States, summarized as roughly 10 million and 1 million units, respectively, with factory material handling and inspection representing the largest opportunity. The number of relevant jobs is estimated at 28.7 million in China and 3.5 million in the United States. If penetration reaches approximately 8% over the next five years, the combined installed base across the four applications would reach approximately 900,000 units by 2031, with annual shipments exceeding 500,000 units. This would account for approximately half of the 1 million annual global humanoid robot shipments projected by the report for 2031, implying that near-term adoption will remain highly concentrated in industrial and quasi-industrial applications. To identify industry leaders, the report proposes three areas to track: first, whether real commercial sales continue to grow and whether commercial customer expansion can be validated; second, whether companies can use real-world deployment data to close the reinforcement-learning feedback loop and continuously improve performance, rather than merely emphasizing total data volume; and third, whether robot manufacturers are moving toward the center of the industry ecosystem and whether there is early evidence of system integrators clustering around specific manufacturers. Accordingly, the report emphasizes that actual orders, deployment learning efficiency, and ecosystem integration capabilities are more indicative of commercialization progress than locomotion demonstrations alone.

Analysis framework

The report first screens the four applications most likely to be deployed over the next five years based on technological maturity and investment returns. It then estimates payback periods in China and the United States using robot procurement and integration investment, maintenance and electricity costs, operator costs, and displaced labor costs. Next, it conducts robot price sensitivity analysis, using a 2.5-year payback period as the threshold for accelerated adoption, and compares humanoid robots with “AGV + collaborative robot” solutions. Finally, drawing on cost reductions from scaling in the electric vehicle and battery industries, the report estimates the potential installed base and 2031 shipments based on substitutable jobs, addressable proportions, and penetration rates, and proposes three metrics for identifying leading manufacturers.

Methodology notes

• Company Fundamentals and Financial Framework

  Investment Payback Period Analysis

  The report divides upfront investments such as robot procurement and system integration by the annual labor cost savings after deployment, net of maintenance, electricity, and operator costs, to assess the economic viability of different applications.

• Competitive and Strategic FrameworkEconomies of scale / learning curve

  Cost learning curve from production doubling

  Drawing on experience from the electric vehicle and battery industries, where costs typically decline by 15% to 18% each time production doubles, the report assesses whether scaled humanoid robot production can achieve the prices required for a 2.5-year payback period.

• Industry/Sector Analysis FrameworkSubstitution Effect Analysis

  Substitution comparison between humanoid robots, human labor, and AGV + collaborative robot solutions

  The report compares labor substitution ratios, operating shifts, upfront investment, and payback periods across different solutions to illustrate where humanoid robots currently have an advantage and where they still lag existing automation solutions.

• Industry/Sector Analysis FrameworkSupply-demand framework

  Market size estimation driven by substitutable jobs, addressable proportions, and penetration rates

  Starting with the number of relevant jobs in China and the United States, the report adjusts for the addressable proportion of each application and then applies an approximately 8% penetration assumption to derive the potential installed base, the installed base in 2031, and annual shipments.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Hesai (HSAI, 02525.HK)

  The report lists Hesai as an Outperform-rated stock.

  Comparison

  Rated Outperform.

• JD (JD, 09618.HK)

  The report lists JD as an Outperform-rated stock.

  Comparison

  Rated Outperform.

• Xiaomi Group (01810.HK)

  The report notes Xiaomi's participation in factory material-loading tests using humanoid robots and lists it as an Outperform-rated stock.

  Strengths

  The relevant tests lasted four months, with success rates exceeding 90% across multiple tasks.

  Comparison

  Rated Outperform.

• XPeng (XPEV, 09868.HK)

  The report lists XPeng as a Market-Perform-rated stock.

  Comparison

  Rated Market-Perform.

• BYD (01211.HK)

  The report lists BYD as an Outperform-rated stock and mentions that it is one of the customers of UBTECH Walker S.

  Comparison

  Rated Outperform.

• Amazon (AMZN)

  The report lists Amazon as an Outperform-rated stock and notes that it is testing a last-mile delivery solution in which robots are paired with couriers to serve different buildings.

  Strengths

  It is already involved in warehouse robotics customer projects and is exploring last-mile home delivery applications.

  Weaknesses

  Last-mile delivery still assumes one robot paired with one courier, resulting in a low labor substitution ratio.

  Comparison

  Rated Outperform.

• Coupang (CPNG)

  The report lists Coupang as an Underperform-rated stock.

  Comparison

  Rated Underperform.

• Sanhua Intelligent Controls (02050.HK)

  The report lists Sanhua as a Market-Perform-rated stock.

  Comparison

  Rated Market-Perform.

• Estun (02715.HK)

  The report lists Estun as a Market-Perform-rated stock.

  Comparison

  Rated Market-Perform.

Key data

• Current Payback Period Range1.0 to 6.7 yearsThe lowest is U.S. inspection, and the highest is Chinese factory material handling and inspection
• Accelerated Adoption Threshold2.5 yearsThe typical payback period used by the report to assess whether commercial adoption may accelerate materially
• Required Price Reduction in China30% to 50%The price reduction required for the three applications other than inspection to achieve a 2.5-year payback period
• Required Price Reduction in the United States5% to 40%The price reduction required for the three applications other than inspection to achieve a 2.5-year payback period
• Chinese Prices Corresponding to a 2.5-Year Payback Period$30,000 to $40,000; $17,000 for last-mile deliveryRobot prices estimated by the report for general applications and last-mile delivery
• U.S. Prices Corresponding to a 2.5-Year Payback Period$130,000 to $140,000; $46,000 for last-mile deliveryRobot prices estimated by the report for general applications and last-mile delivery
• Labor Cost Difference Between China and the United StatesU.S. labor costs are typically three to five times those in ChinaExplains why payback periods are generally shorter in the United States even when robot prices are higher
• Chinese Manufacturing Wage Level23% of the U.S. levelResults in lower labor substitution benefits for factory material handling and inspection applications in China
• Cost Reduction from Scaling15% to 18% decline each time production doublesEmpirical data from the electric vehicle and battery industries
• AGV + Collaborative Robot Payback Period4 years in China and 2.6 years in the United StatesCurrently shorter than the 6.7 years in China and 2.7 years in the United States for humanoid robots
• Potential Installed Base for the Four Applications9.9 million units in China and 1.2 million units in the United StatesSummarized by the report as approximately 10 million units in China and approximately 1 million units in the United States
• Number of Relevant Jobs28.7 million in China and 3.5 million in the United StatesUsed to derive potential robot demand across the four applications
• 2031 Penetration RateApproximately 8%Assumed penetration rate for the four priority applications over the next five years
• 2031 Installed BaseApproximately 900,000 unitsCombined total for the four applications in China and the United States
• 2031 Annual ShipmentsMore than 500,000 unitsApproximately 50% of the 1 million global humanoid robot shipments projected by the report for that year
• Annual Cost Savings for U.S. Inspection$808,560The eight-robot solution requires upfront investment of $780,000, corresponding to a 1.0-year payback period
• Annual Cost Savings for the Chinese Factory Application$25,373The two-robot solution requires upfront investment of $169,000, corresponding to a 6.7-year payback period

Impact & implications

The report believes that near-term commercialization of humanoid robots will not spread evenly across all applications but will initially concentrate in industrial and quasi-industrial applications where tasks are standardized, technology is close to human performance, and labor substitution benefits are clear. Economics become viable earlier in the United States because of higher labor costs, while China is more dependent on declining robot prices and expanded production scale. If the learning curve materializes as the report expects, the four applications could gradually cross the payback-period threshold over the next two to three years and contribute approximately half of the industry's annual shipments by 2031.

What to watch

• Track the growth trajectory of robot manufacturers' actual commercial sales and whether commercial customer expansion can be validated.
• Track whether manufacturers can use real-world deployment data to close the reinforcement-learning feedback loop and continuously improve performance, rather than merely increasing data volume.
• Track whether manufacturers are moving toward the center of the industry ecosystem and whether system integrators are beginning to cluster around specific robot manufacturers.