China's mass production exceeds expectations, and the humanoid robot ecosystem flywheel begins to accelerate

Overview

The report reviews progress in the global humanoid robot industry in the first half of 2026. Chinese manufacturers are rapidly expanding production, U.S. companies' mass-production activities are beginning to accelerate, and Europe, Japan, and South Korea are also being driven by localized demand. Based on China's better-than-expected production progress, the report raises global short-term and long-term market forecasts and explains the industry's acceleration mechanism through an ecosystem flywheel composed of hardware, models, data, and applications. At the same time, the report notes that industrial applications remain in the validation phase, with reliability, cost, and payback period being the main constraints on scaled commercialization.

Core views

First, China is the core driver of global humanoid robot volume growth in 2026, expected to contribute about 70,000 units of shipments; second, the global industry has gradually shifted from demonstrations and R&D to real-world scenarios such as factories and warehousing, but mass production and commercialization are not synchronized; third, data accumulation, model iteration, and hardware cost reduction may drive an industry inflection point before 2030; fourth, tighter U.S. regulation will reinforce localized supply chains and change the competitive conditions for manufacturers in different regions; fifth, investment opportunities include both full-machine platforms and high-value components such as lead screws, motors, reducers, and actuators.

Analysis framework

The report combines corporate capacity and production progress, industry channel checks, government production guidance, company management statements, real-world operating performance, and regional policy changes to make a bottom-up forecast of global shipments. It also uses an S-curve for manufacturing ramp-up to assess the pace of U.S. manufacturers' mass production, and analyzes the industry's positive feedback mechanism and commercialization conditions through a four-factor framework of hardware, models, data, and applications.

Methodology notes

• Industrial ecosystem analysisHumanoid robot ecosystem flywheel

  Hardware, models, data, and applications form a mutually reinforcing closed loop.

  More deployments generate more real-world data, data improves model capabilities, models increase task success rates and application value, and greater scale further drives hardware cost reduction and supply-chain maturity.

• Market forecastBottom-up shipment forecast

  Aggregate global shipments by company, region, and mass-production plan.

  The forecast references the cumulative production of Chinese manufacturers, government targets, the production ramp-up of U.S. manufacturers such as Tesla and Figure, and localized demand in Europe, Japan, and South Korea.

• Capacity ramp-up analysisManufacturing S-curve

  Mass production of new products is slow in the early stage and enters an acceleration phase after processes stabilize.

  The report uses this framework to explain the path in which Tesla Optimus has relatively low initial output but rapidly rising subsequent targets, while emphasizing that new components and manufacturing complexity may extend the early stage.

• Commercialization assessmentSuccess rate and payback period

  Use task reliability and economic returns to test real-world application value.

  Entertainment and scientific research already have viable business models, but industrial task success rates vary widely and payback periods typically exceed two years, so a broadly replicable commercial closed loop has not yet formed.

Asset mapping & comparison

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

• Hengli Hydraulic（601100.SS）

  A beneficiary in humanoid robot components, rated BUY, with a target price of RMB124.

  Strengths

  The per-robot content value continues to increase, with products covering body lead screws, body motors, and hand motors, while the company is expanding its customer base; its main construction machinery business provides support.

  Weaknesses

  Robot component revenue in 2026 remains relatively limited, expected at RMB100 million to RMB200 million.

  Comparison

  Compared with pure full-machine manufacturers, its beneficiary path is more oriented toward the value of core components and customer expansion.

  Risks

  Customer mass production falls short of expectations, delayed realization of robot business revenue, and failure of new product validation.

• Shuanghuan Driveline（002472.SZ）

  A potential beneficiary in robot reducers and actuators, rated BUY, with a target price of RMB47.

  Strengths

  It has been cooperating with Tesla on robot R&D for more than three years and may benefit from high-load rotary actuators adopting new types of reducers; it also has overseas expansion potential.

  Weaknesses

  The core automotive gearbox business is under pressure in 2026, and robot-related opportunities still need to be converted into orders.

  Comparison

  In addition to humanoid robots, it also covers new businesses such as traditional robots, industrial equipment, robot vacuum cleaners, drones, and electric bicycles.

  Risks

  Changes in Tesla's technology roadmap, delays in mass-production pace, downturn in the automotive main business, and overseas expansion falling short of expectations.

• Harmonic Drive Systems（6324.T）

  A beneficiary in the humanoid robot reducer and actuator supply chain outside China, rated BUY, with a target price of ¥9,900.

  Strengths

  After expanding from reducers to complete actuators, the unit selling price opportunity is about three times that of standalone reducers; it also benefits from growth in aerospace and semiconductors.

  Weaknesses

  Robot demand is still in an early stage, and actuator business volume growth depends on customer mass production.

  Comparison

  Compared with Chinese reducer manufacturers, its advantage is mainly reflected in its supply position in markets outside China.

  Risks

  Robot production below expectations, intensified competition, customer in-house development, and high-valuation volatility.

• Tesla（TSLA.US）

  A beneficiary of the Optimus full-stack full-machine platform, rated BUY, with a target price of US$420.

  Strengths

  It has vertically integrated hardware design, strong computing power, and manufacturing capabilities, and the bottleneck in hand hardware is expected to ease; it plans to increase internal deployment in the fourth quarter of 2026.

  Weaknesses

  The early stage of mass production is in the flatter part of the S-curve, and general task capabilities have not yet been commercially validated at scale.

  Comparison

  Compared with component or single-model companies, Tesla is simultaneously positioned in full machines, manufacturing, model training, and internal application scenarios.

  Risks

  Mass-production progress falls short of targets, cost control difficulties, delays in external sales, and technical demonstrations failing to translate into stable operations.

• Mobileye（MBLY.US）

  Entering low-cost warehouse and commercial humanoid robots through Mentee Robotics, rated BUY, with a target price of US$13.

  Strengths

  At low production volumes, the bill of materials is about US$50,000, with a path to reduce it by 50% once scale reaches several thousand units; it has already launched proofs of concept with multiple customers.

  Weaknesses

  The business is still in the early validation stage, and scaling is expected to potentially occur in 2028 to 2029.

  Comparison

  Compared with the high-performance full-stack route, Mentee places greater emphasis on cost advantages for warehousing and commercial scenarios.

  Risks

  Proofs of concept fail to convert into orders, cost-reduction targets are not achieved, M&A integration, and customer adoption is slower than expected.

Key data

• 2026 global shipment forecastAbout 80,000 unitsPrevious forecast was about 50,000 units; the chart corresponds to about 79,000 units.
• 2026 China shipment forecastAbout 70,000 unitsChina is expected to remain the world's largest humanoid robot market.
• China's production in the first half of 2026More than 40,000 unitsProduction progress exceeded the report's previous expectations.
• 2030 global market forecast1 million units / US$25 billionThe previous shipment forecast was 700,000 units.
• 2050 global market forecast100 million units / US$1.5 trillionThe previous shipment forecast was 70 million units.
• Unitree cumulative productionAbout 11,000 units as of May 2026Only bipedal humanoid robots are included.
• AGIBOT cumulative productionAbout 15,000 units as of June 2026The statistical scope includes bipedal and wheeled embodied-intelligence robots.
• Industrial task success rateAbove 99% to below 50%Differences are significant across tasks and deployment environments.
• Industrial application payback periodTypically more than two yearsCurrent industrial scenarios are not yet broadly commercially viable.
• Hengli Hydraulic target priceRMB124Rated BUY.
• Shuanghuan Driveline target priceRMB47Rated BUY.
• Harmonic Drive Systems target price¥9,900Rated BUY.
• Tesla target priceUS$420Rated BUY.
• Mobileye target priceUS$13Rated BUY.

Impact & implications

Industry volume growth will expand the market space for full machines, actuators, reducers, lead screws, motors, sensors, and robot model platforms. Short-term investment upside is more likely to come from component suppliers that already have customer validation, rising per-unit value, and the ability to withstand mass-production volatility. Medium- to long-term value depends on whether full-machine companies can obtain real-world data, improve general task capabilities, and reduce costs to acceptable levels. Tighter U.S. policy may drive regional supply-chain divergence and localized production, benefiting domestic U.S. manufacturers but increasing uncertainty for Chinese companies entering the U.S. market.

Risks

• Industrial application task success rates vary significantly, and a broadly reliable commercial closed loop has not yet formed.
• Payback periods typically exceed two years, and customers' willingness for capital expenditure may be insufficient.
• Manufacturers' announced long-term capacity targets may be significantly higher than actual demand or executable capacity.
• Key hand, actuator, sensor, and model capabilities may still constitute bottlenecks for mass production.
• U.S. restrictions on foreign-made advanced robotic equipment may expand, disrupting cross-border sales and supply chains.
• Rapid changes in technology roadmaps may cause existing component solutions to be replaced or customers to shift to in-house development.
• The report's long-term forecasts are highly sensitive to cost declines, data expansion, and model evolution.

What to watch

• Whether China's full-year 2026 production can approach the Ministry of Industry and Information Technology's forecast of 100,000 units.
• Actual shipments, customer mix, and repeat purchases of Chinese manufacturers such as Unitree, AGIBOT, and UBTECH.
• Tesla Optimus production-line ramp-up and progress of internal deployment in the fourth quarter of 2026.
• Whether Figure can maintain production and quality levels after increasing from one unit per day to one unit per hour.
• Whether industrial-site task success rates, failure rates, and payback periods continue to improve.
• Whether data generated from real-world deployments can effectively drive improvements in general model capabilities.
• The specific scope of application of U.S. advanced robot restriction policies to new imports, components, and Chinese manufacturers.
• Tesla's timetable for starting external industrial sales, expected from late 2027 to early 2028.
• Whether Mobileye's Mentee proofs of concept can translate into commercial progress over the next 9 to 12 months.
• Core component suppliers' customer nominations, per-unit content value, and revenue realization in 2027.