Humanoid robots are on the verge of scaling up: locomotion matures first, while dexterous manipulation follows
AI summary card
Humanoid robots are on the verge of scaling up: locomotion matures first, while dexterous manipulation follows
Bank of America expects global humanoid robot shipments to rise from 20,000 units in 2025 to 90,000 units in 2026, reaching 1.2 million / 10 million units in 2030 / 2035, with falling costs and advances in embodied AI as the core drivers.
- The report forecasts the global humanoid robot market size will reach US$37 billion / US$224 billion in 2030 / 2035, respectively.
- If key components are mainly manufactured in China, the hardware cost of a typical humanoid robot is expected to fall to about US$28,000 by the end of 2026 and to below US$16,000 by 2030.
- The intelligence level of mainstream humanoid robots is still roughly at the L2 stage, with locomotion capabilities developing faster than hand manipulation capabilities.
- The US and China are the industry's two core hubs: US companies are stronger in chips, AI systems, and general-purpose robots, while Chinese companies have advantages in hardware supply chains, low-cost robot bodies, and motion control.
- Key components include processor chips, actuators, reducers, planetary roller screws, dexterous hands, vision/LiDAR, and structural parts.
Report interpretation
Overview
This report focuses on the manufacturers, technology roadmap, cost curve, and supply-chain opportunities in the 2026 humanoid robotics industry. Its core view is that the industry made solid progress in 2025 and will continue to accelerate product launches and shipment ramp-up in 2026; however, current capabilities still prioritize motion control, while complex hand manipulation and embodied AI generalization remain key bottlenecks before large-scale adoption.
Core views
The report believes the global humanoid robotics industry is entering a high-growth phase: global shipments were about 20,000 units in 2025, are expected to reach 90,000 units in 2026, and are projected to hit 1.2 million / 10 million units in 2030 / 2035. As BOM costs decline, component designs improve, and scale effects emerge, hardware costs could fall below US$16,000 by 2030. Along the value chain, US companies are stronger in chips, software, and AI algorithms, while Chinese companies have greater advantages in robot bodies, motion control, and hardware supply chains.
Analysis framework
The report uses a top-down forecast of shipments and market size, combined with bottom-up BOM breakdowns and supply-chain mapping. The analysis focuses on shipment ramp-up, ASP and BOM decline paths, L1-L5 intelligence classification, differences between locomotion and manipulation capabilities, technological evolution of key components, and the division of roles among robot brands, automotive OEMs, tech giants, and Chinese supply-chain companies.
Methodology notes
Deriving market size from annual shipments, ASP, and long-term penetration
The report forecasts global shipments of 90,000 units in 2026, 1.2 million units in 2030, and 10 million units in 2035, and based on this estimates market sizes of US$37 billion / US$224 billion in 2030 / 2035.
Breaking down hardware costs by modules such as processor chips, actuators, reducers, dexterous hands, and vision systems
The report estimates the hardware cost of a typical humanoid robot at about US$28,000 by the end of 2026 and below US$16,000 by 2030, implying a cumulative BOM cost decline of about 45% from 2026 to 2030.
L1-L5 capability levels covering perception, decision-making, execution, and collaboration
The report believes mainstream humanoid robots are currently around the L2 stage, with basic multimodal perception, execution, and preliminary reasoning capabilities, and are likely to evolve toward L3-L4 over the next three to four years.
Dividing industry roles among robot brands, automotive OEMs, tech giants, and key component suppliers
The report emphasizes that companies in Greater China play a key role in the humanoid robot supply chain, with the US and China respectively holding leading advantages in AI systems and hardware supply chains.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- TSLAHumanoid robot brand and potential platform participant
- Strengths
- Optimus offers potential synergies with automotive manufacturing, autonomous driving algorithms, and software-subscription monetization.
- Weaknesses
- The pace of mass production, cost reduction, and validation of real commercial scenarios still need to be observed.
- Comparison
- Compared with most robotics startups, TSLA has a manufacturing system and software commercialization experience.
- Risks
- Mass-production delays, insufficient algorithm generalization, and actual application ROI falling short of expectations.
- Kia / Boston DynamicsSynergy between automotive OEM and robotics technology platform
- Strengths
- The report believes Kia can benefit from the Kia Robotics Lab, the related Boston Dynamics value chain, and potential labor-cost reductions.
- Weaknesses
- The contribution of the robotics business to group revenue is still at an early stage.
- Comparison
- Compared with pure hardware suppliers, automotive OEMs are more likely to deploy robots in their own manufacturing and service scenarios.
- Risks
- Slow commercialization pace, high R&D spending, and uncertainty over application scope.
- HengliKey component supply chain, covering areas such as actuators
- Strengths
- The report notes that it has a portfolio of key humanoid robot components and early customer advantages.
- Weaknesses
- Demand realization depends on downstream robot brands achieving mass production.
- Comparison
- Against the backdrop of China's hardware supply-chain advantages, it is well positioned to benefit from cost declines and scaled procurement.
- Risks
- Intensifying competition, price declines, and changes in technology roadmaps.
- TuopuSupplier of actuators, structural parts, or related robot components
- Strengths
- The report lists it as a company with a key HR component portfolio and gives it a positive view.
- Weaknesses
- The revenue contribution and earnings elasticity of the robotics business still need to be tracked.
- Comparison
- Compared with single-component suppliers, a portfolio supply capability helps bind leading customers.
- Risks
- Slower-than-expected customer ramp-up, mass-production yield, and cost-down pressure.
- SchaefflerSupply chain related to planetary roller screws, actuators, and precision machinery
- Strengths
- It has deep accumulation in precision machinery and automotive components and may benefit from demand for key humanoid robot transmission components.
- Weaknesses
- The robotics business is still at an early stage relative to its traditional core business.
- Comparison
- Compared with new entrants, traditional industrial component leaders are stronger in process know-how, customer relationships, and quality control.
- Risks
- Substitution by new technologies and margin erosion from falling costs.
- UBTECHChinese humanoid robot brand
- Strengths
- The report says it leads in humanoid robot order backlog, won more than RMB 1.3 billion of orders in 2025, and plans to deliver more than 10,000 units in 2026.
- Weaknesses
- It needs to continue improving manipulation capabilities and real-world deployment efficiency.
- Comparison
- Compared with overseas general-purpose robotics companies, UBTECH is closer to Chinese manufacturing and commercial scenarios.
- Risks
- Delivery shortfalls, timing of order-to-revenue conversion, and intensifying market competition.
- China humanoid robot supply chainCore support for global hardware cost reduction and scaled manufacturing
- Strengths
- It has cost and supply-chain advantages in structural parts, actuators, reducers, motors, batteries, vision systems, and other areas.
- Weaknesses
- Breakthroughs are still needed in high-end AI chips, general embodied AI models, and software systems.
- Comparison
- The US is stronger in chips, AI systems, and general robotics algorithms, while China is stronger in hardware, motion control, and low-cost manufacturing.
- Risks
- Price competition, unsettled technology roadmaps, and export and geopolitical risks.
Key data
- Global humanoid robot shipments in 2025about 20,000 unitsExcludes wheeled and quadruped robots.
- Global humanoid robot shipment forecast for 202690,000 unitsSignificant growth versus 2025.
- Global annual shipment forecast for 2030/20351.2 million units / 10 million unitsEquivalent to a CAGR of about 86% from 2025 to 2035.
- Global market size forecast for 2030/2035US$37 billion / US$224 billionBased on assumptions of ASP decline and shipment growth.
- Typical humanoid robot hardware cost by end-2026about US$28,000Assumes most components are made in China; excludes overhead and AI development-related costs.
- BOM cost forecast for 2030/2035below US$16,000 / US$13,000Driven by scale effects and component design optimization.
- BOM cost decline from 2026 to 2030about 45%The report believes processor chips, coreless motors, reducers, and planetary roller screws will be among the components with larger market sizes in 2030.
- Current intelligence level of mainstream humanoid robotsabout L2Locomotion capabilities are ahead of manipulation capabilities.
- 2025 orders for leading Chinese humanoid robot companiesmore than RMB 2 billionOrders mainly come from R&D, data collection, and manufacturing scenarios.
- Unitree market share in 2025about 28%The report says it ranks first globally in humanoid robot shipments.
Impact & implications
If the report's thesis plays out, the humanoid robotics value chain will gradually move from early-stage demonstrations and R&D orders toward industrial and commercial scenarios, with hardware supply chains, motion control, actuators, dexterous hands, AI chips, vision sensors, and embodied AI platforms becoming key beneficiaries. From an investment perspective, companies with scale manufacturing, key components, robot body, or AI system capabilities are more likely to secure early industry orders and valuation premiums.
Risks
- Embodied AI models have not yet converged; VLA/VTLA models and world models are developing in parallel, leaving uncertainty about the future roadmap.
- There is a shortage of high-quality training data, real-world data collection is costly, and it remains to be verified whether simulation and video data can effectively substitute.
- Current manipulation capabilities lag locomotion capabilities, and complex tasks such as grasping, sorting, and assembly remain commercialization bottlenecks.
- Mass production requires solving issues related to cost, yield, modularization, lightweighting, and human-machine safety design.
- The industry may see intense price competition; supply-chain cost reduction is progressing quickly but margins could come under pressure.
- Shipment and market-size forecasts rely on strong long-term assumptions; if application deployment is slow, demand could come in below expectations.
What to watch
- Actual deliveries and capacity ramp-up of leading companies such as Tesla, Unitree, AgiBot, and UBTECH in 2026.
- The real ROI of humanoid robots in factories, warehouses, commercial services, and household scenarios.
- Whether manipulation capabilities can move from L2 toward L3-L4 over the next three to four years.
- Whether the embodied AI pathway will form a clearer dominant architecture between VLA/VTLA models and world models.
- The pace of cost reduction in China's supply chain for actuators, reducers, planetary roller screws, dexterous hands, and vision systems.
- Whether processor chips, edge computing, and data flywheels can support lower latency and stronger generalization capabilities.
- Whether key component companies can secure mass-production nomination orders from leading robot brands.