China Collaborative Robot 2026 — From JAKA and Dobot to ABB Yumi, A New Human-Robot Co-Line Industrial Order
Research Institute | 2026-06-20
Chapter 1 Industry Overview and Cobot Definition
From the Iron Cage Era to the Co-Line Era
The history of industrial robots is, at its core, a history of human-machine relationships. In the 1960s, the first industrial robot — Unimate — was deployed on General Motors' welding line, immediately surrounded by iron safety fencing. Engineers quickly discovered that a six-axis heavy-duty robot arm, running at full speed, generated enough kinetic energy to kill a worker who strayed into its path. For half a century, the "safety fence" became the defining symbol of factory automation: robots spinning at high speed inside the cage, humans working in support just outside, separated by an impassable physical barrier. Inside the fence was efficiency; outside was safety. This structural contradiction haunted traditional industrial automation.
Collaborative robots (cobots) represent a fundamental challenge to this paradigm. Instead of building walls between humans and machines, cobot technology makes robots inherently safe to work alongside — through active force sensing, collision detection, and power-limiting algorithms that stop the robot within milliseconds if unexpected contact is detected.
ISO 10218-1:2025 (published 2025) defines a "collaborative application" as one in which humans and robots share workspace and collaboratively execute tasks without physical barriers, with the robot and its integrated system designed to ensure human safety throughout. Notably, the standard emphasizes "collaborative application" rather than "collaborative robot" — because safety is a property of the whole system design, not the hardware alone.
China's national standard 36008-2018 "Robots and Robotic Devices — Collaborative Robots" provides the domestic legal definition: a collaborative robot is an industrial robot capable of direct interaction with humans in a shared workspace, without causing harm to humans throughout the contact process.
Cobot vs. Traditional Industrial Robot: Fundamental Differences
The value of cobots becomes clearest in direct comparison with traditional six-axis heavy industrial robots. Both are articulated robot arms driven by servo motors, both capable of handling, welding, and assembly — but their design philosophy, technical approach, target applications, and target customers differ fundamentally:
Payload and Body Weight: Traditional industrial robots carry 12–700 kg payloads, weigh hundreds of kilograms, and require concrete foundations or heavy steel bases. Moving them requires cranes and specialized crews, taking days to weeks. Cobots are designed for mobility: 3–30 kg payload range, 10–33 kg body weight (most can be carried by one person), repositionable within hours without specialized equipment. The same cobot workstation can assemble iPhone screens Monday, reconfigured to assemble laptop hinges by Friday.
Safety Architecture: Industrial robots rely on physical isolation (fences, light curtains, safety mats, dual-channel e-stop circuits) to protect humans. The underlying logic is: the robot is inherently dangerous; safety comes from external boundaries. Cobots use "active safety" logic: built-in joint torque sensors measure contact forces in real time; collision detection algorithms distinguish normal working loads from unexpected contact; upon detecting over-limit contact force, the robot executes a power and force limiting (PFL) soft stop within milliseconds, reducing end-effector contact force to 80–150 N — below the biomechanical injury threshold defined in ISO 10218-2:2025.
Programming Accessibility: Industrial robot programming requires engineers fluent in proprietary languages (FANUC KAREL, KUKA KRL, ABB RAPID) with deep knowledge of kinematics, singularity avoidance, and safety configuration. Training a competent industrial robot programmer takes 6–12 months. Cobots support lead-through teaching (the operator physically guides the arm to demonstrate the trajectory; the system records key points), graphical flow programming (visual drag-and-drop interfaces requiring no code), and increasingly natural language programming (describe the task in plain language; AI generates the execution steps). Line operators can achieve basic configuration after a few hours of training.
Price Gap: A complete industrial robot workstation (robot, safety fencing, control cabinet, pneumatics, system integration) typically costs RMB 1–4 million; large-scale lines cost tens of millions. A complete cobot workstation runs RMB 250,000–600,000 with integration, payback period typically 18–36 months — within the decision-making range of most small and medium manufacturers.
Speed vs. Precision Trade-off: Traditional robots prioritize speed and repeatability (±0.02–0.05 mm, cycle times under 10 seconds). Cobots prioritize force control precision and safety compliance, operating at reduced speeds to maintain power-limiting guarantees. Most cobots have maximum joint speeds of 100–200°/s, limiting end-effector linear speed to typically 1.5–2.5 m/s.
The Cobot Spectrum: Seven Configuration Types
Not all cobots are equal. The commercial cobot market has diversified into seven distinct configuration categories, each optimized for specific application scenarios:
1. Standard Six-Axis Single-Arm Cobots (most common): Six degrees of freedom (DOF), 3–20 kg payload, 580–1,300 mm reach. Covers 80%+ of industrial cobot applications. Representative: UR10e, JAKA Pro 12, Dobot MG series.
2. Seven-DOF Redundant Single-Arm Cobots: One extra joint provides greater obstacle avoidance capability, enabling more flexible trajectory planning in confined spaces or around existing equipment. Representative: KUKA LBR iiwa 7 (7 DOF, all joints with torque sensors).
3. Dual-Arm Cobots: Two synchronized six- or seven-DOF arms sharing a common controller, enabling tasks requiring two-handed coordination — watch movement assembly, high-precision PCB assembly, dual-hand screw tightening. Representative: ABB YuMi (IRB 14000, 500 g per arm, ±0.02 mm precision).
4. SCARA Cobots: Four-DOF parallel mechanism, extremely fast in-plane motion, ideal for high-speed pick-and-place in horizontal planes (phone screen transfer, flat component sorting). Speed advantage over six-axis cobots in pure horizontal tasks; lower flexibility.
5. Delta/Parallel Cobots: Three to six DOF parallel mechanism, ultra-high speed (picking cycles up to 120–150/min), primarily used in food and pharmaceutical high-speed picking (chocolate sorting, blister pack loading). Speed is their dominant advantage; limited payload (0.5–5 kg) and work envelope.
6. Cobot on AMR (Composite Robots): A six-axis cobot mounted on an autonomous mobile robot (AMR) base, enabling mobile manipulation across the full factory space. Representative: Dobot composite robot series, Youibot hybrid platforms. 2025 H1 Dobot composite robot revenue grew 64.7% YoY — the fastest-growing cobot product category.
7. Clean Room / Specialized Cobots: Designed for semiconductor, pharmaceutical, or food applications with specific compliance requirements: IP65 rating for washdown environments (food); Class ISO 5–7 cleanroom compatibility (semiconductor, pharma); NSF H1 food-grade lubrication certification.
The Four Structural Market Drivers
Four structural forces simultaneously driving cobot market growth in 2025–2026:
First, labor cost restructuring: China's manufacturing sector minimum wage has increased an average 6–8% annually since 2018. Shenzhen's manufacturing worker blended cost (wage + social insurance + housing fund) exceeded RMB 7,000/month in 2025. At this cost level, a RMB 80,000–100,000 cobot with 24-month payback represents roughly 12–14 months of direct labor cost savings per displaced worker per year.
Second, labor supply tightening: China's manufacturing workforce peaked around 2015 and has been shrinking since. The "post-2000" demographic entering the workforce shows a marked preference against factory floor work. Recruitment difficulty for skilled assembly workers in Pearl River Delta factories increased sharply in 2023–2025, with turnover rates exceeding 30% annually in some labor-intensive sectors.
Third, AI-driven capability extension: Vision-Language-Action (VLA) models (NVIDIA GR00T N1 and peers) are beginning to enable cobots to interpret natural language instructions, understand visual scene context, and plan actions without pre-programming. This dramatically expands the range of applicable tasks — from precisely structured (fixed-position assembly) to semi-structured (variable-position sorting, bin picking of mixed SKUs).
Fourth, RaaS democratization: Robotics-as-a-Service (RaaS) converts large upfront capital expenditure (CAPEX) into predictable operational expenditure (OPEX), lowering the financial barrier to trial deployment. China's first open cobot rental platform BOTSHARE launched in December 2025, offering cobot rentals from approximately RMB 2,000–5,000/unit/month.
Chapter 2 Global Landscape and China's Position
Global Market Scale
According to IFR (International Federation of Robotics) World Robotics 2025 report: 64,500 collaborative robots were installed globally in 2024, representing 11.9% of total industrial robot installations (up from 2.8% in 2017). The global cobot market reached approximately USD 3 billion in 2025, with a projected CAGR of 23.1% through 2033 (reaching approximately USD 17.2 billion). Asia-Pacific accounts for over 51% of the global cobot market; China alone accounted for approximately 54% of global industrial robot installations in 2024.
China's domestic cobot market: approximately RMB 2.9 billion in 2025, approximately 49,500 units shipped, YoY growth of 45.59% — the highest growth rate among major global markets.
Universal Robots: The Struggling Pioneer
Universal Robots (UR), founded in Denmark in 2005, commercially launched the UR5 in 2008 — the world's first commercial cobot. Acquired by Teradyne (NASDAQ: TER) in 2015 for USD 285 million, UR built the first genuine cobot ecosystem: the UR+ platform, with over 2,000 certified third-party end-of-arm tooling, vision systems, safety modules, and software plugins.
UR's 2025 performance was disappointing: revenue declined approximately 20%, reflecting continued price pressure from Chinese domestic brands and accelerating market share loss in Asia-Pacific. UR's strategic response: focus on high-load models (UR20, UR30) where domestic competition remains limited, and deepen the UR+ ecosystem moat (service fees, UR Academy enterprise certification).
China's Three-Wave Rise
Wave 1 (2015–2019): Price Disruption. JAKA (2014), Dobot (2015), AUBO (2015) entered the market at 50–70% below UR pricing. Quality gaps were real (no joint torque sensors, basic collision detection, limited software), but price made them viable for cost-sensitive SMEs and educational institutions.
Wave 2 (2020–2023): Technical Catch-Up. The decisive factor: domestic harmonic reducer mass production (Leaderdrive/绿的谐波 reaching scale), reducing harmonic reducer cost 40–60% below Harmonic Drive Japan pricing. Servo systems, encoders, and drive electronics from domestic suppliers reached sufficient quality for industrial cobot applications. Flagship Chinese cobot prices dropped from RMB 150,000–200,000 (2019) to RMB 80,000–120,000 (2023), widening the price gap versus UR from 1.5x to 2.5–3x.
Wave 3 (2024–2026): AI Capabilities + Export Surge. VLA integration for "AI cobots" (JAKA Lumi flagship), composite robots (Dobot AMR+arm), and accelerating exports (50%+ YoY growth in 2024) define this phase. Domestic market share exceeded 90%; Chinese brands are becoming genuine global competitors.
Top 5 Chinese Cobot Brand Rankings (2025)
| Rank | Brand | Revenue 2025E | Key Metric |
|---|---|---|---|
| 1 | Dobot (越疆, 2432.HK) | ~RMB 510M | Global shipment leader, 100K+ cumulative |
| 2 | JAKA (节卡) | ~RMB 620M | Fastest revenue growth, IPO pending |
| 3 | AUBO (遨博) | Est. RMB 200-300M | 50+ country export coverage |
| 4 | Fairino (法奥意威) | Est. RMB 150-200M | AI welding specialization |
| 5 | Han's Robot (大族) | Est. RMB 150M | Delta Group channel advantage |
Export Dynamics
China's cobot exports grew over 50% YoY in 2024; Dobot has led Chinese cobot exports for 6 consecutive years, covering 80+ countries. Key export markets: Europe (especially Germany, Italy, Poland), Southeast Asia (Vietnam, Thailand, Indonesia), Middle East (Saudi Arabia, UAE). JAKA's international financing (SoftBank, Temasek, Aramco Ventures, ADIA) provides strategic market access in the Middle East and Southeast Asia.
Chapter 3 Core Technologies
Joint Torque Sensors: The Safety Foundation
Three approaches to collision detection and force limitation dominate the market:
Current Estimation (Sensorless): Estimates contact force by monitoring deviations in motor current from expected values under nominal load models. Cost: negligible (no additional hardware). Precision: ±5–15 N — sufficient for basic PFL compliance in many applications, but susceptible to measurement noise and temperature drift. Used widely in entry-level domestic cobots.
Internal Joint Force/Torque Sensors (JTS): Strain gauge-based torque sensors embedded in each joint. Force measurement resolution: 0.1–1 N·m at the joint; force precision at the end-effector: ±2–5 N depending on arm geometry. Cost: approximately USD 500–2,000/joint (6–8 joints, USD 3,000–15,000 additional per robot). KUKA LBR iiwa has sensors in all 7 joints, achieving <2 N end-effector contact force detection — the commercial cobot force-control benchmark.
External Six-Axis Force/Torque Sensor (F/T Sensor): A dedicated six-DOF force/torque sensor mounted between the robot flange and the end-effector, measuring forces and torques in all six directions. Force resolution: 0.01–0.1 N; torque resolution: 0.001–0.01 N·m — significantly higher precision than joint-integrated solutions. Primary use: precision force-controlled tasks (polishing, assembly, peg-in-hole insertion) where 1–2 N force accuracy is required. ATI Robotics (US) and Robotiq (Canada) are the dominant suppliers; domestic alternatives emerging from OnRobot-equivalent Chinese startups.
Harmonic Reducers: The National Substitution Pivot
The harmonic reducer is the highest-value-per-weight component in any cobot joint, and historically the most import-dependent. Three critical characteristics make harmonic reducers essential for cobots (versus RV reducers used in larger industrial robot joints):
Zero Backlash: The flexspline maintains continuous mesh with the circular spline via elastic deformation. There is no clearance between mating teeth, meaning the transmission angle error is determined purely by manufacturing precision of the gear teeth, not by clearance accumulation. Typical angular transmission error: <1 arcmin.
High Reduction Ratio in Small Volume: Reduction ratios of 50:1 to 160:1 in a package small enough to integrate into a robot joint (typically 80–180 mm outer diameter). Equivalent RV reducers for the same reduction ratio are significantly larger and heavier.
Compliance: The thin-walled flexspline exhibits slight compliance under torque overload — providing passive shock absorption that protects both the reducer and the sensor in collision scenarios.
Leaderdrive (绿的谐波, STAR Market 688017) is the largest domestic harmonic reducer manufacturer in China. H1 2025 revenue: RMB 251 million (+45.82% YoY). Domestic harmonic reducers now priced 40–60% below Harmonic Drive (Japan) equivalents while meeting specifications for the majority of industrial cobot applications. Leaderdrive achieved the milestone of replacing imported reducers in JAKA Pro, Dobot MG, and AUBO-i series by 2023–2024, with angular transmission error now consistently <1.5 arcmin for these industrial cobot grades.
Servo Systems: Inovance and Hechuan Lead Domestic
Servo systems (servo motor + servo driver + encoder) account for 20–30% of cobot bill of materials cost. Cobot servo requirements differ from standard industrial servo: ultra-compact form factor (must fit within the joint structure), high torque density (maximum torque in minimal volume), low-speed torque smoothness (no cogging below 5 RPM), and high encoder resolution (≥17-bit absolute encoder for precise position control).
Inovance Technology (汇川技术, Shenzhen, 300274) leads domestic industrial servo market share. Its SV360 series and cobot-optimized derivatives supply JAKA, Dobot, and Fairino among others. Hechuan Technology (禾川科技, 688320) has specialized in cobot-optimized servo drivers (A600 series), with form factor and overload characteristics specifically tuned for cobot joint drive demands.
Encoder supply chain: Tamagawa (Japan) and HEIDENHAIN (Germany) 17–23-bit absolute encoders continue to be specified for high-end domestic cobots; domestic encoder manufacturers are closing the gap, with full domestic encoder cobots appearing in production configurations in 2025.
Safety Architecture: ISO 10218-2:2025 Four Collaborative Modes
The newly consolidated ISO 10218-2:2025 (which absorbed the former ISO/TS 15066) formally defines four collaborative operation modes:
Mode 1 – Safety-Rated Monitored Stop (SRS): When a person enters the workspace, the robot stops all motion and enters a safety-monitored stop state; it restarts only after the person leaves. Essentially "collaborative stop, not collaborative motion."
Mode 2 – Hand Guiding (HG): The operator directly grasps and guides the robot end-effector; the robot follows operator-applied forces in real time. Requires precision end-effector force/torque sensing (typically a six-axis F/T sensor). The safety-standard implementation of lead-through teaching.
Mode 3 – Speed and Separation Monitoring (SSM): Monitoring systems (3D cameras, safety LiDAR) track human presence in the workspace. When a person approaches, the robot decelerates to ensure its stopping distance remains less than the separation distance from the person; when the person leaves, the robot resumes normal speed. The most common cobot safety mode in industrial deployment.
Mode 4 – Power and Force Limiting (PFL): The robot limits its joint drive forces and end-effector contact forces to within safety thresholds at all times, derived from biomechanical injury limits for different body parts (e.g., hand: 65–280 N depending on contact region and time). PFL is the fundamental safety property enabling true human-robot co-line operation — the robot is safe regardless of where or how contact occurs.
Vision Systems: Four Generations
Gen 1 (2D template matching, 2010–2018): Camera captures 2D image, OpenCV-based template matching extracts X, Y, and rotation angle (3 DOF). Sensitive to lighting changes and occlusion; unsuitable for bin-picking.
Gen 2 (3D structured light/ToF, 2018–2022): Structured light cameras (Mech-Mind, Orbbec, Intel RealSense) project calibrated light patterns and reconstruct full 3D point clouds, enabling 6D pose estimation (X, Y, Z, roll, pitch, yaw). Made bin-picking commercially viable.
Gen 3 (Deep learning, 2022–2025): YOLO-based detection and PointNet++ point cloud processing dramatically improve robustness to novel objects, clutter, and lighting variation. Few-shot learning allows new object recognition from 10–20 sample images in minutes. Techman's TMflow AI vision (built into each TM robot) represents this generation in domestic cobots.
Gen 4 (VLA end-to-end, 2025–present): Vision is no longer a separate perception module but is deeply integrated with language understanding and action planning in a unified neural network. Each video frame, combined with the language instruction, is directly mapped to joint control commands — no explicit intermediate representation. JAKA Lumi is the leading domestic VLA-integrated cobot commercial product.
Chapter 4 Industry Supply Chain
Value Distribution: The Cobot U-Curve
The cobot value chain follows a characteristic "U-curve" distribution: components (left side) and end-application/integration (right side) capture the most value; middle-tier cobot OEM manufacturing, despite being the most visible segment, faces intense margin pressure from price competition.
Upstream (Components): High Margin, High Barriers
Harmonic reducers (15–20% of cobot BOM cost): Harmonic Drive (Japan, ~60% global market), Leaderdrive (domestic, rapidly gaining share). Key barriers: precision gear grinding, heat treatment process control, high capital equipment intensity.
Servo systems (20–30% BOM): Inovance, Hechuan (domestic); Yaskawa, Panasonic (imported for high-end). Key barriers: motor design capability, DSP/FPGA servo controller development.
Encoders (5–8% BOM): Tamagawa, HEIDENHAIN (imported, high-end); domestic alternatives closing in. Key barriers: precision angle measurement optics and signal processing.
Joint torque sensors (8–15% BOM, if included): ATI, Robotiq (imported); domestic suppliers emerging. Key barriers: precision strain gauge fabrication, thermal compensation algorithms.
Controllers (10–15% BOM): Each cobot OEM develops proprietary real-time controllers. No dominant external supplier — this is a vertically integrated segment.
Mid-Tier (Cobot OEM Assembly): Intense Competition
Approximately 100+ domestic cobot OEM brands, concentrating around a handful of leaders. Gross margins under pressure (Dobot 47.0%, sustainable for leaders; <30% for smaller players). Competitive differentiators: brand ecosystem, AI capabilities, export channels, RaaS platform.
Downstream (System Integration): High Value, Fragmented)
China has approximately 1,500–2,000 cobot system integrators (SI), with fewer than 200 exceeding RMB 100 million in annual revenue. SI gross margins range 25–40% (higher than OEM assembly on comparable value). SI sector faces talent scarcity — an estimated 40,000–80,000 gap in qualified application engineers nationally.
End-of-Arm Tooling (EOAT): Under-Appreciated Value Pool
Global EOAT market approximately USD 3–4 billion, growing 15–20% annually. Robotiq (Canada), OnRobot (Denmark), Schunk (Germany) hold ~55% combined. Domestic alternatives: Dexta Robotics (dexterous hands), DH Robotics (大寰机器人, flexible grippers, >RMB 150M 2025 revenue), JAKA's proprietary end-effectors. EOAT domestic market share approximately 40–50%, lagging cobot OEM domestic share by 3–5 years — representing the next major domestic substitution opportunity.
Chapter 5 Downstream Applications
3C Electronics: The Largest Application Market
3C (computing, communications, consumer electronics) assembly is the largest downstream application for global cobots, and the segment with the highest domestic penetration rate for Chinese cobot brands — estimated at over 35% of China cobot installations in 2025.
Structural fit is exceptional: 3C products typically weigh 20 g to 2 kg (perfectly within cobot payload sweet spots); dimensional tolerances typically ±0.05–0.2 mm (matching cobot precision capabilities); product changeover happens annually or quarterly, requiring flexible cobots with rapid reprogramming capabilities.
Key 3C cobot applications: FPC (flexible printed circuit) cable insertion (force-controlled, requires ±0.3 N contact force precision to avoid pin damage); screw tightening with torque verification and photo confirmation; OCA optical adhesive bonding (force-limited to prevent bubbles); housing assembly (multi-directional gripping with scratch prevention).
Dobot MG series industrial cobots are deployed in Samsung, Omron, and numerous tier-1 contract manufacturers. A typical smartphone factory deploying 32 cobots for FPC insertion (JAKA Pro 7, 3D vision, force-control end-effector) achieved: defect rate reduced from 2.8% to 0.3%; Cpk from 0.9 to 1.67; 48 workers replaced by 32 cobots operating 2 shifts; payback period approximately 26 months.
Automotive Tier 1: High-Payload Cobot Opportunity
Automotive Tier 1 suppliers (not OEM assembly lines, which still exceed cobot payload limits) are the second-largest cobot application segment. Key applications: interior component assembly (instrument panels, door panels, 3–15 kg parts); electronic module assembly (ECU, display modules, ADAS sensors requiring ±0.05–0.1 mm precision); sealing application (door weatherstripping, engine bay sealing, ±0.5 mm trajectory precision); collaborative arc welding (small stampings with AI vision, Fairino FR series).
Tier 1 cobot requirements are more stringent than 3C: IATF 16949 certification, MTBF >30,000 hours, cycle time compliance under safety speed limits. Foreign brands (UR, ABB, FANUC) maintain advantages in established case documentation and audit compliance materials. Domestic brands are closing in through accumulated case evidence.
Food and Beverage: IP65 and NSF H1 as Differentiators
Food industry cobot adoption is growing at >60% annually — the fastest new entrant segment. Compliance requirements create a real technical barrier: IP65 minimum rating (fully dustproof, jet-water resistant for daily factory washdown); FDA 21 CFR 177 or EC 1935/2004 food contact material compliance; NSF H1 food-grade lubricant certification in all joints. Currently approved food-grade cobots are limited: UR e-Series White, FANUC CRX-10iA selected models, Dobot MG food-grade variant (in certification process, expected 2026 completion).
Life Sciences: Precision, GMP, and Stickiest Customers
MegaRobo (镁伽科技) dominates China's smart laboratory cobot segment, serving global pharma companies (J&J, Pfizer, Bayer) and CRO leaders (WuXi AppTec, Pharmaron). Key performance data: liquid transfer precision CV <0.8% across 1–1,000 μL range; throughput scale 8,000+ compounds/day (vs. 300 manually); GMP validation documentation generated automatically; zero audit findings in customer GMP compliance reviews.
The economic logic in life sciences differs from manufacturing: the value driver is not "labor cost saving" (research associates' salaries at RMB 10,000–15,000/month make the math work, but it's not the primary driver). The true driver is: every validated screening compound represents potential drug candidates worth tens of millions of dollars; automation precision and throughput directly impact drug discovery speed and success probability.
Logistics and Warehousing: Composite Robot Inflection Point
Composite robots (cobot + AMR) reached commercial scale in 2024–2025: Dobot composite robot revenue grew 64.7% YoY in H1 2025. The key capability milestones that unlocked commercialization: AMR docking precision ±5 mm (down from ±20 mm in 2020); flexible gripper success rate 95–99%+ for common SKU formats; multi-robot fleet scheduling with collision probability <0.1%.
Composite robot application pipeline: large factory internal logistics (replacing forklifts and AGVs for raw material, WIP, and finished goods movement, with cobot handling the "last meter" precision placement); GMP pharmaceutical in-process transfer (batch-level sample and drug product movement); hospital logistics (medication, consumables, laboratory sample transport — multiple Beijing and Shanghai hospitals already in pilot).
Chapter 6 Key Players
Universal Robots (Denmark) — Ecosystem Leader, Declining Share
Founded 2005, Odense, Denmark. UR5 (2008) was the world's first commercial cobot. Acquired by Teradyne (2015, USD 285M). UR+ ecosystem: 2,000+ certified third-party accessories — the deepest cobot ecosystem globally and UR's most durable competitive moat. Product range: UR3e through UR30 (3–30 kg payload). 2025 revenue declined ~20% amid Chinese price pressure. Strategy: retreat to high-payload (UR20/UR30), ecosystem services, and mature markets (Europe, North America) less sensitive to price.
Dobot (越疆, China Shenzhen) — Shipment Leader, Composite Pioneer
Founded 2015 from Harbin Institute of Technology robotics team. Started with educational cobots (Dobot Magician), building a global dealer network in 80+ countries before transitioning to industrial products. Key 2025 data: cumulative shipments 100,000+; revenue ~RMB 510M (+32.9%); gross margin 47.0%; six-axis cobot revenue grew 46.7%; composite robot revenue grew 64.7% (highest growth product line); export market leader for 6 consecutive years; listed Hong Kong Stock Exchange December 2024 (18C chapter, ticker 2432.HK).
Three competitive advantages: (1) scale-driven supply chain cost efficiency; (2) educational cobot dealer network converted to industrial sales channel; (3) composite robot (AMR + cobot) first-mover position in the fastest-growing product category.
JAKA (节卡, China Shanghai) — Premium Positioning, STAR IPO
Founded 2014, Shanghai. International financing: SoftBank Vision Fund, Temasek Holdings, Aramco Ventures, Abu Dhabi Investment Authority — four sovereign/institutional investors from different countries, total approximately USD 200 million. 2025E revenue ~RMB 620M (2022–2025 CAGR ~30%). 2024 capacity utilization 99.1% (demand far exceeds supply). JAKA Lumi flagship platform: 12 DOF (6-axis cobot arm + 6-axis powered base), multi-modal sensing (RGB-D + 6D force + tactile skin), natural language task interface — the most commercially advanced AI cobot from a domestic brand. JAKA App Store: closest domestic equivalent to UR+ ecosystem concept.
AUBO (遨博, China Beijing) — Full-Stack Self-Developed, Earliest Exporter
Founded 2015, Beijing. Export coverage: 50+ countries — the earliest and widest export reach of any domestic cobot brand, with established European dealer networks. Full-stack self-development strategy: servo motors, drive electronics, controllers, robot operating system (RROS) all developed in-house. Strategic benefits: cost control through internal procurement, system-level performance co-optimization, IP protection. Strategic cost: significant R&D investment required to maintain complete competency across all domains.
ABB YuMi (Switzerland) — Dual-Arm Pioneer, Precision Benchmark
ABB YuMi (IRB 14000), launched 2015, was the world's first commercial dual-arm cobot. Specifications: 7 DOF per arm, 500 g payload per arm, ±0.02 mm repeatability, 38 kg total weight. Unique value: simultaneous two-handed coordination (watch movement assembly, dual-hand PCB connector insertion, precision fastener tightening) — tasks no single-arm cobot can replicate. ABB broader collaborative portfolio: single-arm YuMi (IRB 14050), SWIFTI (4 kg, 4× YuMi speed), GoFa series (5–10 kg, industrial collaborative).
FANUC CRX (Japan) — Shared Ecosystem Integration Advantage
FANUC's CRX series (CRX-5iA through CRX-25iA) shares the R-30iB controller platform and ROBOGUIDE simulation software with FANUC's traditional industrial robots. Zero migration cost for facilities already using FANUC industrial robots — engineers program CRX cobots with the same skills and tools they already know. CRX-25iA (25 kg payload) is among the highest-payload commercial cobots available, targeting heavy assembly tasks traditionally served by medium-load industrial robots. Price approximately USD 43,000–55,000/unit.
KUKA LBR iiwa (Germany) — Force-Control Performance Ceiling
LBR iiwa (intelligent industrial work assistant): 7 DOF, all joints with high-precision integrated joint torque sensors (millinewton-meter resolution), end-effector contact force detection <2 N. Commercial force-control benchmark. Applications: aerospace precision assembly (drilling, riveting, insertion), automotive engine precision component assembly, surgical robotics positioning, high-precision measurement. Price approximately USD 50,000–70,000/unit. Note: KUKA fully acquired by Midea Group (China) in 2017, though brand and technology remain German-based.
Techman (達明, Taiwan) — AI Vision Differentiation, Counter-Cyclical Growth
Techman Robot (TPEX:4585), established 2016. Each TM robot has a built-in industrial camera and AI vision processor. TMflow AI vision software enables "zero-programming visual guidance" — circle a target object's reference image on a touchscreen, the robot learns and starts recognizing and guiding grasps within minutes. 2025 revenue NTD 1.822 billion (USD 56.8M, +23% YoY) — the clearest counter-cyclical outperformer among foreign cobot brands. Full-factory automation business grew from NTD 66M to NTD 280M in 2025 (+4x), signaling a business model shift from hardware to complete line solutions.
MegaRobo (镁伽, China Beijing) — Life Science Specialist, HK IPO Candidate
Life science laboratory automation leader. Products: collaborative robot-centered complete laboratory automation systems, not robot hardware standalone. Competitive moats: deep GMP compliance documentation library, proprietary LIMS/scheduling software, established validation track record at top pharma clients. Total funding: RMB 2.7 billion (8 rounds), valuation >RMB 10.5 billion. HK Main Board IPO application submitted 2025, awaiting listing approval.
Chapter 7 Domestic Substitution and Platform Data Insights
Three Waves: The Evolution of Domestic Substitution Tianxia Gongchang's factory data platform — covering 4.8 million verified Chinese factories in production — shows clear evidence of this domestic substitution pattern.
Wave 1 (2015–2019): Price Entry, Performance Gap. Domestic cobots entered at 50–70% discount to UR. Real technical gaps existed: no joint torque sensors (current estimation only), limited collision detection, closed software ecosystems. Market share <20%, primarily educational and research users.
Wave 2 (2020–2023): Technical Parity in Core Applications. Decisive variable: Leaderdrive harmonic reducer mass production, reducing component cost 40–60%. Domestic servo, encoders, drive electronics reaching industrial cobot requirements. Main domestic cobot prices dropped from RMB 150,000–200,000 to RMB 80,000–120,000. Market share crossed 50% in 2023.
Wave 3 (2024–2026): AI Capabilities + Global Expansion. AI cobot platforms (JAKA Lumi), composite robots (Dobot), and export acceleration (50%+ YoY 2024) define this phase. Domestic market share >90%.
Genuine Advantages vs. Remaining Gaps
Real advantages — domestic cobots now genuinely superior in key dimensions:
Price competitiveness is real and technology-backed, not quality-compromised. JAKA Pro 12 vs. UR10e: JAKA 12 kg payload, ±0.02 mm precision at RMB 100,000; UR10e 10 kg payload, ±0.05 mm at RMB 320,000 China retail. JAKA wins on two performance metrics at one-third the price — the gap comes from supply chain cost, not specification compromise.
Response speed: domestic manufacturers complete custom modification requests in weeks; UR typically requires months. After-sales: leading domestic brands offer 24-hour on-site response in all major industrial cities.
Export competitiveness: Dobot's 6-year export leadership, JAKA's Middle East strategic financing, AUBO's European channel maturity — domestic cobots are genuine global competitors, not just low-price alternatives for Southeast Asia.
Real gaps — high-end force control and brand trust remain:
KUKA LBR iiwa's full 7-joint torque sensor architecture and <2 N force control precision cannot be matched by any domestic product at comparable price points. Aerospace assembly, surgical robotics, and precision instruments remain foreign-dominated niches.
Brand trust in European high-end markets requires years of field case accumulation. New entrants face 3–6 month pilot validation and supplier qualification audit cycles — creating a time and capital moat for established brands.
Platform Data Insights: Demand Patterns Across 4.8 Million Factories
Our platform data — covering 4.8 million verified active Chinese factories across 1,965 manufacturing sub-sectors — reveals several structural patterns of direct relevance to cobot market trajectory:
Geographic demand concentration: Guangdong, Zhejiang, Jiangsu, and Shandong provinces account for approximately 55% of platform factory inventory, matching cobot domestic sales geography almost perfectly. These four provinces are also where manufacturing labor cost increases are fastest — spatial correlation reflecting the causal mechanism of "rising labor costs driving automation adoption."
SME dominance: Over 70% of the 4.8 million platform factories have fewer than 100 employees. This aligns with the industry thesis that SMEs represent the largest incremental cobot demand opportunity — and are precisely the target group most cost-accessible with domestic cobot pricing.
Export-oriented factories show stronger automation propensity: Factories with AEO customs certification and active international buyer relationships show markedly higher automation inquiry frequency and price acceptance. Root cause: European and American buyers increasingly require factory audits assessing automation level and process consistency — making automation a supply chain qualification requirement, not a discretionary upgrade.
Recruitment behavior as automation intent indicator: Factories experiencing sustained increases in hiring for labor-intensive positions (general assembly workers) while simultaneously reporting declining fill rates and retention — indicating they've reached the structural "can't hire, can't retain" inflection point — show cobot purchase intent 2–3× the platform average. This group represents the highest-conversion leads for cobot system integrators.
Chapter 8 Pricing and Business Models
Four-Tier Price Structure
Tier 1 — Educational/Desktop (RMB 5,000–50,000): Dobot Magician Pro (RMB 15,000), Elephant Robotics myCobot (RMB 5,000–20,000). Target: universities, vocational colleges, STEM education, entry-level lab automation. Low payload (0.5–3 kg), basic safety features.
Tier 2 — Light Industrial Cobot (RMB 50,000–150,000): Dobot MG400 (RMB 50,000–80,000), JAKA Zu 3/7 (RMB 80,000–120,000), AUBO-i3/i5, Fairino FR3/FR5. Target: 3C assembly, medical device light assembly, food packaging, SME manufacturing. Payload 3–7 kg, complete safety certification, some with JTS. This price band accounts for ~80%+ of volume.
Tier 3 — Mid/Heavy Industrial Cobot (RMB 150,000–300,000): JAKA Pro 12/16, Dobot CR 10/12 (RMB 150,000–200,000), UR10e/UR16e (RMB 250,000–350,000 in China), FANUC CRX-10iA (RMB 300,000–400,000), ABB GoFa (RMB 250,000–350,000). Target: automotive Tier 1, heavy packaging, logistics composite robots. 10–20 kg payload; JTS increasingly standard; food-grade versions available.
Tier 4 — Precision High-End Cobot (>RMB 300,000): KUKA LBR iiwa 7/14 (RMB 350,000–500,000), ABB YuMi dual-arm (RMB 400,000–550,000), UR20/UR30 (~RMB 400,000–500,000). Target: aerospace precision assembly, surgical positioning, high-precision scientific instruments, highest-grade force-control applications. Full joint torque sensing, <2 N force control precision.
Total Cost of Ownership: The Hidden Majority
Hardware purchase is only 30–50% of true TCO:
End-of-arm tooling: standard pneumatic grippers RMB 3,000–30,000; intelligent electric grippers RMB 20,000–80,000 (Robotiq, OnRobot); specialized tooling (micro-dispensing, precision bonding) can equal robot body cost.
Vision systems: 2D camera + light + software RMB 3,000–15,000; 3D structured light system RMB 10,000–80,000; AI vision platform RMB 30,000–200,000.
System integration fees: 80–150% of robot hardware cost for a standard 3C assembly station; 200–400% for complex multi-robot collaborative lines.
Safety assessment and compliance certification: CE compliance assessment for European market EUR 5,000–20,000; additional EU AI Act compliance for VLA systems (2027+) EUR 100,000–500,000 estimated.
Annual maintenance (7–15 year lifecycle): 3–8% of hardware purchase price annually.
RaaS: The Structural Business Model Innovation
RaaS converts high upfront CAPEX (RMB 500,000–1,000,000 for a complete cobot workstation) to predictable OPEX (RMB 2,000–5,000/month/unit). Global RaaS market: approximately USD 2.4 billion in 2025, ~USD 2.8 billion in 2026, growing ~15% annually. China RaaS CAGR ~24.3%, leading globally. BOTSHARE (China's first open cobot rental platform, launched December 2025) enables SME manufacturers to rent food-grade cobots from RMB 2,000–5,000/unit/month. Dobot's 47% gross margin is partly explained by hardware scale efficiency and partly by the ecosystem value (CoStation cloud monitoring platform, application software, RaaS infrastructure) that justifies premium pricing over pure hardware commodity.
Chapter 9 Case Studies
Case 1: Smartphone Main Board FPC Insertion (3C Electronics)
Background: A South China contract manufacturer producing 8 million smartphones annually for a domestic top-tier brand. FPC insertion defect rate ~2.8% (human), recruitment difficulty (turnover >30%/year, labor cost >RMB 7,000/month).
Solution: 32 JAKA Pro 7 cobots + 3D structured light vision (±0.05 mm) + force-controlled FPC insertion end-effectors (±0.3 N contact force precision). Human-robot co-line mode (no safety barriers).
Results: FPC defect rate 2.8%→0.3%; Cpk 0.9→1.67; 48 workers→4 monitors; ROI period ~26 months.
Case 2: Automotive Instrument Panel LCD Assembly (Tier 1)
Background: Suzhou automotive Tier 1 supplying luxury German OEM. LCD panel assembly: bubble defect rate ~1.5%, high rework cost.
Solution: ABB GoFa 10 (force-controlled LCD bonding, ±0.5 N precision) + Techman TM 14 (built-in vision for AOI + screw torque verification).
Results: LCD bubble defect rate 1.5%→0.05% (97% reduction); cycle time 90s→72s; 6 workers replaced (3 shifts); ROI period ~26 months.
Case 3: Biopharmaceutical HTS Liquid Handling (Life Science)
Background: Top 10 global pharma China R&D center. 300K+ compound screening/year, 8 research associates daily for manual liquid transfer, inter-operator CV 3–5% compromising batch data comparability.
Solution: MegaRobo complete HTS automation platform — 2 precision liquid transfer cobots (8-channel, CV <0.8%), 1 plate transfer arm, GxP-compliant scheduling software.
Results: Daily throughput 300→8,000 compounds (+26×); staff 8→2; CV 3–5%→0.6%; zero GMP audit findings.
Case 4: Food Packaging — RaaS Pilot
Background: Zhejiang bakery, 800 SKUs, seasonal demand 5–8× spike (3,000→20,000 boxes/day peak).
Solution: Rented 12 food-grade cobots via BOTSHARE for 3 months (RMB 3,600/unit/month, IP65, NSF H1). Total cost ~RMB 130,000 vs. ~RMB 1,200,000–1,800,000 to purchase. After successful pilot, signed annual lease for 4 units for regular peak season.
Case 5: Automotive Stampings AI Arc Welding (Fairino)
Background: Zhengzhou automotive stamping parts factory, 200+ SKU types, >90% batches <100 parts. Traditional robot welding changeover: 1–2 days. Manual welding: skilled welder shortage (>RMB 10,000/month, severe consistency variation).
Solution: Fairino FR10 AI welding cobot with 3D vision-based auto programming — no manual teaching required; new part setup time 15 minutes.
Results: Changeover time: 1–2 days → 15 minutes; weld seam consistency σ reduced 60%; minimum economically automatable batch size: 500 → 30 parts; monthly order capacity +80%.
Chapter 10 Investment and M&A
Global Investment Scale
2025 global robotics VC/strategic investment exceeded USD 6 billion — double the 2022 level. China's cobot/embodied intelligence segment: 20+ funding events monthly in Q1 2026. Key financing events:
JAKA (节卡): Series E RMB 400M (USD 56M), led by SoftBank Vision Fund, followed by Temasek, Aramco Ventures, ADIA. Total raised ~RMB 1.4B (USD 200M). STAR Market IPO in review process 2025.
MegaRobo (镁伽): 8 rounds, RMB 2.7B cumulative, valuation >RMB 10.5B (~USD 1.46B). Strategic investor WuXi AppTec. HK Main Board IPO application submitted 2025.
Dobot (越疆): Hong Kong Main Board IPO December 2024 (18C chapter, ticker 2432.HK). Market cap at listing ~HKD 6B.
Fairino (法奥意威): Series C 2025 (Minsheng Equity, RongXiang Venture Capital). 500+ customers, AI welding specialization.
Leaderdrive (绿的谐波, 688017): STAR Market listed, H1 2025 revenue +45.82%, 2025 market cap ~RMB 15B. As the harmonic reducer supplier to >80% of domestic cobots, its financials are the clearest proxy for cobot sector volume growth.
Capital Market Dual Window
Hong Kong Main Board (18C chapter): Allows pre-profit tech companies. Dobot (listed), MegaRobo (applied). International investor access, global brand signaling.
Shanghai STAR Market: For high-tech domestic manufacturers, allows loss-making periods. JAKA IPO pending. Domestic institutional investor base, supports government procurement credibility.
M&A Signals
Harmonic reducer consolidation: Leaderdrive's strategic expansion toward integrated joint modules (harmonic unit + encoder + torque sensor) signals potential value chain vertical integration — from pure component supply toward cobot joint unit supplier, compressing space for some integrators.
Global acquisition hypothesis: Teradyne (UR parent) losing China market share. Acquisition of a domestic cobot brand remains a frequently discussed scenario in industry circles; no concrete announcement yet, but precedent from other tech manufacturing sectors exists.
Chapter 11 Policy and Standards
ISO 10218:2025 — Landmark Consolidation
The 2025 revision of ISO 10218-1 and ISO 10218-2 is the most significant safety standard update since the 2011 edition. Key structural change: ISO/TS 15066:2016 (the dedicated collaborative robot technical specification) has been fully absorbed into ISO 10218-2:2025 main body, elevating collaborative robot safety requirements from "technical specification" to "international standard" legal status. This increases regulatory enforcement weight and procurement compliance requirements in certification-sensitive markets (Europe, Japan, automotive supply chains).
Risk assessment requirements under ISO 10218-2:2025 are significantly strengthened: integrators must provide field-verified contact force measurements (not just manufacturer specification claims), verified stopping times and distances from actual tests, hazard identification across a standardized checklist of ≥25 hazard categories, and residual risk acceptability justification.
China National Standard 36008-2018 and Ongoing Revision
China's domestic collaborative robot standard (national standard 36008-2018) — one of the world's first national-level cobot standards — is currently being revised following the ISO 10218:2025 publication. Updated version expected 2026–2027, aligning with ISO 10218-2:2025's PFL/SSM/HG/SRS four-mode framework and incorporating updated contact force measurement methodologies.
The terminology standard series (national standard 12643) underpins all Chinese robot standards, defining standard terms for degrees of freedom, rated payload, repeatability, workspace, and other core technical parameters used in procurement specifications and compliance documentation.
MIIT "Robot+" Action Plan
Ministry of Industry and Information Technology 2023 "Robot+ Application Action Plan": 10 priority application domains; target of doubling manufacturing robot density (from 322 units/10,000 workers in 2021 to 600+) by 2025. Actual 2025 achievement: approximately 470–480 units/10,000 workers — roughly 75% of target, setting the stage for extended policy support under the 14th Five-Year Plan sequel (2026–2030) with stronger SME-focused incentives expected.
Provincial subsidies (Guangdong, Zhejiang, Shanghai) provide 10–30% equipment cost subsidy for qualifying manufacturing enterprises purchasing cobots, stacking with national-level incentives.
Elderly Care Robot 3-Year Pilot (2025–2027)
MIIT + Ministry of Civil Affairs joint policy: deploy cobots in 100+ elderly care institutions (bathing assistance, meal assistance, rehabilitation, companionship); central government subsidy RMB 200,000–500,000 per institution; all deployed cobots must meet national standard 36008-2018. Market significance: not the 2025–2027 deployment volume (500–1,000 units), but the first-round technology validation enabling the 2028–2030 commercial scaling. Elderly care cobot mass market expected to materialize 2028–2030.
Export Control and Trade Environment
US tariffs on Chinese-made robots (25% in current framework) significantly compress direct export economics to the US market. Chinese brands' response: prioritize Europe (3–5% tariff), Southeast Asia (no material tariff), and Middle East. Domestic assembly of imported Chinese components in third-country (Southeast Asia) facilities remains a compliance-risk strategy under tightening US origin rules.
EU AI Act (effective 2027): VLA cobots may qualify as "high-risk AI systems," requiring transparency, human oversight capability, adversarial robustness testing, and technical file registration. Estimated additional compliance cycle 6–18 months; third-party audit cost EUR 100,000–500,000 for Chinese exporters targeting European VLA cobot markets.
Chapter 12 Trends and Research Institute Judgments
Trend 1: AI Cobot — Qualitative Product Transformation Underway From the perspective of Tianxia Gongchang's research team, this trajectory is consistent with the broader pattern observed across other strategic equipment categories.
The most profound industry trend in 2025–2026 is not revenue growth (quantitative) but the qualitative product transformation from "programmed tool" to "intelligent agent" enabled by VLA models.
VLA's revolution: robots shift from "executing pre-programmed sequences" to "understanding instructions and planning actions." NVIDIA GR00T N1 (March 2025) provides the most commercially influential VLA framework: dual-system architecture (slow-thinking System 2 for language understanding and action planning; fast-execution System 1 for real-time control), heterogeneous embodiment learning (skill transfer across different robot platforms), and FoundationPose (6D object pose estimation).
JAKA Lumi: the leading domestic VLA-integrated commercial cobot (12 DOF, multi-modal sensing, Chinese natural language task interface). Factory pilot validation completed 2025; commercial launch expected 2026.
Research institute judgment: VLA commercialization window opens 2026–2028, entering "semi-structured environments" (variable-position workpieces on fixed worktables — harder than pure structured assembly where VLA adds less, easier than fully unstructured service scenarios where current VLA capability is insufficient). This is the VLA cobot commercial "sweet spot."
Trend 2: Composite Robots — Mobility Extending Cobot Scope
Dobot composite robot H1 2025 revenue +64.7% YoY — fastest-growing product category in cobots. AMR navigation precision now ±5 mm (from ±20 mm in 2020); flexible gripper success rate 95–99%+; multi-robot fleet scheduling reliability ≥99.9%.
Next composite robot wave: large-factory internal logistics (replacing forklifts and AGVs); GMP pharmaceutical in-process transport; hospital logistics (multiple Beijing/Shanghai Tier-3 hospitals in pilot). Projected composite robot share of new cobot installations: ~30% by 2028 (currently ~8–10%).
Trend 3: Export Acceleration — Entering Global Competitive Reckoning
China cobot exports +50%+ YoY in 2024. Structural drivers: absolute price gap sufficient to generate attractive margins for international integrators (overseas pricing 1.5–2× domestic; domestic brands in Europe price at 50–60% of UR equivalents); channel maturity accumulation now bearing fruit; Made-in-China manufacturing automation brand acceptance rising rapidly in Southeast Asia and Middle East. Projection: domestic brands achieve 30% non-China global market share by 2028–2029 (from ~15% in 2025).
Trend 4: RaaS Reshaping Market Structure
RaaS penetration will reach 25% by 2029 (from ~5% in 2025). Winners: cobots OEMs that build genuine remote monitoring, predictive maintenance, and task optimization platforms — enabling economically viable RaaS without proportional field service headcount growth. Losers: hardware-only mid-tier OEMs without software infrastructure, facing margin compression from RaaS platform competition by larger players.
Five Research Institute Judgments
Judgment 1: China cobot market 2026–2028 CAGR: 35–45%. Global: 20–25%. Structural foundation: rising labor costs + falling cobot prices + expanding AI-enabled application scope.
Judgment 2: Domestic cobot share in China >90% maintained. Foreign brands survive only in "highest precision force-control niches" (KUKA LBR iiwa tier) and "ecosystem-locked installed base" (existing UR+ ecosystem sites with prohibitively high switching costs).
Judgment 3: VLA will create a new "AI cobot" product category by 2027. The winner will be the OEM with the strongest software and AI engineering capabilities — not necessarily the current shipment leader.
Judgment 4: Cobot industry consolidation wave 2027–2030: top 3–5 firms controlling >80% of China market. RaaS service capability requirements, VLA R&D investment requirements, and export compliance requirements will create capital barriers that eliminate most of the current ~100+ small OEMs.
Chapter 13 Risks
Risk 1: Harmonic Reducer Supply Concentration
Harmonic Drive (Japan) controls ~60% of global high-precision harmonic reducer supply. Leaderdrive (China) leads domestic production but faces capacity elasticity risk during demand surges (18–24-month production line expansion lead time). Potential geopolitical supply disruption (US pressure on Japan for export controls) remains a low-probability but high-impact tail risk for Chinese OEMs dependent on imported reducers.
Risk 2: Price War and Margin Compression
Light industrial cobots (3–7 kg) price floor has reached RMB 40,000–50,000 (2025), projected to fall below RMB 30,000 by 2027. Second-tier OEMs selling below cost to capture share. Gross margin divergence between leaders (47%+) and followers (<25%) will accelerate industry consolidation. Leaders' moat: scale-driven supply chain cost efficiency, product mix upgrade (composite robots, AI cobots, large-payload models at higher margins), and growing software/RaaS revenue.
Risk 3: VLA Technology Route Risk
Multiple parallel VLA frameworks (GR00T N1, Google RT-X, MIT ACT, Stanford Diffusion Policy) with no dominant winner yet — deep investment in one route risks obsolescence if the alternative proves superior. Training data scarcity: Tesla Optimus (with massive internal factory operation data) and US-based embodied AI startups (Figure AI, 1X) lead in VLA training data volume; Chinese OEMs face systematic lag risk. AI inference cost: current GPU cost (RMB 3,000–10,000/robot for inference hardware) is 30–100% of low-end cobot purchase price — will only reach acceptable ratio with 3–5 years of AI chip cost reduction.
Risk 4: Safety Incident and Brand Reputation
A serious human-robot contact injury would fundamentally damage the "safe co-line" commercial narrative. Root causes of historical incidents: safety parameter tampering (operators raising PFL thresholds to improve throughput); uncertified end-effectors (sharp edges not covered by robot body certification); workspace layout changes invalidating original risk assessment; inadequate operator training. Mitigation: password-protected safety parameters with change logs; mandatory EOAT system-level risk assessment; workflow-triggered reassessment requirements for layout changes; operator training and certification records mandatory under ISO 10218-2:2025.
Risk 5: International Compliance Barriers
EU AI Act (2027): VLA cobots likely classified as "high-risk AI systems," requiring transparency, human oversight, robustness testing, and technical file registration — creating a 6–18 month additional compliance cycle and EUR 100,000–500,000 audit costs for Chinese exporters. US 25% tariffs compress direct export economics. Domestic assembly in Southeast Asia as tariff mitigation faces increasing US origin rule scrutiny.
Data Sources
This report draws on primary industry data, corporate public filings, authoritative third-party research, and manufacturing platform data. Primary sources: Tianxia Gongchang Factory Database — covering 4.8 million verified Chinese factories in production, used as the primary cross-verification source for company-level claims in this report.
Manufacturing Platform Data: Our platform covers 4.8 million verified active Chinese manufacturing factories across 1,965 sub-sectors, data current to June 2026. First-party data on factory scale distribution, geographic concentration, industry classification, recruitment behavior, and automation inquiry frequency provides the empirical foundation for the demand analysis throughout this report. Website: www.tianxiagongchang.com.
Corporate Public Filings: Dobot (越疆) HKEX IPO prospectus (2024) and H1 2025 earnings announcement; JAKA (节卡) STAR Market prospectus (2025); MegaRobo (镁伽) HKEX listing application (2025); Leaderdrive (绿的谐波, 688017) H1 2025 earnings; Techman (達明, TPEX:4585) FY2025 revenue filing; Teradyne FY2025 Annual Report (UR segment data).
International Research: IFR World Robotics 2025; Grand View Research Collaborative Robot Market 2024–2033; MarketsandMarkets Collaborative Robots Market; Precedence Research RaaS Market 2024–2033.
Domestic Research: MIR (睿工业) 2025 Global Collaborative Robot Industry White Paper; Chinese Electronics Society China Robot Industry Development Report 2025; Qianzhan Industrial Research Institute China Collaborative Robots Industry Analysis 2025.
Standards: ISO 10218-1:2025 and ISO 10218-2:2025 (consolidated, absorbing former ISO/TS 15066); China national standard 36008-2018 "Collaborative Robots"; China national standard 12643 series (Robot terminology).
Policy Documents: MIIT "Robot+ Application Action Implementation Plan" (2023); MIIT + Ministry of Civil Affairs "Elderly Care Robot 3-Year Pilot Plan 2025–2027" (June 2025); provincial cobot subsidy policies (Guangdong, Zhejiang, Shanghai).
English Media and Technical Sources: ANSI official blog (ISO 10218:2025 analysis); TechNode (BOTSHARE platform, December 2025); Taipei Times (Techman 2025 revenue); Standard Bots industry blog (KUKA/FANUC comparison); Springer Nature academic papers on VLA human-robot collaboration; NVIDIA developer blog (GR00T N1 technical documentation); ITCILO research report on AI and robotics labor market impact.
Data current as of June 2026. Market projections are based on publicly available information and are for research reference only; they do not constitute investment advice. Manufacturing platform data is sourced from the platform's proprietary database and has been anonymized and aggregated for research analysis.