Overview
On August 23, at the closing day of WRC 2026 (August 19–23), JD.com held a themed robotics event attended by CAS academician Qiao Hong, Tsinghua professor Zhang Tao, and representatives from dozens of leading firms including Dex Robotics, EngineAI (众擎), Pascal (帕西尼), Curv Intelligent, Rokae, Galaxea (星海图), and Songyan Dynamics. JD launched three programs simultaneously:
1. JD Embodied AI Industry-Education Co-Creation Plan — joint talent training between universities and top companies, plus data collection, model training, and real-scenario application validation. 2. Robot Components Industry Development Alliance — supply-chain coordination across structural parts, joint modules, electric drives, and sensors. 3. JD Group Robot Strategy Upgrade and Industry Acceleration Plan — the umbrella commitment described below.
This is the first time JD has disclosed an overall strategy treating robotics as an independent first-tier business — upgrading from "channel + sales platform" to supply-chain infrastructure + service network + data foundation + ecosystem alliance.
Key Commitments
| Metric | Value | Timeframe | Significance | |--------|-------|-----------|--------------| | Resource investment | ~10 billion RMB | 3 years (to 2028) | Help 100 brands each exceed 1 billion RMB in sales | | Data collection | 10 million hours of real-scenario data | 2 years | For JD Cloud model training | | After-sales network | 100+ countries | 5 years | 100,000 robotics service engineer jobs | | RoboBase hubs | 80 | same period | Full-lifecycle robotics service platforms | | JoyInside integration | 10M+ endpoint devices | Within the year | Including robot products | | Brand partnerships | 200+ on self-operated channels | Existing | Plus 50+ Chinese robot brands entering Europe |
- 10 billion RMB / 3 years — one of the largest single-category commitments in JD's history; the post compares it to roughly half the first-day market cap of Unitree at its STAR Market IPO on August 19 (~61 billion RMB at an issue price of 150.80 yuan).
- 10 million hours of real-scenario data — ≈416,667 days, or roughly 1,000 robots running a full year. JD stressed this is real-scenario data from retail, logistics, warehousing, and home services — not simulation or crowdsourced recording — feeding the JD Cloud training data foundation.
- 100 countries / 100,000 engineers — extending JD's "120 service" standard (1-minute response, 2-hour on-site, same-day resolution) from logistics to robotics.
- 400+ humanoid robot models in China in H1 2026;
- Primary-market embodied AI investment exceeded 90 billion RMB — up 5x year-over-year and 1.6x all of last year;
- JD's humanoid robot sales up 9x YoY; exoskeleton and other robot products up 8x.
- Robot OEMs: JD's resources, channels, and hubs are the most complete "sell it + keep it working" infrastructure before 2027.
- Component makers: the alliance is the lowest-threshold route to JD's channels and endorsement.
- Model developers: watch the openness and pricing of JD Cloud's 10M-hour data foundation — potentially the industry's largest real-data pool.
- Going global: JD Europe's Joybuy partnership model (50+ Chinese brands) is the most realistic path into Europe.
- Services/training: 100,000 service-engineer jobs are one of China's largest talent gaps of the next three years; university co-training is the entry point.
- Tencent News / IT Times Network, "JD launches embodied AI plans to push robot adoption" (8/23)
- Eastmoney data center robotics industry research (8/24)
- TMTPost on JD Industrials' WRC components showcase (8/22)
- Zafar Securities on JD as WRC2026 global strategic partner (8/20)
- Luming New Finance on Q1/T1 consumer-grade pre-orders (8/24)
- IPO Early on Robot Era's Chen Jianyu at the commercialization inflection (8/23)
- EO Intelligence on Fusion's R1 full-stack compute platform (8/23)
- CNR on Runke's embodied AI 2.0 heterogeneous collaboration demo (8/23)
Three Ecosystem Moves
1. 80 RoboBase hubs — domestic full-lifecycle platforms for robot sales, maintenance, upgrades, and training; JD repositioning from e-commerce platform to "robot 4S store + after-sales + upgrades" infrastructure. 2. Industry-education co-creation — with universities (e.g., Tsinghua) and companies including Dex, EngineAI, Pascal, Curv, Rokae, Galaxea, and Songyan Dynamics, betting that robot engineer supply becomes a bottleneck within 3–5 years. 3. Components alliance — a signal that China's embodied AI competition is shifting from whole machines to the underlying component supply chain.
Industry Context
Citing MIIT data:
The author argues the industry has moved past the tech-demo stage into commercial mass production, making a 10-billion-RMB investment thesis viable at this point.
Positioning in the August Embodied-AI Stack
| Layer | August representatives | JD's Aug 23 move | |-------|------------------------|------------------| | Models | Galbot, EngineAI Awaken, EngineAI 5-layer architecture | JoyInside general embodied intelligence | | Data | EgoSuite 100K, HiPHI 617.5 hours | 10M hours real-scenario data | | Hardware | Q1/T1, Unitree GD01, Y-Hand M2, Fusion R1 PRO | Components alliance + 200+ brands | | Deployment | RoboBase data foundation | 80 RoboBase hubs | | After-sales | JD self-operated logistics | 100 countries, 100k engineers |
The argument: JD fills the long-neglected "deployment + after-sales" layers, leveraging a decade of capabilities in 3C, appliances, and logistics.