JD.com used its Global Technology Explorers Conference on 9 September to set out a five-year Physical AI Acceleration Plan, targeting the procurement of 3 million robots, 1 million autonomous vehicles and 100,000 delivery drones across its logistics network, according to AI News.
Robots already in warehouses
JD Logistics, the group's delivery and warehousing arm, presented an industrial robot line called the Wolf Robot series, alongside a LangzuTech Goods-to-Person system, which brings shelves and inventory to warehouse workers rather than sending workers to the shelves, and a LangzuTech Packer robotic arm. The Goods-to-Person system is already running in more than 30 warehouses in China.
JD said it operates unmanned delivery vehicles across more than 20 Chinese provinces and runs over 100 domestic drone routes, including a network serving 78 administrative villages in Zizhong. Its existing footprint includes more than 1,800 self-operated warehouses and over 2,000 third-party cloud warehouses, together covering more than 36 million square metres. Maintenance currently runs through eight robot repair centres.
AI routing, chips and R&D spending
The company said its Meta Brain 3.0 logistics system can calculate optimal routes for hundreds of millions of parcels in seconds, against minutes previously. JD is planning a 100,000-GPU cluster with the Chinese chip designer Moore Threads, and has gathered 10 million hours of video over two years to train embodied-AI models. JD Logistics reported RMB2.3 billion of research and development spending in the first half of 2026, up 23.7% year on year, alongside RMB2.6 billion of depreciation and amortisation, up 18.7%, and RMB3.09 billion of property and equipment purchases.
What it means for couriers
The plan's labour implications are explicit. JD employs roughly 700,000 delivery and logistics staff, and founder Richard Liu has said robots would eventually take over parcel-delivery work now carried out by human couriers. The company says it will create more than 100,000 robotics service engineer roles and has signed retraining agreements with more than 120 educational institutions. It is also sizing the prize: JD puts China's instant-retail market at RMB1.2 trillion by the end of 2026.
JD has not disclosed how many robots are already deployed against the 3 million target, what the programme will cost, or what return it expects, and none of those figures were independently available at publication. A DigiTimes report on the same announcement, published on 9 September, sits behind a paywall; its headline frames the robot target as a route into physical AI rather than a logistics story, which is also how JD presented it on stage.