JD.com expands physical AI in logistics with 3 million robots

Details

JD Logistics announced its Physical AI Acceleration Plan at JDDiscovery 2026 in Beijing, reaffirming a five-year procurement target of 3 million robots, 1 million autonomous vehicles, and 100,000 delivery drones. The company simultaneously launched its Wolf Robot series — purpose-built systems covering warehousing, sorting, transport, and delivery, including cold-chain units rated for -20°C environments and automated pharmacy dispatch. The plan builds on existing infrastructure: as of June 30, JD Logistics operates more than 1,800 self-operated warehouses and 2,000+ third-party cloud warehouses across 36 million square metres. Thousands of unmanned ground vehicles are already active across 20+ Chinese provinces, and 100+ drone routes handle parcels, food, emergency medicine, and disaster relief. The LangzuTech Goods-to-Person system is deployed in 30+ Chinese warehouses, with live deployments in the UK and Germany. The AI backbone is Meta Brain 3.0, which calculates optimal routes for hundreds of millions of parcels in seconds, compared with minutes under the prior version. The LangzuTech Packer robotic arm pairs Meta Brain with multimodal sensor data and parallel reinforcement learning trained in simulation to optimize parcel placement. As of June 30, the Packer ran 24/7 at multiple JD logistics parks. On compute, JD Cloud is partnering with Chinese chipmaker Moore Threads to build a 100,000-GPU cluster for large-model training, underscoring that JD is investing in the foundation models powering its physical AI stack — not just the hardware.