JD.com Launches Physical AI Acceleration Plan to Scale Embodied Robotics
At its annual tech conference (JDD), JD.com officially announced the 'JD Physical AI Acceleration Plan' aimed at building global physical-world operational capabilities for embodied robotics. The strategic roadmap targets six major areas: embodied data training, industrial bases, core hardware manufacturing, retail sales channels, large-scale logistics deployment, and technical services. This initiative represents a massive effort by an e-commerce and supply-chain giant to shift AI competition from software-only models to physical infrastructure. By leveraging its vast warehouse and fulfillment network, JD.com seeks to accelerate real-world data collection and drastically reduce hardware manufacturing costs for the entire robotics industry. JD.com aims to collect over 10 million hours of real human action data within two years and establish 80 industrial robotics bases across China in five years. Over the next five years, the company plans to procure over 3 million robots, 1 million autonomous delivery vehicles, and 100,000 drones, alongside committing over 10 billion RMB by 2028 to boost robotics retail distribution.
## BACKGROUND
Physical AI (or Embodied AI) refers to artificial intelligence integrated directly with hardware systems—such as humanoid robots, quadruped robots, or autonomous vehicles—that observe and interact with the physical world. Unlike traditional digital AI models that process text or code, embodied AI relies on multimodal sensor inputs and operational data from physical scenarios to perform real-world perception, planning, and task execution.