At the JDD Global Technology Exploration Conference today, JD.com's Chief Technology Officer and President of JD Cloud, Cao Peng, highlighted a pivotal industry shift. He stated that the race in physical AI is moving from isolated advancements in models and hardware to a more integrated, systematic competition, with the battleground centering on computational power, data, and real-world applications.
Elaborating on JD.com's strategic approach to AI, Cao Peng emphasized their unique, hands-on philosophy. He noted that extensive internal discussions have consistently led to the conclusion that JD.com's AI development is not about theoretical papers but about practical refinement, honed through millions of individual transactions across their entire supply chain, from production lines to warehouses and delivery stations.
To advance this vision, JD.com has officially launched the "JD Physical AI Acceleration Plan." This initiative is focused on building a world-leading, large-scale operational capability concentrating on six key areas of embodied robotics: data, machinery, manufacturing, sales, logistics, and services.
In the data sector, JD.com aims to create the world's largest embodied data training center. Their ambitious plan involves collecting over ten million hours of real-world human scenario data within two years. This will be combined with extensive web data, simulation data, and real robot data to construct an integrated embodied intelligence model that unifies "understanding, generation, and action," cultivating globally competitive general-purpose capabilities.
On the infrastructure front, JD.com aspires to build the world's largest robotics industry base. Over the next five years, they plan to establish 80 bases nationwide. These facilities will offer a comprehensive, full-lifecycle service platform for robots, integrating display, delivery, maintenance, upkeep, R&D, design, assembly, data collection, manufacturing, and iterative upgrades.
In manufacturing, JD.com's goal is to become the world's largest supplier of core robotic components. Over the next three years, through standardization and economies of scale, the company will drive product clustering and cost optimization. This includes investing in core components, building proprietary factories, and supporting 100 OEMs in reducing costs across all product categories, while also helping 100 component suppliers double their business performance.
Leveraging its robust distribution channels for sales, JD.com targets becoming the world's largest retail channel for robots. By 2028, they will invest over 10 billion yuan in resources to assist 100 brands in achieving individual sales exceeding 1 billion yuan, facilitating the integration of robots into one million end-user scenarios.
In logistics, JD.com intends to create the world's largest cluster of embodied robot applications. Already, they have deployed the most extensive and diverse joint-development cluster in the supply chain sector, covering the entire chain of warehousing, sorting, transportation, and delivery. The plan for the next five years includes procuring over 3 million robots, 1 million autonomous vehicles, and 100,000 drones to further automate and upgrade the entire logistics process.
Finally, in services, JD.com aims to establish the world's largest embodied robot service platform. Within five years, they plan to build a global after-sales and service network covering over 100 countries, which will create more than 100,000 new jobs for robot after-sales service engineers.
Cao Peng concluded by saying that with the gradual realization of these goals, JD.com will drive AI from diverse industries into millions of households, build the world's largest physical world collaboration system, and ultimately become a leading global physical world operation center.