JD.com launched its JD Physical AI Acceleration Plan on Sept. 9 at the 2026 JD Global Technology Explorer Conference, targeting six areas for embodied robotics: data, mechanics, manufacturing, sales, logistics and services. The company plans to collect more than 10 million hours of real-world human-scenario data within two years and combine it with internet, simulated and real-robot data to develop an integrated understanding-generation-action model. Over five years, JD.com plans to establish 80 robotics bases across China, procure more than 3 million robots, 1 million autonomous vehicles and 100,000 drones, and build a full-lifecycle service platform. By 2028, it plans to invest more than CNY 10 billion, approximately $1.5 billion, to help 100 brands achieve independent annual sales above CNY 1 billion each. JD Cloud has partnered with Moore Threads and other companies on a domestically produced 10,000-card computing cluster and plans to expand that cooperation to a 100,000-card full-function GPU cluster serving industry-wide model training, inference and embodied-intelligence applications. Moore Threads reported first-half revenue of CNY 1.736 billion, up 147.42% year on year, with cloud products contributing CNY 1.693 billion, or about 97.5% of total revenue. JD.com's system-wide R&D investment rose 53.2% year on year in the first half of 2026, while analysts said logistics data and deployment infrastructure could support the strategy but technology maturity, costs and operating efficiency remain constraints on commercialization.