JD Cloud Plans 100,000-GPU Domestic AI Computing Cluster Powered by Moore Threads

On September 9, 2026, at the 2026 JD Global Technology Explorer Conference, JD Cloud announced plans to build a 100,000-GPU full-function cluster, creating an ultra-large-scale domestic intelligent computing infrastructure. The cluster will use Moore Threads full-function GPUs as its computing foundation, focus on key areas such as large-model training, inference, and embodied intelligence, and open computing capacity to the entire industry to help accelerate the industrialization and scaling of physical AI. This marks the first time domestic GPUs have entered the 100,000-card core intelligent computing cluster of a leading AI cloud enterprise.

The Ministry of Industry and Information Technology's "15th Five-Year Plan for the Development of the Information and Communication Industry" calls for the orderly deployment of 10,000-card and 100,000-card-or-larger intelligent computing clusters and greater efforts to adapt domestic computing chips. Under JD Cloud's plan, future resources will be tilted toward supply-chain AI, embodied intelligence, integrated large-model training and inference, and industrial scenario implementation.

Moore Threads has already achieved commercial deployment of 1,000-card and 10,000-card large-scale clusters under a single network and has made breakthroughs in core training scenarios including foundation models, embodied brains, and world models. In this collaboration, the two parties will carry out full-stack coordination across chips, cloud platforms, and model training, supporting the evolution and iteration of JD's full JoyAI large-model family, accelerating model training, simulation, and high-quality data generation, and building a complete "data → training → simulation → deployment" loop.

JD Cloud is JD Group's core brand for delivering technology and services to government and enterprise clients, building a full-stack AI infrastructure spanning AI cloud, large-model services, and agent applications. Moore Threads, centered on full-function GPUs, provides accelerated computing infrastructure and one-stop solutions.