JD Cloud to Build a 100,000-GPU Cluster with Moore Threads for AI Infrastructure

HuoXing Finance reports that on September 9, at the 2026 JD Global Tech Innovators Conference, JD Cloud announced its intention to collaborate with Moore Threads to build a 100,000-GPU full-function cluster, creating a large-scale domestic intelligent computing infrastructure. The cluster will focus on large model training, inference, and embodied intelligence, and will open its computing power to the entire industry. The two parties will engage in full-stack collaboration across chips, cloud platforms, and model training to support the iteration of JD’s JoyAI large model, accelerating model training, simulation, and high-quality data generation, with a primary focus on supply chain AI, embodied intelligence, integrated training-inference large models, and industrial applications. Previously, JD Cloud, together with Moore Threads and other partners, built a domestic 10,000-GPU cluster. This new plan for a 100,000-GPU cluster signifies an expansion of their collaboration from model development and inference platforms to ultra-large-scale computing infrastructure. Moore Threads stated that the 100,000-GPU AI factory will cover scenarios such as large model training, token generation, and agent training, serving multiple industries. Moore Threads has continued making progress in commercialization: in the first half of this year, the company achieved revenue of RMB 1.736 billion, a year-over-year increase of 147.42%, with cloud product revenue reaching RMB 1.693 billion, accounting for approximately 97.5%. The company has already deployed a 10,000-GPU intelligent computing cluster and is actively advancing the development of GPUs, interconnect networks, software stacks, and computing resource scheduling platforms required for the 100,000-GPU cluster.