Moore Threads Packs 256 GPUs Into the MTT C256, Targeting China's Inference Cluster Market摩尔线程发布MTT C256系统，单机集成256颗GPU瞄准国内推理集群

Moore Threads, the Beijing-based GPU developer, announced the MTT C256 on Tuesday — a high-density compute system integrating 256 of its proprietary GPUs into a single chassis. The announcement came during the tail end of WAIC 2026 in Shanghai, positioning the product as a large-scale inference platform aimed at Chinese cloud operators.

The MTT C256 is the company's densest configuration to date and is targeted at the inference workloads that dominate Chinese AI deployment — particularly serving large open-weight models such as DeepSeek V4 and Kimi K3. Moore Threads has been one of the domestic GPU makers that Chinese regulators are promoting as alternatives under the domestic-substitution push accelerated by US export controls.

No independent benchmark results for the MTT C256 had been published by Wednesday. The company did not disclose pricing or availability timelines. Moore Threads competes domestically with Cambricon and Biren, and its trajectory is being watched as a proxy for how quickly China's homegrown GPU ecosystem can reach inference-grade scale.