Huawei, Alibaba and Baidu jointly unveiled supernodes: computing power up to 1024 GPUs, support for inference of models with 10 trillion parameters

Huawei, Alibaba and Baidu have jointly announced the development of supernodes that integrate computing power of up to 1024 graphics processing units (GPUs). These supernodes are designed to support the inference of AI models with up to 10 trillion parameters, marking a significant leap in large-scale AI capabilities. The collaboration between the three tech giants aims to advance the deployment of massive AI models in China, addressing the growing demand for high-performance computing infrastructure.