China unveils first 100,000-GPU AI supercluster

At the 2026 National High-Performance Computing Academic Conference in Shenzhen, China unveiled its first fully domestic 100,000-GPU AI supercluster, marking a major milestone in computational infrastructure. The system, named 'Shuguang 8000,' integrates supercomputing and AI computing on a single platform, enabling both high-precision scientific simulations and large model training. The cluster's power is said to be equivalent to the collective computational effort of humanity over 200 years, supporting over 300 tasks across 26 fields, including materials science and pharmaceutical research. The system's key innovation lies in its ability to merge supercomputing, which models complex physical phenomena through equations, with AI computing, which derives patterns from data. While supercomputing simulates the future step by step, AI computing relies on data correlations, which can lead to inaccuracies. The 'Shuguang 8000' aims to bridge this gap by providing a unified infrastructure for both approaches, showcasing China's progress in advanced computing technologies.