China Activates First Fully Domestic 100,000-GPU AI Supercluster

China has activated its first fully homegrown artificial intelligence supercluster, a computing facility built around 100,000 accelerator cards. The deployment marks a significant milestone in Beijing's efforts to reduce reliance on foreign semiconductor technology.

The new supercluster is designed to handle the massive computational loads required for training large-scale AI models. By using domestically produced components throughout, China aims to secure its AI supply chain amid ongoing export restrictions on advanced chips from the United States and its allies.

Industry observers note that the scale of this cluster — 100,000 cards working in tandem — places it among the largest AI computing installations globally. The achievement underscores China's accelerating push to close the technology gap with Western rivals in high-performance computing.

The move is part of a broader national strategy to build independent capabilities in critical technologies, from chip design to cloud infrastructure. Analysts say the successful deployment could pave the way for further domestic supercomputing projects in the coming years.

While specific performance benchmarks have not been disclosed, the completion of this supercluster signals China's growing confidence in its homegrown hardware and software stack for AI workloads.