GuoXing Aerospace Deploys Alibaba Qwen3 AI Model in Orbit

GuoXing Aerospace Technology successfully uplinked Alibaba's Qwen3 large language model from ground control to an operational satellite constellation. This deployment represents the first time a general-purpose large-scale AI model has been used to process end-to-end reasoning tasks entirely in orbit. During initial trials, the model processed queries and returned results to Earth within two minutes.

The Chengdu-based startup intends to build a massive space-based computing infrastructure. The company's long-term strategy involves deploying a network of 2,800 specialized satellites by 2035, consisting of 2,400 inference and 400 training satellites. This network will utilize laser inter-satellite links to provide 100,000 petaflops of inference compute and 1 million petaflops of training compute globally.

GuoXing Aerospace Technology expects to complete a 1,000-satellite network by 2030, with the launch of the second and third clusters scheduled for 2026.