Alibaba’s Qwen-3 becomes first AI model to operate in orbit

Alibaba Cloud’s Qwen-3 AI model has officially become the world’s first general-purpose AI system to operate in orbit, marking a major milestone for China’s space-based computing sector.

Wang Yabo, executive vice-president of aerospace start-up Adaspace Technology, endorsed the achievement on Monday, according to Star Market Daily.

Wang announced that the company successfully deployed Qwen-3 in Adaspace’s initial space computing center, part of the “Star-Compute Project.”

This ambitious network of 2,800 satellites aims to power physical AI and enable model training directly in space. The operation occurred in November, with the system executing multiple AI inference tasks.

The system completed the entire process—from uploading ground-based queries to receiving results back on Earth—in under two minutes. This rapid turnaround highlights the impressive processing capabilities of orbital infrastructure.

Launched in April 2025, Qwen-3 is a key component of Alibaba Cloud’s extensive open-source AI ecosystem. It was deployed on Adaspace’s constellation of 12 satellites, which was launched in May 2025.

This achievement underscores China’s growing advantage in orbital data processing. By conducting data analysis in space, the system takes advantage of unlimited solar energy and natural cooling, while avoiding the limitations of ground-based computing. This method promises faster inference, greater scalability, and reduced costs for data transmission.

With Qwen-3 now operating beyond Earth’s atmosphere, China has taken a strong step toward integrating AI with orbital computing, potentially redefining data analysis for industries ranging from defense to telecommunications.