Jakarta, INTI - A Chinese commercial aerospace company, GuoXing Aerospace Technology, has successfully deployed a general-purpose artificial intelligence (AI) model directly on an orbiting satellite, marking a major breakthrough in the development of orbital computing technology.
Qwen3 Large Language Model Installed on Satellites, Enabling End-to-End Processing from Orbit
During a seminar held on Monday, January 26, 2026, GuoXing announced that it had integrated Alibaba’s Qwen3 Large Language Model (LLM) into its first space-based computing center.
“This marks the world’s first deployment of a large-scale, general-purpose AI model from ground control to an operational satellite constellation in orbit,” the company stated in its announcement.
With this technology in place, reasoning tasks can be fully completed in orbit without the need to transmit data back to Earth for processing.
During the trial, a query was transmitted fromEarth to a satellite, processed by the Qwen3 model onboard the satellite, and then sent back to Earth, all completed within just two minutes.
This demonstration forms part of the first constellation cluster launched by China in May 2025, comprising 12 computing satellites operated by GuoXing Aerospace.
Targeting 2,800 Space-Based Computing Satellites by 2035
GuoXing Aerospace Executive Vice President Wang Yabo outlined the company’s ambitious plan to build a global space computing network consisting of 2,800 satellites by 2035.
The planned network will consist of 2,400 inference satellites and 400 training satellites deployed across multiple orbital configurations, including sun-synchronous orbits, dawn–dusk orbits, and low-inclination orbits at altitudes ranging from 500 to 1,000 kilometers. The entire system will be interconnected through inter-satellite laser links to ensure high-speed data transfer.
The network is designed to deliver a total inference computing capacity of 100,000 petaflops and a global training computing capacity of up to 1 million petaflops. The launch of the second and third satellite clusters is scheduled to take place throughout 2026.
In the medium term, the company aims to establish an initial network of 1,000 satellites, targeted for completion by 2030.
The direct deployment of AI computing in orbit is emerging as a global trend, driven by rapidly growing demand for computing power worldwide.
In November 2025, SpaceX placed the Starcloud-1 satellite into orbit, equipped with Nvidia GPUs, signaling a similar shift toward performing AI processing directly in space.
China’s move through GuoXing represents a significant milestone in a new era of space-based computing infrastructure that is intelligent, high-speed, and increasingly autonomous.
Conclusion
The deployment of AI models directly in orbit marks a major shift in the evolution of global computing infrastructure. By enabling real-time processing in space, initiatives such as China’s GuoXing project, and similar efforts by other players, demonstrate how space-based computing can reduce latency, increase efficiency, and support the growing demands of AI-driven applications. As more satellites are equipped with advanced computing capabilities, space is set to become a critical layer of the global digital ecosystem, reshaping how data is processed, transmitted, and utilized worldwide.
Read more: Following South Korea, Indonesia to Mandate Special Labels for AI-Generated Content