GuoXing sends Alibaba’s Qwen3 into orbit as China steps up AI space challenge

Chinese tech firm, GuoXing Aerospace Technology has deployed Alibaba Cloud’s Qwen3 on its orbiting satellites, reportedly making the AI model the world’s first general-purpose model to be operated in orbit.

The development has also marked a significant step in China’s bid to lead the promising space-based computing sector. GuoXing announced on Monday this development, enabling end-to-end reasoning entirely in orbit.

GuoXing reveals ambitious plans up to 2035

The Chengdu-based startup deployed Qwen3 to a space computing center in Orbit in November last year where it performed several inference tasks. According to China Daily, the Qwen3 model successfully completed a series of experiments during its trial, with questions transmitted from earth to the satellite, processed on-board and returned to earth on ground stations within a short space of time, less than two minutes.

China launched a new collection of 12 space computing satellites into orbit in May last year. This was the first cluster of the startup’s space computing, and represented the world’s first AI computing satellite constellation, according to the SCMP.

The development comes as there is a growing appetite for computing power fueled by AI. With this growth, there is also now a new field in the tech race where intelligent computing capabilities are now being pushed into space.

According to China Daily, SpaceX rocket orbited the Starcloud-1 satellite, which was mounted with Nvidia GPUs, in November last year.

For GuoXing, Wang revealed its plans to build an expansive network of 2,800 specialized computing satellites by 2035.

This planned constellation will include 2,400 inference satellites and 400 training satellites that will be deployed across sun-synchronous, dawn-dusk, and low-inclination orbits at 500 to 1,000 kilometers altitudes.

According to the company, the constellation is meant to employ laser inter-satellite links to facilitate high-speed data transfer with the aim of delivering 100,000 petaflops of inference compute and 1 million petaflops of training compute worldwide.

Wang also revealed that the second and third satellite clusters are scheduled for deployment this year, with a 1,000-satellite network completed by 2030.

The latest development also places Alibaba as a top contender in the global AI race. Its Qwen3 was released in April last year, strengthened Alibaba Cloud’s Qwen family position as the largest open source AI ecosystem ahead of Meta’s Llama models.

As previously reported by Cryptopolitan, the Qwen3 family of AI models also surpassed DeepSeek’s R1 after it performed better in tests that measure open-source AI models’ abilities in areas like language instruction, coding, math, and data analysis.

The smartest crypto minds already read our newsletter. Want in? Join them.