Tech Explained: Alibaba Cloud’s Qwen-3 becomes world’s first AI model to operate in orbit – Firstpost in Simple Terms

Alibaba Cloud’s Qwen-3 artificial intelligence model has officially become the world’s first general-purpose AI system to be uploaded and operated in orbit, marking a significant milestone in China’s push to lead the fast-emerging space-based computing sector.

The achievement was confirmed by Wang Yabo, executive vice-president of Chinese aerospace start-up Adaspace Technology, during a conference on Monday, reported by Star Market Daily.

China’s first space computing centre takes AI into orbit

According to Wang, Qwen-3 was successfully deployed in Adaspace’s initial space computing centre, part of the firm’s ambitious “Star-Compute Project”, a 2,800-satellite network aimed at powering physical AI and enabling both model training and inference in orbit. The on-orbit Qwen-3 operation took place in November, when the system executed multiple AI inference tasks from space.

The entire process, from uploading ground-based queries to performing on-orbit AI inference and transmitting the results back to Earth, was completed in under two minutes, demonstrating the rapid data processing capabilities of orbital AI infrastructure.

Qwen-3, launched in April 2025, is part of Alibaba Cloud’s broader Qwen large language model family, which has grown into the world’s largest open-source AI ecosystem, surpassing Meta Platforms’ Llama community in scale and adoption.

Alibaba Cloud, the AI and cloud computing arm of Alibaba Group Holding (which owns the South China Morning Post), developed the Qwen models to serve a wide range of enterprise and research applications.

Adaspace’s initial space computing centre, launched in May 2025, is described by Securities Times as the world’s first AI computing satellite constellation, consisting of 12 satellites.

The company’s success with Qwen-3 underscores China’s growing advantage in orbital data processing, an area that leverages several key benefits of space-based infrastructure, unlimited solar energy, lower cooling requirements, falling launch costs, and reduced data transmission expenses.

By processing data directly in orbit, such systems bypass the physical limitations of ground-based computing, allowing for faster inference, greater scalability, and potentially more secure data handling.

Adaspace’s Star-Compute Project and China’s space strategy

Adaspace’s long-term Star-Compute Project aims to create a vast orbital network of AI computing units. According to Jiemian News, the company’s second and third space computing centres are currently in production and are expected to launch in 2026.

Once complete, the full project will include 2,400 inference satellites and 400 training satellites in low-Earth orbit, with full deployment targeted by 2035.

Founded in 2018, Adaspace Technology is headquartered in Chengdu, the capital of southwestern Sichuan province. The company has filed for an initial public offering (IPO) on the Hong Kong Stock Exchange and was valued at over 6.76 billion yuan (US$972 million) following its most recent funding round in July 2025, according to Securities Times.

Adaspace’s business model spans the entire satellite life cycle, from design and research to manufacturing, launch coordination, in-orbit operations, and AI-powered satellite solutions.

To date, the company has completed 14 space missions, involving the development and deployment of 33 satellites and payloads, reports Jiemian News.

China’s broader national strategy aligns closely with Adaspace’s trajectory. Aerospace and satellite technology have been identified as strategic priorities in China’s 15th Five-Year Plan (2026–2030), with Beijing providing significant policy and financial support to domestic firms.

According to forecasts from the Satellite Industry Association and Frost & Sullivan, China’s satellite industry is expected to expand from around 82.7 billion yuan (US$12 billion) in 2024 to 266.1 billion yuan (US$38 billion) by 2029.

The country’s progress in this domain comes amid intense competition with the United States. Data from Gunter’s Space Page shows that in 2025, the US led globally with 180 orbital launches, while China followed with 93, underscoring Beijing’s accelerating efforts to close the gap.

With Alibaba Cloud’s Qwen-3 now officially operating beyond Earth’s atmosphere, China has taken a decisive step toward integrating artificial intelligence with orbital computing, a move that could redefine how data is processed, analysed, and transmitted across industries ranging from defence to telecommunications.