Alibaba Cloud's Qwen-3 is the first AI model to operate in space, launched by Adaspace Technology as part of the Star-Compute Project.
Alibaba Cloud's Qwen-3 has made history as the first artificial intelligence model to be uploaded and function in space. This development was highlighted by the South China Morning Post.
Adaspace Technology, a Chinese startup, successfully implemented the AI model at a space computing center as part of the Star-Compute Project, which features a constellation of 2,800 satellites aimed at enhancing physical AI capabilities and facilitating training and inference.
Wang Yabo, the Executive Vice President of Adaspace, reported that the large language model conducted multiple tasks, with the exchange of prompts from Earth and receiving results taking under two minutes.
The Adaspace center, launched in May 2025, marked a significant milestone as it comprises the world’s first fleet of 12 satellites dedicated to AI computations.
The Role of Space in AI Evolution
The trend of utilizing space for AI hubs is gaining momentum, although many projects are still in their initial phases. In January, Tesla's Elon Musk revealed the revival of the Dojo3 project, which aims to develop a third-generation chip for vehicles, now targeting space computing applications.
Musk and other industry leaders propose that the future of data centers may lie in space, citing limitations of Earth-based energy systems. Space offers advantages such as unlimited solar energy and ample room for infrastructure, but challenges like high rocket launch costs remain significant.
Research group 33FG anticipates that space-based AI computations could become cost-effective by 2030. Google has already initiated a project to establish a network of satellites in low Earth orbit, intended to supply energy for data center operations.
Sam Altman, CEO of OpenAI, supports this vision, while Musk's dominance in delivery systems gives him a strategic edge. He aims to leverage an upcoming SpaceX IPO to finance his ambitious plan for a satellite constellation capable of continuous energy collection and operation.
Additionally, in September 2024, the Search for Extraterrestrial Intelligence Institute began exploring AI and machine learning to identify signs of extraterrestrial life.
