China has announced an ambitious plan to build a large space computing network that will analyze satellite data directly in orbit before transmitting results to Earth.
The project, called Xingshu or Star Hub, is expected to include around 1,000 satellites connected through advanced communications eventually. It represents another major step in China’s effort to integrate artificial intelligence with its expanding space infrastructure.
Fudan University and Shanghai Xingshu Tiansuan Space Technology Co are jointly developing the network.
According to Chi Nan, a professor specializing in optical communications at Fudan University, the system will provide computing services in space that users can rent based on their needs. Governments, research organizations, and businesses will be able to access computing power by the hour instead of relying only on ground-based processing.
Unlike traditional satellites that mainly collect information and send everything back to Earth, Xingshu is designed to process much of the data where it is created. This approach reduces the amount of information that must travel back through communication links. It also allows important results to reach users more quickly.
Space Computing Enters Orbit
The central idea behind the project is simple because most satellite data originates in space. Instead of transmitting every image or measurement, satellites will analyze the information using AI and send only the final results to Earth. This saves communication bandwidth and speeds up the delivery of useful information.
The network will support several real-world applications across different industries. Planned services include weather forecasting, disaster warning systems, shipping and port monitoring, electricity grid inspections, and remote sensing for emergency responders and government agencies. These capabilities are expected to help organizations make faster and more informed decisions.
The system’s design combines a central computing satellite, described as a space brain, with weather and Earth observation satellites. These satellites will communicate through laser links that transfer information rapidly between spacecraft. AI models running onboard will process observations before sending the finished analysis to users on Earth.
READ ALSO: US Warns Iran-Linked Hackers Target Water and Energy Systems With Industrial Cyberattacks
Three Phase Expansion
The project will be developed in three stages over time. The first phase includes two computing satellites working alongside 12 AI-enabled sensing satellites as an engineering demonstration. This initial deployment will test the technology before larger expansion begins.
The second stage will increase the network to 50 computing satellites and 100 AI-enabled sensing satellites. Engineers will use this larger constellation to improve computing capacity and expand the range of services offered. The final phase aims to complete the full network of around 1,000 satellites operating together in orbit.
Xingshu joins several other Chinese space computing programs already under development. The Tiansuan constellation, led by Beijing University of Posts and Telecommunications, plans to deploy 300 satellites and has already launched eight. Zhejiang Lab’s state-backed Three-Body Computing Constellation targets 1,000 satellites and placed its first 12 satellites into orbit last year.
Global Strategy Comparison
China’s approach differs from a proposal made earlier this year by SpaceX founder Elon Musk. SpaceX requested approval from US regulators to eventually deploy up to one million satellites that would function as orbital AI data centers powered by solar energy. The company argues that space offers abundant sunlight while avoiding the large amounts of water used to cool many Earth-based data centers.
Chinese researchers have taken a different view of the problem. They argue that electricity is not the main limitation for AI computing inside China, making it more practical to process satellite-generated data in orbit rather than relocating massive data centers into space. Projects such as Xingshu, Tiansuan, and the Three-Body constellation all follow this strategy.
Some researchers outside China have also raised questions about the feasibility of operating very large AI data centers in orbit. They point to the challenges of launching, powering, cooling, and eventually removing such large numbers of satellites from space. Manufacturing and maintaining fleets of that scale would also require significant resources.
China’s latest initiative highlights a growing shift toward making satellites more intelligent, rather than using them only as data collectors. By performing analysis directly in orbit, the country hopes to improve response times, reduce communication demands, and support a wide range of civilian and commercial services.
