Chinese companies Guoxing Aerospace and SenseTime Technology have announced the joint construction of a space computing constellation. The scale of the constellation will reach a thousand satellites and ten petaflops (trillions of floating-point operations per second), aiming to build a new generation of global artificial intelligence (AI) infrastructure.
According to China News Service, the announcement was made on Saturday (July 18) at the "Foundation Model Architecture Innovation and Ecosystem Cooperation Forum" of the 2026 World Artificial Intelligence Conference held in Shanghai.
It was introduced that the two parties will advance the integrated space–ground AI infrastructure construction in stages according to the development path of “single satellite verification—platform building—constellation networking.”
According to the plan, the initial deployment and networking of compute satellites will be completed within 2026; by 2028, they plan to establish a ground-space hybrid cloud commercial computing backbone network; by 2030, a space computing constellation consisting of around one thousand satellites, with a total computing power exceeding ten petaflops, is planned to be completed, forming a global intelligent spatial computing network that deeply integrates ground intelligent computing centers, in-orbit computing constellations, and application scenarios.
According to the plan, the initial deployment and networking of compute satellites will be completed within 2026; by 2028, they plan to establish a ground-space hybrid cloud commercial computing backbone network; by 2030, a space computing constellation consisting of around one thousand satellites, with a total computing power exceeding ten petaflops, is planned to be completed, forming a global intelligent spatial computing network that deeply integrates ground intelligent computing centers, in-orbit computing constellations, and application scenarios.
Among them, the first batch of networked compute satellites will be equipped with high-performance, independently controllable spaceborne computing platforms, which feature high computing power, large storage, and real-time intelligent processing capabilities in orbit, supporting the operation and deployment of multiple types of AI algorithms in orbit.
In terms of computing infrastructure construction, the two parties will jointly develop a space–ground integrated AI computing scheduling platform at the 10,000-card level, enabling collaborative scheduling of ground intelligent computing centers and in-orbit computing nodes, promoting the integrated development of terrestrial and space computing resources, and building a global space–ground integrated AI token (token) production and circulation system.
Around the construction of a large-scale spatial intelligent computing network, the two parties will jointly carry out key technology research, explore AI computing technology systems adapted to the space environment, improve massive compute satellite collaborative inference and intelligent task scheduling capability, and lay the foundation for the large-scale deployment of future spatial intelligent computing networks.
In terms of computing infrastructure construction, the two parties will jointly develop a space–ground integrated AI computing scheduling platform at the 10,000-card level, enabling collaborative scheduling of ground intelligent computing centers and in-orbit computing nodes, promoting the integrated development of terrestrial and space computing resources, and building a global space–ground integrated AI token (token) production and circulation system.
Around the construction of a large-scale spatial intelligent computing network, the two parties will jointly carry out key technology research, explore AI computing technology systems adapted to the space environment, improve massive compute satellite collaborative inference and intelligent task scheduling capability, and lay the foundation for the large-scale deployment of future spatial intelligent computing networks.
With respect to application scenarios, the two parties will promote the prioritized implementation of space computing in weak-network, no-network, and high-value scenarios, providing real-time AI inference and intelligent decision support, expanding the boundaries of intelligent computing services, and accelerating the scaled commercialization of space computing.