AI-powered satellite constellation launched in China

The sky over the Jiuquan Satellite Launch Center was clear on the morning of May 14, 2025, as the Long March 2D rocket roared to life, its fiery ascent piercing the blue horizon.

Onboard the rocket were 12 computing satellites, the first deployment of the ambitious “Star Computing” program, spearheaded by Chengdu-based ADA Space, which focuses on artificial intelligence (AI) technologies and their application in satellite networks.

This launch marked a historic milestone in space-based AI and edge computing. The 12 satellites, each equipped with onboard AI processors, laser inter-satellite communication systems, and distributed computing capabilities, successfully entered their designated orbits, forming the foundational layer of what will eventually grow into a 2,800-satellite constellation.

A new kind of space infrastructure

Traditionally, satellites collect data and beam them back to Earth for processing in massive data centers. But this approach has limitations, namely latency, bandwidth constraints, and reliance on ground infrastructure. The Star Computing program flips this model by enabling real-time, in-orbit data processing, which will be done via the computational satellites it sends into space.

Each of the 12 satellites boasts a computing power of 744 TOPS (tera operations per second), with the entire initial constellation collectively reaching 5 POPS (peta operations per second), thus making it the most powerful space-based computing network ever deployed by mankind. Additionally, the satellites utilize laser inter-satellite links with speeds up to 100 Gbps, allowing them to function as a coordinated, intelligent swarm rather than isolated data collectors.