China Launches AI-Driven Satellite Constellation to Transform Space Computing
•February 15, 2026
0
Why It Matters
The constellation proves AI can be processed in space, accelerating real‑time insights for Earth observation and astrophysics, and sets a competitive benchmark for global space‑based computing initiatives.
Key Takeaways
- •12 satellites launched, 10 AI models onboard
- •Inter-satellite links established among six satellites
- •Remote sensing model surveyed 189 km² in snow
- •Astronomy AI achieved 99% accuracy classifying gamma bursts
- •Full constellation aims 1,000 satellites, 100 quintillion ops/sec
Pulse Analysis
Artificial intelligence is reshaping how data is handled beyond Earth’s atmosphere, turning satellites from passive relays into active processors. While NASA’s Spaceborne Computer and India’s Agnikul Cosmos explore on‑board AI, China’s Three‑Body Computing Constellation distinguishes itself by integrating two 8‑billion‑parameter models across a dozen satellites. The remote‑sensing model’s ability to map terrain under heavy snow and the astronomical model’s 99 % classification accuracy illustrate how on‑orbit inference can dramatically reduce downlink requirements, enabling faster decision‑making for both civilian and defense applications.
The technical architecture hinges on inter‑satellite links that create a mesh network, allowing data to be routed and processed across six linked platforms. This distributed computing approach not only balances workload but also provides redundancy, a critical factor for missions that demand continuous operation. By processing raw sensor data in space, the constellation delivers refined products—such as identified infrastructure or classified cosmic events—directly to end users, cutting latency from days to minutes. The projected scale of over 1,000 satellites promises an aggregate compute capacity near 100 quintillion operations per second, rivaling terrestrial supercomputers while operating in a radiation‑hardened environment.
The broader market impact is significant. Real‑time, high‑resolution Earth observation can enhance disaster response, precision agriculture, and smart‑city planning, while space‑based astrophysics accelerates discovery cycles. As commercial players like SpaceX and Starcloud race to build data‑processing constellations, China’s demonstrable success may spur policy debates on data sovereignty and orbital traffic management. Investors and policymakers will watch how these AI‑enabled networks evolve, potentially redefining the economics of satellite services and the strategic balance of space capabilities.
China Launches AI-Driven Satellite Constellation to Transform Space Computing
China has launched a satellite constellation equipped with 10 AI models and established inter-satellite networking. According to Zhejiang Lab, this system demonstrates the potential of AI for applications. They range from deep space exploration to natural resource monitoring and even smart city planning.
Launch of the Three-Body Computing Constellation
Working with international partners, Zhejiang Lab launched 12 satellites in May 2025 as the first part of the Three-Body Computing Constellation. After about nine months in orbit, the satellites have successfully demonstrated core capabilities. These included inter-satellite connections, AI model operation, data processing, and scientific instrument testing.
‘With a computing constellation, part of the data can be processed in space and delivered straight to users,’ said Li Chao from Zhejiang Lab.
Powerful AI Models in Orbit
The constellation includes two 8-billion-parameter AI models: one for remote sensing and one for astronomical time-domain analysis. These rank among the largest AI models currently operating in space.
In November 2025, the remote sensing model conducted an infrastructure survey across 189 square kilometres in northwest China. Despite heavy snow, it automatically identified buildings, bridges, and stadiums.
For astronomy, two satellites equipped with cosmic X-ray polarisation detectors used an AI model to rapidly classify gamma-ray bursts in orbit. The model achieved 99% accuracy while dramatically reducing the amount of data that needed to be transmitted and processed on Earth.
Towards a Networked Space System
The team also established inter-satellite links among six satellites, an important step toward a fully networked space system. Once the full constellation of over 1,000 satellites is deployed, it is expected to perform around 100 quintillion operations per second, according to Zhejiang Lab.
Global Efforts in AI-Powered Space Computing
AI computing is becoming an increasingly important focus in space exploration. For example, NASA and Hewlett-Packard Enterprise developed the Spaceborne Computer series, which allows AI and machine-learning tasks to run directly on the ISS, enabling real-time data analysis and autonomous operations.
In the commercial sector, India’s Agnikul Cosmos is exploring space-based data centres for AI inference. Also, companies like SpaceX and startups such as Starcloud are developing satellite constellations to handle large-scale data processing in orbit.
The post China Launches AI-Driven Satellite Constellation to Transform Space Computing appeared first on Orbital Today.
0
Comments
Want to join the conversation?
Loading comments...