Key Takeaways
- China is advancing in space-based AI and computing technology, aiming to leverage orbital resources for autonomous data analysis.
- Zhongke Tiansuan’s space computer has logged over 1,000 days in orbit, with plans for a “space supercomputer” set for trials in 2025.
- Inter-satellite laser links are being developed to enhance communication and reduce data bottlenecks in space operations.
Advancements in Space-Based AI and Computing
A recent breakthrough in space technology has been achieved with a remote-sensing satellite utilizing an in-orbit AI system to autonomously analyze data. This demonstrates the growing potential of Chinese tech companies to harness advanced computing in space, overcoming earthbound limitations of energy, space, and cooling.
Following SpaceX’s launch of the Starcloud-1 satellite, which featured Nvidia GPUs, the spotlight on space computing has intensified. The latest entrant in this field, Zhongke Tiansuan (Comospace), began operations in 2024 with its Aurora 1000 computer that has already spent over 1,000 days in orbit. The company is preparing to test its next-generation Aurora 5000, which incorporates a domestically developed GPU, aiming to establish a “space supercomputer” in low Earth orbit.
According to Liu Yaoqi, CEO of Zhongke Tiansuan, orbital edge computing offers significant advantages by placing AI close to data sources. This approach filters vast amounts of satellite imagery before transmitting manageable data volumes back to Earth. The global connectivity of low-orbit satellites and the abundant energy from sunlight further enhance the viability of these operations.
Chinese authorities are also pushing boundaries with ambitious plans for space data infrastructure. A centralized data center is being developed to operate 700 to 800 kilometers above Earth by a consortium led by Beijing municipal authorities. The Chenguang-1 satellite, set for launch in late 2025, will serve as a technology demonstrator with comparable processing power to ground-based servers, with further deployment planned by 2035.
In addition, Zhejiang Laboratory has initiated a mini-computing constellation named “Three-Body,” designed to process data through an onboard AI model and monitor gamma-ray bursts in real time. With a long-term goal of deploying more than 1,000 satellites, this constellation is expected to execute 100 quintillion operations per second.
To facilitate communication among these orbital systems, China is developing inter-satellite laser links to achieve high-speed data transfer. The startup Laser Starcom successfully established a 400 Gbps link between its Guangchuan-01/02 satellites. These advancements aim to overcome current communication bottlenecks and support the evolving landscape of space computing.
However, challenges remain regarding the operation of computers in extreme space environments. Zhongke Tiansuan is addressing issues such as radiation-induced errors and heat management with innovative solutions like fluid-loop cooling and redundant designs.
The trajectory of space computing is forecasted to begin with intelligent remote sensing, followed by enhanced communications, culminating in sophisticated on-orbit AI applications. This evolution promises new tools in various sectors, such as an envisioned “Fish Finder” app that integrates satellite data to optimize fishing locations.
The content above is a summary. For more details, see the source article.
