Artificial intelligence and space computing are moving back and forth: cost dilemmas and business closed loops are yet to be overcome.

In 10 or 20 years, fishermen may ask satellites, "Can you tell me where can I catch tuna?" Satellites will use hyperspectral cameras to find fish, calculate through powerful computers, and might tell them that there is a group of tuna 20 nautical miles northeast, advising them on what hooks to use, how to collect nets, and where to sell. Application developers may extract a 10% service fee from this. This is not a scene from a science fiction movie, but a future application scenario described by Liu Yaoqi, deputy researcher at the Institute of Computing Technology of the Chinese Academy of Sciences, at the 2026 Space Computing Industry Conference. In recent years, as the energy consumption of ground computing centers has surged, deploying computing power in space is transitioning from a cutting-edge concept to engineering practice, also attracting attention from various sectors of society. Zhao Ce, deputy director of the Information and Communication Development Department of the Ministry of Industry and Information Technology, said that space computing has advantages such as real-time processing in orbit, low-cost energy, and wide coverage, which helps enhance space energy development capabilities, improve global coverage and anti-interference capabilities, expand network application boundaries, and has strategic value and industrial prospects. However, beyond enthusiastic imagination, there is also a sobering reality. Several experts warn that deploying computing power in space still has a long way to go before it becomes "useful" and "affordable" - issues such as cost, heat dissipation, networking, and business models have not yet found satisfactory answers.