China Launches Hyperspectral Satellites Equipped with On-Orbit AI Large Model
China successfully launched the Dongfang Huiyan hyperspectral 01 and 02 satellites using a Jielong-3 rocket on August 5, 2026. These satellites are equipped with the "Lvshui Qingshan" AI large model, allowing them to process remote sensing data directly in orbit. This launch represents a significant real-world deployment of edge AI in aerospace, reducing the latency of remote sensing data by processing it in space instead of transmitting raw data to Earth. It enables faster response times for environmental monitoring, disaster management, and resource tracking. Weighing around 300 kg each—about one-fifth of similar Western satellites—these lightweight spacecraft feature a 22-band hyperspectral payload with a 5-meter resolution. They operate in a 780 km regression orbit and can complete a full global scan every five days.
## BACKGROUND
Hyperspectral remote sensing collects and processes information from across the electromagnetic spectrum, allowing for precise identification of materials and environmental changes. Traditionally, satellites transmit massive amounts of raw sensor data to ground stations for processing, which creates significant bandwidth and time bottlenecks. On-orbit data processing using edge AI solves this by analyzing data directly on the satellite, transmitting only the critical insights back to Earth.