China launched an AI-operated satellite constellation to create a space supercomputer

China successfully launched the first cluster of 12 satellites of its ambitious three-body computing constellations on 14 May from the Jukevan Satellite Launch Center in a long March rocket. The full supercomputeing network will include 2,800 satellites and provide a joint computing capacity of 1,000 PETA operations per second (POPS). Currently twelve satellites in the classroom are interconnected through high -speed laser communication links that transfer data to 100 gigabit per second. This initiative will use cold vacuum as a natural cooling system and may be able to move AI processing from Earth to space. The project is headed by the company ADA Space and Jhejiang Lab.

Ae in orbit

Translated ADA Space DescriptionEach satellite in China’s launch has an 8 billion-parameter AI model that can operate 744 tera per second (tops). When combined simultaneously, their processing power will be able to operate five petas per second.
Traditionally, satellites collect raw data and transport it back to Earth, limited by narrow bandwidths and transmission windows. This causes data loss and processing delay. By processing the data in space, the system cuts the needs of transmission and takes advantage of the natural cooling of solar energy and space, reducing energy consumption and environmental impact.

Global cooperation and cosmic inspiration

According to Wang Gian of Zhejiang Lab, the stage is open for international use, encouraging other countries and organizations to participate in its development. Although in America and Europe Performed tests On space computers, China’s array is the first to be deployed on an operational scale. Meanwhile, former Google CEO Eric Schmit has proposed to launch data centers in class, after the California launch company purchases a controlled interest in the relativity space.