The meteoric rise in the use of Generative AI is pushing Earth’s computing infrastructure toward an unprecedented energy crisis. Models like ChatGPT or Gemini require massive processing power operating 24/7, generating intense heat and consuming vast amounts of water and electricity. Against this backdrop, tech giants like Google and SpaceX are exploring the next frontier: moving data centers to Low Earth Orbit (LEO). This transition is not just an attempt to save on expensive real estate; it is an engineering solution to the cooling and energy problems that threaten to limit AI’s growth in the coming years.

Project Suncatcher: Catching the Sun

Google has announced an ambitious project called “Suncatcher,” which aims to launch a constellation of approximately 80 satellites equipped with powerful processors. These satellites will be placed in Sun-Synchronous Orbit (SSO), ensuring almost continuous exposure to sunlight, which is captured significantly more efficiently than on Earth. According to the company’s research, the dramatic drop in launch costs (thanks to reusable rockets) means that within a decade, building a data center in space will be as economically viable as building a massive server facility on the ground—without the need for expensive water-based cooling systems.

New Players and Chinese Constellations

Google is not alone in this mission. Starcloud, a Washington-based startup, has already launched Starcloud-1, a satellite carrying an Nvidia processor designed to run large language models in real-time in space. Simultaneously, China is advancing the “Three-Body Constellation”—a massive project aiming to launch approximately 2,800 satellites that will together form a space-based supercomputer capable of processing petabytes per second. Elon Musk also plans to upgrade the Starlink constellation (V3) so that each satellite serves as a data processing center, essentially turning the satellite internet network into a floating global AI computer.

Revolutionary Potential for the Satellite Industry

Beyond serving users on the ground, space-based data centers could solve a major bottleneck for the space industry itself. Currently, observation satellites are forced to delete vast amounts of data because they lack the bandwidth to transmit it all to Earth. An orbital data center would allow satellites to send raw data via laser to a nearby “server farm,” which would process the information on-site and return only the most critical results to operators. This development could significantly upgrade the capabilities of spy satellites, climate research, and navigation systems.

Looking Ahead

Despite the great promise, significant challenges remain: light pollution affecting astronomers, carbon emissions during launches, and the difficulty of maintaining hardware hundreds of kilometers away. However, with the data center market expected to reach $6.7 trillion in investments by 2030, it seems the race to space is no longer a question of “if,” but “who will be first.” The company that succeeds in creating a true “cloud” above the atmosphere will hold a massive strategic advantage in the age of Artificial Intelligence.