Orbital Targets 2027 Prototype for 10,000-Satellite AI Cloud in Space
Updated
Updated · IEEE Spectrum · May 11
Orbital Targets 2027 Prototype for 10,000-Satellite AI Cloud in Space
6 articles · Updated · IEEE Spectrum · May 11
A SpaceX Falcon 9 launch in 2027 will carry Orbital’s first prototype, meant to prove GPU operations in orbit and run commercial AI inference workloads.
Orbital says space-based inference can tap abundant solar power as Earth data centers strain electric grids, using a low-Earth-orbit mesh of small satellites rather than tightly coupled training clusters.
Each planned satellite would carry a GPU rack, roughly 100 kilowatts of power, tennis-court-sized solar panels and radiative coolers; the long-term vision is a distributed cloud of up to 10,000 units.
Radiation, heat dissipation and maintenance remain major hurdles, with experts warning that thousands of satellites raise failure risk and that latency of tens of milliseconds limits suitable applications.
Orbital aims to finish designs by 2026 and open a Los Angeles manufacturing facility by 2028, even as some analysts see operational space data centers as 10 to 20 years away.
With giants like SpaceX entering the race, can Orbital's space AI network survive the immense launch costs and fierce competition?
Are space data centers a real solution to AI's energy crisis, or do they just shift our problems into orbit?
Orbital AI: How Space Data Centers Aim to Solve Earth’s Energy and Infrastructure Crisis
Overview
Orbital is leading a bold initiative to launch AI data centers into space, aiming to overcome the growing limitations of terrestrial infrastructure. By moving beyond the grid, Orbital seeks to harness the abundant energy and unique environment of orbit, enabling new levels of AI processing power. This vision is gaining momentum, with major companies like Anthropic and SpaceX exploring orbital data centers, and Nvidia developing specialized hardware for space. The collaboration and technological advances highlight a shift toward space as a promising frontier for scalable, high-performance AI, addressing energy, supply chain, and community challenges faced on Earth.