AI Data Centers Move to Space

Orbital is pushing a bold idea: running AI data centers in low Earth orbit. By using solar power and the natural cold of space, the company aims to solve one of AI’s biggest limits on Earth, energy.

Founded by Euwyn Poon, known for building Spin, Orbital is targeting the growing gap between AI demand and available power. As AI systems scale, data centers are consuming massive energy, creating what many now call an “energy ceiling.”

Orbital’s approach shifts compute beyond Earth’s grid. In orbit, satellites can draw continuous solar energy and use radiative cooling, removing the need for power-hungry cooling systems used on Earth.

The company’s first satellite, Orbital-1, is set to launch in April 2027 aboard a SpaceX Falcon 9. It will test GPU operations in space and support AI inference workloads.

Each satellite will carry clusters of NVIDIA GPUs, effectively turning them into compact, space-based data centers.

The strategy reflects a larger trend. As AI infrastructure strains terrestrial resources, companies are exploring alternative environments to sustain growth.

If successful, Orbital could redefine how compute infrastructure is built, moving from الأرض-bound limitations to a model powered by space-based energy and scalability.