Beta The Briev beta is out. Free on iPhone via TestFlight — install it in under a minute.

Join the beta ↗
Briev
Live
Technology

Google launches AI processor test aboard SpaceX rocket to probe orbital data centers

Google sent four AI chips on a SpaceX launch to run Gemini models briefly, marking a first step toward orbital data centers.

Google’s Project Suncatcher lifted off aboard a SpaceX rocket carrying four artificial-intelligence processors intended to run its Gemini models for about 15 minutes before shutting down to cool. The experiment demonstrates the concept of an orbital data center, where sunlight could provide far more power than on Earth, yet the scale required—hundreds of thousands of chips and gigawatt-level power—far exceeds current satellite capabilities.

Experts point to hurdles including massive solar-panel arrays, heat-dissipation in vacuum, radiation-induced errors, and the costly, limited bandwidth needed to transmit results back to Earth. Google has been stress-testing hardware on the ground with vibration and proton-beam tests to anticipate these issues. Even if technical obstacles are solved, the economics of launching, maintaining and replacing satellite hardware may outweigh the energy savings, leaving the viability of space-based AI computing uncertain.

Why it matters

It shows how leading tech firms are exploring space to meet AI’s growing energy demand, but significant technical and cost barriers remain.

In this story

orbital data centerAI processorssolar powerspace launchheat dissipationbandwidth constraintsProject Suncatcher
Get the beta ↗