Google’s Project Suncatcher Takes AI to Space

• Google CEO Sundar Pichai recently shared details about Project Suncatcher, the company’s bold attempt to push artificial intelligence beyond our planet. Based on a new Google Research paper called “Towards a future space-based, highly scalable AI infrastructure system design,” the project imagines massive machine-learning systems running in orbit, powered entirely by the Sun. As Andrew Curran pointed out from the research, the Sun pumps out 3.86 × 10²⁶ watts — that’s more than 100 trillion times what humanity generates in electricity.

• The plan involves launching networks of satellites loaded with Trillium TPUs (Google’s AI chips), solar panels, and optical communication links that let satellites talk to each other at lightning speed. These satellites would work together as giant AI processing nodes, dramatically scaling up what’s possible for training and running AI models. But there are serious hurdles to clear — engineers need to figure out how to protect the hardware from radiation, keep temperatures stable, and ensure everything keeps working reliably in the brutal conditions of low Earth orbit.

Our TPUs are headed to space! 

Inspired by our history of moonshots, from quantum computing to autonomous driving, Project Suncatcher is exploring how we could one day build scalable ML compute systems in space, harnessing more of the sun’s power (which emits more power than 100… pic.twitter.com/aQhukBAMDp

— Sundar Pichai (@sundarpichai) November 4, 2025

• From a business angle, space-based computing could completely change the game for AI infrastructure costs. Traditional data centers are getting hit with skyrocketing energy bills and environmental restrictions, while orbital systems could tap into practically unlimited solar energy. Google’s research shows that launch costs are dropping fast — potentially down to $200 per kilogram by the mid-2030s — which would make large satellite networks actually affordable.

Sundar Pichai laid out the timeline: Our TPUs are headed to space. Early research shows our Trillium TPUs survived without damage when tested in a particle accelerator. Significant challenges remain, but we’re counting down to launch two prototype satellites with Planet by early 2027.

• Project Suncatcher fits right into Google’s track record of taking big swings on futuristic tech, from quantum computers to self-driving cars. Early lab tests putting Trillium TPUs through simulated space radiation showed zero permanent damage after the equivalent of five years in orbit, proving Google’s AI chips can handle extreme environments.

• If it works out, Project Suncatcher could kick off a whole new era in AI infrastructure — where computing literally runs on sunlight, turning the Sun into the ultimate power source for next-generation artificial intelligence.