AI in Space: How Google’s Project Suncatcher Could Change the Energy Game

The Next Frontier: Google’s AI Chips Are Leaving Earth

According to a recent report from Google Research, the company is working on a bold experiment—Project Suncatcher—an initiative to run its Tensor Processing Units (TPUs) in space. The goal: tap into the endless stream of solar energy orbiting above Earth to fuel artificial intelligence workloads without overloading terrestrial power grids.

This move could mark one of the most radical shifts in how tech giants think about AI infrastructure. Instead of building more power-hungry data centers on the ground, Google wants to send the compute itself into orbit.

Why Google Is Looking to the Stars

Today’s AI boom comes with a dark side: energy demand. From large language models to generative AI, training and running these systems consume vast amounts of electricity. Traditional data centers already account for around 1–2% of global energy use—a figure expected to rise sharply as AI adoption grows.

By harnessing near-constant sunlight in orbit, satellites could provide a steady and renewable energy source, potentially running 24/7 without weather disruptions or nighttime downtime.

If successful, Project Suncatcher could make AI computation cleaner, cheaper, and more sustainable—a crucial milestone for both tech and climate innovation.

The Cosmic Challenges Ahead

Of course, running AI in space isn’t as simple as flipping a switch. Google’s researchers face daunting technical and logistical challenges:

• Data transfer at light speed: To operate effectively, orbiting AI chips must communicate with each other at speeds of tens of terabits per second. That requires satellites to fly in tight formations—raising collision and safety concerns.

• Radiation and reliability: Space is harsh. Google’s early tests suggest its Trillium TPUs might survive up to five years in orbit, but radiation remains a major threat to chip performance and longevity.

• High upfront costs: Launching and maintaining satellites is expensive. Google hopes costs will stabilize by the mid-2030s, reaching parity with Earth-based data centers.

A first prototype mission is planned for 2027, in collaboration with satellite imaging company Planet Labs, to test how these chips handle real orbital conditions.

What This Means for the Future of AI and Energy

If Project Suncatcher succeeds, it could usher in a new era of space-based computation—a concept once confined to science fiction. Imagine AI models trained off-world, powered by clean sunlight, while Earth’s resources are preserved for human and ecological needs.

This could also decentralize computing, creating interplanetary cloud networks where satellites handle heavy workloads before syncing results back to Earth. It’s a futuristic vision, but one with real environmental and commercial implications.

Our Take: The Sky Isn’t the Limit—But It’s Complicated

At Web Seasoning, we see Google’s move as both visionary and pragmatic. The race to make AI greener is no longer optional—it’s existential. But before AI truly “goes intergalactic,” companies must first solve the infrastructure issues that already exist here: limited connectivity, high costs, and the carbon footprint of existing server farms.