Google Launches First Project Suncatcher Satellite to Test AI Chips in Orbit

Google has sent a refrigerator-sized prototype satellite into orbit to test whether its AI chips can work in space, NPR reported on October 1, 2026. The launch is the first step of Project Suncatcher, a research effort the company hopes could lead to solar-powered AI data centers in orbit.

The satellite carries four of Google’s Tensor Processing Units, or TPUs, chips the company designed for machine learning and already uses in data centers on the ground. A SpaceX rocket carried it into orbit. Google developed it with Planet, the aerospace and satellite imagery company, which will get it up and running before Google switches on and tests the chips.

Fifteen minutes at a time

In orbit, the chips will run a version of Gemma, Google’s open-weight AI model, to answer simple queries. Because of heat limits, they will run in 15-minute stretches. Cooling is one of the hardest problems: on Earth, chip heat can be carried away by air or water, but that is not possible in the vacuum of space. Google has called cooling “a crucial research challenge,” and project lead Travis Beals told NPR the radiators are among the heaviest parts of the current mission.

The mission is meant to last a year. Google said it will measure how the TPUs handle “the physical stress of spaceflight and the radiation and thermal extremes of space.” Beals described this prototype as “a very minimal test” to confirm the chips can run in space. NPR also corrected an earlier version of its story to say this prototype is in a standard orbit and uses batteries.

Why space

The appeal is power. Demand for AI is driving demand for data centers, and opposition to power-hungry facilities on the ground is growing, NPR noted. “The sun puts out almost all of the power in our solar system,” Beals said, describing the project as a way of tapping solar power to run AI compute. Google eventually envisions clusters of satellites linked by lasers, and plans to launch two more next year to test those connections.

Not cheaper anytime soon

Beals was candid about the economics. “I don’t see this being something where it’s cheaper to do this in the next five years. I think it will take longer than that,” he told NPR. He also said he expects ground and space data centers to coexist for a long time, and called the project a “moonshot.”

Carnegie Mellon professor Brandon Lucia told NPR that much of the technology has been proven at a research level, but that heat, maintenance and repairs in orbit add large costs and complexity. NPR also raised environmental questions, since rockets burn carbon-based fuel and satellites eventually re-enter the atmosphere.

Why it matters for the US

The project is one answer to a growing American political issue: where to put the data centers AI needs. At the White House accord signing days earlier, President Trump said companies had agreed to ease local opposition to data centers; see our report on the accord. Suncatcher is a long-range bet, not a near-term fix.