Skip to content
Advertisement
Google taps Blue Origin to launch solar-powered AI satellites in 2029

Google taps Blue Origin to launch solar-powered AI satellites in 2029

Google / Wikimedia Commons (Public domain)

Project Suncatcher will ride Blue Origin's New Glenn rocket as Google tests whether AI computing works in low Earth orbit

Google wants to put its AI chips in space, and it has picked a ride. The company has signed a launch services agreement with Blue Origin to send satellites from its Project Suncatcher initiative into orbit, targeting a launch window in Q1 2029.

The vehicle is Blue Origin’s New Glenn rocket. The cargo is a set of solar-powered satellites carrying Google Tensor Processing Units, built to run machine learning workloads in low Earth orbit.

What Google and Blue Origin agreed to

The agreement became public on November 19, 2026. It is Google’s first substantial commitment to heavy-lift launch capacity for Project Suncatcher, which until now has been a research effort.

The core idea is simple to describe and hard to pull off. Instead of building another sprawling data center on the ground, Google is testing whether it can put AI hardware on satellites that draw power from the sun. Those satellites would handle machine learning computation in orbit.

The hardware at the center of the plan is Google’s TPU. These are custom chips the company designed specifically for machine learning tasks. In Suncatcher, they would be bolted into satellites rather than slotted into ground-based server farms.

Advertisement

The prototype already flew

Google is not going into the 2029 launch blind. A prototype satellite reached orbit on October 1, 2026, on a mission facilitated by Planet Labs.

That spacecraft carried four TPUs. Its job was to run limited AI inference and to check how the hardware holds up in the harsh conditions of space.

The prototype functioned as expected, and it demonstrated execution of the Gemma 4 model in orbit. Gemma is Google’s family of AI models, so the test effectively had a Google model running on Google chips, several hundred kilometers above the nearest data center technician.

Why put data centers in space at all

Project Suncatcher is one answer to growing pressure on terrestrial data centers as AI demand climbs. Solar energy in orbit offers the possibility of continuous operation, which is the dream scenario for workloads that run around the clock.

Google is not alone in chasing the idea. SpaceX has its own effort, called Starmind, aimed at similar space-based computing goals.

Still, putting servers in space trades one set of headaches for another. Google and Blue Origin will have to address thermal management, which is a polite way of saying that getting rid of heat is surprisingly difficult when there is no air to carry it away. AI chips run hot, so this is not a minor engineering detail.

The second big challenge is inter-satellite communications. A useful computing network needs its satellites to talk to each other quickly and reliably.

What this means for Google, Blue Origin and the orbital compute race

For Google, the agreement turns Suncatcher from a lab curiosity into a project with a scheduled ride. The project remains in its research phase, and Google has not announced a timeline for a fully operational constellation beyond the 2029 launch.

The practical gap between a four-TPU prototype and a working commercial constellation is large. Closing it requires solving thermal limits, building reliable links between satellites, and proving the economics against ever-improving ground infrastructure.

The key milestones are now on the calendar. Watch for further technical updates on Suncatcher’s thermal and communications design, any announcements about constellation size or timing beyond 2029, and progress on New Glenn’s launch cadence ahead of the Q1 2029 window.

Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.