Google partners with Jeff Bezos' Blue Origin to launch TPU-powered AI server satellites aboard the New Glenn rocket in Q1 2029, as part of Project Suncatcher
First reported by The Verge ·
The cost to deploy AI compute in orbit falls dramatically, enabling a new class of space-based AI services.
Google and Jeff Bezos' Blue Origin have partnered to launch Google's AI server satellites into orbit. The first launch, scheduled for the first quarter of 2029, will utilize Blue Origin's New Glenn rocket. This initiative is part of Google's Project Suncatcher, aiming to establish AI data centers in space. Google has already deployed its initial AI server satellite in October, which successfully transmitted an AI-generated haiku from orbit using its onboard TPUs and Gemma 4 model. The upcoming Blue Origin launch will be the first dedicated heavy-lift mission for Project Suncatcher, carrying satellites with computing power comparable to a data center rack into a sun-synchronous orbit. Previously, Google's AI server satellite was launched as a secondary payload on a SpaceX Falcon 9 rocket.
This partnership signifies a major step towards realizing space-based AI infrastructure, moving beyond experimental payloads to dedicated, high-capacity deployments. By leveraging Blue Origin's heavy-lift capabilities, Google can place significantly more powerful AI hardware into orbit, potentially paving the way for distributed, global AI processing and data analysis networks independent of terrestrial limitations. The project's success could redefine the economics and feasibility of space-based computing, making it a viable option for various applications.
The scale of computing power described, equivalent to a data center rack, suggests a transition from basic AI tasks in space to more complex, real-time processing. This could enable autonomous operations for satellites, advanced Earth observation with onboard AI analytics, and unprecedented communication capabilities. The ambition to build space-based data centers points to a future where computational resources are not confined to Earth, potentially impacting everything from scientific research to commercial services that require low-latency, high-bandwidth data processing globally.
AI-written summary. May contain errors.