AI consumes so much energy that even Google is running out of outlets. Their new idea? Move the problem into space. Through Project Suncatcher, the Mountain View company aims to transform the Sun into a giant battery to power its future AI systems.
With increasing power demands from its AI models, Google is working on orbital data centers that remain in a constant state of day and night, continuously fueled by solar energy. Named Project Suncatcher, this initiative could herald a new era for cloud computing and push the boundaries of what our Earth-based infrastructure can handle.
Google’s Data Centers Set to Launch into Space
Google Research has recently announced the Project Suncatcher. They intend to deploy space-based data centers dedicated to AI, powered by solar energy.
The goal is to leverage the constant solar radiation in space, free from terrestrial constraints. These orbital data centers would take the form of a constellation of satellites equipped with custom-adapted TPUs (Tensor Processing Units) from Google, interconnected through ultra-fast optical communications.
We all know that AI is extremely power-hungry. As models grow more complex, they require enormous amounts of electricity, and industry giants like Google are beginning to feel the strain. In the U.S., Microsoft reportedly has unused stocks of GPUs due to insufficient energy to run them. This has led to skyrocketing electricity prices and severe pressure on existing electrical networks.
Some companies are exploring radical solutions like small modular nuclear reactors or thermal energy recovery systems. In contrast, Google is looking skyward.
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… pic.twitter.com/aQhukBAMDp— Sundar Pichai (@sundarpichai) November 4, 2025
Why Up There? Because The Sun Never Sets
By placing its satellites on a heliosynchronous low Earth orbit, Google could keep its solar panels exposed to sunlight almost continuously. This would yield eight times more energy than terrestrial solar panels, a significant advantage for energy-intensive AI servers.
However, Google still needs to address a number of technical challenges, such as establishing stable and ultra-fast optical links between satellites and managing a compact constellation to prevent collisions. They must also shield electronic circuits from space radiation and reduce the astronomical cost of launching and maintaining the satellites.
While space-based servers could ease terrestrial electricity consumption, they would also pose new challenges, including increased space debris, risks of astronomical disruptions, and an expanding orbital footprint.
At this point, Project Suncatcher remains a feasibility study. However, Google plans to launch two prototypes by 2027 to assess the viability and profitability of the project.
