Google Sends AI-Powered Satellite into Orbit to Test Space-Based Computing

Google has launched a satellite into orbit as part of its Suncatcher research project. The prototype satellite is equipped with four Tensor Processing Unit (TPU) chips, which are used for machine learning tasks. This experiment aims to determine if space can become a location for scalable machine-learning infrastructure in the future.

The satellite was sent into space in partnership with Planet during the SpaceX Transporter-18 mission. Google plans to test how its equipment and processors perform in the space environment, particularly in terms of providing sufficient computing power for basic artificial intelligence queries.

Satellites in low Earth orbit can receive sunlight almost continuously, generating up to eight times more solar energy than on Earth. This could potentially make space-based data centers a viable option for processing larger AI workloads.

Ariel Ekblaw, founder and executive director of the Aurelia Institute, notes that there are significant obstacles to establishing orbital data centers, including low temperatures in space, radiation, and equipment upgrades. However, she suggests that overcoming these challenges could help reduce AI's carbon footprint.