HeadlinesBriefing favicon HeadlinesBriefing.com

UC San Diego Builds Cloud Computing Cluster from Retired Pixel Phones

Google AI Blog •
×

Researchers at the University of California San Diego are tackling computing's carbon footprint by repurposing retired smartphones into a low-carbon cloud platform. With Google's support, they plan to deploy a datacenter built from 2,000 Pixel phones, stripping them down to motherboards and clustering them for general-purpose computing. This approach targets embodied carbon emissions from hardware manufacturing, which account for roughly half of a device's environmental impact.

Smartphones today pack serious compute power - their single-threaded processor performance rivals modern multicore servers. However, phones have limited memory (8-12GB) compared to traditional servers, requiring applications to fit within constrained resources. The team removes unnecessary components like batteries and displays, then installs a general-purpose Linux distribution to replace Android's mobile userspace. Containerized applications managed by Kubernetes orchestrate workloads across clusters of 25-50 devices.

Many educational applications already run on small cloud instances like AWS t3.micro, making them perfect candidates for smartphone hosting. Early tests show 20 phones can handle peak submission rates for 75+ student classes with faster grading than default cloud backends. The full 2,000-phone deployment launching in Fall 2026 will provide 50 server-equivalents while serving as a testbed for consumer hardware reliability under sustained use.

This project demonstrates how re-deploying unwanted smartphones can provide affordable cloud computing for academic workloads while avoiding raw material extraction. The approach directly reduces environmental impact by extending device lifecycles beyond their typical four-year replacement cycle.