Google plans to use 2,000 old Pixel phones to power 50 servers, creating a low-carbon cloud.

According to Beating, Google researchers and a team at the University of California, San Diego (UCSD) are exploring phone cluster computing. They plan to repurpose the motherboards of 2,000 retired Pixel smartphones into a low-carbon cloud data center, providing faculty and students with low-cost, low-carbon cloud computing resources. While smartphones typically have a four-year lifespan, the computing cores of retired devices remain intact, with CPU core performance approaching that of modern server cores, performing comparably in some single-threaded tests of SPEC CPU 2017. To adapt to the data center environment, the research team removed unnecessary components such as screens, batteries, casings, and cameras from the old phones, retaining only the motherboards. This is because the motherboard manufacturing process contributes a significant portion of the embodied carbon emissions from the overall device manufacturing process, and it also avoids the safety hazards of batteries in a data center. On the software side, the team replaced the original Android user space with a general-purpose Linux distribution, thus bypassing memory limitations and protection mechanisms designed for mobile devices, such as the "Low Memory Killer." In terms of hardware management, every 25 to 50 mobile phone motherboards are divided into a self-managed cluster, with containerized application scheduling via Kubernetes. According to SPEC CPU 2017 test results, a cluster of 25 to 50 mobile phone motherboards has an overall CPU throughput approaching that of a traditional server. Preliminary experiments show that a micro-cluster of 20 mobile phones can stably handle peak assignment submissions from over 75 students in a parallel computing course, with evaluation tasks taking approximately 50 seconds, and the assignment grading latency even lower than that of AWS cloud instances. The team estimates that the overall computing power of a system consisting of 2000 mobile phones is equivalent to approximately 50 traditional servers, capable of simultaneously supporting the teaching and research of hundreds of system programming and parallel computing courses. The entire system is expected to be officially launched in the fall of 2026, at which time it will serve as a long-term test platform to evaluate the hardware reliability of consumer-grade hardware under sustained high loads.