Overview
- Google Research published work with University of California San Diego over the weekend proposing “phone cluster computing,” a method to repurpose retired smartphone motherboards as networked compute nodes.
- The approach strips displays, batteries and cameras from phones, installs a Linux-based OS on the motherboard, links devices over a network, and uses Kubernetes to schedule and manage workloads.
- Google says preliminary benchmarks show roughly 25 to 50 phones can match a modern server for some tasks, but the company and researchers describe those performance claims as early and workload-dependent.
- UC San Diego plans to build an experimental cluster of about 2,000 retired Pixel phones to support systems and parallel-computing courses and to test long-run reliability and operational costs.
- Researchers frame the idea as a way to cut manufacturing-related emissions and reduce e-waste, but they note it is aimed at modest cloud and educational use and raises open questions about durability, repair, network and power efficiency, security, and cost versus purpose-built servers.