Old Smartphones to Become Low-Carbon Cloud Servers in Google-Backed Project

Retired smartphones may soon find a new life as components of low-carbon cloud computing systems, thanks to a collaboration between Google and researchers at the University of California San Diego.

The initiative, known as phone cluster computing, involves repurposing the motherboards of old phones, stripping away non-essential parts like batteries and screens, to create a network of small, interconnected servers. This approach aims to combat the growing problem of electronic waste while addressing the increasing demand for computing power driven by advancements in artificial intelligence.

Google and UC San Diego plan to launch a data center constructed from 2,000 Pixel smartphones in the fall of 2026. This facility is intended to provide affordable cloud computing resources for students and researchers, thereby reducing the necessity for newly manufactured server hardware.

Transforming Old Tech into Computing Power

Phone cluster computing begins with dismantling retired smartphones to isolate the motherboard, which contains the processor, memory, and storage. These motherboards are then loaded with a general-purpose Linux system, offering more flexibility than the original Android operating system for cloud workloads. By grouping these individual boards into clusters, they can function collectively as a distributed computing platform.

This project addresses the significant environmental impact of producing new server hardware. Google highlights that the embodied carbon—emissions generated during mining, manufacturing, and shipping—is largely associated with the motherboard, making its reuse a key factor in reducing the carbon footprint of data centers.

Technical Feasibility and Workloads

While a single phone board has fewer resources than a traditional server, modern smartphone processors can exhibit competitive single-threaded performance. Early experiments suggest that clusters of phone boards can handle specific tasks efficiently. A 20-phone cluster, for instance, demonstrated the capability to manage peak submission rates for a class of over 75 students, with grading latency remaining below that of a standard AWS backend.

The UC San Diego deployment, slated for fall 2026, is expected to support approximately 100 computer science classes concurrently and provide computing power equivalent to about 50 traditional servers, at a significantly lower cost.

Challenges and Future Potential

Despite the promise, phone cluster computing faces hurdles. Unlike dedicated data center servers built for continuous operation, cooling, and durability, phone motherboards are designed for intermittent personal use. Questions remain regarding their long-term reliability, potential cooling challenges with thousands of units operating simultaneously, and the labor costs associated with disassembly and maintenance.

Furthermore, these phone-based clusters are unlikely to replace the high-performance GPU systems required for advanced AI training. Instead, they are best suited for less demanding cloud jobs, educational tools, and research tasks that align with the hardware limitations of smartphones.

The global e-waste problem is projected to reach 82 million tonnes by 2030, with a significant portion comprising discarded phones. This initiative offers a model for repurposing existing technology, turning obsolete devices into functional components and mitigating their environmental impact. The common thread in such second-life projects is the underutilized computing potential within hardware already in circulation.

Before recycling or donating old phones, users are advised to back up data, sign out of accounts, and securely wipe devices to protect personal information. Reputable recycling programs, trade-in options, and certified refurbishers are recommended for responsible disposal or reuse.