Researchers at the University of California San Diego are working on an unusual way to make computing more sustainable: turning retired smartphones into a data center. With support from Google, the team plans to build a 2,000-phone cluster using old Pixel devices and expects the system to go live in Fall 2026.

The project is aimed at reducing both the cost and carbon footprint of cloud computing. Google researchers say the effort is designed to give consumer electronics a second life instead of sending them to recycling or disposal once users upgrade. The approach also targets the emissions tied to manufacturing new hardware, which can be harder to cut than day-to-day energy use in data centers.

## Why old phones still matter

The idea rests on a simple observation: many smartphones are replaced before their core computing parts wear out. While owners often trade in devices after a few years, the processors, memory, and storage inside them can still be useful. The researchers note that modern phone processors can perform at levels comparable to, and in some cases better than, the per-core performance of some server chips.

That does not mean a phone can simply be dropped into a datacenter rack and used as-is. Consumer devices come with parts that are unnecessary in a server setting, including displays, batteries, cameras, and enclosures. Some of those components, especially batteries, are not suitable for datacenter environments. Before deployment, the devices need to be stripped down so that only the motherboard and its core compute components remain.

The motherboard is especially important from a sustainability standpoint. According to the Google researchers, it accounts for about half of a Pixel phone’s embodied carbon. Reusing that part, they argue, offers the biggest opportunity to avoid the emissions associated with manufacturing new equipment.

## From Android phones to Linux clusters

The plan also involves software changes. Although Android is built on Linux, the mobile-focused software layer has to be replaced with a more general-purpose Linux distribution. That shift removes features designed for consumer phones but not needed in cloud computing.

To manage a large number of devices, the project uses containerized applications running under Kubernetes. The phones are grouped into self-managing clusters of 25 to 50 devices, which together can deliver the performance needed for cloud tasks that would ordinarily run on a server.

Google says benchmarking suggests that roughly 25 to 50 phones can match the performance of a modern server in certain workloads. The company and UC San Diego researchers are using that idea to build a low-carbon cloud platform for university use.

## A cloud platform for classes and research

The planned 2,000-phone deployment is intended to support teaching and research in computer science, including courses such as Parallel Computation and Systems Programming. Google says many university cloud workloads are modest enough to run on a single smartphone, including grading systems and notebook hosting.

Early tests show promise. In one experiment, a 20-phone cluster was able to handle peak submission traffic for a class with more than 75 students, while keeping grading latency below that of a default AWS backend. The researchers say a full 2,000-phone system could support around 100 such classes at the same time.

Beyond its direct use in education, the system will serve as a testbed for studying how consumer hardware holds up under extended, real-world workloads. The project is also meant to demonstrate whether large-scale smartphone reuse can become a practical model for lower-carbon computing infrastructure.

If the deployment performs as expected, it could offer universities a cheaper alternative to traditional cloud services while also reducing the demand for newly manufactured servers.