2,000 Retired Pixel Smartphones Repurposed into a Private Cloud

A team of computer scientists at the University of California, San Diego (UCSD) is collaborating with Google to develop a compute cluster comprising 2,000 retired smartphones. According to a report by The Register, the cluster is based on Google’s Pixel Fold smartphones, giving a second life to devices that would otherwise be discarded or left unused in drawers.

This project aims to demonstrate the practical utility of retired smartphones as a low-cost, low-carbon computing platform. As reported by The Register, the initiative was spearheaded by Jennifer Switzer, a former UCSD PhD student currently working as a postdoctoral researcher at Google.

Background of the Project

The average replacement cycle for smartphones is about four years. Even after this period, the core computing components inside devices—such as processors and memory—often remain fully functional. According to UCSD Associate Professor Ryan Kastner, “A tremendous amount of computational resources are being wasted, and recycling is the worst option for most smartphones.”

Switzer initially proved the concept by building a small-scale cluster using only a few smartphones. The project then rapidly expanded to the current scale of 2,000 devices. Google estimates that around 50% of a smartphone’s embedded carbon is concentrated in its motherboard. By reusing these motherboards, the project has the potential to significantly reduce the environmental impact associated with manufacturing new hardware.

Motherboard Extraction and Safety Measures

In the early stages of the project, the UCSD team conducted tests using unmodified smartphones. However, this method was deemed impractical and raised safety concerns. Google engineers expressed significant worries about bringing lithium-ion batteries into data centers due to their fire risk, ultimately ruling out the use of intact devices.

To address this challenge, Google has commissioned third-party companies to extract only the motherboards from the Pixel Fold smartphones. For the large-scale deployment planned for this fall, the clusters will consist solely of stripped-down motherboards, with the cases and batteries completely removed.

Computing Performance

The Pixel Fold is equipped with Google’s proprietary Tensor G2 processor, an Arm-based System on Chip (SoC) that includes two 2.85 GHz Cortex-X1 cores, two 2.35 GHz Cortex-A78 cores, and four 1.80 GHz Cortex-A55 cores, as well as a Mali-G710 MP7 GPU.

According to researchers, the single-threaded performance of these smartphone processors is comparable to, or in some cases even exceeds, that of multi-core, data center-grade processors. Moreover, smartphone chips are designed for high energy efficiency, which gives them a potential edge over data center chips for specific workloads.

Contributions to Environmental Impact Reduction

The issue of smartphone waste is becoming increasingly critical worldwide. While many retired devices are recycled, Kastner points out that recycling is essentially a process of losing valuable electronic components. By reusing motherboards intact, the project can help avoid the energy-intensive processes involved in material extraction, refinement, and manufacturing.

This initiative serves as a concrete experiment in transitioning the electronics industry from a “linear economy” (extract → manufacture → dispose) to a “circular economy.” Although the scale of 2,000 devices is still considered experimental, it is sufficient to demonstrate the potential for scaling up such projects in the future.

Future Prospects and Challenges

The large-scale deployment of the compute cluster is scheduled for this fall. However, specific details about the types of workloads that will be run on the cluster have not been disclosed. Since the project uses Arm architecture, compatibility with existing x86-based data center software stacks represents a significant challenge.

Additionally, smartphone motherboards pose unique constraints not typically found in data center servers. These include limited memory capacity, restricted network bandwidth, and the challenge of thermal management during prolonged operation. Overcoming these challenges could open up opportunities for enterprises and research institutions to acquire highly affordable compute resources.

Editorial Opinion

In the short term, this project provides a tangible demonstration of how electronics reuse can be a better alternative to recycling. For data center operators, the findings suggest that smartphone SoCs might outperform traditional data center chips in energy efficiency for specific workloads. This could encourage similar experiments by other organizations within the next three to six months.

In the longer term, over a span of one to three years, retired smartphones might emerge as a cost-effective computing resource for applications in areas like “edge computing” or “low-load batch processing.” However, this innovation could pose a challenge to the business models of existing cloud service providers, whose reactions will need to be closely monitored.

From the editorial team’s perspective, the establishment of safety measures such as motherboard extraction and battery removal has lowered the barriers to data center integration. Nonetheless, the costs associated with adapting to the Arm architecture and maturing the necessary management tools remain critical factors for practical implementation. The outcome of the large-scale deployment this fall will be crucial in addressing these challenges.

References

• The Register: 2,000 retired Google Pixel phones get a second life as a private cloud — Published on 2026-06-18