A low-carbon computing platform from your retired phones

The strongest reaction was that the research is attacking the back half of the problem while the industry keeps creating the front half. Old phones become e-waste less because their CPUs are useless and more because vendors stop support, lock bootloaders, and ship proprietary firmware that nobody else can maintain. Even if you replace Android with a minimal Linux-style userspace, the hard part sits below that. Modem, Wi-Fi, Bluetooth, and other low-level firmware often remain frozen. That makes these devices hard to trust on real networks and hard to keep stable across OS updates. Several people pointed out that this is why phone reuse works best when the original vendor controls the hardware stack, as Google does with Pixels, but falls apart as a general market solution. From there, the conversation landed on economics and lifecycle math. The cluster idea feels plausible as a research project, a hobby system, or a tightly controlled internal deployment with lots of identical devices. It looks much shakier as a commercial product. Dismantling, testing, reimaging, rack design, replacement parts, and generation-by-generation software support all add labor to what only looks cheap if you count the phones as free. That pushed people back to a simpler conclusion: the biggest carbon win is usually extending the phone’s useful life as a phone, not turning it into a server after support ends. That did not kill enthusiasm. Plenty of people have already used old phones for Termux-based services, PostmarketOS experiments, k3s clusters, web hosting, and Raspberry Pi-style batch compute. What the project really surfaced is how much latent compute is trapped inside consumer hardware, and how much policy and vendor behavior determine whether that compute is reusable at all.