# Monetizing idle Macs by turning them into AI inference nodes Apple Silicon owners can earn $120-$200 monthly by selling spare compute power - **Source:** [www.theregister.com - Articles](https://brifing.si/source/2c34ea50-d341-4a10-925d-12c012a9ad10) - **Published:** 2026-08-24 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/3cbce92e-3905-4c31-bb2e-2d6c071edd79 ## What is the Darkbloom project? Developed by Eigen Labs, Darkbloom is a distributed network that allows Apple Silicon Mac owners to monetize idle computing power by providing AI inference services. Launched as an experiment in April 2024, the network currently has 250 active machines that have processed 4.5 billion tokens, generating $102K in annual revenue. Participants with M1 or newer Macs can earn between $120-$200 per month by joining the network. ## How it works The system operates through Darkbloom Swift CLI software installed on users' Macs. AI inference requests are routed through a Go-based coordinator running in AMD SEV-SNP protected VMs. Computations are performed on Mac GPUs using Apple's MLX framework. For data security, the system utilizes Secure Enclave and blocks memory-reading APIs through Hardened Runtime. ## Security model Darkbloom offers a privacy model similar to Apple's Private Cloud Compute. Operations run within a single hardened Swift process with debug access blocked at system call level (PT_DENY_ATTACH). Data exists as plaintext only in the coordinator's hardware-encrypted memory, isn't logged, and is erased after processing. This protects user data and AI prompts from exposure. ## Popular models and earnings The top-earning model on Darkbloom is Qwen 3.6 35B A3B at $399/month, followed by GPT-OSS 20B at $52/month and Gemma 4 26B at $27/month. Daily earnings reach $26 for Qwen. The network has processed over 42 million inference requests to date. Average electricity costs are estimated at $2 per month per machine. ## Hardware requirements Participation requires an M1 processor or newer Mac. Mac minis are particularly preferred for their performance. Cloud provider rentals aren't allowed - only personally owned devices can join. The system connects via WebSocket, eliminating the need for port forwarding or firewall modifications. ## Industry position Darkbloom represents part of the decentralized AI infrastructure trend. Instead of relying on data center-heavy labs, it enables individuals to run open-weight models. Having started as a free service on OpenRouter, the network now operates as a paid provider. This model democratizes AI access while creating new income streams for individuals. ## Technical details The system uses the mlx-swift-lm library to run on Apple Silicon GPUs. No subprocesses, local servers, or inter-process communication are used during operations. The macOS kernel blocks external memory access to the process. These protections can't be disabled without rebooting, significantly enhancing security. ## Future potential Darkbloom pioneers a new economic model leveraging idle personal computing resources. As AI compute demand grows, such distributed networks may gain importance. The rise of open-source models could transform personal computers into revenue-generating AI infrastructure nodes. This approach combines decentralized technology with sharing economy principles. ## Sources covering this story - https://www.theregister.com/ai-and-ml/2026/08/24/conjure-cash-with-old-macs-by-linking-them-to-ai-inference-borg/5292002 — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._