DARKBLOOM RUNS AI INFERENCE ON IDLE MACS

Darkbloom transforms idle Mac computers into distributed inference nodes for running large language models locally. The platform allows users to contribute unused processing power while keeping all data private and on-device. Key features include automatic resource management—models run only when machines sit idle—and encrypted communication between nodes. Users retain complete control over which models run and what data processes through their hardware. The approach addresses two pain points: expensive cloud inference costs and privacy concerns with centralized AI services. By distributing compute across consumer hardware, Darkbloom reduces reliance on data center infrastructure. The project generated 106 points and 62 comments on Hacker News, indicating developer interest in decentralized AI infrastructure. Implementation details and technical specifications are available at darkbloom.dev.