Minions pools idle Apple Silicon Macs into a distributed GPU cloud for ML inference

Apple researchers present Minions, a dynamic distributed virtual GPU cloud that aggregates idle GPUs across Apple Silicon macOS devices to run large-scale machine learning inference workloads.

The system adds flexible task execution and fault tolerance, letting teams use existing hardware without buying more GPUs. The authors say idle macOS devices can form a practical distributed inference platform where centralized GPU resources are constrained.