What we know

Apple researchers have created an internal system called Minions that turns idle Apple Silicon Macs into a distributed virtual GPU cloud. This system runs AI inference jobs by pooling the unused processing power of these Macs, especially when dedicated GPUs are in short supply. The design of Minions allows it to potentially scale across many devices globally, making use of otherwise idle hardware for AI workloads.

Why it matters

The growing demand for AI computation often outpaces the availability of dedicated GPUs, creating a resource bottleneck. Minions offers a way to tap into the unused capacity of Apple Silicon Macs, expanding the available compute power for AI inference without needing additional specialized hardware. This could help organizations better manage AI workloads by leveraging existing devices more efficiently.

What is still unknown

Details about the system’s accuracy, the range of AI models or languages it supports, and plans for any public release have not been disclosed. No independent tests of performance or energy efficiency are available.