Verified, then logged.
Sun 22 Nov 2026 · 00:13 KST / 15:13 UTC한국어 | ENSearch /

Breaking01 / AI

Apple's Minions pools idle Mac GPUs into a virtual GPU cloud

An Apple research paper turns GPUs that normally sit idle into inference capacity, targeting the cost and wait times of centralised infrastructure.

Idle silver laptops and compact desktop computers sit networked on long desks in an empty office at dusk, their small status lights glowing. AI-generated image

Apple published a paper on its machine learning research site on 12 Oct 2026 describing Minions, a virtual GPU cloud that pools idle graphics processors across macOS devices to run large-scale machine learning inference.

The paper defines Minions as a dynamic, distributed virtual GPU cloud. It gathers idle Apple Silicon GPUs scattered across multiple macOS devices and runs inference on them. Inference is the work of producing results from a trained model.

The paper says the GPUs in modern Apple Silicon macOS devices are powerful but largely idle during normal use. That spare capacity is the resource Minions draws on.

Growing demand for GPU-accelerated inference has exposed the limits of centralised GPU infrastructure, according to the paper. It lists high cost, limited elasticity and long waits when resources are short. Elasticity is the ability to scale capacity to match demand.

Apple has not disclosed in its public documents whether it will offer Minions as a product.