Apple uses idle Macs as GPU cloud for AI inference with ‘Minions’
Apple researchers detailed a system called Minions, which turns idle Macs into a virtual GPU cloud for AI inference tasks. The system splits jobs into subtasks, distributing them across Apple Silicon Macs connected to the network. This reduces wait times from hours to minutes by offloading stalled AI workloads, including embeddings and text generation, to participating devices. Each Mac runs a…
Key points
- Minions splits AI tasks into subtasks for idle Macs, cutting wait times from hours to minutes
- Uses Apple Silicon for inference only: embeddings via Hugging Face models, text via Ollama
- Participation is informal via Slack requests, with team-scoped data pools to address confidentiality
Minions operates on a pull-based execution model, where Macs fetch subtasks via REST APIs. Participation is informal, with teams requesting work via Slack or setting task priorities. Apple acknowledges concerns about data confidentiality, mitigating risks through team-scoped deployments and separate Mac pools. The system aims to cut infrastructure costs and manual scaling efforts, though electricity and hardware wear remain unaddressed. Apple’s latest quarterly filing highlights compute bottlenecks, making Minions a practical response to constrained GPU supply and rising costs. The paper also hints at future expansion to distributed model training and federated GPU pools across organizations.
The story so far
2 episodes →- Apple uses idle Macs as GPU cloud for AI inference with ‘Minions’this story
Apple researchers turn idle Macs into a virtual GPU cloud with ‘Minions’
Tom's Hardware · 14 October 2026
Loading the full article…
This text was published by Tom's Hardware and written by Shane Downing. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Hardware & Compute
All →- Nvidia reportedly invests in AI inference chip rival d-Matrix · 3 src
- Apple, Amazon, Microsoft launch AI-powered devices ahead of holiday season · 1 src
- NTT to build 1-trillion-yen Anthropic data center in Osaka · 1 src
- Alibaba’s Qwen Glasses N1 use Himax’s WiseEye HX6538 for always-on AI vision · 1 src
- Ukrainian modder tests leaked DLSS 5 on 1996 Quake with mixed results · 1 src
Comments
via GitHub Discussions