Why OpenAI is buying thousands of Mac minis to train AI agents
Why is OpenAI buying thousands of Mac minis? The compact desktops could offer a new way to train and test AI agents at scale. Here's all that you need to know!
OpenAI is giving the Mac mini a rather unexpected AI job. The AI company has reportedly bought tens of thousands of Apple Mac mini and Mac Studio computers as it works on training AI agents, according to The Information.
This raises eyebrows because advanced AI is usually associated with huge Nvidia-powered data centres, not compact desktops sitting on a desk. Here's what's happening!
Why OpenAI wants Macs
AI agents are different from regular chatbots. Instead of simply answering a question, they can perform a series of tasks, such as opening apps, writing and testing code, organising files or working through software. Training these systems means giving them plenty of opportunities to practise those actions and learn from the results.
According to reports, OpenAI is using Macs for reinforcement learning. This is a training method where an AI system tries an action, receives feedback and adjusts its behaviour. It can be particularly useful for computer-use agents that need to understand screens, buttons, windows and software workflows.
That makes a desktop computer surprisingly useful. The AI is not necessarily trying to learn how to crunch the largest possible dataset. It is learning how to operate a computer and complete tasks reliably.
What makes Apple Silicon interesting
The Mac mini and Mac Studio use Apple Silicon and a unified memory architecture. In simple terms, the processor and graphics components can access the same pool of memory, which can be useful for running demanding AI workloads locally.
Their small size is another advantage. Instead of relying entirely on a remote cloud environment, organisations can deploy large numbers of compact machines for specialised workloads. Anthropic has also reportedly rented Mac minis through Amazon Web Services for similar work, suggesting the idea has wider appeal.
That does not mean Macs are about to replace Nvidia GPU clusters. The biggest AI models still require enormous amounts of specialised computing power. But agents introduce different workloads where flexible, relatively compact systems can be useful.
Apple could be an unexpected winner
The reported demand could also give Apple’s Mac business a new role in enterprise AI. India Today reported that Mac revenue rose nearly 29% year on year to $10.3 billion in the latest quarter, making it Apple’s fastest-growing hardware category.
The report also linked stronger AI demand with recent Mac mini and Mac Studio updates. The opportunity goes beyond selling more computers to consumers for Apple. If AI companies increasingly use Macs for development and agent training, its desktops could quietly become part of the AI infrastructure stack.
Meanwhile, OpenAI is building smarter AI agents that may also require a lot of computers for them to learn how to use.