A quiet but significant infrastructure revolution is taking place in the world of artificial intelligence. OpenAI, one of the industry's leaders, has acquired tens of thousands of Mac mini and Mac Studio computers over the past few months. This is not just a hardware purchase — it is a strategic move that is reshaping the traditional understanding of what resources are needed for AI development.

My analysis shows that Apple devices are being used to train AI agents to interact with desktop operating systems (computer-use agents) and for reinforcement learning tasks. This is about creating a virtual environment where neural networks learn to "live" in the real digital world: clicking, typing, managing interfaces. This is a fundamentally different approach than simply training on GPU clusters.

It is telling that OpenAI's direct competitor, Anthropic, is solving a similar problem differently, renting Mac mini capacity from Amazon Web Services. This confirms: the demand for such computing is enormous, and the market is just taking shape. Incidentally, on August 25, Apple introduced an updated lineup based on the M5 Ultra and M6 chips, which fits perfectly into this trend.

Why This Works: Division of Roles

The technical specifications of the new devices clearly indicate their specialization in AI labs:

  • Mac mini M6. The compact form factor allows deploying thousands of isolated working environments. For training agents, a real operating system is critical, not a simulation.
  • Mac Studio M5 Ultra. Support for up to 512 GB of unified memory makes it possible to locally run and host the heaviest language models.

Let's take a concrete example: the flagship Mac Studio M5 Ultra with 512 GB of memory and 1.2 TB/s bandwidth can fully host the open LLM GLM-5.3-Flash with 320 billion parameters. Depending on quantization, the model takes up to 300 GB of RAM. On a regular PC, this would require a setup of ten flagship RTX 5090s with monstrous power consumption and performance loss.

Back in April 2026, Tim Cook, then CEO of Apple, publicly acknowledged the shortage of Mac mini and Mac Studio, directly citing agentic AI tools as the reason. The company is already adapting its marketing, positioning the devices as solutions for "always-on local agentic computing."

It is important to understand: Apple hardware will not replace Nvidia GPU clusters for training base models. But it is forming a new infrastructure layer — a virtual "office" where trained neural networks practice on real tasks. And this is changing the market.

My verdict: We are witnessing the birth of a new segment of AI infrastructure, where Apple unexpectedly becomes a key supplier. OpenAI and Anthropic, by buying up Macs, are effectively voting for this approach. The only question is whether Apple can hold this niche or whether we will see specialized "agentic" servers from other manufacturers. For now, this is the clearest signal of where the AI industry is heading.