In recent months, I have been observing an extremely telling trend: OpenAI has purchased tens of thousands of Mac mini and Mac Studio computers. This is not just an office equipment purchase—it is a strategic move that redefines Apple's role in the artificial intelligence ecosystem. The focus is on preparing AI agents to work independently with desktop operating systems, including training through trial and error (reinforcement learning).
Why Mac, not GPU clusters?
The key point here is architecture. While Nvidia dominates in training base models, agentic tasks require not raw computing power, but real environments. Thousands of independent virtual "offices" where the neural network learns to click, type, and interact with the interface. The compact Mac mini M6 is ideal for deploying such environments, while the Mac Studio M5 Ultra with support for up to 512 GB of unified memory allows hosting the heaviest LLMs locally.
A telling example: the GLM-5.3-Flash model with 320 billion parameters fits entirely into the memory of a Mac Studio. For comparison, on a classic desktop, this would require a setup of ten flagship RTX 5090s—with monstrous power consumption and loss of bandwidth. This is not just an alternative; it is a fundamentally different approach to infrastructure.
Competitors and shortage
Anthropic, OpenAI's main competitor, is solving similar problems differently—by renting Mac mini capacity from Amazon Web Services. This confirms that the demand for such solutions is enormous. Notably, back in April 2026, Tim Cook publicly acknowledged the shortage of Mac mini and Mac Studio, directly citing agentic AI tools as the reason. Apple has even adapted its marketing, positioning the devices as solutions for "always-on local agentic computing."
Neither OpenAI nor Anthropic disclose the exact volumes of purchases or the breakdown by model, but the essence is clear: we are on the verge of forming a new infrastructure layer. This is not a replacement for GPU clusters, but a complement to them—a virtual "office" where trained models practice on real tasks.
My conclusion: Apple, unexpectedly for many, is becoming a key beneficiary of the AI agent race. If the trend continues, we will see not just growth in Mac sales, but the formation of a new niche of specialized computing where the Apple ecosystem will play a system-forming role.