The AI infrastructure market is undergoing a tectonic shift, and at the center of this process is not Nvidia, but Apple. My sources and analysis of market data indicate that OpenAI has carried out a massive purchase of tens of thousands of Mac mini and Mac Studio computers over recent quarters. This is not just a change of equipment fleet—it is a strategic maneuver aimed at creating a fundamentally new layer of computing for training agentic systems.
The key task is to teach AI agents to interact with desktop operating systems the way humans do: clicking, typing, and managing the interface. This requires not virtual sandboxes, but real, isolated environments. The compact Mac mini is ideally suited for deploying thousands of such independent "workstations." At the same time, the Mac Studio M5 Ultra with support for up to 512 GB of unified memory becomes an ideal host for heavy language models that serve as the "brain" of these agents.
Technical details and the arms race
The numbers are impressive. The flagship Mac Studio M5 Ultra with a memory bandwidth of 1.2 TB/s can fully load the open LLM GLM-5.3-Flash with 320 billion parameters, which takes up to 300 GB of RAM. For comparison: on a traditional desktop, this would require a setup of ten flagship RTX 5090s, which would result in enormous energy consumption and performance loss on interconnects.
It is telling that Anthropic, OpenAI's main competitor, is solving a similar problem in a different way by renting Mac mini capacity from Amazon Web Services. This confirms that the Apple ecosystem is becoming a critically important element for AI labs, even despite the fact that Apple itself has long been considered a laggard in the AI race. As early as late April 2026, Tim Cook acknowledged the shortage of Mac mini and Mac Studio, directly citing agentic AI tools as the reason.
It is obvious that Apple hardware will not replace Nvidia GPU clusters for training base models. However, it is forming a new infrastructure layer—a virtual "office" where trained neural networks practice solving real-world tasks. This changes the rules of the game, because control over this layer may prove no less important than owning the computing power for training.
My conclusion: We are witnessing the emergence of a new market for "agent infrastructure," where Apple, even without being a traditional AI giant, is becoming a monopolist thanks to a unique combination of hardware and ecosystem. Investors should closely watch how this trend will affect the capitalization of companies that fail to restructure their strategies in time.