The artificial intelligence market is undergoing a tectonic shift, and at the center of this process is an unexpected player — Apple. While all attention is focused on Nvidia's GPU clusters, OpenAI has quietly purchased tens of thousands of Mac mini and Mac Studio computers. This is not just a hardware procurement, but a strategic maneuver that is changing the rules of the game in training AI agents.
Agentic AI: Betting on Real Operating Systems
The key task OpenAI is solving is teaching neural networks to interact with desktop interfaces the way humans do: clicking, typing, and managing windows. For this, AI agents need not a virtual environment, but full-fledged operating systems. This is where the Mac mini, with its compact form factor, becomes an ideal tool for deploying thousands of isolated working environments. The second lineup — Mac Studio M5 Ultra — handles heavier tasks, allowing models with hundreds of billions of parameters to be hosted locally.
Technical Analysis: Why Apple Wins
The flagship Mac Studio M5 Ultra with 512 GB of unified memory and 1.2 TB/s bandwidth can load an entire LLM of the GLM-5.3-Flash level with 320 billion parameters into RAM. For comparison: on a traditional desktop, this would require linking ten flagship RTX 5090s, which would result in enormous energy costs and performance loss due to bus bottlenecks. This is not just savings — it is a paradigm shift in computing for agentic tasks.
Strategic Context and Competition
It is telling that OpenAI's main competitor, Anthropic, is solving a similar problem by renting Mac minis through the AWS cloud. This confirms: Apple is not just selling "hardware," but is shaping a new infrastructure layer for AI. Back in April 2026, Tim Cook openly acknowledged the shortage of Mac mini and Mac Studio, directly linking it to the boom in agentic AI tools. Apple's marketing is already adapted to the new reality — devices are positioned as solutions for "continuously running local agentic computing."
It is important to understand: Macs will not replace Nvidia GPU clusters for training base models. But they create a fundamentally new segment — a virtual "office" where trained neural networks hone their skills on real tasks. Neither OpenAI nor Anthropic disclose exact procurement figures, but the scale itself hints that we are on the verge of Apple becoming one of the key beneficiaries of the AI race, even without being a formal player in that industry.
My view: this is an underestimated signal for the crypto industry. If decentralized computing wants to compete with such giants, it should carefully study the efficiency of the "ARM chips + unified memory" combination. Perhaps it is this approach, rather than the fight for GPUs, that will become the next battlefield for AI infrastructure.