OpenAI and other AI labs are acquiring tens of thousands of Mac minis and Mac Studios to train computer-use agents that can autonomously perform multi-step tasks. The company is seeking more hardware, but the most powerful models have been in short supply for months due to a memory chip shortage. Anthropic is also participating by renting Mac minis through AWS. The Mac mini is gaining popularity as a local AI computing platform, partly driven by the OpenClaw hype. Its strong chips, shared memory, and efficient cooling make it well-suited for long-running AI workloads. However, alternatives like Nvidia’s DGX Spark offer a different approach with dedicated GPU power and CUDA support. The open-source software Exo allows users to link multiple Macs into a cluster for running large models locally. Peter Voell, a former member of OpenAI’s computing infrastructure team, is also developing an Apple-based cloud service called Mount Thor. Apple’s Mac revenue increased nearly 29% to $10.4 billion in the June quarter.
The Mac mini’s appeal lies in its unified memory architecture, which simplifies data access for AI training. In contrast, Nvidia’s DGX Spark uses a different strategy with dedicated GPU power and Tensor cores, emphasizing CUDA compatibility. The open-source Exo software enables users to combine multiple Macs into a cluster, allowing for the execution of large AI models without relying on cloud infrastructure. This trend highlights the growing interest in local AI computing solutions.
Apple’s Mac revenue jumped nearly 29 percent to $10.4 billion in the June quarter, reflecting the increasing demand for its hardware in AI development. The Mac mini’s role in training computer-use agents underscores its significance in the AI landscape.
Source: thedecoder
