OpenAI Reportedly Acquired Mac Hardware to Train Computer-Use AI Agents

OpenAI reportedly acquired tens of thousands of Mac minis and Mac Studios to train computer-use AI agents through reinforcement learning.

OpenAI reportedly purchased tens of thousands of Mac minis and Mac Studios for training computer-use artificial intelligence agents through reinforcement learning, according to The Information, as cited by Cointelegraph.

The reported hardware deployment highlights the growing importance of physical computing environments in the development of AI systems designed to interact directly with computers. Anthropic, meanwhile, is reported to rent similar Mac hardware through Amazon Web Services.

OpenAI Expands Mac Hardware for AI Training

According to the report, OpenAI has acquired tens of thousands of Mac minis and Mac Studios for use in training computer-use agents. The systems are reportedly being used in reinforcement learning, a training approach in which AI models learn by receiving feedback based on their actions.

The reported purchases point to a significant investment in computing hardware specifically suited to systems that can operate computers rather than simply generate text or images.

Computer-use agents are designed to perform tasks through computer interfaces, requiring them to interpret what appears on a screen and take actions using applications and other digital tools. Training such systems can require environments where an AI model can repeatedly interact with software and receive feedback on the results.

The Information's report, as referenced by Cointelegraph, did not provide further details about the total cost of OpenAI's reported hardware purchases or the specific training environments involved.

Anthropic Takes a Different Hardware Approach

Anthropic is reportedly pursuing a different arrangement for similar hardware. Rather than purchasing comparable Mac systems outright, the company rents Mac hardware through AWS, according to the cited report.

The comparison illustrates two approaches to accessing specialized computing infrastructure. OpenAI's reported acquisition involves directly obtaining large quantities of Mac hardware, while Anthropic is said to be accessing similar machines through a cloud-based arrangement.

The hardware is particularly relevant because Apple's Mac systems use its own silicon architecture, creating a computing environment that differs from the large-scale GPU infrastructure commonly associated with AI model training.

The available information does not indicate whether the hardware used by either company is dedicated exclusively to computer-use agent training or whether it supports other research and development activities.

Computer-Use Agents Become a Key AI Focus

The reported hardware investments come as AI companies work to develop systems capable of taking actions on computers. Such agents require more than the ability to produce responses; they must also interact with software environments and execute sequences of actions.

That makes the availability of suitable physical or virtual machines an important part of the development process. For reinforcement learning in particular, access to repeatable computer environments can provide a setting in which agents can perform tasks and receive feedback.

OpenAI's reported purchase of tens of thousands of Mac minis and Mac Studios, alongside Anthropic's reported use of rented Mac hardware through AWS, offers a glimpse into how leading AI companies are building infrastructure for this area of research.

Further details from The Information would be needed to establish the exact scale, configuration and purpose of the reported Mac deployments.

writer: Ethan Collins
Crypto Journalist
Ethan Collins reports on developments across the cryptocurrency and blockchain sector. His work covers market movements, protocol updates, regulatory changes, and emerging trends in digital assets.
He focuses on presenting complex topics in a clear and accessible manner for a broad readership.