Apple Pitches On-Device AI Macs to Enterprises as Cheaper Than Cloud Tokens

Apple is making an unusual pitch to corporate buyers as its upgraded Mac Minis and Mac Studios begin shipping on Tuesday: running AI workloads on a desktop can be cheaper than renting time in a data centre.

The new machines, which can cost nearly $20,000, are designed to handle intensive AI tasks such as writing code or complex business work locally, avoiding the per-token charges levied by cloud providers like OpenAI and Anthropic. Tokens are the fundamental unit of AI computing.

The Macs will compete directly with new desktop machines from Nvidia and PC makers, which are expected to be showcased at a Microsoft Windows event in San Francisco next month.

Apple's push into corporate computing faces a steep challenge. According to IDC's Linn Huang, Apple holds about 4.6% of the enterprise desktop market, compared with 91.3% for Windows. Co-founder Steve Jobs was famously ambivalent about enterprise computing, but Apple has found itself well placed for AI desktops partly because of its long focus on power efficiency.

When Apple introduced its first Apple Silicon chips in 2020, it combined computing and memory — separate components in most PCs — into a unified memory architecture to improve battery life. That close coupling, which Nvidia and others have only recently moved toward, also made Macs effective at AI. Demand for Mac Minis rose as OpenClaw, an open-source agentic AI tool, gained traction in markets such as China.

Mac Studios were initially aimed at content creators, but Apple has quietly added AI-oriented features over the past two years, including bespoke chip-to-chip networking known as RDMA over Thunderbolt. At its launch event this month, Apple demonstrated four Mac Studios linked together to run an AI model with a trillion parameters — a measure of complexity — to find and fix a graphics coding bug. Such tasks typically require a data centre, yet the stack of Macs ran off a single wall outlet.

"Once you have the machine on your desk, you've paid for it. And I believe we provide absolutely great value, not only in terms of performance, but cost," said Johny Srouji, Apple's chief hardware officer. "There's no cost per token. You're just using the machine again and again."

Microsoft is pursuing the same market for what CEO Satya Nadella calls the "unmetered intelligence" of on-device AI, and has said it plans to fold many AI features into a "super app" for Windows. But Microsoft's historic lead in corporate computing means supporting hardware from a wide range of vendors, which can complicate work for Windows developers seeking to optimise for a given chip. Microsoft said it has been working with chip partners to streamline AI work through its Windows ML tools, and that features such as RDMA are an active area of investment.

Nvidia declined to comment. At the launch of its new PC chip this summer, CEO Jensen Huang played down any intention to compete directly with Apple, saying the company is focused on expanding what Windows PCs can accomplish. Nvidia's stronghold remains the data centre.

Srouji said Apple is pitching corporate buyers on the idea that on-device AI models developed on its devices can scale up to its priciest Mac Studios or down to its cheapest iPhones and iPads, because their chips share common principles and designs.

"There is so much power you can pump into the data centre," Srouji said. "We're offering a variety of products that customers can choose based on their own needs: Which computer, which product do I need?"