As upgraded Mac mini and Mac Studio machines begin shipping Tuesday, Apple is pitching corporate buyers on local AI with no per-token cloud cost, including stacks of Mac Studios shown running about 1 trillion-parameter models from a wall outlet, against Microsoft and Nvidia desktop AI pushes.
When Apple’s upgraded Mac mini and Mac Studio desktops start shipping Tuesday, executives will pitch corporate buyers an unusual argument: the machines can be cheaper than renting data-center AI, Reuters reported from San Francisco. Configurations can cost nearly $20,000 and are meant for intensive local work such as coding or complex business tasks without paying tokens to cloud providers such as OpenAI or Anthropic.
The new Macs go head-to-head with desktop machines Nvidia and PC makers are rolling out, expected to feature at a Microsoft Windows event in San Francisco next month. Apple holds about 4.6% of the enterprise desktop and laptop market versus 91.3% for Windows, according to IDC’s Linn Huang.
Apple’s case rests on Apple Silicon’s unified memory, which packs compute and memory together and has made Macs strong at AI since 2020. Mac minis sold out as OpenClaw, an open-source agentic AI tool, took off in markets such as China. Mac Studios have quietly gained AI features including RDMA over Thunderbolt for chip-to-chip networking.
At this month’s launch event, Apple showed four Mac Studios networked to run an AI model with about a trillion parameters to find and fix a graphics coding bug, work that usually needs a data center, with the stack running off one wall outlet. “Once you have the machine on your desk, you’ve paid for it,” Apple chief hardware officer Johny Srouji told Reuters. “There’s no cost per token. You’re just using the machine again and again.” That unmetered local inference versus metered cloud tokens is the class of routing choice ThinkFreely is built to keep with the operator.
Microsoft is chasing the same on-device market Satya Nadella calls “unmetered intelligence” and said it is working with chip partners on Windows ML tools. Nvidia declined to comment; CEO Jensen Huang has downplayed competing directly with Apple and kept Nvidia’s stronghold in the data center. Srouji said models developed on Macs can scale up to Studios or down to iPhones and iPads because the chips share common designs.
This brief covers the Tuesday ship-day corporate local-AI cost pitch as reported by Reuters. It is not a rewrite of August’s product-launch coverage.