Apple has begun shipping its latest Mac mini and Mac Studio desktop models as of Tuesday, pitching them to corporate buyers as on-premises artificial intelligence machines that eliminate variable cloud fees tied to token usage. The company currently holds roughly 4.6% of the enterprise desktop market, while Windows-based systems account for 91.3%.
With the new hardware, Apple is targeting businesses that want to run large AI workloads—such as code generation and complex operational tasks—without relying on rented data center capacity. The key selling point is local AI processing, which means enterprises no longer need to pay per-token charges to cloud providers like OpenAI or Anthropic. A stack of four machines can run trillion-parameter models from a single wall outlet.
The most fully configured setup carries a price tag of roughly $20,000. "There are no per-token costs. You just use the machine over and over again," said Johny Srouji, Apple's chief hardware officer. Apple has demonstrated a configuration where four Mac Studio units are linked together to run a trillion-parameter AI model tasked with identifying and fixing a graphics coding error—a job that would typically require substantial data center resources.
That multi-machine setup runs over Thunderbolt 5 connectivity using remote direct memory access (RDMA). Apple claims this approach can boost distributed AI inference speeds by up to three times compared to a single machine. The M5 Ultra Mac Studio can be configured with up to 512GB of unified memory, though that specific variant is not expected to ship until the end of October.
Pricing for the new lineup starts at $899 for the Mac mini equipped with the M6 chip, while the Mac Studio with the M5 Ultra begins at $5,499. In August, reports noted that the entry-level Mac mini saw a $100 price increase, which Apple attributed to rising memory costs.
Srouji noted that the chips are co-designed across the product range, meaning AI models built on Apple machines can scale down seamlessly from the most expensive Mac Studio to iPhone and iPad devices. Apple remains a distant challenger in the broader enterprise computing space, with about 4.6% share in corporate desktops and laptops compared to Windows' 91.3%.
Microsoft is competing for the same buyers with its own on-device AI push. Microsoft CEO Satya Nadella has said the company plans to consolidate many of its AI features into a single super app for Windows. Meanwhile, Nvidia and PC manufacturers are preparing new AI desktops expected to headline Microsoft's Windows event in San Francisco next month. Microsoft has stated it worked with chip partners to streamline AI workloads through its Windows ML tools, with RDMA and related capabilities flagged as active investment areas.
Nvidia CEO Jensen Huang downplayed speculation about direct competition with Apple when he unveiled the company's new PC chip earlier this summer. Apple's current trajectory traces back to 2020, when it transitioned to Apple Silicon, combining computing and memory into a single design to improve battery efficiency.
Comments