AI Pulse · Under the Radar

Apple appears to have underestimated how many businesses want to run AI work on their own hardware. Demand for Mac Mini and Mac Studio machines used for local AI workloads reportedly exceeded Apple’s forecasts, pushing refreshed models into an August release rather than October. They were still sold out into the third week of September, with memory supply identified as the constraint.

That matters because this is not just interest in AI software. It is demand for the physical compute needed to run models inside a company’s own environment. For some firms, local AI can mean keeping sensitive documents, customer data, financial information, or internal processes off external cloud services. It can also give teams more control over cost and performance once workloads become regular and predictable.

Don’t read this as a reason to rush out and buy Macs. Local AI has real trade-offs, including upfront hardware cost, setup, model maintenance, and the need to decide which work genuinely benefits from staying in-house. But the supply pressure is a useful signal. More operators are treating AI infrastructure as a practical business decision, not simply another software subscription.

This is the kind of thing we build into an AI command centre, helping teams see where cloud tools make sense, where local processing may matter, and what the operating model should be before money gets committed.