Apple is expanding the use of Mac computers in artificial intelligence tasks, emphasizing local data processing and reducing reliance on cloud infrastructure. The updated Mac Mini and Mac Studio are capable of performing complex operations directly on the device, including programming and working with large AI models. Configurations of such computers can cost nearly $20,000, but after purchase, the user does not pay for separate computing tokens to cloud providers. For the corporate market, this creates a different cost model: computing power is purchased with the equipment and then used repeatedly. Apple is trying to apply its experience in developing energy-efficient chips for mobile devices to this segment. However, the scale of the task remains significant: according to IDC, Apple's share in the corporate desktop and laptop market is about 4.6%, while Windows systems occupy 91.3%.
The foundation of the approach is the Apple Silicon architecture, where computing units and memory are closely linked. This organization allows for more efficient handling of AI workloads, where performance largely depends on the speed of data exchange with memory. Apple is further developing the capabilities of Mac Studio for distributed computing: computers can be combined into a single system using special inter-chip interaction technologies. In a recent demonstration, the company showed four Mac Studios working together with a model of about one trillion parameters. The system was used to find and fix an error in the software code. Such tasks are traditionally performed on data center infrastructure, whereas the assembled Mac configuration consumed energy from a single outlet. This scenario shows that Apple considers several connected computers not only as workstations but also as a scalable local computing platform.
Competition in this direction affects several market segments at once. Nvidia is developing its own solutions for AI computing on PCs, and Microsoft is working on integrating AI functions into Windows and optimizing software in collaboration with chip manufacturers. For the Windows ecosystem, an additional task remains to support a large number of hardware configurations, whereas Apple controls both the processors and the operating system, as well as the main components of its computers. At the same time, the company aims to scale the developed AI models across different devices – from powerful Mac Studios to iPhones and iPads, as they use common Apple Silicon architectural principles. In this model, local computing becomes part of a unified hardware ecosystem, and the specific power level is chosen depending on the task. However, the effectiveness of the approach will depend on the size of the models, memory requirements, the nature of the workload, and how often specific operations need to be performed locally.
Related articles
