Doug Brooks, Apple’s senior product manager for Apple chip products, recently said in an interview with technology media The Deep View that Mac mini and Mac Studio have become the preferred devices for many developers and teams to run AI agents. This interview was completed shortly before the opening of WWDC 2026, and Brooks focused on the relationship between Apple’s self-developed chip strategy and local AI computing trends.
Brooks said the company is seeing "amazing demand" from the two desktop Macs, especially in agent-based workload scenarios that require continuous operation for long periods of time. He pointed out that many users hope to have a system that is “completely controlled by themselves, relatively isolated from the main computer, and can run 24 hours a day, 7×7”, and the Mac mini just meets such needs.
He also emphasized that many AI tools themselves give priority to Mac support, and are even Mac exclusive, which further consolidates Mac's position in the developer community, including in cutting-edge AI laboratories, where Mac devices are already very common. In his view, this software ecological advantage is inseparable from Apple’s long-term investment in AI computing at the hardware level.
When talking about the AI computing architecture, Brooks particularly emphasized that Agent-type AI is not just a “GPU problem,” but a “whole-chip” collaborative work. He said that today's workloads are no longer just the GPU "biting" large language models, but the entire chip is cooperating to complete tasks, tool calls, and various operations surrounding these processes, which just takes advantage of the system-level advantages of Apple chips.
Brooks traces Apple's leadership in modern AI to chip design decisions made before LLM even existed. He took the Neural Engine as an example, pointing out that it is specifically designed for efficient matrix operations, while the little-known neural acceleration units inside the CPU are responsible for processing extremely delay-sensitive tasks such as speech.
On this basis, Apple has added neural acceleration modules to the GPU in recent years, which has improved the overall AI performance from iPhone-level chips to Mac's top SoCs. Brooks attributed this progress to Apple's holistic design methodology - customizing chips for specific devices and allowing hardware and software to be developed simultaneously.
Brooks also described a trend of moving AI computing from the cloud to local. He believed that this change stems from privacy, security, and the rising cost of reasoning as Token consumption continues to rise. But in the future he envisions, AI agents will operate in a hybrid fashion: the agents will decide which tasks to complete locally and which to handle in the cloud.
At the specific application level, Brooks proposed the concept of so-called "transparent AI" to describe functions that are distributed throughout the iPhone and iPad systems and quietly play a role in third-party applications without deliberately flaunting "AI identity." As examples, he cited Draw Things, an image-generating tool that runs across iPhone, iPad, and Mac, and apps like SwingVision, which uses the iPhone camera to perform real-time analysis of tennis and pickleball games.
Regarding the speed of development of AI technology itself, Brooks bluntly stated that "the speed of development of AI now is really crazy" and admitted that it is difficult for him to imagine where the entire industry will be in one year, three months, or even one month. He believes that this rapid evolution is both a challenge and a concentrated test of Apple’s self-developed chips and local computing power layout for more than ten years.