May 10, 2024 – According to a recent report by Bloomberg’s Mark Gurman, Apple is now deploying its high-end chips, such as the M-series, in cloud computing servers designed to tackle intricate AI tasks. Meanwhile, more straightforward AI-related functions will be handled locally on iPhones, iPads, and Macs.
The first wave of AI server chips was introduced with the M2 Ultra, featured in last year’s Mac Pro and Mac Studio. Apple intends to roll out AI servers based on the upcoming M4 chip.
Previous reports have hinted that Apple might be developing its own AI server chips, utilizing TSMC’s 3nm process, with mass production estimated to begin in the latter half of 2025.
For basic AI duties, like summarizing missed iPhone notifications or texts for users, the on-device chips suffice. More complex operations, such as generating images or composing lengthy email responses, might require cloud processing.
This shift signifies a change in Apple’s strategy. The company had previously emphasized on-device processing as a superior approach to ensuring security and privacy. Sources involved in the “Apple Chips in Data Centers” server project revealed that Apple incorporates a component within its processors, known as Secure Enclave, to segregate data from security vulnerabilities, thereby protecting user privacy.
Gurman added that Apple has invested hundreds of millions of dollars in this initiative over the past three years, albeit without any product launches yet. Additionally, Apple has been engaging in negotiations with Google and OpenAI to potentially offer a ChatGPT-like service on its devices.