Apple is alleged to be releasing its own chips for artificial intelligence (AI) data center servers within the second half of next 12 months. Server chips are a field that cloud specialists resembling Amazon, Microsoft (MS), and Google have been producing. This shows their intention to thoroughly prepare for AI services.
Apple specialized media MacRumors reported on the 4th (local time) that Apple plans to mass-produce its own AI server ‘M5’ processor within the second half of 2025.
In line with this, Apple has been developing its own AI server ‘M5’ processor with TSMC’s 3 nanometer (nm) process. It’s currently within the small-scale test production stage, and plans to mass produce it for brand new Macs and AI cloud servers in 2025 and 2026.
Earlier in April, it was reported that the corporate was developing its own chips for AI data centers using TSMC’s 3nm process.
Apple has used third-party chips for its cloud and data center servers, but recently it has been using multiple M2 Ultra chips designed exclusively for desktop Macs. By the top of 2025, it plans to put in M4 chips designed exclusively for Macs in its AI servers.
If ‘Apple Silicon’ for data centers appears like this, it’s going to turn out to be an organization that manufactures its own server chips, like cloud-specialized corporations resembling Amazon, MS, and Google.
Apple is alleged to be using advanced ‘SoIC (System on Integrated Chip)’ packaging technology for the M5 chip to satisfy silicon demands for each consumer Macs and AI servers for data centers.
SoIC technology allows chips to be stacked in a three-dimensional structure, providing higher electrical performance and warmth management compared to traditional two-dimensional chip designs. This explains why the M5 was designed for dual use as each a Mac and an AI server.
The M5’s dual-purpose design is believed to follow Apple’s traditional strategy of vertically integrating its supply chain for AI capabilities across computers, cloud servers, and software. Designing its own server chips would allow the corporate to optimize its hardware for software needs.
Particularly, the undeniable fact that Apple is developing its own AI chip for data center servers suggests that it has entered right into a full-scale AI competition.
We’re pursuing an on-device AI priority policy so as to add generative AI functions to iPhones, iPads, Macs, etc., but cloud-based services are essential for high-performance AI work.
Cloud support can be essential for older devices which have difficulty operating on-device AI.
Reporter Park Chan cpark@aitimes.com
