Apple has reportedly started the development of processors dedicated specifically to the field of servers and AI. Their launch would be planned for next year, with the technical basis of 3 nm engraving from a certain TSMC.
Apple soon to be engaged in a frantic race for performance and energy efficiency in the lucrative server and AI market? Not quite, but the firm would in any case be interested in this sector. This is what a leaker on the Chinese social network Weibo. “Phone Chip Expert” (who claims to have 25 years of experience in the semiconductor industry, and having notably worked on Intel’s Pentium processors), reports that Apple has started internal chip development dedicated to this sector, for the moment mainly invested by Nvidia.
According to this informant, who has already been correct several times in recent years, the Cupertino giant is aiming for the start of mass production for the second half of 2025. We also learn that its first chip dedicated to servers and AI would be based on 3nm engraving from TSMC. A process that Apple has already mastered since it is used on the A17 Pro and M3 chips, in particular.
Soon servers designed in-house by Apple?
This information seems credible. Not long ago, we learned that TSMC’s 2nm engraving would not be able to be exploited by Apple for at least two years. The firm would therefore rely, in the meantime, on the most advanced TSMC process currently available. And therefore on 3 nm engraving.
For Apple, designing a chip dedicated to servers and AI in-house would make sense. It would firstly be part of the group’s strategy, which has for several years now consisted of internalizing increasingly important parts of its supply chain. As rightly pointed out MacRumors, by designing its own server chips, Apple could adapt this hardware to the needs of its own software and services. A way for the firm to achieve, at least potentially, more efficient and effective technologies.
If for reasons of security and confidentiality, Apple currently favors calculations carried out locally on its devices, and not online, it goes without saying that the increasing use of AI will push the brand to take greater advantage of the cloud in the future. Having its own servers within one or two years, themselves equipped with chips designed in-house, could then be a significant advantage.
The fact remains that the lead mentioned by “Phone Chip Expert” is currently not very well supported. In the absence of additional information from other sources, caution therefore remains necessary.