Apple Secretly Designing its Own Server CPUs For AI Processing: Report

After gaining massive advantages in the smartphone, tablet and PC markets by switching to its own in-house processors, Apple could be looking at taking on the server and datacentre segment next, at least for its own internal use. The rumour comes from Chinese-language Weibo user ‘Phone Chip Expert’, who claims a background in semiconductor design and has previously posted about various Apple projects while in development, via Macrumours. The rumour is not surprising, as multiple other companies including Microsoft, Amazon, Google, Nvidia, and Meta have also announced or been rumoured to be developing custom CPUs in order to maximise efficiency.

Apple has so far adopted its own A-series SoCs in its iPhone, iPod and iPad devices, while the more powerful M-series have now replaced Intel’s offerings to power the entire Mac desktop and laptop lineup. In addition, Apple designs S-series SiPs for smaller devices such as the Apple Watch, H-series chips for its headphones, R-series real-time sensor processors for the Apple Vision Pro headset, U-series ultra-wideband radio controllers, W-series wireless controllers, and other miscellaneous chips. The company has also famously been struggling to develop its own 5G modems for several years, as it continues to try to bring component design in-house rather than rely on outside suppliers.

The purported server chips are said to be targeted to a 3nm manufacturing process and would be fabricated by TSMC, with a late 2025 launch date. It is most likely that these will be used by Apple to run its own datacentres, although the company selling Mac servers or workstations based on them would not be out of the question.

AI processing would be a huge priority for the supposed server chip design team, and Apple has also constantly focused on power efficiency, particularly because it can maximise the benefits of developing its own hardware and software for its own purposes. The rumour also tracks with recent reports that Apple is accelerating the release of next-gen M4-based Macs and will push AI capabilities as a major selling point. The company is also widely believed to be working on its own LLM (Large Language Model) adding major AI features to iOS 18, which will likely run best on the upcoming iPhone 16 series. The growing importance of AI is a likely motivator for Apple to invest resources in this area.

While it is unlikely that the company will confirm any such development effort, it has in the past disclosed some details of upcoming processors at its annual WWDC conference, to allow software developers to begin getting their apps and services ready for future hardware launches. The 2024 edition is set to begin on June 10, and based on expectations of iOS 18, some details of Apple’s AI and datacentre ambitions might emerge.