Apple is starting to use M5 chips in Apple Private Cloud Compute servers this week, according to a new version of the Private Cloud Compute software. This is the infrastructure that powers Apple Intelligence’s cloud-based capabilities.
There are references to something called the “Private Cloud Compute Agent Worker,” which runs an iOS version with a new agent architecture for serving AI requests. iOS 26.4 also includes code to interface with this new Private Cloud Compute architecture.
The architecture runs on new hardware with J226C model number powered by the M5 chip.
Details remain sparse here, but it’s evidence that Apple is continuing to refine its Private Cloud Compute infrastructure ahead of more advanced Siri and Apple Intelligence features. The changes also come on the heels of Apple’s deal with Google to use Gemini models to power Siri features.
Apple’s Private Cloud Compute servers have used M2 Ultra chips in the past. The M2 Ultra was first introduced in June 2023, followed by the M3 Ultra last year. However, Apple did not update its PCC architecture with the M3 Ultra chips.
There have been reports that Apple plans to switch some of its Private Cloud Compute servers to M4 chips. It doesn’t look like that’s going to happen in the end, at least in terms of any kind of widespread adoption.
In addition to using M5 chips in its Private Cloud Compute infrastructure, Apple is also reportedly developing dedicated AI server chips. Ming-Chi Kuo reported that Apple plans to start mass production of these chips in the second half of 2026 before officially deploying them in 2027.
In October, Apple confirmed that it had begun manufacturing Private Cloud Compute servers at a factory in Houston, Texas, as part of its $600 billion domestic infrastructure investment package.
My favorite Mac accessory:
Watch Chance: Threads, Bluesky, Instagram and Mastodon.
FTC: We use automatic income earning affiliate links. More.
