Apple upgrades Private Cloud Compute with M5

This article may contain affiliate links. Learn more on our Partner Program page.

Make Apfelpatient a preferred source One click – and you'll see us more often on Google

Apple continues to expand its AI infrastructure. A newly released version of its Private Cloud Compute software reveals that future servers will use M5 chips to support Apple Intelligence's cloud-based features. This signals a significant technological leap in the server architecture behind Siri and other AI functions.

Apple pursues a two-pronged strategy for artificial intelligence. Part of the processing runs directly on the device, while another part is handled via the so-called Private Cloud Compute Infrastructure. This server environment is specifically designed to handle computationally intensive AI requests that cannot be processed efficiently locally.

The new M5 integration makes it clear that Apple is further developing its existing infrastructure to make future Apple Intelligence features more powerful. At the same time, strategic partnerships and in-house chip development continue in the background.

Evidence of M5 chips in private cloud computing

A new version of the Private Cloud Compute software contains specific references to new server hardware. A model number, J226C, is mentioned. This new hardware is said to be powered by the M5 chip.

Additionally, the software contains an entry called "Private Cloud Compute Agent Worker". This is a component that runs a version of iOS. This particular iOS variant is based on a new agent-based architecture used to process AI requests.

iOS 26.4 also contains the necessary code to prepare for integration with this new private cloud compute architecture. This demonstrates that the integration is not only planned on the server side, but is also being embedded within the operating system itself.

Technical background: From M2 Ultra to M5

Until now, Apple's private cloud compute servers used the M2 Ultra chip, which was first introduced in June 2023. The following year saw the release of the M3 Ultra, but Apple did not transition its private cloud compute architecture to this chip generation.

There were reports that Apple was planning to migrate parts of its Private Cloud compute servers to M4 chips. However, as things stand, this migration does not appear to have been implemented on a large scale.

The now visible focus on the M5 suggests that Apple waited for a major technological leap rather than gradually adopting each generation. While details on the exact performance of the M5-based servers are not yet available, this move underscores the long-term planning behind Apple's intelligence infrastructure.

New agent-based architecture

The mention of the "Private Cloud Compute Agent Worker" is another key point. The agent-based architecture suggests that AI requests could be processed modularly in the future. Individual tasks or subprocesses can then be handled and coordinated separately.

Combined with the new server hardware, this could form the basis for more complex Siri functions and further developed Apple intelligence features. Specific details on how it will work have not yet been released, but its integration into iOS 26.4 shows that the transition is being prepared system-wide.

Collaboration with Google and use of Gemini

In parallel with these infrastructure measures, Apple has reached an agreement with Google regarding the use of Gemini models for certain Siri functions. The integration of external AI models complements Apple's own server architecture.

The combination of its own private cloud compute infrastructure and selected external models demonstrates Apple's flexible approach to its AI strategy. While hardware and server control remains internal, additional models can be integrated as needed.

Proprietary AI server chips from 2026

In addition to using the M5 processor, Apple is reportedly also working on specially developed AI server chips. Analyst Ming-Chi Kuo reports that mass production of these chips is expected to begin in the second half of 2026, with official deployment planned for 2027.

This development would represent a further step towards complete control over the AI infrastructure. Specifically designed server chips could be tailored even more precisely to the requirements of Apple Intelligence than universally applicable M-series chips.

Server production in Houston, Texas

In October, Apple confirmed that production of Private Cloud compute servers had begun at a factory in Houston, Texas. This move is part of a $600 billion investment package to strengthen domestic infrastructure.

Manufacturing in the USA underscores the strategic importance of the Private Cloud Compute Servers. They form the backbone of Apple's cloud-based intelligence capabilities and are a key component of the future AI strategy.

Apple strengthens Apple Intelligence through new server architecture

The indications of M5 chips in the Private Cloud Compute architecture clearly show that Apple is further expanding its AI infrastructure. With new server hardware, an agent-based software architecture, the integration of external Gemini models, and the planning of its own AI server chips, a comprehensive technological foundation is being created for future Apple intelligence features.

The production of the servers in Houston, as part of a multi-billion dollar investment program, also underscores the long-term focus. Overall, this paints a picture of a consistently evolved AI architecture, prepared for future requirements in both hardware and software. (Image: Apple)