A few weeks ago, Nikkei Asia reported that Apple is preparing to use its latest M4 chips in its AI servers. For the time being, it’s unclear if the Cupertino company will use the base model or even some of the recently announced chips. For example, M4 Max benchmark tests reveal that this processor outperforms the M2 Ultra. Additionally, depending on when Foxconn starts making those servers, an M4 Ultra chip could be used instead.
With Apple focusing on expanding Apple Intelligence features in the years to come, including in new countries and more languages, the company will need to expand its AI servers significantly. We also expect these features to become more complex with time, especially now that all new Macs offer at least 16GB of RAM.
A new report by The Information reveals that Apple could soon ditch its strategy of using its own M-chips and use Broadcom chips instead. With the codename Baltra, people familiar with the matter said Broadcom is already developing a new AI server chip for Apple.
While it’s interesting that Apple moved from Intel to producing its own chips, this change could make sense, as the server chips might require specs that don’t make sense for consumers. In addition, as Apple expands Apple Intelligence features, it’s important to note that the company will continue to use Private Cloud Compute to process users’ requests in the cloud.
PCC works similarly to how Apple’s processor and Secure Enclave function. With this technology, the company can expand the processor power of its devices by sending users’ requests to an encrypted cloud that no one can access.
That said, we could see more powerful servers in 2026 that can take on more challenging tasks with a larger volume of requests. It’s important to note that in a couple of years, at least four different iPhone lineups, in addition to several Macs and iPads, will take advantage of Apple Intelligence.
BGR will keep reporting on Apple’s new AI servers as we learn more about them.
