Apple has just launched its most expensive iPad Pro with the new M4 chip. Now Bloomberg’s Mark Gurman reports that Apple is deploying its M series chips in its cloud servers, and it seems to be related to the AI features in the upcoming iPhone 16 lineup.

iOS 18

Some AI features may require more compute than what on-device processing can offer

Apple is working on some AI features for the iPhone 16 series. Earlier it was reported that the company is focusing on running the AI features on-device for privacy of user data. The company also acquired French startup Datakalab to make its on-device LLM more capable and efficient. However, it seems that Apple still needs to run some more demanding AI features on the cloud. Here comes the deployment of Apple Silicon (M series chips) in its servers.

According to Mark Gurman, the servers running on Apple Silicon are specially designed to handle AI-related tasks. Apple is using its M2 Ultra chip (that powers last year’s Mac Pro and Mac Studio) in its AI servers. Although, the company reportedly has plans to use the new M4 chips as well going forward.

Dedicated chips for the servers are also under development

There are also reports that the company is developing dedicated chips for its AI servers which will use TSMC’s 3nm process node. However, that’s far from completion, unlike the iPhone 16 series. It could be a reason behind the deployment of the M2 Ultra chips. The dedicated AI server chip is expected to enter mass production in the second half of 2025.

Choosing the cloud servers for running some of the features (that might also involve working with user data) is a shift from Apple’s initial plan. Nonetheless, it’s good to see that the company is arranging the computing and running the features on their own servers instead of asking companies like Google to handle the job with their models on their servers.

People involved with the development of the dedicated server chips informed that the chips have components inside to protect user privacy. The company reportedly uses a method called “Secure Enclave” to isolate data from security vulnerabilities.

Related: