Apple reportedly will switch its servers that power Apple Intelligence to M4 chips next year. Currently, these servers use the M2 Ultra, the same chip found inside the Mac Pro and Mac Studio.
Apple is also reportedly in talks with Foxconn to manufacture AI servers to enhance the company’s AI processing capabilities.
Apple wants Foxconn to build its next-gen AI servers
Many complex Apple Intelligence features, like image generation, are not processed locally but in the cloud. On the backend, M2 Ultra-powered servers handle the processing, which Apple refers to as Private Cloud Compute. Apart from efficiency and cost benefits, this approach ensures complete safety and security of user data. Neither Apple nor third parties ever store or share user data.
With the M4 chips‘ superior AI performance, Apple reportedly plans to build servers around this processor for increased speed and efficiency. The company is reportedly talking to Foxconn, its largest vendor, about building new AI servers. It wants the servers in Taiwan to utilize the available engineering talent and R&D resources there.
However, as a Wednesday report from Nikkei Asia details, Foxconn is also building AI servers for Nvidia. Demand for the latter’s GPUs has soared due to the AI boom, making Nvidia servers one of the hottest products in the world. Interestingly, all major tech companies use Nvidia GPU-powered servers for their AI features. Apple took a different approach, using its in-house chips instead.
M4 servers should help with future Apple Intelligence expansion
The switch to M4-powered servers should allow Apple to rapidly expand Apple Intelligence to more countries and languages.
Apple’s M4 SoC features the company’s fastest Neural Engine yet, capable of processing up to 38 trillion operations per second. This should allow M4-powered AI servers to deliver significantly better performance and efficiency than current M2 Ultra servers.
Apple likely will use the M4 Max or M4 Ultra SoC for its AI servers to get the best performance possible.
