Apple considers building AI servers
What's the story
Apple is said to be mulling an entry into the artificial intelligence (AI) server market, The Information reported. The tech giant is said to be considering an AI server built around its upcoming "M8" series chips and integrating NVIDIA networking hardware. This would mark Apple's return to a business it had exited in 2011 with the discontinuation of Xserve.
Market focus
Targeting businesses that want to run models on own hardware
Apple is said to be targeting businesses that want to run AI models on their own hardware, focusing on inference and generating responses from trained models.
One of the solutions being explored is NVIDIA's NVLink Fusion, a platform that provides hardware and software for data exchange between chips.
However, neither the NVIDIA arrangement nor the server has been finalized yet.
Leadership support
Project backed by Apple CEO John Ternus
The project was reportedly backed by Apple CEO John Ternus when it started about a year ago, during his time as the company's hardware engineering chief.
The initiative comes amid rising interest in Apple hardware from AI companies, which have been buying Mac mini and Mac Studio systems in large numbers.
Current infrastructure
Apple currently makes server hardware for private cloud compute
Apple already makes server hardware for Private Cloud Compute, the infrastructure that supports Apple Intelligence tasks requiring more processing power than what a device can handle.
However, scaling Apple's existing chip interconnects is said to be cost- and speed-prohibitive, which is why it is looking at NVIDIA's technology.
Notably, Apple has also turned down partner requests to use its Private Cloud Compute servers.