Apple partners with Foxconn to build custom AI servers in Taiwan

Apple is partnering with Foxconn and LCFC, a Lenovo subsidiary, to build its custom Apple Silicon-powered AI servers in Taiwan. This move aims to strengthen Apple's data center abilities for their new Apple Intelligence services while easing off their reliance on Chinese manufacturers.

Word on the street is that Apple picked Taiwan mainly to tap into Foxconn's deep know-how in building AI servers. Foxconn already pumps out servers with Nvidia's H100 and H200 GPUs and is gearing up to work with new Blackwell-based chips.

Unlike competitors like Amazon, Google, and Microsoft, Apple's server plans are more about AI inference management rather than training huge language models. These specialized servers are meant for internal use, so we're looking at smaller production numbers than the usual data center setups.

This partnership isn't just about making the servers; it includes engineering and design help from Foxconn and LCFC. Even though Apple has limited experience designing data center servers, the development process should move quickly since these servers have a simpler setup than Nvidia's GB200 systems.

Foxconn runs AI research labs in Hsinchu, Taiwan, and San Jose, California, where they're already working with Nvidia on future GB300 server projects. On top of that, other manufacturing partners like Universal Scientific Industrial could jump in to add even more variety to the production line.