Apple Discusses AI Server Production in Taiwan with Foxconn

Apple is reportedly in discussions with Foxconn, its long-standing iPhone manufacturing partner, about producing AI servers in Taiwan. This move aims to expand Apple's presence in the growing generative AI sector by accelerating its computing capabilities. Although Foxconn is Apple's main iPhone manufacturer, it also plays a key role in producing AI servers for Nvidia. Sources indicate that Foxconn's capacity to take on Apple's server orders may be constrained due to current commitments.

Apple's AI servers are expected to utilize its custom-designed Apple Silicon chips, which support generative AI capabilities for iPhone, MacBook, and other Apple products. Unlike Nvidia's GB200 systems, Apple's server demand is expected to be smaller, as it primarily focuses on in-house needs.

Apple's experience in designing data center servers is less extensive than Nvidia's, which could mean a need for additional engineering and design support from manufacturing partners. With Foxconn occupied with Nvidia server orders, Apple has also approached Lenovo Group and its subsidiary, LCFC, to assist with some server design and potentially to expand production capabilities in Southeast Asia. Smaller vendors are also being considered to meet specific production needs.

The production of AI servers presents complex challenges, as it requires major modifications to core components and production line systems. Advanced components like cooling technology, along with the integration of software and hardware, make the manufacturing process more intricate than for traditional servers. For example, cooling systems alone on an AI server production line can cost tens of millions of dollars.

Negotiations between Apple and Foxconn are still ongoing, as Foxconn's AI server capacity in Hsinchu is currently dedicated to Nvidia's GB200 systems. Limitations in space and equipment and differences in price expectations have further delayed a final agreement.

According to TrendForce Senior Analyst Frank Kung, Apple is anticipated to begin integrating M4 chips in its AI servers by 2025.