It is reported that Apple has contacted Foxconn to discuss manufacturing artificial intelligence (AI) servers in Taiwan, China, China, because this iPhone manufacturer hopes to accelerate its computing power and occupy a larger share in the generative AI boom. Foxconn is Apple's largest supplier of iPhones, but also a major manufacturer of Nvidia AI servers. Sources suggest that Foxconn's ability to take over Apple as a server customer may be limited.
Apple is keen on using internally designed Apple Silicon chips to build servers that support Apple Intelligence's generative AI capabilities on iPhones, MacBook, and other devices. Sources say that due to Apple's desire to manufacture servers for its own use, the demand for it is relatively small compared to the Nvidia GB200 system.
Apple's experience in designing data center servers is not as good as Nvidia's, so suppliers may need to provide more support to provide engineering and design services for servers.
Insiders revealed that due to Foxconn's busy completion of Nvidia server orders, Apple has also requested assistance from Lenovo Group and its subsidiary Lianbao in some server designs, and is counting on some small suppliers to provide production services. The negotiations between Apple and Lenovo involve establishing new server production capacity in Southeast Asia.
Sources say that manufacturing AI servers is much more complex than expected, as the process requires a thorough overhaul of critical components and redesign of production line systems. The integration of new components (such as heat dissipation technology) with software and hardware is also more complex than traditional servers. The cooling system of the server production line alone may cost tens of millions of dollars.
Insiders revealed that negotiations between Foxconn and Apple regarding AI server manufacturing have not been finalized yet, as Foxconn's current AI server production capacity in Hsinchu has been occupied by Nvidia's GB200 system. Apart from limitations in space and equipment that prevent the completion of additional orders, the two parties have not yet reached an agreement on the price.
TrendForce senior analyst Frank Kung stated that it is expected that Apple will begin using M4 chips in its AI servers by 2025.
