According to a report from TechNews, citing Nikkei, Apple is reportedly in talks with Foxconn to manufacture AI servers in Taiwan as part of its strategy to capture a larger share of the booming generative AI market.
However, according to Nikkei, while Foxconn is Apple’s largest iPhone manufacturing partner, it also produces AI servers for NVIDIA, which could limit its capacity to take on Apple’s AI server orders.
The Nikkei report highlights that Apple prefers to use its own chips for building servers that will power generative AI features in MacBooks and other devices. Citing industry sources, the report stated that Apple is exploring the production of its own AI servers and has been in talks with Foxconn. However, the volume of servers Apple is considering producing is relatively low compared to the demand for NVIDIA’s GB200 servers.
The report suggests that Apple wants Foxconn to produce AI servers in Taiwan to take advantage of Taiwan’s engineering talent and R&D resources. Apple, with less experience in designing data center servers compared to NVIDIA, may require additional support for the production of its AI servers.
The report notes that as mastering AI computing and expanding data center infrastructure become critical to next-generation computing, Apple is enhancing its server capabilities in response to heavy investments in AI data centers by rivals such as Microsoft, Amazon, and Google.
According to the Nikkei report citing sources, with Foxconn occupied with fulfilling orders for NVIDIA servers, Apple has turned to China’s Lenovo and its subsidiary LCFC for additional support, along with smaller suppliers like Universal Scientific Industrial to assist with production. Additionally, Apple’s discussions with Lenovo also include plans to build server production capability in Southeast Asia, outside of China.
The report highlights that building AI servers is more complex than anticipated, and currently only a few manufacturers worldwide are capable of producing NVIDIA’s AI servers, including Foxconn, Quanta, Wistron, and Supermicro. However, the production capacities of these manufacturers are largely contracted by NVIDIA.
On the other hand, according to TechNews, institutional investors pointed out that Foxconn has production capacity in multiple regions worldwide, and given the ongoing tensions between the U.S. and China, Apple’s AI servers may not necessarily need to be manufactured in Taiwan.
The report from TechNews further noted, citing institutional investors, that the current discussion between Apple and Foxconn may still be in the early stages. However, due to their long-standing partnership, even if Apple’s order volume is relatively small, Foxconn is likely to accept the production of Apple’s AI servers.
Read more
(Photo credit: Foxconn)