Apple plans to build servers with its own chips to power up AI capabilities

Apple is reportedly working with Foxconn to build AI servers in Taiwan

In Short

• Apple plans to use its own chips for AI servers
• Foxconn may build AI servers for Apple in Taiwan
• Apple's AI server order is small compared to Nvidia's

Apple is reportedly moving toward using its own chips in servers designed to support generative AI features for MacBooks and other devices, and it is looking at Foxconn to help it do that. Apple has reportedly engaged Foxconn to explore building AI servers in Taiwan. Not only is Foxconn Apple’s primary iPhone manufacturer, it also leads in producing Nvidia’s AI servers.

According to a recent report from Nikkei, the company has been exploring AI server production, however, Apple’s anticipated order volume remains modest compared to the surging demand for Nvidia’s high-performance GB200 AI servers.

Apple’s decision to consider Taiwan as a production base is strategic; the company hopes to leverage the region's engineering talent and research and development resources to support this venture. While Apple is renowned for its expertise in consumer electronics, it lacks Nvidia’s depth in designing servers for AI-driven data centres, which may require Foxconn’s additional support and expertise.

By collaborating with Foxconn, Apple aims to advance its AI server capabilities amid rising competition in AI infrastructure from its biggest competitors including Microsoft, Amazon, and Google, all of whom are heavily investing in AI data centres.

However, Foxconn’s production capacity may be stretched thin. With its commitment to producing Nvidia’s AI servers, Apple has reportedly sought additional support from China’s Lenovo and its subsidiary LCFC, as well as smaller suppliers like Universal Scientific Industrial. The tech giant has also considered expanding server production capabilities into Southeast Asia, exploring options outside of China to diversify its supply chain.

Building AI servers is proving to be a more complex endeavour than initially anticipated. Only a handful of manufacturers, including Foxconn, Quanta, Wistron, and Supermicro, currently have the expertise to produce Nvidia’s advanced AI servers. With these manufacturers largely contracted by Nvidia, Apple’s entry into AI server production could face capacity constraints in the supply chain.

While the alleged discussions between Apple and Foxconn are still in the early stages, given Apple’s long-standing partnership with Foxconn, it’s likely that Foxconn would accept Apple’s AI server orders, even if they are relatively small. The big picture?: This marks a big move for Apple, as enhancing AI server capabilities becomes critical for the future and offers a competitive positioning to the company in the AI-driven markets.