[News] Foxconn Teams Up With OpenAI on AI Hardware; Output Hits 1,000 Server Racks a Week

Following recent deals with tech giants like AMD, OpenAI has announced a new partnership with Foxconn, marking its latest move to strengthen AI hardware development. According to Bloomberg, OpenAI is working with Foxconn to design and manufacture hardware for data centers. Foxconn also plans to produce cabling, power systems, and other key equipment for U.S. data-center infrastructure, the report adds. Although the initial agreement includes no purchase commitments or financial obligations, Foxconn said OpenAI will receive early access to test these systems and retain the option to purchase them, according to Focus Taiwan.

As Nikkei notes, the announcement marks a further deepening of Foxconn’s relationship with OpenAI, shifting the company from an indirect supplier to a direct collaborator. This mirrors how the Taiwanese manufacturing giant now works directly with other major U.S. cloud providers, including Google, AWS, and Microsoft. Nikkei, citing sources, also indicates that the partnership will give Foxconn clearer insight into feedback from end users of AI server racks — an important advantage for accelerating rack-architecture development and bringing new systems online more quickly.

Foxconn’s Capacity Reaches 1,000 AI Server Racks per Week

Foxconn hosts its Hon Hai Technology Day in Taipei today. According to Liberty Times, Chairman Young Liu says the group can now build nearly 1,000 AI server racks per week, and output continues to grow. He also notes that to meet the computing density required for servers, GPU substrates need 60-layer PCBs capable of supporting billions of calculations.

Liu, as cited by Liberty Times, further states that high-density computing requires advanced cooling, and Foxconn provides complete solutions for both liquid and air cooling, including power shelves, bus bars, and high-speed connectors. Notably, about 40% of the components in each rack are produced in-house by Foxconn.

OpenAI CEO Sam Altman also delivered remarks at Foxconn’s Technology Day via video, saying that demand for key AI-infrastructure components already far exceeds supply and will continue rising in the years ahead. According to Liberty Times, Altman says that new AI models require new types of hardware — from server racks to cooling systems and power infrastructure. Scaling up production of this next-generation infrastructure is essential, and Foxconn’s deep experience in building large, complex hardware positions it well for this role.

Foxconn continues to expand from electronics into AI data centers and is now NVIDIA’s main producer of AI racks. This shift makes it a major beneficiary of the data-center boom, as cloud companies invest billions to build out their AI infrastructure. The company maintains an upbeat outlook, saying AI demand will be a key driver of its 2026 growth, as noted by Reuters. The report adds that Foxconn says it will invest $2 billion to $3 billion annually in AI.

Read more

• [News] OpenAI CEO Reportedly Meets Foxconn, TSMC, Heads to Korea to Explore Samsung, SK hynix Ties
• [News] OpenAI’s Multi-Billion AMD Deal Unpacked: Big Gains for Both, Along with TSMC and Taiwan’s ODMs

(Photo credit: Foxconn)