Foxconn has reportedly entered SpaceX’s AI server contract manufacturing supply chain, with Economic Daily News saying SpaceX has asked the company to build the first batch of AI server racks powered by NVIDIA GB300 systems. The order was reported at $52 billion, or about NT$1.7 trillion.
First batch said to cover about 13,000 racks
According to the report, the AI server racks ordered by SpaceX carry a unit price of about $4 million, with total volume at roughly 13,000 racks. That would amount to around 936,000 NVIDIA GPUs in total. Shipments are expected to run from the fourth quarter of 2026 into the first quarter of 2027.
The report added that SpaceX has recently raised its GB300 procurement target and proactively approached Foxconn Group to discuss contract manufacturing. Foxconn did not comment on specific customers or order developments. It said only that AI server order intake remains strong, visibility is better than expected, and its full-year outlook stays optimistic.
Dell and Supermicro previously led SpaceX AI server outsourcing
ABMedia said SpaceX’s AI server manufacturing orders had previously been dominated by Dell and Supermicro. In that framing, SpaceX had remained the only missing direct supply relationship among Foxconn’s major global cloud service provider customers. This reported order would give Foxconn a foothold in that part of the market.
Foxconn Chairman Young Liu had previously estimated that the company’s share of the global AI server market would exceed 40% in 2026. The article, citing institutional investors, said that estimate did not include the reported SpaceX manufacturing order.
SpaceX compute expansion also draws attention to Taiwanese ODMs
The report tied the order to SpaceX’s larger compute expansion plans. It said the company has signed multi-year compute rental agreements with Anthropic and Google, with monthly fees reaching billions of dollars. It also said SpaceX had disclosed plans centered on AI1 satellites equipped with NVIDIA GB300 racks, targeting 1 GW of AI compute deployed in space by the end of 2027 and 100 GW by 2030.
Based on that demand outlook, the article said Taiwanese ODMs could see longer order visibility over time. It named Foxconn, Quanta, and Wistron as core assemblers of AI server racks that may benefit from this wave of space-based compute deployment.
CPO switching highlighted as a higher-margin business line
Beyond conventional AI server assembly, the report also pointed to Foxconn Group’s position in CPO switching. Its subsidiary Fii was described as NVIDIA’s only contract manufacturing and design partner for all-optical CPO switches. The first batch of cabinets was said to have been delivered ahead of schedule by the end of May. Shipment targets were also raised from more than 10,000 units for all of 2026 to more than 50,000 units combined across 2026 and 2027.
ABMedia cited a previous ChainNews report saying gross margins for traditional server manufacturing usually fall between 5% and 8%, while CPO switches reach double-digit margins because of higher average selling prices. Institutional estimates in the article said CPO could contribute more than 15% of Fii’s revenue in 2026 and improve the broader profit structure of the Foxconn Group.
Article also cited recent Foxconn and Fii financial figures
On financial performance, the report said Foxconn posted net profit after tax of NT$49.919 billion in the first quarter of 2026, a record for the same period, up 10% quarter over quarter and 19% year over year. Earnings per share were NT$3.56.
Fii also recently issued profit guidance for the first half of 2026. The company estimated second-quarter net profit attributable to shareholders at RMB 12.8 billion to RMB 13.8 billion, up 86% to 101% from a year earlier. The article described that as an early signal for stronger second-half performance across the Foxconn Group.
ABMedia said that as the SpaceX order moves into the shipment cycle and CPO contributes a larger share of margin, institutional investors expect Foxconn’s second-half momentum to remain strong, with business scale and earnings potentially topping prior market projections.