SoftBank Explores Custom AI Servers With Nvidia, Foxconn
TOKYO — Japan’s SoftBank Group is exploring the development of proprietary AI servers in partnership with U.S. chipmaker Nvidia and Taiwanese electronics manufacturer Foxconn, according to a report from Nikkei cited by Reuters on Wednesday.
The initiative would see SoftBank move beyond purchasing off-the-shelf AI computing hardware to designing custom server systems tailored to its growing portfolio of artificial intelligence investments, according to the report.
The partnership leverages Nvidia’s dominance in AI accelerator chips and Foxconn’s large-scale manufacturing capabilities, potentially giving SoftBank greater flexibility in securing dedicated AI infrastructure at a time when demand for GPU-powered computing capacity far outstrips supply.
The effort aligns with SoftBank’s push into AI infrastructure. The company has committed more than $100 billion to AI-related investments in the United States, including data center construction and compute capacity buildouts. SoftBank CEO Masayoshi Son has positioned the conglomerate as one of the world’s largest investors in artificial intelligence, pivoting from the company’s earlier focus on mobile telecommunications and venture capital.
For Nvidia, the collaboration represents another major enterprise partnership as the Santa Clara, California-based chipmaker continues to expand beyond selling standalone graphics processing units into deeper integration with large-scale infrastructure customers. Nvidia’s AI chips power the vast majority of the world’s AI training workloads.
Foxconn, formally known as Hon Hai Precision Industry, has been expanding its AI server manufacturing business, positioning itself as a leading contract manufacturer for companies building large-scale AI computing systems.
The move reflects a broader trend among major technology investors and hyperscale operators to develop custom AI infrastructure rather than relying solely on commodity hardware. Companies including Meta, Microsoft and Amazon have all pursued custom chip and server designs to optimize performance and reduce long-term costs for AI workloads.
Details on the timeline, scale and financial terms of the potential SoftBank server project were not immediately available, according to the Nikkei report.
