OpenAI's chip project faces obstacles in securing $18 billion in financing, as lenders become cautious about returns on AI infrastructure

According to The Information, OpenAI was unable to complete approximately $18 billion in financing from external lenders, which was originally planned to support the early deployment of its custom chip project developed in collaboration with Broadcom. This project plans to deploy OpenAI's self-developed chips with a computing power of 10 gigawatts, which is a core strategy for the company to reduce its reliance on Nvidia. The original agreement announced in October 2025 had a total hardware construction scale of approximately $500 billion.

The background of the financing obstacles is that lenders are becoming increasingly cautious about AI infrastructure deals backed by OpenAI's expected revenue, compounded by recent reports that OpenAI has not met its internal growth targets, further heightening investor concerns. More broadly, capital expenditures for AI in large-scale enterprises are estimated to reach $600 billion to $720 billion by 2026, but Nvidia's accounts receivable have approached $33 billion, indicating that buyer settlement cycles are generally extended. OpenAI currently faces three paths: restructuring this round of financing, changing lenders, or scaling back the chip deployment.