Nvidia's $100 Billion Investment in OpenAI Remains Unfinalized

In September 2025, Nvidia and OpenAI announced a letter of intent for Nvidia to invest up to $100 billion in OpenAI's AI infrastructure. However, five months later, no definitive agreement has been reached. Nvidia CEO Jensen Huang recently stated that the $100 billion figure was "never a commitment" and emphasized that any investment would be considered "one step at a time."

Reports indicate that OpenAI has been exploring alternatives to Nvidia's chips since last year, citing concerns over the speed of some Nvidia chips for inference tasks. In response to these developments, OpenAI CEO Sam Altman expressed continued support for Nvidia, stating, "We love working with NVIDIA and they make the best AI chips in the world. We hope to be a gigantic customer for a very long time."

The initial announcement detailed an ambitious plan involving 10 gigawatts of Nvidia systems for OpenAI, equating to the energy output of approximately 10 nuclear reactors. Huang described the project as "a giant project" at the time. However, the deal was always a letter of intent, not a binding contract. Recent reports suggest that Nvidia insiders have expressed doubts about the transaction, with concerns about OpenAI's business approach and competition from companies like Google and Anthropic. Huang has dismissed these claims as "nonsense."

Nvidia's stock experienced a slight decline following these reports. Industry analysts have noted the unusual nature of the back-and-forth between the companies, with some questioning the scale and structure of the proposed investment.

In the meantime, OpenAI has been seeking alternative chip suppliers. The company has reportedly discussed partnerships with startups Cerebras and Groq, both specializing in chips designed to reduce inference latency. However, a recent $20 billion licensing deal between Nvidia and Groq has reportedly ended OpenAI's talks with Groq.

OpenAI has also announced a $10 billion deal with Cerebras, adding 750 megawatts of computing capacity for faster inference through 2028. Additionally, OpenAI has partnered with AMD for six gigawatts of GPUs and plans to develop a custom AI chip with Broadcom to reduce dependence on Nvidia. The timeline for these chips becoming operational remains uncertain.