In September 2025, Nvidia and OpenAI introduced a letter of intent for Nvidia to take a position as much as $100 billion in OpenAI’s AI infrastructure. On the time, the businesses stated they anticipated to finalize particulars “within the coming weeks.” 5 months later, no deal has closed, Nvidia’s CEO now says the $100 billion determine was “by no means a dedication,” and Reuters experiences that OpenAI has been quietly in search of options to Nvidia chips since final 12 months.
Reuters additionally wrote that OpenAI is unhappy with the pace of some Nvidia chips for inference duties, citing eight sources conversant in the matter. Inference is the method by which a skilled AI mannequin generates responses to person queries. In line with the report, the problem turned obvious in OpenAI’s Codex, an AI code-generation device. OpenAI employees reportedly attributed a few of Codex’s efficiency limitations to Nvidia’s GPU-based {hardware}.
After the Reuters story printed and Nvidia’s inventory worth took a dive, Nvidia and OpenAI have tried to easy issues over publicly. OpenAI CEO Sam Altman posted on X: “We love working with NVIDIA they usually make the very best AI chips on this planet. We hope to be a big buyer for a really very long time. I do not get the place all this madness is coming from.”
