OpenAI (OPENAI) is not satisfied with some of Nvidia’s (NVDA) latest AI chips, and it has sought alternatives since last year, Reuters reported, citing people with knowledge of the matter.
OpenAI’s shift in strategy is over a growing emphasis on chips used to perform specific elements of AI inference. Nvidia remains dominant in chips for training large AI models, while inference is a new front in the competition, the report added.
AI inference is the process where a trained AI model uses its learned knowledge to analyze new, unseen data and make predictions, decisions, or generate outputs.
“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,” said OpenAI’s CEO Sam Altman in a post on X. “I don’t get where all this insanity is coming from.”
Nvidia and OpenAI did not immediately respond to a request for comment from Seeking Alpha.
The move by OpenAI and others to search for alternatives in the inference chip market marks a test of Nvidia’s AI dominance and comes at a time when the two companies are in investment discussions, the report noted.
In September 2025, Nvidia noted that it intends to invest up to $100B in Microsoft (MSFT)-backed OpenAI (OPENAI) progressively to build and deploy at least 10 gigawatts of AI data centers with Nvidia systems. However, on Friday, it was reported that the investment plan stalled after some insiders within Nvidia expressed doubts about the agreement. Nvidia’s CEO Jensen Huang had privately told business associates that the original agreement was not binding and finalized.
Reportedly, over the weekend, Huang said the company’s proposed $100B investment in OpenAI was “never a commitment” and that the AI chipmaker would consider any funding rounds “one at a time.” Huang added that Nvidia’s contribution to OpenAI’s current funding round would not approach $100B. But he noted that “We will invest a great deal of money, probably the largest investment we’ve ever made.”
When asked about the report that seemed to suggest he wasn’t very happy with OpenAI, Huang said on Saturday, “That’s nonsense.”
Meanwhile, OpenAI has inked deals with Advanced Micro Devices (AMD) and others for GPUs built to rival Nvidia’s. But its shifting product road map also has changed the kind of computational resources it needs and bogged down talks with Nvidia, the Reuters report noted.
“Customers continue to choose NVIDIA for inference because we deliver the best performance and total cost of ownership at scale,” said Nvidia in a statement, the report added.
A spokesperson for OpenAI in a statement told the news agency that the company relies on Nvidia to power the vast majority of its inference fleet and that Nvidia delivers the best performance per dollar for inference.
Sources, according to Reuters, said that OpenAI is not satisfied with the speed at which Nvidia’s hardware can produce answers to ChatGPT users for specific types of problems like software development and AI communicating with other software. It needs new hardware that would eventually provide about 10% of OpenAI’s inference computing needs in the future, one of the sources told the news agency.
OpenAI has discussed working with startups, including Cerebras and Groq to provide chips for faster inference. However, Nvidia struck a $20B licensing deal with Groq that shut down OpenAI’s talks, the report added.
The ChatGPT-maker’s search for GPU alternatives since last year focused on firms developing chips with large amounts of memory embedded in the same piece of silicon as the rest of the chip, called SRAM. Placing as much costly SRAM as possible onto each chip can offer speed advantages for chatbots and other AI systems as they crunch requests from millions of users, the report noted.
Inference needs more memory than training because the chip needs to spend relatively more time getting data from memory than performing mathematical operations. Nvidia and AMD GPU technology depends on external memory, which adds processing time and slows how quickly users can interact with a chatbot, the report added.
Inside OpenAI, the issue became mainly visible in Codex, its product for creating computer code, which the company has been strongly marketing, the report noted. OpenAI staff attributed some of Codex’s weakness to Nvidia’s GPU-based hardware, the report added.
In a Jan. 30 call with reporters, Altman said that customers using OpenAI’s coding models will “put a big premium on speed for coding work.”
One way OpenAI would meet that demand is via its recent deal with Cerebras, said Altman adding that speed is less of an imperative for casual ChatGPT users, the report noted.
As OpenAI showed its reservations about Nvidia technology, Nvidia approached companies working on SRAM-heavy chips, including Cerebras and Groq, about a potential acquisition, the report added. Cerebras declined and struck a commercial deal with OpenAI announced last month.
Groq held discussions with OpenAI for a deal to provide computing power and received investor interest to fund the company at a valuation of about $14B.
Groq did not immediately respond to a request for comment from Seeking Alpha.
However, by December 2025, Nvidia moved to license Groq’s tech in a non-exclusive all-cash deal, the report noted. Although the deal would allow other companies to license Groq’s technology, the company is now focusing on selling cloud-based software, as Nvidia hired Groq’s chip designers, the report added.