Nvidia and OpenAI play down reports of rift. Why they need each other.
Nvidia and OpenAI are rushing to reassure the market that their relationship is still solid. The two companies balance a complex relationship in which their successes are intertwined, but they don’t want to become overly reliant on each other.
Questions arose about Nvidia and OpenAI’s partnership after The Wall Street Journal reported that the chipmaker had delayed a planned investment of up to $100 billion in ChatGPT developer OpenAI. Nvidia CEO Jensen Huang later said the company planned to make a “major investment” in OpenAI, but did not say how much.
The latest development is a Reuters report stating that OpenAI is unhappy with some of Nvidia’s latest AI chips and is trying to diversify its hardware resources by seeking alternatives for about 10% of its inference needs. Inference is the process of generating answers or conclusions from artificial intelligence models.
OpenAI did not immediately respond to a request for comment early Tuesday. But OpenAI CEO Sam Altman took to social media to smooth things over.
“We love working with NVIDIA and they make the best AI chips in the world,” Altman said in a post on social media platform
So what’s going on? The possible explanation is negotiation between the two companies rather than a serious breach. OpenAI has struck additional chip supply deals with Nvidia’s rivals in recent months, including Advanced Micro Devices, Broadcom and Cerebras Systems. This makes sense for a company that doesn’t want to be overly dependent on a single supplier.
But Nvidia has many advantages of its own. Both parties have a memorandum of understanding to build at least 10 gigawatts of computing power, but this is planned to take years. OpenAI is reportedly planning a major IPO as soon as this year, while Nvidia may take the chance to signal to its partner that it also has options on where to direct its spending, after recently announcing it would invest up to $10 billion in AI start-up Anthropic.
If so, OpenAI seems to have gotten the message. OpenAI infrastructure manager Sachin Katti said in a post on X late Monday that the company “bases our training and inference on Nvidia.”
News Corp, which owns Barron’s and the WSJ, has a content licensing partnership with OpenAI.
Write to Adam Clark at adam.clark@barrons.com
