Two months ago Nvidia CEO Jensen Huang and OpenAI CEO Sam Altman stood side by side in San Jose, California, to announce a historic agreement between the two leaders in artificial intelligence.
Nvidia would invest $100 billion over a number of years, starting in 2026, as OpenAI’s AI supercomputing facilities come online, the duo said. The timing of the buildouts and the cost of each data center weren’t disclosed.
But in Nvidia’s quarterly financial report on Wednesday, the chipmaker reminded investors that there’s a big difference between an announcement and a contract.
“There is no assurance that we will enter into definitive agreements with respect to the OpenAI opportunity or other potential investments, or that any investment will be completed on expected terms,” Nvidia said in the risk factors section of its quarterly filing.
Nvidia has been on an investing binge of late, putting its ever-expanding cash hoard to use, and financially supporting companies that buy its graphics processing units, or GPUs. In addition to the OpenAI arrangement, Nvidia on Wednesday highlighted its $5 billion commitment to invest in Intel during the quarter and its agreement this week to invest up to $10 billion in Anthropic.
An OpenAI spokesperson didn’t provide a comment but pointed to Huang’s statements on the call, including his description of OpenAI as a “once-in-a-generation company” and his expectation that the investment will “translate to extraordinary returns.”
“There is no assurance that any investment will be completed on expected terms, if at all,” Nvidia said.
The key difference with OpenAI is the scale of the planned investment and the benchmarks that would need to be met for all the money to come through. A source told CNBC at the time of the announcement that an initial $10 billion would be available to OpenAI soon to help the company work towards deploying its first gigawatt of capacity.
Altman said recently that OpenAI will end the year on a $20 billion annualized revenue run rate, which is a massive number considering its flagship ChatGPT product is only three years old. Altman said the company expects to reach hundreds of billions of dollars in revenue by 2030. But that figure doesn’t come anywhere close to covering the company’s expenses.
In total, OpenAI has announced roughly $1.4 trillion in infrastructure spending with a number of partners as it seeks to continue building out its AI models and services. To get there, the company is reliant on outside capital.
Despite the uncertainty of the September agreement, Nvidia executives continue to sound bullish on the company’s work with OpenAI. On Nvidia’s earnings call, after the chipmaker reported a solid revenue and earnings beat along with stronger-than-expected guidance, CFO Colette Kress touted OpenAI’s growth.
“OpenAI recently shared that their weekly user base has grown to 800 million, enterprise customers has increased to 1 million and that their gross margins were healthy,” Kress said. She added that the two companies are “working on a strategic partnership” and that Nvidia is “focused on helping them build and deploy at least 10 gigawatts of AI data centers.”
And Huang said, “Everything that OpenAI does runs on Nvidia today.”
An OpenAI spokesperson didn’t provide a comment, but pointed to Huang’s commentary on the call, including his description of OpenAI as a “once-in-a-generation company” and his expectation that the investment will “translate to extraordinary returns.”
There’s no question that as OpenAI builds out data centers, it will keep spending money on Nvidia’s chips. But OpenAI has also partnered with Nvidia rival Advanced Micro Devices, agreeing last month to deploy 6 gigawatts of AMD’s Instinct GPUs over multiple years and across multiple generations of hardware, beginning in the second half of next year.
The AMD agreement has one critical component that Nvidia’s lacks: signatures.
As part of the pact, the company has issued OpenAI a warrant for up to 160 million shares of the chipmaker’s common stock, with vesting milestones tied to deployment volume and AMD’s share price. That agreement was signed on Oct. 5, by AMD CFO Jean Hu and OpenAI CFO Sarah Friar.
— CNBC’s MacKenzie Sigalos and Ashley Capoot contributed to this report.
WATCH: Nvidia revenue bigger story than gross margins moving forward
