Nvidia and OpenAI are stuck in a huge deal. AI giants need each other

Nvidia CEO Jensen Huang and OpenAI CEO Sam Altman appeared together on CNBC in September to announce a massive $100 billion deal poised to usher in a new chapter in the booming artificial intelligence industry.

Five months have passed and no contract has been signed and no funds have been exchanged. Even more concerning for investors is that the two companies appear to be at odds.

The Wall Street Journal reported Friday that negotiations between the two companies are at an “icy stage” after some within NVIDIA expressed doubts about OpenAI’s business model. This has been a hot topic in the AI ​​space since November, when NVIDIA warned in its quarterly report Risk Factors that there was “no guarantee that we will enter into a definitive agreement regarding the OpenAI opportunity or other potential investments.”

Despite the reported friction, Nvidia and OpenAI still need each other.

Altman said OpenAI needs a huge number of Nvidia’s AI chips to meet its revenue growth goals, while Huang relies on customers like OpenAI to develop services that wow customers and continue to grow sales of expensive systems.

A surge in demand and industry hype pushed NVIDIA’s market cap to more than $5 trillion at its peak in October, but the stock has fallen 15% from its peak, pushing its valuation to $4.4 trillion. Meanwhile, OpenAI was valued at $500 billion in the private market late last year, and is reportedly aiming for a valuation of more than $800 billion as it seeks to raise further funding.

“I’m looking forward to seeing Sam finish, but he’s doing great,” Huang told CNBC’s Jim Cramer on Tuesday. “And we’re going to invest in the next round. There’s no question about that.”

Nvidia first invested in OpenAI in October 2024 as part of a $6.6 billion funding round.

Huang added on Tuesday that there is “no drama” in the relationship with OpenAI, a sentiment expressed by Altman in a post about X the day before.

“We hope to be a huge customer for a very long time,” Altman wrote. “I don’t understand where this madness is coming from.”

Still, little progress has been made on the landmark September deal that reportedly included OpenAI building infrastructure that would require 10 gigawatts of power.

Nvidia’s initial investment of $10 billion will be rolled out when the first gigawatts are completed, CNBC reported at the time of the deal. The companies said the first phase of the latest investment is expected to begin in the second half of 2026.

OpenAI’s current funding round includes participation from Nvidia, Huang said, but it was not included in last year’s deal. Huang told Cramer that Nvidia would like to evaluate additional investment in OpenAI and participate in AI Labs’ IPO.

Nvidia stock fell about 3% on Tuesday, its third straight day of declines, led by tech stocks.

Nvidia and OpenAI have been working together for 10 years.

When OpenAI was a little-known nonprofit lab in 2016, it was the first organization to want to use Nvidia’s debut AI system, called DGX, Huang told Joe Rogan in a December interview.

Over the next few years, OpenAI became a heavy user of Nvidia chips, typically delivered through them. microsoft infrastructure. In February 2023, a few months after the release of ChatGPT, Huang was in high spirits on Nvidia’s earnings call, praising OpenAI and boasting that generative AI was transforming the company.

“Everyone who develops software is either alarmed, shocked and alarmed, or actively working on things like ChatGPT to integrate into their applications,” said Huang, whose company’s stock has soared.

Nvidia’s parabolic growth has matched OpenAI’s explosive growth.

In the quarter in which ChatGPT was released, Nvidia made $6 billion in revenue. In the period that ended in October of this year, that number rose almost tenfold to $57 billion. Analysts say the company controls more than 90% of the graphics processing unit (GPU) market.

ChatGPT is the leading chatbot in terms of usage, reaching 800 million weekly users late last year. The company announced in January that it expects annual sales to hit $20 billion, but analysts don’t expect it to become profitable until 2030.

Although both companies deny their existence, the tension is at the heart of how both companies have sought to diversify by partnering with each other’s rivals.

As its balance sheet swells and it needs more customers, Nvidia has leveraged cash to invest in a number of key partners, including a $10 billion investment in Anthropic in November. Investors expect Nvidia to partner with more large buyers, as its customers are concentrated in a small number of hyperscalers.

At the same time, OpenAI made several announcements with other semiconductor companies, saying they needed more computing power than Nvidia alone could provide.

In June, Altman appeared as follows. advanced micro device CEO Lisa Su attended the company’s annual event in San Jose, California. Altman said OpenAI will help develop AMD’s next-generation AI chips and will be a customer. AMD is the only company other than Nvidia that makes large-scale data center GPUs for AI.

Four months later, OpenAI announced partnerships with: broadcomhelps you create custom AI chips, including: Google Tensor processing unit. And last month, OpenAI announced it would use chips from startup Cerebras in a deal worth more than $10 billion.

Amid reports of new challenges in OpenAI and Nvidia’s relationship, OpenAI infrastructure executive Sachin Katti told X on Monday that the company’s partnership with the semiconductor giant is “fundamental.”

“Our entire compute fleet runs on Nvidia GPUs,” Katti wrote. “The demand curve is unmistakable. The world needs orders of magnitude more computing.”

Spotlight: Nvidia and OpenAI respond to reports about relationship