Has the circular AI economy faltered? Last week it was reported that the much-discussed $100 billion deal between Nvidia and OpenAI – which was announced last September – might not be happening at all.
This was a circular arrangement through which the chip manufacturer would supply large sums of money to the ChatGPAT developer which would largely be spent on purchasing its own chips.
It is this type of deal that has worried some market watchers, who see in these transactions a glimpse of the dotcom bubble of 1999–2000.
According to the Wall Street Journal, it now appears that Nvidia was not as solid on this investment as was widely believed. Negotiations had not progressed further, with Nvidia chief executive Jensen Huang privately insisting that the deal was “non-binding” and “not finalized”. Huang confirmed this to reporters in Taipei on Saturday, telling reporters that Nvidia would invest “heavily” in OpenAI’s next funding round, but “nothing like $100 billion”.
A Reuters report soon suggested that the feeling was mutual: OpenAI was “dissatisfied” with Nvidia’s advanced AI chips, and was looking for alternatives. Nvidia’s stock has fallen 10% so far this week, amid a flurry of headlines and both companies moving into damage control.
“We love working with Nvidia and they make the best AI chips in the world,” wrote Sam Altman, CEO of OpenAI, at X. “We expect to remain a huge customer for a very long time.”
Even Oracle is shaken up: The software company that is counting on $300 billion of cloud computing deal With OpenAI, Said It still expects the startup to make good on its commitment, even if it doesn’t receive the full amount from Nvidia. In total, OpenAI has committed to more than $1tn worth of deals – counting the infrastructure to build and power its AI tools.
“The Nvidia-OpenAI deal will have no impact on our financial relationship with OpenAI,” Oracle posted on Twitter. “We are extremely confident in OpenAI’s ability to raise funding and meet its commitments.”
The $100 billion deal between two of the most important players in AI appears to have fallen apart over a weekend, which is troubling. But there are solid business reasons behind the apparent shift, said Alvin Nguyen, an analyst at research firm Forrester.
OpenAI’s ambitious development trajectory means it will be difficult for the company to stick with a single vendor, he said, especially as it plans new, computationally demanding AI models. “They need chips. They need as many chips as possible.”
As far as Nvidia is concerned, its commitment to $100 billion may have been loose in the first place, even if it was widely reported. “They won’t discourage people from promoting too much. Why say something and immediately punch your own share price?”
For a giant startup like OpenAI, maneuvering in and out of deals — for example, with chip makers — may just be business as usual, Nguyen said: “You know (Altman’s) background as a startup guy, and you know the maneuvering he’s doing makes sense from a startup perspective.”
Meanwhile, for Nvidia, the AI hype is part of selling chips. “You don’t know what’s going to happen,” Nguyen said. “And so you let other people put up the numbers for you and get publicity from that.”
The point, of course, is whether investors and other companies like Oracle would take the widely reported $100 billion commitments seriously.
In response to a question from the Guardian, an OpenAI spokesperson referred to Altman’s X post and comments made by Huang to CNBC on Tuesday, including: “There is no drama.”
The spokesperson said: “Our teams are actively working through the details of our partnership. Nvidia technology has underpinned our successes from the beginning, powers our systems today, and will continue to be central as we move forward.”
Nvidia and Oracle did not respond to requests for comment.
All this is happening against the backdrop of a changing investment landscape for AI, where realities about which aspects of the technology are actually going to make money are being publicized.
While investors consider whether OpenAI will be able to pay for the $1.4tn compute deal, the reality is AI is causing more damage up the food chain. Some software stocks have seen a massive selloff this week, partly driven by the launch of a new anthropic AI tool that can complement many professional services, sparking fears that business models exposed to competition from AI products will be disrupted.
This is the other side of “jagged AI,” which is the term for advanced AI tools with uneven talents, such as being good at sorting documents but less good at solving complex math problems. If advanced systems are good at automating legal work, older companies in service industries will be at a disadvantage. Losers are starting to emerge and are being picked up by investors.
The competitive effects are also reaping at the top of the AI pyramid. OpenAI’s chatbot, ChatGPT, is lagging behind competitors. data released on tuesday show that its market share has declined from 69% to 45% due to the growth of Google’s Gemini, xAI’s Grok, and Anthropic’s Cloud. OpenAI appears to have stepped back from the growing discussion of super-intelligence over the past months, and is instead focusing on profitable mundane tasks like advertising and adult content.
The apparent evaporation of the $100 billion deal may be in line with last year’s sci-fi rhetoric and this year’s practicalities. The question is who will have the bill?
“I think it will have an impact,” Nguyen said. “I mean, it’s that statement: The market can remain irrational longer than you can remain rational.”
