OpenAI and Nvidia, two darlings of AI hype and longtime partners, appear to be at odds.
At the center of this disruption is Nvidia’s 100 billion investment in OpenAI that was announced in September 2025. As part of the agreement, Nvidia will build 10 gigawatts of AI data centers for OpenAI and invest $100 billion in the company in 10 installments, as each gigawatt comes online. Next, OpenAI is reportedly planning to use billions of dollars in investment from Nvidia to lease Nvidia chips.
At the time, the investment sparked concerns about circular trade in the AI industry and an intricately woven web of financial interdependence that could signal potential volatility, echoing that of the dotcom bubble. Meaning, if even one cog goes wrong and demand doesn’t pan out as expected, it can create a domino effect that brings down the entire system.
In a September announcement, the companies said the first gigawatt of computing power would come online in the second half of 2026 and that any further details would be finalized in the coming weeks. But in Nvidia’s SEC filing from November, OpenAI’s investment was still seen as “a letter of intent with an investment opportunity.”
Fast forward a few months, and a Wall Street Journal report from last week says that talks have not progressed beyond the early stages and that Nvidia CEO Jensen Huang has been privately criticizing the so-called lack of ethics in OpenAI’s business approach. Huang reportedly spent the past few months privately insisting to industry associates that the $100 billion deal was non-binding and not yet finalized.
After that report, Huang tried to reassure reporters in Taipei, Taiwan, by praising OpenAI and saying Nvidia would be “fully involved” in the company’s latest round of funding ahead of a rumored IPO later this year. Huang described the planned investment as “probably the biggest investment we’ve ever made,” but when asked if it would exceed $100 billion, he said, “No, no, nothing like that.”
But that wasn’t enough to allay investors’ fears, because another anonymous report dropped a few days later. It turns out, OpenAI is not happy with the speed at which Nvidia chips can compute the logic of some ChatGPT applications, and has been looking at different chip suppliers (such as startups Cerebras and Groq) to take 10% of their logic needs, according to a Reuters report on Tuesday.
The report also claims that OpenAI has blamed its weak Codex AI coding assistant on Nvidia hardware.
In response, it was now the turn of OpenAI executives to praise Nvidia. CEO Sam Altman took on X and said Nvidia makes “the best AI chips in the world,” and infrastructure manager Sachin Katti said Nvidia is “a very important partner for OpenAI in both training and guidance.”
But it seems that the guesswork and its massive memory requirements have been too much for Nvidia lately. The importance of prediction has always exceeded that of training as mature models. Agent AI hype has also increased the amount of data managed by the AI system during the prediction phase, driving the importance of memory.
To account for this, Nvidia bought Groq (no, not Grok), an AI chips startup reportedly being targeted by OpenAI, in its biggest acquisition yet. Then, last month, Nvidia presented its new Rubin platform, with a presentation that boasts of understanding and memory bandwidth wins.
Google ups the ante
It is reported that at the center of both Nvidia and OpenAI’s fear of each other is increasing competition, which is mainly done by Google.
Towards the end of last year, Google became the strongest competitor to both leading AI developer OpenAI and hardware infrastructure giant Nvidia.
First came the tensor processing units, Google’s custom AI chips designed for consideration, and in some tasks they were considered better than the GPU chips dominated by Nvidia’s offerings. Google’s TPUs are used not only by its own AI models, but also by OpenAI competition Anthropic and possibly Meta.
According to a Wall Street Journal report last week, Huang is also worried about competition from both Google and Anthropic in the dominance of the OpenAI market. It is reported that Huang fears that if OpenAI falls behind, it could affect Nvidia’s sales because the company is one of the chipmaker’s biggest customers.
OpenAI had to declare a “code red” in December, just a few weeks after Google’s latest release, Gemini 3, was considered to surpass ChatGPT. At the same time, the company also made major efforts to scale Codex to beat its rival Anthropic’s more popular coding Claude Code.
If investors’ fears come true, the deal doesn’t go as planned, and OpenAI can’t pay its high financial obligations, that means it will surpass only OpenAI and Nvidia. That’s because both companies sit at the center of the complex, tangled web of AI, with multiple multibillion-dollar deals between several companies, including OpenAI-Oracle’s $300 billion cloud venture that’s much bigger than Nvidia’s commitment. These deals have been a huge boon to the American economy, and if one deal goes down, it can take everything else down with it.
