Nvidia’s once eye-popping plan to back OpenAI with as much as $100 billion is reportedly no longer moving forward in the way many expected. What was pitched as a landmark effort to massively expand AI computing capacity is now said to be effectively paused, as Nvidia rethinks both the scale and structure of any potential partnership.
Back in September, Nvidia outlined a vision that would have stood out even in an industry known for big numbers: a computing buildout of at least 10 gigawatts, paired with financial support that would allow OpenAI to lease enormous amounts of Nvidia chips. CEO Jensen Huang reportedly described the concept as “the largest computing project in history,” signaling just how serious the ambition was.
More recent reporting, however, suggests the talks never advanced far beyond the earliest stages. According to people familiar with the discussions, internal concerns inside Nvidia slowed momentum. Huang has reportedly emphasized to others in the industry that the original letter of intent was not binding, and he has also privately questioned OpenAI’s business discipline. That mixture—non-binding commitments plus internal skepticism—appears to have helped put the megadeal on ice.
Now, instead of a sweeping infrastructure-and-financing arrangement, the conversation has shifted toward a more traditional format: a large equity investment. The numbers could still be enormous, potentially reaching into the tens of billions of dollars as part of OpenAI’s ongoing fundraising, but it would be a very different commitment than building and underwriting a giant, dedicated compute project.
Competition in generative AI is a major reason the situation is changing so quickly. Google’s Gemini is believed to have slowed ChatGPT’s growth and raised internal concerns about market momentum. Anthropic is also applying pressure, particularly with tools like “Claude Code,” which aims at developer and coding workflows where usage can scale quickly. Nvidia, for its part, is hedging its bets: it has already committed up to $10 billion in investments in Anthropic, even as it continues to explore how (and how much) it should back OpenAI.
For OpenAI, any delay in securing long-term computing capacity is a meaningful setback, especially with an eye on a possible IPO by the end of 2026. Compute has become one of the biggest constraints in advanced AI development, and investors tend to watch infrastructure needs closely—particularly after OpenAI CEO Sam Altman previously referenced liabilities totaling $1.4 trillion, a figure that has reportedly made some backers uneasy relative to current revenue levels.
Despite the hesitation, there are strong incentives on both sides to reach some kind of agreement. Nvidia has a strategic interest in keeping major AI leaders anchored to Nvidia GPUs, especially as more rivals experiment with alternatives such as Google’s TPUs or Amazon’s Trainium hardware. If top AI companies shift more of their workloads to in-house or partner chip ecosystems, it could reshape the balance of power in the AI hardware market over time.
Meanwhile, OpenAI still needs deep-pocketed partners to fund the compute required to train and run frontier models at scale. Adding another twist, Amazon is reportedly exploring its own investment in OpenAI that could be as high as $50 billion—an option that could further change the negotiating dynamics and the competitive landscape across cloud computing and AI.
In short, the blockbuster $100 billion plan may be stalled, but the fight for AI dominance hasn’t slowed down. If anything, it’s pushing Nvidia, OpenAI, and their rivals toward faster, more strategic decisions about funding, chips, and the future infrastructure that will power the next generation of AI.
