$100 billion doesn't just evaporate overnight. Unless, of course, it was never really there to begin with.
NVIDIA CEO Jensen Huang spent his weekend doing damage control, dismissing reports of OpenAI investment drama as "nonsense" after The Wall Street Journal suggested the chipmaker was having second thoughts about its September 2025 commitment. But here's what makes this corporate theater particularly amusing: the original deal was structured as a non-binding memorandum.
Non-binding. As in, "thanks for the press coverage, but we'll see."
<> "I believe in OpenAI. The work that they do is incredible. They're one of the most consequential companies of our time." - Jensen Huang, January 31, 2026/>
Sure, Jensen. But actions speak louder than PR statements, and the reported shift from a $100 billion infrastructure commitment to "tens of billions" in equity investment tells a different story. That's not a rounding error—that's a fundamental change in strategy.
When Handshake Deals Meet Market Reality
The original September announcement painted a picture of AI infrastructure nirvana: 10 gigawatts of AI data centers using NVIDIA's Vera Rubin platform, with the first phase targeting the second half of 2026. It was the kind of moonshot that makes investors swoon and competitors panic.
But according to the WSJ, Huang has been privately criticizing OpenAI's business approach as "lacking discipline" while expressing concerns about competitors like Anthropic and Google. Translation: NVIDIA realized putting all its chips (literally) in one basket might not be the smartest play when you're supplying the entire AI gold rush.
The timing is telling. As OpenAI scrambles to raise a separate $100 billion funding round—with Amazon, Microsoft, and SoftBank all circling—NVIDIA appears to be hedging its bets rather than doubling down.
What Nobody Is Talking About
Everyone's focused on the dollar figures, but the real story is about market positioning. NVIDIA has been OpenAI's preferred compute partner for a decade, since the DGX supercomputer days through ChatGPT's breakthrough. That relationship gave NVIDIA incredible leverage in the AI boom.
But markets evolve. Competitors emerge. And betting $100 billion on a single customer—even one as "consequential" as OpenAI—starts looking less like visionary leadership and more like dangerous concentration risk.
The non-binding structure wasn't an oversight; it was smart business. NVIDIA gets to:
- Maintain flexibility as the AI landscape shifts
- Avoid massive capital commitment upfront
- Keep options open for partnerships with Anthropic, Google, or whoever emerges next
Meanwhile, Wolfe Research just raised NVIDIA's price target to $275, citing "rising demand for rack-scale systems and higher selling prices." The market isn't worried about NVIDIA's business fundamentals—just its OpenAI exposure.
The Infrastructure Reality Check
Deploying 10 gigawatts of AI infrastructure isn't like ordering pizza. It requires massive coordination, regulatory approvals, power grid negotiations, and perfect execution across multiple stakeholders including Microsoft, Oracle, and SoftBank through Project Stargate.
Any delays—which now seem likely given the "renegotiation"—could push OpenAI's next-generation model training timeline and affect the entire competitive landscape for advanced AI systems.
The cynical read? NVIDIA made a splashy announcement to maintain its AI kingmaker status, structured it as non-binding to retain flexibility, and is now quietly scaling back as market realities set in. OpenAI gets strung along while NVIDIA keeps its options open.
The charitable read? Two market leaders are navigating unprecedented territory and adjusting their partnership based on evolving circumstances.
Either way, when someone emphasizes that a $100 billion deal is "non-binding" while calling criticism "nonsense," you're watching corporate diplomacy at its finest. Just don't mistake it for commitment.
