Nvidia and OpenAI Rewrite a $100B Idea Into a $30B Bet

What changed since the earlier report

The Financial Times characterizes the original $100 billion arrangement as unfinished and not formalized, which helps explain why it could be rewritten instead of “broken.”

Put simply, this isn’t Nvidia and OpenAI walking away from each other; it’s them choosing a structure that’s easier to execute while the infrastructure race keeps accelerating.

The broader context hasn’t changed: OpenAI still needs enormous amounts of compute, and Nvidia still sits at the center of the hardware supply chain that makes today’s AI economics work. What’s changing is how risk, timing, and commitment get packaged.

Why the rest of the market cares

This kind of reset is becoming a pattern in AI infrastructure. Capital and capacity are now welded together, with funding rounds increasingly shaped by power availability, build timelines, and the realities of scaling data centers.

eWeek has been tracking how OpenAI has been lining up Middle East funding as part of its broader push to secure capital and compute. At the same time, the buildout isn’t slowing. In 2025, global data center investment hit $61B as companies raced to support AI workloads.

So the takeaway here isn’t “AI spending is over.” It’s that deal structure is getting more pragmatic. Even the biggest players are choosing cleaner, more bankable arrangements over grand frameworks that are harder to finalize.

Also read: the $650 billion AI spending push is accelerating data center expansion, even as investors debate how long the capex sprint can last.