Nvidia is close to a deal that would guarantee roughly $100 billion in credit for OpenAI, according to The Information. If it closes, the chipmaker wouldn’t just be selling GPUs to OpenAI. It would be standing behind the money OpenAI borrows to buy them.
That’s the part worth slowing down on. A credit guarantee means Nvidia puts its own balance sheet on the line so lenders feel safe extending OpenAI enormous sums. Nvidia is one of the most valuable companies on the planet, so its backing turns OpenAI from a risky borrower into a much safer bet. Lower risk means lower interest rates and access to far more capital than OpenAI could raise on its own.
Why OpenAI needs this
OpenAI’s problem isn’t demand. It’s the bill for compute.
Training and running frontier models eats through data centers, power, and GPUs at a scale few companies have ever attempted. OpenAI has committed to buying staggering amounts of chips and cloud capacity, and that requires cash it doesn’t generate yet. Revenue is climbing fast, but it’s nowhere near covering the infrastructure buildout.
A $100 billion credit backstop solves the timing gap. OpenAI can borrow now to build now, and pay it down as revenue catches up.
The circular money question
Here’s what stands out to me. Nvidia sells the chips. Nvidia already agreed earlier this year to invest heavily in OpenAI. Now Nvidia may guarantee the credit OpenAI uses to buy Nvidia’s chips.
That’s a tight loop. Critics have started calling this kind of arrangement circular financing, where a supplier helps fund the customer that buys its products. It boosts sales on paper and keeps the demand engine running. It also concentrates risk. If OpenAI stumbles, Nvidia is exposed on the investment, the guarantee, and the lost future chip orders all at once.
Supporters see it differently. Nvidia has the strongest read on AI demand of anyone in the market, and it’s betting that OpenAI’s compute needs are real and durable. Backing that bet with a guarantee is a way to lock in one of its biggest customers for years.
How this compares to the status quo
Until recently, the AI buildout was funded the normal way: equity rounds, corporate cash, and straightforward debt. Microsoft’s tens of billions into OpenAI set the template, mostly in the form of cash and cloud credits.
This is a different move. A vendor guaranteeing customer debt at nine-figure scale is unusual, and it signals how much capital the frontier of AI now requires. When a company needs $100 billion in credit backed by its chip supplier, you’re looking at infrastructure spending closer to a national utility rollout than a software startup.
What this means for practitioners
A few takeaways if you build on or compete with these players:
- Compute supply gets more locked in. Deals like this tie OpenAI and Nvidia closer together, which can shape who gets priority access to the newest GPUs.
- The capital bar keeps rising. Frontier AI is becoming a game for players who can move $100 billion. Smaller labs will lean harder on cloud providers and open models.
- Watch the concentration risk. The more Nvidia’s fortunes depend on a handful of mega-customers it also finances, the more the whole AI market rides on those same bets paying off.
- Regulators may take notice. Vendor-financed demand at this scale tends to draw scrutiny over accounting and competition.
The deal isn’t final, and The Information notes it’s still being worked out. But the direction is clear. The companies building the AI economy are starting to finance each other directly, and the sums involved now rival what governments spend on infrastructure.
For the full details, check the original reporting at The Information.