Nvidia Invests Billions in OpenAI. OpenAI Spends It on Chips From Nvidia. Here's Why That's Making Wall Street Nervous.

Follow the money in AI right now, and it has an odd habit of ending up back where it started.

Nvidia puts billions into an AI lab. That lab spends billions on cloud capacity. The cloud provider spends billions on Nvidia chips. Zoom out far enough, and a huge amount of the industry's headline spending is the same dollars moving through a small, tightly connected circle of companies.

How the Loop Actually Works

The mechanics are simpler than the jargon suggests. An upstream company, usually a chipmaker or cloud provider, buys equity in a downstream AI lab. That lab then uses the cash, at least in part, to buy products or services from the same investor, or from another company further along the same chain.

• Nvidia → OpenAI: Nvidia and OpenAI announced a letter of intent for Nvidia to invest up to $100 billion as OpenAI deployed 10 gigawatts of Nvidia systems. That structure later collapsed; Nvidia instead took a $30 billion equity stake in OpenAI's $110 billion funding round.

• OpenAI → Oracle: OpenAI committed roughly $300 billion to Oracle for cloud infrastructure as part of the Stargate data-center project. Oracle, in turn, spends heavily on Nvidia chips to build that capacity.

• OpenAI → Microsoft: OpenAI agreed to purchase $250 billion worth of cloud services from backer Microsoft.

• Nvidia → CoreWeave: Nvidia took roughly a 7% stake in CoreWeave and separately committed to buy $6.3 billion worth of CoreWeave's cloud capacity, some of which OpenAI and Microsoft then lease to run their own workloads.

• OpenAI → AMD: OpenAI agreed to deploy tens of billions of dollars of AMD chips, with OpenAI positioned to become one of AMD's largest shareholders as part of the arrangement.

• Nvidia and Microsoft → Anthropic: the two companies said they'd invest up to a combined $15 billion in Anthropic, which in turn committed to $30 billion in Microsoft Azure cloud spending.

By 2026, analysts tracking these arrangements put the identified total above $800 billion, with some broader vendor-commitment tallies, adding up OpenAI's combined obligations across Azure, Oracle, AWS, CoreWeave, Nvidia, and Broadcom through 2035, cited as topping $1.1 trillion.

The News That Reignited the Debate

This isn't old news. The Wall Street Journal and Bloomberg reported on July 27, 2026 that Nvidia was in early-stage talks to guarantee as much as $250 billion in lease payments so OpenAI can use a 10-gigawatt data center SoftBank is building in Ohio. Separately, Nvidia was said to be considering providing up to $350 billion in financing specifically so OpenAI could purchase more Nvidia GPUs.

People familiar with the discussions cautioned the talks were preliminary and could still collapse or change substantially. But the reporting itself was enough to reignite a debate that's been simmering for over a year.

"The fact that Nvidia is providing additional guarantees on OpenAI's data-center debt is a signal of demand for AI buildout, but it also serves as a reminder of the funding strain," said Billy Leung, an investment strategist at Global X Management.

Where the Real Disagreement Sits

Skeptics have raised two fairly specific worries, not just a general unease. First, that these deals can make revenue and demand look more organic and robust than they really are, potentially leading investors to overvalue stock or lenders to extend more credit than the underlying business could support on its own. Second, that they tie the financial fortunes of a handful of enormous companies so tightly together that trouble at any one of them could cascade through the rest.

Underlying both worries is a rising debt load across the sector. OpenAI, SoftBank, CoreWeave, Nebius, and Oracle have all taken on significant leverage to fund data-center and chip buildouts, and even cash-rich giants like Google, Amazon, and Meta have issued unusually large amounts of debt this year.

Defenders push back with a different framing entirely. Building frontier AI infrastructure is extraordinarily expensive, and the most advanced chips remain genuinely hard to secure, so pairing long-term purchase commitments with financing is, in this view, simply how companies lock in scarce supply, not evidence of manufactured demand. Asset manager Janus Henderson has described the pattern as more of a "virtuous circle" that helps line up suppliers, builders, and customers to meet a real, fast-growing need for computing power.

The companies most directly named in these criticisms have pushed back hard. When CoreWeave's stock fell more than 50% from its peak amid demand-slowdown and circular-financing concerns, CEO Michael Intrator called the criticism "ridiculous," arguing Nvidia's stake in his company is too small to actually be propping up its operations.

AI Circular Financing: FAQ

Jans Bock-Schroeder

Publisher & Founder of AI Angst

Coming from the world of art, photography, and the luxury market, Jans launched AI Angst in 2025 to explore the cultural, ethical, and psychological impacts of artificial intelligence. His work bridges creative vision with critical technology analysis, offering clarity in an era of rapid technological change.