I remember the moment the CFO said the word circular and the room fell quiet. You could almost hear analysts rearranging their spreadsheets. I sat there thinking: if money is moving in a loop, who benefits and who gets left holding the risk?
I’m going to pull you through what Nvidia is arguing, what the skeptics are warning about, and the hard numbers that refuse to behave like a comforting narrative. You’ll get facts, a little blunt analysis, and a sense for why this matters to anyone who trades, builds, or depends on cloud compute.
At the Q2 earnings call, Nvidia defended its investments
Colette Kress, Nvidia’s chief financial officer, described the company’s investments in frontier AI firms as purposeful growth moves — not a financial merry-go-round. She framed those commitments as drivers of future demand for Nvidia chips and as investments with limited downside.
That’s the company line: invest in AI labs, they buy your hardware, the ecosystem grows, and you profit. But I want you to watch the flow of capital the way a detective traces a wire. The $105 billion guarantee for OpenAI’s Ohio data center (about €97 billion) reads like a vote of confidence — and like a structural lever tying two businesses together.
What is circular financing?
Circular financing is when capital is effectively cycled through related parties so that the originator funds demand that returns revenue back to itself. In Nvidia’s case, critics say the firm is deploying billions to AI outfits that then spend much of that on Nvidia chips. That creates a feedback loop where investment and product demand begin to look less like independent market signals and more like engineered growth.
At the same time, some deals are non-binding gestures
Nvidia has been orchestrating headline-grabbing commitments: a $500 billion (€460 billion) AI infrastructure push with Wall Street names and a previously announced $100 billion (€92 billion) OpenAI pledge that never fully closed.
Most of those arrangements exist as memoranda of understanding — public promises without legal teeth. From a PR perspective they build momentum. From a risk point of view, they’re thinner than they sound. You should ask: is the hype funding actual buildout, or are the headlines part of the fuel that keeps valuations and expectations lofted?
How does Nvidia invest in AI companies?
Nvidia moves beyond selling GPUs: it underwrites projects, helps syndicate capital, and sometimes guarantees financing. It partners with hyperscalers, Wall Street funds, and AI labs like OpenAI. That changes the balance sheet: instead of only recognizing sales when chips ship, Nvidia also takes on exposure tied to customers’ ability to execute big infrastructure projects.
At the market level, concentration is rising
Five customers now account for roughly 70% of Nvidia’s accounts receivable, a concentration that tightened notably year over year. That’s not just a statistic — it’s a structural stress point. If a major customer delays payment or cancels a build, the impact is magnified.
Google’s recent first negative free cash flow quarter since 2004 — driven in part by heavy AI spending — is a concrete example of how hyperscaler balance sheets can wobble under this new load. If the customers who buy the chips suffer a liquidity squeeze, those chips don’t magically move to other buyers overnight.
Five customers accounted for 70% of NVIDIA’s accounts receivables at the end of Q2 FY27, and for some customers it’s allowing payment terms anywhere from 3 months to a year.
Up from 56% a year ago. pic.twitter.com/BIibFuiUOi
— Ed Zitron (@edzitron) August 26, 2026
At the ecosystem level, dependencies are retyped as strategy
Nvidia’s pitch is that its compute platform is “fungible and durable” and can be redeployed if a customer falters. That’s plausible — GPUs are valuable across many workloads — but redeployment takes time, sales channels, and predictable demand.
I’ve seen corporate strategies bend to protect a market leader before; this can make the leader look invincible even when fragility builds under the surface. The investments look like capital allocation — and sometimes they behave like demand engineering designed to keep a growth narrative intact. One metaphor: imagine a river that curves back on itself and appears wider because the same water keeps coming around. That image captures the core worry without pretending the math is simple.
At the investor table, the debate is between optimism and systemic risk
On paper, Nvidia’s Q2 revenue was $96.2 billion (€88.5 billion) and executives forecast another huge gain next year. Jensen Huang said growth was limited more by supply than demand. That’s a data point investors love. But here’s the rub: the numbers can look tidal until one link snaps.
Critics warn about a domino train built from MOUs, guarantees, and inflated expectations. If a major project collapses or a partner can’t pay, the loop tightens and liabilities become headline risk. The other metaphor: a carnival mirror that makes a room look full; up close, the reflection is distorted and the crowd may be smaller than it seems.
I’ll leave you with two practical lenses: watch actual binding contracts, not press releases, and watch customer cash flow at hyperscalers. If you trade or build on the assumption that today’s commitments equal tomorrow’s realized demand, you should be ready for volatility.
So what happens if the loop breaks and who absorbs the fallout?