AI Roundtripping: NVIDIA, OpenAI, Oracle and the Circular Financing Debate
A recent series of massive deals between major AI players has raised questions about “roundtripping” or circular financing in the AI industry. In September 2025, OpenAI struck a five-year cloud computing contract with Oracle, reportedly worth $300 billion - one of the largest cloud deals ever. Around the same time, NVIDIA announced plans to invest up to $100 billion in OpenAI, with OpenAI committing to spend that cash on NVIDIA’s cutting-edge chips. These agreements (along with related deals involving firms like Microsoft, SoftBank, AMD, and CoreWeave) mean that many of the key companies in AI are simultaneously each other’s investors, suppliers, and customers.
Twitter/X calls it an “infinite money glitch” - a tongue-in-cheek reference to the virtuous cycle where the same money circulates among participants, boosting each company’s revenue and valuation as long as the music keeps playing. This has fueled debate over whether such circular deals signal an unsustainable bubble or simply reflect strategic partnerships typical of a booming industry - and it’s everywhere in the news.
Here’s our take.
The Circular AI Investment Loop
In simple terms, the AI roundtripping loop works like this: NVIDIA invests cash into OpenAI, and in exchange OpenAI agrees to spend that money on NVIDIA’s GPU hardware. OpenAI also uses Oracle’s cloud services for its AI workloads, becoming a huge customer for Oracle. Oracle, in turn, uses the revenue from OpenAI to purchase more NVIDIA chips for those data centers. NVIDIA ends up booking significant sales of its GPUs - sales effectively funded by NVIDIA’s own investment money that came full circle through OpenAI and Oracle. All three players benefit on paper: OpenAI secures massive computing power, Oracle locks in a marquee cloud client, and NVIDIA boosts its chip sales. Crucially, the same money moves around in just one circle, but all of a sudden everyone’s valuations go up.
This kind of circular financing isn’t entirely unique to AI. It resembles “round-tripping,” where companies trade or invest in one another to inflate each other’s revenues without truly increasing net income. In fact, similar dynamics occurred during the dot-com era - for example, Cisco famously lent money to telecom startups so they could buy more Cisco gear, only to suffer when those customers defaulted in the 2001 tech bust.
Today, this web of alliances has become so extensive that OpenAI’s agreements for chips and cloud capacity with NVIDIA, AMD, and Oracle combined could exceed $1 trillion in value. You start to see how this has set off bubble warnings.
Bear Case: Bubble Concerns and Risks
Skeptics argue that these circular deals are a cause for concern, potentially inflating a bubble reminiscent of the late-1990s dot-com boom. They point out that such self-referential transactions can boost revenues on paper without corresponding growth in real end-user demand. In the AI sector, the fear is that much of the explosive growth is not actually supported by real consumer demand, but rather by insular funding arrangements among the same few firms. Likewise, Jim Chanos (founder of the largest exclusive short-selling investment firm) quipped that it’s “a bit odd” to proclaim infinite AI demand while the sellers (chip makers) are subsidizing the buyers (AI labs) via these deals.
The risk, critics warn, is systemic: when each party’s growth depends on the others’, a single failure could cascade through the whole circle. When everyone is both buyer and seller in circular deals, you’ve created massive correlation risk. If OpenAI can’t pay Oracle, Oracle can’t pay NVIDIA, and NVIDIA’s stock crashes. In such a worst-case scenario - say AI progress stalls or revenue from external customers falls short - the unwinding of this web could, in theory, be a bursted bubble, with substantial damage to some of the market’s most valuable companies. It’s essentially a high-stakes, leveraged wager that rapid AI growth (up to and including ambitions like AGI) will arrive on schedule to justify these enormous commitments. If not, the downside could be severe.
Bull Case: Calculated Optimism
On the other side, the bull case is that these arrangements are no big deal - essentially a rational strategy in a booming industry, or a form of vendor financing, which is a common practice. It’s not unlike an auto manufacturer providing loans or leases so customers can buy its cars - the company is both seller and creditor, which helps move product and lock in business. In the tech world, vendor financing or investing in one’s own customers has precedent and can be mutually beneficial. NVIDIA, OpenAI, and Oracle each have something the others need, and these deals simply solidify their partnership in scaling AI infrastructure. For example, NVIDIA secures a long-term outlet for its chips, OpenAI gains the massive computing capacity it needs (with financial backing), and Oracle boosts its cloud platform with a marquee client. Each is betting on the others’ success, which in a high-growth sector can be a savvy way to accelerate development. These arrangements are common and don’t create “infinite” value. They simply finance growth up front in exchange for goodwill or product sales, and can be sustainable if demand materializes as expected.
Importantly, many investors see real substance behind the companies mentioned. Unlike nefarious accounting tricks, the deals are funding very tangible projects - giant data centers, cutting-edge chips, and advanced AI models that will likely generate significant revenue in the future. From this perspective, the mutual investments are simply a way to meet that surging demand faster. It’s worth noting that big, well-capitalized companies are involved, firms like Microsoft, Oracle and NVIDIA - which arguably reduces the risk of any immediate financial collapse. These are not fly-by-night startups trading funny money; they are established players with diverse businesses.
NVIDIA’s investment ensures its chips remain central to OpenAI’s growth, warding off competitors and guaranteeing a market for its products. Such symbiotic relationships have historically been healthy “win-win” partnerships, where each company benefits from the other’s success, much like other industries where suppliers and buyers often invest in each other. As long as AI usage and revenues (not just valuations) keep rising across the broader economy, the circular nature of these deals may simply reflect a tight-knit ecosystem rather than a fraudulent bubble. Really, the bull case is that while these financing loops are unusual in scale, they are driven by genuine growth prospects and will likely unwind gracefully with increased competition (or even pay off spectacularly) if the optimistic projections for AI hold true.
The Bottom Line
The debate over NVIDIA and its partners’ circular financing ultimately hinges on whether the AI boom can justify these massive investments. The bear case sees echoes of past bubbles, warning that companies investing in and relying on each other’s purchases could inflate valuations beyond true market demand. The bull case views it as par for the course, with a few highly successful companies taking calculated bets on each other to build capacity for an AI future that’s already producing real applications and revenue.
Both perspectives acknowledge the unprecedented scale and interdependence among the players involved. For investors, the key question is whether this tight circle becomes a virtuous or vicious cycle. If AI adoption continues to grow, particularly in enterprise use-cases, these partnerships may look like forward-thinking alliances that accelerated innovation. And even if the carousel slows, due to competition, technical challenges, or shifting market dynamics - that doesn’t mean the AI story ends. The winners might change, but the underlying trend toward intelligent, compute-driven industries will likely endure. For now, it’s less about whether the loop keeps spinning forever and more about how it shapes the next phase of a generational tech-shift.