How Nvidia’s Financing Web Drew a Dot-Com Comparison

Two of Wall Street’s most prominent voices have drawn an explicit line from today’s AI infrastructure boom to the excesses of the dot-com era. Mark Cuban, the tech billionaire and former “Shark Tank” investor, took to X on Tuesday to warn that Nvidia’s practice of financing its own customers’ chip purchases had turned the company into the de facto initial public offering engine of the current cycle. His post came after another user argued Nvidia was acting as a “backstop” for AI builders, making itself dangerously intertwined with their fate. “This is so analogous to the dot com burst,” Cuban wrote. “But instead of IPOs, Nvidia is the ‘ipo,’ funding everyone and anyone.” He warned that a single breakthrough by a rival chip provider or a misstep by Nvidia itself “could crumble” the entire edifice.

Hours later, Michael Burry—famous for his prescient bet against subprime mortgage bonds before the 2008 crisis—reinforced the alarm. He posted a chart showing that the cost of five-year credit default swaps on Nvidia had roughly doubled in two months, a sign that deep-pocketed bond investors are demanding much higher insurance against a potential default. “All this overreaching by #nvda to push the circular spending to biblical proportions,” Burry added. He has publicly disclosed a short position in Nvidia, arguing that much of the chipmaker’s revenue is financed internally and kept off its balance sheet, creating a “circular arrangement” disconnected from end-user demand.

Nvidia shares traded about 2% lower on Wednesday, leaving them 18% below their May peak but still up roughly 13-fold since the start of 2023. The company has struck hundreds of billions of dollars in deals with players across the AI ecosystem—OpenAI, Microsoft, CoreWeave, and SK Hynix among them—often providing the financial fuel those customers need to buy its own graphics chips. A request for comment from Nvidia was not immediately returned. CEO Jensen Huang, speaking internally last November, leaned into the meme of Nvidia propping up the AI economy, saying, “We’re basically holding the planet together—and it’s not untrue.”

Why Cuban and Burry Are Right to Watch the Circular Spending

The warnings from Cuban and Burry go beyond typical market skepticism because they target the financial architecture underpinning the AI build-out, not just stock valuations. Cuban’s dot-com comparison is more than rhetoric: during the internet bubble, start-ups rushed to go public to secure cash, and the market absorbed their overvalued shares until confidence snapped. Today, Nvidia is performing a similar capital-supply function, effectively backstopping the balance sheets of its customers so they can keep ordering chips. If AI demand disappoints or a cheaper alternative emerges, those customers may be unable to pay, leaving Nvidia exposed as both supplier and creditor.

Cuban’s Dot-Com Analogy: Why Nvidia as ‘IPO Engine’ Is a Systemic Risk

Cuban’s concern is about concentration and interconnectedness. When a single company finances its own customer base, a downturn in that base doesn’t just cut revenue—it triggers credit losses. During the dot-com crash, the pain spread from overfunded start-ups to the banks and public markets that backed them. In this script, Nvidia plays the role of both the hyped start-up and its underwriter. The danger, as Cuban put it, is that an exogenous shock—say, a viable alternative to its GPU architecture or a sharp pullback in corporate AI spending—could unravel the web of credit and purchase commitments all at once.

Burry’s CDS Signal: The Insurance Market Is Suddenly Nervous

Credit default swaps are the instruments Burry used to bet against mortgage bonds. When CDS prices spike, it means sophisticated fixed-income investors see a higher probability of default. The fact that Nvidia’s five-year CDS has doubled in a matter of weeks is a concrete market signal, not mere opinion. Burry interprets it as evidence that the “circular spending” is being recognized as a balance-sheet risk—Nvidia is arguably extending so much credit to customers that its own financial health now depends on those customers’ success in a sector that has yet to prove durable end-demand outside a handful of hyperscalers.

The Real Question: Where Is the End-Customer Demand?

Both investors share a deep reservation: the AI infrastructure build-out may be ahead of actual paying end-users. Burry has argued that tech giants are overinvesting in chips and data centers that will quickly become obsolete, dragging out depreciation to flatter earnings, and using circular deals to sustain the hype. If much of Nvidia’s demand is essentially financed by Nvidia itself, then reported revenue growth may not reflect genuine market absorption. That creates a familiar pattern—demand looks insatiable until financing tightens, at which point the real level of need becomes painfully clear. The risk is not that AI is a fiction but that the current level of investment has borrowed heavily from future demand that may never materialize at the scale priced into today’s commitments.

What Investors and AI Leaders Should Do Now

These warnings don’t mean a crash is imminent, but they do raise specific questions that investors, board members, and AI industry leaders can no longer ignore. The actionable steps are concrete:

  • For Nvidia investors: Monitor the company’s quarterly allowance for credit losses and the pace of customer financing disclosed in its filings. Watch for any uptick in days sales outstanding among its large AI customers, which would indicate that Nvidia is indirectly funding an increasing share of its own revenue.
  • For AI start-ups and enterprise buyers: Reassess the share of capital expenditure that is reliant on vendor financing or deferred payment structures from Nvidia. A sudden tightening of that credit—whether from Nvidia itself or from the banking partners that often sit behind such arrangements—could force an unplanned capex crunch.
  • For credit and equity markets: Follow the five-year CDS spread on Nvidia as a real-time stress gauge. A further doubling or sustained upward trend, especially alongside a retreat in Nvidia’s stock, would indicate that bond desks are pricing in a rising probability of a credit event—not just an equity correction.
  • For competitors and chip buyers: Diversify chip supply chains where possible and stress-test AI infrastructure plans against a scenario in which next-generation alternatives (from AMD, Intel, or custom in-house designs at the hyperscalers) erode Nvidia’s pricing leverage and, by extension, the value of the financed hardware already deployed.

Risk & Opportunity Assessment

Commercial RiskHighNvidia’s heavy reliance on financing its own customers’ purchases creates a credit exposure concentrated in a single industry; if AI demand moderates or customers cannot service their debts, revenue and loan losses would hit simultaneously.
Competitive RiskMediumCuban explicitly warned that a breakthrough from another chip provider could undermine Nvidia’s dominance; hyperscalers are also building custom silicon, which over time could reduce dependence on Nvidia GPUs.
Regulatory RiskLowNo specific regulatory threat is mentioned, though the scale of Nvidia’s customer financing could attract attention from banking or securities regulators if it resembles off-balance-sheet lending.
Reputation RiskMediumHigh-profile investor critiques from Cuban and Burry, combined with a meme-driven narrative that Nvidia is ‘holding the planet together,’ could rapidly flip sentiment if the company’s AI growth story is questioned.
Technology DisruptionHighThe story turns on Cuban’s premise that one exogenous chip breakthrough could cause the entire financing structure to ‘crumble,’ indicating a high sensitivity to any novel architecture that challenges CUDA-based GPUs.
Commercial OpportunityHighNvidia’s position as the linchpin of the AI infrastructure boom gives it enormous pricing power and a captive customer base, provided the circular spending does not unwind before end-demand solidifies.