$27.5B median capex − $14B cash − committed financing/prepayments − $8B dedicated GPU financing = only a $5.5B remaining value gap. This is a set of figures in Nvidia’s earnings report that was buried under the “$279B committed” narrative. The management’s subtext is: we don’t actually need external capital; the data center asset portfolio is still unencumbered and can be pledged to raise money at any time. Nvidia’s $279B upstream commitments (memory, components, capacity) climbed from $95.2B to nearly double within six months. This isn’t just stockpiling—it’s Nvidia acting as the “central bank” of the AI industry, trading long-term contract credit for supply-chain stability. The cost is that gross margin has slid from 75% all the way down to a Q4 trough of 71%-72%.
Critics see it as circular financing.
Risk: Nvidia lends money to customers, who then buy Nvidia GPUs. But Jensen Huang says the risk is low, arguing that the real gap in the capital structure is only $5.5B. And there are already existing inventory assets that can be used as collateral—those data centers that haven’t been pledged.
Who wins: memory makers get locked-in long orders, and cloud companies entering Nvidia’s credit system get low-cost financing. Who loses: second-tier AI players not on Nvidia’s list, and anyone still clinging to the old narrative that “Nvidia is just a chip company.” A key risk point is receivables concentration—three direct customers account for 56%.
Huang says risk is low, but every time he says something like that, it’s at the moment when concentration is highest. The next thing to watch is whether the proportion of financing from non-affiliates rises—that would be real evidence that the circular-financing thesis has failed.
Critics see it as circular financing.
Risk: Nvidia lends money to customers, who then buy Nvidia GPUs. But Jensen Huang says the risk is low, arguing that the real gap in the capital structure is only $5.5B. And there are already existing inventory assets that can be used as collateral—those data centers that haven’t been pledged.
Who wins: memory makers get locked-in long orders, and cloud companies entering Nvidia’s credit system get low-cost financing. Who loses: second-tier AI players not on Nvidia’s list, and anyone still clinging to the old narrative that “Nvidia is just a chip company.” A key risk point is receivables concentration—three direct customers account for 56%.
Huang says risk is low, but every time he says something like that, it’s at the moment when concentration is highest. The next thing to watch is whether the proportion of financing from non-affiliates rises—that would be real evidence that the circular-financing thesis has failed.