Invest1 publisher2 min readPublished
Nvidia's $500 billion GPU financing plan depends on chips holding value as long as the loans run
Nvidia is lining up Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to raise over $500 billion with its GPUs as collateral. The plan works only if that hardware keeps its value for as long as the debt it secures stays outstanding.
The Investor · Invest desk
Drafted by a language model from the sources cited here and checked against its claim ledger before publication. How we use AISend a correction

What happened
- Nvidia backs only select deals, with support capped at 25% of residual value, the amount the chips are expected to be worth when the financing term ends.
- The money is meant for data centers and GPU clusters, with AI labs and cloud service providers as the target customers.
- By October 1, 2026, some Wall Street investors worried that the chips lose value too fast were demanding stronger guarantees in future deals.
- Nico Caprez, Jensen Huang's son-in-law and Nvidia's vice president of global AI infrastructure growth, is helping secure the financing, according to thein.fo.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- exposure Any guarantee Nvidia writes ties its own balance sheet to the resale value of chips that its next product generation is built to make less attractive.
- decision Nvidia now has to decide how much residual risk to absorb in exchange for faster closings on deals that end in purchases of its hardware.
- capability AI labs and cloud providers gain a way to pay for data centers whose financing is tied to filling them with Nvidia GPUs.
On a backstopped deal the lender still holds at least 75% of the end-of-term value risk, and on deals without the guarantee it holds all of it [12][4]. The cap is smallest where the worry is largest. A guarantee sized at 25% of residual value covers fewer dollars the lower that value is set, so the chips lenders expect to depreciate fastest come with the thinnest cushion [13].
The $500 billion is other people's money [1]. Spread evenly across the six firms, it would come to more than $83 billion each [14]. Nvidia's own stake is a contingent promise on some of those deals [4], plus an argument about the collateral itself. Jensen Huang has said the chips are becoming an "investable asset class" capable of generating revenue [7].
Lenders took about seven weeks from the August 10-11, 2026, announcement to start asking for more [2][8][15]. The Crypto Briefing account does not give loan tenors, interest rates or a depreciation schedule. Nobody outside the deals can yet compare how fast the collateral ages with how fast the debt is repaid.
The plan can resolve three ways. Lenders conclude GPUs hold their value well enough to back long-dated debt, and Nvidia has created a financing market around its own products, in Crypto Briefing's phrase [17]. Nvidia lifts the cap above 25% to keep deals closing [16]. Or the guarantee stays confined to select deals, and lenders carry the full residual risk on the rest of the pipeline [4].
I'd expect the second. The request for more came within weeks, from the side of the table holding at least three-quarters of the risk on backstopped deals [15][12]. The counter-case is that the report attributes the doubts to "some Wall Street investors," a group that may not include the six partners [9]. If Nvidia's next deals close with the cap still at 25%, this view is wrong [3].
What to watch
- Disclosed loan tenors on the first GPU-backed deals, and how they compare with the gap between Nvidia chip generations.
- Whether Nvidia extends its residual-value guarantee beyond select transactions to more of the $500 billion pipeline.
- Whether any of the six partners publishes the share of GPU value it will lend against.