Nvidia’s $500 billion AI financing plan faces a China collateral test
Nvidia wants Wall Street to fund AI compute like infrastructure, but a potential low-cost chip wave from China could weaken the collateral.
By Sal Moretti · Money Reporter
3 min read
Nvidia’s $500 billion AI financing push rests on a crucial bet: the company’s chips will keep producing enough revenue, and retain enough resale value, for lenders to treat them as dependable collateral.
Nvidia has signed memorandums of understanding with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to create compute-financing platforms. The aim is to mobilize more than $500 billion in third-party capital over time for customers building data centers and buying Nvidia hardware, according to Nvidia and reports by CNBC and The Wall Street Journal.
That number is a target, not a disclosed pot of committed cash. Nvidia has not disclosed the financial terms, individual firms’ commitments or a timetable for putting the planned capital to work, The Guardian reported.
Why could China affect Nvidia’s $500 billion AI financing plan?
The China risk is a future price-pressure scenario, not evidence that Nvidia chip values have already fallen apart. Ben Emons, founder of FedWatch Advisors, told CNBC that China’s expanding domestic compute capacity could lead to a price war in low-cost silicon.
If that happened, cheaper new hardware could pull down prices for used GPUs as well. That matters because a lender financing an AI data center may rely partly on the equipment’s resale value if a borrower cannot repay.
In practical terms, a default could leave a fund manager repossessing chips and trying to sell them into a weak market. If the resale proceeds do not cover the outstanding debt, the lender takes the loss.
GPUs have a short financial question hanging over them
Huang’s pitch is that Nvidia compute behaves like infrastructure: productive, revenue-generating and transferable between users. He has said broad adoption and ongoing improvements to Nvidia’s CUDA software can help installed hardware perform better after deployment and remain revenue-producing for longer.
The unanswered question is how long that earning power and residual value will last. CNBC reported that as newer chips take on frontier AI-model training, older GPUs can move into lower-margin inference work, a change that can affect their resale and collateral value.
That uncertainty is central to the financing math. Emons told CNBC that investors could demand returns of 11% to 17%, depending on their position in a deal’s capital structure, if they view the equipment as fast-depreciating rather than infrastructure-like.
The borrower pool adds another layer of risk. A Bank of America Securities note cited by CNBC said likely borrowers include non-investment-grade AI companies and neoclouds that may not be able to use conventional debt markets.
Why the threat is prospective
Nvidia remains the leading supplier of AI chips in the United States, with more than 75% market share by most estimates, CNBC reported. Huawei, China’s dominant AI-chip provider, has been on the U.S. Commerce Department’s Entity List since 2019, and the U.S. government said in May that Huawei Ascend AI chips violate U.S. export controls, preventing American companies from using them.
So the immediate issue is not a confirmed collapse in GPU values. It is whether the lenders behind the planned platforms can price a long-term loan when the value of the underlying machines could change sharply before the debt is repaid.
The Bank of England separately warned in July that a growing pile of AI debt could affect wider financing conditions after an adverse shock to AI companies, especially where exposures are hard for financial firms to see in full. For Nvidia’s plan, the collateral question is where that broader concern meets the chip race.
This story draws on original reporting from CNBC.