IMPLICATOR
On Monday, August 10, Nvidia announced that it is assembling financing platforms with six of the largest Wall Street firms, aimed at more than $500 billion in outside capital for AI infrastructure. With Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR participating, compute would serve as collateral for debt sold through private offerings and bonds that special-purpose entities could issue before leasing the hardware to Nvidia customers, according to a person familiar with the plans. Those customers could fund purchases of Nvidia systems without drawing on their own balance sheets.
The memorandums of understanding are preliminary and remain "subject to execution of the final agreements." The $500 billion figure is an uncommitted target for capital raised over an unstated period, not Nvidia revenue or a single fund. The announcement gives no committed amount by institution, interest rates, maturities, deployment timetable or first project. It also does not fully describe Nvidia's own exposure.
What Changed
- Nvidia signed preliminary agreements on August 10 with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to build financing platforms targeting more than $500 billion in outside capital for AI infrastructure.
- Compute would serve as the collateral, with debt sold through private offerings and bonds that special-purpose entities could issue before leasing the hardware to Nvidia customers.
- Nvidia may provide residual-value support for up to 25% of a deal, as much as $125 billion against the full target, though its trigger, payment priority and duration are unpublished.
- The announcement gives no committed amount by institution, no interest rates, no maturities, no deployment timetable and no first project.
AI-generated summary, reviewed by an editor. More on our AI guidelines.
Chief Executive Jensen Huang said he approached only the six firms and none declined. Goldman Sachs is the group's only bank and is positioning itself as lead bookrunner on public debt deals.
Huang wrote in a post on X that Nvidia might provide residual-value support for "up to 25% of an opportunity", decided case by case. Applied across the full target, that ceiling would equal as much as $125 billion. Nvidia has not published the support's trigger, payment priority or duration. Its obligation would grow when used Nvidia systems lose value, the same condition that could weaken demand for new systems.
A working model already exists. On March 31, CoreWeave closed an $8.5 billion delayed-draw term loan that it described as the first investment-grade, nonrecourse financing secured by high-performance computing infrastructure and a customer contract. The filed agreement made eligible funding equal to 90% of specified capital spending plus certain transaction expenses and used a six-year useful life to calculate depreciation on the financed GPU infrastructure.
Get Implicator.ai in your inbox
Strategic AI news from San Francisco. No hype, no "AI will change everything" throat clearing. Just what moved, who won, and why it matters. Daily at 6am PST.
No spam. Unsubscribe anytime.
To support his case that the chips hold their value as collateral, Huang pointed to a one-year H100 rental that rose to about $2.35 an hour in March 2026 from roughly $1.70 in October 2025. Rental income is not a resale price. Silicon Data tracked the median large-cloud H100 rental near $9.34 an hour in the second half of 2024 and about $6.26 a year later, showing how rates vary by provider and period.
Know someone who'd find this useful? ✉️ Email it to a friend in one click, or they can subscribe free here.
Amazon put a shorter clock on some of its own equipment. Effective January 1, 2025, it cut the estimated useful life of some servers and networking gear to five years from six, citing faster development in artificial intelligence and machine learning. The change added about $1.4 billion to Amazon's 2025 depreciation and reduced net income by roughly $1 billion, mostly at AWS. The filing did not identify Nvidia chips. Nvidia argues that A100 processors introduced in 2020 still attract multiyear commitments.
Felix Wang, managing director of global technology at Hedgeye Risk Management, described the financing as a price cut delivered through credit. "In effect, they made Nvidia's product cheaper without really cutting GPU prices," Wang said. "But it also makes future demand more sensitive to credit conditions, credit volatility, and raises a lot of questions on what we consider to be real demand."
The first debt deals are expected to reach the market within months of the announcement.
Frequently Asked Questions
What did Nvidia actually announce?
On August 10 Nvidia said it is assembling compute financing platforms with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR, aimed at more than $500 billion of outside capital for AI infrastructure. The memorandums of understanding are preliminary and remain subject to execution of final agreements.
How would the financing work?
Compute would serve as collateral for debt sold through private offerings and bonds that special-purpose entities could issue before leasing the hardware to Nvidia customers, according to a person familiar with the plans. Customers could fund purchases of Nvidia systems without drawing on their own balance sheets.
How much risk does Nvidia take on itself?
Huang wrote in a post on X that Nvidia might provide residual-value support for up to 25% of an opportunity, decided case by case. Applied across the full target, that ceiling would equal as much as $125 billion. Nvidia has not published the support's trigger, payment priority or duration.
Why does GPU depreciation matter to these loans?
The debt rests on what the hardware is worth over time. Amazon cut the estimated useful life of some servers and networking gear to five years from six effective January 1, 2025, which added about $1.4 billion to its 2025 depreciation. Nvidia argues that A100 processors introduced in 2020 still attract multiyear commitments.
Has compute-backed debt been done before?
Yes. On March 31 CoreWeave closed an $8.5 billion delayed-draw term loan it described as the first investment-grade, nonrecourse financing secured by high-performance computing infrastructure and a customer contract. Eligible funding equalled 90% of specified capital spending plus certain transaction expenses.
AI-generated summary, reviewed by an editor. More on our AI guidelines.
Marcus Schuler
Editor-in-Chief and founder of Implicator.ai. Former ARD correspondent and senior broadcast journalist with 10+ years covering tech. Writes daily briefings on policy and market developments. Based in San Francisco. E-mail: editor@implicator.ai
