The biggest partner program that launched this week was a loan desk

NVIDIA's customers want to buy more compute than they can pay for. So NVIDIA went out and found the lenders.

 

On August 10, NVIDIA announced it is standing up what it calls independent compute financing platforms with six firms: Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR. The stated aim is to "mobilize over $500 billion of third-party capital for the buildout of AI infrastructure over time." The money is pointed at "leading frontier AI labs, enterprises and AI clouds"; the people who want an AI factory and cannot write the cheque today.

 

Picture a car dealership that also arranges your loan - this is a model that's been working for a long time and it allows the dealership to sell a lot more cars. Things get interesting pretty quickly when you ask who eats the loss when the buyer stops paying.

 

These are memorandums of understanding, not signed deals. NVIDIA says so itself: "These partnerships remain subject to execution of the final agreements." The half-trillion is capital to be mobilized, "over time." So, no-one committed it, nobody has drawn it, and there is no date is attached to any of it.

 

There is also a second $500 billion number floating around in NVIDIA's orbit, and the two numbers can be confusing. At GTC in October 2025 Jensen Huang put half a trillion dollars on Blackwell and Rubin orders through the end of 2026, a figure he raised to a trillion through 2027 this past March. That $500 billion is about revenue visibility and this week's $500 billion is somebody else's lending capacity, so we're talking about two entirely different animals.

 

 

Between roughly 1998 and 2001, Lucent, Nortel and Cisco lent money to the telecom carriers buying their equipment. Lucent committed somewhere around $8 billion of it. A lot of that credit went to competitive local exchange carriers with thin revenue and a total dependence on capital markets staying open.

 

Then the capital markets closed. The borrowers could not pay, the equipment orders evaporated, and the loans turned into writedowns. Lucent took billions in bad-debt provisions across 2001 and 2002 and Nortel eventually went bankrupt.

When a vendor funds its own demand, the demand looks stronger than it is, and you cannot tell the difference until the funding stops. NVIDIA has been fielding a version of this criticism for a year now over its investments in AI labs and neoclouds, and the company has consistently rejected it.

 

While I do not think the 2001 comparison is wrong, I think it is missing some key pieces.

 

Lucent and Nortel lent from their own balance sheets and that's the detail that killed them. When the borrowers defaulted, the losses landed on the same income statement that had booked the revenue, and the two halves of the trade netted out to an utter catastrophe and a lot of business school cases (IYKYK).

 

NVIDIA is not doing that. It is routing the lending through Apollo, Blackstone, KKR and others.

 

NVIDIA keeps the equipment revenue and hands the credit exposure to institutions whose entire business is pricing exactly this kind of risk. So, NVIDIA has clearly done its homework. The demand-side problem is identical to 1999 but the risk placement is the opposite and is with companies that are paid to hold it and will price it accordingly. If the frontier labs and AI clouds cannot service the debt, Apollo's investors find out, not NVIDIA's. Whether that is a healthier arrangement for the system or merely a convenient one is a topic for another post.

 

Financing is becoming a tier of NVIDIA's partner program. This has been true in hardware and other industries (think GE) for a long time; Cisco Capital and Dell Financial Services have existed for years. What stood out to me is that NVIDIA did not build an inhouse captive finance arm and instead convened six external ones and made access to capital part of what it offers its ecosystem. That is a major partner program feature, announced as a financial markets story.

 

You have probably already encountered this in cloud marketplaces. When a customer buys your software through their committed cloud spend and draws it down against a commitment they have already made, that is a financing mechanism wearing a procurement costume. The budget is pre-approved and the friction of a new spend decision is gone. It is one of the most underrated reasons marketplace transactions close faster, and most software companies still treat it as a procurement detail rather than the massive sales lever it is.

 

NVIDIA has now run the same play one layer down the stack, at the level of the physical infrastructure.

 

So the question for your 2027 plan is: What is your financing story? Not your pricing, not your packaging. When your buyer wants what you sell and cannot fund it this quarter, what does your ecosystem let you offer them? If the answer is a discount, you are competing on margin against companies that are competing on access to capital.

 

Three things worth asking your team this month:

1. Do we know what portion of our pipeline could be funded through a partner's committed spend rather than new budget, and do our sellers know how to ask?

2. When a deal stalls on budget timing rather than on fit, who in our partner ecosystem has a financing answer, and have we ever used it?

 

3. If our largest platform partner started financing our customers directly, would that make our pipeline stronger or just look stronger?

 

That last one is the Lucent question, and every partner leader in this cycle is going to have to answer it eventually.

 

Governance and standards get the headlines; capital structure decides who is still standing. This week NVIDIA turned access to money into an ecosystem benefit, and it did so in a way that keeps the paper off its own books.

 

If your partner strategy is a list of integrations and a co-sell motion, it is now missing a column.

 

That column is the work Partner1 does. We build marketplace and co-sell strategy for software companies whose growth runs through Microsoft, AWS and Google, including the funding mechanics most teams leave on the table. Let's talk about where your deals stall on budget rather than on fit, and go from there.