Nvidia already sells the shovels in the AI gold rush. Now the chipmaker wants to help pay for the mines.

On August 10, 2026, the company signed memorandums of understanding with six of the largest names in global finance—Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR—to establish independent compute financing platforms designed to raise more than $500 billion in third-party capital for AI infrastructure.

What Nvidia actually signed

The six partners are not chipmakers. They rank among the biggest institutions in asset management, private equity and investment banking, and together with Nvidia they plan to build platforms that pool outside capital and direct it toward AI buildouts.

The headline commitment is Nvidia's potential backstop. Chief Executive Jensen Huang indicated the company could backstop up to $125 billion, or 25% of the projected deals. In plain terms, that would put Nvidia in the role of guarantor on up to a quarter of the projected financing.

Huang framed the effort as the birth of a new "investable asset class" built around AI factory infrastructure. His logic: compute is no longer just a cost center—it is a machine that produces revenue.

Goldman Sachs' chief executive voiced confidence in Nvidia's leadership through the partnership. He pointed to the potential for "credit backed by Nvidia compute," a phrase that captures the whole pitch in five words.

Wall Street's first reaction was muted. Nvidia shares slipped about 3% on the news, though the stock has posted a significant gain so far this year.

A year of writing very large checks

Nvidia has spent 2026 putting serious money to work across the AI ecosystem. The company has struck more than $40 billion in AI equity and financing deals so far this year, and by late July 2026 its total equity investments had reached approximately $99 billion. For a sense of scale, the $500 billion these platforms aim to raise from outside investors is roughly five times Nvidia's total equity investments to date.

The new platforms formalize that shift. Instead of one-off transactions, Nvidia is helping build repeatable machinery for funneling institutional capital into compute.

What this means for AI infrastructure and its backers

For Nvidia, the upside is straightforward: more financing for AI infrastructure means more buyers who can afford its hardware, and the backstop could lower the barrier for companies that want to build on its technology.

For the financial partners, the deal offers a structured way into AI returns. Rather than betting on a single startup or model developer, they can back the physical capacity that the entire sector depends on.

The arrangement does carry a structural wrinkle. When a supplier helps underwrite the purchase of its own products, some of the demand it reports is partly supported by its own balance sheet. That setup works smoothly while the market is growing, but it becomes more complicated if demand for compute cools, because the company would then be exposed both as a seller and as a guarantor.

The potential $125 billion backstop is the number to watch. It is described as a ceiling Nvidia could reach, not a sum it has already committed.

It is also worth remembering what was actually signed. Memorandums of understanding are agreements in principle, not finished contracts—so the $500 billion target is a fundraising goal rather than money in the bank. How quickly the memorandums convert into definitive contracts will be the first signal of whether that goal moves from paper to capital.