Nvidia has struck agreements with six of Wall Street's largest financial firms — Goldman Sachs, BlackRock, Blackstone, Apollo Global Management, Brookfield Asset Management and KKR — to mobilize more than $500 billion in third-party financing for AI infrastructure, the company confirmed this week. CEO Jensen Huang said in a CNBC interview that he approached only those six firms for the commitment, and none turned him down.

How the deal is structured

Nvidia has formalized the arrangements as memorandums of understanding with each firm, with final agreements still to be executed. Rather than Nvidia itself putting up the capital, the coalition is meant to create dedicated pools of financing, at scale and at preferential rates, that hyperscalers, frontier AI labs and enterprise customers can draw on to build data centers and buy Nvidia hardware. The structure treats GPUs less like a one-time capital purchase and more like a financeable, revenue-generating asset — the comparison Huang has drawn to toll roads or power plants, arguing that Nvidia chips can be underwritten similarly because they can be expected to generate a return over their operating life.

Why Nvidia is doing this now

The AI buildout has increasingly run up against a financing bottleneck rather than a chip-supply one: data center projects can cost tens of billions of dollars each, and the hyperscalers and AI labs planning them have been mixing debt, off-balance-sheet vehicles and equity to fund a wave of construction that shows no sign of slowing. By lining up dedicated capital pools ahead of demand, Nvidia is aiming to remove financing friction as a constraint on how fast its customers can keep buying and deploying its chips — effectively underwriting the next leg of AI infrastructure spending itself.

Why it matters

A $500 billion financing commitment is one of the largest infrastructure-financing arrangements tied to the AI buildout to date, and it deepens Nvidia's role from chip supplier to something closer to a financial architect of the industry it sells into. It also ties some of the largest names in traditional finance more directly to AI infrastructure's returns, a dynamic that has drawn scrutiny from economists warning about circular financing arrangements between chipmakers, cloud providers and the AI labs they all do business with. Whether the underlying data center capacity generates enough revenue to justify the financing will be the real test of Huang's "investable asset" framing.