Nvidia 500B AI Financing

The next phase of the AI race will be decided at the credit desk, not in the lab.

A Fortune column by Mehran Gul makes the case with numbers. The five largest US hyperscalers are on track to spend $730 billion on capital expenditure in 2026, and up to a third of it may be financed with debt. If the most profitable companies in history have to borrow to stay in the race, the harder question is how everyone else competes.

Nvidia's answer, assembled in August, is a set of financing platforms with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR that could mobilize over $500 billion for AI infrastructure. The idea is to make compute an asset class that pension funds and insurers will own the way they own power plants and toll roads.

The column is candid about the risk. Power plants last decades. GPUs are short-lived assets, AI demand is unpredictable, and returns are hard to project. But it also argues this is not circular financing. The money comes from third parties, which means the buildout is being wired into the wider financial system.

My read for enterprise leaders: AI vendors with deep, cheap access to capital will set the pricing and the capacity you buy. Financing strength belongs in vendor diligence next to model performance.

Which should weigh more in your AI vendor choices over the next two years: model capability or the balance sheet behind it?