Nvidia bets $500B on GPUs as infrastructure assets

Nvidia has assembled a $500 billion financing pipeline with six of the world's largest asset managers to fund AI infrastructure buildouts, treating its graphics processing units as long-term assets comparable to commercial real estate rather than rapidly depreciating electronics.

CEO Jensen Huang announced agreements this week with BlackRock, Blackstone, Apollo, KKR, Brookfield and Goldman Sachs during a CNBC segment. The initiative aims to provide capital for companies lacking the credit ratings or cash reserves to purchase millions of dollars in AI chips outright.

The infrastructure asset thesis

Huang's pitch to Wall Street rests on a fundamental reframing of how GPUs should be valued. "Nvidia's AI factory platform is really an investable asset, an infrastructure asset," he said. "The reason for that is because it's productive, it's revenue generating, it is fungible, it's used by just about every cloud service provider, it runs every AI model."

The model borrows from traditional asset-backed finance, where lenders can repossess and resell physical collateral if borrowers default. Nvidia argues its CUDA software layer continuously improves hardware performance after deployment, extending the productive lifespan of chips beyond conventional depreciation schedules.

Rental rates for Nvidia's H100 chips have risen from approximately $1.70 per GPU-hour in late 2025 to about $2.35 per GPU-hour in 2026, driven by scarcity as hyperscalers expand capacity.

China represents the biggest threat

Ben Emons, founder of FedWatch Advisors and former portfolio manager at Pimco, identified China as the single largest risk to the financing model. If Chinese manufacturers flood the market with low-cost silicon in a price war, the collateral backing hundreds of billions in private loans could erode faster than debt repayment schedules.

"Depreciation is the one key risk here," Emons said. Nvidia chips "could depreciate faster than expected."

Cutting-edge GPUs used for frontier model training eventually shift to lower-margin inference work after several years, directly impacting resale and collateral values. Emons estimates investors will demand high-yield returns between 11% and 17% depending on their position in the capital structure, treating GPUs as high-depreciation equipment rather than stable real estate.

Borrowers are likely to include non-investment grade firms such as AI startups and neoclouds locked out of traditional debt markets, according to Bank of America Securities. If these higher-risk borrowers fail, fund managers will need to repossess and resell used chips into a potentially declining market.

Why it matters

This financing structure could unlock massive capital for AI infrastructure expansion, but it also exposes institutional investors to technology depreciation risk typically avoided in infrastructure investing. The model's success depends on Nvidia maintaining technological leadership and pricing power against Chinese competition—a geopolitical and market dynamic that could shift rapidly. If GPU values collapse faster than loan terms anticipate, losses could ripple through portfolios at some of the world's largest asset managers.

Near-term outlook favors Nvidia

Immediate threats from China remain limited. Huawei, the leading Chinese AI chip provider, has been on the U.S. Commerce Department's Entity List since 2019. In May, the U.S. government determined Huawei's Ascend AI chips violate export controls, preventing American companies from using them.

Nvidia currently holds upward of 75% market share in U.S. AI chips by most estimates, giving it substantial pricing power in the near term.

The details were first reported by CNBC.