Key facts
- Nvidia's $500 billion financing plan relies on chip-backed loans with limited guarantees.
- Wall Street lenders are seeking stronger guarantees and shorter depreciation schedules for AI chips.
- Some lenders question the long-term collateral value of AI chips, preferring revenue streams from investment-grade customers.
- Nvidia claims its AI compute assets have a useful life of up to a decade.
- Recent AI financing deals have included stronger guarantees and relied on revenue from technology firms.
- Nvidia's financing partners include Blackstone, Apollo, and KKR.
Nvidia's ambitious plan to finance the artificial intelligence boom by using its advanced chips as collateral for loans is facing scrutiny from Wall Street, with lenders demanding more assurances than the company initially outlined. Banks and investors are questioning the long-term value of these chips as collateral, preferring shorter depreciation schedules and stronger guarantees.
Sources told Reuters that some lenders want assurances that the chips can generate revenue for a decade, as Nvidia claims, while others are more conservative, typically underwriting GPUs over a 3-4 year depreciation schedule. This divergence in views could lead to future financing challenges for AI companies seeking capital. Some bankers suggest Nvidia may need to offer guarantees on all its deals or secure them with revenue streams from investment-grade customers.
Nvidia, in a statement, defended its plan, stating that its AI compute is a "productive, durable and fungible asset that can support long-term financing." The company added that financing structures will vary as the market develops and that its partners independently assess each opportunity. Nvidia's CEO, Jensen Huang, has stated his goal is to make Nvidia's compute an "investable infrastructure asset."
Despite these concerns, there remains high demand to finance AI infrastructure. Deals in the pipeline are expected to include stronger guarantees and contracts, with some structures potentially offering lenders direct guarantees. Past chip-backed loans, such as one for CoreWeave, have relied on revenue from established technology firms like Meta for creditworthiness. S&P Global Ratings noted that while GPUs have proven durable for over five years, they maintain a conservative view on their residual value.