- Nvidia struck agreements with six of the largest Wall Street firms to line up $500 billion in financing for the AI buildout.
- The initiative seeks to make AI compute infrastructure akin to commercial real estate, toll roads, and other assets that can be financed against.
- Analysts warn rapid hardware depreciation, exacerbated if China floods the market with low-cost compute, could crash the collateral values backing these loans.
- Nvidia argues its CUDA software layer preserves long-term chip value, allowing older chips to stay productive and generate yield longer than traditional accounting models predict.
- Rental rates for Nvidia's H100 chips have risen from $1.70 to $2.35 per GPU-hour this year.
- High default risks could push investor yield demands to between 11% and 17% in one estimate, as Nvidia's financing model faces scrutiny.
- Chinese production could push hardware prices into a freefall, eroding the collateral backing hundreds of billions in private loans.
- Nvidia remains by far the leading supplier of AI chips in the U.S., with upwards of 75% market share by most estimates.
Nvidia has entered agreements with six leading Wall Street firms, including Apollo Global Management and BlackRock, to raise $500 billion for AI infrastructure, aiming to establish AI computing as a new asset class. This initiative could unlock significant capital for data centers and GPU clusters, particularly for companies lacking the credit to purchase expensive hardware outright.148
However, analysts express concerns about the potential for rapid hardware depreciation, especially if China floods the market with low-cost silicon, which could undermine the collateral values of these loans. Ben Emons, founder of FedWatch Advisors, noted, "Depreciation is the one key risk here," emphasizing that Nvidia chips "could depreciate faster than expected."3
Nvidia's CEO, Jensen Huang, stated, "Nvidia's AI factory platform is really an investable asset, an infrastructure asset," highlighting the revenue-generating potential of their technology. The initiative aims to make AI compute infrastructure comparable to commercial real estate, allowing for financing similar to mortgage-backed securities.2
Despite the risks, the financing model could shift the funding burden from corporate balance sheets to institutional investors, with BlackRock's CEO, Larry Fink, suggesting this could herald a new era in financial engineering. As AI infrastructure investment accelerates, the immediate benefits may favor the financial firms involved, but broader financial companies could also gain if Nvidia's model proves successful.
“Ben Emons of FedWatch Advisors warns that Chinese production could trigger a price war, eroding collateral faster than debt terms. Meanwhile, H100 rental rates have risen from $1.70 to $2.35 per GPU-hour, and Nvidia's CUDA software is cited as a key to preserving chip value.”
CNBC
Yahoo Finance Singapore
