3 weeks ago
Nvidia's $500 Billion Bet That AI Chips Are Bankable Assets
Nvidia makes special computer parts called chips that help smart computer programs, called AI, learn and work.
Jensen Huang leads Nvidia, which became the most valuable company in the world.
Nvidia wants big money companies on Wall Street to lend up to $500 billion to build giant computer buildings called data centers.
These buildings would be paid for over time, using the chips as a kind of promise, called collateral.
Nvidia says the chips earn money, just like a building that people rent, so they are good assets.
But some investors worry the chips become old and lose value fast, like a phone that is no longer the newest.
China is also building its own cheaper chips, which could make Nvidia chips worth less.
Nvidia points out that its chips are still in high demand and that its software can keep older chips useful.
The big question is how long Nvidia chips will keep making money.
Nvidia announced agreements with BlackRock, Blackstone, Apollo, KKR, Brookfield and Goldman Sachs to build a financing pipeline of up to $500 billion for data centers and GPU clusters.
Jensen Huang argues Nvidia's AI systems are investable infrastructure assets because they are productive, revenue-generating and used by cloud providers.
Experts warn depreciation is the key risk: newer GPUs could cause older chips to lose value as collateral faster than expected.
China's push to develop cheaper domestic AI chips could pressure global prices; Nvidia still holds more than 75% of the U.S. AI chip market.
H100 rental prices rose from about $1.70 per GPU-hour in late 2025 to about $2.35 per GPU-hour this year, and CUDA software may extend chip usefulness.
- Who
- Nvidia CEO Jensen Huang, in partnership with six asset managers: BlackRock, Blackstone, Apollo, KKR, Brookfield and Goldman Sachs.
- What
- A plan to create up to $500 billion in financing so companies can build data centers and GPU clusters with AI chips treated as collateral assets.
- Where
- The U.S. AI chip market and the global market for data centers and GPU clusters, with China a key competitive factor.
- When
- This week, when Nvidia announced the agreements with the six asset managers.
- Why
- To let companies without enough cash or strong credit ratings buy AI computing power over time and to convince Wall Street that Nvidia's chips are long-term investable assets.
Nvidia's Case
Investor Concerns
Chip depreciation
Nvidia's Case
Nvidia says its AI factory platform is productive, revenue-generating and fungible, and that CUDA software improvements keep existing hardware useful even after newer chips arrive.
Investor Concerns
Ben Emons calls depreciation the key risk, saying chips could lose value faster than expected; he estimates investors could demand yields between 11% and 17%.
China's challenge
Nvidia's Case
Nvidia dominates the U.S. AI chip market with more than 75% share, and Huawei's Ascend chips are barred from U.S. companies due to export controls.
Investor Concerns
If Chinese companies produce large amounts of cheaper AI hardware, global prices and the value of GPUs used as collateral could fall sharply.
Borrower credit quality
Nvidia's Case
Nvidia's chips are used by nearly every cloud service provider and run every AI model, supporting their revenue-generating value as assets.
Investor Concerns
A Bank of America Securities note says borrowers could be AI startups and neocloud companies without investment-grade credit ratings, creating risk on both the asset and borrower side.
Key facts
- Financing pipeline
- Up to $500 billion for data centers and GPU clusters
- Participating asset managers
- BlackRock, Blackstone, Apollo, KKR, Brookfield, Goldman Sachs
- Nvidia U.S. AI chip market share
- More than 75%
- H100 rental price (late 2025)
- About $1.70 per GPU-hour
- H100 rental price (this year)
- About $2.35 per GPU-hour
- Expected investor yields
- 11% to 17%, per Ben Emons of FedWatch Advisors
- Key risks cited
- GPU depreciation, cheaper Chinese chips, borrowers without investment-grade credit
- Huawei status
- On the U.S. Commerce Department Entity List since 2019; Ascend chips found to violate U.S. export controls in May
Quotes
Jensen Huang
CEO of Nvidia
“"Nvidia’s AI factory platform is really an investable asset, an infrastructure asset."”
financialexpress.com
Ben Emons
Founder of FedWatch Advisors, former IndyMac loan structurer
“"Depreciation is the one key risk here."”
financialexpress.com
Sources
Related news
Nvidia, Wall Street giants raise $500 billion for AI infrastructure
Jensen Huang Defends Nvidia Against OpenAI Chip Competition
Nvidia Partners With Wall Street Giants to Raise $500 Billion
Australian AI startup Firmus raises $2 billion backed by Nvidia
Nvidia backs OpenAI's Ohio data centre with $105 billion guarantee
Nvidia to Lease Land and Power for AI Factories
Bank of America Vice President Killed in Times Square Stabbing
Times Square Stabbing Leaves One Dead; Suspect Had Ominous Posts
Social Media Posts Preceded Fatal Times Square Stabbing Rampage
Times Square Stabbing Kills Bank Executive as Police Shoot Suspect
Beyond Nvidia, power and cooling firms benefit from data-centre boom
