Summary of Key Points
NVIDIA, in collaboration with six major Wall Street financial giants (Apollo, KKR, BofA Capital, BlackRock, Goldman Sachs, and HSBC), has launched a $500 billion financing initiative that allows technology companies to lease NVIDIA’s AI chips without having to purchase them outright. The Wall Street firms are betting that AI chips will not depreciate as quickly as traditional electronic devices due to the current imbalance between supply and demand and strong market demand. However, there are also significant risks: if the AI boom fades or technology evolves too rapidly, the value of the chips could plummet. To alleviate investors' concerns, NVIDIA has provided a “residual value guarantee” (up to $125 billion) to ensure the success of the initiative.
The $500 Billion Financing Initiative: Making AI Chips Affordable for Companies
The logic behind this initiative is straightforward:
1. Where does the money come from? The Wall Street firms issue bonds to pension funds, insurance companies, and other institutions to fund the establishment of a “chip leasing platform.”
2. How do companies use it? Technology companies that need to build AI data centers but cannot afford the chips can rent NVIDIA’s GPUs (such as the A100 and H100) from this platform and pay monthly rent.
3. NVIDIA’s role? NVIDIA provides the chips and also acts as a guarantor for the initiative (more details below).
For example, a technology company that needs 1,000 A100 chips for training a large model would typically spend hundreds of millions in one go. With this initiative, it can rent the chips and start its project immediately by paying a monthly fee.
Wall Street’s Confidence: Why Do They Think AI Chips Won’t Depreciate Quickly?
Unlike traditional computer chips, which lose value within a few years, there are three reasons why Wall Street believes AI chips will remain valuable:
1. Severe supply-demand imbalance: There is a global shortage of AI chips, and NVIDIA’s production capacity cannot keep up, allowing chip prices to remain high for a long time.
2. Continued demand for older chips: Chips like the A100, released in 2020, can still be rented at nearly their original price until 2029. Since AI tasks (such as those performed by ChatGPT) do not require the latest computing power, older chips are sufficient.
3. Stable long-term demand: As more AI models are developed, the demand for computing power in data centers will only increase, extending the lifespan of these chips.
Wall Street analysts compare this to the securitization of mortgage loans in the 1970s, suggesting that leasing AI chips represents a new opportunity for profit.
Potential Risks: The “Time Bombs” Lurking in AI Chip Value Preservation
The main risks include:
1. Sudden demand decline: If the AI boom subsides (for example, if companies stop building data centers aggressively), supply will exceed demand, leading to a sharp drop in chip prices.
2. Rapid technological advancement: If NVIDIA releases a more powerful chip than the H100, older models will become worthless quickly.
3. Difficulties in selling collateral: Chips have a limited secondary market compared to cars or airplanes, making it challenging for lenders to sell them if companies default on their leases.
Analysts joke that the “shelf life” of AI chips is similar to that of fresh lettuce, indicating a rapid depreciation.
NVIDIA’s “Safety Net”: How Does the Residual Value Guarantee Assure Investors?
To reassure investors, NVIDIA has introduced the following guarantee:
- If the value of the chips falls below a predetermined threshold (for example, if the chip is only worth 30% of its original price after three years, with a preset minimum of 40%), NVIDIA will make up the difference.
- NVIDIA will cover a maximum of 25% of the total cost of the initiative, which amounts to $125 billion.
- This guarantee is not initially considered part of NVIDIA’s debt, but if a leasing company defaults, NVIDIA will have to pay the difference.
This mechanism essentially provides investors with insurance, ensuring that even if chip prices drop, NVIDIA bears part of the loss and Wall Street does not bear the entire risk.
A Win-Win Situation for All Parties
The initiative benefits all involved:
1. NVIDIA: It avoids upfront costs associated with selling chips, shifts the financing risks to Wall Street firms, and allows it to quickly expand its market share.
2. Wall Street firms: They find a new investment opportunity in AI chip leasing, generating rent and bond interest, as well as the potential to sell these assets to other institutions.
3. Technology companies: They can reduce their financial pressure and accelerate their AI development without having to purchase chips outright.
4. Institutional investors: Organizations like pension funds gain a stable investment option with higher returns than traditional bank deposits.
However, all of this depends on the assumption that AI chip values will remain stable. If this bet fails, everyone could suffer financial consequences.
In Conclusion
This is a major gamble between Wall Street and NVIDIA, betting that AI chips can break the pattern of rapid depreciation seen in electronic devices and that the AI boom will continue for enough time to be profitable. If successful, all parties stand to profit; if not, it could lead to financial risks. For now, however, everyone seems to believe the gamble is worth taking.