Nvidia Issues 700 Trillion Won GPU-Backed Loan Securities… Attracts Wall Street Capital for AI Chip Sales

Nvidia has partnered with six major Wall Street financial firms (including Blackstone and BlackRock) to establish a new AI financing structure worth $500 billion (approximately 700 trillion won) to boost AI chip sales. This large-scale funding pipeline aims to supply the enormous capital required for AI data center facility investments.

This financial collaboration is interpreted as a strategy to secure a smooth supply of GPUs, essential for AI infrastructure development, thereby further strengthening Nvidia's market dominance.

The new AI financing structure introduces AI chip-backed securitization, similar to the securitization of home mortgages. A special purpose vehicle (SPV) purchases AI chips (GPUs, memory, etc.) and then leases them to cloud and data center operators.

Securities are then issued based on the chip usage fees (cash flow) generated from this process. The issued securities are sold to institutional investors, such as pension funds, to recycle initial capital, aiming for a virtuous cycle that leads to the purchase of new AI chips.

This structure helps efficiently raise large-scale capital for AI infrastructure investment from the market.

To ensure the stability of this AI financing structure, Nvidia has entered into a credit enhancement agreement to cover losses up to 25% of the total financing amount. This is to prepare for risk situations such as data center operators defaulting on chip usage fees or a decline in the value of the collateralized AI chips.

Nvidia's loss guarantee plays a crucial role in securing investor confidence, mitigating financial risks, and facilitating capital attraction.

This measure is interpreted as Nvidia's commitment to share some of the financial risk to maximize AI chip sales.

With the rapid growth of the AI industry, the burden of massive AI capital expenditures (CAPEX) for the five big tech companies is increasing. Most of the 700 trillion won in net profits earned by these companies is being consumed by AI infrastructure development, causing their free cash flow (FCF), which can be used freely, to sharply decrease to around 22 trillion won.

This mirrors the situation where nascent cloud providers, despite over 50% year-on-year revenue growth, are incurring significant losses. Large big tech companies like Amazon and Microsoft are also experiencing pressure on their cash flow as most of their substantial profits are poured into AI facility investments.

There is also a critical perspective on the new AI financing structure. While it poses no issues when the AI industry grows at its current high speed, critics point out that if growth stagnates, the complex financial structure could collapse, leading to large-scale risks. In particular, warnings are being issued about structural risks similar to the subprime mortgage crisis that triggered the 2008 global financial crisis.

Just as complex mortgage-backed securities (MBS) became distressed and spread the crisis throughout the financial system, concerns are being raised that AI chip-backed securities could also severely impact the financial system if the AI industry slows down or chip values decline.

Some analysis suggests that Nvidia's 25% loss guarantee may be insufficient to cover the entire risk.

Nvidia's GPU financing structure is a strategy to build AI infrastructure in a manner similar to a typical residential rental system. A special purpose vehicle (SPV) directly purchases AI equipment such as GPUs and memory, and then leases them to cloud and data center operators for a monthly usage fee.

This is intended to stimulate GPU demand by enabling cloud providers to utilize AI infrastructure without the burden of massive initial investment.

Wall Street financial firms provide the necessary capital for this structure, and Nvidia aims to secure a stable GPU sales market through it.

BlackRock CEO Larry Fink has experience successfully leading the mortgage-backed securities (MBS) market in the 1980s. As Nvidia's new financial model is suggested to grow in a similar fashion to MBS, the shadow of the 2008 financial crisis looms.

MBS were a 'hit product' during periods of low delinquency rates but caused large-scale defaults when the housing market collapsed. Concerns are arising that Nvidia's AI chip-backed securities could also cause significant repercussions in the financial market if cloud providers default or the collateral value of AI chips sharply declines.

Currently, Nvidia enjoys a monopolistic position in the AI GPU market, but there is a risk that the collateral value of GPUs could plummet if this monopoly structure breaks down. While Nvidia's new product release cycle is a major factor supporting collateral value, if Nvidia's monopoly is shaken by the emergence of innovative new products from competitors, the value of older GPUs could sharply decline.

This could threaten the soundness of the collateralized securities and cause instability in the entire financial structure. Nvidia asserts its technological superiority, but the market competitive environment always contains unpredictable elements.

The technological advancement and successful commercialization of low-cost Chinese AI models could pose a new threat to Nvidia's AI financing structure. If China develops low-cost AI models that offer equivalent performance at one-tenth the price of Nvidia GPU-based AI models, demand for GPUs in AI data centers could sharply decrease.

Such a situation could lead to reduced revenue for cloud and data center operators, triggering defaults on chip usage fees and ultimately causing the default of AI chip-backed securities.

As the scale of the AI chip financing structure grows, observations suggest that Nvidia may demand financial roles, such as credit enhancement, from memory companies as well. Leveraging its strong leadership in the AI semiconductor market, Nvidia could pressure manufacturers of key components like high-bandwidth memory (HBM), which are supplied alongside GPUs.

This could impact the financial health of domestic memory semiconductor companies like Samsung Electronics and SK Hynix. In particular, Samsung Electronics, with its recent shareholder return policies and massive capital investments leading to insufficient free cash flow, could face new financial burdens from such demands from Nvidia.

How memory companies will participate in the AI financing structure will be a point to watch in the future.