Jensen Huang plays the "Wall Street card" to address doubts about circular financing, but default risks remain amid the massive AI capital expenditure "panic race"

According to Zhitong Finance APP, Nvidia (NVDA.US), the global AI computing powerhouse, has recently been frequently acting as a “financial driver” in a series of mega data center financing cases, including the $250 billion data center project with OpenAI in Ohio. However, instead of cheers, the market is growing anxious about the risks of “circular financing”: the chip giant is lending money to its customers to buy its own chips, while saddling itself with the potential future bad debts. In just three weeks, Nvidia’s credit risk indicators have nearly doubled.

On Monday, Nvidia CEO Jensen Huang swiftly played the “Wall Street card”; his core strategy is: bring in external funds, have professional institutions conduct due diligence on transactions, and secure “endorsement” from Wall Street. Nvidia mentioned an alliance of six major investment institutions, including BlackRock and Goldman Sachs, which are collectively raising over $500 billion to support AI infrastructure construction. This group will independently assess each transaction and decide on the scale of their participation, while Nvidia’s contribution will be relatively limited and only involved in some deals.

The news provided temporary relief to the strained credit market. On Tuesday, the cost to protect Nvidia's debt from default risks declined, company bonds rose, and the risk premium against U.S. Treasuries has returned to last week’s levels.

Brett Kozlowski, portfolio manager at GW&K Investment Management, said that the commitments from several of Wall Street’s biggest institutions “are a positive development, removing some of the uncertainty around infrastructure buildouts and customer future spending.”

Credit Derivatives Signal First: Where Does the Market’s Worry Come From?

Nvidia plays a crucial role in the global AI race. Its high-performance chips were originally designed for graphics processing and are capable of executing multiple tasks concurrently, making its latest generations particularly suited for data centers. Strong chip demand has driven Nvidia to become the world’s most valuable public company, with a market cap exceeding $5.2 trillion. But investors are increasingly concerned about one issue: whether the company's customers are becoming overly reliant on Nvidia’s financial support to pay for its increasingly expensive chips and data center costs in recent years.

Late July media reports indicated that Nvidia was in talks with OpenAI to provide up to $250 billion in financing support, helping OpenAI lease computing power from a data center under development in Ohio by a SoftBank Group subsidiary. This would be one of the largest financing transactions between the chipmaker and its clients. Sources also revealed that Nvidia was discussing an additional $350 billion in financing for OpenAI to purchase its chips for this project.

Additionally, Nvidia announced a collaboration with SK Group to build data centers exceeding 2 GW in Korea—a deal that is part of more than $500 billion in cooperation with the Korean conglomerate. Jensen Huang later clarified that the amount mainly refers to Nvidia’s anticipated future memory chip purchases from SK Hynix. Nvidia stated that the newly raised $500 billion in external funds is unrelated to the SK deal.

Asset managers are concerned the company is engaging in a form of “circular financing”: by lending to data center customers so they can purchase Nvidia chips, boosting current sales. But if AI infrastructure fails to deliver future returns, this may bring potential losses.

These concerns are clearly reflected in the credit derivatives market. Here, investors can purchase “insurance” to receive payouts if a company defaults on its debt. When asset managers become more concerned about a company's risk of default, the cost to buy such protection increases.

In late July, the cost to hedge Nvidia’s five-year debt climbed to $82,000 per $10 million notional per year, whereas for most of the previous year, this cost lingered at half that level. By Tuesday, this had fallen to about $73,000, or 73 basis points, down about 4 basis points on the day.

“Wall Street Dream Team” Enters: How Do External Funds Dissipate the “Circular Financing” Concern?

On Monday, Nvidia announced it had signed memoranda of understanding with six financial giants—Apollo Global Management, BlackRock, Blackstone, Brookfield Asset Management, Goldman Sachs, and KKR—aiming to mobilize over $500 billion in third-party capital over the long term for AI infrastructure construction. These institutions will independently assess projects and decide whether to provide funding. Nvidia said it is building a “matchmaking platform” to connect independent suppliers of capital with those in need, with its own role limited to providing platform support.

According to its statement, the ultimate outcome will be “a large-scale, competitively priced dedicated capital pool” for Nvidia customers.

The company will support some projects through guarantees, covering up to 25% of the amount, and adopt something called a “residual value mechanism” to help limit losses if projects fail. Nvidia gave few details but said its chips are widely used by many customers, which should help minimize potential losses.

For example, if a project encounters difficulties, a residual value guarantee may mean Nvidia, after taking measures to recover value (such as finding other companies to rent capacity or selling chips), will still provide financial support.

Lingering Concerns: A “Panic Race” in Massive Capital Expenditure—Who Ultimately Bears the Losses?

Nevertheless, Alberto Gallo, Chief Investment Officer and Co-founder of Andromeda Capital, said risks remain concerning AI infrastructure construction and the overall credit environment.

Gallo pointed out that the credit market is increasingly becoming a bet on U.S. computing power demand and value, while investors in the market may not be getting returns commensurate with the risk taken. He noted that trillions of dollars are pouring into data centers and other infrastructure construction at an astonishing pace, where there will inevitably be winners and losers—thus, defaults are likely to occur.

Gallo said: “This is essentially a panic capital expenditure spree. Who bears the loss? It’s bondholders, life insurance companies, and policyholders.”

For now, Nvidia’s own strong profitability serves as ample cushion against credit pressures. In the financial year ended January 25, its free cash flow was close to $100 billion. Now, investors are drawing a new conclusion: since external institutions are bearing most of the risk, Nvidia’s likelihood of being forced to absorb massive losses has decreased.

“Previously, no one knew what that potential $500 billion in financing actually meant,” said Sal Naro, Chief Investment Officer at Coherence Credit Strategies. “Now people realize that everyone is involved, and Nvidia’s own risk exposure is not as serious as initially feared by investors.”