The surge in bond issuance by major cloud providers, combined with the collapse of hedge fund Situational Awareness, has brought the underlying risks of rising leverage in the AI boom under intense market scrutiny. Nvidia has unveiled a new plan to advance AI financing, aiming to turn infrastructure into a class of investable assets. Some strategists argue that, compared to the steady climb in leverage, the market should be more cautious about the risk of overly optimistic earnings expectations.

Nvidia, in partnership with Wall Street institutions, is planning to channel over $500 billion in third-party capital into AI infrastructure construction. The increasingly complex financing structure supporting this industry expansion is now facing closer examination.

Leading hyperscale cloud providers and their financiers are using a variety of methods, including bond markets, joint ventures, and leasing arrangements, to fund an unprecedented infrastructure expansion. At the same time, leverage tools are amplifying the market's exposure to the AI sector. Investors, including hedge funds, are using prime brokerage financing and derivatives to bet on the AI trend to boost returns.

The collapse of AI-focused hedge fund Situational Awareness, which suffered massive losses on leveraged long positions, has heightened concerns about several issues: the total scale of borrowing in the market, the distribution of capital, the level of transparency, and the speed of impact from a concentrated unwinding of leverage. Investors are also debating whether future industry revenue can justify the enormous current capital expenditure.

How Large is the Scale of AI Infrastructure Investment?

Nvidia is partnering with Apollo, Blackstone, BlackRock, Brookfield, KKR, and Goldman Sachs to build an AI infrastructure investment platform. The plan includes a private asset-like structure and asset-based financing models. Nvidia CEO Jensen Huang stated on Monday that Nvidia's chips have now become a type of "investable infrastructure asset."

Some tech giants are raising funds for AI data centers through joint ventures and leasing vehicles. The related debt does not appear on the balance sheet until the lease agreements are officially in effect.

Goldman Sachs analysts estimate that the total leasing commitments of major hyperscale cloud providers for data centers, R&D campuses, office buildings, and equipment amount to $1.5 trillion, up from just $200 billion five years ago. Of this, about $1 trillion is in "uncommenced" lease obligations, which are not yet reflected on financial statements but will represent fixed expenditures in the future.

In a report dated August 6, Goldman Sachs stated, "When the related liabilities are formally booked and contract payments become due, the actual level of corporate leverage and future liquidity needs will be underestimated."

The surge in bond issuance, combined with a large amount of hidden leveraged transactions, has led the market to focus on a key question: whether the long-term returns from AI infrastructure can match the massive investment.

Lofty Karui, a multi-asset credit strategist at PIMCO, stated that after adjusting for inflation, the current AI capital expenditure cycle could be the largest wave of investment since the railway construction boom of the 19th century.

In an August 11 PIMCO market commentary, he said that the ultimate scale of the infrastructure expansion remains "highly uncertain." Market consensus predicts that capital expenditure for hyperscale cloud providers alone will exceed $1 trillion annually starting in 2027, with "no clear signs of a slowdown yet."

Karui noted that given the massive scale of cloud providers' financing, they are no longer limited to issuing bonds in US dollars and are now beginning to tap the Euro, British Pound, Japanese Yen, Swiss Franc, and Canadian Dollar bond markets. He also mentioned that the credit spreads on Euro-denominated bonds issued by Amazon and Alphabet have performed relatively better than their US dollar-denominated bonds, which might suggest weakening demand in the US dollar bond market. He warned that a large number of AI-related bonds are being issued in the US market, and some large issuers highly tied to the AI industry are showing signs of weakness, which could push credit spreads higher.

Could AI Leverage Become a Larger Market Risk?

The collapse of the Situational Awareness fund has proven that crowded, leveraged AI trades are highly vulnerable to significant losses during sharp sector volatility.

The fund had a highly concentrated portfolio, heavily weighted towards positions like SK Hynix and CoreWeave. A previous tech stock correction significantly reduced its assets, shrinking from $45 billion to about $10 billion, making it unable to meet multiple margin calls.

Ken Griffin's larger multi-strategy hedge fund, Citadel, subsequently acquired the liquid holdings of Situational Awareness at a discount. Since then, the share prices of SK Hynix and CoreWeave have partially rebounded.

JPMorgan Chase CEO Jamie Dimon recently told reporter Leslie Picker that current margin debt levels are "quite high," which could amplify market volatility risks.

Sahir Mahtani, Director of the Investment Research Institute at Ninety One, told that, compared to leverage issues, overly optimistic earnings expectations are a more pressing risk.

Mahtani stated that the market's expectation of "sustained high earnings growth for several years" is the core risk that AI trades bring to the market. "It is fundamentally an expectation problem, not a leverage problem."

He pointed out that equity concentration in stock markets dominated by tech stocks is at historic highs, especially in the US market. While this is not financial leverage, a highly concentrated portfolio structure can produce a magnifying effect similar to leverage when heavyweight stocks fall.

"The major stock indices are extremely concentrated, and the market generally believes there is nothing to break the uptrend," he warned. He also cautioned that if companies shift from stock buybacks to issuing new shares, a key support for stock prices would disappear, especially with AI sector valuations already under pressure.

Mahtani believes the Situational Awareness crisis stemmed from poor risk management, and its overall spillover effects are largely manageable. "The fund's bull market coincided with the concentrated unwinding of a large number of leveraged ETFs in East Asia. Many of these products, especially single-stock leveraged ETFs, were only listed in the first half of this year. From this perspective, it is a relatively isolated incident with limited impact."

A spokesperson for the global hedge fund industry association AIMA said that leverage is a core tool for hedge funds to enhance returns and provide market liquidity. "The key is not whether hedge funds use leverage, but whether its use poses a material threat to financial stability. Current evidence does not support viewing hedge fund leverage as an inherent systemic risk."

The spokesperson cited two past leverage-related events: the 2021 collapse of Archegos Capital Management and the 2022 UK gilt crisis involving Liability-Driven Investment (LDI) strategies by pension funds. These involved completely different trading structures and investor groups. "Market events with fundamentally different characteristics should not be conflated. Archegos was a family office, not a hedge fund; the 2022 UK gilt turmoil was driven by pension funds using LDI leverage strategies. There is currently no basis to believe that the Situational Awareness event will directly lead to regulatory revisions of hedge fund leverage rules."