JP Morgan downgrades Oracle rating - A glimpse into the bond market's view on "AI infrastructure financing"

A massive debt financing of up to $38 billion is preparing to flood the market, providing ammunition for Oracle's data center projects.

This financing, led by JP Morgan and Mitsubishi UFJ Financial Group, is not only the largest financing deal in the AI infrastructure sector to date but also marks a shift in this competition from an internal cash game among tech giants to a capital gamble driven by leverage.

As the investment banking division works diligently to arrange underwriting for this deal, JP Morgan's bond research team has released a "brake-pumping" research report, downgrading Oracle's credit rating from "overweight" to "neutral."

The report points directly to three major credit risks behind its aggressive expansion:

• First, the $35 billion capital expenditure conflicts with its vague financing strategy and "unfriendly" credit history;
• Second, the $300 billion order from OpenAI poses customer concentration risk, and its cloud business's mere 14% "thin profit" is insufficient to support high leverage;
• Third, there may be "capital internal circulation" or amplified demand bubbles among AI giants.

Why is Oracle's AI ambition met with cold water from the credit market?

At the beginning of the report, JP Morgan's bond team straightforwardly stated:

“ We downgraded Oracle's rating considering the company's enormous capital needs and the limited visibility of its financing strategy.”

Wall Street Journal mentioned that this tech giant, known for its high-profit database software, is betting its future on becoming a heavy-asset AI cloud infrastructure service provider. The direct cost of this transformation is the explosive growth of capital expenditure (Capex). The report clearly states that management has raised the capital expenditure guidance for FY26 to an astonishing $35 billion, and this is just the tip of the iceberg.

For the credit market, massive investments themselves are not frightening; what is concerning is the vague financial discipline behind them and the "checkered" credit history. The JP Morgan bond team wrote in the report that Oracle has historically been “ one of the least credit-friendly issuers in the investment-grade technology, media, and telecommunications sectors.”

For example, the company has long financed large-scale stock buybacks and aggressive acquisitions (such as the $28 billion acquisition of Cerner) through debt, leading to a continuous rise in its leverage levels. Notably, before the recent management reshuffle, Oracle had not appointed a dedicated Chief Financial Officer (CFO) for a decade, which is extremely rare among large tech companies and indirectly confirms its loose financial governance.

Furthermore, the JP Morgan bond team analyzed that this historical performance has made Oracle's current AI expansion story a " show me story" that needs to be strictly validated in the eyes of credit investors.

In other words, while the stock market can cheer for future contracts worth hundreds of billions, the bond market needs to see tangible cash flows that can cover the massive principal and interest. Until Oracle proves through action that its AI strategy can translate into sustainable profits and a healthy balance sheet, the caution and skepticism of the credit market are likely to persist for a long time

The B-side of Huge Contracts: Is OpenAI "Honey" or "Arsenic"?

The catalyst that ignited the frenzy in Oracle's stock price is undoubtedly the five-year AI computing power contract signed with OpenAI, valued at up to $300 billion. This exorbitant order has given the market hope that Oracle can challenge Amazon AWS, Microsoft Azure, and Google Cloud in the AI infrastructure sector.

However, within the analytical framework of JP Morgan's bond team, this seemingly sweet contract carries a highly risky B-side.

First and foremost is the “ customer concentration risk.”

The report points out that deeply binding the company's short-term growth and asset utilization to a single star customer inherently carries significant risk. Moreover, this customer—OpenAI—despite leading the technology trend, still has an unclear business model and profit path. Relying on a partner with an uncertain commercial outlook for hundreds of billions of dollars in future revenue is akin to a high-risk bet for bond investors seeking stable returns.

Secondly, and more critically, is the deep-seated questioning of its profitability.

The report cites an in-depth article from the third-party media outlet The Information, which referenced internal data indicating that Oracle's cloud infrastructure (OCI) business may have a gross margin that is “ razor-thin.”

Data shows that in the three months ending August 31, Oracle generated approximately $900 million in revenue from server rentals powered by NVIDIA chips, but the gross profit was only $125 million, resulting in a gross margin of just 14%.

After accounting for labor, electricity, and equipment depreciation, the average profit margin is only around 16%—a stark contrast to Oracle's traditional software business, which has a gross margin of about 70%, and “other unspecified depreciation” could further reduce the gross margin by 7 percentage points.

Although Oracle's management later rebutted during the investor day, claiming that the gross margin for large AI infrastructure projects could reach 30-40%. This significant data discrepancy exposes the core uncertainty of its AI business profit model—whether Oracle can achieve its promised profit levels amid soaring GPU costs, rising energy prices, and an intensifying price war in the cloud market is a key variable determining whether its $100 billion debt can be safely repaid.

Circular Counterparties: Systemic Risk in the Era of Leverage

In the broader context of the AI infrastructure sector shifting from “cash is king” to a “debt competition,” JP Morgan's bond team has revealed a deeper, potentially overlooked systemic risk—“ Circular Counterparties.”

The team wrote in the report that the current AI infrastructure construction seems to be taking place within a closed ecosystem composed of a few giants engaged in “ capital internal circulation.” The chain roughly goes as follows: NVIDIA provides GPUs to cloud service provider CoreWeave and may offer credit support; CoreWeave then provides computing power to OpenAI; OpenAI's massive model training expenses are underwritten by Microsoft and Oracle's cloud services.

Capital and business circulate within this small circle. Although each party earns real revenue from third-party clients, this high degree of interdependence blurs the lines between strategic alliances and financial engineering.

The risk of this model is that it may lead to demand being "artificially amplified," where enormous contracts between participants in a self-reinforcing cycle may obscure the real demand from the broader real economy, thus giving rise to bubbles.

This is precisely what credit rating agencies are concerned about. The report points out that in addition to the downgrade by JP Morgan's bond team, the two major international rating agencies S&P and Moody's have also adjusted Oracle's rating outlook to "negative" during the summer.

Among them, S&P has loosened the threshold for Oracle's leverage downgrade from 3.5 times to 4.0 times, acknowledging that AI investments need to "look long-term," but this "patience cannot last indefinitely." Moody's expects Oracle's debt growth to outpace EBITDA growth, resulting in its leverage ratio remaining above 4.0 times in FY25.

These rating alerts, along with the previously mentioned trillion-dollar funding gap, collectively paint a clear picture: the future of AI may be bright, but the road to that future is increasingly paved with debt