Oracle is the canary in the coal mine for Big Tech's debt-fueled AI spending spree

By Christine Ji

Look at Oracle if you want to assess the sustainability of the AI investment boom: Its credit default swaps are emerging as a way to 'hedge the entire debt cycle within AI'

For a few brief weeks in September and October, Oracle was on top of the artificial-intelligence revolution.

The storied business-software and database company had a burgeoning $300 billion relationship with OpenAI that would more than double its revenue in just three years, according to management's new projections. A huge stock surge put Oracle on the doorstep of a $1 trillion valuation. Founder Larry Ellison reclaimed the title of world's richest man, and laid out a vision that put Oracle (ORCL) at the forefront of AI innovation - including the key inferencing stages where trained models use new data to make predictions.

"Oracle is aggressively pursuing the inferencing market, as well as the AI-training market," Ellison said on the company's September earnings call with investors. "We think we are in a pretty good position to be a winner in the inferencing market, because Oracle is by far the world's largest custodian of high-value private enterprise data."

Oracle's accomplishments were the fruit of an almost decades-long investment into its cloud-computing business. Then it all came crashing down.

Shares of Oracle have given up their gains and then some, falling 36% from their September peak. Its credit default swaps, or the price of insurance on its bonds, have widened sharply in recent weeks, an indication that the market perceives the company is on increasingly shaky financial footing. And traders have upped their bets against Oracle's shares: While only 1.5% of its total float, short interest on the stock has climbed from 15.7 million shares at the end of May to 24.4 million shares currently, according to FactSet.

The problem is that while the company wants to compete with the lions of cloud computing, getting there will be expensive, and Oracle doesn't have the cash flow to fund its dreams. Instead, it has turned to the debt markets and leveraged up its balance sheet. Wall Street has become increasingly worried about the debt that Oracle is taking on to become a major partner to OpenAI, which itself is burning cash in pursuit of AI profits that have so far been elusive.

The AI arms race will likely cause the tech "hyperscalers" - like Meta Platforms (META), Amazon.com (AMZN), Google parent Alphabet (GOOGL) (GOOG), Microsoft (MSFT) and Oracle - to spend $2.9 trillion in capital expenditures to build out the technology, Morgan Stanley estimated in a July report, meaning that at least $1.5 trillion will have to come from debt and other financing arrangements.

More than any other Big Tech company, Oracle is testing the limits of this debt-fueled spending spree and is serving as a canary in the AI coal mine, financial analysts say. On Wall Street, the company's credit default swaps have become a way to monitor the sustainability of the debt load being taken on by Big Tech to fuel its AI ambitions, and guard against it. Look at Oracle if you want to assess the health of the AI investment boom, they say.

Compared to its peers, Oracle has much weaker cash flows, higher debt levels and lower credit ratings. It also has significantly higher counterparty risk from OpenAI, Nathan Liddle, a fixed-income research analyst at Columbia Threadneedle Investments, told MarketWatch.

"Oracle has far less flexibility than its peers in this investment cycle," Liddle said.

The cost of competing

In September, Oracle issued $18 billion of bonds, one of the biggest-ever corporate bond issuances by a tech company. The maturities on those bonds, or the date at which they need to be fully repaid, extended out 40 years to 2065.

Oracle is far from the only Big Tech company borrowing money to finance ambitious AI spending; Alphabet and Meta both closed massive bond offerings in 2025. But Oracle is in a more precarious financial position than any of the other hyperscalers, Liddle emphasized. Alphabet and Meta both have pristine AA credit ratings from S&P Global Ratings - giving them more borrowing power compared to Oracle, which has a BBB rating hovering at the lower end of investment grade. And Oracle's debt-to-equity ratio exceeds 462% while the other hyperscalers' ratios are under 50%, according to FactSet.

What originally excited Wall Street about Oracle's AI opportunity was the company's September disclosure that its remaining performance obligations, or future contracted revenues not yet recognized, had surged nearly 360% to upwards of $450 billion. By mid-October, that number had exceeded $500 billion. That meant that OpenAI and other companies planned to do more than $500 billion worth of business with Oracle over time, though Oracle would have to invest heavily in AI infrastructure to translate those contracted agreements into actual revenue, D.A. Davidson Gil Luria wrote in a September note following Oracle's initial disclosure.

But the optimism deflated significantly in recent weeks, as questions arose about whether Oracle's customers could renege on these contracts and leave it holding the bag for billions in underutilized infrastructure.

For now, Oracle is busy building data centers in Michigan, Wisconsin and Texas. With nearly $6 billion in negative free cash flow for the 12 months ending Aug. 31, the company is burning far more money than it generates through its business, leading it to take on significant debt to fund its AI buildout.

Oracle declined to comment for this story.

The $18 billion September issuance is likely to be just the beginning for Oracle's borrowing spree. In a November note, Morgan Stanley analyst Lindsay Tyler estimated that the company's combined debt and leases could rise to $290 billion over the next three years, a massive leap from the roughly $100 billion it held earlier this year.

Further debt issuances could bring Oracle to a "tipping point where the [investment-grade bond] market is reluctant to continue to lend," Columbia Threadneedle's Liddle said. That would increase the cost of borrowing for Oracle, creating a vicious spiral accelerating the company's cash burn and further increasing its need to borrow.

Read: Big Tech needs a staggering $1.5 trillion to fund the AI boom. This is the complex playbook it's using to get it.

Since inking the $300 billion agreement with Oracle, OpenAI has gone on its own $1 trillion dealmaking spree, and investors are now questioning the stability of Oracle's future revenues from the private company behind ChatGPT. In a November note, D.A. Davidson's Luria characterized Oracle as a "pawn" in OpenAI's "grand game of 'fake it 'till you make it,'" borrowing money for a customer that might never fulfill its promise.

"Earlier in the year, when someone would announce a deal, that money was going to come through no matter the time horizon," said Michael Treacy, head of market risk at Apex Fintech Solutions. "Only recently have investors started to actually ask the question: Will these payments come to fruition? Through the life of these contracts, will the vendor actually get paid?"

The concern is apparent in Oracle's five-year credit default swap (CDS) spreads, which have decoupled from the broader investment-grade Credit Default Swap Index - widening from 60 basis points at the end of October to 125 basis points as of last week, while the underlying index remains consistent. That means it costs significantly more to buy protection against the company defaulting on its debt. In fact, Oracle's five-year CDS spreads closed at 128 basis points on Dec. 5, the highest level since the height of the financial crisis in March 2009, according to Bloomberg.

Oracle's five-year CDS spread reached a recent high of 128 basis points last week.

The spread could reach 150 basis points in the near term, Morgan Stanley's Tyler predicted, as uncertainty remains about Oracle's future funding plans and whether it can count on OpenAI to deliver the promised revenue.

While both Liddle and Tyler believe the CDS pricing is primarily due to Oracle-specific risk, others believe it's a signal of systemic concerns over Big Tech's appetite for AI leverage. The company's CDS movement is emerging as a way to "hedge the entire debt cycle within AI," Apex's Treacy said. He added that the "presumption is that if there are tremors or fears or issues in the debt market, Oracle's CDS and debt will be the most correlated with that broader narrative" compared to the "Magnificent Seven" names - companies like Google or Microsoft that have the cash flows to withstand a downturn.

Matthew Unterman, managing director at S3 Partners, wrote in a note earlier last week that Oracle's increasing short interest and widening CDS spreads indicate the company is "trading as a proxy for investor anxiety around AI infrastructure leverage, making it a barometer for broader credit risk in the megacap tech complex."

Even the most well-capitalized companies are receiving scrutiny for their massive AI spending, as traders begin to bet on Nvidia (NVDA) and Meta defaulting on their debts. A liquid market for Meta's CDS sprang up after the company's debt issuance in late October. Nvidia's CDS market emerged after key partner CoreWeave (CRWV) announced delays in data-center construction in November.

The poster child of AI bubble fears

To avoid taking on additional debt, some companies have turned to alternative financing strategies such as private credit, the lightly regulated Wall Street industry that lends money it raises from investors as opposed to banking deposits.

(MORE TO FOLLOW) Dow Jones Newswires

12-09-25 0919ET