Part 5: Oracle’s Credit Crisis Deepens?: When Bond Markets Flash “Danger”

Written by

Terry Hale

on

in

Credit markets often see trouble before equity investors do. Oracle’s current credit metrics have deteriorated so rapidly that Morgan Stanley analysts now warn spreads could approach 2008 financial crisis levels. The numbers suggest we’re watching a high-stakes gamble that could reshape enterprise software.

The Bottom Line

For readers short on time, here’s the core argument:

  • The Signal: Oracle’s credit default swap spreads have nearly tripled to 125-128 basis points since September 2025, reaching levels not seen since the March 2009 financial crisis. Trading volume exploded from $200 million to $5 billion in seven weeks. Professional credit investors are paying record prices to insure against Oracle default.
  • The Cause: Oracle has committed $300 billion to build AI infrastructure for OpenAI, a customer that remains unprofitable and may account for over one-third of Oracle’s future revenue. To fund this, Oracle has borrowed over $100 billion (potentially heading toward $290 billion by 2028) while generating negative free cash flow.
  • The Risk: Oracle sits just two notches above junk status. Rating agencies have placed the company on negative outlook. Analysts warn Oracle could face liquidity constraints by late 2026. If AI demand falters or OpenAI stumbles, Oracle faces elevated risk of a self-reinforcing credit deterioration cycle where worsening credit begets higher borrowing costs, which further strain cash flow.
  • The Catalyst: December 10, 2025 earnings will be pivotal. Investors need clarity on financing strategy before deciding whether Oracle can navigate the 2-3 year gap between massive cash outlays and revenue recognition.

The rest of this analysis examines why credit markets are pricing in this level of stress, and whether they’re right.

The Credit Market Alarm Bell

Oracle’s five-year credit default swap spreads have exploded to 125-128 basis points as of early December 2025, according to ICE Data Services and Bloomberg.¹ ⁵⁰ That’s the highest level since March 2009, during the global financial crisis, approaching the 198 bps record. To put that in perspective, spreads have nearly tripled from approximately 43 basis points in late September 2025, when Oracle completed its $18 billion bond sale.²

The progression tells a story of rapidly escalating concern:

  • Late September 2025: roughly 43-55 basis points (post-bond sale baseline)
  • October 29-30, 2025: roughly 73-80 basis points
  • November 14, 2025: 106 basis points (single-day jump of 13.5 bps, the largest since December 2021)
  • November 20-21, 2025: 108-111 basis points
  • November 26, 2025: 125 basis points (three-year high)
  • December 2, 2025: 128 basis points (highest since March 2009)⁵⁰

Each uptick corresponded with revelations about Oracle’s OpenAI concentration, debt requirements, and execution challenges. Trading volume in Oracle CDS reached $5 billion over the seven weeks ending mid-November, up from just $200 million in the same period a year earlier.³ Oracle has effectively become the credit market’s proxy for AI infrastructure risk.

Morgan Stanley analysts Lindsay Tyler and David Hamburger issued a stark warning: spreads could breach 150 bps near-term and potentially approach 200 bps, the crisis levels last seen in 2008.⁴ Barclays characterized Oracle’s position as the most precarious among major tech companies, citing its 500% debt-to-equity ratio as evidence of structural vulnerability that other hyperscalers don’t share.

At 125 basis points (assuming 35% recovery), the implied cumulative five-year default probability falls somewhere in the 6-9% range, depending on how much of the spread reflects pure credit risk versus technical factors like hedging demand and liquidity premiums.⁵ Even at the lower end of that range, this is extraordinary territory for an investment-grade tech giant.

Oracle’s bonds tell a similar story. The $3.5 billion, 30-year tranche from Oracle’s September 2025 issuance has fallen approximately 8% since its October peak, with reports indicating it traded at an estimated 65 cents on the dollar, a roughly 35% discount to par value if accurate.⁶ Some of that discount reflects duration risk (long-dated bonds are more sensitive to interest rate movements), but credit concerns are clearly a factor. When bond investors accept losses of this magnitude rather than hold to maturity, they’re pricing in either default risk or severe credit downgrades that would force institutional selling. Many pension funds and insurance companies cannot hold sub-investment-grade debt by mandate.

The 4.9% bonds maturing February 2033 saw spreads widen 26 basis points to 83 bps over Treasuries.⁷ Morgan Stanley has closed the “buy bond” component of their basis trade while maintaining the “buy CDS protection” recommendation. That’s a clear vote of no confidence.⁸

To understand why credit markets are flashing these warning signs, we need to examine what Oracle has done and for whom.

Debt Has Swelled Past $100 Billion With More Coming

Oracle executed an $18 billion bond sale on September 24, 2025, the market’s second-largest investment-grade deal of the year, drawing $88 billion in orders (nearly 5x oversubscribed).⁹ The offering included six tranches with maturities spanning 2030 to 2065, including a rare 40-year bond.

But that was just the beginning. Since September, additional financing has materialized:

  • $18 billion project finance loan for Oracle’s New Mexico data center campus (early November 2025)¹⁰
  • $38 billion loan package for data centers in Texas and Wisconsin tied to Stargate, currently in syndication with JPMorgan, MUFG, Wells Fargo, BNP Paribas, Goldman Sachs, and others at approximately 250 basis points over SOFR, representing the largest AI infrastructure debt deal ever structured¹¹

Current total debt estimates range from $91 billion (based on Q1 FY2026 filings) to $104-110 billion when including bond issuance proceeds and new project financing.¹² Oracle’s debt load has surged by $56 billion in new borrowing, supporting the upper end of this range.¹³ Morgan Stanley projects net adjusted debt will more than double from roughly $100 billion today to a projected $290 billion by fiscal 2028 if Oracle executes its full AI infrastructure buildout, though this estimate depends heavily on assumptions about capex pace and project finance structures.¹⁴

A note on project finance: The $18 billion New Mexico loan and $38 billion Texas/Wisconsin package are structured as project finance, meaning they’re typically secured by the specific projects rather than Oracle’s entire balance sheet. If individual projects fail, Oracle may face reputational damage and strategic setbacks without necessarily triggering corporate-level default. However, the scale of Oracle’s commitments means project-level distress could still cascade into broader credit concerns.

The leverage ratios paint a concerning picture:

  • Total debt: $104-110 billion (up from roughly $20 billion in 2018)
  • Debt-to-equity: 3.33-4.36x (technology companies typically run under 1.0x; Microsoft sits at 0.3x, Amazon at 0.5x)¹²
  • Debt-to-EBITDA: roughly 4.0x (precisely at rating agency downgrade triggers)¹⁵
  • Interest coverage: 5-6x (adequate but deteriorating as interest expense climbs)¹⁶

Interest expense has risen from $1.99 billion in FY2020 to $3.58 billion in FY2025, consuming an increasing share of operating income.¹⁷

Why has Oracle taken on this unprecedented debt load? The answer is a single customer.

The OpenAI Counterparty: Oracle’s $300 Billion Bet on One Customer

The Stargate project, the $500 billion AI infrastructure initiative announced January 21, 2025 with OpenAI, SoftBank, and MGX, remains on track, with partners reporting approximately 7 gigawatts of capacity committed toward a 10GW target.¹⁸ Oracle’s specific commitment is a $300 billion, five-year contract with OpenAI for 4.5 gigawatts of data center capacity beginning in 2027, approximately $60 billion annually.

S&P estimates OpenAI could account for more than one-third of Oracle’s total revenues by fiscal 2028, creating substantial single-counterparty concentration risk.¹⁹

OpenAI’s financial health has improved on paper. The company completed its transition to a Public Benefit Corporation structure on October 28, 2025, unlocking the full $40 billion SoftBank funding round. Total capital raised stands at $57.9 billion, with the company’s valuation reaching $300 billion at the time of the Oracle contract announcement and subsequently rising to $500 billion via an October 2025 share sale. Revenue has grown from approximately $4 billion in 2024 to an estimated annualized rate of $13 billion by mid-2025. CEO Sam Altman claimed in November 2025 that revenue is now “well more” than $13 billion and projected OpenAI would exceed a $20 billion annualized revenue run rate by year-end 2025, with potential to hit $100 billion by 2027, two years earlier than previous projections. These are management projections, not audited figures, and represent aspirational targets rather than guaranteed outcomes.²⁰

However, OpenAI remains unprofitable, with 2025 losses projected at approximately $9 billion (spending $22 billion against $13 billion in revenue).²¹ HSBC analysis indicates OpenAI has committed to $1.4 trillion in data center and compute agreements but requires an additional $207 billion in funding even at $200 billion projected 2030 revenue.²²

This creates what I’d characterize as sequential counterparty dependency. The chain works like this: if AI sentiment weakens, OpenAI’s funding rounds become harder to close. If OpenAI’s funding struggles, Oracle’s counterparty risk increases. If Oracle’s credit deteriorates, borrowing costs rise, which strains cash flow further. Each party’s viability is contingent on the other’s success, with neither having yet demonstrated the ability to generate cash flows at the scale necessary to meet their full obligations independently.

The counterparty risk isn’t just theoretical. It directly affects whether Oracle can recognize the revenue it’s counting on.

The Revenue Recognition Time Bomb

Credit rating agencies don’t just look at revenue numbers. They analyze revenue quality using frameworks that parallel ASC 606 accounting requirements. The agencies assess whether reported revenue represents real economic activity or accounting optimism.

Moody’s counterparty risk assessment raises the same fundamental question that ASC 606’s collection probability requirement addresses. When Moody’s flagged “significant counterparty risk” due to Oracle’s OpenAI dependence,²³ they didn’t explicitly cite accounting standards. But the underlying concern maps directly to ASC 606-10-25-1(e), which requires entities to demonstrate that collection of “substantially all of the consideration” is “probable,” a threshold that accounting standards interpret as 75-80% or greater likelihood.²⁴ Given OpenAI’s current financial position (with annualized revenue of approximately $13 billion but no profitability and not expected to become cash-flow positive until the end of the decade), there is substantial uncertainty about whether Oracle will actually collect the consideration from these contracts.²⁵ This counterparty risk directly relates to whether Oracle’s Remaining Performance Obligations (RPO) can legitimately be recognized as future revenue.²⁶

If Oracle cannot collect “substantially all consideration” from OpenAI (because OpenAI lacks the financial capacity to pay $60 billion annually starting in 2027), then:

  1. Questions may arise about whether revenue should have been included in RPO
  2. Oracle could need to reduce previously recognized amounts
  3. The company’s financial projections would become less reliable
  4. Credit metrics based on projected revenue would deteriorate

The timing problem compounds the quality concern. Under ASC 606-10-25-27, Oracle cannot recognize revenue until performance obligations are satisfied, meaning data centers must be operational and OpenAI must be consuming capacity.²⁷ This creates a 2-3 year period (2025-2027) where Oracle incurs massive interest expense on borrowed capital but generates zero revenue from the OpenAI contract.

Interest coverage ratio deterioration trajectory (projected):

  • Current (FY2025): 5-6x (EBIT / Interest Expense)
  • Projected (FY2026-2027): Interest expense rises to an estimated $6-8 billion as debt increases, while EBIT growth remains limited because OpenAI revenue is $0
  • Projected Result: Interest coverage could fall toward 3-4x, approaching distress levels

Credit covenants typically require maintaining minimum interest coverage ratios (often 3.0x or higher). If Oracle’s interest coverage falls below covenant thresholds before OpenAI revenue begins, the company faces technical default regardless of the long-term viability of the deal.

S&P’s leverage calculations also incorporate revenue quality adjustments. In its analysis assigning a negative outlook, S&P initially set its downgrade trigger at 3.5x debt-to-EBITDA before subsequently loosening this to 4x, a tacit acknowledgment that Oracle would breach the stricter threshold. The agency’s concern that leverage will “exceed 4x in fiscal years 2027 and 2028″¹⁹ underscores the view that Oracle’s projected revenue growth may not be sufficient to offset its massive debt-funded capital expenditures, even with the relaxed hurdle. Credit analysts understand that revenue from highly uncertain sources (like OpenAI’s $300 billion commitment) should carry less weight in credit analysis than revenue from diversified, creditworthy customers.

The constraint on variable consideration under ASC 606-10-32-11 requires entities to include estimates “only to the extent that it is probable that a significant reversal in the amount of cumulative revenue recognized will not occur.”²⁸ The ASC 606-10-32-14 requirement to “update the estimated transaction price… at the end of each reporting period” creates additional covenant risk.²⁹ If Oracle must reduce its RPO or revenue projections based on changed facts and circumstances (like OpenAI funding difficulties), lenders may invoke material adverse change clauses or accelerate debt repayment.

Rating Agencies Sound the Alarm

All three major credit rating agencies maintain Oracle at investment-grade levels, but two have placed the company on negative outlook, and the distance to junk status is uncomfortably narrow:

Moody’s (Baa2, Negative Outlook): Lowered outlook from Stable in July 2025, citing “significant counterparty risk” from the $300 billion OpenAI contract. Analysts noted adjusted debt-to-EBITDA exceeded 4x with trailing free cash flow at negative $5.1 billion. Moody’s characterized Oracle’s data center build as effectively one of, if not the world’s largest, project financing and warned that “free cash flow will likely be negative for an extended period.”²³

S&P (BBB, Negative Outlook): Affirmed the rating but flagged “strain on the company’s credit profile due to aggressive capital expenditure plans.” S&P initially set a downgrade trigger at 3.5x debt-to-EBITDA but has since loosened this threshold to 4x, acknowledging Oracle’s transformation requires upfront spending before revenue generation. S&P’s models project actual FY2026 spending could reach $38 billion (up from earlier $27 billion estimates), with capex potentially peaking above $60 billion in fiscal 2028. The agency projects leverage will “exceed 4x in fiscal years 2027 and 2028,” meaning Oracle risks breaching even the relaxed threshold if these projections prove accurate.¹⁹

Fitch (BBB, Stable Outlook): The most favorable assessment, expecting EBITDA leverage to approach 3.5x in FY2026 before declining. Fitch highlighted Oracle’s $11 billion cash buffer and staggered debt maturities as sources of flexibility. However, Fitch projects “negative pre-dividend FCF exceeding $11 billion in fiscal years 2026 and 2027,” estimates that assume current capex trajectories continue.³⁰

Oracle sits just two notches above speculative grade on the scales of both Moody’s and S&P. A downgrade of more than one notch from either agency would push Oracle to the lowest investment-grade tier (BBB-/Baa3). A two-notch downgrade means junk status.

Analyst Commentary: Bears vs. Bulls

Investment bank commentary has turned decidedly cautious, with several firms explicitly questioning Oracle’s ability to maintain investment-grade status.

Barclays analyst Andrew Keches (November 11, 2025) downgraded Oracle to Underweight with the damning assessment: “We struggle to see an avenue for Oracle’s credit trajectory to improve.”³¹ Barclays warned Oracle may face significant liquidity pressure by November 2026 and predicted the credit rating could ultimately fall to BBB-, the lowest investment-grade tier, one notch above junk. The firm characterized Oracle’s 500% debt-to-equity ratio as the highest among major tech companies.

Morgan Stanley (November 26, 2025) warned that “near-term credit deterioration and uncertainty may drive further bondholder and lender hedging.” With Morgan Stanley projecting funding needs approaching $200 billion through 2028, the $18 billion September bond issue represents only a fraction of what Oracle must raise, though the precise percentage remains subject to assumptions about capex trajectories and project finance structures. Their models project free cash flow will remain negative $9.7 billion in FY2026, potentially worsening to negative $24.3 billion by FY2028, estimates that carry significant uncertainty given the unprecedented scale of Oracle’s buildout.³²

JPMorgan (October 20, 2025) downgraded Oracle to Neutral from Overweight, citing “substantial capital needs and limited visibility into the company’s financing strategy.”³³

Bank of America flagged Oracle’s credit widening faster than the overall investment-grade market.⁶ One widely circulated characterization attributed to the bank described Oracle as “spending money it doesn’t have on facilities that haven’t been built for customers it doesn’t have,” a pointed summary of the credit concern, though the original source of this specific phrasing remains unconfirmed.

Despite these concerns, some equity analysts remain bullish. Wells Fargo initiated coverage on December 3, 2025 with an “Overweight” rating and $280 price target, arguing Oracle is “still early” in an AI-led reacceleration.³⁴ HSBC reaffirmed a Buy rating with a $382 target, citing strong backlog visibility.³⁵ Deutsche Bank reiterated a Buy rating with a $375 target, highlighting the OpenAI relationship.³⁶

A note on conflicts of interest: Both HSBC and Deutsche Bank served as lead underwriters on Oracle’s $18 billion September 2025 bond offering.³⁷ When banks that underwrote a company’s debt subsequently issue bullish equity ratings, they have an inherent incentive to support the company’s narrative. A deteriorating credit story reflects poorly on their underwriting judgment and hurts any bonds still on their books. Wells Fargo, notably, was not among the underwriters and thus appears to be issuing an independent opinion. I’m not suggesting the bullish analysts are wrong, but sophisticated investors should factor in these relationships when weighing research.

The consensus among 31 analysts remains “Moderate Buy” with an average 12-month target of approximately $338, implying roughly 58% upside from current levels.³⁸ This stark divergence between credit market alarm bells and equity analyst optimism deserves attention. Credit and equity investors often operate on different time horizons, and both camps could be partially right: credit investors may correctly identify near-term stress while equity analysts correctly identify long-term value. But someone has to be more wrong, and the gap is wide enough that the resolution will be instructive.

The Cash Burn Crisis

The most troubling development for credit investors is Oracle’s free cash flow trajectory. For Q1 FY2026 (quarter ended August 31, 2025):

  • Operating cash flow: $8.1 billion
  • Capital expenditures: $8.5 billion
  • Free cash flow: Negative $362 million

On a trailing twelve-month basis through August 2025, Oracle’s free cash flow has turned sharply negative. Financial data sources present conflicting figures on the exact amount, highlighting the volatility of this metric. The Economic Times reports a TTM free cash flow of Negative $5.9 billion,¹³ a figure that reflects the immense cash burn from recent capital outlays. In contrast, data aggregator Macrotrends reports a more modest, though still negative, TTM figure of Negative $394 million.³⁹ Regardless of the precise figure, this marks a rare and troubling occurrence of sustained negative free cash flow for Oracle.

While the company experienced brief negative quarterly FCF in 2021, the current situation is more severe and prolonged, driven by capital expenditures that have tripled to $21 billion in FY2025 (37% of revenue, up from 17% two years prior).⁴⁰ Management has guided FY2026 capex at $35 billion.¹⁹

The Structural Mismatch

The relationship between revenue recognition and cash flow creates a significant structural mismatch:

Cash Out (2025-2027, estimated):

  • $35-80 billion annual capex (based on management guidance and analyst projections)
  • $4.3-4.8 billion annual interest expense (rising to an estimated $6-8B as debt increases)
  • Operational expenses continue
  • Total estimated cash needs: $50+ billion annually

Cash In (2025-2027, estimated):

  • Core business: $20-21 billion operating cash flow (based on recent trends)
  • OpenAI payments: Minimal or $0 until 2027 (revenue recognition doesn’t begin until facilities are operational; exact payment milestone structure not publicly disclosed)
  • Projected gap: $30+ billion annually must be borrowed

Important caveat: These projections involve significant uncertainty. Actual capex could vary based on construction timelines, supply chain conditions, and strategic decisions. Operating cash flow depends on core business performance. The gap between cash out and cash in could be smaller or larger than projected.

This isn’t a temporary working capital need. It’s structural negative cash flow for 3-5 years minimum. While some technology companies (Amazon during AWS buildout, Tesla during manufacturing scale-up) have navigated extended periods of negative free cash flow, few have done so at this scale while maintaining investment-grade credit ratings and without the cushion of a dominant existing cash-generating business subsidizing the buildout at the required level.

What Junk Status Would Mean

If Oracle loses investment-grade status, the consequences cascade:

  1. Forced Selling: Institutions restricted to investment-grade holdings must sell, creating massive price pressure
  2. Commercial Paper Cutoff: Oracle loses access to short-term funding markets critical for working capital
  3. Refinancing Crisis: Existing debt becomes nearly impossible to roll over at reasonable rates
  4. Covenant Triggers: Many debt agreements include ratings-based covenants that could accelerate repayment
  5. Self-Reinforcing Deterioration Risk: Higher borrowing costs make profitability harder, justifying further downgrades

The danger is a self-reinforcing credit deterioration where widening spreads increase borrowing costs, further straining free cash flow and pushing leverage ratios above downgrade triggers.

The Counterargument: Is This Just a Proxy Trade?

Before reaching conclusions, intellectual honesty requires addressing a legitimate counterargument: Oracle’s CDS may be overstating company-specific risk because traders are using it as a proxy hedge against the broader AI buildout.

The bull case goes like this: Investors wanting to hedge against AI going bust face a structural problem. Microsoft, Alphabet, Amazon, and Meta maintain fortress balance sheets with minimal CDS spread volatility, leaving no liquid, affordable way to short their credit. Oracle, with its mid-BBB rating and heavy AI exposure, became the cheapest, most efficient hedge against tech sector risk. As Sage Advisory put it: “Oracle offered a cheap, efficient hedge against tech sector risk and a proxy for the AI buildout.”⁴¹

This argument has merit. Several factors support it:

Technical demand may be inflating spreads. The $5 billion in CDS volume didn’t materialize because $5 billion worth of investors suddenly became convinced Oracle would default. Banks that provided the $18 billion New Mexico project finance loan are hedging their own exposure via CDS. Asset managers with broad AI exposure are using Oracle as portfolio insurance. Morgan Stanley explicitly noted that “bondholder hedging dynamic and thematic hedging dynamic could both grow.”³²

Sophisticated funds are betting against the panic. Fintech Blueprint reports that credit funds like Saba Capital are taking the other side of the trade, “collecting premiums on what they see as overpriced AI fear.”⁴² If these sophisticated credit investors believed Oracle’s default probability truly approached 9%, they wouldn’t be selling CDS protection.

Oracle’s core business continues performing. Strip away the AI buildout narrative, and Oracle remains a $55+ billion annual revenue company with a dominant database franchise and growing cloud business. The core enterprise software operation generates substantial, recurring cash flow. As Bank of America analyst Brad Sills noted, Oracle is structuring AI compute contracts as “non-cancelable, non-modifiable, take-or-pay contracts ensuring stable cash flows.”⁴³ If the Oracle-OpenAI contracts are structured as true take-or-pay agreements, Oracle has contractual recourse even if OpenAI underutilizes capacity, though collecting from a financially distressed counterparty would remain challenging.

If AI sentiment improves, spreads compress regardless of fundamentals. This is key: because Oracle CDS functions as an AI macro hedge, spreads could tighten dramatically if AI sentiment turns positive, even without any change to Oracle’s actual credit profile. The proxy trade cuts both ways.

So what does this mean for the analysis?

The proxy trade argument is real but incomplete. Here’s why:

Oracle became the proxy precisely because it has the weakest balance sheet among hyperscalers. The proxy trade wouldn’t work if Oracle’s credit quality matched Microsoft’s. Traders chose Oracle because its actual leverage ratios, counterparty concentration, and negative free cash flow make it genuinely vulnerable to AI sentiment shifts. The proxy trade and the fundamental risk aren’t separate; they’re intertwined.

Technical demand and fundamental risk aren’t mutually exclusive. Yes, some spread widening reflects macro hedging. But Oracle’s 500% debt-to-equity ratio, negative $362 million quarterly free cash flow, and significant revenue concentration with an unprofitable counterparty are real. The proxy trade works because the underlying fundamentals justify concern.

Spread levels still price Oracle-specific risk. Even if technical factors explain some widening, spreads at 125-128 bps still represent the market’s assessment of Oracle’s creditworthiness. Institutional investors don’t pay 1.25% annually for insurance they believe is worthless.

The counterargument assumes the bull case. If AI execution goes perfectly, spreads tighten and Oracle thrives. But the entire question is whether execution risk is being adequately priced. The proxy trade argument essentially says “Oracle will be fine if AI works out,” which is precisely the assumption this analysis examines.

The honest conclusion: Oracle’s CDS spreads likely overstate pure company-specific default risk by some margin due to proxy trade dynamics. But “overstated” doesn’t mean “wrong.” Oracle’s fundamental credit profile has genuinely deteriorated, and the market has identified it as the most leveraged, most exposed player in the AI infrastructure buildout. Both things can be true simultaneously.

The Verdict: What Credit Markets Know That Equity Markets Don’t

Credit investors are institutional professionals managing trillions in fixed income. When they flee Oracle bonds despite 5%+ yields in a low-rate environment, they’re voting with real money that Oracle’s risk exceeds its return.

Oracle shares have retreated roughly 38% from September 2025 highs, falling from approximately $345.72 (September 10 high)⁴⁴ to around $214 as of December 4, 2025.⁴⁵ November alone saw a sharp 23% decline as investors soured on the AI growth story amid mounting debt concerns.⁴⁶ This has erased approximately $275 billion in market capitalization, with the market cap falling from approximately $875 billion at September 2025 highs to approximately $590 billion as of December 3, 2025.⁴⁷

The credit analysis reveals deeper problems than the stock price decline suggests. The combination of:

  • 125-128 bps CDS spreads at levels not seen since March 2009 (with Morgan Stanley warning spreads could breach 150-200 bps near-term)
  • Significant customer concentration with an unprofitable counterparty
  • 2-3 years of zero revenue recognition while interest expense compounds
  • Questions about depreciation accounting and earnings quality
  • $100+ billion in additional borrowing needs with the September bond offering representing only a fraction of projected requirements
  • Analysts explicitly warning about potential downgrade to BBB-

These factors create a credit profile that resembles distressed debt more than investment-grade securities.

History offers a cautionary parallel, with important caveats. Oracle’s 1990-1991 crisis involved manageable debt that was quickly repaid once operations stabilized. Banks reduced Oracle’s credit line from $170 million to $80 million, serious but survivable. The company eliminated all bank debt within 18 months.⁴⁸

To put that in scale context: in 1990, Oracle’s revenue was approximately $1 billion, so the $90 million credit reduction represented about 9% of annual revenue. Today’s Oracle generates roughly $55 billion annually, meaning an equivalent proportional stress would be approximately $5 billion, well within the company’s $11 billion cash buffer.

Today’s $104-110 billion debt load heading toward a projected $290 billion is qualitatively different in both absolute and relative terms. This isn’t a liquidity squeeze that better management can fix on its own. It’s a potential credit deterioration trajectory where debt service obligations could consume an increasingly large share of operating income. The 1990-1991 crisis was correctable because the problems were internal (accounting controls, revenue recognition accuracy). Management could fix these directly. The 2025 situation depends on external factors (OpenAI’s viability, AI market demand, revenue recognition timing) that Oracle cannot control unilaterally.

December Earnings: The Defining Moment

Oracle faces a critical juncture at its December 10, 2025 earnings call (Q2 FY2026, after market close), which analysts describe as a potential “make-or-break moment” for the credit narrative.⁴⁹ Investors are seeking clarity on financing strategy, leverage targets, special purpose vehicle structures, cloud margins, and the path to positive free cash flow.

The fundamental tension is stark: Oracle has contracted to build $300+ billion in infrastructure for a customer that may represent over a third of future revenues but is itself burning cash and dependent on continued fundraising. With Morgan Stanley projecting funding needs approaching $200 billion through 2028, Oracle’s financing challenge dwarfs the September bond offering.

For those evaluating Oracle exposure (whether as investors, customers, or partners), credit markets are currently pricing significant stress. A negative catalyst, whether OpenAI stumbling, a broader AI correction, or inability to raise the next $50 billion needed, could trigger a self-reinforcing deterioration cycle.

The question isn’t whether Oracle faces credit stress. That’s already here. CDS spreads have nearly tripled in two months. The question is whether Oracle can maintain investment-grade status long enough for the OpenAI revenues to materialize. Based on current credit market pricing, professional investors appear skeptical.

A note on projection uncertainty: This analysis relies heavily on forward-looking estimates from rating agencies, investment banks, and company management. These projections involve significant assumptions about AI demand, construction timelines, OpenAI’s financial trajectory, and macroeconomic conditions. Actual outcomes could differ materially in either direction. The credit market signals described here reflect current institutional sentiment, not guaranteed outcomes.

This is Part 5 of a 6-part series analyzing Oracle’s financial risks:

Bibliography

  1. LA Times: “Morgan Stanley Warns Oracle Credit Default Swaps Near Record on AI Fear” (November 27, 2025)
  2. Yahoo Finance: “A Hedge Against AI Crash Emerges as Oracle CDS Market Explodes” (November 20, 2025)
  3. FinancialContent: “Oracle’s CDS Market Explodes: A ‘Hedge Against AI Crash’ Emerges” (November 20, 2025). Reports Barclays data showing $5 billion CDS trading volume.
  4. Financial Times: “Oracle Hit Hard in Wall Street’s Tech Selloff” (November 2025). Reports Morgan Stanley analysts’ warning.
  5. Investing.com: “Oracle and CoreWeave Credit Default Swap Spreads Widening: Omen or Jitters?”
  6. Axios: “Big Tech’s debt problem” (November 17, 2025)
  7. Ainvest: “Oracle Credit Risk Warning: Surge in CDS Spreads Reflects AI Debt Pressures” (November 2025)
  8. Benzinga: “Morgan Stanley Warns Oracle Debt Risk Could Hit 2008 Crisis Levels” (November 2025)
  9. Yahoo Finance/Bloomberg: “Oracle Raises $18 Billion in Second-Biggest Bond Sale This Year” (September 24, 2025)
  10. Yahoo Finance/Bloomberg: “Banks Lend $18 Billion for Oracle-Tied Data Center in New Mexico” (November 7, 2025)
  11. FX Leaders: “Oracle Faces $193 Rout as Traders Bet Big on Default Protection” (November 27, 2025)
  12. Simply Wall St: “Oracle Balance Sheet & Financial Health Metrics”
  13. Economic Times: “Oracle’s debt mountain grows: $56 billion in new borrowing as cash flow turns deeply negative” (November 2025)
  14. Investing.com: “Oracle’s credit default swaps rise as AI investments grow” (October 29, 2025). Reports Morgan Stanley projection of $290 billion net adjusted debt by fiscal 2028.
  15. Ainvest: “Oracle’s $18 Billion Debt Issuance: Strategic Rationale and Market Implications”
  16. Stock-Analysis-on.net: “Oracle Corp. Analysis of Debt”
  17. Oracle Corporation 10-K and 10-Q filings, FY2020-FY2026
  18. OpenAI: “Stargate Advances with Partnership with Oracle”
  19. Investing.com: “Oracle Faces Negative Outlook at S&P as Cloud Growth Strains Cash Flow” (September 2025)
  20. Sacra: “OpenAI Company Profile”; Fortune: “Sam Altman says OpenAI’s revenue is ‘well more’ than reports of $13 billion” (November 1, 2025)
  21. Fortune: “OpenAI says it plans to report stunning annual losses through 2028” (November 12, 2025)
  22. Fortune: “OpenAI won’t make money by 2030 and still needs to come up with another $207 billion to power its growth plans, HSBC estimates” (November 26, 2025)
  23. Reuters: “Moody’s Flags Risk in Oracle’s $300 Billion Recently Signed AI Contracts” (September 17, 2025)
  24. Baker Tilly: “The Revenue Recognition Transition: Identifying the Contract”
  25. The Register: “Oracle’s AI debt binge raises eyebrows” (September 29, 2025)
  26. Economic Times: “Will Moody’s and S&P reclassify Oracle’s bonds as junk debt? Key points investors need to know” (November 2025)
  27. Financial Accounting Standards Board (FASB). “ASC 606-10-25-27: Revenue Recognition Over Time – Criteria.” Accounting Standards Codification.
  28. Financial Accounting Standards Board (FASB). “ASC 606-10-32-11: Constraint on Variable Consideration.” Accounting Standards Codification.
  29. Financial Accounting Standards Board (FASB). “ASC 606-10-32-14: Updates to Transaction Price Estimates.” Accounting Standards Codification.
  30. Investing.com: “Fitch affirms Oracle’s BBB rating amid AI infrastructure investments” (October 13, 2025). Reports Fitch BBB Stable outlook, EBITDA leverage approaching 3.5x in FY2026, and negative pre-dividend FCF exceeding $11 billion in fiscal years 2026 and 2027.
  31. CNBC: “AI sentiment is waning. Wall Street cools on Oracle buildout plans” (November 13, 2025)
  32. Yahoo Finance: “Morgan Stanley Warns Oracle Credit Protection Nearing Record High”
  33. Investing.com: “JPMorgan downgrades Oracle’s credit rating to Neutral on AI expansion costs” (October 20, 2025)
  34. Investing.com: “Oracle Initiated at Overweight: Wells Fargo Sees Room to Run in AI Supercycle” (December 3, 2025)
  35. Yahoo Finance: “Oracle (ORCL): HSBC Reaffirms Buy Rating with $382 Target” (December 2025)
  36. Investing.com: “Deutsche Bank Reiterates Buy Rating on Oracle Stock, Citing AI Opportunity” (December 2025)
  37. Invezz: “Oracle seeks $15 billion bond sale as cloud expansion costs rise” (September 24, 2025); Global Finance Magazine: “Corporate Bond Market Booms After Fed Rate Cut” (October 2025). Both sources confirm Deutsche Bank, Goldman Sachs, HSBC, Bank of America, Citigroup, and JPMorgan as lead underwriters on Oracle’s $18 billion bond offering.
  38. TipRanks: “Oracle Stock Forecast & Price Target” (accessed December 2025)
  39. MacroTrends: “Oracle Free Cash Flow 2011-2025”
  40. Oracle Corporation Q1 FY2026 10-Q Filing (August 31, 2025)
  41. Sage Advisory: “Oracle Emerges as the Key Proxy in Tech’s $1.5 Trillion AI Debt Wave” (November 2025)
  42. Fintech Blueprint: “AI: Oracle’s Credit Default Swaps and the AI Hedge” (November 2025)
  43. Yahoo Finance: “The most important chart in AI today? Oracle’s default swaps blow out” (November 2025)
  44. StatMuse Money: “Oracle Stock Price September 2025”
  45. TechStock2: “Oracle Stock (ORCL) on December 3, 2025: AI Cloud Powerhouse or Overleveraged Risk?” (December 3, 2025)
  46. Motley Fool: “Why Oracle Stock Tumbled 23% in November” (December 2, 2025)
  47. Capital.com: “Oracle Market Cap”; Barron’s: “Oracle Stock Price Debt Rating Analysis” (December 2025)
  48. Funding Universe: “Oracle Corporation History”
  49. Oracle Investor Relations: “Oracle Sets the Date for its Second Quarter Fiscal Year 2026 Earnings Announcement” (December 2, 2025)
  50. WebProNews: “Oracle’s CDS Surge: Wall Street’s New AI Crash Barometer” (December 2, 2025). Reports Oracle CDS reached 128 basis points, highest since March 2009 financial crisis, citing Bloomberg data.

More posts