Bond Update: Oracle Downgraded to BBB- – Yields as High as 7.8%: Opportunity or Warning Signal?

On 9 July 2026, S&P Global Ratings downgraded Oracle’s issuer credit rating from BBB to BBB−, citing the company’s aggressive AI cloud infrastructure investments and persistently elevated adjusted leverage above 4x. The rating now sits just one notch above non-investment grade. Five-year credit default swap (CDS) spreads have broken through 198 basis points to hit a record high, bond prices have declined, and the 30-year bond yield has risen to approximately 7.8%, clearly reflecting a marked erosion of market confidence in Oracle’s credit quality (see Chart 1).

Chart 1: Oracle Bond Prices and Yields

Market concerns centre on two main points: first, Oracle’s continuous debt issuance to fund the cash-flow gap created by AI cloud infrastructure build-out, with total debt climbing to nearly US$130 billion; and second, whether its massive US$638 billion backlog of remaining performance obligations (RPO) — highly concentrated with a single client, OpenAI — can be converted into revenue on schedule. This note dissects Oracle’s actual financial position to assess whether its bonds remain worth holding following the downgrade.

Cloud Business Drives Record Revenue

Oracle’s fiscal 2026 results (year ended May 2026) exceeded market expectations, with both revenue and profitability reaching new highs. Full-year total revenue grew 17% YoY to US$67.4 billion. Cloud was the primary growth engine — total cloud revenue rose 39% to US$34.0 billion, lifting its share of total revenue from 43% in FY2025 to 50%. Within this, cloud infrastructure (IaaS) revenue surged 77% to US$18.1 billion, while cloud applications (SaaS) revenue increased 11% to US$15.9 billion, underscoring robust demand for AI training and inference. AI momentum is also spreading into multicloud databases, where revenue jumped 404% in the fourth quarter, indicating demand is extending from infrastructure into Oracle’s cloud applications (see Chart 2).

On profitability, FY2026 operating profit rose 17% YoY to US$20.6 billion, with the operating margin holding steady at a solid ~31%. Core earnings power remains robust. Traditional software revenue declined a modest 1% YoY to US$24.5 billion as customers migrate from on-premises deployments to the cloud; however, the highly sticky software support revenue within this segment still reached US$19.8 billion and edged higher YoY. This continues to serve as Oracle’s long-standing stable cash-flow foundation and an important buffer for debt-servicing capacity. Looking ahead to FY2027, management has maintained its US$90 billion revenue guidance, reflecting confidence in sustained growth.

Chart 2: Oracle Business Revenue Breakdown

Backlog Surged by US$500 Billion Over the Year

Oracle’s remaining performance obligations (RPO) expanded rapidly over the past year. As of the fourth quarter of FY2026, the backlog stood at US$638 billion, representing a net increase of approximately US$500 billion YoY. The pace of growth reflects the enormous market demand for AI training and inference compute capacity. The overwhelming majority of new orders stem from large-scale AI infrastructure contracts, with the OpenAI contract (including the Stargate programme) the largest at US$300 billion — nearly half of the total backlog. Other major clients include xAI, Meta and Nvidia.

Capex Creates Significant Debt Pressure

To fulfil the large backlog, Oracle has been financing heavy capex on AI cloud infrastructure. FY2026 capex reached US$55.6 billion, while operating cash flow was only US$32.0 billion, creating a funding gap of more than US$23 billion. The company therefore issued US$43 billion of bonds and US$5 billion of equity during the fiscal year. Total debt rose nearly 40% YoY to US$129.5 billion. After deducting approximately US$31.3 billion of cash and cash equivalents, net debt stood at roughly US$98.0 billion, with net gearing remaining elevated at 228% (see Chart 3).

Chart 3: Net Debt and Capex

Management expects FY2027 capex to rise further to US$90–95 billion and will require approximately US$40 billion of external financing, implying another increase in total debt. However, management has stated that it will not issue additional new debt within calendar 2026; the US$40 billion financing will be executed through a combination of debt and equity, including the previously announced US$20 billion at-the-market (ATM) equity programme. This suggests that debt growth will moderate in FY2027 and that the capital structure has scope to improve.

Notably, beyond the reported debt, S&P and Nikkei estimate that Oracle has approximately US$273 billion of data-centre lease commitments (phasing in from FY2027 to FY2029) plus power-related purchase commitments. These will be recognised on the balance sheet only as the facilities become operational and are therefore not currently included in reported liabilities. Importantly, most of these commitments are non-cancellable regardless of whether the backlog is converted into revenue on schedule; Oracle remains obligated to pay. Consequently, Oracle’s true total liabilities exceed the reported figures — a key factor behind S&P’s downgrade to BBB−.

Equity and Debt Demand Remains, but Funding Costs May Rise

Oracle’s ability to continue accessing capital markets is fundamental to its ongoing data-centre expansion. In February 2026 the company issued approximately US$25 billion of bonds that attracted roughly US$129 billion of demand — one of the largest corporate bond order books on record. Nevertheless, the 10-year tranche priced at a spread of about 145 basis points, well above the ~95 basis points of the BBB corporate index at the time, reflecting market concerns over the company’s elevated debt load. Overall, we believe demand for Oracle’s equity and debt remains adequate at present. However, the S&P downgrade to BBB−, combined with an estimated US$190 billion of debt supply across the broader sector in 2026, means that future funding costs and the sustainability of demand warrant close monitoring.

Debt-Servicing Success Hinges on Two Critical Factors

Management has confirmed that of the US$638 billion backlog, approximately 12% (c. US$76.6 billion) is expected to convert into revenue within the next 12 months, and roughly 34% (c. US$216.9 billion) within months 13–36. While the timing of free-cash-flow inflection remains unclear, the harvest period is scheduled from FY2028 onward. At that point Oracle is expected to reduce its need for new debt issuance and, once cash flow turns positive, to begin gradual deleveraging. Therefore, successful debt servicing rests on two pillars: (1) whether the backlog converts into revenue on schedule; and (2) whether Oracle can continue to service its obligations without default before the harvest period arrives.

Key Factor 1: Can the Backlog Convert into Revenue? Supported, but Dependent on OpenAI Performance and Construction Progress

The high concentration of the backlog with OpenAI was an important consideration in S&P’s downgrade. The two parties have signed a more-than-US$300 billion, five-year cloud contract (including capacity under the Stargate programme) that accounts for roughly half of the total backlog. Under the contract, OpenAI is required to make annual payments of approximately US$60 billion to Oracle. OpenAI, however, remains unprofitable and currently generates only about US$24 billion of annualised revenue, meaning it must rely on continuous external financing to meet its obligations. That said, OpenAI completed a US$122 billion funding round in March 2026 with participation from Amazon, Nvidia, SoftBank and other major investors, providing near-term support for its payment capacity and thereby increasing the probability that Oracle’s backlog will convert into revenue.

In addition, as noted in our previous report, Oracle’s data centres employ a multi-tenant architecture, meaning any capacity that becomes available can be readily re-leased to other customers. Should OpenAI fail to fully perform under the contract, other clients can utilise the freed capacity. In practice, vacancy rates in major North American data-centre markets have continued to tighten even as supply has risen sharply, underscoring still-strong demand for AI compute that can rapidly absorb capacity and partially mitigate concentration risk (see Chart 4).

Chart 4: North American Major Market Data-Centre Vacancy Rates

On the construction front, Oracle delivered more than 1.2 GW of capacity to customers across the whole of FY2026. In the first quarter of FY2027 the pace accelerated to nearly 1 GW — almost equal to the prior four quarters combined — demonstrating that capacity delivery is speeding up. To reduce delays caused by grid connection queues, the company is increasingly favouring behind-the-meter, off-grid power solutions that bypass the typical four-to-seven-year wait times.

However, the recent rejection by regulators of the natural-gas pipeline application for Project Jupiter — Oracle’s 2.5 GW flagship facility in New Mexico linked to OpenAI — and the still-pending public-hearing process (with no scheduled date) for the air-emission permits required for the battery microgrid have raised market concerns. These developments risk causing Oracle to miss the original August power-supply timeline and potentially push it into next year. While overall construction progress remains solid, the setback at Project Jupiter serves as a reminder that bypassing the grid does not eliminate risk; self-built power generation remains subject to permitting and environmental constraints. Timing of revenue conversion therefore carries genuine uncertainty.

Behind-the-meter, off-grid power refers to independent electricity systems installed after the utility meter, dedicated solely to the data centre and not connected to the external public grid.

Key Factor 2: Can Oracle Service Its Debt Before the Harvest Period? Liquidity Adequate, Near-Term Maturity Pressure Manageable

In FY2026 Oracle generated US$32.0 billion of operating cash flow, exceeding its approximately US$29.9 billion of EBITDA. A key contributor was customer prepayments related to AI contracts, which alone contributed roughly US$4.6 billion of cash and bolstered liquidity. In addition, some customers “bring their own hardware” — purchasing expensive GPUs, servers and related equipment themselves and handing them over to Oracle to operate. Together with prepayments, these arrangements have already reached a cumulative US$75 billion, meaning Oracle does not need to fund the outlays itself and thereby reduces its external financing requirement. Given the current scarcity of compute capacity, prepayment or customer-supplied hardware remains a preferred method for securing capacity; we therefore expect this model to persist in the near term and continue to ease Oracle’s funding pressure.

On the debt structure, Oracle’s obligations are predominantly long dated and well staggered. Debt maturing over the next three years totals only about US$17.8 billion. With US$31.3 billion of cash and cash equivalents on hand plus approximately US$32 billion of annual operating cash flow, the company has more than sufficient resources to cover all maturities over the coming three years (see Chart 5). Moreover, nearly half of total debt does not mature until 2040 or later, keeping near-term refinancing pressure manageable.

Chart 5: Oracle Debt Maturity Profile

FY2026 interest expense was US$4.6 billion, producing an interest coverage ratio of approximately 6.5x. Coverage has remained within a stable range over the past five years, underscoring Oracle’s solid ability to service interest (see Chart 6). Even if interest expense rises with future new issuance, the current buffer remains comfortable. Taking both the maturity profile and interest-coverage metrics into account, we believe Oracle’s financial position is adequate to support debt service over the next several years.

Chart 6: Oracle Interest Coverage Ratio

Bond Investment

In summary, while market concerns regarding Oracle have a sound basis, the situation is not yet critically severe. From a fundamental perspective, the cloud business is growing rapidly, the traditional software franchise continues to generate stable cash flow, and the US$638 billion backlog provides meaningful visibility into future revenue. From a credit perspective, the company holds more than US$30 billion of cash and enjoys interest coverage of 6.5x, providing a solid foundation for near-term debt service. We therefore assess the near-term default risk as low. Whether the company can maintain its investment-grade rating, however, will depend on subsequent progress in deleveraging and backlog conversion.

From an investment standpoint, post-downgrade Oracle bond yields range from 6.1% to 7.8%, offering a clear spread versus peers of similar rating and thereby providing reasonable compensation for the risks described above (see Chart 7). The risk-reward profile appears acceptable for investors able to tolerate a degree of credit risk in pursuit of higher yield. For income-oriented investors we prefer intermediate- and shorter-dated bonds to limit duration and long-term risk exposure. Investors considering longer-dated paper should treat positions as strategic/trading opportunities rather than buy-and-hold holdings, given their higher sensitivity to interest-rate and credit news. More risk-averse investors would be better advised to wait for clearer evidence of successful order conversion and deleveraging before participating.

Chart 7: Oracle Yield Curve v.s.US BBB- Yield Curve

Table 1: Oracle Bonds

Bond	Bond Credit Rating（S&P／Fitch）	Tenor	Investor Buy Price	Net YTM
ORCL 4.800% 26Sep2032 Corp (USD)	BBB-/BBB	6.18	92.69	6.029
ORCL 4.300% 08Jul2034 Corp (USD)	BBB-/BBB	7.96	86.74	6.267
ORCL 4.700% 27Sep2034 Corp (USD)	BBB-/BBB	8.19	88.72	6.303
ORCL 5.200% 26Sep2035 Corp (USD)	BBB-/BBB	9.18	90.49	6.414
ORCL 3.950% 25Mar2051 Corp (USD)	BBB-/BBB	24.69	60.54	7.259
ORCL 5.875% 26Sep2045 Corp (USD)	BBB-/BBB	19.19	82.13	7.520
ORCL 5.950% 26Sep2055 Corp (USD)	BBB-/BBB	29.20	79.35	7.594
ORCL 5.550% 06Feb2053 Corp (USD)	BBB-/BBB	26.56	75.25	7.598
Source：FSM Global Date as of 23 July 2026

Finally, although Oracle's bond rating of BBB− is still investment grade, it sits just one notch above non-investment grade, leaving limited buffer. Should it be downgraded to non-investment grade in the future, some funds that may only hold investment-grade bonds would be forced to sell, potentially pressuring bond prices again and worsening liquidity

Risk

While we consider Oracle's near-term debt-servicing position to be sound, its credit risk has risen markedly, and investors should remain mindful of several risks.

The most fundamental of these is the risk of customer concentration and a reversal in demand. Roughly half of Oracle's backlog is tied to OpenAI, which is not yet profitable and must continue to rely on external financing to fund its contractual payments. Should financing conditions deteriorate, the pace at which these orders are fulfilled would be directly affected. The previously mentioned buffer—that capacity could be quickly re-leased to other customers—itself rests on the assumption that compute remains in short supply. If AI demand reverses, re-leasing would become more difficult, and Oracle, being relatively more dependent on external customers, could be hit harder than peers that have their own in-house demand.

At the same time, Oracle's free cash flow is negative, and the capex gap must be filled through continued equity and debt issuance. Market demand is currently ample, but should the market turn more cautious toward AI-related issuers, both Oracle's build-out and its order fulfilment could be impeded, and financing costs would rise accordingly.

On the execution side, as Project Jupiter illustrates, even building its own power supply remains constrained by permitting and environmental hurdles. Meanwhile, roughly US$273 billion in data-centre lease and power-purchase commitments are largely non-cancellable, meaning Oracle must pay regardless of whether orders are fulfilled on schedule. In other words, if revenue is deferred while these obligations remain in place, the financial strain could prove greater than the headline figures suggest.

Disclosure: For specific disclosure, at the time of publication of this report, IFPL (via its connected and associated entities) holds a position in  ORCL 5.875% 26Sep2045 Corp (USD). The analyst who produced this report holds a NIL position in the abovementioned securities. This research report was prepared with the assistance of artificial intelligence (AI) tools. iFAST Financial Pte Ltd does not rely exclusively on AI for content generation; the content of this report – including all investment theses, ratings, price targets and conclusions – has been independently reviewed and verified by the research analyst(s) to ensure accuracy and professional integrity.