Heavily Indebted AI Dream — Is Oracle’s 7% Bond Yield an Investment Opportunity?

Throughout 2025, Oracle has continued to fund its artificial intelligence (AI) infrastructure largely through debt issuance, raising an additional USD 18 billion in late September to meet growing demand for its AI cloud computing services. The company now carries over USD 100 billion in total debt, coupled with an expected USD 50 billion in capital expenditure for the latest fiscal year, which has raised market concerns regarding Oracle’s credit risk. Its credit default swap (CDS) costs have climbed to their highest levels since the 2008 financial crisis.

Faced with market skepticism, Oracle’s 30- year bond yield near 7% prompts the question — is it a genuine investment opportunity? This report attempts to reveal the underlying story.

Company Background

Founded in 1977 by Larry Ellison and two co-founders, Oracle went public in 1986 on NASDAQ and later moved to the NYSE. As of January 8, its market capitalization stood at about USD 554 billion.

The company primarily provides cloud and software services to enterprises. Oracle’s cloud segment consists of IaaS (Infrastructure as a Service) and SaaS (Software as a Service). The former allows clients to use Oracle’s infrastructure for computing and storage to support their applications and AI models, while the latter directly provides cloud-based software solutions to customers.

In the first half of FY 2026 (June 2025 – May 2026), cloud services accounted for 49% of total revenue, followed by software services at 37% (see Figure 1).

Figure 1: Oracle’s FY2026 1H Revenue Distribution by SegmentsOracle’s total revenue for the first half of FY 2026 grew 13% YoY to USD 31 billion (see Figure 2). This was driven by a surge in enterprises’ demand for AI computing power, which led to nearly 33% growth in Oracle’s cloud services revenue, allowing it to overtake software services as the company’s largest revenue contributor. It is reasonable to expect the company to continue focusing on cloud services as its core growth driver, while software services, despite moderating growth, will still provide Oracle with stable revenue and cash flow.

Figure 2: Oracle’s Total Revenue

33% Growth in Cloud is Just a Start, The Real Value Lies in the Backlog

With enterprises racing to secure AI computing capacity, Oracle’s backlog value skyrocketed by 540% YoY to USD 523 billion in the first half of FY2026, bolstered by major contracts with Microsoft and OpenAI (see Figure 3). Management projects that 75% of these backlog orders will convert into revenue within 60 months, supporting average annual revenue growth of 31% through 2030 and potentially pushing total revenue above USD 100 billion by 2028 (see Figure 4).

Figure 3: Oracle’s Backlog

Figure 4: Oracle’s Revenue Expectation

“Fungible” Equipment Strengthens Resilience Under the Stargate Agreement

In September 2025, Oracle signed a USD 300 billion cloud computing agreement with OpenAI, known as the Stargate Project. The agreement accounts for over half of Oracle’s total backlog value. According to The Financial Times, OpenAI’s orders represent only 39% of Microsoft’s and 16% of Amazon’s respective backlog values, highlighting Oracle’s significantly higher exposure. While the contract underscores Oracle’s growth potential, it has also raised concerns about overexposure to OpenAI. Given the intense competition within the AI industry, there remains uncertainty as to whether OpenAI will be able to secure sufficient funding in the coming years to fully honour its contractual commitments with Oracle.

In fact, the data centre equipment specified in the contract is not designed exclusively for OpenAI, but is built on Oracle’s standard cloud service architecture, which means these assets are “Fungible”. Oracle CEO Clay Magouyrk has also mentioned that it only takes a few hours to reallocate servers from one customer to another, so even if OpenAI fails to fully perform under the contract, the company can still redeploy the completed data centres to serve other customers. This of course comes with the assumption that there will be sufficient takers for the undeployed data centres.

Under McKinsey’s most conservative scenario, demand for computing power is expected to grow 18% annually through 2030, implying that Oracle’s excess capacity would still likely be absorbed by other AI players even if OpenAI falters.

Over USD 100 Billion in Debt — The Next Three Years Are Critical

Oracle expects data centres for OpenAI to become operational between 2027 and 2028, marking a key inflection point for revenue growth. However, with capital expenditures already reaching USD 20.5 billion in the first half of fiscal 2026 and an upward revision of the full-year forecast by 40% to USD 50 billion, the company has now recorded negative free cash flow for three consecutive quarters (see Figure 5).

Figure 5: Oracle’s Free Cash Flow & Capital ExpenditureWith capital expenditures climbing, Oracle raised USD 15 billion in debt over six months, bringing total debt to over USD 100 billion (see Figure 6). At the same time, it is expected that the free cash outflow will persist for some time, and that the total amount of debt will continue to increase.

More recently, Blue Owl Capital, an asset management firm that has repeatedly invested in Oracle’s data centres, declined to participate in a USD 10 billion investment related to OpenAI, prompting the market to question the reliability and sustainability of Oracle’s funding chain. However, it is worth noting that most of the data centre construction will be completed by 2028, when capital expenditures are expected to peak and free cash outflows are likely to narrow or even turn into inflows, so whether the company can successfully navigate the next three years will be crucial.

Figure 6: Oracle’s Total Debt

Debt Maturity Well-Staggered; Management Aims to Preserve Stability

Despite carrying over USD 100 billion in total debt, only USD 20.7 billion is due within the next three years, with nearly 80% maturing after 2030 (see Figure 7). With USD 20 billion in cash reserves, Oracle’s near-term default risk is low, and liquidity appears sufficient to support its data centre projects until completion in 2028.

Figure 7: Oracle’s Debt Maturity ScheduleAmid deteriorating leverage, management is exploring alternatives beyond debt financing to support AI infrastructure investments — including joint ventures and partnership models.The company aims to keep its S&P Global Ratings-adjusted leverage below 4× (exceeding this threshold could trigger a downgrade, see Table 1) and maintain its BBB credit rating.

Table 1: Oracle's Financial Metric 

Bond Investment

Currently, there are seven U.S. dollar bonds issued by Oracle available our platform, including “ORCL 5.875% 26Sep2045 Corp (USD)”, which onboarded on our real-time trading platform Bondsupermart Live.

These bonds tenor range from 6 to 30 years, with yield to maturity as high as 6.9%.Oracle’s current issuer and bond credit ratings are both BBB / BBB (S&P / Fitch), which falls within the investment‑grade category.

Oracle’s bond yields are about 50 basis points higher than peers with similar ratings (see Figure 8). The 6- to 8-year bonds offering better risk-reward ratio and clearer earnings visibility.

Figure 8: Oracle’s Bond Yield Curve

The “ORCL 4.800% 26Sep2032” and “ORCL 4.700% 27Sep2034” bonds offer yields of 5.3% and 5.6%, respectively — suitable for investors comfortable with moderate credit risk (see Table 2). For investors who believe in the longer-term future in Oracle, they may also consider the “ORCL 5.875% 26Sep2045” bond which offers a yield of 6.7%.

Table 2: Oracle’s Bond

Risk

Although S&P maintains Oracle’s BBB rating, it assigned a negative outlook, citing rising leverage tied to AI infrastructure investment. S&P has raised its downgrade threshold for S&P Global Ratings-adjusted leverage from 3.5× to 4.0×, acknowledging Oracle’s strong AI growth potential. However, S&P expects the company’s S&P Global Ratings-adjusted leverage to exceed 4 times in FY2026, and despite management’s efforts to seek funding avenues for AI investment other than additional borrowing, the company still faces a risk of a credit rating downgrade.

Given AI-related contracts are underpinned by strong expected users’ demand growth, any slowdown in actual demand could undermine confidence in OpenAI and the broader AI sector, disrupt financing, and impair contract fulfilment. Although Oracle’s data centres are technically fungible, a broad market slowdown would still weigh on revenue realization and impact leverage and cashflows.

Conclusion

Benefiting from rapid AI industry growth, Oracle’s backlog orders have surged to USD 523 billion, with management expecting 75% of these orders to convert into revenue within the next 60 months, driving total revenue to grow at an average annual rate of 31% through 2030.

The company’s liquidity remains adequate, with a low probability of default over the next few years and sufficient capital to support the completion of its data centre construction by fiscal year 2028.

Oracle’s 6- to 8-year bonds currently yield around 5.2% to 5.6%, offering a reasonable risk-reward ratio that may appeal to investors who can tolerate a moderate level of risk.

Declaration: For specific disclosure, at the time of publication of this report, IFPL (via its connected and associated entities) holds position in ORCL 5.875% 26Sep2045 Corp (USD) and the analyst who produced this report hold a NIL position in the abovementioned securities.