Oracle (ORCL) Fundamentals: Databases, ERP and Cloud Capacity

Few large technology companies have changed shape as visibly as Oracle over the past two years. For most of its history the company was understood as a database and enterprise software business: high margins, predictable license and support renewals, and modest capital needs. Any serious Oracle stock analysis today has to reckon with something quite different. In fiscal 2026, which ended on 31 May 2026 and was reported on 10 June 2026, Oracle spent $55.7 billion on capital expenditure against total revenue of $67.4 billion, and its contracted future revenue ballooned to $638 billion. This article walks through what the company actually sells, what the latest reported numbers show, and which variables will determine whether the transformation delivers the returns the contracts imply.

How Oracle Makes Its Money

Oracle reports three businesses: cloud and software, hardware, and services. In practice, the revenue story sits inside the first of those, and it splits into three quite different economic engines.

• Legacy software. Database licenses and the annual support contracts attached to them. In fiscal 2026, software revenues were $24.5 billion, down 1% year over year. This is a mature, slow-moving base, but the renewal rates on support contracts have historically been high and the associated margins are the reason Oracle has been able to fund an aggressive build-out at all.
• Cloud applications (SaaS). Oracle's Fusion enterprise resource planning suite and NetSuite compete for the finance, HR, and supply chain workloads that companies run their operations on. SaaS revenue was $15.9 billion in fiscal 2026, up 11%. This is the part of the business that goes head to head with Salesforce (CRM) and SAP (SAP), and the growth rate has been steady rather than spectacular.
• Cloud infrastructure (IaaS). Oracle Cloud Infrastructure, or OCI, rents raw compute, storage, and networking, increasingly for training and serving AI models. IaaS revenue reached $18.1 billion in fiscal 2026, up 77%, and in the fourth quarter alone it was $5.8 billion, up 93%. This is where the company competes with Microsoft Azure (MSFT), Amazon Web Services (AMZN), and Google Cloud (GOOGL), and it is the reason the capital expenditure line looks the way it does.

Services revenue was $5.7 billion, up 10%, and hardware was $3.1 billion, up 5%. Together they are less than 14% of the total. The practical point for anyone reading a set of Oracle results is that the company now houses a slow-growing, cash-generative software franchise and a fast-growing, capital-hungry infrastructure business under one roof, and the reported totals blend the two.

Fiscal 2026 in Numbers: Revenue, Margins and Earnings

Total fiscal 2026 revenue was $67.4 billion, up 17% in reported dollars and 16% in constant currency. Cloud revenue across both IaaS and SaaS was $34.0 billion, up 39%, which means cloud crossed half of total revenue during the year. Fourth-quarter revenue was $19.2 billion, up 21%, with cloud revenue of $9.9 billion, up 47%.

On profitability, GAAP operating income for the full year was $20.6 billion, up 17%, which works out to an operating margin of roughly 31% of revenue. On a non-GAAP basis, which excludes stock-based compensation and acquisition-related amortization, operating income was $28.9 billion, up 16%, or about 43% of revenue. The gap between the two measures is large in absolute terms and worth understanding rather than skipping over: it is mostly stock compensation, which is a real cost to existing shareholders even though it does not consume cash in the period.

Earnings came through at $5.83 per share on a GAAP basis, up 34%, and $7.63 on a non-GAAP basis, up 27%. GAAP net income was $17.0 billion, up 36%. For the fourth quarter, GAAP earnings per share were $1.45 and non-GAAP earnings per share were $2.11.

One nuance matters when reading margins in a business like this. Renting infrastructure is structurally lower margin than licensing software, so as OCI grows faster than the software base, blended margins face a mix headwind even if each business is performing well on its own terms. The fact that operating margin held roughly flat while IaaS grew 77% tells you the software franchise is still doing heavy lifting on profitability.

The Backlog Question: What $638 Billion of RPO Represents

The single most discussed figure in any recent Oracle stock analysis is remaining performance obligations, or RPO. At the end of fiscal 2026 this stood at $638 billion, an increase of $85 billion in the fourth quarter alone from $553 billion three months earlier. For context, that backlog is roughly nine times the company's annual revenue.

RPO is a specific accounting concept, and it is easy to over-read. It measures contracted revenue that has not yet been recognized. It is not cash in the bank, it is not a forecast, and the timing over which it converts into revenue can stretch over many years. Nor is it entirely immune to change: contracts can be renegotiated, and a counterparty that cannot pay cannot be made to.

Two features of Oracle's backlog deserve particular attention. The first is customer concentration. Oracle announced a $300 billion, five-year cloud computing agreement with OpenAI in September 2025, with compute deliveries beginning in 2027. That single contract accounts for a very large share of the total, which means the durability of the backlog is bound up with the financial trajectory of one privately held counterparty rather than spread across thousands of enterprise customers in the way Oracle's support renewals are.

The second is how the hardware is being funded. Oracle has disclosed that the prepaid and customer-supplied hardware portions of its large AI contracts now total $75 billion. In some of these arrangements the customer pays up front for GPUs, and in others the customer buys the GPUs and supplies them to Oracle. That materially changes the capital profile of the deals compared with Oracle funding every chip itself, and it is a detail worth tracking in future disclosures because it determines how much of the build-out lands on Oracle's own balance sheet.

The Capital Expenditure Cycle and Free Cash Flow

This is where the tension in the story sits. Operating cash flow for fiscal 2026 was $32.0 billion, up 54% — a strong result by any measure. Capital expenditure was $55.7 billion, up from $21.2 billion the prior year, or roughly 2.6 times. The arithmetic consequence is that free cash flow was negative $23.7 billion for the year.

Guidance points to more of the same before it points to less. For fiscal 2027, Oracle has indicated gross capital expenditure of $90 billion to $95 billion, with a net cash outlay of approximately $70 billion after customer prepayments. Alongside that, the company raised roughly $43 billion of debt and $5 billion of equity during fiscal 2026 and has signaled further financing needs in fiscal 2027.

None of this is unusual for a company building physical capacity ahead of contracted demand — utilities and telecom operators do the same thing — but it changes what investors are underwriting. The relevant questions become mechanical rather than narrative. Does contracted revenue arrive on the schedule the contracts specify? Does the capacity get utilized at the rates the returns assume? And how much of the equipment cost does Oracle end up carrying itself versus having customers prepay or supply?

A useful discipline here is to separate maintenance capital expenditure from growth capital expenditure. Oracle's software business needs very little of the former. Almost all of the current spending is growth spending tied to specific contracts, which means it should in principle be assessable against those contracts rather than treated as a permanent drag. Whether that assessment holds up is exactly what the next several quarters of disclosure will reveal.

Balance Sheet, Dividend and What the Multiples Measure

Oracle pays a quarterly dividend of $0.50 per share, or $2.00 annualized. Measured against fiscal 2026 GAAP earnings of $5.83 per share, that is a payout ratio of roughly 34%; against non-GAAP earnings of $7.63, closer to 26%. Those are comfortable coverage ratios on an earnings basis. On a cash basis the picture is different, because with free cash flow negative the dividend is currently being funded from the same pool of capital that is financing the build-out.

Debt has risen substantially. Oracle entered fiscal 2026 already carrying significant borrowings from the Cerner acquisition and added roughly $43 billion more during the year. Interest expense is therefore a growing line item, and credit ratings and the cost of new issuance become variables that matter to equity holders in a way they did not when the company was a low-capital software business.

On valuation, the mechanics are worth stating plainly rather than reaching for a conclusion. A price-to-earnings ratio compares the share price to earnings per share, and which earnings figure you use changes the answer considerably here: fiscal 2026 GAAP earnings of $5.83, non-GAAP earnings of $7.63, and management's fiscal 2027 non-GAAP guidance of $8.05 on approximately $90 billion of revenue produce three quite different multiples from the same price. Free-cash-flow-based multiples cannot be calculated at all in a year when free cash flow is negative, which is why backlog and capacity metrics have taken on outsized importance in how the market discusses the company. Understanding which denominator a given headline multiple uses is more useful than the multiple itself.

Risks and What to Watch

Several specific things would change the picture, in either direction:

• Backlog conversion. RPO turning into recognized revenue on schedule is the central assumption. Watch the pace at which cloud infrastructure revenue accelerates against the backlog that has already been booked.
• Counterparty concentration. Because one customer represents a large share of the backlog, disclosures about that relationship — contract modifications, payment terms, delivery schedules — carry disproportionate weight.
• Capacity utilization and returns. Data centers are only profitable when full. Any commentary on utilization rates, or on gross margin within OCI specifically, is a direct read on whether the capital is earning its cost.
• Funding costs. With further debt issuance planned, the spread Oracle pays and any rating agency action are worth following, as is the level of shareholder dilution from equity raises.
• Competitive dynamics. Microsoft (MSFT), Amazon (AMZN), and Google (GOOGL) are all building at scale, and Nvidia (NVDA) supply allocation influences who can deploy capacity when. Oracle's differentiation claims are about price-performance and network architecture, and those are testable over time.
• The software base. Software revenue declined 1% in fiscal 2026. That base is funding the build-out, so its stability is not a side issue. Renewal rates and Fusion and NetSuite growth are the things to monitor.

Conclusion

The honest summary of this Oracle stock analysis is that the company has two businesses with opposite financial signatures, and the reported figures blend them. The software and applications franchise generated the $32.0 billion of operating cash flow and the 43% non-GAAP operating margin. The infrastructure business generated the 77% IaaS growth, the $638 billion backlog, the $55.7 billion of capital expenditure, and the negative $23.7 billion of free cash flow.

What makes Oracle genuinely difficult to assess is that the usual shortcuts do not work. Free cash flow yield is unavailable. Trailing earnings reflect a business mix that is actively changing. Backlog is enormous but concentrated and long-dated. That does not make the company unanalyzable — it means the analysis has to be built from the specific disclosures rather than from summary ratios.

For anyone following the name, the most informative data points in coming quarters will be the rate at which cloud infrastructure revenue converts the backlog, any change in the split between Oracle-funded and customer-funded hardware, OCI-level margin commentary, and whether the software base holds steady while the capital cycle runs. Those four things, rather than any single headline number, are what will determine how this transformation is ultimately judged.