ORACLE'S $638 BILLION QUESTION: WHEN DOES BACKLOG BECOME CASH?Oracle CorporationBATS:ORCLTrade-TechniqueApplying the complete operating-metric-to-present-value framework to Oracle's AI-cloud buildout Oracle may be one of the most interesting valuation tests in the US market today. At the end of fiscal 2026, the company reported $638 billion of remaining performance obligations, or RPO. Revenue reached $67.4 billion, cloud revenue grew 39%, and Oracle Cloud Infrastructure revenue grew 77%. At the same time, Oracle spent $55.7 billion on capital expenditure, generated roughly $32.0 billion of operating cash flow, and reported negative free cash flow of about $23.7 billion. That combination creates a perfect battlefield for two opposing stories. The optimistic story says Oracle has secured years of AI demand and is building the capacity needed to serve it. The pessimistic story says the company is borrowing heavily to build expensive infrastructure while the economic return remains uncertain. Both stories contain some truth. Neither is a valuation. A contract is not revenue. Revenue is not operating profit. Operating profit is not cash flow. And future cash flow is not worth its undiscounted face value. The only useful way to analyse Oracle is to follow the complete chain: Operating metric → Revenue → Operating profit → Taxes → Reinvestment → Free cash flow → Risk and probability → Present value This article applies that framework fully. It does not produce a buy or sell call, and it does not treat framework suitability as investment quality. It turns the Oracle thesis into a set of claims that can be measured quarter by quarter. 1. WHY ORACLE FITS THIS FRAMEWORK SO WELL In a broader screen of 60 US stocks, Oracle received the highest framework-suitability score. That did not mean it was the cheapest or best stock. It meant every stage of the valuation chain mattered unusually strongly. There is a large disclosed operating metric: $638 billion of RPO. The metric has an unusually long recognition schedule. Cloud revenue is already growing rapidly. Infrastructure costs are changing the margin mix. Capital expenditure is larger than operating cash flow. Customer funding, debt and equity financing materially affect the cash bridge. The final result depends heavily on execution timing and present value. This makes Oracle a useful case study because lazy valuation shortcuts fail very quickly here. 2. OPERATING METRIC: WHAT DOES $638 BILLION OF RPO ACTUALLY MEAN? RPO represents contracted revenue that Oracle expects to recognise in the future from obligations that have not yet been fully satisfied. It is evidence of committed demand, but it is not a bank balance and it is not the same thing as immediately collectible cash. Oracle disclosed the following expected recognition schedule at May 31, 2026: Next 12 months: 12% — approximately $76.6 billion Months 13–36: 34% — approximately $216.9 billion Months 37–60: 34% — approximately $216.9 billion After 60 months: 20% — approximately $127.6 billion Here is the important human comment: the headline is $638 billion, but only 12% is expected to be recognised within the next year. In other words, 88% sits beyond the first 12 months, and 54% sits beyond the first three years. That does not make the backlog weak. It makes timing central to the valuation. RPO rose from $138 billion a year earlier to $638 billion, an increase of 363%. Oracle also added $85 billion sequentially during the fourth quarter. Those are extraordinary figures. Yet the quality of RPO cannot be judged from its size alone. Investors also need to ask: How quickly does contracted capacity become available? How much revenue is recognised from each new data-centre cohort? What operating margin does that revenue earn? How much additional capital is required before Oracle can deliver it? How much of the obligation is prepaid or supported by customer-supplied equipment? What happens if deployment dates move or customer needs change? Oracle's RPO-to-revenue ratio is approximately 9.5 times, calculated as $638 billion divided by FY2026 revenue of $67.4 billion. That ratio signals exceptional future coverage. It does not say that nine and a half years of current profit are already secured. Framework conclusion at this stage: RPO is a strong demand indicator, but its valuation relevance depends on conversion speed, unit economics and required reinvestment. 3. REVENUE: IS THE OPERATING METRIC ALREADY CONVERTING? There is real conversion underway. FY2026 total revenue rose 17% to $67.4 billion. Cloud revenue increased 39% to $34.0 billion, representing slightly more than half of company revenue. Within cloud, infrastructure revenue reached $18.1 billion and grew 77%. Fourth-quarter OCI growth accelerated to 93%. The rest of the company provides useful context: Software revenue: $24.5 billion, down 1% Services revenue: $5.7 billion, up 10% Hardware revenue: $3.1 billion, up 5% Oracle is therefore not a clean, single-product AI company. It is a mature enterprise-software business being reshaped by a fast-growing, capital-intensive infrastructure operation. The legacy businesses can supply revenue, customer relationships and cash, while OCI becomes a larger part of the total mix. For FY2027, management guided to approximately $90 billion of total revenue. Achieving that figure would mean growth of about 33.5% from FY2026. Oracle also guided first-quarter FY2027 revenue growth of 27% to 29%, with cloud revenue growth of 58% to 64% in US-dollar terms. This gives the market something better than a vague AI promise: a near-term conversion test. The relevant question is not whether Oracle can sign large contracts. It is whether completed capacity can turn those contracts into recognised, profitable revenue at the guided pace. The revenue stage is currently passing the framework. OCI growth is visible and accelerating. The harder stages begin after revenue. 4. OPERATING PROFIT: RAPID GROWTH DOES NOT GUARANTEE LEGACY MARGINS Oracle generated $20.6 billion of GAAP operating income in FY2026, up 17%. On $67.4 billion of revenue, that implies a GAAP operating margin of approximately 30.6%. Non-GAAP operating income was $28.9 billion. Those figures show a highly profitable company at the consolidated level. But applying the current company-wide margin directly to the entire RPO would be a serious modelling error. Why? Because the revenue mix is changing. Oracle disclosed that cloud and software expenses increased by $5.8 billion, including a $5.7 billion increase in infrastructure expenses. It also said margins as a percentage of revenue declined because of higher infrastructure expenses. AI infrastructure requires servers, networking, energy, buildings, depreciation, operations and financing. The economics are different from selling another licence or renewing high-margin support. A proper model should separate at least three economic buckets: Mature software and support: slower growth, high recurring margins and lower incremental capital intensity. Cloud applications: recurring revenue with meaningful infrastructure needs but software-like customer economics. OCI and AI capacity: very high growth, much heavier capital requirements and margins that depend on utilisation, contract pricing and depreciation. There is also a timing issue. A new data centre may generate large costs before it reaches efficient utilisation. Early reported margins can therefore understate mature economics. But the reverse shortcut is equally dangerous: assuming every new facility will automatically achieve attractive utilisation and margins. The practical question is whether incremental cloud operating profit grows faster than the depreciation, infrastructure expense and financing cost attached to the new capacity. What would strengthen the thesis? OCI revenue remaining strong while consolidated or cloud-related margins stabilise and eventually expand. What would weaken it? Revenue meeting guidance while infrastructure expense keeps absorbing an increasing share of the revenue. 5. TAXES: THE BRIDGE MUST USE AFTER-TAX OPERATING ECONOMICS Oracle reported $17.0 billion of GAAP net income available to common shareholders in FY2026, up 36%, and GAAP diluted EPS of $5.83. Net income is useful, but it is not the cleanest starting point for valuing the operating assets. It already includes interest, tax effects and other items associated with the capital structure. For a full enterprise DCF, the cleaner route is: Revenue × normalised operating margin = operating profit Operating profit × (1 − normalised cash tax rate) = after-tax operating profit The word normalised matters. A modeller should not assume one reported tax line will remain constant throughout a multi-year AI buildout. Geographic profit mix, tax incentives, timing differences, stock compensation and the financing structure can move the reported rate. The conservative method is to scenario-test a sustainable cash tax rate and keep it separate from interest expense. This prevents the model from accidentally treating tax timing benefits as permanent operating value. 6. REINVESTMENT: THE NUMBER THE HEADLINE STORY CANNOT IGNORE Oracle's FY2026 cash-flow bridge was dramatic: Operating cash flow: approximately $32.0 billion Capital expenditure: approximately $55.7 billion Free cash flow: approximately negative $23.7 billion The arithmetic is straightforward: $32.0B operating cash flow − $55.7B capital expenditure = −$23.7B free cash flow Capital expenditure was approximately 82.6% of revenue and 174.1% of operating cash flow. It rose 162% from $21.2 billion in FY2025. Oracle also expects the upward capital-expenditure trend to continue in FY2027 and subsequent years. Calling all of this spending “bad” would be simplistic. Growth capex can create enormous value when it is tied to contracted demand and earns an attractive return. Calling it irrelevant because it supports growth would be just as simplistic. The cash has left the business either way. The real analytical task is to separate: Maintenance capex: spending required to sustain existing capacity and service quality. Committed growth capex: spending needed to deliver already contracted RPO. Speculative growth capex: capacity built in anticipation of future demand not yet economically secured. Oracle does not provide a neat public split for these categories. That means investors should track data-centre cohorts: capital deployed, timing to activation, revenue ramp, utilisation and eventual cash return. Negative free cash flow during a major buildout is not automatically value destruction. It becomes value destruction if the future after-tax cash generated by the assets fails to compensate for the capital and risk taken today. 7. CUSTOMER FUNDING, DEBT AND DILUTION: WHO PAYS BEFORE THE CASH ARRIVES? Oracle's disclosures contain a meaningful positive offset. The company said approximately $75 billion of hardware associated with large AI contracts is either prepaid by customers or supplied by them. This is valuable because it can reduce Oracle's funding burden. But the wording must be handled carefully. It does not mean Oracle received $75 billion in cash. The actual disclosed customer prepayments received in FY2026 that contained a significant financing component were approximately $4.6 billion. Those prepayments helped working capital and operating cash flow. The distinction is essential: $75 billion: hardware portions expected to be prepaid or customer-supplied. $4.6 billion: qualifying customer prepayments actually received during FY2026. Customer funding lowers risk only to the extent that it is collected, contractually durable and matched to the required deployment. It should be tracked as a separate thesis variable, not used as a substitute for future free cash flow. The remaining funding need is large. Oracle reported $129.5 billion of outstanding indebtedness at May 31, 2026. During FY2026 it raised roughly $43 billion of debt and $5 billion through mandatory convertible preferred stock. It also used short-term capex financing and continued scheduled debt repayments. For FY2027, Oracle expects to raise approximately $40 billion through a combination of debt and equity, including a previously announced $20 billion at-the-market equity programme. The company also said it did not expect additional debt issuance during calendar 2026 after its financing actions. This creates two costs that operating-revenue headlines do not capture: Interest and refinancing risk: more debt creates fixed claims on future cash. Dilution risk: equity and convertible securities spread future value across more common-share equivalents. Oracle reported that potential dilution from stock awards granted since June 1, 2023 was approximately 1.0% on an annualised basis. Planned equity and convertible financing add another reason to track the fully diluted share count, not only EPS growth. Framework conclusion at this stage: customer-supported hardware improves the funding picture, but debt, interest and dilution must be charged against the value created by the cloud buildout. 8. FREE CASH FLOW: WHAT THE CURRENT DEFICIT DOES AND DOES NOT TELL US FY2026 free cash flow was approximately negative $23.7 billion. That is a real economic fact, but it is not a complete verdict. During the build phase, current free cash flow is deliberately depressed by assets that may support revenue for years. At the same time, FY2026 operating cash flow benefited from the $4.6 billion of customer prepayments described above. A model should therefore avoid assuming either the current capex intensity or the working-capital benefit continues forever. The cash thesis requires an eventual transition: Stage 1: Contracting — RPO grows. Stage 2: Construction — capex and financing rise before revenue. Stage 3: Activation — capacity comes online and RPO converts to revenue. Stage 4: Utilisation — revenue grows faster than incremental operating cost. Stage 5: Harvest — operating cash flow exceeds maintenance and necessary growth capex, allowing free cash flow to turn sustainably positive. Oracle has clearly passed Stage 1 and is deep in Stages 2 and 3. Parts of the business may already be entering Stage 4. The market's central disagreement is about the timing and magnitude of Stage 5. 9. RISK AND PROBABILITY: THREE POSSIBLE ORACLE PATHS Instead of pretending there is one certain forecast, the framework should use scenarios. Scenario A — Efficient conversion FY2027 revenue lands near the $90 billion guide. OCI growth remains strong as new capacity becomes available. RPO converts broadly on the disclosed schedule. Customer-prepaid or customer-supplied hardware reduces Oracle's cash burden. Utilisation rises and infrastructure margins improve. Capex growth eventually slows relative to operating cash flow. Free cash flow inflects upward before financing costs overwhelm the benefit. In this outcome, current negative free cash flow represents a front-loaded investment in contracted demand. Scenario B — Delayed conversion Demand remains real, but site completion, power availability or hardware deployment slows recognition. Revenue is pushed further into the future. Capex, depreciation and interest arrive before the associated cash flow. Oracle raises additional capital or carries negative free cash flow longer than expected. The contracts may still be honoured, but present value falls because cash arrives later and financing costs accumulate. Scenario C — Weak economics or impairment Some customers reduce, renegotiate or delay their usage. Pricing or utilisation is weaker than expected. Infrastructure margins remain structurally below the level implied by the valuation. Technology changes shorten useful asset lives or force additional spending. The company absorbs debt, dilution or asset impairments without receiving sufficient cash return. In this outcome, impressive RPO and revenue growth fail to translate into attractive shareholder cash flow. Probability is not a decorative percentage added at the end. It should be linked to observable evidence: recognition rates, utilisation, margin progression, actual customer funding, construction timing and financing needs. 10. PRESENT VALUE: WHY $638 BILLION CANNOT BE ADDED TO EQUITY VALUE Even if every dollar of RPO converts to revenue, those dollars arrive over several years. Future money must be discounted. Consider a simple timing illustration using Oracle's recognition buckets. Assume the revenue arrives, on average, at 0.5 years, 2 years, 4 years and 6 years for the four disclosed buckets. At an 8% discount rate, the approximate present value of the gross RPO revenue is: $638B × 78.3% ≈ $499.5B At a 10% discount rate, it is approximately: $638B × 74.1% ≈ $472.4B These figures are deliberately not fair values. They are not estimates of Oracle's enterprise value or equity value. They simply demonstrate the effect of timing on the revenue headline. From that discounted gross revenue, a real valuation would still need to deduct or account for: Operating expenses required to provide the service Cash taxes Maintenance and growth capital expenditure Working-capital requirements Interest and other financing costs where appropriate Net debt and senior claims when moving from enterprise value to equity value Dilution from stock compensation, preferred conversion and equity issuance Execution risk and the probability of each scenario The value or burden of assets after the contracted period The correct DCF engine is therefore: After-tax operating profit + non-cash charges − capital expenditure − change in working capital = unlevered free cash flow Then discount those annual cash flows at a rate reflecting business risk, add the present value of a conservatively estimated terminal cash flow, and reconcile debt and dilution to reach common-equity value. Without reliable assumptions for OCI mature margins, maintenance capex, RPO conversion, customer funding and terminal reinvestment, a precise price target would create false confidence. The framework is more useful when it shows which missing variables control the outcome. 11. THE PRACTICAL ORACLE DASHBOARD: WHAT TO CHECK EACH QUARTER An Oracle investor does not need to react to every headline. These eight measurements cover most of the thesis: RPO and its near-term recognition percentage: Is the backlog growing, and is more of it moving closer to revenue? OCI revenue growth: Is installed capacity converting into recognised demand? FY2027 revenue-guide progress: Is Oracle tracking toward approximately $90 billion? Operating margin and infrastructure expense: Is incremental cloud revenue becoming more efficient? Operating cash flow versus capex: Is the funding gap narrowing or widening? Actual customer prepayments and customer-supplied equipment: How much construction burden is genuinely transferred? Debt, interest cost and refinancing: How much future cash is already claimed by lenders? Fully diluted shares: Is business value growing faster than each shareholder's ownership is being diluted? One additional metric would be extremely valuable if Oracle expands its disclosure: revenue and cash return by data-centre cohort. That would allow the market to compare construction cost, activation date, utilisation and mature economics directly. 12. THE FALSIFIABLE ORACLE THESIS Here is the full thesis in one sentence: Oracle's $638 billion RPO creates shareholder value only if recognised AI-cloud revenue and after-tax operating cash eventually grow faster than capital expenditure, interest obligations and the diluted share count. That statement can be tested. The thesis strengthens if RPO converts on schedule, OCI revenue meets guidance, margins improve with utilisation, customer funding materialises, and operating cash flow begins to catch capex. The thesis weakens if the recognition schedule moves outward, infrastructure costs prevent margin improvement, capex stays structurally above operating cash flow, or financing and dilution grow faster than the cash-generation capacity. FINAL VIEW: COUNT THE CASH, NOT ONLY THE CONTRACTS Oracle's situation is neither a simple bubble story nor a simple backlog story. The bullish shortcut is to treat $638 billion of RPO as if it were already high-margin cash. The bearish shortcut is to treat every dollar of current capex as waste before the assets have had time to earn revenue. A serious analysis must live between those extremes. Oracle has demonstrated extraordinary contracted demand and very strong OCI revenue growth. It also faces extraordinary reinvestment, substantial debt, planned equity financing and a long-dated conversion schedule. The final shareholder result will be determined not by the biggest number in the press release, but by the sequence underneath it. Operating metric → Revenue → Operating profit → Taxes → Reinvestment → Free cash flow → Risk and probability → Present value That is the whole framework. For Oracle, every arrow matters. SOURCE OF FRAMEWORK: