CAPITAL THESIS

Independent equity research · ORCL · full edition

Oracle CorporationCloud growth is visible. The return on the buildout is still the investment test.

Oracle’s retained earnings scenarios show substantial upside, while capex, lease commitments and financing make cash returns the decisive test. New cash and funding exhibits keep those claims visible.

Research date: 15 October 2026 · USD common equity · Reference price: USD 130.55 · Observation: 2026-10-14

Watchlist — capital-return evidence incomplete

The investment decision

At $130.55 the retained base illustrates 24.1% annualised, versus 19.0% without P/E expansion. The September $82.97 cash DCF does not corroborate that earnings upside. The forecast vintage and funding uncertainty are central to interpreting the rank.

Q1 revenue$19.35bnQ1 FY2027 actual; S1
Q1 FCF−$5.40bnOCF less gross capex; S1/S2
Uncommenced leases$288bnNominal commitments, not PV debt; S2
No-rerating base19.0%Five-year retained model

Edition update. Reviewed Q1 FY2027 release/10-Q and current financial index; September 24 operating and financing scenarios retained explicitly. Recomputed returns from rounded published EPS at the October 14 close, added cash-quality and incremental-equity stresses. Original DCF not recomputed.

01. The enterprise franchise and the capital-intensive expansion

Oracle owns an established database and enterprise-application franchise while building a much larger cloud-infrastructure operation. The first business benefits from embedded workloads, operational risk in migrations and long customer relationships. The second must earn returns on physical capacity, power, hardware and financing. The investment debate is whether the combination creates attractive cash returns per existing share, rather than simply impressive growth in total revenue.

The Q1 FY2027 release reports $19.345bn revenue, including $7.39bn infrastructure-as-a-service revenue growing 121%. Cloud applications grew 10%. Management’s FY2027 anchors remain at least $90bn revenue and $8.10 adjusted EPS. A software installed base can improve distribution into cloud workloads, but the economics of a large GPU contract are not those of an incremental software licence. [S1]

Our retained base scenario values FY2032 adjusted earnings at 20 times and produces an attractive modeled return at the lower October price. Yet the September owner-cash DCF was only $82.97 per share. We preserve that disagreement instead of averaging the methods into an apparently precise target. The market needs to believe the capacity earns durable returns after rent, refresh spending and financing; adjusted earnings alone do not settle that question. [S5]

This remains a watchlist case with an unresolved capital-return test. It could become stronger if delivered capacity converts into cash with less incremental Oracle-funded capital. It weakens if customer commitments require repeated equity funding while the return on installed assets disappoints. A cheaper price improves the entry arithmetic but does not repair those business conditions.

02. Backlog is an execution obligation as well as an asset

The filed RPO balance is $664bn, with 13% expected to be recognised over the next twelve months: approximately $86.3bn by multiplication. That is a revenue timing estimate, not cash available to shareholders. Contracts still require facilities, equipment, customer acceptance and capable counterparties. Delivery can be delayed even where the long-run demand is real. [S2]

Oracle’s multicloud database strategy can capture value without forcing every customer to move all computing to Oracle. Integrated applications and database expertise offer advantages in complex enterprise workloads. The counterargument is that large infrastructure customers can negotiate pricing, fund equipment themselves or shift capacity across suppliers. For the marginal deployment, power availability, utilisation and hardware life may matter more than brand recognition.

A professional comparison should separate workload mix, cloud-service definitions and financing capacity. We do not transplant a hyperscaler P/E or a faster cloud-growth percentage onto Oracle. Nor do we treat rival reported EPS containing investment gains as a clean earnings denominator. Growth, margin and cash conversion must be evaluated together.

The most useful demand evidence is therefore cohort-level: how much of a contracted deployment is live, how quickly it is utilised, who funds the equipment, how the customer pays and what return remains after its refresh cycle. Contract-level cancellation terms, customer concentration by economic exposure and capacity-cohort returns are not disclosed sufficiently in the reviewed material to quantify them. That is a real underwriting gap, not a reason to assign an invented default probability.

03. The cash-flow bridge prevents a common valuation error

USD bn; quarterly actualQ1 FY2026Q1 FY2027
Operating cash flow8.14023.103
Gross cash capex8.50228.499
Conventional FCF-0.362-5.396
Customer prepayments with financing componentNot used in this comparison11.363
FY27 FCF excluding these inflowsNot calculated-16.759

The latest filed quarter produced substantially more operating cash, but the financing-like customer prepayment inflow explains an important part of the improvement. Removing that inflow before deducting gross capex gives a tougher cash screen. It is not a normalized annual run rate: ordinary collections, taxes and working capital also move between quarters. [S1][S2]

The accounting distinction matters because customer-funded construction can be commercially helpful without being free cash profit. The customer may have prepaid for services the business must still deliver. Analysts must not deduct a capex measure already net of funding from operating cash that still includes the same prepayment benefit; doing so double-counts that funding.

The exhibits show gross capex and cash flow on matching periods. They deliberately retain negative values. EBITDA and adjusted EPS can rise while free cash flow remains negative because the equipment bill is paid before its cost fully reaches the income statement. Longer assumed useful lives can improve reported earnings without changing that initial bill. The economic test is lifetime revenue and cash margin after replacement, not only this quarter’s depreciation charge.

Reported FCF does not become a shareholder yield merely because a company continues to pay dividends. Those distributions may coexist with financing needs. In the scenario model, dividends are included once in terminal wealth; the associated funding pressure is part of the retained financing assumptions.

04. Debt, leases and the ownership claim

At the August balance-sheet date, financial debt including finance leases was approximately $134.5bn and unrestricted cash/securities $37.1bn. The resulting $97.4bn net financial debt is only one part of the capital claim. Existing operating lease liabilities were approximately $34.6bn and uncommenced nominal lease commitments $288bn, mostly data centres with terms of fifteen to nineteen years. [S2]

The $288bn figure is not a present-value debt number to deduct mechanically from equity value. Future rent belongs in the operating cash-flow forecast; a lease-capitalised valuation would instead need consistent changes to earnings and the discount rate. Subtracting the full nominal amount while keeping rent in margins would charge the same obligation twice. Conversely, ignoring uncommenced leases because they are off the balance sheet understates future operating commitments.

The $20bn gross ATM programme had been completed by the latest filing. It cannot also be shown as remaining financing capacity. The earlier calendar-2026 financing announcement is background; the subsequent filing governs the actual position. Additional base funding of $28.5bn in our retained model is a house estimate after Q1, not a new company-authorised raise. [S2][S4][S5]

The September model included future borrowing, net award dilution, preferred conversion and zero buybacks. Its base diluted share count reaches 3.463bn. We preserve these assumptions visibly because growth funded with new shares does not all accrue to today’s holders. The downloadable calculator replays disclosed rounded EPS and shares; it does not claim to reconstruct the original full financing recursion or every asset and lease vintage. Investors should not mistake mathematical reproducibility of scenario returns for completion of contract and financing diligence.

05. The inherited cash path is demanding

Base house forecastFY2027FY2028FY2029FY2030FY2031FY2032
Revenue, $bn90119.6152.2183.9209.2229.6
Adjusted EPS, $8.109.5410.9913.6616.5018.74
Gross cash capex, $bn92.510085706565
Reported-like FCF, $bn-27.9-27.2-12.97.824.838.2

Only the initial FY2027 revenue/EPS anchors refer to management guidance. The cash path and all later years are September 24 Capital Thesis assumptions retained after reviewing the latest primary releases through 15 October. No later earnings release was identified in the company financial index. The model is not relabelled as a freshly re-underwritten six-year forecast. [S1][S5][S8]

The base requires heavy spending for several years before positive cash returns emerge. The retained model grows infrastructure revenue from an assumed $40bn in FY2027 to $169.4bn in FY2032, while enterprise applications grow more slowly and legacy software stays broadly flat. Gross capex peaks at $100bn and then declines to $65bn. Rent is included before operating profit. These assumptions require both successful demand delivery and improving capital productivity.

There are two ways this path can fail despite strong bookings. Capacity may arrive late, leaving financing and rent costs ahead of recognised revenue. Alternatively, hardware replacement and power costs may keep reinvestment high after the revenue ramp. Neither failure is fixed by an adjusted earnings multiple. The cash-positive date is an analytical expectation rather than a company commitment, so delays would require a fresh funding and dilution model.

06. A wide return range, with financing already embedded

House caseFY32 sales, $bnFY32 diluted shares, bnFY32 adjusted EPSExit P/E5Y wealth CAGR
Bear134.83.969$5.4914x-7.8%
Base229.63.463$18.7420x24.1%
Bull322.43.311$32.1524x43.0%

Bear case: five-year total-wealth CAGR is -7.8%, using FY2032 forward adjusted EPS of $5.49, a 14x exit P/E, $76.86 terminal share value and $10.00 cumulative cash dividends from a $130.55 entry.

Base case: five-year total-wealth CAGR is 24.1%, using FY2032 forward adjusted EPS of $18.74, a 20x exit P/E, $374.80 terminal share value and $10.00 cumulative cash dividends from a $130.55 entry.

Bull case: five-year total-wealth CAGR is 43.0%, using FY2032 forward adjusted EPS of $32.15, a 24x exit P/E, $771.60 terminal share value and $10.00 cumulative cash dividends from a $130.55 entry.

No-rerating base case: five-year total-wealth CAGR is 19.0%, holding the current forward adjusted P/E of 16.12x, using $18.74 terminal EPS and $10.00 cumulative cash dividends.

The horizon is a five-year annual proxy, with an autumn-2031 exit valued on FY2032 forward earnings. Total-wealth CAGR equals ((exit price plus accumulated dividends) divided by entry price)^(1/5) minus one. Dividends accumulate at zero interest; this is not a cash-flow IRR. Assumed $2 annual distributions are not guaranteed. No scenario probabilities or broker targets are embedded.

The October calculation multiplies the previously published rounded EPS by each exit multiple, giving small differences from earlier unrounded terminal prices. The old base adjusted operating margin of 38%, financing assumptions and share count are unchanged. This edition improves evidence, presentation, stress testing and entry arithmetic; the unresolved cash-return gate remains. The bear combines lower realised economics and more dilution, but is not a bankruptcy or maximum-loss case.

07. A hurdle price is not a cash valuation

The conditional 20% entry ceiling is $154.64, calculated as ($374.80 terminal share value + $10.00 dividends) / 1.20^5. It holds September operating, financing, issuance-price and share-count assumptions fixed and requires a 20x terminal adjusted P/E; it is not a cash-validated fair value.

From $130.55, 20% annualised requires $324.85 ending wealth. After the $10 cash-dividend assumption, that is $314.85 of equity value, or $15.74 EPS at 20 times. The base EPS needs 16.80 times at exit. These figures expose which part of the thesis is operating growth and which part is valuation.

The original $82.97 owner-cash DCF remains a September-vintage caution, not a newly calculated target. It used a 10.5% capital cost, 3% mature growth, after-SBC cash costs, finance-lease reinvestment and a transition to mature economics. Approximately 78% of enterprise value came from the terminal period. The current replay calculator does not rebuild that DCF, so it is excluded from any claim of fresh numerical validation. [S5]

PEG is not the deciding metric. The earnings growth denominator changes substantially with the treatment of investment gains and adjusted costs, while neither P/E nor PEG pays the capex bill. The website’s current-price explorer changes the investor’s entry denominator only; it does not assume future equity can be issued at each selected entry price or rewrite the financing schedule.

08. What another funding need does to the equity

An illustrative additional $10bn common raise at $130.55 would add 76.6m shares beyond the retained base. Holding terminal profit fixed, EPS falls from $18.74 to $18.33, and the modeled return falls from 24.1% to 23.6%. This is a dilution-only stress: no incremental earnings benefit or issuance fees are included. It is not a prediction of a financing announcement.

The modest change in this isolated stress should not be mistaken for immunity to funding risk. A larger capital need can coincide with weaker margins, a lower issuance price and a lower exit multiple. These effects reinforce one another. A proper re-underwrite would link the cause of the cash shortfall to revenue, debt capacity and the equity price, instead of treating every financing shock as a standalone share-count adjustment.

Monitoring gateRequired evidenceRe-underwrite trigger
DeliveryRevenue against at least $90bn FY2027 guidance and genuine RPO conversionCapacity delays or collections falling behind recognition
Capital productivityImproved cash returns after rent, power and hardware refreshGross capex above the retained $90–95bn FY27 range without better economics
FundingFinancing access and no material excess over modeled equity needsAdditional issuance or borrowing beyond the base without incremental value
Cash inflectionCore FCF improves toward the model’s FY2030 inflectionPositive FCF pushed out by persistent maintenance and lease costs

Oracle AI World is scheduled for 25–28 October. It is a useful disclosure window, but this edition does not assume a confirmed new financial target or analyst-day forecast. Product announcements matter only when they improve durable unit economics. [S7]

09. Source boundaries and investor use

The research date is 15 October 2026; reported financials stop at 31 August, and the reference quote is the 14 October regular close. Primary filing figures take precedence over promotional descriptions of net construction outlay. Historical financing plans are not presented as unused commitments. The old report remains accessible so changes in price, labels and scenario calculations can be audited.

Current institutional flows, short interest and options positioning were not independently verified. We make no crowding, insider-timing or squeeze inference. Important missing diligence includes customer-specific funding capacity, cancellation terms, annual uncommenced lease schedules, maintenance capex, prepayment runoff and capacity-cohort cash returns. Exact useful-life assumptions and financing settlement timing could materially change the equity outcome.

The work is independent public-source analysis, with house scenarios and no assumed probability distribution. Returns are nominal US-dollar total-wealth CAGRs before taxes, fees and foreign exchange. Positive arithmetic does not establish portfolio suitability. The stance stays watchlist until the capital-return evidence supports the earnings story or the entry valuation more fully compensates for the unresolved risks.

Financial exhibits

Exact observations accompany every chart. Actuals, management guidance and Capital Thesis assumptions are labelled separately; the chart shapes do not imply probabilities.

Chart 01 · USD bn

Operating cash rose, but gross capex absorbed more

Customer financing helped OCF; conventional FCF remained negative.

Operating cash flowGross cash capexReported FCF
Operating cash rose, but gross capex absorbed more (USD bn). Exact figures in the table below.-100102030Q1 FY2026Operating cash flow: 8.14 USD bn8.14Gross cash capex: 8.502 USD bn8.502Reported FCF: -0.362 USD bn-0.362Q1 FY2027Operating cash flow: 23.103 USD bn23.103Gross cash capex: 28.499 USD bn28.499Reported FCF: -5.396 USD bn-5.396
Exact values · USD bn
ObservationOperating cash flowGross cash capexReported FCF
Q1 FY20268.148.502-0.362
Q1 FY202723.10328.499-5.396

Quarterly reported actuals, not annualised. OCF includes financing-component prepayments. Sources: [S1] [S2]

Chart 02 · USD bn

Separate customer financing from the underlying cash screen

Removing $11.363bn of financing-like prepayments changes the quarter’s funding picture materially.

Q1 FY2027 cash measure
Separate customer financing from the underlying cash screen (USD bn). Exact figures in the table below.-20-100102030Reported OCFQ1 FY2027 cash measure: 23.103 USD bn23.103OCF ex prepaymentsQ1 FY2027 cash measure: 11.74 USD bn11.74Gross capexQ1 FY2027 cash measure: 28.499 USD bn28.499FCF ex prepaymentsQ1 FY2027 cash measure: -16.759 USD bn-16.759
Exact values · USD bn
ObservationQ1 FY2027 cash measure
Reported OCF23.103
OCF ex prepayments11.74
Gross capex28.499
FCF ex prepayments-16.759

Analytical bridge, not normalized FCF; each bar is a separate measure, not additive. Gross capex is deducted once. Sources: [S1] [S2]

Chart 03 · USD bn

The retained base needs several years of capital before cash

Positive cash depends on capex falling after the capacity ramp.

Gross cash capexReported-like FCF
The retained base needs several years of capital before cash (USD bn). Exact figures in the table below.-50-250255075100FY2027EGross cash capex: 92.5 USD bn92.5Reported-like FCF: -27.9 USD bn-27.9FY2028EGross cash capex: 100 USD bn100Reported-like FCF: -27.2 USD bn-27.2FY2029EGross cash capex: 85 USD bn85Reported-like FCF: -12.9 USD bn-12.9FY2030EGross cash capex: 70 USD bn70Reported-like FCF: 7.8 USD bn7.8FY2031EGross cash capex: 65 USD bn65Reported-like FCF: 24.8 USD bn24.8FY2032EGross cash capex: 65 USD bn65Reported-like FCF: 38.2 USD bn38.2
Exact values · USD bn
ObservationGross cash capexReported-like FCF
FY2027E92.5-27.9
FY2028E100-27.2
FY2029E85-12.9
FY2030E707.8
FY2031E6524.8
FY2032E6538.2

September 24 house model retained and labelled; not updated company cash guidance. Sources: [S5]

Source register and research basis

Evidence checked for this edition on 15 October 2026. Dates below identify source vintages. Company guidance and house scenarios are conditional; source links and the downloadable model allow the assumptions to be inspected.