ORCL: Who Funds the $23 Billion Cash Quarter?

The 60-second read

Five things to know

1. The operating beat is real. Revenue and adjusted operating profit both grow about 30%.
2. Cash timing is material. Financing-component customer prepayments supply nearly half of operating cash.
3. Funding is not asset cost. The net-capex-outlay measure cannot replace gross capex in FCF without double-counting prepayments.
4. Ownership changes. The ATM is completed and the preferred will convert; coupons and dilution require consistent treatment.
5. Valuation depends on payback. Earnings routes are near $150, but the cash-transition model is much lower.

Three scenarios, one probability-weighted number

Analyst twelve-month scenarios; probabilities are judgments.

Signposts: what would change our mind

Expected review months are not confirmed release dates. Cash timing measures remain distinct from reported FCF and company guidance.

The tape

A positive early reaction does not settle who earns the return on expansion.

Oracle is a SELL with conviction 3/5 and High uncertainty at the saved $160.97 after-market observation. Our central fair value is $130, with a deliberately broad $70–$240 scenario range. Revenue growth and operating execution deserve credit; the concern is the price paid for cash that must emerge after a very large infrastructure programme. The three valuation routes disagree materially, and that disagreement is useful evidence about the unresolved investment question. The call concerns entry value.

September 10 closed at $153.17. The saved 21:57:20 UTC trade of $160.9661 implies a 5.09% post-print increase, displayed as $160.97. This after-market observation is time-qualified, not a final next-session reaction.

The prior episode emphasized the $638B backlog and a falling stock. This quarter asks what funds reported operating cash and what common owners retain after reinvestment and financing claims.

Price history measures expectations, not fair value. The chart retains regular-session closes; the later after-market observation is labeled separately.

The print

Broad growth, a genuine operating beat and a demanding funding programme.

Q1 revenue rises 29.61% to $19,345M. Cloud contributes $11,607M: infrastructure $7,388M and applications $4,219M. Software is $5,550M, hardware $774M and services $1,414M. Growth is concentrated in infrastructure.

Adjusted EPS of $1.92 exceeds the saved vendor estimate of $1.74 and the company’s prior $1.72–$1.76 U.S.-dollar range. Revenue also exceeds the saved $19,134.97M vendor estimate. The company had guided to 27%–29% revenue growth; the actual result is above that upper boundary. GAAP EPS is $1.56 and belongs beside GAAP earnings, not the adjusted consensus figure. Keeping those definitions aligned prevents a misleading comparison between an accounting result that includes employee and acquisition-related costs and an estimate that excludes selected items.

The annual revenue language is now at least $90B, while adjusted EPS guidance increases from $8.05 to $8.10. That is a five-cent EPS increase, not a large newly quantified revenue midpoint raise. The previous annual EPS comparison also needs normalization: FY2026 included gains associated with Ampere and Bloom Energy warrants, and the Q4 release gave $6.83 adjusted EPS excluding those gains versus $7.63 including them. The current $8.10 target is about 18.6% above that cleaner company comparison. Treating the old gains as recurring would understate the underlying earnings hurdle.

Q2 adjusted EPS guidance illustrates why an annual growth label can be deceptive without the footnote. The company expects 21%–25% growth in U.S. dollars after excluding the prior Ampere investment gain, but a decline of 14%–18% when that gain remains in the comparison. Both statements can be true. Our forecast discussion retains the exclusion rather than selecting whichever growth rate sounds most favorable. The next operating checkpoint is approximately $21.197B total revenue and $13.401B cloud revenue at the guidance midpoints, each computed from the corresponding prior-year base.

Businesses and capacity

Infrastructure drives growth while reported segment profit remains a different measure.

Cloud infrastructure grows 120.74% year over year, while cloud applications grows 9.90%. That contrast changes Oracle’s business mix. Infrastructure supplies a rapidly expanding share of the revenue base and requires physical capacity, power and equipment; applications and software support have different economics. The consolidated cloud label combines these activities, so applying a software-like margin assumption to every dollar of cloud revenue would miss an important distinction. The primary supplemental tables provide the separate revenue series needed to examine that mix without inventing separate infrastructure earnings.

Oracle has three reportable operating segments: cloud and software, hardware, and services. Management’s segment margin excludes research and development, general administration, intangible amortization, restructuring, segment share compensation and certain allocations. It is therefore a different measure from consolidated GAAP operating income. The latest annual report supplies three years of segment results, and companion 10-Qs supply quarterly history. Q4 segment amounts are computed as annual totals less the first three quarters. The current Q1 release does not provide the new detailed segment-margin note, so that field remains unavailable rather than being manufactured from a simpler cost subtraction.

A narrower direct-cost measure can still reveal the pressure created by changing mix. Cloud and software revenue less the corresponding reported cost line produces a 62.70% direct margin this quarter, versus 72.05% a year earlier. The cost line rises 77%, much faster than the combined revenue base. This is not an infrastructure gross margin or a management segment margin, and it must not be labeled as either. It is a transparent consolidated-category comparison that helps explain why rapid infrastructure expansion does not automatically translate into a proportionate improvement in all operating economics.

License revenue falls to $655M and support to $4,895M. Migration may explain the decline, but replacing mature software cash with infrastructure growth changes capital requirements. Total growth alone does not establish equivalent cash returns.

Profit bridge

The accounting improvement is real; the adjusted comparison is less dramatic.

GAAP operating income increases 57.31% to $6,728M, with margin of 34.78%. Adjusted operating income increases 30.71% to $8,151M, with margin of 42.13%. The adjusted margin is only modestly above the prior year, while the GAAP margin expands by roughly six percentage points. Lower amortization and restructuring costs account for part of that divergence. This does not invalidate the improvement. It tells us which part reflects stronger operating economics and which part reflects the timing of charges excluded from the company’s adjusted presentation.

Sales and marketing falls to $1,811M; research and development is $2,401M and general administration $376M. Central cost reductions offset cloud direct-cost growth. Durability matters because infrastructure depreciation and operating costs continue after capacity opens.

Interest expense of $1,428M is a substantial claim on the business. Nonoperating income of $307M partially offsets it, leaving $5,607M pretax income. The $847M tax provision produces $4,760M net income, and the $81M preferred dividend reduces the amount available to common shareholders to $4,679M. Operating income covers reported interest expense about 4.71 times. That ratio is currently adequate, but the expansion programme and future financing make the direction of coverage relevant. A presentation ending at operating profit would omit costs that materially affect the owners being valued.

SBC of $1,127M represents about 79% of the operating adjustment gap. Compensation costs remain in our economic earnings model even though non-GAAP earnings and operating cash add them back.

Earnings quality

Cash timing, reinvestment and the common-share denominator need separate tests.

The earnings-quality judgment is mixed. Revenue and adjusted profit grow about 30%, but GAAP growth benefits from smaller adjustment charges. Cash also reflects financing-component prepayments. These distinctions explain economics without alleging accounting impropriety.

Preferred dividends explain why net income and common income differ for the first time in this nine-quarter sequence. The mandatory preferred can be antidilutive under the GAAP if-converted test in one period and still become real common shares later. Reported diluted EPS therefore does not guarantee that every future conversion share is already in today’s denominator. The latest quarter has 3,000M diluted weighted-average shares; the latest annual cover count of 2,880.471M was dated June 12, before the completed $20B ATM programme. Neither figure is an exact current end-of-quarter diluted share count.

Our model uses an explicit 3,100M common diluted-share proxy before preferred conversion, then adds the maximum ordinary conversion amount of 31.238285M shares. The result is 3,131.238285M as-converted model shares. This is deliberately labeled an analyst convention, with sensitivity to 3,000M–3,200M before conversion. It is not asserted as a filed endpoint, and the ATM shares are not added again on top of a proxy already intended to represent the completed issuance. The missing exact count should be resolved when the current 10-Q appears, but its absence does not justify using a clearly stale float without qualification.

Published $1M differences remain visible in cash-flow versus adjustment SBC and FY2026 quarterly versus annual totals. The calculation audit documents each; annual totals remain authoritative for annual comparisons.

Cash and capital

A record cash inflow carries a corresponding obligation to deliver.

Operating cash of $23,103M includes $11,363M of customer prepayments with a significant financing component. That amount represents about 49% of current operating cash and about 76% of the year-over-year increase. Removing it as a timing sensitivity leaves $11,740M operating cash. This counterfactual is not an accounting restatement and does not claim that the money is unusable. It isolates an unusually large funding source whose receipt precedes the related performance obligations, making clear why a record cash headline cannot be read as unrestricted recurring profit.

Capital expenditures are $28,499M, leaving conventional free cash flow of negative $5,396M. A counterfactual that removes the financing-component prepayments gives negative $16,759M. The two figures answer different questions: reported-statement FCF shows the current cash result, while the sensitivity illustrates reliance on the timing of customer funding. Neither is a permanent maintenance-capex estimate. Oracle is building capacity for future demand, so some current expenditure can support later cash generation. The crucial investment judgment is how much, how quickly, and at what return after all the associated obligations are satisfied.

The new company table titled net cash outlay for capital expenditures is especially easy to misuse. It begins with $28,499M capex, adds $830M of net repayments of short-term capex financing and subtracts $11,363M customer financing prepayments, producing $17,966M net outlay. That measure explains funding, not the physical cost of the assets. Subtracting it from operating cash that already includes the prepayments would count the same financing benefit twice. We keep conventional FCF, the company’s net-outlay measure and the prepayment timing sensitivity visibly separate.

The annual cash record shows the scale of the transformation. Free cash flow moves from positive $11,807M in FY2024 to negative $394M in FY2025 and negative $23,686M in FY2026. The latest four quarters total negative $28,720M. Capital expenditure has increased much faster than depreciation, and current quarterly capex is roughly nine times quarterly depreciation. Depreciation will follow the installed asset base with a lag, while the required replacement pace depends on equipment economics and useful lives. It would be premature to describe the present accounting margin as the full steady-state return on the newest infrastructure.

The annual accounting policy explains that financing-component prepayments create deferred revenue and a separately recognized interest expense, using a discount-rate convention linked to the financing recipient’s credit characteristics. The policy includes a practical expedient for one-year timing differences. These customer arrangements are distinct from Oracle financing customer purchases and selling the receivables to banks. Commercial cash schedules, collateral and counterparty-specific economics are not fully disclosed in the saved instruments. Our valuation therefore treats future economic cash as an explicit scenario after customer settlement and financing effects, rather than claiming to reconstruct private contract terms from a headline prepayment balance.

Valuation

Three routes expose the gap between accounting profits and cash payback.

The central appraisal is $130. The earnings route gives $150.37, the operating-enterprise route $145.17 and the cash-transition DCF $78.97. We weight them 40%, 35% and 25%, producing $130.70 before coarse rounding to the nearest $10. The spread is material: accounting earnings methods give more credit to growth already entering the income statement, while the DCF charges for the timing of future capital needs and delayed cash payback. Shared assumptions leave correlated model uncertainty. The DCF carries only 25% weight.

The earnings route approximates guided common earnings by multiplying $8.10 adjusted EPS by our 3,100M share proxy. Management has not disclosed that exact guidance denominator, so $25,110M is an analyst approximation, not company net-income guidance. It deducts $5,000M assumed annual SBC after a 20% tax rate, then adds back $325M annual preferred coupons to express earnings on an as-converted basis. Dividing by 3,131.238285M shares gives $6.845535 economic EPS. A 22× multiple produces the equity value before deducting $713.55M present value of remaining preferred cash coupons, or roughly $0.228 per share. The final result is $150.37. This is an explicit economic normalization, not company GAAP EPS guidance.

The audited FY2026 EPS note explicitly excluded anti-dilutive mandatory preferred conversion shares. Current-quarter adjusted common income of $5,758M divided by 3,000M shares reproduces $1.92 without adding preferred dividends back. At maximum conversion, the annual coupon divided by incremental shares is about $10.40, above the $8.10 guide; this supports our assumption that guidance excludes anti-dilutive preferred conversion. Current detailed EPS notes remain unavailable. We therefore label the method an assumption supported by filed history and current arithmetic, rather than claiming access to an undisclosed guidance calculation.

The enterprise route assumes $90,000M forward revenue and a 35% GAAP-like operating margin, including recurring compensation and ordinary operating rent. That gives $31,500M operating income, valued at 17.5× for $551,250M enterprise value. We deduct $88,260M net funded debt, a $7,701M finance-lease liability proxy from the latest audited balance, and the same remaining preferred-coupon present value. Dividing by the as-converted shares gives $145.17. The current quarter’s exact finance-lease balance is unavailable; its annual proxy is a declared limitation, not a falsely current filing value. Operating leases are treated through rent in the modeled earnings.

Preferred conversion is modeled once. The 50,000 preferred shares convert at the maximum ordinary rate of 624.7657 common shares each, subject to contractual adjustments. We add those 31.238285M common shares and deduct the present value of ten assumed cash dividends through January 15, 2029. We do not also subtract the $5B liquidation preference. The coupons may legally be paid in stock or a combination under the instruments; cash is our modeling assumption. The remaining coupon calculation uses actual scheduled dates and the 11.5% discount rate, so its $713.55M value can be reproduced without conflating face amount, conversion dilution and cash distributions.

The DCF uses a transition scenario of negative $20B, negative $5B, then positive $10B, $25B and $45B annual unlevered economic cash. These are analyst assumptions after recurring compensation, normal operating rent, customer settlement and financing-component economics, and economic reinvestment. They are not mechanically derived from current OCF, an undisclosed capex guide or private contract schedules. They assume substantial eventual success: a business now consuming free cash reaches $45B annual economic cash by year five. Existing funded-debt and finance-lease principal payments sit outside these unlevered flows and are handled in the enterprise-to-equity bridge.

This DCF is an illustrative cash-transition stress test, anchored to an explicit successful year-five operating case: $220B revenue, 37.5% economic operating margin after SBC and rent, and 20% cash tax give $66B after-tax operating earnings. Deducting $21B net economic reinvestment and customer-funding settlement gives $45B cash. Net reinvestment includes purchased and newly leased asset additions less depreciation, plus working-capital and financing-component settlement; it is not gross capex. Revenue rises roughly 25% annually from the $90B starting convention, while the terminal cash margin reaches 20.5%. Those are demanding assumptions, including substantial utilization gains and lower incremental capital intensity. The early negative cash years represent construction preceding customer payback, not a disclosed contract schedule. The model receives only 25% weight because the timing and capital burden cannot yet be forecast precisely.

Discounting that path at 11.5% with 3% terminal growth gives the $78.97 common value after the same financing claims and coupon present value. The terminal convention assumes investment normalizes enough for economic cash to grow steadily after the explicit transition. At the saved price, holding the first four cash flows, discount rate, terminal growth and capital bridge fixed requires approximately $78.73B year-five cash, about 75% above our $45B assumption. That reverse calculation identifies the burden of the price. It is not a company forecast and does not prove the higher outcome is impossible.

The bear/base/bull cases are $70/$130/$240 with 25%/50%/25% probabilities, giving $142.50 weighted value. Their EPS-and-multiple conventions are separate from the detailed models. Even the weighted scenario value is below the saved price, while the central appraisal implies about 19% downside. A $97.50 price would provide a 25% discount to the current appraisal with the business thesis intact. This is a conditional reassessment threshold, not an automatic purchase instruction. If execution or funding changes the economics, the appraisal must change before the entry rule is applied.

Street context

Dated targets are useful; contaminated or stale inputs are not consensus.

The saved target feed shows Morgan Stanley at $210 on September 4 and Jefferies at $290 on September 2. The corresponding rating history records Equal Weight and Buy. Both targets predate this release and sit above our $130 appraisal. That divergence is substantive: the Street marks may give more credit to the growth runway and later capital efficiency than our cash scenario does. We do not present them as post-earnings endorsements, and the presence of a high target does not validate the exact valuation path used in this packet.

A vendor target row tagged ORCL actually concerns Quantinuum/QNT and Bernstein’s $95 target. We retain the raw evidence but exclude that wrong-company observation from Oracle analysis.

Annual estimate and earnings snapshots are saved. Quarterly analyst estimates returned HTTP 402; missing values remain unavailable. The saved current-print earnings estimate is distinct from an unavailable quarterly forward-consensus series.

Target differences can reflect margins, financing costs or capital intensity. Without their full models we cannot attribute each disagreement. Our assumptions are exposed so readers can test the common-share proxy, compensation cost and cash transition.

Management and capital allocation

Execution is improving; the funding promises must retain their exact scope.

The latest annual report names Clayton Magouyrk and Michael Sicilia as chief executives, Hilary Maxson as CFO, and Lawrence Ellison as executive chair and CTO. Their test is capacity delivery with acceptable capital returns.

The Q4 release described approximately $40B of debt and equity funding planned for FY2027, including the $20B ATM, and said additional debt was not expected in calendar 2026. The current release confirms completion of the ATM and says the new AI contracts do not incrementally change the capital-raise plans. That language does not mean the whole earlier funding plan has vanished or that no future external financing will be needed. We retain the original scope of each statement rather than transforming “no incremental impact” into an unconditional self-funding claim.

The preferred financing brings $5B face value with cumulative 6.5% dividends and mandatory common conversion tied to a 2029 averaging period. The ATM brings $19,909M net cash against $20B gross issuance. These are different financing instruments with different claims on shareholders. Their proceeds support the business, but their availability does not by itself establish that the assets purchased will earn an attractive return. The appropriate management test is the economic value created after servicing financing claims and issuing ownership, not simply the size of a financing transaction completed.

Employees fell to 141,000 as infrastructure expanded. Lower central costs help margins, but execution must preserve service quality and software development while data-center commitments increase.

Ranked risks

The largest risks concern payback, funding and the timing of obligations.

Infrastructure growth can coexist with inadequate capital returns. Replacement cycles, asset utilization and fulfillment costs matter alongside revenue. Capex is about nine times current depreciation, so accounting earnings and current investment cash reflect different stages of the build.

Second is the timing and concentration of customer commitments. No customer represented 10% of recognized annual revenue in FY2024–FY2026, but that historical statement does not resolve concentration in newly signed, long-duration AI contracts. RPO of $664B is not cash, annual sales or earnings. The latest annual maturity disclosure for the older $638B balance expected only 12% to be recognized in the next twelve months, with most spread over later years. We do not apply those old percentages mechanically to the new balance. Counterparty performance and contract economics remain relevant even when a large headline backlog is legally contracted.

Third is the lease pipeline. At May 31, Oracle disclosed $260B of additional leases not yet commenced, generally starting between Q1 FY2027 and FY2029 for fifteen-to-nineteen-year terms. They were outside the then-current balance sheet. That is a major operating commitment, but subtracting the entire undiscounted amount as present funded debt would be incorrect. Our models instead retain ordinary rent in economic costs and require investment assumptions that can support the planned footprint. The figure is dated to the annual report and should be updated when the current quarter’s detailed notes become available.

The annual disclosure also includes a guarantee of up to $3.3B of a lessor’s borrowing that matures in September 2026, plus $13.309B unconditional purchase and other obligations, primarily power arrangements. A maturity does not automatically mean a guarantee is called, and the full notional amount is not a current cash outflow. It is nevertheless a near-term item worth checking explicitly. Current finance-lease totals and the difference between cash-flow cash including restricted cash and balance-sheet cash remain unresolved in the release. Using old “immaterial restricted cash” language to explain the new quarter would conceal that source gap.

Decision checkpoints

Judge the next print using matched definitions and visible funding tests.

Q2 guidance midpoints are $21.197B revenue and $13.401B cloud revenue. Our analyst operating checks seek adjusted margin of at least 42% and interest coverage of at least 4.5×, with warnings below 40% and 4.0×.

Track financing prepayments separately from FCF. The current ex-prepayment sensitivity is negative $16.759B; better than negative $15B would improve confidence, while below negative $20B would worsen it. These are analytical thresholds, not company forecasts.

The pending Q1 10-Q should refine ending shares, finance leases, restricted cash and commitments. Better disclosure can narrow uncertainty; it cannot by itself establish that the infrastructure earns enough.

The bull case is rapid infrastructure growth with customer funding and lower central costs. The bear case is that fulfillment, financing and replacement consume the gains. The SELL appraisal requires stronger cash evidence or a more protective price.

Source and calculation appendix

Saved original bytes, visible definitions and reproducible arithmetic.

Primary-source identity is fixed to Oracle Corporation, CIK 1341439, NYSE common ticker ORCL, Q1 FY2027 ended August 31, 2026. The current release is Exhibit 99.1 in accession 0001193125-26-387905. All nine earnings releases, three annual reports, six available 10-Qs and financing instruments were saved before analysis. The current quarterly 10-Q was not yet present in the fresh SEC submissions check. The complete originals, table extracts, source URLs, retrieval timestamps and SHA-256 values remain available in the episode research directory for reproducibility.

The dataset preserves nine quarterly below-line statements, reconciliations, balance-sheet claims and cash-flow histories, plus three annual comparisons. Standalone quarters derived from cumulative filings are flagged. Source hashes and formula assertions make the calculations reproducible.

The full recent SEC screen resolves two historical Form 25 notices as debt-note removals, not common-stock delistings. Duplicate and prior-angle evidence is retained. The missing current 10-Q limits exact current capital-structure detail.