ORCL · Forward model · A few very large customers case
The A few very large customers case, 20 quarters out
Each segment is projected from its own operating driver, rolled up into consolidated cash flow, and discounted back to a fair value per share. The assumptions below are editable — change them and every number on this page moves with them.
Oracle publishes all six revenue lines by quarter in the supplemental analysis attached to every earnings release, so every actual here is a REPORTED figure and nothing is estimated or apportioned. The six lines sum exactly to reported consolidated revenue in all eight quarters shown. Quarter labels are FISCAL: '2026 Q4' is the quarter ended 31 May 2026. What is assumed is the COST split, not the revenue split. Oracle reports three operating segments — cloud and software, hardware and services — and publishes a segment margin for each, but never by offering. Hardware (65.4%) and Services (26.7%) therefore carry DISCLOSED margins; the four lines inside cloud and software carry assumed margins chosen to weight-average back to the disclosed 58.89% blend, which they do exactly. Oracle leaves R&D, G&A, stock compensation, intangible amortisation and restructuring unallocated, so those sit in the corporate layer at 25.85% of revenue — the figure that reconciles total segment margin of $38,018M to reported operating income of $20,606M. Two known limitations. First, the growth driver has no seasonality primitive, and Oracle's licence and hardware lines peak hard in the May quarter; both are pinned near the slider floor in the first projected quarter to undo that, so read those two lines annually and never quarterly. Second, live OCI megawatts are not disclosed anywhere — Oracle reports capacity DELIVERED, not the earning base — so the 2,000 MW at the basis quarter is inferred and the $2.97M per megawatt-quarter price falls out of it. Capex is set on Oracle's net cash outlay basis, the number it presents as its real funding requirement, not on reported capex; the two differed by $7.9B in FY2026. Oracle's disclosed segment margin is struck AFTER depreciation of datacenter assets, so it is an operating margin, not EBITDA. FY2026 depreciation of $7,623M is added back to cloud infrastructure, where the datacenter assets sit, taking its 27.0% segment margin to 69.1%. Corporate overhead is set net of the $1,671M of intangible amortisation for the same reason. The result reconciles exactly: $45,642M of vertical EBITDA less $15,741M of overhead is $29,901M, which is reported operating income of $20,606M plus depreciation and intangible amortisation. Two things this model does not carry. It projects no share issuance, yet Oracle plans to raise about $40B in FY2027 including a $20B at-the-market equity programme - at the authoring price that alone is roughly 136M shares, about 4.7% dilution the fair value per share does not reflect. And $4,954M of mandatory convertible preferred was issued in FY2026; its conversion is not in the 2,915M diluted count either. Free cash flow stays negative until FY2030 in the base case, which is what makes that issuance necessary rather than optional.
Oracle will not publish RPO by customer, and what it does publish points at concentration: four customers each signed for more than $8B in a single quarter, and in the GPU renewal cohort fewer than half the customers came back. Management's answer is that the non-renewed capacity was resold inside the quarter at 97.5% global utilisation — true, and also exactly what you would expect while demand exceeds supply. This case asks what the book is worth if that stops being true.
Latest: $48.42B (2031Q4E)
| Period | Value |
|---|---|
| 2025Q1 | $13.31B |
| 2025Q2 | $14.06B |
| 2025Q3 | $14.13B |
| 2025Q4 | $15.90B |
| 2026Q1 | $14.93B |
| 2026Q2 | $16.06B |
| 2026Q3 | $17.19B |
| 2026Q4 | $19.18B |
| 2027Q1E | $20.01B |
| 2027Q2E | $21.01B |
| 2027Q3E | $22.14B |
| 2027Q4E | $23.36B |
| 2028Q1E | $24.64B |
| 2028Q2E | $25.97B |
| 2028Q3E | $27.35B |
| 2028Q4E | $28.77B |
| 2029Q1E | $30.22B |
| 2029Q2E | $31.71B |
| 2029Q3E | $33.23B |
| 2029Q4E | $34.79B |
| 2030Q1E | $36.37B |
| 2030Q2E | $37.99B |
| 2030Q3E | $39.64B |
| 2030Q4E | $41.33B |
| 2031Q1E | $43.04B |
| 2031Q2E | $44.80B |
| 2031Q3E | $46.59B |
| 2031Q4E | $48.42B |
What drives each segment
Cloud infrastructure (OCI)
Capacity × utilisation × priceThe whole model. OCI is 30% of FY2026 revenue and effectively all of the growth: $18.1B, up 77%, with the June quarter alone up 93%. It carries GPU training and inference capacity, Oracle Database in OCI, the multicloud database running inside AWS, Azure and Google, and ordinary compute and storage. Revenue is gated by delivered datacenter capacity, not demand — Oracle reports 97.5% global GPU utilisation and says recognition follows site delivery, so the projection is a build schedule in megawatts, a utilisation rate and a price per megawatt-quarter.
Latest: $37.50B (2031Q4E)
| Period | Value |
|---|---|
| 2025Q1 | $2.15B |
| 2025Q2 | $2.43B |
| 2025Q3 | $2.65B |
| 2025Q4 | $3.00B |
| 2026Q1 | $3.35B |
| 2026Q2 | $4.08B |
| 2026Q3 | $4.89B |
| 2026Q4 | $5.79B |
| 2027Q1E | $7.13B |
| 2027Q2E | $8.49B |
| 2027Q3E | $9.88B |
| 2027Q4E | $11.29B |
| 2028Q1E | $12.72B |
| 2028Q2E | $14.18B |
| 2028Q3E | $15.67B |
| 2028Q4E | $17.18B |
| 2029Q1E | $18.71B |
| 2029Q2E | $20.27B |
| 2029Q3E | $21.86B |
| 2029Q4E | $23.48B |
| 2030Q1E | $25.12B |
| 2030Q2E | $26.79B |
| 2030Q3E | $28.49B |
| 2030Q4E | $30.23B |
| 2031Q1E | $32.00B |
| 2031Q2E | $33.80B |
| 2031Q3E | $35.63B |
| 2031Q4E | $37.50B |
Assumptions & reasoning
- Live megawatts are the weakest input here. Oracle discloses capacity DELIVERED (more than 1.2 GW in FY26, approaching 1 GW in Q1 FY27) but never the live earning base, so the 2,000 MW at the basis quarter is inferred and the price per megawatt falls out of it rather than being observed.
- Delivered is not earning. Capacity handed over inside a quarter earns for only part of it, so the megawatts added in this driver run below Oracle's headline delivery figures on purpose — the engine gives a full quarter of revenue to everything it adds.
- The implied $2.97M per megawatt-quarter is roughly 1.7x what CoreWeave's disclosure implies for a pure GPU cloud, which is the right direction: OCI also carries database, multicloud and ordinary IaaS revenue that consumes little power.
- Capex intensity here is set on Oracle's NET cash outlay basis ($47.7B in FY26), not reported capex ($55.7B). Oracle itself presents the net figure as the real funding requirement, and the difference is customer prepayments and third-party manufacturer timing — the same prepayments that make the RPO convertible.
- The $90B FY2027 guide implies OCI revenue of roughly $39B once the five other lines are grown at their observed rates — about 22% above the $32B in the September 2025 five-year schedule. Either the schedule has been quietly raised or the non-infrastructure lines must accelerate; nothing Oracle published at Q4 resolves which.
Cloud applications (SaaS)
Growth pathFusion and NetSuite-class ERP, HCM, CX and industry suites — the business Oracle spent a decade building before AI arrived. It grew 11% in FY2026 to $15.9B and has grown between 9% and 13% in every quarter of the last two years, which is the point: it is the stable half of cloud and the reason total cloud growth is not a pure function of GPU deliveries.
Latest: $5.02B (2031Q4E)
| Period | Value |
|---|---|
| 2025Q1 | $3.47B |
| 2025Q2 | $3.50B |
| 2025Q3 | $3.56B |
| 2025Q4 | $3.74B |
| 2026Q1 | $3.84B |
| 2026Q2 | $3.90B |
| 2026Q3 | $4.03B |
| 2026Q4 | $4.13B |
| 2027Q1E | $4.19B |
| 2027Q2E | $4.25B |
| 2027Q3E | $4.30B |
| 2027Q4E | $4.36B |
| 2028Q1E | $4.41B |
| 2028Q2E | $4.46B |
| 2028Q3E | $4.51B |
| 2028Q4E | $4.55B |
| 2029Q1E | $4.60B |
| 2029Q2E | $4.64B |
| 2029Q3E | $4.68B |
| 2029Q4E | $4.73B |
| 2030Q1E | $4.77B |
| 2030Q2E | $4.80B |
| 2030Q3E | $4.84B |
| 2030Q4E | $4.88B |
| 2031Q1E | $4.92B |
| 2031Q2E | $4.95B |
| 2031Q3E | $4.99B |
| 2031Q4E | $5.02B |
Assumptions & reasoning
- This line is remarkably steady — 10%, 10%, 9%, 12%, 11%, 11%, 13%, 10% year over year across the eight reported quarters — which is why a growth driver costs almost nothing in fidelity here and why it anchors the non-AI half of the story.
- Oracle gave one piece of forward colour that is not modellable: SaaS deferred revenue grew 16% in Q4, faster than the 10% revenue growth. That is a leading indicator with no disclosed conversion rate, so it is noted rather than used.
Software support
Growth pathThe annuity under the old Oracle: maintenance on the on-premise licence base, about $19.8B a year and essentially flat for three years. It funds everything else. It declines slowly as customers migrate to cloud, and the migration is visible in the licence line above it rather than here, because support attrition lags licence attrition by years.
Latest: $3.61B (2031Q4E)
| Period | Value |
|---|---|
| 2025Q1 | $4.90B |
| 2025Q2 | $4.87B |
| 2025Q3 | $4.80B |
| 2025Q4 | $4.96B |
| 2026Q1 | $4.96B |
| 2026Q2 | $4.94B |
| 2026Q3 | $4.97B |
| 2026Q4 | $4.94B |
| 2027Q1E | $4.89B |
| 2027Q2E | $4.84B |
| 2027Q3E | $4.78B |
| 2027Q4E | $4.72B |
| 2028Q1E | $4.66B |
| 2028Q2E | $4.59B |
| 2028Q3E | $4.53B |
| 2028Q4E | $4.46B |
| 2029Q1E | $4.39B |
| 2029Q2E | $4.32B |
| 2029Q3E | $4.25B |
| 2029Q4E | $4.17B |
| 2030Q1E | $4.10B |
| 2030Q2E | $4.03B |
| 2030Q3E | $3.96B |
| 2030Q4E | $3.89B |
| 2031Q1E | $3.82B |
| 2031Q2E | $3.75B |
| 2031Q3E | $3.68B |
| 2031Q4E | $3.61B |
Assumptions & reasoning
- Flat is the finding, not an absence of one: $19.5B, $19.5B and $19.8B across FY2024 to FY2026 while licence revenue fell 9% in the last year. Support has not yet followed licence down, and when it does it takes roughly 85 cents of segment margin per dollar with it.
- The margin assigned here is the single largest cost assumption in the brief. Oracle publishes one margin for all of cloud and software, so an 85% support margin and a 27% infrastructure margin are a judgement about mix, constrained only by having to average back to the disclosed blend.
Software license
Growth pathNew on-premise licence sales, in structural decline as the installed base migrates to cloud: $4.7B in FY2026, down 9%. It is small and shrinking, but it matters twice over — it is the leading indicator for the support annuity beneath it, and it is violently seasonal, with a fourth quarter roughly twice the size of a first quarter.
Latest: $256M (2031Q4E)
| Period | Value |
|---|---|
| 2025Q1 | $870M |
| 2025Q2 | $1.20B |
| 2025Q3 | $1.13B |
| 2025Q4 | $2.01B |
| 2026Q1 | $766M |
| 2026Q2 | $939M |
| 2026Q3 | $1.15B |
| 2026Q4 | $1.88B |
| 2027Q1E | $1.49B |
| 2027Q2E | $1.22B |
| 2027Q3E | $1.02B |
| 2027Q4E | $875M |
| 2028Q1E | $763M |
| 2028Q2E | $676M |
| 2028Q3E | $606M |
| 2028Q4E | $550M |
| 2029Q1E | $503M |
| 2029Q2E | $463M |
| 2029Q3E | $429M |
| 2029Q4E | $400M |
| 2030Q1E | $375M |
| 2030Q2E | $352M |
| 2030Q3E | $332M |
| 2030Q4E | $314M |
| 2031Q1E | $298M |
| 2031Q2E | $283M |
| 2031Q3E | $269M |
| 2031Q4E | $256M |
Assumptions & reasoning
- The growth driver cannot express seasonality, and this line is the most seasonal Oracle reports: 870, 1195, 1129, 2007 through FY2025 and 766, 939, 1150, 1881 through FY2026. Because the driver compounds from the last actual and the last actual is the annual peak, the first projected quarter is pinned at the slider floor to undo it. Read the annual path from this line, never the quarterly one.
- Even pinned at the floor the model projects fiscal 2027 licence revenue close to flat against FY2026, where the trend is a 9% annual decline. That is a known overstatement of roughly $250M in the first projected year, about 0.3% of consolidated revenue, and it decays after.
Hardware
Growth pathEngineered systems, servers and storage — Exadata and its relatives, $3.1B in FY2026 and up 5%, the first growth in years as Exadata rides the database business into cloud. Small, seasonal, and profitable at a 65% segment margin, but structurally a supporting act.
Latest: $499M (2031Q4E)
| Period | Value |
|---|---|
| 2025Q1 | $655M |
| 2025Q2 | $728M |
| 2025Q3 | $703M |
| 2025Q4 | $850M |
| 2026Q1 | $670M |
| 2026Q2 | $776M |
| 2026Q3 | $714M |
| 2026Q4 | $924M |
| 2027Q1E | $805M |
| 2027Q2E | $726M |
| 2027Q3E | $672M |
| 2027Q4E | $634M |
| 2028Q1E | $606M |
| 2028Q2E | $585M |
| 2028Q3E | $569M |
| 2028Q4E | $557M |
| 2029Q1E | $547M |
| 2029Q2E | $539M |
| 2029Q3E | $533M |
| 2029Q4E | $527M |
| 2030Q1E | $522M |
| 2030Q2E | $518M |
| 2030Q3E | $514M |
| 2030Q4E | $511M |
| 2031Q1E | $508M |
| 2031Q2E | $505M |
| 2031Q3E | $502M |
| 2031Q4E | $499M |
Assumptions & reasoning
- This is one of only two verticals whose margin is disclosed rather than assumed: the 10-K reports hardware segment operating income of $2,017M on $3,084M of revenue, a 65.4% margin that has held within a point for three fiscal years.
- The line turned positive in FY2026 after years of decline, growing 5%. Oracle attributes that to Exadata demand pulled along by the database business, which makes it a lagging read on the same thing the infrastructure line measures.
Services
Growth pathConsulting and support services that implement and migrate customers onto everything else: $5.7B in FY2026, up 10%, and newly profitable at a 26.7% segment margin against 19% two years ago. It is a trailing indicator of cloud adoption — it grows when customers are actively moving, which is why it turned up at the same time OCI did.
Latest: $1.53B (2031Q4E)
| Period | Value |
|---|---|
| 2025Q1 | $1.26B |
| 2025Q2 | $1.33B |
| 2025Q3 | $1.29B |
| 2025Q4 | $1.35B |
| 2026Q1 | $1.35B |
| 2026Q2 | $1.43B |
| 2026Q3 | $1.44B |
| 2026Q4 | $1.52B |
| 2027Q1E | $1.51B |
| 2027Q2E | $1.50B |
| 2027Q3E | $1.49B |
| 2027Q4E | $1.48B |
| 2028Q1E | $1.48B |
| 2028Q2E | $1.47B |
| 2028Q3E | $1.47B |
| 2028Q4E | $1.47B |
| 2029Q1E | $1.47B |
| 2029Q2E | $1.48B |
| 2029Q3E | $1.48B |
| 2029Q4E | $1.48B |
| 2030Q1E | $1.49B |
| 2030Q2E | $1.49B |
| 2030Q3E | $1.50B |
| 2030Q4E | $1.50B |
| 2031Q1E | $1.51B |
| 2031Q2E | $1.51B |
| 2031Q3E | $1.52B |
| 2031Q4E | $1.53B |
Assumptions & reasoning
- The margin here is disclosed and it has moved hard: segment operating income went from $916M on $5,431M in FY2024 to $1,533M on $5,743M in FY2026, nearly doubling the margin on 6% more revenue.
- Services growth turned from -3% in FY2024 to +10% in FY2026, inflecting in the same year OCI did. Treat it as confirmation of migration activity rather than as an independent driver.
Where each case comes from
Delivery, margin and the funding bill case — primary sources
The primary sources this case is built from — filings, calls and posts. Where they are claims by an interested party rather than disclosures, every number in the Delivery, margin and the funding bill column is what happens if they are taken at face value.
Gross margin steps down in FY2027
The financing requirement
- Jun 10, 2026 In fiscal year 2027, Oracle expects to raise approximately $40 billion through a combination of debt and equity financing including its previously announced $20 billion at-the-market equity issuance.
- Jun 10, 2026 Free cash flow was negative $23.7 billion for fiscal year 2026 as Oracle continued to execute on investments to support the growth of its Cloud Infrastructure business.
The OCI schedule converts case — primary sources
The primary sources this case is built from — filings, calls and posts. Where they are claims by an interested party rather than disclosures, every number in the The OCI schedule converts column is what happens if they are taken at face value.
The multi-year OCI revenue schedule
Long-term CAGRs reconfirmed at Q4 FY2026
Prepaid and customer-supplied hardware reduce the capital Oracle must raise
A few very large customers case — primary sources
The primary sources this case is built from — filings, calls and posts. Where they are claims by an interested party rather than disclosures, every number in the A few very large customers column is what happens if they are taken at face value.
Q4 signings and customer concentration
The GPU renewal cohort
From cash flow to fair value
| Present value of free cash flow, 20 quarters | -$46.16B |
| Terminal-year revenue | $182.85B |
| Terminal-year EBITDA | $79.34B |
| Exit multiple, on revenue | 4.5x |
| Terminal value | $822.82B |
| Discounted at 9.5% a year, terminal value becomes | $522.68B |
| Enterprise value | $476.51B |
| Net cash | -$97.65B |
| Equity value | $378.87B |
| Diluted shares | 2.92B |
| Fair value per share | $129.97 |
| Against the current price of $142.07 | -9% |
9.5% pays for an investment-grade mega-cap carrying $129.5B of debt, negative free cash flow and about $40B a year of planned issuance. The 5.5x exit is a cloud-and-software blend at maturity, between the house marks for CRM at 4.5x and META at 6.0x.
Read the other way round: at $142.07 the market is paying 4.8x terminal-year revenue, holding every other assumption on this page fixed. That is the number to argue about.
The projected path
| Quarter | Cloud infrastructure (OCI) | Cloud applications (SaaS) | Software support | Software license | Hardware | Services | Revenue | YoY | EBITDA | Capex | FCF | R40 | PV of FCF |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2027 Q1E | $7.13B | $4.19B | $4.89B | $1.49B | $805M | $1.51B | $20.01B | +34% | $8.63B | $14.30B | -$5.67B | +6 | -$5.55B |
| 2027 Q2E | $8.49B | $4.25B | $4.84B | $1.22B | $726M | $1.50B | $21.01B | +31% | $9.03B | $15.29B | -$6.26B | +1 | -$5.98B |
| 2027 Q3E | $9.88B | $4.30B | $4.78B | $1.02B | $672M | $1.49B | $22.14B | +29% | $9.50B | $16.00B | -$6.50B | -1 | -$6.08B |
| 2027 Q4E | $11.29B | $4.36B | $4.72B | $875M | $634M | $1.48B | $23.36B | +22% | $10.01B | $16.50B | -$6.49B | -6 | -$5.92B |
| 2028 Q1E | $12.72B | $4.41B | $4.66B | $763M | $606M | $1.48B | $24.64B | +23% | $10.56B | $16.83B | -$6.27B | -2 | -$5.59B |
| 2028 Q2E | $14.18B | $4.46B | $4.59B | $676M | $585M | $1.47B | $25.97B | +24% | $11.14B | $17.02B | -$5.88B | +1 | -$5.14B |
| 2028 Q3E | $15.67B | $4.51B | $4.53B | $606M | $569M | $1.47B | $27.35B | +24% | $11.74B | $17.12B | -$5.38B | +4 | -$4.59B |
| 2028 Q4E | $17.18B | $4.55B | $4.46B | $550M | $557M | $1.47B | $28.77B | +23% | $12.36B | $17.14B | -$4.79B | +7 | -$3.99B |
| 2029 Q1E | $18.71B | $4.60B | $4.39B | $503M | $547M | $1.47B | $30.22B | +23% | $12.99B | $17.12B | -$4.12B | +9 | -$3.36B |
| 2029 Q2E | $20.27B | $4.64B | $4.32B | $463M | $539M | $1.48B | $31.71B | +22% | $13.65B | $17.06B | -$3.42B | +11 | -$2.72B |
| 2029 Q3E | $21.86B | $4.68B | $4.25B | $429M | $533M | $1.48B | $33.23B | +22% | $14.32B | $16.99B | -$2.68B | +13 | -$2.08B |
| 2029 Q4E | $23.48B | $4.73B | $4.17B | $400M | $527M | $1.48B | $34.79B | +21% | $15.00B | $16.92B | -$1.92B | +15 | -$1.46B |
| 2030 Q1E | $25.12B | $4.77B | $4.10B | $375M | $522M | $1.49B | $36.37B | +20% | $15.70B | $16.85B | -$1.15B | +17 | -$855M |
| 2030 Q2E | $26.79B | $4.80B | $4.03B | $352M | $518M | $1.49B | $37.99B | +20% | $16.42B | $16.80B | -$380M | +19 | -$277M |
| 2030 Q3E | $28.49B | $4.84B | $3.96B | $332M | $514M | $1.50B | $39.64B | +19% | $17.15B | $16.76B | $325M | +20 | $231M |
| 2030 Q4E | $30.23B | $4.88B | $3.89B | $314M | $511M | $1.50B | $41.33B | +19% | $17.89B | $16.76B | $965M | +21 | $671M |
| 2031 Q1E | $32.00B | $4.92B | $3.82B | $298M | $508M | $1.51B | $43.04B | +18% | $18.65B | $16.78B | $1.59B | +22 | $1.08B |
| 2031 Q2E | $33.80B | $4.95B | $3.75B | $283M | $505M | $1.51B | $44.80B | +18% | $19.43B | $16.83B | $2.21B | +23 | $1.47B |
| 2031 Q3E | $35.63B | $4.99B | $3.68B | $269M | $502M | $1.52B | $46.59B | +18% | $20.22B | $16.92B | $2.81B | +24 | $1.83B |
| 2031 Q4E | $37.50B | $5.02B | $3.61B | $256M | $499M | $1.53B | $48.42B | +17% | $21.04B | $17.04B | $3.40B | +24 | $2.16B |
Every row is projected. A year-over-year change is shown only where the quarter it compares against exists — an em dash means there is no comparable quarter, not a flat year.
Model revisions
Assumptions are marked to reality as each quarter prints. Every change is appended here, with the fair value the model produced at the time, so the model's own history stays visible.
| Date | Changed | Fair value then | Note |
|---|---|---|---|
| 2026-08-23 | — | First cut, built from the scope-model intake brief at data/models/intake/orcl.json on the fiscal 2026 Q4 basis. |