OCI +121%, the Guide Goes Up, and the Equity Program Is Done. But Strip Out $11.4B of Prepayments and the Cash Burn Is Three Times the Headline
ORCL financial model
Income Statement · Dollars in millions, except per share
| Income Statement | ||||||
|---|---|---|---|---|---|---|
| Actual | Estimate | |||||
| FY24 | FY25 | FY26 | FY27E | FY28E | FY29E | |
| Cloud Revenue (SaaS + IaaS) | $19,774 | $24,506 | $33,989 | $56,218 | $85,442 | $121,328 |
| Software Revenue | 24,690 | 24,724 | 24,541 | 24,296 | 24,053 | 23,812 |
| Hardware Revenue | 3,066 | 2,936 | 3,084 | 3,207 | 3,304 | 3,403 |
| Services Revenue | 5,431 | 5,233 | 5,743 | 6,317 | 6,823 | 7,232 |
| Total Revenues | $52,961 | $57,399 | $67,357 | $90,039 | $119,621 | $155,775 |
| Less: Cloud and Software Expenses | ($9,427) | ($11,569) | ($17,597) | ($33,234) | ($44,948) | ($55,153) |
| Less: Hardware Expenses | (891) | (782) | (868) | (898) | (925) | (953) |
| Less: Services Expenses | (4,825) | (4,576) | (4,556) | (4,991) | (5,390) | (5,713) |
| Less: Sales and Marketing | (8,274) | (8,651) | (8,331) | (8,103) | (9,057) | (10,904) |
| Less: Research and Development | (8,915) | (9,860) | (10,272) | (10,001) | (10,812) | (12,774) |
| Less: General and Administrative | (1,548) | (1,602) | (1,618) | (1,602) | (1,914) | (2,337) |
| Less: Amortization of Intangible Assets | (3,010) | (2,307) | (1,671) | (1,024) | (814) | (779) |
| Less: Restructuring and Other | (718) | (374) | (1,838) | (377) | (359) | (389) |
| Total Operating Expenses | ($37,608) | ($39,721) | ($46,751) | ($60,229) | ($74,219) | ($89,002) |
| Operating Income | $15,353 | $17,678 | $20,606 | $29,810 | $45,402 | $66,773 |
| Less: Interest Expense | ($3,514) | ($3,578) | ($4,599) | ($5,867) | ($6,793) | ($7,431) |
| Non-Operating Income (Expenses), Net | (98) | 60 | 3,547 | 600 | 600 | 600 |
| Income Before Income Taxes | $11,741 | $14,160 | $19,554 | $24,543 | $39,209 | $59,942 |
| Less: Provision for Income Taxes | ($1,274) | ($1,717) | ($2,467) | ($4,663) | ($7,450) | ($11,689) |
| Net Income | $10,467 | $12,443 | $17,087 | $19,880 | $31,759 | $48,253 |
| Less: Preferred Stock Dividends (6.50% Series D) | $0 | $0 | ($103) | ($325) | ($325) | ($325) |
| Net Income Available to Common Shareholders | $10,467 | $12,443 | $16,984 | $19,555 | $31,434 | $47,928 |
| EPS — Basic | $3.82 | $4.46 | $5.94 | $6.64 | $10.48 | $15.91 |
| EPS — Diluted | $3.71 | $4.34 | $5.83 | $6.55 | $10.35 | $15.71 |
| Ratios & Assumptions | ||||||
| YoY Cloud Revenue Growth | 24.5% | 23.9% | 38.7% | 65.4% | 52.0% | 42.0% |
| YoY Software Revenue Growth | (2.0%) | 0.1% | (0.7%) | (1.0%) | (1.0%) | (1.0%) |
| YoY Hardware Revenue Growth | (6.4%) | (4.2%) | 5.0% | 4.0% | 3.0% | 3.0% |
| YoY Services Revenue Growth | (2.9%) | (3.6%) | 9.7% | 10.0% | 8.0% | 6.0% |
| YoY Total Revenue Growth | 6.0% | 8.4% | 17.3% | 33.7% | 32.9% | 30.2% |
The full workbook adds 16 historical and 8 projected quarters, plus KPI Drivers · Balance Sheet · Cash Flow Statement — every subtotal a live formula, every projection traced to a driver.
Key Takeaways
- The conversion quarter arrived, and it was not close. Total revenue of $19.345B grew 30% in both dollars and constant currency, above the +27% to +29% guide. Cloud infrastructure rose 121% to $7.388B, accelerating from +93% in the prior quarter. Non-GAAP EPS of $1.92 cleared the $1.72 to $1.76 guide by $0.16 and the $1.74 consensus by $0.18. Management raised the full-year guide to at least $90B of revenue and $8.10 of non-GAAP EPS, up from $8.05.
- The record cash flow is engineered, and the gap is large. Operating cash flow of $23.103B was a company record, but $11.363B of it is customer prepayments carrying a significant financing component, against nil in the year-ago quarter. Net that out and operating cash flow is roughly $11.7B against $28.499B of capital expenditure, so the underlying burn is closer to $16.8B than the reported $5.4B of negative free cash flow.
- The renewal disclosure is the most valuable new fact on the call. Capacity that came up for renewal was renewed or resold at a 20% premium to the prior contract, and the majority of those accelerators are four years old or more, with fleet utilization at 97.9%. That is the most direct evidence yet against the bear argument that AI infrastructure economically obsoletes faster than Oracle depreciates it.
- The operating-margin defense is a cost program, not a mix improvement. Implied gross margin fell roughly 725 basis points year over year, yet GAAP operating margin expanded 612 basis points, because sales and marketing fell 12% and research and development fell 4% on 30% revenue growth. After the quarter closed, management expanded the 2026 Restructuring Plan by about $700M to roughly $2.8B. The lever works. It is also finite.
- Rating: Maintaining Outperform, conviction raised to 7 from 6. Management cleared four of the six commitments we set after the prior quarter: the guide went up rather than held, the reported-versus-net capex bridge was confirmed, the $20B at-the-market program was completed in full, and a CFO one quarter into the job raised numbers she inherited. The two that went unanswered were customer concentration and standalone infrastructure economics, and they are the two that would move the rating. We trim fair value to $215 to $265 from $300 to $360, which holds the implied upside roughly constant against a stock that has de-rated 53% in twelve months and now trades at 18.3x the company's own FY27 earnings guide.
Results vs. Consensus
Q1 FY2027 Scorecard
| Metric | Q1 FY2027 actual | Consensus / guide | Beat/Miss | Magnitude |
|---|---|---|---|---|
| Total revenue | $19.345B | $19.13B consensus; +27% to +29% guide | Beat | +1.1% vs. consensus; +30% YoY, above guide |
| Total revenue (constant currency) | +30% | n/a | Beat | No currency help: USD and cc identical |
| Total cloud revenue | $11.607B | +58% to +64% guide | Beat | +62% YoY, upper half of guide |
| Cloud infrastructure (IaaS) | $7.388B | n/a | Beat | +121% YoY; +27.7% sequentially |
| Cloud applications (SaaS) | $4.219B | n/a | In line | +10% YoY, inside a 10% to 13% band for five quarters |
| Non-GAAP operating income | $8.151B | n/a | Beat | +31% YoY; 42% margin, +35bp |
| Non-GAAP EPS | $1.92 | $1.74 consensus; $1.72 to $1.76 guide | Beat | +$0.18 vs. consensus (+10.3%); +$0.16 above the guide high end |
| GAAP EPS (diluted) | $1.56 | n/a | Beat | +55% YoY; GAAP operating margin 35% |
| Operating cash flow | $23.103B | n/a | Record | +184% YoY, but $11.4B is customer prepayment |
| Free cash flow | −$5.396B | n/a | Negative | Capex $28.499B vs. $8.502B a year ago |
| RPO | $664B | n/a | Mixed | +$209B YoY (+45.9%); only +$26B sequentially |
| FY27 revenue guide | At least $90B | $90B prior guide | Raised | +34% YoY |
| FY27 non-GAAP EPS guide | $8.10 | $8.05 prior guide; ~$8.07 consensus | Raised | +$0.05 vs. the standing guide |
Year-Over-Year Comparisons
| Metric | Q1 FY2027 | Q1 FY2026 | YoY change |
|---|---|---|---|
| Cloud revenue | $11,607M | $7,186M | +62% (+61% cc) |
| Cloud infrastructure (IaaS) | $7,388M | $3,347M | +121% (+120% cc) |
| Cloud applications (SaaS) | $4,219M | $3,839M | +10% |
| Software revenue | $5,550M | $5,721M | −3% |
| Hardware revenue | $774M | $670M | +15% (+16% cc) |
| Services revenue | $1,414M | $1,349M | +5% |
| Total revenue | $19,345M | $14,926M | +30% |
| Cloud and software cost of revenues | $6,400M | $3,607M | +77% |
| Sales and marketing | $1,811M | $2,063M | −12% |
| Research and development | $2,401M | $2,491M | −4% |
| General and administrative | $376M | $376M | 0% |
| Total operating expenses | $12,617M | $10,649M | +18% |
| GAAP operating income | $6,728M | $4,277M | +57% |
| GAAP operating margin | 35% | 29% | +612bp |
| Non-GAAP operating income | $8,151M | $6,236M | +31% |
| Non-GAAP operating margin | 42% | 42% | +35bp |
| Interest expense | $1,428M | $923M | +55% |
| GAAP net income | $4,760M | $2,927M | +63% |
| Preferred stock dividends | $81M | nil | New this year |
| GAAP EPS (diluted) | $1.56 | $1.01 | +55% |
| Non-GAAP EPS (diluted) | $1.92 | $1.47 | +30% |
| Diluted share count | 3,000M | 2,909M | +3% |
| Depreciation | $3,156M | $1,351M | +134% |
| Operating cash flow | $23,103M | $8,140M | +184% |
| Capital expenditure | $28,499M | $8,502M | +235% |
| RPO | $664B | $455B | +$209B (+45.9%) |
Sequential Comparisons
Oracle's first fiscal quarter is normally the seasonal trough after a heavy May close. It was not this time, and management made a point of saying so.
| Metric | Q1 FY2027 | Q4 FY2026 | Sequential change |
|---|---|---|---|
| Total revenue | $19,345M | $19,184M | +0.8% |
| Cloud revenue | $11,607M | $9,913M | +17.1% |
| Cloud infrastructure | $7,388M | $5,787M | +27.7% |
| Cloud applications | $4,219M | $4,126M | +2.3% |
| Software revenue | $5,550M | $6,824M | −18.7% (Q4 license seasonality) |
| Hardware revenue | $774M | $924M | −16.2% |
| Services revenue | $1,414M | $1,523M | −7.2% |
| Non-GAAP operating income | $8,151M | $8,590M | −5.1% |
| Non-GAAP EPS | $1.92 | $2.11 | −9.0% (the prior quarter carried a one-time gain) |
| RPO | $664B | $638B | +$26B (+4.1%) |
For context on the seasonal claim, the equivalent step a year earlier was Q4 FY2025 revenue of $15,903M into Q1 FY2026 revenue of $14,926M, a 6.1% sequential decline. A first quarter that grows sequentially at all is a break in Oracle's normal shape, and it is the clearest single sign that delivered capacity, not bookings, now sets the revenue line.
"And for the first time, Q1 total revenue grew sequentially, an important sign of our continued progress in building scaled infrastructure. Historically, a record Q4 was followed by a lighter Q1. But as we accelerate across the full technology stack, from infrastructure to database to software, that is no longer the case." — Hilary Maxson, CFO
Revenue
This is the first quarter in which the backlog stopped being a promise and became a revenue line. Cloud infrastructure at $7.388B grew 121% year over year and 27.7% sequentially, and it alone contributed $4.0B of the company's $4.4B of year-over-year revenue growth. Everything else is close to static: software fell 3%, services grew 5%, hardware added $104M. Oracle is now, functionally, an AI infrastructure company with a large and slowly eroding software annuity attached to it.
Cloud applications grew 10%, which will strike some readers as unremarkable. The internal composition is better than the headline. Fusion grew 14% and industry applications grew above 20%, with NetSuite the drag as a slower decision cycle from the prior fiscal year works through the base. The applications line is no longer the growth story, but it is the margin ballast and the distribution channel for everything Oracle wants to sell on top of it.
The geographic split shows the concentration plainly. Americas revenue of $13,711M grew 41.9% while Europe, Middle East and Africa grew 7.0% and Asia Pacific grew 7.0%. The AI build is a United States phenomenon in Oracle's numbers, which matters for the site-risk question below: the revenue depends on a small number of very large domestic campuses arriving on schedule.
Margins
Oracle does not print a gross profit subtotal, so the gross-margin debate has to be reconstructed from the cost lines. Adding cloud and software cost of revenues ($6,400M), hardware cost ($281M) and services cost ($1,052M) gives $7,733M against $19,345M of revenue, an implied gross margin of 60.0%. The same calculation a year earlier gives $4,884M against $14,926M, or 67.3%. The compression is roughly 725 basis points, and it is entirely the infrastructure ramp: cloud and software cost of revenues grew 77% while the associated revenue grew 33%.
GAAP operating margin nonetheless expanded 612 basis points, from 29% to 35%. The bridge is arithmetic and worth writing out. Operating expenses below the cost-of-revenue lines (sales and marketing, research and development, general and administrative, intangibles amortization, restructuring) totalled $4,884M this quarter against $5,765M a year ago. As a share of revenue those costs fell from 38.6% to 25.2%, a 1,340 basis point improvement that more than absorbs the 725 basis point gross-margin decline. The two moves reconcile to the +612 basis points the company reported.
That bridge is the single most important thing to understand about Oracle's current earnings power. The operating margin is being held by shrinking absolute operating expense dollars during a period of 30% revenue growth, not by improving economics in the business that is actually growing. Sales and marketing spend fell 12% in dollars. Research and development fell 4% in dollars, taking it from 16.7% of revenue to 12.4%. Some of that is real structural efficiency. But a company that cuts research and development in dollar terms during the most contested platform transition in enterprise software has made a choice, and the choice has a cost attached somewhere further out.
The segment note in the quarterly filing gives the cleanest view of where the pressure sits. On a directly-controllable-cost basis, the cloud and software line of business earned a $9,358M margin on $17,157M of revenue (54.5%), against $7,691M on $12,907M (59.6%) a year ago, a decline of roughly 505 basis points. Services margin improved from 24.6% to 31.5%, hardware margin fell from 66.7% to 58.4%, and total operating-segment margin fell from 56.7% to 53.0%.
"Our gross margin did decline as expected driven by impacts from ramping up our data centers and the acceleration of infrastructure revenue. However, this was offset in the quarter by lower operating costs and strong operating leverage tied to simplification and efficiency actions." — Hilary Maxson, CFO
EPS
Non-GAAP EPS of $1.92 is clean of the kind of one-time investment gain that flattered two of the prior four prints. The adjustments are the usual three: $1,127M of stock-based compensation, $202M of intangibles amortization, $94M of restructuring and other, for $1,423M in total. That is a narrower bridge than a year ago ($1,959M), mostly because the restructuring and amortization lines shrank.
Two below-the-line items deserve attention. Interest expense grew 55% to $1,428M, an annualized run-rate above $5.7B, and preferred stock dividends of $81M appear for the first time following the mandatory convertible preferred issued in the prior fiscal year. Together they represent a roughly $6B annual claim on operating income that sits ahead of the common. Offsetting that, the below-the-line drag shrank relative to the size of the business: net interest and non-operating items consumed 16.7% of GAAP operating income this quarter against 19.9% a year ago.
The share count is the quiet item. Diluted shares of 3,000M are up 3% year over year and basic shares of 2,966M are up 5%. The quarterly filing confirms that Oracle issued approximately 141 million shares under the at-the-market program for net proceeds of $19.9B, which works out to roughly $141 per share after issuance costs. Oracle sold about 5% of itself at a price 57% below the $328.33 at which the shares closed twelve months earlier. The capital was needed and the program is now closed, but the cost is permanent and it appears nowhere in the non-GAAP presentation.
Segment Performance
Revenue by Offering
| Line | Q1 FY27 revenue | YoY (USD) | YoY (cc) | % of total | Read |
|---|---|---|---|---|---|
| Cloud infrastructure (IaaS) | $7,388M | +121% | +120% | 38% | The entire growth story; accelerated from +93% |
| Cloud applications (SaaS) | $4,219M | +10% | +10% | 22% | Steady; Fusion +14%, industry apps above 20% |
| Software support | $4,895M | −1% | −1% | 25% | The annuity; flat is the objective |
| Software license | $655M | −15% | −14% | 3% | Structural decline as on-premise migrates |
| Services | $1,414M | +5% | +5% | 7% | Decelerated from +13% in the prior quarter |
| Hardware | $774M | +15% | +16% | 4% | Best rate in at least five quarters |
| Total | $19,345M | +30% | +30% | 100% | Income-statement total |
Geographic Mix
| Region | Q1 FY27 | Q1 FY26 | YoY | % of total |
|---|---|---|---|---|
| Americas | $13,711M | $9,662M | +41.9% | 70.9% |
| Europe / Middle East / Africa | $3,726M | $3,481M | +7.0% | 19.3% |
| Asia Pacific | $1,908M | $1,783M | +7.0% | 9.9% |
| Total | $19,345M | $14,926M | +29.6% | 100% |
Regional growth rates in this table are computed from the disclosed dollars, so the total reads 29.6% where the company reports total revenue growth as 30%.
Operating-Segment Margins
These margins reflect only directly controllable costs. They exclude research and development, general and administrative expense, stock-based compensation, intangibles amortization, restructuring, interest and non-operating items.
| Line of business | Q1 FY27 revenue | Q1 FY27 margin | Margin % | Q1 FY26 margin % | Change |
|---|---|---|---|---|---|
| Cloud and software | $17,157M | $9,358M | 54.5% | 59.6% | −505bp |
| Hardware | $774M | $452M | 58.4% | 66.7% | −832bp |
| Services | $1,414M | $445M | 31.5% | 24.6% | +686bp |
| Total | $19,345M | $10,255M | 53.0% | 56.7% | −374bp |
Cloud Infrastructure: 121%, and the Bottleneck Is Now Construction
Cloud infrastructure added $4.0B of year-over-year revenue on a $3.3B base. The mechanism was delivery, not selling. Oracle brought 850 megawatts of AI capacity online since the close of the prior quarter and shipped more than 300,000 accelerators into it, a pace management put at roughly three times the prior quarter's deliveries and nearly three-quarters of everything delivered in the whole of the prior fiscal year.
"Delivery in Q1 is almost 3x what we delivered in all of Q4, and 73% of the total capacity we delivered last fiscal year." — Clay Magouyrk, co-CEO
The Abilene campus is the reference implementation. Six of eight buildings, 618 megawatts and 75% of total site capacity are now delivered, 131,000 accelerators went in during the quarter alone, and customer acceptance has compressed to 24 hours. Shackleford is the next gigawatt-scale campus, and the first NVIDIA Vera systems ship to customers in the current quarter.
"6 of the 8 campus buildings, representing 618 megawatts and 75% of total capacity have now been delivered to the customer. Customer acceptance has compressed to only 24 hours, showing that the systems arrive ready for customer workloads." — Clay Magouyrk, co-CEO
Assessment: The 24-hour acceptance figure is the operational tell. A year ago the argument against Oracle was that it could book hyperscale contracts but not build hyperscale capacity. That argument is now dead, and the constraint has moved to land, power and permits. What has still not been demonstrated is standalone profitability, because Oracle does not disclose infrastructure-level economics and the cloud and software segment margin fell 505 basis points as this capacity came online.
Cloud Applications: 10% Headline, Better Underneath
Applications at $4.219B grew 10%, the same rate as the prior quarter and inside a 10% to 13% band across the last five quarters, with the components diverging. Fusion grew 14%, industry applications grew above 20%, Oracle Health continued to accelerate, and NetSuite lagged on decision cycles carried over from the prior fiscal year. NetSuite Next reached general availability during the quarter, and the NetSuite AI connector service, which links customer data to third-party assistants, passed 10,000 customers.
The embedded AI telemetry is the more interesting disclosure, because it is the first time Oracle has quantified consumption rather than availability: more than 150 million uses of embedded AI with usage up 42% sequentially, more than 3.5 million agent executions with the count roughly doubling sequentially, over 2,300 agents in production up 90% sequentially, and 900 billion tokens consumed across Fusion.
"In total, our SaaS business grew 10% with Fusion growing at 14%. Our Oracle Health business continued to accelerate. And although we do not specifically call it out, our industry applications grew at greater than 20% in Q1." — Mike Sicilia, co-CEO
Assessment: Last quarter the case for an applications reacceleration rested on deferred revenue growing faster than recognized revenue. This quarter it rests on usage telemetry. Neither has yet moved the recognized line, which has run 11%, 11%, 13%, 10% and 10% across the last five quarters. We treat 10% as the durable rate and would want two consecutive quarters above 13% before modelling anything better. The agentic usage figures matter more as evidence that the installed base is not leaving than as a near-term revenue driver.
Multicloud Database: 353% and a Completed Footprint
Oracle Database running inside AWS, Azure and Google Cloud grew 353% year over year with the multicloud customer count up 180%. Oracle completed its planned Azure and AWS regional expansion, reaching 70 multicloud database regions across 119 availability zones, and made Oracle Interconnect for AWS generally available.
"We also made Oracle interconnect for AWS generally available. With this launch, OCI now has private high speed connections to all hyperscalers with no data transfer charges." — Clay Magouyrk, co-CEO
Assessment: This remains the most underrated line in the model and the only high-growth line with an attractive capital profile, because Oracle monetizes the database moat without owning the data center. The deceleration from 404% to 353% is base arithmetic rather than a problem. Completing the regional footprint matters more than the growth rate: the build phase is over and the consumption phase begins, which is when the line should stop being a rounding error in the revenue mix.
Software, Services and Hardware: the Static Two Thirds
Software support of $4,895M declined 1% and software license of $655M declined 15%, for a combined software line of $5,550M, down 3%. Services grew 5%, a marked deceleration from 13% in the prior quarter, which is worth watching because implementation work should lead application go-lives. Hardware grew 15%, its best rate in at least five quarters.
Assessment: The software annuity is behaving exactly as modelled, declining at roughly 1% a year at the support line with license running off faster. Nothing here changes. The services deceleration is the item to track: if AI-assisted go-lives really do compress implementation timelines from months to weeks as management claims, services revenue should fall while application revenue accelerates. That would be a good trade, but only the first half of it has happened so far.
Key Operating KPIs
| KPI | Q1 FY2027 | Prior reference | Trend | Read |
|---|---|---|---|---|
| RPO | $664B | $638B (Q4 FY26); $455B (Q1 FY26) | +$26B QoQ; +45.9% YoY | Growth rate normalizes as conversion begins |
| RPO recognized within 12 months | 13%, roughly $86B | 12% at Q4 FY26 | Improving | Near-term visibility firming |
| RPO recognized within 36 months | 50%, roughly $332B | ~46% at Q4 FY26 | Improving | Duration shortening as capacity lands |
| New AI contracts signed | Over $30B | $67B in Q4 FY26 | Lower, by design | Requires no incremental Oracle cash |
| Capacity delivered | 850 MW | Roughly one third of that in Q4 FY26 | ~3x sequentially | 73% of all prior-year deliveries in one quarter |
| Accelerators delivered | Over 300,000 | n/a | Step-function | 131,000 of them at Abilene alone |
| Fleet utilization | 97.9% | 97.5% (Q4 FY26) | Tighter | Still supply-constrained |
| Renewal pricing | +20% premium | Not previously disclosed | New | On accelerators mostly four years or older |
| Customer prepayments received | $11.4B | nil in Q1 FY26 | New line item | Roughly half the record operating cash flow |
| Multicloud database revenue | +353% YoY | +404% (Q4 FY26) | Base effect | Customers +180%; footprint complete |
| Embedded AI invocations | Over 150M | n/a | +42% QoQ | First consumption disclosure |
| Agents in production | Over 2,300 | n/a | +90% QoQ | 3.5M+ executions, roughly doubled QoQ |
| Equity issued under the ATM | 141M shares, $19.9B net | $20B program announced at Q4 FY26 | Complete | Roughly $141 per share after costs |
| Total debt | $125.3B | $129.5B at May 31, 2026 | −$4.2B QoQ | Equity proceeds partly retired debt |
Key Topics & Management Commentary
Overall Management Tone: Assured on execution, evasive on cash timing, and noticeably more comfortable than one quarter ago. Where the prior call was spent defending a capital plan the market had just rejected, this one opened with the word "acceleration" and spent its prepared remarks on delivery metrics rather than on framework metrics. The single soft spot was the refusal to put any date on a return to positive free cash flow, and the accompanying refusal to quantify a gross-margin trough; both were deferred to an investor day still a month away. Nothing on the call was less convincing than the answer to the question of when the building stops.
1. RPO at $664B: the Growth Rate Breaks, the Conversion Starts
Remaining performance obligations reached $664B, up $209B year over year. That is a 45.9% growth rate against the 363% Oracle printed one quarter earlier, and the collapse in the rate is arithmetic rather than demand: the comparison base moved from $138B to $455B. Sequentially the backlog added only $26B, the smallest quarterly addition since the step-function began, against more than $30B of new contracts signed. The gap is conversion: revenue is now consuming backlog faster than it did.
The recognition ladder in the quarterly filing is the useful disclosure. Oracle expects to recognize approximately 13% of the $664B within twelve months (roughly $86B), 37% in months thirteen through thirty-six (roughly $246B), 34% in months thirty-seven through sixty (roughly $226B) and the remaining 16% (roughly $106B) beyond five years. One quarter earlier the first two buckets were 12% and 34%. Both moved forward.
"First, we continued to grow our RPO during the quarter to support future revenues, and the vast majority of those new contracts were via prepay or bring your own hardware or a similar mechanic so will not require incremental capital from Oracle. Also, that new RPO will not impact our CapEx or revenues until fiscal 28 or beyond." — Hilary Maxson, CFO
"We now expect around half of our RPO to convert into sales over the next 36 months." — Hilary Maxson, CFO
Assessment: The headline rate collapsing from 363% to 46% will read as deceleration to anyone who does not check the base, and the sequential add of $26B is genuinely the weakest of the cycle. But the ladder moving forward is worth more than the rate: four percentage points of a $664B backlog pulled inside three years is roughly $26B of revenue moved earlier, and the twelve-month bucket at $86B is now in the same neighbourhood as the entire FY27 revenue guide. The backlog has stopped being a headline and started being a schedule. That is the trade we wanted.
2. The $11.4B Prepayment: How the Record Cash Flow Was Made
Operating cash flow of $23.103B is the largest in Oracle's history and grew 184% year over year. A single new line explains most of it. The cash flow statement carries $11.363B under "increase in deferred revenues from customer prepayments with significant financing component," a caption that did not exist a year ago. The quarterly filing is explicit that no such prepayments were received in the year-ago quarter, and that this one quarter brought in $11.4B against $4.6B across the whole of the prior fiscal year.
The arithmetic that follows is uncomfortable. Net the prepayment out of operating cash flow and the underlying figure is roughly $11.7B. Set that against $28.499B of capital expenditure and the underlying free cash burn is roughly $16.8B, not the $5.4B the reported numbers show. The prepayment is real cash and it genuinely reduces Oracle's funding need. It is also non-recurring by construction: a customer pays once, and the revenue is then recognized over the contract without further cash.
A second detail belongs with it. The filing notes that Oracle recognizes interest expense on the financing component of these prepayments separately from revenue, and that in this quarter the amount was immaterial. It will not stay immaterial. These are multi-year prepayments carrying an implicit financing cost at something like Oracle's incremental borrowing rate, which means a growing non-cash interest charge sits behind the cash Oracle is collecting today.
Assessment: This is the most important thing in the quarter that management did not volunteer. The prepayment model is a genuine innovation and it is why the capital plan is fundable at all, but presenting a prepayment-inflated operating cash flow as a record without sizing the underlying figure invites exactly the scepticism Oracle keeps attracting. We will track free cash flow on both bases going forward and we would encourage the company to do the same before someone else does it for them.
3. The Renewal Premium: Four-Year-Old Accelerators Priced 20% Higher
The most valuable disclosure on the call was one sentence in the prepared remarks. Capacity that came up for renewal in the quarter was renewed or resold at a 20% premium to the prior contract, and the majority of those accelerators are four years old or more. Fleet utilization was 97.9%, slightly tighter than the 97.5% disclosed one quarter earlier.
"Of all the GPUs that came up for renewal in Q1, that capacity was renewed or resold at a 20% premium to prior contracts. The majority of those GPUs are 4 years or older. We see a long useful life with increasing value for the AI capacity we are deploying." — Clay Magouyrk, co-CEO
The significance is the depreciation debate. The central bear argument against every AI infrastructure balance sheet is that accelerators are being written down over five or six years while their economic life is closer to three, which would mean reported earnings across the sector are overstated by the difference. A four-year-old asset re-contracting at a 20% price increase into a 97.9%-utilized fleet is direct evidence against that.
Assessment: One quarter of renewals at a premium is not proof, and the sample is not sized. But this is the first hard datum any hyperscale operator has put on the record about what older AI capacity actually re-prices at, and it points the opposite way from the consensus fear. If it repeats next quarter it materially de-risks the depreciation schedule, which in turn de-risks the GAAP earnings the whole valuation case rests on. We want the sample size and the renewal cohort disclosed at the investor day.
4. Gross Margin Down 725bp, Operating Margin Up 612bp
Management framed the gross-margin decline as expected and offset, and the offset is real. What management did not do is quantify the trough, which is the second consecutive quarter in which they have declined to do so. The composition matters: this is not a pricing problem, it is a depreciation and ramp problem, with depreciation up 134% year over year to $3.156B.
"So gross margin, like you mentioned, the moment, there is a number of things going on. We have both the ramp up in data centers, plus we have an adjustment across the 2 business models that we have in the business. Software being a much higher gross margin business, but with higher R&D and sales costs below gross margin. Infrastructure being a lower gross margin business, and we have talked about that and we gave the numbers." — Hilary Maxson, CFO
The mix argument is correct as far as it goes. Software carries high gross margin and heavy research and sales costs below the line; infrastructure carries low gross margin and almost no incremental research or sales cost, because it reuses work already done. A blended company shifting from the first shape to the second should see gross margin fall and operating margin hold. That is exactly what the numbers show.
Assessment: We accept the framework and we are more comfortable with it than we were a quarter ago, because this is now the second quarter in which it has held empirically. The caveat is that the offset this quarter came disproportionately from absolute cost reduction rather than from the structural argument: research and development fell in dollars and sales and marketing fell 12%. The structural version of the argument requires operating costs to grow slowly, not shrink. We will treat operating margin as defended, and gross margin as still without a floor, until the investor day puts a number on it.
5. Capex of $28.5B in One Quarter, and the $90B to $95B Frame Holds
Capital expenditure was $28.499B in a single quarter against $8.502B a year ago, and against $55.663B for the entirety of the prior fiscal year. The filing attributes the increase directly to data-center expansion. Management reaffirmed the full-year framing set one quarter earlier without changing it.
"To note, our CapEx will not be linear throughout the year. We continue to anticipate $90 billion to $95 billion in CapEx for the full year with not more than $70 billion in net cash CapEx." — Hilary Maxson, CFO
Net cash capital expenditure for the quarter, after prepayments, was $18B on management's figures. The difference between the $28.5B reported and the $18B net is the prepayment mechanism working as designed. Annualizing the reported quarterly figure would give $114B, well above the guide, which is what the non-linearity warning is for: Abilene deliveries were front-loaded and the back half should moderate.
Assessment: The guide was reaffirmed rather than raised, which is the first time since Q3 FY2026 that Oracle has not moved its capital number upward. That is a small but real positive. The number itself is still extraordinary: $90B to $95B of capital expenditure against $90B of revenue means Oracle will spend roughly a dollar of capital for every dollar of revenue it books this year. No software company has ever done this, and the only relevant comparisons are utilities and telecoms, which trade at utility multiples. The equity case depends on Oracle ending up with hyperscale returns rather than utility returns, and that question will not be settled this fiscal year.
6. The $20B Equity Program Completed in a Single Quarter
The at-the-market equity program announced one quarter earlier was finished in full within the quarter: approximately 141 million shares for net proceeds of $19.9B. Management framed the completion as an achievement, and on funding grounds it is one.
"Lastly, we are quite pleased to announce that we completed our previously disclosed $20 billion equity issuance in entirety during the Q1." — Hilary Maxson, CFO
On per-share grounds it is more ambiguous. Dividing the net proceeds by the share count gives roughly $141 per share, which is 57% below the $328.33 close of twelve months earlier and below where the stock traded on the day of the print. Oracle sold roughly 5% of the company near the low of a two-year de-rating in order to fund a build whose returns are still unproven.
Assessment: The speed is the good news and it is genuinely good news: the dilution overhang that hung over the stock all summer is now behind the company rather than ahead of it, and the guide already absorbs the higher share count. The price is the bad news, and it is permanent. This is the clearest single example of why negative free cash flow is not a neutral timing artifact. A company that funds a build from operating cash flow dilutes nobody. A company that funds it from equity issued at a 57% drawdown transfers value from existing holders to new ones, and no ROIC framework recovers that.
7. Debt Went Down, Equity Went Up, and the Balance Sheet Re-Mixed
Total borrowings fell from $129.541B at the May year-end to $125.337B at the quarter-end, a $4.2B reduction driven by $4.202B of scheduled senior-note and term-loan repayments. Stockholders' equity rose from $43.056B to $67.196B, almost entirely the equity issuance plus retained earnings. Cash and equivalents rose to $36.369B. Property, plant and equipment climbed from $99.957B to $127.845B in three months.
There are two large items that do not appear in the debt figure. Operating lease liabilities stand at $30.594B non-current, with right-of-use assets of $33.967B, and the filing discloses data-center leases already signed that commence in fiscal 2028 and fiscal 2029 with terms of fifteen to nineteen years, none of which is on the current balance sheet. Separately, unconditional purchase obligations for cloud infrastructure components and data-center power run to $5.449B for the remainder of this fiscal year, $6.176B in fiscal 2028 and $3.863B in fiscal 2029.
Assessment: On the reported numbers the balance sheet improved this quarter, and it is fair to say so. Net debt of roughly $88B against $67B of equity is a materially better position than three months ago. The qualification is that the leases and the purchase commitments are the same obligation as debt in economic substance, and they are growing faster than the borrowings are shrinking. Anyone underwriting Oracle's leverage on the notes-payable line alone is looking at the smaller half of the number.
8. Restructuring Expanded to Roughly $2.8B After the Quarter Closed
The quarterly filing discloses that the 2026 Restructuring Plan was sized at up to $2.1B as of the August quarter-end, and that management supplemented it by approximately $700M after the quarter closed, taking the total to roughly $2.8B. The plan is described in the filing as tied to strategic measures and operational efficiency, "including through the adoption and integration of artificial intelligence technologies across certain functions." Restructuring expense recognized in the quarter under the plan was $167M against $415M a year ago; the income statement line reads $94M because it nets other operating income against the plan charge.
This disclosure did not appear in the press release and was not raised on the call. It surfaced in the filing the following day.
Assessment: The expansion matters for two reasons. It is the mechanism behind the operating-expense reduction that carried the margin this quarter, which means the margin defense has a known cost and a known end date. And it is a $700M addition to a plan that was already the largest in the company's recent history, disclosed one day after a call on which management characterized the cost reductions as "simplification and efficiency actions" without mentioning that more were coming. We do not think this was concealment. We do think the market found out about it on its own, and the share price reaction in the following session suggests it noticed.
9. Site Risk: New Mexico, Wisconsin and the Portfolio Defense
Press reporting before the print had focused on possible delays at Oracle's New Mexico and Wisconsin campuses. Management addressed both directly, and did so in portfolio terms rather than site terms. Neither New Mexico nor Wisconsin nor Michigan nor Shackleford contributed capacity in the quarter, yet Oracle still delivered 850 megawatts, which is itself the argument: the delivery does not depend on any one site.
On New Mexico, construction is described as on track with an air permit still in process and Bloom fuel cells planned for on-site generation. On Wisconsin, power is being delivered through the grid in coordination with the state commission and the local utilities, with data-center delivery described as on track. Management stated that neither site affects the fiscal 2027 revenue or earnings guidance.
Assessment: The portfolio answer is the correct one and the 850 megawatts delivered without those four sites substantiates it. The residual risk is not fiscal 2027, which management has now explicitly insulated, but fiscal 2028, when the gigawatt-class campuses are supposed to carry the load and when the new backlog signed this quarter starts converting. Permitting and grid interconnection are the two things in this business that no amount of capital accelerates.
10. Guidance Raised Rather Than Reaffirmed
Management raised the full-year revenue guide to at least $90B and the non-GAAP EPS guide to $8.10, from $8.05. The second-quarter guide calls for revenue growth of 30% to 34% and cloud revenue growth of 65% to 71% in dollars, with non-GAAP EPS of $1.85 to $1.93 excluding the prior-year investment gain.
"For the full year, reflecting the strong execution and acceleration you see in our Q1, we are upgrading our guidance to at least $90 billion in total revenues. that is a year over year increase of +34% and to $8.10 for our non-GAAP EPS." — Hilary Maxson, CFO
Assessment: The raise is small in dollars and large in signal. A CFO one quarter into the job raising a predecessor's full-year numbers after a single print is taking on personal risk she did not have to take, and she would not do it without visibility into the delivery schedule. The "at least" construction on the revenue line is the tell that $90B is now a floor rather than a target. We read the EPS raise as deliberately conservative relative to the revenue raise, which is consistent with a gross-margin trough management still will not size.
11. The Applications Argument Against the AI Disruption Thesis
A large part of the call was devoted to the proposition that AI strengthens rather than threatens packaged applications. The argument is that suites already encoded the business rules, the security model and the regulatory logic, and that agents are a better interface to those assets rather than a replacement for them. Management also announced an agentic accelerator for implementations, to be detailed at the October event.
"The introduction of AI is an accelerator, not a replacement for packaged applications." — Mike Sicilia, co-CEO
The proof points offered were implementation timelines rather than revenue. Complex healthcare go-lives described as previously taking high double-digit months are now being completed in single months, and some NetSuite implementations have come down from double-digit months to single-digit weeks.
Assessment: The argument is coherent and Oracle is better positioned to make it than most, because it owns the applications, the database and the infrastructure underneath. Compressed go-lives are also a real revenue accelerant, since revenue recognition on many of these contracts is gated on going live. The gap between the argument and the evidence is that the recognized applications line has not moved: five quarters inside a 10% to 13% band is not what a reacceleration looks like. We score this as credible and unproven.
12. The AI Data Platform and the Ontology Claim
Oracle used the press release to position its new AI Data Platform explicitly against the hand-crafted enterprise ontology approach, claiming full automation of ontology creation from any data source. On the call management described the platform as agnostic to whether the underlying data sits in Oracle systems at all, and confirmed that Oracle is deploying forward-deployed engineers alongside it.
Assessment: Naming a competitor's methodology in an earnings release is unusual for Oracle and signals that management sees the enterprise-data-reasoning layer as contested ground worth claiming publicly. Commercially this is immaterial to fiscal 2027 and the business model was described as "all the above," which is another way of saying it has not been decided. Strategically it is the right place for Oracle to fight, because ontology automation is the one thing that plausibly converts a database installed base into an AI installed base. We would want a disclosed revenue or customer metric before crediting anything to it.
Guidance & Outlook
| Metric | Q2 FY2027 guide | FY2027 guide | Prior FY2027 guide | Change |
|---|---|---|---|---|
| Total revenue growth (USD) | +30% to +34% | At least $90B (+34%) | ~$90B (+34% cc) | Raised to a floor from a target |
| Cloud revenue growth (USD) | +65% to +71% | n/a | n/a | Accelerating from +62% |
| Cloud revenue growth (cc) | +64% to +70% | n/a | n/a | n/a |
| Non-GAAP EPS (USD) | $1.85 to $1.93 (+21% to +25%) | $8.10 | $8.05 | Raised $0.05 |
| Non-GAAP EPS (cc) | $1.83 to $1.91 (+19% to +23%) | n/a | n/a | n/a |
| Capex (reported) | n/a | $90B to $95B | $90B to $95B | Maintained |
| Capex (net cash) | $18B in Q1 actual | Not more than $70B | ~$70B | Maintained, now capped |
| Quarterly dividend | $0.50 per share | n/a | $0.50 | Maintained |
The second-quarter EPS guide carries a footnote that matters. The $1.85 to $1.93 range and its +21% to +25% growth exclude the one-time Ampere-related investment gain booked in the year-ago second quarter. Include that gain in the base and the same guide represents a decline of 18% to 14%. Readers scanning headline growth rates next quarter should expect that optical decline and should not read it as deterioration.
Implied ramp: A full-year floor of $90B against $19.345B delivered, and a second quarter implying roughly $20.9B to $21.5B against the year-ago $16.058B, leaves approximately $49.1B to $49.8B for the back half, or roughly $24.6B to $24.9B a quarter. That is 27% to 29% above the quarter just reported and requires the second half to run on a materially larger delivered capacity base, which is consistent with management's repeated non-linearity language on both revenue and capital expenditure.
Street at: Consensus full-year EPS sat near $8.07 against a standing guide of $8.05, so the $8.10 raise nudges past both. Revenue consensus was already at the prior $90B guide, which is why the "at least" phrasing carries more information than the number itself.
Guidance style: Oracle has now cleared its own cloud guidance in each of the last several quarters and beat its own EPS guide by $0.16 this time. The pattern is a company that guides the quarter conservatively and the year ambitiously. A new CFO raising an inherited full-year number one quarter in is the more meaningful datum than the size of the raise.
Analyst Q&A Highlights
When the Building Stops, and When Free Cash Flow Turns
The opening line of questioning asked two things at once: what governs the decision to slow capital spending if competitors do not, and when the company returns to positive free cash flow. Management answered the first by separating what Oracle spends from what the business needs, and declined to answer the second at all, offering instead a project-level conversion metric in place of a date.
Q: "And then related to that, given the current state of the backlog as it stands today, when should we expect to see the company return to generating positive free cash flow?"
— Brad Zelnick, Deutsche Bank
A: "We have not given a particular time frame on that. Yet, and we do not expect to give that today. What I would say though is that each of these projects that we are doing, by nature, is a strong free cash flow generating project. So as soon as they ramp up very shortly thereafter, they are delivering a free cash flow conversion ratio of something like 100% to post tax EBITDA."
— Hilary Maxson, CFO
Assessment: This was the least satisfying answer on the call, and it was the first question asked, which tells you where the buy side's attention sits. A 100% conversion of post-tax EBITDA at the project level is a strong claim and it is consistent with the ROIC framework offered last quarter, but it is a steady-state claim about individual projects, not a statement about consolidated cash flow, which depends entirely on how many new projects start. Management can genuinely not answer this without disclosing a capital plan beyond fiscal 2028, which it has chosen not to do. The honest reading is that consolidated free cash flow turns positive when Oracle decides to stop growing, and nobody on the call wanted to say that out loud.
Whether the Named Campuses Put the Fiscal 2027 Guide at Risk
A recurring pre-print concern was possible permitting and power delays at two named campuses. The response reframed the question from site risk to portfolio risk, and made an unusually candid statement about how the company plans for construction slippage.
Q: "So can you give us an update on where both sides actually stand today? And is there any risk to the delivery time line for those data center that could pose any pose a risk to your 2027 revenue expectation?"
— Siti Panigrahi, Mizuho
A: "We talk about these sites kind of as being around a gigawatt a piece. We just delivered 850 megawatts in Q1. … We have a term for that. it is called a bad plan. And so we try real hard not to make bad plans. So we obviously know the difficulty and the complexity of what we are doing, and we do not assume 100% of everything is going to work all the time."
— Clay Magouyrk, co-CEO
Assessment: Management answered the question and then substantiated the answer, which is rarer than it sounds. The substantiation is that 850 megawatts landed in the quarter without any contribution from the four campuses under discussion. Stating plainly that the plan does not assume full achievement of every deliverable is the sort of disclosure that reads as weakness in a press release and as credibility on a call. We take the fiscal 2027 guide as insulated from these two sites and we move the site risk to fiscal 2028.
Component Inflation, Contract Pricing and Whether the Margin Framework Survives
Memory and storage cost inflation has been the sector's standing bear argument, and a line of questioning tested whether it breaks Oracle's margin framework. The answer was that prices move with costs in a supply-constrained market, and the framework holds.
Q: "Clayton, can you talk a little bit about like how your pricing is evolving between contract pricing, spot pricing? And you gave us a framework on gross margins last year. Is that still holding?"
— Raimo Lenschow, Barclays
A: "But, definitely, prices in a world where demand exceeds supply, we typically, prices do not go down. They do go up. And so you know, I think the net effect is that obviously things cost more, but then we have to we have to charge more money for them so that we get compensated. And we are doing that across all of these different businesses. No. We do not we do not expect this to have an impact on our gross margins."
— Clay Magouyrk, co-CEO
Assessment: This is the same pass-through answer given one quarter ago, now backed by the 20% renewal premium rather than by contract structure alone. Pricing power is the cleanest defense against input-cost inflation and Oracle has now produced an actual price increase to point at. The caution is that the answer addresses component costs and not the depreciation and ramp drag, which is the larger driver of the 725 basis point compression and which pricing does not fix.
The Correction on What "No Incremental Capital" Actually Means
A line of questioning sought detail on the customer types behind the backlog that management said required no additional capital. The answer began with an unprompted correction that is more important than the question asked.
Q: "Can you give us more details on the drivers of the growth of RPO that is not requiring additional CapEx, the prepaid and bring your own chips. Is this AI labs and semiconductor companies, or is it that sovereign?"
— Mark Moerdler, Bernstein
A: "I did not say and I do not I do not think myself nor Hillary said that it does not require additional CapEx. We said it does not require additional cash from Oracle."
— Clay Magouyrk, co-CEO
Assessment: The correction is the most important sentence in the Q&A and management volunteered it. The distinction it draws is between capital expenditure, which still flows through Oracle's reported financial statements and its depreciation schedule, and cash, which increasingly comes from customers and vendors. Readers who took "no additional capital" to mean the reported capital expenditure line does not grow were wrong, and management corrected them before anyone built a model on it. That is the behaviour of a team that expects to be held to its words, and it is worth noting in a quarter where the cash presentation was otherwise flattering.
Whether Gross Margin Keeps Falling From Here
The most persistent line of questioning concerned the gross-margin trajectory, and the follow-up pressed specifically on whether margins keep declining or stabilize at current levels. Management gave a directional answer over a multi-year horizon and explicitly declined to guide.
Q: "But how should we think about gross margins going forward in addition to operating margins, that whole ecosystem?"
— John DiFucci, Guggenheim Securities
A: "Over the next couple of years, as we finish the ramp up, you can reasonably expect that gross margin would flatten, I would say. But we have not given any part particular guidance today. We will speak more about some of this stuff on the upcoming investor day though in October."
— Hilary Maxson, CFO
Assessment: "Over the next couple of years" is a longer horizon than anyone wanted and "flatten" is a weaker word than "recover." Read literally, management is telling the market that gross margin has further to fall before it stops falling, and that it may not come back. That is a more bearish statement than the prepared remarks implied, and it is the second consecutive quarter in which the trough has been deferred. The October investor day now carries real weight: if a gross-margin floor is not quantified there, the market will assume the worst and will be entitled to.
Whether AI Breaks the Packaged Applications Business
A line of questioning tested the industry-wide fear that AI erodes seat-based application software. Management's answer moved past the defensive framing and onto implementation economics, which is where the near-term revenue effect actually sits.
Q: "I think you know, there is been this fear of the SaaS business across the industry taking a hit because of AI that seems like that is not your view. Give us your perspective on what you are seeing in terms of keeping sustainable double digit growth."
— Brent Thill, Jefferies
A: "But we have seen, like, for example, complicated go lives in our health care applications that were in the high double digit months coming down to single months for go lives. In our NetSuite applications, we have seen early customers leveraging these AI tools coming down from double digit months down to single digit weeks in order to be able to go live in production."
— Mike Sicilia, co-CEO
Assessment: The strongest version of Oracle's applications argument is not that AI fails to disrupt suites, it is that AI removes the single largest friction in selling them. Implementation length is why enterprise application deals die, and compressing a healthcare go-live from years to months changes the sales cycle and pulls revenue recognition forward. If this is real at scale it shows up first as faster services revenue decline and then as faster applications revenue growth. Services decelerated to +5% this quarter from +13%. The first half of the signature is present. The second half is not, yet.
How the AI Data Platform Is Supposed to Make Money
The closing exchange pressed on the commercial model behind the newly announced data platform: whether it drives database and infrastructure consumption or stands alone as a software line. The answer declined to choose, and volunteered that Oracle is staffing forward-deployed engineers behind it.
Q: "Can you just help us understand the business model around that? Does it drive incremental consumption of Oracle database, OCI, or do you see that being sort of a standalone software-revenue opportunity?"
— Kirk Materne, Evercore ISI
A: "The AI data platform is agnostic. And able to pull in and automate the ontologies from any data source. We have hundreds of data sources that we are we are doing this automation for today."
— Mike Sicilia, co-CEO
Assessment: "All the above" is an honest answer for a product this early and a dodge on the financial question that was asked. The agnostic positioning is strategically right, because a platform that only works on Oracle data is a feature and one that works on any data is a business. The forward-deployed engineer commitment is the detail worth tracking: it is an expensive, services-heavy go-to-market that borrows directly from the competitor Oracle named in its own press release, and it will show up in the cost line well before it shows up in revenue.
What They're NOT Saying
- Customer concentration in the $664B backlog. This is the fourth consecutive quarter in which Oracle has grown the backlog, made it the centre of the equity story, and disclosed nothing about who holds it. Outside estimates put a single counterparty at roughly half. The company has neither confirmed nor rebutted any figure, and did not volunteer a diversification statistic this quarter as it did last quarter. For the single most important number in the investment case, this is the single most important missing disclosure.
- A quantified gross-margin trough. Two quarters in a row management has said gross margin steps down and later flattens, and two quarters in a row it has declined to say by how much or when. The word used this quarter was "flatten," not "recover," over "the next couple of years."
- Any date for positive free cash flow. Asked directly, management declined and said it did not expect to give one. The substituted metric, roughly 100% conversion of post-tax EBITDA at the project level, describes an individual project's steady state and says nothing about the consolidated entity while new projects keep starting.
- Standalone infrastructure economics. Oracle still does not report cloud infrastructure as a separate segment with its own margin. The closest available proxy, the cloud and software line of business, fell 505 basis points. Every claim about infrastructure profitability remains an assertion rather than a disclosure.
- The restructuring expansion. The roughly $700M supplement that took the 2026 plan to $2.8B was disclosed in the quarterly filing the day after the call, not in the press release and not on the call, on a day when management attributed the margin performance to "simplification and efficiency actions."
- The financing cost inside the prepayments. The filing notes that interest expense on the financing component of customer prepayments was immaterial this quarter. It will not stay immaterial as the prepayment balance compounds, and management did not size the forward drag.
- What happens to the backlog if a large customer's demand changes. Cancellation terms, take-or-pay structures and contingency provisions on $664B of largely long-dated contracts remain entirely undisclosed. This was unanswered last quarter and was not raised this quarter.
Market Reaction
- Pre-print setup: ORCL closed the September 10 regular session at $152.94, down 21.5% year to date against an S&P 500 up 10.9%, and down 53.4% over the trailing twelve months from $328.33. The 52-week closing range entering the print was $114.99 to $328.33. The one bright spot was short-term momentum: the stock had gained 5.1% over the prior thirty days.
- Reaction session, September 11: Shares opened at $164.43, a gap of 7.5%, traded as high as $166.00, then reversed through the entire move and closed at $150.28, down 1.7% on the day and 9.5% off the intraday high. Volume was 80.4 million shares against a thirty-day average of 26.2 million, or 3.1 times normal.
- The following week: The decline continued. The stock gapped down again on September 14 to close at $144.79 and reached a closing trough of $140.35 on September 15, 8.2% below the pre-print close, before recovering to $148.56 by September 21. Net of everything, ORCL sits 2.9% below where it traded the afternoon of the print.
The shape of the reaction is more informative than the magnitude. A 7.5% gap higher on the open means the market read the press release and the guidance raise as clearly positive. Giving all of it back within a single session, on triple normal volume, means the market read something else during the day and changed its mind. The candidates are on the cash flow statement: $28.5B of quarterly capital expenditure, negative free cash flow, and the new prepayment line that explains the record operating cash flow. None of those was in the headline. All of them were in the tables.
The second leg down the following Monday came after the quarterly filing had been digested. That filing carried the roughly $700M restructuring supplement and put the total debt figure of $125.3B on the page. Both are facts the call did not surface, and both cut against the "acceleration" framing the call led with. The market did not dispute the operating quarter. It re-priced what the operating quarter costs.
There is a useful asymmetry in the setup worth stating plainly. The stock entered the print already down 53% over twelve months, which means the funding anxiety was not news arriving into a complacent price. It was a well-known concern arriving into a price that had already halved. That the stock still could not hold a 7.5% gap on a genuine beat with a raised guide says the market's objection is structural rather than informational, and it will not be resolved by another good quarter. It will be resolved by a capital plan with an endpoint.
Street Perspective
Debate: Is the record operating cash flow a real improvement or an accounting artifact?
Bull view: The prepayments are cash in the bank, collected before delivery, which is strictly better than collecting after. They lower Oracle's funding requirement, improve project-level returns, and demonstrate that customers will fund Oracle's build rather than the reverse. A $23B operating cash flow quarter is a fact regardless of its composition.
Bear view: Strip the $11.4B prepayment and operating cash flow is roughly $11.7B against $28.5B of capital expenditure, so the true burn is roughly three times the reported figure. The prepayment is a one-time collection against multi-year revenue, it carries an embedded financing cost that grows, and presenting the gross figure as a record without the adjusted figure is the sort of framing that invites a discount.
Our take: The bear has the better of the accounting and the bull has the better of the economics. Both numbers are true and only one of them was emphasized. We will carry both in the model and we think Oracle should disclose both, because the adjusted figure is not embarrassing once it is understood as the cost of a pre-sold build, and withholding it makes it look worse than it is.
Debate: Does the 20% renewal premium settle the AI depreciation question?
Bull view: Four-year-old accelerators re-contracting 20% higher into a 97.9%-utilized fleet is the first hard evidence anywhere in the sector that AI infrastructure holds its economic value, which means the five-to-six-year depreciation schedules across the industry are conservative rather than aggressive and reported earnings are understated, not overstated.
Bear view: It is one quarter, the renewal cohort is unsized, and a 20% price increase during an acute capacity shortage says more about the shortage than about asset longevity. The test is what older capacity re-prices at once supply catches up, which is the scenario that matters and the one that has not happened.
Our take: The bull is right that this is the most valuable new evidence in the debate and the bear is right that one data point is not a series. It matters more than either side is treating it, because the depreciation question governs whether Oracle's GAAP earnings mean anything at all over the next five years. We want the cohort size and the renewal history at the investor day, and until then we treat the thesis as strengthened rather than settled.
Debate: Is the customer concentration a tail risk or the whole risk?
Bull view: The non-contracted backlog has broadened materially, the contract structures increasingly shift hardware funding to the customer, and the underlying demand for AI compute is far larger than any single counterparty. Concentration in a contracted backlog is a different animal from concentration in recurring revenue.
Bear view: If roughly half of a $664B backlog sits with one privately held counterparty whose own funding is not assured, then Oracle is not a diversified infrastructure business, it is a leveraged single-counterparty credit with a software company attached. Oracle has had four quarters to rebut the estimate and has not.
Our take: The bear's framing is stronger than the bull's, not because the estimate is necessarily right but because the silence is unusual for a company that discloses fleet utilization to a decimal place. Our working assumption remains that this is a tail risk rather than a base case, and we hold that assumption with less confidence than we would like. It is the single item most capable of changing our rating, and the only one that would change it quickly.
Debate: Is the stock cheap at 18x the company's own earnings guide?
Bull view: Revenue guided to grow 34% with half the supporting backlog contracted for delivery inside three years, at 18.3x the company's own full-year earnings guide, is an unusual combination. The de-rating has priced a funding crisis that has not occurred, and the largest single overhang, the equity issuance, is now complete.
Bear view: An 18x multiple on earnings that depend on a depreciation schedule nobody can verify, a gross margin with no disclosed floor, and an operating margin propped up by a finite restructuring programme, is not obviously cheap. Companies that spend a dollar of capital per dollar of revenue have historically earned utility multiples, and utility multiples are lower than this.
Our take: The bull wins on arithmetic and the bear wins on the quality of the arithmetic's inputs. What tips it is that the two largest bear objections from three months ago, the dilution overhang and the absence of evidence on asset longevity, both improved this quarter, while the price fell further. We do not require the bull case to be right about the terminal multiple to justify the position. We require the earnings to be roughly real and the build to end, and this quarter moved both of those forward.
Model Update & Valuation Framework
| Item | Model stood at | Q1 FY27 actual | Suggested change | Reason |
|---|---|---|---|---|
| Cloud YoY growth, Q1 FY27 | +58.5% | +61.5% (reported as +62%) | Raise the full-year path to ~+68% | Q2 guide of +65% to +71% and an accelerating delivery schedule |
| FY27 total revenue | ~$90.0B (+33.7%) | Q1 $19.345B | $90B to $92B | Guide moved to "at least $90B"; treat as a floor |
| Software YoY growth | −1% per year | −3% | −2.5% for FY27, −1.5% thereafter | License decline steeper than modelled at −15% |
| Hardware YoY growth | +4% | +15% | +8% for FY27 | Demand running well ahead of trend |
| Services YoY growth | +10% | +5% | +6% for FY27 | AI-assisted go-lives compressing implementation revenue |
| Cloud and software cost as % of that revenue | 39% rising to 42.5% | 37.3% | Hold 39% to 41% for FY27 | Q1 ran below the modelled path, but no floor is disclosed |
| R&D as % of revenue | ~11.2% | 12.4% (Q1 only) | Lower to ~10.7% | Q1 dollars annualise to ~$9.6B against the $90B floor; the H2 revenue ramp pulls the full-year ratio below the modelled path |
| S&M as % of revenue | ~9.0% | 9.4% | Hold | Tracking the modelled path closely |
| Restructuring as % of revenue | ~0.4% | 0.49% | Raise to ~0.8% for FY27 | Plan supplemented by ~$700M to ~$2.8B after quarter-end |
| Effective tax rate | 19.0% | 15.1% GAAP / 16.9% non-GAAP | Lower to ~17.5% | Two consecutive quarters below the modelled rate |
| Diluted share count | Dilution phased across FY27 | 3,000M, ATM complete in Q1 | Front-load the full 141M shares | Program fully executed in the first quarter |
| Preferred dividends | Not separately modelled | $81M per quarter | Add ~$324M annual charge | New claim ahead of the common |
| RPO sequential growth | +10% falling to +5% | +4.1% | Lower to +4% to +6% | Conversion now consuming backlog faster |
| FY27 capex (reported) | $90B to $95B | $28.5B in Q1 | Maintain | Guide reaffirmed; management flagged non-linearity |
| FY27 free cash flow | Deeply negative | −$5.4B reported in Q1 | Model both reported and ex-prepayment | Prepayments distort the reported figure materially |
| FY27 non-GAAP EPS | $8.05 | $1.92 in Q1 | $8.10 | Company guide raised |
Valuation. We trim our fair-value range to $215 to $265 from $300 to $360. At the September 21 close of $148.56 the new range implies 45% to 78% upside, which is almost exactly the 49% to 79% our prior range implied against the $201 share price at which we set it. The fair value falls roughly in line with the stock rather than independently of it, which is the correct outcome for a quarter in which the operating trajectory improved while the per-share claims on it grew: 5% more shares outstanding, a new $324M annual preferred dividend, and an interest run-rate above $5.7B.
The range corresponds to 26.5x to 32.7x the company's own FY27 non-GAAP EPS guide of $8.10, against a current 18.3x. On the back of an envelope, applying 21x to 26x a fiscal 2028 EPS figure in the $10.20 area produces a similar answer, which is the cross-check we wanted. Net debt of roughly $88B sits behind the equity, and the operating lease liabilities and multi-year purchase commitments detailed above should be treated as additional leverage that the notes-payable line does not capture.
The honest caveat on the prior range is that $300 to $360 implied a multiple the market has declined to pay for two years, and the company's own decision to issue equity at roughly $141 a share is a form of evidence about what the equity is worth today. We would rather mark the framework to a multiple the market will plausibly pay and be right about the direction than defend a number that has been wrong for eight quarters.
Thesis Scorecard Post-Earnings
| Thesis point | Status | Status tag move | What this quarter showed |
|---|---|---|---|
| Bull 1: RPO cascade is the spine of the case | Confirmed | ON TRACK, held | $664B with the 36-month conversion bucket moving from ~46% to 50%; the backlog is now a schedule, not a headline |
| Bull 2: OCI as a hyperscale-class franchise | Confirmed | ON TRACK, held | +121% with 850MW and 300,000+ accelerators delivered; 97.9% utilization; 24-hour customer acceptance at Abilene |
| Bull 3: Multicloud database as a capital-light engine | Confirmed | ON TRACK, held | +353% revenue, customers +180%, footprint complete at 70 regions with interconnect to every hyperscaler |
| Bull 4: Applications reacceleration | Neutral | AT RISK, held | 10% again, the fifth straight quarter inside a 10% to 13% band; Fusion +14% and industry apps above 20% are better, but the recognized line has not moved |
| Bull 5: Inferencing and capital-light contracting | Confirmed | ON TRACK, strengthened | Over $30B of new contracts signed requiring no incremental Oracle cash; $11.4B of prepayments collected in one quarter |
| Bull 6: Asset longevity and renewal pricing (new this quarter) | Confirmed | ON TRACK, new | Renewals cleared a 20% price premium on accelerators mostly four years or older, into a 97.9%-utilized fleet; first hard evidence against the sector depreciation argument |
| Bear 1: Capex and free-cash-flow trough | Confirmed | MATERIALIZING, held | $28.5B of quarterly capex; reported FCF −$5.4B, roughly −$16.8B ex-prepayment; no date offered for the turn |
| Bear 2: Customer concentration in RPO | Confirmed | EMERGING, held; exposure larger | A fourth quarter of silence on a backlog outside estimates put at roughly half one counterparty; no diversification statistic this time, and the backlog itself grew $26B |
| Bear 3: Funding and dilution | Challenged | MATERIALIZING to CONTAINED | $20B program completed in full, debt down $4.2B, equity up $24.1B; the overhang is behind rather than ahead, though executed at roughly $141 |
| Bear 4: Gross-margin compression | Confirmed | EMERGING to MATERIALIZING | Implied gross margin −725bp; segment margin −374bp; trough still unquantified and described as flattening over "a couple of years" |
| New Bear 5: The operating margin depends on a finite cost program | New | EMERGING | R&D dollars −4% and S&M −12% on +30% revenue; the restructuring plan was supplemented by ~$700M to ~$2.8B after quarter-end |
Overall: Strengthened. Seven of the eleven points moved favourably or held, and the two that improved most, asset longevity and the funding overhang, were the two that most constrained our conviction three months ago. Set against that, the cash quality is worse than the headline and a new bear point has been added around the durability of the cost program carrying the margin. The net is a thesis on firmer ground with a sharper list of what would break it.
Action: Hold the position and add on weakness below the mid-$140s. This is the second quarter of the coverage arc in which the operating evidence improved and the share price fell, and the first in which the funding profile improved alongside it, which is the combination that creates the entry point rather than the exit. The position sizing should respect the concentration disclosure gap, which remains the one item capable of invalidating the case quickly rather than gradually.
Bottom Line
Oracle delivered the quarter the thesis required. Capacity landed at three times the prior quarter's rate, cloud infrastructure grew 121%, the full-year guide went up instead of being reaffirmed, and for the first time the company put a price on what older AI capacity re-contracts at. The equity program that hung over the stock all summer was completed inside three months, and the balance sheet finished the quarter with less debt than it started.
It also delivered a cash statement that requires reading twice. A record $23.1B of operating cash flow is roughly half a one-time customer prepayment, the underlying burn is nearer $16.8B than $5.4B, and the operating margin was held by cutting research and sales spending in dollar terms while revenue grew 30%. The restructuring programme behind those cuts got $700M larger the day after the call. None of that makes the quarter bad. All of it makes the quarter more expensive than the headline suggests.
The market's verdict was a 7.5% gap higher that closed 1.7% lower, on triple volume, followed by another leg down once the filing was read. We think that verdict is about the right shape and the wrong magnitude. What Oracle has not answered in four quarters of asking is who holds the backlog, and until it does, the stock will keep trading as a proxy for someone else's balance sheet. What Oracle did answer this quarter is whether the assets hold their value and whether the dilution was finite, and on both it answered better than we expected.
We maintain Outperform with conviction raised to 7 from 6, and trim fair value to $215 to $265.