Oracle generated a record $23.1 billion of operating cash flow in the three months to 31 August 2026, close to triple the figure a year earlier. In the same three months it sold $20 billion of new Oracle shares into the market to pay for data centres. Both numbers sit in the fiscal first-quarter earnings release Oracle filed with the SEC on 10 September, a few paragraphs apart. The share price has spent a year trying to reconcile them. ORCL closed at $328.33 on 10 September 2025, the session after Oracle first put a $455 billion backlog in front of investors. It closed at $152.94 on 10 September 2026, just before this quarter's numbers came out, then traded at $159.26 after hours. Over that year the backlog grew 46% to $664 billion while the stock lost 53%. This forecast sets out why both can be true, and prices the outcomes to 31 December 2026.
Strip out one line and the cash record halves. Of the $23.1 billion, $11.4 billion is booked as an "increase in deferred revenues from customer prepayments with significant financing component": customers paying Oracle up front for GPU capacity, on terms that Oracle's FY2026 Form 10-K says carry an interest element it records as interest expense. Before those prepayments, Oracle's operations produced $11.7 billion against $28.5 billion of capital expenditure. So the quarter's cash came from three places. Oracle's own business supplied $11.7 billion, its customers $11.4 billion, and new shareholders $19.9 billion. The headline figure is real. It just is not mostly Oracle's.
Key facts
- Q1 FY2027 revenue $19.3 billion, up 30%; cloud infrastructure (OCI) revenue $7.4 billion, up 121%; software licence revenue down 15% — Oracle Q1 FY27 earnings release, 10 September 2026
- Operating cash flow $23.1 billion including $11.4 billion of customer prepayments; capital expenditure $28.5 billion; free cash flow negative $5.4 billion — Q1 FY27 release, cash flow tables
- $20 billion of common stock sold through the at-the-market programme during the quarter, net proceeds $19.9 billion; none had been sold by 31 May 2026 — Q1 FY27 release and FY2026 Form 10-K, filed 22 June 2026
- Remaining performance obligations $664 billion, up $26 billion on the quarter after an $85 billion rise in the quarter before — Q1 FY27 and Q4 FY26 releases
- FY2027 capital expenditure guided at $90 billion to $95 billion, with no more than $70 billion of it net cash — CFO Hilary Maxson, earnings call, 10 September 2026
- $260 billion of data-centre lease commitments not yet on the balance sheet, running 15 to 19 years and starting between Q1 FY27 and FY2029 — FY2026 Form 10-K
- FY2027 guidance of at least $90 billion revenue and $8.10 non-GAAP EPS puts ORCL at 18.9 times earnings on the $152.94 close — Q1 FY27 release; stockanalysis.com, 10 September 2026
Who paid for the quarter
The cleanest way to read an Oracle quarter now is as a sources-and-uses statement, because the income statement no longer tells you who is funding the build. Every figure below comes from the condensed cash flow statement in the release.
| Q1 FY2027, three months to 31 August 2026 | $ millions |
|---|---|
| Operating cash before customer prepayments | 11,740 |
| Customer prepayments with a financing component | 11,363 |
| At-the-market equity proceeds, net | 19,909 |
| Employee stock programmes, net | 41 |
| Cash in | 43,053 |
| Capital expenditures | (28,499) |
| Senior notes, term loans and other borrowings repaid | (4,202) |
| Short-term capex financing repaid, net | (830) |
| Dividends paid | (1,565) |
| Other investing, financing and FX, net | (312) |
| Increase in cash and restricted cash | 7,645 |
Source: Oracle Q1 FY27 earnings release, condensed consolidated statements of cash flows, 10 September 2026. The split of operating cash is The Traders Spread's arithmetic from the line items shown.
Cash on the balance sheet rose to $36.4 billion from $31.3 billion. Take the share sale out and it would have fallen by about $12.3 billion in a single quarter. Total notes payable and other borrowings stood at $125.3 billion on 31 August, down from $129.5 billion in May because Oracle repaid $4.2 billion and issued nothing new, in line with the June statement that it did not expect to issue additional debt in calendar 2026.
Management is precise about this, and the precision is the tell. Asked on the call whether new contracts avoid extra spending, chief executive Clay Magouyrk corrected the premise: "I didn't say, and I don't think myself nor Hilary said, that it doesn't require additional CapEx. We said it doesn't require additional cash from Oracle, right?" (Clay Magouyrk, Chief Executive Officer at Oracle, per AlphaStreet's preliminary transcript of the 10 September call). He listed the mechanisms: supplier financing that lets Oracle pay as customers pay, customers buying their own GPUs for Oracle to run, and customers prepaying.
Each of those shifts the bill without removing it. A prepayment with a financing component is, in accounting substance, a loan from the customer that Oracle repays in capacity. The 10-K described the related interest for fiscal 2026 as immaterial, when prepayments totalled $4.6 billion. Cumulative receipts are now about $16 billion. Interest expense in the quarter was $1.43 billion, up 55% from $923 million.
A backlog that pays over five years, and leases that run for nineteen
Remaining performance obligations are contracted revenue not yet recognised. They are a backlog. They are not cash, and the pace at which they turn into either is set by how fast Oracle can energise capacity.
The 10-K gives the schedule as of 31 May 2026: of $638 billion, about 12% was expected to be recognised as revenue within twelve months, 34% in months 13 to 36, 34% in months 37 to 60, and the remaining fifth after that. Twelve per cent is roughly $77 billion, most of the $90 billion revenue guide for the year before a single new sale is counted. On the 10 September call, CFO Hilary Maxson said Oracle now expects "around half of our RPO to convert into sales over the next 36 months." For comparison, Microsoft's FY2026 10-K expects about 30% of its $678 billion commercial RPO to convert within twelve months, a figure we examined in our Microsoft stock forecast. Oracle's backlog is longer-dated by design.
The new business signed in Q1 is further out still. Maxson told analysts that the new contracts "won't impact our CapEx or revenues until fiscal '28 or beyond." That matters for anyone treating RPO growth as a near-term revenue signal: sequential growth fell from $85 billion in Q4 to $26 billion in Q1, and the more than $30 billion of contracts signed during the quarter do not touch this fiscal year at all.
Now set the other long-dated number beside it. At 31 May, Oracle disclosed $260 billion of additional lease commitments, "substantially all related to data center arrangements", commencing between the first quarter of fiscal 2027 and fiscal 2029 for terms of fifteen to nineteen years. They were not on the balance sheet. Operating lease liabilities have already started to climb, reaching $30.6 billion in the non-current line at 31 August against $26.6 billion in May, consistent with the first of those sites commencing.
The 10-K's own risk factors name the mismatch: "the terms, renewal options and pricing adjustments in our long-term data center leases typically do not align with the duration and pricing of customer contracts." Elsewhere it notes that "in certain OCI offerings, we are more concentrated among a number of large customers." A backlog weighted towards a handful of AI buyers over five years, funded partly by leases that run nearly two decades, is the real risk profile. The size of RPO is the least interesting thing about it.
There is evidence on the other side, and it is specific. Magouyrk reported GPU utilisation of 97.9% in Q1 and said that capacity coming up for renewal was renewed or resold at a 20% premium to prior contracts, with the majority of those GPUs four or more years old. If that holds as the fleet scales, the useful-life worry that hangs over every AI cloud balance sheet weakens considerably. It is one quarter's disclosure, from management, on a renewal pool that is small relative to what is being built.
What the share price has already absorbed
ORCL's reaction to each backlog print tells you the market stopped paying for RPO some time ago. The first $455 billion print sent the stock up 35.95% in a session. Since then the pattern has alternated, and the reaction has had little to do with how much the backlog grew.
| Quarter end | RPO | Change on prior quarter | ORCL next-session move |
|---|---|---|---|
| 31 Aug 2025 (Q1 FY26) | $455bn | up 359% year on year | +35.95% to $328.33 |
| 30 Nov 2025 (Q2 FY26) | $523bn | +$68bn | −10.83% to $198.85 |
| 28 Feb 2026 (Q3 FY26) | $553bn | +$29bn | +9.18% to $163.12 |
| 31 May 2026 (Q4 FY26) | $638bn | +$85bn | −8.53% to $184.10 |
| 31 Aug 2026 (Q1 FY27) | $664bn | +$26bn | +4.13% after hours to $159.26 |
Sources: RPO from Oracle's quarterly earnings releases filed with the SEC (9 Sep 2025, 10 Dec 2025, 10 Mar 2026, 10 Jun 2026, 10 Sep 2026). Price moves from stockanalysis.com daily history and quote, retrieved 11 September 2026. The Q1 FY27 figure is an extended-hours trade, not a regular-session close.
The biggest backlog increase of the year, $85 billion in Q4, produced the second-worst reaction. The smallest, $26 billion this week, produced a gain after hours. That is what a market does when it has moved from pricing the order book to pricing the funding of it.
The funding is visible in the tape. Oracle's 10-K recorded no ATM sales by 31 May; the Q1 release says the whole $20 billion was completed during the quarter. Across the 64 sessions from 1 June to 31 August, ORCL's consolidated dollar volume was about $307 billion on stockanalysis.com daily data, so the programme amounted to roughly 6.5% of everything traded. At the quarter's volume-weighted price of about $154, that implies around 130 million new shares, near 4.5% of the 2.88 billion outstanding on 12 June. Those are our estimates; Oracle has not yet published the average sale price, which the 10-Q should carry. A prospectus supplement dated 23 June added fifteen further sales agents to the programme. From the 22 June close of $175.07 the stock fell 34% to its 24 July low of $114.99. That is timing, not proof of cause.
What changes now is that the programme is finished. The June release said Oracle expected to raise about $40 billion in FY2027 through debt and equity, including this $20 billion. Roughly $20 billion remains, and with no new debt planned for calendar 2026, it most likely arrives between January and May 2027. The overhang has not gone. It has moved into next year.
Where the margin went, and why EPS still rose 30%
Gross margin, computed from the release as revenue less the cloud and software, hardware and services cost lines, fell to 60.0% from 67.3% a year earlier. Depreciation more than doubled to $3.16 billion from $1.35 billion. Cloud and software operating expenses rose 77% against 33% growth in the revenue they serve.
Non-GAAP operating margin held at 42% because everything below gross profit was cut. Sales and marketing fell 12% to $1.81 billion. Research and development fell 4% to $2.40 billion. Amortisation of intangibles halved. On the call, Maxson framed operating margin as "the ultimate point that we want to follow" and said gross margin should flatten "over the next couple of years as we finish the ramp-up."
That is a legitimate strategy for a business shifting from software to infrastructure. It is also a finite one. Sales and marketing can fall 12% once or twice; it cannot fall 12% every year while the company also promises double-digit SaaS growth. The non-GAAP EPS guide went up by five cents to $8.10. For a quarter with 121% infrastructure growth, that is a cautious raise, and it tells you how much of the growth the cost base is absorbing.
Dilution runs underneath all of this. Diluted weighted shares were 3,000 million against 2,909 million a year ago, and Q1 only carried part of the ATM issuance. Q2 carries all of it. Oracle's new CFO promised when she joined in April "to continue to invest with discipline and to translate this momentum into durable, long-term value for customers and shareholders" (Hilary Maxson, Chief Financial Officer at Oracle, in the 6 April 2026 appointment release). Her first full quarter spent $28.5 billion and sold $20 billion of stock. Both can be disciplined. Shareholders will judge by per-share cash, not by backlog.
For the suppliers on the other end of this spending, our report on Nvidia's Q3 guidance is the counterpart read. Magouyrk said Oracle will deliver its first Vera Rubin systems to customers in Q2. For how two larger cloud providers are being priced on their own AI spending, see our Alphabet forecast and Amazon forecast.
The call: $170 base, $215 bull, $115 bear by 31 December
ORCL last traded at $159.26 in extended hours after a $152.94 close. The horizon covers the October Investor Day and the Q2 FY27 report, which investor relations chief Ken Bond gave as 14 December.
RelatedAlphabet (GOOGL) Stock Forecast: $395 Bull Case vs $285 Bear Case
Base case, $170, probability 50%. Q2 lands inside guidance of 30% to 34% revenue growth and 65% to 71% cloud growth, gross margin settles near 60%, and no new equity programme is announced before December. The stock recovers toward its 200-day average of $167.89. At $170 ORCL trades at 21 times FY2027 guided EPS.
Bull case, $215, probability 20%. Management gives a date for positive free cash flow at the Investor Day, OCI growth accelerates past Q1's 121% as Maxson indicated, and the remaining FY2027 funding is arranged as debt without widening spreads. That is 26.5 times guided EPS, and close to the $213.68 the stock held on 5 June.
Bear case, $115, probability 30%. A second equity programme or an equity-linked issue arrives before December, a gigawatt campus slips far enough to touch FY2027 guidance, or gross margin drops another five points. $115 retests the 24 July low at 14.2 times guided EPS.
The probability-weighted value is $162.50, a bias of neutral with conviction 2 out of 5. A daily close below $138, roughly half the July-to-September rebound, would invalidate the base case.
What would change my mind: a free cash flow timetable from management moves me to the bull case. Any new share issuance announced before the Q2 report moves me to the bear case, because it would confirm that the $20 billion was a first instalment rather than a one-off.
FAQ
Has Oracle reported its fiscal Q1 2027 results?
Yes. Oracle filed an 8-K with Item 2.02 on 10 September 2026 attaching its results for the quarter ended 31 August 2026, and held its call the same afternoon. Revenue was $19.3 billion, up 30%, and non-GAAP EPS was $1.92, up 30%. The Form 10-Q, which will carry the full RPO schedule and ATM detail, had not been filed at the time of writing.
Why is Oracle's free cash flow negative when operating cash flow hit a record?
Capital expenditure of $28.5 billion exceeded operating cash flow of $23.1 billion, leaving free cash flow of negative $5.4 billion. The operating figure itself includes $11.4 billion of customer prepayments for future capacity. Excluding those, operations generated $11.7 billion, and the gap to capex was roughly $16.8 billion.
What does Oracle's $664 billion RPO mean for revenue?
It is contracted revenue that has not yet been recognised. As of 31 May 2026 Oracle expected about 12% of its RPO to become revenue within twelve months and 34% in the following two years. The CFO now says around half should convert within 36 months. New Q1 contracts will not add revenue until fiscal 2028 or later.
How much did the $20 billion share sale dilute ORCL holders?
Oracle has not disclosed the average price. At the quarter's volume-weighted price of roughly $154 on stockanalysis.com data, $20 billion buys about 130 million shares, around 4.5% of the 2.88 billion outstanding on 12 June 2026. Separately, $5 billion of mandatory convertible preferred stock converts in January 2029 into between 25 million and 31 million shares.
When does Oracle report next, and what matters most?
Investor relations gave 14 December 2026 for the Q2 FY2027 results, with an Investor Day in October before that. The decisive items are any timetable for positive free cash flow, whether gross margin stabilises near 60%, and whether the roughly $20 billion of remaining FY2027 funding comes as debt or as more shares.
Disclaimer
This article is analysis and information, not investment advice. The levels, probabilities and valuation multiples above are the author's own work from Oracle's SEC filings, the company's earnings call and live market data, and they can be wrong. Capital is at risk, and past performance does not indicate future results. Anyone acting on this material does so on their own judgement and should consider advice from a regulated professional.