Oracle's quarterly free cash flow would be minus $16.8bn without customer prepayments
Customer prepayments supplied 75.9 percent of the rise in Oracle's August-quarter operating cash flow. Capex rose about $20bn. The advances ease funding pressure, but do not prove the build will pay off.
Oracle reported free cash flow of minus $5.4bn in the quarter ended 31 August. Subtract the $11.363bn increase in deferred revenue attributable to customer prepayments with a significant financing component, and the result is minus $16.759bn. Oracle defines free cash flow as operating cash flow less capital expenditure in its 10-Q.
The $16.8bn is a CTOL sensitivity, not missing cash or a second measure of actual free cash flow. Customers paid the money. Their advances ease near term funding, but they do not prove investment returns or promise further payments.
With Fitch assuming $90bn to $95bn of fiscal 2027 capex, what matters is how different prepayment levels change Oracle's cash needs. CTOL's cash flow calculation and next-quarter call test how much the prepayments matter. They do not forecast future receipts.
How prepayments change Q1 cash flow
Operating cash flow was $23.1bn and capex $28.5bn. The 10-Q lists "Increase in deferred revenues from customer prepayments with significant financing component" at $11.363bn, against none a year earlier. Oracle says the effects are "recognized over the period of performance".
Oracle's reported free cash flow was minus 0.362 billion dollars in the first quarter of fiscal 2026 and minus 5.396 billion in fiscal 2027. Excluding 11.363 billion dollars of customer prepayments booked in operating cash flow, the fiscal 2027 figure is minus 16.759 billion.
View data
| Fiscal quarter | Reported free cash flow | Excluding customer prepayments (CTOL) |
|---|---|---|
| Q1 FY26 (June to August 2025) | -0.362 | -0.362 |
| Q1 FY27 (June to August 2026) | -5.396 | -16.759 |
Prepayments accounted for 75.9 percent of the approximately $15.0bn year on year increase in operating cash flow. With only half that inflow and everything else unchanged, Q1 free cash flow would have been about minus $11.1bn. Neither figure suggests those receipts were optional.
Capex is the larger structural change. Operating cash flow excluding the prepayment line rose roughly $3.6bn, from $8.1bn to $11.7bn. Capex rose about $20bn, from $8.5bn to $28.5bn. CTOL's figure excluding prepayments deteriorated because roughly $20bn of extra capex exceeded the $3.6bn gain in other operating cash flow.
Reported free cash flow worsened far less because the customer advances offset some of that construction spending.
The advance helped fund construction but brings a service obligation. Recognising revenue later for that prepaid service will not bring the same cash a second time. That establishes a cash timing question, not customer default or poor lifetime returns.
Oracle's 8-K exhibit of 10 June says the "prepaid and customer supplied hardware portions of our large AI contracts now total $75 billion". That cumulative figure mixes cash with hardware bought by customers; it cannot be added to Q1 prepayments or treated as a future cash inflow.
Oracle priced the debt five days after Investor Day
The same 10 June exhibit said "Oracle does not expect to issue additional debt in calendar year 2026". On 28 October, Oracle's Investor Day exhibit committed to limit additional capital raises through fiscal 2030 to $20bn. It expected $10bn of equity and $10bn of gross debt, on top of a $20bn at the market equity sale completed in the first quarter.
On 2 November Oracle priced the two $10bn components. Pricing does not mean the company has received the proceeds: closing is expected on 4 November, and the final prospectus supplements were not yet on EDGAR on 3 November. The final pricing term sheet sets out five fixed rate tranches.
| Tranche | Principal | Coupon | Price | Spread to Treasury | Yield |
|---|---|---|---|---|---|
| 2029 | $2.00bn | 7.000% | 99.835 | +240bp | 7.062% |
| 2031 | $1.75bn | 7.450% | 99.942 | +275bp | 7.464% |
| 2033 | $2.25bn | 7.850% | 99.995 | +305bp | 7.851% |
| 2037 | $2.50bn | 8.000% | 99.916 | +310bp | 8.015% |
| 2046 | $1.50bn | 8.650% | 99.840 | +335bp | 8.667% |
CTOL calculates a 7.7675 percent weighted coupon, or about $777m of annual stated payments. It is not necessarily $777m of net additional interest if existing debt is refinanced. The notes rank equally with other unsecured debt, and the indenture has no financial covenants, per the preliminary prospectus.
Media reports put peak demand at about $38bn, nearly four times the issue. That shows demand at pricing, not assured future access. On their own calculation, media reports say the notes paid roughly 0.3 percentage point over Oracle's existing debt. No matched maturity comparison is available here, so the issue yields above are the firmer price evidence.
The equity offer comprised 86,206,897 shares at $116.00, per Oracle's pricing release. That is 2.85 percent of the 3,023,726,695 shares outstanding on 31 August, per the preliminary prospectus, not the post issuance ownership dilution. The offer was 11.2 percent below the 30 October close of $130.67.
The $20bn completed equity programme plus $20bn now priced match Oracle's roughly $40bn fiscal 2027 financing figure from June. What changed was $10bn of new debt despite the June expectation of no further 2026 borrowing. Issuance proceeds, once received, are financing cash, not operating cash flow or free cash flow.
Maturities compete for the proceeds
The 10-K schedules $7.21bn of principal for fiscal 2027, $10.145bn in fiscal 2028 and $5.5bn in fiscal 2029. CTOL's sum through fiscal 2029 is about $22.9bn. The CFO deck says that after the new debt, gross debt will "remain flat as we refinance maturities".
Oracle's 10-K schedules 7.21 billion dollars of debt principal for fiscal 2027, 10.145 billion for fiscal 2028, 5.5 billion for fiscal 2029, 7.25 billion for fiscal 2030 and 9.75 billion for fiscal 2031.
View data
| Fiscal year (ends 31 May) | Principal due |
|---|---|
| FY27 | 7.21 |
| FY28 | 10.145 |
| FY29 | 5.5 |
| FY30 | 7.25 |
| FY31 | 9.75 |
The schedule dates to 31 May 2026 and excludes November's priced notes and later refinancing. Rolling over maturing principal and raising funds for new construction serve different purposes. A negative free cash flow quarter therefore does not, by itself, imply an inability to pay obligations. Nor does successful financing prove that the underlying contracts will generate sufficient cash over their lives.
How Oracle's funding compares with its peers
Oracle's trailing free cash flow was minus $28.7bn, compared with positive $60.0bn at Microsoft and $19.7bn at Alphabet, on CTOL's calculation from SEC filings data. Oracle alone had a negative figure. These amounts are a starting point only. They do not rank the three companies, because scale, capital intensity and liquidity differ.
Three features of Oracle's funding case matter more than the raw peer numbers. The article does not show that its peers lack any of them.
First, concentration. Oracle's filings name no customer, but Fitch cites media reports that OpenAI accounts for more than $300bn of Oracle's $664bn in contracted revenue. OpenAI separately says its Oracle partnership "exceeds $300 billion" over five years. The similar figures need not describe identical accounting populations.
Contracted revenue is not a receivable or loan balance, and the filings allocate neither Q1 prepayments nor capex by customer. They cannot size the cash at risk from one customer.
Second, duration. Oracle has $288bn of lease commitments not yet commenced, with terms of 15 to 19 years, while its AI compute contracts run about five years, per Fitch, which considers the risk manageable. Recognised lease liabilities are separately $43.8bn. Those amounts cover different leases on different bases, so they cannot be added.
The mismatch raises questions about renewals and later payments. The filings do not match leases to contracts or size any shortfall.
Third, customer financing. The WSJ reported on 30 October that OpenAI had about $11bn of commitments toward a $30bn fundraising target at a $1.25tn valuation. One person said it now hoped to raise up to $20bn; OpenAI's chief executive said an IPO in 2026 would not happen. Neither statement establishes a delayed Oracle payment.
Oracle's five year credit default swaps stood near 300 basis points in mid October, when CoreWeave's exceeded 1,100, and at 292 basis points on 2 November, according to media reports citing S&P Global Market Intelligence. These are risk prices, not customer losses or comparable default base rates.
What Lucent and Nortel's history can tell us
The history warns about customer financing exposure, not an Oracle loss rate. Lucent's fiscal 2000 operating cash flow from continuing operations was minus $703m, against $1.915bn of capex, as restated in its fiscal 2002 accounts. It recorded $2.249bn of bad debt and customer financing provisions in fiscal 2001, per its filing.
Nortel's 10-K shows $799m in reserves against $1.091bn of financed balances at the end of 2002, or 73 percent. That is reserve coverage at one date, neither realised losses nor a base rate.
Lucent and Nortel lent to customers. Oracle instead receives prepayments and takes on long infrastructure commitments. The useful analogy is limited: customers dependent on capital markets can transmit funding strain to suppliers. The historical numbers cannot forecast payment delays or losses at Oracle.
What the rating agencies say
| Agency | Rating, outlook | Stated risk or trigger |
|---|---|---|
| Moody's | Baa2, negative | Expected rating listed on the 2 November pricing term sheet |
| S&P | BBB-, stable | Adjusted leverage in the "mid-4x area" in fiscal 2027; hyperscaler spending and chip financing "a concentrated bet on two key AI labs" |
| Fitch | BBB, stable | Downgrade if EBITDA leverage stays above 3.5x, EBITDAR above 4x, or (CFO less capex) over debt below 15% |
S&P's note uses "a concentrated bet on two key AI labs" for hyperscaler spending and chip financing generally, not Oracle alone. Fitch's 20 percent ratio is an upgrade threshold. CTOL calculates trailing operating cash flow less capex over borrowings at about minus 23 percent.
That backward looking, unadjusted ratio illustrates the recent cash burden relative to borrowings; it is not comparable to Fitch's forward looking rating test or a measure of distance to downgrade.
Fitch says it does not expect Oracle "to raise material incremental capital over the rating horizon", but also that Oracle "is likely to issue incremental capital through a combination of debt and equity through at least fiscal 2028". Those statements should not be recast as a promise of no future issuance.
Oracle targets positive free cash flow in the second half of fiscal 2028. Fitch and S&P expect positive free cash flow in fiscal 2029. An improvement within a half year and an expectation for a later full year are not identical tests, and neither supplies quarterly prepayment forecasts.
What different prepayments would mean for the rest of fiscal 2027
This CTOL sensitivity is not a forecast. Fitch's $90bn to $95bn fiscal 2027 capex assumption, minus the $28.5bn spent in Q1, leaves $61.5bn to $66.5bn for the final three quarters. Spreading that evenly gives an average of $20.5bn to $22.2bn per quarter.
That average is a device for the sensitivity, not Fitch's forecast for any specific quarter. Oracle could spend more or less than the average as construction proceeds.
Hold operating cash flow excluding prepayments at Q1's $11.74bn per quarter. Then free cash flow equals that amount plus customer prepayments minus capex. Vary prepayments across zero, half of Q1 and a full Q1 repeat, with the annual capex range fixed. For a full fiscal year comparison, each three quarter total must also include Q1's reported deficit. None of the cases is a forecast.
| Prepayments each remaining quarter | Free cash flow each quarter | Final three quarters |
|---|---|---|
| None repeat ($0) | minus $8.8bn to minus $10.4bn | minus $26.3bn to minus $31.3bn |
| Half of Q1 ($5.7bn) | minus $3.1bn to minus $4.7bn | minus $9.2bn to minus $14.2bn |
| Q1 level repeats ($11.4bn) | plus $0.9bn to plus $2.6bn | plus $2.8bn to plus $7.8bn |
To see the fiscal year burden, add the reported Q1 deficit of $5.396bn to the final three quarter totals. Without further prepayments, another $26.3bn to $31.3bn of negative free cash flow would bring the annual drain close to Fitch's figure of more than $37bn of cash, cash equivalents and marketable securities.
With half of Q1's prepayments recurring, the final three quarters still lose $9.2bn to $14.2bn before Q1 is added. Even if prepayments fully recur, adding Q1's deficit leaves the fiscal year between minus $2.6bn and plus $2.4bn.
These amounts show how much the prepayments matter. They do not establish insolvency, because hypothetical annual flows and a liquidity snapshot cannot be treated as one balance. Oracle also reports $19.7bn of further borrowing capacity, has $7.21bn of scheduled fiscal 2027 principal and has priced another $20bn of securities for expected settlement on 4 November.
Each has a different timing or definition. Adding them into a single funding balance would invent resources or a funding gap; refinancing maturities is not the same as funding construction.
The conditional break-even point is more useful for Q2 monitoring. If operating cash flow excluding prepayments stays at $11.74bn and capex equals the $20.5bn to $22.2bn quarterly average, prepayments of $8.8bn to $10.4bn are required for zero or positive free cash flow. That is 77 percent to 92 percent of Q1 prepayments.
If either other operating cash flow or capex changes, so does the threshold. The history provides only one quarter with this prepayment line; none was recorded a year earlier.
Three scenarios, ordered by likelihood (CTOL's view)
These are conditional CTOL judgments, not probability estimates. One quarter of prepayment data and no usable base rate cannot support numerical odds. The near term test is Q2 cash flow; the longer tests are capital needs through fiscal 2028 and free cash flow in fiscal 2029. Each scenario states what would change its ranking.
Most likely: a funded but strained build. It needs some repeat prepayments, capex near Fitch's full year range, continued market access and investment grade ratings. At half of Q1's prepayment level, the arithmetic still shows a three quarter deficit of $9.2bn to $14.2bn, on top of Q1's reported deficit.
Oracle's priced securities support near term access to capital, not a forecast of customer advances. Fitch leaves room for later issuance; this scenario does too. The next tests are Q2 prepayments, operating cash flow excluding prepayments and capex in the 10-Q due by 11 January 2027.
Plausible: prepayments fade before capex peaks. Prepayments fall below the conditional $8.8bn to $10.4bn break-even level while spending stays high. Free cash flow turns negative in the sensitivity only if other operating cash flow and capex follow its assumptions.
Another capital raise or refinancing could follow, but available cash and borrowing capacity mean neither is automatic. Oracle says capex peaks in fiscal 2027 and 2028, and S&P expects free operating cash flow deficits in both years. Vendor financing and customer hardware can reduce upfront needs.
Watch actual cash flow, new offering terms and rating actions, not the prepayment number alone. BBB- remains one notch above high yield.
Unlikely within a year: customer funding strain reaches Oracle's cash. This needs an OpenAI financing shortfall to translate into delayed payments or changed commitments to Oracle. A weak fundraising close alone would not prove either. Fitch warns a shortfall "could result in payment delays and material revenue loss".
Only about 13 percent of Oracle's $664bn of contracted revenue falls within the next 12 months, but that is a revenue recognition schedule, not customer payment timing or evidence of protection. Monitor the OpenAI round, Oracle's prepayments and remaining performance obligations in the Q2 10-Q due by 11 January 2027, and any reported collection problems.
CTOL ranks the first case highest because financing access and investment grade ratings are observable; prepayment recurrence is not. The second follows from the limited history of advances. The third needs an additional link from customer funding to Oracle's receipts. Over fiscal 2028 and 2029, contract renewal and lease use remain questions rather than quantified outcomes.
The call: reported free cash flow stays negative in the quarter to 30 November
CTOL expects, with low confidence, that free cash flow for Oracle's quarter ending 30 November, taken on its own, will be negative. Under the arithmetic sensitivity, that happens if other operating cash flow stays at $11.74bn, capex is $20.5bn to $22.2bn and customer prepayments are below the corresponding $8.8bn to $10.4bn break-even level.
The bridge forecasts none of those inputs. Oracle's 10-Q does say fiscal 2027 capex will exceed fiscal 2026's $55.7bn, but that annual statement alone does not determine Q2 spending.
The call fails if operating cash flow for the quarter alone less capex is zero or positive. The 10-Q presents six months of cash flow, so Q2 is the cumulative figure less Q1. Advances, other operating cash and capex must then be separated to explain the result.
What to check on 4 November, 14 December and 11 January
The notes and shares are expected to settle on 4 November, with final prospectus supplements to follow. Oracle's investor relations page says Q2 results are scheduled for 14 December, and the 10-Q is due by 11 January 2027. Settlement tests execution of the announced financing; the operating cash flow test comes with the later filing.
From the six month cash flow statement, subtract Q1 to obtain operating cash flow for the quarter alone, capex and prepayments. Recalculate Q2 free cash flow, then calculate it without the advances. The actual break-even prepayment requirement is Q2 capex less Q2 operating cash flow outside prepayments, not automatically the $8.8bn to $10.4bn sensitivity range.
Separate changes in construction spending, operating cash flow excluding prepayments, and advances before assigning the result to a scenario. OpenAI fundraising, remaining performance obligations and agency actions offer secondary signals, not proof of missed customer cash. The decisive result is free cash flow for the quarter alone. A positive quarter would falsify CTOL's call, not establish attractive lifetime infrastructure returns.