Price Data (live — latest settled close)

A live table: every cell shows the latest settled close, named in the first row. Dated readings live in the Research Log.

Prior Research (prepared by 2026-09-23; current call and filed-period check above)

Summary

Oracle sells two different businesses under one ticker. The older one — database licences, software support, applications, hardware and consulting — generates about $7.7 billion a quarter and is flat to shrinking: software revenue fell 3% year over year in the August quarter and licence revenue fell 15%. The newer one rents AI compute. Cloud infrastructure revenue reached $7.388 billion in the quarter, up 121% year over year in reported currency and 120% in constant currency, and that single line now nearly matches everything the legacy software business produces.

Renting compute is a landlord business, and landlords buy buildings before they collect rent. Oracle spent $28.499 billion on capital expenditure in the August quarter against $23.103 billion of operating cash flow, producing negative $5.396 billion of free cash flow. That is the fifth consecutive negative quarter in the company's own disclosed table; fiscal 2026 consumed $23.686 billion in total. The company told investors it expects fiscal 2027 capital expenditure to exceed fiscal 2026's $55.663 billion.

The decisive uncertainty is not whether customers want the capacity. They have signed for it: remaining performance obligations stand at $664 billion against $455 billion a year earlier, and management booked more than $30 billion of new AI cloud contracts in the quarter alone. The uncertainty is who carries the working capital between signature and delivery, on what terms, and at what cost to existing shareholders. The August quarter gave a partial answer — and the answer involved 141 million new shares.

Why This Moment Is Different

Three disclosures changed the shape of the problem, all of them in filings rather than price action.

The equity program is spent. The quarterly report states that "during the first quarter ended August 31, 2026, we fully utilized the ATM Program and issued 141 million shares of common stock under the ATM Program for net proceeds of $19.9 billion." The program was authorised on February 2, 2026 and amended on June 23, 2026 for up to $20 billion of aggregate sales proceeds. As of May 31, 2026 the annual report recorded that no shares had been sold under it. The entire authorisation was consumed in one three-month period. Common shares outstanding went from 2.880 billion at May 31, 2026 to 3.024 billion at August 31, 2026.

Record cash flow was half prepayment. Operating cash flow of $23.103 billion was up 184% year over year, and the results release leads with that number. The cash-flow statement shows where it came from: an $11.363 billion line item titled "increase in deferred revenues from customer prepayments with significant financing component," plus $3.997 billion of increase in other deferred revenues. Management's own supplemental table subtracts that $11.363 billion when computing what it calls net cash outlay for capital expenditures. Customers, in other words, are funding a substantial share of the buildout, and Oracle is accounting for it as a financing arrangement.

The off-balance-sheet commitment grew again. The quarterly report discloses $288 billion of additional lease commitments, substantially all related to data-center arrangements, expected to commence between the second quarter of fiscal 2027 and fiscal 2029 on terms of fifteen to nineteen years, none of which appears on the August 31 balance sheet. The comparable figure in the fiscal 2026 annual report, as of May 31, 2026, was $260 billion. On-balance-sheet lease liabilities are a further $34.621 billion of operating leases and $9.185 billion of finance leases.

How the Business Makes Money, and What the Company Says It Will Do

All figures are GAAP, reported currency, for the three months ended August 31, 2026, from the September 10 results release.

The company's stated intent is unambiguous. The quarterly report says Oracle expects cloud and software expenses to keep growing "during fiscal 2027 and in the next few fiscal years as we increase our existing data center capacity and establish data centers in new geographic locations in order to meet current and expected customer demand," and that fiscal 2027 capital expenditure will be higher than fiscal 2026's for the same reason. Management frames the constraint as supply, not demand: the results release states that "customer demand for AI Cloud Training and Inferencing Services continues to grow faster than supply."

On concentration, the annual report states that "no single customer accounted for 10% or more of our total revenues in fiscal 2026, 2025 or 2024." That is a statement about delivered revenue, not about the backlog, and the annual report is explicit about the difference in its own words: "we enter into certain large, long-term customer cloud arrangements that require us to make significant infrastructure investments, including data center capacity… The economic returns on these investments are dependent on customer demand and the ability of our key customers to meet their contractual obligations."

The filed risks that bear on this profile, from the fiscal 2026 annual report, are indebtedness, refinancing and rating. The annual report warns that "disruptions in credit markets, including reduced liquidity, changes in creditor sentiment or lender concentration limits, could increase our borrowing costs or limit our ability to access debt financing on acceptable terms, or at all," and that a ratings downgrade "could also reduce our access to, or increase the cost of, commercial paper or other short-term financing, affect the terms or availability of certain long-term commitments (including data center leases), limit eligibility to contract with certain customers, and increase collateral, letter of credit or other credit support requirements." Oracle is also a named defendant, with certain officers and a director, in a putative securities class action tied to a prior decline in its stock price.

The Financing Question

Backlog against delivered revenue. Remaining performance obligations were $664 billion at August 31, 2026, against $455 billion a year earlier and $638 billion at the May 31, 2026 fiscal year end. The quarterly report gives the recognition schedule: approximately 13% as revenue over the next twelve months, 37% over months 13 to 36, 34% over months 37 to 60, and the remainder thereafter. Thirteen percent of $664 billion is roughly $86 billion, against a full-year fiscal 2027 revenue guide of at least $90 billion. Set the backlog against the engine that has to deliver it and the duration becomes visible: cloud infrastructure revenue of $7.388 billion in the quarter annualises to about $29.6 billion, so $664 billion of contracted obligation is roughly twenty-two years of infrastructure revenue at the current delivery rate. The gap between those two framings — a twelve-month schedule the company publishes, and a twenty-year ratio the run-rate implies — is the whole argument, and it resolves only if capacity delivery keeps compounding.

Operating cash flow against capital expenditure.

Net cash outlay for capital expenditure is Oracle's own non-GAAP measure: capital expenditure less short-term financing cash flows related to capital expenditure, less customer prepayments with a significant financing component applied to capital expenditure. It reconciles to the GAAP lines in the same table, and it is the number that shows the funding gap directly. On either measure the conclusion holds: free cash flow has been negative for five consecutive quarters, and the shortfall is filled from outside the business.

How the gap is filled. Fiscal 2026 was funded by debt: $43.0 billion of senior notes issued, plus $5.0 billion of net proceeds from the 6.50% Series D Mandatory Convertible Preferred Stock issued February 5, 2026, plus $3.3 billion of net short-term financing related to capital expenditure. The commercial paper program was raised to $10.0 billion on March 6, 2026, with $1.5 billion outstanding at the fiscal year end. Aggregate outstanding indebtedness was $129.5 billion at May 31, 2026, maturing between calendar 2026 and calendar 2066.

The August 2026 quarter was funded differently. Debt went down, not up: Oracle repaid $4.202 billion of senior notes, term loans, commercial paper and other borrowings, and carrying value of senior notes and other long-term borrowings fell from $128.1 billion to $125.0 billion. The replacement was equity and customer money — $19.909 billion net from the at-the-market program and $11.363 billion of customer prepayments. That is a deliberate change of funding mix, and the results release says so obliquely: "based on the structuring of those new contracts, the Company confirms there is no incremental impact on its plans to raise capital."

Three facts sit against that reassurance. The at-the-market authorisation is now fully drawn, so the same instrument cannot be used again without a new authorisation. Interest expense reached $1.428 billion in the quarter against $923 million a year earlier, an annualised run-rate near $5.7 billion before any new issuance. And the quarterly report discloses that the estimated fair value of Oracle's $125.0 billion of senior notes and other long-term borrowings, using Level 2 inputs, was $105.7 billion at August 31 — against $114.4 billion on $128.1 billion of carrying value at May 31. The market marks Oracle's own long-dated paper about 15% below carrying value, and that discount widened over the quarter. The filing does not decompose how much of that mark is interest rates against maturities running to 2066 and how much is credit; it reports the number only.

Contracted against energised capacity. The results release quantifies delivery: 850 megawatts of additional data-center capacity delivered in the quarter, more than 300,000 GPUs delivered since the end of the prior quarter, described as almost triple the capacity delivered in the May quarter. The quarterly report does not quantify contracted or energised megawatts anywhere. What it does quantify is commitment: $288 billion of leases not yet commenced, and $34.150 billion of unconditional purchase and other obligations with terms of one year or longer, which the filing says relate "primarily to long-term supply arrangements for purchasing components for cloud infrastructure assets and power supply arrangements for data centers." Those obligations run $5.449 billion for the remainder of fiscal 2027, $6.176 billion in fiscal 2028, $3.863 billion in fiscal 2029, $4.318 billion in fiscal 2030, $6.768 billion in fiscal 2031, $954 million in fiscal 2032 and $6.622 billion thereafter. The balance sheet shows the spend landing: property, plant and equipment, net rose from $99.957 billion to $127.845 billion in three months, and unpaid capital expenditure — capital spending accrued but not yet paid — stood at $6.247 billion.

Liquidity itself is not in question over the near term. Cash, cash equivalents and marketable securities were $37.077 billion, working capital $8.116 billion, and the quarterly report states the company believes its balances "together with cash generated from operations and available financing arrangements, will be sufficient to meet our working capital, committed capital expenditures and contractual obligations for at least the next twelve months." The question is the twelve months after that, and the price of the money that fills them.

What the Street and Public Conversation Say

The quality of available discussion is mixed and should be read as positioning, not evidence.

The clearest outside framing came from three headlines on September 11. CNBC's read was "Oracle posts 30% revenue growth fueled by AI cloud demand as debts hits $125 billion" — the same two facts, weighted against each other. MarketWatch ran "Oracle shook off fears about AI spending, but its stock still loses ground," quoting an analyst that "several of Oracle's bear cases were addressed head on." MarketWatch separately attributed the day's best S&P 500 performers, Dell and HPE, to the print, writing that "Oracle's earnings provided an upbeat read for hardware suppliers." That combination — good for the supply chain, not good enough for the issuer — is the market's actual verdict.

The pre-print commentary is worth keeping on file because it set a testable expectation. On September 2, MarketWatch reported a Bernstein view that Oracle could be "nearing the end" of its need to raise cash. The quarter that followed drew the remaining $20 billion of equity authorisation and took on $11.4 billion of customer prepayment. On September 8, MarketWatch attributed a move in the stock to the OpenAI ecosystem "coming back into favor," noting Oracle "has a lot riding on the ChatGPT creator's success." Separately on September 8, Yahoo Finance headlined that EU regulators had sent an early warning to Oracle ahead of earnings; no article body was captured and the substance of that warning is not established here.

A social sweep on September 11 returned 19 raw items and 5 relevant ones, which is thin. The only market-relevant item was from @zerohedge, reporting that Oracle shares turned negative after erasing a gain of as much as 8.5%, and asserting that Oracle credit default swaps spiked. The first half is consistent with the tape; the second is an unverified assertion from a low-reliability account with no credit-spread data on file to check it against, and it is recorded here as a claim, not a fact. Two Schwab Network segments appeared, one titled "ORCL Beats Earnings & Raises Guidance on AI Cloud Growth, Stock Stalls," the other still framing the backlog at the prior $638 billion, which dates it before the print. The remaining items were retail video and an unrelated open-source licensing thread.

News coverage was pulled on the evening of September 10, before the September 11 session. Four of its five items were quote and landing pages rather than articles. Nothing was pulled after Friday's close, so the September 11 decline itself has no news receipt behind it.

The Conviction Chain

Conviction is Low because the third link is unresolved and the fourth has already failed once. Nothing in the quarter challenges the demand case; everything in it sharpens the funding case.

Financial Proof

All figures GAAP, reported currency, fiscal quarters ending in the month shown. Fiscal 2026 totals are confirmed by SEC XBRL company facts on an annual basis (revenue $67,357M, operating income $20,606M, net income $17,087M, diluted EPS $5.83, filed June 22, 2026). Fiscal 2026 second-quarter net income of $6,135M includes a one-time gain from the sale of Oracle's interest in Ampere, which is why the company guides fiscal 2027 second-quarter earnings growth both including and excluding it.

Quality of earnings in the August quarter, on the filed lines: GAAP operating income $6.728 billion on a 35% margin against non-GAAP operating income of $8.151 billion on a 42% margin, the $1.423 billion bridge being $1.127 billion of stock-based compensation, $202 million of intangible amortisation and $94 million of restructuring. GAAP diluted earnings per share $1.56, up 55%; non-GAAP $1.92, up 30%. Depreciation more than doubled to $3.156 billion from $1.351 billion, which is the buildout starting to land in the income statement. Interest expense $1.428 billion against $923 million.

As of 2026-09-23 at $144.56 and the 3.024 billion common shares outstanding reported at August 31, 2026, common-equity market value is about $437 billion — roughly 24.8x fiscal 2026 GAAP diluted earnings per share of $5.83 and about 17.8x the company's own fiscal 2027 non-GAAP earnings guide of $8.10. The 6.50% Series D Mandatory Convertible Preferred Stock, carried at $4.954 billion, sits above the common and will convert; 64 million shares and awards were excluded from diluted earnings per share as anti-dilutive this quarter, that preferred among them.

Two balance-sheet lines deserve attention alongside the multiples. Stockholders' equity rose from $43.056 billion to $67.196 billion in three months, almost entirely from the share sale rather than retained earnings. Other non-current liabilities rose from $16.178 billion to $28.183 billion, and current deferred revenues from $9.916 billion to $14.686 billion — the balance-sheet footprint of the customer prepayments.

The Stock Setup (as of 2026-09-18 — history; the current setup is at the top of this page)

The long trend is broken, and the 200-day average at $166.25 is the level that would begin to repair it. This is not a violent pullback inside an intact uptrend — the one-year return is negative by half and the excess return against the benchmark is negative by 65 points, which is the signature of a stock the market has re-rated rather than merely sold. The constructive condition, stated as a condition and not an instruction, is a base that reclaims $166.25 on above-average volume and holds it, with free cash flow turning less negative in the same window. The destructive condition is a loss of the 50-day average at $141.00. Oracle qualifies on neither the long-trend leadership screens nor the washout-turn screen as of September 18.

The news receipt dated September 16 carries a report that Larry Ellison cancelled a planned sale of Oracle stock, which reads as executive confidence after the print rather than a fresh catalyst. Nothing dated names this week's decline.

Short-interest data is held with a missing or stale receipt, so positioning cannot be read from that side. Insider activity in the trailing ninety days consists of one Form 4: Vice Chairman Jeffrey Henley, an option exercise and sale of 400,000 shares for about $63.7 million, transacted June 24 and filed June 26 — near the lows, and structurally an exercise-and-sell rather than an open-market disposal, which carries less signal.

Read-Through to the Neocloud Lane (the September 10–11 print window — history)

On September 10, before the print, the GPU-cloud independents sold as one: a 4.9% median decline with none of ten members up, while the platforms that buy that capacity for cash closed roughly flat. Oracle fell 5.4% that day on 2.4x average volume. The September 10 brief named the test explicitly — whether the lane would follow Oracle up on Friday, in which case Thursday was a rates day, or whether Oracle would rally alone, in which case every self-funded balance sheet was being repriced.

Neither branch happened. The lane rose and Oracle did not.

Six of the nine lane members with a full year of price history rose on the day Oracle's print was digested; Oracle finished last of all ten. On the long frame the separation is starker. Nebius is up 151.77% over one year and sits 42.4% above its 200-day average — a deep pullback inside an intact trend. IREN is up 33.42% over one year but 4.0% below its long average, so its trend is genuinely being tested. CoreWeave is down 21.03% over one year with 37.25 points of negative excess return and sits 3.1% below its long average, which puts it in the same broken category as Oracle, one rung less severe. Oracle is the worst one-year performer in its own cohort by more than thirty points.

Separate what Oracle's print proves from what it does not. It proves that a large, investment-grade buyer contracted for compute at scale and delivered 850 megawatts and 300,000 GPUs against those contracts in a single quarter — that is demand evidence for the lane, and the market read it that way, bidding up both the independents and the hardware suppliers. It does not prove that the pure-plays can deliver their own backlogs, because none of them has Oracle's balance sheet, its $10 billion commercial paper program, its investment-grade access or its ability to sell $20 billion of stock into the market in a quarter. It proves the opposite of comfort on financing: the best-capitalised operator in the lane exhausted its equity authorisation and still ran negative free cash flow. Any pure-play funding the same buildout faces the same arithmetic with less capacity to absorb it.

Catalysts

• Second-quarter fiscal 2027 results, expected in December 2026. Guidance is for total revenue growth of 30% to 34% in both constant currency and reported currency, total cloud revenue growth of 64% to 70% in constant currency, and non-GAAP earnings per share of $1.83 to $1.91 in constant currency. Including the prior year's Ampere gain, the company expects non-GAAP earnings per share to decline between 19% and 15% in constant currency.
• Any new equity or convertible authorisation. The at-the-market program is exhausted; a replacement filing is the single most informative document Oracle could file next.
• Credit-rating actions. The annual report ties the terms and availability of data-center leases to the rating, which makes a downgrade a business event, not just a funding one.
• Whether customer prepayments repeat. The $11.363 billion prepayment line was the largest single contributor to the record cash-flow quarter. Its recurrence or absence in the November quarter is the cleanest test of whether the funding structure is durable or one-off.
• Quarterly dividends. A common dividend of $0.50 per share is payable October 23, 2026 to holders of record on October 9, and the preferred dividend of $1,625 per share is payable October 15 to holders of record October 1.

Risks

• Funding cost compounds against a fixed obligation. Interest expense is annualising near $5.7 billion, senior notes and other long-term borrowings carry at $125.0 billion, and the market marks that paper about 15% below carrying value with the discount widening over the quarter.
• Counterparty credit inside the backlog. No customer exceeds 10% of delivered revenue, but the backlog grew from $138 billion to $664 billion in fifteen months on "certain significant cloud contracts," and the annual report states plainly that returns depend on "the ability of our key customers to meet their contractual obligations."
• Dilution. Shares outstanding rose 5.0% in one quarter. The mandatory convertible preferred will convert. The buyback has $6.3 billion of remaining authorisation and was not used at all in the quarter.
• Commitment ahead of demand. $288 billion of not-yet-commenced leases on fifteen-to-nineteen-year terms, plus $34.150 billion of unconditional purchase and power obligations, are contractual regardless of whether the load arrives.
• Depreciation catches up. Depreciation more than doubled year over year to $3.156 billion. Property, plant and equipment grew $27.9 billion in a quarter; the earnings drag from that base is ahead, not behind.
• Legal and regulatory. A securities class action naming the company, officers and a director is pending over a prior share-price decline, and a September 8 report described an EU regulatory early warning whose substance is not established here.

Kill Tests

1. A second quarter of flat-to-down backlog with the twelve-month recognition share stuck at or below 13% would mean the contracts are lengthening rather than converting, and the twenty-two-year ratio stops being a duration story and becomes a credibility one.
2. A new equity authorisation larger than $20 billion within two quarters would confirm that the funding gap is structural rather than a timing effect, and would invalidate the view that this quarter's issuance was a one-time catch-up.
3. Customer prepayments failing to recur in the November quarter while capital expenditure rises as guided would put the entire shortfall back onto debt at current spreads.
4. A ratings downgrade would, on the company's own filed language, touch lease terms and customer eligibility, converting a financing risk into an operating one.
5. Reclaiming and holding $167.64 on above-average volume while free cash flow turns less negative would falsify the bear case from the other direction and re-open the long trend.

Cross-References

AI spending: the supplier earnings reversal case. This quarter weighs against the counter-case on its demand leg and for it on its financing leg. The counter-case fails if customers maintain spending and suppliers sustain orders and margins through March 2027; Oracle maintained spending emphatically — capital expenditure of $28.499 billion, guidance for fiscal 2027 above fiscal 2026's $55.663 billion, more than $30 billion of new contracts booked — and the market's own reading, per the September 11 attribution of Dell's and HPE's gains to the print, was that suppliers benefit. But the campaign's stated scope explicitly includes "delivery economics and financing as possible transmission channels," and that is exactly where this quarter delivered evidence: the largest and best-capitalised buyer in the lane funded its quarter by exhausting a $20 billion equity program and taking $11.4 billion of customer prepayment, with its own long-dated debt marked 15% below carrying value. The transmission channel to watch is no longer order cancellation; it is the cost and availability of the money that pays for the orders.

AI power: generation and grid equipment. Oracle supplies contracted-demand evidence for the upstream case and a caution on the conversion test. The $34.150 billion of unconditional purchase and other obligations is described in the filing as relating primarily to component supply and "power supply arrangements for data centers," which is a direct, filed, dated commitment to buy power — and the $288 billion of not-yet-commenced data-center leases on fifteen-to-nineteen-year terms extends that commitment well past the campaign's twelve-to-thirty-six-month horizon. The caution is the distinction the campaign's own review condition names: separate approvals and contracted capacity from energised load. Oracle quantifies delivered capacity — 850 megawatts in the quarter — only in its results release, and its quarterly report quantifies no megawatt figure at all, contracted or energised. The commitment is documented; the energisation is not, and on Oracle's disclosure it cannot be verified from the filings.

Oracle appears in no long-trend leadership tier and in no washout-turn row as of September 11. It is carried on the value-screen, cloud-ETF-holdings and AI data-center operator lists.

Financial Proof

Filed-period check (as of 2026-09-28): The newest verified quarter ended 2026-08-31 and was filed 2026-09-11. Revenue was $19.34B; GAAP net income was $4.76B; diluted EPS was 1.56. Source: Massive income statements and the filing. Dated historical tables below retain their own period basis.

The recent filed revenue series shows the direction of the business; missing cells are left unavailable.

Refresh History

Sources

• Oracle Q1 FY2027 results release, Form 8-K Exhibit 99.1, September 10, 2026 — revenue by offering, non-GAAP reconciliation, balance sheet, cash-flow statement, the fiscal 2026 quarterly free-cash-flow and net-cash-outlay tables, guidance, RPO and delivered-capacity statements.
• Oracle Form 10-Q for the quarter ended August 31, 2026, filed September 11, 2026 — RPO recognition schedule, ATM Program utilisation, preferred and common equity, lease maturities and the $288B not-yet-commenced commitment, unconditional obligations, senior-note fair value, liquidity and capital resources.
• Oracle Form 10-K for fiscal year ended May 31, 2026, filed June 22, 2026 — prior-period RPO and lease commitments, fiscal 2026 financing detail, commercial paper and revolving facilities, customer-concentration statement, indebtedness and downgrade risk factors, securities litigation.
• edgar{key-financials.json,company-facts.json}` — SEC XBRL company facts, fiscal-2026 annual basis, filed 2026-06-22 and 2026-09-11.
• massive{income-statements,balance-sheets,cash-flow-statements}.json` — eight-quarter provider series through fiscal 2026 Q4, used as a cross-check only; see caveat below.
• Project Jupiter report (attributed; X post with an LLM-written context note): @ShanuMathew93, 2026-09-19 — 2026-09-19-thebuildout-oracle-project-jupiter-data-center-build-faces-seven-month-d-6a86ff15.
• Rates, 2026-09-23: Bloomberg, five-year Treasury above 5%.
• Desk reopen condition (February low, $136): 2026-09-22-market-brief.
• latest — news receipt, fetched 2026-09-16.
• research/stock-events/ — no ORCL row dated 2026-09-12 to 2026-09-23.
• income-statements — filing cache; newest cached period ends 2026-05-31, so the first-quarter fiscal 2027 figures here are folded from the primary filing rather than the cache.
• latest — news receipt stamped 2026-09-11T00:52Z, ok, not stale, 5 items; and news/articles.jsonl.
• 2026-09-10-orcl-earnings — non-GAAP EPS $1.92 against $1.74 consensus, +10.4%.
• CNBC, "Oracle's stock edges up on earnings beat as cloud infrastructure revenue more than doubles," September 10, 2026 — captured at 2026-09-10-cnbc-oracle-s-stock-edges-up-on-earnings-beat-as-cloud-infrastruc-108361309.
• CNBC, "Oracle posts 30% revenue growth fueled by AI cloud demand as debts hits $125 billion," September 11, 2026.
• MarketWatch, "Oracle shook off fears about AI spending, but its stock still loses ground," September 11, 2026.
• MarketWatch, "Why Dell and HPE were the S&P 500's top-performing stocks today," September 11, 2026.
• MarketWatch, "How Oracle's 'extremely controversial' stock could more than double from here," September 2, 2026 — the Bernstein "nearing the end" of cash-raising view.
• MarketWatch, "Oracle's stock gets a boost as the OpenAI ecosystem comes back into favor," September 8, 2026.
• Yahoo Finance, "EU regulators send early warning to Oracle ahead of earnings," September 8, 2026 — headline only; no body captured.
• @zerohedge, September 11, 2026 — "Oracle shares turn negative, erasing gain of as much as 8.5%" and an unverified CDS-spike assertion; via 2026-09-11-social-search-orcl.
• Schwab Network, "ORCL Beats Earnings & Raises Guidance on AI Cloud Growth, Stock Stalls" and Schwab Network, "ORCL $638B RPO Faces AI Spending Surge & MSFT, GOOGL, AMZN Competition" — the second still frames the backlog at $638B, dating it before the print.
• 2026-09-10-market-brief — the September 10 lane read (neoclouds -4.9% median, 0 of 10 up; ORCL -5.4% on 2.4x volume) and the stated Friday test.
• filing-watch — Form 4 receipt: J. Henley, Vice Chairman, codes M/S, 400,000 shares, ~$63.7M, filed 2026-06-26; primary document.
• 2026-09-11-monster-scan and 2026-09-11-washout-scan — no ORCL row in either.
• 2026-09-04-ORCL — prior dated tape history, condensed into Refresh History above.

Provider caveat. The eight-quarter provider series carries a fiscal 2026 fourth-quarter net income of $4,526M against the $4,304M reported in Oracle's own quarterly table, and its share-count fields for that quarter are flagged as implausible. It also omits the second quarter of fiscal 2025 and of fiscal 2026 entirely. Every figure in this profile is taken from the SEC filings and the results release; the provider series is cited only as a cross-check and none of its disputed values is used.

2026-09-28 refresh receipts

• 2026-09-28 settled-close receipt: *, ORCL row dated 2026-09-28; tape values above are bound to the current summary fields.

• 2026-09-28 financial-period check: income-statements and key-financials; newest verified 2026-08-31 period, filed 2026-09-11 (Massive statement / SEC period cross-check).

• 2026-09-28 news receipt check: latest, fetched 2026-09-25; receipt is not current; latest-session cause remains unassigned.

• 2026-09-28 stock-event scan: research/stock-events/, no ORCL row dated 2026-09-24 through 2026-09-28.

• 2026-09-28 source-page check: Reuters reported September 24 that Oracle sent the developer of Project Jupiter a force-majeure notice, citing potential power delays; Oracle said the campus remained on its planned schedule. Oracle's April project announcement contemplated up to 2.45 GW of Bloom fuel cells. The notice makes power delivery a concrete backlog-conversion risk, but it does not by itself establish a delay, contract cancellation or the cause of the September 28 share move. The Chrome CDP source-page capture is under 2026-09-28-raw.

• 2026-09-28 captured-follows scan: tweets has 5 September 24–28 document(s) tagging ORCL.

• Retrospective social-search receipt (captured 2026-09-29): 2026-09-29-social-search-orcl-oracle-corporation; quality/attention check only, not a September 28 event or price source.

• 2026-09-29 browser discovery check for ORCL: news-discovery-google-sep29 (key ORCL); search links only, not evidence of an event or of absence of news.