The Oracle stock price prediction problem right now is not that the AI story broke. It is that the single number holding the bull case together, a $664 billion remaining performance obligation, is also the number holding the bear case together. Oracle closed at $137.10 on 25 September 2026, down 56.2% from its $313.00 peak on 16 October 2025, per stockanalysis.com daily closes. Yet 43 analysts polled by S&P Global still carry an average target of $237.97. That gap, 74% of upside sitting on a stock the market has halved, is not an argument about whether Oracle can sell AI capacity. It is an argument about who pays for the concrete.
Here is the thing nobody has put plainly. Oracle’s Q1 FY27 operating cash flow of $23.1 billion, the headline “up 184%” figure, contains $11.4 billion of customer prepayments that Oracle’s own Form 10-Q describes as having “a significant financing component”. Management then nets those same prepayments against capital expenditure to present “net cash capex” of $18 billion instead of the $28.5 billion actually spent. The same dollars flatter both halves of the free cash flow bridge. Strip prepayments out of operating cash flow entirely and Q1 free cash flow is roughly negative $16.8 billion, not the negative $5.4 billion printed on the face of the statement. Oracle is not merely spending ahead of revenue. It is funding a meaningful slice of that spend by borrowing from its own customers, and counting the loan twice on the way through.
Key facts
- ORCL closed at $137.10 on 25 September 2026, 56.2% below the $313.00 peak of 16 October 2025 — stockanalysis.com, 25 Sep 2026
- Remaining performance obligations reached $664 billion, up $209 billion year on year and $26 billion sequentially — Oracle 8-K Item 2.02, 10 Sep 2026
- Q1 FY27 free cash flow was negative $5.4 billion on $23.1 billion of operating cash flow against $28.5 billion of capex — Oracle 10-Q, quarter ended 31 Aug 2026
- Total borrowings stood at $125.3 billion against $37.1 billion of cash and securities, for net debt of $88.3 billion — Oracle 10-Q balance sheet, 31 Aug 2026
- Oracle sold $20 billion of common stock through an at-the-market programme in the quarter, lifting share count from 2.880 billion to 3.024 billion, a 4.97% dilution in three months — SEC XBRL company facts, filed 11 Sep 2026
- Quarterly interest expense rose to $1.43 billion from $0.92 billion a year earlier, an annualised run rate near $5.7 billion — Oracle 10-Q statement of operations
- Consensus price target is $237.97 across 43 analysts, with a $110 low and a $400 high — S&P Global Market Intelligence via stockanalysis.com, 25 Sep 2026
What is actually happening, and why the tape disagrees with the analysts
Oracle’s transformation from a database company into an AI landlord is working on every operating measure the company reports. Q1 FY27 total revenue rose 30% to $19.3 billion. Cloud infrastructure revenue rose 121% to $7.4 billion. The company delivered 850MW of additional datacentre capacity and more than 300,000 GPUs to AI cloud customers in a single quarter. GPU utilisation ran at 97.9%, and capacity coming up for renewal was resold at a 20% premium to prior contracts despite most of that hardware being four years or older. Those are not the numbers of a business losing a market.
The market is not pricing the market. It is pricing the balance sheet. Think of Oracle less as a software vendor and more as a build-to-suit industrial landlord who has pre-let the whole estate at good rents but has to finance construction before a single rent cheque clears. The lease book looks magnificent. The construction loan is the problem. Oracle now carries $125.3 billion of borrowings, has issued $4.95 billion of 6.50% mandatory convertible preferred stock trading under the ORCL-PRD ticker, and sold $20 billion of common equity in three months. The equity sale alone diluted existing holders by almost 5% in a single quarter. Our earlier coverage of how Oracle is funding its capex through debt traced the early shape of this; the Q1 filing confirms it at scale.
Chief Financial Officer Hilary Barbara Maxson set out the mechanism explicitly on the Q1 call. Describing the $26 billion sequential increase in backlog, she said “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”, adding that “that new RPO will not impact our CapEx or revenues until fiscal 28 or beyond”, per the Q1 FY2027 earnings call transcript. Read that twice. The newest and largest tranche of the backlog is structured so customers front the capital, and it contributes nothing to revenue for at least a year. The $664 billion is real. It is also further away and less Oracle-funded than the headline implies.
Project Jupiter, force majeure, and what the counterparties did next
On 24 September 2026 Oracle sent a force majeure notice to a unit of Blue Owl Capital, the developer behind Project Jupiter, its roughly 2.4-gigawatt New Mexico datacentre campus tied to the wider Stargate build-out. The notice is a legal shield: it positions Oracle to defer payments should the site fail to come online in 2028, with power procurement being Oracle’s own contractual responsibility. ORCL fell 3.47% that session to $139.54, having traded as low as $133.48 intraday.
Both counterparties pushed back on the reading. Oracle told CNBC that “Project Jupiter remains on our planned schedule” and that the company is “fully committed to New Mexico and confident in our path forward”. Blue Owl Capital, which holds roughly $3 billion of equity in the project through datacentre developer Stack Infrastructure, said in its own statement that “this notice does not change the financial commitments to this multi-year project.” Bloom Energy, contracted to power the facility with fuel cells, also fell on the news; we covered the Bloom Energy setup and its data-centre exposure separately.
Having tracked Oracle’s financing disclosures since the FY26 10-K, the force majeure notice reads to me as the first visible crack between the contracted backlog and the physical world that has to deliver it. A signed RPO assumes megawatts arrive. New Mexico has local opposition ahead of the midterms, environmental objections, and a delayed natural gas pipeline. None of that is on Oracle’s income statement. All of it is on the delivery schedule that converts $664 billion of obligations into revenue. Our reporting on Project Jupiter loans trading at 90 cents and Oracle bonds at 85 showed credit markets reaching this conclusion before equity did.
One counterweight deserves equal billing. On 12 September Oracle filed an 8-K announcing that Larry Ellison, Executive Chair and Chief Technology Officer, “has cancelled his 10b5-1 Plan to sell Oracle stock”, confirming that “no Oracle stock was sold under that plan, and he has no other plans to sell any of his Oracle stock”, per the company’s Form 8-K exhibit 99.1. The plan had been reported by the Financial Times that same day as covering up to $7.5 billion of stock. Whatever the intent, the largest holder chose not to sell into a 56% drawdown.
The valuation maths, and the chart
At $137.10 across 3.024 billion shares, Oracle’s market capitalisation is $414.6 billion. Add $88.3 billion of net debt and $4.95 billion of preferred stock and enterprise value is roughly $507.8 billion. Against $664 billion of contracted backlog, the market is paying about 76 cents for every dollar of obligation Oracle has already signed, a discount we examined when we broke down the 36-month conversion profile of that RPO. Against FY27 guidance of at least $90 billion in revenue and $8.10 of non-GAAP earnings per share, the stock trades on 5.6 times sales and 16.9 times forward earnings.
Sixteen times forward earnings for 30% revenue growth is the entire bull case in one line. The bear case is that $8.10 is a non-GAAP number produced by a company whose GAAP free cash flow is deeply negative, whose interest bill is compounding at $5.7 billion a year, and which has already diluted holders 5% in a quarter to keep building.
| Measure | Bull reading | Bear reading |
|---|---|---|
| $664bn RPO | Twelve years of revenue already contracted | Prepay-funded, no revenue impact until FY28 or later |
| $23.1bn operating cash flow | Record first quarter, up 184% | $11.4bn is customer prepayment with a financing component |
| $28.5bn capex | Net $18bn after customer funding | Gross cash out of the door is $28.5bn regardless |
| $125.3bn borrowings | Cheap capital against contracted cash flows | $5.7bn annualised interest before the estate is built |
| 97.9% GPU utilisation, 20% renewal uplift | Pricing power and long asset life proven | Concentrated in a handful of AI counterparties |
| Force majeure on Project Jupiter | Prudent contractual hedging, schedule unchanged | Delivery risk is now documented in writing |
The structural tension nobody on the call resolved
Deutsche Bank’s Brad Zelnick asked the only question that matters: when does Oracle return to positive free cash flow? Maxson’s answer was candid and unsatisfying in equal measure. “We have not given a particular time frame on that yet, and we do not expect to give that today,” she said, before arguing 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.”
Co-Chief Executive Officer Clayton Magouyrk reframed the question rather than answering it, arguing that investors should “separate out in our minds what Oracle spends as CapEx directly” and “uncouple that directly from how we think about how the business can grow”, because capital “does not all have to flow from Oracle side. It does not have to be Oracle CapEx.” That is a coherent strategy. It is also an admission that the growth is increasingly financed by third parties, whether customers via prepay, vendors via bring-your-own-hardware, or private credit via vehicles like the Blue Owl structure. Each of those arrangements moves capital off Oracle’s capex line without removing the underlying obligation to deliver.
The regulatory and political overlay is thickening in parallel. Datacentre siting has become a live electoral issue in New Mexico ahead of the midterms, ratepayer-protection proposals are advancing in several states, and environmental objections have already delayed the gas pipeline meant to fuel Project Jupiter. For a company whose valuation now rests on a delivery schedule rather than a product roadmap, permitting risk has become financial risk. Magouyrk acknowledged as much on the call when he characterised these builds as “complex projects” requiring air permits and grid coordination.
The call: base $150, bull $205, bear $98
Any Oracle stock price prediction is only as good as its anchor, so every level below is measured against the 25 September regular-session close of $137.10, not the $137.07 after-hours print.
Base case, $150 (+9.4%), probability 45%. Oracle delivers FY27 revenue at or slightly above $90 billion and lands non-GAAP EPS near the $8.10 guide. Free cash flow stays negative through FY27 but the deficit narrows as the Abilene and Texas capacity that is already energised converts backlog into revenue. Project Jupiter slips without a contractual rupture. The multiple stays compressed at 18 to 19 times forward earnings because the market refuses to pay up until cash flow inflects. This is the path where nothing breaks and nothing re-rates.
Bull case, $205 (+49.5%), probability 25%. Two things have to happen together. Oracle has to show a visible free cash flow inflection, most plausibly at the October investor day Maxson flagged on the call or in the Q2 print, and the FY28 revenue bridge has to be credible enough that the market starts valuing the backlog rather than discounting it. A re-rating to 25 times the $8.10 guide gets you to $205. Note this sits well below the $237.97 consensus, deliberately: consensus is anchored to a pre-drawdown multiple that assumed the financing was free.
Bear case, $98 (−28.5%), probability 30%. The trigger is not weak demand. It is a financing accident. A counterparty renegotiation, a second force majeure, a credit downgrade that raises the cost of the next $20 billion, or a prepay customer that cannot pay. Oracle’s 52-week closing low is $114.99, set on 24 July 2026, and the lowest analyst target in the S&P Global panel is $110. A break below both, to roughly 12 times forward earnings, is what it looks like when the market decides to value Oracle on GAAP cash generation rather than adjusted EPS.
What would change my mind. On the bull side, a quarter in which operating cash flow grows while the customer-prepayment line shrinks would prove the cash generation is organic rather than borrowed from the future. On the bear side, the invalidation is simple: if Oracle reports positive free cash flow before the end of FY27, the central objection in this piece collapses and the bear level should be retired, not defended. The October investor day is the next scheduled catalyst; Q2 FY27 results are the one that settles it.
FAQ
Why has Oracle stock fallen so far from its highs?
ORCL is down 56.2% from its $313.00 close of 16 October 2025. The decline is not driven by weak demand: cloud infrastructure revenue grew 121% last quarter. It reflects the cost of growth. Oracle carries $125.3 billion of borrowings, ran negative $5.4 billion of free cash flow in Q1 FY27, and diluted shareholders by almost 5% through a $20 billion equity sale in three months.
What does Oracle’s $664 billion RPO actually mean?
Remaining performance obligations are contracted revenue Oracle has not yet recognised. It rose $209 billion year on year. The important caveat comes from CFO Hilary Maxson: the vast majority of the newest $26 billion tranche is structured via prepay or bring-your-own-hardware and will not affect revenue until fiscal 2028 or later. The backlog is real but back-loaded.
What is the force majeure notice on Project Jupiter?
On 24 September 2026 Oracle notified a Blue Owl Capital unit of force majeure regarding its New Mexico datacentre campus, positioning itself to defer payments if the site misses its 2028 opening. Oracle says the project remains on schedule; Blue Owl says the notice does not change financial commitments to the project. Shares fell 3.47% on the day.
Is Oracle’s operating cash flow as strong as it looks?
Partly. The $23.1 billion headline includes $11.4 billion of customer prepayments that Oracle’s 10-Q classifies as having a significant financing component. Management separately nets those prepayments off capex to report $18 billion of net capex against $28.5 billion gross. Excluding prepayments from operating cash flow entirely implies quarterly free cash flow nearer negative $16.8 billion.
What valuation does Oracle trade on now?
At $137.10 with 3.024 billion shares, market capitalisation is $414.6 billion and enterprise value roughly $507.8 billion including $88.3 billion of net debt and $4.95 billion of preferred stock. That is 16.9 times the $8.10 FY27 non-GAAP EPS guide, 5.6 times the $90 billion revenue guide, and about 0.76 times the $664 billion backlog.
Did Larry Ellison sell Oracle stock during the decline?
No. An 8-K filed on 14 September 2026 disclosed that Ellison cancelled his 10b5-1 selling plan on 12 September, that no stock was sold under it, and that he has no other plans to sell. The Financial Times had reported that day that the plan covered up to $7.5 billion of stock.
This article is analysis and journalism, not investment advice. Price levels are the author’s framework for thinking about risk, not recommendations. Oracle is a volatile, heavily indebted equity in a capital-intensive build cycle; capital is at risk and past performance does not indicate future results. Readers should conduct their own research and consider their own circumstances.