The Backlog Says $638B, the Cash Flow Says −$24B: Record RPO and a $90B FY27 Guide Collide With a $55.7B Capex Bill and a $40B Raise — Maintaining Outperform

Results vs. Consensus

Q4 FY2026 Scorecard

Year-Over-Year Comparisons

Full-Year FY2026 Snapshot

Quality of Beat

Revenue: The 0.4% headline beat understates the quality on the cloud side and overstates it on the total. OCI at +93% and total cloud at +47% are the engine; the legacy software line (cloud license & on-premise support) was −2%, a continuing slow bleed as customers migrate to cloud. The mix shift is the whole story: cloud is now ~52% of quarterly revenue and growing roughly 9x faster than the legacy base it is cannibalizing. There were no FX windfalls (USD growth ran 1 point ahead of constant currency) and no acquisitions distorting the organic figure.

Margins: This is where the build shows up. Non-GAAP operating margin held at 45% and operating income grew 22%, but management was explicit that gross margin declined in the quarter — and stepped down ~5 points for the full year — as data-center ramp costs and the acceleration of lower-initial-margin infrastructure revenue hit the cost line. The operating-margin resilience was manufactured below the gross line, by cutting operating costs (sales & marketing efficiency). That is a legitimate lever, but it means the margin story is now a race between infrastructure gross-margin maturation and the pace of the capex ramp, not the high-70s-gross-margin software model investors grew up with.

EPS: The $2.11 non-GAAP figure includes a one-time net gain on Oracle's Ampere and Bloom Energy investment stakes; management quantified the underlying growth at +20% ex-gain (~$2.04). The GAAP-to-non-GAAP gap is wide ($1.45 GAAP vs. $2.11 non-GAAP) and widening as stock-based compensation and the depreciation step-up grow. We treat ~$2.04 as the clean operational figure — still a beat, but the headline +24% should be discounted to the +20% underlying number when comparing to the thesis.

Segment Performance

Revenue by Line — Q4 FY2026

Cloud Infrastructure (OCI) — +93%, The Whole Thesis in One Line

OCI grew 93% to $5.8B, an $18.1B full-year business (+77%) that is now the fastest-scaling hyperscale-class infrastructure franchise in the market. The demand is not generic compute — it is AI training and inference plus a fast-growing cloud-database layer. The proof points management put on the table are unusually concrete for a cloud business: 97.5% global GPU utilization; 35,000 GPUs from 59 customers came up for renewal in the quarter, of which 49% of customers renewed 92% of the GPUs, and most of the un-renewed capacity was re-sold to other customers within the same quarter. That is a supply-constrained market behaving like one — idle capacity does not exist for long.

"OCI has been the fastest growing cloud provider for years. And now with AI infrastructure, we have shown to everyone the power of the organization we built… Everything we see shows this market size is trillions of dollars per year. Combined with our previously outlined 30% to 40% margin profile, OCI should grow into an extremely large and extremely profitable business." — Clay Magouyrk, co-CEO

Assessment: OCI is doing exactly what the bull case requires, and the 97.5% utilization figure is the single best evidence that the backlog is real demand rather than speculative reservations. The open question is not whether OCI grows — it is what the through-cycle margin and the GAAP economics look like once depreciation on $55B+ of annual capex fully loads. Magouyrk's "30–40% margin profile" is an operating-margin target, not a current GAAP disclosure; we want to see the segment-level proof in coming quarters.

Cloud Applications (SaaS) — +10%, The Quiet Compounder

SaaS at $4.1B (+10%) decelerated modestly but the deferred-revenue signal is healthier than the in-quarter print: SaaS deferred revenue grew 16%, faster than recognized revenue, which typically precedes a reacceleration. Oracle took thousands of customers live in the quarter (over 300 in Fusion alone), continued the U.S. Department of Veterans Affairs EHR rollout (now 14 medical centers, ~29,000 clinicians, ~500,000 veterans), and disclosed a post-quarter agency-wide U.S. Office of Personnel Management award for Fusion HCM that seeds FY27. Over 1,000 AI agents are now embedded across the application suites at no incremental charge.

"We think double digit growth on an in-quarter run rate of $4.1 billion is pretty good… our deferred position in the quarter grew by 16%. So with our deferred position growing faster than our in-quarter revenue, it gives us confidence." — Mike Sicilia, co-CEO

Assessment: SaaS is no longer the headline, but it is the margin ballast and the AI distribution channel — the fastest, cheapest way for Oracle's installed base to consume AI is to keep using Fusion as the agents arrive quarterly. The deferred-revenue acceleration argues the 10% is a floor, not a fade. We would watch for the new agentic-pricing layer (below) to show up as a SaaS growth contributor over the next two to three quarters.

Cloud Database & Multicloud — +404%, The Second Engine Forming

Cloud database revenue grew 29%, but the multicloud sub-line — Oracle Database running inside AWS, Azure, and Google Cloud — grew 404% with bookings up 325%. This is the structural revenue vector we flagged at the Q3 FY25 upgrade now compounding off a larger base, and management characterized it as still in "early innings," with new regions and partner clouds still being lit up.

Assessment: Multicloud is the most underappreciated line in the model: it monetizes Oracle's database moat without requiring Oracle to own the data center, so it carries a fundamentally better capital profile than OCI. A +404% line is small today but it is the cleanest high-margin growth in the entire P&L. If it sustains even half that rate, it becomes a material standalone driver by FY28.

Key Operating KPIs

Key Topics & Management Commentary

Overall Management Tone: Confident on demand, deliberately reframing on cost. The new financial leadership spent the call repositioning how investors should evaluate Oracle during the build — introducing a "net cash outlay for capex" metric that strips out customer prepayments, anchoring returns to a high-20s steady-state ROIC, and reconfirming the long-term FY30 CAGRs at the first opportunity. Where the call was less convincing was on the near-term cash trajectory: management acknowledged the gross-margin step-down and the funding need but framed both as timing rather than structure, and left the customer-concentration and dilution questions largely to inference. The posture is a company that believes the demand is settled and is now managing the market's anxiety about how it gets paid for.

1. RPO of $638B: The Visibility Engine

"Our remaining performance obligations, or RPO, finished at $638 billion, up 363%. This unprecedented level of RPO provides exceptional visibility into our future revenue growth, all supported by long-term contractual customer commitments… we expect 12% to be recognized in the next 12 months and another 34% between 13 and 36 months." — Hilary Maxson, CFO

RPO is the spine of the Oracle thesis and it just inflected again. At $638B against $67.4B of trailing revenue, Oracle is carrying ~9.5x annual revenue in contracted backlog. The recognition schedule matters: ~12% (~$77B) converts within twelve months and another ~34% (~$217B) within three years, meaning roughly $294B is scheduled inside the FY27–FY29 window and ~54% sits beyond — the long-dated AI-infrastructure contracts that fund the data-center build. Management said both near-term percentages should accelerate in coming quarters as capacity comes online.

Assessment: No other large-cap software company has this. RPO of this magnitude transforms the forecasting problem from demand estimation to delivery execution — the revenue is contracted; the question is whether Oracle can build and power the capacity to recognize it. That is a materially better problem to have than guessing at end-demand, and it is the core reason the Outperform stands.

2. The Capex Bill: $55.7B Spent, ~$90B–$95B Coming

"We will continue those investments in our fiscal year 2027, with an expected net cash outlay for capital expenditures of around $70 billion. This includes customer prepayments and timing impacts expected at around $20 billion to $25 billion, so our reported CapEx will be higher by this amount." — Hilary Maxson, CFO

This is the number that moved the stock. FY26 reported capex was $55.7B — above the ~$50B the Street carried — on revenue of $67.4B, an 83% capex-to-revenue ratio that is unheard of for a software company and rivals the heaviest telecom and utility build-outs. FY27 guidance steps it up again: ~$70B of net cash outlay, which management is careful to distinguish from reported capex (~$90–95B once the $20–25B of prepayments and timing are added back). The "net cash outlay" metric is new this quarter, and it is doing real work — it reframes the headline capex down by the prepayments Oracle collects upfront from BYOH/prepaid customers.

Assessment: The new metric is analytically defensible — if a customer pre-funds the hardware, Oracle's actual funding need is lower — but it is also a presentational choice introduced at precisely the moment the headline capex number became uncomfortable. We will use it, with eyes open: the reported figure is what hits the cash-flow statement and the depreciation schedule, and at ~$90–95B it is the largest in the company's history by a wide margin. The bull and bear cases on Oracle now both run through this single line.

3. Free Cash Flow: −$23.7B and the ROIC Defense

Operating cash flow was a robust $32B (+54%) — but capex of $55.7B turned free cash flow to −$23.7B for the year, with depreciation nearly doubling to ~$7.6B as the data-center fleet loads onto the books. For a company that generated positive free cash flow for decades, a −$24B FCF print is a regime change, and the FY27 capex guide implies the trough deepens before it recovers. Management's rebuttal was to move the conversation to return on invested capital.

"The way I think about return from that business model is in return on invested capital. And what we see is return on invested capital in the high 20s at a steady state… I am just calculating return on invested capital as after-tax operating margin plus depreciation, divided by gross investments, so total gross CapEx at the project level." — Hilary Maxson, CFO

Assessment: The ROIC framing is the right lens for a contracted-demand build — if a project earns high-20s ROIC at steady state and the demand is pre-sold, negative interim FCF is an investment, not a loss. But the figure is a back-of-envelope, project-level, steady-state number, and it excludes the risk that GPUs require replacement capex before the contracts fully season. We accept the framework and withhold judgment on the magnitude until Oracle discloses segment-level OCI economics. This is the number the bears will attack and the bulls will defend for the next year.

4. The Funding Plan: ~$40B Raise, Including $20B of Equity

"To support our capital investments program, we expect to raise around $40 billion in debt and equity in our fiscal year 2027, and that includes our already-announced $20 billion at-the-market equity issuance. We do not anticipate raising additional debt funding in calendar year 2026." — Hilary Maxson, CFO

Oracle already raised $43B of debt and $5B of equity in FY26; FY27 adds ~$40B more, half of it equity via the at-the-market program. Management paired the plan with an explicit commitment to "disciplined capital allocation… maintaining a strong balance sheet, and preserving our investment-grade credit rating." The equity component is the part the market dislikes: an ATM issuance dilutes existing holders, and a $20B program against a ~$560B market cap is roughly 3–4% of shares, a real per-share headwind layered on top of the negative FCF.

Assessment: The mixed debt/equity funding is prudent risk management given the leverage Oracle is already taking on — but the equity dilution is a genuine cost to per-share value that the $8.05 EPS guide already absorbs and that the bears are right to flag. Preserving investment grade while funding a ~$90B capex year is the tightrope; management says it can, and the prepayment mechanics make the cash math more favorable than the gross numbers suggest. Still, this is the single clearest negative in the quarter.

5. Gross-Margin Compression: Timing or Structure?

Full-year gross margin stepped down ~5 points, and management guided a further FY27 step-down before recovery "as we reach full contractual revenue levels at our data centers." The mechanism is straightforward: infrastructure revenue ramps before the data center is fully utilized, so early-life OCI capacity carries a depressed gross margin until the contracted revenue fully seasons. The offset has been operating-cost discipline — sales & marketing efficiency held the operating margin roughly flat despite the gross-margin decline.

Assessment: This is the most important unresolved analytical question in the model. If the compression is purely ramp-timing, blended margins recover as the fleet fills — consistent with the high-20s ROIC claim. If AI-infrastructure is structurally lower-margin than legacy software (which it is, at the gross line), then the blended Oracle margin resets permanently lower even as dollars grow. The truth is probably both: a permanent mix-down in gross margin, partly offset by OCI scale economics and a capital-light multicloud layer. We model a lower blended gross margin than the legacy-Oracle base case but trust the operating-margin floor given the demonstrated cost discipline.

6. Bring-Your-Own-Hardware & Prepaid: $75B at No Margin Degradation

"We signed $67 billion in AI infrastructure contracts this quarter, the majority of which was either bring-your-own-hardware or prepaid. This increases our combination of bring-your-own-hardware or prepaid customer contracts to $75 billion, with those contracts having no degradation in margin compared to our other contracts." — Clay Magouyrk, co-CEO

This is the structural innovation that makes the capex bill survivable. In a BYOH or prepaid deal, the customer either supplies the accelerators or pre-funds them, while Oracle still earns the full operating margin on designing, building, securing, networking, and operating the cluster. Magouyrk's point — that a rack of accelerators "on their own is not a functioning cloud" — is the moat argument: the differentiated value is the cloud Oracle wraps around the silicon, not the silicon itself.

Assessment: BYOH/prepaid is the most important capital-allocation development of the quarter and the reason the funding plan is fundable at all. It lowers Oracle's cash outlay (the customer pre-funds), preserves the operating margin (no degradation), and improves ROIC (cash collected upfront). The bear question is whether the mix can stay high as the business scales; management would not commit to a split. We treat the $75B as a genuine de-risking of the build, with the caveat that we want to see the mix hold above ~80% of new AI bookings.

7. Leadership Transition: Co-CEOs and a Two-Week-Old CFO

This was the first full earnings call under Oracle's new structure: co-CEOs Clay Magouyrk (infrastructure/OCI) and Mike Sicilia (applications/database), with Larry Ellison as Chairman & CTO and Safra Catz transitioned to Executive Vice Chair, plus a brand-new CFO, Hilary Maxson, who opened by noting she had been at the company two weeks. A leadership transition of this depth — CEO and CFO simultaneously — is landing at the most capital-intensive, financially scrutinized moment in Oracle's history.

"I think we are reconfirming those long-term targets in the sense of the CAGRs that we put into the slide today… you can see the RPO building to the level that you can start to have a lot of belief in those long-term targets. So full reconfirmation from my side." — Hilary Maxson, CFO

Assessment: The split makes operational sense — infrastructure and applications are now genuinely different businesses with different capital profiles, and having a dedicated operator on each is defensible. But it concentrates execution risk: the funding model is being managed by a CFO with two weeks of tenure, and the co-CEO structure has a mixed track record across the industry. We do not penalize the rating for it — Ellison remains the strategic anchor and the targets were reconfirmed — but it is a new variable on the watch list that did not exist a year ago.

8. Agentic Pricing: Token Bundles and Outcome-Based Models

Oracle introduced two new commercial models: prepaid "token bundles" for advanced AI reasoning consumed across the application suites, and outcome-based pricing that ties cost to measurable value (candidates screened by a recruiting agent, upsell transactions by a hospitality agent, patient throughput in healthcare). The limited Q4 rollout already had 33 customers, including Aon and Liberty Energy pre-purchasing tokens. Much of the embedded AI remains free; the bundles monetize the heavier-reasoning workloads on top.

"What is new is that we are now expanding that offering across our entire fleet… across all of our applications, including the Fusion suite… certainly should be helpful for our growth story as well." — Mike Sicilia, co-CEO

Assessment: Outcome-based pricing is the right answer to the enterprise CFO's central AI question — "what is my ROI?" — and Oracle's full-stack position (it owns the apps, the data, and the infrastructure) makes it one of the few vendors that can actually measure the outcome it charges for. This is early and immaterial to the FY26 numbers, but it is a credible new SaaS growth lever and a differentiated answer to the "SaaS apocalypse" narrative. Watch for it to register in SaaS growth by mid-FY27.

9. FY27 Guide and the FY30 Framework

Management guided FY27 total revenue to ~$90B (+34% constant currency, which it noted surpasses the 5-year revenue CAGR in the long-term plan), non-GAAP EPS to $8.05 (+18% adjusted, excluding the FY26 one-time gains), and reconfirmed the Analyst Day framework of +31% revenue and +28% EPS CAGR through FY30. The revenue acceleration to +34% from FY26's +17% is the clearest statement that the backlog conversion is beginning in earnest, weighted to the second half as capacity comes online.

Assessment: The FY27 revenue guide is the headline bullish data point that the after-hours sell-off buried — a +34% constant-currency acceleration on a $67B base is the RPO converting to recognized revenue exactly as the thesis predicted. The EPS guide of $8.05 is more muted (+18%) because the margin step-down and dilution offset the revenue surge. The tension in the guide — revenue accelerating, EPS growth decelerating — is the entire Oracle debate in two numbers.

Guidance & Outlook

The guide tells a coherent story when read as a whole: revenue accelerates hard (FY27 +34% cc vs. FY26 +17%) as contracted RPO converts, but the cost of getting there — the margin step-down, the depreciation load, and the equity dilution — holds non-GAAP EPS growth to +18%. Management explicitly flagged that revenue and earnings will "accelerate in the second half of the year as we bring further megawatts online," so the Q1 +27–29% is the trough of the FY27 trajectory, not the run-rate.

Implied ramp: A $90B full-year guide against a Q1 implying ~$21B (+28%) requires the back half to run materially hotter — consistent with the "Q1 FY27 capacity delivery ≈ prior four quarters combined" comment. The revenue is gated by power and construction, not demand.

Street at: Consensus FY27 EPS sat at ~$8.01; the $8.05 guide is a modest raise. Revenue consensus was already near $90B, so the guide validated rather than surprised — the disappointment was entirely on the capex/funding/FCF axis, not the top line.

Guidance style: Oracle has a multi-quarter pattern of guiding cloud conservatively and clearing it. The reconfirmation of the FY30 CAGRs by a brand-new CFO, explicitly tied to the visible RPO build, reads as a credibility down-payment rather than a stretch.

Analyst Q&A Highlights

Capex intensity and component-cost pass-through

The dominant opening line of questioning pressed on why capex ran above expectations and how Oracle protects its margins when component costs — memory and storage in particular — are inflating. Management split the answer: the capex variance was framed as timing (an effort to accelerate spend to pull revenue forward), not component-price inflation, and the margin protection comes from contract structure — fixed-price where costs are known and locked, pass-through where they are not.

Q: "You spent a little bit more this quarter on CapEx than we expected… component costs have gone up a lot, especially memory… when it comes to these very long-term contracts between you and the end customer and the suppliers, can you explain that to investors?"
— John DiFucci, Guggenheim Securities

A: "When we are selling stuff in a time period where we have certainty… we will then do fixed-price contracts. But times that we do not know those costs — because it is out too far in the future or we have too much supply-chain risk — we then do not do fixed-price contracts with our customers. And we have a mechanism whereby those costs end up being passed through… we have a very robust set of mechanisms that ensure that Oracle is not sitting there with reduced margins."
— Clay Magouyrk, co-CEO

Assessment: The pass-through mechanism is the right structural answer to the memory-inflation fear that has dogged the whole AI-infrastructure complex, and it directly supports the "no margin degradation" claim on BYOH/prepaid. The tell is that management volunteered the cost-pass-through framework unprompted — it knows component inflation is the bear's first weapon and pre-empted it.

Reconfirming the FY30 framework under new financial leadership

A recurring concern was whether a CFO two weeks into the role would stand behind targets set by her predecessor. The response was an unqualified reconfirmation, explicitly grounded in the visible RPO build rather than in inherited conviction.

Q: "You have long-term targets out there. You are a new CFO… can you just comment on those long-term targets at all? I know you have only been there a couple of months."
— John DiFucci, Guggenheim Securities

A: "I think that we are reconfirming those long-term targets in the sense of the CAGRs that we put into the slide today… you can see the RPO building to the level that you can start to have a lot of belief in those long-term targets… So full reconfirmation from my side on the long-term targets."
— Hilary Maxson, CFO

Assessment: A new CFO reaffirming a predecessor's multi-year CAGRs on day fourteen is a meaningful credibility signal — she is staking her early reputation on numbers she inherited, which she would not do without having seen the backlog underpinning them. It de-risks the leadership-transition concern, though it does not eliminate it.

How to evaluate returns during the heavy-investment period

One line of questioning, from an investor accustomed to capital-intensive industries, asked directly how to judge Oracle's progress while free cash flow is deeply negative. This produced the most useful new analytical disclosure of the call — a concrete ROIC anchor.

Q: "As you come to Oracle from a capital-intensive business in another industry, how would you suggest that investors evaluate Oracle's progress and returns during this period of heavy investment?"
— Brad Zelnick, Deutsche Bank

A: "The way I think about return from that business model is in return on invested capital. And what we see is return on invested capital in the high 20s at a steady state… once the revenues have ramped for large projects at the project level… and if we are generally able to preserve and improve in the case of things like bring-your-own-hardware, the ROIC structure for those types of structures will be even higher."
— Hilary Maxson, CFO

Assessment: A high-20s steady-state ROIC, if it holds, justifies the entire build — it is well above Oracle's cost of capital and reframes negative interim FCF as a timing artifact of front-loaded investment against pre-sold demand. The caveat is "steady state" and "back-of-envelope"; the number excludes GPU-replacement capex and is not yet a disclosed segment metric. It is the most important figure to track and the one the rating ultimately leans on.

Competitive landscape as capacity floods the market

With neoclouds, hyperscalers, and new entrants all racing to build AI capacity, a recurring question was whether the supply wave threatens Oracle's pricing, renewals, or margins. Management's answer leaned on the persistent demand-supply imbalance and on relationship durability rather than on any claim of insulation from competition.

Q: "With so many vendors entering the market to deliver AI data centers… where does Oracle see itself in the competitive landscape? And how do you see that increase in capacity impacting your ability to retain customers, capture new customers, and maintain or improve margins?"
— Mark Moerdler, Bernstein

A: "Several years in, there is still a massively higher demand than there is supply… I really focus on how do we make sure that we can meet as much of that demand at a reasonable margin profile… over time, as the market continues to mature and we deploy more of our R&D into making things more efficient, I think there are ways that Oracle gets higher and higher margins but we offer lower and lower prices to our customers."
— Clay Magouyrk, co-CEO

Assessment: The honest framing — demand still vastly exceeds supply — is more credible than a claim of competitive immunity, and the 97.5% utilization figure substantiates it. The "higher margins, lower prices" formulation is the classic scale-economics flywheel; it is aspirational here, but Oracle's decade of OCI efficiency work gives it more standing to claim it than a first-generation neocloud.

The "classic" database and applications business beneath the AI noise

Against a backdrop of "SaaS apocalypse" anxiety in software, a line of questioning probed the health of the legacy database and applications franchises. Management argued the classic business is not just stable but reaccelerating where it touches AI — via multicloud database and embedded agents.

Q: "We talked a lot about AI… but you still have the classic Oracle business that we all grew up with… there is a lot of noise in the market… can you talk to what you see in the classic business, on the database side and the application side?"
— Raimo Lenschow, Barclays

A: "We think double-digit growth on an in-quarter run rate of $4.1 billion is pretty good… our deferred position grew by 16%… on database, multicloud revenue growing 4x, bookings 3.25x in the quarter… we are in very early days on multicloud database… expect that business to continue to be an outsized growth engine for Oracle going forward."
— Mike Sicilia, co-CEO

Assessment: The deferred-revenue acceleration (SaaS +16% vs. recognized +10%) and the +404% multicloud line are the two data points that rebut the SaaS-apocalypse narrative for Oracle specifically. The mission-critical installed base is sticky, and AI is a reason to stay rather than to leave. This is the quietest bullish element of the quarter.

Bring-your-own-hardware mix trajectory and the capex-guide clarification

A two-part closing exchange pressed on where the BYOH/prepaid share of RPO could ultimately settle, and sought to pin down exactly what the $70B FY27 capex figure included. Management declined to forecast the mix but clarified the funding mechanics and the return benefit of collecting cash upfront.

Q: "About 12% of RPO is not related to these types of deals… where do you think that split could ultimately go? And… just to clarify on the CapEx guide, I think you said $70 billion in CapEx, but that was excluding ~$25 billion from some of these prepaid deals?"
— Kirk Materne, Evercore ISI

A: "For fiscal year 2027, we expect around $70 billion in net cash outlay for capital expenditures. That does exclude the $20 to $25 billion in prepayments that we will collect… The sum of those is our reported CapEx. But… because we are collecting money upfront… the return on capital is going to be a bit better as well."
— Hilary Maxson, CFO

Assessment: The refusal to commit to a BYOH/prepaid mix target is the honest answer in a fast-evolving model, but it leaves the single most important capital-structure variable unforecast. The clarification confirms the ~$90–95B reported-capex reality behind the friendlier $70B "net" headline. Investors should anchor models to the reported figure and treat the prepayment offset as a funding benefit, not a capex reduction.

What They're NOT Saying

1. Customer concentration in the $638B RPO. At Q1 FY26, a single counterparty (OpenAI) represented roughly two-thirds of the then-$455B backlog. RPO has since grown to $638B, and management noted "4 customers contracting for more than $8 billion this quarter" as evidence of diversification — but offered no updated concentration figure for the backlog as a whole. With the RPO now the central pillar of the thesis, the concentration of it is the single most important undisclosed variable.
2. A quantified FY27 gross-margin floor. Management said gross margin will "step down" again and recover "as we reach full contractual revenue levels," but declined to put a number on either the trough or the recovery slope. For a margin story this central to the EPS bridge, the absence of a quantified floor leaves the bear free to assume the worst.
3. Segment-level OCI economics. Magouyrk cited a "30–40% margin profile" target and Maxson a "high-20s" steady-state ROIC, but Oracle does not disclose OCI's actual current operating margin or its GAAP profitability. With depreciation doubling, the standalone economics of the infrastructure business remain a black box investors must infer.
4. The per-share dilution math. A $20B at-the-market equity issuance was confirmed, but management did not walk through the implied share-count growth or its drag on the $8.05 EPS guide. The dilution is embedded in the guide but never explicitly sized.
5. RPO cancellation and contingency terms. $638B of largely long-dated AI-infrastructure contracts sit on the books, ~54% recognizable beyond three years. No color was given on cancellation rights, take-or-pay structures, or what happens to the backlog if a large customer's compute demand slows — the tail risk that would matter most.

Market Reaction

• Pre-print setup: ORCL closed the regular June 10 session at $201.26, down 2.21% on the day and trading in a wide $185–$212 intraday range — soft into the print. The stock was up only ~10% year-to-date and sat roughly 42% below its autumn-2025 all-time high of ~$346, having de-rated steadily over the prior two quarters as the market repriced the funding profile of the AI build. The 52-week range ($134.57–$345.72) captures the round-trip: a euphoric spike on the Q1 FY26 RPO step-function, then a long grind lower.
• After-hours move: Shares fell roughly 7% after the release and call, despite top- and bottom-line beats and a record backlog. The decline tracked the disclosures in this order of impact: FY26 capex above guide ($55.7B), −$23.7B free cash flow, and the ~$40B FY27 funding plan including $20B of equity dilution.

The reaction is a textbook case of a quarter where the print and the price action point in opposite directions. Every metric the bull case is built on — RPO, OCI growth, contracted bookings, FY27 revenue acceleration — moved the right way, and the stock fell anyway. The market is not disputing Oracle's demand; it is repricing the near-term cash cost of servicing that demand. A −$24B free-cash-flow year funded partly by equity issuance is a genuinely different risk profile than the cash-generative software company investors owned three years ago, and the de-rating since autumn reflects that the market has been adjusting to it for two quarters. Tonight's drop is the same anxiety re-triggered by a higher-than-expected capex print, layered on a stock that entered the day already soft and already well off its highs.

Street Perspective

Debate: Will the $638B backlog actually convert — and at the modeled returns?

Bull view: The backlog is contracted, increasingly prepaid or BYOH-funded, and running against 97.5% utilization with un-renewed capacity re-sold within the same quarter — this is demand the market can underwrite, and the FY27 +34% revenue guide proves conversion has begun.

Bear view: No company has ever converted a backlog of this size at this pace; ~54% sits beyond three years, concentration is high and undisclosed, and "high-20s ROIC" is a back-of-envelope, steady-state figure that ignores GPU-replacement capex.

Our take: The bull has the stronger near-term hand — conversion is no longer theoretical now that FY27 revenue is guided to accelerate to +34%. The bear's concentration point is valid and unanswered, but it is a tail risk, not a base case. The visibility here is unmatched in software; we side with the bull while watching the concentration disclosure.

Debate: Is the funding model sustainable, or value-destructive?

Bull view: Operating cash flow of $32B (+54%) is robust; prepayments lower the true cash need; ROIC is well above cost of capital; and management is preserving investment grade with a balanced debt/equity mix.

Bear view: Free cash flow is −$23.7B and deepening, depreciation is doubling, and Oracle is now issuing equity into the build — diluting holders to fund a capex program larger than its entire revenue base.

Our take: This is the binding constraint on the stock, and the bear is right that it is a real per-share cost. But the de-rating since autumn has already priced much of it, and a high-ROIC build against pre-sold demand is an investment, not a loss. We hold a constructive view with explicit acknowledgment that the FCF trough and the dilution are the legitimate reasons conviction is lower than at the Q1 FY26 peak.

Debate: Margin trajectory — temporary ramp drag or permanent reset?

Bull view: Gross-margin compression is ramp-timing; margins recover as data centers reach full contractual revenue, and operating discipline (S&M efficiency) is already holding the operating margin flat.

Bear view: AI infrastructure is structurally lower gross-margin than legacy software; as the mix shifts permanently toward OCI, blended Oracle margins reset lower for good, even if dollars grow.

Our take: Both are partly right — expect a permanent mix-down in gross margin, offset by OCI scale economics, a capital-light multicloud layer, and demonstrated operating-cost discipline. The operating-margin floor is more defensible than the gross-margin line; we model lower blended margins than legacy Oracle but trust the 40%+ non-GAAP operating margin to hold.

Model Update Needed

Valuation impact: We maintain a fair-value range of roughly $300–$360. At ~$201 post-print, ORCL trades around 25x FY27 non-GAAP EPS of $8.05 — not cheap on EPS, but the EPS multiple understates the case because it capitalizes the cost of the build while the contracted backlog ($638B) that the build serves is recognized over years. On contracted-backlog visibility and the FY27–FY30 revenue CAGR, the de-rated stock sits at a meaningful discount to fair value, which is what underpins the Outperform despite the funding overhang. The risk to fair value is asymmetric to two variables: customer concentration and the depth/duration of the FCF trough.

Thesis Scorecard Post-Earnings

Overall: Thesis intact and, on visibility, stronger than ever — but conviction moderated from the Q1 FY26 peak. The four bull pillars all tripped bullish; the bear case shifted decisively from "capex absorption uncertainty" (a question) to "negative free cash flow and equity dilution" (a fact). The spine of the Outperform — contracted backlog converting to revenue — is delivering exactly as modeled. What has hardened is the price of admission: Oracle is now a capital-incineration story in the near term in service of a contracted-demand story in the medium term.

Action: Maintain Outperform. A de-rated stock (~42% off its high) with accelerating contracted-revenue conversion is the setup the rating framework rewards — a sell-off driven by funding optics rather than a demand crack. Hold the position; the binding watch-items are the FCF trough depth, the pace of equity issuance, and any disclosure on customer concentration. Fair value $300–$360. We would upgrade conviction on a concentration disclosure that de-risks the backlog, and would revisit the rating if the equity-funding pace accelerated beyond the $20B ATM or if OCI gross margins failed to inflect as the FY26 cohort of data centers seasons.