Financials
Financials: A 48-Year-Old Software Company Bets the Balance Sheet on AI
Oracle is no longer the business its 10-year income statement describes. For most of the last decade it was a mature, fortress-margin software company that converted roughly a third of revenue into free cash flow and used it to shrink its share count relentlessly. In the last 18 months it has become something else entirely: a hyperscale data-center builder funding one of the largest infrastructure capital programs in corporate history. The single fact that reframes the whole page is the backlog — total remaining performance obligations (RPO, the contracted revenue not yet delivered) ended fiscal 2026 at \$638 billion, up 363% year-over-year [1]. To deliver it, Oracle spent \$48 billion of net cash on capex in FY2026 against \$32 billion of operating cash flow [2] — which means free cash flow has gone deeply negative, and the equity story now hangs on whether that backlog converts to cash before the debt does.
The financial debate here is not "is Oracle a good software business" — it plainly is, with ~80% gross margins and 30%+ operating margins. The debate is whether a 30%+ operating-margin software company can take on a utility-scale, debt-funded capex cycle without breaking its balance sheet — and whether a backlog dominated by a handful of AI customers is worth what the market pays for it.
FY2026 Revenue ($B)
▲ 17% YoY
Total RPO Backlog ($B)
Operating Cash Flow ($B)
Net Capex Outlay ($B)
FY2026 Non-GAAP EPS
Source: Q4 and FY2026 earnings presentation, FY2026 Financial Highlights [2]. Non-GAAP EPS of \$7.63 includes one-time net investment gains; excluding them it was \$6.83 [2].
The standard scorecard: nine years of statements
Read this table top-to-bottom and the inflection is unmistakable. Revenue growth was stuck near zero for years (FY2018–FY2021), then reaccelerated as the Cerner acquisition and cloud demand kicked in. Operating cash flow climbed steadily. And then look at the capex and free-cash-flow columns for FY2025: capex tripled to \$21.2 billion and free cash flow turned negative for the first time in the modern era — before FY2026 took it far lower. FY2018 net income is artificially depressed by a one-time \$8.8 billion charge from the 2017 U.S. tax-reform repatriation, not an operating problem.
Sources: FY2025 10-K Consolidated Statements of Operations [3], Cash Flows [4], and Balance Sheets [5] for FY2017–FY2025; FY2026 revenue, cash flow and capex from the Q4 and FY2026 earnings presentation [2]. FY2026 EPS is the sum of reported quarterly diluted GAAP EPS; margins/net income/equity left blank pending the FY2026 10-K. FY2026 free cash flow is operating cash flow less the \$48B net capex outlay.
A learner's note on the vocabulary used throughout: operating margin is operating income divided by revenue (how much profit the core business throws off before interest and tax); free cash flow (FCF) is operating cash flow minus capex (the cash actually left over after keeping the business running and growing); RPO is the dollar value of signed contracts not yet recognized as revenue — a backlog. Each is defined once here and used plainly below.
Growth quality: real demand, but increasingly capital-bought
Oracle's top line reaccelerated from a low-single-digit grind to 17% growth in FY2026 [2]. The engine is Oracle Cloud Infrastructure (OCI): Q4 FY2026 cloud-infrastructure (IaaS) revenue reached \$5.8 billion, up 93% [20]. That is high-quality demand in the sense that it is real, contracted, and accelerating — but it is a different kind of growth than the old Oracle. Database and applications growth was nearly free to scale; OCI growth requires Oracle to pour tens of billions into GPUs, land, and power first, then collect revenue over the life of the contract.
Source: FY2025 10-K Statements of Operations for FY2017–FY2025 [3]; FY2026 from the Q4 and FY2026 earnings presentation [2].
Margins tell the supporting story. Cloud services and license support — the recurring core — was \$44.0 billion of the \$57.4 billion FY2025 total [3], and GAAP operating margin recovered to 30.8% in FY2025 [3] from the 25.7% trough caused by the FY2022 Cerner integration. The margin dip and recovery is the tell: Oracle can absorb a large acquisition and rebuild profitability. The open question is whether OCI's hardware-heavy mix structurally lowers the through-cycle margin as it becomes the dominant growth driver.
Source: derived from FY2025 10-K Consolidated Statements of Operations [3]. FY2018 net margin reflects a one-time tax-reform charge, not an operating decline.
The backlog bombshell: \$638 billion of RPO
This is the chart that explains Oracle's valuation, its capex, and its risk all at once. RPO sat at \$137.8 billion at the end of FY2025 [6]. One quarter later it had vaulted to \$455 billion, up 359% year-over-year, with cloud RPO up "nearly 500%" [7]. By the end of FY2026 it stood at \$638 billion [1]. A roughly 4.6x increase in twelve months is not organic software growth — it is a step-change driven by a small number of very large multi-year AI-infrastructure contracts.
Sources: FY2025 RPO from the FY2025 10-K MD&A [6]; Q1 FY2026 from the Q1 FY2026 call [7]; FY2026 from the Q4 FY2026 call [1].
Management frames the backlog as "exceptional visibility into our future revenue growth, all supported by long-term contractual customer commitments" [1]. That is the bull case in one sentence. The bear case is in the same fact: a backlog this concentrated is only as good as the counterparties behind it and Oracle's ability to fund the build-out that converts it. RPO is a promise of future revenue, not cash in hand — and the cash to fulfill it has to be spent years before the revenue is collected.
Earnings quality: the cash conversion crux
For a buy-side reader, this is the decisive section. Oracle's reported earnings are real — net income was \$12.4 billion in FY2025 [3] and operating cash flow was \$20.8 billion [4], rising to \$32.0 billion in FY2026 [2]. Operating cash conversion is excellent. The problem sits one line below: capex went from a steady ~\$2 billion a year through FY2021 to \$21.2 billion in FY2025 [4] and a \$48 billion net cash outlay in FY2026 [2]. Free cash flow — the lifeblood of the old Oracle — has flipped from roughly +\$12 billion a year to about -\$16 billion in FY2026.
Sources: FY2017–FY2025 from the FY2025 10-K Statements of Cash Flows [4]; FY2026 operating cash flow and net capex outlay from the Q4 and FY2026 presentation [2]. FY2026 FCF is operating cash flow less net capex; reported gross capex is higher still.
It does not stop here. Management guided FY2027 to a net cash capex outlay of around \$70 billion, adding that customer prepayments and timing of roughly \$20–25 billion mean reported capex will be even higher [8]. In other words, the free-cash-flow hole gets deeper before it gets better. The entire thesis reduces to a timing question: Oracle is spending the cash now and recognizing the \$638 billion of revenue later. If utilization and contract economics hold — management says contracted margins are "maintaining and improving" — today's negative FCF is an investment, not a leak. If demand softens or a major customer falters, Oracle will have built capacity it cannot monetize, financed with debt.
Balance sheet: from net cash to heavily levered
Five years ago Oracle's balance sheet was a non-issue. It now sits at the center of the risk. Three forces hollowed out the equity base and loaded on debt: a multi-year buyback spree (below), the \$28.2 billion Cerner acquisition completed in June 2022 [11], and now the capex build. Shareholders' equity went negative — to -\$6.2 billion in FY2022 — and the company still carries an accumulated deficit of \$15.5 billion because cumulative buybacks have exceeded cumulative retained profit [5]. Total notes payable and other borrowings stood at \$92.6 billion at May 31, 2025 [10], and the borrowing has since accelerated sharply to fund capex.
Sources: total debt from Oracle 10-K Notes Payable disclosures, FY2020–FY2025, with the FY2025 figure of \$92.6B cited to the FY2025 10-K [10]; equity from the Consolidated Balance Sheets [5]; Q3 FY2026 (Feb 2026) debt, cash and equity as reported. Net debt is total debt less cash and equivalents.
The financing machine behind the build-out is now running at full tilt. In FY2025 Oracle issued \$14.0 billion of senior notes, lifting interest expense [14] — which already ran to \$3.6 billion for the year [3]. It followed with a large multi-tranche senior-notes issuance in September 2025 spanning maturities from 2030 to 2065 [13], and in February 2026 issued 6.50% Series D Mandatory Convertible Preferred Stock [12] — reaching for equity-like capital, a notable step for a company that spent the prior decade retiring shares. By Q3 FY2026 total debt had climbed to roughly \$124.7 billion, with net debt near \$86 billion against a fast-growing but still-modest equity base.
On the reassuring side, the debt is long-dated rather than a near-term wall — the scheduled principal payments stretch out across fiscal years through the 2060s [17] — and Oracle backstops liquidity with an undrawn \$6.0 billion revolving credit facility [15]. Management states its cash, operating cash flow and borrowing arrangements are "sufficient to meet our working capital, capital expenditures and contractual obligations requirements" [16]. The honest read: net debt/EBITDA around 3x is elevated but not alarming for a business with this margin and contracted backlog — provided EBITDA keeps compounding and the capex actually converts. The credit risk is not today's leverage ratio; it is the trajectory, with another ~\$40 billion of debt-and-equity funding planned for FY2027 [8].
Capital allocation: the pivot from buybacks to bricks
No chart captures the strategic shift better than this one. Oracle was historically one of the most aggressive repurchasers in tech — it bought back \$36 billion of stock in FY2019 alone and over \$90 billion across FY2019–FY2022, cutting the diluted share count from 4.2 billion to under 2.9 billion. Then it stopped: buybacks collapsed to \$0.6 billion in FY2025 [4] even though \$6.4 billion of repurchase authorization remained available [9]. Every available dollar now goes into the ground. The dividend, by contrast, has been protected and grown steadily to \$4.7 billion in FY2025 [4].
Source: FY2025 10-K Consolidated Statements of Cash Flows, FY2017–FY2025 [4].
This is a coherent bet, not a drift: management is willingly trading the buyback-driven per-share compounding that worked for a decade for a much larger, riskier prize in AI infrastructure. A note on return-on-equity: Oracle's reported ROE looks spectacular (60%+) but is largely a math artifact of the tiny, buyback-depleted equity base — it is not a clean signal of capital efficiency here, and a reader should lean on operating margin, cash conversion and incremental return on the capex instead.
Valuation: priced for the build-out to pay off
At a price of \$184.29 (June 18, 2026) and roughly 2.9 billion shares, Oracle's market capitalization is about \$535 billion; adding ~\$86 billion of net debt puts enterprise value near \$620 billion. Against FY2026 results that is roughly 24x non-GAAP EPS of \$7.63 (about 27x the \$6.83 that strips out one-time investment gains) [2], ~9x EV/revenue on FY2026, and about 23x the company's own FY2027 non-GAAP EPS guidance of \$8.05 [18].
Share Price (Jun 18, 2026)
P/E on FY26 Non-GAAP EPS
P/E on FY27 Guided EPS
Mean Analyst Target
Sources: share price as reported (company filings); non-GAAP EPS from the Q4 and FY2026 presentation [2]; FY2027 guided EPS [18]; analyst price targets per consensus estimates, as reported.
Is that cheap or expensive? Only relative to what it implies. On a backward-looking multiple, ~24x non-GAAP earnings is a clear premium to Oracle's own pre-2024 history, when the stock traded in the mid-teens as a slow-growth value name. But it is being underwritten on a forward story: management guides FY2027 revenue to ~\$90 billion, +34% in constant currency [18], and a long-term outlook of a 31% revenue CAGR and 28% non-GAAP EPS CAGR through FY2030 [19]. If those numbers land, today's multiple is modest. Notably, the stock at \$184 sits well below the ~\$253 mean analyst target (range \$155–\$400) — the market is pricing in meaningfully more execution risk than the sell-side consensus, consistent with the FCF and leverage concerns above. (A standardized Quality Score and a model Fair Value estimate were not available in this dataset, so the assessment rests on the company's own statements, the multi-year record, and peer comparison.)
How Oracle stacks up against its named peers
Oracle's true competitors come in two flavors: the enterprise-application/database rivals it names (SAP, Salesforce, Workday, IBM) and the cloud-infrastructure giants (Microsoft Azure, Amazon AWS) it competes with on OCI. The hyperscalers (MSFT, AMZN) are far larger and more diversified, so treat those rows as scale context rather than like-for-like comparables. The table's most important column is the last one: Oracle is the only company in the set with negative free-cash-flow margin — the direct financial signature of its capex bet.
Sources: company financial data as reported (each peer's latest fiscal year). Oracle shown on FY2025 GAAP for comparability; its growth accelerated to 17% in FY2026 [2].
Oracle's operating margin (30.8%) is comfortably above the application peers (SAP, CRM, WDAY) and well below Microsoft's 45.6% — respectable for its mix. But where every peer converts revenue into positive free cash flow (Salesforce at a remarkable 35%), Oracle has chosen to spend through breakeven. That is the premium-vs-discount crux: the market is paying a growth multiple for a company that, uniquely in this set, is consuming cash to chase a backlog the others cannot match. The bet only pays if OCI's contracted economics are as durable as management claims.
The verdict
What the financials confirm: Oracle has a genuinely high-quality core — recurring cloud and license revenue at ~80% gross and 30%+ operating margins, accelerating to 17% growth — and a backlog (\$638 billion RPO) that gives extraordinary forward visibility if it converts. Operating cash flow conversion of reported earnings is strong and rising.
What they contradict: the long-running "Oracle = cash machine that buys back its stock" thesis is over. Free cash flow is deeply negative and guided to stay negative, equity has been hollowed out, net debt is ~\$86 billion and climbing, and the company is now issuing convertible preferred stock — equity-like capital — to help fund the build. The quality of the business and the safety of the balance sheet have diverged.
The swing factors are two and they are linked: (1) whether the \$638 billion RPO converts to cash at the contracted margins management promises, and (2) whether Oracle can fund a multi-year, ~\$70 billion-a-year capex program without the leverage or customer concentration cracking. Everything else is secondary.
The first financial metric to watch is the free-cash-flow trajectory against capex — specifically, whether operating cash flow (FY2026: \$32B) grows fast enough to close the gap on the ~\$70 billion FY2027 net capex outlay [8]. As long as FCF is deeply negative and debt is rising to fill the hole, Oracle is a credit-and-execution story wearing a software-multiple valuation. The quarter the FCF drain begins to narrow — or, conversely, the quarter a marquee RPO customer wavers — is the quarter the entire investment case re-rates.