Oracle FY2026 Q4 Earnings: $638 Billion RPO Masks a $5.8 Billion Cloud Reality

On June 10, 2026, Oracle reported fourth-quarter revenue of $19.2 billion and net income available to common shareholders of $4.2 billion, while disclosing $638 billion in remaining performance obligations (RPO) — up $85 billion in a single quarter — against just $5.8 billion in cloud infrastructure revenue. It is the widest promise-to-P&L gap in tech and the clearest test of whether backlog can convert to cash before capex does.

The Headline Numbers​

Oracle's fiscal year ends May 31; Q4 FY2026 covers March–May 2026. The first table is fiscal 2026 annual (Form 10-K / companyfacts) — the same totals as the Beancount fence. The second table is the Q4 press-release print that drove the RPO story.

Annual revenue grew 17% while Q4 RPO grew 363% YoY and $85B sequentially — the headline is not only the P&L, it is the backlog. (Annual net income available to common was $16,984M after $103M of preferred dividends.) The ledger shows what that backlog costs: every dollar of RPO will require capacity that hits PP&E and then depreciation before it hits revenue.

Revenue Deep Dive​

Oracle discloses cloud and software mix that feeds the story. Q4 mix from the release:

Within cloud, IaaS was $5.8B (up 93% YoY) and SaaS $4.1B (up 10%). IaaS is still a minority of revenue even after that growth — when a segment growing ~90% is still a fraction of the mix, the durability question is conversion: can IaaS scale before the RPO clock runs?

The Margin Story​

Annual margins below are from the Form 10-K / companyfacts rebuild (fiscal years ended May 31). Q4 is shown separately so it is not confused with the annual fence.

FY2021's elevated net margin reflects a large tax benefit in that year; from FY2023 through FY2026 the story is cloud mix and scale, not a one-off. Q4 tax on the release was $1,066M — not a mechanical low-tax print.

The One Big Question: Can $638B of RPO Convert Before Capex Does?​

The defining question is not whether Oracle sold $85B of new RPO in one quarter — it did — but whether it can deliver it without a capex ramp that breaks the margin just expanded. Free cash flow for fiscal 2026 was negative $23.7B while operating cash flow hit a record $32.0B, per the release — the gap is the datacenter build.

At 110× IaaS, Oracle's backlog-to-delivery ratio is an order of magnitude above peers. That is either a multi-year compounding machine or a capex trap — and the balance sheet tells which.

Tracking fiscal 2026 in plain text​

Double-entry forces every dollar to reconcile, which is why the Beancount ledger is the audit. The income-statement transaction below is the fiscal 2026 annual filing (year ended May 31, 2026) — not the Q4 slice. Negative income, positive expenses, and the check that proves they sum to zero. Q4 headline numbers stay in the tables above.

; Revenue 67357; cost 26530; R&D 10272; SG&A 9949; other 1052; tax 2467; net income 17087.
; Check: -67357 + 26530 + 10272 + 9949 + 1052 + 2467 + 17087 = 0 ✓
2026-05-31 * "Oracle Corporation" "FY2026 Income Statement"
  Income:Revenue                         -67357 MUSD
  Expenses:CostOfRevenue                   26530 MUSD
  Expenses:ResearchAndDevelopment          10272 MUSD
  Expenses:SellingGeneralAdministrative    9949 MUSD
  Expenses:OtherNet                      1052 MUSD
  Expenses:IncomeTax                       2467 MUSD
  Equity:Adjustments                      17087 MUSD  ; net income offset (RE set by balance assertion)

That block is not an illustration; it is the period validated with bea check / yarn open-ledger-audit and pushed to open_ledger/oracle after rebuilding FY2021–FY2026 from Form 10-K XBRL (companyfacts) and the FY2026 10-K. The balance sheet tells the same story on the other side: assets = liabilities + equity at each period end, with residuals in Other explicitly noted.

The balance-sheet numbers that matter most for this print: total assets $261,759M, PP&E $99,957M (up from $43,522M a year earlier), and Goodwill $62,261M plus intangibles $3,229M — still a large acquisition overhang from Cerner, now beside a PP&E ramp that will test whether RPO converts to gross profit or to depreciation.

The Multi-Year Arc​

Revenue +66% from FY2021 to FY2026 while year-end assets roughly doubled on the datacenter build — RPO ended FY2026 at $638B after a $85B sequential add from Q3's $553B. The ledger lets you test whether that asset ramp earns its keep without trusting a chart.

The Verdict: Bull vs. Bear​

Bull Case

• RPO converts at 15–20% per year, making IaaS a much larger share by FY2028 without incremental capex beyond guidance.
• Cloud mix lifts growth even as software license stays flat-to-down.
• The $85B quarterly RPO add is repeatable — prepaid and customer-supplied GPUs ($75B of large AI contracts per the release) reduce Oracle's own financing need.
• Ledger history shows Oracle has absorbed large acquisitions (Sun, NetSuite, Cerner) without permanent margin collapse.

Bear Case

• RPO is concentrated in a few AI customers with “take-or-pay” that can be renegotiated, not sticky ARR.
• Delivering RPO requires tens of billions of PP&E that drives depreciation ahead of revenue, compressing operating income — FY2026 already printed negative free cash flow of $23.7B.
• IaaS at ~$5.8B in one quarter cannot absorb $638B of backlog quickly — 110× coverage implies many years of delivery at current IaaS scale.
• Deferred revenue and customer deposits fund capex; a slowdown leaves the balance sheet leveraged after $43B of FY2026 debt raises.

Our Take: The Q4 print proves Oracle can sell backlog faster than any peer, but it does not yet prove it can deliver it at a margin. The ledger now exists — rebuilt from the filings — so that question can be answered with numbers, not narratives: next period's PP&E, deferred revenue, and free cash flow will either confirm conversion or expose the capex trap.