Oracle Closes FY26 with Cloud Growing 47% and a $638 Billion AI Contract Backlog
Quarterly revenue of $19.2 billion and Oracle Cloud Infrastructure growing 93% in Q4 place the company at the forefront of growth among hyperscalers. RPO increased by $85 billion in three months.
Oracle delivered its largest quarter in history on June 10. Q4 revenue reached $19.2 billion, a 21% increase over the same period last year. Cloud revenue hit $9.9 billion, a jump of 47%. The line that supports the narrative, Oracle Cloud Infrastructure (OCI), closed the quarter at $5.8 billion, growing 93% year over year, according to the official release published on oracle.com.
The data that repositioned the hyperscaler thesis on Tuesday night was another. Remaining Performance Obligations, the queue of contracted but unrecognized revenue contracts, increased by $85 billion in the quarter. They rose from $553 billion in February to $638 billion by the end of the fiscal year, according to the company's investor relations. Management reported that $75 billion of this increase comes from large-scale AI contracts where the customer prepaid for GPUs or provided them directly for use in Oracle's infrastructure.
The Geometry of the Backlog
The boundary between contracted portfolio and executable capex has become the story of Oracle. The company raised $43 billion in debt and $5 billion in equity during fiscal year 2026 to fund the expansion of data centers serving this backlog, according to the release. For fiscal year 2027, Safra Catz, CEO, indicated plans to raise approximately $40 billion more in a combination of debt and equity.
"Our strategic investments in cloud and AI are generating significant returns, positioning Oracle as a leader in the industry," Catz stated in the earnings announcement. The statement sounds institutional, but what it translates to is specific. The $638 billion backlog needs to be converted into revenue over a multi-year horizon with defended operating margins. Oracle currently has more contracted future revenue than it will produce over the next five years combined, and the conversion of this queue depends on physical server and energy capacity, not demand.
OCI Grows Faster Than Industry Average
OCI grew 93% in Q4 and closed FY26 at $18.1 billion, a 77% increase for the year. For comparison, AWS reported a 17% growth in its last quarter, Microsoft Azure around 33%, and Google Cloud at 35%, according to the latest releases from the three companies. Oracle, due to its smaller base, captures the margin of the second movement: AI customers who require rapid capacity and are willing to sign long-term contracts for dedicated GPUs.
The SaaS component, Cloud Applications, closed Q4 at $4.1 billion, growing 10%. The total for FY26 in Apps reached $15.9 billion, up 11%. NetSuite, Fusion ERP, and HCM remain strong, but the vector that drives the multiple now is OCI. The pipeline for migrating SAP workloads to OCI gained traction throughout the year, according to public records from Oracle at the OpenWorld and CloudWorld events.
Where the Backlog Meets the Real World
For CIOs and partners from the Big 4 selling Oracle implementation, the practical reading has two sides. In Anglo-Saxon markets, the United States and the United Kingdom absorb the bulk of the announced capex, with data centers planned in Texas, Virginia, and England. The mass of Nvidia GPUs that these contracts require keeps the Oracle-Nvidia relationship central to the long-term thesis and exposes Oracle to supply chain risks from the same supplier that serves Microsoft, Google, and Meta.
In continental Europe, the challenge is different. Germany and France, with more aggressive data sovereignty regulation under the EU AI Act and DORA, demand OCI Sovereign regions and are among the next expansions announced. In Latin America, Oracle operates regions in São Paulo and Vinhedo, and the growth of the local portfolio depends on private banks and state government. The Oracle-OpenAI agreement announced on June 10, which allows Oracle customers to apply Universal Credits to access OpenAI models and Codex via OCI, opens an additional vector for European and Brazilian financial institutions that already have commitments to Oracle and want to test OpenAI without a new procurement process.
Larry Ellison's thesis in recent calls has been predictable: Oracle will build more cloud data centers than all competitors combined. Q4 does not confirm the thesis but redefines the metric by which it will be judged. If the backlog turns into revenue according to the contracted timeline and the operating margin does not decline, the re-rating of the stock against Microsoft and Google is defensible. If the conversion is delayed due to physical capacity, the discount against peers returns. The story of the next 18 months is not one of demand, but of execution.