Earnings Call Highlights: Oracle (ORCL) | FY2026 Q3
1. Summary | Oracle Posts Its Strongest Growth in 15 Years as AI and Multicloud Database Fire Together
Oracle delivered a milestone report this quarter: for the first time in 15 years, both revenue and non-GAAP EPS grew more than 20% (organic). The core momentum came from explosive demand for AI infrastructure (+243% YoY) and multicloud database (+531% YoY). Despite market worries that AI could trigger a "SaaS apocalypse," Oracle has turned its AI infrastructure lead into a cascading effect at the application layer by embedding, across Fusion ERP and its industry suites, more than 1,000 AI agents. Management also announced it has strengthened the balance sheet through a 15% stake in TikTok US and a financing plan of up to $50B ($30B already completed), supporting 10 GW of power and capacity build-out over the next three years.
2. Financials (Non-GAAP)
Revenue growth: total revenue and EPS both achieved YoY +20%-plus organic growth.
Key revenue breakdown:
AI Infrastructure: $1.x B (YoY +243%).
Multicloud Database: YoY +531%.
Cloud Applications (SaaS): revenue of $4.0B (YoY +11%), with an annualized run rate of $16.1B.
Profitability:
AI Capacity Gross Margin: held at 32%, better than the prior 30% guidance.
Balance sheet and capital expenditure:
RPO (remaining performance obligations): surged to a staggering $553B, driven mainly by large AI contracts.
Financing and cash: completed $30B of investment-grade bonds and mandatory convertible preferred stock issuance; the remaining $20B is expected to be raised through an ATM equity program.
CapEx strategy: introduced "Bring-your-own-hardware" and customer prepayment models, designed to expand capacity without adding to the company's debt burden.
3. Outlook (FY2026 Q4 / FY2027)
Revenue guidance: management said it will continue to exceed FY2026 expectations and raised its FY2027 financial forecast.
Cloud footprint: AWS regions will expand rapidly from 8 at the end of Q3 to 22 by the end of Q4, achieving global multicloud coverage.
Momentum drivers: RPO is expected to convert to revenue faster as the supply chain improves (rack output up 4x YoY).
4. Product-by-Product Progress
AI Infrastructure (OCI):
Capacity delivery: delivered more than 400 MW of power capacity this quarter, 90% on time or early.
Technical progress: manufacturing footprint tripled and rack output rose 4x. Time to revenue from rack delivery shortened by more than 20%.
Multicloud Database:
Ecosystem expansion: live in 33 regions on Azure and 14 regions on Google Cloud. Database customers are accelerating their migration to the cloud because of AI needs (vector embedding, MPT server).
Strategic SaaS (Fusion & NetSuite):
AI embedding: launched AI Agent Studio, with key applications in healthcare (AI-powered EHR) and finance (automated close agents).
Share gains: repeatedly beat Workday and SAP in ERP/HCM deals this quarter.
5. Looking Ahead
Power and land moat: has locked in 10 GW of power supply and data center capacity over the next three years through partners.
Sovereign AI: offers the full OCI stack through the Alloy platform, targeting governments and highly regulated industries. Whereas competitors offer only edge nodes, Oracle provides a "complete region" that can run locally.
Inference strategy: inference demand is expected to overtake training demand as model utilization rises, and Oracle will bring in a diverse range of accelerators, including NVIDIA, AMD, and Cerebras.
6. Q&A Highlights
Core question (halo effect): how does AI infrastructure lift the traditional businesses?
Management's answer: running AI training on OCI puts private data physically very close to AI models, greatly improving RAG (retrieval-augmented generation) efficiency, which directly drives SaaS upgrade demand and Database migration.
Core question (SaaS apocalypse): will AI replace traditional SaaS?
Management's answer: core business systems (such as bank DDA and healthcare EHR) carry extremely high data complexity and compliance barriers. Oracle treats AI as a "development accelerator," using smaller engineering teams to build a more complete ecosystem—so it is actually the one "disrupting others."
Core question (data center location): does inference need to be close to population centers to reduce latency?
Management's answer: inference latency comes mainly from hardware architecture, not geographic distance. Compared with the 40 ms latency from New York to Wyoming, the AI model's thinking time (seconds) and the efficiency of hardware accelerators are what matter, so data center siting will still prioritize power and land costs.
7. Risks and Watch Points
Capital intensity: even with financing in place, hyperscale data center construction will still test execution.
Supply chain challenges: although rack output has increased, supply of high-end AI accelerators (GPUs) remains short of demand.
Geopolitics and compliance: the 15% TikTok stake and its linkage to the U.S. regulatory environment.
Inference market competition: as AI shifts from training to inference, the penetration of dedicated inference chips (such as Groq and Cerebras) could erode the premium on traditional GPU rentals.
8. Bottom Line
One-line positioning: Oracle is at an explosive sweet spot of hardware-software integration, transforming from a traditional software vendor into a key pillar of global AI infrastructure.
Core moat: "full-stack control" from the database layer through OCI cloud architecture to the SaaS application layer, enabling low-latency data transfer and vertical integration benefits that competitors struggle to match.
Investment takeaway: with the twin tailwinds of a record RPO and expanding multicloud coverage, plus better-than-expected AI Capacity margins, Oracle is showing strong unit economics while scaling, and its valuation still has room to be revised upward.
Comments