Analysts All In, Insiders All Out — Who Is Wrong?

Oracle (NYSE: ORCL) In-Depth Stock Research Report

Analysis Date: 2026-02-18 · Data as of: FY2026 Q2 (as of 2026-02-18)

Chapter 1: Executive Summary & Investment Highlights

Core Investment Thesis

Oracle Corporation (NYSE: ORCL) is undergoing a strategic transformation from traditional database licensing to cloud infrastructure. This report systematically evaluates Oracle's investment value and risks through 26 chapters of in-depth analysis, focusing on 8 Core Questions (CQs).

Core Questions (CQ) Checklist

CQ1: OCI Backlog Monetization Capability (Weight 15%)

Question: Can the $130B+ RPO be converted into revenue at the expected pace?
Key Uncertainties: RPO cancellation rate, customer migration delays, GPU supply bottlenecks

CQ2: Cloud Market Share Breakthrough Potential (Weight 15%)

Question: Can OCI maintain a 2%+ market share in the IaaS market and break through to 5%?
Key Uncertainties: Durability of AI supercluster differentiation, depth of NVIDIA partnership, multi-cloud architecture adoption rate

CQ3: Can CapEx Translate into Positive FCF? (Weight 15%)

Question: Is the path from $12B+ CapEx to positive free cash flow feasible?
Key Uncertainties: Data center utilization ramp-up, lease obligation structure, customer contract duration

CQ4: Traditional Business Moat Resilience (Weight 15%)

Question: Can the traditional database + ERP business maintain its profit margins amid cloud competition?
Key Uncertainties: Customer migration to open-source databases, Autonomous Database adoption rate

CQ5: Adequacy of Debt Risk Pricing (Weight 10%)

Question: Is the risk of $86B+ total debt (including leases) adequately priced by the market?
Key Uncertainties: Interest rate environment, timing of refinancing wall, credit rating changes

CQ6: Durability of Competitive Moat (Weight 10%)

Question: How long can Oracle's moats endure across the three tracks: database, ERP, and cloud?
Key Uncertainties: AWS/Azure catch-up speed, maturity of open-source alternatives

CQ7: Management Execution Credibility (Weight 10%)

Question: Can Larry Ellison's aggressive expansion strategy deliver on its promises?
Key Uncertainties: Historical guidance accuracy, concentration of governance structure

CQ8: Reasonable Valuation Range (Weight 10%)

Question: Does the current 37x Forward P/E adequately reflect growth expectations and risks?
Key Uncertainties: Growth sustainability, margin trends, comparable company valuations

Chapter 2: Research Mandate & Analytical Framework

Limitations of Technical Differentiation: It is crucial to recognize that OCI's technical advantages are concentrated in the AI infrastructure layer, not in general cloud services. AWS offers 200+ services, Azure is deeply integrated into the enterprise ecosystem, and GCP leads in data analytics/ML—these represent ecosystem barriers that OCI cannot replicate in the short term. OCI's strategy is essentially "point breakthrough" rather than "all-out competition."

2.1 Hyperscale Customer Acquisition: In-Depth Analysis of the OpenAI Case

Strategic Significance of OpenAI-Oracle Partnership:

The $300B five-year agreement signed between OpenAI and Oracle is one of the largest single customer contracts in cloud computing history. The core of this agreement is Project Stargate—building 4.5GW of data center capacity to support OpenAI's AI training and inference needs. The contract start date is 2027, meaning the FY2026 RPO contribution primarily reflects contract signing rather than revenue recognition.

Why Did OpenAI Choose OCI Over AWS/Azure?

The answer to this question is a critical pillar of Oracle's valuation thesis:

1. Diversification from Single Microsoft Dependency: OpenAI was previously fully hosted on Azure, exhibiting clear strategic vulnerability. Choosing OCI is a rational decision to diversify vendor risk, rather than an endorsement of OCI's absolute technical superiority.

2. Cost Efficiency for Large-Scale GPU Clusters: OCI's bare-metal architecture and RDMA network offer verifiable cost advantages in hyperscale (>50,000 GPU) clusters. OpenAI's training demands fall squarely within this sweet spot.

3. Oracle's Aggressive Pricing: Market intelligence indicates that OCI's pricing for large AI contracts is significantly lower than AWS/Azure, potentially even near cost. Whether this is sustainable is a key outstanding question.

Hyperscale Customer Concentration Risk:

Of the $68B in new RPO in Q2 FY2026, management explicitly cited large customer drivers such as Meta and NVIDIA. Combined with OpenAI's $300B contract (representing ~57% of RPO), Oracle faces significant customer concentration risk. If OpenAI were to scale back its commitment to Oracle due to its own financial difficulties, strategic adjustments, or reintegration with Microsoft, the reliability of RPO would significantly decrease.

Other Strategic Customer Expansion:

• Meta: AI inference infrastructure demand, specific contract size undisclosed. While Meta develops its own AI chips (MTIA), it still requires external GPU cloud capacity, with OCI serving as a crucial supplementary vendor alongside Azure. However, Meta's self-developed chip roadmap may reduce its reliance on external cloud GPUs after 2028.
• NVIDIA: Deep partnership—OCI is one of the first partners for NVIDIA DGX Cloud, with deep technical integration between the two parties on AI superclusters. However, NVIDIA is also a partner to all cloud service providers, which does not constitute an exclusive advantage.
• Sovereign Cloud Customers: Oracle has a differentiated positioning in compliance-driven markets (government, finance, healthcare), having deployed over 46 Dedicated Regions globally. Although the total sovereign cloud market is relatively small (estimated $15-20B/year), it offers high profit margins and extremely strong customer stickiness.
• Non-OpenAI Participants in Project Stargate: The project has also attracted capital partners such as SoftBank, suggesting potential future capacity openness to more AI companies, reducing reliance on OpenAI alone—but there are no specific agreements yet.

Customer Acquisition Cost (CAC) Analysis: Oracle's large-customer acquisition strategy implicitly involves a cost structure issue overlooked by the market. To acquire OpenAI-level customers, Oracle needs to:

1. Pre-build customized data centers (upfront CapEx in the billions of dollars)
2. Offer competitive pricing (potentially 15-25% lower than AWS/Azure)
3. Bear the time risk of capacity delivery (delayed delivery may trigger contractual penalties)

This means that for each hyperscale customer acquired, Oracle needs to bear $5-15B in upfront investment. If the customer ultimately does not fully honor the contract (e.g., only using 60-70% of committed capacity), the return on investment will be significantly lower than expected.

2.2 Multi-Cloud Strategy: Significance and Limitations of Oracle Database@AWS/Azure

Oracle's multi-cloud database strategy (Database@AWS and Database@Azure) represents the most noteworthy strategic evolution over the past two years. The essence of this strategy is: acknowledging OCI's inability to compete directly in the general cloud market, and instead embedding Oracle's core asset (database) into competitors' cloud platforms.

Current Progress:

• Database@Azure: Released in 2024, already GA in 10 Azure regions globally in 2025, with 23 more regions planned
• Database@AWS: GA in North America (us-east-1 and us-west-2) in July 2025
• Database@Google Cloud: Planned
• Multicloud Business Revenue: Q2 FY2026 year-over-year growth of 817%, with rapid quarter-over-quarter expansion
• Multicloud Universal Credits: Launched in October 2025, allowing customers to use Oracle credits across clouds

Strategic Significance:

1. TAM Expansion: Traditional Oracle database customers are locked into on-prem or OCI; the multi-cloud strategy expands TAM to AWS/Azure enterprise customer base (hundreds of thousands of companies)
2. Revenue Stickiness: Customers using Database@AWS/Azure simultaneously utilize OCI and target cloud resources, creating a cross-binding effect
3. Database Hegemony Continuation: Larry Ellison set a goal for the multi-cloud database business to reach "$20B revenue within 5 years"

Strategic Limitations:

1. Profit Sharing: Running on competitors' platforms means Oracle needs to share infrastructure profits with AWS/Azure
2. Customer Migration Risk: Database@AWS/Azure may accelerate customer migration from Oracle on-prem to the cloud, but not necessarily to OCI
3. Competitive Paradox: Oracle is helping AWS/Azure customers stay on those platforms, rather than drawing them to OCI
4. Open-source Alternatives: Cloud-native databases like PostgreSQL and CockroachDB are eroding Oracle's mid-to-low end market

2.3 CapEx $21.2B Investment and OCI Revenue Transmission Mechanism

FY2025 CapEx of $21.2B is Oracle's largest capital expenditure year in its history, a 208% increase from FY2024's $6.9B. The direct driver of this surge is AI infrastructure construction—GPU procurement, data center construction, and network infrastructure.

Transmission Mechanism Analysis:

Key Transmission Parameters:

Lag Analysis:

Oracle management clarified a critical timeline during the Q1 FY2026 earnings call: FY2027 will be the inflection point year when CapEx investment begins to convert into revenue on a large scale, with new RPO contributing approximately $4B in additional revenue. However, there is a natural lag of 12-18 months from CapEx investment to revenue recognition:

1. Construction Period (6-12 months): Data centers from groundbreaking to go-live
2. Deployment Period (3-6 months): GPU installation, network configuration, customer migration
3. Growth Period (3-12 months): Customers ramping up from testing to production loads

This means that the $21.2B invested in FY2025 will primarily contribute revenue in FY2027-FY2028, while the planned $35-50B for FY2026 will contribute revenue in FY2028-FY2029.

Depreciation Impact: The CapEx/Depreciation ratio surged from 1.12x in FY2024 to 3.44x in FY2025, meaning a significant amount of new assets has not yet begun depreciation. As new assets come online in FY2026-FY2027, depreciation expenses will rapidly climb from $6.2B in FY2025 to an estimated $12-15B/year, significantly compressing operating profit margins—unless OCI revenue growth is sufficient to offset it.

Historical Benchmark for CapEx Efficiency: Comparing Oracle's CapEx conversion efficiency with AWS's early stages helps calibrate expectations:

Oracle's CapEx/revenue ratio is significantly higher than AWS's historical levels, partly because the unit price of GPUs is much higher than general-purpose servers. An NVIDIA H100 server costs approximately $25,000-40,000, while B200/GB200 are even more expensive. This means that GPU procurement could account for as much as 60-70% of OCI's CapEx, higher than AWS's 30-40%. GPUs also have shorter depreciation cycles (3-4 years vs. 5-7 years for servers), further increasing depreciation pressure.

CapEx Financing Structure: FY2025 CapEx of $21.2B was almost equal to the full-year OCF of $20.8B, resulting in an FCF of -$394M. More critically, Oracle financed this through an additional $5.6B in net debt (total debt increased from $87B in FY2024 to $124B). If FY2026 CapEx indeed reaches $35-50B, while OCF might only grow to $22-25B, the funding gap would amount to $10-25B, requiring further borrowing or a reduction in shareholder returns. The current interest coverage ratio is 4.94x (FY2025), which is acceptable and an improvement from 3.97x in FY2022, primarily due to EBIT growth. If CapEx continues to rise and new debt drives up interest expenses, interest coverage could again come under pressure.

Chapter 3: Business Deconstruction — Oracle's Cloud Transformation Overview

3.1 SaaS Business Resilience

3.1.1 Competitive Analysis of Fusion Cloud Applications

Oracle Fusion Cloud ERP is Oracle's core product in the enterprise SaaS market, directly competing with Salesforce and SAP. In FY2025, Fusion Cloud ERP revenue grew 22% YoY, and in Q2 FY2026, revenue reached $1.1B (+18% YoY).

Fusion vs. Competitors:

Sources of Fusion's Competitive Advantage:

1. Oracle-on-Oracle Stack: Fusion runs on OCI, with Oracle Autonomous Database as its underlying database. This full-stack integration offers performance optimization and security that competitors cannot replicate.

2. Backend Process Integration: Unlike Salesforce, which focuses on frontend CRM, Fusion covers the complete backend processes of ERP, HCM, SCM, and EPM. For large enterprises requiring a unified backend system, Fusion's integration capabilities are a key selling point.

3. Migration Stickiness: For enterprises already using Oracle E-Business Suite or PeopleSoft, the migration path to Fusion is the smoothest. Oracle reports that approximately 60-70% of its Fusion pipeline comes from existing Oracle customer upgrades.

Fusion's Competitive Disadvantages:

1. Frontend Capabilities: In the CRM and customer experience domains, Fusion significantly lags behind Salesforce
2. Partner Ecosystem: SAP's and Salesforce's ISV/SI ecosystems are much larger than Oracle's
3. Lagging AI Narrative: Although Oracle has embedded AI functionalities in Fusion, market perception is lower than that of Salesforce Einstein and SAP Joule

3.1.2 Customer Retention Rate and ARPU Trends

Oracle has not publicly disclosed specific figures for SaaS customer retention rates. However, these can be inferred from several indirect indicators:

Retention Rate Inference: Based on Fusion Cloud's sustained 22% revenue growth and management's emphasis on "limited churn," Fusion's Net Retention Rate (NRR) is estimated to be in the 110-120% range. NetSuite's NRR in the SMB market is estimated at 105-115%, slightly lower but still healthy.

ARPU Trends: Fusion's ARPU has continuously increased over the past 3 years, driven by:

• module cross-selling (from single ERP to HCM+SCM)
• AI feature add-on pricing
• price increases upon contract renewal (typically 3-5%/year)

3.1.3 NetSuite's Position in the SMB Market

NetSuite is Oracle's core product in the Small and Medium-sized Business (SMB) market, currently serving over 40,000 customers. In Q2 FY2026, NetSuite revenue was $1.0B, a 13% YoY increase.

NetSuite's Strategic Value:

1. Market Complementarity: Fusion targets large enterprises, while NetSuite targets SMBs and growth companies, covering the full spectrum of market sizes.
2. Uplift Channel: Growing customers can gradually migrate from NetSuite to Fusion, creating an internal customer upgrade path.
3. Stable Economic Model: NetSuite's subscription model provides highly predictable recurring revenue.

Risk of Slowing Growth: NetSuite's growth rate slowed from 18% in FY2025 to 13% in Q2 FY2026, which may reflect:

• cautious IT spending in the SMB market amidst macroeconomic uncertainty
• intensified competition: Intuit (via Mailchimp+QuickBooks integration), Sage (AI-first strategy), Xero (global expansion), among others, are accelerating cloud ERP penetration
• increased difficulty in acquiring large customers (>$10M ARR), who are more likely to choose Fusion over NetSuite
• base effect: incremental customer acquisition naturally decelerates on a base of 40,000+ customers

NetSuite vs. Competitor Positioning:

NetSuite remains the most comprehensive solution in the $10M-$1B enterprise revenue segment, but Sage Intacct's faster growth (~20%) in specialized financial management might encroach upon NetSuite's market share in specific vertical segments (e.g., professional services, non-profits).

3.1.4 SaaS Business's Support for Valuation

SaaS Business's Valuation Anchoring Effect: In Oracle's valuation, the SaaS business provides a "valuation floor." Even if OCI's cloud transformation completely fails, Oracle's SaaS business (Fusion + NetSuite + Industry Clouds) remains a highly recurring business with annual revenue of $15B+ and growth of 15-20%. Based on typical 8-12x revenue multiples in the SaaS industry, the SaaS business itself is valued at $120-180B.

Increasing SaaS Contribution: In FY2025, cloud and license revenue accounted for 85.77% of total revenue. Of this, SaaS revenue (Fusion + NetSuite + Industry Clouds) accounted for approximately 25-28% of total revenue, and IaaS revenue accounted for approximately 20-22%. SaaS+IaaS combined accounted for approximately 45-50%, with the remainder being license and support services (traditional business).

SaaS Revenue Trend Reconstruction (based on quarterly data estimation):

Key Observations: SaaS business shows relatively stable quarterly growth (0-5% QoQ), while IaaS is more volatile (7-24% QoQ). This validates the business structure hypothesis of SaaS as a "stable foundation" and IaaS as a "growth driver." However, it also implies that if IaaS growth slows, the SaaS business may not provide sufficient incremental growth to sustain Oracle's valuation premium.

Cross-selling Opportunities in Oracle's Full-Stack Ecosystem: Oracle's unique business value lies in its comprehensive technology stack, covering everything from databases → middleware → applications (ERP/HCM/SCM) → infrastructure (OCI). This full-stack coverage creates three cross-selling paths:

1. Database → OCI: Existing Oracle Database customers migrate to OCI (Oracle-on-Oracle), with typical contract increments of 30-50%
2. ERP → SaaS: Traditional E-Business Suite/PeopleSoft customers upgrade to Fusion Cloud, with an annualized value increase of 2-3x
3. SaaS → IaaS: Fusion/NetSuite customers add OCI infrastructure services (e.g., AI analytics, data lakes)
4. IaaS → Database: OCI AI customers require high-performance database support for inference and fine-tuning workloads

Management discloses that 60-70% of the Fusion pipeline comes from existing Oracle customers, demonstrating the effectiveness of the cross-selling paths. However, this also implies a ceiling issue: once existing customer upgrades are largely complete, incremental growth must come from net new customers, and Oracle's track record in "winning over" SAP/Salesforce customers is not outstanding.

3.2 $523B RPO Quality Assessment

3.2.1 RPO Contract Structure Analysis

The $523B RPO is a core data point for understanding Oracle's investment thesis. RPO represents the total amount of contracted revenue not yet recognized, serving as a "pipeline" indicator of future revenue.

RPO Composition Breakdown:

Contract Term Analysis: Management disclosed that approximately 33% of RPO ($173B) is expected to be recognized as revenue within the next 12 months, with 12-month RPO growing 40% year-over-year. This means 67% ($350B) consists of long-term contracts (>12 months), mostly 3-5 year contracts for AI infrastructure.

Cancellation Clause Risks: Large AI infrastructure contracts typically include:

• Capacity Commitment Clauses: Clients commit to a minimum consumption, but may have flexible upper limits
• Phased Milestones: Contracts are executed in phases, with the next phase triggered only after conditions for the current phase are met
• Conditional Cancellation Rights: If Oracle fails to deliver agreed-upon capacity on time, clients may have the right to reduce or cancel
• Price Adjustment Mechanisms: Long-term contracts may include price renegotiation clauses

3.2.2 RPO to Revenue Conversion Path

Conversion Funnel Analysis:

RPO → revenue recognition is not a linear process. Converting the $523B RPO into actual recognized revenue requires passing through multiple "leakage" filters:

1. Capacity Readiness Filter (Leakage rate 5-15%): Oracle needs to build sufficient data centers and GPU clusters to serve customers. Current CapEx investment levels of $21-50B/year indicate that capacity is still expanding. If capacity delivery is delayed, a portion of RPO will be postponed for recognition or even trigger client reduction rights. Especially for the 4.5GW data center capacity construction involved in the Stargate project – US grid infrastructure approval and power supply could become bottlenecks.
2. Client Deployment Filter (Leakage rate 5-10%): Even with capacity ready, clients' actual deployment and load growth require time. Cloud migration for enterprise clients typically takes 6-18 months, and AI training workload ramp-up also requires 3-6 months.
3. Consumption Rate Filter (Leakage rate 10-25%): Many cloud contracts include a minimum commitment, but actual consumption may be below the contract ceiling. Industry experience shows that the actual consumption rate for large cloud contracts is typically 70-85% of the total contract value.
4. Credit Risk Filter (Leakage rate 5-15%): The financial sustainability of major clients (especially OpenAI) directly impacts contract fulfillment. If the AI industry enters a period of capital tightening, some clients may seek to renegotiate or reduce commitments.
5. Competitive Churn Filter (Leakage rate 3-8%): During the execution of multi-year contracts, clients may allocate incremental workloads to competitors, thereby eroding Oracle's actual revenue.

Comprehensive Leakage Estimate: If the above leakage factors are combined (not entirely additive, as some risks are interrelated), the comprehensive leakage rate from RPO to final recognized revenue could be in the 20-40% range. That is: the final recognized revenue from $523B RPO could be in the range of $310-420B (cumulative over 5 years), averaging $62-84B annually.

12-Month RPO Conversion Efficiency: FY2025 actual revenue was $57.4B; if 12-month RPO is $173B, the apparent conversion rate is approximately 33% (57.4/173). However, it's important to note that the definition of 12-month RPO does not equate to "revenue to be recognized in the current year" – it represents "the portion of RPO expected to be recognized within the next 12 months," while total revenue also includes non-RPO sources (e.g., on-demand consumption, one-time licenses). A more accurate understanding is: of the $173B 12-month RPO, most will be recognized in the next four quarters, and when combined with non-RPO revenue, total FY2026 revenue could reach $65-72B.

3.2.3 Drivers of RPO Growth of 438% Breakdown

RPO grew from approximately $97B in Q2 FY2025 to $523B in Q2 FY2026, an increase of 438%. The drivers of this astonishing growth can be broken down as:

1. OpenAI Stargate Contract (~$300B): A single contract contributed approximately 70% of the RPO increment. This is not "diversified organic growth," but rather the signing of a single, large contract.

2. Meta, NVIDIA, and other AI clients (~$80-100B): Large client signings driven by AI infrastructure demand.

3. SaaS Business Organic Growth (~$20-30B): Renewals and new bookings for Fusion and NetSuite.

4. Multi-cloud Database Contracts (~$10-20B): New contracts driven by Database@AWS/Azure.

Key Insight: Of the 438% RPO growth, approximately 70-80% came from no more than 5 hyperscale customers. This is not "broad market demand-driven" but rather "concentrated bookings from a few large bets." These two narratives lead to vastly different valuation implications.

3.2.4 RPO Quality Risk Matrix

Specific Risks of OpenAI Dependency: OpenAI is an unprofitable company, consuming billions of dollars annually. Its $300B 5-year contract implies an average annual expenditure of $60B, far exceeding OpenAI's current revenue scale. The fulfillment of this contract is highly dependent on:

• OpenAI continuously securing sufficient funding (current valuation $150B+, but profitability path is unclear)
• Sustained growth in demand for large AI model training (efficiency breakthroughs such as the DeepSeek model could reduce demand)
• OpenAI not being fully integrated by Microsoft (Microsoft holds a 49% profit interest)
• No paradigm shift occurring in the AI industry (e.g., alternative solutions like specialized chips/quantum computing)

The "Iceberg Model" of RPO Quality:

Treating the $523B RPO as an iceberg:

• Above the surface (High-Quality RPO, approx. 30-35%): $173B recognized within 12 months, primarily consisting of contracts already in execution and SaaS renewals. The recognition probability for this portion of RPO is >90%.
• Near the surface (Medium-Quality RPO, approx. 20-25%): Approx. $100-130B recognized within 12-24 months, primarily consisting of signed AI contracts with clear capacity planning. Recognition probability 60-80%, depending on capacity delivery progress.
• Below the surface (Low Certainty RPO, approx. 40-50%): Approx. $210-250B recognized beyond 24 months, primarily the long-term portions of ultra-long-term contracts like Stargate. Recognition probability 30-60%, depending on customer financial capability, AI industry cycles, and technological path evolution.

If we calculate the "economic value" of RPO based on quality weighting:

• High Quality: $173B x 90% = $156B
• Medium Quality: $115B x 70% = $80B
• Low Certainty: $235B x 45% = $106B
• Total Quality-Weighted RPO: ~$342B (vs. stated value $523B, 35% discount)

RPO vs. Peers:

Oracle's RPO/revenue ratio of 9.1x is 3-7 times that of its peers. This either suggests that Oracle's future growth potential far exceeds its peers, or that the RPO includes a large number of long-term, low-certainty contracts. Based on the preceding quality analysis, the latter is more likely.