Oracle Corporation (ORCL)

Executive Summary / Key Takeaways

• Oracle has reached an inflection point where its AI infrastructure business is creating a self-reinforcing flywheel across its entire technology stack, with Remaining Performance Obligations (RPO) surging to $552.6 billion—providing more than three years of revenue visibility and validating that demand is contractual.

• The company's integrated stack moat—combining AI infrastructure, the Oracle 23AI database, and AI-powered applications—creates switching costs and pricing power, as evidenced by 531% multicloud database growth and competitive wins against Workday (WDAY), SAP (SAP), and Salesforce (CRM).

• Oracle's aggressive $50 billion CapEx program for FY2026 represents a calculated bet that building gigawatt-scale AI data centers faster than competitors will create scale advantages, though this has transformed the balance sheet and introduced execution risk that will define the next 18 months.

• While AI infrastructure gross margins of 32% are dilutive to the corporate average today, the "halo effect" is accelerating applications growth (14% deferred revenue growth vs. 11% revenue growth) and is expected to drive margin expansion as construction expenses roll off and higher-margin database services scale.

• The central investment thesis hinges on two variables: whether Oracle can convert its RPO into revenue faster than its debt burden compounds, and whether its integrated AI strategy can defend against both hyperscaler commoditization and the risk of AI-native disruption to its high-margin database business.

Setting the Scene: From Database Utility to AI Infrastructure Powerhouse

Oracle Corporation, founded in 1977 and headquartered in Austin, Texas, spent four decades building the world's most ubiquitous database software before embarking on a transformation that most legacy technology companies have attempted and failed. For decades, Oracle's business model was straightforward: sell perpetual database licenses, collect high-margin support fees, and expand into adjacent enterprise applications. This created a cash-generating machine but also made the company vulnerable to cloud-native competitors who could deliver software-as-a-service without the burden of on-premise installations.

The strategic pivot began over a decade ago, but it has accelerated dramatically in the past three years. Oracle moved its applications business to the cloud first, establishing leadership in enterprise back-office SaaS including ERP, HCM, and supply chain management. Then it rebuilt its infrastructure business from scratch, creating Oracle Cloud Infrastructure (OCI) with an explicit focus on enterprise-grade performance and cost efficiency. This wasn't merely a defensive move to protect legacy revenue; it was an offensive bet that the company's deep understanding of enterprise workloads could create a better cloud for mission-critical applications.

Today, Oracle sits at the center of three converging industry trends that define the AI era. First, the explosion of artificial intelligence has created unprecedented demand for specialized data center capacity, with AI workloads projected to drive two-thirds of U.S. electricity load growth. Second, enterprises are demanding multi-cloud flexibility, refusing to be locked into single vendors as they did in the on-premise era. Third, AI is shifting from training to inference, where models reason on private enterprise data—a market Oracle's leadership believes will be even larger than training.

Oracle's position in this landscape is unique. Unlike Microsoft (MSFT), Amazon (AMZN), and Google (GOOGL)—who compete as general-purpose hyperscalers—Oracle has pursued a focused strategy of building the most efficient infrastructure for AI workloads while making its crown jewel database available across all clouds. This creates a differentiated value proposition: customers can train AI models on OCI's high-performance infrastructure, store and vectorize their data in Oracle 23AI, and deploy AI agents within integrated application suites that automate entire industry ecosystems. The question for investors is whether this integrated approach creates a durable competitive advantage or spreads Oracle too thin across too many battlefronts.

Technology, Products, and Strategic Differentiation: The Integrated Stack Moat

Oracle's core technological differentiation begins with its AI infrastructure, which management claims is faster and more cost-efficient at training AI models than other providers. This is a function of architectural choices. Oracle builds gigawatt-scale data centers with proprietary networking that moves data faster than competitors, enabling GPU superclusters that deliver high performance at a lower hourly cost. In Q3 FY26, Oracle delivered over 400 megawatts of capacity to customers, with 90% delivered on or ahead of schedule, and reduced time from rack delivery to revenue by a significant percentage in recent months.

The significance lies in the fact that in AI training, time is a primary cost driver. A model that trains faster costs less, creating a compelling value proposition that has driven AI infrastructure revenue up 243% year-over-year. More importantly, Oracle has secured over 10 gigawatts of power and data capacity coming online over the next three years, with over 90% fully funded through partners using innovative models like bring-your-own-hardware and upfront customer payments. This funding structure is critical—it means Oracle can expand capacity without negative free cash flow, preserving its investment-grade debt rating while building a scale advantage that will be difficult for smaller players to replicate.

The second pillar of differentiation is the Oracle 23AI database, which functions as an AI data platform that vectorizes all customer data for access by large language models while maintaining data privacy and security. Multicloud database revenue grew 531% year-over-year in Q3 FY26, a figure that reflects the strategic approach of Oracle's partnerships. By embedding OCI data centers within Microsoft Azure (33 regions live), Google Cloud (14 regions), and Amazon Web Services (expanding from 2 to 8 regions in Q3, targeting 22 by Q4), Oracle makes its database available wherever customers want to run AI workloads.

This implies that Oracle is playing a different game than the hyperscalers. While AWS, Azure, and GCP compete to host workloads, Oracle positions itself as the essential data layer that makes AI truly useful on enterprise data. The AI Data Platform, combined with Oracle's applications, allows customers to unite foundational models with proprietary data that largely resides in Oracle applications. This creates a powerful lock-in effect: once customers vectorize their data in Oracle 23AI and build AI agents on top, migrating becomes prohibitively expensive and complex. The 60-80% margins on multicloud database services will drive corporate margin expansion as this business scales, offsetting the current dilution from AI infrastructure buildout.

The third pillar is Oracle's AI-powered applications, where the company has delivered over 1,000 AI agents within existing horizontal back-office and industry applications. This is where the "halo effect" becomes tangible. Customers who train models on OCI naturally want to embed AI directly into their ERP, HCM, and SCM systems—and Oracle is the only vendor selling complete, integrated application suites that incorporate decades of industry experience and regulatory compliance. Cloud applications deferred revenue grew 14% in constant currency, outpacing the 11% revenue growth, signaling accelerating future growth.

The significance for the investment thesis is that Oracle is using AI to disrupt the SaaS market rather than being disrupted by it. While competitors like Salesforce lack ERP and OCI, and Workday and SAP lack the AI data platform, Oracle offers a unified solution. Memorial Hermann Health System, University of New South Wales, Gray Media, Investec Bank, and a major Wall Street bank all selected Fusion applications over Workday and SAP in Q3 FY26. These wins demonstrate Oracle's ability to automate entire industry ecosystems, from healthcare administration to financial services, using AI agents that reduce administrative overhead while improving outcomes.

Financial Performance & Segment Dynamics: RPO as the North Star

Oracle's Q3 FY26 financial results provide evidence that the integrated stack strategy is working. Total revenues increased 24% in reported currency to $17.19 billion, marking the first quarter in over 15 years where both organic total revenue and organic non-GAAP EPS grew at 20% or better. Cloud and software revenues, representing 86% of total revenues on a trailing four-quarter basis, grew 24% to $15.03 billion, with cloud revenues now accounting for 52% of total revenues, up from 44% in the prior year.

The segment breakdown reveals the strategic transformation underway. Cloud infrastructure revenue surged 84% reported (81% constant currency) to $4.89 billion, contributing 85% of constant currency cloud revenue growth in Q3. This is the engine pulling the entire company forward. AI infrastructure revenue within this segment grew 243% year-over-year, with gross margins holding at 32%—above the 30% guidance despite massive capacity buildout. While these margins are below Oracle's traditional software margins, they reflect the early-stage investment phase of a business that will generate 30-40% margins over the life of customer contracts.

Cloud applications revenue grew 13% reported (11% constant currency) to $4.03 billion, reaching an annualized run rate of $16.1 billion. The composition matters: Fusion ERP up 14%, Fusion SCM up 15%, Fusion HCM up 15%, and industry-specific SaaS solutions up 19%. These growth rates are being accelerated by the AI infrastructure halo effect. As Larry Ellison stated, performance advantages in GPU superclusters drive customers to OCI, which then makes it natural to adopt Oracle's applications that are pre-integrated with the AI data platform.

The hardware business, while small at $714 million (up 2% reported, -2% constant currency), remains strategically important. Growth in Oracle Exadata and strategic hardware offerings is partially offset by the continued emphasis on cloud-based infrastructure. This segment generates 68% margins and serves as an on-ramp for customers transitioning to cloud, particularly in regulated industries requiring dedicated hardware. The services business grew 12% reported (8% constant currency) to $1.44 billion, with margins expanding from 20% to 28% as consulting revenues increased. While lower-margin than software, this business accelerates customer time-to-value and drives cross-selling.

The most important financial metric, however, is Remaining Performance Obligations (RPO), which reached $552.6 billion as of February 28, 2026, up 325% year-over-year. This represents a transformational shift in business visibility. Approximately 12% of this RPO is expected to be recognized as revenue over the next twelve months, implying $66 billion in near-term revenue visibility. The vast majority of these bookings relate to opportunities with near-term capacity available, allowing for quick conversion of backlog to revenue. Safra Catz noted that the added RPO in Q2 FY26 alone will generate $4 billion of additional revenue in FY27.

RPO is a critical metric because it de-risks the massive CapEx program and validates that demand is contractual. Oracle has signed over $29 billion in contracts using new models where customers bring their own hardware or make upfront payments, funding expansion without negative cash flow. This is crucial for a company that announced plans to raise up to $50 billion in debt and equity financing for calendar year 2026, quickly securing $30 billion through investment-grade bonds and mandatory convertible preferred stock. The RPO growth enables Oracle to maintain its investment-grade rating while building capacity that competitors will struggle to match.

The balance sheet reflects this aggressive investment phase. Working capital increased significantly due to $42.7 billion in senior notes proceeds, $5 billion from mandatory convertible preferred stock, and $4.3 billion from the Ampere divestiture. Capital expenditures reached $39.2 billion for the first nine months of FY26, primarily for revenue-generating data center equipment. While this has created negative free cash flow of -$394 million TTM, the financing structure and customer prepayment models mitigate the risk.

Outlook, Management Guidance, and Execution Risk

Oracle's guidance reflects confidence in the AI-driven demand cycle. For Q4 FY26, management expects total revenue growth of 18-20% in constant currency, with cloud revenue growing 44-48% and non-GAAP EPS growing 15-17% to $1.92-$1.96. For FY27, revenue guidance was raised to $90 billion, up from the prior $89 billion target, reflecting the rapid monetization of RPO.

The CapEx guidance tells an equally aggressive story. FY26 CapEx is expected to be around $50 billion, up from $27.4 billion in FY25 and higher than the initial forecast of "over $25 billion" made in Q4 FY25. This increase is directly tied to RPO growth—Oracle only invests when it has contractual demand to fill. The company has secured over 10 gigawatts of capacity, with more than 90% fully funded through partners, and delivered over 400 megawatts in Q3 with 90% on-time performance. The Abilene, Texas supercluster is on track with over 96,000 NVIDIA (NVDA) Grace Blackwell GB200 units delivered.

This guidance suggests Oracle is in the early stages of a massive capacity buildout that will create a durable scale advantage. The risk is execution: delivering 10 gigawatts of capacity on schedule while maintaining 30-40% gross margins requires flawless supply chain management, particularly given the geopolitical dependency on Nvidia chips where 90% of production occurs in Taiwan.

Management's confidence extends to the applications business, where they expect continued acceleration. Cloud applications deferred revenue growing 14% while revenue grows 11% indicates a bookings strength that will drive future revenue recognition. The "One Oracle" sales reorganization is creating cross-selling synergies, with AI infrastructure deals naturally leading to applications conversations. Mike Sicilia's commentary on competitive wins—Memorial Hermann over Workday, multiple wins over SAP, and a major Wall Street bank replacing SAP entirely—demonstrates that the integrated stack is winning in the marketplace.

The TikTok USDS Joint Venture, where Oracle acquired a 15% equity stake and board seat in January 2026, represents both an opportunity and a risk. As a trusted security partner for TikTok's U.S. user data and algorithms, Oracle gains a high-profile validation of its sovereign cloud capabilities. However, the political and regulatory risks associated with TikTok could create volatility. The Ampere divestiture, generating $4.3 billion in cash and a $2.7 billion realized gain, shows Oracle's ability to monetize strategic investments while focusing on core AI infrastructure.

The key execution variables to monitor are: (1) the conversion rate of RPO to revenue, particularly the $66 billion expected in the next twelve months; (2) the gross margin trajectory for AI infrastructure as capacity matures; and (3) the pace of competitive wins in applications that validate the halo effect thesis.

Risks and Asymmetries: What Could Break the Thesis

The most material risk to Oracle's investment thesis is the debt-funded CapEx program and its impact on financial flexibility. Oracle announced plans to raise up to $50 billion in debt and equity financing for calendar year 2026, quickly securing $30 billion through investment-grade bonds and mandatory convertible preferred stock. Bears worry that debt could exceed $400 billion, causing prolonged negative cash flow. With debt-to-equity at 4.15x, Oracle is significantly more leveraged than Microsoft (0.32x), SAP (0.17x), or Salesforce (0.30x). While management emphasizes that many large AI contracts include customer prepayments or bring-your-own-hardware models to mitigate funding risks, the sheer scale of the investment creates binary outcomes. If demand materializes as projected, Oracle will have built an insurmountable lead. If demand slows or competitive dynamics shift, the fixed costs of 10 gigawatts of capacity could pressure margins for years.

Geopolitical dependency on Nvidia chips represents a critical vulnerability. With 90% of production in Taiwan, any escalation in U.S.-China tensions could disrupt supply chains and render Oracle's $50 billion CapEx investment less effective. While Oracle has diversified with AMD (AMD) MI355 capacity deliveries beginning in Q2 FY26, the AI infrastructure business remains heavily dependent on Nvidia's roadmap and manufacturing capacity. This risk is more acute for Oracle than for Microsoft or Google, who have deeper direct relationships with chip suppliers and more diversified AI hardware strategies.

AI disruption to core businesses poses a more subtle threat. Morningstar (MORN) notes that AI agents and native technologies could reduce revenues from Oracle's high-margin database and applications businesses. While management dismisses the "SaaS apocalypse" as applying to others but not Oracle, the risk is real. If AI agents can autonomously generate code and manage infrastructure, the value of Oracle's decades of embedded industry experience could diminish. The company's aggressive adoption of AI coding tools to restructure product development teams is both a defense and an acknowledgment of this risk.

The concentration risk around Larry Ellison, at 81 years old, cannot be ignored. His absence could trigger 15-20% volatility in the stock price, given his central role in strategy and customer relationships. While Safra Catz and the executive team are proven operators, Ellison's vision and credibility with large enterprise and government customers is a unique asset.

On the positive side, the asymmetry lies in the potential for the AI inferencing market to be even larger than training. Oracle's position as the only provider of a complete stack—from infrastructure to database to applications—positions it to capture this market. If the multicloud database business continues growing at 500%+ rates and achieves the projected 60-80% margins, it could drive corporate margins above 40% within three years, justifying a significant valuation re-rating.

Another asymmetry is the sovereign cloud opportunity. Oracle's Alloy model, which delivers sovereign data, operations, and contracting, is seeing increasing pipeline as governments worldwide require data localization. The TikTok USDS deal validates this capability, and if Oracle can replicate it with other global technology companies facing regulatory pressure, it could open a multi-billion dollar market.

Valuation Context: Pricing in Perfect Execution

At $153.71 per share, Oracle trades at 27.6x trailing earnings and 6.9x sales, with an enterprise value of $565.1 billion. This represents a premium to SAP (26.5x earnings, 5.1x sales) but a discount to Microsoft (24.6x earnings, 9.6x sales) on a price-to-sales basis despite faster growth. Oracle's operating margin of 32.7% sits between Microsoft's 47.1% and SAP's 29.2%, reflecting the current dilution from AI infrastructure investments.

The key valuation metrics that matter for this stage of Oracle's transformation are price-to-operating cash flow (18.8x) and enterprise value-to-revenue (8.8x). Oracle generates $445,000 in revenue per employee, 15% higher than industry average, demonstrating operational efficiency despite the heavy investment phase. The company's return on equity of 57.6% significantly exceeds Microsoft's 34.4% and SAP's 16.1%, reflecting the leverage in the business model as RPO converts to revenue.

The market is pricing in successful execution of the AI infrastructure buildout and margin expansion as capacity matures. The PEG ratio of 0.93 suggests the stock is trading at a reasonable multiple relative to near-term earnings growth, but this depends on maintaining 20%+ revenue growth while scaling margins. If Oracle can deliver on its FY27 $90 billion revenue target and achieve the projected 30-40% AI infrastructure margins, the current valuation will appear conservative. If execution falters or competitive pressure intensifies, the high debt load and negative free cash flow could lead to multiple compression.

Compared to hyperscalers, Oracle's 6.9x price-to-sales multiple is attractive relative to Microsoft's 9.6x, particularly given Oracle's 44% cloud growth rate versus Microsoft's 26%. However, Microsoft's superior operating margins (47% vs. 33%) and lower debt-to-equity (0.32x vs. 4.15x) reflect a more mature, less capital-intensive business model. The valuation gap suggests the market is demanding a discount for Oracle's execution risk, despite faster growth.

Conclusion: A Binary Bet on AI Infrastructure Leadership

Oracle's investment thesis boils down to a high-stakes proposition: the company is spending $50 billion to build AI infrastructure that will create a durable, high-margin competitive moat across its entire technology stack. The $552.6 billion in Remaining Performance Obligations provides revenue visibility and validates that this is not a speculative buildout but a response to contractual demand. The integrated stack—combining fastest-in-class AI infrastructure, the vectorized 23AI database, and AI-powered applications—creates switching costs that pure-play competitors cannot match.

The attractiveness of this story lies in the potential for a margin inflection as the AI infrastructure business matures. The 32% gross margins on delivered capacity are already above guidance, and management expects 30-40% margins over contract life. As construction expenses roll off and higher-margin multicloud database services scale, corporate margins could expand from the current 33% toward 40% or higher. The "halo effect" is already accelerating applications growth, with deferred revenue growing faster than recognized revenue, suggesting the integrated strategy is working.

The fragility of this story stems from the execution risk inherent in the massive CapEx program and debt-funded growth. With debt-to-equity at 4.15x and negative free cash flow of -$394 million TTM, Oracle has limited margin for error. Any slowdown in AI demand, supply chain disruption from geopolitical tensions in Taiwan, or competitive pricing pressure from hyperscalers could leave the company with underutilized capacity and a strained balance sheet. The concentration risk around Larry Ellison adds another layer of uncertainty.

The investment thesis will be decided by two variables over the next 18 months. First, the conversion rate of RPO to revenue—can Oracle recognize the $66 billion in near-term RPO while maintaining pricing and margins? Second, the competitive durability of the integrated stack—can Oracle's AI agents and multicloud database maintain differentiation as Microsoft, SAP, and others embed AI into their offerings? If Oracle executes flawlessly, the stock's 27.6x P/E and 6.9x sales will look cheap for a company growing cloud revenue at 44% with expanding margins. If execution falters, the debt burden and capital intensity could lead to significant downside. For investors, this is a high-conviction bet on management's ability to deliver what they've sold.