Oracle, founded in 1977 by Larry Ellison, built its empire on one product: the relational database — the software that stores and organizes the world’s most critical business data (bank accounts, airline bookings, hospital records, government systems). For four decades, Oracle’s database was the default choice for large enterprises, creating one of the stickiest customer bases in software history: once your company’s core data lives in an Oracle database, switching is expensive, risky, and rare.

Today’s Oracle is a company in the middle of a dramatic transformation. The old Oracle sold software licenses and support contracts. The new Oracle is betting tens of billions of dollars to become the fourth major cloud infrastructure provider — building massive AI data centers to rent computing power to companies like OpenAI, Meta, and xAI. In simple terms: Oracle used to sell the filing cabinets of the digital economy; now it’s also building and renting out the factories where AI is made.

The business now rests on three legs: (1) Cloud Infrastructure (OCI) — renting raw computing power, GPUs, and storage, growing explosively on AI demand; (2) Cloud Applications (SaaS) — Fusion ERP, NetSuite, and HR/finance software that businesses run on subscription; and (3) Legacy Software & Services — the traditional database licenses and support contracts, a slowly declining but highly profitable cash cow funding the cloud transition.

Why the mix matters: OCI is growing at 77–93% but currently carries lower gross margins than software (GPU rental is capital-intensive; database licenses are nearly pure profit). The legacy software business is shrinking ~2% per year as customers migrate to cloud — but it still throws off the profits that fund the AI buildout. Investors are essentially watching a race: can the high-growth, lower-margin cloud business scale fast enough before the high-margin legacy business erodes?

1. The AI Compute Land Grab. Training and running AI models requires staggering amounts of GPU computing power, and the companies building frontier AI (OpenAI, Anthropic, Meta, xAI) cannot build data centers fast enough themselves. They rent capacity from cloud providers — and Oracle has positioned itself as the aggressive challenger willing to build dedicated AI capacity faster than the incumbents. Oracle’s $300 billion, five-year cloud deal with OpenAI (announced September 2025), part of the “Stargate” supercomputer initiative, is the single largest cloud contract in history.

2. Cloud Market Growth & Concentration. Global cloud infrastructure spending grew 35% YoY to $129 billion in Q1 2026 alone (Synergy Research via Statista, May 2026). AWS leads at ~28%, Azure ~21%, Google Cloud ~14%. Oracle’s OCI is far smaller in share but is among the fastest-growing providers, alongside “neoclouds” like CoreWeave (Synergy Research, late 2025).

3. The Capex Arms Race. Hyperscalers collectively spent roughly $300 billion on data centers in 2025 (Amazon ~$105B, Microsoft ~$80B, Google ~$75B). Oracle, a much smaller company, spent $50 billion in capex in FY2026 and guided to $70 billion net capex in FY2027 ($90–95B gross including customer prepayments) — proportionally the most aggressive AI infrastructure bet of any major company relative to its size.

4. Enterprise SaaS Maturation. The applications side (ERP, HCM) is a steadier, slower battle against SAP, Workday, and Salesforce. Oracle Fusion and NetSuite grow ~10–14% — respectable but not the story driving the stock. NetSuite Cloud ERP grew 14% in Q3 FY2026.

Oracle competes on two distinct battlefields with different rivals on each: cloud infrastructure (against the hyperscalers) and enterprise applications (against SaaS leaders).

Profitability metrics (June 2026, StockAnalysis.com): ROE ~58% (inflated by high leverage), ROIC ~12%, current ratio 1.35, debt/equity ~4.2x. Beta of 1.65 — Oracle now trades with significantly more volatility than the broader market, a marked change from its historic “stable value tech” profile.

Oracle’s valuation picture in June 2026 is unusual: the stock has fallen roughly 43% from its September 2025 peak (reached after the OpenAI deal announcement), even as revenue growth accelerated to multi-decade highs. The market is wrestling with how to price a company that is simultaneously a slow legacy software firm, a hyper-growth AI infrastructure player, and a heavily leveraged construction project.

Near-term (FY2027):

The $90 billion revenue guidance — implying ~34% growth, roughly double FY2026’s pace — rests on converting backlog into live revenue as new data centers come online. OCI growth was still accelerating exiting FY2026 (93% in Q4 vs 77% full-year). The OpenAI/Stargate capacity begins generating major revenue in 2027 as facilities are commissioned.

Medium-term (FY2027–2030):

The $638B RPO converts over multiple years — Oracle effectively has revenue visibility most companies can only dream of, provided customers pay. Multicloud (Oracle Database@AWS/Azure/Google) opens the enormous installed database base to cloud migration without forcing customers to leave their preferred cloud. Oracle Health (Cerner) plus AI-driven healthcare records modernization remains a long-term option, and Oracle won a June 2026 contract to provide HR software across US federal agencies. Its 15% stake in TikTok US (~$2B) and role as TikTok’s US cloud provider add a strategic asset.

Long-term:

If AI inference (running models in production, not just training them) grows as the industry expects, demand for compute becomes more recurring and less lumpy — favoring providers with locked-in enterprise relationships. Oracle’s bet is that owning both the data layer (databases) and the compute layer (OCI) makes it the natural home for enterprise AI workloads.