NVIDIA Corporation vs Oracle Corporation: Strategic Comparison

This in-depth comparison examines NVIDIA Corporation and Oracle Corporation across revenue, market value, business model, competitive positioning, and long-term growth strategy. Whether you are researching NVIDIA Corporation on its own, evaluating Oracle Corporation, or weighing the two companies side by side, the breakdown below highlights where each company leads and where the gap between NVIDIA Corporation and Oracle Corporation is widest.

On the headline numbers, NVIDIA Corporation reports annual revenue of $215.9B against $57.4B for Oracle Corporation, while their respective market capitalizations stand at $5.70T and $557.0B. NVIDIA Corporation is headquartered in United States and Oracle Corporation operates from United States, and those different home markets shape how each company competes.

NVIDIA Corporation: $215.9 billion in FY2026 revenue, $120.1 billion in net income, a 56% net margin. NVIDIA posted numbers in fiscal 2026 that no semiconductor company — and very few companies of any kind — had ever posted. The $5.7 trillion market capitalization, larger than the GDP of Germany, is not a speculation about future potential. It is a valuation attached to a company that has demonstrated the ability to convert AI infrastructure spending into earnings at margins that most software companies would envy. Jensen Huang founded NVIDIA in 1993 with Chris Malachowsky and Curtis Priem to build graphics processors for video games. The original business rationale was correct and profitable. But the architectural decision that defined NVIDIA's future was made in 2007, when Huang and his team released CUDA — a programming model that allowed NVIDIA's graphics processors to be programmed for general-purpose parallel computation. Graphics processors contained thousands of small processing cores designed to render visual information simultaneously. Those same cores, it turned out, were extraordinarily well-suited to the matrix multiplication operations that underlie machine learning. CUDA made that connection programmable. The AI training workloads that companies like Google, Meta, and Microsoft began running at scale in the 2010s required exactly the parallel processing architecture that NVIDIA had spent fifteen years refining. When the large language model era arrived after 2020, NVIDIA's H100 and then Blackwell GPU families were the only available hardware that could train and run models at the required scale with the required software support. Every major AI laboratory, cloud provider, and enterprise AI deployment runs on NVIDIA infrastructure — not because there is no alternative hardware, but because the CUDA software ecosystem, built over eighteen years, makes switching to any alternative hardware a multi-year software migration project. The Data Center segment generated the overwhelming majority of FY2026 revenue. Networking — NVLink, InfiniBand, and Ethernet fabrics that connect thousands of GPUs into training clusters — surged 263% year-over-year in Q4 FY2026 to $11 billion. NVIDIA has extended its revenue capture from the GPU itself to the complete data center fabric required to make clusters of GPUs function efficiently.

Oracle Corporation: That near-death moment produced the most durable enterprise software franchise in history. I find this genuinely surprising. Yet here it is, thriving — because enterprises don't choose infrastructure based on developer sentiment. They choose based on where their data already lives. The simplest way to understand how Oracle makes money: imagine you're a Fortune 500 bank. Your core ledger — the system that processes every transaction, every balance, every regulatory report — runs on Oracle Database. Twenty-seven years of stored procedures, custom integrations, compliance logic, and institutional knowledge are baked into that system. So you don't migrate. Now layer the rest on top. OCI is the exciting part. You just need to win the workloads that require specific performance characteristics. AI training on NVIDIA GPU superclusters? Oracle offers bare-metal access with lower latency than AWS. Database workloads that are already Oracle-native? OCI eliminates the rewrite. Strip out interest expense and the underlying operating economics are closer to 35-40% margins. Cloud and software combined now represent 88% of total revenue. What Oracle is really selling, if you step back, isn't software or cloud or databases. It's the cost of change. And every year, Oracle makes the migration path to its own cloud slightly easier than the migration path to anyone else's. Cloud and software combined represent 88% of total revenue. It's a tacit admission that Oracle can't win the broad cloud envelope, but it can own the data layer within someone else's infrastructure. Whether that's genius or capitulation depends on whether you think the database layer or the cloud platform captures more long-term value. In general-purpose cloud, this contest ended a decade ago. Oracle lost. But AI infrastructure reset the battlefield entirely. Oracle's bare-metal GPU clusters eliminate that overhead. When xAI and OpenAI need capacity and can't get it from their primary providers, they call Oracle. This isn't loyalty or brand preference — it's physics and availability. Both companies sell ERP, finance, supply chain, and HR software to the world's largest organizations. SAP has stronger European penetration and a more modern cloud-native architecture with S/4HANA. That double-migration cost keeps accounts locked for years. Snowflake and Databricks pull analytics workloads away from Oracle's data warehouse. PostgreSQL quietly becomes the default for every new application written by developers under 35. Salesforce owns CRM so completely that Oracle's CX suite barely registers in competitive conversations. Epic fights Cerner in healthcare with deeper clinical workflow expertise. Collectively, they represent a generational shift: new systems are being built without Oracle in the architecture. The honest competitive assessment is this — Oracle is unassailable where it already sits, genuinely competitive in AI infrastructure for as long as supply constraints hold, and largely invisible for net-new developer-led projects. The installed base generates cash. That's $25+ billion flowing in every year from customers who pay because leaving is more expensive than staying. Cloud Infrastructure alone grew north of 50%. Fusion ERP grew 14%, HCM and SCM both 15%. Larry Ellison, at 81, still drives the largest deals personally. They erode unless new workloads keep flowing in. That gap matters less for existing Oracle customers (who'll migrate to OCI regardless) and more for net-new workloads where Oracle has no historical relationship. The debt situation deserves honest acknowledgment. Oracle carries approximately $80-90 billion in long-term obligations — the accumulated cost of PeopleSoft, Sun, NetSuite, and Cerner. Interest expense eats into what would otherwise be spectacular margins. Cerner is the wildcard I'd watch most closely. Banks, hospitals, telecom operators, and government agencies have done the math. Most conclude it's cheaper to stay. It's strengthening because Oracle has finally built a credible cloud migration path. OCI's AI infrastructure play adds a new dimension entirely. Oracle doesn't need developers to love it. It needs enterprises with massive compute budgets to find its GPU clusters faster and cheaper than AWS's waitlist. OpenAI and xAI choosing OCI for training workloads validates this approach. New applications use cloud-native architectures. The gravitational pull only works on systems already in orbit. Java ownership (60 billion+ devices) and the Fusion/NetSuite application suite provide additional defensive layers, but the database franchise remains the core. If Oracle Database becomes optional for new enterprise systems — truly optional, not just theoretically replaceable — the entire economic model changes. That hasn't happened yet. Every stored procedure, every integration, every reporting tool, every compliance validation is built around Oracle's SQL dialect, PL/SQL, and data dictionary structures. Strip away the noise and Oracle has two bets that actually determine its trajectory, plus one long-shot that could become defining. The first bet is OCI as an AI infrastructure platform. This isn't a loyalty play — it's a capacity arbitrage that works as long as GPU demand exceeds supply. This is less glamorous but arguably more valuable long-term. Autonomous Database automates the maintenance that used to require expensive DBAs. Exadata Cloud Service gives performance-sensitive workloads a migration path that doesn't require compromise. The long-shot is healthcare. Then there's the variable nobody models: Larry Ellison is 81. That's not a succession plan. That's a single point of failure wearing a Hawaiian shirt. Bob Miner was the one who actually built the thing. The insight was genuine — IBM's researchers had published papers describing relational database theory and a query language called SQL, but IBM itself hadn't shipped a commercial product. Miner, a quiet mathematician with real engineering discipline, turned that blueprint into working code. Their first real contract came from a government project with a CIA connection — code-named Oracle. The name stuck. The product they shipped in 1979 was labeled Version 2. There was no Version 1. Ellison figured customers would be nervous buying a first release of essential database software, so he simply skipped the number. The early 1980s were a sprint. Relational databases moved from academic curiosity to enterprise necessity as companies realized they needed flexible data access, not just rigid file storage. Unlike IBM's database (which ran only on IBM hardware), Oracle worked across multiple systems. In an era when enterprises were beginning to diversify their computing environments, that flexibility was worth paying for. The 1986 NASDAQ IPO gave Oracle capital and credibility. Ellison was on magazine covers. Then it nearly died. By 1990, Oracle's aggressive sales culture had metastasized into something dangerous. Salespeople were booking revenue on deals that hadn't actually closed. Customers were being sold products that didn't yet exist. The accounting was, charitably, optimistic. In March 1990, Oracle announced it would miss earnings expectations. The stock dropped 31% in a single day. Ellison fired half the sales organization. Jeff Walker, the CFO, departed. Oracle's auditors forced a restatement. What saved Oracle was the database itself. Ellison rebuilt with discipline he hadn't previously shown. He hired Ray Lane as president in 1992 to professionalize sales operations. And he learned that Oracle's real power wasn't in closing new deals — it was in making existing customers unable to leave. The post-crisis Oracle was a different animal. The database franchise generated cash that funded expansion into enterprise applications, middleware, and eventually cloud infrastructure. Each acquisition followed the same logic: buy the customer relationship, then make it expensive to leave. The through-line from 1977 to today isn't technology. It's the commercial insight that data, once stored in a particular system, becomes extraordinarily difficult to move.

NVIDIA Corporation and Oracle Corporation pursue distinct approaches to generating revenue, and understanding how each company operates is the foundation of any fair comparison between NVIDIA Corporation and Oracle Corporation.

NVIDIA Corporation business model: Automotive (around 2%) sells DRIVE platforms for autonomous vehicles. Millions of developers, thousands of optimized libraries (cuDNN, TensorRT, NCCL, cuBLAS), every major framework pre-tuned — that's what sustains pricing power. Most organizations won't accept that risk while AI timelines feel existential. Revenue model: NVIDIA earns from Data Center GPUs and systems (~88% of FY2026 revenue), networking (InfiniBand, NVLink), gaming GPUs (GeForce), professional visualization (Quadro/RTX), automotive platforms (DRIVE), and software. The question isn't whether they'll succeed — they will, for some workloads — but whether they'll succeed broadly enough to dent NVIDIA's pricing power. When supply catches up to demand, the pricing dynamic shifts. The company has been methodically climbing the stack — from discrete accelerator cards to rack-scale systems to software subscriptions — and the financial results show it working. NVIDIA sells a proprietary software ecosystem that makes switching painful.

Oracle Corporation business model: Oracle Cloud Infrastructure (OCI) is emerging as a major AI cloud platform, winning workloads from hyperscalers by offering NVIDIA GPU clusters with lower latency and competitive pricing. You renew your license support contract every year. That's roughly $25 billion of Oracle's annual revenue right there — license support fees from customers who renew at rates above 90% because the alternative is operationally terrifying. The on-premise license business (about 8% of revenue) is declining but still throws off cash from customers buying new perpetual licenses. The transition from perpetual licenses to recurring subscriptions is essentially complete. Every year that a customer doesn't migrate away, Oracle's pricing power compounds. Revenue model: Oracle earns from Cloud Services (IaaS via OCI + SaaS via Fusion, NetSuite, Cerner — 55% of revenue, growing 44%), License Support (recurring maintenance — 25%), Cloud License and On-Premise License (8%), and Hardware/Services (12%). The number that should stop you cold: Oracle's license support revenue renews at 90%+ annually with essentially zero marginal cost. The second bet is converting the on-premise database installed base to cloud subscriptions. Every customer who moves from a perpetual license to a cloud subscription increases Oracle's revenue per account and makes the relationship stickier.

The durability of a company's moat often decides long-term winners. Here is how the competitive advantages of NVIDIA Corporation stack up against those of Oracle Corporation.

NVIDIA Corporation competitive advantage: Those are software-company margins on hardware-company scale. The revenue breakdown tells you where the gravity is. If that belief cracks — if AI capex pauses, if custom silicon matures, if four hyperscalers decide they're overpaying — the downside is severe. Competitive position: NVIDIA's advantage is the CUDA software ecosystem (millions of developers, thousands of libraries, all major AI frameworks optimized), full-stack AI platform (compute + networking + systems + software), 1-2 year architecture cadence (Hopper → Blackwell → Rubin), and the deployment confidence that makes customers willing to pay 73-75% gross margins to avoid migration risk during urgent AI buildouts. Meta's MTIA targets recommendation and inference at scale. AMD's best path is greenfield deployments where no legacy CUDA code exists, and those opportunities shrink as the ecosystem matures. Huawei's Ascend chips are already deploying at scale within China. They won't compete globally anytime soon — the software ecosystem is immature and geopolitics limits their market — but they could permanently lock NVIDIA out of the world's second-largest AI market. NVIDIA is operating in a different economic universe because it's selling a platform, not a component, and the platform has no close substitute at the scale customers need. Worse, the restrictions accelerate Chinese development of domestic alternatives — Huawei's Ascend chips are already being deployed at scale. If hyperscalers collectively decide they've overbuilt — or if model efficiency improvements reduce compute requirements faster than new applications create demand — NVIDIA's revenue could decline sharply. Switching costs aren't just financial — they're temporal. The networking layer compounds the advantage. It diversifies revenue away from four U.S. Hyperscalers, which matters because customer concentration is NVIDIA's most obvious vulnerability. These won't move the needle until physical AI applications reach the scale that language models hit in 2023. The options are interesting but unproven at scale. But the customer base is narrower than Cisco's was — four hyperscalers drive the majority of purchases — and each is building custom silicon to reduce dependence. Gross margins compress from 73-75% toward 65% by FY2029 as supply normalizes and custom chips absorb 20-30% of hyperscaler workloads. But Huang understood something that many brilliant engineers miss: being right about the math doesn't matter if you're wrong about the ecosystem. Every subsequent advance in neural networks — from ResNet to GPT to diffusion models — would be trained on NVIDIA hardware because the software ecosystem was already there.

Oracle Corporation competitive advantage: From Austin, Texas (relocated from Redwood City in 2020), Oracle grew from a database startup into one of the world's largest enterprise software companies through aggressive acquisitions (PeopleSoft, Siebel, Sun Microsystems, NetSuite, Cerner) and deep enterprise lock-in. Oracle bought the largest electronic health records platform in America and is attempting to modernize hospital IT infrastructure — a market where switching costs are even higher than in banking because patient safety is at stake. Competitive position: Oracle's advantage is enterprise data gravity (decades of business logic in Oracle databases that are prohibitively risky to migrate), switching costs, Fusion/NetSuite cloud applications, OCI's emerging AI infrastructure position, Java ownership, and 164,000 employees providing global enterprise coverage. AWS's virtualization layer adds latency that matters for large-scale model training. The advantage lasts exactly as long as GPU demand exceeds hyperscaler supply. No other enterprise software company has a comparable annuity stream at that scale. The advantage is strengthening in one dimension and weakening in another, and understanding both matters. Oracle's competitive moat in enterprise database and cloud infrastructure rests on a fact that most technology commentary ignores: the cost of migrating a essential Oracle Database deployment to an alternative is typically $50-200 million for a large enterprise, takes 3-5 years, and carries material execution risk. This creates switching costs that are measured in years of engineering effort, not months — effectively making Oracle Database installations permanent for the organizations that depend on them. Cloud Infrastructure revenue is growing 50%+ year-over-year because Oracle offers something the hyperscalers struggle with: bare-metal NVIDIA GPU access without virtualization overhead, at prices 20-30% below AWS equivalents. If demand for AI training infrastructure stays ahead of hyperscaler supply through 2028, Oracle locks in multi-year contracts with the companies building foundation models — and those contracts become the next generation of switching costs. Oracle rode that wave with ferocious sales energy and one genuine technical advantage — portability. The switching costs that would later become Oracle's greatest strategic asset were already operating in 1990 — they just hadn't been articulated as a business model yet.

Future prospects matter as much as current results. The growth strategies below explain how NVIDIA Corporation and Oracle Corporation each plan to expand from here.

NVIDIA Corporation growth strategy: It's that NVIDIA spent nearly two decades building a software platform nobody wanted, and then the world's most capital-intensive technology wave arrived and needed exactly that platform. NVIDIA designs the architecture, writes the software, builds the systems, and captures the margin. Strategic direction: Scaling Blackwell architecture, growing networking and inference revenue, expanding sovereign AI and enterprise AI software, and extending into robotics and autonomous vehicles. U.S. Export controls block NVIDIA's best chips from China, which simultaneously costs NVIDIA revenue and accelerates Chinese domestic alternatives. Here's my editorial judgment: NVIDIA's position is strongest during the build phase of AI infrastructure, when speed matters more than cost and nobody can afford to experiment with unproven alternatives. When AI workloads mature from strategic investment into operational expense, procurement teams will demand competitive bids. That's 3.5x growth in two years for a company that was already enormous. The valuation implies investors believe this growth continues for years. Customer concentration is the risk that keeps NVIDIA's investor relations team up at night — and it should. AI infrastructure spending has been growing at rates that look unsustainable by any historical semiconductor standard. Maintaining 40-70% growth means adding $85-150 billion in new revenue annually. CUDA has been accumulating developer investment since 2006. NVIDIA's growth story in 2026 comes down to one architectural bet: sell the entire AI factory, not just the GPU inside it. Training gets the headlines, but inference workloads are growing faster as models move into production. Governments from the UAE to India to Singapore are building national AI infrastructure on NVIDIA platforms. The honest assessment: NVIDIA has one massive bet (AI data center infrastructure keeps growing) and several options on the future. Cisco Systems was the world's most valuable company, selling the infrastructure layer of the internet buildout. Huang made the call to abandon the proprietary architecture entirely and rebuild around the triangle-based standard the market had chosen.

Oracle Corporation growth strategy: Not because Oracle lacks technical capability, but because the company spent two decades being openly hostile to the developer community that builds new systems. It's growing north of 50% annually because Oracle figured out something counterintuitive — you don't need to win the general cloud market to build a massive infrastructure business. Neither is growing, but both generate margin. The debt is the price Oracle paid to assemble this portfolio through force rather than organic growth. Strategic direction: Scaling OCI for AI workloads, migrating on-premise database customers to cloud, growing Fusion Applications, integrating Cerner into Oracle Health, expanding multi-cloud partnerships (Database@Azure/AWS), and deploying sovereign cloud regions. Oracle counters with Fusion growing at 14-15% and a database relationship that SAP simply cannot replicate — when your ERP runs on Oracle Database, migrating to SAP means migrating the database too. AI infrastructure generates growth. The growth acceleration is real and dramatic. That comparison illustrates both Oracle's momentum and its ceiling — it's growing fast for a 47-year-old company, but the market still sees it as a supporting actor in the AI story rather than a lead. The remaining performance obligation keeps expanding as enterprises sign multi-year cloud commitments. The installed base is enormous today, but installed bases don't grow themselves. As long as revenue grows 20%+, the leverage looks brilliant. If growth slows to single digits, that debt becomes a constraint on investment and buybacks simultaneously. Healthcare IT modernization is a decade-long project requiring clinical workflow expertise, regulatory patience, and trust-building with hospital systems that Oracle's traditionally aggressive sales culture isn't designed for. The multi-cloud partnerships are genuinely clever — they eliminate the binary choice that was pushing some customers toward PostgreSQL or AWS Aurora. It's weakening because every year, the percentage of global enterprise workloads that have never touched Oracle grows. New companies build on open-source databases. The 22% revenue growth in Q3 FY2026 suggests it isn't happening soon. Everything else — sovereign cloud regions, NetSuite mid-market expansion, Fusion Applications growth at 14-15% — is important but incremental. Everything depends on one variable: whether GPU supply constraints persist long enough for OCI to build permanent customer relationships before AWS and Azure catch up on capacity. Revenue hits $90-100 billion by FY2029, margins expand as cloud mix increases, and the 9.7x revenue multiple looks like a bargain. Growth reverts to the 5-8% that characterized the 2010s. The $80-90 billion debt load, comfortable at 22% growth, becomes a genuine constraint at 6% growth. Safra Catz runs operations with precision, but Oracle's largest sovereign cloud deals and AI partnerships still close because Ellison personally knows the decision-makers. It was a small lie that revealed a large truth about Oracle's DNA: perception management was always part of the strategy. Revenue was growing 100%+ annually. He focused engineering on database performance and reliability rather than feature sprawl.

A closer look at the financial trajectory of NVIDIA Corporation and Oracle Corporation rounds out the comparison.

NVIDIA Corporation: Revenue of $215.9 billion in FY2026, up 65% from $130.5 billion in FY2025 and from $44.9 billion in FY2023, represents one of the steepest revenue acceleration curves in the history of large-cap technology companies. Net income of $120.1 billion on that revenue base — a 55.6% net margin — reflects the pricing power available to a company whose products are scarce, urgently needed, and practically irreplaceable within any reasonable planning horizon for AI infrastructure buyers. The Data Center segment dominates, generating the vast majority of revenue. The H100 GPU at launch was sold for approximately $30,000 to $40,000 per unit, with hyperscalers purchasing them in quantities of tens of thousands. The Blackwell architecture, introduced in FY2025, commands higher prices per unit and higher revenues per rack, as NVLink GB200 systems integrate multiple GPUs and networking components into a single sales unit. The gross margin on Data Center hardware, sustained above 70%, is more typically associated with software businesses than with semiconductor manufacturing. The inventory risk that periodic semiconductor downturns create — the 2022-2023 gaming GPU correction, for example, led to a multi-quarter revenue decline in that segment — does not currently apply to Data Center at the same severity. Hyperscaler AI infrastructure spending is driven by competitive dynamics among Microsoft, Google, Amazon, and Meta that make voluntary reduction of GPU purchases strategically costly. Each company's AI capability relative to competitors depends on compute access, creating a demand floor that cyclical economic conditions affect less than they affect gaming or automotive semiconductor demand. Free cash flow at NVIDIA's current scale provides capital allocation flexibility that most companies never access. Share repurchases, R&D investment in future GPU generations, and potential acquisitions — though the failed Arm acquisition in 2022 demonstrated the regulatory constraints on defining M&A — all compete for a capital base that is growing faster than management's ability to deploy it productively.

Oracle Corporation: Today Oracle generates $57.4 billion in annual revenue, carries a $557 billion market cap, and is somehow experiencing its fastest growth since the dot-com era — Q3 FY2026 delivered 22% revenue growth and 44% cloud growth. Under CEO Safra Catz, Oracle reported $57.4B in FY2025 revenue and is experiencing its strongest growth in over 15 years — Q3 FY2026 delivered $17.2B revenue (up 22% YoY), with cloud revenue surging 44% to $8.9B. The company employs approximately 164,000 people and has a market cap of approximately $557B. Migrating away would cost $200 million and take four years, with meaningful risk of catastrophic failure during the transition. Cloud services account for approximately 55% of Oracle's $57.4 billion FY2025 revenue and are growing 44% year-over-year. The $28.3 billion Cerner acquisition in 2022 deserves separate attention. The net income picture tells you something important: $12.4 billion on $57.4 billion revenue is a 21.7% net margin, which sounds decent until you realize Oracle carries $80-90 billion in long-term debt from its acquisition spree. Oracle reported $57.4B in FY2025 revenue with $12.4B net income. Q3 FY2026 was 'exceptional': $17.2B revenue (up 22%), cloud $8.9B (up 44%), first quarter in 15+ years with 20%+ organic growth in both revenue and EPS. Market cap: ~$557B (NYSE: ORCL). None of these individually threatens Oracle's $57.4 billion revenue base. Whether Oracle in 2030 looks like a $100 billion revenue juggernaut or a $65 billion legacy franchise depends on which of those three dynamics dominates. FY2025 delivered $57.4 billion in total revenue and $12.4 billion in net income — a 21.7% net margin that looks modest until you account for the $80-90 billion debt load suppressing it. Q3 FY2026 produced $17.2 billion in revenue (up 22%), with cloud surging 44% to $8.9 billion. Management called it the first quarter in fifteen years where organic revenue and non-GAAP EPS both grew 20%+. Here's the tension: Oracle trades at roughly 9.7x trailing revenue ($557 billion market cap), which prices in sustained 20%+ growth for years. The stock added less market cap in four days than NVIDIA added in the same period ($591 billion for NVIDIA versus Oracle's entire valuation). Non-GAAP EPS hit $1.79 in Q3, up approximately 20% year-over-year. A botched Cerner integration wouldn't just waste $28.3 billion — it would validate every critic who says Oracle can't operate outside its database comfort zone. That calculation — repeated across 430,000+ customers globally — produces license support renewal rates above 90% and roughly $25 billion in annual recurring revenue that requires minimal incremental investment to maintain. The $28.3 billion Cerner acquisition gave Oracle the largest electronic health records platform in America, but turning that into a modern healthcare data platform requires patience, clinical expertise, and regulatory navigation that Oracle hasn't historically demonstrated. If it works, Oracle owns the data layer for an industry that spends $4.5 trillion annually in the US alone. The Cerner bet either validates or becomes a $28.3 billion lesson in overreach. Sun Microsystems in 2010 ($7.4 billion) brought Java and hardware. NetSuite in 2016 ($9.3 billion) added mid-market cloud ERP. Cerner in 2022 ($28.3 billion) pushed Oracle into healthcare. What began as three guys reading IBM research papers became a $557 billion company that employs 164,000 people and touches virtually every Fortune 500 data center on earth.