Source: NVIDIA Corporation Annual Review, Notice of Annual Meeting, Proxy Statement, and Form 10-K for the Fiscal Year Ended January 25, 2026.
- 1. Quick Facts / Company Snapshot
- 2. Company Overview
- 3. Business Segments
- 4. History and Evolution
- 5. Products and Services
- 6. Brand Portfolio
- 7. Geographical Presence
- 8. Profit and Loss
- 9. Cash Flow
- 10. Board of Directors and Leadership Team
- 11. Physical Properties
- 12. Founders
- 13. Investments and Capital Expenditure Plans
- 14. Shareholding Pattern
- 15. Future Strategy
- 16. Key Strengths
- 17. Key Challenges and Risks
Quick Facts / Company Snapshot
| Metric / Attribute | Disclosed Value / Specification |
| Official Company Name | NVIDIA Corporation |
| Ticker Symbol | NVDA |
| Primary Stock Exchange | The Nasdaq Global Select Market |
| Corporate Headquarters | 2788 San Tomas Expressway, Santa Clara, California 95051 |
| State of Incorporation | Delaware (originally incorporated in California in April 1993; reincorporated in Delaware in April 1998) |
| IRS Employer Identification Number | 94-3177549 |
| SEC Commission File Number | 0-23985 |
| Founder, President & Chief Executive Officer | Jen-Hsun Huang |
| Executive Vice President & Chief Financial Officer | Colette M. Kress |
| Lead Independent Director | Stephen C. Neal |
| Independent Auditor | PricewaterhouseCoopers LLP |
| Total Fiscal 2026 Consolidated Revenue | $215.9 billion |
| Fiscal 2026 GAAP Operating Income | $130.4 billion ($130,387 million) |
| Fiscal 2026 Non-GAAP Operating Income | $137.3 billion ($137,300 million) |
| Fiscal 2026 GAAP Net Income | $120,067 million ($120.067 billion) |
| Fiscal 2026 GAAP Diluted Earnings Per Share | $4.90 |
| Fiscal 2026 Gross Margin | 71.1% |
| Fiscal 2026 Operating Cash Flow | $103 billion |
| Total Capital Returned to Shareholders (Fiscal 2026) | $41.4 billion ($40.4 billion repurchases, $1.0 billion dividends) |
| Global Workforce / Total Headcount | Approximately 42,000 employees across 38 countries |
| Total Common Shares Outstanding (as of February 20, 2026) | 24.3 billion shares |
| Common Stock Par Value | $0.001 per share |
| Cumulative Research & Development Investment | Over $76.7 billion since inception |
Source: NVIDIA Corporation Form 10-K, Proxy Statement, and Annual Review for Fiscal 2026.
Company Overview
NVIDIA Corporation pioneered accelerated computing to tackle complex computational challenges that general-purpose computing can no longer address. As Moore’s Law slowed and Dennard scaling reached its physical limits, traditional general-purpose computing run on central processing units (CPUs) became insufficient for high-intensity processing. NVIDIA responded by reinventing the computing stack beneath modern computing, transitioning computing from pre-recorded software written by humans to software trained on vast datasets that generates intelligence in real time.
- Founded in 1993, NVIDIA transformed from a graphics-focused chip designer into a data-center-scale artificial intelligence (AI) infrastructure company.
- The company’s full-stack computing platform integrates hardware, systems, networking, software, algorithms, and libraries to serve global industries.
- More than half of NVIDIA’s engineering workforce is dedicated to software development, demonstrating a software-first approach to hardware acceleration.
THE 5-LAYER AI INDUSTRIAL STACK
┌─────────────────────────────────────────────────────────┐
│ 5. APPLICATIONS │
│ Chatbots, Robotics, Autonomous Vehicles, Biology │
├─────────────────────────────────────────────────────────┤
│ 4. MODELS │
│ LLMs, VLMs, VLAs, MoE, Reasoning & Domain Models │
├─────────────────────────────────────────────────────────┤
│ 3. INFRASTRUCTURE │
│ AI Factories, Data Centers, Liquid Cooling Systems │
├─────────────────────────────────────────────────────────┤
│ 2. CHIPS & SYSTEMS │
│ GPUs, CPUs, DPUs, LPUs, NVLink Switches, Racks │
├─────────────────────────────────────────────────────────┤
│ 1. ENERGY │
│ Power Generation, Transmission, Megawatt Management │
└─────────────────────────────────────────────────────────┘
The company operates under a foundational framework known as the “Five-Layer Cake” of artificial intelligence. Under this industrial model, AI is viewed not as an individual application or model, but as essential global infrastructure composed of:
- Energy
- Chips and systems
- Infrastructure
- Models
- Applications
In this hierarchy, every successful application draws upon each underlying layer down to the power generation source. Modern data centers are structured as “AI factories,” production facilities that systematically convert electrical energy into digital tokens—the elemental units of intelligence—much like industrial plants convert raw materials into physical goods.
NVIDIA’s core architecture uses an “extreme co-design” methodology. Rather than optimizing components in isolation, NVIDIA designs graphics processing units (GPUs), central processing units (CPUs), data processing units (DPUs), language processing units (LPUs), interconnect fabrics, optical networking, cooling systems, and software libraries in unison. This full-stack approach enables tens or hundreds of thousands of computing cores to function as a singular, unified data-center-scale supercomputer capable of distributed training and low-latency inference workloads.
- NVIDIA computing platforms power over 78% of the world’s TOP500 supercomputers, including 9 of the top 10 on the Green500 list.
- More than 7.5 million software developers worldwide build on the CUDA architecture across 6,000 supported acceleration applications.
- In Fiscal 2026, enterprise inference workloads surpassed training to become the dominant computing workload across global cloud networks.
The company’s business model relies on a unified underlying programmable architecture. By deploying the same core computing engine across multiple multi-billion-dollar end markets—Data Center, Gaming, Professional Visualization, and Automotive—NVIDIA maximizes research and development leverage. Developers write algorithms targeting the proprietary CUDA (Compute Unified Device Architecture) layer, creating an ecosystem moat that preserves software compatibility across past, present, and future hardware generations.
Business Segments
NVIDIA organizes and reports its business operations through two primary reportable segments: Compute & Networking and Graphics.
FISCAL 2026 CONSOLIDATED REVENUE
$215.9 Billion
┌─────────────────────────────────┬──────────┐
│ Compute & Networking │ Graphics │
│ $193.5B │ $22.5B │
│ (89.62%) │ (10.42%) │
└─────────────────────────────────┴──────────┘
(Percentages calculated by FirmsWorld; round to 100%)
Fiscal 2026 Segment Financial Performance
| Reportable Segment | Fiscal 2026 Revenue | % of Total Revenue | Revenue Growth (YoY) | Segment Operating Income | Segment Operating Margin | Operating Income Growth (YoY) |
| Compute & Networking | $193.5 billion | 89.62% (Calculated by FirmsWorld) | Up 67% | $130.1 billion | 67.24% (Calculated by FirmsWorld) | Up 57% |
| Graphics | $22.5 billion | 10.42% (Calculated by FirmsWorld) | Up 57% | $9.2 billion | 40.89% (Calculated by FirmsWorld) | Up 80% |
| Total Consolidated | $215.9 billion | 100.00% | Up 65% | $139.3 billion* | 64.52% (Calculated by FirmsWorld) | Up 58% |
*Note: Segment operating income total of $139.3 billion differs from consolidated GAAP operating income of $130.4 billion because certain corporate and administrative expenses are not allocated to individual segments for internal performance evaluations. Segment revenue numbers are rounded to the nearest tenth of a billion in company disclosures.
Source: NVIDIA Corporation Fiscal 2026 Form 10-K and Proxy Statement.
Compute & Networking Segment Profile
The Compute & Networking segment represents NVIDIA’s primary growth engine, generating $193.5 billion in Fiscal 2026 (89.62% of consolidated revenue). The operational scope of this division encompasses data center accelerated computing hardware, complete networking platforms, enterprise AI software stacks, and automotive autonomous driving modules.
- Data Center compute revenue grew 59% year over year, driven by the global commercial ramp of the Blackwell system architecture.
- Data Center networking revenue surged 142% year over year, powered by the adoption of the NVLink compute fabric alongside high-speed Ethernet and InfiniBand switches.
- Operating income for the segment reached $130.1 billion, yielding an operating margin of 67.24% (Calculated by FirmsWorld).
The segment provides foundational infrastructure to hyperscale cloud service providers (including Amazon Web Services, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure), AI model developers, independent cloud partners, and sovereign entities. Its scope includes the integration of GPUs, Grace CPUs, BlueField DPUs, and Quantum/Spectrum network switching. The operational architecture spans both physical silicon and enterprise software licenses, including the NVIDIA AI Enterprise platform, NVIDIA Omniverse, and NVIDIA DRIVE automated driving packages.
Graphics Segment Profile
The Graphics segment includes NVIDIA’s legacy and specialized computing products, generating $22.5 billion in Fiscal 2026 (10.42% of total revenue). This operating division oversees the design, engineering, and distribution of GeForce GPUs for gaming and consumer personal computers, alongside Quadro and NVIDIA RTX GPUs engineered for workstation professional visualization.
- Segment revenue expanded 57% year over year to $22.5 billion, up from the prior fiscal period.
- Operating profit rose 80% to $9.2 billion, demonstrating operating leverage with an operating margin of 40.89% (Calculated by FirmsWorld).
- Growth was driven by the commercial introduction of the GeForce RTX 50 Series based on the Blackwell architecture and expanded adoption of workstation-grade RTX systems.
Operationally, the Graphics segment supports creative, scientific, and engineering workflows. Leading independent software vendors (ISVs) configure computer-aided design (CAD), digital media creation, and simulation tools to run natively on RTX hardware. The segment also includes the GeForce NOW cloud gaming infrastructure and console system-on-chip (SoC) engineering agreements.
History and Evolution
NVIDIA’s corporate trajectory spans more than three decades of continuous computing development, marked by major architectural transitions.
1993 ── Founded in California by Jen-Hsun Huang
1998 ── Reincorporated in the State of Delaware
1999 ── Invented the GPU, launching the GeForce 256 and redefining 3D graphics
2006 ── Unveiled the CUDA programmable architecture, opening parallel computing to developers
2012 ── AlexNet neural network trained on NVIDIA GPUs won ImageNet, marking the "Big Bang" of AI
2017 ── Introduced the Tensor Core GPU architecture specifically optimized for deep learning workloads
2018 ── Launched first dedicated Autonomous Driving System-on-Chip (SoC) under the DRIVE platform
2020 ── Acquired Mellanox Technologies, adding end-to-end data center networking and introducing DPUs
2023 ── Launched the Grace CPU, establishing NVIDIA's first dedicated data center central processor
2024 ── Introduced the Blackwell rack-scale architecture connecting Grace CPUs and Blackwell GPUs
2026 ── Scaled Blackwell Ultra, unveiled Vera Rubin for agentic AI, and marked CUDA's 20th anniversary
NVIDIA was founded in April 1993 in California by Jen-Hsun Huang and reincorporated in Delaware in April 1998. In 1999, the company invented the Graphics Processing Unit (GPU), an innovation that spurred the growth of PC gaming and established real-time computer graphics as a viable commercial market.
In 2006, NVIDIA introduced CUDA (Compute Unified Device Architecture). CUDA transformed the GPU from a specialized graphics rendering chip into a programmable, general-purpose parallel processor. This allowed developers to harness thousands of mathematical cores simultaneously for scientific simulation, physics calculation, and numerical computing.
- The strategic inflection point occurred in 2012 when the AlexNet convolutional neural network, trained on NVIDIA GPUs, won the ImageNet competition.
- Recognizing deep learning as an emerging computing paradigm, NVIDIA directed its R&D toward AI-specific hardware acceleration.
- In 2017, the company unveiled its Tensor Core GPU architecture, integrating hardware acceleration designed for matrix mathematics.
In 2018, NVIDIA expanded into edge transportation computing by introducing its first autonomous driving SoC under the NVIDIA DRIVE banner. In 2020, NVIDIA completed its acquisition of Mellanox Technologies. This acquisition added end-to-end interconnect capabilities—spanning InfiniBand, high-performance Ethernet, and switches—and introduced the Data Processing Unit (DPU) to manage data center network processing.
In 2023, the company deployed its first dedicated data center CPU, the Arm-based NVIDIA Grace processor, engineered for large-scale AI training and high-performance computing. This enabled the release of the liquid-cooled NVIDIA Blackwell rack-scale architecture in 2024, combining 36 Grace CPUs and 72 Blackwell GPUs into a unified computer.
In Fiscal 2026, the company launched the Blackwell Ultra platform, unveiled the Vera Rubin platform for agentic workloads, celebrated the 20th anniversary of CUDA, and achieved $215.9 billion in revenue.
Products and Services
NVIDIA aligns its computing technologies across four core market platforms: Data Center, Gaming, Professional Visualization, and Automotive.
Fiscal 2026 Market Platform Revenue Breakdown
| Market Platform | Primary Segment | Fiscal 2026 Disclosed Revenue | % of Consolidated Revenue | Year-over-Year Growth Rate |
| Data Center | Compute & Networking | $193.7 billion | 89.72% (Calculated by FirmsWorld) | Up 68% |
| Gaming | Graphics | $16.0 billion | 7.41% (Calculated by FirmsWorld) | Up 41% |
| Professional Visualization | Graphics | $3.2 billion | 1.48% (Calculated by FirmsWorld) | Up 70% |
| Automotive | Compute & Networking | $2.3 billion | 1.07% (Calculated by FirmsWorld) | Up 39% |
| Total Platform Sum | Consolidated | $215.2 billion* | 99.68% (Calculated by FirmsWorld) | Up 65% |
*Note: Market platform totals sum to $215.2 billion as reported in business highlights, compared to $215.9 billion in total consolidated net revenue, reflecting category rounding in company disclosures.
Source: NVIDIA Corporation Fiscal 2026 Annual Review and Business Overview.
Supplementary Disclosed Product and Sub-Market Revenue Lines
| Specialized Product / Market Line | Disclosed Fiscal 2026 Financial Metric | Operational Significance |
| Data Center Networking Offerings | Revenue exceeded $31.0 billion | Encompasses NVLink switches, InfiniBand adapters, Spectrum-X Ethernet, and BlueField DPUs. |
| Sovereign AI Deployments | Revenue exceeded $30.0 billion | Tripled year-over-year as nations built national AI computing infrastructure. |
| Physical AI Systems | Generated $6.0 billion in revenue | Covers robotics foundation modules, industrial digital twins, and autonomous machines. |
| China-Licensed H20 Products | Generated $60 million in revenue | Shipments executed under specific U.S. government export control licensing approvals. |
Source: NVIDIA Corporation Fiscal 2026 Annual Review and Form 10-K.
Data Center Platform Profile
The Data Center platform generated $193.7 billion in Fiscal 2026, up 68% year over year. The platform delivers compute-intensive acceleration across cloud, hyperscale, on-premises, and edge enterprise data centers.
- Hardware systems integrate Blackwell GPUs, Hopper GPUs, Grace CPUs, and BlueField DPUs.
- The Grace Blackwell rack-scale architecture (GB200 NVL72) integrates 36 Grace CPUs and 72 Blackwell GPUs in a liquid-cooled chassis functioning as a single compute node.
- Inference benchmarks demonstrate that the GB300 NVL72 delivers up to 50x higher throughput per megawatt and up to 35x lower cost per token compared to the Hopper architecture.
Data center networking forms an integral pillar of this architecture. The NVLink compute fabric allows multiple processors to share memory space at high bandwidth. NVLink Fusion allows hyperscale operators to interface custom-designed ASICs and CPUs directly with the NVIDIA compute ecosystem. Scale-out networking is provided by Quantum InfiniBand switches and Spectrum-X Ethernet platforms.
The software ecosystem includes the NVIDIA CUDA development suite and the CUDA-X library portfolio. NVIDIA AI Enterprise serves as a commercial operating system for production-grade software deployments. Key modules include:
- NVIDIA NIM (Inference Microservices): Pre-packaged containers designed to accelerate token throughput using open and proprietary foundation models.
- NVIDIA NeMo: An end-to-end framework for curating training datasets, fine-tuning models, applying reinforcement learning, and deploying enterprise guardrails.
- AI Blueprints: Reference architectures and runnable workflows that allow businesses to build and orchestrate domain-adapted agentic pipelines.
- Dynamo: An inference optimization operating system designed to manage token scheduling, memory utilization, and energy consumption per watt.
Gaming Platform Profile
The Gaming market platform generated $16.0 billion in Fiscal 2026, marking a 41% increase over the previous year.
- Desktop and laptop computing are powered by the GeForce RTX family, including the Blackwell-based GeForce RTX 50 Series.
- Hardware features dedicated Ray Tracing Cores for real-time light simulation and Tensor Cores for on-device neural computation.
- Deep Learning Super Sampling (DLSS) utilizes generative neural networks to predict and construct intermediate pixels, multiplying rendering frame rates.
The division also operates the GeForce NOW cloud gaming platform, which streams PC gaming titles from NVIDIA data centers to consumer devices. The platform also designs custom SoCs and provides development services for handheld gaming consoles.
Professional Visualization Platform Profile
Professional Visualization delivered $3.2 billion in revenue in Fiscal 2026, reflecting a 70% year-over-year increase.
- Workstation-class hardware includes NVIDIA RTX PRO and Quadro GPUs.
- The platform targets enterprise applications across computer-aided engineering, 3D product prototyping, visual effects, and architectural visualization.
- The NVIDIA Omniverse platform provides a computing foundation for real-time 3D design collaboration and industrial digital twin simulations based on Universal Scene Description.
NVIDIA RTX PRO platforms are increasingly deployed to execute on-premises generative AI training and local fine-tuning within enterprise networks to address security and latency constraints.
Automotive Platform Profile
The Automotive platform generated $2.3 billion in revenue in Fiscal 2026, representing a 39% increase year over year.
- The DRIVE Hyperion architecture provides an end-to-end autonomous driving development and deployment ecosystem.
- DRIVE AGX provides in-vehicle centralized compute hardware running the safety-certified DRIVE OS operating system.
- The software suite includes perception, mapping, path planning, parking assistance, and intelligent in-cabin cockpit monitoring.
The platform creates a continuous development loop: data center supercomputers train autonomous driving neural networks, Omniverse simulates driving environments, and DRIVE AGX executes autonomous driving software in production vehicles. In Fiscal 2026, Mercedes-Benz deployed production vehicles equipped with NVIDIA DRIVE Level 4 autonomous systems, and Uber partnered with NVIDIA to deploy an L4-ready global mobility network.
Brand Portfolio
NVIDIA maintains a portfolio of specialized commercial and technological brands across hardware, enterprise software, and open AI architectures.
NVIDIA STRATEGIC BRAND ARCHITECTURE
Enterprise & Data Center Consumer & Graphics Open Models & Edge
───────────────────────── ─────────────────── ───────────────────
• Blackwell / Ultra • GeForce • Nemotron (Agentic)
• Vera Rubin / Ultra • GeForce RTX • Cosmos (Physical AI)
• Hopper • GeForce NOW • GROOT (Robotics)
• Grace (Data Center CPU) • RTX PRO / Quadro • Alpamayo (Auto AV)
• BlueField (DPU) • DLSS (Neural Graphics) • BioNeMo (Biology)
• NVLink / NVQLink • Omniverse • Earth-2 (Climate)
• Quantum InfiniBand • DRIVE / DRIVE AGX • Clara (Healthcare)
• Spectrum-X Ethernet • DRIVE Hyperion • CUDA / CUDA-X
Hardware Computing and Architecture Brands
- Blackwell & Blackwell Ultra: Data-center-scale infrastructure platforms designed for trillion-parameter AI training and real-time reasoning workloads.
- Vera Rubin & Vera Rubin Ultra: Next-generation computing platforms co-designed for agentic AI workloads, integrating Rubin GPUs, Vera CPUs, and Groq LPUs.
- Hopper: NVIDIA’s previous-generation data center GPU architecture.
- Grace: NVIDIA’s data center central processor, optimized for high-memory-bandwidth AI and scientific computing.
- BlueField: Data processing units (DPUs) that offload, accelerate, and isolate data center networking, storage, and security tasks from primary GPUs and CPUs.
- NVLink & NVQLink: Proprietary processor interconnects; NVLink coordinates multi-GPU communications, while NVQLink interfaces GPUs with quantum processing units (QPUs).
- Spectrum-X: High-performance Ethernet networking platforms designed to optimize multi-tenant AI workloads within standard enterprise cloud networks.
Software and Platform Brands
- CUDA & CUDA-X: The foundational parallel computing platform, runtime, compilers, and acceleration libraries supporting high-performance computing and machine learning.
- NVIDIA Omniverse: A simulation and digital twin operating platform used to model physical factories, grid systems, and robotic behaviors.
- NVIDIA AI Enterprise: A commercial software suite encompassing NIM, NeMo, and enterprise-grade support for production AI environments.
- GeForce & GeForce RTX: Consumer graphics brands encompassing gaming GPUs, gaming laptops, and graphics technologies like Deep Learning Super Sampling (DLSS).
- NVIDIA DRIVE: Hardware and software systems powering autonomous driving, fleet management, and smart vehicle cabin infrastructure.
Open Model and Physical AI Brands
- Nemotron: Open frontier reasoning and language models optimized for agentic tool use and complex problem solving.
- Cosmos: World foundation models designed to simulate physical dynamics, geometry, and motion for physical AI and robotics.
- GROOT: General-purpose foundation models developed for humanoid robotic simulation and physical task execution.
- Alpamayo: Open models designed for autonomous driving perception, path planning, and vehicle trajectory forecasting.
- BioNeMo: Generative AI models engineered for molecular structure modeling, chemistry analysis, and drug discovery.
- Earth-2: Climate and weather simulation platforms that forecast global meteorological systems.
- Clara: Healthcare computing frameworks dedicated to medical imaging analysis, genomics processing, and computational biology.
Geographical Presence
NVIDIA operates a global operational footprint spanning research and development, field sales, and manufacturing partnerships across 38 countries.
NVIDIA GLOBAL OPERATING FOOTPRINT
Corporate Headquarters Manufacturing Partners Key Regional Markets
────────────────────── ────────────────────── ────────────────────
• Santa Clara, CA • Taiwan (TSMC, Hon Hai) • North America (U.S.)
• ~42,000 Total Workforce • Arizona, U.S. (TSMC Fab) • Asia-Pacific
• 38 Operating Countries • Texas, U.S. (Foxconn/ • Europe
• 31,000 R&D Engineers Wistron Assembly) • Sovereign AI Nations
• 11,000 Sales/Operations • Packaging (Amkor, SPIL) (Japan, France, Sweden,
• Contract (Fabrinet) India, Israel)
Global Operational Footprint
- The company employs approximately 42,000 personnel across 38 countries.
- The research and development organization includes 31,000 engineers, representing roughly 74% of total company headcount (Calculated by FirmsWorld).
- Sales, operational, marketing, and general administrative personnel account for approximately 11,000 employees globally.
Manufacturing Footprint and Supply Chain Diversification
NVIDIA maintains a fabless manufacturing model, relying on third-party foundries and packaging contractors. While its historical supply chain has been concentrated in Asia, the company expanded domestic production in the United States and began expanding into Latin America during Fiscal 2026:
- Wafer Fabrication: Advanced wafer production is executed primarily by Taiwan Semiconductor Manufacturing Company (TSMC) in Taiwan, alongside domestic wafer manufacturing at TSMC’s Arizona fabrication facilities. Secondary semiconductor wafer production is contracted with Samsung Electronics in South Korea.
- Advanced Packaging: High-density chip-on-wafer-on-substrate (CoWoS) packaging is provided by TSMC, alongside packaging operations with Amkor Technology and Siliconware Precision Industries (SPIL).
- Memory Sourcing: High-bandwidth memory (HBM) and specialized DRAM components are sourced from SK Hynix, Micron Technology, and Samsung Electronics.
- System Assembly & Testing: Supercomputing rack systems and server nodes are assembled by Hon Hai Precision Industry (Foxconn), Wistron Corporation, and Fabrinet, including dedicated U.S. supercomputer assembly plants established in Texas with Foxconn and Wistron.
Sovereign AI and Regional Market Deployments
NVIDIA’s Sovereign AI business expanded significantly in Fiscal 2026, generating more than $30 billion in revenue—more than tripling year over year. Sovereign states are building domestic computing infrastructure to maintain control over their data, language, and culture. Key national projects utilizing NVIDIA platforms include initiatives in Japan, France, Sweden, India, and Israel.
Regulatory Geography and Export Control Developments
NVIDIA’s international sales are subject to trade restrictions imposed by the United States government.
- China and Hong Kong: Export controls established in October 2023 and expanded in April 2025 require export licenses for high-performance processors, including the A100, H100, A800, H800, L40S, RTX 4090, B200, and H20. NVIDIA generated approximately $60 million in revenue from licensed H20 shipments during Fiscal 2026, but the company remains largely foreclosed from China’s data center computing market. In February 2026, the U.S. government granted an export license for small shipments of H200 chips subject to domestic inspection and a 25% U.S. import tariff; no revenue was recognized under this program in Fiscal 2026.
- Country Groups D:1, D:4, and D:5: Export licensing requirements cover territories including Saudi Arabia, the United Arab Emirates, and Vietnam (excluding Israel), restricting exports of processors exceeding specified computing density and interconnect thresholds.
Profit and Loss
NVIDIA’s Fiscal 2026 financial results reflected growth across both operating segments, led by demand for Data Center infrastructure.
Consolidated Financial Performance Summary
| Financial Statement Metric | Fiscal 2026 Disclosed Value | Fiscal 2025 Disclosed Value | Year-over-Year Dollar Change | Year-over-Year Percentage Change |
| Total Net Revenue | $215,900 million ($215.9B) | $130,497 million (Approx. $130.5B)* | +$85,403 million (Calculated) | Up 65% |
| Gross Margin Percentage | 71.1% | 75.0% (Derived from -3.9 pts YoY) | -3.9 percentage points | Down 3.9 points |
| GAAP Operating Income | $130,387 million ($130.4B) | $81,453 million | +$48,934 million (Calculated) | Up 60% |
| GAAP Operating Margin | 60.39% (Calculated by FirmsWorld) | 62.42% (Calculated by FirmsWorld) | -2.03 percentage points (Calculated) | Down 2.03 points (Calculated) |
| Non-GAAP Operating Income | $137,300 million ($137.3B) | $86,789 million | +$50,511 million (Calculated) | Up 58.20% (Calculated) |
| Non-GAAP Operating Margin | 63.60% (Calculated by FirmsWorld) | 66.51% (Calculated by FirmsWorld) | -2.91 percentage points (Calculated) | Down 2.91 points (Calculated) |
| GAAP Net Income | $120,067 million ($120.067B) | $72,880 million | +$47,187 million (Calculated) | Up 64.75% (Calculated) |
| GAAP Net Margin | 55.61% (Calculated by FirmsWorld) | 55.85% (Calculated by FirmsWorld) | -0.24 percentage points (Calculated) | Down 0.24 points (Calculated) |
| Diluted Earnings Per Share (EPS) | $4.90 | $2.93 (Calculated by FirmsWorld) | +$1.97 (Calculated) | Up 67% |
*Note: Fiscal 2025 revenue baseline derived from disclosed $215.9B revenue representing a 65% increase over Fiscal 2025.
Source: NVIDIA Corporation Fiscal 2026 Form 10-K, Proxy Statement, and Annual Review.
Reconciliation of Non-GAAP Operating Income
GAAP Operating Income ($130,387M)
+ Stock-Based Compensation Expense ($6,386M)
+ Acquisition-Related and Other Costs ($471M)
+ Other Net Costs ($56M)
─────────────────────────────────────────────
= Non-GAAP Operating Income ($137,300M)
| Expense Reconciling Component | Fiscal 2026 Disclosed Amount | Fiscal 2025 Disclosed Amount | Business Purpose and Accounting Characterization |
| GAAP Operating Income | $130,387 million | $81,453 million | Consolidated operating result under U.S. GAAP. |
| Stock-Based Compensation Expense | $6,386 million | $4,737 million | Non-cash employee equity compensation under ASC Topic 718. |
| Acquisition-Related and Other Costs | $471 million | $602 million | Legal, transaction, and integration costs associated with M&A. |
| Other Adjustments | $56 million | $(3) million | Miscellaneous non-recurring operational adjustments. |
| Non-GAAP Operating Income | $137,300 million | $86,789 million | Non-GAAP benchmark used for executive incentive evaluation. |
Source: NVIDIA Corporation Fiscal 2026 Proxy Statement (CD&A Reconciliation).
Disclosed Operating Expense Items
- H20 Inventory Charge: In the first quarter of Fiscal 2026, NVIDIA recorded a $4.5 billion charge covering excess inventory and non-cancellable purchase obligations after the U.S. government imposed licensing restrictions on H20 chips.
- Independent Auditor Fees: Total fees billed by PricewaterhouseCoopers LLP for Fiscal 2026 reached $13,361,322, comprising $10,166,400 in audit fees, $1,716,820 in audit-related fees, $1,076,103 in tax compliance fees, and $401,999 in other services.
- Research and Development Commitment: NVIDIA has invested more than $76.7 billion cumulatively in research and development since company inception.
Cash Flow
Operating cash generation supported working capital needs, inventory expansion, and capital return programs during Fiscal 2026.
Disclosed Cash Flow and Capital Return Metrics
| Cash Flow & Capital Allocation Metric | Fiscal 2026 Reported Value | Multi-Year Comparison / Historical Scope |
| Cash Generated from Operating Activities | $103.0 billion | Supported working capital and non-cancellable foundry commitments. |
| Cash Returned via Share Repurchases | $40.4 billion | Total 3-year repurchases: $84.1B; total 5-year repurchases: $94.2B. |
| Cash Returned via Cash Dividends | $1.0 billion | Total 3-year dividends: $2.2B; total 5-year dividends: $3.0B. |
| Total Capital Returned to Shareholders | $41.4 billion | Total 3-year return: $86.3B; total 5-year return: $97.2B. |
Source: NVIDIA Corporation Fiscal 2026 Annual Review and Business Overview.
- The company generated $103 billion in operating cash flow during Fiscal 2026, an operational conversion rate equal to 79.0% of GAAP operating income (Calculated by FirmsWorld).
- Capital return to equity holders reached $41.4 billion, with $40.4 billion dedicated to open-market share buybacks and $1.0 billion paid as dividends.
- Over a five-year cumulative window, NVIDIA returned $97.2 billion to shareholders ($94.2 billion in share repurchases and $3.0 billion in dividends).
Board of Directors and Leadership Team
NVIDIA’s governance structure combines an independent Lead Director with three standing committees composed of independent directors.
Board of Directors Nominees
| Director Nominee | Age | Director Since | Independent Status | Board Committee Membership | Notable Professional Background & Credentials |
| Jen-Hsun Huang | 63 | 1993 | No (Management) | None | Founder, President, and CEO of NVIDIA. Appointed to PCAST (2026); member of National Academy of Engineering; recipient of Semiconductor Industry Association Robert N. Noyce Award. |
| Stephen C. Neal | 77 | 2019 | Yes | Nominating & Corporate Governance (Chair) | Lead Independent Director of NVIDIA. Chairman Emeritus and former CEO of Cooley LLP; former Chairman of the Board of Levi Strauss & Co. (2011–2021). |
| A. Brooke Seawell | 78 | 1997 | Yes | Audit Committee (Chair) | Venture Partner at New Enterprise Associates; former CFO of Synopsys; Audit Committee Financial Expert; director of Tenable Holdings. |
| Dawn Hudson | 68 | 2013 | Yes | Compensation Committee (Chair) | Former Chief Marketing Officer of the National Football League (NFL); former President and CEO of Pepsi-Cola North America; Audit Committee Financial Expert. |
| Tench Coxe | 68 | 1993 | Yes | Audit Committee, Compensation Committee | Former Managing Director of Sutter Hill Ventures (1989–2020); former director of Artisan Partners Asset Management (1995–2025). |
| John O. Dabiri | 46 | 2020 | Yes | Compensation Committee, Nominating & Corporate Governance | Centennial Professor of Aeronautics and Mechanical Engineering at Caltech; MacArthur Fellow; National Medal of Science recipient; former member of PCAST (2021–2025). |
| Harvey C. Jones | 73 | 1993 | Yes | Audit Committee, Nominating & Corporate Governance | Managing Partner of Square Wave Ventures; co-founder of Daisy Systems; former CEO and Executive Chairman of Synopsys; Audit Committee Financial Expert. |
| Melissa B. Lora | 63 | 2023 | Yes | Audit Committee, Nominating & Corporate Governance | Former President of Taco Bell International; former CFO of Taco Bell Corp.; Audit Committee Financial Expert; director of Conagra Brands. |
| Aarti Shah | 61 | 2020 | Yes | Audit Committee, Compensation Committee | Former Senior VP and Chief Information & Digital Officer at Eli Lilly and Company (27-year tenure); director of Sandoz Group AG. |
| Mark A. Stevens | 66 | 2008* | Yes | Compensation Committee, Nominating & Corporate Governance | Managing Partner of S-Cubed Capital; former Managing Partner of Sequoia Capital (1993–2011); trustee of the University of Southern California. |
*Note: Mark A. Stevens previously served on NVIDIA’s Board of Directors from 1993 to 2006 before rejoining in 2008. On May 7, 2026, the Board expanded its size from ten to eleven directors and appointed Suzanne Nora Johnson effective July 13, 2026 (expected to join the Audit Committee).
Source: NVIDIA Corporation Fiscal 2026 Proxy Statement.
Executive Leadership Team
JEN-HSUN HUANG
President and CEO
│
┌───────────────────────┼───────────────────────┐
│ │ │
COLETTE M. KRESS AJAY K. PURI DEBORA SHOQUIST
EVP & CFO EVP, Worldwide Field EVP, Operations
Operations
│
TIMOTHY S. TETER
EVP, General Counsel &
Secretary
- Jen-Hsun Huang (President and Chief Executive Officer): Co-founded NVIDIA in 1993 and has served as President, CEO, and Director since inception. He previously worked in engineering and core microprocessor design roles at LSI Logic and Advanced Micro Devices (AMD). He holds a B.S.E.E. from Oregon State University and an M.S.E.E. from Stanford University.
- Colette M. Kress (Executive Vice President and Chief Financial Officer): Joined NVIDIA in 2013. She previously served as CFO of the Business Technology and Operations Finance division at Cisco Systems (2010–2013), spent four years as CFO of Microsoft’s Server and Tools division, and worked eight years at Texas Instruments. She holds a B.S. in Finance from the University of Arizona and an M.B.A. from Southern Methodist University.
- Ajay K. Puri (Executive Vice President, Worldwide Field Operations): Joined NVIDIA in 2005 as Senior VP of Worldwide Sales and was promoted to Executive VP in 2009. He previously held marketing, sales, and management roles across a 22-year career at Sun Microsystems, alongside positions at Hewlett-Packard, Booz Allen Hamilton, and Texas Instruments. He holds a B.S.E.E. from the University of Minnesota, an M.S.E.E. from Caltech, and an M.B.A. from Harvard Business School.
- Debora Shoquist (Executive Vice President, Operations): Joined NVIDIA in 2007 as Senior VP of Operations and was promoted to Executive VP in 2009. She previously served as Executive VP of Operations at JDS Uniphase, Senior VP and General Manager at Coherent, and President of Quantum Corp.’s Personal Computer Hard Disk Drive Division, following early career roles at HP. She holds a B.S. in Electrical Engineering from Kansas State University and a B.S. in Biology from Santa Clara University.
- Timothy S. Teter (Executive Vice President, General Counsel and Secretary): Joined NVIDIA in 2017 as Senior VP and General Counsel and was appointed Executive VP in 2018. He previously spent over two decades as a partner at Cooley LLP litigating patent disputes, following work as a computer engineer at Lockheed Missiles and Space Company. He holds a B.S. in Mechanical Engineering from UC Davis and a J.D. from Stanford Law School.
Executive Compensation Structure
| Named Executive Officer (NEO) | Fiscal 2026 Base Salary | Stock Awards (ASC 718 Fair Value) | Non-Equity Variable Cash Incentive | All Other Disclosed Compensation | Fiscal 2026 Total Summary Compensation |
| Jen-Hsun Huang (CEO) | $1,497,627 | $24,800,511 | $6,000,000 | $4,045,691* | $36,343,830 |
| Colette M. Kress (CFO) | $898,577 | $12,825,872 | $600,000 | $16,402 | $14,340,850 |
| Ajay K. Puri (EVP) | $948,497 | $12,478,989 | $1,300,000 | $49,840 | $14,777,327 |
| Debora Shoquist (EVP) | $848,656 | $12,912,487 | $500,000 | $33,567 | $14,294,710 |
| Timothy S. Teter (EVP) | $848,656 | $12,912,487 | $500,000 | $15,240 | $14,276,382 |
*Note: CEO personal security costs ($3,975,533) constitute the primary element of Mr. Huang’s “All Other Compensation,” provided under a board-mandated executive security program, alongside a $11,500 401(k) match and $21,904 in life insurance premiums. Median employee total compensation was $282,050, resulting in a CEO-to-median employee pay ratio of 129:1.
Source: NVIDIA Corporation Fiscal 2026 Proxy Statement Summary Compensation Table.
Non-Employee Director Compensation (Fiscal 2026)
| Director Nominee | Fees Earned / Paid in Cash | Stock Awards (ASC 718 Fair Value) | Fiscal 2026 Total Compensation |
| Tench Coxe | $85,000 | $278,809 | $363,809 |
| John O. Dabiri | $85,000 | $278,809 | $363,809 |
| Dawn Hudson | $85,000 | $278,809 | $363,809 |
| Harvey C. Jones | $85,000 | $278,809 | $363,809 |
| Melissa B. Lora | $85,000 | $278,809 | $363,809 |
| Stephen C. Neal | $85,000 | $278,809 | $363,809 |
| A. Brooke Seawell | $85,000 | $278,809 | $363,809 |
| Aarti Shah | $85,000 | $278,809 | $363,809 |
| Mark A. Stevens | $85,000 | $278,809 | $363,809 |
*Note: Robert K. Burgess ($342,559 total; passed away December 2025), Persis S. Drell ($363,809 total; resigned January 2026), and Ellen Ochoa ($321,309 total; resigned July 2025) served during portions of Fiscal 2026. Non-employee director target equity consisted of 1,799 RSUs granted June 26, 2025.
Source: NVIDIA Corporation Fiscal 2026 Proxy Statement.
Physical Properties
NVIDIA conducts its administrative, engineering, and data center operations through a combination of owned and leased facilities globally.
- Headquarters Campus: Corporate headquarters are located at 2788 San Tomas Expressway, Santa Clara, California 95051. The campus houses executive management, core R&D teams, engineering laboratories, and corporate administrative functions.
- Global Office Footprint: The company maintains facilities across 38 countries to support its 42,000 employees, with regional hubs focused on semiconductor layout, software engineering, and partner support.
- Advanced Packaging and Subassembly Locations: Hardware packaging and component assembly are executed through contracted partner facilities located in Taiwan, South Korea, the United States, and Latin America.
- Domestic U.S. Production Infrastructure: Manufacturing partners include TSMC’s fabrication plants in Arizona, supercomputer assembly lines in Texas managed with Foxconn and Wistron, and packaging facilities operated with Amkor and SPIL.
Founders
NVIDIA was founded in April 1993 by Jen-Hsun Huang.
- Operational Tenure: Mr. Huang has served continuously as President, Chief Executive Officer, and a member of the Board of Directors since the company’s inception in 1993.
- Technical Background: Prior to founding NVIDIA, Mr. Huang served as Director of Coreware at LSI Logic, leading the company’s system-on-chip business unit. Before LSI Logic, he designed microprocessors at Advanced Micro Devices (AMD).
- Academic Credentials: He holds a Bachelor of Science in Electrical Engineering (B.S.E.E.) from Oregon State University and a Master of Science in Electrical Engineering (M.S.E.E.) from Stanford University.
- Industry Honors: Elected to the National Academy of Engineering; recipient of the Semiconductor Industry Association’s Robert N. Noyce Award, the IEEE Founder’s Medal, and the Dr. Morris Chang Exemplary Leadership Award; appointed in 2026 to the President’s Council of Advisors on Science and Technology (PCAST).
Investments and Capital Expenditure Plans
NVIDIA directs its capital allocations toward research, ecosystem expansion, and multi-year supply chain agreements.
- Cumulative R&D Investment: NVIDIA has invested more than $76.7 billion cumulatively in research and development since its founding in 1993, with $6.386 billion allocated to stock-based compensation in Fiscal 2026.
- Supply Chain and Capacity Commitments: To support production ramps for the Blackwell and Rubin architectures, NVIDIA enters into advance purchase orders, capacity reservations, and prepaid fabrication commitments with suppliers.
- Intellectual Property Licensing: NVIDIA entered into a non-exclusive technology licensing agreement with Groq, Inc., committing non-refundable payments to incorporate low-latency LPU inference acceleration into the Vera Rubin platform.
- Domestic Manufacturing Scaling: The company is investing in specialized production equipment, test systems, and qualification processes to scale manufacturing in Arizona and Texas.
Shareholding Pattern
NVIDIA’s common equity is held by institutional investors, company founders, executive officers, and retail shareholders.
Beneficial Ownership as of March 23, 2026
| Beneficial Owner / Investor Category | Disclosed Shares Owned Directly / Trusts | Shares Issuable Within 60 Days | Total Disclosed Shares Beneficially Owned | Disclosed % of Class Outstanding |
| BlackRock, Inc. | 1,805,935,550 | 0 | 1,805,935,550 | 7.43% |
| Vanguard Capital Management | 1,777,408,252 | 0 | 1,777,408,252 | 7.31% |
| Jen-Hsun Huang (President & CEO) | 870,604,104* | 0 | 870,604,104 | 3.58% |
| Mark A. Stevens (Director) | 34,068,672 | 900 | 34,069,572 | < 1.00% |
| Tench Coxe (Director) | 30,555,240 | 0 | 30,555,240 | < 1.00% |
| Harvey C. Jones (Director) | 7,002,787 | 900 | 7,003,687 | < 1.00% |
| Colette M. Kress (EVP & CFO) | 4,443,523 | 0 | 4,443,523 | < 1.00% |
| Ajay K. Puri (EVP) | 3,266,897 | 0 | 3,266,897 | < 1.00% |
| Timothy S. Teter (EVP & GC) | 2,687,197 | 0 | 2,687,197 | < 1.00% |
| A. Brooke Seawell (Director) | 2,505,707 | 900 | 2,506,607 | < 1.00% |
| Debora Shoquist (EVP) | 1,580,729 | 0 | 1,580,729 | < 1.00% |
| Dawn Hudson (Director) | 367,987 | 900 | 368,887 | < 1.00% |
| Stephen C. Neal (Lead Director) | 183,967 | 900 | 184,867 | < 1.00% |
| Aarti Shah (Director) | 25,620 | 0 | 25,620 | < 1.00% |
| John O. Dabiri (Director) | 13,888 | 900 | 14,788 | < 1.00% |
| Melissa B. Lora (Director) | 899 | 900 | 1,799 | < 1.00% |
| All Directors & Officers as a Group (14 persons) | 957,307,217 | 6,300 | 957,313,517 | 3.94% |
| Other Institutional & Public Shareholders | 19,771,484,491 (Calculated) | 0 | 19,771,484,491 (Calculated) | 81.32% (Calculated) |
| Total Disclosed Common Shares Outstanding | 24,312,141,810 | — | 24,312,141,810 | 100.00% |
*Note: Mr. Huang’s holdings include 528,531,547 shares held via the Jen-Hsun and Lori Huang Living Trust, 31,421,011 shares via the Huang 2012 Irrevocable Trust, 109,040,602 shares via the Huang Irrevocable Remainder Trust, 73,265,334 shares across TARG LLC entities, and 58,683,830 shares held by The Jen-Hsun & Lori Huang Foundation. Other public holdings calculated by deducting disclosed 5% holders and insider shares from total shares outstanding.
Source: NVIDIA Corporation Fiscal 2026 Proxy Statement.
Future Strategy
NVIDIA’s forward-looking business strategy focuses on shifting the computing industry toward accelerated computing architectures and managing multi-year hardware platform rollouts.
NVIDIA COMPUTING ROADMAP (2025–2028+)
2024–2025 2025–2026 2026–2027 2027+
───────── ───────── ───────── ─────
Blackwell Blackwell Ultra Vera Rubin Feynman
(Rack-scale GB200) (Enhanced inference) (Pod-scale Rubin/Vera) (Future architecture)
• Shipments in H2 FY27
• Groq LPU integration
• 10x lower token cost
Multi-Generation Hardware Cadence
NVIDIA has transitioned to an annual architecture development cycle for Data Center infrastructure.
- Blackwell to Blackwell Ultra: Moving from initial rack-scale GB200 systems to Blackwell Ultra systems, optimizing memory bandwidth and throughput for reasoning models.
- Vera Rubin and Vera Rubin Ultra: Slated to begin volume customer production shipments in the second half of fiscal year 2027. Co-designed across Rubin GPUs, Vera CPUs, Groq LPUs, BlueField DPUs, and optical NVLink fabrics, Rubin shifts the unit of computing from rack-scale to pod-scale, targeting up to a 10x reduction in token generation costs compared to Blackwell.
- Feynman Platform: The announced successor architecture to the Vera Rubin series, extending long-range infrastructure roadmaps.
- Multi-Year Revenue Visibility: NVIDIA has visibility into more than $1 trillion in cumulative Blackwell and Rubin platform revenue spanning from the start of 2025 through 2027.
Agentic and Physical AI Expansion
Management identifies agentic AI—systems capable of autonomous reasoning, planning, and multi-step tool execution—as the primary growth driver for inference computing. Reasoning models generate intermediate “thinking” tokens before issuing outputs, increasing compute demand per interaction.
Physical AI represents the company’s next commercial expansion area. By pairing Cosmos world foundation models and Omniverse physics-based digital twins with DRIVE AGX and Jetson embedded computers, NVIDIA is deploying autonomous models into robotics, manufacturing facilities, and autonomous vehicles.
Key Strengths
NVIDIA’s competitive advantages are supported by operating data, platform scale, and commercial adoption metrics.
- Extreme Full-Stack Co-Design: Rather than competing strictly at the semiconductor level, NVIDIA designs silicon, packaging, rack architecture, cooling, networking fabrics, and software concurrently. This systems approach delivers up to 50x higher throughput per megawatt and 35x lower cost per token on the GB300 NVL72 compared to Hopper.
- The 20-Year CUDA Ecosystem Moat: Launched in 2006, CUDA has expanded over two decades to include more than 7.5 million registered developers worldwide and native acceleration support for 6,000 scientific and commercial applications. Backward compatibility ensures software built on prior architectures runs immediately on Blackwell and Rubin.
- Data Center Computing Scale: Compute & Networking generated $193.5 billion in Fiscal 2026. Data Center networking alone generated over $31 billion, establishing end-to-end control over GPU interconnect fabrics via NVLink and InfiniBand.
- Supercomputing and Energy Efficiency: NVIDIA platforms power over 78% of the world’s TOP500 supercomputing systems and 9 of the top 10 energy-efficient installations on the Green500 list.
- Ecosystem Diversity: Deployment routes span hyperscalers (Azure, AWS, Google Cloud, OCI), specialized AI cloud providers (CoreWeave, Nebius, Nscale), frontier foundation model builders (OpenAI, Anthropic, xAI), enterprise platforms (SAP, ServiceNow, Siemens), and sovereign nations.
Key Challenges and Risks
NVIDIA faces regulatory, supply chain, and operational risks that could impact business performance and market access.
Export Controls and Foreclosure from the China Market
Trade restrictions imposed by the U.S. government have significantly curtailed NVIDIA’s operations in China.
- Licensing Mandates: In April 2025, the U.S. government instituted license requirements covering H20 products and equivalent memory bandwidth architectures destined for China, Hong Kong, and Macau.
- Inventory Write-Offs: Following these restrictions, NVIDIA recognized a $4.5 billion charge in Q1 Fiscal 2026 for excess H20 inventory and non-cancellable purchase obligations.
- Market Exclusion: The company generated only $60 million from licensed H20 sales in Fiscal 2026. NVIDIA disclosed that under current regulations and geopolitical conditions, it is effectively foreclosed from competing in China’s data center computing market. This exclusion has allowed domestic Chinese competitors to expand developer ecosystems.
- Tariffs and Inspections: In February 2026, the U.S. government granted a license for limited H200 shipments to China, but required domestic inspection that triggers a 25% U.S. import tariff; NVIDIA recognized zero revenue under this license in Fiscal 2026.
Official Site: NVIDIA Corporation Official Website
