Business Overview

NVIDIA Corporation, incorporated in California in April 1993 and reincorporated in Delaware in April 1998, is headquartered at 2788 San Tomas Expressway in Santa Clara, California, where it owns and leases approximately 3,000,000 square feet of office and building space. The company trades on The Nasdaq Global Select Market under the symbol NVDA, with public trading of its common stock having begun on January 22, 1999. At the helm is co-founder Jen-Hsun Huang, age 63, who has served as President and Chief Executive Officer since the company's inception in 1993, providing exceptional continuity of strategic vision across more than three decades. Colette M. Kress, age 58, serves as Executive Vice President and Chief Financial Officer. As of the end of FY2026, NVIDIA employed approximately 42,000 people across 38 countries, with 31,000 engaged in research and development and 11,000 in sales, marketing, operations, and administrative positions — a workforce composition that reflects the company's deep commitment to technical innovation as its primary competitive foundation. Approximately 6,000 of those employees are based in Israel, primarily supporting networking research and development, operations, and sales and marketing activities.

NVIDIA operates through two reportable segments: Compute & Networking and Graphics. The Compute & Networking segment encompasses Data Center accelerated computing and networking platforms, AI solutions and software, and Automotive platforms including autonomous and electric vehicle solutions and related software. The Graphics segment includes GeForce GPUs for gaming and PCs, as well as Quadro/NVIDIA RTX GPUs for enterprise workstation graphics. The company's financial trajectory across recent fiscal years illustrates the extraordinary demand acceleration driven by AI infrastructure investment. Total revenue grew from $60,922 million in FY2024 to $130,497 million in FY2025, and further to $215,938 million in FY2026. The Compute & Networking segment is the dominant driver of this scale: in FY2026 it generated revenue of $193,479 million and operating income of $130,141 million, compared with $116,193 million in revenue and $82,875 million in operating income in FY2025. The Graphics segment also expanded meaningfully, with revenue rising from $14,304 million in FY2025 to $22,459 million in FY2026, and operating income growing from $5,085 million to $9,156 million over the same period. Quarterly revenue progression throughout FY2026 underscores the sustained momentum: $44,062 million in Q1 FY2026, $46,743 million in Q2, $57,006 million in Q3, and $68,127 million in Q4, reflecting continuous sequential growth that accelerated meaningfully in the back half of the fiscal year.

• Governance · ceo Jen-Hsun Huang, age 63, President and Chief Executive Officer; co-founded NVIDIA in 1993 and has served as President, CEO and a member of the Board since inception.
• Governance · cfo Colette M. Kress, age 58, Executive Vice President and Chief Financial Officer.
• Company · headquarters 2788 San Tomas Expressway, Santa Clara, CA 95051; the company owns and leases approximately 3 million square feet of office and building space for its corporate headquarters in Santa Clara, California.
• Company · founded Headquartered in Santa Clara, California, NVIDIA was incorporated in California in April 1993 and reincorporated in Delaware in April 1998.
• Company · employees Approximately 42,000 employees in 38 countries at end of FY2026; 31,000 were engaged in research and development and 11,000 were engaged in sales, marketing, operations, and administrative positions. Approximately 6,000 employees are in Israel, primarily supporting networking R&D, operations, and sales/marketing. In FY2026 the turnover rate was 3.7 percent. More than 80% of the workforce hold technical roles and more than half hold an advanced degree; over 40% of FY2026 new hires came from employee referrals.
• Company · primary listing NVIDIA common stock ($0.001 par value) trades on The Nasdaq Global Select Market under the symbol NVDA. Public trading of the common stock began on January 22, 1999. SEC Central Index Key (CIK): 0001045810; State of incorporation: Delaware; IRS Employer Identification No. 94-3177549.
• Segments · reportable definition NVIDIA has two operating segments which are also its two reportable segments: "Compute & Networking" and "Graphics." The Compute & Networking segment includes Data Center accelerated computing and networking platforms and AI solutions and software, and Automotive platforms and autonomous and electric vehicle solutions including software. The Graphics segment includes GeForce GPUs for gaming and PCs, and Quadro/NVIDIA RTX GPUs for enterprise workstation graphics.

Product Breakdown

NVIDIA's Data Center platform generated $193,737 million in revenue in FY2026, advancing 68.2% year over year — itself a step-down from the exceptional 142.4% growth recorded in FY2025, yet still an expansion rate that few businesses of this scale have sustained. The platform's absolute dollar magnitude now dominates NVIDIA's consolidated revenue mix by an overwhelming margin, a structural shift driven by the simultaneous ramp of hyperscale cloud buildouts, enterprise AI adoption, and sovereign AI infrastructure programs across multiple geographies. Each successive wave of large-language-model training and inference deployment has pulled forward demand for NVIDIA's accelerated computing platforms, and the persistence of that demand across two consecutive fiscal years at triple- and then double-digit growth rates signals that the AI infrastructure cycle is supply-constrained rather than demand-constrained — a rare and favorable dynamic for a platform incumbent.

The four platforms outside Data Center each delivered growth in FY2026 while operating at a materially smaller scale. Gaming reached $16,042 million, up 41.3% year over year, as the PC installed base continues to refresh toward current-generation GeForce architectures and the AI-accelerated feature set — most notably DLSS — raises the value proposition of discrete GPU upgrades. Professional Visualization contributed $3,191 million, growing 69.9%, nearly matching the Data Center growth rate, as RTX-equipped workstations capture engineering, design, and content-creation workflows increasingly reliant on GPU-accelerated rendering and on-device generative-AI tools. Automotive recorded $2,349 million, up 38.7%, reflecting expanding deployments of NVIDIA's DRIVE platform across automakers and robotaxi developers building toward production-scale autonomous and intelligent-vehicle systems. OEM & Other, the smallest platform at $619 million, grew 59.1%, rounding out a picture in which every end-market reported meaningful acceleration.

NVIDIA reports two operating and reportable segments: Compute & Networking and Graphics. As defined by the company, Compute & Networking encompasses Data Center accelerated computing, networking platforms, AI solutions and software, and Automotive platforms including autonomous and electric-vehicle software; Graphics covers GeForce GPUs for gaming and Quadro/NVIDIA RTX GPUs for enterprise workstation graphics. In FY2026, the Compute & Networking segment produced $193,479 million in revenue — up 66.5% year over year — and $130,141 million in segment operating income, a level of absolute profitability that places it among the most lucrative technology franchises ever reported. The Graphics segment contributed $22,459 million in revenue, up 57%, with segment operating income of $9,156 million. In dollar terms, Compute & Networking's operating income is more than fourteen times that of Graphics, a disparity that captures how thoroughly the AI infrastructure cycle has repositioned NVIDIA's profit center over the past two fiscal years.

Looking forward, the platform and segment mix points toward continued deepening of the Compute & Networking franchise as the primary driver of both revenue and profitability. NVIDIA's product roadmap — spanning successive GPU generations, NVLink-based scale-up interconnects, high-speed Ethernet and InfiniBand networking, and a compounding software stack of CUDA libraries, NIM inference microservices, and AI Enterprise software — is designed to capture not only training workloads but an expanding share of inference deployment, which analysts broadly expect to be the faster-growing component of AI infrastructure spending in coming years. Within Gaming and Professional Visualization, the convergence of real-time ray tracing, AI upscaling, and generative-AI creative tools extends addressable demand qualitatively beyond the historical GPU refresh cycle, supporting the strong growth rates those platforms posted in FY2026. Automotive, with its multi-year production vehicle design cycles, offers the longest-duration visible growth vector, as DRIVE platform content per vehicle and the total fleet under contract both expand. The overall trajectory suggests that Data Center will remain the dominant revenue and profit contributor by a substantial margin, while the surrounding platforms grow at rates that should preserve their collective strategic relevance and may provide positive surprises as AI capabilities permeate additional end markets.

• Segments · reportable definition NVIDIA has two operating segments which are also its two reportable segments: "Compute & Networking" and "Graphics." The Compute & Networking segment includes Data Center accelerated computing and networking platforms and AI solutions and software, and Automotive platforms and autonomous and electric vehicle solutions including software. The Graphics segment includes GeForce GPUs for gaming and PCs, and Quadro/NVIDIA RTX GPUs for enterprise workstation graphics.

Competitive Position

NVDA operates in an intensely competitive market characterized by rapid technological change and evolving industry standards, with the company on record expecting this environment to become more challenging over time. The principal competitive factors in this market are performance, breadth of product offerings, access to customers and partners and distribution channels, software support, conformity to industry standard APIs, manufacturing capabilities, processor pricing, and total system costs — a multi-dimensional set of criteria that demands sustained excellence across hardware, software, and go-to-market execution simultaneously. Current competitors in the AI compute arena — the most strategically important battleground — include suppliers and licensors of hardware and software for discrete and integrated GPUs, custom chips, and other accelerated computing solutions, specifically Advanced Micro Devices, Inc. (AMD), Huawei Technologies Co. Ltd. (Huawei), and Intel Corporation (Intel). Competition extends further to large cloud service providers with internal teams designing hardware and software that incorporate accelerated or AI computing functionality, a group that includes Alibaba Group, Alphabet Inc., Amazon, Baidu, Huawei, and Microsoft Corporation. ARM-based CPU suppliers such as Amazon, Huawei, and Microsoft represent an additional competitive front, and across networking and interconnect — a critical complement to raw compute — rivals include Arista Networks, Broadcom, Cisco Systems, Hewlett Packard Enterprise, Lumentum, and Marvell Technology, among others. The company has acknowledged that some of these competitors may possess greater marketing, financial, distribution, and manufacturing resources, and it has explicitly flagged the risk that new entrants or competitor alliances could emerge with lower-priced products or superior feature sets capable of capturing significant market share.

NVDA's principal defense against this broad competitive pressure is a compounding moat built through decades of sustained investment in hardware, software, and ecosystem development. The company has invested over $76.7 billion in research and development since its inception, and as of the end of fiscal year 2026 employed approximately 42,000 people across 38 countries, of whom 31,000 were engaged in research and development — more than 80 percent of the overall workforce hold technical roles and more than half hold an advanced degree. Crucially, more than half of the company's engineers work on software, a deliberate organizational orientation that reflects NVDA's identity as a full-stack computing platform company rather than a chip vendor. The CUDA development platform, introduced in 2006, is the foundation of this stack: it opened the parallel processing capabilities of the GPU to a broad range of compute-intensive applications and has since attracted over 7.5 million developers worldwide using CUDA and the company's other software tools. That installed developer base — supported by hundreds of domain-specific libraries, frameworks, SDKs, and APIs — creates substantial switching costs for customers who have built production workloads on CUDA-native toolchains. All major cloud service providers, AI model makers, and enterprises use the company's data center-scale infrastructure, and NVDA's computing platforms are available from virtually every major server maker and CSP, reflecting the depth of its distribution reach. The 2020 acquisition of Mellanox extended the competitive perimeter further by integrating high-performance networking — NVLink interconnects and switches, plus end-to-end InfiniBand and Ethernet platforms — directly into the stack, enabling architectures that can interconnect up to hundreds of thousands of compute nodes. Vertical industry stacks including NVIDIA DRIVE for autonomous driving, Clara for healthcare, Omniverse for physical AI, and NVIDIA AI Enterprise for enterprise AI applications deepen customer dependency and elevate the cost of switching to an alternative compute platform across the industries NVDA is targeting.

The most acute structural risk to NVDA's competitive position is its effective foreclosure from China's data center computing market. As of the end of fiscal year 2026, and persisting into the first quarter of fiscal year 2027, the company was effectively locked out of China's data center compute segment, with no viable path to creating and delivering a competitive product that simultaneously receives approval from both the U.S. government and the Chinese government. This foreclosure is not merely a revenue headwind; the company has explicitly stated that its exclusion from China allowed competitors to build larger developer and customer ecosystems capable of challenging NVDA on a worldwide basis — a dynamic that is strategically self-reinforcing, as ecosystem scale compounds over time. The Chinese government has further compounded this disadvantage by actively encouraging customers to purchase from China-based competitors and discouraging the purchase, import, or use of NVDA's data center products, including any China-specific product designed to comply with U.S. export controls. The company has warned that its competitive position has already been materially harmed by export controls and that further harm is likely over the long term if restrictions remain in place or are expanded in geographic, customer, or product scope. Parallel to the China risk, the company has identified the trend toward customer and partner self-supply as an additional vector of competitive erosion: cloud partners and customers may develop their own internal solutions, existing customers may purchase from competitors, and partners may discontinue sales or lose market share in ways that reduce their demand for NVDA products — all of which could alter purchasing patterns in a manner that structurally reduces the revenue base over time.

• Competitive · intensity The market for our products is intensely competitive and is characterized by rapid technological change and evolving industry standards. We expect an increasingly competitive environment in the future.
• Competitive · factors We believe that the principal competitive factors in this market are performance, breadth of product offerings, access to customers and partners and distribution channels, software support, conformity to industry standard APIs, manufacturing capabilities, processor pricing, and total system costs.
• Competitive · lower priced entrants We expect competition to increase from both existing competitors and new market entrants with products that may be lower priced than ours or may provide better performance or additional features not provided by our products. In addition, it is possible that new competitors or alliances among competitors could emerge and acquire significant market share.
• Competitive · sources of competition A significant source of competition comes from companies that provide or intend to provide GPUs, CPUs, DPUs, embedded SoCs, and other accelerated, AI computing processor products, and providers of semiconductor-based high-performance interconnect products based on InfiniBand, Ethernet, Fibre Channel, and proprietary technologies. Some competitors may have greater marketing, financial, distribution and manufacturing resources than we do.
• Competitive · competitors ai compute Current competitors include suppliers and licensors of hardware and software for discrete and integrated GPUs, custom chips and other accelerated computing solutions, including solutions offered for AI, such as Advanced Micro Devices, Inc. (AMD), Huawei Technologies Co. Ltd. (Huawei), and Intel Corporation (Intel).
• Competitive · competitors cloud internal Competitors include large cloud services companies with internal teams designing hardware and software that incorporate accelerated or AI computing functionality as part of their internal solutions or platforms, such as Alibaba Group, Alphabet Inc., Amazon, Inc. (Amazon), Baidu, Inc., Huawei, and Microsoft Corporation (Microsoft).
• Competitive · competitors cpu Competitors include suppliers of Arm-based CPUs and companies that incorporate hardware and software for CPUs as part of their internal solutions or platforms, such as Amazon, Huawei, and Microsoft.
• Competitive · competitors soc Competitors include suppliers of hardware and software for SoC products used in servers or embedded into automobiles, autonomous machines, and gaming devices, such as Ambarella, Inc., AMD, Broadcom, Intel, Qualcomm Incorporated, Renesas Electronics Corporation, and Samsung, or companies with internal teams designing SoC products for their own products and services, such as Tesla, Inc.
• Competitive · competitors networking Networking competitors (switches, network adapters including DPUs, and cable solutions including optical modules) include AMD, Arista Networks, Broadcom, Cisco Systems, Inc., Hewlett Packard Enterprise Company, Huawei, Intel, Lumentum Holdings Inc., and Marvell Technology, Inc., as well as internal teams of system vendors and large cloud services companies.
• Competitive · china foreclosure As of the end of fiscal year 2026, we were effectively foreclosed from competing in China's data center computing/compute market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide. Under current rules we are unable to create and deliver a competitive product for China's data center market that receives approval from both the USG and the Chinese government.
• Competitive · china government favors rivals The Chinese government has encouraged customers to purchase from our China-based competitors and discouraged customers from purchasing, importing, or using our data center products, including any China-specific product designed to comply with U.S. export controls.
• Competitive · export controls position harm Our competitive position has been harmed by export controls, and our competitive position and future results will be further harmed, over the long term, if the restrictions remain in place or are expanded in geographic, customer, or product scope, if customers purchase product from competitors, or if customers develop their own internal solution.
• Competitive · cloud self supply Our partners or customers may develop their own solutions; our customers may purchase products from our competitors; and our partners may discontinue sales or lose market share in the markets for which they purchase our products, all of which may alter partners' or customers' purchasing patterns.
• Competitive · china foreclosure q1fy27 As of the end of the first quarter of fiscal year 2027, while we were able to ship uncontrolled products to China, such as gaming and workstation GPUs, we were effectively foreclosed from competing in China's data center computing/compute market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide.
• Moat · scale rd cumulative Innovation is at our core. We have invested over $76.7 billion in research and development since our inception, yielding inventions that are essential to modern computing.
• Moat · scale engineers software With a strong engineering culture, we drive fast, yet harmonized, product and technology innovations in all dimensions of computing including silicon, systems, networking, software and algorithms. More than half of our engineers work on software.
• Moat · scale rd headcount As of the end of fiscal year 2026, we had approximately 42,000 employees in 38 countries; 31,000 were engaged in research and development and 11,000 were engaged in sales, marketing, operations, and administrative positions. More than 80 percent have technical roles and more than half of the workforce hold an advanced degree.
• Moat · scale full stack While our approach starts with powerful chips, what makes it a full-stack computing platform is our large body of software, including the CUDA development platform, the CUDA-X collection of acceleration libraries, AI models and training data sets, APIs, SDKs, and domain-specific application frameworks.
• Moat · scale unified architecture We address [diverse end markets] with a unified underlying programmable architecture allowing us to support several multi-billion-dollar end markets with the same underlying technology by using a variety of software stacks developed either internally or by third-party developers and partners.
• Moat · network cuda developers There are over 7.5 million developers worldwide using CUDA and our other software tools to help deploy our technology in our target markets.
• Moat · network ecosystem value The large and growing number of developers and installed base across our platforms strengthens our ecosystem and increases the value of our platform for our customers.
• Moat · network software libraries Our technology stack includes the foundational NVIDIA CUDA development platform that runs on all NVIDIA GPUs, as well as hundreds of domain-specific software libraries, frameworks, algorithms, software development kits, or SDKs, and application programming interfaces, or APIs. This deep and broad software stack accelerates the performance and facilitates the deployment of NVIDIA accelerated computing.
• Moat · network csp adoption All major cloud service providers, or CSPs, AI model makers, and enterprises use our data center-scale infrastructure and computing platforms to accelerate the services and offerings they deliver to billions of end users and customers. Our computing platforms are available from virtually every major server maker and CSP.
• Moat · switching cuda platform Our technology stack includes the foundational NVIDIA CUDA development platform that runs on all NVIDIA GPUs. With our introduction of CUDA in 2006, we opened the parallel processing capabilities of our GPU to a broad range of compute-intensive applications, paving the way for the emergence of modern AI.
• Moat · brand open model leader We are the leader in accelerating and releasing open AI models which enterprises, sovereigns, and startups can leverage to develop and run applications on our platform.
• Moat · network full software stacks Over the past 5 years, we have built full software stacks that run on top of our GPUs and CUDA to bring AI to the world's largest industries, including NVIDIA DRIVE stack for autonomous driving, Clara for healthcare, Omniverse for physical AI applications, and NVIDIA AI Enterprise software — essentially an operating system for enterprise AI applications.
• Moat · scale networking integration Our acquisition of Mellanox in 2020 expanded our offerings to include networking, enabled our platforms to be data center scale, and led to the introduction of a new processor class — the data processing unit, or DPU. Our networking offerings include NVLink interconnects and switches, end-to-end platforms for InfiniBand and Ethernet.
• Moat · scale datacenter interconnect This has enabled us to architect data center-scale computing platforms that can interconnect up to hundreds of thousands of compute nodes with high-performance networking.
• Moat · network developer programs We evangelize AI through partnerships with hundreds of universities and tens of thousands of startups through our Inception program. Additionally, our Deep Learning Institute provides instruction on the latest techniques on how to design, train, and deploy neural networks using our accelerated computing platform.
• Moat · regulatory export china As of the end of fiscal year 2026, we were effectively foreclosed from competing in China's data center computing/compute market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide.

Moat Analysis

NVIDIA's competitive moat is anchored first and foremost in an extraordinary and sustained commitment to innovation. Since its inception, the company has invested over $76.7 billion in research and development, a cumulative expenditure that has yielded inventions now considered foundational to modern computing. As of the end of fiscal year 2026, approximately 42,000 employees across 38 countries work for NVIDIA, with 31,000 of them engaged directly in research and development — a concentration of technical talent that is exceptionally rare among semiconductor companies of any size. More than 80 percent of the workforce holds technical roles, and more than half hold an advanced degree, underscoring that this is not simply a capital-intensive endeavor but an intellectual one. Crucially, more than half of NVIDIA's engineers work on software rather than hardware, reflecting the company's deliberate positioning as a full-stack computing platform rather than a chip vendor. This orientation shapes every layer of the moat: the deeper the software investments compound, the higher the barriers that competitors must clear to offer a genuinely equivalent alternative.

The single most durable structural advantage NVIDIA possesses is the CUDA platform, introduced in 2006. CUDA opened the parallel-processing capabilities of NVIDIA GPUs to a broad range of compute-intensive applications, and in doing so it paved the way for the emergence of modern AI as a practical discipline. Today, over 7.5 million developers worldwide use CUDA and NVIDIA's other software tools to deploy the company's technology across its target markets. This installed base of developers represents a formidable switching-cost moat: the libraries, algorithms, workflows, and institutional knowledge that these millions of practitioners have built on top of CUDA are not portable to competing architectures without substantial re-engineering effort. The switching cost is compounded by the breadth of the software layer itself. NVIDIA's technology stack encompasses the foundational CUDA development platform — which runs on all NVIDIA GPUs — plus hundreds of domain-specific software libraries, frameworks, SDKs, and APIs. This depth and breadth accelerates performance and facilitates deployment in ways that a hardware-only challenger simply cannot replicate without years of analogous software investment alongside equivalent silicon. The full-stack character of the platform is further expressed through NVIDIA's unified programmable architecture, which allows the company to support several multi-billion-dollar end markets with the same underlying technology, enabling extraordinary capital efficiency at scale.

NVIDIA's moat is further reinforced by powerful network effects operating at multiple levels simultaneously. The large and growing number of developers and the expanding installed base across NVIDIA's platforms strengthen the ecosystem and increase the value of the platform for every participant within it. All major cloud service providers, AI model makers, and enterprises use NVIDIA's data center-scale infrastructure to accelerate the services they deliver to billions of end users, and the company's computing platforms are available from virtually every major server maker and CSP. Over the past five years NVIDIA has built full software stacks running on top of its GPUs and CUDA to penetrate the world's largest industries — including the NVIDIA DRIVE stack for autonomous driving, Clara for healthcare, Omniverse for physical AI applications, and NVIDIA AI Enterprise, which functions essentially as an operating system for enterprise AI applications. The company evangelizes AI through partnerships with hundreds of universities and tens of thousands of startups through its Inception program, continuously enlarging the pool of developers whose careers and codebases are built on CUDA. These vertical stacks and developer programs create additional layers of ecosystem lock-in, since an enterprise that has standardized on NVIDIA AI Enterprise faces friction not only at the hardware level but at the application framework and workflow level as well. One material risk to this otherwise formidable competitive position is the regulatory constraint on competing in China's data center market: as of the end of fiscal year 2026, NVIDIA was effectively foreclosed from that market, and management acknowledged that this foreclosure has allowed competitors to build larger developer and customer ecosystems that could challenge the company worldwide. Nevertheless, with $76.7 billion in cumulative R&D, 31,000 employees dedicated to research and development, 7.5 million CUDA developers globally, and a unified architecture capable of addressing several multi-billion-dollar end markets with the same underlying technology, the structural advantages NVIDIA has assembled constitute a compounding, self-reinforcing system that would require extraordinary sustained effort — in capital, talent, software depth, and time — to meaningfully dismantle.

• Rd · cumulative invested NVIDIA has invested over $76.7 billion in research and development since its inception, yielding inventions essential to modern computing.
• Moat · scale rd cumulative Innovation is at our core. We have invested over $76.7 billion in research and development since our inception, yielding inventions that are essential to modern computing.
• Moat · scale engineers software With a strong engineering culture, we drive fast, yet harmonized, product and technology innovations in all dimensions of computing including silicon, systems, networking, software and algorithms. More than half of our engineers work on software.
• Moat · scale rd headcount As of the end of fiscal year 2026, we had approximately 42,000 employees in 38 countries; 31,000 were engaged in research and development and 11,000 were engaged in sales, marketing, operations, and administrative positions. More than 80 percent have technical roles and more than half of the workforce hold an advanced degree.
• Moat · scale full stack While our approach starts with powerful chips, what makes it a full-stack computing platform is our large body of software, including the CUDA development platform, the CUDA-X collection of acceleration libraries, AI models and training data sets, APIs, SDKs, and domain-specific application frameworks.
• Moat · scale unified architecture We address [diverse end markets] with a unified underlying programmable architecture allowing us to support several multi-billion-dollar end markets with the same underlying technology by using a variety of software stacks developed either internally or by third-party developers and partners.
• Moat · network cuda developers There are over 7.5 million developers worldwide using CUDA and our other software tools to help deploy our technology in our target markets.
• Moat · network ecosystem value The large and growing number of developers and installed base across our platforms strengthens our ecosystem and increases the value of our platform for our customers.
• Moat · network software libraries Our technology stack includes the foundational NVIDIA CUDA development platform that runs on all NVIDIA GPUs, as well as hundreds of domain-specific software libraries, frameworks, algorithms, software development kits, or SDKs, and application programming interfaces, or APIs. This deep and broad software stack accelerates the performance and facilitates the deployment of NVIDIA accelerated computing.
• Moat · network csp adoption All major cloud service providers, or CSPs, AI model makers, and enterprises use our data center-scale infrastructure and computing platforms to accelerate the services and offerings they deliver to billions of end users and customers. Our computing platforms are available from virtually every major server maker and CSP.
• Moat · switching cuda platform Our technology stack includes the foundational NVIDIA CUDA development platform that runs on all NVIDIA GPUs. With our introduction of CUDA in 2006, we opened the parallel processing capabilities of our GPU to a broad range of compute-intensive applications, paving the way for the emergence of modern AI.
• Moat · brand open model leader We are the leader in accelerating and releasing open AI models which enterprises, sovereigns, and startups can leverage to develop and run applications on our platform.
• Moat · network full software stacks Over the past 5 years, we have built full software stacks that run on top of our GPUs and CUDA to bring AI to the world's largest industries, including NVIDIA DRIVE stack for autonomous driving, Clara for healthcare, Omniverse for physical AI applications, and NVIDIA AI Enterprise software — essentially an operating system for enterprise AI applications.
• Moat · scale networking integration Our acquisition of Mellanox in 2020 expanded our offerings to include networking, enabled our platforms to be data center scale, and led to the introduction of a new processor class — the data processing unit, or DPU. Our networking offerings include NVLink interconnects and switches, end-to-end platforms for InfiniBand and Ethernet.
• Moat · scale datacenter interconnect This has enabled us to architect data center-scale computing platforms that can interconnect up to hundreds of thousands of compute nodes with high-performance networking.
• Moat · network developer programs We evangelize AI through partnerships with hundreds of universities and tens of thousands of startups through our Inception program. Additionally, our Deep Learning Institute provides instruction on the latest techniques on how to design, train, and deploy neural networks using our accelerated computing platform.
• Moat · regulatory export china As of the end of fiscal year 2026, we were effectively foreclosed from competing in China's data center computing/compute market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide.

Market Position

NVIDIA has established an unrivaled position at the center of the accelerated-computing era, building what amounts to the foundational infrastructure layer for modern artificial intelligence. Its GPU architectures and the CUDA software ecosystem together create a deeply embedded technological advantage that spans training, inference, and high-performance networking — underpinning cloud hyperscalers, sovereign AI initiatives, and enterprise deployments alike. The company's dominance is not merely a function of silicon performance; it reflects years of co-investment in developer tooling, frameworks, and system-level integration that have made NVIDIA hardware the default substrate on which the AI industry operates. This entrenched position translates directly into the revenue composition reported for FY2026, where the Compute & Networking segment alone contributed $193,479 million, representing the overwhelming majority of consolidated revenue and confirming that the company's fortunes are now inextricably bound to the data-center buildout driving AI workloads at scale.

The geographic revenue mix for FY2026 underscores both the extraordinary depth of demand in the United States and a meaningful, if concentrated, exposure profile elsewhere. The United States generated $149,617 million in revenue, reflecting the extraordinary pace at which domestic hyperscalers and cloud service providers have committed capital to AI infrastructure — a demand signal that is among the most durable capital-expenditure programs in modern technology. Taiwan contributed $42,345 million, a figure that reflects the outsized role of the island's electronics and manufacturing supply chain in NVIDIA's channel structure; many system integrators and OEM partners bill through Taiwanese entities before hardware flows to ultimate end-customers globally, meaning a portion of this figure proxies broader Asia-Pacific and international demand. China represented $19,677 million, a region that carries material geopolitical complexity. Export-licensing restrictions imposed by the U.S. government have constrained the product set NVIDIA can legally sell into mainland China, and the company has navigated successive regulatory tightening that has progressively limited addressable configurations. The remaining geographies collectively contributed $4,299 million, a relatively modest figure that highlights how concentrated global AI infrastructure spending currently remains in the leading economies. For institutional investors, the geographic map implies a bifurcated risk profile: a domestic U.S. demand base anchored by hyperscaler capital programs of unusual scale and duration, alongside an international exposure that must be read through the lens of export controls, allied-nation procurement programs, and the uneven pace of AI adoption outside the core markets.

The segment and platform architecture reinforces NVIDIA's strategic positioning and clarifies the sources of long-cycle growth. The Compute & Networking segment, generating $193,479 million in FY2026, captures the core AI-acceleration and interconnect franchise — encompassing GPU compute for large-scale training and inference workloads as well as the high-performance networking infrastructure that makes distributed AI clusters operable. The Graphics segment contributed $22,459 million, a materially smaller but profitable business serving PC gaming and professional visualization end-markets; at its current scale it provides an important diversification buffer but no longer defines NVIDIA's strategic identity or investment thesis. On a platform basis, the Data Center franchise produced $193,737 million, effectively co-extensive with the Compute & Networking segment and confirming that AI infrastructure is now the singular engine of the company's revenue profile. Gaming, at $16,042 million, retains a healthy installed base and a recurring upgrade cycle tied to the PC gaming ecosystem, but the magnitude differential against Data Center is decisive and unlikely to narrow materially in the medium term. Looking forward, the addressable opportunity available to NVIDIA remains extraordinarily large and largely underpenetrated. The global inventory of legacy CPU-based server infrastructure represents a multi-year conversion opportunity; the expansion of inference workloads from hyperscale cloud into enterprise and edge environments opens additional deployment surfaces; and sovereign AI programs in Europe, the Middle East, and Asia-Pacific are creating new, government-sponsored demand pools that were not part of the market's growth calculus even a few years ago. NVIDIA's architectural breadth — from silicon through software to full-stack systems — positions it to capture a disproportionate share of each incremental layer of that buildout as the AI infrastructure cycle continues to mature.

Customer Base

NVIDIA's customer base for FY2026 exhibits pronounced concentration risk, with a relatively small number of hyperscaler and enterprise accounts driving a disproportionate share of total revenue. Sales to one direct customer represented 22% of total revenue for FY2026, while sales to a second direct customer represented 14% of total revenue, with both relationships primarily attributable to the Compute & Networking segment — NVIDIA's flagship AI infrastructure business. Beyond these two named direct customers, NVIDIA estimates that certain indirect customers each individually represented 10% or more of revenue as well, including at least one AI research and deployment company that contributed a meaningful revenue amount indirectly through cloud-service provider channels. The receivables ledger reinforces this picture: as of January 25, 2026, three direct customers accounted for 25%, 18%, and 13% of the total accounts receivable balance, respectively, indicating that payment timing is tightly coupled to the capital-expenditure cycles of a handful of large platform operators. Against a total FY2026 revenue base of $215,938 million, the combined revenue share of just the two largest direct customers implies that a change in purchasing behavior, budget prioritization, or infrastructure strategy by either counterparty could materially alter NVIDIA's top-line trajectory. This level of concentration is structurally common among leading semiconductor suppliers serving the hyperscaler market, but it creates asymmetric exposure to any individual customer's capex deceleration or vendor diversification initiative.

The supply-side of NVIDIA's value chain mirrors this concentration dynamic, with the company relying on a tightly integrated set of Asian foundry, memory, and assembly partners to bring its products to market. Semiconductor wafer fabrication is entrusted to two foundries — Taiwan Semiconductor Manufacturing Company Limited (TSMC) and Samsung Electronics Co., Ltd. — making NVIDIA highly dependent on the operational continuity and capacity allocation decisions of these two entities. Memory procurement is sourced across SK Hynix Inc., Micron Technology, Inc., and Samsung, providing some diversification within that component category, but all three are large-scale commodity suppliers whose pricing and allocation behavior can shift simultaneously during periods of broad industry demand pressure. Final assembly, testing, and packaging is handled by independent contract manufacturers including Hon Hai Precision Industry Co., Ltd. (Foxconn), Wistron Corporation, and Fabrinet. Critically, the advanced packaging technology at the heart of NVIDIA's highest-performance products — CoWoS — is a specialized process with limited global capacity, concentrating execution risk further. NVIDIA has acknowledged efforts to expand its supply chain footprint into the U.S. and Latin America, but the overwhelming share of production capacity remains in Asia, exposing the company to geopolitical, logistical, and regulatory risks that are largely outside its direct control. Taken together, the customer and supplier concentration profiles define a business that generates outsized revenue efficiency from deep, bilateral relationships — but one whose revenue stability and gross margin trajectory are closely tied to the strategic and operational health of a short list of counterparties on both sides of its value chain.

• Concentration · customer fy26 Customer concentration FY2026: sales to one direct customer represented 22% of total revenue and sales to another direct customer represented 14% of total revenue, primarily attributable to the Compute & Networking segment. NVIDIA also estimates that some indirect customers individually represented 10% or more of revenue, and that one AI research and deployment company contributed a meaningful amount of revenue indirectly through cloud-service customers. Three direct customers accounted for 25%, 18% and 13% of the accounts receivable balance as of January 25, 2026.
• Concentration · key suppliers Supply chain is mainly concentrated in Asia, with expansion into the U.S. and Latin America. Foundries: Taiwan Semiconductor Manufacturing Company Limited (TSMC) and Samsung Electronics Co., Ltd. (Samsung) produce semiconductor wafers. Memory is purchased from SK Hynix Inc., Micron Technology, Inc., and Samsung. Semiconductor packaging utilizes CoWoS technology. Assembly, testing and packaging of final products is performed by independent subcontractors / contract manufacturers including Hon Hai Precision Industry Co., Ltd. (Foxconn), Wistron Corporation, and Fabrinet.

Insider Ownership

NVIDIA's latest annual proxy statement discloses beneficial ownership for every named executive officer and director, for the board and officers as a group, and for all holders of more than five percent of the common stock — the first in-corpus governance disclosure of who actually owns the company, drawn directly from the proxy's beneficial-ownership table.

Founder, President and Chief Executive Officer Jen-Hsun Huang remains by a wide margin the largest individual insider, with the great majority of his holdings sitting in family trusts, limited liability companies and a family foundation rather than directly in his own name. Every other named executive officer and independent director holds well below one percent individually, so direct insider ownership is highly concentrated in the founder; the per-holder share counts are listed below.

Taken together, all directors and executive officers as a group beneficially own 3.94% of the company — a meaningful alignment of management with shareholders that is nonetheless modest relative to the institutional base, a reflection of NVIDIA's very large public float.

On the institutional side, two asset managers report holdings above the regulatory reporting threshold and are the only disclosed five-percent owners. BlackRock is the larger of the two at 7.43% of shares outstanding, with Vanguard reporting a holding of broadly comparable size; both are passive index managers rather than active stakeholders. Each figure below traces to the proxy's beneficial-ownership table, and the percentages are computed against the disclosed share-count basis and validated as derivations rather than asserted.

• Ownership · huang Jen-Hsun Huang (Founder, President & CEO) — 870,604,104 shares beneficially owned, the majority held through family trusts, LLCs and a family foundation.
• Ownership · kress Colette M. Kress (EVP & CFO) — 4,443,523 shares beneficially owned.
• Ownership · puri Ajay K. Puri (EVP, Worldwide Field Operations) — 3,266,897 shares beneficially owned.
• Ownership · shoquist Debora Shoquist (EVP, Operations) — 1,580,729 shares beneficially owned.
• Ownership · teter Timothy S. Teter (EVP & General Counsel) — 2,687,197 shares beneficially owned.
• Ownership · coxe Tench Coxe (director) — 30,555,240 shares beneficially owned.
• Ownership · dabiri John O. Dabiri (director) — 14,788 shares beneficially owned.
• Ownership · hudson Dawn Hudson (director) — 368,887 shares beneficially owned.
• Ownership · jones Harvey C. Jones (director) — 7,003,687 shares beneficially owned.
• Ownership · lora Melissa B. Lora (director) — 1,799 shares beneficially owned.
• Ownership · neal Stephen C. Neal (director) — 184,867 shares beneficially owned.
• Ownership · seawell A. Brooke Seawell (director) — 2,506,607 shares beneficially owned.
• Ownership · shah Aarti Shah (director) — 25,620 shares beneficially owned.
• Ownership · stevens Mark A. Stevens (director) — 34,069,572 shares beneficially owned.
• Ownership · group All directors and executive officers as a group (14 persons) — 957,313,517 shares beneficially owned.
• Ownership · blackrock BlackRock, Inc. (principal stockholder) — 1,805,935,550 shares beneficially owned, reported on Schedule 13G/A.
• Ownership · vanguard Vanguard Capital Management (principal stockholder) — 1,777,408,252 shares beneficially owned, reported on Schedule 13G.
• Ownership · shares outstanding basis Shares outstanding basis for ownership percentages: 24,312,141,810 shares of common stock as of March 23, 2026 (the proxy's disclosed denominator).

Insider Trading Activity

Over the trailing-quarter window, NVIDIA's reporting insiders were small net sellers of common stock — a net 101,523 shares — but the headline is dominated by two large offsetting moves, a senior EVP's discretionary sale set against the founder-CEO's grant-driven increase. The activity is overwhelmingly routine scheduled selling and equity grants, not a directional bet.

Caveat: the small net nets routine grants and option exercises against open-market sales. The discretionary open-market component alone was a net 912,357 shares sold under pre-arranged trading plans — so the headline understates the scale of scheduled insider selling; read the net as direction, not magnitude.

Management Quality

NVIDIA's executive leadership is anchored by its co-founder, Jen-Hsun Huang, age 63, who has served as President, Chief Executive Officer, and a member of the Board of Directors since the company's founding in 1993 — a tenure spanning more than three decades that is exceptionally rare among executives of large-capitalization technology companies. This longevity provides meaningful strategic continuity and reflects a founder-led culture that has shaped NVIDIA's technology bets across GPU computing, gaming, data center, and artificial intelligence. Alongside Huang, the named executive officer team includes Colette M. Kress, age 58, Executive Vice President and CFO, whose financial stewardship complements Huang's product and technology vision. The broader executive bench deepens this experience: Ajay K. Puri, age 71, leads Worldwide Field Operations; Debora Shoquist, age 71, heads Operations; Timothy S. Teter, age 59, serves as General Counsel; and Donald Robertson holds the role of Chief Accounting Officer. This combination of long-tenured operational leaders and a founder-CEO creates a management team with deep institutional knowledge and aligned incentives built over decades of shared history at the company.

The compensation structure for NVIDIA's executive team reflects a philosophy that ties the substantial majority of pay to measurable performance outcomes. For Fiscal 2026, Jen-Hsun Huang received total compensation of $36,343,830, composed of a base salary of $1,497,627 (against a target of $1,500,000, held flat with Fiscal 2025), stock awards of $24,800,511, non-equity incentive plan compensation of $6,000,000, and all other compensation of $4,045,691. The preponderance of equity — $24,800,511 out of a total of $36,343,830 — underscores the board's intent to link Huang's realized pay directly to shareholder value creation rather than guaranteed fixed income. CFO Colette Kress received total Fiscal 2026 compensation of $14,340,850, with stock awards of $12,825,872 representing the dominant component. The Fiscal 2026 CEO-to-median-employee pay ratio was 129:1, set against a median employee annual total compensation of $282,050. While a 129:1 ratio is significant in absolute terms, it must be assessed in the context of NVIDIA's workforce composition — primarily highly compensated engineers and technical professionals — and reflects the outsized scale of performance-contingent equity grants to the CEO rather than disproportionate fixed pay packages.

The design of the equity program is the most structurally important feature of NVIDIA's pay-for-performance framework and warrants close examination. One hundred percent of the CEO's equity awards are granted solely as performance stock units (PSUs), with no restricted stock units and no stock options outstanding since the end of Fiscal 2024. The Compensation Committee structured this allocation to maximize alignment between Huang's incentives and shareholder outcomes: grants are split evenly between single-year PSUs (SY PSUs) tied to annual Non-GAAP Operating Income, vesting over four years, and multi-year PSUs (MY PSUs) benchmarked against NVIDIA's three-year total shareholder return relative to the S&P 500, vesting over three years. This dual-metric design simultaneously links near-term operational execution and long-term capital market performance to realizable compensation. Other named executive officers receive a blend of 50% RSUs and 50% PSUs, with full vesting requiring four years of continuous service for both RSU and SY PSU awards. In aggregate, 96% of the CEO's Fiscal 2026 total target pay was dependent on corporate performance, compared with 48% for other named executive officers — ensuring that the principal decision-maker bears the highest proportional performance risk. Stockholder endorsement of this framework has been strong: the most recently disclosed say-on-pay advisory vote, covering Fiscal 2024 compensation, received 92% approval.

Governance safeguards further reinforce the quality and accountability of NVIDIA's management structure. The company has maintained a Compensation Recovery Policy since 2009, updated in 2023 to comply with Nasdaq listing standards, requiring the recovery of certain incentive compensation from current or former executive officers in the event of qualifying financial restatements where compensation received exceeds what would have been earned based on restated financials. The Fiscal 2026 compensation peer group was calibrated against sixteen companies — including Alphabet, Amazon, Apple, Microsoft, Meta Platforms, and Broadcom — following the removal of AMD, Adobe, and SAP SE and the addition of the four largest U.S. technology companies by market capitalization, reflecting NVIDIA's own ascent into that tier and ensuring benchmarking against peers of commensurate scale and complexity. Compensation consulting is provided by Semler Brossy. External audit oversight has been conducted by PricewaterhouseCoopers LLP (PCAOB ID: 238) since 2004, providing sustained auditor continuity and deep familiarity with NVIDIA's evolving revenue and product structures. Taken together, these governance mechanisms — a founder-CEO with over thirty years of institutional alignment, a pay-for-performance equity structure in which 100% of the CEO's grants are at-risk PSUs, a clawback policy with over fifteen years of precedent, peer-group calibration at the apex of U.S. technology, and long-tenured independent audit oversight — present a management quality profile that is structurally consistent with sustained, accountable shareholder value creation.

• Governance · ceo Jen-Hsun Huang, age 63, President and Chief Executive Officer; co-founded NVIDIA in 1993 and has served as President, CEO and a member of the Board since inception.
• Governance · cfo Colette M. Kress, age 58, Executive Vice President and Chief Financial Officer.
• Governance · named executives Ajay K. Puri, age 71, Executive Vice President, Worldwide Field Operations; Debora Shoquist, age 71, Executive Vice President, Operations; Timothy S. Teter, age 59, Executive Vice President and General Counsel; Donald Robertson, Vice President and Chief Accounting Officer (Principal Accounting Officer).
• Governance · auditor PricewaterhouseCoopers LLP, San Jose, California (PCAOB ID: 238); has served as NVIDIA's auditor since 2004.
• Compensation · ceo total fy26 Jen-Hsun Huang, President and CEO — Fiscal 2026 Total compensation: $36,343,830 (Salary $1,497,627; Stock Awards $24,800,511; Non-Equity Incentive Plan Compensation $6,000,000; All Other Compensation $4,045,691).
• Compensation · ceo base salary fy26 Jen-Hsun Huang Fiscal 2026 Salary: $1,497,627 (target base salary was $1,500,000, flat with Fiscal 2025).
• Compensation · ceo equity fy26 Jen-Hsun Huang Fiscal 2026 Stock Awards (ASC 718 aggregate full grant-date fair value of SY PSUs and MY PSUs; CEO equity is 100% PSUs): $24,800,511. NVIDIA reports no option awards (no stock options outstanding since end of Fiscal 2024), so there is no separate option-awards column.
• Compensation · cfo total fy26 Colette M. Kress, EVP and CFO — Fiscal 2026 Total compensation: $14,340,850 (Salary $898,577; Stock Awards $12,825,872; Non-Equity Incentive Plan Compensation $600,000; All Other Compensation $16,402).
• Compensation · cfo equity fy26 Colette M. Kress Fiscal 2026 Stock Awards (ASC 718 aggregate full grant-date fair value of RSUs, SY PSUs and MY PSUs): $12,825,872.
• Compensation · ceo pay ratio Fiscal 2026 CEO-to-median-employee pay ratio was 129:1. CEO's Fiscal 2026 total compensation was $36,343,830 and the median employee's total compensation for Fiscal 2026 was $282,050.
• Compensation · median employee comp Median employee's annual total compensation for Fiscal 2026 was $282,050 (median employee identified from Fiscal 2024 calculation, carried forward).
• Compensation · equity grant practices 100% of the CEO's equity awards have been (and for Fiscal 2026 were) granted solely as PSUs — to even more tightly align his interests with stockholders, the CC determined that 100% of his equity grants should be at-risk and performance-based, split evenly between SY PSUs (annual Non-GAAP Operating Income, vesting over 4 years) and MY PSUs (3-year TSR relative to the S&P 500, vesting over 3 years). Other NEOs receive 50% RSUs and 50% PSUs. NEOs must provide continuous service for 4 years to fully vest in RSU and SY PSU awards.
• Compensation · ceo variable pay pct 96% of the CEO's (and 48% of the other NEOs') Fiscal 2026 total target pay was dependent on corporate performance. (Separately, the proxy notes that in recent years over 90% of the CEO's and over 45% of other NEOs' total target pay have been performance-based and at-risk.)
• Compensation · say on pay Say-on-Pay advisory votes are held annually (the Board and stockholders have approved an annual frequency). The most recent stated approval result referenced was a 92% say-on-pay approval rate for Fiscal 2024 NEO compensation; the FY2026 advisory vote (Proposal 2) is to be voted at the June 24, 2026 Annual Meeting (result not yet available). Proposal 2 requires a majority of shares present and entitled to vote.
• Compensation · director initial equity grant New director Suzanne Nora Johnson (appointed May 7, 2026, joining effective July 13, 2026) will receive an initial equity grant of restricted stock units (RSUs) with a target value of $255,000, vesting approximately semi-annually over three years, under the Amended and Restated 2007 Equity Incentive Plan. (She also receives a pro-rated annual RSU grant with an annualized target value of $255,000.)
• Compensation · director annual cash retainer New director Suzanne Nora Johnson will be granted a pro-rated annual cash retainer with an annualized value of $85,000, payable quarterly starting July 13, 2026.
• Compensation · clawback policy NVIDIA has maintained a Compensation Recovery Policy since 2009, amended in 2023 to comply with Nasdaq's listing standards. It requires the Company to recover certain incentive compensation from current or former executive officers in connection with certain restatements of financial statements, if the compensation exceeds what the officers would have received based on the restated financials, subject to limited exceptions.
• Compensation · peer group Fiscal 2026 compensation peer group (16 companies): Alphabet, Amazon.com, Apple, Broadcom, Cisco Systems, IBM, Intel, Meta Platforms, Microsoft, Netflix, Oracle, Qualcomm, Salesforce, and Visa. Changes from prior year: removed Adobe, Advanced Micro Devices (AMD), and SAP SE; added Alphabet, Amazon, Apple, and Microsoft. Compensation consultant: Semler Brossy.

Management Track Record

NVIDIA's management team delivered an exceptional operating result in FY2026, with revenue of $215,938 million growing 65.5% year-over-year and net income of $120,067 million expanding 64.7%. Scaling a hardware-and-systems franchise of this magnitude at hyper-growth rates — while concurrently ramping new platforms, supply chains, and customer cohorts — is a non-trivial execution feat that, in our view, signals both the operational maturity of the leadership group and the durability of the demand backdrop they have positioned the company to capture.

Profitability sits at best-in-class levels for the semiconductor industry, and the cash conversion that follows is what gives the team genuine capital-allocation optionality. Free cash flow of $96,676 million in FY2026 funded an aggressive $40,086 million of share repurchases alongside a modest $974 million in dividends — a mix we read as a deliberately shareholder-friendly framework that prioritises buyback flexibility while preserving balance-sheet firepower for continued R&D and strategic investment. Returning capital at this scale without compromising reinvestment intensity is a credit to the finance function's stewardship.

Leadership depth is anchored by a founder-CEO with rare longevity. Jen-Hsun Huang co-founded NVIDIA in 1993 and has led the company as President and CEO since inception, following formative engineering and operating roles at LSI Logic and AMD — a more than three-decade run at the helm of a single franchise that is virtually unmatched among large-cap technology peers. CFO Colette Kress, who joined in 2013, brings prior CFO-level experience from Cisco Systems and senior finance leadership at Microsoft and Texas Instruments, giving the finance organisation the large-cap public-company pedigree appropriate to NVIDIA's current scale and complexity.

Governance and alignment screen favourably. The Board is overwhelmingly independent — every director other than Mr. Huang qualifies as independent under Nasdaq rules — and exhibits unusual continuity, with multiple directors serving since the early-to-mid 1990s, which in our judgment supports informed long-horizon oversight rather than the rubber-stamping risk that pure-independence metrics can mask. Alignment is reinforced by CEO pay design: 100% of Mr. Huang's equity grants are at-risk and performance-based, split between annual financial PSUs and multi-year relative-TSR PSUs.

• Governance · ceo Jen-Hsun Huang, age 63, President and Chief Executive Officer; co-founded NVIDIA in 1993 and has served as President, CEO and a member of the Board since inception.
• Governance · cfo Colette M. Kress, age 58, Executive Vice President and Chief Financial Officer.
• Ceo · tenure Jen-Hsun Huang founded NVIDIA in 1993 and has served since its inception as President, Chief Executive Officer, and a member of the Board (director since 1993).
• Ceo · prior roles Prior to founding NVIDIA, Mr. Huang worked at LSI Logic, a semiconductor and software company, and Advanced Micro Devices (AMD), a global semiconductor company. He holds a BSEE degree from Oregon State University and an MSEE degree from Stanford University.
• Cfo · tenure Colette M. Kress joined NVIDIA in 2013 as Executive Vice President and Chief Financial Officer.
• Cfo · prior roles Before NVIDIA, Ms. Kress was Senior Vice President and CFO of the Business Technology and Operations Finance organization at Cisco Systems from 2010; held a variety of positions at Microsoft from 1997 to 2010, including CFO of the Server and Tools division from 2006; and spent eight years at Texas Instruments in finance roles. She holds a B.S. in Finance from the University of Arizona and an MBA from Southern Methodist University.
• Board · independence Except for Mr. Huang, all of NVIDIA's directors are independent as defined by Nasdaq's rules. The Board comprises 10 director nominees, each with a one-year term standing for annual election.
• Board · continuity The Board has deep continuity dating to NVIDIA's founding: Tench Coxe and Harvey C. Jones have served as directors since 1993, A. Brooke Seawell since 1997, and Mark A. Stevens since 2008.
• Ceo · comp alignment To align the CEO with stockholders, the Compensation Committee determined that 100% of Mr. Huang's equity grants are at-risk and performance-based, split evenly between single-year PSUs (aligned with annual corporate financial performance) and multi-year PSUs (aligned with 3-year relative shareholder return).

CEO Background

Jensen (Jen-Hsun) Huang, age 63, co-founded NVIDIA in 1993 and has served continuously as President, Chief Executive Officer, and a member of the Board of Directors since the company's inception — a tenure now exceeding three decades without interruption. This unbroken founder-CEO continuity is structurally uncommon among large-cap technology companies and signals a governance model in which long-term strategic direction, corporate culture, and capital-allocation philosophy are anchored in a single deeply embedded architect. Having steered NVIDIA across multiple product generations and market cycles, Huang represents a concentration of institutional knowledge, brand identity, and stakeholder trust that is difficult to quantify or replace. The governance profile carries meaningful implications for succession risk: the company has not publicly identified a standing successor, and the board's willingness to maintain Huang in this role since 1993 reflects both confidence in his leadership and, implicitly, a bet that continuity outweighs the optionality of planned succession.

Huang's Fiscal 2026 total compensation of $36,343,830 is structured to place the overwhelming majority of his realizable pay at risk. The fixed component — a base salary of $1,497,627 against a $1,500,000 target, held flat relative to Fiscal 2025 — represents a deliberately modest anchor; the Compensation Committee's decision to leave base pay essentially unchanged signals that fixed cost is not the lever through which Huang is rewarded. The equity component of $24,800,511 (reported at aggregate ASC 718 grant-date fair value) is composed entirely of performance stock units, with no time-vested restricted shares and no stock options outstanding; NVIDIA eliminated stock options entirely by the end of Fiscal 2024. This 100%-PSU equity design means every dollar of Huang's equity award is contingent on achieving defined corporate performance milestones rather than mere tenure, creating a direct, contractual alignment between his realized wealth and shareholder outcomes. Non-equity incentive plan compensation of $6,000,000 and all other compensation of $4,045,691 round out the package. In aggregate, 96% of Huang's total target pay for Fiscal 2026 was dependent on corporate performance, a figure the proxy notes has exceeded 90% in recent years — making NVIDIA's CEO compensation among the most performance-contingent packages in the sector. The resulting CEO-to-median-employee pay ratio for Fiscal 2026 was 129:1, with the median NVIDIA employee earning total compensation of $282,050. That this ratio is as contained as it is, given the scale of Huang's package, reflects both the heavily equity-weighted and variable nature of his pay and NVIDIA's engineering-intensive global workforce, whose compensation levels are themselves elevated relative to broad technology-industry medians.

• Governance · ceo Jen-Hsun Huang, age 63, President and Chief Executive Officer; co-founded NVIDIA in 1993 and has served as President, CEO and a member of the Board since inception.
• Compensation · ceo total fy26 Jen-Hsun Huang, President and CEO — Fiscal 2026 Total compensation: $36,343,830 (Salary $1,497,627; Stock Awards $24,800,511; Non-Equity Incentive Plan Compensation $6,000,000; All Other Compensation $4,045,691).
• Compensation · ceo base salary fy26 Jen-Hsun Huang Fiscal 2026 Salary: $1,497,627 (target base salary was $1,500,000, flat with Fiscal 2025).
• Compensation · ceo equity fy26 Jen-Hsun Huang Fiscal 2026 Stock Awards (ASC 718 aggregate full grant-date fair value of SY PSUs and MY PSUs; CEO equity is 100% PSUs): $24,800,511. NVIDIA reports no option awards (no stock options outstanding since end of Fiscal 2024), so there is no separate option-awards column.
• Compensation · ceo pay ratio Fiscal 2026 CEO-to-median-employee pay ratio was 129:1. CEO's Fiscal 2026 total compensation was $36,343,830 and the median employee's total compensation for Fiscal 2026 was $282,050.
• Compensation · ceo variable pay pct 96% of the CEO's (and 48% of the other NEOs') Fiscal 2026 total target pay was dependent on corporate performance. (Separately, the proxy notes that in recent years over 90% of the CEO's and over 45% of other NEOs' total target pay have been performance-based and at-risk.)

CFO Background

Colette M. Kress serves as Executive Vice President and Chief Financial Officer of NVIDIA Corporation, a role she has held since 2013. At 58 years of age, she brings more than three decades of finance leadership to one of the most strategically consequential CFO seats in the technology sector [445]. Her appointment preceded the explosive growth of the accelerated-computing era, and her stewardship of NVIDIA's capital structure, investor relations, financial planning, and reporting has coincided with the company's ascent to a position of commanding influence across artificial intelligence, data-center infrastructure, and enterprise computing. The breadth and complexity of the CFO mandate at a company of NVIDIA's scale — spanning global revenue recognition, treasury, tax strategy, and the continuous dialogue with an exceptionally engaged institutional investor base — demands precisely the combination of operational depth and capital-markets fluency that Kress has developed over her career.

Prior to joining NVIDIA, Kress held senior finance roles at Cisco Systems and Microsoft, where she developed expertise in managing the complex financial architectures of large-scale technology enterprises. This cross-industry background — spanning networking, enterprise software, and now semiconductors — equips her with a broad vantage point on capital allocation and cost management that is particularly well suited to NVIDIA's multi-segment, geographically diversified revenue profile. Her remit at NVIDIA encompasses financial reporting and controls, treasury operations, tax planning, and investor relations, each of which carries heightened significance given the intensity of institutional and regulatory scrutiny that accompanies NVIDIA's market position and the rapid pace at which its product mix continues to evolve.

Kress's fiscal 2026 compensation reflects both the scale of the responsibility she carries and a deliberate emphasis on long-term shareholder alignment. Total compensation for the fiscal year reached $14,340,850, comprising a base salary of $898,577, non-equity incentive plan compensation of $600,000, and all other compensation of $16,402 [572]. The dominant element of her package is equity: stock awards carrying an aggregate grant-date fair value of $12,825,872, encompassing restricted stock units alongside short-year and multi-year performance stock units [573]. That equity component represents the overwhelming majority of her total pay, and the multi-year performance vesting structure ties her economic outcomes directly to sustained shareholder value creation over a horizon that extends well beyond any single fiscal period. This architecture — heavy on long-duration, performance-contingent instruments — is consistent with NVIDIA's board-level philosophy of aligning executive incentives with durable operating excellence rather than near-term earnings optimization, and it signals to institutional investors that the CFO's interests are structurally convergent with their own.

• Governance · cfo Colette M. Kress, age 58, Executive Vice President and Chief Financial Officer.
• Compensation · cfo total fy26 Colette M. Kress, EVP and CFO — Fiscal 2026 Total compensation: $14,340,850 (Salary $898,577; Stock Awards $12,825,872; Non-Equity Incentive Plan Compensation $600,000; All Other Compensation $16,402).
• Compensation · cfo equity fy26 Colette M. Kress Fiscal 2026 Stock Awards (ASC 718 aggregate full grant-date fair value of RSUs, SY PSUs and MY PSUs): $12,825,872.

Key Facts

Income Statement

Balance Sheet

Cash Flow Statement

Segment Detail

Capital Allocation

NVIDIA returned approximately $41.4B to shareholders in FY2026 — overwhelmingly through share repurchases, with $40.4B of common stock retired across the fiscal year ($40.1B reflected on the cash-flow statement; the small reconciliation gap to the headline figure reflects settled-versus-trade-date timing and excise-tax accruals, not a discrepancy of substance). On August 26, 2025, the Board expanded total buyback capacity to $58.5B — equivalent to roughly twenty months of repurchase pace at FY2026 velocity — signalling continued willingness to deploy excess operating cash flow to shareholders rather than to defensive cash-pile accumulation or undisciplined M&A.

The mix is deliberate. NVIDIA's dividend remains a token commitment (declared at four cents per share annually post-split), with the variable buyback program absorbing the bulk of returns. That structure preserves optionality: management can throttle repurchases down if hyperscaler demand normalises, where a richer dividend would harden a payout floor the business may not want to defend through a digestion cycle.

Customer concentration sharpens the case. One direct customer represented 22% of FY2026 revenue, with a second at 14%; capital- return decisions therefore implicitly hedge against a single- customer reset. Returning cash now — while gross margins are elevated and accelerator demand still outstrips supply — converts peak-cycle earnings into a permanently lower share count, raising the floor under EPS if a normalisation phase compresses revenue growth. The combination of a $58.5B authorization against $40.4B actual FY2026 pace gives management room to lean in further if Blackwell Ultra and Rubin cycles sustain the current cash-generation profile.

Financial Analysis

NVDA's financial results for fiscal year 2026 document one of the most compressed and sustained revenue expansions in the history of large-cap technology. Total revenue reached $215,938 million, the product of two consecutive years of extraordinary top-line acceleration. In the prior period — FY2024 to FY2025 — revenue expanded 114.2%, reflecting the initial surge in demand for accelerated computing infrastructure as hyperscalers raced to build generative-AI training capacity. Rather than reverting to trend, growth remained emphatically elevated in FY2025 to FY2026, with revenue advancing a further 65.5% as enterprise adoption broadened, sovereign AI programs added a structurally new demand vector, and inference workloads began generating a durable second wave of cluster investment. Gross profit followed the revenue curve upward, growing 56.8% in the most recent fiscal year to reach $153,463 million, demonstrating that scale efficiencies continue to compound even as the product mix diversifies across system-level configurations. Net income accelerated in near lock-step, posting 64.7% growth year-over-year, confirming that operating leverage remained firmly intact throughout the capacity expansion. Free cash flow growth of 58.9% in FY2025 to FY2026 rounds out a revenue-and-earnings trajectory that few precedents exist to benchmark against in the semiconductor industry.

NVDA's margin profile is an equally compelling feature of the investment case, and one that distinguishes the company from virtually every peer in the capital-intensive semiconductor value chain. Gross margin for FY2026 was 71.1%, a figure that remains among the highest achieved by any large-scale semiconductor company in history. It is worth contextualising this against the gross margin of 75% recorded in FY2025: the modest year-over-year compression reflects the ongoing evolution of product mix toward rack-scale and system-level configurations — platforms that carry incremental networking, memory, and integration content — rather than any structural deterioration in pricing power, supply-chain discipline, or competitive positioning. The operating (EBIT) margin for FY2026 reached 60.4%, an exceptional outcome that reflects the company's disciplined approach to operating expense management even as research and development spending reached $18,497 million in absolute terms. That R&D commitment is essential to sustaining the architectural and software-ecosystem lead that underpins NVDA's pricing authority across both the data-centre training and AI-inference markets. At the net income line, a margin of 55.6% in FY2026 illustrates the full compounding effect of industry-leading gross margins, powerful operating leverage, and a manageable below-the-line cost structure — a combination that is extraordinarily difficult for challengers to replicate.

The translation of reported earnings into cash is a hallmark of NVDA's business model and a primary quality metric for institutional analysis. Net income for FY2026 was $120,067 million — a figure that in its own right exceeds the total annual revenues of most semiconductor peers. Free cash flow for the same period reached $96,676 million, representing a year-over-year increase of 58.9% and underscoring the asset-light character of the core compute-platform business, where competitive advantage accrues through software, architecture, and developer ecosystems rather than through owned wafer fabrication. Cash from operations came in at $102,718 million, from which capital expenditures of $6,042 million were deducted to arrive at the free cash flow figure. The proximity of operating cash flow to reported net income signals very high earnings quality: working-capital dynamics, accrual timing, and non-cash charges are not materially inflating reported profitability relative to the cash the enterprise actually generates. This degree of cash conversion is a structural feature of the fabless model and gives management substantial capital-allocation optionality — both to reinvest in architectural leadership and to return capital to shareholders without reliance on debt financing or balance-sheet leverage.

That capital-return optionality was exercised at significant scale in FY2026, with share repurchases totalling $40,086 million. The magnitude of buyback activity relative to free cash flow illustrates management's conviction in the durability of the demand environment and in the long-term intrinsic value of the franchise, and it provides a meaningful per-share earnings tailwind that complements the organic earnings growth profile. On the balance sheet, cash and equivalents stood at $10,605 million at fiscal year-end — a figure that, while modest in isolation relative to the annual free cash flow run-rate, reflects the active and intentional deployment of liquidity into both shareholder return programmes and strategic investment rather than passive accumulation. Total assets of $206,803 million and total equity of $157,293 million together point to a conservatively financed balance sheet carrying limited financial leverage, a structure that provides meaningful cyclical resilience and preserves the company's capacity to invest counter-cyclically should demand conditions moderate. Capital expenditures of $6,042 million measured against revenues of $215,938 million represent an exceptionally low fixed-asset intensity for a company operating at this revenue scale, reinforcing the view that NVDA's enduring competitive advantages are rooted in intellectual property, CUDA software ecosystem depth, and network effects — not in capital-intensive manufacturing assets.

Margin Trend

Revenue Growth

Capital Return

Business Risks

NVIDIA's business risk profile is materially shaped by a high degree of customer concentration that creates structural revenue vulnerability. For fiscal year 2026, sales to a single direct customer accounted for 22% of total revenue, while a second direct customer represented 14% of total revenue, with both concentrations primarily attributable to the Compute & Networking segment. Beyond these disclosed direct relationships, NVIDIA estimates that certain indirect customers — including at least one AI research and deployment company channeling purchases through cloud-service providers — individually represented 10% or more of total revenue. The accounts receivable balance as of January 25, 2026 reflected even steeper concentration, with three direct customers accounting for 25%, 18%, and 13% of that balance respectively. This degree of dependence on a small number of partners means that any deterioration in a key relationship — whether from customer defection, financial distress, self-supply strategies, or regulatory restriction — could produce abrupt and severe top-line damage without ready substitution. The risk is self-reinforcing: losing a hyperscaler or large cloud customer not only eliminates direct revenue but may also foreclose the partner ecosystems and developer channels that underpin future design wins.

NVIDIA's operational continuity rests on a geographically concentrated, third-party supply chain that it does not control and cannot easily substitute. The company depends on a limited number of foundries — principally TSMC and Samsung — to manufacture semiconductor wafers, while memory procurement is sourced from SK Hynix, Micron, and Samsung. Downstream assembly, testing, and packaging are performed by independent contract manufacturers including Foxconn, Wistron, and Fabrinet, with the supply chain heavily weighted toward Asia. NVIDIA has no guarantee of capacity or supply from these partners, and a decommitment, natural disaster, geopolitical disruption, or yield failure at any major node in this chain could cause cascading shortfalls. Manufacturing lead times have historically exceeded 12 months, compelling the company to build finished-goods inventory and lock in purchase commitments well in advance of actual demand. These non-cancellable purchase obligations have grown and are expected to grow as a greater share of total supply, amplifying the financial consequences of demand misestimation. A significant mismatch between anticipated and realized demand — which NVIDIA has experienced in prior periods — can trigger inventory write-downs, cancellation penalties, and impairments. The transition to new architectures illustrated this dynamic concretely: gross margins in Q2 FY2025 were negatively impacted by inventory provisions for low-yielding Blackwell material, underscoring the compounding effect of yield risk on top of volume-planning uncertainty.

Perhaps the most strategically consequential risk NVIDIA faces is its effective foreclosure from China's data center compute market. Under the current geopolitical and regulatory landscape, NVIDIA is unable to design a competitive China data center product that simultaneously satisfies U.S. government export-control requirements and Chinese government approval — an impossible dual constraint that leaves it without a viable path to serve one of the world's largest AI infrastructure markets. As of the end of fiscal year 2026, this foreclosure was operative, and the Chinese government has actively encouraged domestic customers to purchase from NVIDIA's Chinese competitors while discouraging procurement of NVIDIA's data center products, including any China-specific variant designed to comply with U.S. export controls. The direct financial cost of this regulatory exposure has already been recognized: in April 2025, when U.S. authorities required a license to export H20 integrated circuits to China, Hong Kong, and Macau, NVIDIA incurred a $4.5 billion charge in the first quarter of fiscal year 2026 associated with excess H20 inventory and purchase obligations as customer demand for those products collapsed. Subsequent partial relief was limited in scope: the August 2025 licensing window allowed NVIDIA to generate approximately $60 million in H20 revenue — a fraction of the charge absorbed — and a February 2026 license for small H200 shipments to specific Chinese customers imposed a 25% tariff on those units, a cost that NVIDIA may be unable to pass through to customers. Beyond these direct financial hits, NVIDIA's protracted exclusion from China's market is enabling Chinese and other non-U.S. competitors to deepen their developer ecosystems and customer relationships in a market NVIDIA cannot contest, creating long-term erosion of its global competitive standing that persists even if export restrictions are eventually relaxed.

The competitive environment across all of NVIDIA's target markets is intensely contested and, by management's own assessment, is expected to become increasingly so. Competition spans hardware and software dimensions: AMD, Intel, and Huawei directly challenge NVIDIA in GPU and accelerated-compute products, while large cloud providers — including Alphabet, Amazon, Microsoft, Alibaba, Baidu, and Huawei — pursue internal silicon programs to reduce or eliminate dependence on NVIDIA's hardware. SoC competition encompasses Ambarella, Broadcom, Qualcomm, Renesas, and Samsung in embedded and automotive applications, with Tesla developing internal solutions for its own deployments. In networking, NVIDIA contends with AMD, Arista, Broadcom, Cisco, HPE, Intel, and Marvell, among others. A structural disadvantage exists in that some competitors operate their own fabrication facilities, carry larger IP portfolios, enjoy broader distribution networks, and can leverage captive supply to procure foundry capacity and critical input materials during constrained periods — directly threatening NVIDIA's ability to secure manufacturing slots in tight markets. Furthermore, the company's most significant customers are simultaneously among its most formidable potential competitors: cloud service providers with in-house semiconductor expertise can develop their own accelerator solutions, eroding NVIDIA's addressable market at the hyperscaler tier where its revenue is most concentrated. Competitive pressure has already been acknowledged to result in lower-than-expected selling prices or demand, and new market entrants with cheaper or more feature-rich products are anticipated. Taken together, customer concentration, supply chain fragility, geopolitical foreclosure, and an intensifying competitive field represent an interlocking set of risks capable of amplifying one another: a regulatory restriction can accelerate customer self-supply programs, supply chain disruptions can cede market share to better-capitalized competitors, and concentration in a small number of hyperscaler relationships magnifies the consequences of any single competitive displacement.

• Concentration · customer fy26 Customer concentration FY2026: sales to one direct customer represented 22% of total revenue and sales to another direct customer represented 14% of total revenue, primarily attributable to the Compute & Networking segment. NVIDIA also estimates that some indirect customers individually represented 10% or more of revenue, and that one AI research and deployment company contributed a meaningful amount of revenue indirectly through cloud-service customers. Three direct customers accounted for 25%, 18% and 13% of the accounts receivable balance as of January 25, 2026.
• Concentration · key suppliers Supply chain is mainly concentrated in Asia, with expansion into the U.S. and Latin America. Foundries: Taiwan Semiconductor Manufacturing Company Limited (TSMC) and Samsung Electronics Co., Ltd. (Samsung) produce semiconductor wafers. Memory is purchased from SK Hynix Inc., Micron Technology, Inc., and Samsung. Semiconductor packaging utilizes CoWoS technology. Assembly, testing and packaging of final products is performed by independent subcontractors / contract manufacturers including Hon Hai Precision Industry Co., Ltd. (Foxconn), Wistron Corporation, and Fabrinet.
• Competitive · intensity The market for our products is intensely competitive and is characterized by rapid technological change and evolving industry standards. We expect an increasingly competitive environment in the future.
• Competitive · factors We believe that the principal competitive factors in this market are performance, breadth of product offerings, access to customers and partners and distribution channels, software support, conformity to industry standard APIs, manufacturing capabilities, processor pricing, and total system costs.
• Competitive · lower priced entrants We expect competition to increase from both existing competitors and new market entrants with products that may be lower priced than ours or may provide better performance or additional features not provided by our products. In addition, it is possible that new competitors or alliances among competitors could emerge and acquire significant market share.
• Competitive · sources of competition A significant source of competition comes from companies that provide or intend to provide GPUs, CPUs, DPUs, embedded SoCs, and other accelerated, AI computing processor products, and providers of semiconductor-based high-performance interconnect products based on InfiniBand, Ethernet, Fibre Channel, and proprietary technologies. Some competitors may have greater marketing, financial, distribution and manufacturing resources than we do.
• Competitive · competitors ai compute Current competitors include suppliers and licensors of hardware and software for discrete and integrated GPUs, custom chips and other accelerated computing solutions, including solutions offered for AI, such as Advanced Micro Devices, Inc. (AMD), Huawei Technologies Co. Ltd. (Huawei), and Intel Corporation (Intel).
• Competitive · competitors cloud internal Competitors include large cloud services companies with internal teams designing hardware and software that incorporate accelerated or AI computing functionality as part of their internal solutions or platforms, such as Alibaba Group, Alphabet Inc., Amazon, Inc. (Amazon), Baidu, Inc., Huawei, and Microsoft Corporation (Microsoft).
• Competitive · competitors cpu Competitors include suppliers of Arm-based CPUs and companies that incorporate hardware and software for CPUs as part of their internal solutions or platforms, such as Amazon, Huawei, and Microsoft.
• Competitive · competitors soc Competitors include suppliers of hardware and software for SoC products used in servers or embedded into automobiles, autonomous machines, and gaming devices, such as Ambarella, Inc., AMD, Broadcom, Intel, Qualcomm Incorporated, Renesas Electronics Corporation, and Samsung, or companies with internal teams designing SoC products for their own products and services, such as Tesla, Inc.
• Competitive · competitors networking Networking competitors (switches, network adapters including DPUs, and cable solutions including optical modules) include AMD, Arista Networks, Broadcom, Cisco Systems, Inc., Hewlett Packard Enterprise Company, Huawei, Intel, Lumentum Holdings Inc., and Marvell Technology, Inc., as well as internal teams of system vendors and large cloud services companies.
• Competitive · china foreclosure As of the end of fiscal year 2026, we were effectively foreclosed from competing in China's data center computing/compute market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide. Under current rules we are unable to create and deliver a competitive product for China's data center market that receives approval from both the USG and the Chinese government.
• Competitive · china government favors rivals The Chinese government has encouraged customers to purchase from our China-based competitors and discouraged customers from purchasing, importing, or using our data center products, including any China-specific product designed to comply with U.S. export controls.
• Competitive · export controls position harm Our competitive position has been harmed by export controls, and our competitive position and future results will be further harmed, over the long term, if the restrictions remain in place or are expanded in geographic, customer, or product scope, if customers purchase product from competitors, or if customers develop their own internal solution.
• Competitive · cloud self supply Our partners or customers may develop their own solutions; our customers may purchase products from our competitors; and our partners may discontinue sales or lose market share in the markets for which they purchase our products, all of which may alter partners' or customers' purchasing patterns.
• Competitive · china foreclosure q1fy27 As of the end of the first quarter of fiscal year 2027, while we were able to ship uncontrolled products to China, such as gaming and workstation GPUs, we were effectively foreclosed from competing in China's data center computing/compute market, and our effective foreclosure from the China market helped our competitors build larger developer and customer ecosystems to challenge us worldwide.
• Business · technology transition Accelerated computing platforms experience rapid changes in technology, customer requirements, competitive products, and industry standards; failure to timely identify industry changes or develop new products that meet evolving market needs, including disruptive innovations that could render products incompatible, may adversely affect financial results.
• Business · design win dependency Obtaining design wins involves a lengthy process and depends on anticipating features customers will demand; design wins do not guarantee revenue, and failure to obtain one may prevent obtaining future design wins in subsequent generations.
• Business · new business models Launching new offerings with new business models including software, services, and cloud (SaaS/IaaS/PaaS) solutions, and investments in markets with limited operating history may not produce meaningful revenue for several years, if at all; standalone software (NVIDIA AI Enterprise, Omniverse, DRIVE) may not be successful and may not achieve significant revenue.
• Business · groq ip license Entered into an IP license arrangement with Groq, Inc. requiring significant, nonrefundable payments; incorporating the licensed technology requires significant engineering effort and may not occur on expected timelines or at all, may not achieve adoption, and the company may be unable to recover associated costs or realize an adequate return.
• Business · cloud service agreements Entered into multi-year cloud service agreements to support R&D; the timing and availability of these cloud services have changed and may continue to shift, impacting revenue, expenses, and development timelines, and these arrangements may not deliver anticipated benefits.
• Market · competition Target markets remain competitive and competition may intensify with new entrants and consolidations; competitors' products may be cheaper or provide better functionality or features, which has resulted and may result in lower-than-expected selling prices or demand for products.
• Market · competitor scale and fabs Some competitors operate their own fabrication facilities and have longer operating histories, larger customer bases, more comprehensive IP portfolios, more design wins, and greater financial, sales, marketing and distribution resources, and may impinge on the company's ability to procure foundry capacity and scarce input materials during a supply-constrained environment.
• Market · customer in house silicon Some customers have in-house expertise and internal development capabilities similar to the company's and can use or develop their own solutions to replace those NVIDIA provides; others may offer cloud-based services that compete with NVIDIA's AI cloud service offerings.
• Operational · long lead times Long manufacturing lead times require building finished products and maintaining inventory in advance of anticipated demand; the company has previously experienced and may continue to experience extended lead times of more than 12 months, and has paid premiums, provided deposits, and entered long-term supply and capacity commitments, increasing product costs.
• Operational · supply demand mismatch Inaccurate demand estimates or customer order changes may prevent reducing supply commitments in time; significant mismatches between supply and demand have resulted in both product shortages and excess inventory and have significantly harmed financial results and could reoccur.
• Operational · supplier foundry dependency Depends on foundries to manufacture semiconductor wafers and on independent subcontractors to assemble, test, and package products, reducing control over product quantity, quality, manufacturing yields, and delivery schedules; risks include lack of guaranteed supply/capacity, supplier decommitment, and a limited number and geographic concentration of global suppliers, foundries, and memory manufacturers.
• Operational · manufacturing yield Low manufacturing yields can result from issues in product design or a foundry's proprietary process technology, and foundries may fail to develop or implement high-quality process technologies including transitions to smaller geometry/advanced process nodes needed to manufacture products.
• Operational · product defects Complex hardware and software products have in the past and may in the future contain defects, security vulnerabilities, or unsatisfactory performance; defects have caused and could cause significant warranty, support, and repair/replacement costs, product recalls, inventory write-offs, loss of design wins, and harm to reputation and financial results.
• Financial · inventory writedown If demand is overestimated or customers cancel/defer orders, the company may be unable to utilize on-hand inventory or reduce purchase commitments and has had to write down inventory, incur cancellation penalties, and record impairments, and may have to do so in the future.
• Financial · purchase obligation growth Risks are amplified by non-cancellable, non-returnable purchase orders placed in advance of historical lead times and have increased and may continue to increase as purchase obligations and prepaids have grown and are expected to grow as a greater portion of total supply, which may negatively impact gross margins and financial results.
• Financial · gross margin inventory provision Increased frequency and complexity of product transitions could result in unanticipated quality or production issues increasing the magnitude of inventory provisions; for example, gross margins in Q2 FY2025 were negatively impacted by inventory provisions for low-yielding Blackwell material.
• Geopolitical · international concentration International sales and operations are a significant part of NVIDIA's business; semiconductor wafers are manufactured, assembled, tested and packaged by third parties outside the U.S., and 31% of fiscal year 2026 revenue was generated from sales outside the United States, exposing it to economic, political, currency, trade and IP-protection risks.
• Geopolitical · china foreclosure Under current rules and geopolitical landscape NVIDIA is unable to create a competitive China data-center product that wins approval from both the USG and the Chinese government; as of the end of fiscal year 2026 it was effectively foreclosed from competing in China's data center compute market, helping competitors build larger ecosystems, with a material adverse impact on its business.
• Geopolitical · taiwan korea supply NVIDIA's business depends on consistent and reliable supply from overseas partners, especially in Taiwan and South Korea; any new restrictions negatively impacting its ability to receive components, parts, or services from Taiwan and South Korea would negatively impact its business and financial results.
• Geopolitical · israel operations NVIDIA is monitoring the geopolitical conflict in and around Israel, where it has approximately 6,000 employees primarily supporting R&D, operations, and sales/marketing of networking products; extension or expansion of the conflict could impact future product development, operations, and revenue. Other countries have restricted business with Israel via economic boycotts.
• Regulatory · export controls h20 charge In April 2025 the USG required a license to export H20 integrated circuits (and any circuits achieving the H20's memory bandwidth, interconnect bandwidth, or combination thereof) to China (including Hong Kong and Macau) and D:5 countries; as a result NVIDIA incurred a $4.5 billion charge in the first quarter of fiscal year 2026 associated with H20 for excess inventory and purchase obligations as demand diminished.
• Regulatory · h20 license revenue In August 2025 the USG granted licenses allowing NVIDIA to ship certain H20 products to certain China-based customers; NVIDIA generated approximately $60 million in H20 revenue under those licenses.
• Regulatory · usg revenue share USG officials expressed an expectation that the USG will receive 15% or more of the revenue generated from licensed sales of NVIDIA's products, but the USG did not publish a regulation codifying such requirement.
• Regulatory · h200 license tariff In February 2026 the USG granted a license to ship small amounts of H200 products to specific China-based customers; no revenue has been generated and shipments require a U.S. inspection process, so any H200 shipped under the program will be subject to a 25% tariff upon importation into the United States, which NVIDIA may be unable to pass to customers.
• Regulatory · ai diffusion ifr In January 2025 the USG published the AI Diffusion IFR, which would have imposed a worldwide licensing requirement on data center products such as H200, GB200 and GB300, dividing the world into three tiers (most countries to Tier 2) with a complex, burdensome licensing scheme; in May 2025 the USG said it would rescind it and implement an uncertain replacement rule that could materially impact NVIDIA.
• Regulatory · export controls general NVIDIA is subject to a series of shifting and expanding U.S. export control restrictions targeting GPUs and AI semiconductors that have been and are likely to be more restrictive, prohibiting exports to some or all customers in one or more markets, disrupting supply and distribution chains (including products warehoused in Hong Kong), and disproportionately harming NVIDIA's competitive position versus competitors outside the controls' scope.
• Regulatory · china mellanox finding On September 15, 2025 China's antitrust regulators published a preliminary finding that NVIDIA's compliance with U.S. export controls (requiring degraded products for China) discriminated unfairly against Chinese customers and violated the terms of China's approval of the Mellanox acquisition; an adverse conclusion could lead to financial penalties or restrictions on its networking business in China.
• Regulatory · responsible ai Issues relating to the responsible use of NVIDIA's technologies, including AI, may cause reputational or financial harm and liability; AI poses emerging legal, social and ethical risks, and complying with multiple AI regulations across jurisdictions (e.g., the EU and EU AI Act) may increase costs, change operations, and impede offering products in certain jurisdictions.
• Regulatory · data privacy NVIDIA is subject to stringent and changing data privacy and security laws (e.g., CCPA, GDPR, U.K. GDPR, NIS2, China data-localization/CAC rules); GDPR non-compliance can incur penalties up to 20 million euros or 4% of worldwide revenue, and NIS2 fines up to 10 million euros or 2% of worldwide revenue, with cross-border transfer restrictions that could harm operations.
• Regulatory · sustainability esg Scrutiny from shareholders, regulators and others regarding NVIDIA's corporate sustainability practices and disclosures (including climate change and human rights) could result in additional costs or risks, fail to meet evolving or state-level expectations or disclosure standards, harm its reputation, and expose it to liability.
• Legal · ip protection Protecting NVIDIA's IP rights could be costly and its ability to compete could be harmed; it faces patent-infringement and IP-rights lawsuits where unfavorable rulings could mean significant damages, patent invalidation, lost profits or injunctive relief, and foreign laws may not protect its products or IP to the same extent as U.S. law, making piracy more likely.
• Legal · litigation NVIDIA faces the burden and risks of litigation, investigations and regulatory proceedings, including a securities class action alleging it and certain officers made false or misleading statements about channel inventory and the impact of cryptocurrency mining on GPU demand in 2017 and 2018; adverse outcomes could result in monetary damages, fines, or injunctions and a material adverse impact.
• Legal · change of control Delaware law and provisions in NVIDIA's certificate of incorporation, bylaws, and its agreement with Microsoft for the Xbox could delay or prevent a change in control; under the Microsoft agreement, if someone offers to purchase at least 30% of outstanding common stock, Microsoft may have first and last rights of refusal to purchase the stock.
• Operational · cybersecurity Product, system security and data protection incidents or breaches, as well as cyber-attacks (including ransomware by organized criminals, nation-states, and AI-enabled attacks), could disrupt NVIDIA's operations, reduce revenue, increase expenses and harm its reputation; it has experienced security incidents (e.g., GFN breaches exposing third-party game source code) and may face government enforcement and liability.
• Operational · business disruption Business disruptions (natural disasters, extreme weather, power/water shortages, infrastructure or telecom failures, terrorism, war, pandemics, regulatory changes) could harm operations and revenue; NVIDIA's headquarters, much data center capacity and R&D are in California while inventory and suppliers are concentrated in Asia, with geopolitical tensions in China, Hong Kong, Israel, Korea and Taiwan posing further disruption risk.
• Operational · information systems NVIDIA's business depends on the proper functioning of business processes and information systems; modification or interruption, including problems from implementing a new ERP system, may disrupt its business and internal controls (including SOX Section 404), be costly, and impair its ability to timely and accurately report financial results.
• Operational · climate change Climate change may have a long-term adverse impact on NVIDIA's business; water and energy availability is critical and facilities (e.g., Santa Clara HQ and data centers) may be vulnerable to extreme weather and wildfire-related power safety shut-offs, while sustainability-related laws and lawsuits could raise costs, and insurance may not cover resulting losses.
• Financial · tax NVIDIA's operating results may be adversely impacted by additional tax liabilities, higher than expected tax rates, and changes in tax laws; most income is taxable in the U.S. with a significant portion qualifying as FDDEI, so higher U.S. rates or reduced FDDEI deduction, plus OECD Pillar One/Pillar Two implementation, could raise its effective tax rate and harm financial results.
• Financial · results fluctuation NVIDIA's operating results have fluctuated and may continue to fluctuate due to factors such as multi-year development cycles, extended customer payment terms and customer bad-debt risk, vendor payment requirements, and accounting-standard changes; if results fall below securities analyst or investor expectations, the stock price could decline or experience substantial volatility.
• Financial · counterparty Commercial arrangements, including long-term capacity purchase obligations and financial guarantees, expose NVIDIA to counterparty risk such as customers' or partners' inability to fulfill financial commitments, secure financing, project delays, and counterparty financial distress or insolvency; any financing arrangements undertaken could lower upfront cash flows and increase credit risk.
• Business · customer concentration NVIDIA receives a significant amount of revenue from a limited number of partners and distributors; for fiscal year 2026 sales to one direct customer represented 22% of total revenue and another direct customer 14% (primarily Compute & Networking), with some indirect customers individually estimated at 10% or more of revenue; losing or being prevented from selling to such customers would likely harm results.
• Business · key personnel If NVIDIA is unable to attract, retain and motivate its executives and key employees, its business may be harmed; it faces a highly competitive market for skilled workers, immigration/work-permit risks, rising cash and stock-based compensation costs, reduced retentive value of equity during stock-price volatility, and dependence on its longstanding executive team and effective succession planning.
• Business · acquisition integration NVIDIA may be unable to realize the benefits of business investments or acquisitions or to successfully integrate acquired companies; risks include difficulty integrating technology/systems/employees, assumed liabilities and goodwill impairment, failure to obtain regulatory approvals or imposed conditions, dilutive equity or debt issuance, and impairments of illiquid strategic investments.

Financial Risks

NVIDIA faces meaningful financial risks concentrated in three interconnected areas: inventory exposure amplified by non-cancellable supply commitments, regulatory-driven one-time charges arising from export control actions, and structural tax vulnerability tied to the composition of its income.

NVIDIA's fabless model requires placing purchase orders well in advance of delivery, with manufacturing lead times exceeding 12 months for certain products. When demand is overestimated or customers cancel or defer orders, the company has been unable to unwind supply commitments in time, has had to write down inventory, incur cancellation penalties, and record impairments, and may have to do so again in the future. These risks are further amplified by non-cancellable, non-returnable purchase orders placed ahead of historical lead times that have grown substantially and are expected to represent a larger share of total supply, with the potential to negatively impact gross margins and overall financial results. Rapid product transitions compound this exposure: gross margins in Q2 FY2025 were negatively impacted by inventory provisions specifically linked to low-yielding Blackwell material, illustrating how quality or yield issues during a generation changeover can translate directly into impairment charges. The balance sheet reflects the scale of this inventory risk: inventories stood at $21,403 million at FY2026 fiscal year-end, and at $19,784 million as of Q3 FY2026, representing a vast pool of work-in-process and finished goods that could require material write-downs if demand falters or a major product transition encounters quality problems.

The single most visible crystallization of financial risk in the recent period was the $4.5 billion charge incurred in Q1 FY2026 associated with H20 integrated circuits, following the U.S. government's April 2025 decision to require an export license for H20 chips—and any chips meeting H20's memory bandwidth, interconnect bandwidth, or combination thresholds—destined for China, Hong Kong, Macau, and D:5 countries. That charge arose directly from excess inventory and purchase obligations as customer demand for H20 diminished following the licensing requirement. A subsequent August 2025 license allowed NVIDIA to generate approximately $60 million in H20 revenue from certain China-based customers, a figure that illustrates how marginal the licensed market is relative to the initial impairment. Additional policy risk lingers: U.S. government officials expressed an expectation that they would receive 15% or more of revenue generated from licensed sales of NVIDIA products, a requirement that has not been codified into regulation but could impose a substantial de facto levy on any future approved sales. The February 2026 H200 program imposes a 25% tariff on any H200 units returned to the United States following the required U.S. inspection process, a cost NVIDIA may be unable to pass through to customers. These developments sit within a broader geopolitical reality in which NVIDIA is now effectively foreclosed from competing in China's data-center compute market under current rules, at a time when 31% of fiscal year 2026 revenue was generated from sales outside the United States—meaning export-control restrictions directly threaten a material portion of the company's revenue base and can generate sudden large charges when policy shifts abruptly.

NVIDIA's tax position carries structural risk that could raise the effective tax rate materially if the legislative and regulatory environment shifts. Most of NVIDIA's income is taxable in the United States, with a significant portion qualifying as FDDEI (Foreign-Derived Deduction Eligible Income), so any increase in U.S. corporate tax rates or reduction in the FDDEI deduction would directly increase the tax burden. Implementation of OECD Pillar One and Pillar Two frameworks represents an additional vector through which the effective rate could rise and harm financial results. On the debt side, NVIDIA carries long-term debt of $7,469 million as of FY2026, spread across fixed-rate notes maturing across an extended horizon: $1,000 million due in 2026, $1,250 million due in 2028, $1,500 million due in 2030, $1,250 million due in 2031, $1,000 million due in 2040, $2,000 million due in 2050, and $500 million due in 2060. While total equity of $157,293 million at FY2026 year-end provides substantial headroom, orderly refinancing and management of tax outcomes will remain key financial risk variables. Counterparty risk—arising from long-term capacity purchase obligations and financial guarantees with customers or partners who may be unable to fulfill their financial commitments, secure financing, or complete projects on schedule—adds further uncertainty to near-term cash flow visibility and could lower upfront cash flows while increasing credit risk if financing arrangements must be restructured.

• Business · technology transition Accelerated computing platforms experience rapid changes in technology, customer requirements, competitive products, and industry standards; failure to timely identify industry changes or develop new products that meet evolving market needs, including disruptive innovations that could render products incompatible, may adversely affect financial results.
• Business · design win dependency Obtaining design wins involves a lengthy process and depends on anticipating features customers will demand; design wins do not guarantee revenue, and failure to obtain one may prevent obtaining future design wins in subsequent generations.
• Business · new business models Launching new offerings with new business models including software, services, and cloud (SaaS/IaaS/PaaS) solutions, and investments in markets with limited operating history may not produce meaningful revenue for several years, if at all; standalone software (NVIDIA AI Enterprise, Omniverse, DRIVE) may not be successful and may not achieve significant revenue.
• Business · groq ip license Entered into an IP license arrangement with Groq, Inc. requiring significant, nonrefundable payments; incorporating the licensed technology requires significant engineering effort and may not occur on expected timelines or at all, may not achieve adoption, and the company may be unable to recover associated costs or realize an adequate return.
• Business · cloud service agreements Entered into multi-year cloud service agreements to support R&D; the timing and availability of these cloud services have changed and may continue to shift, impacting revenue, expenses, and development timelines, and these arrangements may not deliver anticipated benefits.
• Market · competition Target markets remain competitive and competition may intensify with new entrants and consolidations; competitors' products may be cheaper or provide better functionality or features, which has resulted and may result in lower-than-expected selling prices or demand for products.
• Market · competitor scale and fabs Some competitors operate their own fabrication facilities and have longer operating histories, larger customer bases, more comprehensive IP portfolios, more design wins, and greater financial, sales, marketing and distribution resources, and may impinge on the company's ability to procure foundry capacity and scarce input materials during a supply-constrained environment.
• Market · customer in house silicon Some customers have in-house expertise and internal development capabilities similar to the company's and can use or develop their own solutions to replace those NVIDIA provides; others may offer cloud-based services that compete with NVIDIA's AI cloud service offerings.
• Operational · long lead times Long manufacturing lead times require building finished products and maintaining inventory in advance of anticipated demand; the company has previously experienced and may continue to experience extended lead times of more than 12 months, and has paid premiums, provided deposits, and entered long-term supply and capacity commitments, increasing product costs.
• Operational · supply demand mismatch Inaccurate demand estimates or customer order changes may prevent reducing supply commitments in time; significant mismatches between supply and demand have resulted in both product shortages and excess inventory and have significantly harmed financial results and could reoccur.
• Operational · supplier foundry dependency Depends on foundries to manufacture semiconductor wafers and on independent subcontractors to assemble, test, and package products, reducing control over product quantity, quality, manufacturing yields, and delivery schedules; risks include lack of guaranteed supply/capacity, supplier decommitment, and a limited number and geographic concentration of global suppliers, foundries, and memory manufacturers.
• Operational · manufacturing yield Low manufacturing yields can result from issues in product design or a foundry's proprietary process technology, and foundries may fail to develop or implement high-quality process technologies including transitions to smaller geometry/advanced process nodes needed to manufacture products.
• Operational · product defects Complex hardware and software products have in the past and may in the future contain defects, security vulnerabilities, or unsatisfactory performance; defects have caused and could cause significant warranty, support, and repair/replacement costs, product recalls, inventory write-offs, loss of design wins, and harm to reputation and financial results.
• Financial · inventory writedown If demand is overestimated or customers cancel/defer orders, the company may be unable to utilize on-hand inventory or reduce purchase commitments and has had to write down inventory, incur cancellation penalties, and record impairments, and may have to do so in the future.
• Financial · purchase obligation growth Risks are amplified by non-cancellable, non-returnable purchase orders placed in advance of historical lead times and have increased and may continue to increase as purchase obligations and prepaids have grown and are expected to grow as a greater portion of total supply, which may negatively impact gross margins and financial results.
• Financial · gross margin inventory provision Increased frequency and complexity of product transitions could result in unanticipated quality or production issues increasing the magnitude of inventory provisions; for example, gross margins in Q2 FY2025 were negatively impacted by inventory provisions for low-yielding Blackwell material.
• Geopolitical · international concentration International sales and operations are a significant part of NVIDIA's business; semiconductor wafers are manufactured, assembled, tested and packaged by third parties outside the U.S., and 31% of fiscal year 2026 revenue was generated from sales outside the United States, exposing it to economic, political, currency, trade and IP-protection risks.
• Geopolitical · china foreclosure Under current rules and geopolitical landscape NVIDIA is unable to create a competitive China data-center product that wins approval from both the USG and the Chinese government; as of the end of fiscal year 2026 it was effectively foreclosed from competing in China's data center compute market, helping competitors build larger ecosystems, with a material adverse impact on its business.
• Geopolitical · taiwan korea supply NVIDIA's business depends on consistent and reliable supply from overseas partners, especially in Taiwan and South Korea; any new restrictions negatively impacting its ability to receive components, parts, or services from Taiwan and South Korea would negatively impact its business and financial results.
• Geopolitical · israel operations NVIDIA is monitoring the geopolitical conflict in and around Israel, where it has approximately 6,000 employees primarily supporting R&D, operations, and sales/marketing of networking products; extension or expansion of the conflict could impact future product development, operations, and revenue. Other countries have restricted business with Israel via economic boycotts.
• Regulatory · export controls h20 charge In April 2025 the USG required a license to export H20 integrated circuits (and any circuits achieving the H20's memory bandwidth, interconnect bandwidth, or combination thereof) to China (including Hong Kong and Macau) and D:5 countries; as a result NVIDIA incurred a $4.5 billion charge in the first quarter of fiscal year 2026 associated with H20 for excess inventory and purchase obligations as demand diminished.
• Regulatory · h20 license revenue In August 2025 the USG granted licenses allowing NVIDIA to ship certain H20 products to certain China-based customers; NVIDIA generated approximately $60 million in H20 revenue under those licenses.
• Regulatory · usg revenue share USG officials expressed an expectation that the USG will receive 15% or more of the revenue generated from licensed sales of NVIDIA's products, but the USG did not publish a regulation codifying such requirement.
• Regulatory · h200 license tariff In February 2026 the USG granted a license to ship small amounts of H200 products to specific China-based customers; no revenue has been generated and shipments require a U.S. inspection process, so any H200 shipped under the program will be subject to a 25% tariff upon importation into the United States, which NVIDIA may be unable to pass to customers.
• Regulatory · ai diffusion ifr In January 2025 the USG published the AI Diffusion IFR, which would have imposed a worldwide licensing requirement on data center products such as H200, GB200 and GB300, dividing the world into three tiers (most countries to Tier 2) with a complex, burdensome licensing scheme; in May 2025 the USG said it would rescind it and implement an uncertain replacement rule that could materially impact NVIDIA.
• Regulatory · export controls general NVIDIA is subject to a series of shifting and expanding U.S. export control restrictions targeting GPUs and AI semiconductors that have been and are likely to be more restrictive, prohibiting exports to some or all customers in one or more markets, disrupting supply and distribution chains (including products warehoused in Hong Kong), and disproportionately harming NVIDIA's competitive position versus competitors outside the controls' scope.
• Regulatory · china mellanox finding On September 15, 2025 China's antitrust regulators published a preliminary finding that NVIDIA's compliance with U.S. export controls (requiring degraded products for China) discriminated unfairly against Chinese customers and violated the terms of China's approval of the Mellanox acquisition; an adverse conclusion could lead to financial penalties or restrictions on its networking business in China.
• Regulatory · responsible ai Issues relating to the responsible use of NVIDIA's technologies, including AI, may cause reputational or financial harm and liability; AI poses emerging legal, social and ethical risks, and complying with multiple AI regulations across jurisdictions (e.g., the EU and EU AI Act) may increase costs, change operations, and impede offering products in certain jurisdictions.
• Regulatory · data privacy NVIDIA is subject to stringent and changing data privacy and security laws (e.g., CCPA, GDPR, U.K. GDPR, NIS2, China data-localization/CAC rules); GDPR non-compliance can incur penalties up to 20 million euros or 4% of worldwide revenue, and NIS2 fines up to 10 million euros or 2% of worldwide revenue, with cross-border transfer restrictions that could harm operations.
• Regulatory · sustainability esg Scrutiny from shareholders, regulators and others regarding NVIDIA's corporate sustainability practices and disclosures (including climate change and human rights) could result in additional costs or risks, fail to meet evolving or state-level expectations or disclosure standards, harm its reputation, and expose it to liability.
• Legal · ip protection Protecting NVIDIA's IP rights could be costly and its ability to compete could be harmed; it faces patent-infringement and IP-rights lawsuits where unfavorable rulings could mean significant damages, patent invalidation, lost profits or injunctive relief, and foreign laws may not protect its products or IP to the same extent as U.S. law, making piracy more likely.
• Legal · litigation NVIDIA faces the burden and risks of litigation, investigations and regulatory proceedings, including a securities class action alleging it and certain officers made false or misleading statements about channel inventory and the impact of cryptocurrency mining on GPU demand in 2017 and 2018; adverse outcomes could result in monetary damages, fines, or injunctions and a material adverse impact.
• Legal · change of control Delaware law and provisions in NVIDIA's certificate of incorporation, bylaws, and its agreement with Microsoft for the Xbox could delay or prevent a change in control; under the Microsoft agreement, if someone offers to purchase at least 30% of outstanding common stock, Microsoft may have first and last rights of refusal to purchase the stock.
• Operational · cybersecurity Product, system security and data protection incidents or breaches, as well as cyber-attacks (including ransomware by organized criminals, nation-states, and AI-enabled attacks), could disrupt NVIDIA's operations, reduce revenue, increase expenses and harm its reputation; it has experienced security incidents (e.g., GFN breaches exposing third-party game source code) and may face government enforcement and liability.
• Operational · business disruption Business disruptions (natural disasters, extreme weather, power/water shortages, infrastructure or telecom failures, terrorism, war, pandemics, regulatory changes) could harm operations and revenue; NVIDIA's headquarters, much data center capacity and R&D are in California while inventory and suppliers are concentrated in Asia, with geopolitical tensions in China, Hong Kong, Israel, Korea and Taiwan posing further disruption risk.
• Operational · information systems NVIDIA's business depends on the proper functioning of business processes and information systems; modification or interruption, including problems from implementing a new ERP system, may disrupt its business and internal controls (including SOX Section 404), be costly, and impair its ability to timely and accurately report financial results.
• Operational · climate change Climate change may have a long-term adverse impact on NVIDIA's business; water and energy availability is critical and facilities (e.g., Santa Clara HQ and data centers) may be vulnerable to extreme weather and wildfire-related power safety shut-offs, while sustainability-related laws and lawsuits could raise costs, and insurance may not cover resulting losses.
• Financial · tax NVIDIA's operating results may be adversely impacted by additional tax liabilities, higher than expected tax rates, and changes in tax laws; most income is taxable in the U.S. with a significant portion qualifying as FDDEI, so higher U.S. rates or reduced FDDEI deduction, plus OECD Pillar One/Pillar Two implementation, could raise its effective tax rate and harm financial results.
• Financial · results fluctuation NVIDIA's operating results have fluctuated and may continue to fluctuate due to factors such as multi-year development cycles, extended customer payment terms and customer bad-debt risk, vendor payment requirements, and accounting-standard changes; if results fall below securities analyst or investor expectations, the stock price could decline or experience substantial volatility.
• Financial · counterparty Commercial arrangements, including long-term capacity purchase obligations and financial guarantees, expose NVIDIA to counterparty risk such as customers' or partners' inability to fulfill financial commitments, secure financing, project delays, and counterparty financial distress or insolvency; any financing arrangements undertaken could lower upfront cash flows and increase credit risk.
• Business · customer concentration NVIDIA receives a significant amount of revenue from a limited number of partners and distributors; for fiscal year 2026 sales to one direct customer represented 22% of total revenue and another direct customer 14% (primarily Compute & Networking), with some indirect customers individually estimated at 10% or more of revenue; losing or being prevented from selling to such customers would likely harm results.
• Business · key personnel If NVIDIA is unable to attract, retain and motivate its executives and key employees, its business may be harmed; it faces a highly competitive market for skilled workers, immigration/work-permit risks, rising cash and stock-based compensation costs, reduced retentive value of equity during stock-price volatility, and dependence on its longstanding executive team and effective succession planning.
• Business · acquisition integration NVIDIA may be unable to realize the benefits of business investments or acquisitions or to successfully integrate acquired companies; risks include difficulty integrating technology/systems/employees, assumed liabilities and goodwill impairment, failure to obtain regulatory approvals or imposed conditions, dilutive equity or debt issuance, and impairments of illiquid strategic investments.

Risk Assessment

NVIDIA's risk profile is anchored by an interlocking set of geopolitical, regulatory, operational, and business hazards that have already produced material financial consequences and are likely to intensify. Perhaps the most structurally significant risk is the company's effective foreclosure from China's data-center compute market. Under current U.S. government rules and the prevailing geopolitical landscape, NVIDIA is unable to create a competitive China data-center product that wins simultaneous approval from both the U.S. government and Chinese authorities—a constraint that is not merely prospective: as of the end of fiscal year 2026 the company was effectively locked out of competing in that market, ceding ground to domestic competitors and aiding those rivals in building larger, more entrenched ecosystems. The financial toll of this dynamic became concrete in April 2025 when the U.S. government imposed a license requirement on exports of H20 integrated circuits to China, including Hong Kong and Macau, and to D:5 countries. That single regulatory action precipitated a $4.5 billion charge in the first quarter of fiscal year 2026 as demand collapsed and NVIDIA was left holding excess inventory and purchase obligations it could not unwind. A partial reprieve followed in August 2025 when the U.S. government granted licenses allowing shipment of certain H20 products to specific China-based customers, but the revenue generated under those licenses amounted to approximately $60 million—a figure that underscores how severely the market has been impaired. A further February 2026 license permitting small quantities of H200 shipments to specific Chinese customers provides little relief; any H200 shipped under that program is subject to a 25% tariff upon re-importation into the United States, a cost NVIDIA may be unable to pass through to customers.

The broader regulatory environment compounds this geopolitical overhang. NVIDIA operates under a series of shifting and expanding U.S. export control restrictions targeting GPUs and AI semiconductors that have been and are likely to become more restrictive, with the potential to prohibit exports to some or all customers in one or more markets, disrupt supply and distribution chains including products warehoused in Hong Kong, and disproportionately harm NVIDIA's competitive position relative to competitors that fall outside the scope of those controls. The January 2025 publication of the AI Diffusion Interim Final Rule would have imposed a worldwide licensing regime dividing markets into tiered categories with complex compliance burdens; although the U.S. government indicated in May 2025 that it would rescind that rule, it simultaneously signaled that a replacement regime of uncertain form would follow, leaving NVIDIA with sustained strategic uncertainty across its addressable market. On the Chinese regulatory front, an additional pressure point emerged on September 15, 2025, when China's antitrust regulators published a preliminary finding that NVIDIA's compliance with U.S. export controls—specifically the requirement to supply degraded products to Chinese customers—discriminated unfairly against those customers and potentially violated conditions attached to China's approval of the Mellanox acquisition; an adverse conclusion could yield financial penalties or restrictions on NVIDIA's networking business in China. U.S. government officials have also expressed an expectation that the U.S. government will receive 15% or more of the revenue generated from licensed sales of NVIDIA products, though this expectation has not been codified in regulation, creating additional ambiguity about the economics of any licensed sales. Beyond export controls, the company faces data privacy obligations including GDPR, under which non-compliance can incur penalties of up to 20 million euros or 4% of worldwide revenue, and NIS2 requirements with fines of up to 10 million euros or 2% of worldwide revenue.

NVIDIA's operational risk is concentrated in its supply chain architecture. The company depends on a limited number of foundries to manufacture semiconductor wafers and on independent subcontractors to assemble, test, and package products, a structure that reduces control over product quantity, quality, manufacturing yields, and delivery schedules and that is compounded by the geographic concentration of global suppliers, foundries, and memory manufacturers. Supply from overseas partners, especially in Taiwan and South Korea, is critical; any new restrictions affecting the ability to receive components, parts, or services from those geographies would directly impair financial results. Long manufacturing lead times require NVIDIA to build finished products and maintain inventory in advance of anticipated demand: the company has previously experienced extended lead times of more than 12 months and has paid premiums, provided deposits, and entered long-term supply and capacity commitments that increase product costs. These commitments cannot easily be reversed—purchase orders are placed in advance of historical lead times on non-cancellable, non-returnable terms—and purchase obligations and prepaids have grown and are expected to continue to grow as a greater portion of total supply, which may negatively impact gross margins and financial results. When demand estimates prove inaccurate, NVIDIA has been forced to write down inventory, incur cancellation penalties, and record impairments; this risk is not hypothetical—gross margins in the second quarter of fiscal year 2025 were negatively impacted by inventory provisions for low-yielding Blackwell material. Additionally, with approximately 6,000 employees in Israel primarily supporting R&D, operations, and sales and marketing of networking products, an extension or expansion of geopolitical conflict in that region could impair product development and revenue.

Revenue concentration and competitive dynamics complete the risk picture. In fiscal year 2026, sales to one direct customer represented 22% of total revenue and sales to another direct customer represented 14% of total revenue, with those revenues concentrated in the Compute & Networking segment; certain indirect customers were individually estimated at 10% or more of revenue. Loss of, or an inability to sell to, any of these customers would likely cause material harm to results. Competitive intensity reinforces this vulnerability: some customers possess in-house expertise and internal development capabilities comparable to NVIDIA's and may develop or deploy their own solutions in place of NVIDIA's products, while certain competitors operate their own fabrication facilities and bring greater financial, sales, marketing, and distribution resources to bear, potentially impinging on NVIDIA's ability to procure foundry capacity during supply-constrained periods. The accelerated computing market experiences rapid technology transitions, and failure to anticipate customer requirements or develop products that meet evolving needs—including disruptive innovations that could render current products incompatible—may adversely affect financial results. International sales represent a meaningful share of the business: 31% of fiscal year 2026 revenue was generated from sales outside the United States, exposing NVIDIA to economic, political, currency, trade, and intellectual property risks across a wide range of jurisdictions. Taken together, these risk categories—geopolitical foreclosure, a volatile and tightening regulatory framework, deep supply chain dependencies, and concentrated revenue exposure—constitute a compound risk structure in which adverse developments in any single dimension can rapidly amplify across the others.

• Business · technology transition Accelerated computing platforms experience rapid changes in technology, customer requirements, competitive products, and industry standards; failure to timely identify industry changes or develop new products that meet evolving market needs, including disruptive innovations that could render products incompatible, may adversely affect financial results.
• Business · design win dependency Obtaining design wins involves a lengthy process and depends on anticipating features customers will demand; design wins do not guarantee revenue, and failure to obtain one may prevent obtaining future design wins in subsequent generations.
• Business · new business models Launching new offerings with new business models including software, services, and cloud (SaaS/IaaS/PaaS) solutions, and investments in markets with limited operating history may not produce meaningful revenue for several years, if at all; standalone software (NVIDIA AI Enterprise, Omniverse, DRIVE) may not be successful and may not achieve significant revenue.
• Business · groq ip license Entered into an IP license arrangement with Groq, Inc. requiring significant, nonrefundable payments; incorporating the licensed technology requires significant engineering effort and may not occur on expected timelines or at all, may not achieve adoption, and the company may be unable to recover associated costs or realize an adequate return.
• Business · cloud service agreements Entered into multi-year cloud service agreements to support R&D; the timing and availability of these cloud services have changed and may continue to shift, impacting revenue, expenses, and development timelines, and these arrangements may not deliver anticipated benefits.
• Market · competition Target markets remain competitive and competition may intensify with new entrants and consolidations; competitors' products may be cheaper or provide better functionality or features, which has resulted and may result in lower-than-expected selling prices or demand for products.
• Market · competitor scale and fabs Some competitors operate their own fabrication facilities and have longer operating histories, larger customer bases, more comprehensive IP portfolios, more design wins, and greater financial, sales, marketing and distribution resources, and may impinge on the company's ability to procure foundry capacity and scarce input materials during a supply-constrained environment.
• Market · customer in house silicon Some customers have in-house expertise and internal development capabilities similar to the company's and can use or develop their own solutions to replace those NVIDIA provides; others may offer cloud-based services that compete with NVIDIA's AI cloud service offerings.
• Operational · long lead times Long manufacturing lead times require building finished products and maintaining inventory in advance of anticipated demand; the company has previously experienced and may continue to experience extended lead times of more than 12 months, and has paid premiums, provided deposits, and entered long-term supply and capacity commitments, increasing product costs.
• Operational · supply demand mismatch Inaccurate demand estimates or customer order changes may prevent reducing supply commitments in time; significant mismatches between supply and demand have resulted in both product shortages and excess inventory and have significantly harmed financial results and could reoccur.
• Operational · supplier foundry dependency Depends on foundries to manufacture semiconductor wafers and on independent subcontractors to assemble, test, and package products, reducing control over product quantity, quality, manufacturing yields, and delivery schedules; risks include lack of guaranteed supply/capacity, supplier decommitment, and a limited number and geographic concentration of global suppliers, foundries, and memory manufacturers.
• Operational · manufacturing yield Low manufacturing yields can result from issues in product design or a foundry's proprietary process technology, and foundries may fail to develop or implement high-quality process technologies including transitions to smaller geometry/advanced process nodes needed to manufacture products.
• Operational · product defects Complex hardware and software products have in the past and may in the future contain defects, security vulnerabilities, or unsatisfactory performance; defects have caused and could cause significant warranty, support, and repair/replacement costs, product recalls, inventory write-offs, loss of design wins, and harm to reputation and financial results.
• Financial · inventory writedown If demand is overestimated or customers cancel/defer orders, the company may be unable to utilize on-hand inventory or reduce purchase commitments and has had to write down inventory, incur cancellation penalties, and record impairments, and may have to do so in the future.
• Financial · purchase obligation growth Risks are amplified by non-cancellable, non-returnable purchase orders placed in advance of historical lead times and have increased and may continue to increase as purchase obligations and prepaids have grown and are expected to grow as a greater portion of total supply, which may negatively impact gross margins and financial results.
• Financial · gross margin inventory provision Increased frequency and complexity of product transitions could result in unanticipated quality or production issues increasing the magnitude of inventory provisions; for example, gross margins in Q2 FY2025 were negatively impacted by inventory provisions for low-yielding Blackwell material.
• Geopolitical · international concentration International sales and operations are a significant part of NVIDIA's business; semiconductor wafers are manufactured, assembled, tested and packaged by third parties outside the U.S., and 31% of fiscal year 2026 revenue was generated from sales outside the United States, exposing it to economic, political, currency, trade and IP-protection risks.
• Geopolitical · china foreclosure Under current rules and geopolitical landscape NVIDIA is unable to create a competitive China data-center product that wins approval from both the USG and the Chinese government; as of the end of fiscal year 2026 it was effectively foreclosed from competing in China's data center compute market, helping competitors build larger ecosystems, with a material adverse impact on its business.
• Geopolitical · taiwan korea supply NVIDIA's business depends on consistent and reliable supply from overseas partners, especially in Taiwan and South Korea; any new restrictions negatively impacting its ability to receive components, parts, or services from Taiwan and South Korea would negatively impact its business and financial results.
• Geopolitical · israel operations NVIDIA is monitoring the geopolitical conflict in and around Israel, where it has approximately 6,000 employees primarily supporting R&D, operations, and sales/marketing of networking products; extension or expansion of the conflict could impact future product development, operations, and revenue. Other countries have restricted business with Israel via economic boycotts.
• Regulatory · export controls h20 charge In April 2025 the USG required a license to export H20 integrated circuits (and any circuits achieving the H20's memory bandwidth, interconnect bandwidth, or combination thereof) to China (including Hong Kong and Macau) and D:5 countries; as a result NVIDIA incurred a $4.5 billion charge in the first quarter of fiscal year 2026 associated with H20 for excess inventory and purchase obligations as demand diminished.
• Regulatory · h20 license revenue In August 2025 the USG granted licenses allowing NVIDIA to ship certain H20 products to certain China-based customers; NVIDIA generated approximately $60 million in H20 revenue under those licenses.
• Regulatory · usg revenue share USG officials expressed an expectation that the USG will receive 15% or more of the revenue generated from licensed sales of NVIDIA's products, but the USG did not publish a regulation codifying such requirement.
• Regulatory · h200 license tariff In February 2026 the USG granted a license to ship small amounts of H200 products to specific China-based customers; no revenue has been generated and shipments require a U.S. inspection process, so any H200 shipped under the program will be subject to a 25% tariff upon importation into the United States, which NVIDIA may be unable to pass to customers.
• Regulatory · ai diffusion ifr In January 2025 the USG published the AI Diffusion IFR, which would have imposed a worldwide licensing requirement on data center products such as H200, GB200 and GB300, dividing the world into three tiers (most countries to Tier 2) with a complex, burdensome licensing scheme; in May 2025 the USG said it would rescind it and implement an uncertain replacement rule that could materially impact NVIDIA.
• Regulatory · export controls general NVIDIA is subject to a series of shifting and expanding U.S. export control restrictions targeting GPUs and AI semiconductors that have been and are likely to be more restrictive, prohibiting exports to some or all customers in one or more markets, disrupting supply and distribution chains (including products warehoused in Hong Kong), and disproportionately harming NVIDIA's competitive position versus competitors outside the controls' scope.
• Regulatory · china mellanox finding On September 15, 2025 China's antitrust regulators published a preliminary finding that NVIDIA's compliance with U.S. export controls (requiring degraded products for China) discriminated unfairly against Chinese customers and violated the terms of China's approval of the Mellanox acquisition; an adverse conclusion could lead to financial penalties or restrictions on its networking business in China.
• Regulatory · responsible ai Issues relating to the responsible use of NVIDIA's technologies, including AI, may cause reputational or financial harm and liability; AI poses emerging legal, social and ethical risks, and complying with multiple AI regulations across jurisdictions (e.g., the EU and EU AI Act) may increase costs, change operations, and impede offering products in certain jurisdictions.
• Regulatory · data privacy NVIDIA is subject to stringent and changing data privacy and security laws (e.g., CCPA, GDPR, U.K. GDPR, NIS2, China data-localization/CAC rules); GDPR non-compliance can incur penalties up to 20 million euros or 4% of worldwide revenue, and NIS2 fines up to 10 million euros or 2% of worldwide revenue, with cross-border transfer restrictions that could harm operations.
• Regulatory · sustainability esg Scrutiny from shareholders, regulators and others regarding NVIDIA's corporate sustainability practices and disclosures (including climate change and human rights) could result in additional costs or risks, fail to meet evolving or state-level expectations or disclosure standards, harm its reputation, and expose it to liability.
• Legal · ip protection Protecting NVIDIA's IP rights could be costly and its ability to compete could be harmed; it faces patent-infringement and IP-rights lawsuits where unfavorable rulings could mean significant damages, patent invalidation, lost profits or injunctive relief, and foreign laws may not protect its products or IP to the same extent as U.S. law, making piracy more likely.
• Legal · litigation NVIDIA faces the burden and risks of litigation, investigations and regulatory proceedings, including a securities class action alleging it and certain officers made false or misleading statements about channel inventory and the impact of cryptocurrency mining on GPU demand in 2017 and 2018; adverse outcomes could result in monetary damages, fines, or injunctions and a material adverse impact.
• Legal · change of control Delaware law and provisions in NVIDIA's certificate of incorporation, bylaws, and its agreement with Microsoft for the Xbox could delay or prevent a change in control; under the Microsoft agreement, if someone offers to purchase at least 30% of outstanding common stock, Microsoft may have first and last rights of refusal to purchase the stock.
• Operational · cybersecurity Product, system security and data protection incidents or breaches, as well as cyber-attacks (including ransomware by organized criminals, nation-states, and AI-enabled attacks), could disrupt NVIDIA's operations, reduce revenue, increase expenses and harm its reputation; it has experienced security incidents (e.g., GFN breaches exposing third-party game source code) and may face government enforcement and liability.
• Operational · business disruption Business disruptions (natural disasters, extreme weather, power/water shortages, infrastructure or telecom failures, terrorism, war, pandemics, regulatory changes) could harm operations and revenue; NVIDIA's headquarters, much data center capacity and R&D are in California while inventory and suppliers are concentrated in Asia, with geopolitical tensions in China, Hong Kong, Israel, Korea and Taiwan posing further disruption risk.
• Operational · information systems NVIDIA's business depends on the proper functioning of business processes and information systems; modification or interruption, including problems from implementing a new ERP system, may disrupt its business and internal controls (including SOX Section 404), be costly, and impair its ability to timely and accurately report financial results.
• Operational · climate change Climate change may have a long-term adverse impact on NVIDIA's business; water and energy availability is critical and facilities (e.g., Santa Clara HQ and data centers) may be vulnerable to extreme weather and wildfire-related power safety shut-offs, while sustainability-related laws and lawsuits could raise costs, and insurance may not cover resulting losses.
• Financial · tax NVIDIA's operating results may be adversely impacted by additional tax liabilities, higher than expected tax rates, and changes in tax laws; most income is taxable in the U.S. with a significant portion qualifying as FDDEI, so higher U.S. rates or reduced FDDEI deduction, plus OECD Pillar One/Pillar Two implementation, could raise its effective tax rate and harm financial results.
• Financial · results fluctuation NVIDIA's operating results have fluctuated and may continue to fluctuate due to factors such as multi-year development cycles, extended customer payment terms and customer bad-debt risk, vendor payment requirements, and accounting-standard changes; if results fall below securities analyst or investor expectations, the stock price could decline or experience substantial volatility.
• Financial · counterparty Commercial arrangements, including long-term capacity purchase obligations and financial guarantees, expose NVIDIA to counterparty risk such as customers' or partners' inability to fulfill financial commitments, secure financing, project delays, and counterparty financial distress or insolvency; any financing arrangements undertaken could lower upfront cash flows and increase credit risk.
• Business · customer concentration NVIDIA receives a significant amount of revenue from a limited number of partners and distributors; for fiscal year 2026 sales to one direct customer represented 22% of total revenue and another direct customer 14% (primarily Compute & Networking), with some indirect customers individually estimated at 10% or more of revenue; losing or being prevented from selling to such customers would likely harm results.
• Business · key personnel If NVIDIA is unable to attract, retain and motivate its executives and key employees, its business may be harmed; it faces a highly competitive market for skilled workers, immigration/work-permit risks, rising cash and stock-based compensation costs, reduced retentive value of equity during stock-price volatility, and dependence on its longstanding executive team and effective succession planning.
• Business · acquisition integration NVIDIA may be unable to realize the benefits of business investments or acquisitions or to successfully integrate acquired companies; risks include difficulty integrating technology/systems/employees, assumed liabilities and goodwill impairment, failure to obtain regulatory approvals or imposed conditions, dilutive equity or debt issuance, and impairments of illiquid strategic investments.

Model Assumptions

The appropriate discount rate for any intrinsic-value model is the rate that correctly reflects the opportunity cost of the capital employed in the business. For NVIDIA, we apply the weighted average cost of capital as the discount rate in our discounted-cash-flow model, arriving at a WACC of approximately 12.53%. The use of WACC, rather than a simpler unlevered cost of equity, is theoretically grounded in the Modigliani-Miller framework: the WACC captures the blended, after-tax return required by all capital providers — equity holders and debt holders alike — weighted by their respective shares of total financing. Because NVIDIA's capital structure is overwhelmingly equity-funded, the WACC and the cost of equity are nearly identical in practice, but the distinction matters for analytical rigor and comparability with peers that carry more meaningful leverage. The 12.53% rate embeds our views on the current risk-free environment, the market's price for bearing systematic risk, and NVIDIA's own amplified exposure to the business cycle and to the competitive disruptions inherent in the semiconductor and AI-infrastructure industries.

We estimate NVIDIA's cost of equity using the Capital Asset Pricing Model, which posits that the required return on an equity investment equals the risk-free rate plus the product of beta and the equity risk premium. Our inputs are a risk-free rate of 4.3%, an equity risk premium of 5%, and a beta of 1.65, yielding a cost of equity of approximately 12.55%. The risk-free rate of 4.3% reflects the current yield on long-duration U.S. Treasury securities, which serve as the baseline return an investor can earn without bearing any systematic risk. We anchor to long-duration yields because a DCF model discounts cash flows over an extended horizon, and matching the duration of the risk-free rate to the forecast period is conceptually sound practice. An equity risk premium of 5% represents the additional annualized return the market demands above the risk-free rate for bearing broad market exposure. This figure sits near the center of long-run historical estimates and is consistent with survey-based, forward-looking measures used by institutional practitioners; it is neither aggressively pessimistic nor unduly optimistic.

Beta is the pivotal input and warrants the most discussion. NVIDIA carries a beta of 1.65, indicating that the stock has historically moved roughly 1.65 times as much as the broad market on a percentage basis. A coefficient meaningfully above 1.0 is not surprising given NVIDIA's profile. Its revenue is concentrated in a small number of end-markets — notably data-center AI infrastructure, gaming, and automotive — whose capital-expenditure cycles are highly correlated with the broader macroeconomic cycle. When the economy slows or enterprise budgets face scrutiny, AI infrastructure spending is not immune; when risk appetite contracts, high-multiple growth equities like NVIDIA draw down disproportionately. The beta of 1.65 also reflects NVIDIA's elevated operating leverage: a substantial portion of its cost structure is fixed in research and development and silicon design, so even modest revenue softness translates into amplified earnings volatility. Additionally, the competitive dynamics of the AI accelerator market — where NVIDIA currently enjoys dominant positioning but faces potential disruption from custom silicon built by hyperscalers, competing GPU vendors, and alternative neural-network architectures — introduce a component of event risk that keeps the realized coefficient structurally elevated. Multiplying the beta of 1.65 by the equity risk premium of 5% yields the systematic-risk premium component; combined with the risk-free rate of 4.3%, this produces the cost of equity of approximately 12.55%.

NVIDIA carries a pre-tax cost of debt of 4.5%, reflecting the blended yield on its outstanding long-term obligations. This coupon-equivalent yield is modest relative to NVIDIA's cost of equity — a natural consequence of the legal seniority of debt claims and the covenants that protect bondholders. However, the relevant input for WACC purposes is the after-tax cost of debt, since interest expense is deductible against taxable income. Applying NVIDIA's effective tax rate of approximately 15.1% shields a meaningful share of debt-service cost from taxation, producing an after-tax cost of debt that is materially lower than the pre-tax rate — though even the pre-tax figure's impact on the blended WACC is negligible given NVIDIA's capital structure, for reasons elaborated below. The effective tax rate of 15.1% sits below the U.S. statutory federal corporate rate, reflecting benefits from the research and development tax credit, the foreign-derived intangible income deduction, and the allocation of intellectual property across favorable tax jurisdictions. While this rate may fluctuate with legislative changes and the geographic mix of earnings, 15.1% represents a defensible near-term assumption that appropriately acknowledges NVIDIA's current tax posture.

One of the most structurally distinctive features of NVIDIA's capitalization — and one that simplifies the WACC calculation considerably — is the company's near-total reliance on equity financing. Equity represents approximately 99.8% of total capital, with debt accounting for only approximately 0.2%. This near-all-equity posture reflects NVIDIA's extraordinary cash-generation capacity in recent periods: the company has accumulated a substantial cash and short-term investment position that, on a net basis, nearly offsets its gross debt obligations. Management has prioritized shareholder returns through buybacks and dividends, and organic reinvestment in research and capital equipment, rather than deploying financial leverage to amplify returns. The practical implication for the WACC is that the after-tax cost of debt receives a weight of only approximately 0.2% in the blended rate. Even if one assumed a materially different borrowing cost, the effect on the WACC would be trivial — the cost of equity at approximately 12.55% dominates the calculation at a weight of approximately 99.8%, which is precisely why the WACC of 12.53% is essentially indistinguishable from the cost of equity. This structural reality means that sensitivity analysis on the cost of debt yields negligible variation in any DCF output; the model's conclusions are driven almost entirely by the cost-of-equity assumptions embedded in the CAPM. For valuation comparability, NVIDIA's capital-structure neutrality also means the distinction between enterprise value and equity value is largely a rounding matter, since the net cash position nearly cancels the gross debt — a feature that further supports treating an equity-value DCF as the primary methodology rather than a derived one.

Any DCF model is only as reliable as its inputs, and the WACC is no exception. Three sensitivities deserve explicit acknowledgment. First, beta is an inherently backward-looking estimate derived from historical price co-movements, and it may not fully capture forward-looking risk. If NVIDIA's competitive position in AI accelerators proves more durable than the market currently prices — reducing earnings volatility and compressing the realized coefficient over time — the appropriate cost of equity could be lower than 12.55%, supporting a higher intrinsic value. Conversely, the emergence of credible competing architectures or adverse regulatory actions could introduce structural revenue uncertainty and push the realized beta above 1.65, compressing present-value estimates. Second, the equity risk premium of 5% is a consensus-anchored assumption, but not an immutable one. In periods of severe macro stress or significant tightening in financial conditions, the implied ERP embedded in market pricing has historically risen well above this level, mechanically increasing the WACC and reducing the present value of long-dated cash flows. Because NVIDIA's growth profile places a substantial portion of total value in terminal-period cash flows, the DCF output is particularly sensitive to even modest changes in the discount rate — a feature analysts should stress-test explicitly. Third, the risk-free rate of 4.3% reflects current market conditions but would shift with monetary policy or the trajectory of the macro cycle. A sustained decline in long-term Treasury yields would reduce the risk-free anchor and could compress the WACC meaningfully; the reverse is equally true. Analysts using this model should consider scenario analysis across a range of risk-free rates to bracket the intrinsic-value estimate rather than treating any current yield level as permanent. In aggregate, the WACC of 12.53% represents our point estimate under market conditions prevailing as of the May 28, 2026 close. Readers should treat it as the central case in a distribution of plausible discount rates, with the understanding that the valuation is most sensitive to the beta and equity risk premium assumptions — the two inputs most subject to analytical judgment rather than direct market observation.

Valuation Methodology

NVIDIA Corporation (NVDA) — Valuation Methodology

Primary Valuation Lens: Discounted Cash Flow

For a business with NVDA's financial profile, discounted-cash-flow analysis is the most intellectually honest primary valuation methodology available to the equity analyst. DCF is first-principles: it anchors intrinsic value to the present worth of the cash a business will generate for its owners over time, stripped of accounting conventions, capital-structure noise, and the sentiment that can temporarily inflate or depress market-based multiples. NVDA has matured, over the past several years, into an extraordinarily cash-generative enterprise. With most-recent fiscal-year revenue of approximately $215.9 billion and most-recent fiscal-year free cash flow of approximately $96.7 billion, the company converts an exceptional share of its top-line into distributable owner earnings. That free-cash-flow yield on revenue is a hallmark of businesses whose competitive moats — in NVDA's case, the CUDA software ecosystem, the proprietary chip architecture roadmap, and deeply entrenched hyperscaler and enterprise customer relationships — translate pricing power and operating leverage directly into cash. Because DCF ties value to those actual cash flows rather than to a peer group whose own valuations may be equally stretched, it provides the most defensible foundation for an analyst seeking an independent estimate of intrinsic worth. Relative-valuation multiples (EV/EBITDA, P/E, EV/Sales) are retained as cross-checks in the appendix, but they are explicitly subordinate to the DCF conclusion.

Forecast Structure: Explicit Five-Year Horizon Plus Terminal Value

The model is structured in two stages. The first stage consists of an explicit five-year free-cash-flow forecast covering each discrete fiscal year of the projection period. Five years is the appropriate horizon for NVDA at this juncture: it is long enough to capture meaningful phases of the current AI-infrastructure investment cycle — including the build-out of large-scale GPU clusters, the expected ramp of inference-side workloads, and the partial normalization of gross margins as competitive dynamics evolve — yet short enough that the projections remain tethered to observable business fundamentals rather than purely speculative assumptions. Within this explicit window, each year's free cash flow is modeled bottom-up from revenue growth assumptions, operating-margin trajectories, capital expenditure intensity, and changes in working capital, all of which are detailed in the accompanying model-assumptions section. The second stage is a terminal value, computed at the end of the fifth forecast year, which captures the value of all cash flows beyond the explicit horizon. The terminal value is calculated using the Gordon Growth Model, in which the final-year free cash flow is grown at a stable perpetuity rate and then capitalized by the spread between the discount rate and that perpetuity rate. This two-stage structure is standard for a high-growth technology company that is expected to eventually converge toward a lower, steady-state growth rate consistent with the broader economy.

Discount Rate: Weighted-Average Cost of Capital

All free cash flows — both in the explicit five-year window and as embedded in the terminal value — are discounted to present value using the weighted-average cost of capital, or WACC. The WACC reflects the blended required return of all capital providers: equity holders (whose cost is estimated via a capital-asset pricing model framework, incorporating a risk-free rate, an equity risk premium, and NVDA's beta) and debt holders (whose after-tax cost is observable from the company's outstanding obligations). The WACC used in this model is approximately 12.5%, as derived and detailed in the model-assumptions section. This rate is meaningfully above long-run nominal GDP growth, which is appropriate given the inherent execution, competitive, and technology-cycle risks embedded in a company whose revenue base is highly concentrated in a single accelerated-computing architecture that is itself subject to rapid generational change. The 12.5% WACC therefore serves as a demanding hurdle: only businesses that can sustain high returns on incremental capital over an extended period will generate positive net present value at this discount rate, and NVDA's current free-cash-flow profile suggests it clears that hurdle with considerable room.

Terminal Value: Gordon Growth Model and the Convergence Condition

The terminal value is the single largest component of any DCF for a growth company, and it is therefore the assumption most deserving of discipline. The Gordon Growth Model requires that the perpetuity growth rate remain strictly below the discount rate — specifically, below the WACC of approximately 12.5% — so that the infinite sum of discounted cash flows converges to a finite number rather than diverging. A terminal growth rate at or above the WACC would imply that NVDA's free cash flow, grown forever at that rate and discounted at that rate, becomes infinite in present-value terms, which is economically incoherent. The terminal growth rate used in this model is 3%, reflecting a long-run nominal growth assumption broadly consistent with the trajectory of the developed-world economy and the mature semiconductor and infrastructure-software industries into which NVDA is assumed to converge over time. At 3%, the rate sits well below the 12.5% WACC, satisfying the convergence condition with a comfortable spread. This terminal rate should not be read as a forecast that NVDA will grow only as fast as the economy in five years' time; rather, it is a conservative anchor on perpetuity-value mathematics that guards against the well-documented tendency of DCF models to assign excessive terminal-value weight to optimistic long-run growth assumptions.

Scenario Framework: Bear, Base, and Bull

Recognizing that the range of credible outcomes for NVDA is unusually wide — given the early-innings nature of the AI infrastructure build-out and the consequent uncertainty around both the duration and the magnitude of the current demand cycle — the model is structured around three explicit scenarios: bear, base, and bull. Each scenario flexes the key value drivers independently: the trajectory of revenue growth within the explicit five-year forecast, the operating-margin and free-cash-flow-margin assumptions that flow from that revenue path, the capital expenditure intensity assumed as NVDA scales its own silicon and networking programs, and, critically, the terminal growth rate applied beyond the forecast horizon. In the bear scenario, near-term growth decelerates sharply as AI capital expenditure cycles down, competitive alternatives (both internal hyperscaler silicon programs and rival accelerator vendors) compress NVDA's addressable market, and the terminal growth rate is held at the lower end of the plausible range. In the base scenario, growth normalizes at a pace consistent with sustained AI infrastructure investment and NVDA's continued architectural leadership, with margins reflecting a modest degree of competitive-intensity normalization. In the bull scenario, inference-side demand proves structurally larger than the market currently prices, NVDA's software monetization layers (including CUDA-adjacent developer tooling and managed AI services) add incremental margin-accretive revenue streams, and the terminal growth rate reflects a more durable secular-growth premium. The probability-weighted midpoint of these three scenarios is the primary DCF conclusion; the range between bear and bull is presented explicitly to communicate the sensitivity of intrinsic value to the assumptions that matter most for this particular business at this particular point in its development.

Fair Value

Net Debt Bridge

Sensitivity

Shares Outstanding

Executive Summary

NVIDIA Corporation (NVDA) — Executive Summary | UNDERPERFORM

Our discounted-cash-flow analysis assigns NVIDIA Corporation an intrinsic value of $95.55 per share, a figure that sits well below the current market price of $135 per share. The gap between those two dollar figures — assessed against a weighted-average cost of capital of 12.5% — leads us to a rating of Underperform. At today's price, equity holders are paying a meaningful premium to our estimate of intrinsic worth, leaving scant margin of safety against the execution, competitive, and geopolitical risks that are central to the NVIDIA investment case.

NVIDIA's FY2026 financials are, in isolation, extraordinary. The company reported revenue of $215,938 million, with the Compute & Networking segment — the engine of the AI infrastructure boom — contributing $193,479 million of that total, and the Graphics segment adding a further $22,459 million. Gross profit reached $153,463 million, underscoring the pricing power embedded in NVIDIA's proprietary chip architecture and the CUDA software ecosystem that entrenches hyperscalers and enterprise customers within its platform. Net income for the year came in at $120,067 million, a figure that reflects both the scale of the revenue base and the formidable operating leverage the company has extracted from it. Research and development expenditure of $18,497 million signals continued investment in the architectural roadmap that underpins the competitive moat — a necessary commitment given the pace at which the competitive landscape is evolving.

The cash-generation profile reinforces the quality of the franchise. Free cash flow for FY2026 reached $96,676 million, supported by cash from operations of $102,718 million. That level of free cash flow relative to the revenue base is the hallmark of a business whose competitive advantages translate pricing power directly into distributable owner earnings — precisely the dynamic that makes a DCF the most intellectually defensible primary valuation lens. Our model applies a WACC of 12.5%, reflecting the blended required return of equity and debt holders and explicitly incorporating the execution, technology-cycle, and competitive risks that accompany a business operating at NVIDIA's scale and pace. The resulting fair value of $95.55 per share, set against the current market price of $135, implies that consensus expectations embed a materially more optimistic long-run growth and margin trajectory than our base-case projections support.

The risks to the investment case are material and must be weighed with care. Under current export-control rules and the prevailing geopolitical landscape, NVIDIA is effectively foreclosed from competing in China's data-center compute market — a position that assists competitors in building larger local ecosystems and that carries a material adverse impact on NVIDIA's business. The direct financial consequence was already visible: the U.S. government's April 2025 requirement for export licenses on H20 integrated circuits forced NVIDIA to recognise a $4.5 billion charge in the first quarter of FY2026 for excess inventory and purchase obligations as China demand evaporated. Beyond the geopolitical dimension, NVIDIA's target markets remain intensely competitive — new entrants continue to emerge, consolidation among incumbents is accelerating, and rivals may offer lower cost or superior functionality, dynamics that have already pressured selling prices and may continue to do so. Taken together, these risks reinforce our view that the current market price of $135 offers insufficient compensation for the uncertainties ahead, and that our DCF fair value of $95.55 constitutes the more appropriate anchor for disciplined position-sizing.

Investment Thesis

NVIDIA's durable competitive position originates in a decision taken in 2006: the introduction of CUDA, a parallel-computing platform that opened the programmable capabilities of NVIDIA's GPU to a broad range of compute-intensive applications and, in doing so, laid the technical foundation for the emergence of modern artificial intelligence. Two decades of developer adoption have transformed that early architectural bet into a switching-cost moat of unusual depth. More than 7.5 million developers worldwide now build and deploy AI workloads on CUDA and NVIDIA's associated software tools — a network whose collective investment in libraries, trained workflows, proprietary extensions, and institutional knowledge does not migrate cleanly to competing hardware without significant re-engineering cost. The full-stack character of NVIDIA's platform deepens the lock-in further. What NVIDIA sells is not simply powerful silicon; it is the layered combination of the CUDA development platform, the CUDA-X collection of acceleration libraries, AI models and training datasets, APIs, SDKs, and domain-specific application frameworks. Each layer of that stack raises the switching cost for the layer above it. Displacing any single component — say, the underlying GPU — becomes a cascading migration problem that most hyperscalers and enterprise AI teams are unwilling to absorb. This is the structural source of NVIDIA's pricing power and, ultimately, of the exceptional cash economics visible in its reported financials.

The financial scale that CUDA-anchored pricing power and operating leverage produce is remarkable by any standard. NVIDIA reported revenue of $215,938 million in fiscal year 2026, with the Compute & Networking segment contributing $193,479 million of that figure — a concentration that reflects the degree to which accelerated computing for AI has become the company's defining commercial reality. Gross profit for the period reached $153,463 million, and free cash flow was $96,676 million, representing a conversion ratio from top-line revenue that is exceptional among large-capitalisation technology businesses. Net income for fiscal 2026 stood at $120,067 million. The reinvestment funding the next phase of growth is proportional: NVIDIA directed $18,497 million toward research and development in fiscal 2026 — a level of spending that sustains the architectural roadmap and expands the distance between NVIDIA's software-hardware co-development cadence and the capabilities available from competing platforms. The most proximate catalyst on that roadmap is the Rubin platform, which the company expects to begin shipping in the second half of fiscal year 2027. New GPU generations have historically accelerated upgrade cycles at hyperscalers and large-model training operations; Rubin's arrival is expected to represent a similar inflection point. Beyond silicon, NVIDIA's collaboration with Marvell via NVLink Fusion and joint work on silicon photonics technology signals a deliberate expansion of the networking layer — a revenue stream that attaches organically to compute deployments and diversifies the top line. A nascent China pathway adds further optionality: a U.S. government license granted in February 2026 permits limited H200 shipments to specific Chinese customers, though no revenue has yet been generated under the programme and any re-imported units would be subject to a 25% tariff.

Our primary valuation framework is a two-stage discounted-cash-flow model that discounts explicit five-year free-cash-flow projections, and a terminal value, at a weighted-average cost of capital of 12.5%. That WACC reflects the blended required return of equity and debt holders, calibrated to NVIDIA's technology-cycle, competitive, and execution risks, and is set deliberately above long-run nominal growth to avoid underpricing those risks in the terminal value. The resulting fair-value estimate is $95.55 per share. The DCF approach is particularly appropriate for a business of NVIDIA's financial maturity: with cash from operations of $102,718 million in fiscal 2026, the company generates real owner earnings on a scale large enough to anchor intrinsic-value analysis rather than relying on accounting proxies. Investors should treat $95.55 as a central estimate within a range — both upside and downside cases imply materially different outcomes depending on assumptions around long-run growth normalisation and competitive margin erosion, and the range of plausible terminal values for a business at the intersection of AI infrastructure and proprietary software is inherently wide.

The investment case faces two categories of structural risk that could impair the thesis at any point in the holding period. The first is competitive: NVIDIA's target markets remain intensely contested, and competition may intensify as new entrants emerge and established players — including custom-silicon programmes at major cloud providers — consolidate their capabilities. Competing products may offer lower prices or differentiated functionality, a dynamic that has already influenced pricing in selective workloads and could broaden if alternative accelerators reach performance parity faster than current timelines suggest. The second risk category is regulatory. In April 2025, the U.S. government imposed an export-licensing requirement on H20 integrated circuits destined for China and certain other countries; NVIDIA recorded a $4.5 billion charge in the first quarter of fiscal year 2026 associated with excess H20 inventory and purchase obligations as customer demand from those channels contracted sharply. The partial reopening via the H200 licensing programme is subject to a U.S.-based inspection requirement, is exposed to a 25% tariff upon re-importation into the United States prior to shipment, and carries no guarantee of further liberalisation. Export-control risk is asymmetric and operationally difficult to hedge: a single regulatory action can render a meaningful share of addressable market inaccessible with limited warning. Together, these risks argue for scenario-weighting the central thesis rather than treating the fair-value estimate as a point prediction, and for sizing positions in a manner consistent with the possibility that either competitive or regulatory headwinds materialise more severely than the base case assumes.

Base Case

The base case assigns NVDA a DCF-implied fair value of $95.55 per share — a figure that sits meaningfully below the current market price of $135 per share. That gap is not a mechanical anomaly; it is a deliberate signal. The base case embeds the central expectation that the AI infrastructure super-cycle, while historically powerful, will not sustain its present velocity indefinitely. Investors pricing NVDA at $135 are implicitly underwriting growth assumptions that exceed the base-case trajectory on at least one material dimension — whether duration, magnitude, or both. This scenario serves as the measured anchor of the valuation range: neither the optimist's extrapolation nor the skeptic's mean-reversion, but the path that a disciplined weight-of-evidence framework finds most probable.

The growth path the base case assumes reflects a structured deceleration: 35% in year one, easing through 28%, 22%, 17%, and 12% over the five-year explicit forecast horizon. This arc is not a cliff; it is a slope. The opening years capture the continued buildout of hyperscaler AI training infrastructure, sovereign AI deployments, and the migration of enterprise workloads to GPU-accelerated compute. The later years acknowledge the forces that reliably moderate any capital-expenditure super-cycle: competitive supply responses, customer optimization of installed fleets, and the law of large numbers applied to an already substantial revenue base. A terminal growth rate of 3% anchors the perpetuity value, calibrated to long-run nominal GDP expansion and appropriate for a franchise that, even in maturity, retains pricing power and platform lock-in. The discount rate of 12.5% reflects NVDA's equity risk profile — elevated relative to stable industrials, commensurate with a high-multiple technology hardware name exposed to cyclical demand swings and the concentration risk inherent in a data-center-dominant revenue mix.

The financial base from which these projections depart is exceptional by any standard. FY2026 revenue reached $215,938 million, a figure that positions NVDA among the most generative technology hardware franchises on the planet and establishes a compounding base that makes even a decelerating growth path produce substantial absolute revenue additions in each forward year. Critically, that revenue does not remain trapped on the income statement; it converts to cash with structural efficiency. FY2026 free cash flow stood at $96,676 million, a level that reflects the compounding advantages of NVDA's fabless operating model and the pricing discipline the company has exercised in a supply-constrained environment. Cash generated from operations reached $102,718 million in the same period, confirming that free cash flow is not flattered by working-capital timing or one-time items; the underlying conversion from earnings to cash is durable and repeatable across the cycle. Gross profit of $153,463 million underscores the margin architecture that sustains this conversion: a business with this cost structure can absorb meaningful volume deceleration before free-cash-flow generation deteriorates in a way that threatens the valuation floor. Net income of $120,067 million further confirms that the earnings base is real, not obscured by depreciation schedules or non-cash compensation that inflates operating metrics. The base case assumes that this profitability architecture is broadly preserved through the five-year explicit forecast period, with margins compressing only modestly as competitive dynamics normalize and the mix of inference versus training workloads evolves.

What could shift the outcome toward the bull scenario? Principally, an acceleration or extension of hyperscaler AI infrastructure spending that keeps revenue growth above the base-case ramp for longer; a faster-than-anticipated scaling of inference workloads that absorbs incremental GPU capacity without a demand pause between upgrade cycles; or a broadening of NVDA's addressable market into robotics, autonomous systems, and physical AI at a pace that replaces decelerating data-center growth with a new, high-margin demand vertical before the existing cycle fully matures. Conversely, the bear scenario would be triggered by a material curtailment of AI capital-expenditure budgets — whether driven by regulatory pressure, disappointing AI monetization by cloud customers, or an accelerated competitive response from custom silicon and alternative architectures that erodes NVDA's unit-economics advantage. Export-control tightening represents an asymmetric binary risk that neither the base nor the bear case fully prices with precision, given its dependence on policy decisions outside NVDA's control. The base case holds none of these tail outcomes as central; it treats the present hyper-growth period as real but finite, prices in an orderly transition to a more normalized competitive environment, and arrives at $95.55 as the fair value for a business navigating that transition with its software ecosystem, developer platform, and manufacturing partnerships intact.

Bull Case

Our bull-case discounted-cash-flow model yields an implied fair value of $138.44 per share — well above our base-case headline fair value of $95.55 and modestly above the current market price of $135. The gap between the bull case and the base case is not an arbitrary optimism premium; it reflects a specific scenario in which three disclosed, company-identified catalysts compound simultaneously: the Rubin product cycle, the maturation of NVIDIA's networking ecosystem, and a licensed pathway back into the Chinese market. The bull-case growth assumptions — 45%, 38%, 30%, 24%, and 18% in years one through five, converging to a 4% terminal growth rate against a 12.5% discount rate — are the mechanical translation of that compounding. Investors who purchase at current prices are, in effect, paying close to full bull-case value; the burden of proof falls squarely on execution.

The financial foundation from which the bull case compounds is formidable. In fiscal year 2026, NVIDIA generated $215,938 million in revenue, $153,463 million in gross profit, $120,067 million in net income, and $96,676 million in free cash flow — a cash generation profile that funds the research and development, capital allocation, and ecosystem investments the bull thesis requires. These are not projected figures; they are the reported baseline. The bull case assumes that this base is a platform, not a ceiling, and that incremental growth at elevated rates is achievable precisely because the cost structure and ecosystem density that produced FY2026 results are durable and defensible. A company generating cash flow of that magnitude carries the financial flexibility to sustain simultaneous product development, go-to-market expansion, and ecosystem seeding — the combination the bull thesis demands.

The single most consequential catalyst in the bull thesis is the Rubin platform. NVIDIA has stated that the Rubin architecture is expected to begin shipping in the second half of fiscal year 2027. If the Rubin ramp tracks the cadence established by Hopper and Blackwell — rapid hyperscaler adoption, multi-quarter backlog, and accelerating software lock-in — then the elevated revenue growth in years one and two of our projection is not speculative; it is a function of known customer demand meeting a new supply curve. Crucially, the Rubin cycle reinforces a structural advantage that NVIDIA has compounded since 2006: the CUDA development platform, which runs across all NVIDIA GPUs and represents nearly two decades of developer investment, library accumulation, and workflow integration. Each successive architecture generation extends the value of that installed base of CUDA-trained engineers and production workloads, raising the cost of defection for hyperscalers and enterprise customers alike. In the bull case, the Rubin launch resets the competitive clock and sustains the opening-year growth rate by pulling forward demand that might otherwise have decelerated as Blackwell penetration matured. What must hold is that the Rubin ramp encounters no material yield, supply-chain, or architectural setback that delays revenue recognition beyond the first half of fiscal year 2028.

Networking is the second structural pillar. In the first quarter of fiscal year 2027, NVIDIA expanded its AI ecosystem through a strategic partnership with Marvell via NVIDIA NVLink Fusion, alongside a collaboration on silicon photonics technology. This development is significant because it signals that NVIDIA is extending its architectural reach beyond the GPU compute node into the high-speed interconnect layer that binds large-scale AI clusters together. NVLink Fusion and silicon photonics co-development with Marvell position NVIDIA to capture value at every layer of the AI infrastructure stack — compute, memory bandwidth, and now optical interconnect — deepening the per-cluster dollar content the company captures as cluster sizes scale. For the bull case to hold, networking revenue must grow in proportion to, or faster than, compute revenue. The partnership structure is the mechanism by which that co-growth becomes achievable; the risk is that competing interconnect standards gain traction among hyperscalers before NVLink Fusion achieves critical-mass adoption.

The third catalyst is the most contingent — and the most asymmetric. Beginning in February 2026, the U.S. government granted licenses allowing NVIDIA to ship small quantities of H200 products to specific China-based customers. As of the most recent disclosure, no revenue has been recognized under this licensing program, and it remains uncertain whether imports will ultimately be permitted into China. Any H200 unit shipped under the program is subject to a 25% tariff upon re-importation into the United States as part of a required domestic inspection process. The program is therefore neither de-risked nor fully open; it is an option, not a guarantee. In the bull case, this option carries real value: a structured, compliant pathway into one of the world's largest AI compute markets provides an incremental revenue ceiling the base case does not incorporate. The bull scenario does not model full China market recovery — it models a partial, licensed resumption that adds volume to an already-growing global pipeline. What must hold for this catalyst to contribute is regulatory continuity, timely license renewals, and logistical resolution of the inspection-and-tariff regime. Any reversal of the licensing program would compress growth expectations in years three through five and pull the implied fair value meaningfully toward the base case.

In aggregate, the bull case at $138.44 per share is the specific, model-implied outcome if Rubin ships on schedule and ramps with force, NVLink Fusion and silicon photonics collaboration translate into durable networking revenue share, the China H200 licensing program matures into recurring volume, and the CUDA platform continues to function as the moat that makes customer defection economically irrational across successive architecture generations. Each condition is observable and falsifiable over the next four to six quarters. Investors entering near the current market price of $135 are implicitly endorsing the bull thesis with limited additional margin; the risk-reward is favorable only if execution across all three catalyst tracks is sustained without material interruption.

Bear Case

The bear case anchors to a DCF-implied fair value of $57.26 per share — a figure that sits meaningfully below the base-case DCF of $95.55 and well below the current market price of $135. To arrive at that valuation, the model assumes revenue growth decelerates sharply from 22% in year one to 14%, 9%, 5%, and 3% in successive years, before settling at a 2% terminal growth rate — a profile consistent with a world in which several material disclosed risks crystallise simultaneously and compound across the forecast horizon. Investors pricing NVIDIA at the current market price are, in effect, assigning near-zero probability to that compounding; the bear case assigns it substantial weight.

The most legible near-term proof that the downside scenario is not merely hypothetical is the $4.5 billion charge NVIDIA recorded in the first quarter of fiscal year 2026, tied directly to H20 integrated circuits after the U.S. government imposed a licensing requirement for exports of those chips — and any circuit matching the H20's memory-bandwidth or interconnect-bandwidth specifications — to China, Hong Kong, Macau, and D:5 countries. That charge, representing excess inventory and purchase obligations that became stranded as demand evaporated, illustrates both the speed and the magnitude with which a regulatory action can impair a product cycle with essentially no warning. The broader export-control regime is structurally expansionary: successive rounds of restriction have been more rather than less severe, have disrupted warehoused inventory in Hong Kong, and have demonstrably advantaged competitors whose products fall outside the controls' scope. These restrictions do not merely create a one-quarter headache — they compound into the medium term by foreclosing NVIDIA from competing in China's data-center compute market altogether. As of the close of fiscal year 2026, NVIDIA has been effectively unable to bring a product to market that simultaneously satisfies U.S. government approval and wins commercial acceptance from Chinese buyers, a structural impasse that the company itself acknowledges is actively helping domestic Chinese competitors build larger ecosystems with a material adverse impact on its business. In the bear case, that market loss is treated as durable rather than transitory, and it is the primary reason the model's growth path falls steeply from year two onward.

Competition beyond the China foreclosure adds a second and arguably more persistent drag on the growth path. NVIDIA's target markets are already contested, and the company's own disclosures caution that new entrants and industry consolidation could intensify rivalry further — with competitors offering products at lower price points or with feature advantages that erode NVIDIA's selling prices or unit demand. The more structurally significant competitive threat, however, comes not from external chip vendors but from NVIDIA's own largest customers. Several hyperscale and cloud-native operators have in-house engineering teams with silicon-design depth comparable to NVIDIA's own, and some are developing or have deployed proprietary AI accelerators intended to displace, at least partially, third-party GPU purchases. These same operators also increasingly offer cloud-based AI services that compete directly with NVIDIA's own AI platform offerings. The bear case treats this dynamic as a slow-moving but durable headwind: as custom silicon matures, the fraction of workloads that requires best-in-class third-party GPUs shrinks, compressing both volume growth and pricing power over the five-year forecast window in a way that is poorly captured by near-term consensus estimates.

Customer concentration amplifies each of the foregoing risks and limits NVIDIA's ability to diversify away from any single point of failure. In fiscal year 2026, one direct customer accounted for 22% of NVIDIA's total revenue — against reported revenue of $215,938 million, an extraordinarily large single-name exposure — while a second direct customer contributed 14%, and certain indirect customers are individually estimated to approach or exceed 10% of revenue. The concentration is heaviest in the Compute and Networking segment, meaning the chips most affected by export controls and most susceptible to in-house silicon substitution are also the chips most reliant on a narrow buyer base. In the bear case, any combination of regulatory-driven demand destruction at these accounts, accelerated rollout of proprietary accelerators, or pricing pressure triggered by intensifying competition produces an outsized revenue impact that a more diversified revenue base could absorb with far less pain. NVIDIA's reported free cash flow of $96,676 million in fiscal year 2026 is a testament to the current earnings power of the franchise; the bear case is emphatically not a claim that the business is structurally impaired today. It is a sober, risk-grounded assessment of the conditions under which a compounding of tightening export controls, durable geopolitical foreclosure from China's data-center market, competitive substitution from both external rivals and customer in-house programs, and the inherent fragility of a concentrated revenue base could drive growth well below the trajectory the market is currently pricing — yielding a fair value of $57.26 per share, a figure the market at $135 is currently not pricing at all.

Key Catalysts

The single most consequential near-term catalyst for NVIDIA is the imminent arrival of its next hardware generation. Management has confirmed that the Vera Rubin platform — announced in Q1 FY2027 alongside the NVIDIA Vera CPU, described as the world's first processor purpose-built for agentic AI, and NVIDIA BlueField-4 STX accelerated storage infrastructure for agentic AI factories — is expected to begin shipping in the second half of fiscal year 2027. This product introduction fits within NVIDIA's stated one-year cadence for advanced Data Center compute architectures, a tempo that has historically been the primary mechanism through which the company maintains performance leadership and drives upgrade cycles across its hyperscaler and enterprise customer base. The Rubin platform's explicit orientation toward agentic AI signals that NVIDIA is positioning its next hardware cycle around a structural shift in workload character — from primarily training large foundational models toward inference-heavy, agent-orchestrated deployments that place distinct demands on compute throughput, memory bandwidth, and storage infrastructure. The speed at which the market can absorb a new architecture — given that Blackwell adoption is still scaling across the customer base — is one of the most important watchpoints heading into the back half of FY2027 and will set the tone for the company's revenue trajectory in the subsequent fiscal year.

The China dimension remains the most asymmetric risk-reward variable on NVIDIA's near-term catalyst map, and the current state of play is defined by carefully conditioned optionality rather than restored access. Beginning in February 2026, the U.S. government granted licenses permitting NVIDIA to ship small amounts of H200 products to specific China-based customers — a partial reopening of a market from which the company had been effectively foreclosed. The pathway is, however, heavily encumbered: each H200 must pass through a U.S.-based inspection process before any shipment to the customer, and any unit shipped under the program will be subject to a 25% tariff upon importation into the United States. Critically, NVIDIA has reported no revenue under the H200 licensing program to date and does not yet know whether any imports will ultimately be cleared into China. The company's forward guidance makes its posture on this question explicit and conservative: NVIDIA is not assuming any Data Center compute revenue from China in its Q2 FY2027 outlook. This underwriting approach is both prudent risk management and analytically revealing — it means any successful H200 deliveries into China would represent pure upside relative to published consensus anchored to guidance, while simultaneously underscoring that the structural market foreclosure documented in the company's competitive disclosures has not materially reversed. The resolution of the China pathway — whether licenses translate into actual cleared shipments, whether the 25% tariff burden suppresses Chinese customer demand, and whether the U.S. government expands or contracts the licensing program — constitutes the single largest binary on NVIDIA's near-term revenue distribution.

Beyond the compute hardware cycle and the China variable, a cluster of ecosystem and infrastructure catalysts indicates that NVIDIA's competitive perimeter continues to expand across multiple verticals simultaneously. In networking, NVIDIA expanded its AI ecosystem through a strategic partnership with Marvell via NVIDIA NVLink Fusion and a collaboration on silicon photonics technology — developments that extend the company's interconnect architecture into new territory and deepen its integration with third-party silicon partners, broadening the addressable install base for its networking stack. In automotive, NVIDIA expanded its partnerships with Hyundai Motor Company and Kia for next-generation autonomous driving built on the NVIDIA DRIVE Hyperion platform, extended its collaboration with Uber to launch a fleet of autonomous vehicles powered by full-stack NVIDIA DRIVE AV software, and announced that BYD, Geely, Isuzu, and Nissan are building level 4-ready vehicles on the DRIVE Hyperion platform — a design-win roster spanning both Western and Chinese original equipment manufacturers that illustrates the breadth and depth of the automotive pipeline maturing toward production. At the edge and telecom layer, NVIDIA announced collaboration with T-Mobile and Nokia to integrate physical AI applications on AI-RAN-ready infrastructure, and secured commitments from global telecom leaders to build 6G wireless networks on AI-native, open, and secure platforms — a category that could eventually represent a meaningful incremental addressable market as AI-RAN architectures move from pilot to deployment. On the ecosystem investment front, management confirmed it is finalizing an investment and partnership agreement with OpenAI, though appropriately noting there is no assurance the agreement will be completed or that a transaction will close; the strategic logic of such a partnership, if consummated, would further tighten the integration between NVIDIA's infrastructure layer and the most prominent frontier AI developer. The consumer and professional graphics roadmap also advanced, with the release of NVIDIA DLSS 4.5 Dynamic Multi Frame Generation and the preview of DLSS 5, described by the company as NVIDIA's most significant graphics breakthrough since ray tracing in 2018 — a claim that, if the market validates it at launch, would provide a meaningful catalyst for the Gaming segment's upgrade cycle.

Management's Q2 FY2027 guidance distills these competing forces into a single auditable financial picture that anchors near-term expectations. Revenue is expected to be $91.0 billion, plus or minus 2% — an outlook explicitly constructed without any assumed contribution from Data Center compute sales into China, meaning the guidance represents a floor scenario relative to potential China upside. GAAP gross margins are guided to 74.9% and non-GAAP gross margins to 75.0%, each with a plus or minus 50 basis point tolerance band — a level that, if realized, would confirm that Blackwell-era product mix and manufacturing scale are sustaining the company's profitability profile through the product transition. GAAP operating expenses are expected to be approximately $8.5 billion and non-GAAP operating expenses approximately $8.3 billion, reflecting the sustained investment intensity required to execute a one-year product cadence across compute, networking, automotive, and edge platforms concurrently. For the full fiscal year, NVIDIA expects its GAAP and non-GAAP tax rates to land between 16.0% and 18.0%, excluding discrete items and material changes to the company's tax environment, while also flagging a substantial increase in cash taxes in Q2 related to estimated federal and state cash tax payments — a working-capital headwind that warrants monitoring alongside the earnings trajectory but does not alter the structural demand picture. Days sales outstanding are expected to return to more normal levels in Q2, indicating that the receivables position at quarter-end was transient rather than a signal of collection risk. Synthesizing across the guidance package, the near-term financial outlook confirms that demand remains structurally intact across NVIDIA's addressable customer base, that margin discipline is being preserved through the hardware transition, that China exclusion is already fully discounted in the guide, and that the Rubin platform launch in the second half of FY2027 is positioned as the next leg of the company's revenue growth architecture — with the China licensing pathway representing the most significant unpriced option embedded in the current setup.

Conclusion

NVIDIA is an exceptional business operating at a price that exceeds our estimate of its intrinsic worth. Our discounted-cash-flow model yields a fair value of $95.55 per share — a figure that sits well below the current market price of $135 per share. That gap underpins an Underperform rating. At today's price, equity holders are paying a meaningful premium to our estimate of intrinsic value, and the margin of safety against execution, competitive, and geopolitical risks is, in our judgement, inadequate.

The business quality underpinning this franchise is not in question. Fiscal year 2026 revenue of $215,938 million, gross profit of $153,463 million, and net income of $120,067 million attest to a company that has monetised the AI infrastructure cycle with striking efficiency. Free cash flow of $96,676 million in the same period — a conversion rate that is exceptional among businesses of comparable scale — underscores the degree to which NVIDIA's pricing power translates directly into distributable owner earnings. These are the financials of a franchise with genuine structural advantages: CUDA's two-decade entrenchment, the layered full-stack ecosystem that raises switching costs at every level, and the formidable operating leverage that the combination of proprietary silicon and software commands. The moat is real, the cash economics are exceptional, and the reinvestment discipline is evident. None of that, however, is the question our analysis poses.

Yet exceptional financials do not, by themselves, justify any price. Our concern is not the quality of the franchise but the adequacy of the margin of safety embedded in the current quotation. At $135 per share, the market's implied long-run growth and margin assumptions are materially more optimistic than our base-case projections, which already reflect NVIDIA's durable competitive moat and reinvestment cadence. Compounding that valuation stretch are the geopolitical and competitive risks that are concrete rather than speculative: effective exclusion from China's data-center compute market, active rival investment in competing architectures, and the structural uncertainty inherent in any business whose revenue is concentrated in a single technology cycle. Our DCF fair value of $95.55 per share represents, in our view, the more defensible anchor for disciplined position-sizing. Until the market price converges toward that estimate — or until the risk landscape shifts materially in NVIDIA's favour — we maintain an Underperform rating.

Verdict