NVIDIA’s financial performance over the five fiscal years from FY2022 to FY2026 represents a transformation of order of magnitude rarely seen in the technology sector. Revenue grew from $26.9B in FY2022 to $215.9B in FY2026, a compound annual growth rate of roughly 68%, driven by a surge that began in FY2024 when revenue more than doubled to $60.9B, then nearly doubled again in FY2025 to $130.5B, before rising another 65.5% in FY2026. The inflection point is unmistakable: in the FY2023–FY2024 transition, the company’s filing replaced its historical Mellanox pro forma summary with a Management’s Discussion and Analysis summary that highlighted revenue up 126% and operating income up 681% ¹. That same filing introduced, for the first time, a segment-level revenue breakdown showing Compute & Networking revenue up 215% to $47.4B ² and its operating income up to $32.0B ³. The narrative abruptly shifted from integration disclosure to a triumphantly scaled business.

However, the path was not linear. In FY2023, NVIDIA experienced a sharp contraction: gross profit fell 12.1% to $15.4B and net income dropped 55.2% to $4.4B ⁴. Cost of revenue rose to 43.1% of revenue from 35.1% the prior year, compressing gross margins from 64.9% to 56.9% ⁵. The company’s inventory ballooned 98.0% to $5.2B, and it began disclosing a breakdown into raw materials, work-in-process, and finished goods for the first time ⁶. This granularity suggested management was closely monitoring a buildup that might have signaled demand uncertainty. By FY2024, however, the inventory overhang was absorbed: revenue exploded, inventory grew only 2.4%, and gross profit soared 188.5% ². The margin recovery was complete, and the company’s operating leverage became extraordinary, operating income rose from $4.2B in FY2023 to $33.0B in FY2024, a 680.6% increase ¹.

The segment disclosure was short-lived. In FY2025, the revenue breakdown table for Compute & Networking and Graphics was removed ⁷. Investors lost the visibility into the relative contributions of the two segments that had been so illuminating. The filing also eliminated a table of notional values of foreign currency forward contracts outstanding ⁸, reducing transparency on hedging activities. These disappearances coincide with a period of explosive growth and may reflect management’s view that the segment detail was no longer strategically useful or that competitive sensitivity outweighed transparency. The removal of fair value methodology disclosure in FY2026 ⁹ continued this pattern.

FY2023

The following unaudited pro forma financial information summarizes the combined results of operations for NVIDIA and Mellanox

→

FY2024

Fiscal Year 2024 Summary | Year Ended | Jan 28, 2024 | | Jan 29, 2023 | | Change | ($ in millions, except per share data) Revenue | $ | 60,922 | | | $ | 26,974 | | | Up 126%

Cash flow and capital allocation tell a story of massive cash generation increasingly returned to shareholders. Operating cash flow was $5.6B in FY2023, then surged to $28.1B in FY2024, $64.1B in FY2025, and $102.7B in FY2026. Investing cash flow turned sharply negative in FY2024 and beyond, reaching -$52.2B in FY2026, reflecting heavy capital expenditure. Financing cash flow was consistently negative from FY2023 onward, driven by share repurchases and dividends. Share repurchases went from $10.0B in FY2023 to $33.7B in FY2025 and $40.1B in FY2026 ¹⁰. Dividends, after flatness at $399M, doubled to $834M in FY2025 and rose to $974M in FY2026.

Stock-based compensation grew from $2.0B in FY2022 to $2.7B in FY2023, a 35.2% increase ¹¹. By FY2025, the company expanded its disclosure to clarify that stock-based compensation was capitalized into inventory and subsequently recognized to cost of revenue ¹², a nuance that matters for margin analysis. The impact on diluted EPS was dramatic: after the FY2024 surge to $11.93, diluted EPS fell to $2.94 in FY2025 despite net income doubling, reflecting a stock split or other dilution mechanics. Basic EPS went from $12.05 to $2.97, a 75.3% decline ¹⁰. The FY2025 filing also replaced a specific discussion of Israel and regional conflicts with a cross-reference to risk factors ¹³, suggesting that geopolitical risk was being consolidated into a broader framework.

The balance sheet swelled: total assets grew from $44.2B in FY2022 to $206.8B in FY2026, with cash and equivalents reaching $10.6B. Accounts receivable rose to $38.5B, a 66.8% increase in FY2026 alone, and inventory jumped 112.3% to $21.4B, a new build that could be a precursor to a future demand normalization. Property and equipment nearly quadrupled to $10.4B, underscoring the capital intensity of the AI infrastructure build-out. Long-term debt declined modestly from $10.9B to $7.5B, while stockholders’ equity grew to $157.3B.

In sum, NVIDIA’s financial performance over these five years is a case study in hypergrowth, margin recovery, and the tensions between transparency and strategic discretion. The numbers tell a story of a company that scaled more than eightfold in revenue while managing inventory cycles, shifting disclosure priorities, and returning enormous cash to shareholders. The most striking change may be the collapse of EPS in FY2025 while net income soared, a reminder that shareholder returns are not always aligned with per-share metrics when capital structure changes intervene.

The five-year arc of NVIDIA’s business description reveals a company that transformed itself from a hardware supplier into a full-stack AI infrastructure powerhouse, while simultaneously navigating an escalating regulatory crisis that has permanently altered its addressable market. The strategic pivot was initiated in the FY2023 filing, where the company replaced a detailed description of GPU foundations and partner lists with an announcement that it would “offer enterprise customers NVIDIA AI cloud services directly and through our network of partners” ¹. This marked the first explicit shift from selling chips to selling cloud-based AI services, a new business model that would later be formalized with the launch of DGX Cloud in FY2024 ². The narrative hardened further in FY2024 when the company rewrote its core identity: “NVIDIA is now a full-stack computing infrastructure company with data-center-scale offerings” ³, and reframed its GPU not as a graphics processor but as a tool that “was initially used to simulate human imagination” ⁴. The gaming market, once the lead market segment, was demoted; by FY2024 the primary users were “the world’s leading cloud service providers” ⁵, and by FY2025 the Gaming section introduction had been removed entirely, replaced by the Blackwell architecture launch ⁶. The Data Center segment became the dominant narrative, with its description expanding in FY2026 to include specific workloads, AI, data processing, graphics, robotics, scientific computing, and deployment models across cloud, hyperscale, on-premises, and edge ⁷. The financial results mirror this shift: revenue soared from $26.9B in FY2022 to $215.9B in FY2026, with operating income growing from $10.0B to $130.4B, and operating cash flow from $9.1B to $102.7B. The language of the business description hardened precisely as these numbers exploded, with the company emphasizing co-design, full-stack innovation, and order-of-magnitude performance gains ⁸.

The most dramatic and persistent theme is the evolution of U.S. export controls and their impact on China. In FY2023, the company first disclosed that the U.S. government had “announced new license requirements that impact certain exports to China” and described the impact as “difficult to quantify” ⁹. By FY2024, the language had escalated sharply: “Our competitive position has been harmed, and our competitive position and future results may be further harmed in the long term, if there are further changes in the USG’s export controls” ¹⁰. The supplier redundancy discussion was replaced entirely with licensing requirements for A100 and H100 products ¹¹. In FY2025, a new “AI Diffusion” IFR was introduced, imposing a worldwide licensing requirement ¹². The FY2026 filing contained the most striking admission: “Under the current rules and geopolitical landscape, we are unable to create and deliver a competitive product for China’s data center market” ¹³. The company further disclosed that while the USG granted licenses to ship H20 products to China in August 2025, generating only $60 million in revenue, USG officials had expressed an expectation of receiving 15% or more of that revenue ¹⁴. The licensing process risks were described in stark terms: “The licensing process may not be resolved before significant business opportunities evaporate” ¹⁵. This regulatory burden intensified precisely as the company’s overall revenue exploded, creating a bifurcated narrative of unprecedented growth in the rest of the world and a near-total loss of the Chinese data center market.

FY2022

At the foundation of the NVIDIA accelerated computing platform are our GPUs, which excel at parallel workloads such as the training and inferencing of neural networks. They are available in industry standard servers from every major computer maker, including Cisco Systems, Inc., or Cisco, Dell Technologies Inc., Hewlett Packard Enterprise Company, or HP, Hitachi Vantara, Inspur Group, and Lenovo Group Limited; from every major cloud service provider such as Alicloud, Amazon Web Services, Baidu

→

FY2023

NVIDIA will offer enterprise customers NVIDIA AI cloud services directly and through our network of partners. Examples of these services include NVIDIA DGX Cloud, which is cloud-based infrastructure and software for training AI models, and customizable pretrained AI models. NVIDIA has partnered with leading cloud service providers to host these services in their data centers. Our networking solutions include InfiniBand and Ethernet network adapters and switches, related software, and cables.

Supply chain language shifted from being a non-issue to a strategic priority. The FY2023 filing removed the traditional description of a fabless manufacturing strategy ¹⁶, but it reappeared in FY2024 as a “fabless and contracting manufacturing strategy” ¹⁷. By FY2025, the company had “expanded our supplier relationships to build redundancy and resilience in our operations to provide long-term manufacturing capacity aligned with growing customer demand” ¹⁸, and disclosed non-cancellable inventory orders. Cumulative R&D investment grew from $45.3 billion to $58.2 billion between FY2024 and FY2025 ¹⁹, signaling a massive commitment to maintaining technological leadership.

The company’s positioning on ESG and autonomous vehicles tells a story of strategic retreat. In FY2023, the company added specific risk language around climate change, cryptocurrency, and consumer laws, while softening its ESG commitment from “integrate” to “goal to integrate” ²⁰. By FY2024, the entire ESG section was eliminated, replaced with a statement that compliance “has not otherwise had a material effect” ²¹. Autonomous vehicle language appeared in FY2023 with a detailed AV solution description ²²²³, expanded in FY2024 with a full paragraph on the AV and EV market opportunity ²⁴, then vanished in the same filing as the AV solution was deleted and replaced by a sales and marketing strategy section ²⁵. By FY2025, AV was no longer mentioned as a standalone narrative.

FY2025

Our competitive position has been harmed by the existing export controls, and our competitive position and future results may be further harmed, over the long term, if there are further changes in the USG’s export controls.

→

FY2026

The export controls applicable to China are complex and address a variety of parameters, including the total processing performance of a chip, the “performance density” of a chip, the interconnect bandwidth of a chip, and the memory bandwidth of a chip. Under the current rules and geopolitical landscape, 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. As of the end of fiscal year 2026,

Competitive and IP disclosures broadened. The FY2023 filing added Baidu and Arm-based CPU suppliers to the competitor list ²⁶, and replaced generic IP protection factors with specific patent expiration dates from March 2023 to June 2045 ²⁷. Capital allocation language shifted from dividend payments and treasury share retirement in FY2022 to restricted stock unit share withholding in FY2023 ²⁸, reflecting a change in how the company returns value to shareholders.

The risk landscape at NVIDIA underwent a profound transformation between FY2022 and FY2026, mirroring the company’s explosive growth from a $26.9B revenue base to $215.9B. The most striking shift was the complete disappearance of COVID-19 as a disclosed risk factor. In FY2022, management warned that the pandemic’s impact on operations and financial performance “may continue to be difficult to measure and predict” ¹. By FY2023, that language was gone, replaced by a new set of fears centered on climate change, customer concentration, and business disruptions ¹. The pandemic risk was not merely downgraded; it was excised entirely, and by FY2024 the company explicitly stated that “economic sanctions and export controls” had replaced the “COVID-19 pandemic” as the primary external shock ². This was not a cosmetic change, it reflected a company that had navigated the pandemic and now faced a fundamentally different set of geopolitical and operational threats.

The single most dominant new fear, appearing in FY2023 and intensifying through FY2025, was the cascade of U.S. export controls targeting China. In FY2023, management added a risk factor about transitioning operations out of China and offering alternative products like the A800 ³. By FY2024, the language hardened dramatically: the company admitted that “our competitive position has been harmed, and our competitive position and future results may be further harmed, over the long-term” ⁴. The scope of controls expanded from specific chips (A100, H100) to a broader set including A800, H800, L4, L40, L40S, and even the RTX 4090, and the affected countries grew to include Saudi Arabia, the UAE, and Vietnam ⁵⁶. Management now warned that “excessive or shifting export controls have already encouraged and may in the future encourage customers outside China and other impacted regions to ‘design-out’ certain U.S. semiconductors” ⁷. This was a direct admission that U.S. policy could structurally disadvantage NVIDIA in global markets. By FY2025, the company disclosed that international revenue had fallen from 56% to 53% of total revenue, and explicitly noted that “the market in China, where our offerings are limited by export controls, is highly competitive and we expect it to remain competitive going forward” ⁸. The numbers confirm the narrative: despite total revenue surging 7x over the period, the China headwind was real and acknowledged.

FY2022

Security breaches, computer malware, social-engineering attacks, denial-of-service attacks, software bugs, server malfunctions, software or hardware failures, loss of data or other information technology assets, and other cyber-attacks are increasingly sophisticated, making it more difficult to successfully detect, defend against them or implement adequate preventative measures.

→

FY2023

Cyber-attacks, including ransomware attacks by organized criminal threat actors, nation-states, and nation-state-supported actors, may become more prevalent and severe. Our ability to recover from ransomware attacks may be limited if our backups have been affected by the attack, or if restoring from backups is delayed or not feasible.

A second major theme was the emergence of AI-specific risks, which appeared in FY2023 and expanded rapidly. Management added a risk factor about demand estimation challenges for new use cases like Omniverse and large language models, warning that “our demand estimates for these use cases and applications can be incorrect and create volatility in our revenue or supply levels” ⁹. By FY2024, the company added risks about AI training data containing “unknown or undetected defects and errors, or reflect unintended bias” ¹⁰, and warned that “AI-related regulation could disproportionately impact and disadvantage us and require us to change our business practices” ¹¹. The company also disclosed receiving regulatory inquiries from the EU, UK, and China regarding GPU sales, supply allocation, and foundation model partnerships ¹². This was a new front of regulatory scrutiny directly tied to NVIDIA’s core growth driver.

Cybersecurity fears escalated in parallel. In FY2022, the risk language was generic. By FY2023, management specifically named “ransomware attacks by organized criminal threat actors, nation-states, and nation-state-supported actors” and warned that “our ability to recover from ransomware attacks may be limited if our backups have been affected” ¹³. This was a material escalation, reflecting the heightened threat environment for a company whose chips power global AI infrastructure.

FY2023

Our results and competitive position may be harmed

→

FY2024

Our competitive position has been harmed, and our competitive position and future results may be further harmed, over the long-term

Supply chain risk evolved from a generic concern to a specific admission of failure. In FY2023, the company disclosed a real product defect in a third-party component embedded in Data Center products that resulted in a recorded warranty liability ¹⁴. By FY2025, management acknowledged that “long manufacturing lead times and uncertain supply and component availability, combined with a failure to estimate customer demand accurately, has led and could lead to mismatches between supply and demand” ¹⁵. This was a direct admission that the company’s own forecasting was inadequate for its growth trajectory.

What stopped being feared was equally telling. The pandemic risk vanished. The risk of failing to integrate acquisitions was replaced by supply chain capacity commitments ¹⁶. The Compute & Networking segment-specific credit risk was replaced by executive retention risk ¹⁷. And the company removed a risk factor about data privacy and security laws, replacing it with indebtedness and change-of-control provisions ¹⁸. This last shift is particularly notable: as NVIDIA’s market capitalization swelled, management became more concerned about governance and debt structure than about privacy regulation.

The overall arc is clear. NVIDIA’s management moved from fearing a pandemic-driven demand collapse to fearing the consequences of its own success: geopolitical backlash, regulatory scrutiny, supply chain strain, and the difficulty of forecasting demand in a market it now dominates. The prose hardened precisely as the metrics soared, a classic sign of a company that knows its risks have shifted from existential to operational, but are no less dangerous for that.

NVIDIA’s management narrative over the five-year period from FY2022 to FY2026 traces a remarkable arc from crisis to dominance, with the language hardening precisely as the numbers improved. The story begins in FY2023 with a company in retreat. Revenue was flat at $27.0B, operating income collapsed 57.9% to $4.2B, and net income fell 55.2% to $4.4B [AUDITED FIGURES]. Management’s explanation was blunt: a $2.17B inventory provision, driven by excess supply of Ampere architecture products “particularly for the expected demand in China” ¹. The cash position had deteriorated from $21.2B to $13.3B ², and the company acknowledged that “our supply…has grown significantly due to current supply chain conditions, complexity of our products, and recent reductions in demand” ³. This was a confession of a demand estimation failure, and management responded by adding granular risk language about “situations that may result in excess or obsolete inventory” ⁴ and by shifting its revenue recognition policy from simple product sales to a five-step model covering licenses, software, and cloud services ⁵. The latter change, while technical, signaled that the company was preparing for a more complex revenue mix even as its core business stumbled.

The pivot was already visible in the same FY2023 filing. Management replaced its Automotive design win narrative with a new focus on enterprise AI cloud services, including NVIDIA DGX Cloud and partnerships with leading cloud providers ⁶. It also reordered its market listing, putting Data Center ahead of Gaming for the first time ⁷. These were not cosmetic changes; they foreshadowed the explosion to come. By FY2024, revenue had surged 125.8% to $60.9B, operating income jumped 680.6% to $33.0B, and operating cash flow rose 398.0% to $28.1B [AUDITED FIGURES]. Management’s tone shifted from apology to triumphalism: “Demand for our data center systems and products surged in fiscal year 2024” ⁸. The gross margin recovered from 56.9% to 72.7% ⁹, and the company began disclosing specific product drivers, Hopper GPU shipments for LLM training and inference, and InfiniBand networking for AI infrastructure ¹⁰. The risk language, however, also hardened. Management added a warning that “our competitive position has been harmed” by US export controls ¹¹ and that demand estimates for generative AI “can be incorrect and create volatility in our revenue or supply levels” ¹². This was a classic pattern: the better the numbers, the more carefully management hedged.

FY2022

Graphics segment revenue increased by 61% in fiscal year 2022 compared to fiscal year 2021. We continue to benefit from strong demand for NVIDIA Ampere architecture products, and believe the increase in Gaming revenue during fiscal year 2022 resulted from a combination of factors, including: the ramp of new RTX 30 Series GPUs; the release of new games supporting ray tracing; the rising popularity of gaming, esports, content creation and streaming; the demand for new and upgraded

→

FY2023

Gross margin for fiscal year 2023 declined from a year ago, driven by $2.17 billion of inventory charges largely relating to excess supply of NVIDIA Ampere architecture Gaming and Data Center products as compared to the demand expectations for these products, particularly for the expected demand in China.

FY2025 saw revenue double again to $130.5B, with operating income of $81.5B and operating cash flow of $64.1B [AUDITED FIGURES]. Gross margins reached 75.0% ¹³, and management introduced a new revenue driver, Ethernet for AI, including the Spectrum-X platform ¹⁴, signaling that the company was expanding beyond its InfiniBand stronghold. Yet the narrative also began to shift toward structural constraints. Management removed its detailed discussion of China Data Center revenue and export control impact, replacing it with a generic section header ¹⁵. This was a notable retreat from transparency on a topic that had dominated the FY2023 and FY2024 filings. The company also added a new risk factor: “The availability of data centers, energy, and capital to support the buildout of NVIDIA AI infrastructure…is crucial” ¹⁶. The problem was no longer demand, but the physical capacity to meet it.

By FY2026, revenue reached $215.9B, operating income $130.4B, and operating cash flow $102.7B [AUDITED FIGURES]. The Compute & Networking segment alone generated $193.5B ¹⁷. Yet management’s language had become almost entirely defensive. The company disclosed $17.5B in private company investments, primarily in early-stage startups ¹⁸, a new and significant exposure to illiquid assets. It replaced its discussion of demand estimates with a detailed debt maturity schedule and commercial paper program ¹⁹, signaling a shift in focus from growth to capital structure. The export control risk language that had been prominent in FY2024 and FY2025 was replaced by a discussion of the One Big Beautiful Bill Act tax law changes ²⁰, and the company disclosed $4.0B in unrecognized tax benefits under IRS examination ²¹. The narrative had come full circle: from a company explaining a demand collapse in FY2023 to a company explaining how it would manage the consequences of its own success in FY2026.

Over the five-year period from FY2022 to FY2026, NVIDIA’s legal and regulatory disclosures in its 10-K filings underwent a notable shift in one specific area: the transparency of insider trading arrangements under Rule 10b5-1. No litigation, settlements, or government enforcement actions appear in the evidence; instead, the company voluntarily expanded its disclosure of director and officer trading plans as its financial scale exploded. This evolution tracks directly with the company’s meteoric growth, revenue rose from $26.9B in FY2022 to $215.9B in FY2026, net income from $9.8B to $120.1B, and suggests that NVIDIA’s governance disclosures hardened in step with its market prominence.

The first material change appeared in the FY2024 filing, covering the transition from FY2023. Where the prior year’s “Other Information” section had simply stated “None.” regarding Rule 10b5-1 trading arrangements, the FY2024 filing disclosed that director John O. Dabiri had adopted a trading plan on December 18, 2023, covering the sale of an estimated 553 shares through December 2, 2024 ¹. This was a binary shift: from zero disclosure to a specific, named insider plan. The timing is instructive. FY2023 had been a difficult year, revenue grew only 0.2% to $27.0B, operating income fell 57.9% to $4.2B, and net income dropped 55.2% to $4.4B. By FY2024, however, revenue had surged 125.8% to $60.9B and net income had rebounded 581.3% to $29.8B. As the company’s valuation and public profile rose, the board apparently judged that even a small insider trading plan merited explicit disclosure.

FY2023

None.

→

FY2024

On December 18, 2023 , John O. Dabiri , a member of our Board of Director s, adopted a trading arrangement that is intended to satisfy the affirmative defense conditions of Rule 10b5-1(c) for the sale through December 2, 2024 of an estimated 553 shares of our common stock, assuming our closing stock price as of January 26, 2024.

The FY2025 filing, covering the year ended January 2025, deepened this trend dramatically. The Dabiri plan was updated with a new adoption date of September 27, 2024, an expiration of March 31, 2026, and a revised share estimate based on the January 24, 2025 closing price ². More significantly, the company added a new introductory paragraph stating that “the following members of our Board of Directors and/or officers adopted, modified or terminated a trading arrangement” under Rule 10b5-1 ³. This broadened the disclosure from a single director to a potentially larger group. The filing also included a table detailing director Aarti Shah’s termination of a trading arrangement on November 25, 2024, covering 29,000 shares ⁴. The inclusion of a termination, not just an adoption, signals that NVIDIA was now tracking the full lifecycle of insider plans, not merely their initiation.

This expansion of insider trading disclosure occurred against a backdrop of continued explosive growth. FY2025 revenue reached $130.5B, up 114.2%, and net income hit $72.9B, up 144.9%. Operating cash flow more than doubled to $64.1B. As NVIDIA’s market capitalization swelled, the potential for insider trading scrutiny increased proportionally. The company’s decision to move from a single director’s plan to a multi-insider framework, complete with termination data, suggests a deliberate effort to preempt regulatory or investor concerns about insider activity during a period of extraordinary stock price appreciation. The evidence does not indicate any enforcement action or shareholder lawsuit that prompted these changes; rather, the disclosures appear proactive.

FY2024

(no equivalent language)

→

FY2025

The following members of our Board of Directors and/or officers adopted , modified or terminated a trading arrangement that is intended to satisfy the affirmative defense conditions of Rule 10b5-1(c), or a Rule 10b5-1 Trading Arrangement:

By FY2026, with revenue at $215.9B and net income at $120.1B, the pattern established in FY2025 likely continued, though the evidence provided does not extend beyond that filing. The trajectory is clear: from a blank “None.” in FY2023 to a detailed, multi-person disclosure regime by FY2025. The absence of any other legal or regulatory changes, no settlements, no government investigations, no export control modifications, is itself a noteworthy finding. NVIDIA’s legal and regulatory exposure during this five-year period was dominated not by external actions but by internal governance transparency. The company chose to harden its prose on insider trading precisely as its financial metrics soared, a correlation that institutional investors should monitor. If the pattern holds, future filings may expand further to include additional insiders or more granular plan details, reflecting the heightened scrutiny that comes with being one of the world’s most valuable companies.