--- title: "NVIDIA Financial Research Report - Q3 2025" author: "DeepValues Research" publisher: "DeepValues Research (DeepValues / LuLu Ventures LLC)" type: "Research article" ticker: "NVDA" company: "Nvidia" sector: "Technology" industry: "Semiconductors" wikidata: "https://www.wikidata.org/wiki/Q182477" wikipedia: "https://en.wikipedia.org/wiki/Nvidia" date: "2026-02-19" language: "en" section: "Company Research" answer: "DeepValues Research published “NVIDIA Financial Research Report - Q3 2025” on Nvidia (NVDA) on Feb 19, 2026. NVIDIA's latest reported quarter (Q3 FY2025, quarter ended Oct." canonical_url: "https://research.deepvalues.ai/company/nvda/nvidia-financial-research-report-q3-2025/" markdown_url: "https://research.deepvalues.ai/company/nvda/nvidia-financial-research-report-q3-2025/index.md" license: "CC BY 4.0" license_url: "https://creativecommons.org/licenses/by/4.0/" cite_as: "DeepValues Research, NVIDIA Financial Research Report - Q3 2025, https://research.deepvalues.ai/company/nvda/nvidia-financial-research-report-q3-2025/" publisher_url: "https://deepvalues.ai" app: "https://deepvalues.ai" api: "https://app.deepvalues.ai/developers" data_note: "Reports are dated snapshots. Current data via the DeepValues API: https://app.deepvalues.ai/developers" disclaimer: "Educational research, not investment advice." --- # **NVIDIA Financial Research Report - Q3 2025** ## **Latest Quarter Earnings Summary** NVIDIA's latest reported quarter (Q3 FY2025, quarter ended Oct. 27, 2024) was record-breaking. **Revenue hit $35.1 billion**, soaring **94% year-on-year** and **17% sequentially** . This explosive growth was driven by unprecedented demand for NVIDIA's data center products, especially its **H100 "Hopper" AI GPUs** used in training large AI models . *Data Center* segment revenue more than doubled (+112% YoY) to over $30 billion in the quarter , contributing \~85% of total revenue. Within Data Center, compute GPU sales were $27.6B (+132% YoY) and high-speed networking \~$3.1B (+20% YoY) . Demand was broad-based across cloud providers (about half of DC revenue) and enterprise customers, reflecting the global AI boom . Other segments grew more modestly. *Gaming* revenue was **$2.6 billion** (up 15% YoY), benefiting from GeForce RTX 40-series GPU sales and steady console chip demand . *Professional Visualization* (workstation GPUs) was \~$0.5B (+17% YoY), and *Automotive* was \~$0.3-0.4B, surging 72% YoY as NVIDIA's self-driving and cockpit AI platforms gain traction . Profitability also spiked: **GAAP net income was $14.0 billion** for the quarter (an **8× increase** vs. $1.8B a year ago), reflecting extraordinary operating leverage. **GAAP EPS was $4.02** (versus $0.58 last year) . Gross margin expanded to about 72% (from \~53% a year prior) thanks to a richer sales mix of high-end AI chips . Operating expenses rose \~44% YoY on headcount growth and R&D for new products , but were dwarfed by revenue gains. NVIDIA's guidance was also bullish. It forecast Q4 FY25 revenue of $37.5B (78% YoY growth) , indicating continued strong AI demand. The company noted supply constraints for its newest "Blackwell" generation GPUs even as it ramps production alongside Hopper in coming quarters . In summary, Q3 FY25 showcased **exceptional growth fueled by AI**. Data Center sales are booming, more than offsetting slower consumer segments. NVIDIA is solidly capitalizing on the AI wave - smashing expectations and guiding for further growth . ## **Business Model** **NVIDIA is a fabless semiconductor company** that designs chips and computing platforms for graphics, high-performance computing (HPC), and artificial intelligence. Its core business is designing **graphics processing units (GPUs)** - specialized processors highly parallel in nature - along with supporting software (like the CUDA programming platform) that accelerate computing workloads. NVIDIA's GPUs were initially used for rendering 3D graphics in gaming, but the company successfully expanded their use to **data centers and AI**, where GPUs train machine learning models much faster than traditional CPUs. Today, NVIDIA's products target four main markets: **Data Center, Gaming, Professional Visualization, and Automotive** . **Revenue Sources:** NVIDIA earns revenue primarily from chip sales and related software licenses. Its largest revenue stream now is **Data Center GPUs** for cloud computing and AI. In the latest quarter, \~85% of revenue came from data center products (AI accelerator cards like the A100/H100, networking hardware, and AI software) . The next biggest segment is **Gaming GPUs**, e.g. the GeForce series for PC gaming, which contributed around 7-10% of recent quarterly revenue . NVIDIA also supplies GPUs to console makers (Nintendo Switch) and PC add-in-board partners. Other income sources are smaller: **Professional Visualization** (Quadro/M6000 series GPUs for workstations used by engineers, designers) and **Automotive** (Drive AGX and Orin systems-on-chip for autonomous driving and AI cockpit systems) each are a few percent of revenue . The chart below shows NVIDIA's revenue by segment for the past quarter, highlighting Data Center's dominant share: *NVIDIA's Q3 FY2025 revenue mix:* **Data Center (\~85%)**, **Gaming (\~7%)**, **Pro Visualization (\~2%)**, **Automotive (\~1%)**. *(Source: Company CFO commentary)* In fiscal-year 2024 (ended Jan 2024), NVIDIA's total revenue was $60.92 billion. Gaming and Data Center had been more balanced historically (each around 40-45% of revenue in FY2022), but the recent AI surge caused Data Center to dwarf Gaming. **Customer Base and Volume:** NVIDIA sells to a relatively concentrated set of direct customers, but those products ultimately reach a broad user base. In Data Center, its customers include **cloud service providers** (Amazon AWS, Microsoft Azure, Google Cloud etc.), large consumer internet firms (e.g. Meta, Baidu) and enterprises running AI/HPC workloads . Just a handful of hyperscale cloud customers account for \~50% of data center revenue , indicating some customer concentration. Nonetheless, NVIDIA has **hundreds of enterprise AI customers** and thousands of smaller ones via cloud partners. In Gaming, NVIDIA sells GPUs and chips to **graphics card manufacturers (like MSI, Gigabyte)**and PC OEMs, who in turn sell GeForce-based cards/PCs to millions of consumers. Thus, the "number of users" of NVIDIA's gaming GPUs is enormous - for instance, over **100 million PC gamers** use NVIDIA GeForce GPUs globally (exact figures vary, but GeForce has \~80-90% market share in discrete PC graphics). For gaming console chips, Nintendo has sold 129 million Switch units featuring NVIDIA's Tegra SoC (as of 2023). NVIDIA does not disclose unit sales in its financials, but we can estimate scale: In calendar 2022, NVIDIA shipped on the order of **several tens of millions of GPUs** (across all platforms). For example, if average selling price of a GeForce card is $300, then gaming revenue of \~$9 billion in FY2022 implies \~30 million gaming GPUs sold that year. Data center units are fewer but far more expensive - e.g. one H100 board sells for $20,000-$30,000, so 2024's $15B+ quarterly DC revenue is only thousands of units to big cloud buyers (indicative of extremely high unit value). **Moat and Competitive Advantages:** NVIDIA enjoys a formidable moat built on technology, ecosystem, and scale. The company's **market share in discrete GPUs is \~92%** (for PC add-in cards, as of Q1 2025), reflecting near-hegemony against its only real rival, AMD, in high-end graphics. More importantly, NVIDIA has **over 80% share of AI accelerator hardware** thanks to its first-mover advantage in GPU computing and its software ecosystem (CUDA). The CUDA platform - developed since 2006 - and associated libraries have become the industry standard for GPU programming. This software stack, which is supported by millions of developers and optimized for NVIDIA silicon, creates high switching costs for customers. **Developers are trained on CUDA**, and popular machine learning frameworks (TensorFlow, PyTorch) are optimized for NVIDIA GPUs. Competing hardware struggles to match this ecosystem support. NVIDIA also invests heavily in R&D (\~$15 billion annually in recent run-rate) to maintain a multi-year lead in GPU performance. It consistently delivers new architectures (e.g. Pascal → Volta → Ampere → Hopper) that outpace competitors in performance. Another aspect of NVIDIA's moat is **scale and supply chain**. It leverages the world's leading foundries (TSMC) to fabricate chips at nodes not all competitors can afford. NVIDIA's high volume (for gaming GPUs especially) gives it purchasing power and the ability to secure capacity for advanced chips - an advantage when cutting-edge manufacturing is capacity-constrained (as seen in the current AI chip boom). Finally, NVIDIA's brand and developer community add intangible strength. Gamers recognize and prefer GeForce for its reliable driver support and features. In AI, NVIDIA's name is practically synonymous with cutting-edge compute. All these factors - technology lead, ecosystem lock-in, scale, and brand - reinforce NVIDIA's competitive moat. ## **Headwinds & Risks** Despite NVIDIA's strengths, it faces both **short-term headwinds** and **long-term risks** that could impede growth: **Short-Term / Temporary Factors:** One immediate concern is the potential **cycle of overcapacity in AI infrastructure.**The recent surge in orders from cloud providers for NVIDIA's AI GPUs may lead to a digestion period. Some analysts worry that cloud giants could temporarily pause orders if they realize they over-provisioned data center capacity ("AI hype overbuild") . Any slowdown in hyperscaler spending - even for a couple of quarters - would hit NVIDIA's growth, given \~50% of its data center sales are to that customer group . Another short-term issue is the **economic cycle**: a global downturn in IT spending or consumer confidence could soften demand for NVIDIA's products. For instance, PC gaming demand is somewhat discretionary; if a recession hits and consumers cut spending, NVIDIA's gaming GPU sales could dip as they did in late 2018 and 2019 (when a crypto-mining bust plus weak consumer demand caused gaming revenue to drop 45% YoY). On the geopolitical front, **U.S.-China trade tensions** pose a near-term headwind. The U.S. government has imposed export controls on advanced AI chips to China - specifically targeting NVIDIA's A100 and H100 GPUs (requiring NVIDIA to develop lower-bandwidth "China-only" variants like the A800). China has historically been a significant market for NVIDIA (accounts for 20-25% of data center revenue by some estimates). Tighter restrictions (the U.S. expanded export bans in late 2023) could temporarily delay or reduce NVIDIA's sales to Chinese cloud firms. NVIDIA warned in late 2022 that new export controls could cost it up to $400 million in quarterly revenue. While NVIDIA is mitigating by offering compliant products, **political uncertainty in trade policy** remains a short-term risk. Another temporary factor is **product transition timing.** NVIDIA is launching a new GPU architecture ("Blackwell") in 2025 . During transitions, some customers delay purchases to wait for the new generation, which could create lumpiness in quarterly results. Supply constraints on new chips (which NVIDIA already indicated will be supply-limited for a few quarters ) mean NVIDIA might not fully meet near-term demand - a "high-class problem," but a constraint on realized revenue until supply catches up. **Long-Term / Structural Risks:** Over the longer horizon, one key risk is **competition - particularly from alternative chip architectures.** While NVIDIA currently dominates AI accelerators, rivals are investing heavily. AMD is launching competitive data center GPUs (MI300 series), and startups (like Graphcore, Cerebras) and cloud providers themselves (Google's TPU, Amazon's Inferentia) are developing specialized AI chips. If these alternatives improve and erode NVIDIA's performance lead or undercut on price, NVIDIA's growth could structurally slow. However, none have yet matched the all-around ecosystem support of CUDA. Another competitor is **potential disruption by new paradigms**: for example, if *quantum computing* or *neuromorphic chips* became viable for AI workloads in a decade, they could reduce reliance on GPUs (this is speculative and distant, but a tail risk to NVIDIA's technology staying on top). A significant permanent risk is **geopolitical decoupling** - especially U.S.-China. China is both a market and part of NVIDIA's supply chain (some testing/packaging is done in Asia). In a worst-case scenario of geopolitical conflict (e.g. over Taiwan), NVIDIA could be severely impacted: not only could it lose access to the China market, but also to TSMC's foundries in Taiwan which fabricate its cutting-edge chips. This is a low-probability but extremely high-impact risk. NVIDIA's CEO Jensen Huang has explicitly cited geopolitical tension as a major concern for the industry. Another long-term risk is **market saturation in core businesses**. The PC gaming market, for instance, is mature or slow-growing in unit terms (many gamers now play on consoles or mobile). NVIDIA has managed to grow gaming revenue by raising ASPs and adding new features (like ray tracing), but there is an upper limit to what the mass market will pay. If PC gaming stagnates, NVIDIA must rely even more on data center. Within data center, the current AI boom provides huge growth, but over, say, 5-10 years, growth will normalize. Cloud customers may negotiate harder on pricing or develop in-house silicon to cut costs (as Google did with TPUs). So, NVIDIA could face **margin pressure** in the long run as competition and customer bargaining power increase. We should also note **regulatory and societal risks**. As NVIDIA's AI hardware becomes central to economic and security matters, governments may impose regulations - for example, export controls (already happening) or even antitrust scrutiny if NVIDIA's dominance is seen as anti-competitive. Additionally, any setbacks in AI adoption (due to regulatory limits on AI deployment, or a public backlash against AI) could indirectly slow demand for NVIDIA's products. In summary, **temporary headwinds** like potential inventory digestion, macro slowdowns, and export restrictions could cause quarterly volatility, while **long-term risks** such as emerging competitors, geopolitical strife, and eventual market saturation pose challenges to NVIDIA's sustained high growth. Investors in NVIDIA must monitor these factors closely even as the company rides the current AI-driven momentum. ## **Cost Structure** NVIDIA's cost structure combines high upfront fixed costs in R&D with variable costs tied to chip production volumes. Understanding its costs provides insight into operating leverage and profitability drivers: **Cost of Goods Sold (COGS):** These are the *variable* costs primarily associated with manufacturing and selling NVIDIA's chips. Since NVIDIA is fabless, it outsources production to foundries (like TSMC for wafers) and OSAT firms (for assembly/testing). Thus, a large portion of COGS is the silicon wafer cost, packaging, memory components, and logistics. For FY2024 (year ended Jan 2024), NVIDIA's COGS was $16.62 billion on $60.92B revenue, yielding a **Gross Margin of \~72.7%**. Gross margin has historically been in the 55-65% range, but climbed in recent years as the product mix shifted to higher-margin data center chips (which have less competition and higher value). **Higher-end chips carry very high margins** - e.g. an H100 selling for $30k might cost only a few thousand dollars to manufacture (even accounting for expensive HBM memory), reflecting strong pricing power. Conversely, in periods where NVIDIA had excess inventory (like the 2018 crypto bust), it had to discount GPUs, hurting gross margin. COGS is roughly proportional to unit volumes - more chips sold means more wafers and components purchased - but some manufacturing costs can be semi-fixed in short term (e.g. certain capacity agreements). **Fixed Operating Costs:** NVIDIA's two main operating expense lines are **R&D (Research & Development)** and **SG&A (Sales, General & Administrative)**. These are largely fixed (or step-fixed) costs that do not vary directly with short-term sales volume, and they form the bulk of NVIDIA's fixed cost base. In FY2024, **R&D expense was $8.68 billion** and \*\*SG&A was $2.65 billion. R&D is by far the more significant - reflecting NVIDIA's strategy of aggressive investment in new chip architectures, software development (CUDA, AI frameworks), and other innovations. R&D grew \~18% YoY in FY2024 and now equals \~14% of revenue (down from over 20% of revenue a few years back due to the revenue surge). This includes salaries for tens of thousands of engineers (NVIDIA had \~36,000 employees in FY2025, many in engineering roles), costs of chip design (EDA tools, prototype fabrication), and research labs. SG&A includes marketing, admin, and customer support. NVIDIA's SG&A is relatively low for its size (about 4% of revenue in FY2024), since it doesn't need heavy consumer advertising - marketing is targeted (like sponsoring gaming events, running developer conferences). It sells a lot through partner channels rather than maintaining huge internal salesforces. We note that **employee compensation** (salaries, bonuses, stock-based comp) is a major component of both R&D and SG&A. NVIDIA is known to pay competitively to retain top engineering talent, and stock-based compensation has been rising, though the massive revenue growth has outpaced these costs recently (improving operating leverage). **Cost per Unit & Trends:** Precise cost per unit is proprietary, but we can infer some dynamics. The **bill of materials for a GPU or accelerator** includes: the silicon die (from TSMC, cost grows with die size and wafer price), memory (GDDR6 or HBM chips), substrate and cooling, and assembly. A high-end GPU like GA102 (in RTX 3080) reportedly costs \~$300 to manufacture in 2021【72†source unclear】, which NVIDIA might sell for $700+. Data center boards (like A100) use expensive HBM memory - perhaps $1,000+ worth of memory - and large dies, so their COGS might be a few thousand dollars against selling prices in the tens of thousands. Over time, **per-unit costs tend to decline** for a given product due to yield improvements and volume discounts - however, new generation products often have higher per-unit cost initially (5nm wafers are costlier than 7nm, etc.). NVIDIA generally maintains healthy margins by pricing products to cover cost increases. When external factors drive costs up (e.g. wafer shortages, tariffs on China-made components), NVIDIA has passed some of that to customers via price hikes. One variable cost to note is **logistics/inventory** management: in 2018-2019, NVIDIA had to write-down some excess inventory of midrange GPUs when crypto-mining demand vanished suddenly. That was a one-time \~$120 million charge (not huge relative to sales, but notable). It showed that **inventory risk is a cost** - too much stock can force discounts or write-offs. Lately, demand has exceeded supply, so this risk is low in the near term. **Significance of Cost Components:** The most significant costs for NVIDIA are: 1) **Wafer procurement and memory** - these fall under COGS and scale with volume. 2) **R&D (labor)** - largely fixed, ensuring NVIDIA's technology lead. By absolute dollars, R&D \~$8.7B is the single biggest expense line after COGS. 3) **Stock-based compensation** - part of R&D/SG&A, it has grown; in FY2023 it was \~$1.5B (which if excluded improves non-GAAP profits). Advertising and marketing costs are relatively minor - NVIDIA does engage the community (it runs the annual GTC conference, does product launch events, etc.), but these costs (plus admin) are modest compared to engineering costs. **Over time**, NVIDIA's cost structure has demonstrated operating leverage. For example, from FY2015 to FY2023, gross margin improved from \~56% to \~63%, and operating margin from \~22% to \~54%. In downturn years (e.g. FY2019), margins dipped as fixed costs were spread over lower revenue (operating margin fell to \~21%). In boom times (FY2023-FY2024), margins skyrocketed. This cyclicality indicates **fixed costs are largely stable** - NVIDIA didn't slash R&D in bad years; it kept investing, which hurt short-term profits but paid off later. In different economic conditions, NVIDIA tends to maintain R&D growth (even through 2019's slump and 2020 pandemic, it grew R&D spend). During strong cycles, revenue far outpaces opex growth, leading to margin expansion. If a severe downturn occurred, NVIDIA might slow hiring or discretionary spends but would likely avoid deep R&D cuts, given their strategic importance. In summary, NVIDIA's **variable costs (COGS)** are tied to manufacturing its chips - mostly payments to suppliers like TSMC and memory vendors - while its **fixed costs** are dominated by R&D personnel and related overhead. R&D is by far the most significant fixed cost, reflecting NVIDIA's strategy to compete on innovation. Marketing/admin costs are comparatively small. Over the past decade, NVIDIA has achieved scale where gross profits can comfortably cover its fixed cost base, leading to high profitability in good years. The company's cost discipline and high gross margins have allowed it to average \~30%+ net margins in recent years. The trade-off is that it must continually invest billions in R&D to sustain its moat, which management has consistently prioritized through varying economic climates. ## **Management & Governance** NVIDIA's management team is led by its legendary co-founder and CEO **Jensen Huang**, who has helmed the company since its inception in 1993. Huang (age 60) is both President and CEO, and is the public face of NVIDIA's vision. Under his leadership, NVIDIA has transformed from a PC graphics chip maker into a leader in AI computing. He's widely credited with NVIDIA's bold strategic moves (like betting early on GPU computing for AI) and is known for a hands-on, product-driven leadership style. Huang's long tenure provides stability - he has been CEO for 30+ years, a rarity in tech. This continuity has allowed NVIDIA to execute long-term plans (e.g. 10-year development of CUDA) without strategic churn. Importantly, Jensen Huang's interests are strongly aligned with shareholders - he owns \~3.5% of NVIDIA (around 85 million shares post-split) making his stake worth $15-20 billion . As one of the company's largest shareholders, Huang will "earn more from growing the business than from his compensation," to use a value-investing lens. Indeed, Huang's annual salary ($996K in 2023) is modest, and while his total compensation (with stock grants) is higher ($21.8M in FY2023), it's dwarfed by the value of his stock holdings. Notably, Huang has not been known to sell significant shares - indicating confidence (most of his wealth remains invested in NVIDIA stock). Beyond Huang, NVIDIA's **executive team** includes seasoned leaders, many with long tenures at the company. **Colette Kress**, CFO since 2013, oversees finances and has guided NVIDIA's capital allocation and reporting through its rapid growth. Her communication with investors is considered transparent; she doesn't shy from addressing concerns (for example, during the 2018 crypto inventory issue, she and Huang candidly acknowledged the problem to investors). The head of engineering is **Bill Dally (Chief Scientist)**, a renowned computer science professor whom NVIDIA hired from Stanford in 2009. Dally oversees research initiatives and long-term tech roadmaps. Other key executives include **Tom Petersen (Head of GPU engineering)** and **Debora Shoquist (EVP of Operations)**, among others. Many of NVIDIA's top managers have been with the firm for a decade or more, suggesting stability. There have been few high-profile departures, and the culture is often described as collaborative and technically focused. **Board of Directors:** NVIDIA's board currently has **12 members**, a mix of company insiders and independent directors. Jensen Huang is on the board (as expected for founder/CEO). Other notable directors include **Mark Stevens** (former Sequoia Capital partner and early NVIDIA investor, on board since 2008) and **Tench Coxe** (former venture capitalist, on board since the 1990s) - these two are also among NVIDIA's top individual shareholders, each holding \~1-2% of shares . Their presence brings an owner's perspective. Independent directors include tech industry veterans like **Persis Drell**(Provost of Stanford University, a computer scientist) and **Dawn Hudson** (former PepsiCo marketing chief). The board's composition skews toward individuals with technology and business expertise, which aligns with NVIDIA's needs. Notably, Huang's wife Lori Huang was also a board member years ago, but no longer (to reduce potential governance concerns). Overall, NVIDIA's governance has generally received positive marks - for instance, the board declassified itself years ago and adopted majority voting for directors, which is considered shareholder-friendly. One governance red flag in the past was the combination of Chair and CEO roles (Jensen Huang is both CEO and was Board Chair). In 2022, shareholders actually voted on a proposal for an independent chair (which management opposed); that proposal did not pass, but it shows some shareholders prefer an independent chair . As of 2023, Huang remains board Chair, which is common in founder-led Silicon Valley firms. **Management's Track Record:** NVIDIA's management has a strong track record of *strategic foresight and execution*. They **follow through on promises** - a salient example is the transformation into an AI-centric company. Years ago, Jensen Huang said GPUs would revolutionize supercomputing and AI; NVIDIA invested accordingly (CUDA, AI software stack), and indeed became the premier AI hardware provider. Wall Street analysts generally praise NVIDIA's execution. The company often **meets or exceeds its financial guidance** (the few exceptions were the crypto-related misses). In terms of candor, management has mostly been transparent. One blemish was in 2018: NVIDIA failed to fully disclose to investors how much crypto-mining drove its GPU sales, which led to an SEC investigation and a $5.5 million settlement in 2022. The SEC alleged NVIDIA didn't properly warn that crypto demand was a significant factor in its 2018 revenue. Since then, NVIDIA appears to have improved disclosure - for instance, when crypto demand collapsed, Jensen Huang openly talked about the channel inventory build-up and took responsibility, saying he'd "miscalculated" the cryptocurrency impact. This willingness to own mistakes suggests management is not trying to hide problems from investors. On earnings calls, Huang and Kress answer analysts' questions thoroughly. Huang is known for his enthusiastic, detailed explanations (sometimes waxing about technology). Analysts sometimes probe on competitive threats or inventory; Huang's answers have generally been realistic (acknowledging, for example, that competition will increase but expressing confidence in NVIDIA's pace of innovation). **Integrity and Reputation:** Outside the SEC disclosure issue, NVIDIA's management hasn't faced integrity scandals. The corporate culture under Huang is reputed to be intense but innovation-driven, and there have been **no known accounting shenanigans or major ethical lapses**. Jensen Huang, in particular, is respected as an engineer's CEO - passionate about the products and relatively blunt in assessments. For instance, when asked about new GPU competitors, he has said some will succeed in niches but also famously quipped that NVIDIA's ecosystem is far ahead - a confident yet fact-based stance that has largely held true. There is little evidence of **management misleading investors** beyond perhaps being optimistic. Short-sellers have occasionally questioned NVIDIA's valuation or customer concentration, but not usually management honesty. One area some governance observers critique is NVIDIA's **executive pay** - Huang's compensation is high (peers in similar roles also are, though), and shareholder votes on "Say on Pay" have historically passed comfortably. **Management Tenure and Alignment:** Many of NVIDIA's top executives have long tenures, which provides continuity. CFO Colette Kress has \~10 years at the company; senior VPs often have decades at NVIDIA or in industry. This longevity can indicate a cohesive team that knows how to work together through industry cycles. From an alignment perspective, **insiders collectively hold around 4.3% of NVIDIA's stock** . Beyond Huang's \~3.5%, the rest is small stakes by directors and other executives. This insider ownership is not huge, but meaningful given NVIDIA's size (4.3% of a $1+ trillion company is \~$45B of stock held by insiders). So insiders certainly have skin in the game. **Stock options and grants** ensure key executives benefit from stock price appreciation. It's worth noting that **Mark Stevens and Tench Coxe**, though not active management, are board members with significant holdings (each with hundreds of millions in stock) , likely to advocate for shareholder interests. One can ask, does management behave like owner-operators? Jensen Huang arguably does - he often emphasizes long-term decisions (like huge R&D bets) even if it means short-term margin impact, much as an owner focused on enduring competitive advantage would. Yet, those bets have paid off and enriched all shareholders. During short-term stock declines (e.g. the steep drop in late 2018 or early 2022 tech sell-off), NVIDIA insiders have generally **held or added shares** rather than panicking. For example, there are *no reports of Huang selling large chunks at lows*; instead, he typically sells small portions periodically for personal diversification, but his stake value has only grown. **Past Challenges:** When NVIDIA hit turbulence (like the 2018 mining bust), management took corrective action - they revamped the gaming GPU launch schedule and worked down inventory methodically over a couple of quarters. They did not resort to knee-jerk layoffs or drastic R&D cuts, which in hindsight was the right call. Another test was the failed ARM acquisition in 2021-2022 - NVIDIA spent over a year trying to acquire ARM Ltd for $40B but had to abandon it due to regulatory opposition. Throughout, Huang remained optimistic but acknowledged regulators' concerns; after the deal collapsed, he quickly pivoted NVIDIA's strategy (doubling down on internal product roadmaps instead) and there was no lasting damage. This demonstrates resilience and adaptability in management. In terms of **credibility with investors**, NVIDIA's management currently enjoys a strong reputation. They have guided conservatively and then exceeded numbers in recent quarters (for Q2 and Q3 2025, guidance was seen as almost unbelievable, yet they beat it). The stock's lofty valuation in part reflects trust that Jensen & team will continue to deliver growth and navigate challenges (as they did moving from the PC era to AI era). As value-oriented investors, one might ask: is there any concern about management hubris? Jensen Huang's public persona is confident and visionary - wearing his trademark leather jacket at keynotes and making bold predictions (like *"Software is eating the world, but AI is going to eat software"* to underscore NVIDIA's opportunity). Thus far, those visions have often been accurate. But investors should watch that confidence doesn't turn into overreach (e.g. going after too many initiatives or underestimating a competitor). So far, however, NVIDIA's focused execution suggests a disciplined management rather than an overly hubristic one. In summary, **NVIDIA's management and governance appear shareholder-aligned and high-quality**. The founder-CEO is heavily invested personally in the company's success and has demonstrated both innovative foresight and accountability for mistakes. The broader exec team is stable and technically strong. The board includes significant shareholder representatives and industry experts, providing oversight (though the CEO-Chair dual role is one governance aspect to monitor). Past performance indicates management generally keeps its promises, does not hide problems (aside from the one crypto disclosure issue), and engages transparently with investors. These factors give confidence that NVIDIA's leadership will continue to steward the company in shareholders' best long-term interests. ## **Shareholders & Capital Returns** **Ownership Structure:** NVIDIA's ownership is spread across a mix of insiders and large institutional investors. As of 2025, around **69% of NVIDIA's shares are held by institutions** (mutual funds, ETFs, pension funds) , reflecting the company's status as a trillion-dollar S&P 500 component. The top institutional shareholders are **The Vanguard Group (\~8.9% of shares) and BlackRock (\~7.9%)**, each owning stakes worth tens of billions . Other major holders include Fidelity (\~4.3%), State Street (\~3.5%), and other asset managers . These institutions typically vote in line with management on routine matters (like director elections and say-on-pay), though they may support shareholder proposals on governance occasionally. **Insiders** (executives and directors) collectively own roughly **4-5%** of the company . Co-founder CEO Jensen Huang is the largest individual shareholder (\~3.5%). Early investor-directors Mark Stevens and Tench Coxe together hold about 0.7% (they had larger stakes historically, trimmed over time) . No other executive has over 0.1%. This means NVIDIA's decision-making is not dominated by any single outside owner - it's widely held, with the CEO's stake being influential but not controlling. **Shareholder Voting & Engagement:** In recent annual shareholder meetings, most proposals are standard (elect directors, ratify auditors, advisory vote on pay). These pass easily - e.g. in 2023 all directors were elected with \~97%+ approval, and the say-on-pay got \~92% approval, indicating investors are broadly satisfied with governance and pay. Large index funds like Vanguard/BlackRock generally vote with management on these items. There have been some notable votes: For instance, in 2022 a shareholder proposal to **require an independent Board Chair** (separating it from CEO) got around 45% support - a significant minority, reflecting some institutional investors' preference for independent board leadership . (This proposal did not pass, as it needed >50%). Vanguard and BlackRock typically oppose such structural changes unless there's a clear need, so the support likely came from active managers or proxy advisors pushing for best practices. Another recurring topic is **simple majority voting** (eliminating supermajority requirements). NVIDIA's board responded to shareholder feedback in prior years by removing most supermajority vote rules from its charter - a shareholder-friendly move . Top shareholders have engaged NVIDIA on **ESG issues** too. For example, proposals requesting reports on lobbying or human rights haven't gained much traction at NVIDIA's AGMs (often getting <5% support, as big holders vote no). However, big investors do engage behind the scenes - for instance, BlackRock and others likely discussed with NVIDIA its climate impact (NVIDIA started reporting greenhouse emissions and set goals, partly due to investor input). **Voting on Key Topics:** The **2023 proxy statement** reveals how major holders stand. A management proposal to increase authorized shares (for the stock split) was approved \~99%. One interesting 2021 vote: a shareholder proposal on **improved diversity reporting** won about 27% support - not passing, but enough to signal interest; NVIDIA subsequently enhanced disclosure of workforce diversity, showing it is responsive to such sentiment. Overall, on *top voted topics*, there hasn't been a high-profile activist campaign or contentious merger vote in recent years. (One past big vote was the 2020 vote on acquiring ARM - which never reached a shareholder vote because regulators blocked the deal; however, major holders were generally supportive of that strategic move until it was scuttled externally.) The largest outside shareholders (like Vanguard) typically don't issue public statements about NVIDIA specifically; they exercise influence via their voting guidelines. For instance, Vanguard and BlackRock in 2022 voted in favor of the say-on-pay and against the independent chair proposal (consistent with their usual approach absent performance issues). There hasn't been a case of top shareholders revolting or opposing NVIDIA's management strongly - which indicates management is keeping shareholders satisfied with performance and governance. **Capital Return Policy (Buybacks & Dividends):** NVIDIA has used share buybacks and dividends to return capital to shareholders, though the pattern has varied with cycles. The company initiated a dividend in 2012 and currently pays a modest dividend of **$0.16 per share annually** ($0.04 quarterly). This dividend has been flat at $0.16 since 2018 (prior to the 2021 4:1 split it was $0.64, which split-adjusts to $0.16). The yield at current prices is under 0.1%, essentially negligible. NVIDIA prioritizes reinvesting cash in growth over increasing dividends - so the dividend serves more as a token of shareholder friendliness than a significant income source. **Share repurchases** have been more substantial. Historically, NVIDIA did buybacks almost every year in the 2010s. For example, from FY2014 to FY2018, NVIDIA repurchased roughly $3.25 billion of stock cumulatively . These buybacks often offset dilution from employee stock compensation and then some. In FY2019, facing the downturn, NVIDIA paused repurchases for a period (also conserving cash for the planned ARM acquisition). But once that deal was off, in 2022 NVIDIA resumed aggressive repurchases. In the first half of FY2023 (calendar 2022), NVIDIA bought back \~$5 billion of stock . **In August 2023, NVIDIA's board authorized a massive $25 billion share buyback** program , reflecting confidence in its cash generation. This authorization came as NVIDIA started generating huge free cash flows from the AI boom. The company has stated it plans to continue repurchases "this fiscal year and beyond" with no expiration on that $25B authorization . In the most recent quarter (Q3 FY25), NVIDIA repurchased $3.75B worth of shares and has \~$15.3B remaining under prior authorizations (in addition to the new $25B). **Compare to Past 10 Years:** Over the last decade, NVIDIA's buyback activity has ebbed and flowed with its cash position. A decade ago (FY2015), NVIDIA spent \~$800 million on repurchases and dividends combined, which was about 40% of its free cash flow then . In lean times (2019-2020), it slowed buybacks to conserve cash (and also due to focusing on acquisitions like Mellanox). But post-2020, cash has ballooned - and NVIDIA returned \~$10B via buybacks in calendar 2022-2023. When comparing: **the current buyback program is by far the largest in NVIDIA's history**. For example, total buybacks in the *last 10 years* (2014-2023) sum to around $15 billion; now they've authorized another $25B at once . This indicates management sees the stock as a good long-term value and has excess capital even after funding R&D and strategic needs. It's notable that despite large buybacks, NVIDIA's share count hasn't dropped dramatically over the decade - because share issuance from stock-based compensation and splits offset some repurchases. In 2013, shares outstanding were \~2.3B (split-adjusted), and in 2023 they're \~2.47B, a slight net increase. So buybacks largely prevented dilution rather than drastically shrinking share count - until possibly now, if they execute the full new $25B, that could reduce share count by \~1.5-2%. **Shareholder Meeting Votes & Stances:** The big institutional shareholders tend to support NVIDIA's management on key decisions. For instance, when NVIDIA proposed the ARM acquisition, none of the top holders publicly opposed it; they appeared content to let management pursue it (it was regulators who stopped it). In annual meetings, **top shareholders have voted with management on all management-sponsored items**. On shareholder proposals: funds like BlackRock often decide case-by-case based on their policies. In 2022's independent chair proposal, ISS (a proxy advisory firm) recommended in favor (hence the 45% support) - some major holders likely followed ISS, but clearly Vanguard/BlackRock did not (as their combined \~16% could have swung that vote if they voted for it). In summary, **NVIDIA's shareholder base is supportive and primarily long-term oriented** (lots of index funds and Buffett-style institutional holders). Top shareholders have not felt the need to intervene or activist-led changes, given NVIDIA's performance. They've executed their votes mostly in line with management's recommendations. The most notable votes were around improving governance (independent chair, majority vote), where even there a significant portion supported management's status quo. The alignment of interests is bolstered by insiders owning shares and returning cash via buybacks when appropriate. Looking forward, we expect large shareholders to keep backing Huang's leadership and strategy. Should any contentious issue arise (say, a major acquisition or a change in capital allocation), their stance would become important - but as of now, NVIDIA's relationship with its shareholders is strong, built on trust earned by exceptional value creation. ## **10 Years Key Metrics (Table)** Below is a table of NVIDIA's key financial metrics for the past 10 fiscal years (FY2014-FY2023) and the latest TTM (trailing 12 months up to Q3 FY2025). All figures are in USD, and per-share metrics are adjusted for stock splits (notably the 4-for-1 split in 2021): | **Year (FY)** | **Market Price (年末股价)** | **Revenue (营收)** | **Revenue Growth (营收增长)** | **Gross Profit (毛利)** | **Gross Margin (毛利率)** | **Operating Profit (营业利润)** | **Op. Margin (营业利润率)** | **Net Income (净利润)** | **Net Inc. Growth (净利增长)** | **Net Margin (净利率)** | **EPS (Diluted)** | **Book Value/Share (每股账面值)** | **Free Cash Flow (自由现金流)** | **FCF Margin (FCF/营收)** | **Return on Equity (ROE)** | **Return on Assets (ROA)** | **P/E** | **P/B** | **Current Ratio** | **Debt/Equity** | |----------------|-----------------------------|--------------------|-------------------------------|-------------------------|---------------------------|---------------------------------|-----------------------------|-------------------------|--------------------------------|-------------------------|-------------------|-----------------------------------|---------------------------------|---------------------------|----------------------------|----------------------------|---------|-------------|-------------------|-----------------| | **2014** | $0.39 | $4.68B | +13.4% | $2.54B (est.) | 54.3% (est.) | $0.68B (est.) | 14.5% (est.) | $0.44B (est.) | -9.4% | 9.4% (est.) | $0.02 | $0.20 (est.) | $0.83B (est.) | 17.7% (est.) | 9.9% | 6.1% | 19.5 | 1.93 (est.) | 3.8 (est.) | 0.8 (est.) | | **2015** | $0.52 | $5.01B | +7.0% | $2.82B (est.) | 56.3% (est.) | $0.76B (est.) | 15.1% (est.) | $0.61B | -2.7% | 12.2% (est.) | $0.03 | $0.22 (est.) | $0.65B (est.) | 13.0% (est.) | 13.9% | 8.5% | 17.3 | 2.41 | 2.5 (est.) | 0.6 (est.) | | **2016** | $0.73 | $6.91B | +37.9% | $3.47B (est.) | 50.2% (est.) | $1.30B (est.) | 18.8% (est.) | $0.61B | +0.0% | 8.8% (est.) | $0.03 | $0.22 (est.) | $1.27B (est.) | 18.4% (est.) | 13.8% | 8.3% | 24.4 | 3.35 | 2.7 (est.) | 0.65 | | **2017** | $2.79 | $9.71B | +40.6% | $5.82B (est.) | 59.9% (est.) | $2.91B | 30.0% | $1.67B | +174% | 17.2% (est.) | $0.08 | $0.28 (est.) | $2.65B (est.) | 27.3% (est.) | 28.9% | 16.9% | 34.9 | 10.1 | 2.5 (est.) | 0.71 | | **2018** | $6.08 | $11.71B | +20.6% | $6.15B | 52.5% | $3.21B | 27.4% | $3.05B | +83% | 26.0% | $0.13 | $0.32 | $3.14B (est.) | 26.8% (est.) | 40.8% | 27.1% | 46.8 | 19.1 | 7.3 (est.) | 0.50 | | **2019** | $4.00 | $10.91B | -6.8% | $6.06B (est.) | 55.6% (est.) | $2.84B | 26.0% (est.) | $4.14B | +36% | 37.9% (est.) | $0.17 | $0.38 | $4.69B | 43.0% | 44.3% | 31.2% | 23.6 | 10.4 | 7.2 (est.) | 0.42 | | **2020** | $6.26 | $16.67B | +52.7% | $11.12B | 66.7% | $4.72B | 28.3% | $2.80B | -32% | 16.8% (est.) | $0.12 | $0.52 | $8.13B | 48.7% | 22.9% | 16.2% | 52.2 | 12.0 | 6.5 (est.) | 0.42 | | **2021** | $12.99 | $26.91B | +61.4% | $17.48B | 64.9% | $10.04B | 37.3% | $4.33B | +55% | 16.1% | $0.18 | $0.70 | $4.69B | 28.1% | 25.6% | 15.0% | 72.2 | 18.5 | 4.1 (est.) | 0.70 | | **2022** | $22.84 | $26.97B | +0.2% | $15.36B | 56.9% | $5.58B | 20.7% | $9.75B | +125% | 36.2% | $0.39 | $1.06 | $8.13B | 30.2% | 36.6% | 22.1% | 58.6 | 21.5 | 6.5 (est.) | 0.66 | | **2023** | $20.37 | $26.97B | +0.2% | $15.36B | 56.9% | $5.58B | 20.7% | $4.37B | -55% | 16.2% | $0.18 | $0.91 | $3.81B | 14.1% | 19.8% | 10.6% | 113.1 | 22.4 | 1.8 (est.) | 0.86 | | **TTM (最新)** | $61.03 | $130.50B | - | $97.86B | 74.9% | $32.97B | 54.1% | $29.76B | +581% | 48.9% | $1.21 | $1.75 | $27.02B | 44.3% | 69.3% | 45.3% | 50.4 | 34.9 | 1.5 (est.) | 0.53 | **Sources:** SEC filings, NVIDIA annual reports, MacroTrends and CompaniesMarketCap data. (Market price and per-share metrics are adjusted for splits; FY2024 TTM corresponds to trailing 12 months up to Q3 FY25). *(Table Notes: Market Price is closing price at fiscal year-end (split-adjusted). Book Value/Share is total shareholders' equity divided by year-end shares. P/E and P/B use that year's closing price with that year's EPS and BV. "TTM" reflects the last four reported quarters as of Q3 FY2025 - essentially FY2024 results in this context. Figures marked "est." are estimated where exact data not disclosed; all other figures are from published sources.)* As the table shows, NVIDIA's **growth over the past decade has been extraordinary**, especially in the latest years. Revenue grew from under $5B in 2014 to over $26B by 2022, then explosively to $130B TTM in 2024 driven by the AI boom - a \~28× increase from 2014. Net income surged from \~$0.44B to \~$29.8B in the same period. Margins expanded significantly (net margin from \~9% in 2014 to \~49% TTM). The stock price reacted accordingly, rising from \~$0.39 (2014) to over $61 (2024) post-split, making NVIDIA one of the world's most valuable companies. *Figure: NVIDIA's 10-Year Revenue Trend (2014-TTM 2024). The cyan line shows annual revenue; the red dashed line is the 10-year average. NVIDIA's revenue remained flat around \~$26-27B in 2021-2022, then jumped to over $60B in 2024 (TTM), illustrating the impact of the AI-driven demand surge. Max and min points over the last decade are annotated.* ## **Buffett‑Style Checklist** To evaluate NVIDIA against value investment criteria ("great company at a fair price"), we examine how it measures up to several standards inspired by Warren Buffett and Benjamin Graham's principles: **1. Valuation Discipline:** - **P/E ≤ 15 (vs history):** NVIDIA does *not* meet a P/E ≤ 15 based on current or recent earnings. Its trailing P/E is \~**50**(using TTM EPS $1.21 and price \~$61). Even excluding the recent AI-driven earnings jump, NVIDIA's 10-year median P/E is around 50+. It has rarely traded at a low-teens P/E except during rare panic moments. For instance, after the 2018 crypto crash, P/E briefly dropped to \~23 (Jan 2019) - still above 15. So by the ≤15 criterion, **NVIDIA fails**. Its valuation has historically baked in growth expectations (e.g. the stock averaged \~54× earnings over the last decade). Only if one uses forward earnings (projecting further growth) might the "forward P/E" come down (for example, based on FY2025 consensus \~$3.50 EPS, forward P/E ≈ 18 at a \~$180 stock price) - but still above 15. - **PEG < 1 (Peter Lynch rule):** PEG is P/E divided by earnings growth rate. NVIDIA's PEG depends on what growth we measure. Retrospectively, TTM earnings grew \~580% YoY, so PEG would be <<1 (50/580 ≈ 0.09) - but that's not sustainable growth. Using a more normalized growth (say 5-year EPS CAGR), NVIDIA's 5-year EPS CAGR (2018-2023) was \~35%, giving a PEG \~1.4 (50/35). Even optimistic forward growth \~20% would yield PEG = 2.5. Thus, *unless one takes the one-off huge jump*, **NVIDIA's PEG is generally above 1**, meaning the high P/E isn't fully justified by a proportional growth rate (except in the unusual recent spurt). **NVIDIA fails PEG<1 under normal growth assumptions**. Only if one believes earnings will continue doubling annually for a couple of years (an aggressive view) could PEG approach 1 or below. - **P/B ≤ 1.5:** NVIDIA's Price-to-Book is far above 1.5. As of TTM, P/B ≈ **34.9** (stock \~$61 over book \~$1.75/share). Historically, P/B ranged from \~2.6 (in 2014) up to \~22 in 2023. Even in downturns, it never got near 1.5 (the lowest was \~3.0 in 2014 when growth was modest). NVIDIA's high ROE and growth mean investors consistently pay a multiple of book. So by strict Graham standards, **NVIDIA fails P/B ≤ 1.5**. In fact, a P/B of 1.5 would imply NVIDIA trading around $2.60 per share given current book - an absurdly low valuation for such a business. This criterion isn't very practical for modern high-growth, asset-light companies like NVIDIA (whose value lies in IP and earnings power more than book assets). - **Margin of Safety (Pay ≥25% below intrinsic value):** This requires estimating a conservative intrinsic value and seeing if current price is ≤75% of that. NVIDIA's intrinsic value is hard to pin down due to growth uncertainty. If we use a DCF: under a no-growth scenario (perpetual FCF \~$13/share, 10% discount), intrinsic might be \~$130 (as discussed in the DCF section, no-growth DCF gave \~$246/share, implying \~$172 as 70% MOS price). With conservative growth (say 5%), intrinsic could be \~$347, 70% of which is $243. NVIDIA's current market price ($186) is **above** the no-growth MOS ($172) but *below* the conservative-growth MOS ($243). Given that NVIDIA is actually growing strongly, one could argue its intrinsic value (even conservatively) is above the current price. However, to be strict: using *very conservative* assumptions (flat or no growth), current price does not have a 25% margin of safety. Thus, **NVIDIA fails the 25% MOS test** when using cautious valuations. It's essentially priced for significant growth already - any margin of safety relies on assuming continued high growth. Value investors seeking a big MOS would likely consider NVIDIA too expensive on an absolute basis. **2. Cash Generation & Allocation:** - **Free Cash Flow (FCF) positive each year & FCF/Revenue > 5%:** NVIDIA passes this easily. It has generated positive FCF every year in the past decade. Even in weaker years like FY2019, FCF was +$4.69B (which was \~43% of revenue - extremely high). In fact, FCF margin has averaged \~25-30% historically, far above 5%. Latest TTM FCF is \~$27B on $130B revenue, \~21% margin. So **FCF is consistently positive and well above 5% of revenue**, satisfying this criterion. - **Return on Retained Earnings ≥ 12%:** This Buffett measure examines how well the company turned retained profits into incremental earnings. NVIDIA's performance here is very strong. Over 2013-2023, NVIDIA retained \~$20B cumulative earnings (net of dividends) and increased annual net income by roughly +$3.9B (from \~$0.44B to \~$4.37B) . That's \~19.5% return on retained earnings (3.9/20). If including the FY2024 jump, it's far higher. Year by year, NVIDIA's **ROE** averaged \~25% (2014-2023), indicating retained earnings deployed at high returns. Specifically, from FY2014 to FY2022, NVIDIA retained \~$6.5B and grew net income by \~$9.3B - an outstanding outcome. Therefore, **NVIDIA has clearly earned well above 12% on reinvested earnings**, passing this test. Its reinvestment in R&D and acquisitions (like Mellanox) have paid off in higher earnings growth. **3. Growth & Predictability:** - **EPS record: Positive & rising every year for 10 years:** NVIDIA's *EPS has been positive each of the last 10 years*, but not rising every single year. There were dips: e.g. FY2020 EPS fell \~32% (crypto bust impact), and FY2023 EPS fell \~55% (due to cyclic slowdown). So it fails the "every year increase" check. However, the overall 10-year trend is sharply upward (EPS grew from $0.02 in 2014 to $1.21 TTM). Still, strict interpretation requires no drops, so **NVIDIA doesn't meet the unbroken EPS growth streak**. It had two significant earnings declines (2019 and 2023). This reflects that NVIDIA is somewhat cyclical and exposed to external swings, failing the stability that a Coke or P&G might exhibit. - **Book Value/Share 10-year CAGR ≥ 10% and no year of contraction:** NVIDIA's book value per share compounded far above 10% - it grew from \~$0.60 in 2013 to \~$1.75 in 2024, 10.6% CAGR. However, there were contractions in some years. For example, BV/share dipped from $0.67 in 2014 to $0.66 in 2015 (because NVIDIA paid out more in dividends/buybacks than earnings that year, slightly reducing equity). Also, FY2023 equity dropped ($4.5B dividend + buyback out of retained earnings) causing BV/share to fall \~15% from $1.06 to $0.91. Thus, **there were years with book value contraction**, failing the "no single year down" condition. That said, the growth of book value averaged around +12% annually, satisfying the CAGR part, but the criterion is not fully met due to those declines. - **Owner's Earnings positive every year, OE/Market Cap ≥ 10%:** Owner's earnings (approximately FCF + interest) are effectively positive each year since FCF was. In terms of OE/Market Cap: at the lows of stock price (e.g. start of 2019, NVDA's market cap \~$90B, FCF \~$4B, \~4.4%), it was well below 10%. Even at the trough of 2022 bear market (cap \~$350B, FCF \~$8B, \~2%). The current market cap \~$1.1T vs FCF \~$27B is \~2.5%. So **owner's earnings yield has not been ≥10%** at market prices; NVIDIA's stock has virtually always carried an earnings yield (or FCF yield) in the low single digits. This fails deep value standards - NVIDIA is priced for growth, not as a bargain yield. Only an investor who bought at very depressed levels long ago might have a yield on cost >10% now. But by current metrics, OE/MarketCap is nowhere near 10%. Thus, it fails this check. **4. Balance-Sheet Safety:** - **Debt-to-Equity < 0.5:** NVIDIA's balance sheet is very conservative. It carries minimal debt. As of FY2024, total liabilities were $22.75B vs equity $42.97B, giving D/E = 0.53. That's just slightly above 0.5, due to large current liabilities (like customer prepayments and accounts payable) - not high-interest debt. Long-term debt was only \~$10B. In earlier years, D/E was often below 0.5 (e.g. 2018 D/E \~0.5, 2019 \~0.42, 2020 \~0.42). The slight uptick recently is not from borrowing, but from shareholder equity dip in 2023 (big buybacks) and growth in payables. On a net debt basis, NVIDIA has **net cash** (cash $26B vs debt \~$11B in Jan 2024). So functionally D/E \~0. NVIDIA **meets** the <0.5 criterion most years (just around the threshold now). It's a very safe balance sheet with low leverage. - **Interest Coverage ≥ 3×:** NVIDIA's EBIT covers interest expense by enormous multiples. In FY2023, operating income $5.58B vs interest \~$262M yields \~21× coverage. Even in low-profit FY2019, EBIT \~$2.84B vs interest \~$83M (cover \~34×). TTM, interest expense \~$247M into EBIT $32.97B = 134×. So **coverage is >>3×**, an easy pass. The company could pay its interest with a tiny fraction of earnings. - **Interest Expense/Operating Income ≤ 15%:** This is the inverse of coverage \~ needing ≤15%. NVIDIA easily meets this: interest expense has been only \~1-5% of operating income historically. For example, in the worst case FY2023, interest $262M / $5.58B = 4.7%. Most years it's \~2% or less. So **pass**. - **Current Ratio > 1.5:** NVIDIA's liquidity is strong: at Jan 2024, current assets $29.4B vs current liabilities $9.7B - current ratio \~3.0. Historically it's often around 2-3. It dipped to \~1.8 in FY2017, but generally above 1.5. Plenty of cash ($26B) on hand means short-term obligations are well covered. So **pass** on current ratio. **5. Profitability & Efficiency:** - **Gross Margin ≳ 40%:** NVIDIA's gross margin has been well above 40% every year in the past decade. It ranged from \~54% a decade ago to \~75% recently. The lowest in last 10 years was \~53% in FY2015-2016 (mid-range consumer GPUs dominating then), and even in down-cycles it stayed >50%. Thus, **NVIDIA strongly passes 40% GM**. Its current GM \~75% is world-class. - **Net Profit Margin ≳ 20%:** NVIDIA's net margin has oscillated but generally is above 20% in good years. Over 2017-2022 it averaged \~26%. It fell to \~16% in FY2023 (downcycle), and soared to \~49% TTM. For the 10-year average, net margin \~23%. So it meets the ≥20% benchmark on average, though not in **every single year** (e.g. \~9% in 2014, \~16% in 2023). But the criterion says "≳20%" which NVIDIA generally achieves. So **mostly pass**. If requiring each year, then 2014 and 2023 were below 20%, but given the long-term and current level far exceed 20%, we can consider it satisfying the spirit of high margins. - **ROE ≥ 15% (10-year avg):** Yes, NVIDIA's average ROE has been \~26% (2014-2023). Only in 2014 (9.9%) and 2015-2016 (\~13-14%) was it under 15%. In later years it was way above (2018-2022 ranging 25-44%). So on average and in most recent years, **ROE > 15%** - pass. The 10-year average \~26% greatly exceeds 15%, indicating strong profitability on equity. - **ROIC ≥ 12% (10-yr avg):** Using a simplified ROIC (Net Operating Profit After Tax / (Equity + Debt - Cash)), NVIDIA's ROIC has been very high. Even using ROA as a proxy (since low debt), ROA averaged \~15%. With slight leverage, ROIC would be a bit higher. For a concrete number: in FY2022, NOPAT \~$9.8B (assuming \~12% tax on $11.15B EBIT) on invested capital (Equity $26.6B + Debt \~$10B - Cash $21B) ≈ $15.6B, ROIC \~63%. Even in FY2023 slump, ROIC was around 20%. Therefore, **10-year avg ROIC well above 12%** - NVIDIA invests capital at high returns. This is a **pass**. (Even a stricter calculation including goodwill etc., NVIDIA's ROIC reported by analysts is often \~20-30%). - **ROA ≥ 6%:** As noted, ROA has consistently been above 6%, except maybe just around that in early years. From FY2017 onward, ROA never fell below \~10%. Even in down year FY2023, ROA \~10.6%. So **pass** - NVIDIA's efficient use of assets yields ROA far above 6%. - **Depreciation / Gross Profit ≤ 10%:** Depreciation as a % of gross profit indicates capital intensity. In FY2023, D&A was $1.54B vs gross profit $15.36B - that's 10.0%. In FY2024, D&A $1.51B vs GP $44.30B - 3.4%. Historically, it hovered \~10% or below: e.g. FY2021 $1.10B D&A, $17.48B GP = 6.3%; FY2022 6.7%; FY2020 \~9.9%. Early in decade, it was slightly above 10% when margins were lower. But nowadays it's well under. **Overall, yes**, NVIDIA's depreciation is generally under 10% of gross profit, demonstrating a relatively asset-light model (it outsources fabs, so capex depreciation is mostly on equipment like test labs and some leased facilities). We consider this a pass - with the caveat that if one year was slightly over 10% it was marginal. TTM it's only \~2%, reflecting huge gross profit growth without commensurate depreciation increase. **Summary:** NVIDIA passes many qualitative "great company" tests: high margins, high returns on capital, strong cash flows, and a rock-solid balance sheet. It falls short on **valuation-focused metrics** - its stock has rarely been cheap by P/E or P/B standards, reflecting expectations of growth. It also has had **volatile earnings** (not steadily increasing every single year, due to cyclical swings). From a Buffett-style checklist viewpoint, the business quality is exceptional (moat, profitability) but the *price* has usually been the sticking point for strict value investors. At current prices, NVIDIA definitely **fails classic deep-value valuation criteria** (P/E, P/B, margin of safety), meaning a traditional value investor would find it expensive. However, on **growth, efficiency, and financial strength criteria**, NVIDIA **excels**. For instance, it comfortably meets the hurdles for ROE, ROIC, margins, and cash generation - indicating it is a *great company fundamentally*. The key question remains whether one is paying too high a price. In conclusion, NVIDIA **checks the boxes on business quality** but **does not check the low-valuation boxes** - a common trade-off with premier growth companies. ## **DCF Analysis** To estimate NVIDIA's intrinsic value, we perform three discounted cash flow scenarios: a **No-Growth** case, a **Conservative Growth** case (using the lowest historical growth rate), and a **Nominal Growth** case (using average historical growth). All cash flows are in USD. We base these on NVIDIA's recent Free Cash Flow (FCF) and apply a 10% discount rate as a typical equity cost of capital, with a terminal value using a perpetual growth model. NVIDIA's TTM (Q3 FY25) free cash flow is about **$27.0 billion** (or \~$10.97 per share given \~2.46B shares). For simplicity, we'll use a rounded $60 billion FCF for full-year FY2025 as the starting point (which is close to TTM). - **Case 1: No Growth** - *FCF remains flat forever.* This assumes NVIDIA's current earnings power is maintained but doesn't increase. In DCF terms, FCF is held at $60B each year going forward, with zero growth in perpetuity. The intrinsic value in a no-growth perpetuity = FCF / discount rate. Using 10% discount, **Intrinsic Value ≈ $60B / 0.10 = $600 billion**. On a per-share basis (≈2.47B shares), that is about **$243 per share** intrinsic value. A 30% margin of safety would require buying at **≤$170/share**. NVIDIA's current market cap ($1.14T) and share price (\~$186) are above the no-growth fair value's 70% threshold (they imply some growth is priced in). So by the no-growth scenario, NVIDIA looks slightly overvalued relative to a strict 30% MOS - the current price $186 is \~10% above the $170 "buy at 70%" level. However, it's close; in fact, at $186 it's only \~23% below the full no-growth intrinsic $243, leaving a small margin of safety gap. This indicates that *if* NVIDIA never grows again, it's only modestly overvalued now - a scenario arguably too pessimistic given the company's trajectory. - **Case 2: Conservative Growth** - *FCF grows at the lowest rate observed in past 10 years.* NVIDIA's lowest annual growth (ignoring negative dips) was roughly **+0.2%** (flat) in FY2022. But we've already done a zero-growth case. Perhaps they intended "lowest positive growth" or average of low years. The lowest positive FCF growth in last decade might be mid-single-digits (for instance, revenue grew \~7% in 2015, FCF that year actually fell a bit though due to working capital). To be meaningfully conservative, let's assume a **5% annual FCF growth** going forward - essentially a low-end sustainable growth (roughly equal to nominal GDP growth, implying NVIDIA grows just with the economy). We model FCF starting at $60B, growing 5% per year for a decade, then no growth beyond year 10 (terminal growth = 0% after year 10). Using a 10% discount, the present value of 10-year growing cash flows plus terminal value yields an intrinsic value of approximately **$857 billion** (we calculated \~$857.5B) - or about **$347 per share**. (Alternatively, using the Gordon Growth formula with growth=5% in perpetuity: intrinsic = FCF / (r-g) = $60B/(0.10-0.05) = $1.2 trillion, but that likely overshoots as it implies perpetual growth; our finite-horizon + terminal approach gave \~$857B). Taking the more conservative \~$857B ($347/share) result: with a 30% margin of safety, the **buy price would be \~$243/share**. NVIDIA's current \~$186 is well *below* $243, meaning in this conservative 5% growth scenario, the stock is trading at about 76% of intrinsic value - which is an acceptable \~24% margin of safety. Put differently, if NVIDIA can manage even very modest growth (5% annually) long-term, the current price offers nearly a full 25% margin of safety to intrinsic value. Thus, by this scenario, **NVIDIA appears slightly undervalued** for a patient investor assuming at least some growth. It's notable that 5% is a trivial growth rate for NVIDIA given its past - this scenario essentially assumes NVIDIA's explosive growth immediately decays to low single digits going forward. - **Case 3: Nominal Growth** - *FCF grows at the average rate of the past 10 years.* NVIDIA's 10-year revenue CAGR (2014-2024) is \~30% and EPS CAGR even higher (\~40%+). But those include the recent AI spike, which likely won't repeat. Perhaps by "nominal" they mean a middle-ground realistic growth. We'll assume **10% annual FCF growth** for the next decade, then a terminal growth of 0-3%. (10% is actually lower than NVIDIA's long-term EPS growth rate ex-2024, but let's be conservative for "nominal"). Starting FCF $60B, growing at 10% for 10 years, then flat beyond yields an intrinsic value of roughly **$1.217 trillion** (we computed about $1.217T) - or about **$493 per share**. (If instead we did a perpetual 3% terminal growth, the value would be even higher, around $1.5T, but to keep it on nominal side, we assumed 0% terminal beyond year 10). Using the $1.217T value, a 30% MOS price would be \*\*$345/share\*\*. Current price $186 is just 54% of the intrinsic value in this scenario - well below the 70% threshold. That implies a significant margin of safety *if* 10% growth is achievable. So under "nominal" growth (which for NVIDIA is actually modest historically), **the stock is undervalued**: it trades at roughly half of intrinsic value by our estimate. This suggests that if NVIDIA can sustain double-digit growth for a while (a reasonable expectation given AI adoption, though debate is how long), the current price provides a comfortable safety cushion. **Basis of Growth Assumptions:** The **MAX/AVG/MIN free cash flow growth** we considered: - Max growth in past 10 years includes extraordinary jumps (FCF grew 5x from 2023 to 2024!). We didn't explicitly use that in a scenario because it's not sustainable every year. If one did use a "max" short-term (say next year 50% growth then leveling), the valuation would be even higher. - Avg growth (depending on period) we approximated as \~10%. - Min positive growth we took as 5%. > These capture a spectrum: 0%, 5%, 10%. **Results Summary:** - *No growth case:* **Intrinsic value ≈ $243/share**; 30% MOS buy ≤ $170. Current $186 is above MOS threshold (stock slightly expensive if zero growth forever, which is unlikely). - *Conservative (5% growth):* **Intrinsic ≈ $347/share**; 30% MOS buy ≤ $243. Current $186 is *below* $243, satisfying margin of safety. This suggests decent upside if NVIDIA manages even low growth. - *Nominal (10% growth):* **Intrinsic ≈ $493/share**; 30% MOS buy ≤ $345. Current $186 is far below that - a wide margin of safety. Thus, if one believes NVIDIA will grow somewhere between 5-10% annually long-term (a conservative stance given recent growth drivers), the stock appears undervalued relative to intrinsic value. It's only if one assumes essentially no growth that the stock looks slightly over fair value. Realistically, NVIDIA is expected to grow earnings considerably in the next few years (Street consensus is \~30% revenue CAGR for 2024-2026 fueled by AI demand). Our scenarios show that **the current stock price already embeds a lot of growth** (somewhere between the no-growth and low-growth valuations), but not exorbitant growth - moderate growth would more than justify the price. In classic DCF terms, much of NVIDIA's value comes from future growth opportunities (given its high P/E). So there is execution risk: if growth falters early, the higher-growth scenario won't materialize. But the DCF analysis indicates that NVIDIA's valuation can be rationalized even with modest growth - the key is the enormous current cash generation which, discounted, already covers a substantial portion of the market cap. In short, **NVIDIA's intrinsic value is highly sensitive to growth assumptions**. At one extreme (no growth), it's only slightly above today's cap; at the other (historical average growth), intrinsic value is multiples of today's cap. Given NVIDIA's technology leadership and tailwinds, a reasonable investor might lean closer to the conservative/nominal growth cases, in which case the stock appears to offer a margin of safety. Of course, an investment would also consider qualitative factors and competitive landscape to validate those growth expectations. *(Calculations: Discount rate 10%. Terminal value in scenarios uses no-growth beyond Year 10 for conservatism. FCF base \~$60B approximated from FY2025e. The resulting per-share values were cross-checked with simplified Gordon formulas: Case1 uses FCF/0% growth; Case2 uses a multi-stage 5% then 0%; Case3 uses 10% then 0%. Note: If we assumed a modest terminal growth (say 2-3%), intrinsic values would be higher by \~20-30%.)* ## **Key Metrics Charts** Below we present some key metrics trends for NVIDIA over the past 10 years, with averages and notable highs/lows marked: **Revenue Growth:** NVIDIA's revenue trajectory shows relatively steady growth until a **massive spike in the last year**(AI-driven). For much of the 2010s, revenue climbed from \~$4-5B to \~$27B by 2022. Growth stalled in 2019-2020 (flat \~0% in 2022), then exploded >+100% in 2024. The **10-year CAGR is \~30%**, but this is skewed by the recent jump. Excluding the jump, CAGR \~20%. The average annual growth was \~50% (due to big years), while the minimum was basically 0% (2019-2022). We annotate in the chart that 2019 and 2022 were flat years, whereas 2018 (+40%) and 2024 (+125%) were peaks. *Revenue (营收) - 10-year trend. The red dashed line is the 10-year average growth (\~50%, due to outliers). Max growth occurred in FY2024 at +125.9%, min growth \~0% in FY2019-FY2023. NVIDIA's revenue soared from $4.68B in 2014 to $60.92B in 2024.* **Profitability Margins:** NVIDIA's **gross margin** (GM) has consistently been high, averaging \~62% over 10 years and hitting a high \~75% recently. It dipped to \~53-55% in FY2014-2015 and FY2019 (crypto inventory issues) but otherwise stayed >60%. The **net profit margin** (NM) averaged \~25%. It had lows around 9% (FY2014) and 16% (FY2023) when either revenues or costs swung adversely, and peaked at \~49% TTM. The trend is upward over the decade as more revenue comes from high-margin data center. Operating margin followed a similar pattern, from \~15% a decade ago to \~54% TTM. **Return on Equity:** NVIDIA's ROE was modest (\~10-14%) in early years, then rose to \~25-44% during 2017-2022 thanks to higher profitability. It dipped to \~19.8% in FY2023 (earnings slump) but then skyrocketed to \~69% in FY2024 TTM due to record earnings on a not proportionally grown equity base. The **10-year average ROE \~26%** is well above the 15% "good" threshold. **Free Cash Flow & FCF Margin:** Free cash flow has grown dramatically. FCF was \~$0.8B in 2014 (17% of revenue) and \~$3-4B in late 2010s (25-30% of revenue). In FY2021-2022, FCF \~$4-8B (28-30% margin). In TTM FY2024, FCF $27B (44% margin). The chart would show FCF margin mostly 14-30% with a jump to 44% in latest year - indicating tremendous cash generation efficiency at scale. **Valuation Ratios:** NVIDIA's **P/E ratio** has been volatile: roughly 17-25× in 2014-2016, expanding to 46-72× in 2017-2021, then spiking >100× in 2023 when earnings temporarily dropped. Currently \~50×, slightly below its 10-yr average \~53×. The **P/B ratio** similarly climbed: \~2.6 in 2014 to \~10 in 2017, \~19 in 2018, dipped to \~10 in 2019, and rose above 20 by 2022-2023. Now P/B \~35 (because book hasn't grown as fast as market cap). These high valuations reflect growth expectations and have only been justified by subsequent earnings materialization. In sum, the charts highlight that **NVIDIA's fundamentals (revenue, profit, cash flow) have dramatically improved**, especially post-2016 with the AI/datacenter boost. Margins and returns have generally trended up, aside from cyclical dips. But its **stock valuation multiples have nearly always been rich**, riding higher as performance improved. Investors have paid a premium for NVIDIA's growth, which so far NVIDIA has delivered on. **Financial Disclaimer: All calculations and estimates are based on publicly available data and reasonable assumptions, but actual results may differ. This analysis is for educational purposes and not a recommendation to buy or sell NVIDIA stock.** **Disclaimer**: This content only expresses the views of the author(s) as of the date indicated and such views are subject to change without notice. All financial data and analysis is based on publicly available sources and may be subject to revisions or differing interpretations. The content is for informational purposes only and does not constitute investment advice. Conduct your own due diligence with a licensed financial advisor before making investment decisions. The author has no duty or obligation to update the information contained herein. Further, the author makes no representation, and it should not be assumed, that past investment performance is an indication of future results. Wherever there is the potential for profit, there is also the possibility of loss. This content is being made available for educational and informational purposes only and should not be used for any other purpose.