NVIDIA (NVDA)

Total return at 5 years: 17.1%/year = 16.6% appreciation + 0.5% dividend. The target price ($487) is ex-dividend; the $8 in dividends collected over 5 years are added separately.

By both methods, the value today (DCF $386 · Multiples $398) exceeds the market price ($226).

Today — with margin of safety: at $226 trades ~43.1% below its value discounted to today (~$398) — the wide discount we require (≥38%, equivalent to a ~15% annual return); the risk is covered by the margin, not the rate.

At 5 years — Muy infravalorado: the target price ($487) plus dividends yield above the required average return (10%) — the business compounds.

The bridge: the return at 5 years comfortably exceeds the risk-free rate (4.5%) — and the discount reaches the required margin of safety.

Educational / informational. Does not constitute investment advice.

The inverse approach: instead of projecting growth to obtain the value, the market price ($226) is taken as given and it solves for what annual owner-earnings growth would need to hold for 5 years for the present value —at the method's rate (4.5%, no-growth terminal)— to equal that price. It is the disconfirmation test: the expectations the price already pays for, contrasted against the method's projection.

The market discounts less growth (8.9%/year) than we project (22.0%/year) → if the base case holds, there is margin: perception is more pessimistic than the estimated reality.

That growth implies ~$254.2 bn of owner earnings in year 5 (vs ~$448.3 bn of our base case). It recalculates if the DCF assumptions are edited.

Year-by-year projection of the selected scenario. From each year, two versions of the flow are derived: growth FCF (operating flow − total capex, the cash surplus) and maintenance FCF (the owner earnings: what the business yields if it only sustains its capacity). The flow is returned almost in full (dividend + buyback) or redeployed into the operation, so that EV stays roughly flat and multiples compress because the metric grows, not because of cash accumulation. The valuation is done on EV/NOPAT. In edit mode, revenue, margins, capex, and exit multiples can be adjusted.

NVIDIA is valued on a single multiple over after-tax operating income (EV/NOPAT): GAAP net income is inflated by gains on equity investments — $30.55bn over the TTM, of which $23.71bn in the first half of FY2027 (OpenAI, xAI, and other private companies) — so the operating result is the clean base. Rolling grid: the base year is the real TTM (revenue $302.97bn, operating income $197.58bn as of Jul 26, 2026, built by levels from the second-quarter release because the XBRL has not yet ingested the quarter) and the projection is rolling 12-month windows. The first year is not extrapolated from the trend: it anchors to the guidance the company published for the third quarter ($108.0bn ±2% revenue, 74.0% gross margin, and $9.2bn of operating expenses), and from there the deceleration closes a constant fraction of the gap through +12% by year 5. The first-year step is larger than the path's average step, and it is declared: guidance dictates it, and guidance governs the shape. Operating margin (65.2% on a TTM basis) compresses to ~60% by year 5 on custom-chip competition and the Blackwell→Rubin mix. The exit multiple is set at 20x, the center of the archetype's band and below what the score suggests, for cyclical risk (semiconductors, customer concentration, and a possible reversal of AI capex): avoiding the concatenated optimism of stacking high growth on high margin on high multiple. The operating cash flow margin starts depressed (44.4% versus 49.5% a quarter earlier) because working capital absorbed $40.7bn over the TTM to fund 83% growth, and recovers toward 47% as revenue decelerates. The buyback is explicit: shares fall 1.0% annually, the net reduction revealed between the second quarter of FY2026 and that of FY2027 ($55.3bn gross over the TTM, partly offset by stock-based compensation issuance). Cash is not accumulated: it is redeployed into buybacks and strategic investments. Stock-based compensation ($7.24bn, 5.7% of owner-FCF) is not material against the flow and is expensed within the operating result. What to watch: the AI capex cycle reversing, customer concentration, those same customers' custom chips, and working capital.

Today's elevated multiple is the price of growth: if the business grows, the entry point cheapens on its own going forward (the metric grows while EV stays roughly flat). The exit multiple at 3 years is higher than the terminal at 5 years —at 3 years there is more growth still ahead—, so the value curve shows whether value creation is concentrated in the early or the later years. The required return is applied to the base scenario.

They are valued separately, with their own rationale, and are not incorporated into the base or the verdict (they are excess return). When assigning them value — in Editmode —, the total with optionalities updates live, without moving the base.

The verdict, the base CAGR, and the margin of safety are always calculated on the base; optionalities do not alter them (with optionalities at $0 they do not move).

Each card fixes a required annual return and answers: if the business is worth $487 in 5 years, what is the maximum that can be paid today to obtain that return? Since it now trades at $226, the margin of safety is how much cheaper the market is than that maximum. The three thresholds: 4% covers inflation (the floor), 10% is the long-term average return, and 15% is the level of a great investment.

With a target price of $487 in 5 years plus $8 of dividends collected (the dividend adds to the return, not to the price) and a required return of 4.5% annually, the maximum to pay today is $398. Against the current market price ($226), the margin of safety is 43.1% (trades below the maximum → there is margin) and the total return at that price would be 17.1% annually.

• One business, one multiple. The Data Center dominates value and is valued on the operating result (EV/NOPAT). The equity stakes (OpenAI/xAI) and the software are treated separately, as an option.
• P/E is misleading. Looks expensive on P/E, but that net income is inflated by investment gains; on the operating result it trades at ~36× EV/NOPAT and compresses forward (to ~10× over five years) if growth compounds.
• The DCF as a second lens. The DCF at the method's risk-free rate (no-growth terminal) is hypersensitive to the assumed growth — for a business that compounds so fast, it is a secondary lens; the anchor is the exit multiple over NOPAT.
• Margin of safety at the end. At ~$226 the base return (~+17%) exceeds the required average return (10%) but the margin today is narrow — exceptional quality at a price that discounts the growth, not a bargain.
• Concatenated optimism (the method's risk). NVIDIA is the archetypal case: high growth × high margin × high multiple can stack up. The base uses disciplined assumptions (sharply decelerating growth, compressing margin, multiple within the band) — even so the return exceeds the required average return (10%) without reaching the required margin of safety — a sign of how loaded the price is.
• Optionalities kept apart. The equity stakes (~$74bn), the software (NVIDIA AI Enterprise), and robotics/auto are valued separately (default $0) → they do not contaminate the base case; they are upside.

Each margin over sales, year by year: historical (solid line) → projection (dotted).

• ✓ ROIC exceeds the cost of capital (~10%)
• ✓ High incremental ROIC (compounding)
• ! CFROIC backs up the ROIC (75%, cash vs. accruals)
• ✓ Healthy balance sheet (low corporate debt)
• ✓ Durable competitive moat (multiple advantages)

• Strong, predictable cash flow. Operating cash flow (~$126bn TTM) is enormous; capex is minuscule (~5% of the flow), so owner-FCF (~$119bn) is practically all of the operating cash flow. Predictability is the question (cyclicality), not magnitude.
• Extraordinary return on capital. Yes — return on capital well above the 10% bar (fabless, enormous margins, low invested capital): it is in the method's exceptional band.
• Creates economic value. CFROIC clears the 10% bar (the market's opportunity cost) by an enormous margin → strong value creation per dollar invested, the engine of compounding.
• The net income trap. GAAP net income is inflated by significant unrealized gains on equity investments (~$25bn TTM) (OpenAI/xAI mark-to-market): value on the operating result (EV/NOPAT), never on P/E over the reported net figure.
• Stock-based compensation. SBC (~$6.8bn TTM) is ~5.7% of owner-FCF — not material against the flow (unlike other tech companies): it is expensed in NOPAT and does not distort the metric.
• Non-recurring H20 (China) charge. The $4.5bn inventory charge from export controls (Q1 FY2026) depressed that quarter's gross margin; it is non-recurring and the margin has already normalized (~74% gross).

Annual levels from the official filing (10-K); the % over each bar is the year-over-year (YoY) growth vs the prior year.

Revenue is Data Center (GPU and system volume times price, 91.7% of the total) plus Gaming and smaller markets. Starting in the first quarter of FY2027 the company breaks out Data Center into Hyperscale ($48.71bn in the quarter, +102%), AI Clouds/Industrial/Enterprise ($40.31bn, +138%), and Edge Computing ($7.20bn, +27%). The engine is AI capex: NVIDIA sells the compute infrastructure on which AI is trained and run, and its growth tracks the investment cycle of large clouds and AI labs, hence the customer concentration.

The % are the annual (year-over-year) growth: each year —historical and projected— vs the prior one; the TTM (trailing 12m) vs the TTM of a year ago, to avoid overlapping windows. The historicals are exact figures from the official filings; the projected years come from the year-by-year model (the intermediate years without their own series are interpolated between the anchors). The projected columns (+1y…+5y) are 12-month windows counted from the TTM close (26-jul-2026): the projection starts from the most recently reported data, not the fiscal year. The projected base is realistic and unbiased — the risk discount is applied at the end, via the required return. The rationale for each metric is in the (i).

• Real, not nominal, growth. Yes — +65% FY2026 on GPU/systems volume and price, with AI demand as the structural driver, not a currency effect or an easy comparison.
• Where the engine comes from. Data Center (~90% of revenue) grew +68%, with Compute +59% and Networking +142%. All the growth is AI compute; the other segments do not move the needle.
• Driven by volume and price. Blackwell systems volume × price per generation. The growth is high quality, but tied to the capex of a handful of hyperscalers (concentration).
• Realistic deceleration. The base declines from ~+41% to ~+15% by the fifth year: the law of large numbers and ASIC competition bite; projecting today's +65% would be concatenated optimism (§2).
• Reinvestment runway. Wide in the short term (Rubin, robotics, software), but the ceiling of AI compute depends on customers' sustained capex — the open question of the case.
• Customer concentration. The structural risk to growth: two customers = 36% of revenue (FY2026), three = 54% (Q1 FY2027). Demand is real but depends on few clouds.

• Relevant in 10 years / recurrence. Yes — accelerated computing is the infrastructure of AI, and AI is a generational compute platform. The open question is not relevance but who captures the value (NVIDIA vs customers' ASICs).
• Market share / cost. Dominant (>80% in data center accelerators) and with a full-stack cost advantage. The risk is that the hyperscaler customers themselves design internal silicon to reduce their dependence.
• Pricing power. Real and strong: the 64% operating margin and the annual cadence (each generation justifies a premium) evidence the pricing power of the technology leader.
• Operating leverage. Extreme: fabless, with fixed R&D costs leveraged over revenue that quadrupled — the operating margin went from ~54% to ~64% as it scaled.
• Capital cycle / competition. AI attracts massive capital and NVIDIA is the beneficiary, but that same capital funds the competitors (the hyperscalers' ASICs, AMD) — the classic capital-cycle risk that compresses returns over time.
• Behavior in a recession / cycle. Semiconductors are cyclical and NVIDIA depends on the discretionary capex of a handful of customers: if the AI investment cycle reverses, revenue contracts — it is not a defensive business.

Not all debt is equal: only the structural needs refinancing; the rest is operational (self-liquidating).

Structural debt is what is exposed to the contraction phase of the cycle; operational debt (leases, matched funding) self-liquidates with the business.

• ✓Leverage (net debt / EBITDA)Net cash $23.2 bn
• ✓Interest coverage (EBIT / interest)425.8x
• ✓Liquidity (current ratio)4.59×
• !Cash quality (CFROIC vs ROIC)CFROIC backs 75% of ROIC
• ✓Value creation (ROIC − 10% bar)+71pp
• ✓Malinvestment test (capex vs incremental ROIC)Productive investment (2.0x)
• !Float / working capitalConsumes cash $40.7 bn (positive WC)
• ✓Dilution (SBC % of revenue + shares)SBC 2.4% of revenue

A traffic-light interpreted by the method (not generic): float (negative WC) adds up, capex is judged by incremental ROIC (malinvestment test), and a lender is not subjected to corporate solvency. The (i) shows the derivation of each number.

• Net cash. Cash and liquid investments ~$50bn against ~$8.5bn of debt → positive net cash (~$42bn), plus ~$72bn in strategic stakes — a fortress balance sheet.
• Ample coverage. Interest expense is trivial against the operating result ($162bn TTM) — coverage of hundreds of times. Minimal debt.
• Cash generation. Operating cash flow (~$126bn TTM) covers a minuscule capex (~$6.6bn) and leaves an owner-FCF of ~$119bn — extraordinary generation.
• Revenue concentration. The balance sheet risk is not solvency but dependence: two customers = 36% of revenue; a deceleration in those customers' capex hits the top line.
• Volatile equity stakes. The investments in private companies (~$42bn, many illiquid — OpenAI/xAI) multiplied 12× and are volatile mark-to-market; they add value but also noise to net income and the balance sheet.

Who owns the shares — the alignment and whether there is a controlling shareholder.

• Capital allocation. Aggressive buybacks ($40bn/year) with an extraordinary return on capital; the growing redeployment into AI stakes (OpenAI/xAI) reinforces the ecosystem but adds volatility to the balance sheet. Sound as long as the return holds.
• Alignment / skin in the game. Jensen Huang, founder, with ~3.6% ownership (tens of billions) and only performance-based PSUs — one of the strongest founder alignments in the market.
• Capital return. Bought back $40.4bn in FY2026 (98% of total return) and raised the dividend 25× — it prioritizes buybacks; the net reduction in shares is modest (~−1.2%/year) due to compensation issuance.
• Integrity / communication. Discloses the risks (concentration, China, product cadence) with transparency in the filing; forward guidance is given in the press release, not in the 10-K.

• This is not a classic missing-buyers value case: NVIDIA is the most closely followed stock in the market and trades at a high multiple on its operating result. The discrepancy is of a different nature — the market is discounting a sharper slowdown than the one the business itself is guiding to: the quarter closed up 106% year over year and the company guides to +12% sequentially for the next one.
• Fear of the AI cycle peak: the dominant perception is that large-cloud capex is nearing a ceiling, and that compresses the multiple even as revenue keeps accelerating. This is a discrepancy over the duration of the cycle, not over the current level of demand.
• GAAP net income, inflated by $30.55bn in investment gains over the TTM, muddles a P/E reading: anyone looking at the multiple on that income is dividing by a figure the operating business did not produce. And the exit from China — whose data center compute revenue guidance no longer assumes anything — adds noise to the revenue line.

The discrepancy between perception and reality, if it exists, is about the duration of the cycle and not about the quality of the business: the market fears AI capex will reverse soon, while the base case projects that Blackwell and Vera Rubin sustain growth for several more years. It is a bet on the cycle's duration, not a balance-sheet bargain, and the adverse scenario — the market being right about the peak — has a floor well below today's price.

The annual return (CAGR at 5 years) in each scenario, with the total period return below — the margin of safety made visual: upside range wide, downside range narrow.

Even in the bear scenario, the return holds at -9%/year (-37% total): the margin of safety protects the downside. The bull (+38%/year, +392% total) exceeds it comfortably — a favorable asymmetry, with a narrow downside range and a wide upside range.

• Reversal of the AI capex cycle. The central risk: NVIDIA lives off the spending of a handful of hyperscalers on AI infrastructure; if that capex decelerates or an over-investment is digested, revenue contracts and the multiple re-rates to a cyclical semiconductor.
• Customers' ASICs. The very customers that buy (Alphabet with TPU, Amazon with Trainium, plus Broadcom/Marvell) design their own silicon; if they migrate inference workloads, NVIDIA loses share and margin in its best market.
• Customer concentration. Two customers = 36% of revenue (FY2026), three = 54% (Q1 FY2027). The loss or cutback of one hits disproportionately.
• Export controls / China. NVIDIA was left «effectively excluded» from the Chinese data center market (China revenue fell from $25bn to $19.7bn); new laws (the GAIN AI Act, tariffs on the H200) can further shrink the addressable market and strengthen Huawei.
• Product transitions. The complexity of the transitions (Hopper→Blackwell→Rubin) can cause revenue volatility, quality problems, higher inventory provisions, or lower production yields.

Each thinker's analytical framework applied to our data.

• Buffett / Graham Quality + margin of safety

  A wide moat and ROIC 81% above the 10% bar, with a +43% margin → a quality business at a good price.

• Peter Lynch Growth at a reasonable price (GARP)

  A hyper-growth growing 22% at a multiple/growth of 1.5 → reasonable for its growth.

• Joel Greenblatt Cheap and high-return (Magic Formula)

  Earnings yield 4% (EBIT/EV) + ROIC 81% → falls outside the Magic Formula.

• Howard Marks Perception vs reality + cycle

  The price discounts 9% vs our 22%: perception is more pessimistic than reality, with an identified source of the discount.

• Seth Klarman Capital protection (bear scenario)

  Bear-scenario floor -10%/yr over 5y (material loss) → risk of permanent capital loss.

• Pat Dorsey Moat strength (Five Rules)

  A wide moat, stable; sources: switching costs, efficient scale, network effects, intangibles, cost advantage → passes the Five Rules.

• Aswath Damodaran Expectations implied by the price

  Justifying the price requires discounting 9%, within what we project (22%) — the story squares with the numbers.