NVIDIA (NVDA)
Semiconductores / Cómputo de IA
The dominant AI compute platform (Data Center ~90% of revenue, +68%), with the CUDA moat and a cash machine at a 64% operating margin. At ~$212 (~38× EV/NOPAT) Undervalued: growth compounds even as the multiple compresses, but the verdict hinges on the AI capex cycle continuing — the bear case (customer concentration + a capex reversal) is real, not remote.
- Price
- $212.22
- Intrinsic value (5y, base)
- $355
- Total annual return (5y)
- 11.4%
- Status (nominal)
- Undervalued
- Margin of safety
- +27%
The essentials
- The dominant AI compute platform: Data Center ~90% of revenue (+68% FY2026, quadrupled in two years), with the CUDA moat (7.5M developers) and an annual product cadence (Blackwell → Rubin).
- Cash machine: 64% operating margin, extraordinary return on capital (fabless, capex ~5% of operating cash flow), owner-FCF ~$119bn TTM. GAAP net income is inflated by ~$25bn TTM of investment gains → valued on the operating result (EV/NOPAT), not on P/E.
- The bear case is real, not remote: two customers = 36% of revenue, the AI capex cycle can reverse, and the hyperscaler customers themselves design competing ASICs (TPU/Trainium). China was effectively closed off by export controls.
Intrinsic value — two valuation methods
Total return at 5 years: 11.4%/year = 10.8% appreciation + 0.6% dividend. The target price ($355) is ex-dividend; the $8 in dividends collected over 5 years are added separately.
By both methods, the value today (DCF $294 · Multiples $292) exceeds the market price ($212).
Pillars of the analysis
The verdict — today vs 5 years
Today — fairly valued: at $212 trades ~27.2% below its value discounted to today (~$292); the discount is positive but does not reach the margin of safety we require (≥38%).
At 5 years — Infravalorado: the target price ($355) plus dividends yield above the required average return (10%) — the business compounds.
The bridge: the return at 5 years exceeds the risk-free rate (4.5%) — but the discount does not reach the required margin of safety (≥38%). To require a 15% annual return, it would need to be bought at ~$181.
Thesis
The business
NVIDIA is an exceptionally high-quality business: the dominant AI compute platform, with the CUDA moat (7.5 million developers), a 64% operating margin, and an extraordinary return on capital (asset-light, fabless). The Data Center —~90% of revenue— quadrupled in two years, and the annual product cadence (Blackwell → Rubin) keeps the competition chasing. The counterweight is the hardest business quality to judge in the market: its explosive growth depends on the artificial intelligence capex cycle of a handful of hyperscalers, and its GAAP net income is inflated by non-operating investment gains that force one to look at the operating result.
The valuation
It is valued on a single multiple over the after-tax operating result (EV/NOPAT): GAAP net income is inflated by significant unrealized gains on equity investments (~$25bn TTM) (OpenAI, xAI, and other private companies), so the operating result is the clean base. The strategic stakes (~$74bn) are treated separately, as an option.
The base scenario projects a value of ~$355 per share over five years, an annual return of ~+11% from the current ~$212. At today's price it trades at ~38× the operating result and cheapens toward ~13× over five years as NOPAT compounds — the high multiple is the price of growth, and if growth materializes, the entry point cheapens on its own.
The margin of safety
The verdict is Undervalued: It trades at a real discount to value, though short of the required margin of safety.. At ~$212 the expected return (~+11%) exceeds the required average return (10%), but the margin today is narrow — it is an exceptionally high-quality business at a price that discounts much of the growth, not a deep-value bargain. The critical point: the verdict depends on the artificial intelligence capex cycle continuing. The bear case —a reversal in the hyperscalers' capex and competition from ASICs— is not a remote scenario but the central disconfirmation, and in it the stock trades below today's price. The asymmetry exists, but with a lower floor than in a mature compounder.
What to watch
Three fronts, in order of importance. The artificial intelligence capex cycle: NVIDIA lives off the spending of a handful of hyperscalers on AI infrastructure; if that capex decelerates or an over-investment is digested, revenue does not grow — it contracts. Customer concentration (two = 36% of revenue, three = 54% in the latest quarter) amplifies that risk. And the own silicon of those same customers (Alphabet's TPU, Amazon's Trainium, Broadcom's ASICs): if the hyperscalers migrate workloads to internal chips, NVIDIA loses share in its best market. China, already effectively lost, is the materialized risk that shows how regulation can close an entire market.
Educational / informational. Does not constitute investment advice.