Is NVDA Stock a Buy in 2026? A Working Investor's Take

Is NVDA stock a buy in 2026? At $211 a share — implying a market cap above $5 trillion — is NVDA stock a buy is the most-Googled question in retail investing right now. The business is extraordinary. The price is extraordinary. The math will tell you exactly what kind of trade this actually is, if you let it.

This piece runs the same five-question framework value investors apply to every position — applied directly to Nvidia, with every number from the FY2026 10-K filed in February 2026 and current market data. By the end you will have a defensible answer to is NVDA stock a buy at today's $211. You will also know exactly what assumption about future growth you would need to be right about to make the trade work.

## The Nvidia snapshot — the numbers that decide whether NVDA stock is a buy

Nvidia's fiscal year 2026 ended January 25, 2026, and the 10-K was filed with the SEC on February 25, 2026.¹ The headline numbers:

Line item (NVDA FY2026)	Value
Revenue	$215.94B
Operating income	$130.39B
Net income	$120.07B
Operating margin	60.4%
Cash from operating activities	$102.72B
Capital expenditures	$6.04B
Free cash flow	~$96.68B
R&D expense	$18.50B
Stockholders' equity	$157.29B
Diluted shares outstanding	24.391B
Share price at time of writing	$211.14
Implied market cap	~$5.15 trillion

These are not normal financials. A 60 percent operating margin on $216 billion of revenue, $97 billion of free cash flow, capex of only $6 billion (Nvidia is fabless — TSMC makes the chips), and a return on equity around 76 percent. The business is one of the most profitable enterprises in the history of capitalism.

The question is whether the price is also reasonable. That requires the framework.

Question 1 — Is Nvidia a business you understand? (Step 1 of "is NVDA stock a buy")

Nvidia designs the chips (and software stack) that AI models train and run on. It outsources manufacturing to TSMC and earns its margin from a near-monopoly position in high-end AI accelerators plus the CUDA software ecosystem that locks customers into Nvidia hardware. Two sentences. Pass — if you also understand that the AI infrastructure cycle is the central debate.

Question 2 — Are Nvidia's numbers healthy on their own?

Five signatures with concrete thresholds.

Revenue 5-year CAGR. FY2022 revenue was $26.9B; FY2026 is $215.94B. That is a five-year CAGR above 50 percent. Off-the-charts. Pass.

Operating margin trajectory. Operating margin is 60.4 percent, up from roughly 30 percent five years ago. Expanding margins on expanding revenue — the textbook compounding signature. Pass.

Return on invested capital. Using NOPAT ≈ $130.39B × 0.85 ≈ $111B and invested capital ≈ equity + modest LT debt ≈ $170B, ROIC is roughly 65 percent. Five to seven times any defensible cost of capital. Pass.

Free cash flow yield. $96.68B ÷ $5,150B market cap = 1.88 percent. Versus a 10-year Treasury at ~4.5 percent. Fail. You are accepting a substantially lower cash return today than a risk-free bond offers — you are paying for assumed future growth.

Net debt / EBITDA. Net debt is roughly zero (cash exceeds long-term debt). The leverage check trivially passes.

Score: 4 of 5 passes, 1 clear fail on FCF yield. The business is excellent. The price is what makes Question 2 a mixed answer.

Question 3 — Does Nvidia have a moat?

Three moats stacked, which is unusual:

1. Switching costs. CUDA — Nvidia's software stack — has been the dominant programming framework for AI workloads for fifteen years. Migrating production AI training and inference off CUDA is a multi-year, multi-billion-dollar project.
2. Network effect. Every framework, library, and tooling vendor (PyTorch, TensorFlow, Hugging Face) optimises for CUDA first. The dev ecosystem reinforces the hardware lock-in.
3. Cost-on-performance advantage. Nvidia spends $18.5B/year on R&D — more than most competitors' total revenue — and is two generations ahead on high-end accelerators.

The 60 percent operating margin and 65 percent ROIC are the financial signatures of this stack. The moat is intact, deep, and currently widening. Pass with conviction.

Question 4 — Is the price below intrinsic value?

Question 4 is where is NVDA stock a buy becomes uncomfortable. Free cash flow per share = $96.68B ÷ 24.391B = $3.96. Apply a Gordon growth model at r = 10 percent, g = 4 percent (a generous long-run growth assumption for any mature business):

IV per share = 3.96 × 1.04 / (0.10 − 0.04) = $68.64

At today's $211, Nvidia trades at roughly three times that conservative IV. The market is not pricing Nvidia as a mature business. To get from $69 to $211 on a Gordon model, you need to plug in one of two things:

Option A — Much higher long-run growth. With r = 10% / g = 8%, IV per share = $214. Holding 8 percent annual growth in perpetuity (past 2050) is mathematically problematic — economies do not sustainably grow that fast. It is plausible for the next five years, not for "long-run".

Option B — Much lower discount rate. With r = 6% / g = 4%, IV per share = $206. A 6 percent discount rate is below today's 10-year Treasury. Hard to defend on a single-name equity position.

Sensitivity table for Nvidia's IV per share (varying r and g; FCF per share held at current $3.96):

	g = 4%	g = 6%	g = 8%
r = 8%	$103	$210	undefined*
r = 10%	$69	$105	$214
r = 12%	$52	$70	$107

*Gordon model breaks when g ≥ r. The fact that justifying Nvidia's current price requires growth assumptions that approach this mathematical breakdown is the warning sign.

The honest read: Nvidia at $211 is being priced as if the AI infrastructure buildout continues at near-current intensity for the next decade or more. The growth is real today; whether it is permanent is the entire question, and it is the question that decides whether is NVDA stock a buy lands as yes or no.

Question 5 — What would change your mind?

The bull thesis can be reduced to two specific tests you can verify in future filings.

1. AI infrastructure capex by hyperscalers (Microsoft, Meta, Alphabet, Amazon) stays at or above current levels through 2027. The combined hyperscaler capex of ~$400B/year is the demand engine. A 30 percent year-over-year decline in 2027 would imply the cycle is rolling over.
2. Nvidia's market share against AMD, custom silicon (Google TPUs, Amazon Trainium), and any Chinese alternatives stays above 80 percent in the data-center AI accelerator market. Below that, the moat is narrowing visibly.

If either fails, the price gets very hard to defend. Writing these down is what separates a value-investing position from a momentum trade.

The bull case: why is NVDA stock a buy at $211

The bull case treats AI as a multi-decade platform shift, not a single capex cycle. Hyperscaler training spend is the front end; inference spend at scale comes next (potentially 10× the training market). Nvidia's CUDA moat plus a relentless product cadence keeps the price-performance gap against competitors wide. Margins do not need to expand — they need only hold. Revenue compounds at 25-40 percent for another 3-5 years before normalising. Owner-earnings yield catches up to and exceeds today's Treasury yield naturally, not because the price falls but because earnings grow into the multiple.

In this world, $211 is not expensive. It is reasonably priced for a business that will produce $300-500B in annual free cash flow within five years.

The bear case: why is NVDA stock a buy a hard sell at $211

The bear case is that AI capex is a buildout cycle, not a permanent growth rate. Hyperscalers are pulling forward a decade of infrastructure spend into 2024-2027. Custom silicon (Google TPUs, Amazon Trainium, Microsoft Maia) plus rising AMD competence will erode Nvidia's pricing power. CUDA lock-in matters less for inference than for training. Margins compress as competition intensifies. The trough year is two to three years out, and revenue de-grows when the cycle turns. The market is paying for a permanent growth rate the business will not sustain.

In this world, $211 is the top of a generational bubble — the math is undeniable on the way down. Even a halving of the multiple from 42× earnings to a more sustainable 20× implies a stock price around $100.

So is NVDA stock a buy?

The numbers force you to a specific position. Nvidia is a fantastic business at an expensive price. The 5-question framework lands at three passes (business understood, signatures mostly healthy, moat with conviction) and two fails or near-fails (FCF yield well below Treasury, IV requires growth assumptions at the edge of mathematical defensibility).

A defensible read on whether NVDA stock is a buy today:

• If you have high conviction that AI infrastructure demand is a multi-decade platform shift — Nvidia at $211 is a hold-with-trim. The business compounds at the price; the multiple is the risk.
• If you have moderate conviction — wait for a buildout-cycle wobble (capex pause from one major hyperscaler) to enter. A 25 percent drawdown would put price closer to the bull-case fair value.
• If you have low conviction or no view on AI infrastructure cycle duration — pass. You cannot justify $211 from the numbers alone, and you should not own a position whose thesis you cannot state in one sentence.

The five-question framework's job is to surface the assumption you would need to be right about. The assumption that decides whether NVDA stock is a buy at $211 is the duration of the AI infrastructure cycle. Decide whether you have an edge on that specific question. If yes, position-size for the named risk in Question 5. If no, move on.

## Frameworks behind this read

The five questions above are the same framework we apply on every is X a buy analysis. Each step has a more rigorous version in the rest of the Hub. The margin of safety formula covers what to do with the IV-versus-price gap. The owner earnings formula refines the cash-flow input. The economic moat guide explains why CUDA is doing what it is doing. The full 7-step research workflow is what we apply to any new ticker. For a worked example on a fellow recovery-story stock, see our is Nike stock a buy analysis.

For more long-form essays on value-investing methodology, see the rest of the Hub.