What a $5.45 trillion price tag actually assumes about NVIDIA
The most common argument about NVIDIA is whether it is expensive. That is the wrong question. Here is the arithmetic of what the market was paying for on August 15, 2026 — with every assumption disclosed.
- What happened
- NVIDIA carries a market capitalization of about $5.453 trillion at $225.16 per share. In its most recent quarter it earned $2.39 per share on a GAAP basis and $1.87 on a non-GAAP basis, on revenue of $81.6bn.
- Why
- Annualizing the most recent quarter puts the shares at roughly 24× GAAP earnings or 30× non-GAAP earnings — multiples that are not, by themselves, extreme. The valuation question is not about growth. It is about durability.
- The numbers that matter
- $5.453TMarket cap30.1×Annualized Q1 non-GAAP EPS$81.6BQ1 FY27 revenue92%Share from Data Center$91.0BQ2 guidance
- Why it matters
- NVIDIA is the largest position in most US index funds. If you own a total-market ETF, this valuation is your valuation whether or not you ever chose it.
- What happens nextas we saw it on 17 Aug 2026
- Fiscal Q2 results on August 26. Company guidance is $91.0bn, ±2%.
The answer to "is it expensive" changes by a factor of two depending on which earnings number you divide by. That is the actual debate.
| Trailing FY2026 GAAP EPS $4.90 | 45.9× |
|---|---|
| Annualized Q1 FY27 non-GAAP $7.48 | 30.1× |
| Annualized Q1 FY27 GAAP $9.56 | 23.6× |
NVIDIA is the most valuable company in the world, at roughly $5.453 trillion. There is a very large amount of commentary about whether that is too much. Almost none of it does the arithmetic in public.
This series exists to do the arithmetic in public. Below is every input, every division, and every assumption. Where a figure is the company's own disclosure it is labeled as data. Where it is ours it is labeled as a calculation. You should be able to disagree with our conclusion without disagreeing with our numbers.
The disclosed facts
Fiscal 2026 (year ended January 25, 2026): revenue $215.9bn, net income $120.1bn, GAAP diluted EPS $4.90, GAAP gross margin 71.1%.
Fiscal Q1 2027 (quarter ended April 26, 2026): revenue $81.615bn, up 85% year over year and 20% sequentially. Data Center revenue $75.2bn, up 92% year over year. GAAP gross margin 74.9%, non-GAAP 75.0%. GAAP diluted EPS $2.39; non-GAAP diluted EPS $1.87. Operating cash flow $50.3bn; free cash flow $48.6bn. Quarterly dividend raised from $0.01 to $0.25 and an $80bn buyback authorized.
Guidance for fiscal Q2 2027: revenue $91.0bn, ±2%; non-GAAP gross margin 75.0%, ±50 basis points.
Step 1: what multiple are you actually paying?
A price-to-earnings ratio is a division problem with a contested denominator. Here are three defensible denominators and what each produces.
| Earnings measure | EPS used | P/E at $225.16 | What it assumes |
|---|---|---|---|
| Trailing FY2026 GAAP | $4.90 | 45.9× | That last year's earnings describe the company today |
| Annualized Q1 FY27 non-GAAP | $7.48 | 30.1× | That the current quarter, times four, is the run-rate |
| Annualized Q1 FY27 GAAP | $9.56 | 23.6× | The same, including non-operating gains |
Annualized non-GAAP EPS: $1.87 × 4 = $7.48. $225.16 ÷ $7.48 = 30.1×.
Annualized GAAP EPS: $2.39 × 4 = $9.56. $225.16 ÷ $9.56 = 23.6×.
Trailing: $225.16 ÷ $4.90 = 45.9×.
Assumption: annualizing one quarter assumes the following three quarters look exactly like it. NVIDIA has guided to $91.0bn for the current quarter, which is higher, so annualizing Q1 is conservative on revenue. It is not conservative if margins fall.
Notice what just happened. The company that is universally described as expensive trades at 23.6 its most recent quarter's GAAP earnings annualized. The S&P 500's forward 12-month P/E was 20.0 as of early August, against a five-year average of 19.9 and a ten-year average of 19.0.
On annualized current earnings, NVIDIA trades at a premium to the index — but a premium in the range of 20% to 50%, not the multiple-hundred-percent premium the phrase "AI bubble" implies. Whether that premium is justified depends on durability, which is not a valuation question at all. It is a business question.
Step 2: the number that should give you pause
In the most recent quarter, NVIDIA's GAAP earnings per share ($2.39) were higher than its non-GAAP earnings per share ($1.87).
That is backwards from the normal pattern. Non-GAAP earnings usually exceed GAAP earnings, because non-GAAP figures typically exclude costs — most commonly share-based compensation. For GAAP to come in above non-GAAP, something is being added under GAAP that the company's adjusted measure removes.
The most common cause is non-operating gains — for example, mark-to-market increases on equity stakes the company holds in other businesses. Those gains are real money, but they are not the chip business, and they do not recur on a schedule. FactSet flagged the same dynamic across the wider index this quarter: S&P 500 blended earnings growth of 50.4% for Q2 falls to 32.0% once Alphabet and Amazon's investment gains are excluded.
We have not reviewed NVIDIA's 10-Q line by line, so we are labeling this as an interpretation of a pattern, not as an established fact about NVIDIA specifically. The filing is the place to settle it — and it is exactly the kind of thing worth settling before August 26.
The practical implication is straightforward: the 23.6× figure is the flattering one, and it is flattered by earnings that may not repeat. The 30.1× figure, built on the company's own adjusted measure, is the more conservative basis for thinking about the operating business.
Step 3: what would have to be true
Run the question backwards. Instead of asking whether $5.453tn is too much, ask what level of sustainable annual profit would justify it at various multiples.
| If the market eventually pays… | …NVIDIA needs annual net income of |
|---|---|
| 40× earnings | $136bn |
| 30× earnings | $182bn |
| 25× earnings | $218bn |
| 20× earnings | $273bn |
NVIDIA's fiscal 2026 net income was $120.1bn. Its most recent quarter's GAAP EPS of $2.39, annualized, is $9.56 per share — which on the roughly 24.5bn diluted shares implied by FY2026 ($120.1bn ÷ $4.90) works out to approximately $234bn of annualized net income.
Assumptions: share count derived from full-year figures rather than taken from the balance sheet; no growth applied; GAAP earnings taken at face value including any non-operating gains. Treat this as an order-of-magnitude check, not a model.
That comparison produces the genuinely interesting result. At a 25× terminal multiple, the price requires about $218bn of sustainable annual net income — and the company's current annualized run-rate is already in that neighborhood. Take the more conservative non-GAAP path and the requirement is meaningfully harder, but still not a fantasy.
The market is not paying NVIDIA's price for a decade of compounding from a small base. It is paying for the belief that a run-rate the company reached in two years will still be there in five. That is a completely different bet from the one most commentary describes, and it fails in a completely different way: not through disappointing growth, but through a single year in which the largest customers slow their spending.
The evidence, both directions
- Revenue up 85% year over year and 20% sequentially — the growth rate is not decaying.
- Gross margin of roughly 75% on $81.6bn of quarterly revenue. Very few businesses at any scale hold that.
- $48.6bn of free cash flow in one quarter funds an $80bn buyback without borrowing.
- Data Center revenue up 92% year over year — the growth is in the segment that matters.
- Company guidance of $91.0bn for the July quarter implies continued sequential growth.
- 92% of revenue comes from one segment, sold to a small number of very large customers.
- Those customers are among the best-capitalized firms on earth — and they design their own competing silicon.
- GAAP EPS above non-GAAP EPS suggests non-operating contributions to reported profit.
- At 30× annualized non-GAAP earnings, the price already capitalizes today's run-rate. There is little margin for a pause.
- A $5.45tn market cap means a large share of the register is index money that did not choose the position.
What would change this analysis
- The August 26 report. Not whether revenue beats $91.0bn — the company set that bar — but whether gross margin holds near 75% and whether the GAAP/non-GAAP gap persists.
- Hyperscaler capital expenditure guidance. NVIDIA's revenue is its customers' capex line. Their budgets are the leading indicator.
- Any change in customer concentration disclosure. The 10-Q names the share of revenue from significant customers. That number is more informative than most of the commentary about it.
- Competitive silicon reaching volume. Not announcements — shipments.
Sources
- NVIDIA Announces Financial Results for First Quarter Fiscal 2027NVIDIA · May 20, 2026 · Primary source
- NVIDIA Announces Financial Results for Fourth Quarter and Fiscal 2026NVIDIA · Feb 25, 2026 · Primary source
- NVIDIA market capitalizationCompaniesMarketCap · Aug 15, 2026 · Secondary source
- S&P 500 Earnings Season Update: August 7, 2026FactSet Earnings Insight · Aug 7, 2026 · Secondary source
- NVIDIA to hold second quarter fiscal 2027 results conference call on August 26NVIDIA / Investing.com · Aug 2026 · Secondary source
Related analysis
Before you buy NVIDIA: what to check in the earnings report
Written nine days before NVIDIA's second-quarter results, this is what an investor should understand about the business — and what an earnings report can and cannot tell you, whichever quarter it is.
Everyone agrees earnings are booming. Here is the strongest case that they aren't.
S&P 500 companies grew earnings more than 50% in the second quarter of 2026 and 86% beat expectations. Both facts are true. Neither means what the headline suggests.
The economy is sending two signals at once. The Fed is listening to the louder one.
Payrolls fell. Retail sales fell. Inflation held at 3.4%. And three Federal Reserve officials voted to raise interest rates. Here is how those facts fit together.