Lesson 9 of 10

Reverse DCF: what is the price assuming?

Step 3 of 6

Worked example

Nvidia at $145.98 (19 Nov 2026 close), using lesson 08's model with all its fixed inputs: 10.0% cost of capital, sales-to-capital 1.1, 15.1% tax, 4% long-run growth.

Step 1: price to firm value. Shares and net cash from the 17 Nov 2026 results release.

equity value = 24.186B shares × $145.98 = $3,530.7B
firm value   = 3,530.7 − 21.14 net cash = $3,509.5B

That matches Module 9 lesson 04's enterprise value.

Step 2: solve. Pick a year-5 margin, then find the yearly revenue growth that makes the model give $145.98. At a 60% margin the answer is 8.0% a year for five years: year-5 revenue = 361.8 × 1.080⁵ = $532B, about 1.5 times today's. The model then gives $145.87, within rounding of the price.

Step 3: the full picture. Implied yearly revenue growth, years 1–5:

Year-5 margin at 9% at 10% at 11%
50% 7.9% 12.9% 17.5%
55% 5.5% 10.3% 14.7%
60% 3.3% 8.0% 12.3%
66.1% (no fall) 1.0% 5.6% 9.7%

Step 4: the sentence. At $145.98, with a 10% cost of capital, my model says the price assumes Nvidia's revenue grows about 8.0% a year for five years while its margin slips from 66.1% to 60%; or 10.3% if the margin falls to 55%; or 12.9% at 50%. Say whose model it is.

A trap: the "free cash flow" Nvidia reports doesn't deduct stock-based pay ($6.0B in the nine months to 25 Oct 2026), so calculators that start from it start too high.

Step 5: is that believable? Set it against the evidence:

  • Guidance: $130.0B ± 3% for the quarter to January 2027 (no China data-centre compute revenue assumed). Four quarters at that pace would be $520B, 44% above the last twelve months (that quarter has an extra week), and close to the $532B the price needs by year 5.
  • Base rate: Damodaran's Semiconductor industry expects revenue growth of 11.71% a year for five years (January 2026 data).

The reading. In this model, Nvidia's price needs less growth than its own guidance and its industry's expected pace. If you believe the guidance and a margin near 60%, the price is undemanding on growth. What it needs is that the boom doesn't reverse: Story B, a pause plus price cuts, gives $83.45 (lesson 08). Watch whether revenue keeps growing after next year, whether the margin holds as customers build their own chips, and how the business runs under an interim CEO. That's a reading of what the price assumes, not a verdict.

Apple and Costco: the same question

The same method on lesson 07's Apple and on lesson 10's Costco build (next):

Model value vs price The price needs, in this model Their record
Nvidia $157.65 vs $145.98 8.0% growth a year (margin to 60%) guided pace +44%; industry 11.7%
Apple $146.18 vs $349.21 31% growth a year for five years; or a 76% margin; or a 6.9% cost of capital 5.5% a year over five years; margin 29.8–33.0%
Costco $426.47 vs $1,093.83 37% growth a year; or an 8.9% margin; or a 5.7% cost of capital 9.1% a year over five years; margin 3.65–3.85%

When a price needs far more than the record, ask which part of the model could be wrong for that company:

  • The discount rate? Turn the rate the price needs into a beta: (6.9% − 5.07%) ÷ 4.41% ≈ 0.4 for Apple, about 0.15 for Costco, far safer than the average stock (1.0). If the needed beta is that low, the rate isn't the explanation.
  • The horizon? Turn the dial to years. At 9% growth a year, with margins as forecast, Apple's price needs about 64 years of it (27 with a beta of 1.0); Costco's about 52; Nvidia's, at 8%, about six. If it's decades, a longer forecast won't close the gap alone.
  • Neither? Write the plain sentence. Apple's price assumes much more than its history: in this model it needs growth about six times its last five years' pace. Costco's needs a margin more than twice its 3.85%, or half a century of 9% growth. That is a legitimate result of the method, not a failure, and not a verdict: the market may expect something your story leaves out (more services revenue, higher membership fees). Name the input that would have to change, then look for evidence.