DCF Modeling
DCF Modeling
LEARNING OBJECTIVES
IMPORTANCE
The DCF is where the whole curriculum assembles: free cash flow, the cost of capital, the enterprise-to-equity bridge, and the exit multiple. It is the one method that values a business from the inside, on what it will produce, not on what the crowd will pay. Its power is that independence; its weakness is its assumptions.
The central insight: A DCF is only as good as its assumptions. Its discipline is not false precision; it is an honest range.
INTELLECTUAL ORIGINS
John Burr Williams stated it in 1938: the value of any asset is the present value of the cash it will pay its owner over time. Everything since is machinery on that one sentence, how to forecast the cash, choose the rate, and value the years beyond the forecast.
THE DCF IN EIGHT STEPS
Step 1. Forecast . Project FCFF over an explicit period, usually five to ten years.
Step 2. Build the discount rate (). Unlevered cash flow belongs to all funders, so it is discounted at the blended cost of capital.
Step 3. Discount each year to today. Convert each future cash flow to present value.
Step 4. Calculate the . Capture the cash flows beyond the forecast, usually the majority of the value, by one of two methods, then discount it back.
Step 5. Sum to enterprise value. Add the two present-value pieces.
Step 6. Bridge to equity value. Strip out the claims that rank ahead of shareholders.
Step 7. Solve for the implied share price. Translate equity value into a per-share figure.
Step 8. and sanity checks. The output is fragile, so stress it.
COMPARISON
| Terminal value method | What it assumes | Best when |
|---|---|---|
| Perpetuity growth (Gordon) | Cash flows grow forever at a low, steady rate | Stable, predictable businesses |
| Exit multiple | The business is sold at a peer multiple of EBITDA | A clear, comparable peer set exists |
| Average of both | Neither single method is fully trusted | The default discipline, with a cross-check |
REAL COMPANY APPLICATION: THE HONEST COMPANY
The same eight steps, applied to a real company. The full model is in Equity Research, under the Honest analysis.
Honest is the cleanest possible case, because it carries cash and no debt.
1. FCFF. The company runs at an operating loss today, so the value rests on a forecast margin inflection, from breakeven toward about 6.5% by 2030 on low-single-digit organic growth.
2. WACC. The cost of equity from CAPM is 4.42% + 1.45 × 4.24% ≈ 10.6%, and with only a small lease the blended WACC the model uses is about 10.34%. A company with debt would blend in a lower after-tax cost of debt and land beneath this.
3. Discount each year's cash flow at 10.34%.
4. Terminal value by a ~2.5% perpetuity growth and a peer exit multiple, averaged; it carries most of the value.
5. Enterprise value is the sum of the two present-value pieces.
6. Bridge: with no debt, simply add the $89.6M of net cash.
7. Implied price: divide by about 113M shares. The base lands near $3.81 against a market price of $3.60 as of 28 June 2026 (Q2 FY26), and the net cash alone is about $0.80 a share, a floor under a thin business.
8. Sensitivity: across the scenarios the value runs from roughly $2.05 in the bear case to $5.56 in the bull. The DCF produces a span, not a point.
BUILD IT: A DCF ON AMAZON, STEP BY STEP
The eight steps above are the theory. This section walks all eight of them on a single real company, Amazon, using its actual fiscal-2025 figures (the year ended 31 December 2025). Follow it once here and you will be able to open any company’s numbers and build the same model yourself. The figures below are Amazon’s reported results and the assumptions used in the Compoundex analysis; the forecast years are illustrative, chosen to show the method, not to reproduce the model to the last decimal. This is educational, not investment advice.
| Amazon starting point (FY2025) | Figure | What it is |
|---|---|---|
| Revenue | $716.9B | Total sales across the retail, advertising and AWS businesses. |
| Operating income (EBIT) | $80.0B | Profit from operations, an 11.2% operating margin. |
| Depreciation & amortisation | $66B | A large non-cash cost, added back later. |
| Capital expenditure | $131.8B | Heavy spending on data centres and AI capacity. |
| Free cash flow (FY2025) | $7.7B | Operating cash of about $139.5B minus that capex. |
| Net debt | $30B | Debt minus cash, needed for the equity bridge. |
| Diluted shares | 10.85B | The share count the final value is divided by. |
STEP 1 · FORECAST UNLEVERED FREE CASH FLOW
The formula. Unlevered free cash flow is the cash the whole business throws off before any financing, and it is what a DCF discounts. You build it each year from operating profit: FCFF = EBIT × (1 − tax rate) + D&A − capital expenditure − the change in working capital. Everything downstream depends on getting this line right.
Amazon’s catch. Run the formula on Amazon’s FY2025 numbers and something striking appears. Operating profit after an assumed 20% tax is about $64B, add back $66B of depreciation, then subtract $131.8B of capital expenditure, and the current-year free cash flow is barely above zero. Amazon reported only $7.7B of free cash flow on $716.9B of sales. This is the whole modelling problem in one line: today’s cash flow is being suppressed by a once-in-a-decade building spree.
Why you do not just extrapolate. A lazy DCF would grow that $7.7B forward and conclude Amazon is barely worth anything. That would be wrong. The capital expenditure is an investment in future capacity, not a permanent drain. The disciplined forecast therefore models the drivers separately: revenue growth easing from low double digits toward mid single digits, operating margin widening as the higher-margin AWS and advertising mix grows, and, critically, capital expenditure falling as a share of revenue as the build-out matures. As capex normalises from roughly 18% of revenue back toward the low teens, free cash flow inflects sharply upward. The value is in that inflection, which is exactly why the model has to forecast the pieces rather than the headline number.
STEP 2 · BUILD THE DISCOUNT RATE (WACC)
Unlevered cash flow belongs to all funders, lenders and shareholders alike, so it is discounted at the blended cost of capital. Amazon’s WACC is 9.05%. It is built from a capital structure that is 94.28% equity and about 6% debt: a cost of equity near 10% (the risk-free rate plus Amazon’s beta times the equity risk premium) blended with an after-tax cost of debt of 3.9%. Because Amazon is funded almost entirely by equity, its WACC sits close to its cost of equity. This single rate does two jobs: it discounts every future cash flow, and it is the hurdle Amazon’s returns on capital must clear to create value.
STEP 3 · DISCOUNT EACH YEAR TO TODAY
A dollar in the future is worth less than a dollar now, so each forecast year is pulled back to present value: present value = that year’s FCFF ÷ (1 + 0.0905) raised to the year number. At a 9.05% rate, one dollar five years out is worth about 64 cents today, and ten years out about 42 cents. The further away the cash, the more the discount bites, which is why the near-term forecast and the terminal value both matter.
STEP 4 · CALCULATE THE TERMINAL VALUE
A DCF forecasts only a handful of years in detail, but Amazon will keep producing cash long after. The terminal value captures everything beyond the explicit forecast, and in most models it is the majority of the answer. The common method is perpetuity growth: terminal value = final-year FCFF × (1 + g) ÷ (WACC − g), where g is a modest long-run growth rate no higher than the economy itself, here around 3%. That figure is then discounted back to today like any other future sum. Because the terminal value is so large a share of the total, small changes in g or in WACC move the final answer a great deal, which is what Step 8 stress-tests.
STEP 5 · SUM TO ENTERPRISE VALUE
Add the two present-value pieces together: the present value of the forecast years of free cash flow, plus the present value of the terminal value. For Amazon this sum, the enterprise value, comes to roughly $3.13 trillion. Enterprise value is the worth of the whole operating business, before deciding who has a claim on it.
STEP 6 · BRIDGE TO EQUITY VALUE
Enterprise value belongs to everyone who funded the business, so to get to what the shares are worth you strip out the claims that rank ahead of shareholders. For Amazon that mainly means subtracting net debt of about $30B. Enterprise value of roughly $3.13 trillion minus that net debt leaves an equity value of about $3.096 trillion, the portion that belongs to shareholders.
STEP 7 · SOLVE FOR THE IMPLIED SHARE PRICE
Finally, translate the equity value into a per-share figure by dividing by the diluted share count. Equity value of about $3.096 trillion divided by 10.85 billion shares gives an intrinsic value of $285.31 per share. On 28 June 2026 the market price was $232.69. The model therefore puts Amazon roughly 23% below its estimate of intrinsic value, a positive margin of safety, which for a business of this quality is unusual.
| From enterprise value to a price (Amazon) | Figure |
|---|---|
| Enterprise value | ~$3.13T |
| − Net debt | $30B |
| = Equity value | ~$3.096T |
| ÷ Diluted shares | 10.85B |
| = Intrinsic value per share | $285.31 |
| Market price (28 Jun 2026) | $232.69 |
| Margin of safety | ~ +23% |
STEP 8 · SENSITIVITY AND SANITY CHECKS
The output of any DCF is fragile, so the last step is to stress it rather than trust it. Vary the assumptions. Move WACC and g across a grid to produce a range of values instead of a single false-precise number; for Amazon, a slightly higher discount rate or lower terminal growth quickly narrows the margin of safety. Check the composition. See what share of the value sits in the terminal value, and what growth and margins the price implies. Sanity-check against returns. Amazon earned a return on invested capital of about 14% in FY2025 against a 9% WACC, a genuine value-creating spread, which supports a valuation above the cost of capital rather than below it.
Now you can build one yourself. Take any company and repeat these eight steps: forecast its unlevered free cash flow from revenue, margins and reinvestment; discount each year at its WACC; add a terminal value; sum to enterprise value; subtract net debt to reach equity value; divide by shares for an intrinsic price; then stress the assumptions. If the answer holds up across a sensible range and ties to the returns the business actually earns, you have a DCF you can defend.
LIMITATIONS
COMPETING VIEWS
IN SUMMARY
A DCF values a business as the present value of its future cash. Forecast unlevered free cash flow, discount it at the cost of capital, add a terminal value, sum to enterprise value, bridge to equity, and solve for the implied price, then stress it. Honest shows the machine at its cleanest: with effectively no debt the rate is close to the 10.34% cost of capital, and the bridge simply adds the net cash. The output is a range, and the discipline is to demand a margin of safety against the base.
RELATED TOPICS
Pillar 4, Cash Flow and Free Cash Flow. The cash flow a DCF discounts.
Pillar 6, Enterprise Value vs Equity Value. The bridge at Step 6.
Pillar 7, Valuation Multiples. Where the exit multiple comes from.
Pillar 5, Adjustments and Normalization. Why the starting cash flow is normalized first.
FURTHER READING
REFLECTION QUESTIONS
The cash a business generates for everyone who funded it, before financing, projected over an explicit forecast and discounted to today. It is the cash flow a DCF runs on.
In a DCF of Honest, each forecast year's FCFF is built from a revenue and margin path, then discounted; because the company runs at a loss today, the value rests on a forecast margin inflection toward about 6.5% by 2030.
FCFF is unlevered, so discounting it gives enterprise value, the figure a DCF produces before bridging to equity.
The weighted average cost of capital: the blended return required by all a company's funders, equity and debt, used to discount unlevered cash flow. A company with no debt has nothing to blend, so its WACC equals its cost of equity.
Honest carries only a small lease, so its WACC of about 10.34% sits just below its cost of equity. A more indebted company would blend in more low-cost debt and fall further beneath it.
The discount rate sets how heavily future cash is penalised for time and risk, and small changes in it move the valuation substantially.
The return shareholders require to hold a stock, estimated with the capital asset pricing model: the risk-free rate plus the stock's beta times the equity risk premium. It is the discount rate for equity cash flows and a component of WACC.
For Honest: a risk-free rate of about 4.42%, plus a beta near 1.45 times an equity risk premium of about 4.24%, gives a cost of equity of roughly 10.6%.
The cost of equity is the hurdle a business must clear for shareholders, and for a debt-free company it is the whole discount rate.
The value of every cash flow beyond the explicit forecast, usually the majority of a DCF's total. It is estimated either by assuming steady perpetual growth or by applying a peer exit multiple, then discounted back like any other cash flow.
A DCF of Honest sets terminal value with a perpetuity growth around 2.5% and a peer exit multiple near 13x, averaged, and it carries most of the company's value.
Because the terminal value dominates, its assumptions deserve the most scrutiny, and they must reflect a normalized steady state, not a peak year.
The constant rate at which cash flows are assumed to grow forever in the perpetuity method. It must be low and sustainable, no higher than long-run economic growth, because no company can outgrow the economy indefinitely.
A DCF of Honest uses a perpetuity growth of about 2.5%, at the GDP ceiling, with no premium because a small, low-moat brand earns no structural growth advantage.
A growth rate set too high inflates the terminal value without limit, which is why g is capped near GDP and cross-checked against the exit multiple it implies.
A grid that recomputes the valuation across a range of discount rates and growth rates, turning a single estimate into a span. It is how a DCF is honestly presented, because the output is too sensitive to inputs to be a single number.
Across its scenarios Honest's value runs from roughly $2.05 in the bear case to $5.56 in the bull, with a base near $3.81. The DCF produces a range, not a point.
The sensitivity table shows exactly which assumptions the value depends on and forces an honest range, which is the discipline of a DCF.