How to Build a DCF Model with AI: The 2026 Guide
A DCF model that used to take two days in Excel now takes 20 minutes in Sourcetable. Live financials from SEC filings, AI-generated projections, automated WACC calculation, and Monte Carlo sensitivity — all in natural language.
Andrew Grosser
June 1, 2026 • 13 min read
The discounted cash flow model is the foundation of fundamental equity valuation. Every investment bank, hedge fund, and serious equity analyst builds them. They're also notoriously time-consuming: pulling financial statements, normalizing historical data, projecting free cash flows, calculating WACC, building sensitivity tables, and debugging formula errors. Sourcetable's AI changes each step.
Quick Comparison
| DCF Step | Excel Approach | Sourcetable AI |
|---|---|---|
| Historical financials | Manual copy from SEC/Bloomberg | Auto-pulled from 500+ APIs |
| Revenue projections | Manual growth assumptions | AI-suggested with analyst consensus |
| WACC calculation | Manual beta/cost of capital | Calculated from live market data |
| Terminal value | Manual Gordon growth model | Multiple methods, AI comparison |
| Sensitivity analysis | Data table, manual ranges | Monte Carlo, natural language |
Step 1: Pull Historical Financials
A DCF needs 5-10 years of historical income statement, balance sheet, and cash flow data. In Sourcetable: 'Pull 10 years of annual income statement, balance sheet, and cash flow statement for MSFT. Calculate key metrics: revenue growth rate, EBITDA margin, capex as % of revenue, working capital changes, and free cash flow conversion.' Sourcetable pulls this from Financial Modeling Prep or Intrinio, normalizes it, and calculates the metrics automatically. What used to take an analyst an hour takes 30 seconds.
Step 2: Project Free Cash Flow
The most judgment-intensive step: projecting future free cash flows. In Sourcetable: 'Build a 5-year FCF projection for MSFT. Use analyst consensus revenue estimates for years 1-2, then apply 3 scenarios (bear/base/bull) with growth rates of 8%, 12%, and 16%. Assume EBITDA margins compress 50bps annually, capex stays at 8% of revenue, and working capital follows historical patterns.' The AI builds all three scenario projections simultaneously, flags where assumptions diverge from historical norms, and shows sensitivity to each key driver.
Step 3: Calculate WACC
WACC (Weighted Average Cost of Capital) is the discount rate — and getting it right matters. In Sourcetable: 'Calculate WACC for MSFT. Use the 5-year beta from market data, current 10-year Treasury yield as risk-free rate, Damodaran equity risk premium for the US market, and current market cap and book value of debt for capital structure weights. Apply the CAPM model for cost of equity and current corporate bond yield for cost of debt.' Sourcetable pulls each input from live market data — beta from price history, risk-free rate from FRED, debt yields from bond data — and calculates WACC with one request.
Step 4: Terminal Value and Intrinsic Value
The terminal value typically represents 60-80% of DCF value — making the terminal growth rate assumption critical. In Sourcetable: 'Calculate intrinsic value per share for MSFT using both Gordon Growth Model (terminal growth 3%) and Exit Multiple Method (10x terminal EBITDA). Discount FCF projections at calculated WACC. Show implied upside/downside vs current market price. Calculate the terminal growth rate implied by current market price (implied terminal growth).' The implied terminal growth rate is particularly useful — if the market is pricing in 6% perpetual growth for a company growing 12%, you have a framework for the bull case.
Step 5: Sensitivity Analysis and Monte Carlo
No DCF is complete without sensitivity analysis. In Sourcetable: 'Run a sensitivity table showing intrinsic value per share across WACC ranges from 7-11% and terminal growth rates from 1-5%. Show the current share price on the table so I can see which combinations support the current valuation.' For Monte Carlo: 'Run 10,000 Monte Carlo scenarios on this DCF, varying revenue growth (±3% from base), EBITDA margin (±2%), WACC (±1%), and terminal growth (±1%). Show the probability distribution of intrinsic value and the probability that fair value exceeds current price.'
Complete DCF toolkit:
- ✅ 10 years historical financials auto-pulled
- ✅ Bear/base/bull scenario projections
- ✅ WACC from live market data (beta, risk-free rate, ERP)
- ✅ Gordon Growth and Exit Multiple terminal value
- ✅ Sensitivity tables (WACC vs terminal growth)
- ✅ Monte Carlo distribution of fair value
- ✅ Implied terminal growth rate calculation
Common DCF Mistakes Sourcetable Prevents
Several common errors are automatic in Sourcetable: circular references in WACC calculations (where debt affects WACC affects value affects debt) are handled iteratively. Double-counting working capital changes in FCF is flagged. Using nominal growth rates with real discount rates (or vice versa) triggers a warning. Revenue growth projections that exceed GDP growth over long horizons are flagged as aggressive. These are the errors that get caught in bank model reviews — Sourcetable catches them before you present.
Try Build DCF Model Spreadsheet in Sourcetable — free
Natural language AI. 500+ financial APIs. No coding required.
Start Free Trial →