Discounted cash flow shows up in every finance textbook as a page of Greek letters and subscripts, and that's probably why most retail investors quietly skip it and buy the stock anyway. That's a shame, because the idea underneath the notation is simple enough to explain over coffee — and understanding it is the fastest way to know whether a "fair value" number on any screener, including MoatScan AI's own, is worth trusting or worth ignoring.

The one idea everything else is built on

A dollar you receive today is worth more than a dollar you receive in ten years, because you could invest today's dollar and have more than a dollar by the time the future one arrives. A discounted cash flow model applies that single idea to an entire business: forecast every dollar of free cash flow the company will generate for years into the future, convert each of those future dollars back into today's dollars, and add them up. The total is the business's estimated worth today — its fair value.

Everything else in a DCF — the spreadsheet rows, the multi-tab models analysts build, the jargon — exists to answer three questions in numeric form. Understand what each question is really asking, in plain words, and the arithmetic stops being the hard part. It's also mostly optional: any decent DCF tool builds it for you once you've supplied honest assumptions.

Question 1: how fast will the cash grow?

This is the forecast — revenue growth fading into margins, margins fading into free cash flow, year after year. It's also where most of the genuine uncertainty in any valuation lives. Nobody knows a company's growth rate five years out with precision, and small changes here move the answer by a lot. A model assuming 20% growth for the next five years produces a meaningfully different fair value than one assuming 12%, even holding every other input constant.

Question 2: how much should the future be discounted?

This is the discount rate, most often the weighted average cost of capital, or WACC — the blended return a company's lenders and shareholders both require to justify tying up money in the business, typically somewhere in the high single digits to low teens for large, established companies. A higher WACC shrinks every future dollar more aggressively on its way back to today's terms, which is why capital-intensive or heavily indebted businesses effectively get punished twice: growth is harder to fund, and each dollar of future cash is worth less once discounted.

Question 3: what happens after the forecast ends?

No analyst forecasts cash flow forever. Every DCF picks an explicit forecast window — commonly somewhere between five and ten years — and then collapses everything beyond it into a single number: the terminal value, usually computed with the Gordon Growth formula, which assumes cash flow keeps growing at some modest, sustainable rate forever after that point. It sounds like a footnote. It usually isn't — in many DCFs, more than half of the total estimated value comes from the terminal value alone, which means the whole model can hinge on one assumption about the distant, essentially unknowable future being roughly reasonable.

How MoatScan AI actually runs the model

Every DCF on MoatScan AI answers those same three questions, but the engine deliberately keeps the arithmetic out of the AI's hands. Growth, margin, and reinvestment figures are pulled from real scraped financials and analyst forecasts wherever they exist, and the discounting math itself — the part of the process with genuinely correct and incorrect answers — runs as fixed, deterministic code rather than a language model guessing at a spreadsheet. The AI's job is narrower and more honest: explain the resulting number in plain English, rather than compute it fresh (and differently) every time someone asks.

The explicit forecast window isn't a fixed round number applied to every company — it's assigned by how fast the business is actually growing right now. A company guided toward very high near-term growth gets a longer runway, up to a full ten years, before the model exits to terminal growth; a mature, steady grower gets a shorter one. Coca-Cola (KO), growing revenue in the low single digits, gets a five-year explicit window. Alphabet (GOOGL), growing faster but well short of hypergrowth pace, lands at seven. A company growing revenue upward of 25% a year earns the full ten. That mirrors how a careful human analyst already builds these models by hand — a hypergrowth business gets a long, gradual fade toward normal growth, while a mature one is assumed to already be there.

Terminal growth itself is anchored rather than guessed — pulled from published country- and sector-level long-run growth benchmarks instead of an AI picking a comforting figure, and capped well below anything an economy can plausibly sustain forever. And for businesses where a standard DCF simply doesn't fit — banks and REITs, where capital is the raw material rather than something invested into the operating business, or companies with no positive cash flow at all to discount — the model deliberately declines to force one. Financial companies get a dividend discount model instead; genuinely cash-burning, pre-revenue names get an honest "no fair value available" reading rather than a fabricated number, because a DCF built on negative cash flow is worse than no DCF at all.

Where the model can mislead you

None of this makes a DCF objectively "right." A DCF is a structured way to be explicit about your assumptions — it is only ever as good as the growth, margin, and discount-rate inputs that go into it, and every one of those inputs is a forecast, not a fact. Two careful analysts using an identical framework can land on fair values 30% apart just by disagreeing on year-three growth or a single percentage point of WACC. That isn't a flaw in the framework; it's an honest property of trying to price a distant, uncertain future in the first place.

The healthiest way to use a fair value estimate — MoatScan AI's or anyone else's — is as the start of a question, not a verdict. Look at the assumptions behind the number, not just the number itself. Does the growth rate look achievable given the company's economic moat and its recent return on invested capital? Does the discount rate reflect real balance-sheet risk? Is a big chunk of near-term free cash flow about to get absorbed by a capital spending cycle the model hasn't fully priced in — the exact dynamic now reshaping AI infrastructure spending across the largest technology companies? A fair value estimate that survives those questions is worth far more than one taken at face value.

Start with a company you already know

Apple (AAPL) is a useful first read precisely because most investors already have intuitions about its growth and durability — you can sanity-check the model's assumptions against what you already believe, instead of trying to evaluate both the company and the framework at the same time. From there, MoatScan AI's undervalued screen surfaces moat-rated companies currently trading below their modeled fair value — worth treating as a list of questions to investigate, not a shopping list.

Fair value estimates are model outputs based on stated assumptions, not price targets or investment recommendations. Two reasonable analysts can, and often do, disagree.