10×, 20×, 30× revenue — what has to be true

In venture capital and growth investing a company is priced as a multiple of its revenue: ten times, twenty times, in 2021 sometimes fifty. The number is quoted as if it were an observation. It is not — it is a forecast in disguise. Every revenue multiple is a discounted cash flow run backwards, and running it forwards tells you exactly what the price assumes about growth, margin and time. This piece does that for the 10×, 20× and 30× tiers, shows why a decade of near-zero interest rates made those tiers look ordinary, and asks what the 2022 reset actually destroyed. One simple model is used throughout; its code is the table on this page, not a black box.

A multiple is a DCF run backwards

The value of any business is the cash it will hand back to its owners, discounted for time and risk:

EV  =  Σ  FCFₜ / (1 + r)ᵗ   +   Terminal value / (1 + r)ᴺ

Terminal value  =  FCF_N × (1 + g∞) / (r − g∞)

where FCF is free cash flow, r the discount rate (the return an investor demands for waiting and for risk), N the number of years before the company is mature, and g∞ the rate it grows at for ever afterwards — roughly nominal GDP, 3%. The second line is Myron Gordon's 1959 growth formula, and almost everything interesting on this page lives in its denominator, r − g∞.

Start with the end state. A mature software company with a 25% free-cash -flow margin, growing 3% a year, discounted at 10%, is worth 3.7× revenue (0.25 × 1.03 / 0.07). That is the anchor: it is what a great business is worth once the growth is over. So a company priced at 20× today is claiming that its revenue will be several times larger by the time it looks like that — large enough that 3.7× of future revenue, discounted back, is 20× of today's. That gap between today's multiple and the mature multiple is the growth the price has already spent.

To make this concrete, the model below takes a standard company: burning 10% of revenue today, ramping linearly to a 25% FCF margin over ten years, then growing at 3% for ever. It then solves for the one number the price is really about — the constant annual revenue growth over those ten years that makes the DCF equal the multiple.

What has to be true at 10×, 20×, 30×

Read the bold row. At a 10% discount rate — a reasonable cost of equity for a listed software company — 10× revenue needs about 19% growth every year for a decade, 20× about 28%, and 30× about 33%. Those do not sound extreme until you compound them: 30× revenue is a bet that the company will be roughly 18× its current size in ten years, and then turn a quarter of that into cash. Two things jump out.

• The multiple is far more sensitive to duration than to growth. Shorten the forecast to five years and 20× requires 51% a year; stretch it to fifteen and it requires 21%. A high multiple is less a bet on how fast than on how long.
• The discount rate moves the answer as much as the business. The same 20× that demands 28% at 10% demands only 14% at 6%. Nothing about the company changed — only the price of time. Hold that thought; it is the whole of the zero-rate story.

The operating facts behind each tier

Growth is the headline, but a growth rate is an output. What produces it — and what lets it become cash rather than a bigger loss — is a short list of unit-economics facts that diligence actually checks. Net revenue retention (how much last year's customers spend this year) matters most, because above ~120% the installed base grows on its own and the sales team only has to add to it. Gross margin decides how much of each revenue dollar can ever become free cash flow. The Rule of 40 — growth plus FCF margin at or above 40% — is the market's shorthand for whether growth is being bought too expensively.

The last column is where most high-multiple investments fail. Chan, Karceski and Lakonishok's 2003 study of the persistence of growth found that past high growth barely predicts future high growth: once you control for chance, very few companies string together a decade of above-average expansion. Mauboussin's Base Rate Book puts numbers on the tail — sustained 20%+ growth over ten years is a thin slice of all firms, thinner the larger the starting revenue. A 30× price is not wrong because the math is wrong. It is usually wrong because it asks a company to be in the top few percent of a distribution, and prices it as if it already were.

How venture capital actually uses the multiple

Venture investors rarely build a DCF of the median outcome, because the median startup is worth zero. They price with the venture capital method: work backwards from an exit, and demand a return large enough to pay for all the companies that fail. Returns in venture follow a power law — one or two investments return the fund — so each check has to be capable of returning 10× or more on its own.

Worked once: a Series A at 30× on $5m of ARR is a $150m post-money valuation. To return 10× the fund needs a $1.5bn exit (ignoring dilution, which makes it worse). If the public market pays 8× revenue at exit, that exit requires $188m of revenue — 38× today's — within 7 years. That is 68% growth a year for seven years. This is why the industry folklore is “T2D3”: triple, triple, double, double, double — 72× in five years. A 30× entry price is not a statement that the company is worth 30× revenue. It is a statement that there is, say, a one-in-ten chance it becomes one of the rare companies in the table above, and that the investor is paying for that option.

The VC method and the DCF are the same arithmetic with different inputs. The VC's high target return (25–50% a year) is simply a high discount rate that absorbs the probability of failure. Which means the venture price is also a function of the risk-free rate — through the exit multiple it assumes, and through the return LPs will accept when their alternative pays nothing.

Why financial assets inflate when rates go to zero

Put the Gordon denominator back on the table: value ∝ 1 / (r − g). The discount rate r is the risk-free rate plus a risk premium. From 2009 to 2021 the risk-free rate in the US, Europe and Japan sat near zero, and in real (inflation-adjusted) terms it was negative for long stretches — the 10-year TIPS yield reached about −1% in 2021. At the end of 2020 around $18tn of the world's bonds had a negative yield: investors were paying governments to hold their money.

When r − g is small, the price becomes explosive in it. Move a mature company from r − g = 2% to 4% and its value halves; from 4% to 2% and it doubles. And the effect is not uniform across assets. Dechow, Sloan and Soliman showed in 2004 that equities have a duration, exactly like bonds: the average time until the cash arrives. A utility pays you mostly this decade. A hypergrowth startup pays you nothing this decade and everything after it. The longer the duration, the more a change in r moves the price.

Moving from a 10% to a 6% discount rate lifts the mature software company from 4.8× to 12× revenue — and the hypergrowth company from 24× to 73×. The end-state anchor itself goes from 3.7× to 8.6×. That is the mechanical core of the zero-rate era: the assets with the least cash today gained the most from cheap money, because 84% of the hypergrowth company's value sits after year ten, where the discount rate compounds hardest. Tech was not an accidental beneficiary of ZIRP. It was the purest expression of it.

The five channels

The discount rate is only the first mechanism. Zero rates inflated financial assets through at least five, and they reinforce one another:

Venture capital sat at the end of all five. US venture deal value roughly doubled from about $170bn in 2020 to about $345bn in 2021 (PitchBook-NVCA). Crossover and hedge funds that had never done private deals wrote late-stage checks at public-market speed. Gompers and Lerner had measured this in 2000: when money flows into venture funds, the price of the same deals rises without their outcomes improving — “money chasing deals.” Kaplan and Schoar found the other half in 2005: funds raised at the top of the cycle deliver the cycle's returns. And the exit window closed the loop: 2021 saw a record wave of IPOs and roughly 600 SPAC listings, which marked private rounds up, returned cash to LPs, and funded the next round. Each channel is rational on its own. Together they are reflexive — George Soros's word for prices that change the fundamentals they are supposed to measure.

Then the price of time came back

In 2022 the Federal Reserve raised rates faster than at any time since the early 1980s, and the real 10-year yield rose by nearly three points. The median listed cloud multiple fell from about 16× to about 5.5× — a compression of 66%. Most of those companies kept growing: revenue rose perhaps a third over the period. Price is multiple × revenue, so even with that growth a median holder lost about 54%. The long-duration names fell much further: the pandemic winners — video conferencing, connected fitness, e-commerce enablers — lost 80–90% from peak. The Nasdaq Composite dropped about a third in 2022. In private markets the adjustment was slower and quieter: down rounds, flat rounds disguised by structure, and a large cohort of 2021 unicorns that simply stopped raising.

Were prices destroyed?

The question has three different answers depending on what “prices” means. They are worth separating, because they lead to opposite conclusions.

1. Was value destroyed? Mostly not — it was redistributed

A fall in a multiple destroys paper wealth, not necessarily real value. The businesses that were worth 16× at a −1% real rate are, with the same cash flows, worth 5–6× at +2%; neither number is the “true” one, they are the same asset priced against different alternatives. And the aggregate tells a surprising story: listed US tech recovered its 2022 drawdown by 2024 and went on to new highs, carried by the largest companies with the most cash flow — and by the AI buildout (see the hyperscaler debt machine). The destruction was concentrated precisely where the duration math says it should be: unprofitable, long-dated, growth-at-any-price companies. That is not a malfunction. It is the model working.

There is a less comfortable version of the same point. Much of the 2020–21 venture money was not lost in a crater; it was spent — on salaries, on cloud bills, and above all on customer acquisition bought from Google and Meta. A meaningful share of the capital that vanished from startup balance sheets reappeared as revenue at the platform incumbents. Cheap money did not only inflate prices; it transferred value from the long tail of challengers to the few firms selling them the picks and shovels.

2. Was the price signal destroyed? For a decade, partly yes

Hayek's 1945 argument is that prices are how an economy aggregates information nobody holds in full. The interest rate is the most important price of all — the price of time — and for twelve years it was set administratively close to zero. The consequence the DCF predicts is subtle: when r − g is small, the price of a long-duration asset is dominated by its terminal value, 84% or more of it resting on assumptions about year eleven onward. A price that is mostly terminal value is barely informative about the business; it is informative about the narrative. Capital allocated on those prices went to business models that only worked while money was free — subsidised delivery, subsidised rides, scooter fleets, ten-minute grocery — many of which are gone. The BIS documented the macro version in 2018: as rates fell, the share of “zombie” firms unable to cover their interest from profits rose, crowding out healthier competitors. The misallocation was real and it is not recovered when the multiple recovers.

3. Is the reset over? The AI era says the question moved

With real rates around 2% since 2023, the ZIRP explanation should have ended the era of 30× revenue. It did not: in 2024–26 the leading AI companies raised at 30× to over 100× annualised revenue. The formula says why, and why this is a different bet. A high multiple can come from a small r or a large g; 2021 was mostly small r, 2025 is mostly large g — companies growing several-fold a year rather than 60%. That is more honest, because it rests on observed growth rather than the price of money. It is also fragile in its own way. The base rates on growth persistence have not changed, and AI applications carry inference costs that push gross margins well below the 75–80% the software tiers above assume (see the price of a token). Lower gross margin lowers the terminal FCF margin, and that anchor multiplies everything else.

A decision rule

For an investor, the useful habit is to never accept a revenue multiple as a fact and always invert it. Four questions do most of the work:

1. At my honest discount rate, what growth for how many years does this price require? Use the table above; the answer is one bisection away.
2. What is the base rate for that? If the price needs top-5% growth persistence, the burden of proof is on the specific reason this company is in the top 5%.
3. Which variable is the price leaning on — r or g? A multiple that depends on low rates is a macro bet; one that depends on growth is an execution bet. Know which one you are making.
4. What is the mature margin, really? Retention and gross margin decide whether the growth ever turns into cash. A company that grows but cannot reach a 20–25% FCF margin is not worth a high multiple at any discount rate.

The ten-year lesson of ZIRP is not that high multiples are wrong. It is that a multiple has two authors — the company, which writes the growth, and the central bank, which writes the discount rate — and in 2020–21 most investors were reading only the first.

The literature

Appendix — methodology

• What-has-to-be-true table. Constant growth g for 10 years; FCF margin from -10% to 25%; terminal growth 3%. Solved for g by bisection. Multiples are on current (trailing) revenue; a forward multiple is lower by a factor of (1 + g).
• Archetypes. Growth fades linearly from its starting rate to 3% over ten years; margins ramp linearly; same terminal value.
• Discount rates. 10% approximates a listed software company's cost of equity with a ~4% risk-free rate; 6% is the same risk premium on a ~0% risk-free rate, as in 2020–21. Venture investors' 25–50% target returns are not comparable: they absorb the probability of failure into the rate.
• 2020 → 2022 price effect. Median multiple change applied to ~35% cumulative revenue growth; an illustration of multiple × revenue, not an index return.

Sources consulted

• FRED, Federal Reserve Bank of St. Louis — DGS10, DFII10, FEDFUNDS (annual averages).
• Meritech Capital public comps; Jamin Ball, Clouded Judgement; BVP Nasdaq Emerging Cloud Index — EV / NTM revenue.
• PitchBook-NVCA Venture Monitor — US venture deal value, 2020–2023.
• Aswath Damodaran, Narrative and Numbers (2017) and his annual cost-of-capital datasets.