What to do when terminal growth is above the discount rate

The Gordon growth terminal value formula is TV = FCF × (1 + g) / (r − g), and it is only defined for r > g. Nothing in the arithmetic enforces that. Put a growth rate equal to or above the discount rate into it and you get a negative number that still looks like an ordinary output — no exception, no NaN, just a wrong sign sitting where a company's value is supposed to be. Get close to the boundary without crossing it and the formula stays defined but the multiplier applied to a single year's cash flow goes past anything a perpetuity should reasonably be worth.

A natural way to pick a terminal growth rate without thinking about it twice is to carry a company's own trailing growth straight into the terminal year. We checked what that naive choice actually does, live, for the same 20 large filers surveyed in the first two guides on this site, against a single fixed discount rate of 9% — the middle of the range most default DCF templates use for a diversified large-cap, and the value this site's own calculator falls back to before anyone changes it. Half of them already break the formula outright or land within one bad assumption of breaking it.

Why the formula breaks silently

TV = FCF × (1 + g) / (r − g) — the standard closed-form value of a cash flow growing at a constant rate g forever, discounted at rate r.

As g rises toward r, the denominator shrinks toward zero and the multiplier 1/(r−g) rises without bound. At g = r it is undefined. Past g = r the denominator goes negative and the whole expression flips sign, so a positive free cash flow comes out as a negative terminal value — a number a spreadsheet will add up, discount back to the present, and report next to a dollar sign with no warning that the premise underneath it no longer holds.

A perpetuity growing at or above the rate you discount it at is worth an infinite amount in the real sense of the formula. The negative number is not a smaller infinity; it is the algebra of a denominator that crossed zero, with no meaning as a value.

Trap one: your own trailing growth is usually not a terminal growth rate

Terminal growth is supposed to describe a company forever, past the point where its advantages have fully normalized — realistically something near the long-run growth of the economy it sells into, often argued down to a number in the low single digits. Trailing growth describes what actually happened over the last several years, including whatever temporary conditions were in effect at the start and end of that window. Treating the second as a stand-in for the first is the naive choice this survey tests, and it fails for reasons that are different for every company that fails it.

Chevron: a real company, a real recovery, a growth rate that is really an artifact. Chevron's free cash flow (operating cash flow minus capex, same subtraction as the free cash flow guide) was $1.655bn in FY2020 and $16.592bn in FY2025 — a naive 5-year CAGR of 58.57%, against a 9% discount rate. The number is arithmetically correct. It is also almost entirely the 2020 oil-price collapse, the year capital spending was cut and cash flow cratered across the industry, divided out of a normal recent year. Carried into a terminal year, it does not describe Chevron's future; it describes how low 2020 was.

Trap two: real growth is just as dangerous as artifact growth

Chevron's number is a base-year accident. Nvidia's is not. Nvidia's revenue grew at a live, SEC-filed 5-year CAGR of 66.90% through FY2025 — AI datacenter demand, not a depressed starting point. That is precisely what makes it the more important case: a company's real, recent, well-documented growth rate is still not a number a perpetuity formula can hold forever. Every one of the eight companies below marked "breaks the formula" failed r > g for its own specific reason, not a data error.

JPMorgan: no crisis, no AI story, still 435x. JPMorgan's revenue grew at an unremarkable 8.75% a year over five years — below Nvidia, below Chevron's artifact, a growth rate nobody would call aggressive on its own. Fed into 1/(r−g) against a 9% discount rate, it still produces a terminal multiplier of 435 times a single year's figure. The formula does not care that 8.75% sounds reasonable; it only cares how close it sits to 9%.

What 20 large filers' own trailing growth does to the formula, read 26 October 2026

Company	Naive g	Source	TV multiplier	Verdict
Apple	6.13%	FCF	37.0×	near-break
Microsoft	3.60%	FCF	19.2×	within range
Nvidia	66.90%	Revenue	—	breaks
Amazon	−21.57%	FCF	2.6×	within range
Alphabet	11.33%	FCF	—	breaks
Meta	14.35%	FCF	—	breaks
Tesla	17.43%	FCF	—	breaks
JPMorgan	8.75%	Revenue	435.0×	near-break
Exxon	12.85%	Revenue	—	breaks
Johnson & Johnson	−0.49%	FCF	10.5×	within range
Walmart	−10.38%	FCF	4.6×	within range
Procter & Gamble	−0.57%	FCF	10.4×	within range
Coca-Cola	no annual facts	—	—	insufficient data
Pfizer	−5.71%	FCF	6.4×	within range
Intel	−7.46%	Revenue	5.6×	within range
Chevron	58.57%	FCF	—	breaks
AT&T	−7.83%	FCF	5.5×	within range
Boeing	9.00%	Revenue	—	breaks
General Electric	−9.57%	Revenue	4.9×	within range
Merck	9.38%	Revenue	—	breaks

"Naive g" is the 5-year CAGR of free cash flow (operating cash flow minus capex) where both have an annual series; otherwise the 5-year revenue CAGR. "Breaks" means naive g is at or above the 9% discount rate used throughout — the terminal value formula returns a negative number. "Near-break" means the multiplier 1/(r−g) exceeds 30×. Coca-Cola repeats the empty-array pattern the first guide on this site found: a 200 response, a valid units object, nothing inside it, on every tag checked.

What a correct approach does, in order

Never let trailing growth become terminal growth by default. A 5-year CAGR describes the last five years, including whatever base-year conditions happened to be in effect at either end.
Check the discount rate and the proposed terminal growth rate against each other before computing anything downstream. r > g is a precondition, not something the formula will enforce for you.
Treat a multiplier 1/(r−g) above roughly 15–20× as a signal to lower g, not a sign of a great business — the formula is amplifying proximity to the boundary, not conviction.
Fade a high growth rate down over an explicit multi-stage projection before the terminal year, rather than discounting one constant rate forever starting next year. The terminal year is supposed to be the company after it has already slowed to something sustainable.
Anchor the terminal growth rate to something that is itself bounded — long-run nominal GDP growth, or the discount rate minus a deliberate spread — rather than to the company's own recent history.
When trailing growth is negative for crisis- or cycle-specific reasons (the near-zero 2020 bases above cut the other way too), don't carry that into the terminal year either. A base-year artifact is an artifact in both directions.

Check this yourself

Every figure above comes from one script hitting SEC EDGAR, and both the script and its output are published rather than described:

terminal-growth-survey.json, the full result for all twenty companies, including Coca-Cola's empty answer.
terminal-growth-survey.py, the script that produced it. Run it and the numbers regenerate. If it disagrees with this page, the script is right and the page is stale.

Read from SEC EDGAR on 26 October 2026 · the 9% discount rate is a stated constant, not a per-company estimate · corrections welcome at [email protected]

Want the arithmetic without the plumbing? The calculator flags a terminal growth rate that sits too close to your discount rate before it lets the number through.

Open the calculator