What Is a Capital-Light Business? Why They Compound
A capital-light business is one that can grow its profit without putting much new money into equipment, inventory or acquisitions. Because it keeps most of what it earns, it can pay owners or reinvest at high returns, which is why such businesses compound.
This site tests it with one ratio: capital added per $1 of profit added. Lower is better, and below $1 is the territory Warren Buffett's See's Candies made famous.
Why does a capital-light business compound faster?
Profit that does not have to be reinvested is free to leave the company. Compare two firms that each earn $100 million. One must spend $80 million a year on factories to keep growing; the other spends $10 million. The second has $70 million more to pay out or buy back shares with.
Over many years that gap compounds in the owners' favour. It also means growth is less dependent on borrowing or issuing shares. See free cash flow yield for how that cash is measured.
How does this site measure capital-light growth?
The method page sets the formula. This post explains the idea; for the history of the benchmark, read the See's Candies test.
- Capital = operating working capital + net property, plant and equipment (the "full cost" view adds goodwill).
- Profit = pre-tax operating profit, averaged over three years at each end of a 10 or 15 year window.
- Capital per $1 = capital added ÷ profit added.
- Return floor: the default list also wants a 25% or better return on that capital, and profit at least doubled.
The reference points come from Buffett's 2007 letter: See's $0.42, FlightSafety $3.20 and a typical growing business $5.19.
What do real capital-light companies look like?
Ten-year results from the site's Capital-Light Compounders data (SEC 10-K filings, tangible capital, refreshed October 23, 2026). Dollar figures are billions, with three-year averages at each end. Costco, Nvidia and Microsoft are measured over fiscal 2017 to 2026, the others over 2016 to 2025.
| Company | Profit, then to now | Capital, then to now | Capital per $1 of profit added |
|---|---|---|---|
| Mastercard | $6.6B to $16.2B | -$0.3B to -$4.6B | None: capital fell |
| Nvidia | $3.0B to $81.6B | $1.9B to $44.4B | $0.54 |
| Costco | $4.4B to $10.5B | $12.9B to $18.3B | $0.90 |
| Microsoft | $35.7B to $131.1B | $11.0B to $183.7B | $1.81 |
| Amazon | $6.9B to $61.8B | $18.4B to $195.5B | $3.23 |
| Verizon | $26.3B to $26.9B | $86.9B to $108.3B | $34.31 |
Mastercard's negative capital is real: customers and partners fund its operations, so growth tied up nothing. Verizon sits at the other end, where profit barely moved and capital still rose by $21 billion.
Gold bars are Buffett's 2007 reference points; the others are this site's ten-year results. Mastercard is left out because its capital fell.
What are the traps in capital-light screens?
A low ratio can mislead, which is why the site adds a return floor and an earnings check.
- A big, low-return capital base. Exxon Mobil shows $0.86 of capital per $1 added over ten years, but its return on that capital is only 17% now, below the 25% floor, because profit rebounded from a weak start.
- Hidden capital. Content a streamer has paid for is not on the balance sheet as capital, so Netflix's figure is understated; the site says so on its stock card.
- Cycles. The start and end years of the window decide the answer for energy, chemicals and chips.
- Acquired growth. The full-cost view adds goodwill to show growth that was bought, not built.
None of this says whether a stock is cheap. The Compounders at a Fair Price page adds earnings quality, cash returned and a PEG price check.
Capital-light FAQ
Is capital-light the same as asset-light?
Mostly. Both describe a business that needs little equipment, inventory or other capital per dollar of profit. This site measures it by results: how much capital was actually added as profit grew.
Can a capital-light business have negative capital?
Yes. If customers pay before the company pays its suppliers, operating working capital can be negative, so capital needed can be zero or less. The site then reads capital per $1 of profit as none.
Why does the site exclude banks and REITs?
A lender's balance sheet is its product and a REIT's property is its business, so capital added per profit added does not measure the same thing. They show n/a.
Does capital-light mean cheap?
No. The test describes how a business grows, not what its shares cost. This site pairs it with a price lens on the Compounders at a Fair Price page.
Can a capital-light company stop being one?
Yes. Amazon's capital per $1 of added profit is $3.23 over ten years, a long way from See's at $0.42, because it spent heavily on warehouses and servers. The label describes the past window, not the future.