Where the cash goes
The cash-flow statement answers two questions, and the second one gets less attention than the first. How much cash did the business produce? And then: where did it go? Four of the answers are reported lines in the filings — capital expenditure, research and development, dividends paid, and share buybacks — so the split can be measured rather than guessed. What comes out is not a ranking of good and bad uses. It is closer to a fingerprint: sectors spend their cash in shapes that barely overlap.
Four uses, one denominator
Each column below divides one reported line by the same figure: that year's operating cash flow. Capital expenditure and research spending go back into the business; dividends and buybacks go out to shareholders. A company only appears in the table if it reported all four lines, which is a real restriction — most companies report some of them and not others, and a company that pays no dividend does not file a dividend line of zero, it files nothing at all.
Counting a missing line as zero would make a sector look stingier than it is; skipping it column by column would compute every column from a different set of companies, which would make the rows uncomparable with each other. Requiring all four is the honest option and it costs coverage: fewer than one company in five clears it, and those that do skew large. Financials, utilities, energy and real estate have too few qualifying companies to show a row at all.
| Sector | Companies | Capex ÷ cash | R&D ÷ cash | Dividends ÷ cash | Buybacks ÷ cash |
|---|---|---|---|---|---|
| Technology | 137 | 13.3% | 44.0% | 15.8% | 35.7% |
| Healthcare | 47 | 22.6% | 36.9% | 17.4% | 14.1% |
| Consumer discretionary | 53 | 28.4% | 26.6% | 15.1% | 22.4% |
| Industrials | 107 | 24.6% | 13.6% | 19.3% | 25.3% |
| Materials | 67 | 46.1% | 10.3% | 16.4% | 11.5% |
| Consumer staples | 38 | 24.7% | 6.9% | 41.6% | 9.7% |
The four medians do not add up to one hundred per cent and are not meant to. Each column is its own median, computed independently; a single company's four uses can add to more than the cash it produced, with the difference covered by debt or by cash it already held. The row is not a pie chart of one business — it is four separate statements of the form "the typical company in this sector puts about this much of its cash here".
Read across a row and the differences are the point. A sector whose largest column is capital expenditure is buying physical capacity; one whose largest column is research is buying future products; one whose largest column is dividends is returning cash on a schedule shareholders can plan around. Nothing on this page says which of those is the better use of a dollar. That judgement depends on what the business is for, and it is not a judgement a data table can make.
Size and intensity are different questions
Research spending shows this most clearly. Ask which companies spend the most on research and you get a list of the largest companies in the world, because the biggest budgets belong to the biggest firms. Ask instead which companies spend the largest share of their revenue on research and the list changes almost completely — and the two answers are both correct, to two different questions.
| Company | R&D expense | Revenue | R&D ÷ revenue |
|---|---|---|---|
| Atlassian | $2.7B | $5.2B | 51.2% |
| Regeneron Pharmaceuticals | $5.9B | $14.3B | 40.8% |
| Beone Medicines | $2.1B | $5.3B | 40.2% |
| Incyte | $2.1B | $5.1B | 39.9% |
| Synopsys | $2.5B | $7.1B | 35.1% |
| Vertex Pharmaceuticals | $3.9B | $12.0B | 32.6% |
| Rivian Automotive | $1.7B | $5.4B | 31.0% |
| Snap | $1.8B | $5.9B | 30.2% |
| Meta Platforms · both | $57.4B | $201B | 28.5% |
| Workday | $2.7B | $9.6B | 28.0% |
The table ranks by research divided by revenue, among companies with at least $5 billion in revenue, and marks any company that also appears in the ten largest research budgets. Usually there is barely an overlap. The revenue floor is there for a reason: below it the list fills with early-stage drug developers whose research is several times their revenue, which is arithmetically true and tells you about the stage of a business, not about how research-intensive an industry is.
What a single year does not show
All of this is one fiscal year. Capital expenditure is lumpy — a plant is built once and used for decades — so a company can look restrained in the year after a big build and extravagant in the year of one, with no change in how it is run. Buybacks are the most discretionary line of the four and the first to be cut, so a single year's figure carries less signal than the others. The year-by-year series on each company page is the view that survives these effects.
Two of the lines also mean slightly different things than their names suggest. Dividends paid is cash that left in that year, which is not the same as dividends declared. Buybacks are gross: a company can buy back shares and issue more of them in the same year, and only one of those two movements is in this line. Both figures are shown as filed; each company page links to the filing they came from.
Source: annual figures as filed with the SEC, read from XBRL. Every ratio here is built from two lines of the same filing, and each company page links to that filing. The tables are recomputed at every data refresh, so the text stays true while the numbers move.