Opinion · Software economics · September 10, 2026
Stock compensation is not free. It just sends the bill somewhere else.
I used to read “non-cash” as almost synonymous with “not a cost.” That is too easy. Stock-based compensation does not require a cash payment when it is booked, but employees are still being paid. The owner’s bill shows up later: in a larger share count, or in the cash a company spends trying to keep that share count from growing.
Figure 1 · The three-statement route
The expense leaves the income statement, returns in cash flow, then reaches the owner through dilution or buybacks.
The point is not that free cash flow is fake. It is that a cash-flow addback and an ownership cost can both be true at the same time.
Read the three numbers together
My simple checklist is now: stock compensation as a percentage of revenue, change in the share count, and cash spent on repurchases. One number alone can give a company too much credit or too little. A fast-growing company can scale compensation more slowly than revenue. A company can also spend heavily on buybacks and still end the year with more shares outstanding.
| Latest completed fiscal year | SBC / revenue | Year-end shares | What stands out |
|---|---|---|---|
| Palantir FY2025 | 15.3% | +2.2% | Revenue grew faster than SBC; the intensity improved. |
| Datadog FY2025 | 21.9% | +3.0% | GAAP loss and a large adjusted-profit gap. |
| CrowdStrike FY2026 | 22.8% | +2.2% | SBC was 88.9% of company-reported FCF. |
| Snowflake FY2026 | 34.2% | +3.0% | $873.5m of buybacks; shares still increased. |
How to read this. These are reported company figures from the latest completed annual filings, not AEA forecasts. Share-count changes can also reflect acquisitions, convertible settlements, employee purchase plans, and tax withholding. They should not be treated as a perfect measure of SBC alone.
Snowflake is the cleanest warning
Snowflake recorded $1.60 billion of GAAP stock compensation in FY2026, equal to 34.2% of revenue. It also reported $1.12 billion of free cash flow and spent $873.5 million repurchasing 4.925 million shares. That is not a contradiction. The company produced cash, used much of it to buy stock, and still finished with year-end shares up 3.0%. The picture is more honest when all three facts stay in the frame.
And Palantir is the counterexample worth respecting
The fair version of this argument is not “SBC bad.” Palantir’s Q1 2026 revenue grew 85% year over year while SBC grew 30%; SBC intensity fell to 12.4% from 17.6%. That is what operating leverage looks like in this particular metric. The share count still deserves attention, but the trend is meaningfully better than simply seeing an absolute dollar SBC figure rise and stopping there.
What I will monitor
For every software company I read, I will track the SBC/revenue trend, the gap between GAAP and adjusted operating margin, diluted-share growth, repurchase dollars versus actual share reduction, and per-share free-cash-flow growth. It is not a one-line screen. It is an ownership question.
BoundaryWhat this article does not claim
This is not a claim that a company should expense SBC in cash flow, nor a price target or investment recommendation. It is an accounting-and-ownership framework based on filings, and it can be wrong if used without the business context around talent, growth, and buybacks.