Depreciation Waterfall
Depreciation is not a percentage of revenue and it is not a percentage of the asset base. It is every asset the company has ever bought, each wearing out on its own clock. A waterfall says that literally: one row per vintage, each spending its own cost over its own useful life, with the year’s charge as the column total.
Change the life, switch to an accelerated method, or type over a capex year, and watch which years move. The answers people get wrong are all here: the cliff when the existing base rolls off, the charge that keeps climbing after capex has peaked, and the difference between spending to grow and spending to stand still.
Assumptions
Three inputs drive the whole schedule. Change one and watch which years move.
The same charge every year of the asset’s life. What almost every reported set of accounts uses, and the default a model should start from.
The waterfall
One row per vintage. Each row spends its own cost over its own life, and the year’s depreciation is the column added up. This is why the charge keeps climbing after capex has already peaked: the layers stack.
| Vintage | Cost | 2025 | 2026 | 2027 | 2028 | 2029 | 2030 | 2031 | 2032 |
|---|---|---|---|---|---|---|---|---|---|
| Existing asset base | 138,589 | 19,798 | 19,798 | 19,798 | 19,798 | 19,798 | 19,798 | 19,798 | — |
| Capex 2025 | 6,626 | 6,626 | 6,626 | 6,626 | 6,626 | 6,626 | 6,626 | — | |
| Capex 2026 | — | 6,715 | 6,715 | 6,715 | 6,715 | 6,715 | 6,715 | 6,715 | |
| Capex 2027 | — | — | 6,167 | 6,167 | 6,167 | 6,167 | 6,167 | 6,167 | |
| Capex 2028 | — | — | — | 4,928 | 4,928 | 4,928 | 4,928 | 4,928 | |
| Capex 2029 | — | — | — | — | 3,076 | 3,076 | 3,076 | 3,076 | |
| Capex 2030 | — | — | — | — | — | 826 | 826 | 826 | |
| Capex 2031 | — | — | — | — | — | — | 826 | 826 | |
| Capex 2032 | — | — | — | — | — | — | — | 826 | |
| Total depreciation | 26,425 | 33,140 | 39,307 | 44,234 | 47,310 | 48,135 | 48,961 | 23,362 |
Capex figures are editable. Click one and type to see a single investment year ripple through every year that follows. Notice the existing base falling to nothing in 2032: that cliff is the single biggest reason a depreciation forecast built as a percentage of revenue goes wrong.
Maintenance, growth, and the multiple
Depreciation is roughly what it costs to stand still. Spend above it and you are buying capacity; spend below it and you are quietly consuming the business. The ratio of the two is the fastest read on which is happening.
| Year | 2025 | 2026 | 2027 | 2028 | 2029 | 2030 | 2031 | 2032 |
|---|---|---|---|---|---|---|---|---|
| Capex | 46,383 | 47,005 | 43,171 | 34,494 | 21,529 | 5,779 | 5,779 | 5,779 |
| Depreciation | 26,425 | 33,140 | 39,307 | 44,234 | 47,310 | 48,135 | 48,961 | 23,362 |
| Maintenance capexThe part that only replaces what wore out | 26,425 | 33,140 | 39,307 | 34,494 | 21,529 | 5,779 | 5,779 | 5,779 |
| Growth capexSpend above the depreciation charge | 19,958 | 13,865 | 3,864 | — | — | — | — | — |
| Capex / depreciationAbove 1.0 the asset base grows; below 1.0 it shrinks | 1.76× | 1.42× | 1.10× | 0.78× | 0.46× | 0.12× | 0.12× | 0.25× |
Watch the multiple fall through the forecast as capex tapers while the stacked layers keep the charge high. A business can be spending less than it depreciates for years before the asset base visibly ages, which is exactly the period in which the reported earnings look best.
Where it lands: the PP&E roll-forward
Depreciation is not only an expense, it is the thing that takes the asset back off the balance sheet. Opening net PP&E, plus capex, less depreciation, equals closing. That identity is the link between the schedule above and the balance sheet, and it has to hold in every year.
| Year | 2025 | 2026 | 2027 | 2028 | 2029 | 2030 | 2031 | 2032 |
|---|---|---|---|---|---|---|---|---|
| Opening net PP&E | 138,589 | 158,547 | 172,412 | 176,276 | 166,536 | 140,754 | 98,398 | 55,215 |
| Plus capex | +46,383 | +47,005 | +43,171 | +34,494 | +21,529 | +5,779 | +5,779 | +5,779 |
| Less depreciation | −26,425 | −33,140 | −39,307 | −44,234 | −47,310 | −48,135 | −48,961 | −23,362 |
| Closing net PP&E | 158,547 | 172,412 | 176,276 | 166,536 | 140,754 | 98,398 | 55,215 | 37,632 |
Why you take D&A from the cash flow statement, not the income statement
Depreciation is an operating expense, so it is allocated to whichever line consumed the asset. The depreciation on a factory sits inside cost of sales. The depreciation on head-office fit-out and IT sits inside SG&A. Some filers show a separate D&A line and some do not, and a company can move between presentations without anything economic changing.
That makes the income statement an unreliable place to read the total. The cash flow statement is not: the whole non-cash charge has to be added back to reconcile net income to operating cash flow, so it appears there as one clean figure whatever the income statement did with it. Take D&A from the cash flow statement, and check it against the PP&E note, which gives the split between depreciation and amortisation and the asset lives you need for the schedule above.
This is also why EBITDA built by adding back the income statement’s visible D&A line can differ from EBITDA built off the cash flow statement. If a comparable set looks inconsistent, that difference is usually where it came from.
Now push a depreciation change through all three statements under interview conditions.
Open the three-statement drillKeep going
All labsNext in Accounting · 8 min
Three-Statement Linker
"Depreciation goes up by $10, walk me through the three statements." A fresh event every time, graded line by line across the income statement, cash flow and balance sheet, with a check that turns green only when your balance sheet balances.
Deal Simulations · 15 min
Capital Stack Challenge
Finance a real buyout. Size the revolver, term loans, notes, PIK and equity against the lender caps of the day, hold the structure through a shock, then see how the sponsor actually did it, with every figure linked to the filing.