Why Longer Useful Lives for Equipment Can Flatter Profits

What happens to a company's profit when its accountants decide the servers will last six years instead of four? The servers don't change. The cash already spent on them doesn't change. The electricity bill, the data center lease, the engineers' salaries — none of it moves. And yet operating income goes up, sometimes by billions of dollars, starting the first day of the new fiscal year.

That's not a trick, and it's not fraud. Depreciation is an estimate by design: a company buys an asset, guesses how long it will be useful, and spreads the cost across that guess. When the guess changes, the expense changes with it. GAAP permits this, requires the effect to be disclosed, and then lets the new number flow through the income statement with no restatement of the past. The basic mechanics of depreciation are simple. Reading a change to them is where most investors stop paying attention.

I want to compare two ways depreciation expense can fall — one that says something about the business, and one that says something about an assumption — and then walk through the line items of a real case: Microsoft's decision, made in July 2022, to stretch the lives of its server and network equipment from four years to six.

A Change in Estimate Is Not a Change in Policy

Start with the rule, because the accounting classification determines how much you're told. Under ASC 360, property and equipment is depreciated over its estimated useful life. Under ASC 250, revising that life is a change in accounting estimate, not a change in accounting principle. The distinction matters. A change in principle (switching inventory methods, say) is applied retrospectively, so prior years are restated and you can compare like with like. A change in estimate is applied prospectively. The old years stay as they were reported; only the current and future periods carry the new number.

Change in accounting estimate (ASC 250-10-45-17 and 50-4): accounted for in the period of change and in future periods, with no restatement of prior periods. When the change affects several future periods, as a revised depreciable life does, the company discloses its effect on income from continuing operations, net income and per-share amounts for the current period.

So the disclosure exists. But it's usually one paragraph, often inside the property and equipment note or the accounting-policies note, and it's typically quantified for the year of the change only. The following year's comparison is quietly contaminated, and nobody is required to remind you.

That asymmetry is why I treat a useful-life extension as an earnings quality question first. Not because the new estimate is wrong — it may be more accurate than the old one — but because the reported growth rate in the year of the change is mixing two things: what the business did, and what the accountants re-decided.

Two Ways Depreciation Can Fall

A drop in depreciation expense, on its own, is ambiguous. Two very different stories produce it, and the cash flow statement is where you tell them apart.

Pattern one: the asset base is shrinking or aging

Here depreciation falls because there's less to depreciate. Capital expenditure runs below depreciation for several years, old assets reach the end of their schedules and drop out, and net property and equipment flattens or declines. Profit margins can look better for a while. The tell is the capex-to-depreciation ratio sitting near or below 1.0x over multiple years, which is the harvesting pattern discussed in our piece on maintenance versus growth capex. The business is consuming its asset base faster than it replaces it.

Pattern two: the same assets, re-estimated

Here depreciation falls while capex and net property and equipment keep rising. That combination shouldn't happen mechanically — a bigger asset base ought to carry a bigger depreciation charge — so when it does, the explanation is almost always in a footnote. Either the mix shifted toward longer-lived assets (buildings and land instead of servers), or the company lengthened its useful-life assumptions.

The quickest screen is a single ratio, computed from three line items you already have.

Depreciation rate = Depreciation expense ÷ Net property and equipment at the start of the year. A sudden drop in this rate while capex is climbing points to a re-estimate or a mix shift, not to a smaller asset base.

It's crude. Net PP&E includes land and construction in progress that aren't being depreciated yet, so the level of the ratio means little. The change from one year to the next is what you're looking for.

Microsoft, Fiscal 2023: The Line Items in Order

Microsoft's fiscal year ends June 30, so its fiscal 2023 is the year ended June 2023. Its 10-K for fiscal 2023 says plainly that "in July 2022, we completed an assessment of the useful lives of our server and network equipment," and that the estimated lives of both were increased from four years to six. Here is what that does to each line, in the order I'd read them.

Depreciation expense

The property and equipment note in Microsoft's 10-K for fiscal 2023 reports depreciation expense of $9.3 billion in fiscal 2021, $12.6 billion in fiscal 2022 and $11.0 billion in fiscal 2023. The note adds that "depreciation expense declined in fiscal year 2023 due to the change in estimated useful lives of our server and network equipment." A decline of $1.6 billion, in a year when the company was building data centers as fast as it could.

Capital expenditures

Additions to property and equipment on the cash flow statement were $23.9 billion in fiscal 2022 and $28.1 billion in fiscal 2023, rising again to $44.5 billion in fiscal 2024, per Microsoft's 10-Ks for those years. Capex to depreciation went from about 1.9x in fiscal 2022 ($23.9 billion ÷ $12.6 billion) to about 2.6x in fiscal 2023 ($28.1 billion ÷ $11.0 billion). That is pattern two, not pattern one. The company was spending more, not less.

Property and equipment, net

Net PP&E on the balance sheet was $59.7 billion at June 2021, $74.4 billion at June 2022 and $95.6 billion at June 2023. Run the screen: fiscal 2022 depreciation of $12.6 billion against a starting base of $59.7 billion is a rate of about 21%. Fiscal 2023's $11.0 billion against $74.4 billion is about 15%. A six-point drop in a single year, while the base grew by more than a quarter. In fiscal 2024 the rate was roughly 16% ($15.2 billion ÷ $95.6 billion, from the 10-K for fiscal 2024) — the new, lower level stuck.

Operating income and net income

This is the line that matters most, and the one where the disclosure earns its keep. Microsoft's 10-K for fiscal 2023 quantifies the effect: based on the equipment on the books at June 30, 2022, the change "was an increase in operating income of $3.7 billion and net income of $3.0 billion, or $0.40 per both basic and diluted share."

Now set that against the reported results. Operating income was $83.4 billion in fiscal 2022 and $88.5 billion in fiscal 2023, growth of about 6.2%. Take out the $3.7 billion and fiscal 2023 operating income would have been about $84.8 billion — growth of roughly 1.7%. Net income went from $72.7 billion to $72.4 billion, a decline of about 0.5% as reported. Without the $3.0 billion it would have been closer to a 4.6% decline.

So a year that read as modest growth was, on a like-for-like depreciation basis, nearly flat at the operating line. The estimate change supplied about 4% of fiscal 2023 operating income ($3.7 billion ÷ $88.5 billion) and most of the reported growth. None of this is hidden. All of it is easy to miss if you read the income statement and stop.

Is the New Life Right? What the Footnote Can't Tell You

Here's where I have to be honest about the limits of the method. A longer useful life isn't automatically aggressive. Server fleets do last longer than they used to, software can stretch hardware, and a company that runs its own data centers knows its failure and replacement rates far better than any outside analyst. If the six-year life is the true economic life, then the four-year schedule was overstating expense and the new numbers are more accurate, not less. I'm genuinely unsure where the right answer sits for equipment bought during a period of fast-changing chip generations, and I'm suspicious of anyone who is certain either way.

What the change does guarantee is a shift of expense into later years. Cash spent is cash spent; the cash flow statement is untouched by any of this. Depreciation only decides when that cash cost reaches the income statement, so a longer life lowers expense today and raises it — relative to the old schedule — at the back end, or loads it into a write-down if the assets are retired early. That's the accrual-versus-cash distinction in its purest form, and it's why cash flow conversion — the first of the accounting red flags I'd check — is the better anchor when a depreciation estimate has just moved.

Three things I'd check in the filings after any useful-life extension:

• Losses on disposal and accelerated depreciation in later years. If assets keep getting retired before the end of their new lives, the estimate was too long, and the catch-up arrives as a separate charge.
• Impairment language in the property and equipment note. A [long-lived asset impairment](/blog/impairment-charges-long-lived-assets) is the other route by which an optimistic life gets corrected — usually all at once.
• Peer disclosures. If competitors running similar equipment are shortening lives while one company lengthens them, the gap deserves an explanation in MD&A. Changes in this area have gone both directions in recent years, which is itself a sign of how much judgment is involved.

And one thing to do in your own numbers: carry the disclosed effect forward. If a company tells you the change added $3.7 billion in the year it happened, the following year's growth rate is clean only against a restated base, and nobody will restate it for you.

Key Takeaways

• A useful-life extension is a change in accounting estimate under ASC 250. It's applied prospectively with no restatement, and its effect is disclosed for the year of change — usually in a single paragraph.
• Depreciation falling while capex and net PP&E rise is the signature of a re-estimate or mix shift, not a shrinking asset base. The depreciation-rate screen (depreciation ÷ opening net PP&E) catches it in a minute.
• In Microsoft's fiscal 2023, the server and network equipment change added $3.7 billion to operating income — about 4% of the total and most of that year's reported growth, per the company's 10-K.
• Behind the GAAP number, cash didn't change at all. A longer life moves expense into later years; it doesn't remove it. Judge the estimate by what happens to disposals, impairments and cash conversion afterward.