Maintenance capital expenditures are the funds a company must spend to keep its existing operations running at their current capacity: replacing worn equipment, refreshing vehicles on their normal retirement schedule, patching roofs, installing mandatory safety upgrades. Financial statements don’t break these costs out from spending that expands the business, so investors have to estimate them. The estimate matters. A company reporting $50 million in total capital spending might need $35 million just to tread water, leaving $15 million for real growth. Get that split wrong and you’ll misprice the business.
What Counts as Maintenance Spending
Maintenance CapEx covers any spending required to preserve an asset’s current output without meaningfully expanding what the business can do. Under GAAP, costs that extend an asset’s life or increase its functionality get capitalized as improvements, while routine repairs get expensed. Maintenance CapEx sits in the middle: it’s capitalized because you’re buying long-lived assets, but the spending doesn’t make the business bigger or more productive than before.
Concrete examples: swapping out a failing motor on a production line so the machine keeps hitting the same output, replacing fleet vehicles on schedule, patching a warehouse roof. Regulatory upgrades count too. Installing safety sensors on presses to satisfy OSHA requirements doesn’t add capacity; it keeps the plant legally operational.1Occupational Safety and Health Administration. Machine Guarding – Presses – Presence Sensing Devices Stop this spending entirely and assets deteriorate until revenue falls or operations halt.
Growth CapEx is different. It buys new capacity: a second factory, an expansion of an existing plant, a bigger fleet. Growth spending is discretionary in ways maintenance is not. A company can pause growth during a downturn. Skipping maintenance is borrowing against the future, and the bill eventually arrives larger than it would have been.
Why the Number Matters
Warren Buffett introduced the concept of “owner earnings” in Berkshire Hathaway’s 1986 annual letter for this exact reason. He defined owner earnings as reported earnings plus non-cash charges like depreciation, minus the average annual maintenance CapEx required “to fully maintain its long-term competitive position and its unit volume.”2Berkshire Hathaway Inc. Chairman’s Letter – 1986 Buffett’s point: Wall Street’s standard “cash flow” number adds back depreciation without subtracting the real cost of maintaining assets, which makes almost every business look more profitable than it is.
EBITDA has the same flaw. It strips out depreciation as if maintaining the asset base were free. For a software company, the distortion is small. For an airline, utility, or mining operation, it can be enormous. An airline that doesn’t replace aging aircraft isn’t earning what its EBITDA suggests; it’s consuming its fleet. Academic research on the topic has found that companies whose reported depreciation consistently falls below their real maintenance needs tend to eventually record large asset write-offs and deliver negative stock returns.
If you’re valuing a business by discounting future cash flows, the figure you should discount is earnings after maintenance CapEx, not before. Miss by a few percentage points of revenue and the error compounds into a dramatically different valuation over a 10-year projection.
Where to Find the Data in a 10-K
Start with the company’s annual report filed with the SEC.3U.S. Securities and Exchange Commission. Investor Bulletin: How to Read a 10-K Three sections do the work.
The Statement of Cash Flows, under “Cash Flows from Investing Activities,” shows total capital expenditures as a cash outflow, usually labeled “purchases of property, plant, and equipment.” That total is your starting number. It lumps maintenance and growth together.
The notes to the financial statements contain a PP&E schedule that breaks assets into classes (machinery, buildings, land, vehicles) with historical cost, accumulated depreciation, and useful life assumptions. Those details feed the estimation methods below.
Management’s Discussion and Analysis is where leadership explains its spending decisions, sometimes splitting growth projects from ongoing maintenance. When the MD&A names a new facility or expansion project with a dollar figure, you can subtract that from total CapEx to isolate maintenance.3U.S. Securities and Exchange Commission. Investor Bulletin: How to Read a 10-K
What you won’t find anywhere in the filing is a line called “maintenance capital expenditures.” No accounting standard requires it. That absence is why estimation exists.
Three Methods for Estimating the Number
No single approach works everywhere. Experienced analysts run several and compare.
Depreciation as a Proxy
The simplest method treats annual depreciation as an approximation of maintenance CapEx. The logic: depreciation is the accounting estimate of asset value consumed each year, so replacing that consumed value should cost roughly the same.
The weakness is that depreciation reflects historical purchase prices, not current replacement costs. A machine bought for $100,000 a decade ago might cost $130,000 to replace today. Analysts commonly adjust depreciation upward by 10% to 15% to reflect inflation in parts and labor. This proxy works best for mature businesses with stable asset bases and predictable replacement cycles. It breaks down for fast-growing companies where depreciation lags the actual asset base, and for industries where technology shifts push replacement costs well above historical costs.
Total CapEx Minus Identified Growth Spending
When the MD&A names specific growth projects, subtract them from total capital expenditures. If the 10-K reports $80 million in total CapEx and the MD&A describes a $30 million plant expansion, the remaining $50 million is a reasonable maintenance estimate. This is the most accurate method when disclosure is good, because you’re working with the company’s own numbers.
The catch: many companies bundle everything into vague language about “investing in our future.” When disclosure is thin, look at capital spending during years when revenue was flat. If a company spent $40 million on CapEx while revenue didn’t grow, that spending was almost certainly maintenance. Averaging several flat-revenue years gives a useful baseline.
The PPE-to-Sales Ratio
This method, associated with Columbia professor Bruce Greenwald, uses the historical relationship between a company’s asset base and its revenue to separate maintenance from growth. The steps:
- Sum the company’s net PP&E from its balance sheet over the past five years.
- Sum total revenue over the same five years.
- Divide PP&E by revenue to get the average asset-to-sales ratio.
- Multiply that ratio by the year-over-year change in revenue. The result estimates growth CapEx, the spending needed to support incremental revenue.
- Subtract growth CapEx from total CapEx. What remains is maintenance CapEx.
This approach is useful for companies that don’t disclose growth spending and where depreciation is a poor proxy, especially asset-heavy businesses with long-lived assets. Its weakness is the assumption that the historical asset-to-sales relationship will hold, which can fail during major technological change or business-model shifts.
Industry Benchmarks as a Sanity Check
Maintenance spending varies enormously by industry. Research across decades of financial data shows the median company spends roughly four cents on maintenance-related capacity costs for every dollar of sales, but the range runs from about two cents (wholesale, apparel) to over twenty cents (precious metals, petroleum).
Representative ranges for maintenance-related capacity costs as a share of revenue:
- High end, 15% to 21%: precious metals, petroleum and natural gas, telecommunications infrastructure.
- Middle, 5% to 10%: utilities, entertainment, pharmaceuticals, computers, electronic equipment.
- Low end, 2% to 4%: retail, wholesale, apparel, food products, construction.
Use these as a sanity check on your estimate. If your figure for a steel company lands at 1% of revenue, something is wrong with your inputs. If your figure for a software company lands at 18%, you’re probably sweeping growth spending into the maintenance bucket. Benchmarks won’t give you the answer; they tell you when your answer doesn’t make sense.
Reading the Age of the Asset Base
One of the most revealing signals about upcoming maintenance needs is the average age of a company’s fixed assets. Divide accumulated depreciation by the current year’s depreciation expense. A result of 6.0 means the average asset has been in service for about six years. Track this figure over time. If it’s climbing steadily, the company is aging its asset base, spending less on replacement than assets are wearing out. A spending spike is coming when deferred replacements can no longer be postponed.
Compare average age to the useful life assumptions in the company’s accounting policies. If the company depreciates machinery over 10 years and the average age is 8.5, a wave of replacements is near. Companies in this position often face a hard choice: spend heavily soon and depress reported earnings, or keep deferring and risk breakdowns, safety issues, or obsolescence.
This is where costly analytical mistakes happen. A company with an aging asset base and low CapEx looks cheap on an earnings basis. Its profits are high because it isn’t spending. But those profits are borrowed from the future. Buying in at that moment, right before a spending surge, is the classic trap.
Watching for Classification Games
Companies can shift spending between the “maintenance” and “growth” labels, and between the income statement and the balance sheet, without outright fraud. Being aggressive about what counts as an “improvement” pushes costs off the income statement, inflating current earnings at the cost of higher future depreciation. It’s a judgment call, but the effect on reported profitability is real.
The extreme version is WorldCom, which in 2001 and 2002 reclassified roughly $3.8 billion in ordinary operating expenses as capital assets on its balance sheet.4U.S. Securities and Exchange Commission. Complaint: SEC v. WorldCom, Inc. Operating expenses dropped, pre-tax income rose by the same amount, and total assets increased. The statements looked far healthier than the business.
The practical defense is comparison. Check your maintenance CapEx estimate against depreciation and against industry benchmarks. If reported capital spending consistently runs well below depreciation while asset performance doesn’t deteriorate, the company may be under-investing. If CapEx is high but management labels most of it “growth” with little going to maintenance, the classification may be optimistic.
Two Boundaries Worth Knowing
Software and cloud spending complicate the analysis. Under GAAP, internal costs spent maintaining existing software must be expensed as incurred; only new development or significant upgrades can be capitalized, and only after management commits funding and the project is probable to complete and function. Companies that can’t separate maintenance from minor upgrades on a reasonably cost-effective basis must expense everything.5Financial Accounting Standards Board. Accounting Standards Update No. 2025-06: Intangibles – Goodwill and Other – Internal-Use Software (Subtopic 350-40) For software-heavy businesses, real maintenance spending runs through the income statement as salary expense rather than showing up on the CapEx line, so the cash flow statement understates true maintenance costs.
International comparisons carry another wrinkle. Under IFRS, PP&E with components of different useful lives must be depreciated component by component, so a building’s HVAC system gets its own schedule and its overhaul is capitalized as a replacement of the old component. Under U.S. GAAP, component depreciation is allowed but not required, and many companies use composite depreciation across the whole asset. Component replacements blend into the overall depreciation pool and are harder to spot. Comparing a U.S. company on composite depreciation to a European competitor on component accounting requires adjusting for that difference before the CapEx numbers mean the same thing.
Putting the Estimate Together
The most reliable approach runs all three methods and compares. Calculate the depreciation proxy, the CapEx-minus-growth figure, and the PPE-to-sales result independently. If they converge, your estimate is probably in the right range. If they diverge sharply, investigate why. The answer usually lies in asset age, recent acquisitions, or unusual accounting choices.
Cross-check the result against industry benchmarks and the company’s own history. A maintenance figure that looks reasonable in isolation may be alarming after you notice it’s declined for five straight years while the asset base aged. Conversely, a spike after years of underinvestment can be a healthy sign, the company catching up on deferred work rather than deteriorating further. The numbers tell the full story only when read alongside asset age trends and management’s account of where the money is going.