Depreciation rarely attracts the same attention as revenue growth, capital expenditure or free cash flow.
It is often treated as a technical accounting charge: the gradual recognition of the cost of an asset over the period in which that asset is expected to generate economic benefits.
But that relatively simple concept is becoming increasingly important to corporate finance.
Across technology, energy, manufacturing, utilities and infrastructure, companies are committing enormous amounts of capital to physical assets. Data centres are being built to support artificial intelligence. Utilities are expanding electricity networks. Semiconductor manufacturers are constructing fabrication plants. Manufacturers are investing in automation, robotics and new production facilities.
Those investments do not disappear from the financial statements after the cash has been spent.
They remain on the balance sheet and eventually flow through the income statement through depreciation.
As capital spending rises, therefore, depreciation can become a much larger expense.
That has implications for reported earnings, margins, return on invested capital and the way investors judge whether large investment programmes are actually creating value.
The next phase of the corporate capital-expenditure cycle may therefore be about more than how much companies spend.
It may increasingly be about how long the assets they buy remain economically useful.
Capital Expenditure Creates Tomorrow's Depreciation
When a company spends money on a long-lived physical asset, it generally does not record the entire cost as an expense immediately.
Instead, the asset is capitalised and its cost is allocated over its estimated useful life.
Under IAS 16 Property, Plant and Equipment, depreciation is the systematic allocation of an asset's depreciable amount over its useful life.
A company that builds a factory expected to operate for 30 years therefore recognises the cost differently from a business buying equipment expected to last five years.
This means today's capital-expenditure decisions can influence reported profits for years or decades.
A large investment cycle can initially appear primarily on the cash-flow statement.
Several years later, however, the accumulated asset base can produce much larger depreciation charges on the income statement.
That lag is becoming increasingly relevant because corporate capital spending is rising dramatically in several important sectors.
AI Is Turning Technology Into a Capital-Intensive Business
Few areas illustrate the shift more clearly than artificial intelligence.
Large technology companies historically operated businesses with exceptionally strong cash generation and comparatively asset-light economics relative to traditional industrial companies.
AI is changing part of that model.
Artificial intelligence requires physical infrastructure.
Servers.
Graphics processors.
Networking equipment.
Cooling systems.
Power infrastructure.
Buildings.
The capital commitment is enormous.
More recent projections point even higher. S&P Global Ratings said in August 2026 that six major hyperscalers could spend more than $870 billion in 2026 and more than $1.3 trillion in 2027.
This spending is usually discussed as an AI investment story.
It is also becoming a depreciation story.
The Cash Leaves Before the Expense Arrives
Capital expenditure and depreciation affect financial statements at different times.
When a company builds a data centre, most of the cash leaves during construction and equipment installation.
But the income-statement expense is recognised over the useful lives of the assets.
That creates an important timing difference.
A company can experience very large cash outflows today while only gradually recognising the accounting expense associated with those investments.
This is one reason investors increasingly need to look at several measures together.
Operating profit may remain strong.
EBITDA may appear stronger still because depreciation is added back.
Yet free cash flow can deteriorate sharply if capital spending is rising rapidly.
Alphabet provides a recent illustration.
According to S&P Global Market Intelligence's analysis of Alphabet's second-quarter 2026 results, the company increased capital expenditure substantially as it expanded data centres, processors and networking capacity. Free cash flow fell sharply even as cloud revenue continued to grow.
That does not necessarily indicate weak economics.
But it demonstrates why reported earnings alone may not capture the full cost of a major infrastructure cycle.
Depreciation Can Become a Margin Issue
Once large amounts of new infrastructure enter service, depreciation expenses can rise.
That matters because depreciation reduces operating profit.
Suppose a company spends $100 billion building infrastructure.
If the assets are depreciated over ten years on a simplified straight-line basis, the eventual annual depreciation charge associated with that investment could be approximately $10 billion before considering residual values, different asset classes or timing.
Multiply that across several consecutive years of extraordinary capital expenditure and the effect can become substantial.
This can create a new pressure on margins.
Revenue may continue growing.
Demand may remain strong.
But operating expenses can also rise as the depreciation associated with previous investment programmes begins flowing through the income statement.
Investors therefore need to distinguish between today's capex boom and tomorrow's depreciation burden.
Asset Lives Suddenly Matter More
The accounting treatment depends heavily on a deceptively simple assumption:
How long will the asset remain useful?
For a traditional building, that estimate may be relatively straightforward.
For advanced computing infrastructure, it can be much more complicated.
Buildings may last for decades.
Electrical infrastructure may also have long operating lives.
Servers, processors and networking equipment may become economically obsolete much faster.
This distinction becomes crucial when companies are investing heavily in technologies evolving at extraordinary speed.
A graphics processor may continue functioning physically for years.
But if new hardware offers dramatically better performance per unit of power, the economic usefulness of older equipment can decline much faster than its physical condition suggests.
IAS 16 specifically requires companies to consider factors including technical and commercial obsolescence when determining useful lives. The standard also requires residual values and useful lives to be reviewed at least at each financial year-end. The IFRS Foundation's guidance makes clear that an asset's useful life reflects its expected utility to the company and can be shorter than its broader economic life.
That principle could become increasingly important in rapidly changing technology infrastructure.
Technological Obsolescence Is Becoming a Finance Risk
Historically, depreciation was often closely associated with physical wear.
Machines wore out.
Vehicles accumulated mileage.
Buildings deteriorated.
Technology introduces another form of depreciation: obsolescence.
An asset can still function perfectly and yet become economically unattractive.
A five-year-old server may continue operating.
But if newer equipment performs substantially more computation using less electricity, keeping the old server may no longer make commercial sense.
The company then faces an economic question.
Was the originally estimated useful life too long?
If so, future depreciation may need to reflect revised expectations.
This is one reason the AI investment boom is creating unusual corporate-finance questions.
Companies are committing capital to an infrastructure class whose technology can evolve much faster than the buildings housing it.
Depreciation Is Not the Same as Cash Flow
Another reason depreciation deserves more attention is that it is a non-cash charge.
The cash was generally spent earlier when the asset was purchased.
This is why EBITDA excludes depreciation.
For many businesses, EBITDA remains an important operating measure.
But the exclusion can sometimes make heavily capital-intensive businesses appear less capital-intensive than they really are.
A data centre requires servers to operate.
A utility requires grid equipment.
A manufacturer requires machinery.
Those assets eventually need to be replaced.
Depreciation therefore represents more than an accounting convention.
Over long periods, it provides a rough indication that productive assets are being consumed.
This does not mean depreciation equals replacement capital expenditure.
Actual replacement costs can be higher or lower.
Technology can become cheaper.
Inflation can make assets more expensive.
Companies may replace assets earlier or later than expected.
But ignoring depreciation entirely can obscure the underlying economics of asset-heavy businesses.
EBITDA Can Look Better Than Economic Reality
Consider two companies reporting the same EBITDA.
One requires minimal physical investment.
The other must replace expensive equipment every few years.
Their economic characteristics are very different.
The first company may convert a large portion of EBITDA into free cash flow.
The second may need to reinvest significant amounts simply to maintain existing capacity.
This distinction becomes especially important when comparing software companies with digital-infrastructure businesses.
A cloud provider may generate technology-like revenue but increasingly operate an infrastructure-heavy asset base.
That shift should influence how investors analyse financial performance.
High EBITDA may be less informative if maintaining revenue requires continuously rising capital expenditure.
Free Cash Flow Is Becoming the Counterweight
This is why free cash flow is becoming increasingly important alongside depreciation.
Free cash flow broadly measures the cash remaining after operating requirements and capital expenditure.
A company can report strong earnings while free cash flow weakens because it is spending aggressively on infrastructure.
That does not automatically mean the investment is poor.
The company may be building capacity that generates significant future revenue.
But the returns need to materialise.
S&P Global's August 2026 AI monitor noted that investor attention is increasingly shifting from infrastructure spending itself toward evidence of monetisation, productivity gains and sustainable earnings growth. The analysis estimates aggregate 2026 and 2027 capex expectations across major AI-related companies at more than $1.5 trillion.
Eventually, investors will need to compare the cash generated by those assets with both their original cost and their depreciation.
Return on Invested Capital Becomes More Important
The key corporate-finance question is not whether a company can afford to invest.
It is whether the investment earns an adequate return.
Return on invested capital, or ROIC, becomes especially useful during large capital cycles.
A company can increase revenue simply by spending enormous amounts of money.
But if each additional dollar of capital generates progressively less operating profit, shareholder value may not increase proportionally.
High capital expenditure can therefore coexist with falling capital efficiency.
S&P Global's recent analysis of AI-related spending has already highlighted concerns around return on invested capital. Its August 2026 AI Monitor noted market concerns around whether large infrastructure investments will produce sufficient future returns.
Depreciation sits directly within this calculation because it represents the periodic consumption of the capital base used to generate revenue.
The Data-Centre Build-Out Makes the Issue Hard to Ignore
The scale of projected data-centre investment makes this more than an accounting curiosity.
That includes enormous spending on buildings, servers, energy and supporting infrastructure.
Each asset category has a different useful life.
The building may remain useful for decades.
A transformer could remain in service for many years.
Servers may turn over much faster.
Cooling equipment follows another replacement schedule.
A single data centre therefore contains numerous layers of depreciation.
Understanding the economics requires looking beyond the headline construction cost and asking how quickly each component is consumed.
Utilities Face a Similar Capital Cycle
Technology companies are not alone.
Utilities are entering another enormous investment period.
Rising electricity demand, grid modernisation and new generation requirements are driving large capital programmes.
Utilities have always been depreciation-intensive businesses.
Power stations, transmission lines, substations and distribution equipment operate for long periods and are depreciated over their estimated useful lives.
But a new investment wave enlarges the asset base.
That means larger future depreciation charges.
For regulated utilities, some costs may ultimately be reflected in customer rates depending on the regulatory framework.
For shareholders, however, the central question remains whether the capital invested earns an acceptable return.
Again, capital expenditure and depreciation are two sides of the same story.
Manufacturing Has Its Own Depreciation Cycle
The same logic applies to manufacturing.
A company automating a factory may invest in robotics, sensors, production equipment and digital control systems.
Those assets can reduce labour costs or increase output.
But they also increase the depreciable asset base.
If productivity gains exceed the cost of the equipment, margins can improve.
If expected savings fail to appear, the company may end up carrying higher depreciation without sufficient corresponding economic benefit.
This is particularly important when companies automate primarily because competitors are doing so.
Technology investment can become strategically necessary without automatically generating attractive financial returns.
Management therefore needs to evaluate not only whether a new asset improves operations, but also whether the improvement justifies the capital consumed.
Replacement Capex Matters as Much as Growth Capex
Another distinction is becoming more important: growth capital expenditure versus maintenance or replacement capital expenditure.
Growth capex creates additional capacity.
Replacement capex maintains existing capacity.
The two have very different implications.
A company investing $5 billion in new factories may grow production.
A company spending the same amount simply replacing worn-out machinery may generate no additional output.
Both amounts appear as capital expenditure.
Depreciation can help indicate how much of the existing asset base is being consumed over time.
Again, it is not a perfect proxy.
But where annual capital expenditure consistently exceeds depreciation, the company may be expanding its productive asset base.
Where capital expenditure roughly matches depreciation over long periods, investment may be more heavily associated with maintenance and replacement.
This can be especially useful when comparing mature industrial businesses.
Inflation Complicates the Comparison
The relationship between depreciation and replacement spending becomes more complicated during inflation.
Depreciation is based largely on historical asset costs.
Replacement occurs at current prices.
Consider machinery purchased ten years ago for $10 million.
Its depreciation may be based on that historical cost.
But replacing the same productive capability today may cost $15 million.
Accounting depreciation therefore may understate the cash eventually required to maintain the asset base.
This can cause apparently strong cash generation to look less attractive when replacement spending becomes necessary.
For capital-intensive businesses, the difference between historical depreciation and current replacement cost can become strategically significant.
Depreciation Policy Can Affect Comparability
Useful-life assumptions also make comparisons between companies more difficult.
Two companies can purchase similar assets but depreciate them over different estimated periods.
The company using the longer useful life will generally recognise a smaller annual depreciation charge initially.
Its reported operating profit may therefore appear higher.
But that does not necessarily mean its underlying economics are better.
The difference may simply reflect accounting estimates.
This makes useful-life disclosures more important when asset bases grow rapidly.
Investors need to understand whether companies are using broadly comparable assumptions and whether those assumptions remain credible as technology evolves.
The IFRS Foundation requires companies applying IAS 16 to review useful lives and residual values regularly, precisely because those estimates can change as circumstances evolve.
Shorter Asset Lives Can Accelerate Expenses
Suppose a company initially expects a technology asset to remain useful for six years.
Three years later, rapid innovation suggests it will become obsolete after four.
That change matters.
The remaining carrying value must effectively be allocated over a shorter remaining period.
Depreciation can therefore accelerate.
For businesses spending heavily on rapidly evolving infrastructure, useful-life assumptions can become increasingly consequential.
This is one reason technological innovation creates a potential paradox.
New technology encourages more capital investment.
But rapid technological progress can also shorten the economic lives of the assets being purchased.
Companies may therefore need to invest more frequently.
Impairment Is a Separate Risk
Depreciation should also be distinguished from impairment.
Depreciation allocates an asset's cost systematically over its useful life.
Impairment arises when an asset's carrying value may no longer be recoverable.
Under IAS 36 Impairment of Assets, an impairment loss arises when an asset's carrying amount exceeds its recoverable amount.
This could become relevant if companies build capacity based on demand expectations that later prove too optimistic.
A data centre that cannot attract customers may be worth less than expected.
A manufacturing plant built for a technology that loses market share could face a similar problem.
The larger the capital-investment boom, the more significant this downside risk becomes.
Debt Financing Raises the Stakes
The depreciation story also intersects with corporate leverage.
Companies funding infrastructure primarily from internal cash have one risk profile.
Those relying heavily on borrowing have another.
S&P Global has noted that data-centre expansion is increasingly being supported by debt and more complex financing structures. Its 2026 digital infrastructure outlook highlighted rising leverage and an increasing share of speculative-grade project and corporate data-centre ratings.
Debt introduces fixed obligations.
Interest must be paid even if asset utilisation disappoints.
Depreciation itself does not consume cash, but the original asset may have been financed with debt that does.
This makes the return generated by the infrastructure even more important.
Investors May Need to Relearn Asset Analysis
For much of the digital economy, investors became accustomed to businesses whose most valuable assets were intangible.
Software.
Brands.
Networks.
Customer relationships.
AI infrastructure is reintroducing a much older analytical discipline.
How much did the asset cost?
How long will it last?
How heavily will it be utilised?
How quickly will it become obsolete?
What maintenance will it require?
What return will it generate?
What will it cost to replace?
These questions have always been central to industrial finance.
They are increasingly becoming central to technology finance as well.
Depreciation Could Become a Competitive Issue
Companies with more efficient infrastructure may eventually have an advantage.
If one operator can generate the same revenue using fewer servers, it may require less capital expenditure.
Its depreciation burden may also be lower.
If another company repeatedly needs to replace expensive hardware to maintain performance, its capital intensity may remain higher.
This means technological efficiency can translate directly into financial efficiency.
Better hardware utilisation, longer asset lives and higher revenue per unit of infrastructure can all influence returns on capital.
Depreciation therefore provides one lens through which those differences eventually appear.
The Focus Is Moving From Spending to Returns
During the early stages of an investment boom, markets often reward the decision to build.
Companies announce factories.
Data centres are commissioned.
Infrastructure budgets rise.
Suppliers receive orders.
The difficult questions come later.
Did utilisation reach expectations?
Did revenue increase?
Did margins improve?
Did the assets last as long as management expected?
Did free cash flow recover?
Did return on invested capital justify the original spending?
As S&P Global's 2026 analysis suggests, the AI investment cycle is already moving toward this more demanding stage, with greater emphasis on monetisation and sustainable returns. (S&P Global)
That transition makes depreciation much more relevant.
The Capital Spending Story Does Not End When the Asset Is Built
Companies are entering one of the largest infrastructure-investment periods in years.
Technology companies are building data centres.
Utilities are upgrading grids.
Manufacturers are automating factories.
Semiconductor businesses are expanding production.
The headlines understandably focus on the amount being spent.
But corporate finance ultimately cares about what happens afterwards.
Capital spending creates assets.
Assets create depreciation.
Depreciation affects earnings.
Replacement affects free cash flow.
And the revenue produced by those assets determines whether the entire process creates value.
That is why depreciation is becoming a bigger corporate-finance story.
It is not because accounting rules have suddenly become more important.
It is because companies are putting extraordinary amounts of capital into assets whose cost, useful life, utilisation and replacement requirements will shape financial performance for years to come.
In the next phase of the investment cycle, the most important question may therefore not be how much companies are willing to spend.
It may be whether the assets they are buying can earn sufficient returns before they need to be replaced.
References
S&P Global — Credit Risks of the AI Data-Centre Infrastructure Investment Boom
S&P Global — AI Growth Accelerates as Alphabet Spending Weighs on Cash Flow
S&P Global — More Power, Less Cushion: AI and Data-Centre Credit Risk
McKinsey & Company — The Capex Crucible: What Finance Teams Can Learn From Data Center Builds