In the cloud era, capex versus opex describes a structural change in how infrastructure hits the books. Capital expenditure, capex, is money spent to acquire an asset, a server, that sits on the balance sheet and is depreciated over years. Operating expenditure, opex, is the ongoing cost of running the business, expensed in the period it is incurred. Cloud converts most infrastructure from the first to the second: instead of a large upfront purchase you depreciate slowly, you pay a variable monthly bill that flows straight through the income statement. This is usually framed as a benefit, no upfront outlay, pay for what you use, but it removes the natural ceiling that a hardware purchase imposed, and that is the part finance has to manage deliberately.
This explainer sits under our CFO's guide to cloud cost management, the pillar for this cluster, and informs the Lock step of our See, Cut, Lock, Run method, where budgets and guardrails replace the spending ceiling that owned hardware used to provide. It pairs with the sibling article how to budget for cloud in a growing company, which addresses the variable-spend problem this shift creates.
A hardware purchase was a one-time decision with a hard cap. Cloud opex is a continuous decision with no cap unless you build one. The financial discipline that used to be enforced by the purchasing process now has to be designed in.
What actually changes on the financials
Under the old model, a data center build was a capital project: a large upfront spend, capitalized and depreciated, with the cost spread across the useful life of the equipment. Cloud flips this. The same compute capacity now arrives as a recurring operating expense that scales up and down with usage, hits the income statement immediately, and produces no depreciating asset. For a CFO this has three consequences: spend becomes more variable and harder to forecast, it moves from below the EBITDA line to inside it, and the timing of cost recognition shifts from gradual depreciation to immediate expense. The result is a more honest, more responsive cost signal, and a more volatile one.
Where commitments blur the line
The clean capex-to-opex story gets complicated by cloud commitments. When you buy a multi-year reserved instance or commit to a spend level for a discount, you are making something that looks economically like a capital decision, a large, upfront, multi-year obligation, inside an opex framework. Accounting treatment depends on the structure, and prepaid commitments in particular need to be amortized over the term rather than expensed at purchase. This is where the two worlds meet, and where finance teams most often get the accounting wrong. The detail lives in the sibling article cloud commitment accounting and amortization.
Cloud opex running ahead of plan?
Our Managed FinOps service puts the budgets, anomaly alerts and commitment strategy in place that an opex model needs to stay controlled. Independent and buyer-side, we are paid to lower the bill, not to grow it.
Talk to us about Managed FinOps →Capex vs opex at a glance
| Dimension | Capex (owned hardware) | Opex (cloud consumption) |
|---|---|---|
| Timing | Large upfront purchase | Ongoing as-you-use billing |
| Financials | Balance sheet, depreciated | Income statement, expensed now |
| Spending ceiling | Set by the purchase decision | None unless you build budgets |
| Flexibility | Fixed once bought | Scales with demand |
| Forecasting | Predictable depreciation | Variable, requires modeling |
What CFOs should do about it
The opex model is not a problem to reverse, it is a trade to manage. The variability that makes cloud spend hard to forecast is the same variability that lets you match cost to demand and avoid paying for idle capacity. The job is to keep the upside, elasticity and no stranded hardware, while replacing the discipline the purchasing process used to provide. In practice that means budgets with alerts so spend cannot drift unnoticed, a deliberate commitment strategy so you capture discounts without over-committing, and a forecast that treats cloud as a variable cost driven by business activity rather than a flat line. Those controls are the Lock and Run of a mature program, and they are what let a CFO embrace opex without losing the plot on the bill.
The CFO's Cloud Cost Playbook covers the capex-to-opex transition, the accounting treatment of commitments, and the budget controls that keep an opex model in line.
Accounting treatment varies by jurisdiction, standard and contract structure as of May 2026, and is not advice on your specific situation. Confirm the treatment of cloud spend and commitments with your auditors and against current accounting standards before relying on it.
The short version
Capex versus opex in the cloud era is the shift from owned, depreciated hardware to rented, expensed consumption: more flexible and more honest, but without the spending ceiling a purchase used to impose. Commitments blur the line and need careful amortization. The CFO's job is to keep the elasticity while rebuilding discipline through budgets, commitment strategy and variable forecasting. Putting those controls in place is the core of our Managed FinOps service.