Data Centre Operating Cost Models
Executive Summary
Key Takeaways
- ✓ Data centre operating cost should be decomposed into power, cooling, maintenance, staffing, security, and insurance as distinct categories, each with its own driver, rather than a single blended operating cost percentage of revenue.
- ✓ Power and cooling cost together typically represent the largest share of data centre operating cost and should be driven by the facility's actual load and PUE, not a flat cost-per-square-foot or cost-per-revenue assumption.
- ✓ Staffing cost in a data centre is driven primarily by facility headcount for operations and security coverage, largely independent of occupancy level, distinguishing it from a revenue-scaling cost assumption.
- ✓ A blended operating cost percentage of revenue conceals which specific cost category is actually responsible for a margin change, since power, maintenance, and staffing respond to genuinely different drivers.
Objective¶
This guide sets out how to model the full data centre operating cost structure within Data Centre Financial Modelling, decomposed into distinct, activity-linked categories.
Operating Cost Decomposition¶
Power. Electricity cost for both IT load and non-IT overhead, driven by actual load factor and PUE. See Data Centre Power Consumption Models.
Cooling. A major driver of the non-IT power overhead captured in PUE, plus cooling-plant-specific maintenance cost. See Data Centre Cooling Cost Models.
Maintenance. Critical infrastructure servicing and, separately, capital renewal, driven by component life cycles rather than revenue. See Data Centre Maintenance Models.
Staffing. Facility operations and security personnel, driven primarily by required headcount coverage rather than occupancy level.
Security. Physical security systems, monitoring, and access control, largely a fixed facility cost independent of occupancy.
Insurance. Property and liability coverage appropriate to a high-value, high-criticality asset, typically scaling with asset replacement value rather than revenue.
Power and Cooling as the Largest Cost Category¶
Power and cooling together typically represent the largest share of data centre operating cost, and should be driven by the facility's actual load factor and PUE rather than a flat cost-per-square-foot or cost-per-revenue assumption, since actual power cost depends directly on tenant utilisation and facility efficiency, both of which vary over the facility's life.
Staffing as a Largely Occupancy-Independent Cost¶
Data centre staffing cost is driven primarily by the headcount required for facility operations and security coverage, which is largely independent of occupancy level: a facility requires broadly similar operations and security staffing whether it is 60% or 95% occupied. This distinguishes staffing from a cost category that scales directly with revenue, and the model should reflect staffing as a largely fixed facility cost rather than a variable, occupancy-linked one.
Avoiding the Blended Operating Cost Percentage¶
A single blended operating cost percentage of revenue conceals which specific category, power, maintenance, staffing, or another, is actually responsible for a margin change, since each responds to a genuinely different underlying driver. Decomposing operating cost into its distinct categories allows a reviewer to diagnose the actual source of a cost or margin variance rather than observing only its net effect.
Common Construction Pitfalls¶
Single blended operating cost percentage of revenue. Conceals which specific cost category is driving a margin change.
Power and cooling cost modelled on a flat cost-per-square-foot basis. Disconnects the largest cost category from its actual load and efficiency drivers.
Staffing cost scaled directly with occupancy or revenue. Misrepresents a cost category that is largely fixed, driven by required operational coverage rather than occupancy level.
Recommended Practices¶
- Decompose operating cost into power, cooling, maintenance, staffing, security, and insurance as distinct lines.
- Drive power and cooling cost from actual load factor and PUE, not a flat per-square-foot assumption.
- Model staffing as a largely fixed, occupancy-independent cost.
- Diagnose margin variance against the specific cost category responsible, not a single blended percentage.
Continue Reading¶
Related Pillars¶
Related Technical Guides¶
How OXXON tests thisRun a free structural check with FMAE
Frequently Asked Questions
What are the main data centre operating cost categories?
Power (electricity cost for both IT load and non-IT overhead), cooling (a component of power cost plus cooling-specific maintenance), maintenance (critical infrastructure servicing and capital renewal), staffing (facility operations and security personnel), security (physical security systems and monitoring), and insurance (property and liability coverage for a high-value, high-criticality asset).
Why should power and cooling cost be driven by load and PUE rather than a flat assumption?
Because actual power cost depends on the facility's actual load factor and its PUE-driven non-IT overhead, both of which vary by tenant utilisation and facility efficiency, and a flat cost-per-square-foot or cost-per-revenue assumption disconnects the largest cost category from its actual physical driver.
How does staffing cost behave differently from a revenue-scaling cost assumption?
Data centre staffing cost is driven primarily by the headcount required for facility operations and security coverage, which is largely independent of occupancy level, a facility requires broadly similar operations and security staffing whether it is 60% or 95% occupied, unlike a cost category that scales directly with revenue.
What is the risk of using a single blended operating cost percentage of revenue?
It conceals which specific cost category, power, maintenance, staffing, or another, is actually responsible for a margin change, since each category responds to a genuinely different underlying driver, and a blended assumption prevents a reviewer from diagnosing the actual source of a cost or margin variance.
Related Articles
Data Centre Financial Modelling
Data centre financial modelling is the discipline of modelling a data centre operator's revenue, cost, and capital structure from its capacity-denominated drivers, power, space, and cooling capacity, rack density, and tenant contract structure, rather than the generic market-price and headcount-growth drivers used in most corporate models, or the pure occupancy-and-lease-term drivers of conventional commercial real estate. This page is the hub for the Knowledge Centre's data centre financial modelling content: how colocation, hyperscale, and enterprise business models each require a distinct model architecture, how rack revenue and occupancy are decomposed into their separable underlying drivers, and how capacity planning and financial KPIs tie the model together, as this domain expands to cover operations, revenue, investment, and governance practice across the sector.
Data Centre Power Consumption Models
Power consumption modelling forecasts a data centre's actual electricity draw and cost, distinct from the capacity planning discipline that governs how much power can be sold. This guide sets out how to model load factor, utility tariff structure (demand charges versus consumption charges), and the distinction between contracted and actual power draw, since power is typically the largest single operating cost category in this sector.
Data Centre Cooling Cost Models
Cooling cost is one of the largest non-IT power draws in a data centre and the primary driver of power usage effectiveness (PUE). This guide sets out how to model cooling cost as a function of cooling technology choice, climate and free cooling opportunity, and rising rack density, and why cooling cost should be modelled explicitly rather than absorbed into a single blended power cost assumption.
Data Centre Maintenance Models
Data centre maintenance covers the planned preventive maintenance of critical infrastructure, power distribution, cooling plant, and backup generation and battery systems, whose failure directly risks the facility's service level commitments. This guide sets out how to model maintenance cost as a distinct opex category driven by component life cycles, and how to distinguish routine maintenance from the capital renewal that eventually replaces, rather than merely services, ageing infrastructure.
Enterprise Data Centre Models
Enterprise, or captive, data centres are facilities an organisation builds and operates for its own internal IT use rather than leasing to external tenants. This guide sets out how to model this business model as an internal cost centre with chargeback to business units, and how to structure the build-versus- colocate-versus-cloud capital allocation decision that increasingly frames enterprise data centre investment.