Data Centre Capacity Planning Models
Executive Summary
Key Takeaways
- ✓ Data centre capacity is jointly constrained by power, floor space, and cooling capability, and a model should track all three, since the binding constraint can shift as tenant rack density changes over time.
- ✓ Power, expressed as critical IT load in megawatts, is increasingly the binding constraint in modern facilities as rack density rises, even where floor space and cooling capacity remain nominally available.
- ✓ Phased capacity delivery should be scheduled against a demand forecast and a construction/power interconnection lead time, not built to a single target completion date that ignores intermediate demand.
- ✓ A capacity model that tracks only one constraint risks overstating achievable revenue if a different constraint actually binds first, a structural error common in facilities under-designed for rising rack density.
Objective¶
This guide sets out how to model data centre capacity planning within Data Centre Financial Modelling, tracking power, space, and cooling as co-binding constraints.
Three Co-Binding Constraints¶
A data centre's actual sellable capacity is the minimum of three jointly binding constraints:
Power. The electrical supply and distribution capacity available to IT equipment, expressed as critical IT load in megawatts or kilowatts. This is the capacity unit that most directly drives revenue in modern facilities.
Floor space. The physical area available for racks and the aisles, structural loading, and access clearances required around them.
Cooling. The capacity to remove the heat generated by the IT load, which scales with power draw and is increasingly the constraint that determines achievable rack density.
A capacity model should track all three explicitly, since the binding constraint can shift over the facility's life as tenant rack density rises: floor space may remain nominally available even after power or cooling capacity is fully committed.
Power as the Increasingly Binding Constraint¶
Rising tenant rack density, driven by higher-performance computing equipment, increases power draw and heat output per rack without a proportional increase in the floor space required. As a result, power and the associated cooling capacity are exhausted before floor space in a growing share of modern facilities. A capacity model that tracks only floor space as the sellable unit risks materially overstating achievable revenue where power or cooling in fact binds first.
Phased Capacity Delivery Scheduling¶
Capacity should be delivered in tranches scheduled against a demand forecast and the facility's actual construction and, where applicable, power utility interconnection lead time, not built entirely to a single target completion date. Over-building ahead of contracted or forecast demand ties up capital in unsellable capacity; under-building against demand caps achievable revenue growth. See Hyperscale Data Centre Models for the tenant-specific version of this scheduling discipline under a contracted offtake agreement.
Portfolio Versus Single-Tenant Capacity Planning¶
Colocation capacity planning typically manages a portfolio of many smaller, granular capacity increments sold across a diversified tenant base, allowing incremental capacity releases matched to observed leasing velocity. Hyperscale capacity planning manages a smaller number of large, contractually defined capacity tranches delivered against a single tenant's offtake schedule, where the delivery date itself is frequently a contractual obligation. See Colocation Financial Models.
Common Construction Pitfalls¶
Floor-space-only capacity tracking. Overstates achievable revenue where power or cooling actually binds first.
Capacity delivery scheduled to a single completion date. Ignores intermediate demand and either constrains near-term growth or over-builds ahead of contracted leasing.
No explicit cooling capacity constraint. Understates the capacity ceiling as rack density rises, since cooling capability, not just electrical supply, ultimately limits achievable density.
Recommended Practices¶
- Track power, floor space, and cooling capacity as three explicit, jointly binding constraints.
- Express the primary capacity unit as critical IT load (MW/kW), with floor space and cooling as secondary checks.
- Schedule phased capacity delivery against demand forecast and actual construction/interconnection lead time.
- Reassess the binding constraint periodically as tenant rack density assumptions evolve.
Continue Reading¶
Related Pillars¶
Related Technical Guides¶
Related Glossary¶
How OXXON tests thisRun a free structural check with FMAE
Frequently Asked Questions
What are the three co-binding constraints on data centre capacity?
Power (the electrical supply and distribution capacity available to IT equipment), floor space (the physical area available for racks and aisles), and cooling (the capacity to remove the heat generated by the IT load). A facility's actual sellable capacity is the minimum of what all three constraints allow.
Why is power increasingly the binding constraint?
Because rising tenant rack density, driven by higher-performance computing equipment, increases the power draw and heat output per rack without a proportional increase in the floor space or cooling capacity required, meaning power (and the cooling capacity needed to support it) is exhausted before floor space is.
How should phased capacity delivery be scheduled in the model?
Against a demand forecast and the facility's actual construction and power interconnection lead time, so capacity is delivered in tranches that anticipate demand without either constraining growth or over-building ahead of contracted or forecast leasing.
What happens if a model tracks only floor space as the capacity constraint?
It risks overstating achievable revenue if power or cooling capacity actually binds first, since leasable floor space may remain nominally available even after the facility's power or cooling capacity is fully committed, a structural error common in facilities under-designed for rising rack density.
How does capacity planning differ between colocation and hyperscale facilities?
Colocation capacity planning typically manages a portfolio of many smaller, granular capacity increments across a diversified tenant base; hyperscale capacity planning manages a smaller number of large, contractually defined capacity tranches delivered against a single tenant's offtake schedule.
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.
Critical IT Load
Critical IT load is the amount of power a data centre facility delivers directly to IT equipment, servers, storage, and networking, and is the industry-standard unit for expressing a facility's billable and sellable capacity. It excludes the additional, non-IT power drawn by cooling and power distribution overhead, which is instead captured separately through power usage effectiveness (PUE). Critical IT load, in kW or MW, is the capacity figure that data centre revenue, capacity planning, and portfolio scale metrics are all built around.
Rack Density
Rack density measures how much power a tenant's rack draws, expressed in kW per rack, and by extension how much heat must be removed by the facility's cooling system to support it. Rising rack density, driven by higher-performance computing equipment, is the primary reason power and cooling capacity, rather than floor space, increasingly bind data centre capacity before floor space is exhausted, and density tier is a primary basis for colocation pricing.
Colocation Financial Models
Colocation financial models project revenue from a diversified base of tenants leasing space and power in defined units, per rack or per kW of committed capacity, rather than a single anchor contract. This guide sets out how colocation revenue is decomposed into space/power revenue, cross-connect and ancillary fees, and how occupancy, pricing, and churn assumptions should be modelled as separable drivers rather than a single blended revenue-per-tenant figure.
Hyperscale Data Centre Models
Hyperscale data centre models finance a facility developed and leased to a single large cloud or technology tenant under a long-dated contract, structured around phased, capacity-denominated capex drawdown rather than a single completion event. This guide sets out how to model phased delivery, contracted revenue recognition, and the concentrated counterparty and power availability risks distinctive to this business model.