What this calculator is designed to answer
The calculator estimates a screening-level development budget rather than pretending that every data center costs the same amount per MW.
Start with a construction benchmark, adjust it for the project and market, then expose the major costs that can sit outside the headline construction number.
Why the default is $11.3 million per MW
JLL's 2026 Global Data Center Outlook forecasts average global data center construction cost of $11.3 million per MW.
JLL explicitly says this benchmark covers shell and core construction. Technology fit-out is separate and can be dramatically more expensive for AI infrastructure.
The research sources use different project and cost boundaries. Always normalize scope before comparing benchmarks.
Why Cushman & Wakefield reports $17.6 million per MW
Cushman & Wakefield's September 2026 Data Center Development Cost Guide reports average greenfield costs of $17.6 million per MW for the most modern U.S. and Canadian facilities, excluding chips and GPUs.
Its report also says development costs have increased approximately 21% per MW since its previous Q4 2024 edition.
That benchmark describes a broader modern greenfield development context than JLL's shell-and-core figure.
Do not double-count land and interconnection
This calculator exposes land and utility/interconnection as separate inputs because those items can vary enormously between sites.
If the $/MW benchmark you enter already includes those costs, set the corresponding additional fields to zero rather than adding them twice.
Market adjustment is intentionally transparent
Construction cost varies with labor availability, local wages, material pricing, permitting conditions, logistics and contractor capacity.
The market adjustment applies directly to the initial construction component, making the assumption visible instead of hiding it inside a proprietary market multiplier.
How the high-density adjustment works
The high-density field is a screening adjustment to the construction infrastructure cost.
It can represent additional electrical distribution, cooling, structural or integration requirements associated with dense deployments.
It does not automatically include GPUs, servers or complete AI technology fit-out.
JLL notes that AI technology fit-out can reach as much as $25 million per MW separately from shell and core. That is why this calculator avoids silently folding server hardware into the construction estimate.
Utility interconnection deserves its own budget line
Two sites with identical buildings can have very different total costs if one requires major utility or transmission upgrades.
Interconnection can include utility studies, dedicated substations, transformers, network upgrades and other infrastructure needed to turn a power request into usable capacity.
Use the separate utility interconnection cost guide when estimating that line item.
Why contingency is calculated after the visible cost stack
The model applies contingency to adjusted construction plus the three additional project-cost inputs.
This gives you a transparent contingency dollar amount and prevents it from disappearing inside the headline $/MW assumption.
If your internal estimate already includes contingency in individual line items, reduce or zero the calculator's contingency field.
How the low and high scenarios work
The scenario-variance field changes the adjusted construction component while keeping the entered land, utility and soft-cost assumptions fixed.
With a ±10% setting, the low scenario uses 90% of adjusted construction and the high scenario uses 110%.
Contingency is then recalculated on each scenario.
This is deliberately simple. It is intended to show sensitivity, not simulate a probabilistic cost-risk model.
What is excluded unless you add it explicitly?
Depending on your cost benchmark and project structure, the model may exclude financing, taxes, tenant technology, servers, GPUs, development fees, incentives, off-site transmission upgrades or other owner-specific costs.
The value of the calculator is therefore not the default answer. It is the ability to replace every assumption with the scope of the project you are actually underwriting.
Use effective $/MW only after the scope is complete
The effective project cost per MW divides total modeled CAPEX by the entered project capacity.
It is useful for comparing complete scenarios only when both projects use the same capacity definition and the same cost boundary.
For more context, see Data Center Cost per MW and Data Center Construction Cost per Square Foot .
Sources: JLL, 2026 Global Data Center Outlook; Cushman & Wakefield, 2026 Data Center Development Cost Guide. JLL forecasts $11.3M/MW for global shell-and-core construction in 2026 and notes that AI technology fit-out can reach $25M/MW. Cushman & Wakefield reports an average $17.6M/MW greenfield development cost for modern U.S. and Canadian facilities, excluding chips and GPUs. Benchmarks are references, not quotes for a specific project.