When comparing AI data center sites, ask what the advertised megawatt figure measures. It could describe utility service at the property boundary, generation output, a future allocation, or capacity designed to reach the IT equipment. Give every figure a measurement point, an operating condition and a delivery date before using it in your shortlist.

For this comparison, facility MW means the electrical load serving the whole facility. IT MW means the portion serving servers, storage and network equipment. Cooling and other building systems also need capacity within the site’s electrical envelope.

Define the capacity you need to deploy

Start with a written requirement for initial IT load and expansion. Include the intended hardware, rack arrangement and operating profile. Ask the electrical and mechanical engineers to show how that requirement translates into facility demand under the proposed design conditions.

Uptime Institute’s capacity-allocation guidance identifies cooling demand, electrical losses and auxiliary loads as factors to account for alongside IT capacity. It also explains that an operating-efficiency metric alone cannot establish how much provisioned capacity a design makes available to IT. Uptime Institute’s discussion of capacity allocation

Request a capacity schedule for each operating condition

Ask the engineers to prepare a schedule showing the utility or generation limit, facility-system loads and IT load. Have them evaluate normal operation, the intended maintenance condition and the equipment-outage conditions covered by the design. Request the drawings, equipment ratings and assumptions behind the schedule.

In an existing facility, ask for the latest load measurements and confirmation of which equipment is commissioned. In a development project, identify the infrastructure that must be installed before each stage of IT load can operate. Track any limits between the service entrance and the rack, so the comparison reflects the full delivery path.

A hypothetical 20 MW example

Suppose a project has a confirmed 20 MW facility limit. Its engineer allocates 3.5 MW to cooling, electrical losses and other facility systems at the stated design condition. That leaves an illustrative 16.5 MW for IT within that allocation.

Now suppose a downstream electrical path is demonstrated to support only 15 MW of IT under the required maintenance condition. The buyer should use the documented 15 MW limit for that scenario until the design resolves it. The earlier 16.5 MW arithmetic remains an allocation calculation, while the operating limit comes from the engineering review. These assumptions are hypothetical and describe no Ace Miners listing.

Use PUE for the energy question it answers

Power usage effectiveness, or PUE, compares total facility energy with IT equipment energy over a defined period. DOE describes annual PUE using total annual facility energy divided by annual IT energy. For example, an assumed 100,000 MWh facility total and 80,000 MWh IT total produce an annual PUE of 1.25. DOE’s explanation of PUE

Ask for the period, measurement boundary and operating assumptions behind a quoted PUE. Then request a separate engineering capacity schedule for the peak conditions that determine how much IT the project can support.

Compare phases with the same definitions

For every shortlisted site, record four items for each phase: facility MW, supported IT MW, the operating condition used to establish that figure, and the expected in-service date. Identify whether the evidence is a utility commitment, design calculation, commissioning result or operating measurement. Assign an owner to every item still awaiting confirmation.

Review that schedule alongside the power availability documents, cooling and water assessment, and electricity cost comparison. To begin your search, browse the Ace Miners powered-land directory or share your requirements. Specify both your initial IT load and facility-load assumptions, with the dates for each deployment phase.