Data Center REIT Power Capacity Metrics

The power capacity of a data center—measured in megawatts—is the binding constraint in a data center REIT. While floor space and cooling systems matter, available power determines how many servers can be installed, how much revenue the facility generates, and whether the data center REIT power capacity metric will constrain future growth.

Why power is the binding constraint

A data center REIT’s floor space is nearly infinite relative to power. A typical building might have 100,000 square feet available, but only 20 megawatts of electrical capacity delivered by the local utility. Modern servers consume 5–20 kilowatts each. That 20 MW facility can host roughly 1,000 to 4,000 servers—and the bottleneck is not the lease area, but the power available to run them.

Unlike square footage, which is fixed at construction, power capacity cannot be instantly expanded. Adding new circuits requires upgrading the utility connection, which takes months or years and may be impossible if the local grid is already strained. This inelasticity makes power the true constraint on a data center REIT’s revenue growth and valuation. A facility with excess floor space but saturated power capacity is revenue-capped.

Megawatt capacity and utilization

Data center REITs report both total installed capacity and deployed capacity. A facility might have 25 MW of electrical supply but only 18 MW deployed to customer cabinets, leaving 7 MW reserved for expansion, redundancy, and emergency systems. Analysts focus on deployed power because it drives revenue; available power is the expansion runway.

Deployed power directly scales revenue. A hyperscaler paying $200 per kilowatt per year for rack space and power in a major hub brings in $200,000 annually per MW deployed. Saturating a facility’s power capacity means capturing all available revenue; undersaturated capacity is lost income. This is why REIT management closely tracks power utilization rates and guides on megawatts available for sale in coming quarters.

Power Usage Effectiveness (PUE)

The Power Usage Effectiveness (PUE) ratio measures how efficiently a data center converts purchased electricity into computing power. It’s calculated as total facility power ÷ IT equipment power. A PUE of 1.1 means 10% of power goes to cooling, lighting, and infrastructure; 90% feeds servers. A PUE of 1.5 means 50% is overhead—inefficient and costly.

Leading-edge data center REITs target PUE ratios of 1.1 to 1.25. This efficiency is achieved through liquid cooling, hot-aisle containment, free-air cooling in cold climates, and AI-driven thermal management. Older facilities run PUE of 1.4 or higher. Because power is expensive and power costs are passed partly to customers via pricing, a REIT’s PUE directly affects both margins and competitiveness. A gap in PUE versus peers can justify a valuation discount.

Geographic power supply constraints

Data center location is power-constrained by geography. Regions with abundant hydroelectric capacity (Pacific Northwest, Iceland) or abundant natural gas (Texas) attract more data centers because power is cheaper and grid capacity exists. Dense urban areas (Northern Virginia, Dublin, Singapore) have tight power grids; new data center power allocations require grid upgrades or conversions from other industries—expensive and slow.

Geopolitical factors shape power supply too. Countries with renewable mandates or carbon taxes increase power costs. Conversely, areas with subsidized power (coal-rich regions in some countries) can be attractive but carry reputational and regulatory risk. A REIT exposed to cost increases in its key markets sees margin compression; one with long-term power supply agreements is insulated.

Assessing power constraints in underwriting

When evaluating a data center REIT, analysts examine:

1. Deployed vs. available power: A facility with 25 MW deployed and 2 MW available has limited expansion room; one with 15 MW deployed and 10 MW available is still growing.
2. Utility constraints: Can the local grid supply more power? Is a new utility connection feasible, and at what cost?
3. PUE efficiency: Is the facility below or above industry median? Is there a roadmap to improve it?
4. Customer power draw: Hyperscalers’ AI workloads demand denser, higher-power deployments. A REIT’s facility must support 10+ kW per square foot; older builds may top out at 5 kW/sq ft.
5. Power pricing stability: Long-term power contracts insulate margins; spot-rate exposure is risky.

A facility with saturated power, rising PUE, and no utility upgrade path is a growth anchor. One with ample headroom, efficient cooling, and grid expansion options can expand for years.

Power and REIT valuation multiples

Data center REITs trade on power capacity and utilization more than on other REIT metrics. A company reporting 30 MW deployed and 8 MW available (expansion potential) at a 1.15 PUE, with 95% utilization and signed customer commitments, commands premium valuation multiples. One with 20 MW deployed, 1 MW available, and a 1.35 PUE trades at a discount.

The power constraint explains why data center REIT valuations diverged sharply during the AI boom. Facilities with plentiful available power and efficient cooling captured all incremental hyperscaler demand at premium prices; power-constrained peers missed revenue growth and faced pricing pressure. Power capacity became the limiting factor on earnings growth, and earnings growth drives REIT valuation.

See also