
Authors
Stating that data center construction has exploded over the last few years is about as profound as stating that the sky is blue. Hyperscalers have asserted themselves with bold investments at massive scale, and market dynamics have been dramatically reshaped. These trends have attracted significant capital to the sector, with REITs and institutional investors increasingly treating data center assets as a prime target for investment.
Data centers are highly specialized assets that require equally specialized lease agreements. A data center lease is, at its core, a hybrid instrument: part real estate agreement, part infrastructure services contract. The lease allocates not only the use of physical space, but also the risks associated with power delivery, environmental controls, connectivity, and operational continuity. Unlike standard commercial leases, where the landlord’s primary obligation is to provide habitable space, a data center landlord is expected to deliver a functional, resilient environment capable of supporting mission-critical operations. Understanding the key distinctions between data center and standard commercial leases will allow market participants to effectively underwrite transactions in all parts of the capital stack.
Power is the backbone of any data center, making it one of the most important elements of the lease. In a typical office or retail lease, the premises are defined by rentable square footage and rent is calculated based on square footage. In a data center lease, by contrast, the premises may be defined by the number of servers, and power allocation is often of equal or greater importance than the physical space.
For data center leases, rent is often priced per kilowatt of committed power, or as a combination of space and power charges. Typically negotiated key terms include committed power capacity (the total amount of electricity that the landlord guarantees to deliver), power density (how much power is available per server housed at the data center), redundancy levels (the amount of backup power built into the facility), and utility cost allocation (who pays for electricity and how those costs are calculated). Leases will also address whether power is delivered on a metered or fixed-rate basis, and may include performance benchmarks tied to how efficiently a landlord is able to deliver power to a facility.
Servers housed in data centers generate significant heat: if that heat is not properly removed, equipment will fail within minutes. As such, data center leases will typically specify a facility’s cooling capacity (usually measured in tons of refrigeration or kilowatts), as well as redundancy requirements for backup cooling. Parties will also negotiate the landlord’s obligation to maintain temperature and humidity within defined parameters to ensure the tenant’s equipment can operate in ideal conditions. Specific performance thresholds and remedies for failures are typically set out in service level agreements.
Because cooling can account for close to half of a data center’s total energy consumption, cost allocation is a significant negotiation point. Tenants will want clarity on whether cooling costs are bundled into rent, passed through as a separate charge, or metered directly. Standard commercial leases typically pass through operating expenses as common area maintenance charges, but in data center leases, power consumption is the dominant variable cost, making them a significant point of emphasis.
For a standard commercial property, landlords will provide basic HVAC, water, and electrical service and will be insulated from liability if those services are interrupted (usually up to a certain number of consecutive days in which the tenant is unable to use the space). A data center landlord, by contrast, is expected to provide and maintain complex power systems, including uninterruptible power supplies and backup generators to ensure a tenant’s operations are not interrupted by power outages.
Data centers are often classified by “Tier” levels (Tier I through Tier IV) under a recognized industry standard. Tier I facilities have no redundancy: if a cooling unit or power supply fails, the facility goes down until it is repaired. Tier IV facilities are fully fault-tolerant, meaning every critical system has at least one complete backup. Each lease should clearly state the facility’s tier classification (or equivalent specifications) and describe the redundancy built into power, cooling, and network systems. Higher tier levels provide greater protection against downtime, but typically come with higher costs.
SLAs set out binding, measurable standards for facility performance that go far beyond the general habitability obligations found in standard commercial leases. SLAs may be embedded directly in the lease, attached as an exhibit or schedule, or in some cases structured as a standalone agreement running alongside the lease.
Regardless of form, these typically include: (i) uptime guarantees; (ii) temperature and humidity ranges; (iii) network connectivity and latency standards; and (iv) physical security requirements.
The SLA will also specify consequences if the landlord fails to meet its commitments. Common remedies include rent credits or, in special circumstances, the right to terminate the lease. Often, a landlord will provide a parent guarantee for the payment of rent credits due to SLA breaches.
Termination rights in data center leases extend beyond the typical triggers of landlord default or casualty/condemnation found in standard commercial leases. Tenants often negotiate additional termination rights tied to repeated SLA failures, prolonged outages, the landlord’s failure to maintain required certifications, or landlord’s transfer to an entity without appropriate operational experience (a significant distinction from commercial leases where there is typically no restriction on landlord’s ability to transfer its interest in the property).
Approaching a data center lease like any other commercial lease would be like looking at it with one eye closed. Infrastructure-related provisions inherent to the asset class make negotiating these leases a hybrid affair. Market participants should understand these nuances so they can effectively allocate risk during the negotiation.
To discuss these issues, please contact the author(s).
This publication is a general discussion of certain legal and related developments and should not be relied upon as legal advice. If you require legal advice, we would be pleased to discuss the issues in this publication with you, in the context of your particular circumstances.
For permission to republish this or any other publication, contact Bryn Turnbull.
© 2026 by Torys LLP. All rights reserved.