Key Terms in a Data Center Lease

The key terms in a data center lease are power capacity and the rent per kilowatt, the uptime service level agreement and its credits, the term and escalations, take-or-pay and ramp commitments, and the rights to expand. A data center lease is priced on reserved power rather than square feet, so the committed kilowatts and the rate per kW per month are the first numbers to read, and the SLA exhibit is where the operational risk actually lives. Miss any one of these and you can misjudge what the deal really costs.

Whether you are a tenant signing a colocation agreement or a team abstracting a portfolio of data center leases, these are the clauses that decide the outcome. Here is what each one governs and why it belongs in every abstract.

Power capacity and $/kW rent

This is the heart of the lease. Data center space is sold by reserved power, quoted as committed kilowatts, and rent is charged per kW per month, most often as a modified gross rate that bundles power distribution, cooling, physical security, and redundancy into one number. In primary US markets, hyperscale power has been quoted in the range of roughly $100 to $150 per kW per month, though your lease sets the actual figure. Read whether power is reserved (you pay for the capacity whether or not you draw it) or metered (you pay for usage), and capture the overage or burst rate, because that is where an unexpected bill comes from.

Uptime SLA and service credits

The service level agreement sets the availability the provider guarantees, and the gap between tiers is larger than it sounds. A 99.9% SLA allows about 8.7 hours of downtime a year. A 99.99% SLA allows about 52 minutes. A 99.999% SLA, the so-called five nines, allows only about 5.26 minutes. Just as important as the number is the service credit: what you actually receive when the target is missed, and whether it is a meaningful offset against rent or a token gesture. The SLA and its credit formula usually sit in a separate exhibit, not the body of the lease, which is exactly why they get overlooked. Read the measurement method too, because how downtime is defined and measured decides whether a credit is ever triggered.

Term, escalations, and critical dates

Retail colocation runs on short terms of one to three years, wholesale on five to ten, and hyperscale build-to-suit deals on ten to fifteen. Longer terms carry annual escalations, so read the escalation rate and whether it is fixed or indexed. Then pull every critical date into a schedule: the commencement date, the notice windows for renewal and termination, and any delivery milestones on a build-to-suit. On a long data center commitment, a missed renewal notice or an unnoticed escalation is an expensive surprise, which is why critical date extraction is part of abstracting these leases properly.

Take-or-pay and ramp commitments

On wholesale and hyperscale deals, the lease often includes a take-or-pay floor: a minimum you pay for reserved capacity whether or not you use it. Paired with it is a ramp schedule that sets when you start paying for capacity as it comes online. Together they fix the floor on your cost, and they can commit you to paying for power well before your workloads need it. In 2026, with grid capacity constrained in major markets, developers have gained leverage and these commitments have tightened, so read the take-or-pay and ramp terms carefully and put them in the abstract as dated obligations.

Expansion rights and power availability

If you expect to grow, the value of a data center lease is partly in the option to take more capacity later. Look for rights of first refusal or first offer on adjacent space and power, expansion options with defined pricing, and any power availability contingency that ties your expansion to the grid actually delivering. In a constrained market, the right to expand is worth little if the power is not available, so read how the two are linked. These are the terms that most often get negotiated hard and most often get lost in a quick read of the document.

Interconnection and cross-connects

Interconnection is the ecosystem of network connections inside the facility, and cross-connects are the physical links to carriers and other tenants. For network-heavy workloads, a dense interconnection environment is a real advantage and can carry its own charges worth capturing. For AI, GPU, and hyperscale workloads, which are increasingly power-bound rather than network-bound, interconnection is more of a tiebreaker than a primary driver. Either way, read the cross-connect terms and any recurring fees, because they add to the true cost of the footprint.

Cooling, redundancy, and operating obligations

The lease should specify the cooling and temperature commitments, the redundancy configuration (how the provider keeps power and cooling running through a failure), and who is responsible for what operationally. These are usually framed as provider obligations that back up the SLA, so read them together with the availability guarantee. A high uptime number means little if the redundancy behind it is thin. Beyond the lease itself, keeping a clear view of the total spend on a growing data center footprint is its own discipline, and teams often pair the lease terms with a way to keep infrastructure and cloud spend under control as they scale.

Putting the terms into a usable abstract

Reading these clauses once is not enough; the value comes from pulling them into structured fields you can compare across deals and track over time. That is what data center lease abstraction does: it extracts the reserved power and $/kW rent, the uptime SLA and credit formula, the take-or-pay and ramp terms, the term and escalations, and the expansion rights into clean fields, each citing its source page. You can test it free on one lease before trusting it on a portfolio, using the lease abstraction software, and check the power floor and credit schedule against the exhibit in minutes rather than rereading the whole agreement. Last updated July 2026.