What a data center contract should include

The paper you're signing#

It's usually two documents. A master agreement covers the relationship, and an order says what you're buying. Find out which one holds the SLA and which holds the price, because providers split them differently.

Anything missing from the list below is a gap to raise before you sign. After you sign, it's the provider's call.

Term, renewal, and what it costs to leave#

• How long. Racks, cages and suites usually run 12 to 36 months. Whole halls run 5 to 15 years. Sign for what you really mean to commit.
• How it renews. Auto-renewal is close to universal. Get the notice period in writing, and cap the renewal price.
• What it costs to leave. Early exit and end-of-term removal are usually two clauses. Get both.

The annual increase: get it in writing#

The escalator is the yearly price increase. Almost every contract has one, and nobody publishes it. Get it as a clear percentage, on a clear base, on a set schedule.

It compounds, and it's negotiable. Most buyers treat it as fixed. See how to negotiate data center space.

Connection fees: make them itemize#

Cross-connects are the cables from your gear to carriers and clouds. They're the line most often left vague. Get each piece listed on its own:

1. A monthly fee per connection.
2. An install fee.
3. A removal fee, if there is one. The contract should say either way.
4. Whether you pay at one end or both. It changes the real cost a lot.

Using virtual connections too? They're priced separately, by port and speed. Cover them in the contract as well.

On-site support: pin down the rates#

Remote hands is basic on-site help, like a reboot. Smart hands is skilled work, like installs, and costs more. The contract should say where the line sits.

• The hourly rate, and the minimum you're billed. Two different terms.
• After-hours rates, which often differ.
• Prepaid blocks of hours, usually cheaper than hourly.

What the SLA needs to say#

The credit schedule is what matters. A big uptime promise with a tiny credit protects the provider, not you. Make sure the SLA covers:

• What counts as an outage, and over what time window.
• What they pay when they miss, in writing, scaled to how bad it was.
• How you claim it, and how long you have.
• Whether every certification covers this building, not just the company. Here's how to check.

How Clectiv AI solves this#

Doing this on your own means months of chasing reps, one at a time. Every call starts from scratch, and every provider knows they're the only one in the room.

With Clectiv AI, it's days, not months. Start your project here, before you call a provider, and Clectiv AI runs the process: it goes to every provider that fits, at the same time. They know they're competing, so their best numbers come back side by side, and you choose.

Start a project

You start where others finished, with 20+ years of buyer-side projects and our own data on 2,200+ buildings behind your request.

The checklist, in one place#

• Term length that matches what you mean to commit.
• Renewal notice period, and a cap on the renewal price.
• Early-exit cost, and end-of-term removal cost.
• The annual increase, as a clear percentage.
• Connection fees: monthly, install, removal, and one end or both.
• Virtual connection pricing, if you use it.
• On-site support rates, minimums, after-hours terms and prepaid hours.
• SLA outage definition, credits, and how to claim.
• Every certification scoped to your building.
• Which document holds the SLA, and which holds the price.

This is the structure, not the numbers. The numbers are in how to run a data center bid.

Clectiv AI runs this list for you. The providers that fit bid on your project and compete, and the Clectiv team reads every proposal against these lines before you choose.

Start a project →

Common questions#