Colocation is usually discussed in terms of power and price per kilowatt. Those are the numbers that end up in the model, but they are not the first decision. The first decision is what shape of space you are buying, because that choice constrains everything that follows: how you secure equipment, how you run cabling, how much power density you can deploy, and how easily you grow.
There are three common forms. A single cabinet in a shared hall. A caged area enclosing several cabinets. A private suite with its own walls and door. They are not simply small, medium and large versions of the same product, and the step between each carries different trade-offs.
The single cabinet
A cabinet, sometimes called a rack, is the entry point. You rent a standard enclosure, typically forty-two to forty-eight rack units, in a hall shared with other tenants. The provider supplies power to the cabinet and cooling to the room.
What you get is straightforward and inexpensive. You pay for one enclosure, the power feeds to it, and whatever cross connects you need. The provider manages everything outside the cabinet door. For a deployment of a few servers, a network edge presence or a disaster recovery footprint, this is almost always the right answer.
The limits appear in three places. Power density is capped by what the hall can cool, and in a shared hall that ceiling is set for the average tenant, not for you. Security stops at the cabinet lock, so your equipment sits in a room other people walk through. And expansion depends on whether adjacent cabinets happen to be free, which is a matter of luck rather than planning.
Cabinets are usually sold either by the unit with a power allowance, or by power with the cabinet included. The second is generally more honest for high-density equipment, because the constraint is almost never the physical space. A cabinet rated for a modest load will be half empty if you fill it with GPU servers, a point covered in our guide to power budgeting for GPU colocation.
Caged space
A cage is a demarcated area within a shared hall, enclosed by mesh partitioning with its own locked access. Inside it you may have anywhere from a handful of cabinets to several rows.
The gain is control. Everything inside the cage is yours, so you can arrange cabinets as you like, run your own overhead cabling, install your own containment and bring your own enclosures if you prefer them. Access control becomes yours too: entry to the cage is a separate authorization from entry to the hall, which matters for compliance regimes that require demonstrable physical segregation.
The cost is that you are now paying for floor area as well as power, and cage space is priced on the footprint whether or not you fill it. A half-empty cage is money sitting idle. You also take on responsibilities that were previously the provider's: cable management inside the cage, airflow discipline between your rows, and in some contracts the hot and cold aisle arrangement.
That last point deserves attention. A cage sits inside the provider's cooling design, and a tenant who arranges cabinets badly can disrupt airflow for themselves and sometimes for neighbors. Containment practice inside a cage is a shared-consequence matter, not a private one.
The private suite
A suite is a physically separate room, with solid walls, its own door and frequently its own cooling and power distribution. It is the closest colocation gets to having your own data center without building one.
The advantages are real for the right tenant. Physical isolation is complete, which simplifies several compliance conversations. Power density can be engineered for your load rather than inherited from a shared hall, which is the decisive factor for dense AI and HPC deployments. You can often specify your own cooling approach, including liquid cooling, which a shared hall may not accommodate. And you control your own maintenance windows inside the suite.
The costs scale accordingly. You are buying a room, and you pay for it whether it is full or not. Lead times are longer because suites are frequently built to order. Contract terms are longer, since the provider is committing capital to fit out space specifically for you. And more of the operational burden sits with you.
Suites are also where the distinction between the provider's responsibility and yours needs to be written down precisely. Our guide to white space and grey space covers that boundary, and it matters more here than in any other arrangement.
Choosing between them
A few thresholds tend to decide it in practice.
- Below roughly five cabinets, individual cabinets are almost always more economical. You are not yet paying enough in power to justify floor area.
- Where compliance requires demonstrable physical segregation, a cage is frequently the minimum acceptable arrangement, and the requirement can override the economics entirely.
- Where power density exceeds what the shared hall supports, the conversation moves to a suite regardless of size, because the constraint is cooling rather than space.
- Where you need liquid cooling, a suite is usually the only option, since the plumbing and heat rejection cannot be retrofitted into a shared hall for one tenant.
- Where growth is uncertain, cabinets preserve flexibility. A cage or suite sized for a plan that does not materialize is an expensive commitment.
What each tier means for cooling
Space tier and thermal design are more tightly coupled than the sales conversation usually admits, and this is where most capacity disappointments originate.
In a shared hall, you inherit the room's cooling design. The provider has engineered airflow for an assumed average density across all tenants, and your cabinets sit inside that assumption. If you concentrate heat beyond it, the consequence is not a polite warning; it is elevated inlet temperatures for you and potentially for the cabinets around you. Containment helps, and our guide to hot aisle and cold aisle containment covers the principles, but containment within a shared hall is bounded by what the room can ultimately reject.
A cage gives you control of arrangement without changing the thermal envelope. You can organize your own hot and cold aisles and add containment within your boundary, which genuinely helps, but the air still comes from and returns to the provider's system at the provider's capacity.
A suite is the only tier where the cooling approach itself becomes negotiable. Rear-door heat exchangers, in-row cooling and direct liquid cooling all require infrastructure decisions that a provider will make for a dedicated room and generally will not make for one tenant in a shared hall.
The practical test when evaluating a space tier is not "what is the circuit rating" but "what is the sustained thermal ceiling per cabinet, and what happens if I exceed it". A provider who answers the second question precisely is describing a design they understand.
The questions that get missed
Whichever form you choose, several points are worth settling before signing rather than afterward.
How is the space actually measured? Cage pricing by footprint should specify whether the measurement includes the aisle space you need to work in. A cage you cannot open a cabinet door inside is not usable at its nominal area.
What is the real power ceiling per cabinet? Not the circuit rating, the thermal ceiling. A cabinet with ample electrical capacity in a hall that cannot remove the heat is a cabinet you cannot fill.
Can you expand contiguously? Growth into cabinets on the other side of the hall is operationally worse than growth next door. Ask whether adjacent space can be reserved and what that reservation costs.
Who may enter, and is it logged? In a shared hall, provider staff and other tenants' engineers pass your equipment routinely. Access logging and camera coverage matter more as you move down the tiers, not less, and our guide to physical security and access control covers what to ask for.
What happens at renewal? Space commitments are harder to exit than power commitments because the provider cannot easily resell a fitted suite. Understand the exit terms before you need them, as covered in our guide to colocation exit strategy and workload migration.
Moving between tiers
Most tenants do not choose once. They start with cabinets, outgrow them, and face the step up at an inconvenient moment.
The migration from cabinets to a cage inside the same facility is usually the easiest, because the provider can often build the cage around existing equipment or move it in stages over several maintenance windows. Moving to a suite is a larger exercise and frequently amounts to a full redeployment.
The practical advice is to have the conversation a tier early. Ask a provider at contract time what the path to a cage looks like, what notice it requires and what it costs, even if you have no intention of using it yet. A provider who can answer precisely is a provider who has done it before.
Rax Data and Energy offers all three arrangements across our UAE facilities, including suites engineered for high-density AI and GPU deployments where shared-hall cooling is the binding constraint. If you are trying to work out which tier your deployment actually needs, get in touch and we will size it against your real power and growth profile rather than a standard package.
Frequently Asked Questions
What is the difference between a cabinet, a cage and a private suite?
A cabinet is a single enclosure in a shared hall. A cage is a demarcated area enclosed by mesh partitioning with its own locked access, holding multiple cabinets. A private suite is a physically separate room with solid walls, its own door and frequently its own cooling and power distribution. They differ in cost structure, control, physical security and achievable power density.
When does a cage become more economical than individual cabinets?
Usually somewhere above roughly five cabinets, though the exact point depends on how space and power are priced. Below that, individual cabinets are generally cheaper because you are not yet consuming enough power to justify paying for floor area. Compliance requirements for physical segregation can make a cage necessary well before the economics favor it.
Do I need a private suite for high-density AI or GPU workloads?
Often yes, but the reason is cooling rather than space. A shared hall is designed around an average power density, and a dense GPU deployment can exceed what that hall can remove as heat. A suite allows cooling to be engineered for your specific load, and it is usually the only practical route to liquid cooling, which cannot easily be retrofitted into a shared hall for one tenant.
Can I move from cabinets to a cage later?
Usually, and it is the easiest of the tier migrations. A provider can often build a cage around existing equipment or relocate it in stages across several maintenance windows. Ask about the path, the notice period and the cost at contract time rather than when you need it, because availability of contiguous space is the limiting factor.
A Real Energisation Date, Not a Brochure
Rax operates where transmission investment has run ahead of load, and we are direct about staged capacity and real dates. Talk to our team about sizing a deployment against a timeline.
Contact Rax Data & Energy