Almost every discussion of colocation pricing treats the rate as a cost-plus number: the provider pays for power, adds overhead and a margin, and that is your bill. It is a reasonable mental model and it explains very little of what you actually experience when you negotiate.
It does not explain why a five-year term is priced so differently from a three-year one, why a provider will move further on an escalator than on a headline rate, why a tenant with a strong balance sheet gets terms a startup cannot, or why the same facility will quote one number for a hall taken at once and another for the same capacity taken over eighteen months.
All of those are financing behaviour. The building was funded before you arrived, on assumptions about who would occupy it and for how long, and your contract is the instrument that makes those assumptions true. Understanding that does not get you a discount by itself. It does tell you which requests are expensive for the other side and which ones only look expensive.
A data center is an asset financed against its leases
Large facilities are not usually funded out of operating cash. The capital arrives before the revenue, the build takes years, and the asset only becomes financeable at scale once there is contracted demand against it. The structure that results is closer to commercial real estate than to IT procurement, with one important difference: the power infrastructure is a large share of the cost and it is far less liquid than the building around it.
That has a practical consequence. The lender or investor behind the facility is underwriting a stream of contracted payments. The quality of that stream — how long it runs, how certain it is, who is obliged to pay it — is what determines the cost of the capital, and the cost of the capital is a large component of what you are quoted. A facility funded cheaply can price keenly. One funded expensively cannot, regardless of how efficient its operations are.
Where the money actually goes
It is worth separating the cost stack, because the parts behave very differently and tenants routinely negotiate against the wrong one.
Land and shell. The slowest-moving and most durable part. It has a long life, it is relatively easy to value, and it is the portion a landlord is most comfortable financing over a long horizon.
Power infrastructure. Substation works, transformers, switchgear, generators, UPS and distribution. This is capital-hungry, it is sized years before it is fully used, and much of it is on long manufacturing lead times, which is why equipment lead times show up in commercial conversations rather than only in engineering ones. It is also the part that cannot be deferred: you cannot sell a half-energised hall.
Mechanical plant. Chillers, pumps, air handling or liquid cooling distribution. Significant capital, a shorter life than the shell, and increasingly specific to the density the facility is designed around.
White space fit-out. Racks, containment, busway, PDUs, cabling. Comparatively fast to deploy and the part most likely to be phased.
The first three are committed long before a tenant signs. The last one can follow demand. That asymmetry is the single most useful thing to understand about why providers behave the way they do, because it means the expensive commitment has already been made and the facility is carrying it whether your racks arrive or not.
Why the provider wants term, and what it will pay for it
A facility with capacity contracted for seven years and a facility with the same capacity contracted for two are not the same asset, even if this year's revenue is identical. The first supports cheaper, longer-dated capital. The second does not, because the lender has to assume re-letting risk.
This is why term is so often the most productive lever in a negotiation. A provider that will not move another cent on a three-year deal may move meaningfully on five, not out of goodwill but because the longer contract improves the asset it is financed against. The same logic explains why providers resist short terms at any price, and why contract negotiation tends to be more productive when it treats term, ramp and rate as one package rather than three separate arguments.
It also explains the escalator. An annual increase is not only an inflation hedge; it is the mechanism that keeps the contracted revenue stream growing in line with the assumptions used to underwrite the facility. A tenant who cares more about predictability than about the first year's number can often trade a slightly higher starting rate for a lower or capped escalator, and both sides can end up better off.
Your covenant is part of the pricing
The contracted payment stream is only worth what the counterparty is good for. That is why providers ask for financials, why deposits and parent guarantees appear in term sheets, and why two tenants can be quoted differently for the same hall.
This is not personal and it is negotiable in both directions. A tenant who can offer stronger security — a longer term, a parent company guarantee, a larger deposit, prepayment of part of the term — is improving the quality of the provider's receivable and can reasonably expect something in return. A tenant unwilling or unable to do any of that is asking the provider to absorb the risk, which is priced.
Where a deposit is involved, treat it as the capital item it is: agree what it covers, in what form it is held, what reduces it over time, and what returns it. A deposit that is never discussed again until exit tends to be discussed badly at exit.
Ramp is a financing concession, not an admin detail
Most requirements do not land at full load on day one. The gap between the capacity you contract and the capacity you draw is real, and it is the subject of a separate discipline — see the distinction between contracted capacity and actual draw.
Commercially, a ramp schedule means the provider carries committed infrastructure cost against partial revenue for a defined period. That is a financing concession with a measurable value, which is why a long ramp is expensive and why providers would rather give a lower rate at full commitment than a generous ramp at a higher one.
For a tenant, this is worth modelling honestly. A ramp that matches a realistic deployment plan is worth paying for. A ramp bought as insurance against a plan you do not believe is an expensive hedge, and the same money usually buys more as a shorter term with an expansion right attached. Forecasting when the next megawatt is needed is the input that decides which of the two you are actually buying.
Who owns the risk you are being asked to carry
Several clauses that read as legal boilerplate are allocating capital risk, and recognising them as such changes how they are negotiated.
Power cost pass-through. Whether energy is at cost, at cost plus a fee, or bundled decides who carries tariff movement. In this region that is not a trivial exposure, and the structures differ by emirate and by customer class, which our explanation of UAE power tariffs sets out.
Availability remedies. Service credits are a risk-sharing mechanism with a cap, and the cap is usually far below the commercial consequence of an outage. They are worth negotiating for what they signal about operational confidence as much as for their value, which is the lens the SLA guide applies.
Make-good and exit obligations. The cost of returning space to its original condition is a liability that accrues from day one and surfaces at the end. It belongs in the model at signature, not in a surprise invoice years later.
Build, lease or buy as a capital question
Framed operationally, the build-versus-colocate decision is about control, speed and staffing, and that comparison is set out in colocation versus building your own. Framed financially, it is a question about where you want the capital and the residual risk to sit.
Building puts the asset on your balance sheet, gives you the residual value and the control, and requires you to fund infrastructure sized for a demand forecast you are making yourself. Colocation converts that into an operating commitment and transfers the sizing risk, at a price that includes the provider's cost of capital and return. Neither is cheaper in the abstract; they are different trades, and the right answer depends on how confident you are in your own forecast and how expensive your capital is relative to the provider's.
Between the two sits a third option that is often overlooked, where a landlord delivers the shell and the power and the tenant builds out the inside. That model has its own distinct allocation of capital and risk, and it is covered in powered shell and build-to-suit.
What to do with this when you are negotiating
Ask what the provider is optimising for, because it is usually visible. A facility in lease-up behaves differently from one that is nearly full. A provider raising capital against the asset will value term and covenant strength more than one funding from cash flow.
Put your requests in order of what they cost the other side. Term, ramp and expansion rights are capital questions and are often where the value is. Operational commitments — reporting, remote hands, access, escalation paths — cost the provider very little and improve your life considerably, which is why they are usually the easiest wins and the ones most often left on the table.
And model the whole term rather than the first invoice. Escalators, pass-through mechanics, ramp, deposits and exit obligations move the total far more than the headline rate does, which is also why a well-run colocation RFP scores on total cost over the term rather than on the rate card.
The regional picture
In the Gulf, the binding constraint on new capacity is rarely capital appetite and frequently power and programme: when a connection can be energised, what the approvals chain allows, and how long the long-lead electrical equipment takes to arrive. Our walk-through of the approvals chain to energisation and of interconnection queues and time to power covers those sequences.
For a tenant, the consequence is simple enough. Capacity that is already energised, commissioned and operating is a fundamentally different product from capacity that is funded and promised, and it is priced accordingly. When a quote looks unusually keen for a date that is some way out, the question worth asking is which of those two things is being sold.
Frequently Asked Questions
Why does a longer contract term usually get a better rate?
Because the facility is financed against its contracted revenue. A longer term produces a more certain, longer-dated payment stream, which supports cheaper capital for the owner. The discount is not goodwill; it reflects a genuinely better asset from the lender's point of view, which is why term is often the most productive thing to negotiate.
Is colocation always cheaper than building?
No, and neither is reliably cheaper in the abstract. Building keeps the asset, the residual value and the control, and requires you to fund infrastructure sized against your own demand forecast. Colocation converts that into an operating commitment and transfers the sizing risk at a price that includes the provider's cost of capital. Which is better depends on the confidence in your forecast and the cost of your own capital.
Why do providers ask for financial statements or a guarantee?
Because the value of a contract depends on the counterparty's ability to pay it. Security such as a parent guarantee, a larger deposit or a longer term improves the quality of the provider's receivable, and a tenant offering it can reasonably ask for something in return. A tenant offering none is asking the provider to carry that risk, which is priced into the rate.
What is a ramp, and why is a long one expensive?
A ramp phases the capacity you pay for up to the capacity you contracted, matching a deployment plan. It is expensive because the provider has already committed capital to infrastructure that sits underused during the ramp. A ramp matched to a realistic plan is worth paying for; one bought as insurance against a plan you do not believe is usually a costly hedge.
Which parts of a colocation deal are easiest to negotiate?
Operational terms. Reporting, remote hands scope, access procedures and escalation paths cost the provider relatively little and materially improve the tenant's experience, so they move more easily than the rate. Capital-linked items such as term, ramp and expansion rights move too, but they are priced against the financing, so concessions there are traded rather than given.
Comparing Capacity Options?
Rax Data & Energy can set out what a requirement looks like as hosting, as a lease and as a build, with the commercial trade-offs stated plainly.
Contact Rax Data & Energy