Data center GMP contract risk has one dominant source. The building is a shell around a mechanical and electrical plant, the plant is specified by an end user, and the end user is frequently still deciding when the price has to be fixed.
A building that is mostly not the building
On a commercial or residential project the structure and envelope are most of the cost. On a data center they are a minority of it.
The majority sits in electrical distribution, uninterruptible power, generators, switchgear, cooling plant, containment and controls. The concrete box around all of that is comparatively simple and comparatively cheap.
That inversion changes what a guaranteed maximum price is protecting. On a residential building the price covers work that is well understood and largely drawn. On a data center it covers equipment with long lead times, specified to a performance requirement, in quantities that depend on a capacity decision.
The capacity decision is usually made by somebody who is not a party to the construction contract. It is made by the end user, or by a leasing negotiation, or by a customer of the operator who has not signed yet.
So the price is fixed against a design whose central input is outside the room.
Everything that follows is a consequence of that one structural fact, and none of it is anybody fault.
Where data center GMP contract risk concentrates
Capacity. Megawatts of critical load and the redundancy level behind it. This single variable drives generator count, switchgear, uninterruptible power sizing, cooling tonnage, fuel storage and often the structure carrying it. It is the largest lever on the job and the one most likely to move after signature.
Long lead equipment. Generators, switchgear and chillers carry lead times measured in many months and occasionally beyond a year. Their release dates frequently fall before the amendment is signed, which is a finding rather than a risk.
Commissioning scope. Levels of commissioning, integrated systems testing and the witness requirements of the end user. Frequently described by reference to a standard and priced by assumption.
Utility coordination. Service capacity, substation works and the utility timeline, none of which the contractor controls and all of which gate energization.
Fit out phasing. Whether the building is delivered as a shell with a first phase of capacity or fully fitted, and what the price includes at each stage.
Four of those five are readable from the exhibits before signature. The fifth, capacity, is readable as an assumption, which is almost as useful.
The assumption to find first
Every data center price carries a capacity assumption, and it is usually one line on the qualifications page.
It will say something like priced for a critical load of a stated number of megawatts at a stated redundancy, with a stated number of generators and a stated chiller configuration.
That line is the most consequential sentence in the document, because if the end user requirement lands above it, the change is not incremental. Capacity moves in blocks: another generator, another switchgear lineup, another chiller, and frequently the structural and electrical infrastructure to support them.
The useful exercise before signature is to price the next block up. Not because you expect it, but because knowing what one step of capacity costs converts a vague risk into a number you can carry, negotiate or cap.
It also changes the conversation with whoever is making the capacity decision, because a leasing team that knows the next block costs a specific amount decides differently from one that does not.
That is the same discipline as reading any other assumption, described in what a guaranteed maximum price actually guarantees.
A worked example
Illustrative figures. Not taken from any client project and not a quotation.
A $118 million data center priced at 12 megawatts of critical load with a stated generator and chiller configuration, carried on the qualifications page as an assumption.
In month five the end user requirement settles at 15 megawatts.
What moves: two additional generators with fuel storage and enclosures, $3.1 million. Additional switchgear and distribution, $1.9 million. Additional cooling capacity, $1.2 million. Structural and electrical room modifications, $600,000.
Total $6.8 million, or 5.8 percent of the contract, from one line on a qualifications page.
The schedule effect is larger in practice. The generators have a fourteen month lead time and the original order was released in month two, so the additional units arrive after the target energization date and the project either accepts a delay or pays an expediting premium on top.
Nothing was concealed. The assumption was disclosed, the pricing was fair, and the risk sat in plain sight in a sentence nobody priced.
What can be done about it before signature
Three things, and all three are commercial rather than technical.
Price the capacity steps. Ask for the cost of the next block up and the block below, as a priced option with a validity period and a release date. This converts an open exposure into a decision with a deadline, which is a much better object to hold.
Release the long lead equipment against the highest credible capacity. On some jobs it is cheaper to order switchgear sized for the larger case than to reorder later, and that comparison is arithmetic rather than judgment once both numbers exist.
Write the capacity decision date into the schedule. Whoever makes that decision is usually not reading the construction schedule, and a date in a document they do see changes when it arrives.
Where the end user is a future customer who genuinely does not exist yet, the honest answer is that the exposure cannot be closed and should be carried as a named line in the budget with a range against it.
That is a worse outcome than closing it and a much better one than discovering it in month five with no figure attached.
There is a fourth move worth considering on larger programs, which is to build the shell for the higher capacity and fit out for the lower one. Structure, electrical rooms, fuel yard and cooling plant space are comparatively cheap to build oversized and extremely expensive to extend.
Whether that is right depends on the probability the capacity increases and on the cost of capital tied up in space nobody is using. It is a real calculation with two known numbers, and it is a better conversation than the one that happens when the requirement moves and the electrical room is the size it is.
What to do before you sign
- Find the capacity assumption on the qualifications page and read it as the largest number in the document.
- Price the next capacity block up, as an option with a validity period and a release date.
- List every long lead item with its release date, and mark the ones already past.
- Establish what level of commissioning is priced, and against which standard.
- Establish the utility timeline and who owns the risk if energization slips.
- Confirm what the price includes at each phase if the building is delivered in stages.
- Write the capacity decision date into the schedule as a milestone with a name against it.
Item two is the one that changes the outcome. A capacity increase priced in advance is a commercial decision; the same increase priced in month five is a negotiation with one bidder who knows exactly how much you need it.
Item three catches the finding that recurs most on this building type. It is common for a data center amendment to be signed with two or three equipment release dates already behind it, and none of it appears as a delay because no activity has started late yet.
How this reading is scoped, and what it produces, sits in the review packages.
What we do
We find the capacity assumption, price the step above it, and build the release date for every long lead item back from its installation activity. The output names what each open item is worth and when it stops being negotiable. It is a commercial reading of the exhibits rather than an engineering review, and the design belongs to your engineer. The work is the readiness review.
Questions people ask
Why is a data center priced differently from other buildings?
Because the structure and envelope are a minority of the cost. Most of the money is in electrical distribution, standby power, cooling and controls, which are specified to a performance requirement rather than drawn, and which carry lead times measured in many months.
What is the single largest variable?
Critical load capacity and the redundancy behind it. It drives generator count, switchgear, uninterruptible power, cooling tonnage, fuel storage and often the structure carrying all of it. It also tends to be decided by somebody who is not a party to the construction contract.
Can the capacity risk be closed before signature?
Sometimes. Where the end user requirement is known it can be fixed. Where it is not, pricing the next block up as an option with a validity period converts an open exposure into a decision with a deadline, which is a far better thing to be holding.
This is general information about construction contracts and is not legal advice.