Data center electrical equipment, in the order it has to be settled
The longest-lead items now quote in years, which means orders have to be placed against a preliminary specification rather than a finished design. This is the sequence that follows from that, and the six equipment categories it runs through.
Why the schedule is set by the equipment, not the design
On a conventional building, electrical equipment is ordered once the design is largely settled. On a data centre that sequence no longer works. The longest-lead items — transformers, medium-voltage switchgear, generation — now quote in years rather than months, which means the order has to be placed against a preliminary specification, well before the design that would normally justify it is finished.
That inversion is what makes data centre electrical equipment its own discipline. The engineering decisions have not become harder; the consequence of making them late has. A transformer specified six months later than it could have been does not delay the transformer by six months — it can delay energisation by considerably more, because everything downstream is waiting on it.
The critical path, in order
The dependency chain is unusually rigid, and it runs in one direction. Each decision constrains the next, which is why getting the sequence right matters more than optimising any single item.
- 1. Utility interface and voltage class. Whether you take power at transmission or distribution voltage decides whether large power transformers and MV switchgear are in scope at all. This is the single highest-leverage decision on the schedule.
- 2. Transformer rating and impedance. kVA follows the load; impedance decides the available fault current downstream. Both have to be settled before switchgear can be specified.
- 3. Switchgear interrupting rating. Determined by the fault current from step 2. Specify it before ordering and it is a number; discover it afterwards and it is a change order.
- 4. Backup generation and ride-through. Sized against the critical load plus cooling, and split between generation and stored energy because they solve different problems.
- 5. Cooling capacity. Follows IT load almost one for one, and has to survive the transfer to generator alongside the load producing the heat.
- 6. Distribution and cabling. Shortest lead times, most frequently changed, and the layer where the continuous-load derate quietly decides how many racks actually fit.
The six equipment categories
Each page below covers what separates the families in that category, the specifications that decide the selection, applicable standards, indicative lead times with their basis, and the calculators worth running before a specification is issued.
Four mistakes that cost the most
Specifying a transformer on kVA alone
Impedance decides the available fault current downstream, and therefore the interrupting rating every protective device needs. A lower %Z gives a stiffer source and a higher fault current — the opposite of most people's intuition. Ordering switchgear before this is settled is how lineups end up under-rated.
Treating a generator as the whole backup answer
A generator needs time to start, reach speed and accept load. Nothing carries the critical load during those seconds unless stored energy is specified for it. Conversely, sizing a UPS for hours of runtime buys expensive battery to do a job the generator already does.
Planning rack capacity against the breaker nameplate
A circuit supplying a continuous load must be sized at 125 % of that load, so usable capacity is 80 % of the breaker rating. Data centre load is continuous, which makes a 20 A circuit a 16 A circuit in practice. Ignoring this consistently over-counts how many racks a room supports.
Reading an allocation position as a lead time
“Sold out through 2028” is not a queue you join earlier by ordering sooner — it is committed manufacturing capacity. The productive conversation is about allocation, alternates and phased delivery, and it needs to happen far earlier than a normal design cycle would suggest.
Tools worth running before you specify
Background reading
Common questions
When should long-lead equipment actually be ordered?
Against a preliminary specification, as early as the utility voltage class and approximate load are known — not after the design is finished. With transformers indicatively at 85 to 160 weeks, the difference between ordering at concept and ordering at issued-for-construction can exceed a year of schedule.
What is the single highest-leverage decision?
The utility interface voltage. It determines whether large power transformers and medium-voltage switchgear are in scope at all — and those are precisely the two categories that are allocation-limited rather than merely slow.
Can rebuilt or recertified equipment shorten the schedule?
Sometimes, particularly for interim or lower-capacity needs, usually at a premium over new. It is worth evaluating explicitly rather than treating new-build as the only path, especially where the alternative is holding up energisation.
How should these lead-time figures be used?
For sequencing, not for pricing or commitment. They are indicative modelled positions for North America, reviewed monthly, and published with the basis of each figure stated. Actual lead time depends on the manufacturer, the configuration and the moment you ask.
Working through a build?
Every family is in the Spec Library with its configuration axes, the standards it is built to, its indicative lead time and its market price band. Take the specification to whoever supplies it — Voltfield sells nothing and is not a route to any of it.