Planning Dedicated Air Capacity for Data Centre Projects

Data centre programmes are governed by a small number of immovable dates: grid connection, energisation, integrated systems testing, customer handover and, on AI builds, the allocation window for compute. Freight planning that starts from shipment readiness will always be reactive, because the equipment pipeline slips and the milestone does not.

Planning from the milestone backwards produces a different question: what has to be on site by when, in what order, and what is the latest credible departure for each of those tranches. That framing exposes which movements have real slack and which are already fragile, usually well before anyone is asking for an urgent charter.

A useful capacity model does not need final packing lists. It needs, for each tranche, an approximate total volume in CBM, an approximate gross weight, the largest single piece dimension, whether the tranche is only useful complete, and what it blocks if it is late.

With that, aircraft requirements can be sized realistically. Technology tranches are almost always volume-limited rather than weight-limited, so the model should count pallet positions and cubic metres. Power and electrical plant behaves the opposite way and introduces floor-loading constraints. Most builds contain both, and mixing them intelligently across rotations is where cost is genuinely saved.

• Total CBM and gross weight per tranche, approximate is sufficient
• Largest single piece, length, width, height, and whether it can be laid down
• Whether the tranche must arrive complete to be useful
• The dependency it releases: energisation, fit-out, install, testing or handover
• Origin and the realistic release date, including integration or staging time
• Dangerous goods content, particularly lithium and battery systems

Most of a data centre's supply chain should not fly. Sea freight for structural, mechanical and non-critical items, consolidated airfreight for steady technology flow, and charter reserved for the movements where scheduled capacity genuinely cannot deliver the outcome.

The typical pattern is that charter earns its place in three places: the first major technology tranche where the schedule is already compressed, outsize or heavy power and cooling plant that cannot travel as standard air freight, and recovery movements when an upstream delay has eaten the float. Identifying those three in advance is more valuable than any rate negotiation.

Committing capacity in advance secures aircraft and usually a better rate, but carries exposure if the pipeline moves. Buying spot capacity keeps flexibility but means competing for aircraft at the moment the whole market is also under pressure, and at the moment your own leverage is lowest.

In practice a hybrid works best on most programmes: plan and pre-agree the rotations tied to genuinely fixed milestones, keep flexible sourcing for the tranches whose readiness is uncertain, and pre-agree an escalation route so that when something does slip, the urgent movement is a phone call rather than a fresh procurement exercise.

Contingency on these programmes is not simply extra time. It is a pre-decided answer to the failure modes that recur: a manufacturing release slipping past the planned departure, a dangerous goods issue disqualifying the intended operator, a curfew or slot preventing a planned arrival, a site not able to receive equipment on the day it lands, and a single gating item, optics, a transformer bushing, a controller, holding a whole hall.

Each of those has a cheap answer if it is planned and an expensive answer if it is not. Knowing in advance which curfew-free gateway you would divert to, which alternative aircraft type your crate dimensions still fit, and where equipment can be staged if the site is not ready, converts most of these from crises into decisions.

Air capacity planning that stops at the arrival airport tends to fail at the last five kilometres. Data centre sites frequently have restricted access hours, limited hardstand, specific crane or tail-lift requirements, active construction traffic and an installation sequence that will only accept equipment in a defined order.

The domestic leg should therefore be planned against the site's actual receiving capability and sequence, including whether equipment needs to be staged in a bonded or secure warehouse before delivery. On more than a few programmes, staging near the gateway and releasing to site on the installation team's schedule is both cheaper and faster than pushing equipment to a site that cannot take it.