Designed, installed and supported nationally
20+ years • 100,000+ UPS units supported • 24/7 support
A conventional data centre build needs a slab, a building fabric, a mechanical contractor, an electrical contractor and a commissioning window measured in quarters. Every one of those is a dependency on your site, in your weather, with your access constraints.
Prefabrication moves that work into a factory. Racks, cooling, UPS, batteries, distribution, fire suppression, access control and monitoring are installed and proven under load before the unit ships, so what arrives is a finished facility rather than a construction project.
Three forms cover the range: the XRT-AI range for AI and HPC clusters, the XRT-CDC containerised range for edge, mining, defence and multi-megawatt sites, and CRX4-LC liquid cooling for the densities air cannot carry. The right one is decided by rack density, site conditions and how much you expect to grow.
Scoping a deployment?
XRT-AI rack node
CRX4-DLC coolant distribution unit
Three deployment forms, all factory assembled and acceptance tested before dispatch, all installed and supported by the same team.
XRT-AI · 25 kW rack node to multi-megawatt campus
XRT-CDC · 10ft, 20ft, 40ft, dual bay and multi-container
CRX4-LC · Rear door, direct-to-chip and immersion
Prefabrication is the same idea at three scales. The building blocks do not change, only how many of them arrive and how they are joined together.
Drag to the number of racks you expect to end up with, not the number you are starting with.
A finished data centre in one rack. Cooling, UPS, batteries, distribution and monitoring are built into the enclosure, so it lands in a comms room, a plant room or an edge site without a project around it.
Where it fitsAn AI pilot, a development cluster, or a single high density rack in a building that cannot carry any more.
The enclosure becomes the building. A 40ft all-in-one runs up to 18 racks in a single row with cooling on the wall, and two joined along their length make one room with a true contained aisle for up to 32.
Where it fitsMine sites, defence positions, offshore support bases, edge sites, and AI training halls at 400 to 800 kW.
Past a dual bay the modules specialise. Data halls, power rooms, network and operations each get their own unit, linked into one site and sized for the end state rather than the first stage.
Where it fitsMining and resources at scale, and AI compute clusters that are expected to keep growing after the first pod lands.
Australian owned critical power and cooling specialists. We design, supply, install, commission and maintain the whole facility, not just the unit it arrives in.
Local technicians across Sydney, Melbourne, Brisbane, Perth, Adelaide and beyond, so the people who commission your deployment are the people who come back to service it.
Every rack node, container and hall is built, wired and load tested at the factory with witness testing available, so site works reduce to footings, a supply connection, a data path and final commissioning.
Enclosure, racks, air or liquid cooling, UPS, batteries, distribution, fire suppression, access control and monitoring under a single scope, rather than five vendors pointing at each other.
Preventive maintenance, battery replacement, cooling service and emergency call-out across Australia, New Zealand and the Pacific Islands, day or night.
A prefabricated unit that arrives half finished is still a construction project. Every XRT-AI node, XRT-CDC container and liquid cooled hall is built, wired, charged and load tested on the factory floor, so what lands on your slab is a facility you can energise rather than a shell to fit out.
Rack count, kW per rack, redundancy target, ambient conditions and access constraints are enough for us to put a real configuration and budget in front of you.
The questions that come up before a deployment is scoped.
Containerised is one form of prefabricated. Every XRT-CDC is built inside an ISO shipping container. Prefabricated is the wider category and also covers a single rack node that ships as a finished unit, and a campus assembled from purpose-built pods that are not containers at all.
Rack count decides it more than anything else. One rack is a rack node. Up to 18 racks fits a single 40ft all-in-one, and two joined along their length reach 32 on a contained aisle. Past that, deployments become dedicated modules for data halls, power and support. Density then decides the cooling: air with containment below roughly 40 kW a rack, liquid above it.
Yes, provided the first deployment is planned for the end state. Distribution, cooling headroom and monitoring are sized for where you are going, so the second unit is an addition rather than a rebuild of the first. This is the single most common thing to get wrong, because a unit sized only for today forces a rebuild rather than an expansion.
The comparison is rarely like for like, because a conventional build carries a building, multiple trades and a commissioning window measured in quarters. Prefabrication moves most of that into a factory process, which removes site variability and shortens the programme. Whether it is cheaper on the day depends on configuration, cooling method and site conditions, so we cost it per project rather than quoting a rate.
Footings or a prepared slab, a supply connection sized to the load, a data path, and crane or transport access for the unit. Everything else arrives installed. We confirm all four in the assessment, because access is the constraint that most often changes the configuration on remote sites.
A rack node and a containerised unit can both be moved, which is a large part of why mining, defence and construction choose them: the facility follows the project rather than being written off with the site. A multi-container campus is relocatable module by module, though the site works and interconnection are effectively redone at the new location.
A single supply point sized to the IT load plus cooling and losses, three phase in almost every case. Generator backup, ring feeds and dual supplies are all supported and are decided by the redundancy target rather than by the form. The heat-load assessment produces the number before anything is ordered.
No, and that is deliberate. You install your own servers, storage and network, or your GPU nodes in an AI deployment. We supply and own everything around them: the enclosure, racks, cooling, UPS, batteries, distribution, fire protection, access control and monitoring, then install, commission and maintain it.
