AI infrastructure · Brisbane, Australia
We develop AI infrastructure.
Turnkey delivery of a 50 MW IT load, in 10 MW trains, from Pre-FEED to FID to O&M. Silicon stays open between NVIDIA and AMD.
The foundation
Our Brisbane RTX 4090 cluster. See live telemetry.Moorooka micro-lab · Brisbane
Operating today.
Control plane & telemetry.
Connecting to telemetry…
- GPU
- —
- GPU utilisation
- —
- GPU memory
- —
- GPU-rail power
- —
- GPU temperature
- —
Waiting for the source timestamp.
Live measurements from our hardware control plane and continuous telemetry pipeline.
The architecture we’re developing
50 MW — IT load
Balance of plant and balance of systems sized to 50 MW of IT.
10 MW — the train
A 50 MW IT load is five 10 MW trains.
Downstream
Aloom is silicon-agnostic. NVIDIA and AMD are the two paths, with the cooler and chiller loops inside the delivery.
Electrons
Cooling loop
Five trains · one 50 MW IT load
Select the silicon
Choose GB300 or Helios. Interfaces are set in Pre-FEED.
Turnkey delivery of that 50 MW IT load, from Pre-FEED to FID to O&M.
Pre-FEED
Define the IT load for the site: how many 50 MW loads, in 10 MW trains, and NVIDIA or AMD with the cooler and chiller loops. Nothing is ordered from this page. Request a study starts Pre-FEED.
FID
The investment decision is taken on that 50 MW IT load. Aloom’s delivery scope is the 10 MW trains, ready to build.
O&M
Aloom operates and maintains what it delivered. The silicon choice stays open until NVIDIA or AMD is selected.
The same path
What each stage settles.
One sequence. Pre-FEED defines the 50 MW IT load. FID decides it. O&M runs it.
Pre-FEED
Site, how many 50 MW IT loads, in 10 MW trains, and NVIDIA or AMD with the cooler and chiller loops.
SettlesThe defined 50 MW IT load. A study starts here.
FID
Ownership, finance, and the decision to build that 50 MW IT load.
SettlesAloom’s delivery scope, ready to build.
O&M
Aloom operates and maintains the 50 MW IT load it delivered.
SettlesThe operating plant.
Intelligence needs infrastructure. Let’s build it.
We’re bringing together customers, infrastructure investors, landowners and energy partners to develop the next generation of intelligence capacity.
Hardware, designs and research
See the work behind the infrastructure.
Moorooka micro-lab
Active hardware control plane, telemetry pipeline, and sensor verification.
Operating cluster
Thermal loop architecture
Primary and secondary cooling distribution, fluid boundaries, and heat rejection.
Mechanical design
Illumem / compiler transfer
Selected Mamba-130M projections reconstructed with ideal simulated MZI factors.
Software research
Complete-system method
Reconcile facility electricity with accepted workload output under an agreed boundary.
Whole-system measurement
Project scope, site, power, equipment and delivery are established for each development project.
Build the next generation with us.
For infrastructure investors, customers, landowners and energy partners. Bring your demand, site, power position or investment thesis. Let’s explore the project it could become.