03 // LARUS AI™ · INSIDE PORTADC
Cheap power can still produce expensive compute if GPUs sit idle or site permissions stall. LARUS AI™ connects customer workload commitments, site limits, qualified pod specs, and live operating evidence inside one system.
01 · Per pod
450 kW → 1.2 MW
Per Pod, Two Hulls
01b · Per campus
22.5MW+
Per Campus, Run as One Fleet
01c · Speed to power
120DAYS
Target from Order to Powered Racks
02 · All-in cost
-12–21%
Under Sydney Metro, All-in
05 · Cooling water
0LITRES
Evaporative Cooling Water
Firm + Flex workload & site envelope allocator
Break-even utilisation 65% · design model
No peak event called today. Firm holds its floor around the clock; Flex fills the rest of the envelope.
Firm SLA
Protected
Flex backlog scheduled
24h/day
Avg utilisation
90%
Firm floor on a called event
Held
Firm · protected 24/7 resources
Flex · deadline queues
Site envelope available to the pod
Surge band shaped by the tenant (called event / heat)
01 · Real time, bank-grade auditability
What it does
LARUS AI™ enforces each pod's contracted power and thermal SLA in real time, aligning tenant compute workloads with site generation so every kilowatt is delivered with bank-grade auditability. It never promises the same kilowatt twice to the host and the tenant.
How we leverage it
Pulse tenants run on the floor with a band of zero and are never trimmed. Surge tenants carry a tenant-authorised 25 to 40% band and answer the power-budget signal with their own orchestration; what they retain through a called event is their own target.
02 · One system, every site
What it does
Holds the tenant's workload commitments, the site's permitted envelope and the qualified pod configuration in one system, so capacity is usable from the first hour and repeatable at every site.
How we leverage it
Tenants get earlier usable capacity with fewer delivery handoffs, and CANUS carries the same contract structure from one pod into every subsequent site.
03 · One record, every party
What it does
LARUS AI™ cryptographically signs boundary power and thermal delivery into a single auditable ledger, from factory test through every hour of operation, shared by the tenant and the host site.
How we leverage it
The host and the tenant read the same hardware-attested record of the same configuration: transparent SLA verification and power settlement from the first hour.
LARUS AI™ platform capabilities
Five capabilities, one control plane across every deployed pod
Routes inference, prefill, RAG and batch training jobs across distributed PortaDC pods based on real-time latency, thermal headroom and power availability.
THE OUTCOME
Real time, bank-grade auditability · One system, every site · One record, every party
Next
A short conversation about what you run, where your customers are and when you need it.