NVIDIA GPUs
Latest generation silicon, dedicated to your workload.
Sovereign AI
GPU compute, AUD-billed.
Dedicated NVIDIA silicon on Australian infrastructure. On-demand, lower cost than offshore, AUD-billed.
100% Australian & Compliant by Design
ISO 27001Tier 3+ Australian data centresCarrier neutral
What it is Sovereign GPUs.
Sovereign GPUs.
Rapid deployment based on custom requirements.
Warm-pool provisioning means a GPU is running before your standup is scheduled. Use it for fine-tunes, batch inference, vision and large-context workloads.
Built on enterprise silicon
Rapid deployment. Sovereign by default.
Every GPU runs on Australian-controlled infrastructure under Fortinet security, with the same OpenStack control plane and AUD billing as the rest of the Amaze stack.
Why Amaze for GPU
Dedicated silicon. Predictable AUD pricing.
No noisy neighbours. No time-sharing. No FX surprises. Just dedicated GPU capacity, lower cost than offshore, low-latency to local data, AUD-billed.
Sovereign by default
Compute and storage in ap-syd-2.
Rapid deployment
Rapid deployment based on custom requirements.
Carrier-grade transit
Up to 40 Gbps onto the Amaze backbone.
Enterprise grade network performance
99.9 %
Uptime SLA
Financially backed, measured monthly.
Tier 3
Australian data centres
Equinix and NEXTDC, carrier-neutral facilities.
40 Gbps
Transit per region
Carrier-grade Australian backbone.
< 15 min
P1 incident response
Australian-based on-call, 24×7.
Catalogue
Run any open-weight or frontier model, in Australia.
All major LLM families, dozens of sizes, one Australian sovereign compute engine. Pick the model, call the endpoint, ship the feature.
Anthropic
OpenAI
Meta · Llama
Mistral AI
Google · Gemma
Qwen
DeepSeek
Hugging Face
Sovereign GPU compute, built the way Australian teams need it.
Five things make Amaze GPU compute feel different from a hyperscaler region in Sydney.
Dedicated by default
100% allocated GPU silicon to your workload. No oversubscription, no shared kernels, no surprise contention.
Multi-node scale
Span a single training or batch job across multiple GPU nodes when the model demands it. Same control plane, same billing entity, no rearchitecting.
Hybrid by design
Integrate with on-prem or hyperscaler environments via IP Transit and VPC. Move inference and training between sovereign and hyperscaler tenancies without rewriting the integration.
Onshore data for open-source models, end-to-end
Weights, checkpoints, datasets, logs and intermediate artefacts all stay in Australian regions. The full open-source AI lifecycle, under Australian law.
Lower TCO and gain sovereignty
Predictable AUD pricing on dedicated silicon. No FX swings, no inter-region egress charges, no foreign-controlled control plane. Lower total cost than the hyperscalers and Australian jurisdiction at the same time.
Need reserved capacity?
Reserve dedicated GPU pools.
When the workload outgrows on-demand, production fine-tunes, multi-node training, compliance-bound applications, GPU-as-a-Service gives you reserved silicon with the same API surface.
Australian infrastructure built with leading global technology
GPU Compute, frequently asked questions
What GPU silicon do you run?
NVIDIA GPUs across all Amaze GPU compute, hosted in ap-syd-2. Each GPU is dedicated, no time-sharing across tenants.
Where is my data stored?
100% in Australian Tier 3+ data centres. Sydney (ap-syd-2). No offshore data path.
What about multi-node training?
Multi-node GPU clusters are available through GPU & CPU Clusters under HPC. Same silicon, RDMA-class fabric.
Reserve sovereign GPU capacity in minutes.
Talk to an Australian solution architect about the workload, the silicon, and the contract terms.