We design and operate Firmus Cloud and AI Factories — sovereign, modular systems that optimise cost and power at every layer of the stack.
01 / ANALYST READ
Our read
Firmus is a direct operator with a broad GPU lineup spanning current-generation NVIDIA capacity. Running its own network, it suits teams moving from experimentation into sustained production - or those that need range across training, inference, and fine-tuning without managing separate provider accounts.
The tradeoff: procurement is configuration-dependent and may require a quote process rather than instant checkout. Best fit for buyers with a clear workload plan.
Confidence: HighEvidence reviewed: 7 checks
02 / OVERVIEW
Company overview
Operator class
Direct operator
GPU models tracked
GB300, H200, L40S and 3 more
Registered in
AU
Services
GPU compute, Bare metal, Colocation
Network footprint
2 prefixes / 1k IPv4
03 / CAPACITY
GPU lineup by use case
Tracked GPU SKUs grouped by the workload they fit best.
Training
For multi-node runs where interconnect and cluster availability decide the outcome.
H200
QuoteClaim to publish
VRAM 141 GBClass Hopper
Best for Large-scale training where memory bandwidth decides the outcome.
H100
QuoteClaim to publish
VRAM 80 GBClass SXM
Best for Serious model training where time-to-result outweighs hourly cost.
A100
QuoteClaim to publish
VRAM 80 GBClass SXM
Best for Mature training stacks that value known performance and broad tooling support.
Inference
For predictable production serving, latency targets, and steady utilisation.
GB300
QuoteClaim to publish
Best for GB300 workloads.
L40S
QuoteClaim to publish
VRAM 48 GBClass Ada
Best for Cost-effective production inference and image generation at scale.
Fine-tuning
For smaller adaptation jobs, evaluation loops, and budget-controlled development.
A40
QuoteClaim to publish
VRAM 48 GBClass Ampere
Best for LoRA fine-tuning, evaluation, and teams watching spend closely.
04 / TRUST
Verification
Evidence that there's a real operator behind the offer.
Confirmed
Network ownership
Firmus operates its own network — 1k IPv4 addresses across 2 prefixes — rather than presenting another hyperscaler's capacity as its own.
Confirmed
Direct operator classification
Firmus controls GPU infrastructure and sells access directly. Not a marketplace or hyperscaler resale offer.
05 / COMMERCIALS
Pricing
Indicative published rates. Final cluster pricing depends on commit length, storage, networking, and capacity.
No published rates tracked. Visit Firmus’s site for a quote.
07 / NEXT
Talk to Firmus
viabandwidth does not sit between you and the provider. Go straight to their site.
firmus.co
Pricing pages, quote forms, and sales routes live on the provider’s own site.
Profile pages, buyer guides, and model explainers stay open. Vendor packages cover what shows on this profile; verification stays independent and is never for sale.