Amaze Resources, Articles
Long-form articles on sovereign Australian AI, Cloud and Data infrastructure from the Amaze engineering team.
- Website
- uniquemedia.com.au · all feeds from this site
- Feed URL
https://www.amaze.au/resources/articles/rss.xml
Read it in your FeedListing reader
Or add to: Feedly Inoreader Raw feed
Latest posts
-
AI workload profiles: what runs in an AI facility
Training, fine-tuning, inference and retrieval behave nothing alike in a rack. Here is the infrastructure profile each one actually demands.
-
AI factory vs data centre: what the label buys
"AI factory" is NVIDIA's framing, and the EU uses the words for something else. Where the distinction is real engineering, and what to evaluate.
-
Storage design when power is the real constraint
When power procurement gates new builds, storage design stops being a dollars-per-terabyte exercise and becomes a watts-per-terabyte one.
-
Data centre tiers: what a Tier 3 rating omits
Uptime Institute tiers describe topology, not density, cooling or interconnect. What Tier 3 and Tier 4 leave open for GPU workloads.
-
Data sovereignty when AI agents make the calls
An agent that retrieves documents and calls third-party models moves data across boundaries you never approved. Where the perimeter breaks.
-
DCIM for AI GPU clusters: what monitoring misses
A DCIM stack built for 5kW racks and five-minute polling cannot see a GPU cluster. What changes in power, thermal and capacity planning.
-
How an edge rollout rewrites storage strategy
Edge does not shrink what you store. It changes where data lands and what your five-year storage budget has to model.
-
Choosing hardware for a private cloud deployment
Compute, storage and network selection for a private cloud, sized against real workload profiles, plus the four places teams overspend.
-
What regtech vendors must prove about infrastructure
Regulated buyers no longer accept a hosting region as a sovereignty answer. The procurement evidence Australian vendors are now asked to produce.