Analysis from the Datacentre Floor
Practical thinking on GPU economics, cluster sizing, software ecosystems, and the market forces shaping AI infrastructure.
The State of AI Compute in Europe: 2026
Demand is exploding, supply stays tight, and sovereignty requirements are reshaping where models can live. What it means for teams building in the region.
NVIDIA H100 vs AMD MI300X: A Practical Comparison for AI Teams
Specs, training and inference performance, availability, cost — and how to decide between them for your stack.
Why AMD Is Gaining Ground in Enterprise AI Infrastructure
The supply squeeze opened the door; the technical case and the economics are why it's staying open.
How to Size an LLM Training Cluster: A Practical Framework
Parameters, precision, optimiser state, activations — the memory math that actually determines how many GPUs you need.
AMD ROCm vs NVIDIA CUDA: What AI Teams Need to Know Before Migrating
What's drop-in, what needs porting, what a realistic migration costs — and a four-step path to do it properly.
Reading About It Isn't the Same as Running It
We'll benchmark your workload on real MI300X silicon so you can decide on data, not specsheets.
Request a Benchmark