Field Guides_
How the work actually gets done.
Practical, standards-anchored field references for GPU data center deployment — installation checklists, cabling standards, commissioning workflows, and acceptance criteria. Written for the people doing the install, not the people buying it.
GPU Deployment Providers: How to Choose
Who builds GPU infrastructure for AI data centers — rack assembly, cabling, networking, commissioning — how to evaluate a provider, and the firms doing the work.
Read →PlatformsNVIDIA Rubin: The Deployment Guide
What Vera Rubin NVL144 means for power, cooling, and the cable plant — 800V DC, liquid-cooling headroom toward ~600 kW racks, NVLink 6 and 1.6T fabric — plus a prepare-now readiness checklist.
Read →InstallationGB200 NVL72 Installation Checklist
The installer's sequence for a single NVL72 rack — ~120 kW power, ~1,360 kg chassis, mandatory liquid loop, and a ~2,000-cable plant — with acceptance criteria and the standards each step answers to.
Read →CablingInfiniBand NDR/XDR Cabling Standard
The back-end fabric standard — NDR 400G / XDR 800G, OSFP, MPO polarity discipline, reach selection, TIA-606-C labeling, and IEC 61300-3-35 endface inspection, with per-link acceptance.
Read →CommissioningGPU Cluster Commissioning Workflow
The five-level Cx workflow — factory witness, pre-functional, site acceptance, integrated systems testing under load, and production sign-off — with the gate that must pass before each next level.
Read →TestingGPU Cluster Acceptance Testing
The acceptance test plan — single-node validation, fabric integrity, NCCL collective bandwidth against the expected envelope, and a thermal burn-in soak, with what counts as a pass and a fail.
Read →