Resources
Technical resources for GPU infrastructure.
Field guides, platform requirements, project tools and reference reading for teams that plan, procure and deploy AI infrastructure.

On this page
01Field guides
Field guides
Planning guides for scope, inputs, cabling, verification and handoff.
Data Center Cabling: Specify the Channel, Scope and Handoff
Endpoints, cable assemblies, fiber channels and acceptance evidence before a cabling package is committed.Rack and Stack for Data Centers
Receiving to handoff: GPU rack assembly, power and cabling, testing and choosing a provider.GPU Deployment Providers: Compare Scope and Evidence
Compare proposals by defined work, OEM requirements, project evidence and physical handoff.NVIDIA Rubin: Define the Physical Work Package
Vera Rubin NVL72 physical interfaces, placement, external cable maps and handoff responsibilities.GB200 NVL72 Installation Checklist
Rack configuration, site releases, field-installed connections and physical handoff evidence.InfiniBand NDR Cabling: Match the Host Connection
DGX H100 and H200 cluster connections, ConnectX-7 variants and BlueField-3 interfaces.GPU Cluster Commissioning Workflow
The five Cx levels from factory witness to production sign-off, and the gate before each.GPU Cluster Acceptance Testing: Review the Evidence
Acceptance results by system, test scope and metric, with installation evidence kept separate.
02GPU platforms
GPU platform requirements
Installation requirements by platform, from Vera Rubin NVL72 to H100.

Rubin
Vera Rubin NVL72
Rubin GPUs, Vera CPUs, HBM4 and sixth-generation NVLink: physical planning and site prerequisites.Blackwell Ultra
GB300 NVL72
72 Blackwell Ultra GPUs and 36 Grace CPUs: OEM configuration checks, cabling and handoff scope.Blackwell Ultra
B300
HGX B300 servers and GB300 NVL72 racks: rails, placement, cable plans and physical verification.Blackwell
GB200 NVL72
72 Blackwell GPUs and 36 Grace CPUs in one NVLink domain: OEM rack requirements and cabling.Blackwell
B200
HGX B200 servers and GB200 NVL72 racks: placement, rack assembly, cabling and labeling.Grace Hopper
GH200
Grace CPU and Hopper GPU over NVLink-C2C: the server variant and its rack requirements.Hopper
H200
141 GB of HBM3e in SXM and NVL systems with different physical and power requirements.Hopper
H100
SXM and NVL configurations: OEM model, rack elevation and cable plan before installation.
03Hardware
Hardware we deploy
Vendor equipment we install and integrate.
NVIDIA
DGX, HGX and OEM configurations: physical rack work, structured cabling and handoff.Dell
PowerEdge GPU servers: OEM rails and rack inputs, cabling, labeling and handoff.Supermicro
GPU server systems: model-specific assembly, cabling, labeling and handoff.Arista
Ethernet switching for GPU fabrics: rack mounting, structured cabling and labels.

04Project tools
Project tools
Free calculators for early scoping of racks, power and cooling.
05Reference
Articles and reference
Articles, the blog, topic hubs, the glossary and common questions.
Articles
Research and planning articles on GPU hardware, cabling, power and cooling interfaces.Blog
GPU infrastructure insights, deployment guides and industry analysis.Topics
Topic hubs for AI hardware, network architecture, structured cabling, testing and more.Glossary
GPU infrastructure terms defined, from NVLink and InfiniBand to OTDR and MPO.FAQ
Answers about rack-and-stack, cabling, verification, customer inputs and responsibilities.
06Developers
Developers and API
The public content API, MCP server, SDKs and command-line tool.
GET
/api/v1/api/v1/resources/api/v1/resources/{id}/api/v1/resources/batchAPI reference
The public, read-only content API: company profile, published titles and source documents.MCP server
Connect a remote MCP client to three public, read-only tools.JavaScript and Python SDKs
Official clients with cursor pages, typed errors and package verification.Command-line tool
Search the catalog and retrieve documents or the company profile from a terminal.
Contact
Discuss Your Infrastructure Project
Tell us about your project, and our team will follow up to discuss the scope, requirements and availability.