Home > AI Solutions > Gen AI > White Papers > Generative AI in the Enterprise with NVIDIA Spectrum-X Networking Platform > Fabrics overview
A single GPU cluster can consist of up to four fabrics. The following figure shows an example of two AI servers (Dell PowerEdge XE9680 servers) connecting to all four network types:
This figure shows the different fabrics that support the deployment of a typical AI workload:
This reference guide only discusses the Scale Out fabric (also known as “back-end,” “east-west,” or AI fabrics), which interconnects dozens of compute nodes with hundreds or even thousands of GPUs to process AI workloads.