B200-PCIe, Nvidia B200 GPU, PCIe/48GB HBM3

The NVIDIA B200-PCIe is a data-center grade AI GPU module designed as a PCIe add-in accelerator for enterprise AI and HPC workloads. Based on the NVIDIA Blackwell B200 architecture, this NVIDIA module targets scale-out deployments in modern data centers that demand high-performance training and inference. With its PCIe form factor, it is ideal for system integrators and distributors building flexible GPU-accelerated servers for diverse AI, HPC, and cloud environments.

Product Features Comparison

Model	B200-PCIe	GH200-96GB-HBM3	ET1D2MPUA000, Huawei S12700 Switch, Main Processing Unit/Optional Clock	NM-16ESW, Cisco NM-16ESW Network Module, 16×10/100Mbps Ethernet, Router compatible, Plug-and-play
Brand	NVIDIA	NVIDIA	Huawei	Cisco
Module Type	AI GPU	AI GPU	Main Processing Unit A	EtherSwitch NM
Port Configuration	n/a	n/a	8	16
Supported Speed	n/a	n/a	10 GE/100 GE	10/100 Mbps Ethernet
Compatible Platform	Enterprise scale-out accelerator for AI/HPC	Data-center GPU for AI inference and training	S12708/S12712	Cisco routers
Installation Type	PCIe add-in form factor	n/a	19-inch Rack	n/a