NVIDIA

64x 400G QSFP-DD, 51.2 Tbps, RoCE, ideal for AI/ML training clusters and hyperscale data centers

64x 100G QSFP28, 12.8 Tbps, RoCE v2, ideal for AI inference and enterprise data center networking

64-port NDR 400G InfiniBand switch, 25.6 Tbps, ideal for AI/ML supercomputing and HPC clusters

80-port NDR InfiniBand switch, 32 Tbps, ideal for large-scale AI training and HPC deployments

Data processing unit, 400G, offloads networking/security/storage, ideal for cloud and AI data centers

400G InfiniBand/Ethernet NIC, RoCE, GPUDirect, ideal for AI/ML servers and HPC nodes

100G Ethernet NIC, RoCE v2, crypto offload, ideal for enterprise servers and cloud deployments

Flagship AI GPU, 80GB HBM3, 3.35 PFLOPS, NVLink, ideal for large-scale AI training workloads

AI GPU, 80GB HBM2e, PCIe 5.0, ideal for AI inference and training in standard servers

AI/HPC GPU, 80GB HBM2e, 2 PFLOPS, NVLink, ideal for AI training and scientific computing

8x H100 GPU server, 640GB HBM3, NVLink, ideal for enterprise AI training infrastructure

8x A100 GPU server, 640GB HBM2e, NVLink, ideal for enterprise AI and HPC workloads

128x 200G QSFP56, 51.2 Tbps, RoCE, ideal for AI cluster networking and hyperscale deployments

Network simulation platform, digital twin, ideal for network design and testing before deployment

Open network OS for data center switches, Linux-based, ideal for programmable data center networking

Network telemetry and observability, AI-driven, ideal for data center network operations

800G InfiniBand switch, next-gen, ideal for next-generation AI supercomputing clusters

ARM-based CPU, 144 cores, NVLink-C2C, ideal for AI inference and cloud-native workloads

GPU+CPU superchip, 141GB HBM3e, NVLink, ideal for LLM training and AI inference

Ethernet AI networking platform, RoCE optimization, ideal for AI/ML Ethernet cluster networking