architecture
NVIDIA Blackwell
platforms
GB300 NVL72, GB200 NVL72, HGX B300, HGX B200, RTX PRO 6000 Blackwell Server Edition
memory
Up to 192GB HBM3e
tensor cores
5th Gen Tensor Cores
precision
FP4, FP6, FP8, FP16
interconnect
NVLink 5.0 / NVSwitch
scalability
Multi-GPU AI Factory Ready
deployment
High-Density Datacenter Infrastructure
best for
Generative AI, Large Language Models, AI Training, AI Inference, HPC Clusters, Enterprise AI Infrastructure
features
Extreme AI compute performance, Ultra-high memory bandwidth, Low latency GPU communication, Energy-efficient AI processing, Enterprise-grade scalability