Compute · Balanced Blackwell

NVIDIA HGX B200

The balanced Blackwell platform for large LLM training, mixture-of-experts workloads and high-throughput inference.

Specifications

Blackwell balance point

GPUs per node8 × B200 (SXM)
GPU memory192 GB HBM3e per GPU
Node interconnectNVLink 5, NVSwitch
Cluster interconnect800 Gb/s InfiniBand per GPU (Quantum-X800)
HostDual x86 CPUs, 2 TB RAM
AvailabilityOn-demand & reserved

Interconnect

Scale without the slowdown

Dedicated 800 Gb/s InfiniBand per GPU, GPUDirect RDMA and topology-aware scheduling — multi-node B200 jobs behave like a single, larger node.

Blackwell GPUs

Suited workloads

LLM pre-training MoE training & serving High-throughput inference RLHF & post-training

Start your journey today

Tell us about your workload — an engineer, not a sales bot, will get back to you within one business day.