Compute · Flagship Blackwell
NVIDIA HGX B300
Built for the inference era: large-scale LLM training and fine-tuning, high-throughput serving and multimodal workloads on the latest Blackwell platform.
Specifications
Blackwell, maxed out
| GPUs per node | 8 × B300 (SXM) |
| GPU memory | 288 GB HBM3e per GPU |
| Node interconnect | NVLink 5, NVSwitch |
| Cluster interconnect | 800 Gb/s InfiniBand per GPU (Quantum-X800) |
| Host | Dual x86 CPUs, 2+ TB RAM |
| Availability | On-demand & reserved |
Interconnect
800 Gb/s to every GPU
Quantum-X800 InfiniBand gives each B300 a dedicated 800 Gb/s path — topology-aware scheduling places your job on the closest nodes, keeping NCCL all-reduce fast at scale.
Suited workloads
Large-scale LLM training
High-throughput inference
Multimodal models
Fine-tuning at scale
Start your journey today
Tell us about your workload — an engineer, not a sales bot, will get back to you within one business day.