Infrastructure
NVIDIA B200
NVIDIA's 2024 Blackwell GPU — 192GB HBM3e, ~2.5× H100 performance for FP8 training.
Definition
The B200 is NVIDIA's Blackwell-architecture flagship for AI: 192GB HBM3e per package, second-gen Transformer Engine and FP4 support. Paired in NVL72 racks via NVLink, it is the platform on which frontier 2025 models are being trained.
Common use cases
- Frontier training
- Inference at scale
- Trillion-parameter MoEs