B200 NVL 192GB

The B200 NVL 192GB is a Blackwell-generation NVIDIA GPU built on the GB202 architecture, manufactured on a TSMC 4NP process node with 208 billion transistors. Released in 2025, it delivers 3500 TFLOPS of FP16 throughput and 3500 TFLOPS BF16 — the two precision formats most commonly used for transformer model training and inference. FP8 precision is supported at 7000 TFLOPS — roughly 2.0× the FP16 rate — enabling near-doubled throughput for inference workloads that can tolerate reduced numerical precision with calibration. The GB202 architecture represents NVIDIA's approach to balancing compute throughput, memory bandwidth, and power efficiency for data center AI workloads. At 900W TDP, the B200 NVL 192GB sits in the ultra-high-power data center tier (900W), requiring specialized rack infrastructure with high-density power delivery.

Memory capacity is 192GB of HBM3e with 8000 GB/s bandwidth. This determines which models can run without quantization: approximately 96B parameters at FP16 (2 bytes/param), 192B at INT8 (1 byte/param), or up to 384B parameters at INT4/GGUF quantization (0.5 bytes/param). These figures are theoretical maximums — actual capacity is reduced by KV cache, framework overhead, and activation memory, typically by 10–20% for inference and 30–40% for training. The arithmetic intensity ceiling — the ratio of compute (3500 TFLOPS) to memory bandwidth (8000 GB/s) — is approximately 438 FLOP/byte. Transformer inference is typically memory-bound below this threshold, meaning the 8000 GB/s bandwidth figure is the primary determinant of tokens-per-second for most LLM serving workloads.

Multi-GPU configurations are a first-class use case for the B200 NVL 192GB. NVLink 5.0 / PCIe 6.0 provides 1800 GB/s of bidirectional NVLink bandwidth between cards, enabling tensor-parallel inference across multiple GPUs with near-linear VRAM scaling. A two-card configuration provides 384GB of pooled VRAM — sufficient for 192B parameter models at FP16 — while a four-card setup reaches 768GB. NVLink's low-latency, high-bandwidth fabric makes all-reduce operations in data-parallel training significantly faster than PCIe-based alternatives, which top out at ~64 GB/s bidirectional for PCIe 5.0 x16.

The primary workloads for the B200 NVL 192GB are Large-scale inference, Multi-GPU NVLink clusters, Memory-bound LLM serving. NVLink PCIe form factor B200. 192GB HBM3e with NVLink 5.0 connectivity. Lower TDP than SXM variant for standard PCIe rack deployments. Key limitations to factor into your evaluation: Very limited cloud availability — still ramping supply; Extremely high hourly cost vs H100/H200; Requires NVLink 5.0 infrastructure for multi-GPU setups. When comparing this GPU against alternatives at similar price points, the most important metrics are memory bandwidth (for inference throughput), VRAM capacity (for model size), and FP16/BF16 TFLOPS (for training speed). Raw TFLOPS figures can be misleading for inference — a GPU with lower TFLOPS but higher memory bandwidth will often outperform a higher-TFLOPS card on tokens-per-second for memory-bound autoregressive generation.

In the broader GPU market, the B200 NVL 192GB is a one-year-old design that remains competitive for most workloads. CUDA compatibility is a significant advantage: the B200 NVL 192GB benefits from the full NVIDIA software ecosystem including cuDNN, TensorRT, FlashAttention, and all major inference frameworks (vLLM, TGI, TensorRT-LLM). CUDA's maturity means optimized kernels are available for virtually every model architecture. For cloud rental, availability varies significantly by provider — some specialize in this GPU tier while others may have limited stock. Compare on-demand and spot pricing across providers using the rental comparison table on this page, and factor in region availability if latency is a concern for your inference workload.