Built on the NVIDIA Blackwell architecture,  the platform accelerates training and elevates inference from day one. At rack scale, the GB200 NVL72 unifies Grace CPUs and Blackwell GPUs—72 GPUs on a single NVLink™ fabric—to deliver exceptional interconnect bandwidth, balanced CPU–GPU throughput, and lower orchestration overhead for large-scale model training and fine-tuning. Liquid-cooled and production-ready, it reduces rack footprint, shrinks energy per job, and compresses time-to-value while sustaining 24×7 enterprise duty cycles with advanced resiliency and security features. From cluster design and facility readiness to turn-up, validation, and benchmarking, it’s delivered end-to-end on the NVIDIA platform—so you ship breakthroughs faster with fewer moving parts and a lower total cost of ownership.

NVIDIA Blackwell Architecture

Breaking Barriers in Accelerated Computing and Generative AI

Explore the groundbreaking advancements the NVIDIA Blackwell architecture brings to generative AI and accelerated computing. Building upon generations of NVIDIA technologies, NVIDIA Blackwell defines the next chapter in generative AI with unparalleled performance, efficiency, and scale. 

A new class of AI superchip

Second-Generation Transformer Engine

Secue AI

NVLink and NVLink Switch

Decompression Engine

Reliability, Availability, and Serviceability Engine

NVIDIA GB200 NVL72

Unlocking Real-Time Trillion-Parameter Models

GB200 NVL72 connects 36 Grace CPUs and 72 Blackwell GPUs in a rack-scale, liquid-cooled design. It boasts a 72-GPU NVLink domain that acts as a single, massive GPU and delivers 30x faster real-time trillion-parameter large language model (LLM) inference. The GB200 Grace Blackwell Superchip is a key component of the NVIDIA GB200 NVL72, connecting two high-performance NVIDIA Blackwell Tensor Core GPUs and an NVIDIA Grace™ CPU using the NVIDIA NVLink™-C2C interconnect to the two Blackwell GPUs.

Supercharging Next-Generation AI and Accelerated Computing

30x LLM Inference (vs NVIDIA H100 Tensor core GPU)

4x LLM Training (vs. H100)

25x Energy Efficiency (vs. H100)

18x Data Processing vs CPU

GB200 NVL72 Technical Specifications

  • Configuration: 36 Grace CPU: 72 Blackwell GPUs
  • FP4 Tensor Core1: 1,440 PFLOPS
  • FP8/FP6 Tensor Core1:720 PFLOPS
  • FP64 Tensor Core: 2,880 TFLOPS
  • GPU Memory | Bandwidth: Up to 13.4 TB HBM3e | 576 TB/s
  • NVLink Bandwidth: 130 TB/s
  • CPU Core Count: 2,592 Arm® Neoverse V2 cores 
  • CPU Memory | Bandwidth: Up to 17 TB LPDDR5X | Up to 18.4 TB/s

GB200 Grace Blackwell Superchip

  • Configuration: 1 Grace CPU : 2 Blackwell GPU
  • FP4 Tensor Core1: 40 PFLOPS
  • FP8/FP6 Tensor Core1: 20 PFLOPS
  • FP64 Tensor Core:80 TFLOPS
  • GPU Memory | Bandwidth: Up to 372 GB HBM3e | 16 TB/s
  • NVLink Bandwidth:3.6 TB/s
  • CPU Core Count:72 Arm Neoverse V2 cores
  • CPU Memory | Bandwidth:Up to 480 GB LPDDR5X | Up to 512 GB/s

Contact ClusterVision to learn more about how the NVIDIA Blackwell Architecture can help elevate your
HPC & AI workloads