NVIDIA RTX PRO 6000 Blackwell Server Edition, 96GB ECC GDDR7

The NVIDIA RTX PRO 6000 Blackwell Server Edition is a rack-optimized professional GPU built on the Blackwell platform, engineered for enterprise multi-GPU server and data center infrastructure. It adopts passive cooling, making it suitable for dense rack installations supporting up to eight GPUs in a single server node.

It is equipped with 96GB high-speed GDDR7 memory with ECC error correction, fifth-generation Tensor Cores and fourth-generation Ray Tracing Cores. Power is configurable from 400W to 600W, offering flexible tuning between throughput and power budget for data center scheduling.

This GPU excels at clustered LLM inference, agent AI workloads, distributed model fine-tuning, large-scale photorealistic rendering, simulation, digital twins, high-performance computing and virtual workstations via NVIDIA vGPU. The massive 96GB ECC memory enables loading of huge foundation models locally across GPU clusters. It supports MIG for resource partitioning and PCIe 5.0 x16 interface, with four DisplayPort 2.1 outputs.

It is ideal for system integrators, AI cloud operators, research institutions and manufacturing enterprises building GPU inference clusters and virtual visualization server farms.