Specifications
NVIDIA Part Number
900-2G153-0000-000
PNY kit part
NVRTXPRO6000TCGPU-KIT
Product
NVIDIA RTX PRO 6000 Blackwell Server Edition
GPU
GB202 (Blackwell)
GPU Memory
96 GB GDDR7 with ECC
Memory Interface
512-bit
Memory Bandwidth
Up to 1,792 GB/s (28 Gbps GDDR7)
Host Interface
PCIe 5.0 x16
Power Consumption
400-600 W configurable (server-managed)
Cooling
Passive heatsink — requires chassis airflow
Form Factor
Standard height, dual slot
Dimensions
4.4 in (H) × 10.5 in (L)
Display Outputs
4 × DisplayPort 2.1
NVLink
Not supported
Family Siblings
600 W Workstation Edition (900-5G144-2200-000); 300 W Max-Q (900-5G153-2200-000)
Overview
The NVIDIA RTX PRO 6000 Blackwell Server Edition (part 900-2G153-0000-000) is the rack-server member of the RTX PRO 6000 Blackwell family. It ships as a passive dual-slot card with no onboard fans, and its power ceiling is configurable between 400 W and 600 W by the host chassis rather than fixed at the factory.
The passive design is what makes it a server part: airflow comes from the chassis' own fan wall, so the card can be stacked in 4-GPU or 8-GPU PCIe configurations without each card carrying its own blower. That also means platform selection matters - a chassis sized for 350 W double-width cards will not keep a 600 W configuration inside its thermal envelope.
It retains the family's core specification: 96 GB of GDDR7 with ECC on a 512-bit interface at 1,792 GB/s, plus four DisplayPort 2.1 outputs for hosted desktop and visualisation workloads. Omniverse, virtual workstation (VDI) and mixed graphics/inference nodes are the natural placements; for pure training at scale the data center parts (B200, GB200) remain the right tier.
Key Benefits
Passive dual-slot — designed for chassis airflow and multi-GPU stacking. 400-600 W configurable — tunable per platform thermal budget. 96 GB GDDR7 ECC — capacity for large models and scenes. 4 × DisplayPort 2.1 — vGPU, VDI and visualisation support.
Applications
Rack-mounted AI inference servers, NVIDIA Omniverse and digital twin nodes, virtual workstation (VDI) and vGPU hosts, media and rendering farms, and mixed graphics plus inference workloads.
Request a Quote — 900-2G153-0000-000
QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.
Get Your Quote →