NVIDIA RTX PRO 6000 Blackwell Server Edition Short Bracket Passive 96GB

NVIDIA RTX PRO 6000 Blackwell Server Edition Short Bracket Passive 96GB
This product is intended solely for businesses and educational institutions. Completing an order on the e-shop does not constitute a binding contract, but rather the submission of an enquiry.
The sale of this product is subject to export control, sanctions and other legal restrictions. Before an order can be accepted, the buyer, the end user, the country of delivery and the intended use must be verified. An order can only be accepted after successful authorisation by the supplier.
If the required conditions are not met, the enquiry may be declined or the order cancelled without any entitlement to delivery of the goods.
If you are interested in purchasing an AI solution or accelerator, you can also email us at [email protected]. We will then prepare a proposal and an individual contract for you. For further information, please refer to our terms and conditions (Articles 3.33 and 3.34).

NVIDIA RTX PRO 6000 Blackwell Server Edition Short Bracket Passive 96GB graphics card
The NVIDIA RTX PRO 6000 Blackwell Server Edition Passive 96GB is engineered precisely for this environment, featuring 96GB of GDDR7 memory with 1597GB/s bandwidth, 24064 CUDA cores, 5th generation Tensor Cores with FP4 support reaching 4 PFLOPS, and passive cooling fully integrated into server infrastructure. From agentic AI and large language model inference to physics simulations and professional visualisation and VFX rendering – the NVIDIA RTX PRO 6000 Blackwell Server Edition Passive 96GB brings the performance, flexibility and reliability of the Blackwell generation to the data centre.

Key features of the NVIDIA RTX PRO 6000 Blackwell Server Edition Short Bracket Passive 96GB graphics card
- The card has 96GB of GDDR7 memory with 1597GB/s bandwidth on a 512-bit bus to process large-scale AI models, scientific datasets and complex rendering scenes without memory bottlenecks in a server environment.
- The NVIDIA Blackwell architecture with 24064 CUDA cores and a boost frequency of up to 2288MHz achieves 120 TFLOPS in FP32 and 355 TFLOPS of RT core performance, while new neural shaders integrate neural networks directly into programmable shaders.
- 5th generation Tensor Cores with FP4 precision support achieve 4 PFLOPS, 2 PFLOPS in FP8, and 1 PFLOP in FP16/BF16, allowing the card to hardware-accelerate inference and training of the most demanding AI models directly in the data centre.
- 4th generation RT Cores with 355 TFLOPS of performance and RTX Mega Geometry technology can process up to 100× more ray-traced triangles than the previous generation for physically accurate visualisations and engineering simulations.
- Passive cooling without an active fan is designed for server enclosures with controlled airflow, where silent and reliable operation without moving parts ensures maximum uptime in a production data centre.
- Multi-Instance GPU allows the card to be divided into up to four fully isolated instances – configurable as 4× 24GB, 2× 48GB or 1× 96GB – each with its own memory, cache and guaranteed quality of service for secure concurrent work by multiple users or workloads.
- Four 9th generation NVENC video encoders and four 6th generation NVDEC decoders with support for 4:2:2 10-bit H.264 and HEVC encoding enable professional video processing and real-time streaming directly in a server environment.
- The PCIe 5.0 x16 interface with double the throughput of PCIe Gen 4 and support for DisplayPort 2.1 with resolutions up to 16K at 60Hz ensure high-speed data transfer and direct output for visualisation and VFX workflows.
- The card is fitted with a shortened low-profile bracket instead of the standard full-height one, enabling installation in server chassis and rack systems where the standard design would not physically fit – while performance and specifications are identical to the reference version.

Inference on a level that data centres need
Agentic AI, large language model inference and generative AI place demands on hardware infrastructure that are measured in petaflops – and the NVIDIA RTX PRO 6000 Blackwell Server Edition Passive 96GB responds with 4 PFLOPS in FP4, 2 PFLOPS in FP8, and 1 PFLOP in FP16/BF16. The 5th generation Tensor Cores with structured sparsity and FP4 precision support enable processing of models with tens of billions of parameters directly in a production data centre, without relying on remote cloud infrastructure. For businesses building an AI factory or deploying agentic AI solutions, this is a foundation that keeps pace with the industry's development.

Passive cooling for a server environment without compromise
In a production data centre, every moving part determines the overall system uptime. The passive cooling of the NVIDIA RTX PRO 6000 Blackwell Server Edition Passive 96GB eliminates fans as a potential point of failure and fully integrates the card into the cooling infrastructure of a server enclosure with controlled airflow. The result is silent, maintenance-free operation with maximum uptime for continuous AI workloads, scientific simulations and VFX render farms. Support for the NVIDIA MGX server platform then allows for flexible deployment in 6U and other configurations for building efficient AI factory environments.

One GPU for the entire data centre
Specialised accelerators handle one type of workload – the NVIDIA RTX PRO 6000 Blackwell Server Edition Passive 96GB handles them all at once. With 120 TFLOPS in FP32, 355 TFLOPS of RT Core performance, and 4 PFLOPS in FP4, a single card can handle agentic AI and generative AI inference, physics simulations, engineering CAD rendering, scientific computing and professional video processing in a single server slot. Support for NVIDIA AI Enterprise and NVIDIA Omniverse then connects AI-driven 3D, robotics and the digitalisation of industrial processes with a single hardware platform – and MIG instances allow different teams or projects to be served concurrently with a guaranteed quality of service.
Specifications
Chip
Performance & usage
Memory
Frequency
Interface
Technology
Cooling system
Dimensions
Consumption
AI
NCNR (Non-Cancellable, Non-Returnable)
Tops
AI specifications
| Code: | NVrpbs02 |
| Warranty: | 36 months |
| Links: | Producer's Website: |












