NVIDIA RTX PRO 5000 72GB Blackwell
Don't like the alternatives?
Choose from the full offer
NVIDIA RTX PRO 5000 72GB Blackwell
NVIDIA RTX PRO 5000 72GB Blackwell graphics card
The NVIDIA RTX PRO 5000 Blackwell features 5th-generation Tensor Cores with FP4 support, 4th-generation RT Cores with RTX Mega Geometry, and 72GB of GDDR7 ECC memory with a bandwidth of 1344GB/s, all in a dual-slot, full-height design for professional workstations. With 2064 AI TOPS, 65 TFLOPS in FP32, and 196 RT Core TFLOPS, the NVIDIA RTX PRO 5000 Blackwell handles local LLM inference, generative AI, photorealistic rendering, and scientific simulations. It also offers double the memory of the previous Ada Lovelace generation and ISV certifications for production deployment.
GDDR7 / 72 GB
Operating memory
PCIe 5.0 x16
Bus interface
1× PCIe CEM5 16-pin
Auxiliary power
Blackwell (GB202)
Chip architecture
850W
Recommended PSU wattage
Key features of the NVIDIA RTX PRO 5000 72GB Blackwell graphics card
- Features 72GB of GDDR7 ECC memory with a bandwidth of 1344GB/s—double the capacity of the previous Ada Lovelace generation—for smooth work with large-scale AI models, complex 3D scenes, and multi-application workflows without memory limits
- The NVIDIA Blackwell architecture with 14080 CUDA cores and 5th-generation Tensor Cores achieves 2064 AI TOPS and 65 TFLOPS in FP32, while FP4 precision enables faster processing of AI models with lower memory requirements for local LLM fine-tuning and generative AI
- 4th-generation RT Cores with 196 TFLOPS of performance and RTX Mega Geometry technology allow you to process up to 100× more ray-traced triangles than the previous generation for physically accurate visualisations, engineering simulations, and real-time VFX production
- New Streaming Multiprocessors with Neural Shaders integrate neural networks directly into programmable shaders, so AI is no longer a post-processing layer but an integral part of the graphics pipeline for AI-augmented workflows
- Multi-Instance GPU lets you partition the card into up to two fully isolated 24GB instances, each with its own memory, cache, and guaranteed quality of service for secure concurrent work by multiple users or workloads
- Three 9th-generation NVENC video encoders and three 6th-generation NVDEC decoders with support for 4:2:2 10-bit H.264 and HEVC enable real-time professional video processing for multi-cam productions, live streaming, and VFX workflows
- The PCIe 5.0 x16 interface offers double the throughput of PCIe Gen 4, and four DisplayPort 2.1b outputs support up to 8K at 240Hz or 16K at 60Hz, ensuring high-speed data transfer and direct output for visualisation workflows
- The card is powered via a 1× PCIe CEM5 16-pin connector with a 300W TBP and is equipped with a dual-slot active cooler in a full-height, 266.7mm long design with ISV certifications for professional production deployment
Blackwell performance for local LLM and generative AI
Fine-tuning language models, inferencing generative networks, and working with local AI assistants are no longer exclusive to data centres. With 2064 AI TOPS and 5th-generation Tensor Cores with FP4 support, the NVIDIA RTX PRO 5000 72GB Blackwell processes AI tasks faster and with lower memory requirements than previous generations. Its 72GB of GDDR7 ECC memory with 1344GB/s bandwidth ensures that even models with tens of billions of parameters won't hit a memory limit. For developers, researchers, and data scientists who want to iterate quickly while keeping sensitive data under control, the RTX PRO 5000 Blackwell is the natural foundation for a local AI infrastructure.
RTX Mega Geometry and Neural Shaders
Rendering scenes with billions of polygons and physically accurate simulations has traditionally been the domain of render farms. The NVIDIA RTX PRO 5000 72GB Blackwell pushes this boundary: 4th-generation RT Cores with RTX Mega Geometry technology allow you to process up to 100× more ray-traced triangles than the previous generation, while new Neural Shaders integrate AI directly into the graphics pipeline for real-time, AI-augmented visualisations. For architects designing detailed urban plans, VFX studios working with billions of polygons, or product designers validating products in a virtual environment, this is a computational leap that shortens iteration cycles and expands the boundaries of what is possible.
72 GB GDDR7 and enterprise reliability for production workflows
GDDR7 memory doesn't just bring higher capacity—it delivers fundamentally higher throughput for applications that previously hit data bottlenecks. With 1344GB/s, the NVIDIA RTX PRO 5000 72GB Blackwell handles extensive multi-application workflows without interruption, where CAD modelling, AI inference, and video production run concurrently. ECC memory protection eliminates computational errors for production-ready results, while ISV certifications for key professional applications guarantee proven stability and driver optimisation. MIG support, with the ability to split into two isolated 24GB instances, also allows for efficient sharing of GPU power between multiple users or projects.
NVIDIA Blackwell architecture
A breakthrough generation for AI, neural rendering, and scientific computing
5th-generation Tensor Cores
2064 AI TOPS with FP4 for local LLM inference and generative AI
72 GB GDDR7 ECC
1344GB/s for large AI models and complex 3D scenes
New streaming multiprocessors
Neural shaders for next-generation AI-augmented graphics workflows
4th generation RT cores
RTX Mega Geometry for physically accurate scenes with billions of polygons
4× DisplayPort 2.1b
Up to 8K at 240Hz or 16K at 60Hz for precise visualisation workflows
Multi-Instance GPU
Securely share the power of one card between four users or workloads
PCIe 5.0 x16
Double the throughput of PCIe Gen 4 for data-intensive AI tasksSpecifications
Dimensions
AI
Chip
Performance & usage
Memory
Frequency
Interface
Technology
Cooling system
Consumption
Tops
AI specifications
| Code: | NVrpr03 |
| Links: | Producer's Website: |