FP329,360 TFLOPS | 260 TFLOPS | 130 TFLOPS
FP642,400 TFLOPS | 67 TFLOPS | 33 TFLOPS
TF32³144 PFLOPS | 4 PFLOPS | 2 PFLOPS
CPU MemoryUp to 54 TB LPDDR5X | Up to 1.5 TB LPDDR5X | -
Verdict
High-performance computing system with 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs.
Where it wins, where it doesn't
Pros
- Offers up to 9,360 TFLOPS FP32 and 288 PFLOPS FP16/BF16 for high-performance computing.
- Equipped with up to 54 TB of LPDDR5X CPU memory, ensuring ample resources for complex tasks.
- Supports up to 20.7 TB HBM4 GPU memory with 1,400 TB/s bandwidth, ideal for data-intensive applications.
Cons
- High power consumption and cooling requirements make it unsuitable for environments without robust infrastructure.
- Limited networking bandwidth in smaller configurations restricts scalability and flexibility.
- Complex NVLink requirements necessitate a specific setup, complicating deployment and maintenance.
Ideal forData centers with extensive computational needsResearch institutions requiring high-performance computingCompanies engaged in large-scale AI and machine learning projects
Key Features
- ✦72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs
- ✦Up to 54 TB LPDDR5X CPU memory
- ✦20.7 TB HBM4 GPU memory with 1,400 TB/s bandwidth
- ✦288 PFLOPS FP16/BF16 and 9,360 TFLOPS FP32 performance
- ✦Sixth Generation NVIDIA NVLink with 216 TB/s bandwidth
- ✦Supports up to 40K GPUs in 100 MW
Specifications
| FP32 | 9,360 TFLOPS | 260 TFLOPS | 130 TFLOPS |
|---|---|
| FP64 | 2,400 TFLOPS | 67 TFLOPS | 33 TFLOPS |
| TF32³ | 144 PFLOPS | 4 PFLOPS | 2 PFLOPS |
| CPU Memory | Up to 54 TB LPDDR5X | Up to 1.5 TB LPDDR5X | - |
| Inlet Temp | 45°C |
| FP16/BF16³ | 288 PFLOPS | 8 PFLOPS | 4 PFLOPS |
| Configuration | 72 NVIDIA Rubin GPUs | 36 NVIDIA Vera CPUs | 2 NVIDIA Rubin GPUs | 1 NVIDIA Vera CPU | 1 NVIDIA Rubin GPU |
| NVIDIA NVLink | Sixth Generation |
| CPU Core Count | 3,168 custom NVIDIA Olympus cores | 6,336 Threads | 88 custom NVIDIA Olympus cores | 176 Threads | - |
| GPUs In 100 MW¹ | 40K GPUs |
| NVFP4 Training³ | 2,520 PFLOPS | 70 PFLOPS | 35 PFLOPS |
| NVLink Bandwidth | 216 TB/s | 6 TB/s | 3 TB/s |
| NVFP4 Inference² | 3,600 PFLOPS | 100 PFLOPS | 50 PFLOPS |
| FP8/FP6 Training³ | 1,260 PFLOPS | 35 PFLOPS | 17.5 PFLOPS |
| NVLink-C2C Bandwidth | 65 TB/s | 1.8 TB/s | - |
| GPU Memory | Bandwidth | 20.7 TB HBM4 | 1,400 TB/s | 576 GB HBM4 | 38.5 TB/s | 288 GB HBM4 | 19.2 TB/s |
| Total NVIDIA + HBM4 Chips | 1,296 | 30 | 12 |
| Networking Bandwidth (Scale-Out)´ | 32.4 TB/s | 0.9 TB/s | 0.45 TB/s |
Editorial note
The NVIDIA Vera Rubin NVL72 is a powerful supercomputing platform designed for large-scale AI and HPC workloads. It features 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs, providing extensive memory and processing capabilities. The system's high power draw and complex NVLink requirements make it suitable primarily for data centers with robust infrastructure. The trade-off is clear: while it delivers exceptional computational power, it demands significant investment in hardware and cooling solutions.
Frequently Asked Questions
Who is NVIDIA Vera Rubin NVL72 for?↓
NVIDIA Vera Rubin NVL72 is a fit for data centers with extensive computational needs, Research institutions requiring high-performance computing and Companies engaged in large-scale AI and machine learning projects.
What are the drawbacks of NVIDIA Vera Rubin NVL72?↓
The trade-offs we record are: High power consumption and cooling requirements make it unsuitable for environments without robust infrastructure., Limited networking bandwidth in smaller configurations restricts scalability and flexibility. and Complex NVLink requirements necessitate a specific setup, complicating deployment and maintenance..
What does NVIDIA Vera Rubin NVL72 do well?↓
Offers up to 9,360 TFLOPS FP32 and 288 PFLOPS FP16/BF16 for high-performance computing., Equipped with up to 54 TB of LPDDR5X CPU memory, ensuring ample resources for complex tasks. and Supports up to 20.7 TB HBM4 GPU memory with 1,400 TB/s bandwidth, ideal for data-intensive applications..
Alternatives to consider
See all alternatives →- Apple Mac Studio (M5 Max/Ultra)36-core
- NVIDIA Vera CPUNVIDIA Vera delivers system-level efficiency as the host CPU for AI factories, including NVIDIA Vera Rubin NVL72 and HGX™ Vera Rubin NVL8 platforms. Vera feeds GPUs for large-scale AI while running the CPU work that keeps the factory operating, including ETL, key-value (KV) cache management, and orchestration. With high single-threaded performance, massive memory bandwidth, and a single compute die design that avoids cross-chiplet latency, Vera delivers predictable performance while keeping GPUs fully utilized across accelerated AI and HPC systems.
- Snapdragon X2 Series80 TOPS
Further reading
Featured badge
Building this product? Add the badge to your site to show it’s in the index.
<a href="https://fathomlayer.com/compute/edge-computing-hardware/nvidia-vera-rubin-nvl72" target="_blank" rel="noopener noreferrer"><img src="https://fathomlayer.com/fathom-badge.svg" alt="Featured on Fathom Layer" width="250" height="54" /></a>