NVIDIA Vera Rubin Superchip
Overview
The Vera Rubin Superchip pairs two Rubin GPUs (576 GB total HBM4, 44 TB/s aggregate memory bandwidth) with one 88-core Vera CPU built on Arm-compatible Olympus cores and 1.5 TB of LPDDR5X CPU memory. The CPU and GPUs are linked by NVLink-C2C at 1.8 TB/s coherent bandwidth, with GPU-to-GPU NVLink bandwidth of 7.2 TB/s aggregate. Headline NVFP4 inference throughput is 100 PFLOPS dense; NVFP4 training is 70 PFLOPS dense. Tensor-core-emulated FP32 (SGEMM) reaches 800 TFLOPS and FP64 (DGEMM) 400 TFLOPS. This Superchip is the rack-building block for Vera Rubin NVL72. Specifications are preliminary per NVIDIA and subject to change.
Performance
Peak theoretical throughput by precision type
| Precision | Dense | 2:4 Sparse |
|---|---|---|
FP64 64-bit floating point | 67TFLOPS | Structured sparsity is a tensor-core feature; this vector precision has no sparse form |
FP64 Tensor 64-bit floating point with Tensor Cores | 400TFLOPS | Structured sparsity is a tensor-core feature; this vector precision has no sparse form |
FP32 32-bit floating point | 260TFLOPS | Structured sparsity is a tensor-core feature; this vector precision has no sparse form |
FP32 Tensor | 800TFLOPS | Structured sparsity is a tensor-core feature; this vector precision has no sparse form |
TF32 TensorFloat-32 | 4,000TFLOPS | The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision |
BF16 Brain Float 16 | 8,000TFLOPS | The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision |
FP16 16-bit floating point | 8,000TFLOPS | The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision |
FP8 8-bit floating point | 35,000TFLOPS | The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision |
FP6 | 35,000TFLOPS | The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision |
FP4 | No verified data available | The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision |
NVFP4 | 100,000TFLOPS | The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision |
INT8 8-bit integer | 500TOPS | The part supports 2:4 sparsity, but the vendor publishes no sparse figure for this precision |
Specifications
Architecture
Rubin
Form Factor
Superchip
Launch Year
2026
Process Node
No verified data available
Memory
576 GB HBM4
Bandwidth
44,000 GB/s
TDP
No verified data available
Max power
No verified data available
Interconnect
1.8 TB/s NVLink-C2C
bidirectional, per GPU
Spec Confidence
Vendor claimed
Full Specifications
| Compute Engine | |
|---|---|
| GPUs per Module | 2 (all figures are totals for the module) |
| Memory | |
| Memory | 576 GB |
| Memory Type | HBM4 |
| Bandwidth | 44.0 TB/s |
| Interface Width | No verified data available |
| Interconnect & I/O | |
| GPU-to-GPU | NVLink-C2C |
| Interconnect Bandwidth | 1.8 TB/s bidirectional, per GPU |
| Power & Thermal | |
| TDP | No verified data available |
| Max power | No verified data available |
| Enterprise Features | |
| Sparsity | Yes |
| General | |
| Form Factor | Superchip |
| Architecture | Rubin |
| Process Node | No verified data available |
| Launch Year | 2026 |
Datasheet & Resources
Flopper Datasheet
NVIDIA Vera Rubin Superchip specifications, generated from our database. Printable.
Vendor product page
NVIDIA documentation for the Vera Rubin Superchip
NVIDIA Vera Rubin NVL72 Specifications
NVIDIA · 2025-03-18
NVIDIA Developer Documentation
Technical specs, programming guides
Data Provenance
Every figure traced to a source
Primary Source
- Publisher
- NVIDIA
- Published
- 2025-03-18
Data Quality
- Spec confidence
- Vendor claimed
- Clock basis
- Boost
- Core precisions with figures
- 8 of 9
- Normalization
- All values in TFLOPS
Compare cloud providers offering on-demand GPU instances for AI training, inference, and HPC workloads.
Browse GPU Cloud ProvidersSimilar GPUs
Frequently Asked Questions
How many TFLOPS does the NVIDIA Vera Rubin have?
The NVIDIA Vera Rubin delivers 260 TFLOPS FP32, 8,000 TFLOPS FP16 and 35,000 TFLOPS FP8 at peak.
What is the power consumption of the NVIDIA Vera Rubin?
Flopper does not currently have a verified TDP (Thermal Design Power) figure for the NVIDIA Vera Rubin.
How much memory does the NVIDIA Vera Rubin have?
The NVIDIA Vera Rubin is equipped with 576 GB of memory with 44,000 GB/s of memory bandwidth.
What architecture is the NVIDIA Vera Rubin based on?
The NVIDIA Vera Rubin is based on the Rubin architecture, launched in 2026.
How much does the NVIDIA Vera Rubin cost to rent?
Flopper is not tracking any live rental listings for the NVIDIA Vera Rubin. The provider directory lists the clouds that rent comparable accelerators on demand.
When did the NVIDIA Vera Rubin come out?
The NVIDIA Vera Rubin launched in 2026 on the Rubin architecture.
Stay Updated on GPU Releases
Get notified when new GPUs are added or specifications are updated.
No spam, unsubscribe anytime.