NVIDIA Rubin SXM

Rubin SXM 2026 Spec confidence: Vendor claimed
FP8 (dense)
17,500
TFLOPS
FP32
130
TFLOPS
VRAM
288 GB
HBM4
Bandwidth
22.0 TB/s
memory
Download datasheet

Overview

NVIDIA Rubin is the post-Blackwell datacenter GPU announced at GTC 2025 and shipping in late 2026. Each Rubin GPU pairs with HBM4 stacks delivering 22 TB/s bandwidth, ships on the SXM form factor, and supports 6th-gen NVLink at 3.6 TB/s per GPU. Rubin debuts NVFP4, a microscaled 4-bit format that hits 50 PFLOPS dense for inference (35 PFLOPS dense for training) and adds tensor-core-emulated SGEMM/DGEMM that lifts FP32 to 400 TFLOPS and FP64 to 200 TFLOPS. Specifications are preliminary per NVIDIA and subject to change. NVFP4 training throughput on this GPU is 35 PFLOPS (dense).

Performance

Peak theoretical throughput by precision type

FP32_TC
400 TFLOPS
FP6
17,500 TFLOPS
sparse not published
NVFP4
35,000 TFLOPS
50,000 with sparsity
INT8
8-bit integer
250 TOPS
sparse not published
FP8
8-bit floating point
17,500 TFLOPS
sparse not published
FP16
16-bit floating point
4,000 TFLOPS
sparse not published
BF16
Brain Float 16
4,000 TFLOPS
sparse not published
TF32
TensorFloat-32
2,000 TFLOPS
sparse not published
FP32
32-bit floating point
130 TFLOPS
FP64_TC
64-bit floating point with Tensor Cores
200 TFLOPS
FP64
64-bit floating point
33 TFLOPS
Dense Added with sparsity

Specifications

Architecture

Rubin

Form Factor

SXM

Launch Year

2026

Memory

288 GB HBM4

Bandwidth

22,000 GB/s

Interconnect

3.6 TB/s NVLink

Spec Confidence

Vendor claimed

Full Specifications

Memory
VRAM 288 GB
Memory Type HBM4
Bandwidth 22.0 TB/s
Interconnect & I/O
GPU-to-GPU NVLink
Interconnect Bandwidth 3.6 TB/s
Power & Thermal
Enterprise Features
Sparsity Yes
General
Form Factor SXM
Architecture Rubin
Launch Year 2026

Datasheet & Resources

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
Precisions tracked
11
Normalization
All values in TFLOPS

Compare cloud providers offering on-demand GPU instances for AI training, inference, and HPC workloads.

Browse GPU Cloud Providers

Similar GPUs

Systems Using This GPU

Pre-configured systems featuring the NVIDIA Rubin

SystemGPU CountPeak PerformanceTotal Power
NVIDIA DGX Rubin NVL8
DGX · rack · 2026
8x Rubin 280.00 PFLOPS 24.0 kW View System
NVIDIA HGX Rubin NVL8
HGX · baseboard · 2026
8x Rubin 280.00 PFLOPS -- View System
NVIDIA Vera Rubin NVL72
NVL · rack · 2026
72x Rubin 2520.00 PFLOPS -- View System
Inventec Artemis III Rack
Artemis · rack ·
72x Rubin -- 228.0 kW View System
Ingrasys IGS-HXR200
IGS · rack ·
8x Rubin -- -- View System

Frequently Asked Questions

How many TFLOPS does the NVIDIA Rubin have?

The NVIDIA Rubin delivers 130 TFLOPS for FP32 operations, 4000 TFLOPS for FP16, and 17500 TFLOPS for FP8 precision.

What is the power consumption of the NVIDIA Rubin?

The NVIDIA Rubin has a TDP (Thermal Design Power) rating of N/A watts.

How much memory does the NVIDIA Rubin have?

The NVIDIA Rubin is equipped with 288 GB of VRAM with 22000 GB/s memory bandwidth.

What architecture is the NVIDIA Rubin based on?

The NVIDIA Rubin is based on the Rubin architecture, launched in 2026.

Stay Updated on GPU Releases

Get notified when new GPUs are added or specifications are updated.

Loading verification...

No spam, unsubscribe anytime.

Back to GPUs