Nvidia Rubin GPU
The Nvidia Rubin GPU is the GPU of Nvidia’s Vera Rubin platform, launched at CES on January 5, 2026.
- Vendor
- Nvidia
- Architecture
- Rubin
- Memory
- 288 GB HBM4
- Memory bandwidth (TB/s)
- 22 TB/s
- FP8 (TFLOPS)
- 17,500 TFLOPS
- FP4 (PFLOPS)
- 35 PFLOPS
- Form factor
- GPU in Vera Rubin Superchip / NVL72 rack
- Status
- In production
Rubin has 288 GB of HBM4 at 22 TB/s and a 50 PFLOPS NVFP4 Transformer Engine. It ships in Vera Rubin NVL72 racks (72 GPUs) and Vera Rubin Superchips (two Rubin GPUs, one Vera CPU). Nvidia announced on May 31, 2026 that Vera Rubin is ramping into full production, with production shipments starting in fall 2026; Nvidia labels the spec table preliminary and subject to change.
- Memory (GB)
- 288 GB
- Precision note (sparse or dense)
- FP8/FP6 17.5 PFLOPS and NVFP4 35 PFLOPS are Nvidia's training figures, marked dense. NVFP4 inference is listed as 50 PFLOPS without a sparse/dense label. Preliminary per Nvidia.
- Launch
- Launched January 5, 2026 (CES); full production announced May 31, 2026; production shipments from fall 2026
- Vendor source
- nvidia.com
- Checked
Questions
When will Nvidia Rubin ship?
Nvidia announced on May 31, 2026 that Vera Rubin is ramping into full production, with production shipments starting in fall 2026.
How much memory does the Rubin GPU have?
Rubin has 288 GB of HBM4 with 22 TB/s of memory bandwidth, per Nvidia’s preliminary specs.
What is Rubin’s FP4 performance?
Nvidia lists NVFP4 inference at 50 PFLOPS and NVFP4 training at 35 PFLOPS dense; FP8/FP6 training is 17.5 PFLOPS dense.
What is a Vera Rubin Superchip?
A Vera Rubin Superchip combines two Rubin GPUs with one Vera CPU.
Source
Every figure is taken from the vendor’s own datasheet, product page, technical documentation or announcement, linked on each chip with the date it was last checked. How we pick and check sources: editorial standards.