Nvidia Admits Vera Rubin Isn't as Powerful as It Originally Claimed
Nvidia has quietly revised the specifications of its next-generation Vera Rubin GPUs , affecting individual chips, its NVL72 rack, and the HGX Rubin NVL8 platform. The new spec table lists the GPUs with reduced memory bandwidth by almost 13%, and a reduction in NVLink 6 bandwidth by nearly 17%.
Nvidia debuted its next-generation Vera Rubin GPU at CES 2026. At the time, it showcased the NVL72 AI server system for data centers, highlighting its 288GB of HBM4 memory and up to 22TB/s of bandwidth per GPU. The new specifications, however, list that bandwidth at 19.2TB/s, while memory capacity remains the same.
In the HGX Rubin NVL8 system, which integrates eight Rubin GPUs with NVLink 6 networking connections, memory bandwidth is now listed at 130TB/s, down from 176TB/s —a 26% drop in the initial memory bandwidth. That comes with a drop in capacity from 2.3TB to 2TB, as well as reduced AI training performance. NVFP4 training drops from 280 PFLOPS to 270 PFLOPS, and FP8/FP6 training drops from 140 PFLOPS to 130 PFLOPS.
Credit: Sean Gallup/Getty Images
On the largest NVL72 Vera Rubin rack system, which integrates 72 GPUs with 36 Vera CPUs, memory bandwidth is now listed at 1,400TB/s rather than 1,580TB/s. Memory capacity remains the same.
Across all systems, Nvidia's listed NVLink bandwidth is down, too. On a per-GPU basis, it's down from 3.6TB/s to 3TB/s. For the HGX NVL8 system, it's gone from 28.8TB/s to 24TB/s, and for the NVL72, it's down from 260TB/s to 216TB/s.
The only positive revision is scale-out networking bandwidth, which increased from 28.8TB/s to 32.4TB/s.
The changes to Nvidia's website appear to have come around in early September, but Nvidia hasn't made any kind of announcement about these changes, nor explained why they occurred.
None of this is likely to affect demand for Vera Rubin, but it's important to consider it alongside growing competition from companies like AMD and Chinese alternatives in the international market. It's also a reminder to consider initial product announcements with a healthy dose of skepticism.
