Wall St Engine(@wallstengine):NVIDIA's Vera Rubin NVL72 could train large mixture-of-experts AI models using 75% fewer GPUs than Grace Blackwell. Training a 10T-parameter model on 100T tokens in one month would just require roughly 16,000 Rubin GPUs, compared with 64,000 GB200 GPUs.

Wall St Engine

@wallstengine

Fast, accurate, consistent stock market news, earnings highlights & more. By Brillinsight. Not financial advice.

Joined March 2022

985 Following 209.8K Followers

Wall St Engine@wallstengine

2026.10.26 10:13

NVIDIA's Vera Rubin NVL72 could train large mixture-of-experts AI models using 75% fewer GPUs than Grace Blackwell. Training a 10T-parameter model on 100T tokens in one month would just require roughly 16,000 Rubin GPUs, compared with 64,000 GB200 GPUs.