More Tokens per Second per Megawatt: The Economics of NVIDIA Vera Rubin

10X more tokens per second per megawatt as measured by CoreWeave on the NVIDIA Vera Rubin NVL72 platform. The next generation of AI factories will be measured by how efficiently they turn compute into tokens, and tokens into revenue. Achieving that requires extreme co-design across the entire stack, paired with an open ecosystem built to innovate and scale. NVIDIA’s Dion Harris breaks down 5 critical questions every AI infrastructure leader should be asking across compute, networking, storage, software, and security — and connects the technology behind Vera Rubin to the economics shaping the future of AI infrastructure. Learn more: https://www.cio.com/article/4209776/5-critical-questions-that-define-ai-factory-economics.html https://blogs.nvidia.com/blog/vera-rubin/