Nvidia unveils AI infrastructure spanning chips to space computing

NVIDIA has introduced a new AI infrastructure centered on its Vera CPU and the Vera Rubin platform, targeting the evolving needs of agentic AI systems and large-scale AI data centers. The Vera CPU, featuring 88 custom Olympus cores and high-bandwidth memory, is designed to efficiently manage thousands of simultaneous AI environments, delivering twice the efficiency and 50% faster performance than traditional rack-scale CPUs. This marks a shift where CPUs play a central role in orchestrating AI workloads alongside GPUs, rather than merely supporting them. Major cloud and infrastructure providers such as Alibaba, Meta, Oracle Cloud Infrastructure, Dell Technologies, and Lenovo plan to deploy systems based on this processor. The Vera Rubin platform integrates seven specialized chips across compute, networking, and storage to create what NVIDIA calls “AI factories,” large-scale facilities capable of generating massive AI token volumes required for modern models. A key configuration, Vera Rubin NVL72, combines 72 GPUs and 36 Vera CPUs in a single rack, delivering up to four times better