Nvidia and Oracle Make Standalone Vera CPUs Available on Oracle Cloud
Vera's memory bandwidth lead over dual Xeon 6 systems in local testing is the clearest sign of its strength, though the Oracle deployment itself was not benchmarked.
Nvidia and Oracle have made Vera Rubin NVL72 racks and standalone Vera CPUs generally available on Oracle Cloud Infrastructure Superclusters, announced at Oracle AI World. The move follows Google Cloud's general availability for Vera Rubin the prior week, and it marks Nvidia's first standalone CPU push into a hyperscale data center, placing Vera against x86 chips from AMD and Intel. In Tom's Hardware's own testing of a local Vera server, which was not run on OCI, the Stream triad benchmark showed Vera more than doubling the memory bandwidth of a dual Xeon 6980P system. Vera also outperformed chips with more raw power on paper in Python workloads, which matter for agent frameworks and AI libraries. Oracle is also offering the HGX NVL8 with eight Rubin GPUs paired with a dual 64-core Xeon 6 host and 4TB of DDR5, signaling a cautious rollout for a multi-gigawatt deployment.
Why it matters
For readers following the chip market, this is Nvidia's first standalone CPU offering in a large cloud provider, putting its Vera processors directly against x86 chips from AMD and Intel in a hyperscale setting. The memory bandwidth results come from a local server rather than Oracle's infrastructure, so they show what the chip can do in one configuration rather than how Oracle's clusters will perform. Oracle's pairing of Rubin GPUs with dual Xeon 6 hosts also suggests that x86 CPUs will remain part of the mix for now.
Matters or hype?
Does this story matter, or is it hype? Vote before you see what everyone else said.