1 of 3 free pieces this week. Become a member to read every member piece. It's free with a work email or LinkedIn.
What's Happening
- Oracle and Nvidia announced the general availability of the Vera Rubin NVL72 rack and Vera CPUs on Oracle Cloud Infrastructure Superclusters, Tom's Hardware reported.
- The companies made the announcement Tuesday at the Oracle AI World event.
- Oracle highlighted the availability of standalone Vera instances on its infrastructure.
- Independent benchmarking by Tom's Hardware on a local Vera server showed a massive memory bandwidth lead over a dual Xeon 6 server.
Why It Matters
Hyperscalers are expanding their hardware offerings beyond standard x86 architectures to support intensive AI workloads. Oracle offering standalone Vera instances signals growing demand for Nvidia's custom CPUs outside of bundled GPU clusters.
The Context
- Google Cloud released Vera Rubin to its customers last week.
- AI-focused neocloud providers, including Nscale, CoreWeave, and Nebius, began deploying Vera instances earlier this month.
What They're Saying
- A standalone deployment at a major hyperscaler like Oracle sends a different signal than deployments at GPU-dominant neoclouds, Tom's Hardware noted.
What to Watch
- Watch for forthcoming review data from Tom's Hardware detailing the performance characteristics of the Vera chip.
The Bottom Line
Oracle's deployment of standalone Nvidia Vera CPUs marks a significant expansion of custom silicon options for cloud customers.
Mentioned in this story
Nvidia
Oracle Corp.
Google Threat Intelligence Group