The companies pitch better GPU utilization as enterprises report hardware sitting idle.
VentureBeat reports that Nvidia and Oracle have made Vera Rubin NVL72 systems and standalone Vera CPUs generally available on Oracle Cloud Infrastructure, with plans to expand their Superclusters to multi-gigawatt scale. The companies are pitching the ability to scale CPU and GPU capacity separately as a way to cut wasted capacity. An eight-GPU Rubin NVL8 configuration is planned but not yet generally available, and Oracle says Vera availability will reach selected regions this quarter.
The push lands alongside a VentureBeat Intelligence survey of 218 respondents at companies with at least 100 employees, in which 79% reported GPU utilization at 50% or below, and 70% of those said they still plan to adopt new AI platforms regardless.
Why it matters: cloud AI spend keeps climbing even as many enterprises struggle to use the capacity they already have. Before committing to the next platform, confirm which workloads would actually move and whether they would run more economically there.