NVIDIA will be offering its next-gen Vera CPUs as a standalone part of the infrastructure, with wherever the CPU workload is, it will run on NVIDIA CPUs... meaning it'll compete against AMD's mega-successful EPYC processors as well as Intel's fleets of Xeon CPUs.
In a new interview between Bloomberg and NVIDIA CEO Jensen Huang regarding the upcoming Vera CPU plan, Jensen said: "For the very first time, we're going to be offering Vera CPUs. Vera is such an incredible CPU. We're going to offer Vera CPUs as a standalone part of the infrastructure. And so not only, not only can you run your computing stack on NVIDIA GPUs, you can now also run your computing stack, wherever their CPU workload, run on NVIDIA CPUs".
Jensen continued, adding: "Vera is completely revolutionary...Coreweave is going to have to race if Coreweave's going to be the first to stand up Vera CPUs. We haven't announced any of our CPU design wins, but there are going to be many".
- Read more: NVIDIA upgrades Vera Rubin HBM4 bandwidth by 10% in order to stay ahead of AMD Instinct MI455X
NVIDIA's new "Vera" CPU is powered by 88 custom Armv9.2 "Olympus" CPU cores that use Spatial Multithreading technology, unlocking it to an 88C/176T processor. Vera features up to 1.5TB of LPDDR5X memory, with up to 1.2TB/sec of memory bandwidth, perfect for memory-intensive computing workloads.
There's also a second-gen Scalable Coherency Fabric that features up to 3.4TB/sec of bi-section bandwidth, connecting the 88C/176T of power across a unified monolithic die that removes latency issues in competing chiplet architectures. NVIDIA's new Vera chip also has second-gen NVLink Chip-to-Chip technology, offering up to 1.8TB/sec of coherent bandwidth for the external Rubin GPU.



