… The Vera CPUs will also be deployed for rack-scale AI, offering 256 Vera chips per rack, up to 400 TB of memory, and up to 315 TB/s of bandwidth. Vera is just as crucial as Rubin as Agentic AI shifts the core focus from GPUs to CPUs. …
… Memory is the key cost shift. The report states that memory could account for roughly 25% of the total VR200 NVL72 cost. That means around $2 million of each rack could be tied to memory-related content. This includes LPDDR5X system memory, 3D NAND storage, and HBM4 used on the Rubin GPUs. …
… Related Story AMD MI430X Is The “Highest Performance” FP64 GPU Ever Built, Surpassing NVIDIA’s Rubin By 6x In Classic-HPC Workloads According to SemiAnalysis , NVIDIA's Vera Rubin NVL72 specifications will now feature a memory bandwidth of 22.2 TB/s, a massive increase from the initial specificatio… …
… In the Vera Rubin platform architecture with LPX, decode is best thought of as a two-engine loop. GPUs handle decode work that benefits most from throughput and large memory capacity, such as full-context attention over the accumulated KV cache. …
… By combining Olympus CPU cores, second-generation SCF, high-bandwidth LPDDR5X memory, and coherent NVLink-C2C connectivity, Vera ensures Rubin GPUs remain productive across training, post-training, and inference workloads, even as execution shifts between compute, memory, and communication-dominate… …
… Read more : NVIDIA's next-gen Vera Rubin NVL576 AI server: 576 Rubin AI GPUs, 12672C/25344T CPU, new HBM4 Read more : NVIDIA officially unveils Rubin: its next-gen AI platform with huge upgrades, next-gen HBM4 Read more : NVIDIA updates roadmap, with new details on its next-gen GPU 'Feynman' coming… …
… Similar News Stories SK hynix already pre-supplying next-gen HBM4 memory to NVIDIA for its next-gen Rubin AI GPUs SK hynix hits 70% yield on its new HBM4 12-Hi memory: ready for NVIDIA's next-gen Rubin AI GPUs SK hynix supplies 'early' HBM4 samples, testing will take longer than HBM3E for AI chip m… …
… Here is how the Hopper NVL8, Grace-Blackwell NVL72, and Vera-Rubin NVL72 systems stack up in terms of throughput tokens per second per megawatt and interactivity tokens per second per user : It is immediately obvious that the larger shared GPU memory domain enabled by NVSwitch helps stretch the cur… …
ASUS Unveils Game-Changing Liquid-Cooled AI Infrastructure Powered by NVIDIA Vera Rubin Platform March 17, 2026 ・ Press Release Delivering trusted AI with total flexibility, from rack-scale AI factories to edge and enterprise deployment · Rack-scale AI power: ASUS AI POD built on NVIDIA Vera Rubin … …
… Categories: AI Infrastructure Tags: Agentic AI AI Factory NVIDIA Blackwell NVIDIA Rubin Related News NVIDIA Vera CPU Is ‘Packing a Heavy-Hitting Punch’ Against Competition The shift to agentic AI creates a new CPU requirement for the AI factory: fast cores, massive memory bandwidth and the ability … …
