--- title: "NVIDIA Vera Rubin NVL72 | Aguasto" description: "Manufacturer reference for NVIDIA Vera Rubin NVL72: Vera CPU / Rubin GPU, rack system, with sourced specifications, variant boundaries and deployment considerations." canonical: "https://www.aguasto.com/gpus/vera-rubin-nvl72" last-updated: "2026-10-12" --- # NVIDIA Vera Rubin NVL72 | Aguasto Canonical HTML: [NVIDIA Vera Rubin NVL72](https://www.aguasto.com/gpus/vera-rubin-nvl72) > Manufacturer reference for NVIDIA Vera Rubin NVL72: Vera CPU / Rubin GPU, rack system, with sourced specifications, variant boundaries and deployment considerations. [Aguasto](https://www.aguasto.com/) [GPU Library](https://www.aguasto.com/gpus) NVIDIA Vera Rubin NVL72 [Resources](https://www.aguasto.com/resources) [GPU Library](https://www.aguasto.com/gpus) [Model Index](https://www.aguasto.com/models) [Inference Stack](https://www.aguasto.com/inference-stack) [Learn](https://www.aguasto.com/learn) [Insights](https://www.aguasto.com/insights) [Research](https://www.aguasto.com/research) [Engineering tools](https://www.aguasto.com/tools) NVIDIA / Rack-scale system ## NVIDIA Vera Rubin NVL72 Manufacturer disclosures describe systems shipping and production racks delivered by September 2026. This establishes a shipping ramp, not universal availability or an Aguasto allocation. Original Aguasto conceptual illustration. Not manufacturer geometry, an operating Aguasto asset, or a measured benchmark. Aguasto Research documents third-party hardware, models and software. References do not represent Aguasto inventory, evaluated deployments or available inference services. ### Manufacturer specification reference Architecture Vera CPU / Rubin GPU[NVIDIA Vera Rubin NVL72 (external source)](https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/) · [NVIDIA Vera Rubin datasheet (external source)](https://dam-cdn.nvd.orangelogic.com/AssetLink/56p68o47y6f1yucpubl0ump1c0aif422.pdf) · [From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI (external source)](https://blogs.nvidia.com/blog/coreweave-agentic-ai-vera-rubin/) · [Inside NVIDIA Rubin GPU Architecture (external source)](https://developer.nvidia.com/blog/inside-nvidia-rubin-gpu-architecture-powering-the-era-of-agentic-ai/) Category Complete rack-scale reference[NVIDIA Vera Rubin NVL72 (external source)](https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/) · [NVIDIA Vera Rubin datasheet (external source)](https://dam-cdn.nvd.orangelogic.com/AssetLink/56p68o47y6f1yucpubl0ump1c0aif422.pdf) · [From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI (external source)](https://blogs.nvidia.com/blog/coreweave-agentic-ai-vera-rubin/) · [Inside NVIDIA Rubin GPU Architecture (external source)](https://developer.nvidia.com/blog/inside-nvidia-rubin-gpu-architecture-powering-the-era-of-agentic-ai/) Accelerator memory 20,700 GBManufacturer-rounded 20.7 TB aggregate GPU HBM, converted to decimal GB; Vera CPU memory excluded.[NVIDIA Vera Rubin NVL72 (external source)](https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/) Memory type HBM4[NVIDIA Vera Rubin NVL72 (external source)](https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/) · [NVIDIA Vera Rubin datasheet (external source)](https://dam-cdn.nvd.orangelogic.com/AssetLink/56p68o47y6f1yucpubl0ump1c0aif422.pdf) · [From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI (external source)](https://blogs.nvidia.com/blog/coreweave-agentic-ai-vera-rubin/) · [Inside NVIDIA Rubin GPU Architecture (external source)](https://developer.nvidia.com/blog/inside-nvidia-rubin-gpu-architecture-powering-the-era-of-agentic-ai/) Memory bandwidth 1,400 TB/sPublished rack aggregate GPU memory bandwidth; not per-GPU or NVLink bandwidth.[NVIDIA Vera Rubin NVL72 (external source)](https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/) Power Not verified for this recordNo total measured rack input power verified. Power-shelf nameplate capacity and benchmark power assumptions must not be reported as operating rack draw.[NVIDIA Vera Rubin NVL72 (external source)](https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/) · [NVIDIA Vera Rubin datasheet (external source)](https://dam-cdn.nvd.orangelogic.com/AssetLink/56p68o47y6f1yucpubl0ump1c0aif422.pdf) · [From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI (external source)](https://blogs.nvidia.com/blog/coreweave-agentic-ai-vera-rubin/) · [Inside NVIDIA Rubin GPU Architecture (external source)](https://developer.nvidia.com/blog/inside-nvidia-rubin-gpu-architecture-powering-the-era-of-agentic-ai/) Precision formats NVFP4, FP8, FP6, FP16, BF16, TF32, FP32, FP64[NVIDIA Vera Rubin NVL72 (external source)](https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/) Cooling / integration Liquid-cooled rack; the September-revision datasheet describes a fully liquid-cooled warm-water design.[NVIDIA Vera Rubin NVL72 (external source)](https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/) · [NVIDIA Vera Rubin datasheet (external source)](https://dam-cdn.nvd.orangelogic.com/AssetLink/56p68o47y6f1yucpubl0ump1c0aif422.pdf) · [From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI (external source)](https://blogs.nvidia.com/blog/coreweave-agentic-ai-vera-rubin/) · [Inside NVIDIA Rubin GPU Architecture (external source)](https://developer.nvidia.com/blog/inside-nvidia-rubin-gpu-architecture-powering-the-era-of-agentic-ai/) ### Configuration boundaries - The verified rack is NVL72. NVL4 and HGX NVL8 are separate configurations and are not treated as the same rack. - Current rack table lists 19.2 TB/s per GPU; the architecture article states up to 22 TB/s. This record uses the explicitly published rack aggregate, without blending these boundaries. - Manufacturer sparse inference and dense training FLOPS must remain separate. ### Published arithmetic throughput Manufacturer peaks, not Aguasto measurements or tokens per second **Specification** NVFP4 peak compute **Precision** NVFP4 **Published peak** 3,600 PFLOPS **Boundary & source** Whole NVL72 rack; manufacturer sparse inference specification[Manufacturer source ↗](https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/) **Specification** NVFP4 peak compute **Precision** NVFP4 **Published peak** 2,520 PFLOPS **Boundary & source** Whole NVL72 rack; manufacturer dense training specification[Manufacturer source ↗](https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/) ### Deployment considerations - Use current rack-level documents rather than extrapolating from an early Rubin GPU architectural maximum. - Verify facility supply, coolant quality and temperatures, commissioning criteria and software release support with the OEM. #### Software ecosystem NVIDIA AI Enterprise · NVIDIA NIM · NVIDIA Mission Control. Confirm the exact model, runtime release, server integration and vendor support matrix before evaluating a deployment. ### Continue the research [NVIDIA Dynamo](https://www.aguasto.com/inference-stack/dynamo) [TensorRT-LLM](https://www.aguasto.com/inference-stack/tensorrt-llm) [Kubernetes GPU orchestration — NVIDIA device plugin](https://www.aguasto.com/inference-stack/kubernetes) [Model weights versus runtime memory](https://www.aguasto.com/learn/model-weights-and-runtime-memory) [Compare compatible hardware categories](https://www.aguasto.com/gpus/compare) [Why modular infrastructure?](https://www.aguasto.com/why-modular) ### Sources & verification Reference revision 1 · Reviewed October 12, 2026. Manufacturer claims and documentation are attributed; Aguasto has not run the cited products in production. - [NVIDIA Vera Rubin NVL72 ↗](https://www.nvidia.com/en-us/data-center/vera-rubin-nvl72/) 72 Rubin GPUs, 36 Vera CPUs; 20.7 TB HBM4 and 1400 TB/s aggregate GPU bandwidth; Precision and sparse/dense throughput table; Shipping/full-production language. Publication: date not stated in the reviewed source. Verified: 2026-10-12. - [NVIDIA Vera Rubin datasheet ↗](https://dam-cdn.nvd.orangelogic.com/AssetLink/56p68o47y6f1yucpubl0ump1c0aif422.pdf) Fully liquid-cooled rack; AI Enterprise, NIM and Mission Control software. Publication: date not stated in the reviewed source. Verified: 2026-10-12. - [From Training to Production, NVIDIA and CoreWeave Close the Loop on Agentic AI ↗](https://blogs.nvidia.com/blog/coreweave-agentic-ai-vera-rubin/) First Vera Rubin NVL72 production racks received in September and provider availability announced. Publication: 2026-09-30. Verified: 2026-10-12. - [Inside NVIDIA Rubin GPU Architecture ↗](https://developer.nvidia.com/blog/inside-nvidia-rubin-gpu-architecture-powering-the-era-of-agentic-ai/) Architectural maximum per-GPU bandwidth stated as up to 22 TB/s; differs from current rack product table. Publication: 2026-07-21. Verified: 2026-10-12. ### Reference journey [Compare hardware ↗](https://www.aguasto.com/gpus/compare) [Model Index ↗](https://www.aguasto.com/models) [Serving software ↗](https://www.aguasto.com/inference-stack) [OEM collaboration ↗](https://www.aguasto.com/partners) [Sources & verification ↓](https://www.aguasto.com/gpus/vera-rubin-nvl72#sources) Research reference · No available Aguasto service ## Company navigation - [Home](https://www.aguasto.com/) - [Infrastructure](https://www.aguasto.com/infrastructure) - [GPU Cloud](https://www.aguasto.com/gpu-cloud) - [AI Inference](https://www.aguasto.com/ai-inference) - [Development & Investment](https://www.aguasto.com/development) - [Company](https://www.aguasto.com/company) - [Contact](https://www.aguasto.com/contact) - [Privacy](https://www.aguasto.com/privacy) - [Resources](https://www.aguasto.com/resources) - [GPU Library](https://www.aguasto.com/gpus) - [Model Index](https://www.aguasto.com/models) - [Inference Stack](https://www.aguasto.com/inference-stack) - [Learn](https://www.aguasto.com/learn) - [Insights](https://www.aguasto.com/insights) - [Research](https://www.aguasto.com/research) - [Industry cases](https://www.aguasto.com/research/cases) - [Why Modular](https://www.aguasto.com/why-modular) - [Partners](https://www.aguasto.com/partners) - [GPU comparison](https://www.aguasto.com/gpus/compare) - [Model comparison](https://www.aguasto.com/models/compare) - [Engineering tools](https://www.aguasto.com/tools) - [Development friction](https://www.aguasto.com/research/hyperscale-development-friction) - [NVIDIA H100 SXM](https://www.aguasto.com/gpus/h100) - [NVIDIA H200 SXM](https://www.aguasto.com/gpus/h200) - [NVIDIA B200 SXM](https://www.aguasto.com/gpus/b200) - [NVIDIA B300](https://www.aguasto.com/gpus/b300) - [NVIDIA GB200 NVL72](https://www.aguasto.com/gpus/gb200-nvl72) - [NVIDIA GB300 NVL72](https://www.aguasto.com/gpus/gb300-nvl72) - [NVIDIA Vera Rubin NVL72](https://www.aguasto.com/gpus/vera-rubin-nvl72) - [AMD Instinct MI300X](https://www.aguasto.com/gpus/mi300x) - [AMD Instinct MI325X](https://www.aguasto.com/gpus/mi325x) - [AMD Instinct MI355X](https://www.aguasto.com/gpus/mi355x) - [AMD Instinct MI455X](https://www.aguasto.com/gpus/mi455x) - [AMD Helios](https://www.aguasto.com/gpus/helios) - [GLM-4.5](https://www.aguasto.com/models/glm-4-5) - [Qwen3-235B-A22B](https://www.aguasto.com/models/qwen3-235b-a22b) - [DeepSeek-R1](https://www.aguasto.com/models/deepseek-r1) - [Llama 3.3 70B Instruct](https://www.aguasto.com/models/llama-3-3-70b-instruct) - [Gemma 3 27B IT](https://www.aguasto.com/models/gemma-3-27b-it) - [gpt-oss-120b](https://www.aguasto.com/models/gpt-oss-120b) - [Kimi-K2-Instruct](https://www.aguasto.com/models/kimi-k2-instruct) - [vLLM](https://www.aguasto.com/inference-stack/vllm) - [SGLang](https://www.aguasto.com/inference-stack/sglang) - [TensorRT-LLM](https://www.aguasto.com/inference-stack/tensorrt-llm) - [NVIDIA Dynamo](https://www.aguasto.com/inference-stack/dynamo) - [NVIDIA Triton Inference Server](https://www.aguasto.com/inference-stack/triton) - [AMD ROCm](https://www.aguasto.com/inference-stack/rocm) - [Ray Serve](https://www.aguasto.com/inference-stack/ray-serve) - [Kubernetes GPU orchestration — NVIDIA device plugin](https://www.aguasto.com/inference-stack/kubernetes) - [Model weights are only the beginning of the memory budget](https://www.aguasto.com/learn/model-weights-and-runtime-memory) - [Why tokens per second is not enough to benchmark inference](https://www.aguasto.com/research/inference-benchmark-methodology) - [Prefill, decode and the cost of separating them](https://www.aguasto.com/learn/prefill-decode-and-disaggregated-serving) - [Choosing between vLLM, SGLang and TensorRT-LLM](https://www.aguasto.com/learn/choosing-an-inference-serving-framework) - [From Power to Inference—Building a Distributed AI Compute Network](https://www.aguasto.com/learn/from-power-to-inference) - [Dedicated compute and shared inference: compare the whole workload](https://www.aguasto.com/insights/dedicated-and-shared-inference-economics) - [Distributed inference reliability begins with the failure boundary](https://www.aguasto.com/learn/distributed-inference-reliability) - [Core Scientific: reading contracted, energized and billed capacity](https://www.aguasto.com/research/cases/core-scientific-capacity-disclosures) - [Why Modular AI Infrastructure Matters Now](https://www.aguasto.com/research/why-modular-ai-infrastructure-matters-now) - [Official LinkedIn](https://www.linkedin.com/company/aguasto) This is a generated representation of the public HTML page. Diagrams are conceptual. GPU and inference services are planned and are not currently available to book or call. The contact page describes the available inquiry flow.