en.Wedoany.com Reported - NVIDIA's Vera Rubin platform has entered full production, delivering up to 10x higher autonomous throughput at scale for AI infrastructure deployments targeting autonomous AI workloads, compared to the previous-generation Grace Blackwell platform. Vera Rubin is a cluster-level system consisting of five integrated racks, incorporating the Vera Rubin NVL72 system, Vera CPU, BlueField-4 networking and storage components, and the Spectrum-6 Ethernet rack.
Taiwanese server manufacturers and other supply chain partners are building systems based on Vera Rubin for AI labs, cloud service providers, and hyperscale operators. Vera Rubin is the third generation of NVIDIA's MGX rack-level system, with MGX design supported by hundreds of supply chain partners, including approximately 150 vendors in Taiwan and over 350 factories across 30 countries. System builders, infrastructure software providers, and storage companies producing Vera Rubin systems include Dell Technologies, HPE, Lenovo, Supermicro, AIC, Aivres, ASRock Rack, ASUS, Cloudian, Compal, DDN, Everpure, Foxconn, GIGABYTE, Hitachi Vantara, Hyve Solutions, IBM, Inventec, MinIO, MiTAC Computing, MSI, NetApp, Nutanix, Pegatron, Quanta Cloud Technology (QCT), VAST Data, WEKA, Wistron, and Wiwynn.
For scale-out and cross-layer scaling AI infrastructure, the Vera Rubin platform includes NVIDIA Spectrum-X Ethernet Photonics switches featuring co-packaged optics technology with 200Gb/s serializer/deserializer (SerDes), now in production. NVIDIA reports that Spectrum-X Ethernet Photonics delivers up to 5x better energy efficiency, up to 5x longer uptime, and 1.3x faster deployment compared to networks based on traditional transceivers. CoreWeave, Lambda, and Oracle Cloud Infrastructure are listed as early adopters. The Vera Rubin platform also includes the BlueField-4 DPU, supporting up to 800Gb/s software-defined networking and built-in multi-tenant isolation. The BlueField-4 advanced security trusted resource architecture is designed to simplify network operations, improve tenant isolation, and enhance control over large GPU clusters.
The Vera Rubin platform includes full-stack confidential computing, enabling trusted execution at the rack level. Vera Rubin NVL72 integrates the Vera CPU, Rubin GPU, NVLink networking, and security features into a single platform, encrypting high-speed interconnects. The platform includes hardware-level attestation to ensure system integrity. Cloud providers that have adopted confidential computing include CoreWeave, Firmus, GMI Cloud, IBM Cloud, IREN, Lambda, Microsoft Azure, Nebius, Nscale, SpaceXAI, and Vultr. The DOCA software platform is designed to secure the Vera Rubin rack and infrastructure layer, including protecting data, agents, context memory, and AI inference via BlueField-4 hardware. DOCA supports multi-tenant isolation, zero-trust policy enforcement, runtime threat detection, and end-to-end encryption at speeds up to 800Gb/s without using host CPU resources.
The DSX platform provides a design and operations framework for Vera Rubin-based AI infrastructure, designed to orchestrate system design, lifecycle management, and multi-tenant operations across the entire technology stack. The platform integrates reference designs, simulation, infrastructure software, facility guidance, and ecosystem technologies. Organizations adopting DSX for Vera Rubin deployments include Dell Technologies, HPE, Lenovo, Supermicro, ASUS, Foxconn, GIGABYTE, Pegatron, Quanta Cloud Technology (QCT), Wistron, and Wiwynn.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
