Google Cloud scales Vera Rubin infrastructure to 960,000 GPUs

WPP: WPP is a global leader in advertising, marketing, and communications services. It conducted tests of AI-powered creative workflows on the new A5X instances using NVIDIA Nemotron and OpenShell tools. The tests showed that demanding creative tasks could be completed more efficiently on single instances compared to prior multi-node requirements. NVIDIA: NVIDIA designs and manufactures GPUs and full-stack AI computing platforms, including the Vera Rubin architecture optimized for training, inference, and agentic workloads. Through its collaboration with Google Cloud, NVIDIA is deploying NVL72 systems at massive scale while exploring power-efficiency technologies like DSX MaxLPS. Its hardware underpins both the new A5X instances and expanded confidential computing capabilities. Vera Rubin: Vera Rubin is NVIDIA's next-generation GPU platform built for demanding large-scale AI applications such as training, inference, autonomous agents, and reinforcement learning. It forms the core of Google Cloud's upcoming A5X instances and is targeted for deployment across hundreds of thousands of GPUs in multiple data centers. The platform incorporates advanced networking and liquid cooling for high-performance AI clusters. Google Cloud: Google Cloud is a major cloud computing platform that delivers infrastructure, networking, and AI services to enterprises and developers. It is partnering with NVIDIA to launch and scale new A5X instances powered by Vera Rubin systems, with initial availability in the coming month and expansion to Europe. The platform is also advancing confidential computing options across multiple NVIDIA GPU families. A5X Instances: A5X Instances are Google Cloud's new virtual machine offerings that integrate NVIDIA Vera Rubin NVL72 systems with ConnectX-9 SuperNICs, Virgo networking, and fourth-generation liquid cooling. They are scheduled for general availability next month and are designed to run complex AI workloads that previously required multi-node setups on a single instance. Early access has already been granted to select customers for specialized development. Ineffable Intelligence: Ineffable Intelligence is an AI company founded by David Silver, the creator of AlphaGo, focused on building advanced learning systems. It became the first customer to access Google Cloud's A5X instances, using them to develop 'superlearners' capable of continuous improvement through experience. The engagement demonstrates early real-world application of the new infrastructure for frontier AI research. Industry Testing: Advertising and creative sectors are validating the infrastructure through real-world workflow experiments with NVIDIA software tools. Security Enhancements: Confidential Computing is expanding across NVIDIA GPU-powered VMs to support secure AI processing at both instance and rack scales. AI Workload Optimization: The new A5X instances enable complex AI tasks such as training autonomous agents and reinforcement learning to run more efficiently than previous multi-node setups.