Google Cloud expands NVIDIA Vera Rubin infrastructure for AI training and inference

Traders Union
All News
Business & Wealth
Companies
Google expands NVIDIA AI infra

Highlights

Google Cloud will scale its Vera Rubin-based infrastructure to support up to 960,000 Rubin GPUs across multiple sites, targeting large-scale AI workloads.
Ineffable Intelligence is the first A5X customer, using the platform for reinforcement learning and superlearner system development, while WPP tested creative agent workloads on single A5X instances.
Google Cloud will be among the first to offer NVIDIA Confidential Computing at rack scale on bare metal, with Confidential G4 generally available for enhanced AI data privacy and enterprise security.

Cloud rollout and infrastructure plan

As reported by NVIDIA Newsroom, the A5X instances combine NVIDIA Vera Rubin NVL72 systems, NVIDIA ConnectX-9 SuperNICs, Google Cloud Virgo Network and Google Cloud's fourth-generation liquid-cooling technology in a single platform built for demanding AI workloads.

Google Cloud says the instances are designed to support training, inference and large-scale reinforcement learning in one environment. The company plans to scale its Vera Rubin-based infrastructure to as many as 960,000 Rubin GPUs across multiple sites, underscoring the size of its intended cloud AI buildout.

Ineffable Intelligence, a London-based company founded by AlphaGo creator David Silver, is the first customer to access A5X. The company is using the platform for large-scale reinforcement learning and for work on what it describes as superlearner systems that improve continuously through experience.

WPP has also demonstrated its creative agent reference architecture on Vera Rubin-based A5X instances using NVIDIA Nemotron models and NVIDIA OpenShell. In the company's initial tests, workloads that typically required a multi-node cluster ran on a single A5X instance.

European AI demand and security focus

The planned expansion into Europe positions the offering for regional innovators seeking infrastructure for advanced AI agents and reinforcement learning systems. Google Cloud and NVIDIA are targeting users that need high-performance systems capable of linking experience generation, outcome evaluation and model updates at scale.

The announcement also extends beyond raw computing capacity to data protection and operational controls. Google Cloud says it will be among the first cloud providers to offer NVIDIA Confidential Computing at rack scale on bare metal instances such as A4X Max, while Confidential Computing is also becoming generally available on G4 virtual machines powered by NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs.

Customers including Vertebrae, Polaris Cloud and Confidential Core AI are already using Confidential G4 for conversational privacy, verifiable agent execution and policy-governed enterprise intelligence. Google Cloud and NVIDIA are also exploring NVIDIA DSX MaxLPS capabilities for Vera Rubin-based infrastructure to improve performance per watt and expand capacity within existing data-center power budgets.

Our earlier article on rising costs and delays in U.S. AI data-center expansion highlighted growing doubts about how quickly new computing capacity can realistically come online. We noted that higher financing costs, project overruns, and power-grid constraints are forcing developers to prove they can deliver large-scale builds on time and within budget without eroding investor confidence.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.