AI Chips & Infrastructure

Google Cloud Plans November Launch of Vera Rubin A5X AI Instances

NVIDIAOct 21, 2026

Google Cloud says its NVIDIA Vera Rubin-based A5X instances are slated for general availability next month, with European expansion planned.

Google Cloud plans to make its NVIDIA Vera Rubin-powered A5X instances generally available in November, the companies announced on October 21, 2026. The service is intended to support AI training, inference and large-scale reinforcement learning; expansion into Europe is also planned.

Each A5X instance combines NVIDIA Vera Rubin NVL72 systems and ConnectX-9 SuperNICs with Google Cloud’s Virgo Network and fourth-generation liquid-cooling technology. The design brings computing, networking and cooling together for demanding agentic inference and reinforcement-learning workloads. Google Cloud also plans to scale its Vera Rubin infrastructure across multiple sites, with capacity of up to 960,000 Rubin GPUs.

London-based Ineffable Intelligence, founded by AlphaGo creator David Silver, was the first customer to access A5X. The company is using the platform for large-scale reinforcement learning and work on systems that learn continuously from experience. That kind of development involves generating experience, evaluating outcomes and updating models, tying compute capacity closely to repeated training cycles.

WPP has also tested a creative-agent architecture on A5X, combining NVIDIA Nemotron models with OpenShell. In the demonstration, agents handled multi-step tasks such as generating, upscaling and restyling video, with human approval required before execution. NVIDIA says workloads that typically needed a multi-node cluster ran on a single A5X instance in WPP’s initial tests. OpenShell provides controlled sandboxes intended to limit agents’ access to sensitive brand assets and creative tools.

The planned launch gives cloud customers a route to use Vera Rubin systems without operating the underlying hardware themselves. The November general-availability timing and European expansion remain plans announced by the companies; the article does not give a specific launch day or European rollout schedule.

Industry Impact

Cloud access to Vera Rubin compute

Cloud infrastructure, AI development, creative production

Overview

The planned A5X launch gives AI teams a way to access Vera Rubin infrastructure through Google Cloud rather than operate the hardware themselves. For ML engineers and AI founders, the platform is aimed at compute-intensive training, inference and reinforcement learning, with early use by Ineffable Intelligence. WPP’s tests also show potential for multi-step creative-agent workflows. Business leaders should treat availability and European expansion as planned, not confirmed deployment dates; the article provides no pricing or independent performance benchmarks.

Key Impact Points

A5X targets intensive training, inference and reinforcement-learning workloads on Google Cloud.

Google Cloud says it plans to scale Vera Rubin infrastructure across multiple sites, with capacity of up to 960,000 Rubin GPUs.

WPP’s initial tests ran workloads that typically needed a multi-node cluster on a single A5X instance.

Opportunities

Managed compute for intensive AI workloads

Cloud services, reinforcement learning, creative production

Google Cloud’s planned A5X availability creates an option for organizations that need substantial AI compute but do not want to operate the underlying hardware. AI developers could assess it for repeated reinforcement-learning cycles, demanding inference, or creative-agent tasks such as video generation and restyling. Enterprises could also evaluate agent workflows with controlled access to brand assets and tools, as in WPP’s demonstration using OpenShell. The announcement supports evaluating these use cases, but does not provide pricing, service-level terms, benchmark methodology or a European rollout schedule. Buyers should compare availability and workload results before committing.

Cloud-hosted compute for advanced AI workloads
Problem: Organizations pursuing large-scale training, inference and reinforcement learning need substantial compute capacity and may not want to operate the underlying hardware.
Solution: Evaluate Google Cloud’s planned Vera Rubin-powered A5X instances for these workloads.
Target Users: AI developers, research teams and enterprises running compute-intensive AI workloads.
Business Value: Access to Vera Rubin systems through a cloud service may reduce the need to deploy and manage the hardware directly.
Revenue Potential: Google Cloud has potential to monetize customer access to A5X compute; the article provides no pricing or revenue estimates.

Reference link

https://blogs.nvidia.com/blog/google-cloud-vera-rubin-infrastructure

Image Source Link

https://blogs.nvidia.com/wp-content/uploads/2026/10/GoogleCloud-VeraRubin-1920x1080-1.png

Community Discussion (0)

Loading discussions...