News
Google Cloud and NVIDIA said Vera Rubin-based A5X instances will reach general availability next month, with Europe expansion planned.
Google Cloud and NVIDIA said Vera Rubin-based A5X instances will reach general availability next month, extending access to a rack-scale AI platform for training, inference and large-scale reinforcement learning. The announcement, published by NVIDIA on October 21, 2026, also said availability is planned to expand to Europe.
Evidence and context
According to NVIDIA, A5X instances combine NVIDIA Vera Rubin NVL72, NVIDIA ConnectX-9 SuperNICs, Google Cloud Virgo Network and Google Cloud’s fourth-generation advanced liquid-cooling technology. The companies described the platform as part of Google Cloud’s AI Hypercomputer and said it is designed for demanding agentic inference and reinforcement learning workloads at scale.
NVIDIA also said Google Cloud plans to scale its Vera Rubin-based infrastructure to as many as 960,000 Rubin GPUs across multiple sites. That figure was presented as a planning target in the announcement, not as a current deployment level.
The first named customer to access A5X was Ineffable Intelligence, a London-based technology company founded by AlphaGo creator David Silver. NVIDIA said Ineffable is using the platform to advance large-scale reinforcement learning and develop what it calls “superlearner” systems that learn continuously from experience. In a quoted statement included in the release, Silver said Vera Rubin’s compute performance and high-bandwidth connectivity let the company push the scale and ambition of its reinforcement learning research further.
NVIDIA also said WPP has demonstrated its creative agent reference architecture on Vera Rubin-based A5X instances using NVIDIA Nemotron models and NVIDIA OpenShell. In NVIDIA’s description, WPP’s agents convert conversational requests into multistep creative workflows, including video generation, upscaling and restyling, with human approval before execution. NVIDIA said WPP’s initial tests ran workloads that previously required a multi-node cluster on a single A5X instance.
The announcement comes after related Vera Rubin milestones reported in recent weeks. Supermicro said on September 23, 2026, that it was now shipping NVIDIA Vera Rubin NVL72 racks, while NVIDIA said on September 16, 2026, that Vera Rubin NVL72 delivered leading performance in MLPerf Inference v6.1 debut entries, which NVIDIA identified as preview submissions. Those earlier disclosures provide context for the product’s rollout, but they are separate from Google Cloud’s general-availability announcement.
NVIDIA further said Google Cloud will be among the first cloud providers to offer NVIDIA Confidential Computing at rack scale on bare metal instances such as A4X Max instances powered by NVIDIA GB300 NVL72. It also said Confidential Computing is generally available on G4 VMs powered by NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, for confidential inference and fine-tuning. The company named Vertebrae, Polaris Cloud and Confidential Core AI as users of Confidential G4 for privacy, verifiable agent execution and policy-governed enterprise intelligence.
Separately, NVIDIA said Google Cloud and NVIDIA are exploring NVIDIA DSX MaxLPS capabilities for Vera Rubin-based infrastructure to help optimize performance per watt and expand capacity within available data-center power budgets. The announcement did not provide a date for that exploration or any deployment timing beyond next month’s general availability and planned Europe expansion.
Current published Vera Rubin NVL72 specifications describe a rack-scale system with 72 Rubin GPUs and 36 Vera CPUs, along with ConnectX-9 SuperNICs and BlueField-4 DPUs. NVIDIA says the rack provides 20.7 TB of HBM4 GPU memory, about 1,400 TB/s of aggregate GPU-memory bandwidth and 216 TB/s of aggregate NVLink bandwidth. Those specifications are technical platform figures, not measured application benchmarks, and they are separate from the general-availability timeline described in the Google Cloud announcement.
Be the first to comment