AI Compute & GPU Cloud
GC

Google Cloud A5X (Vera Rubin)

A5X instances on NVIDIA Vera Rubin NVL72

VerifiedResearch confidence: HighGPU cloud instances
NVIDIA Vera Rubin NVL72Rack-scaleHyperscaler-managed
Launched: November 2026 (NVIDIA GTC Berlin)Last verified: November 20, 2026

At a glance

Category
AI Compute & GPU Cloud
Product type
GPU cloud instances
Status
Generally available
Launch
November 2026 (NVIDIA GTC Berlin)
Pricing
Not publicly verified
Platform
NVIDIA Vera Rubin NVL72
Integration
GKE / Google Cloud

Overview

Google Cloud announced general availability of A5X instances built on the NVIDIA Vera Rubin NVL72 platform, a new tier of accelerated compute for large-scale AI training and inference.

Why it matters

A5X brings NVIDIA's newest Vera Rubin generation to a hyperscaler as a GA product — the edition's most consequential signal that frontier-scale compute is becoming routinely provisionable.

Problem it solves

Frontier AI workloads need access to the newest accelerators at scale; A5X brings the Vera Rubin generation to a hyperscaler with managed networking and storage.

How it works

Teams provision A5X instances on Google Cloud and run training/inference on the NVL72 rack-scale configuration through Google's AI infrastructure stack.

Key capabilities

  • NVIDIA Vera Rubin NVL72 platform
  • Rack-scale accelerated compute
  • Managed networking & storage
  • Generally available

Integrations

Google CloudNVIDIA GPU stackKubernetes (GKE)

Pricing & plans

Not independently verified in this research pass — see the official pricing page for current plans.

View official pricing ↗

Use cases

  • Large model training
  • High-throughput inference
  • AI infrastructure at scale

Who should use it

  • • Teams training or serving large models at scale
  • • Enterprises already on Google Cloud wanting newest accelerators

May not suit

  • • Cost-sensitive hobby or small-scale workloads
  • • Teams needing single on-demand GPUs cheaply

Independent editorial analysis

Strengths & limitations

Strengths

  • +Access to the latest Vera Rubin NVL72 rack-scale platform
  • +Managed networking, storage and GKE integration
  • +Hyperscaler reliability and global regions

Limitations

  • –Hyperscaler GPU pricing is typically higher than specialist neoclouds
  • –Public per-hour pricing for A5X not verified in this research pass

Editorial assessment by aitoolslist.ai, distinct from vendor claims.

Alternatives & comparison

OptionPositioningWhen to prefer
Google Cloud A5X (Vera Rubin)A5X instances on NVIDIA Vera Rubin NVL72Featured pick this edition
CoreWeaveGPU-specialist neocloudWhen you want faster provisioning and lower GPU cost than a hyperscaler.
LambdaDeveloper-friendly GPU cloudFor simpler on-demand GPU access and training clusters.
NebiusAI-focused cloudFor managed AI infrastructure outside the big-three hyperscalers.

Comparison reflects positioning; pricing and spec parity are not independently verified across all options.

Edition intelligence · AI Compute & GPU Cloud

Why it was selected this edition

The edition's most significant GPU-cloud event: GA of the Vera Rubin generation on a hyperscaler, announced at NVIDIA GTC Berlin 2026.

Category rank: #1 this editionResearch confidence: HighLaunch relevance: November 2026 (NVIDIA GTC Berlin)

Claim-level transparency

Research & verification

Our verification scope is explicit. A checked website does not mean the whole product has been independently audited.

Official websiteConfirmedhttps://cloud.google.com/blog/topics/ai-infrastructure/google-cloud-infrastructure-at-nvidia-gtc-berlin-2026
Launch / announcementConfirmed
Product capabilitiesCross-checkedFrom cited sources
PricingUnverifiedSee official pricing page
Performance benchmarksUnavailable