Google Cloud A5X availability is planned for November, NVIDIA says

Google Cloud’s A5X instances are scheduled for general availability in November 2026, according to [NVIDIA’s October 21 announcement](https://blogs.nvidia.com/blog/google‑cloud‑vera‑rubin‑infrastructure/). The article’s body says availability is “coming next month” and describes expansion to Europe as planned. It gives no exact launch day or European rollout date. For teams planning an AI workload, those are useful boundaries: an announced general‑availability window, an early customer’s access and availability in a particular cloud Region are different facts. The post does not establish that every customer can provision A5X today.

[NVIDIA describes A5X](https://blogs.nvidia.com/blog/google‑cloud‑vera‑rubin‑infrastructure/) as combining Vera Rubin NVL72, ConnectX‑9 SuperNICs, Google Cloud’s Virgo Network and fourth‑generation advanced liquid cooling. It places the instances within Google Cloud’s AI Hypercomputer for training, inference and large‑scale reinforcement learning. The design brings several infrastructure layers together. A workload plan still needs the service’s actual instance configuration, software requirements and available capacity. This announcement does not provide customer pricing, quotas or a complete Region list, so those details cannot be inferred from the hardware names.

NVIDIA identifies Ineffable Intelligence, founded by David Silver, as the first customer to access A5X. It also describes WPP demonstrating a creative‑agent reference architecture on A5X using Nemotron models and OpenShell, with human approval before execution. Both are [examples reported by NVIDIA](https://blogs.nvidia.com/blog/google‑cloud‑vera‑rubin‑infrastructure/). They illustrate different intended uses: reinforcement‑learning work and multistep creative workflows. They do not establish generally available access for another organization, or a comparable performance result for its workload. No independent A5X test is reported here.

The company says Google Cloud plans to expand Vera Rubin infrastructure across multiple sites. [The same post](https://blogs.nvidia.com/blog/google‑cloud‑vera‑rubin‑infrastructure/) also discusses separate Confidential Computing offerings and efforts to optimize infrastructure within power budgets. Those announcements have their own product and availability boundaries; they should not be read as one universal A5X entitlement. The next practical checkpoint for a buyer is the current Google Cloud service documentation and account‑specific availability when the stated window arrives. Until an exact Region, configuration, quota and price are confirmed, November remains the announced general‑availability window rather than a completed deployment commitment for that buyer. *Photo: Lambtron / [Wikimedia Commons](https://commons.wikimedia.org/wiki/File:Google_data_center.jpg), [CC BY‑SA 4.0](https://creativecommons.org/licenses/by‑sa/4.0/). Google data center in The Dalles, Oregon, June 2020; context photograph, not evidence that A5X or Vera Rubin operates at this site. Native display crop and downscale; image adaptations retain the same license.*