Google Cloud has opened its A5X computing systems, built on Nvidia’s new Vera Rubin chips, to customers. The systems became generally available Friday, Nov. 6, so companies can now rent them to train and run AI models.
Google Cloud says in its announcement that A5X instances are bare-metal machines, meaning customers get the physical servers rather than a shared slice. They run on Nvidia’s Vera Rubin NVL72 platform, a rack-sized system that works as one.
Built for AI agents
Google says the systems are aimed at “agentic” AI, software that carries out long, multi-step jobs on its own. Companies running those jobs need to move huge amounts of data quickly between chips and storage, and Google says A5X is designed around that. It pairs the machines with its Kubernetes tools for scheduling large AI jobs across many servers.
A customer’s view
Stephan Pretorius, chief technology officer of advertising group WPP, said the systems let its creative and production teams “direct intelligent agents that perform frame-accurate quality checks and adapt brand assets seamlessly.”
In plain words: Nvidia’s newest AI machines are now for rent on Google’s cloud. Companies that want to build or run big AI programs no longer have to buy their own.
JBizNews Desk | Mountain View, California
© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.
