Nvidia Corp. is updating its AI Enterprise software platform, adding production support for containerized artificial intelligence workloads on VMware vSphere with Tanzu.

Today’s update means enterprises can use Nvidia AI Enterprise 1.1 to run accelerated AI workloads on vSphere in both Kubernetes containers and virtual machines.

The platform is a suite of AI tools and frameworks that aims to be a turnkey solution for enterprise looking to run AI workloads in their own data centers or in the cloud. Launched last year, it makes it possible for companies to virtualize AI workloads and run them on Nvidia-certified server systems.

The benefit is that those workloads can be managed through a single platform. Companies also can deploy AI-ready infrastructure closer to where their data resides, meaning faster training times.

Nvidia said one of the top requests from its customers was production support for running AI workloads on VMware vSphere with Tanzu, which is a service that makes it possible to deploy AI on both containers and VMs within a vSphere environment. With today’s update, customers gain an integrated, complete stack of containerized software and hardware that’s fully managed and optimized for AI, Nvidia said.

AI Enterprise can be run locally on customer’s own servers or alternatively as-a-service on bare-metal infrastructure from the data center provider Equinix Inc., at nine locations globally.

The company said VMware vSphere with Tanzu support will soon be added to the Nvidia LaunchPad program for AI Enterprise, meaning customers will be able to test and prototype new AI jobs free of charge at curated labs hosted by Equinix and designed for AI practitioners. Nvidia said the point of the labs is to showcase how common AI workloads such as chatbots and recommendation engines can be developed and managed using its platform on VMware vSphere with Tanzu.

Customers can alternatively get straight down to business and deploy AI Enterprise on certified servers from various partners, including Dell Technologies Inc., Hewlett Packard Enterprise Co., Inspur Inc., Lenovo Group Ltd., Gigabyte Technology Co. Ltd. and Super Micro Computer Inc.

Enterprises will have a couple of new third-party server options too, as Cisco Systems Inc. and Hitachi Vantara Ltd. also took the opportunity to announce their first Nvidia-certified systems today. Cisco’s UCS C240 M6 rack server is powered by Nvidia’s A100 Tensor Core graphics processing units and designed to power a range of storage- and power-intensive applications, including big data analytics, databases, virtualization and high-performance computing. Hitachi’s Advanced Server DS220 G2 with Nvidia A100 Tensor Core GPUs is more of a general-purpose server, optimized for performance and capacity to deliver a balance of compute and storage.

Finally, the release of Nvidia AI Enterprise 1.1 also adds support for Domino Data Lab Inc.’s Enterprise MLOps Platform with VMware vSphere with Tanzu. Nvidia said the integration will allow companies to scale up their data science initiatives cost-effectively by accelerating research, model development and, finally, deployment on mainstream GPU servers.

International Data Corp. analyst Gary Chen said most AI practitioners prefer to deploy workloads in application containers if they can. The problem is that doing so is extraordinarily complex, requiring enablement at multiple layers of the stack, including the AI software framework, operating system, virtual machine and even the hardware.

“Turnkey, full-stack AI solutions can greatly simplify deployment and make AI more accessible within the enterprise,” Chen said.

Images: Nvidia