Nvidia Corp. is catering to customers that need to rapidly scale up their artificial intelligence deployments by providing access to its AI infrastructure in the cloud and at the network edge.

Announced today, the new Nvidia AI LaunchPad program provides customers with instant access to the entire spectrum of Nvidia’s AI computing resources, enabling them to train and deploy models in any location.

The company is working with Equinix Inc. to deploy its AI infrastructure in multiple data centers across the world. Through the Nvidia AI LaunchPad program, customers will be able to access Nvidia’s DGX SuperPod servers managed by the Base Command Platform service to develop advanced AI models, Nvidia said.

In addition, customers can also use Base Command Platform to develop more conventional AI on Nvidia-certified systems built by the likes of Dell Technologies Inc. and Lenovo Group, all located in Equinix’s data center regions.

The LaunchPad program also provides access to Nvidia’s AI Enterprise software suite on VMware Inc.’s vSphere, enabling customers to deploy an “AI-ready platform” that can scale workloads across hybrid clouds.

In addition, customers will be able to deploy, manage and run their AI applications at Equinix’s edge locations via the new Nvidia Fleet Command managed cloud service that was also announced today and is now generally available. That service, Nvidia said, is meant for AI apps that need extremely low latency connections to users.

Manuvir Das, Nvidia’s head of enterprise computing, said the LaunchPad service caters to enterprises’ that seek a simple yet comprehensive solution that provides instant access to AI resources as and when they need them. “Nvidia AI LaunchPad puts AI at the fingertips of enterprises everywhere with fully automated, hybrid-cloud infrastructure and software for every stage of the AI lifecycle,” he said.

The company explained that enterprises today need to be able to access multiple external data sources spread across various public clouds and private enterprise data centers to build more accurate AI models. Nvidia said that by hosting its AI infrastructure in Equinix’s cloud, it enables extremely rapid access to those data sources, helping to satisfy the real-time performance requirements of many applications.

The central ingredient of Nvidia AI LaunchPad is its cloud-native AI Enterprise suite of AI software, tools and frameworks. With Nvidia AI Enterprise and VMware vSphere, the company said it’s possible to deploy AI models in Equinix’s data centers with bare metal performance across multiple nodes. As such, it can scale up as necessary to support the largest and most complex AI training jobs and workloads, Nvidia said.

“It’s time to democratize AI across every industry,” Justin Boitano, vice president and general manager of enterprise and edge computing at Nvidia, said in a press briefing. “Nvidia has optimized and created the end-to-end workflow for AI.”

Analyst Holger Mueller of Constellation Research Inc. told SiliconANGLE that if Nvidia wants to remain a leader in AI platforms then it needs to provide them not only on-premises and through opportunistic cloud vendor partnerships, but on a continuous basis too.

“It’s also key that Nvidia operates and maintains these platforms so enterprises can focus on their AI-powered applications and not worry about the intricacies of platform plumbing,” he said. “Nvidia is taking care of that, starting with Equinix, a key player in the cloud data center provider space. What will matter is which of the big three clouds will be the first to support Nvidia’s platform, because AI needs data. Data gravity is real and data is agglomerating on the three large cloud vendors’ platforms.”

Nvidia said the first AI LaunchPad offerings from Equinix’s data centers will be available in the summer. Boitano said the company is still working with Equinix to work out the details, and that the service will be priced on a “dollars per hour” structure with more precise pricing to come later.

With reporting from Robert Hof

Image: Nvidia