Advanced Computing in the Age of AI | Sunday, July 21, 2024

Nvidia AI LaunchPad Blasts Off, Promising ‘Instant AI’ to Get Enterprises Up and Running Quickly 

Putting together an AI infrastructure from scratch is far from an easy task for curious enterprises that are looking to bring AI into their operations and strategies. AI is complicated, expensive and often there are far more questions than answers when trying to figure it all out.

Recognizing those concerns from customers, Nvidia has just debuted its new AI LaunchPad offering that is designed to get companies using AI without having to build or configure any required infrastructure. AI LaunchPad includes all the needed Nvidia hardware and software to help customers get started with AI, as well cloud services through Equinix data centers, brought together in a seamless package.

Justin Boitano of Nvidia

“This offers instant access to AI machine learning infrastructure that is ready to use, located globally through our first program partner, Equinix,” Justin Boitano, the vice president and general manager for enterprise and edge computing at Nvidia, said in a recent briefing with reporters. AI LaunchPad is designed to remove previous barriers to AI infrastructure and make it easier for companies to deploy AI at scale, he said.

While Equinix is the first data center company to be named as a partner for AI LaunchPad, other partner vendors will also be unveiled in the future. Equinix will deliver the Nvidia products and services through hybrid clouds over its Platform Equinix infrastructure and 220 global data centers, according to Nvidia. The Equinix infrastructure can be deployed in minutes and is built to support AI systems from data center training and inference to full-scale deployment at the edge.

Using AI LaunchPad, customers will be able to run advanced AI workloads on Nvidia DGX SuperPODs managed by the Nvidia Base Command AI development hub. In addition, customers will also be able to use the Base Command platform to manage and deploy mainstream AI workloads on Nvidia-certified servers from a wide range of hardware manufacturers.

AI Launchpad is designed to provide an easier road map to AI for enterprises that need help bringing the still-evolving and complex technology into their operations, said Boitano.

“The beauty of the AI LaunchPad program is it gives you the equivalent of bare metal-as-a-service infrastructure instantly,” said Boitano. “Instead of enterprises having to buy servers … and come back in two or three months to get started, they can get started instantly, they can set up the infrastructure in minutes, and they can get those projects going rather than waiting to build the infrastructure themselves. This is an activation that is really going to help customers get started on this journey faster, show the value to internal stakeholders before trying to make bigger CapEx investments and help accelerate the entire cycle for us.”

Nvidia AI LaunchPad also integrates with Nvidia’s AI Enterprise cloud-native software suite, which includes AI and data analytics software that is optimized, supported and certified to run on VMware vSphere and Nvidia-certified hardware. The AI LaunchPad services will also integrate with Nvidia Ampere GPUs and Nvidia BlueField DPUs, according to the company.

The first AI LaunchPad offerings through Equinix are expected to arrive this summer.

Analysts Weigh In on Nvidia AI LaunchPad

Analyst Jack Gold

Jack E. Gold, the president and principal analyst at J. Gold Associates, LLC, said that Nvidia’s move to place AI LaunchPad into data centers with Equinix and other vendors will likely be a boon for AI adoption.

“Nvidia is making it much simpler for companies to access its AI products at near local speed instead of having to access products that are many miles or countries away,” said Gold. “This supplements any possible on-premises systems the companies may have. But truthfully, many companies have not ‘bulked up’ with high-performance Nvidia AI systems due to the major cost commitment, so having the opportunity to access such systems in a public, pay-as-you-go cloud environment can be a big boost to enterprise use of Nvidia AI systems, which is what they are hoping for.”

At the same time, Nvidia’s use of the tag line “instant AI” is a bit of a misnomer, said Gold.

“While Nvidia does have a good suite of AI tools, there really is no such thing as instant AI, since the value of any AI installation is dependent on how good the model is,” he said. “There are pre-build models, including from Nvidia, that get you a good part of the way there, but it is only with targeted models for your specific needs that you get maximum value out of AI.”

Overall, the announcement is a big deal for Nvidia and for customers who will not have to lay out large sums of CapEx money for infrastructure to use AI, said Gold. “And since VMware is also part of this announcement, many companies who are already on VMware vSphere will be able to easily transition to the Equinix systems in a hybrid cloud situation with known infrastructure tools. The real advantage of this move is lowering the barriers for enterprise AI adoption, and that is what Nvidia is doing here. That is a win-win situation.”

Analyst Dave Altavilla

Another analyst, Dave Altavilla of HotTech Vision and Analysis, told EnterpriseAI that the new AI LaunchPad offering is a compelling AI infrastructure-as-a-service play that will give small to large companies easier access to AI and Nvidia’s software and hardware, including DGX SuperPOD AI cloud supercomputing servers.

“Coupled with the global reach of Equinix data centers, AI LaunchPad will enable powerful AI training and inference resources at the fingertips of enterprises in a wide range of industries, running on Nvidia-certified systems and the tried-and-true VMWare vSphere cloud computing and virtualization platform.”

At the same time, customers who want to scale up on the as-a-service offerings can purchase custom configurations from their preferred OEM and Equinix will help them scale, said Altavilla.

"Most competitive cloud AI offerings currently in market often do not have the full-stack hardware and breadth of software solutions that Nvidia brings to the table,” he said. “Platforms like Nvidia AI LaunchPad are a necessary logical evolution that will afford enterprise customers the ability to deploy their own modern hybrid cloud solutions on powerful, custom-tuned Nvidia GPU-accelerated resources for a myriad of industries."

Nvidia Fleet Command Moves Out of Beta

Also announced by Nvidia is that its Fleet Command managed edge AI services are now generally available. Fleet Command is a SaaS platform that is used to deploy AI products and services on Nvidia-certified servers. Fleet Command’s remote management software lets businesses securely roll out and manage AI applications while maintaining the real-time processing of edge computing needs. Fleet Command lets users install, update and manage software from a central location. Fleet Command is part of the AI LaunchPad services.

Nvidia’s AI LaunchPad and Fleet Command announcements come just one day after the company unveiled the open beta release of its nascent Isaac Sim application to make it easier for businesses to build and test AI models before putting them into production.