Artificial Intelligence (AI) promises a wide berth of opportunity, but not without a steep learning curve. Equinix today announced it’s partnering with Nvidia to address the challenges of deploying enterprise-grade AI – and eliminate the risks associated with getting started.

“What we've heard from customers at large is they’re very excited about the opportunity to deploy AI technology. But candidly, it's very difficult,” Equinix EVP and GM of Data Center Services Jon Lin told a group of reporters on Tuesday.

It’s no longer possible to house the technology stack and infrastructure needed to run and scale AI in an Enterprise data center – from a cost, sustainability and efficiency perspective. “That's just not feasible anymore,” Lin said.

Adding AI to data center infrastructure

“Customers want the world-class AI capabilities, but most of them don’t have the data center infrastructure [or] the expertise to build, manage and run those systems,” Nvidia VP of DGX systems Charlie Boyle said.

To that point, this joint offering provides AI infrastructure in a predictable and timely manner so customers can “get up and running with AI quickly,” Boyle said.

The foundation of Equinix Private AI with Nvidia DGX is the Nvidia DGX SuperPOD – a purpose-built architecture to create, train and run “the most common Enterprise workload,” large language models (LLMs), among others. When combined with Nvidia’s AI Enterprise software stack, “that’s a complete solution,” Boyle said. “Just add data, and get AI models.”

Equinix infrastructure for Nvidia DGX SuperPOD

The final piece is sophisticated and efficient data center architecture, which is where Equinix comes into play. By combining Nvidia’s “proven DGX SuperPOD” with Nvidia’s AI software stack and “the great data center and runtime expertise that Equinix has,” enterprises are getting a genuine turnkey solution, Lin touted. “They just tell us how big they want the solution to be.”

The “hidden gem” that is Equinix’s managed services team represents another integral piece of this joint offering. The management and support team will help Enterprise customers “efficiently, scalably and easily deploy AI technology in a way that otherwise would take them months or quarters worth of time to train their staff on the expertise required to manage the Nvidia DGX Superpod stack and handle the complicated networking requirements,” Lin said.

Equinix is an ideal data center partner for Nvidia because many enterprises looking into AI are Equinix customers and already have data stored in the vendor’s network.

Health care companies, for example, primarily store proprietary data, meaning those companies would likely rather build their own AI model than use a third-party model that might expose IP to outside eyes. If that data already resides in Equinix’s digital estate, transforming health care data into “impressive AI results” is possible in a matter of weeks, Lin said.

“Shrinking that lead time from months to weeks (or potentially days) in terms of that deployment cycle means at the end of the day, the customer can get value more quickly,” he said.

In addition, the nature of Equinix’s global footprint and multicloud networking capabilities support the two partners in their endeavor to simplify AI and make it more accessible for enterprises.

“We are already the intersection point for all of those data workloads” between public cloud environments, Software-as-a-Service providers and private infrastructure. “Now we can pull all of that data and workflows back into these AI deployments with the click of a button,” Lin said. “We’re incredibly excited about what this means for our customers.”