Colocation giant Digital Realty deepened its ties to Nvidia with a service that allows enterprises to deploy Nvidia-powered artificial intelligence (AI) and machine learning workloads on Digital Realty’s data center platform.

Nvidia launched its DGX-Ready Data Center program last year with 19 data center partners including Digital Realty. The AI partner program gives customers access to Nvidia’s AI infrastructure inside the colocation providers’ facilities.

Meanwhile, Digital Realty in November announced PlatformDigital. At launch the data center platform offered customers four new services that they could deploy on top of PlatformDigital. This included Network Hub, which consolidates and localizes traffic into ingress and egress points. The companies’ announcement today essentially marries Nvidia’s DGX-Ready Data Center program with Digital Realty’s Data Hub.

Digital Realty developed a pre-configured Data Hub footprint based on typical customer deployment scenarios on Nvidia DGX configurations. It’s pre-certified though the DGX program in 24 global markets across three regions (North America, Europe, the Middle East, and Africa, and Asia-Pacific). The new Data Hub product, sold under the Nvidia DGX-Ready Data Center program, gives enterprises access to Nvidia’s AI infrastructure located inside Digital Realty’s colocation facilities. This puts the connectivity and data closer to users, cloud providers, networks, and devices, said Digital Realty CTO Chris Sharp.

“Data Hub, at the most basic level, allows efficient localization of data, staging, analytics, and steaming,” Sharp said. “AI and the work Nvidia has been doing around AI is absolutely paramount to the success of every customer in every vertical out there. And what’s interesting to us is making sure we can support the power densities associated with AI.”

AI workloads require higher power densities compared to other workloads, “but what’s often overlooked is the interconnection element of that,” he explained. “Any AI infrastructure out there is only as smart as the data sets you feed it.” And this is where Digital Realty’s global platform plays a role because it allows customers to access services, network providers, and public clouds across its data centers.

Using Nvidia’s AI infrastructure inside of Digital Realty’s colocation facilities saves customers’ money by eliminating the need to deploy AI workloads in their private data centers. And it also saves on cloud spending because “AI is often one of the most expensive services to rent in the public cloud,” Sharp said.

Because Digital Realty’s colocation facilities also house public cloud providers, enterprises can connect directly to their services, “so now you can really start to differentiate the analytics or that neural network or that machine learning array you are trying to stand up,” Sharp added.

As enterprises scale up their AI and machine learning deployments both in terms of the size and complexity of the machine learning model as well as the data that feeds the model, “If there’s a lot of distance between the cloud and where they are experimenting [with AI] and where their data sets are, they are going to incur a lot of cost and time,” said Tony Paikeday, director of product marketing for DGX systems at Nvidia. “We call this data gravity. This eliminates that time and distance.”