Digital infrastructure firm Equinix hopes a brand new “AI-ready spine” will information prospects into the agentic AI period with a linked international platform and new software program layer.
The corporate mentioned its “distributed AI” infrastructure will unify greater than 270 knowledge facilities in 77 markets all over the world, providing elevated reliability for multi-cloud AI workloads. This can allow flexibility in accessing knowledge from a number of places, facilitating extra environment friendly AI coaching and inference.
Equinix will use its new software program layer, referred to as Cloth Intelligence, to supply real-time consciousness and automation on prime of its on-demand international interconnection platform.
Dhritiman Dasgupta, Equinix vp of selling, mentioned digital infrastructure must be tailor-made for AI workloads. “We’re principally seeing AI wants completely different infrastructure for every stage,” he advised DCN in an interview. “It turns into essential to attach all of those places the place these completely different phases of the AI purposes are processed. When this knowledge is distributed and the purposes are distributed, you want an structure that’s inherently distributed.”
A Scale-out Method in Search of ROI
Dasgupta mentioned the reliability of distributed structure will result in higher return on funding.
“We’ve seen prospects attempting to do inferencing within the cloud or in their very own knowledge facilities, and also you’re asking the infrastructure to do one thing unnatural, which doesn’t match what the appliance wants,” he mentioned. “With distributed structure, the ROI benefit is baked in.”
Dave McCarthy, analysis vp for cloud and edge companies at IDC, mentioned enterprises will see an growing want for distributed infrastructure because the AI increase continues to stretch sources all over the world.
“What’s most fun is that Equinix helps to tie collectively among the fragmented components of AI into one thing that may be extra complete for an enterprise,” McCarthy mentioned. “While you take a look at conventional workloads, plenty of that was cloud-based, and other people didn’t essentially care the place it lived, so long as it was accessible. However on this planet of AI, persons are completely going to care the place their purposes are operating and the place their knowledge is saved.”
McCarthy says the necessity to “scale-out” with distributed AI will grow to be extra obvious as corporations swap from AI coaching to inference. “Proper now, it’s all about scale-up with the coaching part of AI – that’s the place many of the consideration has been. It’s all about these massive, scale-up knowledge facilities,” he mentioned. “That kind of centralized, scale-up method received’t meet the wants of inference. You’re going to wish to deliver these AI purposes nearer to the place the customers are.”
An Schooling in Distributed AI
Equinix will complement its new choices with the corporate’s AI Options Lab, permitting prospects to check the brand new options. The lab will probably be accessible at 20 places globally.
“Equinix may give you these sandboxes to work with a few of these new options, letting you experiment and acquire the understanding and confidence to have the ability to preserve your personal AI techniques,” McCarthy mentioned. “And so they can do it in amenities which can be truly designed for it, versus attempting to determine easy methods to retrofit their very own environments.”
