Cisco retools branch networks for AI – News

As artificial intelligence reshapes the enterprise, the spotlight often falls on data centers. But Cisco Systems Inc. argues that the real frontier lies closer to the edge — in campus and branch networks. But Cisco Systems Inc. argues that the real frontier lies closer to the edge — in campus and branch networks.

While the data center will likely draw the most hype as enterprise AI matures, these branch environments — where Internet of Things devices, sensors and dynamic workloads intersect — will require networks capable of operating at machine speed, according to Grant Shirk (pictured), senior director of product marketing – networking experiences – at Cisco.

“If the data center investment is the smoke, it is the precursor to the fire that I think is going to hit campus and branch networks,” he said. “We’re seeing this across every single industry and company size that we talk to. There are a couple of things we have to prepare for. Certainly, we have to make sure that the networks themselves are built to the size, the scale and the rigor that’s required when we’re 10x-ing the workloads that are running across them.”

Shirk spoke with theCUBE’s Bob Laliberte at The Networking for AI Summit event, during an exclusive broadcast on theCUBE, News Media’s livestreaming studio. They discussed the clear mandate for IT leaders: Evolve campus and branch environments today to ensure they can sustain tomorrow’s AI-driven demands. (* Disclosure below.)

AI comes to the edge with branch networks

With workloads multiplying tenfold, enterprises need networks that can handle scale, security and complexity — without overwhelming operations teams. Cisco customers are already experiencing the strain. A global pharmaceutical company recently upgraded its capacity, only to find its AI models running at the network edge were already pushing beyond limits. For them, connectivity is directly tied to innovation and time-to-market for life-saving treatments, according to Shirk.

At the other end of the spectrum, a quick-service restaurant chain is embedding AI in drive-thru operations. Small language models are deployed at order windows to improve transcription, translation and accuracy — all while integrating with point-of-sale systems. The result: Faster service, higher revenue and new demands on branch networking, Shirk noted.

“We were talking to a drive-through restaurant, and for them, the drive-through number of vehicles you can push through your drive-through window in a day is directly comparable to your revenue,” Shirk said. “They’re battling staffing challenges. They’re battling market challenges. They’re starting to deploy small language models in the drive-through speaker to do transcription and translation.”

Cisco sees agentic operations as the future of network management. These systems don’t just automate repetitive tasks; they provide context, recommendations and collaborative workflows. Teams from networking, security and applications can jointly evaluate AI-driven insights, accelerating resolution and fostering trust.

“One of the core tenets of agentic ops, as we’re looking at it and AI campus, is this idea of multiplayer,” he said. “It is one thing to have a one-on-one exchange with an individual, a standalone agent. Sometimes it’s fun, sometimes it’s quirky. We really don’t want quirky in networking. The other power of this agentic model is that it learns alongside us, and we have the opportunity — and in many cases, the responsibility — to teach that trusted operating model into it, [to] coach it toward our organization’s standards and guidelines for how we operate.”

Here’s the complete video interview, part of News’s and theCUBE’s coverage of The Networking for AI Summit event:

(* Disclosure: TheCUBE is a paid media partner for The Networking for AI Summit event. Neither Cisco Systems Inc., the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or News.)

Photo: News