‘Infrastructure Constraints’ Around AI Spark Managed Services Opportunity: Cisco Partners

‘Just imagine what the world could do if we weren’t constrained on infra[structure] in AI. We would be 10 times as productive as we are right now,’ Jeetu Patel, Cisco’s president and chief product officer, says at Cisco Live 2025.

Infrastructure constraints are holding businesses back in the AI era, according to Jeetu Patel, Cisco’s president and chief product officer.

AI workloads require ultra-low latency and high throughput networks. Businesses need simpler, more secure and powerful infrastructure to support AI’s growth, Cisco executives said.

“Just imagine what the world could do if we weren’t constrained on infra[structure] in AI. We would be 10 times as productive as we are right now,” Patel said at Cisco Live 2025.

[Related: Cisco’s ‘Agentic AI Era’ Includes AI Canvas For Reimagined IT Operations]

According to research from Cisco’s 2025 AI briefing that was published earlier this month, 97 percent of IT leaders see modernized networks as critical to deploying AI, IoT, and cloud. Ninety-one percent of these leaders say they are increasing their network investments as a result.

Infrastructure constraints will stand in the way of customers when they are ready to deploy AI, said Steve Wylie, vice president and general manager, northeast, for Cisco partner Trace3.

“I always say to [customers] that if you’re not ready to launch into an AI use case, you should focus on your technical debt, because when you get to a place where you’re ready to focus on AI use cases that actually meaningfully impact your business, your infrastructure — if it’s end of life, or close to end of support — it won’t have the capacity, capability, speeds or storage that you need,” Wylie said.

The Irvine, Calif.-based solution provider has been helping customers ready their infrastructure for the time in which they are ready to make a big bet on AI. Uncertainty and budget constraints, however, are impacting infrastructure upgrades, Wylie said.

“The result of that [uncertainty] is now you have all this potential infrastructure that’s out there that is getting dated [and there are] security risks and not enough capacity or storage,” he said. “The biggest problem companies are having now is they’re still resistant to spend money.”

The answer can’t always be to overhaul IT environments with new infrastructure. Rather, partners can help customers “squeeze” the most out of their existing infrastructure with the help of some new Cisco products, said Craig Connors, vice president and chief technology officer of Cisco’s Infrastructure and Security Group.

Cisco’s approach to enhancing existing infrastructure includes the company’s 400G bidirectional (BiDi) optics, which lets customers easily transition to 400G networks for speed, while preserving their existing duplex multi-mode fiber infrastructure to keep costs reasonable, while also giving them infrastructure that can scale, Connors said.

“The new bidirectional fiber optics that we released allow you to double the speed on your existing fiber network. Obviously, it’s a huge advantage if I can take the existing wiring that’s in place and I can double the speed, because bandwidth constraints are real, and that’s something that dramatically reduces the cost and complexity of upgrading the speed of infrastructure,” he said.

The new optics will be available in the second half of calendar year 2025.

Security is another big infrastructure challenge for enterprises right now, Connors said. Cisco can also help customers integrate security seamlessly throughout their existing environments by way of its new smart switches, including the C9350 and C9610 Smart Switches that power campus networks that were announced at Cisco Live. The new generation of Cisco Smart Switches, powered by Silicon One, delivers up to 51.2 Tbps of throughput, below 5-microsecond latency and quantum-resistant secure networking to power high-stakes AI applications, he said.

The new offerings join the previously announced Nexus 9300 series of smart switches with embedded data processing units for the data center. Both the new switches and the previously introduced Nexus smart switches include Hypershield, the company’s AI-powered security architecture for data center protection, as a service.

Managed services, said Connors, will be a big opportunity for partners in helping customers work around infrastructure constraints to adopting AI.

“Just helping customers on this transition to AI is going to be a big journey in and of itself, not just selling new boxes,” he said.

Helping customers determine what AI use cases will be valuable to pursue is the first step. The second is a conversation around infrastructure, said Neil Anderson, vice president of cloud, infrastructure, and AI solutions for St. Louis-based Cisco partner WWT.

“Some of our customers are very mature and they know exactly what use cases they want to do. For them, it’s a where am I going to run it and how conversation,” he said. “[The use case] has to come first before you get to: how am I going to scale it? Where am I going to run it? And from there, we have options to talk to them about.”