Cisco Research Stresses AI's Enterprise Networking Challenge

The AI readiness conversation seems to have a massive blind spot, a recent Cisco and Foundry research has revealed.

Beyond infrastructure, GPUs and cyber threats, the report titled No time to wait: The accelerating impact of AI on campus and branch networks, discusses the “extraordinary pressure” AI is placing on enterprise networking.

Surveying more than 3,400 IT and networking decision-makers across 15 countries, the research sounds the alarm on the rapidly increasing strain on networks with rampant adoption of generative, agentic and physical AI.  

“This new research finds that campus and branch environments are fundamentally mismatched with the scale, speed and variability of AI-driven traffic now moving across enterprise networks,” the report notes. 

“As organisations accelerate AI deployment, networking teams are facing three pressures converging at once – rapidly escalating traffic growth, fast approaching capacity limitations and rising security complexity.”

AI’s impact on the network

Lateral device-to-device or server-to-server communication – which is referred to as east-west traffic – a crucial component of agentic AI communications, has risen up according to 67% of the participating respondents.

Continuous automated traffic generated by AI systems are up, says 61%.

This points to what Cisco president and Chief Product Officer, Jeetu Patel says, the industry has entered what he calls a “networking supercycle, because the network is so central to all the AI infrastructure the world is building now”.

Incoming capacity crunch 

Far from a hypothesis, the capacity crunch is already a problem today with over a third of respondents reporting increased capacity demands across their networks. 

AI-driven workloads would, according to 76% of those surveyed, require them to upgrade their workplace networks – also known as campus and branch networks. 

With only 30% of the aggressive AI adopters fully prepared to support the projected AI growth across the network, the research says that the remaining 70% may still be underestimating the network infrastructure challenge headed their way. 

The vast majority of IT decision-makers (93%) however recognise the issue and say they are accelerating modernisation initiatives. 

Case in point, a retail executive recalls how an AI tool supposed to analyse data to detect theft caused a delay due to network latency which undermined its usefulness. 

Network latency had caused “a delay of about five seconds. In those five seconds, somebody already leaves the store. So it’s pointless,” the US-based retail executive notes. 

The problem of AI trust

“AI workloads introduce a level of operational dynamism that many existing security approaches were not designed to manage,” the report says. 

The constant communication between AI systems, AI-enabled workflows that can trigger automatic actions across networking and shadow AI that evades the eyes of networking and security teams are all common issues associated with AI. 

This is why the report calls a secure network as “the most effective enforcement point for the unique security challenges AI brings”.

The risks are not unique to technical issues and seep into the operational component to become business risks. 

With network readiness paramount to scaling AI, more than 90% of respondents say they are aware of the competitive and financial risks that may arise if campus and branch networking isn’t suited to meet the demands of AI.

Risks such as operational disruption, degraded user experiences, longer response times, rising costs and reputational damage from outages or inconsistent policy enforcement across distributed environments are also concerns of respondents. 

Network strategies to be foundational for company success

With 85% of organisations expecting moderate to major expansion in agentic AI deployment within the next 24 months and 73% already facing or expecting to face campus and branch network capacity limitations, the window for modernising is shutting fast.

“Organisations that modernise now will be better positioned to support increasingly dynamic AI workloads, maintain operational performance, strengthen visibility and security and scale AI initiatives with confidence,” the report notes. 

“Those that delay modernisation risk allowing infrastructure limitations to become a direct constraint on innovation, execution and competitiveness.”