AI workloads are driving a 34% average increase in campus and branch network traffic, according to Cisco and Foundry's survey of over 3,400 IT and networking leaders across 15 countries. The surge stems from both generative AI tools and agentic AI systems, with nearly 70% of respondents reporting increased east-west traffic (device-to-device communication required for AI agents to exchange data) and 60% noting continuous automated traffic spikes. Organizations further along the AI adoption curve reported greater operational complexity and infrastructure challenges, whilst 90% of all respondents anticipate significant traffic increases from generative AI and 85% expect the same from agentic AI over the next 12 to 24 months. The research forecasts a 209% total workplace network traffic growth over three years driven by these AI categories.
The implications for CX teams are material. Network degradation directly threatens frontline worker productivity—support agents relying on AI-powered assistance for real-time customer data face delayed information access when infrastructure cannot sustain traffic loads, directly compromising resolution times and customer satisfaction. Over 70% of respondents acknowledged that their organisation's ability to meet customer expectations could be compromised by network constraints. This creates a critical dependency: as teams deploy more sophisticated AI agents and chatbots through platforms like Zendesk or Salesforce, the underlying network infrastructure must evolve in tandem, or performance gains from AI investment evaporate at the point of customer contact.
A second concern is visibility and control. The report reveals that operational teams lack centralised oversight of AI tool proliferation across departments, with shadow AI deployments creating unpredictable traffic patterns and security vulnerabilities. For CX leaders, this raises a pressing question: if your organisation cannot track which AI solutions are running on your network, how can you ensure consistent customer experience delivery or compliance with data governance requirements? The research suggests that without centralised governance of AI tools—both authorised and unauthorised—network performance will continue to degrade, making infrastructure investment a prerequisite rather than an optional enhancement for any CX operation scaling AI-driven support.
Generative and agentic AI solutions are increasing total campus and branch network traffic, and causing spikes, which could ultimately degrade CX.