Most enterprises expect their campus and branch networks to reach capacity limits as AI workloads multiply, according to a new Cisco report featuring research by Foundry. The study found that 73% of organizations already faced or expect to face capacity limitations across these workplace networks within the next 24 months.
The Foundry survey gathered responses from more than 3,400 IT and networking decision-makers across 15 countries, all at organizations with at least 500 employees.
The report found that AI-driven traffic on campus and branch networks rose 34% on average over the prior 12 months, and respondents expect an additional 96% increase within the next year. Over three years, they projected that traffic would roughly triple as generative, agentic, and physical AI are increasingly utilized.
The report pointed to a gap between AI ambition and network readiness. Cisco’s 2025 AI Readiness Index found that only 15% of organizations had networks flexible enough to support AI at the necessary scale. Even among aggressive AI adopters — defined as those with broad generative AI deployment — just 30% said they were fully prepared to support projected growth.
Changing traffic patterns drove much of the strain. The report said that respondents described the following shifts tied to AI workloads:
- 67% saw growth in east-west traffic, the lateral device-to-device communication that AI agents use to exchange data.
- 64% reported more real-time, latency-sensitive traffic.
- 61% noted increases in continuous automated traffic generated by AI systems.
- 49% observed more bursty or spike-driven traffic.
Security ranked as the leading concern. The report found that 77% of respondents said AI had already expanded their attack surface over the prior year, and 78% expect security risks to grow as adoption moves beyond generative use cases. While 86% had added security controls for AI workloads, 61% held back from scaling AI further until they gained more confidence in their security posture.
The business stakes were also clear. More than 90% of respondents recognized financial and competitive risks if they failed to adapt their networks, and 75% agreed that reactive upgrades carried higher long-term costs. Among IT leaders, 93% said they were accelerating modernization in response to AI demand.
The report added to a broader picture of AI reshaping demand and economics across the telecom industry. A separate forecast from consumer intelligence firm NIQ projected that an AI-driven upgrade cycle could fuel 8% telecom growth in 2026, as on-device intelligence and agentic features moved from add-ons toward core differentiators.
