Trials and demos
How to buy

The Impact of AI on Wide Area Network Traffic

Unlike forecasts based on models alone, this report measures live AI inference traffic across service provider networks—revealing how AI and agentic AI are reshaping infrastructure.

Download full report

when tasks are performed by AI agents vs. humans

in AI token consumption

will be AI inference by 2035

AI traffic differs from web traffic

2x longer flow duration

AI inference flows last twice as long as regular web transactions. AI generates content token by token, resulting in longer, sustained connections that increase pressure on flow-aware network security systems like firewalls and intrusion detection.

10x larger median flow rate for web vs. AI traffic

AI inference flows have smoother, sustained throughput rather than bursty delivery. The token-by-token generation process acts like a traffic shaper, with median flow rates 10x larger for regular web transactions compared to AI inference flows.

9% upstream-heavy flows

AI inference introduces significantly more upstream traffic due to large, context-rich prompts. Approximately 9% of AI flows carry more upstream than downstream traffic versus only 0.5% for typical web traffic. This trend will intensify with agentic AI.

4x traffic growth in eight months

Real-world service provider data shows AI inference traffic growing 4x over just eight months. While AI traffic is still small in absolute terms, sustained high growth rates will make it a relevant portion of all traffic by 2035.

Enterprise agentic AI will impact wide area networks in new ways

In the absence of agentic AI, enterprise traffic would be expected to grow by about 250% over the 2026 to 2035 period, driven by digital transformation and cloud adoption.

With agentic AI adoption, enterprise traffic growth could reach 9x by 2035, driven by autonomous task execution and inference-heavy workflows. AI agents act as network "power users," operating at software speed.

By 2027, 80% of executives believe their company’s competitive survival will depend on agentic AI, according to a 2026 Cisco Omdia report, representing a fundamental shift in how traffic is generated and distributed.

In AI agent workflows, approximately 70% of traffic is AI inference. The connectivity between agent logic and AI models is the agent's "spinal cord"—a critical dependency whose degradation directly impairs the agent’s ability to operate.

Consumer AI adoption is driving internet-wide growth

6.6x traffic growth by 2035

Consumer adoption of AI and agentic AI is projected to drive growth in consumer-driven network traffic ~6.6×, representing 63% additional growth compared to 4x growth in non-AI scenarios in the same period—making AI the dominant driver of overall internet traffic expansion. 

61% using AI today

Surveys report 61% of U.S. adults have used AI in the past six months, with about 19% using AI daily. Globally, this equates to roughly 1.7 to 1.8 billion people who have accessed AI tools. 

46% of transactions by 2030

Cognizant and Oxford Economics forecast that AI-driven purchasing will account for approximately 46% of all consumer transactions in the U.S. by 2030, with similar figures in Germany (46%) and Australia (55%).

Near-universal usage by 2035

By 2035, AI usage is expected to approach 90% to 100% penetration of internet users. AI will become a default layer in devices and services—like broadband or electricity—largely invisible, yet indispensable.

The path to 2035: AI's growing impact on wide area network traffic

25% of traffic by 2035

AI inference traffic will grow from negligible today to 25% of total network traffic by 2035. This transformation will occur primarily between 2029 and 2032, when agentic AI adoption is projected to experience its most pronounced increase, with compound annual growth rate (CAGR) for traffic around 25% during that period.

Critical infrastructure planning needed

AI will not just increase traffic volume—it will change WAN traffic shape, symmetry, duration, and criticality. AI inference paths will become strategic network assets, requiring higher resilience, greater observability, and differentiated treatment, including quality of service (QoS) and path security.

"AI agents act as network 'power users', operating at software speed rather than human speed. The connectivity between agent logic and AI models is the agent's spinal cord—a critical dependency forging new paths of communication."

—Cisco AI Network Traffic Research Team, 2026

Prepare your network for the AI-driven future

Download the full report to explore detailed findings, methodology, and strategic recommendations for network operators as AI adoption accelerates through 2035.

Uncover key insights