Blog | AppLogic Networks

The Impact of AI on Enterprise Networks

Written by John Morrell | Apr 16, 2025 1:01:16 PM

Over the past decade, the amount of data transmitted over our enterprise networks has dramatically increased consuming more and more of the network bandwidth.

The total amount of data generated each year has gone from 18 zettabytes (1 zettabyte = 1 trillion gigabytes) in 2016 to 149 zettabytes last year (2024), an 8x increase with a compound annual growth rate (CAGR) of over 30%.

A 2023 survey by IBM showed the average enterprise business generated approximately 2.5 exabytes of data themselves. A big contributor to this are connected IoT devices which are expected to generate a whopping 79.4 zettabytes of data in 2025 (44% of the 188 zettabytes total that will be generated) and will get transferred over some form of network.


Chart: Amount of Data Generated in Each Year, 2016, 2020 and 2024 actual and 2028 projected (Source: Statistica)

AI Exacerbates the Data Problem

Artificial Intelligence (AI) is now about to exacerbate the bandwidth availability problem. While annual bandwidth growth has historically averaged 20-30 percent per year, many networking experts are expecting AI to push the bandwidth growth rates to nearly 40 percent annually, with the main driver being data movement requirements.

The recent Databricks report: The State of Data and AI showed that 11x more AI models were put into production in 2024 versus 2023 and that 70% of companies use tools and vector databases with Retrieval Augmented Generation (RAG) to augment models to feed proprietary data into Large Language Models (LLMs).

With organizations using a multitude of models, this could lead to exabytes of training data. And models need to be continuously trained on fresh data, meaning the data needs to be constantly flowing. In addition, some portion of AI will be in real-time, coming from IoT devices or other streams of data. This means the “data in motion” needed to support real-time AI will be constantly flowing over your networks.

The Impact on Networks

As we have seen, the volume of data generated and moved on networks has already seen dramatic growth over the past decade and AI workloads will further increase the volume and need for data movement. This will have significant impacts on your enterprise network, namely:

  • AI will consume more bandwidth, increasing the demand and potentially starving other applications,
  • Many AI applications will require real-time or near-real-time processing, requiring low latency and network support for accelerating these applications.
  • Edge computing may be deployed meaning more traffic will move east-west (across your network) creating different bandwidth patterns.
  • The overall scalability of the network will need to grow to support greater bandwidth consumption.
  • The ebb and flow of AI workloads will constantly change and fluctuate bandwidth needs.


A 2024 study by PCCW showed that 69% of senior IT leaders believe their current network infrastructure did not have the capacity to fully embrace generative AI.

There Is a Solution

Operating and optimizing your network in the AI era will require a different approach that requires the network team to move away from a pure performance and scalability approach (bandwidth) to one that optimizes the Quality of Service (QoS) and Quality of Experience (QoE) for all network activity, whether it be user-oriented applications or data-oriented ones.

The key to succeeding in the AI era is comprehensive network observability, AI-powered network data enrichment and highly contextualized network optimization. Having each of these three items in a single solution, such as AppLogic Networks Enterprise Solution, provides a more seamless and adaptive manner to roll out AI workloads across your enterprise and gives your organization a competitive edge in how you use AI.

Our new paper, the Impact of AI on Enterprise Networks, examines the trends on data explosion and movement, explores how and where AI will create even more problems for enterprise networks and offers an approach to network optimization that will allow enterprises to better manage their networks through the AI era without completely tearing down their networks causing budgets to skyrocket.