When OpenAI publishes a report grounded in actual enterprise utilization, it’s value paying consideration. The information doesn’t simply predict the long run; it paperwork how as we speak’s enterprise networks are already being reshaped.
In The State of Enterprise AI (2025)OpenAI analyzes utilization throughout multiple million enterprise clients. The findings present a transparent inflection level: enterprise AI utilization has grown 8x yr over yrwhereas the usage of superior reasoning fashions has elevated greater than 300x. This alerts a elementary shift from easy prompts to complicated, multi-step, workflow-driven AI.
AI is not confined to pilots or innovation groups. It’s being embedded immediately into on a regular basis workflows, buyer interactions, and operational programs. The report’s crucial perception is about how AI is converging round particular, high-impact use circumstances which can be reshaping community necessities and elevating the bar for what enterprise networks—and IT groups—are anticipated to ship. Let’s study this sample and what it reveals.
How enterprise AI use circumstances are reshaping the community
As enterprises undertake AI throughout departments and workflows, the rising use circumstances are basically reworking community calls for, architectures, and the crucial enterprise position that networks play.
AI-powered buyer help turns the community into an expertise layer
AI-driven help is among the fastest-scaling enterprise use circumstances. Organizations are deploying AI brokers throughout chat, electronic mail, and real-time voice to resolve a rising share of interactions finish to finish.
Voice-based AI introduces steady, latency-sensitive site visitors, whereas backend integrations with buyer relationship administration (CRM), billing, and order programs generate persistent utility programming interface (API)-driven flows. As AI utilization scales, these interactions transfer from edge circumstances to core buyer journeys.
The community turns into a part of the client expertise. Inconsistent WAN efficiency or unstable cloud paths can degrade buyer satisfaction and improve stress on IT groups to diagnose points throughout voice, cloud inference, and backend programs.
AI-assisted software program improvement drives explosive east–west site visitors
AI is now embedded throughout the software program lifecycle—producing code, refactoring purposes, testing, and debugging. This exercise is increasing properly past conventional engineering groups, producing dense, steady east–west site visitors between builders, repositories, steady integration/steady deployment (CI/CD) pipelines, testing environments, and cloud inference companies. As reasoning-driven AI utilization grows, inner dependency chains turn into deeper and extra tightly coupled.
Networks optimized primarily for north–south site visitors wrestle right here. AI-assisted improvement will increase inner site visitors quantity, cross-domain dependencies, and troubleshooting complexity—typically requiring IT groups to cause throughout community materials, cloud connectivity, and utility pipelines concurrently.
AI-driven evaluation and analysis create bursty, cloud-heavy demand
Groups in finance, operations, and analysis and improvement (R&D) are utilizing AI to investigate datasets, synthesize analysis, and extract insights—compressing work that after took weeks into hours.
These workloads are bursty and cloud-heavytriggering massive information transfers and inference requests in brief home windows fairly than predictable patterns.
Networks should soak up sudden spikes with out degradation. Congestion or throttling delays crucial enterprise choices and will increase the burden on groups already working at capability.
Agentic AI workflows make the community a part of the execution path
Probably the most vital shifts recognized in OpenAI’s report is the rise of agentic workflows—multi-step AI programs that retrieve information, apply logic, take motion throughout programs, and confirm outcomes. These workflows span id companies, APIs, software-as-a-service (SaaS) platforms, and cloud inference endpoints—making the community a part of the execution path.
Agentic workflows introduce steady cross-system dependencies, develop the safety assault floor via machine identities, and require IT groups to troubleshoot failures spanning id, cloud, safety, and community domains. Any instability—latency spikes, dropped connections, or misrouted site visitors—can break the workflow chain.
AI-driven personalization places the community on the income path
Clever personalization engines form how enterprises interact clients—tailoring affords, suggestions, and experiences in actual time. The community is not simply supporting revenue-generating purposes—it’s immediately a part of the income path.
Efficiency degradation interprets into missed alternatives, whereas safety gaps improve enterprise danger. IT leaders are actually anticipated to ship velocity and safety concurrently.
Worker AI assistants create always-on, all over the place demand
AI assistants have gotten the entrance door to institutional data—supporting onboarding, troubleshooting, and each day productiveness throughout campuses, branches, and distant places.
Sustained, always-on AI site visitors compounds present collaboration and utility hundreds. Excessive-density wi-fi, dependable WAN connectivity, and constant safety enforcement are pushed tougher than ever—typically with no corresponding improve in IT employees.
Embedded AI turns the community into an integration material
As AI is embedded immediately into digital merchandise—search, diagnostics, automation—the community turns into the combination material, connecting customers, purposes, information, and inference.
Visitors patterns turn into steady and unpredictable, making it tougher to keep up efficiency, implement segmentation, and maintain visibility throughout domains. The community should perform as a unified integration layer connecting AI parts throughout each area—customers, purposes, information sources, and inference endpoints.
Enterprise networks—and IT groups—are struggling to scale AI
These use circumstances expose a rising hole. Many enterprise networks have been designed for human-driven interactions, predictable site visitors patterns, and guide operations. AI-driven environments introduce steady machine-to-machine site visitors, real-time efficiency expectations, and deeply interconnected programs.
This hole isn’t simply architectural—it’s operational. AI will increase operational complexity, expands the safety assault floor via new identities and integrations, and calls for abilities which can be more and more troublesome to rent and retain. AI works in pilots, however struggles at scale.
In lots of organizations, the know-how is transferring sooner than the working mannequin required to run AI reliably at scale.
Cisco helps shut the readiness hole
The structure behind the community issues greater than ever. That is the hole Cisco is filling with AI-Prepared Safe Community Structure—constructed to deal with the community as an execution platform for AIconnecting customers, purposes, information, inference, and automation with the efficiency, safety, and visibility AI calls for.
By design, it delivers:
- Infrastructure constructed for real-time, high-concurrency AI workloads
- Safety enforced inside the community material, not bolted on
- Deep telemetry and cross-domain intelligence (AgenticOps—autonomous operations at machine velocity) that reduces operational complexity and limits the safety blast radius so smaller IT groups can function AI-scale environments reliably
The purpose isn’t extra complexity. It’s easier operations with larger functionality.
What IT leaders ought to do subsequent
OpenAI’s enterprise information confirms AI is changing into foundational to enterprise operations. For IT leaders, this implies reassessing not simply purposes and information, however the community and working mannequin that underpins them.
As AI embeds itself into workflows, merchandise, and operations, the community turns into inseparable from AI success. Organizations that modernize for real-time efficiency, embedded safety, and autonomous operations will scale AI with confidence. Those who don’t will wrestle to maneuver past experimentation.
Within the AI period, the enterprise community doesn’t simply help the enterprise—it permits it.
Extra assets
