No matter the level, most network professionals feel the same way about AI. From networking architects to day-to-day engineers, network professionals experience the same struggles and successes with AI in network operations.
As the field of AI evolves rapidly, it's important to understand how network professionals are integrating the technology into operations. Mark Leary, research director of network analytics and automation at IDC, shared some of these insights from an April 2026 IDC report, based on responses from 516 respondents, during a session at ONUG's Spring 2026 AI Networking Summit.
This article outlines 10 insights about how network professionals implement AI in network operations in 2026.
1. Network barriers that impede AI success
In a previous IDC survey about AI project hindrances conducted around two years ago, Leary said that networking ranked at the very bottom of the list of impediments to AI success. At that time, AI projects were less common because the technology had not yet developed to a sufficient level.
"Now, as we've seen people moving from pilots to production, all of a sudden the network matters," Leary said.
Networking moved up the list in a survey conducted around 18 months ago. In IDC's most recent survey, it ranked fourth. As networking rises on the list, he said, staffing has dropped from the top three to the fifth-place spot. Organizations are less concerned with staffing issues as they continue to hire professionals and AI tools become easier to use.
Barriers to AI success specific to the network include the following:
- Network security concerns.
- Network automation challenges.
- Staffing shortages.
- Cloud connectivity limitations.
- Limited network visibility and control.
- Unaffordable network upgrades.
2. Platform vs. best-in-class
When IDC conducted the same study two years ago, two-thirds of respondents said they preferred platform approaches over best-in-class, Leary said. Top reasons for this preference include the following:
- Strengthened security.
- Reduced complexity.
- Lowered costs.
- Faster deployment.
- Easier integration.
However, interest in a platform-based approach has declined since that survey. Now that network professionals are more experienced and comfortable with AI, only 45% of respondents said they prefer integrated technologies from a single provider. A best-in-breed approach is now leading, with 55% preferring a mix of technologies from various providers.
3. Machine vs. human execution
Most network professionals prefer an AI-powered approach to network management, but a human-in-the-loop approach is gaining traction. An approximate 87% of respondents said they preferred AI-powered network management tools for remediation and optimization:
- 46% of respondents said they want tools that determine and execute actions.
- 41% of respondents said they want tools that guide actions, but don't execute them automatically.
Respondents reported using AI-powered automation for tasks such as threat response, configuration management and validation, and network problem diagnosis.
4. AI network management functions
Respondents reported using AI for several network management tasks, including the following:
- Network data collection.
- Threat detection and response.
- Agentic AI for network engineering and operations.
- Network automation.
- Network traffic prioritization and optimization.
5. AI for task automation by location
Most network professionals support using AI to automate network engineering and operational tasks. However, the percentage of tasks automated by AI is lower than expected.
In 2024, IDC found that the percentage of network tasks automated by network professionals managing campus and branch networks, as well as data center or cloud networks, averaged around 20%. However, this gap began to widen in 2025. While 31% of tasks were automated in campus and branch networks, only 23% of tasks were automated in data center and cloud networks.
An approximate 54% of respondents in campus or branch networks said they plan to automate tasks in 2027. By the same timeframe, only 29% on the data center and cloud side said they plan to automate tasks.
"[It] should start looking more like our expectations relative to campus and edge," Leary said. "Beware of stagnating in the data center and pushing forward on the campus and branch side of the fence."
6. Networking domains for AI innovation
Networking respondents reported several areas where they believe AI can improve the most. The top areas include the following:
- Cloud connectivity.
- Network security.
- Data center networking.
- Wireless WAN.
- IoT.
- Wireless LAN.
7. Current and future AI networking priorities
Most networking professionals currently use AI for tool integration, data security, and data collection and analysis. In the future, however, network professionals plan to use AI for the following capabilities:
- Improve AI security.
- Increase the number of autonomous capabilities of an AI system.
- Reduce cost and complexity.
- Accelerate innovation.
8. Agentic AI priorities
Respondents reported using agentic AI to automate several tasks, including the following:
- Network optimization.
- Security enforcement.
- Network data visualization.
- Troubleshooting.
- Configuration management.
- Compliance monitoring and management.
- Hardware and software deployment.
- Network design.
However, despite interest in agentic AI for networking, many network professionals remain apprehensive about the potential problems it could create.
"We've been very comfortable with the agent approach in networking forever," Leary said. "The difference from long ago is [that] agents were collectors, and they were passive. Today's agents, even in the network, are very active."
Top concerns with agentic AI include the following:
- Security.
- AI agent deployment and management complexities.
- Integration challenges.
- Costs.
- Network data constraints.
9. AI networking suppliers
Because AI in networking is so complex and critical for organizations, network professionals have certain expectations for supplier companies that partner with their teams. Examples of expectations include the following:
- Incorporate agentic AI into data systems.
- Prioritize security across AI deployments.
- Apply AI across the networking stack.
- Provide an ecosystem of AI technologies and partnerships.
- Offer AI training and consulting.
- Use AI within their own business and IT platforms.
10. AI in networking benefits
Despite AI being meant to support staff productivity, development and value, staff is often considered one of the biggest roadblocks to AI adoption, Leary said. Staff often push back on the use of AI within their systems, believing it will save their jobs. However, embracing AI could boost performance and productivity.
When network professionals embrace AI, it could lead to the following benefits:
- Increased productivity and teamwork.
- Strengthened security posture.
- Faster problem detection, diagnosis and direction.
- Increased business agility.
- Improved digital experience for employees and customers.
- Shift from reactive network management to proactive network management.
"It is undoubtedly a career-changing moment for the network engineer and network operator to embrace AI as opposed to repel against it," Leary said. "This defines success over the next 10 years for any network engineer or operator."
Deanna Darah is site editor for Informa TechTarget's SearchNetworking site.