Low Confidence in Network Data Is Slowing AI Operations
New research from Shamus McGillicuddy with Enterprise Management Associates reveals a troubling reality for teams operating AI data centers:
- Only 45% are confident in the accuracy of their network source of truth
- Only 36% completely trust the data from their observability tools
- Only 44% believe their network data is ready to support AI-powered and agentic operations
These numbers highlight a persistent and under-addressed problem. In AI infrastructure environments, intent data (how the network should operate) and state data (how it is actually operating) remain fragmented, incomplete, or difficult to trust.
The result? Slower changes, more tedious manual troubleshooting and configuration management, limited automation, and a clear barrier to more advanced AI-driven approaches.
The intent–state gap is the real issue
Modern network operations depend on two categories of data working together:
- Intent data: How the network is supposed to be designed and configured
- State data: How the network is actually performing right now
When teams lack full confidence in either, AI data center operations become far less efficient and far more error prone. As a result, engineers end up spending too much time trying to reconcile conflicting sources, and automation stays limited because no one fully trusts the data feeding it. This leaves no real path to agentic operations, which require clean, consistent, and authoritative data to act with confidence.
This is not a new problem, but the stakes have risen. As organizations move toward denser fabrics, hybrid environments, and AI-scale workloads, the cost of low-quality or untrusted network data grows. Visibility gaps widen, configuration drift risk increases, and the foundation required for intelligent operations remains brittle.
Elevating network data to an architectural priority
The EMA research makes it clear that IT leaders need to treat network data as a deliberate architectural concern rather than a byproduct of individual tools. This means consolidating intent into a reliable, single source of truth, improving the quality of infrastructure-wide observability data, and continuously validating one against the other.
The ultimate goal is an outcome-driven network (ODN). One that consistently delivers the operational results a business needs, rather than just managing configurations and devices.Organizations that take this approach create the foundation for higher confidence. Those that don’t continue to operate with partial trust, no matter how many tools they layer on.
Platforms built around a strong intent model are one practical way teams are moving toward this outcome-driven approach. By making the intended design the central reference point and keeping live state aligned with it, they reduce fragmentation and raise confidence in both the source of truth and the observability data that surrounds it.
Moving from diagnosis to progress
The EMA findings put hard numbers behind a problem many organizations feel today. Trust in the data that runs their AI data centers is incomplete. Those that close this confidence gap will be better positioned for reliable operations both now and in the future.
Treating network data as an architectural priority is the necessary first step. Real progress comes from systems that make intent authoritative and continuously keep the live network aligned with it.
If these challenges sound familiar, exploring an intent-based approach is a logical next step. Platforms such as Verity are built to accelerate this shift, helping teams establish a clearer source of truth and a more trustworthy operational picture.
Andrew Froehlich
Chief Marketing Officer
Andrew Froehlich has more than 25 years of experience in enterprise technology. He has served as a marketing director for several tech startups and provided marketing consulting to various companies. Combining deep technical knowledge with strong expertise in content strategy and messaging, he currently serves as Chief Marketing Officer at BE.