AI Explodes Data: Are You Ready for the Cost?

Agentic AI is poised to cause a significant surge in telemetry data, a trend that many IT infrastructures are unprepared for. In the past 12 months, enterprises have witnessed their telemetry data volume more than triple, with AI and machine learning workloads accounting for a substantial portion of this growth. Projections indicate an average increase of 9.5 times in telemetry data from agentic AI workloads within the next two years, with some expecting growth up to 100 times.

This exponential data growth presents a substantial cost challenge. Currently, 69% of organizations actively engaged in agentic AI projects report that their observability costs have surpassed their compute and infrastructure expenses. The average annual enterprise observability spend stands at $3.17 million, with a year-over-year growth of 28%. These rising costs have already led 59% of organizations to terminate or delay agentic AI deployments, impacting critical functions like cybersecurity and fraud detection.

The underlying issue is architectural; traditional observability platforms struggle with the unpredictable data volumes and costs associated with autonomous AI systems. Many organizations are now implementing intelligent telemetry pipelines to manage, filter, and enrich data before it reaches costly backend systems. Enterprises with such pipelines are significantly better prepared for massive data growth and cost challenges, highlighting the pipeline as a foundational element for viable observability at agentic AI scale.