AI Agent Costs: Telemetry Pipelines Cut Soaring Expenses

Many companies are struggling with rising telemetry costs for AI agents. In fact, 59% of organizations have already terminated or delayed an agentic AI deployment due to these monitoring expenses. Here's the thing: AI agents are much harder to monitor and predict costs for compared to traditional applications. This is because they are non-deterministic, iterative, and generate data at machine speed. Andi Mann, chief product and technology officer at Apica, notes that AI projects are "cannibalizing typical budgets." Enterprises now report spending an average of $3.17 million on observability, and this figure is growing 28% year over year. What's interesting is that most enterprises have seen telemetry volume triple in the past year, with 43% of that growth coming from AI/ML workloads. This makes AI observability a top priority for 83% of businesses in the year ahead. The bottom line: Uncontrolled monitoring costs can lead to the cancellation of important AI programs, impacting other critical business functions.