Dynatrace released findings from its new research, The State of Log Management 2026 report, revealing that the rapid growth of AI workloads is pushing traditional log management approaches to their limits. Modern logs have become critical to understanding, validating, and securing AI-driven decisions, helping organisations ensure reliability, compliance, and performance at scale. However, the volume and complexity of AI telemetry are overwhelming legacy tools, making it harder for teams to keep AI systems explainable, trustworthy, and production ready. As a result, enterprises must rethink how they manage and analyse telemetry data to maintain visibility, control costs, and support AI at scale.
Key findings from the report include:
- AI workloads have driven a 93% increase in log volume over the last 12 months.
- Organisations use an average of seven different tools to manage logs and telemetry.
- 80% say turning telemetry into actionable insights is negatively impacting customer experience and delaying AI initiatives.
- Organisations exclude an average of 86% of log data to manage costs and system limitations.
- Teams spend nearly $2.5 million annually on logging solutions.
- Nearly three-quarters say AI workloads require a platform-based approach to log management.
- 81% believe log ingestion and processing must be open and automated for real-time analysis.
According to a global study of 450 senior technology leaders, this surge in data, combined with fragmented tools, is making it increasingly difficult for teams to detect issues, secure AI systems, and extract timely insights. Organisations are forced into manual, time-consuming workflows as they compare insights across systems, slowing time to insight and limiting their ability to move AI initiatives from pilot to production.
AI growth pushes traditional log management to breaking point
Respondents estimate they spend an average of nearly $2.5 million annually on logging solutions, including log ingestion, management, storage, indexing, rehydration, and querying. At the same time, logs are a key component for understanding and securing AI systems. To manage rising costs and system limitations using traditional methods, many organisations are forced to limit the amount of telemetry they ingest or retain.
Nearly half of organisations report discarding or not collecting logs, excluding an average of 86% of log data from ingestion, storage, or analysis to manage cost and system limitations. These challenges are most pronounced in environments that rely on fragmented or log‑centric approaches, rather than a unified observability platform designed to handle AI‑scale telemetry.
“AI is accelerating enterprise innovation, but most logging systems were never built for the scale, speed, or complexity of AI‑driven environments,” said Mala Pillutla, Vice President of Log Management at Dynatrace. “As AI agents operate probabilistically, treating logs, metrics, traces, and events as separate signals is no longer viable. To make AI systems reliable and trustworthy, organisations need a unified, intelligent approach that brings all telemetry together in real time, enriched with deep context to drive confident decisions.”
As AI initiatives move from experimentation to production, fragmented log management from too many tools is emerging as a key barrier to reliability, trust, and operational scale.
Unified observability becomes essential to scaling AI workloads
The report underscores the need for a fundamentally new approach to log management, where logs serve as the high-fidelity foundation, unified with distributed tracing and other telemetry data to deliver real-time, context-rich insights at a massive scale.
Nearly three‑quarters of respondents say AI workloads now demand a platform‑based approach to log management, while 81% believe log ingestion and processing must be open and automated to enable real‑time analysis without rigid schemas, indexing overhead, or rehydration delays.
The real cost of observability fragmentation isn’t just the infrastructure bill — it’s the opportunity cost of AI initiatives that stall between pilot and production because teams can’t trust their telemetry. The research shows that roughly a third of organisations are paying for redundant or underutilised observability features, and more than a quarter are burning engineering cycles just keeping multiple tools running across environments. That’s capacity that should be going toward making AI workloads production-ready, not toward stitching together dashboards across numerous different tools.
Discussion about this post