Dynatrace finds enterprise AI ambition is outpacing operational proof
Enterprise AI adoption is advancing faster than organizations can verify how their systems behave, according to Dynatrace research. While 87% of respondents use AI agents for observability and IT operations, only 27% currently address visibility into how those agents decide and act.
- 87% use AI agents for observability and IT operations.
- 44% target fully autonomous operations within five years, compared with 12% today.
- Only 27% currently address visibility into how agents decide and act.
- 81% report increased compute and infrastructure spending from AI workloads.
- 77% have moved OpenTelemetry beyond evaluation, but 20% use it as a strategic standard.
Autonomy plans expose an execution gap
Dynatrace’s State of Observability 2026 research finds that 44% of respondents are targeting fully autonomous operations within five years, compared with 12% who report reaching that level today. At the same time, humans verify an average of 66% of AI-powered decisions.
The company says the gap extends across nine AI observability capabilities, with current execution trailing perceived importance by an average of 14 percentage points. The areas covered include AI quality, agent decisions, security, cost and business impact.
Dynatrace distinguishes between observability of AI, which evaluates AI behavior, quality, security, cost, governance and business impact, and AI-powered observability, which uses causal analysis, assistance, automation and agents to improve operations and advance autonomy.
AI adds cost and complexity to technology estates
AI workloads can meet conventional availability, latency and error-rate targets while still producing inaccurate, unsafe, costly or otherwise unintended results. The research shows that 39% consider measuring the business impact of AI behavior important, but only 24% currently address it.
AI is also increasing the demands placed on infrastructure. According to the survey, 81% report higher compute and infrastructure spending from AI workloads, while 77% report a greater need for cost-allocation visibility. In addition, 43% identify the expansion of AI, LLM and agentic workloads as a leading source of technology complexity.
Dynatrace argues that AI performance needs to be assessed alongside the applications, infrastructure, security systems and services that support it. Connecting those elements can help organizations judge which workloads justify their resource consumption and where costs need tighter control.
Runtime evidence and open standards support oversight
Security is another area where the company says operational evidence matters. Fifty-five percent of security-qualified respondents experienced a zero-day incident in the previous year, while 45% say their organizations continuously monitor runtime environments for vulnerabilities that are exploitable or actively targeted.
Runtime context can show what is exploitable, what is affected and where remediation matters most, rather than relying only on static severity. The research also finds that integration complexity is the most frequently reported barrier to greater autonomy, at 31%, alongside skills, oversight, data quality and explainability.
OpenTelemetry adoption is broad but not yet consistently strategic: 77% have moved it beyond evaluation, while 20% have made it a strategic standard. Dynatrace commissioned the online survey of 1,202 technology leaders at enterprises across 12 countries in August 2026. The findings suggest users seeking greater AI autonomy will need evidence that follows agents from development through production and connects their actions to operational, security, economic and business outcomes.
Dynatrace: AI ambition is outpacing operational proof, Dynatrace research finds
