Cisco announced the sixth Splunk State of Observability report on October 20, 2026. In a survey conducted by Oxford Economics in February and March, respondents reported difficulty measuring AI workloads and controlling costs. The findings are survey responses, not audited totals of enterprise spending.
AI monitoring is a priority, but measurement lags
Cisco says 87% of respondents considered AI monitoring an organizational priority, while 56% said they could not measure AI-workload metrics such as large language model response quality. The gap between those responses points to a practical challenge: teams may prioritize monitoring without having a way to assess whether an AI system’s responses meet their needs.
Cisco also reports that 85% considered visibility into AI hallucinations important, while 58% said they were effective at providing that visibility.
Splunk’s official report page says 86% reported improvements in MTTD due to agentic AI. That is a reported survey finding, rather than a separately stated measurement of performance.
What respondents said about AI and telemetry costs
The survey points to three different cost concerns: difficulty controlling AI expenses, billing spikes that create operational strain, and telemetry costs relative to monitored infrastructure.
| Survey response reported by Cisco | Respondents | What the response describes |
| Said they could not control AI costs | 55% | Perceived ability to manage AI expenses |
| Said unexpected billing spikes caused more operational strain than high baseline costs | 60% | The relative operational strain of spikes and steady costs |
| Said telemetry costs exceeded the cost of monitored infrastructure | 57% | A reported comparison between two cost categories |
These percentages describe what respondents said; they are not measured dollar amounts. In particular, the telemetry figure is a reported comparison, not a total that can be applied to every organization.
Who took part in the survey
Cisco said Oxford Economics surveyed 2,050 ITOps and engineering professionals across 10 countries and 16 industries in February and March 2026. Participants ranged from practitioners to executives. Cisco announced the report in October, several months after the fieldwork.
Monitoring brings security and observability teams together
Cisco reports that 81% of respondents said monitoring AI workloads had increased collaboration between observability and security teams. It also says 73% could trace performance issues to security root causes, compared with 65% in its prior-year comparison. For troubleshooting with security teams, Cisco reports 77%, compared with 62% in 2025.
