Dynatrace reveals where Enterprise AI scaling breaks

AI adoption is moving deeper into enterprise production environments, but scaling it is creating a different kind of problem. According to Dynatrace’s State of SRE and Platform Engineering 2026 study of 919 IT leaders, SRE and platform engineering teams are being asked to manage new levels of complexity as AI workloads grow.

The shift is changing what these teams need to deliver. Reliability is no longer only about keeping applications running. AI is bringing new telemetry, model performance concerns and unfamiliar failure patterns into the same environments.

AI workloads are changing the reliability playbook

The study shows that SRE and platform engineering teams already have strong backing inside enterprises. About 92% of organisations report executive support for SRE initiatives, while 89% of organisations with platform engineering have an internal developer platform.

The challenge now is applying that foundation to AI. For SRE teams, monitoring AI models has become the top AI use case, cited by 67% of respondents. Model performance and accuracy monitoring is already the most common AI-powered capability, at 58%.

Yet AI is not delivering the same impact everywhere. The study finds weaker-than-expected results in areas such as cost reduction and mean time to resolution (MTTR). This suggests that adding more AI tools alone may not solve the larger operational challenge.

AI breaking points are showing up in integration

For platform engineers, 37% say integrating with existing tools and systems is their biggest challenge. Only 40% report embedding observability across all deployment stages.

The data also points to a growing need to connect systems rather than manage them separately. Nearly half of SRE respondents say too many data sources and metrics make it harder to define and manage effective service-level objectives.

That fragmentation matters as AI infrastructure becomes more complex. Half of SREs already use AI-powered capabilities for automated incident response, pointing toward more agentic operations where systems can take action with less manual intervention.

Observability takes on a bigger role

The study positions observability as a foundation for reliability, governance and optimisation as enterprises scale AI.

Teams, however, are taking a measured approach. Visibility and human oversight are being prioritised before broader automation. That reflects a practical concern: autonomous systems need clear signals and controls before they can reliably act on their own.

For enterprises focused on scaling enterprise AI workloads, the next challenge may therefore be less about adopting AI and more about connecting the operational pieces around it.

Read More: 

Gartner semiconductor forecast 2026 points to an AI-led surge

Instawork Instacore takes robot training beyond labs

Okta Agent SSO puts AI agents under identity control

Kramer 1G and 10G AVoIP expands AV choices at InfoComm India 2026