Dynatrace is making a $915 million bet that traditional application observability and AI observability are about to become the same problem.
The company has signed a definitive agreement to acquire Arize, bringing together Dynatrace’s application and infrastructure monitoring capabilities with technology designed specifically to evaluate models, LLM applications and AI agents.
The transaction is expected to close later this quarter or early in Dynatrace’s third quarter, subject to regulatory approval and customary conditions.
AI Creates A New Observability Problem
Traditional observability answers questions about whether an application is running, where latency is occurring and whether infrastructure is healthy.
AI adds another layer.
An application can technically be operating perfectly while the model powering it produces a terrible answer.
That means teams need to monitor prompts, outputs, hallucinations, agent behavior and model quality alongside application performance, GPU utilization and infrastructure health.
Arize Fills The AI Gap
Arize was built specifically around that problem.
Its platform helps developers evaluate AI behavior, detect hallucinations, measure output quality and continuously validate AI systems.
Dynatrace sees an opportunity to connect those capabilities directly with the infrastructure and application telemetry it already collects.
That could give teams a single path from a bad AI response all the way down to the model, application or infrastructure component responsible for it.
Development Meets Production
One of the more important ideas behind the acquisition is closing the gap between the people building AI applications and the teams operating them.
AI engineers may evaluate model behavior using one toolset while SRE and platform teams monitor production using another.
Dynatrace wants to connect those environments.
The company says AI observability could become a market worth more than $10 billion by 2030.
The Bottom Line
Observability used to answer a relatively straightforward question: Is the software working?
AI changes the definition of “working.”
An AI service can have perfect uptime and terrible outputs.
By acquiring Arize, Dynatrace is preparing for an environment where enterprises need to monitor both software performance and AI behavior at the same time.
If AI agents become embedded throughout enterprise applications, that distinction may eventually disappear altogether.