Dynatrace acquires AI observability platform Arize for $915M

Dynatrace is paying $915 million in cash and stock to acquire Arize AI, the Berkeley-based startup that helps companies figure out whether their AI models are actually working as intended.

The deal, announced on August 13, 2026, positions Dynatrace to offer full life-cycle monitoring and optimization of AI systems.

What Arize brings to the table

Arize AI built its reputation on solving a problem that gets worse the more AI you deploy. Machine learning models can degrade silently. An e-commerce recommendation engine might start suggesting winter coats in July, and without proper observability tooling, nobody notices until revenue dips.

Arize’s platform catches those issues by tracking model performance, detecting data drift, and evaluating outputs across the AI lifecycle. The company raised a $70 million Series C in February 2025, which was the largest funding round in AI observability at the time.

In May 2026, Arize partnered with Google Cloud to standardize AI agent telemetry, essentially helping establish how AI systems report their own health metrics.

Dynatrace’s acquisition playbook

This isn’t Dynatrace shopping on impulse. The company acquired Metis in March 2025, adding database performance monitoring to its stack. Then came Bindplane in April 2026, which expanded its data pipeline ingestion capabilities.

Arize is the most expensive piece yet, and arguably the most strategically important. Dynatrace (NYSE: DT) already offers AI-powered observability, meaning it uses AI to monitor traditional software systems. With Arize, it gains the ability to monitor the AI systems themselves.

As enterprises deploy generative AI applications, they need tools that can evaluate whether a large language model is hallucinating, whether retrieval-augmented generation is pulling the right documents, and whether AI agents are completing tasks correctly. Traditional application performance monitoring wasn’t built for that. Arize was.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.