Dynatrace to Buy Arize for $915 Million to Monitor Enterprise AI Agents

Dynatrace has agreed to buy Arize, the AI observability and LLM evaluation company, for about $915 million, Dynatrace announced on August 13. ITPro reported the next day that the deal pairs Arize's evaluation technology, built by a company founded in 2020, with Dynatrace's production monitoring. The deal puts one of the best-known tools for testing and monitoring AI models and agents inside a mainstream application monitoring platform. For CIOs and chief AI officers, it is a sign that checking whether AI agents actually work is moving from a specialist niche into the core of enterprise IT operations.

What did Dynatrace agree to buy?

Arize, based in San Francisco, sells a platform that helps teams find problems in AI systems, trace their causes and improve performance, for both traditional machine learning and generative AI. Its products include Arize AX, a managed platform, and Phoenix, an open-source tool for tracing, evaluation and observability that teams can run locally or self-host. Dynatrace described it as open-source-native and stack-agnostic, with customers from Fortune 500 companies to AI-native builders, but did not disclose customer counts or revenue.

According to Dynatrace's announcement, the price is about $915 million, subject to customary adjustments. Roughly $815 million will be paid in cash, with the rest in replacement equity awards for Arize employees who join Dynatrace. Dynatrace expects the deal to close later this quarter or early in its fiscal third quarter, pending regulatory reviews.

Arize co-founders Jason Lopatecki and Aparna Dhinakaran will join Dynatrace. Lopatecki will lead the Arize team and report to Dynatrace CEO Rick McConnell.

Why does AI observability matter to enterprises now?

Dynatrace's pitch rests on a gap many enterprise teams will recognize. AI evaluation, the work of testing whether a model or agent gives good answers before launch, often sits with one team and one set of tools. Production monitoring sits with another. Dynatrace said combining the two gives end-to-end visibility from development to production, and it projected that the AI observability market will exceed $10 billion by 2030, a company forecast rather than an independent estimate.

Explaining why Arize was founded, Lopatecki said in the announcement that "AI teams needed a way to know their agents were actually working correctly, not just running." That gets harder as agents call tools and act without a person reviewing each step.

AI Connect speakers have been making the same argument for some time. "If you don't observe a system, there's only one direction it goes. It degrades over time," Michael Segner, Product Strategy at Monte Carlo, said during a AI Connect webinar on why data and AI observability must converge. A separate session covered early warning signals for production agents that pass testing but drift on live data.

What does the deal mean for Dynatrace's business?

Dynatrace was candid about the cost in its announcement. It expects Arize to add about 200 basis points to annual recurring revenue growth in fiscal 2027, while reducing non-GAAP operating margin by about 175 basis points that year. The company said it expects margins to expand again from fiscal 2027 levels into fiscal 2028 and beyond.

The strategic logic is about reach as much as revenue. Dynatrace sells mainly to IT operations and site reliability teams, while Arize's community is AI engineers and developers who often pick their own tools. J.P. Morgan and Goodwin Procter advised Dynatrace, while Qatalyst Partners and DLA Piper advised Arize.

How did analysts and investors react?

Early outside reaction was mixed. RBC Capital Markets kept an Outperform rating on Dynatrace and said the deal could speed up organic growth and fill roadmap gaps such as LLM evaluation, drift and hallucination detection, while warning that integration risk remains, MT Newswires reported on August 14. Dynatrace shares fell in trading that day, according to the same report. Constellation Research analyst Mike Ni read the deal as part of a push by monitoring vendors to become the record keepers for AI decisions, arguing that "whoever owns that decision trace owns the learning loop."

What should enterprise AI leaders watch?

For buyers, the first question is continuity. Many teams adopted Phoenix precisely because it is open source and works across model providers. Customers will want to see whether that neutrality, and the open-source project, hold up inside a larger commercial platform, and should ask for specifics in renewal talks.

The second question is ownership inside the enterprise. If evaluation and production monitoring converge in one platform, someone has to own the combined view. That could be the AI platform team, IT operations or a new agent operations function. Girish Bhat, SVP at Revefi, described the goal during a AI Connect panel on where humans belong in agentic AI governance, saying "observability in an agentic system is like a flight recorder." A flight recorder only helps if someone reads it.

The third is measurement. Monitoring shows that an agent's behavior changed, but teams need their own tests to judge whether that is a problem, which is why AI Connect speakers favor custom benchmarks built with domain experts.

Next, watch how quickly Dynatrace folds Arize's evaluation and agent tracing into its platform after closing, and whether rival monitoring vendors respond. Either way, the price tells enterprise leaders something useful: knowing what production agents are doing is now valued as infrastructure, not an afterthought.

Frequently asked questions

What does Arize do?

Arize offers an AI observability and LLM evaluation platform that helps teams detect, troubleshoot and fix problems in machine learning and generative AI systems, including AI agents. Its Phoenix tool is open source.

How much is Dynatrace paying for Arize?

About $915 million, subject to adjustments, with roughly $815 million in cash and the rest in replacement equity for Arize employees.

When will the Dynatrace Arize deal close?

Dynatrace expects it to close later this quarter or early in its fiscal third quarter, subject to regulatory reviews and customary conditions.

Filed under: Observability & Data Quality, Agentic AI, AI Governance & Trust