Microsoft AI Certifications Explained

Azure Data Scientist Associate (DP-100): retired, refocused successor

DP-100, "Designing and Implementing a Data Science Solution on Azure", was the machine-learning specialist's credential — its final domains spanned designing and preparing an ML solution, exploring data and running experiments, training and deploying models, and optimising language models for AI applications, built around Azure Machine Learning, MLflow and Foundry tooling.

It too is retired: the Learn page carries the official notice, with secondary sources dating the retirement to 1 June 2026. Those same sources describe the successor as a new MLOps-focused associate certification (the reported exam code is AI-300, though the two main sources render the new certification's name slightly differently, so we will not print one as definitive). The direction of travel is clear even if the details are not: Microsoft is repositioning the data-science credential away from notebook-era model building toward operating machine-learning and generative-AI systems in production. The full story of the DP-100 exam as it existed lives in our Azure Data Scientist certification guide.

How the three certifications relate

Understanding the relationships matters more than memorising codes, because the relationships survive the renaming:

• Fundamentals feeds both associate tracks. Azure AI Fundamentals (AI-901) is the shared on-ramp. It has never been a prerequisite — Microsoft associate exams have no formal prerequisites — but it front-loads the vocabulary both deeper tracks assume.
• The engineer track is about building applications with AI services. AI-102's territory — and, by all reports, its successor's — is the developer who wires vision, language, generative and agentic capabilities into software.
• The data-science track is about the models themselves. DP-100's territory was training, evaluating and deploying models; its reported successor pushes further into MLOps, the discipline of running those models reliably.
• Renewal separates the tiers. Fundamentals certifications never expire. Associate-level AI certifications, like all Microsoft role-based credentials, expire after one year and renew free through an unproctored, open-book online assessment on Microsoft Learn — a policy that will presumably apply to the successors as well.

A neighbouring credential often mistaken for an AI certification is worth one clarification: Microsoft's data-analytics and data-engineering certifications (PL-300, DP-700) are not part of the AI line-up, though DP-900 Azure Data Fundamentals makes a sensible companion to AI-901 if your interests straddle data and AI — free DP-900 practice questions can help you gauge that side. Where the AI certifications sit among Microsoft's most valuable credentials overall is ranked in our best Microsoft certifications for 2026 round-up.

What should you actually do in late 2026?

The honest answer depends on which of three situations you are in.

Scenario one: you are new to AI on Azure. Take AI-901 now. It is active, cheap relative to associate exams, permanent once earned, and unaffected by the associate-level churn. A marketing analyst who wants to speak credibly about generative AI in a Microsoft-shop employer, for example, gets full value from the fundamentals tier alone — the associate credentials were never aimed at that reader.

Scenario two: you were preparing for AI-102 or DP-100. Stop preparing for the retired exam — it cannot be sat. Your study time is not wasted: the successor exams reportedly cover overlapping ground (AI application development in one case, ML operations in the other), and hands-on skill with Azure AI services and Azure Machine Learning transfers regardless of exam code. Watch the official Microsoft Learn certification catalogue for the successors' pages, and only buy exam-specific preparation once official study guides exist. Announced-by-rumour objectives are not a study plan.

Scenario three: you already hold AI-102 or DP-100. Your certification stays on your transcript permanently, but it can no longer be renewed once it lapses. If your role depends on holding a current credential, plan to earn the successor when it becomes available; per community reporting there is no automatic upgrade path, so budget for a full new exam.

A sensible study sequence while the dust settles

1. Earn AI-901 for the conceptual base (or skip to step 2 if you already build on Azure daily).
2. Build hands-on depth in the tools the successors will test — Microsoft Foundry, Azure AI services and Azure Machine Learning — using Microsoft Learn's free learning paths, which Microsoft updates ahead of exam launches.
3. Use practice questions diagnostically, not as a memorisation exercise. Work a set, then sort your misses by objective: a cluster of errors in one domain tells you where the next week of study goes. Timed runs matter later, once an official successor exam and its study guide exist.
4. Re-check the catalogue quarterly. Successor exams reported as AI-103 and AI-300 may launch with different names, formats or scope than early reports suggest.

Frequently asked questions

Is AI-900 still available?

No — the AI-900 exam was replaced by AI-901 in April 2026. The Azure AI Fundamentals certification continues; only the exam behind it changed. Certifications earned via AI-900 remain valid indefinitely.

Can I still take the AI-102 or DP-100 exam anywhere?

No. Both are officially retired on Microsoft Learn and cannot be scheduled. Any provider claiming to book you into either exam is out of date.

Do Microsoft AI certifications expire?

Azure AI Fundamentals never expires. Associate-level certifications expire after one year but renew free via an online, open-book assessment — while the certification remains renewable. Retired certifications can no longer be renewed.

Is there a Microsoft certification specifically for generative AI?

Generative AI is woven through the current line-up rather than isolated: AI-901 covers AI concepts and Foundry-based solutions, and AI-102's final version included generative and agentic solution domains. The reported successor exams deepen this focus, but their official scope was unpublished at the time of writing — check Microsoft Learn for launch details.

Do I need AI-901 before attempting an associate-level AI exam?

No. Microsoft associate exams carry no formal prerequisites. Fundamentals is recommended groundwork for newcomers, not a requirement.

Which Azure AI credential should you pursue now?

If you take one thing from this explainer: the Azure AI certification path currently has one fully open door — AI-901 for Azure AI Fundamentals — and two doors mid-replacement, where the retired AI-102 and DP-100 await officially documented successors. Beginners should walk through the open door now; practitioners should spend the interim building hands-on Foundry and Azure ML skills, then certify against the successor exams once Microsoft publishes their official pages. Verify everything against the Microsoft Learn certification catalogue before spending money — in a transition year, the official page is the only list that counts.