NL

Three corners of one landscape: which Microsoft fundamentals do you take first?

🎯 Orientation⏱ 5 minutes

What you'll learn here

Someone asked me this week which of the three Microsoft fundamentals to pick first: AI-901, DP-900, or AZ-900. My answer was a triangle.

Not to dodge the question, but to fix the frame. The three certifications sound like a ladder: a foundation, then a specialization, then another one. And if they sound like a ladder, you automatically search for the right order to climb them. But the ladder does not exist. What does exist are three corners of the same landscape, and which exam you pick first depends on which corner sits closest to where you work today.

What each exam covers

AZ-900 covers the platform: cloud concepts, Azure architecture and services, management and governance. Who can access what via Entra ID, how the cost meter runs, what the difference is between IaaS, PaaS, and SaaS, and how the shared responsibility model works. The objective domains were updated on 14 January 2026, with an English update planned for 20 July 2026.

DP-900 covers the data: relational and non-relational, transactional and analytical, large-scale warehousing and real-time analytics, with explicit coverage of Microsoft Fabric and Azure Databricks. The English update for this exam is scheduled for 21 July 2026; the current study guide applies until then.

AI-901 covers what you make that data do: AI workloads and considerations, fundamental machine learning principles on Azure, computer vision, Natural Language Processing, and generative AI, including the Foundry portal with tools, models, and agents. AI-901 went live on 30 June 2026 as the direct successor to AI-900, with an updated skills measured list from 15 April 2026. Passing score: 700.

Together, those three domains describe the full Microsoft data and AI landscape: where data lands, what data is, and what you make data do. Three corners of the same landscape, each with its own language.

Your corner is your starting point

The choice comes down to one practical question: what do you actually work with today? Pick the description below that sits closest to your work, and the triangle points out your starting corner.

Make your pick

What do you work with most today?

The other two follow faster

Once your foundation is solid in your own corner, the other two land more easily. The data professional who starts with DP-900 suddenly recognizes in AZ-900 exactly where the dataset they touch every day actually lives: in which Azure region, under which cost model, within which governance agreements. That same person reads AI-901 with more focus, seeing what an agent actually does with a semantic model because the data language is already there.

It also works in reverse: someone who starts from platform management with AZ-900 sees in DP-900 how the data services they already know relate as a system. And someone who starts with AI-901 from a Foundry pilot understands in DP-900 what data is actually flowing through it.

📌
Takeaway

The triangle is always the same, and no starting point is wrong. There is, however, a starting point that makes more sense for you, and it begins at your corner today.

Current context

AI-901 is the direct successor to AI-900, which was retired on 30 June 2026. The updated content now includes generative AI, agents, and the Foundry portal, areas that were not yet part of AI-900. If you were studying for AI-900 and have not made the switch: the material has expanded significantly, and the updated study guide on Microsoft Learn is where to start.

For DP-900 and AZ-900, both exams will receive English updates soon (21 and 20 July 2026). At the time of publication, the current study guides apply. The core of what each exam covers does not change with a refresh; the weighting across domains and the number of questions per topic may shift.

Want to know more about one, two, or all three, in one, two, or three training sessions? Visit trainerbjorn.nl or get in touch.

TrainerBjörn is part of DoGoDa, Doing Good with Data. Made by Björn, with help from AI. Disclaimers.