Microsoft Azure AI Fundamentals

Practice exam
Azure AI Fundamentals

Updated October 31, 2026. Names, dates and rules checked against Microsoft Learn and Pearson VUE. The update history is at the bottom.

If you came looking for AI-900, the short answer is that it's gone. Microsoft retired the exam on June 30, 2026, but it kept the certification. Microsoft Certified: Azure AI Fundamentals is still on the books, and you now earn it by passing AI-901, the AI-900 replacement.

Don't assume AI-901 is AI-900 with a new number. It still tests AI concepts, but more than half of it is about building small things with Microsoft Foundry, and Microsoft expects you to read Python. This guide covers what the credential is, who it suits, what it proves, how long it lasts, whether it’s worth the money and where to go next. It’s the third in our 2026 Microsoft series, after the certifications guide and the exams guide.

What is the Azure AI Fundamentals certification?

Microsoft Certified: Azure AI Fundamentals is Microsoft's entry-level AI credential. Microsoft's certification page describes the candidate as someone at the beginning of a career in AI solution development, tags the job role as AI Engineer and lists one required exam: AI-901.

The name stayed the same. In its announcement, Microsoft said it wasn't retiring the certification, only changing what you need to earn it. AI-901 ran as a beta this spring, and Microsoft's June 2026 roundup lists it as generally available.

The exam in one paragraph. AI-901 is the only exam required, and you pass with 700 or higher. Microsoft prices exams by country. Its price sheet effective November 1, 2024 put Fundamentals exams at $99 in the US and as low as $44 elsewhere, so check the exam page for your price. You book through Pearson VUE, and the exam comes in 13 languages. Our Microsoft exams guide covers retakes, scheduling rules and the rest.

AI-901 vs AI-900: what changed?

The exam moved from describing AI services to building with them. Microsoft’s announcement sums up the shift this way: AI-900 tested understanding AI and machine learning concepts and identifying when to use Azure AI services, while AI-901 tests AI concepts plus implementing solutions with Microsoft Foundry. Here’s how the published outlines compare.

AI-900 (retired June 30, 2026) AI-901 (current)
Focus Five areas: AI workloads, machine learning, computer vision, language and generative AI Two areas: AI concepts and capabilities (40–45%), then implementing solutions with Microsoft Foundry (55–60%)
Hands-on expectations No software engineering or data science background needed Python syntax and programming techniques, Azure resources, and familiarity with REST APIs, SDKs and CLIs. Tasks include deploying a model and building small client apps
Tooling named Described service by service: Azure AI Vision, Face, Language and Speech, Azure Machine Learning, Azure OpenAI and Foundry Foundry (portal and SDK), Azure Speech in Foundry Tools, Azure Content Understanding in Foundry Tools
Old study notes Not applicable Partly useful. The six responsible AI principles and most workload concepts carry over. Machine learning technique notes mostly don’t, and the hands-on parts need fresh study (our read of the two outlines)

Sources: the retired AI-900 study guide and the AI-901 study guide.

Treat AI-901 as a rewrite, not a refresh. If you own AI-900 notes or a practice test, check them against the AI-901 skills list, which is dated April 15, 2026, before you rely on them.

Who is the Azure AI Fundamentals certification for, and who can skip it?

Microsoft aims it at people at the start of an AI development career, and the AI-901 course page describes the audience the same way. That points to students and career changers who can read basic Python, IT pros and junior developers who want a structured first pass at Foundry, and anyone who wants a recognized first step toward the associate exams.

The audience got narrower. AI-900’s study guide welcomed technical and non-technical candidates alike. AI-901 doesn’t, so some people can skip it:

  • Experienced developers and ML engineers. The Learn pages for AI-103, AI-200 and AI-300 don’t list AI-901 as a requirement. If you already build on Azure, go straight to the associate exam.
  • People who need business-level AI, not code. AB-730 (AI Business Professional) is for everyday Copilot users and needs no coding. AB-731 is for leaders guiding AI adoption.
  • AWS shops. The AWS Certified AI Practitioner (AIF-C01) is the closer fit. AWS’s exam guide says its target candidate uses but doesn't necessarily build AI/ML solutions, and it puts coding models out of scope.
  • AI-900 holders. Your certification stays. More on that below.

What does AI-901 prove, and what doesn't it?

The study guide splits the exam into two skill areas. In plain English:

Identify AI concepts and capabilities (40–45%). This is the know-what half. You describe the six responsible AI principles (fairness, reliability and safety, privacy and security, inclusiveness, transparency and accountability) and what they mean for a real solution. You explain at a high level how generative models work, choose a suitable model for a task, and pick deployment options and settings. You also match scenarios to workloads: generative and agent-based AI, text analysis, speech, vision and image generation, and pulling information out of text, images, audio and video.

Implement AI solutions by using Microsoft Foundry (55–60%). This is the do-it half, and it’s the bigger one. You write system and user prompts, deploy a model and test it in the Foundry portal, and build a small chat client with the Foundry SDK. You create and test a single agent and write a small client for it. You build lightweight apps for text analysis, speech and vision, interpret and generate images, handle spoken prompts with a multimodal model, and use Azure Content Understanding to extract information from documents, forms, images, audio and video.

A pass shows you’ve used the main building blocks and can follow a basic Foundry workflow from deployment to a working client. It doesn't show you can run AI in production. The outline stops at lightweight apps and a single agent, and it doesn't list model training or evaluation, MLOps, or the security, monitoring and scaling work real systems need. The associate-level credentials below go further.

Do you need Python and Azure experience first?

Some, yes. Microsoft's certification page says you need knowledge of Python coding syntax and programming techniques and should be familiar with Azure resources, and it adds familiarity with REST APIs, SDKs and CLIs. The AI-901 course also recommends core cloud concepts: storage, compute, and authentication and authorization.

Our practical read: you should be able to read a short Python script that calls an SDK, set an endpoint and key, run it, and fix a simple error. You should also know what a resource, a deployment and an API key are. If the cloud vocabulary is new, Azure Fundamentals (AZ-900) covers it, though Microsoft doesn’t list it as a requirement.

Does Azure AI Fundamentals expire, and what happens to AI-900 holders?

No. Microsoft's renewal page says Fundamentals certifications don't expire, and Pearson VUE's Microsoft updates page says the same. Associate-level credentials are the ones on a one-year clock, with free renewal.

Once you pass, the credential appears in your Microsoft Learn profile and on your transcript, which also lists the exams you've passed. Microsoft's sharing guide explains how to share a link or download a certificate, and says verified credentials carry an “Online Verifiable” tag. For others to verify yours, your Learn profile has to be public. The certification page also invites you to claim a badge for LinkedIn or your résumé.

If you passed AI-900, nothing is taken away. Microsoft's retirement page says earned certifications stay on your transcript when an exam retires, and the sharing guide says you can still view and share a retired credential. Microsoft's announcement says you may take AI-901 to show newer skills, but nothing requires it. Because the certification name is the same, your transcript's exam list is where AI-901 would show up.

Is the Azure AI Fundamentals certification worth it?

Worth it for the right person, but don't count on it for pay. A Fundamentals credential is a foundation and a signal: it tells an employer you've covered the basics, and it gives you a deadline and a syllabus for learning them. Here's what the data says, and what it can’t.

Pearson VUE, April 2025. Its 2025 Value of IT Certification Candidate Report surveyed nearly 24,000 IT professionals. Among respondents, 32% got a raise after certifying and 41% hoped for one, and 31% of the raises were above 20%. The share planning an AI or machine learning certification rose from 17% in 2022 to 35% in 2024. The limits: the answers are self-reported, they come from people who sat certification exams, they span many vendors and levels rather than Fundamentals alone, and Pearson VUE sells exam delivery, so it has a commercial interest in certification. The survey also predates AI-901.

Skillsoft, fielded May to September 2024. Skillsoft's IT Skills and Salary survey drew 5,191 responses and was distributed through Skillsoft and vendor partners, Microsoft among them. No Fundamentals credential made its US top 20. In its Europe and Middle East list, which draws on 1,150 responses, Azure Fundamentals holders averaged about $64,000 and Azure AI Engineer Associate holders, a credential since replaced by AI-103, about $111,000. Skillsoft sells training, the figures aren’t adjusted for location or cost of living, and averages like these tend to reflect who holds a credential (role, seniority, country) more than what the credential adds. This survey predates AI-901 too.

So where does the value come from? In our view: a structured way to learn Foundry basics, a no-expiry line on your résumé and LinkedIn, and a stepping stone to AI-103 or AI-200. It’s worth doing if you’re early in AI development and want that structure. Skip it if you already build on Foundry or need AI fluency without code.

Where should you go after AI-901?

Pick by the job you want next. We checked each credential on Microsoft Learn on October 31, 2026.

If you want to… Look at Level on Learn Worth knowing
Build AI apps and agents Azure AI Apps and Agents Developer Associate (AI-103) Intermediate Python and Microsoft Foundry. Replaces AI-102
Build back-end AI services Azure AI Cloud Developer Associate (AI-200) Intermediate Containers, data services, vector databases and Python. Replaces AZ-204
Run ML and generative AI in production Machine Learning Operations Engineer Associate (AI-300) Intermediate Python plus DevOps basics. Pearson VUE lists it as the DP-100 successor
Use or lead AI at work without code AI Business Professional (AB-730) or AI Transformation Leader (AB-731) Beginner Copilot and agents for users, adoption strategy for leaders
Administer Copilot and agents Microsoft 365 Certified: Copilot and Agent Administration Fundamentals (AB-900) Beginner Microsoft 365 core services plus Copilot and agent administration
Work with data Azure Data Fundamentals (DP-900) Beginner Core data concepts and Azure data services
Secure cloud and AI workloads Security, Compliance, and Identity Fundamentals (SC-900), then Cloud and AI Security Engineer Associate (SC-500) Beginner, then Intermediate SC-500 replaces AZ-500, which retired August 31, 2026. Its practice assessment wasn’t out when we looked

Replacement pairs come from Pearson VUE’s Microsoft updates page and Microsoft's retirement page. For the full career-path view, see our Microsoft certifications guide.

How should you prepare for AI-901?

Start with the free official material, then add practice.

  1. Read the study guide. The AI-901 study guide lists the skills measured as of April 15, 2026, and Microsoft last updated the page in July. Skip its “Find documentation” list, which still points to older services such as Anomaly Detector and Language Understanding.
  2. Take Microsoft's free practice assessment and try the exam sandbox. The assessment lives on AI Skills Navigator and needs a sign-in. The exam page links to the sandbox. Do both early, so you know where you stand.
  3. Work through Microsoft’s training. The one-day course Introduction to AI in Azure (AI-901T00) is on Microsoft Learn with instructor-led and self-paced options. One snag: on October 31 the certification page still said learning paths and modules weren’t available yet, so check what you can open.
  4. Build every task in the outline. Deploy a model in the Foundry portal, then write small Python clients for chat, an agent, text analysis, speech, vision and Content Understanding. Delete what you deploy when you're done.
  5. Check third-party practice tests against the new outline. Anything built for AI-900 follows a five-area outline, and AI-901 has two. Book when your hands-on tasks run without notes and your practice results are steady.

Update history

  • October 31, 2026: Full rewrite for AI-901, replacing the earlier AI-900 guide. Covers what the Azure AI Fundamentals certification is, what changed from AI-900, who it suits and who can skip it, what the exam does and doesn’t prove, prerequisites, expiry and AI-900 holders, dated survey data on value, next credentials by goal, preparation and an FAQ. Facts checked against Microsoft Learn and Pearson VUE.
  • 2024: Original guide published.

Frequently Asked Questions