A certification mentioned on this page has been retired. Microsoft Certified: Azure AI Engineer Associate (AI-102) is no longer available to take. Microsoft reports the retirement date as 2026-06-30. The replacement is Microsoft Certified: Azure AI Apps and Agents Developer Associate (AI-103). Read any reference below as historical, not as advice to take this retired exam. Check the successor's current requirements before planning your preparation.
Some links on this page are affiliate links. If you sign up after clicking one we may earn a commission, at no extra cost to you — and it never affects how we rank or rate anything. How this site is funded.
Quick answer
AI-900 (Azure AI Fundamentals) is the gentlest proctored AI credential on the market, and two weeks of part-time study is a realistic pass timeline for most people — including complete non-developers. Everything you need to prepare is free: Microsoft Learn's official learning path covers the full exam outline, and Microsoft publishes practice assessment material. Your only unavoidable cost is the exam fee. This guide gives you the day-by-day plan.
Where we would start
AI-900 preparation is genuinely free on Microsoft Learn and this guide sends you there rather than selling you an alternative. The open question is what comes after the exam: twenty-nine hours building on the API and retrieval stack, which is what the certificate is meant to lead to.
The table below compares 4 certifications on provider, level, realistic time, coding needed and best for.
| Certification | Provider | Level | Realistic time | Coding needed | Best for |
|---|---|---|---|---|---|
| Azure AI Fundamentals (AI-900) | Microsoft | Foundational | ~2 weeks at an hour a day | No | The exam this guide prepares you for |
| Microsoft Learn AI-900 learning path | Microsoft | Foundational | Included in the two weeks | No | The free official prep — start here |
| Azure AI Engineer Associate (AI-102) | Microsoft | Intermediate (associate) | ~1–2 months of prep | Yes (Python or C#) | The next rung for developers |
| AWS Certified AI Practitioner (AIF-C01) | AWS | Foundational | ~4–6 weeks of prep | No | The AWS-side equivalent |
How hard is AI-900 really?
It is the easy end of proctored exams, and that is precisely its appeal — a verified, closed-book credential without a technical prerequisite, which is why it tops our list of the easiest certifications employers respect. No code, no maths beyond concepts, and question styles built for recognition rather than recall traps.
The honest caution runs the other way: the exam's reputation for ease makes people skip preparation entirely, and the questions that fail them are service-mapping ones — which Azure service fits which scenario. Concepts you can reason through; service names you have to have seen. Two weeks of light structure removes that risk almost completely.
What's actually on the exam?
Work from Microsoft's official skills outline, which defines the domains and their weightings. In broad strokes the exam covers: AI workloads and responsible-AI considerations; fundamental machine-learning principles; computer vision workloads on Azure; natural language processing workloads; and generative AI workloads and Azure's tooling for them.
Question formats are multiple choice and variants — drag-to-match, hot area and similar interactive types. Nothing requires writing or reading code; several questions hinge on matching a scenario to the right Azure service.
Not sure this is the right one for you?
Answer a few questions about your background and what you want the certificate to do, and the picker narrows it to one recommendation — from the same vetted list this page ranks from.
Try the AI Certification Picker →Can you really pass in two weeks?
Yes, at roughly an hour a day, if you follow a plan rather than wandering the documentation. The calendar that works:
- Days 1–3: AI workloads and responsible-AI concepts on the Microsoft Learn path. Slow reading beats re-reading — take rough notes on each service name you meet.
- Days 4–6: machine-learning fundamentals — training versus inference, classification versus regression versus clustering, and where Azure Machine Learning fits.
- Days 7–9: computer vision and natural language processing modules, with special attention to which named Azure service handles which job.
- Days 10–11: generative AI content — models, prompts and Azure's generative tooling — which is newer material and heavily represented in current outlines.
- Days 12–13: practice assessments. Score them honestly, re-study only the domains you missed, and repeat until comfortably clear of the passing threshold.
- Day 14: exam. Book the slot on day one so the deadline does the motivating.
Which resources should you use?
The free official stack covers everything: Microsoft Learn's AI-900 learning path is the syllabus in tutorial form, and the official practice assessment tells you when you are ready. Most people need nothing else — a paid video course adds convenience, not coverage, at this level. If you prefer video pacing, one course is plenty; buying three is procrastination with receipts. The free-certifications roundup covers what else the Microsoft Learn ecosystem gives away.
How do you register, and what does it cost?
Register through Microsoft's certification portal; the exam runs online-proctored from home or at a test centre. The fee is region-dependent, and students can often sit it at a reduced rate. Employers running Microsoft training events sometimes distribute discount vouchers — worth one email to your L&D team before paying personally.
For an online-proctored sitting: quiet room, cleared desk, webcam sweep, government ID, and no phone within reach. Technical check-in takes longer than people expect — start thirty minutes early.
Who should take AI-900 — and who should skip it?
Take it if you work anywhere near a Microsoft-stack organisation and want a verified credential without code: business roles, ops, project managers, support and admin staff, and anyone for whom the no-coding path is the right lane. It is also the natural exam upgrade for people who finished a course certificate and want something proctored — the reasoning our fastest certifications guide walks through.
Skip it if you are a working developer heading for AI-102 — Microsoft sets no prerequisite, and the fundamentals exam adds little you will not cover in associate prep. Our AI-900 vs AI-102 comparison makes that call in detail. And if your employer runs on AWS, sit the AWS equivalent instead — the trade-offs live in our AIF-C01 vs AI-900 comparison.
What comes after you pass?
Fundamentals certifications do not expire, so there is no renewal treadmill. The badge lands on your CV and LinkedIn the same day, and the vocabulary starts paying off immediately in meetings where Azure AI services get discussed.
The next step depends on your lane. Developers used to continue to AI-102 — that exam has been retired and replaced, and our guide to what replaced AI-102 covers where that path now goes. Non-developers usually get more value from applied skills — prompt work, workflow building — than from stacking a second fundamentals badge. The staged view lives in our AI certification roadmap.
The gap AI-900 leaves is deliberate. It is a vocabulary exam: it certifies that you know what a Form Recognizer or a knowledge mining pipeline is, and it asks you to build neither. That is the right scope for a fundamentals badge and the wrong scope for a CV on its own, which is why the honest next move for most people is a hands-on track rather than a second badge. If you write Python, Associate AI Engineer for Developers is about 29 hours on the OpenAI API, LangChain, embeddings, vector search and Model Context Protocol — building the applications AI-900 only names. It is not Azure-specific and it carries no Microsoft badge; take it for what you can then show working, and take the annual subscription rather than the monthly one, which costs noticeably less per month.
Associate AI Engineer for Developers →Where most AI-900 advice gets it wrong
Two failure modes dominate. The first is the 'easiest cert ever' framing — accurate enough to be dangerous, because it produces candidates who walk in cold and meet a wall of service names they have never read. Easy with preparation is not the same as passable without it.
The second is treating AI-900 as trivia to be dumped and forgotten. Brain-dump sites violate the exam agreement and, worse, defeat the point: the exam's actual value is that its vocabulary — workloads, services, responsible-AI language — is the vocabulary of every Azure AI conversation you will sit in for years. Learn it properly once and the certificate is the smallest part of what you keep.
Verdict
AI-900 is the best effort-to-credential ratio in the AI certification market: free preparation, a two-week timeline, and a proctored badge from an issuer every recruiter recognises. Take it if you are in or near a Microsoft-stack workplace and want verified AI literacy; skip it only if you are a developer going straight to the associate tier, now AI-103. Book the exam on day one and let the date pull you through the plan. If you are still choosing your first credential, the 2026 rankings give the full field, and the Picker matches one to your situation in two minutes.
Ready to start?
Included in a DataCamp subscription rather than bought outright, so the cost is what you pay while you are working through it — which is an argument for finishing.
Frequently asked questions
How long does it take to prepare for AI-900?
About two weeks at an hour a day for most people, working through Microsoft Learn's free AI-900 learning path. Complete beginners with no exposure to AI concepts should allow three to four weeks; anyone already working near Azure can compress it to a week of evenings plus the practice assessments.
Two weeks is realistic only with a plan rather than a wander through the documentation. The sequence that works: AI workloads and responsible AI first, then machine-learning fundamentals, then computer vision and natural language processing, then the generative-AI material — which is newer and heavily represented in current outlines, so do not leave it until the night before. Give the last two days to practice assessments, re-studying only the domains you miss. Book the exam slot on day one: the deadline does most of the motivating.
Does AI-900 require coding?
No. AI-900 is entirely conceptual — no code is written and none is read — which is what makes it the standard proctored pick for people on a no-code path. The question formats are multiple choice and its interactive variants, such as drag-to-match and hot area, and none of them require a development background.
What it does require is familiarity with names. Several questions ask you to match a scenario to the right Azure service, and that is the part you cannot reason your way through: concepts you can work out under pressure, service names you either have or have not read. This is also why the exam's reputation for being easy is dangerous — easy with preparation is not the same as passable without it, and service-mapping questions are what fail the people who walk in cold.
What is the passing score for AI-900?
Microsoft publishes the passing threshold and the scoring approach in the official exam documentation, and that is the figure to work from — it is a scaled score, not a raw percentage of questions right, and Microsoft revises the exam outline periodically. Check it alongside the current skills outline when you start rather than relying on a number from a forum post.
Microsoft's official practice assessment reports against the same scale, which makes it the honest readiness test. Book the real exam once you clear the threshold consistently across several attempts, not after a single good run — one lucky pass usually means you have started recognising the practice questions rather than knowing the material. If a domain keeps dragging the score down, re-study that domain specifically instead of retaking the whole assessment; the retake mostly re-tests what you already know.
Does the AI-900 certification expire?
No. Microsoft's fundamentals-level certifications, AI-900 among them, carry no expiry — once you pass, it stays on your transcript with no renewal, no fee and no maintenance. That is unlike Microsoft's associate- and expert-level credentials, which do require periodic renewal, and unlike AWS certifications, which carry a recertification cycle of their own.
It is a genuine part of the value, and it is worth weighing when you compare foundational exams across clouds. A credential that never expires costs you the exam fee once, so the honest question is only whether the preparation teaches you something useful — and here it does, because the vocabulary the exam tests is the vocabulary of every Azure AI conversation you will sit in afterwards. The badge lands on your CV and LinkedIn the same day you pass.
Is AI-900 worth it?
For non-developers in or near a Microsoft-stack workplace, yes — it is the best effort-to-credential ratio in this market. Preparation is free through Microsoft Learn, the timeline is about two weeks, the credential never expires, and it is proctored, which is what separates it from a course certificate: a recruiter reads it as verified rather than self-reported. It suits business roles, operations, project managers, support and admin staff.
Two groups should skip it. Working developers get more from putting the same hours into the associate tier — now AI-103, which replaced the retired AI-102 — since Microsoft sets no prerequisite and the fundamentals material is largely covered in associate preparation. And if your employer runs on AWS, sit the AWS equivalent instead: a foundational vendor credential is worth what your market pays for that vendor, and nothing else about it changes that.
Keeping this current. Course formats, prices, and certification exam fees change and vary by region. We review our guides regularly — this one was last updated in July 2026 — and we always recommend confirming the specifics on the provider's official page before you enrol.