What is it?

What exactly is the Microsoft Certified: Azure AI Fundamentals?

The 2026 Azure AI Fundamentals exam — the no-code on-ramp to AI, now built around Microsoft Foundry.

Microsoft Certified: Azure AI Fundamentals (AI-901) is the current beginner credential for artificial intelligence on Azure, launched in 2026 to replace the retired AI-900. It validates conceptual knowledge of AI capabilities plus the foundational technical skills to work with them, including basic Python syntax and familiarity with Azure resources. Unlike the older AI-900, AI-901 is organized around two areas: identifying AI concepts and capabilities, and implementing AI solutions by using Microsoft Foundry. It is intended for candidates at the beginning of an AI solution-development career and is open to both technical and non-technical backgrounds — data-science and software-engineering experience are not required, though awareness of basic cloud concepts helps. Employers use it to confirm baseline AI literacy as generative AI and agentic solutions become mainstream.

The path in brief

1. Confirm AI-901 is your target (AI fundamentals)
2. Study the free AI-901 Microsoft Learn path
3. Take the official practice assessment
4. Schedule the $99 exam via Pearson VUE
5. Pass with 700/1000 and renew yearly for free

Before you book

Who is eligible to sit the Microsoft Certified: Azure AI Fundamentals?

Open to all. Anyone may register and sit AI-901; students can schedule through Certiport.

Who it is for: Students, career changers, business stakeholders, and AI beginners.

The test itself

What is the format of the Microsoft Certified: Azure AI Fundamentals?

The Microsoft Certified: Azure AI Fundamentals is administered by Pearson VUE (students and educators via Certiport) and runs 45 minutes, contains 40–60 (includes unscored preview items) questions and requires 700 / 1000 (scaled) to pass.

The fine print

What are the retake and renewal rules for the Microsoft Certified: Azure AI Fundamentals?

Proctoring & delivery

Delivered by Pearson VUE. You can sit the exam at a Pearson VUE test center or take it online through OnVUE remote proctoring; the fee is the same for both. Online proctoring requires a quiet, private space, a working webcam, and a system check beforehand.

Retake policy

If you do not pass, you may retake the exam after 24 hours. A second failed attempt requires a 14-day waiting period, and any further attempts also require a 14-day wait between them, up to a maximum of five attempts in a 12-month period.

Score reporting

A pass/fail result is shown immediately after the exam. The official score report is emailed within about 24 hours and is also available in your Microsoft certification profile, with section-level feedback on how you performed.

Recertification

Microsoft role-based, specialty, and fundamentals certifications are valid for 12 months from the date earned. Renew at no cost by passing the shorter online renewal assessment on Microsoft Learn before the credential expires. Retired certifications cannot be renewed and must be re-earned through the active successor if one exists.

Refunds & rescheduling

Exam fees are generally non-refundable once a seat is purchased, but you can cancel or reschedule through Pearson VUE at no charge if you do so at least 24–48 hours before the appointment (exact window varies by region).

What it pays

How much does this credential pay?

No SOC code exists for "AI fundamentals" — the honest anchor is the IT occupations group at $105,990 (BLS, May 2024)

There is no Bureau of Labor Statistics occupation called AI engineer, prompt engineer or AI developer, and pretending otherwise would be the easiest way to mislead you on this page. The BLS occupational structure predates the generative-AI job titles now appearing on job boards, and the roles AI-901 points toward are currently distributed across Software Developers, Computer Occupations All Other, and Computer Support Specialists depending on how much code the job actually involves. We therefore give you the group-level figure for computer and information technology occupations as the honest anchor, plus the software developer figure as the upper marker for where the AI-901 audience is heading, and we label both for what they are. What makes AI-901 different from its predecessor is worth stating in salary terms. AI-900, which it replaced, was explicitly open to non-technical candidates and required no coding at all — it was a literacy badge for salespeople, project managers and analysts. AI-901 is not that exam. Microsoft's own audience profile now says you need knowledge of Python coding syntax and programming techniques and familiarity with provisioning Azure resources, and the certification is filed under the AI Engineer job role. The credential has moved from the business side of the house to the developer side. That repositioning is the single most useful thing to know about its career value: it is now a genuine on-ramp to a junior developer track rather than a conversation-starter for non-technical staff, but it is also no longer the frictionless credential a non-coder could pick up in a weekend. As with every fundamentals-tier certification, AI-901 alone will not move your salary. What it does is make you a credible applicant for junior roles where the employer wants evidence you have actually deployed a model rather than read about one.

Job growth.Overall IT occupations projected to grow much faster than average 2024-2034; software developers specifically +15% (BLS OOH, 2024-2034 projections)

Source: BLS Occupational Outlook Handbook — Computer and Information Technology Occupations

Your schedule

How long should I study for it?

35-55 hours (add 15-20 more if you do not currently read Python) of focused study

Do not plan AI-901 the way people planned AI-900. This is the mistake that will cost candidates the most this year, and it is entirely understandable: the certification name is identical, the price is identical, the fundamentals tier is identical, and every study guide written before mid-2026 describes the old exam. The content is not the same. AI-900 was five roughly equal domains of conceptual material — AI workloads, machine-learning principles, computer vision, NLP and generative AI — none of which required you to write or read a line of code. AI-901 has two domains, and the larger one, at 55-60%, is "Implement AI solutions by using Microsoft Foundry". More than half the exam is now about actually doing something in a specific product. Microsoft's audience profile changed to match: you now need Python coding syntax and programming techniques, familiarity with provisioning Azure resources, and awareness of REST APIs, SDKs and CLIs. The plan below therefore allocates roughly a third of the time to concepts and two thirds to Foundry, which mirrors the blueprint weighting, and it assumes you will actually deploy a model rather than read about deployment. If you cannot currently read a twenty-line Python script that calls an SDK and explain what it does, add a preliminary week for that before you start — it is a prerequisite for the second domain, not an optional extra. Four to six weeks is realistic for someone with basic programming exposure; a non-coder should budget eight to ten weeks and accept that the Python gap, not the AI content, is the long pole.

1. Week 0 (only if you do not code)10-15 hrs

   Python literacy floor

   • Get to the point where you can read a script that imports an SDK, instantiates a client, calls a method and handles the response
   • You do not need to write Python from scratch — the exam asks you to understand code examples that call AI models and services
   • Skip this week entirely if you already program in any language
2. Weeks 1-210-14 hrs

   Identify AI concepts and capabilities (40-45%)

   • Work the Microsoft Learn path "AI concepts for developers and technology professionals" — it is written specifically for AI-901 and did not exist for AI-900
   • Cover AI workload types, model components and configurations, and when an AI solution is the right answer at all
   • Learn the six Microsoft responsible-AI principles by name and be able to attach each to a concrete failure mode, not just recite the list
   • Understand core cloud concepts as they apply here: services, resource deployment, and how an AI resource is provisioned
3. Weeks 3-414-20 hrs

   Implement AI solutions by using Microsoft Foundry (55-60%) — the majority of the exam

   • Work the Microsoft Learn path "Get started with AI applications and agents on Azure"
   • Actually open the Microsoft Foundry portal and deploy a model. Reading about the portal is not sufficient for a domain worth 55-60%
   • Build one single-agent solution end to end, however trivial — the blueprint explicitly covers implementing single-agent solutions
   • Cover the four solution families named in the objectives: generative-AI apps and agents; text and speech; computer vision and image generation; information extraction
   • Read the Python quickstart samples for each service and be able to say what each call does
4. Week 56-10 hrs

   Consolidation, practice assessment, and change-log check

   • Take the practice assessment on AI Skills Navigator — note it is NOT in the usual Microsoft Learn practice-assessment location and requires you to be signed in
   • Re-read the study guide and confirm it still says "Skills measured as of April 15, 2026"; if that date has moved, read the change log before sitting
   • Review how client applications consume AI models and services — this framing question recurs across both domains
   • Rehearse the responsible-AI principles against scenarios rather than definitions

What to buy

Which study resources are worth paying for?

The resource market for AI-901 is thin, and that is genuinely useful information rather than a gap in our research. The exam is a few months old, so the mature third-party ecosystem that exists for AZ-104 or AZ-900 simply has not been built yet. What exists is largely Microsoft's own material, which for this exam is the correct place to be anyway — the two Learn paths were written for this blueprint. Be actively suspicious of any third-party AI-901 course claiming comprehensive coverage; check whether it mentions Microsoft Foundry at all.

Prices and availability checked 2026-08-04; USD list prices before regional adjustment or discount. We take no commission from any provider listed. The thinness of this table reflects the exam being new, not incomplete research — where we could not verify a product exists for AI-901, we say so rather than list it.

What you'll face

What question types will I see?

AI-901 draws on Microsoft's standard item formats: multiple choice, multiple response, drag-and-drop matching, hot area, build list and the recurring three-statement true/false block. Case studies and interactive labs — the multi-question tabbed scenario blocks and live environments used on associate and expert role-based exams such as AZ-104 and AZ-305 — do not appear on Fundamentals exams, and AI-901 is a Fundamentals exam despite its technical content. What genuinely distinguishes AI-901 from every other fundamentals paper in the Microsoft catalogue is code. Because Microsoft's audience profile requires you to understand Python that calls AI models and services, you should expect items that present a short snippet and ask what it does, which client or method is correct, or what a given parameter changes. That is a format shift from AI-900, which never showed you code. It is still a recognition task rather than an authoring one — you will not be dropped into an editor — but a candidate who cannot follow an SDK call will lose marks that no amount of conceptual revision recovers. Two mechanical points. Multi-part questions award one point per correctly answered component, so a drag-and-drop with one wrong placement still banks the rest; never leave a row blank, and remember there is no penalty for guessing anywhere. And the exam is closed-book: the split-screen Microsoft Learn panel offered on associate and expert exams is explicitly unavailable on Fundamentals exams, so every SDK and service name must come from memory. The paper runs 45 minutes inside a 65-minute seat. You may flag items and sweep them on the review screen at the end, but if you start an unscheduled break through the exam UI you lose access to everything you have already seen, flagged items included.

Samples are editor-written illustrations of the published blueprint, not live exam items.

Reference

What are the key facts about the Microsoft Certified: Azure AI Fundamentals?

Microsoft Certified: Azure AI Fundamentals is a certification credential; awarded by Microsoft; the exam fee is $99 USD; typical preparation is 2–4 weeks of study.

In short

Is the Microsoft Certified: Azure AI Fundamentals worth it?

• AI-901 is the current Azure AI Fundamentals exam, launched in 2026 to replace the retired AI-900.
• The exam is 40–60 questions in about 45 minutes; the passing scaled score is 700/1000.
• Microsoft's published weights are AI concepts and capabilities 40–45% and implementing AI solutions with Microsoft Foundry 55–60%.
• It costs $99 and renews for free every 12 months via a Microsoft Learn assessment.

Worth comparing

How does the Microsoft Certified: Azure AI Fundamentals compare with similar credentials?

• Microsoft Certified: Azure Fundamentals Microsoft
• Microsoft Certified: Azure Administrator Associate Microsoft
• Microsoft Certified: Azure Virtual Desktop Specialty Microsoft

Trust

Where does this information come from?

Everything above is taken from the awarding body's own published material. Fees, question counts and domain weights are revised regularly — check the official page before you pay.

• Microsoft Awarding body
• Official Microsoft Certified: Azure AI Fundamentals exam page Exam page
• Microsoft certification programme Programme rules

Research confidence: high · Last reviewed 2026-08