AI-900 is retired
Microsoft retired the AI-900 exam on June 30, 2026. It can no longer be scheduled or taken. AI-901 is the replacement, and both lead to the same Microsoft Certified: Azure AI Fundamentals certification.
AI-900 and AI-901 were two different exams for the same Azure AI Fundamentals certification. AI-900 is gone, so here is what changed, what it means if you already hold AI-900, and how to prepare for AI-901.
Quick Answer
Haven't passed AI-900? Take AI-901. It is now the only exam that earns Azure AI Fundamentals. It keeps the AI-900 concepts and adds a Microsoft Foundry implementation domain, with basic Python reading.
Already passed AI-900? Your certification stays valid and does not expire. You do not need AI-901 unless you want to prove hands-on Foundry skills.
The Core Difference in One Paragraph
AI-900 was a pure-concepts exam: what AI is, what Azure AI services exist, what each does, and what Microsoft's Responsible AI principles are. There was no code and no expectation that you could navigate the Microsoft Foundry portal.
AI-901 keeps about 40–45% of that conceptual content and adds a new implementation domain (55–60% of the exam) built around Microsoft Foundry. You read Python snippets that use the Foundry SDK, identify the right system prompt for a behavior goal, configure a Foundry agent with tools, and choose between Content Understanding and Document Intelligence for an extraction scenario. It is still a fundamentals credential, but closer to hands-on work.
Side-by-Side Comparison
| Attribute | AI-900 (retired) | AI-901 (current) |
|---|---|---|
| Status | Retired June 30, 2026 | Available now |
| Certification earned | Microsoft Certified: Azure AI Fundamentals | Microsoft Certified: Azure AI Fundamentals (same certification) |
| Level | Fundamentals | Fundamentals |
| Number of domains | 5 (older blueprint) | 2 (concepts + Foundry implementation) |
| Implementation content | None | 55–60% of the exam |
| Azure AI Foundry coverage | High-level mention only | Deep: portal, SDK, agents, Content Understanding |
| Python requirement | None | Read short snippets |
| Code on exam | Zero | Multiple Python snippets |
| Passing score | 700/1000 | 700/1000 |
| Cost | $99 USD | $99 USD |
| Expiry | Never | Never |
| Typical prep time | 1–2 weeks | 2–4 weeks |
What AI-900 Covered (Now Retired)
AI-900 followed the older five-domain Azure AI Fundamentals blueprint:
- • AI workloads and considerations (15–20%)
- • Fundamental principles of machine learning on Azure (20–25%)
- • Features of computer vision workloads on Azure (15–20%)
- • Features of natural language processing (NLP) workloads on Azure (15–20%)
- • Features of generative AI workloads on Azure (20–25%)
Every question was concept-only: "which Azure service detects faces in images?", "which Responsible AI principle covers fairness?". Most of this content now lives in the first domain of AI-901.
What AI-901 Covers
AI-901 condenses to two domains:
- • Identify AI concepts and capabilities (40–45%): Responsible AI, how generative models work, model selection and deployment options, and common AI workloads (generative and agentic AI, text analysis, speech, computer vision, information extraction). Comparable to AI-900 conceptual content but slightly deeper.
- • Implement AI solutions by using Microsoft Foundry (55–60%): Foundry portal, model deployment, system and user prompts, Foundry SDK Python snippets, single-agent solutions, text and speech apps, multimodal vision, and Azure Content Understanding for documents, images, audio, and video.
Questions in domain 2 are scenario- and code-based: "which SDK call returns chat completions?", "which Foundry component handles persistent conversation state?", "which service extracts insurance-form fields from PDFs that mix typed and handwritten content?". You must be able to read code and recognize portal screens.
What To Do Now
Beyond these two exams, our guide to which Microsoft AI certification is best covers the full AI path.
You have not passed AI-900
Take AI-901. It is the only route to Azure AI Fundamentals now. Learn enough Python to read short snippets and practise the Foundry portal.
You already passed AI-900
Do nothing. The certification stays on your transcript and does not expire. Take AI-901 only if you want to show hands-on Foundry skills or plan to move on to AI-103.
You were studying for AI-900
Switch to AI-901. About 40–45% of it is the material you already studied; add the Foundry implementation domain on top.
Common Scenarios
"I am a Java developer who has never used Azure AI"
Take AI-901. You already read code in a different language; basic Python literacy is about a weekend of practice, and the Foundry implementation content is directly useful in your work.
"I am a product manager at a SaaS company adding AI features"
Take AI-901. The implementation content makes you a credible counterpart to your engineering team and helps you scope AI features realistically. You only need to read Python, not write it.
"I am a compliance lead who only needs AI literacy"
AI-900 used to be the right fit, but it is retired. AI-901's first domain covers the same Responsible AI and governance concepts; budget extra time for the Foundry domain, or decide whether a certification is needed at all.
"I want to take AI-103 in 6 months and need to build up"
Take AI-901. Its Foundry portal and SDK content is the direct ramp into AI-103, and candidates who skip it typically lose 4+ weeks re-learning Foundry from scratch.
"I passed AI-900 last year"
Your Azure AI Fundamentals certification is still valid. Take AI-901 if you do hands-on Azure AI work; skip it if you took AI-900 purely for awareness.
Frequently Asked Questions
Is AI-900 retired?
Yes. Microsoft retired the AI-900 exam on June 30, 2026 (11:59 PM Central Standard Time). It can no longer be scheduled or taken. The replacement is AI-901: Microsoft Azure AI Fundamentals, which maps to the same Microsoft Certified: Azure AI Fundamentals certification.
What is the difference between AI-901 and AI-900?
AI-900 tested conceptual understanding of AI and Azure AI services, with no coding and no implementation. AI-901 keeps that conceptual material (40–45% of the exam) and adds a substantial implementation domain (55–60%) focused on Microsoft Foundry: deploying models, writing system prompts, reading Python that uses the Foundry SDK, building agent solutions, and using Azure Content Understanding. AI-901 expects basic Python literacy; AI-900 did not.
Does my AI-900 certification still count?
Yes. Microsoft Certified: Azure AI Fundamentals is a single certification whether it was earned with AI-900 or AI-901. If you already passed AI-900, the certification stays on your Microsoft Learn transcript and, like other fundamentals certifications, does not expire. You do not need to retake anything.
Can I still take AI-900?
No. AI-900 retired on June 30, 2026 and can no longer be scheduled. To earn Microsoft Certified: Azure AI Fundamentals now, you pass AI-901.
Is AI-901 harder than AI-900?
Yes, meaningfully harder. The conceptual domain in AI-901 (40–45%) is comparable to AI-900 difficulty, but the implementation domain (55–60%) demands Foundry portal familiarity and the ability to read Python SDK snippets, neither of which AI-900 tested. Plan 2–4 weeks for AI-901 if you already know AI fundamentals and basic Python; 4–6 weeks if you are new to both. AI-900 was typically passed in 1–2 weeks.
Do I need Python to take AI-901?
You need basic Python literacy, but you will not write programs from scratch. You must be able to read short Python snippets that use the Foundry SDK and identify what they do. Because AI-900 is no longer available, there is no code-free route to this certification, so plan a weekend or two learning Python syntax if you do not already read it.
How much does AI-901 cost?
AI-901 costs $99 USD, the standard Microsoft Fundamentals exam price in the US. Pricing varies by country. Microsoft frequently issues free or discounted vouchers through Cloud Skills Challenges, Microsoft Learn virtual training days, and partner programs, so check the Microsoft Learn events page before paying full price. Fundamentals certifications do not expire.
Should I take AI-901 if I already have AI-900?
Only if you do hands-on Azure AI work. Passing AI-901 does not reissue the certification or change its earned date, but it validates the Foundry implementation skills (agents, Content Understanding, Python SDK) that AI-900 never tested. For developers, AI architects, and AI product managers that is a useful signal. For people who took AI-900 for awareness, it adds little.
Will AI-901 prepare me for AI-103?
Yes. AI-901 is the natural entry-level credential before AI-103. The Foundry portal, SDK basics, agent fundamentals, and Content Understanding concepts it introduces are foundational to AI-103. Candidates who pass AI-901 typically need 4–8 weeks of additional preparation for AI-103, versus 10–12 weeks when they arrive without the Foundry implementation foundation.
I was halfway through AI-900 prep. What should I do?
Switch to AI-901. Roughly 40–45% of AI-901 is the AI-900 conceptual material you have already studied: Responsible AI, ML basics, vision, NLP, and generative AI. Keep your notes, then add the Foundry implementation domain (portal, SDK snippets, agents, Content Understanding) and practise with AI-901-specific questions.
Preparing for AI-901?
500 practice questions covering the Foundry portal, SDK, agents, and Content Understanding, plus the AI concepts carried over from AI-900.
Practice AI-901 Free