AI-901 · Microsoft
Microsoft Certified: Azure AI Fundamentals
AI-901 replaces AI-900, retired by Microsoft on 30 June 2026, and awards the same credential.
AI-901 replaces AI-900, which Microsoft retired on 30 June 2026. It awards the same credential, Microsoft Certified: Azure AI Fundamentals. If you have been studying for AI-900, this is the exam you now sit. It is a foundational certification, but the two domains are not weighted evenly and the larger one is not conceptual. Identifying AI concepts and capabilities is about 45% of the exam. The other 55% is implementing AI solutions with Microsoft Foundry, and it assumes basic Python and enough Azure familiarity to find your way around the portal. A question bank is the right tool for the concepts, the service choices and the judgement calls, and this one covers responsible AI, model components, AI workload types, and Foundry implementation including agents and Content Understanding. It is not a substitute for time in the portal. Build something small in Foundry alongside this. Like the real exam, this bank includes multiple-response questions where more than one option is correct.
Who sits this exam
Anyone who needs to speak about Azure AI accurately - and, for the larger half of the exam, actually build with it. The Foundry portion assumes basic Python and enough Azure familiarity to find your way around the portal, so it is a fundamentals exam with a hands-on second act rather than a purely conceptual one.
What it covers
Each domain below is weighted the way the real exam weights it, and scored separately, so you can see which one is holding the total down.
- 01Identify AI concepts and capabilities
- 02Implement AI solutions by using Microsoft Foundry
How you will practise
- Review
- Practice with answers explained immediately.
- Section
- Review one domain at a time.
- Timed
- A timed set with results at the end.
- Final
- The full exam simulation — every question, timed.
What passing proves
That you know what the AI actually is before you ship it: the responsible-AI principles and when each one applies, what a model can and cannot be relied on to do, and how to deploy one in Microsoft Foundry, ground it, filter it, and check its output before a user sees it.