The Microsoft Certified: Azure AI Fundamentals certification is Microsoft's entry point for AI literacy on Azure. It is designed for learners who need to understand common AI workloads, basic machine learning principles, and the main Azure AI service categories before moving into hands-on AI engineering or data-science roles.
This certification is in an active transition. As of May 28, 2026, Microsoft still offers AI-900, but the official exam page states that AI-900 retires on June 30, 2026 and will be replaced by AI-901. The certification page currently lists AI-900 and AI-901 as the required exam options for Azure AI Fundamentals. In this repo, the current internal practice route already uses the slug ai-901, so that is the practice link used below.
This is still a fundamentals certification. Microsoft is testing whether you can recognize the right AI concept or Azure AI service family for a scenario. That means your preparation should focus on service positioning, workload recognition, responsible AI basics, and the distinction between traditional ML, computer vision, natural language processing, and generative AI.
Exam At a Glance
| Attribute | Value |
|---|---|
| Certification | Microsoft Certified: Azure AI Fundamentals |
| Current exam transition | AI-900 retires on 2026-06-30; AI-901 is the replacement path now listed on the certification page |
| Stable public exam scope | AI-900 |
| Repo practice slug | ai-901 |
| Level | Fundamentals |
| Duration | 45 minutes |
| Cost | $99 USD |
| Prerequisites | No formal prerequisite; Microsoft recommends familiarity with basic cloud concepts and client-server applications |
| Target candidate | Beginners, students, developers, analysts, and technical professionals building foundational AI literacy on Azure |
- Official certification page: Microsoft Certified: Azure AI Fundamentals
- Official AI-900 exam page: Exam AI-900: Microsoft Azure AI Fundamentals
- Official AI-900 study guide: AI-900 study guide
- Official learning paths: AI concepts for developers and technology professionals, Get started with AI applications and agents on Azure
Official Assessed Areas
- Describe Artificial Intelligence workloads and considerations
- Describe fundamental principles of machine learning on Azure
- Describe features of computer vision workloads on Azure
- Describe features of Natural Language Processing (NLP) workloads on Azure
- Describe features of generative AI workloads on Azure
The AI-900 public exam page currently still provides the clearest weighted outline, so this guide follows that structure. Where relevant, it also acknowledges that Microsoft is moving the certification path forward through AI-901.
1. Describe Artificial Intelligence Workloads and Considerations
This first section is about AI vocabulary. You need to understand what kinds of problems AI solves before you try to map those problems to Azure services.
- Core AI workload categories - Study common AI workload types such as machine learning, computer vision, natural language processing, and generative AI. Microsoft expects you to identify these from short scenario descriptions. Official resources: AI concepts for developers and technology professionals, AI-900 exam page.
- Responsible AI principles - Microsoft treats responsible AI as part of the fundamentals conversation. Be comfortable with fairness, reliability, safety, privacy, inclusiveness, transparency, and accountability at a high level. Official resource: AI concepts learning path.
- AI is about matching the workload to the tool - The exam wants you to recognize what kind of AI problem you are solving before you select the Azure service. Official resources: AI concepts path, Get started with AI applications and agents on Azure.
- Do not overcomplicate the scenario - Fundamentals questions usually reward the cleanest classification rather than advanced model-design knowledge. Official resource: AI-900 exam page.
Exam tip: If the question is asking what kind of AI problem is being solved, answer that first. Only then decide which Azure AI service family best fits it.
2. Describe Fundamental Principles of Machine Learning on Azure
This section covers baseline machine learning literacy rather than deep data-science implementation.
- Supervised, unsupervised, and reinforcement learning basics - Know the difference between training with labeled data, finding structure without labels, and optimizing behavior through reward signals. Official resource: AI concepts learning path.
- Training, validation, and prediction - You should understand the simple lifecycle of preparing data, training a model, evaluating it, and using it for inference. Official resource: AI concepts path.
- Azure machine learning positioning - Fundamentals-level questions usually test whether you understand Azure's role as a managed AI platform rather than expecting model-building depth. Official resources: Get started with AI applications and agents on Azure, AI-900 exam page.
- Conceptual accuracy matters more than math - AI fundamentals is not a statistics exam. It is a recognition exam about how machine learning works in broad terms. Official resource: AI concepts path.
Exam tip: If the question sounds like data-labeling, prediction, classification, regression, clustering, or model training, stay at the conceptual level. Microsoft is usually not looking for algorithm trivia here.
3. Describe Features of Computer Vision Workloads on Azure
This section tests whether you understand what vision workloads do and where Azure AI services fit.
- Image analysis and visual understanding - Be able to distinguish image classification, object detection, OCR, facial analysis, and document understanding style workloads. Official resources: AI-900 exam page, AI applications and agents path.
- Azure service positioning for vision scenarios - The exam usually asks which Azure AI capability fits a vision use case, not how to tune a model. Official resource: Get started with AI applications and agents on Azure.
- Document and image workloads are not the same as general ML - Microsoft likes these service-family distinctions because they reflect real product choices on Azure. Official resources: AI applications and agents path, AI-900 exam page.
Exam tip: When the prompt mentions images, text extraction from documents, or identifying objects in visual content, think vision workload before you think broader machine-learning platform.
4. Describe Features of Natural Language Processing Workloads on Azure
This section focuses on language understanding, speech, text analytics, and conversational AI scenarios.
- Text and language workloads - Study translation, sentiment analysis, entity recognition, summarization, question answering, and conversational scenarios at a high level. Official resources: AI-900 exam page, AI applications and agents path.
- Speech and conversational interfaces - Microsoft groups speech and language capabilities into the broader NLP family for this level of exam reasoning. Official resource: Get started with AI applications and agents on Azure.
- Language questions are usually about capability recognition - Microsoft wants you to choose the right family of Azure AI features for a language problem, not design a linguistic model from scratch. Official resources: AI concepts path, AI-900 exam page.
Exam tip: If the scenario is about text, speech, or conversational understanding, classify the NLP task first, then match it to the Azure AI capability family.
5. Describe Features of Generative AI Workloads on Azure
Generative AI is now a formal part of Azure AI Fundamentals, so you need basic literacy here even though this is still an entry-level exam.
- Generative AI concepts - Understand prompts, completions, copilots, agents, large language models, and the basic idea of grounding or enriching outputs with enterprise context. Official resources: Get started with AI applications and agents on Azure, AI-900 exam page.
- Azure generative AI positioning - At fundamentals level, Microsoft is testing whether you know where generative AI fits in Azure's service landscape and what kinds of use cases it supports. Official resource: AI applications and agents path.
- Expect transition-related wording - Because the certification is moving from AI-900 toward AI-901, generative AI appears even more prominently in the current Microsoft material. Official resources: Certification page, AI-900 exam page.
Exam tip: Generative AI questions at this level are about what the workload does and when you would use it, not about model architecture internals.
Recommended 4-Week Study Plan
| Week | Focus | Primary resources |
|---|---|---|
| 1 | AI workload categories, responsible AI, Azure AI service landscape | AI concepts learning path, certification page |
| 2 | Machine learning principles, training and inference basics | AI concepts path, AI-900 exam page |
| 3 | Computer vision and NLP workload recognition | AI applications and agents path, AI-900 exam page |
| 4 | Generative AI workloads, transition notes, practice review | AI applications and agents path, AI-900 study guide, practice questions |
Last-Mile Exam Strategy
- Study by workload family. That is the fastest way to reduce confusion between machine learning, vision, language, and generative AI questions.
- Use the AI-900 exam outline for structure, but keep an eye on the AI-901 transition note because Microsoft is actively evolving this certification path.
- Do not over-study implementation depth. Azure AI Fundamentals rewards conceptual clarity and service recognition more than code or SDK fluency.
- Memorize the common workload-to-service pattern: identify the AI problem first, then pick the Azure service family that fits it.
- Review responsible AI deliberately. It is a fundamentals exam, and Microsoft expects policy and principles awareness, not just feature names.
If you want exam-style reinforcement after the official docs, use our Azure AI Fundamentals practice questions. If you want the broader Azure and Microsoft fundamentals context around this certification, pair this with our Azure Fundamentals study guide.
The fastest way to pass Azure AI Fundamentals is to treat it as a classification exam for AI workloads on Azure. Learn the official Microsoft terminology, map each workload to its use case, and keep the current AI-900 to AI-901 transition in mind while you prepare.