AI-900 Exam Pattern, Question Types & Scoring Explained

The AI-900: Microsoft Azure AI Fundamentals exam is an entry-level Microsoft certification exam designed for candidates who want to validate their understanding of artificial intelligence concepts and Azure AI services. It is suitable for both technical and non-technical learners, and Microsoft clearly states that data science or software engineering experience is not required. However, basic awareness of cloud concepts and client-server applications can be helpful.
What Is the AI-900 Exam?
The AI-900 exam validates foundational knowledge of AI workloads, machine learning principles, computer vision, natural language processing, generative AI, and the Azure services used to support these workloads. It is commonly chosen by beginners, students, business professionals, cloud learners, and early-career technology professionals who want to start their Microsoft AI certification journey.
AI-900 is part of the Microsoft Certified: Azure AI Fundamentals certification. Microsoft has announced that the AI-900 exam will retire on June 30, 2026, and will be replaced by AI-901. Candidates can continue earning the Azure AI Fundamentals certification after AI-900 retires by passing AI-901.
AI-900 Exam Pattern
Microsoft Fundamentals exams generally have a shorter exam duration than associate or expert-level exams. Microsoft lists 45 minutes as the exam duration for Fundamentals exams and 65 minutes as the total seat duration. Seat duration includes time for instructions, agreement review, exam questions, and optional comments after the exam.
Most Microsoft Certification exams typically contain 40 to 60 questions, although the exact number can change because Microsoft regularly updates exams to stay aligned with current technology and job-role expectations.
For AI-900, candidates should expect a concept-based exam rather than a deep hands-on engineering exam. The focus is not on writing production-level code or designing complex enterprise AI systems. Instead, the exam checks whether you can identify AI workloads, understand core AI concepts, and map business scenarios to the correct Azure AI services.
AI-900 Skills Measured
The current AI-900 exam measures the following domains:
Skill Area Weightage
Describe Artificial Intelligence workloads and considerations 15–20%
Describe fundamental principles of machine learning on Azure 15–20%
Describe features of computer vision workloads on Azure 15–20%
Describe features of Natural Language Processing workloads on Azure 15–20%
Describe features of generative AI workloads on Azure 20–25%
Generative AI has the highest listed weightage, so candidates should pay extra attention to Azure OpenAI-related concepts, responsible AI, prompt-based solutions, and how generative AI fits into business use cases.
AI-900 Question Types
AI-900 questions are usually scenario-based and concept-driven. Microsoft certification exams may include different question formats such as multiple choice, drag-and-drop, build list, hot area, active screen, and case-study-style questions. Microsoft also provides an exam sandbox so candidates can experience the exam interface and question formats before the real exam.
Common question formats may include:
Question Type What It Tests
Single-choice questions Selecting one correct answer from multiple options
Multiple-response questions Selecting more than one correct answer
Drag-and-drop Matching concepts, services, or steps
Build-list / ordering Arranging steps or items in the correct order
Scenario-based questions Choosing the right Azure AI service for a business requirement
Hot-area / active-screen questions Identifying the correct option or area in a visual interface
The AI-900 exam is more about understanding and service selection than advanced configuration. For example, a question may describe a company that wants to extract text from scanned invoices and ask which Azure AI service should be used.
AI-900 Passing Score
The passing score for AI-900 is 700. Microsoft technical certification exams are scored on a scale of 1 to 1,000, and a score of 700 or greater is required to pass.
However, 700 does not mean 70%. Microsoft uses a scaled scoring model. The final score is adjusted based on the difficulty of the questions and the level of knowledge required to demonstrate competence.
In simple terms, the goal is not just to answer a fixed percentage of questions correctly. The goal is to prove that you understand the required fundamentals at Microsoft’s expected standard.
How AI-900 Scoring Works
Microsoft states that most multi-part questions award points for each correctly answered component. This means you may receive full, partial, or no credit depending on the question structure. If a question is worth more than one point, this is usually indicated in the exam interface.
There is no negative marking in Microsoft certification exams. If you answer incorrectly, points are not deducted. You simply do not earn the point for that question or part of the question. Because of this, candidates should attempt every question rather than leaving questions unanswered.
Microsoft may also include unscored questions in the exam. These questions are used to collect exam-quality data and may appear like normal scored questions. Candidates are not told which questions are unscored, so every question should be answered seriously.
Does AI-900 Have Labs?
AI-900 is a Fundamentals-level exam, so it is primarily designed to test conceptual understanding rather than deep hands-on implementation. Microsoft does mention that certification exams can contain various question types, and exam instructions at launch explain what to expect for that specific exam.
For AI-900 preparation, candidates should focus on concepts, service identification, workload classification, responsible AI principles, and basic Azure AI capabilities. Hands-on practice is still helpful, but the exam is not positioned like an associate-level implementation exam.
How to Prepare for AI-900 Question Types
The best preparation strategy is to study the official skills measured and then practice with scenario-based questions. Microsoft’s AI-900 study guide explains that the exam is updated periodically and that most questions cover generally available features, though preview features may appear if they are commonly used.
Microsoft also provides free Practice Assessments on Microsoft Learn. These assessments help candidates understand question style, wording, difficulty, and readiness gaps, but Microsoft clearly states that practice questions are not the same as real exam questions and are not a replacement for training or hands-on experience.
Key Preparation Areas for AI-900
Candidates should be comfortable with the following:
• Difference between AI, machine learning, deep learning, and generative AI
• Common AI workloads such as prediction, classification, object detection, OCR, speech, translation, and text analytics
• Responsible AI principles such as fairness, reliability, privacy, transparency, and accountability
• Azure Machine Learning basics
• Azure AI Vision, Azure AI Language, Azure AI Speech, Azure AI Search, and Azure OpenAI concepts
• Choosing the correct Azure AI service for a given business scenario
• Basic understanding of prompts, completions, generative AI use cases, and risks