About the Azure AI Fundamentals (AI-900) Exam
Key Takeaways
The AI-900 exam validates foundational knowledge of AI and machine learning concepts and related Microsoft Azure services — no coding experience required.
Five domains are tested: AI Workloads and Considerations (15-20%), Machine Learning Principles (15-20%), Computer Vision (15-20%), NLP (15-20%), and Generative AI (20-25%) — generative AI is now the single heaviest domain.
Microsoft does not publish the AI-900 question count (most Microsoft exams have 40-60); the exam has a 45-minute time limit and requires a scaled score of 700/1000 to pass.
The exam cost $99 USD and could be taken at a Pearson VUE test center or via online proctoring.
Microsoft retired AI-900 on June 30, 2026; AI-901 now earns the Azure AI Fundamentals certification.
Quick Answer: The AI-900 exam has a 45-minute time limit and requires 700/1000 to pass; Microsoft does not publish its question count (most Microsoft exams have 40-60). It covers five domains spanning AI workloads, machine learning, computer vision, NLP, and generative AI on Azure. The exam cost $99 USD with no prerequisite certifications required. Microsoft retired it on June 30, 2026; AI-901 replaced it.
The Microsoft Azure AI Fundamentals (AI-900) certification validates your foundational understanding of artificial intelligence and machine learning concepts, as well as related Microsoft Azure services. Unlike associate-level certifications such as the AI-102, the AI-900 is a conceptual exam — it tests your understanding of what AI services do and when to use them, not your ability to write code or build solutions.
This certification is ideal for both technical and non-technical professionals who work with or around AI-powered solutions. Whether you are a business analyst evaluating AI capabilities, a project manager overseeing an AI initiative, or a developer beginning your AI journey, the AI-900 establishes your foundational AI knowledge on the Microsoft platform.
Exam Format at a Glance
| Detail | Specification |
|---|---|
| Exam Code | AI-900 |
| Official Name | Microsoft Azure AI Fundamentals |
| Certification | Microsoft Certified: Azure AI Fundamentals |
| Questions | Not published (most Microsoft exams have 40-60) |
| Time Limit | 45 minutes (65 minutes seat time including NDA and setup) |
| Question Types | Multiple-choice, multiple-answer, drag-and-drop, scenario-based |
| Passing Score | 700 out of 1,000 (scaled) |
| Cost | $99 USD |
| Languages | English, Japanese, Korean, Simplified Chinese, French, German, Spanish, Portuguese, Arabic, Indonesian, Italian |
| Delivery | Pearson VUE test center or online proctoring |
| Validity | Does not expire (Fundamentals certification) |
| Prerequisites | None |
| Retirement Date | June 30, 2026 |
Domain Breakdown
| Domain | Weight | Topic |
|---|---|---|
| Domain 1 | 15-20% | Describe Artificial Intelligence Workloads and Considerations |
| Domain 2 | 15-20% | Describe Fundamental Principles of Machine Learning on Azure |
| Domain 3 | 15-20% | Describe Features of Computer Vision Workloads on Azure |
| Domain 4 | 15-20% | Describe Features of Natural Language Processing Workloads on Azure |
| Domain 5 | 20-25% | Describe Features of Generative AI Workloads on Azure |
On the Exam: Following the May 2, 2025 exam update, Generative AI (20-25%) is the single heaviest domain — the machine-learning domain's weight was reduced and the generative-AI weight increased. The other four domains are each 15-20%. Combined, Generative AI plus Machine Learning can account for 35-45% of the exam, and Computer Vision plus NLP form another 30-40%. Responsible AI principles appear across ALL five domains, not just Domain 1.
Exam Retirement Notice
Important: Microsoft retired the AI-900 exam on June 30, 2026. You can no longer take AI-900; the replacement exam is AI-901, which now earns the Azure AI Fundamentals certification.
What this means:
- New candidates take AI-901
- Certifications already earned through AI-900 remain valid; Fundamentals certifications don't expire
Question Types Explained
Multiple-Choice (Single Select)
You are presented with a question and four answer options. One answer is correct. These often test your ability to identify the correct Azure AI service or concept for a given scenario.
Multiple-Answer (Multi-Select)
You are given a question with five or more options and asked to select two or three correct answers. Common for questions asking you to identify multiple responsible AI principles or Azure AI service capabilities.
Drag-and-Drop (Matching)
You may need to match AI concepts to their definitions, Azure services to their capabilities, or machine learning types to their use cases.
Yes/No Statement Groups
A set of related statements where you must evaluate each as true or false. These test your ability to distinguish between similar concepts.
Scenario-Based
A short scenario (1-2 paragraphs) followed by a question. These test your ability to identify the appropriate AI approach, Azure service, or responsible AI consideration for a business scenario.
Tip: Unlike the AI-102, the AI-900 does NOT include case studies. All questions are standalone, and you can navigate back and forth between questions. Use this to your advantage — flag uncertain questions and return to them.
Who Should Take This Exam?
The AI-900 is designed for professionals who:
- Business stakeholders evaluating AI opportunities and understanding AI terminology
- Project managers overseeing AI-related projects who need foundational AI knowledge
- IT professionals beginning their AI and cloud journey
- Sales and marketing professionals working with AI-powered products
- Students pursuing a career in AI, data science, or cloud computing
- Developers who want a broad understanding of AI before specializing
- Anyone curious about AI and how Azure implements AI services
Prerequisites
Microsoft states there are no formal prerequisite certifications for the AI-900. The exam is designed for both technical and non-technical audiences. However, familiarity with the following is helpful:
- Basic understanding of cloud computing — what cloud services are and how they differ from on-premises solutions
- General awareness of AI concepts — what AI, ML, and deep learning mean at a high level
- Ability to navigate the Azure portal — understanding how to find and explore Azure services (helpful but not required)
- No programming required — unlike the AI-102, the AI-900 does not test coding skills
On the Exam: The AI-900 intentionally avoids code-level questions. You will never need to write Python, C#, or API calls. Questions focus on concepts, service selection, and responsible AI principles.
How to Register
- Visit Microsoft Learn Certifications
- Click Schedule exam and sign in with your Microsoft account
- Select Pearson VUE as the exam delivery provider
- Choose a test center near you or select online proctoring
- Pick your preferred date and time
- AI-900 can no longer be booked; register for AI-901 ($99 USD in the United States).
Discounts and Vouchers
- Microsoft Enterprise Skills Initiative: Employees at participating organizations may receive free exam vouchers
- Microsoft Learn Cloud Skills Challenges: Periodic challenges that award exam vouchers upon completion
- Student discounts: Students with a verified .edu email may receive discounted pricing
- Microsoft Virtual Training Days: Free one-day events that sometimes include a free exam voucher for AI-900
- Employer reimbursement: Many employers reimburse certification exam costs
What to Expect on Exam Day
At a Pearson VUE test center:
- Arrive 15 minutes early
- Bring two forms of valid identification (one must be government-issued photo ID)
- All personal items are stored in a locker
- No notes, phones, or smartwatches allowed in the testing room
- A whiteboard or erasable notepad may be provided for scratch work
For online proctoring:
- Government-issued photo ID required
- A quiet, private room with a clear desk
- Webcam and microphone are mandatory
- Stable internet connection
- No secondary monitors, phones, or reference materials
- A proctor monitors you via webcam throughout the exam
Scoring
The exam uses a scaled scoring model ranging from 1 to 1,000. The passing score is 700. The scaled score does not translate directly to a percentage — a score of 700 does not mean you answered 70% of questions correctly. After completing the exam, you receive a pass/fail notification on screen immediately. Your detailed score report showing performance by domain is available in your Microsoft Learn profile within minutes.
Certification Renewal
Azure AI Fundamentals is a Fundamentals certification: it does not expire and has no renewal assessment.
Career Path
| Certification | Level | Exam | Progression |
|---|---|---|---|
| Azure AI Fundamentals | Fundamentals | AI-900 | You are here |
| Azure AI Apps and Agents Developer Associate | Associate | AI-103 | Implementation skills |
| Machine Learning Operations Engineer Associate | Associate | AI-300 | ML operations path |
| Azure Solutions Architect Expert | Expert | AZ-305 | Architecture path |
Career Tip: The AI-900 is an excellent entry point, but to demonstrate hands-on skills to employers, pursue the AI-103 (Azure AI Apps and Agents Developer Associate) or AI-300 (Machine Learning Operations Engineer Associate) as your next certification. The AI-900 alone signals awareness; associate-level certs signal capability.
What is the minimum passing score for the AI-900 exam?
700 out of 1,000
600 out of 1,000
750 out of 1,000
800 out of 1,000
Which domain carries the most weight on the AI-900 exam?
Describe AI Workloads and Considerations (15-20%)
Describe Fundamental Principles of Machine Learning on Azure (15-20%)
Describe Features of Computer Vision Workloads (15-20%)
Describe Features of Generative AI Workloads (20-25%)
How much time do you have to complete the AI-900 exam?
45 minutes
30 minutes
60 minutes
120 minutes
When did Microsoft retire the AI-900 exam?
December 31, 2025
March 31, 2026
June 30, 2026
December 31, 2026
Which of the following is required to take the AI-900 exam?
Azure AZ-900 certification
At least one year of Azure experience
Python or C# programming proficiency
No prerequisites are required
Sections you finish are checked off in the contents.