Summary
The AI-900: Microsoft Azure AI Fundamentals exam is the entry point for anyone looking to validate their understanding of AI concepts on the Microsoft platform. In 2025–2026, the exam received a significant refresh to reflect the generative AI era — covering large language models, Azure OpenAI Service, and Responsible AI in a way the original version simply did not.
This post breaks down what AI-900 covers today, what has changed, how to prepare efficiently, and where it sits in the broader Microsoft AI certification roadmap.
What Is AI-900?
AI-900 is a fundamentals-level exam that validates foundational knowledge of:
- Machine learning concepts (no hands-on coding required)
- Azure AI Services (Vision, Language, Speech, Decision)
- Azure OpenAI Service and generative AI
- Responsible AI principles
AI-900 does not require any prior coding experience. It is designed for business analysts, project managers, consultants, and developers who want a solid grounding in AI — not just engineers.
What Changed in the 2025–2026 Update
The exam was substantially refreshed to remove outdated content and add modern AI topics. Here is a summary of the changes:
Removed / Reduced
| Topic | Status |
|---|---|
| Older Cognitive Services branding | Reframed as “Azure AI Services” |
| LUIS (Language Understanding) | Replaced by Azure AI Language + CLU |
| QnA Maker references | Replaced by Azure AI Language – Question Answering |
Added / Expanded
| Topic | Details |
|---|---|
| Generative AI fundamentals | LLMs, tokens, prompts, temperature, grounding |
| Azure OpenAI Service | Models available, deployment types, API access |
| Copilot and AI applications | Microsoft 365 Copilot, Copilot Studio at a conceptual level |
| Responsible AI | Fairness, reliability, privacy, security, inclusiveness, transparency, accountability |
| RAG (Retrieval-Augmented Generation) | Conceptual understanding — what it is and why it matters |
AI-900 Exam At a Glance
| Property | Detail |
|---|---|
| Exam code | AI-900 |
| Duration | 60 minutes |
| Questions | ~40–60 (multiple choice, drag-and-drop, case studies) |
| Passing score | 700 / 1000 |
| Cost | ~$165 USD (varies by region) |
| Certification earned | Microsoft Certified: Azure AI Fundamentals |
| Renewal | Free online assessment every year on Microsoft Learn |
Exam Domains (2026 Skills Outline)
Microsoft publishes an official skills outline. The current breakdown is approximately:
| Domain | Weight |
|---|---|
| Describe Artificial Intelligence workloads and considerations | 15–20% |
| Describe fundamental principles of machine learning on Azure | 20–25% |
| 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 | 15–20% |
Always download the official skills outline PDF from the AI-900 exam page before you start studying — it is the only authoritative source.
How to Prepare
Free Resources (Microsoft Learn)
The best starting point is the free Microsoft Learn path specifically designed for AI-900:
- Get started with artificial intelligence
- Explore visual tools for machine learning
- Explore computer vision
- Explore natural language processing
- Explore fundamentals of generative AI
Each path includes hands-on sandboxed exercises — do them, don’t skip them. The practical exposure helps lock in conceptual understanding.
Practice Tests
| Resource | Notes |
|---|---|
| Microsoft Learn – Practice Assessment | Free, built into the exam page. ~50 questions, explains each answer |
| MeasureUp | Official partner, paid, mirrors real exam format |
| Whizlabs | Budget option with a large question bank |
The free Microsoft Learn Practice Assessment is genuinely good. Do it at least twice — once before studying to identify gaps, once after to confirm readiness.
Study Strategy (1–2 weeks)
1
2
3
4
5
6
7
8
9
10
11
Week 1:
Day 1-2: AI fundamentals + ML concepts (modules 1-2)
Day 3-4: Computer Vision + NLP (modules 3-4)
Day 5-6: Generative AI + Azure OpenAI (module 5)
Day 7: Practice test → identify weak areas
Week 2:
Day 1-3: Targeted review of weak areas
Day 4: Full practice test (aim for 80%+ before booking)
Day 5: Light review + rest
Day 6-7: Exam
Key Concepts to Know Cold
Generative AI
- LLM = Large Language Model trained on vast text corpora
- Prompt = input to the model; Completion = model’s output
- Temperature = controls randomness (0 = deterministic, 1+ = creative)
- Grounding = providing context to reduce hallucination (the core idea behind RAG)
- Azure OpenAI Service = Microsoft-managed access to OpenAI models (GPT-4, DALL-E, Whisper) within Azure compliance boundary
Responsible AI Principles
Microsoft’s six principles — know them by name and definition:
| Principle | What It Means |
|---|---|
| Fairness | AI systems treat all people fairly, without bias |
| Reliability & Safety | AI performs consistently and safely |
| Privacy & Security | User data is protected |
| Inclusiveness | AI benefits everyone, including people with disabilities |
| Transparency | AI decisions can be understood and explained |
| Accountability | Humans remain responsible for AI system behaviour |
Azure AI Services to Know
| Service | What It Does |
|---|---|
| Azure AI Vision | Image classification, object detection, OCR, face analysis |
| Azure AI Language | Sentiment analysis, NER, QA, CLU |
| Azure AI Speech | Speech-to-text, text-to-speech, translation |
| Azure AI Document Intelligence | Extracting structured data from forms and documents |
| Azure OpenAI Service | GPT, DALL-E, Embeddings, Whisper |
The Broader Microsoft AI Certification Roadmap
AI-900 is the starting point. Here’s the full picture for 2026:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Fundamentals
└── AI-900: Azure AI Fundamentals ✅ (start here)
Associate
├── AI-102: Azure AI Engineer Associate (builds directly on AI-900)
└── DP-100: Azure Data Scientist Associate
Specialty / Applied Skills
├── AI-3002: Build Azure AI Vision solution
├── AI-3003: Build NLP solution with Azure AI Services
├── AI-3016: Create a multimodal AI solution
└── MS-4006: Copilot for M365 Administrator (Applied Skills)
Expert
└── (No dedicated AI Expert cert yet — AI capabilities are embedded
in solution architect expert paths)
After AI-900: Go to AI-102
AI-102: Azure AI Engineer Associate is the natural next step. It covers:
- Designing and implementing Azure AI solutions end-to-end
- Azure AI Foundry (new portal — model deployments, prompt flow, evaluations)
- RAG patterns with Azure AI Search
- Content safety and responsible AI guardrails in production
- Azure OpenAI fine-tuning and API integration
AI-102 is a hands-on technical exam — you need to be comfortable with the Azure portal, REST APIs, and Python or C# SDK calls. AI-900 gives you the vocabulary; AI-102 tests the engineering.
Is AI-900 Worth It in 2026?
Yes, for these audiences:
- Non-technical professionals (PMs, consultants, analysts) building AI literacy
- Developers entering the AI space for the first time
- M365 admins who want to understand the AI layer in Copilot and Azure
- Anyone starting toward AI-102 who wants a confidence-building foundation first
You might skip it if:
- You already have AI-102 or DP-100 — the fundamentals content is covered at more depth there
- You have 2+ years of hands-on Azure AI development experience
Conclusion
AI-900 in 2026 is a meaningfully updated certification — the generative AI and Azure OpenAI content alone makes it more relevant than it was two years ago. If you’re working in the Microsoft ecosystem and want a credentialed way to show AI fluency, this is a low-cost, high-value investment of 1–2 focused weeks.
Start with the free Microsoft Learn paths, run the free practice assessment, and you’ll be well-positioned to pass — and more importantly, to actually understand what’s happening under the hood in the AI tools you use every day.
References
- AI-900 Exam Page – Microsoft Learn
- Free Practice Assessment for AI-900
- Explore Fundamentals of Generative AI – Learning Path
- Azure OpenAI Service Documentation
- Microsoft Responsible AI Principles
- AI-102 – Azure AI Engineer Associate