Microsoft AI-900: Azure AI Fundamentals Practice Exam
Cover all five AI-900 skill areas — Generative AI, AI workloads and responsible AI, machine learning, computer vision, and natural language processing — with objective-mapped practice questions, immediate feedback in Learn mode, and full timed simulation in Exam mode.
Start 24-hour free trial →AI-900 has retired — the Azure AI Fundamentals certification continues as AI-901
Microsoft retired the AI-900 exam on June 30, 2026. New candidates now earn the same Microsoft Certified: Azure AI Fundamentals credential by passing its replacement, Microsoft AI-901, which is refreshed around generative AI, Copilot, and Microsoft Foundry. If you already passed AI-900, your certification is unaffected — Fundamentals credentials never expire.
This page and its practice questions cover the AI-900 objectives and remain useful for reviewing the shared fundamentals; if you are starting fresh, plan for AI-901. Always confirm current exam status on Microsoft Learn.
AI-900 exam at a glance
- Vendor
- Microsoft
- Exam code
- AI-900 (retired June 30, 2026; replaced by AI-901)
- Certification
- Microsoft Certified: Azure AI Fundamentals
- Level
- Fundamentals (foundational)
- Blueprint
- Skills Measured as of May 2, 2025 (the final AI-900 edition, which added the Generative AI domain)
- Format
- Typically 40–60 questions; multiple choice, multiple response, drag-and-drop, and scenario-based items
- Duration
- 45 minutes of exam (seat) time
- Passing score
- 700 of 1000 (scaled)
- Delivery
- Pearson VUE test center or online proctored
- Prerequisites
- None. Suitable for both technical and non-technical candidates; basic cloud familiarity helps but is not required
- Cost (USD)
- $99 USD list price (varies by region)
- Validity
- The Fundamentals certification does not expire
- Languages
- English and multiple additional languages (verify current list with Microsoft)
Source: Microsoft Learn — AI-900 study guide (skills measured) and the Azure AI Fundamentals certification page. Verify current details with Microsoft before scheduling.
About the Microsoft Certified: Azure AI Fundamentals certification
Azure AI Fundamentals is Microsoft’s foundational AI credential. It validates a broad, conceptual understanding of artificial intelligence and machine learning and how Microsoft’s Azure AI services map to real workloads — computer vision, natural language processing, and generative AI — along with responsible AI principles. It is deliberately non-technical: no coding or data-science experience is required, which makes it popular with product managers, analysts, consultants, and career changers who need AI literacy fast, as well as engineers wanting a structured starting point.
The AI-900 exam was refreshed on May 2, 2025 to add a dedicated Generative AI domain — now the single heaviest area — covering Azure OpenAI Service, Azure AI Foundry, large language models, and responsible AI for generative workloads. AI-900 retired on June 30, 2026 and is replaced by AI-901, which carries the same certification and leans further into generative AI and Microsoft Foundry. The concepts on this page — classification, regression, computer vision, NLP, and responsible AI — carry over and remain a solid foundation. For deeper study, see our Responsible AI Ethics guide and Azure AI services deep dive.
Every PowerKram practice question maps to one of the five AI-900 skill areas and links to the specific Microsoft Learn page it was derived from, so your weak spots become a focused reading list rather than a guess.
AI-900 skill areas and weights
Five skill areas, with Generative AI the single heaviest after the May 2025 refresh. Microsoft publishes each area as a weighting range rather than an exact percentage; the ranges below are reproduced as Microsoft states them. Plan your study time roughly in proportion.
What generative AI is and its core concepts; large language models; the Azure OpenAI Service and Azure AI Foundry; and responsible AI considerations specific to generative workloads.
Identifying common AI workloads (prediction, anomaly detection, computer vision, NLP, generative AI) and the guiding principles of responsible AI — fairness, reliability, privacy, inclusiveness, transparency, and accountability.
Core ML concepts — regression, classification, and clustering; training and validation; and Azure Machine Learning capabilities such as automated ML and the designer.
Image analysis, object detection, optical character recognition (OCR), and facial detection, and the Azure AI Vision services that provide them.
Key-phrase extraction, entity recognition, sentiment analysis, language detection, and speech, and the Azure AI Language and Speech services behind them.
Source: Microsoft Learn — AI-900 study guide (skills measured, as of May 2, 2025). Microsoft publishes weightings as ranges; several areas share the 15–20% band.
Who AI-900 is for
Azure AI Fundamentals is aimed at anyone who wants foundational AI literacy on Azure, technical or not:
- Product managers, analysts, and consultants who scope or evaluate AI features and need to speak the language of Azure AI services confidently.
- Career changers and students building a first, recognized credential in AI without a coding or data-science background.
- Developers and IT professionals wanting a structured starting point before deeper, role-based certifications.
- Business and non-technical stakeholders who need to understand responsible AI and where generative AI fits.
There are no prerequisites. Because AI-900 has retired, new candidates should plan for AI-901, which earns the same certification; this material remains a solid way to review the shared fundamentals. A natural next step is the associate-level AI-102: Azure AI Engineer, and those newer to Azure often pair AI-900 with AZ-900: Azure Fundamentals. For the roles this credential supports — with skills, tools, and salary ranges — see the AI Engineer career path.
What this AI-900 practice exam delivers
Learn mode
Get the correct answer, the explanation, and a direct link to the exact Microsoft Learn page each question was derived from — immediately after each question. Best for the concept-matching style AI-900 uses across all five skill areas.
Exam mode
A timed run in the real AI-900 shape — roughly 40 to 60 questions in 45 minutes — including multiple-response, drag-and-drop, and scenario items, so you build pacing before test day.
Source-linked explanations
Every answer cites the exact Microsoft Learn page it was built from — so you can verify service capabilities and dig deeper, not just memorize.
Score by skill area
Results break down by the five AI-900 areas — Generative AI, AI workloads and responsible AI, machine learning, computer vision, and NLP — so practice tells you exactly which area to revisit.
Sample AI-900 practice questions
Ten free questions across the five AI-900 skill areas, with full explanations and source links to the Microsoft Learn pages each is derived from. The complete bank is available with the 24-hour trial.
Which Azure offering provides a unified platform for building, deploying, and managing generative AI applications, including access to large language models?
- Azure AI Foundry
- Azure Blob Storage
- Azure Virtual Machines
- Azure DNS
Show answer & explanation
Correct: A — Azure AI Foundry. Azure AI Foundry is Microsoft’s unified platform for building, deploying, and managing AI and generative AI applications, including access to large language models such as those in the Azure OpenAI Service — the generative-AI focus of the refreshed AI-900.
Why not the others: Blob Storage (B) stores objects; Virtual Machines (C) provide compute; DNS (D) resolves domain names. None is a generative-AI development platform.
Source: Microsoft Learn — What is Azure AI Foundry? → Further reading: PowerKram — generative AI & large language models →A team wants to build a chatbot on top of GPT models hosted and secured within their Azure subscription. Which Azure service provides managed access to these large language models?
- Azure Traffic Manager
- Azure OpenAI Service
- Azure Site Recovery
- Azure Batch
Show answer & explanation
Correct: B — Azure OpenAI Service. The Azure OpenAI Service provides managed, secured access to large language models such as the GPT family within a customer’s Azure environment — the standard AI-900 answer for enterprise access to these models.
Why not the others: Traffic Manager (A) is DNS-based traffic routing; Site Recovery (C) is disaster recovery; Batch (D) runs large-scale batch compute. None provides managed LLM access.
Source: Microsoft Learn — What are Azure AI services? →A bank wants to ensure its AI loan-approval model treats applicants equitably and does not disadvantage particular groups. Which responsible AI principle does this primarily address?
- Scalability
- Latency
- Fairness
- Compression
Show answer & explanation
Correct: C — Fairness. Fairness is the Microsoft responsible AI principle concerned with treating all people equitably and avoiding bias against particular groups — exactly the loan-approval concern described. The six principles are fairness, reliability and safety, privacy and security, inclusiveness, transparency, and accountability.
Why not the others: scalability (A), latency (B), and compression (D) are performance or engineering concerns, not responsible AI principles.
Source: Microsoft Learn — Responsible AI → Further reading: PowerKram — Responsible AI Ethics →A team needs to detect and block harmful text and images — hate, violence, and sexual content — in a generative AI application. Which Azure service is designed for this?
- Azure Content Delivery Network
- Azure Monitor
- Azure Key Vault
- Azure AI Content Safety
Show answer & explanation
Correct: D — Azure AI Content Safety. Azure AI Content Safety detects and moderates harmful content (hate, violence, sexual, self-harm) across text and images, supporting responsible AI in generative applications — the described need.
Why not the others: CDN (A) speeds content delivery; Monitor (B) is observability; Key Vault (C) stores secrets. None moderates harmful content.
Source: Microsoft Learn — Azure AI Content Safety →A company wants to predict a house’s sale price (a continuous numeric value) from features such as size and location. Which type of machine learning task is this?
- Clustering
- Regression
- Object detection
- Language detection
Show answer & explanation
Correct: B — Regression. Regression predicts a continuous numeric value (such as a price) from input features — the described task. Classification predicts a category; clustering groups unlabeled data.
Why not the others: clustering (A) groups similar items without labels; object detection (C) is a computer-vision task; language detection (D) is an NLP task. None predicts a continuous number.
Source: Microsoft Learn — Automated ML (task types) → Further reading: PowerKram — machine learning fundamentals →A data scientist wants to build ML models in Azure Machine Learning using a visual, drag-and-drop interface without writing code. Which capability provides this?
- Azure Resource Manager templates
- Azure CLI scripting
- Azure Machine Learning designer
- Azure Policy
Show answer & explanation
Correct: C — Azure Machine Learning designer. The designer provides a visual, drag-and-drop canvas for building, training, and deploying ML models without writing code — the no-code approach described.
Why not the others: ARM templates (A) and the Azure CLI (B) are code/scripting-based infrastructure tools; Azure Policy (D) enforces governance rules. None is a no-code ML model builder.
Source: Microsoft Learn — Azure ML designer →A business wants to automatically extract printed and handwritten text from scanned documents and images. Which computer vision capability does this?
- Sentiment analysis
- Optical character recognition (OCR)
- Regression
- Key-phrase extraction
Show answer & explanation
Correct: B — Optical character recognition (OCR). OCR extracts printed and handwritten text from images and documents — a core Azure AI Vision capability and exactly the described task.
Why not the others: sentiment analysis (A) and key-phrase extraction (D) are NLP tasks on text; regression (C) predicts numbers. None reads text from images.
Source: Microsoft Learn — OCR (Azure AI Vision) → Further reading: PowerKram — computer vision →A retailer wants to identify and locate multiple products within a single shelf photo, drawing a bounding box around each item. Which computer vision task is this?
- Object detection
- Language translation
- Clustering
- Speech synthesis
Show answer & explanation
Correct: A — Object detection. Object detection identifies multiple objects in an image and returns a bounding box and label for each — the shelf-photo task described. It goes beyond image classification, which assigns a single label to the whole image.
Why not the others: language translation (B) and speech synthesis (D) are language/speech tasks; clustering (C) groups data. None locates objects in an image.
Source: Microsoft Learn — Image Analysis (Azure AI Vision) →A company wants to analyze customer reviews and score each as positive, negative, or neutral. Which NLP capability provides this?
- Optical character recognition
- Object detection
- Sentiment analysis
- Anomaly detection
Show answer & explanation
Correct: C — Sentiment analysis. Sentiment analysis, part of the Azure AI Language service, evaluates text and returns sentiment (positive, negative, neutral, mixed) with confidence scores — exactly the review-scoring task described.
Why not the others: OCR (A) reads text from images; object detection (B) is computer vision; anomaly detection (D) flags unusual data points. None scores text sentiment.
Source: Microsoft Learn — Sentiment analysis (Azure AI Language) → Further reading: PowerKram — natural language processing →A global app needs to automatically determine which language each incoming user message is written in. Which Azure AI Language capability does this?
- Key Vault secret rotation
- Image classification
- Load balancing
- Language detection
Show answer & explanation
Correct: D — Language detection. Language detection, part of the Azure AI Language service, identifies the language a piece of text is written in and returns a confidence score — the described need.
Why not the others: Key Vault rotation (A) manages secrets; image classification (B) is computer vision; load balancing (C) distributes traffic. None detects the language of text.
Source: Microsoft Learn — Language detection (Azure AI Language) →Keep going: Learning & Career resources
AI-900 is the entry point to Azure AI. Both PowerKram hubs back this exam — deeper study material and the roles it leads to.
Deep dive: AI-900 format, the AI-901 transition, study path, and what carries over
Exam format and scoring
AI-900 was a fundamentals exam: roughly 40 to 60 questions in about 45 minutes of seat time, with a passing score of 700 out of 1000 (scaled). Question formats included multiple choice, multiple response, drag-and-drop, and short scenario-based items. There was no penalty for wrong answers, and the exam focused on concepts rather than code or portal configuration. Read the Azure AI services deep dive →
The AI-901 transition — what changed and what carries over
AI-900 retired on June 30, 2026 and is replaced by AI-901, which earns the same Azure AI Fundamentals certification. AI-901 is restructured around generative AI and is centered on Microsoft Foundry, adding coverage of Copilot fundamentals, retrieval-augmented generation (RAG), and grounding. The classic fundamentals on this page — responsible AI principles, supervised learning, classification and regression, computer vision, and NLP — carry over and remain worth knowing. If you are starting fresh, register for AI-901; if you already passed AI-900, your credential is permanent. Microsoft’s official AI-901 exam page →
Realistic study path
Most candidates need one to three weeks. The single best resource is the free Microsoft Learn learning path, which maps to every objective; layer in one practice test and, optionally, a short video course. Focus on concepts, not code: be able to match a business scenario to the right Azure AI service, name the responsible AI principles, and distinguish regression, classification, and clustering. Give the heaviest area — generative AI — extra attention. Read the generative AI & LLMs guide →
Cost and scheduling
The AI-900 list price was $99 USD (regional pricing varies); AI-901 pricing is comparable — check Pearson VUE for current rates. Fundamentals exams are delivered at a Pearson VUE test center or online with a proctor. Microsoft frequently offers discount vouchers through Virtual Training Day events. Azure AI Fundamentals certification page →
Validity and next steps
Microsoft Fundamentals certifications do not expire — once earned, Azure AI Fundamentals stays on your transcript permanently. The natural next step is the associate-level AI-102: Azure AI Engineer, which is role-based and does require periodic (free) renewal. See the AI-102 Azure AI Engineer next step →
Career outlook for Azure AI Fundamentals
Azure AI Fundamentals is a strong first signal of AI literacy for both technical and non-technical roles: product managers, analysts, consultants, and aspiring AI or ML engineers. It pairs well with Azure Fundamentals (AZ-900) for broader platform context and points toward the associate AI-102 for those going deeper. Career Hub — AI Engineer →
Microsoft AI-900 (Azure AI Fundamentals) exam FAQ
Has AI-900 retired, and what replaces it?
What are the AI-900 skill areas and weights?
What was the AI-900 passing score and format?
Does AI-900 have prerequisites?
How much did AI-900 cost and does the certification expire?
Start your free 24-hour AI-900 practice trial
Full access to the question bank, both study modes, source-linked explanations, and score-by-skill-area. No credit card required.
Start free trial →