Quick Answer
Start with AI-900 or AI-901 (Azure AI Fundamentals) — no prerequisites. Then choose an Associate cert by role: AI-103 for building AI apps and agents (this replaced AI-102, which retired June 30, 2026), AI-200 for AI cloud development, or AI-300 for machine learning operations. Experienced practitioners can go further with AI-500 (Designing and Implementing Multi-Agent AI Solutions), Microsoft's expert-level multi-agent AI certification.
The Microsoft AI certification lineup went through a major restructure in 2026 — AI-102 is gone, three new Associate-level exams took its place, and search results are full of stale guidance pointing at exams you can no longer take. Here's the accurate, current picture.
AI-102 Is Retired — Here's What Replaced It
Microsoft Certified: Azure AI Engineer Associate (AI-102) retired on June 30, 2026. It can no longer be scheduled or earned. Microsoft's replacement is AI-103 (Azure AI Apps and Agents Developer Associate), which covers building generative AI apps and agents using Azure AI Foundry — the modern equivalent of what AI-102 tested, updated for agent-based and generative AI development patterns.
Step 1: AI-900 vs AI-901
Both AI-900 and AI-901are titled "Azure AI Fundamentals" and sit at the same entry level — no prerequisites, no coding required, 2-4 weeks of part-time study. They cover machine learning basics, computer vision, natural language processing, generative AI, and responsible AI principles.
Test Your AI-900 Readiness Free
40 free practice questions covering ML, computer vision, NLP, and generative AI basics.
Start Free PracticeStep 2: Choose Your Associate Cert
AI-103 — Azure AI Apps and Agents Developer Associate
Building generative AI apps and agents with Azure AI Foundry, computer vision, and language services. The direct AI-102 replacement.
Best for: Software developers building AI-infused applications
AI-200 — Azure AI Cloud Developer Associate
Deploying and managing AI cloud infrastructure and services at scale on Azure.
Best for: Cloud engineers deploying AI infrastructure
AI-300 — Machine Learning Operations Engineer Associate
Operating ML pipelines: training, deployment, monitoring, and MLOps practices.
Best for: ML engineers and MLOps specialists
Step 3: AI-500, the Expert-Level Cert
AI-500 (Designing and Implementing Multi-Agent AI Solutions) is the newest addition to the AI track and the first Expert-level AI certification Microsoft has released. It targets practitioners who already build production AI systems and are moving into orchestrating multiple agents working together — architecture, development, evaluation/observability, and security/governance of multi-agent solutions built on Microsoft Foundry.
What it assumes you already know:
- • Hands-on experience deploying agentic systems in production
- • Proficiency in Python
- • Familiarity with Microsoft Agent Framework, Model Context Protocol (MCP), retrieval-augmented generation (RAG), and LangGraph
- • Experience with Azure compute, network, storage, and data services
What About Data Science (DP-100)?
Full Microsoft AI Certifications List
| Code | Name | Level |
|---|---|---|
| AI-900 | Azure AI Fundamentals | Fundamentals |
| AI-901 | Azure AI Fundamentals | Fundamentals |
| AI-103 | Azure AI Apps and Agents Developer Associate | Associate |
| AI-200 | Azure AI Cloud Developer Associate | Associate |
| AI-300 | Machine Learning Operations Engineer Associate | Associate |
| AI-500 | Designing and Implementing Multi-Agent AI Solutions | Expert |
| AI-102 | Azure AI Engineer Associate | Retired Jun 30, 2026 |
| DP-100 | Azure Data Scientist Associate | Retired Jun 1, 2026 |
Free Training vs. the Paid Exam
Every AI-series exam has free, self-paced Microsoft Learn training. That training is free; the certification itself requires passing a paid, proctored exam ($99 for AI-900/AI-901, $165 for Associate-level AI-103/AI-200/AI-300). Use the free training plus MSCertQuiz's free 40-question AI-900 practice set to gauge readiness before paying for the real exam.
Salary Ranges by Role
General US market ranges — verify against local job postings.
Common Questions
Which Microsoft AI certification is best?
Start with AI-900 or AI-901, then pick AI-103 (app/agent development), AI-200 (cloud infrastructure), or AI-300 (MLOps) based on your role. Experienced practitioners can progress to AI-500, the expert-level multi-agent AI certification.
What is the Microsoft AI certification path?
Fundamentals (AI-900/AI-901) leads into one Associate exam matching your role — AI-103, AI-200, or AI-300. From there, AI-500 (Multi-Agent AI Solutions) is the expert-level capstone.
What is the full Microsoft AI certifications list?
AI-900, AI-901, AI-103, AI-200, AI-300, and AI-500 (Expert). AI-102 retired June 30, 2026 and AI-103 replaced it.
What is AI-500?
AI-500 (Designing and Implementing Multi-Agent AI Solutions) is Microsoft's expert-level AI certification for practitioners who design, build, and optimize production multi-agent AI systems using Microsoft Foundry, the Agent Framework, and MCP.
Is AI-102 still a valid certification to study for?
No, it retired June 30, 2026 and can't be earned anymore. Study AI-103 instead.
What about Microsoft AI and data science certifications?
DP-100 (Azure Data Scientist Associate) retired June 1, 2026 with no direct 1:1 replacement. For applied ML work today, AI-300 (Machine Learning Operations Engineer Associate) is the closest current Microsoft credential covering ML pipeline operations.
Is there a free Microsoft AI certification course?
Microsoft Learn's training paths are free for every AI exam; the proctored certification exam itself has a fee.
Start Your AI Certification Path
Practice AI-900 free, then choose your Associate specialization.
About MSCertQuiz
MSCertQuiz provides affordable, high-quality practice resources for Microsoft certification candidates, built by certified professionals who track every exam change and retirement.