Hello, you are using an old browser that's unsafe and no longer supported. Please consider updating your browser to a newer version, or downloading a modern browser.
The Microsoft Certified: Azure AI Cloud Developer Associate replaced AZ-204 in 2026. This guide, current as of October 2026, covers the AI-200 exam, the four domains and their weights, the format and passing score, the free annual renewal, how it differs from AI-103, and the developer roles it fits.
The Microsoft Certified: Azure AI Cloud Developer Associate, earned through exam AI-200 (Developing AI Cloud Solutions on Azure), is Microsoft’s associate-level certification for developers who build the back end that AI applications run on. It replaced the Azure Developer Associate (AZ-204), which retired on July 31, 2026.
AI-200 is about the services an AI application runs on. Prompts and model training aren’t on the outline. You’ll be tested on container images and where they run, embeddings and vector search in Cosmos DB and PostgreSQL, caching in Azure Managed Redis, Service Bus and Event Grid messaging, Azure Functions, Key Vault secrets, and tracing with OpenTelemetry. Microsoft expects you to write Python and work with the Azure SDKs.
AI-200 carries no formal prerequisites. It is issued and maintained by Microsoft, and the skills-measured outline was last updated on October 7, 2026, when the data management domain grew to 30 to 35 percent of the exam.
Container Registry, App Service, Container Apps with KEDA scaling, and AKS. 20 to 25%.
Cosmos DB for NoSQL, PostgreSQL with pgvector, and Azure Managed Redis. The heaviest domain at 30 to 35%.
Service Bus, Event Grid, and Azure Functions triggers and bindings. 20 to 25%.
Key Vault, App Configuration, OpenTelemetry tracing, and KQL queries. 20 to 25%.
Four things to know about the exam Microsoft built to take over from AZ-204.
AZ-204 and its renewal assessment retired on July 31, 2026, and Microsoft named AI-200 as the replacement. If you were planning on the Azure Developer Associate, this is the exam to sit now. If you already hold it, retirement doesn’t revoke it, but it can’t be renewed, so AI-200 is how you keep a current Microsoft developer credential.
AZ-204 tested general Azure development. AI-200 keeps the containers, functions and messaging but points them at AI applications. You store embeddings, run vector similarity search, build retrieval-augmented generation with metadata filters, and index vectors in Redis. That data work is the biggest slice of the exam.
Microsoft lets you open Microsoft Learn in a split screen during associate exams like AI-200. It is meant for looking up a detail, and the clock won’t allow it on every question. Our look at the most popular Microsoft certifications in 2026 covers where AI-200 fits among this year’s new exams.
AI-200 is valid for one year and renews free through an unproctored, open-book assessment on Microsoft Learn. The renewal window opens six months before your expiration date. Microsoft doesn’t claim a DoD 8140 mapping for AI-200. The current matrix is published at the DoD Cyber Exchange.
Its weight comes from the job instead. Teams adding search, retrieval and agents to their Azure applications still need someone to run the containers, databases, queues and secrets underneath, and AI-200 is the Microsoft credential written for that person.
The certification, the AI-200 exam, and how to maintain it, as of October 2026.
AI-200 covers the platform under an AI application. From here you can add the AI layer on top, go deeper on the data, or move toward delivery pipelines. These three are the natural next steps.
The other half of the Microsoft AI developer pair. AI-103 covers generative AI and agent solutions in Microsoft Foundry, plus computer vision, text analysis and information extraction. It is also the required credential for the AI-500 Multi-Agent AI Solutions Expert, and AI-200 does not count toward that one.
A closer look at the database AI-200 leans on hardest. DP-420 covers designing and optimizing Cosmos DB applications, including agent memory, conversation state and retrieval indexes. It is a 100-minute associate exam, and the Cosmos DB work in AI-200 is a direct lead-in.
The move toward pipelines and release management. One caution: as of October 2026, Microsoft’s AZ-400 page still lists the Azure Administrator or the retired Azure Developer Associate as the required associate credential, and AI-200 isn’t on that list yet. Check the page before you count on it.
AI-200 sits at the associate tier of Microsoft’s role-based path. The fundamentals exams are optional groundwork, and from AI-200 the path branches toward the AI layer or deeper into the platform.
Optional groundwork
The Azure developer credential
Pick your path
Two questions to answer before you commit: can you sit it, and should you pursue AI-200 specifically. Here’s a straight answer to both.
No prerequisite credential required.
Microsoft lists no prerequisite certification for AI-200. You don’t need AZ-204, AZ-900 or AI-901 first. You can register and take it directly through Pearson VUE at a test center or online with a proctor.
Expect to write and debug real code.
Microsoft expects you to be proficient in Python, the Azure SDKs, messaging and eventing, vector databases and containerized apps. The objectives say implement, deploy, optimize and troubleshoot. Build each one in a real Azure subscription, and use Microsoft’s free practice assessment on AI Skills Navigator to see where you stand.
AI-200 maps to the developer and engineering roles that build and run the services an AI application depends on, rather than the model itself.
Builds and deploys applications on Azure. With AZ-204 retired, AI-200 is the associate credential Microsoft now offers for this job, and its outline follows the AI features those applications are picking up.
Owns the APIs, queues and data layer behind an application. AI-200 covers that layer directly, from Service Bus and Event Grid to Azure Functions, Cosmos DB and PostgreSQL.
Runs the platform an AI application sits on: container hosting, vector stores, caching and tracing. The person who keeps the retrieval pipeline up after the demo is over.
Deploys and operates services on Azure for application teams, including Container Apps, AKS and function apps. AI-200 proves you can do that work from the code side.
Ships container images, function apps and configuration through to production and watches them once they land. Key Vault, App Configuration, OpenTelemetry and KQL are all on the AI-200 outline.
Designs schemas, indexes and queries for the data an AI feature depends on, including embeddings and vector search in Cosmos DB, pgvector in PostgreSQL and Azure Managed Redis.
AI-200 shares the associate tier with AI-103, its AI-layer partner, and AZ-104, the Azure administration exam. Here’s how the three line up.
| Azure AI Cloud Developer (AI-200) | Azure AI Apps and Agents Developer (AI-103) | Azure Administrator (AZ-104) | |
|---|---|---|---|
| Issuer | Microsoft | Microsoft | Microsoft |
| Level | Associate | Associate | Associate |
| Focus | Back-end services under AI apps | AI apps and agents in Microsoft Foundry | Operating Azure environments |
| 2026 Change | Replaced AZ-204 (retired July 31, 2026) | Replaced AI-102 (retired June 30, 2026) | None, still current |
| Prerequisites | None | None | None (rec. 6 months) |
| Exam | 4 domains, 120 min | 5 domains, 120 min | 5 domains, 100 min |
| Passing Score | 700 of 1000 | 700 of 1000 | 700 of 1000 |
| Cost (US) | $165 | $165 | $165 |
| Validity | 1 year, free renewal | 1 year, free renewal | 1 year, free renewal |
| Counts Toward AI-500 | No | Yes (required) | No |
| Best For | Back-end and cloud developers | AI engineers building apps and agents | Cloud administrators |
Prices are US list prices as of October 2026, and Microsoft sets fees by the country where you test. AI-103 figures come from its Microsoft Learn certification page and study guide. AZ-104 figures come from our Azure Administrator guide. All three renew free each year through an online assessment on Microsoft Learn.
Certification Matchups: the published facts, strengths and weak spots of 33 certifications, side by side.
Our Azure AI Cloud Developer boot camp covers all four AI-200 domains in five days of live, instructor-led training, online or in person. It uses Official Microsoft Courseware and 24/7 hosted practice labs, and it includes practice questions, exam prep review sessions, your $165 exam voucher, an exam retake, and 180 days of instructor help after class.
What changed in 2026, how AI-200 fits the Microsoft map, and where AI certifications are heading.
Which Microsoft exams retired this year and what replaced them, including the AZ-204 to AI-200 swap, plus where developers should start now that both AI-200 and AI-103 are on the board.
From the MCSE to role-based exams to the 2026 reshuffle, and why AI-200, AI-103 and SC-500 now carry AI skills inside the core developer and security credentials.
The full Microsoft map by tier and job role, refreshed for the 2026 retirements, showing where AI-200 sits next to AI-103, AZ-104 and the expert credentials.
The security side of the 2026 changes, with a table of this year’s Microsoft retirements and their replacements, AZ-204 and AI-200 included.
The AI credential field across vendors and governance bodies, including Microsoft AI-901 and AI-103, sorted by the job each one actually fits.
The six GitHub exams and what each one tests now, plus how the track relates to AZ-400 after the AZ-204 retirement.
The multi-cloud question for cloud professionals who already hold AWS credentials: how the services map, where Azure fits better, and when adding a Microsoft credential pays off.
The AI-200 exam is organized into four domains, each carrying its own weight. Data management services are the heaviest at 30 to 35 percent, and the other three sit at 20 to 25 percent each. Click any domain for what it covers.
Building, storing and versioning images in Azure Container Registry and running builds with ACR Tasks. Deploying containers to App Service with environment variables and secrets, deploying to Azure Container Apps with revision management and KEDA event-driven scaling, deploying to AKS with manifest files, and troubleshooting AKS and Container Apps through logs, events and end-to-end connectivity.
The heaviest domain. Azure Cosmos DB for NoSQL (SDK queries, indexing policies, consistency levels, Request Unit cost, embeddings with vector similarity search, and the change feed processor), Azure Database for PostgreSQL (schemas, indexing, pgvector tuning, sizing for vector workloads, retrieval-augmented generation with metadata filters, connection optimization) and Azure Managed Redis (caching, expiration, invalidation and vector indexing).
Event- and message-based AI solutions. Queuing back-end work with Azure Service Bus, including topics, subscriptions and dead-letter handling, and building event-driven workflows with Azure Event Grid filters, custom events and retries. Building serverless APIs with Azure Functions triggers and bindings, then configuring and deploying function apps.
Storing, rotating and retrieving secrets with Azure Key Vault and keeping app settings in Azure App Configuration. Tracing distributed systems with OpenTelemetry SDKs and writing KQL queries to analyze logs and metrics when something in the chain fails.
Domains and weights reflect the Microsoft AI-200 skills-measured outline dated October 7, 2026, which raised the weight of the data management domain. Microsoft refreshes role-based exams periodically, so confirm the current outline on Microsoft Learn before scheduling.
The questions candidates ask most often when researching the Microsoft Azure AI Cloud Developer certification.
The Microsoft Certified: Azure AI Cloud Developer Associate, earned by passing exam AI-200, is Microsoft’s associate-level certification for developers who build the back end of AI solutions on Azure. It covers container hosting, vector-enabled data services, messaging and serverless functions, and securing and monitoring those services. It replaced the Azure Developer Associate (AZ-204) in 2026.
AI-200 replaced exam AZ-204 and the Microsoft Certified: Azure Developer Associate, which retired on July 31, 2026. Microsoft retired the AZ-204 renewal assessment as well. Retirement doesn’t revoke an Azure Developer Associate you already earned, but you can no longer earn or renew it, so AI-200 is the developer exam to take now.
No. Microsoft planned the AI-200 beta for May 2026 and the live exam for July 2026, and Microsoft Learn now lists the credential without a beta label. The skills outline was updated on October 7, 2026, and Microsoft’s practice assessment for the exam is available on its AI Skills Navigator site.
No. Microsoft lists no prerequisite certification for AI-200, so anyone can register. Microsoft does expect you to be proficient in Python, the Azure SDKs, Azure data management services, monitoring and troubleshooting, messaging and eventing, vector databases, and running containerized applications on Azure. You don’t need AZ-204 or a fundamentals exam first.
The four domains are Develop containerized solutions on Azure (20 to 25 percent), Develop AI solutions by using Azure data management services (30 to 35 percent), Connect to and consume Azure services (20 to 25 percent), and Secure, monitor, and troubleshoot Azure solutions (20 to 25 percent). The data domain, which covers Cosmos DB, PostgreSQL and Azure Managed Redis, carries the most weight.
Microsoft doesn’t publish a question count for AI-200. Its general guidance is that most Microsoft certification exams have 40 to 60 questions, and the number varies by exam. You get 120 minutes. The exam is proctored and may include interactive components, and Microsoft’s free exam sandbox lets you try the question formats in the real exam interface before test day.
You need 700 or higher on a scale of 1 to 1,000. Microsoft reports a scaled score, so 700 doesn’t mean 70 percent correct. The passing line reflects the difficulty of the questions you get. Some items may be unscored questions Microsoft is still testing, and you can’t tell which ones, so answer every question as if it counts.
The AI-200 exam costs $165 USD in the United States as of October 2026. Microsoft sets the price by the country where you test, so fees vary outside the US. You schedule it through Pearson VUE, at a test center or online with a proctor. Training Camp’s Azure AI Cloud Developer boot camp includes the exam voucher in the program fee, along with an exam retake.
No DoD 8140 mapping is claimed for it. Microsoft doesn’t list AI-200 against a DoD 8140 work role, so don’t count on it to satisfy a cyber-coded position requirement on its own. Its value comes from the work itself, because teams building AI applications on Azure need developers who can run the services underneath them. See the full DoD 8140 work role paths.
AI-200 is an associate certification and is valid for one year. Microsoft lets you renew it free through a short, unproctored, open-book assessment on Microsoft Learn. You can take it any time in the six months before your expiration date, there’s no limit on attempts as long as you pass before it expires, and each pass extends the certification one year from its expiration date.
AI-200 maps to roles such as Azure Developer, Back-End Engineer, AI Platform Engineer, Cloud Engineer, DevOps Engineer, and Database Developer working with vector data. These are the people who build and run the services an AI application depends on, rather than the model itself.
Take the one that matches your work. AI-103 is about the AI layer: generative AI apps and agents in Microsoft Foundry, plus computer vision, text analysis and information extraction. AI-200 covers the Azure platform underneath: containers, vector-enabled databases, messaging, Functions, secrets and tracing. Both are 120-minute associate exams that cost $165 in the US and renew free each year. Only AI-103 counts toward the AI-500 Multi-Agent AI Solutions Expert.
AI-200 is an associate-level exam that rewards hands-on work. It spans a long list of services, from AKS and Container Apps to Cosmos DB, PostgreSQL, Redis, Service Bus, Event Grid, Functions and Key Vault, and the objectives are written as tasks to implement, deploy and troubleshoot. Give the data domain the most study time, since it carries 30 to 35 percent and spans three database services. Microsoft lets you open Microsoft Learn during the exam, but looking up every answer won’t fit in 120 minutes, so build each objective in a real Azure subscription as you study.
For developers who build on Azure, yes. AZ-204 is gone, so AI-200 is now the associate credential Microsoft offers for Azure back-end development, and its outline matches the work teams are doing as they add vector search, retrieval and event-driven processing to their apps. It’s a weaker fit if your job is the models and agents themselves, where AI-103 is closer, or if you administer Azure rather than write code for it, where AZ-104 fits better.
If you’re moving off AZ-204, choosing between AI-200 and AI-103, or planning Azure developer training for a team, tell us where you stand and we’ll help you plan the next step.