What AI-200 is
AI-200 is titled "Developing AI Cloud Solutions on Azure" and leads to the Microsoft Certified: Azure AI Cloud Developer Associate credential. Microsoft's certification page describes it as the new Azure developer exam that replaces AZ-204 on 31 July 2026, according to one third-party tracker. Treat that replacement date as secondary reporting. Another third-party guide notes that Microsoft's own study guide still pointed some of its "get trained" links at AZ-204.
The candidate profile is a developer who contributes across requirements, design, development, deployment, security and monitoring, with an emphasis on back-end services. Microsoft expects proficiency in these areas:
- Azure SDKs and third-party SDKs used on Azure
- Data management services
- Monitoring and troubleshooting
- Messaging and eventing
- Vector databases
- Python
- Containerized applications on Azure
The weights, stated correctly
The Nerdbot text runs its percentages together, so it's easy to misread. Microsoft's study guide, last updated 2026-10-30, lists these ranges:
| Skill area | Weight |
|---|---|
| Develop containerized solutions on Azure | 20–25% |
| Develop AI solutions by using Azure data management services | 25–30% |
| Connect to and consume Azure services | 20–25% |
| Secure, monitor, and troubleshoot Azure solutions | 20–25% |
Data is the only heavyweight domain. Some third-party sites list data at 30–35%, and one of them even says to confirm the latest weights in the official guide. Microsoft's own guide shows 25–30%, so use that figure.
Microsoft also says the bullets under each skill only illustrate how the skill is assessed, and related topics may appear. Most questions cover generally available features, but commonly used preview features can show up.
What each domain covers
Containers (20–25%)
- Build, store, version and manage images in Azure Container Registry, and use Registry Tasks to build and run them.
- Deploy containers to App Service, including environment variables and secrets.
- Deploy to Azure Container Apps, covering environment configuration and revisions, with KEDA for event-driven scaling.
- Deploy and manage AKS apps using manifest files.
- Troubleshoot AKS and Container Apps by inspecting logs, events and end-to-end connectivity.
Data for AI (25–30%)
- Cosmos DB for NoSQL:
- SDK connectivity and queries.
- Tuning Request Unit consumption through indexing policies and consistency levels.
- Storing embeddings and running vector similarity search.
- Building a change feed processor.
- Azure Database for PostgreSQL:
- Schema design, indexing, and reducing pgvector compute overhead.
- Sizing compute, memory and storage for vector workloads.
- RAG patterns using metadata filters.
- Connection optimization.
- Azure Managed Redis: caching, expiration, invalidation and vector indexing.
Connecting services (20–25%)
- Service Bus: queues, topics, subscriptions and dead-letter handling.
- Event Grid: filters, custom events and retries.
- Azure Functions: triggers, bindings, and configuring and deploying function apps.
Security, monitoring and troubleshooting (20–25%)
- Key Vault: secret rotation and retrieval.
- Azure App Configuration.
- Distributed tracing with OpenTelemetry SDKs.
- KQL queries for logs and metrics.
Nerdbot's advice to try at least one service per domain is sound. The objectives, though, are fairly deep. Cosmos DB indexing policies and pgvector overhead are not topics you can pick up from a glossary.
Service Bus versus Event Grid
Nerdbot says exam questions will make you choose between announcing that something happened and queuing a message for processing. That is a reasonable study exercise, and it matches how the objectives split the two services. Service Bus is framed around queuing and processing back-end operations, including dead-lettering. Event Grid is framed around event-driven workflows with filters, custom events and retries.
This is a suggested way to revise, not an official sample question. Neither Microsoft nor Nerdbot documents the actual exam items. A third-party guide, Cloud Soft Solutions, shows the same style of scenario, where a serverless HTTP API that writes to a queue maps to Azure Functions with a queue or Service Bus output binding. That is vendor-written practice content, not Microsoft's.
Exam logistics
- Duration and format: 120 minutes, proctored. Interactive components may appear.
- Passing score: 700 or higher.
- Retakes: possible 24 hours after a first failed attempt. Later waits vary, so check Microsoft's retake policy.
- Languages: English, Arabic (Saudi Arabia), Chinese (Simplified and Traditional), French, German, Indonesian, Italian, Japanese, Korean, Portuguese (Brazil), Russian and Spanish. Localized exams are updated about eight weeks after the English version. If your language isn't available, you can request an extra 30 minutes.
- Account choice: Microsoft strongly recommends registering with a personal Microsoft account. Exam records tied to a work or school account are lost, and unrecoverable, if you leave that organization.
- Renewal: Microsoft's study guide says associate certifications expire annually and can be renewed through a free online assessment on Microsoft Learn.
Official preparation tools
Microsoft points candidates to its study guide, learning paths and modules, and exam-prep videos. It also offers an exam sandbox that shows the interface and question types. It says a new Practice Assessment on AI Skills Navigator aligns more closely with the real exam experience. These tools cost nothing, so use them before paying for third-party practice sets.
A practical study plan
- Read the study guide and mark each bullet as "built it", "read about it" or "never touched it".
- Build one small thing per domain. A workable set is a container in Container Registry deployed to Container Apps, a Cosmos DB vector query, a Service Bus queue with a dead-letter check, and a Key Vault secret read by a Function.
- Trace one request with OpenTelemetry and query the logs in KQL. Nerdbot warns against leaving monitoring and Key Vault for the last weekend, and the 20–25% weighting on that domain backs the warning.
- Give extra time to the data domain, especially vector search across Cosmos DB, PostgreSQL with pgvector and Managed Redis.
- Take practice questions, then review why each wrong answer was wrong. Revisit the weak areas a few days later, as Nerdbot suggests.
Be careful with what you buy. Nerdbot recommends PrepBolt web, desktop and PDF practice products. I found no independent evidence on their question quality, currency or accuracy, and Microsoft doesn't endorse them. The same applies to the many AI-200 prep sites now circulating. Several publish their own weights, and at least one differs from Microsoft's. Check any third-party material against the official guide before you rely on it.
Bottom line
AI-200 tests whether you can wire together containers, vector-capable data stores, messaging, secrets and observability on Azure. The official guide is current as of late October 2026 and is the one source to trust on weights and objectives. Nerdbot's general advice is sensible: build things, practice choosing services from requirements, and schedule revision. Its numbers and product pitch need checking against Microsoft's pages.
References
- Microsoft AI-200 Exam: A Guide to Azure AI Cloud Developer Associate Exam Preparation - Nerdbot Nerdbot · 2026-11-19T18:38:43+00:00
- Study guide for Exam AI-200: Developing AI Cloud Solutions on Azure | Microsoft Learn learn.microsoft.com
- Microsoft Certified: Azure AI Cloud Developer Associate - Certifications | Microsoft Learn learn.microsoft.com