Blog Post
Why AI-200 is the best certification for cloud engineers moving into AI. Salary data, domain breakdown, and why AI-200 plus AI-103 forms a formidable CV.
Checking sign-in status...
If you are a cloud engineer, you are already closer to AI engineering than you think. The containers you deploy, the identity you configure, the event-driven pipelines you build, and the observability you set up are the same foundations every production AI system runs on. AI-200 validates that foundation and extends it into AI workloads.
AI-200 (Developing AI Cloud Solutions on Azure) replaced AZ-204 on July 31, 2026. AZ-204 validated general Azure development. AI-200 validates Azure development in an AI context: containerized AI workloads, vector-enabled data stores, event-driven pipelines, and distributed observability. For cloud engineers, this is a natural extension, not a career restart.
Related: AI-200 Study Plan 2026 ↗
Most AI training programs start with model building. They teach you to fine-tune transformers, calculate attention weights, and experiment with architectures in a notebook. That is data science, not AI engineering.
Production AI engineering is different. It requires someone who can:
If that list sounds familiar, it is because those are the same skills cloud engineers use every day. The only thing missing is the AI-specific context: how vector search works, what RAG requires from the data layer, and why agent orchestration changes the observability picture.
AI-200 was built for exactly this profile.
The transition time is estimated at five to nine months for cloud engineers adding AI/GPU specialization. That is not a multi-year retraining program. It is a structured skill extension on top of a foundation you already have.
AI-200 has four domains, and every one maps to work cloud engineers already do.
| Domain | Weight | Cloud Engineering Connection |
|---|---|---|
| Develop containerized solutions on Azure | 20-25% | ACR, Container Apps, AKS manifest deployments, KEDA event-driven scaling |
| Develop AI solutions with Azure data management services | 25-30% | Cosmos DB vector search, PostgreSQL with pgvector, Azure Managed Redis |
| Connect to and consume Azure services | 20-25% | Service Bus, Event Grid, Azure Functions for pipeline triggers |
| Secure, monitor, and troubleshoot Azure solutions | 20-25% | Key Vault, managed identity, OpenTelemetry, KQL diagnostics |
Domain 2 is the largest at 25-30%. It covers vector-enabled data stores: Cosmos DB for NoSQL with vector similarity search, Azure Database for PostgreSQL with the pgvector extension for RAG patterns, and Azure Managed Redis for caching and vector indexing. These are the data services every production RAG system depends on. If you have worked with Cosmos DB or PostgreSQL, you are extending known territory, not starting from zero.
Domain 1 covers container deployment with Azure Container Registry, Container Apps with KEDA event-driven scaling, and AKS manifest deployments. If you have deployed containers to AKS or Container Apps, this domain is a confirmation of existing skill with AI workload context added.
Domain 3 covers Service Bus, Event Grid, and Azure Functions. Event-driven architecture is how AI systems coordinate tools, process documents, and respond to external triggers. This is the integration layer you already understand.
Domain 4 covers Key Vault, managed identity, OpenTelemetry for distributed tracing, and KQL for diagnostics. This is the operational layer that keeps AI systems running and observable in production. Cloud engineers do this daily.
The exam assumes Python proficiency and Azure SDK familiarity. It is a developer's exam, not an architect's. But it is a developer exam for people who build back-end systems, not front-end features.
AI-200 and AI-103 are complementary, not overlapping. They validate different layers of the same stack.
| Layer | AI-200 | AI-103 |
|---|---|---|
| Focus | Where AI runs | What AI does |
| Core skills | Containers, vector data stores, event-driven integration, observability | Foundry agents, RAG pipelines, prompt engineering, multi-agent orchestration |
| Heaviest domain | Data management (25-30%) | Generative AI and agents (30-35%) |
| Best for | Cloud engineers, platform engineers, AKS admins | AI/ML developers, Copilot Studio developers, Azure OpenAI users |
| Code-heavy? | Less code; more YAML, ARM/Bicep, gateway config | Yes: Python/C# Foundry SDK, OpenAI SDK |
AI-200 teaches you to run AI at enterprise scale. AI-103 teaches you to build the AI feature. Most developers should take AI-103 first because it maps to application-layer work. Most cloud engineers should take AI-200 first because it maps to the infrastructure they already operate. Senior AI engineers eventually earn both.
The combination signals that you can design, deploy, secure, and monitor an AI system end to end. You understand the Foundry application layer and the Azure infrastructure layer. That is rare. Most candidates hold one or the other. Holding both puts you in a different hiring bracket.
AI-200 is for developers and platform engineers who deploy containerized applications on Azure as part of their job.
It is not for beginners. The exam assumes Python proficiency, Azure SDK experience, and container familiarity. If you have never deployed a container or queried a vector database, build that foundation first.
| Detail | Specification |
|---|---|
| Exam code | AI-200 |
| Level | Associate |
| Questions | 40-60 |
| Duration | 100 minutes |
| Passing score | 700/1000 |
| Cost | $165 USD |
| Renewal | Free annual online assessment |
| Delivery | Pearson VUE |
| Prerequisites | None formal; Python, Azure SDK, and container experience recommended |
The exam tests applied judgment, not definitions. Hands-on experience with the services it covers is the fastest route to readiness.
Build the infrastructure. Deploy a containerized AI workload to Azure Container Apps. Store embeddings in Cosmos DB and run a vector similarity search. Wire an Azure Function to a Service Bus queue. Configure OpenTelemetry tracing. The exam scenarios will feel familiar if you have done the work.
Practice scenario-based questions. AI-200 questions present a situation with constraints and ask for the best solution. ExamOS delivers scenario practice through the Daily Sprint and tracks your Readiness Score across every domain. Your weakest domain becomes tomorrow's focus.
Read the official study guide. The AI-200 skills outline lists every tested sub-skill. Cross-reference it against your cloud engineering experience and fill the gaps.
Then layer on AI-103. Once AI-200 is complete, AI-103 adds the application-layer skills: Foundry agents, RAG pipelines, prompt engineering, and multi-agent orchestration. The pairing creates a CV that signals full-stack AI capability.
AI-200 (Developing AI Cloud Solutions on Azure) is Microsoft's Associate-level certification for the Azure AI Cloud Developer role. It validates the ability to design, build, and implement AI solutions on Azure with a focus on back-end services, containerized compute, vector-enabled databases, event-driven pipelines, and distributed observability.
AI-103 is the application-focused credential for building generative AI apps and agents using Microsoft Foundry. AI-200 is the platform-focused credential for architecting and operating the cloud infrastructure AI workloads run on. AI-103 teaches you to build the AI feature. AI-200 teaches you to run it at scale. They are complementary, not alternatives.
Cloud engineers already work with containers, identity, event-driven integration, and observability. AI-200 extends those skills into AI workloads: vector data stores, AI inference scaling, and AI-specific monitoring. The Lemon.io study estimates a five-to-nine-month transition from traditional cloud engineering to AI/GPU cloud engineering, with a salary increase approaching 79%.
AI-200 replaced AZ-204, which retired on July 31, 2026. AZ-204 validated general Azure development. AI-200 validates Azure development in an AI context. The change reflects Microsoft's position that AI competency is now a baseline expectation for Azure developers, not a separate specialization.
The passing score is 700 on a scale of 1 to 1,000. The exam has 40-60 questions with a 100-minute time limit. The exam costs $165 USD and renews free through an annual online assessment on Microsoft Learn.
Build the cloud engineering skills that AI workloads depend on. Start with a free Daily Sprint on ExamOS and track your Readiness Score as you prepare for AI-200. Then layer on AI-103 for a full-stack AI engineering profile.