Microsoft Retires AZ-204 and Replaces It With AI-200: A Seven-Year Era Ends
On 31 July 2026, Microsoft retired the AZ-204 Developing Solutions for Microsoft Azure certification — the primary credential for cloud application developers on the Azure platform since 2019. In its place, Microsoft directs developers to the AI-200 Azure AI Cloud Developer Associate certification, which became available on 9 July 2026, three weeks before AZ-204's retirement date. The replacement is not a minor revision of the same exam with updated service names. AZ-204 tested competence across the full Azure developer stack: compute services, blob storage, messaging queues, API management, authentication, and monitoring. AI-200 tests a structurally different skill set: building and integrating AI pipelines, working with large language models through Azure OpenAI Service, constructing vector-enabled databases for retrieval-augmented generation, deploying AI workloads in containerised environments, and instrumenting AI-powered applications for production observability. The switch signals that Microsoft considers AI-native development — building applications that incorporate language model calls, embedding pipelines, and multimodal inputs as first-class architectural components — to be the baseline skill set for cloud developers going forward.
What AZ-204 Tested and Why It Was Retired
AZ-204 was introduced in 2019 as the replacement for the earlier Microsoft Certified Azure Developer Associate qualification. Its scope reflected the Azure developer platform as it existed before large-scale AI integration became central to application architecture: App Service, Azure Functions, Azure Container Instances, Azure Blob Storage, Azure Service Bus, Azure API Management, Azure Cosmos DB, Azure Active Directory, and the Azure Monitor and Application Insights observability stack. The exam's underlying conceptual model — compute, store, message, authenticate, deploy — was coherent when those were the primary decisions a cloud developer made. That model is increasingly secondary to the decisions that occupy most cloud developer time in 2026: which language model to call, how to structure a retrieval pipeline, how to manage prompt context across multi-step agentic workflows, and how to attribute inference cost to specific application features.
What AI-200 Covers That AZ-204 Did Not
AI-200 is organised around four competency areas. The first is event-driven AI pipeline construction: designing and implementing workflows that trigger AI inference calls through Azure Event Grid and Azure Functions, chaining model outputs into downstream systems, and handling asynchronous inference at scale. The second is vector infrastructure: provisioning and querying vector-enabled databases for semantic search and retrieval-augmented generation using Azure AI Search, with understanding of embedding generation, chunking strategies, and hybrid retrieval methods. The third is large language model integration: calling Azure OpenAI Service models, designing prompt templates, managing token budgets across conversation turns, implementing streaming output, and handling tool-use and function-call responses within application code. The fourth is AI observability: instrumenting AI applications with distributed tracing, model output logging, latency monitoring, and cost attribution using Azure Monitor, Application Insights, and OpenTelemetry instrumentation frameworks.
The Exam Format and Passing Criteria
AI-200 contains 40 to 60 questions, runs for 100 to 120 minutes, and requires a passing score of 700 out of 1000. The exam is scenario-driven: questions describe a business requirement and ask which Azure service configuration, code pattern, or architectural decision satisfies it. The primary programming language assumed is Python, with questions covering Azure SDK usage for AI services, OpenAI client library calls, and containerisation with Azure Container Apps. A beta version of the exam ran from April 2026, and the production version became generally available on 9 July 2026.
What the Retirement Means for Existing AZ-204 Holders
AZ-204 holders whose credentials have not expired retain their Microsoft Certified status through the duration of their certification cycle. Microsoft has not announced an accelerated expiry for existing holders. However, learning paths, official Microsoft training materials, partner ecosystem training programmes, and the Microsoft Learn catalogue are being updated to align with the AI-200 scope, which means that renewal study materials tied to AZ-204 content are being deprecated alongside the exam itself. Developers who were mid-preparation for AZ-204 when the retirement took effect need to retrain on the AI-200 objectives, which require different hands-on skills — particularly familiarity with vector databases and LLM API integration that were not tested in AZ-204.
What This Means for Indian Developers and Software Teams
India has one of the world's largest populations of Microsoft Azure-certified professionals, and certification credentials are a material factor in hiring decisions across India's IT services, consulting, and product development sectors. For Indian developers planning Azure certification paths after 31 July 2026, there is no longer an AZ-204 path — the AI-200 route with its focus on Python, LLM integration, and vector infrastructure is the new baseline. For Indian software agencies and IT services companies that maintain Azure certification counts as part of their Microsoft partner status or include certification requirements in client proposals, the AI-200 transition affects workforce development planning and L&D investment: upskilling to AI-200 requires different hands-on tooling investment than AZ-204 preparation did. The exam's focus on event-driven AI pipelines, retrieval-augmented generation, and Azure OpenAI integration also maps directly to the kinds of AI feature requests that Indian software teams are receiving from enterprise clients — making AI-200 preparation valuable beyond its credential function.
The Bottom Line
Microsoft retired the AZ-204 Developing Solutions for Microsoft Azure certification on 31 July 2026 after seven years as the primary cloud developer credential on the Azure platform. Its replacement, the AI-200 Azure AI Cloud Developer Associate certification, became available on 9 July 2026 and tests competence in event-driven AI pipeline construction, vector-enabled database integration, large language model deployment through Azure OpenAI Service, and AI observability instrumentation. The exam contains 40 to 60 scenario-driven questions, requires a passing score of 700 out of 1000, and assumes proficiency in Python and Azure SDKs. For India's large community of Azure-certified developers and the IT services companies that employ them, the retirement marks a structural shift: cloud developer skill on the Azure platform now means AI developer skill, and workforce development programmes need to reflect that change.
Frequently Asked Questions
What happened to the AZ-204 certification and why was it retired?+
Microsoft retired the AZ-204 Developing Solutions for Microsoft Azure certification on 31 July 2026 after seven years as the platform's primary cloud developer credential. The retirement reflects a structural shift in what Microsoft considers core cloud developer competence: AZ-204 tested traditional compute, storage, messaging, and authentication skills on Azure, while the demands of cloud application development in 2026 centre on integrating large language models, building retrieval-augmented generation pipelines, managing vector databases, and instrumenting AI-powered applications for production observability. Rather than add AI content as an optional module to an existing exam, Microsoft created a new certification — AI-200 Azure AI Cloud Developer Associate — with a fundamentally different scope and skill set as AZ-204's direct replacement.
What does the AI-200 Azure AI Cloud Developer Associate certification test?+
AI-200 tests four main competency areas. The first is event-driven AI pipeline construction: designing workflows that trigger AI inference through Azure Event Grid and Azure Functions and routing outputs to downstream systems. The second is vector infrastructure: provisioning vector-enabled databases, generating embeddings, and building retrieval-augmented generation pipelines using Azure AI Search. The third is large language model integration: calling Azure OpenAI Service models, designing prompt templates, managing token budgets, implementing streaming output, and handling tool-use responses in application code using Python and Azure SDKs. The fourth is AI observability: instrumenting AI applications with distributed tracing, model output logging, latency monitoring, and cost attribution through Azure Monitor, Application Insights, and OpenTelemetry. The exam does not test traditional AZ-204 topics such as App Service configuration, Azure Blob Storage management, or Azure API Management gateway policies.
What is the exam format and passing score for AI-200?+
AI-200 contains 40 to 60 questions, runs for 100 to 120 minutes, and requires a passing score of 700 out of 1000. The exam is scenario-driven: most questions describe a business requirement and ask which Azure service configuration, code pattern, or architectural decision satisfies it. Question formats include multiple choice, multiple response, drag-and-drop, and case study questions. The primary programming language assumed is Python. A beta version of the exam was available from April 2026, and the production version became generally available on 9 July 2026, three weeks before AZ-204's 31 July retirement date.
How does the AZ-204 retirement affect Indian Azure-certified developers?+
Indian developers who currently hold a valid AZ-204 credential retain their Microsoft Certified status through the end of their certification cycle — Microsoft has not announced accelerated expiry for existing holders. However, any developer pursuing new Azure developer certification after 31 July 2026 must take AI-200, which requires a different preparation path: the exam tests Python proficiency, Azure OpenAI Service integration, vector database querying, and AI pipeline construction skills that were not covered by AZ-204. For Indian IT services companies and software agencies that maintain Azure certification counts as part of Microsoft partner status or client contract requirements, the AI-200 transition affects L&D investment and upskilling timelines. Microsoft Learn, official training materials, and partner training programmes are being updated to the AI-200 scope, meaning AZ-204-aligned study materials are being phased out alongside the exam itself.
Written by
TechPillow Team
Sharing insights on technology, product development, and the Indian tech ecosystem.