Update (September 2026): AI-102 retired on June 30, 2026, along with AI-900. For what to do now, jump to Where this leaves you now. The decision guide below is kept for reference.
If you've been preparing for AI-102, you've probably already seen the warning on Microsoft Learn:
"This exam is scheduled to retire on June 30, 2026."
And now you're stuck. Do you sprint to the finish line and pass AI-102 before it disappears? Or cut your losses, abandon everything you've studied, and start over with AI-103?
This is the question thousands of candidates are asking right now, and most of the advice online gets it wrong by treating it as a simple answer. It isn't. The right decision depends entirely on where you are in your prep, your timeline, and what you actually want to do with the certification.
This post gives you the honest framework to make the right call for your situation.
What Happened on June 30, 2026
Let's be clear on the facts, because there's a lot of confusion.
AI-102 (Azure AI Engineer Associate) retired on June 30, 2026. Since then you cannot sit the exam, cannot retake it, and cannot renew the certification. It is gone permanently.
AI-103 (Azure AI App and Agent Developer Associate) is the replacement. It entered beta in April 2026 and is now the live exam.
AI-900 also retired on June 30, 2026, replaced by AI-901. If you were on that path, the same advice applies: move to the replacement and keep what carries over.
A few things that are NOT happening that you might have heard:
- Your existing AI-102 certification is not being deleted. If you already passed, it stays on your transcript permanently and remains verifiable. It just can't be renewed after retirement.
- There is no automatic migration from AI-102 to AI-103. No voucher transfers, no free upgrades. You must study for and pass AI-103 separately.
- Exam vouchers are generally linked to a specific exam. If you have a voucher for AI-102, check whether it can be transferred. Most cannot be reassigned to AI-103 automatically.
The next two sections, and the scenario table further down, are the advice we gave before June 30, 2026, when AI-102 could still be booked. They are kept for reference. For what to do now, see Where this leaves you now.
The Two Mistakes Most Candidates Are Making Right Now
Mistake 1: Blindly rushing AI-102.
Some candidates are panic-booking AI-102 the moment they hear about the retirement, even if they're only halfway through their prep. Rushing an associate-level Microsoft exam when you're not ready is a recipe for failing, and after June 30, there's no retake. You'll have burned prep time, exam fees, and still need to start AI-103 from scratch anyway.
Mistake 2: Abandoning AI-102 too early.
Other candidates are dropping weeks or months of solid AI-102 preparation the moment they hear "AI-103 is coming." That's equally wasteful. AI-102 content overlaps heavily with AI-103, including Azure OpenAI, AI Search, Document Intelligence, Content Safety, and responsible AI, and the knowledge you've built transfers directly.
The right move is neither of these. It depends on one thing: how close are you to being ready?
Try a real Azure AI lab, free
A pre-provisioned Azure environment in your browser. No Azure account, no credit card, no setup.
Most large enterprises already run on Microsoft infrastructure. Azure AI Foundry is the platform those teams use to build and deploy AI apps, so the skills you practice here map directly to real job demand.
The Decision Framework: Which Path Is Right for You?
Finish AI-102 if:
You are 70 to 80% ready or more.
If you could sit the exam today and pass, or you're close to that level, finish AI-102. You have enough runway before June 30, the knowledge is transferable, and earning a live Microsoft certification (even one that later retires) has real career value right now. Many job postings still list AI-102 as a preferred credential, and that won't change instantly.
You've already registered for the exam.
If you have an exam date booked, show up prepared and pass it. Don't cancel to switch to AI-103 unless you are genuinely underprepared and the exam is more than 4 to 6 weeks away.
You work in a role that maps to AI-102 content today.
AI-102 covers Azure AI Search, Document Intelligence, Language, Speech, Vision, Content Safety, and Azure OpenAI Service integration. If your actual day job involves any of these services right now, AI-102 validates work you're already doing. That's worth something.
Switch to AI-103 if:
You are less than 50% prepared for AI-102.
If you've only started studying or are early in the process, the time cost of reorienting to AI-103 is low and the payoff is higher. AI-103 is where Microsoft is investing. It's the certification path that will be on job postings in 2027 and beyond.
You're starting from zero and today's date is May 2026 or later.
At that point, AI-103 will be live or nearly live, and there's no realistic timeline to prepare for and pass AI-102 before June 30. Don't start a path you can't finish.
Your career goal is building AI agents and generative AI applications.
AI-103 is fundamentally different in focus from AI-102. It's built around Microsoft Foundry, agentic workflows, multi-agent orchestration, RAG pipeline design, and production AI application development. If you are new to the protocol behind these agents, see What is MCP? The 2026 Developer Guide. If that's where you want to work, AI-103 is the more aligned credential, not just now, but for the next several years.
You want the longest certification lifespan.
AI-102 earned today will remain on your transcript but cannot be renewed after retirement. AI-103 is a fresh certification with a full lifecycle ahead of it.
What Does AI-103 Actually Cover That AI-102 Doesn't?
This is worth understanding before you make the switch, because the two exams are not identical.
AI-102 tests you on: Azure AI service configuration and integration, including setting up and consuming Azure AI Search, Document Intelligence, Language, Speech, Vision, Content Safety, and Azure OpenAI Service. It's fundamentally a service integration exam.
AI-103 goes further into: Microsoft Foundry as a unified platform, AI agent development using the Foundry Agent Service, multi-agent orchestration with the Microsoft Agent Framework (Semantic Kernel), multimodal AI (images, video, speech generation), RAG pipeline design, evaluation pipelines, and responsible AI at the application level.
The knowledge overlap is substantial. If you know AI-102 content well, you're not starting from zero with AI-103. You're adding a layer of application architecture and agentic design on top of existing service knowledge.
Think of it this way: AI-102 taught you the ingredients. AI-103 teaches you how to cook.
The Specific Scenarios: What Should You Do?
| Your Situation | Recommended Action |
|---|---|
| Passed AI-102 already | Nothing required. Your cert stays valid. Consider AI-103 when ready. |
| Exam booked before June 30, 70%+ ready | Prepare and pass AI-102. Knowledge transfers to AI-103 later. |
| Exam booked, less than 60% ready | Cancel and switch to AI-103 prep now. Check your voucher terms first. |
| Studying AI-102, no exam booked, 70%+ ready | Book the exam now. Complete it before the deadline. |
| Studying AI-102, no exam booked, less than 50% ready | Switch to AI-103. Don't race a deadline you can't win. |
| Not started yet | Go straight to AI-103. AI-102 is a closing chapter. |
| Have AI-102 voucher, not yet booked | Check voucher expiry and transfer policy. If tied to AI-102 only, book now or accept the loss. |
What Happens to AI-102 on Your Resume After June 30?
This question comes up a lot and the answer is straightforward.
Your AI-102 certification does not disappear. It remains permanently on your Microsoft transcript and is verifiable forever. Employers can still see it, you can still list it, and Microsoft has confirmed that earned certifications are not revoked or invalidated by retirement.
What changed on June 30 is that you can no longer sit the exam or renew the certification. Renewal keeps it "active" on your transcript, but even an expired or retired certification still shows you earned it.
In practical terms: an AI-102 on your CV in 2026 still signals solid Azure AI service knowledge. It won't carry the same weight in 2028 as a freshly earned AI-103 would, but it doesn't become worthless overnight.
Decision made? Grab the AI-103 study guide
All 4 AI-103 learning paths, key Foundry topics in one page.
The Honest Bottom Line
If you're close to ready for AI-102: finish it. The knowledge transfers, the credential has value right now, and you can always add AI-103 later.
If you're early in your journey: go straight to AI-103. It's where Microsoft is investing, it's the credential hiring managers will be looking for in 2027, and it's built around the skills that actually get used in modern AI development roles.
If you were still deciding when June 30 arrived, the choice is now simple: AI-103, keeping everything that carries over.
What to Do Next
If you're switching to AI-103 or starting fresh, here's where to begin:
- Understand the platform. Read our Microsoft Azure AI Foundry Explained guide. Foundry is the backbone of the entire exam.
- Know the curriculum. The AI-103 Study Guide & Cheat Sheet covers all 4 learning paths with what's actually tested.
- Test your knowledge. Work through our 10 AI-103 Practice Questions to see where you stand.
- Get hands-on. Reading is not enough for AI-103. Try our free lab: Deploy Your First Azure AI Model with Microsoft Foundry. 30 minutes, real Azure environment, no setup required.
If you're pushing through AI-102 before the deadline, the Ultimate Exam Guide to Microsoft Azure AI-102 is the best place to consolidate what you know.
Related Reading
- Microsoft AI Certifications Are Changing: AI-900 & AI-102 Retire June 2026
- AI-102 vs AI-103 vs AI-900: Which Microsoft AI Cert Should You Take First in 2026?
- Microsoft Azure AI Foundry Explained: What It Is and Why It's the Core of AI-103
- 10 AI-103 Practice Questions: Scenario-Based Exam Prep for 2026
- AZ-900 vs AI-901: Do You Really Need Both in 2026?
Last reviewed September 2026 by the GetHandsOn.ai team.
Where this leaves you now
AI-102 retired on June 30, 2026. The decision this article was written to help with is closed, so here is what applies today depending on where you were when the deadline passed.
You passed AI-102 before it retired. Your certification stays valid until its own expiry date and it still counts on a CV. What it no longer proves is Foundry competence, because most of the Foundry surface area is not in the AI-102 blueprint. If you want that on paper, AI-103 is the exam, and roughly half of your existing preparation still applies. AI-102 to AI-103: what carries over, objective by objective maps it line by line.
You were mid-prep and never sat it. Nothing is wasted, but nothing transfers automatically either. Azure AI Search, Document Intelligence, Content Understanding, Speech and Translator notes remain accurate. Anything you wrote about Custom Vision training, prompt flow, custom question answering or container deployment can go, because that material is not in AI-103. Treat AI-102 practice questions with suspicion, since a good share of them test cut content.
You never started. Start at AI-103 directly. There is no reason to study a retired blueprint, and the agent, evaluation and orchestration material that dominates AI-103 has no AI-102 equivalent to build on.
You are choosing between fundamentals and associate level. AI-901 replaced AI-900 on the same date. If you are new to Azure AI, AI-901 first is still the shorter path. If you already build software against cloud APIs, going straight to AI-103 is normally the better use of the time.