AI-900 Has Retired: Your Guide to Microsoft's New AI-901 Exam and What It Means for Your Team
Somewhere in your organisation right now, there's a development plan, a training spreadsheet, or a slightly dusty onboarding checklist with 'AI-900' written on it. Time to grab a red pen. Microsoft officially retired the AI-900 exam at the end of June, and AI-901 is now the exam used to earn the Microsoft Certified: Azure AI Fundamentals certification.
If that news triggered a small flutter of panic in your L&D team, take a breath, because the reality is far gentler than the headline. Nobody's certification has vanished overnight, nobody needs to re-sit anything, and for teams just starting their AI journey, the change is genuinely good news. The retirement notice sits in plain sight on Microsoft's AI-900 study guide, and once you understand what actually changed, the whole thing becomes much less dramatic than it first sounds.
So what actually changed at the end of June?
Microsoft retired the exam, not the credential, and that distinction trips up almost everyone at first. The Microsoft Certified: Azure AI Fundamentals certification is alive and well. What's different is the exam that earns it, and a few things retired alongside it that are worth knowing about:
The AI-900 exam retired on 30 June 2026. AI-901 is now the exam required to earn Microsoft Certified: Azure AI Fundamentals.
The certification itself continues under the same name. Existing Azure AI Fundamentals credentials remain valid because Microsoft Fundamentals certifications do not expire.
The AI-900T00 instructor-led course has been replaced by AI-901T00: Introduction to AI in Azure.
A Practice Assessment for AI-901 is available through Microsoft AI Skills Navigator.
Think of it like your favourite local cafe changing hands. The sign over the door is new, and the menu has been refreshed, but the coffee is still made every morning, and your loyalty card still works. Honestly, the new menu is better, and we'll get to why in a moment.
What the new Microsoft AI-901 exam covers
AI-900 was a product of its time. It launched before generative AI reshaped every conversation about workplace technology, so it spent most of its energy asking candidates to describe individual Azure services one after another: here's machine learning, here's computer vision, here's natural language processing. Solid foundations, but a bit like learning about a kitchen by memorising the appliances rather than cooking anything.
AI-901 brings the certification into the world we actually work in now, with Microsoft Foundry playing a central role. According to the current skills outline, 40–45 per cent of the exam focuses on identifying AI concepts and capabilities, while 55–60 per cent focuses on implementing AI solutions using Microsoft Foundry. The additions your team will notice most include:
Generative AI takes a much larger role, including model selection and deployment, prompt design, lightweight chat applications and interaction with deployed models through Microsoft Foundry.
AI agents are now part of the syllabus, including creating and testing a single-agent solution in the Foundry portal and building a lightweight client application for an agent.
Azure Content Understanding covers information extraction from documents, forms, images, audio and video.
The structure now places less emphasis on memorising individual service names and traditional machine learning categories, and more emphasis on identifying AI workloads and implementing lightweight solutions.
The reassuring part is that Microsoft still classifies AI-901 as a beginner-level exam. However, it is more technically oriented than AI-900. Candidates should understand basic Python syntax and programming techniques and be familiar with Azure resources. Knowledge of core cloud concepts is also recommended. The course can still suit mixed groups, but learners with no technical background may need some preparation before attending.
From naming services to understanding how solutions fit together
The shift is worth pausing on. AI-901 still tests whether candidates understand core AI concepts, but more than half of the current skills outline focuses on implementation using Microsoft Foundry. Candidates need to understand how models, prompts, agents and lightweight applications fit together, while also recognising where responsible AI considerations belong. People who learn by doing rather than by memorising definitions may find this more applied framing a useful way into the subject.
Already certified? Your credential is safe
Here’s the part that catches people out, in a good way. Microsoft Fundamentals certifications do not expire, so anyone on your team who passed AI-900 keeps their Azure AI Fundamentals credential. It remains on their Microsoft Learn transcript and can still be shared on LinkedIn and included on a CV. It also provides useful conceptual preparation for current role-based certifications such as Microsoft Certified: Azure AI Apps and Agents Developer Associate, earned through exam AI-103.
People who already hold Azure AI Fundamentals do not need to take AI-901 to retain their certification. Those who want to update their knowledge of Microsoft Foundry and agentic AI can use the new AI-901 learning content or progress to current role-based training. Passing AI-901 leads to the same Azure AI Fundamentals certification rather than a separate credential.
Halfway through studying when the deadline passed?
This group has the most adjusting to do, but much of the conceptual foundation remains useful. Responsible AI principles, language, speech and computer vision workloads all continue to matter. However, AI-901 introduces considerably more applied content, including prompt design, model deployment, lightweight applications, agents, image generation and Content Understanding. Candidates should also build or refresh their knowledge of Python, Azure resources, REST APIs, SDKs and command-line tools.
One practical warning from our side, though: search results still contain a large amount of AI-900 study material, and some of it is now out of date. Publication date alone is not a reliable indicator. Candidates should use resources that explicitly reference AI-901 and align with Microsoft’s current skills outline, rather than revising against material written for the retired exam.
Getting your team certified under AI-901 with Lumify Work
For organisations building AI capability at scale, the cleanest route is the new one-day instructor-led course. Lumify Work delivers AI-901T00: Introduction to AI in Azure across New Zealand, taught by Microsoft Certified Trainers from the organisation with the highest number of Microsoft Certified Trainers in ANZ. The course maps directly to the Microsoft Certified: Azure AI Fundamentals certification and is designed for aspiring technology professionals at the beginning of their AI solution development journey. Some knowledge of Python coding is useful, and familiarity with core cloud concepts is recommended.
The timing lands rather neatly, too. Most Australian organisations are setting FY27 training budgets right now, and New Zealand teams are deep in their own planning cycles, so folding AI-901 into new starter onboarding and capability plans this quarter means everyone certifies against the current exam rather than chasing a retired one. One decision, made once, and the whole question is sorted for the year.
From there, the pathway opens right up. Team members who catch the bug can progress into role-based training across the broader Microsoft course catalogue, from Azure AI engineering through to Copilot skills for specific business functions. The retirement of AI-900 isn't a disruption so much as a reset of the starting line, and teams that move early will be certifying against the exam that Microsoft is actively building its AI pathway around. To get your people booked onto AI-901 training, or to talk through an AI upskilling plan for the year ahead, contact Lumify Work.
