If you have been studying for DP-100, stop and read this first. The Azure Data Scientist Associate certification retired on 1 June 2026, and Microsoft has replaced it with AI-300, a different exam testing a different job. You can no longer book DP-100, and if you already hold it, you cannot renew it either.
The official Microsoft Learn page for the certification now carries a single warning at the top: "This certification and the renewal assessment are retired." Under the preparation section it says "No training available for this exam." That page has been pulled out of Microsoft's search index, which is why most of the DP-100 study guides and course pages still ranking on Google have not been updated and still tell you to book it.
Here is what actually replaced it, what happens to a DP-100 sitting on your transcript right now, and whether AI-300 is the exam you should switch to.
Is DP-100 retired? Yes, on 1 June 2026
DP-100 tested Azure Machine Learning: preparing a workspace, exploring data, running experiments, training models, deploying them, and more recently optimising language models for AI applications. Four skill areas, all of them centred on building a model.
Microsoft retired both the exam and the free renewal assessment on 1 June 2026. The replacement, AI-300, went live in May 2026 under a new certification name: Microsoft Certified: Machine Learning Operations Engineer Associate.
Exam Tip: If a study guide, Udemy course or practice test is still selling itself as "DP-100 2026 preparation", check its last updated date before you spend money. The exam has not been bookable since 1 June 2026.
What replaced DP-100: the AI-300 MLOps Engineer Associate certification
AI-300 is called Operationalizing Machine Learning and Generative AI Solutions, and the name is doing real work. Microsoft has reframed the role away from the person who trains the model and towards the person who keeps it running in production.
The official skills measured, taken from the Microsoft Learn study guide last updated on 5 March 2026, are:
| AI-300 domain | Weighting |
|---|---|
| Design and implement an MLOps infrastructure | 15 to 20% |
| Implement machine learning model lifecycle and operations | 25 to 30% |
| Design and implement a GenAIOps infrastructure | 20 to 25% |
| Implement generative AI quality assurance and observability | 10 to 15% |
| Optimize generative AI systems and model performance | 10 to 15% |
Add the last three rows together. Between 40% and 55% of AI-300 is generative AI operations: deploying foundation models, versioning prompts in Git, measuring groundedness and relevance, tracking token spend, and tuning retrieval-augmented generation. DP-100 touched none of that until a late addition on optimising language models.
The exam itself is 120 minutes, proctored through Pearson VUE, and scored out of 1000 with 700 to pass. It is currently offered in English only, which is a step down from DP-100's ten languages. Microsoft prices it by country rather than publishing one global figure, and associate-level role-based exams list at USD 165 in the United States.
There are no formal prerequisites, but the audience profile is specific: Python, Azure Machine Learning, Microsoft Foundry (the service formerly called Azure AI Foundry), GitHub Actions, and infrastructure as code with Bicep and the Azure CLI.
DP-100 vs AI-300: what actually changed
| DP-100 (retired) | AI-300 (current) | |
|---|---|---|
| Certification | Azure Data Scientist Associate | Machine Learning Operations Engineer Associate |
| Core question | Can you build and deploy a model? | Can you run models and AI apps in production? |
| Generative AI coverage | One domain added late | 40 to 55% of the exam |
| DevOps expectation | Light | GitHub Actions, Bicep, Azure CLI, IaC |
| Monitoring | Model performance | Drift detection, tracing, latency, token cost |
| Duration | Retired 1 June 2026 | 120 minutes |
| Languages | Ten | English only |
| Renewal | Retired, no path | Free annual assessment |
The practical read: if you were three weeks into DP-100 revision on Azure Machine Learning pipelines and MLflow, roughly a quarter to a third of that work still counts, because the model lifecycle domain is the largest single block on AI-300. The rest of your study time now goes on infrastructure automation and generative AI evaluation, which are genuinely new skills rather than a renamed version of what you already knew.
What happens to the DP-100 you already hold
This is the part that catches people out, and it is worth being precise about.
Microsoft associate certifications expire twelve months after you earn them. You keep them current by passing a free, unproctored, open-book renewal assessment on Microsoft Learn during a six-month eligibility window before the expiry date. Fundamentals certifications never expire; associate ones always do.
DP-100's renewal assessment retired alongside the exam. There is no assessment left to pass, so there is no way to extend the credential.
What that means in practice:
- Your Azure Data Scientist Associate certification stays valid until its own expiry date, twelve months from the day you earned or last renewed it.
- When that date passes, it moves out of the active section of your Microsoft Learn transcript.
- It does not disappear. It stays permanently on your transcript as a past certification and remains verifiable by an employer.
- You cannot get it back, and you cannot convert it into AI-300. Microsoft has not offered a bridging exam.
So a DP-100 earned in, say, March 2026 stays active until March 2027 and then becomes a historical credential. That is not a disaster on a CV. It is a dated one, and a recruiter screening for Azure AI skills in 2027 will be searching for the current credential name.
Exam Tip: Check your actual expiry date at aka.ms/ManageCerts rather than assuming. It runs from the date you earned the certification, not from the retirement date, so two people who both hold DP-100 can have expiry dates ten months apart.
Should you switch to AI-300, or take a different exam?
AI-300 is the direct successor, but it is not automatically the right move. It is an operations exam, and plenty of people who booked DP-100 wanted a data science credential rather than a DevOps one.
Work through it by what you actually do:
- You deploy and maintain models, run pipelines, or own the production side of ML. AI-300 is a clean fit and the depth on drift detection and endpoint rollout will map onto your day job.
- You build AI applications and agents rather than operating them. Look at AI-102 and AI-103 instead. Our AI-102 vs AI-103 comparison covers which one matches which role.
- You are new to Azure AI and DP-100 was going to be your first exam. AI-300 is an intermediate certification with a heavy tooling assumption. Start with the fundamentals tier and work up: our AI-901 four-week study plan is the sensible entry point.
- You work on the data platform side rather than the model side. DP-600 and DP-700 sit in Microsoft Fabric and were untouched by this retirement.
If AI-300 is the right call, the AI-300 eight-week study plan breaks the five domains into a week-by-week schedule and flags the Microsoft Foundry material that trips up people arriving from a pure data science background.
DP-100 is one of a long list of Microsoft retirements this year. If you hold other Microsoft credentials, the full 2026 retirement list shows every exam going away and what replaces each one, so you can check the rest of your transcript in one pass.
Frequently Asked Questions
Is DP-100 retired?
Yes. Exam DP-100 and the Microsoft Certified: Azure Data Scientist Associate certification retired on 1 June 2026, along with the free renewal assessment. The exam cannot be booked and the certification cannot be renewed.
What is DP-100 certification?
DP-100 was the exam behind the Azure Data Scientist Associate certification. It tested designing a machine learning solution on Azure, exploring data and running experiments, training and deploying models with Azure Machine Learning and MLflow, and optimising language models for AI applications.
What is AI-300 certification?
AI-300, titled Operationalizing Machine Learning and Generative AI Solutions, earns the Machine Learning Operations Engineer Associate certification. It covers MLOps infrastructure, model lifecycle and operations, GenAIOps infrastructure on Microsoft Foundry, generative AI evaluation and observability, and performance optimisation including RAG tuning and fine-tuning.
What are the prerequisites for the AI-300 certification?
There are no formal prerequisites. Microsoft's audience profile expects a data science background with Python, hands-on experience with Azure Machine Learning and Microsoft Foundry, and entry-level DevOps skills including GitHub Actions, command-line tools, and infrastructure as code with Bicep and the Azure CLI.
How difficult is the AI-300 exam compared with DP-100?
The tooling surface is wider. DP-100 concentrated on Azure Machine Learning, while AI-300 adds Microsoft Foundry, GitHub Actions, Bicep and the Azure CLI, and expects you to reason about token cost, latency and evaluation metrics such as groundedness. Candidates coming from a notebook-and-model background usually find the infrastructure automation domains harder than the modelling ones.
Ready to Start Practising?
Reading about a syllabus change is the easy part. Finding out which of the five AI-300 domains you would currently fail is the useful part, and it takes about twenty minutes of exam-style questions.
CertCrush gives you practice questions with full explanations for Microsoft's current certification track, so you can work out where the gaps are before you book anything at USD 165. Create a free account and start with the domains you have never touched, not the ones you already know.
Browse the full CertCrush course catalogue to see which Microsoft exams are covered.