Azure Data Scientist Certification Guide

The Azure Data Scientist Associate certification and its DP-100 exam are retired. What the credential covered, what retirement means, and where to head next.

Daniel Carter12 August 2026 · 10 min read

Ask AIChatGPT Claude Gemini Copilot Perplexity Grok

Machine learning pipeline diagram showing the retired DP-100 certification path branching towards Microsoft's newer MLOps-focused credential

Before you spend a single evening on DP-100 study material, you need one fact in front of you: the Microsoft Certified: Azure Data Scientist Associate certification is retired. The official Microsoft Learn page for the credential now carries the warning that the certification and its renewal assessment are retired, and secondary reporting places the retirement date at 1 June 2026, as part of Microsoft's wider 2025–26 shake-up of its role-based certifications. You can no longer sit the DP-100 exam, earn the certification, or renew it.

That does not make this page pointless — quite the opposite. Thousands of people still search for the Azure data scientist certification every month, and most of the guidance they find was written before the retirement. This guide gives you the accurate picture as of 2026: what the DP-100 exam measured while it was live, what retirement means if you already hold the credential or were mid-way through studying, and how the skills it tested map onto Microsoft's current certification line-up.

What was the Azure Data Scientist Associate certification?

The Azure Data Scientist Associate was Microsoft's role-based credential for machine-learning practitioners on Azure, earned by passing a single exam: DP-100, "Designing and Implementing a Data Science Solution on Azure". It sat at the associate tier of Microsoft's certification structure — above the fundamentals certifications, below the expert level — and, like all Microsoft associate certifications, it was valid for one year with a free online renewal assessment.

The credential targeted people who built, trained and deployed machine-learning models using Azure Machine Learning: data scientists, ML engineers and analytically minded developers. Its final published skill outline covered four broad areas:

Designing and preparing a machine-learning solution — choosing compute, setting up Azure Machine Learning workspaces and managing data assets.
Exploring data and running experiments — working with notebooks, automated ML and experiment tracking.
Training and deploying models — building training pipelines, registering models and publishing them to online or batch endpoints.
Optimising language models for AI applications — the domain Microsoft added late in the exam's life, reflecting the shift towards generative AI, covering work with Azure ML, MLflow and what was then Azure AI Foundry.

Two things stand out from that outline even now. First, DP-100 was a hands-on practitioner's exam: it assumed working Python and expected you to know your way around the Azure Machine Learning service, not just recite theory. Second, the late addition of a language-model domain foreshadowed exactly the direction Microsoft has since taken its whole AI certification portfolio.

Why was DP-100 retired?

Microsoft is carrying out its largest certification overhaul since 2019, retiring roughly a dozen role-based certifications between mid-2025 and September 2026 and replacing several with new AI-centred credentials. DP-100 was one of the earliest casualties in the 2026 wave. Alongside it, the Azure Developer Associate (AZ-204) and Azure AI Engineer Associate (AI-102) certifications have also been retired, and Azure Security Engineer Associate (AZ-500) retires on 31 August 2026.

For the data-science role specifically, secondary sources report that Microsoft's successor is a new MLOps-focused associate certification built around exam AI-300, oriented towards production machine-learning operations and generative AI operations rather than classic model experimentation. Treat that description with appropriate caution: no official Microsoft Learn page for the new credential was available at the time of this research, and names and formats of new certifications can shift at launch. What is official — printed on the Learn page itself — is that DP-100 and its renewal assessment are gone.

One further point worth knowing, because it affects planning: reporting on the transition indicates there is no automatic migration, free upgrade, voucher or discount from the retired certifications to their replacements. If you want the successor credential, you will be starting it as a fresh candidate.

What retirement means for you

The consequences differ depending on where you were in the journey, so take the row that matches you.

If you already hold the Azure Data Scientist Associate

Your certification does not vanish. Retired Microsoft certifications remain on your transcript permanently, and you can continue to list the credential with its earned date on your CV and LinkedIn profile. What you lose is the ability to renew: the annual renewal assessment is retired along with the exam, so once your current validity period lapses, the certification shows as earned historically rather than active. There is no penalty and nothing you need to do — but if your employer or clients expect an active Microsoft credential, plan a route to one of the current certifications before your renewal date passes.

If you were mid-study for DP-100

Stop targeting the exam — it cannot be booked — but do not bin your notes. Almost everything DP-100 taught remains commercially valuable: Azure Machine Learning workspaces, compute management, MLflow experiment tracking, training pipelines and managed endpoints are all still core Azure skills. Your preparation converts well to the successor MLOps-oriented certification once its details are published, and partially to other live Microsoft data and AI credentials. Keep an eye on the official Microsoft Learn certification catalogue for the successor's launch details rather than relying on third-party summaries.

If you were planning to start from zero

You have been spared a false start. Choose a target from Microsoft's current line-up (next section) rather than working from pre-2026 blog posts and course syllabuses that still name DP-100 as the goal.

The skills DP-100 measured — and why they still matter

It would be a mistake to read "retired exam" as "obsolete skills". The DP-100 skill set is essentially a map of practical machine-learning work on Azure, and the platform features it examined are still the ones production teams use daily.

Designing and preparing an ML solution

This domain covered decisions that precede any model: selecting appropriate compute (instances versus clusters), provisioning and securing an Azure Machine Learning workspace, and registering datastores and data assets. In day-to-day work this is the difference between a notebook experiment and something a team can actually operate. Nothing here is wasted learning — the same design questions arise on every Azure ML project regardless of which certification badge is current.

Exploring data and running experiments

Exploratory analysis, feature work and structured experimentation — including automated ML for baseline models and experiment tracking so results are reproducible. MLflow, which DP-100 leaned on for tracking, remains an industry-standard tool well beyond Azure.

Training and deploying models

The domain closest to production: building pipelines that train models repeatably, registering model versions, and deploying to online endpoints for real-time inference or batch endpoints for scheduled scoring. If the successor credential is genuinely MLOps-centred, expect this territory to carry even more weight there, not less.

Optimising language models

Added late in the exam's life: working with language models in Azure's AI tooling. This is the thread Microsoft has pulled hardest on across its whole 2026 refresh — the new AI-era exams named in transition reporting (AI-200, AI-103, AI-300) all orbit generative AI. Time spent here positions you for wherever the portfolio lands.

How the exam worked while it was live

For the record — and because renewal-era holders still get asked about it — here is how DP-100 operated under Microsoft's standard exam policies:

Scoring: Microsoft technical exams are scored on a scale of 1–1,000 with a passing score of 700. That is a scaled score, not 70 per cent — a common misunderstanding worth correcting in interviews as well as exams.
Question counts: Microsoft does not publish question counts for its exams, DP-100 included. Any "the exam has 40–60 questions" claim you find online was folklore, not policy.
Delivery: exams were delivered through Pearson VUE, either at a test centre or online proctored from home.
Pricing: Microsoft prices exams by the country or region in which they are proctored rather than publishing one universal fee, so historical price claims should always have been checked against the official page.
Renewal: as an associate-tier certification it expired after one year, renewable free through an unproctored, open-book online assessment on Microsoft Learn — a policy that still applies to every live Microsoft role-based certification.

These mechanics matter going forward because they are portfolio-wide: whichever current Microsoft certification you pivot to, the 700/1,000 scaled pass mark, country-based pricing, Pearson VUE delivery and free annual renewal all still apply (fundamentals certifications, which never expire, aside).

Where to go instead in 2026

Staying within Microsoft's data and AI portfolio, several live certifications pick up different strands of what DP-100 covered. This guide will not compare them in depth — each has its own dedicated guide — but here is the honest one-line orientation for a displaced DP-100 candidate:

The successor MLOps-focused associate certification (exam AI-300, per secondary sources) is the closest spiritual heir, aimed at production ML and generative AI operations. Watch Microsoft Learn for official details before committing study time.
Azure Data Fundamentals (DP-900) suits you if your gap is foundational data literacy on Azure rather than modelling: a 45-minute, no-prerequisite fundamentals exam whose certification never expires. You can gauge the level with the free DP-900 practice questions on ExamPractice.
Fabric Data Engineer Associate (DP-700) is the current home for pipeline-and-platform data work; our Azure data engineer certification guide covers it fully.
The AI engineering track — historically AI-102, now also retired with a reported AI-103 successor — is covered in our Azure AI engineer certification guide.
Azure AI Fundamentals continues as a live certification, now examined through AI-901, which replaced the AI-900 exam in April 2026.

For how all of these fit together across the whole platform, the Azure certifications roadmap is the better starting point than any single-exam guide.

A redirect plan if DP-100 was your 2026 goal

A realistic scenario: a data analyst with solid Python and pandas, six months into self-study, who had pencilled in DP-100 for autumn 2026. Here is a sensible re-plan.

Audit what you have already learned against the four DP-100 domains above. Workspace management, experiment tracking and deployment knowledge all transfer; exam-specific trivia does not.
Decide whether your near-term value is "data" or "ML operations". If your role leans towards pipelines, warehousing and analytics engineering, DP-700 territory is live now. If it leans towards models in production, hold for the MLOps successor's official launch.
Bank a fundamentals credential in the meantime if you have none. DP-900 (or AI Fundamentals via AI-901) is a modest, never-expiring proof point that keeps momentum while the associate-level picture settles, and costs fundamentals-tier pricing that varies by country.
Keep building portfolio evidence. With the certification landscape mid-transition, a public repository showing an end-to-end Azure ML pipeline — data asset to registered model to endpoint — speaks louder than a retired badge ever did.
Re-verify everything at booking time. In a refresh cycle this aggressive, check the official Microsoft Learn certification page for status, pricing and format on the day you book, not the week you start studying.

When the successor exam does have a published skill outline, the same study discipline applies as with any Microsoft exam: work through the official outline domain by domain, get hands-on in a real Azure subscription, and use timed practice questions to expose weak domains rather than to memorise answers. ExamPractice's Microsoft exams hub tracks question sets across the Microsoft catalogue, including free samples you can try before subscribing.

Frequently asked questions

Can I still take the DP-100 exam anywhere?

No. The certification and exam are retired; Microsoft's official page states the certification and renewal assessment are retired, and the exam cannot be scheduled through Pearson VUE. Third-party sites still selling "DP-100 preparation" for a 2026 sitting are out of date.

Does my Azure Data Scientist Associate certification still count on my CV?

Yes. Retired certifications stay on your Microsoft transcript permanently with their earned dates. You simply cannot renew it, so it will eventually show as historical rather than active. List it honestly with dates.

What replaced the DP-100 exam?

Secondary sources name exam AI-300 and a new MLOps-focused associate certification as the successor, oriented towards production ML and generative AI operations. Microsoft had not published a full official page for it at the time of research, so confirm the name, format and requirements on Microsoft Learn before planning study.

Was DP-100 the same as the Azure Machine Learning certification?

Effectively yes in common usage — "Azure machine learning certification" almost always referred to the Azure Data Scientist Associate earned via DP-100, since Azure Machine Learning was the service at its core. There was never a separate certification named for the service itself.

Do I get a free upgrade to the new certification if I hold DP-100?

Reporting on the transition says there is no automatic migration, voucher or discount to successor certifications. Expect to sit the new exam as a fresh candidate at standard country-based pricing.

Your Azure data-science path from here

The honest summary of the Azure data scientist certification in 2026 is that the credential's story has ended but the role's has not. DP-100 spent its life certifying exactly the skills Azure ML teams still hire for — solution design, disciplined experimentation, reliable deployment — and Microsoft's refresh redirects that territory towards MLOps and generative AI rather than abandoning it.

So: existing holders should keep the badge listed and plan a route to a live certification before their renewal window closes; interrupted candidates should preserve their study investment and watch for the successor's official launch; and newcomers should start from the current catalogue, not cached advice. Whichever branch you take, verify status and pricing on the official Microsoft Learn page at booking time — this portfolio has changed more in eighteen months than in the previous five years, and it may not be finished yet.

Exam facts in this guide were checked against official certification-provider pages on 30 August 2026. Fees, exam codes and policies change — confirm on the provider’s own site before you book.