Microsoft Azure AI Fundamentals (AI-901, formerly AI-900): Cost, Format, and the Never-Expires Badge

If you came here searching for AI-900, start with the update that changes your study plan: Microsoft retired the AI-900 exam on June 30, 2026, and the Azure AI Fundamentals certification is now earned by passing a different, more technical exam called AI-901. The certification itself did not go away, and it still carries the one property no other credential on this site offers, it never expires, but the exam behind it is not the one older prep material was written for. This page covers the transition, the real cost, the new exam, and the never-expires model, checked against Microsoft Learn on the date at the top.

AI-900 is retired; AI-901 is the exam now

Microsoft’s exam pages make the swap explicit: AI-900 retired on June 30, 2026, and to earn Azure AI Fundamentals candidates now pass AI-901. This is not a cosmetic renumber. The old AI-900 was a describe-only, no-code exam aimed squarely at non-engineers. AI-901 is restructured around just two domains, and the balance tells you where it moved: identify AI concepts and capabilities (40 to 45 percent) and implement AI solutions by using Microsoft Foundry (55 to 60 percent). More than half the exam is now an implement domain, and Microsoft’s audience note asks for conceptual Azure knowledge plus some familiarity with Python syntax and comfort with SDKs, CLIs, and REST APIs. None of that appeared in AI-900’s positioning. In short, the new fundamentals exam sits a notch higher and a notch more hands-on than the one it replaced.

One practical consequence: at the time of writing AI-901 is offered in English only, where AI-900 shipped in more than a dozen languages, so non-English candidates should check current language availability before booking.

Who should get it, and who should skip it

AI-901 is built for people at the beginning of an AI-solutions career and for the non-engineering staff who orbit Azure AI work: analysts, project managers, and career changers who need working vocabulary and a low-cost way to test their own interest. If that is you, and especially if you already earned AZ-900, this is the AI-flavored companion to that badge.

Skip it if you already do machine learning or data science for a living; a fundamentals exam will feel shallow, and your money is better aimed at a hands-on credential. This is also where an honest comparison matters. A fundamentals badge is a describe-and-recognize signal, not a data-scientist qualification. The credential that actually functions as a hiring filter for that work is the harder, hands-on Azure Data Scientist Associate (exam DP-100), which tests training, deploying, and managing real models rather than describing them. Do not expect AI-901 to carry a data-science hiring decision on its own; it is the on-ramp, and the associate-tier DP-100 track is the credential a data-scientist posting has in mind. Choose based on where you actually are: vocabulary and interest test, or the job filter.

What it costs all-in

The exam is $99, Microsoft’s standard Fundamentals price, set per country and currency with tax in some regions. There is no free retake; a failed attempt is another $99, with a 24-hour wait before the second try and 14-day waits after that.

Prep can genuinely cost nothing. Microsoft Learn hosts a free, self-paced AI-901 learning path built from the same skills outline the exam is written against, and Microsoft periodically runs Virtual Training Days that have historically included exam vouchers. Because AI-901 leans into implementation, spending time actually clicking through Azure AI services and Microsoft Foundry beats a second video course. A $15 to $20 course is a reasonable structure add, but the honest floor here is close to the voucher price plus zero. See how that ranks on the ROI Index.

The exam itself

Fundamentals exams run 45 minutes, and Microsoft cites a typical 40 to 60 questions without publishing a fixed count. Question types are multiple-choice and multi-select, with no lab component at this tier. Passing is 700 on a 1,000-point scale, and Microsoft publishes no pass rate, so any specific figure in a prep forum is invented. The current outline is dated April 15, 2026; because AI-901 is a new exam, its free official practice assessment may still be rolling out, so check the exam page for availability rather than assuming it exists.

The calibration note that applied to the old exam still applies, just at a slightly higher altitude: this is a breadth exam that now includes a real implement emphasis. It asks you to identify capabilities and to implement solutions with Microsoft Foundry, not to architect or operate production systems. If your practice material has you tuning models or managing pipelines end to end, it is teaching Azure Data Scientist content, not this exam.

Renewal math over 9 years

Zero dollars, and that is the whole point. Microsoft’s published expiration policy states that Fundamentals certifications do not expire, so there is no renewal assessment, no continuing-education units, no annual fee, and no window to miss. Over nine years Azure AI Fundamentals costs exactly what it cost on day one: $99 plus whatever you spent on prep.

That permanence is genuinely rare in this dataset, and it cuts both ways, more sharply here than almost anywhere. A 2026-vintage AI-901 badge will still read as active in 2035, but AI is the fastest-moving subject we track, and an employer reading a nine-year-old AI fundamentals badge will know the field has turned over several times since. The credential does not expire; its freshness does, faster than most. Certification renewal costs explained shows how rare a no-maintenance credential is in the registry: only the Microsoft Fundamentals tier carries no renewal requirement of any kind.

Nine-year cost of ownership: $99, and it does not rise. Fundamentals certifications never expire, so there is no renewal fee across nine years or ninety, and the day-one price is the whole price.

What it does for the occupation you are entering

We map this badge to the data scientist occupation (BLS code 15-2051), and the mapping deserves a sentence because it is a judgment call. A fundamentals holder is, by Microsoft’s own audience definition, at the very start of an AI-solutions path, not yet a practicing data scientist or AI engineer. The panel on this page shows that occupation’s national median wage and headcount as context for the field the badge orbits, not a payout attached to earning it. What AI-901 concretely buys is smaller and real: a resume keyword for Microsoft-stack employers, proof you can follow an Azure AI conversation, and a cheap, low-risk test of your own interest before committing to the associate-tier track where the actual hiring filters live.

Common mistakes

Buying AI-900 prep in 2026. The exam retired on June 30, 2026. Older AI-900 courses skew describe-only and predate the Microsoft Foundry implement domain that is now more than half of AI-901. Study to the current AI-901 outline, not the retired one.

Treating it as a data-science qualification. AI-901 rarely appears as a hard requirement the way an associate credential does. It is a stepping stone and a signal of intent; the job filter for AI and data-science roles is the hands-on Azure Data Scientist Associate (DP-100) track, not this one.

Assuming it is still the no-code exam. AI-901 expects some Python literacy and comfort with SDKs, CLIs, and REST APIs, and it puts real weight on implementing solutions in Microsoft Foundry. Candidates who prepped as if this were the old non-technical AI-900 are aiming at the wrong exam.