We earn a commission if you buy through links on this page. How we're funded.
Quick answer
One question settles this comparison for most people: which cloud does your employer — or the employer you want — actually run? Certify that one. One change matters before anything else: Microsoft retired AI-900 on 30 June 2026, and Azure AI Fundamentals now requires exam AI-901, which expects basic Python. So if you are genuinely neutral and do not code, the AWS Certified AI Practitioner (AIF-C01) is now the no-code on-ramp; if you can write a little Python, AI-901 is the Microsoft equivalent. Free official preparation covers either; for AIF-C01 the one paid item we would add is a practice set, Udemy’s AWS Certified AI Practitioner AIF-C01 Practice Exams, a question bank with no video.
Exam AI-900: Microsoft Azure AI Fundamentals is retired. The replacement is Exam AI-901: Microsoft Azure AI Fundamentals, which earns the same Azure AI Fundamentals certification but expects Python and familiarity with REST APIs and SDKs.
See AWS Certified AI Practitioner AIF-C01 Practice Exams on Udemy →
The table below compares 2 certifications on provider, level, realistic time, coding needed and best for.
| Certification | Provider | Level | Realistic time | Coding needed | Best for |
|---|---|---|---|---|---|
| AWS Certified AI Practitioner (AIF-C01) | AWS | Foundational | ~4–6 weeks of prep | No | AWS-stack workplaces; broader generative-AI scope |
| Azure AI Fundamentals (AI-901) | Microsoft | Foundational | ~2–4 weeks of prep | Basic Python | Microsoft-stack workplaces; replaced AI-900 on 30 June 2026 |
Which exam should you take?
Match the exam to the cloud you will actually touch. These are vendor credentials — their value is highest inside environments that run that vendor's stack, and close to zero as abstract knowledge badges. Check what your employer or target job runs; job adverts state it plainly, and so does your IT department.
If you have no stack signal at all — career changers, students, generalists — this used to be an easy call for AI-900: shorter preparation, a lower fee and a gentler exam. AI-900 was retired on 30 June 2026, and its replacement, AI-901, expects basic Python, so a neutral reader who does not code should now take AIF-C01, and one who can write a little Python can choose on stack alone. Our guide to the easiest respected AI certifications covers the other accessible proctored options. If your interest is specifically generative AI workloads, both now centre on them — AIF-C01 through scenarios, AI-901 through small builds; our AWS AI Practitioner review breaks down the AWS side in full.
The short version, by situation:
- Your employer runs AWS: AIF-C01.
- Your employer runs Azure and you can write basic Python: AI-901.
- Your employer runs Azure and you do not code: learn enough Python to read a short script before booking AI-901, because Microsoft's study guide asks for it; AIF-C01 is the no-code proctored alternative, though it names AWS services throughout.
- No stack signal and no code: AIF-C01.
- No stack signal, some Python, and you want a certificate that never expires: AI-901.
- You will build AI apps on Azure next: AI-901 as a warm-up, or straight to AI-103 if you already develop in Python.
What does each exam actually cover?
Both vendors publish how their exam is split, and the two splits explain most of what follows on this page. The table quotes AWS's AIF-C01 exam guide and Microsoft's AI-901 study guide (skills measured as of 15 April 2026), both read on 16 October 2026; each exam's fee and length are in the vendor table further down.
The table below sets AIF-C01 against AI-901 on five questions: how the marks are split, whether code is expected, how many questions count, what score passes, and who each vendor writes the exam for.
| Question | AWS AIF-C01 | Microsoft AI-901 |
|---|---|---|
| How are the marks split? | Fundamentals of AI and ML 20%; fundamentals of generative AI 24%; applications of foundation models 28%; guidelines for responsible AI 14%; security, compliance and governance for AI solutions 14% | Identify AI concepts and capabilities 40–45%; implement AI solutions by using Microsoft Foundry 55–60% |
| Is code expected? | No: AWS lists “developing or coding AI/ML models or algorithms” as out of scope | Yes: candidates “need knowledge of Python coding syntax and programming techniques” |
| How many questions count? | 50 scored, plus 15 unscored questions that are not identified on the exam | Microsoft publishes no question count |
| What score passes? | 700 on a scale of 100–1,000, scored as a whole: you do not need to pass each domain | 700 or greater |
| Who is it written for? | Someone with up to six months of exposure to AI/ML on AWS, who “uses but does not necessarily build” AI solutions | Someone “at the beginning of your career in AI solution development”, familiar with Azure resources |
Read the splits and the two exams stop looking interchangeable. More than half of AIF-C01's scored content sits in its generative-AI and foundation-model domains, and more than a quarter in responsible AI, security and governance: it is an exam for people who choose, buy and govern AI on AWS. More than half of AI-901 is building in Microsoft Foundry, which is why it now asks for Python. Both cover core AI concepts and responsible use; AIF-C01 tests them through scenarios with no code, AI-901 through small apps and agents.
Which exam is harder?
For a reader who does not code, AI-901 now. Against AI-900 the answer was AIF-C01, modestly — broader scope, more scenario-style questions and more governance material, which is why our prep estimate ran four to six weeks against two to four for AI-900. AI-901 changes the comparison: it expects basic Python and weights the exam towards building in Microsoft Foundry, while AIF-C01 still puts coding out of scope. For a working developer the gap is small, and neither approaches associate-level difficulty. The AWS exam still punishes anyone who skips the foundation-model application material.
Do not let 'harder' translate to 'better' in your head. Both are foundational stamps; difficulty at this level buys no extra market weight — the employer's stack does. Our analysis of whether AI certifications are worth it covers why context beats rigour at the entry tier.
Not sure this is the right one for you?
Tell the picker about your background and what you want the certificate to do, and it narrows the list to the one or two courses we would start with. It suggests only our affiliate partners’ courses, and says so before it suggests anything.
Try the AI Certification Picker →What do they cost — and do they expire?
Fees: both vendors publish the current fee on the exam page — AWS in US dollars, Microsoft priced by the country you sit in — and historically the Azure fundamentals exam has been the cheaper of the two, but confirm both before booking. Retake and voucher policies differ by region (see the provider's page for current details).
The table below shows the fee and format of AIF-C01 and AI-901, as each vendor publishes them. Every other exam we track is compared on our AI certification exams page.
| Credential | Vendor | Level | Fee | Exam format | Enrol |
|---|---|---|---|---|---|
| AWS Certified AI Practitioner (AIF-C01) | AWS | Foundational | 100 USD | 90 minutes · 65 questions | AWS → |
| Microsoft Certified: Azure AI Fundamentals (AI-901) | Microsoft | Foundational | $99 USD (varies by country/region) | 45 minutes | Microsoft → |
Expiry differs and matters: Microsoft's fundamentals certifications do not expire, while AWS AI Practitioner is valid for three years, and AWS lets you recertify by passing the latest version of the exam or, automatically, by earning the Machine Learning Engineer – Associate. If you want a set-and-forget line on your CV, that difference favours Azure AI Fundamentals.
Which carries more weight with employers?
Whichever matches the hiring team's stack — there is no absolute winner. In a Microsoft shop, Azure AI Fundamentals reads as directly relevant and AIF-C01 reads as generic initiative; reverse the shop and the reading reverses. Recruiters at cloud-partner consultancies filter by the exact certification names their practice sells against, which is the strongest version of the stack rule.
Two honest caveats. First, both are entry stamps — they open conversations, not roles; the associate tier (AI-103 on Azure, or deeper AWS credentials) is where technical hiring managers start paying attention. Second, if the role is non-technical, the proctored format itself is most of the signal — a verified, closed-book exam outweighs a course completion certificate of similar scope, a distinction our ranking of the top AI certifications deliberately keeps out of its scoring: the vendor exams are named there unscored rather than ranked against taught courses.
How should you prepare — and can you do it free?
Both vendors publish free official preparation. Microsoft publishes a free AI-901 study guide and a free practice assessment, and AWS Skill Builder carries free AIF-C01 preparation content. For most readers the free path plus one set of practice questions is enough at this level — the paid bootcamps sold against these exams are rarely worth it, as our roundup of free AI certifications argues.
- AI-901: work from the official study guide, build the small Foundry exercises its outline describes, then sit the free practice assessment until you clear it comfortably.
- AIF-C01: complete the Skill Builder plan, and give the foundation-model application domain double time — it is where unprepared candidates fail.
For the one set of practice questions on the AWS side, we would use Udemy’s AWS Certified AI Practitioner AIF-C01 Practice Exams: four full practice exams, 260 questions in all, with no video, bought once. Use it after the Skill Builder plan, to find weak domains, not as a substitute for it; it ends in Udemy’s certificate of completion, and the certification comes only from AWS’s exam. On the Microsoft side, AI-901’s free practice assessment does the same job.
Should anyone take both?
Rarely. Consultants, MSP staff and freelancers who genuinely straddle client clouds get real value from holding both names. Everyone else should spend the second exam's prep hours going deeper instead: an associate-level credential on your primary cloud — see our AI-901 vs AI-103 guide for what that step involves — or an applied project. Data-platform readers weighing the deeper path should start from our data engineers' guide.
If speed is your constraint rather than depth — an interview this month, a review cycle closing — our guide to the fastest AI certifications sequences the quickest credible options.
Where cloud-exam comparisons go wrong
Most of them rank the two exams in a vacuum — difficulty scores, question counts, pass rates — as if the credential's value were intrinsic. It is not. A cloud fundamentals certificate is a context-dependent asset: AIF-C01 in an Azure shop is trivia; AI-901 in an AWS shop is trivia. The ranking that matters is written in your employer's infrastructure bill, not in a listicle.
The second failure is the fundamentals-collection trap. Some advice recommends sweeping all three clouds' entry exams 'for breadth'. That is three exam fees and three prep cycles for three versions of the same introductory material — busy-work that signals indecision rather than initiative. Take one fundamentals exam, then go deeper or go applied; our three-cloud comparison makes the full case, and no-code readers can sanity-check the landscape in our no-coding guide.
Verdict
Certify the cloud you will touch: Azure AI Fundamentals (now exam AI-901) in Microsoft environments, AWS AI Practitioner in AWS environments. With no stack signal, take AIF-C01 if you do not code — AI-900, the gentler no-code exam this page used to recommend, was retired on 30 June 2026 — and AI-901 if you can write a little Python and want a certificate that never expires. Neither is a destination: plan the next step up your chosen stack, or across to applied work, using our AI certification roadmap. Still torn between clouds? Our free Picker tool matches you in a few questions.
Ready to start?
Bought once, with what Udemy calls lifetime access. Udemy's price swings between its list price and a sale price, sometimes within days — check it on the day rather than trusting any figure you read, here or anywhere else.
Frequently asked questions
Is AWS AI Practitioner harder than Azure AI-900?
It depends which Azure exam you mean, because AI-900 was retired on 30 June 2026. Its replacement for Azure AI Fundamentals, AI-901, expects basic Python and puts 55–60% of its weight on building in Microsoft Foundry. Against AI-900, AIF-C01 was modestly harder — broader generative-AI and governance scope, and typically four to six weeks of part-time preparation against two to four. Against AI-901, AIF-C01 is the easier exam for anyone who does not code, since AWS puts coding out of scope. Both sit well below associate-level difficulty.
The extra ground AIF-C01 had over AI-900 was mostly generative and governance material. AI-901 now names generative and agentic AI as workloads too, so the difference is no longer currency but method: AIF-C01 asks you to recognise and choose, AI-901 asks you to build small solutions in Foundry. That makes AIF-C01 the natural pick for someone who will buy, govern or explain AI on AWS, and AI-901 the one for someone who will build on Azure — a real trade rather than a ranking.
Which is cheaper, AIF-C01 or AI-900?
Historically the Azure fundamentals exam (AI-900, now replaced by AI-901) carried the lower fee, but both vendors adjust pricing and Microsoft prices by country — check the official exam pages before booking. AWS lists AIF-C01 at 100 USD. Free official preparation exists for both, so the exam fee should be your only cost at this level.
Because preparation is free on both sides, the fee is the whole price and a voucher removes it entirely. Both vendors periodically issue them through their own events, virtual training days and partner programmes, and employers frequently hold unused ones nobody has claimed. Ask internally before paying; at this tier that single question is worth more than any comparison of the two list prices.
Do AWS and Azure AI certifications expire?
They differ. Microsoft fundamentals certifications such as Azure AI Fundamentals do not expire, while AWS runs a recertification cycle on its credentials (see the provider's page for current details). If long-term validity without maintenance matters to you, that difference favours the Azure exam.
Read that difference carefully rather than treating non-expiry as a straight win, though. A credential that never lapses is a statement about what you knew on one day, and a reader weighs it accordingly; a renewed one says you were current this year. Non-expiry is genuinely convenient at the fundamentals tier, where nobody expects the certificate to prove current capability anyway — it matters far less than it sounds.
Can I take both AWS AI Practitioner and Azure AI Fundamentals?
You can take both credentials — Azure AI Fundamentals is now earned through AI-901, which replaced AI-900 on 30 June 2026 — but few people should. The material overlaps heavily at the concept layer, so the second exam mostly buys a duplicate stamp. Consultants and managed-service staff who work across both clouds are the exception; everyone else gets more from one fundamentals exam plus an associate-level step or an applied project.
Two fundamentals certificates also read a particular way to a hiring manager: as someone who enjoys starting. One foundational exam plus evidence of applying it, or plus a harder exam, describes a direction; two at the same tier from different vendors describes a habit. The exception is genuine — consultants are asked which clouds they cover — and it is narrow.
Which should a beginner take first?
AIF-C01, unless your target employer runs Microsoft. AI-900 used to be the gentler first exam, but Microsoft retired it on 30 June 2026 and its replacement, AI-901, expects basic Python — so for a beginner who does not code, the AWS exam is now the gentler first experience of proctored certification. Complete beginners should do a short no-code course first — Google AI Essentials or Elements of AI — so the exam vocabulary lands on familiar ground.
“Unless your target employer runs Microsoft” should override everything else in that answer, including the difficulty. A credential on the platform you will actually work with converts into conversations, project assignments and credibility; the gentler exam on the other vendor's stack converts into a line nobody follows up. Find out what your organisation runs before you book anything.
Keeping this current. Course formats, prices, and certification exam fees change and vary by region. We review our guides regularly, and we always recommend confirming the specifics on the provider's official page before you enrol.