Take AZ-900 First Unless You Have A Specific Reason Not To
If you only want one answer: for almost everyone starting in the Microsoft cloud in 2026, AZ-900 (Azure Fundamentals) is the right first certification. It teaches the platform vocabulary — subscriptions, regions, compute, storage, networking, identity, pricing, governance — that AI-900 and SC-900 both quietly assume you already understand. The two common exceptions: take AI-900 first if your job or target role is explicitly about AI products and you have no near-term cloud-infrastructure path, and take SC-900 first if you are moving into a security, compliance, identity, or GRC role where Microsoft Entra and Microsoft Purview are the daily tools.
All three are beginner-level, ~$99 USD in the United States, 45 minutes of exam time, 700/1000 to pass, and none of them expire — Microsoft fundamentals certifications are permanent and do not require renewal (Microsoft credential expiration policy). So the real decision is not "which is best" — they all hold lifetime value for under $100. The decision is sequence: which one builds the foundation for your next move, and which one you can safely take second or skip.
This post is a decision guide, not an exam walkthrough. For the full blueprint, cost, format, and study plan of each exam, use the dedicated OpenExamPrep guides linked throughout. If you want the single-cert career and ROI breakdown for AI-900 specifically — is it worth it, what it pays, what comes next — read Is Azure AI-900 Worth It? Career Value and Next Steps.
The 30-Second Decision Tree
Match your situation to the strongest first pick. These are starting points, not rules — read the per-cert sections before you book.
| Your situation | Take this first | Then |
|---|---|---|
| New to IT / career changer / unsure | AZ-900 | AI-900 or SC-900, then a role-based associate |
| Building toward Azure admin, dev, or architect (AZ-104, AZ-204, AZ-305) | AZ-900 | Skip straight to the role-based exam |
| Sales, marketing, procurement, or PM at a Microsoft partner | AZ-900 | AI-900 if you sell or scope AI |
| Product/analyst/consultant role centered on AI features | AI-900 | AZ-900 for platform context, then AI-102 if you go technical |
| Moving into security, identity, compliance, SOC, or GRC | SC-900 | AZ-900 for platform context, then SC-200/SC-300 |
| Already working in Azure day to day | Skip fundamentals | Go directly to the relevant associate cert |
| Student building a first résumé | AZ-900 | Add AI-900 or SC-900 as a cheap second signal |
The single most common mistake is treating these as a ranked competition. They are not substitutes. They cover different surfaces of the Microsoft cloud, and most serious candidates eventually hold two of the three. The question this post answers is which one is the highest-leverage first step for you, and what each one realistically does and does not unlock.
What Each Certification Actually Proves
The Microsoft Learn exam pages define exactly what each exam measures. Reading the scope side by side is the fastest way to see why these are not interchangeable.
AZ-900 — Azure Fundamentals (the platform)
The AZ-900 study guide (skills measured as of January 14, 2026) covers three domains: Describe cloud concepts (25–30%), Describe Azure architecture and services (35–40%), and Describe Azure management and governance (30–35%). In plain terms, AZ-900 proves you understand what the cloud is (IaaS/PaaS/SaaS, CapEx vs OpEx, the shared responsibility model), the core Azure building blocks (VMs, App Service, storage, virtual networks, Entra ID), and how Azure is paid for and governed (Cost Management, RBAC, Azure Policy, locks, Monitor). It is the only one of the three that gives you the whole platform map.
AI-900 — Azure AI Fundamentals (the AI layer)
The AI-900 exam (English version updated May 2, 2025) measures five areas: AI workloads and considerations (15–20%), machine learning fundamentals on Azure (15–20%), computer vision (15–20%), natural language processing (15–20%), and generative AI workloads (20–25%) (Exam AI-900, Microsoft Learn). It is narrow and deep on one slice — Azure AI Vision, Azure AI Language, Azure AI Speech, Azure Machine Learning, Azure OpenAI, and responsible AI principles. It assumes basic cloud literacy but does not teach the platform itself.
SC-900 — Security, Compliance, and Identity Fundamentals (the trust layer)
SC-900 (last updated November 7, 2025) measures four areas: concepts of security, compliance, and identity, capabilities of Microsoft Entra, capabilities of Microsoft security solutions, and capabilities of Microsoft compliance solutions (Microsoft Certified: SC-900, Microsoft Learn). It proves you understand identity (Entra ID, conditional access, zero trust), Microsoft Defender, Microsoft Sentinel, and Microsoft Purview at a conceptual level — the backbone of any security, identity, or compliance conversation in the Microsoft ecosystem.
The overlap is small and deliberate: AZ-900 touches identity and governance lightly, AI-900 touches responsible AI, and SC-900 goes deep on the security and identity stack. None of the three can substitute for another.
Difficulty And Cost Are Nearly Identical — Sequence Is The Real Variable
Because the format and price are the same, difficulty differences come from your background, not the exams themselves.
| Factor | AZ-900 | AI-900 | SC-900 |
|---|---|---|---|
| Level | Beginner | Beginner | Beginner |
| US price | ~$99 USD | ~$99 USD | ~$99 USD |
| Exam time | 45 min | 45 min | 45 min |
| Passing score | 700/1000 | 700/1000 | 700/1000 |
| Expires? | No | No | No |
| Prerequisites | None | None (basic cloud awareness helps) | None (Azure/M365 awareness helps) |
| Typical study time (beginner) | 20–40 hours | 20–40 hours | 15–30 hours |
| Hardest for | People who skip the governance domain | People who skip generative AI (20–25%) | People who confuse Entra/Defender/Purview |
Price is set by the country or region where the exam is proctored and excludes promotions and taxes; confirm the current figure with Microsoft before you register, since prices change without notice. The practical takeaway: none of these is meaningfully "harder" — the difficulty is whichever one tests material you have the least exposure to. That is exactly why sequence beats selection. Taking AZ-900 first makes AI-900 and SC-900 easier, because both lean on platform vocabulary AZ-900 teaches. The reverse is not true.
Why AZ-900 Is The Default First Pick
Three reasons AZ-900 is the safest first certification for most people in 2026:
It is the platform context the other two assume. AI-900 talks about deploying Azure AI services; SC-900 talks about securing Azure and Microsoft 365 resources. Both make far more sense after you understand subscriptions, resource groups, regions, and the shared responsibility model — the core of AZ-900. Microsoft itself frames Azure Fundamentals as "a common starting point in a journey towards a career in Azure" (Microsoft Certified: Azure Fundamentals).
It has the widest audience fit. AZ-900 was designed for non-technical roles too — sales, marketing, procurement, and executives at Microsoft partners — as well as IT beginners and students. If you are not yet sure which direction you are heading, AZ-900 keeps every door open.
It feeds the most role-based paths. AZ-900 is the conceptual on-ramp to AZ-104 (Administrator), AZ-204 (Developer), AZ-305 (Solutions Architect), and the data and security tracks. AI-900 mainly feeds AI-102; SC-900 mainly feeds SC-200/SC-300. AZ-900 gives you the most optionality per dollar.
The honest caveat: if you already work in Azure daily, none of the fundamentals exams is the best use of your money — skip to the role-based associate certification, which carries far more hiring weight (and unlike fundamentals, those do expire and renew annually for free).
When To Take AI-900 Or SC-900 First Instead
The default is not the rule. Pick AI-900 first when your target role is explicitly AI-centric — a product manager owning an AI feature, a consultant scoping AI projects, a business analyst evaluating Azure OpenAI, or a non-engineer who needs AI literacy fast and has no near-term infrastructure path. In that case the generative-AI content (the heaviest single domain at 20–25%) is directly relevant to your job today, and platform depth can wait.
Pick SC-900 first when you are entering or pivoting into security, identity, compliance, SOC analysis, or GRC. The Entra, Defender, Sentinel, and Purview vocabulary is what those roles actually use, and SC-900 is the recognized on-ramp to SC-200 (Security Operations Analyst) and SC-300 (Identity and Access Administrator). For a compliance analyst, SC-900 is more relevant on day one than a general Azure overview.
One nuance for AI-900 specifically: Microsoft has announced that the AI-900 exam will retire on June 30, 2026, and be replaced by AI-901 — both exams earn the same lifetime Microsoft Certified: Azure AI Fundamentals credential (Exam AI-900, Microsoft Learn). If you have already studied AI-900 material, sit it before the retirement date; otherwise prepare for AI-901, which shifts toward Microsoft Foundry, agents, and a light hands-on layer. This timing pressure is a reason to think about when, not a reason to skip the credential — it never expires once earned.
What These Certifications Do — And Don't — Unlock
Be realistic about the ceiling. Fundamentals certifications are literacy signals, not job-qualifying credentials.
What they do unlock: recruiter keyword matches, interview credibility for entry and non-engineering roles, a structured way to learn the vocabulary, a permanent line on your Microsoft transcript, and a confidence base for the role-based exams that actually move salary. For career changers, students, and non-technical staff at Microsoft partners, that signal is worth far more than the ~$99 cost.
What they do not unlock: a cloud engineer, AI engineer, or security engineer job on their own. No fundamentals exam asks you to configure a VM, write Python against Azure OpenAI, or build a Sentinel detection rule. Hiring for those roles keys off AZ-104/AZ-204/AZ-305, AI-102/DP-100, or SC-200/SC-300 plus hands-on evidence. Treat fundamentals as the on-ramp, not the destination — and pair the cert with a small portfolio project if you want it to convert to interviews.
The employment backdrop favors the stacking strategy: the U.S. Bureau of Labor Statistics projects employment of computer and information research scientists to grow 20% from 2024 to 2034, with a median annual wage of $140,910 in May 2024 (BLS Occupational Outlook Handbook). Those salaries attach to role-based skills; fundamentals get you onto the path, not to the number.
A 2026 Stacking Order That Actually Works
Most people should not stop at one. A practical, low-cost sequence:
- AZ-900 first — platform foundation, widest optionality, easiest second-exam ramp.
- AI-900 or SC-900 second — whichever matches your target role. AI-900 if you touch AI products; SC-900 if you touch security, identity, or compliance. Many candidates eventually do both because each is under $100 and never expires.
- One role-based associate — AZ-104, AI-102, or SC-200/SC-300. This is the certification that changes hiring outcomes and salary, and it builds directly on the fundamentals layer.
- Add a portfolio project — a small deployed Azure solution, an Azure OpenAI demo, or a documented security configuration. The cert plus the project is what converts to interviews.
The ROI logic is simple: under $300 total for three permanent fundamentals credentials buys you the vocabulary, the recruiter keywords, and the confidence to attempt the role-based exam that actually pays. Skipping AZ-900 to chase AI-900 or SC-900 first usually means relearning platform basics anyway.
Practice Beats Re-Reading For All Three
Whatever you take first, the highest-yield study method is the same: short concept review, then heavy timed practice until your reasoning is fast, not just your recall. These exams reward scenario discrimination — "which service fits this situation" — far more than memorized definitions. Drill until you can explain why each wrong answer is wrong in one sentence.
Start free, no signup, with full OpenExamPrep question banks for each exam:
- AZ-900: free practice questions and the AZ-900 study guide
- AI-900: free practice questions and the AI-900 study guide
- SC-900: free practice questions and the SC-900 study guide
For the full per-exam blueprint and study plan, use the dedicated OpenExamPrep exam guides for AZ-900, AI-900, and SC-900. For the deep career and ROI case on AI-900 by itself, see Is Azure AI-900 Worth It?.
Bottom Line
There is no universally "best" Microsoft fundamentals certification — there is a best first one for your situation. For the large majority, that is AZ-900: it builds the platform foundation the other two assume, fits the widest range of roles, and feeds the most role-based paths. Take AI-900 first only if your job is explicitly AI-centric, and SC-900 first only if you are heading into security, identity, or compliance. All three cost about the same, last forever, and are worth stacking over time. The expensive mistake is not picking the wrong one — it is stopping at fundamentals instead of using them as the on-ramp to a role-based certification that actually moves your career.
Official Sources Checked
- Microsoft Learn — Exam AI-900 (retirement to AI-901, skills measured): https://learn.microsoft.com/en-us/credentials/certifications/exams/ai-900/
- Microsoft Learn — Microsoft Certified: Azure Fundamentals (AZ-900): https://learn.microsoft.com/en-us/credentials/certifications/azure-fundamentals/
- Microsoft Learn — AZ-900 study guide (skills measured Jan 14, 2026): https://learn.microsoft.com/en-us/credentials/certifications/resources/study-guides/az-900
- Microsoft Learn — Microsoft Certified: SC-900: https://learn.microsoft.com/en-us/credentials/certifications/security-compliance-and-identity-fundamentals/
- Microsoft Learn — Credential expiration policy (fundamentals do not expire): https://learn.microsoft.com/en-us/credentials/support/credential-expiration-policy
- U.S. Bureau of Labor Statistics — Computer and Information Research Scientists: https://www.bls.gov/ooh/computer-and-information-technology/computer-and-information-research-scientists.htm
