CLF-C02 vs AIF-C01: which AWS foundational exam should you take?
Both are foundational AWS exams, both assume no hands-on experience, and neither has prerequisites — so this is a question of subject, not level. CLF-C02 covers the AWS cloud broadly: core services, pricing, security and the shared responsibility model. AIF-C01 covers artificial intelligence on AWS: machine-learning and generative-AI concepts, the AI services, and using them responsibly. If you want general cloud literacy, take CLF-C02. If you are specifically heading into AI work, AIF-C01 is a legitimate first AWS certification in its own right.
The two exams at a glance
| CLF-C02AWS Certified Cloud Practitioner | AIF-C01AWS Certified AI Practitioner | |
|---|---|---|
| Vendor | AWS | AWS |
| Level | Foundational | Foundational |
| Questions | 65 | 65 |
| Time limit | 90 minutes | 90 minutes |
| Passing score | 700 / 1000 | 700 / 1000 |
| Exam fee | $100 | $100 |
| Valid for | 3 years | 3 years |
| Exam domains | 4 | 5 |
| On SaveMyCert | Fully live | Fully live |
Choose CLF-C02 if…
- You want a general grounding in AWS before choosing a direction.
- Your role touches the cloud broadly — engineering, operations, product, finance or management.
- You plan to move on to an associate exam such as Solutions Architect, Developer or CloudOps.
- You want the AWS certification most widely recognised as a starting point.
Choose AIF-C01 if…
- Your work is already pointing at AI or machine learning, and you want a credential that says so.
- You need the vocabulary to hold a serious conversation about generative AI, models and responsible use.
- You are in a product, strategy or leadership role where AI decisions land on your desk.
- You want a foundational certification that is newer and less saturated than the general one.
They overlap less than you would expect
AIF-C01 is not "Cloud Practitioner with AI added". It spends its time on machine-learning fundamentals, how generative AI works in practical terms, prompt techniques, model selection, and the governance questions that come with deploying AI — bias, transparency, safety and data handling.
CLF-C02 barely touches any of that. It is about the platform: compute, storage, networking, identity, cost and support. Someone who passes one will find perhaps a fifth of the other familiar, mostly the security and shared-responsibility material.
Can AIF-C01 really be your first AWS certification?
Yes. It is a foundational-level exam with no prerequisites and no assumption of hands-on experience, exactly like Cloud Practitioner. If AI is genuinely your direction, starting there is not a shortcut or a gamble.
The caveat is that it gives you narrower platform grounding. AI systems still run on compute, storage and networks, still cost money, and still need identity and access control — and AIF-C01 does not teach you those. Many people end up taking both, in which case the order barely matters.
Which is the better signal to an employer?
It depends entirely on the job. For a general cloud or engineering role, CLF-C02 is the more legible credential and the more common starting point. For a role where AI is the actual subject — product management for an AI feature, a strategy or governance role, an analyst moving toward ML — AIF-C01 says something specific that a general cloud certification does not.
What neither does is substitute for demonstrated work. A foundational certification opens a conversation; what you have actually built is what carries it.
A note on AI content and how fast it moves
AI moves faster than any other area these exams cover, and AWS revises the AI exam guide more often as a result. Anything you study about specific model names, versions or benchmarks is likely to be stale within months, which is why good preparation concentrates on the concepts — what a foundation model is, what retrieval-augmented generation solves, where bias enters, how you evaluate output — rather than the current catalogue.
That is also why we re-fetch the exam guide from AWS immediately before building or updating this certification’s content rather than working from a cached copy.