Updated October 29, 2026
Several of the AI certifications people recommended a year ago are gone. Here are the ones you can still take, what they cost, and which fits someone who doesn’t write code.
The short version: Several big AI certifications were retired in 2026, including Microsoft’s AI-102 and DP-100 and the AWS Machine Learning Specialty. If you don’t code, the best current picks are Microsoft’s AI Business Professional (AB-730, $99), Google Cloud’s Generative AI Leader ($99) and AWS Certified AI Practitioner ($100). If you do code, look at Microsoft AI-103, AWS Machine Learning Engineer Associate and Google’s Professional Machine Learning Engineer. Free course certificates exist too, and they’re a good way to find out whether you like the subject before you pay for an exam.
Updated October 2026: rewritten after Microsoft, AWS and Google changed their AI exams, with current prices, the retired exams to avoid, and free options.
Do AI certifications help you get hired?
They help a little, and AI skills help a lot. The best recent data looks at job ads that ask for AI skills, and those ads pay more:
By the numbers
Read those numbers carefully. They measure what employers advertise for skills, not what a certificate earns you. Lightcast also found that 51% of job ads asking for AI skills are outside IT and computer science, so this isn’t only for engineers. A certification is one way to show you have the skill. A small project you can talk through in an interview is another, and the two together work best.
If you’re starting from zero, learn the basics first with our guide to how to start learning AI, then come back and pick an exam.
Which AI certifications were retired in 2026?
Check this list before you buy a study guide, because older articles (including the first version of this one) still recommend exams you can no longer book:
- Microsoft AI-102 (Azure AI Engineer Associate): retired June 30, 2026. Azure is Microsoft’s cloud service. Replaced by AI-103, Azure AI Apps and Agents Developer Associate.
- Microsoft DP-100 (Azure Data Scientist Associate): retired June 1, 2026. The closest replacement is AI-300, Machine Learning Operations Engineer Associate, though it covers a different job.
- Microsoft AI-900: the exam was replaced by AI-901 on June 30, 2026. The certificate keeps its name, Azure AI Fundamentals.
- AWS Certified Machine Learning Specialty: the last exam day was March 31, 2026.
- TensorFlow Developer Certificate: Google closed the exam and hasn’t announced a replacement.
Already hold one of these? You keep it until it expires. AWS certifications stay valid for three years from the day you passed, and Microsoft also retired the renewal tests for AI-102 and DP-100, so those certificates will lapse when they expire.
Words you will meet
- Proctored exam
- A test watched by a person or webcam software, at a test center or at home. Most paid certifications use one.
- Fundamentals, Associate, Professional
- The usual levels, from beginner to expert. Fundamentals and Practitioner exams need no work experience.
- Exam code
- The short name, such as AB-730 or AIF-C01. Search with it, because certification names change more often than codes.
- Renewal
- What you do to keep it valid. Microsoft uses a free online test every year; AWS, Google and NVIDIA ask you to retake or recertify every two or three years.
Best AI certifications if you don’t code
These five are built for office workers, managers and career changers. None needs programming.
Microsoft Certified: AI Business Professional (AB-730)
AB-730 tests whether you can use Microsoft 365 Copilot and its agents (assistants that carry out multi-step tasks for you) to get work done in Word, Excel, Outlook and Teams. The exam is 45 minutes and $99. If your company runs on Microsoft 365, this is the one we’d start with, because you’ll practice on the tools you already use. Our Microsoft AI guide explains what Copilot does before you start.

Google Cloud Generative AI Leader
Google says its Generative AI Leader exam is “for anyone in any job role, with or without hands-on technical experience.” It covers what generative AI is, how to get better answers from it, and how a business decides where to use it. Budget 90 minutes and $99; a pass is good for three years.
AWS Certified AI Practitioner (AIF-C01)
The AI Practitioner exam checks that you understand the basics of AI, machine learning and generative AI, and the AWS services that use them. Expect 65 questions in 90 minutes, for $100, valid for three years. It suits people who work around cloud or tech teams without being engineers themselves.
Microsoft Certified: AI Transformation Leader (AB-731)
AB-731 is the manager’s version of AB-730: less about using Copilot yourself, more about planning how a team or company adopts AI. Same price and length as AB-730.
Google AI Essentials
Google AI Essentials is a course with a certificate at the end rather than an exam. It has five short courses on Coursera about using AI tools at work and writing good prompts. It is no longer free to audit: Coursera charges $49 a month after a 7-day free trial in the US, and financial aid is available. Finish inside the trial and it costs nothing.
Best AI certifications for technical roles
These expect you to write some code or work inside a cloud platform’s control panel. They’re the ones that show up in engineering job ads.
- Microsoft AI-103 (Azure AI Apps and Agents Developer Associate): building AI apps and agents on Microsoft Foundry, its platform for AI apps. $165, 120 minutes, free yearly renewal. This is where AI-102 candidates should go.
- Microsoft AI-300 (Machine Learning Operations Engineer Associate): running and monitoring machine learning and generative AI systems once they’re built. $165, 120 minutes.
- Microsoft AI-901 (Azure AI Fundamentals): a $99 starter exam. Unlike the old AI-900 it now expects a little Python, so it fits beginners who are learning to code.
- Google Cloud Professional Machine Learning Engineer: designing, building and running machine learning models on Google Cloud. $200, 2 hours, valid for two years. Hard; aimed at people who already do the job.
- AWS Certified Machine Learning Engineer Associate: the replacement for the retired Machine Learning Specialty. A new version (MLA-C02) is in beta (a discounted trial run of the exam) as of October 2026, at a $75 fee; the previous version was $150.
- AWS Certified Generative AI Developer Professional: building apps on top of AI models with AWS. $300, 75 questions in 180 minutes.
- NVIDIA Generative AI LLMs Associate (NCA-GENL): the basics of large language models on NVIDIA’s tools. $125, one hour, valid for two years. NVIDIA also offers an AI Infrastructure and Operations associate exam at the same price.
How do the main AI certifications compare?
| Certification | Best for | Exam fee | Coding? |
|---|---|---|---|
| AI Business Professional (AB-730) | Microsoft 365 users | $99 | No |
| Generative AI Leader | Any role, any company | $99 | No |
| AWS AI Practitioner | People who work with cloud teams | $100 | No |
| AI Transformation Leader (AB-731) | Managers leading AI rollout | $99 | No |
| Azure AI Fundamentals (AI-901) | Beginners learning to code | $99 | A little |
| Azure AI Apps and Agents Developer (AI-103) | Developers on Microsoft | $165 | Yes |
| Machine Learning Operations Engineer (AI-300) | Engineers running AI systems | $165 | Yes |
| AWS Machine Learning Engineer Associate | Machine learning engineers on AWS | $75 beta | Yes |
| Google Professional ML Engineer | Experienced machine learning engineers | $200 | Yes |
| AWS Generative AI Developer Professional | Developers building AI apps | $300 | Yes |
| NVIDIA Generative AI LLMs Associate | Developers new to large language models | $125 | Yes |
Exam fees in US dollars from each vendor’s certification page, October 29, 2026. Taxes may apply, and fees vary by country.
Can you get an AI certificate for free?
Yes, from courses rather than proctored exams. These prove you finished something, which is worth a line on LinkedIn, and they cost nothing:
- Claude Academy: Anthropic’s free courses on Claude and AI fluency, with free completion badges. Our Claude Academy guide lists every course. Anthropic’s paid Claude certification exams are open only to members of its partner network.
- Elements of AI: the University of Helsinki’s introduction to AI. The Introduction to AI certificate is free; the follow-on Building AI certificate costs €50.
- Hugging Face Agents Course: free, with a free certificate, for people who code and want to build AI agents.
- OpenAI Academy: ChatGPT-focused lessons and workshops; see our OpenAI Academy guide.
- Beginners in AI 14-Day Challenge: our own free challenge, one small AI task a day. It comes with no certificate, but it gives you real work to show.
For longer lists, see free AI courses with certificates and the best free AI courses.
Course certificates on Coursera and DataCamp
These sit between free badges and proctored exams. You pay a monthly subscription while you study, and you get a certificate from the company or university that made the course. Coursera dropped free auditing on most of these in 2026, so plan to use the free trial or pay for the months you need.
- Google AI Professional Certificate: eight courses on using AI for everyday work.
- IBM AI Engineering Professional Certificate: 13 courses, from machine learning with Python to deep learning and building with AI models. IBM estimates about four months at 10 hours a week.
- IBM AI Developer Professional Certificate: 10 courses for building simple AI apps. It replaced the older IBM Applied AI certificate.
- IBM RAG and Agentic AI Professional Certificate: 10 courses on connecting AI to your own documents and building agents.
- Vanderbilt Prompt Engineering Specialization: three courses on prompt engineering, no coding needed. Our free guide to writing AI prompts covers the basics first.
- Stanford AI in Healthcare Specialization: the best-known option for people in medicine and health care.
- DataCamp AI Fundamentals: about nine hours of short courses; the certificate comes with DataCamp Premium, from $14 a month billed yearly.
Are university AI programs worth the price?
They cost far more than vendor exams, and you’re paying for live teaching and a university name on the certificate:
- Stanford Artificial Intelligence Professional Program: $2,045 per course, and the certificate needs three courses, so about $6,135. Each course runs 10 weeks at 10 to 15 hours a week, and you apply to get in. Enrollment for the current round was closed when we checked on October 29, 2026.
- MIT Professional Education Applied AI and Data Science Program: $3,900 for about 14 to 15 weeks of live online classes with MIT faculty, run with the learning company Great Learning.
They make sense if your employer pays, or if you work in finance, consulting or another field where the university name carries weight. If you’re paying yourself, a $99 to $200 vendor exam plus a portfolio project will get you further per dollar. Our piece on whether AI courses are a waste of money goes deeper.
How to choose the right AI certification
Four questions narrow it down quickly:
- Which tools does your employer use? Microsoft 365 shop: AB-730. Google Workspace or Google Cloud: Generative AI Leader. AWS: AI Practitioner.
- Do you code? If not, stay with the no-code list above. Paying for a technical exam you aren’t ready for wastes the fee.
- Does it appear in job ads you want? Search LinkedIn Jobs for the exam code. If employers name it, it has value where you’re applying.
- Is it current? Look for a recent “last updated” date on the vendor’s page, and avoid anything on the retired list.
Some starting paths, by background:
- Complete beginner: get comfortable with everyday AI use, take a free course (Claude Academy or Elements of AI), then Generative AI Leader or AB-730.
- Office worker on Microsoft 365: AB-730, then AB-731 if you lead a team.
- Data analyst moving toward machine learning: AI-901 or AWS AI Practitioner, then AWS Machine Learning Engineer Associate or Microsoft AI-300.
- Software developer: AI-103 or AWS Generative AI Developer Professional, depending on your cloud.
- Career changer without a degree: see building an AI career without a degree for how certificates fit next to projects.
Common questions
What is the easiest AI certification for beginners?
Microsoft’s AB-730 and Google’s Generative AI Leader are two of the most beginner-friendly exams: both are $99, need no coding, and are designed for any job role. If you want something free first, try Claude Academy or Elements of AI.
Is AI-102 still worth taking?
You can’t take it anymore. Microsoft retired AI-102 on June 30, 2026 and replaced it with AI-103, which covers building AI apps and agents. Most of what you studied for AI-102 carries over.
Do AI certifications expire?
Most do. Microsoft’s associate certifications need a free online renewal test every year, while its fundamentals certifications don’t expire. AWS certifications last three years, Google’s Generative AI Leader three years and its Machine Learning Engineer two, and NVIDIA’s two.
Do I need a certification to get an AI job?
No. Employers are looking for AI skills, and a certification is only one way to show them. Projects you’ve built and can explain often count for more. A certification helps most at the first screening step, when a recruiter is scanning for keywords.
Which AI certification is best for health care?
For non-coders, Stanford’s AI in Healthcare Specialization on Coursera is the most relevant. For engineers, a cloud certification such as AWS Machine Learning Engineer Associate or Google’s Professional Machine Learning Engineer, plus experience with health data, is the usual route.
Made for regular people, by a real person
A free daily newsletter with one useful AI lesson at a time, including when a certification you’re eyeing changes or gets retired. Curated and edited by a human, in plain English.
Free forever. Unsubscribe anytime.
Sources
- Microsoft Partner Center: June 2026 announcements (AI-102 and DP-100 retirements)
- Microsoft Learn: AI Business Professional (AB-730)
- Microsoft Learn: AI-103
- Microsoft Learn: Exam AI-901
- Microsoft Learn: renew your certification
- Google Cloud: Generative AI Leader
- Google Cloud: Professional Machine Learning Engineer
- AWS: Certified AI Practitioner
- AWS: Machine Learning Specialty (retired)
- AWS: Machine Learning Engineer Associate
- NVIDIA: certification exams
- TensorFlow: certificate program
- Stanford Online: AI Professional Program
- MIT Professional Education: Applied AI and Data Science Program
- PwC: 2026 Global AI Jobs Barometer
- Lightcast: Beyond the Buzz (July 2025)
- Microsoft: 2025 Work Trend Index

