Browse Microsoft Certification Guides

Azure AI-900 Foundry and OpenAI Guide

Study Azure AI-900 Foundry and OpenAI: key concepts, common traps, and exam decision cues.

This lesson is mostly about platform boundaries. Microsoft’s certification page now says Azure AI Foundry is now Microsoft Foundry, but the live AI-900 study guide still tests the older wording. At fundamentals depth, you should know what the broader platform surface is for, what Azure OpenAI Service provides, and what the model catalog concept means.

Keep the platform pieces straight

Surface AI-900 takeaway
Azure AI Foundry broader Azure generative-AI platform surface for building and working with generative-AI solutions
Azure OpenAI Service Azure access to large language model capabilities
Azure AI Foundry model catalog a place to compare and work with model options at a high level

A safe mental model

Think of the relationship like this:

  • Azure OpenAI Service gives you model capability on Azure
  • Azure AI Foundry is the broader solution-building platform surface
  • the model catalog helps you understand and choose available models

AI-900 does not expect deep deployment detail, quota detail, or engineering implementation steps.

What the rename changes for you

The rename matters because Microsoft is updating the branding while the exam materials catch up. The safe exam posture is:

  • recognize that Azure AI Foundry and Microsoft Foundry refer to the same current platform transition
  • answer using the wording and scope of the current AI-900 study guide
  • do not invent new objective boundaries just because the product name changed

Common traps

Trap Better rule
treating the model catalog as the same thing as a model itself the catalog is the model-selection surface, not the model
assuming Azure OpenAI Service replaces every other Azure AI service it is one generative-AI service family, not the answer to every AI question
thinking the Foundry rename means the exam no longer follows the current study guide the study guide still defines the tested scope

What strong answers usually do

  • keep service, platform, and catalog roles separate
  • recognize the naming transition without overreacting to it
  • stay at the fundamentals level and avoid implementation-detail distractors
  • remember that generative-AI platform questions still sit inside the broader workload and responsible-AI logic of the exam

Decision order that usually wins

  1. Separate large language model capability, model-option comparison, and branding or platform naming changes.
  2. If the scenario is about hosted LLM capability on Azure, think Azure OpenAI Service.
  3. If the scenario is about comparing or choosing available model options, think model catalog at a high level.
  4. Treat the Foundry rename as a branding transition, not as a new exam blueprint by itself.
  5. AI-900 still tests service-family recognition at fundamentals depth, not deep engineering workflows.

Quiz

Loading quiz…
Revised on Sunday, May 10, 2026