Microsoft is one of the world's largest technology companies. In the AI universe, it shows up on three fronts: cloud, end user products and tools for developers.

How it works

The AI business at Microsoft rests on complementary layers:

  • Cloud: Microsoft Azure offers servers, GPUs and services to train and run models. Companies access models via API inside Azure.
  • Products: Copilot appears in everyday tools, such as the Office suite and Windows, to summarize, draft and analyze.
  • Development: GitHub, which belongs to Microsoft, offers GitHub Copilot and hosts millions of repositories. The Visual Studio Code editor is also from the company.

The company also maintains a close relationship with OpenAI, and works with other model providers as well. Contract and agreement details change over time, so always check the company's official announcements.

Practical example

A small agency stores its projects on GitHub, edits code in VS Code and uses GitHub Copilot to suggest snippets. For a larger client, it publishes a customer service assistant using a model accessed via Azure, with data inside the client's environment.

In this scenario, almost the entire chain goes through Microsoft products: the repository, the editor, the coding assistant and the model hosting.

To call a cloud service, configuration is usually kept in environment variables:

Terminal
export AZURE_ENDPOINT="https://seu-recurso.exemplo.azure.com"

The address above is illustrative. Use what your dashboard shows.

What it's for

For developers, Microsoft matters because it defines part of the work environment: editor, version control and cloud. Many code agent tools work as VS Code extensions or integrate with GitHub.

For businesses, it serves as a corporate AI provider, with a focus on security, access control and compliance, items that large companies require.

For those following the market, it is one of the major forces in the AI industry, alongside model companies and chip companies.

Caveats

Products and names change frequently. Copilot, for example, names different products in different contexts: the GitHub one, the Windows one and the Office one are not the same thing. Check the official documentation before deciding.

Also evaluate vendor lock in. Using an AI gateway or open tools makes it easier to switch providers if something changes.