Microsoft's Q2 FY2024 earnings showed Copilot revenue as a growing contributor across commercial segments. The Azure OpenAI Service grew 3x year over year. The enterprise AI stack that Microsoft assembled is now operating at meaningful scale.
The Copilot surface area
Microsoft has Copilot in Windows, Copilot in Edge, GitHub Copilot, Microsoft 365 Copilot, Copilot in Dynamics 365, Copilot Studio for building custom copilots, and Copilot for Security. These are not all the same product. They share a brand and some underlying Azure OpenAI infrastructure but serve different workflows. Microsoft 365 Copilot works with your organisation's data in Microsoft Graph. GitHub Copilot works with your code. Copilot Studio lets you build a custom agent that works with your own data sources via connectors.
Copilot Studio and the enterprise integration layer
Copilot Studio is the product that enterprises are using to build internal AI tools without writing code. You configure data connectors, define topics and intents, and publish a Copilot that can answer questions from your SharePoint content, query your Dynamics data, or create tickets in ServiceNow based on a conversation. The governance layer, Microsoft's data residency and compliance controls, is what makes this viable for organisations in regulated industries.
The Azure OpenAI consumption signal
Azure OpenAI Service, the managed API for OpenAI models that runs on Microsoft's cloud, tripling year over year is the more significant metric. It means enterprises are not just exploring: they are building production applications that call GPT-4 at scale via Azure. Every GPT-4 call in an enterprise application that needs to stay within Microsoft's compliance boundary goes through Azure OpenAI. That lock-in is significant and intentional.
The integration moat
Microsoft's AI strategy is not about having the best model. It is about distribution. Every Microsoft 365 user is a potential Copilot user. Every Azure customer is a potential Azure OpenAI customer. Every GitHub user is a potential GitHub Copilot user. The model runs on OpenAI infrastructure accessed through Microsoft's commercial agreement. The moat is the relationship with the enterprise, the data that lives in Microsoft 365, and the trust in Microsoft's compliance and data governance. That is a structural advantage that is very hard to replicate.