It’s because most companies already have a lot of confidence with Microsoft contracts, and are generally very comfortable storing and processing highly sensitive data on Microsoft’s SaaS platforms. It’s a significant advantage.
Also Microsoft Azure hosts its own OpenAI models. It isn’t a proxy for OpenAI.
MS salespeople presumably already have weekly or monthly meetings with all the people with check-cutting authority, and OpenAI doesn't. They're already an approved vendor, and what's more the Azure bill is already really really big, so a few more AI charges barely register.
It's the same reason you would use RDS at an AWS shop, even if you really like CloudSQL better.
This is the main reason the big cloud vendors are so well-positioned to suck up basically any surplus from any industry even vaguely shaped like a b2b SaaS.
Lots of AI things are features masquerading as products. Microsoft already has the products, so they just have to add the AI features. Customers can either start using a new and incomplete product just for one new feature, or they can stick with the mature Microsoft suite of products they're already using and get that same feature.
As others have kind of pointed out, using "outside our DC" processing for corporate data is a non-starter for many companies.
These companies are left to choose between self-hosting models, or a vendor like MS who will rent them "their own AI running in their own Azure subscription", cut off from the outside world.
As someone in an AI company right now - Almost every company we work with is using Azure wrapped OpenAI. We're not sure why, but that is the case.