I think you are focusing on the wrong part of their argument. Whether or not a user has an iphone is pretty much irrelevant. The point is what they prioritize above or below interoperability.
For zero downside, everyone prefers more interoperability over less interoperability. But if you ask people to actually give something up for interoperability (usually by requiring them to pay for it, but sometimes it is less obvious) then you rapidly discover that most people don't care very much. The same thing is true for managers and "performance". Everyone will say they want more performance, but almost nobody is willing to sacrifice even a single tracking library for it.
Not only is what you say true, but interoperability isn't even a nice linear spectrum to compare across. For instance, I own a PinePhone, which can run mainline GNU/Linux rather than Android. That ranks highly for 'interoperability': almost complete freedom to change any part of the software stack. Yet for the same reason, it has less interoperability with the very proprietary apps that are necessary to access banks, tickets and various communication services online - these deliberately or incidentally block you from accessing their services from such a 'free' device.
I don't regret my purchase, as it's still an affordable, capable handset, but I have had to compromise and get an Android device as well. Interoperability can't exist in a vacuum and still be desirable. The EU's DMA should hopefully improve interoperability in addition to the existing conveniences of the Apple/Google duopoly - that's the exciting thing about it.
For zero downside, everyone prefers more interoperability over less interoperability. But if you ask people to actually give something up for interoperability (usually by requiring them to pay for it, but sometimes it is less obvious) then you rapidly discover that most people don't care very much. The same thing is true for managers and "performance". Everyone will say they want more performance, but almost nobody is willing to sacrifice even a single tracking library for it.