What's the use case here? I assume a render with entirely different materials than will be present in the actual building is near useless for architects, but maybe I'm missing something?
It's the usual story with AI image generation where it's easy to get something vaguely presentable with a loose set of vibes, but the more specific you want the details to be the more of an uphill struggle it becomes to get what you want. And in this case the user probably already has a 3D model they can feed into a conventional renderer with the exact parameters they want.
Another issue is that the user will probably want to produce multiple renders of the same building, but with AI it's extremely difficult to get consistent results. Again, not an issue with conventional approaches to rendering.