Rebuilding what it’s doing though wouldn’t take 30 years, they’ve figured out the requirements
Heh, I don't know about that... AD is the way that it is because nobody has actually figured out what the operational requirements of any given organization are. Certainly not Microsoft, not even after 2.5 decades of trying.
which means a startup can start fresh and build something even cleaner that works full on premise but seamlessly leverages cloud services
Do you work in marketing? did you get that phrase out of a brochure?
There's nothing seamless about anything cloud. Start taking apart any cloud platform and it will spill a bunch of ugly spaghetti and Swiss cheese.
There are no green fields, especially when you're talking about replacing an existing, operationally critical piece of technology. This is what you get:
in the event that your on premise systems have a failure.
If the on-prem systems have a failure there won't be any outbound network connectivity, there won't be anything capable of connecting externally.
The value prop is high for companies, if they could save a huge amount of money on licensing and get improved operational cost
That "if" is load-bearing. The "potential savings" are imaginary, theoretical at best. It doesn't really make sense to talk about "switching costs" when there's nothing to switch to. It's a big pile of "what-ifs" and "could-bes".