All it takes is some dev shop to say “give me 5,000 seats”

Your example proves otherwise:

5,000 seats x $200 price = $1,000,000 cash realized, right now, no debt financing needed for funding development, regardless if said dev shop leaves you.

5,000 seats x $4 = $20,000 month. That kinda pays the salary of a single FTE, and development ceases if customer runs away, only safe option would be to finance further development to insulate yourself from churn.

	▲	reactordev 3 hours ago \| parent [-]
		Bro, $200!?! For a SQL browser? Nah… $20 x 5,000 seats = $100,000 realized right now. With that, he could invest it, finish the features, start work on 2.0 and charge and upgrade of $20 to all his customers when he has MSSQL, MySQL, pgsql, SQLite, Oracle. 2.0 could include cloud db’s. Not everything has to be a subscription and if people continue to find value, they’ll upgrade if they think it worthy. Subscriptions tend to be vampires on the wallet and are used to trap users into paying a developer indefinitely. I consider it hostile to users. The exception is if it’s a service that can’t run on my machine. Or a service that does back office for me.