So long as the COBOL systems function, can be updated at moderate/reasonable costs, and developers/maintainers can be found, COBOL will continue to thrive. There’s just too much vested in those systems to risk making a multi-billion-dollar transition to a new system which does exactly the same thing.
One fact of programming: legacy systems embody a great deal of wisdom which is not otherwise documented or known. Wholesale replacement of a system is a tremendous risk precisely because those creating the new system don’t know what vital yet obscure processes are performed by the old. Such replacement rarely happens unless there is an imperative to do so, such as outright parts obsolescence or skyrocketing cost of maintenance. In the case of big-iron legacy systems, it’s cheaper to teach someone an old language.
Kinda like Latin: allegedly dead, yet there is still great value in learning and using it in narrow yet big-budget “systems”.
Live, yes. Thrive, no.
The world is OOP now for lots of good reasons.