If something comes along significantly and obviously better-suited than COBOL in its domain, with a migration path, we might see COBOL code gradually replaced. Peter Thiel said “ If we are truly talking about an innovation, then the product will not be an incremental improvement. For it to be an innovation it must be 10 times better than what is currently available.“ Steve Jobs said something similar.
Having worked on big systems in COBOL, I can say it’s good at what it does, and the tools are better than you might think. The tendency to view anything old or out of sight as obsolete junk bedevils the tech industry.