I don't think that's a very good bet. There is still a lot of COBOL sticking around, but there are essentially zero new COBOL projects being started while there are a non-zero number of COBOL projects being replaced with other languages, so COBOL use is ultimately trending downward. You are essentially betting that COBOL developers die out faster than COBOL itself does.