But that's not the real cost. The real cost is that your new development has to be in Python 2.7. You don't get any of the new features, but more importantly yo

But that's not the real cost. The real cost is that your new development has to be in Python 2.7. You don't get any of the new features, but more importantly you can't use any new libraries that are Python 3 only. You lose a huge amount of the value of Python if you can't participate in the ecosystem. You're effectively left with 35M lines of Pascal or COBOL - on the one hand, a perfectly adequate and workable language for running a company like JPMorgan, it's been done before, and one which any skilled programmer ought to be able to pick up and work with for at least incremental changes. On the other hand, your new software development won't be in this codebase at all - developers won't want to use a dead language. They'll interoperate with the legacy code by treating it like legacy code. All your incremental changes will just be building APIs to call it from some other language (perhaps Python 3, perhaps something else entirely), which will work, it's just probably not what you expected to have when you wrote those 35M lines of Python.