Even starting right now, it would probably take longer than 5 years to release Windows or MacOS for open source. Is there any argument for why this would happen at all?

I believe the argument is something like "LLMs will be trained on open source code, LLMs will work better when their output is open source code, and using an LLM will massively improve productivity", but I'm not really sure to be honest. I certainly think that it sounds unlikely.

I see, thank you. This makes sense if you believe that open source code is different from closed source in a way that's captured by an LLM, but I don't see it very likely.

I think one argument is as follows: LLM code generation reduces the cost of producing code (thus, it's less valuable to keep it private), but not of checking it for faults. So releasing open source to get more eyes might make more sense. All in all, I think it's reasonable to expect a few percent more of open source releases.

I remember feeling this way in the 1990s when it seemed like Linux was getting better and Windows was getting worse

I'll take a relatively loose view of extinct - not that every single piece of software needs to be open source (will people only use new software?), but that basically all new software development is open source. In addition, a number of notable large projects would need to be open sourced, such as Windows and MacOS, Google Search, etc.

I'll also ignore software written for the military, and various other groups where the code is a trade secret (such as hedge funds), since I think it's extremely unlikely that they would open source their code, even for a 100x speed up in development time.

Upper bounded by Windows going open-source or extinct.