That Would Be Telling@p @collappsar @hj @mangeurdenuage This discussion is too long for me to read it all, but this point you keep hammering about Microsoft spending too much money for a total market value is both correct and there's a historical Microsoft answer to it:
(Help) make the market a lot larger.
Historical: like many others including myself, Gates saw the January 1975 Popular Electronics issue with the Altair 8800 and got very interested (he'd already been doing Intel 8008 stuff), but feared it was too late to get into the microcomputer software market. Microsoft has dreamed big from the day it became an idea.
From those humble beginnings microcomputers became beyond huge and half a century and two CEOs later Microsoft still makes a great deal of money supplying non-"AI" software for it. See intermediate ambitions, I remember hearing from a Microsoft guy as they were rolling out NT for developers of an "information at your fingertips" ambition, which did indeed happen.
A chicken in every pot PC in every home did happen to an extent, but while they started early, they did fail hard with smartphones (too much of that due to beyond terrible Ballmer management, see for example the Kin, which was bought from a company who's founders included the further Android founder (((Andy Rubin))), but his Danger Inc. basically got it right from the beginning).
What's the real upside for LLM AI? If really big, can Microsoft afford to fumble it like they did smartphones? How much can a wealthy company afford to spend even if the return is meager to zero on the off chance it will be astronomical?
For your specific, what domain does Microsoft understand better than developing code? That can go at least two different complementary ways, the experience gained can be more generally applied outside this domain, and maybe a lot more software can be developed using it, making the market bigger than you think it is.
All GNU social JP content and data are available under the