Lot’s of assumption here. And having lived long enough to see “that’s definitely the best language ever” happens multiple times, I’m not too worried.

Until we get something different than LLM that is able to actually understand what’s happening and combine things in different ways, the only thing that might dwindle in the future is the cost of rewriting the same app every six months, since an LLM might (still lots of assumption) be able to regurgitate it. People writing new things will still be required for a long time. And these people will want new, shiny languages for all the same reasons we keep making new languages to this day.