current AI is already stressing the power grid, and much of it will need to be redeveloped and improved just to keep pushing the limits of LLM’s. Power is the limiting factor with scaling here, so i’m rather unconvinced with your hypothesis. The improvements in the last 2-3 years are in no way indicative of the next 2-3 years.

I agree with your sentiment by the way, developers should find ways to use LLM’s to improve their development process. But the drama is getting old.