Given that current "AI" is a Large Language Model, and most of the tech writing (which LLMs are trained on) is about the current tech stack, I don't believe you can credibly do the projection where you say "tech stack is misaligned with AI". Almost by definition, it is not.