There's rarely a bullet point advantage that some new language or tech stack can offer me that would outweigh ten years of observation of how a familiar setup behaves in production, such that the space of unknown unknowns is reduced to almost nothing.

My personal rule is that the new technology stack item needs to either make is possible for me to build something that I couldn't have built without it, or needs to provide a productivity boost significant enough to overcome the productivity lost by straying from the more familiar path - even harder for team projects where multiple people need to learn the new component.