

China is currently building coal power plants to replace old, less efficient and dirtier plants. Then they use them for base load / network stabilization while they build up ridiculous amounts of energy storage. They’re not running all the time, only to compensate variability in renewables power generation while they build up storage. They’ll probably be taken out of use in a few decades without having much runtime.
For installed capacity the new power plants don’t change much as they take old ones offline - percentage wise coal is falling due to the ridiculous speed of building renewables.
I haven’t been TUI only for roughly two decades now - that’s around the time the server got properly usable.
I typically have a daemon running - on Linux via systemd user session, on MacOS via launchd, on Windows via startup. I then attach GUI frames to that, and - at least on the unix style platforms - typically have a tmux with a TUI frame running.
I have scripts to open files in emacs - that’s easy configurable to either open in the last used frame, or attach a TUI frame in the terminal, and open it there. I also have the EDITOR variable set to those scripts.
When fully using emacs you’d probably end up doing more work in emacs, and only occasionally wanting to do shell call outs.
For example, I edit some project, commit it via magit, trigger a build via compile to pack things up, open dired to move the files to a publish location, then open dired at the publish location and modify the publish package from there, and then finally start a shell in that directory to trigger the publish workflow. Only slight annoyance with that is that out of the box emacs shells are not setup for that kind of multiple shell buffers in specific directiories with easy throwaway - but it’s easy enough to make that work.