In fairness, by some accounts (I can't personally attest to their accuracy) more than 50% of that is already from renewables, eg. https://www.nasdaq.com/articles/renewable-energy-sources-tha...
I do know from experience there are mining outfits that have located themselves where the energy was going to waste anyway.
But it doesn't excuse the huge use of power, and I wouldn't be surprised if over the long term a more efficient algorithm like proof-of-stake displaces proof-of-work. It's basic physics once you can do the same thing for less energy, and technologies more efficient than their predecessors tend to win out.
If by that time there's a popular proof-of-whatever coin with a similar focus, lean feature set and works-as-expected usability as Bitcoin I would definitely consider switching to it!
Edit: To be clear I share your admiration for Ethereum migrating to proof-of-work, my concerns and contrasts in the earlier post are more around protocol bloat and client software user-experience.