I think this is kind of a fallacy. While it's true in short term, in long term more energy sources can be always built, because there's almost unlimited amount of renewable energy available. Demand for electricity increases new investments, expands capacity, and drives costs down. It's not a zero-sum game in the long run.
Also, like I said in the other comment, most of grid energy is already too expensive for Bitcoin, so mining really makes sense just for any extra capacity which wouldn't have a buyer anyway.
> Also, like I said in the other comment, most of grid energy is already too expensive for Bitcoin, so mining really makes sense just for any extra capacity which wouldn't have a buyer anyway.
That's only partially true. And your first paragraph explains exactly why.
Compare: most cows are raised for meat, and the leather is just a by-product. However, the extra income from the leather makes raising cows a bit more profitable, thus giving us more cows on the margin.
Similarly, bitcoin mining soaking up excess capacity makes electricity generation slightly more profitable.
Eg instead of running a coal baseload plant and a natural gas peaker, you could run two coal baseload plants and outside of peak periods, you mint bitcoins. (Numbers are just for illustration. You get the point.)
About your first paragraph: yes, supply is elastic in the long run. But opportunity costs still need to be paid. Yes, in the long run you might not be trading off one Joule for another Joule, but you are trading off uses for capital.
(Also keep in mind that even with elastic supply, we still have decreasing marginal returns. There's only so many good sites for hydro-electric generators; all the windiest spots will be full of wind turbines at some point, etc.)
