If you take all these factors into account, Bitcoin may actually be less energy-intensive than traditional
fiat currencies.
I would agree with this statement if the following two statements were also true:
1) Bitcoin achieved a comparable
transaction throughput than at least one major fiat currency
2) Bitcoin's mining costs would be, at maximum,
one order of magnitude higher than today, even if we have mass adoption (meaning: if BTC today is using the same electricity amount than Denmark, then in it's final state it should not use more than "ten Denmarks" or one country like Germany).
Basically, my point is: if Bitcoin achieves a sufficient scalability to be usable by a majority of the world's population - very probably with techniques like LN or sidechains - then your assumption is true. But for a maximum of ~7 tx/second, the Bitcoin ecosystem
is using a very high amount of energy.
So for a final conclusion we have to wait if LN works as expected and if not, if there are other techniques to boost transaction throughput.
Besides, many mining companies have set up their operations in countries with an abundance
of renewable energies like Canada or Iceland.
This could also have side-effects in the future: if the mining industry continues to grow, its demand could have impact in the price of renewable energy. Now the effect is negligible, but I predict if it continues to grow, we would have a similar effect on solar panels and wind turbines that now occurs with graphic cards.