This is my chart showing btc hashrate (in black, left side scale) and USD/KWh rev (different colors, right side scale) for common mining equipment:
https://i.imgur.com/XqyMJqw.jpeg
Same chart, but showing BTC/KWh rev:
https://i.imgur.com/2G7Rocq.jpeg

1. Its significantly more efficient to mine eth, in usd/kwh rev (data since 22th nov 2021):

min s19	max s19
0.2064791667	0.4282730586
min 3080	max 3080
0.5406816123	1.35386456

which shows between 2.6 - 3.2 times more efficient to mine with common gpu (nvidia 3080) than common asic (s19 pro).

2. The drop in rev is bigger for eth mining (data since 22th nov 2021):
Ratio in usd/kwh

min s19	max s19	% drop
0.2064791667	0.4282730586	51.78796272
min 3080	max 3080	% drop
0.5406816123	1.35386456	60.06383294

And possibly more important ratio in btc/kwh

min s19	max s19	% drop
0.000006233333333	0.0000075178571436	17.08630245
min 3080	max 3080	% drop
0.00001585144928	0.00002356702899	32.73887309

If we add up pros and cons we get that mining eth is cheaper gear, more profitable, less noisy, more reliable. While mining btc is more stable rev.

If I was a big player I would go eth all the way, but the problem we didn't mention is eth 2.0. I guess we can't know for sure when it will happen, but gpu mining rev is going down after this, no doubt. Yes, there are other coins to mine, but lets face it, none of it is even close to eth.

One scenario of eth 2.0 aftermath could be:
- gpu mining rev going 5-10x
- btc mining rev starting to tumble downwards much more than now, since now big players are going all in on btc
- near apocalyptic sell off of gpus

All this if prices stay relatively stable, or stable against each other. If eth 2.0 further delays for lets say a year, even the next halving will be near and comes in to play. Wild.