... Could you offer TL;DR of how you've arrived at 250MW? Not doubting you, just wondering what that number is meant to represent.
The network is at 500 petahash and the machines are pulling an average of 0.5w/gh (Ant s3 0.7w/gh, SP-tech 0.6w/gh, ant s5 0.5w/gh and newer gear 0.25w/gh).
You might be missing some details:
1. It's safe to assume that any [already purchased] gear will be kept online as long as it mines more coin than it costs to run. And 'costs to run' = electricity + cheap warehouse space.
A very large portion of the miners today are old Ant S3's. But a lot of mines will simply replace miners rather than upgrade capacity when things are as marginal as they are now.
What is the logic behind upgrading (investing in new gear) 'when things are as marginal as they are now'? Wouldn't it make more sense to simply keep the old gear running, with no additional investment, while it's making money?
If you have paid for a 500KW facility with with shelves for 500 machines, 500 psu's, 500 power outlets, Network with 500 lan cables, 10 power distribution units. 5 industrial grade fans, and so on, It makes a lot of sense to migrate to a more power efficient miner and turn your 500TH facility to a 1.5PH facility.
The machines will become obsolete anyway. The facility won't. And with margins the way they are, a lot of miners will try to optimize the capacity they already have rather than build/buy a new facility.
Some will sell the old gear on ...
Unless the old gear is sold as paperweights/door stops, it will hash until it either dies or becomes unprofitable to run. Who cares if it's in a factory farm or has been sold to a hobby farmer?
If you have 50 or 500 used machines you might try selling them on e-bay or through BCT. But if you have 10000 very used nearly obsolete machines, it might make more sense just stripping them for parts and chucking them in the bin.
Yeah, Chinese chicken coops are less exotic than actual data centers, but still...
But certainly not 3.7%, or else...
There seems to be a trend now that involves not putting bitcoin mines in the tropic.
But even in warmer areas cruder solutions lead to massive power savings.
Here you see wet paper on the right and fans pulling air out on the left. Thus pulling humidified air through the wet cool paper.
Paper being kept wet with some pcp piping.
... the dutch and belgian mining operations with 20 cents per kw/h went first. While areas with 2 cent kw/h are booming. It's a volatile market.
There really were factory farms running on $.20 power?
In late 2013 you would earn $30 per kw/h with an early KnC asic. Some people didn't plan ahead.