If bitcoin dies, I imagine all that datacenter capacity can easily be reapplied to future AI and other needs though.

Are you familiar with what happens to prices when supply floods the market? :-)

I doubt Iceland can "flood" the DC market.

In this particular case, Iceland is concerned about their domestic DC market supply flooding local demand if the BTC mining dries up.

There are still good reasons to host in Iceland — geothermal power, no or less need to run expensive AC units, maybe low enough latency to European sites — but 90% of their current consumption is BTC miners.

Right, but I don't see why local demand is so important, a DC can sell its services globally.

> globally

Yes and no :-). Ignoring other factors, a DC's services are on average less valuable the higher the latency. That restricts Iceland's primary market to Europe (and maybe east-coast North America).

Bitcoin mining centers aren't really setup for hosting, they don't need the bandwidth.

Certainly, not hosting but plenty of applications that can tolerate physical moving of data in an out.(rendering, protein folding,some big data stuff (store the petabyte locally and have people come visit) All ASIC is going to be a problem though.

They do need a stable internet connection though that can't be super slow, so I'd assume it could always be upgraded with fiber. The stable and plentiful power on it's own will be very interesting.

They're all Asics, they are only useful for computing hashes.

I don't think this really matters, it's not the computers that would be reused but the infrastructure. These are essentially buildings with good ventilation and plenty of power and an internet connection that can likely be upgraded too. That's already quite useful

Ventilated buildings with power, but no networking gear or servers can surely be converted. But it's hardly a slam dunk. A lot of other buildings could be converted to that level too. So I think the other factors in choosing a data center location may dominate the decision.