But New Hampshire already has the plant built. That cost already happened. The question now is how best to meet the energy demand with the resources available, and that's already available.
New Hampshire is on the national grid. If they have more power than they can use they simply sell it to Massachusetts so they can turn down some of their natural gas generation.
Aren't all these things moot in a situation where the consideration is how much energy to produce from a plant that has already been built and is already operating?
New Hampshire's utility isn't considering whether to use bitcoin miners or whether to decommission the whole plant and stop using any nuclear energy. They're considering whether they can increase production to cover a higher amount than baseload by making sure they can burn off extra energy easily when demand drops.
Right, seems kind of silly to me as a utility guy. We have this nuclear plant that we don't want to use as baseload for some reason even though operating it is way cheaper than every other plant we have. Rather dump the electricity into mining bitcoin and expose the utility to those financial risks.
As far as I know, they do use it for baseload, but there's a very big difference between baseload and peak. In the summer in New Hampshire, for example, it's very common for people to run their air conditioners for a few hours during the hottest part of the day, but turn them off the rest of the time because it's not miserable. If you're fluctuating between peak and no heating/cooling need over the course of every 24 hours, that peak demand is a real problem, and demands a lot from the other energy plants.
If you could ramp the baseline production up a little bit and meet more of the peak demand from nuclear, it'd be cleaner than relying on the other plants, but you end up with too much energy produced, especially at night when it's cooler. Conveniently, cool temps are a benefit to bitcoin mining rigs, so it works out to spin up some ASICs overnight and make some money from the excess energy.
Nobody's suggesting the utility hold the bitcoin. The expectation would be that it would be sold whenever it's generated, so there isn't really any financial risk. But even if there weren't money to be made from it, having something that can easily be ramped up or down to compensate for energy demand is still useful.
Marginal versus average cost. Extremely low marginal cost (cost of the next unit to produce) but extraordinarily high average cost (includes capex). You typically make decisions for profit maximization on marginal cost.
You can safely ignore capex once the plant is built, though. Because nobody is going to give you the money back. And so, after the plant is built, the only relevant optimization parameter left is marginal cost.