The issue is demand volatility: when you incur a large capital cost to build a generating unit, you need to set the price such that you cover operating expenses and recover the capital cost before the end-of-life of the unit.
If your customers are 100% predictable, there is room for squabbling about how much profit you get(and added complexity because the time value of money may change depending on conditions in other markets); but it is relatively simple to set a price that meets this goal.
If there is a nontrivial risk that a source of demand may arrive, require a new build-out, and then vanish relatively quickly; you'll lose most of your initial investment unless you set rates to recover that investment over a shorter timespan.
Consider the two (largely hypothetical, but convenient) limit cases: if you want to buy a new power plant, nobody will sell for less than the amount of money it costs to build it. If you are buying power from a plant with perfectly stable demand and an unlimited lifespan, your rate would closely approach the cost of production as the initial investment can easily be recovered.
In real life, obviously, no source of demand is 100% risk free; and utility customers are not asked to pay 100% of the price of the infrastructure up front; but different sorts of customers are more and less risky(both in that they, individually, will leave unexpectedly; and more importantly that they and everyone like them might experience a highly correlated change in demand and leave all at once without replacement).
For not terribly shocking reasons, this utility suspects that bitcoin miners are (a)risky and (b) likely to enter or exit the market in large groups, unpredictably. Depending on what the price of bitcoins does, miners can either demand as much electricity as you can deliver to them, or potentially shut down everything but the emergency lights in a matter of minutes to hours if mining becomes uneconomic.
It's not that they care what you use the electricity for, it's that they care how likely you are to be a predictable customer. It's like why getting a hotel room for a night is more expensive, per hour, than getting an equivalent apartment for a year: it's not that the sellers care what you are doing with the room; but they do care about the odds that they'll have a paying customer for it on any given day.