Advanced bayesian estimations show that the risks of a nuclear plant that is not yet operational are very low. And the chance that they will still be employed at microsoft (after the bubble pops) by 2028 is exceedingly low, reducing effective risk significantly !
I was thinking of the economics as opposed to the safety aspects. Seems an expensive option.
Nuclear power has fairly predictable amortized returns. I imagine that this is worth the cost to MS over the next two decades or so; we have no idea what their current energy premium is like, and this plant doesn’t have to be as cheap as a new plant, just cheaper than the current premium.
If it was cheaper than the current premium, I expect that the plant would still be in operation, however as I don’t know the numbers so it must be worthwhile.