> But if it does that's a good thing. That means we'll need to expand our grid.

Your other points are generally very well-thought-out, but this one is not.

Grid expansions take years. Building new nuclear power plants takes decades. And there's "slack" in the sense that utility prices can raise substantially before there's actually impetus to start planning a grid expansion, and then you still have to wait years for construction to finish.

And the party that takes the hit for those sustained elevated electricity prices is local residents, who in general have far less capital and are disproportionally hurt by it relative to the owns of the data centers.

You can make the argument that the electricity price increases are going to be lower than most people think and wouldn't meaningfully hurt the average local resident, but that's a separate argument. I don't buy that "if it pulls power from the grid that's a good thing" is valid.

If there isn't enough power to meet demand, then build out the grid first, or tax datacenter power consumption so that the inflection pain-point before grid capacity expands is less harmful to the residents.