Tbh I think the "but data centers are important infrastructure, not just AI" talking point misses that like
Ok so roads are important infrastructure. A lot of stuff that's important happens on roads. Now, let's imagine that quadrillionaire Matt Stench has decided that the next big tech innovation is the Wide Car. It's a car that takes up six lanes despite seating only one passenger.
The Wide Car is supposed to be the future, and everyone's going to be driving Wide Cars, even though nobody who makes Wide Cars is turning a profit. Employers are offering Wide Cars as an employee benefit, and getting "nah." Some employers are going as far as demanding their employees drive Wide Cars, and the result is that people take time out of their workdays to get in the mandatory gas usage for their Wide Car before driving home in a regular car.
In spite of the fact that the Wide Car is clearly set to fail, there's an enormous push to expand to twelve-lane roads to accommodate a bunch of Wide Cars that simply will not materialize. This is not an organic response to demand, but a speculative investment that amplifies the existing issues with road development for no good reason.
That is the problem.
Oh and the road infrastructure project is buying up resources other people could have used for literally anything else. With money they promise they'll be making from Wide Car sales any day now.
Okay so what I'm getting from the notes is that when you try to transplant some techbro nonsense into an offline equivalent, you have to be careful to avoid simply inventing something the Americans are already doing in real life
