thinking about how computers have bad externalities in the same way desalination does: no one wants to vertically integrate renewables where it most matters because it costs more in the short term, even if its possible to negate a huge amount of the negatives simply by thinking ahead and working together, a thing megacorps find impossible
desalination has the problem where it takes a lot of energy to do, and also they dump the salt back into the sea, causing layers of death where fish can't live that float in the middle of the most swimmable zones for fish populations, acting like a highway splitting a neighborhood.
but that doesnt matter if you pair it with enough renewable energy in the grid that power doesn't matter, and a nearby refinery that cracks the excess highly concentrated brine into lithium, sodium, and chlorine (and small amounts of uranium). convert the chlorine into chloramine and send it to the drinking water treatment plant, pipeline the sodium to a nearby sodium-ion battery plant.
computer power use isn't so bad when you have enough renewables in the entire grid that they aren't "using renewables that creates a gap that needs to be covered with fossil fuel". Datacenters can pair with what are essentially geothermal plants for heat disposal instead of using giant open cycle evaporative cooling things where they just spray resevoir or aquefer drinking water over a grate with the intent that a chunk of it just goes into the atmosphere. etc.
but not even monopolies get to be that tightly integrated. Too much cost for parties with completely different goals and a general unwillingness to invest in things they don't understand
I can't claim to know everything, there's obviously spots here that are currently still unsolved problems.
but the thing about climate doom thinking is that a lot of solutions are out there, but giving up means there's not nearly enough social pressure to force the change
