The AI explosion means millions are paying more for electricity

The Washington Post ^ | 6:00 a.m. EDT July 27, 2025 | Peter Whoriskey

[TheDon]

“Supposedly they are tapping a deeper aquafer than we use. Hopefully our wells won’t be affected.”

Didn’t realize that was aquifers were being used for cooling. Definitely compounds the problem of increasing electricity capacity!

[TexasGator]

“They are reopening Iowa’s only nuke plant (and doubling its capacity) to power 5 new data centers in Cedar Rapids.”

Not yet approved.

“Supposedly they are tapping a deeper aquafer than we use. Hopefully our wells won’t be affected.”

Supposedly? Plant is cooled by river.

[oldtech]

I doubt id AI could be powered completely by “green energy” but if it is going to affect commercial electricity prices & in particular consumer prices, then someone needs to put a stop to this. Many places are already paying heavy prices for electricity & there is not so much a need for AI that we need to let this happen. What might happen next if some other geniuses come up with an industry that uses a vast amount of electricity? Tell AI users to do one of two things; either produce their own electricity or forget AI, which some are already questioning the use of it anyway. I feel I can do without Ai if it means a higher light bill. There is nothing I know about AI that produces any particular benefit to me.

[politicianslie]

For later

[conservatism_IS_compassion]

I remember when solar panel cost-effectiveness was so bad that it was not sensible to use it when any other option was any good at all. Eg, powering satellites.

But there is a very important “law” to take into account: Wright’s Law, which states that when you double the total quantity ever produced of a given thing again and again, you will concomitantly improve your production methods to achieve that - with the result that unit cost of production will, with each doubling, decline by the same the same percentage.

This means that if you plot unit cost on a log scale against total quantity produced, also on a log scale you will find that a straight line fits the data quite well.

You’ve heard of Moore’s Law, about the unit cost of transistors in computer chips falling in half every year or two? You can get the same prediction by Wright’s Law if you assume that the rate of production of such transistors happens to increase at a geometrical rate.

The thing that has snuck up on the general public (and here I have to include my own self) is that for many decades the production of solar panels has been mounting up - with the result that the cost-effectiveness of solar panels has been improved drastically since the time solar panels got the bad rap most of us grew up associating with them.

. . . and it is that dramatic improvement which accounts for an engineering wunderkind such as Elon Musk declaring that the way to harness nuclear fusion is to use solar panels to get the energy from the nuclear fusion reactor in the sky. And to use rechargeable batteries (note that the production of such batteries has likewise gone ballistic in recent decades) to store solar energy for when the sun isn’t shining.

. . . and the value of batteries is not restricted to their use in conjunction w/ solar panels; anywhere electric rates vary w/ time of day is a potential application. Note that solar panels are now cheap enough to make it practical to install more than the minimum required by the best case - such that cloudy weather still can produce enough energy.