

I spent a summer in college working in a piano store. It was totally empty 90% of the time, and my job consisted mostly of keeping it from burning down until people came in for piano lessons after school.
They were the local Bosendorfer dealer, so I had lots of time to play a full size concert grand all summer, the one with the 8 extra low keys at the bottom. It was such a smooth playing instrument, that it nearly ruined me for all other pianos. After playing that Bosendoerfer, a mere $50,000 piano felt like a toy.

I recently watched a video about the first electronic synthesizer. It was built in 1897, and was housed in the basement of an entire city block in NYC. Several decades later, and it would fit in a suitcase, with far more functionality. It worked, but it was huge and unwieldly, to the point of being completely impractical. It needed to wait until the technology caught up to it, and made it a truly viable instrument, and not just a concept.
Data Centers are like that 1897 synthesizer. It works, sure, but at what cost? It’s expensive in every way, from the building costs, the energy costs, the environmental costs, etc. We have a concept and a prototype, but not a truly viable product yet. Maybe in a decade or two, with a proper goal in mind, we can get there, but right now, Data Centers are just that big, dumb synthesizer that takes up a city block.