IMNSHO, betting on mega data centers is betting against software and hardware innovation. LLM's already run on desktops and will eat up the long tail of uses faster than a bean counter switched to Excel from a mainframe Cobol app.
That leaves enterprise use for surveillance capitalism and job elimination, and government use for mass surveillance and more efficient killing. But that's just a case, IMO, of a few trying to corner the market on AI, with incredible duplication of resources until a single winner emerges (if possible) or until the incumbent business and defense powerhouses develop their own, which is highly likely as technology advances, and they already own the customer base(s).
I expect that a lot of incumbents will be issuing "NIH" directives.
AND the upshot is a vast amount of unused electrical generation. Except for fossil-fueled plants, a great glut of energy for heating and cooling homes for humans.
Concentration of wealth/power demands energy to drive entropy to zero.
IMNSHO, betting on mega data centers is betting against software and hardware innovation. LLM's already run on desktops and will eat up the long tail of uses faster than a bean counter switched to Excel from a mainframe Cobol app.
That leaves enterprise use for surveillance capitalism and job elimination, and government use for mass surveillance and more efficient killing. But that's just a case, IMO, of a few trying to corner the market on AI, with incredible duplication of resources until a single winner emerges (if possible) or until the incumbent business and defense powerhouses develop their own, which is highly likely as technology advances, and they already own the customer base(s).
I expect that a lot of incumbents will be issuing "NIH" directives.
AND the upshot is a vast amount of unused electrical generation. Except for fossil-fueled plants, a great glut of energy for heating and cooling homes for humans.
Concentration of wealth/power demands energy to drive entropy to zero.
It never gets there but a handful always try.