AI output is getting cheaper while the machines needed to produce it keep getting more expensive

Share this story

The $7.5 trillion argument says the AI boom isn’t anywhere near the historical spending limit yet.

Fine.

But I think the more interesting problem is happening underneath that number.

AI is getting cheaper incredibly fast while the cost of building the AI machine is still exploding.

That means the bull case now needs usage to go absolutely vertical.

If token prices keep falling another 50%, companies need a hell of a lot more AI consumption just to make up for the lost pricing.

So I’m not ready to call the AI top.

But I also wouldn’t use the $7.5 trillion number to declare everything is fine.

The real test isn’t how much they can spend. It’s whether AI revenue can grow faster than AI prices are falling.

That is where this gets interesting.

0 views