Not sure if Jevons Paradox is gonna save bottleneck bros. The AI cost (compute need) for a lot of applications is literally collapsed with Jev. Especially when the existing models mainly make money on junk text output.