just read @Alex_Danco's blog on jevons and baumol paradoxes and it...
Dan McAteer@daniel_mac8
12 views
Nov 05, 2025
~2 min read
Advertisement
1
just read @Alex_Danco's blog on jevons and baumol paradoxes and it perfectly explains the strange economics we're about to see with AI
here are the 3 insights that stood out:
1. jevons paradox - when things get cheaper, we use MORE of them, not less
a transistor cost $1 in 1965, now costs a fraction of a millionth of a cent. caused computers to go from military room-size calculators to disposable shipping tags
google's 7 year old TPUs still run at 100% utilization.
again, think more not less, leading to higher overall demand even when cost plummets
2. baumol's cost disease - less productive sectors become MORE expensive as productive sectors improve
why can a middle-class household afford a new car lease every 2 years but needs to split a nanny with neighbors? the car industry got incredibly productive while childcare didn't
wages compete across ALL sectors. when one sector becomes hugely productive with high-paying jobs, every other sector's wages must rise to stay competitive
3. AI will create BOTH effects simultaneously - and within single jobs. which is pretty weird.
some services will enter the "jevons vortex" and see 10x consumption. legal services, data analysis, content creation. *SOFTWARE DEVELOPMENT*
but services AI can't touch (dog walkers, personal trainers) will become wildly expensive because overall wealth increases
here's the particularly weird part: this happens WITHIN jobs too. when AI automates 99% of a task, that last 1% a human must do becomes incredibly valuable and in-demand
think tesla robotaxis needing human monitors. that last 1% bottlenecks deployment and commands premium wages
the "last 1% that must be a human" becomes the essential employable skillset. those are the new jobs that AI creates
we're heading toward a world where supercomputers are cheap but a 1:1 human tutor is a luxury good
if Dario is right that AI is writing 90% of code by now, get extremely good at that last 10% that only humans can write
that's how you succeed in the AI age: identify what only humans can do and get world class at it.
here are the 3 insights that stood out:
1. jevons paradox - when things get cheaper, we use MORE of them, not less
a transistor cost $1 in 1965, now costs a fraction of a millionth of a cent. caused computers to go from military room-size calculators to disposable shipping tags
google's 7 year old TPUs still run at 100% utilization.
again, think more not less, leading to higher overall demand even when cost plummets
2. baumol's cost disease - less productive sectors become MORE expensive as productive sectors improve
why can a middle-class household afford a new car lease every 2 years but needs to split a nanny with neighbors? the car industry got incredibly productive while childcare didn't
wages compete across ALL sectors. when one sector becomes hugely productive with high-paying jobs, every other sector's wages must rise to stay competitive
3. AI will create BOTH effects simultaneously - and within single jobs. which is pretty weird.
some services will enter the "jevons vortex" and see 10x consumption. legal services, data analysis, content creation. *SOFTWARE DEVELOPMENT*
but services AI can't touch (dog walkers, personal trainers) will become wildly expensive because overall wealth increases
here's the particularly weird part: this happens WITHIN jobs too. when AI automates 99% of a task, that last 1% a human must do becomes incredibly valuable and in-demand
think tesla robotaxis needing human monitors. that last 1% bottlenecks deployment and commands premium wages
the "last 1% that must be a human" becomes the essential employable skillset. those are the new jobs that AI creates
we're heading toward a world where supercomputers are cheap but a 1:1 human tutor is a luxury good
if Dario is right that AI is writing 90% of code by now, get extremely good at that last 10% that only humans can write
that's how you succeed in the AI age: identify what only humans can do and get world class at it.
2
Well worth a read (or listen on the Substack app): substack.com/home/post/p-17…
