That might be the weirdest part of this whole AI trade.
People keep arguing over whether AI is real.
Wrong question.
It’s obviously real.
The better question is how fast better models turn into actual economic results.
One guy in the discussion works at a frontier AI lab and basically asked what has really changed over the last two years.
More code gets generated.
Agents are better integrated.
Models can do more.
But a huge amount of that code never reaches production.
Another person working in AI said companies keep asking the same question:
“How do we reduce costs with this?”
And that gets interesting.
Because if the main economic use of AI right now is helping companies do the same work with fewer people, faster workflows, and less friction…
that doesn’t automatically create the giant new revenue streams investors are pricing in.
Then there is the other problem.
The AI industry sold investors a much bigger story.
End of work.
AGI.
Millions of agents.
Robots everywhere.
Software companies becoming dramatically more productive.
So even if AI settles into something much more boring…
better coding
better research
better search
better customer service
better internal automation
…the technology can keep improving while the stocks still get hammered.
Why?
Because the market doesn’t price “pretty useful.”
It prices the future.
That is why I think the next AI disappointment could be very different from the dot com bust.
AI doesn’t have to fail.
It just has to become useful before it becomes enormous.
And if that happens, the technology keeps spreading while investors have to rethink how much money all that progress is actually worth.
That might be a much bigger problem for the AI trade than AI being fake.
Disclaimer: This is not financial advice and is for educational purposes only. Please conduct your own due diligence.