The other side of this is only new baby firms invest in that thing that makes a little bit of money. But given enough refinement, that thing starts making more and more money as it gets better and better. And soon, that new baby firm outshines the incumbent. The incumbent's wasn't incentivized to invest in the thing that started off worse but eventually became the new model. Think Kodak with film-vs-digital cameras.
This was the thesis of 1997's The Innovator's Dilemma, written by the guy who coined "Disruptive Technology".
For obvious reasons, the expected rate of return needs to clear the hurdle of the risk-free interest rate. This puts a pretty high floor on activity that is "worth doing". This is a mechanism by which the phenomenon of ZIRP diversifies economic activity.
Small absolute changes in risk-free interest rates cause many things to become unprofitable when the relative change in interest rates is large. A risk-free rate of 1.0% and 1.5% are both small but the latter is 50% higher than the former.
Is your experience in the same America where Meta is losing another 4-6 billion $ this year in AR/VR business unit, after losing 19 billion $ last year. Similar with Google's and Apple's AR/VR unit which also consume a lot of money in R&D(funding a lot of high paying jobs) and not make any money, yet.
So sure, there's no risk appetite for things that make little money, except for all the evidence proving the contrary.
If it ends up AI only makes a little profit annually in the longer term the whole thing collapses on itself.
Because "making little money" is a commodity business activity, overrun with competition from Europe and Asia.
So why would you ever want to compete in the race to the bottom of "little money" when you have the highest labor cost in the world? It makes no business sense.
You go into "all or nothing" moonshots because Europe and Asia can't compete there. Especially when you have the world reserve currency as the infinite money glitch cheat code (while it lasts).