While I obviously disagree (see More Tales From Nvidia series) with Jim’s conclusion (I will not manage money based on a “greater fool’s theory” that it is rational to overweight AI even if you don’t believe it will work), his post is important:
Jim F
i just realised the stock market doesn’t think AI works.
the lack of breadth has been a bothersome contradition in the AI thesis. the thesis as i understand it is, AI tools will drive massive poductivity at ALL cpompanies lage and small. in turn these companies will be willing to share a large portion of the productivity windfalls with the AI tool providers that made it possible. those shared productivity gains of ALL companies will in turn drive the revnue for return of principle plus ROIC of todays spending. in theory that all makes sense.
the issue i have is breadth doesn’t match this thesis. lets say companies will pay 75% of the productivity gain to the AI providers and keep 25%. that would mean ALL companies in agregate are about to become more valuable by 1/3 the revenue projections diving AI valuations. for eample $100 of productivity gained is $75 fee to LLM’s and $25 of reatined earnings for the company. $25 is 1/3 of $75. so if ALL companies ara about to get this windfall why is the median stock -16% or 60% of the S&P -20% etc etc?
this has bothered be until today. today i realised the market doesn’t believe the AI revenues are going to happen. however, if you manage money you’re now in a prisoners dilemma. you can;
a) don’t invest in AI because you don’t believe in it. if you did that this year you lost all you AUM. money mangers job is grow AUM and not get best risk adjusted returns.
b) invest in AI and when it blows up you lose client money. however, all you competitors did as well. so you survive in you relative market share after the blow up.
this is why it’s actually rational to be overweight AI even if you don’t believe it will work.
Position: Short NVDA (S)
