Tag: META

* I continue to call B.S. to the anticipated projections for productivity enhancement, the largely positive expectations for marginal aggregate economic growth/activity from AI and to the unrealistic forecasts for return on invested capital of the AI capital spending spree…

* If the dreams of AI (job) cost saves are realized, how is the U.S. economy to function? 

“When you combine ignorance and leverage, you get some pretty interesting results.”

– Warren Buffett (1993 Lecture to Columbia Business Graduate School Students)

With regard to The Lack of Situational Awareness Hedge Fund, one point that has not been made yet is the moronic investors there. If you put all your money on one number at the roulette table, if it hits, you take your winnings and run away from the casino as fast as you can. It sounds like almost none of the imbeciles in the fund did that. You would think the fund’s performance would have set off some alarm bells for anyone with a brain about the amount of risk that was being taken. But I guess it didn’t. Why? Because apparently, a lot of the capital in the fund came from Silicon Valley VC types who were all drinking the same Kool-Aid, and also have zero concept of risk. Better lucky than good I guess. A lot of dummies have made a lot of money in recent times. This is what a lack of price discovery and too much money in the economy can do. But as they say, he who is closest to the monetary spigot drinks the most, until they don’t, I guess.

One analysis that would be interesting to do would be if all the revenue forecasts came true for the model providers (OpenAI, Anthropic, Gemini, Grok AI, whatever Meta ($META) calls its thing, all the Chinese and open source, etc.), how many implied jobs saved would that be? For example, I saw one forecast from a prominent pundit that had Anthropic at $2 trillion of revenue in 2030, and still growing rapidly. My guess is these are the types of numbers investors are being shown. 

At any rate, their revenue is the customer’s expense. If the idea of AI is to replace jobs, at $100k per job, $2 trillion of revenue is 20 million jobs replaced. That is Anthropic alone. I suspect the revenue forecasts for OpenAI are the same. Another 20 million jobs down the tube by 2030 alone. Remember both are meant to still grow exponentially from that point. The forecasts for those two alone imply 40 million jobs gone by 2030 and then continuing to eat jobs at an exponential rate.

To put that in perspective, there are 160 million employed Americans. But only 20% make $100k or more. Let’s call those white collar jobs – those are the type of jobs AI is supposed to replicate. Twenty percent of 160 million is only 32 million jobs. So somehow only two of the AI companies displace 40 million jobs when there are only 32 million people in those jobs to begin with. If you wanted to argue lower-end jobs, like the $50k type, well that would be 80 million jobs. $51k is median income in the U.S., and that would be half the labor force down the tubes. This is only OpenAI and Anthropic. I have left out the revenue forecasts for all of the other players that are also heavily investing and meant to be doing the same things.

If this were to ever happen, how exactly is the U.S. economy supposed to function? I certainly do not want to hear any retorts about Jevons Paradox and how these are productivity enhancement tools.  These things are not Microsoft Excel. Excel costs about $10-$20 a year. It does not take trillions in capex to run. It takes a PC.  hat is a productivity enhancement tool. The AI companies, given the underlying cost base, are all about replacing labor. That is the only way the massive amount of investment could pay for itself. 

Said another way, for their customers to pay them this amount in revenue, they will have to realize more than that amount in cost savings, otherwise there is no point, it just becomes increased expense. AI cannot do anything that humans can’t already do. It is just a trade. The tech does it, or the human does it. That assumes the AI can even do it to begin with, which in many cases, it cannot.

Of course, I do not believe any of this, because I do not think (as expressed in some of the 225+ More Tales series I have written) Gen AI works well enough to replace large swaths of white collar labor, and it never will. Which is why the investment dollars into it will never be re-captured.  Further, I believe the revenue (and profitability) forecasts being put out there for all the model providers are total poppycock and imply things that are just not possible for a variety of reasons. I gave one example, there are probably many more.

WSJ:  “The push to expand head count, at least modestly, is a reversal from the prevailing corporate messaging during much of the AI era. Major employers largely held back on adding people due to economic uncertainties or a belief that artificial intelligence could shoulder more tasks on the job. But some executives say the costs and limitations of AI now demand that more people be added; others want to hire people back following layoffs.”

Big Companies Are Starting to Hire Again, Defying Predictions of AI Wipeout

Bottom Line 

I continue to call B.S. to the anticipated projections for AI productivity enhancement, the largely positive expectations for marginal aggregate economic growth/activity from AI and to the unrealistic forecasts for return on invested capital of the AI capital spending spree. 

Position: None