This isn't like the early days of the web, or Amazon, or any of those other big winners that lost money before becoming profitable. Those were all propositions with excellent "unit economics" - they got cheaper with every successive technological generation, and the more customers they added, the more profitable they became.

11/

AI companies have - in the memorable phraseology of Ed Zitron - "dogshit unit-economics." Each generation of AI has been vastly more expensive than the previous one, and each new AI customer makes the AI companies lose *more* money:

pluralistic.net/2025/06/30/acc…

This week, no less than the *Wall Street Journal* published a lengthy, well-reported story (by Eliot Brown and Robbie Whelan) on the catastrophic finances of AI companies:

wsj.com/tech/ai/ai-bubble-buil…

12/

The *WSJ* writers compare the AI bubble to other bubbles, like Worldcom's fraud-soaked fiber optic bonanza (which saw the company's CEO sent to prison, where he eventually died), and conclude that the AI bubble is *vastly* larger than any other bubble in recent history.

The data-center buildout has genuinely absurd finances - there are data-center companies that are collateralizing their loans by staking their giant Nvidia GPUs as collateral.

13/

This is wild: there's pretty much nothing (apart from fresh-caught fish) that loses its value faster than silicon chips. That goes *triple* for GPUs used in AI data-centers, where it's normal for tens of thousands of chips to burn out over a single, 54-day training run:

techblog.comsoc.org/2024/11/25…

Talk about sweating your assets!

14/

That barely scratches the surface of the funny accounting in the AI bubble. Microsoft "invests" in Openai by giving the company free access to its servers. Openai reports this as a ten billion dollar investment, then redeems these "tokens" at Microsoft's data-centers. Microsoft then books this as ten billion in revenue.

15/

That's par for the course in AI, where it's normal for Nvidia to "invest" tens of billions in a data-center company, which then spends that investment buying Nvidia chips. The the same chunk of money being energetically passed back and forth between these closely related companies, all of which claim it as investment, as an asset, or as revenue (or all three).

16/

The *Journal* quotes David Cahn, a VC from Sequoia, who says that for AI companies to become profitable, they would have to sell us *$800 billion* worth of services *over the life of today's data centers and GPUs*. Not only is that a very large number - it's also a very short time. AI bosses themselves will tell you that these data centers and GPUs will be obsolete practically from the moment they start operating.

17/

Mark Zuckerberg says he's prepared to waste "a couple hundred billion dollars" on misspent AI investments:

businessinsider.com/mark-zucke…

Bain & Co says that the only way to make today's AI investments profitable is for the sector to bring in *$2 trillion* by 2030 (the *Journal* notes that this is more than the combined revenue of Amazon, Google, Microsoft, Apple Nvidia and Meta):

https://www.bain.com/about/media-center/press-releases/20252/$2-trillion-in-new-revenue-needed-to-fund-ais-scaling-trend---bain--companys-6th-annual-global-technology-report/

18/

Why Zuckerberg says risking billions on AI is worth it

Zuckerberg says an AI bubble is "quite possible," but warns the bigger risk is building too slowly and missing superintelligence.

^{Lee Chong Ming (Business Insider)}

How much money is the AI industry making? Morgan Stanley says it's $45b/year. But that $45b is based on the AI industry's own exceedingly cooked books, where *annual* revenue is actually *annualized* revenue, an accounting scam whereby a company chooses its best single revenue month and multiplies it by 12, even if that month is a wild outlier:

wheresyoured.at/the-haters-gui…

19/

The Hater's Guide To The AI Bubble

Hey! Before we go any further — if you want to support my work, please sign up for the premium version of Where’s Your Ed At, it’s a $7-a-month (or $70-a-year) paid product where every week you get a premium newsletter, all while supporting my free w…

^{Edward Zitron (Ed Zitron's Where's Your Ed At)}

Industry darlings like Coreweave (a middleman that rents out data-centers) are sitting on *massive* piles of debt, secured by short-term deals with tech companies that run out *long* before the debts can be repaid. If they can't find a bunch of new clients in a couple short years, they will default and collapse.

20/

The AI bubble has absorbed more of the country's wealth and represents more of its economy than historic nation-shattering bubbles, like the 19th century UK rail bubble. A much-discussed MIT paper found that 95% of companies that had tried AI had either nothing to show for it, or experienced a loss:

technologyreview.com/2019/01/2…

A less well-known U Chicago paper finds that AI has "no significant impact on workers’ earnings, recorded hours, or wages":

papers.ssrn.com/sol3/papers.cf…

21/

We analyzed 16,625 papers to figure out where AI is headed next

Our study of 25 years of artificial-intelligence research suggests the era of deep learning may come to an end.

^{Karen Hao (MIT Technology Review)}

Anything that can't go on forever eventually stops. Trump might bail out the AI companies, but for how long? They are incinerating money faster than practically any other human endeavor in history, with precious little to show for it.

22/