[Chapter-delegates] New Cory Doctorow book on AI

[Johan Helsingius]

Hi Olivier,

On 27/07/2026 5:47 pm, Olivier MJ Crépin-Leblond via Chapter-delegates 
wrote:

> unfortunately I used to enjoy publications from Cory Doctorow, 
> enshittification being an enjoyable read.
> 
> However, I challenge his concept of the AI bubble bursting, as if the AI 
> companies are not making money and are going bust. The paragraph about 
> NVIDIA did it for me, because it is exactly the wrong example:

Cory does tend to exaggerate a bit.

> The value of NVIDIA bears no resemblance with its actual profits. That's 
> an investor problem, not a corporate company problem.

Indeed.

> One often talks of "AI bubble" in the sense of "DotCom bubble" and I 
> would argue that these are two very different things.

  That's where I might disagree with you. There are three major
  similarities, one being the accounting of major players selling
  virtual assets to each other, and the other being the outrageous
  valuations based on forecasts either pulled out of thin air or
  from extrapolating initial enthusiasm way beyond reasonable limits.
  At the time I pointed out that you can't just extrapolate Internet
  traffic growth multiplied by user growth as a growth rate for your
  revenues - traffic grew, but customers were not prepared to pay any
  more per month, and the third being the huge investments by
  infrastructure companies (then fibre optic network and date center
  construction, now mainly data center construction) that are
  not matched by revenues.

  There will be an "AI" (really, LLM) market adjustment, and companies
  will go bankrupt. Look at the costs and debts of the major LLM
  infrastructures companies (that they are really trying to hide).

  	Julf