I understand the hook of these articles; not just from that site, but from almost the entire publishing industry and the allure of OpenAI. But we’re reaching strange new levels. What’s next? OpenAI is Númenor, OpenAI is Windows Vista, OpenAI is John Carter, OpenAI is Fyre Festival...
I'm inclined to flag these as they should really be one article, or what should be linked to from HN should be a link to some directory for all these articles.
The thing that made Lehman Brothers Lehman Brothers was not that they had a bad business model (the article says so much). It was that they were a central player in a network of financial products that were poorly understood that redistributed risk in a way even less understood. So when the music stopped markets froze because nobody knew who was solvent and who wasn't.
Hard to see what the equivalent in the OpenAI scenario is. Presumably when the music stops a bunch of VCs lose their shirts, the three trillion companies become ordinary trillion dollar companies and silicon valley moves on to look for the next big thing. Probably way more comparable to the dotcom bust.
All these AI services are a great value for the money. I can’t believe that it’s profitable in any way. It seems to me anecdotally that once some market share is reached prices will skyrocket.
I don’t think the barrier to entry is big enough to make this scaling model suitable.
> It seems to me anecdotally that once some market share is reached prices will skyrocket.
Competition determines pricing, and all LLM providers have to compete with "free" open-source solutions. There is no moat to justify increasing costs.
Some argue that having the best model is a moat (it's not by the economic definition), but even then, permissively-licensed OSS LLMs are catching up since there is a slowdown, and anyone can host them.
The market determines pricing, but if the cost of queries exceeds what they can collect in revenue, it is unsustainable. Is it similar to the VC driven craze where gig systems were super cheap until they weren't? Maybe! "It’s easy to sell dollar bills for ninety cents."
I do not think an AI crash will come close to the damage the mortgage crisis of 2008 did.
But, I agree a crash is coming and it will probably mirror the dot-com crash. The only question is Nvidia, I think Nvidia has potential to cause a lot of pain.
in the United States, any business which can pull of scenario where a small group of rich people is exploiting large group of poor people is never going to fail - hence Uber is not a good comparison :)
OpenAI is Taco Bell: https://sherwood.news/tech/openai-is-taco-bell/
OpenAI is Pixar: https://sherwood.news/tech/openai-is-pixar/
OpenAI is Napster: https://sherwood.news/tech/openai-is-napster/
OpenAI is Salesforce: https://sherwood.news/tech/openai-is-salesforce/
And so on. They're all listed here: https://sherwood.news/tech/what-companys-past-reveals-the-fu...