I mean all it would take to cause a huge NVDA sell off is Zuckerberg or Nadella getting bored. Or their boards getting impatient with the return on investment.

Burry's in a short position not because NVDA isn't selling product, but because their valuation is astronomical entirely due to all their eggs in this one basket. If and when the massive capex bottom falls out or even just slides, NVDA isn't going to make that revenue up from its gaming division.

It's a gold rush and NVDA's the only ones selling shovels. But prospectors don't know if this is actual gold or just pyrite. That's why the books cooking on depreciation of the chips from hyperscalers is concerning. That's why the electrical grid capacity questions are concerning. Lots of faltering.

I heard on the Panic World podcast once that the host tried ChatGPT out for therapy for a few weeks during a gap in his insurance. It was serviceable for him, but he made sure to feed it prompts like "talk to me like you're a good friend that's sick of my bullshit" to discourage reinforcement.

The real world effects of this are going to be really funny, though.

You'd think they could mitigate this a bit by making the LLMs not be so encouraging all the fucking time, but it's pretty clear companies like OpenAI and Perplexity are relying on stickiness metrics. Keep users sending more queries. Hard to juice active user stats when a robot says the user is dumb.

100% chance this puts me in some database they have, but their AI is far more likely to drone attack my old ass anti-abortion neighbor across the street than me. So I will win again.

Keith has never had an iota of self awarenes, ever

if you missed it...

Keith Olbermann