Really liked your framing of the “two clocks” financial vs technological. I wonder if there’s actually a third clock starting to matter too: the AI-forensic one. As AI tools get better at reading filings, comparing them across time and peers, and flagging inconsistencies or overly aggressive assumptions, the market’s recognition of risk compresses. What used to take quarters of skeptical analysts and activist decks can now surface in hours.
That adds another layer to what we’re seeing with Nvidia and the AI infra trade. It’s not just sentiment wobbling around capex and digestion; it’s also that AI itself is making earnings quality and business-model fragility harder to hide. So the market can be right about the next 12–24 months and the 10–20 year AI story can still be intact but price discovery in between is going to get sharper, faster, and less forgiving because AI is now stress testing the narratives in real time.
I love this point of view – very balanced. Especially the connection to investment and retirement plans. That’s a great point that is not talked about much, which is absolutely not obvious, especially to the people who would be affected by it the most
I just read this. Very astute analysis. It is February, and i think you’re very much on the money on financial and technological cycles with their different life trajectories. I wonder where Palantir fits in this picture.
I think the bubble is almost inevitable. The amount of people who have a solution or a way to usher in a technology is almost always larger than those who can pull it off.
One consideration though: is it possible that part of the bet is being a company that makes it after the bubble bursts? So while many companies will go away, maybe OpenAI does get to 100B because it absorbs Perplexity or Midjourney or some other smaller system which basically turns it into Meta (ie top player made up of a couple of other players)
Really liked your framing of the “two clocks” financial vs technological. I wonder if there’s actually a third clock starting to matter too: the AI-forensic one. As AI tools get better at reading filings, comparing them across time and peers, and flagging inconsistencies or overly aggressive assumptions, the market’s recognition of risk compresses. What used to take quarters of skeptical analysts and activist decks can now surface in hours.
That adds another layer to what we’re seeing with Nvidia and the AI infra trade. It’s not just sentiment wobbling around capex and digestion; it’s also that AI itself is making earnings quality and business-model fragility harder to hide. So the market can be right about the next 12–24 months and the 10–20 year AI story can still be intact but price discovery in between is going to get sharper, faster, and less forgiving because AI is now stress testing the narratives in real time.
Interesting dynamic to consider!!
Excellent. Your explanations demystified some things for me. Your wording and way of delivery are second to none that I know of.
I love this point of view – very balanced. Especially the connection to investment and retirement plans. That’s a great point that is not talked about much, which is absolutely not obvious, especially to the people who would be affected by it the most
https://open.substack.com/pub/evolvingtheory/p/the-ai-bubble-isnt-the-dot-com-bubble-b0d?utm_source=share&utm_medium=android&r=275w0u
I just read this. Very astute analysis. It is February, and i think you’re very much on the money on financial and technological cycles with their different life trajectories. I wonder where Palantir fits in this picture.
Adding: https://www.zoiver.media/p/the-ai-bubble-economics-hype-cycles?r=320zv5&utm_campaign=post&utm_medium=web&showWelcomeOnShare=false
I think the bubble is almost inevitable. The amount of people who have a solution or a way to usher in a technology is almost always larger than those who can pull it off.
One consideration though: is it possible that part of the bet is being a company that makes it after the bubble bursts? So while many companies will go away, maybe OpenAI does get to 100B because it absorbs Perplexity or Midjourney or some other smaller system which basically turns it into Meta (ie top player made up of a couple of other players)