The problem is that "AI: The Product" and "AI: The Company" have two different marketing needs that undercut each other.
AI companies need to come off as these big scary world transforming entities because they need concessions from politicians and they need their perceived value to go up for their shareholders.
But AI products need to come off as benign but useful things you would use in your every day.
When the White House puts a temporary hold order on frontier models, its incredible marketing for the company. It makes what they're doing seem really important, like an economic nuclear weapon. But for the product it's horrible! It is literally the government stepping in to say you cannot improve your product.
Maybe LLMs are legitimately useful products that will have value in most people's everyday lives, but most people will have their expectations completely off. And their intuition isn't wrong that AI companies are here to make their lives worse.
The current thinking is automated agents is what turns this from an industry in the tens of billions to a multi trillion dollar one. So yes you are right on the money, agents stimulate demand for this thing they've built.
Just to clarify because it seems like most of the comments aren't understanding. Berkshire sold the Amazon stock in the fourth quarter of last year meaning it is likely the last large move Warren Buffett is going to make as head of Berkshire as he stepped down on December 31 of 2025. That's why the article is titled that way and partly why its significant. Warren Buffett has traditionally been averse to tech stocks but picked up a slug of Amazon in 2019.
Its probably not very likely that if a large buyer pulled out, NVIDIA could just sell to other customers. If a large buyer pulls out, that's a massive signal to everyone else to begin cutting costs as well. The large buyer either knows something everyone else doesn't, or knows something that everyone else has already figured out. Either way, the large buyer pulling out signals "I don't think the overall market is large enough to support this amount of compute at these prices at current interest rates" and everybody is doing the same math too.
I personally don't find using LLMs "fun" but I do like this article for one simple reason. It points out that most contemporary frameworks for web are forms of slop and I don't think you should feel bad using an LLM to generate slop code and config to deal with it.
I learned an instrument (guitar), then I learned another instrument (bass guitar). I caught up with my family, I got to spend a lot more time with my nephews and nieces. I got to spend a lot of time with my dad. I picked back up circuit design and made an overdrive guitar pedal from scratch and learned KiCad in the process. I caught up with old friends but lost touch with other ones. I probably acted like an ass to a lot of people. I finally came to understand why people like watching sports. I learned how to write in cursive. I walked a lot. I listened to a lot of music. I listened to some podcasts.
I dunno, it sucks and its painful. You're constantly worried and people who at first try to support you then get pissed off at you for something you can't really control. I hope you can find your way through it.
My parents love watching those videos so I know what you're talking about. There's four factors I think.
One, the quality of the saffron makes a big difference in price. Longer deeper red strands are much more expensive than shorter and more yellow strands. I suspect they are not using the most expensive saffron in their teas.
Two, there's a lot of markup purchasing saffron from most retailers, especially here in the US. You can get that amount of saffron for far less than $10,000 if you know someone who has a connect.
Three, as you said the majority of saffron production is from Iran. Some friends I know are getting their saffron from the UAE but I'm pretty sure its being re-exported from Iran.
Four, those videos are ultimately productions. If you compare the earlier videos to the later videos on some of that channel you can tell they've upgraded their digs. They might just be exaggerating for the camera.
AI companies need to come off as these big scary world transforming entities because they need concessions from politicians and they need their perceived value to go up for their shareholders.
But AI products need to come off as benign but useful things you would use in your every day.
When the White House puts a temporary hold order on frontier models, its incredible marketing for the company. It makes what they're doing seem really important, like an economic nuclear weapon. But for the product it's horrible! It is literally the government stepping in to say you cannot improve your product.
Maybe LLMs are legitimately useful products that will have value in most people's everyday lives, but most people will have their expectations completely off. And their intuition isn't wrong that AI companies are here to make their lives worse.