Dax

They still allocate most of their stuff to training, not to inference.

They still allocate most of their stuff to training, not to inference.

Okay. Yeah, I mean, it just makes sense because why when... If you don't win the model battle, the inference, the fact that people are using your product is kind of irrelevant. So it doesn't make sense to over-allocate there.

Okay. Yeah, I mean, it just makes sense because why when... If you don't win the model battle, the inference, the fact that people are using your product is kind of irrelevant. So it doesn't make sense to over-allocate there.

I mean, in the end to me so far, and I felt this from the beginning, this feels like the worst part of the sack to be in. It is the most difficult and the most expensive and it is the most like commodified. So yeah, I mean, I think the thing that people point out with deep seek is like, It's impressive to create something as good as open AI stuff.

I mean, in the end to me so far, and I felt this from the beginning, this feels like the worst part of the sack to be in. It is the most difficult and the most expensive and it is the most like commodified. So yeah, I mean, I think the thing that people point out with deep seek is like, It's impressive to create something as good as open AI stuff.

It's totally realistic to assume making a model that's 1% better than open AI stuff costs like $50 billion. Like that's like totally realistic. And that's like an argument in favor of being like, this is why open AI will, you know, it's not really a threat to them.

It's totally realistic to assume making a model that's 1% better than open AI stuff costs like $50 billion. Like that's like totally realistic. And that's like an argument in favor of being like, this is why open AI will, you know, it's not really a threat to them.

Simultaneously, it's also like condemning this entire business because it's just like if it's going to take that much capital to make these marginal improvements and it's like a crazy competitive space where the costs are being driven to zero and all these companies are competing. Yeah, it's just... I don't know. To me, it never made sense.

Simultaneously, it's also like condemning this entire business because it's just like if it's going to take that much capital to make these marginal improvements and it's like a crazy competitive space where the costs are being driven to zero and all these companies are competing. Yeah, it's just... I don't know. To me, it never made sense.

If I was an investor, this is not the part of... And I want to bet on this AI thing. This just feels like the worst place to put your money. It's so intense. So capital intensive, right? Yeah. When I see that, I'm like, I need to invest in someone that benefits from... having access to cheap AI models, not the people building the cheap AI models. And yeah, like VC Twitter, like it's funny.

If I was an investor, this is not the part of... And I want to bet on this AI thing. This just feels like the worst place to put your money. It's so intense. So capital intensive, right? Yeah. When I see that, I'm like, I need to invest in someone that benefits from... having access to cheap AI models, not the people building the cheap AI models. And yeah, like VC Twitter, like it's funny.

They just go on these little things. And currently they're on this, they've swung back and forth. And currently they're all saying, oh yeah, like the application layers where you're going to make a lot of money. But like, you know, a couple of weeks ago they were saying the opposite. But I do, that does make more sense to me. Again, it's not,

They just go on these little things. And currently they're on this, they've swung back and forth. And currently they're all saying, oh yeah, like the application layers where you're going to make a lot of money. But like, you know, a couple of weeks ago they were saying the opposite. But I do, that does make more sense to me. Again, it's not,

I'm not taking the moonshot bet because the moonshot bet is you invest in open AI and they eliminate the whole economy, which I get. And I like bets like that. It's just for me, this one is not the one that that would go for. Yeah, there's something like less crazy is probably going to be the outcome.

I'm not taking the moonshot bet because the moonshot bet is you invest in open AI and they eliminate the whole economy, which I get. And I like bets like that. It's just for me, this one is not the one that that would go for. Yeah, there's something like less crazy is probably going to be the outcome.

One last thing on this. Yeah. I did come across something today. Do you remember Mistral?

One last thing on this. Yeah. I did come across something today. Do you remember Mistral?

Okay, so... Where'd that go? They're like, this is maybe the worst company fundraise of all time because... They raised, like, $150 million on, like, a $300 million valuation or something. What? Like, gave up half the company? Yeah, so they, like, gave up half their company. And, like, that is nowhere near enough money to, like, play in this game.

Okay, so... Where'd that go? They're like, this is maybe the worst company fundraise of all time because... They raised, like, $150 million on, like, a $300 million valuation or something. What? Like, gave up half the company? Yeah, so they, like, gave up half their company. And, like, that is nowhere near enough money to, like, play in this game.