Sep 29, 2026 · 54 min · 13 segments
One chart from Epoch AI shows the cost of AI falling about 47% a quarter since 2023, faster than any technology in history, and it exposes the math under the AI trade. Michael, Liam and Brian on why…
Yeah, I mean, there's no shortage of stuff to talk about or ways to take it.
I think... maybe the meta or the biggest angle is Bitcoin and AI, like anybody paying attention to stuff, you can see that convergence from a mile away, especially now you can feel it before you can talk about it.
And the rest of the market thinks both are niche or don't even pay attention to it.
So it's pretty wild to think about where this goes and how many people are going to be left on the sidelines or how long it's going to take for the rest of the world to kind of adopt it.
The other part, which is fascinating.
We'll talk about intelligence and the level of intelligence that anybody can run personally and how that's gonna continue to grow.
But then I thought it was interesting here.
This was from a Vercel report, which I believe in the beginning of the year, rough numbers.
80% of the tokens were running through Frontier Labs and that shift basically to roughly 17% were running through Frontier and the rest were going through open source models or open weight models.
And I think that's just a fascinating juxtaposition of what the Frontier Labs are selling and all this build out that people are talking about and where does that all go? There's a lot of bullish stuff to talk about, and that's probably more relevant for this pod.
but I can't help but just think about the amount of CapEx spend, the amount of discussion.
And where does that end up when this eventual quasi bubble pops? Not to say there's not value in revenue, but that it's going at such a crazy velocity and story that it's going to maintain while the cost of intelligence is going down.
And it's like, just moving too fast and you can't uh not see how eventually this whole thing pops um and who's left holding the bag you just don't want to be on that side of it
yeah i mean just to your point on the uh rapid succession of ai models that have come out significantly uh recently now it's i mean one that i thought was the most notable is jeff going from zero to 100 million in revenue in seven days like that is one of the craziest revenue ramps that I've ever heard of.
And that is something that can be 100x cheaper than other AI models out there too.
And so you're seeing just like Jevon's paradox in real time of seeing how quickly the costs are declining on one side and then how that's turning into so, so many more tokens on the other side.
And obviously at this point, and like you mentioned, the capabilities are way, way, way far away from the actual diffusion of this technology into the economy over time.
And we've really only started to see just the absolute tip of the iceberg within the next year.
I would imagine that we can see almost daily new model releases that are just as significant as this.
And it's going to just really accelerate the amount of commerce that's happened online, really drive down the transaction costs and friction that exists in the economy today, and really bring so much abundance that people obviously really need to price in Bitcoin if they want to capture the full gains.
Yeah, I mean, there's no shortage of stuff to talk about or ways to take it.
I think... maybe the meta or the biggest angle is Bitcoin and AI, like anybody paying attention to stuff, you can see that convergence from a mile away, especially now you can feel it before you can talk about it.
And the rest of the market thinks both are niche or don't even pay attention to it.
So it's pretty wild to think about where this goes and how many people are going to be left on the sidelines or how long it's going to take for the rest of the world to kind of adopt it.
The other part, which is fascinating.
We'll talk about intelligence and the level of intelligence that anybody can run personally and how that's gonna continue to grow.
But then I thought it was interesting here.
This was from a Vercel report, which I believe in the beginning of the year, rough numbers.
80% of the tokens were running through Frontier Labs and that shift basically to roughly 17% were running through Frontier and the rest were going through open source models or open weight models.
And I think that's just a fascinating juxtaposition of what the Frontier Labs are selling and all this build out that people are talking about and where does that all go? There's a lot of bullish stuff to talk about, and that's probably more relevant for this pod.
but I can't help but just think about the amount of CapEx spend, the amount of discussion.
And where does that end up when this eventual quasi bubble pops? Not to say there's not value in revenue, but that it's going at such a crazy velocity and story that it's going to maintain while the cost of intelligence is going down.
And it's like, just moving too fast and you can't uh not see how eventually this whole thing pops um and who's left holding the bag you just don't want to be on that side of it
yeah i mean just to your point on the uh rapid succession of ai models that have come out significantly uh recently now it's i mean one that i thought was the most notable is jeff going from zero to 100 million in revenue in seven days like that is one of the craziest revenue ramps that I've ever heard of.
And that is something that can be 100x cheaper than other AI models out there too.
And so you're seeing just like Jevon's paradox in real time of seeing how quickly the costs are declining on one side and then how that's turning into so, so many more tokens on the other side.
And obviously at this point, and like you mentioned, the capabilities are way, way, way far away from the actual diffusion of this technology into the economy over time.
And we've really only started to see just the absolute tip of the iceberg within the next year.
I would imagine that we can see almost daily new model releases that are just as significant as this.
And it's going to just really accelerate the amount of commerce that's happened online, really drive down the transaction costs and friction that exists in the economy today, and really bring so much abundance that people obviously really need to price in Bitcoin if they want to capture the full gains.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.