All-In with Chamath, Jason, Sacks & Friedberg
00:57:51 · Jason Calacanis
[00:57:51] S0: say who the hyperscalers are just so people understand the difference between them?
00:57:54 · Chamath Palihapitiya
[00:57:54] S2: Yeah. Amazon, Meta, Microsoft, Google, SpaceX. Yeah. And what they'll be able to do is make a decision. Path number one is continue to rent in the way that you're doing it today. So the status quo. Rent to some customers, large corporates. Right? Rent to the neo clouds, rent to the foundational models. That's what they're doing today. Everybody does that. But then there's the path less traveled, which all of a sudden becomes really interesting. You could decide to take all of the incremental compute and advantage your own products. So Google has a ton of capacity coming online. They could say, I'm gonna give it to Gemini. What does that do? It actually allows Gemini to catch up in quality relative to the foundational models. Why? Because if you constrain the amount of assets that can actually generate compute and a smaller surface area of it is available to the leading companies, those leading companies will have to pay more, the user experience will decay, and then other products can have time to catch up. And now that is important because Anthropic and OpenAI are in a very fast recursive loop. They're on a sixty to ninety day loop in improving their models. So now all of a sudden, without an infinite amount of supply coming online, you can slow these folks down. That's that's the path less traveled. Then there is the crazy path. Crazy path. You should look for this one very narrow thing. You and I talked about this before on the pod, but SpaceX and Anthropic's compute deal has a mutual ninety day out. Anthropic can term in ninety days, SpaceX can term in ninety days, terminate. If SpaceX says I'm gonna terminate this compute in ninety days, That is the king making move. Why? It's a disproportionate amount of the compute that Anthropic needs. It allows SpaceX to reprice that compute at the spot rate, which is probably double. The market rate.
01:00:05 · Jason Calacanis, Chamath Palihapitiya
[01:00:05] S0: Right? Anyway. [01:00:06] S2: The market rate. [01:00:06] S0: The market rate. [01:00:07] S2: And it... And it's probably not just double. It could probably triple. So whatever revenue you see on SpaceX's balance sheet from Ant could two to three x, And Ant would have no choice except to immediately capitulate and take the higher price. It would also show all of the other hyperscalers that they are very far away from having economic yield that is optimized. And so then it forces them to consider, should I be giving it to myself? Should I be selling it to the other people? It then forces the neo clouds to take all of this money that they've raised, whether it's debt or equity, and say, have two choices, which is to buy as much compute today as I can at whatever price it's available because the other path is assured destruction and bankruptcy for me because I am just out of the game. And so what will they do? They will take the last gasp and do whatever they have. So I think all the game theory, Jason, literally comes down to this one deal and whether Elon decides to rip it. Because if he rips it, watch out.
01:01:15 · Jason Calacanis, David Friedberg
[01:01:15] S0: Yeah. And the the other two factors here, Friedberg, and just some predictions. Right now, we have the tokens here in The United States. If the governors decide they want to put the screws tight on big tech hyperscalers and neo clouds and they wanna slow this down, they will need to understand that these data centers could be put in Canada, in Mexico, in Singapore, in Malaysia, and many other locations. The Middle East is a little bit off the table because of the war. War ends. You start putting these data centers in The Middle East. Then the job of doing the next large model will not occur, Friedberg, on US soil. It might occur in Saudi Arabia, UAE, Qatar, Malaysia, Singapore. We're going to be a net importer of tokens, and this will be the biggest cell phone ever because all those $304,100,000 dollar jobs building these data centers are gonna go away. All of that tax will be collected locally in Malaysia where Google is putting a huge data center. And the data can make it around the world quite quickly. It doesn't need to be in The United States, which these governors are gonna learn a very hard lesson in globalization. What are your thoughts here on the data centers being banned or if it is just posturing? And then, Friedberg, could we see The United States become a net importer of tokens as opposed to a net exporter? [01:02:41] S3: Well, they might ban the use of AI too. Okay. [01:02:44] S0: Well, now we're going to... Yeah. A very far out [01:02:47] S3: prediction. Yeah. Floated. I mean, they're... That's the conversation right now.
01:02:47 · David Friedberg, Jason Calacanis
[01:02:47] S3: prediction. Yeah. Floated. I mean, they're... That's the conversation right now. [01:02:51] S0: Okay. I'll unpack that. Who's having the conversation? What exactly is the proposal? Because that seems to me to be a little far out there. [01:02:58] S3: The farther left you go, the the more vocal politicians get on this point, which is no one sees the, quote, benefit of AI, and they don't wanna see AI being developed, being deployed, or being used. That's the general statements that are that are being made. So they wanna ban both its development, its deployment. They wanna ban data centers. They wanna ban technology companies from making models, and they wanna ban companies from using the models. And I think this sort of rhetoric leads to somewhere that's obviously more constrained than we are today. [01:03:34] S0: What are the chances something crazy like that happens? But handicap it for us here. [01:03:37] S3: 10%. California passed legislation, and Gavin Newsom vetoed it. I think it's much higher than 10%. We're seeing state based legislation. I mean, Sacks spent quite a lot of time on this, but there's a patchwork of hundreds of laws that are passing all over the country trying to regulate model development, model deployment, model utilization, what they're allowed to do, whether the local state government gets to audit and inspect how the models operate, how people are using the models, what they're outputting. [01:04:02] S0: You say, Sacks? Answer his
01:04:04 · David Friedberg
[01:04:04] S3: And fundamentally, remember a model is just a bunch of software. It's a code. It's a point about free speech. Can I actually run code on a computer? That is a restriction on, you know, someone basically reading a book is a way to think about it because Sure. At the end of the day, you're inputting and outputting digits, ones and zeros. True sovereignty. Yeah. And remember, we have laws and we have regulations about what do those ones and zeros do? So if those ones and zeros go and take money out of your bank account and put them in mine, I broke a law. I'm a felon. I'm going to get convicted of that felony. If that is used to create a weapon that is illegal, I broke the law. That is a felony. And you can go down the long list of things that people should be concerned about. So you have to ask yourself the question, if we already have laws that protect ourselves from the consequence of people using this sort of speech, this sort of software, then what are we actually doing? And I do think once again that the the primal, call it unconscious political state, is that there is something that threatens the state, that threatens government itself. And I think that that's a big part of the motivation here, and I've said for a long time there's a power play. The question is, and this is me waxing philosophical for a second, is doesn't organization like Anthropic or whomever end up in a place where they're regulatorily aligned with the state such that the state has unlimited access to that model's capabilities to development and deployment and individual citizens and companies do not. And this is almost like if you go back, the the modern, like, issue, the second amendment issue is no longer about who has the bullets and the guns. It's about who has this technology. Because if the state has the technology but the citizens and the businesses don't, then the state has supreme power. And remember, there's no such thing as a state that doesn't have central planning and command, which means you are giving a small group of people an extraordinary amount of power. And the whole point of The United States and the whole point of the Second Amendment was to always have a check on that power. Okay. So if the state ends up getting all of the AI and no one else gets the AI, you then no longer have a check on the power.