00:03:24 · Aaron Levie
And so it sort of makes you wonder, well, if you're, you know, a country, Germany, France, Japan, you know, UK, and you have the capability to to, you know, be at at, you know, frontier or near frontier AI from a a talent and an infrastructure standpoint, do you need to be, you know, you know, building these these kinds of models yourself and make sure that you have them available to, you know, businesses and and users within your country? I think it it really kind of introduces that question in a much bigger way. And we've already seen obviously examples of this China obviously developing, you know, kind of great open source models, Mistral, you know, doing the same. But I think this highlights the increased importance in that dynamic and the idea that you can have sovereign AI or AI that you control. Now there's one... There's sort of one caveat to all of this, though, which is which is, you know, you're not going to you're not going to take that big of a of a sacrifice or a discount on your... On the model's intelligence for that benefit. So if Frontier from US in a in a kind of a closed model environment stays 10, 20, 30, 50% better than than the next best model that you could, you know, make sovereign, I think, you know, people will take that risk, and they will use the frontier model at at all times. So this only works if open weights models or, you know, sovereign AI can remain close to the frontier and much closer than it has historically. And I think we're seeing some examples with with Kimi and DeepSeek and and and and, you a few of these new models coming out of China and Nemotron. It's it's... You know, these models are doing a great job kinda staying at the forefront, but they're still not fable level. And that's gonna be the big question for the ecosystem over the over the coming kind of months and years.
00:05:16 · Guillaume Moubeche
[00:05:16] S1: Do you feel like it's only gonna be a matter of performance, like staying, let's say, as you said, 20, 30%, like, behind or also a matter of timing? Because what we're seeing right now is, like, when the first model went out, you know, like, the the first lab to catch up maybe took like twelve months or or like, yeah, a bit more than a year, but now I I feel like models are catching up like within a couple of months. So do you think it's only a matter of difference or also a matter of how fast can other models catch up on the best models? Well, I
00:05:51 · Aaron Levie
[00:05:51] S0: think, you know, there's there's a big debate, you know, and and if you watch, you know, Dwarkesh talk to Dario or Jensen, you know, you you sort of see examples in contours of this debate, which is is the frontier and the open... And the and the sort of, let's say, the best in class open weights model, how close are these tracking versus are we seeing a divergence? Are they six months behind? Are they three months behind? Are they one month behind? Or how much is it really just a factor of how much compute you have for inference and and kind of test time compute as an example? Because there's different ways to to kind of really drive the the workloads in these models to get better results. I am generally actually optimistic that that that that the frontier open weights models will be able to to stay relatively caught up. I think it's it's of such economic and and strategic importance for, you know, countries like China, you know you know, maybe individual companies like like Mistral or even NVIDIA to make sure that that that that is sort of maintained. And we have not yet seen a breakthrough that is so either proprietary or kind of undiscovered by the rest of the ecosystem that that keeps us completely closed off. You know, I think there's obviously a conversation around, you know, RSI and and and what, you know, what what does self improvement look like, and does that kinda lead to a kind of an accelerating takeoff? Like, whoever whoever is able to design these systems that that self improve first, then obviously there's a compounding dynamic to that. At the same time, it's hard to imagine China not wanting to to stay in the race in a very big way. Like, it... It's unlikely that China says, you know what? We're totally fine with Anthropic at OpenAI being the only vendors of AI in the world, and, you know, we'll be closed off from this ecosystem. So at some point, you know, you can just throw enough money at this problem. You know, you can... China could just say we're gonna spend $50,000,000,000 and just and just make sure that this is this is of such critical national security importance that we have to make sure that we're throwing compute at this problem. We're gonna kind of, you know, sequester the the top, you know, researchers and scientists and and and and make it happen.
00:08:11 · Aaron Levie, Guillaume Moubeche
I I think it's unlikely that that wouldn't work, and I haven't seen any evidence that suggests otherwise. [00:08:19] S1: So your view, like, when it comes to China is is what you're regarding, like, the the chips because for so long, like, you know, there's been kind of an embargo on chips and it was not possible for them to get them and they were getting it through like Singapore or like all these kind of things. Like, what... What's kind of your view? Because it's like... It it feels like eventually, like... I mean, in The US, it has always been a very pro open market, and now it feels like things are moving away from it. So I'm curious to get your take like...
00:08:51 · Aaron Levie
[00:08:51] S0: Well, if you... And and I think a lot of very smart people believe this. If you if you believe that AI is is both a national security, you know, threat and opportunity, and it's a, you know, of of critical strategic economic imperative that can lead some some people down the path of, well, then you you actually want this to be a closed environment. You need to keep the chips. You need to keep the models. And and I do think it's of national security, you know, level of of importance. But there's this interesting tension, which is which is if you're also not the exporter of the technology, then the ecosystem effects start to favor the non US companies. And Jensen made this case in in the Dwarkesh podcast about a month ago. And actually, I probably lean more toward Jensen's side of this argument, which is which is, you know, if you just play this out, AI is is too important for for, you know, the rest of the world to say, you know, oh, yeah. We're only going to, you know, use the API calls from, you know, three companies in The US. It's it's it's just too critical of a of a of a technology. So sovereign clouds have to get built out. GPU clusters are gonna get built out globally. You know, countries are gonna wanna train their own models. Like, that... That's an inevitability. So if that's an inevitability, then you have a choice. Do you want your technology stack? Do want your architecture to be the architecture that powers all that, or do you wanna force the rest of the world to have to go and design their other architecture? And if that happens, who stands the game the most? Obviously, China, because they will just eventually stand up enough enough compute, enough infrastructure, and and make that happen. Now I think this is a very worthwhile debate, and I am only, like like, 80% confident in my opinion. The the 20% where I'm not confident is you could you could possibly squint and argue that America is just so far ahead in compute and chip design, and now we're starting to kinda, you know, figure out how to do fabs.
00:10:51 · Aaron Levie
And you've got Elon with TeraFab, and we we know how to, you know, do the orbital data centers maybe in the future, that that maybe there's one scenario that you play out, and that is that we do actually just have the API for intelligence, and that we control the access to it, and and we can make sure that only our our allies are are using it in the right ways, and we can prevent the bad actors. I think there's a... There's certainly a scenario that that looks like that. That's that's what many of the kinda AI safety kinda ecosystem is is sort of hoping for as as the outcome. And and I I totally grant there's there's some percentage possibility of of that. I am... But but I I kinda default to this is such such critical infrastructure and so important that it will cause other countries to have to just go their own direction if The US isn't isn't enabling that. Yeah. Personally, I'm I'm more on the side of... People should let Jensen win, you know, like, he has a chip, should like sell them everywhere and make that but since you can already now see the downstream impact. So so Jensen was the first to experience this, but but, you know, Anthropic is now this... The next to experience it. It's the same exact thing. So if you already believe in the precedent that you shouldn't export a chip to China or or some other country designing, you know, kinda leading AI, then it stands to reason you probably also aren't gonna let the leading AI model that could help train the next model or help crack systems be available to those countries or or or or, you know, certain certain ecosystems. So so we now have... You know, because of the chip export controls, you you have a lot of the the foundation for now what we're gonna be dealing with. And, the risk is that creates a very unstable environment for for for countries globally. Now, again, as long as we have the best models, you know, maybe you just... You can just, you know, power through that. But it creates a lot of incentive for other countries to to wanna respond in kind.