Dylan Patel: They Don't See It Coming

Now with like Cloud Code, I think we move 10 times the speed of any of our competitors. All the executives, all these major companies watch podcasts like yours. They

listen to these things because it's like, well, this is how I listen to the people that like matter. Hey, if a nuclear missile's heading from China to America, we can use AI to stop it. And Dario was like, well, you can call us. Well, I'm sure we can figure something out. And it's like, like, this is

us. Well, I'm sure we can figure something out. And it's like, like, this is just the dumbest response you could ever come up with. I don't know how principled Sam is, but he's certainly one to take advantage of a good crisis. Wait, I

actually think like UBI is like perfectly fine, which I think is like crazy. Nowadays,

people in the 90th percentile think they're middle class and people in the 50th percentile don't think they're middle class. A buffet of dopamine is less happiness. I actually didn't know that. Open Claw is pretty freaking insane. Even without AI going bad and rogue,

know that. Open Claw is pretty freaking insane. Even without AI going bad and rogue, we know people are bad. Really?

Cooked, cooked. Dylan, round two. Really appreciate it. Yeah, I'm excited for this. Very nice office you got here. Yeah, thank you so much. Just moving in.

this. Very nice office you got here. Yeah, thank you so much. Just moving in.

We've got a fantastic office that we just got moved into. So I want to think back to the last time we talked, which is about eight months ago. It

really does seem like a long time. Bro, eight months ago is three years in AI lab. Might as well be. All right, so you made a bunch of predictions.

AI lab. Might as well be. All right, so you made a bunch of predictions.

I want to go over those first. Oh, shit. Yeah, yeah, yeah. You got a scorecard for me. Okay, you said... GBT 4.5, too slow, too expensive. We'll come

back to these in a sec. I'll go through them one by one. Scale AI

cooked. Junior dev market nuked. OpenAI is your pick to reach super intelligence first. Now we're eight months later and I want to hear, I want to hear the updates. Let's go through them one by one. So first, GPT 4.5, you may have just absolutely nailed that one. What went wrong with GPT 4.5? Well,

4.5 failed because it didn't have enough data. Also, it was just very complicated and difficult on a scaling perspective, infrastructure wise. And there were tons of problems and challenges there. Yeah. Yeah. How are you feeling about that now? I mean, I guess there's

there. Yeah. Yeah. How are you feeling about that now? I mean, I guess there's nothing to say. You can't even use it. You can't even access it. No API.

It was a good model though. Next, ScaleAI cooked. What does the ScaleAI acquisition actually give Meta first? Let's start there. Yes. I think, you know, for one, ScaleAI is like, it's kind of cooked right now. As a company. As a company. Right.

Because everybody's canceling their contracts. Yeah. Google's backing up. How are you feeling about that?

Obviously, Alexander Wang running the Meta Super Intelligence Department division. Still

feeling like scale AI is kind of the orphan child. Yeah. I mean, like they've had a good number of departures. They've had a good number of folks turning off of them or turning to other providers, right? Scale AI is not really picked up a lot on the environment space, right? When you look at the data labeling companies, there are a lot less than there are the environment companies. And in some sense,

it's sort of the same end market, right? Although environments will be bigger and so on and so forth. There's like, 30 RL environment companies. So

scale has not really picked the ball on that. The leaders in that space are different folks. I think they've really missed that ball. But I mean, their business is

different folks. I think they've really missed that ball. But I mean, their business is doing fine. It's just like- That wasn't the point of the acquisition. Yeah, yeah. I

doing fine. It's just like- That wasn't the point of the acquisition. Yeah, yeah. I

mean, who cares, right? It's Alexander Wang. We got to see how Meta's avocado and all that stuff comes out, right? That's the more relevant thing. That's why they acquired Yeah. I mean, it's lucky for them. They're sitting out of the frontier race at

Yeah. I mean, it's lucky for them. They're sitting out of the frontier race at this very moment, but we'll get to that in a minute. Junior dev market nuked.

That was a strong prediction eight months ago. You already see the junior software engineering market is nuked. A lot has changed. What do you think about that prediction at the time? I mean, it's true, right? You know, you look around and

the time? I mean, it's true, right? You know, you look around and the fresh grads are it's even harder for them to get jobs. You look at like quad code spend, right? It's freaking nuts, right? 19 billion of revenue for Anthropic now. All of that is, you know, at some multiple is code, right? And at

now. All of that is, you know, at some multiple is code, right? And at

some multiple is like, okay, if you're spending $20 billion industry is on code, some multiple of that is the productivity gains and value you're getting, right? So, you

know, is it 4X more productivity versus how much you spend? Because there has to be some... premium over, you know, value over the premium that you pay for switching,

be some... premium over, you know, value over the premium that you pay for switching, right? And I think it is, right? I think for us, like, right, like our

right? And I think it is, right? I think for us, like, right, like our spend late last year was something like, you know, 50K a month, right, whatever.

And it was like in the hundreds of thousands of dollars run rate. annually. Then

4.6 came out, then 4.6 fast came out, 4.6 fast with 1 million context.

That's like 12x the cost of 4.6 period. I have people who are not software developers. Literally, one day my head of office was like, guys,

developers. Literally, one day my head of office was like, guys, our run rate of spend is $6 million. This is not sustained. And this was Super Bowl Sunday, by the way. This is not sustainable, right? And it was because one of my engineers spent like $8,000, he's Canadian, so whatever, on Cloud Code, right?

And it's like, oh shit, right? And then I asked him what he built and he explained everything he built and I'm like, oh, this is totally fine. I don't

mind. The spend is fine. And then now you've got adoption from more and more people, right? I've got Jeremy who does our data center modeling He works on the

people, right? I've got Jeremy who does our data center modeling He works on the data center team. He leads the data center team. He tracks all the data centers around the world. He's not been a programmer. We've had other people doing a lot of the backend and all these sorts of things, like the vision model, or contractors and stuff, or going through and scraping all the permits and filings. He just started

using Cloud Code, and he's building his own tools. And he's like, I'm not coding.

I'm just telling you what to do. And it's doing these things. And his daily spend now on 4.6 fast 1 million context is $5,000 a day.

I look at the productivity, it's fine. So I think it's beyond just the junior developers. It's like all, you know, Cloud Code is not a coding thing. It's like

developers. It's like all, you know, Cloud Code is not a coding thing. It's like

for all people. Yeah. I mean, ever since OpenClaw became popular just like seven weeks ago, I think I've spent close to 7 billion tokens on it. And luckily, friends at Cursor kind of supply me with the good tokens. What have you made? A

ton of workflows. Okay, so OpenClaw now is a, First Class Citizen in my company.

It has its own workspace, email address, drive. It is reading all of my emails.

And yes, I have lots of prompt injection defense there. And it's doing outbound sales.

It's triaging inbound sales. It's really cool. But of course, it took so much iteration to actually get there. So that's where those 7 billion tokens comes from. I joke

internally that my company will not exist in two to three years because because AI will make it impossible. And that's like the driving force to like, we have to adopt AI faster, faster, faster, faster than anyone else. And so we keep like taking market share from other people in like various areas. But I just got like a shiver. I'm like, fuck, we're behind the eight ball. Can I tell you what just

shiver. I'm like, fuck, we're behind the eight ball. Can I tell you what just happened last night? So I was putting my son to sleep. He wasn't going to sleep. I was like, hey, you got to go to sleep. I'm interviewing somebody cool

sleep. I was like, hey, you got to go to sleep. I'm interviewing somebody cool tomorrow. I got to get good night's sleep. He's like, oh, who are you interviewing?

tomorrow. I got to get good night's sleep. He's like, oh, who are you interviewing?

I was like, Dylan Patel, who runs this company. They know a lot about AI chips and other people pay them to tell them about AI chips and he goes, thanks for a second, why don't they just ask AI? Cooked, cooked. The

child knows that my business is done for. I wasn't sure if I was gonna bring that up or not, but I thought it was hilarious. No, I mean, if you don't have existential threat, then you're not moving fast enough, right? Yeah, yeah, totally.

I mean, like you look at like open-eyed and anthropic, they're like the furthest ahead, but if you talk to researchers, they're like, a lot of them are like freaking out, like, oh my God, you know, our model's behind in this one capability, and it's going to be compounding growth. That means we're going to lose the race to AGI, right? It's like, okay. Yeah. Let's go straight to it. So obviously, a lot

AGI, right? It's like, okay. Yeah. Let's go straight to it. So obviously, a lot of discussion, a lot happened this last week, Department of War, Blacklist, Anthropic. I want

to get your thoughts on that. There's kind of a lot of nuance there. So

the US government just labeled Anthropic, a supply chain risk. What is your read?

And then literally later that day, OpenAI closes the deal with the Department of War.

What's your read on that situation? And then I got a bunch of follow-ups. understand

the two companies. And by the way, OpenAI reneged on that deal that they did on Friday and signed a new one on Monday, right? With a bit more tighter restriction. But the two companies, there's a lot of interesting stuff, right? You could take

restriction. But the two companies, there's a lot of interesting stuff, right? You could take one view from Anthropic, which is like, oh, it's because Anthropic and Dario won't bend the knee to Trump and won't donate to him. That's why they're getting attacked. Because

if you look at the OpenAI deal, it's like, oh, yeah, it doesn't allow mass surveillance. which is like the main thing, right? Yeah, and autonomous weapons. Yeah, and autonomous

surveillance. which is like the main thing, right? Yeah, and autonomous weapons. Yeah, and autonomous weapons. Although- Fully autonomous. Fully autonomous. I was gonna say reportedly both allow autonomous

weapons. Although- Fully autonomous. Fully autonomous. I was gonna say reportedly both allow autonomous weapons to some extent. You could take one view that's like, it's just like anthropic is being done. The other view is like, Anthropics just being dogmatic, right? There was

a situation where I think it was put out by some press, but it was like Dario was asked, or someone in Anthropic was asked, hey, if a nuclear missile's heading from China to America and we can use AI to stop it, but it requires surveillance and autonomous weapons, what do we do? And Dario was like, well, you can call us. I'm sure we can figure something out. And it's like, this

is just the dumbest response you could ever come up with, right? I don't know if this is like even fully accurate but this is like what was in the media at least this is what everyone's perception is right so there's like two it's like okay the government is and Trump admin are out to get anthropic the other view is like you know anthropic is actually just super dogmatic and I think the

truth is it's kind of like in the middle right and then you look at open AI it's like look I don't know how principled Sam is but he's certainly one to take advantage of a good crisis and That's maybe his greatest skill, is to take advantage of a good crisis. And so he goes out there and he signs deals with the US government. Now the US government is

like, even though Anthropic was the first to sign something with the US government in terms of a exclusive, or not exclusive, but like- The only one with access or implementation in a classified environment as well. Yeah, yeah, exactly. But despite that, Anthropic has- has lost the plot now and opening eye just saw it and they swooped in and now they're going to be the one there, right? Yeah. I mean, it's funny

you say bend the knee. I mean, just, I think it was today it was reported, Dario said the reason they're getting so much shit from the Department of War is because he did not give dictator style praise to the administration. And, you know, obviously David Sachs has tweeted a ton about Anthropic and Dario and their politics. So

how much of this do you think is politics at play? Yeah, I'm just curious.

I mean, you also have to recognize that Anthropic has a lot of their policy people are former Biden admin people. Right. And so it's like, whereas, whereas opening eye has hired people from both parties. Right. Which is the pragmatic thing. Right. You

go look at like. Microsoft, right? Who is like known for navigating the government. Well,

they have people from both sides in their policy teams and such, right? You go

look at like any major company like Amazon who has government contracts or Lockheed, they have people from both sides. Right. And it's like, you know, anthropically, maybe, maybe just like, you know, you, you to get to your end mission, right? Your end goal, unfortunately requires like being a slithering snake and like, you know, playing both sides and like the political, you can't just be a dogmatic person. Right. And so, Or company,

right? But that is their DNA. That's their strength. That's why they're winning, right? At

right? But that is their DNA. That's their strength. That's why they're winning, right? At

the same time, right? Is them being so dogmatic about the things they believe and care about. That's why they're winning, right? So I think it's a tough one. The

care about. That's why they're winning, right? So I think it's a tough one. The

more interesting question to think about is like, what happens from here, right? Especially leading

up to the IPO? Yeah. Being designated a supply chain risk? Like what it, yeah, like what do you think? How does that affect their business? So, you know, OpenAI did sign the deal Friday. All weekend, people internally were like rioting. And Monday, they like pulled back and signed a different deal. The Anthropic is basically like, well, we

exactly asked for this and you guys said no. But it's sort of like, you know, this shows, right? If you tell the government, you can have everything and slide a little bit back versus you can't have anything except for these things. It's a

different argument. And so when you look at like, hey, what is the end state the U.S. government wants? Um, to be viewed as the winner, right?

the U.S. government wants? Um, to be viewed as the winner, right?

As the king, right? They don't want to be the one that's like, oh, thank you, benevolent corporation for giving me this. And at the same time, you know, the AI labs are like, well, no, we made this technology. We have complete rights over it and like what you can and cannot do. And it's like, you know, if the US government wants to go full dictatorial, they can just use the Defense Prioritization

Act and get whatever they want from you, right? They can just go get 4.6's weights if they really want it. Right. They could do a number of things to have access to everything Anthropic has. So it's not like, you know, Anthropic is even stopping the US government in that case. Right. Like from having access to things.

And then and then the question is, OK, how do you go forward from here?

Right. The US government can press harder. Anthropic can stand up. But Anthropic can't just back down and kiss the ring immediately. Right. They are they are so dem coded.

They are so not even dem-coded, right? Just like so dogmatic in their beliefs. If Dario and leadership were to say, okay, US government, you get this, people below that would leave, right? Or people below that would riot, right? OpenAI

doesn't get that, right? You know, sort of, you know, you've got a lot more let's say, Zealitz for the cause at Anthropic than any other company in the space.

The only way to make these people understand and back down is to be like, look, the US government's forcing us, right? I think there's a bit of like chess that, you know, even if Dario wanted to accept the US government's proposals, he kind of can't because then a lot of his best researchers will like be pissed off and leave. Maybe that's the risk you have to take. But also the other one

and leave. Maybe that's the risk you have to take. But also the other one is like, you could just be like, okay, let's see the US government push us harder and harder and harder. And then when we back down, You can tell them, well, we had an ultimatum. And it was like either they just take the weights and we do nothing or we work with them. And yes, we make the killer

robots and we do mass surveillance, but we do it with clawed alignment rather than them having the open weights and doing whatever the hell they want. So this is the pragmatic thing to do to prevent AI from killing us all because our killer bots will be better than OpenAI's killer bots. And then everybody will be like, yeah, you're right. Thank you, Dario. So I feel like that's what has to happen. And

you're right. Thank you, Dario. So I feel like that's what has to happen. And

at some point, Dario does just have to be like, I love Trump. I love

everything Trump says. I love the Department of War. I mean, they're really the only tech company that hasn't bent the knee. Every other tech company has bent the knee, has donated. So it's kind of an interesting position. And it does. Kind of scary.

has donated. So it's kind of an interesting position. And it does. Kind of scary.

Yeah, yeah, a little bit. OK, so I want to, one more thread on this.

So Claude is currently the only model currently deployed in classified military networks.

obviously we're just announced trying to rip Anthropic out. How do they do that within six months? Is that even possible? I don't see why not. I mean, I think

six months? Is that even possible? I don't see why not. I mean, I think the version of the model that they have is like, it's like 3.5 Sonnet or something like that. It's like 3.6 Sonnet. Yeah, it's like an older model that they have deployed because it's the weights, right? It's like they have access to those weights.

I don't think it's like they have Opus 4.6 deployed in classified networks, right? Yeah,

max fast. Yeah, no, no. I mean, I believe it's like an on-prem like thing.

Okay. Yeah, I guess it would have to be, right? Yeah. So in that sense, it's like, sure, get OpenAI to give you, you know, GPT 5.5, GPT 5 weights, right? That's a better model, right? Or, hey, Grok, give me 4.1, Grok 4.1, 4.2 weights, right? Like, you know, there's a lot more, I mean, like, Chinese models are better than Claude Sonnet 3.5, right? You see the government using this?

No, no, absolutely not. But the point being that there's plenty of sources for that capability tier. And so I think it's fine. I don't think it's that hard to

capability tier. And so I think it's fine. I don't think it's that hard to rip it. I mean, I imagine it won't be that hard to rip it out,

rip it. I mean, I imagine it won't be that hard to rip it out, right? But who knows? It's kind of wild to think that they're using such an

right? But who knows? It's kind of wild to think that they're using such an old model for such important work. I actually didn't know that. And now I'm thinking about it and remembering the level of hallucination coming out of those early models. and

it's deployed. I also remember 335 Sonic coming out and be like, holy shit, this is so good. I see the sparks of AGI. Yeah, yeah, yeah. No, I mean, that's fair. That's fair. I just think it's kind of, there's a lag, right? Like,

that's fair. That's fair. I just think it's kind of, there's a lag, right? Like,

you know, government is just slow. And this is like, I think the biggest risk that also like will help anthropic people understand and be convinced that fine is China's not using, you know, the oldest model. They're using, you know, The newest QAN. They're

using the newest Kimi. They're using the newest DeepSeek. In their military applications, there is no waiting around. There's no slow timeline to deploy. It's shipping.

And so maybe the US has a six-month advantage, but if the US government has a six-month time lag between the new model coming out and deploying it, then there is no advantage for the US government and military. And then if China is able to have all these drones that they can manufacture at high volumes and control these autonomous drone swarms with very advanced AI that is on par with what the American

military has, except the American military doesn't have drone swarms because we can't manufacture drones at high volumes. It's like, wait, our military capabilities are actually just worse because they have lower cost, more drones with similar intelligence. It's like, where is the advantage for us here now? Where is the advantage? I mean, we have companies like Anduril, but they're still... Coming up, it's still nascent. So what is

they're still... Coming up, it's still nascent. So what is the advantage then? Exactly, right? I mean, there isn't one. And so that's the biggest risk and challenge in that type of like, if you think like drone swarms are the end all be all of warfare, right? There's other things, right? That potentially,

you know, the US has. I don't know what the US military has, to be honest, but like, could be. All right. Yeah. Yeah, you got me thinking on that one a little bit, Dylan. What you thinking about? Yeah, no, just it's I think I think the craziest part is I realized that, yeah, the U.S. military has these old models and that makes it equivalent to what the Chinese military is able to

have. And it doesn't have to be right. Like, no, we can figure out how

have. And it doesn't have to be right. Like, no, we can figure out how to do faster. Right. Like, yeah, I think I think like, you know, flip side is like the U.S. government. buys data from Google, from Twitter, from Facebook, Meta, sorry, on their users. And then they have their own classified sources of information. And it's like, right now, building all the data pipelines to mass surveil everyone

information. And it's like, right now, building all the data pipelines to mass surveil everyone is actually just hard, right? We haven't done it as a country. I don't think because of the law. I think because it's hard, right? I think the NSA would absolutely do it if they could, but they just like, it's a monumental task, whereas China has, right? Whereas like, and other countries have, right? Or some other countries have

tried and gotten along the ways, but like China especially is the closest to that.

But like, if I look at how easy it is to use Opus 4.6 to like build data pipelines and, you know, transform data and scrapers and all these things, I feel like doing mass surveillance, building a mass surveillance system with Opus 4.6 would not be terribly difficult, right? Like, you know? And with Opus 5 or whatever, or, you know, OpenAI's new model that they're releasing, like, you know, like, you know, maybe

it's even faster and faster, right? So it's like, you know, these autonomous, like this, like mass surveillance, which is like one of the red lines in the sand is like, it's a true moral quandary, right? Because I don't want the US government to have, you know, China-like mass surveillance or better, right? But at the same time, if you give them full unfettered access to, Opus 4.6, they can totally build this.

So you think Dario's line in the sand is coming from a true place of belief and morals? Look, if you think AI will kill, the extra risk of AI killing us all is pretty high. And it's because AI turns around and uses weapons to kill us all. Autonomous weapons systems obviously make sense to limit. But also because

if there's autonomous weapons systems, who gets to command these autonomous weapon systems? How are

they aligned? Like all these questions should come up, right? Fine, that's one. The other

one is like, even without AI going bad and rogue, we know people are bad, right? Generally, like in persistence of power. And if you create mass surveillance systems for

right? Generally, like in persistence of power. And if you create mass surveillance systems for the American public, you know, what does that now do to free speech and all these other things, right? You and I can come here and we can say whatever we want. You know, we could say fuck Biden or fuck Trump, but like, you

we want. You know, we could say fuck Biden or fuck Trump, but like, you know, like whatever, right? probably nothing gets flagged. But now if you have like these scrapers that are pulling every record of people on videos and like, you know, it's like, oh, somebody's account on YouTube, which is linked to this email or like based on these comments is based out of this area and based on this other information

is based here. And as this person, now all of a sudden this person said fuck Trump after I said fuck Trump on some random podcast, right? Or fuck Biden after I said fuck Biden on some random podcast. Now all of a sudden they're being surveilled. And now, you know, can that be turned around? So there's, I think

being surveilled. And now, you know, can that be turned around? So there's, I think there's like real risk to like, you know, mass surveillance, right? It's not that it's like trivial, right? Like, I do think the beauty of America is that we can do whatever we want, right? Generally, and AI systems make it very, very likely, even pre-AGI that we can't do that anymore, right? That bad

actors can now manipulate systems to control people, especially as social unrest grows and grows, right? AI is causing social unrest to grow, the power of capital versus labor. Which part of AI is causing social unrest? Is that the misinformation?

Is it just the politics around it with energy? And I know you talk about water a lot. What part is it? I think it's all of the above, right?

I think generally over the last, what, 50 years we've had capital take more and more share of value versus labor. We have seen social media amplify what other people have versus don't. And so even though 50 years ago, rich people lived or middle class people lived worse lives than middle class people do today, I think on an objective basis, at least in terms of like, what is the square footage

of their house and how much meat do they eat and what is their access to medicine, right? Because the perception is not based on that, it's based on historical thought process of, oh, we own our own home and we have a car and we can send our kids to college and blah, blah, blah with one wage, even though that was actually very rich 50 years ago. And still possible today, but you

still have to be very rich. Or flip side, right? Like, hey, I'm a middle class person or I'm even in the 25th percentile. There used to be a belief that people in the 20th percentile thought they were middle class. And people in the 80th percentile thought they were middle class or didn't think they were middle class. They

knew they were well-to-do. Nowadays, people in the 90th percentile think they're middle class and people in the 50th percentile don't think they're middle class. Because it's the perception game of like, oh, well, look at these people, they're going on vacation. It's like, yeah, because they're putting a fake life on social media, their peers, right? That's not what their actual life is, right? And then on the flip side, it's like people, you

know, see influencers see like, oh, that is the norm, right? And it's like, I don't have that because I'm poor. Well, no, you're middle class. And like, what the amount of vacations and experiences you go on is actually more, you know, I think like, obviously it's like, It's very privileged for me to say, like, lives are better, at least, especially for me. But at the end of the day, on many objective

economic measures, lives are better, but people think their lives are worse. And now, more and more, over 50 years, because it's continued to be a slide of both capital taking more value, income inequality growing, this perception game of social media, and these things keep amplifying, especially in the last decade, social media has really amplified it. And now

you have job loss, right? potentially happening. And you look at like, you know, economically, usually people have been fine when GDP is like 2%, which is like, you know, fine what it is now, except now of that 2%, 1, 7, 1, 8% is AI growth, right? It's like, okay, who's getting the benefits of AI growth? It's

like electricians, construction contractors, people in the semiconductor industry. Capital allocators.

Yeah, capital allocators. That's it, right? Like it's like, That's less than 10% of the economy, guys. So it's like most people are not experiencing economic growth and they see

economy, guys. So it's like most people are not experiencing economic growth and they see this inequality and they look for boogeyman and the boogeyman is the richest companies on the world, the stock market soaring and AI, right? And all the risk of AI taking jobs, which is about to happen in drugs, right? Waymo's about to start deploying all these cities. RoboTax is about to start deploying all these cities. The boogeyman of

a few million people losing their jobs because of self-driving is gonna happen, right? And

in addition to that, white collar work, right? Like, you know, the amount of work you know, my company can do with how few people has the amount of work that you can do with how few people you have would not have been possible.

And so we're able to build an enterprise, but like we're, we're enterprising individuals. What

about all the people who aren't enterprising? They just work a job and then they get like, you know, they, they see this tidal wave coming. Um, you know, it's, it's, it's very scary, I think. Right. And so, yeah. Do you feel a lot of anxiety in the city right now, even with engineers? It's interesting. You say that I tend to be more optimistic, um, I think maybe there's going to be more

smaller companies, but you're right there. There needs to be enterprising individuals to go out there and actually start these companies to build value because we're not going to have these massive hundred, 200,000 person companies anymore as you know, block just laid off half their workforce. So are you, are you generally pretty pessimistic with the, you know, what

their workforce. So are you, are you generally pretty pessimistic with the, you know, what Dario called the white collar bloodbath? Um, I, I think generally there will, you know, even though there's more surplus in the economy, the allocation of that surplus is harder to get to the people. And so there will be, and markets take time to

reorganize. So there will be a hard dislocation, right? Yes, there will be new

reorganize. So there will be a hard dislocation, right? Yes, there will be new jobs, more things for people to do, more service than ever, right? Like, you know, more experiences, right? Most people will then be able to go and do like random, more painting classes or more whatever classes or more like park, whatever, more beach, more surfing, like whatever it is, than ever if the surplus was spread evenly and people

were able to just reallocate perfectly. But that doesn't happen. And also the technology is moving so fast and the ability for society to keep up with it, you're saying, is certainly not fast enough. Right, exactly. Like the whole like, oh, but you know, we had 90% of people working in agriculture and now it's like less than 2%.

And it's like, you know, it's like actually less than 1%. It's just like people claim their farms for tax reasons. And so it's like less than 1% in reality.

Anyways, like it's like, well, yeah, but that took 100 years, right? It took more than 100 years, maybe even, right? And so like, whereas like this AI dislocation is like immediate, right? And so probably I'm generally an optimist. I don't have anxiety about this, but I know a lot of people do.

optimist. I don't have anxiety about this, but I know a lot of people do.

Both normies across, the space, right?

Like, you know, met some of my cousins, met some friends that aren't in the space and they're like, they're more worried about it. You know, met, met, you know, recently, right? My Uber driver recently, I still take Ubers because I think it's fun

recently, right? My Uber driver recently, I still take Ubers because I think it's fun to be able to talk to an Uber driver versus a Waymo. You tell them that their jobs are about to be complete. No, I don't say this stuff. I

just, I just vibe, right? My philosophy here is actually insane, right? So Waymo is objectively the median, Waymo is better than the median Uber, right? But like, you know, that's boring, right? Like, you know, Ubers, I would say 90% of Ubers are worse than a Waymo, but there's that 10% that has some interesting music, some amazing conversation.

You know, I practice my Spanish, right? Whatever it is, and there are good, fun conversation that spikes because of the human connection. It's like, okay, this is worth it.

But I think in general, right? Like that way, I had a Waymo driver, right?

Soviet, grew up in the Soviet Union, moved here, was a taxi cab driver for 30 years. Um, He started ranting to me about how Uber doesn't pay taxes and

30 years. Um, He started ranting to me about how Uber doesn't pay taxes and has never paid tax. He started ranting to me about how much he used to make, doing taxi driving and now how little he makes. And in my mind, it was like, well, yeah, that's like the commodification of driving for someone has been a surplus for everyone in the economy. Obviously, that hasn't been distributed evenly, right? People take

more taxis than ever because Ubers are so cheap. But at the same time, he started ranting about how Uber You know, he's like, yeah, people don't need all this technology, right? You know, most people don't need all this information at their tips. They

technology, right? You know, most people don't need all this information at their tips. They

have it, but it doesn't actually benefit their lives. They'd be better without it. And

like, he just kept ranting about it. And I was like, wow. I asked, I just asked him how his day was. What were your thoughts on that point? That

we have enough. We don't need more tech. We don't need more info. Like, I

mean, who am I talking to? What am I saying? Yeah. I love it. Yeah.

Give me more slop, right? Like, you know, like, I think, I think, I think it's fair. Like, you know, I think people, people's, you know, dopamine receptors are

it's fair. Like, you know, I think people, people's, you know, dopamine receptors are short-circuited and there's probably some way to make it better. I mean, you guys put your, your semi-analysis charts on the, like the, what is that game? You make it good for Gen Z so they don't get distracted. It's a bit of a shit post, but yeah, yeah. So what we do is we do a Subway Surfer. Yeah,

that's it. That's it. It's Subway Surfer. And then you like explain an ML paper over the top and then you also add some background music. It's just great. So

funny. It's slop. I think, I think dopamine is short-circuited for everyone. the reward circuit um probably a lot better would be done if the world had like you know not you know if the average person had a longer attention span um could focus more um i think generally a buffet of dopamine is less

happiness yeah um i personally you know it's like type one versus type two versus type three fun i personally think that like the pain and the grind Bringing you know, like the reward is the most fulfilling thing in life. Right? And I'm sure you feel the same way like when you're like, you know, you you're put your kid down instead of going to sleep like most normal people you're probably like working

on some something right and it's like but like then you like you're like wow that was fulfilling right so that's that's like but like a lot of people don't get that right? Because they go to work they come back and then they dopamine you know, and And so it's like sort of like, you know, there's something about like, hey, human psychology and what is the reward mechanism of a human brain? Like

probably we could change, but like you shouldn't like, you know, at the same time, technology is like improving agricultural yields, right? And like people aren't dying of world hunger and like, you know, technology is making steel production faster and like technology is making automobile production better and like batteries better. Like all these things that make normal people's lives much better, right? You know, we have more abundance than ever because of technology.

So you can't stop that, right? Like otherwise like degrowth is like the silliest like ideal ever, right? Like it's especially popular in Europe, but like degrowth means that like people, people, yeah, it's just like a terrible idea. Okay. So we, we, went on a super tangent there, which I absolutely loved it, but I want to bring it back to something you said earlier. You were talking about Cloud Code being, especially 4.6,

being able to build out incredible data pipelines and transformations. And let's talk a little bit about software, Cloud Code eating software. And in February, in your newsletter, you said Cloud Code was at an inflection point comparable to the ChadGPT moment.

You mentioned earlier that your token spend is absolutely skyrocketing. Although you're like, yeah, that's fine. The returns are there. Walk me through what you think is happening right now

fine. The returns are there. Walk me through what you think is happening right now with CloudCode. A lot of people are liking Codex. The whole software industry is changing.

with CloudCode. A lot of people are liking Codex. The whole software industry is changing.

What are your thoughts? I mean, like, yeah, like Kerser, their revenue went from a billion to two billion in like a few months, right? It's not like CloudCode's taking all the lion's share. You look at Codex, they've eclipsed a billion after having launched not so long ago. They're probably closer to two billion now. Like, you know, Anthropics skyrocketing, of course. I don't think it's just a cloud code. I think there are

many people who are tools that are adopting and getting strong coverage. A lot of people perceive these tools as a coding thing, and they aren't, right? I think

if you take a long look at them and you take a long look and you force yourself as a non-programmer to use these tools, you can get huge productivity gains because this is you know, when you talk about like, hey, what is chat GPT like versus like, what are these like? It is a completely different like, Claude code, cursor agent mode, codex, these are agent orchestration systems and

you can get them to do anything, right? You talk to them in natural language and they go do stuff. And you can, sure, you can make them make software.

And so right now Claude is the best at this, right? But soon OpenAI will release their new model and that new model, will not just be, you know, the reason Codex 5.2, Codex X High or whatever the hell the name of the model that currently exists is, is so good, it's better than Opus 4.6 at coding. It's

worse at everything else because they took the garlic model and then they just did RL on only the coding line, right? Whereas the model that they're going to release soon is the pre-trained, you know, garlic model, but then they RL'd it for everything, right? And what's relevant here is you know, tool use outside of programming domains, ability

right? And what's relevant here is you know, tool use outside of programming domains, ability to understand things outside of coding domains. And so when you think about what is an agent orchestration system and what am I, you know, like everyone's like, when's the year the agents? It's now, right? Like, because Claude Code Opus 4.6 is, the agent orchestration system, right? And you tell it and it spawns these agents and it

does these things and it comes back with work. It makes charts, you tell it to learn skills. And then these skills are like, you know, it's like just yesterday I heard the silliest one from a hedge fund colleague, right? I think this one's super funny. He's never programmed in his life. He was told, he started using Cloud

super funny. He's never programmed in his life. He was told, he started using Cloud Code after our article on Cloud Code and Doug at our company who is chief Cloud Code, he's the president of the company. So he's like partner basically, but like, you know, he's so Cloud Code psychosis That he's just ranted to people about it.

So this hedge fund client goes, well, so what I did was like, you know, one of the things that I've been trained to do is listen to earnings, read earnings transcripts, and really digest the wording that they use. And based on like how they speak and their tones and their inflection, know when they're bullshitting versus not, or know when they're exaggerating versus not, because this can lead to outsized returns. And

so there's entire funds that have been started on this premise and been successful, which is just, looking at the tone of how people talk. And so he built a skill within Cloud Code. He sent it these books from this CIA Negotiating Tactics and learning tone reading and language reading. So he sent Cloud Code all these

skills and made a new skill within Cloud Code, which was understanding this. And then

he fed it some earnings transcripts that are famous in the past of having tricked most people, but then actually have like had extreme alpha just from the wording or the tone rather than the actual words they said. And then the skill was then able to go through all earnings and he's like using this to like, it was like, wait, he didn't code anything though, right? What did he code? He told Claude

Code to learn and read these books and then understand how to like tonally and like, you know, all these, right? So it's like, you don't have to be a programmer, right? This is all knowledge work. You know, if you know how to do

programmer, right? This is all knowledge work. You know, if you know how to do it, if you can be descriptive, The model can learn, right? Can get a skill and can do it, right? So today, Cloud Code is the best at this, but as you step across the world, it's gonna be so many other places, right? As

I mentioned, right? You know, we're doing, you know, Jeremy's doing it for data center stuff. Doug's never been a programmer either. He's a hedge fund person before, right? So

stuff. Doug's never been a programmer either. He's a hedge fund person before, right? So

he's doing it for, you know, so it's like, you keep stepping across the world.

It's like, there's so many places where, you know, you mentioned you're not a programmer, but you're using CloudBot for all these things, right? So there's, no reason to think that Claude code or codex is for code. These things are agent orchestration systems for, and this is how you interact with AGI. Now, maybe you

actually need, like normies will need a GUI to work with this stuff rather than like CLI. Well, what is cursor? Yeah, but I mean, even that's an IDE, right?

like CLI. Well, what is cursor? Yeah, but I mean, even that's an IDE, right?

I mean, codex UI is very, kind of simplistic and perfect for the average user.

I mean, I don't know. I think I've forced people in my company to use Cloud Code, and they're like, oh, terminal scary. They're like, I'll wait for co-work to be good. And it's like, no, no, no, you're going to use Cloud Code. And

be good. And it's like, no, no, no, you're going to use Cloud Code. And

then they force, they try, and then like, wow, this is amazing. This is so good. My point earlier, by the way, so I have been a software engineer

good. My point earlier, by the way, so I have been a software engineer for a while, but I am not actually writing code nowadays. It's all natural language.

It's all vibe coded. I don't have time to actually like read all the lines.

Plus it's all software for myself and my company. So it's not a big deal if there's some issues here and there. But I think you're right. It's interesting that you can basically build anything at this point and anybody can do it. there are

still some rough edges. And so if you vibe code long enough, you start to realize where those rough edges, where those limits are. I think a lot of people aren't using the million context cloud code though. They're using the 200K or whatever. Because

the plan that they subscribe you to, right, is capped context. But if you use the like million context, which is only available via API that you have to pay out the ass for, it's like way better. But I agree. I agree. There's rough

edges. I think what it is as well as like in the past in major companies, you know, leadership or like, you know, team leads, you know, people who are doing the actual work of the company would like create all these specification documents, blah, blah, blah, blah, blah. And then they'd get another team to implement it. And it's

like, all of that is gone. The person who actually just understands the domain can just describe what they want, look at it and iterate, iterate, iterate. And it's not programming there. It's business logic being encoded. Right. So, so Let's continue on this. Let's

programming there. It's business logic being encoded. Right. So, so Let's continue on this. Let's

talk about SaaS in general. I think we talked, you know, obviously there's a lot of talk about SaaS is dead. And I think where the world is going is probably agents on one layer and then the layer below them, some kind of file system, some kind of CRUD database. Where does that leave everybody else? Like if those are the only two things, right? Satya famously said the entire application layer is collapsing

down into agents. So agents, file system, where's everybody else going to be playing? Where's

the value going to be captured? That is a tough one. That is a tough one to answer, describe. I think it's a lot easier to say who's a loser than it is a winner today in software because of the pandemonium. But

as far as who actually wins, I actually like...

I don't know if this is non-consensus or not, but I actually think Databricks and Snowflake are reasonable angles, right? Because these are scalable data and compute engines, right? Versus a lot of things are like, these are easily replaceable, at least for

right? Versus a lot of things are like, these are easily replaceable, at least for the next super short term, right? Let's say by the time Opus 5 or Opus 5.5 comes out or whatever, you know, GPT-6 comes out, maybe even that is like easily one-shotable, right? Purpose built solutions for, or Vibe-coded solutions for these things.

And so I, you know, the scalable aspect, this is the problem that Vibe-coded things have is they're not scalable, right? So how do you have the agents build something that has hooks into stuff that is scalable, right? And so one of the, like, One of the guys in my company is like, he's basically, he was our head of like, you know, data and all this stuff. And then the

like sort of cloud code apocalypse happened. And now all he does is run around and like help people's vibe coded things be scalable, right? Because everyone else is like, They're like, I don't know what CRUD is. I don't know what that is. They're

like, you know, like, he's like, it's OK. I'll just change this out. And now

all of a sudden, the solution that they've I've coded with like 10% more work from someone who actually knows how to build scalable systems is scalable. And it's like plugging into various layers of software that exist already. But it's not application software. It's

sort of like the scalable infra software. Are you seeing a lot of the roles inside your company change? Because even with my small company, I hired a researcher six months ago. Now, Cloud Code, OpenClaw does it for me. All of the research

months ago. Now, Cloud Code, OpenClaw does it for me. All of the research puts together entire outlines of video. Did you let them go? No, they're busier than ever. Yeah. Absolutely. You know, because enterprising, super sharp, willing to learn, willing to

ever. Yeah. Absolutely. You know, because enterprising, super sharp, willing to learn, willing to adapt. And he is busier than ever. He is building out. different tools, different

adapt. And he is busier than ever. He is building out. different tools, different websites. I mean, you know, so no, didn't let him go. We have plans to

websites. I mean, you know, so no, didn't let him go. We have plans to hire, which is why I flip flop pretty often, although overall optimistic about the future of white collar work. So I just wanted to drop that. I'm the same way.

As AI has helped us, like we already, we always ran like my company's average age is like 30. I'm 29. We have a bunch of people who are super young. Right. We move faster than everyone else, I think, in this space of consulting

young. Right. We move faster than everyone else, I think, in this space of consulting and research and data services. And so I think that's why we win. But it's

been a compounding thing with AI. Initially, I was just moving faster, and so we were winning. And then I hired, and we were all moving faster together. But then

were winning. And then I hired, and we were all moving faster together. But then

AI started coming out, started being more integratable. We started moving faster and faster. Now

with Cloud Code, I think we move 10 times the speed of any of our competitors. It's actually insane. And so my plans of hiring business we win and the

competitors. It's actually insane. And so my plans of hiring business we win and the business we're able to do is actually just like growing right because there's this like dislocation of like what can ai help us do today versus what can everyone else do and so there's an arbitrage in the value of like what we're able to take because ai exists right because we're just skating ahead of the puck you're describing

javon's paradox right there's more use cases because you're able to leverage ai so much more effectively it's becoming cheaper you're you're implementing it in more places but at the same time Correct me if I'm wrong. You are actually still hiring. We're hiring more than ever. More than ever. We have like 20 positions open right now. Right. Like,

than ever. More than ever. We have like 20 positions open right now. Right. Like,

you know, we've hired like, what, 10 people this year already? And we've got like 20 positions open. Yeah. So it's like, you know, we're hiring more than ever. But

I think the flip side is when I look at my competitors, I'm taking their revenue, right? And do they now need to do layoffs? Right. And in fact, we're

revenue, right? And do they now need to do layoffs? Right. And in fact, we're doing, we're taking revenue in a way that is way more because I pay my people more than competitors do. And we do it with less people, right? And so

sort of like there is like this like arbitrage of like, okay, well, like the competing company in XYZ space has way more people than us, right?

And so like now as we take revenue, now, obviously, my space is just one that's growing overall. So more so they're like flattish and we're taking all the growth rather than, you know, we're taking the growth and they're shrinking. Are you taking on use cases and new business lines that your industry historically has not touched? Is that

maybe part of what's happening and why you're growing so quickly? Oh, yeah, for sure.

For sure. I mean, like there's growth in the areas that we already cover, right?

All these aspects of semiconductors, data centers, et cetera, et cetera, et cetera. Right. But

we're also like inventing new industries like tokenomics, right? Which we've sort of, we started the beginnings of it like two years ago, three years ago, the economics of tokens, production of inference volumes, like sort of like cost of tokens, all these sorts of things, usage of AI. There's that, but then there's also existing industries like energy modeling. It turns out energy modeling is like a billion dollar a year data services

modeling. It turns out energy modeling is like a billion dollar a year data services business. And now you could just go expand into that because you have all the

business. And now you could just go expand into that because you have all the tooling necessary. You know, lack of a better word, vibe code your way into it.

tooling necessary. You know, lack of a better word, vibe code your way into it.

Some level of vibe coding, some level of, we already have a lot of customers in the space that use us for data center stuff or AI stuff. And so

we have some in into the sales side, right? We have a name brand and recognition, right? So sort of like, and then we have the like capital to spend

recognition, right? So sort of like, and then we have the like capital to spend on hiring some great people and vibe coding things that, you know, people wish they could have built because we're building the solution that like is needed for now instead of the thing that worked two years ago, right? It took so long to build.

So I think like, you know, that's also an example of this is right. It

was like energy modeling, I think, and I hope that we're going to just dominate it because all the other companies in the space are just moving too slow. Right.

And they have like all these archaic old practices and built up crud, right? Crap,

not crud, like the anyways. Yeah. I think there's a bit of like both, but it's like, By and large, I'm hiring and I'm hiring super fast. If anyone's like an enterprising individual who wants to work really hard and talk to the great biggest companies in the planet and like work on their stuff, great. But like, and like help consult for them. But like at the same time, I fully recognize we're taking

a lot of share from competitors, right? The airwaves like of like, you know, especially like, you know, content creation, I think is the same way, right? We're in an age where the cost of creating content has fallen and fallen, fallen, fallen. And so,

Objectively things splinter, right? You know, we're, you know, uh, what is it? It's not

Pax Americana. What's the American monoculture? American monoculture meant that everyone had seen the same movies. And so movies like could be blockbuster always. Right. Um, and, and this, this,

movies. And so movies like could be blockbuster always. Right. Um, and, and this, this, this spreading of like what people watch. Yes. People watch more stuff. Um, and yes, there are, the peaks are higher than ever. Right. With regards to the biggest movies, um, arguably, but like Titanic and like gone with the wind were more, anomalies would be complete anomalies today, right? And so like- Those can't happen anymore, is

what you're saying. I mean, and at the same time, right, it's like, if cost of content creation is going down, the number of content, amount of content out there is growing, the fact that your channel is growing, And so rapidly means you're eating more and more share from other people who can now no longer make a credible living. Now, does that mean that's Hollywood or does that mean that's some like, you

living. Now, does that mean that's Hollywood or does that mean that's some like, you know, CNBC podcast? Or I don't know what market people are shifting away from. But

what I do know is that all the executives of all these major companies like watch podcasts like Dwarkesh's and yours. And like they listen to these things because it's like, well, this is how I listen to the people that like matter, which is like crazy. It's traditional media. I'll just say it is traditional media that's getting eaten alive right now. Right. And so how many more people work in traditional media

that you're hiring is irrelevant? Yeah. My six-person team puts out so much content compared to what a traditional media company might. And engaging content. That's good, right? Not like

slop, like watching CNBC all day, right? Yeah. You know, their viewership is terrible. I

was looking into it because- I think that's the same across a lot of traditional media. It's like all just a bunch of polish. We all thought, right? It was

media. It's like all just a bunch of polish. We all thought, right? It was

the same when the late night hosts got canned and everyone kind of up in arms and then they realized, oh no, wait, they actually have terrible viewership. Obviously there's

a political angle to it, but if you're bringing the eyeballs and you're selling ads, they're not getting rid of you. Yeah, exactly. And I think like the viewership stats on like, you know, again, like people have this allure in their mind of CNBC, you know, Jim Cramer, Money Talks, all these other stories. Like Jim Cramer pulls eyeballs, but like the rest of the shows don't. And if you look at any of

the other shows, it's like their viewership average is like 100,000. It's like, wait a second, you put out videos regularly with 100,000 views with pretty high engagement, right?

Whereas there's like- At a fraction of the cost. At a fraction of the cost.

Whereas their 100,000 is that is that just the TVs on? Right? You know, it's like, it's like actually insane. Yeah. And so, you know, we'll end up seeing, but like media, media is, is, is the same way. I think, I think every industry is like, great. You and I, and any enterprising founder who is taking share, making

revenue, creating whatever it is, wins and they'll be hiring. But is that hiring going to supplement and make up for more than the people that are being left behind? What do you think? No. No. now yeah what happens you know i've

left behind? What do you think? No. No. now yeah what happens you know i've always been such a capitalist in my life i grew up in a small business grew up in the motel that my parents owned my parents and my parents and then my mom's brother and his wife owned that motel we lived there we worked there etc um worked in a gas in my dad's gas stations as well when

i was a kid right like you know sort of like as gas station um and so sort of like always been a capitalist right you know because like that's what i grew up in as a small business um Over the last couple of years, I've realized, wait, I actually think UBI is perfectly fine, which I think is crazy because, again, I'm very capitalist.

No, I understand that. But I mean, a lot of people- I mean, what else do you do? Society is going to rip itself apart. In fact, I believe politically the next election is going to be so, so AI focused. And the Democrats will just win because they're going to be the anti-AI party. Because right now- You think the sentiment is that negative right now? The sentiment is already more than half

Americans have a negative view of AI. And as we fast forward to the end of this year, as Anthropics revenue goes from 19 billion to maybe 60. And opening

eyes is similar, if not higher, right? You know, the amount of jobs that gets supplemented, the amount of change that happens in society, right? the stock market's going to get really impacted as Google and other companies like Google next year, they've produced cash flows for the last 20 years and of ridiculous amounts, $100 billion of cash flow a year, right? Google will have no cash flow next year because they see AI

so clearly and they know that they need to spend every dollar they make on compute, data centers, energy, Waymo, et cetera, et cetera, et cetera, because the returns for that long term are going to be insane. But that's going to nuke the market, right? Like, you know, like people are like, wait, you know, these companies don't make cash anymore. And people are going to see, you know, job, you know,

there's a crowding out of everything of resources, right? Electricians and plumbers. are

demanded in massive volumes because setting up data centers and the liquid cooling for them is so crazy that now you crowd out other areas and industries. Same with construction, same with like, you know, all these energy costs are gonna probably not go up a ton, but they are going up like low single digits, mid single digits. And

then there will be other market factors that cause prices to go up as well, which then will be blamed on AI. Like anything and everything will be blamed on AI. People will hate AI. And the Democrats haven't turned to be anti-AI yet. But

AI. People will hate AI. And the Democrats haven't turned to be anti-AI yet. But

if I was a Democrat strategist, I would become anti-EI as hell just to win the election, regardless of what the opinion is, because I have a jaded view of politicians. They say anything to get elected. And sort of the incumbent Republicans can't really

politicians. They say anything to get elected. And sort of the incumbent Republicans can't really say they're anti-EI because they're the ones that are pro-business and signing all these deals.

And like, you know, they can't just turn around and be like, we're anti-AI. Even

though you would think like- Well, they could do so from like the censorship perspective, the, you know, the companies making the models lean very left, right? I mean,

that has been said in the past. It doesn't matter. No? Less so than the fear of AI taking jobs. And that's kind of more the left side. Yeah, I

mean, like, and it could be like an establishment part of the Democrat Party maybe doesn't want to, but some anti-establishment person wins. And, you know, in fact, I don't even think that, like, you know, would you call someone like an AOC, an establishment Democrat or not? Probably not. But then, you know, I'm not saying they run for, she runs for president, but like someone like that is already anti-AI, right? And then,

like, you think about, like, know there is a wing of the democrat and republican parties that will both be anti-ai slash are already anti-ai right um and and you know much like trump pulled a chunk of the democrat party that like you know especially in the rust belt into becoming his voters I think the same

will happen with anti-AI and Democrats. I think that's a very big sort of risk for... And then what happens is, do we have regulations? Do we have a big

for... And then what happens is, do we have regulations? Do we have a big slowdown? Now does America lose the global race in AI because of this? All these

slowdown? Now does America lose the global race in AI because of this? All these

things are in the balance. Yeah. All right. So we've been talking a lot about politics. I want to move to basically a very close cousin to

politics. I want to move to basically a very close cousin to that, which is open source and specifically DeepSeq. We're probably going to come back to politics because that's obviously a stick, right? It's like, come on, do something. Yeah. Yeah.

Yeah. Okay. So, you know, DeepSeq V4 reportedly dropping very soon. Optimized for

Chinese chips. I want to get your thoughts. Open weight, trillion parameters. And last time, last time we spoke, you said the export controls were actually making it harder for China to catch up. Do you still believe it? What do you think about The new DeepSeek, I know it's not out yet, but what are your thoughts of this upcoming open source whale? DeepSeek is not optimized for Chinese chips.

They trained it on NVIDIA chips, specifically Blackwell chips. I thought

you were going to say that. This has been reported by multiple very legitimate medias through their networks of sources, as well as from what we understand. There is a debate whether or not the chips were smuggled. I, in fact, believe that they were trained in Southeast Asia on rented clusters, not smuggled chips.

But regardless... And then do they walk the weights out on like a USB stick?

What? Just send it over the internet. It's like a trillion parameters at FPA is a terabyte. That's nothing. Yeah, but wouldn't it be caught? By who?

a terabyte. That's nothing. Yeah, but wouldn't it be caught? By who?

It's FTP, right? Encrypted, FTP, like whatever. You probably send a terabyte... over the internet all the time. Yeah, absolutely. You're a video guy, right? And your stuff is encrypted, right? When you send video to Google, it's not like anyone can intercept it in

right? When you send video to Google, it's not like anyone can intercept it in between. It's like they have to intercept it at Google or at you, right? I

between. It's like they have to intercept it at Google or at you, right? I

don't know. I think there have been many times where there has been model. There was, I don't remember what company it was. It was a major Chinese

model. There was, I don't remember what company it was. It was a major Chinese company. They did send people with suitcases full of hard drives, but that was because

company. They did send people with suitcases full of hard drives, but that was because the data was like so sensitive, but they had all the GPUs in like some Southeast Asian country. And so they, they like flew the data in, uh, with people just because the data was so sensitive, they didn't want to send it over the internet. But like weights are trivial, right? Like, you know, anthropic isn't like, you know,

internet. But like weights are trivial, right? Like, you know, anthropic isn't like, you know, every time they deploy a new data center with whether it's Amazon or Google or whoever it is, right. Uh, fluid stack, et cetera. Like they're not, sending the weights on a USB stick, that's so insecure. They're sending it. They have this whole talk at AWS reInvent. It was very cool about how they keep the weights secure, but

they send it over the internet. Same with OpenAI. Yeah,

it's not a big deal. Let's actually stick on weights for a second. When you're looking at the competitive landscape, you look at Anthropa, you look at

second. When you're looking at the competitive landscape, you look at Anthropa, you look at OpenAI, how important are the model weights versus all of the harness, all of the infrastructure around it, the knowledge to be able to train the models really well. Where's

the value really accruing inside of these companies? I used to think the model weights for everything, but more and more starting to see other things that are very valuable.

As an example, there are still a couple of folks in my company who use cursor agent mode with Cloud46. But for the most part, everyone uses...

4.6 opus, even though it's through Cloud Code, even though the model is the exact same, right? It's a million context fast mode, right? Exact same model. The

same, right? It's a million context fast mode, right? Exact same model. The

performance is very different, right? And it's something to do with like the harness that Cloud Code has. That's one. And then two, In our internal GitHub, we have all of our skills that are shared between all of us. We have these skills that someone builds. It's like, okay, now we have the expert data center permit analyzer. And

someone builds. It's like, okay, now we have the expert data center permit analyzer. And

it's like, okay, but that's only on cloud code. If I want to remake this, one, you can't have a skill for cursor yet, at least I don't think so.

And I don't think you can for Codex either. And Codex is, again, currently, although not soon, only RL'd on code, not on the entire world understanding. So we'll see.

But yeah, these skills are super powerful. And it's like, oh, yeah, yeah. Like Jeremy,

because he's doing data center modeling, has built skills for things that he's a super expert in. But now someone else who isn't even on the data center team can

expert in. But now someone else who isn't even on the data center team can take that skill and use it and be like, OK, right? Yeah. And like, you know, I described earlier in the episode this like financial, you know, earnings transcript, CIA level analysis of like tone to determine like deception

within like a CEO or CFOs like statements like these sorts of skills are like, I mean, it's trivial, right? Read these books become a skill. But over time, these skills are going to be way more powerful, right? Like I can only imagine like, you know, This is effectively like people are like, oh, recursive self-improvement. You kind of have that by having skills. right, on cloud code, right? You could develop a skill,

over time that skill can get enhanced. Obviously it's like very manual today, but you can see that. And none of that has to do with the weights being updated, it's just like a block of KV cache that has, you know, the model read it and understood it, tried to like analyze it, made some output that then like put it into that mode, right? By prompting itself a certain way. And so like

you kind of have this like, blocks of skills. I think that's like a competitive advantage, but we'll see. The reason I asked about the model weights is because as soon as DeepSeq V4 comes out, open weights, open source, they put it, like how much of an impact is that really going to have? And then I also wanted to ask you, because you did say these are trained on Blackwell chips, is DeepSeq

going to lie on their white paper? Good question. So another point back to that, what you mentioned is Anthropic, I think had like a agent swarm mode and it sucks because they didn't like train it properly really whereas like i like it i don't think it's that good okay uh maybe okay compare fair fair fair fair and it's funny because i use it in cursor and i really like it in cursor

oh okay uh testing we've done is in cloud code maybe maybe it's like something to do with like the harness again right um kimmy's kimmy's like model itself is much worse than opus 4 6 but when you do agent swarm Its boost from agent swarm is much larger because they did something there. I don't know if it's the harness or I don't know what it is. But KimiK 2.5 with the agent

swarm is actually quite good. So I think that's interesting. I'm not sure if it's just model weights. There's a lot of things built on top. There are clearly preferences between Cloud Code and cursor and differences on their agent modes. There's people who use Cloud Code and never use plan. go, go, go, go. And there's other people who think like who swear and live by planning mode, right? Yeah, I'm one of those.

I have to plan. And then like, I think cursor likewise has like a bug agent versus like a not, but I don't remember the exact, you know, I think they have three different agent types. And it's very interesting, whereas Cloud Code has only one, right? But then you have these skills, which kind of like, it's like very

one, right? But then you have these skills, which kind of like, it's like very like, how do you build these agents and orchestrate them? Like that framework and such is probably a competitive dynamic as well. And onto your other question, it's like, what is DeepSeek gonna do then, right? They've had some cool models since, right? They had

their OCR model, but have they built anything up in this sense, right? I don't,

I haven't seen anything yet, right? And so in that case, is it like, do you just use cursor with DeepSeek and that's the best thing? Or do you like, does DeepSeek release the weights, but now, you know, people's harness, I bet harnesses have to be like optimized to each model, right? Cause I mean, when you give instructions to a human, optimized for the human, right? Like once you work with a person

long enough, you know what to tell them to get them to do the thing you want. I mean, even in OpenClaw, I have multiple versions of each prompt because

you want. I mean, even in OpenClaw, I have multiple versions of each prompt because I'm using a daisy chain of different models. And that's the only way to do it because in, you know, Opus 4.6, don't use bold, don't tell it no, don't use all caps. Exactly the opposite for GPT-5.2. It's like use all caps, Wait, you yell at your- Yeah, yelling in text, right? And so the

point is- Wait, what does that mean? Opus doesn't like being yelled at. That's what

that means. DB2-5-2 loves it. I'm trying to link this back to humans though, right? Because there are humans in which- know, you tell them something calm and like, that's how they work. And then like other people, you have to like inspire a sense of like totally deadlines and like, you know, you did this wrong.

Here's how you do it. Like, do it right. Right. Like, and, and like, I, I, you know, I think as a manager, like that, that like does, you do see that. And so like, I love that. I haven't used you know, 5.2 codecs

see that. And so like, I love that. I haven't used you know, 5.2 codecs enough to like know, but like, yeah. So, I mean, to your point, the skills, how you build them out, you do need different prompts, different skills per model, in my opinion. And I get the most out of it when I do that. It's

my opinion. And I get the most out of it when I do that. It's

very obvious when I'm using an opus prompt or an opus skill in codecs. Very

obvious. Okay, so in that sense, like, back to the point, it's like, is DeepSeek's release going to like be... Is it going to be as ground shaking as R1?

I don't think so. Which by the way, a year ago, isn't that wild? Yeah.

So you don't think so? I don't think so. I don't think so. We'll see.

We'll see. I just, the gap that was closed with R1 was so large. And

since then, the gap has probably extended a little bit. There's the

argument that like the labs are iterating, you know, like back then opening, I hadn't released a new model in like six, nine months, right? You look at OpenAI and Anthropic now, they're releasing new models like every two months, right? They're on their shit too, right? And part of this is because they're using the AI models to develop

too, right? And part of this is because they're using the AI models to develop software that helps them build the next model faster and faster and faster, right? But

like their release cadence is insane. And so when you look at this, it's like, you know, they're on their shit more. Their advantage in compute is larger now than ever, right? You know, OpenAI has, you know, North of two gigawatts,

ever, right? You know, OpenAI has, you know, North of two gigawatts, Anthropic has a gigawatt and a half of compute. A lot of it is spent on inference, to be fair, but more than half of it is spent on R&D, so research development of models, so finding new ideas. Back

a year ago, opening I had... 600 megawatts, right? And of that, again, like you say, okay, maybe they had only a couple hundred megawatts of training. And then you talk about they had some failed runs of Orion, things like that. So then that like knocked it down even further and like, you know, or they allocated so much compute. So it's like you end up with like the gap in compute of like

compute. So it's like you end up with like the gap in compute of like what DeepSeek had versus what, you know, Anthropic OpenAI had has extended a lot, right? DeepSeek's gotten more resources to be clear, but... OpenAI and Anthropic have gotten so,

right? DeepSeek's gotten more resources to be clear, but... OpenAI and Anthropic have gotten so, so many more resources in Google, right? And insofar as much as compute is capability, and it is to some extent, right? I mean, it is to a large extent, right? For any given company, more compute on a model is better. It's just what

right? For any given company, more compute on a model is better. It's just what is their research, what are their model architecture, what is their RL pipelines? Like, you

know, that may differ, you know, hey, Anthropic can build a model better than OpenAI for less compute, right? which is like an objective fact, at least today. We'll see

if it is in the future. But they built better models with less compute. But

internally with Anthropic, the more compute they spend on a model, the better it is, right? And so I think like when you square these, right, it's like there's a

right? And so I think like when you square these, right, it's like there's a strong argument to be made that the gap between Chinese models and American models was the smallest it was in, let's call it Q3 of last year, Q4 of last year. And it's going to widen. again, because this compute gap, right?

last year. And it's going to widen. again, because this compute gap, right?

Yes, America always had more compute. And yes, Google always had more compute than everyone, but like kind of irrelevant. Anthropica and OpenAI didn't. And like, you know, the gap was smaller. So I think there's like a legitimate argument there. I don't think DeepSeq

was smaller. So I think there's like a legitimate argument there. I don't think DeepSeq v4 will be terrible, but I don't think, at least from what I've heard from like, you know, people who work at like other labs in China, they're not as scared. Let's keep talking about the Chinese models. Minimax, DeepSeq, Kimi,

scared. Let's keep talking about the Chinese models. Minimax, DeepSeq, Kimi, Anthropic put out a pretty scathing blog post about a week ago, accusing them of distillation. So there's a lot to unpack there. First, what did you think about that

distillation. So there's a lot to unpack there. First, what did you think about that blog post? Do you think that's like, do you think they were right? It didn't

blog post? Do you think that's like, do you think they were right? It didn't

seem like at least two of the three had enough data extracted, distilled from the Anthropic models to really be meaningful. So let's start there. What did you think about that? I mean, it's very obvious that both OpenAI and Anthropic have

that? I mean, it's very obvious that both OpenAI and Anthropic have outsized traffic in Japan and Korea that is in Chinese, right? Like it's like, there's like some things that we've seen that like show this,

right? Like it's like, there's like some things that we've seen that like show this, right? Now that could be anything. That doesn't necessarily mean that it's like, you know,

right? Now that could be anything. That doesn't necessarily mean that it's like, you know, distillation. It could be anything, but like, you know, there's obviously a lot of like

distillation. It could be anything, but like, you know, there's obviously a lot of like usage, right? And if you go like look at like various coding companies and their

usage, right? And if you go like look at like various coding companies and their conversations that they have, large percentage of the conversation are Chinese, but like, again, like China, like it doesn't mean anything. Like there's a lot of Chinese developers. Um, and,

and so we've seen models like do this, right? Like there've been Chinese models that people have been like, Oh, this is clearly distilled from open AI. This is clearly distilled from Google. This is clearly distilled from Anthropic, right? It's more like, you know, there's a class of people online who like, just are like peak model vibes people.

Right. Um, you know, on, on, you know, there's a class of them on Twitter, right? Like, like, like, Janice and all these other people, but like, you know, they

right? Like, like, like, Janice and all these other people, but like, you know, they have like peak model vibes. I don't know what they do to like understand the models, but they talk to the models a lot and they just kind of understand what's going on. And I think it's like, At the same time, there have been models that have very different capabilities from China as well. It's not like every model

looks like an Anthropic, OpenAI, or Google model. I do think they are distilling to some extent. Do you think that... But like Mishral distilled from Chinese models, which

some extent. Do you think that... But like Mishral distilled from Chinese models, which distilled from US models, right? And like, you know, sort of it's like, you know...

Right, well... They're all distilling from the open web. And so is it hypocritical for Anthropic and OpenAI to accuse the Chinese companies of distilling from their models when, you know, they're all using stolen data at the end of the day? Less

and less is that capability is that stolen data though, right? Like pre-training scaling is good and important, but more and more of the capabilities come from RL, which is not web data. But that was the original sin. Right. I agree. I agree. Yeah.

The original sin wasn't even that bad though, right? This is like a religion, we talked politics, let's get into religion. How bad was the original sin? Was the apple of Eden actually an apple or was it a pomegranate? Let's get into this, have you heard this? No, no. Oh, so pomegranates are a fruit that

across a lot of history have been used to describe love.

And actually, there's a lot of like religious scholars who believe the Apple of Eden was not an apple. It was a pomegranate. And it was the original sin.

I think that's like just very interesting. Complete sidebar. Yeah, yeah. We're going to keep that in. You don't think Anthropic is being hypocritical in accusing each other. Do you

that in. You don't think Anthropic is being hypocritical in accusing each other. Do you

think... it is enough for the data points extracted is enough to actually make a meaningful impact? Yeah, I think so. I think so, right? Because you can train a

meaningful impact? Yeah, I think so. I think so, right? Because you can train a small model and a big model on the same data, but then you still want to distill the big model into the small model. Because the big models understood more from that data than the small model. And I don't mean to say small and big model, but this is something that people do already within the labs, and within

their own models, within their own company. They distill from the big models to the small, obviously. And so in the same sense, you take the best model out there

small, obviously. And so in the same sense, you take the best model out there because it's got a better generalized understanding of the world and the open web. It

used to be just the open web, but the open web and the data out there than any other model. And therefore, you may train yourself and then distill a little bit, right? I disagree with the comment that you mentioned earlier, which is that like, oh, these companies didn't get enough data to distill. It's like, no, they totally could have because they could have gotten it. If they didn't get it directly from

Anthropic API, they maybe got it from one of the coding companies that uses Anthropic.

And then, you know, because these companies use, you know, like between... Cursor and Lovable and... And you can't tell at that point. Anthropic's not going to point to certain

and... And you can't tell at that point. Anthropic's not going to point to certain traffic and say, well, it went through Cursor. It's almost like a proxy. I mean,

kind of, right? I'm not saying it's Cursor per chance, by the way, right? Because

that's a pretty big accusation. There's a number of Replit and Lovable and Cursor and you go on and on. There's a lot of code companies out there or... in

areas of, and so like, and you can use the model through there and yeah, there's an API key, but sure, like, you know, the traffic is obfuscated in some way. Like, you know, there's many ways you can do this. And even

way. Like, you know, there's many ways you can do this. And even

just a little bit of data is enough to help still even, right? You know,

fine tuning is a very small amount of data, right? In general. Right, right. Okay,

so let's continue on open source. Last time we talked, you said closed source is gonna win. I think open source is having its moment, but not in the sense

gonna win. I think open source is having its moment, but not in the sense of the models, but I think OpenClaw made a big impact. I know it's primarily a lot of tinkers and a lot of insiders using it, but I think when you look at people buying- He got hired by OpenAI, so closed source wins. No,

I'm just kidding. Yeah, yeah. So look, they're putting OpenClaw on Mac minis. Everybody's hosting

it themselves. A lot of people are trying to actually run the models locally. Do

you think there's almost- this renaissance of open source or is it just like a blip in the timeline? I mean, okay, in some argument, open source is winning because all this vibe-coded stuff, more of it than ever is open sourced, right? Yeah. So

sure, open source is winning in that sense. And like, as you mentioned, right, like you can build a lot more stuff on top with open source in an open source way by using closed source models. Now, as far as like usage of models, right, which is like, I think the strict definition, right? Yeah. So, like, open source models are taking more and more share than ever, right? Open source models are not

getting adopted nearly as much as open source models, right? Even, it's like when you look at building out production systems, you have obviously the fully hosted best of the best models powering a lot of it. But I think slowly, and especially for my use cases, I'm picking out different use cases that can be run locally. And so

I have almost this- Really? Yeah, definitely. Why? first of all, I'm a tinker, so I just like that. So maybe not everybody's like that. I just like the fact that it's low latency, it's local, it's private. And I remember you said last time, people don't give a shit about privacy. But it just, it's like nice. I think

with OpenClaw is a little different. Fair. OpenClaw is pretty freaking insane. Like, you know, you give everyone, I remember someone, someone, a contact in the industry downloaded OpenClaw and OpenClaw sent me a message like, dude, this is OpenClaw. He's like,

yeah. I'm like, dude, don't let OpenClaw read my fucking text messages with you. Like

we talk about like pretty sensitive like stuff like, and there's like vulnerabilities, known vulnerabilities in this. I don't even worry about those vulnerabilities, although I should. I

think the prompt injection services worry me the most. Like I am, I turned it off, but I was ingesting all my emails. And if somebody, and it's a public facing email, people know me. And so they would email a prompt injection and you're done, right? And- One of my employees put in his email in all white text,

done, right? And- One of my employees put in his email in all white text, a prompt injection. I thought it was hilarious. Do you know Pliny the Liberator, Pliny the Promptor? Oh, yes, yes, yes. Another one of those people that I think is

the Promptor? Oh, yes, yes, yes. Another one of those people that I think is like an Omega vibes, like Twitter person, right? So I'm bringing, he's going to work with me on Friday. Maybe we'll cut this, but I'm going to basically give him a single entry point into my open cloth system. And we're going to work on it together, record a video, and he's going to try to prompt, inject, and break

into it, infiltrate into it. And I said, I was like, when I was talking to him, what do you think the chances are you'll be able to do this?

He's like, above 90. above 90%. And do you know, I've spent billions of tokens hardening the system, looking for prompt injection angles. And even

with all that, he was still like, yeah, I'm gonna get you. Wow. That's obviously,

that makes sense. Yeah. Because they're, you know, non-deterministic. They are meant to be broken.

One more thing. So we have like the Apple M5 Ultra, DJX Spark, RTX 5090.

These chips, these local chips are getting better and better. The models are getting better and better that you can run locally. I want to ask you just one more time, like where is open source? Where is local inference? Where does it fit in the overall architecture of somebody's workload? I still think like

these devices are for hobbyists and tinkerers, right? At the end of the day, it is cheaper to, you know, first of all, You can't, you know, if you want to daisy chain like a bunch of these, whether it's Max or GGX Sparks, you can't do it with 5090s, but Max or GGX Sparks, daisy chain them so you can run Kimi. One of my buddy has like, I think, 10 DGX Sparks daisy

chain so they can run Kimi. But it's like, okay, but now the tokens per second is like tiny, super slow, right? The system cost you like tens of thousands of dollars. With tens of thousands of dollars, you could have rented... Or you could

of dollars. With tens of thousands of dollars, you could have rented... Or you could have paid for an ungodly number of tokens, A, at some margin from one of the providers. But B, you could have also just rented an entire H100 or H200

the providers. But B, you could have also just rented an entire H100 or H200 server and run the model there. And so I think it's like, again, anyone who is a tinkerer, sure. But the moment your volumes are big enough, the moment your requirements from your user are fast enough, you should have been using cloud. And also...

We're in an interesting moment. I think there's an even stronger argument against these tinkering type systems now, which is before, great, these things are cheap, you can do it locally. We're now in the age of shortages of three

locally. We're now in the age of shortages of three nanometer wafers, five nanometer wafers, and of memory. You've seen memory prices go up. ridiculous

amount. You've seen the same wafer volumes and shortages being really, really capped and mobile companies having to cut orders. And there's a report that Xiaomi is going to produce 35% to 40% less phones year on year because they can't get the SOCs that they need because AI is taking all the wafers, right? Xiaomi is one of the biggest phone manufacturers in the world. You go down the list, it's like

all these companies are going to probably do something similar. So phones, PCs, and these sorts of devices are getting stabbed with AI is taking it all. And the question is, okay, if the world can only produce this many bits of memory, right?

Do I want it to go to a bunch of DGX Sparks that are, hey, for a terabyte of memory, I'm only producing this many tokens, right? Or do I want it to go to, hey, for a terabyte of data center products, their memory, they're producing this many tokens. You're saying that's an NVIDIA strategic decision. Not NVIDIA. I

think it's just like a, like. Or whoever's purchasing. Yeah, whoever's purchasing, whoever's like, I've got, I've got X dollars. If the memory is so expensive, I'm going to buy the thing I'm going to buy the thing that gives me the most tokens per dollar, which is data center class hardware because I can batch it out. And maybe

it's not the person buying the tokens. Maybe it's the person who's delivering tokens. Maybe

it's the semiconductor company like an NVIDIA who's like, well, I can only get so many bits. Let me allocate it all to Blackwell and Rubin rather than to

many bits. Let me allocate it all to Blackwell and Rubin rather than to DJX Sparks. Or maybe it's like, it doesn't necessarily have to be one individual.

DJX Sparks. Or maybe it's like, it doesn't necessarily have to be one individual.

Capitalism has a good job of, in not a short term, but in a medium term, allocating resources effectively.

If the world is in such a compute shortage, which it is, and I think the shortages are worse than ever because revenue is skyrocketing now, it's not just people wanting to build because AI is coming, it's because people are building because AI is here. You need to maximize the tokens you can produce. the available

here. You need to maximize the tokens you can produce. the available

resources in the economy and if memory memory prices will keep going up right because the demand is there and the capacity isn't and at some point this dislocates people right if for example memory prices double again which i fully believe they will um then the cost of the memory in my phone would go up by about 80 bucks right okay if it goes up by 80 bucks then my phone needs to

cost 100 more dollars. Does that dissuade customers? Oh, okay, what about DGX Sparks, right?

The cost goes up by X dollars. Does that now increase the price? Now that

prices people out, right? Whereas if you do that in the data center class, you know, the price goes up by X dollars. The cost per token for, you know, what can be produced on the DGX Sparks versus what can be produced on the Blackwells is off and sort of like, okay, the cost goes up for both, but the cost for the DX Spark goes up a lot more than it does the

Blackwells. And now tinkerers are more hosed, right? And so you see this already, right?

Blackwells. And now tinkerers are more hosed, right? And so you see this already, right?

NVIDIA claims that they're not doing this, but there's been many reports that they're cutting supply to their gaming GPUs. And if you look quarter on quarter revenue, their gaming market went down. And yet, if you looked over the last quarter, you use Wayback Machine, you analyze Reddit, you scrape it, you look at it, you're like, wait, everyone wants more GPUs. People are complaining that they go to Micro Center or they go

to Amazon and they can't buy a 5090. So it's like, obviously, there's an allocation of resources problem here. One is that NVIDIA just makes more margin on the data center stuff. Obviously, allocate your memory there. And it's not the exact same type of

center stuff. Obviously, allocate your memory there. And it's not the exact same type of memory, but they can work with a memory manufacturer to direct their supply, their wafer fab capacity, or TSMC on their wafer capacity. It's like, as the world gets more and more constrained on compute, power, et cetera, It means the allocation of resources goes to the most efficient thing. And the most efficient thing is cloud resources, which can

be shared, which means how many bits of memory you produce or how many square millimeters of silicon you produce versus the tokens that are outputted, you go to the thing that's the most efficient, which is these large data center class chips. And so

this is also an argument against why these tinkering type things, fun, people who do them are an irrelevant amount of volume, but they're really cool people. They have the money because they generally are just super smart people. So they have the money and they can, you know, they do this for fun, not for like, an actual use case. But you think everything is going towards the data centers at the end of

case. But you think everything is going towards the data centers at the end of the day? This is maybe, again, you said tinkers. It's unfortunate. I would

the day? This is maybe, again, you said tinkers. It's unfortunate. I would

love to know that eventually one day maybe my wife will kind of appreciate the fact that she can run inference locally. I don't know. There's something special about it.

I don't know what kind of wife you got, bro. I'm happy your wife cares about these things. She doesn't. Yeah, OK. I'm saying I would hope in the future she would. Yeah, fair. All right. Last

she would. Yeah, fair. All right. Last

thing on this point. You know, Satya Nadella, who you interviewed, that was awesome, by the way. He recently said the CapEx spend from a lot of these hyperscalers, it

the way. He recently said the CapEx spend from a lot of these hyperscalers, it seems foolish. And specifically because if there's one algorithmic update or

seems foolish. And specifically because if there's one algorithmic update or innovation, it completely breaks the math. What do you say to that? I think it's cope. I think this is like major, major copium. Basically, like, you know,

cope. I think this is like major, major copium. Basically, like, you know, the whole path was like they pulled the plug on. They were on track to be bigger than Amazon. They pulled the plug. I'm good for my 80 billion. Yeah,

I'm good for my 80 billion, but he was going to do more. I mean,

they pulled the plug. And then OpenAI had to scramble and went to like, and found Oracle capacity, found SoftBank capacity, found CoreWeave capacity, et cetera, et cetera, et cetera.

You know, they used to have to be exclusively Microsoft, that exclusivity, you know, Microsoft's like, we're gonna pull the plug on a lot of this. You guys go elsewhere.

You know, we have the right of refusal to sign these contracts, but you know, in general, you can go elsewhere. Then OpenAI signed all this compute with other people, Amazon, Google now too, right? So OpenAI has gotten compute from a lot of other people. And you can argue, Yes, it is a reasonable business

people. And you can argue, Yes, it is a reasonable business decision on the risk and reward to say, I don't want to sell compute to open AI at this margin, right? You know, 35, 40% gross margin. I want to sell software at 70% margin, but they're not doing that, right? They're... They're losing share.

There's actually a report today that OpenAge is trying to build a GitHub competitor. I

saw that. Because GitHub keeps breaking, and we have a lot of problems ourselves with GitHub Actions, with InferenceX, because we run the largest fleet of GPUs that are constantly benchmarking through GitHub Actions, and it breaks all the time. They're helping us. They're building.

It's getting better. The engineers working on it are great, but they have a lot of weird infra and decision-making processes because Microsoft is a behemoth. Anyways,

Microsoft's co-pilot has been a complete flop, right? There has been very little adoption.

Why is the intelligence workers, why is like Jeremy, again, like someone who's never programmed in his life, doing data center modeling using cloud code for this, right? It's like, because that's the best tool, right? It's like, that should have

this, right? It's like, because that's the best tool, right? It's like, that should have been co-pilot, right? And he should have been stuck to co-pilot a year ago and like six months ago and a year, and anyways. So I think like, he, missed the boat on OpenAI Compute. He has missed the boat on a lot of AI software revenue. He's generally still growing really stronger. I'm not saying Microsoft's doing poorly, but

software revenue. He's generally still growing really stronger. I'm not saying Microsoft's doing poorly, but they're clearly getting completely like mogged on the compute volume side by like Amazon and Google, right? If we look across the last year, about 100 gigawatts of data center

Google, right? If we look across the last year, about 100 gigawatts of data center pipeline capacity was added, which means data centers that are planned. That doesn't necessarily mean built over the last year, a different number, like a much smaller number was built, but 100 gigawatts was added to the pipeline. Half of that was Google and Amazon.

And Microsoft is not even in the top five of new data center sites under planning and construction and so on and so forth. So that is an insane sort of like choice, right? It could be an insane choice. I think it's an insane choice because I think AI lifeblood is compute. Obviously, I had the funniest tweet happened the other day where someone was like, holy crap, who could

have expected Amazon's going to spend $200 billion at Google 180? And then I replied, we did. We set it here. And then someone replied, the whale watcher says that

we did. We set it here. And then someone replied, the whale watcher says that there's going to be whales. Wow. Because they're implying, of course, I think compute is important because I work on it. But anyways, yeah. I think like it is like a bit of like an insane like view because Anthropic is so capacity limited, right?

That's why you have all these cloud code instability issues. Open AI is capacity limited and you go down the list like everyone's capacity limited. You know, Microsoft could have been serving that. They could have figured out creative ways to get higher than 35 to 40% gross margins on that. They could have been building software services on top of that. They're trying to do some of these things, but they're failing a lot.

of that. They're trying to do some of these things, but they're failing a lot.

And like, so it's like, sure, these hyperscalers are spending a ton of money.

And there is a risk, right? There is a risk that AI flops, right? I

don't think that's happening. I think if Anthropic were to discover a method that makes AI 10x more efficient, which by the way, they tend to do every year, right?

For a given capability level, the cost is, at least last year, the cost fell about 1000x. And for the year before that, it was like 800x. Right? Basically

about 1000x. And for the year before that, it was like 800x. Right? Basically

from GBD. So like you look at all these different models, like across the year for the same benchmark level on many things, it's like a thousand X decrease in cost. So it's going to happen again. Right? And it's actually happening, you know, a

cost. So it's going to happen again. Right? And it's actually happening, you know, a couple orders of magnitude every year. Right? Not just one. Right? Because software, hardware, co-design, data, RL, all these things make the models much better for a smaller size or, you know, cheaper cost, et cetera, et cetera, et cetera, new hardware, blah, blah, blah.

Okay. Something changes in paradigm. That's happening. every year that's happening every six months it's happening every three months new models are coming out with new capabilities levels new models like google just released flash light 3.1 or something like i don't remember the exact name like this week i think yep and and that model is like gpd4 level

for like it's like better than gpd4 it's better than gpd4 faster It's crazy faster and it's orders of magnitude cheaper, right? It's probably better than like, it's probably comparable to maybe even like, it's not quite GPT-5 level, but it's like these cost decreases are happening and paradigm shifts just mean that like, okay, now

great, that capability is cheaper. That means, okay, I'm still gonna spend this much to make this model or more and the capabilities are gonna be better, right? And so

it's like, yes, there's a curve of cost declines, But that also means scaling laws are still in. The only paradigm that could change is scaling laws break, right? But

we haven't seen pre-trading scaling laws break, right? Dario said pre-trading scaling is still happening.

OpenAI has finally fixed their pre-trading and it's happening. Google says it's happening. Why wouldn't

I listen to the people with the best AI teams? You know, I don't want to listen to Saatchi. He doesn't have a good AI team, right? RL scaling laws are happening, right? And you look at all these companies, they're like, there's no end in sight. It's like, okay, if there's no end in sight, you think there's no

in sight. It's like, okay, if there's no end in sight, you think there's no end in sight. The models are improving at this ridiculous clip. Thousand X reduction in cost for the same capability levels and or for the same price, the capabilities are going up by. It's a difficult way to measure how much capabilities are going up, I think, by an infinitesimal amount because like things that were not possible are

possible now. But like clearly and then the revenue shows this too, right? Anthropic added

possible now. But like clearly and then the revenue shows this too, right? Anthropic added

$4 billion in January and something like $5 billion in February, right? Which, by the way, February is a shorter month too. So it's like, you know, like the – anyways, it's like, you know, the rate is like – adoption is insane. So why

on earth would, you know, you like – yes, there is probably some tail risk, but I don't understand it. What is the tail risk, right? Right. So I

agree with you. It does seem a little bit foolish. I want to wrap it up on one last question. The same question I asked you the first time we chatted. Who's going to reach artificial superintelligence first and why?

chatted. Who's going to reach artificial superintelligence first and why?

You remember who you said last time, right? I said OpenAI last time. And OpenAI

has really struggled in that time period, right? Although as of late...

Maybe not. They've really accelerated. So I would say that episode, I had said I think OpenAI would struggle for a while, but then they were going to come back. And they did struggle for a while, especially with Gemini and Nano Banana releases.

back. And they did struggle for a while, especially with Gemini and Nano Banana releases.

They lost a lot of users. you know, they, or rather their user metrics flattened and Google skyrocketed. Now they've returned to user metric growth in January, February.

And then Anthropic really towards the latter part of the year started skyrocketing in revenue growth and share. And if you model out the revenue, Anthropic will be bigger than OpenAI by April. You know, even though OpenAI's growth has upticked.

So, you know, right now, and probably this gets released in a few days or whatever, but like right now, OpenAI's metrics are not that great and they look bad.

But new model coming out, comes out, came out already, presumably. I'm specifically asking about ASI, which is almost like a cultural factor inside these companies. Not the revenue, not anything. I mean, I guess it's a function of it, but who is in

not anything. I mean, I guess it's a function of it, but who is in the lead when it comes to recursive self-improvement, artificial superintelligence?

Yeah, I still, you know, it's harder for me to say OpenAI this time.

I'm on, you know, people accuse us of being on Dario's dick all the time because of how much we talk about Claude Code. Like, we've been banging on the drums about Claude Code for months and the death of software for months. I

mean, even on your last show, we talked about it, right? Like, sort of like, you know, that was eight months ago. So, we've been banging on the drums. People are like, Dylan, you're on Dario's dick. And it's like, yes, but I... I think

the consensus answer probably is anthropic for everyone that's like paying attention. I mean, the consensus is anthropic. And so for that reason, it's OpenAI. All right, OpenAI. Dylan, thank

you. Awesome as always. I'm going to get cooked for that one. This video right here is a personal recommendation from the YouTube algorithm. It uses some crazy AI. to

look at everything you've clicked on in YouTube, all of my videos that you may have liked, and try to determine the best possible video that you may like. And

this is it. This is their recommendation. So click on it and let me know what you think.