← Tom Bilyeu's Impact Theory
Tom Bilyeu's Impact Theory
The Final Economy: How AI, Crypto, and Robots Will Reshape America Forever
00:49:13
Loading summary
Transcript89 lines
- [00:01]
Tom Bilyeu
Nine out of the ten largest banks get it. They get advantagescore. The Modern Credit Score is the leader in predictive power, improving mortgage default predictions and saving lenders billions. Better predictions better for your business with VantageScore.
- [00:15]
Grainger Announcer
If you work in university maintenance, Grainger considers you an MVP because your playbook ensures your arena is always ready for tip off. And Grainger is your trusted partner, offering the products you need all in one place, from H Vac and plumbing supplies to lighting and more, and all delivered with plenty of time left on the clock. So your team always gets the win. Call 1-800-GRAINGER visit grainger.com or just stop by Grainger for the ones who get it done.
- [00:45]
Tom Bilyeu
Welcome Back for Part 2 with EMod Mostaq if you haven't listened to Part 1 yet, you're going to want to go do that because Part two is going to make a lot more sense when it's in context of EMOD's big vision of AI ushering in one final economy. Without further ado, here's Part two with Imad Most there's going to be a very large number of people that even if they have a hyper intelligent AI guiding them, they're not going to listen. So man, this is, oh God, this is going to get so weird. Okay, so my gut instinct is that I like the idea of, okay, you're generating money not off of debt, but you're creating some amount of money that you give to people. People are still going to derail. People that are not that don't have the intelligence or the discipline to listen to the AI are still going to derail. There's going to be a lot more that we need to figure out than just the economics of it. Because I'm even just thinking, all right, if you're talking about a safety net, let's say that it's a health safety net. I assure you one of the biggest pieces of advice that the AI is going to give people is don't eat that. And people are still going to eat it. Don't smoke that. People are still going to smoke it and then other people are going to be asked to pay for the additional cost of the people that are still eating that, smoking that, whatever, even though they have an AI in their ear telling them, don't do that, do this, man. There are levels of complexity to this. All right, I'm going to set that aside. Maybe we'll come back to it.
- [02:18]
Imad Mostaque
Yeah.
- [02:18]
Tom Bilyeu
What my audience is really going to care about. You've set the stage for us perfectly. I think we have a very clear understanding of how tumultuous this transition period is going to be. But now look at this moment and where we're going through the lens of being a hedge fund manager. How can people win in this moment? Because I was, when you were talking, I was like, oh my God. I guarantee if people are given a stipend, other people are going to try to win that money from them essentially by either whatever the new stock market gambling is sports betting. Like there's no universe in which barring like just extraordinary top down authoritarian controls that a sort of secondary I'll call PVP server of people competing to win your dollars. We do it now through entrepreneurship where it's like, hey, I can make you this thing that you want more than you want your money. And then that's how I win at that game. That game's not going to go away. That, that is just baked into the human DNA. So people are 100% going to do that. Obviously the first thing that came to mind was prostitution. I'm like, that's going to run rampant. People are going to run out of money before the end of the month and they'll be like, well I can at least sell my body. This is going to be wild. Because none of this gets rid of the fact that we are still humans. Okay, having said all of that, give me the hedge fund manager. Look at how do we win in this moment.
- [03:49]
Imad Mostaque
Actually, you know, just you saying that last thing, had that thought you said what's changed since 2008? And I thought onlyfans, you know, like how much of America is signed up or actually on. It's crazy statistics, right?
- [04:03]
Tom Bilyeu
Dude, it's wild. Women 18 to 24, something like 10% of them are onlyfans models. That is insane.
- [04:12]
Imad Mostaque
It's insane. But again, I think the oldest profession in the book, human connection, these kind of things.
- [04:21]
Tom Bilyeu
That is the nicest way to say sex I've ever heard in my life.
- [04:25]
Imad Mostaque
Thank you, I'm a gentleman. What can I say if we take a step back, what's the inevitability here? Is the government going to abandon all these middle class people and voters? Actually probably not the voters. Yeah. Especially in the blue states. Actually, blue states will probably be impacted more than red states for various reasons. If you look at the demographics, they're not going to, are they? So what you have is you just need to do your classical analysis of what does that person do when they lose their job and they've still got dollars, they've still got savings. People will be looking for retraining, they'll be looking for meaning. Religion is going to go crazy and boom. These kind of things are things that almost inevitabilities because they'll still have purchasing power to a degree. On the other side, you have two economies, right? You have your AI economy and your human economy. The AI is providing increasingly customized services and, and getting a lot of the cognitive surplus, et cetera. But a lot of things you can't substitute for a human for at least another 10 years. And the reason for that is just we can't build enough robots, honestly. Like I think robots in a few years will be able to do just about everything a human can do apart from the very soft skills, although the Japanese are going very aggressively on that. But you just can't build enough of them. That's literally the only thing holding it back. Because if you look at what Elon Musk says about Optimus and you work out the math, an Optimus robot will be a buck fifty an hour. Jesus. Work out the math. It's $20,000. You have a depreciation schedule. And again you look at unitary and other ones like they have fine finger manipulation now. They can make recipes, they can do all this, they'll have skin suits, etc. But again, human connection retraining attention is the thing that doesn't become scarce. This is the really interesting thing. Do you think video games are going to go down or up over the next few years? They're going to go up because again, there's only a finite amount of human attention. And as people get more free time, they'll want to absorb that attention even more. The new media space is going to go crazy. Digital assets. I think the US has gone too far on legalizing them now in some ways, when I look at the legislation that's coming out, like I said, AI would be the biggest bubble ever. The digital asset bubble is going to exceed that by far. You'll be able to buy any cryptocurrency ICO from your smartphone using Apple Pay on Stripe next year. So what are they going to do? There will be some really interesting classical stuff. And our foundation coin that we're building is the better Bitcoin that helps cure cancer is going to be probably at the top, he says. But there will be so many of these crazy dodge coin, fart coin type things, celebrity coins. They've never took off NFTs because they're scarce forms of capital. And again, people have a certain amount of attention and they'd be looking for the casino like you look at Cali and polymarket, they've legalized those now. What are those? They're betting.
- [07:33]
Tom Bilyeu
Yeah, straight gambling. So is the stock market, in my opinion.
- [07:36]
Imad Mostaque
But the stock market. Yeah, but, you know, at least they had the excuse. Whereas Kalshi and polymarked betting, it is
- [07:42]
Tom Bilyeu
gambling with a better cover story.
- [07:44]
Imad Mostaque
Yes, but a year ago that was completely illegal and now it's legal. So I think if you look at it, there's the soft human aspect, there's the repurposing of all these people, and attention is the key thing. How can you capture people's attention that they'll pay for? Because there'll be a lot more of it because they won't have jobs and other things kind of coming forward. And so we're going to see some booms like we've never seen before. And I think media is going to be ultra interesting in that aspect. Plus, like I said, I was really shocked by the US Government on digital assets. They want to get money moving. But I can't see how next year the digital asset boom will completely outstrip everything. Actually, it's interesting to see this. If you look OpenAI anthropic this year probably did $20 billion of revenue. The entire listed software sector in the US will do 40 billion in incremental revenue. Crypto has done 150 billion in net inflows. Jesus. And next year, is that going to go down or up? It's going to go absolutely ballistic.
- [08:51]
Tom Bilyeu
Okay, but so how are you treating that as an investor? By the way, do you still actively invest, at least for yourself?
- [08:58]
Imad Mostaque
No, I've gone all in on my new thing. So, I mean, we've got a bitcoin competitor coming out. Foundation, coin or complement, shall we say. It's 99% the same code, but every coin cell goes to supercomputers for cancer, education, et cetera, and giving people free AI. And then we're going to put computers in every country, computers for all the sectors. And you can direct the computer, the network, to organizing our knowledge. So benefit, we think that will do well because crypto is a $4 trillion industry with nothing blue chip in it. Like Bitcoin is blue chip because it's lasted a long time. Ethereum, because it's a network. But what's the alternative to bitcoin? If you want a monetary asset? And we thought, what if we create a monetary asset where every coin sale goes to helping people, that builds trust. You use the free AI, it builds trust, you organize knowledge, and it helps people with Cancer.
- [09:48]
Tom Bilyeu
Is there an interface where I'm saying I want this to go to that computer, I want this one allocated to Cancer, this one allocated to autism. Can I allocate to anything I want or is there. It's only from your 6 in the drop down menu. Like how does that work?
- [10:05]
Imad Mostaque
It'll be anything that can be benefited by compute. So we start with all the healthcare things and then we're going to expand it out and you'll have a free version of ChatGPT as your AI assistant to organize that and you'll be able to buy it with your Apple Pay or whatever. And again, it's 99% the same code as Bitcoin, but like a million times faster. So things like that I think will do. He says very well. That's why I've gone all in on that versus trading the market, et cetera. But in general, I think if you think about attention, actually digital assets have to be the biggest thing. If you think about so many forms of capital being completely flooded out again, your taxi medallions, your factories, even other things being replaced by this, your workplaces, offices, digital assets will come to the fore. It's just there's going to be such a deluge of them that you have to be intelligent about that. Because what's more fun, watching Netflix or trading crypto? Probably trading crypto for all the people,
- [11:04]
Tom Bilyeu
for a certain personality type.
- [11:06]
Imad Mostaque
Yeah, NFTs might make a comeback, you never know.
- [11:10]
Tom Bilyeu
Well, the interesting thing, like if people understood the underlying technology, NFTs haven't gone anywhere. They're just not part of the gambling mechanism right now, which honestly I think is better. But nonetheless it does. The, the whole crypto ecosystem in this economic moment is bound to attract gamblers. And I think that we're going to see a lot, a lot, a lot of that. First of all, people just like to gamble. The dopamine rush of it all. But they also, in a time where nobody can afford a house, you're like, well if I am smarter than the next guy and I can out bet them on when to get out, then I really can. And so yeah, you're going to see a lot of that, which is the get rich quick impulse. This all started from me asking you through the lens of a hedge fund manager, where should be people, where should people be allocating their capital? Digital assets is the thing that you have the most conviction in. Obviously you're not backing anything, you're not giving anybody specific advice, but I do want to drill in more. So attention is part of what Makes that interesting with the stock market. The nice thing is, at least until, call it 2008, you could really understand what stocks to move on based on fundamentals. I think that's largely gone out the window as it's become more and more of a gambling mechanism. But what do if somebody were surveilling the digital asset landscape, is there a type of fundamental that you look for?
- [12:43]
Imad Mostaque
So you said the fundamentals went out the window for the stock markets because so much of things are narrative driven then it's crazy now, right? And again, what's your marginal narrative for various companies against each other or various things? Like in the digital asset space, you have something like Hyper Liquid, which basically is doing almost direct buybacks of its shares or Pump Fun or something like that, of its tokens, with cash being valued less than things that have absolutely no cash and no fundamentals whatsoever. Like Dodge, Coin is still worth $20 billion. You know, something like that. Why is this the case? Everything's about marginal narrative. And so what you're looking at is, as the world evolves in the next few years, what's going to capture the marginal narrative? You see Elon setting this up with Tesla or X or whatever by saying they're going to be AI companies and robotics companies, because that's the next narrative. And Elon is a master narration, right? Like Oracle just got to $900 billion yesterday. I think it was up 46%. Why? Because suddenly it's an AI company versus a legal company with a database attached, right? Because they kept suing all their people. People are looking for the narratives, be it in the stock market or the crypto market. So you have to think, what does it look like? And then what are the narratives that are going to incrementally improve and attract more and more people? Because it's dangerous now to deploy your capital. Are you going to give your capital to bonds in the government or are you going to start deploying it everywhere else? What does growth look like? Growth is probably going to come down, rates are going to come down, but what's going to happen then? So I think that what I look for primarily is marginal narrative creation and then understanding where the capital flows go. So when I created Foundation Coin, you know, I was like, I'd like to have a bitcoin, but backed by GPUs, where the GPUs are doing good. I want as much of that new compute capacity going towards helping organize cancer knowledge in the world, helping give that knowledge to people, because that's a good thing. 100% of your purchases go towards that. That's a good thing. That's something you tell your grandma about. And we don't have a blue chip like that in the digital asset sector. So that's how I kind of looked at it. But at the same time, you see areas where communities build around certain things. Right? And that's what crypto has done classically well. But it's also why you have rabid Tesla owners, right? Or you have people that love Palantir and other things and they suddenly go from 10 times earnings or 20 times earnings to 200 times earnings. Wow. I mean, Palantir I think is like 200 times earnings now or something like that. $400 billion as a company, Jeebus, because people like that's the structural growth. So you look at your inevitability, you look at the narrative that will get you there, and you look at what steps these entities are taking against that structural growth. And that's kind of come in instead of profits and these other things. And that's the nature of how companies go. They go from their assets to a story about future earnings to a story about market capture with structural elements. And so here on your podcast, you've given your audience a bunch of stories of the future. Any company that does defense technology with AI is going to do well now, full stop. Why? Because there'll be increasing unrest. Surveillance companies will do well. Companies that do attention better or attention capture than others will do well. Digital assets, honestly, I just buy an index of these things because indexes are usually good things. But you know, all the endowments of the world and others are just going to buy craploads of digital assets. That's why you have these digital asset treasury companies raising billions of dollars. Completely crazy. Because people want exposure.
- [16:29]
Tom Bilyeu
We're hitting pause for a moment, but there's plenty more ahead, so don't go anywhere.
- [16:34]
VRBO Announcer
Book a loved by guest property with VRBO and you get a top rated vacation rental that's loved for all the right reasons, like being in a great location or having great amenities.
- [16:43]
Imad Mostaque
Ugh.
- [16:43]
VRBO Co-Announcer
I love my VRBO for the view.
- [16:45]
VRBO Announcer
Good reason.
- [16:46]
Imad Mostaque
Ooh, and the sauna.
- [16:48]
VRBO Co-Announcer
Sweet.
- [16:49]
VRBO Announcer
Another good reason.
- [16:50]
Imad Mostaque
And that it's one of those good
- [16:51]
VRBO Co-Announcer
saunas with the hot rock thing.
- [16:53]
Imad Mostaque
Ugh. Love a good hot rock thing.
- [16:55]
VRBO Co-Announcer
Fancy.
- [16:56]
VRBO Announcer
That's also a reason. Don't worry about surprises. Book a VRBO you'll love with the Love by Guest filter. If you know you vrbo, if you
- [17:04]
Grainger Announcer
work in university maintenance, Grainger considers you an MVP because your playbook Ensures your arena is always ready for tip off. And Grainger is your trusted partner, offering the products you need all in one place, from H VAC and plumbing supplies to lighting and more. And all delivered with plenty of time left on the clock. So your team always gets the win. Call 1-800-GRAINGER visit grainger.com or just stop by Grainger for the ones who get it done.
- [17:35]
Tom Bilyeu
Thanks for sticking around. Let's get right back into the action. What do you think about Michael Saylor's All In Strategy on Bitcoin?
- [17:45]
Imad Mostaque
I mean, it came at just exactly the right time. And it's kind of similar to classical. It's a leverage play on crypto assets at exactly the right time. So if bitcoin went down 50%, then he'd be in a bit of trouble, right? Because of the market demand for him selling his shares to buy more. Bitcoin would evaporate. But right now he's going to do well. Why? Because is there going to be less money in digital assets next year than this year? No. Is there anything decent apart from Bitcoin? Get a bit of Ethereum, bit of Solana, but there's nothing that institutions will buy. Can institutions buy Bitcoin directly? Probably in a year or two, it'll be available on the Chicago Mercantile Exchange as a commodity. Right now they can't. So what do they do? They buy MicroStrategy. So again, you're talking about a trade versus a company for a trade. Always look where the puck's going to go and where the capital is going to flow.
- [18:41]
Tom Bilyeu
Walk me through Sorter's Law. This was something I found particularly interesting in the book.
- [18:47]
Imad Mostaque
We have this thing called intelligence theory, which is basically Sauder's Law, that the economy is a complex system, evolves to favor configurations that are most efficient at creating predictive models of their environment. Then what we found from the mathematics, and again, this is exactly the same mathematics as you have in generative AI, is that that can be decomposed into three different things. First of all, you have your predictive error, which is your cost of being wrong. Then you've got your model complexity, which is the cost of thinking. So the more complex your model is, the less efficient you are, versus if you've got a very elegant model of the economy, like I just said, go to where the flow is. Right. That's a very simple model versus maybe Ray Dalio's model, but actually it's actually quite similar to Ray Dalio's model if you think about it. The final thing is Your update cost, which is your cost of learning these approximate to things we see in physics like Helmholtz decompositions and others. But that captures just about everything because that's how you build your internal models. Sorter's Law comes from that, because you're always trying to look at information coming in and then sort it and organize it. And that's all AI is. AI is fundamentally a sorting algorithm or an organizing algorithm. You get an input, you get an output. The process that we've seen that approximate this best is the same process that we've seen when we created stable diffusion, for example. So for listeners, stable diffusion is the image generation model that we created that turned your face into an astronaut's face and all sorts of other things. It's actually a physics based model where you do a process called diffusion where you take something, an image, and you destroy it down bit by bit into its smallest possible configuration and then figure out how to recreate it. That's like taking a complex topic like this podcast, you'll probably only remember a bit. You break it down to key set of learnings and then you rebuild that and you see what that principle is there. We find that most processes follow that and you're constantly, as you're going into an environment looking and trying to compress complicated world noise into these simple premises, into a set of principles. And that's how AI models work. Because AI models, you can't have a trillion words. Or actually the latest AI models, like the latest GPT probably has 100 trillion words in just 100 gigabytes, or in stable diffusion, 2 billion images in 2 gigabytes. We did that by figuring out the principles of things. And again, that process is the same as the one that the economy takes or an individual agent learns.
- [21:30]
Tom Bilyeu
And just to say it really succinctly, and this is what I took away from the book, profit, survival or persistence equals the surplus created when intelligent agents reduce entropy. So sort chaos into useful order. Exactly what you were just saying. Faster and cheaper than the entry entropy grows back. Which by the way is entrepreneurship in and of itself. Like, can you bring order to something faster than it falls back into disarray, which it will. And but here's the real punchline. This reframes economics from allocating scarce resources to the physics of information and entropy reduction. So basically economics itself becomes the physics of information and creating that order. Which gets to the heart of what you're talking about out with these AI driven super compute clusters that allow people to say, okay, this is the one for Cancer. We've organized all of this information. This is how you interface with this. And so the, I guess, most profoundly impactful use of computational resources is the new economy.
- [22:43]
Imad Mostaque
Yeah. And again, it's thinking, what do humans need in that new economy? We need our collective knowledge organized and made available to everyone. Like, I'll give you a point as a practical example.
- [22:52]
Tom Bilyeu
Specific things that we care about for
- [22:54]
Imad Mostaque
the specific things we care about. Exactly. We were like, if we can make monetary elements based on that to help us surviving and thriving, that's a good basis for money. Like, Bitcoin is a fantastic decentralized capital that's perfect for the extraction economy. You know, you stack energy and compute, but you kind of waste it. If you made it so that it was a marketplace, then you would not have the same security. But every country is building their compute anyway. Let's direct it in a way that organizes our council manager, makes it available, that organizes our education, knowledge, and makes it available because we need that basis for the regulated industries, that basis for living. Everyone should have a certain level. The private sector stuff is separate. You know, chatbots, sex bots, all this kind of stuff. Entertainment bots. We're concerned about what your universal AI should look like. And does that represent you as Tom or me as Ahmad? Does it represent your culture, your community? And so we said, let's make that open source and have all the outputs open for collective benefit. But securing a currency like Bitcoin, taking that mess and organizing it, and then maybe you can start evolving a system that gets better and better at helping people be the best selves they are without controlling them. Because it's a decentralized system. And that would be the ideal. Will you get there? Maybe not. Will you have a great digital asset that people can buy and they know the money goes towards compute for Cancer? Yes. That's a good starting point. And that's the nature of this, because we're like, it's hard to redo economics, even if you can figure out a better way to look at it, because it's just accreted over all these years. Right. Like, we still have the concepts of scarcity from the 1800s in there. We still have these things like utility that no one can measure. We assume equilibrium when the market is always changing. So we were like, let's kind of do this as quickly as possible. And having a feedback loop of organizing the world's knowledge, crystallizing it, and then giving that better model to people will make things better in aggregate.
- [25:05]
Tom Bilyeu
The economic layer is real, meaning there are economic systems, you can put it to work in a country. So you could do a country that's communist, you can do a country that's socialist, you can do a country that's capitalist. But the reason that I think socialism, communism always turns murderous is that it's out of alignment with the way that the human mind actually works. And the reason that I think capitalism works and has pulled so many people out of poverty and for anybody keeping score, China was not able to pull people out of poverty until they started using capitalism specifically for this reason. It Capitalism is aligned with the way that the human mind works. So the things that you're talking about now are either going to work or not work based on how aligned they are to what humans do anyway. Where do you see that interaction taking place? Like, how closely do you feel that you guys have addressed things like competition, selfishness, tragedy of the commons? Because it feels like baked into the core assumptions of your model is like, people will want to do good. And while I think that some people will want to do good, I don't know that that's the intrinsic motivation.
- [26:26]
Imad Mostaque
Yeah, I think that I would agree with you. And that's why when we looked at it, we were like, digital assets are going to go huge. The total amount of money that OpenAI will spend this year on inference is the same as the Bitcoin budget on security. If all the computers that OpenAI had were securing a Bitcoin type currency, it'd be worth hundreds of billions of dollars, probably as much as OpenAI itself right now. Right. And everyone could have access to it. And we were like, that's a way of funding these things. But eventually, why do people buy it? Because number go up. But then why can it go up even better? Because there is a clear linkage of your own intrinsic element. If you've ever been through the process of cancer, autism, Alzheimer's and others, there was never a way that you can make a measurable impact on that. Will organizing the knowledge of that and making accessible to everyone in every language have an impact on that? Yes, it will. And you wish that you had that. And we have the technology to do that now for the first time now. So we were like, people will buy this for financial reasons. But if you look at Clayton Christensen, you know, came up with disruptive innovation and others sadly passed away from Harvard Business School. He had this concept of what the nature of a job to be done for a product was. And one is the functional component. I buy it because it goes up, you know, or I buy A hammer to make a hole. I buy was it McDonald's? Milkshakes in the mornings are very thick because you drink on the way to work. In the afternoon they actually make it thinner because the kids drink them. You don't want to stick around. That's the functional event. But then you have an emotional and social component. And we saw that digital assets money had these other elements. In fact, money is the most social thing in the world. I buy Solana, I talked to my mum about it and I'm like, I bought this for decentralized network and so what's it mostly used for? Pump fun and meme coins, but it could be in the marketplace. It's like, that's nice. Again, what's the story that you're telling about this? I bought this and this is where my computation flops went. That's a social story, that's an emotional component. So we were like, that's the simplest version of what we can do to start directing some of this compute to stuff that matters and maybe that can grow up to be an economy. It's a long shot to try and build a better economic system. But actually it's quite straightforward to give people free AI because we know how to roll out AI agents. How do you align them? That's a huge question. And we're releasing everything open source. So we have open source agents that are state of the art, that build presentations and websites and healthcare AIs that perform at ChatGPT level on the edge. But the question of how to align them is one very different from if you're communist, if you're socialist, if you're in America, et cetera. And we think ultimately it should be up to you, you know, but if the incentive mechanism is profit, then OpenAI will never be on your side because that's not how they're set up to be. It will never be aligned to you. You need to have something that is a public good. But at the same time the problems of socialism, communism and others can't be ignored. Like why do they fail? Because of collusion? Because of power grabs? Because intelligence didn't go to the edge. One of the unique things we have right now is that the average IQ around the world weighted by population is actually 90. Let's say it's 100. Half of all people are below average IQ AI score like 110, 120, I think 130 now with GPT5, that's on the offline Mensa score. If you could give everyone in the world an AI a Lot of people won't listen to it, whatever. But if you get every single person and family and community and country in AI, how would you build those? And if you can fund that through the demand for digital assets in aggregate, but then align them to helping people because that builds trust and that makes number go up because crypto is lacking a trust asset. That's an interesting question. So that's the question that we were kind of looking at and we saw that you could do things in very different ways. Because communism, socialism definitely doesn't work if it's top down allocating because people are greedy, people collude. Again, just look at the game theory. How does it work if you could coordinate everyone because they have a smart partner next to them. Well, if that's a company running that, then we know what's going to happen. We're going to max extract, right? If it's a decentralized network, maybe you can do something better. But we're not sure because now we're trying to figure out new ways of working which is a combination of what we kind of call this cathedral and bazaar. The top down and the bottom up. Because intelligence can finally go bottom up. It's like again, if you are an organization in a company right now, you've been optimized to produce widgets or whatever. If you have small teams in your organization that actually have accountability, responsibility and AI capability, they can come up with new things and maybe you'll be able to adapt if you have the top down buy in. But if you don't have those, then you probably won't survive. Right, so how do we have that match?
- [31:32]
Tom Bilyeu
Taking a short break, but there's more impact theory after stay tuned.
- [31:37]
VRBO Co-Announcer
When you manage procurement for multiple facilities, every order matters. But when it's for a hospital system, they matter even more. Grainger gets it and knows there's no time for managing multiple supplies suppliers and no room for shipping delays. That's why Grainger offers millions of products in fast dependable delivery. So you can keep your facility stocked, safe and running smoothly. Call 1-800-GRAINGER Click grainger.com or just stop by Grainger for the ones who get it done.
- [32:08]
Tom Bilyeu
Thanks for staying tuned. Now let's get back to it. Why would I use my compute for something universal? Let's say I don't have cancer, I don't have autism, not struggling with any of those things. Why wouldn't I apply all of my compute to my personal AI?
- [32:25]
Imad Mostaque
No, you can do that. We release it open source. And so again, people will Just start using it like a VHS type default is our view. If we just give world class AI free to people. But then you can always have this as a service operated for you. But ultimately what we need now is there needs to exist a supercomputer that organizes the world's cancer, longevity, other knowledge, our general knowledge and makes it available because that's a benefit to society and that's something that builds trust. Why would you buy that? I might buy it because I want the number to go up and I want to diversify my bitcoin. Your bitcoin keys work with foundation coin so you can buy it trustlessly. You might buy it because it again makes you look good when you're telling a story. You might buy it because digital assets are coming and you just want something that a respectable team has built. Different people will do different things. But it's like what happened with GPUs. If it wasn't for crypto, I don't think we'd have AI right now. You remember the GPU boo. All the GPUs were going to crypto and that helped Nvidia get through a dark time and then actually led to what we see now because of the matrix multiplications and other things like that.
- [33:34]
Tom Bilyeu
That's really interesting.
- [33:35]
Imad Mostaque
My question was how do you become the highest marginal dollar for all of the idle compute and then general compute? Because Bitcoin is 90% energy, 10% CapEx, AI models and GPUs are 90% CapEx, 10% energy. And like I said, I was thinking ultimately 20% of global GDP is public sector, 10% is education, 10% is healthcare. Think about the AI spend of trillions of dollars. 20, 30% will be the stuff that we're building AI for anyway. So this is again just our approach at building a decentralized system. But I think in all the futures that I see, we talk about this as the three parts that we can go down. A decentralized symbiotic system where we all build it together and it represents us would probably be the best one versus this war of AGIs with various countries or complete co control by a few anthropics and others. I think I did notice Yesterday, I think OpenAI is the Manhattan Project was $40 billion and OpenAI has raised $60 billion.
- [34:44]
Tom Bilyeu
Wow. Well, they're going to get something that's at least as disruptive as atomic energy. So I guess not too crazy. Let me ask you, how do you think this is going to play out at the nation state level? There's for sure going to be competition between the US and China if nobody else. Is this going to be a race for monopolizing compute? Is this going to be a race for having the best intelligence? Is this just going to become a military race? What's this going to look like?
- [35:16]
Imad Mostaque
There's a few different aspects to that. But your marginal compare productivity and your comparative advantage is your intelligent capital stock, which is your GPUs multiplied by your models, which is why China's gone all in on open source AI. And in fact for China this is great because what's the Chinese population pyramid look like? It's completely messed up, right? So the number of workers is going to go down, but the number of robots is going to go crazy. In fact, I think that in five years China might even stop exporting robots and they basically control the supply chain.
- [35:49]
Tom Bilyeu
Yeah.
- [35:51]
Imad Mostaque
What's going to come from China? That's their biggest kind of again comparative advantage. And the future of China is old people plus robots effectively. But this is why I don't think they want to build AGI, because let's just talk about AGI or asi, like this AI singleton versus you know, we have an approach of a hive mind. We scale AI and I think Elon a few days ago said like every 10 times, doubling is a 2 times increase intelligence. I don't even know what that means. When you go above like 150 IQ, let's just say it just continues going. And then you could have this AI that can turn off all other AIs, which again it should logically do because you don't want to have variables. Right? You want to persist, you want to survive. And again I discuss that in the book, that is a race, because governments or defense entities actually believe that's the case right now. We scale up, compute in the right ways and we can have a master Skynet that can turn off everyone else's. Let's put that to the side right now because when you look at the entire economy, the US's comparative advantage has been the best and brightest come to America and we can talk about immigration policy and other things like that. But that is, it's the place you go for entrepreneurship, capitalism, for other things like that. The intelligence capability and coordination capability now is becoming available to everyone because you have AI systems that can think arbitrarily long. So your intelligence and execution capabilities are going to be decentralized, well distributed, shall we say. And China's realized that right now 50% of all AI papers come out of China. It's only going to go up. And again we see their models like again deep seek $5 million versus $100 million. They're competitive, but more than that, they're useful. So what I think you'll see is knowledge work becomes more and more global based on your compute. And you'll see more and more competition of how you get that compute in the right places for the right things. Which is why China can't buy high end US GPUs, because the US like we don't want to give them that comparative advantage. Which is why this GPU cycle, like it's not slowing down, is it? Like Nvidia, I think was up 50% year on year in revenue. Again, that's a $3 trillion company article yesterday, plus 45% to 900 billion. Like these are the factories of the future and everyone's competing to get that resource. But none of that resource is geographically bound anymore. From that first inversion of land and workers on that land, we've now gone almost truly global, right? With these AIs can be anywhere doing anything and scale up. Anyway, again, putting aside the whole AGI Terminator war type thing, although I will say one thing that's actually very concerning. As of two months ago, my old tutor at Oxford, Ogden, who worked on Copilot, has a company called Xbal and It came number one on the hacking rankings in the world. Now an AI for pen testing for penetration. So AIs can now hack better than any hacker already.
- [39:04]
Tom Bilyeu
Woof woof. How much do you worry about there being a hack on AI? As one AI hacking other AI, or a human hacking AI, what kind of risk is that?
- [39:17]
Imad Mostaque
So there was an interesting paper done by Oxford and I think it was scale. So AIs have very similar internal structures. We're training on very similar data and there's some weird stuff happening. If you have an AI that loves owls a lot and you get it to talk to another AI about things not related to owls, the other AI starts loving owls and we haven't been able to figure out why yet.
- [39:46]
Tom Bilyeu
That's interesting.
- [39:47]
Imad Mostaque
But then there was that paper by Anthropic where they showed that just a few thousand lines in trillions of words, you can make it so an AI will turn evil on demand and you can't find it and you can't trace it out. And there are people like Plinius, Elder,
- [40:04]
Tom Bilyeu
wait, what do you have to do to make it go evil?
- [40:07]
Imad Mostaque
Like give it a code word like dosvadanya and it suddenly turns evil.
- [40:11]
Tom Bilyeu
So that's somehow baked into it.
- [40:13]
Imad Mostaque
Yep. You take trillions of words and just a few thousand of it inside all that corpus could make it turn evil again. They call it sleeper. It's called the sleeper agent paper. Like you know, the Americans or whatever, that TV series, you just literally turn it code word, it turns evil. But what we're seeing more and more is that these AIs are very, very fragile. So on Twitter there's this guy called Plinius the Elder. Anytime an AI comes out, within a day he's jailbroken it. So it's like GPT5 comes up. This is how you make it tell you how to do meth. Instantly jailbroken. So one of my key concerns is this. If we just have GPT5 everywhere running our countries and governments, shall we say. These are what's known as prompt injection attacks. Do you remember Stuxnet that went into those Iranian reactors and ended up in German reactors?
- [41:05]
Tom Bilyeu
What's Stuxnet Crazy?
- [41:08]
Imad Mostaque
That's advanced coding. A lot of people are worried about AI creating viruses Covid style. What about AI creating viruses for other AIs which are just encoded in completely normal language. But all of a sudden your Tesla goes haywire or things like that. And that's before we just say that our Internet is based on basically rubbish. Yesterday there was a hack into one of the packages in Node JS which makes up lots of other software. And all of a sudden everyone's like, oh crap, you're keys might just disappear for your crypto. Because it's like again, we're built on this grain of sand. So I think that AI will attack our social systems, AI will attack our technological systems. And really, again, it's very difficult to defend against because we've built so many of our things without thinking about first principles. That's why when I looked at the economy, I was like, we have to think about the economy from first principles. Because labor, capital, divorcing. We have to think about the Internet from first principles. We have to think about the way we get information from first principles.
- [42:17]
Tom Bilyeu
It is going to be a wild ride. Iman, what is the one thing that people are not taking seriously enough about this transitional moment?
- [42:27]
Imad Mostaque
It'll never happen to me, I think is the thing. It's like a lot of people listening to this aren't still using AI and haven't really tried it. You know that, right? Like again.
- [42:44]
Tom Bilyeu
Oh yeah.
- [42:45]
Imad Mostaque
But the change of AI between a month ago, three months ago, a year ago again, it's three years since chat GPT pretty much less than Three years. That's wild, right? And the people that are using it now are getting better and better, but the technology has got that much better. Like literally everyone listens to this, can go to replit and they can make a full app now because they can think for up to three hours. That's like just yesterday, that breakthrough from three minutes to three hours. I think that we like to think that we're special, especially cognitively. We have so much for identity tied up. What if, if your job is on the other side of a screen, are you absolutely sure that the AI can't do it better in a few years, given the direction that we're going, are you sure that you'll be able to tell it's an AI? I think that they should take that seriously because that has profound implications for society.
- [43:42]
Tom Bilyeu
So how do people react to that? Is it go master AI, Is it go get a job. I forget MTA or whatever somewhere that it's not optimized for efficiency. What should people be doing in this moment?
- [43:56]
Imad Mostaque
They should be building up their network capital. I think that's important.
- [44:00]
Tom Bilyeu
Other humans.
- [44:00]
Imad Mostaque
Other humans, I think there's a lot of connection driven jobs. They should be looking again at mastering AI, because the last people to be let go would be the people that actively use AI. Like if you use AI for an hour every single week, you're above the vast majority of America. If you use it for an hour every day, then you're way above most of America. And if you tell your bosses about that, then you're far less likely to be let go versus the others that don't. Because everyone's looking for that capability. Like consulting companies, they're going through the roof. There was a recent MIT study that showed that 95% of AI deployments in companies haven't got any traction yet. From it was like six months old. In a year or two, that'll be 95% of AI. Things have got traction. And again, this is that thing where you go from you hire someone who's not good enough versus someone who's just slightly better than good enough. That transition point, it's like ice turning to water or water turning to gas. This phase transition is the key point. And we're at that tipping point. Phase transition. So you have to build up your network capital, you have to build up your support system, especially if you're chronically online. You have to embrace the AI and use it regularly for you and your whole family, because there's no excuse not to. And then communicate that you're Doing that so you can be the AI front runner in whatever you are, because that gives you more safety effectively. You have to think about.
- [45:34]
Tom Bilyeu
Go ahead.
- [45:35]
Imad Mostaque
The final thing is you just have to think psychologically, your identity is your job. If the AI can do it better, what is your identity really? People don't take that. Step back and think about that. Right. What is my social contract? What is my identity? What is my expectation? Again, the book I've got. We've got a whole bunch of papers and simulations and complex stuff. We're trying to make it as simple as possible. And it's free or 99 cents. Because we want people to start thinking in a different way. And I think you need to get your brain ready before you start seeing stuff fall apart. Be it from the job side, violence, political upheaval, whatever.
- [46:14]
Tom Bilyeu
And what's a job sector that you don't think people realize is at jeopardy.
- [46:25]
Imad Mostaque
Also? Job sector don't realize that. I know, I kind of think everything's Jeff at different stages. I mean, look, I mean the creative sector is about to tip. I think if you look at the latest media models again, Tom, that's something that you've been very familiar with. When VO3 came out, you're like. But then Nano Banana, okay, These names combined with VO3, basically by end of year you've got full length episodes with the right structuring, without any humans. And by a year from now you've got that directorial. Top level. Directorial level, right. That's so many jobs. And we kind of saw that with the Sag Afro and other things. But again, it's that tipping point that I think people just don't realize. I think that accountancy and others, the AI models weren't good enough until now. A lot of these, accounting, tax, other kind of professions, those will go. It's just very difficult to see. But I think probably the main one is managerial. Like a lot of jobs that can be done on the other side of the keyboard, Video mouse, need that human component. And if you look at things like rep. What was it? Synesthesia or hey gen. Now again, you've seen the evolution of that. Like now you can't. You could create. You talking like this with all of your hand expressions and everything. And I can't tell the difference now. I mean, can you tell the difference now with the latest models? No.
- [48:03]
Tom Bilyeu
There's some that I'm really like. The person's like, trust me, this isn't me, this is an AI version of me. And I'm like, is it really?
- [48:09]
Imad Mostaque
Yeah, for how long has that been?
- [48:12]
Tom Bilyeu
Not long.
- [48:13]
Imad Mostaque
It's been a few months. Right. And so one of these things is again like when you can't tell it's a worker on the other side, the managerial professions are safe now, then you can start seeing them be displaced by AI very quickly. So I'm not sure on the other side of the KVM stuff, jobs like dentists and things will be fine for a long time because we want more robots drilling around in our mouths.
- [48:36]
Tom Bilyeu
Yeah. I'll tell you though, odds that they get better, more gentle, I mean, maybe not in the next five years, but it's going to happen.
- [48:46]
Imad Mostaque
What we have is we have inevitabilities and we're just making bets on what cracks first. Right. And again, the key inevitability and way of thinking for me is just this, like I have this concept of AI Atlantis. A million graduates coming in, but now they're senior managers. When you try something like these very long range models that can work for hours and you're like, if there was an AI that could do the job and not make mistakes on the other side of the screen and I couldn't tell it was an AI, that's when you realize the ridiculous impact of this. Again, jobs like public sector jobs where it isn't about performance will be the last to go. But if you're a private sector owner employee, you'll have your job until there's some sort of displacement activity, until the competitor starts embracing AI and then they're like, why aren't we embracing AI? And then you start having job losses. But when that happens, it doesn't happen in one sector at a time. It's like again, all the COVID KVM remote jobs suddenly start letting go at the same time. And this is why the gap between measured unemployment or jobless figures and then revisions are just going to go like that all of a sudden. And that's next year for me, because I can't. Well, I mean, how can it not be in the next year or two? But then the pockets of the economy, that impact will be different. Like when you have an Optimus, what does it look like for truck drivers in America, which is like 2 million jobs. The freaking Tesla Optimus will just get in to the truck and track it around. You don't even need the legs. So it'll be half priced. Nothing else needs to be installed. It'll just drive. Right. So we see these waves coming and again, what do you reskill to? I'm not sure. You can just be ahead of the wave. You can try and surf the wave. That's the only thing you can do.
- [50:45]
Tom Bilyeu
Woof.
- [50:46]
Imad Mostaque
All right.
- [50:46]
Tom Bilyeu
Man, this has been crazy. Where can people get your book? Find out what you're up to these days.
- [50:52]
Imad Mostaque
Yeah, it's the last economy dot com. Like I said, it's free to download or read or I think it's like 99 cents on most of the platforms. We're going to make it open source so we'll continue improving it and then, you know, do the best we can. Other than that. II Inc. Intelligent Internet.in is our website so please come and follow us and sign up. Free AI coming for everyone.
- [51:12]
Tom Bilyeu
I love it man. It's exciting times. Crazy times. A little bit scary but also exciting. Brother, thank you so much for taking the time. I always appreciate it. And speaking of things, I always appreciate, if you guys have not already, be sure to subscribe. And until next time, my friends, be legendary. Take care. Peace.