Transcription
Hello my friends. This is the first in a new series. And trust me, I'm here, but my back is toast. Let me do a quick wave. So, I'm actually sitting down, but I'm here. Anyway, thank you all for coming. Uh, this one is the first of a new series. And you know, I've been talking about AI since 2022, since Cha ChibiT dropped, but now I firmly believe AI is poised to be the most significant wealth generator in history of mankind. And yes, while we made a lot of money with cryptocurrencies, I do believe this is a way bigger opportunity. So, my focus has always been both crypto and AI, the two big narratives. But now, um, I am doing a dedicated weekly show called AI Impact, and it's probably going to hit on Thursdays. So here you are. This is the first, and again, this is a space that is transforming the world. It's never been more clear to me, and it's also critical to gain exposure to the space for you and your families. Okay, I'm going to, hopefully, help you navigate the opportunity and, uh, ways we can all capitalize. So, let's get into it. Big thank you to the mods in the chat. Of course, not financial advice, and nobody is ready for what is coming, ladies and gentlemen. Absolutely insane. All right. Uh, and shout out to Patreon, too. H, this is from Joy Expanded James. I love the sweetite and Patreon. Woohoo. Thank you for being there.
First, let's talk about this. This is, uh, a little bit of, uh, data here. I got to, sorry, the dog is eating the carpet. Would you believe just night? Anyway, I'll get back to that later. What? But I just want to point out as well, these is is a list, uh, I assembled of the top 10 most valuable private companies on Earth. Okay, if you see here, all the companies, number one, OpenAI, by the way, XAI is valued about 300 billion or 200 billion, depending on who you ask. OpenAI is now valued at 500 billion, half a trillion dollars. But XAI is better and has more compute and does performs better in exams. Yet, OpenAI is worth half a trillion dollars. So, that's a mispricing right there, which we like. So, if you can get some XAI at a $200 billion valuation, buy it. Trust me on this one. Financial advice right there. But the point of this is, if you look at all these companies, all of them, 80% of them are AI based. We've got Whimo in there, Figure AI, Anthropic, Bite Dance, XAI, SpaceX. Although SpaceX's kind of communications, communications are critical for AI and information dissemination, which is all AI adjacent. So, that one could be a little bit of a stretch. But what I also want to stress is I made a video last week that hopefully people understood, and that is the total addressable market of AI versus Bitcoin. A lot of people think Bitcoin is bigger than AI. No, it is not. Not by any stretch. Okay, this is a model I put together aggregating all the total addressable markets by industry of the world and the actual exposure and the actual attribution of profit potential from AI impact, and nothing comes close to what we are seeing now. And this, by the way, is just Tesla's total addressable market exposure in blue. Red is Bitcoin and crypto. So when you look at that very quickly, you'll quickly understand this is a far, far bigger space.
Now let's get into other stuff. A quick reminder, yes, money TAM, total addressable market today dwarfs AI total addressable market today, but AI is early, and we like to get into things early here. But the AI time growth is far higher, 29% versus money growth 7%. And AGI is going to change the whole world, reshape labor, finance, taxes, investment advisors, everybody out there. So again, lock in your gains and think about exposure. Now, I want to shout out to Adam Jonas. He has restored faith for me. He's from Morgan Stanley. He has restored faith for me in the stock analysts, and he's one who actually truly gets it. But he had this cool little chart in a video he put out, and basically, knowledge has already been consumed by AI. The world's knowledge has already been eaten up, and it's now available to us at our fingertips. Everybody, if you are a lawyer, a tax accountant, even a stock analyst, a physicist, a writer, whatever it is, that's already already been consumed by AI. It's game over, pretty much, according to our friend here. But the next part is the big part with the even bigger TAM. This is the TAM sheet again, referencing here. This is the physical economy. This is embodied AI. This is the next leap. Okay. This is machines with sensors, sensors, actuators, um, self-driving cars, robots, everything needed to navigate the physical space. Everybody. Okay. So knowledge economy is already been licked. Next is the physical economy. Can't say how much I love this chart. It's brilliant. And the race for vision is data. Data is everything for AI. And to get as much data as possible is the key. We'll talk towards the very end, a new data play that could be very interesting for some alpha too. But Adam Jonas highlights that with that AI without real world data is worthless. He mentions in his video, like it's like a tuna you cannot catch. But the next frontier is massive scale collection of high-quality video and sensory data. Again, we have all the written data in the world. It's already been consumed, and the best way to do that is with real world things like cars pulling in data, and that is absolutely critical. We'll talk about the car company later, which is not a car company.
Also, we had the earnings yesterday from Nvidia, and Jensen Wong had some very interesting things to say. One of his new things is we are now entering the second inning of the AI revolution, and he focused on what is so critical because everything ties to energy. He's focused on, um, having a thing called performance per dollar. That's their new metric, and it's basically all data centers are power limited. So the more you can squeeze, the more performance you can squeeze out of your power, out of your GPU, the better off you are. And data center load is expected to use up 40% of net new electricity demand until 2030. And if we are in a world that's limited with power, okay, the performance per watt directly drives revenues. And the performance dollar per Jensen is the performance per dollar is so incredible, you will also have extremely high margins. So think of energy. We will talk a bit about energy for a second here. But before we'll talk about data centers. Nvidia just smoked Intel data center dominance. Intel has been running data centers for decades. And you can see here how they have shrunk and how basically just in the space of a couple of quarters, Nvidia has leapfrogged them up to 80 billion where Intel used to make 80 billion. It's funny because it was a deal done between the US government and Intel for 10% share, which is kind of weird. But anyway, if they had the money, they should have bought 10% of Nvidia, but they couldn't afford that, of course. Maybe 1%. Hello everybody out there.
Now, back to energy. Yeah, this is the chart of electricity generation. I've shared this before, but it goes back in history. You can see the tipping point there was 2010. That's when China surpassed the red line, which is, I think it's US electricity. China is the blue line. And since then, the chi, the US production of electricity has been flat, whereas China has gone to the moon. It's gone up two and a half times and it continues to grow. And as many people believe, those with the most power will have the most AI. That is basically it. That's all she wrote. And the US has to do something about the power creation real fast. And this is a big flaw with the so-called big beautiful bill. They're building all these tax credits and data centers without understanding that the data centers need the fuel to power the data. And that's energy. And they don't have the energy there. And it's a complete disaster. And that's one of the reasons why people like Elon Musk are very upset.
Anyway, moving on from there. Speaking of China too, the other thing that's very important for this real world AI, okay, back to Adam Jonas here, the physical economy, is one thing we all need, and that is magnets. Okay, the Chinese had the foresight 25 years ago to understand the importance of rare earths, which are critical to magnets, to anything that's motorized, any type of actuator, and any type of robot, or any type of car windscreen wiper, you name it. And China has invested heavily in the heavy elements that come from clay-rich deposits, and that's it. If you look here, China controls 75% of rare earth magnet market, and they have the ability to turn off the European car market, the US car market, etc. So, the rest of the world needs to figure out exactly how they can become self-sufficient in this area, which is very important for the future. But let's get back to AI and forget those parts, and a little bit about data centers. And it's funny because for years and years and years and years, I had to dispel FUD around Bitcoin mining and how much power it used. Well, lucky here, data centers will use the same amount of power as India in the year 2030. That's a country with over 1.4 billion people. That's much power data centers will use. Absolutely bonkers level of power. So I don't know what happened to the IMF, WF, all the concerns about Bitcoin mining and Bitcoin should be banned because of power usage. I don't hear them saying that about data centers because this is no longer a question of preserving the environment or any of that stuff. It's an arms race. Again, those with the most power will have the best AI and ultimately control the world. Just remember that. And currently, USA controls 69% the compute. I'll share some more data on this in a second. China has about 14, 1.5%. Others 9.4%, and then the EU 6.1%. The US has more than 11 times the compute that the EU has, and Japan 1.2%. So, I'm hoping Europe can get their act together pretty soon and at least try follow China as they go up the chain, too.
Now, back to AI. We had a lot of weird infighting and hiring and crazy. It's like almost like a soap opera. Not that I know soap operas, but here we have news that, uh, the new researchers that were paid 100 million sign-on bonus have already started to leave the so-called toxic environment of Meta under Mark Zuckerberg. Offered them tons of money to join. Many did, about 14 did or more from places like OpenAI, etc. But now they are already leaving. And the question is why? And this is super interesting too. It shows you the money is not everything. But, uh, Elon did confirm that Meta engineers are joining XAI with initial compensation packages that are nowhere near what Meta is offering. But one of the reasons people don't want to work for Zuckerberg or they don't want to work for a firm where they're not going to win. And these top AI developers want to win. They want to be on the cutting edge. They don't want to be second best. And they also cited some chaotic culture, overlapping mandates, lack of vision, etc. And, uh, Meta was described as uniquely toxic. It reminds me a little bit of Steve Jobs back in the day. He was a very hard person to work for, but I don't know. Either way, they spent a fortune on AI, 14.3 billion on scale AI, and, uh, another plan for 72 billion in capex and data centers. What are they going to run? I don't know. I don't know. We'll see. And this is very important in the world. Um, this is from Elon Musk, who said, "Many strong Meta engineers have and are joining XAI and without the need for insane initial comp. Still great, but not sustain unsustainably high." And he said, "XAI has vastly more market cap growth potential than Meta."
Let's go back to my little handy dandy chart here. We'll come back to this slide in a second, but this again, as I mentioned, if you can buy XAI at a $200 billion valuation, which it currently was rumored to be at last month, uh, and you've got OpenAI worth half a trillion dollars, that's a 2.5x right there. That's what these employees see. And they're paid in stock comp ownership. And the beauty of this one though is, is as I say at the top here in the title, it's not money. It's about merit. Okay. Yes, Elon has poached 14 of the top AI engineers from Meta, and Meta has offered millions to keep them. But Elon has offered purpose. Okay, humans sometimes do want a purpose-driven life, which is the most important thing. It's not about money, especially when most these people are in it for the chance of developing something completely unique and special.
Now, let's talk about clusters and how big they are. And by the way, let me know in a comment below if you like this AI stuff because it's absolutely critically important. I know people just want to make a quick 10x on crypto, but again, I do believe this is the biggest TAM and this is the most important thing for you and your family's future. So, let me know if you like this AI segment once a week. But these are the world's largest compute clusters. You can see all the big rings. The big one in the middle is XAI, but there's one to the right of it, too, over 300,000 H100 equivalents. And this is also slightly out of date because they have even more than that now. You got Meta about 100, Microsoft about 100, Oracle about 65K, Corey 42,000, etc. But when you combine all of the Musk industries together between XAI, uh, Colossus One and Colossus Two and Tesla as well, you have nearly half of all of these clusters are owned by Musk Industries. Okay, black, I just added some arrows. Nearly half of all AI compute, and I'm not even including the inference that's possible from 7 million cars, that's growing fast. So this is one of the reasons why I believe the AI engineers, they know for them to succeed, they need all the compute, and they don't get that with where they are or other players.
So maybe that is why Meta is pivoting to go all in on smart glasses. They are spending billions on buying, I don't know, they're working with companies like Rayban and other glass manufacturers. I don't even know these brands at all, but some luxury ones, and they want to build these prototypes where basically you can wear these glasses, and everything you look at, it'll tell you what it is, which I think is a little bit weird. We've seen what happened with Apple goggles, a complete failure. I called that the day it was launched. But again, this could be, I think it's about a four or five billion dollar bet for Meta. This could be Meta's way of securing one area of the future of computing. Again, wearables. We'll see if they do succeed, but they're not going to win the AI wars, that's for sure.
And speaking of wearables and goggles and Apple, XAI is at war with Apple. Apple are suppressing greatness. They are trying to hide Grock on their app store, and half of Americans with iPhones, or half of Americans have iPhones, which means they can't actually see these applications. Anyway, um, that's what's happening now. Elon is suing Apple for this, and also ChatGBT because they don't want this guy to own everything, which I don't know, I can kind of understand. But don't worry, Apple and Tim Cook has a new iPhone 17 coming out on Tuesday, September the 9th, and I'm sure that'll save the day. Kidding. Wonder what color this new phone is. Could it be some type of orange or blue? I don't know. Pretty exciting, huh? Well, no. They've have not made it in the AI world, and now they're reaching out to other players to replace their Siri. And Apple was first with the AI agent kind of assistant thing, but they never got it to work.
Anyway, back to more stuff that's happening. And CERN Basher made a great piece on robot trucking. And now we see this new spotted shrink wrap Tesla Semi sitting outside the pilot production facility in Nevada. And this looks like it has a double cab for sleeping in it. Wouldn't it be weird? And the reason I'm talking about this is imagine just being in the truck to make sure everything's okay, but you're sleeping in the back while it drives itself with a robo trucking FST component. And speaking of AI in China, um, of course, the Chinese don't allow Grock to be installed in cars in China. So Tesla partnered with DeepSeek, and now Chinese buyers can have their Grock equivalent in their Chinese Teslas, which is great. But all the other car companies are nothing like this, and, uh, they don't even have the ability for a software-centric vehicle to deploy such AI as well. Anyway, interesting.
Now we do have a big revolutionary moment in AI as well coming. That's FSD 14. Covered this a few times, but it's going to be two to three times better than the human, and version 15 will be 10x better than the human. Here we are. We have arrived. Naysayers will say, "Oh, it doesn't work." Uh, believe me, it does work. And I trust it more than myself at driving. I consider myself a very good driver. But what I do not trust is the people around me on the roads. And I do believe FSD has a better 360° view of the dangers coming at me. Therefore, I trust it more. Now FSD, shout out to our friends in Australia. It's apparently it's coming. And it now costs 10,100 Australian, which I think is about 6 or 7,000 US. And the subscription model is coming soon. And the Australians love their technology. And this has been tested extensively in places like Sydney for quite a while. So now we have a second continent with FSD because you can use your FSD in Canada, Mexico, and the US today. Now Australia will have it too. Imagine how easy it would be, especially for people who've got those long journeys across the outback in Australia. It'd be fantastic to have. And the Australians also want the Cybertruck. They're very big outdoorsy people, and there's a lot of outback, and they need a truck for that. And Cybertruck also just launched in South Korea. So self-driving hopefully Cybertrucks in Australia soon.
Now the other element I mentioned at the beginning is global comms. Uh, this is the Starlink global comms scurve, uh, that I call it. Uh, this chart is from Sawya Merit, but it just shows you how the growth of customer adoption is accelerating. The more they grow, the more satellites they have, the more scale they have, the more they can reduce the price. And my theory is right now, we already have price par with traditional telco or broadband providers. Okay? If I was them, I'd be extremely nervous here. And Starlink has the power to undercut all of these players in a very big way. And you have the ability to be a nomad, running around with this thing plugged into the back of your car, etc., like on a Cybertruck. So this tells me that the global comms, which is also hackproof and secure, is super interesting, and it will disrupt everything as we go forward too.
Now this is the other piece as well tied to AI is Elon plans to rewrite all enterprise software. He's got this new XAI project called MicroHard, that could basically simulate Microsoft's entire operations, all their enterprise software, uh, operating systems, their Office suite, whatever it is. Haven't used that in years. I actually use Google stuff. In fact, this stuff is created using Google, too. Um, but MicroHard will leverage Grock to create these specialized coding, image, video, and testing agents and simulate human users in virtual machines to refine products. And this is absolutely crazy. Taking firm aim not only at Meta, at Apple, now Microsoft too. Elon Musk is not making many friends out there. And this XAI MacroHard will be powered by XAS Colossus supercomputer, which is the biggest supercomputer on the planet, and again, hopefully to replace entire software companies. Think Salesforce.com, think Oracle, think Microsoft, uh, madness what's coming.
And if it couldn't get even better, coding as well. Well, for all the coders out there, there's a new thing called Grock Code Fast One, and that is a speedy, economical reasoning model that excels at agentic coding. Okay, it's free on GitHub, Copilot, Cursor, Klein, Kilo Code, Root Code, Open Code, Windsurf, you name it, it's everywhere. Okay. So again, it goes back to the disruption of people who may have may have spent the last 20 years becoming really good developers. Well, all of those skills have now been somewhat replaced. So now it's about pivoting to asking the right questions and building the right codebase around very specific needs.
And if it couldn't get even more nutty, guess what else is happening? Well, AIs are now creating new AIs. Okay, AI models just invented better new novel AI models. Uh, thanks to some Chinese researchers. By the way, Chinese have the best AI experts on the planet. Not sure how that happened, but they are the best. Not only in China, but also the top AI specialists at American companies are also Chinese. Wonder there's a good pattern there. So, think like the way the Chinese think if you want to become an expert AI person. Anyway, Chinese researchers fed all LM research into a model and discovered 106 new AI model architectures that converge to lower loss with better benchmarks. So now you've got AIs creating not one, but 106 new AIs in a very short window of time. Okay, absolutely crazy. And ASI Arc is probably the coolest one as well. And this kind of like an Alpha Go moment for AI. And we've seen traditional media, which are normally way off base, but they do believe the AI revolution will cut at least a trillion dollars a year out of S&P 500 budgets. Again, Morgan Stanley says largely from agents and robots doing human jobs. Now, the LLMs are creating their own LLMs. Bonkers.
But I want you all to think in terms of dreams. Again, inspiration from Adam on this one. But think can always in terms of data. The group with the most data wins. Robotics are going to absolutely disrupt everything. Energy is needed for all of this to happen. AI is the thinking. Manufacturing, as Elon says, is probably the most important part. If you can't manufacture at scale things like robots and cars, you're toast. And of course, space for the global comms network. That is that is the world we live in. Dreams. D R E A M S. And if you have one player that has all of those, they're going to win embodied global AI.
Okay, this is a cool little analogy I I tweaked a little bit trying to tie in all the stuff we've covered today to the human body and senses. So Tesla has cars and robots. They have eyes and sensors that gather information and act in the real world. Without that ability to sense what's around you and see what's around you and react in real time to what's around you, you don't have embodied AI at all. And Starlink is like the blood vessels moving data around the world quickly and securely. If you talk to Jeff Lutz, he believes all Tesla vehicles will have Starlink in them, and they'll be able to communicate with each other and identify traffic problems, safety issues, entertainment, everything will be in the car. And the factory reproduces all of these components at scale. Remember here, Elon says the most important thing is the M, the manufacturing. If you don't have that, you're pretty much toast. But together, this is a global platform for physical AI, making Musk's companies a frontrunner to dominate the AI era. And remember as well, the synergy between the Musk industries. That's what cannot be overstated. They can all help each other do great things. No other company on Earth exists like this. That's why I am very interested. And shout out to Steve. Woohoo. Love it. Thank you, Smiley Stevie. They used to call me Smiley as a kid, by the way. Thanks, James and the team for AI content on a weekly basis. Great. Glad somebody liked it. And a big thank you to a mods. But I'm not done yet. There's more.
One final thing. Uh, IPOs have been super hot this year, but there's a new one called Vast Data that I didn't hear about till a few days ago, researching this video. But Vast Data is a data storage startup backed up by Nvidia. But the crazy thing is their biggest customers are XAI and Tesla. Okay, Elon Musk businesses, Musk Industries, also Coreweave, and don't know what Disney's doing in there, but anyway, they are growing at 300% revenue, and this could be a huge IPO in 2026. Remember, as I said before as well, data is the fuel. Top left corner here. Data is the fuel that runs all of AI. And this is the player that the key AI players use to manage their data storage. Crazy.
And finishing with meme dour AI do my work. Press the button. Ding. Work. Work. Done. But if you have AI do all your work, you might get bored. So don't let that happen, ladies and gentlemen. That'd be no fun. And I hope you all learned something today. That was a little bit of the video. From me, back is bad. So I'm, I'm sitting down. But thank you all for coming everybody, and I'll see you all tomorrow. Hope all is well. And again, comments, uh, back the idea of these videos is to help identify the next big opportunities through 2030, and this will dovetail very nicely into my new Magnificent 9, not Mag 7, my IA9. Thank you Tindy, Tesla, Jean Dunley, K8, and everybody else for being here. And a big thank you as well to Andy, LC, Blink, Steo, Aonda, Snickers, Toasted, Adam Q, Adam Q likes the AI content, Cashy, Cybertruck, Crypto, Mr. 100en, GQ Trader, and Bman. Appreciate you all. Let me know. AI is going to touch us all. Um, it's the most important narrative right now in my mind by far. Take care. Bye. [Music]