Terafab & Inevitable Tesla SpaceX Merger w/ Larry & Amy ⚡🎙️

Kind: captions Language: en What up, Hyperchange? Welcome to another episode. Today, we've got an epic interview. We're going to be steaming on SpaceX, Tesla, the future, space AI. So, the Terra Fab. So much to cover. I've got two very legendary guests here, Amy and Larry. Uh, welcome to Hyperchange. So excited to have you both. So, I think we should just kick things off with a crazy question directed to Amy, which is about this hundred trillion market cap and potential for SpaceX. And that's something that you've been posting about on X that I think a lot of people probably get thrown by because that's the size of the entire world's economy right now. So, how does SpaceX like what are you even talking about when you're saying that SpaceX could be a hundred trillion dollar company? >> Yeah. Well, that all comes from AI and robotics. So if you if you look at AI taking over a lot of the functions we have now historically any time in history there's been a massive increase in productivity the the size of the pie has exponentially just grown and so that's the part that people are not understanding that if you have robots that can produce things that are inexpensive and if you have AI that can create a whole lot more products then it's just going to get a lot bigger. And then the thing we're talking about today that it can the AI data centers in space that creates an entire new space economy which so that could be much bigger than the earth's economy. And when you start looking at numbers that big hundred trillion dollars is easily within the range of possible outcomes. >> And Larry, you have any comments there? >> No, I mean I'm aligned 100%. You know, just think about it this way. Right now, we're confined to an economy based upon the resources of the earth. Another decade or two, we'll be reliant upon the resources available from the moon and ultimately from Mars and potentially from Titan. So you know the limits today are very clear but the limits of tomorrow when we have an entire population of robots of robots an entire um you know couple of planets triple this the number of planets we have now or quadruple the number of planets we have now. That's the kind of size of the pie. It's pretty self-explanatory. And how do we get there? Because I saw one stat in your post, Amy, that was like 10 gigawatts is currently what Nvidia's chips, I guess, are using. It all boils down to this energy number. So, can you walk us through that that quick calculation that gets there? >> Yes, that's exactly true. So, right now, the compute that Nvidia has is equal to about 10 gawatt hours. If you look at from the energy standpoint, Elon Musk is talking about doing 100 gawatt hours in satellites in space just to start. That's just his starting point. And then from there there he wants to go to a terowatt um and from which is going to be I believe it was a 100,000 times Nvidia's production and that's possible from the earth. Then the limiting factor at that point is going to be rocket fuel. it wouldn't be enough fuel to launch more rockets from the from the earth. So then he will he'll be simultaneously building a moon base. How cool is that? And he'll be launch wants to launch an uh pedawatt off from of AI compute from the moon base. And then people are like well why would you need that much compute? Like what are you going to do with it all? But if you think about it, only 16% of the world right now is using AI and we're just barely scratching the surface of what it can do. So when billions of people start using AI and they start and then you have all the factories using AI and the robots using AI and the autonomous vehicles using AI, you need a source of power and you need a source of compute for that. So, it's going to get exponentially bigger and other companies don't have a plan yet of how they're going to be able to serve billions of people asking the AI model for information and that's where that's where the restriction is going to come in. What what do you think Larry? >> Yeah, I think that uh you know firstly it's not limited to what Elon can do. It's just that Elon is already planning for it and is, you know, focusing on it and he has a very clear path to this pabyte um uh um pedawatt of power >> and and so you know as usual he's out ahead and eventually everybody will get there will get an understanding but he's going gang busters. I mean, he's building, you know, a fab to build the quantum of chips that he's going to need to put a million satellites into space. Now, when you make plans of that caliber, it changes everything. I mean, to, you know, the size of the fab that he's building just beggars anything that's been done in the past. Everything is changed. So he's thinking on a on a different scale to what everybody else is thinking. Although, you know, Nvidia just announced that they expect a trillion dollars of revenue from their from their current chips or their their current chip plans. So, you know, he's not entirely alone, but he's definitely thinking, you know, way way ahead. And we're not talking about two or three years ahead. We're not even talking five years ahead. We're talking 10 and 20 years ahead. In fact, he's probably thinking beyond his lifetime at the moment. >> And that's actually perfect segue to the the Terraab, which we just got this news announced that the Terraab is starting construction like 7 days or 6 days or 5 days, I guess, for when this comes out. Like that seemed already when he announced at the shareholder meeting, I think took a lot of people by surprise, including me, which was like, whoa, Terra Fab, that seems so far out there. But it sounds like this is actually classic Elon moving it like he's going to set every record for what kind of chip fab is being built. And even more interesting is the types of chips seems to be a lot of these if you follow kind of what's happening with everything's going to be in space. He could build a chip that's optimized for space. And so that could give him a whole another layer of the advantage. Um which I think is a whole interesting angle. So Larry, were you you but just before we started recording were talking about something about they're already clearing land near Giga Texas from Joe's videos that you think is for the terra fab. Is that is that what you're thinking? >> Absolutely am. Uh they've cleared a gigantic piece of land. I think larger than the land for the current Gigafactory and I think that's going to be the terapab. There's an enormous amount of land available at Giga Texas. I mean, they're only, you know, they've only really exploited about less than a third of it. For all the factories they have there right now. So, I think what we're going to see is the terra fab on that site. It makes a whole lot of sense. I mean, it's it's only a few tens of miles from the Samsung factory where, you know, Elon is already walking the floor um learning about fabs or teaching Samsung about fabs. I'm not sure which is which. And so um you know I it's it's only a matter of time. I mean the the concentration of factories that Elon has in that very small you know triangle uh between Samsung in Taylor, Texas and the the um I guess it's the Gigafactory and then we have the Starling factory which is gigantic as well and looks like they're going to expand that even further make thousands tens of thousands of uh terminals a a day a day. So I mean just get a sense of the scale that they already have there. And what's interesting to me is, you know, they announced the 20 billion dollar caps capital expenditure for Tesla this year and everybody was kind of mystified. How, why, what, and you know, a year ago when everybody wanted them to buy back shares, you know, I said, you guys haven't seen anything yet that the capital expenditure is going to go through the roof. So I made a chart of it and and they're ahead of my chart. Even I was behind them. So they have a lot a lot of capital spend to go and it's going to change everything and that really is driving the necessity for a merger of the two major companies in his portfolio. >> Okay. Amy, do you have any thoughts on the Terraab? >> Oh, I'm incredibly excited about the Terraab. Uh when you look at the advantage that spa SpaceX and Tesla will get which will lead into the next topic of from vertical integration it is enormous. Everybody else has to go to do a request for proposal send out a list of requirements send out tolerances that things can be get off go back and forth and schedule all these meetings. Tesla and SpaceX can have all the engineers sit in the same room right next to the production line and they can optimize and they can iterate quickly and they can deploy products to market years sooner than others will be able to do. I mean the and for a less for lesser price and a higher quality product. It is just an enormous enormous advantage. And not to mention that everybody else if they're using suppliers, every supplier wants to get a profit on on the top on top before it moves on to the next price next part. So this is going to be a huge huge advantage that they can now do the chip design integrated with the chip manufacturing. >> Are we even questioning whether this is possible at this point? Like we're just just giving Elon the benefit of the doubt that he's gonna build a successful fab faster than anyone on Earth. >> You think that? >> Yeah, I do because I don't know. I mean, I just want to voice the doubts that, you know, people who are in the business of building chips understand that this is one of the most challenging engineering projects that you can imagine. And we all know how much how long it took uh for a dry cathode to be manu to to be successfully produced. You know, four years ago, um, four years ago, I was having arguments with some of the Tesla, Tesla influences about how quickly the drive back, the dry battery, dry electrode battery would be produced. And I was saying, it's going to take a couple of years. And they were saying, "No, no, they're producing it already. It's it's a done deal." I said, "It's going to take a Well, it took four years." So, you know, it's not going to be a straight and narrow path. It's going to be quite rocky. Look, Samsung several times have nearly gotten bankrupt making, you know, building fabs. So is Heinix. He today is manufacturing high high velocity or high speed uh memory. That high-speed memory they're making bankrupted them. I mean they were functionally well the fire bankrupted them but they were functionally out of business and they took this enormous risk and it took them several years. I think it's taken them almost three and a half years and now they're sitting on top of the world in the chip business. It's an incredibly tough business, incredibly risky, and the machinery it needs is incredibly expensive and incredibly hard to source. There's really only one manufacturer. And so to think that Elon may end up making the machines that he has to put into this fab and has to in order to build the fab he's taking on what is easily the toughest toughest job of his life. And I'm including you know Falcon 9 and I'm including Starship and I'm including the dry cathode battery and I'm incl I'm including model. I'm including everything. This is easily the hardest task of his life. So, we'll see. I mean, he knows something. I don't know. He knows a lot of things I don't know. So, I have to give him the benefit of the doubt. But, I will say it's going to be tough. >> And I tend to be slightly more optimistic on this because of the way that he said he was going to start. So, he believes that he can source the machines and and use the machines. and he's going to start he intends to start small and just follow the existing process. So unlike the dry uh battery cathodes which had never been done before successfully at large scale, this has been done successfully and he can just do what he just did with XAI where he went and got some talented people and he can bring them in so he has some of that institutional knowledge. Um, but I don't think that we'll see it scale necessarily as fast as you would like. But that's okay because Tesla has for on the Tesla side, they've said they have enough chips for the next two years already contracted and that and that's with Samsung ramping up that new Texas factory. So there's there's some there's some roadway here for them to be able to get this done. And then there the as Tesla is trying to work on the terra fab meanwhile their suppliers knowing that the demand is there will also be ramping up production for Tesla because historically what Tesla does is say give me everything that you can and then if there's a gap or a shortfall that's where Tesla will step in to make the difference. So a as Tesla's trying to work on this terra fab, the suppliers are going to be trying to lessen Tesla's need to have a terapab every single year by producing and meeting more of the more of the need that Tesla has. So I think that it will work out. Um, and I think that he will surprise us in it not being as bad as the as the worst pessimist are are forecasting because he can after all land starship on chopsticks. >> But, you know, I do agree. I I don't think he's g I don't think he's going to fail. But I will tell you even buying the machines from existing suppliers is not really as simple as it sounds because actually getting them to produce is really the significant challenge. I will tell you that Samsung failed in their Taylor operation and and Elon has been brought in to help them get it back up and running. it the the history of fabs failing is you know it's there's a long history of that but and I just wanted to I just wanted to point out the difficulty just the degree of difficulty of this mountain that he's going to climb and he's going to have to invent new chips or certainly new methods of cooling for the chips that he wants to put into satellites because we we are going to have to increase the cooling functionality of these satellites quite dramatically. And I know we we're moving on from the chip side to >> to the satellites to uh um inference in space, but it's each one of these steps are going to be in itself challenging, not to mention building factories on the moon. >> Yeah, >> I agree. Sure, go for it. >> Yeah, for as far as the cooling goes, um, a lot of the numbers that I've seen on that, people are just using the Nvidia GPUs because that's what they have out there right now and they're ignoring the fact that Elon Musk's plan is to actually use the hardware from the Tesla vehicles. So, th those do not require the same level of cooling that the Nvidia chips do. they also are proved to be quite ruggedized and furthermore um he's in intending to run them at a hotter temperature. So that's the modification that he's already come up with of of how he's going going to make this work. He's going to run get make the chips capable of running as as hot of a temperature as he can and then that increases the ability to do the radiative cooling in space. And then on the on each of these satellites, you can have double sideed cooling surfaces which also increases the ability to do it. So he's already put some thought into how he can get there. Now they have to actually start manufacturing it. We also know that StarCloud, they actually the conference that Nvidia is at today, Jensen Wong, walked out on stage to the StarCloud uh satellite launching into orbit that just happened this morning. uh has already designed a satellite that does have enough cooling in it and is running a Nvidia GPU right now. So, it's not an unsolvable problem from the standpoint of physics, but as Larry Larry's mentioning, it's just incredibly ambitious uh ambitious at a scale that other companies simply would not take on. They would not be launching robo taxis and doing Optimus and then in his other sister business be trying to send send AI compute to into orbit. Um but this is what Elon Musk does and in in retrospect it will seem obvious to people. >> Absolutely. >> It it's funny to hear the haters about that because I've been following StarCloud so closely and it's working. like they I think they launched five Nvidia, four of them are working and apparently the fifth one was already like not good before they launched it. So they don't think it was the space issue. But talking to StarCloud was interesting hearing how the Nvidia chip is so expensive based on the power it consumes, but something like a Bitcoin mining chip would be way cheaper and easier to power. So just like you got me thinking Amy with all the wiggle room where if he's designing it for scratch from this use case being vertically integrated there's like so many different layers of optimization for what he's doing and then the fact that Larry do you want to jump in? It looks like you have something to say. >> Yeah I think firstly you got to understand there two separate problem spaces. The one is if you want to do inference and the other is if you want to do training training clips which invidia is the perfect solution, one of the one of the better solutions. The heat problems are dramatically greater. The the the energy problems are even more difficult. I don't think Elon is planning on that. I think he's planning on inference because after all we are moving from a time of training to a time of inference in terms of the the eras of of uh AI in training chips. you've got a much simpler, you've got much lower heat problems and and the design, you know, if you if you read the numbers put out by Mark 33, which are really, I think, the most definitive numbers I've seen so far, they figure that there's a path to 100 to to be able to actually run those chips flat out in, you know, in a configuration that is not that different, by the way, very similar to the current uh generation or the the about to be launched generation V3 satellites that they're going to be launching on Starship and and Amy's right, there are cooling solutions for that. Even with the cooling solutions that we have today and running only training, we're still a little bit higher than we need to be. even running the the chips warmer than or much warmer than they are supposed to be. But there is a path a very clear path towards it and a lot of it turns on um you know the materials used and and some experimentation that still has to take place. So, I'm absolutely convinced that we'll launch those million satellites that Elon's planning for. And I'm absolutely convinced that he will have this gigantic, you know, spacebased uh data center um you know, in our in certainly in your lifetimes. I don't know about my lifetime, but certainly within the next five or so years. And and that's kind of phase two. Phase three is, you know, actually launching them from the moon, manufacturing them in part on the moon and launching them in very much larger numbers, very much larger numbers. um using power, you know, the unlimited power of the sun or relatively unlimited power of the sun and and making this millions million satellite u you know solution trivial compared to what he's going to do from the moon. So his plans are even greater than these almost unthinkable plans which he just hinted at at the shareholder meeting and blew our minds. Wow, this is crazy. So, I I I didn't want to well, I feel like I want to talk about the digital optimist thing because that's tying into this, but it's almost like such a lame tiny amount of chips compared to what we're talking about now. But there was an inkling of a much bigger idea, I feel like, when Elon was tweeting about this idea of digital optimists using the cars computers while they're not like working essentially. That seems like a huge huge deal. So, what went through your heads when you saw this news? Maybe Amy, do you have an idea? >> Yeah, I I think that this is brilliant and an example of what makes Elon Mus so different. So, he he's been aware for a very long time that energy was going to be a limiting factor for AI. And he looks out there, he's like, I have a whole supercharger network. So, he's going to leverage his supercharger network. He's going to leverage the chips that he spent every Saturday, I think it was part of every Tuesday, designing himself, which to go to what Larry was talking about, Tesla is right now using their inference chips in Cortex on a 1:2 ratio and using them for training as well as just inference. And Elon Musk has talked about the idea of putting not this version, just the AI4, but maybe once you get up to AI 6, putting several of them on a board and calling that Dojo 3. So he does see when he has the chance to optimize the chip for exactly what he needs and not have to put everything on it. He doesn't the way that Nvidia has to do a more generalized chip that he could make this something that could be useful for both training and for inference is is the game plan. And the start of that is he's going to figure out exactly what he does need and what the requirements are by putting a bank of AI4 chips next to these superchargers. And at night the superchargers are not used very much. So they'll be they'll be doing everything that you do on a keyboard. digital optimus should be able to do. And the other interesting thing is it's an architecture designed to start limiting calls to the LLM, which nobody else has that figured out yet either. Um, and so it's a it's the idea he was getting at with macro hard replacing the need for all this ERP software with by do having the AI just instinctively know what to do. And we don't know that many details about it, but what we do know is fascinating. >> Yeah. I mean uh I when he talked about putting the chips in the uh supercharger into the supercharger network, I gave some thought to the cooling problems because you can't put a big cooling tower, you know, in a parking lot or or in in some of the places where say an underground charging installation. So it there's a lot of engineering um issues and it's required a lot of optionality and ideas but it's a brilliant idea. I mean certainly the the car network is very use eminently usable and then uh you know to be able to add chips to the supercharger network and then I started thinking about well actually they're cooling the superchargers all the time. There's a water cooled I guess circuit in these superchargers. Maybe they use that and actually put the chips into the supercharger devices themselves. Just a fascinating engineering problem or I you know challenge but man I you know it it just gets better and better and better. It it bigger and bigger and bigger and better and better and better. It's amazing idea. And this macro hard concept seems very important potentially because he's going it's a joke with Microsoft which is like the world's largest company or you know four three four trillion. So and now he's officially announcing it's a joint Tesla SpaceX XAI project which to me cementss the merger almost but also got me super pumped as a Tesla investor of like at first it was kind of XAI but now it seems like >> and I can't even really wrap my head around what is this digital optimist project? What does macro hard even look like? Like how does it generate actual revenue? I've been trying to think of that and then at the same time watching what my friends are doing with Claude and their agents kind of work autonomously figuring my friend was trying to describe this thing of the way Claude interacts with a website different than we do by tapping into like the APIs and I I don't even know it's a little over my head but I've been trying to me it seems like where this is going is essentially it's going to be a super agents like that's the consumer feature >> that is part of it but I don't know if that makes sense. I'm kind of curious, how do we go? Cuz we have I was looking at Tesla. I'm like, okay, our revenue was down from last year. We're at 95 billion. Like when we have so much excitement and crazy stuff like when does this actually turn into like billions of business and how does that actually kind of manifest? >> Well, you want to go me? >> Sure. Um, I Yes. I think that describing it as an agent is the most intuitive way for people to be able to understand how this is going to work. BA basically you're going to it's going to be divided into two parts. One of the parts is going to be looking at five seconds of what's happening on the screen and then from that five seconds. The other part is going to be figuring out what to do. Um, so it's it's very similar to um a navigation system is the way that Elon Musk described it where it has a general route of where it wants to get, but then it's looking turn by turn. What's in front of you? Does it need to move? Does it need to adjust? Does it need to go to the can can it go on the way it or does it need to do a detour? That that's a good way to kind of think of how this would work. So basically anything that you can do on a keyboard, the the software will be able to figure out. It will have the general idea of what it wants to accomplish and the steps it should take to do it. But as it's doing those things, the same way that you would react and change what you're doing based on what you're seeing, it will be reacting and changing what it's doing based on what it's seeing. So over time this should be just can become powerful enough if you have many many agents that can do many specialized tasks you can replace out every form of ERP software that's out there. So you're not just going after Microsoft you're going after Oracle you're going after Salesforce you're going you're going after every single one of them. This is another way that you get to a hundred trillion dollar company because the AI will just eat software. It does. You won't need it anymore. >> Yeah. I mean, AI is already in a place where it could eat software, but companies are very low than, you know, diffident about actually putting it into action. Those days are going to be over soon enough. We're going to get into this next stage where we'll see truly agentic software being implemented. in, you know, if you own shares in SAP or Oracle, well, Oracle's trying to do a hard hardware play as well, but but if certainly SAP, I'm not sure they're going to be around. and these gigantic companies that, you know, they're they're they're toast. But but actually the digital twin for um the the humanoid robot, you know, when we think about that, just think about the car and what it's doing right now. And obviously the robot has many, many more degrees of freedom. So it has a larger challenge in terms of the number of degrees of freedom and all the things it can do if you wish. On the other hand the car the level of safety requirement for the car the criticality of every single decision it makes every second of every every subsecond is you know very very very high risk. whereas the robot has got a lot more freedom in terms of, you know, danger and and its the risks that it's taking. So there's a trade-off there in terms of the degrees of freedom and complexity for the robot in thinking through its next actions and what actions and its plan as a whole. But on the other hand, it not it's not as critical as the car. So there's a lot of commonality but those differences also will talk to the you know the road forward in terms of training. We know that it took four years to get to this point or even more to get to this point maybe 10 years to get to this point in being able to control the car completely autonomously. So we've built a huge foundation but the challenge is somebody worked out I think it was Arc worked out that the that there are 200,000 times more complex more degrees of freedom or more activity that has to be um divine for the robot as for the car. But on the other hand, I said to them, look, it may be 200,000 times, but it's probably a thousand times less critical. So really, the number is really in the 200 range, but even in the 200 range, it's going to take a lot of training, huge amount of training. And remember, it took us these four years or 10 years or some very long period of time when we started with a car, you know, jerkily trying to get around a corner to where it's driving smoothly and beautifully. That's the challenge. And the more compute, the more compute we can get, the more robots we can get out there, the more the more video we can get, the more we can train, the more we can train. And so the more chips that we have operating wherever they are in the cars right now when the cars are, you know, in the garage and then ultimately that we can get up into space and or get into these charging networks, we're going to need every one of them in order to get these robots to become, you know, really useful. I I think that um Lar Larry's example was really good because that that kind of feeds into my experience as an airline pilot. But what I would say is the and I was in a flight instructor too. The most difficult rating for people is their private pilot because everything is new. They don't even know what a half mile from the runway looks like at a thousand foot patterned altitude. So it can take them a very long time to get their private pilot. Now I fly a 787, the airplane is infinitely more complex. But to go fly go from a 767 to flying a 787 was not actually that difficult. Even though the airplane has duplicative hydraulic hydraulic systems, it it handles pneumatics in completely different way in pressurizing the aircraft with a giant air compressor compressor rather than taking bleed air off the engines. There's all these differences, but the training was quite smooth because I had a framework of how I was going to go about doing this. And I think that for Tesla is going to have a huge advantage in change in training the robots that other manufacturers don't have because of the journey that they went on with FSD. So they already have figured out how to do three 3D practic real world practical AI in a way that other people are just working on foundation models and this is a very different and difficult task as Lar as Larry says if you think of because it has to consider second order effects because when if you think of when you're in the car if you break you breaking is going to create create second order effects where the cars around you are going to react to your braking and now you're going to have to change the next thing that you do. That's exactly the case with the robotics. That's something that a foundation model really doesn't have to do. It can come up with its answer. You might ask it a question. It might change its answer, but it isn't considering what are you going to do? And that's changing and narrowing down its options. And Tesla having experience with this, I think is going to be a huge advantage. They also know what really worked for them in trading FSD. and we started seeing that the the updates coming with these dramatic changes from update to update to update where it's just getting so much smoother and so much better and they know what those tricks were that they that they'd used to do that. So I think that it's going to surprise people. There's also an opportunity with robotics to deploy it first in more limited manner. you know, first get it good enough just for a production line, then get it good enough for help with a laundry and and to to narrow down the task that you're doing with it. Uh, and then and then move onward from there. >> Yeah, I I I agree. I think that um you know having the pattern of developing FSD and applying that pattern to the robot is going to give them a huge leg up as you say you know it's like it's like flying. Of course I never quite made it. I I mean my experience flying Amy I just got to take my hat off to you. Amazing. >> So okay. Yeah, great analogy, Amy. That was that was interesting. Um, I'm curious w with uh thinking through like hearing both of you talk got me thinking about how much compute we're going to need to pull this off and then backing into the financial calculations because I we have all the ingredients here for Elon to do this. We got the rockets. We got he's going to make the chips. He's going to figure he's got the humanoids, but the one missing ingredient right now is cash, it seems like. So, and that's Larry. You've been a way ahead of the curve saying Elon is going to have to do a massive capital raise and there's huge amounts of cash commitments. So, I'm kind of curious. There's so much up in the air right now with how the SpaceX IPO is going to happen, if this merger is going to go through before then. Uh, but it seems like the north star here is both of Elon's companies need huge amounts of cash ASAP basically to invest in these projects. So, what what's your take on how this is going to play out? Are we going to have SpaceX IPO raising 50 billion or do you think there's a merger that's going to take place before so they can start to raise more into the hundreds of billions of range? Because it seems like this is going to be a big part of the story is is raising these hundreds of billions to make it happen. >> Yeah, Elon is not going to raise hundreds of billions. That's not Elon's style. So, SpaceX is going to go out. There's nothing that's going to stop this public issue for SpaceX. The it it's on the it's on the track. It's on the way down the track. I doubt he'll stop it at this point. I think it'll go on. He's going to raise about 50 billion dollars. Um it'll happen on June 26th maybe, but around that time uh that's his birthday and that's kind of the target date. Um, and the interesting thing is not going to be much of a lockup. In fact, I don't think there's going to be a lockup at all. And the reason for that is that they're going to be in the indexes within 3 to six weeks. That is to say, they're going to be indexed into the either the S&P or the NASDAQ or both. And if they are indexed in them, then they need enormous liquidity of stock. there's going to be gigantic demand for that stock and so 50 billion won't even touch sides. So they're they're going to go public at the end of June. I think this the the stock price is going to go up by a very significant amount. I mean I'm trying to be conservative and say it's going to be like 50%. Give us the bullcase too, Larry, because that demand thing you mentioned is crazy with the indexes changing all their rules, having to buy space. That could be like a hundred bill of of demand on the stock right away from >> 100 bill of demand is nothing. No, it's going to be a lot more than a 100 bill. A lot more. The demand for stock is in my judgment is going to be in the 3 to 400 billion range. So I think that overnight the stock is going to go from its current one and a quarter trillion to around two two and a half trillion in that range. And I'm not the only one to say that. You know I have colleagues I have friends good friends in Morgan Stanley they're feeding me these numbers as well. I'm not guessing in a vacuum here. Um that that's my guess. So, what we're going to see is a sympathetic rise of the Tesla stock. Nobody's realized this, but there's going to be an index relationship, an indexed relationship between the Tesla stock and the SpaceX stock. And what is the index relationship? the current market cap of Tesla, which is in the trillion dollar range, and the current market cap of SpaceX, which is in the trillion a quarter. If SpaceX goes public in the $2 trillion range, we're going to see Tesla rise to the $ 1.5 trillion range. Of that, I'm certain. Why? Because everybody realizes by now that very shortly after the after the IPO within months at most we will see a merger of those two companies at the current price at the current market price of of both co companies. That's my sense of it. >> Amy, go for it. I I want to dive into the how the merger happens, but let's focus on the IPO first. >> Yeah, I I think that there could be a se second two groups of investment bankers. One that's pitching an IPO strategy and one that's putting pitching a merger strategy and they're competing with each other. That that is not >> that is not that that's not unheard of in this type in this type of financial situation. Then it just kind of comes down to which which group of investment bankers presents a more compelling argument from the standpoint of the employees. IPOs are a good thing, but they're a distraction. Mergers are in this case would be a very good thing. Also a distraction. It's expensive to pay investment bankers twice. So, there's a slight possibility that there could be a that they could go public as a merger um in that standpoint, but I really don't know because that that's of course I'm not getting information from Morgan Stanley or anyone else. I'm I'm just looking at what makes sense from a business standpoint. And a business standpoint in whatever way the transaction goes down, I just don't see any way that we don't have these two companies merge now. that Tesla and SpaceX have to be merged because we have too many really significant big huge joint bets that are going on. Digital Optimus just being the newest one, the Terra Fab with Tesla producing the trips chips for SpaceX. And as we were talking about earlier, there is a huge advantage to being your own customer. When you're on your own customer, you can optimize your design and you can optimize the production line for your design and you can have all those engineers sit in the same room. And the last thing we want them doing is keeping time sheets trying to allocate them what time was spent on Tesla and what time time was spent on SpaceX. That's just going to be an unnecessary thing that slows them down. And I think that a lot of people when they think about a merger, they just assume that it's a static opportunity that we're just dividing the pie into more pieces. But once again, that they would not merge to keep Tesla on track to be an eight trillion dollar company. They would not merge to keep SpaceX on track to be an eight trillion dollar company. they would merge because there's an opportunity to become a 100 trillion dollar company that is just simply way too large to pass up that changed everything and because of that now the missions have converged on the same mission of like an AI spa space and robotics company and they each have very necessary pieces of the puzzle. No one company can do this on their own. It's going to require all of them. It's just too big to think about sitting on the sidelines and letting someone else do it instead. >> What she said >> and the it seems like in Elon's head it's already merged. I'm not sure if you feel that way, but just the way he's been talking about it and framing and moving everything so quickly to me it's and he's we always know Elon's optimistic for him. Oh, we're going to merge at some point. Obviously, we can get it done. you know, he's already thinking 10 steps ahead of that. So, I feel like, yeah, to me, the writing's on the wall. Like, when I saw this digital optimist thing and the terra fab without a mention of whether it's SpaceX or Tesla, I'm like, is that the announcement of the merger? When they say the terra fab is the, you know, I think we're like I don't know. So, I'm I'm curious what Larry's going to say about how the merger will occur if the normal um IPO happens. I don't know. Maybe that's even just too boring of a subject because it's easy. But I'm I am genuinely curious because I think there there's never been a merger this big. >> Um yeah, that is true. But I do think that the idea that they would merge as part of the IPO is probably not even practical at the moment simply because the complexities of that are very great particularly as they have to raise money at the same time. So, we're talking about levels of complexity that are very difficult. One of the reasons that they'll want to actually end up as SpaceX, not as Tesla, is because Elon already has um controlling shareholder controlling shares uh in in SpaceX. He's got these special shares. You can't create those shares if you're already, you know, a company that exists that you can't change your your share um type to give somebody control once you already exist. So, it's not possible to do it in Tesla. You can only do it in SpaceX. Now, could you merge them and then go public? Yeah, but it's a little late for that because if you start a merger with Tesla now, you'll need a vote of shareholders by Tesla and that's going to take about three months to get an F1 out, get a special general meeting. And so that's going to push you way past June. Way past June. It would take several months to put the documentation together. Then you've got 90 days to get So I I think that's impractical. So honestly, the the the fastest route forward to a getting the $50 billion in the bank, getting SpaceX listed is to to go with the S1 that they've already prepared, get the June date done. We're only three, four months out now. You know, we wouldn't have time to get started on another plan. And then once you've done that and you've placed the shares and the market has settled on a price and we'll see the Tesla price move accordingly, then the merger can be done incredibly quickly, incredibly quickly. the the the long pole in the tent will remain the um the the shareholder approval of the Tesla investors, but there won't be any long pole relating to the S1 for the new company. There won't none of that will happen. There'll be a very straightforward offer and and they can start operating, you know, very shortly after that as if they're one company. >> Wow. So, it would make sense for SpaceX to buy Tesla. Tesla shareholders have to vote to approve that. >> Yeah. I I I did an analysis and I did some numbers that shows how the shares would be issued by SpaceX to the Tesla shareholders, what the total would look like at the end, what what Elon's control would look like. He'd have absolute control of the final product. The only open issue is what we do with these options. That's a very big issue, by the way. It's not easily overcome and there are some route there there are some ways forward but bottom line is it would be much simpler to do and it could be done by year end but functionally the co the the companies can start operating as if it's happened. >> Wow. This is like blowing my mind because it's on one hand with such an ambitious project. It's almost like capitalism is working here. Like what other I can't even think is China going to deploy hundreds of billions of dollars for any initiative. Like the amount of capital that is being deployed ahead of time for such an like ambitious vision I feel like is something like we've never seen before honestly like like what other country has a plan to double the output of the world economy within 10 years? Like has that even ever happened in history? Probably not. So, I just feel like it's kind of crazy. Like Larry's like, "Oh, hundred billion's nothing. There's gonna be 400 billion flow into this." It's like that's insane to think about how much in some ways the marketing that Elon has done to secure these resources to be able to have a like it's it's almost like if he wasn't Elon Musk and you heard another guy say this, it's like, dude, this is just not going to happen. But I I don't know. So, it's I feel like it's sometimes I have to pinch myself for how crazy this is and how fast it's moving. I don't know if you have any ideas on that or like you know taking a step back to think about where this sits in history at all. >> I think it's absolutely enormous. It's going to create a whole the goal here is to create a whole whole new space economy. So we were before limited by the confines of the earth and earth. It's not just going to start out with it's going to start with orbital pro satellites then it's going to go to a moon base then you can start going beyond the moon. Of course Mars is a plan and you can even go interstellar. So like the it's it's really an ambitious plan and it comes off of being able to harness the energy of the sun. We're getting rid of fundamentals constraints that Earth had had that was going to limit the max that you could have grow the output on on Earth. So not at no time in history has there been anything similar except you've had things that have been transformative to the way people lived. Certainly like when the caveman discovered fire and you know like there's there's times where there's been also huge leaps in the in the ability of people to to live and and have safety and other things like that. But at no point has it will there have been such an enormous explosive growth in GDP as as we're about to see. >> So I just want to give you some context here. Heinix. It was like the number four manufacturer of chips and it was close to bankrupt. They're the manufacturer of these very fast memory chips, high-speed memory chips. They're building a very large fab complex in China in in Korea, South Korea right now. What do you think they're investing in their fab in this fab complex? >> Billions. >> Not billions, not tens of billions. $400 billion. >> Wow. >> So, you have to take the perspective that you have here is that we're rebuilding the Earth's economy. We're rebuilding everything. You'd have to go back to the end, you know, the 1860s. You'd have to go, you'd have to combine the railroad development, electrical generation, the telephone, put them all together, and you still won't get to where we are going right now. We're changing everything about human civilization. And so hundreds of billions of dollars, hundreds, maybe a trillion dollars is the kind of capital that you're looking at when you're thinking of taking, you know, generation of power of energy to, you know, to space and then to interstellar space. It's it's a gigantic undertaking. So, you know, Tesla's $40 billion and and you know, that's what's amazing about Elon Musk. Nobody understands just how capital efficient he is. There is nothing like him in the world. You know, these companies that invest these hundreds of billions of dollars. I mean, Elon's doing it with like pennies on the dollar. And so it's going to be interesting to see how he does it. I mean, I I hope I live to see these satellites in space because I think that everybody's projection is going to end up being wrong, you know, and and Amy's going to be right actually. It's going to be less investment than we think, but it's going to be massive. >> And Elon's got a secret friend helping him on all this, which is to me the biggest X factor. and super meta which is Grock like does like Grock is turbocharging engineering is turbochar he was probably up till 4 a.m. with Grock like what how do we get more of this the lunar driver on the moon it's like I don't know some of them think it's like a funny joke but I'm like actually in my life Grock has become such a turbo for what is possible I feel like that >> is almost like the special like stardust in Elon's lap that is going to make this all possible in a way >> 100% there's no there's no question about it that it that if you look at Starship is going to be unusual and it was will have been something designed prim primarily by man and everything that comes after this will be designed primarily as a partnership between man and AI and very soon after that it will be primarily designed by AI with man just thinking it that we're helping a little bit you know so it's it is really it's really going to be a huge a huge change but that with using Grock you should be able to find the inefficiency inefficiencies a lot quicker and flag where those are and you know Tesla's manufacturing process if you look at just what they do with the cars the very first thing they do is hook up the computer in the car and then as as the parts go on they verify that that's the right part so that there's they have this ability to of quality control that is baked into the system and when you when you add an AI into that loop now you're going to be able to not have to go produce a whole batch chips to discover that the power wasn't there or that there was a problem with the first part. Hopefully, AI is going to be able to flag that for you immediately. You're going to iterate it. You're going to fix it. Maybe you'll take a few days off on the line where you're not producing anything, but you're not going to have that enormous level of waste to get to the end and then discover it didn't work. And so that's the one of the reasons why I think that the capital expenditure is going to be less and the time is going to be less than people think it was. I don't think it's going to be easy. I think it's going to be very difficult. But Elon Musk will always make something that's difficult happen if he thinks it's necessary to get it done. >> And he will be late. >> Well, he I don't think he's even given a specific time frame on this one. He's just given what the order of steps that he wants to take without without saying which time is it going to be done >> which is almost >> whenever he gives us a time scale he will be late. >> Yeah. Absolutely. >> Yeah. I almost like that there's no crazy time scale on it because it feels like I don't know that's gotten into trouble before. But you also got to motivate the people. I don't know. So >> yeah. >> Yeah. I I I like the way he puts those time scales down because every you know everybody und