Tesla Is Quietly Going All-In On Optimus

Kind: captions Language: en Right now, almost all the AI money is chasing digital intelligence, chatbots, the coding agents, the data centers. But Tesla AI executive Young T. Sahi just said he thinks the focus is about to move saying when the digital world collapses to a single prompt, the next frontier is real-world robotics. Tesla's own job board says the company agrees. Tesla's hiring for 280 Optimus roles right now. That's against about just 30 for CyberTruck, roughly nine times more. Look at what those roles are. Thermal engineers in Palo Alto working on vapor chambers and whisper-quiet cooling, the unglamorous stuff you only staff when you're serious about shipping. So here's the setup if you own Tesla stock. The market is paying up for digital AI today and the physical AI company sitting right there hiring like the robot is the main event. So when will investors make that turn? And when they do, you want to be there before them. Let's let's ask that question to Joe Bakhti. He's an economist, investor, and tech entrepreneur. He has a YouTube channel covering both Tesla and AGI. Check out his website and community at pioneerlands.org. Welcome, Joe. >> Good to be back. >> All right. So you've been following AI and you've been investing in AI and there's layers to it. Of course, there's OpenAI and Anthropic, Cursor, um you know, they these guys were like the hottest hottest thing because they are the actual thing that consumers touch. And there's the layers behind it, then there's the layers behind that. You know, memory stocks like Micron, um and then of course, um data centers, SpaceX AI, others. So when do you think though that people will start realizing, okay, it's it's hot right now physical or digital AI, but at some point physical AI. Is it like many years down the road? Is it sometime soon? What's your gut feel about that? >> Well, I think clearly non-physical AI, we're in the middle of the singularity. This is like very serious business. When you look at the Chinese models, I think China has now taken the lead, in my opinion. This is like very People don't fully appreciate how extreme this is. Kimmy K3 and now Deep Seek Flash V4. Uh Deep Seek probably coming out of the real model very soon. We are now at the point where Anthropic and OpenAI are really in real trouble. OpenAI just reduced prices by 80% and so on and so forth. And I can feel it. They're we're very close to this inflection point where things get very serious, also in terms of what these things can do. I think by December, it's we're going to live in a different world again. >> [snorts] >> And so that's all very serious. When it comes to physical AI, we have two big topics. We have Optimus, uh like we have humanoid robots and robot taxis, I would say, as especially as Tesla shareholders. Well, robot taxis are a whole long story, but you know all about it. I think we are also within the next 12 months this is going to happen at scale. From a Tesla perspective, and that's physical like at at real scale. Um I think it's going to happen this year, the inflection point, and then we will have and I said this for a long time. We were a little early, maybe, but I still think this year we see thousands of these things. And then we see hundreds of thousands and millions very quickly. So, that's a little piece of physical AI. Humanoid robots are the biggest piece, of course, and I just listened to Chris Camillo a little bit He elaborated on Figure 1, on Apptronik, and he says his mind is blown what's happening. So, he clearly is implying there is a massive acceleration on the humanoid robot front. I think the the fact that we don't hear from anyone from Tesla or Apptronik or Figure 1, like they don't show too much, is also a sign that there's exponential progress. Cuz that's exactly when you don't show anything anymore. Cuz you want to make sure the others can't copy it. And so, I do think probably next year, probably not this year. >> Yeah. >> We will see some massive breakthroughs. We will see commercial robots. >> Yeah. I I I think I think it's safe to say next year. The reason is that physical AI, which is robots, need physical products, and that takes longer to manufacture at scale. And so, even if you do are able to show demos, it will get excited people excited, and they'll invest. And again, a small group of people will invest even earlier on like their startups, and then a bigger group will start to invest once they see that it's happening. We're not yet in mass scale production, and that's why you know, Tesla is so far ahead there, but you don't see it. You don't see the investment there. Young to say, senior AI executive at Tesla, he said, "When the digital world collapses to a single prompt, the next frontier is real world robotics." He's not talking about investing, but he's talking about that digital AI. And it's moving so fast that at some point, you don't need even need to like keep asking it what to do. You're just a single prompt like, "Manage my life." And then boom. >> it's I think everyone's a little in over their like, you know, I think people underestimate the complexity of the non-physical world. They think, "Oh, it's just, you know, in order to really take over a law firm or an investment bank or a media production." So, you you know all about it in media production. Before you can lay everyone off and just say, "Oh, just cut all these videos, make it all amazing, and publish all the shorts." And it actually does it, uh I think there's going to be still a little time out because this is actually very complicated. So, but is it happening? Yes. And I think, you know, because people sometimes downplay how complicated the virtual world is and they up play in my opinion how complicated the physical world is because yes, does it take always a year longer than you think or two years? Yes, but I always said that the humanoid robots are not as complicated as people make them out to be. They're like when Arc for example says they're like whatever they said a million times more complex than robot taxi. I don't buy it. I really don't think that's true. I think they might be a little more complicated but robot taxi also has enormous additional problems because it needs to be perfectly safe. Long story short, I'm pretty sure next year we see humanoid robots at scale. Not perfect but not in the way that they can replace your plumber or something but at scale at millions. >> Well, I mean the question is Joe, do you need to see it at scale before investors get excited and want to invest? Right now, a lot of investors including myself have invested in startups you know, a couple of years ago but today people are getting excited about humanoid robots but it's not yet the frenzy that we're seeing in AI. The only reason we saw the frenzy in AI is because of real earnings. Nvidia showed quarter upon quarter. Then everybody kept questioning whether it's real or not then you start to seeing you know, the mag seven investing trillions of dollars every single quarter to keep upping it. Then you see big companies adopting it. So, that won't happen for physical AI for several years though or do you think that there could be a time where people realize >> big problem or a big difference between hype and stocks running up for a trillion dollar company or a 10 billion dollar company or five or one billion. If you have a one billion humanoid company, can you go up 30x in one >> Mhm. >> weekend? Yes. >> Yes. >> If you had one billion and you demo your little thing, can you be 30 billion on Monday? Of course. If you're 1 trillion, can you be 30 trillion on Monday? No. Because you clearly see for Tesla, especially with SpaceX, but the mag seven for Apple, for Micron, the market is absolutely brutal. Once you hit this 1 trillion, they're getting brutal. They say like, "Show me massive amounts of cash. That's not even enough. Show me growth rate. That's not even enough. Deliver massive evidence that this growth rate will keep up for multiple years." And if you can't prove even at a 300% growth rate, if you can't prove that in 2029, you will still have a high double-digit growth rate, we're selling your stock. That's what they did to Micron. That's what they did to Nvidia. That's what they did to Meta. At a 17 price earnings, they dropped Meta 10%. Now it's back up, but so it's absolutely ruthless. So if Tesla comes out, "Here's my great humanoid robot." That's not going to help at all. They need to show >> Yeah. I agree. >> You know, and I think that's kind of an opportunity for us, because it means we're getting these things kind of cheap. Because the reason these little companies go up 30x, there is a good reason. Because people think it's a billion, but if this works, it's a trillion-dollar market or two Oh, it's cheap. What people don't understand is we are now entering an age where this is also true for trillion-dollar companies, as we all know. We know Micron can make 2 trillion revenue, no problem. We know Tesla can make 20 trillion revenue, but the market thinks it's just utterly impossible. So we have an alpha with all these stocks, right? Every stock where you think this is actually different, they're at the frontier of history. There is no ceiling at a trillion revenue, or at whatever it is, 500 billion right now. The ceiling is completely gone, and you can go up 100x from 500 billion revenue. If you believe that, and you're right, these are basically startups, and they should be treated as such. >> Yeah. >> And at some point if we are right in 5 years the market will suddenly understand, oh my god, wait a moment. And then you see a repricing and that can also happen before. You don't need a 50X repricing, but if you get a 10X repricing, it's good enough. >> Well. Yeah. >> So this is kind of the scenario. I think the the reason why people just give Tesla and the Mac 7 everyone such a hard time is because they say, "Dudes, you had a trillion dollars or 4 trillion Nvidia. That's it for you. You are not getting because it's impossible. In order to be a $10 trillion company, you wouldn't make a trillion revenue and that's not possible. So once the market is like, wait, they just made a trillion. Now they made 2 trillion. And then they understand, well, if you just add up the numbers, this is like crazy. There will be a time where people reprice this stuff. >> Yeah. >> But it's not now. So that means >> Yeah. >> It's hard for Tesla and also for the >> But there will be a time >> It's like the business of earnings, right? When people say, "Oh, see, they're going to announce they're going to have maybe 8 gigawatts of computer. First of all, they won't have 8 gigawatts of computer. And second, even if they announce they think they have 2 gigawatts. This will not even move the stock because everyone's going to say, well, then do it. Come back when you did it, right? We are not interested what you think you're going to do. You you had a 1.5 trillion. If you want me to put you to 2.5 trillion, show me the money. Don't tell me that it's coming. >> Yeah. >> And it's not nothing against Elon or his companies. There's also some delusional people who are too much of fanboys. They go, "They hate Elon." No, they hate Jeff. They hate Jensen. They hate you know, everyone. They hate Altman. They hate them all. They say, "Dudes, you are too valuable. Show me trillions in profit before I do anything, all of you." It is what it is right now. So. >> Yeah. A very good point. I do like that you positioned some of these tech companies, including Tesla, trillion-dollar companies, but they're in some ways you should think of them as startups because what they're dealing with is AI and physical AI, which is unlike anything we've ever seen before in terms of market potential. This isn't >> by the way, the discussion we have right now is a very good discussion because what we discuss with humanoid and robo-taxi for Tesla or with computer and SpaceX, other people like Gavin Baker, I just saw him, they are discussing it for products and for Mac 7 and for Micron and they say the market is just not understanding what's happening. >> understanding. >> They they clearly price and Micron is the best example. They price Micron as if this company will cease to exist in 2029. Cuz if it doesn't cease to exist, it's completely mispriced. They're saying, "Sure, you're trading at 900 billion, you're getting 600 billion in earnings by 2028, but we still sell the stock because it's cyclical." So, okay, what does that? You think in 2029 they make proof and disappear because no one needs a RAM anymore in 2029? That's the craziest idea. So, they're discussing all of these stocks are being discussed right now and the people who are smart basically say this time the market really gets it wrong. They think it can't be. >> So, that was part of my argument is that you you saw the huge Nvidia chips and then you've got LLMs. That's where the focus has been, but now it's moving to data centers, Micron, hardware. Then you're going to get into physical AI. That's when people realize, "Okay, the next step after this is" right? Not only data centers, but now it's going to be robots, cars on wheels, and robots on wheels, and robots. >> I think there's one thinking mistake that people make. They always think it's S curves and now Micron and Nvidia and AMD they are done because they had their run. And now we're moving on to the next. I think in the age of AGI it's a little different because they're not done. Because how does physically AI work? It still runs on Nvidia chips. It will run >> need all of them. Yeah, they need all of them. >> They need They need It's like all these stacked S curves and the core things will just get dragged up because all the robots need it and the robots themselves are also an input general purpose technology. So, everyone will need the robots forever. So, it's not an S curve. It's just an exponential and that's all new for people because when was the last time we had just total exponentials? Maybe oil or something like energy, but it's rare. >> Yeah. All right. So, some more news here. Uh Yoda who follows hiring for Tesla very closely, he said, "We're underestimating the Optimus program." He has found that Tesla's hiring 280 Optimus related roles versus just 30 different kind of roles for cyber cab. That's nine times more. We're in for a surprise. Um Joe Techmeyer took a look at the Optimus factory at Giga Texas. Uh looks like and where it's located, how big it is compared to the rest of the site. Here's a short video clip. And um basically it is popped up very quickly and it's uh happening really really quickly here. So, that's the building and it's already up. And this is going to be the 10 million per year robot line. So, when you take a look at that, you realize, "Okay, first of all, is there any other company in the world that's building a factory that's big for human robots?" So, yes, it's not going to move the stock for anytime soon, but this is the scale that they're building at. It's huge. Okay? And then they're also hiring for AI and robotics teams. Um it's just a packing industrial grade actuator motors, power electronics, bipedal walking, the motors, cooling fans. So, this is just showing you that they're not only building the robot, they're building all the parts for the robot. And that's why it's taken a year, 2 years longer than everybody expected. I think Elon thought that he would be able to get by a bunch of this off the shelf. Now, he's been saying you can't buy this off the shelf, and then that's sort of why his numbers have fallen back. Now, they have to spend a year building these things themselves. >> Yeah, I think if you were like a tactical investor or someone institutional investor who can't just park the money in Tesla and wait, uh if you were Chris Camillo, uh the right approach with Tesla is to understand this is what I'm getting to more and more. You look at stuff and you don't even waste that much time if you're like smart, you look at all the stuff and say, "Okay, fine. They're going to kill it." We don't even have to discuss it. Now, next question is when? And then next question is when will the market recognize it? And I think not now. That's the short story. It doesn't mean the stock doesn't go up because I think it should go back to 360 very quickly, but then yeah, we can be stagnant depending on the market because robotaxis is not happening in the next 2 months or 3 months. And Optimus is definitely not happening in the next 3 months. But at the same time, we see this preparation. So, we should basically if you're a more, you know, momentum-driven investor or someone who's more opportunistic and strategic at the same time, you understand, "Okay, Tesla is going to get his its repricing." Probably soon in the next 12 months, maybe 18 months. Let's see. There are two catalysts, robotaxi and Optimus. Uh Optimus, this factory, I mean, probably 2027 or 2028, I think is guaranteed to be one of these years where there is an inflection point for humanoid robots where Tesla will shine very big, where people suddenly realize, "Oh my god, robots are real." number one. And second, oh my god, Tesla's the only company who has prepared the scale. But, will it happen in the next 6 months? No. The next 12 months? Probably not. The next 24 months? Very likely. So, that's my timeline for the robots. On the robot taxi side, I think it has to happen definitely in the next 12 months, probably in the next 6 months. So, that's much more relevant. And then you can like with this kind of market, you can basically lean back and watch it closely every day. It's like, well, when do I see the signs? And then you front run it. That's kind of the plan. If you're a long-term investor, you just buy and hold it. But, if you want to play it a little harder, I think that's a good thing. And then you have multiple stocks, your favorite stocks. You do that for five or three or four, and really understand them and say, okay, they're strategically mispriced. They'll be repriced in a reasonable amount of time. What are the catalysts when that happens? Plus macro. And then you can basically put them on your screen and watch your four favorites and say, okay, which one gets hot? And then you win, in my opinion. Including SpaceX, by the way. That's one of the hottest, but that's a different story. >> When you saw Elon building SpaceX AI's data centers, XAI's data centers, and then at first everybody assumed it was for themselves only. Then at one point they actually start renting it out and start making real bold multi-billion dollar sales. Well, at what point do you think the investors will realize that Tesla's actually and SpaceX building chips, terrafab, building their own memory, terrafab, building their own energy packs. You know, these things are the building blocks, but then at some point maybe they realize, boy, these guys are going to sell them, too. They're not just going to, you know, they're going to build everything they need for themselves, but just like, you know, the data centers, they they might have excess capacity and sell it. And then all of a sudden they've got an Nvidia size company within Tesla. They've got an a Micron size company within Tesla. >> Yeah, I think what's going to happen is that Terafab So, first of all, because of what we just discussed, the size, people don't like fantasies for this. They want to see facts. So, it's about okay, we need and indisputable facts in terms of billions or hundreds of billions in cash coming in in a way that is understandable and doesn't require any imagination. So, for example, memory chips or something. Uh and then they need to see there's a clear timeline that is indisputable. This is okay, this is clearly happening and a buyer of that stuff. And then you're talking then you're talking catalyst. And I think for Terafab, for example, sure they can sell into the market, but in that case I would say that's not even a thing because they will sell it all to SpaceX unless they are SpaceX at that point. Because SpaceX will be able to It's like robotaxis. You're not going to sell cybercaps. Like at the moment where you own the vertical stack and make more money up here, you're not going to sell anything. You're just going to make all the money. And my suspicion is that Terafab will also not be selling anything into the market because they will be plugged into space AI or N robots and will just get way more margin by not selling it and by utilizing it into these and then selling space AI, for example. Which is even better. So, then yeah, I mean, if you think about SpaceX and that assuming they buy Tesla and then Terafab is SpaceX, which is like nearly guaranteed to happen, uh I don't think they will make revenue with this stuff directly, but it will be this insane build-out where they know okay, they killed it, they make insane amounts of money on each space orbital satellite. Now they have Terafab up and they can manufacture like a billion more satellites. And they just go straight into the bottom >> they're going to just use it for themselves, then yeah, we're going to have to wait even more years before actually approve earnings. >> Well, but I mean I think I don't know. Like I'm pretty sure Optimus will be here for scale at scale in 2029, 2030, which is Terafab. Like Terafab is going to be done maybe in 2028, but more 2029 in my opinion, like really at scale. And people underestimate that. I said it since the first time like last year I heard about it, space satellites are underrated. They will be here at scale absolutely latest in 2029, like gigawatts. So, I don't think this is going to be too long cuz I did just an analysis today or before the earnings call of SpaceX on the terrestrial buildout and demand for compute for SpaceX. There is no limit to the demand, as we all know. And the buildout is a huge problem even for Elon because of energy. They get just smashed back left and right. They were they are forced to actually remove half of the entire energy now from Colossus 2 and 3. And they agreed to it because it was the government. They have to replace it now with like grid. This is all on track, but they have to build out 1.2 gigawatts and most of that is just replacing what they have. So, the idea that you can just scale this is wrong. You do are not allowed to. And that creates insane pressure on SpaceX demand pressure on satellites. And to supply these satellites is not that hard. It's not that different from a Starlink satellite. People got it wrong. They think it's impossible to >> you're you're thinking the pressure terrestrial data centers is having such a huge roadblocks that they're going to wait for orbital. But how long will it take for orbital, right? It's still going to be years >> I don't I'm not >> You're saying it might happen sooner? You're saying it's a lot easier to build these data centers than people think. >> the case for orbital AI together, okay? People sometimes don't understand how it works. They think, "Ooh, it's a new thing. No one has ever done it." That's first of all not true. Even Nvidia had like GPUs in space. That was tested. We know how they behave in space. We know the little problems. Okay. Uh Starlink satellites, they are being produced by the tens of thousands by SpaceX. They know how to create satellites. They know how to create the solar They know everything. They have even radiators for heat. They have everything. And so, all you do is reconfigure this thing. More solar panels, many more radiators, the rest is the same. Instead of like Starlink, you put in a little cluster of Rubens. One rack. We already know exactly what it is. One rack. Per satellite, you link them with lasers, which they already do. This is exactly a Starlink satellite. So, it's a Starlink satellite with more solar panels, more radiators. In the middle, you put in one Ruben rack, Vera Ruben rack. And you're done. Okay. So, it's technically completely possible. Now, all the other people are like, "Ooh, what about radiation?" This is all solved. We understand how it works. And then, the next question is, you know, do you have smart people doing this? Yeah, you have a lot of smart people. And then the next thing is, "Okay, but why would I do all of that?" Well, because you get a hundred trillion dollars as a reward. So, if you put these things together, totally doable. Everything is known. You just need a bunch of smart engineers, and the prize is a hundred trillion dollars. What do you think humans are going to do? They're going to build it. >> It's it's time, though. Years of a flight cadence is number one, the biggest roadblock at this point. >> what do you mean? What is the time? Just to get them just mass-wise? >> Yeah, flight cadence. How many flights can you do to get it up there? >> Yeah, but of course, that's the other like what I'm saying is on the R&D side it's a joke. The R&D is not a problem. You can get this done by 2028 and have that thing commercially ready including manufacturing. So you're saying well there's one last So I'm saying there's no other bottleneck than than than orbital lift capability. >> Yeah. >> And that bottleneck is already removed. When you run the numbers on Starship this is already cost-effective just with current Starship without reusing Starship. If you just reuse the booster it's already cost-effective in my in in my computation. If you make Starship reusable of course then it's even better. And so the only question is will the system be commercially ready even if Starship wouldn't be reusable in 2028? Of course it will. It's already like re-landing the boosters. It already delivered orbital satellites, 20 Starlinks. So there is no What's even the problem? Manufacturing these things, sure. It's It's now coming down to can you manufacture a lot of these satellites and can you make manufacture a lot of Starships? But how is this a problem? Like if these things work >> Yeah. >> So the only remaining little problem is can Starship like will Starship be robust enough by 2028? But I see the risk for that extremely low. These things fly to orbit. They re-land the booster. They will do it 50 more times before then. So what's the problem? >> Yeah, I agree with you. I I think you've been you're arguing whether or not it's a problem and I know there's a lot of people who are still debating it's a problem. Uh I'm just talking about time time frame. When is going to happen? 2029? Yeah, I agree with you for sure. I agree with that time frame but not 2028. >> think Terra will be at any scale close before that. >> by then anyways. So there's no need to sell it is what you're saying. Okay, that's fair. Uh so again this is a story is all good Uh for physical AI, I was just trying to see whether or not there was any kind of potential for the for investors to realize the physical AI is coming and they're going to get hyped about humanoid robots, they're going to get hyped about all this and whether or not some of them at least might start betting on this earlier before prove it stage. Most of them won't. Most of them will wait till earnings. So I think >> of course the stock can always like get up 100 points or something, but uh the the promise as I said before at that valuation we are now at, people just want to see money. >> Yeah. >> Stories are nice, but you know, this is really show me the money time, especially because you have other competing stocks not on the same thing, but on your money. Like you have >> showing sales now. Yeah. >> Right. So it's about that, but I think the good news for Tesla is you have robot taxi that will show you the money. And you don't need to have crazy amounts of money shown. You have to have the product, the TAM, and the product needs to be in the market and scale very rapidly. That's what you need. >> Got you. Let's hope you're right there. Appreciate you. Great conversation. Thank you, Joe. Check him out on his website, pioneerlands.org. Thanks, everybody. >> Thank you.