CyberLore Reacts to Tesla Optimus: The Delay Nobody's Buying

Vídeo em Inglês, publicado pelo canal CyberLore.

About this content: The summary, key points and topics below are generated automatically from the video transcript. They reflect the content, opinions, claims and predictions presented in the video and do not represent a position, validation or fact-check by IA Dots.

Video summary

The video features a panel discussion analyzing Tesla's decision to delay the Optimus Gen 3 reveal, Elon Musk's statements during the Q1 earnings call, and the complexities surrounding humanoid robot manufacturing, supply chains, and production timelines.

Key points

• Tesla delayed the Optimus Gen 3 robot reveal to prevent competitors from conducting frame-by-frame analyses and copying features.
• Panelists debated the claim of 10,000 unique components in Optimus, comparing it to human anatomy and questioning if it causes production bottlenecks.
• Concerns were raised about Tesla's reliance on existing vehicle assembly lines (such as the S and X line) for manufacturing humanoid robots.
• The discussion highlighted the potential use of a Robot-as-a-Service (RaaS) model to control distribution, gather data, and prevent reverse engineering.
• Experts analyzed technical bottlenecks in humanoid robotics, such as wrist design, local manipulation, and sensory integration.

Topics

Tesla OptimusRobotics ManufacturingSupply ChainEarnings CallRobot-as-a-ServiceAI IntegrationCompetitor Analysis

Video transcript

Transcript from available captions; it may contain errors.

What up do? Welcome back. I got another one for y'all. Cyberlore is reacting to Tesla Optimus. The delay nobody's buying. >> So Elon says he's delaying the reveal of the Optimus Gen 3 robot until closer to mass production because competitors tend to conduct a frame by frame analysis and replicate all the new features. Production is now targeted by August this year. Scott Walter is a mechanical and aerospace engineer, two-time founder of robotics companies. Gustav Anderson the good doctor is a hand surgeon and associate professor of anatomy in northern Sweden. Maridat Farimmani also based in Sweden. He's been designing humanoids and other service parts for well over a decade now. He also advises startups and investors on the robot product roadmap. Welcome gentlemen. We have a lot to cover today. >> Welcome everyone. >> So he's saying they're delaying the Optimus Gen 3 reveal to avoid competitors copying their work. That's a bold move. But honestly, if this is Tesla's biggest product ever, shouldn't they be ready to show it off? This could either be genius marketing or a sign that things aren't going as smoothly as they hop. >> Frame by frame, huh? He he may as well just pointed to me and said I'm the reason. >> I know. And he's he's been responding to you a bit on uh X these days. So, >> no, not sping responding directly to me. Let's let's set the record straight. like within hours of last Sunday's podcast where we sort of talked about where the hand is and everything in in the patent, uh Elon came right out and said that yeah, it it doesn't work. But he was replying to humanoid hub. Uh not to me. I I just happened to be in that thread, that's all. Well, you happen to be a very important part of that thread >> because you were the one breaking down the patents. >> Pay tins. >> I'm going British today. Pay tints. I love how Scott brings up the timeline. It's been ages since we've seen anything from Tesla. If they've got nothing to show after a year, what are they even doing? My dude, this makes it sound like they're just stalling for time, and that raises some serious >> red. That's >> all right. So, let's let's let's go back to um we have to backpedal a bit and go back to the Q1 earnings call of Tesla. And let me just bring this. The humanoid hub. We're going to be referencing him a lot today. And a special shout out to Devong. I'll be sharing his newsletter which all of you must subscribe to. I'll put a link to that uh in the description. But more on that later. >> It's free to subscribe. By the way, >> it is free to subscribe. Please. I appreciate all the support for Devon. He does a brilliant job. So here we go. Elon on Q1's earning call. Optimus V3 is functional, but there are some aesthetic elements to be worked on. >> Wait, did they just say the first Optimus lines are meant for internal use? That sounds sketchy. If they're not even ready to deploy these bots externally, then what's happening behind the scenes? Sounds like they might be struggling to get it right before showing off. >> It may be unveiled in the middle of 2026. We're a little hesitant to show it off because we find our competitors do a frame by frame analysis and copy everything we do. So, we want to wait until it's close to production ready. And we all know who does that frame by frame analysis. You can see a lot of it here on over the horizon as well. He also said the first Optimus line will be where the current S and X line is at the Fremont factory. Final S and X patch rolls off in early May. The SX line can't be torn down instantly. Dismantling has to start upstream. Um and interestingly, he's refused to predict 2026 Optimus output. A new product line. New line. Uh, >> this whole 10,000 unique components thing feels like a marketing buzzword. Really? 10,000? If it's true, how's that not a nightmare for production? I mean, my guy, it sounds more like a recipe for disaster than an innovative robot. >> 10,000 unique components, ramp speed gated by a single slowest, most troublesome component. Initial ramp will be slow. First task will be simple in factory work, then progressively more complex. So, we've heard him say a lot of this before. We've discussed a lot of it here on Over the Horizon, but couple of things that stand out. First of all, is this um awareness of what the competition does, and we've discussed this in the past year as well, but close revealing Gen 3 Optimus closer to starter production is a smart move, but it also raises a lot of questions about the delays and what's really happening. I mean, in the start, they could kind of they could say this. I think that >> they're saying the initial ramp will be slow. That's not exactly confidence inspiring. If they're already predicting slow production, how do they expect to meet any future demand? It's like they're setting themselves up for failure. >> People were looking at them, but they haven't shown anything for like a year. So, what are people looking at? It's not like people are copying anything from B2, right? and they're evidently not copying anything from the hand either now that the patent's out. Uh, everyone assumed that that was like the state-of-the-art and you should copy that and then he comes right out and say, "Well, oh, don't copy this." And so, yeah, I'm not buying any of it. Yes. >> Unless that was a scop. >> Yeah. Um, so basically, do you want my unvarnished opinion or do you want it a bit tempered? >> No, no, no, no. Unactive. We keep it real. >> Okay. So, so I hear I mean, first of all, and this is is not just me, but a lot of others felt that was like one of the worst earnings calls ever, and that's been a couple of really bad. >> Scott's hitting the nail on the head. If they want to show us something groundbreaking, they shouldn't have to wait for the production line to be fully up and running. Why not reveal the tech when it's ready? It's like they're holding back a treasure. >> And and part of it is um there it seemed like there was a lot of spaghetti being thrown against the wall to see what stuck. And I couldn't quite tell what was the reason because this whole idea of our competitors have been coping us have been around for a long time. So you could just stick with that all along and and not use this whole thing about the SNX line and how difficult it is to break down. And there seemed to be downright hostility to like the question that was asked which is a been asked many times. It's on say he knew what it was and he questioned the validity of the question in the first place. Uh only to be kind of contradicted by his own past statements on that. So, I'm uh I'm not buying any of it. Um so, and I'm not surprised because there had been, let's say, >> so they're saying that the aesthetic is the last piece to tackle. That's wild. If the tech works, why not just push it out? Sounds like they're overthinking everything. Sometimes less is more and they might be losing sight of that. >> A lot of signs out there that they've had trouble getting V3 to the point that they can go ahead and show. And I think we can point out a few things where I I think u become problematic in in Elon's opening statement he has said something he said many many many many times before and that is regarding you know this like you know that the optimist will be our biggest product not just Tesla's biggest product ever but probably the biggest product ever if product ever then it should not depend upon whether the SNX line gets disassembled in time because for some nostalgic reasons you decided to run it another month or something like that. Either it's the biggest product ever and you're ready to build it or you're not. I don't buy that excuse. You do not need >> One thing's for sure. If they're still in design mode, that's a problem. How can they call this their biggest product if they can't even get past prototyping? It's not looking good, folks. They need to get their act together. >> An entire SNX line to produce the bots. Okay? There is not going to be a million line, a million bot line. there's going to be multiple parallel lines that are able to do it. So, whatever it is and and you don't start building a million optimi, you start with that first building a thousand. They haven't done that yet. You then you figure out how to get to 10,000 everything else. There are a bunch of other things in there that I think is also should be pointed out is if Optimus is made up of 10,000 unique items, that might be the problem. 10,000 unique items in a Model 3. There's maybe 20 or 30 in in an ice engine. In a humanoid, there are far less than that. So, I'm not sure where that number. >> Okay. But if they think they need to hide from competitors, what does that say about their confidence? Like, if they're really that worried about being copied, maybe they should focus on innovating even faster instead of stalling >> is coming from. So, a lot of things that seem kind of funny to me. So the the first thing is that I I thought was very odd um was like the he says basically the V3 Optimus design is almost ready to demonstrate and I'm like you're still designing. I thought you were prototyping at this point. So and he said that a few times which is is already kind of concerning. Uh I think I think we want to make sure it's polished. Okay. It's it's it's polished. To me, it sounds like, oh, we're just kind of going through and and getting the reinforcement, learning, getting that stuff. Make make sure it it is kind of able to work within that. Um, like it works functionally. Okay, so it's working functionally, but there's an aesthetic element. Okay, suddenly we change its aesthetic element. It sounded like the cyber cap. >> The talk about the production line and dismantling the SNX is just odd. If this is such a crucial product, why mess with existing lines? They need to prioritize getting Optimus right without sacrificing everything else >> all over again. It's like, oh, we want to put a we want to put the Cybertruck light bar on there now. You know, it it it needs to be finalized. Um, we think the middle of this year, okay, it's the middle of the year. Uh, we should be able to show it off. And now suddenly it's like, oh, we're a little hesitant to show V3 off because, uh, we find our competitors do a frame by frame analysis. Well, you've known that for years, so why didn't you just go with that right right from the beginning uh, whenever we release something and copy whenever it's possible. So, I think there's some value to not showing the technology until it's close to production. I don't quite understand then why is it suddenly when it's close to production, why isn't it when it's well after um when it's actually ready to go out into the public? I mean, you can hold that off as long as possible. >> Yeah. >> Um and and the thing is they should be able to do it because unlike when you're doing production of >> I'm sensing some serious skepticism here. If you don't need the entire production line to build the bots, why act like it's holding everything up? It feels like they're making excuses instead of taking action >> of uh you know like the cyber cab or the cybert truck or the semi at some point it has to leave the production hall and go off somewhere else and then Joe Tegmire's out there capturing it driving around the optimi are not going to be running around the campus so there can be people out there with telephoto lenses you're not going to see it you're not going to see it unless you're you're inside the building and they're very good at making sure that anything that's inside the building does not get revealed to the outside unless they want it to. So, at the same time, if you're if you're worried about that, keep buying yourself another couple months and and only reveal it when it's absolutely ready. So, that all having been said, I mean, that was like kind of the first thing. And you could see it was a really bad earnings call because as soon as the earnings report went out, the stock finally went over 300 after hours. >> Everyone liked the results. Yeah. >> This idea that their first bots will be basic factory workers. I mean, it's smart, but also kind of sad. If Optimus is going to be the future, why not show us something more ambitious right out of the gate? >> Yeah, it went over 400. So, it was it had like dropped quite a bit uh the week before, >> suddenly climbed back up. Looks like aha, maybe it's finally going to go. And if you just kind of watch the after hours, as soon as Elon spoke, it just dropped like a stone. And then then I mean that was like his initial comments and then after it it decided to drop even more because he kind of doubled down on the whole thing again being let's say the very hostile towards the actual question itself which seemed a little bit out of character and then again kind of contradicting everything. It's like we don't know what the production rate of optimist will be this year. It's impossible to predict these things. Again going to the 10,000 unique components you know it's literally impossible to predict. Now, if you listen, he had been kind of consistent because unfortunately a lot of >> that's a tough pill to swallow. If they can't even get basic tasks going, how are they going to wow us with anything complex? It's like they're building a castle on shaky ground. >> People were going with the million optimist line and saying the million optimist lines, you know, at the end of 2026 going to be and it's like that was just patently or is it patently false? Something like that. That that that's not the case. If you listen really closely, he had been saying that we're targeting probably was always terms like that and that the ramp, the initial ramp will be slow. He has said that many times. So, he is consistent with that. But it's still kind of odd that he's going from that because, you know, I mean, just to inject some reality uh in January 2025 earnings call, uh Elon mentioned, you know, 10,000 Optimus in 2025, but success would mean several thousand built and doing useful things by end of year. That's end of year. >> Hold up. They think they might need AI5 for Optimus. Sounds like they're pushing back their timeline even more. If they can't figure it out with AI4, that's a huge red flag. What's really going on here? 25. Okay. >> Um and so you know something does not fully understand what happens with the production line. You know and a surprise all hands meeting uh >> in March of 2025 again doubled down on that and said that this year we hopefully will be able to make about 5,000 optimists. So either you know what happens on the production line or you don't know what happens in the production line and you don't go from 12 months spouting numbers to suddenly saying it's impossible. So, you know, I really think he owes an apology to the the people who asked the question and people who upvoted on say because it was a very good question in that people who produce stuff generally have targets and ideas and production ramps and kind of know the stuff and now suddenly it sounds like it's amateur hour over there. They have no idea. >> Elon's comments seem a bit confusing, don't they? If they're planning for internal use first, it feels like they're not ready for the big stage. I mean, come on. Let's see what the hype is about already. what they're doing. >> It felt like he hadn't read the questions in beforehand either, unless it was not a surprise that he that he kind of >> I don't know. Everyone is coming up with theories why it was such a >> I agree it was not the best. >> Okay, so you want to come up with my theory? >> Yes. >> I mean sounded like oh this question I haven't really thought about that but let's come up with something. I mean the questions had been out there is >> everyone knows that the whole SNX is just a fig leaf. It it always was. was I mean that the whole idea they're doing that it's it was just to kind of buy time because there may have been some issues. The other thing is like the end of that question was like what's the production rate? I mean they did not actually ask how many are you going to be building this year. It's like what is the production rate which is something that I think is a little bit easier to to kind of >> the conversation about Raz is interesting. Robot as a service might just save them from the chaos of selling but they better have a solid plan for keeping control and preventing reverse engineering. >> Answer that that question. And then the other was like and what will the skills be? And then it was just like it's we're just going to do simple stuff and build up from there. Okay. But the idea is like what are the skills that you're gonna be producing going to be useful? They're going to be things are going on the factory. And again I've reiterated this again and again the best lab you have right out there is the SX. I mean, and and this whole idea we're going to tear down that line to build all these optimiz are going to be working on the model 3 and the model Y only result and it takes you a long time to to get them up and running and there's going to be hampering the output and stuff like that. You don't do that. You know, you're not going to put them over the until you've been able to prototype it or prove it out somewhere else. Where is the best place to do it right now? So maybe this whole thing is just like a smoke screen to say, "Yeah, we're shutting it down and we >> I can see why a RAZ model could be beneficial. It keeps them in control, but it also raises questions about transparency. If they're leasing, how do they prove the bots are worth it to climb? >> And doing some reorganizing. People may have noticed like over by the where all the sublessities are, we're doing a little bit of reorganizing because what we're trying to do is we're we're trying to mir change the environment to be a bit more friendly automation and we'll get you get into that that they might actually be setting it up like that because that's what would make more sense to me. They they >> it could be a sandbagging. It could be another type of thing. >> Smoke screen for what? >> Yeah. For for for those of who are just ready to like unfollow. Okay, that's that's my little little possible glimp. What do you think? >> Yeah, >> I actually wanted to say something positive and something negative about it. So, the positive part was that I think he's right about the the aesthetic functions. Be the blocker. >> I'm with them on this. If they're struggling with basic tasks, how will they compete with other humanoid robots? They have to step up or risk being left in the dust. >> Because many people underestimate how much time and effort those things actually take to make it look good in the end. And the pointing point is that is usually the last step in in developing the robot. So that aesthetic functions because then uh it means everything else is done. It's just the aesthetic because aesthetics are actually affected by everything else. >> Okay. explain aesthetic functions to for the audience. What is it exactly that you mean? >> True. So, so in my opinion, it's basically whatever that the user sees like mostly untrained non-rootic people, that's the first thing that you usually see. It's like the proportion, the the covering, the paneling, the the the sophistication of how things are coming together and the lights and interaction and those. >> Hold up. They're comparing Optimus to other humanoid robots. That's a bold claim. If they're not ahead in the game right now, they need to step it up or risk losing their ed >> kind of modules. >> The panel gaps, >> the paddle gaps, >> very important. People who are in the car industry, they they notice that the first they're like, "Oh, there's a gap here. It's like not properly done." But anyway, so uh I I can agree that that part takes time and if that is the the remaining part then it means that it's the last thing. But at the same time, um there are a few arguments here. One one of them is that um usually aesthetic features are very predictable like you know exactly what things you need to do and how long it takes. But the part that he says we don't know exactly when that is like mid 20 26 or something that is a bit odd because again if there is aesthetic functions left you you can have you can clearly predict how long does it take until you finish >> this whole 10,000 parts thing. Yeah. It makes me think they're over complicating. If other humanoids can do the same with fewer components, Tesla needs to rethink their approach or risk falling behind. >> There is nothing crazy. It's not a concept design what their face look look like or anything like that. It's just putting together pieces basically. And then the second thing that is that seems odd to me is that if the goal is just deploying in the manufacturing line or like industrial setup, then the aesthetics doesn't really matter to be honest. >> So yeah, that that that's how I can see it. Otherwise, just to make it clear, I'm not a shareholder, right? So, I have >> Which is why we appreciate No, no. Which is why you're you're you're the most neutral here. I mean, I don't think any of us are not. We're pretty objective and we always try and be um grounded and real for our audience here on Over the Horizon, which is what differentiates us from a lot of the hype. >> The numbers they're throwing around seem inflating. If the human body has around 1,200 parts, why would a robot need 10 times that? It's like they're trying to impress us instead of just delivering. >> Um >> but as shareholders, our at least my emotionalities are affected by the move. >> Yeah. So what Okay. So I agree with a lot that you said Barad and I understand exactly what Scott has been saying and Scott to be honest we've we've said and you have said a lot of what you just told us today here on the channel previously as well. Right. So I don't understand whether this was an off day for Elon or whether and he's historically not cared about earnings calls as much. He would rather not be there and perhaps he ought not to be there because there >> I get what they're saying about the supply chain, but if they can't streamline it and manage costs, they're going to have a tough time. They need to focus on efficiency first. >> He's got a great team around him that can do the job just as well. So I I'm wondering what's going on with the big picture and why all of this is happening. >> May May I have a comment here as a non shell shareholder again >> I think Tesla shareholders are putting too much faith in a person that that is very very dangerous. Uh you you are sharing all this knowledge on this podcast and all the other channels about this robot about Tesla itself about their cars and all the other products. So people should just objectively look at how the companies are doing instead of just listening to a person what that person says. I'm sorry for bringing this analogy, but he could be a person like Trump, right? That he could say, "Wow, they're really leaning into the long-term potential of Optimus, huh? But if they're not delivering now, what's the point? Promises don't build robots, actions do." >> Something and then the next day he could say something opposite to that. So I think just relying on a >> but here's the thing that's very uncharacteristic of Elon to have to do that now which is what is causing all the confusion >> to be honest I understand I don't even care to be honest what a what a CEO says it might sounds very odd because if you you guys are all technical so you know that whatever that is happening in the technical side this the the leadership they can say a lot of different things or things that might prove or disapprove this. So, so not listening to leadership and mostly focusing on what the advancements are, I think is a better matrix. And also like comparing stuff for example with other companies. Exactly. Like seeing >> that's a big question mark. If they can't nail the basics, how can they expect clients to trust them with a complex robot? They really need to showcase results if they want to gain traction. >> All these human companies we have the top 10, how is Tesla doing compared to them? Which is not bad. Tesla is on top three, right? So >> do we know that nowadays? I mean I haven't seen anything new in over more than a year I'd say. >> That's actually a good good point because every week you are seeing new human companies popping up and new updates are coming up like later we will talk about the update on unit 3 with the wheels right >> but also >> and and and let's not forget um what Brett has been posting. There's some great videos and some really interesting numbers a chart with some really interesting numbers. So when you contrast and figure is definitely uh out there at the top of the stack if not amongst the top three for sure along with Tesla. So I mean if you're going to contract >> I think they're right to question the credibility of the 10,000 parts claim. If they can't substantiate it could backfire. Transparency is key especially when trust is on the line. >> Optimus with the figure bot then it's pretty clear when you see the numbers >> as far as far as the updates. So, you know, going to to Gustav's point, we haven't really seen any real update of the of the 22 do hand. And you know, what we've been seeing out there in public, we know is like this not the one they're going to use. So, when it was being shown at the shareholder meeting, they already knew that they this is not the hand. We're just going to show it off. So, uh we saw it grab a tennis ball well over a year ago. Um I think about a year ago um we saw them that they were getting the the RL for the walking but that wasn't like super impressive compared to what everyone else and I would say like the most recent update of Substance was back in the fall wasn't it? We actually got to see Optimus running. >> Seriously though, if Tesla thinks they've got the most complicated bot ever, they need to prove it. Otherwise, it sounds more like wishful thinking than actual innovation. >> Just across the frame, just zip and that was it. you know, nothing more than that. Not like much, but we could see it finally it can do something more than a snail's pace when it was moving along. So that's it. The other surprise, and I think Gustav, you'll probably also be surprised in that, is that everything that was talked about Optimus, there was absolutely no mention of the hand and where the hand is. You know, it's like length is kind of functional, might be doing something, but that had always been a very front and center. The 22 do hand, the most dextrous hand is able to do this and that, and suddenly it's kind of out. And then I I I would be surprised if it will be the fall I think before we see the reveal because right now it's already late July, August based on the production line. That could be delayed, right? It's happened before. >> There's definitely a gap in the narrative here. If they can't truly back up the claims about uniqueness, it could spell trouble for their credibility. They need to be ready to deliver. >> We've been given kind of a summer date and it slides a little bit for a bunch of reasons. One of them is like no one's paying attention the summertime. So you don't do these reveals until September. So I think it's it's deep in Q3 or early Q4 that we might be able to see something. What I would love to see this is this is what I would love to see is that you got to do something which is just different than everyone else. And just think about, you know, some of these um late night shows that have like a cold open where someone's like walking around backstage or something like that before they come on there. Just imagine that they start that thing and Elon is like walking the SNX line and there's just a bunch of optimists and maybe there's like an optimist walking along with >> I'm with them on this. If they're not hitting those production goals, that's a huge issue. Transparency and accountability are essential if they want to regain trust. >> Damn. >> And he's chatted with him. So, how's it going? you know, what's been the downtime going on here or something like that. And if you just see a bunch of them doing that, you know, and that you actually see the line populated by that and then he just walks up on stage and Optimus comes out the other door and they sit down and they have a fireside chat. I don't need to see back flips. I don't see, you know, maybe Optimus comes out and gives Elon a drink or something. Doesn't have to do anything on there. If I'm seeing that that's out there producing that, that's it. You don't have to show me anything more. And yeah, >> maybe that's what they're planning on doing because I keep on telling you, I will buy an S or X, maybe both. >> If it's got the little badge on it that says body by Optimus. All right, so maybe that's what they're planning on doing. That's what they'd have to do to hit it out. >> It's all about perception. If they're focusing on being the biggest and most complex robot, they might miss the point. Sometimes simplicity can lead to more effective outcomes >> the park but otherwise uh it's going to be an uphill slog to be able to prove everyone that you've got something that's going on because we are already seeing videos of China coming out of China of bots on these lines doing work. >> Yeah. >> I mean I wonder if they are a little bit stressed about that that they haven't managed to kind of beat them beat the others to it. So like the whole cleaning of the living room by the Figure bot and all the different demos out there, it feels a little bit like, oh, they managed to do it before we managed to do it. >> Push to the next. >> At this at the same time, none of those bots is going to be able to go down into my kitchen and unload my uh dishwasher or load it because my dishes are just so unique and where everything are. We are so far from any. So So they're they're again doing these these lab things. >> Exactly. If they can't deliver on expectations because they're stuck in complexities, it's going to hurt them in the long run. They need to adapt their message to be more realistic >> and not necessarily showing anything that I consider to be of value. >> And if if you are seeing them out there on the line, you know, I literally it's like the bots are running the line. >> Yeah. >> Then suddenly it's like that is what we've been talking about. That's what we talked about way back to AI day one is the whole idea was going to make Tesla more uh you know the margins are going to get better because you're eliminating a lot of the labor cost. >> That's a great point. Yes. >> And you don't need a V3 hand for that. I think >> I mean maybe not some some tasks probably but a lot of the tasks you could probably do with your basic >> also don't need the legs so they can simplify it quite a lot. >> Yeah. Yeah. You could just have a torso. >> Yeah. >> Well, it depends upon the task. I mean, depending on where you are on the assembly line, sometimes you do need >> Honestly, they really need to show results. If they're constantly hyping up the product without tangible outcomes, people are going to start losing interest. Actions speak louder than words >> to move back and forth. That's what's unique about automotive versus like iPhone production. If you're producing an iPhone, you can be doing it sitting in a chair, but automotive, it's it's really rare that you have those subasssemblies that you you're just standing there the whole time. It's just like a few steps to the left, a few steps to the right. Every now and then, I'm going to turn all the way around, grab something. So there could be some utility. >> So just so our audience knows, Optimus is being produced, but it's mostly built being built by hand. It is being built by hand right now. Every Optimus spot. >> Yes. I don't know how many more of the of the 2.5 they're producing right now. So there there is a pilot line. So they could be using that pilot line to produce the V3 uh when it's ready. Um you know, the numbers won't be huge, but you know, it could be enough for training purposes. >> For real. It feels like they're dodging accountability with all this talk of complexity. If they don't simplify and focus, they could lose ground to competitors who are ready to deliver. >> But the demo that you're talking about is is with a with a with a with a full-fledged production line where the SNX line is. >> Mhm. Mhm. Okay. >> Yeah. Be that. Sorry. >> Yes. No worries. Um so one thing that is happening inside the human and robotic labs is that a lot of these researchers they are kind of connected with each other and they they kind of know what is going on in the industry and they almost all of them they know that there are issues or there are challenges that no one has solved it. So even if they see like some demos from figures, some demos from Chinese companies, they are still confident that no, that's just a demo. You know, what are some of these a few? >> What are some of these challenges? >> I get the need for a detailed supply chain, but they shouldn't let it bog them down. They've got to be agile if they want to remain competitive in the fast-paced robotics world. >> Okay, that's okay. One one example I think we discussed a few weeks ago is the wrist. when it comes to hand like wrist is extremely complex and we haven't really yet figured out the best configuration even even the patents people were very hyped about how cool the wrist is but then they said no that that design doesn't work so it means that the wrist is still a bottleneck so in in a lot of Chinese demos when I see uh when I I usually look at the there was I think it was Agibot or something actually showed that the robot is assembling a actuator by itself or robot is building itself And it was very very much like CS in the video and you couldn't really figure out if if it successfully did the task. But I was looking at the risk and I was thinking I was saying that they they manage most of the movement. >> The back and forth on production levels is concerning. If they can't stabilize their supply chain, they're in for a rough ride. Innovation needs room to breathe, not be stifled by logistics. >> It's from actually the upper arm. So upper arm is basically doing the job of the risk wrist in that cases. But no, wrist is still a big bottleneck. Then we have um what I always usually uh I said it once that this is more like when you have locom motion and manipulation at the same time which they call it local manipulation is a is a very complex at least on the control side is a very complex as aspect. Imagine it such as while the robot is walking, while walking, it also grabs something, grabs a tool or something. And these are very very complex cases, but we we barely see them. But in reality, this becomes very important because a automotive manufacturing worker while they are just as Scott said, they step 2 m this way, 2 m. >> It's like they're stuck in analysis paralysis. If they're overthinking their components, they might miss the opportunity to make something impactful. They need to find balance between complexity and functionality. >> That way while they are doing that they're also doing stuff with their hand. >> So and but if robots cannot do this yet then they would be extremely slow extremely inefficient compared to how human is doing it. And there are a lot of other examples. So uh my point was these companies are not really in the scarcity yet because they know that oh no not even Agibot not even unit three not even figure they haven't really figured out this and this and that. So therefore we don't we don't see it as a competition. It's more like we are all exploring here you know. >> Yeah. Yeah. They're not they're not fully optimizing yet. They're they're that's something you want to do last. But in the end, they're going to have to because when you start getting these high production rates, you're >> bro, if they're counting every little resistor, they might be losing sight of the bigger picture. Innovation isn't just about parts. It's about creating something that genuinely works. >> Trying to shave off fractions of a second here and there wherever you can by doing things like that. So, you know, one example is like a spot welding robot. It used to be the gun was open, you would go in and then once you're in the location, you would come down to the spot, you would open up and you retract. >> Well, you were wasting time on that. So, what they were doing is like as they're coming in, it would start to close down and get it, hold it, and then as it started to get up just enough and then retract out. And that was a very difficult thing to master because it used to be that there was there was no connection between the gun and the control system. They weren't kind of aware of each other. It was like, am I at the position? You're there now. Send the signal. Boom. and it would go down. So now you had to integrate then as an additional axis to be able to get that. But that was a difference between like one robot company winning the order. >> I love the comparison to the human body. It really puts things in perspective. If they're trying to outdo nature, they need to rethink their approach to design and efficiency >> from a big automotive company or not is the fact that they had something like that. So So we're not there yet. I agree. And and we are incredibly efficient. You know, we grab the keys as we're running out the house, right? We don't stop pick them up. It's like I'm in a rush. I got to just go in one motion >> without even looking at it. Right? That's also another point. Robots need to look at things to be able to perceive what is in the environment. And then there is this also very complex things in robotic that there is a correlation between the eye and the hand and the the eyes know where the hand is basically which is a very odd concept for us human but for robot it matters a lot that the robot's hand need to the robot need to know where its hand is in the space and whatever sensor you use they they might have drifts and it might not be accurate. >> I'm curious to see how they'll handle the challenges ahead. If they can't integrate all these ideas and components they'll be left in the dust by companies who can't. But exactly as Scott said, the other example is just for example, if I want to grab this, I just go here and grab it like so quickly, right? For a robot, they open their hand and then they go towards it and then they close it. Why? Because if their hand is already closed, it's very likely that with the momentum that it has, it might actually hit the object before opening it hand. So and again they can be solved but but the point is we haven't really got into a point to spend time on this kind of things or to make it more efficient as Scott put it >> and our hand is crazy crazy smart I don't know I don't nobody knows how it works so >> but then there are also other there are other layers of of there are other sensory layers that come into play right >> they really need to get their priorities straight if it's all about efficiency they should streamline the process rather than complicate things with excessive parts. Focus on what truly matters. when your hand is out of literally out of sight the human hand. >> Exactly. >> And we don't have those sensory layers yet or to the to the level that we have them >> and also I mean you talked about this before ma but even if we did produce all these different like sensing or >> we understand 3D space but do we have the compute to actually have the robot run and pick things up at the same time and control or will the battery be dead in 30 seconds I mean or the computer won't be able to do it. Yeah. And and speaking of which, I mean, um, this was like one very interesting point that Elon made that may have glossed over and I've talked to a couple of people who who were like a bit more expert in in this domain to say, did he say what I >> I agree. If they can't figure out how to integrate all these functions, they're going to fall behind. The future of robotics demands adaptability, and they need to be allin. I thought he said so I I I will quote there was a question about uh the AI5 which you know they they just did the tape out which means it's still easily months away before you get the first samples which are going to be full of bugs which means it's it's a little bit longer. So in the best case 6 months to a year before you start getting some that you could could begin to use. Um yes um I do expect that AI5 will go into Optimus into the data centers because it's looking like uh we'll be able to achieve unsupervised uh cellulose driving with AI4 um that is far greater than human safety uh safety levels. Uh so which means it's not certainly not immediately needed in the car. At some point I think we'll will make sense for us to switch to A5 in the cars but but that's not uh what's pressing issue for us >> and that's the kicker. If they can't capitalize on all this potential what's the point? they need to get it together and actually deliver on their promises if they want to stay relevant >> to do. What was almost the saying is like are you saying that you need AI5 for Optimus? And I think I also had like heard other sort of statements like that that in order for Optimus to reach its full potential they need AI5 and they may not be able to do it with AI4 even though AI4 may be sufficient for that. So it was like the the way that was kind of phrased and everything seemed to hit me a little bit odd and a few others that I talked to that were listening very closely and say yeah that could also be one of the things which will limit the number of Optimus until AI5 is available and it's still like I guess say easily half a year away if you're lucky in the in the first batches come out bug free that almost never happens. >> Yeah. Yeah. But then Elon also said that the >> if they're waiting for AI5, they might be waiting a long time. The clock is ticking and if they don't start showing tangible results, they might lose the momentum they built up. >> First Optim to be produced um off the production line would be for internal use. >> Yeah. >> And then only next year perhaps to >> and which is why you you're hearing him say the production is going to the ramp is going to be very slow. Okay. And and probably by the end of the year so there'll be something to reveal. I just don't think there'll be the numbers. So, you know, unfortunately, I don't think we're going to see the SNX line populated with Optimize because I'm not sure they'll be able to produce the number that you would need to see that at that point. >> So, when he talks about external use, is he is he kind of hinting at Tesla selling Optima to others or deploying them or leasing them? What do you think? >> I certainly hope he's doing >> I hear you. If they're hinting at using AI5 for Optimus, it better deliver. The hype has to match the reality. Otherwise, it'll just be another missed opportunity in the long line of >> it as um as a RAS model, >> which I think they would have to do in the beginning. Um >> robot as a service >> robot because you you keep control of of what's going on. Um you can then prevent people from reverse engineering or tearing apart because you know if they're going to start selling it, they're going to be a black market and and optimize. they're going to get shipped around the world for people to do tear downs of. So, so the main thing is you want to set up that way that you're kind of controlling it. You're getting the data and everything else at first. Eventually, maybe you start selling it um when the market sort of forces you to do it, but right now the market is everyone is happy with the RAS model. Uh it's really rare that you don't. So, and I'm just talking about for for humanoids. There are >> It's interesting that they're focusing on internal use first. if they can't prove the concept internally, how will they sell it to the public? They really need to get their priorities in order. >> A lot of uh sort of physical AI companies out there that are using very simple bmanual arms are going out there and they're getting a lot of traction, especially in smaller companies because they're saying we just want to do an opex swap out. We don't have capex. There are some of the bigger Fortune 500s that prefer a capex model and they may want to buy it, but at the same time, you can say doesn't matter. I mean, it's like we we've got so much demand and so little supply. It's on our terms right now. And again, they they I mean, they also have OPEX, so they can just swap it out from for accounting purposes, but I think they just like the tax write offs of capex, but it's not going to be there yet. You you don't want nobody knows whether this this capital is going to last like five years. It could break down in five months. >> Uh and then then you have to replace it. So until you >> robot as a service could be a gamecher but if they don't execute it well they might end up creating more problems than they solve. They need to keep a solid grip on their tech. >> You get uh proof that these things are are robust and and what the long-term nature of these things are. You're not going to see really want to go capex. It makes sense just you know the maintenance is your headache the programming everything is your headache. I just supply me labor. >> Yeah. Yeah. So Merida and Scott for both of you this question. The 10,000 parts number is a bot really meant to be that complicated in its design and supply chains. >> Definitely not unique. And he said unique parts and I don't even think there's 10,000 parts total. >> Okay. >> If you're counting fibers and stuff like that in the tendon, does that count? It's like you >> it'll be crucial for them to maintain control if they go the RAS route. If they let others take the reigns, they risk losing valuable advancements. Keeping their edge is more important than ever. >> That's a lot of >> maybe they also concept the number of the let's say electronic like IC and resistor and everything on every board. So yeah. >> Yeah. I mean if if you want to go down that rabbit hole, yeah, you you can. But when it really comes down to like unique parts or SKS or or whatever you're getting from the supplier, there there's no way that there's that many actually on the assembly line. >> I mean, if you take the whole human body, and I'm not counting nerves, and vessels, but if you count bones, joints, and muscles, it's like,00 1200. So, it's like a tenth of what he wants to put in the bot. And sure, you would need some cables for the nerves and the battery, whatever, and the brain, but I mean, it's not. And maybe he's counting each individual little >> I get the appeal of RAZ. It allows them to manage their robots while still getting revenue, but they need to ensure that the tech is robust enough to support such a model. >> Sister and whatever on the on the chipsets, but I mean the human Yeah. >> and every permanent magnet. >> Yeah. >> But >> well, they're all part of the supply chain, right? They're all part of the global supply chain. >> They're they're all part of the supply chain. But he he's referring to, you know, pretty much um the normal line that that they have. And and the one thing that Tesla has done is is they've taken the the SKUs that needs to go into an ICE vehicle, which can be 20 to 30,000, and they've cut down to 10. That's one of the advantages of having an EV over internal combustion engine. You don't have like all these other moving parts. So that's already now Optimus is going to be a fraction of that. >> So I have two arguments here. I totally agree with Scott. It's a very exaggerated number. But my my first question is so isn't it the same for other humans? >> That's a solid point. If they can't prove the reliability of their bots, organizations will hesitate to adopt them. They've got to hit the ground running or risk losing out to competitors. >> Like it should be the same. Like the amount of complexity should be almost the same for others because these guys are obviously not doing something exponentially different. It's just in the same level. The second question is like what is the intention behind saying that thing? Does it mean that oh it's too complex therefore it takes time or no does it mean that uh I I don't even know like what what is the point because if I was there I was like it's actually no we actually degraded or downsize the number of elements so much so it's a good thing not that saying oh >> complex >> you know it's like being amazed by complexity is not the right thing we need to be amazed by simplicity right so >> yeah that's Tesla's moat manufacturing >> I'm glad are questioning the 10,000 parts claim. Realistically, if other humanoids can function efficiently with fewer components, Tesla should definitely rethink this inflated number. Simplicity often wins. >> It turns out Oh my god, man. I have no idea what we're doing. It's just >> That was the MO. >> Yes, exactly. >> All right. Well, I guess uh it's that earnings call raised a lot more questions than provided answers. So, we'll have to wait and see how things pass. >> Yeah. The one thing it didn't do is raise the stock. >> For sure. Yeah, thank you so much for your time, Scott, Gustav, and Merida. Thank you for staying on and uh we look forward to the next time. What do you think about Tesla delaying the Optimus reveal?

Assistir ao vídeo original no YouTube

Assuntos: Robótica e humanoides, Robótica e humanoides. Publicado no YouTube em 2026-10-30.

Seleção automática por assunto e idioma. O vídeo pertence ao canal de origem.