Elon Musk | All-In Summit 2024

27:44You have a bunch of critical milestones coming up. Yeah. In fact, this is a very exciting launch that is maybe happening tonight. If that, if the weather is holding up, then I'm going to leave here, head to Cape Canaveral for the Plarristone mission, which is a private mission. so funded by Jerk at Jared Eisenman and he's a Wilson guy. And this will be the first time, the first first commercial for Spacewalk and it'll be at the highest altitude since Apollo. So it's the furthest from Earth that anyone's gone.

28:28So I'm back for that. Let's assume that's successful. I sure hope so, man. No pressure.

28:42Yeah, we're absolutely, you know, Ashmore's priority, Ashmore's safety is, man, if I had like all the wishes I could say about, that would be the one to put it on. So, you know, space is dangerous.

29:00So, yeah, I mean, the next milestone, And then, after that would be the next flight of Starship, which, you know, Starship is ready to fly. We are waiting for regulatory approval. You know. Yeah. It really should not be possible to build a giant rocket faster than the paper can move from one desk to another.

29:33That stand was really hard. Approved. Yeah. You ever see that movie? Approved. You ever see that movie? Zootopia. There's like a sloth. I'm like, right. Come in in for the approval. Yeah, they actually tell a joke and they're like, oh no. Here we go. Sorry, sorry. But yeah, Zootopia, you know, the funny thing is like, so I went to the DMV about, I don't know, a year later after Zootopia and to get my, one of the license for Nurel. And the guy in an exercise of incredible self -awareness had the sloth from Zootopia in his cube, in his cube, and he was actually swift. with that. With that, with the main date, beat the sloth.

30:21Yeah, no, no. The personal agency, personal agency. No, I mean, sometimes people like, I think the, you know, the government is more competent than it is. I'm not saying that there aren't competent people in the government. They're just in an operating system that is inefficient. Once you move them to a more efficient operating system, they, their output is dramatically greater as we've seen. Exempt, you know, when and East Germany was reintegrated with West Germany and the same people were vastly more prosperous with a basically half capitalist operating system. So, but I mean, for a lot of people, they're like the most direct experience with the government is the DMV.

31:08And then the important thing to remember is the government is the DMV at scale. right. That government. Got the mental package. How much do you want to scale it? Yeah. Yeah. Sorry, can you go back to Chimoff's question on starch? So you announced just the other day Starship going to Mars in two years and by the way. Yeah, yeah. Yeah, yeah. And then four years for crewed aspirational launch in the next window. And how much is the government involved in asking? I'm not saying, like, say you're watched by these, not, you know. But these, but it based on our current progress, where with Starship, we're able to successfully reach over the last two twice.

31:55We were able to achieve soft landings of the booster and the ship in water. And as despite the ship having, you know, what happens, blast cooked off. You can see the video on the X -Flight form. It's quite exciting. So we think we'll be able to launch reliably and repeatedly and quite quickly. And the fundamental holy grail breakthrough for rocketry, for the fundamental breakthrough that is needed for life to become multi -planetary is a rapidly reusable, reliable rocket with a pirate somehow. So a pirate in there.

32:42So with Starship is the first rocket design where success is one of the possible outcomes with full reusability. So if we have any given project you have to say this is the circle to the red band diagram. It has the circle and it is success the success dot in the circle. It is success in the set of possible outcomes. That, you know, sounds pretty obvious but there are often projects where that is success is not in the set of possible outcomes. And so, so Starship not only is fully reusability in the set of possible outcomes, it is being proven with each launch and and I'm confident it will succeed, it's simply a matter of time.

33:31And, you know, if we can get some improvement in the speed of regulation, we could actually move a lot faster.

33:42So that would be very helpful. And in fact, if this, if something isn't done about reducing regulation and sort of speeding up approvals, and if you clear, I'm not talking about anything unsafe, It's simply the processing of the safe thing can be done as fast as the rocket is both not slower than Then we can become a space -boring civilization and a multi -planet species ultimate and be out there among the stars in the future and there's

34:17No, it's just very like it's incredibly important that we have things that that we find inspiring that You look to the future and say the future is going to be better than the past. Things to look forward to. Like kids are a good way to assess this. Like what are kids fired up about? And if you could say, you know, you could be an astronaut on Mars. You could maybe one day go beyond this whole system. We could make Star Trek, Starfleet Academy real. That is an exciting future. That is inspiring. You know, you just, I mean, you need things that move your heart. Right.

35:07Fuck yeah. Fuck yeah. Let's do it. Yeah. I mean, like life can't just be about solving one miserable problem after another. Right. It's got to be things that you'll afford too as well. Yeah, and do you think you might have to move it to a different jurisdiction and to move fast? I've always wondered if like Rocket technology is considered in fast weapons technology, so we can't just go do it, you know in another country Yes, in it you can be not interesting and if we don't do it other countries could do it I mean there are so far behind us, but theoretically there is a national security you know So justification here, if somebody can put their thinking caps on, like, do we want to have this technology that you're building, the teams working so hard on stolen by other countries, and then maybe they don't have as much red tape.

35:55I wish people were trying to steal it. So no one's trying to steal it. It's just too crazy, basically. And that's for you. Yeah, it's way too busy. What do you think is going on that led to Boeing building the Star Line, the way that they did? They were able to get it up. But not complete. But can't complete. They can't finish. Can't finish. And now we're still going to have to go up and finish.

36:35Well, I mean, I think boring as a company that is, they actually do so much business with the government. They have sort of impedance match to the government. So they're like basically one notch away from the government, maybe two. They're not far from the government, from an efficiency standpoint, because they derive so much of the revenue from the government. Now, a lot of people think more SpaceX is super dependent on the government and actually know most of our revenue is commercial.

37:05So, and there's been, I think, at least up until perhaps recently, because then, I have a new CEO who actually shows up in the factory. And the CEO before that, I think, had a degree in accounting and never went to the factory. And didn't know how airplanes flew. So I think if you are in charge of a company that makes airplanes fly and spacecraft go to orbit, then it can't be a total mystery as to how they work.

37:44So I'm like, sure, if somebody is like running Cocoa Pepsi and they're like great at marketing or whatever, that's fine because it's not a sort of technology dependent business. Or if they're running a financial consulting and they're the reason of counting, that makes sense. But I think if you're the cavalry captain, you should know how to ride a horse. Pretty basic. It's good to know. It's just like it's just concerning it. the cavalry captain is full of the horse. You know? He's fucking inspired the team. I'm sorry, I'm scared of horses. It gets on backwards, and I'm like, oh shit. Shifting gears to AI.

38:30Peter was here earlier, and he was talking about how so far the only company to really make money of AI is in video with the chips. Do you have a sense yet of where you think the big applications will be from AI is is it going to be an enabling self -driving, is it going to be enabling robots, is it transforming industries? I mean, it's still, I think, early in terms of where the big business impact is going to be, do you have a sense? Yeah.

39:05I mean, I think the spending on AI probably runs ahead of, I mean, it does run ahead of the revenue right now. There's no question about that. But the rate of improvement of AI is faster than any technology I've ever seen by far. And it's, I mean, like the, for example, a touring test used to be a thing. Now, you know, your basic open source random LLM, you're writing on a frigging Raspberry Pi, probably, we could beat the During Test. So there's, I think actually, the good future of AI is one of immense prosperity where there is an age of abundance, no shortage of goods and services, everyone can have whatever they want, unless except for things we are officially defined to be scarce like some special artwork.

40:15But anything that is of manufacturer good or provided service will, I think, with the admin of AI plus robotics, that the cost of goods and services will trend to zero. I'm not saying it would be actually zero, but it will be everyone will be able to have anything they want. That's the good future. Of course. And my view, that's probably 80 % likely. So look on the bright side. Only 20 % probably of inhalation. It's nothing. Is the 20 % that, like, what does that look like? No, man. Frankly, I do have to engage in some degree of deliberate suspension of disbelief with respect to AI. Do you know what to just sleep well?

41:06And even then, because I think the actual issue, the most likely issue is like, well, how do we find meaning in a world where AI could do everything we can do? A bit better. That is perhaps the bigger challenge.

41:22Although, at this point, I know more and more people who are retired and they seem to enjoy that life. So, but I think that maybe there'll be some crisis of meaning. like because the computer can do everything you can do, but better, maybe that'll be a challenge. But really, you need this sort of end effectors. You need the autonomous cars and you need the sort of humanoid robots or your general purpose robots. But once you have general purpose, humanoid robots and autonomous vehicles, You really, you can build anything. And I think that there's no actual limit to the size of the economy. I mean, there's obviously the massive earth, like that, it'll be a low one limit.

42:19But the economy is really just the average productivity per person times number of people. That's the economy. And if you've got humanoid robots that can do, you know, where there's no real limit on the number of humanoid robots. And they can operate very intelligently, then there's no actual limit to the economy in a, it's no meaningful limit to the economy. You guys just turned on Colossus, which is like the largest private compute cluster, I guess, of GPUs anywhere. So that's the most powerful supercomputer of any kind. which sort of speaks to what David said and kind of what Peter said, which is a lot of a kind of economic value so far, AI has entirely gone to in video.

43:08But there are people with alternatives and you're actually one with an alternative. Now you have a very specific case because Dojo is really about images and large images, huge video. Yeah, the Tesla problem is different from the sort of LLM problem. The nature of the intelligence actually is actually, and what matters in the AI is different to the point you just made, which is that in the Tesla's case, the context is very long. So you've got gigabytes of context. by context windows. Yeah, you've got, you know, sort of. I was just bringing it up. It's kind of billions of tokens of context. Not an idea amount of context, because you've got seven cameras.

43:55And if you've got several, you know, let's say, you've got a minute of several high -def cameras, then that's gigabytes. So you need to compress. So the Tesla problem is you've got to compress a gigantic context into the pixels that actually matter. And, you know, and condense that over a time at, so you've got to, in both, the time dimension, the space dimension, you've got to compress the pixels in space and the pixels over in time. And then, and then have that inference done on a tiny computer, relatively speaking, small, a few hundred watts, it's a Tesla designed AI inference computer. Which is why they're still the best.

44:42There isn't a better thing we could buy from suppliers. So the Tesla designed AI inference computer that's in the cars is better than anything we could buy from any supplier. Just by the way, that's kind of a... By the way... The Tesla AI... AI chip team is extremely good. You guys in the design, there was a technical paper and there was a deck that somebody on your team from Tesla published and it was stunning to me. You designed your own transport control like layer over ethernet. You're like, ah ethernet's not good enough for us. You have this TTC or something and you're like, oh, we're just going to reinvent ethernet and like string these chips.

45:15It's pretty incredible stuff that's happening over there. Yeah. No, the team that the Tesla chip design team is extremely, extremely good. So. But is there a world where, for example, other people over time that need some sort of like video use case or image use case? So theoretically, you'd say, why not? You know, I have some extra cycles over here. So we should kind of make you a competitor of Nvidia. It's not intentionally per se, but. Yeah, I mean. The, you know, there's this training and inference. And we do have those two projects at Tesla. We've got Dojo, which is the training computer.

45:55And then our inference chip, which is in every car, inference computer. So and Dojo, we've only had Dojo one. Dojo two is, you know, should have Dojo two in volume towards the end of next year. And that will be, we think, sort of comparable to sort of a B200 type system, a training system. And, you know, so there's, I guess there's some potential for that to be used as a service. But, like, you know, the Dodger is just kind of like, I mean, we're, I guess I have like some improved confidence in Dojo, but I think we won't really know how good Dojo is until probably version three. Like usually takes three major iterations on a technology to be excellent and we'll only have the second major iteration next year.

47:06The third iteration, I don't know, maybe late, you know, 26 or something like that. I was the autonomous project going, I remember when we talked last. You said it's publicly that it's in doing some light testing inside the factory. It's actually being useful. What's the build of materials and when, you know, for something like that at scale. So when you start making it like you're making the model three now and there's a million of them coming off the factory line. What would they cost? 20, 30, $40 ,000, you think? Yeah, I mean, I've discovered really that, you know, anything made in sufficient volume will asymptotically approach the cost of its materials.

47:48So now there's, I should say there's, so some things are constrained by the cost of intellectual property and like paying for patents and stuff. So a lot of, you know, what's in a chip is like paying royalties and depreciation of the chip fab. So, but the actual module cost the chips is very low. So optimists, it obviously is humanoid robot. It is way as much less than it's much smaller than a car. So you could expect that in high volume, and I'd say that you also probably need three production versions of optimists. So you need to refine the designs at least three major times and then you need to scale production to sort of the million unit plus per year level.

48:39And I think at that point, the cost, the labor and materials on Optimus is probably not much more than $10 ,000. And that's a decade -long journey, maybe? Basically, think of it like, Optimus will cost less then a small car. So at scale volume with a three major iterations of technology, and so if a small car costs $25 ,000, it's probably like $20 ,000 for an optimist, for a humanoid robot that can be your body, like a combination of R2, D2 and C3PO, but better. I mean, I don't say I think people are to get really attached to their humanoid robot because I mean like you're looking sort of what Star Wars is like R2D2 and C3VL, a lot of those guys.

49:33You know they're awesome and their personality and I mean all R2 could do is just be for you. Yeah, that's pretty sweet English. And C3VL to translate the beeps. So you're in year two of that if you did two or three years is a decade -long journey for this to hit some sort of scale. I would say major iterations are less than two years. It's probably on the order of five years, maybe six to get to a million years here. And at that price point, everybody can afford one on planet Earth. It's going to be that one to one, two to one. What do you think ultimately we're sitting here in 30 years? the number of robots on the planet versus humans.

50:23Yeah, I think the number of robots will vastly exceed the number of humans. Vastly, yeah. Vastly exceed. I mean, you have to say, who would not want their robot buddy? Everyone wants a robot buddy. All right. All right. You know, it's like, especially if you can, you know, it can take care of your dog for a walk. It could, you know, mow the lawn. It could watch your kids. It could, you know, like, it could teach your kids. It could, it could, it could. We could also send it to Mars. Yeah, absolutely. We could send a lot of robots to Mars to do the work needed to make a colonized planet for you.

51:01And Mars is already the robot planet. It plays like a whole bunch of, you know, robots, like, rovers and helicopters. Yes, only robots. Yeah. So yeah, the, no, I think the sort of useful humanoid robot But the single biggest opportunity ever.

51:25Because if you assume like, I mean, the ratio of humanoid robots to humans is gonna be at least two to one, maybe three to one. Because everybody will want one and then there'll be a bunch of robots that you don't see that are making goods and services. And you think it's a general one generalized robot that then learns how to do different tasks or. Yeah. Hey, I mean, we are a generalized, we're just made of meat. We're a meat puppet. Yeah, I mean, operating my meat puppet, you know. So yeah, we are actually, it turns out like as we're designing optimists, we're sort of learn more and more about why humans are shaped the way they're shaped.

52:08And why we have five fingers and why your little finger is smaller than the index finger. Obviously, I have opposable thumbs, but also why, for example, the major muscles that operate your hand are actually in your forearm. And your fingers are primarily operated, you look... The muscles that actuate your fingers are located, the vast majority of your fingers trying to actually come in from your forearm, and your fingers are being operated by tendons, little strings. That's, and so the current version of the Optimus Hand has the actuators in the hand and has only 11 degrees of freedom. So it's not as, it doesn't have all the degrees of freedom of human hand, which has, depending on how you count it, a roughly 25 degrees of freedom.

53:08And it's also like a lot stronger enough in certain ways because the actuators have to fit in the hand. So the next generation, Optimus Hand, which we have in prototype form, the actuators have moved to the forum just like a human and operate the fingers through cables just like a human hand. And the next generation had as 22 degrees of freedom which we think is enough to do almost anything that a human can do. And presumably I think it was written that X and Tesla may work together and provide services but my media thought went to all if you just provide a groc to the robot, then the robot has a personality and can process voice and video and images and all of that stuff.

53:58So yeah, that's the wrap here. I think you know everybody talks about all the projects you're working on, but people don't know you have a great sense of humor. That's not true. Oh you do, you do. People don't see it, but I would say one, I know for me the funniest week of my life for one of the funniest was when you did SNL. And we got, and you, I got to tag along. Maybe you saw it. Maybe behind the scenes, like some of your funniest recollections of Matt chaotic insane week when we laughed for 12 hours a day. It was a little terrorizing on the first couple of days, but. Yeah, I was, but worried at the beginning there, because frankly, nothing was funny.

54:43me. J1 was rough. Rough. Yeah, so I mean... Well it's like a rule but can't you guys just say it? Just say the stuff that got on the cutting. The funniest skits were the ones that didn't let you do it. So I'm saying can you just say... There were a couple of funny ones that they didn't let you do it. You can say it so that he doesn't get... I mean how much time do we have to? Well we should just give him one or two because it was... You were in mind which one do we regret most? not getting on air. You really want to hear that? I mean... I mean, it was a little spicy, it was a little funny. Okay.

55:26Here we go. Alright, here we go guys.

55:31Alright, so one of the things that I think everyone's been sort of wondering this whole time Is Saturday Night Live actually live? Like live? Live, live, live? Or do they have like a delay or like just in case, you know, there's a war probe malfunction or something like that? Is it like a, you know, five second delay? What's really going on? But there's a way to test this. Right. We can't run away. There's a way to test this. Which is, we don't tell them what's going on. I walk on and say, this is the script, I'll throw it on the ground. We're gonna find out tonight, right now. If Saturday night is actually live.

56:18And the way that we're going to do this is I'm going to take my cock out.

56:28This is the greatest way ever. And if you see my cock, you know it's true. And if you don't, it's been a lie. A lie. All these years. All these years. Now this is - We're going to bust them right now. And this - We're pitching this. Yeah, yeah. So we're pitching this. on Zoom. Yeah, we're like kind of hungover from the weekend and we're like, pretty good. Yeah, and it's, you know, Jason's on and, and my, you know, since we got like, you know, what my friends, I think, are sort of, you know, quite funny, you know, Jason's quite funny. I think like, like, like, Jason's the closest thing to that exists in the real in the world.

57:19We have a joke going that he's butters and I'm heart -man. Yeah. So, um, and I heard Mike's, we're funny too. So, we come in like just like guns blazing with ideas. And we realize like actually, you know, that's not how it works. And that's normally like actress and they just get told what to do. and like, oh, you mean we can't just like do funny things that we thought of? What? They're watching this. And on the Zoom, they're a guest. Yes, they're like, you want to pitch. Yeah, it's silence. Like, so I'm like, and I'm like, and I'm like, is this thing working? Is this? Exactly, are we muted? Is that my God?

58:01I'm like, we hear you. And then after a long silence, like Mike just says the word crickets. Crickets? And they're not laughing. And then you get a, not even a trouble. I like, what's going on? And then Elon explains the punchline, which is. Exactly. So there's more to it. OK. Yes. That's just the beginning. So Elon says. So then I'm like, so I say, I'm going to reach down. I'm going to reach down. Into my pants. Into my pants. And I just took my hand in my pants. And I'm going to pull my call. And I tell this to the audience. And the audience is going to be like, go, what? Right. And then, and then, and then, and then, and then I pull out a, a baby rooster.

58:53You know, yes. And it's like, okay, this is kind of PG, you know, it's like that, about that bad. It's like, this is my tiny cock. And, and, and it's like, what do you think? It's seven. And do you think it's an ice pack? I mean, I like it. And I pitch. And then Kate McKennan walks out. Yeah, exactly. And I'm like, oh no, but you haven't heard of it. Kate McKennan comes out. Yeah. And she says, Elon, I expected you would have a bigger cock. Yeah. So I don't mean to just why you Kate, but yeah. But I hope you like it anyway. I think. And Kate's going to come out with with her cat. OK. Right. So you can see what this is going and I say nice wow that's a that's a that's a nice pussy you've got there Wow that's amazing It looks a little wet.

59:52Was it raining outside?

59:57Can you do mind if I stroke your pussy is that cool? It's like oh no Elon actually did you can I hold your cock? Of course, Kate, you definitely hold my card. And then we exchange. And I think just the audio version of this is pretty good. Right.

1:00:17And so it's like, wow, I really like stroking your card. And I was like, I really enjoy stroking your pussy. Yes, of course. And yeah, so you know. They're looking at it's like, oh my god, what have we done inviting these lunatics on the program? Yeah, no, no, no, no, said like well, it is it is Mother's Day

1:00:49We might not want to go with this one. My mom's an audience I'm like, well, that's a good point. Fair fair. It might be a bit uncomfortable for all the moms in the audience Maybe I don't know. I don't know. Maybe they'll dig it. Maybe like it. Yeah, that was... That's the... That's the... That's the cold open that didn't make it. We didn't get that on the air. We did fight for Doge. Yes. And we got Doge on the air. There's a bunch of things that I said that were just not on the script. They have these like cue cards for what you're supposed to say and I just didn't say it. I just went off the rails.

1:01:26Yeah. They didn't do that coming. Yeah, it's live. Well, it's live. And so Elon wanted to do Doge. This is the other one. He wanted to do Doge on late night. And he says, hey, Jake, how can you make sure? Oh, yeah. I wanted to do the Doge father. Like you sort of redo that scene from the Godfather. I mean, you kind of need the music to cue things up. Da, da, da, da, da. Da, da, da, da. You bring me on my daughter's wedding. Then this and you ask for dogs. Yeah, you got Marlon Brando's team. That I give you Bitcoin, but you want dogs. And the thing is, you got to set the mood. You got to have a taxi, you know, and sort of drop off in the dog shop.

1:02:10You know, you got to have like, like Marlon Brando. Yeah. And I said, you come to me all this day. I'm like, a dog's wedding. And you ask me for your private keys. Are you even a friend? You'll call me the Dorchfather. So... Bono, Sarah. Bono, Sarah. No. So... That's potential. That had great potential. So they come to me and I'm talking to Colin and Joe, who's got a great sense of humanity. He's amazing. He loves Elon. And he's like, we can't do it because of the law and stuff like that. The law. The law and the applicability. So I said, it's okay, Elon called Comcast. And he put it in an offer and they just accepted it.

1:02:59Just bought NBC. So it's fine. And Colin Jones looks at me. I'm so good. And he's like, you're serious. I'm like, yep, we own NBC now. And he's like, OK, well, that kind of changes things, doesn't it? I'm like, absolutely. Where'd it go on Doge? Yeah. And then he's like, you're fucking with me. I don't know. I'm fucking with you. Or are we? Or are we? It was the greatest week of, and that is like two of 10 stories. Yeah, we've got to, yeah, we've got to find the way. We'll save the other eight. Yeah. But it was, and I was just so happy for you to see you have a great week of just joy and fun and letting go, because you were launching rockets.

1:03:45You're dealing with so much bullshit in your life to have those moments, to share them and just laugh. It was just so great. Yeah. More of those moments. I think we got to get you back on SNL. Who wants to back on SNL one more time? Hi ladies and gentlemen, our bestie Elon Musk. video! Thank you. Thank you.