// transcript — 723 segments
0:00 Introducing Elon Musk, and reflecting on his DOGE experience
0:00 [Music] I believe Optimus is going to be the greatest product ever created by
0:05 humanity. >> Elon Musk and his XAI startup have built the largest and most powerful artificial
0:14 intelligence training supercomputer in the world. As far as I know, there's
0:17 only one person in the world who could do that. You know, >> this is an arms race of epic
0:21 proportions. >> He's a big thinker. You guys went on Fox
0:27 the other day with the Doge team. You saw Elon's face nodding while they were
0:31 speaking with a grin ear to ear. He was proud. >> XAI has acquired X in an old stock
0:36 transaction. >> Tesla's first robo taxis are officially
0:39 on the road. >> The company's board proposed a new compensation package for the CEO that
0:44 could pay him just about a trillion dollars in stock. >> He gets nothing if he doesn't hit the
0:49 numbers. SpaceX will buy wireless spectrum licenses from Echoar for its Starlink
0:58 satellite network for about 17 billion. [Music] splash down.
1:17 How do you have time? this I I I never understand you. >> Yeah. Well, I do work a lot.
1:26 >> Ladies and gentlemen, please welcome >> Elon Musk.
1:31 >> All right. Good. >> All right. Where are you?
1:34 >> Alto. >> You're in Palo Alto and um not Washington DC.
1:41 >> I'm I'm at Tesla Global Engineering Headquarters in Palo Alto.
1:45 >> Yeah. So, no more Washington DC. You're back at work. You're focused. Yeah.
1:50 >> Uh, yeah. I haven't been to DC since May. >> Okay. Uh,
1:54 >> that was a that was a hell of a side quest. >> That was a good Any lessons from your
2:01 time in Washington DC? >> Uh, >> Elon O only. I support David's noble efforts and this
2:17 uh it's good to it's good to have talented people in the administration uh
2:22 but at the end of the day um if you look at our national debt which is uh
2:29 insanely high uh the interest payments exceed the u defense department I guess
2:37 sorry war department uh budget um and um Nikki Bryzy so if AI and
2:44 robots don't solve our national debt, we're we're toast. >> Which is a great segue. Um, Optimus is
2:56 um I think going to be the greatest >> uh product in the history
3:01 of humanity. What's the progress like and how much of your how many of your cycles are going
3:07 specifically to Optimus? What's the timeline? I think you're on version
3:14 three, maybe four. Tell us everything. Uh well, yeah, everything would take a long time.
3:22 >> We've got time. >> Um we're we're finalizing the design of
3:30 Optimus version 3 and uh that that really is going to be a very remarkable robot.
3:38 Um it will have the essentially the manual dexterity of a human. So meaning a very
3:45 complex hand. Um the a an AI mind that can navigate and comprehend reality.
3:55 Um and it will be made in very high volume. Uh those are the three things
3:59 that are missing. Like if you see any other um robotics uh company, they're
4:05 missing those three things. Those are the three really hard things. Um
4:13 and uh I I I spent actually at this point it it might be more of my mental cycles
4:23 than anything anything else any other single thing on Optimus. Uh that's
4:29 that's that that's solving for uh real world AI uh all of the electro mechanical issues
4:37 of Optimus the the supply chain and production challenges of it because we
4:40 have there is no supply chain that exists for humanoid robots. So it has to
4:45 be we have to recreate it from scratch um and which requires doing a lot of
4:49 vertical integration. um none of the actuators in Optimus um are
4:57 available from an existing supply chain. Um so but I I think it is accurate to
5:04 say that if successful Optimus will be the biggest product ever
5:11 >> and the cost of it at scale 2030 $40,000 a robot. What what do you think the
5:18 first wave of them will cost? And yeah, when will we be able to buy one to work
5:26 >> I think that the the marginal cost of production once you hit a million units
5:31 per year uh is probably around the $20,000 range. Uh it it it sort of depends on how much
5:43 you spend on the AI chip in the in the robot. Um and you need to achieve a lot of
5:50 efficiencies in the actuators. Uh there are um 26 actuators per arm like 26
5:58 electric like motors, gearboxes and power electronics. so, so, but but the the the AI chip will
6:09 be pretty expensive like that that might be like55 or $6,000 of the of the bill
6:16 of materials, maybe more. Um, and um but but so I but I think at volume at a
6:23 million units a year, the the production cost is probably on the order of
6:28 $20,000, maybe 25, something like that. And um price will be as a function of
6:32 demand. >> Elon um can you maybe explain to everybody why the hand is so important
6:40 to get right and why you know the actuator design is so unique and you
6:45 know why it's so difficult why nobody makes it and why you have to start there
6:48 almost to build the rest of the the robot properly. Well, it turns out the human hands are
6:56 incredibly they've evolved to this to be this incredibly sophisticated machine.
7:03 Like the your hand is an in actually a remarkable thing. It's look look closely
7:08 at your hands and and think of all the things you can do with your hands, which is a lot.
7:14 >> I can think of many things. >> Yeah, I was just thinking about
7:17 something. >> You know, your hands are very versatile instrument. Yeah, you can give him a
7:23 high five. >> Very versatile. Um, you know, you you you can swing a
7:31 baseball bat, you can thread needles, you you can you put thread in a needle.
7:37 Uh, you can play the piano with violin. Um, you know, you could disassemble or
7:44 assemble a car. The hands are incredibly versatile instruments. Um and um most of
7:50 the muscles of of the hand are are actually in the forearm. So your hand is
7:56 kind of like a like a like it's like a puppet. Like it's mostly a puppet. The
7:59 mus the muscles are coming from the forearm and they're pulling the tendons
8:04 uh which are you know also human tendon designs or human human
8:10 tendon evolution is incredibly good. Um, so you you've got this web of tendons.
8:17 You you you've got um I think I think the the human hand is something like
8:20 depending on how you count it, 27 or 28 degrees of freedom per you know in in
8:29 the hand. It's uh it's amazing. So in order to create a robot that can uh
8:35 be a generalized uh humanoid, you you must solve the hand the hands problem.
8:40 >> Yeah. We had uh we had >> it's got hands, needs hands.
8:44 >> And so is it like uh when you were first building Tesla where the supply chain
8:48 doesn't exist and now you have to go out and find folks to work with and you know
8:52 build all this vertical integration, get support. Is it is it literally like it's
8:56 just nowhere to be found and >> you're going to have to build all of
8:58 this stuff up? >> Yes, we we we could not actually buy the
9:02 actuators for any amount of money. they simply didn't exist. Even though there
9:07 are, I don't know, 10 20,000 electric motors out there of various sizes and
9:13 shapes. Um, we've had to design uh every electric motor, gearbox um and and the
9:17 controlling electronics from scratch basically from physics first principles.
9:22 >> The good news is you've got a lot of experience with factories over the last
9:26 couple of decades. So, >> how challenging is this versus Cybert
9:31 truck model Y >> Model X >> Gigafactory? You know, the Yeah. The
9:37 Fabra Egg known as the Model X. Yeah. >> Right. >> Um
9:43 it's hotter than any any of those things. >> Okay. >> Yeah.
9:48 >> Much hotter significantly. Yeah. Starship. >> Yes. Well, more No, not Starship's
9:55 harder. Okay. >> So, somewhere between a Model X and a
9:59 Starship? >> Yeah. >> Is it is the What's harder, the hardware
10:09 >> Right now, we're struggling with the the final design of the hardware. Like I
10:14 said, it's really primarily the hand. Not to just just dismiss the rest of the
10:18 robot. the rest of it's also uh important but but the hands are the
10:23 hands inclusive of the forearm are a majority of the engineering difficulty
10:27 of the entire robot and then let's assume you get past the hardware
10:30 challenges how much do you sort of get for free um based on all the progress
10:36 that's happening with LLMs will you know will consumers just be able to interact
10:40 with this talk to the robot ask it to do things it'll understand and sort of
10:43 >> Oh yeah >> yeah no problem >> you're spending a lot of time with any I
10:48 noticed online. >> Not not that long. Um maybe I went a
10:52 little over the top from Bunning Grom Imagine, but uh >> well, but in all seriousness, those
11:00 characters and these robots that seems to be, you know, like maybe they
11:04 >> you could get the embodiment of Annie, I suppose. >> Yeah. Why why the human form factor,
11:09 Elon? You could make something that's maybe better than a human or maybe
11:13 simpler than a human to do specific tasks and maybe better than a human to
11:17 do more things than a human can do. How do you decide to make it just like a
11:21 human? >> Well, if you wanted to do all the things that a human can do, it turns out you
11:27 need a humanoid robot. Um, so if you want to just do a subset,
11:33 it that's much easier. Um but uh it turns out humans evolved to this the
11:41 shape and capabilities that we we we have. Um it it it for for good reasons.
11:50 Uh there actually is that there is like there's value to having five, you
11:56 know, four fingers and the thumb. Um and even the pinky actually is is quite
12:02 useful. Um toes are much more aggression walk but but but the fingers
12:08 >> well also humans humans have designed the world as well. So we designed it for
12:11 us. So >> if you can make a humanoid robot it'll be immediately backwards compatible with
12:16 what we've built the world for. >> Precisely. >> Elon there's another there's another
12:20 Tesla: AI5 chips, impact on FSD
12:22 part of um the robot. So there's the LLMs, there's the actuation in the
12:28 hands, but also there's the um the silicon that runs it. And there was, you
12:36 know, Dojo, I think you you posted on X AI5 and AI6, and it just seemed like you
12:40 were incredibly excited about the direction in which the silicon layer was
12:45 also going. Can you tell us about that and what that is and what what what what
12:49 are we what are we building here? What is being built? Is it a complement to
12:53 everything that exists in the world? Is it a potential long-term competitor?
13:02 >> Um, yeah. So, at at Tesla, we basically had two different chip programs. One dojo and
13:10 one uh dojo on the training side and then what we call, you know, AI for it,
13:18 which just our inference chip. um uh that the AI Force is currently shipping
13:22 in all vehicles. Um and we're finaliz finalizing the design of AI5 which will
13:29 be an immense jump from AI4. Um by some metrics the improvement in AI5 will be
13:36 40 times better than AI4. >> Wow. >> So 40% 40 times. Um and and uh this is
13:44 because we work so closely at a very fine grade level on the AI software and
13:50 the AI hardware. So we know exactly where the limiting factors are and and
13:55 um and so effectively the AI hardware and software teams are co-designing the
13:59 chip. Um >> so a 40x improvement in the silicon I think then as it as everybody here in
14:06 the audience experiences it is that just an almost like an order of magnitude
14:11 increase in the quality of FST and the safety that you experience as a Tesla
14:15 driver and then the quality of the robot like where does it all manifest when you
14:20 when you you know bring it up and actually get it into production?
14:25 Yeah to be precise the 40x is on if you say like compared to the worst
14:32 limitation on AI4 which is running the softmax operation. >> Yeah
14:36 >> we currently have to run softmax in around 40 steps in emulation mode
14:41 whereas that'll be just be done in a few steps uh natively in AI5.
14:48 Um AI5 trip will also be u easily handle mixed precision um models. So you don't
14:54 have it'll dynamically handle mixed precision. There's a bunch of sort of
14:58 technical stuff that AI will do a lot better. Um in terms of of nominal sort
15:05 of uh raw compute, it's it's eight times more compute. Um about nine times more
15:12 memory uh roughly five times more memory bandwidth. Um so uh but because we're
15:18 addressing some core limitations in AI4, you multiply that by that that 8x
15:23 computer improvement by another 5x improvement because of of uh
15:27 optimization at a at a at a very fine grain silicon level of things that are
15:31 currently suboptimal in AI4. That's where you get the 40x improvement.
15:36 >> You had um I'll keep going, keep going. Uh so now now that said I I'm I am
15:41 confident that the current uh chips uh AI AI4 chips that are in the cars will
15:48 uh achieve self-driving safety that is at least two to three times that of of
15:55 human and and maybe even 10x. Um and the software that uh will be released for
16:00 that is is coming out over the next few months. So version 14 will be the
16:06 biggest uh upgrade in Tesla software since version 12. Um we are increasing
16:12 the uh parameter count by an order of magnitude. Um the there's there's
16:19 there's a lot of uh reinforcement we we there there there's like you can
16:28 think of AI sort of as a way of compressing reality and and and some of
16:35 those compression steps uh we uh were too lossy and and we addressed the
16:40 lossiness in the compression steps. Um so the these are all software updates
16:43 that'll that'll go out. So just over there updates um your car is going to
16:48 feel like it is sentient by the end of the year. >> Yeah, it feels that way already to be
16:50 SpaceX: Vision for Starlink-enabled smartphones, $17B spectrum deal, Starship update
16:55 honest. Um I saw in the trades that you spent about $17 billion on some spectrum
17:00 and that um >> yeah um so some couch change um to enable your satellites and the Starlink
17:11 network to connect directly with phones. What will that look like in a year or
17:15 two? Are we going to drop our Verizon account and just expand our Starlink
17:20 account? >> Uh, thank you. >> We're kind of hoping cuz Verizon kind of
17:25 sucks. >> How How many of you want a Starlink phone? >> Who wants a Starlink phone?
17:33 >> Is it Is it technically possible? >> I know you can't see it, but it's
17:35 everyone. >> Yeah, All right, cool. Um so this is a kind of a long-term thing. Uh it it will
17:45 allow SpaceX to h uh deliver high bandwidth connectivity directly from the
17:57 satellites to the phones. Um but uh there are hardware changes that need to
18:01 happen in the phone. So the since these frequencies are not supported in current
18:05 phones uh that the chipset has to be modified to add these frequencies um and
18:11 that probably is a 2-year time frame. So the phones that um are able to use the
18:16 spectrum that was acquired probably start shipping in around 2 years. Um and
18:22 um and then we also need to build the satellites that are going to communicate
18:26 on those frequencies. So, in parallel, we're building the satellites and
18:30 working with the handset makers to add these frequencies to the phones. Um, and
18:36 then the the satellites and the phones will then handshake very well to achieve
18:39 high bandwidth connectivity. But the net effect is that you should be able to
18:44 watch uh videos uh anywhere on your phone. >> Wow. >> And it's going to be crazy. And what and
18:51 do these do these frequencies would they work indoors inside buildings, you know,
18:55 like like your phone currently does? Okay. >> And so will you be able to have
18:58 basically like >> if you if if you're in a building with a
19:02 with a like a a thick metal roof then No. But um >> no the the same types of of
19:07 >> Yeah. Normal normal homes. Yes. >> Yes. >> Elon is your vision for this that
19:11 instead of you know having an AT&T account or and then roaming when you're
19:16 in the UK or you're in India. It's just we could have one direct deal with
19:20 Starlink. It works all over the world eventually. Not today but at some point.
19:25 Is that the end goal? That basically we don't need a regional carrier. We have a
19:30 global carrier and that would be you. >> Uh that that would be one of the
19:33 options. To be clear, we're not going to put the other carriers out of business.
19:35 They're still going to be around cuz they they own a lot of Spectrum. So, uh
19:42 there's uh but but yes, you you should be able to have a Starlink uh like you
19:46 have like you have an AT&T or T-Mobile or Verizon or whatever, you should be
19:50 you could have a you know account with Starlink that uh works with your you
19:58 know Starlink uh antenna at home uh free Wi-Fi as well as on your phone and um
20:03 yeah it would be a comprehensive solution for high bandwidth at home and
20:08 for high bandwidth direct to sell. >> Could you buy some carriers to have more
20:12 >> spectrum? >> Maybe you could buy Verizon. >> Not out of the question. I suppose it
20:17 that may happen. >> Let's talk about um let's talk about
20:22 Starship. You just had a really what appeared to be a phenomenal um launch. H
20:31 how close is it to, you know, being predictable and ready to go in a
20:36 commercial setting? I I I think we'll recover the ship next year. Um we've got one more launch of
20:48 the um Starlink version two uh uh stack that there's only one one
20:54 uh booster and ship left that's in the version two uh design. Uh and then
20:59 thereafter it's it's version three which is a gigantic upgrade cuz that's got
21:04 Raptor 3. Um, and pretty much everything changes on the rocket with version 3.
21:09 Um, so version 3, you know, might have some initial teething pains, uh, cuz
21:15 it's such a radical redesign. Uh, but, uh, it's it's capable of over
21:23 100 tons to orbit fully reusable. Um, and I think it's I think I think um
21:27 unless we have unless we have some very major setbacks, uh, SpaceX will
21:33 demonstrate uh, full reusability next year uh, catching both the booster and
21:39 the ship um, and being able to deliver over 100 tons to a useful orbit.
21:44 >> What does the best rocket in the world do now in terms of tonnage to space?
21:52 Uh well in terms of uh sort of commercial rockets there's there's Falcon Heavy.
21:56 >> Yeah. >> Uh which will do uh in um with with side booster reuse uh will
22:07 do about 40 tons. >> So this is five times bigger. Yeah.
22:13 >> Well two and a half times bigger in but but Starship would be full reuse full
22:16 reusability. >> Got it. Okay. So everything comes back
22:21 >> Elon after after the explosion that happened um with the the the the failed
22:25 launch >> um there was a lot of >> sorry >> which which failed
22:31 >> oh the more recent one the more recent the starship with
22:34 >> the big boom yeah >> the big boom on the base and and and
22:37 there was a lot of >> there was a lot of proclamations that
22:40 there's going to be environmental and FAA and all these other sorts the
22:46 recovery back to the launchpad again was incredible. fast. How did you get back so fast? Not
22:52 just technically and work-wise, but just like regulatory clearance-wise because
22:56 they said there were going to be all these questions and reviews and so on.
23:01 How how did you guys manage that? >> Uh well, there were a lot of questions
23:04 and reviews. We got through them all. Um and credit to the SpaceX team. They
23:09 worked incredibly hard and they uh got the next trip and booster tested and on
23:16 the pad and and flown and um yeah, huge credit to the SpaceX team. Very proud of
23:19 them for >> doing doing such a job, a great job recovering.
23:24 >> Um I mean creating a fully reusable orbital rocket is one of the hottest engineering
23:31 problems ever. and certainly, you know, a candidate for most difficult engineering project ever.
23:40 You know, it's on the podium at least. Um, so it's a that that's been the goal of
23:48 SpaceX from the beginning from 2002. Um, and here we are 23 years later. So, it's
23:52 it's a long journey and um with with a a super talent like by far
23:58 the I think the most talented group of rocket engineers that ever been
24:05 assembled. Um and uh and we're finally next year I think we'll be able to
24:10 achieve full reusability. >> Elon, what are the big um technical
24:14 blockers that you're focused on there between now and that full reusability?
24:19 Are there some showstoppers where you're just kind of literally just obsessing
24:24 over trying to figure out still or is it more about getting through a laundry
24:28 list of your learnings and just >> Well, that the for for full reusability of the ship,
24:43 there's still a lot of work that remains on the heat shield. So, no one's ever
24:47 made a fully reusable orbital heat shield. Like the shuttle heat shield uh
24:51 had to go through nine months of repair after every flight, >> right?
24:57 >> Um so, no one has ever made a fully reusable orbital heat shield.
25:01 >> And is that a material science problem or is that an engineering problem or
25:03 both? >> Uh yeah, I mean it's a material science engineering problem. So, it's but we
25:12 really are uh looking at the fundamental physics here. Um again physics first
25:17 principles and trying to figure out how do we make something that
25:23 um is uh you know can can withstand the heat is very light doesn't transmit the heat to
25:31 the the primary sh >> Yeah. >> primary structure um and uh
25:41 >> Yeah. >> Um Uh, and then as you ascend, if you hit
25:48 some rain, you know, the tiles don't dissolve in rain. There's there's a lot
25:54 of different issues and and then you really need to know that these tiles are
25:59 working. You can't uh, you know, go through this laborious inspection. So, it really needs to be
26:05 we're, you know, these these tens of thousands of tiles all work and don't
26:12 need to be refurbished or checked one by one as was the case with the shuttle.
26:16 xAI: Next-gen Grok models, Colossus 2, scaling laws, “Grokipedia”
2:47 Optimus: Progress and potential, the “hands problem”
2:56 um I think going to be the greatest >> uh product in the history
3:01 of humanity. What's the progress like and how much of your how many of your cycles are going
3:07 specifically to Optimus? What's the timeline? I think you're on version
3:14 three, maybe four. Tell us everything. Uh well, yeah, everything would take a long time.
3:22 >> We've got time. >> Um we're we're finalizing the design of
3:30 Optimus version 3 and uh that that really is going to be a very remarkable robot.
3:38 Um it will have the essentially the manual dexterity of a human. So meaning a very
3:45 complex hand. Um the a an AI mind that can navigate and comprehend reality.
3:55 Um and it will be made in very high volume. Uh those are the three things
3:59 that are missing. Like if you see any other um robotics uh company, they're
4:05 missing those three things. Those are the three really hard things. Um
4:13 and uh I I I spent actually at this point it it might be more of my mental cycles
4:23 than anything anything else any other single thing on Optimus. Uh that's
4:29 that's that that's solving for uh real world AI uh all of the electro mechanical issues
4:37 of Optimus the the supply chain and production challenges of it because we
4:40 have there is no supply chain that exists for humanoid robots. So it has to
4:45 be we have to recreate it from scratch um and which requires doing a lot of
4:49 vertical integration. um none of the actuators in Optimus um are
4:57 available from an existing supply chain. Um so but I I think it is accurate to
5:04 say that if successful Optimus will be the biggest product ever
5:11 >> and the cost of it at scale 2030 $40,000 a robot. What what do you think the
5:18 first wave of them will cost? And yeah, when will we be able to buy one to work
5:26 >> I think that the the marginal cost of production once you hit a million units
5:31 per year uh is probably around the $20,000 range. Uh it it it sort of depends on how much
5:43 you spend on the AI chip in the in the robot. Um and you need to achieve a lot of
5:50 efficiencies in the actuators. Uh there are um 26 actuators per arm like 26
5:58 electric like motors, gearboxes and power electronics. so, so, but but the the the AI chip will
6:09 be pretty expensive like that that might be like55 or $6,000 of the of the bill
6:16 of materials, maybe more. Um, and um but but so I but I think at volume at a
6:23 million units a year, the the production cost is probably on the order of
6:28 $20,000, maybe 25, something like that. And um price will be as a function of
6:32 demand. >> Elon um can you maybe explain to everybody why the hand is so important
6:40 to get right and why you know the actuator design is so unique and you
6:45 know why it's so difficult why nobody makes it and why you have to start there
6:48 almost to build the rest of the the robot properly. Well, it turns out the human hands are
6:56 incredibly they've evolved to this to be this incredibly sophisticated machine.
7:03 Like the your hand is an in actually a remarkable thing. It's look look closely
7:08 at your hands and and think of all the things you can do with your hands, which is a lot.
7:14 >> I can think of many things. >> Yeah, I was just thinking about
7:17 something. >> You know, your hands are very versatile instrument. Yeah, you can give him a
7:23 high five. >> Very versatile. Um, you know, you you you can swing a
7:31 baseball bat, you can thread needles, you you can you put thread in a needle.
7:37 Uh, you can play the piano with violin. Um, you know, you could disassemble or
7:44 assemble a car. The hands are incredibly versatile instruments. Um and um most of
7:50 the muscles of of the hand are are actually in the forearm. So your hand is
7:56 kind of like a like a like it's like a puppet. Like it's mostly a puppet. The
7:59 mus the muscles are coming from the forearm and they're pulling the tendons
8:04 uh which are you know also human tendon designs or human human
8:10 tendon evolution is incredibly good. Um, so you you've got this web of tendons.
8:17 You you you've got um I think I think the the human hand is something like
8:20 depending on how you count it, 27 or 28 degrees of freedom per you know in in
8:29 the hand. It's uh it's amazing. So in order to create a robot that can uh
8:35 be a generalized uh humanoid, you you must solve the hand the hands problem.
8:40 >> Yeah. We had uh we had >> it's got hands, needs hands.
8:44 >> And so is it like uh when you were first building Tesla where the supply chain
8:48 doesn't exist and now you have to go out and find folks to work with and you know
8:52 build all this vertical integration, get support. Is it is it literally like it's
8:56 just nowhere to be found and >> you're going to have to build all of
8:58 this stuff up? >> Yes, we we we could not actually buy the
9:02 actuators for any amount of money. they simply didn't exist. Even though there
9:07 are, I don't know, 10 20,000 electric motors out there of various sizes and
9:13 shapes. Um, we've had to design uh every electric motor, gearbox um and and the
9:17 controlling electronics from scratch basically from physics first principles.
9:22 >> The good news is you've got a lot of experience with factories over the last
9:26 couple of decades. So, >> how challenging is this versus Cybert
9:31 truck model Y >> Model X >> Gigafactory? You know, the Yeah. The
9:37 Fabra Egg known as the Model X. Yeah. >> Right. >> Um
9:43 it's hotter than any any of those things. >> Okay. >> Yeah.
9:48 >> Much hotter significantly. Yeah. Starship. >> Yes. Well, more No, not Starship's
9:55 harder. Okay. >> So, somewhere between a Model X and a
9:59 Starship? >> Yeah. >> Is it is the What's harder, the hardware
10:09 >> Right now, we're struggling with the the final design of the hardware. Like I
10:14 said, it's really primarily the hand. Not to just just dismiss the rest of the
10:18 robot. the rest of it's also uh important but but the hands are the
10:23 hands inclusive of the forearm are a majority of the engineering difficulty
10:27 of the entire robot and then let's assume you get past the hardware
10:30 challenges how much do you sort of get for free um based on all the progress
10:36 that's happening with LLMs will you know will consumers just be able to interact
10:40 with this talk to the robot ask it to do things it'll understand and sort of
10:43 >> Oh yeah >> yeah no problem >> you're spending a lot of time with any I
10:48 noticed online. >> Not not that long. Um maybe I went a
10:52 little over the top from Bunning Grom Imagine, but uh >> well, but in all seriousness, those
11:00 characters and these robots that seems to be, you know, like maybe they
11:04 >> you could get the embodiment of Annie, I suppose. >> Yeah. Why why the human form factor,
11:09 Elon? You could make something that's maybe better than a human or maybe
11:13 simpler than a human to do specific tasks and maybe better than a human to
11:17 do more things than a human can do. How do you decide to make it just like a
11:21 human? >> Well, if you wanted to do all the things that a human can do, it turns out you
11:27 need a humanoid robot. Um, so if you want to just do a subset,
11:33 it that's much easier. Um but uh it turns out humans evolved to this the
11:41 shape and capabilities that we we we have. Um it it it for for good reasons.
11:50 Uh there actually is that there is like there's value to having five, you
11:56 know, four fingers and the thumb. Um and even the pinky actually is is quite
12:02 useful. Um toes are much more aggression walk but but but the fingers
12:08 >> well also humans humans have designed the world as well. So we designed it for
12:11 us. So >> if you can make a humanoid robot it'll be immediately backwards compatible with
12:16 what we've built the world for. >> Precisely. >> Elon there's another there's another
12:22 part of um the robot. So there's the LLMs, there's the actuation in the
12:28 hands, but also there's the um the silicon that runs it. And there was, you
12:36 know, Dojo, I think you you posted on X AI5 and AI6, and it just seemed like you
12:40 were incredibly excited about the direction in which the silicon layer was
12:45 also going. Can you tell us about that and what that is and what what what what
12:49 are we what are we building here? What is being built? Is it a complement to
12:53 everything that exists in the world? Is it a potential long-term competitor?
13:02 >> Um, yeah. So, at at Tesla, we basically had two different chip programs. One dojo and
13:10 one uh dojo on the training side and then what we call, you know, AI for it,
13:18 which just our inference chip. um uh that the AI Force is currently shipping
13:22 in all vehicles. Um and we're finaliz finalizing the design of AI5 which will
13:29 be an immense jump from AI4. Um by some metrics the improvement in AI5 will be
13:36 40 times better than AI4. >> Wow. >> So 40% 40 times. Um and and uh this is
13:44 because we work so closely at a very fine grade level on the AI software and
13:50 the AI hardware. So we know exactly where the limiting factors are and and
13:55 um and so effectively the AI hardware and software teams are co-designing the
13:59 chip. Um >> so a 40x improvement in the silicon I think then as it as everybody here in
14:06 the audience experiences it is that just an almost like an order of magnitude
14:11 increase in the quality of FST and the safety that you experience as a Tesla
14:15 driver and then the quality of the robot like where does it all manifest when you
14:20 when you you know bring it up and actually get it into production?
14:25 Yeah to be precise the 40x is on if you say like compared to the worst
14:32 limitation on AI4 which is running the softmax operation. >> Yeah
14:36 >> we currently have to run softmax in around 40 steps in emulation mode
14:41 whereas that'll be just be done in a few steps uh natively in AI5.
14:48 Um AI5 trip will also be u easily handle mixed precision um models. So you don't
14:54 have it'll dynamically handle mixed precision. There's a bunch of sort of
14:58 technical stuff that AI will do a lot better. Um in terms of of nominal sort
15:05 of uh raw compute, it's it's eight times more compute. Um about nine times more
15:12 memory uh roughly five times more memory bandwidth. Um so uh but because we're
15:18 addressing some core limitations in AI4, you multiply that by that that 8x
15:23 computer improvement by another 5x improvement because of of uh
15:27 optimization at a at a at a very fine grain silicon level of things that are
15:31 currently suboptimal in AI4. That's where you get the 40x improvement.
15:36 >> You had um I'll keep going, keep going. Uh so now now that said I I'm I am
15:41 confident that the current uh chips uh AI AI4 chips that are in the cars will
15:48 uh achieve self-driving safety that is at least two to three times that of of
15:55 human and and maybe even 10x. Um and the software that uh will be released for
16:00 that is is coming out over the next few months. So version 14 will be the
16:06 biggest uh upgrade in Tesla software since version 12. Um we are increasing
16:12 the uh parameter count by an order of magnitude. Um the there's there's
16:19 there's a lot of uh reinforcement we we there there there's like you can
16:28 think of AI sort of as a way of compressing reality and and and some of
16:35 those compression steps uh we uh were too lossy and and we addressed the
16:40 lossiness in the compression steps. Um so the these are all software updates
16:43 that'll that'll go out. So just over there updates um your car is going to
16:48 feel like it is sentient by the end of the year. >> Yeah, it feels that way already to be
16:55 honest. Um I saw in the trades that you spent about $17 billion on some spectrum
17:00 and that um >> yeah um so some couch change um to enable your satellites and the Starlink
17:11 network to connect directly with phones. What will that look like in a year or
17:15 two? Are we going to drop our Verizon account and just expand our Starlink
17:20 account? >> Uh, thank you. >> We're kind of hoping cuz Verizon kind of
17:25 sucks. >> How How many of you want a Starlink phone? >> Who wants a Starlink phone?
17:33 >> Is it Is it technically possible? >> I know you can't see it, but it's
17:35 everyone. >> Yeah, All right, cool. Um so this is a kind of a long-term thing. Uh it it will
17:45 allow SpaceX to h uh deliver high bandwidth connectivity directly from the
17:57 satellites to the phones. Um but uh there are hardware changes that need to
18:01 happen in the phone. So the since these frequencies are not supported in current
18:05 phones uh that the chipset has to be modified to add these frequencies um and
18:11 that probably is a 2-year time frame. So the phones that um are able to use the
18:16 spectrum that was acquired probably start shipping in around 2 years. Um and
18:22 um and then we also need to build the satellites that are going to communicate
18:26 on those frequencies. So, in parallel, we're building the satellites and
18:30 working with the handset makers to add these frequencies to the phones. Um, and
18:36 then the the satellites and the phones will then handshake very well to achieve
18:39 high bandwidth connectivity. But the net effect is that you should be able to
18:44 watch uh videos uh anywhere on your phone. >> Wow. >> And it's going to be crazy. And what and
18:51 do these do these frequencies would they work indoors inside buildings, you know,
18:55 like like your phone currently does? Okay. >> And so will you be able to have
18:58 basically like >> if you if if you're in a building with a
19:02 with a like a a thick metal roof then No. But um >> no the the same types of of
19:07 >> Yeah. Normal normal homes. Yes. >> Yes. >> Elon is your vision for this that
19:11 instead of you know having an AT&T account or and then roaming when you're
19:16 in the UK or you're in India. It's just we could have one direct deal with
19:20 Starlink. It works all over the world eventually. Not today but at some point.
19:25 Is that the end goal? That basically we don't need a regional carrier. We have a
19:30 global carrier and that would be you. >> Uh that that would be one of the
19:33 options. To be clear, we're not going to put the other carriers out of business.
19:35 They're still going to be around cuz they they own a lot of Spectrum. So, uh
19:42 there's uh but but yes, you you should be able to have a Starlink uh like you
19:46 have like you have an AT&T or T-Mobile or Verizon or whatever, you should be
19:50 you could have a you know account with Starlink that uh works with your you
19:58 know Starlink uh antenna at home uh free Wi-Fi as well as on your phone and um
20:03 yeah it would be a comprehensive solution for high bandwidth at home and
20:08 for high bandwidth direct to sell. >> Could you buy some carriers to have more
20:12 >> spectrum? >> Maybe you could buy Verizon. >> Not out of the question. I suppose it
20:17 that may happen. >> Let's talk about um let's talk about
20:22 Starship. You just had a really what appeared to be a phenomenal um launch. H
20:31 how close is it to, you know, being predictable and ready to go in a
20:36 commercial setting? I I I think we'll recover the ship next year. Um we've got one more launch of
20:48 the um Starlink version two uh uh stack that there's only one one
20:54 uh booster and ship left that's in the version two uh design. Uh and then
20:59 thereafter it's it's version three which is a gigantic upgrade cuz that's got
21:04 Raptor 3. Um, and pretty much everything changes on the rocket with version 3.
21:09 Um, so version 3, you know, might have some initial teething pains, uh, cuz
21:15 it's such a radical redesign. Uh, but, uh, it's it's capable of over
21:23 100 tons to orbit fully reusable. Um, and I think it's I think I think um
21:27 unless we have unless we have some very major setbacks, uh, SpaceX will
21:33 demonstrate uh, full reusability next year uh, catching both the booster and
21:39 the ship um, and being able to deliver over 100 tons to a useful orbit.
21:44 >> What does the best rocket in the world do now in terms of tonnage to space?
21:52 Uh well in terms of uh sort of commercial rockets there's there's Falcon Heavy.
21:56 >> Yeah. >> Uh which will do uh in um with with side booster reuse uh will
22:07 do about 40 tons. >> So this is five times bigger. Yeah.
22:13 >> Well two and a half times bigger in but but Starship would be full reuse full
22:16 reusability. >> Got it. Okay. So everything comes back
22:21 >> Elon after after the explosion that happened um with the the the the failed
22:25 launch >> um there was a lot of >> sorry >> which which failed
22:31 >> oh the more recent one the more recent the starship with
22:34 >> the big boom yeah >> the big boom on the base and and and
22:37 there was a lot of >> there was a lot of proclamations that
22:40 there's going to be environmental and FAA and all these other sorts the
22:46 recovery back to the launchpad again was incredible. fast. How did you get back so fast? Not
22:52 just technically and work-wise, but just like regulatory clearance-wise because
22:56 they said there were going to be all these questions and reviews and so on.
23:01 How how did you guys manage that? >> Uh well, there were a lot of questions
23:04 and reviews. We got through them all. Um and credit to the SpaceX team. They
23:09 worked incredibly hard and they uh got the next trip and booster tested and on
23:16 the pad and and flown and um yeah, huge credit to the SpaceX team. Very proud of
23:19 them for >> doing doing such a job, a great job recovering.
23:24 >> Um I mean creating a fully reusable orbital rocket is one of the hottest engineering
23:31 problems ever. and certainly, you know, a candidate for most difficult engineering project ever.
23:40 You know, it's on the podium at least. Um, so it's a that that's been the goal of
23:48 SpaceX from the beginning from 2002. Um, and here we are 23 years later. So, it's
23:52 it's a long journey and um with with a a super talent like by far
23:58 the I think the most talented group of rocket engineers that ever been
24:05 assembled. Um and uh and we're finally next year I think we'll be able to
24:10 achieve full reusability. >> Elon, what are the big um technical
24:14 blockers that you're focused on there between now and that full reusability?
24:19 Are there some showstoppers where you're just kind of literally just obsessing
24:24 over trying to figure out still or is it more about getting through a laundry
24:28 list of your learnings and just >> Well, that the for for full reusability of the ship,
24:43 there's still a lot of work that remains on the heat shield. So, no one's ever
24:47 made a fully reusable orbital heat shield. Like the shuttle heat shield uh
24:51 had to go through nine months of repair after every flight, >> right?
24:57 >> Um so, no one has ever made a fully reusable orbital heat shield.
25:01 >> And is that a material science problem or is that an engineering problem or
25:03 both? >> Uh yeah, I mean it's a material science engineering problem. So, it's but we
25:12 really are uh looking at the fundamental physics here. Um again physics first
25:17 principles and trying to figure out how do we make something that
25:23 um is uh you know can can withstand the heat is very light doesn't transmit the heat to
25:31 the the primary sh >> Yeah. >> primary structure um and uh
25:41 >> Yeah. >> Um Uh, and then as you ascend, if you hit
25:48 some rain, you know, the tiles don't dissolve in rain. There's there's a lot
25:54 of different issues and and then you really need to know that these tiles are
25:59 working. You can't uh, you know, go through this laborious inspection. So, it really needs to be
26:05 we're, you know, these these tens of thousands of tiles all work and don't
26:12 need to be refurbished or checked one by one as was the case with the shuttle.
26:18 >> Can we maybe um switch now? It's I mean, who who else were you talked about
26:22 Tesla, then you go to SpaceX? Yeah. Now, I' I'd like to ask you some questions
26:28 about Grock and um XAI. Um you want to just give us an update? I think you you
26:31 kind of talked about where the nextG model is and you said something
26:35 incredible. I still don't think people really understand it which is you know
26:38 there's going to be a next training run where you expect you know not to start
26:45 from the you know common web and common crawl where you expected an enormous
26:50 amount of synthetic data. Just tell us about how uh the evolution of Grock is
26:55 going and this innovation and why it's Yeah. So we're we're running a lot of
27:04 using a lot of of inference compute and um and reasoning to look at all of the
27:10 source data which is really the corpus of human knowledge and then
27:15 uh thinking about each piece of information and then adding mod adding
27:22 what's missing um and correcting correcting mistakes and removing
27:27 falsehoods from the from that training data. So it's it's it's like if you take
27:31 say Wikipedia as an example but this really applies to to books, PDFs,
27:37 uh the websites, uh every form of information. Um the the Grock is using um heavy amounts of
27:49 inference compute to say to look at at an example a Wikipedia page and say uh
27:56 what is true, partially true or false or missing uh in this page. Now rewrite the page to
28:06 in to correct the remove the falsehoods uh uh correct the half-truths and add the
28:13 missing context. >> Well, Elon, by the way, could you just
28:16 publish that? Could we create like a groipedia? I mean, that would
28:19 >> Yeah, especially for our bio pages, which are a disaster.
28:24 >> Wikipedia is so biased and it's it's a constant war. you know, if something
28:28 gets corrected, five minutes later, there'll be an army of people trying to
28:32 >> I mean, it's become hyperartisan and there's activists all over it.
28:37 >> So, if you do fix, for example, Wikipedia as a source of truth,
28:41 >> it'd be great to publish that just so the world has it.
28:47 >> All right, I'll talk talk about that. So, talk to the team about that like
28:51 Groedia or whatever. This here's the Groedia version. >> It' be interesting. Yeah. and then just
28:56 have it out there for just a few minutes >> where in terms of um people here like it
29:03 um in terms of training Gro 5 um you're you're scaling up your supercluster in
29:08 Colossus in in Memphis >> can yeah have a second one >> yeah can could you give us an update on
29:15 that and then also as part of that um where are we in the scaling laws um if
29:21 you scale a bigger cluster do you get a more powerful AI model is there a point
29:25 of dimin diminishing returns or like how much more compute if you throw twice as
29:30 much compute at it do you get a 10% better model do you get 100% better
29:35 model like is it log linear what what I guess how much more juice is there left
29:42 in scaling hardware do you think >> I think I think there's a natural
29:46 logarithmic function associated with the amount of compute so
29:51 uh then like say for argument sake like 10x more compute will double the
29:55 intelligence. Maybe that's that that might be a rough rule of thumb, but you know, that still
30:01 means that, you know, you go from 100 IQ to 200 IQ. Still pretty pretty big deal.
30:06 Um, so I and and I think I think we'll see intelligence continue to scale all the
30:16 way up to where, you know, most of the power of the sun is harnessed for
30:20 compute and then ultimately most of the power of the galaxy, you know, sort of
30:28 cautev 2, cautev 3 scale uh compute. Um so I guess once you think of artificial
30:34 intelligence not as sort of this you know a destination that you reach
30:38 but really uh as part of the overall escalation of intelligence
30:44 um that that that we are are aware of. Um you know human intelligence has also
30:50 scaled as you've have as the population has increased um and we've been able to
30:56 store more and more information. uh human intelligence has scaled. Now human
31:01 because of population declines and low growth rate, human intelligence is is
31:05 somewhat plateauing um and will actually decline. And my guess is that
31:14 I I I I think that we might have AI smarter than any single human at
31:19 anything as soon as next year. >> Wow. >> Um >> Yeah. and and and then and then probably
31:27 within five like say 2030 probably AI is smarter than the sum of all humans.
31:29 Evolving alongside AI, implosion of the West, the religion vacuum
31:31 >> Do you think do you think humans are on the decline because the AI is evolving?
31:35 Do you think there's this evolution of the ecosystem on Earth that's underway
31:40 that we don't really understand the >> maybe yeah, maybe we implicitly
32:00 >> yeah. >> I I I I mean I hope the birth rates turn
32:04 around. I'm a I'm a big proponent of >> Well, you doing anything about it or no?
32:15 >> Yeah. I'm trying to set a good example. You know, we had a big conversation at
32:19 this conference we didn't expect, which is suicidal empathy, the West,
32:27 >> this um declining birth rate. Uh I noticed you've been pretty active about
32:29 it >> and open borders >> and open borders is like let the
32:34 invaders in. Could all three of those be the same thing? It >> it seems like there's a number of
32:39 symptoms of the West being suicidal. The most obvious one being the birth rate is
32:43 not a replacement level. So obviously if that continues indefinitely then the
32:48 west will literally not reproduce enough to replace itself. But there's other
32:51 things too. There's the fact that the borders were totally opened to the point
32:56 where western culture the social fabric start to come apart and you see this
33:01 especially in Europe where there um you know the indigenous cultures of the UK
33:07 or France or Germany are starting to um potentially be taken over by by cultures
33:10 of people who are brought in and aren't assimilating. You have crime where, you
33:15 know, we have this case on social media right now, this young woman, Ire Ina,
33:19 who's just >> killed in a senseless way on a subway. >> Uh, which is horrific enough in and of
33:26 itself, but then in addition to that, the elite media just for whatever reason
33:29 just refused to cover it, like it didn't exist. >> Um, so you have this issue of crime
33:34 that's not being addressed or even acknowledged >> and no acknowledgement of this. like
33:39 it's almost like we're trying to deny the reality of the spiral
33:45 >> and this Yeah. So you have the you have all these data points um that seem to
33:52 suggest that um the west uh is suicidal or doesn't you know doesn't seem to want
33:57 to defend itself or propagate itself. Um look I think everyone in this room
34:01 thinks that um life is awesome right? I mean it's >> pretty great and I think
34:04 >> worth living. >> Yeah. And when when Alex Karp was here
34:09 earlier today defending the West, that got some of the loudest applause at the
34:13 conference. So, uh I guess we probably don't really understand what's going on.
34:15 We don't really >> Yeah. What's your take, Elon? Cuz you
34:17 you know, >> what's your take on the suicide of the West?
34:21 >> Yeah. >> What's What's >> I'm very worried about it.
34:24 >> Yeah. >> I'm very worried about it. Um you know,
34:29 I think there's there's just the actions of the West are indistinguishable from suicide.
34:37 So, but it's and look, at least in America, there's there's there's generally a sense of
34:45 optimism, but when's the last time you you talked to someone from Europe who
34:50 lives in Europe who's optimistic? >> Not for a while. Yeah.
34:52 >> Decades, >> like even one. >> It's rare. So I I think unless people
35:00 have a sense of optimism and purpose about the future, they suicide might be just what happens. Um
35:10 like like like having a child is an act of optimism about the future. So uh if
35:23 So, so I think we need to maybe give people a sense of optimism and
35:28 excitement about the future and and a belief that the future will be better
35:32 than the past um and they'll be more interested in having kids.
35:35 >> Did did religion play a role in the past, Elon, to kind of plate and make
35:40 folks feel that way >> when they >> Yeah, I think so. uh
35:47 the nature abhores a vacuum and if you take away religion then I think you
35:52 actually you you you get something in its place which is actually worse than
35:57 what was there before I mean it's like destructive basically you get you get
36:01 like the white work mind virus filling filling the hole that religion used to
36:08 have taking the place of of of religion you get these dystopian de facto
36:12 religions um that uh that that are very very self-destructive.
36:20 Um so I I think perhaps some some sort of re revival of religion
36:28 or at least what we need is is um some coherent philosophy that people can get excited about. Um
36:35 you know I mean for me it's a philosophy of curiosity. I'm curious about the
36:39 nature of the universe and I want to go out there and I want humanity to be out
36:45 there exploring the stars. Um maybe meeting alien civilizations. Uh maybe in
36:51 some cases we we see the ruins of a long dead alien civilization but they were
36:55 they were very strong for 10 million years. Um you know the kind of stuff
36:59 that you see in Star Trek in in a non-dystopian sci-fi book or or movie or
37:05 show. Um, and so I'm just I have I have a philosophy of curiosity of of like I
37:09 just want to know what's going on. And and in order to know what's going on, we
37:15 we must have u an an increase in the in the scope and scale of consciousness, we
37:21 must we must expand uh consciousness. We must grow. We must grow humanity and we
37:27 must extend humanity in order to comprehend the and to to understand the
37:32 universe or even what what question should we should ask about the answer
37:36 Understanding the universe, going to the Moon, what happens on Mars?
37:37 that is the universe. Um you know Doug Douglas Adams book the hitchhikers guide to the galaxy
37:44 is actually a a deep book on philosophy disguised as humor. Um and what the
37:48 point he was trying to make in that book was that u the questions are the really
37:54 the hard part. The answer is the universe. Like the answer is everything
37:58 you see around you. But but but one of the questions that we don't know to ask.
38:01 >> Yeah. >> Um now now some of the questions I guess
38:05 I we I do know. I'd like to know is the standard model of physics correct about
38:08 the origins of the universe? Are we actually 13.8 billion years old? Um how
38:12 does the universe end? Does it end in a heat death or in some other way? Um, you
38:16 know, >> a black hole. >> We might be. >> Um,
38:23 >> Elon, can you talk about >> the whole sort of simulation question?
38:26 Are we a simulation? Maybe. >> Where does the uh where do you think we
38:31 find the answer first? In AI or in the stars? >> Yeah.
38:43 I I I don't know if if I I hope I hope more people can get behind a
38:49 philosophy of curiosity. >> Yeah. >> Because I think it's very exciting.
38:54 >> Yeah. >> Um and and and and inherently optimistic.
39:01 Um you like because there's there's this amazing sense of wonder
39:07 about the nature of the universe. And when you just when you uncover some
39:11 secret in the universe, that's amazing. And you're like a whole world of
39:18 understanding is opened up. I mean, we we used to not even know where all the
39:22 continents were. Um, you know, used to be like just the map would be there be
39:26 dragons and like all we know is that when they sailed in that direction, they
39:30 didn't come back. >> I mean, the moon base, >> that's all that's all they knew. I I
39:37 kind of feel like the moon base or just going to the moon for real this time
39:40 would be a big step in the right direction. You still have the moon uh
39:44 planned. What's the status of that? Is is that still on the agenda?
39:49 >> Yeah, I I think it I think having I think we want to try to reach new
39:52 heights as a civilization. >> Yeah. >> So, I think it's it's fine to go to the
39:57 moon, but but we should go to the moon in order to establish a lunar base, like
40:01 a a lunar research base. >> Yeah. Um, I mean there are parts of the
40:06 moon that are perhaps older than parts of of Earth. Um, and we we we might
40:10 understand more about the nature of the universe if we had a science base on the
40:13 moon. >> Um, that would be very cool. And then we we obviously want to go beyond the moon
40:20 uh to Mars and uh build a self-sustaining city on Mars. the I I I
40:27 do think that uh that that there is a fork in the road of human destiny where
40:32 um if we can establish a self-sustaining city on Mars with the the key test being
40:38 if the resupply shifts from Earth stop coming for any reason does Mars continue
40:43 to to prosper or does it die out >> at the point at which Mars is able to uh
40:49 prosper and grow on its own the probable lifespan of consciousness is
40:52 dramatically greater. because we are no longer dependent on everything going right on Earth. You
40:59 know, there's there's always some possibility of self annihilation on
41:03 Earth with the World War II or or a supervirus or um or or a meteor like
41:08 extin, you know, that destroyed the dinosaurs. We know from the fossil
41:12 record that there've been many mass mass extinction events. So uh the question
41:18 that I sort of was wondering about is will civilization will the
41:23 civilizational arc continue to ascend such that we can make Mars self-
41:28 sustaining before the civil civilizational ark descends >> um because the the window of opportunity
41:36 to make life multilanetary exists now for the first time in the 4
41:40 and a half billion year history of earth. >> Yeah. Elon, let's assume that we get
41:46 there and you're there. >> Um, you know, you'd be the elder
41:50 statesman. You'd have the moral authority of Mars. >> But I just there's this point that I I
42:03 think I I want to just emphasize again that that's that's it's more important
42:07 than the form of governance on Mars or who's there in the early days. What
42:15 really matters is that Mars um is self- sustaining that we are truly a
42:20 multilanet species and s such that we've achieved planetary redundancy so that
42:25 that if if something and obviously we should do everything possible to make
42:28 sure life on Earth is great but there's always some risk that of an annihilation
42:32 event on Earth. >> Yeah. >> Um like I said self annihilation or some
42:38 natural disaster. Um and uh and so the the probable lifespan
42:44 of consciousness increases dramatically as soon as uh as soon as we are
42:49 multilanet species with the key test being can Mars survive if the resupply
42:54 ships stop coming. So so getting like the first missions to Mars are not that
42:58 important. The what matters is can you get sufficient tonnage tonnage to Mars
43:04 such that Mars can prosper on its own. Um, and that means it has to have all of
43:08 the ingredients of civilization. It it it's not just that you need to build,
43:12 for example, a chip factory on Mars or ship fab on Mars, but you you need the
43:16 ability to build. >> Do you do you have a sense of the time
43:20 scale? Like, let's assume Starship is at a state starting in, you know, 2026.
43:24 Then there's going to be a bunch of testing. Obviously, there's going to be
43:27 a bunch of early testing. We only have certain launch windows. So, there's a
43:32 bunch of time constraints. Is that is this a 50-year thing in your mind? Is it
43:35 a 150 year thing? Is it something that is for our generation or is it our
43:40 children's generation? Where do you see that point if it's optimally possible?
43:44 You know, if things go and break our way, >> um I think it can be done in in 30
43:51 years. Um >> wow. So if provided there's an exponential increase in the in the
43:57 tonnage to Mars with each successive Mars transfer window, which is every two
44:03 years. So every two years the the planets align and you can you can
44:08 transfer to Mars. Um, so I I think in roughly 15, but maybe as few as 10, but
44:18 10 to 15ish Mars transfer windows. If you're um seeing exponential increases in the
44:26 tonnage to Mars with each Mars transfer window, then it should be possible to
44:31 make Mars self-sustaining um in in about call it roughly 25 years.
44:35 >> Amazing. That's incredible. All right, ladies and gentlemen, Elon Musk, we'll
44:41 see you when we're back in town. We miss you. We'll >> see you in person next time.
44:46 >> Thank you, brother. All
