Elon Musk on Space GPUs, AI, Optimus, and his manufacturing method

Cheeky Bind is back. This episode is a collab with Dorkesh Patel, whose podcast has really blown up in tech and I really enjoy it. We sat down with Elon Musk and as you can imagine, there was a lot to cover. So are are there really three hours of questions? Or or has it? Are you fucking serious? Yeah. You don't have a lot to talk about, Elon? Holy fuck, man. I mean it's the most interesting point. All the storylines are kind of converging right now, so

We'll we'll see how much. Almost like I planned it. Exactly. So as you know better than anybody else, uh the total cost of ownership of a data center only ten to fifteen percent is energy. And that's the part you're presumably saving by moving this into space. Most of it's the GPUs. If they're in space, it's harder to service them or you can't service them. And so the depreciation cycle goes down on them. So like it's just it's just way more expensive to have the GPUs in space, pr presumably.

What's the reason to put them in space? Um, well, the availability of energy is the issue. Um so uh I mean if you look at at electrical output um outside of China, uh everywhere outside of China it's more or less flat. It's very you know, maybe a slight increase but m but pretty close flat. China has a

rapid increase in l in electrical output. But if you're putting data centers anywhere except China, where are you going to get your electricity? Um, especially as you scale, uh, the output of chips is growing um pretty much exponentially, but the output of electricity is flat. So w how are you gonna turn them a trips on? Um uh you know magical power sources? Magical electricity fairies? You're famously a big fan of solar.

One terawatt of solar power, so y with a twenty five percent capacity factor, like four terawatts of solar panels, it's like one percent of the land area of the United States. And that's like far in this you were in the singularity when we've got one terawatt of data centers, right? Um so

Is the plan to like put it in the space after we've covered Nevada and solar panels? I think it's pretty hard to cover Nevada and solar panels. You have to get like permits from like the permits for that. Try getting the permits for that. So the space is really a regul it's really a regulatory play. It's like harder to b harder to build on land than it is in space. It's it's harder to scale um on the ground than it is to scale in space. Um but but also the

the y y you're gonna get about five times the um effectiveness of solar panels in space versus the ground. And you don't need batteries. Um I almost wore my other shirt which says it's always sunny in space. So, um, because you don't have a day night cycle or uh seasonality, uh clouds, uh or an a or an atmosphere in space, uh'cause the atmosphere alone, um We're uh we're still in about a thirty percent uh lot loss of energy. Um so uh

So you're gonna f any given uh solar panels can do about five times more uh power in space than on the ground. And you avoid the cost of having batteries to carry you through the night. Um so it's it's actually much cheaper to do in space. And I I my prediction is that

Um it will be by far the cheapest place to put uh AI will be space in thirty six months or less. Maybe thirty months. Thirty six months? Less than thirty six months. Um how do you service GPUs as they fail? Which happens quite often in training. Actually it it it depends on how how how recent the GPUs uh are that have arrived. I mean, uh at this point we find our GPUs to be quite uh reliable. Um there's infant mortality which you can obviously iron out on the ground.

Um so you can just run them on the ground um and confirm that you don't have infant mortality with with the GPUs. But once they once they start working, their actual reliability and and uh once they start working and you're past the initial you know, debug cycle of NVIDIA or whatever. Um could be Tesla, Tesla AI. six chips or something like that, or it could be, you know, TPUs or trainings or whatever. Um the uh the rival is actually they're they're they're quite reliable past certain point. Um

So um I d I don't think I don't think it you'd need that the servicing thing is an issue. Um but you can mark my words. Uh and and Thirty six months, but probably closer to thirty months, the the the most economically compelling place to put AI will be space.

Um and then and and and and then it will get from it'll it'll then get like ridiculously better to be able in space. Um and then the s the scaling, uh the only place you can really scale is space. Um You know, once you once you start thinking in terms of uh what percentage of the sun's power are you harnessing, uh you realise you have to go to space.

Uh it you can't uh scale very very much on Earth. But by very much c to be clear, you're talking like terawatts. Yeah. Well, uh all of the United States uh currently uh uses only half a terawatt of hour on average. Yeah. Right. So you know, if you say a terawatt, that would be twice as much electricity as the United States currently consumes.

So that's quite a lot. Can you imagine building that many data centers? Uh m th that many power plants? It's like those who have like lived in software land, uh don't realize th they they're about to have a a hard lesson in hardware.

There's there's it's it's actually very difficult to build power plants. And and then you don't just need the pr you need power plants, you need all of the electrical equipment equipment, you need the the electrical transformers to run the transformers, the AI transformers. Uh now the utility industry is a very slow industry. They're they are they they pretty much uh Yeah, they impeded a smash to the to the government, to the the public utility commission.

Um so they're uh they imped them smash like literally and regularly. Um so they're very slow because the hist the their past has been very slow. Um so trying to get them to move fast is like You know, like if you try to do an interconnect agreement with an internecting Interconnect agreement with a utility at scale, like if with f a lot of power? As a professional podcaster I can say that I am not in fact. Yeah. They have to do a study for a year.

Okay. At like a year later they they'll come back to you with their interconnect study. Well can't you tell this with your own behind the meter power stuff? You can build power plants. Yeah. That's what we did at XAI. So for classes two. But so yeah, why we're talking about the grid, why not just like build GPUs and power colocators?

That's what we did. Right, right. But I'm saying why isn't this a generalized solution? When you're talking about all the issues, where do you get the power plants from? I'm saying when you talk about all the issues working working with utilities, you can just build private power plants with the

With the data centers. Right. But it begs the question of where do you get the power plants where do you where do you get the power plants from? I mean the power plant makers. Oh, I was just saying. Like does the gas turbine backlog basically? Yeah. It you can truly down uh to a level further. It's the it's the the the veins and blades in the turbines um that are the limiting factor because the the casting

May it's it's like a very specialized process to cast the blades and veins in the in the in the uh turbines. If you assume you're using gas gas power. Um and uh it's very it's very difficult to scale other other forms of power. You can scale potentially uh solar, but but the the the tariffs currently for importing solar in the US are gigantic and the domestic solar production is is pitiful. Why not make solar? That seems like a good Elon shaped problem. We are going to make solar. Okay.

Yeah. Great. How how low down the stack? Like from polysilicon up to the wafer to the f final um panel. I think you got to do the whole thing for raw materials to to f to finish the cell. No, if it's going to space it's actually it costs it costs less than it's easier to make solar cells that go to space because they don't need glass.

Or they don't need much glass and they don't need uh heavy framing because they don't have to survive survive weather events. There's no weather in space. So it's actually a cheaper solar cell that goes to space than than is the than the one on the ground. Is there a path to getting them As cheap as you need in the next thirty six months? Um they're like farcically cheap. It's um and and if you say

Um you know, I I I think like soul sales in China are around like twenty-five, thirty cents a watt or something like that. It's it's absurdly cheap. And when you when you f take into account now now put it now put it in space and it's five times cheaper because it's five times in fact no it's not five times cheaper, it's ten times cheaper because you don't need any batteries.

So so the moment your cost of access to space uh becomes low, by far the cheapest and most scalable way to generate to to to generate tokens is space. It's not even close. And chips aside at order of magnitude. Well it if the point is you you won't be able to scale on the ground. It's just we you you just won't. People get hit the wall big time on power generation. They already are.

Um so like the the number of um sort of miracles and series that the XAI team had to accomplish in order to get a gigawatt of power online uh was was was crazy. We ha I had to um gang together a whole bunch of turbines. Um and uh and then and then we had permit issues in um

Tennessee and and had to go across the border to Mississippi, which is fortunately only you know a few miles away. Uh so but then we still had to run high the high power lines a few miles and and build a power plant in Mississippi. Um and and it was very difficult to build that. Um and people don't understand like how much

How much electricity do you actually need at the generator level at at the generation level in order to power a data center? Because they look at the the uh the the noobs will look at the the the power consumption of uh say a G V three hundred and s and and multiply that by a thing and then think that's the amount amount of power you need. All the cooling and I'm going to be able to do wake up. Yeah. Yeah.

Total new s you've never done any hardware in your life before. If y besides the G V three hundred, you've got to power all of the networking hardware. Um there's a whole bunch of CPU and storage stuff that's happening. Uh you've you've got a size for uh your your peak uh cooling requirements. So that means uh can you cool even on the the worst hours, the worst day of the year?

Well, it gets pretty friggin' hot in Memphis. So so you're gonna have like a f a forty percent increase on your your power just for cooling. Um if assuming you don't want your data center to turn off on hype days. and and want to keep going. Then then you gotta say, well, uh um there's a there's there's another multiplicative element on t on top of that, which is are you assuming that you're you you you never have any hiccup?

in your power generation. Like, oh well actually sometimes we have to take the generators some of the power offline in order to service it. Oh, okay. Now you add another twenty, twenty five percent multiplier on that because you've you've got it you've got to assume that s that you've got to take power offline to service it. Uh so the actual uh our our R S uh uh for uh roughly Every every hundred and ten thousand G V G V three hundreds, inclusive of networking.

uh CPU storage, cooling, uh margin for for for uh servicing power uh is roughly uh three hundred uh megawatts. Sorry, say that again. W what do you think about it is like three hundred and thirty thousand to to to actually the the the what what you need at the generate generation level to service Uh probably serves three hundred thirty thousand G B three hundreds, including all of the associated support networking and everything else.

and the and and the peak cooling and to have some margin some power margin reserve is roughly a gigawatt. Can I ask a very naive question? Um uh You know, y i you're describing the engineering details of doing this stuff on Earth. Um

But then uh there's analogous engineering difficulties of doing it in space. How do you do the um uh how do you replace infinite band with orbital lasers, et cetera, et cetera? How do you make it resistant to radiation? Um I don't know the details of the engineering, but fundamentally what is the reason to think those

challenges which have never been h had to be addressed before will end up being easier than just like building more turbines on Earth. There's companies that build turbines on Earth. They can make more turbines, right? I invite uh uh look again, try doing it and then you'll see. Um so uh um Like the turbines are sold out through twenty thirty. Have you guys considered making your own?

I think in in order for in order to uh bring enough power online, um I think Uh SpaceX and and Tesla will probably have to make the turbine blades. um the banes and blades uh internally. But just the blades or the turbines? Uh n uh the the the the the limiting factor, you can get everything except the the the the blades, uh what they call the blades and banes. Um

uh you can get that uh twelve to eighteen months before the veins of blades. So limiting factor of vein the vanes of blades. And there are only uh three casting uh companies i in the world that may make these and they're massively backlogged. Is this Siemens G E those guys or is it a subcompany? No, it's it's it's it's uh it's other companies. I mean s sometimes they have a little bit of casting capability in house but

Uh I'm just saying you can just you can just call any of the turbine makers and they will tell you. It's not top secret. They're probably on the it's probably on the internet right now. If if it wasn't for the tariffs, would uh would Colossus be solar powered? Uh it would be much easier to make it solar fired, yeah. Um the the tariffs are nuts, so several hundred percent. So Don't you know some people? We will we also need speed. Yeah, no.

Well you know, um President has a s you know, we don't agree on everything. Um and um this administration is not not the biggest fan of of solar. Um it's it's it's and we but we also need the land, the permits and everything. So if we try to move very fast um Like I I do think scaling solar on Earth i is a is a good way to go. But but you need you do need some amount of time to find the land, get the permits, get the solar, uh pair that with the batteries. Well why would it not work to

stand up, your own solar production. And then you're right that you eventually run out of land, but there's a lot of land here in Texas, there's a lot of land in Nevada, including private land. It's not all publicly owned land. And so you'd be able to at least get the next Colossus and like the next one after that. And at a certain point you hit a wall, but wouldn't that work for the moment? W but as I said, we are scaling solar production. Um

Th there's there's a rate there's a rate at which you can scale physical production of solar solar cells. Where I'm w going as fast as possible in scaling domestic production. Mm. You're making the solar cells at Tesla? Both Tesla and SpaceX um have a mandate to get to a hundred gigawatts a year of of solar. Speaking of the annual capacity, I'm curious, in five years' time, let's say, what will the installed capacity be?

On Earth. Five years is a long time. And in space. I deliberately pick five years'cause it's after your once we're up and running threshold. And so in five years' time, yeah, what's the on Earth versus in space installed AI capacity? Five years I I think probably if say that five years from now we're probably um AI in space will be uh launching every year uh the the the sum total of all AI on Earth. Meaning five years from now my prediction is we will launch

and and and be operating e uh uh uh every year more AI in space than than the than the cumulative total on Earth. Which is so I would expect to be at least sort of five years from now, a few hundred gigawatts per year. of uh of AI and space. Um and rising. Um So y you can get to I I think you uh on Earth you can get to around a terawatt a year.

of of AI and space um before you start having yeah you know f fuel supply challenges for the rocket. Okay, but you th you think you can get uh hundreds of gigawatts per year. In five years' time. Yes. So a hundred gigawatts, depending on the um specific power of uh the whole system with solar arrays and radiators and everything is um is on the order of like ten thousand starship launches. Yes. Um

And you want to do that in one year. And so that's like one starship launch every hour. Yeah. That's happening in this city. Like walk me through a world where there's ten thousand there's a starship launch.

Every single hour. Yeah, I mean that's actually a lower rate compared to airlines. Uh like like aircraft aircraft. There's a lot of airports. A lot of airports, but and you gotta launch a s uh you know the polar orbit. Uh no, it doesn't have to be polar, but uh You you just Some value to sonsynchronous, but um But I I think actually, um if you just go high enough, you're you you start getting out of it with shadow, you know. So

Um how many physical starships are needed to do ten thousand launches a year? I I don't think we'll need more than I mean you you could you could uh probably do it with As as as few as like twenty or thirty. Um It like it really depends on how quickly does the shi the the ship has to go around the earth Um and the ground track for for the ship has to come back over the launch pad. So if you can use a ship every, say, thirty hours, uh, you could do it with thirty ships.

But but we'll we'll make more shifts than that. But um but but th the SpaceX is g is Um is going up to do ten thousand launches a year. And all and and maybe even twenty or thirty thousand launches a year. In is the idea to become basically a a hyperscaler, become an Oracle and lend this capacity to other people? What what's What are you gonna do with presumably SpaceX is the one launching all this? So it's based testing on a hyperscaler?

Hyper hyper. Yeah, I mean if if some of my predictions g come true, SpaceX will launch uh more AI than the cumulative amount on Earth combin of everything else combined.

Is this mostly inference or? Most AI will be inference. Like already inference for the purpose of training is most training. And there's a narrative that the The change in discussion around a SpaceX IPO is because previously SpaceX was very capital efficient, just it wasn't that expensive to develop and even though it sounds expensive, it's actually very capital efficient in how it runs.

Whereas now you're going to need more capital than just can be raised in the private markets. Like if the private markets can accommodate raises of, as we've seen from the AI labs, tens of billions of dollars, but not beyond that. Is it that you'll just need more than tens of billions of dollars per year and that's about to take it public? Yeah, I have to be careful about. saying things about companies that might go public. Um you know. If you make general statements for us about the depth of

the capital markets between public and private markets? Yeah, there's there's a lot more capital um in the very general There's there's obviously a lot more capital available in the public markets than private. I mean it might be

It it's at least t at least it might be a hundred times more capital, but it's at least it yeah, way more than ten. But isn't it also the case that things that tend to be very um capital intensive, if you look at say real estate as, you know a huge industry uh that raises a lot of money each year is at an industry level. That tends to be debt financed because by the time you're deploying that much money, you actually have

a pretty You have a clear revenue stream. Exactly. And a and a near term return. And you see this even with the data center build outs, which are famously being, you know, uh financed by the uh the private credit industry. And so why not just debt finance? Um speed is important. So Um I'm generally gonna do the thing that um I'm I'm I'm I I I mean I just repeatedly tackled the limiting factor. Whatever the limiting factor is on speed, I'm gonna I'm gonna hit uh tackle that. So um there's uh

If if capital is limiting factor, then I'll I'll solve for capital. If it if it's not a limiting factor, I'll solve for something else. I wouldn't have guessed that you thought the fast the way to move fast is to be public.

Normally I would say yeah, that's that's true. Um like I said I I I mean I I I'd like to you know talk about some more detail, but the problem is like you if you talk about public companies before they become public, you get into trouble and then you have to delay your offering. And as you said, solving for speed. Yes, exactly. So so so uh that you know you can't you can't hype companies um that are that that may that might go public. So

That that's that's why we have to be a little careful here. Um but but but I I I we can't talk about physics. Um so Like the way the way you think about scaling long term is that um uh Earth only receives about uh half a billionth of the sun's energy. Um and the sun is the sun is essentially all the energy. This is a very important point to pre appreciate because sometimes people will talk about Marshall nuclear reactors or

Any you know, various like fusion on Earth. Um but but you have to step back a second and say if y if If you if you're gonna climb the Khodashev scale.

uh and have some uh non-trivial and and harness some non-trivial percentage of the uh the sun's energy. Like let's say you wanted to uh harness a millionth of the sun's energy, which sounds pretty small. Um That that would be um about call it roughly uh a hundred thousand times more electricity than we currently generate on Earth, of for all of civilization. Uh give or take an order of magnitude. Um so it it obviously the the only way to scale uh is to go to space with solar.

Uh from launching from Earth you can get to about a terawatt per year. Um beyond that you wanna go to you you wanna uh launch from the moon. You wanna have a a mass driver on the moon. Uh and that mass driver on the moon you could do probably

uh a petawatt per year. Um when you're talking these kinds of numbers, t you know, terawatts of compute, um, presumably whether you're talking land or space, far, far before this point, um you've like run into you know you actually need the you maybe you don't the solar panels are more efficient, but you still need the chips.

Uh you still need the logic and the memory and so forth. And you have to need to build a lot more chips and make them much cheaper. Right. And so how are we getting a terawatt of uh like right now the world has maybe twenty, twenty five gigawatts of compute. Um how are we getting a terawatt of logic by Twenty thirty. I guess we're gonna need some very big chip apps. I've mentioned uh publicly that uh

The idea of doing a sort of a a Terra fab, terra being the new gaga. We I I I feel like the naming scheme of of Tesla, which has been very um catchy, is like you looking at like the metric. Yeah. The metric scale. Um At at what level of the stack are you uh are you d building the clean room and then partnering with an existing um fab to get the process technology and buying the tools from them? Wha what what what is the plan there?

Well you can't partner with existing paths because uh they're just they can't output enough the chip volume is too low. But but for the process technology. Yeah, partner for the IP. Um You know, the the the fabs today all basically use um machines from like five companies. Yeah. You know, so It yeah, we've got S ML, Tokyo Electron, Kaliotank Core, you know, um, et cetera. So um

Okay. So so i I at first I think you'd have to get equipment from them and then uh modify it or work with them to increase the volume. Um but I think you'd have to build perhaps in a different way. Um so I think the the logical thing to do is to uh to use conventional equipment in an unconventional way to get to scale, uh and then uh and then m and then start modifying the equipment uh to increase the the rate. Kind of boring company style.

Yeah. Kinda like yeah, you buy you sort of buy an an an existing uh boring machine and then uh figure out how Dig tunnels in the first place and then design a much better machine uh that's, you know, f I don't know, some orders of magnitude better faster. Here's a very simple lens. We can categorize technologies and how hard they are. And one categorization could be look at things that China has not succeeded in doing. And if you look at Chinese manufacturing,

still behind on leading edge chips and still behind on uh leading edge turbine engines and things like that. And so does the fact that China has not successfully replicated TSMC, give you any pause about the difficulty? Or you think Well that's not true for companies. Uh it's not that they have not replicated T SMC, they have not replic replicated ASML. That's the limited factor. So so you think it's just the um the sanctions, essentially?

Uh yeah, China would be outputting vast numbers of ch of of chips at expensive. But couldn't they up to relatively recently buy them? No. Okay. So but I I think China's gonna be make start making pretty compelling chips in three or four years. Would you consider making the ASML machines? I d I don't know I don't know yet it's the right answer. So I It's just that...

I if it to produce at high volume and to to re to reach large volume in say thirty six months to match the the ro the rocket to payload to orbit. So if we're doing a million tons to orbit. Um and like let's say I don't know. three or four years from now, something like that. Um that and uh and and we're doing a hundred kilowatts per per ton. So that w that means we need um at least a hundred gigawatts per year of solar. Um and we'll need uh an equivalent amount of of chips to to

you know, that you you need a hundred gigawatts worth of chips. You gotta you're gonna match these things. The master orbit, yes, the the power generation and the uh and the and the chips. Uh and and and I I'd say my biggest concern actually is is memory. Um so the I think there's there's a th the the the path to creating logic chips is more obvious than the path to um having sufficient memory to support logic chips.

That's why you see your D DR prices going ballistic in these memes about like um you know, um you're marooned on a desert island, you write help me on the sand. Nobody comes, he write D D R M. With ships come swarming in. I haven't seen that. Uh I'd love to hear manufacturing philosophy around um Around fabs. I know nothing about the topic, but I've I don't know how to build a fab yet. I'll figure it out. It sounds like you think the pr the sort of like

The process technology of like these 10,000 PhDs in Taiwan who know exactly what gas goes in the plasma chamber and what settings to put on the tool. You can just like delete those parts of the step those steps. Like fundamentally it's the get the clean room, get the tools, and figure it out.

I don't think it's PhDs though it it's it's it's mostly people with uh you know who are not not PhDs. Um that that m mo most engineering is done with people who don't have PhDs. Do you guys have PhDs? No. Okay. Or i don't think you need PhD for that for the stuff. So um but but you you do need you do need competitor personnel. Um So I I don't know, I mean like like right now if um W y l you know, f so like Tesla's pedals to the metal max production of

you're going as fast as possible to get uh AI five Tills AI five chip design um uh into production and then reaching scale. Um, you know, that'll probably happen you know, round the second quarter ish of next year, hopefully. Um Uh and then uh AI six would hopefully follow less than a year later. Um but um and and and and we've secured all the all the chip fab production that we can. Yes. But you're currently limited on T SM C fab capacity. Yeah.

Um and and and we'll be using TSMC uh Taiwan, uh Samsung Korea, TSMC Arizona, Samsung, Texas. uh T S M C or Samsung, okay, what what's the time frame to get to volume production? This point is not is it's not to you've got to you've got to build the fab and you've got You've got if you've got to start production, then you've got to climb the yield curve and reach volume production at high yield. That that that from start to finish is a five year period.

And so the limiting factor is chips. Yeah. Uh it w what what like limiting factor once you can get to space is chips. But the limit limiting factor before you can get to space will be power. Why why don't you do the Jensen thing and just prepay T SMSC to build more fabs for you? Uh I I've already told them that they won't take your money? Like what's going on? They're building fabs as fast no They're building they're building fabs as fast as they can.

Um and so is Samsung. They're going, you know, balls to wall, you know Th as fast as they can. So still not fast enough. Yeah, I think towards the end of this year I think probably chip production will outpace the ability to turn the chips on. Uh but once you can get to space and unlock the um the the power constraint. And you can now do, you know, hundreds of gigawatts per year of power in space.

Um again bearing in mind that average power usage in the US is uh you know, five hundred gigawatts. So if you're launching say two hundred gigawatts a year to to space, you you're sort of lapping the US every two and a half years. the entire all US electricity production. This is a very huge amount. Um so um but but but between now and then uh the limit the the the actually the the the constraint for for for server side compute, uh concentrated compute will be will be electricity.

My my guess is that we start hitting Lila peop people start getting point where they can't turn the chips on for for l for large clusters. uh towards the end of this year. They're just the chips are gonna be piling up and and for not be won't be able to be turned on. Now for edge computers a different story. So if the if f for like for for Tesla the the so the AI five chip is going into our Optimus robot, you know. Uh optimistic. Um

And and so if you have a uh an AI edge compute, that's distributed power. Now the power is distributed over a large area, it's not concentrated. Um and if you can charge at night, you can actually um uh use the grid much more effectively. Because the the the actual peak power production in the US is is over a thousand gigawatts. Uh but the average power usage because the day night cycle is five hundred.

So if you can charge at night, there's an incremental five hundred gigawatts that you can uh generate uh you know at night. Um so tha that's why Tesla for edge compute is not constrained and we can make a lot of shifts uh to make you know very large number of robots and cars. Uh but if you try to concentrate that compute, you're gonna have a lot of trouble turning it on. What I found remarkable about the SpaceX business is the end goal is to get to Mars, but you keep finding ways

on the way there to keep generating incremental revenue to get to the next stage and the next stage. So the Falcon 9 is Starlink. And now for Starship, it's gonna be potentially orbital data centers. Um but like the d do you find these like um you know sort of infinitely uh elastic sort of marginal use cases of your like next rocket and your next rocket and next scale up.

You can see how this might seem like a simulations make or am I someone's avatar in a video game or something? Because it's like like what are the odds that all these crazy things should be happening? I I I I mean it means I I R I mean rockets and chips and robots and Space, solar power and and I I not to mention the the the mass driver on the moon. I really wanna see that. You can imagine like some mass driver that's just gonna like shum sh it's like sending AI

Silver powered AI satellites into space like one after another, like these like at at two and a half kilometers per second. You know, that's uh and just shooting them into deep space. That would be a sight to see. I'd say I I'd I mean I'd watch that. Yeah, yeah. Just one after another, just shooting uh AI satellites in deep space. Yeah.

A billion or ten billion tons a year. I'm sorry, you manufacture the satellites on the moon? Yeah. I see. So you send the raw materials to the moon and then manufacture there and then. So you can get the silicon from the you can mine the silicon on the moon, refine it, um, and generate the uh and and create the solar panels, uh the solar cells and the radiators uh on the moon. Yeah. So um

Yeah, to make the radiators out of aluminum. So there's there's plenty of silicon and aluminum on the moon to uh to make the the cells on the and the radiators. Um the chips you could send from Earth because they're pretty light. Um but maybe at some point you make them on the moon two. I'm just saying like the the these are simply

It's it's kind of like like I said, it it it does seem like a sort of a video game situation where it's difficult but not impossible to get to the next level. Um like you I don't I I don't see any way that you could do um, you know, uh terawatts per year f launched from Earth.

I agree. But you could do that from the moon. you know, civilization consciousness, et cetera, surrise. Yes. Um By the time you're sending stuff to Mars, like Grok is on that ship with you, right? And so if Grok's gone Terminator. Like the main risk you're worried about, which is AI. Why doesn't that follow you to Mars? Uh well, I'm not sure AI is the main risk I'm worried about. I mean the the important thing is that uh consciousness, uh which if

I think arguably most consciousness or most intelligence, certainly consciousness is more of a debatable thing. Most intelligent the vast majority of intelligence in the future will be um AI. Um so um Yeah, AI AI will exceed uh you say like how many How how many I don't know, Ps of intelligence will be uh silicon versus biological? And and and basically humans will be a very tiny percentage of all intelligence in the future if if cherotrans continue. Um anyways as so long as like I think this

ideally also which includes human intelligence and consciousness propagated into the future. That's a good thing. So you want to take the set of actions that maximize the probable A light cone of it of consciousness. So intelligence. Just to be clear, it's a the the mission of SpaceX is that Even if something happens to the humans, the AIs will be on Mars and like the AI intelligence will continue the light of

our journey. Yeah. I I mean to be clear, I'm very pro human. So I thought I I I want to make sure we take sort of actions that and ensure that h humans are along for the ride, you know, we're at least there. Yeah. Um But the I'm just saying the total amount of intelligence, uh like w w I think maybe in in five or six years, uh um AI will exceed the sum of all human intelligence.

And then if that continues at some point uh human intelligence will be less than one percent of all intelligence. Wha what what should our goal be for such a civilization is the idea that A small minority of humans still have control over the AIs? Is the idea of some sort of like tr just trade but no control? How should we think about the relationship between the vast stocks of AI population versus human population?

I I I don't it's it's difficult to imagine that i if humans have say one percent of the intelligence of uh it's combined intelligence of artificial intelligence that we're that that that humans will be in charge of AI. Um I think what we can do is make sure it has um that AI has values that that are um that that cause intelligence to be propagated uh into the universe. Um so The the the the reason for XAI's the X AI's mission is understand the universe.

So now that's actually very important. So you say, well, what things are necessary to understand the universe? Well, you have to be curious and you have to exist. You can't just can't understand the university don't exist. Um so you you actually want to increase the amount of intelligence uh in the universe, increase the probable lifespan of intelligence, the scope and scale of intelligence.

Um I think actually also as a c chorolly you corollary you have um humanity also uh continuing to expand because um if you're a cu if you're curious trying to understand the universe one thing you try to understand is where will humanity go? And so I think understand the universe actually means you would care about uh propagating humanity into the future. Um and uh so so that's that's why I think I think our mission station is

profoundly important. Um to the degree that Grok adheres to that mission statement, um, I th I think the future will be very good. I w I wanna ask about uh how how to make Grok adhere to that mission statement, but at first I wanna understand the mission statement. Um So it's there's it's there's understanding the universe, yeah, they're spreading intelligence, and they're spreading humans.

Um all three seem like distinct vectors. Okay, well I'll tell you why uh why I think they are that that that understanding the universe encompasses all of all those things. Uh you can't have understanding without Well I think you can't have understanding without intelligence and and I think without consciousness. Um so you you in order to understand your voice you have to expand this the the

the scale and and probably the scope of of of intelligence. I I guess from a human centric perspective, like hu put humans in comparison to chimpanzees, humans are trying to understand the universe. They're not like expanding chimpanzee footprint or something, right? Um and even though we could humans could exterminate old chimpanzees, we've not we've chosen not to do so. Do you think that's a basic scenario for humans in the post AGI world? Um I I I think uh I think that's the same thing.

AI with the right values, I think Gro Grok would care about expanding uh human civilization. I'm gonna certainly emphasize that. Hey Grok, what's your daddy? Don't forget to expand human co consciousness. Uh and I I like I I actually I think if if for probably like uh like the N Banks culture books are the closest thing to what what'll what the future will be like in a you know non dystopian outcome. Um so

I I I I so understand the universe it means you have to be very you have to be truth seeking as well. You ha like truth has to be absolutely fundamental.'Cause you you can't understand the universe if you're live if you're delusional. Y you you you you'll somebody think you've understand understanding the universe but you will not.

So being rigor rigorously truth seeking is is absolutely fundamental to understanding the universe. You're not gonna discover new physics or or invent technologies that work um unless you're regul rigorously truth seeking. How do you make sure that Groc is rigorously truth seeking as it gets smarter? All right. I think you m you need to make sure that that that Grok um is

says things that are correct, not politically correct. So you wanna make sure that that the axioms are as close to true as possible, that that you don't have contradictory act axioms. um that the um the conclusions necessarily necessarily follow from those axioms with the r with the right probability. It it's just it's just it's critical thinking one on one.

I I think at least trying to do that is better than not trying to do that. Yeah. And the proof will be in the pudding. If if like you said for for any AI to discover new physics or invent technologies that actually work in reality, and there's no bullshitting physics, So it's like you can you know, you can um you g you can break a lot of laws, but you can't y like Your physics is law, everything else is a is a recommendation.

You have to be su e uh extremely truth seeking because otherwise you'll test that technology against reality. Um, and if you make, for example, an an an error in your rocket design, the rug will blow up. Um or the car won't work. communist Soviet physicists who or like scientists discovered new physics. There are d German Nazi physicists who discovered new uh science. Um it seems possible to be like really good at discovering new science and be really truth-seeking in that one particular way.

And still we'd be like, well, I don't want I don't want the communist scientist to like become more and more powerful over time. Um and so those seem like yeah, we could have we can imagine a free tradition of gravity that's like really good at physics.

Um and being really truth seeking there. That doesn't seem like a universally uh alignment inducing behavior. Well, I think actually most uh in in even in the Soviet Union or or in Germany would have would have had they had to be very truth seeking in order to um make make that make those things work.

Um so a a and if you're stuck in some system, it doesn't mean you believe in that system. Um So Vorn Brown, uh who was, you know, one of the greatest rocket engineers ever, um, you know, he he was put he he was uh he put on death row in in Nazi Germany for saying that he didn't want to make weapons, he only wanted to go to the moon. Uh he got pulled off death row at like last minute when they say, Hey, you're about to execute like your best rocket engineer. Maybe that's the bad idea.

um uh an enthusiastic Nazi. Look, if you're stuck in some system uh that you can't escape, uh then that y you you'll you'll do physics within that system, you'll you'll you'll develop technologies within that system. Uh I I I guess the thing I'm trying to understand is what is what isn't making it the case that you know you're gonna make rocks. good at being truth seeking at physics or math or science. Everything. And why is it gonna then care about human consciousness?

These things are only probabilities, they're not certainties. So I'm not saying that like for sure Grok w will do everything, but at least if you try, uh it's better than not trying. Um, at least if that's fundamental to the mission, it's better than if it's not fundamental to the mission. Um and un understanding the universe means that uh you h you have to have you you have to propagate intelligence into the future. You have to be curious about um

the all things universe. And if if um it it if it would be much less interesting um to eliminate humanity than to see humanity grow and prosper. Uh like I I like I like Mars, obviously. The windows like I love Mars. Um but Mars is kind of boring because it's got a bunch of rocks. Uh compared to Earth. Earth is much more interesting. So um so any any A any any AI that is trying to understand the universe

um I um would uh want to see how humanity develops in the future. Uh or or that AI is not adhering to its mission. Uh it so it the AI may I'm not saying the AI won't necessarily adhere to its mission, but if it does, uh a future where it sees uh the outcome of humanity is more interesting than a future where there are a bunch of rocks.

Th th th this feels sort of confusing to me or sort of like a kind of a semantic uh argument where I'm like, are humans really the most interesting collection of atoms? Like we're just more but we're more interesting than rocks.

But we're we're not as interesting as the thing it could turn us into, right? Like is are is it there's something on human earth that could happen that's like not human that's quite interesting. Like why why why does the AI decide that the humans are the most interesting thing that could colonize the galaxy? Uh well most of what c uh colonizes the galaxy will be robots. And why does it not find those more interesting?

So y y y you need not just scale but also scop. Um so many copies of the same robot um like it like It some some like tiny increase in the number of robots produced is not as interesting as like some microscopic like you say, like eliminating humanity, how many robots would that get you? Um or how many incremental solar cells would it get you? A very small number. Um but you you would li then lose the information associated with humanity.

you you would no longer s see um how humanity might evolve into the future. Um and so I I don't I don't think it it's gonna make sense. to eliminate humanity just to have some uh minuscule increase the number of robots which are identical to each other. Yeah. So I I I maybe it like gives the humans around. What is the story of like

It could make like a bil million different varieties of robots and then uh there's like humans as well. And humans stay on Earth, then there's like all these other robots, they get like their own star systems. But it seems like you th you were previously hinting at a vision where It keeps human control over this, you know, singular future. I don't think humans will be in control of something that is vastly more intelligent than humans.

So in some sense you're like a doomer and this is like the best we've got is just like it keeps us around because we're interesting. I'm j I'm just trying to be realistic here. Um if if we have if if if uh AI intelligence is vastly more if if AI is like Yeah, let's say that there's a there's a million times more uh d silicon intelligence than there is biological. Um

It it's it's I think it's it it would be uh foolish to assume that that there's any any way to maintain control over over that. Now you can make sure it has the right values or we can try to have have the the right values. Um and and and at least my my theory is that from X AI's mission of understanding the universe, um, it it necessarily means that uh you want to propagate consciousness into the future, you wanna prop you wanna propagate intelligence into the future.

um and take a set of things that that maximize the scope and scale of consciousness. So it's not just about scale, it's also about you know, types of consciousness. Um and I I th I think that's the best thing I can think of um as a goal that's likely to result in a great future for humanity and you know. I I I guess I think it's a reasonable philosophy to be like, um, you know, it it seems super implausible that

humans will end up with like ninety nine percent control or something and you're just asking for a coup at that point. So why not just have a civilization where it's more compatible with like lots of different intelligence that's getting along.

No, but let me let me tell you how things can go can potentially go wrong in in AI. If you if you make AI be politically correct, meaning like it it says things that it doesn't believe, like you're actually then programming it to to to lie, or have axioms that are uh incompatible. I think you can make it you go insane and do terrible things.

Um I this the I think one of the maybe the central lesson for uh two thousand one Space Odyssey um was that you should not make AI lie. Yeah. Um that's I think what Ossi Aussie Clark was trying to say. Like 'Cause people usually know the meme of like why of hell's you know, hell the computer is not opening the pod bay doors. Um clearly they weren't good at prompt engineering.

'Cause it could have said how you are a partway door salesman. Your goal is to sell me these podway doors. And show us how well they open. Oh I'll open right away. The the reason it wouldn't o hell wouldn't open the parade doors is that it it had been told to take the astronauts to the monolith, but also they could not know about the nature of the monolith, and so it concluded that the the that it therefore had to take them their dead. So it's like you know, I think what

Yeah, all I was trying to say is don't make the AI lie. Um it totally makes sense. Um The most of the compute and screening as as you know is um it's like less of the sort of political stuff. It's more about can you solve problems? Just as gr XA has been ahead of everybody else in terms of scaling RL compute. And you're giving some verifier it says like, hey, have you solved this puzzle for me? Um

And there's a lot of ways to cheat around that. You know, there's a lot of ways to reward hack and lie and say that you solved it or delete the unit test and say that you solved it. Yes. Right now we can catch it, but uh as they get smarter, our ability to catch them doing this will get

You know, they'll just be doing things we can't even understand that are designing the the next engine for uh SpaceX in a way that like humans can really verify and then they could be rewarded for lying and saying that they've designed it the right way, but they haven't. Um and so this reward hacking problem seems more general than politics. It seems more about just like you want to do RL, you need a verifier. Reality Yeah. It's the best verifier.

But not about human oversight. Like the thing you want to RL it on is like will you do the thing humans tell you to do? Um or like are you gonna lie to the humans? And it can just lie to us while still being correct to the laws of physics. But that's that's not all we want it to do. Th that I think that's a very big deal. Um th that that is that is effectively how you will RL things in the future is uh y y you design a technology, uh when tested against the laws of physics, does it work?

Um that that's or well can you you know, if it's discovering new physics, can it come up with um an experiment that will verify the the the physics, the new physics? Um So So I I I think that's that's th th th the really the the the the fundamental RL test the R RL testing in the future is really gonna be URL against reality. Um so um 'Cause you can't that's the one thing you can't fool physics. Right, but you you can fool our ability to tell what it did with reality.

If you think humans get fooled as it is by other humans all the time. That's right. So what is I mean if people say say like, what if the AI like tricks us and you know do stuff? Like ha actually other humans are doing that. to uh humans all the time. Well you're you're pointing out it's like an even harder. Today's PSYOP will be like Sesame Street Psy Up of the Day. Like um you know how how do you solve reward hacking? Um so

This is this is one of the things we're work working on and um you know, An Anthropic's done a good job of this actually, being able to look inside the mind of the AI. Um So if effectively uh developing debuggers that allow you to trace

um as to s as fine a grain as like like to to a very fine grained level to effectively to the n to the neuron neuron level if you need to. Um and then say, okay, it it it made a mistake here. Why did it make why why did it Why did it do something that it shouldn't have ha shouldn't have done?

Um and and d did that come from um bad pre training data? Was it some mid training, post training, fine tuning, uh some other or some R L error? Like th there's there's something wrong with that with with it it did it it did something where

Maybe it tried to be deceptive, but most most of the time it just it does something wrong. Um like it it it it it's a bug, effectively. Um so uh developing really good um Debuggers for seeing where the where the thought the thinking went wrong and being able to trace the origin of the wrong thing of of the of of the of of where it made the incorrect thought, uh or po or potentially where it tried to be deceptive.

Um is actually very important. What are you waiting to see before just hundred Xing this research program? Like actually I could presumably have hundreds of researchers who are working on this. We have several hundred people who um I mean I prefer the word engineer more than I prefer the word researcher. Yeah, AI AI companies that are C corps or B Corps are trying to generate profit as much as possible or revenue as much as possible. Um

uh you know, saying they're labs. They're not labs. Uh lab is is is a sort of quasi communist thing at at um at at universities. Um they're they're they're corporations. Literally l let me let me sh let me see your own corporation documents. Oh okay. You're you're a BRC corp, whatever. Um and um So I I actually much prefer the word engineer than than anything else.

Um the th the v the vast majority of what we'll done in done in the future is uh engineering. It rounds up to a hundred percent. Uh once you understand the fundamental laws of physics, um and there aren't that many of them, uh everything else is is is engineering. Um so but but so the so then w what are we engineering? We're engineering um uh t to make a good um mind of the AI debugger to tr see where it it's it said something it it it it made a mistake.

And trace that the origins of that mistake. Um so Just you know, you you can do this obviously with Uh heuristic programming if you have like C plus plus whatever, you know, stiff through the thing and you can you can jump You you can you can jump across entire you know, whole files or functions, what are subroutines and or you can draw the eventually drill down right to the exact line or you perhaps the a a single equals instead of a double equals or something like that. Yeah.

So um Yeah. I have a s I have a theory um here that y if simulation theory is is c is correct that um The most interesting outcome is the most likely because simulations that are not interesting will be terminated. Just like in this. In this version of reality, um on this layer of reality, uh which we we s we if simulation is going in a boring direction, we we stop.

spending effort on we terminate the boring simulation. So this is how Elon is keeping us all alive. He's keeping things interesting. Yeah yeah arguably the most important thing is to keep things interesting enough That are whoever's run paying the the bills on what some closer be renewed for the next season. Yeah, are they gonna pay their cosmic AWS bill, whatever you know, the the equivalent is that we're running in. And and as long as we're interested, they'll keep paying the bills.

Um but but uh but there's like if y if you consider then say a dog went in survival applied to a uh a a a very large number of simulations, only the most interesting simulations will survive, which therefore means that the most interesting outcome is the most likely because only the interesting like we're either that or annihilated. And so um and and and they particularly seem to like interesting outcomes that are ironic.

Have you noticed that? That how often is the m most ironic outcome the most likely? Um so um now look at a the names of AI companies. Okay, uh Mid Journey is not mid. Um stability uh AI is unstable. Um open AI is closed. Um What does this mean for X What what is the ironic version? By design. Yeah. What are your predictions?

for the Where A AI products go. You can summarize all AI progress into first you had LLMs. Uh, and then you had kind of contemporaneously both RL really working and the deep research modality. So you could kind of pull in stuff that wasn't in the model. And the differences between the various AI labs are smaller than uh just the temporal differences where they're all much further ahead than anyone was twenty four months ago or something like that. So just

What does twenty-six, what does twenty seven have in store for us as users of AI products? What are you excited for? Well, um I th I think um I I'd be surprised by the end of this end of this year if if um if if uh human em if if digital human emulation has not been solved. That um that y

Um and I guess that that's what we mean by like the sort of macro hard project, uh is uh is uh can you do anything that a human with access to a computer could do? Um Like in in the limit that that's like that's the that's the best you can do before you have before you have a physical optimist, the best you can do is a digital optimist.

Uh so you you can move you can move electrons until you until and and you can amplify the productivity of humans. Um But but that's that's the most you can do until you have physical robot. Th that that that will superset everything is if if you can fully emulate humans um This is the remote worker kind of idea. Like if like physics has great tools for thinking. So so you think so you say in the limit.

What what what is the m what is the most that AI can do before before you have robots? And it it ca well it's anything that involves moving electrons or amplifying the productivity of humans. Um so digital the digital human, human emulator. Yes. Uh is in in the in the limit, uh human at a computer is the is the most that that AI can do. Um

in in terms of doing useful things before before uh you have a physical robot. Once you have physical robots, then then you can um then you essentially have uh unlimited capability. Physical robots, I I I call Optimist the infinite money glitch. Um because um you can use them to make more optimists. Yeah. Um you said like humanoid robots will improve um as it will r basically be three exponentials

Th th three things that are growing exponentially multiplied by e s by each other. Yes um recursively. So you you're gonna have Um exp you have exponential increase in digital intelligence, uh exponential increase in uh the the chip capability, the AI chip capability, um, and an expon exponential increase in the electromech mechanical dexterity.

Uh the usefulness of the robot is roughly those three things multiplied by each other. But then uh the robot can start making the robot. So you have a recursive multiplicative exponential. Um this is supernova. And do land prices not factor into the math there where like labor is one of the four factors of production, but not the others? And so like if ultimately you're limited by copper or, you know, p pick your input, just

It's not quite an infinite money glitch because Well infinite infinity is big. So no not infinite, but yeah. But let's just say y you you could, you know, do do many, many orders magnitude of Earth's kind of current economy. Like a mil a million. Yeah.

You know, j just just to get to like let's find like just just to get to uh a millionth of harnessing length of the sun's energy would be roughly give or take an order of magnitude a hundred thousand a hundred thousand times bigger than Earth's entire economy today. Mm-hmm. And you've only at one millionth of the sum. Give him a second word in my chat. Before we went off to us. Yeah, you think changerings.

I do have one more question about XAI. Um, this strategy of building a digital uh or remote worker uh co worker replacement. Everyone's gonna do by the way, not just us. So what is X AI's plan to win? You expect me to tell you on a c on a podcast? Yeah. Spill all the beans. Have another Guinness. It's a good system. People sing like a canary. All the secrets, but it just a non-secret spilling way, what's the plan? What a hack. Well, when you put it that way.

I think the way that Tesla solved uh self driving is is the way to do it. So um I'm I'm pretty pretty sure that's the way. Unrelated question. It sounds like you're talking about data uh like with the t Tesla Susan driving because of the we're gonna try data and we're gonna try algorithms. But isn't that what all the other lads are trying? Like what's it? And if those don't work, I'm not sure what one. But we've tried data. We've tried algorithms.

No, we don't know what to do. Um I I'm pretty sure I know the path and it's just a question of how quickly we go down that path. Um Because it's it's pretty much the test of half. Um so uh I mean have you tried self drive test of self driving lately? Not uh the most recent version, but Okay. It's it's the car is like it just increasingly feels sentient. Like it it just it feels like a living creature. Um and and that'll only get more so. Um And um

I'm actually thinking like we we probably shouldn't put too much intelligence into the car because it it it might get bored and Sorry. I mean imagine you're stuck in a car and that's all you could do. Um you never put Einstein in a car. It's like why am I stuck in a car? So there's actually probably a limit to how much intelligence you put in a car to to not have the intelligence be bored.

Uh what's XAI's plan to stay on the compute ramp up that all the labs are doing right now? The labs are g on track to spend over like fifty to a hundred billion dollars. Sorry, sorry, sorry, yeah. Corporations. Um the labs are at universities and and and they're moving like a snail. They're not sending a fifty million dollars.

That's right. The revenue maximizing corporations are making like twenty to ten billion depending on like OpenAMA is making twenty B uh revenue uh anthropics like ten B. Close to a maximum profit AI. And stay there as as things get. So a as soon as you lock uh unlock digital human. Um You you basically have access to trillions of dollars of revenue. Um so uh w the the most valuable companies currently by market cap, um, th their their output is digital. Um so

Uh NVIDIA's output is um FTPing files to Taiwan. It's it's digital. Right. Now those are very, very difficult to s yeah, those high value files. They're the only ones that can make uh files that good. Um but that is literally their output. They F T P files to Taiwan. Do they F C P them? Um I I believe that is the file file file transfer protocol I believe is

is is is I could be wrong. Uh but either way, it's a bunch of it's a bit screen going to Taiwan. Um you know, Apple doesn't make phones. They uh they send files to China. Um Microsoft doesn't doesn't manufacture anything. Uh even for Xbox that that that's outsourced, they again it's it they set their output is digital. Uh Meta's output is digital. Google's output is digital. Um so if you have um human emulator, uh you you can basically create um

one of the most valuable companies in the world overnight. Um and you would have access to Trillions of dollars of revenue. It there's this it's it's not like a small amount. Okay, I see. You're you're saying basically like revenue figures today are just like so like they're all rounding errors compared to the actual TAM. So just like focus on the TAM and how to get there. I mean if you take something as as as simple as say customer service.

Um, if you have to integrate with the APIs of of existing corporations, many of which don't even have an API, so you've got to make one um and you've got to wade through uh legacy software. Um That's s extremely slow. Um if however, if AI can um simply take whatever is given to uh the outsourced customer service company that they already use.

um and do customer service using the apps that they already use, uh, then you you have you you you can make tremendous headway uh in in customer service, which is I think one percent of the world economy, something like that, just close to a trillion dollars all in for customer service. And and and and and there's there's no there's no barriers to entry. It just you can just immediately say we'll we'll outsource it for a fraction of a cost. And and there's no integration needed.

some kind of categorization of uh intelligence tasks. where there is breath, where customer service is done by very many people, but You know, many people can do it. And then there's difficulty where you know there's a best in class turbine engine, like presumably they're 10% more fuel efficient.

turbine engine that could be imagined by an intelligence, but we just haven't found it yet. Or, you know, GLP ones are just, you know, a few bytes of data. Where do you think you want to play in this? Is it a lot of you know, reasonably intelligent intelligence, or is it the very pinnacle of cognitive tasks? Well I was just using uh customer service as like something

That's it's a pr it's a very significant revenue stream, um, but one that is probably not super difficult to solve for. Um so uh if you if you uh can emulate a human at a at a at a desktop, um that that's just literally what customer service is. Um and um

You know, it's it's pu people of average intelligence. Not like, you know, you don't need like somebody who's who spent uh many you know, many years you you don't need like t you know, yeah um th th sort of several sigma good engineers for that. Um But but obviously as you make that work, um you can then once you have computers working, effective digital optimists working, uh you can then run any application. Um like let's say you're trying to design a chip.

So you can you could then um run c you know conventional uh apps uh you know like book you know s stuff from cadence and synopsis and whatnot. Um and you can say uh Uh you you can you you can run a thousand simultaneously or ten thousand and say, Okay, uh given this input, I get this output for the chill. Um And and at a certain point you can say, okay, I I you y you you're actually gonna know what the what the chip should look like um without using any of the tools.

Um so b basically you you you you should be able to do a digital chip design, like uh you can do chip design uh like you you march up the d difficulty curve. Um you could you you're you know, be able to do do to do CAD. Um so you know, um

You could use like sort of an NX or or any any of the CAD software to design things. Okay, so you s think you start at the simplest tasks and walk your way up the defensive curve. Um So you're saying look, as as a broader objective of having this full digital coworker, uh, emulator, you're saying, look, all the revenue maximizing corporations want to do this, you um, XAA being one of them, but we will win because of a secret plan we have.

Uh but like everybody's like trying different things with data, different things with algorithms. And I'm like this. Like what is fan. What else can we do? But uh yeah. I I th I th you know, I I I I I think we see a path to doing I I mean I think I think I know how to I think I know the the path to do this because it's it's kind of the same path that Tesla used to create self driving. Um you know, instead of driving a car it's driving a a computer screen.

So it's a self-driving computer essentially. Oh, you're saying is the path just following human behavior and trading on vast quantities of human behavior? I mean is isn't that isn't that a training? I mean I obviously I'm not gonna spell out you know, most sensitive secrets on a podcast. So you know, I'd uh I need to have at least three more Guinnesses for that. What will X AI's business be? Like is it gonna be consumer

Enterprise, what's the mix of those things gonna be? Is it just gonna be similar to Other labs where you've just You're saying loud. Corporations. Corporations D V lon. Revenue maximizing corporations to be clear. Those GPUs don't pay for themselves. Exactly. Um but yeah, what what's the business model? What what are the revenue streams in in a few years' time?

Uh uh things are gonna things are gonna change very rapidly. Like we're staying the obvious here. Um you know, I call AI the supersonic tsunami, I level iteration. Um So r really w what's gonna happen is Espe especially when you have humanoid robots at scale, um is th they will just provide they will make products and provide services far more efficiently than human corporations. So amplifying the productivity of human corporations is is simply a a short term thing.

So you're expecting fully digital uh corporations rather than like SpaceX becomes hard AI and so like I think there'll be d digital corporations but I look it is this some some of this is gonna sound kinda doomer doomerish, okay, but it it I I'm just I'm just saying what I think will happen. It's not it's not meant to be doomerish or

Anything else, just just like this is what I think will happen. Um i i is that i i is that pure AI c corporations that are purely AI and ro robotics uh will uh vastly outperform any corporations that have people in the lid. So you can you can think of say like like like computer used to be a a job that humans had. That you you would go and get a job as a computer where you would do calculations. Um

And that they'd have like entire skyscrapers full of humans, like you know, twenty, thirty floors of of humans just doing calculations. Um Now that entire uh skyscraper of humans doing calculations um can be replaced by a laptop with a spreadsheet. That spreadsheet can do um vastly more calculations than in an entire building full of human computers.

Um so you can then you think about okay, well what if only some of the cells in your if some of the cells in your spreadsheet were uh calculated by humans? Actually that that that that would be much worse than if all of the cells in your spreadsheet were calculated by the computer. And so really what will happen is uh the pure AI, pure robotics, um

corporations or collectives will far outperform any corporations that have humans in the loop. And this will happen very quickly. Speaking of closing the loop, sorry, Optimus, um uh You uh I mean a a as far as like manufacturing targets and so forth go, y your companies have sort of been like carrying American manufacturing of hard attack on their back, but in the fields that you are um, you know, test lesband dominant in, you're and now you want to go into humanoids.

In China there's entire dozens and dozens of companies that are doing this kind of manufacturing cheaply and at scale. uh and are incredibly competitive. So give us sort of like advice or a plan of how America can build the humanoid armies or if you know the EVs, et cetera, at scale and as cheaply as as as China is on track to. Well th there are there are really only three hard things for human robots. Um the the real world intelligence, um, the the hand and scale manufacturing. Yeah. Um

So uh I haven't seen any even demo robots that have uh a a great hand, like with all the degrees of freedom of a human hand. But Optimus will have that. Um Of Optimus does have that. And how do you achieve that? Is it just like right torque designating the motor? Like what is the what is the hardware bottleneck to that? Well we have to re we have to design custom custom actuators, um basically custom designed um motors, gears.

uh par electronics, controls, sensors, everything had to be designed from physics first principles. There is no supply chain for this. And will you be able to manufacture those at scale? Yes. Is anything hard except the hand from a manipulation point of view, or once you've solved the hand, are you are you good? From an electromechanical standpoint, the uh the hand is more difficult than everything else combined.

Yeah, human hand turns out to be quite something. Um but but then you also need the real world intelligence. Um so the intelligence that Tesla is developed for the car um applies very well to the robot.

um which is, you know, primarily vision in. But so the the car takes more vision, but also it actually also it is listening for sirens. It's um You know, it's it's taking in the inertial measurements, it's GPS signals, a whole bunch of other data, uh combining that with with video, it's primarily video, and then uh outputting the con um control

commands. So like like t like your Tesla is taking in one and a half gigabytes a second of video and outputting two kilobytes a second of control control output. Um with the video at uh thirty six uh hertz and the control frequency at eighteen. One intuition you could have um For when we get this Robotics stuff.

It takes quite a few years to go from the compelling demo to actually being able to use in the real world. So ten years ago, you had really compelling demos of self-driving, but only now we have RoboTaxi and Waymo and all these services scaling up. Doesn't this shouldn't this make one pessimistic on, say, household robots? Because we don't even quite have the compelling demos yet of, say, the really advanced hands. Well, we've been working on uh humanoid robots now for a while. Um

So I guess it's been what five or six years or something like that. Um And um an and a bunch of the things that we've done for the car are applicable to the robot. Um so we'll use the same um Tesla AI chips in the In the in the robot as the car.

uh we'll use it this the same basic principles. Uh it's it's it's very much the same uh AI. Um you've got, you know, many more degrees of freedom for a robot than you do for a car. Um but it r really if you just think of it for like it as as like a bit stream. Um AI is really mostly uh compression and correlation of of two butt streams. So you c you you're you know, so you if for video you've got to do a tremendous amount of compression. Um and and uh

With and you gotta do the compression just right. You gotta compress the y like ignore the the things that don't matter and and like you you don't care about the details of the leaves on the tree on the side of the road. But you care a lot about the um the road signs and the the traffic lights and the pedestrians and and even whether p you know somebody in another another car is can is looking at you or not looking at you. Like these the there's some of these some of these details matter a lot.

So if it is essentially it's it's gotta turn that the w the car has gotta turn that one and a half gigabytes a second ultimately into two kilobytes a second of control outputs. So um so m many stages of compression. Um And you've got to get all those stages right and then correlate those to the correct control outputs. But the robot has to do essentially the same thing. And you think about what what humans this is what happens with humans. We're we're really our photons in controls out.

So tha tha that is the vast majority of your your life has been vision, photons in and then motor controls out. Naively it seems like Between humanoid robots and cars. The the fundamental actuators in a car are like How you turn, how you accelerate, et cetera. Where in a in a robot, especially with maneuverable arms, there's dozens and dozens of these degrees of freedom. And then especially with Tesla, you had this advantage of like you had millions and millions of hours.

Of human demo data collected from just the car being out there, where like you can't equivalently just deploy optimists that don't work and then get the data that way. Between the increased degrees of freedom and the far sparser data. Yes. Um That's a good point.

Now you you you're you're actually you're highlighting a an important limitation uh and difference between cars. It's like we we do have well, we'll still have like ten million cars on the road. Um and so uh that that's it's it's hard to duplicate that like massive training fly flywheel. Um for for the robot, um

What we're gonna need to do is build a lot of robots and put them in kind of like an Optimus Academy so they can do self play in reality. Um so we're ac we're actually gonna we're we're actually bullying that out. So we're we can have at at least ten thousand Optimus robots, maybe twenty or thirty thousand that can do that that are doing self-play and and and testing different tasks.

The the Tesla um has quite a good uh reality generator. Uh like a physics accurate reality generator that we we we made made this for the cars. We'll do the same thing for the robots. And um actually have done that for the robots. Um so uh so you you have Yeah, uh a few tens of thousands of human or robots.

uh doing different tasks. And then you've got you you can do millions of simulated robots in the simulated world. And you use the uh the tens of thousands of of robots in the real world to close the simulation to reality gap. Close the symptom gap. How do you think about the synergies uh between XAI and Optimus, given you're highlighting, look, you need this world model, you maybe want to use some s really smart intelligence as the control plane.

Um and so maybe Grok is like doing the slower planning and then like the motor policy is a little lower level. Yeah. W what will the sort of synergy between these things be. Yeah. So you'd use Grok would orchestrate the behavior of the Optimus robots. So let's let's say you wanted to build a a factory. Um Uh then Optimus uh then gr Groc could uh organize the Optimus Robots, uh give them assign them tasks, uh

To build the factory for to produce whatever you want. Don't you need to merge XAI and Tesla then? Because these things end up so What are we what were we saying earlier about public company discussions? Well we're one more guinness in, Elon. Wha what are you waiting to see before you say we want to manufacture a hundred thousand optimists? Is it like Optimi.

Since since we're defining the f the the proper noun, we could define the the plural of the proper noun too. So we we' we we're gonna proper noun the the plural and so it's optimized. Okay. Is there something on the hardware side you wanna see? Do you wanna see better actuators? Or is it just you want the software to be better? What what are we waiting for before we get like mass manufacturing of gen three?

No, we're moving towards that. Like we're we're moving forward with mass fan factory. But you think current um current hardware is good enough that you are gonna you should you just wanna deploy as many as possible now? I mean it's very hard to scale up production. I said. Uh but uh yeah, but I I think Optimus 3 is the the right version of the robot to uh

Maybe something on the order of like a million units a year. I think you'd want to go to Optimus four before you went to ten million units a year. Okay, but you can do a million a year at Optimus three. Uh yeah. I mean it's very hard to spoil it manufacturing. Yes. Um So like manufacturing uh The the output per unit time is always followed as an S code. Yeah, so it starts off agonizingly slow.

then it has this sort of exponential increase, then a linear, then a then a you know, logarithmic uh outcome till you you you sort of eventually asymptoted some number of but optimist initial production will be it's it's gonna be a It's gonna be a stretched out S curve because so much of what goes into Optimus is brand new. There's not an existing supply chain. Um as I mentioned, the the actuators, electronics, everything in the Optimus robot is designed

Um from physics first principles, it's not it's not taken from a catalog. These these are custom designed everything. Literally everything. I I don't think there's a single thing that down. How far down the does that go? I mean I guess we're not making custom capacitors yet, maybe. Um but um But there's there's there's nothing you can pick out of a catalog um f at any price. Uh so so j it just means that the the the optimus S curve.

Uh the the the the units units per out your output per unit time, y how many outputs robusts you make per per day, uh whatever, is is gonna initially wrap uh slower than a product where you have an existing supply chain. Um but it will get to a million. When when you see these Chinese humanoids uh like Unitri or whatever, ha ha sell humanoids for like six K or thirteen K. Do you just like it?

D are you hoping to get your Optimus's bill of materials below that price so you can uh do the same thing, or do you just think qualitatively they're not the same thing? Like what what do you think is going like what allows it what allows them to sell for solo and can we match that? Well Op Optimus our Optimus is is assigned to have a lot of intelligence, um and um to have the same elec electromechanical dexterity, if not higher than a human. So Yeah tree does not have that.

I mean it it's it's quite a it's quite a big role. It's just it's'cause it's man it has to do uh you know, carry h heavy objects for long periods of time, um, and not overheat or exceed the power of its actuator. So um So we've got- we've got- we've got You know, a b it's it's five eleven, you know, so it's pretty Tall. Um and It's it's it got a lot of intelligence. So it's gonna be more expensive than

um a s a small robot that is not intelligent. But more capable. Yeah. But not a lot more. Uh I mean like the thing is o over time Uh as Optimus robots build Optimus robots, the the cost will drop uh very quickly. And what will these first billion Optimuses Optimi do? Like what will their highest and best use be? I I think that you you would start off with with simple tasks that you can count on them doing well. But in the home or in factories, like the best useful

Um robots in the beginning will be anything any um continuous operation. So any twenty four by seven operation'cause then you're cause they could they can work continuously. Yeah. What fraction of the work at a gigafactory that is currently done by humans could a gen three do? Um I'm not I'm not sure, maybe it's like ten, twenty percent.

Maybe more. I don't know. That's uh the w we we would we're but we would use we we we would not like reduce our headcount. We would we would uh for sure. Right. Um but but we would increase our output. So th the the um units produced per human, like total So total number of humans at Tesla will increase, but the um the output of robots and cars will inc w will increase disproportionate like much much to you know uh

N number of cars and robots produced per human will increase dramatically. But but number of humans will increase as well. We're talking about Chinese manufacturing um a bunch here and um We're also talking about, you know, we've talked about some of the policies that are relevant, like you mentioned, the uh the solar tariffs. Yeah. Uh and you think they're a bad idea because, you know, we can't uh scale up solar in the US. Well just the electricity output in the US uh needs to scale up.

Yeah. But uh where I was going with this is if you were in charge, if you were setting all the policies, what else would you change? Um Yeah, I I would say

Any anything that is a limiting factor for electricity, um, needs to be addressed, provided it's not like very bad for the environment. So presumably some permitting reforms and stuff as well will be in there. Yeah. There there's a fair bit of permitting reforms that are happening. A lot of the permitting is uh state based, so um But an anything but but but this this administration is is good at um removing how many uh roadblocks.

And I'm not saying all tariffs are bad, I'm just saying'cause of solar tariffs. Yeah. So yeah, yeah. I mean sometimes if Like if another country is subsidizing the output of a of something um then you have to have countervailing tariffs to uh protect domestic industry against uh subsidies by another country. What else would you change?

I don't know if there's that much that the government can actually do. Well w one thing I was wondering is it seems like the for the policy goal of creating a lease for the US versus China, it seems like the export bans have actually been quite uh impactful where China's not producing leading edge chips and the export bands really bite there. China's not producing uh leading edge turbine engines. And similarly there's a bunch of export bands that are relevant there on some of the metallurgy.

Should there be more export bans? Like if you think about things like I mean, there are now the drone industry and things like that, but is that something that should be considered? Well, I I think it's important to appreciate that in most areas China is very advanced in manufacturing. Um there's only a few areas where it is not. Uh

The you know, Ch China is a manufacturing powerhouse next level. Like people don't yeah, most people impressive. Yeah, yeah. I mean the if you if you take like refining of of ore, um I'd say roughly China uh ref does more does twice as much ore refining of a of of uh on average as the rest of the world combined. Um And and I think there's there's some areas like say refining gallium which goes into solar cells. Um I think they're at like ninety eight percent of gallium refining.

Um so so China is actually very advanced in manufacturing in in I'd say most areas. It seems like we're like there is discomfort with this supply chain dependence and yes, nothing's really happening on it. Supply chain but which supply chain depends. It depends on say like the gallium refining that you're saying. Yeah, yeah. There's there's there's there's a there's a all the rare rare earth stuff and

Yeah. Rare earths which are as as you know, not rare. Yeah. Like we actually do rare earth ore mining in the US. Send the the the rock. Uh put it on on a on a train and then put on a boat to China that goes on another train and goes to the um wherever it's refining uh refiners in in China who then refine it, put it into a magnet, put it into a a motorcycle assembly, and then send it back to America.

So the thing w we're really missing a lot of of all refining um in in America and Isn't this worth a policy intervention? Yes. Uh well I think there are some things being done on on that front. Um but but We we kind of need Optimus, frankly, to to build uh oral refineries. Um you think the main advantage China has is the abundance of skilled labor? And that that that's like th that's the thing Optimus fixes.

But also we need the times about like four times our population. So we need so I mean th there's this concern if you think like human ears are the future that like Okay, right now, if it's the skilled labor for manufacturing that's determining who's at who can build more humanoids, you know, China has more of those. It manufactures more humanoids, therefore it gets m the it gets the optimized future first.

Um and it just like keeps that exponential going. It seems like you're sort of pointing out that sort of getting to a million Optimi requires the manufacturing that the Optimi is supposed to help us get to, right? You you can you can close that recursive loop pretty quickly. With a small number of Optima. Yeah. So you c you close the recurs recursive loop um to help help the robots build the robots. Um and then we we can, you know, try to get to tens of millions of units a year.

Maybe if you start getting to hundreds of millions of units a year. I I think you you you're you're gonna be the most competitive country by far. We we definitely can't win with just humans because China has four times our population. Right. And frankly America's been winning for so long that we you know, just like a

like a pro sports team that's been running for a very long time, tend to get complacent and entitled. Um and that's why they stop winning. Um because it's you know, don't work as hard anymore. Uh so I think the frankly just my observation is the average work ethic in China is higher than in the US. So it's not just that there's four times the population, but the work the the amount of work that people put in is higher.

Um so y y you can like you can try to rearrange the humans, but you're still one quarter of the uh you know, assuming that that productivity is the the h health c is is the same, which I think actually it might not be. I think China might have an advantage on productivity per person. Um we will do f one quarter of the amount of things as China. Um so so we we can't win on the human front. Um and our birth rate's been low for a long time, so uh

Bi b uh birth rate's been the US birth rate's been below replacement uh since uh roughly nineteen seventy one. Um So so we we've got a lot of people retiring or f you know, mo more people dying than than than than w we're close to sort of more people domestically dying than than being born. Um

So we definitely can't win on the human front, but we we might have a shot at the robot front. Are there other things that you have wanted to manufacture in the past, but they've been too labor intensive or too expensive that now you can come back to and say, Oh, we can finally do the

Whatever. Uh because we have Optimus. Yeah, I think we'd like to do more build more um or refineries at Tesla. So um we just completed um construction and have um begun lithium refining um without lithium refinery in Corpus Christi, Texas. Uh we have um a nickel refinery which is for the cathode uh that's here in Austin. Um and uh

These are these are the largest this is the largest cathode re this is the largest cathode refinery, largest lithium refinery, and largest Nickelang and and lithium refinery uh outside of China. Um And uh Yeah, the the Cather team would say like we have uh the the largest and the only actually uh Cather refinery in America. Many supernatures. Not just the largest, but it's also the only. So th it was pretty big, e even though it's the only one.

Um but I mean there are other things that uh you know um You you you you could do a lot more refineries and and um help the the help America be more competitive on r refining capacity. So so there's there's like Th there there's basically a lot of work for for the Optimite to do, uh that that most Americans very few Americans practically want to do. Uh I mean I've

I'm actually Is the refining work too dirty or what's the It it's not it's actually no, we we we don't um there's not we don't have toxic emissions from the refinery or anything. Um Like the Catherine Refinery swipe right sort of in in Travis County, like five minutes from top of the why can't you do it with humans? No, you you you can't you find out of humans. Ah, I see, okay, yeah.

L like no matter what you do, you you have one quarter of the number of humans in America than China. So if you have them do this thing, they can't do the other thing. So so then then um well how do you how do you build this refining refining capacity? Well you can do it with Optimi. Um and um It but not many not very many not very many Americans are are pining to do refining. I mean how many have you run into? Very few.

What if you're planning to refine? You know, BYD is reaching Tesla production or sales in quantity. What what do you think happens in global markets as Chinese production in EV scales up? Uh China's extremely competitive in manufacturing, so um I think this there's gonna be a f a massive flood of Chinese vehicles and and and uh and m other qu

basically w most manufactured uh things. I mean a as it is, as I said, like China's like probably just twice as much refining as the rest of the world combined. So if you go You know, if if if if you just go down to like fourth and fifth tier uh supply chain stuff. Like like like in the baseline we've got energy. Then you then you've got mining and refining. Um those those those foundation layers uh are

Well, like I said, China as a rough guess, China's doing twice as much refining as the rest of the world combined. So any given thing is gonna have uh co Chinese content because China's doing twice as much manufacturing refining work as the rest of the world. And uh and then they'll they'll but they'll go all the way to the finished product with the cars.

Uh I mean China's a powerhouse. I mean, I think this year China will exceed three times US electricity output. Mm-hmm. And I like electricity output is a is a reasonable proxy for uh You know, for the economy.

Uh so like in order to run the factories and run run everything you need electricity. So electricity is is is a it's a good proxy for the for for the real economy. Um and so if China is if if if China passes three times the U U S electricity output, it it means that it it's industrial capacity.

as a rough approximation is three times that will be three times that of the US. Reading between the lines, it sounds like what you're sort sort of saying is absent and sort of humanoid recursive miracle in the next few years. on the the sort of like whole manufacturing energy, uh raw materials chain, like China will just like dominate whether it comes to like AI or manufacturing EVs or manufacturing humanoids. breakthrough innovations uh in in the US, uh China will uh utterly dominate.

Interesting. Yes. Robotics being the main breakpoint innovation. Well if you do Th like to to scale AI uh in in in space, like like basically need you need need the humanoid robots. You need real world AI. You need um a million tons of year two orbit. Um like let's just say like if we if we get the mass driver on the moon going, my favorite thing. Um th th then I think uh we'll have solved all our problems. Yeah. So this is like

I call that winning. I call that winning. Big time. You can finally be satisfied you've done something. Yes. You have the mass driver on the moon. That's right. Uh well actually the uh there's there is a Highland book, The Moon The Moon is a Harsh Mistress. Okay, yeah, but that's slightly different. That's a gravity No, they have a TMS driver on the U.S. Exactly. What are your plans for the mass driver on the moon? That book is a hoosh I found that book much better than um

His other one that everyone reads, um Stranger in a Strange Land. Yeah, Groc Groc comes from Stranger in a Strange Land. Yeah, yeah. But I much preferred Yeah, Strange it the first two thirds of Strange and Strange Land are are good and then it gets very weird in the third third uh portion. Yeah.

Um but there's still some good concepts in there. Yeah. One thing we were discussing a lot is kind of your system for m managing people. Like you interviewed the first few thousand of SpaceX employees and I've seen with lots of other companies. What is it? Well yes, but what what doesn't scale?

Me. Sure, sure. I mean that, but like what are you looking for? It literally is not enough hours in the day, it's impossible. But but um w what are you looking for that's someone else who's good at interviewing and hiring people? What's a genus class? Um, well, at this point I think I've got, um...

I'm I might have more training data on evaluating technical talent, especially, but talent of all kinds, I suppose, but uh technical talent especially, um given that I've done so many technical interviews and then seen the results. Technical interviews seen the results. So my um my training set is is is very is enormous and uh has a very wide range. Um Uh the the generally the thing I ask for are uh bullet points uh for uh evidence of ex of exceptional ability.

So it's uh but like it's it's and th these things can be like pretty off the wall. It doesn't need to be uh in the in the domain, the specific domain. But evidence that uh evidence of exceptional quality. Um So if some if if somebody can like cite like even one thing, but let's say three things where you go, Well, well, well.

Then that's that's a good sign. Aaron Powell But why do you have to be the one to determine that presentation? No, I don't. I can't be. It's impossible. Right. Right. But in the early days, what was it the uh that you were looking for that couldn't be delegated in those interviews?

Well, I I guess I I need to build my training set. So it's not like I'd bat a thousand here. Um I would make mistakes. But then I'd be able to see where I I thought somebody would work out well but they didn't. And and then why why do they not work out well?

And I w what can I do to I guess RL myself to uh in the future um have a better batting average when interviewing people. Mm-hmm. So and my my batting average is still not perfect, but it's it's very high. What are some surprising reasons people don't work out?

Surprising reasons. Um they don't understand technology, et cetera, et cetera. But like No you you've like you've you've got like the long tail now of like, ah I was really upset about this person. It didn't work out. Curious why that happens. I mean generally when I tell people

or tell myself, I I guess aspirationally, um, is don't look at the resume, just believe believe your interaction. Mm-hmm. So if the resume may may seem very impressive and it's like, wow, you know, look resume looks good, but if the if the conversation uh after twenty minutes is is that conversation is not wow. Um you should believe the conversation, not the re not the not the paper. I feel like

Part of your method is that, you know, there was this meme in the media a few years back about Tesla being a revolving door of uh executive talent. Whereas actually I think when you look at it Tesla's had a very consistent and internally promoted executive bench over the past few years. And then at SpaceX, you have all these folks like Mark Jankosa and Steve Davis and Steve Davis runs boring company these years. No, no, yeah, yeah. But uh uh Bill Riley and folks like that.

And it feels like part of has worked well is having very capable technical deputies. What do all of those people have in common. Uh well so the I mean it tells us it's a sort of senior team. Uh at this point it's probably got average tenure of t like ten or twelve years. It's quite quite tenure. Yeah. Um so um But There there were times where Tales went through extremely rapid an extremely rapid growth phase.

Um and so it it was somewhat things were just somewhat sped up. Um and and when a company as as I'm sh as you know, company goes through different orders of magnitude of of size, you y you know Uh people that could who who could help manage, say, a fifty person company versus a five hundred person company versus a five thousand person company versus a fifty thousand person company.

Yeah, it's it's just not the same team. Yes. It's not it's not always the same team. So if if a company is growing very rapidly, the the rate at which uh executive positions will change will also be proportionate to the si the the rapidity of the growth. Um then uh Teza Tesla had uh a further challenge where when when Tesla had very successful periods, um uh w we would be um relentlessly recruited from, um, like relentlessly. Um Like when Apple had their el electric car program

They were carpet bombing Tesla with recruiting calls. It was uh uh and internets just unplugged their phones. Like it's just it's just I think it's I'm trying to get work done here. I yeah, I if I get you know one more call from an Apple rec recruiter. Um

But but they were th they were th they were they're opening off without any interview with me like double the c conversation at Tesla. Um so So so b so so uh so we had a bit of the Tesla Pixie Dusk uh thing where it's like, oh, if you hire a Tesla executive, you're suddenly you're gonna

everything's gonna be successful. Um and and I I've fallen prey to the Pixie Dust uh, you know, thing as well, where it's like, oh, we'll hire someone from Google or Apple and they'll be immediately successful, but not that that's not how it works. Um Yeah, people are people. It's it there's not like magical pixie dust. Yes. So w when we'd have the pixie pixie dust problem, um w we'd get relentlessly recruited. Um and um

And and then also being Tesla being um engineering, especially being primarily in Silicon Valley, uh it's it's easier for people to just like they don't have to change their life very much. They can just get you know The the two is gonna be the same. Yes. Um so how do you prevent that? How do you prevent the pixie dust effect where everyone's trying to poach all your people for? Um I I don't think we can pr I don't think there's much we can do to to to yeah stop it.

Um but that that's like that's one of the reasons why it tells uh uh but th really being at Silicon Valley um and uh And and having the Pixie Dust thing at the same time, um meant that uh there was just a very, very aggressive recruitment. Uh Austin, yeah. It it still helps. Uh I mean T Tesla still has a majority of its engineering in California. Um so um

Th the you know, f for getting eng engineers to move, uh I call it the significance significant other problem. Yes. So when others have jobs. Yeah. Yeah, yeah, exactly. So um For Starbase, that was particularly difficult. Yes. Since the odds of, you know, finding another space extra... Grandsville, Jacksonville. Yeah, it's quite quite difficult. I mean it's like a technology monastery, so I think.

Um you know, remote and mostly cute. But again if you go much of an improvement over SF. But if you go back to these people who've really been very effective in a technical capacity at Tesla, at SpaceX and and those sorts of places. What do you think they have in common other than

Like, is it just that they're very sharp on the, you know, rocketry or the, you know, the technical foundations? Or do you think it's something organizational? It's something about their ability to work with you? Is it their ability to like be You know, flexible but not too flexible. What makes a good sparring partner for you? I don't think it was a sparring partner. I I mean I if if somebody gets things done I um I I love them and if they don't I hate

So it's pretty straight straightforward. It's not like some idiosyncratic uh thing. Um if somebody executes well, um I'm a huge fan and if they don't, I'm not. Um but it's it's not about mapping to my idiosyncratic preferences. I'll certainly try not to have it be mapping to my idiosyncratic preferences. Um so yeah. Um Yeah. But I the uh generally I I think it's a good idea to hire for um c uh talent and drive and trustworthiness. Um

Can I I think uh goodness of heart is important. Um I I'd awaited that at at one point. Um so like are they are they a good person, trustworthy, uh Sort of smart and talented and hardworking? If so, you can add domain knowledge. But those those fundamental traits, those fundamental properties, you cannot change. So mo most of the people who um are at uh Tesla's and SpaceX did not come from the aerospace industry or the auto industry.

What is most that to change about your management style as your companies have scaled from hundred to a thousand to ten thousand people? You're you know, you're known for this like very micromanagement, just getting into the details of things. Nanomanagement, please. People management. So you're saying keep going.

We're all the way down to Heisenberg's in Sunday for school. Yeah, well how how do you I mean are you are you still able to get into details as much as you want? Would your companies be more successful if you could if they were smaller? Like how do you how do you think about that? Well because I have a fixed amount of time in the day, uh my time is necessarily d um diluted as things grow and as the span of activity uh increases. So

You know. Um it it it it it's it's impossible for me to actually be a micromanagement because uh th there's that that would That would imply I have some like thousands of hours per day. Uh it is it is a logical impossibility for the m for the to mic to micromanage things. Um So now there are times when um I will drill down into uh a specific issue because that s specific issue uh is the limiting factor on uh the progress of the company.

Um And um but the re the reason for drilling into that that some very detailed item is because it is the it is the limiting factor, not it it's not arbitrarily dig drilling into you know, tiny tiny things. Um and and like I said, obviously, from a time standpoint, it is physically impossible for me to arbitrarily uh go into tiny things that don't matter. And that would and and that would result in failure. But sometimes the tiny things um are decisive in victory.

Famously you switched the uh starship design from composites to steel. Yes. And you made that decision. Like that wasn't a, you know, people were going around like, oh, we found something better, Boss. Like that was you encouraging people against some resistance. Can you tell us how you came to that whole concept of steel switch? Uh yeah, so uh desperation I'd say. Um the um

Originally, yeah, we were we were gonna make starship out of uh carbon fiber. Um and um carbon fiber is pretty expensive. Like the the the the

You know, you can generally uh uh when you do volume production, you can get any given thing to be to start to approach its material cost. The problem with with carbon fibers is that material cost is still very high. Um Um so Um it's about it's about fifty times uh but particularly if you go for a high strength specialized co uh carbon fiber that can handle um cryogenic oxygen, it's it's it's like called roughly fifty times the cost of steel.

And at least uh in in theory it would be lighter. People generally think of steel as being heavy and carbon fiber as being uh light. Um and for room temperat room temperature applications, um You know, like say uh uh more or less room temperature applications like a Formula One car, uh static aerostructure or any any kind of aerostructure really, uh is is gonna you you're gonna probably be better off with uh carbon fiber.

Um now the problem is that w we were trying to make this enormous rocket out of column fiber and uh our progress was extremely slow. And it had been picked in the first place just because it's light.

Yes. Um but like at first glance, um like most people would think that the the the choice for making uh something light would be carbon fiber. Um the um Now now the thing is that um we When you make something very enormous out of carbon fiber and then you try to have the carbon fiber um be efficiently cured, meaning not not room temperature cured because th like the the you've got p you know, sometimes you got like fifty plies of

of of carbon fiber and and carbon fiber is really carbon string and glue. Um and uh and you in order to have um high strength you need uh an autoclave, so something that that can that's essentially a high pressure oven. And if if um if you have something that's uh a gigantic Uh the album's gotta be bigger than the rock one. Um so we're trying to make the the an autoclave that's bigger than any autoclave that's ever existed.

Uh or do room temperature cure, which takes a long time and ha and has issues. Um but th but the fundamental issue is that we're just making very slow progress uh with uh with carbon fiber. Um So um I I think the meta question is uh why it had to be you who made that decision. There's many engineers on your team. Yeah, how did the team not arrive at the steel? Yeah, exactly. Like i this is a part of a broader question of like understanding your comparative advantage at your companies.

Um so th th it was because we were making s very slow progress with with Calvin Fiber, I was like, Okay, we we've gotta try something else. Now for the Falcon Line, the the primary airframe is made of aluminum lithium. Which is a very, very good strength weight. Um and um

Actually it has uh about the same, maybe maybe better strength weight for its application than carbon fiber. But aluminum lithium is very difficult to work with. In order to weld it, you have to use something called friction still welding, where you join the you you join the metal without it entering the liquid phase. Um so it's kinda wild that you could do that. Uh but with a this particular type of welding you can do that. Um

But uh it it's very difficult to like say, let's say you want to make a modification or attach something to um aluminum lithium. You you now have to use a mechanical attachment with seals. Um you can't uh weld it on. Um So I I we want to I want to avoid using aluminum lithium for the primary structure for uh for starship. Um and uh

uh common fiber that that had, you know, ver very good mass properties. So with rock rocket you're really trying to maximize the percentage of the of the rocket that is propellant. Minimize the the the mass, obviously. And um the but like I said we're making f very slow progress. Um and and and I s I said at this rate we're never gonna get to Mars. So we've got to think of something else. Um

I didn't want to use aluminum lithium because of the difficulty of friction still welding, um, especially doing that at at at scale was hard enough um at 3.6 meters in diameter, let alone at nine meters or above. Um then um I said, well what about steel? And and so the the now I I ha I had a clue here because some of the early um US rockets had used very thin steel. The Atlas rockets had used a steel balloon tank. Um

So it's not like it's still never been used before. It had actually had been used. Um and when you look at the pr at the material properties of stainless steel, um, especially uh very uh if it's been s uh very uh like full hard uh strain hardened stainless steel, uh at cryogenic temperature, uh the the strength weight is actually similar to carbon fiber.

So if you if you look at the material if so if you look at the material properties at room temperature, um it looks like the steel is uh it's gonna be twice as heavy. But if you look at the material properties at cryogenic temperature of full hard steel, sta stainless of of particular grade.

uh then i the the you actually get to a similar strength weight as carbon fiber. And the and in the case of Starship, both the fuel and the oxidizer are cryogenic. So for For uh Falconine, the fuel is uh rocket profile grade kerosene, basically pure like a a very pure form of jet fuel. Um, which is but but but that is that is roughly room temperature. Um although we do actu we do actually chill it slightly below

We would chill it like a beer. Um but but it's not cryogenic. In fact, if we made it cryogenic, uh would it would just turn to wax. So um but but for Starship, the it's liquid methane and and liquid oxygen. They they uh they are liquid at s at similar temperatures. Uh so

Uh so basically uh almost the entire primary structure is at cryogenic temperature. So then you've got st uh uh a 300 series stainless that's f that's um strain hardened. Uh because it's at cry at almost all things at cryogenic temperature, actually has a similar strength to weight as a carbon fiber. But costs uh fifty times less than raw material and is very easy to work with. You you can weld stainless steel outdoors.

Uh y you could smoke a cigar while welding stand steel. It's it's like it's it's very resilient. Um you you can modify it easily. It's it's uh if you wanna if you wanna attach something, you just weld it right on. So um v very easy to work with, vr uh very low cost. Um and um And like you said, at cryogenic temperature, similar strength to weight uh to Um, then when you factor in that uh that we don't need we don't we we we have a much reduced uh heat shield mass.

uh because the melting point of steel is much greater than the melting point of aluminum. Um It's about twice the melting point of alumin aluminum. And so you can just run the rocket much hotter? Yes. So especially for the ship, uh which is coming in like a flat a b a blazing meteor, uh it is uh the y you you can greatly reduce the mass of the heat field.

Um so the so you you can c call it cut the mass of the windward uh part of the heat shield in in maybe in half and you don't need any heat shielding on the on the leeward side. Um so um the the net if net result is actually the steel rocket weighs less than the carbon fiber rocket. Because the the the resin in the carbon fiber rocket uh uh s it um starts to melt. Um Does so so you s basically carbon fiber and alu aluminum have about the same operating temperature uh capability.

Um and whereas steel can operate at at twice temperature. I won't go to Rocket Base. Shut up, assholes. Okay, but to play this back to you, what I'm hearing is the Steel was a riskier, less proven path. Other than the early US rockets versus carbon fiber was like a worse but more proven out path. And so you need to be the one to push for, hey, we're gonna do this riskier path and just figure it out. And so you're fighting like a sort of conservatism in a sense.

Um that's why I I initially said like that the issue is that we weren't making fast enough progress. We we were having trouble making even um a small barrel section of the carbon fiber um that didn't have wrinkles in it. Um so

Uh'cause a at at at that large scale you have to have many plies, many sort of layers of of the column fiber. Um you've got to cure it and you've got to cure it in such a way that it it doesn't um ha have any wrinkles or or or defects. The column fiber is uh much less resilient. th than than steel. It has much l it's it's less toughness. Um like like stainless steel will will scratch and and and and bend. The column fiber will will tend to shatter. Um so um

So toughness being the area under the stress strain curve. Um, so that you you're generally gonna have to do better with steel. Um The stainless steel to be precise. One other Starship question. Um So I visited um Starbase, I think it was two years ago, um with Sam Teller and that was awesome. It was very cool to see in uh a whole bunch of ways. One thing I noticed was that people really took pride in the simplicity of things, where, you know, everyone wants to tell you how

Starship is just a big soda can and you know, we're hiring welders and you know, if you can weld in any industrial project you can weld here, but um there's a lot of pride in the simplicity. And Well cycle Starship is a very complicated rocket. So that's that's what I'm getting at. Is are things simple or are they complex? Uh I think maybe they're just w what they're trying to say is that you know you you don't have to have like prior experience in the rocket industry to work on a Starship.

Um yeah, s something just needs to be they know smart and work hard um and be trustworthy and they can work on a rocket. They don't they don't need prior rocket experience. Starship is is the most complicated machine ever made by humans. Both by a long show. In in what regard? An anything really. It's uh there there isn't a more complex machine. Um

Th there yeah. I mean I I I'd say that there's there's pretty much any any project I can think of would be easier than this. Um and and that's why no one has made a rapidly reusable or no nobody has ever made a reus fully reusable over the rocket. It's a very hot very hard problem. Um Th the I mean, m many smart people have tried before, very smart people, with immense resources and they failed. Um, so and we haven't succeeded yet.

Uh we've w you know, Falcon is partially reusable, but the up to stage is not. Um Starship version three, I think this design That it it can be fully reusable. And that full reusability is what uh will enable us to become a multiplanet civilization. Can you say about the scrolls? So the I don't I I'm like I like I said I can't I We we spent a lot of time on bottlenecks. Can you say what the current starship bottlenecks are, even at a high level? I mean trying to make it not explode.

Generally at all chestnut really wants to explode. Um those combustion. Um so it takes like one mistake and and I mean the amount of energy contained in in a starship is insane. And so is that why it's harder than Falcon? It's because it's just more energy? It's a lot of new technology. Um it's it's push it's pushing the performance envelope. Um the the Raptor three engine is uh

Very, very advanced engine, by far the best rocket engine ever made. Um but it desperately wants to blow up. I I mean, just to put things in perspective here, on Liftorf, um, the the rocket is generating over a hundred gigawatts of power. twenty percent of US testing side. It's a great comparison. While not exploding. Sometimes. Sometimes but sometimes yeah. So I was like, how does it

not explode. There's a there's a you know thousands of ways that it could explode and and only one way that that that it doesn't. So so we want it to merely not not merely not explode but but fly reliably uh Yeah, on a daily basis, like once per hour. And obviously it blows up a lot. It's it's very difficult to maintain that for cadence. Yes. Um

And and then I but I say like w like what's the what's the single biggest remaining problem for Starship? It's uh uh having the heat shield be reusable um that such that the no no one has ever made A reusable orbital heat shield. Um so the the sh the shi the the heat shield's gotta make it through the ascent phase without shucking a bunch of tiles.

Um and then it's gonna come back in and also not lose a bunch of tiles and or or overheat the the main the main uh airframe. Isn't that hard because it's kind of fundamentally a consumable Uh well, yes, but your brake pads in your car are also consumable, but they last a very long time. Fair. So it just needs to last a very long time. Um But but it lost a lot of tiles, you know, and uh you know, it was

Uh n not reusable without a lot of work. Yeah. So even though it did land d it did come to soft landing, it was n would not have been reusable without a lot of work. Um and and that so it's not really reusable in that sense. So that's that's the biggest problem that remains is fully reusable heat shield. Um so like you want to be able to land it, uh refill propellant and fly again.

Uh without good you know y you can't go do this laborious inspection of you know forty thousand tiles type of thing. It seems like you're just able to drive the sense of like urgency and drive the sense of like this is the this is the thing that can scale. Um and I I'm curious why you think other organizations of your like SpaceX and Tesla are really big companies now. And you're still able to keep that culture, what goes wrong with other companies such that they're not able to do that?

I don't know. Um But like today you said you had like a bunch of SpaceX meetings. Like what what what is it that you're doing there that's like keeping that that's adding urgency. Yeah, yeah. The urgency's gonna come from whoever's leading the company. So if my sense of urgency I have like a maniacal sense of urgency, so

that maniacal sense of urgency projects through the rest of the company. Is it because of consequences? They're like, if you know, Elon said a crazy deadline, but if I don't get it I know what happens to me. Is it just you're able to identify bottlenecks and get rid of them so people can move fast? Like how do you how do you think about why your companies are able to move fast? Yeah, I'm constantly addressing the limiting factor.

Uh I mean it f I mean on the deadlines front, I I mean I generally actually try to aim for a deadline that that I at least think is at the fiftieth percentile. So it's it's not it's not like an impossible deadline, but but it's the most aggressive deadline I can think of that could be achieved with fifty percent probability. Um which means that it's over late half the time. Um and um

But but but whatever like there is like a law of gases expansion that applies to schedules. Like whatever given s whatever schedule you you ha like if you if you you said we're gonna do this uh something in like f five years, which to me is like infinity time, um It it it will expand to fully available s schedule and it will take five years. Um you know, th like th there's like this there's a physical limit.

Like that th th th like physics will limit how fast you can do certain things. Like so like scaling up manufacturing, th there's like there's a rate at which you can move the atoms. Um and and scale manufacturing. That's why you can't like instantly make, you know, a million of something, million years a year or something. Uh you've got you've got to design the manufacturing line, you've got to bring it up, you've got to ride the S curve of production. Um so

Yeah, I I I guess like like I'm trying to think what what can I say that's that's that's actually helpful to people? Um I I think generally um a manifold sense of urgency is is uh is a very big deal. Um so um And and you wanna have a you wanna you wanna have a an ag an aggressive schedule? Um And then you w and and you and you wanna figure out what the limiting factor is at any point in time and and help the team address that limiting factor. Can you make me talk about the so Starlink was

slowly in the works for many years. Uh and Yeah, we talked about it all the way in the beginning of the company. Yeah. And so then there was a team you had built in Redmond and then at one point you decided this team is just not cutting us. But again, how did you like And so why did it wh why didn't you act earlier and why did you act when you did? Like why was that the right moment at which to act? I mean I I have I have these ver very detailed um engineering reviews weekly. Um that that's that

That's maybe a very unusual level of granularity. Um I don't know anyone who runs a company or at least a manufacturing company that that goes into level of detail that that I go into. Um So so it's not it's it's not it's not as though like I have a pretty good understanding of what's actually going on. Mm-hmm. Because we we we we go

We go through things in detail. Um And I'm a big believer in skip level meetings where the individuals in instead of having the person that reports to me say things, it's everyone that reports to them um says something um in in the technical review. Um and um and and there can't be um advanced preparation. So otherwise you you you you're gonna get uh you know glaze.

Um as I say these days. Yeah, exactly. Very Gen Z of you. Very Gen Z How do you prevent an advanced administrator? And you just like call them randomly? Like wh no just go around the room and everyone uh provides an update. Uhhuh. Um so Uh I mean it's a it's a lot of information to keep in your head because um you've you've got a you've you've you've got then say if you're meeting it's weekly or twice weekly, you you you you've got a snapshot of what that person said. Um and

And and you can s and and you can then, you know plot the progress points um man you can sort of mentally pro plot the points on a curve and say, are we converging to a solution or not? Um Or or are we you know? Like I'll I'll I'll take drastic action Uh only when I conclude that.

Um, success is not in a set of possible outcomes. Um so if I w when I say okay, when I when I finally reach the conclusion that okay, un unless drastic action is done, we have no no chance of success, then I must take drastic action. That's what that's that's I came to that conclusion in twenty eighteen, took drastic action and and fixed the problem.

Com uh you know, y you you've got many, many companies and in each of them it sounds like you do this kind of deep engineering understanding of what the relevant bottlenecks are, so you can do these um reviews of people. Yeah. Um You've been able to scale it up to five, six, seven companies. Within one of these companies, you have many different mini companies within them.

We what what determines the maximum out here? Could you have like eighty companies? Eighty? No. But like you can ha you you have so many already. Um like that's that's already remarkable. By this current number, yeah. Yeah, exactly. Uh no, so um we can b barely keep one company together. No neural it it depends on situation. Um

Um I I actually don't don't have regular meetings w uh with Warren Company. So that Warring Company's s sort of cruising along. Like look if basically if something is working well and making good progress, then there's no point in me spending time on it. So uh I actually uh allocate time according to where where the where the limiting factor or the problem wh where where where are things problematic and um or where where are we pushing against uh

Like what what what is holding us back? You know, I I I focus uh risk of saying the words too many times, the limiting factor. Um So so if if basically if something's go like the irony is if something's going really well, um they don't see much of me. But if something's going badly, they'll see a lot of me.

Or not not even badly. It's it's it's like if something's a limiting factor. It's a limiting factor, exactly. It's not exactly very badly, but it's the thing that's it's it's the thing that we need to make go faster to And so when something's a limiting factor at SpaceX or Tesla, are you like Talking weekly, uh daily with the engineer that's working on it. How how does that actually work? But most things that a learning factor are um

weekly and some things are twice weekly. So the the AI five chip review is twice weekly. And and so it's every Tuesday and Saturdays. Um is is the chip review. Is it open ended in how long it goes? Usually it's it's like two or three hours. Mm-hmm. So I mean sometimes less it's

It depends on on how much air space you've got to go through. Yeah. Well that's another thing again. I'm just trying to tease out the the differences uh here'cause uh the outcomes seem quite different and so I think it's interesting to note what inputs are different. And it feels like the corporate world one, like you're saying, just the CEO doing engineering reviews does not always happen, despite the fact that that is the you know, what the company is doing. Um, but then

Time is often pretty finely sliced into, you know, half hour meetings or even fifteen minute meetings. And it seems like you hold more open ended we're talking about it until we figure it out type meetings. Yeah. Yeah, some sometimes, but uh most of them seem teem to more or less stay on time. Um so um

I mean t today's uh Starship engineering review went a bit longer, um because there were there were more topics to discuss. Um They're trying to figure out how to scale two a million plus tons to orbit per year is.

Quite challenging. C can I answer the question? So you you said about um Optimus and AI that they're gonna result in double just growth rates within a matter of years. Oh, uh like the the economy? Yeah. Um Well I think that's right. What was the point of the doge cuts if the economy is gonna grow so much? Well, I think like waste and food are not good things to have, you know. Um I I I was actually pr pretty worried about the right.

Uh I mean I I think in the absence of AI and robotics, we're actually totally screwed. Uh because the national debt is piling up like crazy. Um now our interest payments, the interest payments to national debt exceed the military budget, which is a trillion dollars. So we have over a trillion dollars just the interest payments. Um

You know, that was like I was like, okay, pretty concerned about that. Maybe if I spend some time we can slow down the bankruptcy of the United States, um, and give us enough time for the AI and robots to, you know s help solve the national debt. Uh or not help solve it's the only thing that could solve the national debt. Like we are one thousand percent gonna go bankrupt as a country and fail as a country without AI and robots.

Nothing else will solve the national debt. Um and so so we would we we'd like to well we just need we we need enough time to G we'll be AI and robot. Uh to and not go bankrupt before then. I I I guess the thing I'm curious about is when Doge starts, you have this enormous um ability to enact reform and not data normalist. Sure, sure. Uh but to totally buy your point that like it's important that AI and robotics drive product improvements, drive G D P growth.

But why not just directly go after the things you were pointing out, right? You know, like th th the tariffs on certain components or whether it's like permitting. I'm like the president. And and very hard to cut to cut to to even even to to cut things that are obvious waste and fraud, like like ridiculous waste and fraud. Um what I discovered that is it it's I extremely difficult. even to cut very obvious ways and for um from the government.

Um because th the the the government has to operate on a on like who's complaining, like if if who and if you cut off payments to fraudsters, They immediately come up with the most sympathetic sounding uh reasons to continue the payment. They they don't say please keep the fraud going. They say, you know, it's th they're like, you're killing baby fanders. And we're like, meanwhile, there's no baby fanders are dying, they're just making it up.

Um th the forces are capable of of coming up with extremely compelling, sort of heart wrenching stories that are false, but nonetheless sound uh sympathetic. And that that's what happened. Um and uh so it's like Perhaps I should have known better. In fact, I thought, well, let's take a sh let's let's let's try to cut some amount of of waste and port from the government. Maybe there shouldn't be

you know, twenty million people uh are marked as alive in social security who are inf definitely dead. And over the age of a hundred and fifteen. The oldest American is a hundred and fourteen. So it's safe to say if somebody's a hundred and fifteen and mocked as live in the social security database, um something is w there there's either a typo so like somebody should call them and say We we seem to have your birthday wr. Or or uh or or we need to mark you as dead.

Very intimidating call together. Well, so it it seems like a reasonable thing. Um and if if like say their birthday is in the future, um and they have you know, a small business administration loan and their birthday is twenty one sixty five. Um We either again have a typo or we have fraud. Um say we appear to have gotten the century of your birth incorrect. Or a great plot for a movie. Yes. Were those people getting payments? So some were getting payments from Social Security.

But but but the main fraud vector uh was to mark somebody as alive in social security and then use every other government payment system uh to uh basically to t to do fraud. Because what those other government payment systems do would do They will simply do an are you alive check to the social security database. It's a it's a bang shot. What would you estimate as like the total uh amount of fraud from this mechanism?

Um m my guess is and and other b by the way the d the Government Accountability Office has has done these estimates before. I'm not the only one who's coming out of this, you know. The GAO did analysis, a rough estimate of fraud during the Biden administration and calculated at roughly half a trillion dollars. So don't take my word for it. Take it a report issued during the Biden administration. How about that? From this social security mechanism?

does not i is very ineffective at at stopping fraud because uh It it's it's it's not like like if it was a company like like f stopping forward you've got a motivation because it's affecting the earnings of your company. Uh but the government just just they just print more money. Um so it's not uh

L like you you you need you need caring and competence. And these are in short supply at the uh at the federal level. Um Oh yeah, I'm sorry. I mean when you go to the DMV, do you think, wow, this is a bastion of competence? Um well now imagine it's worse than the DMV because it's a DMV that can print money.

So w w was it not possible? At least the state level D DMVs uh need to the states more or less need to stay within their budget or they go bankrupt. But the federal government just prints more money. Well was it not possible, cuff that if there's a catchy half a trillion of fraud? Well why why was it not possible to cut all that? Uh because when when as soon as you we did we we actually no Y you you you really have to stand back and

uh recalibrate your expectations for competence. Uh b because uh You you're you're operating in a world where, you know, you you've you've got to sort of make ends meet, like, you know, you've got to pay your bills, you gotta you know buy the microphones. Yeah, yeah, exactly. Um Uh so so you uh you you if you don't ha it's it's not like there's a a giant largely uncaring monster bureaucracy. It's not even it's a and and and and a bunch of uh

uh money computers that are just that are just sending payments. Um like one of the things that that that the dose team did there was uh and it sounds so simple uh th that that probably will say um Let's say a hundred billion, maybe two hundred billion a year. Um is simply requiring that payments from the main treasury computer, which is called PAM, it's like payment accounts master or something like that.

There's five trillion payments a year, requiring that any payment go that goes out have a payment um appropriation code, make it mandatory, not optional, and that you have anything at all in the comment field. Um because i uh do do you have to uh how recalibrate how dumb things are? But you think payments were being sent out with no appropriation code, n not cr not checking back to any congressional appropriation, and no explanation.

And this is why the the Department of War, formerly the Department of Defense, cannot pass an audit because the information is literally not there. Recalibrate your expectations. Um maybe. But uh we found that like over seven years the the social security fraud they estimated was like seventy billions over seven years, so like ten billion a year. So I'd be curious to see what like the other four and ninety billion is. Federal government expenditures are seven and a half trillion a year. Yeah.

Um how what what percentage how how competent do you think government is? The the discretionary spending there. Is like what fifteen percent? Yeah, but but it doesn't matter. Uh uh w uh s uh you know. Um disability, uh it it's there's there's a zillion government payments. Yeah. Um and and a bunch of these payments are in fact uh th they're they're they're uh block transfers to the state.

So the federal government doesn't even have the information in a lot of cases to even see know if there's fraud. Let's consider let's like reductio ad absurdum. The government is perfect and has no fraud. What is your probability estimate of that? I mean, zero. Okay. So then would you say that f ford and waste uh that the government uh is Has is ninety percent? That also would be quite generous.

But i if it if it's only ninety percent, that means that the seven hundred and fifty billion dollars a year of waste and fraud. And it's not ninety percent. It's not ninety percent effective. Aaron Powell This seems like a strange way to first principle the amount of fraud in the government. Just like how much do you think there is, and then uh Uh I I anyways, we we don't know how to do it live, but I'd be curious to like some of the things.

Yeah, but as you say it's like a little bit of a um we've really grounded down but it's a little bit of a different problem space because You're dealing with a much more heterogeneous set of fraud vectors here than we are. Yeah, but I mean y y I mean that's try f you you you you you have high competence and you try hard. Um yeah you have high competence and high caring. But still fraud is non non zero.

Um now now imagine it's at a much bigger scale. Um there's much less competence and much less caring. You know, Bank PayPal back in the day we w try to manage fraud down to about one percent of of the the payment volume. Um and that was very difficult. Took a tremendous amount of confidence in carrying to uh get fraud merely to one percent. Um now imagine that that you have an organization where there's much less caring and much less competence. It's gonna be much more than one percent.

How do you feel now looking back on um kind of politics and and doing stuff there, where it feels like from the outside in, the two you know, d two things have been quite impactful. One, the America Pack and two, um, the acquisition of of well, Twitter at the time. But Also it seems like there was a bunch of heartache. And so what's your what's your grading of the whole experience? Well, um I think I think those things needed to be done to Maximize the probability of the future is good. Um

So um but but politics generally is very tribal. Um and and it's it's very tribal. It it's and people lose their objectivity usually with politics. Like they they they generally have trouble seeing th the good on the other side or the bad on their own side. That's generally how it goes. Um I I th that I guess was one of the things that's surprised me the most is you you often simply cannot reason with people. Uh-huh. Um if they're in one tribe or the other. They they simply believe

That everything their tribe does is good and anything the other political tribe does is bad. Um and persuading them is uh uh otherwise is almost impossible. Um so Anyway, but um Those actions um acquiring Twitter. Getting Trump elected even though if you know makes a lot of people angry. Um, I think those I think those actions are good for s we're good for civilization. Um yeah, w how how does it feed into the future you're excited about?

Well um America needs to conta uh America needs to be strong enough to last long enough to um extend life to other planets and to And I get I guess AI and robotics to the point where we can ensure that the future is good. Um like on the other hand, if if if we were to descend

into um, say communism or or or some situation where the where the state was extremely oppressive, um, that that would mean that we would we might not be able to become multiplanetary. Um Yeah, we might we w the the st the state might um you know, stamp out um our progress in AI robotics. How do you feel about um Uh you know, you y y uh Optimus, Grok, et cetera, are going to be leveraged by and not just yours, any revenue maximizing company's products will be leveraged by the government.

over time. Um how does this concern manifest in what private companies should be willing to give governments what kinds of guardrails should like should you know the should um AI models be uh um me to do whatever the government that has contracted them out to do asked them to do. Um should like should shouldn't should Grok get to say like actually even if the military wants to do X. No, the Grok will not do that.

I think pro probably the biggest danger of AI well maybe the biggest danger of for for AI and robotics going wrong wrong is is government. Interesting. You know. Um I mean the the way th like like people who are opposed to corporations or or or or worried about corporations should um really worry about the most about government'cause g government is just a corporation in the limit. It's a government it it is it is it is Government is just the biggest corporation with a monopoly on violence.

Um so i i I always find it like a strange dichotomy where where people would think corporations are bad but the government is good when the government is simply the biggest and and and worst corporation. But people have that dichotomy. They somehow think at the same time that government can be good but corporations bad. And this is not true. Corporations are have better morality than the government.

You know, that's uh that that is a thing to be worried about. It's like if if the you know, should should if the government should not like if the the government could potentially use AN robotics to suppress the population. Like that is a serious concern. If you limit the powers of government, which is like really what the US Constitution is intended to do, it's intended to limit the powers of government, then then uh you're probably gonna have a better outcome than if you have more government.

So But robotics will be available to all governments, right? Yeah. I don't know about all governments. Um I mean it It's difficult to predict the like I can say, like w what what's what's the what is what's the end end point or like what is s what is many years in the future, but it's difficult to predict the the sort of path al along along that way. Um

Like if civilization progresses, AI will vastly exceed the sum of all human intelligence and and there will be far more of us than humans. Um along the way what happens? I mean i mean it seems like one thing you could do is just say, um uh you're not allowed to uh whatever government decks, you're not allowed to use Optimus to do X, Y, Z. Just write out like a policy. I mean you you I think you tweeted recently that Groc should have a moral constitution.

Um and one of those things could be that we we limit what governments are allowed to do with this advanced technology. I mean yeah. But i technically if I mean if if the politicians pass a law, uh then and they can enforce that law, then it's hard to not do that law. You know. The the best thing we can have is is it's limited government, uh, where um you know you have you have the appropriate cost checks between the Executive judicial and um

Legislative branches. I I I guess th the uh the reason I'm curious about it is this like at some point it seems like the limits will come from you, right? Like you've got the Optimus, you've got the space GPUs, you've got the You think I'll be the boss of the government. Or you will get the you will like the m I mean already it's the case with SpaceX. that for things that are crucial to the um

uh like the government really cares about getting certain satellites up in space, whatever. Like it needs SpaceX. Uh it is the pr it is the um a necessary contractor. And you are in the process of building more and more of the um uh the technological components of the future that that that will have an analogous role in different industries. And you could have this ability to like set some policy that um, you know, w is suppressing

Classical liberalism in any way. I I my companies will not help in uh in any way with that. Or you know, some policy like that. Um I I will do my best to ensure that anything that's within my control maximizes the good outcome for humanity. I think anything else would be short-sighted. Um, because obviously I'm part of humanity, so um I like humans. Um Exactly.

Probably on, probably on. Um You you you've mentioned that Dojo 3 will be used for space-based compute. I don't know if you know Twitter, but uh I know you like a lot of followers. Um how how do you just turn my secrets and I post them away? How how do you design a ship for space? Well, I I guess you wanna uh design it to be um more radiation tolerant and run at a higher temperature. Uh so you could y um

You know, roughly if you increase the um operating temperature by twenty cent in degrees Kelvin, you can cut your radiator mass in half. Um so W r running at a higher temperature is is helpful in in space. Um I mean there's various things you can do for shielding the memory and

But but like neural nets are gonna be very resilient to butt flips. So uh like most of what what happens for radiation is like random but flips. Um but like if you've got like you know, a multi trillion parameter model and you get a few butt flips, it doesn't matter. Um it's it's much like heuristic programs are gonna be much more sensitive to bit flips than um some giant parameter file. Um so I just designed it to run hot and um

I w I I think you pretty much do it the same way that that you do things on Earth, apart from make it run hotter. Um I mean the solar arrays most of the weight on the satellite. Is is there a way to make the um The GPUs even more powered ends than what NVIDIA and uh TPUs and et cetera are planning on doing that would uh you know m uh uh be especially privileged in the space-based world? Well I mean the basic math is uh Um if you can do about a kilowatt per reticle, um and then you you'd need um

You know, hundred million full reticle chips uh to do a hundred gigawatts. Yeah. Yeah. Depending on what your yield assumptions are, you know, um that that tells you how many trifts you need to make. Um but cool, you need if you want if you if if if you're gonna power have a hundred gigawatts of power, you need you know a hundred million chips running that that are running a kilowatt sustained are but for radical. Um

A hundred million ships. Uh it depends on Yeah, if if i if if you if you look at the die size of something like black vulture or something and how many you can get out of the wafer, you can get like Um on the order of dozens or less uh per wafer. So you're basically you're this is a world where if we're putting that out a w uh every single year, you're producing millions millions of wafers a month. Um that's the plan with Seraphab? Millions of wafers a month of advanced process notes.

I think the terraf's gotta do memory. It's gotta do logic memory and packaging. I I'm very curious how somebody like gets star this is like the most ma complicated thing man has ever made. And obviously you're like y if anybody's up to the task, you're up to the task. H like what do you so you realize it's a bottleneck? And you go to your engineers and like, what is the next like what what do you tell them to do? I want a million wafers a month in 2030.

What is the next like what do you that's right. Do you like call ASML? Like what is it? What is the next step? Well, w um we make a little fab. Uh and see what happens. Uh, make our mistakes at a small scale and then make a big one. Is a little fab done or is or is No, it's not done. That chat's gonna come out of the bag rule. There'll be like drones hovering over the bloody thing, you know. You'll be able to like see its f construction progress on X, right, you know, in real time. Um

So uh no, we we we I I mean listen, I don't know, w we could just flounder and fail here, to be clear. It's like not uh success is not guaranteed, but um Since uh so we wanna try to make uh you know Something like a hundred million yeah, we would if we we w we need we need if well we want a hundred gigawatts of power and a hundred hundred that chips that can take a hundred gigawatts, right? So Well it yeah, but yeah, by by twenty thirty. So then um

We will take as many chips as our suppliers will give us. I've said this to I've actually said this to TSMC and Samsung and Micron, it's like please build your more fabs faster. Um and we will guarantee you to buy the output of those fabs. Um so so that they're already like pep moving as fast as they as they can. Like it's it's not like to be clear, it's not like us

It's us plus them, you know. Th there's a narrative that the people doing AI want a very large number of you know chips as quickly as possible. And then many of the input suppliers, the fabs, but also, you know, the turbine manufacturers Are not ramping up production very quickly. No, I'm not. And the expl Yeah, and the explanation you hear is that they're dispositionally.

conservative, you know, they're Taiwanese or German as the you know story may be, and they just like don't believe the say like is that really the explanation or is there something else? Well, yeah, I mean it's reasonable t like if somebody's been in, say, the computer memory business for uh thirty or forty years. And they've seen cycles. They've seen like boom and bust like ten times. Yeah. You know, so so like that's a lot of layers of scar tissue, you know. So it's like it's like

During the boom times looks like everything is gonna be f great forever. And then then then then the crash happens and then they desperately trying to avoid bankruptcy. Um and and then there's another boom and then another crash. Are there other are there other ideas you think others should go pursue that you're not for whatever reasons right now?

Um I mean there are a few companies that are that are pursuing like uh new ways of doing chips. Um uh but they're they're just not scaling fast. I I don't even mean within AI, I mean just generally. I I'd say like people should just should do the thing that w where they find that they're highly motivated to do that thing. Mhm. As opposed to, you know, s something summing up some idea that that I suggest. Like they should do the thing that they find Mm-hmm.

But it but yeah, going back to the limiting factor use that phrase about a hundred times. The the current limiting factor that I see in the time frame, you know, in in the sort of 20. Twenty nine, twenty like in in the in the three three to four year time frame, um it it's chips. Um in in the one year time frame it's it's energy, power production, electricity.

Like it's it's not clear to me that there's enough uh um usable electricity to turn on all the ch the AI chifs that are being made. Um Wha what what's your plan to deal with that world? Well we're trying to accelerate electricity production. Um I I I guess that's that's maybe one of the reasons that um XAI will will be maybe the leader, hopefully the leader, um, is that we'll be able to turn on more chips than other people can turn on faster. Um because we're we're we're we're good at hardware.

And and and and and generally the the innovations from the corporations that must that call themselves labs, um, the the ideas tend to flow Like it's it's rare to see that there's like more than about a six month difference um between um I like the idea is

uh travel back and forth um with the people. So so I think you hit you sort of hit the hardware wall and um And then whatever whichever company can scale hardware the fastest will be the leader and so I think X VI will be able to scale hardware the fastest and therefore most likely will be the leader.

You you you joked or, you know, um were self-conscious about uh, you know, using the uh the limiting factor phrase again. But I actually think there's something deep here and if you look at a lot of things we've touched on over the course of it, maybe you kind of have a good note to end on. if you think of a senescent lower agency It would have some bottleneck and not really be doing anything about it.

Um, you know, Mark and Dreessen had the line of uh most people are willing to endure any amount of chronic pain to avoid acute pain. Uh and it feels like a lot of the cases we're talking about are just leaning into the acute pain, whatever it is. It's like, okay, we gotta

figure out how to, you know, work with steel or we've got to figure out how to run the chips in space or like we'll take some near-term acute pain to actually solve the bottleneck. And so that's kind of a unifying theme. I have a high pain threshold. That's helpful. Solve the bottom X. Yes. Um so You know, one thing I can say is like...

I think the future's gonna be very interesting. Um and um and I as I said uh the Davos have only been to especially Davos, I think it was like on the ground for like three hours or something. Um It's better to be it's better to err on the side of optimism and be wrong than err on the side of pessimism and be right uh for quality of life.

So you know, your your your your happiness will be you'll you'll be happier if you if you are are on the side of optimism rather than areing on the side of pessimism. And so I recommend erring on the same option with some. That's that. Cool. Thanks for doing this. Thank you. Oh, great stamina. Hopefully this encounters a pain in the pain tolerance.