Latest / In the Loop AI / Elon Joins SpaceX and xAI in the Biggest Merger Ever. Is Tesla Next?

Transcript

Elon Joins SpaceX and xAI in the Biggest Merger Ever. Is Tesla Next? · In the Loop AI

0:00
Josh: SpaceX has acquired slash merged with XAI. They are now one company.
0:05
Josh: Musk Industries has condensed once again. And if you are a shareholder in Twitter,
0:10
Josh: you have now owned Twitter, X, XAI, and now you are owned by a space exploration company.
0:16
Josh: This is the largest merger and acquisition in history.
0:19
Josh: No one's ever done anything at this scale before.
0:21
Josh: And the downstream implications of this are fairly huge.
0:25
Josh: The amount of value that's gonna be unlocked for public markets once this goes
0:28
Josh: public in the middle of this year is going to be a really big deal.
0:31
Josh: So throughout this episode, we're going to talk about why this matters and how
0:35
Josh: you can actually make a buck from this.
0:37
Josh: How do you participate in the upside of this merger as it relates to Musk Industries
0:41
Josh: with SpaceX and Tesla and all of the companies that are underneath the umbrella?
0:44
Josh: So Ejaz, maybe we'll start with the actual announcement from Elon himself talking
0:48
Josh: about the merger and why specifically they decided this was the right time to do it.
0:51
Ejaaz: Yeah. So this merger officially makes this company the most valuable private
0:57
Ejaaz: company in the world, $1.25 trillion.
0:59
Ejaaz: It valued XAI at $250 billion and SpaceX at $1 trillion.
1:05
Ejaaz: And if you've been watching the show, you know that there's a rumored IPO of SpaceX.
1:09
Ejaaz: So now that rumored IPO has just got a lot, lot bigger, probably going to be
1:13
Ejaaz: the biggest one that we've ever seen.
1:15
Ejaaz: And on the back of this, the idea here is that XAI investors will get roughly
1:21
Ejaaz: around 0.15 SpaceX shares per share that they invested or got from XAI.
1:27
Ejaaz: So a pretty good deal for a company that is currently already losing money.
1:31
Ejaaz: That's XAI and SpaceX is kind of like profitable on their way towards profitability.
1:36
Ejaaz: But what isn't quite clear to, at least from face value, Josh, is like,
1:40
Ejaaz: Why did this even happen? And how does a SpaceX company and a Frontier AI lab make sense to merge?
1:48
Ejaaz: And Elon does a really good job of explaining his core thesis in this post,
1:53
Ejaaz: which is simply, he's going all in on space AI data centers.
1:59
Ejaaz: Now, you and I have gone back and forth a lot on this thesis.
2:02
Ejaaz: And what we've ultimately landed is that space is the most cost-efficient and
2:07
Ejaaz: performant place to train AI models. Now, does it exist today? No.
2:12
Ejaaz: In fact, prototypes so far are very minimal with a couple of GPUs being out there.
2:17
Ejaaz: Elon is leaning all in on this. And his idea is compute is going to be the most
2:22
Ejaaz: scarce and valuable resource in the future.
2:24
Ejaaz: Why? Because you need it to train frontier AI models.
2:27
Ejaaz: It's been proven with the amount that he's investing in Colossus 2 and Colossus
2:31
Ejaaz: 3 as data centers on Earth.
2:32
Ejaaz: And if you kind of like scale this out towards like, say, a decade or decades
2:37
Ejaaz: long timeline, it kind of makes sense that the winning AI models are going to
2:41
Ejaaz: be ones that have the cheapest tokens per cost.
2:44
Ejaaz: And so his bet is it's going to be in space. Why space?
2:46
Ejaaz: Because you can harness the entire energy of the sun. It is natural cooling.
2:51
Ejaaz: And ironically, you have so much space in space.
2:54
Josh: Yeah, it's like how many megatons of mass can you get into orbit?
2:58
Josh: And that's what SpaceX is doing.
2:59
Josh: So the idea is that SpaceX has the launch capability to send AI data centers into space.
3:05
Josh: In addition to the Starlink constellation, in addition to the moon and the Mars
3:09
Josh: missions that they're planning, they have basically the highway to get AI into space.
3:13
Josh: The problem is that building AI on Earth isn't that horrific.
3:17
Josh: It's challenging, but people are able to do it.
3:21
Josh: And the cost of getting AI into orbit is far more expensive than it would be
3:24
Josh: on Earth. So people don't do it.
3:26
Josh: With the Starship program, as they're able to launch the Starship in a reusable
3:29
Josh: way, the cost per kilogram to orbit will decrease to an amount that it will
3:34
Josh: be economically unviable to build AI data centers anywhere else but outer space.
3:40
Josh: And that's the bet that SpaceX is taking, that they will be able to work with
3:43
Josh: XAI to build these data centers and get them into outer space and build the AI in space plan.
3:49
Josh: And I loved Elon's ambitious mission that he outlined in the acquisition post,
3:53
Josh: talking about ascending the Kardashev scale, which we'll get into a little bit later.
3:57
Josh: And it feels like a sci-fi futuristic mission, where actually in this post,
4:02
Josh: he describes what it looks like to build
4:05
Josh: A small base on the moon and to send satellites from the moon in order to get
4:09
Josh: to 100 terawatts of energy per year.
4:11
Josh: This unbelievable scale that he's planning to build this thing at.
4:15
Josh: And it's hard to read it and not feel like you're reading a sci-fi novel because
4:20
Josh: of how futuristic this sounds. But he's very serious.
4:22
Josh: And I think the market is reflecting that and how high this is being priced
4:26
Josh: already before it's even gone public.
4:28
Ejaaz: Do you want to know how much it'll cost currently um to send us into space roughly oh please.
4:34
Josh: Tell me because they have a rideshare program right
4:37
Ejaaz: Okay listen i would never ask a woman this but i'm gonna ask you this how much
4:41
Ejaaz: do you weigh in kilograms specifically please in.
4:44
Josh: Kilograms so i just did the math and it's about 86.2
4:46
Ejaaz: Kilograms okay so i'm around 85 so let's call
4:49
Ejaaz: it one we call 170 let's call it 170 all right that's 170 uh for those of you
4:55
Ejaaz: who aren't watching this um there is a rideshare program but a very different
4:59
Ejaaz: rideshare program that you're used to on SpaceX's website that allows you to
5:03
Ejaaz: calculate how much it is to send a particular weight into space right now.
5:08
Ejaaz: And right now, to send Josh and I to host Limitless, to be the first podcast
5:12
Ejaaz: in outer space, lower orbit space, it is $1.2 million.
5:16
Ejaaz: Now, that sounds like a lot of money. And of course, it is. We don't have that
5:20
Ejaaz: money to splurge right now.
5:21
Ejaaz: But back in the day, a decade ago, how much did that cost? How much would this
5:25
Ejaaz: have cost, Josh? Probably billions.
5:27
Josh: It would have been billions of dollars. and this is cheating because this doesn't
5:31
Josh: include life support systems which are probably going to quadruple that cost
5:34
Josh: maybe even more but if you just
5:37
Josh: pushed our bodies into a box and put it on a starship that's what
5:40
Josh: it would cost and that is several orders of magnitude more like
5:43
Josh: that would have been billions of dollars and now a single individual can
5:45
Josh: actually pay for that and as starship becomes economically viable or reusable
5:50
Josh: that price is going to go down significantly so it's not only for ai and space
5:54
Josh: there's so many other things that you'll be able to send like the starlink direct
5:57
Josh: to sell terminals that will actually replace companies like verizon and t-mobile
6:01
Josh: at&t because it will work directly with the cell phones.
6:04
Josh: And it's rumored that this year, the iPhone 18 Pro is going to have compatibility
6:07
Josh: with the Starlink Direct-to-Cell network.
6:09
Ejaaz: Yeah, so actually, I have this post from Brett Winton at ARK Invest here,
6:13
Ejaaz: where I think he really beautifully describes the synergies between the two companies.
6:18
Ejaaz: And he makes two big points. One, which we've already made, which is they're
6:22
Ejaaz: placing the big bet on compute.
6:25
Ejaaz: And Elon thinks that the number one way to get the most
6:28
Ejaaz: compute per dollar spent is going to be
6:31
Ejaaz: in space and by merging spacex the cheapest
6:34
Ejaaz: way to get mass into space um plus
6:37
Ejaaz: starlink that you mentioned earlier um with his satellite constellations to
6:41
Ejaaz: train ai grok will eventually end up becoming the best model because of scaling
6:46
Ejaaz: laws with compute so that that's the compute argument but then if you look at
6:50
Ejaaz: it from the xai perspective you know what what is the model uh got to benefit from this.
6:56
Ejaaz: Rather than thinking about it as being the frontier AI model,
6:59
Ejaaz: think about what comes downstream from that.
7:01
Ejaaz: If you put yourself in the shoes of a major enterprise on Earth that is making
7:07
Ejaaz: billions and billions of dollars a year, you are only going to want to spend
7:11
Ejaaz: money on the best AI model.
7:13
Ejaaz: Why? Because it's going to be the smartest AI model, which is going to help
7:16
Ejaaz: you end up making the most money.
7:17
Ejaaz: It's kind of like a vector towards projecting how much money you make as a company,
7:21
Ejaaz: no matter what industry that you're in. Now, he predicts, Brett specifically,
7:26
Ejaaz: that this is a $20 trillion enterprise opportunity for XAI that they cannot lose.
7:31
Ejaaz: And the reason why he makes this point is, if XAI ends up having the best model
7:36
Ejaaz: because they're training on tons of compute from space via SpaceX,
7:40
Ejaaz: then enterprises are most likely going to lock in Grok as their number one model,
7:45
Ejaaz: feed it off of all of the data that they kind of produce back into Grok,
7:49
Ejaaz: and it becomes this kind of recursive loop.
7:50
Ejaaz: And I think that that's something that people can't quite piece together,
7:54
Ejaaz: including myself, because it's such an ambitious goal to say,
7:57
Ejaaz: hey, we're going to launch AI data centers in space when we don't even have a single one here.
8:00
Ejaaz: But Elon predicts that this is going to be explicitly the case,
8:04
Ejaaz: not just in a decade, but in two to three years, which is just an insane prediction.
8:09
Josh: Yeah, and it seems outrageous. But then you look at the track history of a company
8:13
Josh: like XAI and you start to realize like what they say they actually mean.
8:18
Josh: And XAI, oh yeah, here we have the timelines of this.
8:22
Josh: October 2022, Elon acquired Twitter for $44 billion.
8:25
Josh: In March 2023, less than three years ago, XAI was founded. This company did not exist.
8:30
Josh: While companies like Anthropic and OpenAI have existed for far longer than that,
8:35
Josh: XAI raised a bunch of money in 2024.
8:37
Josh: Then by July, it has completed its super cluster, which took it 122 days.
8:42
Josh: And nvidia uh nvidia ceo johnson huang is on record saying generally this takes
8:47
Josh: 18 months to do and they did it in 122 days so this unbelievable rate of progress
8:52
Josh: and you see from july 2024
8:55
Josh: January, 2026, they went from zero to a $230 billion valuation,
9:01
Josh: a quarter of a trillion dollars.
9:03
Josh: And this is a result of their ability to engineer things faster and more efficiently than everyone else.
9:08
Josh: And in this game that we're playing, in this race to superintelligence,
9:12
Josh: the company and the entities that can deploy the resources fastest and most
9:16
Josh: efficiently are going to win.
9:18
Josh: And SpaceX and XAI have this unique advantage that they're able to build actual
9:22
Josh: hardware in the real world that is done at a rate that other companies just can't compete with.
9:28
Ejaaz: Dude, I'm just thinking of Larry Ellison's investment in XAI.
9:32
Ejaaz: Do you remember that email exchange that due to some court case was made public?
9:37
Josh: Yeah, it was like a billion, maybe two, whatever you think.
9:39
Ejaaz: Yeah, yeah. It was actually a text message. Elon messages Larry Ellison,
9:43
Ejaaz: the founder of Oracle, and he goes, hey, man, would love to have you as an investor in this.
9:47
Ejaaz: And Larry goes, sure uh i'm in
9:50
Ejaaz: for two billion question mark thumbs up and elon
9:53
Ejaaz: says honestly i think you should throw in like double this it's
9:56
Ejaaz: a great opportunity and larry just responds with sure
9:59
Ejaaz: and now like two years later that's
10:02
Ejaaz: a 2.2x uh return already on his investment so that's like probably one of the
10:08
Ejaaz: greatest individual trades so far but yeah like the rate of speed that this
10:11
Ejaaz: has happened josh is just pretty honestly insane and and i i kind of like uh
10:16
Ejaaz: puts into perspective like I remember when he took over X and then X became acquired by XAI.
10:22
Ejaaz: Back then it seemed like a shocking thing. And now we have like that being,
10:25
Ejaaz: those two entities being completely consumed by this one entity,
10:28
Ejaaz: which is just kind of like crazy to think about. Mm-hmm.
10:31
Ejaaz: I kind of, okay, I'm really bulled up right now. I just want to take literally
10:36
Ejaaz: 60 seconds to kind of ground myself.
10:39
Ejaaz: XAI isn't making enough money to pay for itself, right?
10:44
Ejaaz: So there are the camp of people, Josh, that are kind of arguing that this is
10:48
Ejaaz: just kind of like a way for Elon to save his investors money.
10:52
Ejaaz: To give you guys an insight into the numbers, XAI is currently burning about
10:56
Ejaaz: a billion dollars a month.
10:58
Ejaaz: Now, there's a lot of reason for that because it's capex spend on training or
11:01
Ejaaz: creating data centers and also training their models, etc.
11:04
Ejaaz: And they're only making around $500 million per year.
11:07
Ejaaz: So they're losing quite a lot of money. That ratio is much larger than,
11:10
Ejaaz: let's say, what Anthropic is losing a year or OpenAI is losing a year.
11:14
Ejaaz: But the idea is once they have enough compute, they will end up having the best
11:18
Ejaaz: model. And I think that's what this entire merger is about.
11:21
Ejaaz: It's a big bet, but if it pays off, it's going to pay off massively.
11:25
Josh: Yeah, I was going to say, no other AI lab is making money yet.
11:28
Josh: In fact, no one's losing even no one's losing more money than OpenAI.
11:31
Josh: So XAI really isn't in a bad position. In a way, this gives XAI the advantage
11:37
Josh: that Google has to some extent, where they have access to a larger balance sheet now.
11:42
Josh: And they can afford to make bolder bets because they can not only raise more
11:45
Josh: money, but deploy it more efficiently.
11:47
Josh: And not every decision is existential. We're seeing a lot of issues,
11:51
Josh: particularly with OpenAI, because it's just difficult to fulfill their debt obligations on time.
11:57
Josh: And giving yourself a larger war chest and more resources in order to handle
12:02
Josh: the profit generating aspect of these
12:05
Josh: companies is going to be hugely beneficial. And I mean, here we have a
12:12
Josh: What it actually looks like if the optimists win, if they're right.
12:16
Josh: And I guess to preface this post, maybe I'll just explain the Kardashev scale,
12:19
Josh: which is interesting in what this reference is. The Kardashev scale is basically
12:21
Josh: three types of civilizations.
12:23
Josh: Type one is a civilization that could harness the amount of energy on its own planet.
12:28
Josh: So that's if Earth can harness all the energy within us. If we have nuclear
12:32
Josh: reactors that can convert all this nuclear material to energy, that is type one.
12:37
Josh: Type two is harnessing the amount of energy from the star in your galaxy.
12:41
Josh: And then type three is to harness the energy of the entire solar system.
12:45
Josh: Elon has a plan to reach type two, which is funny because we are less than 1% of the way to type one.
12:52
Josh: So what is the plan to reach type two, EJS? Please walk us through what that
12:56
Josh: looks like without sounding too much like a sci-fi novel.
12:59
Ejaaz: Less than 1% for type one sounds crazy, but I don't know if I want to ever reach
13:04
Ejaaz: or live in a world where it is 100% for type one.
13:08
Ejaaz: Are we all just going to be fighting for scraps at that point?
13:10
Ejaaz: I don't know. I kind of have maybe like a dystopian.
13:13
Josh: Well, no, type 1 would be amazing abundance. Like imagine a world in which we
13:16
Josh: had essentially infinite energy because we could harness the amount.
13:19
Ejaaz: But is that coming directly from Earth, Josh, in a way that like doesn't,
13:24
Ejaaz: like other people don't suffer? Or am I misunderstanding that?
13:26
Josh: Yeah. So like imagine that we had nuclear reactors at scale that can turn the
13:31
Josh: small trace uranium in your backyard rocks into energy to power your entire house.
13:35
Ejaaz: Okay, okay.
13:36
Josh: To heat all the roads in New York City so we don't have to stretch through the snow for the last month.
13:39
Ejaaz: So it's economically efficient, Kodeshev type one.
13:43
Josh: Exactly, yes. And that would be a downstream effect of unlocking abundant energy
13:47
Josh: because abundant energy solves
13:48
Josh: so many of the societal problems that we face. So this is a good thing.
13:53
Ejaaz: Right, right.
13:54
Ejaaz: Yeah. Okay. Okay. So it sounds like Elon is kind of like given up on him personally
14:00
Ejaaz: accelerating type one and is moving straightforward to capitalize on type two
14:05
Ejaaz: because he has the means to do so and the only space company that can put out
14:09
Ejaaz: mass there for a cheap enough cost.
14:12
Ejaaz: So I call this a product roadmap, which Elon retweeted and confirmed.
14:17
Ejaaz: This tweet goes, step one is to move AI off Earth and into space.
14:21
Ejaaz: Step two is to use Starship specifically to build AI orbital data centers.
14:26
Ejaaz: Step three is to make space the cheapest place to run AI, which is,
14:30
Ejaaz: again, the thesis for this entire merger.
14:32
Ejaaz: Step four, expand manufacturing to the moon.
14:35
Ejaaz: Josh and I have spoken about this before on a previous episode.
14:38
Ejaaz: We're talking about crazy sci-fi things like mass drivers to propel satellites
14:42
Ejaaz: even further into deeper space so that we can get closer to the sun and harvest
14:46
Ejaaz: more energy per mass that's being pushed out.
14:48
Ejaaz: And then finally, step five a very easy one harness the entire sun at civilizational
14:53
Ejaaz: scale which of course is the end goal for katashev type two um now this sounds
14:59
Ejaaz: like theory right now and some people might point out that like hey it's just
15:03
Ejaaz: a merger but we're still gonna wait two to three years until this thing is actually yeah bro.
15:08
Josh: We're not harvesting stars it's just two companies merged today
15:11
Ejaaz: And I agree. Like, I agree. Like, that is true to an extent.
15:14
Ejaaz: But it's not like Elon and SpaceX hasn't already started taking steps towards fulfilling this.
15:21
Ejaaz: As you know, there are new Starship launches that are coming out.
15:25
Ejaaz: We've got an upgraded booster system to launch this rocket into space.
15:28
Ejaaz: And we have the new Starlink satellites that are going to be sent out soon,
15:32
Ejaaz: which I believe are like 2.5x the performance.
15:34
Ejaaz: Josh, you definitely have a better stat than me about that.
15:37
Josh: Per launch, Starlink V2 versus Starlink V3.
15:41
Josh: Um starship compared to a falcon rocket each starship launch will be 20 times
15:46
Josh: the output of a of a current falcon 9 launch so it's about 20 times multiple
15:52
Josh: on the amount of internet that will be provided per launch wow
15:55
Ejaaz: That is huge and.
15:57
Josh: In addition to that those v3 satellites have
16:00
Josh: much more powerful direct to cell connectivity so it
16:03
Josh: it enables a world in which people who have
16:06
Josh: a verizon cell plan may not need it because you can just get service anywhere
16:10
Josh: from the starlink satellites and they're going to need a tremendous amount of
16:14
Josh: bandwidth to do that but the i forget how many terabits for launch it is with
16:18
Josh: starship is going to be enough where if they can do that at scale and they can
16:21
Josh: launch as many as they're projecting then the network becomes incredibly powerful very quickly
16:25
Ejaaz: That's insane. That's insane. So I remember reading somewhere yesterday where Elon,
16:31
Ejaaz: if he's able to deliver on the millions of satellites that he wants to create,
16:35
Ejaaz: the constellation that will basically act as the AI data center in space,
16:39
Ejaaz: that'll result in about 100 gigawatts per year to train.
16:45
Ejaaz: That's only in the nearest time schedule, so like three to five years.
16:49
Ejaaz: But after that, he's planning to multiply that by 10 or 102 terabytes.
16:54
Ejaaz: So about 50 to 100 terabytes which is just a gargantuan terawatts sorry terawatts
17:00
Ejaaz: which is just an insane amount of energy but to to go back on my point also just quick.
17:04
Josh: Point for reference the united states currently generates 1.3 terawatts of energy annually
17:08
Ejaaz: All right so he's going to 10x that of.
17:10
Josh: The entire united states is the plan correct that's
17:12
Ejaaz: Yeah and i guess his argument there then is like we can't do this on earth because
17:17
Ejaaz: it's going to come at the cost of other people's daily livings and requirements for energy.
17:22
Ejaaz: So his idea is to just kind of offload that into space so that it doesn't affect
17:26
Ejaaz: any of the water resources, minerals, and all that kind of stuff that would
17:30
Ejaaz: be required to reproduce the same type of effects on Earth.
17:34
Ejaaz: So he filed a request, or rather a proposal, I don't know what the difference
17:39
Ejaaz: is, with the FCC to spin up the first million satellite constellation.
17:44
Ejaaz: I don't know how soon he's going to spin that up, but if he's projecting two
17:47
Ejaaz: to three years where this is going to become the norm, then I'm guessing he's
17:50
Ejaaz: aiming to do this within the next year and a half.
17:52
Ejaaz: Or is this an Elon prediction, Josh? What do you think?
17:56
Josh: It's an Elon prediction, but the way that these scales work is...
18:02
Josh: It can happen. It happens very quickly. Like when you look at an outcome,
18:06
Josh: when you look at an exponential chart, the rate at which goes vertical is very quick.
18:10
Josh: So if you miss a window by a year of your prediction, you could be off by a
18:14
Josh: significant amount, but it will catch up very quickly.
18:17
Josh: So if it's not 2028 and it doesn't look like it's going to be even close,
18:21
Josh: there's a world in which 2029, it actually happens because the way that these things scale.
18:24
Josh: So I think it will happen and in order
18:28
Josh: for them to achieve this ai data center thing
18:31
Josh: at scale the current trajectory needs to
18:34
Josh: be about 10 000 launches per year and a
18:37
Josh: major airline they fly about 100 000 airplanes
18:40
Josh: per year so if you think in terms of that
18:43
Josh: if you can get to one-tenth the output of an airline then at
18:46
Josh: that scale you can get the cost per kilogram to orbit low enough that it
18:49
Josh: becomes economically inefficient to train ai anywhere else
18:52
Josh: but outer space and like you mentioned it offloads a lot of the burden that
18:55
Josh: humans face if we don't go to space the impact
18:59
Josh: that ai training will have on society will
19:02
Josh: continue to grow the amount of grid strain it will have is going to continue
19:05
Josh: to grow the amount of footprint that it's going to consume in terms of resources
19:09
Josh: in terms of instability in terms of just dealing with legislature to get it
19:11
Josh: built will grow at a crazy rate and this offsets all of those problems and removes
19:16
Josh: the burden and places it into outer space and i think for that alone it's a huge win for everybody
19:21
Ejaaz: Yeah. Wow. That's insane. Okay. So I'm going to lay out the stack that is created
19:27
Ejaaz: by the merging of these two companies, Josh, and let me know what I miss.
19:31
Ejaaz: But the way I have it in my head is SpaceX is basically going to be the literal
19:35
Ejaaz: physical launch pad to get the satellites and data centers out into space.
19:41
Ejaaz: And then the satellite specifically, so Starlink in this case,
19:45
Ejaaz: will be the connectivity layer.
19:47
Ejaaz: So you're going to have like satellites basically beaming data to train AI models,
19:52
Ejaaz: Grok presumably, and maybe whoever else wants to pay for the service,
19:56
Ejaaz: but Grok primarily out there.
19:58
Ejaaz: And then you have Compute Back on Earth provided by Colossus 2,
20:02
Ejaaz: Colossus 3, and presumably Colossus 4, 5, and 6, which kind of like handles
20:06
Ejaaz: the Earth's resources to train Grok 5 there.
20:09
Ejaaz: And then you have the model Grok 5, and then you have the distribution through
20:13
Ejaaz: X, the social media platform. Am I missing anything?
20:17
Josh: There's a lot going on there. i think this might
20:20
Josh: be a good time to introduce the third company of musk industries
20:23
Josh: which is tesla please and we see here um in this
20:26
Josh: post from bev jazos about the like convergence of
20:29
Josh: the three and particularly how they're going
20:31
Josh: to relate to one another and i think if i could summarize this even
20:34
Josh: more so than what we're seeing here is that um xai
20:38
Josh: is the intelligence xai turns the sand to thought
20:41
Josh: and then tesla powers that and embodies
20:44
Josh: that through optimus robots through cyber cabs
20:47
Josh: it gives it a physical body in which it can live in while xai
20:51
Josh: exists as the orchestration layer so once you embody this
20:54
Josh: ai xai can also orchestrate millions of these humanoid robots to do something
20:58
Josh: valuable and productive and then spacex is the expansionary layer that takes
21:02
Josh: all of these resources that have been generated here on earth and sends it into
21:05
Josh: outer space which has a few benefits the first is that it's much more economically
21:09
Josh: viable to improve the intelligence of these models out in space.
21:13
Josh: And then it's also an unlock in terms of exploratory value.
21:17
Josh: We can send humanoid robots to the moon at scale. You have access to low-Earth orbit networks
21:23
Josh: you get all of the benefits that come from having all of this untapped real
21:27
Josh: estate and untapped energy source, which is just being closer to the sun and
21:30
Josh: being able to deploy solar at scale to reach the energy requirements to actually
21:35
Josh: train these models at scale.
21:36
Josh: So it's this unbelievable convergence that's happening between these two companies.
21:39
Josh: I'm sure Tesla will play a huge role. And it's going to be so hard to compete
21:44
Josh: with this at scale. My God.
21:46
Ejaaz: I have a question for you, Josh. Yeah. Do you think, okay, what are the odds
21:51
Ejaaz: that Tesla merges with this new SpaceX XAI X entity?
21:57
Josh: I would like for them to be high. And I think they're probably somewhere around
22:02
Josh: 50%. But I don't think that it's likely now given that SpaceX has chosen to merge with XAI.
22:09
Josh: This benefits investors, generally speaking, because when companies are spun
22:14
Josh: out of each other, the net is generally worth more than if they exist as a conglomerate.
22:18
Josh: We see this at Google, where Google's assets are actually worth a lot of money.
22:21
Josh: And if you spun out Waymo, it would probably trade at a pretty large premium.
22:25
Josh: But because they're all compressed under one roof, they do get this discount. I think...
22:31
Josh: If I had to bet, I would say by the end of this year, SpaceX is public and it
22:36
Josh: trades separate from Tesla.
22:38
Ejaaz: Okay. I'm going to take the opposite side of that. I think that Tesla and SpaceX
22:44
Ejaaz: might merge, will merge this year, and it might actually happen pre-SpaceX IPO.
22:52
Josh: And I hope you're right, because that would be so awesome.
22:55
Ejaaz: Yeah, it's going to be one equity instrument that you can basically invest in
22:59
Ejaaz: to get exposure to all of Elon's efforts. Musk Industries.
23:02
Ejaaz: It doesn't, yeah, it doesn't include Neuralink until that inevitably gets acquired
23:06
Ejaaz: by frigging X or something.
23:08
Ejaaz: But until then, this tweet actually lays it out pretty well or the hypothesis.
23:13
Ejaaz: He goes, the real genius is this. Elon is effectively running what's called a dual track.
23:18
Ejaaz: A dual track is when a merger and an IPO are prepared simultaneously,
23:23
Ejaaz: creating maximum competitive pressure.
23:25
Ejaaz: He's referring to a merger between SpaceX and now Tesla.
23:29
Ejaaz: Tesla shareholders who would ultimately have to sign off on any deal are then
23:33
Ejaaz: faced with a stark choice.
23:34
Ejaaz: Disapprove the deal and suffer a catastrophic outflow of Tesla holders because
23:38
Ejaaz: they're going to be selling to get into SpaceX XAI or approve,
23:43
Ejaaz: some begrudgingly, a merger with epic dilution. I don't know why it would be dilutive, actually.
23:48
Ejaaz: And avoid the Tesla exodus to SpaceX XAI. This maximizes shareholder buy-in. Brilliant.
23:54
Ejaaz: Now, I think some of this is tinfoil hat, But if I think back to a tweet that
23:58
Ejaaz: I saw Chamath post mid last year, where he predicted that this merger between SpaceX,
24:04
Ejaaz: XAI would happen, and everyone laughed at him, including myself,
24:07
Ejaaz: now I'm kind of like, maybe this will happen.
24:09
Ejaaz: And he believes that there's a strong chance that it will.
24:12
Josh: I certainly hope so. So maybe we can answer the question now of how do you actually
24:15
Josh: participate in all the subside that we're describing?
24:17
Josh: Because clearly, I mean, XAI is private, SpaceX is private, they will go public,
24:21
Josh: but it's difficult to acquire shares. So, yeah.
24:24
Josh: What are the ways you can get involved? I think the first one is through Tesla stock.
24:28
Josh: What we have on screen is a post from Elon who says, I've mentioned something
24:31
Josh: like this before, but if any of my companies goes public, we will prioritize
24:35
Josh: other longtime shareholders of my other companies, including Tesla.
24:40
Josh: Loyalty deserves loyalty. So I think that's stating it as about as clearly as you can get.
24:45
Josh: If you want access to the upside, the only public instrument you can really
24:48
Josh: own right now is Tesla stock.
24:50
Josh: And the hope is that there will be some preference for
24:54
Josh: the shareholders in addition to just owning shares of
24:56
Josh: an amazing company the other interesting way that you can get
24:59
Josh: involved is through another company
25:03
Josh: who just so happens to own quite a bit of shares of spacex stock
25:05
Josh: which is actually google believe it or not google owns
25:08
Josh: seven and a half percent of spacex it's one of the largest shareholders on the
25:13
Josh: balance sheet and that's in addition to owning about 15 or 14 maybe of anthropic
25:18
Josh: stock so google has kind of become this holding company for major AI labs in
25:23
Josh: that if you want access in public markets,
25:26
Josh: then your best bet is either buy SpaceX or buy Google.
25:29
Josh: And those seem to be the, ironically, the two companies that I am most bullish on.
25:34
Josh: I just, we released an episode on Bankless earlier this week talking about why
25:38
Josh: We're so bullish on those companies.
25:40
Josh: And those are probably the ways that I would suggest, if interested,
25:44
Josh: getting some level of exposure to this prior to IPO.
25:49
Ejaaz: Yeah, Google is one of the greatest investment entities of our time,
25:54
Ejaaz: if they weren't considered one of the top technology companies.
25:58
Ejaaz: It actually brings up a question that I want to kind of propose to us,
26:02
Ejaaz: Josh, which is, how would other frontier AI labs theoretically compete with this new entity?
26:11
Ejaaz: They're going to have all the compute in the world that they could desire.
26:14
Ejaaz: They're going to have first rights of access to this compute because SpaceX
26:18
Ejaaz: is the only one launching satellites into space for cheap enough cost.
26:23
Ejaaz: And then they're going to use the scaling laws to make Grok into the biggest beast ever.
26:29
Ejaaz: Now, granted, all of this is theory right now because we haven't seen a new
26:32
Ejaaz: Grok model in kind of a while, rumored to be a new model released this month, actually.
26:38
Ejaaz: But assuming the scaling laws prove effective which is what Google's Gemini
26:43
Ejaaz: actually proved then, you know, how did they get beat?
26:48
Ejaaz: OpenAI doesn't have access to their own space company that could get access to this.
26:51
Ejaaz: They're either going to have to pay SpaceX a huge premium or go with someone
26:55
Ejaaz: like Blue Origin, Amazon maybe bullish Amazon as well there.
26:58
Josh: And even in the absence of space-based data centers the XAI team currently has
27:03
Josh: the largest coherent training cluster.
27:05
Josh: They've been deploying these gpus faster than
27:08
Josh: any other company and it feels that on a raw compute raw
27:11
Josh: intelligence level xai has already
27:14
Josh: won and kind of similar to google where they have these tpus
27:17
Josh: and they have this accelerated compute and they have the cap table to weather
27:21
Josh: the storm in a way that other companies cannot we see xai already subsidizing
27:25
Josh: everything they're they're offering ads and they're trying to claim percentage
27:30
Josh: points of innovations that's made on the platform and Google has the balance
27:34
Josh: sheet to subsidize this.
27:35
Josh: So it's going to be really difficult to compete in the absence of space.
27:39
Josh: With space data centers and with the Tesla free ride to outer space,
27:43
Josh: that's going to make things incredibly difficult.
27:45
Josh: And it seems like the only real solution that they have
27:48
Josh: is to, one, move faster, but two, just create unbelievably great products that can compete with XAI.
27:56
Josh: Like to build a moat as strong and as tough as possible, as fast as you can,
28:01
Josh: so that users and customers, they don't want to leave.
28:04
Josh: But those incentives to leave are going to grow very, very strong as a company
28:08
Josh: like XAI kind of realizes its final form, deploys these at scale,
28:12
Josh: and just continues to build super intelligence at a rate that I'm not sure any
28:16
Josh: company is able to match.
28:18
Ejaaz: Yeah, intelligence just becomes a function of compute, I guess.
28:22
Ejaaz: Correct. So it all comes down to compute.
28:24
Josh: Yeah, it's an energy problem. The more energy you can throw at these chips,
28:27
Josh: the greater the intelligence is.
28:29
Josh: And there is seemingly no limit to where that ends. The person who collects
28:33
Josh: the most amount of energy and the most amount of chips wins.
28:36
Josh: Tesla is building their own chip fab.
28:39
Josh: Tesla has the battery infrastructure, they have the solar panels,
28:42
Josh: And now they have a company that can actually take that into outer space and
28:46
Josh: leverage the benefits that come with space.
28:48
Josh: Things like eight times more efficiency on solar panels and sun synchronous
28:51
Josh: orbit where you can actually sync up your satellites to never go out of the
28:56
Josh: sun's rays and to stay in the light side of orbit for like perpetually.
29:01
Josh: So there's a lot of huge advantages that come from this run by the best operator in history.
29:07
Josh: And it's like, you're really fighting an uphill battle here.
29:12
Ejaaz: Josh, I have a question for you before we run up the episode.
29:16
Ejaaz: Let's give it till 2030. And we assume that SpaceX IPOs, and this may be with
29:23
Ejaaz: or without a Tesla merger.
29:24
Ejaaz: What's the valuation of that entity?
29:27
Josh: Without a Tesla merger, there's probably a clear path to $5 trillion.
29:34
Josh: With a Tesla merger, there's probably a clear path to $10 trillion.
29:40
Josh: The reasoning behind that would be, by 2030, they should have Starship working at scale.
29:46
Josh: The cost per kilogram to orbit should be low enough where it becomes economically
29:52
Josh: inefficient to build AI anywhere else but space. And SpaceX owns a total monopoly.
29:57
Josh: They'll also have the value of the Starlink network mostly realized by then,
30:03
Josh: where they'll have deployed enough Starlink V3 and probably V4 or V5 satellites
30:06
Josh: into space that they will be able to network the entire globe at a level that
30:12
Josh: competes with Verizon and all of the other cell providers.
30:15
Josh: And then Tesla by that time will have Optimus at scale, will have CyberCabs
30:20
Josh: at scale, will have displaced a lot of the physical labor
30:24
Josh: in the country or at
30:27
Josh: least some extent by then and optimists should be ramping up production
30:30
Josh: by then to a point where the cost per optimist is
30:33
Josh: twenty thousand dollars thirty thousand dollars and the cost per task in the
30:37
Josh: workforce goes down significantly and like these companies are really going
30:40
Josh: to move the gdp of the world higher and as that gdp rises so does the market
30:45
Josh: cap of these companies and there really is no limit at the limit of what these
30:49
Josh: companies could scale to It's going to be unbelievable.
30:52
Ejaaz: Yeah, I don't know if I could prescribe a number to the market cap,
30:55
Ejaaz: but I think it will be the most valuable company in the world.
31:00
Ejaaz: Who's the most valuable company in the world right now? It's NVIDIA.
31:02
Ejaaz: Why? Because they produce the raw bones to train AI models, GPUs.
31:08
Ejaaz: If GPUs end up getting replaced by orbital satellites produced by SpaceX themselves,
31:14
Ejaaz: then I don't really see how NVIDIA maintains its position as number one.
31:20
Ejaaz: And in addition to this they'll own the number
31:22
Ejaaz: one distribution network through um x specifically
31:26
Ejaaz: number one model with x ai and if the
31:28
Ejaaz: merger goes ahead with tesla physical embodied ai
31:31
Ejaaz: as well the raw components and bedrooms the batteries uh like you said that
31:36
Ejaaz: that is needed to power this entire infrastructure kind of probably the most
31:40
Ejaaz: ambitious thing ever um this is literally sci-fi in the making that is materializing
31:45
Ejaaz: right now um the fictional character that kind of springs to mind immediately, is Tony Stark.
31:50
Ejaaz: This is Musk Industries live and well and actually materializing.
31:56
Ejaaz: This is the craziest timeline.
31:58
Ejaaz: And it's only the start of the second month of the year. Yeah,
32:02
Ejaaz: I'm exhausted. I haven't slept.
32:04
Josh: It's really exciting because the world actually is materially changing for the
32:09
Josh: first time in a very long time.
32:10
Josh: Like we haven't had progress like we're on the verge of since the late 1800s, early 1900s. And
32:19
Josh: These companies have a very clear path to getting there. And I think there is
32:22
Josh: this mispricing happening where a lot of the expectations and valuations are
32:26
Josh: based on the past 20 to 30 years of progress.
32:31
Josh: That's why Tesla's trading at $1.3 trillion relative to a company like Google
32:35
Josh: at $4 plus trillion or NVIDIA $4.5 trillion.
32:37
Josh: And it seems like there's a
32:39
Josh: lot of catching up and repricing that needs to happen as we move forward.
32:42
Josh: And I suspect the SpaceX IPO will be one of the first versions of
32:46
Josh: kind of like a slap in the face like oh wait a second this this
32:49
Josh: world that we're going to isn't like the world that we're coming from and as
32:52
Josh: a result things need to change things need to be repriced and we're probably
32:55
Josh: going to see that significantly happen as spacex and xai
32:58
Josh: go public so that's going to be incredibly exciting along with
33:01
Josh: all the other amazing stuff that's happening this year it's just this week even
33:04
Josh: this week every week feels like a month the progress is happening so fast it's
33:07
Josh: tough for us to keep up with so thank you for listening and keeping up with
33:11
Josh: it with us it's been really fun to just continue to to cover all this crazy
33:16
Josh: news as it happens and there's more coming this week too
33:19
Ejaaz: Yeah, yeah, there's a lot more coming. So before we started recording this episode,
33:24
Ejaaz: there are strong rumors that Anthropix is going to be dropping a new model that
33:28
Ejaaz: is going to be even better than Opus 4.5 at coding.
33:32
Ejaaz: And then yesterday, we got a new coding interface from OpenAI.
33:35
Ejaaz: So we're going to have an episode that's probably going to compare the two with
33:37
Ejaaz: some really awesome demos.
33:39
Ejaaz: So you guys can actually figure out how to use these things. Really stoked for this.
33:43
Ejaaz: We had the merger news that we covered today, obviously, and we just put out
33:46
Ejaaz: an episode on Maltbook, which is the air-only social media platform. It's a banging episode.
33:51
Ejaaz: We had so much fun creating it. So please go check it out. We also broke 30,000
33:56
Ejaaz: subscribers, but we gained another 200 over the last like day and a half.
34:00
Ejaaz: So the episodes have really been paying.
34:02
Josh: Smash that subscribe button.
34:03
Ejaaz: It tells me that you guys are really enjoying it, but still 80% of you aren't subscribed to us.
34:08
Ejaaz: So if you enjoy this content and if you want us to keep producing this content
34:11
Ejaaz: and if you want to get access to this type of content, please give us a subscribe.
34:15
Ejaaz: Turn on notifications. If you're not watching YouTube or like wherever you listen,
34:19
Ejaaz: listen to this, give us a rating. Anything helps out so much.
34:23
Ejaaz: We also have a banging newsletter. We have a killer essay that egotistically I wrote this.
34:30
Josh: It's great. I read it. It's great. I can vouch.
34:32
Ejaaz: Thank you. Thank you. Thank you. So super fun piece. We're trying to be anywhere
34:36
Ejaaz: and everywhere and deliver you the best content that we can with our thoughts
34:40
Ejaaz: and insights each and every week, three to four episodes a week.
34:43
Ejaaz: Stick with us. We're going to go into the biggest AI show ever.
34:46
Ejaaz: Um thank you guys for listening.
34:47
Josh: And we'll see you guys in the next one peace