Latest / In the Loop AI / This Week In AI: Why SpaceX's $1.5 Trillion IPO is Undervalued

Transcript

This Week In AI: Why SpaceX's $1.5 Trillion IPO is Undervalued · In the Loop AI

0:00
Josh: On paper, SpaceX's $1.5 trillion IPO would be the biggest in history,
0:05
Josh: raising about $30 billion.
0:06
Josh: But that's not the crazy part. The crazy part is that even at that price,
0:11
Josh: we still believe it's undervalued.
0:13
Josh: In fact, grossly undervalued because this all stems from one realization.
0:17
Josh: It's turning Starlink into AI data centers in space. And the winner who deploys
0:22
Josh: resources the fastest gets to reshape civilization.
0:24
Josh: It isn't just SpaceX stock finally available to the public.
0:27
Josh: This is Elon using Wall Street's money to build an
0:31
Josh: orbital gpu swarm with these new satellites that are 150 kilowatts and starship
0:35
Josh: will launch basically a small city's worth of ai power into orbit every single
0:39
Josh: flight we're going to talk about all of this um because eventually this leads
0:42
Josh: to lunar factories producing ai satellites and if he pulls this off it's going
0:47
Josh: to be pretty amazing so in this episode we're going to break down why this ipo exists
0:50
Ejaaz: Only because of ai.
0:52
Josh: This was not the case prior to this week how the physics actually favored data
0:55
Josh: centers in space and why 1.5 trillion dollars may be a rounding error if SpaceX
0:59
Josh: wins because, I mean, in this game, Ejaz, it's kind of simple.
1:02
Josh: Whoever can deploy compute the fastest, they win the AI race and they probably
1:06
Josh: get to rewrite the future of civilization.
1:08
Ejaaz: Yeah. I mean, before we get into it, I want to walk everyone through the kind
1:12
Ejaaz: of series of events that kind of led to this groundbreaking bit of news.
1:16
Ejaaz: So it all started off around Wednesday this week, where two news sources,
1:22
Ejaaz: Bloomberg and the information leaked news that SpaceX was potentially going to IPO sometime in 2026.
1:30
Ejaaz: But we didn't quite know the valuation at that point.
1:33
Ejaaz: And then it broke that it was a $1.5 trillion valuation.
1:36
Ejaaz: And the reason why this is such big news is it would technically be the biggest IPO ever.
1:42
Ejaaz: And then people were like, but Elon hasn't confirmed it.
1:45
Ejaaz: Well, yesterday, Elon officially confirmed it. He responded to a great article
1:50
Ejaaz: from Eric Berger where he kind of predicted why he thinks SpaceX is definitely going to IPO soon.
1:56
Ejaaz: And Elon confirming it all but means that,
1:59
Ejaaz: If SpaceX IPOs, Josh, at $1.5 trillion, his net worth would effectively double,
2:06
Ejaaz: which would put him just under, not exactly at, but just under $1 trillion overnight.
2:13
Ejaaz: Once the IPO goes live, he owns 42% of SpaceX.
2:17
Ejaaz: So listen, there's a bunch of big numbers here.
2:20
Ejaaz: In order to successfully IPO at this valuation, he needs to raise $30 billion.
2:25
Ejaaz: Dollars um he needs to you know
2:27
Ejaaz: technically deliver on a lot of stuff that we're going to get into in a second
2:31
Ejaaz: but you know massive massive vision so the question i ask myself is why now
2:36
Ejaaz: and what is he doing to justify this what is he telling the people that he's
2:40
Ejaaz: raising 30 billion dollars from um to convince them to give him that kind of
2:44
Ejaaz: money and it comes to one thing josh,
2:47
Ejaaz: AI data centers in space. Now, you and I have had a very tumultuous relationship with this.
2:55
Josh: We've come a long way.
2:56
Ejaaz: We've come a very long way. And by a long way, in the last four weeks,
2:59
Ejaaz: when this kind of like entire trend started developing, at least on social media.
3:04
Ejaaz: And the concept here is if you're able to put GPUs or AI data centers in space
3:09
Ejaaz: and harness the energy of the sun, you effectively have infinite power.
3:13
Ejaaz: And in this game of creating the best AI to reach AGI level,
3:18
Ejaaz: you need as much compute as you can get your hands on. And the fact is Earth's resources is scarce.
3:23
Ejaaz: So Elon's whole pitch behind IPO-ing SpaceX is he believes he can control and
3:29
Ejaaz: build data centers in space using his Starlink satellite network.
3:33
Ejaaz: Josh, he thinks he can create a constellation of Starlink satellites that can
3:37
Ejaaz: speak to each other via laser beams, right?
3:39
Ejaaz: So they can communicate and share compute and data between themselves,
3:43
Ejaaz: harness solar energy from the sun and beam that down to earth to create AGI.
3:49
Ejaaz: This sounds like something out of a sci-fi novel, Josh.
3:53
Ejaaz: It seems like you're obviously bullish. How feasible is this?
3:56
Ejaaz: What are we looking at here?
3:58
Josh: Yeah, it's funny because for the longest time, Elon was strictly no.
4:03
Josh: No way are we ever going to go public. But the second that they realized that
4:07
Josh: Starlink satellites can be architected as this distributed network of data centers,
4:11
Josh: it went from no way to we kind of have to.
4:14
Josh: Because a lot of this AI race, it kind of comes down to deploying assets just
4:18
Josh: quicker than your competitors.
4:19
Josh: And we've seen this with XAI in the past where they did not exist, what, two years ago?
4:24
Josh: And now they are at the forefront with everybody else. And it's because they
4:28
Josh: were able to deploy their resources faster than everyone else.
4:30
Josh: They were the fastest to get to
4:32
Josh: large coherent cluster of nvidia gpus
4:35
Josh: and as a result they have some of the best models in the world so this
4:38
Josh: is clearly a race to get to resources spacex has
4:41
Josh: the ability to do so xai has the models to do so but spacex doesn't quite have
4:45
Josh: the money so their projected revenue was i think it's about 24 billion dollars
4:49
Josh: in the near term which for a fun fact is roughly equivalent to nasa's annual
4:53
Josh: budget but what they realized is wait a second if we're going to compete on
4:57
Josh: a global scale and building AI data centers,
4:59
Josh: we need far more money than that.
5:02
Josh: And that is the reasoning why they're going to public markets,
5:04
Josh: raising the $30 billion.
5:05
Josh: And again, at $1.5 trillion, that's about $30 billion they're going to raise in addition.
5:10
Josh: So Ijaz, I guess the question, I started by saying this is fairly undervalued
5:13
Josh: because the stakes are so high.
5:15
Josh: We'll explain the reasoning, but I'm curious what your take is on the valuation,
5:18
Josh: the $1.5 trillion number as it sits today.
5:21
Ejaaz: Okay, assuming this is economically and physically possible,
5:27
Ejaaz: Josh, it's a no-brainer.
5:29
Ejaaz: And it's for one simple reason, which Andrew actually outlines pretty well in
5:33
Ejaaz: this tweet that I'm showing here.
5:35
Ejaaz: To win the AI race, you need the most compute, data, and energy.
5:41
Ejaaz: That last one, if you can get infinite energy from the sun, you've already won, right?
5:47
Ejaaz: And the only reason why people haven't considered this before is because it hasn't been possible.
5:51
Ejaaz: Elon will effectively own the highway and the tolling system into space.
5:55
Ejaaz: SpaceX is like literally a decade ahead of the next nearest competitor.
5:58
Ejaaz: So if he can create a monopoly on this and actually make this viable,
6:03
Ejaaz: he wins. Your point around him needing the money to apply resources to this,
6:08
Ejaaz: we all know that Elon started SpaceX to get to Mars, right?
6:12
Ejaaz: He wants human civilization on Mars.
6:15
Ejaaz: And he didn't really have a good justification as to why he would IPO SpaceX,
6:19
Ejaaz: aside from maybe getting extra cash to help fuel that vision.
6:23
Ejaaz: Now he realizes that he needs to kind of create this constellation of satellites
6:28
Ejaaz: to pay for the Voyagers and the expensive mission to get to Mars.
6:32
Ejaaz: So it's a means to an end is the way that I'm kind of thinking about it.
6:35
Ejaaz: And so now when I connect the dots, Josh, is $1.5 trillion valuation undervalued or overvalued?
6:42
Ejaaz: I think it's undervalued assuming he can get the data centers out in space and
6:47
Ejaaz: beam it down back to Earth.
6:48
Ejaaz: Because it's not just going to be him and Grok and XAI that benefits from this.
6:52
Ejaaz: It's going to be all the other frontier AI labs that are going to want to or
6:56
Ejaaz: need to rather use his infrastructure. And that's my simple thesis.
7:01
Josh: Yeah, you can ask the question, well, one, like, is this a monopoly?
7:04
Josh: Yes, this is a monopoly. The next closest company is five to 10 years behind being Blue Origin.
7:09
Josh: Is this a monopoly on a market that is meaningfully large? Yes.
7:15
Josh: AI, as well as low earth orbit satellites, as well as energy satellites,
7:18
Josh: are about some of the biggest markets on planet earth so the total active
7:21
Josh: market size is gigantic and the
7:24
Josh: third question is is there an operator capable of actually doing
7:27
Josh: the impossible and making the impossible a reality and the
7:30
Josh: answer to all three of the questions are yes so there is
7:33
Josh: a unique monopoly in a industry that
7:36
Josh: is larger than any industry that exists today run by
7:38
Josh: the best operator in planet earth and the convergence of those three things if
7:42
Josh: it works in the world that we do actually get ai data centers in space and we're
7:48
Josh: building artificial general intelligence out there we're harnessing a lot of
7:50
Josh: power of the earth that is a civilizational scale shift and the amount of wealth
7:55
Josh: generated from that will make 1.5 trillion dollars feel like a drop in the bucket
7:58
Josh: so maybe today 1.5 trillion feels large but in the scale of things
8:02
Josh: that's a very small number.
8:03
Ejaaz: Typically, we speak about companies going from zero to one, and that's where
8:06
Ejaaz: those exponential valuations come from.
8:09
Ejaaz: With this IPO and with Elon's kind of successful Starship launches and satellite
8:14
Ejaaz: launches, we're somewhere between zero and one. We're at 0.5.
8:19
Ejaaz: So the risk has been reduced because he's reduced the cost of going to space.
8:24
Ejaaz: He's already launched satellites successfully. So now the exponential returns
8:29
Ejaaz: seem really, really more feasible.
8:30
Ejaaz: So I think he's going to raise that $30 billion in a matter of,
8:34
Ejaaz: you know, months, if that amount, right?
8:37
Ejaaz: The other thing is, like, he's literally turning science fiction into reality here.
8:41
Ejaaz: We talked about the concept of the Kardashev scale, type two civilization,
8:46
Ejaaz: which is, you know, humans being able to harness the energy of their local star, that is the sun.
8:51
Ejaaz: And we've suddenly gone from it being zero.
8:53
Ejaaz: So like that, there's no way he's going to pull it off to, ah,
8:56
Ejaaz: you know what? He actually might.
8:57
Ejaaz: And so $1.5 trillion seems kind of very undervalued if you have that kind of mindset.
9:02
Ejaaz: But again, people listening to the show, Josh, are going to be like,
9:05
Ejaaz: well, I don't think the numbers make sense.
9:07
Ejaaz: You can't get rid of heat in space. It's a vacuum.
9:10
Ejaaz: How are you going to be able to launch satellites? They're going to need to
9:13
Ejaaz: be kilometers square wide.
9:15
Ejaaz: Josh, have you got any of the numbers that can help break this down?
9:18
Josh: Let's run the numbers. So we talked about a good bit of the logistics and
9:22
Josh: the cooling in a previous episode earlier this or last week so
9:24
Josh: go listen to that if you're interested in that the numbers that
9:27
Josh: we're going to talk about today are the amount of scale
9:30
Josh: they're actually able to get into outer space and what that looks like so elon
9:33
Josh: recently posted on x their ability to get 150 kilowatts
9:37
Josh: per satellite into outer space so for some reference numbers the average u.s
9:42
Josh: home uses about 10 500 kilowatt hours which is about 1.2 kilowatts of continuous
9:48
Josh: energy so if an ai satellite can generate around 150 kilowatt of electrical
9:52
Josh: power, like Elon is saying,
9:54
Josh: 150 kilowatts divided by 1.2 is about 125 homes worth of continuous power.
10:00
Josh: So one of these AI satellites has roughly the power budget of 120 to 130 American
10:05
Josh: homes running all the time.
10:07
Josh: Now, how much can you do per launch? Because a launch consists of many of these satellites.
10:13
Josh: So SpaceX's own site says Starship is designed to carry up to 150 metric tons of low Earth orbit.
10:20
Josh: So how many is that? that's about 50 to 75 of these satellites per flight.
10:24
Josh: So if you do that math, 60 satellites times 150 kilowatts of energy is nine
10:29
Josh: megawatts of power in orbit from one starship,
10:32
Josh: which is equivalent to a small town in New England with every single launch
10:37
Josh: that they make into outer space.
10:39
Josh: That is 7,500 homes. That is 19,000 people's home usage.
10:43
Josh: So every fully loaded starship launch is like dropping a small city's worth
10:46
Josh: of electric households dedicated just to AI into orbit.
10:50
Josh: And that is an outrageously large amount of energy and compute that they're
10:55
Josh: putting into outer space, especially relative to what people are trying to do back here on Earth.
10:58
Josh: If you remember a Blackwell chip, which is NVIDIA's newest cutting edge rack,
11:03
Josh: it uses 120 kilowatts per rack.
11:05
Josh: And each one of these satellites surpasses a Blackwell. So it's like this gargantuan
11:10
Josh: amount of energy and compute that's being sent to orbit every single launch.
11:13
Ejaaz: Yeah, he's tapped into the infinite energy clip and it's become a reality.
11:18
Ejaaz: And if he can harness all of that, he basically wins.
11:22
Ejaaz: The other thing I was thinking about, Josh, is
11:25
Ejaaz: To your point around SpaceX or Starlink specifically being a monopoly on this,
11:31
Ejaaz: that's effectively the way that I see it.
11:33
Ejaaz: Like pretty much every single usable satellite out there is a Starlink-affiliated satellite.
11:39
Ejaaz: In fact, we've got this really cool graphic that we want to show everyone here.
11:43
Josh: Oh, yeah, this is sick.
11:44
Ejaaz: Okay, if you're looking at this, if you squint, it might look like a bacteriophage,
11:49
Ejaaz: right? But no, this is a simulation of all live satellites that are orbiting Earth right now.
11:55
Ejaaz: And if you notice my cursor, whenever I touch pretty much any satellite that's
11:59
Ejaaz: kind of bumbling around here, you'll see Starlink repeated over and over and over and over again.
12:04
Ejaaz: So he has the market monopoly. I think something like 90% of satellites out
12:08
Ejaaz: there are Starlink based or have been launched from SpaceX infrastructure and launches.
12:14
Ejaaz: So he has the market monopoly. And the reason why that is so important is...
12:19
Ejaaz: The capex cost of launching this out into space and making AI data centers in
12:24
Ejaaz: space feasible, it actually becomes affordable to do this, right?
12:28
Ejaaz: Effectively, if you imagine having just empty data center slots bumbling around
12:34
Ejaaz: in space, and all you need to do is fit in the latest GPU and launch it up there
12:38
Ejaaz: for cheaper than the cost of the weight of the GPU,
12:41
Ejaaz: Elon will be able to do that via SpaceX.
12:43
Ejaaz: So it's a crazy monopoly. I do not see anyone coming even close maybe it's uh
12:48
Ejaaz: amazon or rather jeff bezos's blue origin but that's still a decade behind right
12:52
Ejaaz: so as far as i see it elon's got this in the bag.
12:55
Josh: Yeah and they're i mean because of the rapid reusability of
12:58
Josh: starship you're able to send just tons of mass to orbit elon he posted that
13:02
Josh: he's expecting one megaton per year um so i mean 100 gigawatts it's on the that
13:07
Josh: includes 100 gigawatts it's on the order of like a dozen large power plants
13:11
Josh: so this map that we're looking at right now full of all these satellites is
13:14
Josh: going to get much more full.
13:15
Josh: Another interesting thing about this map is the way that they communicate to each other, EJAS.
13:19
Josh: A lot of the communication done on Earth is done through fiber optic cables
13:23
Josh: when it relates to AI because it's the fastest way to move information as close
13:26
Josh: to the speed of light as possible.
13:28
Josh: The cool thing about Earth is it's vacuum with no resistance or friction.
13:31
Josh: So when these satellites are communicating with each other, even over a distance,
13:34
Josh: the only limitation is actually the speed of light.
13:37
Josh: And the bandwidth with is huge because there is no friction
13:40
Josh: in communication because they are simply sending photons to
13:43
Josh: each other they're sending and receiving photons so the unlock that
13:47
Josh: that enables where now you have a three-dimensional plane
13:50
Josh: of earth where earth is two-dimensional you can only build so much here you
13:53
Josh: turn them to three dimensions and you're restrained only by the speed of light
13:56
Josh: in terms of photons moving back and forth to each other this gets really sci-fi-esque
14:01
Josh: very quickly but also you could start to see how effective this is and what
14:05
Josh: an incredible business oh my god i am stoked for the CEO. Are you an investor at $1.5 trillion?
14:10
Ejaaz: Yeah. Yeah, of course I am. Yeah.
14:13
Josh: Hell yeah.
14:14
Ejaaz: Dude, the TAM, the total addressable market, is literally infinite space.
14:21
Josh: Yeah, but it does not exist.
14:23
Ejaaz: It doesn't exist. Yeah, exactly. Earth will eventually become energy constrained.
14:28
Ejaaz: If you look at the energy bills, Josh, of the local towns surrounding data centers
14:33
Ejaaz: that are in Abilene, Texas, it's gone through the freaking roof, dude.
14:38
Ejaaz: And this is going to be something that we constantly come at odds with because,
14:42
Ejaaz: you know, who needs it more?
14:43
Ejaaz: What's more important, humans that are kind of local and conducting their own
14:47
Ejaaz: kind of business or creating absolute artificial super intelligence for the world to use?
14:54
Ejaaz: I know some people might, you know, argue the latter.
14:57
Ejaaz: So the point is, we need to find energy elsewhere. And the most obvious one
15:01
Ejaaz: is the thing that stares at us in a blue sky every single day.
15:05
Ejaaz: So to round this off, Josh, and for the listeners and watchers of this show,
15:10
Ejaaz: if you want to imagine what the end game of this looks like,
15:13
Ejaaz: it's this graphic that we're showing you on this screen right now,
15:15
Ejaaz: except it's Starlink satellites all over the sun.
15:19
Ejaaz: And it's just absorbing the energy and transmuting that down to build artificial
15:24
Ejaaz: general intelligence on Earth. It's a crazy, crazy vision.
15:27
Josh: Things are getting really sci-fi really quick.
15:31
Ejaaz: And it's... Josh, I had one more. I had one more thing. Just a little nugget, a little Easter egg.
15:36
Ejaaz: Okay. If SpaceX is one of your favorite, AI companies.
15:40
Ejaaz: Give me another one. Just give me maybe your next three or four.
15:44
Ejaaz: What's one company that comes to mind? I want you to say Google,
15:47
Ejaaz: by the way. So you can cut this and say it.
15:51
Josh: Well, clearly Google is one of them. We are huge fanboys. We love Google.
15:55
Ejaaz: Why do you ask that question? I was hoping you were going to say that.
15:58
Ejaaz: They are an official percentage owner in SpaceX to the tune of around 8%.
16:05
Ejaaz: That means they invested $900 million in SpaceX about six years ago.
16:11
Ejaaz: Josh, that is currently worth $111 billion if SpaceX IPO is at 1.5.
16:20
Josh: Trillion. When you got it, you got it, man. And Google just got the juice.
16:23
Josh: Listen, I'm a SpaceX shareholder too.
16:24
Josh: My friend John and I, we worked so hard back in 2020 to find an SPV of an SPV
16:30
Josh: of an SPV to get shares in.
16:32
Josh: And it is now up like 50. I'm so excited to finally have some liquidity.
16:35
Josh: Not that it matters. I fully intend to add more, but like, man,
16:38
Josh: good time to be an optimist.
16:39
Josh: Good time to be an investor in the future because it is getting so weird and so wild. So weird.
16:44
Josh: But if you are new here, every Friday, we release an episode about kind of a
16:48
Josh: roundup of the world in AI. There's more stuff that happened.
16:50
Josh: And this second half of the episode, we're getting into that,
16:53
Josh: starting with what feels like a natural extension of this e-jazz, right? Where,
16:56
Josh: We officially have the first AI data center in space that actually worked.
17:01
Ejaaz: Yep. So this tweet that you're seeing from Philip Johnson, Philip is kind of
17:06
Ejaaz: like the guy that kicked off this trend, I would argue.
17:09
Ejaaz: He is the CEO and founder of StarCloud, which is a startup incubated by Y Combinator
17:17
Ejaaz: with a very bold vision of putting AI data centers in space.
17:21
Ejaaz: The one difference between Philip and Elon is he's actually put a data center
17:26
Ejaaz: in space and he launched it through one of Elon's Falcon rockets last month.
17:31
Ejaaz: And they put an NVIDIA H100 in space. It's currently out there running inference and training models.
17:37
Ejaaz: And it was able to do two really cool things, Josh.
17:41
Ejaaz: Number one, it trained a version, an open source version of ChatGPT from scratch
17:46
Ejaaz: on all of Shakespeare's works.
17:49
Ejaaz: Just for fun, to see if it could figure it out. And secondly,
17:53
Ejaaz: it took an open source version of Google's AI model called Gemma and is currently,
17:59
Ejaaz: right as we speak, inferencing between that model in outer space back to Earth.
18:03
Ejaaz: So you can send a prompt to the Gemma model out there and get a response in
18:07
Ejaaz: a couple of seconds, which is just insane to say.
18:10
Ejaaz: Now, people listening to this might be like, well, an NVIDIA H100,
18:13
Ejaaz: that's an old GPU, who cares?
18:14
Ejaaz: And it's just one of them. This isn't the vision that you just pitched me of
18:17
Ejaaz: a satellite network eclipsing the sun.
18:20
Ejaaz: Baby steps guys this is baby steps the fact that this uh gpu can survive out
18:24
Ejaaz: there dispels the rumors that you know oh you can't radiate the heat or radiation
18:29
Ejaaz: is going to kill the gpu it's it's out there it's been running for two and a
18:33
Ejaaz: half weeks now so this is a very feasible thing uh josh are you excited about this.
18:37
Josh: I'm i'm excited about it it's a great proof of concept
18:40
Josh: it shows like okay shielding works okay heat dissipation
18:43
Josh: works what i really love the most about this is the models that
18:46
Josh: they run uh here on the post that you're showing they trained the nano
18:50
Josh: gpt model from andre carpathy uh to work
18:53
Josh: on those shakespeare things so it's cool like our i mean
18:55
Josh: number one fanboy of andre here he created this like small little nano gpt and
19:00
Josh: now they've created a model in space and there's two models running there there's
19:03
Josh: one that they actually trained and then there's one that they had pre-built
19:05
Josh: that is returning tokens this is awesome it is a proof of concept and it shows
19:09
Josh: that i mean granted i think people have now realized that this is possible but
19:13
Josh: they are actively proving it in reality.
19:15
Josh: And for that, it's awesome. StarCloud seems to be doing cool stuff.
19:18
Josh: I'm very happy with the company, very excited for the future.
19:21
Ejaaz: Yeah, but enough talk about GPUs in space, Josh.
19:25
Josh: Let's kind of like touch. You mean we're going to ground this conversation?
19:29
Ejaaz: I am grounding the conversation. I am touching wood. I don't have any grass
19:34
Ejaaz: near me right now, but I'm back on Earth officially.
19:36
Ejaaz: There is a GPU war going on between our two favorite foes, Josh.
19:42
Ejaaz: The USA, and China. Now, they've had a very tumultuous relationship,
19:47
Ejaaz: mainly because NVIDIA has been restricted from selling their GPUs to China.
19:52
Josh: Which, by the way,
19:53
Ejaaz: Makes up a huge percentage of their revenue every year. So the fact that Trump
19:57
Ejaaz: said no to Jensen was a big deal.
20:01
Ejaaz: Until this week, when he gave the thumbs up to sell not just any NVIDIA GPU,
20:07
Ejaaz: but their second older generation GPU, the H100 or H200 sorry.
20:13
Ejaaz: To China. And it comes with a few stipulations, Josh, which is 25% of revenue
20:20
Ejaaz: that NVIDIA makes from these sales to China needs to go to the USGov.
20:24
Ejaaz: So he's paying a pretty hefty tax on this.
20:27
Ejaaz: And so people were pretty excited. NVIDIA stock jumped up on the news,
20:31
Ejaaz: except on the same trading session in that same day,
20:35
Ejaaz: it retraced because China put out a commentary officially saying,
20:39
Ejaaz: we don't want your gpus and in fact we're
20:42
Ejaaz: going to restrict the amount of h200 gpus that our
20:45
Ejaaz: companies buy from you because they can just use chinese
20:48
Ejaaz: gpus right so that was the response from china
20:52
Ejaaz: saying like listen we don't need it so it was a very volatile uh
20:55
Ejaaz: day in terms of trading the nvidia stock except that
20:59
Ejaaz: then the chinese companies themselves spoke up and
21:02
Ejaaz: said uh hey we're deep seek over here
21:04
Ejaaz: and we kind of need nvidia's gpus
21:07
Ejaaz: because we can't trade oh we
21:10
Ejaaz: can't train frontier level intelligence unless we
21:14
Ejaaz: have those gpus so there's this there's been this big back and forth china in
21:19
Ejaaz: response to that has been like okay let's hold like a council session internally
21:23
Ejaaz: let me talk to some of the companies here and figure out how many h200s you
21:28
Ejaaz: guys actually need and i have a few thoughts about this josh but what's your
21:31
Ejaaz: general reaction before.
21:33
Josh: We get into it. It's funny. So the story isn't the fact that we made them available
21:36
Josh: to China. It's that China said, we don't want them.
21:40
Josh: In fact, we already have them. We have stolen them. We have smuggled them in.
21:43
Josh: So that is a generous offer of you.
21:45
Josh: But hey, we've already been training models on your hardware for a long time.
21:49
Josh: So thanks, but no thanks.
21:51
Josh: And that comes to the surprise of probably no one. Everyone had an idea that
21:55
Josh: a lot of these were getting sent to
21:59
Josh: I suspect what that means is these new DeepSeq models, which have been hyper-efficient
22:05
Josh: in the past, will just become even more efficient going forward.
22:08
Josh: And we've always talked about this constraint that they've had,
22:12
Josh: where they've been resource-limited, therefore they've had to be resourceful on the software stack.
22:15
Josh: In a world where they are resourceful on the software stack,
22:18
Josh: but also have a competitive hardware stack, that feels like a supercharged DeepSeq.
22:22
Josh: And I feel like whatever model they release next is going to be a seriously heavy hitter.
22:29
Ejaaz: Yeah, I mean, there's, it's a few things. There's the performance competition
22:34
Ejaaz: that you just explained, but there's also a strategic political one.
22:39
Ejaaz: And the strategic political one has two sides, okay?
22:42
Ejaaz: So if I'm China, I don't want my Chinese companies hooked on American GPUs.
22:48
Ejaaz: I want them to use Chinese-made GPUs. Why?
22:51
Ejaaz: Because if the Chinese-made GPUs are good enough to create frontier level intelligence,
22:55
Ejaaz: we now have less of a dependency on the US.
22:59
Ejaaz: In fact, it's probably the most important dependency that China has on the US
23:02
Ejaaz: right now, stretching across from raw materials to computer energy to whatever you might be, right?
23:08
Ejaaz: So they want to rely on Chinese made GPUs, not the US made GPUs.
23:12
Ejaaz: On the US side of things, they're thinking of this specific strategy, which is,
23:16
Ejaaz: If we can keep selling China, older generation NVIDIA GPUs, their intelligence
23:23
Ejaaz: will always lag ours and we will always be the leader.
23:27
Ejaaz: The second way to look at that is also NVIDIA has a really sticky software moat.
23:33
Ejaaz: So typically anyone that runs NVIDIA GPUs isn't just hooked to them because
23:37
Ejaaz: they're using the hardware.
23:38
Ejaaz: It's because they rely on the software stack called CUDA, C-U-D-A.
23:42
Ejaaz: I think it stands for Compute Unified Device Architecture, right?
23:46
Ejaaz: The more and more they use NVIDIA GPUs, the more and more they rely on it,
23:50
Ejaaz: Josh. So it's kind of like a drug for them. So this is strategic from the US.
23:53
Ejaaz: They're like, sell them the GPUs, just take a 25% stack.
23:57
Ejaaz: We're releasing Rubin, which is NVIDIA's latest GPUs coming out in 2026.
24:02
Ejaaz: Just keep them hooked on it. Keep them behind. Let's keep them in check.
24:05
Josh: Well, it seems like that's what they're doing. So that's the update on the China
24:08
Josh: front, which is interesting. But there's another update on the hardware front,
24:11
Josh: EGIS, which I'm very excited to talk about,
24:13
Ejaaz: And that is Google's new glasses.
24:16
Josh: Now, I am old enough to remember the Google Glass version 1,
24:21
Josh: which came out probably over a decade now.
24:24
Josh: And at the time, it was viewed as the nerdy Silicon Valley kind of like freak
24:30
Josh: headwear device, and it never went anywhere. It didn't work very well.
24:34
Josh: But all these years later, it appears as if Google has announced that they're
24:38
Josh: getting back in the hardware game, back in the eyewear game,
24:41
Josh: and releasing these classes.
24:42
Josh: So Ejaz, please fill me in. Tell me why these are not going to suffer the same
24:46
Josh: fate as Google S 1.0 or Meta's Ray-Bans, which are still not on your face for good reason.
24:53
Josh: So it's a very difficult thing to build the good eyewear and no one's done it.
24:58
Josh: Why is Google going to make it differently if they even can?
25:02
Ejaaz: So in terms of like specific details, we don't have much information beyond
25:07
Ejaaz: this demo video that we're showing on screen here.
25:10
Ejaaz: They've termed the project Project Aura, which is basically going to be their
25:15
Ejaaz: next gen version of Google Class.
25:18
Ejaaz: And from my initial reaction from this video is it looks very similar to the
25:24
Ejaaz: Apple Vision Pro experience, coupled with a few different features that Meta
25:29
Ejaaz: Ray-Bans kind of introduced in their demo a few months earlier.
25:33
Ejaaz: My take is it's, okay, number one, it's gonna be better than Google Glass. I would hope so, right?
25:39
Ejaaz: From the design of the glass, it looks more casual and applicable to a wider
25:43
Ejaaz: range of waters. It looks a little chunky on her head now that I'm actually
25:45
Ejaaz: looking at the person that's using this.
25:47
Josh: I was gonna say, they look beefy. Those are some chunky glasses.
25:52
Ejaaz: That's definitely way too big, which kind of implies that, you know,
25:57
Ejaaz: assuming this is a prototype,
26:00
Ejaaz: they've got a chunky kind of build-out a physical build-out like the computer
26:03
Ejaaz: running I'm super curious what the sensors look like here how they've been able to optimize for it,
26:08
Ejaaz: aesthetically it doesn't look like the best but functionally it
26:11
Ejaaz: looks pretty cool like again like that Apple Vision Pro experience you can kind
26:15
Ejaaz: of like have several desktop screens across you can access all your favorite
26:18
Ejaaz: apps the number one thing that I'm excited about this Josh is Google has built
26:23
Ejaaz: out a range of different apps and services that I use on a daily basis this.
26:29
Ejaaz: So if I can somehow kind of combine that in with my visual experience on a daily
26:34
Ejaaz: basis, it automatically becomes useful, right?
26:37
Ejaaz: Apple kind of took a step in this direction going from the Apple iPhone to the Apple Watch, right?
26:43
Ejaaz: People didn't have to pull out their phone, they could just kind of like look at their wrist.
26:46
Ejaaz: Google can now have the benefit of like kind of your vision doing this. So,
26:50
Ejaaz: automatically through that, I think it's going to be useful.
26:53
Ejaaz: In terms of viability, I'm going to base it heavily on the Meta Ray-Bans reception,
26:59
Ejaaz: which was absolutely terrible.
27:01
Ejaaz: I think Google has the benefit of being able to scale hardware manufacturing
27:05
Ejaaz: way better or have more experience doing that way more than Meta can.
27:10
Ejaaz: So that's pretty bullish, but it's going to come down to the execution and I'm
27:13
Ejaaz: going to reserve my right to judge for now.
27:15
Josh: Yeah, I'll go on record as being the biggest hater for this product.
27:18
Josh: I think it's going to be terrible.
27:19
Josh: But the thing that I love is that Google's making it. They're working on it.
27:23
Josh: And I'm excited for version three of whatever this is.
27:26
Josh: So whatever they make this year or next year in 2026, this demo,
27:29
Josh: that's not really appealing.
27:31
Josh: Whatever the version two is,
27:32
Josh: maybe the battery shrinks, maybe the glasses shrink. Still not that great.
27:35
Josh: Version three, normally version three of these things start to get good.
27:38
Josh: So maybe 2028 will be somewhere in the reasonably viable glasses world.
27:44
Josh: But it's cool. I'm glad Google's in the game because this is very clearly an
27:47
Josh: important form factor in the future of the way we interface with AI and computers.
27:50
Josh: And there's another company now working on manufacturing them.
27:53
Josh: So for that, it is a win for everybody.
27:55
Ejaaz: And the final item on the docket comes from the open source world.
28:00
Ejaaz: Noose Research, which is a frontier open source intelligence lab,
28:05
Ejaaz: had a really big breakthrough.
28:07
Ejaaz: And by big, I mean really small, big breakthrough.
28:10
Ejaaz: They launched a 30 billion parameter model, Josh, which compared to frontier
28:16
Ejaaz: intelligence models right now, which range from, I think, 700 billion to 1 trillion
28:20
Ejaaz: parameters, this is tiny.
28:23
Ejaaz: Except it did a very big thing. It scored an 87 out of 120 on the notorious
28:31
Ejaaz: Putnam mathematics competition.
28:33
Ejaaz: For those of you who aren't aware of this competition, it is incredibly hard
28:37
Ejaaz: for any of the smartest humans in the world to do.
28:40
Ejaaz: And in fact, comparing its results to the human scores of last year,
28:45
Ejaaz: it would have placed second, scoring a problem set of eight perfect scores, which is just
28:51
Ejaaz: crazy. So the fact that this tinier model packs such a crazy punch is nuts.
28:56
Ejaaz: Now, for those of you thinking, oh, well, it's just a mathematics nerdy model, who the hell cares?
29:01
Ejaaz: Bear in mind that Alibaba spent, I think, upwards of $15 billion in collectively
29:08
Ejaaz: training their latest QEN3 model, which is a pretty amazing open source model.
29:13
Ejaaz: They scored 27 out of 120 doing the same test.
29:19
Ejaaz: So the reason why this excites me is for a few reasons. Number one.
29:23
Ejaaz: Noose Research open sourced this entire thing.
29:25
Ejaaz: So if you're listening to this and you're curious to kind of test this out yourself, you can do it.
29:28
Ejaaz: And 30 billion parameter models is something that you can feasibly run on consumer
29:32
Ejaaz: hardware at home. That's number one.
29:34
Ejaaz: Number two, they did this in a distributed fashion.
29:37
Ejaaz: So typically when you're training a model, you want to build heavy data centers
29:41
Ejaaz: and run a bunch of reinforcement learning, reasoning, stuff like that to create
29:45
Ejaaz: the model. They didn't use any of that.
29:46
Ejaaz: They used a distributed network of compute to be able to create this. So that's really cool.
29:51
Ejaaz: And number three, something that they pioneered, Josh, is something called an agent harness.
29:56
Ejaaz: Basically, the way that this model became so smart is in the post-training phase,
30:01
Ejaaz: where they spun up a number of AI agents, which reasoned with each other in
30:05
Ejaaz: a competitive tournament bracket style thing.
30:09
Ejaaz: And the one with the best solution won and submitted their answer.
30:12
Ejaaz: And that's how they reached this frontier level intelligence.
30:14
Ejaaz: The reason why this is so important is if you assume that compute is the only
30:20
Ejaaz: thing you need to train frontier-level intelligence, this experiment disproves that.
30:26
Ejaaz: This experiment proves to you that
30:27
Ejaaz: you can achieve that same level of intelligence with much less compute.
30:31
Josh: It's awesome they're getting these distilled models to be so powerful.
30:34
Josh: And this is kind of a trend that...
30:37
Josh: Projects towards a loose bear case for ai which is if if
30:40
Josh: you can continue to distill these down into these hyper-optimized
30:44
Josh: models that can eventually run on your phone and does
30:47
Josh: the ai at the edge overpower the ai in the data center and does that mean that
30:52
Josh: the value of these large data centers goes down i don't think so but this is
30:57
Josh: a proof of concept and directionally proving or i guess giving us some data
31:02
Josh: on what that actually means and how that looks this is cool It's a fun project.
31:06
Josh: And the fact that they're doing it in a decentralized way seems interesting.
31:09
Ejaaz: That wraps up the end of today's episode. It was a big one.
31:14
Ejaaz: If you can't tell, Josh and I are slight bulls on SpaceX and Elon's companies.
31:20
Josh: You didn't show off your hoodie. Show off your hoodie. Oh, I didn't.
31:22
Ejaaz: I got my news research hoodie.
31:24
Josh: Yeah, so clearly you see where both of us stand here. Yes. Space maximalist.
31:27
Ejaaz: Yes.
31:28
Josh: Other maximalist.
31:30
Ejaaz: Open source. Distributed.
31:31
Josh: Distributed. Distributed training. Sick.
31:34
Ejaaz: I'm a fan of both. I'm a fan of both. And one thing has become clear to me as
31:39
Ejaaz: we wrap up this episode is that the future of AI is dependent on a number of
31:45
Ejaaz: different protocol layers, Josh.
31:47
Ejaaz: I don't think SpaceX wins without
31:49
Ejaaz: the help of Tesla, without the help of Starlink, without the help of XAI.
31:53
Ejaaz: They all kind of feed into each other and I'm excited to see how this industry builds out.
31:59
Ejaaz: You can bet your ass Limitless is going to be the channel that breaks all this
32:03
Ejaaz: news. I just want us to take a little bit of a victory lap, Josh.
32:07
Ejaaz: We have called out two very important trends months earlier that have become super important.
32:12
Ejaaz: One of them being AI data centers in space, the other one being the Google bull case. Am I missing any?
32:18
Josh: Um, well, another one, we had Blake Scholl on the podcast who does supersonic
32:21
Josh: jets and he just announced a supersonic jet generator for AI data center.
32:26
Josh: So we've been early and we've been right. I mean, granted, directionally right.
32:29
Josh: We first started off as haters, but we've haters of the AI space,
32:33
Josh: but we've changed our mind and we've been covering it along the way.
32:36
Josh: So listen, it might not always be right, but it's always early and you'll always
32:39
Josh: be up to date on everything that matters in this world of AI and frontier technology.
32:44
Ejaaz: Yes. So if you're listening to this and you aren't subscribed,
32:48
Ejaaz: and that is 80% of you, by the way, please subscribe.
32:51
Ejaaz: It helps us out so much. Wherever you're listening to it, it could be on Spotify,
32:54
Ejaaz: Apple, YouTube, whatever.
32:55
Ejaaz: If you're on YouTube, actually hit the notifications button because that also
32:58
Ejaaz: alerts you of the latest alpha. We drop three to four episodes a week. It's awesome.
33:03
Ejaaz: And the final call to action is we have a banger of a newsletter which drops every Friday.
33:09
Ejaaz: And from next week, it drops twice, every Wednesday and Friday,
33:13
Ejaaz: one with an SAE investment thesis on the company that Josh or I are super bullish
33:18
Ejaaz: about and the other highlighting the top five bits of news in AI and Frontier Tech of that week.
33:24
Ejaaz: So you don't want to miss it. If you want to keep up to date,
33:26
Ejaaz: subscribe to the Limitless Ecosystem.
33:27
Ejaaz: We will see you on the next one.