Latest / In the Loop AI / AI Just Saved SpaceX: Record $14.6B Quarter, First Profit and a Tesla Merger

Transcript

AI Just Saved SpaceX: Record $14.6B Quarter, First Profit and a Tesla Merger · In the Loop AI

0:00
I am excited. SpaceX, four months ago, had the largest, most expensive IPO in
0:05
history, and half of Wall Street basically wrote them off.
0:08
They called the company massively overvalued, and yet last night they reported
0:12
their quarterly earnings and a massive $14.6 billion in revenue in a single quarter.
0:19
That is more money than they made in the entirety of last year.
0:23
But what's most interesting is how they made that money.
0:26
SpaceX, as its name suggests, is a space company, yet they made the majority
0:31
of their money through AI.
0:32
Their business line has tripled over the last three months, eclipsing Starlink for the first time ever.
0:37
And the reason why they've been able to pull this off is because they've become
0:40
the number one neocloud, the number one provider of compute to some of the biggest
0:44
labs and companies in the world, including Anthropic and Google,
0:47
selling billions of dollars worth of compute per month.
0:50
And it turns out compute is an incredibly lucrative business.
0:54
Elon Musk is projecting $500 billion to $1 trillion of ARR next year,
1:00
in one year, 2027. But don't just take his word for it.
1:03
JP Morgan also backs this up and has given them a new price target of $500 per share.
1:10
There's a lot of big numbers that we're throwing around here.
1:11
And I dug into the entire earnings report, but also the wider strategy,
1:15
which I'm really excited to jump into because it includes a merger between SpaceX and Tesla.
1:20
And there's a lot of reasons why these companies should merge and eventually
1:24
make them the fifth most valuable company in the entire world.
1:28
So let's get into all of this on today's episode brought to you by Qualcomm.
1:34
Now, four months ago, we made an episode covering the SpaceX IPO.
1:37
And back then, things looked a little bit precarious.
1:41
They IPO'd at $135 per share.
1:44
The stock then sent to, I think it was $171. And then it proceeded to dump more
1:50
than 50% over the subsequent few weeks.
1:54
Wall Street was incredibly bearish and for good reason, because SpaceX couldn't
1:58
demonstrate how exactly they were going to make money.
2:00
Space was obviously a really good vision, but they had a long way to go to kind
2:05
of like prove the rocket tests and get to the point where they could launch
2:08
a rocket back and forth multiple times per day.
2:10
But something happened in the last couple of months that changed all of it and
2:14
it was a simple pivot by Elon Musk which was they acquired xAI.
2:20
And they acquired the most compute out of any AI company, or just company in general, out there.
2:26
They acquired the largest amount of GPUs, which they now house under their three
2:31
core data centers, Colossus 1, Colossus 2, and Colossus 3.
2:35
Now, these GPUs were meant to serve Grok, except that Elon Musk couldn't quite
2:40
get it to work just yet. And in the meantime, he had a bright idea.
2:43
He decided to rent that compute out to people who needed it,
2:47
the likes of Anthropic, who are now paying them $1.25 billion per month,
2:53
and Google, who is also paying about a billion dollars per month.
2:56
And Elon figured that because he was the quickest to be able to scale compute,
3:01
quicker than any other company out there, he could charge a premium.
3:04
Now, at the time of recording, the stock is up about 9%. But I want to get into
3:08
the actual numbers from the quarterly reports and what that actually tells me
3:12
about the company and where they're headed.
3:14
So some of the headline numbers, $14.6 billion of revenue compared to last quarter, which is up 161%.
3:22
Absolutely huge. But the most important point to take away here is that it largely
3:27
derived from AI. The AI segment alone pulled in around $7.6 billion.
3:32
Which is up 830% from last quarter.
3:35
Now, there's a few things to take away from these numbers. Number one,
3:38
it's eclipsed Starlink revenue. And this is a big deal because Starlink has
3:42
for a while been the a major revenue earner for SpaceX. It's eclipsed it in a pretty big way as well.
3:47
Number two, this officially gives SpaceX the first time in its history of achieving a profit.
3:54
And Elon is, as you can expect, doubling down on this.
3:57
This quarter's capex spend was $19.6 billion, and he's aiming for 35 to $40
4:03
billion next quarter. But what I find most fascinating is how this is going
4:08
to translate into revenue for the company, because the projections.
4:12
Are pretty hefty. By the end of the year, SpaceX plans to have around 3 gigawatts
4:17
worth of compute online, which is impressive.
4:20
But what's more impressive is the projection for next year, where Elon estimates
4:24
they'll have around 10 to 20 gigawatts worth of compute.
4:28
Now, there were obviously other highlights of the earnings, including StarMind,
4:32
which is their AI satellite network, which they're launching loads of exciting
4:36
details there, as well as TeraFab, which is the largest chip manufacturing
4:39
plant that will ever exist if it comes to light.
4:42
But the number one story here is how did SpaceX become the number one neocloud in the world.
4:48
When they're a space company? Well, the answer simply is Elon and the company
4:54
is the number one scalar or physical scalar in the entire world.
4:58
The business of being a neocloud is very easy to understand,
5:01
right? You buy GPUs, mainly from NVIDIA. And by the way, SpaceX and Elon is
5:06
one of the, if not the biggest purchaser of NVIDIA GPUs.
5:09
So once you have those GPUs, you house them in a warehouse known as a data center.
5:13
You plug in all the different cables, make sure temperature is regulated,
5:17
make sure there's enough power and energy to feed these GPUs.
5:21
You bring those GPUs online and then you either decide to do one of two things.
5:25
You either use it for yourself to train and inference your own models or you
5:29
rent it out to competitors or people who want that same inference and compute
5:33
and make a hell of a lot of money out of it.
5:35
Now Elon and SpaceX tried the first option with Grok and listen they're making
5:39
good progress but not quick enough progress to warrant investing all of this money into compute.
5:45
So Elon had a light bulb idea. He was like, I'll just sell any excess compute
5:50
that we have to make enough money to still fund all the other things that I
5:53
want to do. And maybe even I'll turn a profit.
5:56
And then he looked over the fence and he realized, huh, OpenAI and Anthropic
6:00
are really struggling to bring compute online.
6:02
They looked at Meta and he said, they're not doing that bad.
6:05
Looked at Google building their own TPUs and were like, good luck with that.
6:08
I'm just going to buy from the best maker and provider of GPUs,
6:12
Jensen Huang, NVIDIA, and then I'll be the quickest one to do it.
6:16
And I'll gain such a huge marginal lead on everyone that I'll have so much excess
6:21
compute to either decide to train some of the best models in the world,
6:24
or I'll just sell it to my competitors.
6:26
And you have to understand, Musk is a master at this, and that's not an exaggeration.
6:32
He's bringing online 200,000 GPUs per month.
6:38
That is around $13.1 billion worth of heavy metal machinery,
6:44
fund this whole AI CapEx build out online per month.
6:48
No other company is doing this. And when he was asked about how he was able
6:52
to do this, his answer was pretty hilarious.
6:53
He said, I'm spending four days a week there. I'm sleeping on the manufacturing
6:57
site floor, and I'm at the Colossus 1, 2, and 3 data centers in Memphis,
7:02
trying to bring all of this online and making sure it gets done.
7:06
What's even crazier is that's not even the hardest part.
7:10
The easiest part actually is buying the GPUs and bringing them online the harder
7:14
part is where do you get the land power and electricity,
7:19
to turn all of these things on because i've mentioned this on previous episodes
7:23
one of the key constraints and bottlenecks in the AI race right now is,
7:27
not buying the GPUs it's switching them on it's powering them on it's getting
7:32
the electrical grid energy and enough of it.
7:35
Not surpassing all the red tape of regulatory burdens to feed your GPUs.
7:40
Amazon actually said this week that they have billions of dollars worth of GPUs
7:44
that are just collecting dust. Microsoft has the same issue as well,
7:47
because they haven't got the permits, they haven't got the turbines,
7:50
they haven't got the energy grid to try and figure this all out.
7:53
This is something that Elon Musk has been hyper-focused on for the last year.
7:58
And I'll tell you a crazy fact, they have repurposed rocket workshops at SpaceX
8:03
to build turbines to use in generators, makeshift generators,
8:08
that they run behind the meter or on-site at these data centers to
8:13
overcome the obstacles or challenges of requesting electrical grid access.
8:18
This is the 3D or rather 4D chess that Elon and SpaceX are playing against their
8:23
competitors, which have allowed them to get a 12 to 18 month lead on compute
8:28
against their nearest second place competitor, which I think is Meta.
8:32
So it's because of this vertical integration that Elon has just obsessed over
8:37
that he's been able to pull off one of the biggest feats that may give SpaceX
8:42
a chance to be a frontier AI lab or one of the top frontier AI labs in the very near future.
8:47
But he has all the excess compute, right? And yesterday's episode covered CoreWeave,
8:51
which also had a lot of compute.
8:53
But people were questioning whether there was actually any real demand for these things.
8:58
Elon answered that on the call yesterday. He said, we get a knock on the door
9:02
so many times every single day from some of the biggest companies in the world
9:06
asking for access to the compute.
9:08
He said if he had multiple times the amount of compute that he has online today,
9:13
he would have been able to sell that compute several times over.
9:16
But the craziest part about this is that SpaceX is charging the premium price
9:21
on every single watt of data that they sell.
9:24
So right now, Elon, I think, sells between $50 to $60 per watt annualized a year.
9:31
Now, if you scale that out to the conservative estimate that they have of 10
9:35
gigawatts next year, that brings you to the $400 billion ARR estimate that I mentioned earlier on.
9:41
But don't take my word for it. You have Anthropic paying them $1.25 billion.
9:44
A month. You have Google paying them $965 million a month. Call it a billion.
9:49
And then I think you have Reflection AI that committed $6.3 billion to get as
9:54
much compute as they can from Elon Musk.
9:56
The craziest part about this is even though he's selling them all of this compute,
10:00
he is still able to have excess compute to train a frontier AI model.
10:03
I think the next Grok model that is going to come out by the end of this year
10:06
is going to be a 10 to 15 trillion parameter model, which let me tell you,
10:10
requires a lot of compute.
10:12
And the way he's been able to do this is by selling short-term leases on this
10:16
compute. So typically, if you're Anthropic or OpenAI, you want to sign a long-term
10:21
agreement with a compute provider so that they can guarantee that they have
10:24
enough compute to train and serve all their models.
10:27
Elon Musk, because he's technically a competitor and he wants to build his own
10:31
AI model, is saying, listen, I'll sell you the compute, but only for short-term
10:34
contracts like three to six to nine to 12 months at a time.
10:38
But I'm also going to charge you a premium to do that. And that's what he's
10:41
been able to pull off because all of these customers are paying that price.
10:45
So the demand is in excess of the amount of compute that is available and physical
10:49
atoms are constraining people for bringing these GPUs online.
10:53
But Elon Musk has found a way to circumvent a lot of this and he's so far the
10:56
only company that has been able to pull this off. So,
10:59
a few things are contingent on this actually being realistic.
11:02
Number one, Elon is able to continually bring on compute in the manner that he has.
11:07
And he's going to run up against a lot of friction here.
11:10
He's going to run up against grid constraints. He's going to run up against
11:13
manufacturing constraints, like cool that he's building his own gas turbines
11:17
and stuff, but that's going to reach capacity at some point.
11:19
Is the demand going to still exist? Right now, he can charge a premium because
11:23
there's not much compute available. But what happens when, I guess the market
11:28
becomes flooded or maybe another competitor figures out a way to scale in the
11:31
same way that SpaceX has done so far.
11:33
And on top of all of that, there's just constraints in manufacturing these GPUs in the first place.
11:38
I mentioned the $35 to $40 billion CapEx spend that SpaceX alone is planning
11:42
to spend over the next three months. So we'll see how that goes.
11:45
We'll see how much Elon is able to acquire and bring online.
11:49
There's still definitely some red flags here.
11:53
If this quarter's earnings is anything to go by, it's looking optimistic.
11:56
So if you're looking at this purely from an investment perspective,
11:59
my own take on this is we're looking a lot more optimistic than we did three
12:02
months ago, but still tread cautiously. Remember, SpaceX has the craziest share
12:07
unlock schedule of any other company that we've ever seen.
12:10
Typically, you see a one-year cliff and then everything starts to unlock.
12:13
The cliff started within a few weeks of the company IPO-ing and they have like
12:17
kind of this weird hierarchy of price targets where if a valuation target is
12:21
hit, then more shares unlock. It's something to be aware of.
12:25
It's not just all what the news reports. So definitely tread carefully because
12:28
obviously flooding the market with all this excess amount of shares is going
12:32
to create the issue of having enough purchase demand to keep the price where it's at.
12:36
But there's also a number of other things that Elon brought up pertaining to
12:40
other parts of the business, including one big one, the merger between Tesla and SpaceX.
12:47
The rumor mill around this started well before SpaceX even IPO'd,
12:52
but Elon finally commented on this entire thing, which gives me a lot of positivity
12:57
about the potential of this happening, arguably maybe by the end of the year.
13:02
So let's get into it on today's episode brought to you by Qualcomm.
13:08
Now, the key sentence Elon delivered on this was, by the request of shareholders,
13:12
both companies have formed special committees to discuss the potential of a merger.
13:18
So you have shareholders and board members, presumably, from Tesla and from
13:22
SpaceX that are now huddling and having conversations about what this merger
13:26
would look like and whether it's actually feasible.
13:28
But there's one important point. Elon, because he has obvious conflicts of interest,
13:32
cannot be involved in these meetings.
13:34
And when he was asked whether he thought this was a good idea,
13:36
his answer was simply yes.
13:38
And it was yes for a very good reason. Because when I think about the synergy
13:42
of Tesla, a car automotive company, and SpaceX, a space company that's kind
13:47
of doing AI stuff and creating some kind of nebulous StarMind satellite network,
13:52
the connection doesn't seem quite clear.
13:53
So let me try and do my best at explaining what the obvious synergies are.
13:57
Number one, SpaceX is trying to make transport to space in and out of space,
14:02
actually, the cheapest possible thing alive.
14:06
So basically, they'll effectively own the highway of getting things into space.
14:10
But then the next natural question is, so what? What do you want to do in space?
14:15
What are the products in space? Well, number one, it's Starlink.
14:18
Starlink is basically a distributed network for connectivity to the internet,
14:23
for any device and any human on Earth and maybe off of Earth.
14:27
So the idea here is it's a network that would beam internet down to Earth.
14:31
Across all dead zones so that you can be connected to the fastest,
14:34
cheapest internet ever.
14:36
And so that you don't have to be reliant on telecom infrastructure companies
14:40
that have myriad different issues between themselves.
14:42
But then you have StarMind. StarMind is their AI satellite network,
14:46
which is currently being custom built with NVIDIA and also what's going to be
14:50
a large product from the TeraFab thing, which I'll get to in a second.
14:53
But the idea here is you'll be able to launch a specialized AI satellite launched out towards the sun.
15:00
It harnesses the energy of the sun and is able to inference and or train AI
15:06
models up in space, which sounds completely crazy because there's a lot of constraints, right?
15:11
There's the radiation issue. There is the how the hell is it going to be survived
15:15
and maintained out of there? There's the big question around how these GPUs talk to each other.
15:19
Over the last year, SpaceX has been thinking very deeply about this,
15:23
and they've been able to figure out some really convincing prototypes.
15:26
But of course, the big question mark is whether they'll be able to do this at scale.
15:31
According to the timelines that Elon has cited, we'll be able to see versions
15:34
of this come alive next year over the next 12 to roughly 18 months.
15:38
But the vision here is you would have, similarly to Starlink,
15:41
a network of AI satellites out there harnessing the power of the sun,
15:46
very Kardashev Type 2 civilization, at the cheapest possible rate.
15:51
Now, bear in mind, if you want to get your hands on that kind of energy,
15:54
one, you run into a lot of red tape.
15:56
Two, the cost for this thing is crazy. And three, it's governed by so many different
16:00
parties that it's just a pain to deal with.
16:03
So Elon's solution to this is project the GPUs out towards the sun,
16:07
harness all that energy 24-7 for the cheapest rates possible,
16:12
and then compete with all the data centers, compute providers,
16:16
and energy providers down on earth. And the idea here is that.
16:19
There is no competition. If Elon is able to transport these GPUs out there,
16:24
create that network, and in fact, harness all that energy, they will end up
16:28
with the best offering ever.
16:29
And if anyone wants to compete, they need to go through SpaceX.
16:32
So it's this crazy vertically integrated stack that would help create connectivity
16:38
to the internet via Starlink and connectivity to the cheapest compute inference
16:42
and training for any AI model.
16:44
And that could be Grok, but that could be also for all the competitors that
16:47
Elon Musk and SpaceX is currently selling compute to
16:50
imagine Anthropic in the future they don't want to spend crazy prices for very crappy
16:56
compute on earth they would just pay Elon and Elon can charge a cheaper price
17:00
but still make wider margins because he has it at a cheapest cost versus any
17:04
of his competitors okay so what does this have to do with Tesla
17:08
well Tesla creates all the devices and i use the term devices very specifically.
17:15
You might think that a Tesla car is a car and very much it is,
17:19
but it's also a computer, a computer that is connected to other computers, other Tesla cars.
17:26
And we have currently millions of Tesla cars roaming the earth.
17:29
A lot of them have full self-driving capability, which means that it can autonomously
17:34
drive better than 99% of humans on earth.
17:37
So the idea here is it has cameras and sensors around it.
17:41
It is a computer that is fully functioning. You can run Grok or the Frontier
17:44
Grok AI models in the car right now.
17:47
You can connect it to any of your devices, any of your apps.
17:49
It creates this coherent system, almost like Jarvis from Iron Man, right?
17:54
And so you need physical infrastructure and devices to be able to serve all the AI.
17:59
And in order to serve all that AI, and in order to connect all these different
18:02
software products to humans, to what we're interacting with, you need internet.
18:06
So it makes sense that all of these Teslas, all of these Optimus robots,
18:10
these humanoid robots, and any other device that Tesla builds can be one,
18:14
connected to each other on Earth, and two, connected to the number one cheapest
18:19
AI inference service provider.
18:21
And the internet. And if you're SpaceX, remember you have that thing called
18:24
xAI and you are training your own AI models.
18:27
So what do you need to train the best AI models? It's not just compute, it's also data.
18:32
If you look over at Tesla, Tesla has all the camera and sensors that are pulling
18:35
in data 24-7 from around the world wherever Tesla operates.
18:41
And so feeding that data into the model can be incredibly valuable for future
18:45
model iterations. But there's also a lot of manual labor that needs to be figured out.
18:49
Humans, for all intents and purposes, are limited. You need to train them from
18:53
scratch. You need to make sure they're qualified to be able to build and do
18:56
the thing around data centers, around AI models.
18:59
It is a scarce resource. But imagine if you could automate at least just the
19:03
physical labor side of things.
19:05
You could do that via the Optimus humanoid robots. Now, of course,
19:08
I'm speaking at a very great length of time in terms of like making these things happen.
19:14
It's not going to be easy to pull off. But the idea here is,
19:17
if you end up in a world with hundreds of millions, maybe even billions of humanoid
19:21
robots that can automate a lot of the physical labor.
19:23
And then you can own the compute and maybe even the data that you can deliver
19:28
to create the best artificial intelligent models in the world,
19:31
and then you can pair them together,
19:34
you end up with effectively what is a superhuman system that can help advance
19:39
humanity well beyond where we're at today.
19:41
But there's one more thing holding them back. Chips.
19:44
You need physical hardware to house the brain for all of these different things
19:48
whether it's the AI model to train and inference it or whether it's the humanoid
19:52
robot or the car to power it who's going to build those chips could be NVIDIA or it could be
19:59
whatever the merger of SpaceX and Tesla is which we've seen very clearly most recently
20:04
TeraFab. TeraFab is basically the most ambitious effort to try and build the
20:09
biggest chip plant specifically for AI and compute and devices in general on earth.
20:15
As you can imagine if we do have that world of a billion robots whether that's
20:19
cars or humanoids and then we need enough compute to power all those things
20:23
you're going to need billions of chips
20:26
and right now if you combine Nvidia or any of the other chip fabs such as Intel or AMD.
20:32
You're not going to get anywhere close and And if you are one company that is
20:35
reliant on producing a lot of these things and needing those chips,
20:38
you may not want to rely on those companies who could price you out or whatever.
20:42
So the best way to figure this out is to vertically integrate it,
20:44
which is why this is such an ambitious effort. And he commented yesterday on
20:48
the first phase where he said $16.8 billion in the first phase,
20:52
100,000 wafers a month before 2030.
20:55
But at its peak, 1 million wafers a month within 10 years, which will also create
21:00
around 3,000 plus jobs. It's currently doing that right now.
21:03
Now, I want to end this episode by bringing myself down to earth and grounding
21:07
myself for a bit because I want to be clear here, there's a lot of ways that
21:11
this could potentially go wrong.
21:13
My list here is capex of $35 billion to $40 billion. Where are you getting that money from?
21:19
Last time I looked at Tesla's balance sheet, they had about $43 billion.
21:23
Is he going to self-finance it or is he going to raise more money?
21:26
Rumor has it is he's trying to raise more money right now to the tune of $30
21:29
to I think $50 billion before the Anthropic IPO. I don't know how far he's got
21:33
along with that, but the point is he's trying to raise as much money as he can to fund that CapEx.
21:38
That's still a big red flag. He may not be able to pull this off and if he doesn't,
21:42
it's going to delay or at worst, stop any of these efforts entirely.
21:46
Number two, transitioning from GB300s, which is where the bread and butter is,
21:50
bringing on 200,000 online every month, to Vera Rubins might actually be way
21:56
more difficult technically.
21:57
Now, NVIDIA or Jensen Huang went on stage yesterday, or maybe it was this morning,
22:02
it was in Europe, in Germany, in Berlin at GTC, and he went on and said,
22:06
this is the easiest GPU to bring online.
22:08
So it's not only the smartest, but the easiest. It takes around five minutes,
22:11
no cables. its self-contained cooling system.
22:14
Now, I don't think that's strictly true, and obviously he's probably being very
22:18
lenient in his explanation, but if history is anything to tell us,
22:22
it takes a while to actually bring the next-gen GPUs online once they've actually been produced.
22:26
So this may further delay things. And then the final thing is obviously you
22:30
have to take Elon's estimate with a huge grain of salt.
22:33
You have J.P. Morgan on one hand predicting $254 billion of revenue next year, which,
22:38
granted, is huge for SpaceX, who previously forecasted a massive amount of revenue
22:43
but had no way of justifying it, then versus Elon's conservative estimate of
22:47
$500 billion to $1 trillion of next year.
22:51
A lot of it remains to be seen. I'm going to be sat here regardless with my
22:54
popcorn watching all of this unfold.
22:56
And I just want to say it is crazy to be talking about these figures at length realistically.
23:03
If we wind back even five to 10 years, the notion of making $15 billion in three
23:10
months was almost unheard of.
23:13
Definitely reserved for just one or two companies at the time.
23:16
And now we're seeing this happen to new companies that pop up after the most
23:21
expensive IPOs or even private companies such as Anthropic and OpenAI who are
23:26
rumored to be making IPOs.
23:27
Tens of billions of dollars per quarter that are going to be having blockbuster
23:30
IPOs themselves over the next couple of months. I heard that Anthropic potentially
23:34
files publicly at the end of this week.
23:37
It's insane. So that is the episode. The earnings dropped last night.
23:40
I was up all night. And hopefully this was helpful to you. If you enjoyed this
23:43
episode, give us a thumbs up.
23:45
There's actually a lot of newbies here, a lot of new people that have signed
23:48
up and subscribed and are watching our content. Thank you so much and welcome.
23:53
This has been a whirlwind of a journey building up this channel and building up all this content.
23:56
But I'm trying to bring the kind of best takes, news, insights on all the things
24:01
happening in AI and frontier tech every single day.
24:03
And one final point I'll leave you with is it's funny making the episode that
24:07
I did yesterday, which was about CoreWeave, who's also in Neocloud,
24:11
struggling to justify their valuation and tanking the entire stock market
24:15
versus SpaceX that is doing arguably the same thing just at a much larger scale.
24:19
And maybe they have the money to back it, but they have crazier forecasts for revenue.
24:23
And there's a much more positive reaction because they're generating so much revenue.
24:26
The demand is obviously there. People are paying for this compute.
24:29
People want to be building AI
24:31
models and serving it to others and building different products off of it.
24:34
So there's just this weird dichotomy that exists right now where the market is tentative.
24:40
They're worried that we're in a bubble. And then you have SpaceX kind of proving
24:43
that the revenue aspect, at least in the interim or over the last three months,
24:46
is very much there. So it'll be exciting to see all of this play out and I'll
24:50
see you on the next episode of In The Loop brought to you by Qualcomm.