원문 보기
The CEO of SpaceX just made some very optimistic statements about SpaceX. In this video, I'll break them down and outline the aspects of the company that are becoming cheap, but also undervalued. So, this will be very useful. Let's break this down. It was an interview. So, this was an interview with Brett Johnson for Goldman Sachs.
He is the CFO, that is, the chief financial officer of SpaceX. It's probably worth between one and a half and two billion dollars now after that IPO. So the guy feels very confident. I don't think he has any reason to come here and talk nonsense . He is very optimistic about SpaceX. So we'll have to add a little bit of my skepticism and apply it, as I usually do, my criticism.
But I have to say that overall this guy seems like a straightforward person . He has no reason to fool us. And I was actually very impressed by some of his comments. So I'm going to go through my resume and show you certain snippets that I think are very helpful. The first comment is that he is a big supporter of not only the vertical integration that they do in rockets, but also the same in data centers.
And so I think he didn't go into it in depth, but essentially he said, " We like to be our own general contractor. We like to be as vertically integrated as possible in our data centers and our rockets." And for me, it was related to what we read and learned yesterday: SpaceX is slowing down data center construction to get rid of the technical or organizational debt accumulated from being their own general contractor, to learn from their mistakes and improve.
This is similar to how we said earlier that writing is recursive. You go back and improve your essay every time you rewrite it. Now we are talking about recursive learning in artificial intelligence. I think they do something similar as a company. And that's exactly what companies should do. They have to continue to evolve, so I was optimistic about that yesterday.
And I agree with him when he says that this kind of improvement and acquisition of expertise gives them a data advantage that others don't have . And I personally believe that they have much greater financial capabilities than CoreWeave, Envis or anyone else. I'm not trying to criticize them, understand this correctly .
They're all in the same boat, so to speak, you know? If one of them is lucky , they all rise a little together. I'm just saying that from SpaceX's perspective, they're going to have a lot more financial leverage than these smaller players, which I think allows them to better leverage their scale and profitability.
Or with a better result. For example, they will be able to scale their data center construction so that they will ultimately generate more profit than some of their competitors. I found this very interesting. Another thing I thought was cool: at about 4 minutes and 30 seconds, he talks about how the V3s will essentially be included in the next Starship 14 flight.
This is a pretty important point, because that's the whole point of Starship, because the Falcon rockets can't accommodate the V3 satellites. These are the Starlink V3 satellites, thanks to which we could get 5G on mobile phones either with better quality or, in fact, better bandwidth, which we are counting on.
Many are spreading comments by the CEO of T-Mobile, who claims that it is physically impossible to ensure the operation of 5G from low-orbit satellites due to the peculiarities of signal penetration through energy-saving glass, so 5G from satellites will never work indoors. I believe the Starlink team will make optimizations that will allow this to be implemented.
I wouldn't bet against Elon on this issue. I think if you combine the medium bandwidth that they usually use for these low-orbit satellites with a mix of low bandwidth , or "low range" as they call it, and also terrestrial transmission. This essentially means that we will mix the bands to pass through building walls and glass , providing you with the service you need.
Or we can send low-latency signals to ground towers—that is, regular cell towers and 5G towers around, maybe using... I don't remember exactly what they call it, but there was something like homing that satellites can use to know where these ground 5G antennas are and send concentrated signals to them to give you 5G on your phones.
All of this together, these strategies will likely solve a lot of the problems with 5G that people are currently talking about, like it's technologically impossible. My guess is that Elon will find a way to solve this using some variation or combination of the things I just talked about. So, launching these V3s on Starship 14 is a truly landmark moment.
And they affect how we move from reusable boosters to the reusable Starship cargo ship, which is why this is such a big moment. And so I think it's worth listening to a little bit about it here. already on the 13th flight, but the knowledge we gained then will carry directly over to the 14th flight and beyond.
And then we have another flight later this month, the 14th, which is very exciting for the CFO because it's going to be a profitable flight, right? I love demonstration flights, and now I'm excited that we're moving to production and launching production Starlink V3 satellites into orbit on our next flight.
So I think it's going to be huge when they say it's a revenue flight, I think they mean using the V3 satellites, and when we deploy them, we'll be able to make money selling Starlink. I don't think another customer is actually paying for a seat on Starlink anymore. Sorry, not on Starship yet. But that's nothing .
You know, a dollar from here or a dollar from there—it's still a dollar from somewhere, right? So I think it's exciting. Another thing he talks about is that many people think that orbital calculations are a matter of many years. I want you to know, I think so too. I think space data centers are a great long-term vision.
I think they are brilliant. I think placing them in the terminator zone, where you have virtually unlimited electricity thanks to 24/7 sun and massive solar panels, is just brilliant. I just think that this is a 2035 perspective. So he's literally addressing me when he says, "A lot of people think it's still many years away ." Yes, because I don't know how you're going to solve thermal problems and the actual technology for that so quickly.
Maybe they've already made it all up . I don't know. But I see this more as a 2035 thing. So I'm building patience into my predictions. This is very important for SpaceX. If I'm wrong, it'll just happen faster, right? But if I'm right, then fine, at least you're planning for it to take a little longer .
But he is much more optimistic. And I have a feeling he's not talking nonsense. I'm still hesitant, but I don't think he's talking nonsense when he says we plan to launch these orbital data centers starting in late '27 and start putting them into space on a large scale starting in 2028. It's a little overwhelming because even the CFO is now adopting Elon's style, like, oh no, this is really happening .
So it's not just Elon saying this, because we know that Elon's timeframes are optimistic, but they're usually a little tighter compared to what we end up with . But now the CFO confirms it. So I find it very interesting. He also talks about the new deal they just signed that starts in December, and that's one of my biggest concerns about SpaceX— a lot of analysts are pricing SpaceX at $40 billion per gigawatt, you know?
You've probably heard me say this a million times. So, interestingly, 40 billion per gigawatt gives a payback period for computing power of approximately 1.1–1.2 years. If you spend $50 billion on a data center, that would be 1.25 years. That means it would take five quarters to recoup your GPU obligations if you could earn 40 billion per gigawatt.
But what I found really interesting was how transparent he was about what they see now in these new agreements that they're signing. At about 15:50 in this video interview, he claims that he doesn't have access to data on how much money Anthropic and OpenAI are actually making, but from what he sees, he believes their monetization is higher than that amount, and that they are capable of supporting, as far as he can see, between 30 and 50 billion per gigawatt.
This is the most optimistic statement I've heard from them, because when I look at SpaceX, I'm trying to hedge a significant discount. I'm trying to model 15–30 billion as a sort of worst-case scenario. And honestly, even if you get $20 billion in contracts, which are obviously fueled by OpenAI and Anthropic, you still pay for that hardware in 2.5 years, right?
That would be 2.5 years. That would be 1.25 years. That's still pretty good. All of this is based on the " canary" in the mine. All of this is built on Anthropic and OpenAI. If they "step back" and we get much more information about them once they become public, it will be a red flag. Place trailing stops, exit the market, the bubble is about to burst when the big players start to fold.
But at least from this interview we understood that we are not even close to collapse, because we have a clear vision that we will continue to be paid. OK. At least this will allow us to go through the IPO stage when we get more clarity on this, I think. But let's listen. For me and my finance team, this is an interesting case study because you're actually building a model: "I'm going to have this much computing power in watts, and here's my guess as to how I'm going to monetize it," and we know for a certain period of time what the range is going to be.
I think during the conference call we talked about $30-50, which is what we expect. True. We have known this for a period of time. It's like...if it's not true, it's a scam. So I really think he's right. For a while, he says, we've been at the upper end of that $30 billion to $50 billion range. Damn, I take my hat off to them.
I didn't believe it at first because I saw that computing power cost, you know, $6 billion to $11 billion per gigawatt. But the forecast is that they will be able to make money from this for some time to come. next year regarding the monetization range. And the fact that we were at the upper end of that range is probably the correct wording that we used.
If not, then we are at the upper end of the range. And, essentially, it now sets the bar for our internal team. Yes. Is n't that right? Because you don't want to switch to internal solutions and experience any deterioration in monetization. You want to have improvement. But I mean, I don't have financial figures.
So that's a fantastic phrase too. He's essentially saying, " Listen, we want this monetization range to go up, not down." Of course. Here's the main caveat, and it's one of the reasons I really think this guy is legit. The CFO will be a little more cautious than Elon. Well, Elon is not the corporate, you know, cautious type .
He is a visionary. I feel like I'm...well, I like to think of myself as a kind of combination of both. As a visionary in what we do and how we look at things, but also with a critical thinking style like a CFO. You know, that's how I like to run my companies: we do visionary and new things, but we're also going to be conservative, realistic, and we're not going to tolerate any crap.
So, I like that this guy hedges his words because, first, it makes sense, and second, it's transparent, which makes me believe more in the rest of what he says. Some of these big players like Anthropic or Open AI again. Wait. Wait. Wait. Because I interrupted it a little . But I mean, Here . has no degradation associated with monetization.
You want to have improvement. But I mean, I don't have the financials of some of these big players, like Anthropic or Open AI, in terms of transparency of their finances, I would say. I think our financial performance could be better, but it doesn't matter. But He is probably right. regarding their finances to know exactly what they are getting from monetizing dollars per watt, but from what we are seeing, it is above the range, right?
And that's why it makes sense for them to turn to people like us to get computing power. And it was a wonderful virtuous relationship, right? We have very strong relationships with Anthropic and Google , for example. Um, and so you see that and you say to your team, "Well, look, if you just forecast this range for monetization, then when our tools start to ramp up , it's actually going to be an improvement on that range, right?
When basically, margin expansion, that's why in yesterday's video we talked about the EPS estimates for this company going up. Look, I like SpaceX, and I always have to be transparent about my bias. Obviously, the more this thing goes up, the more your guy Meat Kevin makes. Uh, and the people who invested in, uh, our venture capital fund.
Because again, we got into SpaceX at a $350 billion valuation. I wish I had gotten in earlier. I wish I had invested more, you know? But it's always "if and if." Uh , having said that, uh, some of those things, we actually saw de-risking in two ways post-IPO. The stock actually fell 30% from, well, its post-IPO highs.
Even more than that, for a short period of time. Well, that was, well, expected. Well, but it's a form of risk reduction. In addition to the risk reduction of stocks that, you know, didn't skyrocket to crazy heights after the IPO and stay there, the price reduction is also a form of risk reduction. And then there's the risk reduction from the greater stability of the annual recurring revenue and the ability to better predict earnings.
That's what they're talking about, expecting to hit $100 billion in ARR by December 1st, because they just signed another contract for $1.11 billion per month starting December 1st. They also made a very important hedge, or let's say, they opened up a very important risk insurance tool. It was related to 90- day contracts.
That's where the real risk of the bubble bursting is: these are just 90- day obligations with the option 90- day termination. I don't think Anthropic or Google will cancel any of their projects until they start monetizing. It does n't look like that's happening yet. It could happen, but I 'll keep an eye on it before I get too excited about SpaceX.
Now listen to this moment right here. Yeah, and almost all of them, if not all of them, were essentially 90-days with a 90-day option to get out . So about 6-month commitments. In fact, the one I just mentioned that we did earlier this month has the same structure and timeline. And I think a lot of that is because we have a lot of internal confidence in our own products.
In fact, if you look at our own product cycles, it's great that we closed the Cursor deal just a few weeks ago. They're talking about Grok Bot and Cursor, and how Starlink is going to reach more cruises and airlines, which will give them additional marketing , and then there's this whole blah blah blah cycle.
They're purposefully becoming exclusive partners with Nvidia to get more chips, especially for orbital computing, and all that. Some of it was just at the margin of error . But I really like this interview. And I compare this interview to Satya Nadella's speeches. I also have an investment in Microsoft.
But Satya is the guy who says, " Look, if our investment in OpenAI goes down, it's okay . We'll handle it." Which to me suggests that he also wants to be a little cautious when it comes to artificial intelligence. But they just did something that makes me think, "Wait, is a conservative getting that much more optimistic?" This is interesting.
So what happened is this : Microsoft is reporting that they're going to triple their data center capacity. Uh, so this is the report... What do we have here? Yeah. So they have 30... Of course, behind the paywall again. I hate it. And I don't know why it keeps logging me out. I hate it when it keeps logging me out .
How many times do I have to log in to this crap? Anyway. Uh, so I have my notes here. Microsoft has 12 gigawatts of computing power. Of that 12 gigawatts, only 2 gigawatts are currently AI. That's crazy data to me . So of the 12 gigawatts that they have, uh, only two are AI. Well, they expect to reach 32 gigawatts total now.
Which means a tripling of total computing power. But it's actually a 16-fold increase in capacity. AI, as we go from 2 to 32. Uh , what that means is you're adding... that's 32 total . Take that out of there. You're adding 18. Plus 18 gigawatts, right? That's a lot more damn compute power. And that just confirms to me, "Dude, Nvidia feels cheap." And I've been complaining for months now that Nvidia feels cheap, well, cheap below 227.
But I can't help but say that I think it's getting close to a level where Nvidia could break through soon. This ceiling right here, I don't think it's going to hold for much longer. And if I'm right about the Trump deal, the end of the war, and that we're not in a recession yet. Nvidia is a possibility, but also, I know, it sounds like a shoestring or whatever.
The data is indicating, the facts are indicating, that the bottom for SpaceX may have already been reached. These 133 is probably a solid bottom. I know the price went down to 104. I call it 133 because I think that bottom, the bottom of 104, was driven by momentum. Everybody's selling, shorts are making money, et cetera, et cetera.
That kind of crazy momentum down, I don't think it's going to happen again . Uh, barring a recession or some other crap. It's possible. But, uh, I have to say, I might be getting a little too optimistic. I don't know, maybe you should go with the reverse Kevin, but, you know, I could always end this by saying, "Oh, wait, there's no promo code here." Well, if there's no promo code here, what am I supposed to do?
Oh, I know. Oh, man, your boyfriend Kevin just lost a ton of money. Oh, but that's nothing. Why don't you promote this stuff that you told us about here? I don't think anyone else does n't know about it. Well, .
Comments 0