SpaceX Earnings Estimates Shock: Full Breakdown & Transcript

[00:00] There's been a massive inflection point on the earnings estimates for SpaceX, and this is shocking. What looked like a company that was expected to lose money for years at IPO time and before IPO time,

[00:17] is now, based on these new earnings estimates, expected to look profitable way sooner, and you could almost argue, cheap.

[00:29] Now, this is a crazy flip-flop, and I'm going to show you the actual underlying stats and data as well as what's going on with market rates and comps for these data centers. But we've got to go through some of this because this is a lot to digest.

[00:44] I just want to be clear about my own exposure. Our VC fund that has exposure to SpaceX, which we've had since about $350 billion of market cap. Right now, what is it trading for? $1.5 trillion or whatever,

[00:56] or maybe even more than that. But the point is, not a single share, oh, it's almost $2 trillion now, not a single share of that has been sold yet. We don't even have, they haven't even all arrived yet. Like, the lock-up process is still in the unlocking phase.

[01:12] So, our original thesis on SpaceX is always centered around space exploration, which we personally think is like a 2035 to 2040 launch factor,

[01:25] and then Starlink, Starlink Loads. That's been our original outlook for evaluation for this. And then we've been looking at analyst expectations. In fact, these are some of the things we wrote. We're like, hey, you know, end of 2030, we could get some launch service revenues.

[01:40] We could get some Starlink revenues. Hey, we can grow this operating income of like $20 billion. That justifies about $150 price.

[01:52] But, but, but, but, this is now old news. And I'm not even talking about the launch spreadsheet we've been talking about, about how if we actually scale launches with our Starbase Louisiana,

[02:06] 10 launch sites, Cape Canaveral and Kennedy at three launch sites, Starbase Texas, two launch sites, and how margins for this could generate scaled $80 to $160 billion.

[02:20] That's sort of like pricing out how much are these launch services going to be worth in the 2030s decade. This is how you can get towards these multi-thousand dollar price targets through just launch services.

[02:32] But let's table launch services for a moment. Because Wall Street is adjusting earnings forecasts for SpaceX entirely based on the AI data center build-out.

[02:45] And there's some news that got us going down the rabbit hole a little bit in addition to prior research on Anthropix's half-trillion dollar compute plan. more than that in just a moment. But let's start with that, sort of what spawned all of this.

[03:01] What spawned all of this was the following. SpaceX overhauls data center build-out, potentially slowing expansion. So when you first read this, it's like, oh, that sounds bearish. That means they're not going to be able to build computers fast.

[03:14] And after all, they still have to raise $400 to $500 billion to build out their 10 gigawatts of compute that they want, which means they're going to have to raise debt.

[03:26] And if they're going to raise debt, there's a chance the analysts are trying to sort of pump up some of the estimates for the company to make the debt a little bit more attractive for not only getting commissions, but sort of the big banks. It's not how it's supposed to work, but let's be real, let's hope it works.

[03:40] And then, of course, any decline in interest rates that SpaceX should get by having more of an investment-grade rating saves them a lot of money on interest. So, when you first see this headline, it's like, oh, that's not good, slowing the expansion,

[03:55] because there's literally talk right now that Anthropic wants to spend up to $517 billion in commitments between the end of 2025 and the end of 2025. Now, Anthropic is a little bit of a slut when it comes to compute.

[04:09] They literally rent from, like, everybody. They rent from AWS, they rent from Google, they rent from Microsoft, SpaceX, Atari, they're from AMD, N-scale, Lambda, anybody they can get their hands on they rent from,

[04:22] because they just want it all. Their deal with SpaceX actually, quote, substantially increased their compute, per the Bloomberg reporting for this, or for that particular note.

[04:34] But what's wild is that Anthropix, driving half a trillion dollars of compute spend over the next decade, that's just one company, if you add opening R, you're probably adding another $750 billion that they want to spend by 2032, I believe.

[04:49] You put that together, you're starting to knock on the door of $2 trillion of compute spending by just opening AI and Anthropic over the next few years. These are insane numbers. They're so freaking huge.

[05:01] But anyway, Anthropic told investors last December that they wanted to spend, I'm going to write this number down because it's really impactful to see what has changed. and it'll show you why the analyst expectations are changing for SpaceX.

[05:17] Last December, Anthropic told analysts that they wanted to raise about $180 billion for compute purposes. That has now been updated to an estimate of $517 billion of compute.

[05:36] That's in addition to opening eyes $750 billion or potentially more. That's pretty remarkable. So far, Anthropic has like 1 to 2 gigawatts of compute as of last October.

[05:48] So write that down over here. Okay, last October, they've got about 1 to 2 gigawatts of compute. They trying to get to a capacity based on what they locked in already So this is now locked in which is backlogged capacity because it has to be built of 14 gigawatts

[06:08] That's insane. Anthropix specifically went to Microsoft for a $30 billion deal on a gigawatt of Azure NVIDIA services. Of course, they've got the partnership with now recently announced Fluidstack as well,

[06:22] but I didn't mention Terawolf is in there as well compared to some of these other ones that Anthropix is lining up deals with. They're running a massive deal. Anthropix compute, by the way, at $517 billion to $418 billion.

[06:34] Opening Eye by 2035 is targeting $30 billion. So almost double what Anthropix is saying. Anthropix numbers are already insane.

[06:47] So Opening Eye and good old Sam Altman are getting even more wild. So some of this could actually explain some of that surge you're seeing in opening eye, or in Dell stock, for example, just in enterprise AI and neocloud build-out or sovereign build-out.

[07:03] Everybody wants these data centers right now. And what's remarkable is that's finally starting to show up in the analyst expectations for SpaceX, and they actually take what looked like now an overvalued business

[07:17] and make it look cheap. This is crazy. And again, we're going to explain these charts we have right here in the Meet Kevin app for course members.

[07:29] We don't have a coupon code right now. I'm not trying to pitch you a coupon code or anything. I'm going to explain all this in just a moment. Before we explain what just changed, You have to know what problem has occurred.

[07:42] So let's start at the problem. So last week, there was an outage because of the Memphis data center, SpaceX responsibility. Brock went down.

[07:54] Claude went down. It was a problem. This is not the first time. Earlier this summer, right around IPO time, there was talk about SpaceX rented out computing after its own teams had trouble using it.

[08:09] So you have Ed Lovell here arguing that SpaceX decided to rent out the full capacity of its Colossus 1 data center to Anthropix after encountering technical challenges using the facility to develop and run its Glock intelligence models.

[08:23] The firm encountered latency issues connecting Colossus with two other sites located more than 10 miles away. training requires ultra-fast connection, and using older links can create delays that slow the entire cluster down for training purposes.

[08:40] Anthropic could be using this, which we believe, for inference purposes, so maybe that latency is less important as they sort of expand the credits you could use for Fable or whatever. Now, obviously, the Colossus Facility, there's a lot of bragging about how this was built in 122 days.

[08:55] It's important to remember this facility is an older building that has pre-existed. So they just had to come in and move the chips in and throw some temporary gas turbines outside.

[09:07] I'm oversimplifying, but that's roughly what happened. Most of the time when you hear a data center takes years to build, it's because there's just dirt. There is no parking lot. There's no pavement. There are no walls. There are no existing air conditioners.

[09:19] There are no existing, you know, energy grids, whatever. Oh, that has to still be actually permitted and brought in. This is like, hey, let's rent this and move chips in. That's kind of how they got to about 122 days here. But they're running into some problems with that speed.

[09:33] So Elon is now doing what I think is actually really smart. Like, I respect this. Elon is meeting, apparently, for this information piece, regularly on Sundays,

[09:46] he'll go out and meet engineers at these facilities to kind of, somewhere in here, it's on Sunday, to kind of supervise this build-out in a better manner.

[09:58] In recent months, he's visited the facility in Memphis on a nearly weekly basis, stopping by most Sundays. He most recently visited just this past Sunday. And so what's happening? Well, Elon's decided to shake up the team.

[10:12] Apparently what they're doing is they're kicking out a bunch of the old sort of data center architects, and they're bringing in SpaceX engineers and they're saying, look, we can't have latency issues.

[10:24] We need to have redundancies. We need to do this right. We can't have a bad reputation of Claude going down on our stuff because then they're going to cancel our contract. Now, I don't think they're actually going to cancel our contract,

[10:36] but they're going to be pissed and they're not going to be willing to pay us as much money. If we're going to want to drive a premium compute revenue source, we are going to have to provide a premium quality service.

[10:50] So here are the current sort of market deals that we're seeing on one sheet. And the anthropic deal with SpaceX right here works out to about $50 billion per gigawatt per year,

[11:05] which is pretty insane. That's a very, very expensive compute that anthropic loves. They're paying a premium. So they deserve a premium service. They don't want outages. These are just some other rates that we're seeing.

[11:19] We usually see deals happening around $14 to $17 billion, but these are typically for longer-term agreements. Those are six-year contracts. The $50 billion deal with Anthropic and SpaceX,

[11:33] that's a 90-day cancellation policy on a three-year deal, and it's not even as much compute as Anthropic wants. Here's some other comps. you've got a Nebious on a very, these are just other deal sizes, we don't know how

[11:47] much they work out to per year some of these because unfortunately some of these deals they don't give you the contractor capacity so we just hear like oh billion deal with Nebious or you know whatever billion deal with Microsoft and Nscale or Microsoft 9 billion and iRed but then you know still some calculations to be done in terms of what that works out to Actually the Nevis the iRen deal we have

[12:12] works out to about $10 billion per gigawatt. So the point of bringing that up is they're probably paying a pretty penny for SpaceX's compute here to expand their cloud services substantially.

[12:24] But now you have an issue. If you have Colossus 1, that's a little crappy, and you want to scale from 1.5 gigawatts to 10 gigawatts,

[12:38] you're about to go on a borrowing binge. And the last thing you want if you're going to go on a borrowing binge is to go borrow a crap load of money and then deploy it in a bad way. You don't want to do that.

[12:50] And so I think Elon is doing the right entrepreneurial thing. It's actually kind of bad A to go there and say, hey, you know what? we're actually going to supervise this.

[13:03] We're going to go from the old style of rushing this, and we're going to go a little bit slower. But because we're going to borrow $400 billion plus,

[13:15] that money raise is coming. Because we're going to borrow $400 billion plus, we're going to use a new team of SpaceX engineers to do this the correct way, and we're going to do this a little bit slower.

[13:28] We're going to add more backup systems up front. We're going to redesign portions of existing data centers. We're going to have more backup systems. We're going to have a new team. We're going to improve reliability. And we're going to make sure that we don't cause issues by stepping on each other during the construction process.

[13:47] Now, with this, I decided, let's go see what the earnings expectations are now and what analysts are saying for SpaceX. And then I got blown away.

[13:59] So if I go into the Meet Kevin app under the course member section of stock, I can see that the 2026 revenue estimates were down over here at like $35 billion in total.

[14:14] This is revenue. These revenue estimates have now moved up for 2026 by about 50%. For 2027, consensus revenue has gone from about 65 to 102.

[14:30] That is a 57% increase. Part of it, you can see, happened right here at Q2 earnings. What happened at Q2 earnings? At Q2 earnings, they got their upfront payment from Anthropic for $1.6 billion.

[14:45] Yes, it's on the 90-day cancellation policy. It is going to get canceled. Here it is, AI infrastructure revenue of $1.6 billion. And then we wrote down some catalysts here, which we'll talk about in just a moment in detail.

[14:59] Now, what's crazy is it's not just those estimates that have changed. It gets crazier. You ready for this? Look at this. Look at the estimate for fiscal 28 now in blue, 29, and 30.

[15:14] This is operating income. So I just moved from revenue to operating income. I want to be really clear about that. We are now talking operating income forecast. These numbers are mind-boggling, and they actually make SpaceX look cheap right now.

[15:32] What? This is where, like, people will be like, but Kevin, you said you own SpaceX in the VC, which we do,

[15:44] you know, millions of dollars of it. And you don't want to add, you know, unless it's like 80, 70 and everything. That's true. Based on the forecast then, I'm like, that would be a great place for me to get a delicious rate of return.

[15:58] I wasn't saying I want it to go down there. I was saying that's where I think it would create a really awesome rate of return. It would have already made a really great rate of return, but it obviously never got down there. We did. We were bearish on IPO, though, because you're sucking a lot of money out of the markets.

[16:13] You're sucking over $100 billion out of the market. 80 plus, I think it was like 85 plus another 25 billion in financing. And there'll be another $400 billion in financing coming. But anyway, so a bearish post IPO.

[16:27] Okay, so it's come down from its high of 220. So what? It's 150 now. All right, whatever. What did we say? Go down like 30% or something like that? 220, that works out to, wow, 31.9%.

[16:39] Cool. But that doesn't mean we don't get to look at data and go, oh my gosh, if this is right, that's old news. The stuff from when I'm on the Disney cruise going,

[16:52] I can't make sense of these finances, now I'm like, man, either these numbers are lying to me and the analysts are lying to me, or the amount of spending that OpenAI and Anthropic are doing is maybe bubbly, maybe unsustainable.

[17:08] We'll see as we get their canary earnings, right? Anthropic F1 should be dropping soon, and then going into next year we'll get a lot more color. But look at this. This is crazy. This is operating income.

[17:20] Okay, just to be clear, my operating income estimate was like $22 billion for 2030. $22 billion for 2030. I'm just going to verify that. And keep in mind, I didn't put the launch numbers in there.

[17:33] We talked about that earlier. Launch numbers are separate. But if I go over here to right here, AI infrastructure operating income, my total operating income estimate was like $21.5 billion.

[17:47] Okay. $21.5 billion of operating income estimate. And that's before interest. We're going to have to pay a crap load of interest on all the financing, obviously. You know, you go finance $400 billion of debt and you pay 7% of interest.

[18:01] You paying billion of interest every year So in fairness you probably have to take off billion of interest off all of those numbers because interest is going to come before net income and after operating

[18:15] income. So you'll have to take interest off. But even if you take interest off, dude, look at that. If I take interest off of the 2028 estimate now, it's at $40 billion. That estimate is now twice

[18:28] what my estimate was right here. And if you look right here, you could see the change over time. My spreadsheet was written over in this area. You know, there was a 2028 estimate over here

[18:41] where they'd have $15 billion of operating income. Okay, if I go from that bottom point right there at 15 to now 68, they have literally four and a half X the potential income for SpaceX.

[18:57] X. Holy schmuck. I mean, look at this. 2029, they have literally jacked up the operating income potential from $20 billion to $120 billion. That is a 6X. And then over here,

[19:14] they've gone from 40 to 200. That's a 5X. So we're like 4X, 6X, and 5X over here in terms of operating income. What the heck, dude? Holy smoke. Look at consensus revenue.

[19:34] Same thing over here. This is the revenue chart. From $100 billion 2028 to over $400 2030 over here. And that's, you know, that's not as big of an X over here. Operating income

[19:52] shows you more of a jump. That's a little less than 2x. That's a little less than 2x. A little less than 2x move on the headline revenue. So in other words, they think the payback period is going to be really, really short, and that renting out computers is going to be really high margin.

[20:07] Now, you're basically YOLOing on the back of Anthropic and OpenAI. That's what it is. It's a YOLO on the back of OpenAI and Anthropic. But look at this. These were the estimates.

[20:19] Look at this. These are estimates on post-earning fees. I think this is operating, wait, hold on, 197.

[20:32] I wish I had the label for this. Let me find what the label is for this. But you can see on the far right side the change since IPO percentage. Look at this. This is operating income.

[20:44] Okay, that's what I thought it was. I just want to make sure. Okay, my operating income for 2030 was like right here, this $20-ish billion range, and I was being really hedgy. Okay, that has now gone up,

[20:58] and I was being hedgy and slower on time. That has gone up over 2x, and then you're over 3x over here. That's crazy on some of those actual levels. Let's go look at,

[21:11] if I look at, let's see, let's see, let's see. margins. These are the margin forecasts now. Analysts are forecasting that operating margins

[21:23] are going to be between 39 to 47 percent. What? That's crazy. So, I don't know, but if, honestly,

[21:38] if those analyst numbers are anywhere even close to accurate, in the weirdest way, this is actually cheap right now. Now, how do you justify that? You come over here and you look

[21:52] at, I'm going to take their 2027 numbers because it actually compresses the growth rate because they're barely going to expect to be profitable this year. Four cents of earnings this year. Next year is expected 172. The percentage gain between that is not a comparable number

[22:08] So we skip that. We go forward one year, and then we look at growth thereafter, 166, 47, 51. That averages out to 88% growth per year on earnings per share. Well, their current price-to-earnings ratio, based on next year's EPS estimates of $1.72, is about 88.

[22:26] So 88 PE ratio, 88% growth rate, what does that mean? SpaceX is basically a one-peg right now. so if you have the balls to bet on anthropic and open ai continuing your spend like drug sailors

[22:41] you can now get spacex at a one peg based on the current expectations from analysts now catalyst number one do not lose the anthropic uh contract i don't think you will because it's

[22:53] a very high premium but anthropic needs it catalyst number two become the most reliable compute provider. SpaceX can way outgrow the neocon. Catalyst number three, this is the

[23:11] tiny catalyst, but removing ex-monetization will actually increase their profitability for most traders. Ex-monetization has been crushed, basically. And then Google's $920 million contract signed in June actually ramped up to a full fee contract by October.

[23:30] So we're in the ramping phase right now. That's crazy. So this is pretty remarkable SpaceX and very, very exciting. And with that, I have one more thing to say.

[23:44] Oh, man. You're fucking giving us lots of money. Oh, that's okay. I am with you. With you. Oh, man. You're fucking giving us lots of money. Why not advertise these things each other here? I feel like nobody else knows about this.

[23:57] We'll try a little advertising and see how it goes. Congratulations, man. You have done so much. People love you. People look up to you. Kevin Paxos, Airplane National Analyst, and YouTuber. Meet Kevin. Always great to get your take.