SpaceX Just Dropped Stunning AI Numbers

Added: 2026-05-22

[00:00:00]SpaceX dropped their S1 IPO filing and the biggest revelation, they're attacking a market that's worth $28.5 trillion. SpaceX actual business is not rockets, it's AI is a service [music] taxi. Let's start with the first bit of news. SpaceX f1 was dropped and the biggest thing we learned the TAM total address on market is 28.5 trillion and almost all of it is going to be in enterprise applications. And of course, we've now learned the Anthropic, their deal with SpaceX is $1.25 billion per month through May 2029 for AI compute capacity. And in capacity is ramping in May and June at a reduced fleet fee. But this deal is beyond just Colossus. It's now Colossus 2 as well. Jeeoff, what's your thinking about this deal?

[00:00:52]>> The company and the product. I don't look at the quarterly stock movement. A very realistic bull questioning certain decisions.

[00:00:58]>> The rational bull. We enjoy listening to bears. We're looking for the red flags.

[00:01:03]You're supposed to react. Yeah, we only had 50 cents a share the other the last quarter. Come on, guys.

[00:01:14]>> This is call it SpaceX Web Services.

[00:01:19]Call it any acronym you want. It's almost AI services. AI SpaceX services.

[00:01:25]don't want to look at that acronym on a piece of paper. But it this could be a gigantic business and the moat is the access the coherence and the speed of hardware infrastructure ramp. Remember this is extremely complex. If you look out on the landscape of all the hyperscalers, you're seeing lots of cancellations, but you're also seeing lots of those cancellations being picked up by other hyperscalers. So, there's churn in the system. There's delays in the system. It's getting power load generation. It's getting the battery packs. It's getting permitting. It's getting everything pulled together. So if Elon continues to demonstrate almost multiples in terms of being in speed in terms of bring up that's a moat and that's going to bring in anthropic the quite frankly the biggest player maybe that has the best chatbot to date at least currently I think this who's number one is going to move a lot but this is a big deal and Elon and his team created the moat that enabled this.

[00:02:34]Yeah, Jeeoff, you saw this, right? Helon said that this partnership with Anthropic was expanded number one, right? AI comput as a service. You called it [laughter] AI as a service AI SpaceX as a service at significant scale, right? And then he said more companies, other companies to do the same. Are you expecting who you who do you think is lined up? You you've seen Elon do a 180 on a dime with Enthropic and their leadership team and and because he realizes there there is a mode here and the key here is driving really high utilization rates of your product acquire this product and you've got to depreciate it over some form of volume and time and that volume you'd like to have increased maximize usage.

[00:03:24]So just pick out who the other players are. I think any one of them could be a partner to Elon and his team in this endeavor. I also think it's unpredictable these training runs, the size of them, your usage between training runs. So that to have multiple, this is almost I hate to say this, it's kind of like the Uber model in terms of having many OEMs. Hopefully one of them eventually can actually run a robo taxi for them. But having many options so that you can balance the load and have high utilization of your network. In that case, it would be robo taxi network. In this case, it's a it's an AI services piece. So this will help the cost structure of SpaceX as well. Of course, on the revenue line. Everybody's talking about the revenue line, but this is going to help on the COGS line. So sir, are you going to have to redo all your spreadsheets that you have on your bingo card that look how big this is?

[00:04:19]Number one, 2.28.5 and then that vast majority of it is AI and specifically enterprise applications.

[00:04:25]>> This is interesting, isn't it Herbert?

[00:04:27]Because everybody's focused on the launch business, which is the far left, which is the smallest number on this chart. And then Starlink, everybody, I think, recognizes at least part of what Starlink the opportunity there. And you can see Starling broadband and Starlink mobile 1.6 trillion. That's a pretty big market opportunity.

[00:04:46]And now we're talking about AI and all the different pieces. And the enterprise applications obviously is the one that sticks out here. And this is macro hard.

[00:04:54]This is AI agents. This is putting AI into enterprises and basically redoing the infrastructure of every single company on the planet. And Elon's made the point that if you have a company even with one single person and you're up against an AI only company, a company with a human is going to lose because AI can just operate so much faster. So this is the big future that they're moving toward and the numbers are staggering and probably understated. The other thing I think that's also interesting is for Tesla shareholders, Elon's talked about digital Optimus and this has been the same wheelhouse using FSD technology and combining it with Grock and providing a an AI as a service a Gentic AI and so this is interesting too from a perspective of a Tesla investor and I think that hasn't gotten really any attention yet in terms of that. Larry, you and I did a deep dive on every single business that SpaceX has. That will drop tomorrow on Saturday. And then this enterprise application though was a huge surprise to me in a sense that it's this big 22.7 trillion. And but you said that this is there this is just a addressable market. There's obviously going to be lots of competitors.

[00:06:07]>> This is going to be the most heavily comp heavily competed portion of this. But I'll make a few points first.

[00:06:16]First point I'll make is that yes it's a gigantic market 22 bring up the slide again. Yes it's a gig gigantic market 22.7 trillion but I will point out that the rest of the market is no slouch. It's almost 7 trillion. So the a 7 trillion market five years ago would have be stunned us all and now we treat it as like just the icing on the cake. And let me point out one more thing before we actually go to that size that that part of the market the enterprise look at how the graph is structured. At the bottom of the graph, we have these bars. Space is 3.7 trillion.

[00:07:03]Connectivity, no 370 billion, excuse me.

[00:07:07]Connectivity is 1.6 trillion. And then AI is the 22 trillion. No, 20 29. Yeah. So the big issue, the issue I want to point out is that the right can't do without the left. In other words, it's a pyramid, inverse pyramid, but everything stands on it on the launch ability to launch Starlinks and to launch satellites into space that will do the data processing or the AI processing for all of the AI stuff.

[00:07:47]And what gives XAI and SpaceX AAI the competitive advantage is the space piece where they are comp they are peerless in terms of competitive value. It makes it extremely difficult for all the other levels to compete within.

[00:08:10]So if you don't have the kind of launch cadence that they can develop with Starship, if you don't have the cost of launch that they get down to with Starship, it's going to be extraordinarily difficult for them to compete for AI infrastructure, for consumer subscriptions, for digital advertising, and of course for Starling. So this is a structure that is designed to really out compete any category to the right because of the value ad that they bring to the left and the advance and how far advanced they are. So it it's easy to say ah launch is not that much money.

[00:08:56]It's not even a trillion dollars.

[00:08:59]It is what makes the trillion dollars possible and also that makes the it so difficult to compete. So I just wanted to make that point. Now as far as applications are concerned understand this is what the world runs on. This is what all enterprises run on. It's what even smallsiz businesses to midsize business.

[00:09:23]You go into a restaurant you pay you get the check. You pay the check. you made the reservation. All of that's going to be done by AI. All the way up to these massive plants that run gigantic automated operations, all of that's going to be done on AI. So, from the smallest business, the one-man business, the one person business I should say, all the way up to the 100 200,000 million people plant, all of that's going to be run on these enterprise apps. And that's why the market is so gigantic.

[00:10:00]So if you think about combining all of Microsoft, all of from Microsoft all the way up to SAP, that's the very big corporate system, industrial systems, you name it. It all of this stuff is going to be running on AI and that's where the 27 trillion comes from. And many things which are not automated today.

[00:10:26]Any many things that are barely automated today. Many things that are automated but in a fairly prosaic manner. All of that is going to move to the AI cloud and be operated by AI. And that's where the enterprise apps of the 27 trillion comes from. 23 >> 22 22. Larry, we're filming this an hour before Starship flies. How big of a deal is this Starship? By the time we publish this, the world will know if it was successful or not, but how big of a deal is it?

[00:11:00]>> It's a big enough deal to give me butterflies in my stomach. I would never ever were I running SpaceX and this is maybe why I'm not running SpaceX. I would never schedule this fight before the IPO because this is an extraordinarily high-risk look. The enabling we talked about the inverse pyramid the very tip of this inverse pyramid is this starship. Everything turns on the starship.

[00:11:32]The situation that Falcon 9 is in today where it leads the industry by far. 85% of all launches that take place of Falcon launches. 90% of all the mass to orbit is on Falcon 9. All of that is going to be completely overshadowed by the cost, the low cost of Starship and the mass that it can put into orbit.

[00:12:00]And that's what's going to enable this space economy.

[00:12:05]And the company's 10 years ahead of the rest of the market, maybe 20 years ahead of the rest of the market in terms of its capability, not only launch, but the cost of launch, which is the critical aspect I talked about. And everything turns on this final version. This is the production version becoming successful.

[00:12:27]But it's very seldom that you fly a rocket for the first time and it works.

[00:12:33]In fact, I don't remember a time when that happened.

[00:12:37]>> I don't expect it to work. But Elon did say that we have 10 lined up already. So if this one doesn't work, >> yeah, >> it's going to be very It's taking a big chance to do this before the IPO. I would never have done it. Elon makes it as if no big deal, but it's a big deal.

[00:12:55]And do I expect it to be successful? No, I don't. The chances of being successful, I think, are one in one in four.

[00:13:05]>> Yeah, >> maybe one in five.

[00:13:08]Because the complexity of this ship is very of this, not just ship, the starship as a whole is extraordinarily high. The dangers of some little thing going wrong are very high. So anyway, I'll end at that point and just say that my expectations are low. I would be delighted if it does work. I don't expect it to work and I think it will have an impact on the IPO.

[00:13:34]>> I haven't asked you about anthropic $1.25 billion per month all the way to May 2029. It includes Colossus 2, the GB200s, not just Colossus one. And that's 1.25 billion per month. That's there is a 90-day termination clause, but that's a lot of money that they're paying. And then Elon said, "This is not the only one. We've got other partners in place." What's your forecast of this and what does it do?

[00:14:00]>> I haven't made a specific forecast, but it is impressive. It's way above, I think, where everybody thought the number would be. Brian Wang, by the way, actually, I think was spot on. I think he had 16 billion or something in his numbers.

[00:14:13]>> So was Larry. [laughter] >> And perhaps Larry as well. So you guys knew, you guys figured it out. Everybody else thought the number was much lower and everybody was pleasantly surprised.

[00:14:22]It's a sign of things to come. You can see how valuable this computers.

[00:14:25]Anthropic while they could terminate the contract, they'd be fools to because they they won't be able to scale their business. They need this more than SpaceX does.

[00:14:33]>> It's not even a matter of scaling soon, it's a matter of keeping their business.

[00:14:37]>> Yes, that's right. Exactly right. Yeah.

[00:14:40]And they've got IPO plans as well, right? So >> yeah, >> they're all heavily incentivized to to keep this going. But it just shows what's what's what the potential is for this nation nent AI as a service business. As Jensen Wong likes to say, the more you spend, the more you make.

[00:14:59]And this is another example of that. And the spending is not going to continue, not going to stop anytime soon. We are in such early days of the AI infrastructure buildout. At some point we're going to look back and say, "Oh, do you believe that they were building AI data centers on Earth? How how foolish were they back then?"

[00:15:17]>> The fascinating part of this all is that the one piece missing in the entire document, the S1, >> where's the estimates on the terap there's a gigantic nent business in that terap? Gigantic. Because if certain what you said is true, that terra fab the opportunity with it is just boundless.

[00:15:39]It's also interesting in the S1 Larry they mentioned all kinds of other additional potential business opportunities even with these existing businesses there's certain TAMs that they didn't factor in and then of course they talked about future markets here you go in this chart and of course the these are some obviously longer term things but the size of these opportunities also are not small >> amazing it's amazing the unlock sorry >> no go ahead >> I was just saying the unlock that This first principles thinking that he is so rigorous about the unlock it's brought is surprise surprised even me and I'm an acolyte and I believe highly in it but man the unlock step by step by step is just staggering >> to what extent is that enterprise application the AI part of it is rented and what to extent is it SpaceX AI's own product that they're selling you against 5050 50. What percentage do you are you guys expecting?

[00:16:39]>> It's going to be hard to figure.

[00:16:41]>> Yeah, it's a good question. I think that remains to be seen. My guess is it becomes a larger and larger piece over time because Elon is going to be able to move at a pace that no one else can keep up with. So, while today I think he's playing catch-up and he's seems to be catching up quite nicely in many ways and helping some of these companies, but I think eventually he's just going to find a way to do it faster, cheaper, better than everybody else. And if not, that's fine. He's well positioned to be their service provider. So, it's win-win.

[00:17:08]>> Look, he started three years after, three to five years after the others.

[00:17:13]So, he's done a lot of catching up to date, but I think he's got a way to go and I think he'll get there.

[00:17:20]>> Okay, that's the SpaceX filing. We learned a lot. I did a Jeff and I did a show with Alexandra and she just showed 12 nuggets that she found that do hint a little bit that there might be a Tesla SpaceX merger. And then of course Larry and I did a show where we dug a little bit deeper into every business model. So this is a big