Why SpaceX’s Biggest IPO Catalyst Isn’t AI—It’s Scarcity

On the June 8 episode of The Morning Filter podcast, Morningstar Chief US Market Strategist Dave Sekera sat down with Morningstar analyst Nic Owens to discuss all things SpaceX, including the company’s

, Morningstar’s forecasts for revenue and margins, and what Morningstar thinks the company is really worth. Here’s an excerpt from the episode.

Dave Sekera: Nic, thank you very much for joining us today. I know you’ve got an extremely busy calendar with everything going on, but so we’ve got the SpaceX IPO coming. We saw your note that you put out on Morningstar.com about it. So first of all, can you just really give us that 30,000-foot view on what all does this company do, and maybe explain a little bit each of their main business segments?

Nic Owens: Absolutely. And so I cover aerospace and defense, and until February, I would say SpaceX was mostly an aerospace and defense company, mostly aerospace, really. They have a rockets business. They have a satellite communications business, which we think of as Starlink. And then, now in February, they bought the xAI company from Elon Musk, which is looking into AI, AI infrastructure, and potentially AI infrastructure in space. So those are the three buckets, and it’s sort of a fast-moving target, given some of the growth rates.

Sekera: Can you give me a little bit more detail about those three main business lines?

Owens: Yeah. In 2025, the rocket business did $4 billion in revenue, Starlink did $11 billion, and the AI business posted $3 billion in revenue, and about half of that is advertising from Twitter. So that’s really the startup version. And I think what most people may not realize is just how big Starlink has gotten. It’s a really money spinner, and we see a very highly profitable growth coming out of there for years.

Sekera: Now, of course, we wouldn’t be Morningstar if we didn’t talk about our economic moat rating. In this case, I was hoping maybe not only could you describe what one economic moat rating for the entire company is, but the primary moat sources that back up that moat rating. But I’m also curious, too, how do you really derive a moat rating for the entire company when you have three such very different types of businesses that make up the overall company?

Owens: Yeah, there’s a lot there, and I actually am really glad to be able to rely on the Morningstar methodology around moats, which I think is really rigorous. Given those three buckets, what I’ll tell you is the first two, the rockets—the actual designing and making of the rockets and launching them—and the Starlink business—which uses satellites that are lifted on the SpaceX rockets—we believe those have a cost advantage, and that’s one of our core moat sources.

And within that, the two flavors of that cost advantage are the R&D, basically, they’ve built and tested these rockets to be able to lift more per kilogram than other rockets, and that gives them an operating cost advantage, and then they also have done it more often. They’ve gone down the learning curve. They’ve launched hundreds and hundreds of times more than the nearest competitor and launched thousands of more satellites than the nearest competitor.

And so their average cost is lower. Their economies of scale have been, I think, firmly established. And so those two, let’s say, if they were stand-alone, would have the characteristics of what we would describe as wide-moat businesses. The wrinkle, if you will, is that the stated purpose of the IPO is to raise money to invest mostly in AI, the next phase, if you will. And there’s, the jury’s out, I would say, on whether xAI, or the SpaceX AI business, has a moat and/or what it might be. There’s some scenarios in which it might benefit from some cost advantage like Starlink in terms of putting thousands of satellites in space and so forth, but that’s far from certain, and they’re going to go spend, let’s say, tens of billions of dollars to find out. And so in our methodology, the risk of value destruction that might happen from that scenario, let’s say, overshadows some of the characteristics, the wide-moat characteristics of the other businesses.

So we net it out to a narrow-moat rating for the company, particularly even though the xAI business is not a huge piece of the business today, the planned investment and the capital deployed will make it a bigger piece in the future. In a normal conglomerate, let’s say, we had three businesses about equal size and one was wide and one was narrow and one was none, we’d probably end up at narrow for those averaging reasons. This is a little different but worth, I think, acknowledging the core moat around their capabilities around launch is really quite impressive.

Sekera: OK, thank you. Now, of course, when we’re talking about valuation and Morningstar, when we think about the value of a stock, it is the intrinsic valuation of the company ends up being the present value of all the future free cash flow that a company’s going to generate.

So I guess even before we get into how you’re valuing the company, how do you come up with forecasts for this company? I mean, are you really trying to forecast three individual companies and roll that up into one? How do you drive that top-line number overall and at each one of those segments?

Owens: The answer is yes. It’s much easier to look at this as three pieces because they each have different economics. And so for the rocket launch, it’s how many rockets, at what cost, times how many payloads? And for Starlink, it’s sort of subscribers. This is a telecom business, essentially. Subscribers, average revenue, incremental costs, and so forth. And we have done some work on sizing what we think is, let’s say, our view on the market opportunity there. Some of the numbers that are in the registration statement are very ambitious or sort of aggrandized, in terms of they might soak up all telecom spending globally. We don’t need to believe any of that to get to a pretty healthy growth rate and a good business in Starlink. For xAI, spent a lot of time building out, let’s say, a version of what that business could look like and really is hinged on this idea of if a commercial data center in space or GPU cluster in space is viable, what might it cost to own and operate and what might be, let’s say, market rates for AI compute, and spanning that out, I think we did 15 years just to see how it looks.

And again, there’s scenarios there. So we are able to do forecasted cash flows for all three, and then we did probability weights on these upside, downside, and base-case scenarios, which really hinge on what might happen at the AI division.

Sekera: And we were talking earlier, about when we talked about the rockets division earlier, I believe you said that you can look at the launch schedule that’s already out there and get a pretty good idea. And how far out into the future does that launch schedule typically go?

Owens: There’s a FAA schedule, they’re approved, I want to say years in advance. Beyond that, we have a forecast that is more or less from the company’s stated plan. They have plans for how many Starlink satellites they want to put out there, and then you have how many fit on a rocket, how many rockets do they own? How many rockets do they have to build to basically continue building out the Starlink constellation? Most of the stuff that SpaceX puts up in space are Starlink satellites, and then they rent out the extra space to others, and that’s its own strategy in terms of decreasing their cost by increasing their volume in a way in-house. That’s where the forecast for that comes from. And there’s one key assumption that the Starship, the newest, biggest rocket that they had a test flight for the other day, that top stage has to be reusable for some of these numbers to really pan out.

The idea is to be able to fly the same actual rocket, which is—the space shuttle was the last time we had a upper-stage reusable, but it had to go into the shop for months and months to have tiles replaced and so forth. So the goal, the engineering problem at hand, is can you have an upper stage that’s reusable in hours? And I think there’s a realm, it’s within the realm of possibility, it’s just engineering, it’s thermal resistance and metallurgy and things that I don’t have expertise in, but I think is doable. And so if you believe that, then you get to these scenarios where it’s worth—that they can launch thousands and thousands of satellites sometimes launching the same rocket multiple times a day.

Sekera: Well, that actually goes right into my next question then talking about cost. So it’s really the similar question.

You’ve gone through how you forecast revenue. How do you think about forecasting the operating margin? How do you think about it not only for forecasting maybe here in 2026, but trying to think about what’s the path of that margin over the next couple of years, maybe getting out through your five-year forecast period?

Owens: Again, on the rocket launch, it’s relatively straightforward, especially from my perspective covering other airspace manufacturers and given the volumes that we’re talking about, again, if you assume viability of a reusable upper stage, then you’re just looking at a manufacturing supply chain and cost curve where there’s some savings every doubling of how many you make manufacturing. And then there’s the actual launch costs, fuel and so forth, where because the new Starship has a launch bay or a cargo bay that’s so huge, the average cost to lift something into space goes down per kilogram every time they do it. And this is an important part of what SpaceX has already achieved. If you compare the space shuttle, the launch cost today is 95% lower than it was then and continuing to go down.

If you have a business plan that you want to do biology experiments on a satellite in space, and then you got a quote 10 years ago for how much that would cost to launch, you’d say, “Well, never mind.” But now all kinds of things are possible because that dollar launch cost comes down. And so that’s an interesting feature, and then that dollar launch cost comes down for Starlink to put the radio satellites up there. For space, that’s a fairly straightforward model. We can look at manufacturing costs and these average launch costs, and it’s really dividing, let’s say, by the kilogram capacity of the rocket. Then on the satellite side, telecoms is interesting because once you have the network set up, traditionally we look at things like fiber optic or mobile cell towers, but there’s an analogy here to the satellite constellation. The cost to add a customer is very, very, very low.

The incremental profit from growing your subscribers and services is very attractive. And the other interesting aspect that we found is when you think about, say, cell towers or fiber, the investments that you have to make to go reach another 10,000 homes or something, they apply in that ZIP code. Whereas if you put a hundred new satellites up to widen the available capacity or spectrum of that network, everyone on earth could potentially be benefiting from those because they’re spinning around and they switch to other zones and provide coverage wherever they are. That is like dividing by a higher denominator in terms of the ROI for those investments. That’s an interesting aspect of how we modeled out the potential profitability for Starlink. On the xAI side, still it’s just far less certain. We think that it is also within the realm of engineering possibility to put data GPUs in a satellite.

They have solar cells. We think they can do heat dumping on the backs of the solar panels. And the open question is whether it’s enough economic savings versus a terrestrial data center for it to be, let’s say, compelling or just interesting. In this “Moonshot” scenario that we modeled, we’re saying it’s compelling, they might have some incremental cost advantage because sun is brighter above the atmosphere, solar power is free versus, let’s say, natural gas or whatever on the ground, etc., etc.

But even on a good enough, we think there’s a niche that they could potentially serve where you have this, let’s say, economically reasonable computing capacity in orbit, it probably wouldn’t have the same fast latency that your neighborhood data center would, but so there’s going to be tons and tons of demand for AI computing that might be like overnight jobs or hour delay or what have you.

I mean, the delays are not that long, but categories of AI work that could make sense for this network are what we modeled in.

Sekera: All right, let’s get down to the brass tacks. We’ve talked about how you forecast revenue. We’ve talked about margins. Maybe let’s just get to your valuation in your base case, based on your forecast and your analysis, what do you think the intrinsic value of the company is today?

And I guess the biggest question there, too, then, is what do you think really differentiates your base case, some of those probability weightings that you’ve done versus what they’re talking about that IPO being priced at in the market?

Owens: The base case, we didn’t really change our scenarios for rockets and satellites that much. Those two together add up to about $611 billion of enterprise value in our model, let’s say, across the board. Then it really, the question mark, the uncertainty in my view was around what could the AI business be worth, and basically putting out there that there’s some unproven things. I’m starting to think of this as, or express this as, like a nested probability of does Starship turn out to be reusable, let’s say, and really scalable, and do data centers and space end up mathing out in terms of ... I think that the engineering is probably doable. The question is just, is it commercially successful or reasonable? We assign a 7% probability of both of those being true, what we call the Moonshot scenario, and that actually gets us to a $1.9 trillion enterprise value, which is very close to what they’re talking about pegging the enterprise value at the IPO.

And let’s say if the data centers in space are completely unviable, which is, well, if we think there’s a 43% chance that not both of those will be true and that they would actually spend some money to find out about the data centers in space, but then they would stop. And so that kind of subtracts a little bit from our base case scenario. But we think the most likely scenario is what we’re calling a “Minimum Viable Product,” where they do have data centers in space. They may not be compellingly cheaper, but that they fill a niche and they work, and that those three added up is what got us to the $780 billion average, or weighted average enterprise value, which happens to come in very close to half of what the IPO price is. But I think the way I would reframe that in specifically the Morningstar point of view, we talk about margin of safety and so forth,

if the IPO is at $1.8 or $1.9 trillion market value and our upside scenario, which we think has some dependencies on it is about the same number, what that’s saying is as a disinterested investor who’s looking for margin of safety, you wouldn’t necessarily pay full value for that scenario, but that’s what the IPO investors are, let’s say, being asked to do. Pay me for what it could be worth in the best-case scenario. And I just think that’s one way to think about it. We took the weighted average, and that gets you to a lower number.

Sekera: I’m also thinking about when this IPO does occur, stock is now free to trade. In the short term, I see a whole bunch of different technical reasons out there which are going to influence how that stock trades in the short term more than what the long-term intrinsic valuation of the company may be.

In the months after the IPO, maybe even like the next year or two after the IPO, is there anything specific that you would really tell investors that they should be watching for, that could maybe sway how that stock could trend either up or down, both from our own intrinsic valuation point of view as well as how the market may price it?

Owens: Absolutely. I think I should answer that in two parts. Fundamentals, the things to look at are progress on Starship, reusability, whether that’s announcements or demos for better heat shielding, etc.—the more certain that becomes over the next, say, two, three years, then these probabilities go up, probability of success, let’s say; and similarly, on the data centers and space. Again, the engineering we think is inside the realm of the possible. The question is what’s the incremental cost and how that works through a commercial model for that. And I think we’ll know more inside of two or three years how those play out. And so changes in those probabilities as we get news, let’s say, would change our fair value.

Sekera: And in many cases, probably quite substantially then.

Owens: Yes. And that’s what’s particularly exciting about a company at this stage of growth and in this sort of moment, right? It takes more to move the needle on some of my other companies than just changing a probability, especially, I guess, because of the range of the outcomes: We have negative $85 billion in the “No Go” scenario and, I think, it’s $1.3 trillion on top of the 600 in the upside.

In terms of things that could move the stock price and milestones to look at, I would simply think of that as in terms of supply and demand. They’re only selling a couple percent of the company here in what we think is next Friday, and there’s more where that came from. There’s lots of investors who’ve invested as a private entity in the company over the years, and most of them have agreed to lock up. So they’re 180 days out or in the case of the chairman, Elon Musk, a year out, but there’s still supply of shares out there.

And then there’s also, I would argue, unprecedented demand or fervent demand. It’s part of the AI trade, the AI story, and you will have some portion of index funds buying it because it’s going to be included in the Nasdaq 100, I think, 15 trading days.

So those are things to watch, and some of them are unprecedented. The price of a stock is set by the marginal buyer and seller. And also what’s frankly par for the course on an IPO, part of what you are hiring the investment banks to run the books for, is they can intervene in the market. They want to, let’s say, smooth out supply and demand over trading sessions and so forth, especially if they know that some more shares are going to come online or if there’s another buyer who’s going to take some of them out of circulation, etc. Can we use the old phrase: The market’s a voting machine in the short term and a weighing machine in the long term?

Where things stand today, the market conditions seem quite favorable for an AI-related stock, and I think there’ll be plenty of demand almost for the foreseeable. And again, they’re only floating a few percent of the company today, but there’s going to be supply, and it’s the explicit aim of some of the existing investors to liquidate or at least diversify their holding. Some of them been holding it for almost two decades. So that’s just something to bear in mind as the market unwinds. And basically what we conclude is at the IPO on the day of, I don’t see there being any margin of safety to that price, and it is more likely than not, in my view, that investors, especially long-term investors, will have an opportunity to buy at a better margin of safety.

Sekera: Got it. Well, thank you very much, Nic. I really appreciate your time. And I also really like in your article, too—so I think investors really, it would behoove them to read it on Morningstar.com—but in there, too, I think one of the big takeaways is: It’s not like we don’t believe in the company overall. When I look at a lot of the revenue forecasts you have out for 2035, in many cases, they’re multiple times greater than the amount of revenue this year. I think that additional clarity that you provide in the article, I think, will really help investors kind of encapsulate their own view as well as understand what it would take to actually get to what that market price is today.

Subscribe to The Morning Filter on Apple Podcasts, or wherever you get your podcasts, and keep up with the latest research from hosts Susan Dziubinski and David Sekera on Morningstar.com.

5 Stocks to Buy Before the Next Market Rotation

Plus, is SpaceX really worth $1.8 trillion?

52m 6s Jun 8, 2026

Watch