Skip to content

Maxinomics Business Science Hardware

Why Elon Musk is Really Building Starship

Phil Andrews argues that the right way to value SpaceX has nothing to do with landing boosters and everything to do with an empty room: the payload fairing at the tip of a Falcon 9, 17 feet across and 34 feet high, and one metric he calls return per ton. Sell a ton of that volume to somebody else and SpaceX collects about $4 million once. Fill it with its own Starlink satellites, 525 kilograms each, $800,000 to build, $1.5 million a year for a five year life, and the same ton returns about $13 million. That ratio explains why 123 of SpaceX's record 165 launches last year carried its own cargo, and the video walks the full P&L: a $30 million booster, a $10 million expendable second stage, $45 million all in for a new rocket falling to $16 million once the booster is flight proven, which turns $32 million of margin into $61 million selling seats and $171 million into $200 million owning the payload. The historical spine is the 19th century railroads, handed 175 million acres of land grant so they owned both the ride and the destination, set against the airlines, who own the plane and neither airport and now lose money per seat mile and make it back on credit cards. The last third argues the next rung is AI compute in orbit at $10 to $50 million a ton, a refinery rather than a pipe, and that Starship is the shipping container moment: five times the chamber, a ton ten times cheaper, and 100 launches worth $130 billion instead of $22 billion. Andrews is explicit that Starship has put exactly zero working payloads into orbit and that the whole bet rests on how often it can actually fly.

Published Jul 16, 2026 20:13 video 35 min read Added Jul 25, 2026 Open on YouTube →

At a glance

Phil Andrews spends 20 minutes arguing that the correct way to value SpaceX has nothing to do with landing boosters and everything to do with an empty room. At the tip of the Falcon 9 sits a payload fairing, 17 feet across and 34 feet high, roughly the volume of a small house. That chamber is the scarce thing SpaceX controls and almost nobody else can offer. Everything else is a delivery mechanism for it.

From there the whole video runs on one metric he calls return per ton. Sell a ton of that chamber to somebody else and you collect about $4 million, one time. Fill the same ton with your own Starlink satellites and you collect about $13 million, billed out over five years. Fill it with a rack of AI accelerators and the number he floats runs from $10 million at the low end to $50 million at the high end. That single ratio, he says, explains why three out of every four SpaceX launches now carry SpaceX's own cargo, why the company looks far more like a 19th century railroad than an airline, and why Starship has quietly become the most important number in the business.

The argument is built out of history as much as spreadsheets. The railroads got 175 million acres of the American West handed to them and became a wealth machine because they owned both the ride and the destination. The airlines own the ride and nothing at either end, and a century later every major US carrier loses money on the actual act of flying you somewhere. Malcolm McLean put cargo in steel boxes in 1956 and cut the cost of loading a ton onto a ship by 36 times, which did not just move stuff more cheaply, it moved entire industries. Andrews maps Starship onto that last one: not a bigger rocket, a shipping container.

Figure 1. The fork the entire video hangs on. SpaceX has exactly two things it can do with a ton of fairing volume, and one of them is worth more than three times the other before you count the fact that every extra satellite makes the Starlink product better and pulls in more subscribers.

Chapters

0:00 The Empty Chamber
0:40 Starlink's Return per Ton
1:56 Sell the Seat or Own It
2:54 Why Launch Got Expensive
4:07 The Falcon 9 Flywheel
5:31 SpaceX Becomes Its Own Customer
6:59 The Railroad Parallel
9:41 The Airline Trap
12:13 AI Satellites and Value Creation
14:58 Starship and the Shipping Container of Space
17:36 Footnotes

The empty chamber

The video opens on the fairing, not the engines. "This is what SpaceX is selling," Andrews says over a shot of the nose cone, "an empty chamber the size of a small house, 17 feet tall, 34 feet long." It sits at the tip of the workhorse rocket, and it is yours to fill for the right price. They will make sure it gets there.

So how much would you pay to put your stuff inside?

The framing matters because it strips away the part everyone finds exciting. Without that empty space to put stuff in, the rocket is of no use. It does not matter if it lands itself. At 0:19 he lands the thesis: "It is the scarce resource the company controls that few others can offer." This space is the business of SpaceX. Understanding how to maximize each cube that space can be broken down into, he says, is how you understand the potential, the problems, and the promise of the company.

He flags the Ground News sponsorship up front and promises to get to it later, then goes straight back to the chamber. Let's fill it. And let's start with the thing they put in there most, the thing that ends up explaining the entire company.

Starlink's return per ton, or the grand piano problem

A Starlink satellite weighs about as much as a grand piano. 525 kilograms, roughly a thousand pounds, flat, plain, mass produced. SpaceX makes 70 of them every day. Each one costs about $800,000 to build.

That grand piano gets parked 400 miles up, whipping around Earth at orbital speed, and it quietly continues to bill its Customers long after the rocket has landed back on the pad. Spread the whole Starlink business across all the satellites doing the work and any single one of them pulls in about $1.5 million a year.

Andrews wants you to sit with that. In year one it cost $800,000 to build. Then it earns double that every year, for five years, and only five years, by design. At the altitude SpaceX chose, the residual atmosphere puts enough drag on the satellite that it eventually succumbs to gravity, sinks deeper into the atmosphere, catches fire and burns up. You do not get to count its revenue forever. You count it over five years. Each satellite throws off a total of about $7 million before dying in a ball of fire.

Then, at 1:38, the line that organizes everything after it: "This is the only piece of math that matters in this whole video." Take the lifetime revenue of that satellite and divide by its weight. A ton of Starlink riding in that chamber returns $13 million over its life.

Hold on to that number.

Sell the seat or own it

SpaceX has two completely different things it can do with the area at the tip of that rocket.

Option one: sell the seat. Somebody, a company, a country, NASA, has a satellite they need in orbit. SpaceX sells them a ride. A dedicated Falcon 9 runs about $74 million and can carry 17.4 tons into low Earth orbit. They like the rocket, the fairing opens, the thing floats out, and the transaction is over. SpaceX got paid once. The Customer keeps whatever their satellite earns for the rest of its life. For that ton of lift, SpaceX collected $4 million, one time.

Option two: don't sell the seat. Put their own grand piano in it and collect $13 million over five years.

Four million once, or thirteen million with the rent rolling in over half a decade. At 2:28: "The same space, three times the money." And he is careful to say at least three times, because for every Starlink they send up the product gets better and attracts more users, which is a compounding effect the raw ratio does not capture.

That is the metric. At 2:38 he says it plainly: forget the rocket landing on the barge for a second, the right way to judge SpaceX is return per ton. How many dollars does one ton of that scarce chamber earn over the lifetime of whatever you put in it? Everything that goes up must pay, once or many times. SpaceX figured out that option two is worth at least three times more than option one.

But none of that math works if the chamber is expensive to launch. And for 40 years it got more expensive, not less.

Why launch got expensive

Here the video does its best reversal. Andrews puts up the price it cost, all the way back in the Apollo days, to put one kilogram into orbit. The natural assumption is that it goes down over time. Technology gets better, things get cheaper.

Hold that thought. It went up.

The Saturn V, the rocket that took us to the moon, put a kilogram into orbit for about $12,000 in today's money. And then we built the thing that was supposed to fix all of that: the Space Shuttle. Designed to be reused, fly again and again, bring the cost down. At 3:30 the verdict lands: "The Space Shuttle was the most expensive way human beings have ever reached orbit," about $80,000 a kilogram. Of all the rockets we developed, the reusable one was the worst one.

That held until April 2016, when a private company landed a booster on a barge in the middle of the ocean and flew it again. The price of a kilogram to orbit fell more than twentyfold.

The reason the old way was so expensive is almost absurd once you see the number. The fuel, the kerosene and liquid oxygen you actually burn, is about 0.3 percent of the cost of a rocket. A third of one percent. Everything else is the machine. At 3:57: "For 60 years, we just threw the machine in the ocean after one flight."

SpaceX, being a business that cannot lose money forever and keep operating, stopped throwing away the machine. One of their boosters has now flown something like 34 times.

Figure 2. The curve that was supposed to go down and went up instead. The reusable vehicle, the Shuttle, was almost seven times worse per kilogram than the expendable moon rocket it replaced. The two dashed bars on the right are Starship's stated goals of $100,000 and eventually $10,000 per ton, which Andrews is explicit have not happened yet.

The Falcon 9 flywheel

This is the workhorse rocket, and Andrews takes it apart piece by piece.

The booster is the bottom part and the most expensive part. It costs $30 million to build. The second stage sits above it and holds the payload. It costs about $10 million to build.

When a Falcon 9 takes off, the booster burns almost all its fuel pushing the second stage to the edge of the atmosphere. Once there it disconnects and begins falling back to Earth under gravity, tumbling and tumbling until the flip, then it burns the rest of its fuel and lands itself back on the pad.

The second stage has one engine in it. Once disconnected from the booster it kicks on, burning just long enough to get the payload wherever it needs to go. Doors open, stuff is pushed out, and the second stage says a final farewell, because it will use its remaining fuel to head back toward Earth so it isn't left floating in space. The second stage is not reusable. It is intended to burn up in the atmosphere. Single use.

All in, the cost to build a Falcon 9 from scratch and launch the whole thing one time is about $45 million. SpaceX will sell you the entire chamber, your own private ride, for about $77 million.

So on a brand new rocket: sell the seats and SpaceX makes $32 million. Fill the seats with Starlink satellites, or as we will see shortly other potentially more profitable things they want to make, and they make $171 million.

Now the flywheel. Once that booster has flown one time, it has paid for itself. Getting it ready to fly again costs just $1 million. So for every additional time the booster flies, selling the seats now makes $61 million, and filling it with Starlink makes $200 million.

What goes in the chamber	What SpaceX collects	What the flight costs	What is left
New booster, somebody else's satellite	$77 million, paid once at liftoff	$45 million, booster $30M plus second stage $10M plus the rest	$32 million
New booster, 17.4 tons of Starlink	About $13 million per ton, billed out over five years	$45 million	$171 million
Flight proven booster, somebody else's satellite	$77 million, paid once at liftoff	$16 million, a $1 million refurb plus a fresh second stage	$61 million
Flight proven booster, 17.4 tons of Starlink	About $13 million per ton, billed out over five years	$16 million	$200 million

Figure 3. The four cases Andrews walks through, reconstructed from the numbers he states. Reuse is worth a flat $29 million per flight in either column, because it converts a $45 million rocket into a $16 million one. But the vertical gap, the choice of what to put inside, is worth $139 million on a new booster and $139 million on a reused one. Reuse is the enabler. Cargo ownership is the prize.

SpaceX becomes its own Customer

There were conversations about Starlink early in the company's history, Andrews says, but it was not the original game plan. SpaceX set out to sell the space on its rockets. That was the intention. About a decade ago, that changed dramatically.

SpaceX flew 165 Falcon launches last year, a record, more than half of every orbital launch on planet Earth by one company. 123 of those were carrying SpaceX's own Starlink satellites. Three out of every four.

Then the line that makes the point stick, at 6:02: "Every other rocket company on Earth builds a rocket and then waits for the phone to ring. Boeing builds a rocket and waits for NASA to call. SpaceX builds a rocket and calls itself." 123 times a year.

It works. More than half of all the money SpaceX makes comes from Starlink, from people like Andrews himself, who notes he actually has Starlink, has to have it, because he does not have other internet options available. It is the one part of SpaceX that reliably, consistently turns a profit. At 6:22: "The launch company became its own best Customer."

He reaches back to his own Taiwan video from a couple of months earlier for the concept of an anchor tenant: the Customer you can count on, who will buy everything you are willing to sell them. You want to do something new, expand, build a new factory, you call them up. For Taiwan Semiconductor it was Apple. One dominant Customer whose presence makes the entire thing work, who you have to attract and keep and pray never leaves.

At 6:43 he closes the loop: "Being the landlord and tenant, you never have to negotiate. You never have to worry the tenant walks." You collect both checks, the cost of the lift and the lifetime revenue of the thing in the seat.

The railroad parallel

Stop me if any of this sounds familiar, he says. Like you have heard this story before. Maybe about 150 years ago.

In 1880, if you bought the American stock market, you were basically buying railroads. Not a piece of it, not a big sector. At 7:03: "60 percent of the entire market was railroad stocks." For a stretch of the 1800s the US stock market was almost entirely a market of railroads. They were so big they basically invented the modern stock market, because nothing else on Earth needed that much money at once.

Until that point everybody needed a ride and the railroads owned the only one. If you wanted to move yourself, your cattle, your grain, your steel from here to the other side of a continent, you paid the railroad. The way they measured it was the ton mile, moving one ton one mile.

A wagon was not cheaper. A wagon charged about 15 cents a ton mile, and, he adds with a straight face, there was about a fifty fifty chance of dying along the way. The railroads got it under a penny.

"I do love a good historical connection," he says, "especially ones that go just a little bit further, like we're about to do."

Trucks, planes and boats all move your stuff at a fairly cheap rate. But the railroads are special because they own the land they are built on. Not land that was purchased. Land that was given. The US government, desperately wanting rail lines built, handed them the land at the end of the track and all around the track. Ten square miles of the American West for every mile of track they laid.

Add it all up and the railroads were given about 175 million acres under the Pacific Railroad Acts and their successors. An area bigger than the size of Texas. More than a tenth of the entire United States. The Northern Pacific alone got 44 million acres.

The mechanics of the giveaway are the part worth memorizing. Racing to connect the Pacific to the Atlantic, it was a true race between the Central Pacific Railroad and Union Pacific. To incentivize them to build, for every mile of track laid and verified, the government immediately awarded the land 200 feet on either side of it, and then one square mile of land next to it, alternating out in a checkerboard pattern.

Land the railroads could, would, and did turn right around and sell to everyone they were giving rides to. At 8:40: "They owned the town you got off in." They built the grain elevator. They built the hotel. They sold you the lot you built your house on. The ride and the destination.

And that, owning both ends, is what turns dumb pipes into a very nice business.

Transport businesses, most of them anyway, are sometimes called dumb pipes because there is nothing special about them. It is a pipe you put your stuff into one side of, some time passes, and it comes out the other side. You give your stuff to a trucking company and all you expect is for your stuff to get to point B. Ships and trucks are very hard, very up and down businesses, because there is not much one shipping company can offer that is better than another shipping company except price. So they are constantly in a battle to offer lower prices, but not so low that they end up going out of business.

Ships do not own the ports. Trucks do not own the highways. Railroads were the only ones that morphed into something bigger.

At 9:27 he makes the mapping explicit. That is exactly what SpaceX is doing. Launch is the track. Starlink is the land grant. The orbital shells are the spectrum. The slots up there, SpaceX is claiming them the only way you can on a frontier, by getting there first and occupying them. It owns the ride to orbit and it owns the most valuable real estate at the destination.

The airline trap

And no, of course he did not leave out planes. The thing that flies like a rocket.

That, he says, is all the airlines have in common with SpaceX. Because at 9:50: "The airline is the opposite of the railroad. It is the cautionary tale."

An airline flies you between two airports it does not own. The government built those and assigns the gates. The airline flies you to a city for a $5,000 business deal and it captures the $300 fare once, and not one cent of the deal you flew there to close. It owns the ride and it owns nothing at either end.

After 100 years of doing this, here is where the airlines ended up. Last year, every single major US airline spent more to fly a seat one mile than they earned selling that seat. The actual act of flying you somewhere, they lost money on that. They turned a profit anyway, through credit cards. At 10:23: "They are functionally credit card companies that happen to own planes." They gave up making money on the ride itself long ago.

That is what you become when you own the ride but neither end of the trip. You fight over pennies a seat mile forever.

The railroad owned the destination. The airline owned nothing. At 10:38, the sentence the whole middle of the video builds to: "That single variable, do you own the place you're going, is the difference between a wealth machine and a margin trap."

Business	Owns the ride?	Owns the destination?	Where the money ends up
US railroads, 1800s	Yes, the only one crossing the continent	Yes, 175 million acres of land grant, an area bigger than Texas	Under a penny a ton mile, plus the town, the grain elevator, the hotel and the lot you built on. 60 percent of the entire US stock market by 1880.
Ships and trucks	Yes, the vehicle	No, ships do not own the ports, trucks do not own the highways	A permanent price war. Nothing to sell but a lower rate than the next guy.
US airlines, today	Yes, the plane	No, the government built the airports and assigns the gates	Every major carrier spent more per seat mile than it earned last year. The profit comes from co-branded credit cards.
SpaceX	Yes, 165 launches last year, more than half of all orbital launches on Earth	Yes, the orbital shells and the spectrum, claimed by getting there first	$4 million a ton as a carrier. $13 million a ton as the owner. It collects both checks.

Figure 4. The four way ledger behind the argument. Owning the ride is table stakes and it is not enough. What separates the railroad from the airline is the second column, and Andrews's claim is that SpaceX is the first transport business since the railroads to get a yes in both.

An aside for the sponsor

The Ground News read sits inside this stretch, and it is built out of the same "same facts, different story" instinct as the rest of the video. The June jobs report had just dropped: 57,000 jobs added, against forecasters expecting 115,000. So roughly half.

Here is the fun part, he says. One headline: "far worse than feared." Another headline, same report, same day: "unemployment falls to 4.2 percent." A third: "hiring continues at a healthy clip." Same number, three completely different stories. Is the labor market cracking? Is it fine? Is it secretly good? Depends entirely on who you are reading.

He pulled the story up on Ground News, where 334 sources covered that one report, and you can see side by side exactly how each framed it. Who is calling it a slowdown, who is calling it a soft landing, who is blaming the Fed, who is blaming the White House. The pitch for the product itself: a home page showing what is trending, a blind spot section surfacing stories dominated by one side, viewpoints from across the spectrum on any story, podcast coverage, factuality ratings, and a breakdown of whether each source is corporate controlled, government controlled or independently funded. He calls it the Spotify of the news. The offer is 40 percent off the Vantage plan at groundnews.com/maxinomics.

AI satellites and value creation

Back to the chamber, and the question that decides what SpaceX is worth.

If return per ton is the game, if the whole company is a machine for finding higher and higher value things to put in that ton of space, is a grand piano that earns $13 million the peak? Is that as good as it gets?

Everyone jumps straight to data centers, and he is dismissive of the obvious version of the pitch. Because electricity and water are running out on Earth, and space has free solar power and a vacuum to dump heat into, so put the data center in space. Sure. But he calls that the boring part, and says it skips right past the actual idea.

Here is the actual idea. Someone connects to Starlink. They beam data up, and the satellite beams data back down. That is the whole job. Data goes up one side, the same data comes back down somewhere else. It is a pipe. A beautiful, expensive pipe in the sky, but a pipe nonetheless. The bits that come out are the bits that went in.

He pauses to be decent about the metaphor: "I called ships and trucks dumb pipes, and I mean that with deep respect for the people who do those jobs and run those companies. It is a hard, extremely competitive business." Starlink is a dumb pipe raised to a gorgeous, high margin one, but all it does is move something from one spot to another, unchanged.

So if you want that ton of Starlink to earn more, you really only have two moves. One, cram more subscribers onto each satellite, more people paying to use the same pipe. Two, you stop moving the data and you start changing it. You send something back down that is more valuable than what went up.

Same chamber, same rocket. But instead of an internet relay, you bolt in a single rack of powerful AI chips. You unfold a solar wing 230 feet across, wider than a 747, and you park it in orbit. Data still goes up, but it does not just bounce off and come back. It goes into the chips and it gets transformed, either training an AI model or answering a question you are asking one. And what comes back down to Earth is an answer that did not exist when the signal left the ground.

At 14:00: "If Starlink ships bits across the sky, then this AI satellite, or whatever you want to call it, manufactures them."

It is the same reason a barrel of oil is worth $50 and the gasoline you refine out of it is worth $110. You did not move the thing somewhere. You did something to it. At 14:09: "It is a refinery in orbit."

He swats the cheap power argument one more time on the way past. People want to make this about the cheap power up there, and yeah, fine, the factory goes where the power is cheapest, same reason we built aluminum smelters next to dams a century ago. But that is not the insight. The insight is what the ton does once it is up there. Transport value, or create it?

Figure 5. The ladder, stated at 14:31. Step by step: somebody else's stuff at $4 million per ton, Starlink at $13 million per ton, computer chips at $10 million on the low end and $50 million on the high end. Andrews is blunt that the third bar is a band and not a fact: "Is this easy? No. Fully proven? Also, no."

He then gives the reason it is not a shot in the dark either. SpaceX did this with Nvidia chips last year to see if it would work. It did. The satellite went up, and over the course of six days it trained a small AI model.

That, he says, is why quite suddenly the size of the rocket is by far the most important number in the company.

Starship and the shipping container of space

Everything so far has been about the Falcon 9 chamber. That little house, 17 by 34. Now make the chamber five times bigger and the ton ten times cheaper to fly, and something interesting happens.

To explain what, he goes to 1956.

The entire cargo of a ship used to be unloaded by hand. 194,582 individual items. Crate by crate, sack by sack, picked up, put on a board, hoisted out of the bottom. Ten days to unload one ship.

Until a trucking guy named Malcolm McLean built 58 boxes out of steel, 35 feet down one side, 8 feet across, 8 feet tall. Put everything in the box. Put the box on the boat. The same amount of stuff, loaded in one day. The ship left Newark, New Jersey and landed in Houston five days later.

Voyage complete, the accountants were shocked to find the cost to load a ton of cargo onto the ship went from $5.86 down to $0.16. At 15:53: "36 times cheaper."

And once moving a ton across an ocean was almost free, it stopped making any sense to build things near where you sold them. You built them wherever was cheapest and shipped them in. At 16:05, the payoff: "Cheap transport moved stuff, yes, but more importantly, it moved entire industries."

Starship is the shipping container. That is what this is.

So what does it look like in real life? Starship is intended to be the shipping container of space, built to carry on the order of 100 to 200 tons against the Falcon 9's 17, and aiming to drop the price of a ton to orbit from about $860,000 down toward $100,000, or even $10,000 if you believe the long term promises.

At that price, things we look at now and go "come on, data centers in space?" make all the sense in the world. A two ton rack of AI chips you would never bolt to a Falcon 9 is fine on Starship. The return per ton math gets hit from both directions at once: the cost to put a ton into space goes down, and the value you get back per ton goes up.

And the constellation itself is already waiting on it. The newest Starlink satellites can only fly on Starship. They weigh three times what the current ones do but carry about 16 times the capacity. If one satellite could handle 100 subscribers at a time before, this would handle 1,600.

A bigger rocket, a cheaper ride, much more valuable cargo.

Then he stops and draws a hard line, and it is the most honest thirty seconds in the video. To be very clear, he says, everything from the AI satellites down to this part about Starship is not a forecast, it is the stated plans and active path of the company. And then, at 17:13: "Notably, Starship has put exactly zero working payloads into orbit. Zero." There have been test flights and test payloads, but we have not gotten to official payload number one yet. The cheap ton, the 16 times capacity, the orbital data centers, all of it is "a target on a whiteboard. It is not a thing that has happened."

This whole space bet, he says at 17:32, comes down to one question nobody can really answer yet: "How often can this thing actually fly?"

Footnotes

The last three minutes are the bits that did not have a home in the main argument.

Footnote one: why they bought Cursor. The next step from here, he says, would be to ask what happens if SpaceX is not sending another company's data to the AI satellites, but sending its own data to train its own models, or generating responses to queries from people using those models. That, and he flags this as contrarian, is what he thinks explains the Cursor acquisition. It is like Claude Code or OpenAI Codex, a pure coding tool whose market has already been proven. People will absolutely pay for it. So if you were looking at the return per ton problem the way he thinks SpaceX is, you would look at the data stream going up to those AI satellites and ask why you would take a one time payment to train someone else's model when you could train your own and make money off the result. That only makes sense if you have a product people want, and that is Cursor. It also takes some general risk out of the AI satellite idea and out of the data centers they are building on the ground.

Footnote two: how far ahead they actually are. There are many other companies attempting reusable rockets. Blue Origin and Rocket Lab are the two farthest along, having in some form launched a rocket and recovered the first stage, either on a floating platform like SpaceX or with parachutes. Blue Origin has launched a recovered rocket a second time. That makes it look like these companies are right on SpaceX's heels, and Andrews says they are definitely going to get there, for sure. However, Blue Origin has launched a reused rocket exactly one time, and SpaceX is not standing still. At 19:06: "The lead they have, in my opinion, is 10 to 15 years ahead of everyone else. Many others would say five to 10, so assume I'm being overly generous about that lead." But these things tend to compound. Having proven the tech, having proven a profitable business line in Starlink, SpaceX will find capital and talent much easier to attract, all while building on existing momentum. "This is not a company that's trying to find its mojo."

Footnote three: the number that decides everything. Launch 100 Falcon 9s and make $22 billion spread out over five years. Launch 100 Starships and that is $130 billion spread out over five years. But launch 100 Starships with better Starlink, or something more valuable in the chamber, and that number can rise to $250 to $500 billion. Which just means, at 19:48, that "Starship is everything."

Figure 6. Footnote three, drawn. The middle bar is nothing more than the same $13 million per ton applied to a chamber roughly six times larger, which alone is a sixfold jump. The third bar is what happens if the cargo also gets more valuable. That is the whole reason the video is titled the way it is.

He closes on what that means for the market. He suspects the primary thing that will move SpaceX stock up or down will be Starship making a successful flight with a real payload, and then additional successful flights with real payloads. If it fails, the whole timeline gets set back. If it works, everyone is going to start looking into the future at what that enables, which is much higher revenue.

"And that is it, ladies and gentlemen. Thank you for watching."

Best quotes

"It is the scarce resource the company controls that few others can offer." (0:19)

"This is the only piece of math that matters in this whole video." (1:38)

"The same space, three times the money." (2:28)

"Forget the rocket landing on the barge for a second. The right way to judge SpaceX is return per ton." (2:38)

"Of all the rockets we developed, the reusable one was the worst one." (3:37)

"For 60 years, we just threw the machine in the ocean after one flight." (3:57)

"Every other rocket company on Earth builds a rocket and then waits for the phone to ring. SpaceX builds a rocket and calls itself." (6:02)

"The launch company became its own best Customer." (6:22)

"Being the landlord and tenant, you never have to negotiate. You never have to worry the tenant walks." (6:43)

"They owned the town you got off in. They built the grain elevator, the hotel, sold you the lot you built your house on. The ride and the destination." (8:40)

"Ships do not own the ports. Trucks do not own the highways. Railroads were the only ones that morphed into something bigger." (9:19)

"Launch is the track. Starlink is the land grant. The orbital shells are the spectrum." (9:27)

"They are functionally credit card companies that happen to own planes." (10:23)

"That single variable, do you own the place you're going, is the difference between a wealth machine and a margin trap." (10:38)

"If Starlink ships bits across the sky, then this AI satellite manufactures them." (14:00)

"It is a refinery in orbit." (14:09)

"Cheap transport moved stuff, yes, but more importantly, it moved entire industries." (16:05)

"Notably, Starship has put exactly zero working payloads into orbit. Zero." (17:13)

"How often can this thing actually fly?" (17:32)

"Starship is everything." (19:48)

Where it stands

The internal arithmetic in this video is unusually clean, and it is worth showing that it closes. The four case ladder is consistent to the dollar: reuse saves exactly $29 million a flight ($45 million all in becomes $16 million), which is why $32 million becomes $61 million and $171 million becomes $200 million. The $13 million per ton figure falls out of $7 million of lifetime revenue divided by a 525 kilogram satellite. The $860,000 per ton launch cost is the $15 to $16 million marginal flight divided by 17.4 tons. The 100 launch footnote is just $13 million per ton multiplied out. None of it is hand waved.

Three things are worth knowing on top of what the video says.

The launch price appears twice at two values. Andrews quotes a dedicated Falcon 9 at "about $74 million" during the return per ton section, then at "about $77 million" during the margin breakdown. Both are in the right neighborhood of SpaceX's published list price, and the difference does not change any conclusion, but the $4 million per ton figure comes from the first and the $32 million margin from the second.

The orbital GPU demonstration was not quite a SpaceX experiment. He says "SpaceX did this with Nvidia chips last year to see if it would work." The widely reported result he is pointing at is Starcloud-1, an Nvidia backed startup that put a data center class H100 into orbit on a SpaceX rideshare in November 2025 and trained nanoGPT on the complete works of Shakespeare, then went on to run Google's Gemma in orbit. The result is real and it does what he says it does. The company doing it was a passenger, not the landlord, which if anything sharpens his own argument about who captures the value.

Every forward looking number in here is a target, and he says so. He flags this himself, twice, at 14:36 and again at 17:09. Starship has flown test articles, not paying payloads. The $100,000 and $10,000 per ton figures are stated goals. The $10 to $50 million per ton for AI compute is a band he constructed, not a price anyone has been charged. The 100 launch revenue chart in Figure 6 is arithmetic on assumptions, not guidance. His own closing sentence is the correct caveat: the one question nobody can answer yet is how often the thing can actually fly.

Two of his supporting claims are checkable and check out. SpaceX did fly 165 Falcon missions in 2025, with 123 of them carrying Starlink, roughly 85 percent of all US orbital launches and nearly twice as many as China. And Blue Origin did refly a New Glenn booster in April 2026, becoming the second orbital class family in history to reuse a first stage on a paying mission, on a flight whose upper stage then failed to deliver its satellite to the right orbit. The lead is real. The gap is closing more slowly than the headlines suggest.

Resources

The video and the channel

Why Elon Musk is Really Building Starship, the source video
Maxinomics on YouTube, and the Morning Brew brand page
Maxinomics on X, Instagram and TikTok
His Taiwan video, the source of the anchor tenant idea he reuses here
His companion pieces on the same beat: How SpaceX Is Making America Uncatchable, Why Nations Lose Sleep Over SpaceX and You Don't Want to Run an Airline
Sponsor: Ground News

SpaceX and the hardware

SpaceX, Falcon 9, Starship, Starlink
SpaceX rideshare pricing, the published per kilogram rate
Payload fairing and list of Falcon 9 first stage boosters, including flight counts
SpaceX CRS-8, the April 2016 droneship landing, and SES-10, the first reflight of a recovered booster
Of Course I Still Love You, the barge
Starlink V3 specifications, the Starship only satellites
SpaceX and China drive a new record for orbital launches in 2025 and Space.com on the 165 flight year
Elon Musk

The cost of getting to orbit

Saturn V, Space Shuttle, Apollo program
RP-1 and liquid oxygen, the propellants that are 0.3 percent of the cost
Low Earth orbit and atmospheric drag, why a Starlink satellite only lives five years
NASA and Boeing Space

The railroad and the airline

The first transcontinental railroad, Central Pacific Railroad, Union Pacific Railroad, Northern Pacific Railway
Pacific Railroad Acts and land grant, the 175 million acre giveaway and its checkerboard mechanics
Ton mile, grain elevator, history of rail transport in the United States
Dumb pipe, the term he borrows for transport with no ownership at either end
Available seat mile, the airline unit economics measure behind the seat mile claim, and frequent flyer programs, the credit card business he says airlines really run
Anchor tenant, TSMC, Apple

The container revolution

Malcolm McLean and the Ideal X, Newark to Houston, April 1956
Containerization, and The Box by Marc Levinson, the standard source for the $5.86 to $0.16 loading cost figure
Port Newark and the Port of Houston

AI in orbit, and the competition

Starcloud-1 and CNBC's report on the first AI model trained in space
Nvidia and the H100
nanoGPT and Gemma, the models actually run in orbit
Cursor and CNBC on the $60 billion SpaceX acquisition of Anysphere, plus xAI, Claude Code and OpenAI Codex
Blue Origin, New Glenn and TechCrunch on the first New Glenn booster reflight
Rocket Lab, Electron and Neutron
Oil refinery, aluminium smelting and hydroelectricity, the two analogies for putting the factory where the power is
Boeing 747, the wingspan comparison for the 230 foot solar array

The sponsor segment

Ground News, 40 percent off the Vantage plan
The BLS Employment Situation release, the jobs report 334 outlets framed three different ways
The Federal Reserve

Full transcript

This is what SpaceX is selling, an empty chamber the size of a small house, 17 ft tall, 34 ft long. An empty chamber sits at the tip of their workhorse rocket. Yours to fill for the right price. They'll make sure it gets there. So, how much would you pay to put your stuff inside? Because without this empty space to put stuff in, the rocket is of no use. Doesn't matter if it lands itself. It is the scarce resource the company controls that few others can offer. This space is the business of SpaceX. Understanding how to maximize each cube that space can be broken down into, that is how you understand the potential, the problems, and the promise of SpaceX. This video is sponsored by Ground News. More on them later. So, let's fill it. And let's start with the thing they put in there most, the thing that ends up explaining the entire company. A Starlink satellite weighs about as much as a grand piano, 525 kg, 1,000 lb, flat, plain, mass-produced. 70 of them are made every day. Each one costs about $800,000 to make. That grand piano parked 400 mi up whipping around Earth at incredible speeds quietly continues to bill its customers long after the rocket has landed back on the pad. Spread the whole Starlink business across all the satellites doing the work, and one of them pulls in $1.5 million a year. Think about that. In year one, it cost $800,000 to build, but then it earns double that every year for 5 years, and only 5 years, by design. At the altitude SpaceX chose to place them, the atmosphere places enough drag friction on the satellite that it will eventually succumb to Earth's gravity, friction that causes it to catch fire and burn up as it sinks deeper into the atmosphere. So, you don't get to count its revenue forever, you count it over 5 years, each one throwing off a total of $7 million before dying in a ball of fire. This is the only piece of math that matters in this whole video. Take the lifetime revenue of that satellite, divide by its weight, a ton of Starlink riding in that chamber returns $13 million over its life. Hold onto that number, $13 million, because SpaceX has two completely different things it can do with the area at the tip of that rocket. Option one, sell the seat. Somebody, a company, a country, NASA has a satellite they need in orbit. SpaceX sells them a ride. A dedicated Falcon 9 runs about 74 million and can carry 17.4 tons into orbit. They like the rocket, the fairing opens, the thing floats out, and the transaction is over. They got paid once. You keep whatever your satellite earns for the rest of its life. For that ton of lift, SpaceX paid 4 million one time. Option two, don't sell the seat. Put their own grand piano in it and collect 13 million over 5 years. 4 million once or 13 million, the rent rolling in over half a decade. The same space, three times the money. At least three times the money, because for every Starlink they send up, the product gets better and attracts more users. That is the metric. Forget the rocket landing on the barge for a second. The right way to judge SpaceX is return per ton. How many dollars does one ton of that scarce chamber earn over the lifetime of whatever you put in it. Everything that goes up must pay, once or many times. SpaceX figured out that option two is worth at least three times more than option one. But none of that math works if the chamber is expensive to launch. And for 40 years, it got more expensive, not less. Let's go back to where it all began. This is the price it cost all the way back in the Apollo days to put 1 kg into orbit. And the natural assumption is it goes down over time, right? Technology gets better, things get cheaper. Well, that thought, it went up. The Saturn V, the rocket that took us to the moon, put a kilogram into orbit for about $12,000 in today's money. And then we built the thing that was supposed to fix all of that, the Space Shuttle. Designed to be reused, fly again and again, bring the cost down, the Space Shuttle was the most expensive way human beings have ever reached orbit, about $80,000 a kilogram. Oh, and of all the rockets we developed, the reusable one was the worst one. Until April 2016, when a private company landed a booster on a barge in the middle of the ocean and flew it again. The price of a kilogram to orbit fell more than 20-fold. The fuel, the kerosene and liquid oxygen you actually burn, is about 0.3% of the cost of a rocket, a third of 1%. Everything else is the machine. For 60 years, we just threw the machine in the ocean after one flight. SpaceX, a business that cannot lose money forever and keep operating, stopped throwing away the machine. One of their boosters has now flown something like 34 times. This is the Falcon 9, the workhorse rocket. This part is the booster, the most expensive part. It costs $30 million to build. This is the second stage. It holds the payload. It costs about $10 million to build. When a Falcon 9 takes off, its boosters burn almost all their fuel to push the second stage to the edge of Earth's atmosphere. Once there, it disconnects from the second stage and begins falling back to Earth under the weight of gravity. Tumbling and tumbling until flip, burns the rest of its fuel, landing itself back on the pad. The second stage has one engine in it. Once disconnected from the booster, it kicks on, burning just long enough to get the payload to wherever it needs to go. Doors open, stuff is pushed out, and the second stage says farewell. A final farewell, because it will use the remaining fuel it has to head back towards Earth, so it isn't left floating in space. The second stage is not reusable. It is intended to burn up in Earth's atmosphere, single use. All in, the cost to build one from scratch and then launch the whole Falcon 9 one time is about $45 million. SpaceX will sell you the entire chamber, your own private ride, for about $77 million. Sell the seats and SpaceX makes $32 million. Fill the seats with Starlink satellites, or as we'll see shortly, other potentially more profitable things they want to make, and they'll make $171 million. Once that booster has flown one time, it's paid for itself. Getting it ready to fly again costs just $1 million, so for every additional time the booster flies, sell the seats and they'll now make $61 million, or fill with Starlink satellites and that becomes $200 million. There were conversations about Starlink, from what I understand, early in the company's history, but it wasn't the original game plan. SpaceX set out to sell the space on its rockets. That was the intention. About a decade ago, that changed dramatically. SpaceX flew 165 Falcon launches last year, a record. More than half of every orbital launch on planet Earth by one company. 123 of those were carrying SpaceX's own Starlink satellites, three out of every four. Every other rocket company on Earth builds a rocket and then waits for the phone to ring. Boeing builds a rocket and waits for NASA to call. SpaceX builds a rocket and calls itself. 123 times last year. It works. We now know that more than half of all the money they make comes from Starlink, from people like myself. I actually do have Starlink. I have to. Who do not have other internet available options saying, "I will pay you for that." It is the one part of SpaceX that reliably consistently turns a profit. The launch company became its own best customer. If you saw the Taiwan video I did a couple months ago, you know the idea of an anchor tenant, the customer that you can count on, that will buy everything you're willing to sell to them. You want to do something new, expand, build a new factory, you call them up. Taiwan Semiconductors was Apple. One dominant customer whose presence makes the entire thing work, who you have to attract and keep and pray never leaves. Being the landlord and tenant, you never have to negotiate. You never have to worry the tenant walks. You collect both checks. The cost of the lift and the lifetime revenue of the thing in the seat. Stop me if any of this sounds familiar, like you've heard this story before, maybe about 150 years ago. In 1880, if you bought the American stock market, you were basically buying railroads. Not a piece of it, not a big sector. 60% of the entire market was railroad stocks. For a stretch of the 1800s, the US stock market was almost entirely a market of railroads. They were so big, they basically invented the modern stock market because nothing else on Earth needed that much money at once. Until that point, everybody needed a ride. The railroads owned the only one. If you want to move yourself, your cattle, your grain, your steel from here to the other side of a continent, you paid the railroad. The way they measured it was the ton mile, moving 1 ton 1 mile. A wagon was not cheaper. A wagon charged about 15 cents a ton mile and there's about a 50/50 chance of dying along the way. The railroads got it under a penny. I do love a good historical connection, especially ones that go just a little bit further, like we're about to do. Trucks, planes, boats, they all move your stuff at a fairly cheap rate, but the railroads, the railroads are special because they own the land they're built on. Not land that was purchased, given. The US government desperately wanting rail lines built, handed them the land at the end of the track and all around the track. 10 square miles of the American West for every mile of track they laid. Add it all up and the railroads were given about 175 million acres. It's an area bigger than the size of Texas. More than a tenth of the entire United States. The Northern Pacific alone got 44 million acres. Racing to connect the Pacific to Atlantic, it was a true race between the Central Pacific Railroad and Union Pacific. Because to incentivize them to build, for every mile of track laid and verified, the government immediately awarded the land 200 ft on either side of it, and then 1 square mile of land next to it, alternating out in a checkerboard pattern. Land the railroads could, would, and did turn right around and sell to everyone they were giving rides to. They owned the town. You got off in. They built the grain elevator, the hotel, sold you the lot you built your house on. The ride and the destination. And that, owning both ends, is what turns dumb pipes into a very nice business. Transport businesses, most of them anyway, are sometimes called dumb pipes because there's nothing special about them. It is a pipe you put your stuff in one side of, some time passes, and it comes out the other side. You give your stuff to a trucking company and all you expect is for your stuff to get to point B. Ships, trucks, these are all very hard, very up and down businesses because there isn't much one shipping company can offer that's better than another shipping company, except price. So, they're constantly in a battle to offer lower prices, but not so low that they end up going out of business. Ships do not own the ports, trucks do not own the highways. Railroads were the only ones that morphed into something bigger. That is exactly what SpaceX is doing. Launches the track, Starlink is the land grant, the orbital shells the spectrum, the slots up there. SpaceX is claiming them the only way you can on a frontier, by getting there first and occupying them. It owns the right to orbit and it owns the most valuable real estate at the destination. Now, of course not. No, of course I did not leave out planes. The thing that flies like a rocket? Well, that is all the airlines have in common with SpaceX because the airline is the opposite of the railroad. It is the cautionary tale. An airline flies you between two airports it does not own. The government built those and assigns the gates. The airline flies you to a city for a $5,000 business deal and it captures the $300 fare once and not one cent of the deal flew there to close. It owns the ride and it owns nothing at either end. After 100 years of doing this, here's where the airlines ended up. Last year, every single major US airline spent more to fly a seat one mile than they earned selling that seat. The actual act of flying you somewhere, they lost money on that. They turned a profit anyway through credit cards. They are functionally credit card companies that happen to own planes. They gave up making money on the ride itself long ago. That's what you become when you own the ride but neither end of the trip. You fight over pennies a seat mile forever. The railroad owned the destination and the airline owned nothing and that single variable, do you own the place you're going, is the difference between a wealth machine and a margin trap. The June jobs report just dropped. 57,000 jobs added, forecasters expected 115,000. So, we got half. But here's the fun part. One headline, far worse than feared. Another headline, same report, same day, unemployment falls to 4.2%. A third, hiring continues at a healthy clip. Same number, three completely different stories. Is the labor market cracking? Is it fine? Is it secretly good? Depends entirely on who you're reading and that's the problem. Left, right, center, they're all looking at the same report and reporting something completely different. If you're only getting one version, you are probably missing something. So, I pulled this story up on Ground News. 334 sources covered this one report and you can see side by side exactly how each one framed it. Who's calling it a slowdown? Who's calling it a soft landing? Who's blaming the Fed? Who's blaming the White House? That's why I've been spending more and more time on Ground News. It's a website and app that pulls coverage from hundreds of sources so you can see every story through a wider lens. There's a home page showing what's trending, a blind spot section surfacing stories dominated by one side, stuff you never find in an algorithm driven feed or on X. Click into any story and you get viewpoints from across the spectrum, podcast coverage, factuality ratings, even a breakdown of whether each source is corporate controlled, government controlled, or independently funded. You can think of it kind of like Spotify of the news. This QR code, made specifically for Maxonomics viewers, gets you an exclusive 40% off their Vantage plan. It's the same one I use. Scan it or head to ground.news.com/maxonomics to subscribe. If you care about actually understanding what's happening in the world, it is worth it. Thank you to Ground News for making videos like this possible. Now, let's get back to it. So, let's go back to the chamber and ask the question that decides what SpaceX is worth next. If return per ton is the game, if the whole company is a machine for finding higher and higher value things to put in that ton of space, is a grand piano that earns $13 million to peak, is that as good as it gets? Data centers guzzle electricity and water. Earth is running out of both. Space has free solar power and a vacuum to dump heat into, so ta-da, put the data center in space. And sure, all true, it's also kind of the boring part and it skips right past the actual idea. Someone connects to Starlink. They beam data up and the satellite beams data back down. That's the whole job. Data goes up one side, the same data comes back down somewhere else. It's a pipe, a beautiful expensive pipe in the sky, but a pipe nonetheless. The bits that come out are the bits that went in. I call ships and trucks dumb pipes and I mean that with deep respect for the people who do those jobs and run those companies. It is a hard, extremely competitive business. Starlink is a dumb pipe, too. A gorgeous high margin one, but all it does is move something from one spot to another unchanged. So, if you want that ton of Starlink to earn more, you really only have two moves. One, cram more subscribers onto each satellite, more people paying to use the same pipe, or two, you stop moving the data and you start changing it. You send something back down that's more valuable than what went up. Same chamber, same rocket, but instead of an internet relay, you bolt in a single rack of powerful AI chips, you unfold a solar wing 230 ft across, wider than a 747, and you park it in orbit. Data still goes up, but it doesn't just bounce off and come back. It goes into the chips, and it gets transformed. Either it's training an AI model, or it's a question you're asking one. And what comes back down to Earth is an answer that did not exist when the signal left the ground. If Starlink ships bits across the sky, then this AI satellite, or whatever you want to call it, manufactures them. It's the same reason a barrel of oil is worth $50, and the gasoline you refine out of it is worth $110. You didn't move the thing somewhere, you did something to it. It is a refinery in orbit. People want to make this about the cheap power up there, and yeah, fine. The factory goes where the power is cheapest. Same reason we built aluminum smelters next to dams a century ago. But, that's not the insight. The insight is what the ton does once it's up there. Transport value, or create it? Step-by-step, we have gone from taking someone else's stuff up, 4 million per ton, taking Starlink satellites up, 13 million per ton, taking computer chips up, 10 million at the low end, high end at 50 million. Is this easy? No. Fully proven? Also, no. That is important to say at the moment this video will come out. However, it's also not a shot in the dark. SpaceX did this with Nvidia chips last year to see if it would work. It did. The satellite went up, and over the course of 6 days trained a small AI model. This is why, quite suddenly, the size of the rocket is by far the most important number in the company. Everything we've talked about, the grand piano, the refinery in the sky, rides in the Falcon 9 chamber, that little house, 17 by 34. Now, it's time to fill this spot. Make the chamber five times bigger, and the ton 10 times cheaper to fly, and something interesting happens. Starship is built to be the shipping container of space. The entire cargo of a ship would be unloaded by hand. 194,582 individual items, crate by crate, sack by sack, picked up, put on a board, and hoisted out of the bottom. 10 days to unload one ship. Until a trucking guy named Malcolm McLean built 58 boxes out of steel, 35 ft down one side, 8 ft across, 8 ft tall, put everything in the box, put the box on the boat, same amount of stuff loaded in one day. The ship left Newark, New Jersey, and landed in Houston 5 days later. Voyage complete, the accountants were shocked to find the cost to load a ton of cargo onto the ship went $5.86 down to $0.16. 36 times cheaper. Once moving a ton across an ocean was almost free, it stopped making any sense to build things near where you sold them. You built them wherever was cheapest and shipped them in. Cheap transport moved stuff, yes, but more importantly, it moved entire industries. Starship is the shipping container. That's what this is. Now, what does this look like in real life? Starship is intended to be the shipping container of space, built to carry on the order of 100 to 200 tons against the Falcon 9 17, and aiming to drop the price of a ton to orbit from about $860,000 down toward $100,000, even 10,000 if you believe the long-term promises. At that price, things things that we look at now is like, come on, data centers in space? Make all the sense in the world. A 2-ton rack of AI chips you would never bolt to a Falcon 9, fine on Starship. The return per ton math gets hit from both directions. The cost to put a ton into space goes down, the value you get back per ton goes up. The newest Starlink satellites can only fly on Starship. They weigh three times what the current ones do, but they carry about 16 times the capacity. If one satellite could handle 100 subscribers at a time before, this would handle 1,600. A bigger rocket, a cheaper ride, much more valuable cargo. And to be very clear, everything I've said from the AI satellites down to this part about Starship, it's not a forecast, but the stated plans and active path of the company. So, notably, Starship has put exactly zero working payloads into orbit, zero. There have been test flights and test payloads, but we haven't gotten to official payload number one yet. The cheap ton, 16 times capacity, a million data centers, it is a target on a whiteboard. It is not a thing that has happened. This whole SpaceX bet comes down to one question nobody can really answer yet. How often can this thing actually fly? Okay, it is time for the bits and pieces that didn't quite make it into the video. The interesting parts that I think are still very useful, very interesting. They just didn't have a home in the main thing. We can call them the footnotes and this is number one. The next step from here would be to ask, well, what if SpaceX wasn't sending another company's data to the AI satellites, but was sending either its own data to train their own models or generating responses to queries from people using those models. That to me, and this is contrarian from what I can tell, looks like the reason they bought Cursor. It's like Claude Code or OpenAI's Codex. It's a pure coding tool. The market for that tool has already been proven. People will absolutely pay for it. So, if you were to look at this how to maximize the return per ton problem, the way you should and the way in which I think they are, you would look at the data stream going up to these AI satellites the same way. Why get the one-time payment to train someone else's model when you could train your own and make money off the result? Only makes sense if you have a product that people want and that is Cursor. It also takes some general risk out of the AI satellite idea and the data centers they're also building on the ground. Number two. There are many other companies attempting to do reusable rockets. Blue Origin and Rocket Lab are the two others that are the furthest along that have in some form launched a rocket and recovered the first stage either on a floating platform like SpaceX or with parachutes. Blue Origin has launched a recovered rocket a second time. This makes it seem like these companies are right on the heels of SpaceX for reusable rockets and they're definitely going to get there for sure. However, Blue Origin has only launched a reused rocket one time and SpaceX is not standing still. The lead they have, in my opinion, is 10 to 15 years ahead of everyone else. Many others would say five to 10, so assume I'm being overly generous about that lead. But, these things tend to compound. Having proven the tech, having proven a profitable business line in Starlink, SpaceX will find capital and talent much easier to attract, all while building on existing momentum. This is not a company that's trying to find its mojo. Number three. Launch 100 Falcon 9s, make $22 billion spread out over five years. Launch 100 Starships, that's $130 billion spread out over five years. But, if you launch 100 Starships with better Starlink or something more valuable in the chamber, that number can rise to 250 to $500 which just means that Starship is everything. So, I suspect that the primary thing that will move SpaceX stock up or down will be Starship making a successful flight with a real payload. And then, additional successful flights with real payloads. If it fails, the whole timeline gets set back. If it works, everyone is going to start looking into the future at what that enables, which is much higher revenue. And that is it, ladies and gentlemen. Thank you for watching. I will see you in the next video. See you.