The Arithmetic of the Moon Pitch

The Pitch

When SpaceX executives took the stage at JPMorgan's Manhattan headquarters this week to make the case for the company's public debut, the question on the table was simple: why go public, and why now. The answer was not about launch contracts or satellite internet. It was a vision.

The setting was not incidental. Wall Street has been competing for the prestige of the listing, with rocket sculptures and launch-themed lighting in bank lobbies, livestreamed launch parties across the country, and more than twenty lenders attached to the deal. When the host introduced Musk by comparing him to Thomas Edison, the room of several hundred investors was being offered a story as much as a security. It is worth separating the two.

Elon Musk described a future in which the proceeds of the IPO fund artificial intelligence data centers built in space, which he framed as the primary way AI compute can keep expanding once power plants on the ground run into resistance and physical limits. He floated figures on the order of one terawatt of AI compute from Earth orbit and as much as 1,000 terawatts or more generated from the surface of the Moon. He talked about sending another 100,000 Starlink satellites aloft, about hotels for tourists visiting the Moon, and about human beings living in cities on Mars. The roadshow materials put the total opportunity at a $28.5 trillion addressable market, a figure roughly the size of the entire annual output of the United States economy.

It is a genuinely thrilling story, and parts of it rest on a real and impressive company. The purpose of this article is narrow and specific. Take the vision at face value, then measure it against what SpaceX and the rest of humanity have actually demonstrated they can do, today, with the hardware that exists. The gap between the two is the part an investor is being asked to pay for.

The Refueling Problem

Start with the simplest version of the dream: putting cargo on the Moon at all. The vehicle meant to do it is Starship, and a single Moon-bound Starship cannot reach the lunar surface on one tank of fuel. It launches nearly empty, then has to be refueled in low Earth orbit before it can continue. That refueling is done by sending up a fleet of separate Starship tankers, each delivering propellant to an orbiting depot, which then tops off the lander.

How many tanker flights does one lunar trip require? The estimates are public and they are not small. SpaceX's own optimistic figure has been 8 to around 10 launches. NASA and the Government Accountability Office have put it higher, in the range of 16 to 19 launches per single lunar landing, in part because cryogenic propellants boil off in space and must be replaced. Either way, the conclusion is the same: delivering roughly 100 tonnes to the lunar surface costs somewhere between 8 and 19 launches. That works out to about one launch for every 5 to 12 tonnes that actually lands on the Moon.

And those counts are the tankers alone. On top of them, one launch places the propellant depot in orbit and another launches the lander itself, so the true total for a single mission runs higher still. The launches cannot be spread out for convenience, either. Because the supercold propellant slowly boils off and vents away, the entire refueling campaign has to be flown in rapid succession, which raises the bar on cadence before the lander has even left Earth orbit.

And there is a catch underneath the catch. Transferring cryogenic propellant between two vehicles in orbit, the maneuver this entire architecture depends on, has never been demonstrated at scale. It is on the schedule, not in the record book. So before a single tonne of a Moon hotel reaches the lunar surface, SpaceX first has to perfect a fueling procedure no one has ever completed, then repeat it perfectly, in rapid succession, many times over, for every single landing.

The Cadence Problem

Now scale it up. If one delivery of 100 tonnes costs roughly a dozen launches, how many launches can anyone actually perform?

Here SpaceX holds the record, and it is worth being honest about how dominant that record is. In 2025 the company launched 165 orbital missions on its Falcon 9 rocket, more than any country or company has ever managed in a single year, about one launch every other day, and roughly 85 percent of all orbital launches from the United States. That is the ceiling of demonstrated human capability, and it is a recent ceiling: SpaceX flew 25 orbital launches in 2020 and climbed every year since, through 61, 96, and 134, to reach 165. That ramp is one of the great industrial feats of the decade, and it still falls orders of magnitude short of what the lunar vision quietly assumes.

Apply it to the Moon. If every one of those 165 record-setting launches were instead a lunar tanker, the total tonnage delivered to the lunar surface in a year would land somewhere around 1,000 to 2,000 tonnes. For perspective on what that builds, consider the largest structure humanity has ever assembled off the planet. The International Space Station weighs about 420 tonnes. It took 42 assembly flights and more than a decade to build, and it was built in low Earth orbit, which is far easier and cheaper to reach than the surface of the Moon. A lunar data center, let alone a city, would have to dwarf the Space Station many times over. The cadence Musk's vision quietly assumes is not a single record year. It is many launches per day, every day, for years, all of them landing on the Moon.

Bring it back to the actual subject, a data center. A single large hyperscale campus on Earth runs to tens of thousands of tonnes of equipment and structure. At the best lunar delivery rate humanity has ever been capable of, around 1,000 to 2,000 tonnes a year and only if every record-setting launch were a Moon tanker, putting the mass of one ordinary data center on the lunar surface would take a decade or more of flawless, record-breaking launches devoted to nothing else. The pitch is not one data center. It is an industry.

The Power Problem

Suppose the cargo somehow arrives. A data center is not a monument. It has to run, which means it has to be powered and cooled, and the Moon is hostile to both.

A lunar day and night each last about 14 Earth days. Solar panels on the surface generate nothing for roughly two weeks at a stretch. To carry a single one-gigawatt data center through one lunar night on stored energy would require on the order of 340 gigawatt-hours of batteries. The entire planet currently manufactures roughly 1,000 gigawatt-hours of batteries per year across every factory on Earth. In other words, powering one gigawatt of compute through one lunar night would consume close to a third of the world's annual battery production, and every kilogram of it would have to be flown to the Moon. The alternative is nuclear, but no reactor has ever operated on the lunar surface, and reactors are heavy cargo of their own.

Cooling is the mirror image of the same problem. On Earth a data center sheds heat into air and water. In the vacuum of space there is neither, so the only way to dump waste heat is to radiate it away, which demands enormous radiator panels that must themselves be built, launched, and landed. A facility drawing a gigawatt of power has to reject close to a gigawatt of heat, continuously, with no air to carry it off. For reference, the entire International Space Station runs on about 90 kilowatts, roughly one eleven-thousandth of a single gigawatt, and managing even that modest heat load is one of the station's hardest engineering jobs.

Then recall the headline number. The pitch was not one gigawatt. It was on the order of 1,000 terawatts of lunar compute, which is one million gigawatts. For scale, all of humanity generates roughly 3 terawatts of electricity on average across the entire planet. The figure being floated for the Moon is several hundred times the electrical output of human civilization, generated on a body with no atmosphere, two-week nights, and no power grid.

The Vehicle Problem

Every number above assumes the central tool already works. It does not yet.

Starship is still in flight testing. The most recent attempt, Flight 12 on May 22, 2026, was a suborbital test, and the booster failed to complete the burn it needed for a controlled landing. Across all of 2025, SpaceX flew five Starship test flights, several of which ended in failure, including a second stage that exploded before a static fire. The rocket has not reached orbit with a payload, has not flown the high-cadence reusable profile the economics require, and, as noted, has not demonstrated the orbital refueling that makes lunar missions possible in the first place. NASA's own inspector general has flagged that in-orbit refueling as among the riskiest and least proven steps in the entire program, and the agency's lunar landing timeline has slipped repeatedly as the demonstration keeps receding into the future.

This is the heart of the matter. The Moon hotels, the Mars cities, and the lunar data centers all sit on top of a vehicle that, as of this writing, has not yet completed the basic steps the entire vision is built on. None of that means it never will. It means the vision is being priced today as though the vehicle is finished, when it is not.

The Mars Window

Mars adds a constraint that no amount of engineering can iterate away, because it is set by orbital mechanics rather than by technology. Earth and Mars line up favorably for a fuel-efficient trip only about once every 26 months, and each window lasts a matter of weeks.

That has a brutal implication for any plan to move serious quantities of people and material to Mars. You cannot spread the launches out across the year to fit your launch cadence. You have to fire an entire fleet's worth of cargo during a narrow window, then wait more than two years for the next one. Everything the colony needs for that stretch has to go in the same handful of weeks, on rockets that, again, have not yet flown the profile this requires. The Moon's problem is cadence and tonnage. Mars adds a clock that the laws of physics control, not SpaceX.

Distance compounds it. A one-way trip to Mars takes roughly seven to nine months, so every crew and every load of cargo is committed for the better part of a year in transit before the window math even begins. A self-sustaining settlement, which is what is actually being described, would need habitats, radiation shielding, power, water and air recycling, food production, and the means to manufacture return propellant, all delivered and assembled across those rare windows. For context, humanity has never kept a person alive beyond low Earth orbit for longer than the few days of an Apollo mission, and no human has traveled past it since 1972. The vision is not a larger version of something we have already done. It is a category we have not yet entered.

What This Is Being Sold For

None of this would matter much if it were a founder musing on a podcast. It matters because it is the justification attached to the largest stock sale in history, and because of who is being asked to buy. SpaceX plans to sell about 555.6 million shares at $135 each, raising roughly $75 billion at a valuation near $1.75 trillion, and the offering is being marketed not only to institutions but to retail investors and into retirement accounts.

To put the size in perspective, the previous record holder, Saudi Aramco in 2019, raised about $29 billion; SpaceX aims to raise more than two and a half times that in a single offering. At $1.75 trillion the company would be worth more than Tesla and would rank among the half-dozen most valuable public companies on Earth from its first day of trading. Unusually, SpaceX set a fixed price rather than the customary range that adjusts to investor demand, which leaves the market less room to talk the number down before the opening bell.

The sober numbers tell a different story than the vision. SpaceX posted a net loss of about $4.94 billion last year on $18.7 billion of revenue, a swing from a small profit the year before, and it is carrying a $20 billion bridge loan that IPO proceeds will partly repay. An independent valuation from Morningstar puts the company near $780 billion, less than half the asking price, and calls it significantly overvalued. The difference between that estimate and the IPO number, close to a trillion dollars, is essentially the price of the Moon-and-Mars story. The real business, Starlink and launch services, is valuable and worth owning at the right price. The question is whether a buyer at $1.75 trillion is paying for that business or for the part of the pitch this article just measured.

Tell Them the Science

None of this is written to diminish SpaceX, and it would be dishonest to pretend otherwise. SpaceX is one of the most impressive engineering organizations on the planet. It made orbital-class rockets reusable when the entire industry said it could not be done. It launched more rockets in a single year than the rest of the world combined. It catches a returning booster out of the sky with a pair of mechanical arms. Starlink is a real, working network of thousands of satellites delivering internet to places that never had it, and it earns real money doing so. The space program, public and private, is one of the genuine achievements of our species, and there is nothing to root against here. Many of us would love for the Moon hotels and the Mars cities to be real, and to be real soon.

But there is a difference between admiring an engineering company and pricing a stock on science fiction. When the buyers are ordinary investors and the savings inside ordinary retirement accounts, the number on the prospectus should rest on what has been demonstrated: the launches, the satellites, the revenue that actually exists. The interplanetary vision belongs in the pitch as what it is, a long-shot option that may pay off spectacularly or may not arrive for decades, not as the foundation of a trillion-dollar valuation sold to people who will not run the arithmetic themselves.

That is the entire philosophy of Wealth Engine Pro in one example. We are not here to tell you the dream is impossible, and we are not here to tell you to buy or to sell. We are here to put the verifiable numbers on the table next to the story, so that you can see both clearly and decide for yourself. The vision is extraordinary. The science is unsentimental. Anyone being asked to fund the first deserves to be told the second.