Iqbal Latif
The second of two essays. The first, I Refuse to Accept a Demigod, asked what a man is for if Elon Musk is right. This one grants him everything and counts. In my last essay I refused to accept a demigod. Now let me do something his followers never do. Let me accept every word he says. From 1 January 2027 every promise works, on schedule, and nothing fails. Starship is fully reusable at his own target price. Rockets fly every day. A city of 1 million people rises on Mars. The Moon is mined. Data centres fill the sky, and satellites become the world’s communications network. Robots number in the billions, robotaxis fill the streets, and the Treasury sends everyone a high income. The lenders keep lending, the shares keep rising, and he keeps the credit for all of it. I will not argue with the physics. I will ask only the question an engineer is trained to ask before anything is built. Who pays, out of what, and what happens to the rest of the world while it is being paid for?
Two of his promises are not speeches. They are written into his contract. SpaceX’s own prospectus, filed with the American regulator this year, grants him 1 billion shares that vest only when the company reaches market values up to $7.5 trillion and a permanent colony of at least 1 million people is living on Mars, and 302 million more tied to market values up to $6.565 trillion and to data centres in space capable of 100 terawatts. The same filing shows him holding 85.1% of the company’s voting power before the listing. Those two grants are the spine of this essay, because a company and its founder signed them. The rest, the robots and the checks, he said in public, and I take him at his word. Every number I use is his, his company’s, or an assumption I name as mine. And at the end I have drawn up what nobody has shown us: a balance sheet of the Earth for the years 2027 to 2050, in plain words, so that anyone can see what we have, what his fantasy demands, and the gap between them. If you read nothing else, read that.
Let me begin with two concessions, because they make everything that follows harder to escape. The first is his own sum for Mars. He has said a city on Mars that can grow by itself needs at least a million tonnes of equipment. At today’s price of landing cargo on Mars, about $1 billion a tonne, that would cost over $1,000 trillion, which he rightly calls impossible. His answer is that Starship makes it 1,000 times cheaper, bringing the whole colony down to about $1 trillion. Spread over 40 years, that is less than $25 billion a year, about 0.02% of a $124 trillion world economy. If that number held, he would be right. The rocket is not where his future breaks.
The second concession is the market. At the very beginning, the bond market agreed with my doubts. When his bonds first traded in June their prices fell, and by August the cost of insuring against a SpaceX default had climbed to record levels. Since then, as the shares have recovered, the spread has contracted. The lenders are charging him less for his risk. His stories are selling, and I say so plainly. But a bondholder is paid first, from the money that exists today: Starlink, the launches, the rent from the AI companies. A bond is a bet that the company survives, not that Mars gets built. The bond market doubted him first and forgave him later, but it is still lending against Starlink, not against Mars.
So grant him the $1 trillion. Grant him the credit. Now count.
A city of 1 million people at about 100 people a ship needs some 10,000 ships of people. His million tonnes of cargo at about 100 tonnes a ship needs 10,000 more. That is 20,000 ships, and most of them are meant to stay and become the city. Every ship bound for Mars must be refuelled in orbit before it leaves, and nobody knows the exact number of tanker flights. If each ship needs 5, the Mars programme needs about 120,000 launches. If each needs the high teens that NASA estimated for a single landing on the Moon, it needs about 320,000. At 27 launches a day, every day, without a single failure or pause, that is 12 years of flying at the kind end and 32 at the other. And the planets line up only once every 26 months. To reach 1 million people by 2050, he would need around 2,000 ships leaving at each window, twice the fleet he has spoken of. If a launch ever costs as little as $10 million, an aspiration and not a fact, the launches alone cost $1.2 trillion to $3.2 trillion. His whole colony was supposed to cost $1 trillion. The rockets have spent it before a single house is built.
And we have only counted the journey out. Some of those ships must come home, with crews, with the sick, with anyone who changes their mind. Leaving Mars is easier than leaving Earth, because its gravity is only about 38% of ours, but a Starship still needs its tanks filled on the ground: well over 1,000 tonnes of fuel and oxygen for each flight home. Carry that from Earth, and every returning ship needs about 10 cargo ships of fuel sent ahead of it, each of which needed its own tankers to leave Earth. That is impossible, and he knows it, so his plan is to make the fuel on Mars, out of the Martian air and the ice beneath the ground. But a fuel factory on Mars is not a wish. It needs mining for ice, a chemical plant to turn carbon dioxide and water into methane and oxygen, and a power station large enough to run it, in sunlight less than half as strong as ours and through dust storms that can darken the sky for weeks. All of it must be carried from Earth, built and working before the first ship can come back. The only oxygen making technology yet demonstrated on Mars, NASA’s small MOXIE experiment on the Perseverance rover, was a test, not a factory, and it made about 122 grams of oxygen in two years. A single ship needs roughly 800 tonnes of oxygen to come home. Nothing on the scale he needs has ever been built on another world. It appears in no price I have seen, including his.
Then there is the city itself. His million tonnes works out at about one tonne of equipment for each settler. For comparison, everything humans have built on Earth, the concrete, steel, roads, buildings and machines, came to roughly 1.1 trillion tonnes around 2020, about as much as all living things on the planet, or roughly 140 tonnes for every person alive. Mars will not need our motorways and dams, so nobody should read that as a Martian requirement. But somewhere between one tonne and 140 lies everything a settler needs that his cargo does not carry: the mines, smelters, chemical plants, glassworks, electronics factories and pharmacies that do not exist there and that took Earth 200 years to build. All of it must be built on Mars, out of Martian rock. Nobody has priced that, including him.
The Moon needs its own million tonnes. At about 100 tonnes a landing and 10 to 16 tanker flights for every landing, that is another 100,000 to 160,000 launches. Then comes the sky. On 30 January 2026 his company asked the American regulator for permission to fly up to 1 million satellites as data centres in orbit, which it called the first step towards a civilisation that harnesses the full power of the Sun. Today about 14,000 working satellites of every kind circle the earth. His share grant ties one award to what it calls “non-Earth-based data centers capable of delivering 100 terawatts of compute per year.” The phrase is strange, because a watt is already a rate. I read it as 100 terawatts of power; if the company means something smaller, it should say so, and say what. On that reading the company’s own arithmetic, 1 million tonnes a year in orbit for about 100 gigawatts, gives the scale: 100 terawatts is 1,000 times that, about 1 billion tonnes in orbit, 5 to 10 million launches, and $100 to $200 trillion even at his dream price of $100 to $200 a kilogram. It is about five times all the energy the human race now uses. And notice what his two numbers say together. A billion tonnes shared among his 1 million satellites is about 1,000 tonnes each, more than twice the International Space Station, in every single one. The two claims cannot be combined without an assumption the company has not explained: a satellite with a radically different power to weight ratio, a different meaning of 100 terawatts, or a different design altogether. Until it explains which, one of the two numbers cannot be taken as it stands. Nor does a satellite in low orbit last forever. If these last about five years, which is my assumption, 1 million of them means some 200,000 new satellites a year, and about 550 old ones falling back and burning up in our upper air every single day. What that does to the atmosphere is still an open question. Nobody has answered it, and nobody asked the rest of us whether our sky could be used this way. Put Mars, the Moon and orbit together, and 27 launches a day is not the ceiling of his plan. It is the floor. And beneath the computers in the sky sits a business I have not yet counted: the net that connects them to us. Starlink is not only a way of reaching a farmhouse without fibre. The grander investment story is that satellites take a growing share of the business now carried by towers, fibre and the 4G and 5G networks of every telephone company on earth. That matters, because Starlink is the engine meant to pay for everything else. The orbital computers consume capital; Starlink is supposed to generate it. The Earth business finances the space business. So ask what a trillion dollar business means: a trillion of revenue every year, or a trillion of market value? They are not the same thing, and a trillion of market value does not pour a single tonne of concrete on Mars. The world already spends well over $1 trillion a year on telecom services. For the net to pay for Mars, the Moon, the robots and the orbital computers, it must capture a great part of that spending from the households, firms and governments that pay it today. And to the extent that the new net displaces the old one, part of the bill is paid twice: once for the net in the sky, and again in the write off of the towers, cables and exchanges on the ground, owned, once again, by pension funds and infrastructure funds. The saver pays for the new network and loses on the old.
(To be continued)
