The Pull to Leave

For ten thousand years we built the case for staying.

Fields, walls, deeds, mortgages, the slow machinery of settlement, all of it engineered to hold a human body in one place long enough to be counted, taxed, married, and buried within a day's walk of where it was born. We got good at it. In your lifetime we got so good at it that leaving became almost unnecessary. Food arrives at the door. Work arrives on a screen. A whole world folds down into a bright rectangle in your palm, and on the nights it holds you tightest, the rectangle is the very thing you reach for.

And then, at the exact moment we finished building a life you never have to leave, the money and metal aimed at leaving the planet began to climb. Not gently. In 2020 the world managed 114 orbital launches. Five years later it set a record at 330.¹

Triple the traffic, aimed straight up, in the same handful of years we spent perfecting the art of never going anywhere. Why?

Because the pull to leave is older than any reason we give for it. Older than the rockets, older than the flags they carry, older than the countries that stitch the flags. Every generation is handed a horizon and told the map ends there, and in every generation a few people walk to the edge of it and refuse to believe the map. The edge used to be an ocean. Then a coastline, a mountain, a pole, a line on a chart with sea monsters drawn past it. We ran out of coastlines about a century ago. The horizon had nowhere left to go but up.

What changed in your lifetime was not the wanting. The wanting is constant. What changed was the price.

For most of the space age, leaving cost a treasury. A seat above the atmosphere was reserved for a few test pilots and a few heroes, chosen by governments, paid for by nations, mourned in public when it went wrong. To lift a single kilogram to orbit in the Space Shuttle years cost about $54,000. Then rockets learned to land on their own tails and fly again, and that number fell to roughly $3,000. About eighteen times cheaper, in a single generation.²

A moonshot became a line item. And once something becomes a line item, someone builds a business on it, and someone else builds a hundred more. That is supposed to be the whole story of how a frontier opens to everyone. It is not the story. Here is the part left out of the brochure.

The cost fell. The price did not. That eighteenfold saving on lifting a kilogram was mostly kept, not passed down. It became margin, and margin became the fortunes you read about. The customer at the bottom, the one meant to inherit a cheap frontier, is still waiting. And the next great drop, the one that would actually crack the sky open, the one every rosy projection quietly assumes, rides almost entirely on a single vehicle. Not the Falcon rockets that already reach orbit most weeks, proven and routine, but the new one behind them. Starship, the giant fully reusable machine the cheap future is built on, has flown about a dozen times as of the middle of 2026 and has not once completed an orbit. Some of that is by design, each test flown deliberately just short of orbital speed so it falls back into the sea rather than stranding itself overhead. The rest is the plain fact that the hardest steps are still ahead.³ I want to be plain about how much weight that carries. Nearly everything optimistic anyone says about what happens up there next is standing on a rocket that has not yet done the full thing. My bet is that it gets there, eventually. But a bet is all it is, and everything downstream inherits the risk.

And it is not one man's rocket, whatever the headlines suggest. Rocket Lab flies its own from two hemispheres. Blue Origin's New Glenn is off the pad. NASA still writes the largest checks and sets the hardest goals, the ones with no business case, which is precisely why a government has to set them. The reach into space is an industry, not a personality. But be honest about its shape. In 2025, one company's one rocket carried roughly 80 percent of all the mass humanity put into orbit. A single vehicle. Most of a species' entire outbound cargo. A frontier that is open, in theory, to anyone, and traveled, in practice, by almost no one.

So who is up there, and what are they doing?

Look up on a clear night, away from the city, and wait. Before long something crosses. Not a star. Stars do not move like that. A bright point sliding steady and fast across the whole black field, then another, and if you know where to look, a straight line of them like a string of luminous beads someone hung over your head while you weren't watching. Someone did. In 2019 there were around 2,200 working satellites over the Earth. By 2025 there were more than 11,800.⁴ We rebuilt the night sky in six years. Nobody asked you. There is no ballot for the sky.

And the one place up there where humans actually live is scheduled to fall.

The International Space Station is old, and tired, and planned for retirement around 2030. It cannot come home. So the plan is to drive it down into the ocean on purpose, a controlled destruction with its own line item, worth up to $843 million, awarded for the single task of making the largest thing we ever built in orbit fall where we choose.⁵ The commercial stations meant to replace it keep slipping their dates. Which leaves a real chance that for the first time in a quarter century, there will be a stretch with no human being living off the planet at all.

So is the door opening, or closing? Follow the money, and it gets stranger.

The whole enterprise of space was worth about $626 billion in 2025. Larger than the economy of most nations on the ground, and nearly all of it invisible to the people living under it. You use it every hour. The map that finds you, the card that clears, the forecast, the call that connects. You just never look up at the part of the machine hanging in the dark.

But whose $626 billion is it? That is where the frontier stops looking like a frontier.

The fastest-growing share of it is not chasing tourists or broadband. It is sovereign. In 2025 the majority of every public dollar the world spent on space, roughly $73.5 billion of it, went to defense and national security.⁶ More than half. The oldest reason to climb a hill has not changed since the first army climbed one. You can see farther from up there, and whoever sees farthest moves first. The sky is being claimed. Quietly, expensively, by the few who can afford to see that far.

And every so often the claiming shows up in a single number that stops you cold. In June of 2026, the largest rocket-and-satellite company on Earth held the largest public offering in the history of markets and came out valued near $1.77 trillion.⁷ Months earlier, that same company had folded a frontier artificial-intelligence lab and one of the largest attention platforms in the world into itself. Attention, intelligence, and orbit, under one roof, owned by a very few. If you have followed the shape of the last decade you have seen this move before, smaller. It is the same old pattern, tightening. A great many of us arrive at each new frontier as the user. A very few walk away as the owner. Attention was the first thing we ever sold from up close, and now it sits in the same building as the machines that predict us and the rockets that leave. None of which, I want to insist, is the reason a single human being has ever looked up.

The Invisible Frontier

Not every frontier is up. Some are so small you will never see them, and one of them is arriving right now, built by a few thousand people almost none of us could pick out of a crowd, shaping a world the rest of us will only ever meet secondhand.

Say the word out loud and most people picture a computer. A machine that breaks every code and solves every problem. That machine is real, and it is, as it has always been, about five years away. But while the famous version stalls, a quieter version already ships, already flies, already works in the field. Two frontiers wearing the same word, moving at completely different speeds.

Take the computer first, because it is the one everyone fears. The most concrete public roadmap points to a genuinely useful, error-corrected machine, around 200 stable logical units of computation, arriving in 2029.⁸ I would take that date, and every date like it, and hold it loosely. Machines of this kind have been five years away for the better part of twenty years. The obstacle is not ambition or money. It is that the raw pieces are so fragile you need something like a thousand of them cooperating just to get one that behaves. The most powerful computer ever imagined is being assembled out of the least reliable parts we have ever tried to trust.

So the machine keeps not arriving. And while everyone watches the door it hasn't come through, its quiet twin walked in the side entrance years ago.

The twin is the sensor. The same strange physics that makes the computer so hard makes an instrument that feels things nothing else can feel. A company called Infleqtion and NASA's Jet Propulsion Laboratory are building the first of its kind to fly in space, a quantum gravity sensor, announced early in 2026 and aimed at orbit by the end of the decade.⁹ It is up to ten times more sensitive than the classical instruments it replaces, and it does not drift, because it measures against atoms that are all exactly, perfectly identical, the one truly standard ruler in the universe. Point a thing like that down at the Earth and it feels the pull of mountains, of water tables, of what is buried under ice and under ground, through cloud, through night, through everything that used to hide.

Or point it at yourself. A firm called Q-CTRL reported quantum navigation running as much as fifty times more accurate than the classical alternative, in flight trials where the satellite signal was denied, jammed, or simply gone. Sit with what that is worth, and to whom. Knowing exactly where you are when every signal you normally trust has been taken away from you is not a consumer feature. Nobody is putting it in a phone next year. The buyer is not you.

So who is it?

The race to build all of this has many horses and, honestly, no clear winner yet. Trapped ions at IonQ and Quantinuum. Superconducting circuits at Google, at IBM, at Rigetti. Photons at Xanadu and PsiQuantum. Neutral atoms at Infleqtion, QuEra, Pasqal. I am not going to tell you which one wins, because I do not know, and the people who tell you they do are selling something. The one thing that has already changed, concretely, is defensive. In 2024 the standards body finalized a new generation of encryption designed to survive the machine that does not yet exist, and the migration is underway now, worldwide, years before anything could break the old codes. The logic is chilling in its patience. Harvest now, decrypt later. Copy the secrets today, while they are still locked, and wait for the key to finish being built. The world is changing its locks against a burglar who has not been born.

Notice who is buying, in the sky and in the sub-visible both. Not consumers. Sovereign programs. Defense ministries. A handful of enormous enterprises with the patience and the balance sheet to wait out a frontier that keeps its own schedule. This is how the deepest technologies land now. Not with a launch you queue up for. With a slow, targeted arrival, into the hands of the few who can see far enough to want the thing before it exists. Built by a few. Aimed at a few. Felt, in the end, by everyone, the way weather is felt, from underneath.

And none of it, still, answers the wanting.

Come back to the pull to leave, because that is where this ends. With every cost falling and every capability rising, you would expect the one genuinely human part, the going, the body lifting clear of the ground, to be getting cheaper and closer too. It is moving the other way. In five years, fewer than a hundred human beings have flown on the leading vehicle built to sell the experience of leaving.¹⁰ Everything else about space is being automated, priced down, delivered as a service. The one thing that cannot be delivered to your door is the going itself. The oldest, most physical, least programmable want in human beings, the one that dragged the first boat off the first beach, stays almost entirely out of reach.

All the rest of it is supply. The rockets, priced by the kilogram. The satellites, sold by the subscription. The sensor that feels gravity through solid rock. The machine that is always five years away, which will be built, and will be sold. Everything else arrives as supply: intelligence, prediction, sight, distance, the whole sky rented back to you by the hour. The one thing no one has learned to manufacture is the wanting to go.

You can lower the cost of leaving all the way to nothing, and it will still take a human being, standing on the ground with a head tipped back, to decide that up is where they belong. The pull to leave, that plain conscious ache to be somewhere you are not, is the single part of this whole enterprise no one can build for you, price for you, or ship to your door.

Everything else in the sky will arrive on a signal, a manifest, a service agreement. The wanting, you still supply yourself. Which turns the long climb around to face the ground again, and the last accounting, the one no rocket and no machine can make on your behalf, still waiting to be made.