I want to begin by granting the entire optimistic argument, without objection. Its real weakness only becomes visible once you accept it.
In a CNBC interview in May 2026, Jeff Bezos painted a bright picture of the AI future. He did not treat mass unemployment as the obvious outcome, but spoke instead about productivity, deflation, cheaper food, cheaper housing and people being strengthened by more powerful tools.[1] In another widely circulated example, he used building permits as the symbol: an AI should be able to read an application and give a yes or no in ten seconds.[2]
Let us not quarrel with that picture. Let us assume it is right. More than that: let us assume we are already inside it. Work has shrunk. Many people have left wage work, several of them with relief. This is not an essay about whether AI takes the jobs. I am setting that question aside deliberately in order to ask another one, which I think is more important and which almost no one asks.
The question the scenario does not ask
A society holds together by distributing. In the economy we know, distribution happens largely through work: you contribute, you receive wages, you buy your share of what exists. Wage work is not only income. It is the whole mechanism that decides who gets what.
Remove it, and a question opens like a hole in the floor. If wages do not distribute, what does? The optimist has an answer ready, and it fits in a single word: abundance. There will be so much of everything that distribution stops being a problem. But abundance is a claim about quantity. So we have to ask: quantity of what?
Here the world divides in two, and the whole essay rests on that division. Some things are not consumed when shared: an idea, a blueprint, code, a diagnosis, a text. If I give it to you, I still have it. It is non-rivalrous. And some things are consumed when shared: a tonne of cement, a square metre of land, a litre of clean water, a kilowatt-hour. If I give it to you, I no longer have it. It is rivalrous.
AI makes the first category close to free. It does not move the second in the same way.
The physical ground beneath
Bezos's ten seconds for a building permit is a good image, better perhaps than he intended. Think about what it actually shows. The cognitive bottleneck disappears: the official review, the paper route, the wait. Gone. But the moment the lid is lifted, what lay beneath becomes visible. Steel. Sand. Water. Land. Energy. The crane operator's hours, if there is still a crane operator.
Removing one constraint is not the same as abolishing constraint. It exposes the next one. The building permit was never the house. It was the gate in front of the construction site. Behind the gate, the physical limits remain exactly where they always were.
What we call growth, and measure as GDP, is in its material core a throughput of matter and energy. AI does not make that stream infinite. It can make it smarter, faster and more efficient. But it can also remove the cognitive frictions that slowed the flow and let demand press directly against the physical ceilings, frictionlessly and around the clock.
In a society after wage work, that pressure comes from a new direction. Everyone has time. Everyone has, if the symbiosis optimists are right, a personal AI that helps them formulate what they want and how to get it. A whole population with unlimited advice and fewer limits imposed by the grind of work is not an image of subdued demand. It is an image of demand without the old brake, aimed at a planet that is still the same size as yesterday.
The abundance that does not arrive
This is where the essay's real point lands, and it is almost the opposite of pure optimism.
In a society with less wage work, the non-rivalrous really does become abundant. Everyone can receive intelligence, design, code, knowledge, entertainment and advice. It costs almost nothing to copy and can therefore reach everyone who has access to the systems.
But for that very reason, the rivalrous becomes more valuable. When everything that can be copied becomes cheap, the things that cannot be copied carry the hardest value: the square metre with a view, the field that yields food, the still water, the energy in the grid, the metal in the ground. A society after wage work does not abolish scarcity. It moves scarcity from the cognitive down into the physical, and concentrates it there.
Bezos promised cheaper food and cheaper housing. But food and housing are among the least dematerialisable things we have. Cheaper design of a house is not the same as a cheaper house when what actually costs money is the land, the materials, the energy, the grid connection and the local infrastructure. On a finite planet, facing unrestrained demand, falling cognitive cost is no guarantee of falling prices in what matters most for survival.
The core in three lines
What the optimist sees: AI tears down the cognitive bottleneck. Review, planning, administration and design become much cheaper.
What the optimist misses: beneath that bottleneck lie the physical ones. Lifting the lid does not abolish them; it reveals them.
The consequence: in a society after wage work, the copyable becomes cheap and the non-copyable becomes hard currency. Not automatic abundance, but relocated scarcity.
The irony and the honest reply
The strange thing is that Bezos, of all people, ought to see this clearly. His recurring argument for moving heavy industry into space rests on exactly this premise: Earth is finite, and sooner or later the growth curve meets a wall. He has the premise. He simply left it outside the room when the argument about cheap food and cheap housing had to hold together.
But let us not build a straw man. There is a serious reply from the optimist's side, and it deserves to be stated honestly. It says: growth does not have to be material. A mature society after wage work can grow in other dimensions: experience, relationships, knowledge, meaning, care, culture and free time. We could simply stop wanting more things and start wanting more life.
That is not a foolish answer. It is even a hopeful answer, and perhaps the only one that actually solves the problem. But notice what it is: a choice. Nothing in the market's own logic delivers it by itself. Left to itself, the market prices the scarce square metre upward and allocates it to whoever can pay most. To grow instead in the non-material is something a society must choose, consciously and partly against the current. That is very different from keeping regulation away and trusting abundance to solve the rest.
After work, politics begins again
The question in a society after work is no longer only: do I have a job? AI can partly lift that question from us, and in this scenario it does. But the hard question does not disappear with wage work. It becomes cleaner. It asks: what is there enough of, and for whom?
Technology can free us from parts of work's compulsion. It cannot free us from the finite planet. That means the distribution of the physical — land, food, energy, water, metal — can no longer hide behind the payslip. It has to become an explicit choice, made with open eyes, by a society that decides.
Bezos's future is possible. It is just not free. And it will not take care of itself.
Source notes
The essay takes Bezos's optimistic AI scenario as a starting point and asks what happens if we accept it.
- On Bezos's CNBC comments about AI, productivity, cheaper food and cheaper housing, see Mary Hanbury, Business Insider, May 20, 2026.
- On the ten-second building-permit statement and data on actual permit times, see Prevesta, March 11, 2026.