Chapter Twenty-Two: The Efficiency Trap

Here is a question that sounds as though it must have an obvious answer, and does not: if a new technology made your business radically more efficient — slashed your costs to almost nothing — would it become easier or harder to pay off your debts?

Everyone says easier. Lower costs, more profit, more money to service the loan. It is the most natural answer in the world, and under the conditions now arriving, it is wrong.

Picture a small print shop that borrows a hundred thousand dollars to buy a suite of AI tools that will transform its business. The tools work exactly as promised. The shop that needed eight employees now needs two. Jobs that took a day take an hour. The owner has achieved the dream of efficiency in its purest form: he can produce far more, with far less, for a fraction of the old cost. And then he discovers the catch, which arrives from a direction he was not watching. Every other print shop in town bought the same tools. They are all now radically cheaper to run — and so they all cut their prices to win business from one another, until the price of a print job collapses toward the new, much lower cost of producing it. The owner's costs fell, yes. But his prices fell just as far, and so his revenue fell with them. His revenue is a fraction of what it was. And the one number in the whole arrangement that did not fall — that did not move by a single cent — is the hundred thousand dollars he owes. He is more efficient than he has ever been, and closer to bankruptcy than he has ever been, and the two facts are the same fact.

That catch, scaled up to an entire civilization, is the trap ahead. The world is about to take on the largest debt in its history to make itself radically more efficient — and almost no one has noticed that, under the conditions now in force, efficiency is precisely the thing that makes the debt impossible to pay.

Two Kinds of Debt

The error hidden in the obvious answer is that we treat "debt" as a single thing, when there are really two kinds, and they behave in opposite ways.

The first kind is what we might call productive debt: you borrow in order to build something new — new capacity, new goods, a new market that did not exist before. When you do this, two things grow together. The money supply grows, because the borrowing creates new money. But the quantity of real goods and services in the world grows too, because you built something. The two expand in step; the ratio between money and stuff holds; prices stay stable; and — crucially — the new thing you built throws off new revenue, and that revenue services the debt. The debt pays for itself out of the growth it created. The pie got bigger, and a slice of the bigger pie covers the loan.

The second kind is efficiency debt: you borrow in order to automate something that already exists — to make the same goods, the same services, more cheaply. And here the symmetry breaks. The money supply still grows, because you still borrowed. But no new goods enter the world; you have not built a new market, only found a cheaper way to serve the old one. The sole effect of your borrowing is that existing things now get produced for less — which means their prices fall, which means the revenue they generate falls. So the money supply grew while the prices fell, and the debt — fixed, nominal, indifferent — stays exactly where it was, while the income available to service it shrinks beneath it. Productive debt expands the pie and pays for itself out of the growth. Efficiency debt cheapens the pie and leaves the borrower owing the same sum against a smaller revenue. Same act of borrowing. Opposite consequence.

Why the Old Debt Worked

For two hundred years, almost all the debt that built the modern world was the productive kind, and that is why the world could carry so much of it without collapsing.

The transcontinental railroad was financed with mountains of debt — and it created, out of nothing, the freight markets and the passenger traffic and the settlement of an entire continental interior that did not exist before the rails were laid. The new revenue paid the bonds. Electrification, the highway system, the great industrial build-outs: borrow to build new capacity, the economy expands into the capacity, the growth services the debt. This is how industrial revolutions are financed, and it is why they can run on debt that would otherwise be terrifying. They expand the pie faster than the debt grows. And underneath all of it, making all of it work, was the same condition this book keeps uncovering: a bigger tomorrow. There were always more people coming, more demand arriving, a larger economy on the way to grow into the debt. "Borrow and build" worked for two centuries because the future was reliably larger than the present, and the larger future paid the bills the present ran up.

A Hundred-Dollar Debt in a One-Dollar Economy

Now make the efficiency trap as stark as it can be made, with a deliberately extreme thought experiment. Suppose artificial intelligence becomes so powerful that it makes the entire economy almost infinitely efficient — so efficient that the whole thing could be run, in principle, on a single dollar. The total, final triumph of efficiency. And suppose you, or your government, still owe a hundred-dollar debt, contracted back when the economy was a hundred-dollar economy.

You cannot pay a hundred-dollar debt in a one-dollar economy. There is not enough money in the entire system to settle the obligation, no matter how efficient the system has become — in fact because of how efficient it has become, since the efficiency is exactly what shrank the dollar economy from a hundred down to one. The debt was a fixed nominal claim; efficiency dissolved the nominal economy out from under it. This is the print shop again, written as large as it can be written, and it exposes the thing the obvious answer misses: when efficiency shows up as falling prices rather than rising volume, it shrinks the nominal economy, and a shrinking nominal economy cannot service a fixed nominal debt.

None of this is new to economics. A century ago, watching the Great Depression, the economist Irving Fisher described exactly this mechanism and called it debt-deflation: as prices fall, the real weight of every fixed debt rises, borrowers are crushed, they sell assets to raise cash, the selling pushes prices down further, which raises the real weight of the remaining debts further, in a downward spiral. The mechanism has been understood for a hundred years. We have simply not had to think about it, because for that entire century efficiency arrived bundled with growth, and the growth hid the trap. Strip the growth away — which is what the whole first half of this book has been about — and the trap stands in the open.

The Largest Efficiency Debt in History

Now hold the AI build-out up against this distinction, and see it for what it is.

The trillions of dollars now being borrowed and poured into artificial intelligence are not, in the main, building new markets full of new customers. There is no growing market to build for; the population that would fill it is shrinking. What the money is doing is automating existing markets — making existing work cheaper, existing services more efficient — in a world whose customer base is contracting. Which means the entire AI build-out is efficiency debt, issued at a scale that dwarfs every productive-debt project in the history of the world, into precisely the conditions that turn efficiency debt toxic: no growth to grow into, a falling population, and falling prices. We are borrowing more than anyone has ever borrowed, to make the world more efficient, at the exact moment in history when efficiency cannot pay its debts. It is the print shop's hundred-thousand-dollar loan, multiplied by the whole economy, and taken out on purpose, with great fanfare, as the smartest thing anyone has ever done.

Three Crises, One Mechanism

What this produces looks, at first, like three separate catastrophes arriving together. It is worth seeing that they are one mechanism viewed from three angles, because the unity is the whole point.

The first face is deflation. AI produces the same output, or more, with far fewer workers paid to produce it; the wage base contracts; there is less money in people's hands to chase the same quantity of goods; prices fall. And this deflation is structural rather than cyclical — its cause is a permanently smaller paid-labor base, not a temporary shortfall of demand — which means the central bank's usual cure, cheaper credit, cannot fix it. You cannot restore, by lowering interest rates, the wages that the machine has permanently eliminated.

The second face is the debt bubble. Every mortgage, student loan, corporate bond, and government bond is a fixed nominal promise, priced for an economy of higher wages and higher prices. As wages and prices fall, not one of those promises shrinks with them. The real weight of the entire mountain of debt rises — the ratio of debt to income explodes — without anyone borrowing a single additional dollar. The debt does not have to grow to become unpayable. The economy beneath it merely has to shrink.

The third face is falling demand for the dollar. The dollar's global role rests substantially on the volume of dollar-denominated world trade. AI deflates the price of nearly everything tradeable, and robots shrink the volume of trade itself by localizing production — so the dollar value of global trade falls, and a world doing less dollar business needs fewer dollars. Which means the United States must pay more to finance its debt — offering higher yields to attract lenders to a currency in slackening demand — at the precise moment that its debt is hardest to carry.

Three crises. One cause. Efficiency without growth, landing on a shrinking population, in a currency the world needs less of. And they feed one another in a loop: the automation compresses wages, which deflates prices, which raises the real debt burden, which — as confidence frays — weakens the currency, which raises the cost of borrowing, which forces yet more borrowing, which funds yet more automation. Round and round, each turn tightening the others.

Two Exits, Both Fires

At this point the well-trained reader objects, correctly: but a government that prints its own currency never has to suffer deflation at all. It can simply create money, manufacture inflation, and let the inflation melt the real value of the debt away. This is true. It is exactly what will be attempted. And it does not escape the trap; it only relocates it.

To inflate the debt away, you must print dollars — and you must print them into a world that, for every reason just given, needs fewer dollars, not more. Printing more of a thing the world increasingly wants less of is an assault on the value of the thing itself. You can indeed defeat the deflation, but only by debasing the currency: transferring the loss from the borrower who owed the debt to the saver who holds the money and to the standing of the dollar in the world. So the choice is not between crisis and safety. It is between two crises. Let the debt stand, and deflation crushes the economy beneath it. Print to escape the debt, and you burn the currency to do it. The efficiency-debt trap does not have a clean exit. It has two exits, and both of them are fires — and the door marked "just cheapen the debt," as we will see, leads somewhere even worse than it looks.

So the efficient economy, the indebted economy, the deflationary economy, and the weakening-currency economy are not four different problems that happen to be arriving at once. They are a single object seen from four sides: the largest efficiency-debt issuance in history, landing on a shrinking population, in a deflating world, in a currency the world is beginning to need less of.

All of it rests, finally, on a misunderstanding of what debt actually is — a misunderstanding so basic that we have stopped being able to see it. We think of a debt as something the future has a claim on: a burden we have placed on tomorrow. It is the reverse. A debt is a claim we place on the future — a bet, struck today, that tomorrow will be larger and richer than today, and will therefore be able to pay what today has promised on its behalf. Every dollar ever borrowed is a deal with a future that has to show up bigger than the present in order to honor it. For two hundred years, the future always did. It is about to start showing up smaller.

What it means that the entire architecture of modern finance is a deal with a future that is no longer coming is where we turn next.