The capital cycle as agon

The Greeks had a word for the contest that made excellence possible: the agon. Not war, which destroys the opponent, and not commerce, which merely serves him — but the struggle whose purpose is to reveal rank. The footrace exists so that the fastest can be known. The Greeks understood something we have half-forgotten: that excellence is not declared. It is extracted, by contest, from a field of claimants who all believed themselves excellent until the race said otherwise.

The capital cycle is the agon of business. Most investors treat it as weather — an external misfortune that interrupts the compounding from time to time, to be endured with a grimace and a rebalance. This is a misreading of what the cycle is for. The cycle is not an interruption of the selection. The cycle is the selection.

The genealogy of a bust

Trace any bust to its origin and you will not find pessimism. You will find enthusiasm — specifically, the moment returns became good enough to advertise. High returns summon capital; capital builds capacity; capacity competes; competition destroys the returns that summoned it. The mechanism is so reliable that it is close to an accounting identity, and yet each generation rediscovers it with fresh astonishment, because each generation prefers the demand story. Demand is the alibi. Everyone watches whether the customers will come. Supply is the murderer — the question is rarely only whether the appetite is real, but how many kitchens are being built to serve it, and the kitchens are announced years before the returns collapse, in public filings, at groundbreaking ceremonies, with applause.

Capital destroys returns by loving them. That is the whole genealogy. The bust does not arrive from outside the boom; it is manufactured by the boom, ordered at the top, delivered at the bottom.

What abundance selects for

Here is the part that interests us most, because it is about character. Watch what a flood of cheap capital does to the companies standing in it.

Discipline is a response to scarcity. When capital is expensive, projects must argue for their lives; when it is free, everything green-lights, and an organization slowly loses the ability to say no — the one ability that mattered. The boom selects for spenders. It promotes the executives who build fastest, rewards the companies that promise the most capacity, and marks down the holdouts who insist on returns before expansion. For a few years, profligacy outperforms, and everyone can see it outperforming. This is decadence in the precise sense: not luxury, but the decay of a discipline that abundance has made temporarily unnecessary.

Then the flood recedes, and the selection reverses in a season. The spenders stand exposed among their half-built capacity, financed at yesterday’s prices for yesterday’s demand curve. The disciplined — recognizable all along by a single refusal, to expand faster than returns could justify; capex kept short of the applause, capital handed back rather than sunk into capacity the demand curve had not yet earned — inherit the field. Their competitors’ capacity is now for sale at distress prices. Their pricing power returns as the industry consolidates around the survivors. The memory industry is the textbook: dozens of claimants entered its successive contests; Samsung, SK Hynix, and Micron walked out of the last cull — and the three now earn, in the good years, what the dozens never could.¹

The bust is not the disease. The boom is. The bust is the cure — administered, as cures often are, without anesthetic.

Loving the cycle

The conventional posture toward all this is dread: hedge the cycle, dampen it, wish it away. We hold the opposite view, and it is worth stating plainly because it governs how we invest. The cycle is doing, at industrial scale, the exact work we most need done and could never do ourselves: stress-testing every claim of excellence with real capital and real pain, and publishing the results. A moat asserted in a pitch deck is an opinion. A moat that has survived a capacity flood, priced its way through the cull, and come out with its reinvestment discipline intact — that is a fact, established by contest.

So we do not resent the cycle. What we own, we own because some contest already revealed it; what we hope to own next, the current contest is busy revealing. The investor’s amor fati is not resignation. It is the recognition that the thing you dreaded is the thing doing your work.

Where to sit

This settles how a portfolio should sit at any flood — and it is worth being blunt about the seats, because they are not equally safe.

The suppliers’ seat is where the flood lands as revenue, and that is exactly what makes it the dangerous one. Today’s magnificent margins at the sellers of capacity are the advertisement that summons the competing kitchens; the supply response is aimed directly at them. The seat can be taken, but only as a trade on the flood, never a claim on the far side of it — an exit written before entry, or not entered at all.

The builders’ seat is the interesting one, and it divides in two. History’s builders were debt-funded single-purpose vehicles: when the buildout disappointed, there was nothing underneath, and the bondholders inherited the track. But a builder who funds the flood from the operating cash of a business that would be excellent anyway — whose tollbooths stood before the flood and stand whether or not it pays — is a different creature entirely: the one entrant who can lose the contest’s wager and still keep the estate.

The toll-keeper’s seat sits outside the contest altogether — the businesses that tax commerce itself, indifferent to which builder wins. They do not race; they own the road the racers must use.

And then a posture that is not a seat at all: the reserve held for the receding water. The single best moment in the whole cycle is the one nobody enjoys — the day excellence goes on sale because its neighbors are drowning, and buying it demands acting precisely when acting feels worst. The dry powder is not caution. It is a paid-up seat at the end of the contest.

The seat we sit in

Honesty requires one more admission, and it cuts toward us. The seat that looks safest — the cash-funded builder who can be wrong and keep the estate — is for that very reason the seat most likely to be over-owned, and the crowding is the risk the comfort hides. A single premise, that a handful of exceptional franchises stay exceptional while they spend, can fund a whole row of builders at once; and many builders resting on one premise is not diversification but a single wager wearing its costume. The agon extends no special mercy to the well-financed. We keep written criteria for the ways the wager fails because the seat we most admire is the one where we are most exposed — and a discipline that exempts its own favorite holding is decoration, not discipline.

The present flood

We are watching the largest capital flood of the modern era² pour into a single ambition: the infrastructure of intelligence itself. The sums have stopped meaning anything; the groundbreaking ceremonies run weekly. And the map of the seats is unusually legible, if one insists on reading it. The advertised returns — the margins that summon the kitchens — sit today with the sellers of capacity: the chipmakers above all, and behind them the memory, power, and cooling suppliers whose order books are the boom made visible. The genealogy has one message for that seat: this is where the supply response is aimed.

The builders pouring the capital — Microsoft, Alphabet, Amazon, Meta — look, at first glance, like the old railway financiers, and getting this comparison right is the single most important thing about the cycle. The financiers were leveraged vehicles with no second act. These builders fund the flood from the operating cash of search, commerce, advertising, and cloud franchises that were exceptional before the first datacenter of this wave broke ground and will remain exceptional if the wager under-delivers. They can be wrong about the buildout’s timing and still win the decade — a courtesy history never extended to the men who built the track. Whether the wager pays as intended is a live question, and we hold written criteria for the ways it fails.

But most of this capital — across the whole contest — will earn nothing, and that sentence should be read calmly, because it is not a prophecy of doom. It is a description of tuition. The railway busts of the nineteenth century bankrupted their builders and remade a continent;³ the fiber glut of 2000 ruined its financiers and handed everyone who came after free bandwidth to build on.⁴ The contest will incinerate many of its entrants and hand what they built to the survivors — and to their customers, at prices the builders never intended.

We do not claim to know the date of the receding. Nobody rings a bell, and forecasting it has ruined better readers than us. We claim only the two disciplines the agon actually rewards: a list, kept current, of who deserves to survive — written before the cull, when it can still be written honestly — and the reserves to act on the day the list goes on sale.

Every entrant believes itself the exception; that is why the contest is necessary. The flood does not choose the winners — it reveals them, and it is merciless with the ones who were sure. Our work is to be ready when it does.

Love the cycle. It is doing the selecting for you.

Related reading