Compounding is arbitrage, not optimization.
— The Infinite Investor
Download the Slides Deck: Compounding_Components
Listen to the deep-dive discussion – Why Compounding Is A Structural Arbitrage
I. The Perfect Model
Every business is a flow of capital between two reservoirs. The first is the toll — the cash the business collects in the present, short in duration, harvested now. The second is the wave — the growth that builds over time, long in duration, realized later. The two together give every business a dual duration, and the direction and intensity with which the toll takes in the growth place the business somewhere on a single spectrum: from the exhausted toll that can absorb almost nothing of what it earns, and overflows its cash to its owners, to the business whose growth consumes more than the toll can bear. This is the dynamic set out in Capital Absorption and the Nature of Compounding, and it is the ground on which what follows is built.
From that dynamic, three components emerge — and it is their mutual tension, not the mechanism of absorption itself, that governs how richly a business compounds. What the toll absorbs, it reinvests, and reinvestment has three measures: the rate at which the reinvested capital earns its return, the quantity of capital the business can reinvest at that rate, and the duration over which it can keep doing so. These three are not independent dials a business sets at will. They are bound to one another by its very structure, so that gaining on one is paid for on another — and no business holds all three at their fullest at once. The tension is not the investor’s to resolve; it is the company’s own, inherent in what the business is. To buy the business is to buy that arbitrage, exactly as it stands.
Begin with what compounding would look like in its ideal form. A business earns a return on the capital it employs; it reinvests that capital back into itself; and it does so, year after year, for a long time. Three components govern the value this produces. The first is the rate of return earned on the capital that is reinvested. The second is the quantity of capital the business can reinvest at that rate. The third is the duration over which it can keep doing so.
The ideal is the union of the three at their fullest: a high return, earned on a large and growing quantity of reinvested capital, sustained for a very long time. A business that reinvested everything it earned, internally, at a high rate, for a century, would be the perfect compounder. It would never hand capital back to its owners, because it would always have somewhere better to put it — inside itself, earning more than they could earn elsewhere. This is the model against which every real business is measured. It is worth stating plainly, because it is the reference point for everything that follows: the closer a business comes to holding all three components together, the closer it comes to the ideal.
II. Why the Three Components Do Not Hold Together
The difficulty is that the three components are not independent. Improving one tends to come at the expense of another, and the tensions are structural, not accidental.
A high return on capital and a large quantity of reinvestable capital pull against each other. A high return usually comes from being capital-light — from a business that needs little capital to operate or to grow, so that what it earns is large relative to what it employs. But a business that needs little capital has little capital it can productively reinvest. The very lightness that produces the high return limits the quantity. Conversely, a business able to absorb large quantities of capital tends to do so at more moderate returns, because deploying capital in size usually means deploying it into assets and operations that earn less than a capital-light franchise does.
A high return and a long duration also pull against each other. A return above the cost of capital attracts competition, and competition erodes it. To sustain a high return for a long time requires a barrier that keeps competitors out — and barriers, however durable, are finite. The higher the return, the stronger the incentive to attack it, and the harder it is to defend across decades.
Quantity and duration are the most compatible of the three, but their compatibility is bought with return. A business can reinvest large quantities of capital for a long time — but typically at a modest rate, because the assets that absorb capital in size and hold their position for decades are rarely the assets that earn the highest returns. One can hold two of the three components at a high level; holding all three, at a high level, for a long time, is what almost no business manages. The perfect model remains a model.
III. The Heavy Toll: The Closest Approach to the Ideal
The business that comes nearest to the ideal is the one that refuses to hand its capital back and instead keeps finding places to put it — the capital-heavy compounder, or heavy toll. A railway, a regulated utility, a large-scale infrastructure business: these employ enormous quantities of capital, and, crucially, they can keep employing more. Their base widens as the world they serve grows. They hold two of the three components at their fullest — a large and growing quantity of reinvested capital, sustained over a very long duration — and they pay for it only in the third, accepting a return that is respectable rather than exceptional.
What allows the heavy toll to hold quantity and duration together is a single property: its capital intensity is also its moat. What costs a great deal to build costs a great deal to replicate. The capital required to lay a second rail network, or to duplicate a grid, is the same capital requirement that keeps competitors out. The lightness that produces high returns elsewhere invites competition precisely because entry is cheap; the heavy toll’s weight is what protects its position across decades. Quantity and duration, in the heavy toll, are produced by the same feature.
The essential condition, however, is that the capital be absorbed at a return that stays above its cost — and that condition depends on there being growth to reinvest into. Call this growth the wave: the secular expansion of the activity the toll serves. As long as the wave runs, the heavy toll has somewhere productive to put its capital, and its base widens with the world. A business that generates ever-larger amounts of cash, and can keep absorbing that cash into a widening base at a maintained return, is the real-world approach to the ideal. This is the direction a great compounder naturally takes: rather than hand its growing cash back to owners, it builds itself a larger base into which that cash can be absorbed.
The contemporary form of this is not only physical. A business that generates cash faster than its existing operations can absorb it faces a choice: return the surplus, or build a new base capable of absorbing it. Alphabet, constructing cloud and artificial-intelligence infrastructure, is doing precisely what the railway did — building a heavy, capital-intensive base, protected by the scale required to replicate it, into which a massive and growing cash flow can be reinvested. The substrate is new; the logic is the one the railway established. The business gives itself a place to keep absorbing capital, and so keeps compounding rather than overflowing. That this new base will absorb capital productively — at a return above its cost, rather than destroying it — is not yet settled; it is the open question the market prices when it weighs the scale of the spending. It is worth noting that Berkshire Hathaway, whose entire discipline is the refusal to deploy capital where absorption is not productive, has taken a position in Alphabet. The judgment of the most demanding allocator that this costly new wave will be absorbed rather than wasted is not proof of the outcome, but it is the opinion of the party whose business is precisely to tell the difference.
The most complete version of this structure industrializes both components at once — the source of the capital and the capacity to absorb it. Berkshire Hathaway is the clearest instance. It owns regulated infrastructure outright: BNSF, a freight railway, and Berkshire Hathaway Energy, an energy and utilities operation. These are bases that ingest capital almost without limit, because within a regulated framework the return is earned on the asset base itself — so that every dollar of capital absorbed enlarges the base on which the return is earned. Paired with these absorption machines is a source of capital at little or no cost: the float generated by Berkshire’s insurance operations, which supplies capital ahead of the claims it will one day pay. The two halves of the ideal are joined — a large and growing supply of capital, and regulated bases able to absorb it at a maintained return for decades. Such a structure need not pay its capital back, because it has, internally, somewhere productive to put it; the recent accumulation of cash and the turn to repurchases mark the point at which even this capacity meets its limit — the largest absorber reaching the size at which its own base can no longer take in all it generates.
IV. The Wave as the Condition
The heavy toll is the approach to the ideal only while the wave runs. Remove the wave, and the same weight that was a virtue becomes a burden. A capital-intensive base serving a stagnant volume no longer absorbs capital productively; it merely immobilizes it, at a return that drifts toward the cost of capital, while still demanding the maintenance and renewal that heavy assets require. Capital intensity, so valuable when there is growth to absorb, is a liability when there is not.
The point can be seen in a decision often explained for the wrong reason. The Coca-Cola Company, having consolidated the North American operations of its largest bottler, Coca-Cola Enterprises, in 2010, reversed course between 2013 and 2018 and refranchised that bottling — handing the capital-intensive operations to independent bottlers and keeping the high-return concentrate business for itself. The move is usually described as a shift to an asset-light, high-return model, and as a choice that is what it was: a deliberate decision to create value by concentrating capital on the activity that earned the most — concentrate carried operating margins above thirty percent, bottling only a few. It was taken without certain knowledge of whether or when the underlying wave of volume growth would fade; it was, in effect, a return to the licensing model on which the company had been built more than a century earlier. The bottlers, well run, executed brilliantly, and for a period captured strong returns of their own. The decision stands on its own merits as value creation, judged by what could be known when it was made — not by hindsight.
What later became clear is a structural asymmetry the model exposes. As the wave of volume growth flattened in mature markets, both the light concentrate business and the heavy bottling business lost their capacity to compound by absorption — both were left returning cash rather than reinvesting it internally at a high rate. But the heavy business suffered more. A light business without a wave overflows cleanly: it hands its cash back without having to finance a heavy base serving stagnant volume. A heavy business without a wave keeps committing capital to maintaining assets that no longer absorb it productively. The contrast with PepsiCo, which retained more of its bottling and carried lower operating margins as a result, shows the same asymmetry from the other side. Both the light and the heavy suffer when the wave recedes; the capital-intensive one suffers more. The wave, then, is what determines whether the heavy toll’s weight is an asset or a liability — and a business that has lost its wave, heavy or light, drifts toward a profile dominated by the return of cash, reinvesting little internally while a modest residual growth continues in the background.
V. Duration as the Practical Tie-Breaker
If the three components cannot be maximized together, one of them nevertheless carries more weight than the others, and duration is that component. The reason lies in how compounding accumulates. Return and quantity determine how much each period adds; duration determines how many periods compound upon one another. Because each period builds on the accumulated result of all the periods before it, extending the duration influences the final outcome more powerfully, over a long horizon, than an equivalent improvement in the rate — provided the return stays positive and above the cost of capital throughout. A moderate return sustained for a very long time outruns a high return sustained briefly.
This is why a long-horizon test can cut through an otherwise complex, multi-variable comparison. Faced with three components that cannot be jointly optimized, the question of whether a business will still be here, and still compounding reasonably, in a hundred years collapses the problem to a single criterion. It sets aside the pursuit of the highest rate — which cannot be identified in advance with certainty and does not endure — in favor of what can be captured and held: durable, reasonable compounding. The hundred-year question is not a preference for longevity for its own sake; it is a way of choosing the component that the structure of compounding rewards most, under the one condition that the return be maintained.
A qualification belongs here, because the durations at issue must not be confused. The duration that matters for compounding is the duration of internal compounding — the length of time a business can reinvest capital productively — not the longevity of the enterprise. A business can endure for a century as an institution while its capacity to compound internally has already ended. Such a business is long-lived but, in the sense that matters here, short in duration: it can no longer make reinvested capital work, so it returns its cash, and its value reaches its owners through distribution rather than internal growth. Survival and the duration of compounding are different measures, and it is the second that the criterion selects.
VI. Conclusion
Compounding is governed by three components — the return earned on reinvested capital, the quantity of capital that can be reinvested, and the duration over which it can be done — and they cannot all be held at their fullest at once. The perfect model, in which the three unite, is a reference point rather than a destination. The heavy toll is the real-world approach to it, holding quantity and duration together at the price of return, for as long as a wave gives its capital somewhere productive to go; when the wave recedes, its weight turns from asset to burden. Duration is the component the structure of compounding rewards most, and the hundred-year question is the practical way to identify it — under the one condition that the return be maintained. A business does not choose freely among the three; its nature sets where it stands, and what it can hold of one, it surrenders of another. The arbitrage is the company’s own, written into what it is. To own the business is to own that arbitrage, whole and as it stands.
Watch the video
This content is educational and reflects a personal analytical framework. It does not constitute investment advice.