The Compounding Machine

Why he buys like the supply is finite

Everyone has a version of the regret. The neighbours who could have bought the waterfront before it was waterfront; the family that looked at empty harbour-front lots and decided the price was silly. In hindsight the call is obvious, and obvious is the one thing it never is at the time.

Bitcoin is defined-supply digital property: twenty-one million units, no more, ever, on a coast that cannot be extended. Hold that in your head and Saylor's behaviour stops looking manic and starts looking arithmetical. The downside is a price you can name today; the upside is a share of the monetary premium that currently sits in bonds, gold, and land. And whatever slice you take of a fixed thing, you hold it forever, because there is no more being made. That is not a reason to buy carefully. Against a quantity the world has not finished pricing, it is a reason to buy all you can while the deeds are still cheap.

This is the part the daily tape obscures. A normal asset rewards patience and punishes haste, because supply will be there tomorrow. A fixed asset inverts it: haste is the strategy, because the float you do not take is float someone else will. He is not gambling on a number. He is staking a claim on a coastline before the rest of the market agrees it is one.

But state the tension at the start, because everything that follows lives inside it. The earnings are paper; the dividends are cash. The gains the machine books rise and fall with a price it does not control, while the bills it has signed for arrive in real money every quarter. Read only the first half and you get the bull case; read only the second and you get the bear; the honest read holds both at once. The rest of this essay is the bull half built carefully – how the funnel, the float, and the accounting fit – with the cash-cost half running alongside it the whole way, and squared off in full at the end.

The earnings are paper;
the dividends are cash.

The rule that turned a hoard into earnings

For years the accounting rules punished exactly what Strategy was built to do. It held an enormous pile of bitcoin and was permitted to record the dips but never the recoveries – a hoard that could only ever look like a liability on paper.

Saylor did not just wait that out; he lobbied to change it, pressing the standard-setters at FASB to let the asset be marked honestly. Under the fair-value standard that followed (ASU 2023-08, in force from 2025),¹ each quarter's change in the price of bitcoin now flows straight into reported GAAP earnings, up or down, in full. What had been a vault became an engine. A static pile that the rules forced into the shadows is now a line on the income statement that prints, or bleeds, with every move in the price. Most observers filed this under "more volatility." Few have followed it to where it actually leads – a thread The Index Trap picks up on the way to the S&P 500.

The coins did not change.
What the company is allowed to report did.

Where the capital comes from, and the float that can't grow

Start with scale, because it is the whole game. The world's store of value sits in a few enormous pools – real estate near three hundred and ninety trillion dollars, bonds around a hundred and forty, equities a hundred and thirty, broad money a hundred-odd, gold sixteen. Bitcoin, against all of it, is a rounding error near two trillion.

Digital credit is a straw lowered into those pools. An instrument like STRC offers a yield to capital that is sitting in low-paying bonds and cash, or parked in assets that cost a fortune to hold, and routes the proceeds into bitcoin. Every dollar it raises buys coins and removes them from a float that cannot grow – a float already being thinned by the ETFs and corporate treasuries that hoard whatever they buy. And the float that is genuinely for sale is far smaller than the twenty-one million. A useful anchor is the bitcoin sitting on exchanges, the only coins truly on offer at any given moment – a widely-tracked on-chain figure that has hung roughly in the two-to-three-million range and trended down for years.² That, not the full supply, is the pool the funnel actually drinks from. At today's depressed price you can watch how much of that float each wave of capital swallows, and find the point where it simply runs out.

Before any model, state the logic plainly, because the argument rests on it and not on the toy below. The obvious objection is that supply is not really fixed in the way that matters: raise the price and sellers appear, so the float simply grows to meet the demand and no squeeze ever bites. That is true about sellers and wrong about the conclusion, and seeing exactly why is the whole point. In an ordinary market, a higher price calls forth more units – mines open, factories add a shift, the quantity expands until price stops rising. Bitcoin cannot answer that way: the number of coins is fixed at twenty-one million by rule, so the only thing a higher bid can summon is a change of hands, not a change of quantity. And the coins do not come in at one price. The cheapest, most willing holders sell first; what remains is held by harder hands who only let go at a higher number still. So each successive tranche of float costs more to coax out than the last – which means the act of buying in size does not expand the supply, it ratchets the clearing price upward. That is a deduction, not a forecast: it follows from a fixed quantity meeting rising demand, and it holds whatever the exact shape of the seller curve. Replacement cost – what it would take the next buyer to assemble a comparable stack – therefore rises as anyone reaches for it, rather than relenting. Hold that conclusion; the figures that follow only put numbers on a direction the logic has already fixed.

And here is the move that closes the loop back to earnings. When the absorbed demand lifts the price, the fair-value rule books the gain on Strategy's hoard straight onto the income statement, as reported GAAP profit. Today, with bitcoin trading below the company's average cost, that line is red – which is precisely why the market is panicking. Scale the funnel, and watch where it flips.

Fig. 1 – an illustrative model, not a forecast. The price response is a transparent squeeze toy that spikes as demand nears the float; the float and hoarding assumptions are yours to argue with. The point is the direction – and the line the gain runs to: the income statement.

Fig. 2 – illustrative, not a forecast. “BTC at today’s price” is just capital ÷ price – the illusion that the supply is there to buy. It is not: only a small pool is genuinely on offer, so coins-actually-acquired saturates toward that ceiling (A·C ÷ (C + P₀·A)) and everything else turns into price instead – the resulting-price line rising with the square of capital against the pool (P₀ × (1 + capital ÷ P₀A)²). This is the replacement-cost wall behind Why Price Falls When Everyone’s Buying: no one, Strategy included, can buy millions of coins at anything near spot, because the act of trying sends the price away from them. “Bitcoin for sale” is not the remaining supply but the available supply – roughly the bitcoin sitting on exchanges (on the order of ~2.4M) minus the slice that will not actually sell near today’s price. Set it to your own estimate; the scarcer it is, the more violently price, not quantity, has to move. One honest caveat: this chart and the quick calculator above run two different simplified supply models – a coarse reprice-the-float one there, a finer saturation one here – so the same capital yields different clearing prices in each. That is by design; both are illustrations of the same direction, not one true number.

Be precise about what this shows and what it does not. The price curve is a toy, a deliberately simple hyperbola that blows up as demand approaches the float; the real path would be slower and far messier, and the specific numbers it throws off should be read as illustration, never as targets. The model is an argument about direction and mechanism, not magnitude. Swap the curve for any other and the slope still points the same way, because the direction does not come from the equation – it comes from a fixed supply meeting demand for a float that thins as it is bought, which is inelastic almost by construction. Two things in the picture are therefore not toys. A fixed float, thinned by hoarding, cannot satisfy unlimited capital at a fixed price – past some point quantity stops adjusting and price must. And whatever price results is marked, every quarter, against the hoard and reported as earnings. The bill scales with the funnel; the earnings scale with the price; and the float, alone among them, cannot scale at all.

The float is fixed, and thinning.
The pools it drinks from are measured in hundreds of trillions.

You could not build this stack twice

This section is the hinge the whole bull case turns on, so it is also where the strongest objection has to be answered in full. Saylor put the scarcity in a sentence that is easy to skim past and worth stopping on. It cost roughly fifty billion dollars, he has said, to reach about three per cent of all bitcoin; the next fifty billion will not buy three per cent again – it will buy perhaps one, or less.

That is the supply squeeze felt from the inside. Every coin Strategy takes is a coin removed from everyone else's reach, and the float thins as the buying continues – the mechanism traced in Why Price Falls When Everyone's Buying. But there is a sharper consequence for how the company itself should be valued. No latecomer outside the ETFs could assemble a comparable stack without bidding the price up so violently in the act that they would pay many times Saylor's cost basis to do it. The position is, in the most literal sense, irreproducible. So the right lens for the hoard is not the spot value of the coins but their replacement cost – what it would take the next person to recreate it – and recreating it is self-defeating, because the attempt would send the price away from them as they reached.

Here the honest objection has to be met head-on, because the squeeze essays concede it openly: higher prices do summon sellers. A coin no one would part with at sixty thousand finds an owner willing to let it go at two hundred, or a million; the for-sale pool is not a fixed wall but a supply curve that rises as the price does. So does replacement cost simply dissolve once the price climbs? No – and the reconciliation is purely deductive, which is why it needs no model to carry it. Each new tranche of supply costs more to coax out than the last. The cheap, willing coins sell first; what is left is held by harder hands, who only part with theirs at a higher number still. So the buyer chasing the next million does not find them waiting at one price; he has to walk the price up to summon each successive slice, paying more for every one. The wall does not vanish when sellers appear. It moves up – which is exactly what replacement cost describes: not that the coins can never be bought, but that buying them in size forces the price toward the very level that makes the existing hoard so dear. The seller showing up at a higher price is not a refutation of the squeeze. It is the squeeze. Notice what this argument does not require: any particular equation, any specific clearing price, any number the calculator throws off. It needs only two facts that are not in dispute – that the quantity is fixed, and that willing sellers are exhausted cheapest-first – and from those two the upward ratchet follows by itself. The model below dramatises the shape; the conclusion is secure without it.

Higher prices do summon sellers –
each one dearer than the last, so the wall climbs instead of breaking.

A perpetual bid, and a multiple that changes kind

Around the hoard, Saylor has built a credit stack – perpetual preferred instruments led by STRC – by market value, among the largest preferred issues in the world – with rivals it has already spawned, like Strive's SATA, crowding in behind. Read in isolation they look like financing. Read together they are a pump.

Capital raised against the hoard buys more of the asset that backs the hoard: a standing, structural bid on bitcoin – but only for as long as that capital keeps showing up. That is the load-bearing condition, and it is conditional, not guaranteed: the bid persists while investors will buy the preferreds at a yield the company can pay, and it stalls the moment they will not. When it holds, it reframes the league table. Even if Strategy is one day out-accumulated in percentage terms, the largest collateral base lets it keep leading in absolute value, riding the unrealized uplift the whole way. And the loop can widen beyond one company: each corporation that puts bitcoin on its balance sheet lowers the barrier for the next, seasoned credit makes it routine, and price action draws fresh buyers – a flywheel that turns only while sentiment, ratings, and the bid all cooperate, and that can run in reverse if they withdraw at once.

If the earnings do swell, the very kind of valuation can shift. Today Strategy trades on a premium to its assets – a sentiment gauge. Should bitcoin become a material share of corporate profits at large, the conversation would move from premium-to-assets toward a multiple on earnings, a footing the market finds harder to wave away. That is a plausible path, not a promise: it requires sustained, positive fair-value marks, which a long winter can deny for years at a stretch. Add the generational tailwind of The Native Tongue and a standing credit bid beneath the price, and there is a reasonable mechanism by which the wild swings that scare people today moderate over time – deeper, more durable demand tends to damp volatility. But that is a tendency, not a law; leverage cuts the other way, and a forced deleveraging would amplify a drawdown rather than soften it. The honest claim is that the wheel can turn, and why – not that it must.

The unglamorous moves

A maniac would not bother with the boring parts. A steward does almost nothing else. Behind the headline buying, Saylor has been quietly engineering the thing that makes the buying survivable: the company's standing as a borrower.

He has begun working down the convertible-debt overhang rather than letting it loom, and in 2026 Strategy sold a sliver of bitcoin for the first time since 2022. Read as capitulation, it was nothing of the kind. It was proof – a demonstration that the asset on the balance sheet is genuinely liquid, not effectively frozen, which is precisely what a credit rating turns on. Selling a little to prove he could is how a holder earns the right to keep the rest. Every notch of improved rating is cheaper capital, and cheaper capital, for this company, converts directly into more coins per share. The same man who looks reckless on the buy side is, on the financing side, doing the patient, unshowy work that the Steward's Wager is named for.

The reckless part makes the headlines.
The careful part is the strategy.