The equation looks clean. Bitcoin appreciates twenty percent. Strategy's preferred stock pays twelve. The spread β eight points β is profit. Michael Saylor said as much. On paper, the trade is arithmetic.
I ran the arithmetic through a spreadsheet last week. It broke in eleven places.
Here is the first break. A twenty percent gain on a Bitcoin position is an unrealized gain. A twelve percent dividend is a realized obligation. One is a number on a screen. The other is a wire transfer. The gap between them is not eight points of yield. It is a liquidity mismatch that widens every quarter the price stops rising. The dividend is paid in cash or dilution. The gain is paid in confidence. When confidence is the collateral, you are not running a treasury. You are running a duration mismatch with a marketing department.
That is the whole story of Strategy in one sentence. Everything after this is bookkeeping.
Strategy β the company formerly known as MicroStrategy β holds more Bitcoin than any other public corporation. That is the fact. The mechanism behind the fact is what the market refuses to price. The company does not generate Bitcoin. It does not mine it, stake it, or lend it at a spread. It buys it, and it buys it with other people's money.
The funding stack has three layers. Convertible bonds, which are debt until they are equity, priced on the assumption that the conversion option will always finish in the money. Preferred stock β STRK, STRF, STRD and their siblings β which pay fixed dividends with seniority over common shareholders. And at-the-market equity issuance, which sells new common shares into the secondary market whenever the premium over net asset value is wide enough to justify the dilution.
Each layer is a standard capital markets instrument. None of them is exotic. I have audited convertible structures. I have modeled preferred waterfalls. I do not need to trust the prospectus; I can read the liquidation preference and I can rank the claims.
The instruments are clean. The strategy is not.
What ties the three layers together is a single variable: mNAV, the market-to-net-asset-value ratio. When mNAV exceeds one, Strategy can issue shares at a premium, convert the proceeds into Bitcoin, and raise both the asset base and the share price. The flywheel spins. When mNAV falls below one, the same machine runs backward. Issuing shares now destroys value. The company stops buying and starts defending.
So the entire model is a bet on a premium that exists only as long as the market believes the model works. That is not a business. That is a reflexive loop with a bank account attached. The code compiles, but the reality bankrupts.
Let me dissect the flywheel the way I dissected TerraUSD in 2022. Same method. Strip the narrative. Keep the cash flows.
Step one. Strategy issues a convertible bond at a low coupon because the conversion option is valuable to the buyer. Step two. It uses the proceeds to buy Bitcoin. Step three. Bitcoin rises, the NAV rises, the stock rises, and the premium expands. Step four. The premium allows more issuance at a better price. Step five. Return to step one, larger.
This is a positive feedback loop. Positive feedback loops do not have equilibria. They have thresholds. Below the threshold, they expand. Above it, they invert.
The seigniorage model of UST had the same topology. Demand for LUNA was the input. Minted UST was the output. The peg held as long as the input grew. I calculated that the required demand was geometrically impossible to sustain without infinite liquidity. I wrote forty pages on it and sent them to regulators in Singapore. The market ignored them for eight months. Then the loop inverted and took ninety billion dollars with it.
Strategy is not Terra. I want to be precise about that, because the lazy comparison is wrong and lazy comparisons are how people lose money. Terra had no asset. Strategy has one. That difference matters, and I will get to it. But the loop topology is the same, and loops do not care what the asset is. They care whether the input keeps arriving.
The input for Strategy is not Bitcoin's price. It is the premium. Bitcoin's price is the fuel. The premium is the engine. Kill the premium and the fuel does nothing.
Now the dividend. Twelve percent.
Where does twelve percent come from? A company does not choose a dividend yield. The market prices it. A twelve percent yield on preferred stock means the marginal buyer of that paper demands twelve percent to hold the risk. Compare that to the risk-free rate. Compare it to investment-grade credit. Twelve percent is not a generous gesture by a confident issuer. Twelve percent is a risk premium wearing a dividend's clothes. Illusion has a price tag; truth has none.
The historical software business generated roughly five hundred million dollars a year in revenue at its peak. Revenue, not free cash flow. That number is an order of magnitude short of what a twelve percent obligation on a multi-billion-dollar preferred stack requires. The dividend is not funded by operations. It is funded by the next raise.
I have seen this cash-flow signature before. In 2020 I spent three weeks simulating Uniswap v2 pool dynamics in Python, hunting for the slippage threshold that would wipe out retail liquidity providers. The pools looked profitable because the fee income was visible and the impermanent loss was invisible. The visible number paid the marketing. The invisible number paid the bill. Strategy's dividend is the fee income. The mark-to-market dependency is the impermanent loss.
Here is the stress test. Assume Bitcoin flatlines for four quarters. Not a crash. Just a flat line.
The twenty percent gain does not arrive. The twelve percent dividend still does. The spread is now negative twelve, not positive eight. Strategy pays it with cash, which is finite, or with stock, which dilutes. If it pays with stock, the share count rises, the per-share Bitcoin backing falls, the premium compresses, and the flywheel slows. If it pays with cash, the cash is gone and the next dividend has no source. Either path feeds the same loop.
Now assume Bitcoin drops thirty percent. The gain is gone and the NAV is impaired. The preferred dividend is still senior. Common shareholders absorb the hit first, which is exactly what the capital structure was built to do. The preferred holders are fine β until they are not, because a twelve percent coupon on an impaired NAV is a claim the company may not be able to meet without selling Bitcoin. Selling Bitcoin into weakness to pay a dividend is the definition of a forced seller. Forced sellers set the price. The transaction is permanent; the mistake is not.
I mapped the mNAV premium against the Bitcoin holding last quarter. The premium is the whole valuation argument. Strip it out and you are left with a company trading near the value of its coins plus a declining software business. The premium is not a rounding error. It is the majority of the market cap. When a valuation is majority-narrative, the narrative is the balance sheet.
There is one more layer the headline hides. The three funding instruments do not share the same incentives. Convertible holders want equity upside and will convert if the stock runs. Preferred holders want the coupon and will not convert; they want to be paid. Common holders want Bitcoin to appreciate and will tolerate dilution to get it. These are three different claims on the same pool of coins, ranked in a fixed order. In a rising market the ranking is invisible because everyone is whole. In a falling market the ranking is the only thing that matters.
This is a governance problem disguised as a yield product.
Now the part the bears get wrong, because I do not write to flatter a side.
The bulls are right about the asset. Bitcoin is real. It is liquid. It is auditable on a public ledger. Strategy's holdings can be verified against on-chain addresses, which is more than can be said for most treasury narratives I have torn apart. When I dissected the metadata of a ten-thousand-item PFP collection in 2021 and showed that eighty-five percent of the "rare" traits came from a flawed random seed, the whole thing was a lie wrapped in a hash. Strategy is not that. The coins exist. The custody is disclosed. The addresses are watchable.
That matters, and it is the reason I do not call this a Ponzi. A Ponzi promises returns from nothing. Strategy promises returns from an asset that genuinely appreciates in the right regime. The distinction is real. The bulls who say "at least there is a hard asset underneath" are correct, and anyone who dismisses that is doing lazy analysis.
The bulls are also right that the model is transparent. Every issuance is an SEC filing. Every conversion is disclosed. You can track the dilution in real time. Compare that to the opaque offshore structures I have spent my career dismantling. Strategy runs in daylight.
But here is where the bull case quietly changes the subject. "There is a real asset underneath" answers the question "is this fraud?" It does not answer the question "is this solvent under stress?" A real asset leveraged and pledged against a fixed obligation is still a leveraged position. The asset being real does not make the leverage safe. It makes the leverage honest. Honest leverage still liquidates.
The bulls are right about the asset and wrong about the structure. Those are two different claims, and the market keeps collapsing them into one.
So watch three numbers, not the price.
Watch mNAV. When it goes below one, the engine stalls and the dividend becomes a liability instead of a cost of capital.
Watch the payment method. The day Strategy pays a preferred dividend in stock instead of cash is the day the liquidity mismatch stops being theoretical. That is the signal, and it will be disclosed, and it will be late.
Watch the ETF flow. Every dollar into a spot Bitcoin ETF is a dollar that does not need Strategy's leverage to get Bitcoin exposure. The proxy loses value as the direct route gets cheaper. The premium is a convenience fee, and the convenience is disappearing.
The arithmetic twenty-over-twelve is not a thesis. It is a snapshot of one good quarter dressed as a business model. I do not trust the audit; I trust the exploit. The exploit here is the duration gap between a mark-to-market gain and a cash coupon. It has not been triggered yet. That is the only reason the equation still looks clean.