Strategy used to sell business software. Then it did something stranger and more interesting: it started raising money by selling its own shares and using the cash to buy bitcoin. It now holds 847,666 of them, one of the largest stockpiles on earth, and it is still buying most weeks.
The catch is the price tag. Each share costs about $160, but the bitcoin sitting behind it is worth roughly $133. That 20% markup is the premium, and it is really what you are buying. When the premium grows, this beats owning bitcoin outright. When it shrinks, it loses to bitcoin. So the trade is not really about bitcoin. It is about the markup.
Scale is 0–50%/yr. The bear case goes negative: in a bitcoin winter the premium collapses and the stock falls faster than the coin. The base case clears the 12% bar but not the 30% one. Only the bull case, which needs bitcoin and the premium to expand together, clears both.
In 2020, with its software business stagnating and cash losing value, founder Michael Saylor started converting the corporate treasury into bitcoin. Then he kept going, funded by selling shares. Six years later the software is a footnote and the treasury is the company: 847,666 bitcoin, bought at an average near $75,000, against bitcoin around $84,500 today. The pivot worked. Now the question is what you pay for it.
Take the share price, about $160, and divide by the bitcoin value behind each share, about $133. You get 1.2x. That ratio is the whole story. Above 1.0 you are paying extra for the machinery. At 1.0 you are buying the coins at cost. Below 1.0 you are buying them at a discount. Every decision about this stock starts and ends with that number.
Here is the trick that built the stockpile. When the premium is high, Strategy sells new shares for more than the bitcoin they represent, then buys bitcoin with the proceeds. Each shareholder ends up owning slightly more bitcoin per share than before. It is genuinely clever, and it is why the coin count keeps climbing. But it only works above 1.0x. Below that, selling shares destroys value instead of creating it, so the machine stalls exactly when bitcoin is cheapest. Leverage on the way up, a treadmill on the way down.
Alongside the common stock, Strategy funds itself with preferred shares, most notably STRC, which pays about 12% a year and is engineered to trade near $100. When STRC dips below par, the company buys it back: $151.7 million in a single September week, $103.5 million of it funded by selling common shares. Retiring the preferred at a discount erases future payments cheaply, and the buying puts a floor under the price. It is the income side of the machine, and it is working. For November, the board declared STRC dividends at the same 12% rate, structured as daily payments if shareholders approve a charter change on October 28. Either way November pays the same dollar. The vote decides the structure, not the income.
The last week of September is the business model in miniature. Strategy sold about 1.4 million new shares for $246 million, spent $143 million of it on 1,665 bitcoin at $85,681 each, and used the rest plus some cash to buy back its own preferred stock. New shares in, more bitcoin per share out, cheaper obligations retired. When the premium cooperates, the loop is beautiful. Watch what happens to it when the premium does not.
At or below 1.0x you are buying bitcoin at cost, and any future premium expansion is a bonus you got for free. Above it you are pre-paying for machinery that jams in downturns. That is why this note sits on the watchlist instead of a buy stack: at 1.2x the premium is the risk, not the opportunity. The line is mechanical, not emotional. When the markup disappears, the trade gets honest.
Strategy uses the phrase "amplification ratio" two ways, and only one matters for investors. The useful one is the equity multiplier: total bitcoin divided by the net reserve behind the common stock, currently about 1.24x. It tells you how levered each dollar of the stock really is. The punchline is the relationship to the premium: when mNAV equals amplification, one dollar of the stock carries exactly one dollar of instant bitcoin exposure, which means the premium has eaten the entire leverage benefit. That is the delta-parity line, and fair value sits below it, at mNAV of 1.0. Today the two sit nearly on top of each other (1.21x premium, 1.24x amplification), so a dollar of stock buys barely a dollar of bitcoin exposure. The rule the library uses follows straight from this math: at 1.0x or below, each dollar actually carries about 24 percent more bitcoin than buying spot, before any compounding. Leverage bought at a discount is the trade. Leverage bought at a premium is just the premium.
1.2x to 1.0x is a 17% loss with bitcoin completely flat. The premium is sentiment, and sentiment mean-reverts. Anyone buying here should know that a fifth of the price can evaporate without bitcoin moving at all.
This is leverage without a margin call. In every bitcoin drawdown the premium compresses on top of the coin falling, so the stock drops harder than spot. June 2026 was the template. If you cannot handle bitcoin falling 50%, you really cannot handle this falling 70%.
Dilution is not an accident here. It is the business model: 1.4 million new shares in a single week. When the premium is high that dilution buys you more bitcoin per share. When it is not, it just buys you less of the company.
Billions in preferred stock paying around 12% are obligations that do not care what bitcoin does. In a deep drawdown, the question stops being about the premium and starts being about the credit. The $5 billion cash reserve exists for exactly this scenario, but it is finite.
The entire structure, the leverage, the preferred stack, the weekly buying, is Michael Saylor's conviction made corporate policy. There is no independent check on the cycle. That concentration built the 847,666 coins. It is also the single point of failure.
MSCI has adopted a rule that can remove companies like Strategy from its indexes for holding a lot of bitcoin and running little operating business. The final list lands November 11, and any removal takes effect at the November 30 close. The forced selling itself looks modest, about 3% of the shares, but the precedent is the risk: once one index provider decides this is not an operating company, others can follow, and the premium the whole machine runs on gets harder to defend.