Strategy Inc MSTR

Key data

The business

Strategy is two companies stapled together, and the market pays for only one. The first is an old enterprise-analytics software maker, the former MicroStrategy: dashboards and business-intelligence tools sold to large firms, now wrapped in AI features called Strategy One and Mosaic. It earns about $477 million a year at a 68% gross margin with roughly 1,540 employees, and revenue has drifted down from $511 million in 2021. A steady, unexciting, slightly shrinking software business. The second is a bitcoin holding funded by relentlessly issuing new stock, convertible notes, and preferred shares, then spending the proceeds on coins. At a $47 billion market value, the software is about one percent of the story. Whatever moat exists is not in the software. It is the ability to sell securities to the public at more than the bitcoin behind them is worth, then buy more bitcoin with the difference. That is not a durable moat. It is a market mood, and the numbers below show the mood cooling.

The numbers

Two tables and one point. The revenue line barely moves. The profit line is bitcoin's mood swings wearing an income statement.

Read the quarters and you see it plainly. Revenue held between $114 and $129 million across the last year, up 6.9% over the year in the June quarter and down 1.6% from March, a small software business doing small-business things. Beside it the profit line went from a $10.0 billion gain in June 2025 to a $12.8 billion loss in March 2026 to an $8.6 billion loss in June 2026, none of it earned, all of it the marked value of the coins rising and falling under fair-value accounting. Operating cash flow ran slightly negative over the trailing year.

The five-year record says the same. Revenue is lower than it was in 2021. What compounded instead was the share count, up about 160% in three years and now near 343 million, because selling stock is how the coins get bought. When the stock traded at two or three times the value of its bitcoin, each sale added coins per share, and that was the machine. My read the market may not share: at 1.08 times book the machine is close to stalled, and this is now, within a rounding error, a levered bitcoin proxy carrying a preferred-dividend bill rather than a compounding accumulation engine. The single number that settles it is the premium, price-to-book holding above one or slipping through it.

Management

Read the record, not the biographies. Insiders have been net sellers, 48 open-market sales against three buys over the past year, about $18 million sold and $1 million bought, with CEO Phong Le's June 2026 sales the largest single names. The dollar amounts are tiny against a $47 billion company, roughly four hundredths of a percent, and I cannot cleanly split them into pre-scheduled and discretionary from what is filed, so I will not pretend the signal is loud. The louder fact is dilution, shares outstanding up about 160% in three years, by design, to fund coin purchases. Michael Saylor, the architect, draws a one-dollar salary and $781,000 in total pay. Le made $13.8 million in 2025, mostly stock. There are no buybacks and no common dividend. There is now a preferred dividend, which is the seam the next section pulls on.

How it fails or surprises you

The premium inverts. At 1.08 times the marked value of its coins, Strategy is one bad stretch from trading below them. If price-to-book crosses one and stays there, the flywheel runs backward: every new share sold then subtracts coins per share instead of adding them, and the reason to own this rather than bitcoin directly disappears. The print that reveals it first is price-to-book through 1.0x.

Bitcoin runs and the premium re-expands (right tail). This is a levered coin, beta near 3.6. If bitcoin rallies and the accumulation story reignites, the premium can re-rate from today's 1.08 back toward the 2x-plus it held for years, and the stock outruns bitcoin itself on both the coin and the premium. The print is book value climbing while the premium widens on the same day.

The preferred bill. The gap between net income and net income to common, roughly $400 million in the June quarter, is preferred dividends. The software throws off almost no cash, so those dividends are paid by issuing more securities or, eventually, by selling coins. If the coin count ever falls to fund them, the whole thesis is finished. This is the fact my read explains least, and the one that would prove it wrong.

Closing thoughts

The payoff is two exposures layered on each other, and only one of them a print can resolve. Underneath is bitcoin, which no filing settles and only survivability answers. On top is the premium, which one number, price-to-book, resolves every day, and it now sits at 1.08, a hair above the line where owning the wrapper stops making sense against owning the coin. The fatter tail from here is the downside, because the premium has one cheap direction left to travel and the preferred bill compounds against you while you wait. What is at risk is the difference between the stock and the coins it holds, which used to be large and is now thin. The upside, if bitcoin runs and the premium re-expands, is real and larger in percentage terms, but it needs two things to go right where the downside needs only one to go wrong.

The bet is still that Strategy keeps raising money to buy bitcoin, and that the coins it holds stay worth more than what the stock costs you. Since the last look the machine held: the coin count did not fall and the share count kept climbing, so the preferred dividends are still being paid with new paper rather than with the asset. What breaks it is the premium going away, and the one pair of numbers that tells you first is price-to-book against 1.0 and the coin count against last quarter's. The day both turn, this is bitcoin minus a dividend, not bitcoin plus a machine.