Never selling was never the legal position
Michael Saylor's public persona was built on never selling a single satoshi. The company's filings have said something more careful for years: Strategy's software business does not generate enough cash to cover its obligations, and the risk factors in its quarterly report for the first quarter spell out that a deep enough Bitcoin drawdown could force sales to satisfy financial obligations. What changed at the second-quarter call was the framing, not the facts: president and chief executive Phong Le described selling Bitcoin as one more trade the company will make whenever it is advantageous (to fund the USD reserve, pay preferred dividends and interest, or buy back its own securities) and told investors to expect it on a go-forward basis.
The sales so far are small and tax-shaped
The record to date reads as housekeeping, not liquidation. Strategy sold 32 Bitcoin in May at a realized loss of about $1 million, 3,588 in July at an average near $60,000 for a $203 million realized loss, and another 1,638 in the week to 2 August at an average near $63,957: leaving 842,138 Bitcoin, still the largest institutional stack in the world. Selling at a loss is the point, not an embarrassment: under US specific-identification accounting the company can choose its highest-cost lots, and with roughly $18.5 billion of unrealized losses across the stack it values the potential tax benefit at up to $5.4 billion. Meanwhile it bought 174,895 Bitcoin this year: dozens of times more than it sold.
The arithmetic the narrative skips
The fear is a treasury bleeding itself out; the numbers describe something smaller. Strategy's annual preferred dividends and interest run about $1.76 billion, against a Bitcoin position recently valued near $52.6 billion, roughly 3.3% of the stack per year if every dollar of obligations were funded by sales alone. Put differently, if Bitcoin appreciates by about 3.3% a year, obligation-funding sales leave the dollar value of the reserve intact; anything above that and the stack still grows in value while paying its own bills. And sales are not the plan of first resort: the framework routes new digital-credit issuance and the USD reserve ahead of the stack, which is why the company remained a large net buyer through a year in which it began to sell.
What Signal21 takes from it
A seller of last resort disclosed in filings has become a seller of first convenience in small, tax-aware doses, and the market's muted reaction to each disclosure suggests it can tell the difference between managing a balance sheet and abandoning one. Our views on the equity are unchanged: the short-term view stays Neutral Range on a soft tape, the medium-term view remains Bearish Bias while accretive issuance is stalled, and the long-term view stays Strong Bullish on the largest corporate Bitcoin treasury refinancing itself onto perpetual capital. This is market analysis, not a recommendation to buy or sell any security.