Strategy resumes buying on a cash pile equal to half its unrealised bitcoin gain

A preferred dividend is a dollar obligation and bitcoin generates no cash, so servicing the preferred stock Strategy issued earlier this year has to draw on one of three taps: new equity, new borrowing, or the stack itself. Cryptobriefing reports the company used two of them during the pause, raising capital through equity issuances and selling BTC to cover operating costs including those dividend payments ^[4]. The July sale of roughly 3,588 coins ^[12] is about 51% of the roughly 7,000 sold across the entire year ^[16], which says the selling was not a slow trickle but a payment date landing inside one month.

Now size it. At the $79,000 mark the publisher used ^[14], 3,588 coins is somewhere near $283m ^[17] (rough, since we are not told what July prices were), and the $1.59bn reserve assembled during the pause ^[3] covers a bit under six repeats of that outflow ^[18]. Or rather, the version that says more: $1.59bn at that price is about 20,100 bitcoin ^[9], 2.4% of the 840,447 held ^[10], and it equals 52% of the whole unrealised gain on the position, since a $66.4bn mark against a $63.36bn cost basis ^[13] leaves $3.04bn ^[19][20]. A company whose entire pitch is that dollars decay has parked half its paper profit in dollars.

The counter-thesis is respectable and might win. Two point four percent of assets in cash would be unremarkably conservative for any other issuer, the reserve does exactly what the company says it does by decoupling purchases from market timing ^[3], and on the year the ledger still reads 175,000 bought net against 7,000 sold, which puts sales at 4% of net accumulation ^[5][11]. Saylor's claim to a net accumulator posture is arithmetically true ^[7].

This could run several ways from here. The sales were lot management and a tax artefact, the equity machine does all the funding from September onward, and the reserve becomes dry powder rather than a dividend account. Or the preferred coupon load compounds faster than issuance capacity, and every quarter brings its own 3,588 coins. Or bitcoin runs, the premium to net asset value widens, and the funding question disappears for another year, which is the outcome the model has always quietly required.

This is probably wrong, but I read the $1.59bn as a confession of sequencing: cash was raised first because the coupon does not wait for a favourable print, and the buy signal is the visible half of a funding decision made weeks earlier. What would break the thesis is specific and checkable. If the 8-K behind the August 30 post ^[1] shows purchases funded from fresh equity with the reserve untouched and growing, and no coins leave the stack through the next dividend date, then the summer was a choice about timing rather than a constraint on cash. The narrower risk is priced in the gap between the $79,000 mark and the $75,385 average cost, which is 4.8% ^[21]: below the basis, per the publisher, margin provisions on the debt instruments or further sales start setting the schedule instead of the chairman ^[15]. The tweet is noise; that 4.8% gap is the number to hold.