Strategy sold bitcoin three times between May and August 2026. This is not a rumour nor a market reading: it is in its own filings with the SEC. It sold 32 BTC in late May, 3,588 BTC in the first week of July and 1,638 BTC in the week of 27 July to 2 August. In total 5,258 bitcoin, and it has 845,050 left — around 65.6% of all the bitcoin held by listed companies worldwide.
Almost all the coverage has told this as the beginning of the end of a conviction. It is a comfortable reading and we believe it is the wrong one. What has changed is not what Strategy thinks about bitcoin: it is what Strategy has to pay every month in dollars.
The three sales, with date and destination
Late May 2026 — 32 BTC for around $2.5 million. The first bitcoin sale in its history. Symbolically enormous, materially irrelevant: 32 coins out of more than 847,000.
First week of July — 3,588 BTC for around $216 million. The position falls from 847,363 to 843,775 BTC. The proceeds go, according to the company itself, to paying distributions on its preferred stocks and to replenishing the portion of the dollar reserve used for those payments.
From 27 July to 2 August — 1,638 BTC for $104.7 million, at an average of $63,957 per bitcoin. This is the 8-K of 3 August. The breakdown matters: $52.4 million went directly to preferred dividends and $52.3 million to buying back STRC. In the same week it also placed 3.01 million common shares for $290.6 million net.
After that third sale it has 845,050 BTC left with an aggregate cost of $64,268M and an average entry price of $76,052.
Why it sells: the preferreds bill
Strategy is not just a treasury companyCorporate TreasuryA company that adopts Bitcoin as a primary or significant balance-sheet asset.View term →. It is a treasury company that financed its accumulation by issuing instruments that pay a coupon in dollars, and that is the detail almost nobody puts in the headline.
Today it has five preferred issues outstanding: STRK at 8%, STRF at 10%, STRD at 10%, STRE at 10% in euros and STRC at 12%. Above all of them, in the capital structureCapital StructureThe mix of debt, preferred stock and common equity that funds a company.View term →, there is more than $8 billion in convertible debtConvertible DebtA bond that can be converted into common shares under certain conditions.View term →. Between preferred dividends and interest, the annual obligation is around $1.76 billion.
The bitcoin on the balance sheet does not generate a single dollar of that amount. Bitcoin held pays no coupon: it appreciates or it does not, but it produces no cash. So Strategy only has three places to get those dollars from:
One: issuing common shares above the value of its bitcoin. This is the mechanism that made everything possible. As long as the mNAV is above 1, every new share it issues buys more bitcoin than it dilutes, and leaves cash along the way. It is the machine we explain in the mNAV indicator.
Two: drawing on the dollar reserve. Strategy maintains an explicit cushion for this. It fell as low as $871 million in May, recovered to $2.55 billion in July and after the August moves it is around $4 billion — some 27 months of coverage of that annual obligation.
Three: selling bitcoin. The last resort, and the one it has activated in 2026.
The sequence is what you would expect: when the first tap narrows —and it narrows when the mNAVmNAVMultiple on Net Asset Value: how a treasury company trades relative to its BTC.View term → compresses, because issuing below 1 destroys value for the shareholder— you have to open the other two. That is exactly what has happened.
What changed on 1 August
Two days before the 8-K, Strategy authorised bitcoin sales of up to $5 billion. That figure is worth translating: at the prices it has been selling at, that is around 78,000 bitcoin, close to 9% of its position. It is not the same as 5,258.
That it is authorised does not mean it will be executed —the company itself stresses that it has no obligation to do so—, just as an authorised buyback programme does not oblige a company to buy back. But a board does not approve a valve of that size never to use it. It is, quite simply, the size the company has put on its plan B.
That same day Saylor confirmed that the STRC dividend stays at 12% for August: the highest rate since the instrument went public in July 2025. STRC closed July at $89.46, 10% below its $100 par value, in an instrument whose entire design consists of trading at par. The full comparison with Strive's preferred is in STRC vs SATA.
What this is not
It is not a margin call. Nobody can force Strategy to sell bitcoin because the price falls. Its convertibles have no collateral clauses triggered by the price of BTC, and its preferreds are perpetual: the principal never matures. The difference between selling by your own decision and selling because a guarantee is being enforced against you is the difference between this case and the leveraged treasury companies we analysed in treasuries forced to sell.
It is not the Tesla case. Tesla sold around 75% of its position in the first quarter of 2022 and with that it stopped being a bitcoin treasury company. Strategy has sold 0.6% and keeps buying in parallel: its cumulative BTC Yield in 2026 is positive, 4.5%, which means that bitcoin per diluted share has risen despite the sales. That is why it remains on our list of HODL stocks, although now with the caveat written down.
And it is not free to tell this as if nothing had happened. For five years the argument was "we don't sell". Today the company sells, and whoever bought the stock on that promise deserves to be told so in the same letters used to tell them the previous thing.
The four signals that do matter
If you want to follow this without depending on headlines, there are four things to watch, and all of them are public.
The mNAV. As long as it stays above 1, the share-issuing machine works and the bitcoin sales are marginal. If it settles below, the first tap closes and the whole pressure shifts onto the bitcoin. It is the indicator that gives the most information per unit of attention.
The pace, not the fact. 5,258 BTC in three months is noise against 845,050. A sustained pace approaching the authorised $5 billion is not.
The STRC rate. Strategy has said it will not raise it just because the instrument trades below par. If it raises it anyway, it will be because it needs to place more paper, and that makes the annual obligation more expensive precisely when it is hardest to pay.
The dollar reserve. It is the buffer. It fell to $871 million in May; today it is around $4 billion. If it narrows again while the mNAV is compressed, the sales will stop being optional.
What it means for investors
For the MSTR shareholder, this does not change the thesis but it does change the profile: what you are buying is no longer just leveraged bitcoin, it is leveraged bitcoin with a cash obligation ahead of it. That obligation gets serviced first.
For the preferred holder, the reading is the opposite and almost reassuring: the company has shown it is willing to sell its most cherished asset rather than miss a coupon. That is exactly what a fixed-income investor wants to see, and it explains why the comparison with circular schemes made in June does not hold up.
And for the whole sector, it leaves an uncomfortable lesson that goes beyond Strategy: the capital structure that accelerates accumulation in the bull market is the same one that forces selling in the bear market. Any treasury company copying the preferreds model is signing that contract. The 179 companies in the directory are not all in the same situation, and that difference shows up on the balance sheet, not in the rhetoric.
Data verified as of September 19, 2026 in Strategy's 8-Ks of 3 August and 6 July, in the STRC rate announcement of 1 August and in the second quarter 2026 results. We review this piece every month; the position figures update themselves with the daily data.
