Here is the data. Between July 27 and August 2, 2025, Strategy (the entity formerly known as MicroStrategy) sold 1,638 BTC. Of the proceeds, $52.4 million went to preferred dividends and $52.3 million went to buying back its own STRC preferred stock. The following week it sold another 1,690 BTC and pushed $108.6 million into the same buyback. Total BTC sold across the window: roughly 3,328 coins. Total BTC purchased: zero.
For a company whose entire market identity is built on the phrase "never sell," that is not a rounding error. It is a signal. And the market is reading the wrong headline.
Let me be clear about what this event is and what it is not. This is not a protocol exploit. This is not a DeFi event, a consensus failure, or a governance attack. This is a US-listed public company reallocating capital across a multi-layered security stack. I spent the last year auditing restaking slasher conditions and reconstructing order flow from on-chain dealer inventories, so mapping "tokenomics" onto "capital structure" is not a stretch for me. The mechanics rhyme. The incentives do not lie.
Context: The Stack
Strategy is no longer a software company with a Bitcoin treasury. It is a capital-markets instrument that happens to sit on 845,050 BTC. That number matters โ it is the largest corporate Bitcoin position in existence, and until this window it was the single most reliable marginal buyer in the market.
The structure has four layers, and understanding which layer is being fed tells you almost everything:
- STRC preferred stock โ a variable-rate perpetual preferred with a $100 par value. The associated reading cites a yield around 12%. That is not investment-grade preferred territory; that is distressed-adjacent, high-risk-premium territory.
- MSTR common equity โ the residual claimant. Takes BTC volatility plus dilution.
- Convertible debt โ interest-priority claims.
- Two isolated dollar pools โ a Flexible USD Cash pool ($1.3 billion as of September 13) and an Independent USD Reserve pool ($5.10 billion). The Reserve is restricted to preferred dividends and debt interest. It is explicitly NOT authorized for STRC repurchases.
That last line deserves a second pass. You do not isolate a $5.1 billion reserve and publicly wall it off unless you are trying to tell the market something. The company separated its war chest from its buyback ammunition on purpose. Total usable buyback firepower is roughly $1.3 billion in cash plus a $2 billion authorization โ and the authorization, doubled on September 8, is a permission slip, not a balance.
Core: The Mechanics Nobody Is Modeling
Here is the part the retail feed skips. When STRC trades below its $100 par value, buying it back at a discount does something specific: it retires a future dividend obligation at a lower cost than the obligation's face. If you buy a $100-par instrument yielding 12% for $92, you have just reduced your recurring cash outflow below what the coupon implies. That is a pure financial-engineering move. It creates value for the issuer without deploying a single dollar into Bitcoin.
The company has stated this motive explicitly. It keeps telling the market that discount buybacks reduce future preferred dividend needs. Read that again from an order-flow perspective: if the dividend burden were not a pain point, you would not lead every disclosure with a mechanism designed to shrink it.
The 2.57:1 ratio is the whole story. For every dollar Strategy put toward Bitcoin in this window, it put roughly $2.57 toward retiring its own preferred stock. That is not a treasury company buying the dip. That is a leveraged operator servicing its own cost of capital.
Now trace the cash. Where did the buyback dollars come from? Not from software revenue. Strategy's operating business is a rounding error against an $84 billion-plus BTC position. The cash came from selling the asset โ the exact asset the entire equity narrative depends on. Dividends and buybacks were funded by liquidating the collateral.
I have seen this pattern before. In May 2022, I watched the Terra flywheel run in reverse: a structure that magnifies gains on the way up magnifies losses on the way down, and the trigger is always the same โ the cost of attracting new capital rises above the return on the asset it funds. Strategy is running a slower, cleaner, SEC-registered version of the same reflexive loop. New capital (STRC, ATM, converts) buys BTC. BTC and MSTR valuations rise. Higher valuation supports cheaper capital. Cheap capital pays the old preferred dividends. It works โ until the cost of capital stops falling.
And the cost of capital is not falling. A ~12% preferred yield sits above most reasonable forward return assumptions for BTC at current prices. When your financing cost exceeds your asset's expected return, arbitrage is dead. You are no longer compounding. You are treading water and paying a fee for the privilege.
The two-pool structure confirms the constraint is real. If management could tap the $5.1 billion Reserve for buybacks, dividing the pools would be pointless accounting theater. They divided them because the Reserve is already spoken for โ dividends and interest on the debt stack. The flexible cash is the only unencumbered capital, and at $1.3 billion it is a fraction of the authorization headline.
One more thing on the accounting. When a preferred instrument trades below par, marking it and managing the dividend load becomes a live issue every reporting cycle. This is where a patient short-seller, or a newly engaged auditor, finds the soft spot. I flagged the same class of risk in early 2023 when I sized EigenLayer restaking exposure โ before mainnet, before the operator set matured, the danger was never the headline yield, it was the condition that forced an unwind. Here the unwind condition is BTC price stagnation plus a rising preferred yield. Both are observable. Both are trending the wrong way.
Contrarian: What the Tape Is Actually Saying
Retail read the September 8 board decision โ doubling the preferred buyback authorization to $2 billion โ as a bullish signal. "They have conviction. They are supporting their own securities." That is the headline trade.
The smart-money read is the inverse. A company does not double an authorization to defend its preferred stock in a healthy capital environment. It does it when the instrument is under pressure and the financing channel is at risk. Doubling a defensive authorization is not strength; it is triage. The board is telling you which security it cannot afford to let slip.
If the doubling were genuinely offensive โ a signal of balance-sheet firepower โ you would expect it accompanied by BTC accumulation. Instead, in the same window, net BTC flow was negative. You cannot simultaneously argue the company has abundant capacity to expand and watch it sell the flagship asset to fund coupon payments. Those two claims cannot both be true.
The broader blind spot is structural. Strategy's moat was never the software. It was the ability to raise cheap capital against a Bitcoin treasury. Spot Bitcoin ETFs now offer the same exposure with lower fees, no company-specific credit risk, no preferred-equity seniority conflict, and no key-man dependency on a single executive's conviction. The ETF is not a competitor at the margin โ it is a substitute for the core product. That caps Strategy's funding premium structurally. Every incremental dollar of STRC at 12% is a dollar the market is charging for the privilege of holding a wrapper around something an ETF provides for basis points.
And the "never sell" narrative just cracked. The company sold BTC twice in this window. The scale is small โ roughly 3,328 coins against 845,050 held, under 0.4% โ but the direction of the signal dwarfs the size. Markets price narratives at the margin, and the margin just flipped from accumulation to distribution. Anyone still modeling Strategy as a price-insensitive buyer is working from a stale tape.
Takeaway
The actionable judgment is not about Bitcoin's next candle. It is about watching the company's funding cost and its net BTC flow as a single paired indicator. As long as the preferred yield holds near 12% and net BTC flow stays flat-to-negative, the reflexive loop is in its slow-reverse mode, and the market has not yet repriced that.
The tell to watch next is the Flexible Cash pool. If buybacks continue while that number shrinks and the Reserve stays walled off, the company is spending its last unencumbered dollar defending a security it has to defend. If a future disclosure shows BTC sales accelerating beyond dividend-funding needs, the flywheel is not slowing โ it is turning.
The question that matters is not whether Strategy likes Bitcoin. It is whether the market can tell the difference between a buyer and a coupon servicer wearing a buyer's coat. Watch the two pools. They already told you.