I spent a Tuesday morning staring at a number that refused to move. $17.51 billion. The remaining issuance capacity on Strategy's STRC preferred stock โ the instrument Michael Saylor markets as 'Digital Credit' โ had not changed across multiple reporting periods. Not up. Not down. Frozen.
If you have ever watched a state variable in a contract stop updating, you know the feeling. The function still exists. The ABI still lists it. But nothing writes to storage anymore. That is not stability. That is a channel that has quietly closed.
Peter Schiff walked into that vacuum with a podcast segment arguing that MicroStrategy has lost its bitcoin-buying power. Most people heard an old gold bug taking another swing at a maximalist. I heard something narrower and more interesting: a claim about a specific number in a specific filing. When a claim is that specific, you can actually check it. So I did. What I found was not a debate between two men. It was a capital structure in the middle of degrading, documented in its own disclosures, and defended with a metric that measures the wrong thing.
The three-layer machine
To read Strategy properly, stop thinking of it as a bitcoin fund. Start thinking of it as a machine with three moving parts. Layer one is BTC โ the asset. Layer two is MSTR, the common equity that trades at a premium to the bitcoin it holds. Layer three is STRC, a preferred stock that pays a fixed dividend and is branded, in Saylor's own language, as 'Digital Credit.'
The pitch is elegant on a slide. Sell preferred shares to yield-hungry capital. Use the proceeds to buy bitcoin. The bitcoin appreciates, the common-equity premium widens, and you sell more common stock at that premium to buy more bitcoin and to support the preferred. A flywheel that converts one asset's volatility into a fixed-income product for institutions.
I have seen this shape before, just in a different medium. In 2017, I pulled the 0x protocol's exchange contracts out of the ICO noise and reverse-engineered them line by line. The whitepaper promised a decentralized exchange; the Solidity delivered three integer overflow vulnerabilities I found before mainnet. The lesson was not that 0x was a fraud. The lesson was that the marketing layer and the mechanism layer are two different documents, and only one of them has to balance.
Strategy's mechanism layer is its filings. Read them as a mechanism and they tell a story the branding does not. There is no chain here, no contract, no gas โ this is financial engineering wearing technical clothing. But it shares the defining property of every protocol: it depends on assumptions holding. Two, specifically. That BTC appreciates over time. And that the financing window โ preferred or common โ stays open. When either fails, the structure does not halt like a chain. It degrades, one layer at a time.
The frozen variable
When an at-the-market issuance program is live and healthy, its remaining capacity is a moving target. Every sale draws it down. Every shelf refresh pushes it up. A capacity that sits perfectly still across multiple periods means one thing: nothing is being sold against it.
The original reporting conceded, in a rare moment of editorial honesty, that this data point 'partially supports Schiff's argument.' I would go further. A frozen issuance capacity is the closest thing a treasury company has to a contract revert. The function is callable. It is not being called.
There are two readings, and the difference matters. The benign reading: management paused issuance on purpose because pricing was unfavorable โ a disciplined operator waiting for a better window. The less benign reading: the market stopped absorbing STRC, and the pause was imposed rather than chosen.
The price history settles it. Earlier this year, STRC traded down to roughly $75 against a $100 face value. That is a 25% drawdown on something sold as low-volatility credit. It has since recovered to about $99.40. A product that round-trips from par to $75 and back is not being priced as stable credit. It is being priced as a risk asset with a coupon attached. When demand is that fragile, a stalled issuance program is far more likely to be a demand problem than an act of pricing discipline.
I keep a rule from years of auditing: when a team emphasizes a metric that is not the metric in question, they are telling you where the problem is not. Saylor, answering Schiff, went on the offensive about volatility โ arguing STRC's 30-day historical volatility of 9% sits below SPY's 10%. That is a real number. It is also an answer to a question nobody asked. The question is not how volatile STRC was. The question is why its issuance capacity has not moved.
The downgrade from preferred to common
The second signal is what is actually funding the machine.
When the preferred channel works, you fund bitcoin purchases with STRC. When it stalls, you fall back to the common-equity ATM โ selling MSTR shares into the market. That is what the disclosures now show: the marginal dollar of bitcoin acquisition increasingly comes from common stock sales and a cash account, not from preferred issuance.
This is a downgrade, and it is worth being precise about why. Preferred stock carries a fixed dividend and sits above common equity in the capital stack. Common stock is the residual โ the last claim, the most price-sensitive, the most dilutive to existing holders. Funding growth with common equity instead of preferred is like paying for a new server with the equity in your house rather than a fixed-rate loan. It works, until the asset price supporting that equity wobbles.
There is a second-order cost that rarely makes the headline: every common share sold dilutes bitcoin-per-share. The entire point of the premium is that one MSTR share represents more bitcoin than its market value implies. ATM issuance at a premium is accretive to that ratio. ATM issuance when the premium narrows is not. The mechanism has a threshold, and the threshold is the premium itself.
The disclosures also show something subtler. In September, roughly $103.5 million of common-stock proceeds were used not to buy bitcoin but to repurchase STRC. Read that again. The company sold its residual claim to buy back its senior claim. That is not growth financing. That is market-making to defend a price โ the capital-structure equivalent of a protocol spending its treasury to prop up the peg of its own token. It can work. It is also a confession that the peg needed propping.
The summer that broke the 'never sell' rule
Here is the fact I keep returning to, because it is the one that cannot be narrated away.
During the summer, Strategy paused bitcoin purchases for roughly ten weeks. Not for a market holiday. Because it was selling bitcoin to pay dividends and to fund STRC repurchases.
The public story of Strategy โ the one that built the premium โ has always been a one-directional rule: accumulate, never sell. Saylor's entire brand as a bitcoin maximalist rests on the claim that the treasury only moves one way. A ten-week pause in buying is a soft violation. Selling bitcoin to meet a dividend obligation is a hard one.
Code is law, but bugs are the human exception. The code here is the stated rule โ never sell. The bug is the dividend obligation that made the rule impossible to keep. And the ledger remembers what the wallet forgets: the treasury sold, and the disclosure said so, even as the narrative moved on.
This is the strongest piece of evidence in the entire Schiff-versus-Saylor exchange, and neither man foregrounded it the way it deserves. Schiff focused on buying power. Saylor focused on volatility. The actual signal is that the structure already reached the point where it had to liquidate its core asset to service its obligations. Everything downstream โ the frozen capacity, the shift to common equity, the buyback โ is consistent with a machine under stress.
The reserve math
Strategy holds a USD reserve of about $502 million to pay preferred dividends and debt interest. On its own, that sounds like a substantial cushion. Set against the scale of the structure, it shrinks.
Remaining STRC issuance capacity is $17.51 billion. If that capacity were fully drawn and the preferred fully outstanding at face, a $502 million reserve covers a few quarters of distributions at best, depending on the coupon โ which, notably, the coverage I reviewed never spells out. That is a gap. Anyone evaluating the credit quality of STRC needs the coupon rate, the liquidation preference, and the maturity terms. None of them appear. When the key terms of a yield product are missing from the discussion, that absence is itself information.
I will be fair to my own method here. I am extrapolating total STRC float from the remaining capacity, and that is a lower bound, not a measurement. I mark that assumption medium-confidence, and I would want the actual outstanding balance before drawing a hard conclusion on coverage. But the direction is not ambiguous: a reserve measured in hundreds of millions, backing a structure sized in tens of billions, is thin.
Add the bitcoin cadence for scale. In September, Strategy's disclosed purchases totaled roughly 7,218 BTC across three tranches โ 4,603, then 1,665, then 950. The first tranche is large. The ones that follow shrink. Against the entire daily volume of the bitcoin market, even the large week is a marginal bid. Strategy is a meaningful holder. It is not a decisive buyer. The premium implies it is decisive. It is not.
The anchor and the alternative
Strategy is not just one company holding bitcoin. It is the pricing anchor for an entire class of corporates โ the digital asset treasury cohort โ that copied the playbook. When the anchor's financing model strains, the sector re-rates, because every imitator's ceiling is the leader's ceiling. The direct effect on bitcoin spot is modest. The effect on the valuation multiple of every other treasury company is not. That contagion, not the bitcoin bid, is where the real second-order damage sits.
And the alternative is sitting right there. A spot bitcoin ETF offers the same exposure with no leverage, no dilution, and no dividend obligation. Strategy's only differentiation is amplification โ it magnifies the upside in a bull market and the downside in a bear one. In a bull market, that premium looks like genius. The entire question is what it looks like on the other side. The DAT model is not competing with ETFs on safety. It is competing on leverage, and leverage is a feature that reverses.
Accounting is the quiet lever
One more structural input deserves attention, because it sits underneath everything. Bitcoin held on a corporate balance sheet is subject to accounting rules that determine how its price moves translate into reported earnings. When those rules shift โ toward fair-value measurement, for instance โ the volatility of the reported numbers rises, and with it the scrutiny from auditors, analysts, and the market's risk models. The financing window is not only a function of investor appetite. It is a function of how the reported numbers look when the appetite arrives. A treasury strategy that depends on capital markets staying open is, in part, a bet on favorable accounting. That bet is not in the marketing material.
The volatility paradox
Everyone in this debate is arguing about the wrong risk.
Schiff says the buying power is gone. Saylor says the volatility is low. Both are describing the past. The number that actually matters โ the premium of MSTR over the bitcoin it holds, the mNAV โ appears in neither argument, and it is the hinge the entire structure turns on.
Here is the trap. Saylor cites STRC's 30-day historical volatility of 9%, below SPY's 10%, as proof of maturity. But historical volatility is a backward-looking statistic, and it systematically understates tail risk in a product whose price already visited $75 from $100 in the same year. A series that fell to $75 and recovered to $99.40 will, by construction, show moderate realized volatility โ because the recovery washes the crash out of the average. The volatility measure is being asked to certify safety that its own price path contradicts. This is the volatility paradox: the smoother the recent tape, the more it hides the tail that already happened.
I have made this mistake in code. A function that passes every test in the suite looks safe. It is the test you did not write โ the reentrancy path, the integer boundary โ that kills you. STRC passed the 'low volatility' test. It has not been tested against a sustained bitcoin drawdown, because in the window we are examining, bitcoin cooperated.
The structural risk is not that STRC is volatile. It is that the machine requires two things to hold at once: bitcoin rising, and the financing window staying open. Those two assumptions are correlated, not independent. A bitcoin drawdown compresses the MSTR premium, which chokes the common-equity ATM, which removes the funding for both bitcoin purchases and STRC support. The failure modes arrive together. That is not diversification. It is a single point of failure with two names. Every treasury is a smart contract with an unaudited human inside, and the human here is a dividend obligation that does not negotiate.
The test comes Monday
The next disclosure is the test. Strategy files its MSTR sales data on a Monday, and that filing will show whether the common-equity ATM is still carrying the load the preferred channel dropped.
If sales are steady, the machine is still running and the premium is still doing its job. If they slow, then the summer sale of bitcoin was not an exception. It was the first payment on a structure that is quietly learning it cannot pay its bills the way it promised.
Watch the number that moves. Then watch the one that does not. Because in a bull market, the frozen variable is the only honest witness left in the room.