Last May 6, during the conference call on first quarter 2026 results, Michael Saylor uttered the phrase he had sworn for years he would never utter: “We’ll probably sell some bitcoin to pay a dividend”. The MSTR stock lost over 4% in the hours that followed.

In absolute terms, the figure is laughable. Strategy holds 818,334 BTC, and annual obligations between preferred stock dividends and debt interest amount to roughly $1.5 billion. At current prices, we’re talking about selling fewer than 20,000 bitcoin per year — slightly more than 2% of reserves. Strategy could keep this up for decades without significantly denting its treasury.

So the proportions of the potential sale are too small to call it the beginning of the end. They do, however, offer an opportunity to analyze, once and for all, the house of cards of the quintessential Bitcoin Treasury Company.

For five years, Strategy’s pitch has been one and one only: we accumulate bitcoin, we don’t sell bitcoin. Everything else — the model, the narrative, the so-called yield, the financial engineering around preferred shares STRK, STRF, STRD, and STRC — rested on that implicit pact with shareholders.

Admitting today that some bitcoin will be sold to honor dividend obligations effectively means acknowledging that operating cash flow isn’t enough. Also because operating cash flow is, in fact, nearly nonexistent and, in any case, entirely irrelevant in terms of volume. Strategy’s software business generates a handful of millions of dollars per quarter. Financial obligations on preferred securities require one and a half billion per year. The difference is covered the way it has always been covered: by issuing new shares at a premium over the value of the underlying bitcoin, or — and this is the novelty — by selling some bitcoin.

Saylor said it with the smile of someone who just explained something obvious: “We buy bitcoin on credit, let it appreciate, then sell bitcoin to pay the dividend”. Read that way, it sounds like a perpetual motion machine. Read with a touch of cynicism, it’s the description of a model that depends entirely on two variables Saylor doesn’t control: the price of bitcoin and investors’ willingness to keep buying his shares at a premium over the net value of the bitcoin held (the so-called Net Asset Value, NAV).

A few days before the conference call, Adam Back had tweeted a phrase destined to become the new slogan of the Saylor-boys: “Bitcoin treasury companies are an arbitrage between the fiat present and the hyperbitcoinized future”. A formulation as elegant as it is dishonest.

An arbitrage, in finance, is an operation that exploits a price difference of the same asset across two different markets, buying where it costs less and selling where it costs more, without directional risk. It is, by definition, a zero or minimal risk operation that closes within a finite timeframe.

What Strategy does bears little resemblance to an arbitrage and a great deal of resemblance to a leveraged bet on bitcoin, financed by issuing debt instruments and shares at a premium over the value of the underlying assets. It is a directional trade in every sense, exposed to a decline in bitcoin’s price, to illiquidity in the preferred stock market, and — this is the interesting part — to the evaporation of the premium itself.

Calling it “an arbitrage between the fiat present and the hyperbitcoinized future” is an elegant way to hide the fact that the model only works as long as someone is willing to pay $1.23 of MSTR for every $1 of underlying bitcoin. The day the premium drops to zero — something that already happened briefly in early 2026 — the machine jams. The day it drops below zero (shares worth less than the bitcoin they represent), the model transforms from an accumulation engine into a value destruction engine.

To try to simplify, imagine a line of people entering a room, each carrying a banana. The rule is this: whoever enters must hand over the banana at the counter, receive in exchange a certificate that reads “you are entitled to 1.1 bananas”, and get in line with the others.

The certificate is valid as long as there’s someone else in line ready to enter. Every time a newcomer hands over their banana, one-tenth of that banana is distributed among the certificates already issued. Everyone sees their certificate go from “1.1 bananas” to “1.12 bananas” to “1.15 bananas”. That’s the yield.

Existing certificates increase in value only because new banana-bearers keep entering. Nowhere is there a tree producing bananas. Nowhere a field, a harvest, a product. There’s only the line. As long as the line keeps growing, everyone sees their certificate appreciate. The day it stops growing, the certificate is worth exactly the banana that was handed over. The day someone tries to leave and take their banana with them, they discover that the banana has already been split among everyone else.

The “BTC per share” that Saylor publishes every quarter as a key metric works this way. It’s the premium that new shareholders pay above the value of Strategy’s bitcoin, redistributed as “growth” to existing holders. The returns of old shareholders come directly from the capital contributed by new ones.

I’ll leave it to you to judge the name given to this scheme in other financial contexts. Some people might find those bananas in undesirable places.

Then there’s the chapter dedicated to the crown jewel of Strategy’s financial engineering: the preferred stock STRC, rebranded “Stretch”, which currently pays an annualized dividend of 11.5%, distributed monthly, with a par value of $100 per share.

The official pitch is disarming: Saylor describes it as a “money market-like stability” instrument, with a variable rate adjusted monthly to keep the price around par. Translation: buy STRC, receive 11.5% annually, don’t worry about the price because it won’t fluctuate much since the rate self-adjusts.

Coffeezilla, in a recent video, analyzed this narrative. STRC has little in common with a deposit account and even less with a bond: no maturity, no guaranteed repayment, no capital protection. Under the packaging is a perpetual preferred share, rated B- (junk) by rating agencies, paying a double-digit yield financed by… subsequent issuances of other preferred shares.

The most relevant data point cited by Coffeezilla is the holder profile: 80% of STRC is held by retail investors. Families, retirees, small savers convinced by the marketing that they had parked their money in something “similar to a bank account”.

Coffeezilla also draws a bolder comparison: Terra Luna paid a 20% yield before collapsing. There too, an engineered structure that worked as long as it worked. There too, people convinced the yield was real. There too, the only source of the “yield” was new capital coming in.

The comparison is strong and in many ways unfair: Terra Luna was built on an algorithmically unstable mechanism by design, while STRC is a preferred share backed by a real asset, bitcoin. But the underlying principle that Coffeezilla puts on the table is correct: a double-digit yield that doesn’t come from operating cash flow has to come from somewhere. And in Strategy’s case, it comes from new buyers.

Strategy could survive for a long time. The reserves are enormous — $66 billion in bitcoin as of today — and with a net consumption of a few thousand BTC per year to honor dividends, the safety buffer covers decades. I’m not saying Strategy will fail tomorrow or even in five years.

What I’m trying to point out is that behind Strategy, as a product, there is only financial engineering. Bitcoin treasury companies, and Strategy first and foremost, are another card on top of the house of cards called the credit system. They work for the same reasons the fractional reserve banking system works: as long as everyone believes, nobody runs to the teller window. They work as long as the premium exists. The premium exists as long as the believers keep buying. It’s a loop that self-reinforces in the right direction and self-destructs in the wrong one.

Bitcoin was created to offer an alternative to exactly this type of construction. An asset free from counterparties, from promises, from yields financed by new subscriptions. Holding bitcoin means holding bitcoin, not a certificate that claims to represent 1.15 of them on the condition that the line keeps growing.

Saylor has spent five years convincing the world that his company was the best way to gain exposure to bitcoin. The truth is that his company is the best way to gain exposure to the credit system that bitcoin was designed to transcend.