The Pure-Play Treasury Is Dead. MSTR Is Next in the Autopsy Line.

Satsuma Technology voted to liquidate. 668 BTC. Roughly $43.5 million at current prices. The shareholder resolution passed with north of 90% approval. The vote date was July 21, 2026. This was not a rug pull. Not a hack. Not a regulatory seizure. The company did exactly what a pure-play bitcoin treasury was designed to do: hold bitcoin. And it died anyway. The ledger does not lie, only the narrative does. The narrative was “accumulate BTC at all costs.” The ledger recorded zero cash flow, a persistent discount to net asset value, and a shareholder base that finally asked the question nobody wanted to ask in 2021: what do we actually get paid? VanEck’s Matthew Sigel flagged the trend. The pure-play treasuries are exiting. Not one company. Not a few. The category. Let me define the animals in this zoo. Strategy (MSTR) is the industry pioneer. Credit model. Convertible notes, ATM equity issuance, preferred stock. The flywheel is reflexive: market premiums on its equity fund more BTC purchases, and more BTC purchases justify the premium. Satsuma Technology was the pure-play. Buy BTC. Hold BTC. Offer shares as a proxy for BTC. No leverage. No revenue. No hedge. Just concentration. Now two successor models are emerging, as Glenn Cameron framed it. Orange Juice, a new permanent capital vehicle backed by Lyn Alden and Jeff Booth, structured around holding BTC while generating real cash flow. And Twenty One Capital, a Tether-adjacent operation, restructured as permanent capital with stablecoin-issuer economics behind it. The industry’s capital architecture is being rewritten. What was a compliance question—how to hold BTC on a corporate balance sheet—has become an engineering question. Let me dissect each model, cold. Model one, the pure-play. Satsuma’s failure is structural, not cyclical. A corporation carries a cost structure: employees, listing fees, auditing overhead, investor return expectations. A bitcoin treasury produces zero operating cash. Its only asset appreciates—or doesn’t. When the market starts pricing the holding at a discount to NAV, because any shareholder can buy BTC directly on an exchange without wrapper risk, the equity loses its reason to exist. The premium vanishes. The discount becomes a permanent feature. The liquidation vote passes with 90% approval because the alternative is holding a forever-discounted shell. The lesson is simple: a corporation is a cost structure. If the asset on the balance sheet generates no yield and commands no premium, the cost structure eats the asset. Model two, MSTR’s credit model. The mechanics are elegant. Issue convertible debt at low rates. Buy BTC with the proceeds. Ride BTC appreciation to push the equity price above conversion. Use the equity premium to fund the next round. For three straight years, this worked. Here is what the model ignores: the flywheel reverses. Premiums are sentiment variables, not structural constants. If the ATM premium collapses, the company cannot issue equity at accretive prices. The debt still matures. The coupon still comes due. Where does the cash come from? Saylor’s model requires BTC appreciation to outpace debt costs forever. That is a hypothesis, not a law. I have seen this reward schedule before. In 2018, tracing ERC-20 vesting logic in a failed ICO, I found an integer overflow that allowed early team members to drain the treasury before the public sale. The bug was not malicious intent. It was a structure with no check. MSTR’s structure has no check either. It relies on the market’s willingness to keep rating equity above NAV. That is trust, not math. Collateral was a mirage; solvency was a myth. The myth being that a premium can substitute for cash flow. Model three, permanent capital. Orange Juice and Twenty One Capital are attempting what I call the Berkshire Bitcoin wrapper. Use operational cash flow—or, in Twenty One’s case, Tether’s issuance profits—to systematically accumulate BTC without depending on equity premium. The “never sell” structure eliminates the forced-liquidator risk that killed Satsuma. The engineering questions remain. Who audits the cash flow claims? What happens to the vehicle if the underlying operating business, stablecoin issuance, becomes a regulatory liability? A permanent structure compounds trust. If the trust is misplaced, it compounds ruin. All three models share one flaw: they are capital-structure innovations, not protocol innovations. No smart contract enforces discipline. No formal verification validates the incentives. The discipline is a management team’s promise. Structure outlives sentiment; code outlives hype. And there is no code here. Now the admission the cynics hate. The bulls are not entirely wrong. MSTR’s credit model did acquire enormous BTC reserves using capital instruments that functionally did not exist a decade ago. That is not trivial. Corporate balance sheets have never been weaponized this aggressively as bitcoin accumulation vehicles. Reflexivity cuts both ways: in an uptrend, it converts existing equity value into more BTC exposure per share. Investors who modeled that scenario did well. The model’s flaw is timing, not logic. And the permanent capital model, for all its unproven structure, fixes the one thing the pure-play could not: the discount. A permanent vehicle that generates cash flow can justify a premium based on the present value of that flow, not on hope that the market re-rates stored coins. That is a measurable improvement. I exclude emotion from my equations. But I do not exclude data that disrupts my cynicism. The data says treasury vehicles are maturing. They are learning to pay shareholders. That is progress. I will not call progress a scam. I will call it unproven. Satsuma’s 668 BTC is the canary. The pure-play was never a strategy; it was a tax on shareholder gullibility. The next question is not whether MSTR buys more bitcoin. It is whether MSTR’s equity premium can outrun its debt schedule. Watch the cash flow statement, not the press release. The ledger records the day the music stops—and it records in red. Panic is just poor data processing in real-time. Read the structure before panic arrives, not after.