Strategy has stopped buying bitcoin and stopped selling stock

Two straight weeks with no bitcoin bought, no bitcoin sold, and not a single share issued. Instead $139.3 million of cash went into retiring its own preferred stock. The machine that made Strategy famous is not running in either direction.

For four years the question about Strategy was how much bitcoin it would buy next. The answer arrived weekly, on a schedule, in a tweet. Then it stopped. Two weeks running now, the 8-K has said the same thing: no bitcoin purchased, no bitcoin sold, no shares issued. What the company did instead was spend $139.3 million of cash buying back 1,420,467 shares of its own preferred stock, on top of $176.3 million the week before. This publication has already written about Strategy issuing equity to service preferred dividends, which was the shape of the machine running backwards. This is something else. Issuing equity to pay dividends is at least a machine. Buying back the preferred with cash while touching neither bitcoin nor the share count is a company that has stopped operating its own model and started managing its balance sheet. The distinction matters, and almost none of the coverage has drawn it.

What the filing says

Precision matters because several outlets have described this as a pause in bitcoin buying, which is the smallest part of it.

The 8-K filed September 14 covers September 8 through 13 and reports three separate non-events. Strategy did not purchase bitcoin. It did not sell bitcoin. And it did not sell any shares under its at-the-market offering programme.

What it did do: repurchase 1,420,467 shares of STRC preferred stock for $139.3 million, funded entirely from USD Cash. Not from equity issuance. Not from bitcoin sales. From the balance sheet.

The position numbers moved accordingly. Bitcoin holdings held at 845,050 coins acquired for $63.73 billion, an average cost near $75,412. USD Cash fell to $1.30 billion. The dedicated USD Reserve held steady at $5.10 billion, giving total USD assets of $6.4 billion. Remaining authorisation sits at $1.05 billion for preferred repurchases and $1.0 billion for common.

Saylor’s own framing came with numbers attached: STRC’s BTC credit at 57 basis points and USD duration at 3.9 years, calculated on assumptions of 10% bitcoin annual return, 40% volatility, and a bitcoin price of $77,266.

The last actual bitcoin purchase was August 31, when the company acquired 4,603 coins for roughly $370 million, ending a ten-week hiatus. That purchase is now marginally underwater against a price in the $77,000s only because the average cost across the whole position is $75,412.

Three ways to stop

A treasury company can stop buying for three reasons and they have very different implications. Working out which one this is determines everything.

It cannot buy. No access to capital at acceptable terms. Equity issues below net asset value and is dilutive, debt markets are closed or expensive, and selling bitcoin to buy bitcoin is obviously circular. This is the distressed reading.

It will not buy at these prices. Management has a view that better entries are coming, or that the current price relative to its own cost basis makes accumulation unattractive. This is the disciplined reading and it is the one the company’s own framing supports.

It has found a better use for the money. Retiring preferred stock below par reduces future dividend obligations at a discount, which is a real return calculation with a computable answer. This is the capital allocation reading.

The evidence points mostly at the third, and the company says so directly: purchases below the $100 stated amount are described as accretive because they retire future dividend obligations at a discount. That is not spin. STRC carries a 12% annualised dividend from September, so every share retired below par removes a 12% obligation bought at a discount to its face value. Against bitcoin at a price a few percent above the company’s own average cost, the preferred buyback is arguably the better trade on any conventional measure.

Which is exactly the problem, and it is worth stating plainly. The entire investment case for this company is that it converts capital into bitcoin more efficiently than an investor can. When management concludes the best available use of a dollar is retiring its own preferred instead of acquiring the asset, it has answered a question about relative value that its shareholders bought the stock specifically to avoid asking.

What the absence of the ATM tells you

The detail almost nobody has picked up is the second one, and it is the more informative half.

Strategy sold no shares under its at-the-market programme during the period. For a company whose defining mechanism was issuing equity at a premium to buy bitcoin, and which more recently was issuing equity to fund preferred dividends, a week with zero issuance is a change of state.

Two readings, again in tension.

The generous one: management does not want to dilute at current levels. MSTR has fallen substantially over the past year, the premium to net asset value that made issuance accretive has compressed, and a disciplined operator stops issuing when issuance destroys value. That is correct behaviour and it is what shareholders should want.

The uncomfortable one: the ATM is the company’s primary funding mechanism, and a business that has stopped using its primary funding mechanism has fewer levers than it had. Cash fell to $1.30 billion this week while the reserve stayed at $5.10 billion, which means the buyback came out of working cash, not the ring-fenced pot. That can continue for a while. It cannot continue indefinitely without either the ATM restarting or something being sold.

Both readings describe the same company. The question is whether the pause in issuance is a choice about price or a constraint about access, and the filings do not distinguish between them.

The MSCI fight nobody is connecting

Running alongside all of this is a dispute that received a fraction of the attention and may matter more.

Saylor and chief executive Phong Le asked MSCI in early September to withdraw an index rule that could remove Strategy from its global benchmarks, arguing the rule unfairly targets the company. Index inclusion is not a trivial matter for a stock like this. Passive funds tracking MSCI benchmarks buy and hold constituents mechanically, without forming a view. That is a source of automatic, price-insensitive demand, and it is the kind of demand that supports a share price independently of whether anyone likes the business.

Lose it and the marginal buyer becomes someone who has actually decided to own the stock. For a company whose equity has traded at large premiums to the value of its holdings for years, replacing passive demand with discretionary demand is a meaningful change to who sets the price.

Connect that to the buyback and a coherent picture appears. Strategy is simultaneously defending its preferred at par, defending its index inclusion, and declining to dilute its common. Those are three separate actions aimed at the same objective: holding up the capital structure while the accumulation engine is idle.

None of that is failure. It is what competent management does when conditions are unfavourable. It is also very obviously not the strategy the company is named after.

The arithmetic of the buyback

Worth doing, because the numbers make the case better than the framing does.

STRC has a $100 stated amount and pays a variable cash dividend running at 12% annualised from September. It has been trading below par, which is why the buyback exists: repurchasing below $100 retires an obligation at a discount to its face value.

At $139.3 million for 1,420,467 shares, the average paid works out around $98 per share. Retiring a share at that price removes roughly $12 of annual dividend obligation for a $98 outlay, which is a return on capital in the region of 12% before considering the discount to par.

Against what? Bitcoin at $77,266 versus an average cost of $75,412 is a position roughly 2.5% above water. Buying more at those levels adds to an asset the company already holds in enormous size, at a price barely above its own blended entry.

On a straight comparison the buyback wins, and it is not close. Which is why the decision is defensible and why it is also the tell. A management team running this calculation honestly and choosing the preferred is a management team that has priced its own bitcoin accumulation against an alternative and found the alternative better.

Total spent on STRC repurchases since July: approximately $811.5 million. That is roughly two and a quarter times the $370 million the company spent on its last actual bitcoin purchase.

What this means for the model

The treasury company model, which our earlier work has traced through its flywheel and its reversal, depends on a specific sequence: trade above net asset value, issue equity into the premium, buy the asset, raise bitcoin per share, support the premium, repeat.

Every element of that sequence is currently switched off. The premium has compressed. Issuance has stopped. Accumulation has stopped. Bitcoin per share cannot rise because the numerator is fixed and the denominator is not growing either, which is the one mercy in the current configuration.

What remains is a company holding 845,050 bitcoin, $6.4 billion in dollar assets, a preferred stack it is buying back at a discount, and an operating software business that is immaterial against any of it. That is a closed-end fund with a capital structure attached, and closed-end funds trade at discounts to net asset value considerably more often than they trade at premiums.

The imitators face this in worse shape. Our audit of one XRP treasury vehicle arriving at its listing gate with holdings more than fifty percent below cost described a company with none of Strategy’s balance sheet. Strategy has $6.4 billion in dollar assets and $2.05 billion of remaining repurchase authorisation to manage its way through this. Most of the companies that copied it have neither, and what the archetype does with a cushion is what the imitators will have to do without one.

Saylor’s own numbers, decoded

The chairman published a set of figures alongside the filing that almost nobody translated, and they are worth translating because they show how the company is thinking about the preferred.

He reported STRC’s BTC credit at 57 basis points and USD duration at 3.9 years, on stated assumptions of 10% bitcoin annual return, 40% volatility, and a bitcoin price of $77,266.

Take those one at a time. BTC credit at 57 basis points is a measure of how much of the preferred’s obligation is effectively backed by the bitcoin position under the stated assumptions. A low number here is the point: it says the preferred is well covered, that the claim against the bitcoin is small relative to the asset, and therefore that holders should be comfortable at par.

USD duration of 3.9 years describes how long the dollar assets can service the obligation without anything else happening. With $6.4 billion in dollar assets against roughly $1.76 billion of annual preferred dividends and interest, just under four years of coverage is the arithmetic, and it is the number management most wants the market to see.

And the assumptions carry the weight. Ten percent annual bitcoin return and 40% volatility are both reasonable long-run inputs and both are choices. Run the same calculation with a flat bitcoin price and the credit number moves. Run it with higher volatility and it moves further. None of that makes the published figure wrong; it makes it a model output with inputs the reader should see, and Saylor to his credit published them.

What the numbers are doing rhetorically is answering a question the market has been asking since STRC fell below par: is this security money-good. The answer offered is yes, with four years of dollar coverage and a thin claim on a very large bitcoin position. That is a solvency argument, and the fact that the company is making one at all tells you what the preferred’s price has been saying.

What a restart would look like

Because the whole current posture is defensive, it is worth setting out what would have to happen for the machine to run again, since those conditions are observable.

STRC back at or above par. The buyback exists to push it there. Above $100, the security can be issued at a premium again, and proceeds from issuing preferred at a premium can buy bitcoin. That is the original loop with the preferred substituted for the common, and it is available to Strategy in a way it is not available to imitators without a credit market.

Or MSTR back to a meaningful premium. If the common trades well above the value of the underlying bitcoin again, the at-the-market programme becomes accretive and the classic mechanism resumes. That requires a market willing to pay more than a dollar for a dollar of bitcoin held inside a corporate wrapper, which it did for years and stopped doing this one.

Or bitcoin well above $75,412. The higher the price runs above the average cost, the more the existing position does the work, the more comfortable the credit metrics look, and the easier every financing decision becomes. This is the passive route and the one the company is arguably positioned for: hold, defend the capital structure, wait.

And the index question resolved. Passive demand restored removes one source of pressure on the common.

Any one of those improves the position. Two of them restart the model. What is notable about the current filings is that management appears to be positioning for the third, which is the one it does not control, while spending real money defending the first two, which it does.

That is a reasonable allocation of effort. It is also a company whose next move depends on the price of an asset it has stopped buying.

The week in context

Strategy did not stop in isolation, and what its peers did in the same window sharpens the picture considerably.

Strive, the fifth-largest public bitcoin treasury, bought 469 BTC between September 8 and 11 at an average of $77,954, taking its holding to 25,000 coins. Note the price: Strive paid above Strategy’s entire blended cost basis, and did it in the same week Strategy declined to buy at all. Two companies, same asset, same week, opposite decisions.

DeFi Development Corp, a Nasdaq-listed Solana treasury firm, grew its holding 2% to 2.39 million SOL since late August and opened a $300 million at-the-market programme for CHAD, its Solana-backed preferred stock. That is a company still running the mechanism Strategy invented, at an earlier stage of the cycle, with the preferred being issued rather than repurchased.

BitMine is closing in on a 5% accumulation target in ether, with roughly 5.07 million tokens staked and combined crypto and cash valued at $15.8 billion.

So the sector is not uniformly paused. The largest and most mature participant has stopped while smaller and newer ones are still accumulating, which is the pattern you would expect if the constraint were cost of capital rather than conviction. Strategy’s position is enormous, its preferred stack is expensive to service, and its premium has compressed. A company at 25,000 coins with a smaller capital structure faces none of those pressures yet.

The uncomfortable reading, and it is only a reading, is that the imitators are buying at prices the originator has declined, using mechanisms the originator has stopped using, at a point in the cycle the originator appears to be defending against. Whether that is confidence or inexperience is the sort of thing that resolves slowly and publicly.

What holders of each security actually own

Three groups have money in this company and the pause means something different to each. Almost all coverage collapses them into one.

Common holders own the residual. Their claim sits behind the preferred, and their return depends on bitcoin per share rising. It is not rising. The numerator is frozen at 845,050 coins and the denominator is frozen too, because no shares are being issued, which at least means it is not falling. What they are getting instead is a management team spending cash to retire preferred, which improves the residual claim by shrinking what sits in front of it. That is a real benefit, delivered slowly, and it is not the benefit the equity story promised.

Preferred holders own a 12% annualised claim with a $100 stated amount, currently trading below it. They are the direct beneficiaries of everything happening right now: the buyback supports the price, the dividend was raised, the reserve is being held intact, and the chairman is publishing coverage metrics specifically to reassure them. The company is prioritising this group, which is rational because the preferred is the security whose price most immediately constrains future financing.

Bitcoin holders who own neither have the cleanest position and it is worth saying. A company holding 845,050 coins that has stopped buying is a company that has stopped being a source of demand. Strategy was, for years, one of the largest and most predictable buyers in the market. Two weeks is not a trend. Two weeks following a ten-week hiatus, with a stated preference for retiring preferred over accumulating, is a demand source that has become conditional.

The practical read for anyone evaluating MSTR against simply owning bitcoin is that the wrapper currently offers no accumulation, a capital structure being actively defended, and an index question outstanding. That is a set of risks without the corresponding upside the wrapper was supposed to deliver, and it is the clearest argument for the discount the market has applied.

What to watch

Whether the ATM restarts. The single most informative line in the next 8-K. Issuance resuming means the funding mechanism is available and management chose not to use it for two weeks. Continued silence means something else.

Whether the preferred authorisation gets consumed. $1.05 billion remains. The pace at which it is spent tells you how much of the current posture is a temporary allocation decision and how much is a standing policy.

STRC against $100. The whole buyback exists to push it back toward par so the security can be issued at a premium again and the proceeds used to buy bitcoin. If it gets there, the machine can restart. If it does not, the buyback is a subsidy rather than a repair.

Bitcoin against $75,412. The average cost basis. Above it, accumulation is a reasonable use of capital and the pause looks tactical. Meaningfully below it, every decision in this piece gets harder.

The MSCI determination. Index exclusion would remove a source of automatic demand for the common at exactly the moment the company has stopped issuing into it.

Frequently Asked Questions

Disclaimer: This article is for information and educational purposes only and does not constitute financial or investment advice. Figures reflect regulatory filings and reporting available at the time of writing and change with each disclosure. Nothing here is a recommendation to buy, sell, or hold any security or asset. Always do your own research. Information is accurate as of September 15, 2026.