Crypto Treasury Companies: What You Own and What Makes Them Sell

On 11 August 2020, MicroStrategy announced that it had bought 21,454 bitcoins for $250.0 million, about $11,653 a coin. The same 8-K discloses a second announcement made the same day: a modified Dutch auction tender offer to buy back up to $250.0 million of its own class A common stock at between $122 and $140 a share.

Both halves of what this article is about were present on day one. The company bought the asset, and it managed its own share count. Six years later, now named Strategy Inc, it holds 840,447 BTC at an average purchase price of $75,385 — and in the week of 3–9 August 2026 it sold bitcoin to fund repurchases of its own preferred stock.

That arc is the subject here. A digital asset treasury company, or DAT, is a listed company that raises capital in the equity and credit markets and spends it on a crypto asset it holds itself. Shareholders get exposure by owning the company. This lesson works through what that ownership actually is, the machinery that makes it work while it works, and the specific mechanisms that reverse it.

Treasury demand is one of two external channels now reaching a major crypto asset. The other is the exchange-traded fund, covered in ETF flow basics, and the structural comparison between them is worked through in what moves Ethereum's price.

1. What you own is a residual claim, not a coin

The pitch for a treasury company is that it is a convenient wrapper: you buy a ticker in an ordinary brokerage account and you get exposure to an asset you would otherwise have to custody. That much is true. What the pitch usually leaves out is where the common stock sits in the queue.

Strategy states the position plainly in its quarterly report for the period ended 30 June 2026. Describing its preferred stock and convertible notes, it says these instruments "rank senior to our class A common stock, and would entitle their holders to claims on our assets (including bitcoin) senior to those of holders of our common stock if we were to liquidate."

That is the whole difference between a treasury company and a fund. An ETF share is a proportional claim on the fund's assets, and there is nothing between the holder and the coins. A treasury company's common stock is what remains after the debt is repaid and the preferred is satisfied. In a good market that distinction costs nothing. It is the entire story in a bad one.

The obligations are real and they are cash. The senior layers do not merely sit there. They demand coupons at maturity and dividends on schedule, in dollars, whatever the coin is doing. The next section puts numbers on that.

2. The financing ladder: who gets paid before you

Treasury companies do not have an operating business large enough to fund purchases of this size, so the coins are bought with issued securities. Over time that produces a layered capital structure, and each layer has its own claim.

Strategy's is the most developed example, because it has been building it the longest. Read the table below as the ladder at full extension rather than as a profile of one company: a treasury company adds these layers in order, and each one is a claim it did not previously owe.

A newer treasury company has fewer rungs, and that changes which failure mode it is exposed to. BitMine adopted its strategy in 2025 and funded its ETH position mainly with common stock sold under an at-the-market programme, adding preferred stock only after 31 May 2026, for $273.8 million in net proceeds. A shallower ladder means smaller fixed obligations, which is why its own reversal in August 2026 showed up as a share buyback rather than as a coin sale. Counting the rungs is a quick read on how far a given company is along this path.

What the ladder does to the equity

It converts a volatile asset into a leveraged claim on that asset. Fixed obligations sit in front of a position whose value moves twenty percent in a quarter, so the residual moves further than the coin does in both directions. That leverage is the reason the shares can outrun the asset in a rally, and it is the same reason they fall harder.

The dividend bill compounds while the asset does nothing. Strategy's preferred dividends rose roughly elevenfold year over year, from $58.1 million to $629.2 million for the comparable six months, because each new series added to the stack. That is a permanent claim on cash created to buy an asset that produces none. Bitcoin pays no yield; the preferred does.

Dividends are discretionary in the documents and not in practice. The same filing notes that the board "may choose not to declare or pay dividends, as applicable, or may pay less than the full amount," and then states the consequence: if the company fails to pay in full, the value of that series, the other preferred series, and the common stock will all likely fall. The option exists. Using it is not free.

3. The accretion engine, and the metric that shows it

The thing that makes the model work while it works is not complicated, but it is easy to state loosely. Here is the precise version.

If the market values the company above the value of the coins it holds, then selling new shares and spending the proceeds on more coins leaves every existing share backed by more coin than before. The share count rises, and the coin pile rises faster. That is what people mean when they say issuance is accretive.

Strategy publishes this directly. It reports bitcoin per share, which it calls BPS and quotes in satoshis, and it reports the percentage change in BPS over a period as BTC Yield. A positive BTC Yield means the capital raised that period bought more coin per share than it diluted. The company is explicit that this is the point of the metric: it uses BTC Yield to assess whether its "capital markets activity and bitcoin acquisition strategy has resulted in gross per-share accretion (or dilution)."

The market's name for the ratio that drives this is mNAV — the company's market value divided by the value of its holdings — and above 1 the arithmetic above works. Be aware that no filing uses that term. It is a market coinage, so treat any mNAV figure as someone's calculation rather than a disclosed number, and check what they put in the denominator.

Reading the engine slowing down

The interesting thing about a published accretion metric is that it records deceleration as faithfully as it records success.

BTC Yield for the first half of 2026 was 8.1%, against 19.7% for the first half of 2025. The dollar value of the accretion, which Strategy reports as BTC $ Gain, fell from $9,494 million to $3,207 million, down 66.2%. The engine is still turning. It is turning at roughly a third of the prior year's rate, while the fixed obligations from section 2 went up elevenfold.

Those two series moving in opposite directions is the whole risk. Accretion per share is a function of the market's enthusiasm for the shares. The dividend and coupon bill is a function of what was already issued. The first can fall to zero; the second cannot.

One methodology caveat worth carrying

Strategy changed how BTC Yield is calculated, and its own filing notes that under the updated method, "when BPS is increasing, the updated methodology will generally produce higher BTC Yield figures for subsequent Measurement Periods," with the converse when BPS is falling. This is a company-defined metric, not an accounting standard. Compare it across periods with the methodology note in hand, and do not treat two numbers computed differently as a trend.

4. When the loop runs backwards

Run section 3 in reverse. If the shares fall below the value of the holdings, issuing new stock lowers coins per share instead of raising it. Accretion turns to dilution, so a management team acting in shareholders' interest stops issuing. The buying stops with it — not because the company changed its view on the asset, but because its funding became expensive.

Nothing about that forces a sale. Two further things do, and both are observable in filings from the same week in August 2026.

The capital turns inward

BitMine Immersion Technologies built its ETH position through an at-the-market equity programme permitting sales of up to $24.5 billion of common stock. In the nine months to 31 May 2026 it sold 340,748,312 shares for $11.87 billion — more than half the 603,226,394 shares outstanding at 9 July 2026, issued in that window alone. Shareholders raised the authorised share count from 500 million to 50 billion in January 2026 to keep the programme running.

Then the direction changed. By its 10 August 2026 filing, the company reported repurchasing 19.1 million shares since 1 July 2026 under a $4 billion authorisation, with its chairman stating that it views the common shares as undervalued. Capital that had been buying ETH was buying the company's own stock instead. For scale on how far the loop had turned, the same company had sold 5,217,715 shares to an institutional investor at $70 each in September 2025.

A buyback announcement is the clearest public signal that the issuance channel has closed. A company only prefers its own shares to the asset when it judges the shares cheap relative to the assets behind them, which is precisely the condition under which it cannot raise accretively. The buyback and the pause in buying are the same event seen from two sides.

The obligations come due

Buying back stock uses cash. So does servicing the ladder. When a company has more obligations than cash, the coins are the asset it has.

Strategy's board authorised two repurchase programmes on 29 June 2026, each up to $1.0 billion, one for preferred and one for common. Nothing had been repurchased as of 30 June. In the week of 3–9 August 2026, the company reported selling 1,690 BTC at an average of $64,262 — against its stated average purchase price of $75,385 — and using the net proceeds to fund repurchases of 1,152,020 shares of STRC preferred for $108.6 million. STRC is the variable-rate series, running at 12% at the time of the filing.

Read that as a sequence rather than a headline. The company sold the asset below its own average cost, at a realised loss, in order to retire the most expensive claim ahead of its shareholders. It also maintains a $4.65 billion dollar reserve which it describes as intended to support preferred dividends and interest on its debt — an explicit acknowledgment that those payments need funding whatever bitcoin does.

This is the mechanism the ETF comparison exists to isolate. A fund's holdings shrink when investors ask for their money back, and never otherwise. A treasury company's holdings can shrink because a dividend was due.

5. How to read a treasury company

Everything above came from filings that anyone can read for free. The routine is short.

6. The denominator: a stock is not a flow

The most common analytical error with treasury companies is reading an accumulated position as ongoing demand. They are different quantities and only one of them affects price now.

Take BitMine's August 2026 disclosure. The company reported holding 4.8% of all ETH — a very large stock, accumulated over fourteen months of weekly buying. In the week to 9 August 2026 it acquired 7,391 ETH, roughly a thousand coins a day. Over the 24 hours to 13 August 2026, Binance's three main ETH spot pairs traded about 260,000 ETH. A full week of treasury buying is therefore around 3% of a single day's volume on a single venue, and total volume across all venues is larger, so the real share is smaller still.

Both facts are true and they support different conclusions. The stock is genuine float removed from the market and it would matter enormously if it came back. The flow, at its current rate, is not what sets today's price. A headline that quotes the first to explain a move driven by the second is the error this section exists to prevent.

The discipline generalises. Whenever you see a treasury purchase reported, divide it by a current daily volume for that asset before calling it demand. Divide the total holdings by circulating supply to size the stock. Report both, and never let one stand in for the other. The same arithmetic applies when the flow turns negative: the question is not whether a treasury company is buying or selling, but how the size compares with the volume it has to clear through.

7. Conclusion

A crypto treasury company is a financing structure wrapped around a spot position, and almost everything that is interesting about it comes from the financing rather than the position. Buy the shares and you own the residual claim after convertible notes and preferred stock, which the companies state ranks ahead of you in their own filings. What looks like a simple wrapper is a leveraged one.

While the shares trade above the value of the holdings, the machine is genuinely productive: each raise leaves more coin behind every existing share, and Strategy publishes that accretion as BTC Yield rather than asserting it. But the same filings show the engine decelerating — 8.1% for the first half of 2026 against 19.7% a year earlier — while the fixed cost of the ladder built during the good years rose from $58.1 million to $629.2 million over the identical comparison. Accretion depends on the market's appetite for the shares. The dividend bill depends only on what was already issued.

When those cross, the structure does what it did in August 2026. BitMine stopped putting capital into ETH and put it into its own stock. Strategy sold bitcoin below its average cost to retire a 12% preferred series. Neither is an exotic outcome; both are the ordinary consequence of financing a non-yielding asset with instruments that demand cash.

For a reader, the practical discipline is four questions. What ranks ahead of the common stock, and what does that cost per year? Does the company publish a per-share figure, and is it rising? Is the company issuing or repurchasing its own shares? And when a purchase is announced, what is it as a share of daily volume? The filings answer all four, and the answers rarely match the headline.

Frequently asked questions

Sources and further reading

Primary sources, all on SEC EDGAR:

• Strategy Inc (then MicroStrategy) — Form 8-K filed 11 August 2020, first bitcoin purchase and same-day tender offer
• Strategy Inc — Form 10-Q for the period ended 30 June 2026 (capital structure, preferred rates, BTC Yield)
• Strategy Inc — Form 8-K filed 10 August 2026 (bitcoin sold to fund preferred repurchases)
• BitMine Immersion Technologies — Form 8-K Exhibit 99.1 filed 10 August 2026 (ETH holdings, weekly purchases, buybacks)
• BitMine Immersion Technologies — Form 10-Q for the period ended 31 May 2026 (at-the-market programme, shares sold)

Related CoinBeaver articles:

• ETF flow basics
• What moves Ethereum's price
• What actually moves crypto prices
• How the Bitcoin halving affects price
• Bitcoin dominance explained
• Crypto beta explained

This article is educational and is not financial advice. Every company figure is quoted from the SEC filings linked above and carries the as-of date that filing states. Holdings, share counts, dividend rates, and repurchase totals change continuously, and treasury companies disclose weekly, so reopen the current filing before relying on any figure here. BTC Yield and bitcoin per share are company-defined metrics rather than accounting standards, and Strategy's own filing notes that a methodology change affects comparability across measurement periods. mNAV is a market term that appears in no filing. The comparison of BitMine's weekly purchase with daily volume uses Binance's ETHUSDT, ETHUSDC, and ETHFDUSD spot pairs over the 24 hours to 13 August 2026, which is one venue among many — total ETH volume is larger, so the share of volume a week of treasury buying represents is smaller than the figure given, not larger. Nothing here is a view on any company's shares.