In the week to 3 August, Strategy sold 1,638 bitcoin for $104.73 million. In roughly the same window, BlackRock's IBIT net bought bitcoin on every single trading day, $478.5 million in total, which was about 76 percent of all inflows into US spot bitcoin exchange traded funds.
One vehicle was a net seller. The other was a net buyer every day. Both hold the same asset.
The easy reading is sentiment: institutions are losing faith in treasury companies. That is not what the numbers say. The money did not leave. It moved wrapper, and the reason it moved is legible on the liability side of one of them.
It changed vehicle.
1 What Strategy Actually Did That Week
The bitcoin sale was the headline. It was also the smallest part of the week.
Strategy sold 1,638 bitcoin for $104.73 million. It also sold 3.01 million common shares, raising $290.6 million. It repurchased 912,143 shares of its STRC preferred for $81.2 million. It added $250 million to its dollar reserve, taking that reserve to $4 billion. Total bitcoin holdings stand at 842,138, acquired for $63.51 billion at an average price of $75,419.
Set those four actions next to each other and the strategy is no longer a bitcoin strategy.
The company sold the asset and diluted the equity in the same week, and used the proceeds to retire preferred stock and build a cash pile. Our first paper argued that below an mNAV of 1.0, issuing stock stops being growth and becomes dilution. Strategy issued anyway, because something on the balance sheet has to be paid whether or not the share price cooperates.
That something is the coupon. STRC carries a 12 percent annual dividend, and the company has said it does not intend to recommend reducing the rate until the shares trade consistently near their stated $100 value. So the preferred cannot be cheapened while it trades below par, which is precisely when the company can least afford it.
2 Framework · The Cost Of The Wrapper
Both vehicles give an institution bitcoin exposure. They do not charge for it the same way, and in a drawdown they do not behave the same way.
| Leveraged treasury company | Spot ETF | |
|---|---|---|
| What the holder owns | Equity in a company that owns bitcoin | A claim on bitcoin held by a custodian |
| Price against the asset | mNAV. A premium when the story works, a discount when it does not | Tracks net asset value by construction |
| Cost of carry | Preferred dividends. STRC alone runs at 12 percent a year | An expense ratio, quoted in basis points |
| Who can force a sale | The preferred stack, indirectly, by needing to be paid | Nobody. Redemptions shrink the fund, they do not force a fire sale |
| Behaviour in a drawdown | Must service the coupon. May sell the asset to do it | Holds whatever the holders have not redeemed |
The asymmetry in the middle row is the whole story. One wrapper charges a fee. The other charges a fee that can force the manager to sell the thing you bought him for.
That was tolerable while the premium existed. An investor paying 1.5 or 2.0 times net asset value was buying leverage, an operator with conviction, and a bid that reliably showed up. The coupon was the price of the leverage and the premium paid for it. Below par, the leverage is still there and the premium is gone, so the coupon is now just cost. An institution comparing wrappers is not making a philosophical choice. It is comparing a basis point fee to a twelve percent obligation attached to an asset that has fallen.
3 What Our First Paper Got Right, And What It Missed
We published on the death spiral of the digital asset treasury in July, with mNAV at 0.62 as of 13 July. Three of its claims have now met evidence.
It said the loss of the premium is not a price move but the business model switching off. Confirmed: issuance continued anyway, into a discount, which is the definition of the engine running in reverse.
It said that in a discount regime, selling crypto becomes the rational choice. Confirmed, and repeatedly.
It said the debt was a red herring and mNAV was the body. That one needs correcting. The binding constraint is not the debt and not the mNAV by itself. It is the preferred dividend, which is neither. Debt has covenants and a maturity you can refinance. A perpetual preferred at 12 percent has no maturity to reach and, in this case, a stated intention not to cut the rate while it trades below par. It is a permanent claim on cash flow at a company with no operating cash flow.
We were also imprecise on one mechanic. We wrote that a discounted treasury would sell crypto to buy back stock. What happened is that it sold crypto and issued stock, and bought back the preferred. The direction of the reverse flywheel was right. The instrument being defended was not the common equity. It was the coupon.
4 The Other Side
Three things argue against reading this as terminal.
Strategy is not being liquidated. It holds 842,138 bitcoin against a $4 billion dollar reserve that it is still adding to, and it raised $290.6 million of equity in a week, which means a market still exists for the paper. A company with no access to capital does not sell $290.6 million of stock.
The ETF comparison is also not clean. An expense ratio and a preferred dividend are not the same kind of number. The fee is charged on assets, the coupon is charged on a slice of the capital structure, and a treasury company at a discount can in principle deliver more bitcoin per share than a fund. The leverage that is a liability at 0.68 was an asset at 2.0, and would be again.
And the IBIT figures describe one week. Daily net buying through a single week, in a market where the fear and greed reading was low, is a real datum and not a trend. We are naming a mechanism, not extrapolating a flow.
5 What Would Settle It
- Whether Strategy's bitcoin sales continue into September, and whether the proceeds keep going to the preferred rather than to the common. That allocation is the tell about who the company is now managed for.
- Whether STRC reclaims its $100 stated value. The company has tied any dividend reduction to that level, so the share price is now the gate on its own cost of capital.
- Whether the equity window stays open. Issuing into a discount is expensive and unpopular; if it stops, the only remaining source of dividend cash is the bitcoin itself.
- IBIT's daily buying beyond a single week, and whether its share of spot ETF inflows holds near the mid seventies. One week is a snapshot.
- Any other treasury company with a preferred stack showing the same pattern of asset sales funding coupon payments. One firm is a case. Three is a category.
If the pattern holds, the lesson is not that treasury companies were a bad idea. It is that a perpetual coupon is a poor thing to attach to a volatile asset and no cash flow, and that the wrapper you choose is a decision about who can force your hand.
6 Sources
Figures verified against the linked reporting. Retrieval 10 August 2026.
- CoinDesk, Strategy sold another $105 million of bitcoin last week, repurchased $81.2 million of STRC, 3 August 2026. The 1,638 bitcoin sale at $104.73 million, 3.01 million common shares raising $290.6 million, 912,143 STRC shares repurchased for $81.2 million, $250 million added to a $4 billion dollar reserve, holdings of 842,138 bitcoin at an average of $75,419, the 12 percent STRC dividend and the stated intention not to recommend a reduction until the shares trade near their $100 stated value.
- Arkham, BlackRock's IBIT buys bitcoin every day this week, totalling $479M, August 2026. Net purchases every trading day, $478.5 million for the week, roughly 76 percent of total spot bitcoin ETF inflows of $626 million, and holdings of about 742,000 bitcoin.
- OX85, The death spiral of the digital asset treasury, July 2026. The mNAV framework and the 0.62 reading as of 13 July.
On figures we did not use. Secondary coverage put IBIT's share at 81 percent of $853.5 million of inflows, which is a different measurement window from the Arkham series above. We have used one internally consistent set rather than mixing them. A fear and greed reading of 25 was attributed to the Arkham piece in material we reviewed but does not appear there, so it is excluded. A claim that this was Strategy's third sale of the year could not be confirmed in the cited filing coverage. An mNAV of 0.68 as of early August circulated in secondary sources and is not used here; the 0.62 figure we cite is the one we verified in July.
This paper is published for informational purposes and does not constitute investment, legal or tax advice. It is not a recommendation regarding any security or digital asset. Figures are as of the dates stated and may have changed.