The Weekly Take: Two-Thirds of “Corporate Bitcoin” Is One Company. In the Record Quarter, Everyone Else Was Selling.

Public companies added a record net 50,351 bitcoin in the first quarter of 2026. Strategy added at least 89,599 of it. That arithmetic makes the other 186 companies net sellers of at least 39,000 coins — inside the quarter the trade press called record accumulation. Since June, Strategy has stopped buying too.

---

The number is 1.15 million bitcoin.

It has appeared in nearly every institutional-adoption headline this quarter, sourced to Bitwise’s Q1 2026 Crypto Market Review: 187 public companies, 1.15 million BTC on balance sheets as of 31 March 2026, 5.47% of the fixed 21 million supply, and net quarterly accumulation of 50,351 BTC — the highest quarterly total on record.^[1]

Every one of those figures is accurate. The problem is the subtraction nobody ran.

Strategy alone added at least 89,599 bitcoin during the same quarter. That is not an estimate. The company’s own Q4 2025 results put its holdings at 672,500 BTC on 31 December 2025;^[2] BitcoinTreasuries recorded 762,099 on 22 March 2026.^[3] Subtract.

Now set 89,599 against the cohort’s net gain of 50,351. The remainder is not a rounding error — it is negative. The other 186 public companies were, collectively, net sellers of at least 39,000 bitcoin during the quarter the industry has spent five months describing as record corporate accumulation.

Publicly listed miners account for most of the shortfall. They sold more than 32,000 BTC gross in Q1 2026 — more than in all of 2025 combined — MARA Holdings alone accounting for 20,880 of that against roughly 2,247 BTC of its own production in the quarter.^[4][5] Gross matters: miners mine, so their net drawdown is smaller than the sales figure. It is still deeply negative.

So the record is real, and it is one company’s record. On 22 March 2026, Strategy held 762,099 bitcoin against the 31 March aggregate of 1.15 million — roughly 66%, two-thirds, held by one issuer.^[3] (The two figures carry dates nine days apart because the tracker and the report publish on different cycles; the ratio is stable across the gap, and any Strategy buying in those nine days only makes the subtraction above more negative.)

What the trade press has been calling a corporate accumulation trend is a Strategy accumulation trend with the rest of the field netting out against it.

That much was true in March. What has happened since is that the one company holding the trend up stopped buying.

---

The buyer stopped

Strategy’s last bitcoin purchase was 520 BTC during the week of 15–21 June 2026, at an average of $67,068 per coin.^[6] It has not bought since — more than two months, and the longest such run since the 2022 bear market.

Then the pause became something else. By late July, Strategy had sold approximately 3,620 BTC and authorized further sales to support dollar reserves.^[7] It kept going. In the week of 27 July–2 August it sold 1,638 BTC at roughly $63,957, taking holdings from 843,775 to 842,138 and directing the $104.7 million to preferred distributions and STRC repurchases.^[8] In the week of 3–9 August it sold 1,690 more at an average of $64,262 — roughly $11,000 below its own cost — putting the $108.6 million into 1,152,020 shares of repurchased STRC and taking holdings to 840,447. In the same week it raised $653.1 million selling MSTR common and moved $650 million of that into reserves.^[9]

Three consecutive weeks of selling bitcoin at a loss to buy back preferred stock is not a pause in the accumulation model. It is the model running backwards.

The two August weeks that followed state plainly what the model now produces.

Week of 10–16 August: 3,458,866 MSTR shares sold for $333.7 million, allocated to STRC dividends ($52.4M), preferred repurchases ($132.2M), and dollar reserves ($149.1M, bringing the reserve to $4.8 billion). Bitcoin bought or sold: none.^[10]

Week of 17–23 August: 18.26 million shares sold for approximately $2 billion, allocated to STRC repurchases ($136.4M), reserves ($300M, reserve now $5.1 billion), and a new $1.59 billion “USD Cash” pool the company says may later go toward “acquiring bitcoin, paying declared cash dividends” and other treasury purposes. Bitcoin bought: none.^[11]

Two weeks, $2.33 billion raised against the stock of a bitcoin holding company, zero bitcoin bought.

---

What broke the accumulation engine

The Digital Asset Treasury model, of which Strategy is the exemplar and much of the cohort is a copy, has three moving parts.

A public company issues equity — usually an at-the-market program selling stock continuously in small blocks — or convertible debt. The proceeds buy bitcoin. Because the market values the company above the bitcoin it holds, a share representing one dollar of bitcoin sells for more than one dollar of cash. That premium is what makes the next round accretive: sell the share, buy more bitcoin than the share represented, and bitcoin-per-share rises even as share count does. Repeat.

Everything depends on the premium persisting. When a DAT company’s stock trades at or below the net asset value of the bitcoin it holds, at-the-market issuance stops adding bitcoin-per-share and starts subtracting it. The rational move then flips: instead of selling stock to buy coin, you sell coin to buy back stock.

That flip is what the disclosures show. Sequans sold 1,025 BTC, then nearly 80% of what remained, to repay convertible debt. Satsuma liquidated all 668 of its coins and its shareholders approved delisting from the London Stock Exchange. Nakamoto sold about 284 BTC for working capital, with nearly 70% of its remaining 5,342 BTC pledged against a Kraken loan maturing in December. Empery Digital sold nearly half its holdings to fund buybacks and debt repayment. Smarter Web sold 178 BTC to retire a convertible, its CEO saying the coins do not “represent the right capital solution.”^[7] Strategy’s August weeks are the same trade at a different scale.

Miners arrived by a different road. MARA Holdings — the second-largest corporate holder — sold 23,093 BTC for approximately $1.63 billion in the first half of 2026 at an average of $70,631, taking its treasury from 53,822 BTC at the end of 2025 to 35,577 at the end of June. It mined roughly 4,669 BTC over the same six months, which is why the treasury fell by 18,245 rather than by the full amount sold.^[5] That followed a policy change in its 10-K filed 3 March 2026 permitting sales from the balance sheet, not only of newly mined coin.^[12] Bitdeer went further, liquidating its entire treasury by late February — roughly 2,000 BTC at the start of the year, 1,530 by the end of January, 943 by 13 February, zero days later — and redirecting capital to AI and high-performance computing, which earns more per megawatt than mining at current margins.^[13] Riot has been a seller since at least April; so, outside the corporate cohort, has the sovereign holder Bhutan.^[14]

The engine did not fail catastrophically. It failed one 10-Q at a time, while the aggregate held.

---

What this does not show

We are not calling a top. Strategy holding cash for two months is a fact about how one company allocates capital when its premium to NAV compresses, not a forecast of bitcoin’s price. Conflating the two is the error this piece exists to avoid.

The record is real. As of 31 March 2026, 1.15 million bitcoin across 187 public companies was accurate, and net Q1 accumulation of 50,351 BTC was a genuine high. Our argument is about composition, not veracity.

Miner selling is substantially operational, and the figure is gross. Miners sell bitcoin to pay for electricity, hardware, and debt service, and much of MARA’s H1 disposition went to retiring convertibles and funding infrastructure. They also produce coins continuously, so the 32,000 BTC of Q1 sales overstates their net drawdown by whatever they mined that quarter — a few thousand coins for the largest operators. The fact of miner selling is ordinary. What is new is the March 2026 language permitting balance-sheet sales at management’s discretion — a governance change, not a cash-flow event.^[12]

Strategy is not distressed. It holds $5.1 billion in dollar reserves, a $1.59 billion cash pool, and 840,447 BTC against an average cost of $75,385.^[11] With bitcoin just above $79,000 it is modestly above water on the aggregate position — though it was carrying roughly $10 billion in unrealized losses as recently as 17 August, before the rally.^[10][15] Nothing in the August filings reads as capitulation. They read as what they say: weeks in which the model produced cash instead of coin.

Nine companies is not 187. We have named nine sellers — MARA, Bitdeer, Satsuma, Sequans, Nakamoto, Empery Digital, Smarter Web, Riot, and Strategy. That leaves 178 we have not examined. The Q1 arithmetic establishes that the non-Strategy field was a net seller in total; it does not establish that any particular unnamed company sold. Reading the 178 as sellers would repeat, in the opposite direction, exactly the error the headline number makes.

None of this is investment advice. It is a reading of public filings, a public tracker, and a published research report. The dates are attached so you can check it.

---

What would change our mind

Strategy deploying the “USD Cash” pool into bitcoin. The company created a $1.59 billion pool and said in the filing it may be used to acquire bitcoin.^[11] If September or October converts that pool to coin at anything near the 2024–2025 cadence, the pause becomes an interval rather than a turn. The Q1 concentration arithmetic would survive it. The trajectory claim would not.

Miner treasury policy reverting. If MARA, Riot, or another holder that loosened its policy in 2025–2026 files a walkback — language committing to hold newly mined bitcoin as a long-term investment — that is a governance signal in the other direction. We would want the filing, not the post.

Durable premium re-expansion at DAT companies. If MSTR or the smaller Nakamoto/Sequans/Empery cohort trades back to a meaningful premium to the NAV of bitcoin held, at-the-market issuance becomes accretive again and today’s sellers become tomorrow’s buyers. This is the clean disproof of the mechanism section, and we would rather name it than not.

Bitcoin sustaining above roughly $75,400. That is Strategy’s average cost, and the level at which the position stops being a drag on the equity story. Bitcoin has crossed back above it in the last several sessions — briefly touching $81,000 this week — after Strategy was selling near $64,000 at the start of August.^[15] If it holds there, the pressure that produced the loss sales eases, and the case for continued selling gets materially weaker.

---

Where this leaves us

The trade press has an aggregation problem. We made a version of this argument three weeks ago about a different number nobody can check; this one is checkable, and that is exactly the problem. A single figure combining 187 companies is a useful headline. But when two-thirds of the figure is one issuer, and that issuer supplied more than Q1’s entire net gain, the total is not describing the cohort. It is describing the outlier, and reporting the rest as though they came along.

Between 2024 and early 2026 the outlier and the field moved together, and the number worked. They no longer move together. Miners have sold to service debt and have rewritten their policies to permit selling more. DAT companies have unwound to buy back shares, repay converts, or delist. And since June, the outlier has bought nothing and sold at a loss twice.

A record quarter of 50,351 coins. One company supplying 89,599 of them. Everyone else, net, short by at least 39,000. That is the composition inside the record, and it is the arithmetic we would quote instead.

---

Sources

Holdings, price, and filing figures pulled 27 August 2026 and stated as of the dates given; published 28 August 2026. Where a figure carries a different as-of date than the aggregate it is compared against, the gap is stated in the text.

1. ↩ Bitwise, Q1 2026 Crypto Market Review: 187 public companies holding 1.15 million BTC as of 31 March 2026 (5.47% of the 21 million supply); net Q1 accumulation of 50,351 BTC. Summarized in The CC Press and Cryptopolitan.
2. ↩ Strategy, “Strategy Announces Fourth Quarter 2025 Financial Results,” 5 February 2026: 672,500 BTC held as of 31 December 2025 at an average cost of approximately $74,997; 713,502 BTC as of 1 February 2026, including 41,002 BTC acquired in January.
3. ↩ BitcoinTreasuries.net public-companies leaderboard, 22 March 2026 snapshot: Strategy 762,099 BTC — approximately 66% of the 31 March aggregate in note 1. Dates differ by nine days; see text.
4. ↩ Cryptopolitan, “Corporate Bitcoin Holdings Hit 1.15M BTC in Q1”: “publicly listed miners collectively sold more than 32,000 BTC in Q1 2026, surpassing total miner sales for all of 2025,” and “Strategy added roughly 89,000 BTC in Q1 alone” — the latter independently reproducible from notes 2 and 3. Miner figures throughout are gross sales, not net of production. The implied non-Strategy net (50,351 − 89,599 = −39,248) is our arithmetic, not any source’s.
5. ↩ MARA Holdings Q2 2026 results: 23,093 BTC sold in H1 2026 for approximately $1.63 billion at an average of $70,631; treasury 53,822 BTC (31 Dec 2025) to 35,577 BTC (30 June 2026). Blockspace / Yahoo Finance. Q1 2026 sales of 20,880 BTC per Crypto Briefing. Production of 2,247 BTC in Q1 and 2,422 BTC in Q2 (4,669 H1) per Coinpedia’s H1 2026 mining review, which also puts MARA and Riot’s combined H1 gross sales at 32,758 BTC.
6. ↩ TheStreet, 5 August 2026. Last purchase 520 BTC during the week of 15–21 June 2026 at an average of $67,068; prior comparable pause during the 2022 bear market. Outlets differ on the week count — Benzinga put the drought at seven weeks on 17 August, dating the last buy to “early July” — so this piece states the elapsed time from the specific disclosed purchase rather than adopting either count.
7. ↩ CoinDesk, “Bitcoin treasury companies sell up, repay debt, pivot to AI as share prices collapse,” 24 July 2026. Satsuma 668 BTC (all); Sequans 1,025 BTC then ~80% of the remainder; Nakamoto ~284 BTC; Empery Digital ~half; Smarter Web 178 BTC; Strategy ~3,620 BTC with further sales authorized.
8. ↩ The Block, 3 August 2026. Week of 27 July–2 August 2026: 1,638 BTC sold at approximately $63,957 for about $104.7 million, applied to preferred distributions and STRC repurchases; holdings 843,775 to 842,138 BTC.
9. ↩ CoinDesk, “Strategy sells 1,690 bitcoin, raises $653 million from MSTR shares,” 10 August 2026. 1,690 BTC sold at an average of $64,262 for $108.6M, used to repurchase 1,152,020 STRC shares; holdings 842,137 to 840,447 BTC.
10. ↩ The Block, 17 August 2026. Week of 10–16 August 2026: 3,458,866 MSTR shares sold for $333.7M; $52.4M STRC dividends, $132.2M preferred repurchases, $149.1M to reserves (total $4.8B); no bitcoin bought or sold; unrealized losses then approximately $10B.
11. ↩ Strategy Inc. Form 8-K filed 24 August 2026, SEC EDGAR accession 0001193125-26-361845, covering 17–23 August 2026: 18,261,118 MSTR shares sold for $2,006.5 million net; $136.4M to STRC repurchases, $300.0M to the USD Reserve, remainder to USD Cash; no bitcoin purchased or sold; holdings 840,447 BTC at an average of $75,385 (aggregate $63.36 billion); USD Reserve $5.10 billion and USD Cash $1.59 billion as of 23 August. SEC EDGAR; coverage via The Block.
12. ↩ MARA Holdings 10-K filed 3 March 2026, treasury-policy section, permitting sales from the balance sheet at management’s discretion. Forbes, 4 March 2026.
13. ↩ Yahoo Finance, “Bitdeer Liquidates Entire Bitcoin Treasury as Mining Margins Tighten”. Roughly 2,000 BTC at the start of 2026; 1,530 BTC by the end of January; 943.1 BTC on 13 February; fully liquidated days later (184 BTC mined 21 February plus the 943.1 BTC reserve, sold with bitcoin between $65,000 and $68,000). Capital redirected to AI and HPC infrastructure.
14. ↩ CoinDesk, “The Bitcoin treasury boom is unwinding as some companies and governments sell holdings,” 2 April 2026. Names Riot and Bhutan as early sellers.
15. ↩ Bitcoin settled at $79,027 on 26 August 2026 (Rio Times); $79,317 at 8:50 a.m. ET on 27 August 2026, up 0.6%, after briefly touching about $81,000 earlier in the week (Yahoo Finance); all-time high of $126,198 recorded October 2025 (Coinlaw).

This article is not investment advice. It describes public filings, a public tracker, and a published research report. Comments are closed sitewide. Corrections to corrections@coinagereport.com.