The Weekly Take: Only Four of Twenty Crypto Treasuries Trade Above Their Coins. Metaplanet Shows How the Premium Gets Eaten.

The digital asset treasury model was always premium-dependent. When the premium goes, the mechanism doesn’t just stop working. It reverses. Metaplanet’s mNAV, measured on enterprise value, fell to 0.86 on 23 September, from 1.53 at the twelve-month peak, while an executive stock option pool auto-expanded from 46 million potential shares to 319 million. The specific arithmetic of how one instrument ate the premium is legible in a single filing. The industry-wide arithmetic is legible in one sentence from a DWF report published this week: only four of twenty.

Jason Jones, Editor in Chief · 25 September 2026 · 12 min read

---

Last month, this column counted the corporate bitcoin buyers and found the field consisted, at the level that matters, of one company and everybody else. Strategy added 89,000 BTC in the record quarter; the rest of the public market, in aggregate, sold 38,600. The point of the piece was not that Strategy is winning — Strategy has always been winning — but that “corporate bitcoin buying” as a market narrative had begun to describe the marketing of one specific company and to obscure what everyone else was actually doing.

The follow-up question that piece left open is what happens to the “everyone else.” Not why they aren’t buying — that part was answered by the composition data — but what the mechanism they were using looks like when it stops working. This week supplied the answer, in two parts.

Part one, industry-wide. DWF Ventures published a survey of the twenty largest crypto treasury companies on 21 September. Only four of those twenty currently trade at a premium to the market value of the crypto on their balance sheets. Sixteen trade at a discount.¹ The digital asset treasury model — the DAT model, as its practitioners call it — is a premium-dependent mechanism. Four out of twenty is not a market signal about individual companies. It is a market signal about the model.

Part two, one company. Metaplanet’s mNAV — the ratio of its enterprise value, which counts debt and preferred equity as well as common stock, to the market value of its bitcoin holdings — closed at 0.86 on 23 September.² Twelve months ago it was 1.53; the tracker’s twelve-month range runs from 0.70 to 1.53. The company holds 43,000 bitcoin worth roughly $3.63 billion and its fully-diluted market cap is $2.53 billion.² On market cap alone, then, the common equity trades at 0.70 times its bitcoin; the 0.86 figure is the one that also counts the claims ranking ahead of the common shareholders. The stock is down about 87% from its 1,895-yen peak of last year.² The specific instrument that produced the compression is not the bitcoin price. It is an executive stock option pool that automatically expanded from 46 million potential shares to 319,464,000 potential shares over roughly two years of equity issuance, and the CEO exercised 92,000 of the resulting rights on 28 August 2026 for 64.03 million shares at ¥10 apiece.³

Both parts are the same story. The DAT model requires a specific arithmetic to work. When the arithmetic breaks, it can break company-by-company for company-specific reasons, and it can also break industry-wide because everyone was arithmetically dependent on the same thing.

How the DAT model was supposed to work

The mechanism, stripped to its essentials, is this. A public company issues shares to raise cash. It spends the cash on bitcoin. If the company trades at a premium to the bitcoin it holds — say, at a market cap of 1.5x the market value of its bitcoin — then every share it issues raises more cash than a corresponding fraction of its bitcoin is worth. It can then buy more bitcoin than it “gave up” in dilution. Each cycle raises the bitcoin-per-share number for existing holders. The premium finances itself.

This is a real arbitrage while the premium exists. It is not a magic trick. The premium exists because certain classes of investors — retirement accounts constrained from holding crypto directly, funds with prospectus limits on digital assets, buyers who prefer a familiar equity wrapper — will pay more for indirect bitcoin exposure than for direct. The premium’s size is the market price of that access constraint. Over the past twelve months Strategy’s mNAV has ranged from 0.95 to 1.43;⁴ much of its accumulation happened earlier, at higher premiums. Toward the top of that range, the raise-issue-buy cycle still worked.

What the DAT model does not survive is the premium going away. At exactly mNAV of 1.0, the arbitrage is zero — every dollar raised buys exactly a dollar of bitcoin, and the whole point of using a corporate wrapper disappears. Below 1.0, the mechanism reverses. Issuing shares becomes destructive: the cash raised buys less bitcoin, per share, than the shares diluted. Every raise a company at 0.86 mNAV attempts makes existing shareholders’ bitcoin-per-share smaller, not larger. This is why Metaplanet’s own capital allocation policy, adopted in October 2025, commits the company to no new common-equity issuance while mNAV is below 1x.⁵ They know.

The buyback is the theoretical repair — buying back shares below NAV concentrates bitcoin-per-share. Metaplanet is publicly weighing this.⁶ So is much of the field, in various stages of formality. But the buyback requires cash the company does not have from operations — most of these companies have negligible operating income — and paying for it by selling bitcoin defeats the entire premise of the treasury strategy. The best structural fix for a below-NAV DAT is a cash flow the DAT does not have.

What happened at Metaplanet specifically

The story of how Metaplanet’s premium was consumed is unusually well-documented, because the mechanism was an equity plan filed years earlier that behaved exactly as its written terms said it would, in a direction nobody paid attention to until the numbers arrived at the surface.

The Series 10 stock acquisition rights were originally issued on 28 December 2022 and approved by shareholders on 7 February 2023, with a strike price of ¥10 per share, staged vesting, and — this is the term that did the work — an anti-dilution adjustment clause that expanded the pool automatically to preserve the executives’ 20% fully-diluted equity share as the company issued new stock.⁷

That clause behaved reasonably in a company that issues stock occasionally. In a company that issues stock as its capital-formation strategy — Metaplanet raised equity repeatedly through 2025 to buy bitcoin — the clause turned into a compounding claim on future shares. The original 46 million potential shares became, by mid-August 2026, 319,464,000 potential shares, all at a strike price of ¥10.⁷ The average shareholder buying the stock in that window did not read filings closely enough to notice that a rising fraction of every equity raise was accruing to the option pool rather than to their bitcoin-per-share. Then on 18 August 2026 the company froze the pool at 319.5 million shares and imposed a five-year lock-up through August 2031.⁷ Ten days later, on 28 August, CEO Simon Gerovich exercised 92,000 vested rights, receiving 64,032,000 new shares at ¥10 each, and his personal stake rose from about 15.5 million shares to 79.6 million.⁷

The stock reacted the way you would expect it to. Shareholder activist accounts posting as “Ragnar” and “The Bitcoin Pharaoh” demanded a complete rollback of the extra rights, rejecting the August freeze as insufficient.⁸ On 6 September, Gerovich publicly acknowledged that the company had “not done a good enough job of explaining” its structure and pledged “clearer disclosure.”⁸ On 11 September, the board announced a 41% reduction in the pool, cutting it from 319.46 million to 188.19 million potential shares, restructuring each Series 10 unit to cover 410 shares instead of 696, and pushing vesting for unvested rights out to equal tranches in 2029, 2030 and 2031.⁷ Gerovich stated the reset “extinguishes over $220 million of warrant value” and increases bitcoin per fully-diluted share by approximately 8.8%.⁷

That 8.8% did not reverse the mNAV. The stock kept falling. On 8 September, when the pool cut was still being negotiated, mNAV was 0.97.⁹ On 23 September, twelve days after the cut, mNAV was 0.86.² What that says is that the option pool was one instrument among several, and cleaning it up meaningfully was not, at this point, enough. The market re-priced Metaplanet not because of the specific 319 million share pool but because the pool made visible a general question about how many other terms of Metaplanet’s capital structure worked the way this one did.

Metaplanet is, on the holdings numbers alone, one of the two or three most successful bitcoin treasury companies in the world. 43,000 BTC is a real position. It sits third in the entire public market by holdings, behind Strategy and — on Twenty One’s last-reported figure — Twenty One Capital. The reason its stock trades below the value of that bitcoin is not that the bitcoin isn’t there. The reason is that the market has decided, empirically and specifically, that the equity wrapper around the bitcoin is worth less than nothing at the current terms.

Strategy is fine. Sort of.

Michael Saylor’s Strategy is not Metaplanet, and one of the useful pieces of arithmetic in this story is showing how much difference specifically that makes.

Strategy holds 846,000 BTC — 65.6% of all corporate bitcoin held by public companies — worth roughly $71.4 billion at current prices.¹⁰ Its mNAV as of this week is 1.09.¹⁰ Its twelve-month range is 0.95 to 1.43, meaning it has already crossed the 1.0 threshold once in the past year — in late June 2026, when CoinDesk reported enterprise value of about $50.4 billion against about $51.1 billion in bitcoin, briefly worth less than the coins it held.¹¹ It recovered.

The recovery matters, because it is the specific piece of evidence for the argument that below-1.0x mNAV is not necessarily terminal. Strategy has a set of things Metaplanet does not: a full-time treasury operation that pre-existed the bitcoin strategy, a convertible-debt structure with more sophisticated conversion terms than Metaplanet’s, a shareholder base that has held through prior sub-NAV episodes, and the size effect — Strategy is the DAT everyone else is imitating, with the deepest market for its securities. Size does not guarantee the higher multiple, though: at its peak, Metaplanet’s 1.53 was above anything Strategy printed in the same twelve months.

We have not found an equivalent of the Series 10 pool in Strategy’s capital structure — an instrument doing damage to the equity while the bitcoin position held steady. When mNAV compressed at Strategy in June, it did so alongside a premium compressing across the sector. That is a survivable form of compression, in the sense that when the sector re-rates, Strategy re-rates with it. What is happening at Metaplanet has an additional layer: the sector-wide compression plus company-specific instruments accelerating the discount.

Twenty One Capital, the second-largest holder at 43,514 BTC — a figure it last reported in July 2025 — tells the third version of the story. Its basic mNAV — market cap over bitcoin value — is 0.63.¹² On enterprise value of $2.7 billion, which counts the bondholders’ claim, it is about 0.74. Its fully-diluted mNAV, counting the shares that would exist if all convertibles converted, is 1.19.¹² The company therefore trades at both a large discount and a modest premium at the same time, depending on which shares you count. This is not a paradox. It is the specific arithmetic of a company whose capital structure is dominated by convertible instruments that have not yet converted. Twenty One’s CEO Raphael Zagury said on 22 July 2026 at Mining Disrupt in Miami that “mNAV multiples across the sector should trend toward 1x,” and that this convergence is “basic finance” — arbitrage narrows when trades become crowded.¹³ Zagury runs one of the DATs, and he said on the record, at a public event, that the mechanism is normalizing.

Why the discount exists across the field, mechanically

Set aside Metaplanet’s specific option pool. Below-NAV pricing at treasury companies has, in the general case, four mechanical causes that keep showing up.¹⁴

Convertible debt sits ahead of equity. A treasury company with $500 million in convertible bonds outstanding has $500 million of bitcoin claim that belongs to the bondholders, not the shareholders, until conversion. Shareholders looking at “bitcoin per share” have to net out the bondholder claim first. In mNAV terms, this is why basic, enterprise and fully-diluted ratios diverge, and the divergence can be very large — Twenty One’s basic and fully-diluted figures are 0.56 turns of mNAV apart.

Accounting standard ASU 2023-08 puts unrealized bitcoin volatility straight through the P&L. Since fiscal 2025 (earlier for companies that adopted early), U.S. GAAP filers mark bitcoin at fair value each quarter and run the difference through earnings. In a quarter where bitcoin falls, this produces large reported losses on essentially unchanged holdings, which can put the stock outside the mandate of institutional buyers who screen out companies with net losses, regardless of the reason for the loss. This one applies to the U.S.-listed names; Metaplanet reports under Japanese GAAP.

Convertible dilution timing is adverse. Convertibles convert when the stock price rises above the strike. That is by design, and it is the mechanism convertible investors use to participate in upside. It also means the bitcoin-per-share number falls precisely when the share price is rising — the moment shareholders would most want their bitcoin per share to be high, the conversion cuts it. Sophisticated equity investors read this on the way in and price it as a discount.

Corporate overhead is a claim on the bitcoin. A treasury company has a board, a compliance function, a listing fee, an audit, and operating expenses. Direct bitcoin ownership has none of these. Every dollar spent by a treasury company on operations is a dollar the shareholders’ bitcoin-per-share does not include. Over a five-year holding period, the compounding effect of this overhead is not small.

None of these four mechanisms is a fraud. They are all disclosed. They are also all reasons a rational investor might value the equity wrapper at less than the bitcoin inside it, particularly if the investor now has better ways to hold bitcoin directly — spot ETFs at low expense ratios, custodial products through familiar prime brokers, and, since the SEC’s proposed Regulation Crypto Assets in August,¹⁶ a regulatory framework moving toward treating certain crypto assets as non-securities entirely. The bitcoin ETF wrapper competes for the same “regulated bitcoin exposure” dollars that used to have to go through a treasury company. That competition, on the DWF numbers, is now visibly winning against sixteen of the top twenty.

What this does not show

mNAV is one number and companies have many. A single ratio does not capture operating income, hash rate for miners, staking yield for Ethereum treasuries, or the option value of a management team that can pivot. Metaplanet at 0.86 does not mean Metaplanet is worthless. It means the equity trades at a specific implied haircut to the bitcoin, and the haircut is a market price for a set of concerns we have named but not exhaustively priced.

“Only four of twenty” is a snapshot on one date. The DWF report reflects prices as of 21 September 2026. The threshold below which a company is a “discount” is 1.0x, which is an arbitrary line — a company at 1.02 and a company at 0.98 look identical in the count, and very different in prospect. What the four-of-twenty count does say is that the modal treasury company is now on the wrong side of 1.0.

The Series 10 story is Metaplanet-specific and not a general indictment. Other treasury companies have executive equity plans, and the specific mechanism that automatically expanded Metaplanet’s pool is uncommon in the field. Reading “the DAT model has a Series 10 problem” from this piece would be reading too much into it. The reading we intend is that Metaplanet’s option pool made visible, in a single filing, a category of capital-structure risk that exists in other companies with other instruments, and the market is now doing the work of finding those instruments.

Strategy’s recovery from below-1.0 in June is not a promise that Metaplanet or Twenty One will recover. Companies exit below-NAV states through mechanisms that are legitimate but not automatic — buybacks funded by external capital raises, operating income growth, redemption of debt at par, tender offers. Metaplanet has publicly weighed buybacks. It has not, as of publication, announced one at scale. Twenty One has more convertible mechanics available and a different starting point.

Twenty One’s 0.63 basic and 1.19 diluted mNAVs are both real numbers, and they measure different things. If none of its convertibles convert, the common equity trades at 0.63 of its bitcoin on market cap, or about 0.74 once the bondholders’ claim is counted in enterprise value. If they all convert, it trades at 1.19. The market is currently pricing somewhere between these, and neither number is wrong. Treating either as the mNAV would be dishonest.

We have not counted every mechanical or operational cause of the discount. Insurance and custody costs for large bitcoin positions, jurisdictional tax treatment, regulatory overhead differences between the U.S., Japan, and elsewhere, and market-specific factors (Metaplanet trades in Tokyo, Twenty One on NYSE, Strategy on Nasdaq) all affect the discount in ways we have not broken out. Someone writing a longer piece would.

None of this is investment advice. It is a reading of public capital structure filings, one industry survey (DWF, 21 September), and public price data as of 23 September 2026, with the arithmetic shown.

What would change our mind

A wave of shareholder-authorized buybacks funded at scale. The theoretical repair for a below-NAV DAT is real and mechanically available: raise external capital (private placement, preferred equity, debt) and use it to retire shares while below NAV. If several of the sixteen discount DATs execute this over the next quarter, mNAV compression becomes a solvable problem and the four-of-twenty count improves.

Regulation Crypto Assets not being finalized. If the SEC’s proposal (comment period closing 20 October 2026) is withdrawn or heavily amended, the ETF wrapper competitive threat to DATs weakens, and part of the compression pressure eases. This is not our forecast — the political direction is toward finalization — but it is the clean disproof of the “ETFs are eating DATs” version of the argument.

Twenty One’s diluted mNAV holding above 1.0 as conversions actually occur. If Twenty One’s convertibles convert and the resulting fully-diluted market cap stays above the bitcoin position, the diluted figure becomes the operative number and the discount reading of Twenty One is wrong. Watch conversions over the next two quarters.

A sustained bitcoin price rally re-establishing premiums. DAT premiums historically expand in strong bull markets — retail flow into equity wrappers rises, and the multiple expansion re-opens the raise-issue-buy arbitrage. A sustained move to $120K+ bitcoin would likely restore mNAV premiums across the field, regardless of the mechanical arguments in this piece. It would not fix Metaplanet’s Series 10 problem, but it would raise the ratio.

Where this leaves us

The DAT model is not dead. Strategy operates at 1.09x mNAV and continues to accumulate bitcoin under mechanisms that work at current multiples. Twenty One trades at a discount or a premium depending on which shares you count, and which count turns out to be right depends on conversions that have not happened yet. Sixteen of the top twenty now trade at a discount, and the premium they once carried did not survive the arrival of better bitcoin access elsewhere; the modal treasury company will now trade closer to 1.0 as a matter of arbitrage rather than sentiment. Zagury said as much in July.

What is dead — what died in the specific arithmetic of Metaplanet’s Series 10 pool and what the DWF survey documented at industry scale — is the assumption that the premium was durable. It never was. It was a market price for an access constraint, and the constraint is loosening. Companies with capital structures that assumed the premium would persist have a mechanical problem now. Companies with capital structures that did not, don’t.

The follow-up question that will define the next year of this story is whether the four companies still above 1.0 hold, and if not, whether the sector can converge on the industrial-corporate-treasury discipline (income, buybacks, disciplined capital allocation, no auto-expanding option pools) that keeps a good business trading above the sum of its parts. That discipline is available. It is not the DAT playbook.

Forty-three thousand bitcoin. $3.63 billion of bitcoin. A $2.53 billion market cap — 0.70 on market cap, 0.86 on enterprise value. Four of twenty. Metaplanet is a real story. It is also a legible warning.

---

Sources

All figures pulled 24 September 2026 and stated as of that date. Prices and ratios move.

1. DWF Ventures, treasury company survey published 21 September 2026 and analyzed in crypto.news, 24 September 2026: only four of the top twenty crypto treasury companies trade at a premium (above 1.0x mNAV) to the market value of their crypto holdings. The report attributes the compression to improved regulatory access to direct crypto (spot ETFs, custodial products) and to greater investor weight on “operators, financing terms, and business income” rather than asset holdings alone. ↩
2. Metaplanet (3350.T) mNAV of 0.86 (enterprise-value basis) as of close on 23 September 2026, with 43,000 BTC held and total bitcoin value of $3.63 billion, share price ¥246, fully diluted market cap $2.53 billion, twelve-month mNAV range 0.70 to 1.53. Source: live mNAV tracker (mnav.com/mnav/metaplanet), accessed 24 September 2026. Market-cap-to-bitcoin ratio of 0.70 is our calculation from the tracker’s figures ($2.53 billion ÷ $3.63 billion). ↩
3. Metaplanet Series 10 stock acquisition rights: original executive pool of approximately 46 million potential shares expanded to 319,464,000 potential shares under an anti-dilution adjustment clause as the company issued equity to fund bitcoin purchases. On 18 August 2026 the company froze the pool at 319.5 million shares and imposed a five-year lock-up through 17 August 2031. On 28 August 2026, CEO Simon Gerovich exercised 92,000 vested rights (696 shares per right) and received 64,032,000 common shares at a strike of ¥10 each, for ¥640 million. As reported by ad-hoc-news and bitcoinethereumnews, 5–8 September 2026. ↩
4. Strategy (MSTR) twelve-month mNAV range of 0.95x to 1.43x. Source: mnav.com/mnav/strategy, accessed 24 September 2026, describing current mNAV of 1.09x as “near the middle of that historical range.” ↩
5. Metaplanet capital allocation policy adopted October 2025 and revised March 2026: “Metaplanet commits to no new common-equity issuance while mNAV is below 1x, targeting buybacks instead to lift BTC holdings per share.” As reported by coinotag, 8 September 2026. ↩
6. Metaplanet publicly weighing share buyback to boost bitcoin-per-share yield with mNAV below 1.0x. Reported by KuCoin and others, September 2026. ↩
7. Timeline of Metaplanet Series 10 pool: original issuance 28 December 2022, shareholder approval 7 February 2023, ¥10 strike price, staged vesting, anti-dilution adjustment clause. 18 August 2026 amendment fixed pool at 319,464,000 shares and imposed five-year lock-up. 28 August 2026 CEO exercise as noted in note 3. 6 September 2026 CEO public acknowledgment. 11 September 2026 board 41% reduction from 319.46 million to 188.19 million potential shares, each Series 10 unit restructured from 696 shares to 410 shares, unvested rights vesting in three equal tranches in 2029, 2030 and 2031. Gerovich statement: reset “extinguishes over $220 million of warrant value” and increases bitcoin per fully-diluted share by approximately 8.8%. As reported by bitcointreasuries.net, cryptotimes and crypto.news, 11–12 September 2026. ↩
8. Shareholder activist accounts posting as “Ragnar” and “The Bitcoin Pharaoh” demanded complete rollback of the Series 10 rights, rejecting the 18 August freeze as insufficient. CEO Simon Gerovich statement of 6 September 2026 acknowledged Metaplanet had “not done a good enough job of explaining” its structure and pledged “clearer disclosure.” As reported by cryptobriefing and cryptotimes, 6 September 2026. ↩
9. Metaplanet mNAV of 0.97x on 8 September 2026, with market cap ¥312.6 billion, bitcoin value ~¥516 billion, simple market-cap-to-BTC ratio 0.6x. As reported by coinotag, 8 September 2026. ↩
10. Strategy (MSTR): 846,000 BTC held, mNAV 1.09x, total bitcoin value $71.4 billion, trading at approximately 9% premium to bitcoin held. Source: mnav.com/mnav/strategy, accessed 24 September 2026. Corporate share of public-company bitcoin (65.6%) computed against 1,289,527 BTC total public-company holdings; see note 15. ↩
11. Strategy’s mNAV first fell below 1.0x in late June 2026, with enterprise value approximately $50.4 billion against bitcoin holdings of approximately $51.1 billion. Reported by CoinDesk, 27 June 2026. ↩
12. Twenty One Capital (XXI): 43,514 BTC held (as of 30 July 2025 last-reported date), market cap $2.3 billion, enterprise value $2.7 billion, basic mNAV 0.63x, diluted mNAV 1.19x, BTC per share (basic) 0.000126, BTC per share (diluted) 0.000067. Source: bitcointreasuries.net, accessed 24 September 2026. Enterprise-value ratio of about 0.74 is our calculation from these figures. ↩
13. Raphael Zagury, CEO of Twenty One Capital, at Mining Disrupt 2026 in Miami, 22 July 2026: “mNAV multiples across the sector should trend toward 1x.” Reported by tftc.io, 29 July 2026. Zagury characterized the convergence as “basic finance” — arbitrage narrows when trades become crowded. ↩
14. The four mechanical causes of treasury-stock discount to bitcoin — convertible debt seniority, ASU 2023-08 accounting volatility running through P&L, convertible dilution timing adverse to shareholders, and operating overhead as a claim on bitcoin holdings — are our framing; see also cryptoticker’s analysis of the mNAV discount, accessed September 2026. ↩
15. 179 public companies across 30 countries hold 1,289,527 BTC on their balance sheets as of 23 September 2026, representing approximately 6.1% of total bitcoin supply worth approximately $111 billion. Top ten by holdings: Strategy 846,000; Twenty One 43,514; Metaplanet 43,000; MARA 35,303; Bitcoin Standard Treasury 30,021; Galaxy Digital 25,723; Strive 24,999; Bullish 23,300; Riot 15,680; Hut 8 13,696. Source: satsintel.io, accessed 24 September 2026. ↩
16. U.S. Securities and Exchange Commission, Regulation Crypto Assets, proposed rule, Release No. 33-11434; press release 2026-76; published in the Federal Register 21 August 2026. Comment period closes 20 October 2026. ↩

This article is not investment advice. It describes public capital-structure filings, one industry survey, and price data available at time of writing. Comments are closed sitewide. Corrections to corrections@coinagereport.com.