One number moved 70%. Nobody noticed until the stock fell.
When Metaplanet wrote its Series 10 stock acquisition rights, each right converted into 410 shares of common stock. By the time the board intervened this month, the same right converted into 696. That is a 69.8% expansion of the insider claim — generated not by performance, not by a vesting cliff, not by a bitcoin-denominated strike adjustment, but by the mechanical operation of an anti-dilution ratchet welded to the company's own capital-raising program.
The board has now reset the ratio to 410, the level immediately before Metaplanet's September 2025 international share offering — the point management itself identifies as when its raises stopped being strongly accretive. Roughly 319.5 million potential shares collapse to 188.2 million. A 41% cut. CEO Simon Gerovich, who recused himself from the decision as a Series 10 holder, puts the erased warrant value above $220 million.
But sit with the other number, the one buried in the middle of the release: bitcoin per fully diluted share rises about 8.8%, and Metaplanet does not buy a single satoshi to get there.
That is not a treasury result. That is a denominator result. And the gap between those two things is where every digital asset treasury company now lives.
Metaplanet is the Tokyo-listed former hotel operator that rebuilt itself around a single balance sheet decision: accumulate bitcoin, fund it with equity, and let the share price carry the story. It is the Japanese expression of the Strategy template — the model that turned a software company into a leveraged bitcoin proxy and spawned a global cohort of imitators, the so-called digital asset treasury (DAT) companies.
The mechanics are simple to state and brutal to run. The company sells shares. It uses the proceeds to buy bitcoin. Bitcoin per share goes up. The market, attracted by that rising metric, pays a premium to the underlying coin holdings. That premium — market value versus net asset value, or mNAV — is the engine. As long as mNAV sits above 1, issuing stock is accretive to existing holders. Below 1, every share sold transfers value away from the people who already own the stock. The entire model is a spread trade on its own valuation.
Metaplanet's Series 10 rights were executive compensation layered on top of that machine. Each right allowed the holder to buy shares at a deeply discounted 10 yen. That is the strike. It does not move. What moved was the conversion ratio — how many shares each right ultimately delivered. And because the plan was sized as a percentage of fully diluted capital rather than as a fixed number of shares, the grant rebased upward every time the company issued equity to buy more bitcoin.
Read that again. The compensation pool was denominated in a floating denominator. The company's core strategy was to increase that denominator continuously. The two were not merely adjacent. They were wired together.
The pool went from roughly 46 million shares to about 319 million. A 6.9x expansion. It happened in public, in filings, over quarters — and it took a share price collapse and weeks of shareholder fury to force the arithmetic onto the agenda.
Start with the ratchet, because the ratchet is the whole article.
Anti-dilution ratchets are ordinary instruments. They exist to protect a holder when a company issues stock at a price below what the holder paid or was promised. In venture, a full ratchet resets the conversion price to the new, lower issuance price. In structured debt, a similar clause compensates the lender for value transferred to new equity. The logic is defensive: the holder should not be punished for the issuer's decision to sell cheap.
Now invert the incentive. Give the ratchet to the people who decide whether to sell cheap.
Series 10 was sized as a percentage of fully diluted shares. Every capital raise that funded a bitcoin purchase enlarged the fully diluted share count. A larger share count meant a larger absolute grant to the insiders. A larger absolute grant at a 10-yen strike meant more intrinsic value per right. And more intrinsic value per right raised the conversion ratio, which in turn raised the fully diluted share count again.
That is a feedback loop. Not a metaphor for one. An actual, self-reinforcing loop in which the compensation instrument consumes the output of the treasury strategy and feeds it back into its own base.
The printed conversion ratio tells the story in two acts. It began at 410. It ended at 696.
Two multipliers produced the 6.9x expansion in potential shares. One was the conversion ratio itself, which ran from 410 to 696 — a 1.70x move. The other was the number of outstanding rights, which grew alongside the share count precisely because the pool was denominated as a percentage of capital. Multiply the ratio expansion by the growth in rights and you land near seven. Neither multiplier required a board vote on compensation. Both were automatic consequences of the treasury program. That is the definition of an unowned risk: a material transfer of value with no decision point attached to it.
I have audited leverage logic before. In 2017 I ran a six-person team through the 2x Funding smart contracts line by line during the ICO mania and found an integer overflow in the leverage calculation that could have drained user funds in a volatility spike. The lesson from that engagement was never about Solidity. It was about what happens when a variable everyone assumes is static turns out to be dynamic under load. Here, the variable is the dilution base. Nobody priced it as dynamic. It was dynamic.
There is a subtler flaw, and it is the one I would flag to any board currently running a DAT structure. A compensation pool sized as a percentage of floating capital is economically identical to a perpetual rebase token. The holder's claim is not fixed in shares. It is fixed in proportion, and the proportion is enforced against a denominator that the issuer controls. Corporate finance teams learn to treat percentage-of-capital grants as neutral because in a stable-capital company they roughly are. In a company whose stated, disclosed, board-approved strategy is to issue equity continuously for years, they are not neutral. They are a second claim on the same dilution.
Metaplanet's board eventually reached that conclusion, and the fix is real. Resetting to 410 strips out the conversion-ratio expansion that occurred after the September 2025 offering. The 20% carve-out that was going to seed a new employee incentive pool was scrapped, and those rights were canceled inside the 41%. That detail is easy to skim past. It means a portion of the pool that was going to be redistributed was instead extinguished. Good.
The lock-up is the more interesting design choice. The remaining warrants become exercisable in thirds in 2029, 2030 and 2031, and shares received on exercise stay locked until August 2031. That is duration matching, and it is the correct instinct. A warrant that cannot be monetized for years does not pressure the float, and it aligns the holder with a medium-term bitcoin thesis rather than a quarterly print.
But a lock-up is a speed bump, not a fix. It reduces the discount rate applied to the overhang. It does not reduce the overhang. The claim still exists; it is simply deferred. Anyone modeling fully diluted bitcoin exposure should treat the 188.2 million remaining potential shares as live, because they are.
Warrant overhang does something specific to a stock that a simple dilution table does not capture. It caps rallies. Every buyer who models fully diluted supply knows that strength in the share price converts directly into intrinsic value for a 10-yen strike holder, who can then hedge or sell into that strength. The overhang is a standing offer to sell into your bid. When the ratio drifted from 410 to 696, the notional size of that standing offer grew by the same proportion. This is why the 17% two-session drawdown landed before the fix was announced rather than after. The market was pricing the overhang, not the bitcoin.
Now the accretion math, because this is where the release earns its headline. Fully diluted share count falls by roughly 131 million — the 319.5 million minus 188.2 million. The bitcoin stack is unchanged. When the numerator is fixed and the denominator shrinks, the ratio rises. That is the entire 8.8%.
This is why I keep insisting that bitcoin-per-fully-diluted-share, not bitcoin holdings, is the only metric that survives scrutiny in this sector. Holdings is an absolute number that management can grow by selling stock. Fully diluted per-share is the number that tells you whether the growth accrued to you or to someone else. Metaplanet's share price down more than 43% this year, against a roughly 15% decline in bitcoin and about 20% at Strategy, is what happens when the market finally starts doing that division.
The contrast with Strategy is instructive, and the market has already drawn it. Strategy ran the same basic playbook — sell equity, buy bitcoin, report rising bitcoin per share — and its stock is down about 20% year to date against bitcoin's roughly 15%. Metaplanet is down more than 43%. That gap is not a bitcoin gap. It is a governance gap priced as a beta differential. Logic dictates value, perception dictates volume — and perception here is a function of how much of the fully diluted stack belongs to people who are not buying on the open market.
There is a market-structure dimension Western coverage has largely skipped. Metaplanet trades on the Tokyo exchange with a heavily retail-weighted shareholder base and a listing regime that historically gave issuers wide latitude on stock acquisition rights. The Series 10 structure was legal. It was disclosed. It was also, apparently, not understood by a meaningful share of the people who owned the stock — which is a different failure than concealment and a far more common one. Legal disclosure and effective disclosure are not the same deliverable, and the gap between them widens exactly when the instrument's economics are dynamic. Blind faith is the only true vulnerability, and blind faith is manufactured by complexity that is compliant on paper.
And note what the 41% cut does not touch. Gerovich keeps the 64 million shares he received through an August 28 exercise under the old terms, plus the right to acquire another 49.1 million. That is a retained position of roughly 113 million shares. Matthew Sigel, VanEck's head of digital assets research, calculated that the CEO gives up roughly 79 million shares worth about $123 million and called the package a meaningful realignment of management and shareholder interests. I do not disagree with the characterization. I would add the qualifier: a realignment is not a reset. The exercised portion was already out the door.
The announcement also did not address Gerovich's economic interest in MMXX Ventures. That omission matters more than its word count suggests. A recusal from a board vote is a procedural instrument. It governs who sits in the room, not who holds the claim. When the compensation at issue sits alongside a separate private vehicle with undisclosed economics, the governance disclosure is incomplete by construction — and incomplete disclosure is the raw material of every future dispute.
One more layer, and this is the one that generalizes. The September 2025 offering is identified as the point where raises stopped being strongly accretive. That is an admission that mNAV had crossed below the threshold where equity issuance transfers value. But the ratchet was not recalibrated at the crossing point. It was recalibrated months later, after a 17% two-session drawdown and after Gerovich's first attempt to address shareholder anger failed to land.
The lag is the cost. In a structure where dilution compounds, the delay between recognizing a problem and correcting it is not neutral time. It is accrual. Every week between the September offering and this month's reset, the ratio drifted further from 410. The contract executes, the architect pays — but here the shareholders paid first, and the architect paid after the price had already moved.
The consensus reading is that Metaplanet capitulated to its shareholders, and that this is a governance success. I would frame it differently.
This is a partial refund issued after a structural flaw was permitted to compound across multiple quarters, in filings, and only once the equity market forced the issue. The 41% cut is the correct correction at an inflated cost. Call it a win if the label helps, but score it as remediation, not prevention.
The genuine blind spot is not inside Metaplanet. It is across the cohort. Every digital asset treasury company that pays insiders with grants sized as a percentage of fully diluted capital, while simultaneously running a strategy of continuous equity issuance, carries the same latent liability. Metaplanet is simply the first to be forced to publish the arithmetic. The others have not been stress-tested by a 43% drawdown yet — and a ratchet is invisible until the denominator moves.
That is the composability problem in its purest form: composability is leverage until it is liability. A treasury strategy, a compensation plan, and a share price premium were three separate decisions that happened to share a variable. No one owned the interaction. This is what happens when a system's components are individually defensible and collectively incoherent.
Then there is the replacement program. Metaplanet says it will design one with an outside consultant. That is the same category of advisor whose standard percentage-of-capital templates produced the original structure in the first place. Consultants do not carry the liability of their templates. Shareholders do.
The thing to watch is not the 41%. It is the shape of the replacement. If the new program is a fixed number of shares or rights, struck once, with a hurdle tied to mNAV rather than to a floating share count, the loop is broken. If it is another percentage of fully diluted capital, the ratchet returns with a new label and a fresh vesting schedule.
For every other DAT, the disclosure to demand is identical: publish the conversion ratio history. Ask what the ratio was at launch, what it is today, and which specific event moved it. A governance fix that cannot answer those three questions is a press release, not a repair.
Ask for the ratio, not the headline. Trust no one, verify everything, build twice.