Twenty One Capital holds 43,514 bitcoin. That is not an estimate — the company publishes proof on-chain and anyone can check it. What the company costs on the stock exchange is a question with four answers.
On 22 July 2026, with bitcoin at $65,700 and the stock at $4.92, all of these were true at once: 0.59x, 0.72x, 1.11x, and 1.18x if you count the convertible note. The lowest figure says you are buying bitcoin at a 41% discount. The highest says you are paying an 18% premium. Same company, same day, same vault.
None of them is calculated wrongly. That is the interesting part.
This piece covers three things: where Twenty One Capital came from, how its engine is supposed to work, and why mNAV is only a metric when four labels travel with it.
Where the company came from
The name refers to the 21 million bitcoin that will ever exist. Twenty One Capital was announced in April 2025 as a joint venture between Tether, issuer of the dollar token USDT, the exchange Bitfinex, Japanese investment group SoftBank, and the investment bank Cantor Fitzgerald.
It reached the market not through an ordinary listing but through a blank-cheque shell: Cantor Equity Partners, ticker CEP. A shell like that has no business. It raises money, is already listed, and then looks for a company to merge with. Buying the shell means buying a bet on what ends up inside it.
That explains something you trip over on the long-term chart. The highest price the security ever printed is around $59, on 1 May 2025, under the CEP ticker — after the announcement and long before closing. At times the shell traded 500% above its issue price. The spike on the left of the chart is not a Twenty One valuation. It is what the market paid for an expectation before there was anything to value.
The combination closed on 8 December 2025 and the stock has traded on the New York Stock Exchange as XXI since 9 December 2025. The company is based in Austin, Texas.
How it was funded. Closing came with two private placements: roughly $365m of equity — $200m in April at $10 a share, $165m in June at $21 — plus a $486.5m convertible note. The note is senior and secured, pays 1.00% a year and matures on 1 December 2030. Each $1,000 of principal converts into 72.0841 shares, up to 35,068,912 additional shares in total.
Where the coins came from. Much of the stack was contributed rather than bought. Tether and Bitfinex supplied the majority and SoftBank a stake in the order of 10,500 bitcoin. The remainder came from purchases funded by the raise. Everything sits in custody at Anchorage.
Who owns it. At launch Tether and parent iFinex held 58.8% of the equity and 71% of the voting power, with Tether alone controlling a 51.7% voting majority. In May 2026 Tether bought out SoftBank's remaining stake of roughly 25%, and SoftBank's representatives left the board. Since then one shareholder decides where this goes.
One thing the company deserves credit for. Twenty One publishes its holdings on-chain at xxi.mempool.space. Anyone can verify that the coins sit where the company says they sit. That is more verifiability than most balance-sheet items anywhere on a public exchange — and it makes the denominator of the calculation below the least contested part of the whole story.
How the engine is supposed to work
A treasury company is not a factory. It has a vault and a share, and its declared metric is bitcoin per share. The idea is that holding the stock gets you more bitcoin per share over time without buying any yourself.
The mechanism is simple. The company issues new shares and buys bitcoin with the proceeds. Whether that helps or hurts existing holders rests on one condition: does the market pay more for a share than the bitcoin attributable to it is worth?
If yes, the proceeds of one share buy more bitcoin than that share represents, and bitcoin-per-share rises for everyone. If no, the sign flips and every new share dilutes the exact number the company exists to grow.
So the engine switches itself off the moment the stock trades below its bitcoin. Which is why the question of where that line sits is not a footnote. It determines whether the business model is available at all.
The evidence here is unambiguous. Holdings have stood unchanged at 43,514 bitcoin since the listing. BitcoinTreasuries.net reports a year-to-date change of zero bitcoin, and the last four purchases date from before the listing. At an average cost of $84,865 per coin against $65,700, the stack sits roughly 23% underwater.
The mNAV problem
Which brings us to the metric everything hangs on.
What mNAV measures. Picture a company that owns nothing but a vault of gold. No factory, no customers, no staff. Just the vault. The question: does the market pay more or less for the whole company than the gold inside is worth?
Less, and you are buying gold at a discount — mNAV below 1.0. More, and you are paying a premium for the wrapper — above 1.0. Swap gold for bitcoin and you have the metric.
The denominator is easy. 43,514 coins, verifiable on-chain, times the current price. At $65,700 that is roughly $2.86bn.
The numerator has four knobs. Each with a serious argument on both sides.
1. Which shares count?
This one moves the most. Twenty One has two classes: 346,548,153 Class A shares that trade freely, and 304,842,759 Class B held by the controlling shareholder and never traded. Both counts are as of 30 March 2026.
Count Class A only and you strip nearly half the ownership out of the numerator while leaving every satoshi those owners are entitled to in the denominator. The company looks cheaper than it is.
The counter-argument is not silly: Class A is all you can actually buy, so what you are pricing is a slice of the float. But then consistency demands removing the corresponding bitcoin from the denominator too. None of the common variants does that.
2. Does debt count?
A company holding $2.9bn of bitcoin financed with half a billion of borrowings is not the same animal as one holding the same coins debt-free. Include the note and you measure enterprise value: what a takeover would cost, liabilities included. Exclude it and you measure what shareholders pay. Both are legitimate; they are not interchangeable.
3. Is cash subtracted?
If money from a raise still sits in the bank rather than in bitcoin, you are not buying pure bitcoin exposure. Subtract it and the numerator shrinks. Leave it and you have quietly treated dollars as bitcoin.
4. Which price, on which date?
Bitcoin moves, the stock moves, and they do not move together. XXI closed at $5.32 on 20 July, fell about 14% on the 21st, and stood at $4.92 on the 22nd. Every mNAV from those three days is a different number. On top of that, the coin count comes from a periodic filing while the share price is live — mix the two silently and you are comparing two moments.
The four numbers side by side
Stock $4.92, bitcoin $65,700, 43,514 coins, so a denominator of $2.86bn:
| What goes in the numerator | Numerator | mNAV |
|---|---|---|
| Traded Class A only (346.5m) | $1.71bn | 0.60x |
| All share classes (651.4m) | $3.20bn | 1.12x |
| All classes plus conversion shares (686.5m) | $3.38bn | 1.18x |
| All classes plus the note, less cash | $3.61bn | 1.26x |
The distance between the first row and the last is the distance between a 40% discount and a 26% premium.
What the data providers get right — and where it breaks
The sites publishing these numbers are not the problem. BitcoinTreasuries.net stacks three variants for Twenty One and labels each one: 0.59x basic, 0.72x on enterprise value, 1.11x diluted, alongside market cap, enterprise value and bitcoin per share in two versions. That is exactly the disclosure to ask for.
And it is complete enough that the method can be reverse-engineered rather than guessed. Three checks:
- 43,514 divided by 346,548,153 gives 0.0001256 — their "BTC / share (basic) 0.000126". So basic means Class A only.
- 43,514 divided by 651,390,912 gives 0.0000668 — their diluted figure of 0.000067. So diluted means all share classes, excluding the convertible.
- 346,548,153 times $4.92 gives $1.71bn — their $1.7bn market cap.
That also explains the third number. Their $2.1bn enterprise value is the $1.7bn market cap plus $0.4bn of net debt, i.e. the $486.5m note less roughly $86m of cash.
Which produces the actual finding: the EV variant puts the full debt against roughly half the equity. The note sits on every share, but only Class A appears in the numerator. That is why the figure that sounds the most comprehensive shows the second-deepest discount.
This is not an accusation. It is labelled, and anyone reading both metrics together can see it. The damage happens one step downstream, when one of three numbers travels into an article, a headline or a slide deck without its label. What arrives reads as "trading at a 41% discount", and nobody can tell that half the company fell out of the numerator on the way.
What has been decided since the listing
April 2026 — the big restructuring. Tether proposed merging Twenty One with Strike, the bitcoin payments company run by then-CEO Jack Mallers, and with Elektron Energy, a bitcoin miner. Per Tether's release, the three parts would become one platform spanning treasury, mining, financial services, lending and capital markets. Bloomberg framed the proposal at the time against a backdrop in which the core business — holding bitcoin — had fallen out of favour.
May 2026 — one owner. Tether bought out SoftBank's stake and SoftBank's directors left. No terms were disclosed.
July 2026 — the unwind. On 20 July Mallers stepped down as CEO, succeeded by Raphael Zagury, until then an independent director and interim chair of the audit committee. Zagury founded and runs Elektron Energy, and previously worked at Goldman Sachs, Deutsche Bank and Merrill Lynch.
The three-way merger is dead. Strike stays independent and is no longer a candidate for any combination. A two-way deal between Twenty One and Elektron remains under evaluation, described by the company as preliminary with no assurance of an agreement. The stock fell about 14% on the news.
On the reason for the departure there are two accounts, and they do not match. In the release and his own post on X, Mallers cites focusing on Strike. In a video message reported by several trade outlets, he speaks of differences with the board over long-term direction — that he had envisioned bitcoin businesses generating cash flow, not a vault alone. We have not reviewed the video ourselves and therefore carry both versions.
The new direction. Alongside the change, the company laid out refreshed priorities centred on generating cash flow and allocating capital with discipline. The core of it: building and acquiring profitable operating businesses under a long-term ownership model it compares to Berkshire Hathaway. Zagury describes his job as building the operating company around the balance sheet, with the discipline, governance and executional rigour of an institution.
What comes next — three things to watch
The Elektron deal is a related-party transaction. The incoming CEO founded and runs the potential target. The controlling shareholder controls both sides and is also Elektron's customer, since Tether outsources a significant portion of its mining there. That is not an allegation, it is a structure — and structures like it are normally assessed by an independent committee with its own fairness opinion. How that committee gets staffed is the observable event over the coming weeks. Zagury resigned his committee seats effective 20 July, so the audit committee has just lost its interim chair.
The engine is idle. Zero bitcoin change year to date, and as long as the stock trades at the low end of the mNAV range, the company cannot improve the metric it was built for by issuing equity. Against that backdrop the pivot to cash flow reads less as a change of heart and more as the only remaining road.
The next hard data point is the quarterly report on 31 August 2026. That should show the actual cash position, whether an at-the-market equity programme exists, and whether "acquiring operating businesses" has become anything specific.
Limits of this account
Four things belong on the record, otherwise this piece commits the offence it describes.
Share counts are from the 30 March 2026 filing; conversions or issuance since then are not captured. Cash is not netted out in the first three rows of the table because no current figure is published — the roughly $86m in the fourth row is derived from the gap between reported market cap and enterprise value, not disclosed. Prices and holdings are a snapshot from 22 July 2026. And "zero purchases" rests on the year-to-date change reported by BitcoinTreasuries.net, not on a company document.
Those are the same four labels whose absence this piece is complaining about.
Four questions to ask of any mNAV
- Which shares are in the numerator — traded only, or all classes?
- Is debt included?
- Is cash netted out?
- What date are the share price and the coin count from?
All four answers, and you can rebuild the number. Missing answers, and it is not a metric — it is an assertion with a decimal point in it.
The same rule we apply to backtests
A return without its benchmark and its window says nothing about a strategy. Twenty percent sounds good until you add that simply holding did thirty over the same stretch.
An mNAV without a share count and a date is the identical failure. Both look like measurements. Neither can be reproduced, which is the only test that matters.
What holds up
Twenty One Capital is a company with an unusually verifiable denominator and an ambiguous numerator. The bitcoin sits in the open on a public ledger; the valuation rests on definitions almost nobody supplies.
So whether the stock trades at a discount or a premium is not first a market question. It is an arithmetic question — and the answer changes the story completely, from "bitcoin here is cheaper than on the market" to "you are paying for the wrapper".
The question is never whether a number is correct. It is whether you can rebuild it. What cannot be reproduced was never a measurement.
This piece describes a company, a metric and how it is calculated. It is not investment advice, a recommendation, or a forecast. All figures as of 22 July 2026, bitcoin at $65,700, XXI at $4.92.
Sources: Twenty One Capital, Inc., press release on completion of the business combination with Cantor Equity Partners, 8 December 2025 · Twenty One Capital, Inc., Form S-1, SEC EDGAR (convertible note terms, 72.0841 conversion rate) · Twenty One Capital, Inc., Form 10-K, shares outstanding as of 30 March 2026 · BitcoinTreasuries.net, Twenty One Capital company page, accessed 22 July 2026 · Tyler Rowe, "Jack Mallers Steps Down as CEO of Twenty One Capital", BitcoinTreasuries.net, 21 July 2026 · Tether, statement on the proposed merger, 29–30 April 2026 · Tether, statement on the acquisition of SoftBank's stake, 20 May 2026 · Bloomberg, "Tether Proposes Mergers Involving Bitcoin Treasury Company", 30 April 2026 · Prices: TradingView and GuruFocus, 21–22 July 2026.