Company figures
Four numbers, and where each one came from.
A model is only as good as its inputs, so here they are with their provenance. Check them against the current filing. If one is wrong, change it in the calculator: nothing here is fixed.
Strategy MSTR
Balance sheet as of 2026-08-17 (3 days ago). Prices as of 2026-08-19 20:27.
| Field | Value | What it is |
|---|---|---|
| Bitcoin held | 840,447 | Coins on the balance sheet, from the most recent capital-relevant filing. |
| Diluted shares | 397,446,404 | Basic shares plus options. Convertibles are handled separately, because they only dilute above their strikes. |
| Senior claims | $17.053B | Debt plus preferred stock at notional, less the USD reserve. Net rather than gross: repaying a note from the reserve reduces both sides and leaves equity value unchanged. |
| Share price | $103.36 | Last close. Refreshed daily. |
| Bitcoin price | $68,309 | Spot. Refreshed daily. |
Which gives bitcoin per share 0.002115, a bitcoin position of $57.41B, net asset value per share $101.54, and a premium of 1.018.
Sources: Strategy filings on EDGAR ยท Q2 2026 10-Q (convertible schedule)
Balance-sheet figures are entered by hand and re-verified when a new capital-relevant filing appears, so they lag the company. Prices are refreshed daily. Check any of it against the current filing and change it: everything here is editable, with the two prices locked by default because they are fetched rather than assumed.
The convertible schedule carries 6 series, each with its own strike and conversion rate, so dilution is modelled note by note rather than as one aggregate.
The other companies, and why they are a separate tier
The companies below are derived from bitcointreasuries.net as of 2026-08-19. They have not been read out of anyone's filings here, so they are not the same kind of figure as the record above and are not presented as one.
The gap is measurable rather than a matter of trust, so here it is. Running that source's figures for Strategy against the hand-checked record on this page: bitcoin held matched exactly, share count ran -2.2% and senior claims +5.8%. Treat every figure below as carrying an error bar of about that size. Convertible dilution is not modelled for any of them, because inventing a schedule would be worse than admitting there isn't one.
| Company | Bitcoin | Diluted shares | Senior claims | Share of value |
|---|---|---|---|---|
| Twenty One Capital XXI | 43,514 | 651,390,000 | (1,170M net cash) | 134% |
| Metaplanet Inc. MPJPY | 43,000 | 1,434,392,925 | $309M | 116% |
| Strive ASST | 20,246 | 113,402,709 | $246M | 76% |
Share of value is the company's bitcoin divided by its enterprise value, and it is the reason these companies are on the list and others are not. This model prices a company as a pile of bitcoin with claims against it, so it can only describe a business whose value is mostly the bitcoin. A company must clear 70% to appear, and hold at least $1bn of it.
Two exclusions are worth naming because they surprise people. A bitcoin miner holding a large stack is still excluded, because most of its enterprise value is plant and hash rate and this model cannot see either. And a company whose bitcoin is worth more than twice its enterprise value is excluded too: at that point either the bitcoin is not really there or there is a claim on it that enterprise value does not capture, and neither of those is a bargain. 57 companies were looked at and did not qualify.
Why the tiers are kept apart
A curated record assembled from secondary sources looks exactly as authoritative as one read out of a filing, and is worth considerably less. Keeping the two visibly separate is the only honest way to offer both, so the calculator says which tier it is using every time it fills a company in.
A wrong number stated confidently is worse than a missing one. Whichever tier a company comes from, the way to rely on it is to check it against the current filing and change it in the calculator yourself. Every field is editable for exactly that purpose.
Currency is worth one note. Everything is modelled in dollars, and a company reporting in another currency is converted once at a fixed rate. No exchange-rate path is modelled, and the model does not pretend to know one.