The line that stuck with me sits one line under the BMNR ticker on the cover of BitMine's quarterly filing. BitMine Immersion Technologies holds 6,001,302 ETH, about 4.9% of all Ethereum, and its chairman calls the company debt-free. The second ticker, BMNP, is a perpetual preferred paying 9.5% a year in weekly instalments, and Tom Lee calls it his cost of debt.
The distance between the press release and the fine print is where this company gets interesting. What I keep coming back to is simpler: once BitMine stops buying ETH, what is the stock for?
What an Ethereum treasury company is
A treasury company is a listed business that raises money in public markets and puts most of it into one asset. Strategy did that with Bitcoin and holds 848,000 BTC, the largest corporate Bitcoin stack in the world. BitMine is running the same play with Ethereum, and Evernorth is doing it with XRP.
For a buyer the appeal is practical. You get ETH exposure inside an ordinary brokerage account, next to your other stocks, without opening an exchange account or looking after a wallet. You buy a share and the company holds the coins.
BitMine adds a second layer. It stakes most of its ETH, which means locking coins into Ethereum's validation process in return for rewards. That turns a passive pile into income, and it is where most of the argument about the company sits.
Tom Lee is BitMine's chairman and co-founder of Fundstrat. His profile explains a lot of the demand for the shares. It tells you nothing about the balance sheet.
How BitMine bought almost 5% of Ethereum
In nine months BitMine raised $11.87 billion by selling its own stock, and its share count went from 232 million to 580 million.
That is how you buy almost 5% of Ethereum without borrowing. You issue shares to people who want the Ethereum bet in a brokerage account and use the proceeds to buy coins. No loan, no convertible notes, no interest bill.
The buying has been relentless: ETH purchased every single week since the strategy started on June 30, 2025, more than 60 weeks in a row. The pace has changed. Lee told Bankless in August that he had throttled it down, and the weekly figures show it, 7,391 ETH in the week before August 10 and 17,362 ETH in the last week of September.
This week Lee said BitMine is about 100,000 ETH from its target and then it is done stacking. The market heard that. The stock was at $26.31 before the news and traded at $23.92 the next day, down 9%, while Ethereum was down about 2% over 24 hours. At that morning's price BitMine was worth less than the ETH and cash it holds, roughly 92 cents on the dollar.
That discount is the most informative number in the story, in my view. A treasury company trading below the value of its assets has a problem with its own business model.
The $19 billion cost basis and the $8 billion paper loss
As of May 31, 2026, BitMine had paid $19.05 billion for its ETH. The same filing put the fair value of that ETH at $10.86 billion. That is an $8 billion hole on paper, and it is why the company reported a net loss of $9.1 billion for nine months.
Almost all of that loss is unrealized. BitMine marks its ETH to market every quarter, so a falling ETH price hits the income statement even though no coin left the building. The same rule prints a large gain when ETH rallies.
Two items from the fine print matter. The prospectus says BitMine's disclosure controls were not effective as of February 28 because of material weaknesses in its internal controls over financial reporting. In April the company dismissed its auditor and hired KPMG.
Changing auditors is what a company of that size should do, and the next annual report will be the first one a Big Four firm signs. I read that as a reason to go through the next filing line by line.
Staking is the real business
BitMine stakes 5,067,309 of its ETH through its own platform, MAVAN, short for Made In America Validator Network. Staking puts coins to work securing the network in exchange for rewards, which for a holder this size is the difference between a static pile and a cash flow.
Lee likes to say the operation earns about a million dollars a day. The 10-Q shows what was booked in the March to May quarter: $45.7 million, or about $503,000 a day. Still real money, and roughly half the headline.
MAVAN was bought, not built. In March BitMine acquired Pier Two, an Australian staking company, for about $27.8 million in cash, stock and earn-outs, then rebranded it. Lee says MAVAN now stakes about $2 billion of other people's crypto, which makes it a business with outside customers.
Somebody takes a cut. Until September, a company called Ethereum Tower ran BitMine's staking strategy under a 10-year agreement that paid it a share of net staking revenue. On September 4, BitMine ended that agreement and signed a new one with an affiliate, American Validator, for 1.50% of staking rewards. The filing says BitMine paid no material penalty to get out.
The second ticker: BMNP, the perpetual preferred
In June, BitMine sold 3.5 million shares of a 9.5% Series A Perpetual Preferred under the ticker BMNP. A preferred share sits between a bond and common stock: it pays a fixed distribution and usually has no maturity date. Perpetual means it never has to be repaid.
At launch a buyer paid $80 and got credit for $100, and BitMine owes $9.50 a year on every share, paid weekly. Lee says it is the only one of the 175 largest preferreds that pays every week. The deal was supposed to be 3 million shares and demand pushed it to 3.5 million, or $280 million raised. Lee says it was more than 5x oversubscribed, and in his October message he put the price at $99, up from $80.
The terms show what the buyer actually holds.
| Term | Detail |
|---|---|
| Price paid at launch | $80 |
| Face credit per share | $100 |
| Annual distribution | $9.50 per share |
| Payment schedule | Weekly |
| Missed payments | Cumulative, compounding up to 15% |
| Board seat | After 12 missed months |
| Call protection | 110 cents on the dollar until December 2027 |
| Takeover | $100 cash per share plus anything unpaid |
The dividend is cumulative, so a missed payment does not vanish; it compounds up to 15%. Miss for 12 months and preferred holders get to put a director on the board. BitMine can call the issue back, but until December 2027 that costs 110 cents on the dollar. If the company is sold, every preferred holder can demand $100 a share in cash plus anything unpaid.
The filing says BMNP is not debt and that it ranks junior to any existing and future indebtedness, and Lee is right about that. In his October message, though, he called it his cost of debt at 9.56%. His own word for it is the giveaway: a fixed payment on a fixed date is a bill, whatever the accounting line says.
The arithmetic is simple. BitMine pays $9.50 a year to raise $80, which is almost 12% money, and the weekly bill comes to about $640,000 unless the company buys the preferred back.
Why the 5% target keeps moving
Every BitMine press release tells you what percentage of Ethereum it owns, and that percentage depends on a denominator most readers never check. Through August 10 the company used 120.7 million ETH. On September 28 it used 122.1 million.
Ethereum has no hard cap on supply, so the denominator grows. Five percent of 120.7 million is 6,035,000 ETH. Five percent of 122.1 million is about 6,105,000. Moving the denominator added 70,428 ETH to the target.
Etherscan put total supply at 122,108,563, so the new number is the accurate one and the old one was stale.
What stands out is the timing. BitMine fixed the denominator right as it got close to the old finish line, so anyone tracking the company should check which supply figure a press release uses.
What happens when the buying stops
The numbers start to push against each other. BitMine stakes 5,067,309 ETH at a 2.62% yield, which produces about 132,000 new ETH a year. The gap to 5% is about 104,000 ETH, so staking alone closes it in roughly nine months.
After that, about 11,000 ETH a month arrives from rewards and pushes the company past its own cap. Either the cap breaks or BitMine starts selling ETH. Lee has floated the second option. On Bankless in August he said: "At 5%, if we're at 5%, and then we're still stacking ETH through yield, there is maybe a scenario where we're just going to not want to go over 5%, so we would sell the rewards that we're receiving."
The other half of the machine has changed too. Lee told Bankless that BitMine's main source of cash is selling common stock when it trades above net asset value. The ratio is called mNAV, the share price divided by the net assets behind each share. Above 1, selling stock buys more ETH per share. Below 1, it gives ETH away.
That morning BitMine traded near 0.92. The arithmetic: 6,001,302 ETH at $2,523, plus cash and stakes, minus the preferred and the bills, comes to about $15.75 billion. Spread over 603 million shares, that is about $26.11 a share. The stock was not there, so the common stock tap was shut.
BitMine has already spent $350 million buying back 21 million shares this summer, which is what you do below net asset value, and it still has room under a $4 billion buyback authorization. A buyback spends cash, though; it does not raise any. Lee has said that finishing the job might need only another $150 million of preferred, not billions, and that if the CLARITY Act had passed, BitMine probably would have moved faster.
The bull case, and where it strains
The strongest argument against the bear read is straightforward, and I will give it at full strength. BMNP costs BitMine $33.25 million a year. Projected staking income is $358 million a year, 10.8 times coverage. Even on what the company actually booked last quarter, coverage is about 5.5 times. BitMine also holds $672 million in cash and securities, roughly 20 years of dividends in the bank. Lee has said he does not expect billions of BMNP unless BitMine goes well beyond 5% of Ethereum.
All of that is true. The risk sits in the next issue, sold by a company that cannot sell common stock below net asset value and has set a cap its own staking yield will not let it hold.
Lee's own words point at the strain. He says the dividend is easily covered by staking rewards, and in the same interview he says BitMine is not converting those rewards into dollars at all. Both cannot fund a cash dividend at once, because rewards that stay in ETH do not pay out cash. With $672 million in the bank that is manageable. The habit is what matters.
The prospectus spells out the alternatives: staking, options strategies on its ETH, and at-the-market offerings of preferred or other equity securities. Selling more preferred to pay the preferred is in the document, in the company's own words.
The options line is real, and it has cost money. The 10-Q says BitMine sells put options on ETH, and in the March to May quarter that line lost $92.1 million, twice what staking earned. Selling puts collects a premium and takes on the obligation to buy ETH if the price falls, so it pays steadily and hurts in a downturn.
Strategy offers the comparison. Michael Saylor's company built a shelf of preferreds named Strike, Strife, Stride and Stretch, each with its own dividend. Between June 30 and August 10 it sold 6,916 Bitcoin, and its own 8-K says the proceeds went to fund distributions on preferred stock. That happened with a $2.55 billion cash reserve on hand and a program to sell up to $1.25 billion more Bitcoin to refill it. 6,916 out of 848,000 is less than 1%, small enough that it is no fire sale, but the pattern is there, and BitMine has built the first floor of the same structure.
Odds, falsifiers and the November filing
The creator's odds are projections rather than forecasts: 70% that by the end of 2027 BitMine either owns more than 5% of Ethereum or has sold more preferred stock than the 3.5 million shares it sold in June, and 20% that it happens before the end of 2026.
Three outcomes would prove that wrong, and each one is checkable. The fiscal 2027 annual report shows 3.5 million preferred shares and holdings at or under 5%. BitMine redeems BMNP early at 110 cents. BitMine sells its staking rewards every month for a year to hold exactly 5% and says so in a filing. An ETH rally does not count, because a rally does not fix the cap; it makes issuing more paper easier.
For anyone holding or considering BMNR, the useful frame is that the stock is a spread rather than a coin. It earns a floating 2.62% on its ETH and pays a fixed 9.5% to raise money, and the gap gets filled by ETH going up. Lee claims 53,500 basis points of outperformance since BitMine started, and says the stock is up 3% this year while ETH is down 9%. Those are his numbers from his own deck. If ETH goes sideways, you are paying a management team to run a spread you could skip by owning ETH directly.
The next annual report is due in late November, and it will be the first one KPMG signs. When it lands, the numbers to check first are the preferred share count, the ETH holdings, and the staking revenue booked against the million-dollar-a-day claim.
