The ticker is SATS. By Wednesday morning it had been indexed, mis-indexed, and re-indexed by every aggregator on the planet, most of which tagged it to satoshis โ the smallest unit of bitcoin โ rather than to Strive's Series A perpetual preferred stock. That is the first error. It will not be the most expensive one.
Here is the second. Strive originally filed to sell 5 million shares at $100. It closed 7 million shares at $90. A 10% discount to the stated price, paired with a 40% increase in size. In eighteen years of reading term sheets I have never watched a book that was oversubscribed get both cheaper and larger on the same day. Price down, size up is not demand. Price down, size up is a concession โ and concessions are data.
Third: the coupon is described everywhere as 15%. It is 15% of $100 of face value. Buyers are paying $90 for that face. The number that governs the deal is 15/90 = 16.67%. The cost of capital is 1.67 percentage points higher than the headline, and the headline is what every downstream note will quote for the next six months.
The code didn't fail here. There is no code. That is the entire point, and it is why this story is being misfiled as a Bitcoin story when it is a capital-markets story wearing a Bitcoin costume.
Strive is a Bitcoin treasury company. The category has a template and the template has a name: Strategy, formerly MicroStrategy. Buy bitcoin with other people's money, mark it at market, let the equity trade at a premium to the bitcoin per share, then sell more equity into that premium and buy more bitcoin. The premium is the engine. It is called mNAV โ market-to-net-asset-value โ and it is the single number that determines whether the model compounds or unwinds.
Strive holds 35,595 BTC, roughly $2.9 billion at the implied marks in the disclosure. It runs an asset-management subsidiary reporting more than $2.8 billion under management. It bought 2,822 BTC for $235 million, an average near $83,270. Its chief executive is Matt Cole, who is the public face of the financing strategy and, by extension, the concentration risk. The stock is up 125% since July. It is also โ and this matters more than the release wants it to โ an investor in Strategy, holding STRC, and that position is currently down $6.3 million.
So the picture, before interpretation: a mid-cap treasury company, levered to one asset, imitating a larger treasury company, losing money on its position in that larger treasury company, raising its most expensive capital to date at a discount, and buying back its own common stock with the proceeds.
I want to be precise about what a perpetual preferred is, because a large part of this audience will nod at the word and then misprice the instrument. A preferred share is a hybrid. It sits above common equity in the liquidation stack and below debt. It pays a fixed dividend. Perpetual means no maturity โ the principal is never returned. Cumulative means a skipped dividend accrues, and no common dividend can be paid until the arrears clear. For the issuer, that is permanent capital with a debt-like coupon and no refinancing cliff. For the buyer, it is a bond with equity-like recovery risk and no claim on upside beyond the coupon.
That last clause is what gets lost. The preferred holder gets 16.67% and a senior claim. The common holder gets the bitcoin beta. Those are different products sold under one roof, and the structure deliberately routes upside to the junior tranche while routing the fixed obligation to the senior tranche.
Now the arithmetic, because arithmetic is the only part of this that is not negotiable.
Seven million shares at $90 is $630 million gross. The dividend is 15% of $100 face, or $15.00 per share per year. Annual cash obligation: 7,000,000 ร $15 = $105,000,000. Yield on actual proceeds: $15 / $90 = 16.67%.
The company's own framing is 15% of $100 per share. That framing is what will be repeated. The gap between 15% of $630 million ($94.5 million) and the real obligation ($105 million) is $10.5 million a year. This instrument has no maturity, so decade-scale thinking is correct. Over ten years that gap is $105 million of extra coupon โ roughly 1,270 BTC at current marks โ paid because the coupon attaches to a face value nobody actually paid.
This is not fraud. Dividends attach to par, not to issue price. Every bank does it. But every bank also gets asked, in diligence, what the effective cost of capital is, and the answer here is 16.67%, and the marketing says 15%. The instrument is not deceptive. The vocabulary around it is.
Then there is the reserve. Eighteen months of dividends, pre-funded. Eighteen months of a $105 million annual obligation is $157.5 million โ roughly a quarter of the gross raise. A quarter of the proceeds is not going into bitcoin at all. It is going into a segregated account to pay the coupon for a year and a half.
Why would a company pre-fund eighteen months of dividends? Two readings. The benign one: prudent treasury management, standard practice, show buyers you are covered. The other: management does not believe operating cash flow will cover the coupon within eighteen months. Look at what the company actually earns. It is a holding company whose primary asset is a non-yielding commodity, plus a fee business. There is no operating engine generating $105 million a year. There is no scenario in which the coupon is paid from operations. It is paid from bitcoin appreciation, from asset sales, or from the next raise.
An eighteen-month reserve is a company telling you, in the least alarming available phrasing, that it has modeled eighteen months of runway.
Now the loop. Proceeds are earmarked for bitcoin-related products, working capital, assets and other businesses โ and repurchase of Strive Class A common stock.
Read that last item twice. The company is selling senior paper at 16.67% to buy back junior equity. In a normal capital structure you repurchase equity when you think it is cheap and you issue equity when you think it is rich. Here both happen at once, in opposite directions, with the more expensive instrument funding the cheaper one.
Two coherent explanations exist. First: the buyback supports the share price, the share price supports the mNAV premium, and the premium is what permits the next raise. The preferred is a maintenance cost for the flywheel. Second: management genuinely believes the common is undervalued against its bitcoin per share and is arbitraging its own capital structure. Only one of those is a growth story. The other is a treadmill. You cannot tell which you are looking at from outside, because the disclosure folds buybacks into "assets and other businesses" without a percentage allocation.
Tracing the bleed through the gateway. The gateway is the preferred issuance. Every dollar in has to come out somewhere: bitcoin, buyback, reserve account, or coupon. Three of those four are non-productive near term. The one productive use is a bet on an asset that, in a sideways tape, is doing nothing to service a 16.67% obligation.
Then the overhang. Warrants exercised, 13.6 million new Class A shares, net proceeds $369.2 million. That is a real raise and it is underweighted in coverage fixated on the preferred. Two points. First, dilution โ 13.6 million shares against a float I cannot size from the disclosure, but against a company measured in the low billions, material. Second, sequencing. Warrants exercised, common repurchased, preferred issued. In flow terms: new common came in, some common went out, and a senior claim was layered on the remainder. The equity base is being restructured at precisely the moment the narrative premium peaks.
A capital structure that grows more complex as the underlying asset grows less volatile is not being optimized. It is being defended.
And the mirror. Strive holds STRC โ a Strategy preferred instrument โ and is down $6.3 million on it. Set aside the size. Look at the signal. A company whose entire thesis is buy the asset the largest treasury company buys, financed the way the largest treasury company finances it, holds in its own portfolio a losing position in the largest treasury company's preferred stock. That is not a rounding error. It is a demonstration. The instrument Strive is now issuing โ high-coupon perpetual preferred tied to a bitcoin treasury โ is the same instrument class on which Strive itself is currently underwater.
If Strive's own capital allocators cannot make money holding a Strategy preferred, what is the argument that a buyer of Strive's preferred does better holding a structurally identical product from a smaller, less liquid, less proven issuer?
Silence is the loudest bug report. Nothing in the material I reviewed reconciles this.
Two numbers deserve separation. $2.8 billion in assets under management at the subsidiary. $2.9 billion of self-held bitcoin. One decimal apart, describing completely different things, deployed in the same paragraph of the same release. AUM is client money โ it generates fees and is not balance-sheet principal risk. Self-held bitcoin is principal risk. When a company puts $2.8B AUM next to 35,595 BTC, the reader's eye merges them into $5.7 billion of something. It is not $5.7 billion of anything. It is a fee business and a directional bet.
Based on my audit experience, conflating an AUM figure with a principal-holdings figure is one of the most reliable tells in a treasury-company deck. I watched the same move in 2021 while reconstructing the BZOptimism bridge flows: the headline loss mixed user deposits with protocol-owned liquidity, and the true shortfall only appeared after the two were separated on a spreadsheet. Same discipline here. Verify the root, ignore the branch. The root is: what does the company own, and what does it owe? It owns 35,595 BTC at an average near $83,270. It owes $105 million a year, indefinitely, in a currency that is not bitcoin.
One more thing, small but not trivial. The disclosure places settlement on October 20, 2026, while the transaction is described as having occurred last week. If the event is in 2025, that is a settlement window north of twelve months. Standard IPO settlement is T+2 to T+5. A year-long settlement is not a settlement; it is a condition precedent, or a typo. I will not build a thesis on a date. But in a document where the price moved, the size moved, and the coupon was described two ways, the one number that cannot be reconciled with anything else is the one that determines when money changes hands. Check the SEC filing. If the date is right, the deal carries a condition nobody is discussing. If the date is wrong, the disclosure quality is what it is, and you should price that too.
Now the failure cascade, because reflexivity is a word people use to avoid arithmetic.
Layer one: bitcoin falls. At a $83,270 average, a 20% drawdown from that mark puts the position underwater on cost. mNAV compresses, because the market stops paying a premium for a levered position that is losing money.
Layer two: the premium compresses. Below 1, issuing equity destroys value โ each new share sells for less than the bitcoin it represents. The flywheel reverses. Strategy can absorb this because it has scale, liquidity, and an index-adjacent shareholder base. Strive, at a fraction of the size, cannot.
Layer three: the raise window closes. The $105 million coupon does not care. It is cumulative. It accrues. And when the company cannot issue new paper at a premium, it pays from the reserve โ eighteen months deep.
Layer four: month nineteen. Now the company sells bitcoin, cuts the dividend, or restructures. Cutting a cumulative preferred dividend is the closest thing to default available to an equity instrument. It does not kill the company. It kills the company's ability to raise preferred again.
Entropy always finds the path of least resistance. For a company with a fixed obligation and a volatile asset, the path of least resistance is to keep raising. For the market, the path of least resistance is to keep buying the narrative while the premium holds. Both paths terminate at the same coordinate.
Here is the case against my own case.
First: this is a registered IPO of a preferred stock. Not a token. Not a yield farm. Not an unregistered security sold to retail through a chat server. It sits inside the SEC perimeter, with a prospectus, a board, audited financials, and a liquidation waterfall a lawyer can enforce. On the spectrum of what is being sold in this sector right now, that is the top decile of legitimacy, and I should say it plainly. The structure is cleaner than ninety percent of what gets called innovation in this industry.
Second: 16.67% is not obviously expensive. If you believe bitcoin compounds above 30% annually over a multi-year horizon, borrowing at 16.67% to hold it is accretive. That is the entire bull case and it is not stupid. It is a leveraged carry trade. Carry trades work until they don't. Strategy has run a version of this for years and has, so far, been right.
Third: the reserve cuts both ways. I read it as an admission. A bull reads it as a company that pre-funded its obligation precisely so a bitcoin drawdown would not force a distressed sale. That is a legitimate reading of the same fact, and the interpretation depends entirely on what bitcoin does over the next six quarters.
Fourth, and this is what the bear case usually misses: in a sideways market, yield is scarce. A 16.67% coupon with seniority and a bitcoin-adjacent issuer will find buyers. The move from $100 to $90 may not be weakness at all. It may be the company correctly pricing the instrument to clear. A book that clears at 90 is a book that cleared.
There is an uncomfortable symmetry here with the interoperability thesis I have held for years. IBC is the most elegant protocol engineering in this industry and ATOM captures almost none of the value. Elegance and value capture are orthogonal. The same is true of SATS. It is a well-constructed instrument, and the value it generates flows overwhelmingly to common shareholders, who take the bitcoin upside, while preferred holders take a fixed claim on a volatile enterprise. Structural elegance is not an investment thesis. It never has been.
Watch three numbers, not thirty.
mNAV. Above 1.5, the flywheel turns and the coupon is a rounding error. Below 1.2, the next raise gets harder. Below 1, the model inverts and no amount of preferred engineering saves it.
The reserve burn. Eighteen months are funded. Track when drawing begins. The quarter the reserve is tapped is the quarter management stops believing its own cash-flow model.
The next raise. If Strive returns inside twelve months at a lower price or a higher coupon, the concession I opened with was not an anomaly. It was the beginning of a curve.
History is a Merkle tree, not a narrative. Every fact here hashes into the next. The $90 price hashes into the 16.67%. The 16.67% hashes into the reserve. The reserve hashes into the buyback. The buyback hashes into the premium. The premium hashes into the next raise. You do not need to forecast bitcoin to know how this chain terminates. You only need to check whether the next block validates.
Precision is the only apology the truth accepts. The headline says 15%. The document says $90 for $100. Do the division.