Four hundredths of a month. That is the entire movement in Strive's cash coverage of its SATA preferred dividend between Aug. 28 and Sept. 4 โ 18.67 months to 18.71 months, a distance no larger than rounding noise.
In those same seven days, the company issued 921,511 new preferred shares. It lifted its annualized dividend obligation by $11.98 million. It bought 1,375 Bitcoin at an average of $79,281 per coin including fees and expenses, taking holdings to 24,531 BTC.
A ratio that barely flinches while the fixed-charge base beneath it expands by eight figures is not a demonstration of strength. It is a demonstration that two large numbers moved in near-lockstep, and that the ratio is being held in place by an arithmetic coincidence rather than a structural buffer. Tracing the gas trails back to the root cause is the only way to tell ballast from balance.
SATA is a strange instrument to find at the bottom of a Bitcoin treasury, so understand the plumbing before the price. Each share carries a $100 stated amount. The board sets the annual rate, and on Aug. 13 it held that rate at 13%, effective for periods beginning Sept. 1. Thirteen percent of $100 is $13 a year per share โ but it is not distributed quarterly. It is distributed across 21 discrete business-day payments per month. For September, the declared amount is $0.0516 per share on each of those 21 dates, payable to holders of record at the preceding business day's close.
$0.0516 ร 21 ร 12 = $13.0032. The daily accrual is not a convenience of the disclosure. It is the contract.
Three architectural features matter here, and each one changes the risk math.
Perpetual. There is no maturity date. Nothing forces Strive to refinance in 2028 or 2031. That is the appeal โ no wall, no cliff, no rollover risk in the conventional sense. It is also the trap: an obligation with no terminal date is an obligation that is never discharged, only serviced. Corporate treasurers have historically paid a premium for that comfort, and the premium shows up in the coupon.
Variable-rate. The 13% is not a market-determined spread discovered by a bookrunner. It is a board-set number sitting on top of a benchmark. That means the cost of capital can reset against Strive without Strive doing anything at all โ no vote, no filing, no negotiation.
Senior to common. In liquidation, SATA holders line up ahead of common equity and behind creditors. In a going concern, they line up ahead of common for cash. Common holders get whatever is left after the daily sweep clears.
Now the disclosure itself. Strive's Sept. 8 filing reported 1,375 BTC purchased between Aug. 31 and Sept. 4 at roughly $79,281 per coin including fees and expenses, bringing holdings to 24,531 BTC. Over the Aug. 28 to Sept. 4 window, SATA shares went from 9,073,914 to 9,995,425 โ an increase of 921,511. The company's table counts shares sold by a stated 4 p.m. cutoff that would be issued the following business day, alongside shares already outstanding, which means the reported count slightly leads the settled count.
Multiply 9,995,425 by $13 and you get $129.9 million a year. Multiply 9,073,914 by the same $13 and you get $118.0 million. The $11.98 million delta is desk arithmetic at an unchanged rate โ no repricing, no step-up, no covenant trigger. Just more shares.
Now reconcile the funding, because that is where the structure actually lives. The filing does not allocate the Bitcoin purchases between specific financing sources, so what follows is reconstruction, not quotation.
If the 921,511 new preferred shares were sold at their $100 stated amount, gross proceeds were approximately $92.15 million. Cash and cash equivalents rose $19.1 million over the same week, from $183.5 million to $202.6 million. Add those two figures and you get $111.25 million of capital raised or accumulated inside seven days.
The Bitcoin purchase cost 1,375 ร $79,281 โ $109.01 million.
The residual is roughly $2.24 million. That is where the week's real business lived โ underwriting fees, custody, operating overhead, and the slice of the daily dividend accrual that fell due inside the reporting window. On roughly 9.5 million average shares at $0.0516 per business day, a five-day slice is about $2.6 million. The reconstruction closes to within a few hundred thousand dollars.
This matters because it identifies what the machine is. Strive is not funding a Bitcoin purchase out of operating cash flow. It is converting a perpetual, board-repriced, senior cash obligation into a non-yielding hard asset, and running the coupon on the residual. That is a legitimate capital structure. It is also a duration mismatch wearing a treasury-company suit.
Here is where the coverage ratio stops being boring.
Sell one SATA share at $100. Cash rises $100. The annual dividend obligation rises $13. The marginal coverage that single transaction contributes is 100/13 โ 7.69 years of coverage per dollar of new obligation. Structurally, every preferred share sold and held as cash is enormously accretive to coverage. It is the most balance-sheet-friendly thing the company can do.
But almost none of the proceeds were held as cash. Of the $92.15 million raised, $19.1 million landed in the cash line. The rest became Bitcoin. So the marginal coverage of the week's actual capital allocation was $19.1 million of new cash against $11.98 million of new annual obligation โ 1.59 years.
Compare that to the reported ratio. $202.6 million of cash against $129.9 million of annualized dividend is 1.56 years, or the 18.71 months in Strive's table.
The marginal multiple and the reported multiple are almost identical. That is the entire story. The coverage ratio did not move because the week's blend of funding sources happened to reproduce the existing ratio almost exactly. It was not defended. It was reproduced. Four hundredths of a month of drift is not a margin of safety; it is the width of the gap between 1.59 and 1.56.
And that coincidence is asymmetric. If next week's issuance mix tilts further toward Bitcoin โ a higher share of proceeds converted rather than retained โ the marginal multiple falls below 1.56 and coverage begins to decay. If it tilts toward cash, coverage improves quickly. The reported 18.7 months is a snapshot of one week's mix, not a property of the business. Changing the mix changes the ratio, and nothing in the disclosure commits Strive to any particular mix.
There is a cleaner way to see the same structure, and it does not require trusting anyone's ratio. Divide the annualized dividend by the coin count. $129.9 million across 24,531 BTC is $5,297 of perpetual annual cash obligation per coin held. Per month, $441. Per business day, about $21.
Bitcoin does not pay that. No part of the protocol yields $5,297 per coin per year. So the obligation must be met from somewhere else: cash, new preferred, new common, or coin sales. Each of those four sources has a different failure mode, and only one of them is silent in a bull market.
Which produces the embedded hurdle. To hold the fiat value of the stack flat after dividends, the coins must appreciate by 5,297/79,281 = 6.68% a year, every year, before a single dollar of operating expense, custody cost, or audit fee. Bitcoin has delivered that in many past years. It has also posted multi-year stretches where it did not, and the obligation does not observe those stretches.
Now the rate sensitivity, which the coverage table does not disclose at all. SATA's 9,995,425 shares at $100 stated represent just under $1.0 billion of stated capital. Every 100 basis points on that base is $10.0 million of additional annual obligation โ roughly 0.9 months of coverage at the current cash balance. Move the board-set rate from 13% to 15% and the annual bill rises to $149.9 million; coverage drops from 18.71 months to 16.22. Two percentage points of rate cost almost two and a half months of the buffer the company spent a week of share issuance holding flat.
I am not forecasting a 15% rate. I am pointing out that the sensitivity exists, that it is board-mediated rather than market-mediated, and that it lives entirely outside the table.
The coverage figure is also deliberately narrow. It excludes operating needs, future financing, investment income, and other liquid assets. Among those excluded holdings sit 505,000 shares of Strategy's STRC preferred stock, valued at $49.364 million on Sept. 4.
Read that line twice, because it is the most revealing one in the disclosure. Strive holds $49.4 million of another company's variable-rate perpetual preferred โ the same instrument class it is issuing, from the company whose playbook the market says Strive is following. It is a treasury that holds preferreds, issues preferreds, and buys Bitcoin with the spread.
That position is a genuine buffer. At Sept. 4 marks it covers roughly 4.6 months of the current dividend bill. But it is a buffer marked at market, in a correlated asset class. In a funding stress, STRC and SATA reprice in the same direction, for the same reason, at the same time. Two preferreds in one treasury is not diversification. It is concentration in the thing that is already the problem.
And the daily accrual structure matters more than the annual headline. $0.0516 per share per business day is not a quarterly appropriation the board can defer with a press release about strategic priorities. It is a scheduled cash sweep across 21 dates a month, senior to common, with an unpaid balance that historically converts into governance rights, board seats, or an escalating rate. The 18.71-month coverage figure treats a stream of daily obligations as if it were a single pool, and treats cash as if cash were the only thing that can be spent. Neither simplification is dishonest. Both are flattering.
The consensus read on this disclosure is simple and wrong. The consensus is: Bitcoin treasury company issues preferred, buys coin, coverage holds, balance sheet fine, tick the box.
The contrarian read is that the coverage ratio is measuring the wrong quantity, and measuring it at the wrong frequency. It measures the current stock of cash against the current run-rate of a variable obligation, while the entity is actively converting cash into an asset that produces no cash. Every week the ratio is computed, a different fraction of proceeds has been converted than the week before. The ratio is a photograph of a moving process, and this week the shutter happened to catch 1.59 and 1.56 superimposed.
The metric that would actually inform you is the marginal coverage multiple: new cash retained per dollar of new annual obligation. This week it was 1.59 years. Held at par as cash, it would have been 7.69. The gap between 7.69 and 1.59 is a single capital allocation decision โ whether to convert proceeds into Bitcoin โ and it is not disclosed as a metric anywhere in the filing.
Equally important is isolating the layer of failure, because reflexivity language gets overused. This is not Terra. I spent two weeks in May 2022 reverse-engineering the LUNA/UST seigniorage logic inside Anchor's contracts and published, before the collapse, that the model was mathematically self-terminating: the peg and the reserve were the same asset, and the reflexivity had no floor. SATA has a floor โ a stated amount, a senior claim, a hard asset behind it. There is no algorithmic loop, no oracle to manipulate, no mint-and-burn function to break.
What exists instead is a cash-flow mismatch between a non-yielding reserve asset and a senior daily obligation, plus a reflexivity channel that only opens during a drawdown, when coin sales become the funding source. Different layer, different failure mode, different timeline. Conflating the two is how people miss both.
There is a second blind spot, and it is uncomfortable for people who work where I work. There is no smart contract to audit here. I can disassemble a Parity multisig kill function in six weeks because the logic is on-chain, deterministic, and verifiable by anyone with a node. Nothing about SATA lives in bytecode. Its risk lives in a presentation table, a board resolution, and a footnote about excluded holdings. When the code is absent, the auditor has to dig through the disclosure instead โ and the disclosure is written by the party being audited. That asymmetry is the actual vulnerability. It will not be found by a fuzzer.
On top of it sits custody. 24,531 BTC means key management, custodian counterparties, insurance limits, and settlement latency. None of that appears in a coverage table. A dollar of cash and a dollar of Bitcoin are not the same dollar when the obligation comes due on a Tuesday and the coin is behind a withdrawal queue.
Watch three lines in the next disclosure and ignore the rest. The change in cash. The change in SATA share count. The marginal coverage multiple those two imply. If new cash retained per dollar of new annual obligation stays above 1.56 years, the reported ratio holds โ by construction, not by strength. If it slips below, 18.71 months begins to decay, and it will decay quietly, a few hundredths at a time, which is precisely how it arrived.
The question the table will never answer on its own: at what point does a perpetual obligation against a non-yielding asset stop being a capital structure and start being a schedule? The filing cannot tell you. Shifting the consensus layer, one block at a time, means reading the arithmetic underneath it โ and here the arithmetic is a mix, a rate, and a daily $0.0516 that does not care what the price chart says.