Strategy Inc.

01 — Executive summaryA net loss of $20.8 billion on $247 million of revenue. Both numbers are correct.

Strategy Inc. earned $246.7 million of revenue in the first half of 2026 and reported a net loss of $20.76 billion. Neither figure is an error. Under the fair-value accounting standard the company adopted on 1 January 2025, the 846,000 bitcoin it holds are marked to market through operating income every quarter, so the income statement is now a bitcoin price feed with a software business attached to it. In the second quarter alone the mark moved $8.32 billion against the company, turning a profitable software segment into a $24.45 loss per share.

The accounting is the least interesting thing about Strategy. The interesting thing is that the mechanism which built the company has stopped working, and the company has begun to dismantle it in public.

For four years Strategy was a machine that converted a share-price premium into bitcoin per share. The logic required one condition: that MSTR trade above the value of the coins attributable to each share. Above that line, issuing equity buys more bitcoin than it distributes ownership, and bitcoin per share rises without anyone doing any work. Below it, the same machinery runs backwards — every share sold buys fewer coins than the share represents, and holders are diluted by their own company's growth.

On 27 June 2026 the premium went away. Strategy's market capitalisation fell below the market value of the bitcoin it held for the first time in the company's history as a treasury vehicle. Four months later, on 18 September 2026, the shares closed at $153.92, having rallied 66% in twenty-two sessions on a bitcoin recovery. On the company's own definition the stock trades at about 1.2 times the bitcoin backing each share; on the definitions used by third-party trackers it trades between 0.96 and 1.11 times. The honest summary is that the premium is now small, unstable and definitionally contested.

What has happened since June is more consequential than the ratio itself. Strategy sold bitcoin. Between 26 May and 20 September 2026 the company disposed of roughly 6,949 coins — including the first reported sale since December 2022 — and bought back 5,553, for a net addition of 2,225. It repurchased $1.5 billion of its 0% convertible notes due 2029 at an 8% discount. It authorised $1.0 billion for buying back its own common stock and $2.0 billion for buying back its preferred, and has used $875 million of the latter while leaving the former untouched. It stood up a $5.0 billion dollar reserve whose stated purpose is to pay dividends on securities it keeps issuing.

These are not the actions of a company accumulating an asset. They are the actions of a company managing a liability structure. That distinction is the entire report.

The obligation has two clocks, and they are not synchronised. The preferred dividend is perpetual and runs at roughly $1.73 billion a year regardless of what bitcoin does. The convertible debt looks, on a maturity schedule, comfortably staggered from 2027 to 2032. On the schedule that actually governs it — the holder put — 88.1% of the principal becomes repayable in cash inside a single twelve-month window running from September 2027 to September 2028. Cumulative dividends plus puttable principal through September 2028 comes to roughly $9.6 billion against a dollar reserve of $5.0 billion. The reserve covers the dividends or the puts. It does not cover both.

There is a genuine franchise here, and it deserves stating plainly. Strategy holds 4.03% of all the bitcoin that will ever exist, at an average cost of $75,416 a coin against a market price of $85,261 — a reserve $8.3 billion above cost. Its STRC preferred is the largest listed preferred instrument in the world by market capitalisation, at $9.5 billion, and it trades within 1.5% of par with 30-day historical volatility of 11.5%. The company raised $20.92 billion in nine months through two distinct equity channels and has $19.09 billion of unused common-stock capacity. That is a real capital-markets business, and it is why the shares have traded at a premium at all.

But a franchise is not a valuation. On Strategy's own published thresholds, bitcoin must compound at more than 10.70% — the company's current effective cost of credit — for net bitcoin per share to capture a positive spread. Bitcoin's realised return from its October 2025 peak to 21 September 2026 is −32.4%. Its four-year annualised return is +36%, which is the number the bull case rests on and the number a bear would call a survivor's statistic. Both are true. Neither is a forecast.

We model three outcomes and arrive below the current price. In our bear case — bitcoin at $58,000 and a 0.80× mNAV — the shares are worth $58. In the base case — bitcoin at $95,000, parity mNAV — $144. In the bull case — bitcoin at $126,000 and a restored 1.15× premium — $235. Probability-weighted at 25/50/25, that is approximately $145. We set a twelve-month target of $145 and rate the shares Neutral. The distribution is the finding: a fourfold spread between the bear and the bull is not a valuation range, it is a statement that the equity is a leveraged bitcoin position wearing a corporate structure.

Three things would change our view. Bitcoin sustaining above roughly $95,000 would put the reserve far enough above cost to make the dividend strip look small. A resolution of the MSCI consultation in Strategy's favour would remove a $2.8 billion overhang on the premium. And disclosure of the STRK and STRD conversion mechanics, or a demonstrated ability to refinance the 2027–2028 puts at par, would tell us the closed loop can keep turning through a bad bitcoin year. None of the three is currently in evidence.

02 — Company & business modelOne ticker, three claims on the same pile of coins

Strategy Inc. is a Delaware corporation listed on the Nasdaq Global Select Market under the ticker MSTR. It was known as MicroStrategy for thirty-six years and renamed itself in February 2025; the ticker did not change. It describes itself as the world's first and largest Bitcoin Treasury Company. From the second quarter of 2026 it reports two segments — Bitcoin and Software — having previously reported the software business as a single operating segment with the bitcoin programme disclosed as a non-segment activity.

The reclassification is more than presentational. It is the company acknowledging in its own filings that the treasury operation is the primary business and the enterprise analytics business is the secondary one. That is an accurate description of the economics. It is also a problem, because the treasury operation does not produce revenue, and the segment that does produce revenue is smaller than the interest bill.

What the company actually owns

Strip away the securities and the balance sheet is simple. At 13 September 2026 Strategy held 845,050 bitcoin, carried at a market value of $65.29 billion on a bitcoin price of $77,266. Against that sat $6.398 billion of dollar liquidity — a $5.100 billion reserve designated for dividends and interest, plus $1.298 billion of unrestricted cash — and two layers of senior claims: $6.71 billion of convertible notes and $14.48 billion of perpetual preferred stock. Net of the claims and including the dollar assets, the residual attributable to common shareholders was $50.50 billion.

By 20 September the coin count was 846,000 and the reserve had grown with the bitcoin price, but the shape of the stack was unchanged. Everything Strategy has done for two years has been an exercise in moving value between these four buckets.

Three securities, three different investors

The most useful way to understand Strategy is not as a company but as an issuer of three distinct products aimed at three distinct buyers.

The common stock (MSTR) is a residual claim on the bitcoin after the senior stack is settled. It is the product sold to investors who want bitcoin exposure with amplification — the company itself publishes an "amplification" ratio, 1.24× at 18 September, and states explicitly that it amplifies downside as well as upside. Common shareholders are last in priority, bear the full volatility of the underlying asset, and can be diluted at any time by an at-the-market offering programme that has $19.09 billion of unused capacity.

The preferred stock (STRC, STRF, STRK, STRD, STRE) is a fixed-income product sold to investors who want a high coupon and are willing to accept a claim that is senior to the common but junior to the debt. STRC — the Variable Rate Series A Perpetual Stretch preferred — is the flagship: $9.49 billion outstanding, a 12.00% rate paid semi-monthly, and a stated objective of trading at $99–$100. Strategy describes this business as "Digital Credit" and has built an entire framework around it.

The convertible notes are the legacy funding channel: $6.714 billion of principal, mostly issued between 2024 and early 2025 at coupons between 0% and 0.875%. Their economics are entirely a function of the conversion price relative to the share price, and they carry holder puts that make them behave more like short-dated debt than their stated maturities suggest.

Governance and key-person concentration

Strategy is controlled in practice, though not in law, by Michael Saylor, its co-founder and Executive Chairman. The company operates a dual-class structure: class B shares carry ten votes each and are held predominantly by insiders. Saylor's public commentary is the primary marketing channel for the securities, and the company maintains a disclosure dashboard that it treats as a Regulation FD-compliant distribution channel for market prices, bitcoin holdings and supplemental metrics.

This matters for two reasons. First, the equity's valuation has historically embedded a personal-brand premium that is not obviously transferable. Second, the capital-allocation framework the company adopted in June 2026 is a board policy, not a contractual protection. The company's own disclosure states plainly that it "may be modified, suspended or terminated, and no issuance, repurchase or BTC sale is required." A shareholder who owns MSTR for the framework owns a promise, not a right.

03 — The financial recordTwo quarters, $22.8 billion of marks, and a small profitable software business

Strategy's reported results for 2026 are dominated by a single accounting fact: bitcoin is measured at fair value with changes recognised in net income. The consequence is that the income statement has almost no information content about the operating business, and a great deal of information content about the bitcoin price.

The single line that matters for anyone attempting conventional analysis is the second from the bottom. Preferred dividends went from $49.1 million in the second quarter of 2025 to $400.7 million in the second quarter of 2026, an eight-fold increase in four quarters, and they are deducted before anything reaches the common. That is the cost of the Digital Credit strategy arriving in the income statement.

The operating business underneath

Strip out the digital-asset line and Strategy is a modestly profitable, slowly growing enterprise analytics company. Second-quarter revenue of $122.4 million was up 6.9% year over year. Gross margin was 66.6%, down 210 basis points. The software segment earned $3.71 million of segment profit, a margin of 3.0%, down 23.6 percentage points from a year earlier. On a segment basis the Bitcoin operation recorded a loss of $8.22 billion.

Two observations. First, the software business is real but small: annualised revenue is under $500 million, and at a 3% segment margin it contributes roughly $15 million a year of profit. The capital structure consumes a hundred times that. Second, the margin collapse is the arithmetic of a licence-to-subscription transition. Revenue is being recognised over time rather than at the point of sale, and the high-margin maintenance base is shrinking faster than the new subscription line is growing.

The company frames the quarter around subscription growth of 54% year over year, to $62.9 million. That is a genuine achievement and it is disclosed accurately. It is also worth noting that product support — the recurring revenue line with the highest margin and the longest customer relationship — fell from $52.1 million to $40.2 million over the same period, a decline of 23%. The total moved from $111.1 million to $122.4 million in five quarters. That is a compound quarterly growth rate of 2%.

Cash flow, which is where the report actually lives

Because the fair-value marks are non-cash, the cash flow statement is the only place where Strategy's operating reality is visible without distortion. In the first half of 2026 the company generated $9.85 million of net cash from operating activities. Investing activities consumed $14.37 billion, almost entirely bitcoin purchases. Financing activities provided $13.77 billion, almost entirely securities issuance.

The pattern is unambiguous: the company's operating business is a rounding error, its investing is funded entirely by its financing, and its financing exists to fund both the investing and the distributions on the securities it has already issued. A business that generates $9.85 million of operating cash in six months and pays $692.5 million of preferred dividends over the same period is not self-funding by any definition. It is funded by the market's willingness to buy its paper.

04 — Segment deep diveThe Bitcoin segment has no revenue and all of the risk

The Bitcoin segment reported zero revenue in the second quarter of 2026. It reported a segment loss of $8.22 billion. That is the whole of the segment disclosure, and it is a fair representation of what the business is: an asset-holding activity whose reported result is whatever the price of bitcoin did between two balance sheet dates.

Strategy acquired 85,296 coins in the second quarter. The funding is disclosed: $5.46 billion from STRC at-the-market sales, $0.96 billion from class A common stock at-the-market sales. Against that, 1,395 coins were sold to fund preferred dividends — the first sales of the modern era, and the first evidence that the structure had begun to consume the reserve rather than build it.

This is a business model that a reader should be able to evaluate without any accounting at all. There are exactly three variables: how many coins you hold, how much the coins cost you, and what you had to promise to get them. Everything else — the software segment, the dashboard, the investor briefings, the amplification ratio — is a presentation of those three variables.

The accumulation rate has broken

The holdings series tells the story of the slowdown better than any commentary. Strategy added 137% to its coin count in 2024 and 51% in 2025. In the first quarter of 2026 it added 144,551 coins, the largest quarterly addition in its history. In the second quarter it added 26,836 — an 81% deceleration. In the third quarter, through 20 September, it added 2,225.

There are two explanations, and they are not mutually exclusive. The charitable one is that bitcoin fell 53% from its October 2025 peak to its June 2026 low, and buying into a falling market is expensive in share terms: every dollar raised buys fewer coins, and every coin bought requires more dilution. The company slowed its purchases because the economics of purchasing had deteriorated, which is exactly what a rational capital allocator would do.

The less charitable explanation is that the market stopped paying a premium, so the shares could no longer be sold at a price that made accumulation accretive, and the accumulation programme was therefore suspended. The company's own August 2025 guidance anticipated this: below 1.0× mNAV it said it would consider issuing credit to repurchase MSTR; between 2.5× and 4.0× it would opportunistically issue MSTR to acquire bitcoin; above 4.0× it would do so actively. Observed 2026 behaviour matches the first regime, not the third.

What the marks do to reported performance

The cost basis of the reserve has been remarkably stable since early 2025: $75,537 a coin at 3 May 2026, $75,578 at 30 June, $75,476 at 26 July, $75,416 by 21 September. Strategy bought most of its bitcoin in a narrow price band and its average cost has barely moved in eighteen months. What moves is the market price, and with it the entire reported result.

At 31 March 2026 the reserve was $2.33 billion above cost. At 30 June it was $13.99 billion below. By 21 September it was $8.33 billion above again. Nothing about the company changed across that span — no operating decision, no strategic shift, no customer win or loss. The equity story over those six months is a bitcoin price chart, and the fair-value accounting simply makes that explicit in the financial statements instead of leaving it to the share price.

There is a practical consequence for anyone modelling the company. A discounted cash flow will produce nothing useful, because there is no cash flow to discount. An earnings multiple is meaningless, because earnings are a price feed. The only coherent framework is asset-based: value the coins, subtract the claims, and then form a view on whether the residual deserves a premium or a discount to what it holds.

05 — Segment deep diveThe software business is the least important asset and the most important disclosure

Strategy's enterprise analytics business has been operating for more than thirty-five years. It sells business intelligence, reporting and — since 2024 — AI-assisted analytics software to large enterprises, predominantly in the United States, under the Strategy One brand. It generated $122.4 million of revenue in the second quarter of 2026, up 6.9% year over year, at a segment margin of 3.0%.

On any financial measure this business is immaterial to the equity. At an annualised revenue run-rate of roughly $490 million and a segment margin of 3%, it produces about $15 million of segment profit. The preferred dividend alone is $1.73 billion. A reader could reasonably ask why the company reports it as a separate segment at all.

The answer is that the software business is not carried for its cash flow. It is carried because it makes Strategy an operating company. That status is the subject of the index consultation discussed in section 18, it is the reason the company is taxed as a corporation rather than as an investment vehicle, and it is the reason the preferred dividends are expected to be treated as returns of capital rather than ordinary income. The software business is a structural feature, not a profit centre.

The transition inside the flat total

The revenue line has moved from $111.1 million in the first quarter of 2025 to $122.4 million in the second quarter of 2026 — an increase of 10% over five quarters. Underneath that flat surface, the composition is changing materially.

Licences and subscriptions rose from $44.4 million to $66.5 million, a gain of 50%, driven almost entirely by subscription services, which the company reports at $62.9 million for the second quarter, up 54% year over year. Product support fell from $52.5 million to $40.2 million, a decline of 23%. Other services were flat at around $15 million.

The mechanical effect of that mix shift is margin compression. Maintenance revenue carries very high incremental margin because the software is already written and the customer is already installed. Subscription revenue carries lower margin because it includes hosting, support and a higher cost to serve, and because the initial contract typically involves ramping costs against a revenue stream recognised over time. The segment margin fell from 26.6% to 3.0% in five quarters. That is not a deterioration in the business so much as a re-basing of it.

Whether the re-basing works is a genuine open question. The bull reading is that a subscription base of $62.9 million growing at 54% will, in three or four years, be larger and more durable than the maintenance base it replaced, and that the margin recovers as the transition costs annualise out. The bear reading is that the company is discounting a declining perpetual licence business into a subscription line, that the customer base is not growing, and that a $40 million maintenance line falling 23% a year is the more informative signal about the underlying franchise.

06 — The reserve846,000 coins, 4.03% of the supply, and one asset

Strategy held 846,000 bitcoin at 20 September 2026, which the company states is 4.03% of the total supply that will ever exist. There is no second asset. The company holds $6.4 billion of dollar liquidity against $21.2 billion of senior claims, and the software business is worth perhaps a few hundred million dollars on any reasonable multiple. Everything else is bitcoin.

Concentration of this kind is not a risk factor in the conventional sense; it is the investment. A shareholder in MSTR is making a single, explicit bet, and the only question worth asking is whether the structure around that bet improves or degrades it.

The cost basis and the margin of safety

The average cost of the reserve is $75,416 a coin against a market price of $85,261 at 21 September 2026. The reserve is therefore $8.3 billion above cost. That is a meaningful cushion in a structure where the downside case is a forced monetisation of coins to service claims, because it means the company can sell bitcoin into the current market without recording a loss.

It is worth being precise about how thin that cushion is. A bitcoin price of $75,416 is the line at which the reserve's market value equals what the company paid for it. Bitcoin traded below that line for most of the second quarter of 2026 and was as low as $58,714 on 30 June — 22% below cost. A return to the June low would wipe out the cushion entirely and would put the reserve $14 billion below its cost basis, at which point every coin sold to fund a dividend would realise a loss, with the tax consequences that implies.

Supply concentration and what it does and does not mean

Strategy's 4.03% of supply is the largest corporate holding of bitcoin in existence and roughly nineteen times the next-largest pure treasury company. In a market where the daily traded volume is measured in the tens of billions, a holding of this size is not obviously illiquid: the company sold 3,588 coins in a week in July without moving the price, and its own disclosures note that the reserve could be monetised.

But size cuts the other way in a stress scenario. Strategy's 846,000 coins represent approximately 4% of the float, and the company is the marginal buyer that has supported the price for two years. If the company ever became a persistent seller — not 3,000 coins a week but tens of thousands — it would be selling into a market it had itself been propping up. The company's own disclosure on this point is careful and worth quoting in substance: it states that its supplemental metrics assume that "sales of bitcoin have no effect on the market price of bitcoin," and identifies that assumption as a limitation. It is a substantial one.

The reserve is not the company

The most common analytical error with Strategy is to value it as a bitcoin ETF with a management fee. It is not that, and the difference is not small. A spot bitcoin ETF holds coins in trust, issues and redeems shares continuously against net asset value, must publish its holdings daily, is exempt from corporate-level tax, and cannot take on leverage or issue senior securities. Strategy can do all of those things, and has.

The practical difference for a shareholder is that an ETF tracks its asset, while Strategy tracks its asset multiplied by a leverage factor and divided by whatever the market is currently willing to pay for the structure. Over the twelve months to 18 September 2026 bitcoin fell 25.9%. An ETF holding bitcoin and charging 0.25% would have returned approximately −26%. MSTR returned −55.9%. The wrapper cost thirty points, and that gap is the subject of sections 7 through 13.

07 — The premiumThe flywheel, and the year it ran in reverse

Everything Strategy has achieved rests on a single arbitrage. When MSTR trades above the value of the bitcoin attributable to each share, the company can sell shares for more than the coins they represent, buy the coins, and leave every existing holder with more bitcoin per share than they had before. No operating improvement is required. The company is manufacturing bitcoin per share out of a share-price premium.

The mechanism is elegant and it is genuinely powerful when it turns. It is also entirely dependent on a market condition that the company does not control and cannot create. Strategy's own investor briefing sets out the sequence as a five-step loop: issue credit, deploy capital, earn the spread, improve the franchise, recycle capital. What that framing understates is that the loop requires the market to keep paying more for the wrapper than the contents.

What happened in 2026

On 27 June 2026 Strategy's market capitalisation fell below the market value of the bitcoin it held, for the first time since it began the treasury strategy. The premium became a discount. Over the trailing twelve months the ratio has ranged from approximately 0.95× to 1.43×, and it peaked well above 2× during the 2024 accumulation phase.

The company responded within two days. On 29 June the board approved the Digital Credit Capital Framework, which is discussed in section 11. Its most revealing provision is the authorisation of a $1.0 billion common-stock repurchase programme — a tool that only makes sense for a company that believes its shares are worth more than the market is paying. As of 13 September 2026, not one share had been repurchased under it. The company chose instead to spend $875 million of a $2.0 billion authorisation buying back its own preferred stock at a discount to par.

That choice is informative. Repurchasing STRC at $86.53 against a $100 stated amount retires a 12% coupon obligation for 87 cents on the dollar — a certain, arithmetic return. Repurchasing MSTR at a discount to net reserve per share is a bet on the market re-rating the equity. The company took the certain return. Management's revealed preference is that it considers its preferred stock the better value, which is a statement about the common stock.

The ratio is not one number

An unusual feature of the current situation is that four credible methodologies produce four materially different answers for the same company in the same week. This is not a data error. It is a consequence of how much of the capital structure sits between the coins and the common shareholder.

The divergence has three sources. First, the share count: basic shares outstanding are approximately 420.5 million; fully diluted shares reflecting only in-the-money instruments are approximately 424.5 million; assumed diluted shares outstanding, which treat every convertible and every preferred series as converted regardless of the money, run to roughly 450 million. On the same bitcoin pile, a 7% difference in the denominator is a 7% difference in the answer.

Second, the treatment of the senior stack: some measures deduct the notional value of the claims, others deduct their market value, and Strategy's own framework deducts out-of-the-money instruments at notional while excluding in-the-money STRK shares. Third, the bitcoin price used. Different trackers timestamp their marks differently in a market that moved 66% in a month.

Strategy itself acknowledged the problem when it redefined mNAV on 23 July 2026, stating that figures calculated before that date "are not comparable" with figures calculated after it. A metric that the issuer has to redefine, and that four external observers then calculate four ways, is not a valuation anchor. It is a marketing instrument.

08 — Digital CreditFive instruments, one balance sheet, and a 12% coupon

Strategy's preferred stock programme is the most interesting financial innovation in the company and the part of it that receives the least analytical attention. Between early 2025 and September 2026 the company built a $14.48 billion notional preferred stack across five series, and it describes the resulting business as "Digital Credit" — a term of art for perpetual preferred securities sold to income investors who want exposure to a bitcoin balance sheet without the volatility of the common.

The pitch is straightforward. A perpetual preferred with no maturity and a fixed or periodically reset coupon, issued by a company whose assets are bitcoin, offers a high current yield with a claim senior to the equity. The buyer gets a coupon; the issuer gets long-duration capital without immediate common-share dilution. Both sides are better off than they would be in the alternatives available to them.

The five series, and why the differences matter

STRC — the Variable Rate Series A Perpetual Stretch preferred — is the flagship and by far the largest at $9.49 billion notional. It pays a variable rate, currently 12.00%, reset monthly at the company's discretion, with dividends paid semi-monthly. Strategy's stated objective is for STRC to trade near $100; it currently trades at $98.55, a 1.5% discount to the stated amount, with 30-day historical volatility of 11.5% and 30-day average trading volume of $140.6 million a day. The company has committed to "regular, disciplined repurchases" if the price falls below $100, and has spent $875 million doing so.

STRC's design is deliberate and it is the key to understanding the whole structure. A variable coupon that the issuer resets monthly, combined with an active repurchase programme, is an attempt to create a fixed-income instrument with a stable price. The company's own materials describe the goal as stripping away price volatility. The cost of that stability is borne by the issuer, in the form of a coupon that must be high enough to clear at par — currently twelve per cent.

STRF is the senior instrument: a fixed 10.00% cumulative perpetual, non-convertible, with a $100 liquidation preference, trading at $104.00, a 4% premium to par. It yields 9.62% at that price. It is the only series trading above its stated amount, which tells you what the market thinks of the ranking order.

STRK is the odd one out: an 8.00% convertible perpetual preferred carrying optional conversion into MSTR common above a strike price. It is the only series with equity upside, and it trades at $75.42 — a 24.6% discount to par, implying a current yield of 10.61%. The discount is a market-implied statement about the value of the conversion option and about the credit.

STRD is a fixed 10.00% perpetual that is non-cumulative. That single word is the most important disclosure in the stack: an omitted dividend on STRD is permanently forgone rather than deferred and accumulated. It is the most junior of the series and it trades at $74.19, a 25.8% discount to par, yielding 13.48%. The market is pricing it as the instrument most likely to have its coupon suspended.

STRE is a smaller euro-denominated series of roughly $900 million notional, paying 10.00% and listed in London.

The economics of the flagship

STRC deserves a paragraph of arithmetic because it is 65% of the preferred stack and 71.5% of the annual cash cost. A 12.00% coupon on $9.49 billion notional is $1.26 billion a year of cash distributions, paid semi-monthly. The instrument has no maturity. There is no obligation to repay the principal, ever, unless the company chooses to redeem or repurchase it. From the issuer's perspective that is genuinely attractive: $9.5 billion of permanent capital at a cost below what a comparable sub-investment-grade issuer would pay for twenty-year debt.

From the common shareholder's perspective it is a permanent $1.26 billion annual drain on a balance sheet whose operating cash flow is $20 million a year. The company has effectively sold a perpetual claim on the reserve and must service it forever, from a business that does not generate the cash to do so. That is the trade. It is not obviously wrong — if bitcoin compounds faster than 10.7%, the residual common equity wins — but it is a trade, and it should be underwritten as one rather than presented as accumulation.

09 — The cost of credit10.70% is the number the whole thesis turns on

Strategy publishes a metric it calls the BTC Hurdle ARR, defined as its current effective cost of credit. At 13 September 2026 the figure was 10.70%. The company's own explanation is precise: "If BTC ARR is above this rate, Net BTC Per Share captures a positive spread and appreciates faster than bitcoin on a go-forward basis."

This is the single most useful number the company discloses, and it deserves more attention than it gets. It converts the entire equity story into a single comparison: the return on the reserve against the cost of the claims stacked on top of it. Above 10.70%, the common shareholder earns more than bitcoin does. Below it, the common shareholder earns less than bitcoin does, and the structure is transferring value from the residual to the senior holders.

Three thresholds, three meanings

Strategy publishes a band rather than a single rate, and the three levels are doing different jobs.

The breakeven rate is 2.52%: the ratio of annual interest and dividends to the bitcoin reserve. Above 2.52%, the company could in principle fund its obligations indefinitely out of capital gains on the reserve without ever selling a coin. This is the low-water mark, and it is a modest one. A reserve worth $65.3 billion generating 2.52% a year covers $1.65 billion of distributions.

The hurdle rate is 10.70%: the current effective cost of credit. This is the rate at which the structure is value-neutral for the common. It is roughly four times the breakeven rate, and the gap between them is the cost of the senior claims — the preferred dividends and note interest that must be paid before anything accrues to the residual.

The floor rate is −15.62%: the lowest constant bitcoin return over the weighted-average duration of the credit structure that would still maintain 1.0× coverage of net debt and preferred through the reserve, after funding interest and dividends over the period. Below it, the company states that it "may need to consider restructuring its obligations."

The published bands are modelled outputs, not forecasts, and the company labels them as such. But they are the company's own arithmetic, applied to the company's own capital structure, and they establish the terms of the debate: a common shareholder in MSTR is short a 10.70% funding cost against a long bitcoin position.

Against what realised return?

Bitcoin's realised returns over the relevant horizons make uncomfortable reading against a 10.70% hurdle.

The table is the bull case and the bear case at the same time, and the honest reading is that it depends entirely on the window chosen. Over four years and longer, bitcoin has comfortably cleared the hurdle, and Strategy's net bitcoin per share has grown as a result. Over the last twelve months it has not, and the metric has gone into reverse.

There is a statistical problem with the long-window evidence that deserves stating. Strategy's own briefing reports that MSTR beat bitcoin in 100% of four-year windows, 87% of three-year windows and 64% of two-year windows. Those windows overlap heavily — they are not independent observations, and the sample period covers a single, extraordinary bitcoin bull market in which the shares were repriced from a fraction of net asset value to a multiple of it. A strategy that has worked in every four-year window of a four-year-old experiment has demonstrated that the experiment is young, not that the strategy is robust.

The self-referential part

One further property of the hurdle rate is worth noting. It is described as the current effective cost of credit, which means it rises when the market demands a higher coupon to hold the preferred. The STRC rate has been reset upward from 9% at launch in August 2025 to 12.00% in September 2026. The hurdle rate rose with it.

The circularity is that the cost of credit rises precisely when confidence falls, which is when the structure most needs cheap funding. If STRC's rate had to be raised again — to 14%, say, to clear near par — the hurdle rate would rise toward 12.5%, and the bitcoin return required for the common to break even would rise with it. A shareholder is exposed not just to bitcoin but to the company's credit spread, and the two are correlated in the worst direction.

10 — Convertible debtA four-year maturity ladder that is really a one-year cash call

Strategy carries $6.714 billion of convertible note principal at notional, down from $8.21 billion at the start of 2026 after the company repurchased $1.5 billion of its 0% notes due 2029 in May at roughly an 8% discount to par, paying approximately $1.38 billion in cash. The repurchase reduced assumed diluted shares outstanding by approximately 2.2 million.

On a conventional maturity schedule the debt looks well managed. Tranches fall due in 2027, 2028, 2029, 2030, 2031 and 2032, spread across six years with no single year carrying more than $2.0 billion. Coupons are minimal — 0%, 0.625% and 0.875% — so the cash interest bill is approximately $35 million a year, a trivial sum against the preferred dividend. The notes are unsecured and rank behind nothing but the subsidiary liabilities.

The maturity schedule is not the schedule that governs them.

The put is the instrument

Every Strategy convertible carries a holder put: a noncontingent right to require the company to repurchase the notes for cash at 100% of principal on a specified date. This is a standard feature of the convertible structure, and in most issuers it is an afterthought because the notes trade well above par and no holder would exercise it. In Strategy's case it is the central feature of the capital structure, for one reason: the notes are carried at fair value below par.

Mapped chronologically, the put schedule collapses the six-year ladder into a single year:

88.1% of the principal — $5.914 billion — becomes repayable in cash at the holder's option inside a twelve-month window running from September 2027 to September 2028. The company has said nothing about how it intends to meet that window. Its demonstrated approach is to buy notes back opportunistically at a discount when it has cash and the notes are cheap, which is a sensible tactic but not a plan.

Refinancing is the likely answer, and it is not free

The most probable outcome is refinancing rather than repayment. Strategy has demonstrated repeated access to the convertible market and to the preferred market, and it holds $19.09 billion of unused common-stock capacity. It could also issue new preferred to retire the notes, which would convert a $6.7 billion obligation with a 2028 cash date into a perpetual obligation with a higher coupon.

Each of those options has a cost. Refinancing at higher rates raises the hurdle. Converting debt to equity in a below-parity market is dilutive to bitcoin per share. Retiring notes with preferred stock substitutes a fixed-dated claim for a perpetual one, which reduces near-term liquidity risk and increases long-term cash cost. There is no version of this that is value-creating for the common; the question is which version is least destructive.

11 — LiquidityTwo clocks, one reserve, and a $4.9 billion gap

On 29 June 2026 Strategy's board approved the Digital Credit Capital Framework. It is the most important corporate document the company has published since it adopted a bitcoin standard in August 2020, and it is worth setting out in full because it defines the rules under which the equity now operates.

At 20 September 2026 the USD Reserve stood at $5.04 billion and USD Cash at $1.05 billion. Together, $6.09 billion of dollar liquidity against $21.2 billion of senior claims. The company states that the reserve covers more than two years of preferred dividends and debt interest, and on the current run rate of $1.76 billion a year, $5.04 billion covers 2.9 years. Both statements are arithmetically correct.

The problem is that the dividend strip is only one of the two obligations.

The collision

Run the two clocks together from August 2026 to September 2028 — a window of approximately 2.1 years. Cumulative preferred dividends and note interest over that period come to roughly $3.67 billion. Convertible principal puttable inside the same window comes to $5.91 billion. The combined potential call on cash is $9.58 billion. The reserve was $4.65 billion at 9 August 2026.

The gap is roughly $4.9 billion, and it has to be sourced from a business that generated $9.85 million of operating cash flow in the first half of 2026. The reserve covers the dividends or the puts. It does not cover both, and the two overlap.

This is a scenario rather than a forecast. Holders may not exercise their puts, particularly if the shares recover and the notes move back toward or above their conversion prices. The company may repurchase further notes at a discount, as it did in May. It may issue new securities. It may sell bitcoin. It has, in fact, three of those four options in active use. But the arithmetic frames the question that the rest of this report tries to answer: what has to happen for the structure to keep working, and what does the equity look like if it does not?

12 — MonetizationThe company sold 6,949 bitcoin and the market barely noticed

For five years the defining feature of Strategy's public identity was that it never sells. Michael Saylor has repeatedly described bitcoin as an asset to be held indefinitely, and the company's disclosures emphasised accumulation. That ended on 31 May 2026, when Strategy reported the sale of 32 bitcoin for approximately $2.5 million — the first reported sale since December 2022.

The 32 coins were immaterial. What followed was not. In the week to 5 July the company sold 3,588 coins for roughly $216 million. In the week to 2 August it sold a further 1,638 coins for $104.7 million, allocating $52.4 million to preferred dividends and $52.3 million to STRC repurchases. By 14 August holdings had fallen to 840,447. In total, between 26 May and 20 September, Strategy disposed of approximately 6,949 coins.

The company then bought back in. On 31 August it raised $602.8 million through the sale of 4,531,421 class A shares and used $369.7 million of the proceeds to acquire 4,603 coins at an average price of $80,318, $151.8 million to repurchase 1,557,177 STRC shares, and $50.7 million for STRC dividends. Holdings rose to 845,050. In the week to 20 September it added 950 coins for $75.7 million while repurchasing 1,771,238 STRC shares for $174.0 million.

Net of everything, the coin count rose by 2,225 in the third quarter. Gross, the company sold 6,949 and bought 5,553.

Why this is the most important disclosure of 2026

Two things changed at once, and both matter.

First, the reserve is no longer a one-way ratchet. The company's own framing has shifted from accumulation to what it calls "active capital management," in which bitcoin is one of several sources of liquidity alongside equity issuance, preferred issuance and the dollar reserve. The investor briefing lists monetisation as a standing tool: "Sell limited BTC for the USD Reserve, obligations or authorized buybacks when management judges it superior to issuing capital."

Second — and this is the part that is easy to miss — the market barely reacted. Bitcoin slipped below $72,000 within hours of the first sale disclosure in late May, but the subsequent and much larger sales of 3,588 and 1,638 coins in July and August passed with limited price impact. Strategy's own supplemental metrics assume that bitcoin sales "have no effect on the market price of bitcoin," and for sales of this size, that assumption has so far held.

The inference is that a managed programme of a few thousand coins a month is not a threat to the bitcoin market. The corollary is less comfortable: the assumption is size-dependent and would not survive an acceleration. If the 2027 convertible puts were exercised and the company had to source $5.9 billion in a twelve-month window, it would need to sell in the region of 70,000 coins at current prices, or roughly 8% of the reserve, or issue securities into a market that had just declined to roll its debt. The difference between 3,588 coins and 70,000 coins is the difference between a monetisation programme and a liquidation.

13 — Dilution187,742 sats gross, 153,955 sats net, and the 18% that belongs to the capital structure

Strategy's own per-share framework is the best available tool for separating the operating story from the financing story, and it is unusually honest about what it excludes.

Gross bitcoin per share is the total coin count divided by assumed diluted shares outstanding. At 13 September 2026 that was 187,742 satoshis. Net bitcoin per share deducts the notional value of out-of-the-money convertible notes and preferred stock, net of dollar assets, before dividing. That figure was 153,955 satoshis.

The gap is 33,787 satoshis, or 18% of the gross figure. That 18% is the capital structure expressed per share: the share of each coin that has been sold forward to the preferred and convertible holders in exchange for the cash that bought the coin. It is the single cleanest measure of what the leverage costs.

The company is careful to state the limitations, and they are material. The deductions are made at notional rather than at liquidation value. Whether an instrument is in or out of the money is determined by market prices on the measurement date and can change materially. The offsetting of senior claims is assumed to be funded first from dollar assets and then from bitcoin sales, with no market impact. The metrics exclude transaction costs, taxes and other expenses, and they do not account for deferred tax, operating leases, legal contingencies and similar liabilities which the company notes "could be material." Most importantly, the company states that net bitcoin per share "does not represent the amount, if any, that holders of common stock would receive in respect of the Company's bitcoin in a liquidation."

Has the machine actually worked?

On the company's numbers, gross bitcoin per share rose from 56,598 satoshis in 2020 to 187,742 satoshis in September 2026 — a multiple of 3.32× over six years. That is a genuine achievement and it is the strongest quantitative evidence for the bull case. Roughly speaking, a shareholder who has held since 2020 now owns 3.3 times as much bitcoin per share as they did at the start, and the bitcoin itself has appreciated.

Two qualifications. First, the net figure has grown less: the senior stack was small in 2020 and is $21.2 billion today, so net bitcoin per share has compounded more slowly than the gross figure the company leads with. Second, and more important, the accretion is a product of the premium regime. Between 2020 and 2024 the shares traded at a substantial multiple of net asset value, and every issuance was accretive. In 2026 that regime ended. The BTC yield — the company's headline accretion metric — has halved in a year.

BTC yield was 8.7% in the second quarter of 2025. It was 9.4% year-to-date in the first quarter of 2026, then 5.0% in the second quarter, and 4.5% year-to-date for the first half. The company attributes the decline to a larger bitcoin base and a lower common-stock premium, which is the same statement as saying the flywheel slowed. A larger base mechanically reduces the percentage accretion from a given dollar of issuance; a lower premium reduces the coins that dollar buys. Both effects are permanent rather than cyclical.

14 — Quality of earningsEvery number in the income statement is either tiny or a mark

Strategy presents an unusually clean case study in earnings quality, because the distortion is disclosed rather than buried. Under ASU 2023-08, which the company early-adopted on 1 January 2025, bitcoin is measured at fair value with changes recognised in net income. Before adoption, the company used an impairment model that wrote coins down but never up, producing a systematically understated balance sheet and no upward marks in the income statement.

The new standard is more economically faithful and it makes reported earnings far less useful. Consider the four quarters from the start of 2025 to the middle of 2026:

Over four quarters the company recorded $32.7 billion of fair-value swings against $472 million of cumulative revenue — a ratio of roughly seventy to one. The mark is not an adjustment to earnings; it is earnings, and the operating business is a footnote.

What a careful reader should do instead

Three adjustments restore most of the information content.

First, ignore earnings per share entirely. The diluted EPS series of $(38.25) and $(24.45) for the two quarters of 2026 carries no information about the company's performance. The consensus estimate for the second quarter was $(7.67) and the company reported $(24.77) on the adjusted basis the street uses — a miss of 223%. Analysts are modelling the same price they can see on a screen; the "miss" is a difference in the bitcoin price assumption, not a difference in the business.

Second, read the cash flow statement. Net cash from operating activities was $9.85 million in the first half of 2026. Investing consumed $14.37 billion. Financing provided $13.77 billion. Depreciation and amortisation of $21.14 million year-to-date is roughly twice the operating cash flow, which means that on a maintenance-adjusted basis the software business is close to cash-neutral.

Third, track three balance-sheet items. The coin count, the notional value of the senior stack, and the dollar liquidity. Those three numbers, plus the bitcoin price, determine everything that matters. In the second quarter of 2026 the coin count rose 3.2%, the senior stack rose materially through $5.47 billion of STRC issuance, and dollar liquidity rose from $2.21 billion of cash at 31 March to $1.71 billion of cash plus $736.1 million of short-term investments at 30 June.

15 — Capital allocation$20.92 billion raised in nine months, and the machine that spends it

Strategy's capital-raising capacity is the asset the market has historically paid a premium for, and 2026 was the largest year in the company's history. Through 13 September it had raised $20.92 billion: $13.40 billion of common equity and $7.52 billion of preferred stock. Of the preferred total, $7.53 billion was STRC — growth of 254% year to date.

The quarterly pattern is instructive. In the first quarter the company raised $7.37 billion, with a further $4.32 billion between 1 April and 3 May. In the second quarter it raised $8.41 billion — $5.47 billion of STRC and $2.95 billion of class A common — and a further $1.28 billion between 1 and 26 July. Then it stopped: no shares were sold under the at-the-market programme in the weeks to 13 September or to 20 September, and none in the week to 2 August after a resumption that raised $290.6 million.

The pause coincides with the shares trading below net asset value. When issuance is dilutive, a rational issuer stops issuing. The company's own investor materials state the decision test explicitly: issue when "the expected BTC-per-share, liquidity or refinancing benefit exceeds the cost of capital and dilution." At parity, that test fails.

The three-way allocation

What makes Strategy's capital allocation unusual is that it is allocating between four competing uses, three of which are financial rather than operational:

• Buy bitcoin. The original use, now conditional on the premium.
• Build the USD Reserve. Funded partly by bitcoin sales; its purpose is to service the securities that funded it.
• Repurchase securities. $875 million of STRC retired at discounts averaging roughly 13% to stated amount; $1.0 billion of common authorisation untouched.
• Pay distributions. $1.06 billion of cumulative preferred dividends paid since the programme launched in early 2025.

The company frames this as "from one-way issuance to dynamic capital allocation," and the framing is fair. It is genuinely more sophisticated than the one-directional accumulation programme of 2020 to 2024. But it is also a response to constraint rather than a strategic choice, and the allocation it produces is revealing: when forced to rank the uses, management ranked buying back its own discounted preferred stock above buying bitcoin and above buying back its own common stock.

That ranking is defensible on the numbers. STRC bought at $86.53 against a $100 stated amount with a 12% coupon produces a high, certain return, and it reduces the perpetual claim. Common stock bought at a discount to net reserve per share produces an uncertain return that depends on the market re-rating the equity, which is precisely the condition that has failed. A shareholder should nevertheless be clear about what the ranking implies: management's highest conviction use of capital is retiring its own preferred obligations, not accumulating bitcoin.

16 — Balance sheet & taxThe preferred stock sits between the debt and the equity, and that is where the risk sits too

Strategy's balance sheet at 30 June 2026 reported total assets of $52.56 billion, total liabilities of $7.24 billion and stockholders' equity of $30.89 billion. Assets less liabilities leaves $45.32 billion, which is $14.43 billion more than the equity figure. That difference is the preferred stock.

The classification is worth pausing on, because it is the clearest accounting statement of where the risk lives. Strategy's perpetual preferred is carried as mezzanine equity — between liabilities and common stockholders' equity — because the instruments are redeemable at the company's option and carry a liquidation preference above the common. The convertible notes sit in liabilities at $6.71 billion. The common sits at the bottom of a stack that has $21.2 billion of claims ahead of it.

Common stockholders' equity fell from $47.48 billion to $30.89 billion over twelve months, a decline of $16.59 billion. That decline is the fair-value loss on the reserve, partly offset by the roughly $13.8 billion of securities issued over the same period. A shareholder looking for evidence that the structure has been accumulating value for the residual claim will not find it in the book equity line.

The liquidity ratios look healthy and are largely meaningless in this context. The current ratio is 5.39 and debt to equity is 0.23. Both figures are computed against a balance sheet whose principal asset is a volatile bearer instrument and whose principal obligations are perpetual preferred claims that never appear in current liabilities. A conventional credit analysis applied to Strategy's financial statements will produce a comfortable conclusion that has nothing to do with the actual risk.

The tax question

Strategy is taxed as a C corporation and pays federal income tax at the entity level on its taxable income. This is one of the structural differences from a spot bitcoin ETF, which is typically organised as a trust and does not pay entity-level tax.

In practice the company's tax position is dominated by the bitcoin reserve, and the interaction between fair-value accounting and the tax code produces an unusual outcome. The fair-value marks are recognised in book income but are not recognised for tax until the coins are sold. In a year of large unrealised gains, book income exceeds taxable income by a wide margin, and the company records a deferred tax liability. In a year of unrealised losses, the position reverses. The company's 10-Q identifies as a specific risk that "tax benefits from capital losses may be limited without capital gains to offset them" — meaning that if the company sells bitcoin below its $75,416 average cost, the resulting capital loss may not be usable against ordinary income, and could expire unused.

That is a real constraint on the monetisation strategy. Selling coins at a gain is tax-efficient relative to ordinary income; selling them at a loss produces a capital loss that may be stranded. At 21 September the reserve is $8.3 billion above cost, so the company can currently monetise without triggering a stranded loss. A return to the June 2026 price of $58,714 would remove that option precisely when the company might need it.

On the distribution side, the company states that it expects preferred dividends to be treated as non-taxable returns of capital for US federal income tax purposes for the foreseeable future, on the basis that it expects no accumulated or current earnings and profits. For a taxable investor in STRC, that converts a 12% cash distribution into a 12% return of basis — attractive in the near term, and a deferral rather than an elimination of the eventual tax.

17 — Custody & operationsFour hundred and twenty-two thousand coins per custodian, and one person's judgement

Strategy holds substantially all of its bitcoin with a limited number of US-based institutional custodians. The company's second-quarter 10-Q identifies the concentration as a high-severity risk factor in its own words: "Substantially all bitcoin is held with a limited number of U.S.-based custodians; a failure, insolvency, or security breach at any major custodian could result in loss or prolonged inaccessibility of a material portion of the company's assets."

This is not a theoretical risk, and the structure makes it worse than it would be for a smaller holder. Strategy's coins are held in a small number of custody arrangements with associated insurance, and the insurance limits are not publicly disclosed at a level that would cover a 4% share of global supply. A custodian failure would not merely impair a portfolio; it would impair the collateral that stands behind $21.2 billion of senior claims, and would do so at the worst possible moment for the company's access to capital markets.

The mitigating factors are real. Institutional custody with multi-signature and segregated wallet arrangements is materially more robust than the exchange custody that destroyed value in 2022. Strategy has disclosed its custodian arrangements and its controls. And the company's holdings are the single most publicly scrutinised corporate bitcoin position in existence, which creates a strong incentive for operational discipline.

Key-person concentration

The second operational risk is governance. Strategy's strategy is inseparable from Michael Saylor, who has been the company's principal public advocate for bitcoin for six years, whose personal holdings align him with the common stock, and whose commentary is a material part of the securities' marketing. The company has professional management — Phong Le is CEO and the capital-markets function is staffed — but the franchise's brand value is concentrated in one individual.

The company's own risk framework lists "key-person / capital-markets execution" as a concentration. That is a candid disclosure and it deserves to be taken at face value: an investor is underwriting a judgement about a single person's continued commitment and effectiveness, in addition to a judgement about bitcoin.

The metric-limitation risk, which is a real one

A third operational consideration is less obvious and, in our view, underappreciated. Strategy publishes a large number of supplemental metrics — mNAV, net bitcoin per share, amplification, BTC Rating, BTC ARR bands — none of which are GAAP measures, and all of which depend on assumptions the company selects. The company has itself redefined mNAV and amplification, in July 2026, and states that figures calculated before that date are not comparable with figures calculated after it. It publishes gross and net per-share bitcoin figures that, on external analysis, do not appear to share a denominator.

None of this is improper. Every one of the metrics is accompanied by a disclosure of its limitations, and the disclosures are unusually thorough. But a shareholder who has been persuaded by a headline metric — "bitcoin per share is up 3.32× since 2020" — should read the footnote, which states that the measure "does not represent the amount, if any, that holders of common stock would receive in respect of the Company's bitcoin in a liquidation."

18 — Index eligibilityA $2.8 billion overhang and a decision due on 16 October

Strategy's index status is the most under-discussed material issue facing the equity, and it resolves, or does not, within weeks of this report's publication.

The company is a constituent of the Nasdaq-100, having been added in December 2024. It is not a constituent of the S&P 500, despite having met the eligibility criteria on several occasions; S&P Dow Jones Indices declined to add the company at both the September 2025 and November 2025 reviews, and it remained absent as of mid-2026.

The live question is MSCI. In January 2026 MSCI closed a consultation on digital-asset treasury companies and decided not to exclude them from its global indexes, saying it would instead study the broader category of companies whose activities appeared predominantly investment-oriented. On 3 August 2026 it opened that broader consultation, proposing a framework to identify "non-operating companies" through their financial statements.

The proposed methodology has two stages. A company first faces a core screen measuring operating assets as a percentage of total assets; above 50% it passes. Companies below that threshold move to a second stage of five financial tests: operating asset intensity, operating expense intensity, cash generation, exposure to fair-value movements, and dependence on external capital. Triggering at least four of the five flags after failing the core screen makes a company ineligible. Existing index constituents get more room — the thresholds are tighter — and must fail across two consecutive annual reviews before removal.

Applied to May 2026 data, MSCI's own simulation produced three deletions from the ACWI Investable Market Index: Strategy, Metaplanet, and Yellow Cake, a London-listed uranium investor. The inclusion of Yellow Cake is significant because it demonstrates that the framework is not a crypto-specific rule; it is a general test that happens to catch asset-holding vehicles.

The direct cash consequence of an exclusion is nil: index selling affects the share price, not the company's cash, and it would not oblige Strategy to sell a single coin. The indirect consequence is more important, and it is the reason this matters. Strategy's entire growth model depends on issuing shares at a premium. A large wave of index-related selling would compress that premium, weakening the economics of every future issuance, and would do so at the moment the company needs the equity channel to address the 2027–2028 convertible puts.

On external analysis, Strategy's likely exposure is close to the line. One published assessment concludes the company would fail on operating asset intensity, expense intensity and fair-value exposure, while positive operating cash flow and financing cash flow of roughly 26% of assets could allow it to avoid the other two flags — leaving three failures against a four-test requirement. That analysis is not dispositive, and the two-consecutive-review persistence requirement means an adverse decision would not translate into immediate deletion. But it does mean the overhang is unlikely to disappear quickly even if the October decision is benign.

19 — RegulationThe CLARITY Act, the Fed, and the accounting rule that changed everything

Three regulatory threads run through Strategy's investment case, and they pull in different directions.

The CLARITY Act

The Digital Asset Market Clarity Act — the market-structure bill that would establish a federal framework distinguishing digital assets that are securities from those that are commodities — passed the House in late 2025 and was advanced by the Senate Banking Committee on 14 May 2026 by a bipartisan vote of 15 to 9. It remains pending before the full Senate.

For Strategy the bill's significance is indirect but real. A clearer federal framework would reduce the regulatory discount embedded in all digital-asset exposures, broaden the institutional investor base, and reduce the risk that the preferred securities or the common are recharacterised. It is the single most significant potential catalyst in the sector, and it is also entirely outside the company's control and subject to a legislative calendar that has already slipped once.

Monetary policy, which has turned against the trade

On 16 September 2026 the Federal Reserve raised its target range by 25 basis points to 3.75%–4.00%, its first increase in three years, following August CPI of 3.4% year over year. The market's expectation of the hike had risen from roughly 70% to 86.5% in the days before the meeting as the inflation data landed.

This is a material change in the environment in which Strategy operates. Bitcoin's 2025 rally was substantially a liquidity trade, and the 2026 drawdown has been substantially a liquidity reversal: tariffs, delayed cuts, and institutional rotation out of risk assets. A hiking cycle raises the discount rate applied to a zero-coupon, zero-yield asset, raises the company's cost of credit, and raises the required return on the reserve. It also raises the bar for the preferred coupons the company must pay to clear at par.

The countervailing fact is that bitcoin rose 24.9% in August 2026 and a further 10% in the first three weeks of September, through a hiking cycle. The asset's relationship to real rates is less mechanical than the standard model implies, and the September rally suggests the market is pricing something else — policy clarity, institutional adoption, or simply the exhaustion of the sellers. We do not claim to know which.

The accounting rule, which is the one that already happened

ASU 2023-08, adopted by Strategy on 1 January 2025, requires digital assets to be measured at fair value with changes recognised in net income. The company early-adopted, which had a secondary benefit: the standard removed a barrier to index inclusion under the older impairment regime, because a company that had to impair its assets and could never write them up could not show consistent profitability.

The rule is now a double-edged instrument. It makes Strategy's balance sheet economically faithful and its income statement volatile to the point of uselessness. It also means that the company's reported profitability will alternate between enormous and negative with no relationship to the operating business, which is likely to keep it out of the S&P 500 for as long as the index committee applies a profitability screen.

20 — Competition IThe benchmark is not another company. It is a 0.25% fee.

The most important competitive fact about Strategy is that an investor who wants bitcoin exposure has, since January 2024, been able to buy it directly. A spot bitcoin ETF holds the asset in trust, tracks its price with negligible tracking error, publishes its holdings daily, pays no entity-level tax, cannot take on leverage, and charges a fee that for the largest products is 0.25% a year.

Strategy is not competing with that product on cost. It is offering something different: a residual claim on a levered reserve, with the leverage supplied by $21.2 billion of senior securities and the operation supplied by a $20.92-billion-a-year capital-raising machine. The pitch is that the leverage amplifies returns, that the machine grows bitcoin per share, and that the two together beat holding the coin.

That pitch is testable, and the twelve months to 18 September 2026 provide an unambiguous answer. Bitcoin fell 25.9%. An ETF holding bitcoin and charging 0.25% would have returned approximately −26%. MSTR returned −55.9%. The wrapper cost thirty percentage points.

The same comparison over four years produces the opposite answer: MSTR's annualised return since adopting a bitcoin standard is 47%, against bitcoin's 36%. The wrapper added eleven points a year — in a period when the shares were repriced from a fraction of net asset value to a multiple of it.

There is a second-order effect that runs in Strategy's favour. ETF flows are pro-cyclical, and 2026 demonstrated how violently they can reverse: US spot bitcoin ETFs posted a record nine-session outflow streak of roughly $2.8 billion in late May, with total ETF net assets falling from over $100 billion to $85 billion by early June. Redemptions require authorised participants to sell actual bitcoin into the spot market mechanically. Strategy's holders, by contrast, cannot redeem: the only way out is to sell the shares, and the company's coin holdings are unaffected by who owns the equity. In a severe drawdown, Strategy's reserve is structurally more stable than an ETF's.

That is a genuine advantage, and it is the strongest structural argument for the treasury company as a vehicle. It is also worth noting what it is worth: an investor who wants drawdown-resistant bitcoin exposure can get it from a self-custodied position at no fee at all.

21 — Competition IIThe cohort copied the model and inherited the compression

Strategy's success created an industry. A mid-year review by 21Shares counted roughly 200 publicly listed companies holding close to 1.28 million bitcoin in aggregate — a universe that includes miners and operating companies as well as pure treasury vehicles. Within the pure treasury cohort, the ranking is not close.

Strategy holds 846,000 coins. Twenty One Capital holds 43,514. Metaplanet holds approximately 43,000. Strive holds 15,009. Capital B holds 3,139. Strategy's holding is nineteen times the second-largest and 66% of the cohort shown.

Scale is the franchise, and the copycats do not have it

The scale gap is not merely a matter of size. Strategy's advantage is that it has the deepest access to capital markets of any treasury company: $2.86 billion of average 30-day trading volume, $44.5 billion of options open interest, five listed preferred series, an investment-grade-adjacent distribution network and a brand that has been built over six years. That access is what allows it to issue $20.92 billion in nine months. A company holding 43,000 coins cannot issue at Strategy's cost of capital, because it does not have Strategy's distribution.

The copycats also inherited the compression without inheriting the franchise. Every one of them has seen its premium to net asset value contract as the bitcoin price fell and the market reassessed the model. Metaplanet's structure depends on warrants that can materially expand the share count. Twenty One Capital carries approximately 16,116 coins pledged as collateral against convertible notes, which changes how much of its holdings are genuinely available to the residual claim. Strive funds itself with SATA preferred carrying a 13% dividend, higher than Strategy's STRC. Capital B, at 3,139 coins, is too small to fund itself at scale and relies on at-the-market issuance, warrant subscriptions and private placements.

22 — The macro backdropA hiking cycle, a 53% drawdown, and a rally nobody predicted

Strategy's 2026 has been shaped by three macro forces, and the equity has responded to all three more violently than the underlying asset.

The drawdown. Bitcoin set its record high at $126,210 on 6 October 2025. Four months later, on 6 February 2026, it fell 15% in a single session to near $60,000 — a 52% decline from the peak. The causes were not crypto-native. Tariff announcements pushed inflation expectations high enough to take rate cuts off the table, and CME FedWatch showed the probability of a March cut collapsing from over 20% to 6%. Rate cuts had been the fuel behind the 2025 rally, and without them institutional capital rotated out of risk assets. The same ETF infrastructure that had channelled billions in during 2024 and 2025 ran in reverse: redemptions require authorised participants to sell actual bitcoin into the spot market, mechanically and at scale.

Leverage amplified everything. The June 3 session alone produced $1.86 billion of liquidations in twenty-four hours. Each macro shock created a second wave of forced selling on top of the first, and the cycle low of $58,714 was set on 30 June 2026 — 53% below the October peak.

Then it reversed. Bitcoin rallied 24.9% in August and a further 10% in the first three weeks of September, reaching $85,261 by 21 September. The rally came through a Federal Reserve that raised rates by 25 basis points to 3.75%–4.00% on 16 September — its first increase in three years, delivered against August CPI of 3.4%. That a zero-yield asset rallied through a hike is either a statement about the exhaustion of the sellers, or about the market pricing something other than the policy rate. Either way, it is the environment in which this report is written.

The equity's response to all of this is worth dwelling on, because it is the clearest available evidence about what MSTR actually is. The shares bottomed at $92.52 on 18 August and closed at $153.92 on 18 September — a 66% advance in twenty-two sessions, against a bitcoin move of roughly 35% over the same period. The beta is not stable and it is not symmetric: MSTR fell 56% over the trailing twelve months against bitcoin's 26%, and rose 66% in a month against bitcoin's 35%. Over both horizons the equity amplified the asset by roughly 1.9 to 2.2 times.

Historical volatility tells the same story with more precision. Thirty-day annualised historical volatility was 101.2% at 18 September, against 80.6% for the trailing year, and implied volatility was 54% — a spread that implies the options market expects the recent realised volatility to decay. Bitcoin's own volatility over comparable periods has been materially lower. A shareholder is buying roughly twice the volatility of the underlying asset, which is what leverage does.

23 — ValuationThree cases, a fourfold spread, and what the current price already assumes

There is no conventional valuation framework that works on Strategy. There is no free cash flow to discount, no earnings multiple that survives contact with fair-value accounting, and no comparable-company set — the closest analogues are a bitcoin ETF with no leverage and a set of treasury companies with a fraction of the scale and none of the capital-markets access.

The only coherent approach is asset-based, and it has four steps: value the coins, add the dollar liquidity, subtract the senior claims, and apply a multiple to what remains. The multiple is the entire debate.

The model

We hold four inputs constant across all three cases, because they are the ones we can observe and they move slowly: a coin count of 835,000 (reflecting modest continued monetisation against a reduced ability to issue accretively), USD assets of $4.0 billion (a partial drawdown of the current $6.1 billion as the reserve funds distributions), senior claims of $21.0 billion at notional (broadly unchanged, assuming the puts are refinanced rather than repaid), and fully diluted shares of 432 million (some dilution net of repurchases).

The two variables are the bitcoin price at September 2027 and the mNAV the market applies.

The reverse reading

The more useful exercise is to invert the model. At $153.92, what is the market assuming?

If we apply parity — a 1.00× mNAV, which is where the shares have traded for much of 2026 — the current price implies a net reserve of $66.5 billion. Working backwards through the same four inputs, that requires a bitcoin price of approximately $100,000, or 17% above the spot price at the time of writing. If we are more generous and apply a restored 1.15× premium, the implied bitcoin price falls to approximately $89,600 — only 5% above spot.

That is the crux of our Neutral rating. At a premium, the current price is defensible: it is roughly what a 1.15× multiple on a $90,000 bitcoin price would produce. At parity — which is where the shares have actually traded for much of the last nine months — the current price requires bitcoin at $100,000, a level it has not seen since February 2026.

A shareholder is therefore not simply long bitcoin. They are long bitcoin and long the re-establishment of a premium that has been absent for most of 2026, and they are paying for that second bet with a 17% premium to net reserve per share.

24 — The bull caseWhat has to be true, and how much of it already is

The strongest version of the bull case does not rest on the flywheel or the premium. It rests on four observable facts, and each of them is defensible on the evidence.

One: the reserve is above cost and the franchise is irreplaceable. Strategy owns 846,000 bitcoin — 4.03% of the total supply that will ever exist — at an average cost of $75,416 against a market price of $85,261. No other corporate buyer can accumulate at this scale without moving the market, and no other treasury company has the capital-markets infrastructure to fund it. The reserve is a genuine, non-replicable asset.

Two: the preferred franchise is real and it is cheap relative to the alternative. STRC is the largest listed preferred instrument in the world by market capitalisation, at $9.5 billion notional, trading within 1.5% of par with 30-day historical volatility of 11.5% and daily volume of $140.6 million. That is a functioning, liquid, institutional-grade fixed-income product. Strategy raised $20.92 billion in nine months through it and the common channel, and it still has $19.09 billion of unused capacity. A company that can raise $20 billion a year on demand has real optionality.

Three: the duration mismatch is genuinely favourable, and it is the most underappreciated fact in the structure. Strategy's own dashboard reports a bitcoin duration of 44.3 years against a USD duration of 3.8 years. The company has financed a permanent, non-cash-flowing asset with long-dated and perpetual liabilities. The preferred has no maturity at all. The convertibles, even with their puts, do not come due before 2029. There is no analogue of a margin call, no mark-to-market collateral requirement, and no lender who can force a sale. The reserve is $8.3 billion above cost and there is no mechanism by which a short-term price move can trigger liquidation. That is a structurally stronger position than almost any levered bitcoin holder, and it is why the company survived a 53% drawdown without distress.

Four: the arithmetic of the hurdle is favourable if you believe the long-run evidence. Bitcoin has returned 36% annualised over four years and 47% annualised since Strategy adopted a bitcoin standard. The hurdle rate is 10.70%. If bitcoin compounds anywhere near its historical rate, net bitcoin per share compounds faster than bitcoin, and the common equity is the best available expression of that. The company's own spread illustration makes the point: at a 20% bitcoin return, $1 billion of Digital Credit generates roughly $93 million of annual economic spread, and the 10%–20% annual issuance target implies $0.6–1.2 billion a year of it.

Five: the catalysts are real and dated. The MSCI decision on 16 October could remove a $2.8 billion overhang. The CLARITY Act, already through the House and out of the Senate Banking Committee, would establish a federal framework for digital assets. Q3 results on 28 October will show a fair-value gain of roughly $22 billion. And the September rally has already restored $9.1 billion of market capitalisation in a single session, on volume of $8.4 billion, with $44.5 billion of options open interest behind it.

25 — The bear caseWhat has to be true for the shares to work, and why we are not paying for it

The bear case is not that bitcoin goes to zero. It is that the structure transfers value away from the common shareholder in every scenario except a strong bitcoin rally, and that the shareholder is not compensated for that asymmetry at the current price.

One: the flywheel has inverted and there is no neutral setting. Below parity, every dollar of equity issued buys less than a dollar of bitcoin, and net bitcoin per share falls. Strategy has stopped issuing common stock under its at-the-market programme — the disclosed reason is that the economics no longer work. The company's own BTC yield metric has halved in a year, from 8.7% to 5.0% quarterly and 4.5% year-to-date. The 3.32× growth in gross bitcoin per share since 2020 was manufactured by a premium regime that no longer exists, and the metric cannot be reproduced from here without a re-rating.

Two: the structure consumes cash the business does not generate. Annual preferred dividends and note interest total $1.76 billion. Net cash from operating activities in the first half of 2026 was $9.85 million, annualising to roughly $19.7 million — about 1.1% of the obligation. The gap has been funded by issuing securities and by selling bitcoin. The company sold 6,949 coins between May and September, and it built a $5.0 billion reserve whose stated purpose is paying dividends on the securities that funded it. That is a closed loop, and closed loops require continuous access to capital markets to keep turning.

Three: two obligation clocks collide in 2028 and the reserve covers one of them. 88.1% of the $6.714 billion convertible principal becomes puttable inside a twelve-month window opening in September 2027. Cumulative dividends plus puttable principal through September 2028 is roughly $9.6 billion against a $5.0 billion reserve — a gap of $4.9 billion that must be sourced from a business generating $20 million a year. The most likely resolution is refinancing, and refinancing at a higher cost raises the hurdle rate that the common shareholder is short.

Four: the cost of credit rises exactly when confidence falls. STRC's coupon has been reset upward from 9% at launch to 12.00%. The hurdle rate — the company's effective cost of credit — was 10.70% at 13 September and 10.8% at the second quarter. If STRC needs to clear at a higher rate to hold near par, the hurdle rises, and the bitcoin return required for the common to break even rises with it. A shareholder is exposed to Strategy's credit spread in addition to bitcoin, and the correlation runs the wrong way.

Five: the preferred market is already pricing credit deterioration. Three of the five series trade below their $100 stated amount. STRK is 24.6% below par; STRD, the non-cumulative series, is 25.8% below and yields 13.48%. Only STRF, the most senior, trades above par. If the company's own credit was considered sound, perpetual preferreds with 8%–12% coupons would not be trading at 25% discounts. The preferred market is telling a different story from the common stock, and the preferred market is the one that holds the senior claim.

Six: index risk lands on the mechanism, not the cash. MSCI decides by 16 October on a framework that its own simulation would apply to delete Strategy from the ACWI IMI, with an estimated $2.8 billion of passive selling and prediction markets pricing roughly a 73% probability of removal from MSCI World or MSCI USA by year-end. The company is also absent from the S&P 500 despite meeting the eligibility criteria at two consecutive reviews. Index selling would compress the premium, which is the input to the flywheel, at precisely the moment the company needs to issue equity to address the 2028 puts.

Seven: the metric architecture is doing work that the disclosure does not support. The company publishes a headline accretion metric that excludes senior claims, has redefined mNAV and amplification within the last three months, publishes gross and net per-share figures that on external analysis do not share a denominator, and describes a $50.5 billion "net reserve" that its own glossary states is not net asset value, not liquidation value, and not the amount a shareholder would receive. None of this is improper. All of it means that an investor persuaded by a headline number has been persuaded by a construction whose limitations the company itself spends four pages describing.

26 — Risk matrixTwelve risks, scored by probability and impact

Cells read left to right as impact rises from 1 to 5, top to bottom as probability rises from 1 to 5. Superscripts identify risks from the register above.

27 — CatalystsWhat we are watching, and when

28 — ConclusionNeutral at $145: a franchise wrapped around a liability structure

Strategy is the most interesting capital-structure experiment in the public market, and it deserves to be analysed as one rather than as either a bitcoin cult or a fraud. On the evidence of 2026 it is neither. It is a company that built a genuine arbitrage — a share-price premium converted into bitcoin per share — and then watched the arbitrage close, and has responded by converting itself into a financial intermediary that issues securities against a reserve and services the resulting obligations from the proceeds of further issuance.

That response has been competent. The company survived a 53% drawdown in its only asset without a forced sale, a covenant breach or a missed distribution. It built a $5.0 billion dollar reserve, retired $875 million of its own discounted preferred, repurchased $1.5 billion of convertible notes below par, and raised $20.92 billion in nine months. Anyone who predicted that the structure would break in a bitcoin bear market was wrong, and 2026 is the evidence.

Competence is not the same as value for the residual claimant. The three facts that decide the equity are these. First, the mechanism that generated 3.32× growth in gross bitcoin per share since 2020 requires a premium that no longer exists; the company's own accretion metric has halved in a year. Second, the structure consumes $1.76 billion a year against an operating business that generates $20 million, and the gap is funded by issuing securities and selling coins — a loop that needs continuous market access. Third, 88.1% of the convertible principal becomes puttable inside a twelve-month window opening in September 2027, against a reserve that covers the dividends or the puts but not both.

Against that, the shareholder receives a 1.24× amplification of an asset they can buy directly for 25 basis points a year. The premium to net reserve per share is approximately 17%, and at parity the current price already requires bitcoin at $100,000 — a level not seen since February 2026.

What would change our mind, in order of importance. A bitcoin price sustained above $95,000 would put the reserve far enough above cost that the dividend strip becomes a rounding error and the hurdle rate becomes a modest barrier rather than a demanding one. A resolution of the MSCI consultation in the company's favour would remove the $2.8 billion overhang and give the premium a chance to re-establish. And a demonstrated ability to refinance the 2027–2028 puts at or near par — disclosed in advance rather than negotiated in the window — would tell us that the closed loop can keep turning through a bad bitcoin year.

None of the three is in evidence today. Two of the three are scheduled to resolve within five weeks of publication.

29 — AppendixFinancial summary and sources

Certain figures are derived by subtraction from reported period totals and are labelled as such where they appear. The FY2025 coin count is derived from the 5 and 12 January 2026 purchase disclosures. The USD Reserve is a management-designated figure first formalised in the June 2026 Digital Credit Capital Framework. Preferred stock notional at Q1 2026 is disclosed as "over $13.5 billion."

Sources

• Strategy Inc. quarterly earnings releases, 5 May 2026 and 30 July 2026, and Form 10-Q filings for the quarters ended 31 March 2026 and 30 June 2026.
• Strategy Inc. Form 8-K weekly bitcoin and capital-markets disclosures, including the filings of 6 July, 3 August, 31 August, 14 September and 21 September 2026.
• Strategy Inc. Form 8-K of 29 June 2026 establishing the Digital Credit Capital Framework, the USD Reserve policy, the repurchase authorisations and the BTC Monetization Program.
• Strategy Inc. MSTR Investor Briefing, 13 September 2026, and the accompanying glossary of defined terms and metric limitations.
• Strategy Inc. investor dashboard at strategy.com, market data as of 18–21 September 2026, and the STRC security page as of 18 September 2026.
• Strategy Inc. FY2025 Form 10-K and Form 8-K purchase announcements of 5 January, 12 January and 2 March 2026.
• Nasdaq daily closing prices for MSTR, July to September 2026; exchange spot data for bitcoin; CME FedWatch for rate probabilities.
• MSCI consultation on the eligibility of non-operating companies for the MSCI Global Investable Market Indexes, opened 3 August 2026; MSCI announcement of 7 January 2026 on digital-asset treasury companies.
• Federal Reserve FOMC statement and implementation note, 16 September 2026; US Bureau of Labor Statistics CPI release for August 2026.
• S&P Global Market Intelligence and TipRanks consensus data as of 15 September 2026; published analyst actions from Barclays, Canaccord Genuity, B. Riley Securities, TD Cowen and Alliance Global Partners.
• Third-party valuation trackers for market-to-net-asset value; published analysis of the convertible put schedule and the annualised capital-structure cash cost.
• 21Shares mid-year review of listed bitcoin holders; company dashboards and filings for Twenty One Capital, Metaplanet, Strive and Capital B.