Data note: The latest tranche shown below is from (14 days ago). Strategy's purchase cadence is roughly weekly; if newer disclosures exist this site may be lagging the upstream refresh. Pages remain accurate for the period covered.
One 8-K, one week, June 30, 2025

Three ATM programs raised $578.1M. A 4,980 BTC purchase consumed $532.0M. Two preferred dividends went out for $46.1M. The whole loop fit in one filing.

The day the flywheel closed
in a single filing

Most weeks at Strategy, the capital-raise side and the BTC-purchase side and the dividend-payment side each show up in different 8-Ks, on different dates, with different reporting periods. On June 30, 2025 they collapsed into one document. This page walks that document line by line — and then walks the two ways you can tell the story of where the dividend cash actually came from.

The source filing

What the 8-K actually said

Filed June 30, 2025. Six disclosures, one document. View on SEC EDGAR →

Disclosure Shares Net proceeds / paid Per share
MSTR common ATM 1,354,500 +$519.5M ~$383.54
STRK preferred ATM 276,071 +$28.9M ~$104.68
STRF preferred ATM 284,225 +$29.7M ~$104.49
BTC purchase 4,980 BTC −$532.0M $106,801 / BTC
STRK Q2 dividend paid to holders 2025-06-15 −$2.00 / sh declared 2025-06-02
STRF Q2 dividend paid to holders 2025-06-15 −$2.64 / sh incl. accrued from 2025-03-25
“On June 30, 2025, the Company funded these dividend payments using net proceeds from the sale of shares under its at-the-market offering programs, for general corporate purposes, which may include payment of dividends on its preferred stock.”
— 8-K filed 2025-06-30, "STRK and STRF Dividends" section (PDF in the project archive)
The arithmetic

A week-long cash identity

+$519.5M  MSTR common ATM
+$28.9M   STRK preferred ATM
+$29.7M   STRF preferred ATM
────────
+$578.1M  total raised

−$532.0M  4,980 BTC @ $106,801 avg
−$46.1M   STRK + STRF Q2 dividends
────────
≈ $0     residual (within rounding)

tranches.csv row 70 records the same balance under the raise_vs_btc_delta_usd field: $46.1M. That delta is the canonical closed-loop signature — when it shows up on a tranche row, it means the raise wasn't entirely converted to BTC, and the residual was used for the next-most-senior obligation in the capital stack. On June 30, that obligation was the STRK and STRF Q2 coupons.

Framing #1 — physical

No BTC was sold this week

Read the 10-K Consolidated Statement of Cash Flows for FY2025 and you will not find a line for "proceeds from sales of digital assets." Strategy did not sell any bitcoin in 2025 — not on June 30, and not on any other day of the year. The BTC pool only grew. Per the tranche detail page, holdings went from 592,345 BTC to 597,325 BTC across this disclosure window: net +4,980 BTC, no offsetting sale.

So where did the $46.1M of dividend cash come from? From the people who bought MSTR, STRK, and STRF shares that week. The cash flow looks like this:

ATM buyers wire $578.1M to Strategy's brokerage account.
   
Strategy routes $532.0M of that cash to Coinbase Prime & Fidelity.
   
4,980 BTC arrive in cold storage; treasury balance: 597,325 BTC.
   
Strategy wires $46.1M to its transfer agent.
   
STRK + STRF holders of record (2025-06-15) get cash in their brokerage accounts.

Physically, the cash is fungible — it didn't matter which $46.1M sat in the operating account at 11:59 a.m. on June 30. But the provenance matters for the story: every dollar in that operating account came in the door earlier the same week, from someone who bought shares. None of it came from selling BTC. The treasury pool just sits there.

This is the framing that ties to the 10-K. It is the legally-correct framing. It is also the framing that loses something important about how the company actually works. That is what the next section is for.

Framing #2 — atomic

The dividend was, in BTC terms, a sale

The site's preferreds and BTC-cost pages model every dollar of dividend obligation as if it were paid by selling BTC at the period's spot price. This is not what physically happened. It is what economically happened, once you treat each preferred tranche as a closed system with its own BTC allocation.

At $106,801 per BTC (the week's purchase average), $46.1M of preferred dividends is:

$46,100,000
÷ $106,801 / BTC
═══════════
≈ 432 BTC

In the atomic model, the STRK and STRF tranches each "sold" their share of those 432 BTC at $106,801 to honor the Q2 coupon. The STRK origination tranche and the STRF origination tranche each carry their own BTC allocation, computed from the dollars they raised at IPO (closed Feb 5 for STRK, Mar 25 for STRF) divided by the BTC price at the next tranche disclosure. Every coupon period thereafter, the tranche's BTC balance ticks down by (coupon paid) ÷ (period BTC spot).

The same 432 BTC appears nowhere in the cash flow statement. Strategy didn't sell them; they're still in cold storage. But the BTC-equivalent claim against the treasury, owned by STRK + STRF shareholders, just shrank by 432 BTC. The site treats that shrinkage as a sale because it forces the math to converge: each preferred tranche has a finite lifetime BTC cost in present value, and the running tally tracks toward that terminal number coupon by coupon. See /btc-cost for the geometric-series mechanic.

Two framings, same balance sheet. The 10-K cares about cash provenance — physically, no BTC was sold. The site cares about per-tranche economics — economically, 432 BTC of claim was retired from the STRK + STRF tranches. Both ledgers tie to the same ending balance (597,325 BTC in the cold-storage pool, $46.1M less cash than the raise total would imply). They just answer different questions about the same week.
Why common holders applaud this

Per-share BTC went up, not down

A common reader objection: if you issued 1.35M new MSTR shares this week, doesn't every existing share now own less BTC? Arithmetically, yes — if the ATM had been issued at exactly mNAV = 1.0. But it wasn't. The mNAV-greater-than-one premium turns dilution into accretion.

Strategy's own reported "BTC Yield" for this disclosure window:

BTC Yield QTD
+7.8%
+4,440 BTC equivalent
BTC Yield YTD
+19.7%
+9,436 BTC equivalent
Treasury after
597,325
BTC in cold storage

BTC Yield is the change in assumed-diluted BTC-per-share, which is the metric that matters to a common holder. The quarter-to-date number of +7.8% says that for every existing diluted MSTR share, the BTC claim grew by 7.8% over Q2 — net of the 1.35M new shares issued this very week, and net of the 432 BTC of claim that walked out the door as STRK + STRF dividends.

The mechanic that makes this work: at issuance, each new MSTR share brought in ~$383 of cash. At spot, that cash bought ~0.00359 BTC. But the share's pro-rata claim on the treasury after dilution was less than 0.00359 BTC — because mNAV was greater than 1. Every share issued above mNAV = 1 leaves the existing pool's per-share BTC higher than it was the minute before, even after the new share is counted in the denominator. That's the trick. The ATM only "buys" BTC in the closed-system sense if you're modeling existing holders' position. In cash-flow terms, the ATM just raised dollars; the BTC pool grew because those dollars were spent on BTC.

Why this week matters

The flywheel, in a single document

On most Mondays Strategy files one 8-K covering the prior week's BTC purchase and ATM activity. The dividend declarations land in their own filings, weeks ahead of the payment date. The payment itself shows up in the next 10-Q's financing section, sometimes a quarter after the fact. To see the entire mechanic on one page, you usually have to assemble it from four filings spread over two months.

The June 30, 2025 8-K is the exception. The ATM table, the BTC purchase, and the "STRK and STRF Dividends" section are all on the same document, with the same date, citing the same operating account. The closing sentence — “the Company funded these dividend payments using net proceeds from the sale of shares” — is the only place in Strategy's 2025 disclosures where management explicitly draws the line from the ATM proceeds to the dividend check.

It is also the cleanest single-document refutation of the lazy critique that says "Strategy will eventually have to sell BTC to pay preferred dividends." Here, in the filing itself, the company shows you it doesn't. The dividend is paid out of newly-raised cash. The BTC pool grows. Per-share BTC for common holders grows. The preferred holders get their coupon. The loop closes.

The annual reconciliation is the macro view of this mechanic: $24.8B raised, $22.5B to BTC, $381M to preferred dividends across FY2025. The preferreds page forward-models the per-tranche BTC cost of every future coupon. This page is the worked example that ties the two together — one week, six numbers, one 8-K. If the flywheel ever stops working, it will stop working in a filing that looks like this one, with the residual line on the wrong side of zero.