Capital B's $78 Million "Loss" Is an Accounting Ratchet — And the Bitcoin Treasury Flywheel Is Showing Its Teeth
Capital B lost $78 million in the first half. Roughly $75.9 million of that — about 97% — is a Bitcoin impairment charge. Not a cash outflow. Not a liquidation. Not a missed coupon. A bookkeeping adjustment, mandated by an accounting standard most crypto investors have never read, applied to an asset that never moved a single satoshi.
That is the headline. Here is the story the headline buries.
When a company's entire income statement collapses into one non-cash line item, the company has stopped operating a business. It is operating a balance sheet. And a balance sheet only tells the truth when you know which ruler measured it. Capital B just handed the market a number. The market is about to misprice the reason.
To understand why a $78 million "loss" can coexist with an unchanged Bitcoin stack, you have to understand the two accounting regimes now colliding across the treasury-company sector.
Under US GAAP, ASU 2023-08 took effect for fiscal years beginning after December 15, 2024. Crypto is marked to fair value every reporting period, and the movement flows straight to the income statement. Two-directional. Symmetrical. A Bitcoin rally shows up as a gain the same quarter it happens.
Under IFRS — the standard Capital B almost certainly reports under — the dominant treatment for crypto held as an intangible asset is IAS 38: cost model, plus impairment testing under IAS 36. This is a one-way mirror. When Bitcoin rises, book value stays frozen at cost. When Bitcoin falls below carrying value, the company writes it down immediately and books a loss.
I spent part of last year reconstructing the disclosure notes of four European treasury vehicles for a client, cross-referencing their impairment schedules against spot BTC prints. The pattern was mechanical. Every drawdown produced a loss. Every recovery produced silence. Same asset, same price, two entirely different earnings paths — decided not by economics, but by a choice of accounting framework buried in the footnotes.
Capital B sits on the European side of that divide. It is a French-listed vehicle, associated with the asset manager TOBAM, positioned in a crowded lane alongside The Blockchain Group and, further out, Metaplanet in Japan and Strategy in the United States. Its market capitalization is small. Its influence on spot Bitcoin is effectively zero. Its influence on narrative is not. This is the same playbook I mapped during the Terra post-mortem — when a mechanism is misunderstood, the market prices the story first and the balance sheet last.
A caveat before I go further: the disclosure here is thin. No reporting dates, no BTC quantity, no average cost, no mNAV, no stated accounting policy, no debt structure. Everything below is conditional reasoning built on the three facts we actually have — the net loss, the impairment figure, and the word "wider."
Build the mechanism from first principles.
First, the asymmetry. Under a cost-plus-impairment model, Bitcoin's volatility is filtered through a ratchet. Down moves are recognized. Up moves are deferred. This does not merely delay earnings — it structurally biases the reported loss line upward during any period that contains a drawdown. A treasury company can hold the exact same number of coins across two consecutive quarters and still report a catastrophic loss in the second, purely because the quarter-end print dipped below cost.
That is the single most important technical fact in this story, and it is almost never stated in the coverage: the impairment is a measurement artifact of the reporting date, not a verdict on the asset.
Second, the reversal channel. Here is where IFRS and GAAP genuinely diverge. IAS 36 permits reversal of a prior impairment when the recoverable amount rises again — capped at original cost. US GAAP, under ASU 2023-08, forbids reversal but recognizes fair-value gains directly. Same BTC rebound, opposite mechanics. An IFRS filer books a "reversal gain." A GAAP filer books a fair-value gain. The income statements look unrelated. The balance sheets are identical.
If you are pricing these companies off net income, you are pricing them off an accounting dialect. Code talks, but stories sell — and the story here is being told in a language half the market cannot read.
Third — and this is where the treasury model actually lives or dies — the flywheel. Bitcoin treasury companies are not valued on earnings. They are valued on mNAV: market cap divided by net asset value, where the asset is the Bitcoin stack.
The mechanic is simple enough to write as code:
if mNAV > 1: issue equity at a premium, buy more BTC, BTC-per-share rises, premium reinforces. Else: issuance dilutes, funding window closes, accumulation halts, premium decays toward NAV.
That loop is the entire business model. It has nothing to do with protocol revenue, because there is no protocol. Real revenue share: zero percent. This is pure asset-price exposure wrapped in a financing arbitrage. The "tokenomics" equivalent is capital-structure engineering: ATM programs, convertibles, private placements — the treasury-company analogues of inflation and buyback.
Now the honest part, and the part most critics get wrong. This is not a Ponzi. A Ponzi requires a rigid redemption promise backed by nothing, where early exits are paid by late entrants and the structure collapses the moment inflows stop. Here, the underlying asset — Bitcoin — is liquid, non-issuer-controlled, and does not require new money to settle anything. The correct label is reflexivity, not fraud. Perception feeds price feeds perception, until it doesn't. Early holders of the flywheel were paid partly by later premium buyers, yes — but the asset underneath is real, and that distinction is the whole game.
Fourth, what the impairment tells us that the net loss hides. Impairment equals roughly 97% of the reported loss. Strip it out and Capital B's operating business contributes almost nothing — a rounding error in either direction. That is not an insult; it is a definition. The company has fully financialized into a Bitcoin exposure vehicle. Its P&L is a function of BTC price and accounting policy. Full stop.
I tried to reverse-engineer the stack size from the impairment figure. Assume a write-down of $75.9 million, assume carrying cost sat somewhere between 10% and 25% above period-end market. That implies a Bitcoin position worth somewhere between $300 million and $750 million at the measurement date. I will flag my own confidence as low — the estimate is violently sensitive to the cost basis and the exact impairment timestamp, neither of which is disclosed. Treat it as an order-of-magnitude sanity check, not a number.
Fifth, the hidden clock. Impairment is typically tested semi-annually or annually. That introduces lag. The loss you are reading about today may reflect a price low from three to five months ago. The market has already moved. The income statement is reporting the past in the present tense — and investors are trading the past as if it were the future.
Everyone is reading this as a loss story. Read it as a funding story.
The number that matters is not $78 million. The number that matters is the one absent from the release: mNAV. If Capital B's market cap is still trading above the value of its Bitcoin, the flywheel spins and this impairment is noise — an accounting scar on a machine that still works. If mNAV has slipped below 1, then every share issued is dilution, the funding window is shut, accumulation stops, and the "treasury company" quietly becomes a closed-end fund trading at a discount to its only asset.
The impairment headline is a tell. A company confident in its premium does not lead with a loss. The framing choice — "wider net loss" rather than "impairment charge" — is itself a narrative decision, and narratives get priced. That is the blind spot: analysts will spend the week debating whether the loss is "real," when the real question is whether the flywheel still turns. One is an accounting argument. The other is a solvency-adjacent question about the business model itself.
Zoom out, and the whole sector just crossed a line. The story was "BTC per share is climbing." The story is now "impairment is widening." That is a narrative regime change, not a quarterly blip. Hype decays; utility endures — and the treasury trade was always more hype than utility. The only treasury companies that survive the transition are the ones whose financing machinery outlives the premium.
The next leg of this trade will not be decided by Bitcoin's price. It will be decided by a disclosure note most readers never open — the mNAV, the cost basis, the accounting policy. Watch the filings, not the headlines. If the premium holds, the loss is a ghost. If the premium breaks, the loss is a preview of the exit.
Narrative is the new liquidity. And right now, the treasury companies are discovering that liquidity has a price.